Fine-Tuning with Unsloth does not create any Modelfile

[dave-espinosa]

Hello everyone. I am following the Tutorial: How to Finetune Llama-3 and Use In Ollama, but when running the comand:

model.save_pretrained_gguf("gguf_model", tokenizer, quantization_method = ["q4_k_m"])

The Modelfile is never created:

Same thing happens when this command is run:

model.push_to_hub_gguf(
    "daveespinosa-qs/gguf_model", 
    tokenizer, 
    quantization_method = ["q4_k_m",], 
    token = "hf_myHFtoken", 
    private=True
)

Which gets only:

I have seen that the tutorial mentioned above includes this workaround already, so I am not sure what else it could be.

What could be going on here?

BTW, I am running that notebook from a GCP Vertex AI Jupyter Notebook.

[Coder3333]

I have been using this _ollama_modelfile method to generate a modelfile, while the model and tokenizer are loaded.

print(tokenizer._ollama_modelfile)