This document provides a complete workflow for Wan2.2 Fun (Inpainting) LoRA fine-tuning training, including environment configuration, data preparation, various distributed training strategies, and inference testing.

Note: Wan2.2 Fun is a video inpainting model based on the Wan2.2 architecture, supporting video inpainting tasks. Wan2.2 adopts a dual-Transformer architecture (high-noise/low-noise models), and the 5B version uses a single-Transformer architecture. This guide covers the LoRA fine-tuning training workflow for Wan2.2 Fun, supporting both A14B and 5B model variants.

---

Table of Contents

• 1. Environment Configuration
• 2. Data Preparation
  • 2.1 Quick Test Dataset
  • 2.2 Dataset Structure
  • 2.3 metadata.json Format
  • 2.4 Relative Path vs Absolute Path Usage
• 3. LoRA Training
  • 3.1 Download Pre-trained Model
  • 3.2 Quick Start (DeepSpeed-Zero-2)
  • 3.3 LoRA-specific Parameter Explanation
  • 3.4 Training Validation
  • 3.5 Training with FSDP
  • 3.6 Other Backends
  • 3.7 Multi-machine Distributed Training
• 4. Inference Testing
  • 4.1 Inference Parameter Explanation
  • 4.2 Inpainting Inference
  • 4.3 Multi-GPU Parallel Inference
• 5. More Resources

---

1. Environment Configuration

Method 1: Using requirements.txt

pip install -r requirements.txt

Method 2: Manual Dependency Installation

pip install Pillow einops safetensors timm tomesd librosa "torch>=2.1.2" torchdiffeq torchsde decord datasets numpy scikit-image
pip install omegaconf SentencePiece imageio[ffmpeg] imageio[pyav] tensorboard beautifulsoup4 ftfy func_timeout onnxruntime
pip install "peft>=0.17.0" "accelerate>=0.25.0" "gradio>=3.41.2" "diffusers>=0.30.1" "transformers>=4.46.2"
pip install yunchang xfuser modelscope openpyxl
pip uninstall opencv-python opencv-contrib-python opencv-python-headless -y
pip install opencv-python-headless
pip install deepspeed==0.17.0 numpy==1.26.4

Method 3: Using Docker

When using Docker, please ensure that the graphics card driver and CUDA environment are correctly installed on your machine, then execute the following commands:

# pull image
docker pull mybigpai-public-registry.cn-beijing.cr.aliyuncs.com/easycv/torch_cuda:cogvideox_fun

# enter image
docker run -it -p 7860:7860 --network host --gpus all --security-opt seccomp:unconfined --shm-size 200g mybigpai-public-registry.cn-beijing.cr.aliyuncs.com/easycv/torch_cuda:cogvideox_fun

---

2. Data Preparation

2.1 Quick Test Dataset

We provide a test dataset containing several training samples.

# Download official example dataset
modelscope download --dataset PAI/X-Fun-Videos-Demo --local_dir ./datasets/X-Fun-Videos-Demo

2.2 Dataset Structure

📦 datasets/
├── 📂 my_dataset/
│   ├── 📂 train/
│   │   ├── 📄 video001.mp4
│   │   ├── 📄 video002.mp4
│   │   └── 📄 ...
│   └── 📄 metadata.json

2.3 metadata.json Format

Relative Path Format (example format):

[
  {
    "file_path": "train/video001.mp4",
    "text": "A beautiful sunset over the ocean, golden hour lighting",
    "type": "video",
    "width": 1024,
    "height": 1024
  },
  {
    "file_path": "train/video002.mp4",
    "text": "A person walking through a forest, cinematic view",
    "type": "video",
    "width": 1328,
    "height": 1328
  }
]

Absolute Path Format:

[
  {
    "file_path": "/mnt/data/videos/sunset.mp4",
    "text": "A beautiful sunset over the ocean",
    "type": "video",
    "width": 1024,
    "height": 1024
  }
]

Key Field Descriptions:

• file_path: Video path (relative or absolute path)
• text: Video description (English prompt)
• type: Data type, fixed as "video"
• width / height: Video width and height (highly recommended to provide, used for bucket training. If not provided, they will be automatically read during training, which may affect training speed when data is stored on slow systems like OSS).
  • You can use scripts/process_json_add_width_and_height.py to extract width and height fields from JSON files without them, supporting both images and videos.
  • Usage: python scripts/process_json_add_width_and_height.py --input_file datasets/X-Fun-Videos-Demo/metadata.json --output_file datasets/X-Fun-Videos-Demo/metadata_add_width_height.json.

2.4 Relative Path vs Absolute Path Usage

Relative Path:

If your data uses relative paths, set in the training script:

export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"

Absolute Path:

If your data uses absolute paths, set in the training script:

export DATASET_NAME=""
export DATASET_META_NAME="/mnt/data/metadata_add_width_height.json"

💡 Suggestion: If the dataset is small and stored locally, relative paths are recommended. If the dataset is stored on external storage (such as NAS, OSS) or shared across multiple machines, absolute paths are recommended.

---

3. LoRA Training

3.1 Download Pre-trained Model

# Create model directory
mkdir -p models/Diffusion_Transformer

# Download Wan2.2 Fun official weights
# A14B model (dual-Transformer architecture)
modelscope download --model PAI/Wan2.2-Fun-A14B-InP --local_dir models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP
# or 5B model (single-Transformer architecture)
# modelscope download --model PAI/Wan2.2-Fun-5B-InP --local_dir models/Diffusion_Transformer/Wan2.2-Fun-5B-InP

3.2 Quick Start (DeepSpeed-Zero-2)

After following 2.1 Quick Test Dataset and 3.1 Download Pre-trained Model, you can directly copy the quick start command to launch training.

DeepSpeed-Zero-2 and FSDP are recommended for training. Here we use DeepSpeed-Zero-2 as an example.

DeepSpeed-Zero-2 vs FSDP:

• Both are distributed training strategies that help reduce GPU memory usage
• DeepSpeed-Zero-2: Optimizer states and gradients are sharded across GPUs
• FSDP (Fully Sharded Data Parallel): Model weights, optimizer states, and gradients are all sharded
• If you encounter insufficient GPU memory with DeepSpeed-Zero-2, switch to FSDP for better memory efficiency

About train_lora.sh: The train_lora.sh script in this directory provides a basic training template without DeepSpeed or FSDP. It's suitable for:

• Single-GPU training
• Quick testing and debugging
• Custom modifications for your specific needs

For production training with multiple GPUs, use the DeepSpeed-Zero-2 or FSDP commands below for better performance and memory efficiency.

Wan2.2 Fun LoRA Training Example (DeepSpeed-Zero-2):

export MODEL_NAME="models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"
export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"
# NCCL_IB_DISABLE=1 and NCCL_P2P_DISABLE=1 are used in multi nodes without RDMA. 
# export NCCL_IB_DISABLE=1
# export NCCL_P2P_DISABLE=1
export NCCL_DEBUG=INFO

accelerate launch --use_deepspeed --deepspeed_config_file config/zero_stage2_config.json --deepspeed_multinode_launcher standard scripts/wan2.2_fun/train_lora.py \
  --config_path="config/wan2.2/wan_civitai_i2v.yaml" \
  --pretrained_model_name_or_path=$MODEL_NAME \
  --train_data_dir=$DATASET_NAME \
  --train_data_meta=$DATASET_META_NAME \
  --image_sample_size=640 \
  --video_sample_size=640 \
  --token_sample_size=640 \
  --video_sample_stride=2 \
  --video_sample_n_frames=81 \
  --train_batch_size=1 \
  --video_repeat=1 \
  --gradient_accumulation_steps=1 \
  --dataloader_num_workers=8 \
  --num_train_epochs=100 \
  --checkpointing_steps=50 \
  --learning_rate=1e-04 \
  --seed=42 \
  --output_dir="output_dir_wan2.2_fun_lora" \
  --gradient_checkpointing \
  --mixed_precision="bf16" \
  --adam_weight_decay=3e-2 \
  --adam_epsilon=1e-10 \
  --vae_mini_batch=1 \
  --max_grad_norm=0.05 \
  --random_hw_adapt \
  --training_with_video_token_length \
  --enable_bucket \
  --uniform_sampling \
  --boundary_type="low" \
  --rank=64 \
  --network_alpha=32 \
  --target_name="q,k,v,ffn.0,ffn.2" \
  --use_peft_lora \
  --train_mode="inpaint" \
  --low_vram

Note: The train_lora.sh script in this directory provides a basic training template without DeepSpeed. For better multi-GPU training performance and memory efficiency, use the DeepSpeed-Zero-2 command above.

3.3 LoRA-specific Parameter Explanation

Wan2.2 Dual-Transformer Architecture Explanation:

Wan2.2 adopts an innovative dual-Transformer architecture:

• Low Noise Model: Responsible for handling the low-noise stage (closer to final output)
• High Noise Model: Responsible for handling the high-noise stage (initial generation stage)
• Boundary Type (boundary_type):
  • low: Train low noise model, high noise model uses pre-trained weights (recommended for T2V/I2V LoRA fine-tuning)
  • high: Train high noise model, low noise model uses pre-trained weights
  • full: Single model training (for single-Transformer models like TI2V-5B)

Key LoRA Parameters:

Parameter	Description	Example Value
--config_path	Configuration file path	config/wan2.2/wan_civitai_i2v.yaml
--pretrained_model_name_or_path	Pre-trained model path	models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP
--train_data_dir	Training data directory	datasets/X-Fun-Videos-Demo/
--train_data_meta	Training data metadata file	datasets/X-Fun-Videos-Demo/metadata_add_width_height.json
--train_batch_size	Number of samples per batch	1
--image_sample_size	Maximum training resolution for images	640
--video_sample_size	Maximum training resolution for videos	640
--token_sample_size	Token sampling size	640
--video_sample_stride	Video sampling stride	2
--video_sample_n_frames	Number of video frames to sample	81
--gradient_accumulation_steps	Gradient accumulation steps (effectively increases batch size)	1
--dataloader_num_workers	Number of DataLoader subprocesses	8
--num_train_epochs	Number of training epochs	100
--checkpointing_steps	Save checkpoint every N steps	50
--learning_rate	Initial learning rate (recommended for LoRA)	1e-04
--lr_scheduler	Learning rate scheduler: linear, cosine, cosine_with_restarts, polynomial, constant, constant_with_warmup	constant
--lr_warmup_steps	Learning rate warmup steps	500
--seed	Random seed (for reproducible training)	42
--output_dir	Output directory	output_dir_wan2.2_fun_lora
--gradient_checkpointing	Activation recomputation to save memory	-
--mixed_precision	Mixed precision: no, fp16, bf16	bf16
--adam_weight_decay	AdamW weight decay	3e-2
--adam_epsilon	AdamW epsilon value	1e-10
--vae_mini_batch	Mini-batch size for VAE encoding	1
--max_grad_norm	Gradient clipping threshold	0.05
--enable_bucket	Enable bucket training, no cropping, group by resolution	-
--random_hw_adapt	Auto-scale images/videos to random sizes within [min_size, max_size]	-
--training_with_video_token_length	Train based on token length, supports arbitrary resolutions	-
--uniform_sampling	Uniform timestep sampling (recommended)	-
--low_vram	Low VRAM mode for memory efficiency	-
--boundary_type	Wan2.2 dual-Transformer boundary type: low (train low noise model), high (train high noise model), full (train single model like TI2V-5B)	low
--train_mode	Training mode: normal (T2V), inpaint (video inpainting)	inpaint
--resume_from_checkpoint	Resume training path, use "latest" to auto-select latest checkpoint	None
--rank	LoRA update matrix dimension (higher rank = stronger expression but more memory)	64
--network_alpha	LoRA update matrix scaling factor (usually set to half of rank or same)	32
--target_name	Components/modules to apply LoRA, comma-separated (e.g., q,k,v,ffn.0,ffn.2)	q,k,v,ffn.0,ffn.2
--lora_skip_name	Components to skip in LoRA training, comma-separated	None
--use_peft_lora	Use PEFT module to add LoRA (more memory-efficient)	-
--validation_steps	Run validation every N steps	2000
--validation_epochs	Run validation every N epochs	5
--validation_prompts	Prompts for validating video generation	"A brown dog shaking head..."
--validation_paths	Reference image paths for I2V validation (inpaint mode only)	"asset/1.png"
--use_deepspeed	Enable DeepSpeed for distributed training	-
--use_fsdp	Enable FSDP for distributed training	-
--use_8bit_adam	Use 8-bit Adam optimizer to save memory	-
--use_came	Use CAME optimizer	-
--multi_stream	Use CUDA multi-stream for performance	-
--snr_loss	Use SNR loss function	-
--weighting_scheme	Timestep weighting scheme: sigma_sqrt, logit_normal, mode, cosmap, none	none
--motion_sub_loss	Enable motion sub-loss for better temporal consistency	-
--motion_sub_loss_ratio	Motion sub-loss ratio	0.25

Sample Size Configuration Guide:

• video_sample_size represents the resolution size of videos; when random_hw_adapt is True, it represents the minimum value between video and image resolutions.
• image_sample_size represents the resolution size of images; when random_hw_adapt is True, it represents the maximum value between video and image resolutions.
• token_sample_size represents the resolution corresponding to the maximum token length when training_with_video_token_length is True.
• Due to potential confusion in configuration, if you don't require arbitrary resolution for finetuning, it is recommended to set video_sample_size, image_sample_size, and token_sample_size to the same fixed value, such as (320, 480, 512, 640, 960).
  • All set to 320 represents 240P.
  • All set to 480 represents 320P.
  • All set to 640 represents 480P.
  • All set to 960 represents 720P.

Token Length Training Explanation:

• When training_with_video_token_length is enabled, the model trains based on token length.
• For example: A video with 512x512 resolution and 49 frames has a token length of 13,312, requiring token_sample_size = 512.
  • At 512x512 resolution, the number of video frames is 49 (~= 512 * 512 * 49 / 512 / 512).
  • At 768x768 resolution, the number of video frames is 21 (~= 512 * 512 * 49 / 768 / 768).
  • At 1024x1024 resolution, the number of video frames is 9 (~= 512 * 512 * 49 / 1024 / 1024).
  • These resolutions combined with their corresponding frame counts allow the model to generate videos of different sizes.

3.4 Training Validation

You can configure validation parameters to periodically generate test videos during training, allowing you to monitor training progress and model quality.

Validation Parameters:

Parameter	Description	Default Value
--validation_steps	Run validation every N steps	2000
--validation_epochs	Run validation every N epochs	5
--validation_prompts	Prompts for video generation validation	None
--validation_paths	Reference image paths for I2V validation (inpaint mode only)	None

Validation Example (with inpaint mode):

  --validation_paths "asset/1.png" \
  --validation_steps=100 \
  --validation_epochs=100 \
  --validation_prompts="A brown dog shaking its head, sitting on a light-colored sofa in a cozy room. Behind the dog, there is a framed painting on a shelf, surrounded by pink flowers. The soft, warm lighting in the room creates a comfortable atmosphere."

Notes:

• Validation videos are saved to the output_dir directory
• Multi-prompt validation format: --validation_prompts "prompt1" "prompt2" "prompt3"
• Wan2.2 Fun validation automatically selects single or dual-Transformer based on boundary_type
• Validation uses inpainting mode with reference image as the starting frame when train_mode="inpaint"
• For T2V validation (without reference image), set train_mode="normal" and omit --validation_paths

3.5 Training with FSDP

If you encounter insufficient GPU memory when using multiple GPUs with DeepSpeed-Zero-2, you can switch to FSDP for training.

export MODEL_NAME="models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"
export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"
# NCCL_IB_DISABLE=1 and NCCL_P2P_DISABLE=1 are used in multi nodes without RDMA. 
# export NCCL_IB_DISABLE=1
# export NCCL_P2P_DISABLE=1
export NCCL_DEBUG=INFO

accelerate launch --mixed_precision="bf16" --use_fsdp --fsdp_auto_wrap_policy TRANSFORMER_BASED_WRAP --fsdp_transformer_layer_cls_to_wrap=WanAttentionBlock --fsdp_sharding_strategy "FULL_SHARD" --fsdp_state_dict_type=SHARDED_STATE_DICT --fsdp_backward_prefetch "BACKWARD_PRE" --fsdp_cpu_ram_efficient_loading False scripts/wan2.2_fun/train_lora.py \
  --config_path="config/wan2.2/wan_civitai_i2v.yaml" \
  --pretrained_model_name_or_path=$MODEL_NAME \
  --train_data_dir=$DATASET_NAME \
  --train_data_meta=$DATASET_META_NAME \
  --image_sample_size=640 \
  --video_sample_size=640 \
  --token_sample_size=640 \
  --video_sample_stride=2 \
  --video_sample_n_frames=81 \
  --train_batch_size=1 \
  --video_repeat=1 \
  --gradient_accumulation_steps=1 \
  --dataloader_num_workers=8 \
  --num_train_epochs=100 \
  --checkpointing_steps=50 \
  --learning_rate=1e-04 \
  --seed=42 \
  --output_dir="output_dir_wan2.2_fun_lora" \
  --gradient_checkpointing \
  --mixed_precision="bf16" \
  --adam_weight_decay=3e-2 \
  --adam_epsilon=1e-10 \
  --vae_mini_batch=1 \
  --max_grad_norm=0.05 \
  --random_hw_adapt \
  --training_with_video_token_length \
  --enable_bucket \
  --uniform_sampling \
  --boundary_type="low" \
  --rank=64 \
  --network_alpha=32 \
  --target_name="q,k,v,ffn.0,ffn.2" \
  --use_peft_lora \
  --low_vram \
  --train_mode="inpaint"

Note: FSDP is more stable in this repository and has fewer errors compared to DeepSpeed-Zero-3. Use FSDP when DeepSpeed-Zero-2 encounters memory issues with multiple GPUs.

3.6 Other Backends

3.6.1 Training with DeepSpeed-Zero-3

DeepSpeed Zero-3 is not highly recommended at the moment. In this repository, using FSDP has fewer errors and is more stable.

DeepSpeed Zero-3 is suitable for high-resolution 14B Wan. After training, you can use the following command to obtain the final model:

python scripts/zero_to_bf16.py output_dir/checkpoint-{our-num-steps} output_dir/checkpoint-{your-num-steps}-outputs --max_shard_size 80GB --safe_serialization

Training shell command is as follows:

export MODEL_NAME="models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"
export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"
# NCCL_IB_DISABLE=1 and NCCL_P2P_DISABLE=1 are used in multi nodes without RDMA. 
# export NCCL_IB_DISABLE=1
# export NCCL_P2P_DISABLE=1
export NCCL_DEBUG=INFO

accelerate launch --zero_stage 3 --zero3_save_16bit_model true --zero3_init_flag true --use_deepspeed --deepspeed_config_file config/zero_stage3_config.json --deepspeed_multinode_launcher standard scripts/wan2.2_fun/train_lora.py \
  --config_path="config/wan2.2/wan_civitai_i2v.yaml" \
  --pretrained_model_name_or_path=$MODEL_NAME \
  --train_data_dir=$DATASET_NAME \
  --train_data_meta=$DATASET_META_NAME \
  --image_sample_size=640 \
  --video_sample_size=640 \
  --token_sample_size=640 \
  --video_sample_stride=2 \
  --video_sample_n_frames=81 \
  --train_batch_size=1 \
  --video_repeat=1 \
  --gradient_accumulation_steps=1 \
  --dataloader_num_workers=8 \
  --num_train_epochs=100 \
  --checkpointing_steps=50 \
  --learning_rate=1e-04 \
  --seed=42 \
  --output_dir="output_dir_wan2.2_fun_lora" \
  --gradient_checkpointing \
  --mixed_precision="bf16" \
  --adam_weight_decay=3e-2 \
  --adam_epsilon=1e-10 \
  --vae_mini_batch=1 \
  --max_grad_norm=0.05 \
  --random_hw_adapt \
  --training_with_video_token_length \
  --enable_bucket \
  --uniform_sampling \
  --boundary_type="low" \
  --rank=64 \
  --network_alpha=32 \
  --target_name="q,k,v,ffn.0,ffn.2" \
  --use_peft_lora \
  --low_vram \
  --train_mode="inpaint"

3.6.2 Training without DeepSpeed and FSDP

This approach is not recommended, as without memory-saving backends, it easily causes out-of-memory errors. Only provided here for reference.

export MODEL_NAME="models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"
export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"
# NCCL_IB_DISABLE=1 and NCCL_P2P_DISABLE=1 are used in multi nodes without RDMA. 
# export NCCL_IB_DISABLE=1
# export NCCL_P2P_DISABLE=1
export NCCL_DEBUG=INFO

accelerate launch --mixed_precision="bf16" scripts/wan2.2_fun/train_lora.py \
  --config_path="config/wan2.2/wan_civitai_i2v.yaml" \
  --pretrained_model_name_or_path=$MODEL_NAME \
  --train_data_dir=$DATASET_NAME \
  --train_data_meta=$DATASET_META_NAME \
  --image_sample_size=640 \
  --video_sample_size=640 \
  --token_sample_size=640 \
  --video_sample_stride=2 \
  --video_sample_n_frames=81 \
  --train_batch_size=1 \
  --video_repeat=1 \
  --gradient_accumulation_steps=1 \
  --dataloader_num_workers=8 \
  --num_train_epochs=100 \
  --checkpointing_steps=50 \
  --learning_rate=1e-04 \
  --seed=42 \
  --output_dir="output_dir_wan2.2_fun_lora" \
  --gradient_checkpointing \
  --mixed_precision="bf16" \
  --adam_weight_decay=3e-2 \
  --adam_epsilon=1e-10 \
  --vae_mini_batch=1 \
  --max_grad_norm=0.05 \
  --random_hw_adapt \
  --training_with_video_token_length \
  --enable_bucket \
  --uniform_sampling \
  --boundary_type="low" \
  --rank=64 \
  --network_alpha=32 \
  --target_name="q,k,v,ffn.0,ffn.2" \
  --use_peft_lora \
  --low_vram \
  --train_mode="inpaint"

Note: This is similar to the train_lora.sh script but with the correct dataset paths. The train_lora.sh script can be used as a starting point for single-GPU training.

3.7 Multi-machine Distributed Training

Suitable for: Ultra-large-scale datasets, faster training speed

3.7.1 Environment Configuration

Assuming 2 machines, each with 8 GPUs:

Machine 0 (Master):

export MODEL_NAME="models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"
export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"
export MASTER_ADDR="192.168.1.100"  # Master machine IP
export MASTER_PORT=10086
export WORLD_SIZE=2                  # Total number of machines
export NUM_PROCESS=16                # Total processes = machines × 8
export RANK=0                        # Current machine rank (0 or 1)
# NCCL_IB_DISABLE=1 and NCCL_P2P_DISABLE=1 are used in multi nodes without RDMA. 
# export NCCL_IB_DISABLE=1
# export NCCL_P2P_DISABLE=1
export NCCL_DEBUG=INFO

accelerate launch --mixed_precision="bf16" --main_process_ip=$MASTER_ADDR --main_process_port=$MASTER_PORT --num_machines=$WORLD_SIZE --num_processes=$NUM_PROCESS --machine_rank=$RANK --use_deepspeed --deepspeed_config_file config/zero_stage2_config.json --deepspeed_multinode_launcher standard scripts/wan2.2_fun/train_lora.py \
  --config_path="config/wan2.2/wan_civitai_i2v.yaml" \
  --pretrained_model_name_or_path=$MODEL_NAME \
  --train_data_dir=$DATASET_NAME \
  --train_data_meta=$DATASET_META_NAME \
  --image_sample_size=640 \
  --video_sample_size=640 \
  --token_sample_size=640 \
  --video_sample_stride=2 \
  --video_sample_n_frames=81 \
  --train_batch_size=1 \
  --video_repeat=1 \
  --gradient_accumulation_steps=1 \
  --dataloader_num_workers=8 \
  --num_train_epochs=100 \
  --checkpointing_steps=50 \
  --learning_rate=1e-04 \
  --seed=42 \
  --output_dir="output_dir_wan2.2_fun_lora" \
  --gradient_checkpointing \
  --mixed_precision="bf16" \
  --adam_weight_decay=3e-2 \
  --adam_epsilon=1e-10 \
  --vae_mini_batch=1 \
  --max_grad_norm=0.05 \
  --random_hw_adapt \
  --training_with_video_token_length \
  --enable_bucket \
  --uniform_sampling \
  --boundary_type="low" \
  --rank=64 \
  --network_alpha=32 \
  --target_name="q,k,v,ffn.0,ffn.2" \
  --use_peft_lora \
  --low_vram \
  --train_mode="inpaint"

Machine 1 (Worker):

export MODEL_NAME="models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"
export DATASET_NAME="datasets/X-Fun-Videos-Demo/"
export DATASET_META_NAME="datasets/X-Fun-Videos-Demo/metadata_add_width_height.json"
export MASTER_ADDR="192.168.1.100"  # Same as Master
export MASTER_PORT=10086
export WORLD_SIZE=2
export NUM_PROCESS=16
export RANK=1  # Note: this is 1
# NCCL_IB_DISABLE=1 and NCCL_P2P_DISABLE=1 are used in multi nodes without RDMA. 
# export NCCL_IB_DISABLE=1
# export NCCL_P2P_DISABLE=1
export NCCL_DEBUG=INFO

# Use the same accelerate launch command as Machine 0

3.7.2 Multi-machine Training Notes

• Network Requirements:

  • RDMA/InfiniBand recommended (high performance)
  • Without RDMA, add environment variables:

    export NCCL_IB_DISABLE=1
    export NCCL_P2P_DISABLE=1
    

• Data Synchronization: All machines must be able to access the same data path (NFS/shared storage)

---

4. Inference Testing

4.1 Inference Parameter Explanation

Key Parameters:

Parameter	Description	Example Value
GPU_memory_mode	Memory management mode, see table below for options	model_group_offload
ulysses_degree	Head dimension parallelism degree, 1 for single GPU	1
ring_degree	Sequence dimension parallelism degree, 1 for single GPU	1
fsdp_dit	Use FSDP for Transformer during multi-GPU inference to save memory	False
fsdp_text_encoder	Use FSDP for text encoder during multi-GPU inference	True
compile_dit	Compile Transformer for faster inference (effective for fixed resolution)	False
model_name	Model path	models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP
sampler_name	Sampler type: Flow, Flow_Unipc, Flow_DPM++	Flow_Unipc
transformer_path	Path to load trained low-noise Transformer weights	None
transformer_high_path	Path to load trained high-noise Transformer weights (dual-Transformer models only)	None
vae_path	Path to load trained VAE weights	None
lora_path	Low-noise model LoRA weights path	None
lora_high_path	High-noise model LoRA weights path (dual-Transformer models only)	None
sample_size	Generated video resolution [height, width]	[480, 832] or [832, 480]
video_length	Number of video frames	81
fps	Frames per second	16
weight_dtype	Model weight precision, use torch.float16 for GPUs without bf16 support	torch.bfloat16
validation_image_start	Reference image path for image-to-video (I2V mode)	"asset/1.png"
validation_image_end	Reference image path for end frame (optional)	None
validation_mask	Mask path for inpainting (inpaint mode)	None
prompt	Positive prompt describing generated content	"A brown dog shaking head..."
negative_prompt	Negative prompt to avoid certain content	"low resolution, low quality..."
guidance_scale	Guidance strength	6.0
seed	Random seed for reproducibility	43
num_inference_steps	Number of inference steps	50
lora_weight	Low-noise model LoRA weight strength	0.55
lora_high_weight	High-noise model LoRA weight strength (dual-Transformer models only)	0.55
save_path	Path to save generated videos	samples/wan-videos-i2v or samples/wan-videos-t2v

Memory Management Modes:

Mode	Description	Memory Usage
model_full_load	Entire model loaded to GPU	Highest
model_full_load_and_qfloat8	Full load + FP8 quantization	High
model_cpu_offload	Offload model to CPU after use	Medium
model_cpu_offload_and_qfloat8	CPU offload + FP8 quantization	Medium-Low
model_group_offload	Layer groups switch between CPU/CUDA	Low
sequential_cpu_offload	Layer-by-layer offload (slowest)	Lowest

4.2 Inpainting Inference

4.2.1 Inference Script Selection

Wan2.2 Fun provides multiple inference scripts. Choose based on your model version and task type:

Script	Model Version	Architecture	Primary Use
predict_i2v.py	A14B	Dual-Transformer	Image-to-Video/Inpainting (I2V/Inpaint)
predict_t2v.py	A14B	Dual-Transformer	Text-to-Video (T2V)
predict_i2v_5b.py	5B	Single-Transformer	Image-to-Video/Inpainting (I2V/Inpaint)
predict_t2v_5b.py	5B	Single-Transformer	Text-to-Video (T2V)

Note:

• A14B model uses dual-Transformer architecture (low-noise + high-noise models), requiring both transformer_path and transformer_high_path
• 5B model uses single-Transformer architecture, only transformer_path is needed, keep transformer_high_path as None

4.2.2 A14B Model Inference (Dual-Transformer)

Run the following command for single-GPU inference:

python examples/wan2.2_fun/predict_i2v.py

Modify examples/wan2.2_fun/predict_i2v.py according to your needs. For first-time inference, focus on the parameters below. If you're interested in other parameters, refer to the inference parameter explanation above.

# Choose based on GPU memory
GPU_memory_mode = "sequential_cpu_offload"
# Based on actual model path
model_name = "models/Diffusion_Transformer/Wan2.2-Fun-A14B-InP"  
# Path to trained low-noise weights
transformer_path = None  
# Path to trained high-noise weights
transformer_high_path = None  
# Path to trained LoRA weights, e.g., "output_dir_wan2.2_fun_lora/checkpoint-xxx/diffusion_pytorch_model.safetensors"
lora_path = None
lora_high_path = None
# Starting image for inpainting
validation_image_start = "asset/1.png"
# Mask for inpainting (optional, will be auto-generated if not provided)
validation_mask = None
# Write based on generated content
prompt = "A brown dog shaking its head, sitting on a light-colored sofa in a cozy room. Behind the dog, there is a framed painting on a shelf, surrounded by pink flowers. The soft, warm lighting in the room creates a comfortable atmosphere."  
# ...

Note: Wan2.2 Fun is primarily designed for video inpainting tasks. Use predict_i2v.py for both I2V and inpainting workflows. The model will automatically handle inpainting when a mask is provided or generated.

4.2.3 5B Model Inference (Single-Transformer)

Run the following command for single-GPU inference:

python examples/wan2.2_fun/predict_i2v_5b.py

Modify examples/wan2.2_fun/predict_i2v_5b.py according to your needs, focusing on the parameters below:

# Choose based on GPU memory
GPU_memory_mode = "sequential_cpu_offload"
# 5B model path
model_name = "models/Diffusion_Transformer/Wan2.2-Fun-5B-InP/"  
# Path to trained weights (5B is single-Transformer, only set transformer_path)
transformer_path = None  
# 5B model doesn't use high-noise Transformer, keep as None
transformer_high_path = None  
# Path to trained LoRA weights, e.g., "output_dir_wan2.2_fun_lora/checkpoint-xxx/diffusion_pytorch_model.safetensors"
lora_path = None
# 5B model doesn't use high-noise LoRA, keep as None
lora_high_path = None
# Starting image for inpainting
validation_image_start = "asset/1.png"
validation_image_end = None
# Write based on generated content
prompt = "A brown dog licking its tongue, sitting on a light-colored sofa in a cozy room. Behind the dog, there is a framed painting on a shelf, surrounded by pink flowers. The soft, warm lighting in the room creates a comfortable atmosphere."  
# ...

Note:

• 5B model uses single-Transformer architecture with simpler configuration and lower memory usage
• If you trained with boundary_type="full", only load transformer_path during inference, no need to set transformer_high_path
• For LoRA training, only set lora_path, keep lora_high_path as None

4.3 Multi-GPU Parallel Inference

Suitable for: High-resolution generation, accelerated inference

Install Parallel Inference Dependencies

pip install xfuser==0.4.2 yunchang==0.6.2

Configure Parallel Strategy

Edit examples/wan2.2_fun/predict_i2v.py:

``$\text{python}

\text{Ensure} \text{ulysses_degree} \times \text{ring_degree} = \text{number} \text{of} \text{GPUs} \text{used}

\text{For} \text{example}, \text{using} 2 \text{GPUs}:

\text{ulysses_degree} = 2 # \text{Head} \text{dimension} \text{parallelism} \text{ring_degree} = 1 # \text{Sequence} \text{dimension} \text{parallelism} $``

Configuration Principles:

• ulysses_degree must evenly divide the model's head count
• ring_degree splits along the sequence dimension, which affects communication overhead. Avoid using it when heads can be evenly divided.

Configuration Examples:

GPU Count	ulysses_degree	ring_degree	Description
1	1	1	Single GPU
4	4	1	Head parallelism
8	8	1	Head parallelism
8	4	2	Hybrid parallelism

Run Multi-GPU Inference

torchrun --nproc-per-node=2 examples/wan2.2_fun/predict_i2v.py

---

5. More Resources

• Official GitHub: https://github.com/aigc-apps/VideoX-Fun