Release: v0.17.1

* Correct timestep inpaint

* make style

* Fix

* Apply suggestions from code review

* make style

.github/workflows/build_documentation.yml

@@ -5,6 +5,7 @@ on:
     branches:
       - main
       - doc-builder*
+      - v*-release
       - v*-patch
 
 jobs:
@@ -17,3 +18,4 @@ jobs:
       languages: en ko
     secrets:
       token: ${{ secrets.HUGGINGFACE_PUSH }}
+      hf_token: ${{ secrets.HF_DOC_BUILD_PUSH }}

.github/workflows/upload_pr_documentation.yml

@@ -0,0 +1,16 @@
+name: Upload PR Documentation
+
+on:
+  workflow_run:
+    workflows: ["Build PR Documentation"]
+    types:
+      - completed
+
+jobs:
+  build:
+    uses: huggingface/doc-builder/.github/workflows/upload_pr_documentation.yml@main
+    with:
+      package_name: diffusers
+    secrets:
+      hf_token: ${{ secrets.HF_DOC_BUILD_PUSH }}
+      comment_bot_token: ${{ secrets.COMMENT_BOT_TOKEN }}

examples/controlnet/train_controlnet.py

@@ -55,7 +55,7 @@ if is_wandb_available():
     import wandb
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__)
 

examples/controlnet/train_controlnet_flax.py

@@ -59,7 +59,7 @@ if is_wandb_available():
     import wandb
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = logging.getLogger(__name__)
 

examples/custom_diffusion/train_custom_diffusion.py

@@ -56,7 +56,7 @@ from diffusers.utils.import_utils import is_xformers_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__)
 

examples/dreambooth/train_dreambooth.py

@@ -58,7 +58,7 @@ if is_wandb_available():
     import wandb
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__)
 

examples/dreambooth/train_dreambooth_flax.py

@@ -36,7 +36,7 @@ from diffusers.utils import check_min_version
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 # Cache compiled models across invocations of this script.
 cc.initialize_cache(os.path.expanduser("~/.cache/jax/compilation_cache"))

examples/dreambooth/train_dreambooth_lora.py

@@ -64,7 +64,7 @@ from diffusers.utils.import_utils import is_xformers_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__)
 

examples/instruct_pix2pix/train_instruct_pix2pix.py

@@ -51,7 +51,7 @@ from diffusers.utils.import_utils import is_xformers_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/text_to_image/train_text_to_image.py

@@ -52,7 +52,7 @@ if is_wandb_available():
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/text_to_image/train_text_to_image_flax.py

@@ -33,7 +33,7 @@ from diffusers.utils import check_min_version
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = logging.getLogger(__name__)
 

examples/text_to_image/train_text_to_image_lora.py

@@ -47,7 +47,7 @@ from diffusers.utils.import_utils import is_xformers_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/textual_inversion/textual_inversion.py

@@ -77,7 +77,7 @@ else:
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__)
 

examples/textual_inversion/textual_inversion_flax.py

@@ -56,7 +56,7 @@ else:
 # ------------------------------------------------------------------------------
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = logging.getLogger(__name__)
 

examples/unconditional_image_generation/train_unconditional.py

@@ -28,7 +28,7 @@ from diffusers.utils.import_utils import is_xformers_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.17.0.dev0")
+check_min_version("0.17.0")
 
 logger = get_logger(__name__, log_level="INFO")
 

setup.py

@@ -227,7 +227,7 @@ install_requires = [
 
 setup(
     name="diffusers",
-    version="0.17.0.dev0",  # expected format is one of x.y.z.dev0, or x.y.z.rc1 or x.y.z (no to dashes, yes to dots)
+    version="0.17.1",  # expected format is one of x.y.z.dev0, or x.y.z.rc1 or x.y.z (no to dashes, yes to dots)
     description="Diffusers",
     long_description=open("README.md", "r", encoding="utf-8").read(),
     long_description_content_type="text/markdown",

src/diffusers/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.17.0.dev0"
+__version__ = "0.17.1"
 
 from .configuration_utils import ConfigMixin
 from .utils import (

src/diffusers/models/modeling_utils.py

@@ -17,6 +17,7 @@
 import inspect
 import itertools
 import os
+import re
 from functools import partial
 from typing import Any, Callable, List, Optional, Tuple, Union
 
@@ -162,6 +163,7 @@ class ModelMixin(torch.nn.Module):
     config_name = CONFIG_NAME
     _automatically_saved_args = ["_diffusers_version", "_class_name", "_name_or_path"]
     _supports_gradient_checkpointing = False
+    _keys_to_ignore_on_load_unexpected = None
 
     def __init__(self):
         super().__init__()
@@ -608,6 +610,7 @@ class ModelMixin(torch.nn.Module):
                             " `low_cpu_mem_usage=False` and `device_map=None` if you want to randomly initialize"
                             " those weights or else make sure your checkpoint file is correct."
                         )
+                    unexpected_keys = []
 
                     empty_state_dict = model.state_dict()
                     for param_name, param in state_dict.items():
@@ -615,6 +618,10 @@ class ModelMixin(torch.nn.Module):
                             inspect.signature(set_module_tensor_to_device).parameters.keys()
                         )
 
+                        if param_name not in empty_state_dict:
+                            unexpected_keys.append(param_name)
+                            continue
+
                         if empty_state_dict[param_name].shape != param.shape:
                             raise ValueError(
                                 f"Cannot load {pretrained_model_name_or_path} because {param_name} expected shape {empty_state_dict[param_name]}, but got {param.shape}. If you want to instead overwrite randomly initialized weights, please make sure to pass both `low_cpu_mem_usage=False` and `ignore_mismatched_sizes=True`. For more information, see also: https://github.com/huggingface/diffusers/issues/1619#issuecomment-1345604389 as an example."
@@ -626,6 +633,16 @@ class ModelMixin(torch.nn.Module):
                             )
                         else:
                             set_module_tensor_to_device(model, param_name, param_device, value=param)
+
+                    if cls._keys_to_ignore_on_load_unexpected is not None:
+                        for pat in cls._keys_to_ignore_on_load_unexpected:
+                            unexpected_keys = [k for k in unexpected_keys if re.search(pat, k) is None]
+
+                    if len(unexpected_keys) > 0:
+                        logger.warn(
+                            f"Some weights of the model checkpoint were not used when initializing {cls.__name__}: \n {[', '.join(unexpected_keys)]}"
+                        )
+
                 else:  # else let accelerate handle loading and dispatching.
                     # Load weights and dispatch according to the device_map
                     # by default the device_map is None and the weights are loaded on the CPU

src/diffusers/pipelines/controlnet/pipeline_controlnet.py

@@ -611,7 +611,7 @@ class StableDiffusionControlNetPipeline(DiffusionPipeline, TextualInversionLoade
             and not image_is_np_list
         ):
             raise TypeError(
-                "image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors"
+                f"image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors, but is {type(image)}"
             )
 
         if image_is_pil:

src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py

@@ -638,7 +638,7 @@ class StableDiffusionControlNetImg2ImgPipeline(DiffusionPipeline, TextualInversi
             and not image_is_np_list
         ):
             raise TypeError(
-                "image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors"
+                f"image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors, but is {type(image)}"
             )
 
         if image_is_pil:

src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py

@@ -770,7 +770,7 @@ class StableDiffusionControlNetInpaintPipeline(DiffusionPipeline, TextualInversi
             and not image_is_np_list
         ):
             raise TypeError(
-                "image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors"
+                f"image must be passed and be one of PIL image, numpy array, torch tensor, list of PIL images, list of numpy arrays or list of torch tensors, but is {type(image)}"
             )
 
         if image_is_pil:
@@ -1306,7 +1306,10 @@ class StableDiffusionControlNetInpaintPipeline(DiffusionPipeline, TextualInversi
                     init_mask = mask[:1]
 
                     if i < len(timesteps) - 1:
-                        init_latents_proper = self.scheduler.add_noise(init_latents_proper, noise, torch.tensor([t]))
+                        noise_timestep = timesteps[i + 1]
+                        init_latents_proper = self.scheduler.add_noise(
+                            init_latents_proper, noise, torch.tensor([noise_timestep])
+                        )
 
                     latents = (1 - init_mask) * init_latents_proper + init_mask * latents
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py

@@ -1038,7 +1038,10 @@ class StableDiffusionInpaintPipeline(DiffusionPipeline, TextualInversionLoaderMi
                     init_mask = mask[:1]
 
                     if i < len(timesteps) - 1:
-                        init_latents_proper = self.scheduler.add_noise(init_latents_proper, noise, torch.tensor([t]))
+                        noise_timestep = timesteps[i + 1]
+                        init_latents_proper = self.scheduler.add_noise(
+                            init_latents_proper, noise, torch.tensor([noise_timestep])
+                        )
 
                     latents = (1 - init_mask) * init_latents_proper + init_mask * latents
 

src/diffusers/pipelines/unidiffuser/modeling_text_decoder.py

@@ -61,6 +61,8 @@ class UniDiffuserTextDecoder(ModelMixin, ConfigMixin, ModuleUtilsMixin):
             dot-product/softmax to float() when training with mixed precision.
     """
 
+    _keys_to_ignore_on_load_unexpected = [r"h\.\d+\.attn\.bias", r"h\.\d+\.attn\.masked_bias"]
+
     @register_to_config
     def __init__(
         self,

src/diffusers/utils/dynamic_modules_utils.py

@@ -21,12 +21,12 @@ import os
 import re
 import shutil
 import sys
-from distutils.version import StrictVersion
 from pathlib import Path
 from typing import Dict, Optional, Union
 from urllib import request
 
 from huggingface_hub import HfFolder, cached_download, hf_hub_download, model_info
+from packaging import version
 
 from .. import __version__
 from . import DIFFUSERS_DYNAMIC_MODULE_NAME, HF_MODULES_CACHE, logging
@@ -43,7 +43,7 @@ logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 def get_diffusers_versions():
     url = "https://pypi.org/pypi/diffusers/json"
     releases = json.loads(request.urlopen(url).read())["releases"].keys()
-    return sorted(releases, key=StrictVersion)
+    return sorted(releases, key=lambda x: version.Version(x))
 
 
 def init_hf_modules():