mirror of
https://github.com/THUDM/CogVideo.git
synced 2025-12-02 18:52:08 +08:00
- Integrated progress tracking with upscale model loading. - Implemented conditional latent upscaling using the upscale model. - Processed batch video frames using PyTorch and converted them to PIL images. - load_torch_file for params_ema convert weights