/mlc-llm

Enable everyone to develop, optimize and deploy AI models natively on everyone's devices.

Primary LanguagePythonApache License 2.0Apache-2.0

MLC LLM

Documentation | Blog | Discord

Machine Learning Compilation for Large Language Models (MLC LLM) is a high-performance universal deployment solution that allows native deployment of any large language models with native APIs with compiler acceleration. The mission of this project is to enable everyone to develop, optimize and deploy AI models natively on everyone's devices with ML compilation techniques.

Universal deployment. MLC LLM supports the following platforms and hardware:

AMD GPU NVIDIA GPU Apple GPU Intel GPU
Linux / Win ✅ Vulkan, ROCm ✅ Vulkan, CUDA N/A ✅ Vulkan
macOS ✅ Metal (dGPU) N/A ✅ Metal ✅ Metal (iGPU)
Web Browser ✅ WebGPU and WASM
iOS / iPadOS ✅ Metal on Apple A-series GPU
Android ✅ OpenCL on Adreno GPU ✅ OpenCL on Mali GPU

Quick Start

We introduce the quick start examples of chat CLI, Python API and REST server here to use MLC LLM. We use 4-bit quantized 8B Llama-3 model for demonstration purpose. The pre-quantized Llama-3 weights is available at https://huggingface.co/mlc-ai/Llama-3-8B-Instruct-q4f16_1-MLC. You can also try out unquantized Llama-3 model by replacing q4f16_1 to q0f16 in the examples below. Please visit our documentation for detailed quick start and introduction.

Installation

MLC LLM is available via pip. It is always recommended to install it in an isolated conda virtual environment.

To verify the installation, activate your virtual environment, run

python -c "import mlc_llm; print(mlc_llm.__path__)"

You are expected to see the installation path of MLC LLM Python package.

Chat CLI

We can try out the chat CLI in MLC LLM with 4-bit quantized 8B Llama-3 model.

mlc_llm chat HF://mlc-ai/Llama-3-8B-Instruct-q4f16_1-MLC

It may take 1-2 minutes for the first time running this command. After waiting, this command launch a chat interface where you can enter your prompt and chat with the model.

You can use the following special commands:
/help               print the special commands
/exit               quit the cli
/stats              print out the latest stats (token/sec)
/reset              restart a fresh chat
/set [overrides]    override settings in the generation config. For example,
                      `/set temperature=0.5;max_gen_len=100;stop=end,stop`
                      Note: Separate stop words in the `stop` option with commas (,).
Multi-line input: Use escape+enter to start a new line.

user: What's the meaning of life
assistant:
What a profound and intriguing question! While there's no one definitive answer, I'd be happy to help you explore some perspectives on the meaning of life.

The concept of the meaning of life has been debated and...

Python API

We can run the Llama-3 model with the chat completion Python API of MLC LLM. You can save the code below into a Python file and run it.

from mlc_llm import MLCEngine

# Create engine
model = "HF://mlc-ai/Llama-3-8B-Instruct-q4f16_1-MLC"
engine = MLCEngine(model)

# Run chat completion in OpenAI API.
for response in engine.chat.completions.create(
    messages=[{"role": "user", "content": "What is the meaning of life?"}],
    model=model,
    stream=True,
):
    for choice in response.choices:
        print(choice.delta.content, end="", flush=True)
print("\n")

engine.terminate()

The Python API of mlc_llm.MLCEngine fully aligns with OpenAI API. You can use MLCEngine in the same way of using OpenAI's Python package for both synchronous and asynchronous generation.

If you would like to do concurrent asynchronous generation, you can use mlc_llm.AsyncMLCEngine instead.

REST Server

We can launch a REST server to serve the 4-bit quantized Llama-3 model for OpenAI chat completion requests. The server has fully OpenAI API completeness.

mlc_llm serve HF://mlc-ai/Llama-3-8B-Instruct-q4f16_1-MLC

The server is hooked at http://127.0.0.1:8000 by default, and you can use --host and --port to set a different host and port. When the server is ready (showing INFO: Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)), we can open a new shell and send a cURL request via the following command:

curl -X POST \
  -H "Content-Type: application/json" \
  -d '{
        "model": "HF://mlc-ai/Llama-3-8B-Instruct-q4f16_1-MLC",
        "messages": [
            {"role": "user", "content": "Hello! Our project is MLC LLM. What is the name of our project?"}
        ]
  }' \
  http://127.0.0.1:8000/v1/chat/completions

Universal Deployment APIs

MLC LLM provides multiple sets of APIs across platforms and environments. These include

Citation

Please consider citing our project if you find it useful:

@software{mlc-llm,
    author = {MLC team},
    title = {{MLC-LLM}},
    url = {https://github.com/mlc-ai/mlc-llm},
    year = {2023}
}

The underlying techniques of MLC LLM include:

References (Click to expand)
@inproceedings{tensorir,
    author = {Feng, Siyuan and Hou, Bohan and Jin, Hongyi and Lin, Wuwei and Shao, Junru and Lai, Ruihang and Ye, Zihao and Zheng, Lianmin and Yu, Cody Hao and Yu, Yong and Chen, Tianqi},
    title = {TensorIR: An Abstraction for Automatic Tensorized Program Optimization},
    year = {2023},
    isbn = {9781450399166},
    publisher = {Association for Computing Machinery},
    address = {New York, NY, USA},
    url = {https://doi.org/10.1145/3575693.3576933},
    doi = {10.1145/3575693.3576933},
    booktitle = {Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2},
    pages = {804–817},
    numpages = {14},
    keywords = {Tensor Computation, Machine Learning Compiler, Deep Neural Network},
    location = {Vancouver, BC, Canada},
    series = {ASPLOS 2023}
}

@inproceedings{metaschedule,
    author = {Shao, Junru and Zhou, Xiyou and Feng, Siyuan and Hou, Bohan and Lai, Ruihang and Jin, Hongyi and Lin, Wuwei and Masuda, Masahiro and Yu, Cody Hao and Chen, Tianqi},
    booktitle = {Advances in Neural Information Processing Systems},
    editor = {S. Koyejo and S. Mohamed and A. Agarwal and D. Belgrave and K. Cho and A. Oh},
    pages = {35783--35796},
    publisher = {Curran Associates, Inc.},
    title = {Tensor Program Optimization with Probabilistic Programs},
    url = {https://proceedings.neurips.cc/paper_files/paper/2022/file/e894eafae43e68b4c8dfdacf742bcbf3-Paper-Conference.pdf},
    volume = {35},
    year = {2022}
}

@inproceedings{tvm,
    author = {Tianqi Chen and Thierry Moreau and Ziheng Jiang and Lianmin Zheng and Eddie Yan and Haichen Shen and Meghan Cowan and Leyuan Wang and Yuwei Hu and Luis Ceze and Carlos Guestrin and Arvind Krishnamurthy},
    title = {{TVM}: An Automated {End-to-End} Optimizing Compiler for Deep Learning},
    booktitle = {13th USENIX Symposium on Operating Systems Design and Implementation (OSDI 18)},
    year = {2018},
    isbn = {978-1-939133-08-3},
    address = {Carlsbad, CA},
    pages = {578--594},
    url = {https://www.usenix.org/conference/osdi18/presentation/chen},
    publisher = {USENIX Association},
    month = oct,
}

Links

  • You might want to check out our online public Machine Learning Compilation course for a systematic walkthrough of our approaches.
  • WebLLM is a companion project using MLC LLM's WebGPU and WebAssembly backend.
  • WebStableDiffusion is a companion project for diffusion models with the WebGPU backend.