Datasets:

open-llm-leaderboard-old

/

requests

like 20

oh the problem in this case was on my side sorry... a file seems to be missing. it's just that another submission didn't work either.

hmm... the other one should still be interesting. I have the suspicion it's because I used a modified llama3 model.

https://huggingface.co/datasets/open-llm-leaderboard/requests/commit/73a39439d2a74acceec14d904ac7ef0f5f2840c3

I didn't use the regular llama3-8b, but this one:

https://huggingface.co/imone/Llama-3-8B-fixed-special-embedding

I would not think that this has an impact, it's just the only thing I know which is different.

oh I redid the request for my mixtral model, it failed again

https://huggingface.co/datasets/open-llm-leaderboard/requests/commit/a54455736079ab9efb1754d28aaf3b6f48489f8e

Hi @KnutJaegersberg ,

I resubmitted this your model. Please, open a discussion here if you need help with other models

I think I've now had like 4 models in a row where the eval didn't work, with different architectures. It's weird.

https://huggingface.co/datasets/open-llm-leaderboard/requests/commit/de66a17b0866fcefa8d8d4c770f45e7ad87c938c

Hi @KnutJaegersberg ,

Could you please send a list of the requests files of all your models that failed? I'll check the logs and resubmit then