[Chore] Bump Python version and apply formatting changes for v0.12 release (#126)

Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>

vllm_omni/diffusion/cache/cache_dit_backend.py

@@ -7,7 +7,8 @@ This module provides a CacheDiTBackend class to enable cache-dit acceleration on
 pipelines in vllm-omni, supporting both single and dual-transformer architectures.
 """
 
-from typing import Any, Callable, Optional
+from collections.abc import Callable
+from typing import Any, Optional
 
 from vllm.logger import init_logger
 
@@ -311,8 +312,8 @@ class CacheDiTBackend(CacheBackend):
         super().__init__(config)
 
         # Cache-dit specific attributes
-        self._refresh_func: Optional[Callable[[Any, int, bool], None]] = None
-        self._last_num_inference_steps: Optional[int] = None
+        self._refresh_func: Callable[[Any, int, bool], None] | None = None
+        self._last_num_inference_steps: int | None = None
 
     def enable(self, pipeline: Any) -> None:
         """Enable cache-dit on the pipeline if configured.

vllm_omni/diffusion/cache/teacache/config.py

@@ -2,7 +2,6 @@
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 
 from dataclasses import dataclass
-from typing import Optional
 
 # Model-specific polynomial coefficients for rescaling L1 distances
 # These coefficients account for model-specific characteristics in how embeddings change
@@ -51,7 +50,7 @@ class TeaCacheConfig:
     """
 
     rel_l1_thresh: float = 0.2
-    coefficients: Optional[list[float]] = None
+    coefficients: list[float] | None = None
     model_type: str = "QwenImagePipeline"
 
     def __post_init__(self) -> None:

vllm_omni/diffusion/cache/teacache/extractors.py

@@ -13,8 +13,9 @@ all model-specific information needed for generic caching, including preprocessi
 transformer execution, and postprocessing logic.
 """
 
+from collections.abc import Callable
 from dataclasses import dataclass
-from typing import Any, Callable, Union
+from typing import Any
 
 import torch
 import torch.nn as nn
@@ -146,7 +147,7 @@ def extract_qwen_context(
     hidden_states: torch.Tensor,
     encoder_hidden_states: torch.Tensor,
     encoder_hidden_states_mask: torch.Tensor,
-    timestep: Union[torch.Tensor, float, int],
+    timestep: torch.Tensor | float | int,
     img_shapes: torch.Tensor,
     txt_seq_lens: torch.Tensor,
     guidance: torch.Tensor | None = None,

vllm_omni/diffusion/cache/teacache/state.py

@@ -7,8 +7,6 @@ TeaCache state management.
 This module manages the state for TeaCache hooks across diffusion timesteps.
 """
 
-from typing import Optional
-
 import torch
 
 
@@ -27,9 +25,9 @@ class TeaCacheState:
 
         # Caching state
         self.accumulated_rel_l1_distance = 0.0
-        self.previous_modulated_input: Optional[torch.Tensor] = None
-        self.previous_residual: Optional[torch.Tensor] = None
-        self.previous_residual_encoder: Optional[torch.Tensor] = None
+        self.previous_modulated_input: torch.Tensor | None = None
+        self.previous_residual: torch.Tensor | None = None
+        self.previous_residual_encoder: torch.Tensor | None = None
 
     def reset(self) -> None:
         """Reset all state variables for a new inference run."""