Replying to Avatar iefan šŸ•Šļø

Ok, Iā€™m currently running this model on my phone with one-click deployment. Iā€™m also discussing fine-tuning with my friend.

Do you want an open-source AI assistant on the Nostrnet.work dashboard? I could train it using the Nostr project GitHub repos during fine-tuning.

First, I will share my Colab notebook today, so you all can test it out. nostr:note1ju22sjyahvnmp9wrcfz6apgwj0v46ymn9c5h3em2dxntnlm3q9lqenmqg4
Avatar
cloud fodder 2y ago

This is really really cool. What are you training it with? Would a dataset from a relay be useful, jus lmk. Excited to see where this goes. šŸŒ±šŸŒ“šŸļøšŸ¦œ

Reply to this note

Please Login to reply.

Discussion

Avatar
iefan šŸ•Šļø 2y ago

Currently it's using Llama 2, 16 Billion parameter model. It's already extremely smart, we just have to fine tune it with nostr related data.

https://ai.meta.com/llama/