Spaces:

open-llm-leaderboard

/

open_llm_leaderboard

like 10.1k

Running on CPU Upgrade

While reading the data from open-llm-leaderboard/results using this code for analysis, I noticed that the values are float("nan") instead of None, and the condition doesn't capture them. Probably nothing serious in terms of the leaderboard though :)

edit: small typo fixes

A slight change I believe would be that about 40 out of the 2000 models will have the partial average score, computed as the sum of successful metrics / 4, which is nan in the current code because the condition fails to catch and exclude it.

Hi ! Thank you for this PR ! Do you have an example of a model where this change does catch a nan ?

Closing for inactivity, feel free to reopen if needed!