النوع
Checkpoint
وقت النشر
2023-06-30
النموذج الأساسي
Other
مقدمة الإصدارة
"RMHF" method allowes me to interpolate dozen of models at a time.
This model is a mix of:
(The algorithm is basically like generating potential new merging ratio vector, and show user image comparison and let user choose better one. I named it "Reinforcement Merging with Human Feedback)
