Wan

Alibaba (Wan-Video team)

Structures found in this family (3)

Anisotropy, Linear Separability, Linear Subspace

By model (3)

Wan2.1-1.3B

Anisotropy, Linear Separability

Wan2.1-T2V-14B

Linear Subspace

Wan2.2-5B

Anisotropy

Observations (3)

Video-DiT massive activations follow a token-position hierarchy

Inverting video diffusion transformers makes physical plausibility linearly decodable

An SVD contrastive subspace enables closed-loop control of video diffusion

Papers

Steering Video Diffusion Transformers with Massive Activations (2026), The Invisible Hand of Physics: When Video Diffusion Models Know More Than They Show (2026), Activation Steering of Video Generation Models via Reduced-Order Linear Optimal Control (2026)