Remove the Benchmarks section from the model card
Browse files
README.md
CHANGED
|
@@ -84,21 +84,6 @@ The pipeline and its blocks come from `modular_model_index.json`. Classifier-fre
|
|
| 84 |
|
| 85 |
**Kandinsky S2V 1.0 is not part of a released Diffusers version yet**; it lands as a modular pipeline.
|
| 86 |
|
| 87 |
-
## Benchmarks
|
| 88 |
-
|
| 89 |
-
EMTD, the standard semi-body benchmark; ↑ higher is better, ↓ lower is better,
|
| 90 |
-
best in bold. The paper has more details and more benchmarks.
|
| 91 |
-
|
| 92 |
-
| Model | Sync-C ↑ | Sync-D ↓ | CSIM ↑ | IQA ↑ | ASE ↑ | Min-max ↑ | Mean rank ↓ |
|
| 93 |
-
|---|---|---|---|---|---|---|---|
|
| 94 |
-
| **Kandinsky S2V 1.0 Pro SD (ours)** | **8.03** | **7.65** | **0.751** | 4.72 | 3.52 | **0.876** | **1.8** |
|
| 95 |
-
| MultiTalk | 7.38 | 8.04 | 0.630 | 4.52 | 3.33 | 0.423 | 4.8 |
|
| 96 |
-
| InfiniteTalk | 7.24 | 8.03 | 0.733 | 4.58 | 3.41 | 0.574 | 3.8 |
|
| 97 |
-
| LongCat 1.5 | 7.21 | 8.11 | 0.734 | 4.73 | **3.62** | 0.766 | 2.8 |
|
| 98 |
-
| OmniAvatar | 6.82 | 8.60 | 0.502 | 4.56 | 3.24 | 0.153 | 6.0 |
|
| 99 |
-
| HunyuanAvatar | 6.61 | 8.64 | 0.745 | 4.69 | 3.45 | 0.515 | 4.4 |
|
| 100 |
-
| Wan-S2V | 6.06 | 8.92 | 0.724 | **4.83** | 3.60 | 0.568 | 4.4 |
|
| 101 |
-
|
| 102 |
## Notes
|
| 103 |
|
| 104 |
Euler flow matching, 50 steps, classifier-free guidance 3.0.
|
|
|
|
| 84 |
|
| 85 |
**Kandinsky S2V 1.0 is not part of a released Diffusers version yet**; it lands as a modular pipeline.
|
| 86 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
## Notes
|
| 88 |
|
| 89 |
Euler flow matching, 50 steps, classifier-free guidance 3.0.
|