Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model
active 2024-12-30 → 2026-05-14 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 501 days from 2024-12-30 to 2026-05-14. Pushes: 47 total, peak 8 in a day. Pull requests: 15 total, peak 2 in a day. Issues: 105 total, peak 7 in a day. Comments: 250 total, peak 19 in a day. Stars: 931 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 |
|---|---|---|---|---|
| LiewFeng | 204 | 47 | 9 | 118 |
| zishen-ucap | 18 | 0 | 3 | 13 |
| spawner1145 | 13 | 0 | 3 | 10 |
| FurkanGozukara | 8 | 0 | 0 | 8 |
| nighting0le01 | 7 | 0 | 0 | 5 |
| Oruli | 6 | 0 | 0 | 4 |
| phyllispeng123 | 6 | 0 | 0 | 4 |
| cellzero | 6 | 0 | 0 | 6 |
| YunjieYu | 6 | 0 | 0 | 4 |
| zgcmx | 5 | 0 | 0 | 1 |
| jffu | 5 | 0 | 0 | 3 |
| fffanty | 5 | 0 | 0 | 3 |
| rlsu9 | 5 | 0 | 0 | 3 |
| kijai | 4 | 0 | 0 | 4 |
| syguan96 | 4 | 0 | 0 | 4 |
| asahni04 | 4 | 0 | 0 | 3 |
| cchance27 | 4 | 0 | 0 | 3 |
| Xzh-lab | 4 | 0 | 0 | 3 |
| trouble-maker007 | 3 | 0 | 0 | 1 |
| Zctoylm0927 | 3 | 0 | 0 | 1 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#94quic-amitraj2026-01-25 15:14how to get a model's coefficients param?
- Issue comment#98mr-september2026-01-18 03:45cannot import name 'precompute_freqs_cis' from 'comfy.ldm.lightricks.model'
- Issue#98zhuhw192026-01-07 00:43cannot import name 'precompute_freqs_cis' from 'comfy.ldm.lightricks.model'
- Issue#97BienLuky2025-12-08 10:16Wan2.2 support
- Issue comment#96BenDes212025-11-30 10:43z-image support
- Issue#94hufangjian2025-10-23 06:23how to get a model's coefficients param?
- Issue#92mali-afridi2025-09-12 01:20Teacache with MultiGPU (FSDP) for Wan2.1
- Issue#91MathamPollard2025-09-10 12:16how to get the rescale coefficient of new model?
- Issue#90oagrawal2025-08-29 20:10For the Wan2.1 implementation, it seems to be accumulating TEMNI
- Issue#89ITerydh2025-08-24 11:17when wan2.2 can support
- Issue comment#88eaeful2025-08-08 17:30qwen-image support
- Issue comment#63bookkkker2025-08-07 06:27Is there an example in wan2.1 to add teacache based on xdit usp parallel computing?
- Issue comment#48bookkkker2025-08-07 03:36Can I run TeaCache on multiple GPUs?
- Issue comment#63bookkkker2025-08-07 03:08Is there an example in wan2.1 to add teacache based on xdit usp parallel computing?
- Issue comment#88joelulu2025-08-07 02:39qwen-image support
- Issue#88zoumaguanxin2025-08-06 07:38qwen-image support
- Issue#87Tomer-Solberg2025-07-30 10:03Can TeaCache support LTX video 13B?
- Issue#85xiongjingwu2025-07-24 14:00Is there a plan to support musetalk? Or how can I achieve it? Which layer of results do I need to cache?
- Issue comment#84AuZhoomLee2025-07-24 11:03got "The size of tensor a (xxx) must match the size of tensor b (xxx) at non-singleton dimension 0" error when i use rescale to calculate the rel_l1_distance.
- Issue#83buptmengjj2025-07-21 01:40Floating point exception (core dumped)?
- Issue comment#63myliu-hub2025-07-18 06:50Is there an example in wan2.1 to add teacache based on xdit usp parallel computing?
- Issue comment#63myliu-hub2025-07-18 06:50Is there an example in wan2.1 to add teacache based on xdit usp parallel computing?
- Issue#82philipwan2025-07-16 04:07Will support SD3(.5)
- Issue#74liufeiluo2025-07-04 03:36Can it support Wan2.1 FLF2V
- Issue comment#72tuolaku2025-07-01 07:28TeaCache support for Chroma model.
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 931 stars here means stars gained during the window, not the repo's star count.