Partially downloaded models

[djswagerman]

When I download a model and somehow the download gets interrupted, when I try to use it, llm does not give any error message and responds with newlines. Is there any way I can delete or reload a partially download model?

[mjavadhpour]

Same here.

[mjavadhpour]

I think this line may help to delete the cached but crupted model:

https://github.com/nomic-ai/gpt4all/blob/0d1edaf0294b116d281996e18e027e35b55be1b1/gpt4all-bindings/python/gpt4all/gpt4all.py#L129

So removing the ~/.cache/gpt4all/<THE_MODEL_NAME> will remove the cached model so I can try to download the model again.

In my case prefixing command with SSL_CERT_DIR=/etc/ssl/certs successfully download the model:

SSL_CERT_DIR=/etc/ssl/certs llm -m ggml-all-MiniLM-L6-v2-f16 'Hello World!' 

I used llm-gpt4all plugin.

[djswagerman]

Thanks, that helped!

[doubleplusplus]

I tried to download the gpt4all models but got this error: ValueError: Request failed: HTTP 404 Not Found

[boldandbusted]

I'd just like a way to delete a model, period. :)