theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-18 22:08:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

DiffSynth-Studio 2.0 major update

Browse Source
This commit is contained in:
root
2025-12-04 16:33:07 +08:00
parent afd101f345
commit 72af7122b3
758 changed files with 26462 additions and 2221398 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
examples/wanvideo/model_training/full/Wan2.1-FLF2V-14B-720P.sh
View File

@@ -10,4 +10,5 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--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"
--extra_inputs "input_image,end_image" \
--initialize_model_on_cpu

3
examples/wanvideo/model_training/full/Wan2.1-I2V-14B-480P.sh
View File

@@ -10,4 +10,5 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--remove_prefix_in_ckpt "pipe.dit." \
--output_path "./models/train/Wan2.1-I2V-14B-480P_full" \
--trainable_models "dit" \
--extra_inputs "input_image"
--extra_inputs "input_image" \
--initialize_model_on_cpu

3
examples/wanvideo/model_training/full/Wan2.1-I2V-14B-720P.sh
View File

@@ -12,4 +12,5 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--output_path "./models/train/Wan2.1-I2V-14B-720P_full" \
--trainable_models "dit" \
--extra_inputs "input_image" \
--use_gradient_checkpointing_offload
--use_gradient_checkpointing_offload \
--initialize_model_on_cpu

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

@@ -9,7 +9,7 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--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" \
--output_path "./models/train/Wan2.2-Fun-A14B-Control-Camera_high_noise_full" \
--trainable_models "dit" \
--extra_inputs "input_image,camera_control_direction,camera_control_speed" \
--max_timestep_boundary 0.358 \

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

@@ -9,7 +9,7 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--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" \
--output_path "./models/train/Wan2.2-Fun-A14B-Control_high_noise_full" \
--trainable_models "dit" \
--extra_inputs "control_video,reference_image" \
--max_timestep_boundary 0.358 \

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

@@ -8,7 +8,7 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--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" \
--output_path "./models/train/Wan2.2-Fun-A14B-InP_high_noise_full" \
--trainable_models "dit" \
--extra_inputs "input_image,end_image" \
--max_timestep_boundary 0.358 \

5
examples/wanvideo/model_training/full/Wan2.2-S2V-14B.sh
View File

@@ -1,13 +1,12 @@
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 \
--dataset_base_path data/example_video_dataset \
--dataset_metadata_path data/example_video_dataset/metadata_s2v.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" \

6
examples/wanvideo/model_training/full/Wan2.2-VACE-Fun-A14B.sh
View File

@@ -15,7 +15,8 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 0.358 \
--min_timestep_boundary 0
--min_timestep_boundary 0 \
--initialize_model_on_cpu
# boundary corresponds to timesteps [900, 1000]
@@ -36,5 +37,6 @@ accelerate launch --config_file examples/wanvideo/model_training/full/accelerate
--extra_inputs "vace_video,vace_reference_image" \
--use_gradient_checkpointing_offload \
--max_timestep_boundary 1 \
--min_timestep_boundary 0.358
--min_timestep_boundary 0.358 \
--initialize_model_on_cpu
# boundary corresponds to timesteps [0, 900]

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!")