Hugging Face's logo

Spaces:
Intel
/
low_bit_open_llm_leaderboard
Running

App Files Community
9

My models failed

#8
by Quazim0t0 - opened
Discussion
Quazim0t0

My models failed the quantization, and two of the attempts we're because I forgot to set "trust_remote_code=True" in my model's config. The 3rd attempt was seeing if it would work but it failed too. Is there any way I can get some information on how I might be able to get this to work?

Quazim0t0

I have tried all settings at this point. I will stop attempting to try.
Quazim0t0 changed discussion status to closed
Quazim0t0 changed discussion status to open
lvkaokao
 Intel org

@Quazim0t0 Hi, sorry for the late reply. The error occurs because the tokenizer cannot be found. I noticed that your model card mentions using SpikeTokenizer, but I couldn’t find any corresponding implementation or files in the model repository. Setting trust_remote_code=True also does not resolve or locate this tokenizer.
Quazim0t0

@lvkaokao Hey! Thanks for getting back to me. Oh wow, you're totally right, I wouldn't have noticed unless you pointed that out. Thank you so much for that, will I be able to resubmit the models once I get that sorted?
Quazim0t0

Just fixed it now.
Quazim0t0

I tried again and it still didn't make it through.

Sign up or log in to comment