[ICLR 2026 Oral] DiffusionNFT: Online Diffusion Reinforcement with Forward Process
active 2025-09-22 → 2026-06-25 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 277 days from 2025-09-22 to 2026-06-25. Pushes: 2 total, peak 1 in a day. Pull requests: 0 total, peak 0 in a day. Issues: 33 total, peak 5 in a day. Comments: 44 total, peak 6 in a day. Stars: 295 total, peak 40 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| ChenDRAG | 21 | 0 | 0 | 9 |
| zhengkw18 | 19 | 2 | 0 | 17 |
| zhangquanwei962 | 3 | 0 | 0 | 3 |
| Jayce-Ping | 3 | 0 | 0 | 0 |
| MengHao666 | 3 | 0 | 0 | 3 |
| jaylee2000 | 3 | 0 | 0 | 1 |
| Student-of-Holmes | 2 | 0 | 0 | 1 |
| siruizhang30 | 2 | 0 | 0 | 0 |
| LAW1223 | 2 | 0 | 0 | 1 |
| YibooZhao | 2 | 0 | 0 | 0 |
| TimZhang001 | 2 | 0 | 0 | 0 |
| happynear | 2 | 0 | 0 | 1 |
| littletomatodonkey | 1 | 0 | 0 | 0 |
| carmysteve-lgtm | 1 | 0 | 0 | 0 |
| Dominoer | 1 | 0 | 0 | 0 |
| Henglin-Liu | 1 | 0 | 0 | 1 |
| HiSultryMan | 1 | 0 | 0 | 1 |
| huangxu1991 | 1 | 0 | 0 | 1 |
| Fuzm7 | 1 | 0 | 0 | 1 |
| beaglew | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue#7jaylee20002026-05-25 03:46xFormers can't load C++/CUDA extensions
- Issue comment#35Henglin-Liu2026-03-30 03:13About the Pickscore performance
- Issue comment#46zhengkw182026-03-04 10:59Implicit positive/negative velocities collapse
- Issue comment#45zhengkw182026-03-04 10:09DiffusionNFT Training on Few-Step Distilled Diffusion Models Becomes Progressively Blurry and Collapses: Instability vs. Loss Mismatch?
- Issue#40Jayce-Ping2026-02-10 12:00Huge congratulations on ICLR 2026 Oral!
- Issue comment#40zhengkw182026-02-10 11:18Huge congratulations on ICLR 2026 Oral!
- Issue#40Jayce-Ping2026-02-10 06:07Huge congratulations on ICLR 2026 Oral!
- Issue comment#39zhengkw182026-02-08 14:55Does this framework currently only support SD3.5?
- Issue#24ChenDRAG2026-02-06 12:25About negative-aware
- Issue comment#38zhengkw182026-02-05 14:16Clarification on Evaluation Datasets for Table 1 (PickScore, HPSv2.1, etc.)
- Issue#38carmysteve-lgtm2026-01-26 06:57Clarification on Evaluation Datasets for Table 1 (PickScore, HPSv2.1, etc.)
- Issue comment#36Fuzm72026-01-21 02:58Question about the specific pipelines of Multi-Reward joint training.
- Issue comment#37zhengkw182026-01-20 10:41Questions about training
- Issue#37LAW12232026-01-16 12:02Questions about training
- Issue comment#33zhengkw182025-12-29 09:44about no cfg training
- Issue comment#33zhangquanwei9622025-12-29 03:35about no cfg training
- Issue comment#33zhangquanwei9622025-12-29 03:25about no cfg training
- Issue comment#33zhengkw182025-12-26 04:26about no cfg training
- Issue comment#33zhangquanwei9622025-12-26 04:22about no cfg training
- Issue comment#33zhengkw182025-12-24 15:02about no cfg training
- Issue#34beaglew2025-12-24 07:47ModuleNotFoundError: No module named 'mmcv._ext'
- Issue#18ChenDRAG2025-12-22 13:03NFT is work for DDPM, I have tested. But I have some question. please help me.
- Issue#26ChenDRAG2025-12-22 13:02using samples of other models or even gt samples to form the G samples
- Issue#28ChenDRAG2025-12-22 13:02Policy Loss sometimes value equal zero(0)
- Issue#29ChenDRAG2025-12-22 13:02Questions about the stochastic exploration
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 295 stars here means stars gained during the window, not the repo's star count.