Register Tokens Prove Effective in DiT
serrjoa · x · 2026-07-17
This post recaps a research observation regarding Diffusion Transformers (DiT):
- Previously in Vision Transformers, adding register tokens was found to fix certain artifacts; however, in DiT, the author initially expected these tokens to be unhelpful.
- The actual results were the opposite: even adding just a few empty tokens dropped the FID of pixel-space DiT from 3.52 to 2.69.
- Conclusion: register tokens can also be effective in DiT, significantly improving the generation quality of pixel-space models.
Related event: Register Tokens Prove Effective in Pixel-Space DiT(2 posts)→
More from Research
- SUFLECA shows NOC-based correspondence can improve CAD-to-image alignment — ducha_aiki · 2026-07-21
- OpenAI-style autonomous researchers could become real scientific collaborators — Promptmethus · 2026-07-21
- Soft Clamp cuts tool-call overuse in multi-teacher distillation, from 13.7% to 9.0% — antgroup · 2026-07-21
- ShotPlan adds learnable planning tokens for cinematic multi-shot video generation — Tele-AI · 2026-07-21
- A silicon photonic reservoir chip compensates fiber distortion in real time at 28 Gbps — bravo_abad · 2026-07-21
- A developer maps out six design rules for CLIs that humans and AI agents can both use — yujiezha · 2026-07-21