2017 11 15

wangkuiyi

• Load/save matrices:
  • https://github.com/PaddlePaddle/Paddle/pull/5519#pullrequestreview-75611051
  • https://github.com/PaddlePaddle/Paddle/issues/5528

srinetkavya

• Add documentation for multi-device survey Tensorflow: https://github.com/PaddlePaddle/Paddle/pull/5412
• Few more updates on multi-device proposal Paddle: https://github.com/PaddlePaddle/Paddle/pull/5284
• Rewrite documentation for RNN and Sequence decoder: https://github.com/PaddlePaddle/Paddle/pull/5611
• Write documentation for every layer : https://github.com/PaddlePaddle/Paddle/pull/5529
• Set up the documentation tool on device and debug the rendering of documentation for layers.py
• Working on implementing the FTRL operator (PR is in process).

kexinzhao

• PR
  • [WIP] Add half precision float16 data type: https://github.com/PaddlePaddle/Paddle/pull/5716
  • Add python interface for optimizer: https://github.com/PaddlePaddle/Paddle/pull/5644
• PR review
  • https://github.com/PaddlePaddle/Paddle/pull/5453#pullrequestreview-75583129
  • https://github.com/PaddlePaddle/Paddle/pull/5272#pullrequestreview-76689772

sidgoyal78

• Working on row-convolution operator (WIP): https://github.com/PaddlePaddle/Paddle/issues/5612

abhinavarora

• PR

  • Adding greater than and less than equal ops to compare op https://github.com/PaddlePaddle/Paddle/pull/5609
  • Adding logical operators for beam search and control flow https://github.com/PaddlePaddle/Paddle/pull/5708
  • https://github.com/PaddlePaddle/Paddle/pull/5610

• Code Review

  • https://github.com/PaddlePaddle/Paddle/pull/5639
  • https://github.com/PaddlePaddle/Paddle/pull/5524
  • https://github.com/PaddlePaddle/Paddle/pull/5354

cs2be (Thuan)

• Remove sidebar and permalink navigation on models and mobile https://github.com/PaddlePaddle/PaddlePaddle.org/pull/304
• Add permalink nav to models and mobile. Move navigation.sass after content.sass. FIxed issue with permalink showing when there is only one link. https://github.com/PaddlePaddle/PaddlePaddle.org/pull/313
• Add gzip compression to nginx https://github.com/PaddlePaddle/PaddlePaddle.org/pull/328
• Setup load balancing and redundant servers for PPO https://github.com/PaddlePaddle/PaddlePaddle.org/issues/330
• Add wiki to provision new servers https://github.com/PaddlePaddle/PaddlePaddle.org/wiki/AWS-Provision-New-Server

daming-lu

• Add script to deploy ppo.zip to AWS in a daily cron job https://github.com/PaddlePaddle/PaddlePaddle.org/pull/305
• Add 3 modes to developing doc https://github.com/PaddlePaddle/PaddlePaddle.org/pull/317
• Study different stress test tools, choose blazemeter and artillery.io to generate high QPS for PaddlePaddle.org
• Prepare BIT talk based on PPO development challenges and solutions

Yu Yang

• Dynamic RNN
  • Array <--> LoDTensor conversion op
    • work with tony, jiayi
    • https://github.com/PaddlePaddle/Paddle/pull/5411
  • ArrayRead/Write Operator
    • https://github.com/PaddlePaddle/Paddle/pull/5409
  • ShrinkMemory Operator
    • https://github.com/PaddlePaddle/Paddle/pull/5419
  • WhileOp forward
• If Operator
  • Conditional Block Operator https://github.com/PaddlePaddle/Paddle/pull/5530
  • IfElse Python API
    • Work with qijun
    • https://github.com/PaddlePaddle/Paddle/pull/5624

Several enhancements

• Generate doc string for generated operator
  • https://github.com/PaddlePaddle/Paddle/pull/5585
• assign op
  • https://github.com/PaddlePaddle/Paddle/pull/5531
• add array length op
  • https://github.com/PaddlePaddle/Paddle/pull/5501
• change increase_op, it can only be run in CPU
  • https://github.com/PaddlePaddle/Paddle/pull/5499
• CompareOp
  • https://github.com/PaddlePaddle/Paddle/pull/5465
  • https://github.com/PaddlePaddle/Paddle/pull/5325

jetfuel (Jeff Wang)

• Compress banner.png to banner.jpg https://github.com/PaddlePaddle/PaddlePaddle.org/pull/321
• Update the HTML title tag accordingly. https://github.com/PaddlePaddle/PaddlePaddle.org/pull/315
• Left align the service-desc https://github.com/PaddlePaddle/PaddlePaddle.org/pull/312
• parse the Models homepage so it doesn't take the users to GitHub https://github.com/PaddlePaddle/PaddlePaddle.org/pull/307
• Update translation https://github.com/PaddlePaddle/PaddlePaddle.org/pull/296
• Provide the video tutorial links on the README page. https://github.com/PaddlePaddle/PaddlePaddle.org/pull/277

• Triage issues created by PaddlePaddle users
• Communicated with Baidu AIOP team to prioritize the action needed for Baidu World
• PaddlePaddle.org Stress test
• Communicate and plan the BIT presentation trip.

nickyfantasy

• Finalized iOS AICamera Demo integrated with 4 different models and settings for front/back camera, time refresh rate, accuracy threshold etc https://github.com/PaddlePaddle/Mobile/pull/26

ranqiu

• update convs2s

  https://github.com/PaddlePaddle/models/pull/431 https://github.com/PaddlePaddle/models/pull/467

• update doc of layers.py

  https://github.com/PaddlePaddle/Paddle/pull/5517 https://github.com/PaddlePaddle/Paddle/pull/5678

luotao

• remove usused ProtoDataProvider related codes: https://github.com/PaddlePaddle/Paddle/pull/5345
• MKL
  • remove unused INTEL_MKL_ROOT etc: https://github.com/PaddlePaddle/Paddle/pull/5512
  • fix compile warning in test_MKLDNN.cpp and ROIPoolLayer.cpp: https://github.com/PaddlePaddle/Paddle/pull/5600
  • [WIP] resnet50 benchmark data: https://github.com/PaddlePaddle/Paddle/pull/5578
  • code review on MKLDNN:
    • https://github.com/PaddlePaddle/Paddle/pull/5547
    • https://github.com/PaddlePaddle/Paddle/pull/5543
    • https://github.com/PaddlePaddle/Paddle/pull/5582
• OpenBlas:
  • fix openblas bug: undefined reference to pthread_atfork: https://github.com/PaddlePaddle/Paddle/pull/5552
  • A lot of warnings in -Wunused-but-set-variable and Wunused-variable: https://github.com/xianyi/OpenBLAS/issues/1357
• WebSite:
  • fix typo in faq: https://github.com/PaddlePaddle/Paddle/pull/5536
  • Discuss the directory structure of models repo:
    • https://github.com/PaddlePaddle/models/issues/458
    • https://github.com/PaddlePaddle/models/issues/462
    • https://github.com/PaddlePaddle/models/issues/464

caoying

• PR
  • refine DSSM: https://github.com/PaddlePaddle/models/pulls
  • update introduction to NCE: https://github.com/PaddlePaddle/models/pull/443
  • small fix:
    • https://github.com/PaddlePaddle/models/pull/463
    • https://github.com/PaddlePaddle/models/pull/461
  • fix a bug of NER: https://github.com/PaddlePaddle/models/pull/473
• PR review:
  • generate Chinese poetry:
    • https://github.com/PaddlePaddle/models/pull/439#pullrequestreview-75980596
  • https://github.com/PaddlePaddle/Paddle/pull/5622#pullrequestreview-76344140
  • https://github.com/PaddlePaddle/Paddle/pull/4859#pullrequestreview-76373761

wangmeng

• Finish generating chinese poetry
  • https://github.com/PaddlePaddle/models/pull/439

hedaoyuan

• Mobile
  • https://github.com/PaddlePaddle/Paddle/pull/5679
  • https://github.com/PaddlePaddle/Mobile/pull/28
  • https://github.com/PaddlePaddle/Paddle/issues/5675
  • https://github.com/PaddlePaddle/PaddlePaddle.org/issues/294
• Review
  • https://github.com/PaddlePaddle/Paddle/pull/5165
  • https://github.com/PaddlePaddle/Paddle/pull/5573
  • https://github.com/PaddlePaddle/Paddle/pull/4891
  • https://github.com/PaddlePaddle/Mobile/pull/26
  • https://github.com/PaddlePaddle/Paddle/pull/5605
  • https://github.com/PaddlePaddle/Paddle/pull/5660

qijun

Paddle Book:

• Stack LSTM Net for Paddle Book6
  • https://github.com/PaddlePaddle/Paddle/pull/5503

Refactoring:

• add split and merge lod tensor operator
  • https://github.com/PaddlePaddle/Paddle/pull/5537
• IfElse Python API(work with yuyang)
  • https://github.com/PaddlePaddle/Paddle/pull/5624

Fix and Enhance:

• create learning rate variable for every parameter
  • https://github.com/PaddlePaddle/Paddle/pull/5524
• fix gitignore
  • https://github.com/PaddlePaddle/Paddle/pull/5657

ZhangChao

• Update Bilinear Tensor Product operator.[merged] https://github.com/PaddlePaddle/Paddle/pull/5014

• Fix docstrings of lstmemory_group,lstmemory_unit,lstm_step_layer.[merged] https://github.com/PaddlePaddle/Paddle/pull/5622

• [WIP] Change the epsilon in batch_norm_layer to a variable. https://github.com/PaddlePaddle/Paddle/issues/5548

Dang Qingqing

• [Speed Compiling]: Reduce NVCC compiling files:
  • https://github.com/PaddlePaddle/Paddle/pull/5573
• C-API inference of OCR-based text decetion:
  • https://github.com/qingqing01/PaddleDemo/tree/master/ocr_text_detection
• Code Review:
  • Maxout Op: https://github.com/PaddlePaddle/Paddle/pull/5571#pullrequestreview-76042697
  • SequenceSlice Op: https://github.com/PaddlePaddle/Paddle/pull/5546#pullrequestreview-75700603
  • Add dilations for conv2d and optimize conv2d code：https://github.com/PaddlePaddle/Paddle/pull/5472
  • fix pooling functor parameter order：https://github.com/PaddlePaddle/Paddle/pull/5608
  • Assign Operator: https://github.com/PaddlePaddle/Paddle/pull/5531#pullrequestreview-76009034
  • Stack LSTM Net for Paddle Book6: https://github.com/PaddlePaddle/Paddle/pull/5503#pullrequestreview-75646795
  • Fix seq concat op with refactoring LoD: https://github.com/PaddlePaddle/Paddle/pull/5486

guosheng

• Refine ROIPoolLayer(merged): https://github.com/PaddlePaddle/Paddle/pull/2982
• Convert the detection model of OCR team to PaddlePaddle model
• Fix data order of H0 in GRU Operator: https://github.com/PaddlePaddle/Paddle/pull/5623
• PR review:
  • https://github.com/PaddlePaddle/models/pull/473

wanghaoshuang

• Op porting

  • Add NCE op:
    • https://github.com/PaddlePaddle/Paddle/pull/5480

zhaochengduo

• OpOporting
  • [WIP] Add cudnn pool3d
    • https://github.com/PaddlePaddle/Paddle/pull/5602
  • [WIP] Add conv3d_trans_cudnn_op
    • https://github.com/PaddlePaddle/Paddle/pull/5603
  • [Merged] Fix pooling functor parameter order
    • https://github.com/PaddlePaddle/Paddle/pull/5608
  • [Merged] Add dilations for conv2d and optimize conv2d code
    • https://github.com/PaddlePaddle/Paddle/pull/5472
• Review
  • Add maxout operator
    • https://github.com/PaddlePaddle/Paddle/pull/5571
  • Add sequence slice operator
    • https://github.com/PaddlePaddle/Paddle/pull/5546
  • conv shift op: change to CamelCase & fix bug
    • https://github.com/PaddlePaddle/Paddle/pull/5558
• Issue
  • Some inline functions and structs don't follow "Camel Case".
    • https://github.com/PaddlePaddle/Paddle/issues/5549

dongzhihong

• Refactorization:
  • Python API add numpy/LoDTensor feed/fetch interface
    • https://github.com/PaddlePaddle/Paddle/pull/5620
  • call once in DSO
    • https://github.com/PaddlePaddle/Paddle/pull/5553
  • add Tensor copytovector
    • https://github.com/PaddlePaddle/Paddle/pull/5455
  • elementwise_mod operator
    • https://github.com/PaddlePaddle/Paddle/pull/5454
  • add evaluator python
    • https://github.com/PaddlePaddle/Paddle/pull/5331
  • fix accuracy_op gpu mode bug
    • https://github.com/PaddlePaddle/Paddle/pull/5673
  • fix elementwise_op gpu macro
    • https://github.com/PaddlePaddle/Paddle/pull/5647
  • fix ci compile
    • https://github.com/PaddlePaddle/Paddle/pull/5567
  • fix reduce op in debug mode
    • https://github.com/PaddlePaddle/Paddle/pull/5565
  • unsigned variable compare problem
    • https://github.com/PaddlePaddle/Paddle/pull/5359
  • fix startup_program
    • https://github.com/PaddlePaddle/Paddle/pull/5314
  • update memory design doc
    • https://github.com/PaddlePaddle/Paddle/pull/5682

helinwang

• Paddle autoscaling news article and code polish

  • https://github.com/PaddlePaddle/cloud/pull/493
  • https://github.com/PaddlePaddle/cloud/pull/492

• Python demo API polish

  • https://github.com/PaddlePaddle/Paddle/pull/5615

gongweibao

• Parse job config on controller: https://github.com/PaddlePaddle/cloud/pull/496
• Run PaddlePaddle Cloud on minikube： https://github.com/PaddlePaddle/cloud/pull/463

Yibing Liu

Deep Speech 2：

• fixes:
  • https://github.com/PaddlePaddle/DeepSpeech/pull/10
  • https://github.com/PaddlePaddle/DeepSpeech/pull/7
  • https://github.com/PaddlePaddle/models/pull/453
• Adapt tuning script to padding removing machnism
  • https://github.com/PaddlePaddle/models/pull/447
  • https://github.com/PaddlePaddle/models/pull/449
• Re-train aishell model
• code review:
  • https://github.com/PaddlePaddle/DeepSpeech/pull/2

Op Porting:

• Update and merge LoD Reset Operator
  • https://github.com/PaddlePaddle/Paddle/pull/4747

typhoonzero(wuyi):

• autoscaler scheduler design:
  • https://github.com/PaddlePaddle/cloud/pull/468
• send recv op: ready to review.
  • https://github.com/PaddlePaddle/Paddle/pull/5520
• manylinux compile env and CI:
  • https://github.com/PaddlePaddle/Paddle/pull/5430 merged

qiaolongfei

refactoring

• Beam search decode:

  • BeamSearchDecodeOp
  • Beam search decode op python

• change framework to fluid

  • Change framework to fluid
  • mv test_beam_search_decode_op.py to fluid
  • rm unused dir framework

• bugfix && code refine

  • fix lod_tensor_array

code review

• Fix conv2d bias

Hu Weiwei

• Refine some documents.

  https://github.com/PaddlePaddle/models/pull/470 [Merged] https://github.com/PaddlePaddle/DeepSpeech/pull/6 [Merged]

Yan Chunwei

• PR
  • merged, beamsearch op
  • open, is_empty_op
• Reviews
  • https://github.com/PaddlePaddle/Paddle/pull/5609#pullrequestreview-76320584
  • https://github.com/PaddlePaddle/Paddle/pull/5652#pullrequestreview-76665300
  • https://github.com/PaddlePaddle/models/pull/472#pullrequestreview-76737081
  • https://github.com/PaddlePaddle/Paddle/pull/5611#pullrequestreview-76981991
• VisualDL with longfei
  • overall design of backend architecture
  • feature/add storage protobuf

tonyyang-svail

• Dynamic RNN:
  • while_op: https://github.com/PaddlePaddle/Paddle/pull/5681
• Python API:
  • executor: https://github.com/PaddlePaddle/Paddle/pull/5681

yangyaming

• DS2

  • Publish BaiduEng8k model
    https://github.com/PaddlePaddle/models/pull/469
  • Bug fix
    https://github.com/PaddlePaddle/models/pull/446
  • Adapt to support padding removing
    https://github.com/PaddlePaddle/models/pull/444
  • Performance verification
    Libri data, Aishell data
  • Review
    https://github.com/PaddlePaddle/models/pull/453
    https://github.com/PaddlePaddle/models/pull/449
    https://github.com/PaddlePaddle/models/pull/447

• Paddle

  • Bug fix for chunk_eval_op
    https://github.com/PaddlePaddle/Paddle/pull/5542
  • Bug fix for CTR model
    https://github.com/PaddlePaddle/models/pull/472
  • Review
    https://github.com/PaddlePaddle/Paddle/pull/5546

zhouxiao:

• PR:
  • Merged PnpEvaluator: https://github.com/PaddlePaddle/Paddle/pull/5108
  • Doing Add normalization methods && implement layer norm: https://github.com/PaddlePaddle/Paddle/pull/5686
• Issue:
  • https://github.com/PaddlePaddle/Paddle/issues/5685

Yan Xu

• TeamCity CI
  • Build different version Python package with manlinux standar
  • Adjust trigger to scheulded trigger for all project except PR CI.
• OP port
  • [WIP] Hsigmoid Operator, https://github.com/PaddlePaddle/Paddle/pull/5676
• Review
  • CFS Scheudler design, https://github.com/PaddlePaddle/cloud/pull/468
  • Following question about trainer_count and CPU utils, https://github.com/PaddlePaddle/Paddle/issues/5280

Liu Yiqun

• Mobile deployment
  • SSD demo
    • on linux: https://github.com/Xreki/Mobile/tree/add_ssd_linux_demo/demo/linux/ssd
    • on iOS: https://github.com/PaddlePaddle/Mobile/pull/26
  • Use Paddle to do inference for OCR
    • https://github.com/Xreki/PaddleNativeDemo/tree/master/linux/optical_character_recognizer
  • [Reviewing] Support the build for multiple architectures at one cmake command (iOS)
    • https://github.com/PaddlePaddle/Paddle/pull/5677
  • [Merged] Fix compiling error for Android, and installing error for cmake of low version
    • https://github.com/PaddlePaddle/Paddle/pull/5660
  • [Merged] Fix bug in MergeModel.cpp
    • https://github.com/PaddlePaddle/Paddle/pull/5605

Chenxi

• Prepare Fluid EDL PR and experiment document
• Review Kubernetes controller design

varunarora

• Setup load balanced / scaling servers for Baidu World
• Fixed breaking math symbols issues: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/314
• Started unifying styles to ai.baidu.com specification
• Fixed homepage banner glitches: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/321
• Fixed markdown->HTML conversion breaking issues: https://github.com/PaddlePaddle/PaddlePaddle.org/pull/306