michaelfeil

/

ct2fast-Llama-2-13b-hf

like 1

Thanks for convert and sharing the model.

First, given the sample code of the model card, it will raise the following error:

ValueError: no suitable Tokenizer found.

Set the tokenizer using the AutoTokenizer class doesn't help either.

Hence, I used the LlamaTokenizer and it required the installation of the sentencepiece package.

Lastly, the sentencepiece package will raise the TypeError: not a string, after added the tokenizer.model file (which downloaded using the download.sh) to the model directory, finally it is working fine.

Please test it out and update the model card. I hope it save couple of productive hours of other developers time.

Best regards.