diffusers/pr_12595/en/api/models/wan_transformer_3d.md

WanTransformer3DModel

A Diffusion Transformer model for 3D video-like data was introduced in Wan 2.1 by the Alibaba Wan Team.

The model can be loaded with the following code snippet.

from diffusers import WanTransformer3DModel

transformer = WanTransformer3DModel.from_pretrained("Wan-AI/Wan2.1-T2V-1.3B-Diffusers", subfolder="transformer", torch_dtype=torch.bfloat16)

WanTransformer3DModel[[diffusers.WanTransformer3DModel]]

class diffusers.WanTransformer3DModeldiffusers.WanTransformer3DModelhttps://github.com/huggingface/diffusers/blob/vr_12595/src/diffusers/models/transformers/transformer_wan.py#L501[{"name": "patch_size", "val": ": typing.Tuple[int] = (1, 2, 2)"}, {"name": "num_attention_heads", "val": ": int = 40"}, {"name": "attention_head_dim", "val": ": int = 128"}, {"name": "in_channels", "val": ": int = 16"}, {"name": "out_channels", "val": ": int = 16"}, {"name": "text_dim", "val": ": int = 4096"}, {"name": "freq_dim", "val": ": int = 256"}, {"name": "ffn_dim", "val": ": int = 13824"}, {"name": "num_layers", "val": ": int = 40"}, {"name": "cross_attn_norm", "val": ": bool = True"}, {"name": "qk_norm", "val": ": typing.Optional[str] = 'rms_norm_across_heads'"}, {"name": "eps", "val": ": float = 1e-06"}, {"name": "image_dim", "val": ": typing.Optional[int] = None"}, {"name": "added_kv_proj_dim", "val": ": typing.Optional[int] = None"}, {"name": "rope_max_seq_len", "val": ": int = 1024"}, {"name": "pos_embed_seq_len", "val": ": typing.Optional[int] = None"}]- patch_size (Tuple[int], defaults to (1, 2, 2)) -- 3D patch dimensions for video embedding (t_patch, h_patch, w_patch).

• num_attention_heads (int, defaults to 40) -- Fixed length for text embeddings.
• attention_head_dim (int, defaults to 128) -- The number of channels in each head.
• in_channels (int, defaults to 16) -- The number of channels in the input.
• out_channels (int, defaults to 16) -- The number of channels in the output.
• text_dim (int, defaults to 512) -- Input dimension for text embeddings.
• freq_dim (int, defaults to 256) -- Dimension for sinusoidal time embeddings.
• ffn_dim (int, defaults to 13824) -- Intermediate dimension in feed-forward network.
• num_layers (int, defaults to 40) -- The number of layers of transformer blocks to use.
• window_size (Tuple[int], defaults to (-1, -1)) -- Window size for local attention (-1 indicates global attention).
• cross_attn_norm (bool, defaults to True) -- Enable cross-attention normalization.
• qk_norm (bool, defaults to True) -- Enable query/key normalization.
• eps (float, defaults to 1e-6) -- Epsilon value for normalization layers.
• add_img_emb (bool, defaults to False) -- Whether to use img_emb.
• added_kv_proj_dim (int, optional, defaults to None) -- The number of channels to use for the added key and value projections. If None, no projection is used.0

A Transformer model for video-like data used in the Wan model.

Transformer2DModelOutput[[diffusers.models.modeling_outputs.Transformer2DModelOutput]]

class diffusers.models.modeling_outputs.Transformer2DModelOutputdiffusers.models.modeling_outputs.Transformer2DModelOutputhttps://github.com/huggingface/diffusers/blob/vr_12595/src/diffusers/models/modeling_outputs.py#L21[{"name": "sample", "val": ": torch.Tensor"}]- sample (torch.Tensor of shape (batch_size, num_channels, height, width) or (batch size, num_vector_embeds - 1, num_latent_pixels) if Transformer2DModel is discrete) -- The hidden states output conditioned on the encoder_hidden_states input. If discrete, returns probability distributions for the unnoised latent pixels.0

The output of Transformer2DModel.