X-rayted
#1
by MrDevolver - opened
It's a shame there are no E/RP finetunes of these uncensored models. The models are no longer refusing, so all you need to do is push them a little bit and I believe further E/RP training would be just that extra push they need to do it automatically without user pushing them.
The GPT-OSS are an "odd" model to train (relative to other models). Recent updates at Unsloth may mean more of these fine tunes in the near future.
The GPT-OSS are an "odd" model to train (relative to other models). Recent updates at Unsloth may mean more of these fine tunes in the near future.
You're hopefully right about that. I really want to see a good finetune of this model, because it's definitely not stupid, just needs some guidance in the right direction.