Llm Re-rankers Performing Better Than Cross Encoders

Question

@Logan M aside from token cost... have you notice llm re-rankers performing better than cross encoders?

Logan M · Answer

eh yea, usually LLMs will be better (although this also depends on the LLM you are using lol)

LLM rerankers can also fail though too (outputting some output that can't be parsed). They are also much slower

Find answers from the community

Llm Re-rankers Performing Better Than Cross Encoders