Dictionary.json

{
    "abide": {
        "abide by": {
            "sentences": [
                "First, the tuning system should always abide by the ser\u0002vice level agreement (SLA) while optimizing the resource utilization,",
                "Non-adversarial mining power abides by the protocol."
            ]
        }
    },
    "ability": {
        "ability to": {
            "sentences": [
                "reinforcement learning approaches have long appealed to the data management community due to their ability to learn to control dynamic behavior from raw system performance."
            ]
        }
    },
    "able": {
        "be able to": {
            "sentences": [
                "Our cost model is able to capture the hardware characteristics as well as query statistics to give an accurate estimation.",
                "Results show that our scheme is able to closely match the system mining power and the mining difficulty for each individual chain, thus keeping the chain forking rate stable."
            ]
        }
    },
    "abort": {
        "abort a transaction": {
            "sentences": [
                "The coordinator decides to commit or abort a transaction based on decisions collected from workers in the prepare phase"
            ]
        },
        "abort sth": {
            "sentences": [
                "The coordinator decides to commit or abort a transaction based on decisions collected from workers in the prepare phase"
            ]
        }
    },
    "abstract": {
        "abstract from": {
            "sentences": [
                "Tencent-RB is a real benchmark abstracted from the Tencent billing service platform."
            ]
        },
        "abstract as": {
            "sentences": [
                "We abstract the hash function as a random oracle functionality."
            ]
        }
    },
    "accelerate": {
        "accelerate by": {
            "sentences": [
                "Linet al.[59] accelerated the inference process by constructing a lookup table for both input and function spaces learned by PointNet."
            ]
        },
        "accelerate even further": {
            "sentences": [
                "We expect the trendof using ML for DB will accelerate even further in the future, andhope that our work in initiating a formal study of the learnabilityof DB problems will complement the existing efforts at leveragingdeep learning for improving the performance of DB systems inpractice."
            ]
        },
        "accelerate sth.": {
            "sentences": [
                "GPUs also have significant potential to accelerate memory-bound applications such as database systems."
            ]
        },
        "accelerate the recovery process": {
            "sentences": [
                "Instead, it complements them because it accelerates the recovery process through simultaneous reduction of recovery communication costs and parallelization of the recovery computations"
            ]
        },
        "accelerate sth": {
            "sentences": [
                "Oracle DBIM accelerates HTAP workloads by maintaining an in-memory column store for active data.",
                "To accelerate the tuning process by reducing the budget to tens of iterations"
            ]
        }
    },
    "accept": {
        "be accept in": {
            "sentences": [
                "The possibility of data loss can be accepted in exchange for possibly better throughput"
            ]
        },
        "accept as long": {
            "sentences": [
                "Applications that need the highest level of availability  4.8.1  Ring Membership  can set W to 1, which ensures that a write is accepted as long as a  single node in the system has durably written the key it to its local  In Amazon\u2019s environment node outages (due to failures and  store."
            ]
        }
    },
    "access": {
        "access to": {
            "sentences": [
                "Furthermore, the model can access to a background KG",
                "In DA, a learner usually has access to the labelled source data and unlabelled target data, and it is typically trained to align the feature distribution between the source and target domain.",
                "Jason et al. studied a set of database primitives on the integrated GPU of a coupled architecture and show that transparent access to CPU virtual addresses and very low overhead of computation offloading are important for query co-processing performance.",
                "Since the ground truth for the test set is not available and the access to the test server is limited,",
                "When designing a testing tool, access to source code and some knowledge about the PUT are often assumed.",
                "Note that nr can only be determined by examining the view of all honest parties and is not a quantity that is accessible to any of the honest parties individually.",
                "In other words, we only access to the effect, but not the causes"
            ]
        },
        "access via": {
            "sentences": [
                "the blocks are accessed via two index arrays, a vertex index and an edge index, storing pointers to appropriate blocks by vertex id."
            ]
        },
        "access from": {
            "sentences": [
                "Generally, servers can be accsessed from multiple clients at the same time"
            ]
        },
        "access the main memory": {
            "sentences": [
                "The RDMA feature allows a machine to directly access the main memory of another machine without the involvement of the operating systems of either side"
            ]
        },
        "be access in": {
            "sentences": [
                "A volume can be randomly accessed in 512B alignment"
            ]
        },
        "access as well": {
            "sentences": [
                "The speedup is a function of the fraction of the record accessed as well as Acknowledgements the selectivity."
            ]
        },
        "access in only": {
            "sentences": [
                "Sec. 4.2.1 showed that \u224850% of ETC\u2019s keys are accessed in only 1% of requests, and therefore bene\ufb01t We created functional models by \ufb01tting various distribu-"
            ]
        }
    },
    "accesse": {
        "accesse to": {
            "sentences": [
                "On multicore CPUs, this could present a scalability bottleneck as an SMO will block concurrent accesses to the node."
            ]
        }
    },
    "accommodate": {
        "accommodate the rapidly": {
            "sentences": [
                "Extending these frameworks to accommodate the rapidly diversifyinglandscape of DL models and hardware platforms presentschallenging tradeoffs between expressivity, composability, andportability."
            ]
        },
        "accommodate to": {
            "sentences": [
                "One one hand, LBE can accommodate to a wide range of popular classifiers such as LF and MLP presented in this paper."
            ]
        }
    },
    "accompany": {
        "accompany ..with..": {
            "sentences": [
                "We also accompany it with an online appendix with detailed proofs and numerous additional experiments"
            ]
        },
        "be accompany by": {
            "sentences": [
                "For all nine questions, each response option was accompanied by an explanation to support its understanding."
            ]
        }
    },
    "accomplish": {
        "accomplish by": {
            "sentences": [
                "although the ldbc workload driver provides a time compression ratio parameter to achieve certain throughputs, this is accomplished by coarsening the granularity of scheduled and dependency times."
            ]
        },
        "accomplish with": {
            "sentences": [
                "This task is usually accomplished with help from existing knowledge bases."
            ]
        }
    },
    "accord": {
        "accord to": {
            "sentences": [
                "For example, character \u201c\u8fd8\u201d has two 3Among 7000 common characters in Chinese, there are about 700 characters that have multiple pronunciations, according to the Contemporary Chinese Dictionary",
                "according to [yao et al., 2019], above 40.7% of relations can only be identified at the document level.",
                "According to the correlation between wavelength and fog concentration",
                "Then, according to the correlation between wavelength and color, a transmission estimation algorithm based on maximum fuzzy correlation graph cut is designed.",
                "According to the hypothesis that words with similar context have similar semantics",
                "It should be noted that R\u03b3,b in Algorithm 1 can be applied in Randomize(\u00b7) as a basic randomizer, which accords to estimating with equation (1) for line (18)",
                ":The difficulty of the proposer blocks is adjusted according to the Bitcoin rule [13].",
                "the features are then fed into self-attention computation and each token, according to the vanilla design, can exchange information with all the others at every single layer."
            ]
        }
    },
    "account": {
        "account for": {
            "sentences": [
                "Then we account for the central DP against the analyzer",
                "Follow up works [23, 59] extended the open set framework to multi-class classifier setting with probabilistic models to account for the fading away classifier confidences in case of unknown classes.",
                "does not account for the appearance at other locations of the reference image.",
                "This flexibility allows users to account for changes in their interpretation of policy text as they read and understand the privacy policy to answer successive ques- tions.",
                "To account for such hybrid schemes, we define a generic scheme that combines universe and Bernoulli sampling.",
                "These are called here minimum temporal paths, and account for different measures.",
                "They account for over 50% of total time on 17 of the 22 queries on the TPC.",
                "average errors amount and quality of resources are accounted for jointly.",
                "Further, in addition to traditional position embeddings, this work accounts for the hierarchical nature of tabular data using tree-based positional embeddings.",
                "To account for class imbalances, we evaluate model performance using the average F1-score, weighted by the number of columns per class in the test set.",
                "Each group accounts for a dedicated attention granularity.",
                "Depth residual \u2013 which accounts for the fact that if one is an outlier",
                "Each robot accounts for its own state."
            ]
        }
    },
    "accurate": {
        "be accurate by": {
            "sentences": [
                "Thus, model \u201cpredictions\u201d in inner nodes during traversal are accurate by construction."
            ]
        }
    },
    "achieve": {
        "achieve for": {
            "sentences": [
                "APEX also achieves crash consistency for all operations with low overhead."
            ]
        },
        "achieve by": {
            "sentences": [
                "The invariance to geometry transformation is achieved by learning from relative euclidean and direction distances.",
                "achieving by up to 4x for a broader class of workloads with different index operations.",
                "livegraph is the first system that guarantees these properties, achieved by co-designing a graph-aware data structure (section 3) and the concurrency control algorithm (sections 4 and 5) to ensure purely sequential scans even in the presence of concurrent transactions.",
                "pre-trained models achieve performance improvement by training knowledge"
            ]
        },
        "have achieve": {
            "sentences": [
                "Calvin has already achieved over traditional nondeterministic execution schemes in the presence of distributed transactions."
            ]
        },
        "achieve target": {
            "sentences": [
                "To achieve such target, a common and reasonable idea is to train different embeddings with cause-specific data."
            ]
        },
        "achieve a more": {
            "sentences": [
                "however, all of these approaches are orthogonal to the idea of learned indexes as none of them learn from the data distribution to achieve a more compact index representation or performance gains."
            ]
        },
        "achieve tremendous success on": {
            "sentences": [
                "Recent pre-trained language models (LMs) such as BERT (Devlin et al., 2019) and RoBERTa (Liu et al., 2019) achieve tremendous success on a spectrum of natural language processing tasks, including semantic parsing (Zettlemoyer & Collins, 2005; Zhong et al., 2017; Yu et al., 2018b)."
            ]
        },
        "be achieve by": {
            "sentences": [
                "The best result on SPIDER is achieved by using GRAPPA pre-trained for only 5 epochs on our relatively small pre-training dataset."
            ]
        },
        "achieve much higher performance than": {
            "sentences": [
                "However, a critical difference is that we focus on the supervised learning setting, where the model parameters are fine-tuned to learn from given examples to achieve much higher performance than few-shot learning or non-parametric methods."
            ]
        },
        "achieve up": {
            "sentences": [
                "We show that it achieves up to 10x and 3.5x speed-up on the radix-join against CPU and GPU alternatives, respectively, and 1.6x-8x against state-of-the-art CPU- and GPU-based commercial DBMSs on the selected TPC-H queries."
            ]
        },
        "achieve sth.": {
            "sentences": [
                "The results on all metrics show that our proposed Though using only language-style discriminator achieves model greatly outperforms all the baselines."
            ]
        },
        "achieve in": {
            "sentences": [
                "Convolutional neural networks (CNNs) have achieved great successes in many computer vision problems."
            ]
        },
        "achieve fault tolerance\u3001": {
            "sentences": [
                "One key challenge in such design is to achieve fault tolerance when the CPUs on backup servers do not participate in the replication protocol"
            ]
        },
        "achieve sth": {
            "sentences": [
                "we achieve superior pelf-training"
            ]
        },
        "achieve via": {
            "sentences": [
                "communications between windows are achieved via the proposed msg tokens, which avoid frequent feature dimension transitions and maintain high concision and efficiency."
            ]
        },
        "achieve on": {
            "sentences": [
                "Transformer [24] has achieved surprising success on machine translation"
            ]
        },
        "achieve with": {
            "sentences": [
                "This section shows that the generated algorithms achieve similar optimization effects with their Paxos counterparts"
            ]
        }
    },
    "acquire": {
        "be acquire": {
            "sentences": [
                "For thread safety, the lock head generally stores a mutex object, which is acquired before lock requests and releases to ensure that adding or removing elements from the linked list always occurs within a critical section"
            ]
        },
        "acquire by": {
            "sentences": [
                "The information on which this decision is based is acquired by PREPROCESS and CONFUPDATE, which augment C with this knowledge.",
                "3D data acquired by these sensors can provide rich geometric, shape and scale information [2], [3]."
            ]
        },
        "be acquire by": {
            "sentences": [
                "Further, the semantic space is acquired by jointly embedding each photo with its corresponding contextual sentence such that their correlations are discovered (Liu et al., 2017)."
            ]
        },
        "acquire with": {
            "sentences": [
                "We propose Generative Style Decoder to acquire long sequence output with only one forward step needed."
            ]
        },
        "acquire lock": {
            "sentences": [
                "In the read phase, transactions perform read operations from the database and write operations to local copies of objects without acquiring locks"
            ]
        },
        "acquire sth": {
            "sentences": [
                "In the read phase, transactions perform read operations from the database and write operations to local copies of objects without acquiring locks"
            ]
        }
    },
    "acr": {
        "act as": {
            "sentences": [
                "The second part acts as a solver for non-linear problem."
            ]
        }
    },
    "act": {
        "act as": {
            "sentences": [
                "The BPIIoT platform will act as a key-enabler for cloud-based manufacturing",
                "hash-based sketches act as a key computational bottleneck depending on the workload.",
                "This is achieved through a visibility matrix, which acts as an attention mask so that words in each cell can only aggregate information from other structurally related words during the \ud835\udc40\ud835\udc62\ud835\udc59\ud835\udc61\ud835\udc56\ud835\udc34\ud835\udc61\ud835\udc61 calculation.",
                "Then, the worker adds the value (i.e. number of other workers marking this record) to its local contention counter, which acts as a proxy for record contention",
                "Our approach establishes pointwise correspondences from off-the-shelf local features: it acts as a middle-end between handcrafted or learned front-end and back-end.",
                "which acts as an overflow area."
            ]
        },
        "act like": {
            "sentences": [
                "It acts like a global average pooling but with different attention for different plane instances"
            ]
        },
        "act with": {
            "sentences": [
                "Gato can sense and act with different embodiments across a wide range of environments using a single neural network with the same set of weights."
            ]
        },
        "act as a direct replacement": {
            "sentences": [
                "We introduce GOCor, a differentiable neural network module that generates the correspondence volume between a pair of images, acting as a direct replacement to the feature correlation layer."
            ]
        },
        "act on": {
            "sentences": [
                "Therefore, the inherent negative feedback present in NEFDA is the only force acting on solve times."
            ]
        }
    },
    "adapt": {
        "adapt to": {
            "sentences": [
                "we observe that the replay time for each iteration takes at least 5 minutes to adapt to different types of transactions",
                "A key requirement of practical PoW blockchains is to adapt to mining power variations over time.",
                "we perform a thorough evaluation of the proposed flamingo models, exploring and measuring their ability to rapidly adapt to a variety of image and video understanding benchmarks.",
                "since the resource that adapts to the workload to meet the time condition",
                "building models that can be rapidly adapted to numerous tasks using only a handful of annotated examples is an open challenge for multimodal machine learning research.",
                "they yield satisfactory results on specific datasets, they cannot adapt to the complex environments commonly encountered in autonomous driving.",
                "as such, one may wonder how well these techniques are able to adapt to more distant tasks.",
                "The methods in segmentation almost don\u2019t utilize the distribution matching scheme and can\u2019t adapt to our problem.",
                "In addition, JIT allows the execution model to be adapted to each device providing enough flexibility for efficient inter-operator execution.",
                "By this general learning objective, state-of-the-art PU methods based on optimizing a consistent risk estimator can be adapted to conquer the imbalance",
                "Therefore, concurrency control in transactional databases, in addition to preserving ACID properties, should adapt to workload changes without manual intervention or disrupting performance.",
                "The ability to adapt to new relations is limited for current models",
                "However, if fine-tuning is used, the last layers of the encoder can be tuned to adapt to the recognition task",
                "These procedures are based on the graph transformation approach introduced by Wu for DLTGs, which was adapted to address the temporal graph model of Definition 4.",
                "Specifically, we propose three private dimension selection mechanisms and adapt the gradient accumulation technique to stabilize the learning process with noisy updates.",
                "To avoid such a case, we adapt BERT to a feature-based approach. First we use the fine-tuning approach to train BERT without additional features like in Section 6.1.",
                "In contrast, our proposed adaptive logging can adapt to the dynamic change of the workload properties",
                "GNN models for network localization essentially aim to recognize the most distinctive feature that correspond mostly to the neighbor nodes."
            ]
        },
        "adapt for": {
            "sentences": [
                "APEX further adapts optimistic locking for learned indexes on PM",
                "Recent work have also considered adapting Transformer based models for time series\nforecasting (Li et al., 2020; Zhou et al., 2021), specifically focusing on tackling the quadratic space\ncomplexity of Transformer models."
            ]
        },
        "adapt from only": {
            "sentences": [
                "to  do  so,  they  build  upon  their  prior experience, reusing concepts and abstractions that they have built up over time to ef\ufb01- ciently adapt from only small amounts of new evidence."
            ]
        },
        "adapt from": {
            "sentences": [
                "We propose a new algorithm with the central DP guarantee, LFL-SGDMF, which is adapted from the FMF as shown in Figure 3(d)",
                "A straightforward way to address Semi-Supervised Object Detection (SS-OD) is to adapt from existing advanced semi-supervised image classification methods"
            ]
        },
        "adapt sth for": {
            "sentences": [
                "We then make the case for adapting the logging protocol for NVM."
            ]
        },
        "adapt do": {
            "sentences": [
                "We adopt DQN as our modelfor the following reasons"
            ]
        },
        "be adapt to": {
            "sentences": [
                "The advances in Open Set and Open World image clas- sification cannot be trivially adapted to Open Set and Open World object detection"
            ]
        }
    },
    "add": {
        "add for": {
            "sentences": [
                "add support for updating the index"
            ]
        },
        "add to": {
            "sentences": [
                "Multiple text segments can be added to (and removed from) the selection field.",
                "The first column of the CONFUPDATE section indicates if fuzzers evolve the seed pool during CONFUPDATE, such as adding new seeds to the pool.",
                "The blocks are added to the Blockchain at regular intervals",
                "\u2022 Glyce-BERT layer Position embeddings are first added to the glyph embeddings",
                "The computed clustering loss is added to the standard detection loss and back-propagated to learn the network end-to-end.",
                "Raft also adds another restriction to leader election",
                "These two models are similar in mitigating the uncertainty by adding stochastic perturbation to\nthe encoders.",
                "The remaining loop closures are added to the PCM graph one by one.",
                "adding more context  to the shape description.",
                "In the case of our subset-sum sampling implementation, some of the functions that we added to the Gigascope runtime library are:",
                "As we mentioned above, a small compression ratio results in a smaller sensitivity, and hence less Gaussian noise will be added to the model.",
                "A transaction proposal zone is added to each block, containing txpids \u2014 the first several bytes of transaction hashes\u2014of some possibly fresh transactions.",
                "Further more, a refinement module is added to optimize the pose after the decoupled initialization.",
                "Additional steps for privacy will be considered in future versions of the algorithm, adding noise to the gradients that are shared and also sharing random gradients for items"
            ]
        },
        "add by": {
            "sentences": [
                "Additional dimensions may be added by computing normals and other local or global features."
            ]
        },
        "add into": {
            "sentences": [
                "if so, a new triplet is extracted and added into the resulting set t. 3.2 token pair representation given a sentence of length n, we first map each token wi into a low-dimensional contextual vector hi by a basic encoder.",
                "To make features of adjacent vertices more similar, a graph-signal smoothness prior is added into the loss function."
            ]
        },
        "add sth. to sth.": {
            "sentences": [
                "In every round, for the set of keys that have found their matches, we calculate their sum, add it to a local sum, and reload those SIMD lanes with new keys."
            ]
        }
    },
    "addition": {
        "in addition  to": {
            "sentences": [
                "In addition to the above, we also introduce a novel angle loss regression approach to solve the problem of the large loss generated"
            ]
        },
        "in addition to": {
            "sentences": [
                "In the data set used for training, there will be corresponding RGB images in addition to the point cloud data.",
                "In addition to improving image quality, camera calibration is often used to determine the position of an image in 3D space."
            ]
        }
    },
    "address": {
        "address sth": {
            "sentences": [
                "To address these reasoning challenges, we design two different models: Table-BERT and Latent Program Algorithm (LPA)."
            ]
        },
        "address question": {
            "sentences": [
                "In this paper, we address these questions, show how to integrate the radix join in Umbra, a code-generating DBMS, and make it competitive for selective queries by introducing a Bloom-filter based semi-join reducer."
            ]
        },
        "address the challenge": {
            "sentences": [
                "To address the challenge, previous work [54, 55] of using AR models propose to use the progressive sampling algorithm to estimate selectivities for range queries."
            ]
        },
        "address before": {
            "sentences": [
                "as such, there are a multitude of data challenges that must be addressed before text-to-image models like imagen can be safely integrated into user-facing applications."
            ]
        },
        "address by": {
            "sentences": [
                "This problem has been addressed by using bagging techniques or using least square SVMs."
            ]
        },
        "be address in": {
            "sentences": [
                "The problem of analyzing the difficulty adjustment mechanism in Bitcoin was first addressed in [12] in the lock-step synchronous communication model."
            ]
        },
        "address on": {
            "sentences": [
                "nodes negotiate about the destination memory address on the remote node using Send/Recv verbs"
            ]
        },
        "address as": {
            "sentences": [
                "For instance, collection of financial (Q3) and health information (Q4) practices are often not as explicitly and concisely addressed as col- lection of contact (Q1) or location (Q2) information."
            ]
        },
        "addressed, however": {
            "sentences": [
                "This issue is addressed, however, by the refined partitioning  scheme described in Section 6.2."
            ]
        },
        "address such ifthevaluesofthatcolumnarestoredcontiguously": {
            "sentences": [
                "To address such ifthevaluesofthatcolumnarestoredcontiguously."
            ]
        }
    },
    "adjust": {
        "adjust for": {
            "sentences": [
                "Popular entity experts allow rankings to be adjusted for particular entities that are especially important"
            ]
        },
        "adjust to": {
            "sentences": [
                "Instead, we propose audio-adaptive learning that exploits the supervisory signals from sound to adjust to the distribution shift and handle both audible and silent activities.",
                "It has an autonomic controller to automatically reconfigure itself to adjust to the changes in the resource demand."
            ]
        },
        "adjust to widely": {
            "sentences": [
                "Subsequent methods do notrequire any prior knowledge of the cardinality, and adjust to widely varying cardinalities."
            ]
        }
    },
    "adopt": {
        "adopt in": {
            "sentences": [
                "Blockchain is a nascent technology and is mostly adopted in the financial sector",
                "Additive models are widely adopted in causal inference.",
                "graphbased approaches are now widely adopted in re because of e1e2e3e4e5e6e7e8 e1 e2 e3 e4 e5 e6 e7 e8 na rel2 rel1 figure 2: illustration of the entity-level relation matrix applied in our formulation."
            ]
        },
        "adopt to": {
            "sentences": [
                "researchers adopt the term ontology to describe what can be computationally represented of the world in a program",
                "Specifically, convolution is usually implemented through MLP over spatial neighbors, and pooling is adopted to produce a new coarsened graph by aggregating information from each point\u2019s neighbors."
            ]
        },
        "be adopt to": {
            "sentences": [
                "a MMD metric is adopted by as a cross-domain regularizer and KL divergence is adopted to measure the domain shift by for domain generalization problem.",
                "This technique is adopted to further improve the model with sequence-level ."
            ]
        },
        "adopt in almost": {
            "sentences": [
                "Cost-basedoptimizer studied in this paper is adopted in almost all currentdatabase systems."
            ]
        },
        "adopt sth. to do": {
            "sentences": [
                "In this paper, we adopt a model-based approach to understand when and why the performance gains of running queries on GPUs vs on CPUs vary from the bandwidth ratio (which is roughly 16\u00d7 on modern hardware)"
            ]
        },
        "adopt with": {
            "sentences": [
                "we adopt a more lightweight and  directly interpretable approach"
            ]
        },
        "adopt as": {
            "sentences": [
                "Here we adopt Euclidean Distance as the metric due to its simplicity and prevalence.",
                "OpenAI proposes GPT [9] that adopts the modified Transformer\u2019s decoder as a language model"
            ]
        },
        "adopt sth": {
            "sentences": [
                "Almost every DBMS adopts the steal and no-force policies."
            ]
        },
        "adopt by": {
            "sentences": [
                "Open-source kernels have been adopted by massive downstream vendors on billions of devices."
            ]
        }
    },
    "advance": {
        "advance to": {
            "sentences": [
                "WSI technology has advanced to the point where digital slide scanners can make high-resolution digital images in a relatively short time.",
                "The graph optimization technology has advanced to the point where VSLAM can make high-resolution digital images in a relatively short time."
            ]
        },
        "advance for": {
            "sentences": [
                "Active learning has advanced for more than a couple of decades."
            ]
        }
    },
    "advantage": {
        "take advantage of": {
            "sentences": [
                "so current cloud platforms have trouble taking full advantage of the emerging hardware standards such as RDMA and NVMe SSD."
            ]
        }
    },
    "advertise": {
        "advertise a slowly": {
            "sentences": [
                "Between synchronizations, a daemon advertises a slowly increasing time uncer- tainty.",
                "Between synchro- nizations, Armageddon masters advertise a slowly increasing time uncertainty that is derived from conservatively applied worst-case clock drift."
            ]
        }
    },
    "advise": {
        "have advise": {
            "sentences": [
                "Recent comprehensive studies have advised that the partitioned radix join performs better than the non-partitioned hash join."
            ]
        }
    },
    "affect": {
        "affect by": {
            "sentences": [
                "Moreover, the experimental analysis in [12] shows that the classification performance could be badly affected by an inaccurate estimate.",
                "representations of rare words affected by heavytailed distribution",
                "Only states affected by the block\u2019s transactions are recorded in the root hash",
                "In both scenarios, ARIES logging shows the best recovery performance and is not affected by the percentage of distributed transaction",
                "third they cannot estimate the index maintenance costs which are affected by multiple index utilization factors and can significantly affect the index benefits"
            ]
        },
        "be affect by": {
            "sentences": [
                "Facial landmark detection may be inaccurate or even fail for faces which are affected by various challenging factors, such as strong illumination changes, large pose variations, and heavy occlusions.",
                "The build phase runtimes are less affected by caches as writes to hash table end up going to memory."
            ]
        }
    },
    "afford": {
        "afford to": {
            "sentences": [
                "we can't afford to go abroad.",
                "Practical deployments of detection systems in robotics, self-driving cars, plant phenotyping, healthcare and surveillance cannot afford to have complete knowledge on what classes to expect at inference time, while being trained in-house"
            ]
        }
    },
    "agent": {
        "agent on": {
            "sentences": [
                "which includes a set of masters implemented in micro-service, and agents deployed on all compute and storage nodes."
            ]
        }
    },
    "aggragate": {
        "aggregate into": {
            "sentences": [
                "The global-local graph aggregation module [40] is proposed to model interactions across graphs and aggregate heterogeneous graphs into a holistic graph representation in the video titling task."
            ]
        }
    },
    "aggravate": {
        "aggravate with": {
            "sentences": [
                "This is because the error brought by local noises can be aggravated with a dimensional factor"
            ]
        }
    },
    "aggre": {
        "aggre with": {
            "sentences": [
                "The play does not aggre with the book."
            ]
        }
    },
    "aggregate": {
        "aggregate... into...": {
            "sentences": [
                "The final fully connected layers of the network aggregate these learnt optimal values into the global descriptor for the entire shape as mentioned above (shape classification) or are used to predict per point labels (shape segmentation)"
            ]
        },
        "aggreagate into": {
            "sentences": [
                "Aggregating the computation of all static string data into one function, however, may be un-stealthy."
            ]
        },
        "aggregate with": {
            "sentences": [
                "By aggregating their data with others in the same cluster (same learning task), they can leverage the strength in numbers in order to perform more efficient Federated Learning."
            ]
        },
        "aggregate use": {
            "sentences": [
                "as one of the most popular models in deep learning, cnns construct a hierarchical structure to learn visual features, and in each layer, local features are aggregated using convolutions to produce features of the next layer."
            ]
        },
        "aggregate by": {
            "sentences": [
                "We uncover the potential of aggregating highlighting by multiple, independent annotators (or algorithms) showing that aggregation is practical and useful, somewhat analo- gously to what happens in a crowdsourced classification where we aggregate multiple votes on items."
            ]
        }
    },
    "agree": {
        "agree with": {
            "sentences": [
                "The results of experiment do not agree with the book.",
                "The calculated results agree with experimental results in terms of stress amplitude impacted by machine stiffness."
            ]
        },
        "agree on": {
            "sentences": [
                "However, computing the difficulty of a block must be deterministic and based on data from previous blocks s.t. individual nodes can perform the computation independently and agree on the same results.",
                "All nodes in a blockchain network agree on an ordered set of blocks",
                "We find that it is exceedingly rare for crowdworkers to agree on an inter- pretation that differs from skilled annotators."
            ]
        },
        "agree well with": {
            "sentences": [
                "The calculated mobilities agree well with the experimental analysis."
            ]
        }
    },
    "aim": {
        "aim at": {
            "sentences": [
                "Unlike all these above pre-trained models that aim at natural understanding or generation tasks",
                "In particular, most prior works aim at reducing communication overhead among operators",
                "This paper aims at developing effieient aggregation algorithms for compressed data warehouses.",
                "Deep learning for sets such as point clouds aims at designing permutation equi- or invariant functions by aggregating information across elements.",
                "In this paper, we propose VPFedMF, a new design aimed at privacy-preserving and verifiable federated MF",
                "These works also aim at establishing reliable point to point correspondences between images.",
                "SigGraph aims at deriving unique signatures for individual data structures for brute force kernel memory scanning.",
                "Our Teacher-Student Mutual Learning stage aims at evolving both Teacher and Student models via a mutual learning mechanism",
                "we aim at complementing local information by specifying the details in the feedforward layer.",
                "We first discuss policies that aim at minimizing the number of conflicts",
                "Some recent studies aim at supporting updates.",
                "In this work, we aim at predicting new facts under a challenging setting where only one training instance is available",
                "We aim at formulating an objective based on the PW-bipath constraint",
                "The recent SemTab 2019 challenge also aims at benchmarking systems that match tabular data to KBs."
            ]
        },
        "aim to": {
            "sentences": [
                "We aim to reduce the persistence operations for NVM space management as much as possible.",
                "The task aims to answer questions over KBs in a conversational manner.",
                "While this has a direct impact on an IPs profits, some of the proposals aim to maximise revenue by increasing capacity to service workloads.",
                "AFQMC is the Ant Financial Question Matching Corpus for the question matching task that aims to predict whether two sentences are semantically similar CSL is the Chinese Scientific Literature dataset extracted from academic papers",
                "Several recent efforts aim to crowdsource the interpretation of privacy policies and use the resulting annotations to build more effective user interfaces that provide users with salient policy summaries.",
                "One-shot relational learning for KG, which aims to uncover facts of a new relation given only one training instance, has been proposed recently by Xiong",
                "Aiming to solve this, we adopt Optimal Transport (OT) with Wasserstein distance to align the feature distribution for domain generalization.",
                "We now aim to verify whether the model trained on 100% supervised data can be further improved by using additional unlabeled data",
                "Image captioning aims to automatically generate textual descriptions of an image.",
                "DNN have been applied in the task of sequential recommendation, which aims to capture the dynamic preference characteristics.",
                "We also adopt the Masked Language Model (MLM) objective from BERT, which aims to model the complex characteristics of word use in table metadata.",
                "Through this work, we aim to raise awareness of the importance of the network layer in designing protocols.",
                "In this paper, we propose an efficient algorithm that aims to address one of the major challenges in FL\u2014dealing with heterogeneity in the data distribution.",
                "This paper aims to enhance the performance of modeling stock\nprice movement by mitigating both data and modeling uncertainty.",
                "Saber is designed aiming to hide PCI-e overhead. It adopts a bulk-synchronous parallel model",
                "MIL models for WSI classification essentially aim to recognize the most distinctive patches that correspond mostly to the slide label.",
                "General code obfuscation techniques aim to con- fuse the understanding of the way in which a program functions.",
                "Instead, we aim to estimate the index update costs",
                "This paper specifically aims to study the fact verification given semi-structured data as evidence.",
                "Interleaving aims to improve the overall throughput of reads and writes by spreading sequential data accesses over multiple DIMMs in parallel, similar to RAID 0.",
                "Contrastive learning aims to discriminate instances by attracting the positive pairs and repelling the negative pairs",
                "End-to-end deep learning approaches aim to learn feature correspondences in depth maps",
                "These systems target and aim to disrupt applications which have so far been implemented on top of database systems",
                "we aim to reduce the sparsity of the stores but the fanout is restricted by the memory available for consolidating the stores.",
                "They aim to estimate the size of t = |A ??B|.",
                "This research aims to carry out the experiment on the zero power reactor and obtain import critical parameters."
            ]
        },
        "aim to independently": {
            "sentences": [
                "to show universality, we will aim to independently control information \ufb02ow from x, from y, and from x?"
            ]
        },
        "aim to empirically": {
            "sentences": [
                "then, in section 4.6.4, we will consider the latter two questions, aiming to empirically study the inductive bias of gradient-based and recurrent meta-learners."
            ]
        },
        "aim for": {
            "sentences": [
                "Based on multi-modal fundus imaging comprised of OCT and color fundus photography, we aim for automated categorization of the fundus condition of a specific eye",
                "We aim for as controlled of a comparison with TaBERT as possible, as its performance on table QA tasks indicate the strength of its table encoder."
            ]
        },
        "aim to do": {
            "sentences": [
                "Commercial systems like Omnisci, Kinetica, and BlazingDB aim to provide real-time analytical capabilities by using GPUs to store large parts (or all) of the working set."
            ]
        },
        "be aim for": {
            "sentences": [
                "Some of them belong to PolarFS itself, and others are aimed for database features."
            ]
        }
    },
    "aleviate": {
        "alleviate this issue": {
            "sentences": [
                "To alleviate this issue,we proposed conditional BERT"
            ]
        }
    },
    "align": {
        "align to": {
            "sentences": [
                "Using basic principles, we analyze how visual classification is extended to handle UDA by aligning a source domain distribution to a target domain distribution to make the classifier domain invariant.",
                "entity mentions in the text are firstly recognized and then aligned to corresponding entities in KGs",
                "For secondary indices, if they align to the partitioning of the primary indices, they are partitioned in the same way.",
                "The algorithm aligns spans in statement to the columns by extracting the longest character n-gram that matches a cell."
            ]
        },
        "align with": {
            "sentences": [
                "Structural lines are aligned with the axes of the dominant MF",
                "The key open questions will revolve around making sure the assumptions and settings considered within PU learning align with real-world PU tasks.",
                "both fid and clip scores have limitations, for example fid is not fully aligned with perceptual quality [42], and clip is ineffective at counting [49].",
                "However, an 8 corner encoding does not take into account the physical constraints of a 3D bounding box, as the top corners of the bounding box are forced to align with those at the bottom",
                "A line segment may be aligned with two possible vanishing points"
            ]
        },
        "align\u2026to\u2026": {
            "sentences": [
                "A natural solution is to align all input set to a canonical space before feature extraction."
            ]
        }
    },
    "alleviate": {
        "alleviate on": {
            "sentences": [
                "That is to alleviate domain discrepancy on class level."
            ]
        },
        "alleviate the severity": {
            "sentences": [
                "which also alleviates the severity of the above problem."
            ]
        },
        "alleviate a heavily": {
            "sentences": [
                "InCassandrafailuredetectionisalsousedtoavoidat- Whenanewnodeisaddedintothesystem,itgetsassigned temptstocommunicatewithunreachablenodesduringvar- a token such that it can alleviate a heavily loaded node."
            ]
        }
    },
    "allocate": {
        "be allocate to": {
            "sentences": [
                "we first assume the node will reach the upper density limit du and all free slots are allocated to PA"
            ]
        },
        "allocate memory transparently": {
            "sentences": [
                "\"A unique aspect of such a service is that it not only needs to allocate memory transparently"
            ]
        },
        "allocate...to": {
            "sentences": [
                "We also compared the performance for the case when the memory allocated to each stream can vary while the total amount of memory is kept constant."
            ]
        },
        "allocate to": {
            "sentences": [
                "We also allocate dedicated CPU cores to each cluster for transaction execution",
                "a block reward is allocated to the miner of every blockchain block via a special coinbase transaction in the block.",
                "In the first experiment, we fixed the memory allocated to the approximate join to be 10% of the dataset.",
                "The amount of GPU resources allocated to each kernel is determined by the configuration to minimize the estimated execution time of individual segments."
            ]
        },
        "allocate sth. to do": {
            "sentences": [
                "As we typically allocate memory specifically to build the hash table, we can allocate pinned memory or Unified Memory for use with the Zero-Copy or Unified Memory Migration methods."
            ]
        },
        "allocate time finally": {
            "sentences": [
                "Ansors task scheduler allocates time Finally in f4, we define a function ES(gi,t) that returns an resources to tasks in an iterative manner."
            ]
        },
        "allocate memory accordingly": {
            "sentences": [
                "In the case of operators with dynamic shaped inputs, weneed to insert shape functions before the kernel invocation tocompute the output shapes and allocate memory accordinglyas detailed in Section 3.2."
            ]
        },
        "allocate memory back": {
            "sentences": [
                "Free allocated memory back to memory manager."
            ]
        },
        "be allocate for": {
            "sentences": [
                "A volume is allocated for each database instance, and consists of a list of chunks"
            ]
        }
    },
    "allow": {
        "allow to": {
            "sentences": [
                "it allows us to articulate what design decisions have made Raft more understandable or more efficient than Paxos.",
                "In this work, we design a pretraining methodology specifically for tables that resembles several core tasks in table extraction and decomposition pipelines and allows easy access to representations for different tabular substructures.",
                "The proposed scale-aware formulation allows to accurately estimate both fine details and large-scale geometry",
                "they will no longer allow to tell whether a feature is coming from the source or the target domain.",
                "Our design allows combining hardware-conscious device-specific algorithms to achieve efficient execution across all the compute units of a multi-CPU, multi-GPU server.",
                "Relative metrics allow to an application or application component",
                "A frame is associated with a time interval and allows to retrieve the status of the social network during such interval.",
                "Relative metrics allow to an application or application component.",
                "In order to improve concurrency in single-threaded VLL implementations, we allow transactions to enter a third state.",
                "It allows to restrict the amount of values scanned even further",
                "Crucially, selecting only relevant columns would allow longer examples to fit without discarding potentially relevant tokens.",
                "We uncover a meta-design principle that allow us to design variable mining difficulty protocols for three popular PoW blockchain proposals.",
                "We fo- cused on data practices most frequently mentioned in federal pri- vacy litigation and FTC enforcement actions [26], namely collec- tion of personal information, sharing of personal information with third parties, and whether websites allow users to delete data col- lected about them.",
                "metrics allows to properly control the application response",
                "Building this map allows us to estimate drift-free rotation",
                "This result implies that a good encoding selector allows us to benefit from both good compression ratios and high query performance.",
                "Thus, low utilization allows us to see better the dynamic behavior of the autoscalers by minimizing the number of extreme cases."
            ]
        },
        "allow sb to do": {
            "sentences": [
                "This ease of experimentation allowed us to find the simple upgrade of subset-sum sampling which so improved its accuracy."
            ]
        },
        "allow for": {
            "sentences": [
                "This allows for extremely fast response times, since theapproximate query execution engine can do the bulk of its processingover compact wavelet synopses, essentially postponing the (expensive)expansion rendering step into relational tuples until the end-result ofthe query.",
                "a single cache-miss costs 50-100 additional cycles and would thus allow for even more  complex models.",
                "That allows for the models to learn sufficient meaningful information before being distilled.",
                "This allows for the forecast to be reviewed by a domain expert who\nmay make adjustments when appropriate, a procedure known as human-in-the-loop.",
                "The Transformer allows for significantly more parallelization and can reach a new state of the art in translation quality after being trained for as little as twelve hours on eight P100 GPUs.",
                "Differential privacy may allow for training on data from the true distribution (actual mobile usage) rather than on proxy data from some other source that would produce inferior models",
                "Foremost is that it can potentially allow for greater concurrency than a single-version system."
            ]
        },
        "allow ... to": {
            "sentences": [
                "Our input form of point clouds allows us to achieve this goal in a much simpler way compared with [9]."
            ]
        },
        "allow for extremely": {
            "sentences": [
                "in contrast, our method explicitly trains the parameters for sensitivity on a given task distribution, allowing for extremely ef\ufb01cient adaptation for problems such as k-shot learning and rapid reinforce- ment learning in only one or a few gradient steps."
            ]
        },
        "be allow to": {
            "sentences": [
                "More specifically, our framework is allowed to contrast features across modalities within a domain or across domains using one modality."
            ]
        },
        "allow ab to do": {
            "sentences": [
                "A joint use of Fig. 5(a) and 5(b) allows us to interpret the behavior of a specific head"
            ]
        },
        "allow sb. to do": {
            "sentences": [
                "The programming model allows users to explicitly allocate global memory and shared memory in each thread block."
            ]
        },
        "allow a more": {
            "sentences": [
                "This not only avoids unnecessary runtimeoverheads, but also allows a more costly scheduling policy tofully exploit the inter- and intra- operator parallelism together."
            ]
        },
        "allow for//take into account": {
            "sentences": [
                "Existing learned indexes do not allow for an efficient build."
            ]
        },
        "allow sb to do sth": {
            "sentences": [
                "Initialising the FL approach in this way allows us to pick a fair sample that reflects the different user communities in the domain."
            ]
        },
        "allow us to": {
            "sentences": [
                "This allows us to avoid ambiguous areas and leads to reliable keypoint detections and descriptions"
            ]
        },
        "allow to do": {
            "sentences": [
                "The former policy allows a DBMS to flush the changes of uncommitted transactions at any time."
            ]
        },
        "allow usersto freely": {
            "sentences": [
                "Thesummarizationmethodshould allow usersto freely choosethe attributesandrelationships that are of interest, and then make use of these features to producesmall and informative summaries."
            ]
        },
        "allow for more": {
            "sentences": [
                "(An early Spanner incarnation supported multiple Paxos state machines per tablet, which allowed for more \ufb02exible replication con\ufb01gurations."
            ]
        }
    },
    "alternate": {
        "alternate between": {
            "sentences": [
                "The basic idea of our algorithm is a strategy that alternates between estimating the cluster identities and minimizing the loss functions."
            ]
        }
    },
    "ameliorate": {
        "to ameliorate": {
            "sentences": [
                "To ameliorate this problem,"
            ]
        }
    },
    "amend": {
        "be amend to": {
            "sentences": [
                "In this case, domain discrepancy is amended on sample level."
            ]
        }
    },
    "amortize": {
        "be amortize over": {
            "sentences": [
                "recovery work is amortized over runtime"
            ]
        },
        "amortize by considerably": {
            "sentences": [
                "Our experiments on a real world customer application show that the marginal estimation overhead of less than 1% is wellamortized by considerably faster query execution times."
            ]
        },
        "amortize the cost": {
            "sentences": [
                "Second, it amortizes the cost of system calls by condensing multiple requests into a single one, as in group commit"
            ]
        }
    },
    "analysis": {
        "analysis on": {
            "sentences": [
                "Parameters from program analysis can be obtained from program analysis on the source code. Parameters from query optimizer are determined by query optimizer."
            ]
        }
    },
    "analyze": {
        "analyze sth": {
            "sentences": [
                "We analyze the fine-tuned BERT attention maps and embeddings, and explain what information is captured by BERT.",
                "It then analyzes the log to identify which transactions must be redone and which are uncommitted."
            ]
        },
        "analyze of": {
            "sentences": [
                "We analyze the different ob- fuscation techniques in relation to protection of intellectual property and the hiding of malicious code.",
                "We analyze the different obfuscation techniques in relation to protection of intellectual property and the hiding of malicious code."
            ]
        },
        "analyze sth. to do": {
            "sentences": [
                "We analyze NVLink 2.0 to understand its performance and new functionality in the context of data management."
            ]
        }
    },
    "annotate": {
        "annotate with": {
            "sentences": [
                "Since there does not exist a large scale tabular dataset annotated with concept types, learning-based approaches need to be trained on datasets collected through distant supervision.",
                "Each paper is annotated with the genes it mentions",
                "In this model, nodes and edges are annotated with property-value pairs.",
                "3 describes a node program that executes bfs using only edges annotated with a specified edge property."
            ]
        },
        "annotate sth": {
            "sentences": [
                "In total, they annotated 3120 query-table pairs."
            ]
        },
        "annotate in": {
            "sentences": [
                "in many cases, our model will extract both of them, but normally only one of them has been annotated in the test set."
            ]
        },
        "be annotate by": {
            "sentences": [
                "Each video is annotated by at least 5 different AMT workers, thus resulting in 529 stories in total."
            ]
        },
        "annotate for": {
            "sentences": [
                "they are trained on a carefully chosen mixture of complementary large-scale multimodal data coming only from the web, without using any data annotated for machine learning purposes."
            ]
        },
        "annotate by": {
            "sentences": [
                "Crowdworkers were paid US$6 per annotated privacy policy, and each policy was annotated by ten crowdworkers."
            ]
        },
        "be annotate in": {
            "sentences": [
                "The number of classes that are annotated in standard vi- sion datasets like Pascal VOC [10] and MS-COCO [32] are very low (20 and 80 respectively) when compared to the infinite number of classes that are present in the open world"
            ]
        }
    },
    "answer": {
        "answer question": {
            "sentences": [
                "The goal of each of these tasks are different, however they share the common intention of answering questions when conditioned on visual input."
            ]
        },
        "answer at least": {
            "sentences": [
                "we also used interleaved \u201ccontrol\" trials, and only include rater data from those who correctly answered at least 80% of the control questions."
            ]
        }
    },
    "appeal": {
        "appeal to": {
            "sentences": [
                "reinforcement learning approaches have long appealed to the data management community due to their ability to learn to control dynamic behavior from raw system performance."
            ]
        }
    },
    "appear": {
        "appear to": {
            "sentences": [
                "When training a model to predict only a few missing words per sentence, this task appears to induce sophisticated language understanding",
                "While this appears to be a simple and intuitive algorithm, minor seemingly-innocuous variants turn out to be dangerously insecure.",
                "The idea is to process undisassembled portions of the text segment that appear to be code, in the expectation that they might be the targets of indirect function calls; a \u201cspeculative bit\u201d is set when this is done, and speculative disassembly of a particular region of memory is abandoned if an invalid instruction is encountered.",
                "With adequate training, generated data could look real and may appear to be even more realistic than real data at times."
            ]
        },
        "appear in": {
            "sentences": [
                "besides, the subject and object entities involved in a relation may appear in different sentences.",
                "Stateful functions that appear in SELECT clause will be evaluated last, when the output tuple is created."
            ]
        },
        "appera to": {
            "sentences": [
                "This choice appears to be entirely random."
            ]
        },
        "appear on": {
            "sentences": [
                "whereas Hyperledger confirms a block as soon as it appears on the blockchain"
            ]
        }
    },
    "appeare": {
        "appeare in": {
            "sentences": [
                "Each node in the same row represents a set of fuzzers appeared in the same year."
            ]
        },
        "appeare to": {
            "sentences": [
                "Moreover, data practices involving shar- ing with third parties appeared to be a particular source of disagree- ment among the annotation groups."
            ]
        }
    },
    "append": {
        "append to": {
            "sentences": [
                "When the NVM log buffer size reaches a threshold, its contents are asynchronously appended to an on-disk log file.",
                "only a miner which has successfully solved a computationally hard puzzle (finding the right nonce for the block header) can append to the blockchain",
                "We use special tokens to denote tones, which are appended to the end of the Romanian character sequence",
                "On decoding, LZ77 maintains an in-memory buffer for decoded content, reads the tuple sequence and appends the characters to the buffer"
            ]
        },
        "append sth to": {
            "sentences": [
                "Subsequently, the DBMS appends a record containing cp and cd to the log."
            ]
        }
    },
    "applie": {
        "applie to": {
            "sentences": [
                "PointNet [22] could then be applied to predict 3D bounding boxes"
            ]
        }
    },
    "apply": {
        "apply in": {
            "sentences": [
                "Such attentions always apply in the deep layers.",
                "Such reinforcement learning methods also apply in the path planning.",
                "The originally apply random projection in gradients of the Matrix Factorization and use another value perturbation method.",
                "graphbased approaches are now widely adopted in re because of e1e2e3e4e5e6e7e8 e1 e2 e3 e4 e5 e6 e7 e8 na rel2 rel1 figure 2: illustration of the entity-level relation matrix applied in our formulation.",
                "These techniques can be adapted and applied in OCC as well",
                "Recently one-shot learning has been successfully applied in KG completion"
            ]
        },
        "apply to": {
            "sentences": [
                "directly applying it to fined-grained sentiment analysis shows less significant improvement",
                "The parallel composition can be applied to both the per-rating and per-user privacy",
                "five Conv2D(256, 3, 1(2)) layers are applied in the second and third stages",
                "We apply all viable encodings to each column to find the encoding compressing the column with minimal size",
                "it achieves the best performance on the datasets. Existing DA / DG methods cannot be directly applied to our problem.",
                "a fusion process is applied to map the features between the image and other views.",
                "We demonstrate that SigGraph can be applied to kernel memory forensics, kernel rootkit detection, and kernel version inference.",
                "Secondly, we apply global scaling to all ground truth  boxes bi and to the whole point cloud M.",
                "Cluster level adaptation is applied to facilitate node adaptation and to adhere to any reliability policies used by IPs by adding and/or removing nodes.",
                "it becomes computationally very expensive when applied to the millions of measurements per second",
                "As kernel fusion is applied as a post-processing step, it will miss opportunities where kernel configurations are incompatible.",
                "Note that Lemma 9 applies to both centralized and decentralized settings.",
                "\u732b(cat) in this case, a CNN with width 2 is applied to the sequence of Romanized pinyin letters, followed by max-pooling to derive the final pinyin embedding",
                "For example, PartCC often adopts command logging to record transactions applied to databases, while the logging algorithm of Silo logs record values rather than operations.",
                ". When there are multiple parallelchains, one natural idea is to apply Bitcoin\u2019s difficulty adjustment algorithm to each of the chains independently.",
                "Then the environment applies the action at and returns a reward  to the agent",
                "Latent Program Algorithm applies lexical matching to find linked entities and then use pre-defined APIs",
                "This paper studies how temporal databases concepts can be applied to graph databases, in order to be able to model, store, and query temporal graphs, in other words, to keep the history of a graph database.",
                "We describe the addressed threat model, and discuss our classification method before applying it to the published techniques we survey.",
                "they cannot be applied to a large number of points",
                "In all of our downstream experiments, we apply essentially the same fine-tuning strategy to both TABBIE and TaBERT.",
                "There are two main approaches to obtain those metrics: analytic methods (usually valid for simple models), and simulation (that can be applied to more complex scenarios).",
                "Since the perturbation is applied to each ground truth box  and the associated LiDAR points independently,",
                "The term obfuscation is applied to a broad range of tech- niques, covering both syntactic and semantic transforma- tions with results of varying efficacy.",
                "Crowdsourcing is typically applied to tasks that are still difficult for computers to solve, but can be easily solved by humans [22].",
                "We then apply the CPC (Oord, Li, and Vinyals\n2018) method to mitigate the model uncertainty of the selfsupervised learning."
            ]
        },
        "apply for": {
            "sentences": [
                "I need to apply for the mortgage.",
                "which apply for tasks accentuating semantics of entities such as sentiment analysis and word sense disambiguation",
                "The same methodology can also be applied for using other OpenCL processing engine such as Voodoo"
            ]
        },
        "apply... on...": {
            "sentences": [
                "Our idea is to approximate a general function defined on a point set by applying a symmetric function on transformed elements in the set"
            ]
        },
        "apply... to...": {
            "sentences": [
                "We predict an affine transformation matrix by a mini-network (T-net in Fig2) and directly apply this transformation to the coordinates of input points."
            ]
        },
        "apply onto": {
            "sentences": [
                "As the core layer of EdgeConv, an MLP is used as the feature learning function for each edge, and channel-wise symmetric aggregation is applied onto the edge features associated with the neighbors of each point."
            ]
        },
        "apply as": {
            "sentences": [
                "Unfortunately, these simplifications cannot always be applied as joins appear in many stages of query execution."
            ]
        },
        "apply sth. to do": {
            "sentences": [
                "BlockPred applies the predicate to generate the bitmap."
            ]
        },
        "apply rule recursively": {
            "sentences": [
                "Similarly, we start from the output nodeE(i = 5) and apply rules recursively."
            ]
        },
        "apply optimizations more": {
            "sentences": [
                "Figure 5 showsmean inference speedup relative to no optimizations as Relayapplies optimizations more aggressively."
            ]
        },
        "be apply to": {
            "sentences": [
                "but also demonstrate that our algorithm can be efficiently applied beyond the regime we discussed in the theory.",
                "whatever changes Mencius makes to Phase2b should be applied to these actions as well."
            ]
        },
        "apply on": {
            "sentences": [
                "First, we apply temporal augmentation on multi- level features to construct contrastive pairs that have different motion patterns with the objective designed to distinguish the augmented samples and original ones",
                "this is a train-test mismatch, and since the diffusion model is iteratively applied on its own output throughout sampling, the sampling process produces unnatural images and sometimes even diverges.",
                "Our encoder is a ViT [16] but applied only on visible, unmasked patches",
                "a sequence of operations applied on some states"
            ]
        },
        "apply transformers to": {
            "sentences": [
                "Our work is the first to apply transformers to obtain accurate correspondences."
            ]
        },
        "apply to ana-": {
            "sentences": [
                "Our summarization methods have been applied to ana-"
            ]
        },
        "apply them atomically": {
            "sentences": [
                "Despite this  and then applying them atomically in that order."
            ]
        }
    },
    "approach": {
        "approach by": {
            "sentences": [
                "As such, we approach the task by modeling a binary verdict variable v as p(v|q,T)"
            ]
        },
        "can approach": {
            "sentences": [
                "The additional memory accesses, cache misses, CPU cycles, and critical sections invoked by lock manager operations can approach or exceed the costs of executing  the actual transaction logic."
            ]
        },
        "approach to": {
            "sentences": [
                "In addition to the above, we also introduce a novel angle loss regression approach to solve the problem of the large loss generated",
                "approach to data augmentation that makes full use of the advantages of point clouds."
            ]
        },
        "approach from": {
            "sentences": [
                "The same idea is approached from a different angle by Jain et al."
            ]
        }
    },
    "appropriate": {
        "appropriate for": {
            "sentences": [
                "it is not appropriate for conceptual or factual knowledge"
            ]
        }
    },
    "approve": {
        "approve of": {
            "sentences": [
                "we don't approve of the idea.",
                "This method does not approve of the consistence."
            ]
        }
    },
    "approximate": {
        "approximate \u2026 by...": {
            "sentences": [
                "Empirically, our basic module is very simple: we approximate h by a multi-layer perceptron network and g by a composition of a single variable function and a max pooling function."
            ]
        },
        "approximate joint distributions": {
            "sentences": [
                "AR models are feed-forward deep learning models that predict values from past values, which can be used to approximate joint distributions."
            ]
        }
    },
    "argue": {
        "argue with": {
            "sentences": [
                "It's beneath us to argue with this ugly dirty devil."
            ]
        },
        "argue about": {
            "sentences": [
                "It is beneath us to argue about the trade off between the two indexes."
            ]
        },
        "argue (that)": {
            "sentences": [
                "we argue that the fact verification under structured evidence forms is an equivalently important yet unexplored problem."
            ]
        },
        "argue to": {
            "sentences": [
                "it is important to note that we do not argue to completely replace traditional index structures with learned indexes."
            ]
        },
        "argue to completely": {
            "sentences": [
                "it is important to note that we do not argue to completely replace traditional index structures with learned indexes."
            ]
        },
        "argue for": {
            "sentences": [
                "\ufeffIn this paper, we argue for leveraging attention over the bidirectional sequential encoding of the\ninput, as discussed above."
            ]
        }
    },
    "arise": {
        "arise during": {
            "sentences": [
                "the second stage resolves conflicts that may arise during execution of transactions with imprecise timestamps."
            ]
        },
        "arise from": {
            "sentences": [
                "There are a number of implications arising from the metrics and insights we have derived from our temporal graph analysis.",
                "In addition, we also need to assign cores to partitions such that we maximize throughput for clusters, while avoiding over-assigning cores to limit synchronization overheads that can arise from too many workers",
                "As discussed in Bengio et al. (2013), complex data arise from the rich interaction of multiple sources",
                "These approaches are computationally expensive due to the long sequences that arise from concatenating text with linearized tables, which necessitates truncating the input sequences to make training feasible.",
                "Our probabilistic learning objectives are then derived using the constraints arising from the resulting image triplet",
                "we find that this is due to a train-test mismatch arising from high guidance weights."
            ]
        },
        "arise in": {
            "sentences": [
                "Many statistical and computational challenges arise in Federated Learning, due to the highly decentralized system architecture.",
                "Positive and unlabeled (PU) data naturally arises in many applications."
            ]
        }
    },
    "arrange": {
        "arrange to": {
            "sentences": [
                "I arrange to wash clothes tomorrow morning.",
                "UAVs arrange to take off after initialization."
            ]
        },
        "arrange accord to": {
            "sentences": [
                "entities in y are arranged according to their first appearance in the document."
            ]
        }
    },
    "arrayofagroupcanbeinferred": {
        "arrayofagroupcanbeinferred operation approximately": {
            "sentences": [
                "Notethattheparticipation-arrayofagroupcanbeinferred operation approximately."
            ]
        }
    },
    "arrive": {
        "arrive at": {
            "sentences": [
                "When a transaction arrives at a partition, it attempts to request locks on all records at that partition that it will access in its lifetime.",
                "We arrived at the village towards evening.",
                "Through iterative similarity calculation or probabilistic graphic models they try to arrive at the overall optimal solution.",
                "we arrived at the village towards evening.",
                "We can increase \u02dcC until we arrive at a noise level that does not adversely effect the optimization process"
            ]
        },
        "arrive after": {
            "sentences": [
                "for example, in a nave implementation,  a node program p may arrive after a transaction on shard 2, but before on shard 1."
            ]
        }
    },
    "ask": {
        "ask for": {
            "sentences": [
                "We essentially just add a final item k with wk+1 = 2k(i +1)/2, and ask for the threshold ??which is the adjusted weight of item 0 or item k, whichever is not dropped from the sample.",
                "It is usually the case when queries ask for paths that are valid con- tinuously during a certain interval."
            ]
        },
        "ask to": {
            "sentences": [
                "\"this problem asks to properly store a multiset s of time across users devices and applications.\"",
                "In case of disagreement, a third expert was asked to make the final decision"
            ]
        }
    },
    "assign": {
        "assign to": {
            "sentences": [
                "To make sure no core is assigned to a cluster with low utilization, we put all clusters having utilization that is below a predefined threshold into a pile and merge the two clusters with the lowest utilization.",
                "The intrinsic capability of EBMs to assign low energy values to in-distribution data and vice-versa motivates us to use an energy measure to characterize whether a sample is from an unknown class.",
                "Finally, we assign cores to the cluster that is proportional to their utilization, while factoring the overhead of additional worker contention.",
                "we can assign operators to CPU and GPU devices in a fine-grained manner according to their device-preference",
                "Additionally, every time a new VBlock is assigned to a client, the number of used blocks is incremented in the metadata.",
                "uring training, we assign true or false to the gold table depending on the truth of the query, and irrelevant to every other table.",
                "We also randomly assign a movie to one of the 18 parties in a random split setting",
                "the point cloud data are assigned to  volumetric grid cells via quantization, and 3D CNNs are then applied.",
                "The first relay node is assigned to check whether the value of tracking aggregation functions exceeds user-defined error thresholds."
            ]
        },
        "assign in": {
            "sentences": [
                "however, timestamps assigned in this manner are imprecise and do not give a total order between transactions."
            ]
        },
        "assign\u2026to": {
            "sentences": [
                "The proposed 3D kernel assigns the same weights to all points falling into the same grid."
            ]
        },
        "assign with": {
            "sentences": [
                "Features at each vertex are usually assigned with coordinates, laser intensities or colors, while features at each edge are usually assigned with geometric attributes between two connected points.",
                "for instance, the token pair (city\blasio) is assigned with blue tag 1.",
                "In ACC, we partition the dataset into clusters and assign each cluster with a concurrency control protocol."
            ]
        },
        "assign transactions": {
            "sentences": [
                "We design a thread-aware policy which assigns transactions thread by thread"
            ]
        },
        "be assign to": {
            "sentences": [
                "The probability being assigned to a specific bin is distributed."
            ]
        }
    },
    "assist": {
        "assist in": {
            "sentences": [
                "the derivation assists in constructing the proposed framework and also helps with the corresponding analysis."
            ]
        },
        "assist to": {
            "sentences": [
                "Image descriptions can 119 also help organize our photo collections, aid domestic robots to communicate their visual 120 observations with the users, and most importantly, assist the visually impaired to better 121 understand their surroundings."
            ]
        }
    },
    "associat": {
        "associat with": {
            "sentences": [
                "This is usually done by checking a version number associated with the data item did not change"
            ]
        }
    },
    "associate": {
        "associate with": {
            "sentences": [
                "A natural idea is to use the feature map extracted from each client as the representation since it is directly associated with the raw data.",
                "Each structure line is associated with a dominant direction",
                "Since hash tables associated with the supplier and date table are small, we can assume that they remain in the L2 cache.",
                "Thereby eliminating many of the traditional privacy risks associated with conventional recommender systems",
                "We associate the sampling state with supergroups, and samples with the groups in a supergroup.",
                "Then, they associated each bin with a learnable weighting matrix.",
                "After running the two Transformers over all rows and columns, respectively, each cell (i, j) of a table is associated with a row embedding ri,j as well as a column embedding ci,j .",
                "These known few-shot entity pairs associated with r are called references",
                "Each record is associated with a counter for contention detection and each worker runs this algorithm independently.",
                "[11] highlights the challenges of sampling associated with highly selective predicatesand applies the central limit theorem to provide probabilistic error bounds.",
                "Negative impacts of our research are difficult to predict, however, it shares many of the pitfalls associated with standard deep learning models such as susceptibility to adversarial attacks and lack of interpretablity.",
                "A frame is associated with a time interval and al- lows to retrieve the status of the social network during such interval.",
                "\ufeffHowever, the major challenge associated with retrieval of similar questions is the lexicosyntactic gap between them."
            ]
        },
        "associate to": {
            "sentences": [
                "Frame nodes may become associated to a large number of edges.",
                "each uniquely associated to the spatial coordinates of a particular non-empty voxel."
            ]
        },
        "be associate with": {
            "sentences": [
                "During the redo phase, the DBMS skips replaying the log records associated with uncommitted transactions.",
                "Co-CPC first models the dependence between a certain stock sector and relevant macroeconomic variables that are sequential and heterogeneous, e.g., macrovariables are associated with different time intervals.",
                "Each data column \ud835\udc50\ud835\udc56 is associated with a list of compression scores \ud835\udc46i"
            ]
        }
    },
    "assume": {
        "assume sth": {
            "sentences": [
                "Much of that work assumes a \u201cgreenfield\u201d approach to the problem.",
                "Prior verification systems assume that a single user verifies a short document based on a single data source."
            ]
        },
        "assume to": {
            "sentences": [
                "As the dominant directions are assumed to be perpendicular to each other",
                "Additionally, the adversary is assumed to have all required information for executing the software in question5, but not necessarily any knowl- edge of the protection method used."
            ]
        },
        "assume that": {
            "sentences": [
                "We assume that the adversarial nodes are Byzantine and they do not act rationally.",
                "Conditioned on \ud835\udc3f we assume that the partial-keys are distributed.",
                "We assume that we have a sequence encoder."
            ]
        },
        "assume that probably": {
            "sentences": [
                "while it is reasonable to assume that probably all learned indexes will \ufb01t on the gpu/tpu because of the exceptional compression ratio as shown before, it still requires 2-3 micro-seconds to invoke any operation on them."
            ]
        },
        "assume that even": {
            "sentences": [
                "In  datacenters are typically connected through high speed network  particular, Dynamo\u2019s design assumes that even where there is a  links."
            ]
        }
    },
    "assure": {
        "assure for": {
            "sentences": [
                "An ideal sampling process should assure the consistent ratios for these patterns following the category\u2019s intrinsic distribution, but it\u2019s impractical."
            ]
        }
    },
    "attach": {
        "attach to": {
            "sentences": [
                "Moreover, temporal attributes are attached to people to indicate when they were born and died.",
                "The most common step after detecting robust and repeatable interest points is to attach a fixed dimensional descriptor vector to each point for higher level semantic tasks, e.g., image matching.",
                "attach B to the parent node, update the sibling pointers, and reclaim A",
                "To realize this scenario, we attach a \u201closs prediction module\u201d to a deep network and learn the module to predict the loss of an input data point.",
                "most existing graph databases such as neo4j, titandb, orientdb employ the property graph model, which is a directed, edge-labeled multi-graph with an arbitrary number of key-value pairs attached to vertices and edges."
            ]
        }
    },
    "attack": {
        "attack on": {
            "sentences": [
                "Attack on vulnerabilities, especially on zero day vulnerabilities, can result in serious damages."
            ]
        }
    },
    "attain": {
        "attain of": {
            "sentences": [
                "Though the existing methods match the two domains in different ways, most of them ignore the class-conditional information in the alignment procedure, thus hard to attain the objective\nof discriminative learning."
            ]
        }
    },
    "attempet": {
        "attempet to": {
            "sentences": [
                "Kernel fusion attempted to hide inefficiency associated with running multiple kernels for each query like in the operator-at-a-time model."
            ]
        }
    },
    "attempt": {
        "attempt to": {
            "sentences": [
                "There are many methods that attempt to model the world from a single image.",
                "This is what we attempt to quantify in this study, based on a classifica- tion of attack tools, similarly to the approach in [4].",
                "We ask: what makes masked autoencoding different between vision and language? We attempt to answer this question from the following perspectives: (i) Until recently, architectures were different",
                "When a transaction arrives at a partition, it attempts to request locks on all records at that partition that it will access in its lifetime.",
                "A large number of researchers attempted to propose various algorithms for classification of pathology images.",
                "Thus, we attempt to go beyond pre-processing usage of decomposition and propose a\ngeneric architecture to empower the deep forecasting models with immanent capacity of progressive\ndecomposition.",
                "When the model attempts to judge whether to rely on a given table for verification, other highly-scored tables represent useful contextual information",
                "Several classical methods attempt to learn weights that identify reliable negative examples",
                "Learning algorithms attempting to tackle the transferring problem from a fully labeled source domain to an unlabeled target domain is called unsupervised domain adaptation (UDA).",
                "This work attempts to learn dynamic entity and reference representations by an adaptive attentional network",
                "Existing method attempts to do this by training the pseudo-label generation model using all the available labeled data and is completely frozen afterwards",
                "Most works focused on optimising the distributed consensus problem to speed up training and achieve similar accuracy with centralised models, while attempts to consider privacy have also been made",
                "a large number of researchers attempted to propose various algorithms for classification of pathology images.",
                "Both approaches attempt to predict the workload behaviour to optimise run time performance, however they differ in their methodology.",
                "Thus we can add a data-dependent spatial transformer network that attempts to canonicalize the data before the PointNet processes them, so as to further improve the results.",
                "As we attempt to unify the field, we will start by using 2 to present our fuzzing terminology and a unified model of fuzzing.",
                "It then attempts to learn by combining data sources in a way that produces visual classifiers that are less sensitive to the specific target data that will need to be processed.",
                "Authors attempted to automatically discover 3D lines from the output of point-based SLAM algorithms",
                "For pretraining, TaBERT attempts to recover the name and datatype of masked column headers (masked column prediction), in addition to contents of a particular cell (cell value recovery).",
                "we attempt to make a formal connection between Raft and Paxos using refinement mapping"
            ]
        },
        "attempt on": {
            "sentences": [
                "An initial attempt on end-to-end multi-modal learning is by Wang et al"
            ]
        },
        "attempt to do": {
            "sentences": [
                "Federated adaptation attempts to use the clients with similar distribution to collaboratively adapt the federated model, rather than just adapting it with the data in a single device."
            ]
        }
    },
    "attend": {
        "attend to": {
            "sentences": [
                "While Jamal et al. and Munro and Damen directly penalize domain specific features with an adversarial loss at every time stamp, Chen et al., Choi et al. and Pan et al. attend to temporal segments that contain important cues.",
                "simultaneously self-attending to all mentions for full-abstract biological relation extraction.",
                "For example, caption and section title both attend to themselves a lot in head 9 of the 1st layer.",
                "The conventional Transformer model is a bi-directional encoder, so each element can attend to all other elements in the sequence.",
                "which utilized self-attention mechanism and restricted it to attend to local neighborhoods.",
                "Some methods try to find a good tradeoff between attending to more feature points and increasing efficiency.",
                "Each position in the encoder can attend to all positions in the previous layer of the encoder.",
                "while bidirectional regions can attend to all bidirectional regions, unidirectional regions can attend to all bidirectional regions and previous unidirectional regions."
            ]
        }
    },
    "attract": {
        "attract attention": {
            "sentences": [
                "Recently, disentanglement representation learning has attracted increasing attention due to its robust performance and interpretability"
            ]
        },
        "attract from": {
            "sentences": [
                "The recent progressive development of pre-trained models has attracted much attention from researchers",
                "recently, storage and processing of such graph-structured data attracted significant interest from both industry and academia and have led to the development of many graph analytics systems and graph databases."
            ]
        }
    },
    "attracte": {
        "attracte with": {
            "sentences": [
                "For this reason, learning methods from semi-supervised learn- ing [42, 39, 33, 20] to unsupervised learning [1, 7, 58, 38] are attracting attention along with weakly-labeled or unla- beled large-scale data."
            ]
        }
    },
    "attribute": {
        "be attribute to": {
            "sentences": [
                "This difficulty can be attributed to the typically higher instance densities of pedestrians and cyclists compared with that of cars",
                "\ufeffThis can be attributed to the fact that using state-of-the-art query strategies for transformers induces a prohibitive runtime overhead, which effectively nullifies, or even outweighs the desired cost savings."
            ]
        },
        "attribute to": {
            "sentences": [
                "However, when the classes are not separable, the overlapping parts of the instance space might be attributed to the wrong class.",
                "Prior works[17],[4] have already demonstrated promising results in vari- ous OCT lesion localization, but the success should be attributed to not only recent progress in deep learning techniques but also large volumes of carefully labeled data",
                "This can be attributed to low-degree individuals who appear early or late in the dataset."
            ]
        }
    },
    "augment": {
        "be augment to": {
            "sentences": [
                "Latent factor models for recommendation have been augmented to incorporate additional content information of items and users"
            ]
        },
        "augment with": {
            "sentences": [
                "However, when these same models are augmented with PICARD, we find substantial improvements."
            ]
        },
        "augment with fine": {
            "sentences": [
                "Toachieve higher utilization, the instruction stream must beaugmented with fine-grained synchronization operations."
            ]
        }
    },
    "average": {
        "average over": {
            "sentences": [
                "we can average over the possilble produced datasets given by the random data"
            ]
        }
    },
    "avoid": {
        "\"avoid trap limitationoftheunderlyingrdmanetwork)": {
            "sentences": [
                "butthisisnot\""
            ]
        },
        "to avoid": {
            "sentences": [
                "To avoid physically impossible outcomes, we perform a collision test between any two boxes after the perturbation and revert to the original if a collision is detected."
            ]
        },
        "avoid unnecessary overhead": {
            "sentences": [
                "here we leverage ideas from [25], which already showed how to avoid unnecessary overhead from the spark-runtime."
            ]
        },
        "avoid to": {
            "sentences": [
                "To avoid materialization overhead, we use sideways information passing."
            ]
        },
        "avoid from": {
            "sentences": [
                "Our solution avoids from being brittle to scenarios where single protocols suffer"
            ]
        },
        "avoid interpretation overhead": {
            "sentences": [
                "Instead, DISC generates runtime flow at compile timeto avoid interpretation overhead, and exposes more opportunity of host-device co-optimization."
            ]
        },
        "avoid sth": {
            "sentences": [
                "With NVM, the logging algorithm can avoid this unnecessary data duplication."
            ]
        },
        "avoid go too": {
            "sentences": [
                "We want each step inSection3.1,mostoftheedgesinformationwillbecached to make the smallest move possible, to avoid going too far in the bu(cid:11)er pool."
            ]
        },
        "avoid it as": {
            "sentences": [
                "In the past, centralized control has resulted in outages and  simple design with a single master server for hosting the entire  the goal is to avoid it as much as possible."
            ]
        }
    },
    "backpropagate": {
        "backpropagate into": {
            "sentences": [
                "For both models, we backpropagate the downstream error signal into all of the model's parameters."
            ]
        }
    },
    "balance": {
        "balance by": {
            "sentences": [
                "The certainty is balanced by equally distributing the probability mass."
            ]
        },
        "balance with": {
            "sentences": [
                "First, workloads on each cluster should be balanced with flexible assignment of cores to clusters to balance computing capacity and synchronization overhead."
            ]
        },
        "balance between": {
            "sentences": [
                "Following the analysis proposed in GAN [5], the optimal distribution D will balance between the true distribution pdata and the learned distribution pgf ."
            ]
        }
    },
    "base": {
        "base on": {
            "sentences": [
                "We base our approaches on EM-optimal algorithms to minimize main memory accesses by both the CPU and GPU, and use benchmarks and experimental results to further improve performance on our hardware platform.",
                "even when based on transformer architectures, state-ofthe-art vision architectures often rely on local 2d priors to improve efficiency by making use of inductive biases not suitable for text (liu et al., 2021).",
                "based on our findings, we draw the following conclusions: the lower performance of sparql compared to sql for virtu oso suggests that storing rdf data in a single relational table with extensive indexing is not favourable for transactional graph workloads due to index maintenance and query translation costs.",
                "The principle has three components: (M1) a pivot chain, based on which blocks in all chains choose difficulty.",
                "Based on the semantic summaries, PDiff compares the target kernel with its mainstream version before and after the adoption of the patch, preferring the closer reference version to determine the patching status.",
                "Zimmeck & Bellovin [32] complement ToS;DR data with automated policy analysis based on natural language processing.",
                "Any estimator to |A ??B| produced by Bob that is based on an s-bits summary by Alice must have a variance of at least ?(kt/s).",
                "SECOND incorporates a novel form of data augmentation based on this capability.",
                "leading to encouraging results in 3D detection of pedestrians and cyclists, based on only LiDAR.",
                "We propose a unified news recommendation framework based on federated learning",
                "Specifically, it includes a partition based concurrency control (PartCC ) from H-Store, an optimistic concurrency control (OCC) based on Silo",
                "The user switches the scope of this view based on as few sites as possible, thereby minimizing the amount onstoredprocedureinvocationhandlersorqueryoperation of cross-site messaging between transaction managers.",
                "The theory based on basic material science is summarized.",
                "Specifically, we identify the bottlenecks dynamically based on a cost model and resolve them using ARIES logging",
                "Based on the refinement mapping",
                "An adaptive approach can be used, in which the fuzzy model is repeatedly updated based on online monitored information.",
                "Our experimental setup is based on the LiDAR specifi- cations of the KITTI dataset",
                "For each class, detection outcomes are evaluated based on three difficulty levels",
                "We first preallocate buffers based on the specified limit on the number of voxels",
                "Based on Rocchio classifcation, this methods builds a prototype for both the labeled and the unlabeled examples.",
                "Based on BERT",
                "The cropped areas need to be slightly adjusted based on the voxel size to ensure",
                "The learnable metric model is based on a permutation-invariant network that effectively encodes the one-hop neighbors of entities",
                "Traditional database is designed by database architects based on their experien- ces, but database architects can only explore a limited number of possible design spaces.",
                "The recovery is based on the logs and checkpoints in NVM.",
                "to extract relations among these inter-sentence entity pairs, most current studies constructed document-level graph module based on heuristics, structured attention or dependency structures [peng et al., 2017; christopoulou et al., 2019; nan et al., 2020; zeng et al., 2020; wang et al., 2020a], followed by reasoning with graph neural models.",
                "Hence, the scaling of container based on U generate a more intense workload than sysbench.",
                "This paper has proposed Fed-SPA, a new FL scheme based on sparsification-coded DP mechanism and server-side adaptive update.",
                "Below, we briefly review existing works on 3D object detection based on point cloud data and images.",
                "we crop the point cloud based on the ground-truth  distribution",
                "Based on the analysis, this work explores streamprocessing with SQL on integrated architectures",
                "points can simply be filtered and locations of objects can be easily determined based on 2D detection results"
            ]
        },
        "be base on": {
            "sentences": [
                "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks that include an encoder and a decoder",
                "Our second algorithm is based on this observation",
                "APEX is based on ALEX, a state-of-the-art updatable learned index."
            ]
        },
        "learning-based model": {
            "sentences": [
                "The learning-based models currently integrated a real system arethe light model or one model for one (sub-)query graph [14,98], which can be trained and updated quickly."
            ]
        }
    },
    "be": {
        "be in": {
            "sentences": [
                "for example, if the intrasentence relation (maryland, country, u.s.) has been identi- fied, it is implausible for u.s. to be in any other person-social relationship."
            ]
        },
        "be amoungst": {
            "sentences": [
                "Moving from large cloud infrastructure,inside databases hash-based joins and aggregations are amoungst the most expensive and used operators."
            ]
        },
        "be still": {
            "sentences": [
                "there are still some limitations in cogvideo, e.g."
            ]
        },
        "be briefly describe": {
            "sentences": [
                "The performance goals and its measuring methods are briefly described.",
                "Relationship between mass and energy in nature is briefly described."
            ]
        },
        "be explain in": {
            "sentences": [
                "The chain reaction caused by the uranium nuclei is explained in the critical experiment."
            ]
        },
        "be induce immediately": {
            "sentences": [
                "Research and development tools are induced immediately."
            ]
        },
        "be give": {
            "sentences": [
                "Basic ideas and data collected are given."
            ]
        },
        "be analyze with": {
            "sentences": [
                "The main experimental problems are analyzed with the ultra-low-frequency measurements."
            ]
        },
        "be regard as": {
            "sentences": [
                "Phenomena associated with the breakdown phase are regarded as the initial conditions."
            ]
        },
        "be derive": {
            "sentences": [
                "Lumped parameter of the concrete-structure interaction is derived",
                "The special equations are derived based on the quantum theory."
            ]
        },
        "be briefly discuss": {
            "sentences": [
                "Applications of the two-phase flow in the reactor are briefly discussed."
            ]
        },
        "be investigate numerically": {
            "sentences": [
                "A particular system is investigated numerically."
            ]
        },
        "be study": {
            "sentences": [
                "The function continuity under a certain boundary is studied."
            ]
        },
        "be involve in": {
            "sentences": [
                "\"Many elements are involved in this program such as Fe, Au, Ag, Cu and so on.\""
            ]
        },
        "be abstract": {
            "sentences": [
                "\"All the operating procedures are abstracted such as the general procedures(GP), abnormal operating procedures(AOP), emergency operating procedures(EOP).\""
            ]
        },
        "be outline for": {
            "sentences": [
                "The experimental technique is outlined for measuring these properties."
            ]
        },
        "be review in": {
            "sentences": [
                "The development of CAD is reviewed in the mechanical engineering"
            ]
        },
        "be summarize in detail": {
            "sentences": [
                "The design planning to be used is summarized in detail."
            ]
        },
        "be concentrate on": {
            "sentences": [
                "The research is concentrated on the collisions of neutrons and uranium nuclei.",
                "Attention is concentrated on the quantum theory."
            ]
        },
        "be to": {
            "sentences": [
                "The purpose of this study is to obtain the heavy concrete of much higher stability.",
                "The research is to provide healthy planners with useful information on systematic changes.",
                "The point is to extend the lead time to 24h.",
                "The purpose of this study is to show the methods mentioned above/aforesaid are promising."
            ]
        },
        "be on": {
            "sentences": [
                "Focus is on energy reserving everywhere."
            ]
        },
        "attention be be pay to": {
            "sentences": [
                "Attention is being paid to the management of environmental effluent and disposal of hazardous wastes."
            ]
        },
        "attention be pay on": {
            "sentences": [
                "Much attention is paid on the economic current density in the conductor."
            ]
        },
        "attention be give to": {
            "sentences": [
                "Particular attention is given to the start-up procedure of nuclear reactor operations."
            ]
        },
        "emphasis be pay on": {
            "sentences": [
                "The greatest emphasis is paid on the peaceful utilization of nuclear energy."
            ]
        },
        "attention be concentrate on": {
            "sentences": [
                "Attention is concentrated on the implications of attitudes to nuclear power."
            ]
        },
        "be seek for": {
            "sentences": [
                "The answer to this question is sought for by many investigators."
            ]
        },
        "be construct to": {
            "sentences": [
                "A dissolution diagram is constructed to be used in heat treatment."
            ]
        },
        "be design for": {
            "sentences": [
                "The system is designed for transferring heat from the reactor core."
            ]
        },
        "be develop": {
            "sentences": [
                "Two kinds of boiler design methods are developed."
            ]
        },
        "be establish in": {
            "sentences": [
                "A simulation model is established in a full scope simulator."
            ]
        },
        "be give for": {
            "sentences": [
                "The four-factor formula obtained in the static reactor physics is given for critical calculation."
            ]
        },
        "be provide for": {
            "sentences": [
                "An ideal source of radiation is provided for an optical communication system."
            ]
        },
        "no \u2026 be find": {
            "sentences": [
                "No phase transition evidence is found."
            ]
        },
        "be indicate": {
            "sentences": [
                "The importance of the ratio of water to uranium in reactor physical design is indicated.",
                "One kind of application to the study of muscles mechanics is indicated.",
                "The application fields of these kinds of nuclear reaction are indicated."
            ]
        },
        "be observe": {
            "sentences": [
                "The relationship between the apparent viscosity and particle coupling ration Rn is observed."
            ]
        },
        "be calculate by method of": {
            "sentences": [
                "The physical parameters of the reactor core are calculated by method of the two-group theory."
            ]
        },
        "be calculate": {
            "sentences": [
                "The activation probability per unit time is calculated."
            ]
        },
        "be determine": {
            "sentences": [
                "The condition for obtaining the minimal number of impeller revolutions is determined."
            ]
        },
        "be carefully estimate": {
            "sentences": [
                "The deviation in each method is carefully estimated."
            ]
        },
        "be estimate": {
            "sentences": [
                "The model parameters are estimated using a recursive extended least square method."
            ]
        },
        "be measure": {
            "sentences": [
                "The minimum fluidizing velocities at 1000\u2103 are measured.",
                "Neutron flux distributions in the reactor core are measured."
            ]
        },
        "be apply to": {
            "sentences": [
                "The system is applied to the analysis of a series of practical coal mine problems.",
                "The developed technique is applied to ambient air and automobile exhaust analysis."
            ]
        },
        "be use to": {
            "sentences": [
                "The method is used to measure the neutron flux distribution in the reactor core.",
                "These facilities are used to develop computer hardware for some projects."
            ]
        },
        "be assess strictly": {
            "sentences": [
                "The role of heterogeneous reactions and procedure in the troposphere are assessed strictly."
            ]
        },
        "be compare": {
            "sentences": [
                "Two portable microcomputers are compared by using 6 basic programs."
            ]
        },
        "be compare with": {
            "sentences": [
                "These data are compared with republished results."
            ]
        },
        "be evaluate": {
            "sentences": [
                "Various rate and energy parameters are evaluated."
            ]
        },
        "be assess by compare with": {
            "sentences": [
                "The used assumptions are assessed by comparing with measurement results."
            ]
        },
        "be experiment on": {
            "sentences": [
                "Animals are experimented on in the zoo laboratory."
            ]
        },
        "be process on": {
            "sentences": [
                "The selective agglomeration experiments are processed on kerosene and kerosene is at 14.8% consumption with respect to solid."
            ]
        },
        "be experiment hundreds of time": {
            "sentences": [
                "The ascend rocket are experimented hundreds of times."
            ]
        },
        "be experiment by compare with": {
            "sentences": [
                "The accuracy of 5 solutions is experimented by comparing with the exact model solution."
            ]
        },
        "be carry on": {
            "sentences": [
                "The experiment is carried on an investigated horizontal reactor."
            ]
        },
        "be design to": {
            "sentences": [
                "The test is designed to determine the flow rate of the reactor coolant pump."
            ]
        },
        "be accord to": {
            "sentences": [
                "The analytic method is according to Maxwell equations."
            ]
        },
        "be study accord to": {
            "sentences": [
                "The interactions among basic particles are studied according to the quantum mechanics."
            ]
        },
        "be use as": {
            "sentences": [
                "The diagram of the reactor vessel internals is used as a reference.",
                "Some numerical results of lower activation energies are used as and ending of this paper."
            ]
        },
        "be provide": {
            "sentences": [
                "The suggestions on the further research of the frequency spectrum analysis are provided.",
                "One kind of method calculating the effective multiplication coefficient is provided.",
                "A method to control the corrosion of high temperature section is provided."
            ]
        },
        "be establish": {
            "sentences": [
                "A modular tooling system is established."
            ]
        },
        "be recommend for": {
            "sentences": [
                "A formula calculating the critical mass in reactor is recommended for checking the two-group theory."
            ]
        },
        "be introduce": {
            "sentences": [
                "An alternative theory of generation and migration of petroleum is introduced."
            ]
        },
        "be the essential purpose of": {
            "sentences": [
                "The optimum design is the essential purpose of engineering studies."
            ]
        },
        "be achieve by": {
            "sentences": [
                "The results are achieved by more transparent way."
            ]
        },
        "be necessary": {
            "sentences": [
                "The development of some instruments is necessary when the engineer established this simulation model."
            ]
        },
        "be a relatively": {
            "sentences": [
                "There is a relatively large group of 193 lib- eral blogs with almost no connections to any other blogs, Table 4: The SNAP Results for the DBLP Datasets while such isolated blogs compose a much smaller portion (96 blogs) of all the conservative blogs."
            ]
        },
        "was, conversely": {
            "sentences": [
                "ls.o ary outcomes included the change in 6-minute hall walk distance and   However, there was, conversely, a trend toward an increase  rg Minnesota Living with Heart Failure Questionnaire score between   b in all-cause hospitalizations in the CRT-ICD group (Table 2)."
            ]
        }
    },
    "bear": {
        "bear in": {
            "sentences": [
                "by decoding, 5 triplets can be extracted: (new york, mayor, de blasio), (de blasio, born in, new york), (de blasio, born in, new york city), (de blasio, live in, new york), (de blasio, live in, new york city)."
            ]
        }
    },
    "become": {
        "will become": {
            "sentences": [
                "As the number of cores per machine continues to grow, lock managers will become even more of a performance bottleneck."
            ]
        },
        "become even more": {
            "sentences": [
                "Moreover, with the increasing gap between emerging powerful accelerators and the slow growth of network bandwidth, theembedding communication bottleneck would become even moresevere."
            ]
        },
        "become more": {
            "sentences": [
                "nowadays, the web is becoming more visual than ever before, as images and videos have become the new information carriers and have been used in many practical applications."
            ]
        },
        "become available again": {
            "sentences": [
                "It allows for multiple versions of an object to be present in  \u2022  When a node becomes available again, or a new node is  the system at the same time."
            ]
        }
    },
    "begin": {
        "begin to": {
            "sentences": [
                "We see that worker W2 finished T2 first; thus, it begins to access R1 using OPCC",
                "Aided by the rapid gains in hardware, models today can easily overfit one million images [13] and begin to demand hundreds of millions of\u2014often publicly inaccessible\u2014labeled images [16]"
            ]
        },
        "begin do": {
            "sentences": [
                "In the general version of VLL described above, once a thread begins executing a transaction, it does nothing else until the transaction is complete."
            ]
        },
        "begin with": {
            "sentences": [
                "After all workers acknowledge the switch for RS, the degrade phase begins with the coordinator messaging workers about the degrade to the new protocol.",
                "We begin with a standard statistical learning setting of empirical risk minimization (ERM).",
                "We begin this section with an overview of the runtime operations performed by a WBL-based DBMS.",
                "We begin with an analysis of the storage mechanisms for non-inline attributes in append-only storage.",
                "Because self-modifying code weakens the distinction be- tween program code and data, we begin the paper with an introduction to code and its use on modern comput- ers, followed by an overview of self-modifying code.",
                "we begin with the theory of domain adaptation."
            ]
        },
        "begin by": {
            "sentences": [
                "We begin by evaluating the effectiveness of domain confusion at learning a domain invariant representation."
            ]
        }
    },
    "behave": {
        "behave for": {
            "sentences": [
                "how this weighting behaves for different types of motion"
            ]
        },
        "behave more aggressively": {
            "sentences": [
                "Cumulative Probability for Execution Costthe optimizer behave more aggressively."
            ]
        }
    },
    "believe": {
        "believe in": {
            "sentences": [
                "I still believe in the notion of an egalitarian society."
            ]
        },
        "believe (that)": {
            "sentences": [
                "However, we believe that the proposed solutions could be applied in other languages where a pre-training corpus of text and tables is available, such as the Wikipedia datasets."
            ]
        }
    },
    "belong": {
        "belong to": {
            "sentences": [
                "Pinyin Embedding The pinyin embedding for each character is used to decouple different semantic meanings belonging to the same character form, as shown in Figure 3",
                "Many of the aforementioned methods belong to attention-based MIL.",
                "most popular byte-compression algorithms that belong to the LZ77 family work",
                "enabling multi-versioned sequential accesses to adjacency lists is to our knowledge an open topic encompassing multiple challenges: (1) different versions of edges belonging to the same adjacency list must be stored in contiguous memory locations; (2) finding the right version of an edge should not require auxiliary data structures that, in turn, require random access to be visited; (3) the concurrency control algorithm should not require random access during scans.",
                "For each record access, it first examines the cluster it belongs to and subsequently the protocol managing the cluster",
                "besides, according to the triples that eldersburg is located in maryland and maryland belongs to u.s., we can infer that eldersburg belongs to u.s.. as described above, each relation triple can provide information to other relation triples in the same text.",
                "Many of the aforementioned methods belong to barycentric algothrims.",
                "In summary, our approach belongs to domain discrepancy based methods. Similar to previous approaches, our TPN leverages additional unlabeled target data for learning task-specific classifiers.",
                "as shown in 2, (1) all frames in stage 1 and the 2nd, 4th frames in stage 2 are in the unidirectional region; (2) {frame rate}, {text} and all other frames belong to the bidirectional region.",
                "six protocols belong to the class of non-mutating optimization for Paxos",
                "We merge the line segments belonging to the same straight line which may be divided to several short parts",
                "As another kind of problem, note that in case (c) above, the shortest (or fastest) way to reach one city from another one, varies with time, since a segment belonging to the shortest path may have not existed in the past.",
                "These operators belong to some grammar, either standard like SPARQL or ad-hoc.",
                "each voxel belongs to either one organ or the background",
                "when a shard server fails, the graph data that belongs to the shard is recovered from the backing store.",
                "Given an abstract and some keywords, the goal is to determine whether they belong to the same paper",
                "We utilize ratios of clicked news belonging to each interest channel to generate their quotas",
                "Tables present information in a compact and clean manner, with cells in the same column belonging to the same concept type.",
                "To consolidate the stores, we read a chunk of the input (at a time) in the scratchpad and reorganize it in such a way that elements belonging to the same output partition are located in consecutive scratchpad entries."
            ]
        }
    },
    "benefit": {
        "benefit from": {
            "sentences": [
                "Meanwhile, BERT also benefits from pre-training on a large number of different columns, table names, and values in the generated data, which could potentially improve schema linking in semantic parsing tasks.",
                "It means that each party can benefit from the privacypreserving HFL process with our HFL-SGDMF",
                "This training stability benefits from both accurately identified examples in self-paced training and prediction consistency forced by our self-supervised distillations.",
                "We also present an efficient implementation of VoxelNet that benefits from point cloud sparsity and parallel processing on a voxel grid",
                "Cloud database services also benefit from this architecture",
                "They benefit from using entity features and structure information of triplets",
                "We aim to benefit from more advanced NLP techniques.",
                "Simultaneously, users benefit from private compute resources, avoiding interference of different workloads and organizational units one of the reasons for data marts.",
                "however, their approach does not benefit from the power of large fully-convolutional neural networks",
                "A better regularization method often benefits from longer training of deeper models",
                "The question-specific variations in Figure 8 sug- gest that some questions may benefit from the use of different ma- chine learning methods",
                "Delta-bitpacked hybrid encoding can also benefit from the sorted dataset",
                "However, domains such as image classification and language translation have benefited from shared benchmarks and test sets.",
                "benefiting from the evolving of backbone networks, cnns have largely improved the performance of various vision recognition scenarios including object detection, semantic/instance segmentation, etc.",
                "Both the real-time camera pose estimation process and the backend pose graph optimization process benefit from the global orientational estimation.",
                "To efficiently benefit from these extensions, new temporal querying operators are introduced.",
                "Benefiting from the self-attention mechanism, Transformers obtain great advantage in modeling\nlong-term dependencies for sequential data",
                "Benefiting from multiple pretraining tasks and large scale training corpora, pretrained models can capture complex syntactic word relations",
                "If these clients can be aggregated and benefit from each other, the performance will definitely outperform the local adaptation schemes.",
                "this limits the models to benefit from both image and video data."
            ]
        },
        "benefit for": {
            "sentences": [
                "Parameters are tied cross other time series,\nwhich benefits for subsequent joint distribution learning."
            ]
        },
        "benefit of": {
            "sentences": [
                "To validate the co-running benefits of the two devices",
                "Our analysis shows that the benefit of data partitioning quickly diminishes as soon as we deviate from the optimal parameters."
            ]
        },
        "benefifiting from": {
            "sentences": [
                "While benefiting from direct supervision, the lack of real image pairs often leads to poor generalization to real data."
            ]
        }
    },
    "bias": {
        "bias to": {
            "sentences": [
                "highlighting relevant paragraphs does not seem to bias annotators to ignore non-highlighted parts of the policy."
            ]
        }
    },
    "bind": {
        "bind in": {
            "sentences": [
                "APEX thus bounds S in this range"
            ]
        },
        "bind to": {
            "sentences": [
                "thereby bounding the lookup cost to tree height and significantly improving index performance.",
                "In branch co-running, the hardware resource binded to each branch is based on the characteristics of both the operator and the workload."
            ]
        },
        "be bound by": {
            "sentences": [
                "the node size could be bounded by a predefined hyper-parameter",
                "The total runtime will be bounded by either the device memory bandwidth or the cache bandwidth."
            ]
        },
        "bind on": {
            "sentences": [
                "this puts an upper bound on the throughput that can be achieved while maintaining correct ordering for updates."
            ]
        },
        "bound": {
            "sentences": [
                "The theory bounds the expected error on the target samples by three terms as follows."
            ]
        },
        "be bound": {
            "sentences": [
                "HLC is bounded and the size of the clocks remain small"
            ]
        },
        "bind with": {
            "sentences": [
                "In per-rating setting, global privacy loss can be bounded with the following theorem derived based on the parallel composition"
            ]
        },
        "be bind to": {
            "sentences": [
                "ChunkServer owns a standalone NVMe SSD disk and is bound to a dedicated CPU core,"
            ]
        },
        "be bound in": {
            "sentences": [
                "As virtual labels are bounded in the same set in this formulation"
            ]
        }
    },
    "block": {
        "will be block": {
            "sentences": [
                "Furthermore, all transactions that requested their locks after it will be blocked if their read and write set conflict."
            ]
        }
    },
    "boil": {
        "boil down": {
            "sentences": [
                "Research on SSOD boils down to answering the following two questions"
            ]
        }
    },
    "boost": {
        "boost by": {
            "sentences": [
                "Although the training complexity of NNGP is high, the exact Bayesian inferenceis boosted by highly paralleled basic linear algebra operations."
            ]
        },
        "boost sth": {
            "sentences": [
                "Nowadays, massive peripheral gadgets, such as mobile phones and wearable devices, produce an enormous volume of personal data, which boosts the process of Federated Learning (FL)."
            ]
        }
    },
    "borrow": {
        "borrow from": {
            "sentences": [
                "Here, we borrow\nideas from Bayesian Structural Time Series models",
                "Our key idea is borrowed from the digital signal processing field"
            ]
        }
    },
    "bottleneck": {
        "bottleneck by": {
            "sentences": [
                "at one extreme, one could use the timeline oracle for maintaining the global timeline for all requests, but then the throughput of the system would be bottlenecked by the throughput of the oracle."
            ]
        }
    },
    "bound": {
        "bound by": {
            "sentences": [
                "the accuracy of image-based 3D detection approaches are bounded by the accuracy of the depth estimation"
            ]
        },
        "bound consistency only": {
            "sentences": [
                "It is worth pointing out that Lemma 1 and the per-embeddingclock bounded consistency only describe the observable embeddings."
            ]
        }
    },
    "break": {
        "break down": {
            "sentences": [
                "A common approach is to break the forecast down into components which are\nindividually interpretable.",
                "Unfortunately, classical methods for both problems break down in high-dimensional settings.",
                "We break down the transfer bottleneck into three fundamental limitations for GPU-enabled data processing.",
                "But such techniques often break down when the percentage of missed and irrelevant detections is high",
                "The previous method may break down when the matrix degenerates.",
                "The initial partitioning breaks down big relations into partitions that fit in the GPU memory, while its small fan-out allows for a high throughput on the CPU side.",
                "For example, most Chinese characters can be broken down into semantically meaningful radicals"
            ]
        },
        "break in": {
            "sentences": [
                "Burglars had broken in while we were away on holiday."
            ]
        },
        "break up": {
            "sentences": [
                "She has just broken up with her boyfriend."
            ]
        },
        "break into": {
            "sentences": [
                "we simply break the long line segment into several short segments",
                "We can achieve much higher resilience and potency if the program that computes the string is broken into smaller components and embedded in the normal control flow of a program."
            ]
        },
        "break with": {
            "sentences": [
                "We break with the pre-processing convention of series decomposition and renovate\nit as a basic inner block of deep models."
            ]
        },
        "break  up": {
            "sentences": [
                "Each  database-level  transaction  is  broken  up  into  latency."
            ]
        }
    },
    "bridge": {
        "bridge gap between": {
            "sentences": [
                "To bridge the gap between the practical needs for attractive headlines and the plain HG by the current summarization systems, we propose a new task of Stylistic Headline Generation."
            ]
        }
    },
    "bring": {
        "bring to": {
            "sentences": [
                "From the model aspect, most of the studies assume that\nthe stock price is stationary but neglect the uncertainty that\nits random nature brings to the model.",
                "brings the interpretability for models to some extent",
                "APEX brings persistence to learned indexes which is a missing but a necessary feature"
            ]
        },
        "be bring by": {
            "sentences": [
                "These overheads are all brought by the imprecise predictions of learned models."
            ]
        },
        "bring about": {
            "sentences": [
                "In both stages, all album photos are displayed in fore has the potential to bring about more human- the order of the time that the photos were taken, like intelligence and understanding.",
                "A huge amount of environmental feature point matching has been brought about by the delay.",
                "For example, [30] shows that the conventional loss function such as hinge loss will bring about incorrect decision boundary, and then reveals that the non-convex loss function such as ramp loss is helpful for PU learning."
            ]
        },
        "bring in": {
            "sentences": [
                "Loop closure, however, brings in an extra computational burden and removes the drift only when revisiting the same place."
            ]
        },
        "bring out": {
            "sentences": [
                "Figure 6 nicely brings out the effect of skew in the input data streams on the performance of the algorithms."
            ]
        },
        "bring by": {
            "sentences": [
                "We first perform an ablation study to understand the performance benefits\nbrought by each component in CoST.",
                "The benefits brought by the combination of our probabilistic losses with keypoint annotations are also demonstrated for four recent networks."
            ]
        },
        "bring up": {
            "sentences": [
                "Therefore, we bring up a new perspective of solving efficiency problems:separability."
            ]
        },
        "bring down": {
            "sentences": [
                "However, gathering resource consuming information brings a heavy overhead and brings down the fuzzing efficient."
            ]
        },
        "be eventually bring into scope": {
            "sentences": [
                "If PICARD matches on tid.cid or alias.cid, then guards require that the table tid or the alias alias, respectively, is eventually brought into scope by adding it to the from clause."
            ]
        },
        "bring interpretation overhead": {
            "sentences": [
                "Meanwhile, the VM approach brings interpretation overhead."
            ]
        },
        "bring into": {
            "sentences": [
                "Modern applications often use very complex data structures and parsing on complex data structures are more likely to bring into vulnerabilities."
            ]
        }
    },
    "broadcast": {
        "broadcast to": {
            "sentences": [
                "the center machine selects a random subset of worker machines, and broadcasts the current model parameters to the worker machines."
            ]
        }
    },
    "buffer": {
        "buffer transaction validation request": {
            "sentences": [
                "In validator batching, we buffer transaction validation requests at the validator as they arrive"
            ]
        }
    },
    "build": {
        "build up": {
            "sentences": [
                "the latter one  builds up a classifier upon this bag representation for the bag-level prediction."
            ]
        },
        "build upon": {
            "sentences": [
                "This method is built upon and does not outperform STR, so we only compare our methods with STR instead of Table2Vec.",
                "QReCC builds upon three publicly available datasets.",
                "MultiPaxos builds upon single-decree Paxos to agree on a sequence of operations",
                "We built upon the architecture of the state-of-the-art ORBSLAM"
            ]
        },
        "build on": {
            "sentences": [
                "imagen builds on the power of large transformer language models in understanding text and hinges on the strength of diffusion models in high-fidelity image generation.",
                "The two-step technique builds on the assumptions of separability and smoothness.",
                "transformer models in particular, which the latest lms build on, would suffer from memory limitations with the naive addition of high-dimensional visual data into the sequence \u2013 computation scales quadratically with the sequence length (number of tokens), for example.",
                "The graph database benchmarking architecture built on the existing ldbc social network benchmark.",
                "we build on that idea",
                "Some applications propose to build on blockchain\u2019s immutability and transparency for better application workflows in which humans are the bottlenecks.",
                "RDWS are built on top of relational databse systems.",
                "On the one hand, since Ocelot is built on top of MonetDB, it can utilize optimizations like pre-fetching, data compression etc which are already implemented in MonetDB.",
                "Built on the pioneering work of Nakamoto, Bitcoin [18] is a permissionless blockchain operating on proof-of-work based on the Nakamoto protocol.",
                "Our work builds on recent advances in the training of deep networks on user-partitioned data and privacy accounting for stochastic gradient descent",
                "we have open-sourced our modified driver with this kafka in tegration so that the community can replicate and build on our work.",
                "We build on the recent insights from [49, 26, 62] which compare the importance of example replay against other more complex solutions",
                "\ufeffInteraction-based approaches build on the similarity matrix computed from word pairs between\nthe query and the document, often with countbased techniques to address data sparsity.",
                "To handle low-texture environments, we build on the VIO frameworks"
            ]
        },
        "build a highly": {
            "sentences": [
                "for example, if the goal is to build a highly-tuned system to store and query ranges of xed-length records over a set of  continuous integer keys (e.g., the keys 1 to 100m), one would not use a conventional b-tree index over  the keys since the key itself can be used as an offset, making it an o(1) rather than o(log n) operation to  look-up any key or the beginning of a range of keys."
            ]
        },
        "be build from": {
            "sentences": [
                "Each production rule ROOT\u2192 h\u03b1, \u03b2i is built from some (x, y) \u2208 D by replacing all terminal phrases with non-terminals. ti, ci, and vi stand for any table name, column name, entry value respectively"
            ]
        },
        "build a model": {
            "sentences": [
                "In [30] Yatskar et., build a more comprehensive language generation model from densely labeled images incorporating human annotations for objects, parts, attributes, actions, and scenes."
            ]
        },
        "build for": {
            "sentences": [
                "Technically, cc selector is built for conflict accumulator"
            ]
        },
        "be build on": {
            "sentences": [
                "Although Unified Memory is built on the aforementioned transfer primitives, CUDA hides the type of memory used internally."
            ]
        },
        "build the system": {
            "sentences": [
                "Although the primary contribution of STAR is this phase switching protocol, to build the system, we had to explore several novel aspects of in-memory concurrency control"
            ]
        },
        "be build in": {
            "sentences": [
                "Later work simply counts the number of consistent matches in a certain image neighbourhood [33, 38], which can be built in a scale invariant manner [30] or using a regular image grid [5]."
            ]
        },
        "build sth": {
            "sentences": [
                "Our goal is to build replicated systems that are both multicorescalable and replica-scalable."
            ]
        }
    },
    "bump": {
        "bump into": {
            "sentences": [
                "He is rushing around and bumping into everything.",
                "The neighbor of the UAV is rushing around and bumping into all the data."
            ]
        }
    },
    "bundle": {
        "bundle into": {
            "sentences": [
                "The transactions are time stamped and bundled into blocks where each block is identified by its cryptographic hash"
            ]
        }
    },
    "bypass": {
        "bypass basic heuristics": {
            "sentences": [
                "However, to bypass basic heuristics whenever possible, we distinguish two phases."
            ]
        }
    },
    "cache": {
        "cache for frequently": {
            "sentences": [
                "[13] extended this idea of near-memory-processingby adding memory-side-caching for frequently-accessed entries."
            ]
        },
        "be cache in": {
            "sentences": [
                "all of these metadata can be cached in the main memory of PolarSwitch,"
            ]
        },
        "cache sys- actually": {
            "sentences": [
                "But as Sec. 4 discusses, hit rates do not there are a number of related studies on other caching sys- actually correlate very well with most locality metrics, but tems that can shed light on the relevance of this work and rather, correlates inversely with the size of the pool (com- its methodology."
            ]
        }
    },
    "calculate": {
        "be calculate by": {
            "sentences": [
                "The position of the measure values within the multidimensional arrays can be calculated by the dimension values."
            ]
        },
        "be calculate from": {
            "sentences": [
                "Both the local accumulated vector r and the vector with momentum s are true local vectors that are directly calculated from the private data."
            ]
        },
        "calculate with": {
            "sentences": [
                "FID scores are calculated with 10,000 generated samples and 10,000 real images randomly sampled from the dataset in advance."
            ]
        }
    },
    "calibrate": {
        "calibrate with": {
            "sentences": [
                "However the need for an additional camera that is time synchronized and calibrated with the LiDAR restricts their use and makes the solution more sensitive to sensor failure modes. In this work we focus on LiDAR-only detection."
            ]
        }
    },
    "call": {
        "call out": {
            "sentences": [
                "The lawyer called out Tom's name."
            ]
        },
        "call as": {
            "sentences": [
                "We call a mention of an entity that occurs in the same row as the focus entity a supporting mention, because it increases the chance that we falsify the statement by replacing the focus entity."
            ]
        },
        "be call on": {
            "sentences": [
                ". As the API is called on the CPU, data are pushed to the GPU."
            ]
        },
        "call for": {
            "sentences": [
                "Thus, it calls for the view advisor, which automatically identifies the appropriate views for a given query workload.",
                "This calls for more informed frontier scoring than our current strategy."
            ]
        },
        "be call to": {
            "sentences": [
                "After the volume space grows, the pfs mount growfs should be called to recognize newly allocated blocks and mark them available."
            ]
        }
    },
    "calm": {
        "calm down": {
            "sentences": [
                "You'd better calm down."
            ]
        }
    },
    "can be": {
        "can also be achieve in": {
            "sentences": [
                "The normal operating pressure value can also be achieved in summer."
            ]
        },
        "can be produce in": {
            "sentences": [
                "Highly enriched uranium can be produced in China."
            ]
        },
        "can be determine": {
            "sentences": [
                "The earth-moon distance can be determined to an accuracy of \u00b115cm by this technique."
            ]
        },
        "can be adopt accord to": {
            "sentences": [
                "These remedial can be adopted according to the probabilistic safety analysis (PSA)."
            ]
        }
    },
    "cancel": {
        "cancel out": {
            "sentences": [
                "So gradient clipping limits the effects of gradients with large magnitudes in aggregation, the useful updates may be canceled out"
            ]
        }
    },
    "capitalize": {
        "capitalize on": {
            "sentences": [
                "These tech- niques capitalize on the strength of humans and of machine classifiers to solve difficult tasks"
            ]
        }
    },
    "capture": {
        "capture by": {
            "sentences": [
                "capture structure information by knowledge representation task",
                "for each entity ei in the matrix, their relevance is captured by a d-dimensional feature vector f(es, eo)."
            ]
        },
        "capture correlations": {
            "sentences": [
                "Sampling is an ad-hoc approach that captures correlations among arbitrary numbers and types of predicates."
            ]
        },
        "capture sth": {
            "sentences": [
                "First, how to efficiently capture the index requirements"
            ]
        },
        "be capture by": {
            "sentences": [
                "their relationships to Paxos cannot be captured by refinement mapping."
            ]
        }
    },
    "care": {
        "care about": {
            "sentences": [
                "Therefore, in order to derive more 149 human-centric descriptions, we need to know what people care the most about in images.",
                "Actually, databases such as MySQL and AliSQL do not care about the I/O sequences of underlying storage",
                "However, as F1 is counted as one of the metrics suitable for imbalanced learning, a good AUC does not necessarily mean a good F1, as it cares about the relative order of real outputs, instead of the classification result."
            ]
        }
    },
    "carry": {
        "carry out": {
            "sentences": [
                "The neighbourhood consensus has been typically carried out on sparsely detected local invariant features as a filtering step performed after a hard assignment of features by nearest neighbour matching using the Euclidean distance in the feature space.",
                "In this section, we carry out a thorough ablation study, on the key components introduced during the self-paced learning stage (i.e., the selection of trusted set) and the self-supervised distillation stage (i.e., the diversity of students, the effect of hard sample mining when training students, and the effect of weight averaging further by teachers).",
                "All experiments were carried out with an Intel Core i7-4790",
                "The optimization of the PLM encoder is carried out more carefully with layer-wise learning rate decay coefficient 0.8.",
                "this type of models measure the plausibility of a fact as the distance between the two entities after a translation carried out by the relation",
                "Users of data warehouses typically carry out online analytical porcessing.",
                "In short, dynamic analysis has the shortcomings of slow speed, low efficiency, high requirements on the technical level of testers, poor scalability, and is difficult to carry out large-scale testing.",
                "collect the number of keys No overflowed to SA without actually carrying out any inserts",
                "The authority failed to carry out its statuory duties."
            ]
        },
        "carry on": {
            "sentences": [
                "we should carry on with our research.",
                "Each UAV carries on the localtion algothrim distributedly."
            ]
        },
        "carry out entirely": {
            "sentences": [
                "The key idea in the decomposition algorithms of [284] is to break the standard HWT computationinto hyperplanes (in the chosen dimension order), such that the complete decomposition for each hyperplane can be carried out entirely in-memory."
            ]
        },
        "be carry out": {
            "sentences": [
                "In subsequent rounds, ActvAGG returns a new partition of the clients, where the clustering is carried out over the set of user embeddings.",
                "Our evaluations are carried out using three benchmarks"
            ]
        },
        "carry in": {
            "sentences": [
                "Modifications to the same chunk carried in a single I/O request are atomic."
            ]
        }
    },
    "carve": {
        "carve out": {
            "sentences": [
                "It jointly models natural language question and structured database schema information, and uses some pre-defined relationships to carve out the interaction between them."
            ]
        }
    },
    "cast": {
        "cast xxx into": {
            "sentences": [
                "While previous works also cast SKG tasks into the textto-text format (Hosseini-Asl et al., 2020; Shaw et al., 2021; Liu et al., 2021), their independent choices of pretrained language models (PLMs), input-output formats, and frameworks make our unification non-trivial"
            ]
        },
        "cast as": {
            "sentences": [
                "in principle, the same can be done for many image and video understanding tasks such as classification, captioning, or question-answering: these can be cast as text prediction problems with visual input conditioning.",
                "We cast contrastive learning as training a non-parametric classifier by assigning a unique virtual class to each data in a batch"
            ]
        },
        "cast into": {
            "sentences": [
                "supervised learning, a set of target (class) labels is provided for a set of samples (images) using which we train a model (CNN [26, 18, 53] or Transformer [66, 40]) that casts these samples into some representation space from which predictions are made, e.g. using a linear classifier"
            ]
        }
    },
    "catch": {
        "catch up": {
            "sentences": [
                "the receiver catches up if it is missing entries",
                "He found it is hard to catch up the training speed.",
                "We found it is hard to catch up the change of IMU.",
                "He found it's hard to catch up her",
                "To catch up with the continuously improved quality of generated data, we thus take them as unlabeled data, which consist of low-quality data and high-quality data.",
                "We found it is hard to catch up the training speed."
            ]
        }
    },
    "categorize": {
        "categorize as": {
            "sentences": [
                "They can also be categorized as local features and global features."
            ]
        },
        "be categorize into": {
            "sentences": [
                "It can be further categorized into covariate shift, target/label shift, etc."
            ]
        },
        "categorize into": {
            "sentences": [
                "The survey further categorizes HTAP solutions using \u201cSeparate OLTP and OLAP Systems\u201d into two sub-categories.",
                "For this paper, we have categorized fuzzers into three groups based on the granularity of semantics a fuzzer observes in each fuzz run.",
                "We first categorize them into three groups according to the type of injected knowledge",
                "most existing models in handling entitypairoverlap (epo) and singleentiyoverlap (seo) cases can be categorized into two classes: decoder-based and decomposition-based.",
                "The information on the Web can be roughly categorized into four types, namely, unstructured text, semi-structured data, Web tables and semantic annotations.",
                "Our method can be categorized into the uncertainty ap- proach but differs in that it predicts \u201closs\u201d based on the in- put contents"
            ]
        },
        "be broadly categorize into": {
            "sentences": [
                "Strongly consistent replication is implemented in databases using various schemes, which can be broadly categorized into two groups"
            ]
        }
    },
    "category": {
        "category into": {
            "sentences": [
                "The life cycle management of workloads can be categorised into two overlapping phases."
            ]
        }
    },
    "cause": {
        "cause by": {
            "sentences": [
                "ALEX even has unbounded \u201clast mile\u201c search cost in the leaf nodes since it does not provide any threshold of errors caused by the wrong predictions",
                "To address the challenges caused by diverse graph topology of data, the SGCLL layer in AGCN [101] utilizes a learnable distance metric to parameterize the similarity between two vertices on the graph.",
                "Sometimes it caused by occasional and temporary faults,",
                "to overcome the large memory and time overhead caused by the large model and long sequence, we refer to swin attention [14] and extend it to autoregressive video generation in \u00a7 3.3."
            ]
        },
        "cause ... to": {
            "sentences": [
                "Contention causes OCC\u2019s performance to degrade, however, and recent concurrency control designs, such as hybrid OCC/locking systems and variations on multiversion concurrency control (MVCC), have claimed to outperform the best OCC systems."
            ]
        },
        "cause in": {
            "sentences": [
                "This causes an increase in hash rate and consequently a series of blocks with short solve times."
            ]
        }
    },
    "caution": {
        "caution against": {
            "sentences": [
                "we strongly caution against the use text-to-image generation methods for any user-facing tools without close care and attention to the contents of the training dataset."
            ]
        }
    },
    "center": {
        "center on": {
            "sentences": [
                "lift centers on utilizing human expertise to learn from demonstrations thus lowering online training times."
            ]
        }
    },
    "change": {
        "change only slowly": {
            "sentences": [
                "\"Shared- nothing distributed databases [35\u201337] do not require shared memory"
            ]
        },
        "change in": {
            "sentences": [
                "Architectural changes in modern processors have inspired a significant amount of research on finding the optimal join implementation."
            ]
        },
        "change to": {
            "sentences": [
                "When the workload state changes to the state in Figure 8",
                "Fast Paxos changes the quorum size of Paxos to include a super-majority"
            ]
        },
        "change into": {
            "sentences": [
                "Compared with the non-private FMF, a few additional steps are required in [Stage 2] (s2) to change it into a differentially private mechanism, called VFL-SGDMF"
            ]
        }
    },
    "characterize": {
        "characterize by": {
            "sentences": [
                "we study the long-term forecasting problem of time series,\ncharacterizing itself by the large length of predicted time series."
            ]
        }
    },
    "charge": {
        "in charge of": {
            "sentences": [
                "The server is in charge of maintaining parameters of the recall model"
            ]
        }
    },
    "check": {
        "check for": {
            "sentences": [
                "we check for matches of their constituent planes to the map planes"
            ]
        }
    },
    "choose": {
        "choose between": {
            "sentences": [
                "we use cost models to choose between node expansion and split"
            ]
        },
        "choose to": {
            "sentences": [
                "we choose to port two optimizations as case studies",
                "We choose to use VINS-mono at all stages as the default setting.",
                "an operator may choose to kill an antagonist task.",
                "an operator may choose to kill an antagonist task",
                "As an example, when forecasting the next 24 steps of an hourly time series, the forecaster could\nchoose to set the AR order to 100 with some regularization.",
                "we choose to use VSA at all stages as the default setting",
                "For instance, to measure the temporal proximity between A and D, we may choose to calculate."
            ]
        },
        "choose as": {
            "sentences": [
                "We  choose  deadline  as  the  performance  requirement."
            ]
        },
        "choose from": {
            "sentences": [
                "It relies on DBAs to choose some frequent queries from the query log as the representative work- load, and uses the workload to select indexes.",
                "Each of the k dropped tuples may be chosen from either relation??",
                "Each record can be choosed from random partitions"
            ]
        },
        "choose the best": {
            "sentences": [
                "Humans can quite conveniently detect semantic mistakes in the sentences, or 803 choose the best suited caption for an image."
            ]
        },
        "choose on": {
            "sentences": [
                "We choose the MSE loss function on prediction w.r.t the target sequences, and the loss is propagated back from the decoder\u2019s outputs across the entire model."
            ]
        },
        "choose randomly": {
            "sentences": [
                "during training, aug_level is chosen randomly, while during inference, we sweep over its different values to find the best sample quality."
            ]
        },
        "choose key carefully": {
            "sentences": [
                "Supporting directories allows applications to control the locality of their data by choosing keys carefully."
            ]
        }
    },
    "cite": {
        "cite for": {
            "sentences": [
                "This assumption is simply cited for completeness, and is ignored for the remainder of this survey."
            ]
        }
    },
    "claim": {
        "claim to": {
            "sentences": [
                "HippogriffDB claims to achieve 100\u00d7 speedup over MonetDB when the ratio of memory bandwidth of GPU to CPU is roughly 5\u00d7."
            ]
        }
    },
    "classfy": {
        "classfy into": {
            "sentences": [
                "existing works can be classified into two major categories",
                "They can be classified into two categories"
            ]
        }
    },
    "classify": {
        "classify into": {
            "sentences": [
                "We further classify them into entity features fused and knowledge graph supervised pre-trained models",
                "The aggregation is used to \"collapse\" away some dimentions to obtain a more concise data set, namely, to classify items into groups and determine one value per group.",
                "Other approaches learn to filter matches by classifying them into inliers and outliers [33, 44, 7, 71]."
            ]
        },
        "be classify into": {
            "sentences": [
                "Based on the complexity of the SQL, the examples in SPIDER are classified into four types: Easy, Medium, Hard, Extra Hard."
            ]
        },
        "classify in": {
            "sentences": [
                "compared with sentence-level re that only contains one entity pair to classify in a sentence, document-level re requires the model equal contribution and shared co-first authorship."
            ]
        },
        "be classify as": {
            "sentences": [
                "According to the strategies of exploring the programs, fuzzers could be classified as directed fuzzing and coverage-based fuzzing."
            ]
        }
    },
    "click": {
        "click on": {
            "sentences": [
                "For example, whether someone clicks on a sponsored search ad is influenced by the position in which it is placed."
            ]
        }
    },
    "close": {
        "close to": {
            "sentences": [
                "bringing learned indexing another step closer to practical adoption"
            ]
        },
        "close up": {
            "sentences": [
                "the results also prove that it is of benefit to close up the gap between training and inference."
            ]
        }
    },
    "cluster": {
        "cluster into": {
            "sentences": [
                "Therefore, we clustered each dataset into five groups (based on the attribute values) and generated for each group a random distribution between propensity attribute values {0, 1}.",
                "An efficient mean shift clustering algorithm is employed to cluster the masked pixel embeddings into plane instances"
            ]
        }
    },
    "code": {
        "code which first": {
            "sentences": [
                "For example, if one needsfn main() -> Tensor<10> {let t1, t2 : Tensor<10> = ...;add(t1, t2)}234567fn main() -> Tensor<10> {let t1 = ...; let t2 = ...;let buf1 = alloc_storage(40,64,cpu);let out1 = alloc_tensor(buf1,0,(10),f32);invoke_mut(add, (t1, t2), (out1));out1}The above transformation replaces the operator invocationadd to code which first allocates an output tensor frombacking storage at offset zero, and call to invoke_mut."
            ]
        }
    },
    "coincide": {
        "coincide at": {
            "sentences": [
                "predictions for two different keys might coincide at one position"
            ]
        },
        "coincide with": {
            "sentences": [
                "Differently, we require that the real labels of positive and unlabeled data should be specified in advance and will not change during the entire online learning process, so that our model coincides with the standard PU learning requirements"
            ]
        }
    },
    "collaborate": {
        "collaborate to": {
            "sentences": [
                "Multi-robot autonomous systems with the ability of communication and perception can collaborate to accomplish missions"
            ]
        },
        "collaborate with": {
            "sentences": [
                "Usually the checkpointing techniques need to collaborate with the logging techniques and complement each other to actualize a reliable recovery process"
            ]
        },
        "collaborate with only": {
            "sentences": [
                "The group collaborating with only the P researchers relationships."
            ]
        }
    },
    "collapse": {
        "collapse on": {
            "sentences": [
                "When these implementation choices are made sensibly, OCC performance does not collapse on high-contention TPC-C."
            ]
        },
        "collapse into": {
            "sentences": [
                "the classes with fewer training instances start to collapse into each other"
            ]
        }
    },
    "collect": {
        "collect from": {
            "sentences": [
                "Statistical information is collected periodically from all workers",
                "The user utterance can be either one of the original human utterances collected from the aggregated datasets or the canonical sentences sampled from the SCFG.",
                "First a number of items are collected from the original data stream according to a certain criteria, and perhaps with aggregation in the case of duplicates.",
                "collect data from the Bitcoin network between May 10th and May 18th, 2019.",
                "The server iteratively updates a global model for E epochs based on gradients of the objective function, which are collected from batches of m clients\u2019 d-dimensional local updates.",
                "while initial progress has been made towards a similar capability in computer vision, the most used paradigm still consists of first pretraining on a large amount of supervised multimodal data collected from the web, before fine-tuning the model on the task of interest (lu et al., 2019; wang et al., 2021; zellers et al., 2022)."
            ]
        },
        "collect on": {
            "sentences": [
                "In many cases, it is expensive or difficult to collect annotations on the target domain."
            ]
        },
        "collect to": {
            "sentences": [
                "Collecting the data of all possible domains to train ML models is expensive and even prohibitively impossible."
            ]
        },
        "collect in": {
            "sentences": [
                "unlabeled data collected in the new environment as well as a few human labeled examples from a subset of the categories of interest.",
                "All policies were collected in December 2013 or January 2014."
            ]
        }
    },
    "collecte": {
        "be collect by": {
            "sentences": [
                "This 926 dataset was collected by initially downloading 760 million images by querying Flickr 927 forasetof693nouns."
            ]
        },
        "collect": {
            "sentences": [
                "We collected the first large scale dataset for end-to-end, open-domain conver sational QA that contains question rewrites that incorporate conversational context."
            ]
        }
    },
    "combine": {
        "combine as": {
            "sentences": [
                "In order to get answers, NSM learns to generate a sequence of actions that can be combined as a executable program"
            ]
        },
        "combine to": {
            "sentences": [
                "For a Chinese character, the glyph embedding, the pinyin embedding and the character embedding are combined to form a fusion embedding, which models the distinctive semantic property of that character"
            ]
        },
        "combine data locally": {
            "sentences": [
                "\"Thisisbecause VM instances o\ufb00er ample opportunities to share"
            ]
        },
        "combine with": {
            "sentences": [
                "Harder and Harrison present a preliminary analytical model that combines traditional locks with Oracle Parallel Server locks, which are parallel cache management (PCM) locks held by instances of the parallel server.",
                "Probabilistic-VO [7] combines points together with lines and planes for pose estimation while modeling their uncertainties.",
                "The robustness discussed below is all a consequence of the ipps of goal(i) combined with the non-positive covariances of goal (iii).",
                "We combine four different news ranking models with our proposed privacy-preserving news recall model in Uni-FedRec",
                "We combine it with the feature learning network and convolutional middle layers to form an end-to-end trainable pipeline.",
                "Thus we lastly combine SuperPoint with a descriptor subnetwork (see Figure 2c).",
                "combined with the other two sequences, the decoded triplet should be (de blasio, born in, new york).",
                "The human pose estimation is a typical regression task, and the object detection is a more complex problem combined with both regression and classification.",
                "In this work, we combine the robustness of neighbourhood consensus filtering with the power of trainable neural architectures.",
                "Application topology combined with architectural characteristics",
                "To make TDS self-adapted to data and task, and to combine it with model training.",
                "Hybrid approaches utilise proactive approaches and combine these with reactive approaches, as way to engage adaptation for long and short term time scales.",
                "However virtual machine based methods such as Cryp- texec, CSPIM and Shiva [51, 54, 15], demonstrate that singleton encoding methods can be combined with tempo- rary exposure, by interpreting arbitrary (Turing-complete) programs, instruction by instruction.",
                "For any i, j, pi,j ??pipj combined with Horvitz-Thompson estimators implies nonnegative covariance between ?wi and ?wj.",
                "by re- quiring a recurrent model (santoro et al., 2016) or a siamese network (koch et al., 2015)), and it can be readily combined with fully connected, convolutional, or recurrent neural networks.",
                "We propose to leverage the weight sharing technique in multi-task learning and combine it with IFCA.",
                "This flexibility that combining with other models provides the proposed method a chance to get further improvement on existing excellent models.",
                "We also combine planes with point based SLAM to improve robustness.",
                "To combat this challenge, [38] proposed combining RL with a Simplex method to reduce the search space to a smaller valuable set, and then used online CPU and memory utilisation to guide decision making.",
                "An overall ratio combines the user and elasticity ratios with the geometric mean.",
                "Although entire software-based methods cannot afford perfect protection, we anticipate that the above mentioned techniques significantly raise the difficulty for reverse-engineering as well as for signature-based detection and code detection through pattern matching, especially when these techniques will be combined with the existing code obfuscation techniques.",
                "We combine this retriever with a RoBERTa-based joint reranking-and-verification model"
            ]
        },
        "combine sth. into": {
            "sentences": [
                "Li et.al. combined item con tent information and user-item feedback into a variational autoencoder to learn user preference towards items."
            ]
        },
        "be combine with": {
            "sentences": [
                "This auxiliary task is combined with the main text-to-SQL task in a multitasking way."
            ]
        },
        "combine xxx into": {
            "sentences": [
                "As for the node-centric graph, we combine both local and non-local edge features into the computation."
            ]
        },
        "combine these coarse": {
            "sentences": [
                "To combine these coarse-grained models assuming column independence."
            ]
        },
        "combine into": {
            "sentences": [
                "Our solution combines these threes algorithms into a systhesized one"
            ]
        },
        "be combine into": {
            "sentences": [
                "The device groups global memory loads and stores from threads in a single warp such that multiple loads/stores to the same cache line are combined into a single request."
            ]
        },
        "combine command log": {
            "sentences": [
                "Hence, the intuition of our adaptive logging approach is to combine command logging and ARIES logging"
            ]
        },
        "combine sth with sth": {
            "sentences": [
                "Replicated, in-memory transactional storage systems have emerged as an important piece of infrastructure for datacenter applications because they combine fault-tolerance and strong semantics with high throughput and low latency."
            ]
        }
    },
    "come": {
        "come with": {
            "sentences": [
                "While combining multiple protocols into a single system can potentially allow more concurrency to improve the overall throughput, it comes with the cost of higher concurrency control overhead",
                "Personality Captions16 (Shuster et al., 2019) is a large scale image caption dataset that comes with so-called personality traits that are useful for controllable and style-based image captioning.",
                "Blockbench is a flexible and extensible framework that provides a number of workloads, and comes with Ethereum",
                "We now show that the proposed estimator comes with the following guarantee.",
                "However, when it comes with the proper resource provide a higher level of abstraction for the process lifecycle allocation needed to satisfy service level objectives",
                "they also come with higher overhead for training and inference."
            ]
        },
        "come to": {
            "sentences": [
                "This approach introduces an increased risk when it comes to user privacy",
                "Unlike single key search data structures, spatial data structures are designed in a way to achieve high performance results when it comes to multidimensional range queries such as searching for all cities within a given distance of a specific point.",
                "When compared to the above transformations, this obfuscation technique is the most resilient, even when it comes to reverse engineering.",
                "When it comes to deep learning based models."
            ]
        },
        "come about": {
            "sentences": [
                "A number of SLAM papers have come about as a result of the report."
            ]
        },
        "come from": {
            "sentences": [
                "This dataset comes from the the CSQA dataset5 (Saha et al., 2018), originally cre- ated for a setting similar to conversational KBQA.",
                "Half of the intratable swaps must come from the same row or column to make the objective more challenging.",
                "The 2048-dim feature vector which comes from the last residual block in the RoI Head is used for contrastive clustering",
                "It used to be the case that most of the data in a data warehouse came from sources within the organization.",
                "These traces come from real-world applications",
                "For example, weekly variables (\u03c4 = 7) share the same parameters and the last time step state h\u03c4 comes from 7 days ago.",
                "The plane\u2019s bounding polygon for projection comes from the pop-up process.",
                "During inference, we will not filter any retrieved information, as all the retrieve data only come from training set.",
                "the term ontology comes from philosophy"
            ]
        },
        "come  back": {
            "sentences": [
                "we  will  come  back  to  this  property  in  section  4.4,  when  we  discuss  the  expressive   power of model-agnostic meta-learning algorithms."
            ]
        },
        "come into": {
            "sentences": [
                "multimodal learning has come into prominence recently, with text-to-image synthesis [53, 12, 57] and image-text contrastive learning [49, 31, 74] at the forefront."
            ]
        },
        "come only": {
            "sentences": [
                "they are trained on a carefully chosen mixture of complementary large-scale multimodal data coming only from the web, without using any data annotated for machine learning purposes."
            ]
        },
        "come up with": {
            "sentences": [
                "To solve this problem, they come up with a spatio- temporal model based on graph databases."
            ]
        }
    },
    "comment": {
        "comment on": {
            "sentences": [
                "More and more people like to comment on the models.",
                "more people like to comment on the dishes."
            ]
        }
    },
    "commit": {
        "commit (there": {
            "sentences": [
                "Let e(R ) be the executionof R that committed (there i i is exactly one such execution)."
            ]
        },
        "commit on": {
            "sentences": [
                "a transaction that has committed on the backing store before failure requires no extra handling: the backup server will read the latest copy of the data from the backing store."
            ]
        },
        "epoch-based group commit": {
            "sentences": [
                "The latency that phase switching incurs is no larger than the typical delay used in high-performance transaction processing systems for an epoch-based group commit"
            ]
        },
        "two-phase commit": {
            "sentences": [
                "However,these systems suffer when transactions need to access more than one partition, especially in a distributed setting where expensive protocols like two-phase commit are required to coordinate these cross-partition transactions"
            ]
        },
        "commit to": {
            "sentences": [
                "it commits skip entries to keep log execution progressing forward",
                "A commitment scheme allows one to commit to a message ahead of time"
            ]
        }
    },
    "communicate": {
        "communicate with": {
            "sentences": [
                "They never communicate with each other during the training process",
                "IoTFuzzer targets IoT devices by letting the driver communicate with the corresponding smartphone application.",
                "While modern industrial machines can directly communicate with the interface board",
                "as described in section 3.1, the gremlin server is a layer on top of a tinkerpop3 implementation and enables multiple clients to communicate with the same data base in a platform-agnostic manner.",
                "We communicated with several seasoned medical doctors practicing in AD fields, and they unanimously agreed that AD diagnosis should be described as a PU learning problem rather than (the traditional treatment as) a binary classification problem.",
                "a clear drawback of cnns is that it may take a number of layers for distant features to communicate with each other, hence limiting the ability of visual representation.",
                "The center machine and worker machines can communicate with each other using some predefined communication protocol.",
                "The application communicates directly with a running Neo4j server through the bolt protocol, and automatically populates the database by executing the corresponding Cypher queries."
            ]
        }
    },
    "compare": {
        "compare with": {
            "sentences": [
                "How does our proposed DICE framework perform compared with state-of-the-art causal recommendation methods un der non-IID circumstances?",
                "we compare FineStream on integrated architectures with Saber on discrete CPU-GPU architectures",
                "Compared with a single\nstock, the impact of macro-factors on a certain stock sector is more stable and easy to capture.",
                "In this method, for each algorithm we pairwise compare the value of each metric with the value of the same met\u0002ric of all the other  autoscalers.",
                "We now illustrate the proposed framework in Algorithm 1 and compare it with theflat solutions and the compressed solution.",
                "We compared our results with D2A (Dialog-to-Action) [12], the state-of-the-art end-to-end neural model for conversational KG-QA.",
                "We compare with UbT [11] and SoftT [18], two leading methods for SSOD, as demonstrated by their superior performance on the MS-COCO benchmark, see Table 1",
                "Given an unseen data column at test time, we compare 1, 000 randomly selected column values to each entry of the dictionary, then classify the column as the most frequently matched type.",
                "We then compare with the traditional centralized method.",
                "The idea of using a model in kernel fuzzing originated in Koopman et al.'s seminal work [128] where they compared the robustness of OSes with a finite set of manually chosen test cases for system calls.",
                "At last, we provide data from a real-world implementation and compare them with several high-value proof-of-work projects.",
                "experimental results on three benchmark datasets show that our model docunet can achieve state-of-the-art performance compared with baselines.",
                "we also compare  VoxelNet with HC-baseline on 3D Pedestrian and Cyclist detection.",
                "We then compare our model with Transformer backbones.",
                "To better demonstrate the effectiveness, our SADGA is evaluated on the development set and test set compared with RATSQL according to the parsing difficulty level defined by Yu et al. [36].",
                "On a number of benchmark datasets ranging from shape classification, part segmentation to scene segmentation, we experimentally compare our PointNet with state-ofthe-art approaches based upon multi-view and volumetric representations.",
                "Few other teas can compare with ours either for flavour or in color.",
                "Compared with the case for image perception, training a relational inference system that learns the underlying reasoning process of human beings could be a better choice for learning new relations",
                "we compare the proposed method with several top-performing algorithms",
                "Our experiments show CormCC has significantly higher throughput compared with single protocols and state-of-the-art mixed concurrency control approaches",
                "Compared with training only locally with no communication, our experiments show that our VFL-SGDMF algorithm can provide high accuracy in predictions",
                "Temporal queries based on Allen\u2019s temporal intervals are then used to validate our model and compare with the traditional way of storing temporal intervals.",
                "The sampling process guarantees a bounded memory usage compared with other methods shown in Table 1."
            ]
        },
        "compare in": {
            "sentences": [
                "Nevertheless, MINERVA outperforms DeepPath on their benchmark NELL-995 dataset when compared in their experimental setting"
            ]
        },
        "compare to": {
            "sentences": [
                "The performance of VPU is compared to that of some recently developed PU learning methods.",
                "The assembly code represents the data dependencies and comfort to analyse the data after disassembling the executable as compared to the decompiled code.",
                "Compared to a baseline network based on dense input grids, OctNet requires much less memory and runtime for high-resolution point clouds.",
                "Such assumptions often drive the development of testing tools to have different characteristics compared to those of fuzzers.",
                "Compared to the strongest inductive baseline, GraIL is able to substantially outperform state-of-the-art models",
                "Compared to image based detection, LiDAR provides reliable depth  that can be used to accurately localize objects and characterize their shapes",
                "When compared to the above transformations, this obfuscation technique is the most resilient, even when it comes to reverse engineering.",
                "Compared to the stream processing engine on the discrete architecture, FineStream on the integrated architecture achieves 10.4x price-throughput ratio, 1.8x energy efficiency, and can enjoy lower latency benefits.",
                "Compared to traditional, hand-designed heuristics, our technique learns priors over geometric transformations and regularities of the 3D world through end-to-end training from image pairs",
                "Byte-addressable persistent memory (PM) offers persistence, high capacity, and lower cost compared to DRAM",
                "These methods have been shown to perform poorly for both BEV detection and 3D detection compared  to other methods.",
                "In Table 2, we compare our best-performing models to the closed-setting system from Chen et al.(2020b), as well as to several recent models from the literature",
                "We compared our algorithms to the case where each robot performs independent MSCKF using a single camera and IMU.",
                "This improvement is 13 % as compared to pure-pessimistic approach and is more than 31 % as compared to pure-optimistic approach",
                "None of those two methods show how this library compares to a more traditional way of temporal data representation.",
                "However, in prior work, when a new policy is proposed, it is seldom compared to the state-of-the-art, and is often compared only to static provisioning using a predefined QoS target.",
                "the  convolutional  model  learned  by  maml com- pares well to the state-of-the-art results on this task, narrowly outperforming the prior methods.",
                "which are inexpensive compared to keypoint annotations.",
                "Towards this end, we devote this work to analysing three unique linguistic characteristics of Chinese (writing system) compared to English: 1) The Chinese writing system is morphemic (Hill, 2016), which means the Chinese characters poorly reflect the pronunciation, resulting in the conventional character-based tokenization misses much more phonological information",
                "Additionally, TABBIE uses a simplified training objective compared to masked language modeling.",
                "Compared to the FullText setting, each iteration exhibits less time consumption in PartText setting",
                "Compared to the traditional federated learning, federated adaptation has the following differences.",
                "we analyze the quality of crowd- sourced annotations and compare them to the results of skilled an- notators.",
                "Compared to the MLM objective",
                "based on our findings, we draw the following conclusions: the lower performance of sparql compared to sql for virtuoso suggests that storing rdf data in a single relational table with extensive indexing is not favourable for transactional graph workloads due to index maintenance and query translation costs.",
                "This helps Ocelot in reducing memory transactions when compared to GPL.",
                "Compared to conventional\napproaches, these methods are able to jointly learn feature representations and the prediction function (or forecasting function) by\nstacking a series of non-linear layers to perform feature extraction,\nfollowed by a regression layer focused on forecasting.",
                "For Instacart, it took 65.16 secs from start to finish in the decentralized setting, compared to 99.98 secs in the centralized setting.",
                "our AdaRNN achieves lowest distribution distances compared to MMD and adversarial training.",
                "compared to convolutional networks, transformers enjoy the ability of referring to global features at each stage, yet the attention module brings higher computational overhead that obstructs the application of transformers to process highresolution visual data.",
                "Compared to other state-of-the-art methods based on global rotation estimation, in our method there exists a fallback evel using points and lines to estimate the full pose, in case no Manhattan frame can be found.",
                "We have a slow CPU compared to the arrival rate of the stream.",
                "\"in addition it achieves good performance compared to previously discussed techniques.\""
            ]
        },
        "compare against": {
            "sentences": [
                "We also compare against a recent research project MorphStore",
                "we compare our PLSLAM method against current state-of-the-art Visual SLAM methods",
                "As an upper bound, we can then compare against the performance of the closed-domain system scored using a single table retrieved through an oracle.",
                "In this section, we report the results of a detailed empirical analysis of CoST and compare it against a\ndiverse set of time series representation learning approaches",
                "In order to evaluate our method, on the one hand, we compare against several monocular SLAM frameworks",
                "e choose Faster R-CNN [54] as the base detector as Dhamija et al. [8] has found that it has better open set performance when compared against one-stage Retina-Net detector [31] and objectness based YOLO detector [52].",
                "we compare against learned regular expression and dictionary-based benchmarks representative of the approaches taken by these systems."
            ]
        },
        "cpmpare with": {
            "sentences": [
                "Compared with ORB-SLAM which only initializes after the 628th frame, our method is able to initialization much earlier around frame 110"
            ]
        },
        "compare ... against ...": {
            "sentences": [
                "When we compare VLL (implemented within the Calvin framework) against Calvin\u2019s native lock manager"
            ]
        },
        "as compare to": {
            "sentences": [
                "Table 3 shows the accuracy of our best model as compared to other approaches7 with different level of hardness defined by Yu et al. (2018)."
            ]
        },
        "compare between": {
            "sentences": [
                "Hence we also compare the maximum path length between any two input and output positions in networks composed of the different layer types."
            ]
        },
        "be compare to": {
            "sentences": [
                "Our approach can also be compared to other selfsupervised methods, synthetic-to-real domain-adaptation methods.",
                "One unique aspect of this dataset when compared to previously discussed video description datasets is that that the videos are annotated with procedure segments that contain rich semantic information."
            ]
        },
        "compare ... with": {
            "sentences": [
                "\ufeffIn this Sub-section, we compare our model with state-of-theart models on the paraphrase identification task."
            ]
        },
        "compare against sth.": {
            "sentences": [
                "here is a lack of consensus on how much performance improvement can be obtained from using GPUs. Past work frequently compares against inefficient baselines."
            ]
        },
        "compare sth against sth": {
            "sentences": [
                "We compared it against the canonical write-ahead logging (WAL) protocol that is used in most DBMSs today."
            ]
        }
    },
    "comparison": {
        "comparison to": {
            "sentences": [
                "In comparison to a dense convolution network, our sparse-convolution-based detector achieves a factor-of-4 speed enhancement during training on the KITTI dataset and a factor-of-3 improvement in the speed of inference.",
                "our validation results for car detection for comparison with other methods"
            ]
        }
    },
    "compensate": {
        "compensate for": {
            "sentences": [
                "To compensate for the potential lack of texture, or visual information necessary to determine 2D correspondences, we directly utilize 3D information from the underlying geometry instead."
            ]
        },
        "compensate with": {
            "sentences": [
                "Participants were compensated with $6 USD."
            ]
        }
    },
    "compete": {
        "compete with": {
            "sentences": [
                "The neighbors of the root vertex are transactions that do not compete with the others"
            ]
        },
        "compete for": {
            "sentences": [
                "In an ideal case, where no transactions compete for the same tuple, the maximal parallelism can be achieved by recovering all transactions in parallel"
            ]
        }
    },
    "complain": {
        "complain about": {
            "sentences": [
                "She always complain about her noisy sister.",
                "She always complain about her noisy sister"
            ]
        }
    },
    "complement": {
        "complement with": {
            "sentences": [
                "Complemented with 2D images, 3D data provides an opportunity for a better understanding of the surrounding environment for machines."
            ]
        }
    },
    "complicate": {
        "complicate sth": {
            "sentences": [
                "Relaxing the ordering of writes to durable storage complicates WBL\u2019s commit and recovery protocols."
            ]
        }
    },
    "comply": {
        "comply with": {
            "sentences": [
                "while letting PolarFS comply with traditional I/O semantics."
            ]
        }
    },
    "compose": {
        "compose by": {
            "sentences": [
                "A knowledge graph is composed by a large amount of triples in the form of (head entity, relation, tail entity)",
                "The mini-network itself resembles the big network and is composed by basic modules of point independent feature extraction, max pooling and fully connected layers."
            ]
        },
        "compose of": {
            "sentences": [
                "composed of real road sequences captured by a car-mounted stereo camera rig.",
                "The environment is composed of a flat ground and vertical walls",
                "The remaining system is composed of a single device, and thus homogeneous, subsystems.",
                "[2020] proposes a global context-enhanced graph convolutional network (gcgcn), composed of entities as nodes and the contexts of entity pairs as edges between nodes.",
                "An expert committee composed of four retinal specialists was formed and to label a random subset according to dozens of pre-specified lesions related to varied retinal diseases.",
                "As the core of PointNet++ hierarchy, its set abstraction level is composed of three layers: the sampling layer, the grouping layer and the PointNet based learning layer.",
                "In our case, the empirical risk of classifier is composed of two parts: the loss on the clean positive examples and the loss on the unlabeled examples.",
                "ACC uses a model composed of classifiers to predict the optimal protocol for a workload.",
                "Suppose the dataset O is composed of N instances of (u,i,p)."
            ]
        },
        "compose as": {
            "sentences": [
                "Each time series is composed as the aggregate of multiple\ngenerated time series, each representing a particular component type."
            ]
        },
        "be compose of": {
            "sentences": [
                "DLMM is composed of four main steps.",
                "The working process is composed of four main stages, the testcase generation stage, testcase running stage, program execution state monitoring and analysis of exceptions."
            ]
        },
        "be compose to do": {
            "sentences": [
                "We present Crystal, a library of data processing primitives that can be composed together to generate efficient query code that can full advantage of GPU resources."
            ]
        },
        "compose with": {
            "sentences": [
                "HolE [46] composes the entity representations with the circular correlation operation"
            ]
        }
    },
    "compress": {
        "compress into": {
            "sentences": [
                "We further compress the representations into another space with linear transformation, denoted as {cj } and vi , respectively."
            ]
        },
        "compress the footprint log": {
            "sentences": [
                "Once a transaction completes, it compresses the footprint log, which con\u0002tributes to a slight delay in response"
            ]
        },
        "compress as well": {
            "sentences": [
                "If most components in c are also zero, then it can be compressed as well."
            ]
        },
        "compress sth": {
            "sentences": [
                "Once a transaction completes, it compresses the footprint log, which con\u0002tributes to a slight delay in response"
            ]
        }
    },
    "comprise": {
        "comprise of": {
            "sentences": [
                "Based on multi-modal fundus imaging comprised of OCT and color fundus photography, we aim for automated categorization of the fundus condition of a specific eye",
                "The traditional WAL recovery algorithm (see Figure 6) comprises of three phases."
            ]
        },
        "comprise the complete scheme": {
            "sentences": [
                "Figure 4 comprises the complete scheme, detailed next."
            ]
        },
        "be comprise of": {
            "sentences": [
                "Our general methodology is comprised of three parts, as detailed below."
            ]
        }
    },
    "compute": {
        "compute environment as": {
            "sentences": [
                "\"We believe that there are no fundamental obstacles to making a cloud-computing environment as secure as the vast majority of in-house IT environments"
            ]
        },
        "compute for": {
            "sentences": [
                "In other words, we compute for every entity the best match in the table, and score the table as the sum over the best matches."
            ]
        },
        "compute by": {
            "sentences": [
                "We illustrate the 3D features computed by the selected neurons in the net and develop intuitive explanations for its performance."
            ]
        },
        "be compute": {
            "sentences": [
                "The prototype of each class could then be computed on source-only, target-only and source-target data, respectively."
            ]
        },
        "be compute by": {
            "sentences": [
                "The MLM loss is then computed by the cross-entropy function on predicting the masked tokens."
            ]
        },
        "compute to": {
            "sentences": [
                "Many database operations have a low compute to memory access ratio."
            ]
        },
        "be compute on": {
            "sentences": [
                "these features must be computed on a subset of the records in the column"
            ]
        }
    },
    "concatenate": {
        "concatenate with": {
            "sentences": [
                "We randomly select a window of value vectors, concatenate the column vector with the remaining value vectors, then train a single model to predict the former from the latter.",
                "As described in section 2, an example is built by concatenating the statement with the flattened table."
            ]
        },
        "concatenate into": {
            "sentences": [
                "concatenate these feature maps into one feature map."
            ]
        },
        "concatenate them together": {
            "sentences": [
                "The data feature vector x D has k-dimensions, where each dimension is the distance between a data sample to query q, and weuse k data samples instead of the entire dataset D.By concatenating them together."
            ]
        },
        "concatenate for": {
            "sentences": [
                "We then concatenate the representations for two corresponding images side by side"
            ]
        },
        "concatenate along": {
            "sentences": [
                "We concatenate along the spatial dimension rather than the channel dimension"
            ]
        }
    },
    "concentrate": {
        "concentrate on": {
            "sentences": [
                "because most of the system components of havingblockchain and mining concentrate on the company and community nodes",
                "Initially, Denis et al. [23] concentrate on the learning complexity and show that the function classes learnable under the statistical query model are also learnable from positive and unlabeled data.",
                "universe sampling shows poor performance in approximating queries when the frequencies of keys are concentrated on a few elements.",
                "In this paper, we concentrate on relational queries",
                "Triplet enhanced pre-trained models concentrate on triplets level knowledge and inject knowledge"
            ]
        }
    },
    "concern": {
        "comcern with": {
            "sentences": [
                "we are concerned with the problem of using a team of robots to gain situational awareness over a large environment, under realistic constraints on communication bandwidth, local sensing, and computation at each robot."
            ]
        },
        "concern with": {
            "sentences": [
                "The last two traits concern the data flow in the system. Data locality is concerned with the distance of the data from their consumer.",
                "In this work we are especially concerned with the version of this problem where only very few target labeled samples per class are available.",
                "In general data stream environments, data continuously arrives and users are concerned with real-time analysis results."
            ]
        }
    },
    "conclude": {
        "conclude sth": {
            "sentences": [
                "Though we conclude that the max salience selector with row items is the best method, the conclusion may depend on the corpus."
            ]
        },
        "conclude that": {
            "sentences": [
                "Meanwhile, we can also safely conclude that the over all score of the two causes of a is larger than b."
            ]
        },
        "conclude with": {
            "sentences": [
                "We conclude the paper with open challenges and possible future work",
                "We conclude with a discussion of related work in Section 8.",
                "Finally, we conclude with a discussion of related work"
            ]
        },
        "conclude  widely": {
            "sentences": [
                "This has been  gained by running Dynamo in production and Section 7 concludes  widely acknowledged by both the industry and academia"
            ]
        }
    },
    "condition": {
        "condition on": {
            "sentences": [
                "all diffusion models are conditioned on the text embedding sequence and use classifier-free guidance [27].",
                "We condition on the last embedding hj of the sequence to predict the next token y\u02c6j+1.",
                "3.1 handshaking tagging scheme 3.1.1 tagging according to the insight that a triplet (s, r, o) can be determined by aligning the boundary tokens of subject entity s and object entity o conditioned on the relation r, we realize one-stage joint extraction by tagging token pairs with link labels."
            ]
        },
        "be condition": {
            "sentences": [
                "Similarly, models can be conditioned on labels by prepending labels"
            ]
        }
    },
    "conduct": {
        "conduct on": {
            "sentences": [
                "In addition to a node-level lock to ensure only one thread can conduct SMO on the node",
                "Finally, we conduct experiments on the real-world dataset to validate the effectiveness of the proposed protocols",
                "Therefore, we also conduct experiments on the dataset from another domain",
                "the loop detection is conducted on each newly added keyframe",
                "We conduct experiments on KITTI benchmark and show that VoxelNet produces state-of-the-art results in LiDAR-based car",
                "Its testing process is conducted on a set of new tasks",
                "We conduct an extensive set of experiments on several real-life and synthetic datasets."
            ]
        },
        "conduct over": {
            "sentences": [
                "Then relational reasoning is conducted over the graph to update the nodes\u2019 representations"
            ]
        },
        "conduct so far": {
            "sentences": [
                "Related Works  Various studies have been conducted so far with regard to the scaling of the web  applications."
            ]
        },
        "conduct to": {
            "sentences": [
                "we conducted to demonstrate the effectiveness of incorporating 3D points, lines and planes."
            ]
        },
        "conduct upon": {
            "sentences": [
                "Convolutions are then conducted upon the features concatenated from vertices of spherical lattices and their neighbors through convolution-maxpooling-convolution structures."
            ]
        },
        "be conduct on": {
            "sentences": [
                "Extensive experiments are conducted on the transfers across MNIST, USPS and SVHN datasets, and superior results are reported when comparing to state-of-theart approaches.",
                "Q1, Q2, and Q3 are conducted on the Google compute cluster monitoring dataset."
            ]
        },
        "conduct with": {
            "sentences": [
                "All the experiments in this paper are conducted with a signle 16GB Nvidia P100 GPU.",
                "Pilot tests were conducted with a set of privacy policies different from those used in the actual study.",
                "However, malicious tasks can also be conducted with VPU, such as discovery of harmful chemical substance."
            ]
        },
        "conduct task to": {
            "sentences": [
                "Specifically, we conduct two tasks to evaluate on four criteria: (1) relevance, (2) attractiveness, (3) language fluency, and (4) style strength."
            ]
        },
        "conduct the experiment": {
            "sentences": [
                "We conduct the experiments of variance analysis on the write heavy workload."
            ]
        },
        "be conduct to": {
            "sentences": [
                "Extensive experiments based on benchmark datasets have been conducted to verify the effectiveness of the proposed scheme and numerically show the trade-off between privacy guarantee and model accuracy."
            ]
        },
        "conduct sth": {
            "sentences": [
                "We conduct experiments on various workloads to show its superiority."
            ]
        }
    },
    "configure": {
        "configure with": {
            "sentences": [
                "A workload can be configured with multiple clients and threads per clients to saturate the blockchain throughput",
                "For the general policies configured with service rate 1.0 and for workflow-specific policies in Table 3 and Table 4 job elastic slowdowns show low variability",
                "For example, the first cluster is configured with OCC since it receives transactions having few conflicts."
            ]
        },
        "configure as": {
            "sentences": [
                "Local context\nis introduced with auto-regression and covariate modules, which can be configured as classical\nlinear regression or as Neural Networks."
            ]
        }
    },
    "confirm": {
        "be confirm by": {
            "sentences": [
                "This is confirmed by the results shown in Figure 7b (with log scale on y-axis), that with an increasing number of distributed transactions, the impact on latency contributed by logging becomes secondary",
                "This problem was confirmed by recent work"
            ]
        },
        "confirm that": {
            "sentences": [
                "The above study confirms that filtering algorithms can improve the performance of RNN and Transformer based sequential recommendation models."
            ]
        },
        "confirm as": {
            "sentences": [
                "a block can be confirmed as part of the blockchain only with some high probability"
            ]
        }
    },
    "conflict": {
        "conflict of": {
            "sentences": [
                "The authors declare no conflict of interest."
            ]
        },
        "conflict with": {
            "sentences": [
                "Since a read in t can potentially conflict with all the other transactions in the batch",
                "Log entries not conflicting with any other entry can be applied safely",
                "Increasing the length of the chains means that transactions take longer to execute, thereby increasing the likelihood that a transaction will conflict with another one.",
                "The transactions must not conflict with those in previous blocks of the same chain.",
                "First, for PartCC and 2PL if a transaction ti conflicts with another transaction tj , ti cannot proceed until tj commits or vice versa, so PartCC and 2PL are COCSR. Shang et al"
            ]
        }
    },
    "conform": {
        "conform to": {
            "sentences": [
                "While this is convenient, it does not conform to the motivation of learning from PU data."
            ]
        }
    },
    "confuse": {
        "confuse to": {
            "sentences": [
                "pedestrians and cyclists to be more easily confused with other point clusters and noise"
            ]
        },
        "confuse with": {
            "sentences": [
                "After learning each task, WI and A-OSE metrics are used to quantify how unknown instances are confused with any of the known classes"
            ]
        }
    },
    "connect": {
        "connect into": {
            "sentences": [
                "Network Resource: includes the network cards that connect into servers as well as infrastructure components that include repeaters, load balancers, switches and fire\u0002walls."
            ]
        },
        "connect with": {
            "sentences": [
                "we conducted our experiments on two machines in a local cluster connected with 10 gbit/s ethernet (one for running the sut, the other for running the benchmarking code)."
            ]
        },
        "connect to": {
            "sentences": [
                "in other words, all regular tokens are not explicitly connected to other regions, and msg tokens serve as the hub of information exchange.",
                "GPUs are often used as coprocessors that are connectedto CPUs through PCI-e",
                "as sensed by the sensors on the interface board which is connected to SBC over serial port",
                "Redundant data are also a problem since each frame is connected to all the existing data.",
                "The target patch is connected to spatially similar patches [16].",
                "Indeed, this turns out to be formally connected to measures of domain discrepancy."
            ]
        },
        "connect layer": {
            "sentences": [
                "the architecture in our simulated experiments is the same, except with  16 \ufb01lters in each convolution layer, fea- ture \ufb02attening instead of a spatial soft-argmax, and three fully-connected layers instead of one."
            ]
        },
        "connect xxx through": {
            "sentences": [
                "The best performing models also connect the encoder and decoder through an attention mechanism."
            ]
        },
        "connect up to": {
            "sentences": [
                "NVLink 2.0 connects up to one CPU and six GPUs in a point-to-point mesh topology."
            ]
        },
        "connect by": {
            "sentences": [
                "the embedded entities h and t can be connected by translation vector r"
            ]
        }
    },
    "consent": {
        "consent to": {
            "sentences": [
                "She won't consent to him staying out late.",
                "The upper task can not consent to the error from SLAM."
            ]
        }
    },
    "consider": {
        "consider as": {
            "sentences": [
                "If we consider SA as an oracle, there are two rounds of communication in each iteration",
                "In MCCNN [68], convolution is considered as a Monte Carlo estimation process relying on a sample\u2019s density function (which is implemented with MLP).",
                "When applied to few-shot multi-hop reasoning, triple queries with the same relation are considered as a task",
                "We consider distributed manners as efficiency .",
                "we consider 'poor quality' as 'ignore' .",
                "An anchor is considered as negative if the IoU between it and all ground truth boxes is less than 0.45."
            ]
        },
        "consider sth": {
            "sentences": [
                "In this paper we consider a related but different approach.",
                "Consider the table in Figure 1 and the sentence \u2018[Greg Norman] is [Australian].\u2019(Square brackets indicate mention boundaries.)."
            ]
        },
        "consider with": {
            "sentences": [
                "who considered both open and closed-loop models with the following specific applications."
            ]
        },
        "consider on": {
            "sentences": [
                "although more advanced indexing schemes can be considered on an individual basis, we restricted indexes only on vertex ids for fairness."
            ]
        },
        "consider for": {
            "sentences": [
                "this suggests that TCN models are still a powerful\nbaseline which should still be considered for time series."
            ]
        },
        "consider over": {
            "sentences": [
                "But, the majority of these works do not consider query execution over heterogeneous devices, for example multiple GPUs."
            ]
        },
        "be consider as": {
            "sentences": [
                "In fact, both mixture of regressions and mixture of classifiers can be considered as special cases of our problem in the centralized setting."
            ]
        },
        "consider we formally": {
            "sentences": [
                "For ease of presentation, we only consider We formally de(cid:12)ne the two operations, and prove that the undirectedgraphsinthispaper."
            ]
        },
        "consider a very": {
            "sentences": [
                "We consider a very sim- itsdescription,volume,weightandsales-relateddataarekeptinthe pleworkloadonarelationcontainingN tupleswithfourattributes materialdetailstableandthematerialhierarchy."
            ]
        }
    },
    "consist": {
        "consist of": {
            "sentences": [
                "For these cases an optimistic protocol may be ideal if the workload mainly consists of read operations or has low-skew access patterns",
                "each graph partition consists of a set of vertices, all outgoing edges rooted at those vertices, and associate attributes.",
                "\ufeffThe classification model consists of the respective transformer, on top of which we add a fully \ufeffconnected projection layer, and a final softmax output layer.",
                "The current state of the art consists of two main approaches",
                "we introduce a large-scale dataset called TABFACT, which consists of 118K manually annotated statements with regard to 16K Wikipedia tables",
                "At a high-level, we train subnetworks for each feature category except the statistical features, which consist of only 27 features.",
                "The input to the matrix factorization (MF) [30] problem is a matrix consisting of ratings from n users on m items",
                "Stehle and Jacobsen present an efficient sorting algorithm that consists of two steps.",
                "It consists entirely of feedforward neural networks",
                "The dataset consists of almost 15,000 multiple choice questions and answers obtained from over 400 movies having high diversity.",
                "A conversation in QReCC consists of a sequence of question-answer pairs.",
                "It consists of a tuple heap in NVM, an Met-Cache in DRAM, and per-thread NVM-tuple managers.",
                "Our between-subjects study consisted of a control condition and two treatment conditions that highlighted different numbers of para- graphs (five and ten), in order to investigate the effects of the num- ber of highlights on annotation accuracy and productivity.",
                "\"a sorted array a then member(x) consists of just checking doi:https://doi.org/10.14778/3389133.3389135\"",
                "APEX consists of inner nodes and data nodes.",
                "The main reason behind it is that the encoderdecoder-based model directly adopts the stochastic variables\nas input and the back-propagation loss also consists of future uncertain data.",
                "The convolutional kernels consist of a weighting function (which is learned with MLP layers) and a density function (which is learned by a kernelized density estimation and an MLP layer).",
                "The test dataset consists of the testing triples of r with ground-truth tail entities t_i for each query (h_i, r)",
                "Table 1 systematically shows the different strategies for design points of the current SOTA of SSOD, which consists of strong data augmentations, initialization & update strategy of teacher network, pseudo label generation strategy, unsupervised loss functions",
                "gda [wu et al., 2019] is a dataset in the biomedical domain, which consists of 23,353 training samples.",
                "The agent is trained on a dataset consisting of episodes of the robot stacking objects with a variety of different shapes.",
                "figure 2 shows the data structures of livegraph, which mainly consist of vertex blocks (vb) and tels.",
                "The 31 categories in the dataset consist of objects commonly encountered in office settings, such as keyboards, file cabinets, and laptops.",
                "The stored map consists of 3D map points",
                "GPUs consist of a large amount of lightweight computing cores",
                "The subfield of mathematics and computer vision known as Multiple View Geometry [9] consists of theorems and algorithms built on the assumption that interest points can be reliably extracted and matched across images.",
                "Given that our training data consists of only 100K samples for each query complexity, our goal is to understand whether thedeep learning models can actually compress information and still provide a good accuracy.",
                "Here, the training data consists of the tuples already in the KB.",
                "ERNIE encoder consists of a text encoder",
                "it consists of a Multihead attention(MSA) layer and a MLP block",
                "In addition, an auto-scaling policy consists not only of performance metrics thresholds, but also temporal parameters, which often seem to be neglected.",
                "The algorithm consists of two phases: in the build phase.",
                "where each data point consists of a pair of feature and response denoted by zi = (xi, yi).",
                "The algorithm consists of three major components",
                "Another strategy consists of assuming an underlying (global) structure in the world frame",
                "As show in Figure 1, Prism runs multiple m+1 separate parallel \u201cblocktrees\u201d where one of the trees, called the proposer tree, consists of blocks from which the final transaction ledger is constructed.",
                "The proposed VoxelNet consists of three functional blocks:",
                "A typical smart home consists of three parties: IoT device, service provider and user.",
                "The system consists of three components: processors, storage, and one or more validators",
                "The proposed system works on Assembly code and consists of:",
                "\"This paper consists of 4 parts: introduction, theory experimental method and conclusion.\""
            ]
        },
        "consist of approximately": {
            "sentences": [
                "consisted of approximately 1800 machines."
            ]
        },
        "consist in": {
            "sentences": [
                "One solution for this consists in using a very small number of keypoints"
            ]
        },
        "consist of first": {
            "sentences": [
                "while initial progress has been made towards a similar capability in computer vision, the most used paradigm still consists of first pretraining on a large amount of supervised multimodal data collected from the web, before fine-tuning the model on the task of interest (lu et al., 2019; wang et al., 2021; zellers et al., 2022)."
            ]
        },
        "consist of just": {
            "sentences": [
                "The Chubbyclient library providesconsis- consists of just onetablet."
            ]
        },
        "consist of only": {
            "sentences": [
                "The vast majority of directories (and therefore customers) consist of only 1 fragment, which means that reads and writes to those customers\u2019 data are guaranteed to occur on only a single server."
            ]
        }
    },
    "consolidate": {
        "consolidate into": {
            "sentences": [
                "Migration can also be used to reduce power consumption, by consolidating VMs into fewer nodes and enabling some nodes to be switched off."
            ]
        }
    },
    "constrain": {
        "constrain sth": {
            "sentences": [
                "We constrain the \u27e8select\u27e9 values of the left and right expression to be either both the count or to have the same value for \u27e8column\u27e9."
            ]
        },
        "constrain by": {
            "sentences": [
                "However, volumetric representation is constrained by its resolution due to data sparsity and computation cost of 3D convolution."
            ]
        }
    },
    "construct": {
        "construct with": {
            "sentences": [
                "To overcome the issue of constructing a loss\nfunction with complex-valued representations, each frequency can be uniquely represented by its\namplitude and phase representations"
            ]
        },
        "construct by": {
            "sentences": [
                "Based on the PointNet architecture, a hierarchical learning architecture is constructed by stacking multiple DPAMs."
            ]
        },
        "construct from": {
            "sentences": [
                "Wanget al.[103] proposed an end-to-end spectral convolution network LocalSpecGCN to work on a local graph (which is constructed from theknearest neighbors)."
            ]
        },
        "be construct to": {
            "sentences": [
                "Meanwhile, a semantic metric learning module is constructed to concurrently learn discriminative and class-aware transferable representations for category-level adaptation."
            ]
        },
        "be construct with": {
            "sentences": [
                "Our model is implemented with Pytorch (Paszke et al., 2019), and the original and line graphs are constructed with library DGL (Wang et al., 2019a)."
            ]
        },
        "construct a ensemble": {
            "sentences": [
                "In the following, we describe our procedure that constructs a so called base ensemble for a given database scheme."
            ]
        },
        "construct to": {
            "sentences": [
                "by constructing meta-learning tasks to simulate domain shift."
            ]
        },
        "construct sth": {
            "sentences": [
                "Instead, it constructs transactionally consistent checkpoints."
            ]
        },
        "be construct as": {
            "sentences": [
                "The shared memory is constructed as multiple ring buffers"
            ]
        }
    },
    "consume": {
        "consume by": {
            "sentences": [
                "the ldbc workload driver generates query parameters, which are consumed by operation executors."
            ]
        },
        "consumes... as...": {
            "sentences": [
                "We design a deep learning framework that directly consumes unordered point sets as inputs."
            ]
        },
        "consume for": {
            "sentences": [
                "Gas is the name of the crypto fuel which is consumed for performing the operations on a Blockchain network"
            ]
        }
    },
    "contain": {
        "contain info i++": {
            "sentences": [
                "\"Done: \u2019result\u2019 structure contains info i++; // about counters"
            ]
        },
        "contain complementary then": {
            "sentences": [
                "\"three views of features"
            ]
        },
        "contain with": {
            "sentences": [
                "each nonempty voxel with 6 statistical quantities that are derived from all the points contained within the voxel."
            ]
        },
        "contain as": {
            "sentences": [
                "Weather is an\nhourly-level dataset containing 11 climate features from nearly 1,600 locations in the U.S., and we\ntake \u201cwet bulb\u201d as the target value for univariate forecasting."
            ]
        },
        "contain of": {
            "sentences": [
                "The SSOD setting we discuss is that the training contains a small set of labeled data and another set of completely unlabeled data"
            ]
        },
        "contain semantically information": {
            "sentences": [
                "We select columns with small discrete domain sizes, as these are generally the columns that contain more semantically meaningful information about the data, unlike columns that contain a sequence of identifiers."
            ]
        },
        "contain arbitrarily": {
            "sentences": [
                "thanks to their flexibility, flamingo models can be trained on large-scale multimodal web corpora containing arbitrarily interleaved text and images, which is key to endow them with in-context few-shot learning capabilities."
            ]
        },
        "contain sth": {
            "sentences": [
                "Each record also contains the after-images (i.e., new values) of the tuples modified."
            ]
        },
        "contain at least": {
            "sentences": [
                "First, by building storage as an independent fault- r w of nodes used for a read intersects with the set of nodes used for a  tolerant and self-healing service across multiple data-centers, we  write and the read quorum contains at least one location with the  protect the database from performance variance and transient or  newest version."
            ]
        },
        "contain at most": {
            "sentences": [
                "Each block contains at most 128 keys and is de-"
            ]
        }
    },
    "contend": {
        "contend with": {
            "sentences": [
                "To contend with this issue, we filtered out the types for which at least 15% of the columns did not contain a single word that is present in the GloVe dictionary."
            ]
        },
        "contend that": {
            "sentences": [
                "Given this, we contend that it is better to employ logging and recovery algorithms that are designed for NVM."
            ]
        }
    },
    "contest": {
        "contest along": {
            "sentences": [
                "while this approach has enabled rapid algorithmic advances in recent years, datasets of this nature have been critiqued and contested along various ethical dimensions."
            ]
        }
    },
    "contextualize": {
        "contextualize sth": {
            "sentences": [
                "The cross-attention function which contextualizes retrieved tables in relation to each other, and the joint reranking-and-verification loss."
            ]
        }
    },
    "continue": {
        "continue to": {
            "sentences": [
                "In each active learn- ing cycle, we continue to train the current model by adding K=1,000 labeled data points.",
                "data storage technology continues to change at a rapid pace",
                "after that each thread can continue to insert the record to its own stash array slot.",
                "when \ud835\udc3e continues to increase, the performance drops.",
                "even  if  the  model  was  trained  with  only  one  to  three  inner  gradient  steps,  we  found in practice that it was bene\ufb01cial to take substantially more gradient steps during meta- testing; performance continued to improve with up to 20 steps.",
                "As the price of main memory continues to drop,",
                "In contrast, in our framework, our pseudo-label generation model (Teacher) continues to evolve gradually and smoothly via Teacher-Student Mutual Learning",
                "FineStream updates the window size parameter after the related batch processing is completed, and then continues to run with performance detection.",
                "ACC continues to process transaction requests during the partitioning and switches to PartCC when partitioning is done"
            ]
        },
        "continue with": {
            "sentences": [
                "This is why the time grows exponentially and then continue with a linear behaviour.",
                "We continue with the notation used in the description before."
            ]
        }
    },
    "contract": {
        "be contract": {
            "sentences": [
                "It may be argued that nodes should be either contracted or merged after entry deletion."
            ]
        }
    },
    "contradict": {
        "contradict with": {
            "sentences": [
                "In addition, durability and recovery algorithms can be optimized for a protocol, but contradict with other implementations or optimizations."
            ]
        }
    },
    "contrary": {
        "contrary  to": {
            "sentences": [
                "Contrary to the fully supervised case, here we note that both domain confusion and soft labels contribute significantly to the overall performance improvement of our method."
            ]
        }
    },
    "contrast": {
        "in contrast": {
            "sentences": [
                "In contrast, pointbased methods directly work on raw point clouds without any voxelization or projection."
            ]
        },
        "contrast to": {
            "sentences": [
                "This contrasts to supervised learning, where a learner has a fully labeled training set and to semi-supervised learning, where a learner (usually) observes some labeled examples from each class."
            ]
        },
        "contrast with": {
            "sentences": [
                "A core problem of contrastive IR model training is selecting proper negative samples to contrast with the positive ones"
            ]
        }
    },
    "contribute": {
        "contribute to": {
            "sentences": [
                "Even if we further abandon security, more aggressive re-parameterization does not allow us to fully exploit the nodes\u2019 bandwidth, because orphaned blocks do not contribute to the transaction confirmation, yet still consume bandwidth to propagate [80].",
                "Each sub-model contributes to the teaching materials according to their reliability, which is measured by the similarity between the testing instance and the prototypes learned in the training dataset.",
                "Inspired by their use in [18], we propose to apply a 1 \u00d7 1 convolutional kernel on the output feature maps from each view, as an efficient dimensionality reduction mechanism that learns to select features that contribute greatly to the performance of the region proposal generation",
                "We  contribute to deriving the distributed solver for the localization problem.",
                "All the proposed components contribute to significantly speed-up the computations.",
                "Note that delegate k only contributes to those components, Ik , that it has updated during this round.",
                "This accordingly contributes to reduce the estimation drift in the front end and reduce computation burdens in the back end",
                "The former simply assumes that all neighbors contribute to the entity embedding equally",
                "\ufeffSelecting parameters from the best in-domain development accuracy partially contributes to this result.",
                "Moreover, the coarse-to-fine module also contributes to the estimation accuracy by refining matches to a sub-pixel level.",
                "Table 6 indicates that mining examples with our calibrated loss contribute to better supervision than using only the loss of nnPU.",
                "The increasing prevalence of mobile networks contributes to massive distributed data continuously produced by a collection of geographically monitoring sites (e.g., sensors, mobile phones, etc.)",
                "Admission control, which is the decision to accept a new work\u0002load if it contributes to the current management objectives and resource adaptation.",
                "we also contribute to deriving the instance probability under the framework of attention based MIL.",
                "These factors greatly contribute to the variation in code layout, making the above two approaches ineffective."
            ]
        },
        "contribute to nearly": {
            "sentences": [
                "25% of largequeries contribute to nearly 50% of total execution time."
            ]
        },
        "contribute by": {
            "sentences": [
                "A DAO manages funds contributed by participants and gives its users voting power proportional to their contributions"
            ]
        },
        "contribute for": {
            "sentences": [
                "We contribute an annotated dataset for researchers who want to study the problem."
            ]
        }
    },
    "control": {
        "be control to": {
            "sentences": [
                "In our method, the direction classification weight can be controlled to make the network focus more on overlap maximization  instead of struggling with  direction recognition in difficult cases."
            ]
        },
        "control flow": {
            "sentences": [
                "In our prototype the device provider translates the code generation directives to LLVM IR and the generated code contains control flow statements, such as branches and loops."
            ]
        },
        "control by": {
            "sentences": [
                "Controlled by a constant sampling factor c, we set u = c \u00b7 ln LQ , which makes the ProbSparse selfattention only need to calculate O(lnLQ) dot-product for each query-key lookup and the layer memory usage maintains O(LK ln LQ).",
                "Slock it has developed a smart lock technology called Slocks which enables real-world physical objects to be controlled by the Blockchain"
            ]
        },
        "concurrency control": {
            "sentences": [
                "Meanwhile, single-partition transactions can still be executed without any concurrency control at all"
            ]
        },
        "take control of": {
            "sentences": [
                "The adversary A is adaptive, and allowed to take control of parties on the fly, as well as \u201crushing\u201d, meaning that in any given round the adversary gets to observe honest parties\u2019 actions before deciding how to react."
            ]
        }
    },
    "converge": {
        "converge to": {
            "sentences": [
                "We provide guarantees that BBE will converge to the best estimate achievable over all choices of the bin size and provide consistent estimates whenever a pure top bin exists.",
                "Moreover, policy gradient method with partial data may sometimes converges to a local optima when learning on structured data because there exist many indirect relations among the learning instances."
            ]
        },
        "converge on": {
            "sentences": [
                "Overall, we observe that crowdworkers are able to converge on annotations in a majority of cases."
            ]
        }
    },
    "convert": {
        "convert to": {
            "sentences": [
                "When methods that were originally proposed in a case-control scenario are discussed on a level where this distinction is necessary, we either convert them to the single-training-set scenario or explicitly state that the case-control scenario is assumed.",
                "4.4 comparison models for comparison, we employ the following models as baselines: (1) noveltagging(zheng et al., 2017) applies a novel tagging strategy that incorporates both entity and relation roles, thus the joint extraction task is converted to a sequence labeling problem.",
                "E(x) converts a sequence of tokens to a sequence",
                "Converting a string to its fingerprint allows our algorithm to make faster string comparisons and requires less space for intermediate result.",
                "The two special tokens are also converted to the grayscale image to obtain the corresponding feature maps",
                "Having the time converted to cartesian point simplifies our temporal queries.",
                "For a query call, the operation is converted to access the lookup table which rreturns a pre-computed result.",
                "The fundamental idea is to convert a portion of the code to a different machine code."
            ]
        },
        "convert sth. into": {
            "sentences": [
                "We perform Fast Fourier Transform (FFT) to convert the input representations into the frequency domain."
            ]
        },
        "convert into": {
            "sentences": [
                "It first groups point cloud data into voxels and then applies linear networks voxel by voxel before converting the voxels into dense 3D tensors to be used in a region proposal",
                "MV3D is the first method to convert point cloud data into a BEV representation.",
                "We devised a method that converts the point cloud into a dense tensor structure where stacked VFE operations can be processed in parallel across points and voxels.",
                "To align with our designed input, we convert the SQL into the format we expected.",
                "Point cloud data are converted into a 2D bird\u2019s eye view image [8] or are projected onto an image",
                "We then convert the mapping into a ground-truth probabilistic mapping",
                "In the cooperative learning stage, fixing the user embedding can convert the problem into a convex problem, so that item embeddings\u2019 privacy preserved learning process is easier",
                "These phrases were first normalized (for stemming and capitalization) and then converted into a list of 110 regular expressions, such as:",
                "Most 3D-based methods either use point cloud data directly or require converting these data into 3D grids",
                "we convert the dataset into hourlylevel measurements and set \u201cMT 320\u201d as the target value for univariate forecasting.",
                "A variety of approaches have been proposed, for example hybrid consensus algorithms try to convert the permissionless problem into a permissioned consensus problem by subselecting a set of miners from a previous epoch.",
                "generation by converting the data into the text-to-text format",
                "point cloud data are converted into several slices to obtain height maps"
            ]
        },
        "convert ... into ...": {
            "sentences": [
                "Feature-based DNNs:[6,8] firstly convert the 3D data into a vector, by extracting traditional shape features and then use a fully connected net to classify the shape."
            ]
        },
        "convert them back": {
            "sentences": [
                "Different from bit slicing, we slice a value into groups of bits and convert them back into base-10 integers."
            ]
        },
        "convert ... into": {
            "sentences": [
                "\ufeffMeanwhile, our convolutional layers convert the texts into fix-length vectors and keep the important sequence information which are lost in bag-of-words representations."
            ]
        }
    },
    "converte": {
        "converte into": {
            "sentences": [
                "point cloud data are converted into a front view and a BEV"
            ]
        }
    },
    "convey": {
        "convey a mean": {
            "sentences": [
                "It is quite possible that captions 663 that may contain the same words (lexicons) might convey a totally different meaning."
            ]
        }
    },
    "cooperate": {
        "cooperate with": {
            "sentences": [
                "To process distributed transactions, multiple sites have to cooperate with each other, and as a result, the coordination cost rises with the increased amount of participating sites"
            ]
        }
    },
    "coordinate": {
        "coordinate with": {
            "sentences": [
                "Deterministic active-active replication seeks to reduce the network communication needed to ship logs and coordinating with other replicas in an active-passive scheme",
                "any server in the system that receives the transaction from a client can assign the timestamp, without coordinating with other servers."
            ]
        },
        "be coordinate with": {
            "sentences": [
                "In contrast, transactions involving multiple nodes in existing distributed partitioning based systems are coordinated with two-phase commit"
            ]
        },
        "be coordinate within": {
            "sentences": [
                "Furthermore, because the node has a complete replica, transactions can be coordinated within a single machine without a slow commit protocol like two-phase commit"
            ]
        },
        "coordinate sb to do sth": {
            "sentences": [
                "Besides, there is a server who coordinates the P clients to accomplish the FedMF task with all the distributed data X1, X2, . . . , XP being considered."
            ]
        },
        "coordinate to": {
            "sentences": [
                "This chapter showed a detailed overview of our model and how we can use the cartesian coordinate to design temporal queries."
            ]
        },
        "coordinate  locally": {
            "sentences": [
                "For    example, the background controller checks to see how close the  99th percentile database read latency (over the last 60 seconds) is  coordinated  locally  if  Dynamo  is  using  timestamps  based  to  a  preset  threshold  (say  50ms)."
            ]
        }
    },
    "cope": {
        "cope with": {
            "sentences": [
                "Taken together, it is anticipated to design self-adaptive approximate schemes to cope with the dynamic change of stream data.",
                "To cope with the accumulation of error, we can perform loop-closure detection",
                "A key benefit of temporal graphs is that they rely on standard shortest-path algorithms, hence most existing software tools can easily cope with temporal graph analysis.",
                "which is a popular formalism to cope with the exploration vs. exploitation conflict in decision science.",
                "However, most techniques can cope with only further work in this community, but one foreseeable limited sequences (15-200 frames), mainly due to memory challenge is the lack of ground truth for evaluation.",
                "A disadvantage of the offline approach is it may have a significant overhead and may not cope with dynamic behaviour of some workloads.",
                "A distinguishing feature of TURL is its universal architecture across different tasks - only minimal modification is needed to cope with each downstream task.",
                "To cope with the intensive practical demands as listed above, PU learning is proposed to handle the situations where only positive data and unlabeled data are available."
            ]
        }
    },
    "copy": {
        "copy from": {
            "sentences": [
                "the data must be copied from SSD to DRAM for the DBMS to operate on it."
            ]
        }
    },
    "corpus": {
        "corpus significantly": {
            "sentences": [
                "language models are trained on text only corpus significantly larger than paired image-text data, thus being exposed to a very rich and wide distribution of text."
            ]
        }
    },
    "corre-": {
        "corre- zoom in": {
            "sentences": [
                "The inset Once more, these increased-locality indications also corre- zooms in on the \ufb01ve hours after the \ufb01rst."
            ]
        }
    },
    "correspond": {
        "correspond to": {
            "sentences": [
                "By the linearity properties of the HWT andthe earlier description, an update to the data entries corresponds to only polylogarithmically many coefficients in thewavelet domain.",
                "The resultant vocabulary contains tokens corresponding to flexible combinations of characters and sub-characters",
                "This corresponds to slowly decreasing the probability of accepting worse policies as it explores the state space.",
                "In our problem, each training task corresponds to a KG relations and has its own training/testing triples",
                "However, these methods are not problemless. In classification, as the classes correspond to different semantics and different characteristics, the marginal distribution of the data naturally has a class-conditional multi-modal structure.",
                "for the convenience to illustrate, we show all tags in one matrix, in which each color corresponds to a specific kind of tag.",
                "The authors assume that the history of a node-labeled graph is given in the form of graph snapshots corresponding to the state of the graph at different time instants.",
                "Each item j in cluster i corresponds to an value vij.",
                "examples from the latter consist of text with interleaved images, corresponding to the page layout when viewed in a browser.",
                "Lower FID is better as it corresponds to more similar generated and original samples.",
                "Raft RequestVote corresponds to Phase1a in Paxos",
                "Higher scores correspond to a higher proportion of positives, and when the positive data contains a separable sub-domain, i.e., a region of the input space where only the positive distribution has support, classifiers often exhibit a threshold above which the top bin contains mostly positive examples.",
                "where \ud835\udc60\ud835\udc56\ud835\udc57 corresponds to the relevance of encoding scheme \ud835\udc52\ud835\udc57 to column \ud835\udc50\ud835\udc56",
                "Those components of the parameters w that do not correspond to either user or item embeddings are updated in the same manner as the FedAvg algorithm.",
                "This implies that the output of the learner does not correspond to the final network output",
                "This corresponds to the situation that the network has some base hash rate.",
                "each node corresponds to a combination of existing or candidate indexes",
                "Each layer of the network corresponds to one layer of the octree and a spherical convolutional kernel is applied at each layer.",
                "These 100 data points correspond to approximately 100 minutes of load generation, monitoring, and data gathering for each VM, and for each load type, resulting in approximately 26 hours of data collection.",
                "Finally, keypoints correspond to locations that maximize both confidence maps.",
                "All images which correspond to the above split from Pascal VOC and MS-COCO train-sets form the training data",
                "Matrix factorization (MF) approximates unobserved ratings in a rating matrix, whose rows correspond to users and columns correspond to items to be rated",
                "This content could correspond to item images, reviews, descriptive text, or other item-specific information.",
                "the same edge now correspond to multiple entries in the log, as shown in figure 3.",
                "\ufeffSince a non-leaf node does not correspond to a specific word, we use a special vector x0 t as its input, which is like an unknown word."
            ]
        }
    },
    "cost": {
        "cost 7.5x as": {
            "sentences": [
                "\"Fallacy Since an AWS Lambda cloud function instance with equivalent memory capacity of an on- demand AWS t3.nano instance (0.5 GiB) costs 7.5x as much per minute"
            ]
        },
        "cost of": {
            "sentences": [
                "There have been a number of studies on estimating the cost of individual database operators."
            ]
        }
    },
    "count": {
        "count on": {
            "sentences": [
                "Moreover, comparing the accumulated attention weights per modality reveals which modality the model truly counts on for recognizing specific diseases",
                "All UAVs can count on the computation task.",
                "nobody can count on impunity."
            ]
        },
        "count from": {
            "sentences": [
                "3.3 u-shaped segmentation module taking the entity-level relation matrix f as a dchannel image, we formulate the document-level relation prediction as the pixel-level mask in f. where n is the largest number of entities, counted from all the dataset samples."
            ]
        },
        "count by": {
            "sentences": [
                "The time is counted by iterating over synthetic datasets with variant dimensions from 10 to 10,000."
            ]
        },
        "count to": {
            "sentences": [
                "The rationale behind this decision is to allow branch counts to be considered different only when they differ in orders of magnitude."
            ]
        }
    },
    "couple": {
        "couple with": {
            "sentences": [
                "The method uses direct image alignment coupled with filtering-based estimation of semi-dense depth maps",
                "It enables sampling strategies of selecting multiple positive and negative samples in the loss terms, coupled with memory banks to record large variations in video clips.",
                "This rise in memory capacity, coupled with the ability to equip a modern server with several GPUs (up to 20), means that it\u2019s possible to have hundreds of gigabytes of GPU memory on a modern server.",
                "Machine learning models, coupled with large-scale training and benchmarking corpora, have proven effective at predictive tasks across domains.",
                "Regarding HTAP solutions that are tightly coupled with a SQL query engine, Wildfire and SnappyData are two more recent HTAP systems.",
                "Coupled with elastic-net based learning, PRA then picks more plausible paths using supervised learning",
                "Prior work has generally coupled these two tasks with entity linking."
            ]
        }
    },
    "cover": {
        "cover by": {
            "sentences": [
                "Paragraph vectors aim at capturing semantic meanings of table columns, which can be covered by word embeddings if properly trained."
            ]
        },
        "cover the result of": {
            "sentences": [
                "This paper covers the testing results of about 200 assemblies of fuel and control rods in the reactor core."
            ]
        },
        "cover the research on": {
            "sentences": [
                "The cited reports cover the research on the lower atmospheric air pollution movements by using the tracer elements."
            ]
        }
    },
    "crash": {
        "crash into": {
            "sentences": [
                "For example, a Google self-driving car recently crashed into a bus because it expected the bus to yield under a set of rare conditions but the bus did no."
            ]
        }
    },
    "crawl": {
        "crawl from": {
            "sentences": [
                "Considering the diversity of target application inputs, crawling from the Internet is the most intuitive method.",
                "Paper content and metadata information are crawled from two resources"
            ]
        }
    },
    "create": {
        "create before": {
            "sentences": [
                "in a nutshell, each update transaction is scheduled to be executed after its dependent transactions are guaranteed to be executed, e.g., a post cannot be created before the dependent user and forum are created."
            ]
        },
        "create for": {
            "sentences": [
                "we create a separate lagged regressor module for each of\nthe m covariates x of lenght T ."
            ]
        },
        "be create with": {
            "sentences": [
                "The action sequence is created with our designed grammar."
            ]
        },
        "be create so far": {
            "sentences": [
                "Only a limited set of models have been created so far to handle the task of Language-to-Video Generation."
            ]
        },
        "create sth. instead of": {
            "sentences": [
                "Image paragraph generation aims to create paragraphs instead of generating a single simple description or dense descriptions for an image."
            ]
        },
        "create by": {
            "sentences": [
                "negative examples are created by replacing one entity of pairs"
            ]
        },
        "be create in": {
            "sentences": [
                "when the primary database node executes a create table DDL statement a new file is created in PolarFS"
            ]
        }
    },
    "cross": {
        "cross-partition": {
            "sentences": [
                "Partitioning-based systems work well when workloads have few cross-partition transactions, because they are able to avoid the use of any synchronization and thus scale out to multiple machines"
            ]
        }
    },
    "cur": {
        "cur in": {
            "sentences": [
                "explicit graph reasoning can bridge the gap between entities that occur in different sentences, thus mitigating long-distance dependency and achieving promising performance."
            ]
        }
    },
    "curate": {
        "be curated to": {
            "sentences": [
                "The paper is curated to be self-contained with the most critical material ."
            ]
        }
    },
    "cut": {
        "cut down": {
            "sentences": [
                "Avoid stress foods and cut down on sugar and salt."
            ]
        },
        "cut down on": {
            "sentences": [
                "This setup reduces the sequence length of each Transformer's input, which cuts down on its complexity."
            ]
        }
    },
    "dataset": {
        "dataset by": {
            "sentences": [
                "docred [yao et al., 2019] is a large-scale documentlevel relation extraction dataset by crowdsourcing."
            ]
        }
    },
    "date": {
        "data back to": {
            "sentences": [
                "Their history dates back to Ancient Egypt."
            ]
        },
        "date back to": {
            "sentences": [
                "Their data dates back to last sliding window."
            ]
        },
        "date back": {
            "sentences": [
                "The preliminary researches of PU Learning can be dated back to [23], [24], [25], which reveals that unlabeled data show a great impact on accurately training a binary classifier without the aid of negative examples."
            ]
        }
    },
    "deactivate": {
        "deactivate itself": {
            "sentences": [
                "A vertex can deactivate itself by voting to halt"
            ]
        }
    },
    "deal": {
        "deal with": {
            "sentences": [
                "FPNN [12] and V ote3D [23] proposed special methods to deal with the sparsity problem.",
                "We deal with the commmunication efficiency for classification from another perspective with the proposed distributed framework.",
                "One way of dealing with this problem that has been gaining attention is the semantic annotation of table data columns by using canonical knowledge.",
                "Finally, the M3 translation rule ensures that other aspects of the protocol, such as the confirmation rule are adapted correctly to deal with the variable difficulty regime correctly.",
                "to cover language, image, and video at the same time for different scenarios, a 3d transformer encoder-decoder framework is designed, which can not only deal with videos as 3d data but also adapt to texts and images as 1d and 2d data, respectively.",
                "To deal with the limit for various downstream tasks, inputs are typically truncated into valid lengths or multiple instances for a single document are created",
                "This is done by preventing the model from over-fitting to domain-specific visual content, while simultaneously dealing with imbalanced semantic distributions between domains.",
                "They are thus not directly applicable to deal with the multi-modal input",
                "Positive-unlabeled (PU) learning deals with the binary classification problem when only positive (P) and unlabeled (U) data are available, without negative (N) data.",
                "In this paper, we propose an efficient algorithm that aims to address one of the major challenges in FL\u2014dealing with heterogeneity in the data distribution.",
                "proven to be more effective than LSTM in dealing with long-range text dependencies",
                "we overcome the limitations of previous studies by properly extending the tree structure when dealing with update operations",
                "Multidimensional point data could be represented in a variety of ways such as K-d tree which is preferable when dealing with dimensions greater than 3 [34] or point quadtree.",
                "ines and planes are extracted from the environment to deal with low/non-textured scenes",
                "In our model, we deal with resource constraints by shedding load in the form of dropping tuples from the data streams.",
                "\ufeffHowever, our method can deal with more complicated interactions with the tensor layer than the other methods.",
                "To deal with that, it is necessary to consider the effect both from micro-level and macro-level aspects.",
                "To deal with the scalability issue, we propose a new sampling scheme, a truncation method.",
                "Bogin et al. (2019a) and Chen et al. (2021) deal with the graph structure of database schema via GNN.",
                "Most prior work on differentially private machine learning deals with example-level privacy: two datasets d and d0 are defined to be adjacent if d0 can be formed by adding or removing a single training example from d",
                "Our model deals with long-tail or newly added relations and focuses on one-shot relational learning without any external information",
                "we deal with the MIL problem for WSI classification from another perspective with the proposed double-tier MIL framework.",
                "To deal with large dynamic datasets, Leopard [7] proposes an online partitioning method based on scoring. T"
            ]
        }
    },
    "decide": {
        "decide to": {
            "sentences": [
                "When current frame is decided to be a new keyframe, the observed 3D points will be recovered",
                "For example, if one self-driving car decides to turn left while others turn right for the same input, one of them is likely to be incorrect.",
                "She has decide to travel all over the world.",
                "We decide to make full use of the information from neighbors.",
                "As the virus runs, it must de- crypt itself to attempt to infect a file and there- fore the emulator can scan for the virus signature when the virus decides to execute.",
                "However, we have decided to retain such \"waiting\" hops in our measurements, because they represent distinct events in time and opportunities that arise."
            ]
        },
        "decide on": {
            "sentences": [
                "The adversary can decide on the spot how many honest parties are activated adaptively."
            ]
        }
    },
    "declare": {
        "declare that only": {
            "sentences": [
                "As seen in Figure 4, it involves the following steps: (1)  have the complete data up to LSN 1007, the database may have  receive log record and add to an in-memory queue, (2) persist  declared that only 900, 1000, and 1100 are CPLs, in which case,  record on disk and acknowledge, (3) organize records and identify  we must truncate at 1000."
            ]
        }
    },
    "decline": {
        "decline with": {
            "sentences": [
                "the performance does not decline with the changes"
            ]
        }
    },
    "decode": {
        "decode into": {
            "sentences": [
                "the systems always decode data into memory",
                "Finally, we integrate the fea- ture embeddings learned from both the graph encoding module and the latent memory unit, and then decode them into a human-like, coherent and informative story for the given image stream. The"
            ]
        },
        "decode from": {
            "sentences": [
                "it similarly decode from z to a sequence of tokens, using recurrent network."
            ]
        }
    },
    "decompose": {
        "decompose sth. into": {
            "sentences": [
                "We propose to train different embeddings with cause-specific data, and decompose the problem into four tasks."
            ]
        },
        "decompose into": {
            "sentences": [
                "We decompose the protection goal into following privacy definitions for dimension selection and value perturbation.",
                "In the context of SLAM, which typically [8] decomposes the problem into the visual feature extraction front-end and the bundle adjustment or pose estimation back-end, our network lies directly in the middle \u2013 SuperGlue is a learnable middle-end (see Figure 1)._x000D_",
                "CormCC decomposes a database into partitions according to workload access patterns and assigns a specific protocol to each partition, such that a protocol can be used to process the parts of a workload they are optimized for.",
                "every multivariate CDF could be\ndecomposed into its marginals Fi and a unique copula C",
                "The one-hop latency is decomposed into two parts."
            ]
        },
        "decompose input hierarchically": {
            "sentences": [
                "hierarchical transformer decomposes inputs hierarchically into elements of finer granularity."
            ]
        },
        "be decompose into": {
            "sentences": [
                "In cross-domain visual classification, the domain distribution can be decomposed into class-discriminative latent visual pattern distribution (p(x)) and conditional class distribution (p(yjx))."
            ]
        },
        "decompose xxx into": {
            "sentences": [
                "Recently, Wolfson et al. (2020) proposed the Question Decomposition Meaning Representation, QDMR, which is a way to decompose a question into a list of \u201catomic\u201d steps representing an algorithm for answering the question"
            ]
        }
    },
    "decouple": {
        "decouple from": {
            "sentences": [
                "decoupling storage from compute has become a trend for cloud computing industry."
            ]
        }
    },
    "decrease": {
        "decrease with": {
            "sentences": [
                "However, when time progresses, the JS Divergence decreases with time."
            ]
        },
        "decrease under": {
            "sentences": [
                "This comparison is not sensitive to literal query values and can decrease under semanticpreserving SQL query rewriting."
            ]
        },
        "decrease to": {
            "sentences": [
                "The foreground threshold is set to 0.9 and the data sampling ratio sr is set to 0.2 and gradually decreases to 0 over the last 10k iterations"
            ]
        },
        "decrease and eventually": {
            "sentences": [
                "17 shows, while computing with sparse tensorstorage formats incurs some performance penalty as compared to the same computation withdense formats when the inputs are highly dense, the performance penalty decreases and eventuallyturns into performance gain as the sparsity of the inputs increases."
            ]
        }
    },
    "decrypt": {
        "decrypt to": {
            "sentences": [
                "As the virus runs, it must de- crypt itself to attempt to infect a file and there- fore the emulator can scan for the virus signature when the virus decides to execute."
            ]
        }
    },
    "dedicate": {
        "be dedicate for": {
            "sentences": [
                "However, those methods are dedicated for pixel-level domain adaptation, assuming the domain gap is mainly the \u201cstyle\u201d difference."
            ]
        },
        "dedicate to": {
            "sentences": [
                "Core are dedicated to corresponding worker thread"
            ]
        },
        "dedicate sth. to sth.": {
            "sentences": [
                "NVLink 2.0 is a new interconnect technology that links dedicated GPUs to a CPU."
            ]
        },
        "dedicate their process power": {
            "sentences": [
                "Using ActiveMemory, all replicas dedicate their processing power to executing new transactions"
            ]
        }
    },
    "deem": {
        "deem as": {
            "sentences": [
                "Unbiased PU learning, a batch-mode PU learning algorithm, which can be deemed as the performance upper bound of our OPU algorithm."
            ]
        }
    },
    "defer": {
        "defer a more": {
            "sentences": [
                "\"Although we defer a more thorough discussion of Cloud Computing economics to Section 6"
            ]
        },
        "defer sth. to sth.": {
            "sentences": [
                "We defer giving further details on our measurement setup and methodology to Section 7.1."
            ]
        },
        "defer sth": {
            "sentences": [
                "We defer the description of our experiment environment until Section 7."
            ]
        },
        "defer to": {
            "sentences": [
                "Some less important plots are deferred to \u00a7C"
            ]
        }
    },
    "define": {
        "define as": {
            "sentences": [
                "AFLGo defines some vulnerable code as target locations, and optimally select testcase that are closer to target locations.",
                "The state of the t-th time step is de\ufb01ned as a tuple",
                "We define a soft label for category k as the average over the softmax of all activations of source examples.",
                "formally, three types of links are defined as follows2: 1) entity head 2 for ease of exposition, we use entity/subject/object head to represent the start position of one entity/subject/object.",
                "We suggest the proposal window w, defined as wfar \u2212 wclose + 1, to be at least four to ensure liveness (Sect. V-C).",
                "Under this threat model, security of a blockchain is defined as the safety property of the underlying consensus protocol",
                "In DensePoint [64], convolution is defined as a Single-Layer Perceptron (SLP) with a nonlinear activator.",
                "Rate adjustments at each changepoint can be defined as a vector \u03b4, where \u03b4 is the rate change at the j change-point."
            ]
        },
        "define by": {
            "sentences": [
                "The score of a fact (h, r, t) is defined by a bilinear function",
                "This measure is de\ufb01ned by a weighted combination of simple \u201cpath experts\u201d"
            ]
        },
        "define of": {
            "sentences": [
                "We define each voxel of size vD, vH, and vW accordingly."
            ]
        },
        "be define as": {
            "sentences": [
                "In this work, latency is defined as the end-to-end time from the time a query starts to the time it ends",
                "The client rate is defined as the total number of transactions submitted by all the clients per second",
                "Meta-path is defined as a composite relation linking two objects, which can be used to capture multi-hop semantics."
            ]
        },
        "define the failure recovery problem": {
            "sentences": [
                "We formally define the failure recovery problem in distributed graph processing systems and introduce a partition-based recovery method that can efficiently handle any node failures"
            ]
        },
        "define a functionally": {
            "sentences": [
                "2.1Graph Substitution DefinitionA graph substitution consists of three components: (1) a sourcegraph that is matched to subgraphs in a computation graph;(2) a target graph1 that defines a functionally equivalent newsubgraph to replace the matched subgraph; and (3) a mappingrelation between input/output tensors in the source and target graphs."
            ]
        },
        "be define by": {
            "sentences": [
                "As previously defined by Shokri et al., there are two sources of information that we intend to preserve for the local vector."
            ]
        },
        "define with": {
            "sentences": [
                "In this section, we formally define the active learning scenario with the proposed loss prediction module."
            ]
        }
    },
    "degrade": {
        "be degrade": {
            "sentences": [
                "The generality of the model may be severely degraded when facing the same task of a different domain."
            ]
        },
        "degrade sth by": {
            "sentences": [
                "These unfinished transactions may degrade performance by preventing other transactions from being successfully validated."
            ]
        }
    },
    "delay": {
        "delay in": {
            "sentences": [
                "network delay in which we inject arbitrary delays into messages"
            ]
        }
    },
    "delegate": {
        "delegate by": {
            "sentences": [
                "Knowledge representation learning (KRL) delegated by deep learning focuses on representation learning of entities and relations in the knowledge base"
            ]
        }
    },
    "delete": {
        "delete from": {
            "sentences": [
                "We delete ? from L and insert it in X while adding w? to W."
            ]
        }
    },
    "deliver": {
        "deliver to": {
            "sentences": [
                "These hard negative samples are delivered to user clients and mixed with the observed negative samples from local data",
                "this can effectively exchange information from local regions and delivered to each other in the next attention computation, while the implementation is easy yet efficient.",
                "Unlike SSD, it delivers higher bandwidth and lower latency to CPUs."
            ]
        },
        "be deliver to": {
            "sentences": [
                "Observe that notion of \u201crounds\u201d still exist in the model (since we consider discretized time), but now these are not synchronization rounds within which all messages are supposed to be delivered to honest parties."
            ]
        }
    },
    "delve": {
        "delve into": {
            "sentences": [
                "In this section, we will delve into the Structure-Aware Dual Graph Aggregation Network (SADGA), including Dual-Graph Construction, Dual-Graph Encoding and Structure-Aware Aggregation.",
                "We delved into the transferabil- ity of the model from the perspective on model-level explanation in the training process",
                "We now delve into the details of each part"
            ]
        }
    },
    "demonstrate": {
        "demonstrate by": {
            "sentences": [
                "attacks have been demonstrated by an adversary controlling only 25% of the nodes",
                "As demonstrated by recent learned indexes"
            ]
        },
        "demonstrate sth": {
            "sentences": [
                "We demonstrate significant gains from including multiple tables, and these gains are increasing as more tables are used."
            ]
        },
        "demonstrate that": {
            "sentences": [
                "For building the RSPNs, we use zero as the budget factorto demonstrate that even a base RSPN ensemble providesgood estimates after updates.",
                "The intense research and commercial interest in HTAP demonstrates that data owners have a strong desire to process queries and transactions.",
                "we demonstrate that COTR achieves state-of-the-art performance in both dense and sparse correspondence problems on multiple datasets and tasks",
                "We demonstrated that running an entire SQL query on a GPU delivers better performance than using the GPU as an accelerator.",
                "We further demonstrate that the learned embeddings successfully capture the desired causes."
            ]
        },
        "demonstrate on": {
            "sentences": [
                "We demonstrate the ability of our architecture to improve adaptation performance in the supervised and semi-supervised settings by experimenting with two standard domain adaptation benchmark datasets."
            ]
        },
        "these observations demonstrate that:despite numerous algorithmic and engineer advances,hash-based operations be still expensive": {
            "sentences": [
                ""
            ]
        },
        "demonstrate here": {
            "sentences": [
                "similarly, we demonstrate here that the way we train the flamingo models matters greatly for their final performance."
            ]
        }
    },
    "denote": {
        "denote as": {
            "sentences": [
                "The observed ratings in this sampled submatrix is denoted as I",
                "These relations form a meta-testing set which is denoted as T",
                "equation 1 is also denoted as handshaking kernel in figure 3."
            ]
        },
        "denote to": {
            "sentences": [
                "Hence, the input and output sequence can be separately denoted to X = (\ud835\udc65\ud835\udc50\ud835\udc59\ud835\udc60, \ud835\udc651, \ud835\udc652,"
            ]
        },
        "be denote as": {
            "sentences": [
                "while our method is denoted as \u201cFineStream\u201d",
                "Symbol from the top layer of BERT is denoted as Cij."
            ]
        },
        "denote with": {
            "sentences": [
                "The transactions are denoted with their frequency of occurrence and relationships to entities."
            ]
        },
        "denote sth": {
            "sentences": [
                "which denote the set of most-frequently-used access patterns"
            ]
        },
        "denote for": {
            "sentences": [
                "We will denote the time for M blocks."
            ]
        },
        "denote by": {
            "sentences": [
                "This is the value calculated in the last difficulty update, and we denote it by \u03b40."
            ]
        }
    },
    "depart": {
        "depart from": {
            "sentences": [
                "First of all, GPUs depart from the linear memory hierarchy of CPUs and adopt a fatter cache hierarchy"
            ]
        }
    },
    "depend": {
        "depend on": {
            "sentences": [
                "pop-up process depends on the camera pose",
                "We identify whether a certain kind of information (e.g., user/item embeddings) cross such privacy boundaries depending on the settings below",
                "Depending on the region the hash falls, the block is considered mined of a certain type.",
                "\ufeffThe network structure depends on the available \ufefftraining data.",
                "which is highly dependent on the application scenario and hard to predict.",
                "For example, automated data cleaning facilitates the generation of clean data through validation and transformation rules that depend on data type.",
                "We have tj depends on ti and ti depends on tj",
                "The effectiveness of the pre-trained model largely depends on the representation learning of the model\u2019s encoder",
                "Meanwhile it is possible that one data structure may have multiple unique signatures, depending on how (especially, how deep) the points-to edges are traversed when generating a signature.",
                "DP can either be enforced in a local or global sense depending on whether the server is trusted.",
                "Apparently, the upper limit of performance of the framework depends on the quality of the pseudo label, as we rely on the pseudo label to train the student model",
                "In MCAR data, the missingness of the variable cannot depend on the value of the variable.",
                "We propose a model that depends only on the entity embeddings and local graph structures",
                "That means the encoder and decoder simultaneously depend on the uncertain data.",
                "How does the performance of generic and of workflow-specific autoscalers depend on workflow-based workload characteristics?",
                "These solutions depend on deep ETL pipelines and physical tuning that fundamentally assume predictable, slow-moving, and easily categorized data from largely internal sources.",
                "\"the solution quality depends on well-chosen restrictions di\ufb00erent algorithms can perform best.\"",
                "the collision probability of the model depends on how well the model fits the data.",
                "That depends on how model inputs interact with the context.",
                "feature-based works (yu and lam, 2010; miwa and sasaki, 2014) need a complicated process of feature engineering and heavily depend on nlp tools for feature extraction.",
                "The observable order of memory operations depends on the memory consistency model.",
                "Ae,Be,Ce,De are constants that depend on the distributional information of the tables in v1 and v2.",
                "First, conformity depends on both user and item.",
                "The correct choice of an autoscaler is important but significantly depends on the application type."
            ]
        }
    },
    "depict": {
        "depict sth": {
            "sentences": [
                "Fig. 1 depicts a simplified example in which systems are expected to decide whether the facts in the table support the natural language statement."
            ]
        },
        "depict in": {
            "sentences": [
                "overall, as depicted in figure 3, the joint extraction task is deconstructed into 2n+1 sequence labeling subtasks where n denotes the number of pre-defined relation types, each subtask builds a tag sequence of length n2+n 2 , where n is the length of the input sentence.",
                "though not depicted in figure 2, there is an additional level of indirection between the edge index and tels, called label index blocks, used to separate the storage of per-vertex edges with different labels."
            ]
        },
        "depict two hierarchically-compressed coefficients": {
            "sentences": [
                "4.10 Example Haar-tree structure depicting two hierarchically-compressed coefficients."
            ]
        },
        "be dipicted visually": {
            "sentences": [
                "This effect is depicted visually  in Figure1."
            ]
        }
    },
    "deploy": {
        "deploy software originally": {
            "sentences": [
                "It can also deploy software originally developed for on-premise use completely on the public cloud with little modi\ufb01cation."
            ]
        },
        "deploy on": {
            "sentences": [
                "Contracts deployed on a Blockchain network can send messages to other contracts",
                "Through experiments deployed on, or with data collected from, the Bitcoin network, we unveil the detailed mechanism of how the block size and interval affect NC\u2019s security through the existence of fresh transactions.",
                "The choice of FPFH is based on the required computational efficiency, given the necessity for real-time computation if deployed on an actual UAV"
            ]
        },
        "deploy with relatively": {
            "sentences": [
                "The supervised learning approaches are easy to deploy with relatively low training and prediction overhead, however, they lack robustness to shifting query workloads."
            ]
        },
        "be deploy on": {
            "sentences": [
                "H-Store is deployed on a cluster of 16 compute nodes with the database being partitioned evenly"
            ]
        },
        "be deploy in": {
            "sentences": [
                "The copy instance is deployed in the user's environment"
            ]
        },
        "deploy sth": {
            "sentences": [
                "We deployed AUTOINDEX in the open\u0002sourced database system openGauss"
            ]
        },
        "deploy in": {
            "sentences": [
                "Deep learning (DL) systems are increasingly deployed in safety- and security-critical domains including self-driving cars and malware detection.",
                "All three platforms support smart contracts and can be deployed in a private environment"
            ]
        },
        "deploy as": {
            "sentences": [
                "chaincodes are deployed as Docker images interacting with Hyperledger\u2019s backend via pre-defined interfaces"
            ]
        }
    },
    "deprive": {
        "deprive of": {
            "sentences": [
                "In previous sample selection approaches, once selected, a trusted example will never be deprived of its label during the subsequent training."
            ]
        }
    },
    "derive": {
        "derive from": {
            "sentences": [
                "There are a number of implications arising from the metrics and insights we have derived from our temporal graph analysis.",
                "The account address is derived from the public key",
                "This validate our Theorems 1, 4 and Corollaries 2, 3 are generic for any local randomizer, as long as the amplification bound can be derived from a close-form solution or numerical evaluations",
                "A minimum spanning tree connecting all keyframes is derived from the covisibility graph",
                "The method that merge last two steps into one derives from Silo",
                "Normally, the transferable knowledge across domains are derived from some certain data, while others contribute less and are costly to be learned from (Axelrod et al., 2011; Ruder and Plank, 2017).",
                "Therefore, the signature derived from the mainstream kernel cannot remain in the target kernels.",
                "A bunch of inequalities are derived from our causal model.",
                "The accumulation in step(i) and the momentum in step(iii) derive from the work of Sun et al.",
                "Without loss of generality, we assume these observations\nsatisfy Gaussian distribution (Salinas et al. 2019) whose\nmean and variance are derived from state h\u03c4, which is conductive to preserve their own specialty.",
                "In the image matching problem, some regularities of the world could be leveraged: the 3D world is largely smooth and sometimes planar, all correspondences for a given image pair derive from a single epipolar transform if the scene is static, and some poses are more likely than others.",
                "we implement a strong baseline that is derived from the VoxelNet architecture but uses hand-crafted features instead of  the proposed feature learning network",
                "The difficulty D of a new block is derived from the estimated hash rate.",
                "a comparison of fractional differ\u0002ences between each system and an ideal system derived from the experimental results.",
                "We derive our training set from various structured data collections",
                "To understand how our model derives improvement from the addition of more tables, we compute in Table 3 the performance of our reranking-and-verification model when TF-IDF returns the correct table at rank 1, rank 2-3, or rank 4-5.",
                "The objective is derived from flow constraints obtained when introducing a third image, constructed by randomly warping one of the images in the original pair.",
                "A qualitative nearest-neighbor analysis of embeddings derived from TABBIE confirms that it encodes complex semantic properties about textual and numeric cells and substructures.",
                "The penalties are derived from the class prior."
            ]
        },
        "derive in": {
            "sentences": [
                "The lack of a method for comparing autoscalers derives in our view from scientific and industry practice."
            ]
        },
        "derive by": {
            "sentences": [
                "Thus, we try to\nconstruct a series-level connection based on the process similarity derived by series periodicity."
            ]
        },
        "be derive from": {
            "sentences": [
                "Existing video domain adaptation methods mainly rely on adversarial feature alignment, which has been derived from the RGB image space."
            ]
        },
        "derive xxx from": {
            "sentences": [
                "For example, semantic parsing (Zelle and Mooney, 1996; Zettlemoyer and Collins, 2005) converts natural language questions into formal programs; knowledge-base question answering (Berant et al., 2013) derives answers from tables or knowledge graphs."
            ]
        },
        "derive for": {
            "sentences": [
                "We derive metric equations for three core use cases of Entropy-learned Hashing."
            ]
        },
        "derive sth": {
            "sentences": [
                "We may derive different candidate indexes from those forms"
            ]
        },
        "deriv from": {
            "sentences": [
                "To directly derive real-world knowledge from unstructured text"
            ]
        },
        "derive from conservatively": {
            "sentences": [
                "(cid:2) is derived from conservatively applied worst-case local clock drift."
            ]
        },
        "derive  accurately": {
            "sentences": [
                "Clearly,  this  truncation  scheme  can  lead  to  inefficiencies  in  reconciliation as the descendant relationships cannot be derived  accurately."
            ]
        }
    },
    "describe": {
        "describe in": {
            "sentences": [
                "as described in section 3.1, the gremlin server is a layer on top of a tinkerpop3 implementation and enables multiple clients to communicate with the same data base in a platform-agnostic manner."
            ]
        },
        "describe  above": {
            "sentences": [
                "this is then used to pre- dict a grid which is then fed as an additional channel in to the base cnn architecture described  above."
            ]
        },
        "describe here also": {
            "sentences": [
                "In fact, many of the sketches we will describe here also apply to themore general case where each f (i) can take on arbitrary real values,and even negative values."
            ]
        },
        "describe several widely": {
            "sentences": [
                "The contributions of this paper are:tensor storage formats that separately designate each dimension as dense or sparse and specifythe order in which dimensions are stored, which can describe several widely used formatsbut generalizes to many more (Section 3);iteration graphs that describe how to co-iterate through the multi-level index data structures ofthe sparse operands in a compound tensor expression (Section 4);merge lattices that describe how to merge the index data structures of sparse tensors that areused in the same tensor expression (Section 5); and acode generation algorithm that uses the above concepts to generate efficient code that computesa tensor expression with dense, sparse, and mixed operands (Section 6)."
            ]
        },
        "describe so far": {
            "sentences": [
                "The algorithm described so far is generic, in the sensethat it does not depend on the specific tensor operators used."
            ]
        },
        "describe attribute": {
            "sentences": [
                "the role of this model is to extract semantic spatial features that describe attributes that one would want to query about a visual datum: color, shape, nature, positions of objects, etc."
            ]
        },
        "describe sth": {
            "sentences": [
                "We now describe how a WAL-based DBMS processes and commits transactions."
            ]
        },
        "describe of": {
            "sentences": [
                "which describes the occurrence of things in a definite temporal"
            ]
        },
        "describe as": {
            "sentences": [
                "The paradigm is described as vanilla FMF in Figure 3(a)"
            ]
        },
        "be describe by": {
            "sentences": [
                "In particular, we will argue that the mining of blocks can be described by an inhomogeneous Poisson process."
            ]
        },
        "describe with": {
            "sentences": [
                "In addition, we were interested in how clearly these practices were described with respect to particularly sensitive information types"
            ]
        },
        "describe  earlier": {
            "sentences": [
                "Note  VDL  (the  10  million  limit  described  earlier)."
            ]
        }
    },
    "design": {
        "design to": {
            "sentences": [
                "Our MetaR is designed to \ufb01ll this vulnerability of existing embedding models",
                "as many general autoscalers are designed to stably operate when VM booting times do not exceed a certain threshold.",
                "oint features often encode certain statistical properties of points and are designed to be invariant to certain transformations, which are typically classified as intrinsic [2,21,3] or extrinsic [18,17,13,10,5].",
                "The analytical model is designed to automatically determine these factors accurately without sacrificing performance.",
                "tensor ow was designed to ef ciently run larger models, not small models, and thus, has a signi cant invocation overhead, especially with python as the front-end.",
                "task-specific ontologies are also designed to add reasoning capabilities based on static knowledge",
                "Larger main memory sizes drive database systems from disk-based design to in-memory design.",
                "Blockchain\u2019s key advantage is that it does not assume nodes trust each other and therefore is designed to achieve Byzantine fault tolerance"
            ]
        },
        "design for": {
            "sentences": [
                "However, the challenge is how to design Top-k selection mechanisms under LDP for federated SGD with better selection strategies.",
                "Four of these use cases are explicitly designed for IoT",
                "As an interesting by-product, the proposed methodology achieves state-of-the-art performance on Incremental Object Detection, even though not primarily designed for it",
                "The outdoor urban datasets are designed for autonomous driving, where objects are spatially well separated and these point clouds are sparse.",
                "It is designed for the simpler task of predicting if a fact is true or not",
                "However, as mentioned in the introduction, all above methods are only designed for batch learning mode and cannot be applied to deal with sequential data.",
                "While originally designed for Silo OCC, we extend it to 2PL and PartCC.",
                "Besides, designing pretraining tasks for tables and pretraining BERT on a large table collection could be promising to further improve the performance of BERT on table-related tasks such as table retrieval.",
                "Early PU works designed heuristics for sample selection."
            ]
        },
        "be design to": {
            "sentences": [
                "Homographic Adaptation is designed to enable self-supervised training of interest point detectors. It warps the input image multiple times to help an interest point detector see the scene from many different viewpoints and scales (see Section 5).",
                "As GPUs are designed to handle such high-latency memory accesses, they are well-equipped to cope with the additional latency of NVLink 2.0.",
                "Third, the I/O model of the PolarFS data plane is also designed to eliminate locks and avoid context switches on the critical data path",
                "APEX is carefully designed to address the challenges.",
                "Here the domain discriminator is designed to predict the domain (source/target) of each input sample and is trained in an adversarial fashion, similar to GANs, for learning domain invariant representation."
            ]
        },
        "design a more": {
            "sentences": [
                "Whentheswitch scale clusters; we need to design a more compact copyset (cid:193) processesacachecoherencerequestandmanagestoacquire formatlikecoarsevectors[31]."
            ]
        },
        "be design to be": {
            "sentences": [
                "The VLL protocol is designed to be as general as possible, with specific optimizations for the following architectures"
            ]
        },
        "design the reward": {
            "sentences": [
                "At night ,The scenery paragraphs,we design the rewards with two critic networks, was nice and it would be great if I can visit here again in the future."
            ]
        },
        "be design towards": {
            "sentences": [
                "Federated Learning (FL) is a recently proposed distributed computing paradigm that is designed towards this goal."
            ]
        },
        "be design for": {
            "sentences": [
                "Encodingschemes are designed for datasets with particular characteristics"
            ]
        },
        "be design for sth": {
            "sentences": [
                "That is designed for a hybrid storage hierarchy with NVM and DRAM."
            ]
        },
        "design fo": {
            "sentences": [
                "This fact motivates our task-agnostic design for active learning."
            ]
        },
        "design to reliably": {
            "sentences": [
                "We describe Bigtable is designed to reliably scale to petabytes of several examples of how Bigtable is used at Google dataandthousandsofmachines."
            ]
        }
    },
    "designe": {
        "design": {
            "sentences": [
                "We designed QReCC to overcome both of these limitations."
            ]
        }
    },
    "desire": {
        "desire to": {
            "sentences": [
                "This practice has greatly concerned software companies who desire to protect the in- tellectual property of their products."
            ]
        }
    },
    "detach": {
        "detach from": {
            "sentences": [
                "This is because the model weight of the pseudo-label generation model (Teacher) is detached from the optimized model (Student)"
            ]
        }
    },
    "deteriorate": {
        "deteriorate with": {
            "sentences": [
                "because insert performance will deteriorate with fewer gaps."
            ]
        },
        "deteriorate significantly": {
            "sentences": [
                "However, the accuracy of these estimators will deteriorate significantly in the presence of attributes of large domain size (e.g., continuous attributes), as the search space in query inference will be very large."
            ]
        }
    },
    "determine": {
        "determine to": {
            "sentences": [
                "He determine to learn French."
            ]
        },
        "determine into": {
            "sentences": [
                "The history window determines the sensitivity of the algorithm to local versus global trends, while the size of the adaptation window determines how far into the future the model extends."
            ]
        },
        "determine by": {
            "sentences": [
                "Second, the number of partitions is no longer a constant, but a variable that is determined by both the workload and the number of cores available",
                "The output of the spherical convolutional kernel for a point is determined by the non-linear activation of the mean of weighted activation values of its neighboring points.",
                "3.1 handshaking tagging scheme 3.1.1 tagging according to the insight that a triplet (s, r, o) can be determined by aligning the boundary tokens of subject entity s and object entity o conditioned on the relation r, we realize one-stage joint extraction by tagging token pairs with link labels."
            ]
        },
        "be determine by": {
            "sentences": [
                "Overall, throughput is determined by the proportion of accesses to a given processor."
            ]
        },
        "determine sth": {
            "sentences": [
                "The DBMS must ignore the effects of transactions that fall within such a gap while determining the tuple visibility."
            ]
        }
    },
    "develop": {
        "develop in": {
            "sentences": [
                "Models such as those developed in Neelakantan et al. (2015); Das et al. (2017) use the same set of initially collected paths to answer a diverse set of query types"
            ]
        },
        "develop largely independently": {
            "sentences": [
                "further, these works have been developed largely independently without common terminology or even a common problem state- ment."
            ]
        },
        "develop skills": {
            "sentences": [
                "For example, what 809 we see through our eyes is registered in our brains; what we read, develops not only our 810 language facilities, but various other skills."
            ]
        },
        "develop from": {
            "sentences": [
                "Our prototype develops from Hper-V"
            ]
        },
        "develop a more": {
            "sentences": [
                "(MLPerf is developing a more representative recommendation benchmark forthe next submission round [69], [70])."
            ]
        },
        "develop to automatically": {
            "sentences": [
                "2006], which is a compilerframework developed to automatically optimize dense tensor contractions (multiplications) in theNWChem software."
            ]
        },
        "develop to": {
            "sentences": [
                "we develop a method to port certain Paxos optimizations",
                "R-GCN [54] is developed to deal with the highly multi-relational data characteristic of realistic knowledge bases"
            ]
        }
    },
    "develope": {
        "be develop with": {
            "sentences": [
                "Recently, deep models have been developed with promise in unsupervised domain adaptation to learn expressive features."
            ]
        }
    },
    "deviate": {
        "deviate from": {
            "sentences": [
                "one common challenge is that the generated video frames tend to gradually deviate from the text prompt, making the generated characters hard to perform the desired actions.",
                "As mentioned in Section 3, the parameter h is usually estimated from data in practical situations, so it might be slightly deviated from its real value.",
                "Select, Project, and Join on both CPU and GPU to understand when the ratio of operator runtime on CPUs to runtime on GPUs deviates from the ratio of memory bandwidth of GPU to memory bandwidth of CPU."
            ]
        }
    },
    "devide": {
        "devide into": {
            "sentences": [
                "We divide these methods into two categories based on their semantics, push-based and pull-based."
            ]
        }
    },
    "devise": {
        "devise to": {
            "sentences": [
                "There are many different types of transforma- tions that can be devised to obscure the opera- tions performed on an array.",
                "There are many different types of transformations that can be devised to obscure the operations performed on an array."
            ]
        }
    },
    "devote": {
        "devote to": {
            "sentences": [
                "he rest of this subsection is devoted to prove that (iii+) is preserved by (4).",
                "In this paper, we devote to discuss few-shot link prediction in knowledge graphs",
                "Considering the serious damages caused by vulnerabilities, much effort has been devoted to vulnerability discovery techniques towards software and information systems.",
                "Extensive effort has been devoted to the development of unsupervised representation learning methods for both unstructured text and structured knowledge bases, which in turn can be utilized for a wide variety of downstream tasks via fine-tuning."
            ]
        },
        "be devote to": {
            "sentences": [
                "Only very few efforts have been devoted to respectively bridging the domain shift and semantic gap for cross-domain FER, but they fail to jointly narrow these two discrepancies in one unified learning framework, yielding unsatisfied performance."
            ]
        },
        "devote to formally": {
            "sentences": [
                "In the compiler literature, significant effort has been devoted to formally verifying optimizations [7, 11, 13, 23, 29, 30, 33, 35]."
            ]
        }
    },
    "differ": {
        "differ from": {
            "sentences": [
                "WBL differs from WAL in many ways.",
                "PU learning differs from the former in that it explicitly incorporates unlabeled data into the learning process.",
                "\ufeffJoint Model Pairwise differs from ASR as it concatenates the embeddings of the (q; ci), instead of using max-pooling, and does not use any Answer Support Classifier (ASC).",
                "We propose the notion of dynamic properties in few-shot KG completion, which differs from previous paradigms by studying the dynamic nature of entities and references in the few-shot scenario",
                "Moreover, SigGraph differs from global memory mapping-based approaches that have to start from global variables and require reachability to all data structure instances from them.",
                "Even if a historical graph differs from the current graph or one of the materialized graphs in only a few elements, we would have to set the corrsponding bit appropriately for all the elements in the graph.",
                "Our work differs from Callas and Tebaldi in that our mixed concurrent control execution does not introduce any extra concurrency control overhead, which greatly reduces the mixing overhead."
            ]
        },
        "differ in": {
            "sentences": [
                "Previous unsuper- vised learning methods can differ substantially in model sizes",
                "The latest works of this kind differ in the ways to generate the weighting values.",
                "The presented autoscalers differ in a way they use workflow structural information and task runtime estimates.",
                "CormCC differs in that our framework is more general and can be extended to more protocols.",
                "The value perturbation algorithms for each pair of comparison in Fig.2(h) is the same and only differ in the selection stage.",
                "An alternative is to define X and X\u2032 as neighboring if they have the same number of rows and differ in at most one row",
                "the latest works of this kind differ in the ways to generate the weighting values.",
                "A and B are two images that are almost identical to each other but differ in very subtle details."
            ]
        },
        "differ on": {
            "sentences": [
                "because arrival order may differ on different shard servers, care must be taken to ensure atomicity and isolation."
            ]
        }
    },
    "different": {
        "different from": {
            "sentences": [
                "Different from [5], we increase the number of height channels from 4 to 16 to capture more detailed shape information\u2013 further increasing the number of height channels did not lead to performance improvement.",
                "Different from leveraging lexical-similar entity pairs"
            ]
        },
        "be different from": {
            "sentences": [
                "The M2 monotonicity rule ensures that blocks in non-pivot chain do not refer to stale/old pivot blocks with target which is very different from expected in the present round."
            ]
        }
    },
    "differentiate": {
        "differentiate between": {
            "sentences": [
                "Such a feature helps differentiate between types that contain more repeated values, such as gender, from types that contain many unique values, such as name."
            ]
        }
    },
    "dig": {
        "dig out": {
            "sentences": [
                "Therefore, it is crucial for both normal users and security analysts to quickly profile the behavior of an IoT device and dig out services that are open or halfopen to attackers"
            ]
        },
        "dig into": {
            "sentences": [
                "In this section, we dig deeper into TABBIE's representations by comparing them to TaBERT across a variety of quantitative and qualitative analysis tasks, including our own pretraining task of corrupt cell classification, as well as embedding clustering and nearest neighbors."
            ]
        }
    },
    "diminish": {
        "diminish with": {
            "sentences": [
                "We find that the central DP level depends on the dimension d. Intuitively, from the view of privacy, the amplification effect is diminished with a large d"
            ]
        }
    },
    "disagree": {
        "disagree on": {
            "sentences": [
                "Even experts and trained analysts may disagree on their interpretation"
            ]
        }
    },
    "discover": {
        "discover the best": {
            "sentences": [
                "RCN discovers the best longer sequences."
            ]
        }
    },
    "discuss": {
        "discuss sth": {
            "sentences": [
                "We then discuss logging and recovery implementations in modern DBMSs.",
                "We discuss two methods to create pre-training data that lead to stronger table entailment models."
            ]
        },
        "be discuss": {
            "sentences": [
                "This process is discussed further in Section 3.1"
            ]
        },
        "as discuss in xxx": {
            "sentences": [
                "As discussed in Section 2.1, GRAPPA is also pre-trained on human annotated questions over tables with a MLM objective."
            ]
        },
        "as discuss earlier": {
            "sentences": [
                "As discussed earlier only few models have been created to approach the task of Video Dialog."
            ]
        },
        "discuss the most recent techniques": {
            "sentences": [
                "We briefly discuss the most recent or the most representative state-of-the-art techniques 1217 that are related to these tasks."
            ]
        },
        "discuss with": {
            "sentences": [
                "discuss its relationship with existing research areas in Section II. Section III presents the related theories in domain generalization."
            ]
        },
        "discuss on": {
            "sentences": [
                "We discuss more on sampling in Section 6.2.2",
                "The vulnerability of selfish mining has recently been discussed on several existing blockchain projects with variable difficulty in [19]."
            ]
        }
    },
    "disentangle": {
        "disentangle into": {
            "sentences": [
                "which tries to disentangle the features into domain-shared or domainspecific parts for better generalization."
            ]
        }
    },
    "display": {
        "display by": {
            "sentences": [
                "We display the number of instances selected by the four measurements in Figure 2."
            ]
        },
        "display on": {
            "sentences": [
                "The annotation tool, shown in Figure 1, displays a scrollable pri- vacy policy on the left and one annotation question with multiple response options on the right."
            ]
        }
    },
    "disregard": {
        "disregard for": {
            "sentences": [
                "in general, these terms do not necessarily need to vanish, and likely would further improve the expressiveness of the gradient update, but we disregard them here for the sake of the simplicity of the derivation."
            ]
        }
    },
    "distill": {
        "distill from": {
            "sentences": [
                "We design APEX with a set of principles distilled from the unique properties of PM and learned indexes",
                "Corollary 2 distills the amplification from \u03f5l to \u03f5c"
            ]
        }
    },
    "distinguish": {
        "distinguish into": {
            "sentences": [
                "The Authorship relations are further distinguish into three sub-types"
            ]
        },
        "distinguish a from b": {
            "sentences": [
                "In the application of VAROPTk,k+1 we do not distinguish between items from the previous reservoir and the new item."
            ]
        },
        "distinguish with": {
            "sentences": [
                "it can distinguish a future stock state that\nis inconsistent with the previous context."
            ]
        },
        "distinguish between": {
            "sentences": [
                "The goal of binary classification is to learn a model that is able to distinguish between positive and negative examples."
            ]
        }
    },
    "distribute": {
        "distribute on": {
            "sentences": [
                "The size of the small, the medium, and the large workflows is uniformly distributed on the intervals."
            ]
        },
        "distribute over": {
            "sentences": [
                "Large-scale graphs often have to be distributed over multiple compute nodes to enable in-memory processing."
            ]
        },
        "distribute to": {
            "sentences": [
                "Updated parameters are further distributed to all clients to update local model",
                "Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page."
            ]
        },
        "be distribute as": {
            "sentences": [
                "Furthermore, it is obvious that CAT is compatible to the marginal distribution alignment methods on the tasks where the source data is similarly distributed as the target data."
            ]
        },
        "distribute the recomputation task": {
            "sentences": [
                "Furthermore, we distribute the recomputation tasks for the subgraphs in the failed nodes among multiple compute nodes to achieve better parallelism"
            ]
        },
        "be distribute to": {
            "sentences": [
                "The whole query execution on a batch is distributed to a device",
                "The generated global model is finally distributed to each client and may be further trained to iterate the above steps until convergence."
            ]
        },
        "be distribute among": {
            "sentences": [
                "A volume is divided into chunks which are distributed among ChunkServers"
            ]
        },
        "distribute throughout": {
            "sentences": [
                "Descriptions of a particular data practice may be distributed throughout a policy."
            ]
        },
        "be distribute in": {
            "sentences": [
                "Though the targets are uniformly distributed in the feature space, the distance between two targets can vary widely"
            ]
        },
        "distribute  logically": {
            "sentences": [
                "4.8.2  External Discovery  The mechanism described above could temporarily result in a  Operational experience has shown that this approach distributes  logically  partitioned  Dynamo  ring."
            ]
        }
    },
    "disturb": {
        "disturb with": {
            "sentences": [
                "we disturb the orientation estimation with 1-deg Gaussian noise"
            ]
        }
    },
    "dive": {
        "dive into": {
            "sentences": [
                "To further dive into the effectiveness of work\u0002load characterization"
            ]
        }
    },
    "diversify": {
        "diversify sth": {
            "sentences": [
                "That model is trained with numerous historical index management data and can efficiently diversify the costs of queries under different indexes"
            ]
        }
    },
    "divide": {
        "divide into": {
            "sentences": [
                "We divide models into token-based models and context-based models",
                "Different from [Kiryo et al., 2017], which divided the data into animal and non-animal, we pick only one category from all the ten categories as P , and treat all other data as N.",
                "To capture diverse user interests, we divide user\u2019s clicked news into different interest clusters",
                "Figure 2 shows a sample configuration, where the data store is divided into four clusters such that the records co-accessed by transactions (denoted as solid lines connecting records) across clusters are minimized.",
                "We show that our method, which does not divide the pipeline into multiple stages and is learned end-to-end, can outperform these sparse methods.",
                "We now briefly introduce the limitations of existing innovative protocols, which are further divided into two approaches.",
                "The conversation can be roughly divided into two parts",
                "Previous work on one-shot learning can be divided into two groups",
                "HTAP solutions are divided into \u201cSingle System for OLTP and OLAP\u201d and \u201cSeparateOLTP and OLAP Systems.\u201d",
                "We divide training data into cause-specific parts, and train different embeddings with cause-specific data",
                "Methods using 2D representations of RGB-D data can be divided into two classes",
                "To ensure consistent evaluation across benchmarks, we divided the data into 60/20/20 training/validation/testing splits.",
                "We divide the experiments into two parts, domain adaptation and domain generalization.",
                "These methods can be roughly divided into two categories.",
                "The overall system is divided into four modules",
                "According to the data type of input for neural networks, existing 3D shape classification methods can be divided into multi-view based, volumetric-based and point-based methods.",
                "Existing in-memory hash join algorithms can be divided into two camps.",
                "the approaches of vision-language pre-training can be divided into two categories.",
                "Hierarchical architectures typically divide the infrastructure into multiple clusters, placing an engine (Level 1) in each cluster.",
                "The Stratified-Universe-Bernoulli Sampling Scheme applies to a table T that is divided into K groups (i.e., strata), denoted as G1,G2,...,Gk.",
                "AFLFast divide the paths into high-frequent ones and low-frequent ones."
            ]
        },
        "divide by": {
            "sentences": [
                "4.7 analysis to assess the effectiveness of docunet in modeling global information for multiple entities, we evaluated models respec- tively trained with or without u-shaped segmentation module on different groups of development set in docred, which are divided by the number of entities.",
                "We take the length of the cross sectional diagonal of the voxel, divide it by two and set that as the radius."
            ]
        },
        "be divide into": {
            "sentences": [
                "Existing MDA methods can be roughly divided into two categories: Adversarial Learning and Co-training.",
                "the database nodes of POLARDB are divided into two types:",
                "In the Spider dataset, the samples are divided into four difficulty groups based on the number of components selections and conditions of the target SQL queries.",
                "Thread blocks are further divided into groups of threads called warps (usually consisting of 32 threads)."
            ]
        },
        "divide ... into ...": {
            "sentences": [
                "We divide the labeled dataset at random into training and test, on the basis of patient identities"
            ]
        },
        "divide a ually": {
            "sentences": [
                "For this tiled loop, it divides a ually improved to match the target hardware."
            ]
        },
        "be divide by": {
            "sentences": [
                "we use bandwidth, defined as the transmitted data size divided by the execution time"
            ]
        }
    },
    "do": {
        "do quite well": {
            "sentences": [
                "BayesNet also does quite well, nearly matching supervisedapproaches, but is still significantly outperformed by Narudue to the former uses of lossy discretization and conditional independence assumptions."
            ]
        },
        "do by": {
            "sentences": [
                "this is done by constructing a messenger (msg) token in each local window."
            ]
        },
        "do so": {
            "sentences": [
                "to do so, we freeze the weights of these models so that their initial capacity remains unchanged.",
                "Importantly, while doing so, we disallow self attention among the query points in order to ensure that they are solved independently.",
                "by doing so, flamingo models gain strong generative language abilities and access to a large amount of knowledge stored in the lm weights."
            ]
        },
        "do nothing else": {
            "sentences": [
                "If a database system does nothing else, it must satisfy the contract  3 AZs with 2 copies of each item in each AZ."
            ]
        }
    },
    "docred": {
        "docred with": {
            "sentences": [
                "finetune bert for docred with two-step process."
            ]
        }
    },
    "dominate": {
        "dominate by": {
            "sentences": [
                "The field of forecasting was traditionally dominated by statistical\ntechniques."
            ]
        },
        "be dominate by": {
            "sentences": [
                "This dataset contains images that on average have 1059 4 unique object classes and this dataset is not overly dominated by commonly occurring 1060 classes such as person, car etc."
            ]
        }
    },
    "double": {
        "double in": {
            "sentences": [
                "further, the number of channels is doubled in each down-sampling block."
            ]
        }
    },
    "dovetail": {
        "dovetail with": {
            "sentences": [
                "This dovetails with the previous suggestion in that knowledge base completion is inherently a relational problem."
            ]
        }
    },
    "download": {
        "be download at": {
            "sentences": [
                "The pre-trained embeddings can be downloaded at https://huggingface.co/Salesforce/grappa_large_jnt."
            ]
        },
        "download from": {
            "sentences": [
                "Item dataset I is downloaded from server"
            ]
        }
    },
    "downsample": {
        "downsample to": {
            "sentences": [
                "Spatially sparse convolutional networks are introduced for LiDAR-based detection and are used to extract information from the z-axis before the 3D data are downsampled to  something akin to 2D image data."
            ]
        }
    },
    "drag": {
        "drag down": {
            "sentences": [
                "That means the encoder and decoder simultaneously depend on the uncertain data (we call it model\nuncertainty), which may drag down the improvement from\ntraining examples to unknown validation examples.",
                "However, the discrete memory architectures with relatively low PCI-e bandwidth and high latency have dragged down the benefits of heterogeneous coprocessors."
            ]
        }
    },
    "draw": {
        "draw from": {
            "sentences": [
                "These methods are usually learned on centralized user data, where abundant and representative negative samples can be drawn from the global dataset",
                "as a result, the predicted tokens at training and inference are drawn from different distributions, namely, from the data distribution as opposed to the model distribution(zhang et al., 2019).",
                "where \u2206x, \u2206y, \u2206z are drawn independently from a Gaussian distribution with mean zero and standard deviation 1.0.",
                "The scaling noise was drawn from the uniform distribution",
                "d dimensions of the gradient g is reduced to q dimensions by multiplying the vector g with a public random matrix drawn from the Gaussian distribution with mean 0 and variance 1/q."
            ]
        },
        "draw on": {
            "sentences": [
                "In order to generate high-quality pseudo boxes and to facilitate the training of the student model, we draw on FixMatch [26] which is the latest advancement in semi-supervised image classification task"
            ]
        },
        "draw its name": {
            "sentences": [
                ""
            ]
        },
        "draw randomly": {
            "sentences": [
                "consistent with previous works, we report zero-shot fid-30k, for which 30k prompts are drawn randomly from the validation set, and the model samples generated on these prompts are compared with reference images from the full validation set."
            ]
        }
    },
    "drill": {
        "drill in\u704c\u8f93": {
            "sentences": [
                "To start drilling in at both the source of the problem and our problem."
            ]
        }
    },
    "drive": {
        "drive by": {
            "sentences": [
                "In fact, those interactions are mainly driven by users conformity."
            ]
        },
        "data-driven": {
            "sentences": [
                "Different from DeepDB , additional representations (lineageAACs) have to be constructed on a per-query basis whereas RSPNs are data-driven and thus workload-independent."
            ]
        }
    },
    "drop": {
        "drop in": {
            "sentences": [
                "Just drop in any lime during the data collection.",
                "Just drop in any lime during the day.",
                "All query plans can use it as a drop-in replacement for the non-partitioned hash join used otherwise."
            ]
        },
        "drop off": {
            "sentences": [
                "In this section, we present the DISC-VIO in which the robot-to-robot cross-correlations are conservatively dropped off via covariance intersection."
            ]
        },
        "drop only slightly": {
            "sentences": [
                "Our results in Table 2 show that with a higher fraction of updates,the accuracy drops only slightly."
            ]
        },
        "drop exponentially": {
            "sentences": [
                "It continues to drop exponentially fast with more entropy."
            ]
        },
        "drop out": {
            "sentences": [
                "A user is said to drop out if either of the two cases happens"
            ]
        }
    },
    "due": {
        "due to": {
            "sentences": [
                "This avoids the generation of too many active locations, which can lead to a decrease in speed in subsequent convolution layers due to the large number of active points.",
                "Due to its importance in industrial, society and scientific applications. For instance.",
                "Due to these limitations and uncertainty brought by such radical innovations, NC and its variants remain the foundation of most of the leading cryptocurrencies such as Bitcoin, Litecoin [67], Ethereum [12],Bitcoin Cash [7] and Zcash [76].",
                "due to a distribution shift",
                "Due to the diversity and complexity of the real world",
                "Due to the lack of easy to use vulnerability detection model, static analysis tools are prone to a large number of false positives.",
                "due to factors such as non-uniform  sampling of the 3D space, effective range of the sensors,occlusion, and the relative pose",
                "A major breakthrough in recognition [20] and detection [13] tasks on images was due to moving from hand-crafted features to machine-learned features.",
                "Due to its irregular format, most researchers transform such data to regular 3D voxel grids or collections of images.",
                "In the latter case, approximations arise due to memory restriction.",
                "reinforcement learning approaches have long appealed to the data management community due to their ability to learn to control dynamic behavior from raw system performance.",
                "Now, due to the power wall, the power inefficiency of generalpurpose hardware causes modern servers to change."
            ]
        }
    },
    "duplicate": {
        "duplicate the dataset": {
            "sentences": [
                "we duplicate the real datasets before using them"
            ]
        }
    },
    "easier": {
        "be easier to": {
            "sentences": [
                "Raft, a variant of Paxos,is much easier to understand and implement."
            ]
        }
    },
    "effect": {
        "effect of": {
            "sentences": [
                "HAPE minimizes the effect of heterogeneity and allows the rest of the system to be built by combining existing work on homogeneous systems."
            ]
        }
    },
    "elaborate": {
        "elaborate on": {
            "sentences": [
                "we elaborate on these novelties one by one.",
                "Section 3 elaborates on our system model, threat model, and the detailed construction, followed by the security analysis in Section 4",
                "This section first introduces CB, and then elaborates on how fresh transactions defer the block propagation and enable a transaction withholding attack, followed by experiments that confirm the identified bottleneck."
            ]
        },
        "elaborate in": {
            "sentences": [
                "The key component that helps us realise this is our proposed contrastive clustering in the latent space, which we elaborate in Sec. 4.1."
            ]
        }
    },
    "elicit": {
        "elicit from": {
            "sentences": [
                "In addition to eliciting entity features from the KGs"
            ]
        }
    },
    "eliminate": {
        "can be eliminate": {
            "sentences": [
                "how two-phase commit can be eliminated for distributed transactions"
            ]
        },
        "eliminate the possibility": {
            "sentences": [
                "This eliminates the possibility of performing the recovery progressively"
            ]
        },
        "eliminate dispatch overhead": {
            "sentences": [
                "Discussion An alternative solution to the runtime is aheadof time compilation to eliminate dispatch overhead."
            ]
        },
        "eliminate sth": {
            "sentences": [
                "Meerkat eliminates all cross-core and cross-replica coordination."
            ]
        }
    },
    "elucidate": {
        "elucidate in": {
            "sentences": [
                "As elucidated in Sec. 4.3, we find that Helmholtz free energy is higher for unknown instances."
            ]
        }
    },
    "embed": {
        "embed in": {
            "sentences": [
                "Proposals typically build utility frame\u0002works by constructing a performance model of multi-tier applications embedded in an optimisation problem, where a utility function expresses satis\u0002faction of each workload towards assigned resources."
            ]
        },
        "embed as": {
            "sentences": [
                "By embedding our proposed\ndecomposition blocks as the inner operators, Autoformer can progressively separate the long-term\ntrend information from predicted hidden variables."
            ]
        },
        "embed a bloom": {
            "sentences": [
                "We embed a Bloom-filtered semi-join reducer that significantly reduces materialization overhead for queries with medium and high selectivity."
            ]
        },
        "embed into": {
            "sentences": [
                "We embed feature extraction into transaction workers to parallelize this process."
            ]
        },
        "embed table often": {
            "sentences": [
                "In total, embedding tables oftenrequire up to tens of GBs of storage."
            ]
        },
        "be embed in": {
            "sentences": [
                "The block difficulty of each block is measured in terms of how many times the block is harder to obtain than using the initial target of the system that is embedded in the genesis block."
            ]
        }
    },
    "embedd": {
        "embedd into": {
            "sentences": [
                "In this category, computation nodes can be seen as approximate objects where general AC techniques are embedded into the distributed environment by combining its features."
            ]
        }
    },
    "embrace": {
        "embrace semantic batch": {
            "sentences": [
                "We propose to embrace semantic batching as a core design principle throughout transaction execution for OLTP systems with optimistic concurrency control"
            ]
        },
        "embrace a rapidly": {
            "sentences": [
                "Flex-The deep learning ecosystem is embracing a rapidly growing Tensor [59] proposes a general template to cover multiple diversity of hardware platforms including CPUs, GPUs, FP-operators, but its template is still designed for single operator GAs, and ASICs."
            ]
        },
        "embrace sth": {
            "sentences": [
                "We propose to embrace semantic batching as a core design principle throughout transaction execution for OLTP systems with optimistic concurrency control"
            ]
        }
    },
    "emerge": {
        "emerge as": {
            "sentences": [
                "In recent years, the Path-Ranking Algorithm (PRA) emerges as a promising method for learning inference paths in large KGs",
                "recent years have witnessed the autoregressive transformer emerging as a powerful generative model.",
                "Replicated, in-memory transactional storage systems have emerged as an important piece of infrastructure for datacenter applications because they combine fault-tolerance and strong semantics with high throughput and low latency.",
                "recently, text-to-video generation emerges as a promising direction."
            ]
        },
        "emerge in": {
            "sentences": [
                "Since KBs have a lot of missing information, two natural tasks have emerged in the information extraction community"
            ]
        },
        "emerge to": {
            "sentences": [
                "GPUs have massive threads and high bandwidth, and have emerged to be one of the most promising heterogeneous accelerators to speedup stream processing"
            ]
        }
    },
    "emphasize": {
        "emphasize on": {
            "sentences": [
                "Specifically, we emphasize on two particular cases in MI and MC."
            ]
        },
        "emphasize in": {
            "sentences": [
                "this method was in fact used but not emphasized in previous work [28], and to our knowledge its importance has not been investigated in the context of guided sampling."
            ]
        }
    },
    "employ": {
        "employ to": {
            "sentences": [
                ", 2017) is employed to train the model parameters",
                "it is employed to describe the existence of beings in the world",
                "Our research goal is to understand whetherthe blockchain and Peer-to-Peer approaches can be employed to foster a decentralized and private-by-design IoT",
                "BERT also employs the next sentence prediction (NSP) task to learn the semantic connection between sentences"
            ]
        },
        "employ sth": {
            "sentences": [
                "To model p(v|q,Dq), we employ a RoBERTa-based late fusion strategy."
            ]
        },
        "employ sth. for": {
            "sentences": [
                "Early recommender systems typically employ collaborative filtering for mining users preference from their be havior."
            ]
        },
        "employ sth.": {
            "sentences": [
                "In our experiments, we employ several highly related words."
            ]
        },
        "employ sth. base on": {
            "sentences": [
                "Most of them employ either scene- or object-level image representations based on Recurrent Neural Networks (RNNs)"
            ]
        },
        "employ ... as": {
            "sentences": [
                "\ufeffWe employ bidirectional LSTM (BiLSTM) as one of our basic building blocks for NLI."
            ]
        },
        "employ sth. to do": {
            "sentences": [
                "Third, we employ the cache-coherence and system-wide atomics of NVLink 2.0 to share the hash table between a CPU and a GPU and scale-up data processing."
            ]
        },
        "emply the table": {
            "sentences": [
                "It employs a radix table to quickly find the two spline points."
            ]
        },
        "employ the full objective": {
            "sentences": [
                "The global GOCor module employs the full objective (7), while the local variant uses only the reference term (5)"
            ]
        },
        "employ sth for": {
            "sentences": [
                "For example, diskoriented DBMSs employ different data layouts optimized for non-volatile and volatile storage."
            ]
        },
        "employ a globally": {
            "sentences": [
                "These networks employ a globally consistent protocol  wide-area distributed storage system designed to handle multiple  to ensure that any node can efficiently route a search query to  server failures [23]."
            ]
        }
    },
    "empower": {
        "empower with": {
            "sentences": [
                "This design empowers Autoformer with\nprogressive decomposition capacities for complex time series."
            ]
        }
    },
    "enable": {
        "enable to": {
            "sentences": [
                "The system enables a team of robots to build a semanticallyannotated 3D mesh model of the environment in real-time by leveraging local sensing and computation, and intermittent communication",
                "Conceptual knowledge enables them to function together"
            ]
        },
        "enable re": {
            "sentences": [
                "[2020] proposed a novel model (lsr) that enables relational reasoning across sentences by automatically inducing a latent document-level graph."
            ]
        },
        "enable sth.": {
            "sentences": [
                "The methods can enable many potentially enjoyable interactions."
            ]
        },
        "enable sb./sth. to do": {
            "sentences": [
                "The high bandwidth of NVLink 2.0 enables us to overcome the transfer bottleneck and to efficiently process large data sets stored in main-memory on GPUs."
            ]
        },
        "enable sb to do": {
            "sentences": [
                "Federated learning (FL) enables distributed agents to collaboratively learn a centralized model without sharing their raw data with each other."
            ]
        },
        "enable sth to": {
            "sentences": [
                "It also enables it to recovery nearly instantaneously from failures."
            ]
        },
        "enable  ahead": {
            "sentences": [
                "Graefe  [41]  fine-granularity locking and partial rollbacks using write- describes a system with per-page log record chains that enables  ahead logging."
            ]
        }
    },
    "encode": {
        "encode in": {
            "sentences": [
                "The pattern that anyone closely associated with the Lakers would live in L.A with high probability could be encoded in the embedding space."
            ]
        },
        "encode with": {
            "sentences": [
                "Since each feature is encoded with its label"
            ]
        },
        "encode on": {
            "sentences": [
                "Table 6 shows the performances of different angle encoding methods on the KITTI validation set"
            ]
        },
        "encode into": {
            "sentences": [
                "First, any text-to-SQL parsing model must encode the schema into representations suitable for decoding a SQL query that might involve the given columns or tables."
            ]
        },
        "encode sth. into sth.": {
            "sentences": [
                "In order to ensure a fair comparison across systems, we dictionary encode the string columns into integers prior to data loading and manually rewrite the queries to directly reference the dictionary-encoded value."
            ]
        },
        "encode visually": {
            "sentences": [
                "the image-text training objectives suggest that these text encoders may encode visually semantic and meaningful representations especially relevant for the text-to-image generation task."
            ]
        },
        "be encode into": {
            "sentences": [
                "Moreover, the state of ISP is easily encoded into vector as the input of DQN"
            ]
        }
    },
    "encoders": {
        "encoders for": {
            "sentences": [
                "imagen showcases the effectiveness of frozen large pretrained language models as text encoders for the text-to-image generation using diffusion models."
            ]
        }
    },
    "encompass": {
        "encompass sth": {
            "sentences": [
                "The system encompasses two primary components."
            ]
        },
        "encompass not only ... but also": {
            "sentences": [
                "In order to encompass not only multisets of discrete data, but also a range of important applications including selectivity estimation, time series processing, and OLAP queries, we generalize this setup as follows."
            ]
        }
    },
    "encounter": {
        "encounter in": {
            "sentences": [
                "we provided xes to the issues we encountered in the existing reference implementations.10, 11 a more concerning problem we encountered relates to the performance of the gremlin server.",
                "The \ufb01rst two of these tasks are encountered in preparing a new paper"
            ]
        },
        "encounter with": {
            "sentences": [
                "Federated learning (FL) is encountered with the challenge of training a model in massive and heterogeneous networks."
            ]
        }
    },
    "encourage": {
        "encourage to": {
            "sentences": [
                "encouraging to leverage from label similarity information for better classification boundary.",
                "To encourage the model to learn fine-grained distinctions between topically-similar data, our second strategy produces corrupted candidates by swapping two cells in the same table."
            ]
        },
        "be encourage to": {
            "sentences": [
                "hence, the agent can be encouraged to group them."
            ]
        }
    },
    "encrypt": {
        "encrypt with": {
            "sentences": [
                "The private key is encrypted with the password which is provided while creating the account"
            ]
        }
    },
    "encrypte": {
        "encrypte with": {
            "sentences": [
                "Various characteristics of the system on which the binary is to be executed are collected and the code is encrypted with this information as key."
            ]
        }
    },
    "end": {
        "end up with": {
            "sentences": [
                "We end up with exactly k so (ii) follows. Finally, we need to argue that there are no positive covariances."
            ]
        },
        "end up": {
            "sentences": [
                "Dhamija et al. [8] find that even with this extra training signal, the state-of-the-art object detectors result in false positive detections, where the unknown objects end up being classified as one of the known classes, often with very high probability"
            ]
        },
        "end with": {
            "sentences": [
                "(city, blasio) is tagged as 1 in the st-to-ot sequence, which means that the subject and object are the entities ending with city and blasio, respectively."
            ]
        },
        "end up do": {
            "sentences": [
                "As we show in the next section, using HorseQC ends up being slower than running the query efficiently directly on the CPU."
            ]
        },
        "end up very": {
            "sentences": [
                "Since the number ofinteractions for a categorical feature is often small comparedto the features cardinality (all available products), the binaryvector representing such interactions ends up very sparse."
            ]
        }
    },
    "endow": {
        "endow with": {
            "sentences": [
                "with the proposed multi-frame-rate hierarchical training framework, cogvideo is endowed with a better understanding of text-video relations and abilities to control the intensity of changes during generation."
            ]
        }
    },
    "enforce": {
        "enforce the index": {
            "sentences": [
                "Since the in-memory index is used mainly for estimation pur\u0002pose, it is unnecessary to enforce the index in the master node to be strictly consistent"
            ]
        }
    },
    "engage": {
        "engage in": {
            "sentences": [
                "With a single set of weights, Gato can engage in dialogue, caption images, stack blocks with a real robot arm, outperform humans at playing Atari games, navigate in simulated 3D environments, follow instructions, and more.",
                "For collection and shar- ing, the provided response options allowed users to select whether a given policy explicitly stated that the website engaged in that prac- tice (\u201cYes\u201d), explicitly stated that it did not engage in that prac- tice (\u201cNo\u201d"
            ]
        }
    },
    "enhance": {
        "enhance sb to do sth": {
            "sentences": [
                "We also enhance the framework to avoid the loss of accuracy by accumulating delayed gradients."
            ]
        },
        "enhance sth": {
            "sentences": [
                "we can enhance table joins by building indexes for the driven tables of the join"
            ]
        },
        "enhance with": {
            "sentences": [
                "checkpoints could be enhanced with the support of backend storage clusters"
            ]
        }
    },
    "enhancement": {
        "enhancement to": {
            "sentences": [
                "Future enhancements to the system can include adding a loop closure module"
            ]
        }
    },
    "enqueue": {
        "be enqueued to": {
            "sentences": [
                "The current mini-batch is enqueued to the dictionary, and the oldest mini-batch in the queue is removed"
            ]
        }
    },
    "ensure": {
        "ensure across": {
            "sentences": [
                "A key feature of persistent indexes is to ensure correct recovery across restarts and power cycles"
            ]
        },
        "ensure by": {
            "sentences": [
                "We ensure this by replacing the lowest-scored retrieved table in Dq with the gold table whenever it has not been retrieved."
            ]
        },
        "ensure highly": {
            "sentences": [
                "we demonstrate that this careful mixture of datasets is essential for flamingo\u2019s general few-shot task induction capability and to ensure highly relevant generated outputs for visual inputs."
            ]
        },
        "ensure taht": {
            "sentences": [
                "A DBMS ensures that the database state is not corrupted due to an application, operating system, or device failure."
            ]
        }
    },
    "entail": {
        "entail two major drawbacks": {
            "sentences": [
                "This scheme entails two major drawbacks."
            ]
        },
        "be entail by": {
            "sentences": [
                "Not only does this algorithm avoid the use of a sliding window, but the lack of autocorrelation is an emergent property entailed by history agnosticism."
            ]
        }
    },
    "enter": {
        "enter to": {
            "sentences": [
                "It is entered to a table TH."
            ]
        }
    },
    "enumerate": {
        "enumerate sth": {
            "sentences": [
                "which needed to enumerate each coming query"
            ]
        }
    },
    "equal": {
        "equal to": {
            "sentences": [
                "Actually, the function of worker thread equals to cleanup",
                "In probability proportional to size sampling (pps) with replacement (wr), each sample Sj ??[n], j ??[k], is independent, and equal to i with probability wi/w[n].",
                "which simply equals to the maximal number of available processors.",
                "It is an indicator function which equals to one if its argument is true, and zero otherwise."
            ]
        }
    },
    "equip": {
        "equip with": {
            "sentences": [
                "Due to the intensive cost of labor and expertise in anno- tating 3D medical images at a voxel level, most benchmark datasets are equipped with the annotations of only one type of organs and/or tumors",
                "The head node is a powerful server equipped with an Intel(R) Xeon(R) 2.2 GHz 24-core CPU",
                "Equipped With DP, SDCF allows users to adjust their privacy levels based on their own privacy budgets"
            ]
        },
        "be equip with": {
            "sentences": [
                "The first discrete platform is equipped with an Intel i7-8700K CPU"
            ]
        }
    },
    "escalate": {
        "escalate into": {
            "sentences": [
                "Moreover, since the number of users participating in one iteration is typically a few thousands, the aggregation escalates into a high-dimensional task"
            ]
        }
    },
    "eschew": {
        "eschew distributional assumptions": {
            "sentences": [
                "If one eschews distributional assumptions, then distribution-free bounds can be used instead."
            ]
        }
    },
    "establish": {
        "establish as": {
            "sentences": [
                "this order is then established as a commitment for all time; the timeline oracle will record the happens-before relationship and ensure that all subsequent queries from all shard servers receive responses that respect this commitment."
            ]
        },
        "establish through": {
            "sentences": [
                "Correspondence is established through exhaustively pair-good solutions."
            ]
        },
        "be establish as": {
            "sentences": [
                "Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neural networks in particular, have been firmly established as state of the art approaches in sequence modeling and transduction problems such as language modeling and machine translation [35, 2, 5]."
            ]
        },
        "establish on": {
            "sentences": [
                "It also establishes a record on the recently released Aachen Day-Night localization benchmark._x000D_"
            ]
        }
    },
    "estimate": {
        "estimate sth": {
            "sentences": [
                "We estimate the error margin as half the interquartile range, that is half the difference between the 25th and 75th percentiles."
            ]
        },
        "be estimate as": {
            "sentences": [
                "the cardinality then can be estimated as follows"
            ]
        },
        "be estimate by": {
            "sentences": [
                "In this paper, we assume that the query cost estimated by the optimizer"
            ]
        }
    },
    "evaluate": {
        "evaluate on": {
            "sentences": [
                "We evaluate VoxelNet on the KITTI 3D object detection benchmark [11] which contains 7,481 training images/point  clouds and 7,518 test images/point clouds, covering three categories: Car, Pedestrian, and Cyclist.",
                "We evaluate our models on the same subset of VizNet Web Tables.",
                "Most recent work on GPU-based database evaluates on GPUs which have memory capacity and bandwidth of 4 GB and 150 GBps respectively.",
                "We evaluate on four popular semantic parsing benchmarks in both fully supervised and weakly supervised settings.",
                "We evaluate our proposed Shunted Transformer on various studies.",
                "using this improved architecture, we evaluated on a selection of specialized graph databases, rdf stores, and rdbms adapted for graphs."
            ]
        },
        "evaluate as": {
            "sentences": [
                "We evaluate table pruning techniques as a pre-processing step to select relevant columns that respect the input example length limits."
            ]
        },
        "evaluate ... on": {
            "sentences": [
                "We evaluate our algorithms on a large set of synthetic and real-life data."
            ]
        },
        "evaluate to": {
            "sentences": [
                "B can be split into two basic blocks by inserting an opaque predi- cate PT (which always evaluates to true) in the center of B.",
                "We evaluate the effectiveness of SEAL-Shap to build a straightforward ranker that directly computes the source values without any evaluation target corpus."
            ]
        },
        "evaluate in": {
            "sentences": [
                "\"this way meta-learning techniques can be evaluated in more challenging settings where tasks are diverse.\""
            ]
        },
        "be evaluate": {
            "sentences": [
                "This task is evaluated using only one dataset, e.g., Multi30k-MMT, containing descriptions in three source languages and one target language."
            ]
        },
        "evaluate the effectiveness": {
            "sentences": [
                "Since current automatic metrics may not be sufficient to comprehensively and accurately evaluate the effectiveness of our proposed model."
            ]
        },
        "evaluate sth": {
            "sentences": [
                "To evaluate our approach, we implemented it in the Peloton in-memory DBMS and compared it against WAL using three storage technologies: NVM, SSD, and HDD."
            ]
        },
        "evaluate with": {
            "sentences": [
                "Privacy policies have been evaluated with different readability metrics in different domains, such as energy companies\u2019 terms and conditions"
            ]
        }
    },
    "evict": {
        "evict from": {
            "sentences": [
                "If the function call returns false, the group is evicted from the hash table."
            ]
        },
        "evict sth. to sth.": {
            "sentences": [
                "Following that, we evict \ud835\udc45 and copy the probe-side relation \ud835\udc46 to GPU memory."
            ]
        }
    },
    "evolve": {
        "evolve into": {
            "sentences": [
                "Data management systems evolved into large mono lithic database servers running as stand-alone processes.",
                "In the past decade, special-purpose graphics processing units (GPUs) have evolved into general purpose computing devices"
            ]
        },
        "evolve over": {
            "sentences": [
                "each may be acquired at a specific time interval (e.g., daily,\nweekly or monthly), have distinct stylized fact, hold varied\ndistributions, and evolve over time at the macro-level."
            ]
        },
        "evolve with": {
            "sentences": [
                "This is a scalable approach for keeping track of how the feature vectors evolve with training, as the number of feature vectors that are stored is bounded by C \u00d7 Q, where Q is the maximum size of the queue",
                "An ST object represents an object that evolves with space and time."
            ]
        },
        "evolve from": {
            "sentences": [
                "It has been evolving from basic techniques such as string matching to more complex methods such as object traversal and signature-based scanning."
            ]
        }
    },
    "exacerbate": {
        "exacerbate if short": {
            "sentences": [
                "\"Other challenges of providing distributed management of security primitives"
            ]
        },
        "exacerbate ... by": {
            "sentences": [
                "For instance, some abort mechanisms exacerbate contention by obtaining a hidden lock in the language runtime."
            ]
        }
    },
    "examine": {
        "examine hypothesis": {
            "sentences": [
                "To examine this hypothesis, we conduct several empirical experiments in Section 3."
            ]
        },
        "examine to": {
            "sentences": [
                "their weights can be examined\nto study how different combinations of past observation values affect the forecast."
            ]
        }
    },
    "exceed": {
        "exceed sth": {
            "sentences": [
                "For large tables the example length can exceed the capacity limit of the transformer."
            ]
        },
        "can exceed": {
            "sentences": [
                "The additional memory accesses, cache misses, CPU cycles, and critical sections invoked by lock manager operations can approach or exceed the costs of executing  the actual transaction logic."
            ]
        }
    },
    "excel": {
        "excel in": {
            "sentences": [
                "While no agent can be expected to excel in all imaginable control tasks, especially those far outside of its training distribution, we here test the hypothesis that training an agent which is generally capable on a large number of tasks is possiblel;"
            ]
        },
        "excel at": {
            "sentences": [
                "AR models excel at unconstrained generation but might not retain label information"
            ]
        }
    },
    "except": {
        "except from": {
            "sentences": [
                "The fastest path algorithm is the one with the lowest performance (except from the particular cases mentioned above)."
            ]
        },
        "except for": {
            "sentences": [
                "SS-Topk still provides a strong privacy guarantee since the shuffler knows nothing except for Top-k indexes"
            ]
        }
    },
    "exchange": {
        "exchange faster": {
            "sentences": [
                "this design facilitates the visual information to exchange faster but also increases the computational complexity, as the computational complexity grows quadratically with the number of tokens \u2013 in comparison, the complexity of a regular convolution grows linearly."
            ]
        }
    },
    "exclude": {
        "exclude from": {
            "sentences": [
                "rowdworkers with legal training were not ex- cluded from participation, since they were part of the population sampled.",
                "we also suppose that the model has access to a background KG, which is a subset of G with all the relations excluded from T_mtrand T_mte",
                "Symmetrically qi is the probability that item i is excluded from the sample S and q[J] is the probability that all i ?? are excluded from the sample."
            ]
        }
    },
    "execute": {
        "execute use purely": {
            "sentences": [
                "\"ExCamera exposes the internal state of the video encoder and decoder"
            ]
        },
        "execute on": {
            "sentences": [
                "The contract code is executed on each node participating in the network as part of the verification of new blocks",
                "First, when FineStream starts a new query, the dispatcher executes the query on the CPU for batch1 and then on the GPU for batch2",
                "Many database operations executed on the GPU are performance bound by the memory subsystem.",
                "Unless all to execute on exactly one processor core on a node.",
                "The generated code is executed on the available devices and is responsible for transferring control and data between the devices.",
                "the smart contract is executed on all Ethereum nodes as a replicated state machine"
            ]
        },
        "execute against": {
            "sentences": [
                "we introduce a reference implementation for the ldbc-snb interactive workload in the gremlin query language that can be executed against any tinkerpop3-compliant graph database."
            ]
        },
        "execute very efficiently": {
            "sentences": [
                "While this process might appear expensive at first glance, it canbe executed very efficiently."
            ]
        },
        "execute it efficiently": {
            "sentences": [
                "SPLATT and the MATLAB TensorToolbox (TTB) exemplify different points in the tradeoff space for hand-written libraries: SPLATTonly supports one of the kernels we evaluate but executes it efficiently, while the Tensor Toolboxsupports all of the evaluated tensor operations but does not achieve high performance."
            ]
        },
        "execute rtasks independently": {
            "sentences": [
                "A vDevice furtherpresents multiple parallel virtual execution units (vEUs), eachof them can execute rTasks independently."
            ]
        },
        "execute sth": {
            "sentences": [
                "We executed a write-heavy variation of the YCSB workload on Peloton."
            ]
        },
        "be execute in": {
            "sentences": [
                "Otherwise, the conflicted entries will be executed in a strict sequence as they arrived"
            ]
        },
        "execute consistently faster": {
            "sentences": [
                "M NS 6% 1% (cid:127) When compared to DSM, PAX cache performance is X/ 10% better and queries execute consistently faster because A 4% 20% P PAXdoesnotrequireajointoreconstructtherecords."
            ]
        }
    },
    "exercise": {
        "exercise them regularly": {
            "sentences": [
                "\"We note that if the start-up latency of a cloud function is not good enough for a given application"
            ]
        }
    },
    "exhibit": {
        "exhibit with": {
            "sentences": [
                "After fitting, the AR module may then\nexhibit sparse coefficients with the most significant weights located on a few positions only."
            ]
        },
        "exhibit bias": {
            "sentences": [
                "Many estimators exhibit bias that diminishes linearly (or faster) with increasing sample size."
            ]
        }
    },
    "exist": {
        "exist in": {
            "sentences": [
                "However, few-shot problem widely exists in KGs",
                "We observe that data structures with pointer fields widely exist in OS kernels.",
                "Challenges and constraints do exist in most surveyed methods.",
                "Note that for each candidate protocol, only one protocol instance exists in the whole system regardless of the number of clusters using that protocol.",
                "We show empirically that such a level exists in the experiments"
            ]
        },
        "exit in": {
            "sentences": [
                "If the group with this key exists in the new group hash-table,update group aggregates."
            ]
        },
        "exist at": {
            "sentences": [
                "the shards execute the node program on a version of the in-memory graph consistent with tprog by comparing tprog to the timestamps of the vertices and edges in the multi-version graph and only reading the portions of the graph that exist at tprog."
            ]
        },
        "exist for": {
            "sentences": [
                "Several methods exist for each one of the steps, which are discussed in the following paragraphs.",
                "This difference also exists for \u201cCalendar\u201d in patterns like whether it has digit or not and whether it is daily calendar or monthly calendar."
            ]
        },
        "exist templates deep": {
            "sentences": [
                "Existing be orders of magnitude larger than what existing templates deep learning frameworks (e.g., Tensorflow [1], PyTorch [39], can cover."
            ]
        },
        "exist due": {
            "sentences": [
                "restriction on the length of the input sequence still exists due to the large scale of the model and limitation of gpu memory, and we leave them for future work."
            ]
        },
        "exist on": {
            "sentences": [
                "limited work exists on selecting encodings for a given dataset"
            ]
        },
        "exist with": {
            "sentences": [
                "The existence of a key in an index means that some version exists with that key but the index entry does not contain information about which versions of the tuple match."
            ]
        },
        "exist tablets only": {
            "sentences": [
                "(Chubbyprovidesan ef(cid:2)cient mechanismthatallows atablet servertocheck The set of existing tablets only changes when a ta- whetherit still holdsits lockwithoutincurringnetwork bleiscreatedordeleted,twoexistingtabletsaremerged traf(cid:2)c.)"
            ]
        },
        "exist nod no": {
            "sentences": [
                "Due to the  allocation of key ranges to X, some existing nodes no longer have  2 http://www.oracle.com/database/berkeley-db.html     Figure 4: Average and 99.9 percentiles of latencies for read and  Figure  5:  Comparison  of  performance  of  99.9th  percentile  write requests during our peak request season of December 2006."
            ]
        }
    },
    "existence": {
        "existence of": {
            "sentences": [
                "The major problem we encountered during training was the existence of too few ground truths",
                "We check the existence of the voxels based on a hash table during the iterative process"
            ]
        }
    },
    "expand": {
        "expand to": {
            "sentences": [
                "This series-wise mechanism achieves O(L log L)\ncomplexity for length-L series and breaks the information utilization bottleneck by expanding the\npoint-wise representation aggregation to sub-series level."
            ]
        },
        "be expand on": {
            "sentences": [
                "the capacity can be expanded on demand by appending chunks to the volume."
            ]
        }
    },
    "expect": {
        "expect to": {
            "sentences": [
                "we cannot expect crowdworkers to perform better.",
                "If the input distributions are not too special, then we expect this to give us the best estimates in practice, using variance as the classic measure for estimation quality.",
                "We expect to create a false statement if we replace the focus mention with a replacement mention (e.g., \u2018Billy Mayfair is Australian.\u2019), but there is no guarantee it will be actually false.",
                "The mirror sample is constructed using the local neighbor approximation and equivalence regularization, expecting to approximate equivalent samples in the marginal distributions.",
                "Combined with work showing self-modified software to be semantically equivalent to non-self modified software, self-modifying techniques are not expected to provide a provable level of obfuscation."
            ]
        },
        "except to": {
            "sentences": [
                "expect to explore the most important feature is not blocking the application at any scaling."
            ]
        },
        "expect to accurately": {
            "sentences": [
                "These smaller space sketches will naturally beless powerful than the Bloom filter: when using a data structure that issublinear in size (i.e., o(n)) we should not expect to accurately answerall set-membership queries, even allowing for false positives and falsenegatives."
            ]
        },
        "expect from": {
            "sentences": [
                "Large scale enterprise use of blockchain one has to expect from the middle of the 2020-s."
            ]
        },
        "expect of": {
            "sentences": [
                "The results show that blockchain systems\u2019performance is limited, far below what is expected of a stateof-the-art database system"
            ]
        }
    },
    "experience": {
        "can experience": {
            "sentences": [
                "On high-contention workloads, however, OCC can experience frequent aborts and, in the worst case, contention collapse, where performance for a class of transactions crashes to nearly zero due to repeated conflicts."
            ]
        }
    },
    "experiment": {
        "experiment with": {
            "sentences": [
                "We experiment with two different strategies",
                "We implemented the sampling operator in the Gigascope DSMS in order to experiment with the feasibility and performance of the operator.",
                "Next, we experiment with the semi-supervised adaptation setting."
            ]
        },
        "experiment on": {
            "sentences": [
                "We experiment on two public datasets and demonstrate that our method achieves the best results and generalizes to other domains.",
                "We conduct experiments on KITTI benchmark and show that VoxelNet produces state-of-the-art results in LiDAR-based car,"
            ]
        }
    },
    "explaine": {
        "explaine by": {
            "sentences": [
                "This is likely explained by the limited number of full annual periods\nin the data to fit the model on."
            ]
        }
    },
    "explicitly": {
        "explicitly model occlusion": {
            "sentences": [
                "In order to explicitly model occlusion and non-matching regions into our probabilistic mapping"
            ]
        }
    },
    "exploit": {
        "exploit into": {
            "sentences": [
                "Recent pre-training techniques exploit external knowledge (e.g. entity-level information, commonsense knowledge, knowledge graph) into large-scale pretrained language models (Xiong et al. 2019; Wang et al. 2020; Peters et al. 2019; Rosset et al. 2020)."
            ]
        },
        "exploit in": {
            "sentences": [
                "in this way, bidirectional attention context is fully exploited in text and given frames without interfering with auto-regressive frame prediction."
            ]
        },
        "exploit context in": {
            "sentences": [
                "Exploiting context in this manner allows them to find correspondences at arbitrary locations, including seemingly texture-less areas."
            ]
        },
        "exploit for": {
            "sentences": [
                "TransR exploits separated space for relations to consider different attributes of entities"
            ]
        },
        "exploit to": {
            "sentences": [
                "exploits the power of encoder to capture implicit knowledge of texts under the supervision of KGs"
            ]
        }
    },
    "exploite": {
        "exploite to": {
            "sentences": [
                "Common features tracked over the sliding-window are exploited to improve the localization accuracy for both algorithms."
            ]
        },
        "be exploit to": {
            "sentences": [
                "The KL-divergence is exploited to model the mismatch of score distribution by classifiers on prototypes computed in each domain or their combination."
            ]
        }
    },
    "explore": {
        "explore in": {
            "sentences": [
                "Previous research has explored the benefits of highlight- ing in: supporting workers in digitization tasks by high- lighting target fields",
                "Dynamic properties have also been explored in other contexts outside few-shot relational learning"
            ]
        },
        "explore for": {
            "sentences": [
                "The mutual-instance relation is an important direction to explore for this purpose."
            ]
        },
        "explore from": {
            "sentences": [
                "Therefore, we explore from observational interactions"
            ]
        },
        "be explore so far": {
            "sentences": [
                "This have not been explored so far."
            ]
        },
        "explore to": {
            "sentences": [
                "We observe further VQA/Text-Caption network designs and explore to further improve theText-VQA andText-Caption tasks.",
                "A few methods explore to recommend news in a privacy-preserving way"
            ]
        },
        "explore a well-structured dag": {
            "sentences": [
                "Structural learning isneeded to explore a well-structured DAG to represent potentialconditional independence of attribute."
            ]
        },
        "explore program better": {
            "sentences": [
                "It shows the distribution of performance of the explored programs better."
            ]
        },
        "explore this further": {
            "sentences": [
                "They conjecture that their approach can be extendedto disjunctive binary merges (e.g., matrix addition) by casting them as outer joins [Stodghill 1997,Chapter 15.1], but did not explore this further [Kotlyar 1999, Chapter 11.2]."
            ]
        },
        "explore pretrained": {
            "sentences": [
                "imagen explores pretrained text encoders: bert [15], t5 [51] and clip [46]."
            ]
        },
        "explore even": {
            "sentences": [
                "our observation that scaling the size of these language models have significantly more impact than scaling the u-net size on overall performance encourages future research directions on exploring even bigger language models as text encoders."
            ]
        },
        "explore visually": {
            "sentences": [
                "others have explored visually conditioned language generation (cho et al., 2021; tsimpoukelli et al., 2021;\nwang et al., 2022, 2021; xu et al., 2021) but have not yet shown good performance in low data regimes."
            ]
        },
        "explore that": {
            "sentences": [
                "This paper explores the changes that are required in a DBMS to leverage the unique properties of NVM in systems that still include volatile DRAM."
            ]
        }
    },
    "expose": {
        "expose to": {
            "sentences": [
                "The translation, then, must not only rewrite the query in terms of the Cypher language, but also bridge the gap between the structure exposed to the user, and the model actually stored in Neo4j.",
                "On the other hand, Hyper exposes the operators to parallelism.",
                "language models are trained on text only corpus significantly larger than paired image-text data, thus being exposed to a very rich and wide distribution of text."
            ]
        },
        "expose the fine": {
            "sentences": [
                "RAMMER addresses this issue with a holistic compiler solution that (1) provides an rTask-operator abstraction thatexposes the fine-grained intra-operator parallelism."
            ]
        },
        "expose sth": {
            "sentences": [
                "However, new technologies, like kernelbypass networking and faster network links, have exposed hidden bottlenecks in distributed systems."
            ]
        },
        "expose (e.g.": {
            "sentences": [
                "We  instead  focus  on  reducing  MTTR  to  shrink  the  system like MySQL writes data pages to objects it exposes (e.g.,  window  of  vulnerability  to  a  double  fault."
            ]
        }
    },
    "express": {
        "be express as": {
            "sentences": [
                "Therefore, the distribution of zi can be expressed as the joint distribution of zc and zo."
            ]
        },
        "express over": {
            "sentences": [
                "however, many relations are expressed over multiple sentences in real-world applications."
            ]
        },
        "be express by": {
            "sentences": [
                "This kind of computation logic is expressed by a user-defined function"
            ]
        },
        "express with": {
            "sentences": [
                "It can be noticed that the former two queries would be much difficult and unnatural to express with a duration-labeled representation."
            ]
        }
    },
    "extend": {
        "extend to": {
            "sentences": [
                "However, our method can also be extended to few-shot cases by existing few-shot learning methods",
                "Bendale et al. [3] extend Open Set to an Open World classification setting by additionally updating the image classifier to recognise the identified new unknown classes",
                "However, as shown in Fig. 2, extending it to multiple query points is straightforward",
                "We further plan to extend our annotation efforts to more and other categories of websites.",
                "More recently, these pretraining methods have been extended to jointly learn representations of tables as well as text.",
                "The joint distribution can be extended to multiple tables as follows.",
                "This also extends to Internet of Things settings, where devices need to establish trust among each other",
                "With the development of web browsers, browsers are extended to support more function than ever."
            ]
        },
        "extend for": {
            "sentences": [
                "Future work will extend our framework\nfor other time-series intelligence tasks.",
                "Similarly, Prob and ProbHash cannot be easily extended for multi-way approximate join, unless the multi-way join query plan is decomposed into a series of binary joins."
            ]
        },
        "extend by": {
            "sentences": [
                "this tradeoff ensures that the scalability limits of a centralized timeline service  are extended by adding gatekeeper servers."
            ]
        },
        "extend into": {
            "sentences": [
                "In these real-world applications, one pressing demand\nis to extend the forecast time into the far future, which is quite meaningful for the long-term planning\nand early warning."
            ]
        },
        "extend\u2026 to\u2026": {
            "sentences": [
                "However, it\u2019s nontrivial to extend them to scene understanding or other 3D tasks such as point classification and shape completion."
            ]
        },
        "be extend by": {
            "sentences": [
                "The work is further extended by adapting classifiers through a residual transfer module in [17].",
                "This result is later extended by Muandet et al., and Deshmukh et al.  give a bound for multi-class classification in a similar form."
            ]
        },
        "extend xxx to": {
            "sentences": [
                "We plan to extend the Transformer to problems involving input and output modalities other than text and to investigate local, restricted attention mechanisms to efficiently handle large inputs and outputs such as images, audio and video."
            ]
        },
        "extend ... to": {
            "sentences": [
                "\ufeffFor future research, we will extend our model to capture more detailed interactions according to the category structure of questions."
            ]
        },
        "be extend to sth.": {
            "sentences": [
                "Our strategy could be extended to multi-way joins (e.g., for a star schema) by building hash tables on a different processor in parallel."
            ]
        },
        "extend command log": {
            "sentences": [
                "To address this problem, we first extend command logging to a distributed setting and enable all the nodes to perform their recovery in parallel"
            ]
        },
        "extend sth": {
            "sentences": [
                "It adds the Spark API for OLTP and extends SparkSQL for OLAP queries.",
                "To address this problem, we first extend command logging to a distributed setting and enable all the nodes to perform their recovery in parallel"
            ]
        },
        "extend with": {
            "sentences": [
                "It was already explained that the Neo4j database was extended with temporal capabilities by means of a collection of procedures."
            ]
        },
        "extend from": {
            "sentences": [
                "The HFL setting can be extended from the cross-silo setting to the cross-device setting"
            ]
        }
    },
    "extent": {
        "extent to": {
            "sentences": [
                "we extend it to the monocular case based on the pop-up plane model"
            ]
        },
        "extend to": {
            "sentences": [
                "While ACC currently supports PartCC, OCC, and 2PL, we plan to extend it to more protocols and variants in the future."
            ]
        }
    },
    "extract": {
        "extract from": {
            "sentences": [
                "Planes are extracted from the downsampled 3D point cloud",
                "The first step in geometric computer vision tasks such as Simultaneous Localization and Mapping (SLAM), Structure-from-Motion (SfM), camera calibration, and image matching is to extract interest points from images.",
                "The feature of each node in SOM is extracted from point features associated with this node using channel-wise max pooling.",
                "It is extracted from Flymine, which is an integrated database",
                "the ELMo [8] model extracts context-dependent representations from a bidirectional language model",
                "A natural idea is to use the feature map extracted from each client as the representation since it is directly associated with the raw data.",
                "The content selector extracts informative items (rows, columns or cells) from a table.",
                "ACC monitors transactions\u2019 access patterns and extracts features from the active workload."
            ]
        },
        "be extract from": {
            "sentences": [
                "Traditionally, correspondences between images have been obtained by hand crafted local invariant feature detectors and descriptors [23, 25, 42] that were extracted from the image with a controlled degree of invariance to local geometric and photometric transformations.",
                "The text of both datasets, WikiText103 (Merity et al., 2017) and WikiText2 (Merity et al., 2017), are extracted from Wikipedia.",
                "z is extracted from item content information."
            ]
        },
        "extract by": {
            "sentences": [
                "however, sentence-level re faces an inevitable restriction in that many real-world relations can only be extracted by reading multiple sentences."
            ]
        },
        "extract term": {
            "sentences": [
                "Open-SESAME then graphs such as Visual Genome (Krishna et al. 2016) to extracts terms again from the stories."
            ]
        }
    },
    "extracte": {
        "extracte from": {
            "sentences": [
                "Their analysis is limited to a small number of binary questions for which answers are extracted from privacy policies with vary- ing accuracy."
            ]
        }
    },
    "extrapolate": {
        "extrapolate from": {
            "sentences": [
                "Emulation analyses are more \"messy\" and harder to extrapolate from, precisely because they are data driven."
            ]
        }
    },
    "face": {
        "face with": {
            "sentences": [
                "Faced with specific applications, different preprocessing methods need to be built based on existing cleansing or filtering techniques."
            ]
        },
        "face by": {
            "sentences": [
                "However , deep learning on point clouds is still in its infancy due to the unique challenges faced by the processing of point clouds with deep neural networks."
            ]
        },
        "face in": {
            "sentences": [
                "It also occurs naturally in many activities we are faced in our work as scientists, such as identifying if a paper is relevant to a research topic (Wal- lace et al. 2017)."
            ]
        }
    },
    "facilitate": {
        "facilitate sth. with sth.": {
            "sentences": [
                "These are dedicated hardware units that facilitate both bidirectional transfers and overlapping transfers with computation."
            ]
        },
        "facilitate with": {
            "sentences": [
                "It is desirable to facilitate FL with rigorous differential privacy (DP) guarantee."
            ]
        }
    },
    "fail": {
        "fail to": {
            "sentences": [
                "However, 3D-ResNet [11] failed to con- verge on our training data",
                "First, simple features only capture shallow patterns and often fail to handle the flexible schema and varied expressions in Web tables.",
                "fails to provide upper bound of time complexity on the performance for insertions.",
                "But it fails to scale beyond 16 nodes",
                "Yet,such models assumes prior knowledge and often fail to correctly capture the full complexity and richness of the data.",
                "Crowdworkers generally either match the skilled annotators\u2019 inter- pretation or fail to reach the required agreement level.",
                "Though such protocols achieve ideal performance in their optimized workload scenario, they fail to maintain the same performance with fluctuating workloads.",
                "Using uniform popularity bias fails to distinguish these two users different conformity.",
                "The encoder\ndirectly learned based on the prediction loss fails to learn a\nbetter stock representation for the decoder (predictor).",
                "Multi-core in-memory transactional systems promise significant performance  gains over disk-based systems, but recent proposals often fail to deliver consistently high performance outside of a narrow spectrum of workload characteristics, suffering low performance under high contention [43] and limited scalability with multiple cores [63]",
                "they might fail to capture the long-range temporal dependency in time series. Several methods applied pre-training to RNNs which is similar to CNN-based pre-training.",
                "However, a feature detector may fail to extract enough interest points that are repeatable between images due to various factors such as poor texture",
                "it fails to maintain the interpretability of symbolic knowledge",
                "Images shown in Figure 2 are generated by the proposed method within the architecture of SGAN (PUSGAN), while SGAN failed to generate such high-resolution images, especially on the CA T-256 dataset."
            ]
        },
        "fail to effectively": {
            "sentences": [
                "in table  7, we also include an analysis of the failure modes of our approach, including the number of failures caused by incorrect task identi\ufb01cation \u2013 misidentifying the object \u2013 versus control failures \u2013 when the object was clearly correctly identi\ufb01ed, but the robot failed to effectively push it."
            ]
        },
        "fail for": {
            "sentences": [
                "First, this method fails for detected objects that do not always obey the rule proposed above, such as pedestrians"
            ]
        },
        "fail to accurately": {
            "sentences": [
                "Because the number of possible choices of each de-increase quickly as the programs become complete, whichcision is large, to make the sequential process feasible, this means the cost model performs very well for complete pro-approach keeps only top-k candidate programs after every de-grams but fails to accurately predict the final performance of cision."
            ]
        },
        "fail on": {
            "sentences": [
                "speaking by randomly moving lips), but fail on text prompt such as \u201ca lion is drinking water\u201d."
            ]
        },
        "fail in": {
            "sentences": [
                "As a result, matching individual patch descriptors will often fail in such challenging situations."
            ]
        },
        "fail to do": {
            "sentences": [
                "A common embedding space fails to tailor discriminative visual knowledge for a target task especially when there are a few labeled training data"
            ]
        }
    },
    "fall": {
        "fall into": {
            "sentences": [
                "The utilization of a cluster is defined to be the ratio of number of operations that fall into the given cluster to the total number of operations",
                "Our method also falls into this category",
                "In general, existing online learning algorithms mainly fall into two categories according to the model updating strategy, namely the first-order-based online learning and the second-order-based online learning.",
                "Recent deep learning based few-shot learning approaches fall into two main categories"
            ]
        },
        "fall back to": {
            "sentences": [
                "When not enough inliers are found we fall back to a frameto-frame tracking method"
            ]
        },
        "fall onto": {
            "sentences": [
                "The approaches used in the literature fall onto Reactive, Proactive and Hybrid engagement."
            ]
        },
        "fall through": {
            "sentences": [
                "To ensure that these junk in- structions are unreachable during program execution, these candidate blocks must not allow the execution to fall through to them.",
                "To ensure that these junk in- structions are unreachable during program exe- cution, these candidate blocks must not allow the execution to fall through to them."
            ]
        },
        "fall off": {
            "sentences": [
                "However, while the strict criteria (higher threshold) leads to higher foreground precision, the recall of the retained box candidates also falls off quickly"
            ]
        },
        "fall within": {
            "sentences": [
                "It suggest that the performance of DoDNet tends to become stable when the width of dynamic head falls within reasonable range (\u2265 8)",
                "These works, however, still all fall within centralized training settings and lack scalability for practical deployment",
                "The DBMS must ignore the effects of transactions that fall within such a gap while determining the tuple visibility."
            ]
        },
        "fall off sharply": {
            "sentences": [
                "IBJSfares better at the median, but falls off sharply at tail, because samples of a practical size have a small chance to hit low-density queriesin a large joint space."
            ]
        }
    },
    "fallback": {
        "fallback to": {
            "sentences": [
                "We prioritise to use more specific (level 3) DBpedia classes if possible, and will fallback to less specific (level 2) DBpedia classes only if the hyperlinked Wikipedia entities cannot agree on a more specific DBpedia classes."
            ]
        }
    },
    "favor": {
        "in favor of": {
            "sentences": [
                "In addition, architects explore designs that go beyond the classical system-wide cache-coherence in favor of increased core scalability."
            ]
        },
        "favor to": {
            "sentences": [
                "He et al. [32] claim that the prior of an example being positive is related to the data distribution, and favor to build an optimal Bayesian classifier to tackle this problem."
            ]
        }
    },
    "feature": {
        "be feature with": {
            "sentences": [
                "The structure-aware aggregation method is featured with Global Graph Linking, Local Graph Linking and DualGraph Aggregation Mechanism."
            ]
        }
    },
    "fee": {
        "fee into": {
            "sentences": [
                "The online profiling module is used to analyze input batches and queries for useful information, which is then fed into the performance model.",
                "Speci\ufb01cally, the entity pair embeddings are sequentially fed into a recurrent autoencoder",
                "The CFP and OCT feature vectors are simply concatenated and fed into a regular classification block (a linear layer followed by softmax) for categorizing the input as normal, dryAMD, or wetAMD",
                "we feed into the down stream classifier combines this combination"
            ]
        },
        "feed": {
            "sentences": [
                "The fusion embedding is then added with the position embedding, which is fed as input to the BERT model Since we do not use the NSP pretraining task, we omit the segment embedding"
            ]
        },
        "feed to": {
            "sentences": [
                "The flattened vector is fed to an FC layer to obtain the output glyph vector"
            ]
        },
        "feed into": {
            "sentences": [
                "The average normal vectors of a 3D model sampled in the finest leaf octants are fed into the network, and 3D-CNN is applied on the octants occupied by the 3D shape surface."
            ]
        },
        "be feed into": {
            "sentences": [
                "Second, aggregated vector f(l)i will be fed into a vanilla GRU layer to update the node representation at last step h(l\u22121)i, noted as:"
            ]
        },
        "fee to": {
            "sentences": [
                "The final step of cell contextualization is to compose the row and column embeddings together before feeding the result to the next layer."
            ]
        },
        "be feed to": {
            "sentences": [
                "\ufeffThen, the sentence representation is fed to the decoder to reconstruct the input sequence itself."
            ]
        }
    },
    "feed": {
        "feed as": {
            "sentences": [
                "for the decoder-based method, at training time, the ground truth tokens are used as context while at inference the entire sequence is generated by the resulting model on its own, and hence the previous tokens generated by the model are fed as context."
            ]
        },
        "feed into": {
            "sentences": [
                "the features are then fed into self-attention computation and each token, according to the vanilla design, can exchange information with all the others at every single layer."
            ]
        }
    },
    "fetch": {
        "fetch data cooperatively": {
            "sentences": [
                "An alternative to the shared-nothing approach is tofetch data cooperatively."
            ]
        }
    },
    "figure": {
        "figure out": {
            "sentences": [
                "For example, to identify vulnerabilities in a software system, a hacker has to be able to figure out how it works and where it may be attacked.",
                "granted, we figure out that there is a significant gap between the performance on webnlg and webnlg?",
                "We cannot figure out which one is better at this point, so we can either pick one randomly or use all of them at training.",
                "where we also figure out redundant or negative indexes based on the index benefit estimation results"
            ]
        }
    },
    "fill": {
        "fill with": {
            "sentences": [
                "case (2) needs to fill all consecutive gaps with the adjacent keys to enable exponential search",
                "Unused memory is filled with the INT 3 software interrupt14 and when called, will trigger code that will do decryption and mapping of the missing page for execution to resume.",
                "they are lengthy documents filled with legal jargon and often inten- tional vagueness."
            ]
        },
        "fill in": {
            "sentences": [
                "The slots filled in a linear probing hash table or the length of the linked lists in a separate chaing hash.",
                "Fill in blanks in the sentence.",
                "We fill in the above gaps and make the following contributions."
            ]
        },
        "fill up": {
            "sentences": [
                "Clearly, the queue will fill up over time and overflow.",
                "edge insertions and updates are appended at the tail of the block until it fills up, at which point the tel is upgraded to a larger block."
            ]
        },
        "be fill by": {
            "sentences": [
                "These holes are filled directly by copying from leader\u2019s data blocks"
            ]
        }
    },
    "filter": {
        "filter by": {
            "sentences": [
                "The title words are \ufb01ltered by a stop word list of size 429"
            ]
        },
        "filter out": {
            "sentences": [
                "The downgraded performance of w/o Aggregation Gate indicates the advantages of the gated-based aggregation mechanism, which provides the flexibility to filter out useless local structure information.",
                "We filter out the mismatching lines while preserving the computational efficiency",
                "Filter out box candidates of poor quality",
                "we filter out duplicate indexes",
                "we filter out the mismatching lines while preserving the computational efficiency",
                "We do not perform any universe sampling as it will have a high chance of filtering out all tuples.",
                "To contend with this issue, we filtered out the types for which at least 15% of the columns did not contain a single word that is present in the GloVe dictionary."
            ]
        },
        "filter over": {
            "sentences": [
                "Depth is refined by filtering over many per-pixel"
            ]
        },
        "fifilters possible match": {
            "sentences": [
                "which fifilters possible outlier coarse matches."
            ]
        }
    },
    "find": {
        "find to": {
            "sentences": [
                "Glyce is found to improve a wide range of Chinese NLP tasks"
            ]
        },
        "find out": {
            "sentences": [
                "however, we also find out that there are 127 wrong triplets in webnlg, containing a meaningless empty entity.",
                "Research in developmental psychology [41, 36] finds out that the ability to identify what one doesn\u2019t know, is key in captivating curiosity.",
                "Another unethical scenario is that sometimes the negative pattern could be hidden on purpose for the sake of privacy or other ethical considerations, but with VPU, people might be able to find out about the hidden information.",
                "To find out, we perform clustering experiments on the FinTabNet dataset from Zheng et al. (2021).",
                "we can find out the trend of WSI technology.",
                "Research in developmen- tal psychology [41, 36] finds out that the ability to identify what one doesn\u2019t know, is key in captivating curiosity",
                "We can find out the trend of UAV localization technology."
            ]
        },
        "find sth adj.": {
            "sentences": [
                "Some replacements can lead to ungrammatical statements that a model could use to identify the negative statements, so we found it is useful to also replace the entity in the original positive sentence from Wikipedia with the mention from the table."
            ]
        },
        "find through": {
            "sentences": [
                "therefore, for vertices we use a standard copy-on-write approach, where the newest version of the vertex can be found through the vertex index, and each version points to its previous version in the vertex block."
            ]
        },
        "find with": {
            "sentences": [
                "In the black-box setting, the only information an FCS algorithm can use is the fuzz outcomes of a configuration-the number of crashes and bugs found with it and the amount of time spent on it so far."
            ]
        },
        "find it particularly important": {
            "sentences": [
                "We find it particularly important to include the around 1.4 millions image-text pairs."
            ]
        },
        "find it easy to": {
            "sentences": [
                "While humans find it rather easy to construct sentences from 85 images, this task is not as straightforward for machines."
            ]
        },
        "find the most": {
            "sentences": [
                "Otherwise, the policy will find the mostefficient rKernels and then perform a profiling."
            ]
        },
        "find empirically": {
            "sentences": [
                "within [\u22121, 1], but we find empirically that high guidance weights cause x-predictions to exceed these bounds."
            ]
        },
        "find it critical": {
            "sentences": [
                "We found it critical to use a linear increase in frequency for the positional encoding"
            ]
        },
        "find on": {
            "sentences": [
                "Although the real Ethereum transactions and contracts can be found on the public blockchain"
            ]
        }
    },
    "finetuning": {
        "finetuning for": {
            "sentences": [
                "adaprompt: adaptive prompt-based finetuning for relation extraction."
            ]
        }
    },
    "fire": {
        "fire on": {
            "sentences": [
                "The resulting detections are more repeatable and fire on a larger set of stimuli; thus we named the resulting detector SuperPoint."
            ]
        }
    },
    "fit": {
        "fit for": {
            "sentences": [
                "optimistic locking is usually a better fit for PM and balances programmability and performance"
            ]
        },
        "fit into": {
            "sentences": [
                "Alas, traditional data warehousing systems are struggling to fit into this new environment.",
                "Overall, CoMix is simple and easy to implement which perfectly fits into modern mini-batch end-to-end training.",
                "VLL fits naturally into the design of deterministic systems.",
                "The feature points calculated by this method are not accurate and can not be fitted into lines and planes area."
            ]
        },
        "fit in": {
            "sentences": [
                "For completeness, this section lists PU methods that do not fit in any of the considered categories.",
                "The two input tables are co-partitioned such that for each partition pair the hash table fits in cache."
            ]
        },
        "fit continuous attribute to": {
            "sentences": [
                "We use GMMs to fit continuous attributes to reduce their domain sizes."
            ]
        },
        "fit sth. into sth.": {
            "sentences": [
                "Data compression could be used to fit more data into GPU\u2019s memory."
            ]
        }
    },
    "fix": {
        "fix to": {
            "sentences": [
                "The width is fixed to 8, except for the last layer, which has 2 channels"
            ]
        },
        "be fix to": {
            "sentences": [
                "Here,the interaction function is fixed to cosine in order to directly compare with the baseline model LSTM-RNN, similar results can be obtained for other interaction functions such as bilinear and tensor layer."
            ]
        },
        "fix for": {
            "sentences": [
                "the project matrix can not be fixed for real-world long sequence input."
            ]
        }
    },
    "fixe": {
        "fixe to": {
            "sentences": [
                "Figure 12 presents the results of overall query processing performance when we vary the tile size from 256KB to 16MB while the other parameters are fixed to their default values."
            ]
        }
    },
    "flow": {
        "flow from": {
            "sentences": [
                "This is a one-way communication channel, with information flowing unidirectionaly from the sender to the recipient."
            ]
        }
    },
    "fluctuate": {
        "fluctuate in": {
            "sentences": [
                "The access patterns of a workload may be unknown before deployment of user-facing applications or be unpredictable and fluctuate in response to users\u2019 requests"
            ]
        }
    },
    "flush": {
        "flush m into": {
            "sentences": [
                "Finally, N flushes M into its local storage (line 10), which will be used in case there are further cascading failures"
            ]
        },
        "flush sth to": {
            "sentences": [
                "Using this method, the DBMS flushes the changes to the database before recording them in the log."
            ]
        }
    },
    "focus": {
        "focus on": {
            "sentences": [
                "however, they mainly focus on token-level syntactic features or contextual information rather than global interactions between entity arxiv:2106.03618v2 [cs.cl] 22 aug 2021 pairs, neglecting the interdependency among the multiple relations in one context.",
                "In this section, we focus on the core question: why is approximating common aggregates (e.g., COUNT, SUM and AVG) hard when using a small sample (or more generally, a small summary)?",
                "Some works treat all elements equally, through global pooling [70, 40, 15] or instance normalization [60, 33, 32], while others focus on a local neighborhood in coordinate or feature space [41, 66].",
                "graph databases, on the other hand, focus on real-time querying and manipulation of entities, relationships, and the graph structure.",
                "Past work in the database community has focused on using the GPU as a coprocessor.",
                "Previous few-shot learning research mainly focuses on vision and imitation learning domains",
                "There are some classic methods focus on modeling correlation among multiple variables or time series.",
                "for now we focus on data node operations",
                "Most researches on fuzzing mainly focus on file format fuzzing.",
                "Traditional transac- tion management techniques focus on transaction protocols, e.g., OCC, PCC, MVCC, 2PC.",
                "previous works [zeng et al., 2015; feng et al., 2018] focused on identifying relations within a single sentence, which failed to recognize relations between entities across sentences.",
                "Prior works focus on the SLAM on single robot.",
                "The subsequent works focus on expanding the range of text",
                "tagging is shared by all relations because it focuses on the general entity extraction without considering the specific relation types.",
                "A line of research focuses on learning a single global model from non-i.i.d. data.",
                "while  reinforcement  learning  algorithms provide a framework for learning new tasks, they primarily focus on mastery of individ- ual skills, rather than generalizing and quickly adapting to new scenarios.",
                "We chose to focus on the reconfigured resource and classify the reviewed literature on this dimension, in order to analyse patterns that are specific to a cloud resource.",
                "We focus on the stereo task.",
                "Existing works focus on adding new indexes that can benefit the read queries",
                "Future work will focus on implementation and demonstration of the BPIIoT platform for more realistic solutions such as on-demand manufacturing and device self-service",
                "The first attempts for addressing this issue focused on proposing metrics that could be computed based on the total number of examples and the number of positive examples.",
                "SQLite strongly focuses on transactional (OLTP) workloads, and contains a row-major execution engine operating on a B-Tree storage format.",
                "But our focus on ease-of-use and service experience does not stop at the user interface.",
                "We do not analyze CPU utilization and RAM usage as for the considered workloads CPU and RAM information has low value as we primarily assign one task per VM and focus on performance characteristics from the perspective of job execution times.",
                "we focus on the car detection results because pedestrian and cyclist detection may need images to get better results",
                "Though we focus on table retrieval, our methods to alleviate the long sequence issue are off-the-shelf without any further training and can also be applied to other dofmains.",
                "We focus on HBase, as it presents the most complex architecture, is fromRPCHandletoDataXceive.",
                "For this reason we focus on a different input representation for 3D geometry using simply point clouds \u2013 and name our resulting deep netsPointNets.",
                "Several research projects focus on optimizing concurrency controls for main-memory databases.",
                "although achieving great success, such solutions still have limitations \u2013 they treat images and videos separately and focus on generating either of them.",
                "A dedicated body of work [43, 42, 17] focuses on developing measures of (spatial and semantic) uncertainty in object detectors to reject unknown classes",
                "Recent works such as [7,43] focus on fast checkpointing to support efficient recovery",
                "Our experiments are mainly focused on Spider (Yu et al., 2018), a large multi-domain and crossdatabase dataset for text-to-SQL parsing.",
                "In general, these methods usually focus on developing privacy-preserving model training approaches",
                "The few attempts for learning keypoint detectors [48, 35, 59, 9, 11] have only focused on the repeatability._x000D_",
                "In this paper, we focus on the case where all data and structures of the OLTP engine can fit into NVM memory.",
                "Investigations on video representation learning focus on exploiting temporal coherence among consecutive video frames",
                "prior works focus ononly static feature maps in one attention layer.",
                "This category of methods focuses on manipulating the inputs to assist learning general representations.",
                "Pre-trained language models like BERT and ELMo focus on modeling the syntactic and semantic characteristics of word use in natural sentences.",
                "\ufeffWe focus on neural network approaches below.",
                "These deep UDA methods mainly focus on matching the source and target domains via adversarial training or kernelized training.",
                "Related work on Collaborative SLAM(CSLAM) has focused on developing multi-robot SLAM frontends and back-ends(e.g., distributed pose graph optimization).",
                "Prior research definition and curation of such plot graphs require highly on story generation mostly focused on automatically method.",
                "In order to ensure that the adversary cannot focus the mining power onto a single chain, a \u201csortition\u201d mechanism is used."
            ]
        },
        "focuson": {
            "sentences": [
                "existing works focus on adding new indexes that can benefit the read queries but they have several limitations."
            ]
        },
        "focus on much": {
            "sentences": [
                "focused on much more narrow notions of generalization, such as a varied target object location (levine et al., 2016a) or block stacking order (duan et al., 2017)."
            ]
        },
        "fucus on": {
            "sentences": [
                "This work focuses on avoiding bank conflicts in shared memory and finds that their two algorithms that eliminate or avoid bank conflicts are more efficient than their naive reference implementation."
            ]
        }
    },
    "focuse": {
        "focuse on": {
            "sentences": [
                "n response to this issue, recent efforts have focused on the devel- opment of approaches that rely on crowdsourcing to annotate im- portant elements of privacy policies."
            ]
        }
    },
    "follow": {
        "follow to": {
            "sentences": [
                "Our implementation therefore follows prior work [14] to disallow inner node split and only allow expansion."
            ]
        },
        "follow by": {
            "sentences": [
                "We start with the ubiquitous WAL protocol in Section 3, followed by our new WBL method in Section 4.",
                "followed by a sequence of convolutions of stride 1",
                "these methods are able to jointly learn feature representations and the prediction function by\nstacking a series of non-linear layers to perform feature extraction,\nfollowed by a regression layer focused on forecasting.",
                "We measure the performance of an equi-join over the key columns that is followed by a sum/count aggregation over each payload column.",
                "imagen comprises a frozen t5-xxl [52] encoder to map input text into a sequence of embeddings and a 64\u00d764 image diffusion model, followed by two super-resolution diffusion models for generating 256\u00d7256 and 1024\u00d71024 images (see fig.",
                "it is followed by a number of blocks",
                "We next discuss related work, followed by a technical description of our approach",
                "[2020c] proposed a novel graphbased model with a dual-tier heterogeneous graph (dhg), which contains a structure modeling layer followed by a relation reasoning layer.",
                "Similar to [37], we used 2 convolutional layers with 6 and 16 filters of 5 \u00d7 5 kernels followed by max-pooling layers and 2 fully connected layers with size 120 and 84."
            ]
        },
        "follow with": {
            "sentences": [
                "Following with dual-graph encoding, we devise a structure-aware aggregation method on questionschema linking between question-graph GQ and schema-graph GS.",
                "we use a CNN model that consists of three 3 \u00d7 3 convolution layers, two fully connected layers (each followed with ReLu activation), and a final softmax output layer."
            ]
        },
        "be follow by": {
            "sentences": [
                "\ufeffThis is followed by a hidden and a softmax classification layer."
            ]
        },
        "follow a rather": {
            "sentences": [
                "Fortunately, the hardwarelogic of embedding table lookups follows a rather fixed pattern,thus it is possible to prepare a set of hardware code templates, andcompile the look up logic to hardware using an automated codegenerator."
            ]
        }
    },
    "force": {
        "be force to": {
            "sentences": [
                "As a result, models are forced to learn rich relations among diverse local patches."
            ]
        },
        "force to": {
            "sentences": [
                "many service providers are re- luctant to adopt machine-implementable solutions that would force them to further clarify their privacy practices or to commit to more stringent practices."
            ]
        }
    },
    "fork": {
        "fork off": {
            "sentences": [
                "Although it has the same motivation as in-memory fuzzing, a fork server involves forking off a new process for every fuzz iteration."
            ]
        }
    },
    "form": {
        "form by": {
            "sentences": [
                "These direction-associated features are then aggregated according to the angles formed by the given point and its neighboring points."
            ]
        },
        "form as": {
            "sentences": [
                "as can be seen in figure 3, the module is formed as a u-shaped segmentation structure, which contains two down-sampling blocks and two upsampling blocks with skip connections."
            ]
        },
        "form a fully": {
            "sentences": [
                "DISC enrichesa set of IR to form a fully dynamic shape representation."
            ]
        },
        "be form as": {
            "sentences": [
                "Each item embedding component, i, of the aggregated parameters, w[i], is formed as a weighted average of the parameter values wk[i] learned by each delegate."
            ]
        }
    },
    "formalize": {
        "formalize as below": {
            "sentences": [
                "Embedding Model can be formalized as below."
            ]
        },
        "formalize the problem": {
            "sentences": [
                "We formalize the problem of distributing recomputation tasks for subgraphs residing in the failed nodes as a reassignment generation problem"
            ]
        }
    },
    "formulate": {
        "formulate as": {
            "sentences": [
                "Wikipedia with a test set of 1,000 tables, formulated as a multi-label classification task with 127,656 possible header labels.",
                "To achieve this goal, we leverage the idea of structural time series models (Scott & Varian, 2015;\nQiu et al., 2018), which formulates time series as a sum of trend",
                "Another approach is to formulate Federated Learning with non-i.i.d. data as a meta learning problem.",
                "Multi-Hop Reasoning over KGs aims to learn symbolic inference rules from relational paths in G and has been formulated as sequential decision problems in recent years",
                "After extracting four categories of features describing the values of each column, we formulate type detection as a multiclass classification task.",
                "Convergence is therefore formulated as convergence of both the log likelihood and the propensity model.",
                "we formulate the index selection problem as a reinforcement learning task and employ deep q network (dqn) on it.",
                "In this approach, we formulate the table fact verification as a program synthesis problem"
            ]
        },
        "formulate on": {
            "sentences": [
                "the sql implementation is formulated on a schema where vertices and edges are represented as separate tables."
            ]
        },
        "be formulate to": {
            "sentences": [
                "As such, the general-purpose adaptation is then formulated to minimize the distances between the prototypes measured on source data, target data with pseudo labels, and source plus target data."
            ]
        },
        "formulate sth as sth": {
            "sentences": [
                "In this paper, we formulate the ISP as a DRL problem"
            ]
        }
    },
    "forward": {
        "forward to": {
            "sentences": [
                "The glyph embedding himage from CNNs will be forwarded to an image classification objective to predict its corresponding charID"
            ]
        },
        "be forward to": {
            "sentences": [
                "Therefore, they would be forwarded to the corresponding coordinators for processing"
            ]
        }
    },
    "forwardingtheselocationsto": {
        "forwardingtheselocationsto the gracefully": {
            "sentences": [
                "\"The locations of these buffered pairs on verylargeamountsofdatausinghundredsorthousands the local disk are passed back to the master"
            ]
        }
    },
    "frame": {
        "frame as": {
            "sentences": [
                "which is also commonly framed as the visual dataset bias problem.",
                "Predicting missing facts in KGs\u2014usually framed as relation prediction between two entities\u2014is a widely studied problem in statistical relational learning"
            ]
        }
    },
    "freeze": {
        "freeze in": {
            "sentences": [
                "all the parameters in the cogview2 are frozen in the training, and only the parameters in the newly added attention layer (see the attentionplus in figure 3) are trainable."
            ]
        }
    },
    "fulfill": {
        "be fulfil with": {
            "sentences": [
                "(19) is fulfilled with small enough residual and that the new difficulty is calculated correctly from \u03b1."
            ]
        }
    },
    "fund": {
        "fund by": {
            "sentences": [
                "This research has been partially funded by the National Science Foundation under grant agreement CNS-1330596.",
                "this work is funded by national key r&d program of china (funding no."
            ]
        }
    },
    "fuse": {
        "fuse a with b": {
            "sentences": [
                "In TAP, we design text-aware pre-training cially designed pre-training tasks, Text-Aware Pre-tasks to better fuse scene text with the resentation that leads to significant performance"
            ]
        },
        "fuse this newly": {
            "sentences": [
                "In this subsection, we annotate sketches to into the DAG, then rule 4 can be applied to fuse this newly make them complete programs for fine-tuning and evaluation."
            ]
        },
        "fuse by": {
            "sentences": [
                "these representations are then fused by cross-modal attention."
            ]
        }
    },
    "gain": {
        "gain impressive success": {
            "sentences": [
                "Recent efforts have gained impressive success on this task with deep sequential models."
            ]
        },
        "gain in": {
            "sentences": [
                "Deep learning approaches have gained prominence in various machine learning problems and applications."
            ]
        },
        "gain more and more importance,": {
            "sentences": [
                "As these database frontends gain more and more importance, we hope that future benchmark designers and, in particular, system builders get an impetus for enabling better database support for such system-generated workloads."
            ]
        }
    },
    "gap": {
        "gap at": {
            "sentences": [
                "Our evaluation shows that the consensus protocols account for the performance gap at the application layer for Ethereum and Hyperledger"
            ]
        }
    },
    "gather": {
        "gather by": {
            "sentences": [
                "These models take as input a set of paths which are gathered by performing random walks"
            ]
        },
        "gather from": {
            "sentences": [
                "We evaluate the performance of our algorithm by using datasets gathered from building environments"
            ]
        },
        "gather together": {
            "sentences": [
                "Then a copula function gathers them together to learn a joint\ndistribution with coefficient matrix R"
            ]
        }
    },
    "gear": {
        "gear towards": {
            "sentences": [
                "However, there is no such system yet geared towards analytical workloads."
            ]
        }
    },
    "generalize": {
        "generalize to": {
            "sentences": [
                "it cannot be generalized to other",
                "However, classifiers trained on specific datasets cannot always generalize effectively to new datasets owing to the well-known domain shift problem.",
                "However, existing related studies are restricted to dealing with unstructured text as the evidence, which would not generalize to the cases where the evidence has a highly structured format.",
                "GraIL is naturally inductive and can generalize to unseen entities and graphs after training",
                "Generalizing knowledge to unseen domains, where data and labels are unavailable, is crucial for machine learning models."
            ]
        },
        "generalize xxx to": {
            "sentences": [
                "Therefore, cross-domain Text-to-SQL, aiming to generalize the trained model to the unseen database schema, is proposed as a more promising solution [13, 4, 5, 29, 8, 24, 21, 7]."
            ]
        },
        "generalize to properly": {
            "sentences": [
                "The type relation must be generalized to properly handledynamic shapes."
            ]
        }
    },
    "generate": {
        "generate ... from": {
            "sentences": [
                "Otherwise, ??has its correct new value ??,j, and then we proceed to generate the random number r ??(0, 1) from the original algorithm."
            ]
        },
        "generate by": {
            "sentences": [
                "The positive examples in the unlabeled data are then all those that are likely to be generated by the same distribution as the labeled examples.",
                "first, the static portion of the social network is generated by the ldbc data generator, then loaded into the system under testusing vendor-specific loading mechanisms.",
                "furthermore, making the super-resolution models aware of the amount of noise added, via noise level conditioning, significantly improves the sample quality and helps improving the robustness of the super-resolution models to handle artifacts generated by lower resolution models [29].",
                "for the decoder-based method, at training time, the ground truth tokens are used as context while at inference the entire sequence is generated by the resulting model on its own, and hence the previous tokens generated by the model are fed as context.",
                "Figure 1 is an example of such a case, where the observed time\nseries is generated by a seasonal module and nonlinear trend module.",
                "The correspondence volume generated by our module is the result of an internal optimization procedure that explicitly accounts for similar regions in the scene."
            ]
        },
        "generate from": {
            "sentences": [
                "In summary, the learned representations and prediction associations from the end-to-end training approach are unable to transfer\nnor generalize well when the data is generated from a non-stationary environment"
            ]
        },
        "generate attention completely": {
            "sentences": [
                "distinctively, they drop the generated attention completely and use only the gaussian distribution for attention computation."
            ]
        },
        "generate for": {
            "sentences": [
                "Finally, rather than relying only on the supervision of source categories to learn a discriminative representation, we generate pseudo-labels for the target samples in every batch and then harness the label information using a temporal supervised contrastive term, that pushes the examples from the same class close and the examples from different classes further apart."
            ]
        },
        "generate paragraph": {
            "sentences": [
                "Such two-step Existing works [39, 21, 13] mainly adopt a two-stage manner makes the quality of generated paragraphs highly framework for video paragraph captioning: firstly detect dependent on the accuracy of event proposal detectioning event segments in the video, and then generating the which is already a challenging task."
            ]
        },
        "generate code entirely": {
            "sentences": [
                "Our technique generates code entirely fromtensor index notation and simple format descriptors that fully specify the compressed data structures."
            ]
        },
        "generate highly": {
            "sentences": [
                "these include compositionality, cardinality, spatial relations, the ability to handle complex text prompts or prompts with rare words, and they include creative prompts that push the limits of models\u2019 ability to generate highly implausible scenes well beyond the scope of the training data."
            ]
        },
        "generate more": {
            "sentences": [
                "we introduce dynamic thresholding, a new diffusion sampling technique to leverage high guidance weights and generating more photorealistic and detailed images than previously possible."
            ]
        },
        "generate sth": {
            "sentences": [
                "As transactions tend to generate multiple log records that are each small in size, most DBMSs use group commit to minimize the I/O overhead."
            ]
        },
        "generate with": {
            "sentences": [
                "To generate reasonable stories with commonsense knowledge"
            ]
        }
    },
    "get": {
        "get to": {
            "sentences": [
                "How long will it take to get to your house?"
            ]
        },
        "get over": {
            "sentences": [
                "You can't get over a disease by lying in bed.",
                "You can not get over a disease by lying in bed."
            ]
        },
        "get back with": {
            "sentences": [
                "But, it\u2019s obvious that the disassembled code will be not similar to the original code, as it is not possible to get back with all same functionalities."
            ]
        },
        "get 1000x faster": {
            "sentences": [
                "[5]  moore law is dead but gpu will get 1000x faster by 2025.      https://www.nextbigfuture."
            ]
        },
        "get in": {
            "sentences": [
                "Unfortunately, existing research has only evaluated the performance of the two on synthetic microbenchmarks, which are not representative of what we typically get in real workloads."
            ]
        },
        "get too far": {
            "sentences": [
                "This  records, they gossip with the other members of their PG, looking  limit ensures that the database does not get too far ahead of the  for gaps and fill in the holes."
            ]
        }
    },
    "gine": {
        "gine to more": {
            "sentences": [
                "\"People seemed doomed to gine to more modern ones"
            ]
        }
    },
    "give": {
        "give by": {
            "sentences": [
                "similarly, the decomposition-based method uses the gold subject entity as specific input to guide the model extract object entities and relations during the training process while at inference the input head-entity is given by a trained model, leading to a gap between training and inference.",
                "For reference, the original regularization function proposed in AR-Net is given by:"
            ]
        },
        "give up": {
            "sentences": [
                "neural models for joint relation extraction are investigated in recent studies (gupta et al., 2016; zheng et al., 2017), they show promising results but completely giving up overlapping relations."
            ]
        },
        "give in": {
            "sentences": [
                "We use a YCSB benchmark to test their performance under workload variation, with a detailed configuration given in Section 6."
            ]
        },
        "give sth. to sth.": {
            "sentences": [
                "However, in contrast to PCI-e, NVLink gives the GPU direct access to pageable CPU memory."
            ]
        },
        "give rise to": {
            "sentences": [
                "The final system gives rise to state-of-the-art homography estimation results on HPatches when compared to LIFT, SIFT and ORB."
            ]
        },
        "give time as": {
            "sentences": [
                "The perspective,  it  is  preferred  to  keep  the  number  of  divergent  values of the system size and number of replicas are based on versions at any given time as low as possible."
            ]
        }
    },
    "go": {
        "go over": {
            "sentences": [
                "Usually, we will go over many tasks in a batch and update \u03b8 as follows"
            ]
        },
        "go from": {
            "sentences": [
                "For Workload 2 they slightly increase (the Plan policy in Table 4 is an exception) while going from small to large job sizes."
            ]
        },
        "go through": {
            "sentences": [
                "As explained in the introduction, every transaction goes through three phases",
                "To achieve a specific business goal, IPs go through a decision making process that changes the infrastructure, a process we name Cloud Systems Adaptation.",
                "If so, the input batches sequentially goes through OP1 and OP2 and no pipeline exists between two batches for OP1 and OP2",
                "When a multi-head MM-MIL is used to produce multiple \ud835\udc53 \u0304\n\ud835\udc5a\ud835\udc5a,\nwe let each \ud835\udc53 \u0304 go through a distinct linear layer and use mean pooling"
            ]
        },
        "go beyond": {
            "sentences": [
                "Going beyond Transformers, we design Autoformer as a novel decomposition architecture with an Auto-Correlation mechanism.",
                "Also, semi-supervised learning can go beyond binary classification tasks."
            ]
        },
        "go for": {
            "sentences": [
                "the same logic goes for other relations, but note that the tag 2 has an opposite meaning to the tag 1, which represents a reversal link between tokens."
            ]
        },
        "go back to": {
            "sentences": [
                "because it cannot go back to previous steps",
                "This powerful idea goes back to at least 1990s [5, 33, 34, 38, 46], and is typically known as neighbourhood consensus or more broadly as semi-local constraints."
            ]
        }
    },
    "graduate": {
        "graduate from": {
            "sentences": [
                "It is not ideal to solve a query such as finding the people who graduated from Kennesaw State University after 2018."
            ]
        }
    },
    "grateful": {
        "be grateful to": {
            "sentences": [
                "We are grateful to our colleagues Russ Webb, Barry Theobald, and Jerremy Holland for their valuable input."
            ]
        }
    },
    "grind": {
        "ground in": {
            "sentences": [
                "The training data consists of N pairs of utterances and SQL queries {x(l), y(l)}Nl=1 grounded in database D."
            ]
        }
    },
    "group": {
        "group into": {
            "sentences": [
                "We group classes into a set of tasks T = {T1,\u00b7\u00b7\u00b7Tt,\u00b7\u00b7\u00b7}",
                "edges that are incident to the same vertex are grouped into one adjacency list per label.",
                "All transactions that change the distributed Bitcoin ledger are grouped into blocks.",
                "Machine Learning techniques are grouped into two categories, supervised and unsupervised learning.",
                "These primitives are then grouped into planar regions based on their geometric relationships",
                "It first groups point cloud data into voxels and then applies linear networks voxel"
            ]
        },
        "be group into": {
            "sentences": [
                "Specifically, the transactions are grouped into batches where each batch is executed by all replicas of the database in the same deterministic order"
            ]
        }
    },
    "grow": {
        "grow much faster": {
            "sentences": [
                "As disk capacity and cost-per-gigabyte are growing much faster than network cost-performance\u201410X vs. less than3Xinthelast5yearsaccordingtoTable5\u2014theFedExdiskoptionforlargedatatransferswillgetmoreattractive eachyear."
            ]
        },
        "grow quadratically": {
            "sentences": [
                "this design facilitates the visual information to exchange faster but also increases the computational complexity, as the computational complexity grows quadratically with the number of tokens \u2013 in comparison, the complexity of a regular convolution grows linearly."
            ]
        }
    },
    "guarantee": {
        "guarantee on": {
            "sentences": [
                "a persistent learned index that retains the benefits of learned indexes while guaranteeing crash consistency on PM and supporting instant recovery and scalable concurrency."
            ]
        },
        "guarantee sth": {
            "sentences": [
                "This guarantees that the domains of both expressions are comparable. \u27e8value\u27e9 is chosen as at random from the respective column."
            ]
        },
        "guarantee that": {
            "sentences": [
                "By ordering writes to NVM correctly, the DBMS can guarantee that all transactions are durable and atomic.",
                "Therefore, in addition to reducing lock manager overhead, this technique also guarantees that there will be no deadlock within a partition"
            ]
        },
        "guarantee the continuously": {
            "sentences": [
                "(2) The spline interpolation guarantees the continuously differentiable property, soq is continuous because it is the derivative of the CDF."
            ]
        }
    },
    "guide": {
        "guide to": {
            "sentences": [
                "Entry types are used to guide the index operations to choose different actions. Details are illustrated in Section 4."
            ]
        },
        "guide by equally": {
            "sentences": [
                "however, the authors found that the optimization procedure was better guided by equally weighting the  performance  after  each  gradient  descent  step."
            ]
        }
    },
    "gured": {
        "gured to": {
            "sentences": [
                "all systems were congured to load the entire dataset into main memory before the benchmark execution in order to eliminate the overhead of disk i/o we encountered several roadblocks during our experimental study."
            ]
        }
    },
    "halve": {
        "halve in": {
            "sentences": [
                "different from down-sampling, the number of channels is halved in each up-sampling block, which can distribute the aggregated information to each pixel."
            ]
        }
    },
    "handle": {
        "handle sth by": {
            "sentences": [
                "DBMSs can handle media failures by storing the database, the log, and the archival versions of the database.",
                "The TAPAS model handles this by shrinking the text in cells"
            ]
        },
        "handle by": {
            "sentences": [
                "The scale drift in monocular vSLAM system is mainly handled by two approaches",
                "modality interaction is handled by the cosine similarity of the image and text feature vectors."
            ]
        },
        "handle to": {
            "sentences": [
                "To handle these challenges, many approaches manually crafted feature represen Figure 1. VoxelNet directly operates on the raw point cloud (noneed for feature engineering) and produces the 3D detection results using a single end-to-end trainable network.tations for point clouds that are tuned for 3D object detection."
            ]
        },
        "handle in": {
            "sentences": [
                "Finally, in the GPU SIMT model, each GPU thread has an independent register file but, in contrast to the SIMD model, thread divergence is handled in hardware."
            ]
        },
        "handle well": {
            "sentences": [
                "MSCN can handle well on large domain attributes and strong correlations."
            ]
        },
        "handle machine failures": {
            "sentences": [
                "Strong consistency makes it easy to handle machine failures, since all of the copies of a record are identical at all times"
            ]
        },
        "handle a very": {
            "sentences": [
                "For writes, the system routes the requests to to handle a very high write throughput."
            ]
        }
    },
    "happen": {
        "happen relatively quickly": {
            "sentences": [
                "The re-executionof this map work Averageworkerdeathsperjob 1.2 happens relatively quickly."
            ]
        },
        "happen  just": {
            "sentences": [
                "\"The Comparison of Overhead between AS and ASTAW with  Increased Thresholds   Figure 6 presents a comparison between SLA violation percentages of the system in  AS and increased ASTAW algorithm"
            ]
        },
        "happen inside": {
            "sentences": [
                "Both the requesting of the locks and the adding of the transaction to the queue happen inside the same critical section."
            ]
        },
        "happen in": {
            "sentences": [
                "For example, the \u201cfencing\u201d action usually happens in a stadium, but it can happen in other places such as home or outdoors."
            ]
        },
        "happen to": {
            "sentences": [
                "The relation/edge indicates the changes that happens to the object from time t to time t+1."
            ]
        }
    },
    "hard": {
        "be hard to": {
            "sentences": [
                "Sometimes it is hard to make the decision."
            ]
        }
    },
    "harvest": {
        "harvest sth": {
            "sentences": [
                "To harvest statements of different difficulty levels, we design a two-channel collection process"
            ]
        }
    },
    "have": {
        "have (finally": {
            "sentences": [
                "\"An Old Idea Whose Time Has (Finally) Come CloudComputingisanewtermforalong-helddreamofcomputingasautility[35]"
            ]
        },
        "have a much": {
            "sentences": [
                "\"Although cloud functions have a much lower startup latency than traditional VM-based instances"
            ]
        },
        "have to": {
            "sentences": [
                "This quantity represent the number of times a particular person appears in the dataset, hence it is a good indication of how many opportunities that person will have to receive or transmit information.",
                "scans between successive frames have to overlap significantly to avoid incorrect ICP solutions"
            ]
        },
        "have effect on": {
            "sentences": [
                "An increase in domain size has opposite effects on the performance of OPT and PROB."
            ]
        },
        "have at least": {
            "sentences": [
                "These were then filtered down to keep only those image-sentence 928 pairs that have at least one verb, preposition, or adjective."
            ]
        },
        "have yet": {
            "sentences": [
                "several large-scale datasets for language have been collected semi-automatically by scraping billions of web pages at a large scale (brown et al., 2020; rae et al., 2021; raffel et al., 2019), but an equivalent multimodal dataset which induces similar few-shot abilities has yet to be collected."
            ]
        },
        "have access to": {
            "sentences": [
                "The query engine is blind to the encoding scheme used and has no direct access to the encoded data"
            ]
        },
        "have impact on": {
            "sentences": [
                "Note that the SQL collection process has minor impact on the workload performance"
            ]
        },
        "have be exploit in": {
            "sentences": [
                "\"Moreover, meta information has been exploited in many existing approaches for improving prediction accuracy.\""
            ]
        },
        "have be utilize in": {
            "sentences": [
                "Nowadays rich visual images have been utilized in more and more project profiles for attracting backers"
            ]
        },
        "have be conduct to": {
            "sentences": [
                "little work has been conducted to evaluate their effects towards success prediction"
            ]
        },
        "have be greatly improve": {
            "sentences": [
                "The steam generator design has been greatly improved."
            ]
        },
        "have a good agreement with": {
            "sentences": [
                "The corresponding results have a good agreement with the reactor dynamics."
            ]
        },
        "have a long": {
            "sentences": [
                "(Note that having a long-lived Paxos leader is critical toef\ufb01cientlymanagingthelocktable.)InbothBigtableandSpanner,wedesignedfor long-lived transactions (for example, for report generation, which might take on the order of minutes), which perform poorly under optimistic concurrency control in the presence of con\ufb02icts."
            ]
        },
        "have existence only": {
            "sentences": [
                "A row has existence only if some value (even if it is NULL) is de\ufb01ned for the row\u2019s keys."
            ]
        },
        "have not just": {
            "sentences": [
                "In addition to locally storing each key  potentially result in the system having not just two but several  within its range, the coordinator replicates these keys at the N-1  versions of the same data."
            ]
        }
    },
    "help": {
        "help with": {
            "sentences": [
                "the derivation assists in constructing the proposed framework and also helps with the corresponding analysis.",
                "Semantic types provide finergrained descriptions of the data by establishing correspondences between columns and real-world concepts and as such, can help with schema matching to determine which columns refer to the same real-world concepts, or data cleaning by determining the conceptual domain of a column."
            ]
        },
        "help to": {
            "sentences": [
                "\"the radix table helps to quickly locate the correct spline 20bitsneglectingcommonprefixbitssharedbyallkeys) points for a given lookup key.\"",
                "This is because TaBERT aims at semantic parsing over tables, its column embeddings can help to understand alignments between input texts and column schema.",
                "Such framework can help to well characterize the album representation.",
                "Their work helps to keep track of the evolution of data through time.",
                "Transformer helps to capture both tissue structure and cellular details of the image",
                "As an interest- ing by-product, we find that identifying and characterising unknown instances helps to reduce confusion in an incremental object detection setting",
                "helped to implement the sparse convolution and reviewed the paper.",
                "Such design makes the architecture more flexible and helps to exploit shared storage capabilities:",
                "Besides, it helps to reduce the accumulated estimation drift of camera pose",
                "Neigbors help to capture both feature points and objections of the image."
            ]
        },
        "help on": {
            "sentences": [
                "Unlike TicToc and MVCC, they help on every benchmark we evaluate, not just TPC-C; and they benefit TicToc and MVCC as well as OCC."
            ]
        },
        "help to further": {
            "sentences": [
                "Learning these correlations helps to further improvethe accuracy of queries that span more than two tables."
            ]
        },
        "help improve": {
            "sentences": [
                "furthermore, making the super-resolution models aware of the amount of noise added, via noise level conditioning, significantly improves the sample quality and helps improving the robustness of the super-resolution models to handle artifacts generated by lower resolution models [29]."
            ]
        },
        "with the help of": {
            "sentences": [
                "Coverage-based fuzzers try to solve the problem with the help of program analysis techniques, like program instrumentation."
            ]
        }
    },
    "hide": {
        "hide of": {
            "sentences": [
                "We analyze the different ob- fuscation techniques in relation to protection of intellectual property and the hiding of malicious code.",
                "We analyze the different obfuscation techniques in relation to protection of intellectual property and the hiding of malicious code."
            ]
        },
        "hide sth": {
            "sentences": [
                "Traditionally, the high cost ofnetwork communication between servers has hidden the impact ofcross-core coordination in replicated systems."
            ]
        }
    },
    "highlight": {
        "be highlight in": {
            "sentences": [
                "all memory accesses by a lookup are highlighted in red (blue) for PM (DRAM)."
            ]
        },
        "highlight the most": {
            "sentences": [
                "in the following, we highlight the most important related areas.",
                "Below, we highlight the most related areas."
            ]
        }
    },
    "highlighte": {
        "highlighte in": {
            "sentences": [
                "These relevant paragraphs are high- lighted in the annotation tool"
            ]
        }
    },
    "hinge": {
        "hinge on": {
            "sentences": [
                "the accuracy of our method hinges on the quality of the point clouds",
                "Our proposed method hinges on the notion of focal entities that we introduced in Section 1"
            ]
        }
    },
    "hint": {
        "hint sth": {
            "sentences": [
                "there is some common knowledge that hints when the workload may have significant changes"
            ]
        }
    },
    "hold": {
        "hold for": {
            "sentences": [
                "Note that since T already holds the lock for the data node",
                "The above procedure also holds for the candidate tail entity t"
            ]
        },
        "hold out": {
            "sentences": [
                "Additionally, we also hold out a small test set with 2K samples for human evaluation",
                "These methods hold out a portion of the input sequence and train models to predict the missing content"
            ]
        },
        "hold on": {
            "sentences": [
                "It aims to hold on the computation tasks.",
                "To hold on to one's views."
            ]
        },
        "hold an entire copy": {
            "sentences": [
                "Our system does require a single node to hold an entire copy of the database, as in many other modern transactional systems"
            ]
        },
        "hold for\u4fdd\u6301": {
            "sentences": [
                "Similar statements hold for the false positive rate of Bloome filter and the partitions."
            ]
        },
        "hold to": {
            "sentences": [
                "To consolidate the five skilled anno- tators\u2019 responses, we held them to an 80% agreement threshold: for each policy-question pair"
            ]
        }
    },
    "hope": {
        "hope to": {
            "sentences": [
                "target domains in the hope to reduce \u03b3 and \u03c1 on the representation space."
            ]
        }
    },
    "identical": {
        "identical to": {
            "sentences": [
                "Each lagged regressor module is functionally identical to the AR module, with\nthe only difference being the inputs."
            ]
        }
    },
    "identify": {
        "identify with": {
            "sentences": [
                "Then, we identify the named entities eq within the query q"
            ]
        },
        "identify at": {
            "sentences": [
                "according to [yao et al., 2019], above 40.7% of relations can only be identified at the document level."
            ]
        },
        "identify as": {
            "sentences": [
                "We identified a small number of data practices and information types that prior work identified as primary concerns for users.",
                "We materialize the transactions identified as bottlenecks in ARIES log so that transactions depending on them can be recovered more efficiently"
            ]
        },
        "identify the well": {
            "sentences": [
                "Be-nodes in the DAG, because the rewriting steps across different cause we mostly care about identifying the well-performing nodes usually have less dependency."
            ]
        },
        "identify and then": {
            "sentences": [
                "Traditional operator fusion algorithms resemble instruction selection: A sequence of operatorseligible for fusion is first identified and then replaced witha corresponding handwritten fused implementation, usuallyfrom a vendor-provided library."
            ]
        },
        "identify sth": {
            "sentences": [
                "This allows the DBMS to identify the log records to replay during recovery to restore the page."
            ]
        },
        "identify by": {
            "sentences": [
                "a smart contract is identified by a unique address which has its own money balance"
            ]
        }
    },
    "ignore": {
        "ignore for": {
            "sentences": [
                "postgres provided the best overall performance, followed by virtuoso (sql), which led us to conclude that rdbmses should not be ignored for interactive transactional graph workloads."
            ]
        },
        "ignore sth": {
            "sentences": [
                "The DBMS must ignore the effects of transactions that fall within such a gap while determining the tuple visibility."
            ]
        }
    },
    "ilies": {
        "ilies very much": {
            "sentences": [
                "cluster using consistent hashing [11] but uses an order pre- ilies very much similar to what happens in the Bigtable[4] serving hash function to do so."
            ]
        }
    },
    "illustrate": {
        "illustrate in": {
            "sentences": [
                "Several milestone methods are illustrated in Fig. 2.",
                "The problem setting is illustrated in the top row of Fig. 2.",
                "in this paper, we introduce flamingo, a visual language model (vlm) that sets a new state of the art in few-shot learning on a wide range of open-ended vision and language tasks, simply by being prompted with a few input/output examples, as illustrated in figure 1."
            ]
        },
        "illustrate the randomly": {
            "sentences": [
                "MOSE illustrates the randomly generated datamation accuracy thanks to the additional information about is not always valuable for the model and might even bethe data distribution."
            ]
        },
        "illustrate by": {
            "sentences": [
                "This operation is illustrated by the computation graph in Fig. 3(c)."
            ]
        }
    },
    "imagine": {
        "imagine that": {
            "sentences": [
                "Imagine that rather than viewing a relation R as a pool of |R| tuples that may be sampled, therelation is viewed as a pool of |R|b blocks that might be sampled, andthe sampling process is used to select those blocks at random, ratherthan tuples."
            ]
        }
    },
    "impact": {
        "impact of": {
            "sentences": [
                "However, each algorithm may not exceed main memory capacity, as we do not consider the impact of disk accesses in this work."
            ]
        },
        "impact sth. on": {
            "sentences": [
                "A single branch misprediction does not impact performance on the GPU."
            ]
        }
    },
    "impair": {
        "impair system performance": {
            "sentences": [
                "However, this design increases the latency of replication and impairs system performance"
            ]
        }
    },
    "implement": {
        "be implement as": {
            "sentences": [
                "indexes can be implemented as machine learning models that predict the location of target values given a search key."
            ]
        },
        "implement as": {
            "sentences": [
                "we created a reference implementation of the ldbc snb interactive workload in the gremlin language as well as sql.7 the gremlin implementation contains all the queries of the original benchmark specification implemented as gremlin traversals and helper utilities for graph loading."
            ]
        },
        "implement through": {
            "sentences": [
                "PointCNN [79] transforms the input points into a latent and potentially canonical order through ax-conv transformation (which is implemented through MLP) and then applies typical convolutional operator on the transformed features."
            ]
        },
        "implement to": {
            "sentences": [
                "For this reason, obfuscation techniques are implemented with other approaches, such as code replacement/update, code tampering detection, protections updating (by that the attackers get a limited amount of time to complete their objective) etc."
            ]
        },
        "implement by": {
            "sentences": [
                "This form can be easily implemented by a graph to apply graph algorithms of probability graph and graph theory to solve problems",
                "\"the recurrence is implemented by e.g. an lstm (wang et al. 2016) or a gru (duan et al. 2016).\""
            ]
        },
        "implement base on": {
            "sentences": [
                "4.2 experimental settings our model was implemented based on pytorch."
            ]
        },
        "implement in": {
            "sentences": [
                "the database primitives that we implement in this work, and optimizations related to data transfers."
            ]
        },
        "be implement with": {
            "sentences": [
                "The policy model is implemented with a multiple-RNN architecture."
            ]
        },
        "implement sth. to do": {
            "sentences": [
                "We implement individual query operators to show that while the speedups for selection, projection, and sorts are near the bandwidth ratio, joins achieve less speedup due to differences in hardware capabilities."
            ]
        },
        "implement in approximately": {
            "sentences": [
                "For example, TensorFlow r1.14includes 155 substitutions implemented in approximately53K lines of C++ code."
            ]
        },
        "implement sth in": {
            "sentences": [
                "To evaluate our approach, we implemented it in the Peloton in-memory DBMS and compared it against WAL using three storage technologies: NVM, SSD, and HDD."
            ]
        },
        "implement sth": {
            "sentences": [
                "We have implemented the modules like candidate index generation and index estimatorin an open-sourced database system openGauss"
            ]
        },
        "implement with": {
            "sentences": [
                "We have implemented our method and all the recogni- tion tasks with PyTorch [40]."
            ]
        },
        "implement a consistently": {
            "sentences": [
                "The Paxos state machines are used to implement a consistently replicated bag of mappings."
            ]
        }
    },
    "imply": {
        "be imply by": {
            "sentences": [
                "Because Phase2b in Paxos is implied by multiple sub-actions in Raft*"
            ]
        }
    },
    "imporve": {
        "improve in": {
            "sentences": [
                "Surprisingly, the radix join brings a noticeable improvement in only one out of all 59 joins in TPC-H."
            ]
        }
    },
    "impose": {
        "impose long delay": {
            "sentences": [
                "Second, the communication between primaries and replicas not only imposes long delays to the critical path of transactions, but perhaps more importantly"
            ]
        }
    },
    "improve": {
        "have improve": {
            "sentences": [
                "This has greatly improved performance of OLTP database systems, since disk IO is eliminated as a bottleneck."
            ]
        },
        "improve to": {
            "sentences": [
                "Note that these methods are based on the vanilla\nTransformer and try to improve the self-attention mechanism to a sparse version, which still follows\nthe point-wise dependency and aggregation."
            ]
        },
        "improve upon": {
            "sentences": [
                "Recently, AFLFast by Bohme et al.  has improved upon AFL in each of the three aspects above.",
                "To improve upon the single-modal framework, optical flow is commonly used as the temporal cue to greatly improve the action recognition accuracy."
            ]
        },
        "improve  forward": {
            "sentences": [
                "in such lifelong learning settings, meta-learning has great  potential  for  improving  forward  transfer  to  new  tasks."
            ]
        },
        "improve the performance by": {
            "sentences": [
                "GRAPPA (MLM) usually improves the performance by around 1% such as 1.4% gain on SPIDER (dev), 0.8% on WIKITABLEQUESTIONS, and 1.2% on weakly-sup WIKISQL."
            ]
        },
        "improve over": {
            "sentences": [
                "Our model achieves 28.4 BLEU on the WMT 2014 Englishto-German translation task, improving over the existing best results, including ensembles, by over 2 BLEU."
            ]
        },
        "imporve from ... to ...": {
            "sentences": [
                "Comparing to training from scratch, us- ing the MOTS pre-trained weights contributes to a substan- tial performance gain, improving the average Dice from 85.30% to 86.44%, reducing the average mean surface dis- tance (SD) from 1.46 to 1.17, and reducing the average HD from 19.67 to 15.62"
            ]
        },
        "improve on": {
            "sentences": [
                "Unlike existing works that designed CNN architectures to improve performance on a single task of a single domain and not generalizable.",
                "but also offers insights for platform developers: helping them to identify and improve on the performance bottlenecks"
            ]
        },
        "improve sth via": {
            "sentences": [
                "F1 Lightning improves the data freshness via the integration with a Change Data Capture (CDC) mechanism."
            ]
        },
        "improve by": {
            "sentences": [
                "These experiments show that WBL with NVM improves the DBMS\u2019s throughput by 1.3\u00d7 while also reducing the database recovery time and the overall system\u2019s storage footprint."
            ]
        },
        "improve with": {
            "sentences": [
                "The performance of deep neural networks improves with more annotated data."
            ]
        }
    },
    "in-depth": {
        "in-depth search": {
            "sentences": [
                "readers may wish  to see for an in-depth survey of LSM-trees"
            ]
        }
    },
    "inaccurate": {
        "be inaccurate for": {
            "sentences": [
                "Since the models might be inaccurate for the predicted positions, the learned index needs to search the lookup key in a bounded range around the predicted positions."
            ]
        }
    },
    "include": {
        "include in": {
            "sentences": [
                "The decoupled-based model included in the first category guarantees the independence of knowledge",
                "right: an example of the handshaking tagging scheme, the region in shade is not included in the tag sequence.",
                "Note that we only evaluate test cases that are not included in the training of our predictive model."
            ]
        },
        "include compilation overhead": {
            "sentences": [
                "However, existing AI compiler optimizationsystems suffer a lot from problems brought by dynamic shapemodels, including compilation overhead, memory usage, optimization pipeline and deployment complexity."
            ]
        },
        "include open": {
            "sentences": [
                "these include open-ended tasks such as visual question-answering, where the model is prompted with a question which it has to answer, captioning tasks, which evaluate the ability to describe a scene or an event, and close-ended tasks such as multiple choice visual question-answering."
            ]
        },
        "be include in": {
            "sentences": [
                "In general, a lower entropy value means less information is included in the dataset"
            ]
        },
        "include \u2026 as well as \u2026 and \u2026": {
            "sentences": [
                "The paper includes the configuration as well as the structural design and the results of parameter structural optimization analysis."
            ]
        },
        "include  most": {
            "sentences": [
                "Although  each  instance  still  includes  most  of  the  One approach to tolerate failures in a replicated system is to use a  components of a traditional kernel (query processor, transactions,  quorum-based voting protocol as described in [6]."
            ]
        },
        "include memory however": {
            "sentences": [
                "we performed several additional optimizations, including memory However, it performs very well in practice since very wide rela- alignment, narrow-row padding and cache-set collision avoidance, tions typically consist of a small number of sets of attributes that asdescribedinAppendixA. are frequently accessed together (thus, creating a small number of To evaluate our model we choose a set of queries derived from primarypartitions)andsinceoperationsacrossthosepartitionsare anSAPenterpriseresourceplanning(ERP)applicationthatincludes oftenrelativelyinfrequent(thus, drasticallylimitingthenumberof severalanalyticalqueriesthatmodelreportingovertherecenthistory newpartitionsgeneratedbythesecondphaseabove)."
            ]
        }
    },
    "incorporate": {
        "incorporate sth": {
            "sentences": [
                "A field relevance model is proposed by Kim and Croft to incorporate relevance feedback for field weights estimation."
            ]
        },
        "incorporate into": {
            "sentences": [
                "[31] is one of early works that incorporates MMD into DCNNs with regular supervised classification loss on source domain to learn both semantically meaningful and domain invariant representation.",
                "To the best of our knowledge, this is thefirst time to incorporate attention mechanisms into federated optimization.",
                "To support the use of Sherlock 'out-of-the-box,' we distribute Sherlock as a Python library that can be easily installed and incorporated into existing codebases.",
                "it is incorporated into the global map",
                "We propose a federated contrastive learning method for recommendation, which can incorporate various kinds of negative samples into contrastive model training in a privacy-preserving way"
            ]
        },
        "incorporate sth. in": {
            "sentences": [
                "As the major technical contribution, we incorporate a filter component in each stacked block"
            ]
        },
        "incorporate from": {
            "sentences": [
                "Xu et al. (2018) incorporate signals from social media"
            ]
        },
        "incorporate structural prior": {
            "sentences": [
                "the improvements include incorporating structural prior, designing lightweight architecture, pre-training, and so on."
            ]
        },
        "incorporate xxx from": {
            "sentences": [
                "We believe that incorporating rules from a broader range of datasets (e.g. WIKITABLEQUESTIONS) would further improve the performance."
            ]
        },
        "incorporate in": {
            "sentences": [
                "TVM can be incorporated in our system to tune the query optimizer as well as the compiler optimizations used by the different device back-ends.",
                "A candidate protocol incorporated in CormCC should be COCSR and strict."
            ]
        },
        "incorporate to": {
            "sentences": [
                "Under the SAR assumption, in a similar fashion, the propensity score can be incorporated to enable learning"
            ]
        }
    },
    "increase": {
        "increase the difficulty": {
            "sentences": [
                "We increase the difficulty by decaying margin values."
            ]
        },
        "increase from...to": {
            "sentences": [
                "As a tendency the lines for EXACT and OPT get closer as the domain size increases from 10 to 200, while the lines for PROB and OPT become more and more separated."
            ]
        },
        "increase from": {
            "sentences": [
                "Different from [5], we increase the number of height channels from 4 to 16 to capture more detailed shape information\u2013 further increasing the number of height channels did not lead to performance improvement."
            ]
        },
        "increase with": {
            "sentences": [
                "However by using self-modifying code the in- struction set is increased with the set of bytes that can be written as an exclusive or (XOR) of a printable charac- ter with another one, since instruction XOR is within the printable character set.",
                "The relatively low performance of crowdsourced consensus is perhaps due to the visual overload of selecting from 78 types, such that performance may increase with a smaller number of candidate types."
            ]
        },
        "increase from ... to ...": {
            "sentences": [
                "However, the channel of input increases from 1 to m + 1, leading to a dramatic increase of computation and memory cost"
            ]
        },
        "increase in": {
            "sentences": [
                "it seems that our tagging scheme is extremely inefficient because the length of the tagging sequence increases in a square number with increasing sentence length."
            ]
        },
        "increase the number": {
            "sentences": [
                "As we increased the number of threads further, the RJ slowed down because of memory bandwidth contention."
            ]
        }
    },
    "increment": {
        "increment sth. with sth.": {
            "sentences": [
                "After processing all entries in a tile, we use BlockAgg to aggregate the local sums within a thread block into a single value and increment a global sum with it."
            ]
        },
        "increment sth": {
            "sentences": [
                "Without this information, the DBMS cannot increment cp before the transaction commits."
            ]
        }
    },
    "incur": {
        "incur low overhead": {
            "sentences": [
                "Thecoef\ufb01cientsofthe it incurs low overhead compared to monitoring low-level polynomial are computed using pseudo-inverse solution hardware utilization to trigger the scale-out decisions."
            ]
        },
        "incur at": {
            "sentences": [
                "irrespective of implementation, refinable timestamps rep\u0002resent a hybrid approach to timeline ordering that offers an interesting tradeoff between proactive costs due to periodic synchronization messages between gatekeepers, and the reactive costs incurred at the timeline oracle."
            ]
        },
        "incur a prohibitively": {
            "sentences": [
                "As joins tend to involve columns with high cardinalities, an AR model built on aStoragejoin may incur a prohibitively large size."
            ]
        }
    },
    "index": {
        "index by": {
            "sentences": [
                "the representation consists of two arrays, the first storing the adjacency lists of all vertices as sequences of destination vertex ids, while the second storing pointers to the first array, indexed by source vertex id.",
                "The structure of octreeis encoded efficiently using a bit string representation, and the feature vector of each voxel is indexed by simple arithmetic."
            ]
        },
        "index with": {
            "sentences": [
                "This task is useful when indexing tables with missing column names, which happens relatively often in practice."
            ]
        }
    },
    "indicate": {
        "indicate (that)": {
            "sentences": [
                "Our results indicate that the use of multiple tables can provide contextual clues to the model even when those tables do not explicitly verify or refute the claim, because they can provide evidence for the probability of the claim."
            ]
        },
        "indicate that": {
            "sentences": [
                "The results indicate that the addition of a nucleating agent and a thickening agent can inhibit supercooling and eliminate segregation.",
                "The results indicate that this method can improve the sensitivity and reproductivity.",
                "This third state, \u201cwaiting\u201d, indicates that a transaction was previously executing but could not complete without the result of an outstanding remote read request."
            ]
        },
        "indicate how": {
            "sentences": [
                "Adequacy refers to the semantic correctness of the captions, disregarding the sentence grammar structure, whereas, fluency indicates how grammatically well-formed a sentence is."
            ]
        },
        "indicate strongattribute only": {
            "sentences": [
                "The bold edges between two groups indicate strongAttribute Only k=4 k=5 k=6 k=7 DB 0.84"
            ]
        }
    },
    "induce": {
        "induce by": {
            "sentences": [
                "Different Riemannian metrics can be induced by appropriately selecting the arc length."
            ]
        },
        "induce to": {
            "sentences": [
                "The answer is to induce appearance invariance into CNNs."
            ]
        }
    },
    "infer": {
        "infer from": {
            "sentences": [
                "all of which are inferred from the dataset",
                "Previous works demonstrated that the relationship of two entities could be inferred from paths connecting them",
                "It is easier to infer user private information from representations that can recall more users\u2019 historical clicked"
            ]
        },
        "infer by": {
            "sentences": [
                "Besides, the MF can also be inferred by estimating the vanishing point"
            ]
        }
    },
    "infilling": {
        "infilling in": {
            "sentences": [
                "besides the pure gpt-style generation, cogview2 [6] proposes a new language model coglm for infilling in the image generation."
            ]
        }
    },
    "inform": {
        "inform with": {
            "sentences": [
                "In applications, where forecasts inform business or operational decisions with potentially fatal consequences"
            ]
        }
    },
    "infuse": {
        "infuse into": {
            "sentences": [
                "SenseBERT [59] infuses word-sense information into BERT\u2019s pre-training signal"
            ]
        }
    },
    "ingest": {
        "ingest so far": {
            "sentences": [
                "The same query generator as macrobenchmarks is usedwhere the filters are drawn from tuples in the first partition(true selectivities computed on all data ingested so far)."
            ]
        },
        "ingest into": {
            "sentences": [
                "Using offline ETL to ingest data into columnar file formats tends to suffer from high lag between the OLTP data and the OLAP copy."
            ]
        }
    },
    "inherit": {
        "inherit the advantage": {
            "sentences": [
                "This special class of BDL, known as Neural Network Gaussian Process (NNGP), inherits the advantages of Bayesian approachwhile keeping universal approximation of neural network, and can utilize a much larger model space to model distribution-free dataas a nonparametric model."
            ]
        },
        "inherit from": {
            "sentences": [
                "Since Taiji inherits the parallel-chain structure from Prism, our meta-principles will also apply."
            ]
        }
    },
    "initialize": {
        "initialize with": {
            "sentences": [
                "the weight is initialized with kaiming\nnormal (He et al. 2018).",
                "Joint training KEPTMs initialize triplets embeddings with text encoder"
            ]
        },
        "be initialize by": {
            "sentences": [
                "GRAPPA is initialized by RoBERTaLARGE (Liu et al., 2019) and further pre-trained on the synthetic data with SQL semantic loss and table-related data with MLM loss."
            ]
        },
        "initialize of": {
            "sentences": [
                "For all experiments we initialize the parameters of conv1\u2013 fc7 using the released CaffeNet weights."
            ]
        },
        "initialize by": {
            "sentences": [
                "The classification model is initialized by fitting a balanced model which considers the unlabeled examples as negative."
            ]
        }
    },
    "initiate": {
        "initiater sth. by sth.": {
            "sentences": [
                "The CPU then initiates a PCI-e transfer by accessing the mapped GPU memory with normal load and store instructions."
            ]
        }
    },
    "inject": {
        "inject into": {
            "sentences": [
                "We pre-train GRAPPA on the synthetic data to inject important structural properties commonly found in table semantic parsing into the pre-training language model."
            ]
        },
        "inject by": {
            "sentences": [
                "Entity enhanced pre-trained models focus on entitylevel knowledge and inject knowledge by infusing entity information and language representations"
            ]
        }
    },
    "input": {
        "input to": {
            "sentences": [
                "In both CPUs and GPUs, similar algorithmic approaches can mitigate the problem by, for example, partitioning the input to fit the per-partition hash-tables in a memory (cache) with a higher bandwidth."
            ]
        }
    },
    "insert": {
        "insert to": {
            "sentences": [
                "To insert data into Viper, clients must issue a put(Key, Value) request.",
                "Once this is done, we have to determine which junk instructions to insert so as to confuse the disassembler as much as possible and delay resynchronization for as long as possible.",
                "Once this is done, we have to determine which junk instruc- tions to insert so as to confuse the disassembler as much as possible and delay resynchronization for as long as possible."
            ]
        },
        "insert into": {
            "sentences": [
                "We maintain a bitmap \ud835\udc35, compute a hash value for each record and insert a bit into the corresponding location of the bitmap",
                "If the number of tuples in the reservoir is less than the reservoir size |R|, we insert the tuple into the reservoir (line 5 to 6).",
                "We insert 10 gigabytes records into this table"
            ]
        }
    },
    "insist": {
        "insist on": {
            "sentences": [
                "Considering the stability and for a fair comparison, we insist on using the same batch size for both the real and generated data in the above experiments."
            ]
        }
    },
    "inspire": {
        "inspire by": {
            "sentences": [
                "Our algorithm is inspired by KeOps.",
                "Inspired by linguistic regularities",
                "we propose a new probe-and-stash mechanism inspired by recent PM hash tables",
                "Inspired by the recent advances in image representation using deep convolutional neural networks (DCNNs), a few deep architecture based methods have been proposed for unsupervised domain adaptation.",
                "The architecture of our network (Sec4.2) is inspired by the properties of point sets inRn(Sec4.1).",
                "Inspired by previous work on data synthesis for semantic parsing (Berant & Liang, 2014; Wang et al., 2015b; Jia & Liang, 2016; Herzig & Berant, 2018; Andreas, 2020), we induce a synchronous context-free grammar (SCFG) specific to mapping natural language to SQL queries from existing text-to-SQL datasets, which covers most commonly used question-SQL patterns.",
                "Inspired by Prophet\u2019s impact on the forecasting community, it is our objective\nto continue the democratization of forecasting tooling, by making hybrid models as accessible as\nProphet made classic models.",
                "Inspired by this strong regularity of man-made buildings, we propose a novel visual SLAM method",
                "inspired by the above, we propose a novel model called document u-shaped network (docunet), which formulates document-level re as semantic segmentation.",
                "Inspired by curriculum learning, we adopt an easy-to-hard strategy on training DICE."
            ]
        },
        "be inspire by": {
            "sentences": [
                "Inspired by variance reduction, [Liang et al., 2019] proposed a similar method.",
                "We are inspired by prior work on distilling deep models [3, 16] and extend the ideas presented in these works to a domain adaptation setting."
            ]
        },
        "inspire to": {
            "sentences": [
                "Such different device preferences inspire us to perform fine-grained stream processing on integrated architectures"
            ]
        },
        "inspire sth": {
            "sentences": [
                "These works inspire us to solve the resource-oriented tuning problem"
            ]
        }
    },
    "instantiate": {
        "instantiate that": {
            "sentences": [
                "we instantiate both the encoder and decoder as neural networks"
            ]
        }
    },
    "instead": {
        "instead of": {
            "sentences": [
                "neural networks can express more possible features with distributed vectors instead of sparse vectors",
                "In our method, the direction classification weight can be controlled to make the network focus more on overlap maximization  instead of struggling with  direction recognition in difficult cases.",
                "Instead of performing image feature detection, description, and matching sequentially, we propose to first establish pixel-wise dense matches at a coarse level and later refine the good matches at a fine level."
            ]
        }
    },
    "instructe": {
        "be instruct to": {
            "sentences": [
                "Users are also instructed to ignore statements applying to a limited audience"
            ]
        }
    },
    "instrumentalized": {
        "instrumentalized to": {
            "sentences": [
                "however, it should be pointed  out that lif is still an experimental framework and is instrumentalized to quickly evaluate different index  con gurations (e.g., ml models, page-sizes, search strategies, etc."
            ]
        },
        "instrumentalized to quickly": {
            "sentences": [
                "however, it should be pointed out that lif is still an experimental framework and is instrumentalized to quickly evaluate different index con\ufb01gurations (e.g., ml models, page-sizes, search strategies, etc."
            ]
        }
    },
    "integrate": {
        "integrate with": {
            "sentences": [
                "Symbolic representations have also been integrated with machine learning especially in statistical relational learning",
                "But there is still a lack of research on the comprehensive quality management over online stream processing, especially to integrate with the characteristics of distributed computing environments.",
                "The macro-sector contexts are integrated with the micro-stock context and a self-supervised\ntask captures their couplings.",
                "We have developed a novel approach for annotating various types of table data in which probabilistic-model-based annotation is integrated with highly predictive multi-label classifiers.",
                "ERNIE [85] integrates entity embeddings pre-trained on triplets with corresponding entity mentions in the text",
                "To further improve the utility-privacy trade-off, we integrate the sparsification-coded DP mechanisms with convergence acceleration techniques.",
                "symbolic knowledge can be much easier to integrate with the neural network"
            ]
        },
        "integrate to": {
            "sentences": [
                "Any private blockchain can be integrated to Blockbench via simple APIs and benchmarked against workloads that are based on real and synthetic smart contracts",
                "Then, the glyph features of Chinese characters and corresponding character embeddings are integrated to enrich the representation of Chinese characters",
                "It tightly integrates to the ORB-SLAM2"
            ]
        },
        "integrate into": {
            "sentences": [
                "as such, there are a multitude of data challenges that must be addressed before text-to-image models like imagen can be safely integrated into user-facing applications.",
                "We integrated the relevance model into our privacy policy an- notation tool by color-highlighting the top k-relevant paragraphs in each policy, as shown in Figure 4.",
                "Results based on MF and GCN illustrate that DICE is a general framework, which can be smoothly integrated into various embedding based recommendation algorithms.",
                "And the activation response on global feature map and positive proposal confidence are integrated into the final localization score",
                "Furthermore, to control the complexity and diversity of the generated text, we integrate three types of control codes into the model input:",
                "We make our results publicly available and demonstrate how these results can easily be integrated into existing state-of-the-art analysis tools.",
                "We integrate multiple most representative CardEst methods into an open-source database system PostgreSQL",
                "When integrated into state-of-the-art networks, our approach significantly outperforms the feature correlation layer for the tasks of geometric matching, optical flow, and dense semantic matching.",
                "The significance of our results is that we have developed a simple way in which sophisticated streaming algorithms that returns set results can be integrated into a query system.",
                "When integrated into four recent semantic networks, our approach sets a new state-of-the-art on four challenging benchmarks",
                "The proposed method performs matching in real-time on a modern GPU and can be readily integrated into modern SfM or SLAM systems"
            ]
        },
        "integrate sth. with sth.": {
            "sentences": [
                "Overall, fast interconnects integrate GPUs tightly with CPUs and significantly reduce the data transfer bottleneck."
            ]
        },
        "be integrate into": {
            "sentences": [
                "L is tightly integrated into the hash function",
                "The GPU is integrated into a system-wide address space.",
                "which can be integrated into this work"
            ]
        },
        "integrate them seamlessly": {
            "sentences": [
                "For example, we fuse the element-wise nodesand integrate them seamlessly with existing rules."
            ]
        },
        "integrate sth into": {
            "sentences": [
                "To achieve stronger privacy protection, a few recent works start to integrate local differential privacy (LDP) into federated learning.",
                "The article analyses policies of the world largest oil and gas companies to integrate blockchain into their business structure"
            ]
        }
    },
    "intend": {
        "intend to": {
            "sentences": [
                "It is worth remarking that, of course, this does not intend to be a real-world example of a flight scheduling graph, but a simplified portion of it.",
                "which is intended to be compiled into the  database process and replace the file system interfaces provided by operating system,",
                "\ufeffThus, we intend to map each input triplet to an enriched feature space by training the network simultaneously on all of the inputs \ufeffavailable (and the similarity scores between them), thereby ensuring an enhanced extraction of semantic information from the inputs..",
                "We do not intend to fuse in additional information from sources like images, but instead, all of our proposed tokenzation methods are drop-in replacements to the existing single-character tokenizers, without adding any extra layers or parameters"
            ]
        },
        "intend for": {
            "sentences": [
                ". This method is intended for the one-class classification setting where the distribution of negative examples can be diferent at test time.",
                "In contrast, TABBIE is pretrained strictly on tabular data and intended for more general-purpose tablebased tasks, and uses corrupt-cell classification as its pretraining task.",
                "HistogramWe construct a histogram QuadHist, intended for low-dimensional data and queries."
            ]
        },
        "intend to do": {
            "sentences": [
                "there are two sources of information that we intend to preserve for the local vector."
            ]
        }
    },
    "intensify": {
        "be intensify by": {
            "sentences": [
                "The capacity limitation is inten- sified by database operators that need additional space for intermediate state."
            ]
        }
    },
    "intepret": {
        "interpret as": {
            "sentences": [
                "we can interpret \u03b8 as a distribution over latent topics"
            ]
        }
    },
    "inter": {
        "inter in": {
            "sentences": [
                "he died near , , and is interred in a private cemetery near ."
            ]
        }
    },
    "interact": {
        "interact with": {
            "sentences": [
                "By interacting with the environment designed around the KG, the agent learns to pick the promising reasoning paths",
                "Unfortunately, object detection has some unique characteristics that interact poorly with such methods",
                "It works based on the DRL concept where an agent interacts with the environment",
                "for read only queries, operation executors can directly interact with the sut, whereas updates are processed through a dedicated kafka.",
                "Their plan gen- erator interacts with a cost model that estimates the query cost by considering the potential reuse ability of its sub- queries.",
                "With the use of Block chain technology, the BPIIoT platform enables peers in a decentralized, trustless, peer-to-peer network to interact with each other without the need for a trusted intermediary.",
                "In (D)RL, an agent interacts with environment by actions and re\u0002wards.",
                "In symbolic execution, when target program execution interacts with components out of the symbolic execution environments, such as system calls, handling signals, etc., consistency problems may arise.",
                "The question planning component interacts with human domain experts to verify such elements and the checking results, optimizing verification tasks for maximal benefit.",
                "Transactions are executed to the completion without interacting with clients"
            ]
        },
        "interact  with": {
            "sentences": [
                "7.1.6.3   real-world placing the goal of our \ufb01nal experiment is to evaluate how well a real robot can learn to learn to  interact  with  new,  unknown  objects  from  a  single  visual  demonstration."
            ]
        }
    },
    "interest": {
        "be interest in": {
            "sentences": [
                "For example, mobile phone users who read news articles may be interested in different categories of news like politics, sports or fashion."
            ]
        }
    },
    "interface": {
        "interface to": {
            "sentences": [
                "Interface to a highly sparse LiDAR point cloud with a region proposal network"
            ]
        }
    },
    "interfere": {
        "interfere with": {
            "sentences": [
                "synchronicity assumptions interfere with debugging, and maybe violated by network delays under heavy load and systems running in virtualized environments.",
                "Test inputs do not interfere with each other, increasing the reproducibility of bugs",
                "in this way, bidirectional attention context is fully exploited in text and given frames without interfering with auto-regressive frame prediction."
            ]
        }
    },
    "interleave": {
        "interleave sth. into sth.": {
            "sentences": [
                "As a result, for each set of 8 keys, we do 2 SIMD gathers and then de-interleave the columns into to 8 keys and 8 values."
            ]
        },
        "interleave with": {
            "sentences": [
                "the difference from a lm is that the model must be able to ingest a multimodal prompt containing both image and/or videos interleaved with text."
            ]
        }
    },
    "interpolate": {
        "interpolate\u2026to": {
            "sentences": [
                "By interpolating point features to neighboring discrete convolutional kernel-weight coordinates, Mao et al. [80] proposed an interpolated convolution operator InterpConv to measure the geometric relations between input point clouds and kernel-weight coordinates."
            ]
        }
    },
    "interpret": {
        "interpret as": {
            "sentences": [
                "3D convolution can be interpreted as a weighted sum over a given subset.",
                "Therefore, our LBE algorithm can also be interpreted as a one-sided label noise learning problem as mentioned in Section 2."
            ]
        },
        "interpret as simply": {
            "sentences": [
                "the role of the adaptation loss therefore may be interpreted as simply directing the policy parameter update to modify the policy to pick up on the right visual cues in the scene, so that the meta-trained action output will produce the right actions."
            ]
        },
        "be interpret as": {
            "sentences": [
                "These pages are interpreted as one leader record plus data blocks for each database node."
            ]
        }
    },
    "interprete": {
        "interprete as": {
            "sentences": [
                "These positions can be\ninterpreted as the locations with an auto-correlation"
            ]
        }
    },
    "intersect": {
        "intersect with": {
            "sentences": [
                "The algorithm checks if the node in the opposite end of the edge is in the path, or the interval in the edge does not intersect with the interval in the triplet."
            ]
        }
    },
    "introduce": {
        "introduce for": {
            "sentences": [
                "To achieve this goal, it is necessary to introduce structural priors for time series.",
                "Spatially sparse convolutional networks are introduced for LiDAR-based detection and are used to extract information from the z-axis before the 3D data are downsampled to  something akin to 2D image data.",
                "we introduced a technique for sampling ground truths and their associated point cloud data to construct  better training data on the fly"
            ]
        },
        "introduce into": {
            "sentences": [
                "we randomly selected several ground truths from this database and introduced them into the current training point cloud  via concatenation",
                "As is noted in 7.2, this step can also be part of CONFUPDATE, which is useful for fuzzers that can introduce new seeds into the seed pool throughout the campaign.",
                "Objects sampled from this  database are then introduced into the point clouds during training."
            ]
        },
        "introduce with": {
            "sentences": [
                "Local context\nis introduced with auto-regression and covariate modules, which can be configured as classical\nlinear regression or as Neural Networks."
            ]
        },
        "introduce incorporate": {
            "sentences": [
                "a line of work is dedicated to improving multi-head mechanism by introducing incorporating more sophisticated mechanisms that guide the behavior of different attention heads or allow interaction across attention heads."
            ]
        },
        "introduce in": {
            "sentences": [
                "Then, more partition conflicts are introduced in the workload",
                "There are a number of parameters introduced in GPL.",
                "to offer such an ability, researchers propose to replace convolutions by the transformer, a module which is first introduced in the field of natural language processing."
            ]
        },
        "introduce a filter": {
            "sentences": [
                "we introduced a filter for the probe phase that drops tuples as early as possible to save computation time and reduce unnecessary materialization overhead."
            ]
        },
        "introduce sth.": {
            "sentences": [
                "Liuetal.[23] further improve this model by introducing a more practical text detector."
            ]
        },
        "introduce to": {
            "sentences": [
                "The changes we introduced to Sherlock was helpful in increasing the model's representation power and generalisation capability.",
                "For a particular domain classifier, \u03b8D, we can now introduce our loss which seeks to \u201cmaximally confuse\u201d the two domains by computing the cross entropy between the output predicted domain labels and a uniform distribution over domain labels."
            ]
        },
        "introduce them below": {
            "sentences": [
                "We introduce them below."
            ]
        },
        "introduce as": {
            "sentences": [
                "all models introduced treats triplets as a training unit during KRL procedure"
            ]
        }
    },
    "introducethe": {
        "introducethe evaluation algorithm": {
            "sentences": [
                "In this section, we introducethe evaluation algorithm for After we get the maximum (A;R)-compatible grouping, the SNAP operation."
            ]
        }
    },
    "invent": {
        "invent an entirely": {
            "sentences": [
                "Rather than invent an entirely newlanguage, Relays IR design is based on IRs used by the wellstudied ML family of functional programming languages (e.g.,SML and OCaml)."
            ]
        }
    },
    "investigate": {
        "investigate sth": {
            "sentences": [
                "This paper investigates a very important yet previously unexplored research problem"
            ]
        },
        "be investigate for": {
            "sentences": [
                "Speed of a video is investigated for self-supervised and semi-supervised learning unlike the problem we consider in this paper."
            ]
        },
        "investigate the policies": {
            "sentences": [
                "Therefore, we investigate the following enumeration policies, commonly used for filter ordering."
            ]
        },
        "investigate through": {
            "sentences": [
                "we investigate their performance through a case study on the DVD domain in sentiment analysis."
            ]
        },
        "investigate + clause": {
            "sentences": [
                "In this paper, we investigate how a fast interconnect can resolve these scalability limitations using the example of NVLink 2.0."
            ]
        },
        "investigate in": {
            "sentences": [
                "this method was in fact used but not emphasized in previous work [28], and to our knowledge its importance has not been investigated in the context of guided sampling."
            ]
        },
        "investigate from": {
            "sentences": [
                "First, we investigate the accuracy of crowdsourcing the extrac- tion of data practices from privacy policies by comparing crowd- worker annotations with those of skilled annotators on a set of 26 privacy policies"
            ]
        }
    },
    "invoke": {
        "invoke by": {
            "sentences": [
                "The additional memory accesses, cache misses, CPU cycles, and critical sections invoked by lock manager operations can approach or exceed the costs of executing  the actual transaction logic."
            ]
        }
    },
    "involve": {
        "involve in": {
            "sentences": [
                "Therefore, both works assign static weights to neighbors, leading to static entity representations when involved in different task relations",
                "it is vital to build indexes (e.g., B+Tree) on columns involved in the query",
                "Specifically, all the entity mentions involved in the sentence are selected out to query corresponding triples in KGs",
                "entity mentions and relations only involved in these relation instances are colored.",
                "Let i, j be two records and eij be the access cost when they are involved in a single transaction"
            ]
        },
        "involve sth": {
            "sentences": [
                "Performing fact verification based on structured data is important for many real-life applications and is a challenging research problem, particularly when it involves both symbolic operations and informal inference based on language understanding."
            ]
        },
        "involve do": {
            "sentences": [
                "For distributed transactions that perform remote reads, this may involve sleeping for some period of time while waiting for another partition to send the read results over the network."
            ]
        },
        "involve into": {
            "sentences": [
                "Methods using LiDAR [18] involve the conversion of point clouds into front-view 2D maps"
            ]
        }
    },
    "invovle": {
        "invovle in": {
            "sentences": [
                "In practice, the number of parties involved in the vertical federated learning is limited in most scenarios"
            ]
        }
    },
    "isolate": {
        "isolate from": {
            "sentences": [
                "Chaincode can only access its private storage and they are isolated from each other",
                "In the past, data warehousing solutions were well-hidden back-end systems, isolated from most of the world."
            ]
        }
    },
    "issue": {
        "issue before": {
            "sentences": [
                "to order the new timestamps with respect to timestamps issued before the failure, the vector clocks in weaver include an extra epoch field which the cluster manager increments on failure detection."
            ]
        },
        "issue sth": {
            "sentences": [
                "we will issue an index tuning request to the Index Recommendation module"
            ]
        }
    },
    "iterate": {
        "iterate on": {
            "sentences": [
                "For the silver LF generation, we iterated on a small subset of the datasets.",
                "we iterate on the rules and use  the stored spatial indexes to obtain the output index for each input index"
            ]
        },
        "iterate over": {
            "sentences": [
                "The threads iterate over each tile independently to find matching entries from the hash table.",
                "we iterate over training a model,"
            ]
        },
        "iterate with": {
            "sentences": [
                "It only returns the latter interval, and the algorithm iterates with the last interval in the path."
            ]
        }
    },
    "join": {
        "join with": {
            "sentences": [
                "The problem with this approach is that the entire reservoir needs to be scanned in order to find tuples which can be joined with the newly arrived tuple."
            ]
        },
        "hash join": {
            "sentences": [
                "An efficient implementation of a hash join has been a highly researched problem for decades."
            ]
        }
    },
    "jump": {
        "jump on": {
            "sentences": [
                "he jumps on and begins a romantic adventure through space."
            ]
        },
        "jump with": {
            "sentences": [
                "In the above example, in place of JLE, jump with no carry (JNC) is used. If the carry flag is clear, then JNC transfers the control to the level."
            ]
        },
        "jump between": {
            "sentences": [
                "While annotators must answer all ques- tions before they can complete a policy annotation task, they can jump between questions, answer them in any order, and edit their responses until they submit the task."
            ]
        }
    },
    "keep": {
        "keep away from": {
            "sentences": [
                "Keep away from the blazing building.",
                "Keep away from the neighbor UAVs."
            ]
        },
        "keep on": {
            "sentences": [
                "The UAV keeps on searching until the extration fails."
            ]
        },
        "keep as": {
            "sentences": [
                "For Workload 2, we keep the same job structures as in Workload 1 but change the job execution times using a two-stage hyper-Gamma distribution derived from the model presented in [30]."
            ]
        },
        "keep in": {
            "sentences": [
                "4.2 evaluation in our experiments, to keep in line with previous works, we use partial match for nyt?",
                "This guarantees that all the tuple versions that are used by any running transactions are kept in Met-Cache.",
                "the entire block content is kept in memory"
            ]
        },
        "keep up with": {
            "sentences": [
                "Traditionally, analytical query engines rely on the exponential increase of CPU performance in order to keep up with the, also exponential, data growth.",
                "Query sets which make use of very high speed data streams must often use approximate data reduction strategies to provide complex statistics while keeping up with the offered data load."
            ]
        },
        "keep away": {
            "sentences": [
                "However, the multiple local updates keep the client away from the global optimum."
            ]
        },
        "keep up": {
            "sentences": [
                "Even if the adversary cannot keep up the chain difficulty of its private proposer chain with the heaviest public chain, at the current time on the voter chain, the adversary will refer to this very difficult block on the proposer tree to create avery difficult block on the voter tree."
            ]
        },
        "keep a lower": {
            "sentences": [
                "The potential ith leader keeps a lower bound on the start of a lease vote from replica r as vleader = TT.now().earliest, computed before esend (de\ufb01ned as when the lease i,r i,r ACMTransactionsonComputerSystems,Vol.31,No.3,Article8,Publicationdate:August2013."
            ]
        }
    },
    "keephit": {
        "keephit rat high": {
            "sentences": [
                "notonly di\ufb00erenttra\ufb03cpatterns,butalso di\ufb00erentcaching LikeUSR,SYShasarelatively boundeddatadomain,so properties that can bene\ufb01t from di\ufb00erent tuning, justifying it can easily besized to keephit rates high andstable."
            ]
        }
    },
    "kernels": {
        "kernels by at": {
            "sentences": [
                "Code generated by taco outperforms equivalent Tensor Toolbox kernels by at least about anorder of magnitude on all of our benchmarks."
            ]
        }
    },
    "knock": {
        "knock at": {
            "sentences": [
                "Did you hear the knocks at the door?"
            ]
        }
    },
    "know": {
        "know about": {
            "sentences": [
                "He knows a lot about that subject."
            ]
        },
        "know as": {
            "sentences": [
                "to capture the interdependency among the multiple triples, we reformulate the document-level re task as an entity-level classification problem [jiang et al., 2019], also known as table filling [miwa and sasaki, 2014; gupta et al., 2016], as shown in figure 2."
            ]
        }
    },
    "konw": {
        "know to": {
            "sentences": [
                "Opaque predicates and variables are constructs whose values are known to the obfus- cator, but are difficult for the deobfuscator to deduce."
            ]
        }
    },
    "label": {
        "label with": {
            "sentences": [
                "The first one is labeled with the periods when someone lived somewhere.",
                "Buckets are labeled with bucket id starting from 1."
            ]
        }
    },
    "lack": {
        "lack of": {
            "sentences": [
                "To conquer the challenge of the lack of manual annotation, this paper presents a first study on utilizing semi-supervised object detection (SSOD) for retinal lesion localization in OCT images",
                "However, there is a clear lack of a systematic framework with which different systems can be analyzed and compared against each other"
            ]
        }
    },
    "layer": {
        "layer on": {
            "sentences": [
                "weaver\u2019s key contribution is to reduce the load on a totally ordering timeline oracle by layering on a timestamping service that manages the bulk of the ordering, and leaves only a small number of overlapping transactions to be ordered by the oracle."
            ]
        }
    },
    "lead": {
        "lead to": {
            "sentences": [
                "the quadratic time complexity leads to a much slower inference speed than the dual-encoder models whose time complexity is linear.",
                "Recent works show that decoupling representation learning from classifier learning can lead to good features",
                "Faster block-time also leads to high number of stale blocks to be produced",
                "Causal modeling on the effect (click) and cause (interest and conformity) can lead to more robust models",
                "Mistakes in predicting row/column/cell boundaries (sometimes compounded by OCR errors) can lead to inaccurate extraction.",
                "\ufeffThe task has led to a growing interest in scalable end-toend retrieval systems for question answering.",
                "However, it is possible that two distinct data structures may lead to isomorphic graphs which cannot be used to distinguish instances of the two data structures.",
                "The improved memory bandwidth of GPUs has led some researchers to use GPUs as coprocessors for analytic query processing.",
                "3D data has multiple popular representations, leading to various approaches for learning.",
                "Meanwhile, simply removing Cutout also leads to a drop from 49.34 to 48.22, see the row of UbT w/o Cutout. The result suggests that Cutout has a positive effect on the performance, while the effect of Mixup is negative",
                "it leads to a large number of patches (instances) cropped from this slide.",
                "This leads to CPU-only memory-bound applications having an increasingly non-negligible impact on database query throughput.",
                "On one\nhand, training a stock representation on stochastic stock data\nin an end-to-end manner may lead to excessive modeling,\nwhich involves the model uncertainty.",
                "We also leverage our knowledge of the size of the (approximate) maximum clique in the previous graph to eliminate parts of the search space in our restricted search that cannot lead to larger cliques",
                "this discrepancy leads to error accumulation.",
                "These different constraints lead to different design choices.",
                "Both of the two works use centralized clustering algorithm such as K-means, in which the center machine has to identify the cluster identities of all the users, leading to high computational cost at the center.",
                "This choice is purely result-driven: computing the loss on all pixels leads to a slight decrease in accuracy (e",
                "This issue led to increased attention being paid by academia and industry to semantic annotation of table data.",
                "besides, we observed that the u-shaped segmentation module and balanced softmax module are most important to model performance and sensitive to f1, leading to a drop of 2.18% and 1.32% in dev f1 score respectively when removed from docunet.",
                "As the node with the maximum utility has the largest possibility of leading to the optimal node",
                "Brute force addition of all neighbors to X t will quickly lead to an explosion in its size after a few turns (hugely exacerbated if popular entities are added, e.g. Germany and Barcelona have \u2243 1.6M and \u2243 40k neighbor entities in Wikidata).",
                "Since EDA method modifies tokens at random, it leads to more diverse n-gramo",
                "This could lead to unbalanced load between the leader replica and other replicas",
                "Advances in program analysis and software engineering technol- ogy in recent years have led to significant improvements in tools for program analysis and software development.",
                "Ignoring the time dimension could lead to incorrect results, or prevent interesting analysis possibilities.",
                "To guarantee security, the mining rate remains low [11], which leads to low throughput and high latency.",
                "Thus, a larger output size does not necessarily lead to a better approximation of the query.",
                "Furthermore, as the increase in cores and processors per server leads to an increase in concurrency (and therefore lock contention),",
                "This avoids the generation of too many active locations, which can lead to a decrease in speed in subsequent convolution layers due to the large number of active points.",
                "which will lead to indistinguishable features with an ambiguous classification boundary.",
                "Such rule-based selection leads to unsatisfactory performance",
                "This scenario has clearly placed the threat for the software manufacturers and leads to the development of numerous software protection techniques.",
                "Different from [5], we increase the number of height channels from 4 to 16 to capture more detailed shape information\u2013 further increasing the number of height channels did not lead to performance improvement.",
                "it leads to a large number of patches extracted from the network.",
                "this stage is more expensive and less scalable but leads to a precise ordering of transactions.",
                "For example, an ISBN or credit card number are generated according to strict validation rules, lending themselves to straightforward type detection with just a few rules.",
                "Over long trajectories this leads to scale-drift",
                "csr is very compact, leading to a small storage footprint, reduced memory traffic, and high cache efficiency.",
                "This allows us to avoid ambiguous areas and leads to reliable keypoint detections and descriptions",
                "However, on a fip side, this also leads to increased communication costs",
                "If the batch size is small, even one false pseudo label will lead to the huge deviation between the pseudo-labeled centroid and true centroid.",
                "Returning to the example in Figure 1, inferring ownership on the basis of political affiliation when no other information is available may increase accuracy on average, but it can also lead to erroneous or biased decisions (indeed, for the claim in the example, the prediction would be wrong).",
                "One example is the model-agnostic meta-learning (MAML) which trains model via a small number of gradient updates and leads to fast learning on a new task",
                "text-to-image generation models, in particular, have the potential to extend image-editing capabilities and lead to the development of new tools for creative practitioners.",
                "While this is simpler to apply, compared to fine grain CPU and memory configuration, it may lead to wastage by over-allocating resources to workloads as well consume more power.",
                "Effectively, making the SCAR assumption leads to two important insights.",
                "however, it should be noted that fast always requires to allocate memory in the power of 2 to  use the branch free simd instructions, which can lead to signi cantly larger indexes.",
                "Otherwise, aggressively approximating may cause unacceptable quality loss and even lead to corrupt out\u0002put [17].",
                "At each internal node of the tree, a certain field inside the message is checked and different values will lead to different paths until a corresponding service is invoked."
            ]
        },
        "lead to roughly": {
            "sentences": [
                "this approximation removes the need for computing  hessian-vector  products  in  an  additional  backward  pass,  which  we  found led to roughly 33% speed-up in network computation."
            ]
        },
        "lead to substantially": {
            "sentences": [
                "the  results,  shown  in  figure  41,  indicate that  prior  methods  for  learning  distance  metrics  struggle  to  infer  the  goal  of  the  task, while our approach leads to substantially more successful behavior on average."
            ]
        },
        "lead to significantly": {
            "sentences": [
                "While [25] certainly addresses generic version of theselectivity estimation problem, the models in this paper aremuch more succinct leading to significantly faster estimations.",
                "First, looking up embedding vectors on many tables causesrandom DRAM accesses, leading to significantly under-utilizedmemory bandwidth and low lookup performance."
            ]
        },
        "lead to arbitrarily": {
            "sentences": [
                "WhiletheDataMorphingmodelworks\ufb01ne for simple, non-overlapping queries and small, aligned containers, Q3 Readalladdressesbelongingtothiscustomer(ADRC) select * from ADRC where KUNNR = (...); it can lead to arbitrarily bad results in more realistic cases."
            ]
        }
    },
    "learn": {
        "learn from": {
            "sentences": [
                "They want to cooperate in training a recommendation model, which learns from both users\u2019 behavior on online contents (items such as videos) and items in local stores",
                "We devise a simple pretraining objective (corrupt cell detection) that learns exclusively from tabular data and reaches the state-of-the-art on a suite of tablebased prediction tasks.",
                "Recently, the SuperPoint approach by DeTone et al. [9] tackles keypoint detection as a supervised task learned from artificially generated training images containing basic structures like corners and edges.",
                "To what extent can conclusions and good practices of SSOD learned from the natural-image domain be generalized to the OCT-image domain is mostly untouched.",
                "Our method is trained end-to-end from image pairs \u2013 we learn priors for pose estimation from a large annotated dataset,enabling SuperGlue to reason about the 3D scene and the assignment.",
                "They learn from existing data patterns, efficiently predict future workload trend and effectively schedule them by balanc- ing the conflict rates and concurrency.",
                "the underlying neural layers of deep neural networks transform concrete features learned from data into abstract features in upper layers and keep stable as the local change of the input data",
                "The meta information learned from high-frequency relations is helpful for few-shot relations",
                "We discuss lessons learned from the results in Section 5 and related work in Section 6, and we conclude in Section 7",
                "Online learning aims to learn from the data arriving in a sequential order.",
                "Individual point features are learned from normalized point-to-node coordinates through a series of fully connected layers.",
                "Rajpal et al. (2017) from Microsoft use neural networks to learn from past fuzzing explorations and predict which byte to mutate in input files."
            ]
        },
        "learn to": {
            "sentences": [
                "This motivates us to propose a new problem where a model should be able to identify instances of unknown objects as unknown and subsequently learns to recognise them when training data progressively arrives, in a unified way",
                "The model learns to recover the visual tokens of the original image, instead of the raw pixels of masked patches",
                "SSA learns to aggregate few tokens and preserves more local details.",
                "In this setting, a semantic parsing model learns to map a natural language question to a logical form (LF), i.e. a tree of operators over the KG.",
                "intelligent machines must be able to reason with existing resources, and learn to infer an unknown answer",
                "In this way, the model is learning to address a more challenging task as opposed to predicting word components"
            ]
        },
        "learn about": {
            "sentences": [
                "I need to learn more about that topic.",
                "The UAV swarn needs to learn more about the neightbor information for localization."
            ]
        },
        "learn through": {
            "sentences": [
                "The weighting function is learned through a simple MLP."
            ]
        },
        "be learn through": {
            "sentences": [
                "Such dataset conventions could be learned through fewshot learning, where a small number of in-domain training examples are available."
            ]
        },
        "learn over": {
            "sentences": [
                "The intuition is that learning over a representative subset would be competi- tive over the whole pool.",
                "[2019] proposed a model that combines representations learned over various text spans throughout the document and across the sub-relation hierarchy."
            ]
        },
        "learn for": {
            "sentences": [
                "Owning to joint representation learning for sequences and knowledge",
                "reinforcement learning for relation classification from noisy data."
            ]
        },
        "learn by": {
            "sentences": [
                "It learns by simply looking at images and reading paired questions and answers.",
                "Probabilistic graph models learn feature representation by modeling the posterior distribution of potential variables in sample data"
            ]
        },
        "learn irs only": {
            "sentences": [
                "Currentdeep learning IRs only support codegen when all dimensions of a tensor's shape are known at compile-time, i.e.,static shapes are mandatory for type inference and checking."
            ]
        },
        "learn with": {
            "sentences": [
                "vlmo is jointly learned with three pre-training tasks, namely image-text contrastive learning, imagetext matching, and masked language modeling.",
                "Since embeddings of sequences and knowledge are learned with different algorithms"
            ]
        },
        "learn more": {
            "sentences": [
                "by getting rid of the limited size of image-text pairs and their simple and short captions, stagewise pre-training on large amounts of image-only and text-only data helps vlmo to learn more generalizable representations."
            ]
        }
    },
    "leed": {
        "leed to": {
            "sentences": [
                "These features have led to numerous PM-optimized indexes"
            ]
        }
    },
    "lelize": {
        "lelize largecomputations easily": {
            "sentences": [
                "Programmers speci\ufb01ed mapandreduce operationsallowsus to paral- \ufb01ndthesystemeasytouse:hundredsofMapReducepro- lelize largecomputations easily and to use re-execution gramshavebeenimplementedandupwardsofonethou- astheprimarymechanismforfaulttolerance."
            ]
        }
    },
    "let": {
        "let us further": {
            "sentences": [
                "now, let us further assume that our model generalizes and is able to learn the patterns, which also hold for the future data."
            ]
        }
    },
    "leverage": {
        "leverage to": {
            "sentences": [
                "We leverage the state information from neighbors to pool the sequential information.",
                "Then, well-established 2D or 3D convolutional networks are leveraged to achieve shape classification.",
                "Godfrey & Gashler (2017) presents an initial attempt to use neural\nnetworks to model periodic and non-periodic components in time series data, leveraging periodic\nactivation functions to model the periodic components.",
                "we leverage the sequence pooling method(SeqPool) to pool the sequential information produced by Transformer encoder.",
                "A self-calibration strategy is leveraged to further explore the fine-grained treatment of loss functions over unconfident examples, in a meta-learning fashion."
            ]
        },
        "leverage sth": {
            "sentences": [
                "Table-BERT leverages the state-of-the-art pre-trained language model to encode the linearized tables and statements into continuous vectors for verification."
            ]
        },
        "can leverage": {
            "sentences": [
                "Single-node systems can leverage distributed systems techniques to scale well on multi-core CPUs."
            ]
        },
        "leverage from": {
            "sentences": [
                "Overall, our REINA can effectively leverage the most relevant data from additional datasets while being trained only on the target task."
            ]
        },
        "leverage the highly effective technology": {
            "sentences": [
                "Theoverall hope is to leverage the highly effective technology that has been developed for indexing multi-dimensional data."
            ]
        },
        "leverage on": {
            "sentences": [
                "PT leverages on physical clocks at nodes that are synchronized using the Network Time Protocol (NTP)"
            ]
        },
        "leverage the heavily": {
            "sentences": [
                "The baseline systems all make use of third-party kernel libraries to achieve high-performance by leveraging the heavily hand-optimized operators, which is handicapped when anoperator is not supported on a specific target."
            ]
        },
        "leverage the fine": {
            "sentences": [
                "Even though TVM can make individualoperator run faster through tuning, it still lacks the capability to leverage the fine-grained parallelism as RAMMER."
            ]
        },
        "leverage for": {
            "sentences": [
                "on the other hand, generative methods can be leveraged for malicious purposes, including harassment and misinformation spread [20], and raise many concerns regarding social and cultural exclusion and bias [67, 62, 68]."
            ]
        },
        "leverage the inductive bias": {
            "sentences": [
                "To achieve this, we leverage the inductive bias that densely connected networks possess in representing smooth functions"
            ]
        },
        "leverage sth in": {
            "sentences": [
                "This paper explores the changes that are required in a DBMS to leverage the unique properties of NVM in systems that still include volatile DRAM."
            ]
        }
    },
    "license": {
        "license under": {
            "sentences": [
                "1the source code of this paper can be obtained from https://github.com/131250208/tplinker-joint-extraction this work is licensed under a creative commons attribution 4.0 international license."
            ]
        }
    },
    "lie": {
        "lie in": {
            "sentences": [
                "Point points which lie in sparsely sampled regions are based on the Cloud Library (PCL).",
                "The generality of the proposed framework lies in two aspects.",
                "Critical clues to distinguish them lie in both eyes and areas between eyes.",
                "The main advantages of RBCD over DGS lies in the fact that it has provable convergence guarantees",
                "An important limitation of direct image alignment lies in the inherent non-convexity of the problem",
                "A major factor to their success lies in their innovative scheduling algorithms.",
                "Intuitively, the key of weak-strong data augmentation scheme lies in the difference between weak augmentations and strong augmentations",
                "\ufeffIn general, this paradigm is quite straightforward and easy to implement, however, the main disadvantage lies in that important local information is lost when compressing such a complicated sentence into a single vector.",
                "then they will lie in distinct parts of the feature space, and a classifier will be able to easily separate the domains."
            ]
        },
        "lie on": {
            "sentences": [
                "We show how samples lie on the embedding space using CCSA ."
            ]
        },
        "lie anywhere": {
            "sentences": [
                "for tasks lying anywhere on this spectrum, we demonstrate that a single flamingo model can achieve a new state of the art for few-shot learning, simply by prompting the model with task-specific examples."
            ]
        }
    },
    "light": {
        "light on": {
            "sentences": [
                "By contrast, to shed the light on the relationship between Paxos and Raft"
            ]
        }
    },
    "limit": {
        "limit to": {
            "sentences": [
                "However, these methods are limited to detecting a fixed number of planes with certain learned order.",
                "To address this heterogeneity, we limited the number of columns to at most 15K per class and excluded the 10% types containing less than 1K columns.",
                "Considering the dataset statistics in Table 1, we limit the caption to 20 tokens, section title and page title to 10 tokens each, and table headers to 20 tokens.",
                "They are typically limited to a single local device, which may incur bias or overfitting since data in a single device is extremely limited.",
                "However, prior projects to mix concurrency control either are limited to specific pairs of protocols",
                "They design a model that supports temporal analysis with property graph databases, using a single-graph model limited to structural changes.",
                "They are limited to the transductive setting, where the full set of entities must be known during training"
            ]
        },
        "be limit by": {
            "sentences": [
                "scalability is limited by DRAM's high cost and low capacity"
            ]
        },
        "limit of": {
            "sentences": [
                "the relatively small volumes of pedestrians and cyclists lead to less voxels of them, which limits the power of the CNNs."
            ]
        },
        "limit on": {
            "sentences": [
                "which limits the model scalability on receiving long sequence inputs."
            ]
        },
        "be limit to sth.": {
            "sentences": [
                "Current algorithms and systems for data processing on GPUs are all limited to some degree by the capacity of GPU memory (L2)."
            ]
        },
        "limit sth": {
            "sentences": [
                "To limit the amount of work performed while determining the visibility of tuples, the DBMS\u2019s garbage collector thread periodically scans the database to undo the dirty modifications associated with the currently present gaps."
            ]
        },
        "limit in": {
            "sentences": [
                "Adversaries are severely limited in what they can learn about the original training data based on analyzing the parameters"
            ]
        },
        "limit how far": {
            "sentences": [
                "The database infers this upper bound because it  commits in the writer and use this information to support snapshot  allocates LSNs, and limits how far allocation can occur above  isolation for local transactions that are of course read-only."
            ]
        }
    },
    "limitate": {
        "limitate sb. to": {
            "sentences": [
                "A good automatic storyteller should imitate humans to completely understand the visual contents and their relationships."
            ]
        }
    },
    "linearize": {
        "linearize by": {
            "sentences": [
                "based on this formulation, attention can be linearized by first computing the highlighted terms."
            ]
        },
        "linearize into": {
            "sentences": [
                "While methods such as TaBERT and TaPaS contextualize cell embeddings by linearizing the table into a single long sequence, we take a different and more computationally manageable approach.",
                "We then linearize the sampled candidates into {column name | associated cell value list} and concatenate them into a sequence, separated by <sep> token."
            ]
        }
    },
    "link": {
        "be link together": {
            "sentences": [
                "Extended stash blocks are linked together and reachable via a pointer in the data node."
            ]
        },
        "link to": {
            "sentences": [
                "We hypothesize that this reasoning-like behavior is linked to the learning of useful representations",
                "That is, for each node, the algorithm follows the link to the previous node until there is no previous node, like in the implementation of the Dijkstra algorithms.",
                "Further, we note that there is a lot of related research linked to join processing and radix partitioning.",
                "it first recognizes the entity mentions and links them to Wikipedia entities",
                "The columns containing numerical values are not linked to knowledge, which are called 'literal-columns.'"
            ]
        },
        "link sth. together": {
            "sentences": [
                "And then link these together using the verb where 1 returns the one-hot representation."
            ]
        },
        "link with": {
            "sentences": [
                "This ensures 'sender anonymity', which is also linked with 'untraceability'",
                "In the global indexing approach, indexes are linked with the creation and deletion time of items which can be used at run time to efficiently reconstruct order sets."
            ]
        }
    },
    "list": {
        "list in": {
            "sentences": [
                "the corresponding statistics for the generated datasets and database sizes of the selected systems are listed in the table 1."
            ]
        },
        "list data": {
            "sentences": [
                "We use the listed data only in pre-training and then fine-tune the model"
            ]
        },
        "be list below": {
            "sentences": [
                "The main contributions of our paper are listed below."
            ]
        },
        "list as follow": {
            "sentences": [
                "We list them as follows."
            ]
        }
    },
    "listen": {
        "listen to": {
            "sentences": [
                "She liked to listen to children talk"
            ]
        }
    },
    "live": {
        "live in": {
            "sentences": [
                "by decoding, 5 triplets can be extracted: (new york, mayor, de blasio), (de blasio, born in, new york), (de blasio, born in, new york city), (de blasio, live in, new york), (de blasio, live in, new york city).",
                "We live in the golden age of distributed computing"
            ]
        }
    },
    "load": {
        "load to": {
            "sentences": [
                "Hymem uses cache line-grained loading to extract only the hot data objects from an otherwise cold 16 KB page.",
                "we propagate task assignment and load balancing to the CPU and perform them through a CPU operator, HetExchange\u2019s router, which is a parallelism trait converter."
            ]
        },
        "load into": {
            "sentences": [
                "the generated dataset contains two parts: (i) a static part which is loaded into the system as an intermediate state of the social network, (ii) updates which are played out on this intermediate state."
            ]
        },
        "load from": {
            "sentences": [
                "BlockLoad internally uses vector instructions when loading a full tile and for the tail of the input array that may not form a perfect tile.",
                "we load the data from memory"
            ]
        },
        "load sth": {
            "sentences": [
                "During recovery, the DBMS first loads the latest checkpoint that contains an empty ATT."
            ]
        },
        "loaded, andalso": {
            "sentences": [
                "The tablet server handles read and write pletelymappedintomemory,whichallowsustoperform requests to the tablets that it has loaded, andalso splits lookupsandscanswithouttouchingdisk."
            ]
        }
    },
    "locate": {
        "locate on": {
            "sentences": [
                "For a given point, the mean features of all the neighboring points that are located on the same grid are computed from the previous layer."
            ]
        },
        "locate in": {
            "sentences": [
                "But also locates in n isolated views on each distributed node",
                "We deployed three instances of the modified client on IP addresses 47.251.4.119,\n47.244.165.26, and 8.208.203.101, located in the US, Hong Kong, and the UK, respectively.",
                "( , country, ) triple1 ( , located in, ) triple2 ( , located in, ) triple3 ( , country, ) triple4 ( , country, ) triple5 intra-sentence inter-sentence triple2 triple1 triple3 triple4 triple5 interaction between triples figure 1: example document with entity pairs and relations from docred."
            ]
        },
        "locate the most": {
            "sentences": [
                "Unlike other vision-to-text tasks 1340 (discussed above), a visual grounding system takes a natural language query in the form of 1341 a word or sentence as input and aims to locate the most relevant object or region of interest 1342 in the input image."
            ]
        },
        "be locate in": {
            "sentences": [
                "When replicas are located in different data centers"
            ]
        },
        "locate between": {
            "sentences": [
                "The selection field is intentionally located between question and response options to integrate it into the annotator\u2019s workflow."
            ]
        }
    },
    "lock": {
        "lock: e.g.": {
            "sentences": [
                "Becausetheroottabletcontains lock: e.g., due to a network partition that caused the the names of all METADATA tablets, the master knows aboutallofthemafterithasscannedtheroottablet."
            ]
        }
    },
    "log": {
        "log record asynchronously": {
            "sentences": [
                "Note that the replicas  threads do not pause for commits, they simply pull other pending  consume log records asynchronously from the perspective of the  requests and continue processing."
            ]
        },
        "log into exactly": {
            "sentences": [
                "GFS A merging compaction that rewrites all SSTables tablet log into exactly oneSSTable is called a major compaction."
            ]
        }
    },
    "look": {
        "look up": {
            "sentences": [
                "In these models, each word is usually converted into a discrete vector representation by looking up the word embedding table, and then the context-sensitive word representation is learned by certain structures",
                "For each value in a column, if the value is a single word, we look up the word embedding from the GloVe dictionary."
            ]
        },
        "look at": {
            "sentences": [
                "Look at the mesh map the algorithm produces.",
                "To improve their discriminativeness, some works explicitly look at a wider context using regional features [32] or log-polar patches [20].",
                "Look at the mess you have made!",
                "By looking at the times t1 and t2 for the last two difficulty-retargeting intervals, we can do that.",
                "Next, we analyze how well TABBIE understands trends in numerical columns by looking at specific examples of our corrupt cell detection task.",
                "When looking at low data regimes in Figure 3 we find that pre-training on SQA or our artificial data consistently leads to better results than just training with the MASK-LM objective.",
                "It follows that an adversary looking at the trained model cannot infer whether any specific user\u2019s data was used in the training, irrespective of what auxiliary information they may have"
            ]
        },
        "look over": {
            "sentences": [
                "Some officials will look over our factory.",
                "The central agent will look over the environment to search for feature points."
            ]
        },
        "look into": {
            "sentences": [
                "One recent work from Oriol Vinyals et al [22] looks into this problem."
            ]
        },
        "look more": {
            "sentences": [
                "to probe image quality, the rater is asked to select between the model generation and reference image using the question: \u201cwhich image is more photorealistic (looks more real)?\u201d."
            ]
        },
        "look back": {
            "sentences": [
                "If a tuple refers to a previous occurrence, the algorithm looks back in the buffer to find it"
            ]
        },
        "look for": {
            "sentences": [
                "To evaluate this Cypher query, the engine will look for all the airports in Lon- don and Bariloche, and all the combinations from airports in London to airports in Bariloche.",
                "The PosOnly setting of the relational rule learning system Aleph makes the separability assumption and looks for the simplest theory that covers all positive examples and introduces as few new facts as possible.",
                "A high sparsity ratio means there are many empty entries in the dataset, and implies a better compression ratio with schemes that looks for repetitions in the dataset",
                "Specifically, the role of the keypoint detector is to look for scale-space locations with covariance_x000D_ with respect to camera viewpoint changes and invariance with respect to photometric transformations."
            ]
        },
        "look behind": {
            "sentences": [
                "The look behind buffer contains the LBA modified by the previous N log entries,"
            ]
        }
    },
    "loop": {
        "loop through": {
            "sentences": [
                "We loop through all possible operators."
            ]
        }
    },
    "lose": {
        "lose track of": {
            "sentences": [
                "As such, it is easy to lose track of the design decisions and potentially important tweaks in these fuzzers over time."
            ]
        },
        "get lose in": {
            "sentences": [
                "all committed modifications will not get lost in corner cases"
            ]
        }
    },
    "low": {
        "be low in": {
            "sentences": [
                "stash ratio is low in most cases and extended stash blocks are rarely used as stash ratio is reasonably set based on data distribution."
            ]
        },
        "low by": {
            "sentences": [
                "to extract relations among these inter-sentence entity pairs, most current studies constructed document-level graph module based on heuristics, structured attention or dependency structures [peng et al., 2017; christopoulou et al., 2019; nan et al., 2020; zeng et al., 2020; wang et al., 2020a], followed by reasoning with graph neural models."
            ]
        }
    },
    "machinedaysused": {
        "\"machinedaysused 79": {
            "sentences": [
                "186days properly\""
            ]
        }
    },
    "maintain": {
        "be maintain in": {
            "sentences": [
                "This is because learned indexes require data (key-value pairs) be maintained in sorted order,"
            ]
        },
        "maintain sth.": {
            "sentences": [
                "The need to maintain inter-sentence coherence and topic consistency throughout the whole story."
            ]
        },
        "maintaine with": {
            "sentences": [
                "[7] shows that an offline sample canbe maintained with virtually no overhead."
            ]
        },
        "maintain a local write set": {
            "sentences": [
                "A transaction reads from the database and maintains a local write set in the execution phase"
            ]
        },
        "maintain the globally": {
            "sentences": [
                "The server usually maintains the globally shared parameters byaggregating updates from the workers and updating the globalparameters."
            ]
        },
        "maintain sth": {
            "sentences": [
                "A disk-oriented DBMS maintains two meta-data tables at runtime that it uses for recovery.",
                "A transaction reads from the database and maintains a local write set in the execution phase"
            ]
        },
        "maintain for": {
            "sentences": [
                "Regardless of its implementation, there is common meta-data that a MVCC DBMS maintains for transactions and database tuples."
            ]
        },
        "maintain by": {
            "sentences": [
                "A Blockchain is maintained by a network of nodes and every node executes and records the same transactions",
                "Blockchain network is highly scalable in nature as it is maintained by a network of peers"
            ]
        }
    },
    "make": {
        "make them even": {
            "sentences": [
                "\"Thepossibilitytomaketheautomaticscalingfacilitiesmoreawareofbusiness rules makes them even more useful"
            ]
        },
        "make it inherently": {
            "sentences": [
                "\"This low level makes it inherently dif\ufb01cult for Amazontoofferautomaticscalabilityandfailover"
            ]
        },
        "make sth adj.": {
            "sentences": [
                "To make the groups exclusive, we add a fifth case when more than one operation is needed."
            ]
        },
        "make up": {
            "sentences": [
                "The successfully matched line features make up the initial clusters",
                "But LIPP can make up for its little slower bulkload time.",
                "Besides, as our likelihood function is made up of several elementary functions, so it is continuous on the parameter space."
            ]
        },
        "make of": {
            "sentences": [
                "Cloud systems are typically diverse and made of a heterogeneous set of compute and storage resources."
            ]
        },
        "make use of": {
            "sentences": [
                "In this way, ResTune could accordingly make use of existing data and accelerate the tuning process",
                "Spitfire makes use of the direct data flow path from DRAM to SSD.",
                "Besides, to make full use of negative features, we pro- pose the random interpolation among negatives, which in- tuitively provides diversified negatives for each training step and makes the model more discriminative",
                "To make use of (4) in a streaming context, first as a base case, we assume an implementation of VAROPTk (I) when I has k+1 items, denoting this procedure VAROPTk,k+1."
            ]
        },
        "make sense to": {
            "sentences": [
                "It makes most sense to optimize average variance measures like those above giving each item equal opportunity to be included in the estimated subset."
            ]
        },
        "make adjustment to": {
            "sentences": [
                "To correct this problem, we made a minor adjustment to the dynamic subset-sum sampling."
            ]
        },
        "make a balance between": {
            "sentences": [
                "Most works [9,22] emphasize that it is vital to evaluate the quality of the process of data handling so as to make a better balance between computation quality and system performance."
            ]
        },
        "make for": {
            "sentences": [
                "similar arguments can be made for oltp-like graph workloads; however, there are no comprehensive studies of existing systems for real-world, dynamic graph workloads such as online social networks."
            ]
        },
        "make in": {
            "sentences": [
                "weaver queries wishing to modify the graph must collate the changes they wish to make in a node program and submit them as a transaction."
            ]
        },
        "make to": {
            "sentences": [
                "However the need for an additional camera that is time synchronized and calibrated with the LiDAR restricts their use and makes the solution more sensitive to sensor failure modes. In this work we focus on LiDAR-only detection."
            ]
        },
        "make sure": {
            "sentences": [
                "They are included nonetheless, to make sure the datasets are not biased\ntowards the proposed model NeuralProphet",
                "then apply a sigmoid transform on diagonal elements and plus vector one to make sure values in diagonal\nare greater than 1"
            ]
        },
        "make it surprisingly": {
            "sentences": [
                "in terms of semantic guarantees, indexes are already to a large extent learned models making it surprisingly straightforward to replace them with other types of ml models."
            ]
        },
        "make a comprehensive": {
            "sentences": [
                "This allows us to make a comprehensive comparison between the two join implementations in a much broader scope of workloads and factors than the analysis done by prior work."
            ]
        },
        "make them accessible": {
            "sentences": [
                "Making them easily accessible can therefore have direct commercial and societal impact."
            ]
        },
        "make descriptions more natural": {
            "sentences": [
                "Personalization is what makes descriptions more natural and diverse."
            ]
        },
        "make out way": {
            "sentences": [
                "we made out way down to the lake after leaving our belongings in the lodge."
            ]
        },
        "be make public": {
            "sentences": [
                "However, the dataset including annotations has not been made public yet."
            ]
        },
        "make accelerators more": {
            "sentences": [
                "Intuitively,improving power effciency makes accelerators more appealing for recommendation inference."
            ]
        },
        "make it more": {
            "sentences": [
                "By iteratively mutating high-used to re-train the cost model to make it more accurate."
            ]
        },
        "make gpu less": {
            "sentences": [
                "Finally, the model inference of Tree-LSTM on Nvidia GPUwas omitted as it's hard to saturate GPU compute capability due to excessive control flows and small kernel sizes,making GPU less favorable deployment targets."
            ]
        },
        "make statements": {
            "sentences": [
                "Meta make similar statements."
            ]
        },
        "make towards": {
            "sentences": [
                "while initial progress has been made towards a similar capability in computer vision, the most used paradigm still consists of first pretraining on a large amount of supervised multimodal data collected from the web, before fine-tuning the model on the task of interest (lu et al., 2019; wang et al., 2021; zellers et al., 2022)."
            ]
        },
        "make available": {
            "sentences": [
                "The code and trained models will be made available"
            ]
        },
        "be make from": {
            "sentences": [
                "a checkpoint is made from time to time"
            ]
        },
        "make \u201cgenerally": {
            "sentences": [
                "In addition to the IO  In this section, we will share our experiences in running Aurora as  Subsystem, InnoDB also includes the transaction subsystem, the  a production service that was made \u201cGenerally Available\u201d in July  lock  manager,  a  B+-Tree  implementation  and  the  associated  2015."
            ]
        },
        "make recovery fast": {
            "sentences": [
                "on-demand page-by-page redo that can make recovery fast."
            ]
        },
        "make the especially": {
            "sentences": [
                "At each step, we can only make the Especially for the updates of the data structures discussed decision based on the current grouping."
            ]
        },
        "make this decisionsothatitwouldbepossibletocolocatemultipledirectoriesthatarefrequently": {
            "sentences": [
                "We made this decisionsothatitwouldbepossibletocolocatemultipledirectoriesthatarefrequently accessedtogether."
            ]
        },
        "make leadership long": {
            "sentences": [
                "Spanner\u2019s Paxos implementation uses timed leases to make leadership long-lived (10 seconds by default)."
            ]
        },
        "make it even": {
            "sentences": [
                "For  instance,  multiple storage nodes in read and write operations makes it even  during low loads the imbalance ratio is as high as 20% and during  more challenging, since the performance of these operations is  high loads it is close to 10%."
            ]
        },
        "make progress": {
            "sentences": [
                "With the landmark of AFL, feedback-driven fuzzing, especially coverage-guided fuzzing, has made great progress."
            ]
        }
    },
    "manage": {
        "manage by": {
            "sentences": [
                "these are stored in a single large memory-mapped file managed by livegraph memory allocator.",
                "People\u2019s privacyis particularly at risk when such sensitive data are managed by centralized companies"
            ]
        },
        "be manage in": {
            "sentences": [
                "First, data consistency must be managed in software instead of in hardware."
            ]
        },
        "manage data globally": {
            "sentences": [
                "Given that Spanner manages data globally, there will be only a handful of running universes."
            ]
        }
    },
    "manipulate": {
        "manipulate exist": {
            "sentences": [
                "this paper presents a unified multimodal pretrained model called nuwa that can generate new or manipulate existing visual data (i.e., images and videos) for various visual synthesis tasks."
            ]
        }
    },
    "map": {
        "map to": {
            "sentences": [
                "Mapping all d dimensions to according utility values consumes O(d^2) and sampling requires O(d).",
                "The three embeddings are first concatenated, and then mapped to a D-dimensional embedding through a fully connected layer to form the fusion embedding",
                "During an\u0002swering, each word in q t potentially maps to one or more items in K (q t i 7\u2192 E \u222aP \u222a C \u222aL).",
                "Finally, this feature map is mapped to the desired learning targets",
                "Proposals that forecast workload arrival rate have an additional challenge to map this to a utilisation forecast.",
                "Once a valid nonce is found, the output of the hash is deterministically mapped to either a voter block in one of the m trees or a proposer block.",
                "the value mapped to key k can be accessed by array[k]",
                "each observation sequence is\nmapped to a hidden space to learn its marginal distribution.",
                "How to map the image to sequence is important for the model's performance.",
                "existing work has mapped Paxos, Viewstamp Replication and ZAB to a high-level abstract consensus algorithm"
            ]
        },
        "map into": {
            "sentences": [
                "Specifically, for partition with range size r, a title with release year y was mapped into a subset containing all titles whose release year is y mod r.",
                "They first map queries and tables into a set of word embeddings or graph embeddings.",
                "The operators in phase1 and the operators in phase2 need to be mapped into different compute units",
                "g and h are two-layer multi-layer perception (MLP) with ReLU activation which map the embedding into another space and increase the representation ability of the embedding",
                "because any contract states must be mapped into key-value tuples"
            ]
        },
        "map sth. to": {
            "sentences": [
                "Latent factor models typically map both users and items to a latent factor space with a low latent dimension."
            ]
        },
        "map xxx to": {
            "sentences": [
                "Semantic parsing is the task of mapping natural language utterances to formal meaning representations, and has been studied in tasks including instruction following, evaluating sentence meaning, and building interfaces to knowledge bases."
            ]
        },
        "map ... to": {
            "sentences": [
                "\ufeffIn embedding layer, a model can map tokens to vectors with the pre-trained word representation such as GloVe."
            ]
        },
        "map sth. into sth.": {
            "sentences": [
                "MMIO maps GPU memory into the CPU\u2019s address space."
            ]
        },
        "map sth": {
            "sentences": [
                "Reduce workload size by mapping queries into templates"
            ]
        }
    },
    "marginalize": {
        "marginalize over": {
            "sentences": [
                "At test time, we marginalize over s to obtain a final truth value"
            ]
        }
    },
    "mark": {
        "mark with": {
            "sentences": [
                "The methods marked with q have a public implementation available. The check mark(\udbff\udc01) or commenting text is used when a technique suits the corresponding attribute.",
                "The general autoscalers using the second configuration are marked with a star symbol."
            ]
        },
        "mark in": {
            "sentences": [
                "the overlapping entities are marked in blue."
            ]
        },
        "mark sth": {
            "sentences": [
                "Each of these pages has an entry in the DPT that marks the log record\u2019s LSN of the oldest transaction that modified it."
            ]
        }
    },
    "mask": {
        "mask out": {
            "sentences": [
                ", 75%) is masked out",
                "The technique randomly masks out entities in a table with the objective of recovering the masked items based on other entities and the table context.",
                "(2019a) to use WWM, a strategy to mask out all characters within a selected word, mitigating the easy-predicting shortcoming of the CM strategy",
                "We only allow copying of certain token types, and mask out invalid copying actions, including independent wordpieces from u and TAB and column-type tokens."
            ]
        },
        "be significantly mask by": {
            "sentences": [
                "it can be significantly masked by efficiently leveraging parallelism and concurrency in software"
            ]
        }
    },
    "master": {
        "be master on": {
            "sentences": [
                "At any point in time, each record is mastered on one node, with other nodes serving as secondaries"
            ]
        }
    },
    "match": {
        "match to": {
            "sentences": [
                "The lidar mapping module matches features in current scan to a surrounding point cloud map",
                "To match the workloads used in the original paper, we have used the following query to join build and probe table and count the resulting tuples."
            ]
        },
        "match with": {
            "sentences": [
                "We approximate this post-processing cost by the number of tuples which could not be matched with all their partner tuples because of resource shortage.",
                "Local feature matching is generally performed by i) detecting interest points, ii) computing visual descriptors,iii) matching these with a Nearest Neighbor (NN) search,iv) filtering incorrect matches, and finally v) estimating a geometric transformation.",
                "the compositional vector is then matched with the relation representation to score the triplet",
                "Finding matches consistent with a geometric model is typically performed in a separate post-processing stage [3, 4, 7, 10, 17, 22, 26, 36, 37, 43]._x000D_"
            ]
        },
        "match on": {
            "sentences": [
                "If PICARD matches on tid.cid or alias.cid, then guards require that the table tid or the alias alias, respectively, is eventually brought into scope by adding it to the from clause."
            ]
        },
        "be match with": {
            "sentences": [
                "we can infer the queries that were not matched with any templates in AutoIndex"
            ]
        }
    },
    "materialize": {
        "materialize the command log entries": {
            "sentences": [
                "Similarly, our proposed distributed command logging materializes the command log entries and footprint log before a transaction commits"
            ]
        },
        "materialize sth": {
            "sentences": [
                "Similarly, our proposed distributed command logging materializes the command log entries and footprint log before a transaction commits"
            ]
        }
    },
    "matter": {
        "matter greatly": {
            "sentences": [
                "similarly, we demonstrate here that the way we train the flamingo models matters greatly for their final performance."
            ]
        }
    },
    "maximize": {
        "maximize by fully": {
            "sentences": [
                "Thus, throughput for computeintensive models is maximized by fully utilizing the dataparallel SIMD units and parallelizing queries into multiplerequests across the chip-multiprocessor cores."
            ]
        }
    },
    "maxmize": {
        "maxmize by": {
            "sentences": [
                "First, we maximize domain confusion by making the marginal distributions of the two domains as similar as possible."
            ]
        }
    },
    "mean": {
        "mean to empirically": {
            "sentences": [
                "these task-conditioned models act as an oracle and are meant to empirically determine the depth needed to represent these poly- nomials, independent of the meta-learning process."
            ]
        },
        "mean that generally": {
            "sentences": [
                "As meaning that generally, data is added to the cache after datasizegrowsintheapplication,moreRAMcanbeadded a client hasrequested it and failed."
            ]
        }
    },
    "measure": {
        "measure between": {
            "sentences": [
                "to properly measure the correlation between a training sample and the target domain."
            ]
        },
        "measure sth": {
            "sentences": [
                "In this experiment, we measure the write throughput of a single thread performing synchronous writes to a large file."
            ]
        }
    },
    "mention": {
        "mention in": {
            "sentences": [
                "As mentioned in the introduction, a clean design that differentiates our VAROPTk scheme from preceding schemes is that we can just sample from samples without relying on auxiliary data."
            ]
        },
        "be mention in": {
            "sentences": [
                "as mentioned in Section 2.1."
            ]
        }
    },
    "merge": {
        "merge into": {
            "sentences": [
                "PL-SLAM merges lines into a point-based system.",
                "Therefore, these partitions (e.g. P4 to P6) can be merged into a cluster along with their respective cores"
            ]
        },
        "merge with": {
            "sentences": [
                "We merge it with two other data sources: CSQA, and 3.6M examples generated by random sampling.",
                "The leader candidate receives these entries and merges with its own log entries",
                "For example, if one partition has only a very few transactions, ACC can merge it with another partition as it would be a waste of resource to allocate a core to it",
                "In a distributed setting, distributed summaries are merged with the consideration of communication cost and merge complexity."
            ]
        }
    },
    "migrate": {
        "migrate into": {
            "sentences": [
                "To minimize this contention, we freeze the original indices, migrate data in frozen indices into newly built indices, and operate with tiered indices."
            ]
        },
        "migrate from": {
            "sentences": [
                "Algorithm 1 gives the details to migrate the historical data from the rollback segment to historical tables."
            ]
        },
        "be migrate between": {
            "sentences": [
                "Chunks can be migrated between ChunkServers when hot spots exist."
            ]
        }
    },
    "minimize": {
        "minimize communication overhead": {
            "sentences": [
                "\"Let applications provide a computation graph"
            ]
        },
        "minimize network communication": {
            "sentences": [
                "In essence, these techniques aim to minimize network communication between replicas at the cost of incurring more processing redundancy"
            ]
        },
        "minimize sth": {
            "sentences": [
                "As transactions tend to generate multiple log records that are each small in size, most DBMSs use group commit to minimize the I/O overhead."
            ]
        }
    },
    "miswritten": {
        "miswritten as": {
            "sentences": [
                "were miswritten as 246 in (wei et al., 2020) and (yu et al., 2020), which is the number of the original webnlg dataset instead of the subsets they used."
            ]
        }
    },
    "mix": {
        "mix with": {
            "sentences": [
                "These systems usually use a language generation model that mixes styles with other linguistic patterns.",
                "Each transaction in YCSB consists of 20 operations mixed with read-only and read-modify-write operations.",
                "These hard negative samples are delivered to user clients and mixed with the observed negative samples from local data"
            ]
        },
        "be mix in": {
            "sentences": [
                "data instances are mixed in both the input and virtual label spaces"
            ]
        }
    },
    "model": {
        "model as": {
            "sentences": [
                "they can be considered analogues to jdbc and sql for graph databases, providing a standardized way of querying and processing data modelled as a property graph.",
                "We model the encoding selection as a learning-to-rank problem",
                "This environment is modeled as a Markov decision process (MDP)"
            ]
        },
        "model with": {
            "sentences": [
                "Also, many works based on temporal convolution networks (TCN) attempt to model the\ntemporal causality with the causal convolution."
            ]
        },
        "model the process": {
            "sentences": [
                "The sampling process can be modeled statistically as follows."
            ]
        },
        "model each separately": {
            "sentences": [
                "On the other hand, frequent misses do noticeably increase the load raw data into 24 hourly bins and modeled each separately."
            ]
        }
    },
    "modify": {
        "modify sth": {
            "sentences": [
                "Each of these pages has an entry in the DPT that marks the log record\u2019s LSN of the oldest transaction that modified it."
            ]
        },
        "modify to": {
            "sentences": [
                "we can modify Raft slightly to remove these superficial differences"
            ]
        }
    },
    "monitor": {
        "monitor and automatically": {
            "sentences": [
                "We monitor and automatically repair faults as  primary EBS volume are synchronized with the standby EBS  part of our service."
            ]
        }
    },
    "motivate": {
        "motivate by": {
            "sentences": [
                "Motivated by the above challenges and gaps, this paper aims to enhance the performance of modeling stock\nprice movement by mitigating both data and modeling uncertainty.",
                "Motivated by the above observations, we propose Local Feature TRansformer (LoFTR)",
                "Motivated by the recent explosion of interest around blockchains",
                "Motivated by the empirical findings, we propose a novel Filter enhanced MLP approach for sequential Recommendation"
            ]
        },
        "motivate to": {
            "sentences": [
                "The above observations motivated us to design the obfuscation schemes, which protect sensitive code fragments with satisfying all efficiency criteria mentioned in [1].",
                "The intrinsic capability of EBMs to assign low energy values to in-distribution data and vice-versa motivates us to use an energy measure to characterize whether a sample is from an unknown class.",
                "These considerations motivate us to scale the argument of IM such that the transition happens, for all values of M, at the same position."
            ]
        },
        "motivate sb to do sth": {
            "sentences": [
                "The above drawbacks motivate us to develop a new privacy-preserving and communication-efficient FL scheme."
            ]
        }
    },
    "move": {
        "move away from": {
            "sentences": [
                "First, we move all lock information away from a central locking data structure"
            ]
        },
        "move to": {
            "sentences": [
                "In principle, preserving Invariant 1 requires waiting to verify that at least f other correct nodes are moving to the next round.",
                "the cluster manager imposes a barrier between epochs to guarantee that all servers move to the new epoch in unison.",
                "ACC removes this bottleneck by moving central extraction to concurrency control workers"
            ]
        },
        "move  back": {
            "sentences": [
                "that is, we shift our attention from one object  to  the  other,  and  move  back  and  forth  to  take  glimpses  of  di\ufb00erent \n\nparts of both objects."
            ]
        },
        "move  forward": {
            "sentences": [
                "hence,  a  key  question  moving  forward  is,  can  we 131 also automate the process of task construction?"
            ]
        },
        "move  away": {
            "sentences": [
                "with systems capable of building upon previous experiences rather than learning from scratch,  our  agents  will  be  more  prepared  to  handle  the  diversity  of  the  real  world and  better  equipped  to  improve  themselves;  we  will  be  able  to  move  away  from  end- to-end training from scratch."
            ]
        },
        "move toward": {
            "sentences": [
                "To be more specific, with the EMA mechanism, the new Teacher model is regularized by the previous Teacher model, and this prevents the decision boundary from drastically moving toward the minority classes"
            ]
        },
        "move via": {
            "sentences": [
                "Anchors without 3D points in BEV are removed efficiently via integral images resulting in 80 \u2212 100K non-empty anchors per frame."
            ]
        },
        "move sth. to sth.": {
            "sentences": [
                "Finally, Unified Memory transparently moves CPU pages to GPU memory."
            ]
        },
        "move from": {
            "sentences": [
                "We observe that data points\u2019 colour roughly changes as we move from the bottom-left towards the top-right, hence suggesting that as nodes gain more links they are quicker to reach the network and quicker in being reached."
            ]
        },
        "move very frequently": {
            "sentences": [
                "Oncea server\u2019s(cid:2)le hasbeen cacheentriesareonlydiscovereduponmisses(assuming deleted,themastercanmoveallthetabletsthatwerepre- that METADATA tablets do not move very frequently)."
            ]
        },
        "move data automatically": {
            "sentences": [
                "Second, moving data automatically between datacenters in response to changes in client load has long been a goal of ours, but to make that goal effective, we would also need the ability to move client-application processes be- tweendatacentersinanautomated,coordinatedfashion."
            ]
        }
    },
    "multiplexing": {
        "multiplexing  forward": {
            "sentences": [
                "by multiplexing  forward  information  from  x  and  backward  information  from  y."
            ]
        }
    },
    "mutate": {
        "mutate to": {
            "sentences": [
                "This is an especially interesting technique because it seems to influenced by the way in which viruses themselves can mutate to avoid detection."
            ]
        }
    },
    "name": {
        "name function dramatically": {
            "sentences": [
                "The addition offirst-class named functions dramatically increases Relays expressivity, allowing it to encode generic higher-order functionsand thus capture higher-level program structure."
            ]
        }
    },
    "narrow": {
        "narrow down": {
            "sentences": [
                "In order to narrow down the search space of Brent\u00a1\u00afs method, we use the uncertainty of our observation."
            ]
        }
    },
    "navigate": {
        "navigate between": {
            "sentences": [
                "Annotators could click on the bar to directly jump to highlighted paragraphs or use buttons above the policy to navigate between highlighted paragraphs."
            ]
        }
    },
    "nec-": {
        "nec- 2 essarily": {
            "sentences": [
                "Clinical Outcomes (With HR and 95% CI for CRT-ICD  who had undergone AV junction ablation but who did not nec- 2 essarily have a history of heart failure.12,16 The Post AV-Nodal  Versus ICD) Ablation Evaluation (PAVE) study followed 184 patients,  ICD, %   CRT-ICD, %   88% with a history of symptomatic heart failure, for 6 months  (n=115) (n=114) HR (95% CI) P  Value and found borderline significant improvements in left ven-"
            ]
        }
    },
    "need": {
        "need to": {
            "sentences": [
                "these local attention needs to shift the windows over the image",
                "For multi-step ahead forecasts, we further need to set the forecast horizon h.",
                "In this work, we considered three memory-bound database primitives that are typically not candidates for acceleration on GPUs due to the need to perform slow data transfers.",
                "On the other hand, the proposed solution needs to be robust to missing labels, which is inevitable for distant supervision datasets.",
                "The gas fee paid is proportional to the amount of work that is needed to execute the transaction",
                "rather, learned indexes provide a new way to think about indexing and  much more research is needed to fully understand the implications.",
                "A total of k tuples need to be dropped from the input buffers.",
                "The edges do not need to be translated, since the grammar for the edges matches the Cypher\u2019s grammar.",
                "pre-trained models need to distinguish the semantics of words",
                "For a cluster of data, we need to choose an optimal protocol according to the current workload.",
                "In Paxos any commit needs to collect from a quorum of acknowledgments,",
                "stream data need to be copied from the main memory to GPU memory via PCI-e before GPU processing",
                "Viruses need to deal with the fact that their obfuscations must continually change, or else scanners will have a known signature to match against.",
                "\"our query answerdoesnothavefalsepositives(i.e.wemaymisspoints 9 insertpintoq; inq butwillnotreturnanypointsoutsideq).thissuitsap- 10 if q.size()<k then plications that need to quickly identify regions with pois 11 width\u21902\u00b7width;height\u21902\u00b7height; but may not require all pois in the region e.g. to recom- (cid:112) mend a region with many restaurants to a user.\"",
                "there are various challenges that need to be addressed to ensure its widespread adoption"
            ]
        },
        "need because typically": {
            "sentences": [
                "The sorting is needed because typically ingonotherworkers."
            ]
        },
        "need to fully": {
            "sentences": [
                "Thenorth- needs to fully utilize available resources to execute queries boundmulti-modelaccessindicatesthatasingledatamodel and transactions robustly and e\ufb03ciently."
            ]
        },
        "be need to": {
            "sentences": [
                "An alternative mechanism is therefore needed to determine when blocked transactions can be unblocked and executed"
            ]
        },
        "need for": {
            "sentences": [
                "However the need for an additional camera that is time synchronized and calibrated with the LiDAR restricts their use and makes the solution more sensitive to sensor failure modes. In this work we focus on LiDAR-only detection."
            ]
        },
        "need not only a but b": {
            "sentences": [
                "The model needs not only to extract global, but also local information."
            ]
        }
    },
    "neutralize": {
        "be neutralize": {
            "sentences": [
                "When computing the centroid for each class, pseudo-labeled ( correct or wrong ) samples are being used together and the detrimental influences brought by false pseudo labels are expected be neutralized by correct pseudo labels."
            ]
        }
    },
    "normalize": {
        "normalize form": {
            "sentences": [
                "a word with normalized form inform (e.g., information) must occur within three words of collect."
            ]
        }
    },
    "note": {
        "note that": {
            "sentences": [
                "To see why this is the case, note that each transaction requests all locks and enters the queue all within the same critical section."
            ]
        },
        "as note in": {
            "sentences": [
                "As noted in Table 1, a self-attention layer connects all positions with a constant number of sequentially executed operations, whereas a recurrent layer requires O(n) sequential operations."
            ]
        },
        "note as": {
            "sentences": [
                "Our method noted as \u201clearn loss\u201d shows the highest per- formance for all active learning cycles."
            ]
        },
        "note that only": {
            "sentences": [
                "Note that only few of the edges are shown 2."
            ]
        },
        "note, however": {
            "sentences": [
                "It is important to note, however, that all Memcached  70000 instances in thisstudyran on identical hardware."
            ]
        }
    },
    "observe": {
        "observe in": {
            "sentences": [
                "imagen relies on new sampling techniques to allow usage of large guidance weights without sample quality degradation observed in prior work, resulting in images with higher fidelity and better image-text alignment than previously possible.",
                "However, as observed in this paper, the degrees of domain shift in different modalities are usually diverse.",
                "As we observed in Section 2.1, we need to determine the suitable tile size so that the channel and GPU resources can be fully utilized."
            ]
        },
        "observe that": {
            "sentences": [
                "In Table 53, we observe that only one method utilizing the sequence-to-sequence paradigmTrends in Integration of Vision and Language Research withcontextualinformation(i.e."
            ]
        },
        "observe a much longer train time": {
            "sentences": [
                "Thus, we observe a much longer training time on STATS than onthe IMDB dataset for these two methods."
            ]
        },
        "observe for": {
            "sentences": [
                "The gains of lexical knowledge injection are observed for 9 out of 10 language understanding tasks from the GLUE benchmark"
            ]
        }
    },
    "obtain": {
        "obtain from": {
            "sentences": [
                "The accuracy gain of the shuffle model is obtained from the privacy amplification effect",
                "Platform input notations are platform specific and can be obtained from its specification.",
                "The accuracy of policy annotations obtained from crowdwork- ers has received little prior attention.",
                "1the source code of this paper can be obtained from https://github.com/131250208/tplinker-joint-extraction this work is licensed under a creative commons attribution 4.0 international license.",
                "Both x_1 and x_2 are obtained from the adaptive neighbor encoder",
                "the KG obtains knowledge and corresponding descriptions from the network",
                "While most work focus on single tables, often obtained from the Web, some have extended modeling to more complex structures such as relational databases (Finegan-Dollak et al., 2018; Yu et al., 2018b; Wang et al., 2020).",
                "The adjacency matrix obtained from graph is normalized using Gaussian kernels and learned distances."
            ]
        },
        "obtain by": {
            "sentences": [
                "We also demonstrate signi\ufb01cant gains obtained by ensembling GraIL with various knowledge graph embedding methods in the transductive setting"
            ]
        },
        "obtain sth": {
            "sentences": [
                "To obtain detailed information of the improvements of our approach, we manually annotated 200 random examples with the complex operations needed to answer them."
            ]
        },
        "be obtain to": {
            "sentences": [
                "Since the global scope is used to reinforce each local patch, a better representation is obtained to boost the FER performance."
            ]
        },
        "be obtain from": {
            "sentences": [
                "We use a bidirectional-GRU(Chungetal.2014) eration state, we feed the relation-aware representations v(biGRU) to encode the feature maps obtained from Faster-RCNN into a hierarchical decoder to generate the story.",
                "All these values can be obtained from online profiling"
            ]
        },
        "be obtain by": {
            "sentences": [
                "which are obtained by computing per-class accuracies independently, then averaging over all 31 categories."
            ]
        },
        "obtain by directly": {
            "sentences": [
                "HasFusibleConsumer(S,i) indicates if thegram obtained by directly expanding the loop indices, and the node i in S has only one consumer j and node j can becorresponding computational graph (directed acyclic graph, fused into node i (e.g., matmul + bias_add, conv2d + relu)."
            ]
        },
        "obtain a more": {
            "sentences": [
                "Therefore, we can obtain a more general substitution by replacingthe common subgraph with a fresh input tensor."
            ]
        },
        "obtain independently": {
            "sentences": [
                "these ratings are obtained independently for model samples and reference images, and both are reported."
            ]
        },
        "obtained, however": {
            "sentences": [
                "The speedups obtained, however, are not variable-length attributes."
            ]
        }
    },
    "obtaine": {
        "obtain": {
            "sentences": [
                "Thereby, in total, we obtained 350 \u00d7 32 = 11, 200 such conversations, that we release with this paper."
            ]
        }
    },
    "occupy": {
        "occupy by almost": {
            "sentences": [
                "On x86, theoperator runtime is high due to the relatively low performanceof x86 cores for numerical computations, and the small number of cores can be fully occupied by almost any DNN operators."
            ]
        }
    },
    "occur": {
        "occur (e.g.": {
            "sentences": [
                "\"Formostlines"
            ]
        },
        "occur fre- quently": {
            "sentences": [
                "\"(3) The fast PanguDistributed Storage System CPU Mem moving current problem - due to large numbers of promo- tion events that last over short time periods"
            ]
        },
        "occur in": {
            "sentences": [
                "however, it is more challenging for a model to recognize inter-sentence relations, such as those between eldersburg and u.s. and between baltimore and u.s. because these mentions occur in different sentences and have long-distance dependencies.",
                "For other expression that occur in the GROUP or ORDER clauses",
                "Drift can still occur in translation estimation",
                "Many different kinds of changes may occur in a property graph as the world they represent evolves over time."
            ]
        },
        "occur with": {
            "sentences": [
                "SlowFuzz (Petsios et al. 2017) aims at algorithmic complexity vulnerabilities, which often occurs with a significantly high computing resource consuming."
            ]
        },
        "occur on": {
            "sentences": [
                "If a TLB miss occurs on a GPU access to CPU memory, the NVLink Processing Unit provides the address translation by walking the CPU\u2019s page table."
            ]
        },
        "occur only once": {
            "sentences": [
                "In a study of requests received reads and 70% of its keys occur only once."
            ]
        },
        "occur at later": {
            "sentences": [
                "As a result, no reads can occur at later safe timestamps,evenifthereadsdonotcon\ufb02ictwiththetransaction."
            ]
        }
    },
    "offer": {
        "offer a more powerful generalization capability": {
            "sentences": [
                "Exact Bayesian inference can be used to train this specialGP as a lightweight cardinality estimator, while offering a more powerful generalization capability than a finite wide neural network."
            ]
        },
        "offer sth. to sth.": {
            "sentences": [
                "Hence, although we offer some insights into the best implementations of different operators on CPUs and GPUs."
            ]
        },
        "offer performance": {
            "sentences": [
                "It offers better performance"
            ]
        }
    },
    "offline": {
        "be offline for": {
            "sentences": [
                "For instance, the follower has been offline for days"
            ]
        }
    },
    "offload": {
        "be offload to do": {
            "sentences": [
                "DMA operations can thus be offloaded to copy engines."
            ]
        }
    },
    "ofmorethan80,000samplespersecond)withoutinterfering": {
        "ofmorethan80,000samplespersecond)withoutinterfering set again": {
            "sentences": [
                "DELETEoperationsoccurwhenacached Thisdataistoolargetostorelocallyondisk,addinganother database entry is modi\ufb01ed (but not required to be challenge: howtoo\ufb04oadthismuchdata(atanaveragerate ofmorethan80,000samplespersecond)withoutinterfering set again in the cache)."
            ]
        }
    },
    "omit": {
        "omit a very similar method": {
            "sentences": [
                "We omit a very similar method in [19] as it hasslightly worse performance than NeuroCard."
            ]
        },
        "be omit": {
            "sentences": [
                "RMI is omitted as it has similar results to RadixSpline."
            ]
        },
        "omit to": {
            "sentences": [
                "We omit the details of the DAM and refer interested readers to [31]"
            ]
        }
    },
    "open": {
        "open up": {
            "sentences": [
                "but, this observation is most interesting  because  it  opens  up  a  wide  design  space  of  possible  meta-learning  instan- tiations  and  algorithms:  learning  to  learn  in  some  manner  via  some  supervision.",
                "such integration opens up new opportunities for fine-grained cooperation between different devices",
                "This opens up new possibilities for indexes that operate and persist data directly on the memory bus."
            ]
        }
    },
    "operate": {
        "operate in": {
            "sentences": [
                "Note that many of the recent KG reasoning methods still rely on \ufb01rst learning the PRA paths, which only operates in a discrete space"
            ]
        },
        "operate on": {
            "sentences": [
                "not directly operating on the enormous patch tokens, we propose to use the lightweight msg tokens to exchange information.",
                "yet existing systems either operate on offline snapshots, provide weak consistency guarantees, or use expensive concurrency control techniques that limit performance.",
                "These operate on sets of matches, still estimated by NN search, and thus ignore the assignment structure and discard visual information.",
                "The majority of trainable image descriptors are based on convolutional neural networks (CNNs) and typically operate on patches extracted using a feature detector such as DoG [23], yielding a sparse set of descriptors [3, 4, 10, 17, 36, 37] or use a pre-trained image-level CNN feature extractor [26, 32].",
                "The surgeon operated on the conjoined twins.",
                "In this short paper we propose PDMFRec, a novel, decentralised approach to MF, which can operate on a user\u2019s device",
                "As opposed to patch-based neural networks, our fully-convolutional model operates on full-sized images and jointly computes pixel-level interest point locations and associated descriptors in one forward pass.",
                "These features have led to numerous PM-optimized indexes that directly persist and operate on PM.",
                "Therefore, the data packing trait specifies whether the operators operate on tuples or packets and in the case of the latter, the properties that are common between all the tuples of each packet."
            ]
        },
        "operate at": {
            "sentences": [
                "Most cloud facilities operate at very low utilization.",
                "The algorithm operates as follows."
            ]
        },
        "operate with": {
            "sentences": [
                "Fuzzing test could be operated with files both with or without standard format.",
                "To minimize this contention, we freeze the original indices, migrate data in frozen indices into newly built indices, and operate with tiered indices."
            ]
        }
    },
    "oppose": {
        "oppose to": {
            "sentences": [
                "as a result, the predicted tokens at training and inference are drawn from different distributions, namely, from the data distribution as opposed to the model distribution(zhang et al., 2019).",
                "With this in mind, PDMFRec consists of two extra privacy levels as opposed to past approaches",
                "In this way, the model is learning to address a more challenging task as opposed to predicting word components"
            ]
        },
        "as oppose to": {
            "sentences": [
                "As opposed to existing reviews [19], [20], we specifically focus on deep learning methods for3D point cloudsrather than all types of 3D data."
            ]
        }
    },
    "opt": {
        "opt to": {
            "sentences": [
                "We opt to use the domain adversarial similarity loss in our model."
            ]
        },
        "opt for": {
            "sentences": [
                "Thus, we opt for prudent expansion as follows.",
                "instead opt for hash functions which lack formal robustness guarantees",
                "By contrast, STAC and UbT opt for Burn-in",
                "In the mediated protocol, we always opt for lazy modification, which means storing the modification in a local buffer during execution and applying them when the transaction is committed"
            ]
        }
    },
    "optimize": {
        "optimize with": {
            "sentences": [
                "4.3 implementation details our tplinker is implemented with pytorch and the network weights are optimized with adam (kingma and ba, 2014)."
            ]
        },
        "optimize to": {
            "sentences": [
                "Thus, it can be optimized to achieve very high throughput even for datasets in the order of tens or hundreds of gigabytes."
            ]
        },
        "optimize for": {
            "sentences": [
                "In addition, durability and recovery algorithms can be optimized for a protocol, but contradict with other implementations or optimizations.",
                "during fine-tuning, our model is optimized for the image-text contrastive loss.",
                "In [60], which is an SDA approach, unlabeled and sparsely labeled target domain data are used to optimize for domain invariance to facilitate domain transfer while using a soft label distribution matching loss."
            ]
        },
        "be optimize to do": {
            "sentences": [
                "More recent algorithms saturate the interconnect, and are optimized to access data over a low-bandwidth interconnect."
            ]
        },
        "optimize the disk i/o performance": {
            "sentences": [
                "They also proposed to write log records in batches to optimize the disk I/O performance"
            ]
        },
        "optimize sth": {
            "sentences": [
                "They also proposed to write log records in batches to optimize the disk I/O performance"
            ]
        }
    },
    "orchestrate": {
        "orchestrate  long": {
            "sentences": [
                "For    orchestrating  long-running  operations,  e.g.  a  database  volume  Figure 5: Aurora Architecture:"
            ]
        }
    },
    "order": {
        "order for": {
            "sentences": [
                "this puts an upper bound on the throughput that can be achieved while maintaining correct ordering for updates."
            ]
        },
        "in order to": {
            "sentences": [
                "Different protocolsutilize different types of aggregation rules and semantics in order to consider the final ledger out of these parallel chains.",
                "Typical convolutional architectures require highly regular input data formats, like those of image grids or 3D voxels, in order to perform weight sharing and other kernel optimizations.",
                "They can leverage the strength in numbers in order to perform more efficient Federated Learning.",
                "In the last years, modern servers are adopting hardware acceleators, such as GPUs, in order to improve their power efficiency and computational capacity."
            ]
        },
        "serial order of the transactions": {
            "sentences": [
                "The operations of multiple transactions are interleaved to provide concurrency while ensuring that the state of the database is equivalent to some serial ordering of the transactions"
            ]
        },
        "order sth": {
            "sentences": [
                "The operations of multiple transactions are interleaved to provide concurrency while ensuring that the state of the database is equivalent to some serial ordering of the transactions",
                "By ordering writes to NVM correctly, the DBMS can guarantee that all transactions are durable and atomic."
            ]
        },
        "order by": {
            "sentences": [
                "Elections are ordered by their corresponding term numbers and a node rejects RequestVote if it has already processed a request with a higher term or the same term from a different candidate."
            ]
        }
    },
    "organize": {
        "organize xxx into": {
            "sentences": [
                "A QDMR step can refer to one of the previous steps, allowing one to organize the steps into a graph."
            ]
        },
        "organize into": {
            "sentences": [
                "Processing on the GPU is done by a large number of threads organized into thread blocks (each run by one SM)."
            ]
        },
        "organize sth": {
            "sentences": [
                "The remainder of this paper is organized as follows."
            ]
        },
        "be organize as": {
            "sentences": [
                "The remainder of the paper is organized as follows"
            ]
        },
        "organize as": {
            "sentences": [
                "The rest of the survey is organized as follows",
                "The remainder of the paper is organized as follows."
            ]
        },
        "organize in": {
            "sentences": [
                "the content is organized in a two layered data structure"
            ]
        },
        "organize record sequentially": {
            "sentences": [
                "I/O performance and organized records sequentially on disk ment[20][29][32]."
            ]
        }
    },
    "originate": {
        "originate from": {
            "sentences": [
                "The positive and unlabeled examples in PU data can originate from two scenarios.",
                "The difficulty originates from the domain gap that may adversely affect the performance especially when the source and target data distributions are very different.",
                "The difference between the 2 models originates from the consumed time.",
                "The solution that improve throughput a lot originates from Calvin",
                "Originating from the NNLM, distributed representations of words are generated as a by-product during the training"
            ]
        }
    },
    "outperform": {
        "outperform by": {
            "sentences": [
                "GRAPPA outperforms RoBERTa by 3.0% in fully-sup.",
                "LIPP outperforms ALEX by up to 2.9x"
            ]
        },
        "ourperform by": {
            "sentences": [
                "In Table 2 we find that our approach outperforms the previous state-of-the-art on TABFACT by more than 6 points (Base) or more than 9 points (Large)."
            ]
        },
        "outperform sth.with": {
            "sentences": [
                "Our proposed methodology outperforms state-of-the-art baselines with remarkable improvements on two real-world datasets on top of various backbone models"
            ]
        },
        "outperform the baseline": {
            "sentences": [
                "Our proposed models consistently outperform the baseline in all metrics."
            ]
        },
        "outperform the best-known model": {
            "sentences": [
                "For human evaluation, we employ the ss model already outperforms the best-known model."
            ]
        },
        "outperform on": {
            "sentences": [
                "we show that it underperforms on tasks that operate over tables without any associated text."
            ]
        },
        "outperform deepdb mostly": {
            "sentences": [
                "It can outperform DeepDB mostly because the STATS dataset is highly correlated, so the FSPN in FLAT has a more accurate representation of the data distribution than the SPN in DeepDB."
            ]
        },
        "outperform by a large margin": {
            "sentences": [
                "LoFTR\u2020 also outperforms DRC-Net\u2020 by a large margin in this evaluation"
            ]
        }
    },
    "overcome": {
        "overcome xxx by": {
            "sentences": [
                "In this paper, we overcome this limitation by proposing the UNIFIEDSKG framework, which unifies 21 SKG tasks into a text-to-text format, aiming to promote systematic SKG research, instead of being exclusive to a single task, domain, or dataset."
            ]
        },
        "overcome xxx through": {
            "sentences": [
                "In GAP , we overcome the pre-training data challenge through the use of synthetic data."
            ]
        },
        "overcome to": {
            "sentences": [
                "HAPE has to overcome three challenges to effectively use the underlying hardware."
            ]
        }
    },
    "overhead": {
        "overhead to": {
            "sentences": [
                "Though applying only tiling would incur non-negligible overhead to query execution, combining tiling with data channels as well as concurrent execution can offset those overhead and even introduce more performance improvements over KBE approach."
            ]
        }
    },
    "overlap": {
        "overlap in": {
            "sentences": [
                "those that are concurrent and overlap in their read-write sets."
            ]
        },
        "overlap with": {
            "sentences": [
                "Any lease-quorum must overlap with any Paxos quorum",
                "As both steps can be performed in parallel, the computation overlaps with the transfer."
            ]
        }
    },
    "overwhelm": {
        "overwhelm capacities": {
            "sentences": [
                "Here the high speed data can easily overwhelm the often limited CPU and memory capacities of a stream processor unless AQP methods are used."
            ]
        }
    },
    "owe": {
        "owe to": {
            "sentences": [
                "for example, as shown in figure 1, it is easy to identify the intrasentence relations, such as (maryland, country, u.s.), (baltimore, located in, maryland), and (eldersburg, located in, maryland), owing to the occurrence of entities in the same sentence.",
                "Owing to the wealth of such tables, there has been tremendous progress on a variety of tasks in the area of table understanding."
            ]
        }
    },
    "pace": {
        "pacing, further": {
            "sentences": [
                "Furthermore, even moderately rapid ventricular rates   J u tion \u226435%, and a QRS width \u2265120 ms.1\u20134 Patients with  during AF may lead to significant reduction in  biventricular  n e  2 heart failure who have permanent atrial fibrillation (AF) have  pacing, further reducing any potential benefit of CRT.7 , 202 a reduced survival and more advanced symptoms of heart  Editorial see p 547  2  failure.5,6 Many of these patients have a clear indication for  Clinical Perspective on p 570 an implantable cardioverter defibrillator (ICD), and CRT is  perceived as a modest increment in a patient who is under-"
            ]
        }
    },
    "pack": {
        "pack as": {
            "sentences": [
                "Then we pack the context fields of a table and its selected items as the final table representation."
            ]
        },
        "be pack together": {
            "sentences": [
                "The keys and values are also packed together into matrices K and V."
            ]
        }
    },
    "pad": {
        "pad into": {
            "sentences": [
                "Since all dimension-level datasets are padded into the same size"
            ]
        }
    },
    "parallelize": {
        "parallelize very easily": {
            "sentences": [
                "In addition, the whole architecture according to the predictions from the slowest to the fastest, of Ansor can be parallelized very easily."
            ]
        }
    },
    "parse": {
        "parse with": {
            "sentences": [
                "Figure 1 shows results on Spider without and with PICARD when parsing with guards for different beam sizes and sizes of T5."
            ]
        }
    },
    "participate": {
        "participate in": {
            "sentences": [
                "Hence, we consider all users are willing to participate in each iteration",
                "we en- sured that crowdworkers could not participate in more than one condition.",
                "We refer to those that do not participate in the training process as subordinate users or simply subordinates.",
                "(2017) proposed a unified tagging scheme to convert joint extraction to a sequence labeling problem but lacks the elegance to identify overlapping relations: one entity may participate in multiple relations in the same text (figure 1)."
            ]
        }
    },
    "partition": {
        "be partition": {
            "sentences": [
                "Data is partitioned across processors."
            ]
        },
        "partition a into": {
            "sentences": [
                "We partitioned the titles into subpopulations and computed the sum of the square errors of the estimator over the partition."
            ]
        },
        "partition into": {
            "sentences": [
                "For each dataset illustrated in Table 1, we partition it into five subsets with almost equal size, which facilitates the subsequent five-fold cross validation",
                "Logical tables are horizontally partitioned into chunks of columns which are compressed into physical blocks using light-weight compression methods.",
                "We horizontally partition these datasets into s subsets randomly to simulate the HFL setting in our experiments",
                "In ACC, we partition the dataset into clusters and assign each cluster with a concurrency control protocol.",
                "It works well for partitionable workloads, which means the whole dataset is partitioned into disjoint partitions such that a transaction is highly likely to access only one partition"
            ]
        },
        "partition\u2026into": {
            "sentences": [
                "Leiet al.[77] defined a spherical convolutional kernel by partitioning a 3D spherical neighboring region into multiple volumetric bins."
            ]
        },
        "in partition scheme": {
            "sentences": [
                "We also ensure that at least one of the replicas is partitioned across several processors or machines, as in partitioned schemes"
            ]
        },
        "be partition into": {
            "sentences": [
                "We address the problem of Federated Learning (FL) where users are distributed and partitioned into clusters."
            ]
        },
        "partition an n": {
            "sentences": [
                "At the same time, DSM partitions an n-attribute relation vertically into n allpartsoftherecordareonthesamepage."
            ]
        },
        "partition as well": {
            "sentences": [
                "The main differences between our approaches izontal partitioning as well as additional hybrid-based query opti-"
            ]
        }
    },
    "partitione": {
        "partitione into": {
            "sentences": [
                "Thus, a policy\u2019s text cannot be trivially partitioned into smaller reading tasks for crowdworkers to annotate"
            ]
        }
    },
    "pass": {
        "pass to": {
            "sentences": [
                "then passes the decoded data to the query engine",
                "The tensor is flattened and passed to an FC to obtain the glyph embedding"
            ]
        },
        "pass over": {
            "sentences": [
                "coordinate buffers can be constructed via a single pass over the point list, therefore its complexity is O(n).",
                "As a solution, they advocate for the use of multiple partitioning passes, each producing a smaller number of partitions, reducing TLB misses at the expense of extra passes over the input."
            ]
        },
        "pass through": {
            "sentences": [
                "in general, current approaches obtain entity representation via information passing through nodes on document-level graphs or transformer-based structure learning.",
                "After the local feature extraction, F\u02dcA and F\u02dcB are passed through the LoFTR module to extract position and context dependent local features."
            ]
        },
        "pass from": {
            "sentences": [
                "Message overloaded will be passed from client to server"
            ]
        }
    },
    "patch": {
        "patch or even": {
            "sentences": [
                "As a result, nontrivial extensions require patching or even forking frameworks [27, 47, 52, 41, 55, 38, 51]."
            ]
        }
    },
    "pave": {
        "pave the way": {
            "sentences": [
                "The cause-specific data paves the way for disentangling the two underlying causes, interest and conformity."
            ]
        }
    },
    "pay": {
        "pay attention to": {
            "sentences": [
                "Bad models that pay attention to color or size would perform well on test set.",
                "These embedding-based models usually assume enough training instances for all relations and entities and do not pay attention to those sparse symbols",
                "Instead, data consistency in sensor-based data streams pays more attention to the trend consistency of the parameters being monitored, such as ??-loss consistency (absence of data loss) [20].",
                "Similarly, the adaptation process should also pay much attention to the privacy issue."
            ]
        },
        "pay to": {
            "sentences": [
                "which is paid to the miners and the balance is refunded to the sender"
            ]
        },
        "pay for": {
            "sentences": [
                "We show that a small number of experiments can narrow the parameter space into a regime where we pay for privacy not in terms of a loss in utility but in terms of an increased computational cost"
            ]
        }
    },
    "perceive": {
        "perceive as": {
            "sentences": [
                "The results of our respective user study suggest that highlighting paragraphs increases annotation ac- curacy and that the highlights are perceived as useful by crowd- workers."
            ]
        }
    },
    "perform": {
        "perform sth": {
            "sentences": [
                "We also perform a comprehensive analysis to demonstrate great future opportunities."
            ]
        },
        "perform to": {
            "sentences": [
                "In a cat and mouse game, programs are ob- fuscated involving complex methods such as self-modifying code, to hide pieces of code for a period of time before they become irrelevant, and attacks with varying attack tools are performed to uncover the obfuscated code."
            ]
        },
        "perform as": {
            "sentences": [
                "recurrent meta-learners have shown to perform almost as well as asymp- totically  optimal  algorithms  on  simple  reinforcement  learning  tasks  (wang et al., 2016; duan et al., 2016)."
            ]
        },
        "perform  to": {
            "sentences": [
                "our results also indicate the importance of integrating temporal information when observing the human demon- stration,  as  the  linear  loss  performs  poorly  compared  to  using  a  temporal  adaptation objective."
            ]
        },
        "perform on": {
            "sentences": [
                "To motivate our approach, we first perform a qualitative assessment on the learned attention patterns of the canonical self-attention.",
                "A lookup operation is performed on the table.",
                "Training of the model is performed on a 16x16 TPU v3 slice for 1M steps with batch size 512 and token sequence length \ud835\udc3f = 1024, which takes about 4 days.",
                "We perform experiments on four platforms"
            ]
        },
        "perform against": {
            "sentences": [
                "When there are only 0.5% to 10% of data labeled, our model consistently performs favorably against the state-of-the-art methods"
            ]
        },
        "perform in": {
            "sentences": [
                "different from previous methods, our model performs in one stage and generates triplets without a gap between training and inference."
            ]
        },
        "be perform via": {
            "sentences": [
                "In the first stage, mixed imitation and reinforcement learning is combined, while in the second stage, fine-tuning is performed via newly-introduced unseen triplets."
            ]
        },
        "perform more efficiently": {
            "sentences": [
                "Thus, it does not need to check out the connection and combinations and perform more efficiently."
            ]
        },
        "perform very well": {
            "sentences": [
                "Because the number of possible choices of each de-increase quickly as the programs become complete, whichcision is large, to make the sequential process feasible, this means the cost model performs very well for complete pro-approach keeps only top-k candidate programs after every de-grams but fails to accurately predict the final performance of cision.",
                "We therefore provide an algorithm that finds a good dependency structure using simple heuristic techniques; despite the fact that theoptimal dependency structure is not guaranteed to be produced, the algorithm nevertheless performs very well in practice.",
                "Aurora\u2019s  performance  doubles  for  each  higher  instance size and for the r3.8xlarge achieves 121,000 writes/sec  6.1.5  Throughput with hot row contention  and 600,000 reads/sec which is 5x that of MySQL 5.7 which tops  In this experiment, we report that Aurora performs very well  out at 20,000 reads/sec and 125,000 writes/sec."
            ]
        },
        "perform nod usually": {
            "sentences": [
                "Be-nodes in the DAG, because the rewriting steps across different cause we mostly care about identifying the well-performing nodes usually have less dependency."
            ]
        },
        "perform the calculation": {
            "sentences": [
                "Seperate hash tables perform a hash calculation to get a initial slot."
            ]
        },
        "perform similarly": {
            "sentences": [
                "we also find that while t5-xxl and clip text encoders perform similarly on simple benchmarks such as ms-coco, human evaluators prefer t5-xxl encoders over clip text encoders in both image-text alignment and image fidelity on drawbench, a set of challenging and compositional prompts."
            ]
        },
        "perform for": {
            "sentences": [
                "We perform comprehensive experiments for two tasks: geometric correspondences and optical flflow."
            ]
        },
        "perform with": {
            "sentences": [
                "Furthermore, we claim that descriptors should be learned only in regions for which matching can be performed with high confidence."
            ]
        },
        "perform sth to": {
            "sentences": [
                "It must perform random writes to the database at multiple locations on disk."
            ]
        }
    },
    "performe": {
        "performe at": {
            "sentences": [
                "Static instrumentation is often performed at compile time on either source code or intermediate code."
            ]
        }
    },
    "permit": {
        "permit to": {
            "sentences": [
                "First, the applications should be amenable or permitted to be approximation-enable from users?? or systems?? perspectives."
            ]
        }
    },
    "persist": {
        "persist write re-": {
            "sentences": [
                "\"In particular"
            ]
        },
        "persist in": {
            "sentences": [
                "After all the log records associated with a transaction are safely persisted in a checkpoint, the DBMS can remove those records from the log.",
                "a series of subsequent methods emerge and persist in promoting cnn performance on vision tasks."
            ]
        }
    },
    "pick": {
        "be pick up": {
            "sentences": [
                "More reliable results are then picked up as teaching materials for all sub-models in the group."
            ]
        },
        "pick out": {
            "sentences": [
                "The test selections provided by the skilled annotators were analyzed by a group of five law and privacy graduate students, who picked out phrases (4-10 words) that captured the essence of the response to a specific data practice question.",
                "Each client representation has q elements since we pick out q channels in above steps."
            ]
        },
        "pick up": {
            "sentences": [
                "A similar two-step method can also be found in [28] which utilizes the Rocchio method [29] to pick up the reliable negative examples and then uses a traditional SVM for the subsequent classification."
            ]
        },
        "pick up again": {
            "sentences": [
                "Forexample,over90%ofAPP\u2019skeysare terns, we observe a small tra\ufb03c drop on most Fridays and 31byteslong,andvaluessizesaround270Bshowupinmore Saturdays, with tra\ufb03c picking up again on Sundays and than30%ofSETrequests."
            ]
        }
    },
    "place": {
        "place in": {
            "sentences": [
                "we place inner nodes in PM",
                "APEX places in DRAM certain structures that are (1) easy to reconstruct in case of failures yet are (2) very critical to performance at runtime."
            ]
        },
        "place into": {
            "sentences": [
                "We then examine the contents of these clusters (Figure 7) and observe that TABBIE embeddings can not only be clustered into these sub-types, but also that tables from reports of the same company, but from different financial years, are placed into the same cluster."
            ]
        },
        "place on": {
            "sentences": [
                "As a result, after training, our method places the highest weight on the correct category.",
                "We can place operators on different devices with proper workloads in a fine-grained manner"
            ]
        }
    },
    "plan": {
        "plan to": {
            "sentences": [
                "In the future, we plan to extend this work from both algorithm and application level.",
                "While ACC currently supports PartCC, OCC, and 2PL, we plan to extend it to more protocols and variants in the future.",
                "We plan to extend our model to include a richer KG context.",
                "In future work, we plan to further exploit line features and incorporate other geometric primitives like planes",
                "\ufeffWe are planning to make the code and constructed dataset localgovFAQ publicly available.",
                "He plan to travel around the world.",
                "For the future, we plan to extend this work to consider other application models"
            ]
        },
        "plan to further": {
            "sentences": [
                "As future work, we plan to further refine and validate these conclusions through additional experimentation."
            ]
        },
        "plan to also": {
            "sentences": [
                "This approach works well for development, butwe plan to also implement an LLVM JIT backend that does not need a system compiler."
            ]
        },
        "plan to eventually": {
            "sentences": [
                "In addition, we plan to eventually support direct changesofPaxoscon\ufb01gurations[Lamportetal.2010;Shraeretal.2012]."
            ]
        }
    },
    "play": {
        "play a key role on": {
            "sentences": [
                "The filter component plays a key role in reducing the influence of the noise from item representations."
            ]
        },
        "play out": {
            "sentences": [
                "the generated dataset contains two parts: (i) a static part which is loaded into the system as an intermediate state of the social network, (ii) updates which are played out on this intermediate state."
            ]
        },
        "play an important role": {
            "sentences": [
                "It reveals that the task encoding plays a much more important role than image features in dynamic filer gener- ation"
            ]
        },
        "play a vital role in": {
            "sentences": [
                "Local feature extraction and description play a vital role in several high-order methods in computer vision and has received a continuous influx of attention in the past several years (c.f . surveys in [8, 13, 43, 57])."
            ]
        },
        "play a role in": {
            "sentences": [
                "Index selection plays a substantial role in database performance by reducing the I/O cost"
            ]
        }
    },
    "plot": {
        "plot in": {
            "sentences": [
                "we plot in Figure 3 the t-SNE visualizations of the features learned from the feature extractor in four settings."
            ]
        }
    },
    "plug": {
        "plug in": {
            "sentences": [
                "By plugging in the learned formulas, the system should be able to automatically infer the missing link between a pair of entities."
            ]
        },
        "be plug into": {
            "sentences": [
                "NVM can be plugged into DIMM slots to deliver higher bandwidths and lower latencies to CPUs.",
                "The outcome of GAP is a pre-trained model that can be plugged into neural semantic parsers to compute contextual representations of utterances and schema."
            ]
        },
        "plug into": {
            "sentences": [
                "Plugging the values for our dataset into these formulas yields the same answers as before"
            ]
        }
    },
    "point": {
        "point to": {
            "sentences": [
                "One point to note here is that since the simulation of disassembly scans forward from each candidate to determine the number of \u201cjunk\u201d bytes to be inserted there, it is important to ensure that such deci- sions made for one candidate are not subsequently invalidated by the insertion of junk into subsequent candidates.",
                "Different pronunciations of the same character cannot be distinguished by the glyph embedding since the logographic is the same, or the char-ID embedding, since they both point to the same character ID, but can be characterized by pinyin",
                "where prev_location points to the nearest previous occurrence",
                "an overflow bucket per record needs a 48-bit pointer to point to it",
                "therefore, for vertices we use a standard copy-on-write approach, where the newest version of the vertex can be found through the vertex index, and each version points to its previous version in the vertex block.",
                "However since the template code is shared the original function is generated in memory and the call to the rewriting engine updated to point to it."
            ]
        },
        "point out": {
            "sentences": [
                "The fourth column points out whether fuzzers support in-memory fuzzing.",
                "It is worth pointing out that none of the multi-modal baselines can beat the single-modal baselines for all diseases",
                "but we also point out the advantages and tradeoffs of running VLL in the other two architectures.",
                "It also points out that instead of building models of ever increasing sizes, we can make a decent-size model output high-quality results by leveraging those training data that resemble the instance at hand.",
                "It was later pointed out that the second line in Eq. (1) would become negative due to overfitting complex models",
                "Database research points out that a data transfer bottleneck is the main reason behind the comparatively slow adoption of GPU-enabled databases.",
                "Chaudhuri et al. [17] also point out that a join of independent samples from two relations does not yield an independent sample of their join.",
                "\ufeffSpecifically, Guo et al. (2016) point out three distinguishing characteristics of relevance matching: exact match signals, query term importance, and diverse matching requirements.",
                "As pointed out by Maxwell, a Bitcoin Core developer, resourceful miners can raise their revenues by deliberately packing transactions known only to themselves [52]."
            ]
        }
    },
    "poll": {
        "be poll by": {
            "sentences": [
                "The journal file is polled by database nodes for new transactions"
            ]
        }
    },
    "pool": {
        "pool over": {
            "sentences": [
                "We pool over the entire sequence of data."
            ]
        },
        "pool base on": {
            "sentences": [
                "[2021] proposed a novel transformer-based model (atlop) of adaptive thresholding and localized context pooling based on bert."
            ]
        }
    },
    "pre-built": {
        "pre-built into": {
            "sentences": [
                "Bitcoin nodes execute a very simple state machine pre-built into the protocol"
            ]
        }
    },
    "precede": {
        "precede it as": {
            "sentences": [
                "Given this feedback loop, each hidden state contains traces of not only the previous timestep, but all those preceding it as well."
            ]
        }
    },
    "preclude": {
        "preclude from": {
            "sentences": [
                "which precludes a refinement mapping from Paxos to Fast Paxos."
            ]
        }
    },
    "predict": {
        "predict arrival cally": {
            "sentences": [
                "Then the predicted arrival cally."
            ]
        },
        "predict with": {
            "sentences": [
                "at training time, they predict with the ground truth conditions while at inference it has to make extraction from scratch."
            ]
        },
        "predict by": {
            "sentences": [
                "given a token pair representation hi,j , the link label of token pair (wi , wj ) is predicted by equation 3."
            ]
        }
    },
    "prefer": {
        "prefer sth": {
            "sentences": [
                "Also, checkers prefer working through claims in document order as that avoids context switching overheads."
            ]
        },
        "prefer to": {
            "sentences": [
                "Thus, we prefered to potentially highlight some non-relevant paragraphs rather than omitting relevant ones.",
                "enhanced pretrained models prefer to utilize entity information and belong to coupled-based KEPTMs"
            ]
        }
    },
    "preprocess": {
        "preprocess sth": {
            "sentences": [
                "Better ways to preprocess the inputs beyond length limitation are worth studying since trivially throwing away part of the inputs could lose important information."
            ]
        }
    },
    "prescribe": {
        "prescribe on": {
            "sentences": [
                "An additional upper limit is prescribed on the total size of all newly-proposed transactions in a proposal zone."
            ]
        }
    },
    "present": {
        "present as": {
            "sentences": [
                "Network cards and storage are also virtualised and presented as individual devices to VMs."
            ]
        },
        "present in": {
            "sentences": [
                "further information about the loading phase and the data ingestion performance of the systems are presented in appendix a.",
                "The framework and differences from existing works are presented in Sect.3.1.",
                "The results presented in our studies show promise for using crowd- workers to answer questions about privacy policies.",
                "Here, we focus our discussions on the AMD GPU, and the detailed results on the NVIDIA GPU is presented in Appendix A.",
                "we review MixUp in supervised learning and present i-Mix in contrastive learning"
            ]
        },
        "present a novel model": {
            "sentences": [
                "We present a novel headline generation model, TitleStylist, to produce enticing titles with target styles including humorous, romantic, and click-baity."
            ]
        },
        "present how well": {
            "sentences": [
                "However, they do not present how well their method compared with [91, 92]."
            ]
        },
        "present for": {
            "sentences": [
                "Table 4 presents the results for sentiment analysis. Similar to previous tasks, it is observed that our approach still performs well in this task."
            ]
        },
        "present over": {
            "sentences": [
                "we present multi-class accuracies over the 16 held-out categories and compare our method to a previous domain adaptation method."
            ]
        },
        "present a more": {
            "sentences": [
                "For the sake of space, we present a more complex examplein the Appendix A, which illustrates how to handle memoryallocation when operators have dynamic shaped inputs."
            ]
        },
        "present a self-supervised framework": {
            "sentences": [
                "This paper presents a self-supervised framework for training interest point detectors and descriptors suitable for a large number of multiple-view geometry problems in computer vision."
            ]
        },
        "present sth": {
            "sentences": [
                "In this paper, we present a new protocol, called write-behind logging (WBL)."
            ]
        }
    },
    "preserve": {
        "preserve between": {
            "sentences": [
                "thereby preserving the relationships between classes."
            ]
        },
        "highlight in": {
            "sentences": [
                "We highlight the index corresponding to the monitor category in red."
            ]
        },
        "preserve more coarse": {
            "sentences": [
                "With the help of them, we could preserve more coarse-grained IR at the frontend makingoptimization simpler."
            ]
        }
    },
    "presume": {
        "presume in": {
            "sentences": [
                "However, we also presume a handful of sparse labels in the target domain."
            ]
        }
    },
    "pretrained": {
        "pretrained on": {
            "sentences": [
                "t5), pretrained on text-only corpora, are surprisingly effective at encoding text for image synthesis: increasing the size of the language model in imagen boosts both sample fidelity and imagetext alignment much more than increasing the size of the image diffusion model."
            ]
        }
    },
    "pretraining": {
        "pretraining on": {
            "sentences": [
                "while initial progress has been made towards a similar capability in computer vision, the most used paradigm still consists of first pretraining on a large amount of supervised multimodal data collected from the web, before fine-tuning the model on the task of interest (lu et al., 2019; wang et al., 2021; zellers et al., 2022)."
            ]
        }
    },
    "prevail": {
        "prevail in": {
            "sentences": [
                "it prevails in weakly supervised tasks.",
                "The last part prevails in weakly supervised tasks."
            ]
        }
    },
    "prevent": {
        "prevent from": {
            "sentences": [
                "It uses an EKF framework to prevent drift by integrating constraints",
                "However, in the early phase of training, consistency regularization regularizes the model towards high entropy predictions, and prevents it from achieving good accuracy",
                "Since we were working with a relatively small dataset, we used L1 regularization to prevent the model from overfitting the data.",
                "An effective way to alleviate the damage caused by incorrect labels is to use label smoothing, which can prevent the model from making overconfident decisions after training.",
                "The applying of optimistic concurrency control protocol prevents from high rollback",
                "To prevent privacy leakages from gradients that are calculated on users\u2019 sensitive data, local differential privacy (LDP) has been considered as a privacy guarantee in federated SGD recently.",
                "We also highlight the issues that prevent them from scaling to support larger thread counts and more complex workloads.",
                "which prevents a refinement mapping from Fast Paxos to Paxos.",
                "This definition helps prevent the adversary from distinguishing any individual user i\u2019s ratings from not rating anything",
                "A can inject messages for selective delivery but cannot change the contents of the honest parties\u2019messages nor prevent them from being delivered beyond \u2206 rounds of delay \u2014 a functionality parameter."
            ]
        },
        "prevent to": {
            "sentences": [
                "It prevents each position from attending to coming positions, which avoids auto-regressive."
            ]
        },
        "prevent sth. from do": {
            "sentences": [
                "Memory is \u201cpinned\u201d by locking the physical location of pages, which prevents the OS from moving them."
            ]
        },
        "prevent sth from do sth": {
            "sentences": [
                "These unfinished transactions may degrade performance by preventing other transactions from being successfully validated."
            ]
        }
    },
    "preview": {
        "preview a dataset": {
            "sentences": [
                "Nevertheless it is sufficient for the purpose of previewing a dataset.argue"
            ]
        }
    },
    "probe": {
        "probe sth": {
            "sentences": [
                "In this paper, we particularly probe two approaches to deal with such mixed-reasoning challenge"
            ]
        },
        "probe in": {
            "sentences": [
                "Then, all hash-table accesses during the probing phase are in cache and cache misses are averted"
            ]
        }
    },
    "proceed": {
        "proceed in": {
            "sentences": [
                "linear probing always proceeds in the same direction"
            ]
        },
        "proceed as before": {
            "sentences": [
                "Propagation proceeds as before; Figure 4b shows examples."
            ]
        },
        "proceed to": {
            "sentences": [
                "we proceed to explore how to migrate optimizations done for Paxos to work for Raft."
            ]
        }
    },
    "process": {
        "process (randomly": {
            "sentences": [
                "\"To ensure that two auto-scaling strategies will handle the same situation"
            ]
        },
        "process in": {
            "sentences": [
                "by integrating kafka into the existing workload driver and ingesting updates from a kafka queue, data ingestion and update streams can process in real-time to better re ect the streaming nature of updates."
            ]
        },
        "process by inherently": {
            "sentences": [
                "Compared to the continuousfeatures (i.e., vectors, matrices, images), categorical featuresare processed by inherently different operations."
            ]
        },
        "process bert hence": {
            "sentences": [
                "TensorFlow Foldwas not designed to process BERT hence no result was obtainable."
            ]
        },
        "process sth": {
            "sentences": [
                "In the analysis phase, the DBMS processes the log starting from the latest checkpoint."
            ]
        }
    },
    "produce": {
        "produce a doubly": {
            "sentences": [
                "the assignment of key blocks is controlled by a sorting network, which uses sinkhorn normalization to produce a doubly stochastic matrix as the permutation matrix representing the assignment."
            ]
        },
        "produce by": {
            "sentences": [
                "similarity-based method is produced by concatenating operation result of element-wise similarity, cosine similarity and bi-linear similarity between es and eo.",
                "The difference between a transaction and a message is that a message is produced by a contract while a transaction is produced by an EOA",
                "To this end we compared the annotations of our crowdworkers with those produced by our skilled annotators on the dataset of 26 privacy policies."
            ]
        },
        "produce highly": {
            "sentences": [
                "we corroborate the results of recent text-guided diffusion work [16, 41, 54] and find that increasing the classifier-free guidance weight improves image-text alignment, but damages image fidelity producing highly saturated and unnatural images [27]."
            ]
        }
    },
    "prohibit": {
        "prohibit from": {
            "sentences": [
                "However, intricate temporal patterns of the long-term future prohibit\nthe model from finding reliable dependencies."
            ]
        }
    },
    "project": {
        "project to": {
            "sentences": [
                "The patches are then linearly projected to obtain patch embeddings Expi , where E \u2208 R(P 2C)\u00d7D"
            ]
        },
        "project\u2026into": {
            "sentences": [
                "These methods first project a 3D shape into multiple views and extract view-wise features, and then fuse these features for accurate shape classification."
            ]
        },
        "project onto": {
            "sentences": [
                "In SFCNN [84], a point cloud is projected onto regular icosahedral lattices with aligned spherical coordinates."
            ]
        },
        "project back": {
            "sentences": [
                "the model then concatenates all the outputs and projects them back to a \ud835\udc37 -dimensional representation."
            ]
        },
        "be project into": {
            "sentences": [
                "A test target example is projected into the embedding space to compare to each prototype and the outputs of softmax are taken as predictions."
            ]
        },
        "project into": {
            "sentences": [
                "Item/user embeddings are projected into a subspace so that their norms and thus the sensitivity is bounded by a threshold",
                "the latter two types of models project sequences and knowledge into the same vector space avoiding heterogeneous embeddings"
            ]
        }
    },
    "proliferate": {
        "proliferate across": {
            "sentences": [
                "Many kinds of architectures and systems have proliferated across the HTAP landscape."
            ]
        }
    },
    "promise": {
        "promise for": {
            "sentences": [
                "IoT technologies are promising for industrial and manufacturing systems"
            ]
        }
    },
    "prompt": {
        "prompt with": {
            "sentences": [
                "these include open-ended tasks such as visual question-answering, where the model is prompted with a question which it has to answer, captioning tasks, which evaluate the ability to describe a scene or an event, and close-ended tasks such as multiple choice visual question-answering."
            ]
        }
    },
    "propagate": {
        "propagate up": {
            "sentences": [
                "it is possible for splits to propagate up and grow the RMI by acquiring locks bottom-up"
            ]
        },
        "propagate  to": {
            "sentences": [
                "the  loss  from  the  generated  weights  can  then  be  propagated  backwards  to adjust the embedding space."
            ]
        },
        "propagate  effectively from": {
            "sentences": [
                "however,  this  approximation does  not  permit  uncertainty  to  propagate  effectively  from  \u03b8 to  \u03c6 ."
            ]
        },
        "propagate to": {
            "sentences": [
                "Updates to the primary copy propagate to all the backup copies synchronously such that any failed primary server can be replaced by a backup server"
            ]
        },
        "propagate any correctly": {
            "sentences": [
                "Second, when input shapescontain Any dimension, the type relation needs to propagate Any correctly to the output types and relax typingconstraints that hold in the static cases when necessary."
            ]
        },
        "be propagate to": {
            "sentences": [
                "The request is written to the log block on the disk through SPDK and is propagated to the follower nodes through RDMA."
            ]
        }
    },
    "propose": {
        "propose by": {
            "sentences": [
                "As far as we know, work proposed by Xiong is the \ufb01rst research on few-shot learning for knowledge graphs"
            ]
        },
        "propose to": {
            "sentences": [
                "We propose to train the models by combining IFCA with the weight sharing technique in multi-task learning.",
                "proposed to minimize the semantic alignment loss as well as the separation loss based on deep learning models.",
                "SemtiLARE [60] is proposed to inject sentiment polarity",
                "we propose to apply InputMix together with i-Mix",
                "Then, inspired by Mitrovic\net al. (2020), we propose to simulate interventions on the error variable via data augmentations and\nlearn the disentangled seasonal-trend representations via contrastive learning.",
                "Recently , deep learning on point clouds has become even thriving, with numerous methods being proposed to address different problems in this area.",
                "we propose to use relevance models to select paragraphs that pertain to a specific data practice and to highlight those paragraphs for annotators.",
                "We thus propose to jointly learn keypoint detection and description together with a predictor of the local descriptor discriminativeness.",
                "We propose to use the building structure lines",
                "we first propose five heuristic rules to generate the index candidates",
                "in this paper, we propose to leverage pretrained image generation models instead of image data.",
                "To enforce the model to learn both long-term and short-term dependencies, we propose to alternate pretraining between packed input and single input, where the packed input is the concatenation of multiple sentences with a maximum length 512, and the single input is a single sentence",
                "Although recent studies have proposed some preliminary approaches to support update",
                "Miller et al. [43] propose to use dropout sampling to get an estimate of the uncertainty of the object detection prediction.",
                ", 2019b) proposed to remove the NSP pretraining task since it has been proved to offer no benefits for improving downstream performances",
                "Chai et al. proposed to compute gradients of user and item embeddings in user clients based on locally stored user rating data and upload gradients to the server for federated model updating",
                "we propose to virtually enlarge the number of bags by introducing the concept of pseudo-bags.",
                "Instead of fixing the number of selected confident examples, we propose to dynamically choose the number of confident examples during the self-paced learning."
            ]
        },
        "propose for": {
            "sentences": [
                "Many applications are being proposed for blockchain, leveraging the latter\u2019s two key properties",
                "multi-versioning is a common approach to ensure this property and many efficient concurrency control algorithms have been proposed for relational databases.",
                "We propose two strategies with corresponding loss functions for modeling table fact verification in the open setting"
            ]
        },
        "propose in": {
            "sentences": [
                "To our knowledge, no other works address modeling the work splitting strategy that we propose in this paper.",
                "to mitigate the issue, we propose in this paper a one-stage joint extraction model, namely, tplinker, which is capable of discovering overlapping relations sharing one or both entities while immune from the exposure bias.",
                "we first pretrain vision experts and self-attention modules of mome transformer on image-only data using masked image modeling proposed in beit [2]."
            ]
        },
        "propose a signicant generalization": {
            "sentences": [
                "They also propose a more signicant generalization, whichthey call MHIST."
            ]
        },
        "propose sth.": {
            "sentences": [
                "We propose Crystal, a library of parallel routines that can be combined together to run full SQL queries on a GPU with minimal materialization overhead."
            ]
        },
        "be propose for": {
            "sentences": [
                "Many LDP mechanisms have been proposed for different computational tasks or data types."
            ]
        },
        "propose to do": {
            "sentences": [
                "Specifically, we first propose five heuristic rules to generate the index candidates."
            ]
        },
        "propose on": {
            "sentences": [
                "WPaxosis a recently proposed nonmutating optimization on Flexible Paxos"
            ]
        },
        "be propose to": {
            "sentences": [
                "a slowly progressing key encoder, imple- mented as a momentum-based moving average of the query encoder, is proposed to maintain consistency"
            ]
        }
    },
    "prove": {
        "prove to": {
            "sentences": [
                "This modification proves to be critical, as the distribution holds key information about the relationships between categories and imposes additional structure during the training process.",
                ", 2019b) proposed to remove the NSP pretraining task since it has been proved to offer no benefits for improving downstream performances",
                "Star-Transformer has proven to achieve significant improvements against standard Transformer.",
                "Data-flow problems have been proved to be NP complete or harder."
            ]
        },
        "have be prove to": {
            "sentences": [
                "Curriculum learning has been proven to be an effective method in the field of domain adaptation and semi-supervised learning."
            ]
        },
        "be prove to": {
            "sentences": [
                "The Nakamoto longest-chain protocol was proven to be secure as long as the adversary controlled less than 50% of the mining power in the breakthrough work"
            ]
        },
        "be prove to be": {
            "sentences": [
                "Unsupervised domain adaption has proven to be an ef- fective approach for alleviating the intensive workload of manual annotation by aligning the synthetic source-domain data and the real-world target-domain samples"
            ]
        }
    },
    "provide": {
        "provide to": {
            "sentences": [
                "These methods provide improved performance compared to LiDAR-only 3D detection",
                "The storage layer provides interfaces to manage and access volumes for the file system layer",
                "Although recent work provides easy ways to convert DRAM indexes to work on PM with crash consistency",
                "CodecDB provides support to common operators"
            ]
        },
        "provide for": {
            "sentences": [
                "It provides general-purpose support for a wide range of concurrency control methods.",
                "Our goal is to provide a guide for distributed stream processing systems to make high-efficient systematic strategies.",
                "The learnable positional embedding can provide more distinguish information.",
                "Our proposed Open World Object Detection goes a step further to incrementally learn the new classes, once they are detected as unknown and an oracle provides labels for the objects of interest among all the unknowns",
                "These provided inspiration for the influential DrQA model",
                "Moments accountant can be used to provide tight analysis for the privacy loss of a sequence of adaptive mechanisms",
                "We provide two results for our method: \u2018COTR\u2019, which uses 1,000 random query points for each image pair",
                "the learnable positional embedding can provide more distinguish information for histology classification.",
                "We summarized the support that current graph databases provide for answering the queries defined above.",
                "We provide a convenient tool for non-experts in forecasting by\npre-selecting strong defaults and automating many of the modelling decisions.",
                "This work provides a unified framework for addressing the problem of visual supervised domain adaptation and generalization with deep models.",
                "Shannon\u2019s Entropy provides a lower bound for the theoretical best compression ratio",
                "As part of the annotation interface, we provide definitions for privacy-specific terms used in the questions and the response options (e.g., third parties, explicit consent, core service, etc.)."
            ]
        },
        "provide the best": {
            "sentences": [
                "\" The  method  that  is  proposed  here  is  an  auto-scaling  method  based  on  threshold that seeks to provide the best environmental responses for every mode of the  virtual machines in cloud clusters"
            ]
        },
        "provide the pre-": {
            "sentences": [
                "SLA violation occurs when a provider is unable to provide the pre- defined criteria (i.e. the Service Level Objectives (SLO)) in the SLA for the users."
            ]
        },
        "provide by simply": {
            "sentences": [
                "Theoutputof Although the basic functionality provided by simply acombinerfunctioniswrittentoanintermediate\ufb01lethat writingMapandReducefunctionsissuf\ufb01cientformost willbesenttoareducetask."
            ]
        },
        "provide the middle": {
            "sentences": [
                "the mechanism for reliability and availability provided The middle-left graph shows the rate at which data by our underlying \ufb01le system."
            ]
        },
        "provide from": {
            "sentences": [
                "The camera pose Tw,c could be provided from other sensors or state estimation methods."
            ]
        },
        "provide as": {
            "sentences": [
                "Machine attribute values are provided as integers if there are a small number of values for that attribute name or as obscured strings otherwise."
            ]
        },
        "provide by": {
            "sentences": [
                "for virtuoso and neo4j (cypher), we modified and used the reference implementations provided by the ldbc8 and an open-source cypher implementation (cypher).9 all systems were initially loaded with the synthetic social networks generated using the ldbc data generator to bring the social network into an intermediate state where the daily operations of a social networking application can be simulated.",
                "The input to our RPN is the feature map provided by the convolutional middle layers."
            ]
        },
        "provide... for...": {
            "sentences": [
                "Our network, named PointNet, provides a unified architecture for applications ranging from object classification, part segmentation, to scene semantic parsing."
            ]
        },
        "provide xxx for": {
            "sentences": [
                "Furthermore, UNIFIEDSKG currently provides the correct type of structured knowledge for each task."
            ]
        },
        "provide a way": {
            "sentences": [
                "Instead,itisonlytheviewpointthat spective, describing (captioning) the change between two is different (e.g. a moving robot can see a scene from two images provides a more meaningful way of understanding different viewpoints and should be able to distinguish that the difference between images."
            ]
        },
        "provide errors better": {
            "sentences": [
                "While Perfect Selectivities which isbased on an oracle provides errors better than wavelets it isstill worse than DeepDB since it does not take correlations across tables into account."
            ]
        },
        "be provide by": {
            "sentences": [
                "Second, system-wide atomics are unsupported. Instead, a workaround is provided by first migrating Unified Memory pages to GPU memory"
            ]
        },
        "provide a more": {
            "sentences": [
                "In contrast, our snapshot transactions provide a more natural semantics, because we support ex- ternalconsistencyoveralloperations.",
                "Section VI provides a more detailedanalysis into the implication of hardware heterogeneity onDLRM-RMC2Fig.",
                "For all intents and purposes Relay could be the IR inquestion, therefore Relay complements these systems well byproviding a more expressive IR to map computation onto."
            ]
        },
        "provide not only": {
            "sentences": [
                "These instructions provide not only highlevel information about the dynamic model behavior butalso an architectural level interface for better orchestrationand virtualization of the execution of control logic and optimized operator kernels."
            ]
        },
        "provide guarantee": {
            "sentences": [
                "Given this model,it becomes possible to provide guarantees with larger amounts of independece being more computationally"
            ]
        },
        "provide with": {
            "sentences": [
                "At last, we provide data from a real-world implementation and compare them with several high-value proof-of-work projects.",
                "We design a sparsification coded DP mechanism for FL that provides stronger privacy guarantee with the same amount of random noise.",
                "This provides us with a uniform platform to compare implementations that is not encumbered by other architecture facets."
            ]
        },
        "provide beforehand": {
            "sentences": [
                "however, because these models simply provide a similarity score between a text and an image, they can only tackle limited use cases such as classification, where a finite set of outcomes is provided beforehand."
            ]
        },
        "provide sth for": {
            "sentences": [
                "Unlike Lightning, Databus does not provide a complete service solution for hybrid transactional and analytic processing."
            ]
        },
        "provide a strongly": {
            "sentences": [
                "Figure  1:  Service-oriented  architecture  of  Amazon\u2019s  2.3  Design Considerations  platform  Data  replication  algorithms  used  in  commercial  systems  traditionally perform synchronous replica coordination in order to  provide a response within 300ms for 99.9% of its requests for a  provide a strongly consistent data access interface."
            ]
        }
    },
    "prune": {
        "prune the less": {
            "sentences": [
                "TASO prunes the less general substitution if the more generalone is valid."
            ]
        },
        "be prune by": {
            "sentences": [
                "log entries between them would be pruned by leader"
            ]
        }
    },
    "push": {
        "push forward": {
            "sentences": [
                "In the future, we plan to push forward this research direction by inspiring more sophisticated architectures which can perform both linguistic and symbolic reasoning."
            ]
        },
        "be push to": {
            "sentences": [
                "Consequently, models are pushed to explore diverse local patches adaptively."
            ]
        },
        "push down": {
            "sentences": [
                "We push down to the producers the responsibility of packing data such that the router can take routing decisions in the granularity of packets, without accessing their contents but only packet metadata."
            ]
        }
    },
    "put": {
        "put ... aside": {
            "sentences": [
                "When a transaction triggers this condition and enters the \u201cwaiting\u201d state, the main execution thread puts it aside and searches for a new transaction to execute."
            ]
        },
        "put into": {
            "sentences": [
                "Zen garbage collects old tuple versions and puts them into free lists for tuple allocations."
            ]
        },
        "put on": {
            "sentences": [
                "training naturally puts more weight on the loss of majority classes and less weight on that of minority classes",
                "In industry, much attention has been put on building cloud infrastructures that enable autoscaling as a mechanism."
            ]
        },
        "put back": {
            "sentences": [
                "Otherwise, the new cluster is put back to the pile and the first step is repeated"
            ]
        },
        "put sth. on sth.": {
            "sentences": [
                "The probe phase puts high demands on the interconnect\u2019s bandwidth."
            ]
        },
        "put down": {
            "sentences": [
                "the main difference between the two cases is that, in the former case the first frame already provides sufficient information for the subsequent changes, while in the latter the model has to precisely understand the action \u201cdrink\u201d in order to correctly generate the desired action \u2014 the lion lifts the glass to its lip, drinks and then puts down the glass.",
                "if a \u201cdrinking\u201d video is split into four individual clips of \u201cholding a glass\u201d, \u201clifting\u201d, \u201cdrinking\u201d and \u201cputting down\u201d with the same text \u201cdrinking\u201d, the model will be confused to learn the accurate meaning of drinking."
            ]
        },
        "put forward": {
            "sentences": [
                "Davis et al. [30] puts forward the definition of knowledge representation in 1993"
            ]
        }
    },
    "quarrel": {
        "quarrel with": {
            "sentences": [
                "They often quarrel with each other about trifles.",
                "The neghbors of the UAV may quarrel with each other about trifles."
            ]
        }
    },
    "query": {
        "query on": {
            "sentences": [
                "We note that, if we implement pipelined query processing on early GPUs with the support of only a single kernel."
            ]
        },
        "query about": {
            "sentences": [
                "the role of this model is to extract semantic spatial features that describe attributes that one would want to query about a visual datum: color, shape, nature, positions of objects, etc."
            ]
        },
        "query with": {
            "sentences": [
                "If certain committed transactions are unknown to a CB receiver, the receiver queries the sender with a short timeout (Line 6 to Line 7)."
            ]
        }
    },
    "question": {
        "question sth": {
            "sentences": [
                "One may question whether models are guessing the right answer."
            ]
        }
    },
    "raise": {
        "raise by": {
            "sentences": [
                "As each cluster can have a different protocol, ACC needs to address challenges raised by mixed concurrency control protocol execution"
            ]
        },
        "raise to": {
            "sentences": [
                "In 2017, the numbers raised to 5.7 Mbps and 1425 TPS [34]."
            ]
        }
    },
    "range": {
        "range from to": {
            "sentences": [
                "The algorithms range from the simple CNN architecture to complicate fusion model."
            ]
        },
        "range from": {
            "sentences": [
                "take much longer to complete for reasons that range from there is no interference from external workloads.",
                "We see that the prediction accuracy (i.e. the ratio between the number of cases ACC predicts the optimal protocol over the total number of test cases) ranges from 83% to 97% and F1 scores is from 80% to 98%.",
                "We have a broad experiment on 5 datasets, ranging from single document summarization (XSum) to multi-document summarization (Multi-News), from news domain to wiki knowledge (WikiHow) and patent (BigPatent) domains.",
                "Our approach is not limited to a low dimensional parametric model, but outputs a generic set of locally consistent image correspondences, applicable to a wide range of computer vision problems ranging from category-level image alignment to camera pose estimation. The proposed method builds on the classical ideas of neighbourhood consensus, which we review next.",
                "The delay of patching can range from months to years, exposing a significant attack surface.",
                "Each data point represents a person, and we have colour coded all points according to the number of connections (i.e. an indication of centrality), ranging from red to yellow, orange and blue (Fig. 7.1 and Fig. 7.2)."
            ]
        },
        "range of": {
            "sentences": [
                "Thus, with an extensive range of microbenchmarks, we have isolated the effects of the most important workload factors and synthesized the range of values where partitioning the data for the radix join pays off."
            ]
        }
    },
    "rank": {
        "rank with": {
            "sentences": [
                "(2019) solved this task via ranking with translation mechanism."
            ]
        },
        "rank over": {
            "sentences": [
                "Based on each method's average ROC-AUC ranks over all eight benchmark datasets, the Friedman test [8] rejects the null-hypothesis that all methods perform the same (p < 0.001), regardless of the number of propensity attributes used."
            ]
        }
    },
    "ravel": {
        "ravel out": {
            "sentences": [
                "Further, decomposition can ravel out the entangled temporal patterns and highlight\nthe inherent properties of time series"
            ]
        }
    },
    "reach": {
        "reach at": {
            "sentences": [
                "The map tasks reach at least 20% of completion."
            ]
        },
        "reach to": {
            "sentences": [
                "It is interesting to note that in 7.6 the average path of nodes reaching to the network is about 15 hops with relatively little variability."
            ]
        }
    },
    "react": {
        "react to": {
            "sentences": [
                "Some of the control theory proposals react to monitored metrics such as CPU utilisation thresholds and workload throughput.",
                "In addition, avoiding interval-based autoscaling in real setups could improve the quality of predictions by reacting to changes in the demand more quickly."
            ]
        }
    },
    "read": {
        "read-only": {
            "sentences": [
                "We  ran  the  SysBench read-only and write-only benchmarks for a 1GB data  1 GB  107,000  8,400  set (250 tables) on 5 EC2 instances of the r3 family (large, xlarge,  10 GB  107,000  2,400  2xlarge, 4xlarge, 8xlarge).",
                "These cloud database systems and numberofread-onlynodessupportedforasingledatabaseis technologieshavesuccessfullyservedboththecomplexbusi-",
                "Table 4: Replica Lag for SysBench Write-Only (msec)  Writes/sec  Amazon Aurora  MySQL  1,000  2.62  < 1000  2,000  3.42   1000  Figure 6: Aurora scales linearly for read-only workload  5,000  3.94   60,000  10,000  5.38  300,000  The results are shown in Figure 7 and Figure 6, and measure the  performance in terms of write and read statements per second    respectively.",
                "\"Therefore",
                "\"Data\tFiles As discussed previously",
                "\"In addition"
            ]
        },
        "read from": {
            "sentences": [
                "graph jet supports concurrent read queries, while updates are handled through a single writer reading from a kafka queue."
            ]
        },
        "read by": {
            "sentences": [
                "The transactions on a Blockchain network can be read by all the participant nodes in the network"
            ]
        },
        "read the most": {
            "sentences": [
                "Theclientissuesreadstotheleaderreplicaoftheappropriategroup, which acquires read locks and then reads the most recent data."
            ]
        },
        "read operations first": {
            "sentences": [
                "In this  whose request load is above a certain threshold from the scheme, read operations first check if the requested key is present  average system load) and their corresponding request load."
            ]
        }
    },
    "reaffirm": {
        "reaffirm that": {
            "sentences": [
                "The above analysis reaffirm the central takeaway of our analysis."
            ]
        }
    },
    "realize": {
        "be realize via": {
            "sentences": [
                "It could be realized via supervised learning techniques by using the Cumulative Distribution Function (CDF) of the dataset for training."
            ]
        }
    },
    "rearrange": {
        "rearrange multiple previously": {
            "sentences": [
                "finally,  we  showed  how  our  framework  for  quickly  inferring  goals  can be combined with both reinforcement learning and planning for vision-based manipula- tion skills such as maneuvering a rope into a given shape in simulation and rearranging multiple previously-unseen objects in the real world."
            ]
        }
    },
    "reason": {
        "reason out": {
            "sentences": [
                "Speci\ufb01cally, we design a new method to reason out the missing tail entity of a relation given the head entity, under the inspiration of the dual process theory in cognitive science"
            ]
        },
        "reason over": {
            "sentences": [
                "Our approach naturally generalizes to unseen nodes, as the model learns to reason over subgraph structures independent of any particular node identities"
            ]
        },
        "reason sth": {
            "sentences": [
                "While some information needs can be anticipated by hand-crafted templates, user queries are often surprising, and having models that can reason and parse that structure can have a great impact in real world applications"
            ]
        },
        "reason about": {
            "sentences": [
                "we can reason about the world through thinking without practice",
                "Unlike white-box fuzzers, greybox fuzzers do not reason about the full semantics of the PUT."
            ]
        }
    },
    "recall": {
        "recall sth": {
            "sentences": [
                "Recall that the bi-directional self-attention mechanism in transformers is unaware of order, which motivates the usage of positional and segment embeddings for text in BERT, and generalizes naturally to column and row embeddings when processing tables, in the 2-dimensional case."
            ]
        }
    },
    "recap": {
        "recap sample in dbms": {
            "sentences": [
                "Section 3 introduces our basic notation and briefly recaps sampling in DBMS."
            ]
        }
    },
    "receive": {
        "receive from": {
            "sentences": [
                "whether the data received from a network socket is used in memory comparisons.",
                "The system receives input from a 3D lidar and an IMU"
            ]
        }
    },
    "recognize": {
        "recognize as": {
            "sentences": [
                "For instance, Scrutinizer learns to recognize new types of claims and queries as more claims from a given domain are verified."
            ]
        },
        "be recognize for": {
            "sentences": [
                "The Paxos family of protocols is widely recognized for solving consensus"
            ]
        }
    },
    "recommend": {
        "recommend by": {
            "sentences": [
                "we will also compare the authors of a paper to the experts recommended by the query"
            ]
        },
        "recommend to": {
            "sentences": [
                "A na\u00a8\u0131ve approach is to recommend miners to wait for a fixed amount of time after receiving a transaction before packing it into a block."
            ]
        }
    },
    "reconcile": {
        "reconcile from": {
            "sentences": [
                "However, the remaining challenge in their approach is to reconcile results from multiple questions and segments of policy text into a coherent rep- resentation of a website\u2019s data practices."
            ]
        }
    },
    "reconstruct": {
        "reconstruct a rather than b": {
            "sentences": [
                "We choose to reconstruct X0 rather than y."
            ]
        },
        "reconstruct with": {
            "sentences": [
                "we instantiate msg-transformer as a straightforward case that the features of msg tokens are shuffled and reconstructed with splits from different locations."
            ]
        }
    },
    "record": {
        "record sth": {
            "sentences": [
                "Instead, the DBMS records a transaction\u2019s changes to a log on durable storage.",
                "we record that query as a new template"
            ]
        },
        "record on": {
            "sentences": [
                "The manufacturing information for a product (such as the manufacturing facility details, machine details, manufacturing date and parts information) can be recorded on the Blockchain"
            ]
        },
        "be record as": {
            "sentences": [
                "a transaction is prepared and each update is recorded as a transaction operation"
            ]
        }
    },
    "recover": {
        "recover from": {
            "sentences": [
                "when a shard server fails, the graph data that belongs to the shard is recovered from the backing store.",
                "To recover from a node's failure, our approach needs to retrieve its complete recovery set. For this purpose, it builds a dependency graph",
                "To a certain extent, due to the long recovery time, failures may occur repeatedly before the system recovers from an initial failure",
                "APEX can perform up to ~15x better than existing PM indexes and can recover from failures in ~42ms.",
                "Without sharing the ratings directly, training samples can still be recovered from shared gradients in FL",
                "The DBMS must recover the database from an archival version."
            ]
        },
        "recover performance": {
            "sentences": [
                "It recovers good performance quickly"
            ]
        },
        "be recover by": {
            "sentences": [
                "all committed entries can be recovered by the new leader,"
            ]
        }
    },
    "recruit": {
        "recruit on": {
            "sentences": [
                "Crowdworkers were recruited on Amazon Mechanical Turk."
            ]
        }
    },
    "recycle": {
        "recyclling": {
            "sentences": [
                "A valid question is why not recycling off-theshelf learnt knowledge/models in source domain for new domain(s)."
            ]
        }
    },
    "redirect": {
        "redirect to": {
            "sentences": [
                "redirecting to a descendant node with higher cost reduction"
            ]
        }
    },
    "redistribute": {
        "redistribute to": {
            "sentences": [
                "This adaptive strategy is very light-weight, because we can monitor the ingestion rate during batch loading and redistribute the proportion of reduced computing resources to the branch that has a higher ingestion rate",
                "To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee"
            ]
        }
    },
    "reduce": {
        "reduce to": {
            "sentences": [
                "LogTrans introduces the local convolution to Transformer and proposes the LogSparse attention\nto select time steps following the exponentially increasing intervals, which reduces the complexity to\nO(L(log L)2)",
                "After 20 batches are processed, we can reduce the deviation to less than 10%",
                "We assume that the shuffler and the analyzer are not colluded (otherwise, FLAME is reduced to LDP-FL)",
                "For the compressed solution, d dimensions of the gradient g is reduced to q dimensions by multiplying the vector g with a public random matrix.",
                "Estimating SUM queries is at least as hard as estimating COUNT queries, since any COUNT can be reduced to a SUM by setting all entries in the SUM column to 1."
            ]
        },
        "reduce into": {
            "sentences": [
                "To improve memory and computational efficiency, the 3D convolution is further reduced into two operations: matrix multiplication and 2D convolution."
            ]
        },
        "reduce by": {
            "sentences": [
                "With the same parameter setting, its memory consumption can be reduced by about 64 times.",
                "Raft*-PQL reduces the latency of read requests by performing local read.",
                "ALEX can better trade the search performance for reducing the index size by holding the inaccurate elements in wrong positons",
                "These experiments show that WBL with NVM improves the DBMS\u2019s throughput by 1.3\u00d7 while also reducing the database recovery time and the overall system\u2019s storage footprint."
            ]
        },
        "be reduce by": {
            "sentences": [
                "Class-level domain discrepancy is reduced by learning similar prototypes of each class in different domains, while sample-level discrepancy is by enforcing similar score distributions across prototypes of different domains."
            ]
        },
        "reduce on": {
            "sentences": [
                "Ds will have reduced performance on the target domain."
            ]
        },
        "reduce the overhead": {
            "sentences": [
                "It uses a hybrid replication strategy to reduce the overhead of replication, while providing transactional con\u0002sistency and high-availability",
                "To reduce the overhead of transaction processing, a dependency graph is built offline"
            ]
        },
        "reduce compilation overhead": {
            "sentences": [
                "Lazy compilation[2, 14] can be applied to reduce compilation overhead if unknown shapes are limited."
            ]
        },
        "reduce sth": {
            "sentences": [
                "It uses a hybrid replication strategy to reduce the overhead of replication, while providing transactional con\u0002sistency and high-availability"
            ]
        },
        "reduce three time": {
            "sentences": [
                "The switching time of this new transistor reduced three times."
            ]
        },
        "reduce durability even": {
            "sentences": [
                "If not,  would have reduced durability even under the simplest of failure  it implies that the values of some replicas are different."
            ]
        }
    },
    "reelect": {
        "reelected, though": {
            "sentences": [
                "As leaders get reelected, though, the throughput of the system rises to approximately 100K reads/second be- cause of two artifacts of our experiment: there is extra capacity in the system, and operations are queued while the leader is unavailable."
            ]
        }
    },
    "refer": {
        "refer to": {
            "sentences": [
                "Refer to App. B.1 for details.",
                "Representation learning refers to learning representations of the data that make it easier to extract useful information",
                "the sequence is like the handshaking of all tokens, which is the reason why we refer to this scheme as the handshaking tagging scheme.",
                "These two constraints are referred to as durability ofupdates and failure atomicity, respectively.",
                "where a(s,o) is the attention weight for entity-aware attention and asi refers to the tokens importance to the i-th entity, h is the document embedding, w1, w2 is the learnable weight matrix, k is the number of head in the transformer.",
                "Interested readers may refer to Neurocard [54] for details.",
                "We will therefore refer to it as the MAX-subset measure.",
                "The latter approach, when changing existing code, is referred to as self-modifying code, and is of particular relevance to obfuscation because of the inherent additional complexity it introduces.",
                "While computing the clustering loss with Eqn. 1, we contrast the input feature vector fc against prototype vectors, which include a prototype for unknown objects too (c \u2208 {0, 1, .., C} where 0 refers to the unknown class)",
                "This paper refers to as bag embedding-based GNN models.",
                "Paxos refers to the multi-decree version of Paxos and single-decree Paxos refers to the single-decree version",
                "Send/Recv verbs are commonly referred to as two-sided operations since each Send operation needs a matching Recv operation invoked by the remote process",
                "To convert into stroke sequences, we use Latin alphabet to represent the basic strokes and convert the char- 3 Interestingly, the word \u2018\u9b51\u9b45\u9b4d\u9b49\u2019 is in fact a Chinese idiom, which is now often used to refer to bad people who are like monsters",
                "Blocks mined in any other chain need to refer to a block in the pivot chain and use the target inferred therefrom.",
                "Obviously, referring to the organization-related references would be more bene\ufb01cial",
                "We refer to this data as silver data or silver LFs, as opposed to unknown gold ground truth.",
                "refers to the only source (\u201cen\u201d) results in Wu and Dredze.",
                "Because of this, blockchain is often referred to as a distributed ledger",
                "Window w, which refers to a finite sequence of tuples, which is the data unit to be processed in a query.",
                "We refer to those clients that partake in model training on a particular round as delegates.",
                "we refer to as bag embedding-based MIL.",
                "We refer the reader to [7] for a detailed overview of the field",
                "Refer to (Chen 2009), the macroeconomic variables in\nour experiment contain the 3-Month Treasury Bill Rate\n(TB3-daily, TB3W-weekly and TB3M-monthly)",
                "Module-like binaries mainly refer to Common Gateway Interface (CGI) files written in C programming language.",
                "We refer readers to the supplementary material for more details of training and timing analyses.",
                "we refer to elementwise clipping the x-prediction to [\u22121, 1] as static thresholding.",
                "it refers to that in an ideal scenario supervised contrastive learning should converge to an embedding where the different classes are uniformly distributed on a hypersphere",
                "We refer to [45, 46] for details on dCor.",
                "We briefly review the related work relevant to this study, and refer the readers to those surveys for more details.",
                "The interested reader is referred to these works for details.",
                "We refer to the protected user embedding of ui as \u02c6ui",
                "compared to convolutional networks, transformers enjoy the ability of referring to global features at each stage, yet the attention module brings higher computational overhead that obstructs the application of transformers to process highresolution visual data.",
                "The neural network, which we refer to as 'Sherlock' is implemented in TensorFlow."
            ]
        },
        "refer to as": {
            "sentences": [
                "Lower dimensional structure is often referred to as manifold In this setting, a generalization of convolution in the spatial learning or non-linear dimensionality reduction.",
                "One of the par- ticipant groups is chosen as the coordinator: the participant leader of that group will be referred to as the coordinator leader, and the slaves of that group as coordinator slaves.",
                "this kind of self-attention is often referred to as autoregressive or causal attention  .",
                "The transaction manager is used to im- plement a participant leader; the other replicas in the group will be referred to as participant slaves."
            ]
        },
        "be refer to": {
            "sentences": [
                "Enabling models to generalize from a source domain to a target domain is usually referred to as domain adaptation (DA).",
                "Specificity, as people are referred to as simply man or woman."
            ]
        },
        "refer as": {
            "sentences": [
                "We also refer the chain difficulty of a block as the chain difficulty of the chain ending at this block."
            ]
        }
    },
    "refine": {
        "be refine": {
            "sentences": [
                "The prediction is recursively refined by more accurate models in the next stages."
            ]
        },
        "refine to": {
            "sentences": [
                "After establishing coarse matches, these matches are re- fined to the original image resolution with the coarse-to- fine module"
            ]
        },
        "refine over": {
            "sentences": [
                "The online annotation tool, the instructions, and the wording of the questions and the response options were refined over multiple iterations."
            ]
        }
    },
    "reflect": {
        "reflect with": {
            "sentences": [
                "For example, detecting record conflicts in OCC may be reflected with transaction aborts, whereas in PartCC queue length and lock wait time may be used."
            ]
        },
        "be reflect by": {
            "sentences": [
                "there are a large number of redundant or even negative benefit indexes that cannot be directly reflected by query plans"
            ]
        },
        "reflect that": {
            "sentences": [
                "Instead, this reflects the fact that an- notators were offered answer options that included \u201cunclear\u201d and \u201cnot addressed in the policy.\u201d"
            ]
        }
    },
    "refuse": {
        "refuse to": {
            "sentences": [
                "The miner refuses to provide the full content of proposed transactions.",
                "A set of parties want to cooperate in training but refuse to share data directly"
            ]
        }
    },
    "regard": {
        "be regard as": {
            "sentences": [
                "The key insight is that indexes could be regarded as learned models that predict the position of a lookup key in the dataset.",
                "Our proposed background mixing can be regarded as an extension of this line of research by adding background of a video from one domain to a video from another domain, to explore shared semantics while learning domain-invariant features for action recognition.",
                "For simplicity, all log entries before the checkpoint can be regarded as pruned"
            ]
        },
        "with regard to": {
            "sentences": [
                "While sorting sounds like a simple solution, in high dimensional space there in fact does not exist an ordering that is stable with regard to point perturbations in the general sense."
            ]
        },
        "regard sth. as sth.": {
            "sentences": [
                "In another perspective, we can also regard the Teacher as the temporal ensemble of the Student models in different time steps"
            ]
        },
        "regard as": {
            "sentences": [
                "In ISP, an index configuration can be regarded as a result of several steps of index selection",
                "The recovery installs agreement among all correct nodes regarding the longest possible blockchain prefix as detailed in Algorithm 3.",
                "The most commonly used programming model for integrated architectures is OpenCL [49], which regards the CPU and the GPU as devices",
                "the server regards it as receiving a zero vector s\u2217 = {0}d.",
                "The single-training-set PU methods can be regarded as a special case of class-conditional label noise learning problem by treating the unlabeled data as noisy negative examples.",
                "We therefore can regard the entity \u201cThe Great Gatsby\u201d as the focus of the conversation at this point."
            ]
        },
        "regard with": {
            "sentences": [
                "It is similar to a variety of hard example min- ing [50, 11] since they regard training data points with high losses as being significant for model improvement."
            ]
        }
    },
    "regret": {
        "i regret to tell you that you can't pass the examination.": {
            "sentences": [
                "nan"
            ]
        }
    },
    "relate": {
        "relate to": {
            "sentences": [
                "hence, our work is also related to the study of [liu et al., 2020], who formulated incomplete utterance rewriting as a semantic segmentation task and motivated us to study the re problem from a computer vision perspective.",
                "Conversational KBQA is about answering a sequence of questions related to a KB.",
                "The database primitives that we implement in this work, and optimizations related to data transfers.",
                "The remedy to discover facts for new relations in a data-ef\ufb01cient way could be related to one-shot learning",
                "Characters that share common radicals often share related semantic information, such as the four characters \u2018\u9b51\u9b45\u9b4d\u9b49\u2019 all share the same radical \u2018\u9b3c\u2019 (meaning ghost), and their meanings are indeed related to ghosts and monsters",
                "The input relates to the batch while the output relates to both the operator and the batch",
                "This section provides preliminaries related to the construction of VPFedMF",
                "The second experiment set will demonstrate performance related to temporal ordering.",
                "A structure line is usually related to several segments detected in the image",
                "another ethical challenge relates to the large scale data requirements of text-to-image models, which have have led researchers to rely heavily on large, mostly uncurated, web-scraped datasets.",
                "These datasets have further boosted the research of deep learning on 3D point clouds, with an increasingly number of methods being proposed to address various problems related to point cloud processing, including 3D shape classification, 3D object detection and tracking, 3D point cloud segmentation, 3D point cloud registration, 6-DOF pose estimation, and 3D reconstruction [16], [17], [18].",
                "Here, we discuss a limited set of projects related to partitioning, optimizing protocols for main-memory systems, and mixed concurrency control protocols",
                "Our work also relates to novel class detection"
            ]
        },
        "relate...to": {
            "sentences": [
                "Overall the nodes in the upper half correspond to events related to R-tuples, while the nodes on the lower half correspond to the events related to S-tuples."
            ]
        },
        "be relate to": {
            "sentences": [
                "Lewis et al. (2021) is related to our work by retrieving related questions from pre-built largescale question-answer pairs."
            ]
        },
        "relate with": {
            "sentences": [
                "Among other works related with temporal graphs, Han present an engine for temporal graph mining, and Kostakos shows the use of temporal graphs to represent dynamic events."
            ]
        }
    },
    "relax": {
        "relax sth": {
            "sentences": [
                "Relaxing the ordering of writes to durable storage complicates WBL\u2019s commit and recovery protocols."
            ]
        }
    },
    "release": {
        "release to": {
            "sentences": [
                "The code and pretrained models will be publicly released to facilitate linguistically informed Chinese NLP"
            ]
        },
        "release very quickly": {
            "sentences": [
                "\"Cloud computing is a  model for an easy provision of network access"
            ]
        },
        "release on": {
            "sentences": [
                "[1] is the fourth studio album by american rapper , released on by aftermath entertainment, shady records, and interscope records."
            ]
        },
        "release all acquire lock": {
            "sentences": [
                "Finally, it applies the writes to the database and releases all acquired locks"
            ]
        },
        "release sth": {
            "sentences": [
                "Finally, it applies the writes to the database and releases all acquired locks"
            ]
        }
    },
    "rely": {
        "rely on": {
            "sentences": [
                "With this modification our network is able to predict per point quantities that rely on both local geometry and global semantics.",
                "Reinforcement Learning (RL) seems promising as it does not rely on pre-constructed models of the controlled infrastructure, by discovering system behaviour online without prior training.",
                "Traditional embedding models are heavily rely on rich training instances",
                "It still relys on the sliding windows for graph optimization.",
                "Several emerging proof-of-work (PoW) blockchain protocols rely on a \u201cparallel-chain\u201d architecture for scaling.",
                "In order to transfer data, push-based methods rely on a pipeline to hide transfer latency.",
                "which relies on the power of ensemble to learn a unified and generalized predictive function.",
                "Most existing works rely on a detect-then-describe approach and we focus here on the learning approaches only.",
                "On the other hand, most recent proposals either lack theoretical coherence, rely on heroic assumptions, or depend precariously on tuning hyperparameters that are, by the very problem setting, untunable.",
                "Existing data preparation and analysis systems rely on dictionary lookups and regular expression matching to detect semantic types.",
                "at training time, they predict the triplets with the ground truth tokens or subjects while at inference they have to rely on the predicted results.",
                "Furthermore, crowdsourcing could aid privacy pol- icy analysis and ease the work of regulators, who currently rely on manual inspection by experts for policy sweeps.",
                "Most existing news recommendation methods rely on centrally stored user behavior data to both train models offline and provide online recommendation services",
                "these methods achieve competitive performance compared to vanilla transformers, but halonet still wastes memories and introduces additional cost in the key and value; swin transformer relies on frequent 1d-2d feature transitions, which increase the implementation difficulty and additional latency.",
                "Our first algorithm relies on a SCC partitioning routine that takes linear time in the size of the graph",
                "The state of the art of monocular approaches relying on a MW are based on parallel and orthogonal lines alone",
                "These heuristic methods rely on the estimated cost provided by the query optimizer",
                "The first relies on random load shedding, i.e., tuples are removed based on arrival rates, but not their actual values.",
                "Apparently, the upper limit of performance of the framework depends on the quality of the pseudo label, as we rely on the pseudo label to train the student model",
                "OLTP databases typically rely on logs and checkpoints/snapshots to achieve durability.",
                "However, existing work generally relies on heavily-engineered taskspecific features and model architectures.",
                "It relies on DBAs to choose some frequent queries from the query log as the representative work- load, and uses the workload to select indexes.",
                "Of course, thesizes of relations produced as intermediate results in a query plancan not generally be computed exactly without first executing thequery plan, so in order to produce an estimate for the cost of aquery plan, the query optimizer needs to rely on quickly-compute destimates of the sizes of intermediate relations.",
                "While these approaches work for many scenarios, they all require the number of expected partitions to be known in advance or rely on simple heuristics for adding partitions if a solution cannot be found",
                "they still rely on heuristic-designed windows for attention computation.",
                "Phase1 and phase2 can co-run because they rely on different compute units",
                "General DP mechanisms, such as Gaussian or Laplacian mechanism, rely on the injection of carefully calibrated noise to the output of an algorithm directly.",
                "Since we rely on templates to extract candidate indexes",
                "Additionally, GPU memory transfers can be overlapped and rely on pinned memory buffers to achieve peak throughput.",
                "Strongly-supervised approaches relying on such annotations therefore struggle to generalize across datasets",
                "\ufeffExisting deep models rely on a single sentence representation or multiple granularity representations for matching.",
                "The achievements rely heavily on the requirement to have large quantities of annotated data for deep model learning.",
                "these methods still rely on textures for template matching",
                "Many state-of-the-art solutions rely on the field value invariant exhibited by a data structure as its signature.",
                "Our best-performing model from Table 2 relies on two innovations"
            ]
        },
        "rely upon": {
            "sentences": [
                "One of the building blocks that these analysis systems rely upon is the computation of a precise control flow graph."
            ]
        },
        "rely heavily on": {
            "sentences": [
                "Search engines such as Google 4 rely 116 heavily on the text around the pictures for image retrieval task."
            ]
        },
        "rely on heavily": {
            "sentences": [
                "Tested on theNVIDIA Titan X.works (which rely on heavily optimized libraries)on each back-end?"
            ]
        },
        "rely heavily": {
            "sentences": [
                "another ethical challenge relates to the large scale data requirements of text-to-image models, which have have led researchers to rely heavily on large, mostly uncurated, web-scraped datasets."
            ]
        }
    },
    "remain": {
        "remain to": {
            "sentences": [
                "practical solutions remain elusive due to large training data requirements algorithmic instability and lack of standard tools.",
                "However, due to the lack of rich pictographic evidence in glyphs and the weak generalization ability of standard computer vision models on character data, an effective way to utilize the glyph information remains to be found"
            ]
        },
        "remain sth": {
            "sentences": [
                "there are several remaining challenges",
                "Because solving these entailment problems requires sophisticated reasoning and higher-order operations like arg max, averaging, or comparing, human accuracy remains substantially (18 points) ahead of the best models"
            ]
        },
        "remain for": {
            "sentences": [
                "While previous studies successfully dealt with this problem for specific types of table data such as web tables, it still remains for various other types of table data."
            ]
        },
        "remain tractable": {
            "sentences": [
                "The number of progressive sample paths required to accurately query the model increases modestlywith the number of columns, but remains tractable even as thejoint data space reaches over 10190 (at 100 columns)."
            ]
        },
        "remain largely": {
            "sentences": [
                "while end-user applications of generative methods remain largely out of scope, we recognize the potential downstream applications of this research are varied and may impact society in complex ways."
            ]
        },
        "remain of": {
            "sentences": [
                "Despite these limitations, website privacy policies remain Internet users\u2019 primary sources of information on how companies collect, use, and share their data."
            ]
        }
    },
    "remind": {
        "remind sb to do": {
            "sentences": [
                "It also reminds us to allocate a small portion of privacy budget to the dimension selection."
            ]
        }
    },
    "remove": {
        "remove from": {
            "sentences": [
                "These IDs are never reused; however, machine IDs may stay the same when a machine is removed from the cluster and returned.",
                "Finally, the third column of the CONFUPDATE section shows which fuzzers remove seeds from the seed pool.",
                "besides, we observed that the u-shaped segmentation module and balanced softmax module are most important to model performance and sensitive to f1, leading to a drop of 2.18% and 1.32% in dev f1 score respectively when removed from docunet.",
                "Importantly, we remove all of the tables in the column population test set from our pretraining data to avoid inflating our results in case TABBIE memorizes the missing columns during pretraining."
            ]
        },
        "remove sth": {
            "sentences": [
                "It must also remove the updates of incomplete transactions to satisfy the failure atomicity constraint."
            ]
        }
    },
    "render": {
        "render sth": {
            "sentences": [
                "This model excels the symbolic reasoning aspects by executing database queries, which also renders better interpretability by laying out the decision rationale."
            ]
        },
        "render\u2026 into\u2026": {
            "sentences": [
                "Multiview CNNs:[20,16] have tried to render 3D point cloud or shapes into 2D images and then apply 2D conv nets to classify them."
            ]
        },
        "render progressive sample inaccurate": {
            "sentences": [
                "The large difference may render progressive sampling inaccurate on estimation."
            ]
        },
        "render by": {
            "sentences": [
                "Table 3 presents the test accuracies rendered by the compared methods, which clearly validate the top-level performance of our two proposed methods (LDCE and KLDCE) among all the approaches in this dataset."
            ]
        }
    },
    "renovate": {
        "renovate into": {
            "sentences": [
                "Autoformer still follows residual and encoder-decoder structure but\nrenovates Transformer into a decomposition forecasting architecture."
            ]
        }
    },
    "repeat": {
        "repeat at all": {
            "sentences": [
                "Unique keys (those that do not repeat at all 1e+09 ETC 1e+12 1e+08 VAR within thetrace period) are excludedfrom this count."
            ]
        }
    },
    "replace": {
        "replace with": {
            "sentences": [
                "Instead, we preserve the precise predictions by replacing the current element with a new node that accommodates these two conflicting elements",
                "When we consider interventions on user conformity, we simply replace X with pre-assigned values.",
                "we can replace the last applied instance id with the last applied index in the log with guaranteed correctness",
                "Replacing individual instructions with random data for example, while highly granular, requires a reverse mapping of data back to the original instructions.",
                "We replace it with the full version of g2o",
                "then replaces mentions with other entities of the same type of linked entities after consulting KGs"
            ]
        },
        "replace by": {
            "sentences": [
                "w/o u-shaped segmentation means that our segmentation module is replaced by a feed-forward neural network.",
                "To add support for variable mining power to Prism, a natural first approachis to replace thelongest chain rule [11]by the heaviest chain rule [13] in all the parallel chains.",
                "With probability 0.5 we replace one of both expressions by the values it evaluates to."
            ]
        },
        "replace a with b": {
            "sentences": [
                "We stop as soon as we get a value such that i ?wX[1..i]/?? ??, and then we delete X[i ?1] from X, replacing it by the last item from X in the array."
            ]
        },
        "replace of": {
            "sentences": [
                "Replacing individual instructions with random data for example, while highly granular, requires a reverse mapping of data back to the original instructions."
            ]
        }
    },
    "replay": {
        "replay sth": {
            "sentences": [
                "During the redo phase, the DBMS skips replaying the log records associated with uncommitted transactions."
            ]
        }
    },
    "replicate": {
        "replicate among": {
            "sentences": [
                "Otherwise, if secondary indices are rarely modified, we replicate them among all cores to reduce latch content",
                "A cluster database consists of several database engine nodes where data is replicated among the nodes"
            ]
        },
        "replicate to": {
            "sentences": [
                "In the partitioned phase, transactions that can be executed completely on a single partition are mastered at one of the partial replicas storing that partition, and replicated to all other replicas to ensure fault tolerance and consistency",
                "transactions that can be executed completely on a single partition are mastered at one of the partial replicas storing that partition, and replicated to all other replicas to ensure fault tolerance and consistency"
            ]
        },
        "get replicate to": {
            "sentences": [
                "Every record of the database gets replicated to one or more machines"
            ]
        },
        "replocate to": {
            "sentences": [
                "The elected leader batches client operations and replicates them to all other servers (called followers) using the AppendEntries RPC."
            ]
        },
        "replicate  otherwise": {
            "sentences": [
                "To copy  scalability and availability: Data is partitioned and replicated  otherwise, or republish, to post on servers or to redistribute to lists,  using consistent hashing [10], and consistency is facilitated by  requires prior specific permission and/or a fee."
            ]
        }
    },
    "replsce": {
        "replace sth. with sth.": {
            "sentences": [
                "Finally, we replace the Zero-Copy transfer method with the Coherence transfer method in the Zero-Copy join."
            ]
        }
    },
    "reply": {
        "reply to": {
            "sentences": [
                "I was unsure how to reply to this question."
            ]
        }
    },
    "report": {
        "report to": {
            "sentences": [
                "Using the Wubi scheme \u2014 a Chinese character encoding method that mimics the order of typing the sequence of radicals for a character on the computer keyboard \u2014- is reported to improve performances on Chinese-English machine translation [Tan et al",
                "The basic process of feature extraction involves concurrency control workers collecting workload statistics in parallel and reporting them to a centralized coordinator."
            ]
        },
        "report xxx from": {
            "sentences": [
                "In the top section, we report the results from PEGASUS (Zhang et al., 2020) paper."
            ]
        },
        "report in": {
            "sentences": [
                "the statistics are reported in table 1."
            ]
        },
        "report sth.": {
            "sentences": [
                "We report the quantitative performance of our proposed model compared with different baselines and state-of-the-art approaches across three benchmarks in Table 1."
            ]
        }
    },
    "represent": {
        "represent as": {
            "sentences": [
                "First, we propose to represent video as a graph and then utilize temporal contrastive self-supervised learning over the graph representations as a nexus between source and target domains to align features, without requiring any additional adversarial learning, as most prior works do in video domain adaptation.",
                "A point cloud is represented as a set of 3D points {Pi|i= 1, ..., n}, where each point Pi is a vector of its (x, y, z) coordinate plus extra feature channels such as color, normal etc.",
                "the knowledge graph (KG) is the knowledge base represented as a network with entities as nodes and relations as edges",
                "A KG G is represented as a collection of triples"
            ]
        },
        "represent by": {
            "sentences": [
                "Each workflow task is represented by a circle with an identifier within it and runtime in time units on the right.",
                "based on the information represented by these three sequences, a triplet can be decoded: (new york city, mayor, de blasio).",
                "In the basic setting each point is represented by just its three coordinates(x, y, z).",
                "The scheduling class is represented by a single number, with 3 representing a more latency-sensitive task",
                "A semantic network is a conceptual network represented by a directed graph where nodes represent concepts"
            ]
        },
        "be represent as": {
            "sentences": [
                "a \"cell\" in a logically multidimensional space is represented as a tuple with som attributes ifentifying the location of the cell in the multidimentional space and other attributes containging the values of the measures of the cell."
            ]
        },
        "represent with": {
            "sentences": [
                "To represent each column with a fixed-length numerical vector, we implemented the Distributed Bag of Words version of Paragraph Vector.",
                "3D data can usually be represented with different formats, including depth images, point clouds, meshes, and volumetric grids.",
                "Classical methods for these tasks are based on a detect-then-describe paradigm where separate handcrafted methodsare used to first identify repeatable keypoints and then represent them with a local descriptor.",
                "we represent a string with its Karp-Rabin fingerprint"
            ]
        },
        "represent with fully": {
            "sentences": [
                "the input feature extractor and output function can be represented with fully connected  neural  networks  with  one  or  more  hidden  layers,  which  we  know  are  uni- versal function approximators, while  qn w corresponds to a set of linear layers with                                                    i=1    i non-negative input and activations."
            ]
        },
        "represent image": {
            "sentences": [
                "We need scene graphs to represent images."
            ]
        },
        "represent it explicitly": {
            "sentences": [
                "Thus, from our compactmodel, we can recover the joint distribution; we do not need torepresent it explicitly."
            ]
        },
        "represent the n": {
            "sentences": [
                "meta[Constant][n] represents the n-th constant in the pool."
            ]
        },
        "represent (cid:147)real": {
            "sentences": [
                "Theycanbeas- the same domainnear each other makes some host and signed by Bigtable, in which case they represent (cid:147)real domainanalysesmoreef(cid:2)cient."
            ]
        },
        "represent a very": {
            "sentences": [
                "Inmosten- terpriseapplications,thereareasmallnumberofqueryclasses(and F. HYRISELAYOUTSELECTION henceattributegroups)thatareaccessedfrequently;ad-hocqueries may be run but they will likely represent a very small fraction of In this sectionwe illustratethe in\ufb02uenceof workloadvariations totalaccess."
            ]
        }
    },
    "representation": {
        "representation through": {
            "sentences": [
                "The framework representation organizes knowledge through a hierarchy of frames"
            ]
        }
    },
    "represente": {
        "represente by": {
            "sentences": [
                "To overcome the issue of constructing a loss\nfunction with complex-valued representations, each frequency can be uniquely represented by its\namplitude and phase representations"
            ]
        }
    },
    "reproduce": {
        "reproduce by": {
            "sentences": [
                "results of baselines are all quoted directly from (wei et al., 2020) except for etl-span, of which the results are reproduced by the official implementation."
            ]
        }
    },
    "request": {
        "have request": {
            "sentences": [
                "Once a transaction has requested its locks, it is added to the TxnQueue."
            ]
        },
        "request to": {
            "sentences": [
                "or part replacement requests to the machine maintenance vendors"
            ]
        }
    },
    "require": {
        "be require to": {
            "sentences": [
                "\ufeffTaking question answering as an example,given a pair of question and answer, a matching function is required to determine the matching degree between these two sentences.",
                "Each insertion of a key is required to find a series of non-empty sets"
            ]
        },
        "require to": {
            "sentences": [
                "Before initializing any structure lines, it requires to estimate the dominant directions in the world frame.",
                "but it is required to maintain consensus in the network where there is no centralized authority or a trusted third-party",
                "Many applications require to evaluate the values of a set of sources with respect to a set of targets.",
                "obviously, more investigation is required to better understand the impact of learned indexes for disk-based  systems.",
                "other services required to federate not yet been implemented by mainstream providers.",
                "We remove paragraph vectors from the feature set, as it requires to train a separate table column-specific paragraph vector model.",
                "KB-truncated join approximation problem in which we are required to delete kA, kB tuples from the two joining relations respectively as opposed to k tuples overall.",
                "Paxos requires all clients requests to be sent to a leader for better throughput.",
                "more research is required to better understand this problem.",
                "Last but not least, the better prediction performance requires to separate the learning process of encoder and decoder modules",
                "Par- ticipants were required to be U.S. residents and to have at least a 95% approval rating for 500 completed tasks.",
                "The nature of encoding schemes requires us to choose the encoding scheme correctly for a given dataset, which is not trivial"
            ]
        },
        "require for": {
            "sentences": [
                "However, self-modification for obfuscation is usually ex- pected to raise the adversary\u2019s cost of reverse-engineering by the cost required for deobfuscation."
            ]
        },
        "require to better": {
            "sentences": [
                "obviously, more investigation is required to better understand the impact of learned indexes for disk-based systems."
            ]
        },
        "require not only": {
            "sentences": [
                "This is challenging photostream usually has signifficantly large visual variance because it requires not only the full understanding of each (Fu et al. 2015), such as in Figure 1."
            ]
        },
        "require much more": {
            "sentences": [
                "Second, visual storytelling requires much more complicated textual form."
            ]
        },
        "require sb. to do": {
            "sentences": [
                "Higher interconnect performance requires us to reconsider how well a NP-HJ that spills to CPU memory performs on GPUs."
            ]
        },
        "require to manually": {
            "sentences": [
                "Most DL frameworks have chosen a specific setof fixed quantization schemes and datatypes due to the effortrequired to manually implement operators for each pairing."
            ]
        },
        "require sb to sth": {
            "sentences": [
                "All the above works considered privacy issues in conventional FL that requires parties to share model weights."
            ]
        },
        "require sth": {
            "sentences": [
                "In more detail, firstly, the SAP architecture requires an interface for log shipping inside the OLTP engine."
            ]
        },
        "require by": {
            "sentences": [
                "The transactions include input parameters which are required by the functions in the contract"
            ]
        },
        "require  prior": {
            "sentences": [
                "To copy otherwise, or republish, to post on servers or to redistribute to lists, require  prior  specific  permission  and/or  a  fee."
            ]
        },
        "require  only": {
            "sentences": [
                "Compared  to  Bigtable,  Dynamo  targets  applications  that  require  only  key/value  access  with  primary  focus  on  high  1 http://freenetproject.org/, http://www.gnutella.org  availability where updates are not rejected even in the wake of Table 1: Summary of techniques used in Dynamo and  Key K  their advantages."
            ]
        },
        "require at least": {
            "sentences": [
                "The context encodes system metadata about the  latency sensitive applications that require at least 99.9% of read  object that is opaque to the caller and includes information such as  and  write  operations  to  be  performed  within  a  few  hundred  the version of the object."
            ]
        },
        "require to uniformly": {
            "sentences": [
                "Subsequently, it decides on how many time slices will  An  important  advantage  of  the  client-driven  coordination  be available to background tasks, thereby using the feedback loop  approach is that a load balancer is no longer required to uniformly  to limit the intrusiveness of the background activities."
            ]
        }
    },
    "reshape": {
        "reshape to": {
            "sentences": [
                "The decoder\u2019s output is reshaped to form a reconstructed image"
            ]
        }
    },
    "reside": {
        "reside in": {
            "sentences": [
                "In this model, data primarily resides in CPU\u2019s main memory.",
                "We identify and experimentally prove that the tradeoff resides in the network layer via the existence of fresh transactions"
            ]
        },
        "reside on": {
            "sentences": [
                "We define a federated multi-view setting, the aim of which is to learn a model over data that not only resides on m distributed nodes",
                "A smart contract is a piece of code that resides on a Blockchain and is identified by a unique address"
            ]
        }
    },
    "resize": {
        "resize to": {
            "sentences": [
                "The corresponding regions are then used to extract feature map crops from each view, which are then bilinearly resized to 3 \u00d7 3 to obtain equal-length feature vectors"
            ]
        },
        "resize with": {
            "sentences": [
                "All images are resized with shorter dimensions equal to 480"
            ]
        }
    },
    "reslut": {
        "result in": {
            "sentences": [
                "This easily results in high correlation values at multiple incorrect locations"
            ]
        }
    },
    "resolve": {
        "be resolve": {
            "sentences": [
                "Fortunately, this situation can be resolved by a simple observation"
            ]
        },
        "be resolve within": {
            "sentences": [
                "If no distributed transactions are involved, command logging can reach the recovery performance similar to the other methods, because transaction dependencies can be resolved within each site"
            ]
        }
    },
    "resort": {
        "resort to": {
            "sentences": [
                "the system resorts to the second stage only for a small subset of transactions, i.e.",
                "we resort to a greedy algorithm to solve.",
                "state-of-the-art Transformer models have resorted to various downsampling strategies to reduce the feature size and the memory consumption",
                "To get good accuracy results in the prediction, we resort to active learning.",
                "To circumvent the impossibility of distribution matching at class-level, we resort to pseudo labels.",
                "Some works [22, 23, 24] resort to cryptographic techniques (like powerful yet expensive homomorphic encryption and garbled circuits)",
                "The state-of-the-art Transformer models have resorted to various downsampling strategies to reduce the feature size and the memory consumption"
            ]
        }
    },
    "respect": {
        "with respect to": {
            "sentences": [
                "Theoretically, we provide analysis towards understanding of what the network has learnt and why the network is robust with respect to input perturbation and corruption.",
                "With respect to the dependence on program source code and the degree of program analysis, fuzzers could be classified as white box, gray box and black box.",
                "the only difference with respect to car detection is that the stride of the first convolutional layer in the RPN is 1 instead of 2."
            ]
        }
    },
    "respond": {
        "respond to": {
            "sentences": [
                "We conduct extensive simulation studies to examine how our systems respond to varyingmining power.",
                "The tall box shows a subset of nodes which correspond to the events at a certain time t = 3.",
                "No punishment is prescribed to upstream peers who do not respond to the queries on newly proposed transactions, as it is difficult to locate the responsible parties for the delay."
            ]
        },
        "respond with": {
            "sentences": [
                "they must respond with \u201cyes\u201d, \u201csomewhat\u201d, or \u201cno\u201d."
            ]
        }
    },
    "response": {
        "response to": {
            "sentences": [
                "FineStream responses to these situations with different strategies based on the analysis for dynamic workloads"
            ]
        }
    },
    "restore": {
        "restore from": {
            "sentences": [
                "Even when a test input corrupts the program state by hitting bugs, a fresh target program state can always be restored from an existing snapshot"
            ]
        },
        "restore the last snapshot": {
            "sentences": [
                "Once failure occurs, the database can restore the last snapshot and redo all the transactions in the command log to re-establish the database state"
            ]
        },
        "restore sth": {
            "sentences": [
                "This allows the DBMS to identify the log records to replay during recovery to restore the page."
            ]
        }
    },
    "restrict": {
        "restrict to": {
            "sentences": [
                "which utilized self-attention mechanism and restricted it to attend to local neighborhoods.",
                "we restrict the vocabulary to 30000.",
                "in the official implementation, casrelbert is restricted to processes one sentence at a time, which means it is seriously inefficient and difficult to deploy.",
                "While the FinTabNet dataset is restricted to the homogenous domain of financial tables, these tables cluster into sub-types such as consolidated financial tables, jurisdiction tables, insurance tables, etc."
            ]
        },
        "be restrict to do": {
            "sentences": [
                "For this reason, previous approaches either cannot scale beyond GPU memory, or are restricted to partitioning-based joins."
            ]
        }
    },
    "resubmitted": {
        "resubmitted by": {
            "sentences": [
                "transactions that have not executed on the backing store, as well as all node programs, are reexecuted by weaver with a fresh timestamp after recovery, when resubmitted by clients."
            ]
        }
    },
    "result": {
        "result in": {
            "sentences": [
                "there is no distributed coordination, resulting in high scalability.",
                "Moreover, recently it was shown that KM2 results in more accurate classifier performance than TIcE on SAR PU data.",
                "Any operations that involve writing more than eight bytes could result in inconsistencies as currently only 8-byte PM writes are atomic.",
                "the solution could miss some optimization opportunities or even result in an incorrect protocol.",
                "Under high-contention workloads, this can result in reduced concurrency and poor CPU utilization.",
                "model weight is usually of much higher dimension than model predictions, resulting in huge communication overhead and higher privacy cost.",
                "We ablate contrastive clustering (CC) and energy based unknown identification (EBUI) to find that it results in reduced performance than standard ORE.",
                "results in low throughput and long transaction confirmation latency.",
                "This architecture results in the heuristic being invoked at three different time-scales, evaluating different adaptation decisions on each time period.",
                "Such asynchronous semantics could result in better performance",
                "The matching process resulted in 6, 146, 940 columns matching the 275 considered types.",
                "The itera- tive design resulted in substantial usability improvements.",
                "However, it may result in suboptimality in integrated architectures for mainly two reasons",
                "Testcases generated by black box and gray box fuzzers are hard to pass the validation for a blind generation strategy, which results in quite low efficient fuzzing.",
                "Earliest-arrival path, defined as a path that results in the earliest arrival time starting from a source x to a target y.",
                "As a third contribution, we introduce the new task of actor shift, and a corresponding audio-visual video dataset ActorShift, to challenge our approach when the change in actors results in large variation in activity appearance.",
                "Furthermore, the feature duplication and high correlation among features can also result in overweighting these correlated features in learning user preference, leading to sub-optimal performance and unstable recommendations.",
                "Transformers have to adopt the\nsparse versions of point-wise self-attentions for long series efficiency, resulting in\nthe information utilization bottleneck.",
                "Since OCC and 2PL maintain a different set of metadata, T3 and T2 are not aware of the conflict of T2 reading R1 and T3 writing R1, which may make database result in an inconsistent state.",
                "The abundance of customized solutions in this space also result in duplication due to the lack of standards.",
                "imagen relies on new sampling techniques to allow usage of large guidance weights without sample quality degradation observed in prior work, resulting in images with higher fidelity and better image-text alignment than previously possible.",
                "However, max-pooling only retains the maximum elements from a specific view, resulting in information loss.",
                "prior works show that joint learning can result in a noticeable performance gain."
            ]
        },
        "result into": {
            "sentences": [
                "The mean inter-arrival interval is 117.77 s which results into arrival rate of 30.57 jobs per hour."
            ]
        },
        "result from": {
            "sentences": [
                "Observational interaction data could result from users conformity towards popular items",
                "The final class predictions result from the output of the final softmax layer, corresponding to the network's confidence about a sample belonging to each class, the predicted label then is the class with the highest confidence.",
                "Moreover, specifying a single disk service time distribution does not take into account differences in disk service time that result from data storage and file organization."
            ]
        },
        "result on": {
            "sentences": [
                "As an empirical example, Fig.(1) shows the forecasting results on a real dataset.",
                "they yield satisfactory results on specific datasets, they cannot adapt to the complex environments commonly encountered in autonomous driving.",
                "Disambiguation task and achieves state of the art results on the Word in Context task",
                "SuperGlue outperforms other learned approaches and achieves state-of-the-art results on the task of pose estimation in challenging real-world indoor and outdoor environments."
            ]
        },
        "result in incomparably": {
            "sentences": [
                "we always use our learned model, since having the number of neurons mentioned in [8], results in incomparably\nlarge execution time."
            ]
        },
        "result model consistently": {
            "sentences": [
                "the resulting model consistently outperforms standard transformer and compares favorably to larger et models, with a significantly reduced search cost."
            ]
        },
        "result in correctly": {
            "sentences": [
                "The result of the heterogeneous device placement composes with memory planning and shape functioninsertion resulting in correctly placed allocations."
            ]
        },
        "result in less": {
            "sentences": [
                "is it not possible thatprediction with the embedding parameters that are less frequentlysynchronized may result in less accurate results (i.e., bias)?)"
            ]
        },
        "result in over": {
            "sentences": [
                "nonetheless, static thresholding still results in over-saturated and less detailed images as the guidance weight further increases."
            ]
        },
        "result in well": {
            "sentences": [
                "For  example,  the  service  that  maintains  requirements:  shopping cart (Shopping Cart Service) served tens of millions  requests that resulted in well over 3 million checkouts in a single  Query Model: simple read and write operations to a data item that  day and the service that manages session state handled hundreds  is uniquely identified by a key."
            ]
        }
    },
    "resume": {
        "resume from": {
            "sentences": [
                "the SMO can resume from step 3 by observing stage in the log"
            ]
        }
    },
    "retain": {
        "retain the most important coefficients": {
            "sentences": [
                "The key idea is to deterministically retain the most important coefficients while randomly rounding the other coefficients either up to a larger value (called a rounding value) or down to zero."
            ]
        },
        "retrain": {
            "sentences": [
                "We still retained the question rewrites even if no answer was found on the web."
            ]
        }
    },
    "retrieve": {
        "retrieve from": {
            "sentences": [
                "At each step, query-relevant entities and relations are retrieved from the neighborhood and organized as a cognitive graph",
                "We train Sherlock on 686, 765 data columns retrieved from the VizNet corpus by matching 78 semantic types from DBpedia to column headers.",
                "retrieving related information from the provided corpus",
                "For the deleted records that have been migrated from current data storage to historical data storage, we retrieve them simply from the historical data storage.",
                "Realistic systems must first retrieve evidence from a large data source. That is, realistic systems must operate in an open setting.",
                "Thus, we explore whether retrieving from the training data, which is usually much smaller than a large-scale corpus, can help improve the performance."
            ]
        },
        "retrieve the most": {
            "sentences": [
                "The proposed model not only retrieves the most suitable descriptions for a given 344 image but also generates novel descriptions."
            ]
        },
        "be retrieve from": {
            "sentences": [
                "\ufeffSubtask C is similar to A but comments were retrieved from an external thread, which increases the difficulty of the task."
            ]
        },
        "be efficiently retrieve": {
            "sentences": [
                "By using the index, the time-dependent transactions of a transaction can be efficiently retrieved"
            ]
        }
    },
    "retry": {
        "retry from": {
            "sentences": [
                "It detects this case by checking data nodes's lock status and retries from root if it is set."
            ]
        },
        "keep retry with": {
            "sentences": [
                "If response timeout happens,PolarSwitch would keep retrying with exponential backoff while detecting whether leader election happens"
            ]
        }
    },
    "return": {
        "return to": {
            "sentences": [
                "We return to this point during our discussion.",
                "This is done by rerouting call instructions through a specialized branch function, which branches to its appropriate target function, but returns to some predetermined offset from the original call instruction.",
                "The robot returns to the initial position after 10 minutes of driving"
            ]
        }
    },
    "reuse": {
        "reuse buffer as": {
            "sentences": [
                "The basicoptimization rule is to free buffer as soon as it has no users,and reuse buffers as much as possible according to \"shapecompatibility\"."
            ]
        }
    },
    "reveal": {
        "reveal sth": {
            "sentences": [
                "Our studies also reveal the importance of accurate program acquisition for improving the performance of table fact verification."
            ]
        },
        "reveal that": {
            "sentences": [
                "Furthermore, Fig. 4 reveals that those different types of operators do not occur independently of each other."
            ]
        }
    },
    "revert": {
        "revert to": {
            "sentences": [
                "We cannot be in the simple case, so we revert to the original implementation."
            ]
        }
    },
    "review": {
        "review by": {
            "sentences": [
                "This allows for the forecast to be reviewed by a domain expert who\nmay make adjustments when appropriate, a procedure known as human-in-the-loop."
            ]
        }
    },
    "rewrite": {
        "rewrite as": {
            "sentences": [
                "hence, our work is also related to the study of [liu et al., 2020], who formulated incomplete utterance rewriting as a semantic segmentation task and motivated us to study the re problem from a computer vision perspective."
            ]
        }
    },
    "rewritte": {
        "be rewrite as": {
            "sentences": [
                "With the disentanglement, our model Equation (1) can be rewritten as:"
            ]
        }
    },
    "rid": {
        "rid of": {
            "sentences": [
                "by getting rid of the limited size of image-text pairs and their simple and short captions, stagewise pre-training on large amounts of image-only and text-only data helps vlmo to learn more generalizable representations."
            ]
        }
    },
    "roll": {
        "roll  out": {
            "sentences": [
                "another approach to meta-learning is to train memory-augmented models on many tasks,  where  the  recurrent  learner  is  trained  to  adapt  to  new  tasks  as  it  is  rolled  out."
            ]
        }
    },
    "root": {
        "root at": {
            "sentences": [
                "each graph partition consists of a set of vertices, all outgoing edges rooted at those vertices, and associate attributes."
            ]
        },
        "root in": {
            "sentences": [
                "due to a well-known tradeoff rooted in its security and performance\u2019s conflicting requirements on the block size and the block interval\u2014its security demands small blocks and long block intervals"
            ]
        }
    },
    "round": {
        "round to": {
            "sentences": [
                "We derive this value by rounding to the nearest integer the service rate calculated based on the mean task runtime in the workloads."
            ]
        }
    },
    "rout": {
        "rout to": {
            "sentences": [
                "Transactions are routed to a global pool of transaction workers, each of which is a thread or a process occupying one physical core."
            ]
        },
        "rout information locally": {
            "sentences": [
                "systems  (e.g.,  [14])  employ  O(1)  routing  where  each  peer  Bigtable is a distributed storage system for managing structured  maintains enough routing information locally so that it can route  data.",
                "enough routing information locally to route a request to the  appropriate node directly."
            ]
        },
        "rout request directly": {
            "sentences": [
                "node based on load information, or (2) use a partition-aware client  library that routes requests directly to the appropriate coordinator  To illustrate the use of vector clocks, let us consider the example  nodes."
            ]
        }
    },
    "route": {
        "route to": {
            "sentences": [
                "Depending on the routing policy, a packet may be routed to a consumer that does not have access to its content.",
                "Queries can be routed to the OLAP replicas according to users\u2019 preferred max acceptable staleness."
            ]
        }
    },
    "rule": {
        "rule out": {
            "sentences": [
                "\ufeffThe cost of ensembles, the adverse scaling of network parameters in gradient-based strategies, and a history of deeming neural networks to be overconfident effectively rule out the most predominantly used query strategies.",
                "In particular, differential privacy rules out memorization of sensitive information in a strong information theoretic sense"
            ]
        },
        "be rule by": {
            "sentences": [
                "While the onset of deep- learning was pre-dominantly ruled by supervised learning"
            ]
        }
    },
    "run": {
        "run in": {
            "sentences": [
                "synchronicity assumptions interfere with debugging, and maybe violated by network delays under heavy load and systems running in virtualized environments.",
                "Since both database and application run in the same process and thus address space, there is a unique opportunity for efficient data sharing which needs to be exploited.",
                "DCPMM can run in Memory or App Direct modes"
            ]
        },
        "run around": {
            "sentences": [
                "The two agents are always running around the center node.",
                "the two animals are always running around or fighting each other."
            ]
        },
        "run it ef\ufb01ciently": {
            "sentences": [
                "\"It has beenusedacross a widerange ofdomains within causeitmakesitpossibletowriteasimpleprogramand Google"
            ]
        },
        "run at almost": {
            "sentences": [
                "\"We see no fundamental reason why the cost of serverless computing should be higher than that of serverful computing"
            ]
        },
        "run on": {
            "sentences": [
                "A crypto currency application is an example of a Dapp that runs on a Blockchain network",
                "We run the TAPAS algorithm on top of the input data to limit the input size.",
                "The following experiments run on a machine provisioned with two 12-core Intel Xeon.",
                "However, these experiments were run on a single core machine with no physical contention for lock data structures.",
                "Blockchain is the fundamental system of bitcoin running on non-structured P2P networks."
            ]
        },
        "run at": {
            "sentences": [
                "We can conclude that the resources either significantly over-provisioned (Hist and ConPaaS) or already provisioned resources are running at low utilization (React, Adapt, Reg, Plan, and Token)."
            ]
        },
        "run as": {
            "sentences": [
                "The EVM runs as a sandbox and provides an isolated execution environment",
                "cassandra is run as a separate process, while berkeleydb is used in embedded mode."
            ]
        },
        "run into": {
            "sentences": [
                "The build phase runtimes are less affected by caches as writes to hash table end up going to memory."
            ]
        },
        "run the risk of": {
            "sentences": [
                "On the other hand, any image matching algorithm runs the risk of being used for malevolent tasks"
            ]
        },
        "run out": {
            "sentences": [
                "The privacy budget and the hyper-parameters are predefined, so that privacy budget will not run out during the training"
            ]
        },
        "run query faster": {
            "sentences": [
                "PAX reduces cache delays related to data accesses, and therefore runs queries faster."
            ]
        }
    },
    "s": {
        "s be always": {
            "sentences": [
                "Note that s is always advanced to the highest value in MinNextTS() to preserve max disjointness."
            ]
        }
    },
    "sample": {
        "sample from": {
            "sentences": [
                "We then sample from this grammar given a large amount of tables to generate new synthetic examples.",
                "We randomly sampled 50 news abstracts from the test set and asked three native-speaker annotators for evaluation to score the generated headlines.",
                "Then an index is randomly sampled from q dimensions with its value perturbed.",
                "We first sample uniformly 80k entities from the graph.",
                "In experiments, we evaluate Meta-KGR on two datasets sampled from Freebase and NELL"
            ]
        },
        "sample at all": {
            "sentences": [
                "In the extremecase where only a few values ever appear in the stream, these samplingalgorithms may not even need to sample at all, and the entire streamcan be summarized via a set of (tuple, count) pairs."
            ]
        },
        "sample as well": {
            "sentences": [
                "The evolution starts from the sampled as well as high-quality programs from the previous mea-initial generation."
            ]
        },
        "sample with": {
            "sentences": [
                "we discover that static thresholding is essential to sampling with large guidance weights and prevents generation of blank images."
            ]
        },
        "sample on": {
            "sentences": [
                "For the first challenge, the performance model module estimates the overall performance with the help of the online profiling module by sampling on a small number of batches"
            ]
        },
        "samplefrom": {
            "sentences": [
                "we sample from the column for performance consideration"
            ]
        },
        "be sample from": {
            "sentences": [
                "The \u201ckeys\u201d (tokens) in the dictionary are sampled from data (e.g., images or patches) and are represented by an encoder network"
            ]
        }
    },
    "sanitize": {
        "sanitize into": {
            "sentences": [
                "So we first sanitize them into disjoint intervals."
            ]
        }
    },
    "satisfy": {
        "satisfy sth": {
            "sentences": [
                "It must also remove the updates of incomplete transactions to satisfy the failure atomicity constraint."
            ]
        }
    },
    "saturate": {
        "be saturate as": {
            "sentences": [
                "We use theoretical minimums derived assuming memory bandwidth is saturated as a baseline."
            ]
        },
        "saturate at": {
            "sentences": [
                "However, as the client rate increases, the system with ARIES logging saturates at around the input rate of 20,000 transactions per second for the TPC-C benchmark and around 60,000 transactions per second for the Smallbank benchmark"
            ]
        }
    },
    "scale": {
        "scale up": {
            "sentences": [
                "We show that our method can scale up to large scale knowledge graphs",
                "Previous work has proved that symbolic execution is still difficult to scale up to large applications.",
                "In these tasks, our pre-training achieves better results than its supervised pre-training counterparts, and more importantly, we observe significant gains by scaling up models"
            ]
        },
        "scale with": {
            "sentences": [
                "First, we ran experiments to ensure that our join implementations are competitive to related work , to check how well they scale with the number of threads and in a NUMA system",
                "recovery time scales with thread count, instead of data size."
            ]
        },
        "scale quickly up": {
            "sentences": [
                "\"The opportunity is then to automatically scale quickly up and down in response to load in order to save money"
            ]
        },
        "scale down rapidly": {
            "sentences": [
                "\"The cloud piece needs to both scale down rapidly as well as scale up"
            ]
        },
        "scale over only": {
            "sentences": [
                "\"Other \u201celastic\u201d services require explicit capacity provisioning"
            ]
        },
        "scale down": {
            "sentences": [
                "To reduce power consumption of a node before turning it off, proposals use power management features in modern nodes (DVFS) to scale down both the frequency of the CPU and the voltage used."
            ]
        },
        "scale as": {
            "sentences": [
                "this sharding enables both memory storage capacity and query throughput to scale as servers are added to the system."
            ]
        },
        "scale to more": {
            "sentences": [
                "to study how well maml can scale to more complex deep rl problems, we also study adaptation on high-dimensional locomotion tasks with the mujoco simu- lator   (todorov et al., 2012)."
            ]
        },
        "scale up properly": {
            "sentences": [
                "We can still push sampling below the join in the query plan, as long as we compute our HTestimator properly, that is, as long as we scale up properly."
            ]
        },
        "scale to": {
            "sentences": [
                "In sum, GPU co-processing does not scale to large data volumes.",
                "we show that this approach makes it possible to scale to large inputs while still retaining model expressivity.",
                "Once the sensitivity of the chosen estimator is bounded, we may add Gaussian noise scaled to this sensitivity to obtain a privacy guarantee",
                "It was shown to be effective for clas- sification with small datasets, but according to [45], it does not scale to larger datasets.",
                "To address computational limitation and scale to large data."
            ]
        },
        "scale out": {
            "sentences": [
                "The capacity and throughput of a storage cluster can be easily scaled out transparently.",
                "In order to scale out the performance continuously when the number of nodes increases, it is becoming crucial to provision the graph processing systems with the ability to handle failures effectively"
            ]
        },
        "scale quadratically": {
            "sentences": [
                "transformer models in particular, which the latest lms build on, would suffer from memory limitations with the naive addition of high-dimensional visual data into the sequence \u2013 computation scales quadratically with the sequence length (number of tokens), for example."
            ]
        },
        "scale up to": {
            "sentences": [
                "Our experiments found that Meerkat is able to scale up to 80 hyper-threads."
            ]
        },
        "scale  linearly": {
            "sentences": [
                "In this experiment, we report that throughput in Aurora can scale  linearly with instance sizes, and with the highest instance size can  Table 2: SysBench Write-Only (writes/sec)  be 5x that of MySQL 5.6 and MySQL 5.7."
            ]
        },
        "scale so well": {
            "sentences": [
                "This approach is feasible in Aurora since  concurrency control and recovery from a Data Component (DC)  the underlying storage system scales so well."
            ]
        }
    },
    "scan": {
        "scan forward": {
            "sentences": [
                "A range query first locates the position of the start key, then scans forward until it reaches the end key."
            ]
        },
        "scan sth": {
            "sentences": [
                "To limit the amount of work performed while determining the visibility of tuples, the DBMS\u2019s garbage collector thread periodically scans the database to undo the dirty modifications associated with the currently present gaps."
            ]
        },
        "scan  periodically": {
            "sentences": [
                "Nodes that receive hinted replicas will keep them in a  Subsequently, using the tree traversal scheme described above the  separate  local  database  that  is  scanned  periodically."
            ]
        }
    },
    "schedule": {
        "schedule conflict transactions": {
            "sentences": [
                "Intuitively, for a batch of transactions, we schedule conflicting transactions to be executed on the same thread"
            ]
        }
    },
    "score": {
        "score as": {
            "sentences": [
                "these responses are scored as 100, 50, and 0, respectively."
            ]
        }
    },
    "screen": {
        "screen out": {
            "sentences": [
                "The experiment began by defining the concepts of data and semantic type, then screened out participants unable to select a specified semantic type."
            ]
        }
    },
    "search": {
        "search for": {
            "sentences": [
                "The loop detection module searches for loop candidate keyframes",
                "Using a decision tree induction method, it searches for the subdomain that implies the largest lower bound and returns that as the label frequency estimate.",
                "it decreases the lookup performance heavily since it has to search for a given key in all components with different sizes.",
                "spatial data structures are designed in a way to achieve high performance results when it comes to multidimensional range queries such as searching for all cities within a given distance of a specific point.",
                "Finally, with an extensive series of microbenchmarks we searched for ideal range values of workload properties that emphasize performance advantages of the radix join over the non-partitioned hash join.",
                "To search for a key",
                "Additionally, the anno- tation tool features a search box above the policy, which enables annotators to search for key terms or phrases within the policy be- fore selecting an answer."
            ]
        },
        "search more efficiently": {
            "sentences": [
                "Finding high-performance programs requires the search based approaches, Ansor searches more efficiently, generating strategy to cover a comprehensive space and explore it effi-higher-performance programs in a shorter time, despite its ciently."
            ]
        }
    },
    "secure": {
        "secure by": {
            "sentences": [
                "All the transactions in a Blockchain network are secured by strong cryptography"
            ]
        }
    },
    "see": {
        "see as": {
            "sentences": [
                "The search process over G can be seen as a Markov Decision Process (MDP)"
            ]
        },
        "see in": {
            "sentences": [
                "furthermore, through imagen we re-emphasize the importance of classifier-free guidance, and we introduce dynamic thresholding, which allows usage of much higher guidance weights than seen in previous works.",
                "as can be seen in figure 3, the module is formed as a u-shaped segmentation structure, which contains two down-sampling blocks and two upsampling blocks with skip connections."
            ]
        },
        "see from": {
            "sentences": [
                "as we can see from the left panel of figure 2, the matrix is quite sparse, especially the lower triangular region."
            ]
        },
        "be see in": {
            "sentences": [
                "Improvement is further seen in how images are interpreted."
            ]
        },
        "be see as": {
            "sentences": [
                "thus can be seen as an Alternating Minimization algorithm in a distributed setting."
            ]
        }
    },
    "seek": {
        "seek for": {
            "sentences": [
                "However, the aforementioned algorithms are not applicable to the PU learning problem studied in this paper, and this motivates us to seek for a novel online training strategy for tackling the OPU learning problem",
                "Meta-learning seeks for the ability of learning quickly from only a few instances within the same concept"
            ]
        },
        "seek to": {
            "sentences": [
                "Given the features (f \u2208 F) in the latent space F and their corresponding labels l \u2208 L, we seek to learn an en- ergy function E(F, L)",
                "Coming from a perspective of natural language understanding, a Recursive Grounding Tree (Hong et al., 2019) sought to automatically compose a binary tree structure by parsing the referring expression, in order to perform visual reasoning along the tree in a bottom-up fashion.",
                "The paper seeks to justify the remedial measures in terms of the final safety analysis report(FSAR).",
                "The relation prediction task can also be viewed as a logical induction problem, where one seeks to derive probabilistic logical rules underlying a given KG",
                "we seek to answer the question.",
                "To close this gap, we seek to learn contextual representations jointly from structured tabular data and unstructured natural language sentences, with objectives oriented towards table semantic parsing.",
                "The vision-and-language pretraining seeks to jointly learn representations using both visual and textual content for improving the efficiency of previously discussed vision and language integration tasks.",
                "this technique actually seeks to minimize the number of bits that differ between the crashing input and the original seed file and is not related to reducing input size."
            ]
        },
        "seek to empirically": {
            "sentences": [
                "now, we seek to empirically explore this theoretical \ufb01nding, aiming to answer the ques- tion: is there a scenario for which model-agnostic meta-learning requires a deeper rep- resentation to achieve good performance, compared to the depth of the representation needed to solve the underlying tasks being learned?"
            ]
        },
        "have seek to": {
            "sentences": [
                "While several works have sought to extend this success to video-based tasks like action recognition by aligning appearance (e.g., RGB) features through adversarial learning, challenges persist in video adaptation tasks due to the greater complexity of the video data."
            ]
        }
    },
    "seem": {
        "seem to": {
            "sentences": [
                "Predicting normal maps also seems to generalize better between datasets as depth does.",
                "This seems to be a lower bound which limits all individuals, even those with high probability of finding a path to their recipient."
            ]
        },
        "seem innocuous": {
            "sentences": [
                "such an assumption may seem innocu- ous  at  \ufb01rst  glance,  but  in  practice,  it  requires  for  a  great  deal  of  human  engineering."
            ]
        }
    },
    "select": {
        "select from": {
            "sentences": [
                "Matches with high confidence scores can be selected from the dense matches",
                "synonyms pairs and direct hyponymhypernym pairs are selected from WordNet",
                "In this way, we randomly select q channels (c1, c2, ..., cq ) from the federated model.",
                "The labeled positive examples are selected from the complete set of positive examples according to a probabilistic labeling mechanism.",
                "In cluster sampling, the population is first divided into k clusters and the elements sampled are mi items within the clusters that are randomly selected from all clusters [49,9].",
                "The locations of the spatial part are randomly selected from a unit sphere."
            ]
        },
        "select the top": {
            "sentences": [
                "Only means directly selecting the top-1 retrieved sentence as the caption."
            ]
        },
        "select the most": {
            "sentences": [
                "Multi-faceted attention has also been used to select the most salient visual features or semantic attributes, with which an overall sentence is generated (Long et al., 2018)."
            ]
        },
        "select between": {
            "sentences": [
                "to probe image quality, the rater is asked to select between the model generation and reference image using the question: \u201cwhich image is more photorealistic (looks more real)?\u201d."
            ]
        },
        "select for": {
            "sentences": [
                "The state of the network is changed only by the transactions which are selected for inclusion in the blocks"
            ]
        }
    },
    "send": {
        "send to": {
            "sentences": [
                "Second, it sends the batches to the corresponding operators with proper hardware resources for execution.",
                "When a transaction is sent to an EOA",
                "the server picks the value for its ballot and sends it in an Accept RPC to every server",
                "Advertisement platforms might need to send different categories of ads to different groups of customers."
            ]
        },
        "send by": {
            "sentences": [
                "Transactions are the messages which are sent by Externally Owned Accounts (EOAs) to other EOAs or contract accounts"
            ]
        },
        "send out": {
            "sentences": [
                "In all following settings, we aim to ensure that all the information sent out by one party satisfies DP relative to its user\u2019s data"
            ]
        }
    },
    "sented": {
        "sented with fully": {
            "sentences": [
                "the  input  feature  extractor  and  readout  function  can  be  repre out sented with fully connected neural networks with one or more hidden layers, which we 140 know are universal function approximators, while qn w corresponds to a set of linear layers."
            ]
        }
    },
    "separate": {
        "separate in": {
            "sentences": [
                "PGM [37] uses linear models and separates the keys in different linear segments with given error bound."
            ]
        },
        "separate into": {
            "sentences": [
                "The distributed localization methods can separate into three kinds.",
                "As the first formal analysis against NC, Garay et al. [32] separated NC into (1) its backbone protocol.",
                "ML can separate into three kinds according to the types of data processed."
            ]
        },
        "separate\u2026into\u2026": {
            "sentences": [
                "ConvPoint [66] separates the convolution kernel into spatial and feature parts."
            ]
        },
        "be separate from": {
            "sentences": [
                "Moreover, conventional TDS is separate from model training."
            ]
        },
        "separate the execution": {
            "sentences": [
                "It separates the execution of single-partition transactions and cross-partition transactions using a novel phase-switching protocol"
            ]
        },
        "separate with": {
            "sentences": [
                "These systems separate the query engine and the encoded data file with a decoding layer"
            ]
        },
        "separate from": {
            "sentences": [
                "Mencius separates the execution of a log entry from its commit.",
                "All MVCC DBMSs keep the database's versioning information separate from its indexes."
            ]
        },
        "separate sth": {
            "sentences": [
                "It separates the execution of single-partition transactions and cross-partition transactions using a novel phase-switching protocol"
            ]
        },
        "separate out": {
            "sentences": [
                "Both methods dominate previous approaches empirically, and for BBE, we establish formal guarantees that hold whenever we can train a model to cleanly separate out a small subset of positive examples."
            ]
        }
    },
    "serialize": {
        "serialize into": {
            "sentences": [
                "BRIDGE [21] serializes the question and schema into a tagged sequence and maximally utilizes BERT [12] and the database content to capture the question-schema linking."
            ]
        }
    },
    "serve": {
        "serve as": {
            "sentences": [
                "The front end serves as a real-time VO",
                "It serves as an implicit acknowledgement.",
                "The feature correlation layer serves as a key neural network module in numerous computer vision problems that involve dense correspondences between image pairs.",
                "The PFE module serves as the first step in the proposed framework.",
                "The contrastive loss serves as an unsupervised objective function for training the encoder networks that represent the queries and keys",
                "\ufeffHere, topic models could serve as an additional source for dataset-specific information.",
                "The front end serves as an odometry",
                "The distribution of p serves as a proxy for conformity distribution.",
                "These implicit connections between cell entries along rows and columns of a table can serve as important context for prediction methods for the two tasks.",
                "The class activation map originally serves as a spatial visualization tool.",
                "These phrase selections serve as supporting evidence for provided annotations.",
                "in other words, all regular tokens are not explicitly connected to other regions, and msg tokens serve as the hub of information exchange.",
                "This kind of reasoning machine will potentially serve as an essential components of complex QA systems",
                "The results serve as a baseline for further development of blockchain technologies.",
                "the popular ORB-SLAM2, which also serves as our baseline",
                "each column chunk consists of data pages that serve as a unit of encoding compression",
                "We\nfind empirically that the background score produced by the teacher model with weak augmented image can well serve as a proxy indicator of r and is easily obtained in our end-to-end training framework",
                "this paper aims to alleviate the conflict between efficiency and flexibility, for which we propose a specialized token for each region that serves as a messenger (msg).",
                "The primary contribution of this paper is to serve as a guide to implementors as to what sorts of performance differences one should expect to observe in database implemen- tations on modern versions of these different architectures."
            ]
        },
        "be serve as": {
            "sentences": [
                "After the learning process, such methods are usually served as a pre-training step for the downstream tasks."
            ]
        },
        "serve the purpose": {
            "sentences": [
                "Afterwards, we look at some curves which solely serve the purpose of gaining more insights into sampling and errors produced"
            ]
        },
        "serve depend heavily": {
            "sentences": [
                "Individual nodes be served depends heavily on users interaction with the webservice, query sizes vary."
            ]
        }
    },
    "set": {
        "set to": {
            "sentences": [
                "Also, the aim has been set to implement the proposed idea for large scale software protection and improvement.",
                "The IoU threshold is set to 0.3 as overlap between the prediction and the ground truth at this level is clinically sufficient",
                "grayscale image) is set to \u20191\u2019 where the stroke of Chinese character passes; otherwise, the position is set to \u20190\u2019",
                "In addition, we use 51/5/11 task relations for training/validation/testing in NELL and the division is set to 133/16/34 in Wiki",
                "We set P (\u27e8select\u27e9 \u2192 the count) to 0.2 in all our experiments."
            ]
        },
        "set up": {
            "sentences": [
                "Such kind of methods can also be effective when the cost is appropriately set up but most of the time, setting up the cost accurately is not possible due to lacking of domain knowledge.",
                "First, we set up an annotation task to collect dialogues with question-answer pairs along with question rewrites and answer provenance links.",
                "In the experiment, we set up a simple chain of two kernels, producer and consumer.",
                "Existing benchmark datasets for knowledge graph completion are set up for transductive reasoning",
                "An experienced analyst can easily intercept the traffic by setting up a proxy."
            ]
        },
        "set via": {
            "sentences": [
                "we set S via simulation"
            ]
        },
        "set out": {
            "sentences": [
                "He set out on a round-the-world trip in a small boat."
            ]
        },
        "set out on": {
            "sentences": [
                "We set out on a real-world experiment in a small boat."
            ]
        },
        "set by": {
            "sentences": [
                "Using the pre-training tasks, we set the new state-of-the-art on TABFACT out-performing previous models by 6 points when using a BERT-base model and 9 points for a BERT-large model."
            ]
        },
        "set with": {
            "sentences": [
                "In addition, the val2017 set with 5k images is also provided for validation"
            ]
        },
        "set of": {
            "sentences": [
                "table 2: results (%) on the development and test set of docred."
            ]
        },
        "set aside": {
            "sentences": [
                "They formulate this task as a multi-class classification problem (with 78 classes), with a training set of 64,000 tables and a test set consisting of 16,000 tables."
            ]
        },
        "be set to": {
            "sentences": [
                "The size of a chunk is set to 10 GB in PolarFS",
                "\ufeffThe threshold is dynamically set to the average similarity score across questions and we output only those which have a similarity greater than the average similarity score."
            ]
        },
        "set as": {
            "sentences": [
                "We set comparison methods as recommended and the batch size is 32."
            ]
        },
        "set out to do": {
            "sentences": [
                "In this paper, we set out to understand the true nature of performance difference between CPUs and GPUs."
            ]
        },
        "set pose": {
            "sentences": [
                "although similar in spirit to large language models, the multimodality of our setting poses a number of challenges not present in the language-only domain."
            ]
        },
        "set out to": {
            "sentences": [
                "We set out to design a weakly-supervised loss for probabilistic mappings."
            ]
        }
    },
    "set out": {
        "set out to": {
            "sentences": [
                "we set out to develop a dense matching module capable of effectively utilizing the aforementioned information when predicting the correspondence volume"
            ]
        }
    },
    "setup": {
        "setup on": {
            "sentences": [
                "Our demonstration setup on a table on which a screen, a large dial and four identical benchmark computers are laid out."
            ]
        }
    },
    "share": {
        "share for": {
            "sentences": [
                "we share an accelerator for every 16 records in the PA"
            ]
        },
        "share storage mightactasprimarynodeswritingshared\ufb01lesconcurrently": {
            "sentences": [
                "[5] as the shared storage mightactasprimarynodeswritingshared\ufb01lesconcurrently."
            ]
        },
        "share xxx between": {
            "sentences": [
                "In our model, we share the same weight matrix between the two embedding layers and the pre-softmax linear transformation, similar to [30]."
            ]
        },
        "share by": {
            "sentences": [
                "In the original design, Bitcoin\u2019s blockchain stores coins as the system states shared by all participants",
                "it consists of an adaptive encoder that takes either text or visual sketch as input, and a decoder shared by 8 visual synthesis tasks.",
                "tagging is shared by all relations because it focuses on the general entity extraction without considering the specific relation types."
            ]
        },
        "share across everywhere": {
            "sentences": [
                "The word embedding size is also confirms that stories generated by CVAE provide best 300 and shared across everywhere."
            ]
        },
        "share with": {
            "sentences": [
                "however their main characteristic is that they are centralised, not scalable as the number of users and items increase and require users to share their personal data with the server.",
                "But also opens an entirely new spectrum of privacy risks when sharing information with fellow nodes",
                "For example, in one section a pol- icy may claim that data is not shared with third parties"
            ]
        }
    },
    "sharpen": {
        "sharpen the reader\u00a1\u00afs intuition": {
            "sentences": [
                "Firstly, we want to sharpen the reader\u00a1\u00afs intuition about q-errors produced by sampling."
            ]
        }
    },
    "shed": {
        "shed light on": {
            "sentences": [
                "This sheds light on an elegant way to reduce the domain gap.",
                "This paper presents a summary of the recent advances, analyzes how they improve the fuzzing process, and sheds light on future work in fuzzing."
            ]
        }
    },
    "shift": {
        "shift to": {
            "sentences": [
                "In this paper, we first make the case that in modern RDMAenabled networks, the bottleneck has shifted to CPUs, and therefore the existing network-optimized replication techniques are no longer optimal",
                "Most data points have shifted to the right, such that colour becomes a very good predictor of how easily nodes can be reached from the rest of the network.",
                "ALEX also shifts the elements to make the gap for the inserted key when there is a key in that position."
            ]
        },
        "shift into": {
            "sentences": [
                "The above approach shifts the task of finding offsets into the output array to an optimized prefix sum kernel whose runtime is a function of T."
            ]
        },
        "shift from": {
            "sentences": [
                "Unlike the pure local transaction processing as shown in Figure 5a and Figure 6a, the main bottleneck of distributed pro\u0002cessing gradually shifts from logging to network communication"
            ]
        }
    },
    "ship": {
        "ship with": {
            "sentences": [
                "Execution accuracy compares the results of executing the predicted and ground-truth SQL queries on the database contents shipped with the Spider dataset."
            ]
        },
        "be ship from": {
            "sentences": [
                "For query execution, data is shipped from the CPU to the GPU over PCIe."
            ]
        }
    },
    "shorten": {
        "be shorten by": {
            "sentences": [
                "Moreover, the latency of multi-replica concurrent write can also be shortened by eliminate unnecessary serial restrictions in ParallelRaft for PolarFS"
            ]
        }
    },
    "show": {
        "show to": {
            "sentences": [
                "These methods have been shown to scale excellently [4] and a large abundance of evidence indicates that these pre-trained representations generalize well to various downstream tasks",
                "The use of specialized data engines has been shown to more e\ufb03cient execution of OLTP transactions.",
                "which has shown to bring large improvement on a range of NLP tasks"
            ]
        },
        "show that currently": {
            "sentences": [
                "These examples show that currently existing automatic scalability mecha- nismscanbeimproved."
            ]
        },
        "show the most": {
            "sentences": [
                "Table 4 shows the most popular uses of serverless computing today."
            ]
        },
        "as show in": {
            "sentences": [
                "Existing XSP evaluation data such as Spider simplifies some of these challenges, for example by including utterances that closely match their paired SQL query, as shown in last row of Figure 1."
            ]
        },
        "show in": {
            "sentences": [
                "as it shows in the figure 2, the segmentation area in the entitylevel relation matrix refers to the co-occurrence of relations between entity pairs.",
                "for example, as shown in figure 1, it is easy to identify the intrasentence relations, such as (maryland, country, u.s.), (baltimore, located in, maryland), and (eldersburg, located in, maryland), owing to the occurrence of entities in the same sentence.",
                "the architectural components shown in figure 3 are key to the performance of flamingo models, and these are chosen according to two objectives."
            ]
        },
        "be show to": {
            "sentences": [
                "Such an approach has been shown to effectively inspect local regions and pairwise interactions between them."
            ]
        },
        "be show in": {
            "sentences": [
                "There has been interest shown in jointly addressing multiple tasks that handle video and language."
            ]
        },
        "show more relevant sequence": {
            "sentences": [
                "Our method clearly shows more semantically relevant and coherent sentence sequences in most cases."
            ]
        },
        "show that jointly": {
            "sentences": [
                "As a result, TASO can automatically generate graph substitutions that optimize both graph structures and data layouts,and our evaluation shows that jointly optimizing the twotasks can significantly improve the end-to-end performance,compared to optimizing the them separately."
            ]
        },
        "show promise": {
            "sentences": [
                "in experiments, msg-transformers show promising results on both imagenet classification, i.e., 84.0% top-1 accuracy, and ms-coco object detection, i.e., 52.8 map, which consistently outperforms recent state-of-the-art swin transformer."
            ]
        },
        "show surprisingly": {
            "sentences": [
                "furthermore, it also shows surprisingly good zero-shot capabilities on text-guided image and video manipulation tasks."
            ]
        },
        "show that": {
            "sentences": [
                "These experiments show that WBL with NVM improves the DBMS\u2019s throughput by 1.3\u00d7 while also reducing the database recovery time and the overall system\u2019s storage footprint."
            ]
        },
        "be show to be": {
            "sentences": [
                "Contrastive representation learning has shown to be effective to learn represen- tations from unlabeled data"
            ]
        },
        "show up": {
            "sentences": [
                "Such corner cases were not part of Google's or Tesla's test set and thus never showed up during testing."
            ]
        }
    },
    "shuffle": {
        "be shuffle to": {
            "sentences": [
                "In H-Store and its commercial successor, VoltDB1, command logs of different nodes are shuffled to the master node that merges them based on the timestamp order"
            ]
        },
        "shuffle ...to": {
            "sentences": [
                "We also randomly shuffle these four datasets to simulate the real-world cenarios"
            ]
        }
    },
    "shuf\ufb02ing": {
        "shuf\ufb02ing start as": {
            "sentences": [
                "This shuf\ufb02ing starts as soon as the \ufb01rst derlying\ufb01lesystemusederasurecoding[14]ratherthan map task completes."
            ]
        }
    },
    "sign": {
        "sign by": {
            "sentences": [
                "Transactions are signed by the sender's private key"
            ]
        }
    },
    "siloed": {
        "siloed within very": {
            "sentences": [
                "these existing systems are siloed within very narrow environments, often for the sake of practicality but at the expense of narrow experiences that are insuf\ufb01cient for developing common sense."
            ]
        }
    },
    "similar": {
        "similar with": {
            "sentences": [
                "Similar with state-of-the-art learned index structures"
            ]
        },
        "be similar to": {
            "sentences": [
                "The network is modeled as a diffusion functionality similar to thosein [13, 20]"
            ]
        }
    },
    "simple": {
        "be simple to": {
            "sentences": [
                "This kind of systems are relatively simple to implement"
            ]
        }
    },
    "simulate": {
        "simulate two scenarios": {
            "sentences": [
                "We simulate two scenarios of failures"
            ]
        },
        "simulate the  senario": {
            "sentences": [
                "we simulate a low-resource data scenario"
            ]
        },
        "simulatesth": {
            "sentences": [
                "We simulate two scenarios of failures"
            ]
        }
    },
    "sit": {
        "sit in": {
            "sentences": [
                "Sitting in the core of the Operating System (OS), the kernel is the most important piece of software in various types of computing devices."
            ]
        }
    },
    "size": {
        "size large enough": {
            "sentences": [
                "Onereasonforthisisthatthe in ETC (approximately 30 : 1) is signi\ufb01cantly higher than pool is sized large enough to maximize hit rates, so most synthetic workloads typically assume (Sec. 7).",
                "Indeed, note that the from Sec. 2.2 that USR is sized large enough to minimize diurnaldips correspond to increases in hit rates in Fig."
            ]
        }
    },
    "skip": {
        "skip over": {
            "sentences": [
                "Skip over the next C data records.",
                "This way, it can skip over unnecessary startup code."
            ]
        },
        "skip do sth": {
            "sentences": [
                "During the redo phase, the DBMS skips replaying the log records associated with uncommitted transactions."
            ]
        }
    },
    "slow": {
        "slow down": {
            "sentences": [
                "Therefore, Bitcoin slows down its transaction propagation to protect the users\u2019 privacy [6].",
                "As sparsification can slow down the convergence speed of learning algorithms and increase the total number of communication rounds, we propose to further reduce the end-to-end privacy loss by using convergence acceleration techniques to offset the negative impact of sparsification.",
                "First, maintaining the consistency among replicas incurs a high synchronization overhead and as a consequence slows down the transaction processing",
                "generating region features slows down the inference speed and renders the approach less scalable.",
                "For the high-dimensional case (d>n), random sampling a small fraction \u03b2 of values from a vector slows down the convergence rate of training"
            ]
        }
    },
    "snippet": {
        "snippet with": {
            "sentences": [
                "Basic blocks are code snippets with a single entry and exit point, instructions in basic blocks will be sequentially executed and will only be executed once."
            ]
        }
    },
    "solve": {
        "solve challenge": {
            "sentences": [
                "Efficiently solving the above-mentioned and related challenges can result in many potential applications."
            ]
        },
        "solve them sequentially": {
            "sentences": [
                "treating data layout and graph substitution as separate optimization problems and solve them sequentially [8, 27], asshown in Figure 1a, but this separation misses many possibleoptimization opportunities."
            ]
        }
    },
    "sort": {
        "sort by": {
            "sentences": [
                "Suppose all values are stored in an array sorted by key order"
            ]
        },
        "sort out": {
            "sentences": [
                "Guided by the taxonomy, we conduct an extensive empirical study to reveal the influence of the key modules in SSOD and consequently sort out the good practices for the new task"
            ]
        },
        "sort merge": {
            "sentences": [
                "Over the years, the community has reached the conclusion that hash joins are better than sort-merge joins"
            ]
        },
        "sort to": {
            "sentences": [
                "by using these features, we can train a ranker to sort the sources to unknown targets by predicting their value."
            ]
        },
        "sort the nod": {
            "sentences": [
                "it sorts the nodes according to a policy P, and includes the k top-ranked nodes in the FVS"
            ]
        }
    },
    "sound": {
        "sound appeal to": {
            "sentences": [
                "Every good article needs a good title, which should not only be able to condense the core meaning of the text, but also sound appealing to the read- ers for more exposure and memorableness."
            ]
        }
    },
    "source": {
        "source to": {
            "sentences": [
                "particular source to every possible subset of the sources."
            ]
        }
    },
    "span": {
        "span sth": {
            "sentences": [
                "For long running transactions that span a group commit window, the DBMS also records their commit timestamps in the log."
            ]
        }
    },
    "speak": {
        "speak by": {
            "sentences": [
                "speaking by randomly moving lips), but fail on text prompt such as \u201ca lion is drinking water\u201d."
            ]
        },
        "speak with": {
            "sentences": [
                "one S&P 500 employee we spoke with disclosed that their monthly cloud costs for storing Parquet files are in the sixfigure range"
            ]
        }
    },
    "specialize": {
        "specialize in": {
            "sentences": [
                "graph analytics systems such as pregel and powergraph specialize in olap-like batch-processing of graph-structured data on large computing clusters."
            ]
        },
        "specialize to just": {
            "sentences": [
                "For example, Winograd convolution [30]) and accelerator intrinsics (e.g., the above 10-level loop can be specialized to just a simple TensorCore [37]) require special tile structures to be effec-reorder (k0, j2,i3) by setting the length of other loops to one."
            ]
        }
    },
    "specify": {
        "specify use": {
            "sentences": [
                "the augmentation level is specified using aug_level \u2208 [0, 1]."
            ]
        }
    },
    "speed": {
        "speed up": {
            "sentences": [
                "Consequently, we can only speed up data processing on GPUs by increasing the inter-connect bandwidth.",
                "indexes in data management systems can speed up data retrieval at the cost of index maintenance and storage overhead which can significantl",
                "To improve the privacy, we speed up the convergence by updating the model in an adaptive manner similar as Adam on the server.",
                "Operators, such as aggregation, can use this information to speed up execution",
                "Such a small decoder can further speed up training",
                "While the distributed version of command logging effectively exploits the parallelism among the nodes to speed up recovery",
                "The inference time on the ModelNet and ShapeNet datasets is sped up by 1.5 ms and 32 times over PointNet on a moderate machine.",
                "This recall progress can be speeded up by some search algorithms such as ANN search",
                "In addition, memory is greatly reduced, which can enable training even larger models or speeding up more by large-batch training",
                "Our RJ implementation speeds up by a factor of 7.5 to 9.5 for workloads A and B, respectively.",
                "Indexes in data management systems can speed up data retrieval at the cost of index maintenance and storage over\u0002head"
            ]
        }
    },
    "spend": {
        "be spend": {
            "sentences": [
                "One study reported that 16-25% of transaction time is spent interacting with the lock manager in a main memory DBMS [8]."
            ]
        },
        "spend on": {
            "sentences": [
                "Fundamentally, every scheduling algorithm confronts the same exploration vs. exploitation conflict-time can either be spent on gathering more accurate information on each configuration to inform future decisions (explore), or on fuzzing the configurations that are currently believed to lead to more favorable outcomes (exploit)."
            ]
        },
        "spend to": {
            "sentences": [
                "The x-axis shows the time spent to highlight how expensive each phase of the join is."
            ]
        },
        "spend join sub-": {
            "sentences": [
                "The time spent joining sub- belong to different records) with minimal impact on the relations increases with the number of attributes in the intra-record spatial locality."
            ]
        }
    },
    "spill": {
        "spill to": {
            "sentences": [
                "Higher interconnect performance requires us to reconsider how well a NP-HJ that spills to CPU memory performs on GPUs."
            ]
        }
    },
    "split": {
        "split to": {
            "sentences": [
                "So the input is split to 14 \u00d7 14 image patches, and the same amount of visual tokens"
            ]
        },
        "split into": {
            "sentences": [
                "Observational interaction data O, which is splitted into Otrain and Otest , with non-IID conditions on popularity distribution D1 and D2",
                "We recursively split tables row-wise into the upper and lower half until they have at most 50 cells.",
                "A common approach to crowdsourcing is to split a complex task into smaller subtasks that are easier to solve",
                "We split the collected tables into training, validation and test datasets.",
                "Retrieving records (b) is split into two groups. FASTER, pmemkv, and RocksDB do not perform well for random get request due to inefficient lookups in PMem, all peaking below 4 Mops/s.",
                "First, it splits the stream data into batches with a fixed size",
                "the model may split into multiple words",
                "split the features maps into regions and perform local self-attention within the region",
                "After the encoder, the architecture splits into two decoder \u201cheads\u201d,which learn task specific weights \u2013 one for interest point detection and the other for interest point description.",
                "Split the network into several patches and independent points.",
                "We split the datasets into s disjoint sub-datasets by movie/profiles, such that the ratings of each movie/profile can appear in only one dataset",
                "if a \u201cdrinking\u201d video is split into four individual clips of \u201cholding a glass\u201d, \u201clifting\u201d, \u201cdrinking\u201d and \u201cputting down\u201d with the same text \u201cdrinking\u201d, the model will be confused to learn the accurate meaning of drinking."
            ]
        },
        "split up": {
            "sentences": [
                "Variables of restricted range can be split up into two or more variables."
            ]
        },
        "split up evenly": {
            "sentences": [
                "for  example,  the  mnist  (lecun  et  al.,  1998)  and  cifar- 10 (krizhevsky and g. hinton, 2009) datasets both have 60,000 images, split up evenly within  10 classes."
            ]
        },
        "be split into": {
            "sentences": [
                "When copying the chunks, a whole chunk is split into small 128KB pieces and using lots of small tasks",
                "It is split into the train set (8659 examples), development set (1034 examples) and test set (2147 examples), which are respectively distributed across 146, 20 and 40 database",
                "Here, we assume that the database is split into two partitions"
            ]
        },
        "split from": {
            "sentences": [
                "All images which correspond to the above split from Pascal VOC and MS-COCO train-sets form the training data"
            ]
        },
        "split sth. into sth.": {
            "sentences": [
                "Before we setup the pipeline, we split the data into chunks."
            ]
        },
        "split as": {
            "sentences": [
                "Operators also access their input data in parallel by splitting the input as multiple data blocks"
            ]
        },
        "split in": {
            "sentences": [
                "We could split the errors in two main categories"
            ]
        }
    },
    "spread": {
        "spread to": {
            "sentences": [
                "afterwards the epidemic quickly spread to other communities and there were insert queries that recorded new potentially-infected people (w2).",
                "Afterwards, the epidemic quickly spread to other communities"
            ]
        },
        "spread across": {
            "sentences": [
                "With WBL, the DBMS writes out the changes to locations spread across the durable storage device."
            ]
        },
        "spread out": {
            "sentences": [
                "Our results show that crowdworkers can provide highly accurate privacy policy annotations for some questions, primarily concern- ing collection and deletion, but that they struggle with questions pertaining to sharing practices, which are typically more spread out in the policy."
            ]
        }
    },
    "squeeze": {
        "squeeze sth": {
            "sentences": [
                "For these common application sce\u0002narios, squeezing more throughput from the capacity is not the goal"
            ]
        }
    },
    "stack": {
        "stack to": {
            "sentences": [
                "these methods are able to jointly learn feature representations and the prediction function (or forecasting function) by\nstacking a series of non-linear layers to perform feature extraction,\nfollowed by a regression layer focused on forecasting."
            ]
        },
        "stack on": {
            "sentences": [
                "Blockchain software stack on a fully validating node"
            ]
        }
    },
    "stand": {
        "stand for": {
            "sentences": [
                "Iq stands for the mutual information (MI).",
                "G stands for ground truth.",
                "The results stand for the error declining process."
            ]
        }
    },
    "start": {
        "start from": {
            "sentences": [
                "for example, a breadth first search traversal can explore an expansive connected component starting from a single vertex.",
                "Starting from a real image pair (I, J), we consider the image triplet introduced in [49]",
                "training a multimodal vision and language model, starting from a pretrained lm, has the potential to save the immense computational resources used to train the lm in the first place.",
                "Finally, we introduce a new task of story interpolation, which involves predicting curriculum learning which starts from relatively guage descriptions not only for the given images, but also easy task and gradually increases the difficulty of train- for gaps between the images.",
                "on the language side, we start from an existing autoregressive language model (lm) trained on a large and diverse text corpus (hoffmann et al., 2022).",
                "We expect the agent for r_q can start from the source entity and walk through several edges to reach the target entity"
            ]
        },
        "start to": {
            "sentences": [
                "Node 3 starts to synchronize metadata automatically and it only needs to load incremental entries and replay them in local memory.",
                "before the system can start to serve requests",
                "Only a few works have recently started to focus on increasing the attractiveness of generated headlines.",
                "As a person starts to appear more often (e.g. green, yellow or red) then they are increasing their chances of being able to send and receive data from then network.",
                "interest from the industry has started to drive development of new blockchain platforms that are designed for private settings in which participants are authenticated",
                "Third, after the performance model is built, the dispatcher starts to work",
                "Then, DomCC starts to execute the transaction logic."
            ]
        },
        "start at": {
            "sentences": [
                "It means that if the memory region starting at x is an instance of task struct, the layout of the region must follow the graph's definition.",
                "we start at the root node and use its model to predict which child node to go to"
            ]
        },
        "start with": {
            "sentences": [
                "The algorithm starts with k initial model parameters.",
                "In all experiments, we start with the public TAPAS checkpoint,6 train an entailment model on the data from Section 3 and then fine-tune on the end task (TABFACT or SQA).",
                "We start with the ubiquitous WAL protocol in Section 3, followed by our new WBL method in Section 4.",
                "The test starts with the partially partitionable and highly conflicted workload",
                "Each optimistic period starts with the current proposer broadcasting its block.",
                "We start with I = ? x??m] Ix. Stage (0) is the combined inner sampling, taking us from I to I(1) = ? x??m] VAROPTkx (Ix). Stage (1) is the outer subcall taking us from I(1) to the final sample S = VAROPTk (I(1))",
                "We start with our observation and analysis of the cluster assumption in semantic segmentation, motivating the proposal of our cross-consistency training approach",
                "This trend started with [10] for indoor object detection",
                "The queries below illustrate their syntax and semantics, starting with an example of an earliest- arrival path query.",
                "We initially start with an empty set of edges",
                "for relation mayor, (new, de) is tagged as 1 in the sh-to-oh sequence, which means the mayor of the subject starting with new is the object starting with de.",
                "We start with an overview of the whole ac- tive learning system in Section 3.1, and provide in-depth descriptions of the loss prediction module in Section 3.2, and the method to learn this module in Section 3.3."
            ]
        },
        "start by": {
            "sentences": [
                "We start by quantifying how much NVLink 2.0 improves the GPU\u2019s interconnect performance.",
                "For this, we start by designing Blockbench based on three most mature platforms within our consideration",
                "We start by giving an overview of the complete algorithm"
            ]
        },
        "strart to": {
            "sentences": [
                "We also started to implement clustering techniques on the invariant feature space for better classification and showed better performance on the target domain."
            ]
        },
        "start prototypingal- ever": {
            "sentences": [
                "We have already started prototypingal- ever, thesepapersdid not speci\ufb01cally disclose theworkload ternative replacement schemes, and the initial results are characteristics."
            ]
        },
        "starts, then": {
            "sentences": [
                "ternalconsistency(orequivalently,linearizability[HerlihyandWing1990]):ifatrans- action T commits before another transaction T starts, then T \u2019s commit timestamp 1 2 1 is smaller than T \u2019"
            ]
        }
    },
    "stash": {
        "stash to": {
            "sentences": [
                "we do not expect stashing to be the main storage"
            ]
        }
    },
    "state": {
        "state that": {
            "sentences": [
                "The Zero-Coordination Principle states that scalable systems can be built by following two rules."
            ]
        }
    },
    "stay": {
        "stay in": {
            "sentences": [
                "The upper limit ensures most records stay in PA"
            ]
        }
    },
    "stem": {
        "stem from": {
            "sentences": [
                "This relatively good accuracy mostly stems from the counterfactual data.",
                "The reason for such a gap stems from the difficulty induced by adding a third dimension to the estimation problem, the low resolution of 3D input data, and the deterioration of its quality as a function of distance",
                "The reason may stem from differences in their respective signal spaces"
            ]
        }
    },
    "step": {
        "step away": {
            "sentences": [
                "However, neither of them has stepped away from the positive-negative classification problem initially established in standard GAN."
            ]
        }
    },
    "stitch": {
        "stitch as": {
            "sentences": [
                "Intuitively, we can directly stitch macro-variables as features with micro-ones to predictors."
            ]
        }
    },
    "stop": {
        "stop at": {
            "sentences": [
                "The reward will be 1 if the agent \ufb01nally stops at the right entity"
            ]
        },
        "stop train": {
            "sentences": [
                "We set batch size to 64 and stop training when the metric sentence representations."
            ]
        }
    },
    "store": {
        "store in": {
            "sentences": [
                "facts not directly stored in a KB can often be inferred from those",
                "virtuoso-rdf employs the single table with extensive indexing approach, where the entire graph is stored in a single relational table with multi ple indexes.",
                "\"in fact if the keys in s are stored in fully-dynamiccompressedlearnedindexwithprovableworst-case bounds.\"",
                "Notice that the query instances are stored in the forecaster module and are available to the cost module.",
                "Rather than knowledge stored in KG",
                "by doing so, flamingo models gain strong generative language abilities and access to a large amount of knowledge stored in the lm weights."
            ]
        },
        "store as": {
            "sentences": [
                "A KB is a collection of facts stored as triplets",
                "as The above models embed entities and relations by only facts stored as a collection of triplets",
                "store all values as offsets from a reference value"
            ]
        },
        "store to": {
            "sentences": [
                "we assume unique keys, but non-unique keys can be supported by storing a pointer to a linked list of records as payload.",
                "One such instruction stored in the machine code of a program is the program entry point, i.e the location in the machine code where the machine instructions begin."
            ]
        },
        "store into": {
            "sentences": [
                "All conflicting information will be stored into this procedure for later use",
                "store the corresponding indexes into the rules"
            ]
        },
        "store with": {
            "sentences": [
                "For instance, stock price information may be stored with various column names."
            ]
        },
        "store sth. with sth.": {
            "sentences": [
                "We store the data in columnar format with each column represented as an array of 4-byte values."
            ]
        },
        "sufficient to store": {
            "sentences": [
                "Such high memory instances are sufficient to store 10 kilobytes of online state about each customer in a database with about one billion customers"
            ]
        },
        "be store in": {
            "sentences": [
                "In many applications, data are stored in end users\u2019 own devices such as mobile phones and personal computers."
            ]
        },
        "be store on": {
            "sentences": [
                "Since data are all stored on the storage cluster"
            ]
        },
        "store close e(cid:2)ts": {
            "sentences": [
                "onHTMLpagesbecauseofthewayWebtablerowsare Using one log provides signi(cid:2)cant performanceben- laid out: all pages from a single host are stored close e(cid:2)ts during normaloperation, but it complicates recov- to each other."
            ]
        },
        "store it locally": {
            "sentences": [
                "This requires us to design applications  at nodes C and D in addition to storing it locally."
            ]
        }
    },
    "stress": {
        "stress on": {
            "sentences": [
                "Avoid stress the feature computation on one central machine."
            ]
        }
    },
    "strive": {
        "strive for": {
            "sentences": [
                "This paper strives for activity recognition under domain shift, for example caused by change of scenery or camera viewpoint."
            ]
        },
        "strive to": {
            "sentences": [
                "This locking mechanism limits the scalability of protocols that strive to eliminate lock overhead"
            ]
        }
    },
    "structure": {
        "structure around": {
            "sentences": [
                "the former provides a common set of interfaces for the property graph model, and the latter de nes a procedural query language structured around the gremlin structure api."
            ]
        },
        "be structure as": {
            "sentences": [
                "The remainder of the paper is structured as follows."
            ]
        }
    },
    "struggle": {
        "struggle in": {
            "sentences": [
                "MW-based systems struggle in cluttered environments"
            ]
        },
        "struggle to": {
            "sentences": [
                "Both TaBERT (fine-tuned for corrupt cell detection) and TABBIE FREQ struggle to identify this swap.",
                "However, involving heavy crafting on data pre-processing and labor-intensive feature engineering makes those methods struggling to cope with large-scale data and the performance is also limited for more complex patterns.",
                "Existing engines, such as Spark Streaming [57], struggle to support small window and slide sizes",
                "In our analysis, we grouped \u201cunclear\u201d and \u201cnot addressed in the policy\u201d annotations, since crowdworkers struggled to differentiate between these two options."
            ]
        },
        "struggle on": {
            "sentences": [
                "They are struggled on data with strong correlations"
            ]
        }
    },
    "study": {
        "study on": {
            "sentences": [
                "The recent study on Learned Index has opened up a new way to construct the index for sorted data.",
                "figure 4: case study on our proposed docunet and baseline model.",
                "to study how well maml can scale to more complex deep rl problems, we also study adaptation on high-dimensional locomotion tasks with the mujoco simu- lator   (todorov et al., 2012)."
            ]
        },
        "have be study from": {
            "sentences": [
                "Unsupervised domain adaptation has been studied from multiple perspectives."
            ]
        },
        "study of": {
            "sentences": [
                "this work motivates the study of other fundamental database operations that have not been considered for GPU acceleration."
            ]
        },
        "study in": {
            "sentences": [
                "We are, to the best of our knowledge, the first to systemat- ically study the effect of text highlighting in human com- putation, identifying the quality requirements that algo- rithms for text highlighting should possess to help with text classification and estimating the potential impact of good (and bad) highlighting."
            ]
        },
        "study the quently": {
            "sentences": [
                "In the rest of this section, we study the quently, PAX saves about 75% of the data misses NSM effects of the data placement scheme when running que-"
            ]
        }
    },
    "subcribe": {
        "subscribe to": {
            "sentences": [
                "Similarly, biases arise when people decide to like items online, bookmark web pages, or subscribe to mail lists."
            ]
        }
    },
    "subdivide": {
        "subdivide into": {
            "sentences": [
                "we subdivide the 3D space into equally spaced voxels"
            ]
        }
    },
    "submit": {
        "submit to": {
            "sentences": [
                "meaning that transactions submitted to the systems are processed at a later time",
                "gremlin queries are submitted to the gremlin server, then executed over the underlying tinkerpop3 graph database."
            ]
        }
    },
    "submitt": {
        "be submit to": {
            "sentences": [
                "When a transaction is submitted to the system, its information is inserted into the in-memory index"
            ]
        }
    },
    "succeed": {
        "succeed in": {
            "sentences": [
                "It succeeded in performing the optimization task.",
                "At last he succeeded in performing his task."
            ]
        }
    },
    "suffer": {
        "suffer from": {
            "sentences": [
                "In addition, similar to distributed machine learning, federated learning also suffers from expensive communication.",
                "\ufeffHowever, suffering from inefficient propagation of lower-level features and vanishing gradients, these deeper architectures are harder to train.",
                "Since only user-item interactions are considered, these models usually suffer from sparsity and cold-start problems.",
                "Such approaches suffer from one important limitation",
                "This is why we are suffering from an- notation labor and cost of time.",
                "however, they usually involve sequential interrelated steps and suffer from the problem of exposure bias.",
                "These techniques suffer from several disadvantages.",
                "This design suffers from two drawbacks.",
                "Although checkpoint-based recovery is able to handle any node failures, it potentially suffers from high recovery latency",
                "While the stock movement prediction has been intensively\nstudied, existing work suffers from weak generalization because of the uncertainty in both data and modeling.",
                "transformer models in particular, which the latest lms build on, would suffer from memory limitations with the naive addition of high-dimensional visual data into the sequence \u2013 computation scales quadratically with the sequence length (number of tokens), for example.",
                "Although efficient, in-memory API fuzzing suffers from unsound fuzzing results.",
                "They still suffer from limited expressiveness and poor performance.",
                "MonetDBLite suffer from excessive intermediate result materialization due to its bulk processing model.",
                "It can also occur in survey data that suffers from under-reporting.",
                "But this approach suffers from increasing cross-partition transactions",
                "Supervised models suffer from the problem of domain shifting where distribution mismatch in the data across domains greatly affect model performance.",
                "Moreover, the approach of Chen at al. seems to suffer from large neighbourhoods",
                "This approach suffers from two problems",
                "monocular vSLAM suffers from scale drift",
                "it suffers from inability of understanding the deep semantics of text and logical reasoning",
                "But, this learning paradigms suffers from the domain shift problem, which is an huge obstacle for adapting predictive models to the target domain.",
                "Sometimes the system still suffer from these weaknesses.",
                "for casrel, it is inherently a two-stage method, which suffers from both exposure bias and error propagation.",
                "Although with huge amount of entities, relations and triples, many KGs still suffer from incompletenesss",
                "Sometimes I still suffer from these weaknesses."
            ]
        }
    },
    "suffice": {
        "suffice to accurately": {
            "sentences": [
                "Our scheme is effective: a thousand samples suffice to accurately estimate theaforementioned 1010 -point query."
            ]
        },
        "suffice to": {
            "sentences": [
                "Ad- ditionally, it performs well with a low number of dimensions and when the predictors do not suffice to give more than a probabilistic estimate of the response."
            ]
        }
    },
    "suggest": {
        "suggest sth": {
            "sentences": [
                "This suggests that the amount of guessing decreases substantially."
            ]
        },
        "suggest to": {
            "sentences": [
                "It is also suggested to re-use memory locations for different operations.",
                "We suggest the proposal window w, defined as wfar \u2212 wclose + 1, to be at least four to ensure liveness (Sect. V-C)."
            ]
        },
        "suggest of": {
            "sentences": [
                "In that case they suggest on deriving a key from all the callers."
            ]
        },
        "suggest on": {
            "sentences": [
                "it also suggests that deep SDA approaches have not considered that greater performance could be achieved by encouraging class separation."
            ]
        }
    },
    "suit": {
        "be not well suit for": {
            "sentences": [
                "While classic hand-crafted neighbourhood consensus approaches are applied after a hard assignment of matches is done, this is not well suited for developing a matching method that is differentiable and amenable for end-to-end training."
            ]
        },
        "suit to": {
            "sentences": [
                "they are only suited to attain fixed representations rather than capturing polysemy",
                "This problem is well suited to the techniques we introduce, as differential privacy may allow for training on data from the true distribution"
            ]
        },
        "suit  purely": {
            "sentences": [
                "The  main  reason  for  designing  a  pluggable  For the purpose of avoiding failed attempts at communication, a  persistence component is to choose the storage engine best suited  purely local notion of failure detection is entirely sufficient: node  for an application\u2019s access patterns."
            ]
        }
    },
    "sulted": {
        "sulted in": {
            "sentences": [
                "this  process  re- sulted in randomly generated, seamless environments."
            ]
        }
    },
    "sum": {
        "sum up": {
            "sentences": [
                "Deep sets [53] achieves permutation invariance by summing up all representations and applying nonlinear transformations.",
                "After summing up all reports, the symmetric masks are canceled out if less than wn users drop out, and only the true sum remains."
            ]
        },
        "sum with": {
            "sentences": [
                "At each step, the upsampled feature map is first summed with the corresponding low-level feature map from the encoder, and then refined by a residual block"
            ]
        }
    },
    "summarize": {
        "summarize as": {
            "sentences": [
                "the contributions of this paper summarize as follows: we implement a kafka-based update mechanism on top of the ldbc snb interactive workload to better re ect the real-time, streaming aspect of social networking applications.",
                "We summarize our contributions as the following",
                "our contributions can be summarized as follows: to the best of our knowledge, this is the first approach that regards document-level re as a semantic segmentation task."
            ]
        },
        "summarize below": {
            "sentences": [
                "Various existing measures are summarized below: BLEU [16] is a precision based metric, which counts the n-gram matches between the candidate and reference texts."
            ]
        }
    },
    "summary": {
        "be summarise in": {
            "sentences": [
                "This notation is summarised in the following table."
            ]
        }
    },
    "superced": {
        "superced by": {
            "sentences": [
                "due to poor generalization performance, these approaches have largely been superceded by distributed vector representations"
            ]
        }
    },
    "superior": {
        "be superior to": {
            "sentences": [
                "Our method achieves state-of-the-art results with only LiDAR data and is superior to the original VoxelNet [14] by a significant margin,"
            ]
        }
    },
    "supersede": {
        "supersede the currently": {
            "sentences": [
                "But at the same time it  node in the system, but each node can be responsible for more  shouldn\u2019t supersede the currently unavailable state of the cart,  than one virtual node."
            ]
        }
    },
    "support": {
        "support on": {
            "sentences": [
                "the original Learned Index [22] only supports lookup on read-only datasets"
            ]
        },
        "support a much": {
            "sentences": [
                "\"By allowing users to bring their own libraries"
            ]
        },
        "be  support by": {
            "sentences": [
                "This work was supported by a project of the National Natural Science Foundation of China"
            ]
        },
        "support by": {
            "sentences": [
                "This material is based upon work supported by the National Science Foundation under Grant OAC.",
                "acknowledgements this work was supported by the national key r&d program of china (grant no.",
                "Since both traditional ARIES logging and command logging are already supported by H-Store",
                "Table 1 shows the encodings schemes supported by state-of-the-art columnar databases and storage formats",
                "These applications are currently supported by enterprise-grade database systems like Oracle and MySQL"
            ]
        },
        "support this well": {
            "sentences": [
                "TVM supports this well-known GPU optimization using a schedule primitive to achieve optimal performance."
            ]
        },
        "support sth": {
            "sentences": [
                "Our system intentionally supports looser coupling, yielding two systems that are both of interest."
            ]
        },
        "support for": {
            "sentences": [
                "n par- ticular, the data model implemented by a database play an important role in its selection because this ensure (in advance) an innate support for storing and querying the data for a desired application domain."
            ]
        },
        "support  exactly": {
            "sentences": [
                "We  ran  Aurora  supports  exactly  the  same  isolation  levels  that  are  MySQL on instances that are attached to an EBS volume with  supported by community MySQL in the writer (the standard  30K provisioned IOPS."
            ]
        },
        "support a very": {
            "sentences": [
                "We have empirically demonstrated that Cassandra cluster of 100 nodes time to taken to detect a failed can support a very high update throughput while deliver- node was in the order of two minutes."
            ]
        },
        "support a \ufb01ne": {
            "sentences": [
                "The cost of such operators Our HYRISE prototype supports a \ufb01ne-grained hybrid storage doesnotvarywhenthephysicalstoragelayoutchanges."
            ]
        }
    },
    "suppose": {
        "be suppose to": {
            "sentences": [
                "Observe that notion of \u201crounds\u201d still exist in the model (since we consider discretized time), but now these are not synchronization rounds within which all messages are supposed to be delivered to honest parties.",
                "A decentralized system is supposed to be more reliable but also more complex to implement and reasoning."
            ]
        }
    },
    "surpass": {
        "surpass by": {
            "sentences": [
                "Our best model, GRAPPA with MLM+SSP achieves the new state-of-the-art performance, surpassing previous one (RAT-SQL+BERT-large) by a margin of 4%."
            ]
        }
    },
    "surround": {
        "surround by": {
            "sentences": [
                "However, both of these approaches need the new nodes to be surrounded by known nodes and can not handle entirely new graphs"
            ]
        }
    },
    "survive": {
        "survive in": {
            "sentences": [
                "If less than (1 \u2212 w)n users survive in this iteration, the SA protocol halts, and the server learns nothing in this iteration and execute next iteration"
            ]
        }
    },
    "sweep": {
        "sweep over": {
            "sentences": [
                "during training, aug_level is chosen randomly, while during inference, we sweep over its different values to find the best sample quality."
            ]
        }
    },
    "switch": {
        "switch to": {
            "sentences": [
                "initially, we developed a simple implementation for the tinkerpop3 api on an rdbms but we switched to sqlg due to its better performance.",
                "FineStream shall switch to another alternative query plan topology.",
                "it captures modality-specific information by switching to different modality experts, and uses the shared self-attention across modalities to align visual and linguistic information.",
                "During their execution, the coordinator informs that the protocol managing R1 should be switched to 2PL.",
                "our method switches to feature tracking",
                "ACC continues to process transaction requests during the partitioning and switches to PartCC when partitioning is done"
            ]
        },
        "switch from": {
            "sentences": [
                "ACC merges the partitioned store and switches from PartCC to 2PL"
            ]
        },
        "switch across": {
            "sentences": [
                "ProteusTM is proposed to adaptively switch across multiple transaction memory algorithms for different workloads,"
            ]
        },
        "phase switch": {
            "sentences": [
                "Because this node already has a copy of every record, changing the master for a record (re-mastering) can be done without transferring any data, ensuring lightweight phase switching"
            ]
        }
    },
    "synchronize": {
        "synchronize at": {
            "sentences": [
                "at the other extreme, one could use only gatekeepers and synchronize at such high frequency so as to provide no opportunity for concurrent timestamps to arise."
            ]
        },
        "be synchronize with": {
            "sentences": [
                "Therefore, in our current implementation, each processing node caches its inmemory index which is synchronized with that in the master node periodically"
            ]
        }
    },
    "tackle": {
        "tackle with": {
            "sentences": [
                "To tackle with these issues brought by inaccurate predictions, we propose the Updatable Learned Index with Precise Positions (LIPP)"
            ]
        },
        "tackle as": {
            "sentences": [
                "Deep forecasting has typically been tackled as an end-to-end supervised learning task, where early\nwork considered using RNN based models (Lai et al., 2018) as a natural approach to modeling time\nseries data."
            ]
        }
    },
    "tag": {
        "tag as": {
            "sentences": [
                "3.1.2 decoding in the case of figure 3, (new, york), (new, city) and (de, blasio) are tagged as 1 in the eh-to-et sequence, which means new york, new york city, and de blasio are three entities."
            ]
        },
        "tag each record": {
            "sentences": [
                "we tag each record with the last TID that wrote it and apply a write if the TID of the write is larger than the current TID of the record"
            ]
        },
        "tag as highly": {
            "sentences": [
                "20 papers are tagged as highly proli(cid:12)c (HP)."
            ]
        }
    },
    "tailor": {
        "tailor to": {
            "sentences": [
                ", bronzeware script, seal script, traditional Chinese, etc) to enrich the pictographic evidence in characters; (2) We design CNN structures (called tianzege-CNN) tailored to Chinese character image processing; and (3) We use image-classification as an auxiliary task in a multi-task learning setup to increase the model\u2019s ability to generalize"
            ]
        },
        "tailor sth for": {
            "sentences": [
                "We demonstrate that tailoring these algorithms for NVM not only improves the runtime performance of the DBMS."
            ]
        }
    },
    "take": {
        "take with": {
            "sentences": [
                "We take a convolution layer with stride 2."
            ]
        },
        "take to": {
            "sentences": [
                "How long will it take to train the model?",
                "The synthetic dataset is generated by defining two trend and three seasonal\npatterns, and taking the cross product to form\nsix time series"
            ]
        },
        "take into account": {
            "sentences": [
                "We assume a bound on the maximum number of classes to expect, and modify the loss to take into account only the classes of interest",
                "this has to be taken into account",
                "This can be taken into account by considering input from several statistics modules.",
                "This requires taking into account the database schema, for example by considering that the course table has distinct columns for department and number."
            ]
        },
        "take from": {
            "sentences": [
                "This data is taken from step (1) of constructing a temporal graph.",
                "a function mapping that takes all instances from the few-shot support set and outputs the set of adapted support instance embeddings",
                "we take the rotation estimate from the pose estimate"
            ]
        },
        "take advantage of": {
            "sentences": [
                "GPL takes advantage of emerging GPU hardware features including channels and concurrent kernel execution to improve the resource utilization as well as to reduce memory overhead of the KBE approach.",
                "In our problem, since we only have access tofinite data, we take advantage of the empirical loss functions.",
                "Generative modeling can also be used for semi- supervised semantic segmentation [50, 25] to take advantage of the unlabeled examples",
                "Benefiting from this, we try to take advantage of the series\nperiodicity to renovate the point-wise connection in self-attention.",
                "we take advantage of the useful geometrical information",
                "We propose the first fine-grained window-based relational stream processing framework that takes the advantages of the special features of integrated architectures",
                "a Privacy-preserving Feature Extraction module, which takes advantage of the feature map sparsity to represent the distribution property of clients."
            ]
        },
        "take as": {
            "sentences": [
                "Weather is an\nhourly-level dataset containing 11 climate features from nearly 1,600 locations in the U.S., and we\ntake \u201cwet bulb\u201d as the target value for univariate forecasting.",
                "M is a randomized algorithm that takes a vector v as input and outputs a perturbed vector v'.",
                "Blockbench\u2019s core component is the Driver which takes as input a workload",
                "we will take the known 5 days before the target sequence as \u201cstarttoken\u201d.",
                "Taking LSTM [22] as neural encoder",
                "Our SLAM system takes the stereo camera as the sensor input."
            ]
        },
        "take into": {
            "sentences": [
                "In this way, the decision on matching assignment is done only after taking into account the spatial consensus constraints, hence avoiding errors due to early matching decisions on ambiguous, repetitive or textureless matches.",
                "Proposals that do take the infrastructure capability into account, usually focus on power consumption of nodes.",
                "This conditioned embedding takes other instances in the set into consideration",
                "FineStream needs to take device preference into consideration during placement optimization"
            ]
        },
        "take sth. as": {
            "sentences": [
                "The encoder takes a sequence x as input and outputs the representation of the sequence."
            ]
        },
        "take part in": {
            "sentences": [
                "The sets Ix are at different locations and only local samples VAROPTkx (Ix) are forwarded to the take part in the global sample."
            ]
        },
        "take ... as...": {
            "sentences": [
                "Our PointNet is a unified architecture that directly takes point clouds as input and outputs either class labels for the entire input or per point segment/part labels for each point of the input."
            ]
        },
        "take place": {
            "sentences": [
                "Also, different actions can take place under the same background."
            ]
        },
        "take ... for example": {
            "sentences": [
                "Taking colon tumor segmentation for example, the result of organ prediction is ignored during the loss computation and error back-propagation, since the annotations of organs are un- available."
            ]
        },
        "take up": {
            "sentences": [
                "actually, casrelbert and tplinkerbert both use bert as the basic encoder, which is the most time-consuming part and takes up the most of model parameters, so the time cost of handshaking tagging is not significant."
            ]
        },
        "take away": {
            "sentences": [
                "We take away that, although NVLink 2.0 puts the GPU within a factor of two of the CPUs\u2019 bandwidth, it adds significant latency."
            ]
        },
        "take into accounr": {
            "sentences": [
                "We must take into account data locality, synchronization costs, and the differences in hardware architectures to achieve efficiency."
            ]
        },
        "take on": {
            "sentences": [
                "\ud835\udc4b takes on the \ud835\udc56th symbol in an alphabet."
            ]
        },
        "take of": {
            "sentences": [
                "graph neural network based models take account of the whole structure of the graph"
            ]
        },
        "take up nearly": {
            "sentences": [
                "The rightmost CDF aggregates value sizes by the total amount of data they use in the cache, so for example, values under 320B or so in SYS use virtually no space inthe cache; 320B valuesweigharound 8% of the data, and values close to 500B take up nearly 80% of the entire cache\u2019s allocation for values."
            ]
        },
        "take  almost": {
            "sentences": [
                "has  taken  almost  a  day  to  complete."
            ]
        },
        "take in": {
            "sentences": [
                "Similarly, network protocol fuzzers such as PROTOS [120], SNOOZE [29], KiF [12], and T-Fuzz [114] also take in a protocol specification from the user."
            ]
        }
    },
    "take turns": {
        "take turn do": {
            "sentences": [
                "we take turns selecting a domain as the target domain, and use the union of the rest as the source domain."
            ]
        }
    },
    "talk": {
        "talk about": {
            "sentences": [
                "Talk about future plans.",
                "This part mainly talks about the distributed manners."
            ]
        }
    },
    "target": {
        "target in": {
            "sentences": [
                "so we target it in this paper"
            ]
        },
        "target at": {
            "sentences": [
                "This paper introduces a new concept called federated adaptation, targeting at adapting the trained model in a federated manner to achieve better personalization results.",
                "In this work, we propose IoTBolt, an automatic firmware analysis tool targeting at finding hidden services that may be potentially harmful to the IoT devices.",
                "these efforts are mostly targeted at object detection in natural-scene images as provided by the PASCAL VOC and MS-COCO bench- marks",
                "Targeting at integrated architectures, FineStream adopts continuous operator model"
            ]
        },
        "target to": {
            "sentences": [
                "\"i/os c dling updates one targeted to time series and the other wheremistheminimumnumberof\u03b5-approximatesegments targeted to more general dynamic scenarios.\""
            ]
        },
        "target  eventually": {
            "sentences": [
                "Dynamo targets  eventually-consistent storage system can be used in production  applications that need to store objects that are relatively small  with demanding applications."
            ]
        }
    },
    "tease": {
        "tease apart": {
            "sentences": [
                "We perform an ablation study to tease apart the effectiveness of each component"
            ]
        }
    },
    "tend": {
        "tend to": {
            "sentences": [
                "Using offline ETL to ingest data into columnar file formats tends to suffer from high lag between the OLTP data and the OLAP copy.",
                "PVT and similar models tend to merge too many tokens in such spatial-reduction.",
                "High-degree individuals on the other hand tend to cluster in the top-left corner of 8.2, suggesting both high availability and relatively short temporal proximity.",
                "Another advantage of static disassembly is that it takes time proportional to the size of the program, while the time taken by dynamic disassembly is typ- ically proportional to the number of instructions executed by the program at runtime\u2014the former tends to be considerably less than the latter (often by several orders of magnitude), making static dis- assembly considerably more efficient than dynamic disassembly.",
                "we tend to select the node that is rarely accessed in order to try more possible indexes and avoid falling in local-optimum",
                "TOP05 participants tended to select more information from non-highlighted parts.",
                "They tend to select more data at the beginning stage.",
                "Unfortunately, upsampling layers tend to add a high amount of computation and can introduce unwanted checkerboard artifacts [18], thus we designed the interest point detection head with an explicit decoder1 to reduce the computation of the model.",
                "The features of different layers in the deep neural net- work tend to possess different attributes",
                "\ufeffTSUBAKI\u2019s score tends to be higher when the given query is longer.",
                "Certain classes like name and requirements tended to contain fewer values, while others like year and family contained significantly more values.",
                "Models tend to recommend items according to their popularity values.",
                "ViT-L is very big (an order of magnitude bigger than ResNet-50 [25]) and tends to overfit",
                "The max-pooling operator tends to capture the most special  ping function f:RD\u00d7k\u2192RD\u2032, which maps the input features ei to a new  features.",
                "Both React and Plan tend to under-provision more when the VM provisioning time is longer.",
                "Thus, it is not sufficient for existing deep UDA methods to only minimize the discrepancy between the marginal distributions while neglecting their structures, and such methods tend to fail in challenging cases, such as those in Fig. 1.",
                "Although the micro-level indicators are more relevant to the fluctuation of stock prices\nthanks to their fast updating speed, it tends to capture the\nshort-term and local influence relationships, which is an\nadvantage but can also suffer from overlooking the partial relevance.",
                "It suggest that the performance of DoDNet tends to become stable when the width of dynamic head falls within reasonable range (\u2265 8)"
            ]
        },
        "tend to coauthorstrongly": {
            "sentences": [
                "Next, we want to compare the database community with they tend to coauthorstrongly within their group and they theAIcommunitytoseewhetherthecoauthorshiprelation- have some connection with the HP and P groups."
            ]
        }
    },
    "term": {
        "term as": {
            "sentences": [
                "To eliminate such overhead, we propose a Data-oriented Mixed Concurrency Control (DomCC), where a subset of records (i.e. clusters) are assigned a single protocol to manage all concurrent record read and write operations, which we term as a protocol mananging the cluster."
            ]
        }
    },
    "test": {
        "test on": {
            "sentences": [
                "We test the performance of these methods on KITTII dataset.",
                "We test on l2-regularized Logistic Regression and Support Vector Machine.",
                "The proposed model thus could be tested on this set",
                "For an evaluation with larger, more distinct domains, we test on the recent testbed for cross-dataset analysis.",
                "we test the performance of these methods on PCam dataset."
            ]
        },
        "test with": {
            "sentences": [
                "it was tested with queries implemented in both query languages."
            ]
        }
    },
    "thank": {
        "thank to": {
            "sentences": [
                "As of 2020, most Bitcoin blocks have no fresh transactions, thanks to the long block interval."
            ]
        }
    },
    "think": {
        "think of": {
            "sentences": [
                "We think of the above sampling as divided in two stages."
            ]
        },
        "think about": {
            "sentences": [
                "how to think about the world"
            ]
        }
    },
    "tiered": {
        "tiered storage regardless": {
            "sentences": [
                "More technical details can be tiered storage regardless of data temperatures."
            ]
        }
    },
    "tigger": {
        "trigger undefined behavior": {
            "sentences": [
                "ORDER BY clauses are still missing expressions, triggering undefined behavior due to ties in the sort order."
            ]
        }
    },
    "tioning": {
        "tioning techniques largely": {
            "sentences": [
                "tioning techniques largely ignore the node attributes in the Inthecasewhenthemostmemoryconsumingdatastruc- summarization."
            ]
        }
    },
    "token": {
        "token of": {
            "sentences": [
                "the blue tag shares a similar logic with the red tag, which means two positions are respectively the end token of a paired subject entity and object entity."
            ]
        },
        "token into": {
            "sentences": [
                "we then integrate the msg token into a multi-scale architecture named msg-transformer."
            ]
        }
    },
    "tokenize": {
        "tokenize  with": {
            "sentences": [
                "We tokenize documents with spaCy"
            ]
        }
    },
    "tokens": {
        "tokens into": {
            "sentences": [
                "integrating the transformer with msg tokens into a multi-scale design, we derive a powerful architecture named msg-transformer that takes advantages of both multi-level feature extraction and computational efficiency."
            ]
        }
    },
    "touch": {
        "touch upon": {
            "sentences": [
                "none have touched upon learning representations composed of trend and seasonal features."
            ]
        }
    },
    "track": {
        "be track by": {
            "sentences": [
                "free space is tracked by a bitmap"
            ]
        },
        "track sth. as": {
            "sentences": [
                "Existing methods tracks this problem as elim inating popularity bias,"
            ]
        },
        "track sth": {
            "sentences": [
                "The way that WBL achieves this is by tracking what parts of the database have changed rather than how it was changed."
            ]
        }
    },
    "trade": {
        "trade with": {
            "sentences": [
                "Heuristic based adaptation techniques use problem specific knowledge to provide a quick solution and trade preciseness of the outcome with lower time complexity, which makes them good candidates for dynamic resource allocation on the cloud."
            ]
        },
        "trade off": {
            "sentences": [
                "Our results across the diverse metrics highlight the trade-offs of using the different autoscalers.",
                "We quantify this privacy threat by formalizing index privacy and design a method to flexibly trade off between index privacy and utility"
            ]
        }
    },
    "train": {
        "train on": {
            "sentences": [
                "we trained on one nvidia v100 16gb gpu and evaluated our model with ign f1, and f1 following [yao et al., 2019].",
                "our zero-shot fid score is also better than state-of-the-art models trained on coco, e.g., make-a-scene [22] (at 7.6).",
                "The guiding design principle of Gato is to train on the widest variety of relevant data possible, including diverse modalities such as images, text, proprioception, joint torques, button presses, and other discrete and continuous observations and actions."
            ]
        },
        "train from scratch": {
            "sentences": [
                "It rev- els that, comparing to training from scratch, pre-training the network helps it converge quickly and perform better, particularly in the initial stage"
            ]
        },
        "train retriever separately": {
            "sentences": [
                "2.20GHz, 128 GB memory and jointly trained retriever separately."
            ]
        },
        "train more robustly": {
            "sentences": [
                "be trained more robustly from irregular and lengthy text For the NYC dataset, we perform more thorough comon data of natural blogs."
            ]
        },
        "train with": {
            "sentences": [
                "A separate generator model is trained with BERT's masked language modeling objective to produce candidate corrupted tokens.",
                "We train the model using data sources with extensive coverage and expect users can use it in various scenarios",
                "FastText [7] trains the model with text classification data",
                "Neural networks trained with metric learning losses have recently caught up with these techniques, focusing on learning repeatable saliency maps for keypoint detection and learning descriptors at the detected keypoint locations."
            ]
        },
        "train through": {
            "sentences": [
                "Deep neural models can be trained through large-scale unsupervised text corpus with skills that we call pre-trained models"
            ]
        },
        "train over": {
            "sentences": [
                "training over long sequences"
            ]
        }
    },
    "transfer": {
        "transfer to": {
            "sentences": [
                "The problem of processing unordered sets by neural nets is a very general and fundamental problem \u2013 we expect that our ideas can be transferred to other domains as well.",
                "it is called from one of the locations ai, causes control to be transferred to the corresponding location bi.",
                "The remarkable progress made by deep neural networks on the task of 2D object detection in recent years has not transferred well to the detection of objects in 3D",
                "PolarSwitch transfers the request to the corresponding chunk leader node according to the local cached cluster metadata.",
                "the transaction value is transferred to the recipient",
                "when transferring a NLI model to English and French in Fig 7."
            ]
        },
        "transfer into": {
            "sentences": [
                "The first step involves decompos- ing high level control transfers into a series of if- then-goto statements."
            ]
        },
        "transfer from": {
            "sentences": [
                "Any up- date/delete of a current record produces one or multiple historical records which are then transferred from the current relation to the historical relation simultaneously.",
                "For the GPU, this corresponds to the total amount of data transferred from main memory to global memory."
            ]
        },
        "be transfer to": {
            "sentences": [
                "A main purpose of unsupervised learning is to pre-train representations (i.e., features) that can be transferred to downstream tasks by fine-tuning"
            ]
        }
    },
    "transform": {
        "transform into": {
            "sentences": [
                "We also tried to predict two relevance scores with BERT features and additional features separately, and then linearly transform them into a weighted relevance score.",
                "which can also be transformed into triplets",
                "The purpose of data encoding is to transform the original data into a compact format to reduce storage requirements",
                "Random sparsification transforms a large vector into a sparse one by keeping only a random subset of coordinates while setting other coordinates to zeros.",
                "it could \nbe transformed into predicting whether there is a connection\nbetween them."
            ]
        },
        "transform sth. from sth. to": {
            "sentences": [
                "We trans form the problem from learning absolute values to learning relative relations,"
            ]
        },
        "transform ... to...": {
            "sentences": [
                "Since point clouds or meshes are not in a regular format, most researchers typically transform such data to regular 3D voxel grids or collections of images (e.g, views) before feeding them to a deep net architecture."
            ]
        },
        "transform\u2026into": {
            "sentences": [
                "Ben-Shabatet al.[52] transformed the input point cloud into 3D grids which are further represented by 3D modifiedFisherVector (3DmFV) method, and then learned the global representation through a conventional CNN architecture."
            ]
        },
        "log transform": {
            "sentences": [
                "The output Y is log transformed and also normalized with a StandardScaler."
            ]
        },
        "transform sth. into sth.": {
            "sentences": [
                "Predication transforms the branch (control dependency) into a data dependency."
            ]
        },
        "transform into functionally": {
            "sentences": [
                "Like LLVMIRs, a computational graph can be transformed into functionally equivalent graphs to apply optimizations."
            ]
        }
    },
    "transition": {
        "transition to": {
            "sentences": [
                "The characters started from being easy-to-draw, and slowly transitioned to being easy-to-write"
            ]
        }
    },
    "translate": {
        "translate into": {
            "sentences": [
                "Instead of synthesizing noisy SQLs and then translating into natural language utterances, we propose a Table-to-Text generation model that can directly transform a set of column names and cell values into user queries without query intent constrains.",
                "sparql queries on the rdf graph are translated into sql queries over these indexes and the base table."
            ]
        },
        "translate to": {
            "sentences": [
                "In 2015, 90% of Bitcoin public nodes have bandwidth at least 3.03 Mbps, translating to a synchronization capacity of 758 transactions per second (TPS) [19];",
                "This translates to an 8\u00d76 pixel area in a BEV map with 0.1 meter resolution",
                "For the CPU and GPU, this translates to minimizing the total number of main memory elements loaded or stored."
            ]
        },
        "translate story back": {
            "sentences": [
                "The reconstructor translates story back to the album representations."
            ]
        },
        "translate from": {
            "sentences": [
                "This analysis translate from known data to random data."
            ]
        }
    },
    "transmit": {
        "transmit to": {
            "sentences": [
                "Nevertheless, users\u2019 data are still under threats of privacy attacks if the raw gradients are transmitted to an untrusted server."
            ]
        }
    },
    "traverse": {
        "traverse down": {
            "sentences": [
                "the thread keeps traversing down until reaching the target data node"
            ]
        },
        "traverse through": {
            "sentences": [
                "it easily supports edge insertions but suffers from random accesses during scans when traversing through pointers."
            ]
        }
    },
    "treat": {
        "be treat as": {
            "sentences": [
                "each data node can be treated as a hash table that uses a model as an order-preserving hash function to predict insert location",
                "The dimension values are treated as the indices of the multidimentional arrays.",
                "Each sub-model is treated as a student model to learn knowledge from the others, which assign pseudo labels to the samples with the highest posterior probability and use them to train the student model."
            ]
        },
        "treat with": {
            "sentences": [
                "First, they treat each submitted job independently  auto-scaling mechanism with two other approaches on the cost  using their original algorithms."
            ]
        },
        "treat as": {
            "sentences": [
                "We treat\neach given forecasting horizon as a separate forecasting task.",
                "Or finally, one can just not estimate it but treat as a hyperparameter and use a validation set and tune for it using a PU evaluation metric.",
                "We treat the dataset as a byte stream",
                "Treating column headers as ground truth labels of the semantic type, we formulate semantic type detection as a multiclass classification problem."
            ]
        },
        "be treat with": {
            "sentences": [
                "It corroborates the assumption that local and nonlocal relations should be treated with distinction."
            ]
        },
        "treat it analytically": {
            "sentences": [
                "In general, a model with an adaptive basis function (e.g., neural networks) can be a potential extension, but it is much difficult to treat it analytically [53]."
            ]
        },
        "treat as just": {
            "sentences": [
                "Thus, RAMMER BASE can be treated as just another regular DNNcompiler implemented in the same codebase of RAMMER."
            ]
        }
    },
    "tricky": {
        "tricky to": {
            "sentences": [
                "it is tricky to analyze their index requirements and find the most beneficial indexes within resource limit."
            ]
        }
    },
    "trigger": {
        "trigger by": {
            "sentences": [
                "The contract code execution is triggered by transactions sent by EOAs or messages sent by other contracts",
                "With respect to recommender systems, a click interaction can be triggered by users real interest, or their conformity towards popular items."
            ]
        },
        "be trigger by": {
            "sentences": [
                "The page migration is triggered by a memory access to a page not present in GPU memory."
            ]
        }
    },
    "truncate": {
        "truncate to": {
            "sentences": [
                "We truncate each cell's contents to the first 300 characters, as some cells contain huge amounts of text."
            ]
        }
    },
    "try": {
        "try to": {
            "sentences": [
                "at the inner-level, a new task is presented, and the agent tries to quickly learn the associated concepts from the training observations.",
                "Some other methods try to find a good tradeoff between attending to more tokens and increasing attention distance.",
                "Both studies and try to achieve higher performance on CPUs by improving data sharing and data locality.",
                "Trying to represent temporal data is not straightforward if the relationships are recursive or iterative.",
                "We try to accomplish few-shot link prediction from another perspective",
                "DAN (Liu et al., 2019) tried to recover the positive and negative distributions from the unlabeled data without requiring the class prior.",
                "In this section, the code fraction, which mentioned italic, are tried to obfuscate.",
                "MultiPaxos tries to agree on the operation for each position in the sequence using a separate instance of Paxos",
                "Through iterative similarity calculation or probabilistic graphic models they try to arrive at the overall optimal solution.",
                "tries to the insert the key into that position.",
                "Given pseudo labels produced by the score filter, we tried to include the bbox regression loss l into UbT",
                "Previous DA works usually tried to minimize the domain discrepancies, such as KL-divergence and Maximum Mean Discrepancy (MMD) etc."
            ]
        },
        "try to do": {
            "sentences": [
                "Upon collecting this content, the miner tries to generate a block with target according to the proposer parent via proof-of-work (M1).",
                "Further, DBMSs try to minimize random writes to the disk."
            ]
        },
        "try out": {
            "sentences": [
                "so we need to try out less frequently selected nodes"
            ]
        }
    },
    "tune": {
        "tune to": {
            "sentences": [
                "However, if fine-tuning is used, the last layers of the encoder can be tuned to adapt to the recognition task"
            ]
        },
        "tune on": {
            "sentences": [
                "on many of these benchmarks, flamingo actually surpasses the performance of models that are fine-tuned on thousands of times more task-specific data.",
                "\ufeffThe hyperparameters of the network are tuned on the validation dataset."
            ]
        }
    },
    "turn": {
        "turn to": {
            "sentences": [
                "Having described TABBIE's model architecture, we turn now to its training objective.",
                "We next turn to larger subpopulations.",
                "Motivated by BERT, we turn to the denoising auto-encoding idea to pretrain vision Transformers, which has not been well studied by the vision community"
            ]
        },
        "turn off": {
            "sentences": [
                "Nobody turns off WALL-E.",
                "Failure to send these transactions in time leads to the receiver turning off the HB mode for the sender and turning on the HB mode for the next fastest peer (Line 8 to Line 12).",
                "When the system converges, the YAV turns off the sensors."
            ]
        },
        "turn out": {
            "sentences": [
                "More interestingly, it turns out that our network learns to summarize an input point cloud by a sparse set of key points, which roughly corresponds to the skeleton of objects according to visualization.",
                "It turned out to have a simple signature in the flow record; namely as being udp traffic to port 1434 with a packet size of 404 bytes.",
                "It turns out that there are subtle differences that make it impossible for Raft to have a direct refinement mapping to Pax"
            ]
        },
        "turn for": {
            "sentences": [
                "As a heavyweight operator, the join is extensively studied and tuned for modern CPUs, resulting in multiple variants of the radix hash-join"
            ]
        },
        "turn attention to do": {
            "sentences": [
                "We now turn our attention to evaluating selections, also called selection scans."
            ]
        },
        "turn into": {
            "sentences": [
                "Turning the problem into a convex problem and the smaller noise for each update, we can obtain smaller and more stable losses in training"
            ]
        },
        "turn out to be": {
            "sentences": [
                "While this appears to be a simple and intuitive algorithm, minor seemingly-innocuous variants turn out to be dangerously insecure."
            ]
        },
        "turn on": {
            "sentences": [
                "Failure to send these transactions in time leads to the receiver turning off the HB mode for the sender and turning on the HB mode for the next fastest peer (Line 8 to Line 12)."
            ]
        }
    },
    "underly": {
        "underlie data": {
            "sentences": [
                "For real-world settings, this metricplays an important role as its underlying data always updates withtuples insertions and deletions."
            ]
        }
    },
    "underperform": {
        "underperform on": {
            "sentences": [
                "We show that TaBERT underperforms on downstream table-based applications that operate independent of external text."
            ]
        }
    },
    "understand": {
        "understand as horizontally": {
            "sentences": [
                "this can be understood as horizontally and vertically flattening image pixels in raster-scan order and then applying strided attention with gaps of image width and height, respectively."
            ]
        },
        "understand how well": {
            "sentences": [
                "however, typical measures of accuracy or calibration do not match this discretization procedure, and as such further analysis would be valuable to understand how well model accuracy aligns with it\u2019s suitability as a hash function."
            ]
        },
        "understand on": {
            "sentences": [
                "Furthermore, to give more in-depth understanding on the performance improvement, we further investigate hardware performance counters on cache, memory units and device utilization."
            ]
        }
    },
    "unfold": {
        "unfold or only": {
            "sentences": [
                "The only part of the computational graph is unfolded or only some number of decisions for different programs becomes different."
            ]
        }
    },
    "unify": {
        "unify into": {
            "sentences": [
                "Data structure unified KEPTMs transform sequences and knowledge into a unified structure that links sequences",
                "Future work includes unifying pose graph optimization and mesh deformation into a single optimization."
            ]
        }
    },
    "update": {
        "update to": {
            "sentences": [
                "in addition to assigning timestamps to transactions, the gatekeepers also commit transactional updates to the backing store."
            ]
        },
        "update the record": {
            "sentences": [
                "The core idea of Active-Memory is to use the one-sided feature of RDMA to directly update the records on remote backup servers without involving the remote CPUs"
            ]
        },
        "update with": {
            "sentences": [
                "Node 1 updates the superblocks with the modified metadata"
            ]
        }
    },
    "upload": {
        "upload to": {
            "sentences": [
                "First, each client trains a model locally with a few epochs and only uploads the parameters/gradients of the model to a central server.",
                "These representations are further uploaded to the server to recall a small number of candidate news"
            ]
        }
    },
    "use": {
        "use to": {
            "sentences": [
                "Raft uses AppendEntries to replicate operations to other servers,",
                "Pinyin Embedding The pinyin embedding for each character is used to decouple different semantic meanings belonging to the same character form, as shown in Figure 3",
                "For long-running tasks, this data can be used to build a reasonably accurate stochastic model of CPU usage.",
                "These datasets can be used to develop algorithms for various challenges including similar distractors, shape incompleteness, and class imbalance.",
                "it is the medium of human expression used to express cognition of the world",
                "Meta-learning is based on relation-speci\ufb01c learning and is used to learn a meta-model with parameters",
                "Concurrency control (CC) techniques have been widely used to achieve efficient access (including read and write operations) in the database systems",
                "Additionally, self-modifying code has been used to improve software tamper resistance techniques.",
                "Seqpool uses an attention function to weight the individual information.",
                "Lagged regressors are used to correlate other observed variables to our target time series.",
                "Language-specific compilers for smart contracts (such as Solidity or Serpent) are used to compile the contracts into byte code.",
                "Point and line correspondences are used to estimate translation by a combination of frame-toframe and frame-to-map methods."
            ]
        },
        "use as": {
            "sentences": [
                "word embeddings have been commonly used as text representation in NLP tasks",
                "We use MSE and MAE as evaluation metrics,\nand perform a 60/20/20 train/validation/test split.",
                "DeepSDF [45] uses deep networks as a function that returns the signed distance field value of a query point.",
                "we use the standard learnable 1D position embedding as location embedding and add it to the spreading feature map."
            ]
        },
        "use concepts directly": {
            "sentences": [
                "\"In this approachwe assume that it is far easier for the user to de- \ufb01ne triggers for dynamic resource provisioning"
            ]
        },
        "use a linearly": {
            "sentences": [
                "\"Weevalu- order of resource provisioning con\ufb01gurations"
            ]
        },
        "use endnotes instead": {
            "sentences": [
                "(We are using Endnotes instead of footnotes."
            ]
        },
        "use cpu2 signi\ufb01cantly": {
            "sentences": [
                "\"Ononehand"
            ]
        },
        "use the previously": {
            "sentences": [
                "\"Animplementation of a tool based on this concept"
            ]
        },
        "use for": {
            "sentences": [
                "it replaces the previous keyframe and is used for tracking subsequent new frames",
                "According to the network architecture used for the feature learning of each point, methods in this category can be divided into pointwise MLP, convolution-based, graphbased, hierarchical data structure-based methods and other typical methods.",
                "At this stage, our use of days as a unit of measurement is arbitrary, as hours or seconds can equally be used for the same purpose.",
                "A coordinator thread is used for extracting features for the prediction classifiers from statistics collected by workers and predicting the ideal protocol to be used.",
                "causal order is used for dynamic knowledge",
                "Generally, more and more deep models are used for time\nseries prediction task",
                "In this section, we describe the key concepts related to Blockchain in general and the Ethereumblock chain network which we have used for our proposed BPIIoT platform."
            ]
        },
        "use by": {
            "sentences": [
                "kafka has become the de facto architecture for streaming applications, used by linkedin, 3 twitter, 4 and interest, 5 just to name a few organizations."
            ]
        },
        "make use of": {
            "sentences": [
                "For machine translation, we make use of the datasets from WMT16."
            ]
        },
        "use in": {
            "sentences": [
                "on the other hand, as these studies[nguyen and grishman, 2015; shen and huang, 2016] have done, convolutional neural networks have been long used in the relation extraction area, which enlightens us to pay attention to the role of cnn in extracting information of the image-style feature map.",
                "in our case, we use gaussian noise as a form of augmentation, and apply variance preserving gaussian noise augmentation resembling the forward process used in diffusion models (appendix a).",
                "Ether is the currency which is used in the Ethereum Blockchain network",
                "it is also used in information retrieval and NLP",
                "partitioning is used in many database approximation problems."
            ]
        },
        "be use to": {
            "sentences": [
                "The logged messages are used to reduce recovery workload",
                "The GPU queue is used to sequentially execute the queries on the GPU, while the CPU queue is used to execute the related queries on the CPU",
                "Human correlation and accuracy are commonly used to then use the object detector in [4] to extract initial visual evaluate the image caption evaluation metrics.",
                "Their 1115 systems can be used to automatically generate textual description of a day's events from 1116 images just like diary writing.",
                "As a first step, note that the substitution u = m(t) can be used to calculate."
            ]
        },
        "use for compute": {
            "sentences": [
                "be used for compute-bounded problems [11]."
            ]
        },
        "use gpus exclusively": {
            "sentences": [
                "On the computation side, we use GPUs exclusively for DNNcomputation to avoid the irregular memory lookup operations thatdegrade the SIMD performance."
            ]
        },
        "use a more": {
            "sentences": [
                "One could eliminate these errors using a moreadvanced type system, but at increased complexity."
            ]
        },
        "use only": {
            "sentences": [
                "in contrast to prior work that uses only image-text data for model training [e.g., 53, 41], the key finding behind imagen is that text embeddings from large lms [52, 15], pretrained on text-only corpora, are remarkably effective for text-to-image synthesis.",
                "building models that can be rapidly adapted to numerous tasks using only a handful of annotated examples is an open challenge for multimodal machine learning research."
            ]
        },
        "use sth": {
            "sentences": [
                "To evaluate our approach, we implemented it in the Peloton in-memory DBMS and compared it against WAL using three storage technologies: NVM, SSD, and HDD."
            ]
        },
        "be use for": {
            "sentences": [
                "While the cross-entropy loss is widely used for supervised training of deep neural networks",
                "Paxos be used for some practical purposes?",
                "This configuration is then used for the next fuzz iteration. Notice that SCHEDULE is only about decision-making."
            ]
        },
        "use aurora primarily": {
            "sentences": [
                "Since our customers use Aurora primarily  are not HAT-compliant, while most other isolation levels are  as  an  OLTP  service  backing  production  applications,  any  achievable  with  high  availability."
            ]
        },
        "use by nally": {
            "sentences": [
                "Bigtable is used by nally,Section10describesrelatedwork,andSection11 more than sixty Google products and projects, includ- presentsourconclusions."
            ]
        },
        "use it internally": {
            "sentences": [
                "use it internally for the location column family in the quentialreads, or randomreads of differentcolumnsin METADATAtable. thesamelocalitygroupwithinahotrow)."
            ]
        },
        "use hy- as": {
            "sentences": [
                "increasing Memcached\u2019s hit rate, especially on larger data, ticular large-scale production traces, many works used hy- as well as theupperboundon thepotentialincrease."
            ]
        },
        "use a highly": {
            "sentences": [
                "face at Amazon.com, one of the largest e-commerce operations in  Amazon uses a highly decentralized, loosely coupled, service  the world; even the slightest outage has significant financial  oriented architecture consisting of hundreds of services."
            ]
        }
    },
    "utilize": {
        "utilize to": {
            "sentences": [
                "most existing index management methods utilize cost estimator in the database to estimate the index benefit and greedily select indexes with maximum estimated cost reduction.",
                "CodecDB identifies features that impact datasets\u2019 encoding performance and utilizes machine learning techniques to train a series of models to predict compression and query performance",
                "Once the module is learned, it can be utilized to active learning as shown in Figure 1-(b).",
                "meanwhile DMA is heavily utilized to transfer data between main memory and RDMA NIC/NVMe disks.",
                "The purpose of this work is to design a machine learning model which could be utilized to predict the new facts with few-shot reference instances"
            ]
        },
        "utilize as": {
            "sentences": [
                "The remaining positive entity pairs are utilized as positive query pairs"
            ]
        },
        "utilize sth": {
            "sentences": [
                "Chen et al. utilize matrix factorization to generate additional table headers and then show that those generated headers can improve the performance of unsupervised table search.",
                "Although we utilize a what-if interface to get the cost without building an index physically"
            ]
        },
        "utilize by": {
            "sentences": [
                "the context-based strategy is utilized by default."
            ]
        },
        "be utilize to": {
            "sentences": [
                "Furthermore, under an generate framework allows the use of external resources end-to-end training process, extra plain text and knowledge not only for the enrichment phase, but also for the distilla-base data cannot be utilized to better enrich the story."
            ]
        },
        "utilize a much larger model space": {
            "sentences": [
                "This special class of BDL, known as Neural Network Gaussian Process (NNGP), inherits the advantages of Bayesian approachwhile keeping universal approximation of neural network, and can utilize a much larger model space to model distribution-free dataas a nonparametric model."
            ]
        },
        "utilize a much larger model": {
            "sentences": [
                "Asa nonparametric model, NNGP utilizes a much larger model spaceto model distribution-free data."
            ]
        },
        "utilize a fully": {
            "sentences": [
                "Sup [28] encodes queries as a set of featuresand learns weights for these features utilizing a fully connectedneural network to estimate the selectivity."
            ]
        },
        "utilize the memory access": {
            "sentences": [
                "LIPP utilizes the fewer memory accesses to finish these operations"
            ]
        },
        "utilize mask": {
            "sentences": [
                "for language pre-training, we freeze parameters of the attention module and vision expert, and utilize masked language modeling [10] to optimize the language expert on text-only data."
            ]
        }
    },
    "validate": {
        "validate it empirically": {
            "sentences": [
                "According to our theory, we should still do fine, butlet us validate it empirically with other query distributions Randomand Gaussian, which are independent from the underlying data distribution."
            ]
        },
        "be successfully validate": {
            "sentences": [
                "The TID is assigned to a transaction after it is successfully validated"
            ]
        },
        "validate by symbolically": {
            "sentences": [
                "These properties are manually writtenand reviewed, and are further validated by symbolically executing operators on tensors of small sizes and confirmingthat the operator properties are satisfied for these tensorsizes."
            ]
        },
        "validate by": {
            "sentences": [
                "Since a Blockchain network is not controlled by a central authority and all the transactions are verified and validated by a consensus among the peers,"
            ]
        },
        "validate through": {
            "sentences": [
                "We validate the proposed method through image classi- fication, human pose estimation, and object detection. The human pose estimation is a typical regression task, and the object detection is a more complex problem combined with both regression and classification."
            ]
        }
    },
    "variants": {
        "variants of": {
            "sentences": [
                "Prior work introduces hardware-conscious variants of several operators, including scan-like operators, sort-based operations and index scans."
            ]
        }
    },
    "vary": {
        "vary in": {
            "sentences": [
                "the length varies in different heads for capturing different granularity information.",
                "Since the reported performance of supervised baseline varies in different works, we report the results of the comparison methods and their baseline at the same time",
                "The length varies in different heads for capturing different granularity information."
            ]
        },
        "vary over": {
            "sentences": [
                "Demand for recourses varies over time as users consume and release these resources."
            ]
        },
        "vary on": {
            "sentences": [
                "One user's conformity varies on different items."
            ]
        },
        "vary to": {
            "sentences": [
                "In the second experiment, we vary the amount of memory allocated to the progressive approximate join."
            ]
        },
        "vary with": {
            "sentences": [
                "We believe that the interactions should vary with the different categories or topics of questions.",
                "In ACC we are facing several new challenges. First, cross-cluster access cost varies with multiple concurrency protocols",
                "Although data compression and approximation can reduce transfer volume, their effectiveness varies with the data and query."
            ]
        },
        "vary from sth. to sth.": {
            "sentences": [
                "We vary the thread block sizes from 32 to 1024 in multiples of 2."
            ]
        },
        "vary from": {
            "sentences": [
                "It varies from 10,000 transactions per second to 40,000 (resp. 100,000) transactions per second with the TPC-C (resp. SmallBank) workload in Figure 5a (resp. Figure 6a)",
                "The storage budget varies from 2 units to 8 units where each unit indicates 128.57M.",
                "As TDSQL-T has already maintained the changed accounts automatically, of which the number varies from 2 million to 5 million, TDSQL-T only needs to join the changed accounts with W, which re- sults in much less join computation than that of TDSQL."
            ]
        },
        "vary for": {
            "sentences": [
                "we are confident that our results provide meaningful indications of the annotation difficulty of collection, sharing, and deletion practices,which may vary for individual websites but are largely ubiquitous."
            ]
        }
    },
    "verify": {
        "verify over": {
            "sentences": [
                "Moving beyond the scope of that dataset, a fully open fact verification system should be able to verify claims over even larger collections of tables"
            ]
        },
        "verify with": {
            "sentences": [
                "To the best of our knowledge, this is the first work verified with three different recognition tasks using the state-of-the-art deep network models."
            ]
        }
    },
    "verted": {
        "verted to": {
            "sentences": [
                "5 conclusion in this paper, we propose an end-to-end sequence labeling model tplinker for joint extraction of entities and relations based on a novel handshaking tagging strategy, by which the joint extraction task is converted to a token pair linking game."
            ]
        }
    },
    "view": {
        "view as": {
            "sentences": [
                "Our approach can be viewed as principled discriminative version of this algorithm",
                "We view Prophet as a precursor to such a solution.",
                "each containing multiple transactions, thus the blockchain can be viewed as a log of ordered transactions",
                "RTE and NLI view a premise sentence as the evidence, whereas claim verification views passage as the evidence and commonsense reasoning views a context sentence as the evidence."
            ]
        },
        "view as explicitly": {
            "sentences": [
                "our method can also be viewed as explicitly maximiz- ing  sensitivity  of  new  task  losses  to  the  model  parameters."
            ]
        },
        "be view as": {
            "sentences": [
                "These equations can be viewed as conditional expectations",
                "In this way, the forward process of each layer can be viewed as a motion pattern applied to the input features",
                "It can be viewed as a variant of RATSQL with our tailored grammar-based decoder."
            ]
        },
        "view in": {
            "sentences": [
                "examples from the latter consist of text with interleaved images, corresponding to the page layout when viewed in a browser.",
                "[2020] proposed a hierarchical inference network (hin) to make full use of the abundant information from the entity, deconv country input encoder module u-shaped segmentation module classification module conv conv poolconv conv poolconv conv conv conv deconv conv conv ffd figure 3: architecture of our document u-shaped network (docunet) (best viewed in color)."
            ]
        },
        "view sth. as sth.": {
            "sentences": [
                "Past work generally views GPUs strictly as an coprocessor. Every query ends up shipping data from CPU to GPU over PCIe."
            ]
        }
    },
    "violate": {
        "violate by": {
            "sentences": [
                "synchronicity assumptions interfere with debugging, and maybe violated by network delays under heavy load and systems running in virtualized environments."
            ]
        }
    },
    "visit": {
        "visit it again": {
            "sentences": [
                "If the current typerelation is now completely solved, we discard it to avoidunnecessarily visiting it again."
            ]
        },
        "visit and only": {
            "sentences": [
                "Finally, an output is also produced if b and c have non-zero valuesbut not d. For efficiency, only the non-zero values in the inputs are visited and only the non-zerovalues in the output are produced."
            ]
        }
    },
    "visualize": {
        "visualize with": {
            "sentences": [
                "When trained with all source data, features are visualized with some changes in their distributions where instances from different domains are mixed and closer to the target domain."
            ]
        },
        "visualize the distribution": {
            "sentences": [
                "we visualize the gap distribution of the output values for various datasets and the corresponding proportion of empty slot"
            ]
        }
    },
    "vote": {
        "vote to": {
            "sentences": [
                "exploits sparsity in point cloud data by leveraging a feature-centric voting  scheme to implement novel convolutions"
            ]
        },
        "vote for": {
            "sentences": [
                "Upon receiving a valid proposer block B with chain difficulty higher than the previous heaviest chain, the miner makes B the new proposer parent, and updates allm voter trees to vote for chain difficulties until B."
            ]
        }
    },
    "wait": {
        "wait until": {
            "sentences": [
                "If t(i) and t(j) conflict in requesting partition locks of PartCC, either t(i) or t(j) will wait in the Preprocess phase until the other transaction finishes."
            ]
        },
        "wait for": {
            "sentences": [
                "A na\u00a8\u0131ve approach is to recommend miners to wait for a fixed amount of time after receiving a transaction before packing it into a block.",
                "A transaction is allowed to commit only when its dependency counter is zero, whereupon the DBMS traverses its dependency list and decrements the counters for all the transactions that are waiting for it to finish.",
                "On the CPU, if a thread makes a memory access, the thread waits for the memory fetch to return.",
                ". Finally, the commit phase waits for more than two third of the nodes to confirm before announcing that the value is committed",
                "Therefore, transactions blocked by a protocol will not wait for transactions blocked by other protocols in earlier phases."
            ]
        },
        "wait as long": {
            "sentences": [
                "when using PAX, because the processor does not need to 5.2.2 NSM/PAXSensitivityAnalysis wait as long for data to arrive from main memory."
            ]
        }
    },
    "walk": {
        "walk on": {
            "sentences": [
                "We present agent MINERVA, which learns to do query answering by walking on a knowledge graph conditioned on an input query"
            ]
        },
        "walk through": {
            "sentences": [
                "SigGraph's data structure definition extractor adopts a compiler-based approach, where the compiler pass is devised to walk through the source code and extract data structure definitions."
            ]
        }
    },
    "want": {
        "want to": {
            "sentences": [
                "Most of the time, users want to query temporal data as illustrated in the following examples.",
                "We want to maintain a generic sample of a certain limited size that we can later use to estimate the total weight of arbitrary subsets.",
                "In this work, we focus on the former case as we want to predict the unseen facts of a given relation",
                "Interpreting probabilities as a measure of confidence, we may want to only label samples with high confidence of belonging to a type.",
                "The owners of a Slock who wants to rent their real-world physical objects (such as houses, cars or bikes) set a deposit amount and a price for using the objects",
                "Similarly, we want to maximize the covering of edges in image x direction."
            ]
        }
    },
    "warm": {
        "warm up": {
            "sentences": [
                "we warmed up the system and ensured that all data is in memory."
            ]
        }
    },
    "weight": {
        "weight to": {
            "sentences": [
                "when this assumption does not hold,  one could use randomly generated keys, non-keys generated by a machine learning model [34], importance  weighting to directly address covariate shift [ 18], or adversarial training for robustness [65]; we leave this as  future work."
            ]
        }
    },
    "well": {
        "as well as": {
            "sentences": [
                "On the GPU front, recent work explores optimized algorithms for GPU execution as well as GPU query execution models"
            ]
        }
    },
    "whichcaneasilydou-": {
        "whichcaneasilydou- but especially": {
            "sentences": [
                "opmentanddeploymentisthatonemustbudgetindividual Small values dominate all workloads, not just in count, nodecapacitytoallowforthesespikes,whichcaneasilydou- but especially in overall weight."
            ]
        }
    },
    "wish": {
        "wish to": {
            "sentences": [
                "Since we consider hybrid algorithms, we wish to avoid the additional CPU computation overhead of these techniques, so we do not use them in this work.",
                "Since we wish to apply the algorithm for retaining/deleting k nodes.",
                "It is exactly this impact that we wish to quantitatively analyse.",
                "In our analysis of the Enron and Cityware data we wish to better understand how information flows, or can flow, in these networks over time.",
                "a trusted intermediary is required for transactions between the users who wish to avail manufacturing services",
                "The repository also supports developers wishing to retrain Sherlock using data from their specific data ecologies, such as enterprise or research settings with domain-specific data."
            ]
        }
    },
    "with": {
        "roll with": {
            "sentences": [
                "Under the rolling forecasting setting with a fixed size window."
            ]
        }
    },
    "witness": {
        "witness great success": {
            "sentences": [
                "Recent years have witnessed great success in recommender systems."
            ]
        }
    },
    "wonder": {
        "wonder about": {
            "sentences": [
                "It is natural to wonder about the relative strengths and weaknesses of the various approaches for estimating the class prior."
            ]
        }
    },
    "work": {
        "work under": {
            "sentences": [
                "Data node expansion and split require more care to work under optimistic locking."
            ]
        },
        "work for": {
            "sentences": [
                "This demonstrates that our method can work effectively for the distribued localization.",
                "This demonstrates that our method can work effectively for histology image classification.",
                "This demonstrates that our algorithm can work effectively for distributed solving.",
                "While any general sequence model can work for next token prediction, we chose a transformer (Vaswani et al., 2017) for simplicity and scalability.",
                "While these approaches work for many scenarios, they all require the number of expected partitions to be known in advance or rely on simple heuristics for adding partitions if a solution cannot be found",
                "\ufeffThis idea works for finding the highest scoring sentence in a larger collection.",
                "This ap- proach works well for labeling tasks, such as tagging or categoriz- ing images",
                "AnGrapa is an large-scale graph data management frame- work for analytical processing."
            ]
        },
        "work less well": {
            "sentences": [
                "\"It worked less well for stateful services"
            ]
        },
        "work at all": {
            "sentences": [
                "however, in more complex domains, such as few-shot imitation learning (see chapter 7), we found the \ufb01rst-order approxima- tion to not work at all."
            ]
        },
        "work to": {
            "sentences": [
                "The model will assign more work to the respective architecture."
            ]
        },
        "work group": {
            "sentences": [
                "Each work-group is scheduled as the basic unit to one CU."
            ]
        },
        "work with": {
            "sentences": [
                "Column type annotation and relation extraction both work with table columns.",
                "Additionally, we can found several in-memory graph tools characterized by their restriction to work with small graphs."
            ]
        },
        "work in automatically": {
            "sentences": [
                "TASO's approach to identifying potential substitutions via enumeration of graphs andfingerprinting is similar to work in automatically generatingpeephole optimizers using superoptimization techniques [7]."
            ]
        },
        "work by": {
            "sentences": [
                "the transformer module works by using a token to formulate the feature at each spatial position."
            ]
        },
        "work on": {
            "sentences": [
                "However, existing PU learning methods all work on a batch learning mode, which cannot deal with the online learning scenarios with sequential data.",
                "Finally, none of the prior works on ConvQA produce human-interpretable structures that could assist end users in case of erroneous system responses.",
                "To construct a valid block, miners work on finding the right nonce so that the block hash is smaller than a target T"
            ]
        },
        "work in": {
            "sentences": [
                "and works in three phases in which nodes broadcast messages to each other"
            ]
        },
        "work as": {
            "sentences": [
                "The validation works as a guard of programs."
            ]
        }
    },
    "workload": {
        "workload much more": {
            "sentences": [
                "ThekeyobservationisthatCloudComputing\u2019sabilitytoaddorremoveresourcesata\ufb01ne grain (one server at a time with EC2) and with a lead time of minutes rather than weeks allows matching resources to workload much more closely."
            ]
        }
    },
    "wrap": {
        "wrap around canbevisualizedasacolumnfamilywithinacolumnfamily": {
            "sentences": [
                "Super column families space or \u201cring\u201d (i.e. the largest hash value wraps around canbevisualizedasacolumnfamilywithinacolumnfamily."
            ]
        }
    },
    "write": {
        "write in": {
            "sentences": [
                "DCPMM also writes in 256B units.",
                "The library is wrriten in GO",
                "Contracts can be written in various high-level languages"
            ]
        },
        "write to": {
            "sentences": [
                "Then, we scan the scratchpad and write each tuple to its corresponding output partition, before moving to the next chunk.",
                "In these attacks, the injected code is written to a program\u2019s memory as input data to which control is subsequently transferred."
            ]
        },
        "write out": {
            "sentences": [
                "Assuming the implementations can write out the matched entries efficiently and saturate memory bandwidth"
            ]
        },
        "be write in": {
            "sentences": [
                "The original CPU operators in Saber are written in Java"
            ]
        },
        "be write into": {
            "sentences": [
                "logs are written into NVMe SSD instead"
            ]
        },
        "write but also": {
            "sentences": [
                "only because of how data is written but also because of what data  Unfortunately, this model results in untenable performance for a  is written."
            ]
        },
        "write-only": {
            "sentences": [
                "Table 3: SysBench OLTP (writes/sec)  Connections  Amazon Aurora  MySQL  50  40,000  10,000  500  71,000  21,000  5,000  110,000  13,000  Figure 7: Aurora scales linearly for write-only workload    6.1.4  Scaling with Replicas",
                "shows that for the SysBench write-only workload, Aurora can be  The buffer cache on the r3.8xlarge is set to 170GB.",
                "[9]  representing the same information in multiple ways \u2013 for example,  write-only  workload  with  a  100GB  data  set  for  both  the writes to the double write buffer in order to prevent torn pages  configurations described above: one with a synchronous mirrored  in the storage infrastructure."
            ]
        },
        "write in monotonically": {
            "sentences": [
                "Spannerdependsonthefollowingmonotonicityinvariant:withineachPaxosgroup, Spanner assigns timestamps to Paxos writes in monotonically increasing order, even across leaders."
            ]
        },
        "write will no": {
            "sentences": [
                "(cid:2) (cid:2)(cid:2) (cid:2) (cid:2) (cid:2) Spanner:Google\u2019sGloballyDistributedDatabase 8:11 writes are applied in order, writes will no longer occur at or below tPaxos with respect safe toPaxos."
            ]
        },
        "write be now": {
            "sentences": [
                "All application writes are now by default sent through F1 to Spanner, instead of the MySQL-based application stack."
            ]
        }
    },
    "yield": {
        "yield sth": {
            "sentences": [
                "A statement is redrawn if it yields an error"
            ]
        },
        "yield the the best-in-class quality": {
            "sentences": [
                "For Conviva-A, 2 epochs yields the best-in-class quality and about15 epochs yields the quality of single-digit max error."
            ]
        },
        "yield surprisingly": {
            "sentences": [
                "while conceptually simple and easy to train, imagen yields surprisingly strong results."
            ]
        }
    },
    "\ufb01nd": {
        "\ufb01nd so much": {
            "sentences": [
                "4.2.1, we \ufb01nd so much as models for syntheticworkloads."
            ]
        }
    }
}