Can I get the error log for my failed model?

#27
by heavytail - opened

Thanks for the great leaderboard.

My model adhered the guideline properly,
and I tested the model in huggingface_hub (downloaded weight) by local lm-evaluation-harness environment.
It goes well in local env, but it has been failing in leaderboard continuously.

Could you please provide me the error log for my model?
I apologize for the inconvenience.

MODEL NAME: heavytail/kullm-polyglot-12.8b-S

upstage org

Hello,

Apologies for the delayed response.
I will forward your request to our PIC.

Regards

upstage org

Hello,

The Error log is here.

ValueError: The current architecture does not support Flash Attention 2.0. Please open an issue on GitHub to request support for this architecture: https://github.com/huggingface/transformers/issues/new

I will modify the models that do not support Flash attention to be evaluated without applying Flash attention.

Thanks

@choco9966 Thank you so much for your kind assistance! It has perfectly cleared up my curiosity.

heavytail changed discussion status to closed

Sign up or log in to comment