elinas

/

chronos-mistral-7b

like 3

I don't understand these models with no tokenizer.json. This is an essential file that needs to be provided with every model.
It is required, because inference engines (like TGI, vLLM, ollama, lamma-cpp-server) use it to determine the prompt template for inference.
I always have to download the JSON from your source model, and add it to your repo.
in this case, it is just simply:

https://huggingface.co/mistralai/Mistral-7B-v0.1/resolve/main/tokenizer.json?download=true -O tokenizer.json