Fixed tokenizer

#6
by bullerwins - opened

Hi!

Would this models need to be requantized for the fixed tokenizer?
https://huggingface.co/meta-llama/Meta-Llama-3.1-8B-Instruct/discussions/28/files

Hugging Quants org

Hi here @bullerwins thanks for flagging, in this case there's no need to re-quantize the model weights, but to update the tokenizer to make sure we're aligned, I'll do so and close this issue when done! 🤗

Thanks/Gracias Álvaro!

Sign up or log in to comment