MochiTransformer3DModel

A Diffusion Transformer model for 3D video-like data was introduced in Mochi-1 Preview by Genmo.

The model can be loaded with the following code snippet.

from diffusers import MochiTransformer3DModel

transformer = MochiTransformer3DModel.from_pretrained("genmo/mochi-1-preview", subfolder="transformer", torch_dtype=torch.float16).to("cuda")

MochiTransformer3DModel

class diffusers.MochiTransformer3DModel

< source >

( patch_size: int = 2 num_attention_heads: int = 24 attention_head_dim: int = 128 num_layers: int = 48 pooled_projection_dim: int = 1536 in_channels: int = 12 out_channels: typing.Optional[int] = None qk_norm: str = 'rms_norm' text_embed_dim: int = 4096 time_embed_dim: int = 256 activation_fn: str = 'swiglu' max_sequence_length: int = 256 )

Parameters

patch_size (int, defaults to 2) — The size of the patches to use in the patch embedding layer.
num_attention_heads (int, defaults to 24) — The number of heads to use for multi-head attention.
attention_head_dim (int, defaults to 128) — The number of channels in each head.
num_layers (int, defaults to 48) — The number of layers of Transformer blocks to use.
in_channels (int, defaults to 12) — The number of channels in the input.
out_channels (int, optional, defaults to None) — The number of channels in the output.
qk_norm (str, defaults to "rms_norm") — The normalization layer to use.
text_embed_dim (int, defaults to 4096) — Input dimension of text embeddings from the text encoder.
time_embed_dim (int, defaults to 256) — Output dimension of timestep embeddings.
activation_fn (str, defaults to "swiglu") — Activation function to use in feed-forward.
max_sequence_length (int, defaults to 256) — The maximum sequence length of text embeddings supported.

A Transformer model for video-like data introduced in Mochi.

Transformer2DModelOutput

class diffusers.models.modeling_outputs.Transformer2DModelOutput

< source >

( sample: torch.Tensor )

Parameters

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

Diffusers

MochiTransformer3DModel