Update: Remember I told you about Llama 2 7b & 13b models I was fine tuning for nostr? Actually I went one step further to eliminate trust factor. Let me explain.

Once this project is complete, anyone will be able to host their own LLM on Colab for free or locally (maybe pi/pc) "low specs"

They can also mount this to their cloud storage to some .txt file, where users can keep pasting new data.

The model will automatically clean the data (as much as possible) and start training on the new information.

This way, they can fine-tune the AI and add new information according to their preferences and needs for your personal use.

It's not yet complete, here is a part of it. Hosting, training, and mounting to storage works (sometimes 😂)