Spaces:
Paused
Paused
| # Common | |
| einops # Tensor operations | |
| # Training | |
| torch==2.4.0 # Torch | |
| omegaconf # Configuration system. | |
| lpips # LPIPS model for VAE training | |
| # Dataloading | |
| bson # bson for Image dataloading | |
| tensorflow # Video dataloading | |
| opencv-python # OpenCV | |
| sentencepiece # For Text Encoder | |
| # Modeling | |
| diffusers # Diffusers | |
| rotary-embedding-torch # Rotary positional embedding | |
| transformers # Transformers | |
| torchvision==0.19.0 # Torchvision | |
| tiktoken # Tiktoken for generation | |
| transformers_stream_generator #LLM generation support | |
| # Metrics | |
| torchmetrics # Core module for metric | |
| pycocotools # COCO-related | |
| torch-fidelity # FID-related | |
| # Experiment Tracking | |
| moviepy # WandB Logging Image & Video | |
| imageio # WandB Logging Image & Video | |
| tabulate # Logging Table | |
| deepdiff # Find difference of config | |
| # Testing | |
| parameterized # Define multiple tests through decorators. | |
| numpy<2.0 | |
| # Notebook | |
| mediapy # Notebook Visualization | |
| # DevOPs | |
| black # Code formatting | |
| flake8 # Code style | |
| isort # Import sorting | |
| pre-commit # Pre-commit hooks | |
| expecttest # Pytorch dist tests | |
| hypothesis # Fix randomness | |
| av # Process audio and video |