Improving Text-to-Image Generation with Intrinsic Self-Confidence Rewards
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911668758380544 |
|---|---|
| author | Kim, Seungwook Cho, Minsu |
| author_facet | Kim, Seungwook Cho, Minsu |
| contents | Text-to-image generation powers content creation across design, media, and data augmentation. Post-training of text-to-image generative models is a promising path to improve human preference alignment, factuality, and aesthetics. We introduce SOLACE (Self-Originating LAtent Confidence Estimation), a post-training framework that replaces external reward supervision with an internal self-confidence signal: we re-noise the model's own outputs and measure how accurately it recovers the injected noise, treating low reconstruction error as high self-confidence. SOLACE converts this intrinsic signal into scalar rewards for reinforcement learning, requiring no external reward models, annotators, or preference data. By reinforcing high-confidence generations, SOLACE delivers consistent gains in compositional generation, text rendering, and text-image alignment. Integrating SOLACE with external rewards yields complementary improvements while alleviating reward hacking. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_00918 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Improving Text-to-Image Generation with Intrinsic Self-Confidence Rewards Kim, Seungwook Cho, Minsu Computer Vision and Pattern Recognition Artificial Intelligence Text-to-image generation powers content creation across design, media, and data augmentation. Post-training of text-to-image generative models is a promising path to improve human preference alignment, factuality, and aesthetics. We introduce SOLACE (Self-Originating LAtent Confidence Estimation), a post-training framework that replaces external reward supervision with an internal self-confidence signal: we re-noise the model's own outputs and measure how accurately it recovers the injected noise, treating low reconstruction error as high self-confidence. SOLACE converts this intrinsic signal into scalar rewards for reinforcement learning, requiring no external reward models, annotators, or preference data. By reinforcing high-confidence generations, SOLACE delivers consistent gains in compositional generation, text rendering, and text-image alignment. Integrating SOLACE with external rewards yields complementary improvements while alleviating reward hacking. |
| title | Improving Text-to-Image Generation with Intrinsic Self-Confidence Rewards |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2603.00918 |