castorini / rank_llm

Repository for prompt-decoding using LLMs (GPT3.5, GPT4, Vicuna, and Zephyr)
http://rankllm.ai
Apache License 2.0
277 stars 36 forks source link

create a result class #35

Closed sahel-sh closed 5 months ago

sahel-sh commented 6 months ago

Currently, retrieved_results and rerank_results are dictionaries that get saved as json. communication between retrieve and reranker happens by reading the data from the file. We can create a results class to capture this and additinal info, prompts and responses, token counts, etc all are written in files right now.

sahel-sh commented 5 months ago

https://github.com/castorini/rank_llm/pull/54 and https://github.com/castorini/rank_llm/pull/55 fixes this