Diffusers documentation


Hugging Face's logo
Join the Hugging Face community

and get access to the augmented documentation experience

to get started


A Transformer model for image-like data from CompVis that is based on the Vision Transformer introduced by Dosovitskiy et al. The Transformer2DModel accepts discrete (classes of vector embeddings) or continuous (actual embeddings) inputs.

When the input is continuous:

  1. Project the input and reshape it to (batch_size, sequence_length, feature_dimension).
  2. Apply the Transformer blocks in the standard way.
  3. Reshape to image.

When the input is discrete:

It is assumed one of the input classes is the masked latent pixel. The predicted classes of the unnoised image don’t contain a prediction for the masked pixel because the unnoised image cannot be masked.

  1. Convert input (classes of latent pixels) to embeddings and apply positional embeddings.
  2. Apply the Transformer blocks in the standard way.
  3. Predict classes of unnoised image.


class diffusers.Transformer2DModel

< >

( num_attention_heads: int = 16 attention_head_dim: int = 88 in_channels: Optional = None out_channels: Optional = None num_layers: int = 1 dropout: float = 0.0 norm_num_groups: int = 32 cross_attention_dim: Optional = None attention_bias: bool = False sample_size: Optional = None num_vector_embeds: Optional = None patch_size: Optional = None activation_fn: str = 'geglu' num_embeds_ada_norm: Optional = None use_linear_projection: bool = False only_cross_attention: bool = False double_self_attention: bool = False upcast_attention: bool = False norm_type: str = 'layer_norm' norm_elementwise_affine: bool = True norm_eps: float = 1e-05 attention_type: str = 'default' caption_channels: int = None interpolation_scale: float = None use_additional_conditions: Optional = None )


  • num_attention_heads (int, optional, defaults to 16) — The number of heads to use for multi-head attention.
  • attention_head_dim (int, optional, defaults to 88) — The number of channels in each head.
  • in_channels (int, optional) — The number of channels in the input and output (specify if the input is continuous).
  • num_layers (int, optional, defaults to 1) — The number of layers of Transformer blocks to use.
  • dropout (float, optional, defaults to 0.0) — The dropout probability to use.
  • cross_attention_dim (int, optional) — The number of encoder_hidden_states dimensions to use.
  • sample_size (int, optional) — The width of the latent images (specify if the input is discrete). This is fixed during training since it is used to learn a number of position embeddings.
  • num_vector_embeds (int, optional) — The number of classes of the vector embeddings of the latent pixels (specify if the input is discrete). Includes the class for the masked latent pixel.
  • activation_fn (str, optional, defaults to "geglu") — Activation function to use in feed-forward.
  • num_embeds_ada_norm ( int, optional) — The number of diffusion steps used during training. Pass if at least one of the norm_layers is AdaLayerNorm. This is fixed during training since it is used to learn a number of embeddings that are added to the hidden states.

    During inference, you can denoise for up to but not more steps than num_embeds_ada_norm.

  • attention_bias (bool, optional) — Configure if the TransformerBlocks attention should contain a bias parameter.

A 2D Transformer model for image-like data.


< >

( hidden_states: Tensor encoder_hidden_states: Optional = None timestep: Optional = None added_cond_kwargs: Dict = None class_labels: Optional = None cross_attention_kwargs: Dict = None attention_mask: Optional = None encoder_attention_mask: Optional = None return_dict: bool = True )


  • hidden_states (torch.LongTensor of shape (batch size, num latent pixels) if discrete, torch.Tensor of shape (batch size, channel, height, width) if continuous) — Input hidden_states.
  • encoder_hidden_states ( torch.Tensor of shape (batch size, sequence len, embed dims), optional) — Conditional embeddings for cross attention layer. If not given, cross-attention defaults to self-attention.
  • timestep ( torch.LongTensor, optional) — Used to indicate denoising step. Optional timestep to be applied as an embedding in AdaLayerNorm.
  • class_labels ( torch.LongTensor of shape (batch size, num classes), optional) — Used to indicate class labels conditioning. Optional class labels to be applied as an embedding in AdaLayerZeroNorm.
  • cross_attention_kwargs ( Dict[str, Any], optional) — A kwargs dictionary that if specified is passed along to the AttentionProcessor as defined under self.processor in diffusers.models.attention_processor.
  • attention_mask ( torch.Tensor, optional) — An attention mask of shape (batch, key_tokens) is applied to encoder_hidden_states. If 1 the mask is kept, otherwise if 0 it is discarded. Mask will be converted into a bias, which adds large negative values to the attention scores corresponding to “discard” tokens.
  • encoder_attention_mask ( torch.Tensor, optional) — Cross-attention mask applied to encoder_hidden_states. Two formats supported:

    • Mask (batch, sequence_length) True = keep, False = discard.
    • Bias (batch, 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.

  • return_dict (bool, optional, defaults to True) — Whether or not to return a UNet2DConditionOutput instead of a plain tuple.

The Transformer2DModel forward method.


class diffusers.models.modeling_outputs.Transformer2DModelOutput

< >

( sample: 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.

The output of Transformer2DModel.

< > Update on GitHub