Hacker News Reader

Nevin1901

I fine tuned LLama-2 on code/comment generation (in python) for around $2 and was able to run it natively on an m1 macbook air. I can totally see smaller fine tuned LLM's being used locally on consumer devices in the future. I think people underestimate how cheap and efficient this stuff is.

I've actually built a service which lets you fine tune LLama-2/other llms by uploading a JSON dataset. I'm looking for feedback, the link is https://useftn.com.

View on HN

I've been playing with llama-2, and I've been pleasantly surprised with its ability to process images. So you plan to offer image inputs on your fine tuning service?

That sounds interesting. Do you have more info on how llama2 can process images?

I've been using this [0] locally, and this model [1] on replicate.

[0] https://github.com/ggerganov/llama.cpp

[1] https://replicate.com/yorickvp/llava-13b/versions/6bc1c7bb0d...

What is the cost for a useful run? $2.4/Mb sounds good, but how many megabytes did you need for example to do your fine tuning to your satisfaction.

Reply on news.ycombinator.com