vLLM + `logit_bias`

How to install the forked version

This fork adds support for logit_bias used in OpenAI apis.

conda create -n myenv python=3.8 -y
conda activate myenv

# Install vLLM.
pip install git+https://github.com/kazemnejad/vllm.git # This may take 5-10 minutes.

Examples of usage:

No bias

curl http://localhost:8000/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "meta-llama/Llama-2-7b-chat-hf",
        "prompt": "San Francisco is a",
        "max_tokens": 7,
        "temperature": 0
    }'

Outputs: city in Northern California that is known

Prevent "city"

curl http://localhost:8000/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "meta-llama/Llama-2-7b-chat-hf",
        "prompt": "San Francisco is a",
        "max_tokens": 7,
        "temperature": 0,
        "logit_bias": {
            "4272": -100
        }
    }'

Outputs: popular tourist destination, known for

Prevent "city" and "destination"

curl http://localhost:8000/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "meta-llama/Llama-2-7b-chat-hf",
        "prompt": "San Francisco is a",
        "max_tokens": 7,
        "temperature": 0,
        "logit_bias": {
            "4272": -100,
            "12551": -100
        }
    }'

Outputs: popular tourist attraction, known

Always generate "city"

curl http://localhost:8000/v1/completions \
    -H "Content-Type: application/json" \
    -d '{
        "model": "meta-llama/Llama-2-7b-chat-hf",
        "prompt": "San Francisco is a",
        "max_tokens": 7,
        "temperature": 0,
        "logit_bias": {
            "4272": 100
        }
    }'

Outputs: city city city city city city city

Latest News 🔥

[2023/08] We would like to express our sincere gratitude to Andreessen Horowitz (a16z) for providing a generous grant to support the open-source development and research of vLLM.
[2023/07] Added support for LLaMA-2! You can run and serve 7B/13B/70B LLaMA-2s on vLLM with a single command!
[2023/06] Serving vLLM On any Cloud with SkyPilot. Check out a 1-click example to start the vLLM demo, and the blog post for the story behind vLLM development on the clouds.
[2023/06] We officially released vLLM! FastChat-vLLM integration has powered LMSYS Vicuna and Chatbot Arena since mid-April. Check out our blog post.

vLLM is a fast and easy-to-use library for LLM inference and serving.

vLLM is fast with:

State-of-the-art serving throughput
Efficient management of attention key and value memory with PagedAttention
Continuous batching of incoming requests
Optimized CUDA kernels

vLLM is flexible and easy to use with:

Seamless integration with popular HuggingFace models
High-throughput serving with various decoding algorithms, including parallel sampling, beam search, and more
Tensor parallelism support for distributed inference
Streaming outputs
OpenAI-compatible API server

vLLM seamlessly supports many Huggingface models, including the following architectures:

Aquila (BAAI/Aquila-7B, BAAI/AquilaChat-7B, etc.)
Baichuan (baichuan-inc/Baichuan-7B, baichuan-inc/Baichuan-13B-Chat, etc.)
BLOOM (bigscience/bloom, bigscience/bloomz, etc.)
Falcon (tiiuae/falcon-7b, tiiuae/falcon-40b, tiiuae/falcon-rw-7b, etc.)
GPT-2 (gpt2, gpt2-xl, etc.)
GPT BigCode (bigcode/starcoder, bigcode/gpt_bigcode-santacoder, etc.)
GPT-J (EleutherAI/gpt-j-6b, nomic-ai/gpt4all-j, etc.)
GPT-NeoX (EleutherAI/gpt-neox-20b, databricks/dolly-v2-12b, stabilityai/stablelm-tuned-alpha-7b, etc.)
InternLM (internlm/internlm-7b, internlm/internlm-chat-7b, etc.)
LLaMA & LLaMA-2 (meta-llama/Llama-2-70b-hf, lmsys/vicuna-13b-v1.3, young-geng/koala, openlm-research/open_llama_13b, etc.)
MPT (mosaicml/mpt-7b, mosaicml/mpt-30b, etc.)
OPT (facebook/opt-66b, facebook/opt-iml-max-30b, etc.)
Qwen (Qwen/Qwen-7B, Qwen/Qwen-7B-Chat, etc.)

Install vLLM with pip or from source:

pip install vllm

Getting Started

Visit our documentation to get started.

Performance

vLLM outperforms HuggingFace Transformers (HF) by up to 24x and Text Generation Inference (TGI) by up to 3.5x, in terms of throughput. For details, check out our blog post.

Serving throughput when each request asks for 1 output completion.