A Diffusion Transformer model for 3D data from Latte.
( num_attention_heads: int = 16 attention_head_dim: int = 88 in_channels: typing.Optional[int] = None out_channels: typing.Optional[int] = None num_layers: int = 1 dropout: float = 0.0 cross_attention_dim: typing.Optional[int] = None attention_bias: bool = False sample_size: int = 64 patch_size: typing.Optional[int] = None activation_fn: str = 'geglu' num_embeds_ada_norm: typing.Optional[int] = None norm_type: str = 'layer_norm' norm_elementwise_affine: bool = True norm_eps: float = 1e-05 caption_channels: int = None video_length: int = 16 )
( hidden_states: Tensor timestep: typing.Optional[torch.LongTensor] = None encoder_hidden_states: typing.Optional[torch.Tensor] = None encoder_attention_mask: typing.Optional[torch.Tensor] = None enable_temporal_attentions: bool = True return_dict: bool = True )
Parameters
(batch size, channel, num_frame, height, width) —
Input hidden_states. torch.LongTensor, optional) —
Used to indicate denoising step. Optional timestep to be applied as an embedding in AdaLayerNorm. torch.FloatTensor of shape (batch size, sequence len, embed dims), optional) —
Conditional embeddings for cross attention layer. If not given, cross-attention defaults to
self-attention. torch.Tensor, optional) —
Cross-attention mask applied to encoder_hidden_states. Two formats supported:
(batcheight, sequence_length) True = keep, False = discard.(batcheight, 1, sequence_length) 0 = keep, -10000 = discard.If ndim == 2: will be interpreted as a mask, then converted into a bias consistent with the format
above. This bias will be added to the cross-attention scores.
bool, optional, defaults to True): Whether to enable temporal attentions. bool, optional, defaults to True) —
Whether or not to return a ~models.unet_2d_condition.UNet2DConditionOutput instead of a plain
tuple. The LatteTransformer3DModel forward method.