LatteTransformer3DModel

A Diffusion Transformer model for 3D data from Latte.

mindone.diffusers.LatteTransformer3DModel

Bases: ModelMixin, ConfigMixin

Source code in mindone/diffusers/models/transformers/latte_transformer_3d.py

class LatteTransformer3DModel(ModelMixin, ConfigMixin):
    _supports_gradient_checkpointing = True

    """
    A 3D Transformer model for video-like data, paper: https://arxiv.org/abs/2401.03048, offical code:
    https://github.com/Vchitect/Latte

    Parameters:
        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.
        out_channels (`int`, *optional*):
            The number of channels in the output.
        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.
        attention_bias (`bool`, *optional*):
            Configure if the `TransformerBlocks` attention should contain a bias parameter.
        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.
        patch_size (`int`, *optional*):
            The size of the patches to use in the patch embedding layer.
        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`.
        norm_type (`str`, *optional*, defaults to `"layer_norm"`):
            The type of normalization to use. Options are `"layer_norm"` or `"ada_layer_norm"`.
        norm_elementwise_affine (`bool`, *optional*, defaults to `True`):
            Whether or not to use elementwise affine in normalization layers.
        norm_eps (`float`, *optional*, defaults to 1e-5): The epsilon value to use in normalization layers.
        caption_channels (`int`, *optional*):
            The number of channels in the caption embeddings.
        video_length (`int`, *optional*):
            The number of frames in the video-like data.
    """

    @register_to_config
    def __init__(
        self,
        num_attention_heads: int = 16,
        attention_head_dim: int = 88,
        in_channels: Optional[int] = None,
        out_channels: Optional[int] = None,
        num_layers: int = 1,
        dropout: float = 0.0,
        cross_attention_dim: Optional[int] = None,
        attention_bias: bool = False,
        sample_size: int = 64,
        patch_size: Optional[int] = None,
        activation_fn: str = "geglu",
        num_embeds_ada_norm: Optional[int] = None,
        norm_type: str = "layer_norm",
        norm_elementwise_affine: bool = True,
        norm_eps: float = 1e-5,
        caption_channels: int = None,
        video_length: int = 16,
    ):
        super().__init__()
        inner_dim = num_attention_heads * attention_head_dim

        # 1. Define input layers
        self.height = sample_size
        self.width = sample_size

        interpolation_scale = self.config.sample_size // 64
        interpolation_scale = max(interpolation_scale, 1)
        self.pos_embed = PatchEmbed(
            height=sample_size,
            width=sample_size,
            patch_size=patch_size,
            in_channels=in_channels,
            embed_dim=inner_dim,
            interpolation_scale=interpolation_scale,
        )
        self.patch_size = self.config.patch_size

        # 2. Define spatial transformers blocks
        self.transformer_blocks = nn.CellList(
            [
                BasicTransformerBlock(
                    inner_dim,
                    num_attention_heads,
                    attention_head_dim,
                    dropout=dropout,
                    cross_attention_dim=cross_attention_dim,
                    activation_fn=activation_fn,
                    num_embeds_ada_norm=num_embeds_ada_norm,
                    attention_bias=attention_bias,
                    norm_type=norm_type,
                    norm_elementwise_affine=norm_elementwise_affine,
                    norm_eps=norm_eps,
                )
                for d in range(num_layers)
            ]
        )

        # 3. Define temporal transformers blocks
        self.temporal_transformer_blocks = nn.CellList(
            [
                BasicTransformerBlock(
                    inner_dim,
                    num_attention_heads,
                    attention_head_dim,
                    dropout=dropout,
                    cross_attention_dim=None,
                    activation_fn=activation_fn,
                    num_embeds_ada_norm=num_embeds_ada_norm,
                    attention_bias=attention_bias,
                    norm_type=norm_type,
                    norm_elementwise_affine=norm_elementwise_affine,
                    norm_eps=norm_eps,
                )
                for d in range(num_layers)
            ]
        )

        # 4. Define output layers
        self.out_channels = in_channels if out_channels is None else out_channels
        self.norm_out = LayerNorm(inner_dim, elementwise_affine=False, eps=1e-6)
        self.scale_shift_table = ms.Parameter(ops.randn((2, inner_dim)) / inner_dim**0.5)
        self.proj_out = nn.Dense(inner_dim, patch_size * patch_size * self.out_channels)

        # 5. Latte other blocks.
        self.adaln_single = AdaLayerNormSingle(inner_dim, use_additional_conditions=False)
        self.caption_projection = PixArtAlphaTextProjection(in_features=caption_channels, hidden_size=inner_dim)

        # define temporal positional embedding
        temp_pos_embed = get_1d_sincos_pos_embed_from_grid(
            inner_dim, ops.arange(0, video_length).unsqueeze(1).numpy()
        )  # 1152 hidden size
        self.temp_pos_embed = ms.Tensor.from_numpy(temp_pos_embed).float().unsqueeze(0)

        self.gradient_checkpointing = False

    def _set_gradient_checkpointing(self, module, value=False):
        self.gradient_checkpointing = value

    def construct(
        self,
        hidden_states: ms.Tensor,
        timestep: Optional[ms.Tensor] = None,
        encoder_hidden_states: Optional[ms.Tensor] = None,
        encoder_attention_mask: Optional[ms.Tensor] = None,
        enable_temporal_attentions: bool = True,
        return_dict: bool = False,
    ):
        """
        The [`LatteTransformer3DModel`] forward method.

        Args:
            hidden_states shape `(batch size, channel, num_frame, height, width)`:
                Input `hidden_states`.
            timestep ( `ms.Tensor`, *optional*):
                Used to indicate denoising step. Optional timestep to be applied as an embedding in `AdaLayerNorm`.
            encoder_hidden_states ( `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.
            encoder_attention_mask ( `ms.Tensor`, *optional*):
                Cross-attention mask applied to `encoder_hidden_states`. Two formats supported:

                    * Mask `(batcheight, sequence_length)` True = keep, False = discard.
                    * Bias `(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.
            enable_temporal_attentions:
                (`bool`, *optional*, defaults to `True`): Whether to enable temporal attentions.
            return_dict (`bool`, *optional*, defaults to `False`):
                Whether or not to return a [`~models.unet_2d_condition.UNet2DConditionOutput`] instead of a plain
                tuple.

        Returns:
            If `return_dict` is True, an [`~models.transformer_2d.Transformer2DModelOutput`] is returned, otherwise a
            `tuple` where the first element is the sample tensor.
        """

        # Reshape hidden states
        batch_size, channels, num_frame, height, width = hidden_states.shape
        # batch_size channels num_frame height width -> (batch_size * num_frame) channels height width
        hidden_states = hidden_states.permute(0, 2, 1, 3, 4).reshape(-1, channels, height, width)

        # Input
        height, width = (
            hidden_states.shape[-2] // self.patch_size,
            hidden_states.shape[-1] // self.patch_size,
        )
        num_patches = height * width

        hidden_states = self.pos_embed(hidden_states)  # alrady add positional embeddings

        added_cond_kwargs = {"resolution": None, "aspect_ratio": None}
        timestep, embedded_timestep = self.adaln_single(
            timestep, added_cond_kwargs=added_cond_kwargs, batch_size=batch_size, hidden_dtype=hidden_states.dtype
        )

        # Prepare text embeddings for spatial block
        # batch_size num_tokens hidden_size -> (batch_size * num_frame) num_tokens hidden_size
        encoder_hidden_states = self.caption_projection(encoder_hidden_states)  # 3 120 1152
        encoder_hidden_states_spatial = encoder_hidden_states.repeat_interleave(num_frame, dim=0).view(
            -1, encoder_hidden_states.shape[-2], encoder_hidden_states.shape[-1]
        )

        # Prepare timesteps for spatial and temporal block
        timestep_spatial = timestep.repeat_interleave(num_frame, dim=0).view(-1, timestep.shape[-1])
        timestep_temp = timestep.repeat_interleave(num_patches, dim=0).view(-1, timestep.shape[-1])

        # Spatial and temporal transformer blocks
        for i, (spatial_block, temp_block) in enumerate(zip(self.transformer_blocks, self.temporal_transformer_blocks)):
            hidden_states = spatial_block(
                hidden_states,
                None,  # attention_mask
                encoder_hidden_states_spatial,
                encoder_attention_mask,
                timestep_spatial,
                None,  # cross_attention_kwargs
                None,  # class_labels
            )

            if enable_temporal_attentions:
                # (batch_size * num_frame) num_tokens hidden_size -> (batch_size * num_tokens) num_frame hidden_size
                hidden_states = hidden_states.reshape(
                    batch_size, -1, hidden_states.shape[-2], hidden_states.shape[-1]
                ).permute(0, 2, 1, 3)
                hidden_states = hidden_states.reshape(-1, hidden_states.shape[-2], hidden_states.shape[-1])

                if i == 0 and num_frame > 1:
                    hidden_states = (hidden_states + self.temp_pos_embed).to(hidden_states.dtype)

                hidden_states = temp_block(
                    hidden_states,
                    None,  # attention_mask
                    None,  # encoder_hidden_states
                    None,  # encoder_attention_mask
                    timestep_temp,
                    None,  # cross_attention_kwargs
                    None,  # class_labels
                )

                # (batch_size * num_tokens) num_frame hidden_size -> (batch_size * num_frame) num_tokens hidden_size
                hidden_states = hidden_states.reshape(
                    batch_size, -1, hidden_states.shape[-2], hidden_states.shape[-1]
                ).permute(0, 2, 1, 3)
                hidden_states = hidden_states.reshape(-1, hidden_states.shape[-2], hidden_states.shape[-1])

        embedded_timestep = embedded_timestep.repeat_interleave(num_frame, dim=0).view(-1, embedded_timestep.shape[-1])
        shift, scale = (self.scale_shift_table[None] + embedded_timestep[:, None]).chunk(2, axis=1)
        hidden_states = self.norm_out(hidden_states)
        # Modulation
        hidden_states = hidden_states * (1 + scale) + shift
        hidden_states = self.proj_out(hidden_states)

        # unpatchify
        if self.adaln_single is None:
            height = width = int(hidden_states.shape[1] ** 0.5)
        hidden_states = hidden_states.reshape((-1, height, width, self.patch_size, self.patch_size, self.out_channels))
        hidden_states = hidden_states.transpose(0, 5, 1, 3, 2, 4)
        output = hidden_states.reshape((-1, self.out_channels, height * self.patch_size, width * self.patch_size))
        output = output.reshape(batch_size, -1, output.shape[-3], output.shape[-2], output.shape[-1]).permute(
            0, 2, 1, 3, 4
        )

        if not return_dict:
            return (output,)

        return Transformer2DModelOutput(sample=output)

mindone.diffusers.LatteTransformer3DModel.construct(hidden_states, timestep=None, encoder_hidden_states=None, encoder_attention_mask=None, enable_temporal_attentions=True, return_dict=False)

The [LatteTransformer3DModel] forward method.

PARAMETER	DESCRIPTION
hidden_states	Input hidden_states. TYPE: shape `(batch size, channel, num_frame, height, width)`
timestep	Used to indicate denoising step. Optional timestep to be applied as an embedding in AdaLayerNorm. TYPE: `ms.Tensor`, *optional* DEFAULT: None
encoder_hidden_states	Conditional embeddings for cross attention layer. If not given, cross-attention defaults to self-attention. TYPE: `torch.FloatTensor` of shape `(batch size, sequence len, embed dims)`, *optional* DEFAULT: None
encoder_attention_mask	Cross-attention mask applied to encoder_hidden_states. Two formats supported: * Mask `(batcheight, sequence_length)` True = keep, False = discard. * Bias `(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. TYPE: `ms.Tensor`, *optional* DEFAULT: None
enable_temporal_attentions	(bool, optional, defaults to True): Whether to enable temporal attentions. TYPE: bool DEFAULT: True
return_dict	Whether or not to return a [~models.unet_2d_condition.UNet2DConditionOutput] instead of a plain tuple. TYPE: `bool`, *optional*, defaults to `False` DEFAULT: False

RETURNS	DESCRIPTION
	If return_dict is True, an [~models.transformer_2d.Transformer2DModelOutput] is returned, otherwise a
	tuple where the first element is the sample tensor.

Source code in mindone/diffusers/models/transformers/latte_transformer_3d.py

def construct(
    self,
    hidden_states: ms.Tensor,
    timestep: Optional[ms.Tensor] = None,
    encoder_hidden_states: Optional[ms.Tensor] = None,
    encoder_attention_mask: Optional[ms.Tensor] = None,
    enable_temporal_attentions: bool = True,
    return_dict: bool = False,
):
    """
    The [`LatteTransformer3DModel`] forward method.

    Args:
        hidden_states shape `(batch size, channel, num_frame, height, width)`:
            Input `hidden_states`.
        timestep ( `ms.Tensor`, *optional*):
            Used to indicate denoising step. Optional timestep to be applied as an embedding in `AdaLayerNorm`.
        encoder_hidden_states ( `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.
        encoder_attention_mask ( `ms.Tensor`, *optional*):
            Cross-attention mask applied to `encoder_hidden_states`. Two formats supported:

                * Mask `(batcheight, sequence_length)` True = keep, False = discard.
                * Bias `(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.
        enable_temporal_attentions:
            (`bool`, *optional*, defaults to `True`): Whether to enable temporal attentions.
        return_dict (`bool`, *optional*, defaults to `False`):
            Whether or not to return a [`~models.unet_2d_condition.UNet2DConditionOutput`] instead of a plain
            tuple.

    Returns:
        If `return_dict` is True, an [`~models.transformer_2d.Transformer2DModelOutput`] is returned, otherwise a
        `tuple` where the first element is the sample tensor.
    """

    # Reshape hidden states
    batch_size, channels, num_frame, height, width = hidden_states.shape
    # batch_size channels num_frame height width -> (batch_size * num_frame) channels height width
    hidden_states = hidden_states.permute(0, 2, 1, 3, 4).reshape(-1, channels, height, width)

    # Input
    height, width = (
        hidden_states.shape[-2] // self.patch_size,
        hidden_states.shape[-1] // self.patch_size,
    )
    num_patches = height * width

    hidden_states = self.pos_embed(hidden_states)  # alrady add positional embeddings

    added_cond_kwargs = {"resolution": None, "aspect_ratio": None}
    timestep, embedded_timestep = self.adaln_single(
        timestep, added_cond_kwargs=added_cond_kwargs, batch_size=batch_size, hidden_dtype=hidden_states.dtype
    )

    # Prepare text embeddings for spatial block
    # batch_size num_tokens hidden_size -> (batch_size * num_frame) num_tokens hidden_size
    encoder_hidden_states = self.caption_projection(encoder_hidden_states)  # 3 120 1152
    encoder_hidden_states_spatial = encoder_hidden_states.repeat_interleave(num_frame, dim=0).view(
        -1, encoder_hidden_states.shape[-2], encoder_hidden_states.shape[-1]
    )

    # Prepare timesteps for spatial and temporal block
    timestep_spatial = timestep.repeat_interleave(num_frame, dim=0).view(-1, timestep.shape[-1])
    timestep_temp = timestep.repeat_interleave(num_patches, dim=0).view(-1, timestep.shape[-1])

    # Spatial and temporal transformer blocks
    for i, (spatial_block, temp_block) in enumerate(zip(self.transformer_blocks, self.temporal_transformer_blocks)):
        hidden_states = spatial_block(
            hidden_states,
            None,  # attention_mask
            encoder_hidden_states_spatial,
            encoder_attention_mask,
            timestep_spatial,
            None,  # cross_attention_kwargs
            None,  # class_labels
        )

        if enable_temporal_attentions:
            # (batch_size * num_frame) num_tokens hidden_size -> (batch_size * num_tokens) num_frame hidden_size
            hidden_states = hidden_states.reshape(
                batch_size, -1, hidden_states.shape[-2], hidden_states.shape[-1]
            ).permute(0, 2, 1, 3)
            hidden_states = hidden_states.reshape(-1, hidden_states.shape[-2], hidden_states.shape[-1])

            if i == 0 and num_frame > 1:
                hidden_states = (hidden_states + self.temp_pos_embed).to(hidden_states.dtype)

            hidden_states = temp_block(
                hidden_states,
                None,  # attention_mask
                None,  # encoder_hidden_states
                None,  # encoder_attention_mask
                timestep_temp,
                None,  # cross_attention_kwargs
                None,  # class_labels
            )

            # (batch_size * num_tokens) num_frame hidden_size -> (batch_size * num_frame) num_tokens hidden_size
            hidden_states = hidden_states.reshape(
                batch_size, -1, hidden_states.shape[-2], hidden_states.shape[-1]
            ).permute(0, 2, 1, 3)
            hidden_states = hidden_states.reshape(-1, hidden_states.shape[-2], hidden_states.shape[-1])

    embedded_timestep = embedded_timestep.repeat_interleave(num_frame, dim=0).view(-1, embedded_timestep.shape[-1])
    shift, scale = (self.scale_shift_table[None] + embedded_timestep[:, None]).chunk(2, axis=1)
    hidden_states = self.norm_out(hidden_states)
    # Modulation
    hidden_states = hidden_states * (1 + scale) + shift
    hidden_states = self.proj_out(hidden_states)

    # unpatchify
    if self.adaln_single is None:
        height = width = int(hidden_states.shape[1] ** 0.5)
    hidden_states = hidden_states.reshape((-1, height, width, self.patch_size, self.patch_size, self.out_channels))
    hidden_states = hidden_states.transpose(0, 5, 1, 3, 2, 4)
    output = hidden_states.reshape((-1, self.out_channels, height * self.patch_size, width * self.patch_size))
    output = output.reshape(batch_size, -1, output.shape[-3], output.shape[-2], output.shape[-1]).permute(
        0, 2, 1, 3, 4
    )

    if not return_dict:
        return (output,)

    return Transformer2DModelOutput(sample=output)