Primordial PrivateGPT - No Sentence-Transformer Model Found.

[Riley-Dunnaway]

I am working with the primordial version of PrivateGPT. I am interested in this project to establish an offline LLM that doesn’t conflict with work Firewalls and is safe for possibly sensitive data.

I have setup nearly complete, however I receive an error when ingesting while the program attempts to communicate with huggingface. I receive a message saying “no sentence-transformers model found with name sentence-transformers/all-minilm-l12-v2. creating a new one with mean pooling”

It then struggles for a minute before timing out due to my network firewall blocking requests to huggingface.

Can I adjust something in the basic setup to take the LLM 100% offline and avoid accessing the sentence transformer through API? Could i set up the LLM on a personal device to get the embedder downloaded, then transfer it to a work machine after for further 100% offline use?