Latte: Latent Diffusion Transformer for Video Generation (#8404)

* add Latte to diffusers

* remove print

* remove print

* remove print

* remove unuse codes

* remove layer_norm_latte and add a flag

* remove layer_norm_latte and add a flag

* update latte_pipeline

* update latte_pipeline

* remove unuse squeeze

* add norm_hidden_states.ndim == 2: # for Latte

* fixed test latte pipeline bugs

* fixed test latte pipeline bugs

* delete sh

* add doc for latte

* add licensing

* Move Transformer3DModelOutput to modeling_outputs

* give a default value to sample_size

* remove the einops dependency

* change norm2 for latte

* modify pipeline of latte

* update test for Latte

* modify some codes for latte

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* modify for Latte pipeline

* video_length -> num_frames; update prepare_latents copied from

* make fix-copies

* make style

* typo: videe -> video

* update

* modify for Latte pipeline

* modify latte pipeline

* modify latte pipeline

* modify latte pipeline

* modify latte pipeline

* modify for Latte pipeline

* Delete .vscode directory

* make style

* make fix-copies

* add latte transformer 3d to docs _toctree.yml

* update example

* reduce frames for test

* fixed bug of _text_preprocessing

* set num frame to 1 for testing

* remove unuse print

* add text = self._clean_caption(text) again

---------
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: YiYi Xu <yixu310@gmail.com>
Co-authored-by: Aryan <contact.aryanvs@gmail.com>
Co-authored-by: Aryan <aryan@huggingface.co>

.gitignore

@@ -175,4 +175,4 @@ tags
 .ruff_cache
 
 # wandb
-wandb
+wandb
\ No newline at end of file

docs/source/en/_toctree.yml

@@ -249,6 +249,8 @@
       title: DiTTransformer2DModel
     - local: api/models/hunyuan_transformer2d
       title: HunyuanDiT2DModel
+    - local: api/models/latte_transformer3d
+      title: LatteTransformer3DModel
     - local: api/models/lumina_nextdit2d
       title: LuminaNextDiT2DModel
     - local: api/models/transformer_temporal

docs/source/en/api/models/latte_transformer3d.md

+<!--Copyright 2024 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+## LatteTransformer3DModel
+
+A Diffusion Transformer model for 3D data from [Latte](https://github.com/Vchitect/Latte).
+
+## LatteTransformer3DModel
+
+[[autodoc]] LatteTransformer3DModel

src/diffusers/__init__.py

@@ -88,6 +88,7 @@ else:
             "HunyuanDiT2DMultiControlNetModel",
             "I2VGenXLUNet",
             "Kandinsky3UNet",
+            "LatteTransformer3DModel",
             "LuminaNextDiT2DModel",
             "ModelMixin",
             "MotionAdapter",
@@ -269,6 +270,7 @@ else:
             "KandinskyV22PriorPipeline",
             "LatentConsistencyModelImg2ImgPipeline",
             "LatentConsistencyModelPipeline",
+            "LattePipeline",
             "LDMTextToImagePipeline",
             "LEditsPPPipelineStableDiffusion",
             "LEditsPPPipelineStableDiffusionXL",
@@ -513,6 +515,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             HunyuanDiT2DMultiControlNetModel,
             I2VGenXLUNet,
             Kandinsky3UNet,
+            LatteTransformer3DModel,
             LuminaNextDiT2DModel,
             ModelMixin,
             MotionAdapter,
@@ -672,6 +675,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             KandinskyV22PriorPipeline,
             LatentConsistencyModelImg2ImgPipeline,
             LatentConsistencyModelPipeline,
+            LattePipeline,
             LDMTextToImagePipeline,
             LEditsPPPipelineStableDiffusion,
             LEditsPPPipelineStableDiffusionXL,

src/diffusers/models/__init__.py

@@ -41,6 +41,7 @@ if is_torch_available():
     _import_structure["transformers.dit_transformer_2d"] = ["DiTTransformer2DModel"]
     _import_structure["transformers.dual_transformer_2d"] = ["DualTransformer2DModel"]
     _import_structure["transformers.hunyuan_transformer_2d"] = ["HunyuanDiT2DModel"]
+    _import_structure["transformers.latte_transformer_3d"] = ["LatteTransformer3DModel"]
     _import_structure["transformers.lumina_nextdit2d"] = ["LuminaNextDiT2DModel"]
     _import_structure["transformers.pixart_transformer_2d"] = ["PixArtTransformer2DModel"]
     _import_structure["transformers.prior_transformer"] = ["PriorTransformer"]
@@ -86,6 +87,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             DiTTransformer2DModel,
             DualTransformer2DModel,
             HunyuanDiT2DModel,
+            LatteTransformer3DModel,
             LuminaNextDiT2DModel,
             PixArtTransformer2DModel,
             PriorTransformer,

src/diffusers/models/attention.py

@@ -359,7 +359,10 @@ class BasicTransformerBlock(nn.Module):
                 out_bias=attention_out_bias,
             )  # is self-attn if encoder_hidden_states is none
         else:
-            self.norm2 = None
+            if norm_type == "ada_norm_single":  # For Latte
+                self.norm2 = nn.LayerNorm(dim, norm_eps, norm_elementwise_affine)
+            else:
+                self.norm2 = None
             self.attn2 = None
 
         # 3. Feed-forward
@@ -439,7 +442,6 @@ class BasicTransformerBlock(nn.Module):
             ).chunk(6, dim=1)
             norm_hidden_states = self.norm1(hidden_states)
             norm_hidden_states = norm_hidden_states * (1 + scale_msa) + shift_msa
-            norm_hidden_states = norm_hidden_states.squeeze(1)
         else:
             raise ValueError("Incorrect norm used")
 
@@ -456,6 +458,7 @@ class BasicTransformerBlock(nn.Module):
             attention_mask=attention_mask,
             **cross_attention_kwargs,
         )
+
         if self.norm_type == "ada_norm_zero":
             attn_output = gate_msa.unsqueeze(1) * attn_output
         elif self.norm_type == "ada_norm_single":

src/diffusers/models/transformers/__init__.py

@@ -5,6 +5,7 @@ if is_torch_available():
     from .dit_transformer_2d import DiTTransformer2DModel
     from .dual_transformer_2d import DualTransformer2DModel
     from .hunyuan_transformer_2d import HunyuanDiT2DModel
+    from .latte_transformer_3d import LatteTransformer3DModel
     from .lumina_nextdit2d import LuminaNextDiT2DModel
     from .pixart_transformer_2d import PixArtTransformer2DModel
     from .prior_transformer import PriorTransformer

src/diffusers/models/transformers/latte_transformer_3d.py

+# Copyright 2024 the Latte Team and The HuggingFace Team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from typing import Optional
+
+import torch
+from torch import nn
+
+from ...configuration_utils import ConfigMixin, register_to_config
+from ...models.embeddings import PixArtAlphaTextProjection, get_1d_sincos_pos_embed_from_grid
+from ..attention import BasicTransformerBlock
+from ..embeddings import PatchEmbed
+from ..modeling_outputs import Transformer2DModelOutput
+from ..modeling_utils import ModelMixin
+from ..normalization import AdaLayerNormSingle
+
+
+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,
+        )
+
+        # 2. Define spatial transformers blocks
+        self.transformer_blocks = nn.ModuleList(
+            [
+                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.ModuleList(
+            [
+                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 = nn.LayerNorm(inner_dim, elementwise_affine=False, eps=1e-6)
+        self.scale_shift_table = nn.Parameter(torch.randn(2, inner_dim) / inner_dim**0.5)
+        self.proj_out = nn.Linear(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, torch.arange(0, video_length).unsqueeze(1)
+        )  # 1152 hidden size
+        self.register_buffer("temp_pos_embed", torch.from_numpy(temp_pos_embed).float().unsqueeze(0), persistent=False)
+
+        self.gradient_checkpointing = False
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        self.gradient_checkpointing = value
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        timestep: Optional[torch.LongTensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: Optional[torch.Tensor] = None,
+        enable_temporal_attentions: bool = True,
+        return_dict: bool = True,
+    ):
+        """
+        The [`LatteTransformer3DModel`] forward method.
+
+        Args:
+            hidden_states shape `(batch size, channel, num_frame, height, width)`:
+                Input `hidden_states`.
+            timestep ( `torch.LongTensor`, *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 ( `torch.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 `True`):
+                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.config.patch_size,
+            hidden_states.shape[-1] // self.config.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)
+        ):
+            if self.training and self.gradient_checkpointing:
+                hidden_states = torch.utils.checkpoint.checkpoint(
+                    spatial_block,
+                    hidden_states,
+                    None,  # attention_mask
+                    encoder_hidden_states_spatial,
+                    encoder_attention_mask,
+                    timestep_spatial,
+                    None,  # cross_attention_kwargs
+                    None,  # class_labels
+                    use_reentrant=False,
+                )
+            else:
+                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
+
+                if self.training and self.gradient_checkpointing:
+                    hidden_states = torch.utils.checkpoint.checkpoint(
+                        temp_block,
+                        hidden_states,
+                        None,  # attention_mask
+                        None,  # encoder_hidden_states
+                        None,  # encoder_attention_mask
+                        timestep_temp,
+                        None,  # cross_attention_kwargs
+                        None,  # class_labels
+                        use_reentrant=False,
+                    )
+                else:
+                    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, dim=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(
+            shape=(-1, height, width, self.config.patch_size, self.config.patch_size, self.out_channels)
+        )
+        hidden_states = torch.einsum("nhwpqc->nchpwq", hidden_states)
+        output = hidden_states.reshape(
+            shape=(-1, self.out_channels, height * self.config.patch_size, width * self.config.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)

src/diffusers/pipelines/__init__.py

@@ -209,6 +209,7 @@ else:
             "LEditsPPPipelineStableDiffusionXL",
         ]
     )
+    _import_structure["latte"] = ["LattePipeline"]
     _import_structure["lumina"] = ["LuminaText2ImgPipeline"]
     _import_structure["marigold"].extend(
         [
@@ -485,6 +486,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             LatentConsistencyModelPipeline,
         )
         from .latent_diffusion import LDMTextToImagePipeline
+        from .latte import LattePipeline
         from .ledits_pp import (
             LEditsPPDiffusionPipelineOutput,
             LEditsPPInversionPipelineOutput,

src/diffusers/pipelines/latte/__init__.py

+from typing import TYPE_CHECKING
+
+from ...utils import (
+    DIFFUSERS_SLOW_IMPORT,
+    OptionalDependencyNotAvailable,
+    _LazyModule,
+    get_objects_from_module,
+    is_torch_available,
+    is_transformers_available,
+)
+
+
+_dummy_objects = {}
+_import_structure = {}
+
+
+try:
+    if not (is_transformers_available() and is_torch_available()):
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    from ...utils import dummy_torch_and_transformers_objects  # noqa F403
+
+    _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
+else:
+    _import_structure["pipeline_latte"] = ["LattePipeline"]
+
+if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
+    try:
+        if not (is_transformers_available() and is_torch_available()):
+            raise OptionalDependencyNotAvailable()
+
+    except OptionalDependencyNotAvailable:
+        from ...utils.dummy_torch_and_transformers_objects import *
+    else:
+        from .pipeline_latte import LattePipeline
+
+else:
+    import sys
+
+    sys.modules[__name__] = _LazyModule(
+        __name__,
+        globals()["__file__"],
+        _import_structure,
+        module_spec=__spec__,
+    )
+
+    for name, value in _dummy_objects.items():
+        setattr(sys.modules[__name__], name, value)

src/diffusers/pipelines/latte/pipeline_latte.py

+# Copyright 2024 the Latte Team and The HuggingFace Team.
+# All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import html
+import inspect
+import re
+import urllib.parse as ul
+from dataclasses import dataclass
+from typing import Callable, Dict, List, Optional, Tuple, Union
+
+import torch
+from transformers import T5EncoderModel, T5Tokenizer
+
+from ...callbacks import MultiPipelineCallbacks, PipelineCallback
+from ...models import AutoencoderKL, LatteTransformer3DModel
+from ...pipelines.pipeline_utils import DiffusionPipeline
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import (
+    BACKENDS_MAPPING,
+    BaseOutput,
+    is_bs4_available,
+    is_ftfy_available,
+    logging,
+    replace_example_docstring,
+)
+from ...utils.torch_utils import is_compiled_module, randn_tensor
+from ...video_processor import VideoProcessor
+
+
+logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+
+if is_bs4_available():
+    from bs4 import BeautifulSoup
+
+if is_ftfy_available():
+    import ftfy
+
+
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```py
+        >>> import torch
+        >>> from diffusers import LattePipeline
+        >>> from diffusers.utils import export_to_gif
+
+        >>> # You can replace the checkpoint id with "maxin-cn/Latte-1" too.
+        >>> pipe = LattePipeline.from_pretrained("maxin-cn/Latte-1", torch_dtype=torch.float16).to("cuda")
+        >>> # Enable memory optimizations.
+        >>> pipe.enable_model_cpu_offload()
+
+        >>> prompt = "A small cactus with a happy face in the Sahara desert."
+        >>> videos = pipe(prompt).frames[0]
+        >>> export_to_gif(videos, "latte.gif")
+        ```
+"""
+
+
+# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps
+def retrieve_timesteps(
+    scheduler,
+    num_inference_steps: Optional[int] = None,
+    device: Optional[Union[str, torch.device]] = None,
+    timesteps: Optional[List[int]] = None,
+    sigmas: Optional[List[float]] = None,
+    **kwargs,
+):
+    """
+    Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
+    custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
+
+    Args:
+        scheduler (`SchedulerMixin`):
+            The scheduler to get timesteps from.
+        num_inference_steps (`int`):
+            The number of diffusion steps used when generating samples with a pre-trained model. If used, `timesteps`
+            must be `None`.
+        device (`str` or `torch.device`, *optional*):
+            The device to which the timesteps should be moved to. If `None`, the timesteps are not moved.
+        timesteps (`List[int]`, *optional*):
+            Custom timesteps used to override the timestep spacing strategy of the scheduler. If `timesteps` is passed,
+            `num_inference_steps` and `sigmas` must be `None`.
+        sigmas (`List[float]`, *optional*):
+            Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed,
+            `num_inference_steps` and `timesteps` must be `None`.
+
+    Returns:
+        `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the
+        second element is the number of inference steps.
+    """
+    if timesteps is not None and sigmas is not None:
+        raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values")
+    if timesteps is not None:
+        accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
+        if not accepts_timesteps:
+            raise ValueError(
+                f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
+                f" timestep schedules. Please check whether you are using the correct scheduler."
+            )
+        scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs)
+        timesteps = scheduler.timesteps
+        num_inference_steps = len(timesteps)
+    elif sigmas is not None:
+        accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
+        if not accept_sigmas:
+            raise ValueError(
+                f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
+                f" sigmas schedules. Please check whether you are using the correct scheduler."
+            )
+        scheduler.set_timesteps(sigmas=sigmas, device=device, **kwargs)
+        timesteps = scheduler.timesteps
+        num_inference_steps = len(timesteps)
+    else:
+        scheduler.set_timesteps(num_inference_steps, device=device, **kwargs)
+        timesteps = scheduler.timesteps
+    return timesteps, num_inference_steps
+
+
+@dataclass
+class LattePipelineOutput(BaseOutput):
+    frames: torch.Tensor
+
+
+class LattePipeline(DiffusionPipeline):
+    r"""
+    Pipeline for text-to-video generation using Latte.
+
+    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods the
+    library implements for all the pipelines (such as downloading or saving, running on a particular device, etc.)
+
+    Args:
+        vae ([`AutoencoderKL`]):
+            Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
+        text_encoder ([`T5EncoderModel`]):
+            Frozen text-encoder. Latte uses
+            [T5](https://huggingface.co/docs/transformers/model_doc/t5#transformers.T5EncoderModel), specifically the
+            [t5-v1_1-xxl](https://huggingface.co/PixArt-alpha/PixArt-alpha/tree/main/t5-v1_1-xxl) variant.
+        tokenizer (`T5Tokenizer`):
+            Tokenizer of class
+            [T5Tokenizer](https://huggingface.co/docs/transformers/model_doc/t5#transformers.T5Tokenizer).
+        transformer ([`LatteTransformer3DModel`]):
+            A text conditioned `LatteTransformer3DModel` to denoise the encoded video latents.
+        scheduler ([`SchedulerMixin`]):
+            A scheduler to be used in combination with `transformer` to denoise the encoded video latents.
+    """
+
+    bad_punct_regex = re.compile(r"[#®•©™&@·º½¾¿¡§~\)\(\]\[\}\{\|\\/\\*]{1,}")
+
+    _optional_components = ["tokenizer", "text_encoder"]
+    model_cpu_offload_seq = "text_encoder->transformer->vae"
+
+    _callback_tensor_inputs = [
+        "latents",
+        "prompt_embeds",
+        "negative_prompt_embeds",
+    ]
+
+    def __init__(
+        self,
+        tokenizer: T5Tokenizer,
+        text_encoder: T5EncoderModel,
+        vae: AutoencoderKL,
+        transformer: LatteTransformer3DModel,
+        scheduler: KarrasDiffusionSchedulers,
+    ):
+        super().__init__()
+
+        self.register_modules(
+            tokenizer=tokenizer, text_encoder=text_encoder, vae=vae, transformer=transformer, scheduler=scheduler
+        )
+
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor)
+
+    # Adapted from https://github.com/PixArt-alpha/PixArt-alpha/blob/master/diffusion/model/utils.py
+    def mask_text_embeddings(self, emb, mask):
+        if emb.shape[0] == 1:
+            keep_index = mask.sum().item()
+            return emb[:, :, :keep_index, :], keep_index  # 1, 120, 4096 -> 1 7 4096
+        else:
+            masked_feature = emb * mask[:, None, :, None]  # 1 120 4096
+            return masked_feature, emb.shape[2]
+
+    # Adapted from diffusers.pipelines.deepfloyd_if.pipeline_if.encode_prompt
+    def encode_prompt(
+        self,
+        prompt: Union[str, List[str]],
+        do_classifier_free_guidance: bool = True,
+        negative_prompt: str = "",
+        num_images_per_prompt: int = 1,
+        device: Optional[torch.device] = None,
+        prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        clean_caption: bool = False,
+        mask_feature: bool = True,
+        dtype=None,
+    ):
+        r"""
+        Encodes the prompt into text encoder hidden states.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                prompt to be encoded
+            negative_prompt (`str` or `List[str]`, *optional*):
+                The prompt not to guide the video generation. If not defined, one has to pass `negative_prompt_embeds`
+                instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is less than `1`). For
+                Latte, this should be "".
+            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
+                whether to use classifier free guidance or not
+            num_images_per_prompt (`int`, *optional*, defaults to 1):
+                number of video that should be generated per prompt
+            device: (`torch.device`, *optional*):
+                torch device to place the resulting embeddings on
+            prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
+            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative text embeddings. For Latte, it's should be the embeddings of the "" string.
+            clean_caption (bool, defaults to `False`):
+                If `True`, the function will preprocess and clean the provided caption before encoding.
+            mask_feature: (bool, defaults to `True`):
+                If `True`, the function will mask the text embeddings.
+        """
+        embeds_initially_provided = prompt_embeds is not None and negative_prompt_embeds is not None
+
+        if device is None:
+            device = self._execution_device
+
+        if prompt is not None and isinstance(prompt, str):
+            batch_size = 1
+        elif prompt is not None and isinstance(prompt, list):
+            batch_size = len(prompt)
+        else:
+            batch_size = prompt_embeds.shape[0]
+
+        max_length = 120
+        if prompt_embeds is None:
+            prompt = self._text_preprocessing(prompt, clean_caption=clean_caption)
+            text_inputs = self.tokenizer(
+                prompt,
+                padding="max_length",
+                max_length=max_length,
+                truncation=True,
+                return_attention_mask=True,
+                add_special_tokens=True,
+                return_tensors="pt",
+            )
+            text_input_ids = text_inputs.input_ids
+            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids
+
+            if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
+                text_input_ids, untruncated_ids
+            ):
+                removed_text = self.tokenizer.batch_decode(untruncated_ids[:, max_length - 1 : -1])
+                logger.warning(
+                    "The following part of your input was truncated because CLIP can only handle sequences up to"
+                    f" {max_length} tokens: {removed_text}"
+                )
+
+            attention_mask = text_inputs.attention_mask.to(device)
+            prompt_embeds_attention_mask = attention_mask
+
+            prompt_embeds = self.text_encoder(text_input_ids.to(device), attention_mask=attention_mask)
+            prompt_embeds = prompt_embeds[0]
+        else:
+            prompt_embeds_attention_mask = torch.ones_like(prompt_embeds)
+
+        if self.text_encoder is not None:
+            dtype = self.text_encoder.dtype
+        elif self.transformer is not None:
+            dtype = self.transformer.dtype
+        else:
+            dtype = None
+
+        prompt_embeds = prompt_embeds.to(dtype=dtype, device=device)
+
+        bs_embed, seq_len, _ = prompt_embeds.shape
+        # duplicate text embeddings and attention mask for each generation per prompt, using mps friendly method
+        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
+        prompt_embeds_attention_mask = prompt_embeds_attention_mask.view(bs_embed, -1)
+        prompt_embeds_attention_mask = prompt_embeds_attention_mask.repeat(num_images_per_prompt, 1)
+
+        # get unconditional embeddings for classifier free guidance
+        if do_classifier_free_guidance and negative_prompt_embeds is None:
+            uncond_tokens = [negative_prompt] * batch_size if isinstance(negative_prompt, str) else negative_prompt
+            uncond_tokens = self._text_preprocessing(uncond_tokens, clean_caption=clean_caption)
+            max_length = prompt_embeds.shape[1]
+            uncond_input = self.tokenizer(
+                uncond_tokens,
+                padding="max_length",
+                max_length=max_length,
+                truncation=True,
+                return_attention_mask=True,
+                add_special_tokens=True,
+                return_tensors="pt",
+            )
+            attention_mask = uncond_input.attention_mask.to(device)
+
+            negative_prompt_embeds = self.text_encoder(
+                uncond_input.input_ids.to(device),
+                attention_mask=attention_mask,
+            )
+            negative_prompt_embeds = negative_prompt_embeds[0]
+
+        if do_classifier_free_guidance:
+            # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
+            seq_len = negative_prompt_embeds.shape[1]
+
+            negative_prompt_embeds = negative_prompt_embeds.to(dtype=dtype, device=device)
+
+            negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+            negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
+
+            # For classifier free guidance, we need to do two forward passes.
+            # Here we concatenate the unconditional and text embeddings into a single batch
+            # to avoid doing two forward passes
+        else:
+            negative_prompt_embeds = None
+
+        # Perform additional masking.
+        if mask_feature and not embeds_initially_provided:
+            prompt_embeds = prompt_embeds.unsqueeze(1)
+            masked_prompt_embeds, keep_indices = self.mask_text_embeddings(prompt_embeds, prompt_embeds_attention_mask)
+            masked_prompt_embeds = masked_prompt_embeds.squeeze(1)
+            masked_negative_prompt_embeds = (
+                negative_prompt_embeds[:, :keep_indices, :] if negative_prompt_embeds is not None else None
+            )
+
+            return masked_prompt_embeds, masked_negative_prompt_embeds
+
+        return prompt_embeds, negative_prompt_embeds
+
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
+    def prepare_extra_step_kwargs(self, generator, eta):
+        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
+        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
+        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
+        # and should be between [0, 1]
+
+        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
+        extra_step_kwargs = {}
+        if accepts_eta:
+            extra_step_kwargs["eta"] = eta
+
+        # check if the scheduler accepts generator
+        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
+        if accepts_generator:
+            extra_step_kwargs["generator"] = generator
+        return extra_step_kwargs
+
+    def check_inputs(
+        self,
+        prompt,
+        height,
+        width,
+        negative_prompt,
+        callback_on_step_end_tensor_inputs,
+        prompt_embeds=None,
+        negative_prompt_embeds=None,
+    ):
+        if height % 8 != 0 or width % 8 != 0:
+            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
+
+        if callback_on_step_end_tensor_inputs is not None and not all(
+            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
+        ):
+            raise ValueError(
+                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
+            )
+        if prompt is not None and prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
+                " only forward one of the two."
+            )
+        elif prompt is None and prompt_embeds is None:
+            raise ValueError(
+                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
+            )
+        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
+            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
+
+        if prompt is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `prompt`: {prompt} and `negative_prompt_embeds`:"
+                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
+            )
+
+        if negative_prompt is not None and negative_prompt_embeds is not None:
+            raise ValueError(
+                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:"
+                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
+            )
+
+        if prompt_embeds is not None and negative_prompt_embeds is not None:
+            if prompt_embeds.shape != negative_prompt_embeds.shape:
+                raise ValueError(
+                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
+                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
+                    f" {negative_prompt_embeds.shape}."
+                )
+
+    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._text_preprocessing
+    def _text_preprocessing(self, text, clean_caption=False):
+        if clean_caption and not is_bs4_available():
+            logger.warning(BACKENDS_MAPPING["bs4"][-1].format("Setting `clean_caption=True`"))
+            logger.warning("Setting `clean_caption` to False...")
+            clean_caption = False
+
+        if clean_caption and not is_ftfy_available():
+            logger.warning(BACKENDS_MAPPING["ftfy"][-1].format("Setting `clean_caption=True`"))
+            logger.warning("Setting `clean_caption` to False...")
+            clean_caption = False
+
+        if not isinstance(text, (tuple, list)):
+            text = [text]
+
+        def process(text: str):
+            if clean_caption:
+                text = self._clean_caption(text)
+                text = self._clean_caption(text)
+            else:
+                text = text.lower().strip()
+            return text
+
+        return [process(t) for t in text]
+
+    # Copied from diffusers.pipelines.deepfloyd_if.pipeline_if.IFPipeline._clean_caption
+    def _clean_caption(self, caption):
+        caption = str(caption)
+        caption = ul.unquote_plus(caption)
+        caption = caption.strip().lower()
+        caption = re.sub("<person>", "person", caption)
+        # urls:
+        caption = re.sub(
+            r"\b((?:https?:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
+            "",
+            caption,
+        )  # regex for urls
+        caption = re.sub(
+            r"\b((?:www:(?:\/{1,3}|[a-zA-Z0-9%])|[a-zA-Z0-9.\-]+[.](?:com|co|ru|net|org|edu|gov|it)[\w/-]*\b\/?(?!@)))",  # noqa
+            "",
+            caption,
+        )  # regex for urls
+        # html:
+        caption = BeautifulSoup(caption, features="html.parser").text
+
+        # @<nickname>
+        caption = re.sub(r"@[\w\d]+\b", "", caption)
+
+        # 31C0—31EF CJK Strokes
+        # 31F0—31FF Katakana Phonetic Extensions
+        # 3200—32FF Enclosed CJK Letters and Months
+        # 3300—33FF CJK Compatibility
+        # 3400—4DBF CJK Unified Ideographs Extension A
+        # 4DC0—4DFF Yijing Hexagram Symbols
+        # 4E00—9FFF CJK Unified Ideographs
+        caption = re.sub(r"[\u31c0-\u31ef]+", "", caption)
+        caption = re.sub(r"[\u31f0-\u31ff]+", "", caption)
+        caption = re.sub(r"[\u3200-\u32ff]+", "", caption)
+        caption = re.sub(r"[\u3300-\u33ff]+", "", caption)
+        caption = re.sub(r"[\u3400-\u4dbf]+", "", caption)
+        caption = re.sub(r"[\u4dc0-\u4dff]+", "", caption)
+        caption = re.sub(r"[\u4e00-\u9fff]+", "", caption)
+        #######################################################
+
+        # все виды тире / all types of dash --> "-"
+        caption = re.sub(
+            r"[\u002D\u058A\u05BE\u1400\u1806\u2010-\u2015\u2E17\u2E1A\u2E3A\u2E3B\u2E40\u301C\u3030\u30A0\uFE31\uFE32\uFE58\uFE63\uFF0D]+",  # noqa
+            "-",
+            caption,
+        )
+
+        # кавычки к одному стандарту
+        caption = re.sub(r"[`´«»“”¨]", '"', caption)
+        caption = re.sub(r"[‘’]", "'", caption)
+
+        # &quot;
+        caption = re.sub(r"&quot;?", "", caption)
+        # &amp
+        caption = re.sub(r"&amp", "", caption)
+
+        # ip adresses:
+        caption = re.sub(r"\d{1,3}\.\d{1,3}\.\d{1,3}\.\d{1,3}", " ", caption)
+
+        # article ids:
+        caption = re.sub(r"\d:\d\d\s+$", "", caption)
+
+        # \n
+        caption = re.sub(r"\\n", " ", caption)
+
+        # "#123"
+        caption = re.sub(r"#\d{1,3}\b", "", caption)
+        # "#12345.."
+        caption = re.sub(r"#\d{5,}\b", "", caption)
+        # "123456.."
+        caption = re.sub(r"\b\d{6,}\b", "", caption)
+        # filenames:
+        caption = re.sub(r"[\S]+\.(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)", "", caption)
+
+        #
+        caption = re.sub(r"[\"\']{2,}", r'"', caption)  # """AUSVERKAUFT"""
+        caption = re.sub(r"[\.]{2,}", r" ", caption)  # """AUSVERKAUFT"""
+
+        caption = re.sub(self.bad_punct_regex, r" ", caption)  # ***AUSVERKAUFT***, #AUSVERKAUFT
+        caption = re.sub(r"\s+\.\s+", r" ", caption)  # " . "
+
+        # this-is-my-cute-cat / this_is_my_cute_cat
+        regex2 = re.compile(r"(?:\-|\_)")
+        if len(re.findall(regex2, caption)) > 3:
+            caption = re.sub(regex2, " ", caption)
+
+        caption = ftfy.fix_text(caption)
+        caption = html.unescape(html.unescape(caption))
+
+        caption = re.sub(r"\b[a-zA-Z]{1,3}\d{3,15}\b", "", caption)  # jc6640
+        caption = re.sub(r"\b[a-zA-Z]+\d+[a-zA-Z]+\b", "", caption)  # jc6640vc
+        caption = re.sub(r"\b\d+[a-zA-Z]+\d+\b", "", caption)  # 6640vc231
+
+        caption = re.sub(r"(worldwide\s+)?(free\s+)?shipping", "", caption)
+        caption = re.sub(r"(free\s)?download(\sfree)?", "", caption)
+        caption = re.sub(r"\bclick\b\s(?:for|on)\s\w+", "", caption)
+        caption = re.sub(r"\b(?:png|jpg|jpeg|bmp|webp|eps|pdf|apk|mp4)(\simage[s]?)?", "", caption)
+        caption = re.sub(r"\bpage\s+\d+\b", "", caption)
+
+        caption = re.sub(r"\b\d*[a-zA-Z]+\d+[a-zA-Z]+\d+[a-zA-Z\d]*\b", r" ", caption)  # j2d1a2a...
+
+        caption = re.sub(r"\b\d+\.?\d*[xх×]\d+\.?\d*\b", "", caption)
+
+        caption = re.sub(r"\b\s+\:\s+", r": ", caption)
+        caption = re.sub(r"(\D[,\./])\b", r"\1 ", caption)
+        caption = re.sub(r"\s+", " ", caption)
+
+        caption.strip()
+
+        caption = re.sub(r"^[\"\']([\w\W]+)[\"\']$", r"\1", caption)
+        caption = re.sub(r"^[\'\_,\-\:;]", r"", caption)
+        caption = re.sub(r"[\'\_,\-\:\-\+]$", r"", caption)
+        caption = re.sub(r"^\.\S+$", "", caption)
+
+        return caption.strip()
+
+    # Copied from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_synth.TextToVideoSDPipeline.prepare_latents
+    def prepare_latents(
+        self, batch_size, num_channels_latents, num_frames, height, width, dtype, device, generator, latents=None
+    ):
+        shape = (
+            batch_size,
+            num_channels_latents,
+            num_frames,
+            height // self.vae_scale_factor,
+            width // self.vae_scale_factor,
+        )
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
+        if latents is None:
+            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
+        else:
+            latents = latents.to(device)
+
+        # scale the initial noise by the standard deviation required by the scheduler
+        latents = latents * self.scheduler.init_noise_sigma
+        return latents
+
+    @property
+    def guidance_scale(self):
+        return self._guidance_scale
+
+    # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
+    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+    # corresponds to doing no classifier free guidance.
+    @property
+    def do_classifier_free_guidance(self):
+        return self._guidance_scale > 1
+
+    @property
+    def num_timesteps(self):
+        return self._num_timesteps
+
+    @property
+    def interrupt(self):
+        return self._interrupt
+
+    @torch.no_grad()
+    @replace_example_docstring(EXAMPLE_DOC_STRING)
+    def __call__(
+        self,
+        prompt: Union[str, List[str]] = None,
+        negative_prompt: str = "",
+        num_inference_steps: int = 50,
+        timesteps: Optional[List[int]] = None,
+        guidance_scale: float = 7.5,
+        num_images_per_prompt: int = 1,
+        video_length: int = 16,
+        height: int = 512,
+        width: int = 512,
+        eta: float = 0.0,
+        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
+        latents: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.FloatTensor] = None,
+        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        output_type: str = "pil",
+        return_dict: bool = True,
+        callback_on_step_end: Optional[
+            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
+        ] = None,
+        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
+        clean_caption: bool = True,
+        mask_feature: bool = True,
+        enable_temporal_attentions: bool = True,
+        decode_chunk_size: Optional[int] = None,
+    ) -> Union[LattePipelineOutput, Tuple]:
+        """
+        Function invoked when calling the pipeline for generation.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts to guide the video generation. If not defined, one has to pass `prompt_embeds`.
+                instead.
+            negative_prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts not to guide the video generation. If not defined, one has to pass
+                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
+                less than `1`).
+            num_inference_steps (`int`, *optional*, defaults to 100):
+                The number of denoising steps. More denoising steps usually lead to a higher quality video at the
+                expense of slower inference.
+            timesteps (`List[int]`, *optional*):
+                Custom timesteps to use for the denoising process. If not defined, equal spaced `num_inference_steps`
+                timesteps are used. Must be in descending order.
+            guidance_scale (`float`, *optional*, defaults to 7.0):
+                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
+                `guidance_scale` is defined as `w` of equation 2. of [Imagen
+                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
+                1`. Higher guidance scale encourages to generate videos that are closely linked to the text `prompt`,
+                usually at the expense of lower video quality.
+            video_length (`int`, *optional*, defaults to 16):
+                The number of video frames that are generated. Defaults to 16 frames which at 8 frames per seconds
+            num_images_per_prompt (`int`, *optional*, defaults to 1):
+                The number of videos to generate per prompt.
+            height (`int`, *optional*, defaults to self.unet.config.sample_size):
+                The height in pixels of the generated video.
+            width (`int`, *optional*, defaults to self.unet.config.sample_size):
+                The width in pixels of the generated video.
+            eta (`float`, *optional*, defaults to 0.0):
+                Corresponds to parameter eta (η) in the DDIM paper: https://arxiv.org/abs/2010.02502. Only applies to
+                [`schedulers.DDIMScheduler`], will be ignored for others.
+            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
+                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
+                to make generation deterministic.
+            latents (`torch.FloatTensor`, *optional*):
+                Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for video
+                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
+                tensor will ge generated by sampling using the supplied random `generator`.
+            prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
+            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative text embeddings. For Latte this negative prompt should be "". If not provided,
+                negative_prompt_embeds will be generated from `negative_prompt` input argument.
+            output_type (`str`, *optional*, defaults to `"pil"`):
+                The output format of the generate video. Choose between
+                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
+            return_dict (`bool`, *optional*, defaults to `True`):
+                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
+            callback_on_step_end (`Callable[[int, int, Dict], None]`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
+                A callback function or a list of callback functions to be called at the end of each denoising step.
+            callback_on_step_end_tensor_inputs (`List[str]`, *optional*):
+                A list of tensor inputs that should be passed to the callback function. If not defined, all tensor
+                inputs will be passed.
+            clean_caption (`bool`, *optional*, defaults to `True`):
+                Whether or not to clean the caption before creating embeddings. Requires `beautifulsoup4` and `ftfy` to
+                be installed. If the dependencies are not installed, the embeddings will be created from the raw
+                prompt.
+            mask_feature (`bool` defaults to `True`): If set to `True`, the text embeddings will be masked.
+            enable_temporal_attentions (`bool`, *optional*, defaults to `True`): Whether to enable temporal attentions
+            decode_chunk_size (`int`, *optional*):
+                The number of frames to decode at a time. Higher chunk size leads to better temporal consistency at the
+                expense of more memory usage. By default, the decoder decodes all frames at once for maximal quality.
+                For lower memory usage, reduce `decode_chunk_size`.
+
+        Examples:
+
+        Returns:
+            [`~pipelines.latte.pipeline_latte.LattePipelineOutput`] or `tuple`:
+                If `return_dict` is `True`, [`~pipelines.latte.pipeline_latte.LattePipelineOutput`] is returned,
+                otherwise a `tuple` is returned where the first element is a list with the generated images
+        """
+        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
+            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
+
+        # 0. Default
+        decode_chunk_size = decode_chunk_size if decode_chunk_size is not None else video_length
+
+        # 1. Check inputs. Raise error if not correct
+        height = height or self.transformer.config.sample_size * self.vae_scale_factor
+        width = width or self.transformer.config.sample_size * self.vae_scale_factor
+        self.check_inputs(
+            prompt,
+            height,
+            width,
+            negative_prompt,
+            callback_on_step_end_tensor_inputs,
+            prompt_embeds,
+            negative_prompt_embeds,
+        )
+        self._guidance_scale = guidance_scale
+        self._interrupt = False
+
+        # 2. Default height and width to transformer
+        if prompt is not None and isinstance(prompt, str):
+            batch_size = 1
+        elif prompt is not None and isinstance(prompt, list):
+            batch_size = len(prompt)
+        else:
+            batch_size = prompt_embeds.shape[0]
+
+        device = self._execution_device
+
+        # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
+        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
+        # corresponds to doing no classifier free guidance.
+        do_classifier_free_guidance = guidance_scale > 1.0
+
+        # 3. Encode input prompt
+        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
+            prompt,
+            do_classifier_free_guidance,
+            negative_prompt=negative_prompt,
+            num_images_per_prompt=num_images_per_prompt,
+            device=device,
+            prompt_embeds=prompt_embeds,
+            negative_prompt_embeds=negative_prompt_embeds,
+            clean_caption=clean_caption,
+            mask_feature=mask_feature,
+        )
+        if do_classifier_free_guidance:
+            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
+
+        # 4. Prepare timesteps
+        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
+        self._num_timesteps = len(timesteps)
+
+        # 5. Prepare latents.
+        latent_channels = self.transformer.config.in_channels
+        latents = self.prepare_latents(
+            batch_size * num_images_per_prompt,
+            latent_channels,
+            video_length,
+            height,
+            width,
+            prompt_embeds.dtype,
+            device,
+            generator,
+            latents,
+        )
+
+        # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
+        extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
+
+        # 7. Denoising loop
+        num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
+
+        with self.progress_bar(total=num_inference_steps) as progress_bar:
+            for i, t in enumerate(timesteps):
+                if self.interrupt:
+                    continue
+
+                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
+                latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
+
+                current_timestep = t
+                if not torch.is_tensor(current_timestep):
+                    # TODO: this requires sync between CPU and GPU. So try to pass timesteps as tensors if you can
+                    # This would be a good case for the `match` statement (Python 3.10+)
+                    is_mps = latent_model_input.device.type == "mps"
+                    if isinstance(current_timestep, float):
+                        dtype = torch.float32 if is_mps else torch.float64
+                    else:
+                        dtype = torch.int32 if is_mps else torch.int64
+                    current_timestep = torch.tensor([current_timestep], dtype=dtype, device=latent_model_input.device)
+                elif len(current_timestep.shape) == 0:
+                    current_timestep = current_timestep[None].to(latent_model_input.device)
+                # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
+                current_timestep = current_timestep.expand(latent_model_input.shape[0])
+
+                # predict noise model_output
+                noise_pred = self.transformer(
+                    latent_model_input,
+                    encoder_hidden_states=prompt_embeds,
+                    timestep=current_timestep,
+                    enable_temporal_attentions=enable_temporal_attentions,
+                    return_dict=False,
+                )[0]
+
+                # perform guidance
+                if do_classifier_free_guidance:
+                    noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
+                    noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
+
+                # use learned sigma?
+                if not (
+                    hasattr(self.scheduler.config, "variance_type")
+                    and self.scheduler.config.variance_type in ["learned", "learned_range"]
+                ):
+                    noise_pred = noise_pred.chunk(2, dim=1)[0]
+
+                # compute previous video: x_t -> x_t-1
+                latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
+
+                # call the callback, if provided
+                if callback_on_step_end is not None:
+                    callback_kwargs = {}
+                    for k in callback_on_step_end_tensor_inputs:
+                        callback_kwargs[k] = locals()[k]
+                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
+
+                    latents = callback_outputs.pop("latents", latents)
+                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
+                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)
+
+                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
+                    progress_bar.update()
+
+        if not output_type == "latents":
+            video = self.decode_latents(latents, video_length, decode_chunk_size=14)
+            video = self.video_processor.postprocess_video(video=video, output_type=output_type)
+        else:
+            video = latents
+
+        # Offload all models
+        self.maybe_free_model_hooks()
+
+        if not return_dict:
+            return (video,)
+
+        return LattePipelineOutput(frames=video)
+
+    # Similar to diffusers.pipelines.stable_video_diffusion.pipeline_stable_video_diffusion.decode_latents
+    def decode_latents(self, latents: torch.Tensor, video_length: int, decode_chunk_size: int = 14):
+        # [batch, channels, frames, height, width] -> [batch*frames, channels, height, width]
+        latents = latents.permute(0, 2, 1, 3, 4).flatten(0, 1)
+
+        latents = 1 / self.vae.config.scaling_factor * latents
+
+        forward_vae_fn = self.vae._orig_mod.forward if is_compiled_module(self.vae) else self.vae.forward
+        accepts_num_frames = "num_frames" in set(inspect.signature(forward_vae_fn).parameters.keys())
+
+        # decode decode_chunk_size frames at a time to avoid OOM
+        frames = []
+        for i in range(0, latents.shape[0], decode_chunk_size):
+            num_frames_in = latents[i : i + decode_chunk_size].shape[0]
+            decode_kwargs = {}
+            if accepts_num_frames:
+                # we only pass num_frames_in if it's expected
+                decode_kwargs["num_frames"] = num_frames_in
+
+            frame = self.vae.decode(latents[i : i + decode_chunk_size], **decode_kwargs).sample
+            frames.append(frame)
+        frames = torch.cat(frames, dim=0)
+
+        # [batch*frames, channels, height, width] -> [batch, channels, frames, height, width]
+        frames = frames.reshape(-1, video_length, *frames.shape[1:]).permute(0, 2, 1, 3, 4)
+
+        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloa16
+        frames = frames.float()
+        return frames

src/diffusers/utils/dummy_pt_objects.py

@@ -197,6 +197,21 @@ class Kandinsky3UNet(metaclass=DummyObject):
         requires_backends(cls, ["torch"])
 
 
+class LatteTransformer3DModel(metaclass=DummyObject):
+    _backends = ["torch"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch"])
+
+    @classmethod
+    def from_config(cls, *args, **kwargs):
+        requires_backends(cls, ["torch"])
+
+    @classmethod
+    def from_pretrained(cls, *args, **kwargs):
+        requires_backends(cls, ["torch"])
+
+
 class LuminaNextDiT2DModel(metaclass=DummyObject):
     _backends = ["torch"]
 

src/diffusers/utils/dummy_torch_and_transformers_objects.py

@@ -677,6 +677,21 @@ class LatentConsistencyModelPipeline(metaclass=DummyObject):
         requires_backends(cls, ["torch", "transformers"])
 
 
+class LattePipeline(metaclass=DummyObject):
+    _backends = ["torch", "transformers"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch", "transformers"])
+
+    @classmethod
+    def from_config(cls, *args, **kwargs):
+        requires_backends(cls, ["torch", "transformers"])
+
+    @classmethod
+    def from_pretrained(cls, *args, **kwargs):
+        requires_backends(cls, ["torch", "transformers"])
+
+
 class LDMTextToImagePipeline(metaclass=DummyObject):
     _backends = ["torch", "transformers"]
 

tests/pipelines/latte/test_latte.py

+# coding=utf-8
+# Copyright 2024 Latte Team and HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import inspect
+import tempfile
+import unittest
+
+import numpy as np
+import torch
+from transformers import AutoTokenizer, T5EncoderModel
+
+from diffusers import (
+    AutoencoderKL,
+    DDIMScheduler,
+    LattePipeline,
+    LatteTransformer3DModel,
+)
+from diffusers.utils.testing_utils import (
+    enable_full_determinism,
+    numpy_cosine_similarity_distance,
+    require_torch_gpu,
+    slow,
+    torch_device,
+)
+
+from ..pipeline_params import TEXT_TO_IMAGE_BATCH_PARAMS, TEXT_TO_IMAGE_IMAGE_PARAMS, TEXT_TO_IMAGE_PARAMS
+from ..test_pipelines_common import PipelineTesterMixin, to_np
+
+
+enable_full_determinism()
+
+
+class LattePipelineFastTests(PipelineTesterMixin, unittest.TestCase):
+    pipeline_class = LattePipeline
+    params = TEXT_TO_IMAGE_PARAMS - {"cross_attention_kwargs"}
+    batch_params = TEXT_TO_IMAGE_BATCH_PARAMS
+    image_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+
+    required_optional_params = PipelineTesterMixin.required_optional_params
+
+    def get_dummy_components(self):
+        torch.manual_seed(0)
+        transformer = LatteTransformer3DModel(
+            sample_size=8,
+            num_layers=1,
+            patch_size=2,
+            attention_head_dim=8,
+            num_attention_heads=3,
+            caption_channels=32,
+            in_channels=4,
+            cross_attention_dim=24,
+            out_channels=8,
+            attention_bias=True,
+            activation_fn="gelu-approximate",
+            num_embeds_ada_norm=1000,
+            norm_type="ada_norm_single",
+            norm_elementwise_affine=False,
+            norm_eps=1e-6,
+        )
+        torch.manual_seed(0)
+        vae = AutoencoderKL()
+
+        scheduler = DDIMScheduler()
+        text_encoder = T5EncoderModel.from_pretrained("hf-internal-testing/tiny-random-t5")
+
+        tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-t5")
+
+        components = {
+            "transformer": transformer.eval(),
+            "vae": vae.eval(),
+            "scheduler": scheduler,
+            "text_encoder": text_encoder.eval(),
+            "tokenizer": tokenizer,
+        }
+        return components
+
+    def get_dummy_inputs(self, device, seed=0):
+        if str(device).startswith("mps"):
+            generator = torch.manual_seed(seed)
+        else:
+            generator = torch.Generator(device=device).manual_seed(seed)
+        inputs = {
+            "prompt": "A painting of a squirrel eating a burger",
+            "negative_prompt": "low quality",
+            "generator": generator,
+            "num_inference_steps": 2,
+            "guidance_scale": 5.0,
+            "height": 8,
+            "width": 8,
+            "video_length": 1,
+            "output_type": "pt",
+            "clean_caption": False,
+        }
+        return inputs
+
+    def test_inference(self):
+        device = "cpu"
+
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+        pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        video = pipe(**inputs).frames
+        generated_video = video[0]
+
+        self.assertEqual(generated_video.shape, (1, 3, 8, 8))
+        expected_video = torch.randn(1, 3, 8, 8)
+        max_diff = np.abs(generated_video - expected_video).max()
+        self.assertLessEqual(max_diff, 1e10)
+
+    def test_callback_inputs(self):
+        sig = inspect.signature(self.pipeline_class.__call__)
+        has_callback_tensor_inputs = "callback_on_step_end_tensor_inputs" in sig.parameters
+        has_callback_step_end = "callback_on_step_end" in sig.parameters
+
+        if not (has_callback_tensor_inputs and has_callback_step_end):
+            return
+
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+        pipe = pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+        self.assertTrue(
+            hasattr(pipe, "_callback_tensor_inputs"),
+            f" {self.pipeline_class} should have `_callback_tensor_inputs` that defines a list of tensor variables its callback function can use as inputs",
+        )
+
+        def callback_inputs_subset(pipe, i, t, callback_kwargs):
+            # iterate over callback args
+            for tensor_name, tensor_value in callback_kwargs.items():
+                # check that we're only passing in allowed tensor inputs
+                assert tensor_name in pipe._callback_tensor_inputs
+
+            return callback_kwargs
+
+        def callback_inputs_all(pipe, i, t, callback_kwargs):
+            for tensor_name in pipe._callback_tensor_inputs:
+                assert tensor_name in callback_kwargs
+
+            # iterate over callback args
+            for tensor_name, tensor_value in callback_kwargs.items():
+                # check that we're only passing in allowed tensor inputs
+                assert tensor_name in pipe._callback_tensor_inputs
+
+            return callback_kwargs
+
+        inputs = self.get_dummy_inputs(torch_device)
+
+        # Test passing in a subset
+        inputs["callback_on_step_end"] = callback_inputs_subset
+        inputs["callback_on_step_end_tensor_inputs"] = ["latents"]
+        output = pipe(**inputs)[0]
+
+        # Test passing in a everything
+        inputs["callback_on_step_end"] = callback_inputs_all
+        inputs["callback_on_step_end_tensor_inputs"] = pipe._callback_tensor_inputs
+        output = pipe(**inputs)[0]
+
+        def callback_inputs_change_tensor(pipe, i, t, callback_kwargs):
+            is_last = i == (pipe.num_timesteps - 1)
+            if is_last:
+                callback_kwargs["latents"] = torch.zeros_like(callback_kwargs["latents"])
+            return callback_kwargs
+
+        inputs["callback_on_step_end"] = callback_inputs_change_tensor
+        inputs["callback_on_step_end_tensor_inputs"] = pipe._callback_tensor_inputs
+        output = pipe(**inputs)[0]
+        assert output.abs().sum() < 1e10
+
+    def test_inference_batch_single_identical(self):
+        self._test_inference_batch_single_identical(batch_size=3, expected_max_diff=1e-3)
+
+    def test_attention_slicing_forward_pass(self):
+        pass
+
+    def test_save_load_optional_components(self):
+        if not hasattr(self.pipeline_class, "_optional_components"):
+            return
+
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        for component in pipe.components.values():
+            if hasattr(component, "set_default_attn_processor"):
+                component.set_default_attn_processor()
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(torch_device)
+
+        prompt = inputs["prompt"]
+        generator = inputs["generator"]
+
+        (
+            prompt_embeds,
+            negative_prompt_embeds,
+        ) = pipe.encode_prompt(prompt)
+
+        # inputs with prompt converted to embeddings
+        inputs = {
+            "prompt_embeds": prompt_embeds,
+            "negative_prompt": None,
+            "negative_prompt_embeds": negative_prompt_embeds,
+            "generator": generator,
+            "num_inference_steps": 2,
+            "guidance_scale": 5.0,
+            "height": 8,
+            "width": 8,
+            "video_length": 1,
+            "mask_feature": False,
+            "output_type": "pt",
+            "clean_caption": False,
+        }
+
+        # set all optional components to None
+        for optional_component in pipe._optional_components:
+            setattr(pipe, optional_component, None)
+
+        output = pipe(**inputs)[0]
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            pipe.save_pretrained(tmpdir, safe_serialization=False)
+            pipe_loaded = self.pipeline_class.from_pretrained(tmpdir)
+            pipe_loaded.to(torch_device)
+
+            for component in pipe_loaded.components.values():
+                if hasattr(component, "set_default_attn_processor"):
+                    component.set_default_attn_processor()
+
+            pipe_loaded.set_progress_bar_config(disable=None)
+
+        for optional_component in pipe._optional_components:
+            self.assertTrue(
+                getattr(pipe_loaded, optional_component) is None,
+                f"`{optional_component}` did not stay set to None after loading.",
+            )
+
+        output_loaded = pipe_loaded(**inputs)[0]
+
+        max_diff = np.abs(to_np(output) - to_np(output_loaded)).max()
+        self.assertLess(max_diff, 1.0)
+
+
+@slow
+@require_torch_gpu
+class LattePipelineIntegrationTests(unittest.TestCase):
+    prompt = "A painting of a squirrel eating a burger."
+
+    def setUp(self):
+        super().setUp()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def tearDown(self):
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_latte(self):
+        generator = torch.Generator("cpu").manual_seed(0)
+
+        pipe = LattePipeline.from_pretrained("maxin-cn/Latte-1", torch_dtype=torch.float16)
+        pipe.enable_model_cpu_offload()
+        prompt = self.prompt
+
+        videos = pipe(
+            prompt=prompt,
+            height=512,
+            width=512,
+            generator=generator,
+            num_inference_steps=2,
+            clean_caption=False,
+        ).frames
+
+        video = videos[0]
+        expected_video = torch.randn(1, 512, 512, 3).numpy()
+
+        max_diff = numpy_cosine_similarity_distance(video.flatten(), expected_video)
+        assert max_diff < 1e-3, f"Max diff is too high. got {video.flatten()}"