Closed vince62s closed 1 year ago
I want to do the same thing, do you have any idea?
It's up to the Meta team @glample to release or not. Would be great though.
We are considering to also release things like our eval platform but for now I'd recommend the OpenLLM leaderboard from HF - https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard
It would be great to be able to reproduce the results provided in the paper.
the zero-shot ones but also the other tables with few-shots.
Can this be released ?