Diffusers

You are viewing main version, which requires installation from source. If you'd like regular pip install, checkout the latest stable version (v0.42.1).

Join the Hugging Face community

and get access to the augmented documentation experience

Collaborate on models, datasets and Spaces

Faster examples with accelerated inference

Switch between documentation themes

to get started

LatteTransformer3DModel

Un modelo Diffusion Transformer para datos 3D de Latte.

LatteTransformer3DModel

clase diffusers.LatteTransformer3DModel

< source >

( num_attention_heads: int = 16attention_head_dim: int = 88in_channels: int | None = Noneout_channels: int | None = Nonenum_layers: int = 1dropout: float = 0.0cross_attention_dim: int | None = Noneattention_bias: bool = Falsesample_size: int = 64patch_size: int | None = Noneactivation_fn: str = 'geglu'num_embeds_ada_norm: int | None = Nonenorm_type: str = 'layer_norm'norm_elementwise_affine: bool = Truenorm_eps: float = 1e-05caption_channels: int = Nonevideo_length: int = 16 )

forward

< source >

( hidden_states: Tensortimestep: typing.Optional[torch.LongTensor] = Noneencoder_hidden_states: typing.Optional[torch.Tensor] = Noneencoder_attention_mask: typing.Optional[torch.Tensor] = Noneenable_temporal_attentions: bool = Truereturn_dict: bool = True )

Parámetros

hidden_states (torch.Tensor de forma (batch size, channel, num_frame, height, width)) — Entrada hidden_states.
timestep ( torch.LongTensor, opcional) — Se utiliza para indicar el paso de desruido. Paso temporal opcional que se aplica como un embedding en AdaLayerNorm.
encoder_hidden_states ( torch.FloatTensor de forma (batch size, sequence len, embed dims), opcional) — Incrustados condicionales para la capa de atención cruzada. Si no se proporcionan, la atención cruzada predeterminada es la autoatención.
encoder_attention_mask ( torch.Tensor, opcional) — Máscara de atención cruzada aplicada a encoder_hidden_states. Se admiten dos formatos:
- Máscara (batcheight, sequence_length) True = conservar, False = descartar.
- Sesgo (batcheight, 1, sequence_length) 0 = conservar, -10000 = descartar.
Si ndim == 2: se interpretará como una máscara y luego se convertirá en un sesgo coherente con el formato anterior. Este sesgo se sumará a las puntuaciones de atención cruzada.
enable_temporal_attentions — (bool, opcional, valor predeterminado True): Si se habilitan las atenciones temporales.
return_dict (bool, opcional, valor predeterminado True) — Indica si se debe devolver un ~models.unet_2d_condition.UNet2DConditionOutput en lugar de una tupla simple.

El método forward de LatteTransformer3DModel.

Update on GitHub

←Krea2Transformer2DModel LongCatImageTransformer2DModel→