[docs]: use the FastH3-QAD-v2 name consistently
Browse files
README.md
CHANGED
|
@@ -3,7 +3,7 @@ license: other
|
|
| 3 |
base_model: FastVideo/FastVideo-FastH3-8-Step-V2
|
| 4 |
tags: [video-generation, text-to-video, audio, fastvideo, nvfp4, qad]
|
| 5 |
---
|
| 6 |
-
# FastH3-QAD-v2
|
| 7 |
|
| 8 |
FastH3-QAD-v2 is an 8-step video-and-audio model with NVFP4 weights, trained with **quantization-aware distillation
|
| 9 |
(QAD)**. It runs on one Blackwell GPU (RTX 5090, RTX PRO 6000) or a DGX Spark.
|
|
@@ -26,7 +26,7 @@ model attends to, not just the arithmetic of attention.
|
|
| 26 |
|
| 27 |
Error against the BF16 teacher on 30 held-out prompts (relative L2 of the x0 prediction, lower is better):
|
| 28 |
|
| 29 |
-
| Metric |
|
| 30 |
|---|---:|
|
| 31 |
| Mean (video + audio) | 0.351 |
|
| 32 |
| Video | 0.382 |
|
|
|
|
| 3 |
base_model: FastVideo/FastVideo-FastH3-8-Step-V2
|
| 4 |
tags: [video-generation, text-to-video, audio, fastvideo, nvfp4, qad]
|
| 5 |
---
|
| 6 |
+
# FastH3-QAD-v2, 8-step, NVFP4 for one GPU
|
| 7 |
|
| 8 |
FastH3-QAD-v2 is an 8-step video-and-audio model with NVFP4 weights, trained with **quantization-aware distillation
|
| 9 |
(QAD)**. It runs on one Blackwell GPU (RTX 5090, RTX PRO 6000) or a DGX Spark.
|
|
|
|
| 26 |
|
| 27 |
Error against the BF16 teacher on 30 held-out prompts (relative L2 of the x0 prediction, lower is better):
|
| 28 |
|
| 29 |
+
| Metric | FastH3-QAD-v2 |
|
| 30 |
|---|---:|
|
| 31 |
| Mean (video + audio) | 0.351 |
|
| 32 |
| Video | 0.382 |
|