- Remove redundant comments and debug information
- Adjust default parameters in training scripts
- Clean up code in lora_trainer and trainer implementations
- Add DeepSpeed ZeRO-3 configuration support
- Optimize memory usage during training
- Rename training scripts to reflect ZeRO usage
- Update related configuration files and trainers
- Add SFT (Supervised Fine-Tuning) trainers for all model variants:
- CogVideoX I2V and T2V
- CogVideoX-1.5 I2V and T2V
- Add DeepSpeed ZeRO configuration files:
- ZeRO-2 with and without CPU offload
- ZeRO-3 with and without CPU offload
- Add base accelerate config for distributed training
- Update trainer.py to support SFT training mode
This enables full-parameter fine-tuning with memory-efficient distributed training using DeepSpeed ZeRO optimization.
When loading videos with fewer frames than max_num_frames, repeat the last
frame to reach the required length instead of failing. This ensures consistent
tensor dimensions across the dataset while preserving as much original video
content as possible.
Add a table in README files showing hardware requirements for training
different CogVideoX models, including:
- Memory requirements for each model variant
- Supported training types (LoRA)
- Training resolutions
- Mixed precision settings
Updated in all language versions (EN/ZH/JA).
Before precomputing the latent cache and text embeddings, cast the VAE and
text encoder to the target training dtype (fp16/bf16) instead of keeping them
in fp32. This reduces memory usage during the precomputation phase.
The change occurs in prepare_dataset() where the models are moved to device
and cast to weight_dtype before being used to generate the cache.
When training i2v models without specifying image_column, automatically extract
and use first frames from training videos as conditioning images. This includes:
- Add load_images_from_videos() utility function to extract and cache first frames
- Update BaseI2VDataset to support auto-extraction when image_column is None
- Add validation and warning message in Args schema for i2v without image_column
The first frames are extracted once and cached to avoid repeated video loading.
- Add pipe.remove_all_hooks() after validation to prevent memory leaks
- Clean up validation pipeline properly to avoid potential issues in subsequent training steps
This change enables caching of prompt embeddings in the CogVideoX text-to-video
LoRA trainer, which can improve training efficiency by avoiding redundant text
encoding operations.
- Add validation to ensure (frames - 1) is multiple of 8
- Add specific resolution check (480x720) for cogvideox-5b models
- Add error handling for invalid resolution format
- Add caching for prompt embeddings
- Store cached files using safetensors format
- Add cache directory structure under data_root/cache
- Optimize memory usage by moving tensors to CPU after caching
- Add debug logging for cache hits
- Add info logging for cache writes
The caching system helps reduce redundant computation and memory usage during training by:
1. Caching prompt embeddings based on prompt text hash
2. Caching encoded video latents based on video filename
3. Moving tensors to CPU after caching to free GPU memory
Add two shell scripts to simplify model training:
- accelerate_train_i2v.sh: Launch script for Image-to-Video training
- accelerate_train_t2v.sh: Launch script for Text-to-Video training
Both scripts provide comprehensive configurations for:
- Model settings
- Data pipeline
- Training parameters
- System resources
- Checkpointing
- Validation
- Add field validators for model type and validation settings
- Implement command line argument parsing with argparse
- Add type hints and documentation for training parameters
- Support configuration of model, training, and validation parameters
Add and refactor trainers for CogVideoX model variants:
- Implement CogVideoXT2VLoraTrainer for text-to-video generation
- Refactor CogVideoXI2VLoraTrainer for image-to-video generation
Both trainers support LoRA fine-tuning with proper handling of:
- Model components loading and initialization
- Video encoding and batch collation
- Loss computation with noise prediction
- Validation step for generation
Add Pydantic models to handle:
- CLI arguments and configuration (Args)
- Model components and pipeline (Components)
- Training state and parameters (State)
- Replace bucket-based dataset with simpler resize-based implementation
- Add video latent precomputation during dataset initialization
- Improve code readability and user experience
- Remove complexity of bucket sampling for better maintainability
This change makes the codebase more straightforward and easier to use while
maintaining functionality through resize-based video processing.
