AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-27 17:22:53 +03:00
Code Activity

Fix incorrect tile_latent_min_width calculations (#11305)

Browse Source
This commit is contained in:
Tuna Tuncer
2025-04-13 14:51:50 +02:00
committed by GitHub
parent 97e0ef4db4
commit 36538e1135
3 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/diffusers/models/autoencoders/autoencoder_kl_hunyuan_video.py
View File

@@ -829,7 +829,7 @@ class AutoencoderKLHunyuanVideo(ModelMixin, ConfigMixin):
def _decode(self, z: torch.Tensor, return_dict: bool = True) -> Union[DecoderOutput, torch.Tensor]:
batch_size, num_channels, num_frames, height, width = z.shape
tile_latent_min_height = self.tile_sample_min_height // self.spatial_compression_ratio
tile_latent_min_width = self.tile_sample_stride_width // self.spatial_compression_ratio
tile_latent_min_width = self.tile_sample_min_width // self.spatial_compression_ratio
tile_latent_min_num_frames = self.tile_sample_min_num_frames // self.temporal_compression_ratio
if self.use_framewise_decoding and num_frames > tile_latent_min_num_frames:

2
src/diffusers/models/autoencoders/autoencoder_kl_ltx.py
View File

@@ -1285,7 +1285,7 @@ class AutoencoderKLLTXVideo(ModelMixin, ConfigMixin, FromOriginalModelMixin):
) -> Union[DecoderOutput, torch.Tensor]:
batch_size, num_channels, num_frames, height, width = z.shape
tile_latent_min_height = self.tile_sample_min_height // self.spatial_compression_ratio
tile_latent_min_width = self.tile_sample_stride_width // self.spatial_compression_ratio
tile_latent_min_width = self.tile_sample_min_width // self.spatial_compression_ratio
tile_latent_min_num_frames = self.tile_sample_min_num_frames // self.temporal_compression_ratio
if self.use_framewise_decoding and num_frames > tile_latent_min_num_frames:

2
src/diffusers/models/autoencoders/autoencoder_kl_magvit.py
View File

@@ -887,7 +887,7 @@ class AutoencoderKLMagvit(ModelMixin, ConfigMixin):
def _decode(self, z: torch.Tensor, return_dict: bool = True) -> Union[DecoderOutput, torch.Tensor]:
batch_size, num_channels, num_frames, height, width = z.shape
tile_latent_min_height = self.tile_sample_min_height // self.spatial_compression_ratio
tile_latent_min_width = self.tile_sample_stride_width // self.spatial_compression_ratio
tile_latent_min_width = self.tile_sample_min_width // self.spatial_compression_ratio
if self.use_tiling and (z.shape[-1] > tile_latent_min_height or z.shape[-2] > tile_latent_min_width):
return self.tiled_decode(z, return_dict=return_dict)