5 Commits

Author	SHA1	Message	Date
Mr-Neutr0n	9f548bce11	Fix min-SNR loss weighting: replace min() with torch.clamp() ... In VideoDiffusionLoss.__call__, the weight tensor `w` is a multi-dimensional tensor produced by append_dims(). Using Python's built-in min(w, self.min_snr_value) on a multi-dimensional tensor and a scalar does not perform element-wise clamping — it either raises an error or produces incorrect results depending on the tensor shape. Replace with torch.clamp(w, max=self.min_snr_value) to correctly apply element-wise upper-bound clamping, which is the intended behavior for the min-SNR-gamma loss weighting strategy.	2026-02-11 18:00:28 +05:30
Yuxuan Zhang	39c6562dc8	format	2025-03-22 15:14:06 +08:00
zR	b8001a769f	finetune I2V sat	2024-09-13 16:46:47 +08:00
zR	b43951addd	i2v draft	2024-09-10 07:45:24 +08:00
zR	34dd61430e	init	2024-08-06 02:12:51 +08:00