Connections between reinforcement learning with feedback,test-time scaling, and diffusion guidance: An anthology
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866911138520760320 |
|---|---|
| author | Jiao, Yuchen Chen, Yuxin Li, Gen |
| author_facet | Jiao, Yuchen Chen, Yuxin Li, Gen |
| contents | In this note, we reflect on several fundamental connections among widely used post-training techniques. We clarify some intimate connections and equivalences between reinforcement learning with human feedback, reinforcement learning with internal feedback, and test-time scaling (particularly soft best-of-$N$ sampling), while also illuminating intrinsic links between diffusion guidance and test-time scaling. Additionally, we introduce a resampling approach for alignment and reward-directed diffusion models, sidestepping the need for explicit reinforcement learning techniques. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_04372 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Connections between reinforcement learning with feedback,test-time scaling, and diffusion guidance: An anthology Jiao, Yuchen Chen, Yuxin Li, Gen Machine Learning General Literature Statistics Theory In this note, we reflect on several fundamental connections among widely used post-training techniques. We clarify some intimate connections and equivalences between reinforcement learning with human feedback, reinforcement learning with internal feedback, and test-time scaling (particularly soft best-of-$N$ sampling), while also illuminating intrinsic links between diffusion guidance and test-time scaling. Additionally, we introduce a resampling approach for alignment and reward-directed diffusion models, sidestepping the need for explicit reinforcement learning techniques. |
| title | Connections between reinforcement learning with feedback,test-time scaling, and diffusion guidance: An anthology |
| topic | Machine Learning General Literature Statistics Theory |
| url | https://arxiv.org/abs/2509.04372 |