modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "minimax_h3/MiniMax-H3-Pruned-Ref2VA/*" --local_dir ./data/diffsynth_example_dataset # stage 1 (data process) accelerate launch examples/minimax_h3/model_training/train.py \ --dataset_base_path data/diffsynth_example_dataset/minimax_h3/MiniMax-H3-Pruned-Ref2VA \ --dataset_metadata_path data/diffsynth_example_dataset/minimax_h3/MiniMax-H3-Pruned-Ref2VA/metadata.json \ --data_file_keys "video,input_audio,references" \ --extra_inputs "input_audio,references" \ --height 480 \ --width 832 \ --num_frames 124 \ --dataset_repeat 1 \ --model_id_with_origin_paths "MiniMax/MiniMax-H3:Ref2VA/text_encoder/model*.safetensors,MiniMax/MiniMax-H3:Ref2VA/video_vae/source/model.safetensors,MiniMax/MiniMax-H3:Ref2VA/audio_vae/model.safetensors" \ --processor_path "MiniMax/MiniMax-H3:Ref2VA/processor/" \ --learning_rate 1e-5 \ --num_epochs 1 \ --remove_prefix_in_ckpt "pipe.dit." \ --output_path "./models/train/MiniMax-H3-Ref2VA-full-split-cache" \ --trainable_models "dit" \ --use_gradient_checkpointing \ --task "sft:data_process" # stage 2 (train) accelerate launch --config_file examples/minimax_h3/model_training/full/accelerate_config_zero3.yaml \ examples/minimax_h3/model_training/train.py \ --dataset_base_path ./models/train/MiniMax-H3-Ref2VA-full-split-cache \ --data_file_keys "video,input_audio,references" \ --extra_inputs "input_audio,references" \ --height 480 \ --width 832 \ --num_frames 124 \ --dataset_repeat 100 \ --model_id_with_origin_paths "Comfy-Org/MiniMax-H3:diffusion_models/minimax_h3_ref2va_pruned_bf16.safetensors" \ --processor_path "MiniMax/MiniMax-H3:Ref2VA/processor/" \ --learning_rate 1e-5 \ --num_epochs 2 \ --remove_prefix_in_ckpt "pipe.dit." \ --output_path "./models/train/MiniMax-H3-Ref2VA-full" \ --trainable_models "dit" \ --use_gradient_checkpointing \ --find_unused_parameters \ --task "sft:train"