本地部署 mistralai/Mistral-Nemo-Instruct-2407

1. 创建虚拟环境

conda create -n fastchat python=3.10 -y
conda activate fastchat

2. 安装 fschat

git clone https://github.com/lm-sys/FastChat.git; cd FastChat
pip install --upgrade pip  # enable PEP 660 support
pip install -e .

3. 安装 transformers

# Until a new release has been made, you need to install transformers from source
pip install git+https://github.com/huggingface/transformers.git

4. 安装 flash-attn

pip install packaging
pip uninstall -y ninja && pip install ninja
pip install flash-attn --no-build-isolation

5. 安装 pytorch

pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu121

6. 启动 controller

python -m fastchat.serve.controller

7. 启动 mistralai/Mistral-Nemo-Instruct-2407

python -m fastchat.serve.model_worker --num-gpus 2 --model-names "mistral-7b,gpt-3.5-turbo,gpt-3.5-turbo-16k,gpt-4,text-davinci-003" --model-path mistralai/Mistral-Nemo-Instruct-2407 --controller http://localhost:21001 --port 31000 --worker-address http://localhost:31000 --max-gpu-memory 16Gib

8. 启动 api

python -m fastchat.serve.openai_api_server --host 0.0.0.0 --port 8000

9. 访问 mistralai/Mistral-Nemo-Instruct-2407

使用 chatbox 配置好访问信息，

问它一个问题，

完结！