DPO on top of OAS
I take that Llama3-Unholy-8B-OAS was not uncensored enough? Which was on top of unholy L3, right?
How did it fares in terms of intelligence, will you submit it to Open LLM Leaderboard? I know the benchmark is not that important, but it would be interesting to see it's evolution since Unholy L3.
To be fair, I'm already amazed that it work... Hahaha
No, I don't plan to do anything with it at the moment, I'm more focused around that OAS thingy, I really want to make it work better, or at least understand how it work correctly.
But feel free to put it in some benchmark if you want! Or even run an eval yourself if you have the ressource.
To be fair, I'm already amazed that it work... Hahaha
No, I don't plan to do anything with it at the moment, I'm more focused around that OAS thingy, I really want to make it work better, or at least understand how it work correctly.
But feel free to put it in some benchmark if you want! Or even run an eval yourself if you have the ressource.
Could you maybe upload a q4 as well? Curious about this variant!