Sunset Python 3.8 & get rid of explicit typing exports where possible

.github/workflows/notify_slack_about_release.yml

@@ -15,7 +15,7 @@ jobs:
     - name: Setup Python
       uses: actions/setup-python@v6
       with:
-        python-version: '3.8'
+        python-version: '3.10'
 
     - name: Notify Slack about the release
       env:

.github/workflows/pr_dependency_test.yml

@@ -22,7 +22,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
       - name: Install dependencies
         run: |
           pip install -e .

.github/workflows/pr_tests.yml

@@ -35,7 +35,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
       - name: Install dependencies
         run: |
           pip install --upgrade pip
@@ -55,7 +55,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
       - name: Install dependencies
         run: |
           pip install --upgrade pip

.github/workflows/pr_tests_gpu.yml

@@ -36,7 +36,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
       - name: Install dependencies
         run: |
           pip install --upgrade pip
@@ -56,7 +56,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
       - name: Install dependencies
         run: |
           pip install --upgrade pip

.github/workflows/pr_torch_dependency_test.yml

@@ -22,7 +22,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
       - name: Install dependencies
         run: |
           pip install -e .

.github/workflows/pypi_publish.yaml

@@ -20,7 +20,7 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v6
         with:
-          python-version: '3.8'
+          python-version: '3.10'
 
       - name: Fetch latest branch
         id: fetch_latest_branch
@@ -47,7 +47,7 @@ jobs:
       - name: Setup Python
         uses: actions/setup-python@v6
         with:
-          python-version: "3.8"
+          python-version: "3.10"
 
       - name: Install dependencies
         run: |

.github/workflows/stale.yml

@@ -20,7 +20,7 @@ jobs:
     - name: Setup Python
       uses: actions/setup-python@v6
       with:
-        python-version: 3.8
+        python-version: 3.10
 
     - name: Install requirements
       run: |

LICENSE

@@ -144,7 +144,7 @@
       agreed to in writing, Licensor provides the Work (and each
       Contributor provides its Contributions) on an "AS IS" BASIS,
       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
-      implied, including, without limitation, any warranties or conditions
+      implied, including, without limitation, Any warranties or conditions
       of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
       PARTICULAR PURPOSE. You are solely responsible for determining the
       appropriateness of using or redistributing the Work and assume any

benchmarks/benchmarking_utils.py

@@ -6,7 +6,7 @@
 import threading
 from contextlib import nullcontext
 from dataclasses import dataclass
-from typing import Any, Callable, Dict, Optional, Union
+from typing import Any, Callable
 
 import pandas as pd
 import torch
@@ -91,10 +91,10 @@ def model_init_fn(model_cls, group_offload_kwargs=None, layerwise_upcasting=Fals
 class BenchmarkScenario:
     name: str
     model_cls: ModelMixin
-    model_init_kwargs: Dict[str, Any]
+    model_init_kwargs: dict[str, Any]
     model_init_fn: Callable
     get_model_input_dict: Callable
-    compile_kwargs: Optional[Dict[str, Any]] = None
+    compile_kwargs: dict[str, Any] | None = None
 
 
 @require_torch_gpu
@@ -176,7 +176,7 @@ def run_benchmark(self, scenario: BenchmarkScenario):
             result["fullgraph"], result["mode"] = None, None
         return result
 
-    def run_bencmarks_and_collate(self, scenarios: Union[BenchmarkScenario, list[BenchmarkScenario]], filename: str):
+    def run_bencmarks_and_collate(self, scenarios: BenchmarkScenario | list[BenchmarkScenario], filename: str):
         if not isinstance(scenarios, list):
             scenarios = [scenarios]
         record_queue = queue.Queue()
@@ -214,10 +214,10 @@ def _run_phase(
         *,
         model_cls: ModelMixin,
         init_fn: Callable,
-        init_kwargs: Dict[str, Any],
+        init_kwargs: dict[str, Any],
         get_input_fn: Callable,
-        compile_kwargs: Optional[Dict[str, Any]],
-    ) -> Dict[str, float]:
+        compile_kwargs: dict[str, Any] | None = None,
+    ) -> dict[str, float]:
         # setup
         self.pre_benchmark()
 

examples/cogvideo/train_cogvideox_image_to_video_lora.py

@@ -432,9 +432,9 @@ def get_args():
 class VideoDataset(Dataset):
     def __init__(
         self,
-        instance_data_root: Optional[str] = None,
-        dataset_name: Optional[str] = None,
-        dataset_config_name: Optional[str] = None,
+        instance_data_root: str | None = None,
+        dataset_name: str | None = None,
+        dataset_config_name: str | None = None,
         caption_column: str = "text",
         video_column: str = "video",
         height: int = 480,
@@ -443,8 +443,8 @@ def __init__(
         max_num_frames: int = 49,
         skip_frames_start: int = 0,
         skip_frames_end: int = 0,
-        cache_dir: Optional[str] = None,
-        id_token: Optional[str] = None,
+        cache_dir: str | None = None,
+        id_token: str | None = None,
     ) -> None:
         super().__init__()
 

examples/cogvideo/train_cogvideox_lora.py

@@ -416,9 +416,9 @@ def get_args():
 class VideoDataset(Dataset):
     def __init__(
         self,
-        instance_data_root: Optional[str] = None,
-        dataset_name: Optional[str] = None,
-        dataset_config_name: Optional[str] = None,
+        instance_data_root: str | None = None,
+        dataset_name: str | None = None,
+        dataset_config_name: str | None = None,
         caption_column: str = "text",
         video_column: str = "video",
         height: int = 480,
@@ -428,8 +428,8 @@ def __init__(
         max_num_frames: int = 49,
         skip_frames_start: int = 0,
         skip_frames_end: int = 0,
-        cache_dir: Optional[str] = None,
-        id_token: Optional[str] = None,
+        cache_dir: str | None = None,
+        id_token: str | None = None,
     ) -> None:
         super().__init__()
 

examples/community/README_community_scripts.md

@@ -260,7 +260,7 @@ class SDPromptSchedulingCallback(PipelineCallback):
 
     def callback_fn(
         self, pipeline, step_index, timestep, callback_kwargs
-    ) -> Dict[str, Any]:
+    ) -> dict[str, Any]:
         cutoff_step_ratio = self.config.cutoff_step_ratio
         cutoff_step_index = self.config.cutoff_step_index
         if isinstance(self.config.encoded_prompt, tuple):
@@ -343,7 +343,7 @@ class SDXLPromptSchedulingCallback(PipelineCallback):
 
     def callback_fn(
         self, pipeline, step_index, timestep, callback_kwargs
-    ) -> Dict[str, Any]:
+    ) -> dict[str, Any]:
         cutoff_step_ratio = self.config.cutoff_step_ratio
         cutoff_step_index = self.config.cutoff_step_index
         if isinstance(self.config.encoded_prompt, tuple):

examples/community/adaptive_mask_inpainting.py

@@ -871,7 +871,7 @@ def __call__(
         latents: Optional[torch.FloatTensor] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
         callback_steps: int = 1,

examples/community/bit_diffusion.py

@@ -231,9 +231,9 @@ def __call__(
         height: Optional[int] = 256,
         width: Optional[int] = 256,
         num_inference_steps: Optional[int] = 50,
-        generator: Optional[torch.Generator] = None,
+        generator: torch.Generator | None = None,
         batch_size: Optional[int] = 1,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         **kwargs,
     ) -> Union[Tuple, ImagePipelineOutput]:

examples/community/clip_guided_images_mixing_stable_diffusion.py

@@ -235,8 +235,8 @@ def __call__(
         self,
         style_image: Union[torch.Tensor, PIL.Image.Image],
         content_image: Union[torch.Tensor, PIL.Image.Image],
-        style_prompt: Optional[str] = None,
-        content_prompt: Optional[str] = None,
+        style_prompt: str | None = None,
+        content_prompt: str | None = None,
         height: Optional[int] = 512,
         width: Optional[int] = 512,
         noise_strength: float = 0.6,
@@ -245,8 +245,8 @@ def __call__(
         batch_size: Optional[int] = 1,
         eta: float = 0.0,
         clip_guidance_scale: Optional[float] = 100,
-        generator: Optional[torch.Generator] = None,
-        output_type: Optional[str] = "pil",
+        generator: torch.Generator | None = None,
+        output_type: str | None = "pil",
         return_dict: bool = True,
         slerp_latent_style_strength: float = 0.8,
         slerp_prompt_style_strength: float = 0.1,

examples/community/clip_guided_stable_diffusion.py

@@ -179,9 +179,9 @@ def __call__(
         clip_prompt: Optional[Union[str, List[str]]] = None,
         num_cutouts: Optional[int] = 4,
         use_cutouts: Optional[bool] = True,
-        generator: Optional[torch.Generator] = None,
+        generator: torch.Generator | None = None,
         latents: Optional[torch.Tensor] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
     ):
         if isinstance(prompt, str):

examples/community/clip_guided_stable_diffusion_img2img.py

@@ -316,9 +316,9 @@ def __call__(
         clip_prompt: Optional[Union[str, List[str]]] = None,
         num_cutouts: Optional[int] = 4,
         use_cutouts: Optional[bool] = True,
-        generator: Optional[torch.Generator] = None,
+        generator: torch.Generator | None = None,
         latents: Optional[torch.Tensor] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
     ):
         if isinstance(prompt, str):

examples/community/composable_stable_diffusion.py

@@ -357,13 +357,13 @@ def __call__(
         negative_prompt: Optional[Union[str, List[str]]] = None,
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
-        generator: Optional[torch.Generator] = None,
+        generator: torch.Generator | None = None,
         latents: Optional[torch.Tensor] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
-        weights: Optional[str] = "",
+        weights: str | None = "",
     ):
         r"""
         Function invoked when calling the pipeline for generation.

examples/community/ddim_noise_comparative_analysis.py

@@ -110,7 +110,7 @@ def __call__(
         eta: float = 0.0,
         num_inference_steps: int = 50,
         use_clipped_model_output: Optional[bool] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
     ) -> Union[ImagePipelineOutput, Tuple]:
         r"""

examples/community/dps_pipeline.py

@@ -54,7 +54,7 @@ def __call__(
         batch_size: int = 1,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
         num_inference_steps: int = 1000,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         zeta: float = 0.3,
     ) -> Union[ImagePipelineOutput, Tuple]:

examples/community/edict_pipeline.py

@@ -1,5 +1,3 @@
-from typing import Optional
-
 import torch
 from PIL import Image
 from tqdm.auto import tqdm
@@ -39,7 +37,7 @@ def __init__(
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
 
     def _encode_prompt(
-        self, prompt: str, negative_prompt: Optional[str] = None, do_classifier_free_guidance: bool = False
+        self, prompt: str, negative_prompt: str | None = None, do_classifier_free_guidance: bool = False
     ):
         text_inputs = self.tokenizer(
             prompt,
@@ -141,7 +139,7 @@ def prepare_latents(
         text_embeds: torch.Tensor,
         timesteps: torch.Tensor,
         guidance_scale: float,
-        generator: Optional[torch.Generator] = None,
+        generator: torch.Generator | None = None,
     ):
         do_classifier_free_guidance = guidance_scale > 1.0
 
@@ -194,9 +192,9 @@ def __call__(
         guidance_scale: float = 3.0,
         num_inference_steps: int = 50,
         strength: float = 0.8,
-        negative_prompt: Optional[str] = None,
-        generator: Optional[torch.Generator] = None,
-        output_type: Optional[str] = "pil",
+        negative_prompt: str | None = None,
+        generator: torch.Generator | None = None,
+        output_type: str | None = "pil",
     ):
         do_classifier_free_guidance = guidance_scale > 1.0
 

examples/community/fresco_v2v.py

@@ -1208,7 +1208,7 @@ def apply_FRESCO_attn(pipe):
 
 
 def retrieve_latents(
-    encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample"
+    encoder_output: torch.Tensor, generator: torch.Generator | None = None, sample_mode: str = "sample"
 ):
     if hasattr(encoder_output, "latent_dist") and sample_mode == "sample":
         return encoder_output.latent_dist.sample(generator)
@@ -2064,7 +2064,7 @@ def __call__(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
         ip_adapter_image_embeds: Optional[List[torch.FloatTensor]] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         controlnet_conditioning_scale: Union[float, List[float]] = 0.8,

examples/community/gluegen.py

@@ -597,7 +597,7 @@ def __call__(
         latents: Optional[torch.Tensor] = None,
         prompt_embeds: Optional[torch.Tensor] = None,
         negative_prompt_embeds: Optional[torch.Tensor] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         guidance_rescale: float = 0.0,

examples/community/hd_painter.py

@@ -462,7 +462,7 @@ def __call__(
         num_inference_steps: int = 50,
         timesteps: List[int] = None,
         guidance_scale: float = 7.5,
-        positive_prompt: Optional[str] = "",
+        positive_prompt: str | None = "",
         negative_prompt: Optional[Union[str, List[str]]] = None,
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.01,
@@ -471,7 +471,7 @@ def __call__(
         prompt_embeds: Optional[torch.Tensor] = None,
         negative_prompt_embeds: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
-        output_type: Optional[str] = "pil",
+        output_type: str | None = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         clip_skip: int = None,