diff --git a/README.md b/README.md
index 61bacba9..ec6ff24f 100644
--- a/README.md
+++ b/README.md
@@ -1,100 +1,54 @@
-
-
-🎉 Introduction
-KTransformers, pronounced as Quick Transformers, is designed to enhance your 🤗 Transformers experience with advanced kernel optimizations and placement/parallelism strategies.
-
-KTransformers is a flexible, Python-centric framework designed with extensibility at its core.
-By implementing and injecting an optimized module with a single line of code, users gain access to a Transformers-compatible
-interface, RESTful APIs compliant with OpenAI and Ollama, and even a simplified ChatGPT-like web UI.
-
-Our vision for KTransformers is to serve as a flexible platform for experimenting with innovative LLM inference optimizations. Please let us know if you need any other features.
-
-🔥 Updates
-
-* **Feb 15, 2025**: KTransformers V0.2.1: Longer Context (from 4K to 8K for 24GB VRAM) & Slightly Faster Speed （+15%) (Up to 16 Tokens/s), update docs [here](./doc/en/DeepseekR1_V3_tutorial.md) and [online books](https://kvcache-ai.github.io/ktransformers/).
-* **Feb 10, 2025**: Support Deepseek-R1 and V3 on single (24GB VRAM)/multi gpu and 382G DRAM, up to 3~28x speedup. For detailed show case and reproduction tutorial, see [here](./doc/en/DeepseekR1_V3_tutorial.md).
-* **Aug 28, 2024**: Decrease DeepseekV2's required VRAM from 21G to 11G.
-* **Aug 15, 2024**: Update detailed [tutorial](doc/en/injection_tutorial.md) for injection and multi-GPU.
-* **Aug 14, 2024**: Support llamfile as linear backend.
-* **Aug 12, 2024**: Support multiple GPU; Support new model: mixtral 8\*7B and 8\*22B; Support q2k, q3k, q5k dequant on gpu.
-* **Aug 9, 2024**: Support windows native.
-
-🌟 Show Cases
-
-
-
GPT-4/o1-level Local VSCode Copilot on a Desktop with only 24GB VRAM
-
-
-https://github.com/user-attachments/assets/ebd70bfa-b2c1-4abb-ae3b-296ed38aa285
-
-
-
-- **[NEW!!!] Local 671B DeepSeek-Coder-V3/R1:** Running its Q4_K_M version using only 14GB VRAM and 382GB DRAM([Tutorial](./doc/en/DeepseekR1_V3_tutorial.md)).
- - Prefill Speed (tokens/s):
- - KTransformers: 54.21 (32 cores) → 74.362 (dual-socket, 2×32 cores) → 255.26 (optimized AMX-based MoE kernel, V0.3 only) → 286.55 (selectively using 6 experts, V0.3 only)
- - Compared to 10.31 tokens/s in llama.cpp with 2×32 cores, achieving up to **27.79× speedup**.
- - Decode Speed (tokens/s):
- - KTransformers: 8.73 (32 cores) → 11.26 (dual-socket, 2×32 cores) → 13.69 (selectively using 6 experts, V0.3 only)
- - Compared to 4.51 tokens/s in llama.cpp with 2×32 cores, achieving up to **3.03× speedup**.
- - Upcoming Open Source Release:
- - AMX optimizations and selective expert activation will be open-sourced in V0.3.
- - Currently available only in preview binary distribution, which can be downloaded [here](./doc/en/DeepseekR1_V3_tutorial.md).
-
-- **Local 236B DeepSeek-Coder-V2:** Running its Q4_K_M version using only 21GB VRAM and 136GB DRAM, attainable on a local desktop machine, which scores even better than GPT4-0613 in [BigCodeBench](https://huggingface.co/blog/leaderboard-bigcodebench).
+[🌟 Show Cases](#show-cases) | [🚀 Quick Start](#quick-start) | [📃 Tutorial](#tutorial) | [💬 Discussion](https://github.com/kvcache-ai/ktransformers/discussions) | [🙋 FAQ](#FAQ)
-
-
- 
-
-
-- **Faster Speed:** Achieving 126 tokens/s for 2K prompt prefill and 13.6 tokens/s for generation through MoE offloading and injecting advanced kernels from [Llamafile](https://github.com/Mozilla-Ocho/llamafile/tree/main) and [Marlin](https://github.com/IST-DASLab/marlin).
-- **VSCode Integration:** Wrapped into an OpenAI and Ollama compatible API for seamless integration as a backend for [Tabby](https://github.com/TabbyML/tabby) and various other frontends.
+## 🎉 Introduction
-
+KTransformers, pronounced as Quick Transformers, enhances [🤗 Transformers](https://github.com/huggingface/transformers) with advanced kernel optimizations, parallelism, and placement strategies.
-https://github.com/user-attachments/assets/4c6a8a38-05aa-497d-8eb1-3a5b3918429c
+It includes a Transformers compatible interface, RESTful APIs compatible with OpenAI and Ollama schema, and a simple ChatGPT-inspired web client.
-
+KTransformers aims to provide a versatile platform for experimenting with novel LLM inference optimizations. Please contact us or open an issue if you request any additional features.
-
+- **Local 236B DeepSeek-Coder-V2:** Runs at Q4_K_M with 21GB VRAM and 136GB DRAM, suitable for desktop PCs, outperforming GPT4-0613 in [BigCodeBench](https://huggingface.co/blog/leaderboard-bigcodebench).
+
-More advanced features will coming soon, so stay tuned!
+- **Faster Inference:** Achieves 126 tokens/sec for 2K prompt prefill and 13.6 tokens/sec for generation using MoE offloading and use of optimized kernels from [LlamaFile](https://github.com/Mozilla-Ocho/llamafile/) and [Marlin](https://github.com/IST-DASLab/marlin).
+- **VSCode Integration:** Features an OpenAI and Ollama-compatible API for seamless integration as a backend for [Tabby](https://github.com/TabbyML/tabby) and various other frontends.
🚀 Quick Start
@@ -105,22 +59,17 @@ Getting started with KTransformers is simple! Follow the steps below to set up a
To install KTransformers, follow the official [Installation Guide](https://kvcache-ai.github.io/ktransformers/).
+🚀 Quick Start
+
+At the core of KTransformers is a user-friendly, template-based injection framework. This allows researchers to effortlessly replace original torch modules with optimized variants. It also simplifies the process of combining multiple optimizations to explore their synergistic effects.
-📃 Brief Injection Tutorial
-At the heart of KTransformers is a user-friendly, template-based injection framework.
-This allows researchers to easily replace original torch modules with optimized variants. It also simplifies the process of combining multiple optimizations, allowing the exploration of their synergistic effects.
+
-
-
-
- 
-
-
+Considering vLLM already serves as an excellent framework for large-scale deployment optimizations, KTransformers primarily focuses on local deployments constrained by limited resources. We pay special attention to heterogeneous computing opportunities, like GPU/CPU offloading of quantized models. For example, we support the efficient [LlamaFile](https://github.com/Mozilla-Ocho/llamafile/) and [Marlin](https://github.com/IST-DASLab/marlin) kernels for both the CPU and GPU. More details can be found [here](doc/en/operators/llamafile.md).
-Given that vLLM already serves as a great framework for large-scale deployment optimizations, KTransformers is particularly focused on local deployments that are constrained by limited resources. We pay special attention to heterogeneous computing opportunities, such as GPU/CPU offloading of quantized models. For example, we support the efficient Llamafile and Marlin kernels for CPU and GPU, respectively. More details can be found here.
+### Example Usage
-Example Usage
-To utilize the provided kernels, users only need to create a YAML-based injection template and add the call to `optimize_and_load_gguf` before using the Transformers model.
+To use the provided kernels, users only need to create a YAML-based injection template and add the call to `optimize_and_load_gguf` before using the Transformers model.
```python
with torch.device("meta"):
@@ -130,11 +79,11 @@ optimize_and_load_gguf(model, optimize_rule_path, gguf_path, config)
generated = prefill_and_generate(model, tokenizer, input_tensor.cuda(), max_new_tokens=1000)
```
-In this example, the AutoModel is first initialized on the meta device to avoid occupying any memory resources. Then, `optimize_and_load_gguf` iterates through all sub-modules of the model, matches rules specified in your YAML rule file, and replaces them with advanced modules as specified.
+In this example, the AutoModel is first initialized on the meta device to avoid taking any memory. Then, `optimize_and_load_gguf` iterates through all sub-modules, matcheing the rules specified, and replacing them with the optimized modules.
-After injection, the original `generate` interface is available, but we also provide a compatible `prefill_and_generate` method, which enables further optimizations like CUDAGraph to improve generation speed.
+After injection, the original `generate` interface is available, but we also provide a compatible `prefill_and_generate` method, which enables further optimizations like CUDAGraph to improve inference speed.
-How to custom your model
+### How to Customize Your Model
A detailed tutorial of the injection and multi-GPU using DeepSeek-V2 as an example is given [here](doc/en/injection_tutorial.md).
@@ -142,7 +91,7 @@ Below is an example of a YAML template for replacing all original Linear modules
```yaml
- match:
- name: "^model\\.layers\\..*$" # regular expression
+ name: "^model\\.layers\\..*$" # regular expression
class: torch.nn.Linear # only match modules matching name and class simultaneously
replace:
class: ktransformers.operators.linear.KTransformerLinear # optimized Kernel on quantized data types
@@ -158,17 +107,16 @@ You can find example rule templates for optimizing DeepSeek-V2 and Qwen2-57B-A14
If you are interested in our design principles and the implementation of the injection framework, please refer to the [design document](doc/en/deepseek-v2-injection.md).
-Acknowledgment and Contributors
+## 🙋 FAQ
-The development of KTransformer is based on the flexible and versatile framework provided by Transformers. We also benefit from advanced kernels such as GGUF/GGML, Llamafile, Marlin, sglang and flashinfer. We are planning to contribute back to the community by upstreaming our modifications.
+Some common questions are answered in the [FAQ](doc/en/FAQ.md)
-KTransformer is actively maintained and developed by contributors from the MADSys group at Tsinghua University and members from Approaching.AI. We welcome new contributors to join us in making KTransformer faster and easier to use.
-
-
-Discussion
+## Discussion
If you have any questions, feel free to open an issue. Alternatively, you can join our WeChat group for further discussion. QR Code: [WeChat Group](WeChatGroup.png)
-🙋 FAQ
+## Acknowledgments and Contributors
+
+KTransformers builds upon the flexible and versatile framework provided by 🤗 Transformers. We have also benefited from advanced kernels such as GGUF/GGML, LlamaFile, Marlin, Sglang, and FlashInfer. We plan to contribute back to the community by upstreaming our modifications.
-Some common questions are answered in the [FAQ](doc/en/FAQ.md).
+KTransformers is actively maintained and developed by contributors from the [MADSys group](https://madsys.cs.tsinghua.edu.cn/) at Tsinghua University, along with members from [Approaching.AI](https://approaching.ai/). We welcome new contributors to join us in making KTransformers faster and easier to use.