[pre-commit.ci] pre-commit autoupdate

.pre-commit-config.yaml

@@ -7,7 +7,7 @@ ci:
 
 repos:
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.11.2
+    rev: v0.16.0
     hooks:
       # Run the linter.
       - id: ruff
@@ -18,7 +18,7 @@ repos:
         types_or: [python, pyi, jupyter]
 
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v5.0.0
+    rev: v6.0.0
     hooks:
       - id: trailing-whitespace
         exclude_types:

src/animatediff/__init__.py

@@ -47,21 +47,21 @@ def get_dir(dirname: str = "data") -> Path:
 
 
 __all__ = [
-    "__version__",
-    "version_tuple",
-    "PACKAGE",
-    "PACKAGE_ROOT",
     "HF_HOME",
     "HF_HUB_CACHE",
+    "PACKAGE",
+    "PACKAGE_ROOT",
+    "__version__",
+    "cli",
     "console",
     "err_console",
+    "generate",
     "get_dir",
     "models",
     "pipelines",
     "rife",
-    "utils",
-    "cli",
-    "generate",
     "schedulers",
     "settings",
+    "utils",
+    "version_tuple",
 ]

src/animatediff/cli.py

@@ -1,7 +1,7 @@
 import logging
 from datetime import datetime
 from pathlib import Path
-from typing import Annotated, Optional
+from typing import Annotated
 
 import torch
 import typer
@@ -57,8 +57,8 @@
     rife_app = None
 
 # mildly cursed globals to allow for reuse of the pipeline if we're being called as a module
-pipeline: Optional[AnimationPipeline] = None
-last_model_path: Optional[Path] = None
+pipeline: AnimationPipeline | None = None
+last_model_path: Path | None = None
 
 
 def version_callback(value: bool):
@@ -125,7 +125,7 @@ def generate(
         ),
     ] = 16,
     context: Annotated[
-        Optional[int],
+        int | None,
         typer.Option(
             "--context",
             "-C",
@@ -137,7 +137,7 @@ def generate(
         ),
     ] = None,
     overlap: Annotated[
-        Optional[int],
+        int | None,
         typer.Option(
             "--overlap",
             "-O",
@@ -149,7 +149,7 @@ def generate(
         ),
     ] = None,
     stride: Annotated[
-        Optional[int],
+        int | None,
         typer.Option(
             "--stride",
             "-S",
@@ -229,7 +229,7 @@ def generate(
         ),
     ] = False,
     version: Annotated[
-        Optional[bool],
+        bool | None,
         typer.Option(
             "--version",
             "-v",
@@ -387,7 +387,7 @@ def convert(
         ),
     ] = ...,
     out_dir: Annotated[
-        Optional[Path],
+        Path | None,
         typer.Option(
             "--out-dir",
             "-o",
@@ -417,7 +417,7 @@ def merge(
         ),
     ] = ...,
     out_dir: Annotated[
-        Optional[Path],
+        Path | None,
         typer.Option(
             "--out-dir",
             "-o",

src/animatediff/generate.py

@@ -2,7 +2,6 @@
 import re
 from os import PathLike
 from pathlib import Path
-from typing import Optional, Union
 
 import torch
 from diffusers import AutoencoderKL, StableDiffusionPipeline
@@ -26,7 +25,7 @@
 
 
 def create_pipeline(
-    base_model: Union[str, PathLike] = default_base_path,
+    base_model: str | PathLike = default_base_path,
     model_config: ModelConfig = ...,
     infer_config: InferenceConfig = ...,
     use_xformers: bool = True,
@@ -136,8 +135,8 @@ def create_pipeline(
 
 def run_inference(
     pipeline: AnimationPipeline,
-    prompt: Optional[str] = None,
-    prompt_map: Optional[dict[int, str]] = None,
+    prompt: str | None = None,
+    prompt_map: dict[int, str] | None = None,
     n_prompt: str = ...,
     seed: int = -1,
     steps: int = 25,
@@ -186,7 +185,7 @@ def run_inference(
     # Trim and clean up the prompt for filename use
     prompt_str = prompt or next(iter(prompt_map.values()))
     prompt_tags = [re_clean_prompt.sub("", tag).strip().replace(" ", "-") for tag in prompt_str.split(",")]
-    prompt_str = "_".join((prompt_tags[:6]))
+    prompt_str = "_".join(prompt_tags[:6])
 
     # generate the output filename and save the video
     out_str = f"{idx:02d}_{seed}_{prompt_str}"[:250]

src/animatediff/models/attention.py

@@ -1,7 +1,7 @@
 # Adapted from https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention.py
 
 from dataclasses import dataclass
-from typing import Any, Dict, Optional
+from typing import Any
 
 import torch
 from diffusers.configuration_utils import ConfigMixin, register_to_config
@@ -25,14 +25,14 @@ def __init__(
         self,
         num_attention_heads: int = 16,
         attention_head_dim: int = 88,
-        in_channels: Optional[int] = None,
+        in_channels: int | None = None,
         num_layers: int = 1,
         dropout: float = 0.0,
         norm_num_groups: int = 32,
-        cross_attention_dim: Optional[int] = None,
+        cross_attention_dim: int | None = None,
         attention_bias: bool = False,
         activation_fn: str = "geglu",
-        num_embeds_ada_norm: Optional[int] = None,
+        num_embeds_ada_norm: int | None = None,
         use_linear_projection: bool = False,
         only_cross_attention: bool = False,
         upcast_attention: bool = False,
@@ -86,11 +86,11 @@ def __init__(
     def forward(
         self,
         hidden_states: torch.Tensor,
-        encoder_hidden_states: Optional[torch.Tensor] = None,
-        timestep: Optional[torch.LongTensor] = None,
-        cross_attention_kwargs: Dict[str, Any] = None,
-        attention_mask: Optional[torch.Tensor] = None,
-        encoder_attention_mask: Optional[torch.Tensor] = None,
+        encoder_hidden_states: torch.Tensor | None = None,
+        timestep: torch.LongTensor | None = None,
+        cross_attention_kwargs: dict[str, Any] = None,
+        attention_mask: torch.Tensor | None = None,
+        encoder_attention_mask: torch.Tensor | None = None,
         return_dict: bool = True,
     ) -> tuple[Tensor] | Transformer3DModelOutput:
         # validate input dim
@@ -180,9 +180,9 @@ def __init__(
         num_attention_heads: int,
         attention_head_dim: int,
         dropout: float = 0.0,
-        cross_attention_dim: Optional[int] = None,
+        cross_attention_dim: int | None = None,
         activation_fn: str = "geglu",
-        num_embeds_ada_norm: Optional[int] = None,
+        num_embeds_ada_norm: int | None = None,
         attention_bias: bool = False,
         only_cross_attention: bool = False,
         upcast_attention: bool = False,
@@ -262,11 +262,11 @@ def __init__(
     def forward(
         self,
         hidden_states: torch.FloatTensor,
-        attention_mask: Optional[torch.FloatTensor] = None,
-        encoder_hidden_states: Optional[torch.FloatTensor] = None,
-        encoder_attention_mask: Optional[torch.FloatTensor] = None,
-        timestep: Optional[torch.LongTensor] = None,
-        cross_attention_kwargs: Dict[str, Any] = None,
+        attention_mask: torch.FloatTensor | None = None,
+        encoder_hidden_states: torch.FloatTensor | None = None,
+        encoder_attention_mask: torch.FloatTensor | None = None,
+        timestep: torch.LongTensor | None = None,
+        cross_attention_kwargs: dict[str, Any] = None,
         video_length=None,
     ):
         # SparseCausal-Attention

src/animatediff/models/clip.py

@@ -1,5 +1,3 @@
-from typing import Optional, Tuple, Union
-
 import torch
 from transformers.modeling_outputs import BaseModelOutput, BaseModelOutputWithPooling
 from transformers.models.clip import CLIPTextConfig, CLIPTextModel
@@ -27,14 +25,14 @@ class CLIPSkipTextTransformer(CLIPTextTransformer):
     @replace_return_docstrings(output_type=BaseModelOutputWithPooling, config_class=CLIPTextConfig)
     def forward(
         self,
-        input_ids: Optional[torch.Tensor] = None,
-        attention_mask: Optional[torch.Tensor] = None,
-        position_ids: Optional[torch.Tensor] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
-        return_dict: Optional[bool] = None,
+        input_ids: torch.Tensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.Tensor | None = None,
+        output_attentions: bool | None = None,
+        output_hidden_states: bool | None = None,
+        return_dict: bool | None = None,
         clip_skip: int = 1,
-    ) -> Union[Tuple, BaseModelOutputWithPooling]:
+    ) -> tuple | BaseModelOutputWithPooling:
         r"""
         Returns:
 
@@ -122,14 +120,14 @@ def __init__(self, config: CLIPTextConfig):
     @replace_return_docstrings(output_type=BaseModelOutputWithPooling, config_class=CLIPTextConfig)
     def forward(
         self,
-        input_ids: Optional[torch.Tensor] = None,
-        attention_mask: Optional[torch.Tensor] = None,
-        position_ids: Optional[torch.Tensor] = None,
-        output_attentions: Optional[bool] = None,
-        output_hidden_states: Optional[bool] = None,
-        return_dict: Optional[bool] = None,
+        input_ids: torch.Tensor | None = None,
+        attention_mask: torch.Tensor | None = None,
+        position_ids: torch.Tensor | None = None,
+        output_attentions: bool | None = None,
+        output_hidden_states: bool | None = None,
+        return_dict: bool | None = None,
         clip_skip: int = 1,
-    ) -> Union[Tuple, BaseModelOutputWithPooling]:
+    ) -> tuple | BaseModelOutputWithPooling:
         r"""
         Returns:
 

src/animatediff/models/motion_module.py

@@ -1,6 +1,5 @@
 import math
 from dataclasses import dataclass
-from typing import Optional
 
 import torch
 from diffusers.models.attention import Attention, FeedForward
@@ -125,8 +124,8 @@ def __init__(
     def forward(
         self,
         hidden_states: Tensor,
-        encoder_hidden_states: Optional[Tensor] = None,
-        attention_mask: Optional[Tensor] = None,
+        encoder_hidden_states: Tensor | None = None,
+        attention_mask: Tensor | None = None,
     ):
         assert hidden_states.dim() == 5, (
             f"Expected hidden_states to have ndim=5, but got ndim={hidden_states.dim()}."
@@ -251,7 +250,7 @@ class VersatileAttention(Attention):
     def __init__(
         self,
         attention_mode: str = None,
-        cross_frame_attention_mode: Optional[str] = None,
+        cross_frame_attention_mode: str | None = None,
         temporal_position_encoding: bool = False,
         temporal_position_encoding_max_len: int = 24,
         *args,

src/animatediff/models/resnet.py

@@ -1,6 +1,5 @@
 # Adapted from https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/resnet.py
 
-from typing import Optional
 
 import torch
 import torch.nn.functional as F
@@ -24,7 +23,7 @@ def __init__(
         channels: int,
         use_conv: bool = False,
         use_conv_transpose: bool = False,
-        out_channels: Optional[int] = None,
+        out_channels: int | None = None,
         name="conv",
     ):
         super().__init__()
@@ -75,7 +74,7 @@ def __init__(
         self,
         channels: int,
         use_conv: bool = False,
-        out_channels: Optional[int] = None,
+        out_channels: int | None = None,
         padding: int = 1,
         name="conv",
     ):