A multi-step reasoning workflow for philosophical Q&A based on DeepSeek.
Now support Traditional Chinese.
Set up the required dependencies:
python3 -m pip install openai
Configure the DeepSeek official API key by setting the API_KEY variable in llm_client.py.
In the main program main.py, populate the topics variable with one or more questions in list format, then run the following command:
python3 main.py
The final results and intermediate reasoning steps will be stored in the output folder. The final output for each question will be saved in the stage3/self_expression.txt file within a folder named after the question.
python3 -m pip install pytest pytest-mock openai
python3 -m pytest
