theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-24 10:18:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

diffsynth 2.0 prototype

Browse Source
This commit is contained in:
Artiprocher
2025-11-04 10:59:29 +08:00
parent a30ed9093f
commit 288fb7604c
664 changed files with 3581 additions and 2237905 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
examples/wanvideo/model_training/full/LongCat-Video.sh
View File

@@ -1,12 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "meituan-longcat/LongCat-Video:dit/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/LongCat-Video_full" \
--trainable_models "dit"

13
examples/wanvideo/model_training/full/Wan2.1-1.3b-speedcontrol-v1.sh
View File

@@ -1,13 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_motion_bucket_id.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-T2V-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-1.3B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-1.3B:Wan2.1_VAE.pth,DiffSynth-Studio/Wan2.1-1.3b-speedcontrol-v1:model.safetensors" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.motion_controller." \
--output_path "./models/train/Wan2.1-1.3b-speedcontrol-v1_full" \
--trainable_models "motion_controller" \
--extra_inputs "motion_bucket_id"

13
examples/wanvideo/model_training/full/Wan2.1-FLF2V-14B-720P.sh
View File

@@ -1,13 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-FLF2V-14B-720P:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-FLF2V-14B-720P:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-FLF2V-14B-720P:Wan2.1_VAE.pth,Wan-AI/Wan2.1-FLF2V-14B-720P:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-FLF2V-14B-720P_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image"

14
examples/wanvideo/model_training/full/Wan2.1-Fun-1.3B-Control.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_control.csv \
--data_file_keys "video,control_video" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-1.3B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-1.3B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-1.3B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-1.3B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-1.3B-Control_full" \
--trainable_models "dit" \
--extra_inputs "control_video"

13
examples/wanvideo/model_training/full/Wan2.1-Fun-1.3B-InP.sh
View File

@@ -1,13 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-1.3B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-1.3B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-1.3B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-1.3B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-1.3B-InP_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image"

14
examples/wanvideo/model_training/full/Wan2.1-Fun-14B-Control.sh
View File

@@ -1,14 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_control.csv \
--data_file_keys "video,control_video" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-14B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-14B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-14B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-14B-Control_full" \
--trainable_models "dit" \
--extra_inputs "control_video"

13
examples/wanvideo/model_training/full/Wan2.1-Fun-14B-InP.sh
View File

@@ -1,13 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-14B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-14B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-14B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-14B-InP_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image"

13
examples/wanvideo/model_training/full/Wan2.1-Fun-V1.1-1.3B-Control-Camera.sh
View File

@@ -1,13 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-1.3B-Control-Camera_full" \
--trainable_models "dit" \
--extra_inputs "input_image,camera_control_direction,camera_control_speed"

14
examples/wanvideo/model_training/full/Wan2.1-Fun-V1.1-1.3B-Control.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-1.3B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-1.3B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-1.3B-Control_full" \
--trainable_models "dit" \
--extra_inputs "control_video,reference_image"

13
examples/wanvideo/model_training/full/Wan2.1-Fun-V1.1-1.3B-InP.sh
View File

@@ -1,13 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-1.3B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-1.3B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-1.3B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-1.3B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-1.3B-InP_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image"

13
examples/wanvideo/model_training/full/Wan2.1-Fun-V1.1-14B-Control-Camera.sh
View File

@@ -1,13 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-14B-Control-Camera_full" \
--trainable_models "dit" \
--extra_inputs "input_image,camera_control_direction,camera_control_speed"

14
examples/wanvideo/model_training/full/Wan2.1-Fun-V1.1-14B-Control.sh
View File

@@ -1,14 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-14B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-14B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-14B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-14B-Control_full" \
--trainable_models "dit" \
--extra_inputs "control_video,reference_image"

13
examples/wanvideo/model_training/full/Wan2.1-Fun-V1.1-14B-InP.sh
View File

@@ -1,13 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-14B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-14B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-14B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-14B-InP_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image"

13
examples/wanvideo/model_training/full/Wan2.1-I2V-14B-480P.sh
View File

@@ -1,13 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-I2V-14B-480P:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-I2V-14B-480P:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-I2V-14B-480P:Wan2.1_VAE.pth,Wan-AI/Wan2.1-I2V-14B-480P:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-I2V-14B-480P_full" \
--trainable_models "dit" \
--extra_inputs "input_image"

15
examples/wanvideo/model_training/full/Wan2.1-I2V-14B-720P.sh
View File

@@ -1,15 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 720 \
--width 1280 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-I2V-14B-720P:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-I2V-14B-720P:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-I2V-14B-720P:Wan2.1_VAE.pth,Wan-AI/Wan2.1-I2V-14B-720P:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-I2V-14B-720P_full" \
--trainable_models "dit" \
--extra_inputs "input_image" \
--use_gradient_checkpointing_offload

12
examples/wanvideo/model_training/full/Wan2.1-T2V-1.3B.sh
View File

@@ -1,12 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-T2V-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-1.3B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-1.3B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-T2V-1.3B_full" \
--trainable_models "dit"

12
examples/wanvideo/model_training/full/Wan2.1-T2V-14B.sh
View File

@@ -1,12 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-T2V-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-T2V-14B_full" \
--trainable_models "dit"

16
examples/wanvideo/model_training/full/Wan2.1-VACE-1.3B-Preview.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "iic/VACE-Wan2.1-1.3B-Preview:diffusion_pytorch_model*.safetensors,iic/VACE-Wan2.1-1.3B-Preview:models_t5_umt5-xxl-enc-bf16.pth,iic/VACE-Wan2.1-1.3B-Preview:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.1-VACE-1.3B-Preview_full" \
--trainable_models "vace" \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload

16
examples/wanvideo/model_training/full/Wan2.1-VACE-1.3B.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-VACE-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-VACE-1.3B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-VACE-1.3B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.1-VACE-1.3B_full" \
--trainable_models "vace" \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload

16
examples/wanvideo/model_training/full/Wan2.1-VACE-14B.sh
View File

@@ -1,16 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 17 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-VACE-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-VACE-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-VACE-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.1-VACE-14B_full" \
--trainable_models "vace" \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload

16
examples/wanvideo/model_training/full/Wan2.2-Animate-14B.sh
View File

@@ -1,16 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_animate.csv \
--data_file_keys "video,animate_pose_video,animate_face_video" \
--height 480 \
--width 832 \
--num_frames 81 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-Animate-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-Animate-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-Animate-14B:Wan2.1_VAE.pth,Wan-AI/Wan2.2-Animate-14B:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.animate_adapter." \
--output_path "./models/train/Wan2.2-Animate-14B_full" \
--trainable_models "animate_adapter" \
--extra_inputs "input_image,animate_pose_video,animate_face_video" \
--use_gradient_checkpointing_offload

35
examples/wanvideo/model_training/full/Wan2.2-Fun-A14B-Control-Camera.sh
View File

@@ -1,35 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control-Camera:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control-Camera:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control-Camera_high_niose_full" \
--trainable_models "dit" \
--extra_inputs "input_image,camera_control_direction,camera_control_speed" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control-Camera:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control-Camera:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control-Camera_low_noise_full" \
--trainable_models "dit" \
--extra_inputs "input_image,camera_control_direction,camera_control_speed" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

35
examples/wanvideo/model_training/full/Wan2.2-Fun-A14B-Control.sh
View File

@@ -1,35 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control_high_niose_full" \
--trainable_models "dit" \
--extra_inputs "control_video,reference_image" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control_low_noise_full" \
--trainable_models "dit" \
--extra_inputs "control_video,reference_image" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

33
examples/wanvideo/model_training/full/Wan2.2-Fun-A14B-InP.sh
View File

@@ -1,33 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-InP:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-InP:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-InP_high_niose_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-InP:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-InP:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-InP_low_noise_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

37
examples/wanvideo/model_training/full/Wan2.2-I2V-A14B.sh
View File

@@ -1,37 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-I2V-A14B:high_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-I2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-I2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-I2V-A14B_high_noise_full" \
--trainable_models "dit" \
--extra_inputs "input_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-I2V-A14B:low_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-I2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-I2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-I2V-A14B_low_noise_full" \
--trainable_models "dit" \
--extra_inputs "input_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900)

17
examples/wanvideo/model_training/full/Wan2.2-S2V-14B.sh
View File

@@ -1,17 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset/wans2v \
--dataset_metadata_path data/example_video_dataset/wans2v/metadata.csv \
--data_file_keys "video,input_audio,s2v_pose_video" \
--height 448 \
--width 832 \
--num_frames 81 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-S2V-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-S2V-14B:wav2vec2-large-xlsr-53-english/model.safetensors,Wan-AI/Wan2.2-S2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-S2V-14B:Wan2.1_VAE.pth" \
--audio_processor_config "Wan-AI/Wan2.2-S2V-14B:wav2vec2-large-xlsr-53-english/" \
--learning_rate 1e-5 \
--num_epochs 1 \
--trainable_models "dit" \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-S2V-14B_full" \
--extra_inputs "input_image,input_audio,s2v_pose_video" \
--use_gradient_checkpointing_offload

33
examples/wanvideo/model_training/full/Wan2.2-T2V-A14B.sh
View File

@@ -1,33 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-T2V-A14B:high_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-T2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-T2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-T2V-A14B_high_noise_full" \
--trainable_models "dit" \
--max_timestep_boundary 0.417 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [875, 1000]
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-T2V-A14B:low_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-T2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-T2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-T2V-A14B_low_noise_full" \
--trainable_models "dit" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.417
# boundary corresponds to timesteps [0, 875)

14
examples/wanvideo/model_training/full/Wan2.2-TI2V-5B.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-TI2V-5B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-TI2V-5B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-TI2V-5B:Wan2.2_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-TI2V-5B_full" \
--trainable_models "dit" \
--extra_inputs "input_image"

40
examples/wanvideo/model_training/full/Wan2.2-VACE-Fun-A14B.sh
View File

@@ -1,40 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 17 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-VACE-Fun-A14B:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-VACE-Fun-A14B:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-VACE-Fun-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.2-VACE-Fun-A14B_high_noise_full" \
--trainable_models "vace" \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 17 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-VACE-Fun-A14B:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-VACE-Fun-A14B:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-VACE-Fun-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.2-VACE-Fun-A14B_low_noise_full" \
--trainable_models "vace" \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

22
examples/wanvideo/model_training/full/accelerate_config_14B.yaml
View File

@@ -1,22 +0,0 @@
compute_environment: LOCAL_MACHINE
debug: false
deepspeed_config:
gradient_accumulation_steps: 1
offload_optimizer_device: cpu
offload_param_device: cpu
zero3_init_flag: false
zero_stage: 2
distributed_type: DEEPSPEED
downcast_bf16: 'no'
enable_cpu_affinity: false
machine_rank: 0
main_training_function: main
mixed_precision: bf16
num_machines: 1
num_processes: 8
rdzv_backend: static
same_network: true
tpu_env: []
tpu_use_cluster: false
tpu_use_sudo: false
use_cpu: false

12
examples/wanvideo/model_training/full/krea-realtime-video.sh
View File

@@ -1,12 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "krea/krea-realtime-video:krea-realtime-video-14b.safetensors,Wan-AI/Wan2.1-T2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-5 \
--num_epochs 2 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/krea-realtime-video_full" \
--trainable_models "dit"

38
examples/wanvideo/model_training/full/run_test.py
View File

@@ -1,38 +0,0 @@
import multiprocessing, os
def run_task(scripts, thread_id, thread_num):
for script_id, script in enumerate(scripts):
if script_id % thread_num == thread_id:
log_file_name = script.replace("/", "_") + ".txt"
cmd = f"CUDA_VISIBLE_DEVICES={thread_id} bash {script} > data/log/{log_file_name} 2>&1"
os.makedirs("data/log", exist_ok=True)
print(cmd, flush=True)
os.system(cmd)
if __name__ == "__main__":
# 1.3B
scripts = []
for file_name in os.listdir("examples/wanvideo/model_training/full"):
if file_name != "run_test.py" and "14B" not in file_name:
scripts.append(os.path.join("examples/wanvideo/model_training/full", file_name))
processes = [multiprocessing.Process(target=run_task, args=(scripts, i, 8)) for i in range(8)]
for p in processes:
p.start()
for p in processes:
p.join()
# 14B
scripts = []
for file_name in os.listdir("examples/wanvideo/model_training/full"):
if file_name != "run_test.py" and "14B" in file_name:
scripts.append(os.path.join("examples/wanvideo/model_training/full", file_name))
for script in scripts:
log_file_name = script.replace("/", "_") + ".txt"
cmd = f"bash {script} > data/log/{log_file_name} 2>&1"
print(cmd, flush=True)
os.system(cmd)
print("Done!")

14
examples/wanvideo/model_training/lora/LongCat-Video.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "meituan-longcat/LongCat-Video:dit/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/LongCat-Video_lora" \
--lora_base_model "dit" \
--lora_target_modules "adaLN_modulation.1,attn.qkv,attn.proj,cross_attn.q_linear,cross_attn.kv_linear,cross_attn.proj,ffn.w1,ffn.w2,ffn.w3" \
--lora_rank 32

15
examples/wanvideo/model_training/lora/Wan2.1-1.3b-speedcontrol-v1.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_motion_bucket_id.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-T2V-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-1.3B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-1.3B:Wan2.1_VAE.pth,DiffSynth-Studio/Wan2.1-1.3b-speedcontrol-v1:model.safetensors" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-1.3b-speedcontrol-v1_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "motion_bucket_id"

15
examples/wanvideo/model_training/lora/Wan2.1-FLF2V-14B-720P.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-FLF2V-14B-720P:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-FLF2V-14B-720P:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-FLF2V-14B-720P:Wan2.1_VAE.pth,Wan-AI/Wan2.1-FLF2V-14B-720P:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-FLF2V-14B-720P_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image"

16
examples/wanvideo/model_training/lora/Wan2.1-Fun-1.3B-Control.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_control.csv \
--data_file_keys "video,control_video" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-1.3B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-1.3B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-1.3B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-1.3B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-1.3B-Control_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "control_video"

15
examples/wanvideo/model_training/lora/Wan2.1-Fun-1.3B-InP.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-1.3B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-1.3B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-1.3B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-1.3B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-1.3B-InP_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image"

16
examples/wanvideo/model_training/lora/Wan2.1-Fun-14B-Control.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_control.csv \
--data_file_keys "video,control_video" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-14B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-14B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-14B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-14B-Control_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "control_video"

15
examples/wanvideo/model_training/lora/Wan2.1-Fun-14B-InP.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-14B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-14B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-14B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-14B-InP_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image"

15
examples/wanvideo/model_training/lora/Wan2.1-Fun-V1.1-1.3B-Control-Camera.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-1.3B-Control-Camera_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,camera_control_direction,camera_control_speed"

16
examples/wanvideo/model_training/lora/Wan2.1-Fun-V1.1-1.3B-Control.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-1.3B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-1.3B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-1.3B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-1.3B-Control_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "control_video,reference_image"

15
examples/wanvideo/model_training/lora/Wan2.1-Fun-V1.1-1.3B-InP.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-1.3B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-1.3B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-1.3B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-1.3B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-1.3B-InP_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image"

15
examples/wanvideo/model_training/lora/Wan2.1-Fun-V1.1-14B-Control-Camera.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-14B-Control-Camera:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-5 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-14B-Control-Camera_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,camera_control_direction,camera_control_speed"

16
examples/wanvideo/model_training/lora/Wan2.1-Fun-V1.1-14B-Control.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-14B-Control:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-14B-Control:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-14B-Control:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-14B-Control_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "control_video,reference_image"

15
examples/wanvideo/model_training/lora/Wan2.1-Fun-V1.1-14B-InP.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.1-Fun-V1.1-14B-InP:diffusion_pytorch_model*.safetensors,PAI/Wan2.1-Fun-V1.1-14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.1-Fun-V1.1-14B-InP:Wan2.1_VAE.pth,PAI/Wan2.1-Fun-V1.1-14B-InP:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-Fun-V1.1-14B-InP_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image"

15
examples/wanvideo/model_training/lora/Wan2.1-I2V-14B-480P.sh
View File

@@ -1,15 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-I2V-14B-480P:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-I2V-14B-480P:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-I2V-14B-480P:Wan2.1_VAE.pth,Wan-AI/Wan2.1-I2V-14B-480P:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-I2V-14B-480P_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image"

17
examples/wanvideo/model_training/lora/Wan2.1-I2V-14B-720P.sh
View File

@@ -1,17 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 720 \
--width 1280 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-I2V-14B-720P:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-I2V-14B-720P:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-I2V-14B-720P:Wan2.1_VAE.pth,Wan-AI/Wan2.1-I2V-14B-720P:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-I2V-14B-720P_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image" \
--use_gradient_checkpointing_offload

14
examples/wanvideo/model_training/lora/Wan2.1-T2V-1.3B.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-T2V-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-1.3B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-1.3B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-T2V-1.3B_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32

14
examples/wanvideo/model_training/lora/Wan2.1-T2V-14B.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-T2V-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-T2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-T2V-14B_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32

17
examples/wanvideo/model_training/lora/Wan2.1-VACE-1.3B-Preview.sh
View File

@@ -1,17 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "iic/VACE-Wan2.1-1.3B-Preview:diffusion_pytorch_model*.safetensors,iic/VACE-Wan2.1-1.3B-Preview:models_t5_umt5-xxl-enc-bf16.pth,iic/VACE-Wan2.1-1.3B-Preview:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.1-VACE-1.3B-Preview_lora" \
--lora_base_model "vace" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload

17
examples/wanvideo/model_training/lora/Wan2.1-VACE-1.3B.sh
View File

@@ -1,17 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-VACE-1.3B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-VACE-1.3B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-VACE-1.3B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.1-VACE-1.3B_lora" \
--lora_base_model "vace" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload

18
examples/wanvideo/model_training/lora/Wan2.1-VACE-14B.sh
View File

@@ -1,18 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 17 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.1-VACE-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.1-VACE-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-VACE-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.1-VACE-14B_lora" \
--lora_base_model "vace" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload

20
examples/wanvideo/model_training/lora/Wan2.2-Animate-14B.sh
View File

@@ -1,20 +0,0 @@
# 1*80G GPU cannot train Wan2.2-Animate-14B LoRA
# We tested on 8*80G GPUs
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_animate.csv \
--data_file_keys "video,animate_pose_video,animate_face_video" \
--height 480 \
--width 832 \
--num_frames 81 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-Animate-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-Animate-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-Animate-14B:Wan2.1_VAE.pth,Wan-AI/Wan2.2-Animate-14B:models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Animate-14B_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,animate_pose_video,animate_face_video" \
--use_gradient_checkpointing_offload

39
examples/wanvideo/model_training/lora/Wan2.2-Fun-A14B-Control-Camera.sh
View File

@@ -1,39 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control-Camera:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control-Camera:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control-Camera_high_niose_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,camera_control_direction,camera_control_speed" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_camera_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control-Camera:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control-Camera:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control-Camera:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control-Camera_low_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,camera_control_direction,camera_control_speed" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

39
examples/wanvideo/model_training/lora/Wan2.2-Fun-A14B-Control.sh
View File

@@ -1,39 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control_high_niose_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "control_video,reference_image" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_reference_control.csv \
--data_file_keys "video,control_video,reference_image" \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-Control:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-Control:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-Control:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-Control_low_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "control_video,reference_image" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

37
examples/wanvideo/model_training/lora/Wan2.2-Fun-A14B-InP.sh
View File

@@ -1,37 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-InP:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-InP:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-InP_high_niose_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-Fun-A14B-InP:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-Fun-A14B-InP:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-Fun-A14B-InP:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-Fun-A14B-InP_low_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,end_image" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

39
examples/wanvideo/model_training/lora/Wan2.2-I2V-A14B.sh
View File

@@ -1,39 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-I2V-A14B:high_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-I2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-I2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-I2V-A14B_high_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image" \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-I2V-A14B:low_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-I2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-I2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-I2V-A14B_low_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image" \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900)

19
examples/wanvideo/model_training/lora/Wan2.2-S2V-14B.sh
View File

@@ -1,19 +0,0 @@
accelerate launch --config_file examples/wanvideo/model_training/full/accelerate_config_14B.yaml examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset/wans2v \
--dataset_metadata_path data/example_video_dataset/wans2v/metadata.csv \
--data_file_keys "video,input_audio,s2v_pose_video" \
--height 448 \
--width 832 \
--num_frames 81 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-S2V-14B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-S2V-14B:wav2vec2-large-xlsr-53-english/model.safetensors,Wan-AI/Wan2.2-S2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-S2V-14B:Wan2.1_VAE.pth" \
--audio_processor_config "Wan-AI/Wan2.2-S2V-14B:wav2vec2-large-xlsr-53-english/" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-S2V-14B_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image,input_audio,s2v_pose_video" \
--use_gradient_checkpointing_offload

38
examples/wanvideo/model_training/lora/Wan2.2-T2V-A14B.sh
View File

@@ -1,38 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-T2V-A14B:high_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-T2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-T2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-T2V-A14B_high_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--max_timestep_boundary 0.417 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [875, 1000]
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-T2V-A14B:low_noise_model/diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-T2V-A14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-T2V-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-T2V-A14B_low_noise_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.417
# boundary corresponds to timesteps [0, 875)

16
examples/wanvideo/model_training/lora/Wan2.2-TI2V-5B.sh
View File

@@ -1,16 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--num_frames 49 \
--dataset_repeat 100 \
--model_id_with_origin_paths "Wan-AI/Wan2.2-TI2V-5B:diffusion_pytorch_model*.safetensors,Wan-AI/Wan2.2-TI2V-5B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.2-TI2V-5B:Wan2.2_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.2-TI2V-5B_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "input_image"

43
examples/wanvideo/model_training/lora/Wan2.2-VACE-Fun-A14B.sh
View File

@@ -1,43 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 17 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-VACE-Fun-A14B:high_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-VACE-Fun-A14B:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-VACE-Fun-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.2-VACE-Fun-A14B_high_noise_lora" \
--lora_base_model "vace" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
# boundary corresponds to timesteps [900, 1000]
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_vace.csv \
--data_file_keys "video,vace_video,vace_reference_image" \
--height 480 \
--width 832 \
--num_frames 17 \
--dataset_repeat 100 \
--model_id_with_origin_paths "PAI/Wan2.2-VACE-Fun-A14B:low_noise_model/diffusion_pytorch_model*.safetensors,PAI/Wan2.2-VACE-Fun-A14B:models_t5_umt5-xxl-enc-bf16.pth,PAI/Wan2.2-VACE-Fun-A14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.vace." \
--output_path "./models/train/Wan2.2-VACE-Fun-A14B_low_noise_lora" \
--lora_base_model "vace" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32 \
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
# boundary corresponds to timesteps [0, 900]

14
examples/wanvideo/model_training/lora/krea-realtime-video.sh
View File

@@ -1,14 +0,0 @@
accelerate launch examples/wanvideo/model_training/train.py \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata.csv \
--height 480 \
--width 832 \
--dataset_repeat 100 \
--model_id_with_origin_paths "krea/krea-realtime-video:krea-realtime-video-14b.safetensors,Wan-AI/Wan2.1-T2V-14B:models_t5_umt5-xxl-enc-bf16.pth,Wan-AI/Wan2.1-T2V-14B:Wan2.1_VAE.pth" \
--learning_rate 1e-4 \
--num_epochs 5 \
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/krea-realtime-video_lora" \
--lora_base_model "dit" \
--lora_target_modules "q,k,v,o,ffn.0,ffn.2" \
--lora_rank 32

25
examples/wanvideo/model_training/lora/run_test.py
View File

@@ -1,25 +0,0 @@
import multiprocessing, os
def run_task(scripts, thread_id, thread_num):
for script_id, script in enumerate(scripts):
if script_id % thread_num == thread_id:
log_file_name = script.replace("/", "_") + ".txt"
cmd = f"CUDA_VISIBLE_DEVICES={thread_id} bash {script} > data/log/{log_file_name} 2>&1"
os.makedirs("data/log", exist_ok=True)
print(cmd, flush=True)
os.system(cmd)
if __name__ == "__main__":
scripts = []
for file_name in os.listdir("examples/wanvideo/model_training/lora"):
if file_name != "run_test.py":
scripts.append(os.path.join("examples/wanvideo/model_training/lora", file_name))
processes = [multiprocessing.Process(target=run_task, args=(scripts, i, 8)) for i in range(8)]
for p in processes:
p.start()
for p in processes:
p.join()
print("Done!")

136
examples/wanvideo/model_training/train.py
View File

@@ -1,136 +0,0 @@
import torch, os, json
from diffsynth import load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.trainers.utils import DiffusionTrainingModule, ModelLogger, launch_training_task, wan_parser
from diffsynth.trainers.unified_dataset import UnifiedDataset, LoadVideo, LoadAudio, ImageCropAndResize, ToAbsolutePath
os.environ["TOKENIZERS_PARALLELISM"] = "false"
class WanTrainingModule(DiffusionTrainingModule):
def __init__(
self,
model_paths=None, model_id_with_origin_paths=None, audio_processor_config=None,
trainable_models=None,
lora_base_model=None, lora_target_modules="q,k,v,o,ffn.0,ffn.2", lora_rank=32, lora_checkpoint=None,
use_gradient_checkpointing=True,
use_gradient_checkpointing_offload=False,
extra_inputs=None,
max_timestep_boundary=1.0,
min_timestep_boundary=0.0,
):
super().__init__()
# Load models
model_configs = self.parse_model_configs(model_paths, model_id_with_origin_paths, enable_fp8_training=False)
if audio_processor_config is not None:
audio_processor_config = ModelConfig(model_id=audio_processor_config.split(":")[0], origin_file_pattern=audio_processor_config.split(":")[1])
self.pipe = WanVideoPipeline.from_pretrained(torch_dtype=torch.bfloat16, device="cpu", model_configs=model_configs, audio_processor_config=audio_processor_config)
# Training mode
self.switch_pipe_to_training_mode(
self.pipe, trainable_models,
lora_base_model, lora_target_modules, lora_rank, lora_checkpoint=lora_checkpoint,
enable_fp8_training=False,
)
# Store other configs
self.use_gradient_checkpointing = use_gradient_checkpointing
self.use_gradient_checkpointing_offload = use_gradient_checkpointing_offload
self.extra_inputs = extra_inputs.split(",") if extra_inputs is not None else []
self.max_timestep_boundary = max_timestep_boundary
self.min_timestep_boundary = min_timestep_boundary
def forward_preprocess(self, data):
# CFG-sensitive parameters
inputs_posi = {"prompt": data["prompt"]}
inputs_nega = {}
# CFG-unsensitive parameters
inputs_shared = {
# Assume you are using this pipeline for inference,
# please fill in the input parameters.
"input_video": data["video"],
"height": data["video"][0].size[1],
"width": data["video"][0].size[0],
"num_frames": len(data["video"]),
# Please do not modify the following parameters
# unless you clearly know what this will cause.
"cfg_scale": 1,
"tiled": False,
"rand_device": self.pipe.device,
"use_gradient_checkpointing": self.use_gradient_checkpointing,
"use_gradient_checkpointing_offload": self.use_gradient_checkpointing_offload,
"cfg_merge": False,
"vace_scale": 1,
"max_timestep_boundary": self.max_timestep_boundary,
"min_timestep_boundary": self.min_timestep_boundary,
}
# Extra inputs
for extra_input in self.extra_inputs:
if extra_input == "input_image":
inputs_shared["input_image"] = data["video"][0]
elif extra_input == "end_image":
inputs_shared["end_image"] = data["video"][-1]
elif extra_input == "reference_image" or extra_input == "vace_reference_image":
inputs_shared[extra_input] = data[extra_input][0]
else:
inputs_shared[extra_input] = data[extra_input]
# Pipeline units will automatically process the input parameters.
for unit in self.pipe.units:
inputs_shared, inputs_posi, inputs_nega = self.pipe.unit_runner(unit, self.pipe, inputs_shared, inputs_posi, inputs_nega)
return {**inputs_shared, **inputs_posi}
def forward(self, data, inputs=None):
if inputs is None: inputs = self.forward_preprocess(data)
models = {name: getattr(self.pipe, name) for name in self.pipe.in_iteration_models}
loss = self.pipe.training_loss(**models, **inputs)
return loss
if __name__ == "__main__":
parser = wan_parser()
args = parser.parse_args()
dataset = UnifiedDataset(
base_path=args.dataset_base_path,
metadata_path=args.dataset_metadata_path,
repeat=args.dataset_repeat,
data_file_keys=args.data_file_keys.split(","),
main_data_operator=UnifiedDataset.default_video_operator(
base_path=args.dataset_base_path,
max_pixels=args.max_pixels,
height=args.height,
width=args.width,
height_division_factor=16,
width_division_factor=16,
num_frames=args.num_frames,
time_division_factor=4,
time_division_remainder=1,
),
special_operator_map={
"animate_face_video": ToAbsolutePath(args.dataset_base_path) >> LoadVideo(args.num_frames, 4, 1, frame_processor=ImageCropAndResize(512, 512, None, 16, 16)),
"input_audio": ToAbsolutePath(args.dataset_base_path) >> LoadAudio(sr=16000),
}
)
model = WanTrainingModule(
model_paths=args.model_paths,
model_id_with_origin_paths=args.model_id_with_origin_paths,
audio_processor_config=args.audio_processor_config,
trainable_models=args.trainable_models,
lora_base_model=args.lora_base_model,
lora_target_modules=args.lora_target_modules,
lora_rank=args.lora_rank,
lora_checkpoint=args.lora_checkpoint,
use_gradient_checkpointing_offload=args.use_gradient_checkpointing_offload,
extra_inputs=args.extra_inputs,
max_timestep_boundary=args.max_timestep_boundary,
min_timestep_boundary=args.min_timestep_boundary,
)
model_logger = ModelLogger(
args.output_path,
remove_prefix_in_ckpt=args.remove_prefix_in_ckpt
)
launch_training_task(dataset, model, model_logger, args=args)

25
examples/wanvideo/model_training/validate_full/LongCat-Video.py
View File

@@ -1,25 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="meituan-longcat/LongCat-Video", origin_file_pattern="dit/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/LongCat-Video_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True
)
save_video(video, "video_LongCat-Video.mp4", fps=15, quality=5)

28
examples/wanvideo/model_training/validate_full/Wan2.1-1.3b-speedcontrol-v1.py
View File

@@ -1,28 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="DiffSynth-Studio/Wan2.1-1.3b-speedcontrol-v1", origin_file_pattern="model.safetensors", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-1.3b-speedcontrol-v1_full/epoch-1.safetensors")
pipe.motion_controller.load_state_dict(state_dict)
pipe.enable_vram_management()
# Text-to-video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True,
motion_bucket_id=50
)
save_video(video, "video_Wan2.1-1.3b-speedcontrol-v1.mp4", fps=15, quality=5)

33
examples/wanvideo/model_training/validate_full/Wan2.1-FLF2V-14B-720P.py
View File

@@ -1,33 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-FLF2V-14B-720P_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0],
end_image=video[80],
seed=0, tiled=True,
sigma_shift=16,
)
save_video(video, "video_Wan2.1-FLF2V-14B-720P.mp4", fps=15, quality=5)

32
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-1.3B-Control.py
View File

@@ -1,32 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-1.3B-Control_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(81)]
# Control video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
control_video=video,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-Fun-1.3B-Control.mp4", fps=15, quality=5)

31
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-1.3B-InP.py
View File

@@ -1,31 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-1.3B-InP_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0], end_image=video[80],
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-1.3B-InP.mp4", fps=15, quality=5)

32
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-14B-Control.py
View File

@@ -1,32 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-14B-Control_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(81)]
# Control video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
control_video=video,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-Fun-14B-Control.mp4", fps=15, quality=5)

31
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-14B-InP.py
View File

@@ -1,31 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-14B-InP_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0], end_image=video[80],
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-14B-InP.mp4", fps=15, quality=5)

32
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-1.3B-Control-Camera.py
View File

@@ -1,32 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-V1.1-1.3B-Control-Camera_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0],
camera_control_direction="Left", camera_control_speed=0.0,
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-V1.1-1.3B-Control-Camera.mp4", fps=15, quality=5)

33
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-1.3B-Control.py
View File

@@ -1,33 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-V1.1-1.3B-Control_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(81)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
# Control video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
control_video=video, reference_image=reference_image,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-Fun-V1.1-1.3B-Control.mp4", fps=15, quality=5)

31
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-1.3B-InP.py
View File

@@ -1,31 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-V1.1-1.3B-InP_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0], end_image=video[80],
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-V1.1-1.3B-InP.mp4", fps=15, quality=5)

32
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-14B-Control-Camera.py
View File

@@ -1,32 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-V1.1-14B-Control-Camera_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0],
camera_control_direction="Left", camera_control_speed=0.0,
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-V1.1-14B-Control-Camera.mp4", fps=15, quality=5)

33
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-14B-Control.py
View File

@@ -1,33 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-V1.1-14B-Control_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(81)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
# Control video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
control_video=video, reference_image=reference_image,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-Fun-V1.1-14B-Control.mp4", fps=15, quality=5)

31
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-14B-InP.py
View File

@@ -1,31 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-Fun-V1.1-14B-InP_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0], end_image=video[80],
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-V1.1-14B-InP.mp4", fps=15, quality=5)

30
examples/wanvideo/model_training/validate_full/Wan2.1-I2V-14B-480P.py
View File

@@ -1,30 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-I2V-14B-480P_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=input_image,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-I2V-14B-480P.mp4", fps=15, quality=5)

31
examples/wanvideo/model_training/validate_full/Wan2.1-I2V-14B-720P.py
View File

@@ -1,31 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-I2V-14B-720P_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=720, width=1280)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=input_image,
height=720, width=1280, num_frames=49,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-I2V-14B-720P.mp4", fps=15, quality=5)

25
examples/wanvideo/model_training/validate_full/Wan2.1-T2V-1.3B.py
View File

@@ -1,25 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-T2V-1.3B_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-T2V-1.3B.mp4", fps=15, quality=5)

25
examples/wanvideo/model_training/validate_full/Wan2.1-T2V-14B.py
View File

@@ -1,25 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-T2V-14B_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-T2V-14B.mp4", fps=15, quality=5)

30
examples/wanvideo/model_training/validate_full/Wan2.1-VACE-1.3B-Preview.py
View File

@@ -1,30 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-VACE-1.3B-Preview_full/epoch-1.safetensors")
pipe.vace.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(49)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
vace_video=video, vace_reference_image=reference_image, num_frames=49,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-VACE-1.3B-Preview.mp4", fps=15, quality=5)

30
examples/wanvideo/model_training/validate_full/Wan2.1-VACE-1.3B.py
View File

@@ -1,30 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-VACE-1.3B_full/epoch-1.safetensors")
pipe.vace.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(49)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
vace_video=video, vace_reference_image=reference_image, num_frames=49,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-VACE-1.3B.mp4", fps=15, quality=5)

30
examples/wanvideo/model_training/validate_full/Wan2.1-VACE-14B.py
View File

@@ -1,30 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.1-VACE-14B_full/epoch-1.safetensors")
pipe.vace.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(17)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
vace_video=video, vace_reference_image=reference_image, num_frames=17,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-VACE-14B.mp4", fps=15, quality=5)

33
examples/wanvideo/model_training/validate_full/Wan2.2-Animate-14B.py
View File

@@ -1,33 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-Animate-14B_full/epoch-1.safetensors")
pipe.animate_adapter.load_state_dict(state_dict, strict=False)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/animate/animate_output.mp4", height=480, width=832)[0]
animate_pose_video = VideoData("data/examples/wan/animate/animate_pose_video.mp4", height=480, width=832).raw_data()[:81-4]
animate_face_video = VideoData("data/examples/wan/animate/animate_face_video.mp4", height=512, width=512).raw_data()[:81-4]
video = pipe(
prompt="视频中的人在做动作",
seed=0, tiled=True,
input_image=input_image,
animate_pose_video=animate_pose_video,
animate_face_video=animate_face_video,
num_frames=81, height=480, width=832,
num_inference_steps=20, cfg_scale=1,
)
save_video(video, "video_Wan2.2-Animate-14B.mp4", fps=15, quality=5)

34
examples/wanvideo/model_training/validate_full/Wan2.2-Fun-A14B-Control-Camera.py
View File

@@ -1,34 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-Fun-A14B-Control-Camera_high_noise_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
state_dict = load_state_dict("models/train/Wan2.2-Fun-A14B-Control-Camera_low_noise_full/epoch-1.safetensors")
pipe.dit2.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0],
camera_control_direction="Left", camera_control_speed=0.0,
seed=0, tiled=True
)
save_video(video, "video_Wan2.2-Fun-A14B-Control-Camera.mp4", fps=15, quality=5)

35
examples/wanvideo/model_training/validate_full/Wan2.2-Fun-A14B-Control.py
View File

@@ -1,35 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-Fun-A14B-Control_high_noise_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
state_dict = load_state_dict("models/train/Wan2.2-Fun-A14B-Control_low_noise_full/epoch-1.safetensors")
pipe.dit2.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(81)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
# Control video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
control_video=video, reference_image=reference_image,
seed=1, tiled=True
)
save_video(video, "video_Wan2.2-Fun-A14B-Control.mp4", fps=15, quality=5)

32
examples/wanvideo/model_training/validate_full/Wan2.2-Fun-A14B-InP.py
View File

@@ -1,32 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-Fun-A14B-InP_high_noise_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
state_dict = load_state_dict("models/train/Wan2.2-Fun-A14B-InP_low_noise_full/epoch-1.safetensors")
pipe.dit2.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0], end_image=video[80],
seed=0, tiled=True
)
save_video(video, "video_Wan2.2-Fun-A14B-InP.mp4", fps=15, quality=5)

33
examples/wanvideo/model_training/validate_full/Wan2.2-I2V-A14B.py
View File

@@ -1,33 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-I2V-A14B_high_noise_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
state_dict = load_state_dict("models/train/Wan2.2-I2V-A14B_low_noise_full/epoch-1.safetensors")
pipe.dit2.load_state_dict(state_dict)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=input_image,
num_frames=49,
seed=1, tiled=True,
)
save_video(video, "video_Wan2.2-I2V-A14B.mp4", fps=15, quality=5)

53
examples/wanvideo/model_training/validate_full/Wan2.2-S2V-14B.py
View File

@@ -1,53 +0,0 @@
import torch
from PIL import Image
import librosa
from diffsynth import VideoData, save_video_with_audio, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.2-S2V-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-S2V-14B", origin_file_pattern="wav2vec2-large-xlsr-53-english/model.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-S2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-S2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
audio_processor_config=ModelConfig(model_id="Wan-AI/Wan2.2-S2V-14B", origin_file_pattern="wav2vec2-large-xlsr-53-english/"),
)
state_dict = load_state_dict("models/train/Wan2.2-S2V-14B_full/epoch-0.safetensors")
pipe.dit.load_state_dict(state_dict, strict=False)
pipe.enable_vram_management()
num_frames = 81 # 4n+1
height = 448
width = 832
prompt = "a person is singing"
negative_prompt = "画面模糊，最差质量，画面模糊，细节模糊不清，情绪激动剧烈，手快速抖动，字幕，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走"
input_image = Image.open("data/example_video_dataset/wans2v/pose.png").convert("RGB").resize((width, height))
# s2v audio input, recommend 16kHz sampling rate
audio_path = 'data/example_video_dataset/wans2v/sing.MP3'
input_audio, sample_rate = librosa.load(audio_path, sr=16000)
# S2V pose video input
pose_video_path = 'data/example_video_dataset/wans2v/pose.mp4'
pose_video = VideoData(pose_video_path, height=height, width=width)
# Speech-to-video with pose
video = pipe(
prompt=prompt,
input_image=input_image,
negative_prompt=negative_prompt,
seed=0,
num_frames=num_frames,
height=height,
width=width,
audio_sample_rate=sample_rate,
input_audio=input_audio,
s2v_pose_video=pose_video,
num_inference_steps=40,
)
save_video_with_audio(video[1:], "video_pose_with_audio.mp4", audio_path, fps=16, quality=5)

28
examples/wanvideo/model_training/validate_full/Wan2.2-T2V-A14B.py
View File

@@ -1,28 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-T2V-A14B_high_noise_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
state_dict = load_state_dict("models/train/Wan2.2-T2V-A14B_low_noise_full/epoch-1.safetensors")
pipe.dit2.load_state_dict(state_dict)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True
)
save_video(video, "video_Wan2.2-T2V-A14B.mp4", fps=15, quality=5)

30
examples/wanvideo/model_training/validate_full/Wan2.2-TI2V-5B.py
View File

@@ -1,30 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="Wan2.2_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-TI2V-5B_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=input_image,
num_frames=49,
seed=1, tiled=True,
)
save_video(video, "video_Wan2.2-TI2V-5B.mp4", fps=15, quality=5)

33
examples/wanvideo/model_training/validate_full/Wan2.2-VACE-Fun-A14B.py
View File

@@ -1,33 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/Wan2.2-VACE-Fun-A14B_high_noise_full/epoch-1.safetensors")
pipe.vace.load_state_dict(state_dict)
state_dict = load_state_dict("models/train/Wan2.2-VACE-Fun-A14B_low_noise_full/epoch-1.safetensors")
pipe.vace2.load_state_dict(state_dict)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(17)]
reference_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
vace_video=video, vace_reference_image=reference_image, num_frames=17,
seed=1, tiled=True
)
save_video(video, "video_Wan2.2-VACE-A14B.mp4", fps=15, quality=5)

28
examples/wanvideo/model_training/validate_full/krea-realtime-video.py
View File

@@ -1,28 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="krea/krea-realtime-video", origin_file_pattern="krea-realtime-video-14b.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
state_dict = load_state_dict("models/train/krea-realtime-video_full/epoch-1.safetensors")
pipe.dit.load_state_dict(state_dict)
pipe.enable_vram_management()
# Text-to-video
video = pipe(
prompt="a cat sitting on a boat",
num_inference_steps=6, num_frames=81,
seed=0, tiled=True,
cfg_scale=1,
sigma_shift=20,
)
save_video(video, "output.mp4", fps=15, quality=5)

25
examples/wanvideo/model_training/validate_full/run_test.py
View File

@@ -1,25 +0,0 @@
import multiprocessing, os
def run_task(scripts, thread_id, thread_num):
for script_id, script in enumerate(scripts):
if script_id % thread_num == thread_id:
log_file_name = script.replace("/", "_") + ".txt"
cmd = f"CUDA_VISIBLE_DEVICES={thread_id} python -u {script} > data/log/{log_file_name} 2>&1"
os.makedirs("data/log", exist_ok=True)
print(cmd, flush=True)
os.system(cmd)
if __name__ == "__main__":
scripts = []
for file_name in os.listdir("examples/wanvideo/model_training/validate_full"):
if file_name != "run_test.py":
scripts.append(os.path.join("examples/wanvideo/model_training/validate_full", file_name))
processes = [multiprocessing.Process(target=run_task, args=(scripts, i, 8)) for i in range(8)]
for p in processes:
p.start()
for p in processes:
p.join()
print("Done!")

24
examples/wanvideo/model_training/validate_lora/LongCat-Video.py
View File

@@ -1,24 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="meituan-longcat/LongCat-Video", origin_file_pattern="dit/diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
],
)
pipe.load_lora(pipe.dit, "models/train/LongCat-Video_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True
)
save_video(video, "video_LongCat-Video.mp4", fps=15, quality=5)

27
examples/wanvideo/model_training/validate_lora/Wan2.1-1.3b-speedcontrol-v1.py
View File

@@ -1,27 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData, load_state_dict
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="DiffSynth-Studio/Wan2.1-1.3b-speedcontrol-v1", origin_file_pattern="model.safetensors", offload_device="cpu"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-1.3b-speedcontrol-v1_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
# Text-to-video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
seed=1, tiled=True,
motion_bucket_id=50
)
save_video(video, "video_Wan2.1-1.3b-speedcontrol-v1.mp4", fps=15, quality=5)

32
examples/wanvideo/model_training/validate_lora/Wan2.1-FLF2V-14B-720P.py
View File

@@ -1,32 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-FLF2V-14B-720P_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0],
end_image=video[80],
seed=0, tiled=True,
sigma_shift=16,
)
save_video(video, "video_Wan2.1-FLF2V-14B-720P.mp4", fps=15, quality=5)

31
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-1.3B-Control.py
View File

@@ -1,31 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-1.3B-Control_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1_softedge.mp4", height=480, width=832)
video = [video[i] for i in range(81)]
# Control video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
control_video=video,
seed=1, tiled=True
)
save_video(video, "video_Wan2.1-Fun-1.3B-Control.mp4", fps=15, quality=5)

30
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-1.3B-InP.py
View File

@@ -1,30 +0,0 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth", offload_device="cpu"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth", offload_device="cpu"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-1.3B-InP_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)
# First and last frame to video
video = pipe(
prompt="from sunset to night, a small town, light, house, river",
negative_prompt="色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走",
input_image=video[0], end_image=video[80],
seed=0, tiled=True
)
save_video(video, "video_Wan2.1-Fun-1.3B-InP.mp4", fps=15, quality=5)

Some files were not shown because too many files have changed in this diff Show More