Text Generation
Transformers
Safetensors
English
mistral
text-generation-inference

Running in ollama

#1
by whsinth - opened

Hi, I'm trying to get this model to run in Ollama. It seems that it is the only model out there that is training on public domain code.

I try putting it in llama.cpp's convert_hf_to_gguf.py and I get "NotImplementedError: BPE pre-tokenizer was not recognized - update get_vocab_base_pre()". The BPE pre-tokenizer "48b3395982eda712c0f3aaa4056c791b9a2db4ee474467631e77bb46dab2b235" is also not recognized.

Would it be possible to ship the model in gguf format?

Koala AI org

Hey there,

I have no idea how Ollama works (I assume it is just llama.cpp in a wrapper, since you talk about GGUF files), but in general this model is more of a "proof of concept", and as such is very limited as to what it can do, I have plans to make a proper, larger model also trained on CC0 data, but I have yet to get around to it.

I'll make sure to let you know when I do! :) (and to supply gguf files)

Sign up or log in to comment