problem using the factscorer function with LLaMa

[samuel-LP]

Good morning,

I was trying to use the factscorer function.

With this code :

I received this message error :

If I remove the openai_key=None in factscore = FactScorer(model_name="retrieval+llama", data_dir="/kaggle/working/"), I have this error : AssertionError: Please place your OpenAI APT Key in api.key.

I would like to know if I need to have an OpenAI key for using FactScore even if my model is LLaMa as you can see on the first picture.

Thank you!

[martiansideofthemoon]

Hi @samuel-LP, the OpenAI key is needed for the atomic fact generation step (which uses davinci-003 in both workflows).