Is it compatible with LLaMa 2

[SSusantAchary]

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

[madaan]

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.