search

madaan / self-refine

LLMs can generate feedback on their work, use it to improve the output, and repeat this process iteratively.
https://selfrefine.info
Apache License 2.0
596 stars 48 forks source link

Is it compatible with LLaMa 2 #12

Open SSusantAchary opened 1 year ago

SSusantAchary commented 1 year ago

Was there an attempt to test this library with LLaMa 2 model

madaan commented 1 year ago

We ran some experiments with LLaMA2-70b. Surprisingly, the results look pretty good on some of the tasks, even in an instruction-only setup (no few-shot examples). We will add detailed results shortly, but if you have access to the 70b model, I highly recommend trying Self-Refine on your tasks.