AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-29 07:22:12 +03:00
Code Activity

Fix pipeline return bugs

Browse Source
This commit is contained in:
Daniel Gu
2025-12-23 08:49:11 +01:00
parent 3bf736979f
commit fa7d9f77f1
2 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
scripts/ltx2_test_full_pipeline.py
View File

@@ -191,6 +191,7 @@ def main(args):
guidance_scale=args.guidance_scale,
generator=torch.Generator(device=args.device).manual_seed(args.seed),
output_type="np",
return_dict=False,
)
# Convert video to uint8 (but keep as NumPy array)

4
src/diffusers/pipelines/ltx2/pipeline_ltx2.py
View File

@@ -1033,7 +1033,7 @@ class LTX2Pipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraLoaderMix
# NOTE: currently, unlike the video VAE, we denormalize the audio latents inside the audio VAE decoder's
# decode method
generated_mel_spectrograms = self.audio_vae.decode(audio_latents, return_dict=False)[0]
waveforms = self.vocoder(generated_mel_spectrograms)
audio = self.vocoder(generated_mel_spectrograms)
# Offload all models
self.maybe_free_model_hooks()
@@ -1041,4 +1041,4 @@ class LTX2Pipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraLoaderMix
if not return_dict:
return (video, audio)
return LTX2PipelineOutput(frames=video, audio=waveforms)
return LTX2PipelineOutput(frames=video, audio=audio)