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

bug fix

Browse Source
This commit is contained in:
Artiprocher
2025-11-15 16:47:13 +08:00
parent e3356556ee
commit ea0a5c5908
72 changed files with 481 additions and 472 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
.gitignore vendored
View File

@@ -9,6 +9,7 @@
*.DS_Store
*.msc
*.mv
log*.txt
# Byte-compiled / optimized / DLL files
__pycache__/

2
diffsynth/configs/vram_management_module_maps.py
View File

@@ -46,7 +46,7 @@ VRAM_MANAGEMENT_MODULE_MAPS = {
},
"diffsynth.models.wan_video_dit.WanModel": {
"diffsynth.models.wan_video_dit.MLP": "diffsynth.core.vram.layers.AutoWrappedModule",
"diffsynth.models.wan_video_dit.DiTBlock": "diffsynth.core.vram.layers.AutoWrappedModule",
"diffsynth.models.wan_video_dit.DiTBlock": "diffsynth.core.vram.layers.AutoWrappedNonRecurseModule",
"diffsynth.models.wan_video_dit.Head": "diffsynth.core.vram.layers.AutoWrappedModule",
"torch.nn.Linear": "diffsynth.core.vram.layers.AutoWrappedLinear",
"torch.nn.Conv3d": "diffsynth.core.vram.layers.AutoWrappedModule",

64
diffsynth/core/vram/layers.py
View File

@@ -198,6 +198,66 @@ class AutoWrappedModule(AutoTorchModule):
return getattr(self.module, name)
class AutoWrappedNonRecurseModule(AutoWrappedModule):
def __init__(
self,
module: torch.nn.Module,
offload_dtype: torch.dtype = None,
offload_device: Union[str, torch.device] = None,
onload_dtype: torch.dtype = None,
onload_device: Union[str, torch.device] = None,
preparing_dtype: torch.dtype = None,
preparing_device: Union[str, torch.device] = None,
computation_dtype: torch.dtype = None,
computation_device: Union[str, torch.device] = None,
vram_limit: float = None,
name: str = "",
disk_map: DiskMap = None,
**kwargs
):
super().__init__(
module,
offload_dtype,
offload_device,
onload_dtype,
onload_device,
preparing_dtype,
preparing_device,
computation_dtype,
computation_device,
vram_limit,
name,
disk_map,
**kwargs
)
if self.disk_offload:
self.required_params = [name for name, _ in self.module.named_parameters(recurse=False)]
def load_from_disk(self, torch_dtype, device, copy_module=False):
if copy_module:
module = copy.deepcopy(self.module)
else:
module = self.module
state_dict = {}
for name in self.required_params:
param = self.disk_map[self.param_name(name)]
param = param.to(dtype=torch_dtype, device=device)
state_dict[name] = param
module.load_state_dict(state_dict, assign=True, strict=False)
return module
def offload_to_disk(self, model: torch.nn.Module):
for name in self.required_params:
getattr(self, name).to("meta")
def __getattr__(self, name):
if name in self.__dict__ or name == "module":
return super().__getattr__(name)
else:
return getattr(self.module, name)
class AutoWrappedLinear(torch.nn.Linear, AutoTorchModule):
def __init__(
self,
@@ -366,11 +426,15 @@ class AutoWrappedLinear(torch.nn.Linear, AutoTorchModule):
def enable_vram_management_recursively(model: torch.nn.Module, module_map: dict, vram_config: dict, vram_limit=None, name_prefix="", disk_map=None, **kwargs):
if isinstance(model, AutoWrappedNonRecurseModule):
model = model.module
for name, module in model.named_children():
layer_name = name if name_prefix == "" else name_prefix + "." + name
for source_module, target_module in module_map.items():
if isinstance(module, source_module):
module_ = target_module(module, **vram_config, vram_limit=vram_limit, name=layer_name, disk_map=disk_map, **kwargs)
if isinstance(module_, AutoWrappedNonRecurseModule):
enable_vram_management_recursively(module_, module_map, vram_config, vram_limit=vram_limit, name_prefix=layer_name, disk_map=disk_map, **kwargs)
setattr(model, name, module_)
break
else:

7
diffsynth/diffusion/base_pipeline.py
View File

@@ -220,7 +220,7 @@ class BasePipeline(torch.nn.Module):
module: torch.nn.Module,
lora_config: Union[ModelConfig, str] = None,
alpha=1,
hotload=False,
hotload=None,
state_dict=None,
):
if state_dict is None:
@@ -233,12 +233,15 @@ class BasePipeline(torch.nn.Module):
lora = state_dict
lora_loader = self.lora_loader(torch_dtype=self.torch_dtype, device=self.device)
lora = lora_loader.convert_state_dict(lora)
if hotload is None:
hotload = hasattr(module, "vram_management_enabled") and getattr(module, "vram_management_enabled")
if hotload:
if not (hasattr(module, "vram_management_enabled") and getattr(module, "vram_management_enabled")):
raise ValueError("VRAM Management is not enabled. LoRA hotloading is not supported.")
updated_num = 0
for name, module in module.named_modules():
for _, module in module.named_modules():
if isinstance(module, AutoWrappedLinear):
name = module.name
lora_a_name = f'{name}.lora_A.weight'
lora_b_name = f'{name}.lora_B.weight'
if lora_a_name in lora and lora_b_name in lora:

8
diffsynth/pipelines/wan_video.py
View File

@@ -15,10 +15,10 @@ from ..diffusion import FlowMatchScheduler
from ..core import ModelConfig, gradient_checkpoint_forward
from ..diffusion.base_pipeline import BasePipeline, PipelineUnit
from ..models.wan_video_dit import WanModel, RMSNorm, sinusoidal_embedding_1d
from ..models.wan_video_dit import WanModel, sinusoidal_embedding_1d
from ..models.wan_video_dit_s2v import rope_precompute
from ..models.wan_video_text_encoder import WanTextEncoder, HuggingfaceTokenizer
from ..models.wan_video_vae import WanVideoVAE, RMS_norm, CausalConv3d, Upsample
from ..models.wan_video_vae import WanVideoVAE
from ..models.wan_video_image_encoder import WanImageEncoder
from ..models.wan_video_vace import VaceWanModel
from ..models.wan_video_motion_controller import WanMotionControllerModel
@@ -526,13 +526,13 @@ class WanVideoUnit_FunReference(PipelineUnit):
super().__init__(
input_params=("reference_image", "height", "width", "reference_image"),
output_params=("reference_latents", "clip_feature"),
onload_model_names=("vae",)
onload_model_names=("vae", "image_encoder")
)
def process(self, pipe: WanVideoPipeline, reference_image, height, width):
if reference_image is None:
return {}
pipe.load_models_to_device(["vae"])
pipe.load_models_to_device(self.onload_model_names)
reference_image = reference_image.resize((width, height))
reference_latents = pipe.preprocess_video([reference_image])
reference_latents = pipe.vae.encode(reference_latents, device=pipe.device)

17
examples/test/run.py
View File

@@ -29,7 +29,7 @@ def run_tasks_on_single_GPU(script_path, gpu_id, num_gpu):
os.makedirs(target_path, exist_ok=True)
if script.endswith(".sh"):
cmd = f"CUDA_VISIBLE_DEVICES={gpu_id} bash {source_path} > {target_path}/log.txt 2>&1"
else:
elif script.endswith(".py"):
cmd = f"CUDA_VISIBLE_DEVICES={gpu_id} python {source_path} > {target_path}/log.txt 2>&1"
print(cmd, flush=True)
os.system(cmd)
@@ -58,6 +58,12 @@ def run_train_single_GPU(script_path):
p.join()
def move_files(prefix, target_folder):
os.makedirs(target_folder, exist_ok=True)
os.system(f"cp -r {prefix}* {target_folder}")
os.system(f"rm -rf {prefix}*")
if __name__ == "__main__":
# run_train_multi_GPU("examples/qwen_image/model_training/full")
# run_train_single_GPU("examples/qwen_image/model_training/lora")
@@ -66,3 +72,12 @@ if __name__ == "__main__":
# run_inference("examples/qwen_image/model_training/validate_full")
# run_inference("examples/qwen_image/model_training/validate_lora")
run_train_single_GPU("examples/wanvideo/model_inference_low_vram")
move_files("video_", "data/output/model_inference_low_vram")
run_train_single_GPU("examples/wanvideo/model_inference")
move_files("video_", "data/output/model_inference")
run_train_single_GPU("examples/wanvideo/model_training/lora")
run_train_single_GPU("examples/wanvideo/model_training/validate_lora")
move_files("video_", "data/output/validate_lora")
run_train_multi_GPU("examples/wanvideo/model_training/full")
run_train_single_GPU("examples/wanvideo/model_training/validate_full")
move_files("video_", "data/output/validate_full")

2
examples/wanvideo/model_inference/Wan2.2-Animate-14B.py
View File

@@ -41,7 +41,7 @@ save_video(video, "video_1_Wan2.2-Animate-14B.mp4", fps=15, quality=5)
# Replace
snapshot_download("Wan-AI/Wan2.2-Animate-14B", allow_file_pattern="relighting_lora.ckpt", local_dir="models/Wan-AI/Wan2.2-Animate-14B")
lora_state_dict = load_state_dict("models/Wan-AI/Wan2.2-Animate-14B/relighting_lora.ckpt", torch_dtype=torch.float32, device="cuda")["state_dict"]
lora_state_dict = load_state_dict("models/Wan-AI/Wan2.2-Animate-14B/relighting_lora.ckpt", torch_dtype=torch.bfloat16, device="cuda")["state_dict"]
pipe.load_lora(pipe.dit, state_dict=lora_state_dict)
input_image = Image.open("data/examples/wan/animate/replace_input_image.png")
animate_pose_video = VideoData("data/examples/wan/animate/replace_pose_video.mp4").raw_data()[:81-4]

2
examples/wanvideo/model_inference_low_vram/Wan2.2-Animate-14B.py
View File

@@ -52,7 +52,7 @@ save_video(video, "video_1_Wan2.2-Animate-14B.mp4", fps=15, quality=5)
# Replace
snapshot_download("Wan-AI/Wan2.2-Animate-14B", allow_file_pattern="relighting_lora.ckpt", local_dir="models/Wan-AI/Wan2.2-Animate-14B")
lora_state_dict = load_state_dict("models/Wan-AI/Wan2.2-Animate-14B/relighting_lora.ckpt", torch_dtype=torch.float32, device="cuda")["state_dict"]
lora_state_dict = load_state_dict("models/Wan-AI/Wan2.2-Animate-14B/relighting_lora.ckpt", torch_dtype=torch.bfloat16, device="cuda")["state_dict"]
pipe.load_lora(pipe.dit, state_dict=lora_state_dict)
input_image = Image.open("data/examples/wan/animate/replace_input_image.png")
animate_pose_video = VideoData("data/examples/wan/animate/replace_pose_video.mp4").raw_data()[:81-4]

12
examples/wanvideo/model_training/validate_full/LongCat-Video.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="meituan-longcat/LongCat-Video", origin_file_pattern="dit/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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",

14
examples/wanvideo/model_training/validate_full/Video-As-Prompt-Wan2.1-14B.py
View File

@@ -1,22 +1,22 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="ByteDance/Video-As-Prompt-Wan2.1-14B", origin_file_pattern="transformer/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"),
ModelConfig(model_id="ByteDance/Video-As-Prompt-Wan2.1-14B", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
state_dict = load_state_dict("models/train/Video-As-Prompt-Wan2.1-14B_full/epoch-1.safetensors")
pipe.vap.load_state_dict(state_dict)
pipe.enable_vram_management()
ref_video_path = 'data/example_video_dataset/wanvap/vap_ref.mp4'
target_image_path = 'data/example_video_dataset/wanvap/input_image.jpg'

14
examples/wanvideo/model_training/validate_full/Wan2.1-1.3b-speedcontrol-v1.py
View File

@@ -1,22 +1,22 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="DiffSynth-Studio/Wan2.1-1.3b-speedcontrol-v1", origin_file_pattern="model.safetensors"),
],
)
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(

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)

14
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-1.3B-Control.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)]

14
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-1.3B-InP.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)]

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)

14
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-1.3B-Control-Camera.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
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)

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
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)]

14
examples/wanvideo/model_training/validate_full/Wan2.1-Fun-V1.1-1.3B-InP.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
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)

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
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)

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
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)]

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
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)

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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]

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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]

12
examples/wanvideo/model_training/validate_full/Wan2.1-T2V-1.3B.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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",

12
examples/wanvideo/model_training/validate_full/Wan2.1-T2V-14B.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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",

12
examples/wanvideo/model_training/validate_full/Wan2.1-VACE-1.3B-Preview.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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)]

12
examples/wanvideo/model_training/validate_full/Wan2.1-VACE-1.3B.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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)]

12
examples/wanvideo/model_training/validate_full/Wan2.1-VACE-14B.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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)]

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

@@ -1,22 +1,22 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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]

14
examples/wanvideo/model_training/validate_full/Wan2.2-Fun-A14B-Control-Camera.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,17 +10,16 @@ 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"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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)

14
examples/wanvideo/model_training/validate_full/Wan2.2-Fun-A14B-Control.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,17 +10,16 @@ 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"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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)]

14
examples/wanvideo/model_training/validate_full/Wan2.2-Fun-A14B-InP.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,17 +10,16 @@ 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"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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

14
examples/wanvideo/model_training/validate_full/Wan2.2-I2V-A14B.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,17 +10,16 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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]

5
examples/wanvideo/model_training/validate_full/Wan2.2-S2V-14B.py
View File

@@ -2,7 +2,7 @@ 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
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
@@ -19,7 +19,6 @@ pipe = WanVideoPipeline.from_pretrained(
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
@@ -50,4 +49,4 @@ video = pipe(
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)
save_video_with_audio(video[1:], "video_Wan2.2-S2V-14B.mp4", audio_path, fps=16, quality=5)

14
examples/wanvideo/model_training/validate_full/Wan2.2-T2V-A14B.py
View File

@@ -1,24 +1,24 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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",

12
examples/wanvideo/model_training/validate_full/Wan2.2-TI2V-5B.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +10,13 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="Wan2.2_VAE.pth"),
],
)
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]

14
examples/wanvideo/model_training/validate_full/Wan2.2-VACE-Fun-A14B.py
View File

@@ -1,24 +1,24 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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)]

14
examples/wanvideo/model_training/validate_full/krea-realtime-video.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="krea/krea-realtime-video", origin_file_pattern="krea-realtime-video-14b.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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(
@@ -25,4 +25,4 @@ video = pipe(
cfg_scale=1,
sigma_shift=20,
)
save_video(video, "output.mp4", fps=15, quality=5)
save_video(video, "video_krea-realtime-video.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!")

11
examples/wanvideo/model_training/validate_lora/LongCat-Video.py
View File

@@ -1,20 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="meituan-longcat/LongCat-Video", origin_file_pattern="dit/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
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",

14
examples/wanvideo/model_training/validate_lora/Video-As-Prompt-Wan2.1-14B.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
device="cuda",
model_configs=[
ModelConfig(model_id="ByteDance/Video-As-Prompt-Wan2.1-14B", origin_file_pattern="transformer/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"),
ModelConfig(model_id="ByteDance/Video-As-Prompt-Wan2.1-14B", origin_file_pattern="transformer/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Video-As-Prompt-Wan2.1-14B_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
ref_video_path = 'data/example_video_dataset/wanvap/vap_ref.mp4'
target_image_path = 'data/example_video_dataset/wanvap/input_image.jpg'

14
examples/wanvideo/model_training/validate_lora/Wan2.1-1.3b-speedcontrol-v1.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="DiffSynth-Studio/Wan2.1-1.3b-speedcontrol-v1", origin_file_pattern="model.safetensors"),
],
)
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(

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

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-FLF2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)

13
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-1.3B-Control.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)]

13
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-1.3B-InP.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-1.3B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
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)

13
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-14B-Control.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-Control", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-14B-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)]

13
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-14B-InP.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-14B-InP", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-14B-InP_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

14
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-V1.1-1.3B-Control-Camera.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +10,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-1.3B-Control-Camera_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

13
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-V1.1-1.3B-Control.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-1.3B-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)]

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

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-1.3B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-1.3B-InP_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

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

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +10,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-1.3B-Control-Camera_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

13
examples/wanvideo/model_training/validate_lora/Wan2.1-Fun-V1.1-14B-Control.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-14B-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)]

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

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.1-Fun-V1.1-14B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
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"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-Fun-V1.1-14B-InP_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

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

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-480P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-I2V-14B-480P_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]

13
examples/wanvideo/model_training/validate_lora/Wan2.1-I2V-14B-720P.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,14 +9,13 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-I2V-14B-720P", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-I2V-14B-720P_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=720, width=1280)[0]

11
examples/wanvideo/model_training/validate_lora/Wan2.1-T2V-1.3B.py
View File

@@ -1,20 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-1.3B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-T2V-1.3B_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",

11
examples/wanvideo/model_training/validate_lora/Wan2.1-T2V-14B.py
View File

@@ -1,20 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.1-T2V-14B_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",

11
examples/wanvideo/model_training/validate_lora/Wan2.1-VACE-1.3B-Preview.py
View File

@@ -1,20 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="iic/VACE-Wan2.1-1.3B-Preview", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.vace, "models/train/Wan2.1-VACE-1.3B-Preview_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(49)]

11
examples/wanvideo/model_training/validate_lora/Wan2.1-VACE-1.3B.py
View File

@@ -1,20 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-1.3B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.vace, "models/train/Wan2.1-VACE-1.3B_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(49)]

11
examples/wanvideo/model_training/validate_lora/Wan2.1-VACE-14B.py
View File

@@ -1,20 +1,19 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-VACE-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.vace, "models/train/Wan2.1-VACE-14B_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(17)]

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

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="Wan2.1_VAE.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-Animate-14B", origin_file_pattern="models_clip_open-clip-xlm-roberta-large-vit-huge-14.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-Animate-14B_lora/epoch-4.safetensors", alpha=1)
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]

14
examples/wanvideo/model_training/validate_lora/Wan2.2-Fun-A14B-Control-Camera.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +10,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control-Camera", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-Fun-A14B-Control-Camera_high_noise_lora/epoch-4.safetensors", alpha=1)
pipe.load_lora(pipe.dit2, "models/train/Wan2.2-Fun-A14B-Control-Camera_low_noise_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

13
examples/wanvideo/model_training/validate_lora/Wan2.2-Fun-A14B-Control.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +9,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-Control", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-Fun-A14B-Control_high_noise_lora/epoch-4.safetensors", alpha=1)
pipe.load_lora(pipe.dit2, "models/train/Wan2.2-Fun-A14B-Control_low_noise_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)]

13
examples/wanvideo/model_training/validate_lora/Wan2.2-Fun-A14B-InP.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +9,14 @@ 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"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-Fun-A14B-InP", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-Fun-A14B-InP_high_noise_lora/epoch-4.safetensors", alpha=1)
pipe.load_lora(pipe.dit2, "models/train/Wan2.2-Fun-A14B-InP_low_noise_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)

13
examples/wanvideo/model_training/validate_lora/Wan2.2-I2V-A14B.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +9,14 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-I2V-A14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-I2V-A14B_high_noise_lora/epoch-4.safetensors", alpha=1)
pipe.load_lora(pipe.dit2, "models/train/Wan2.2-I2V-A14B_low_noise_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]

5
examples/wanvideo/model_training/validate_lora/Wan2.2-S2V-14B.py
View File

@@ -2,7 +2,7 @@ import torch
from PIL import Image
import librosa
from diffsynth import VideoData, save_video_with_audio
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
pipe = WanVideoPipeline.from_pretrained(
torch_dtype=torch.bfloat16,
@@ -17,7 +17,6 @@ pipe = WanVideoPipeline.from_pretrained(
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-S2V-14B_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
num_frames = 81 # 4n+1
@@ -48,4 +47,4 @@ video = pipe(
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)
save_video_with_audio(video[1:], "video_Wan2.2-S2V-14B.mp4", audio_path, fps=16, quality=5)

13
examples/wanvideo/model_training/validate_lora/Wan2.2-T2V-A14B.py
View File

@@ -1,7 +1,7 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,15 +9,14 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-T2V-A14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-T2V-A14B_high_noise_lora/epoch-4.safetensors", alpha=1)
pipe.load_lora(pipe.dit2, "models/train/Wan2.2-T2V-A14B_low_noise_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
video = pipe(
prompt="from sunset to night, a small town, light, house, river",

12
examples/wanvideo/model_training/validate_lora/Wan2.2-TI2V-5B.py
View File

@@ -1,7 +1,8 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video import WanVideoPipeline, ModelConfig
from modelscope import dataset_snapshot_download
@@ -9,13 +10,12 @@ 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"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.2-TI2V-5B", origin_file_pattern="Wan2.2_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/Wan2.2-TI2V-5B_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
input_image = VideoData("data/example_video_dataset/video1.mp4", height=480, width=832)[0]

13
examples/wanvideo/model_training/validate_lora/Wan2.2-VACE-Fun-A14B.py
View File

@@ -1,22 +1,21 @@
import torch
from PIL import Image
from diffsynth import save_video, VideoData
from diffsynth.pipelines.wan_video_new import WanVideoPipeline, ModelConfig
from diffsynth.utils.data import save_video, VideoData
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="high_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="low_noise_model/diffusion_pytorch_model*.safetensors"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="PAI/Wan2.2-VACE-Fun-A14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.vace, "models/train/Wan2.2-VACE-Fun-A14B_high_noise_lora/epoch-4.safetensors", alpha=1)
pipe.load_lora(pipe.vace2, "models/train/Wan2.2-VACE-Fun-A14B_low_noise_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(17)]

14
examples/wanvideo/model_training/validate_lora/krea-realtime-video.py
View File

@@ -1,21 +1,21 @@
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 diffsynth.utils.data import save_video, VideoData
from diffsynth.core import load_state_dict
from diffsynth.pipelines.wan_video 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"),
ModelConfig(model_id="krea/krea-realtime-video", origin_file_pattern="krea-realtime-video-14b.safetensors"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="models_t5_umt5-xxl-enc-bf16.pth"),
ModelConfig(model_id="Wan-AI/Wan2.1-T2V-14B", origin_file_pattern="Wan2.1_VAE.pth"),
],
)
pipe.load_lora(pipe.dit, "models/train/krea-realtime-video_lora/epoch-4.safetensors", alpha=1)
pipe.enable_vram_management()
# Text-to-video
video = pipe(
@@ -25,4 +25,4 @@ video = pipe(
cfg_scale=1,
sigma_shift=20,
)
save_video(video, "output.mp4", fps=15, quality=5)
save_video(video, "video_krea-realtime-video.mp4", fps=15, quality=5)

25
examples/wanvideo/model_training/validate_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} 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_lora"):
if file_name != "run_test.py":
scripts.append(os.path.join("examples/wanvideo/model_training/validate_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!")