Anime Flow Matching
Rectified/conditional flow matching. Predicts velocity v = x1 - x0 along the straight path xt = (1-t)x0 + tx1, where x0 is noise and x1 is data.
Model source
- Kind
- Other
- File size
- 250 MB
- Version
- vFm anime
- Updated
- Oct 9, 2026
- Source
- Hugging Face
Source description
Flow Matching — anime faces 64x64
Rectified/conditional flow matching. Predicts velocity v = x1 - x0 along the
straight path xt = (1-t)*x0 + t*x1, where x0 is noise and x1 is data.
- UNet: ch=64, mults=(1,2,2,4), attn @16 — ~16.3M params
- Trained: 30 epochs, final FM loss 0.2059
- Sampling: Euler ODE,
x += v(x,t)*dt, ~10 steps is enough - Checkpoint contains
net,ema,opt,scaler,losses— resumable.
Use the ema weights for sampling.
Sources
1 source
Sources
1 sourceVerified Oct 9
Model artifacts
1 artifact
Model artifacts
1 artifactSource excerpts
2 excerpts
Source excerpts
2 excerpts--- tags: [flow-matching, generative, anime, unconditional-image-generation] --- # Flow Matching — anime faces 64x64 Rectified/conditional flow matching. Predicts velocity `v = x1 - x0` along the straight path `xt = (1-t)*x0 + t*x1`, where `x0` is noise and `x1` is data. - **UNet**: ch=64, mults=(1,2,2,4), attn @16 — ~16.3M params - **Trained**: 30 epochs, final FM loss 0.2059 - **Sampling**: Euler ODE, `x += v(x,t)*dt`, ~10 steps is enough - Checkpoint contains `net`, `ema`, `opt`, `scaler`, `losses` — resumable. Use the `ema` weights for sampling.
Source context: 0 downloads · 0 likes · Pipeline unconditional-image-generation · Repo AnasKAN/anime-flow-matching
