Spaces:

open-llm-leaderboard

/

open_llm_leaderboard

like 11.8k

Running on CPU Upgrade

Hi all,
I would like to know the decoding parameters used when generating outputs from the LLMs.
Were they kept as default, what was the max_new_token length.

My struggle with OS LLMs is that they repeat content too often and do not terminate generation at appropriate internals.
It would be good to learn more about the parameters used during evaluations.

Thanks!