Diffusion forward corruption process. Input: - clean_latent: the clean latent with shape [B, C, H, W] - noise

Update the current scheduler with the interface's static method

q: [B, Lq, Nq, C1]. k: [B, Lk, Nk, C1]. v: [B, Lk, Nk, C2]. Nq must be divisible by Nk. q_lens

Helper function for reproducible behavior to set the seed in `random`, `numpy`, `torch`. Args: seed (`int`): The seed to

Configure LoRA for a WanDiffusionWrapper model

Randomly generate a timestep tensor based on the generator's task type. It uniformly samples a timestep from the range [min_timestep,

Initialize a Per-GPU cross-attention cache for the Wan model.

Lazy initialize the inference pipeline during the first backward simulation run. Here we encapsulate the inference code with a model-

Initialize a Per-GPU KV cache for the Wan model.

Optionally simulate the generator's input from noise using backward simulation and then run the generator for one-step. Input

"Dynamic thresholding: At each sampling step we set s to a certain percentile absolute pixel value in xt0 (the prediction of x_0 at t

Autoencoder3d adapted from Stable Diffusion 1.x, 2.x and XL.

r""" Args: x(Tensor): Shape [B, L1, C] e(Tensor): Shape [B, L1, C]

Sets the discrete timesteps used for the diffusion chain (to be run before inference). Args: num_inference_steps (`int`):

Input: - timestep: the timestep with shape [B*T] Output: the corresponding weighting [B*T]

r""" Reconstruct video tensors from patch embeddings. Args: x (List[Tensor]): List of patchified features

Compute the KL grad (eq 7 in https://arxiv.org/abs/2311.18828). Input: - noisy_image_or_video: a tensor with shape [B, F,

Configure LoRA for a WanDiffusionWrapper model

Configure LoRA for a WanDiffusionWrapper model

Configure LoRA for a WanDiffusionWrapper model

Configure LoRA for a WanDiffusionWrapper model

Configure LoRA for a WanDiffusionWrapper model

Simulate the generator's input from noise to avoid training/inference mismatch. See Sec 4.5 of the DMD2 paper (https://arxiv.org/abs/

Convert flow matching's prediction to x0 prediction. flow_pred: the prediction with shape [B, C, H, W] xt: the input noisy da

r""" Forward pass through the diffusion model Args: x (List[Tensor]): List of input video tensors, each w

r""" Run the diffusion model with kv caching. See Algorithm 2 of CausVid paper https://arxiv.org/abs/2412.07772 for details. T

r""" Forward pass through the diffusion model Args: x (List[Tensor]): List of input video tensors, each w

Randomly generate a timestep tensor based on the generator's task type. It uniformly samples a timestep from the range [min_timestep,

Initialize the step_index counter for the scheduler.

Initialize a Per-GPU cross-attention cache for the Wan model.

Initialize a Per-GPU cross-attention cache for the Wan model.

Initialize a Per-GPU KV cache for the Wan model.

Initialize a Per-GPU KV cache for the Wan model.

we will divide the token sequence into the following format [1 latent frame] [1 latent frame] ... [1 latent frame] We use fle

Compute the DMD loss (eq 7 in https://arxiv.org/abs/2311.18828). Input: - image_or_video: a tensor with shape [B, F, C, H