Evaluation with retriever

[preritt]

Is there a reson to use questions_torchhub_0_shot.jsonl instead of questions_torchhub_bm25.jsonl when evaluating the retriever with torchub.

This is what is mentioned in the documentation for evaluation script. The arguments seem to be the same as the one with zero shot, except that the retriever is also passed.

python get_llm_responses_retriever.py --retriever bm25 --model gpt-3.5-turbo --api_key $API_KEY --output_file gpt-3.5-turbo_torchhub_0_shot.jsonl --question_data eval-data/questions/torchhub/questions_torchhub_0_shot.jsonl --api_name torchhub --api_dataset ../data/api/torchhub_api.jsonl