Survey agent

Launch the agent server

The server uses uvicorn to launch the agent server. Install uvicorn by running the following command

pip install uvicorn

Run the following command to launch the agent server

cd code
uvicorn server:app --host 0.0.0.0 --port 8132

The frontend will send request to the 8132 port, so make sure not to change the port number.

You can run the following command to find the process occupying port number 8132 (may require sudo to see it), and kill it afterwards.

lsof -i :8132

set up openchat model https://github.com/imoneoi/openchat

pip3 install ochat
python -m ochat.serving.openai_api_server --model openchat/openchat_3.5  --engine-use-ray --worker-use-ray --tensor-parallel-size 1

You can choose the num of gpus using parameter "--tensor-parallel-size "

Tokenizer

Need model tokenizer for calculating tokens of chunk or controlling

set up arxiv-sanity

先运行 feature_func.py 生成一下 tfidf 要用的 features.p，之后运行 arxiv_sanity_func.py 就可以

set up config file

Use different models

You can choose form "openchat-3.5", "mixtral", "mistral-0.2", "vicuna-1.5", "chatglm3" Write the model name in config file

Use GPT 4

You are supposed to export OPENAI_API_KEY="your_key" before use GPT-4 model

Update arXiv paper every week day

crontab -l
# Add the following line
0 10 * * 1-5 {work_dir}/survey_agent/code/arxiv_update.sh