Command Line mistral-chat performs better than ChatCompletionRequest on the same text ?

#58
by maniiii - opened

Are there any settings which makes it perform better or Is it taking into account the previous conversational context ?Even with changing prompt with formatting ChatCompletionRequest tends to give lower accuracy.

Mistral AI_ org

Hi maniiii! Thats odd, you can take a look at the implementation of mistral-chat here: https://github.com/mistralai/mistral-inference/blob/main/src/mistral_inference/main.py and it uses ChatCompletionRequest, could you provide an example?

Sign up or log in to comment