Tested this but requires OpenAI for embeddings

[ninadpathak]

You got some crazy awesome stuff going here! But please let's make our embeddings local with ollama models.

It's easy with the nomic models on Ollama.

I tried to modify the code to use OLLAMA embeddings and succeded. But it gets stuck at the step after that:

There's a lot of hardcoded API endpoints and dependencies on GPT 3.5 turbo (let's just move to 4o or 4o mini?)

Is there a way for you to support ollama completely for chat and embeddings?

I have 2000+ notes from just 6 months and definitely see potential in this tool.

But I'd absolutely hate to send those notes to OpenAI for embedding.

[yingapple]

Thank you for your effort. Additionally, we are able to use the locally deployed model from Ollama. You just need to change the API base to:
http://localhost:11434/v1/
This is the OpenAI-compatible version provided by Ollama.
Reference: https://github.com/ollama/ollama/blob/main/docs/openai.md