Ethanfel f99d2666e8 fix: interpolate sync_cond to match audio sequence length in transformer ...

Sync_MLP interpolates sync features based on video duration, but audio
latent length depends on the user-set audio duration. When video != audio
duration, the sequences diverge. Resample sync_cond to x's length before
the gated addition so any video/audio duration combo works.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-03-27 21:21:39 +01:00

..

configs

feat: extract prismaudio_core config and model factory

2026-03-27 17:05:57 +01:00

inference

fix: remove unused tqdm import from sampling.py

2026-03-27 18:01:29 +01:00

models

fix: interpolate sync_cond to match audio sequence length in transformer

2026-03-27 21:21:39 +01:00

__init__.py

feat: extract prismaudio_core config and model factory

2026-03-27 17:05:57 +01:00

factory.py

fix: remove MMDiTWrapper import and dead code paths from factory.py

2026-03-27 19:12:40 +01:00