Skip to content

Latest commit

 

History

History
36 lines (26 loc) · 1.59 KB

File metadata and controls

36 lines (26 loc) · 1.59 KB

AnyFlowTransformer3DModel

The bidirectional 3D Transformer used by AnyFlowPipeline. It is the v0.35.1 Wan2.1 backbone with one structural change: the timestep embedder is replaced by AnyFlowDualTimestepTextImageEmbedding, so every forward call conditions on both the source timestep t and the target timestep r. This is the embedding required to learn the flow map :math:\Phi_{r\leftarrow t} introduced in AnyFlow (Yuchao Gu, Guian Fang et al., NUS ShowLab × NVIDIA).

For frame-level autoregressive (FAR causal) generation, use AnyFlowFARTransformer3DModel instead.

from diffusers import AnyFlowTransformer3DModel

# Bidirectional AnyFlow checkpoint (T2V):
transformer = AnyFlowTransformer3DModel.from_pretrained(
    "nvidia/AnyFlow-Wan2.1-T2V-1.3B-Diffusers", subfolder="transformer"
)

AnyFlowTransformer3DModel

[[autodoc]] AnyFlowTransformer3DModel