https://github.com/mihirp1998/AlignProp

AlignProp uses direct reward backpropogation for the alignment of large-scale text-to-image diffusion models. Our method is 25x more sample and compute efficient than reinforcement learning methods (PPO) for finetuning Stable Diffusion
https://github.com/mihirp1998/AlignProp

alignment diffusion-models reinforcement-learning stable-diffusion text-to-image

Last synced: about 2 months ago
JSON representation

AlignProp uses direct reward backpropogation for the alignment of large-scale text-to-image diffusion models. Our method is 25x more sample and compute efficient than reinforcement learning methods (PPO) for finetuning Stable Diffusion

• Host: GitHub
• URL: https://github.com/mihirp1998/AlignProp
• Owner: mihirp1998
• License: mit
• Created: 2023-10-06T00:03:26.000Z (12 months ago)
• Default Branch: main
• Last Pushed: 2024-03-14T03:47:33.000Z (6 months ago)
• Last Synced: 2024-05-15T23:58:45.185Z (4 months ago)
• Topics: alignment, diffusion-models, reinforcement-learning, stable-diffusion, text-to-image
• Language: Python
• Homepage: https://align-prop.github.io/
• Size: 3.86 MB
• Stars: 189
• Watchers: 7
• Forks: 7
• Open Issues: 8
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-diffusion-categorized - [Code