LocalLLaMA

1 readers

1 users here now

Community to discuss about Llama, the family of large language models created by Meta AI.

founded 10 months ago

MODERATORS

Is Open LLM Leaderboard reliable source ? yi:34B is at the top but I get better results with neural-chat:7B model (alien.top)

submitted 9 months ago by grigio@alien.top to c/localllama@poweruser.forum

23 comments fedilink hide all child comments

I use in both cases q4_K_M

you are viewing a single comment's thread
view the rest of the comments

[–] phree_radical@alien.top 1 points 9 months ago (1 children)

Most of the benchmarks seem to measure regurgitation of factual knowledge, which IMO everyone should accept as a misguided idea for a task, from in-weights learning, instead of testing in-context learning, which I would argue was the goal of LLM training. I'd say they are probably harmful to the cause of improving future LLMs

[–] andrewlapp@alien.top 1 points 9 months ago

I agree, and The Leaderboard's newly added DROP metric is a step in the right direction.