Model "xxx" was not found on hub!

[zyh3826]

Describe the bug

I use git to push my model to the hub, my model it can be seen on the web, but when I submit it to the Leaderboard, this error happened: and I try it this way:

from transformers import AutoConfig, AutoModel, AutoTokenizer
save_dir = 'zyh3826/llama2-13b-ft-openllm-leaderboard-v1'
config = AutoConfig.from_pretrained(save_dir)
tokenizer = AutoTokenizer.from_pretrained(save_dir)

everything is ok

anyone can help me? thanks a lot

Reproduction

submit it on the openllm leaderboard

Logs

Model "zyh3826/llama2-13b-ft-openllm-leaderboard-v1" was not found on hub!

System info

None

[Wauplin]

Hey @zyh3826 , this issue seems to be related to the Open LLM Leaderboard, not the huggingface_hub library itself. It would be better to open a discussion directly on the Space community tab (here).

Regarding the issue, I think it is because your model is gated, meaning users have to accept the license to download the model. So the Open LLM Leaderboard cannot evaluate it properly. Maybe a more explicitly error message would be better though (cc @clefourrier)

[zyh3826]

Hey @zyh3826 , this issue seems to be related to the Open LLM Leaderboard, not the huggingface_hub library itself. It would be better to open a discussion directly on the Space community tab (here).

Regarding the issue, I think it is because your model is gated, meaning users have to accept the license to download the model. So the Open LLM Leaderboard cannot evaluate it properly. Maybe a more explicitly error message would be better though (cc @clefourrier)

thanks for your reply @Wauplin , I will check the gate

[clefourrier]

Closing since an issue was also opened on the Leaderboard. @Wauplin good intuition, it is indeed because the base model is gated

[Wauplin]

Thanks for taking care of it @clefourrier !

[julien-c]

question was also posted in the Forums. Please do not post same question to multiple channels @zyh3826, thanks!