torphix / talking-avatar

TODO

• Create pipeline
• Train using the given pipeline

Run

python3 train.py --use_8bit_adam --use_ema --mixed_precision=fp16 --enable_xformers_memory_efficient_attention --data_dir=./data/vox/processed --pretrained_model_name_or_path=runwayml/stable-diffusion-v1-5

Pickup point

• Train code