safe remove old lora backend 👋

docker/diffusers-pytorch-compile-cuda/Dockerfile

@@ -26,9 +26,9 @@ ENV PATH="/opt/venv/bin:$PATH"
 # pre-install the heavy dependencies (these can later be overridden by the deps from setup.py)
 RUN python3.9 -m pip install --no-cache-dir --upgrade pip && \
     python3.9 -m pip install --no-cache-dir \
-    torch \
-    torchvision \
-    torchaudio \
+    torch==2.1.2 \
+    torchvision==0.16.2 \
+    torchaudio==2.1.2 \
     invisible_watermark && \
     python3.9 -m pip install --no-cache-dir \
     accelerate \

docker/diffusers-pytorch-cpu/Dockerfile

@@ -25,9 +25,9 @@ ENV PATH="/opt/venv/bin:$PATH"
 # pre-install the heavy dependencies (these can later be overridden by the deps from setup.py)
 RUN python3 -m pip install --no-cache-dir --upgrade pip && \
     python3 -m pip install --no-cache-dir \
-        torch \
-        torchvision \
-        torchaudio \
+        torch==2.1.2 \
+        torchvision==0.16.2 \
+        torchaudio==2.1.2 \
         invisible_watermark \
         --extra-index-url https://download.pytorch.org/whl/cpu && \
     python3 -m pip install --no-cache-dir \

docker/diffusers-pytorch-cuda/Dockerfile

@@ -25,9 +25,9 @@ ENV PATH="/opt/venv/bin:$PATH"
 # pre-install the heavy dependencies (these can later be overridden by the deps from setup.py)
 RUN python3 -m pip install --no-cache-dir --upgrade pip && \
     python3 -m pip install --no-cache-dir \
-    torch \
-    torchvision \
-    torchaudio \
+    torch==2.1.2 \
+    torchvision==0.16.2 \
+    torchaudio==2.1.2 \
     invisible_watermark && \
     python3 -m pip install --no-cache-dir \
     accelerate \

docker/diffusers-pytorch-xformers-cuda/Dockerfile

@@ -25,9 +25,9 @@ ENV PATH="/opt/venv/bin:$PATH"
 # pre-install the heavy dependencies (these can later be overridden by the deps from setup.py)
 RUN python3 -m pip install --no-cache-dir --upgrade pip && \
     python3 -m pip install --no-cache-dir \
-        torch \
-        torchvision \
-        torchaudio \
+        torch==2.1.2 \
+        torchvision==0.16.2 \
+        torchaudio==2.1.2 \
         invisible_watermark && \
     python3 -m pip install --no-cache-dir \
         accelerate \

docs/source/en/using-diffusers/custom_pipeline_overview.md

@@ -56,60 +56,6 @@ pipeline = DiffusionPipeline.from_pretrained(
 )
 ```
 
-### Load from a local file
-
-Community pipelines can also be loaded from a local file if you pass a file path instead. The path to the passed directory must contain a `pipeline.py` file that contains the pipeline class in order to successfully load it.
-
-```py
-pipeline = DiffusionPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5",
-    custom_pipeline="./path/to/pipeline_directory/",
-    clip_model=clip_model,
-    feature_extractor=feature_extractor,
-    use_safetensors=True,
-)
-```
-
-### Load from a specific version
-
-By default, community pipelines are loaded from the latest stable version of Diffusers. To load a community pipeline from another version, use the `custom_revision` parameter.
-
-<hfoptions id="version">
-<hfoption id="main">
-
-For example, to load from the `main` branch:
-
-```py
-pipeline = DiffusionPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5",
-    custom_pipeline="clip_guided_stable_diffusion",
-    custom_revision="main",
-    clip_model=clip_model,
-    feature_extractor=feature_extractor,
-    use_safetensors=True,
-)
-```
-
-</hfoption>
-<hfoption id="older version">
-
-For example, to load from a previous version of Diffusers like `v0.25.0`:
-
-```py
-pipeline = DiffusionPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5",
-    custom_pipeline="clip_guided_stable_diffusion",
-    custom_revision="v0.25.0",
-    clip_model=clip_model,
-    feature_extractor=feature_extractor,
-    use_safetensors=True,
-)
-```
-
-</hfoption>
-</hfoptions>
-
-
 For more information about community pipelines, take a look at the [Community pipelines](custom_pipeline_examples) guide for how to use them and if you're interested in adding a community pipeline check out the [How to contribute a community pipeline](contribute_pipeline) guide!
 
 ## Community components

examples/advanced_diffusion_training/train_dreambooth_lora_sd15_advanced.py

@@ -70,7 +70,7 @@ 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("0.27.0.dev0")
+check_min_version("0.25.0.dev0")
 
 logger = get_logger(__name__)
 

examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py

@@ -74,7 +74,7 @@ from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/community/marigold_depth_estimation.py

@@ -40,7 +40,7 @@ from diffusers.utils import BaseOutput, check_min_version
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 
 class MarigoldDepthOutput(BaseOutput):
@@ -170,10 +170,10 @@ class MarigoldPipeline(DiffusionPipeline):
 
         # Normalize rgb values
         rgb = np.transpose(image, (2, 0, 1))  # [H, W, rgb] -> [rgb, H, W]
-        rgb_norm = rgb / 255.0 * 2.0 - 1.0  #  [0, 255] -> [-1, 1]
+        rgb_norm = rgb / 255.0
         rgb_norm = torch.from_numpy(rgb_norm).to(self.dtype)
         rgb_norm = rgb_norm.to(device)
-        assert rgb_norm.min() >= -1.0 and rgb_norm.max() <= 1.0
+        assert rgb_norm.min() >= 0.0 and rgb_norm.max() <= 1.0
 
         # ----------------- Predicting depth -----------------
         # Batch repeated input image

examples/consistency_distillation/train_lcm_distill_lora_sd_wds.py

@@ -72,7 +72,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/consistency_distillation/train_lcm_distill_lora_sdxl.py

@@ -65,7 +65,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/consistency_distillation/train_lcm_distill_lora_sdxl_wds.py

@@ -78,7 +78,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/consistency_distillation/train_lcm_distill_sd_wds.py

@@ -71,7 +71,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/consistency_distillation/train_lcm_distill_sdxl_wds.py

@@ -77,7 +77,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/controlnet/train_controlnet.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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/controlnet/train_controlnet_flax.py

@@ -60,7 +60,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = logging.getLogger(__name__)
 

examples/controlnet/train_controlnet_sdxl.py

@@ -60,7 +60,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/custom_diffusion/train_custom_diffusion.py

@@ -63,7 +63,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/dreambooth/README.md

@@ -376,14 +376,18 @@ After training, LoRA weights can be loaded very easily into the original pipelin
 load the original pipeline:
 
 ```python
-from diffusers import DiffusionPipeline
-pipe = DiffusionPipeline.from_pretrained("base-model-name").to("cuda")
+from diffusers import DiffusionPipeline, DPMSolverMultistepScheduler
+import torch
+
+pipe = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
+pipe.scheduler = DPMSolverMultistepScheduler.from_config(pipe.scheduler.config)
+pipe.to("cuda")
 ```
 
-Next, we can load the adapter layers into the pipeline with the [`load_lora_weights` function](https://huggingface.co/docs/diffusers/main/en/using-diffusers/loading_adapters#lora).
+Next, we can load the adapter layers into the UNet with the [`load_attn_procs` function](https://huggingface.co/docs/diffusers/api/loaders#diffusers.loaders.UNet2DConditionLoadersMixin.load_attn_procs).
 
 ```python
-pipe.load_lora_weights("path-to-the-lora-checkpoint")
+pipe.unet.load_attn_procs("patrickvonplaten/lora_dreambooth_dog_example")
 ```
 
 Finally, we can run the model in inference.

examples/dreambooth/train_dreambooth.py

@@ -63,7 +63,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/dreambooth/train_dreambooth_flax.py

@@ -35,7 +35,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 # Cache compiled models across invocations of this script.
 cc.initialize_cache(os.path.expanduser("~/.cache/jax/compilation_cache"))

examples/dreambooth/train_dreambooth_lora.py

@@ -67,7 +67,7 @@ from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/dreambooth/train_dreambooth_lora_sdxl.py

@@ -68,7 +68,7 @@ from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 

examples/instruct_pix2pix/train_instruct_pix2pix.py

@@ -53,7 +53,7 @@ from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/instruct_pix2pix/train_instruct_pix2pix_sdxl.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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/kandinsky2_2/text_to_image/train_text_to_image_decoder.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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/kandinsky2_2/text_to_image/train_text_to_image_lora_decoder.py

@@ -46,7 +46,7 @@ from diffusers.utils import check_min_version, is_wandb_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/kandinsky2_2/text_to_image/train_text_to_image_lora_prior.py

@@ -46,7 +46,7 @@ from diffusers.utils import check_min_version, is_wandb_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/kandinsky2_2/text_to_image/train_text_to_image_prior.py

@@ -51,7 +51,7 @@ if is_wandb_available():
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/t2i_adapter/train_t2i_adapter_sdxl.py

@@ -49,7 +49,6 @@ from diffusers import (
 )
 from diffusers.optimization import get_scheduler
 from diffusers.utils import check_min_version, is_wandb_available
-from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card
 from diffusers.utils.import_utils import is_xformers_available
 from diffusers.utils.torch_utils import is_compiled_module
 
@@ -60,7 +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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 
@@ -196,7 +195,7 @@ def import_model_class_from_model_name_or_path(
         raise ValueError(f"{model_class} is not supported.")
 
 
-def save_model_card(repo_id: str, image_logs: dict = None, base_model: str = None, repo_folder: str = None):
+def save_model_card(repo_id: str, image_logs=None, base_model=str, repo_folder=None):
     img_str = ""
     if image_logs is not None:
         img_str = "You can find some example images below.\n"
@@ -210,25 +209,27 @@ def save_model_card(repo_id: str, image_logs: dict = None, base_model: str = Non
             image_grid(images, 1, len(images)).save(os.path.join(repo_folder, f"images_{i}.png"))
             img_str += f"![images_{i})](./images_{i}.png)\n"
 
-    model_description = f"""
+    yaml = f"""
+---
+license: creativeml-openrail-m
+base_model: {base_model}
+tags:
+- stable-diffusion-xl
+- stable-diffusion-xl-diffusers
+- text-to-image
+- diffusers
+- t2iadapter
+inference: true
+---
+    """
+    model_card = f"""
 # t2iadapter-{repo_id}
 
 These are t2iadapter weights trained on {base_model} with new type of conditioning.
 {img_str}
 """
-    model_card = load_or_create_model_card(
-        repo_id_or_path=repo_id,
-        from_training=True,
-        license="creativeml-openrail-m",
-        base_model=base_model,
-        model_description=model_description,
-        inference=True,
-    )
-
-    tags = ["stable-diffusion-xl", "stable-diffusion-xl-diffusers", "text-to-image", "diffusers", "t2iadapter"]
-    model_card = populate_model_card(model_card, tags=tags)
-
-    model_card.save(os.path.join(repo_folder, "README.md"))
+    with open(os.path.join(repo_folder, "README.md"), "w") as f:
+        f.write(yaml + model_card)
 
 
 def parse_args(input_args=None):

examples/text_to_image/train_text_to_image.py

@@ -45,7 +45,6 @@ from diffusers import AutoencoderKL, DDPMScheduler, StableDiffusionPipeline, UNe
 from diffusers.optimization import get_scheduler
 from diffusers.training_utils import EMAModel, compute_snr
 from diffusers.utils import check_min_version, deprecate, is_wandb_available, make_image_grid
-from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card
 from diffusers.utils.import_utils import is_xformers_available
 from diffusers.utils.torch_utils import is_compiled_module
 
@@ -55,7 +54,7 @@ if is_wandb_available():
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 
@@ -67,8 +66,8 @@ DATASET_NAME_MAPPING = {
 def save_model_card(
     args,
     repo_id: str,
-    images: list = None,
-    repo_folder: str = None,
+    images=None,
+    repo_folder=None,
 ):
     img_str = ""
     if len(images) > 0:
@@ -76,7 +75,21 @@ def save_model_card(
         image_grid.save(os.path.join(repo_folder, "val_imgs_grid.png"))
         img_str += "![val_imgs_grid](./val_imgs_grid.png)\n"
 
-    model_description = f"""
+    yaml = f"""
+---
+license: creativeml-openrail-m
+base_model: {args.pretrained_model_name_or_path}
+datasets:
+- {args.dataset_name}
+tags:
+- stable-diffusion
+- stable-diffusion-diffusers
+- text-to-image
+- diffusers
+inference: true
+---
+    """
+    model_card = f"""
 # Text-to-image finetuning - {repo_id}
 
 This pipeline was finetuned from **{args.pretrained_model_name_or_path}** on the **{args.dataset_name}** dataset. Below are some example images generated with the finetuned pipeline using the following prompts: {args.validation_prompts}: \n
@@ -119,21 +132,10 @@ These are the key hyperparameters used during training:
 More information on all the CLI arguments and the environment are available on your [`wandb` run page]({wandb_run_url}).
 """
 
-    model_description += wandb_info
+    model_card += wandb_info
 
-    model_card = load_or_create_model_card(
-        repo_id_or_path=repo_id,
-        from_training=True,
-        license="creativeml-openrail-m",
-        base_model=args.pretrained_model_name_or_path,
-        model_description=model_description,
-        inference=True,
-    )
-
-    tags = ["stable-diffusion", "stable-diffusion-diffusers", "text-to-image", "diffusers"]
-    model_card = populate_model_card(model_card, tags=tags)
-
-    model_card.save(os.path.join(repo_folder, "README.md"))
+    with open(os.path.join(repo_folder, "README.md"), "w") as f:
+        f.write(yaml + model_card)
 
 
 def log_validation(vae, text_encoder, tokenizer, unet, args, accelerator, weight_dtype, epoch):

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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = logging.getLogger(__name__)
 

examples/text_to_image/train_text_to_image_lora.py

@@ -45,50 +45,42 @@ from diffusers import AutoencoderKL, DDPMScheduler, DiffusionPipeline, StableDif
 from diffusers.optimization import get_scheduler
 from diffusers.training_utils import cast_training_params, compute_snr
 from diffusers.utils import check_min_version, convert_state_dict_to_diffusers, is_wandb_available
-from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card
 from diffusers.utils.import_utils import is_xformers_available
 from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 
 
-def save_model_card(
-    repo_id: str, images: list = None, base_model: str = None, dataset_name: str = None, repo_folder: str = None
-):
+def save_model_card(repo_id: str, images=None, base_model=str, dataset_name=str, repo_folder=None):
     img_str = ""
     for i, image in enumerate(images):
         image.save(os.path.join(repo_folder, f"image_{i}.png"))
         img_str += f"![img_{i}](./image_{i}.png)\n"
 
-    model_description = f"""
+    yaml = f"""
+---
+license: creativeml-openrail-m
+base_model: {base_model}
+tags:
+- stable-diffusion
+- stable-diffusion-diffusers
+- text-to-image
+- diffusers
+- lora
+inference: true
+---
+    """
+    model_card = f"""
 # LoRA text2image fine-tuning - {repo_id}
 These are LoRA adaption weights for {base_model}. The weights were fine-tuned on the {dataset_name} dataset. You can find some example images in the following. \n
 {img_str}
 """
-
-    model_card = load_or_create_model_card(
-        repo_id_or_path=repo_id,
-        from_training=True,
-        license="creativeml-openrail-m",
-        base_model=base_model,
-        model_description=model_description,
-        inference=True,
-    )
-
-    tags = [
-        "stable-diffusion",
-        "stable-diffusion-diffusers",
-        "text-to-image",
-        "diffusers",
-        "lora",
-    ]
-    model_card = populate_model_card(model_card, tags=tags)
-
-    model_card.save(os.path.join(repo_folder, "README.md"))
+    with open(os.path.join(repo_folder, "README.md"), "w") as f:
+        f.write(yaml + model_card)
 
 
 def parse_args():

examples/text_to_image/train_text_to_image_lora_sdxl.py

@@ -58,33 +58,45 @@ from diffusers.utils import (
     convert_unet_state_dict_to_peft,
     is_wandb_available,
 )
-from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card
 from diffusers.utils.import_utils import is_xformers_available
 from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 
 
 def save_model_card(
     repo_id: str,
-    images: list = None,
-    base_model: str = None,
-    dataset_name: str = None,
-    train_text_encoder: bool = False,
-    repo_folder: str = None,
-    vae_path: str = None,
+    images=None,
+    base_model=str,
+    dataset_name=str,
+    train_text_encoder=False,
+    repo_folder=None,
+    vae_path=None,
 ):
     img_str = ""
-    if images is not None:
-        for i, image in enumerate(images):
-            image.save(os.path.join(repo_folder, f"image_{i}.png"))
-            img_str += f"![img_{i}](./image_{i}.png)\n"
+    for i, image in enumerate(images):
+        image.save(os.path.join(repo_folder, f"image_{i}.png"))
+        img_str += f"![img_{i}](./image_{i}.png)\n"
 
-    model_description = f"""
+    yaml = f"""
+---
+license: creativeml-openrail-m
+base_model: {base_model}
+dataset: {dataset_name}
+tags:
+- stable-diffusion-xl
+- stable-diffusion-xl-diffusers
+- text-to-image
+- diffusers
+- lora
+inference: true
+---
+    """
+    model_card = f"""
 # LoRA text2image fine-tuning - {repo_id}
 
 These are LoRA adaption weights for {base_model}. The weights were fine-tuned on the {dataset_name} dataset. You can find some example images in the following. \n
@@ -94,19 +106,8 @@ LoRA for the text encoder was enabled: {train_text_encoder}.
 
 Special VAE used for training: {vae_path}.
 """
-    model_card = load_or_create_model_card(
-        repo_id_or_path=repo_id,
-        from_training=True,
-        license="creativeml-openrail-m",
-        base_model=base_model,
-        model_description=model_description,
-        inference=True,
-    )
-
-    tags = ["stable-diffusion-xl", "stable-diffusion-xl-diffusers", "text-to-image", "diffusers", "lora"]
-    model_card = populate_model_card(model_card, tags=tags)
-
-    model_card.save(os.path.join(repo_folder, "README.md"))
+    with open(os.path.join(repo_folder, "README.md"), "w") as f:
+        f.write(yaml + model_card)
 
 
 def import_model_class_from_model_name_or_path(

examples/text_to_image/train_text_to_image_sdxl.py

@@ -48,13 +48,12 @@ from diffusers import AutoencoderKL, DDPMScheduler, StableDiffusionXLPipeline, U
 from diffusers.optimization import get_scheduler
 from diffusers.training_utils import EMAModel, compute_snr
 from diffusers.utils import check_min_version, is_wandb_available
-from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card
 from diffusers.utils.import_utils import is_xformers_available
 from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__)
 
@@ -66,45 +65,41 @@ DATASET_NAME_MAPPING = {
 
 def save_model_card(
     repo_id: str,
-    images: list = None,
-    validation_prompt: str = None,
-    base_model: str = None,
-    dataset_name: str = None,
-    repo_folder: str = None,
-    vae_path: str = None,
+    images=None,
+    validation_prompt=None,
+    base_model=str,
+    dataset_name=str,
+    repo_folder=None,
+    vae_path=None,
 ):
     img_str = ""
     for i, image in enumerate(images):
         image.save(os.path.join(repo_folder, f"image_{i}.png"))
         img_str += f"![img_{i}](./image_{i}.png)\n"
 
-    model_description = f"""
+    yaml = f"""
+---
+license: creativeml-openrail-m
+base_model: {base_model}
+dataset: {dataset_name}
+tags:
+- stable-diffusion-xl
+- stable-diffusion-xl-diffusers
+- text-to-image
+- diffusers
+inference: true
+---
+    """
+    model_card = f"""
 # Text-to-image finetuning - {repo_id}
 
-This pipeline was finetuned from **{base_model}** on the **{dataset_name}** dataset. Below are some example images generated with the finetuned pipeline using the following prompt: {validation_prompt}: \n
+This pipeline was finetuned from **{base_model}** on the **{args.dataset_name}** dataset. Below are some example images generated with the finetuned pipeline using the following prompt: {validation_prompt}: \n
 {img_str}
 
 Special VAE used for training: {vae_path}.
 """
-
-    model_card = load_or_create_model_card(
-        repo_id_or_path=repo_id,
-        from_training=True,
-        license="creativeml-openrail-m",
-        base_model=base_model,
-        model_description=model_description,
-        inference=True,
-    )
-
-    tags = [
-        "stable-diffusion-xl",
-        "stable-diffusion-xl-diffusers",
-        "text-to-image",
-        "diffusers",
-    ]
-    model_card = populate_model_card(model_card, tags=tags)
-
-    model_card.save(os.path.join(repo_folder, "README.md"))
+    with open(os.path.join(repo_folder, "README.md"), "w") as f:
+        f.write(yaml + model_card)
 
 
 def import_model_class_from_model_name_or_path(

examples/textual_inversion/textual_inversion.py

@@ -79,7 +79,7 @@ else:
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = logging.getLogger(__name__)
 

examples/textual_inversion/textual_inversion_sdxl.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("0.27.0.dev0")
+check_min_version("0.25.0.dev0")
 
 logger = get_logger(__name__)
 

examples/unconditional_image_generation/train_unconditional.py

@@ -29,7 +29,7 @@ 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("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/wuerstchen/text_to_image/train_text_to_image_lora_prior.py

@@ -50,7 +50,7 @@ if is_wandb_available():
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

examples/wuerstchen/text_to_image/train_text_to_image_prior.py

@@ -51,7 +51,7 @@ if is_wandb_available():
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.27.0.dev0")
+check_min_version("0.26.0.dev0")
 
 logger = get_logger(__name__, log_level="INFO")
 

scripts/convert_diffusers_to_original_sdxl.py

@@ -167,10 +167,7 @@ vae_conversion_map_attn = [
 
 def reshape_weight_for_sd(w):
     # convert HF linear weights to SD conv2d weights
-    if not w.ndim == 1:
-        return w.reshape(*w.shape, 1, 1)
-    else:
-        return w
+    return w.reshape(*w.shape, 1, 1)
 
 
 def convert_vae_state_dict(vae_state_dict):
@@ -324,18 +321,11 @@ if __name__ == "__main__":
     vae_state_dict = convert_vae_state_dict(vae_state_dict)
     vae_state_dict = {"first_stage_model." + k: v for k, v in vae_state_dict.items()}
 
-    # Convert text encoder 1
     text_enc_dict = convert_openai_text_enc_state_dict(text_enc_dict)
     text_enc_dict = {"conditioner.embedders.0.transformer." + k: v for k, v in text_enc_dict.items()}
 
-    # Convert text encoder 2
     text_enc_2_dict = convert_openclip_text_enc_state_dict(text_enc_2_dict)
     text_enc_2_dict = {"conditioner.embedders.1.model." + k: v for k, v in text_enc_2_dict.items()}
-    # We call the `.T.contiguous()` to match what's done in
-    # https://github.com/huggingface/diffusers/blob/84905ca7287876b925b6bf8e9bb92fec21c78764/src/diffusers/loaders/single_file_utils.py#L1085
-    text_enc_2_dict["conditioner.embedders.1.model.text_projection"] = text_enc_2_dict.pop(
-        "conditioner.embedders.1.model.text_projection.weight"
-    ).T.contiguous()
 
     # Put together new checkpoint
     state_dict = {**unet_state_dict, **vae_state_dict, **text_enc_dict, **text_enc_2_dict}

scripts/convert_diffusers_to_original_stable_diffusion.py

@@ -170,10 +170,7 @@ vae_extra_conversion_map = [
 
 def reshape_weight_for_sd(w):
     # convert HF linear weights to SD conv2d weights
-    if not w.ndim == 1:
-        return w.reshape(*w.shape, 1, 1)
-    else:
-        return w
+    return w.reshape(*w.shape, 1, 1)
 
 
 def convert_vae_state_dict(vae_state_dict):

setup.py

@@ -126,8 +126,8 @@ _deps = [
     "regex!=2019.12.17",
     "requests",
     "tensorboard",
-    "torch>=1.4",
-    "torchvision",
+    "torch>=1.4,<2.2.0",
+    "torchvision<0.17",
     "transformers>=4.25.1",
     "urllib3<=2.0.0",
 ]
@@ -249,7 +249,7 @@ version_range_max = max(sys.version_info[1], 10) + 1
 
 setup(
     name="diffusers",
-    version="0.27.0.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="0.26.0.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)
     description="State-of-the-art diffusion in PyTorch and JAX.",
     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__ = "0.27.0.dev0"
+__version__ = "0.26.0.dev0"
 
 from typing import TYPE_CHECKING
 

src/diffusers/dependency_versions_table.py

@@ -38,8 +38,8 @@ deps = {
     "regex": "regex!=2019.12.17",
     "requests": "requests",
     "tensorboard": "tensorboard",
-    "torch": "torch>=1.4",
-    "torchvision": "torchvision",
+    "torch": "torch>=1.4,<2.2.0",
+    "torchvision": "torchvision<0.17",
     "transformers": "transformers>=4.25.1",
     "urllib3": "urllib3<=2.0.0",
 }

src/diffusers/loaders/ip_adapter.py

@@ -166,7 +166,8 @@ class IPAdapterMixin:
                         pretrained_model_name_or_path_or_dict,
                         subfolder=Path(subfolder, "image_encoder").as_posix(),
                     ).to(self.device, dtype=self.dtype)
-                    self.register_modules(image_encoder=image_encoder)
+                    self.image_encoder = image_encoder
+                    self.register_to_config(image_encoder=["transformers", "CLIPVisionModelWithProjection"])
                 else:
                     raise ValueError("`image_encoder` cannot be None when using IP Adapters.")
 

src/diffusers/loaders/single_file_utils.py

@@ -1112,6 +1112,7 @@ def create_text_encoder_from_open_clip_checkpoint(
             text_model_dict[diffusers_key + ".q_proj.bias"] = weight_value[:text_proj_dim]
             text_model_dict[diffusers_key + ".k_proj.bias"] = weight_value[text_proj_dim : text_proj_dim * 2]
             text_model_dict[diffusers_key + ".v_proj.bias"] = weight_value[text_proj_dim * 2 :]
+
         else:
             text_model_dict[diffusers_key] = checkpoint[key]
 

src/diffusers/models/attention_processor.py

@@ -11,7 +11,6 @@
 # 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 inspect
 from importlib import import_module
 from typing import Callable, Optional, Union
 
@@ -510,15 +509,6 @@ class Attention(nn.Module):
         # The `Attention` class can call different attention processors / attention functions
         # here we simply pass along all tensors to the selected processor class
         # For standard processors that are defined here, `**cross_attention_kwargs` is empty
-
-        attn_parameters = set(inspect.signature(self.processor.__call__).parameters.keys())
-        unused_kwargs = [k for k, _ in cross_attention_kwargs.items() if k not in attn_parameters]
-        if len(unused_kwargs) > 0:
-            logger.warning(
-                f"cross_attention_kwargs {unused_kwargs} are not expected by {self.processor.__class__.__name__} and will be ignored."
-            )
-        cross_attention_kwargs = {k: w for k, w in cross_attention_kwargs.items() if k in attn_parameters}
-
         return self.processor(
             self,
             hidden_states,

src/diffusers/pipelines/auto_pipeline.py

@@ -13,6 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import inspect
 from collections import OrderedDict
 
 from huggingface_hub.utils import validate_hf_hub_args
@@ -163,6 +164,14 @@ def _get_task_class(mapping, pipeline_class_name, throw_error_if_not_exist: bool
         raise ValueError(f"AutoPipeline can't find a pipeline linked to {pipeline_class_name} for {model_name}")
 
 
+def _get_signature_keys(obj):
+    parameters = inspect.signature(obj.__init__).parameters
+    required_parameters = {k: v for k, v in parameters.items() if v.default == inspect._empty}
+    optional_parameters = set({k for k, v in parameters.items() if v.default != inspect._empty})
+    expected_modules = set(required_parameters.keys()) - {"self"}
+    return expected_modules, optional_parameters
+
+
 class AutoPipelineForText2Image(ConfigMixin):
     r"""
 
@@ -382,7 +391,7 @@ class AutoPipelineForText2Image(ConfigMixin):
                 )
 
         # define expected module and optional kwargs given the pipeline signature
-        expected_modules, optional_kwargs = text_2_image_cls._get_signature_keys(text_2_image_cls)
+        expected_modules, optional_kwargs = _get_signature_keys(text_2_image_cls)
 
         pretrained_model_name_or_path = original_config.pop("_name_or_path", None)
 
@@ -659,7 +668,7 @@ class AutoPipelineForImage2Image(ConfigMixin):
                 )
 
         # define expected module and optional kwargs given the pipeline signature
-        expected_modules, optional_kwargs = image_2_image_cls._get_signature_keys(image_2_image_cls)
+        expected_modules, optional_kwargs = _get_signature_keys(image_2_image_cls)
 
         pretrained_model_name_or_path = original_config.pop("_name_or_path", None)
 
@@ -934,7 +943,7 @@ class AutoPipelineForInpainting(ConfigMixin):
                 )
 
         # define expected module and optional kwargs given the pipeline signature
-        expected_modules, optional_kwargs = inpainting_cls._get_signature_keys(inpainting_cls)
+        expected_modules, optional_kwargs = _get_signature_keys(inpainting_cls)
 
         pretrained_model_name_or_path = original_config.pop("_name_or_path", None)
 

src/diffusers/pipelines/pipeline_utils.py

@@ -981,9 +981,10 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
             revision (`str`, *optional*, defaults to `"main"`):
                 The specific model version to use. It can be a branch name, a tag name, a commit id, or any identifier
                 allowed by Git.
-            custom_revision (`str`, *optional*):
+            custom_revision (`str`, *optional*, defaults to `"main"`):
                 The specific model version to use. It can be a branch name, a tag name, or a commit id similar to
-                `revision` when loading a custom pipeline from the Hub. Defaults to the latest stable 🤗 Diffusers version.
+                `revision` when loading a custom pipeline from the Hub. It can be a 🤗 Diffusers version when loading a
+                custom pipeline from GitHub, otherwise it defaults to `"main"` when loading from the Hub.
             mirror (`str`, *optional*):
                 Mirror source to resolve accessibility issues if you’re downloading a model in China. We do not
                 guarantee the timeliness or safety of the source, and you should refer to the mirror site for more
@@ -1422,7 +1423,6 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
 
         device_type = torch_device.type
         device = torch.device(f"{device_type}:{self._offload_gpu_id}")
-        self._offload_device = device
 
         if self.device.type != "cpu":
             self.to("cpu", silence_dtype_warnings=True)
@@ -1472,7 +1472,7 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
             hook.remove()
 
         # make sure the model is in the same state as before calling it
-        self.enable_model_cpu_offload(device=getattr(self, "_offload_device", "cuda"))
+        self.enable_model_cpu_offload()
 
     def enable_sequential_cpu_offload(self, gpu_id: Optional[int] = None, device: Union[torch.device, str] = "cuda"):
         r"""
@@ -1508,7 +1508,6 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
 
         device_type = torch_device.type
         device = torch.device(f"{device_type}:{self._offload_gpu_id}")
-        self._offload_device = device
 
         if self.device.type != "cpu":
             self.to("cpu", silence_dtype_warnings=True)

tests/pipelines/pixart/test_pixart.py

@@ -27,13 +27,7 @@ from diffusers import (
     PixArtAlphaPipeline,
     Transformer2DModel,
 )
-from diffusers.utils.testing_utils import (
-    enable_full_determinism,
-    numpy_cosine_similarity_distance,
-    require_torch_gpu,
-    slow,
-    torch_device,
-)
+from diffusers.utils.testing_utils import enable_full_determinism, 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
@@ -338,35 +332,37 @@ class PixArtAlphaPipelineIntegrationTests(unittest.TestCase):
         torch.cuda.empty_cache()
 
     def test_pixart_1024(self):
-        generator = torch.Generator("cpu").manual_seed(0)
+        generator = torch.manual_seed(0)
 
         pipe = PixArtAlphaPipeline.from_pretrained(self.ckpt_id_1024, torch_dtype=torch.float16)
         pipe.enable_model_cpu_offload()
         prompt = self.prompt
 
-        image = pipe(prompt, generator=generator, num_inference_steps=2, output_type="np").images
+        image = pipe(prompt, generator=generator, output_type="np").images
 
         image_slice = image[0, -3:, -3:, -1]
-        expected_slice = np.array([0.0742, 0.0835, 0.2114, 0.0295, 0.0784, 0.2361, 0.1738, 0.2251, 0.3589])
 
-        max_diff = numpy_cosine_similarity_distance(image_slice.flatten(), expected_slice)
-        self.assertLessEqual(max_diff, 1e-4)
+        expected_slice = np.array([0.1941, 0.2117, 0.2188, 0.1946, 0.218, 0.2124, 0.199, 0.2437, 0.2583])
+
+        max_diff = np.abs(image_slice.flatten() - expected_slice).max()
+        self.assertLessEqual(max_diff, 1e-3)
 
     def test_pixart_512(self):
-        generator = torch.Generator("cpu").manual_seed(0)
+        generator = torch.manual_seed(0)
 
         pipe = PixArtAlphaPipeline.from_pretrained(self.ckpt_id_512, torch_dtype=torch.float16)
         pipe.enable_model_cpu_offload()
 
         prompt = self.prompt
 
-        image = pipe(prompt, generator=generator, num_inference_steps=2, output_type="np").images
+        image = pipe(prompt, generator=generator, output_type="np").images
 
         image_slice = image[0, -3:, -3:, -1]
-        expected_slice = np.array([0.3477, 0.3882, 0.4541, 0.3413, 0.3821, 0.4463, 0.4001, 0.4409, 0.4958])
 
-        max_diff = numpy_cosine_similarity_distance(image_slice.flatten(), expected_slice)
-        self.assertLessEqual(max_diff, 1e-4)
+        expected_slice = np.array([0.2637, 0.291, 0.2939, 0.207, 0.2512, 0.2783, 0.2168, 0.2324, 0.2817])
+
+        max_diff = np.abs(image_slice.flatten() - expected_slice).max()
+        self.assertLessEqual(max_diff, 1e-3)
 
     def test_pixart_1024_without_resolution_binning(self):
         generator = torch.manual_seed(0)
@@ -376,7 +372,7 @@ class PixArtAlphaPipelineIntegrationTests(unittest.TestCase):
 
         prompt = self.prompt
         height, width = 1024, 768
-        num_inference_steps = 2
+        num_inference_steps = 10
 
         image = pipe(
             prompt,
@@ -410,7 +406,7 @@ class PixArtAlphaPipelineIntegrationTests(unittest.TestCase):
 
         prompt = self.prompt
         height, width = 512, 768
-        num_inference_steps = 2
+        num_inference_steps = 10
 
         image = pipe(
             prompt,

tests/pipelines/test_pipelines_auto.py

@@ -21,7 +21,6 @@ from collections import OrderedDict
 from pathlib import Path
 
 import torch
-from transformers import CLIPVisionConfig, CLIPVisionModelWithProjection
 
 from diffusers import (
     AutoPipelineForImage2Image,
@@ -49,20 +48,6 @@ PRETRAINED_MODEL_REPO_MAPPING = OrderedDict(
 
 
 class AutoPipelineFastTest(unittest.TestCase):
-    @property
-    def dummy_image_encoder(self):
-        torch.manual_seed(0)
-        config = CLIPVisionConfig(
-            hidden_size=1,
-            projection_dim=1,
-            num_hidden_layers=1,
-            num_attention_heads=1,
-            image_size=1,
-            intermediate_size=1,
-            patch_size=1,
-        )
-        return CLIPVisionModelWithProjection(config)
-
     def test_from_pipe_consistent(self):
         pipe = AutoPipelineForText2Image.from_pretrained(
             "hf-internal-testing/tiny-stable-diffusion-pipe", requires_safety_checker=False
@@ -219,20 +204,6 @@ class AutoPipelineFastTest(unittest.TestCase):
         assert pipe_control_img2img.__class__.__name__ == "StableDiffusionControlNetImg2ImgPipeline"
         assert "controlnet" in pipe_control_img2img.components
 
-    def test_from_pipe_optional_components(self):
-        image_encoder = self.dummy_image_encoder
-
-        pipe = AutoPipelineForText2Image.from_pretrained(
-            "hf-internal-testing/tiny-stable-diffusion-pipe",
-            image_encoder=image_encoder,
-        )
-
-        pipe = AutoPipelineForImage2Image.from_pipe(pipe)
-        assert pipe.image_encoder is not None
-
-        pipe = AutoPipelineForText2Image.from_pipe(pipe, image_encoder=None)
-        assert pipe.image_encoder is None
-
 
 @slow
 class AutoPipelineIntegrationTest(unittest.TestCase):

tests/schedulers/test_schedulers.py

@@ -36,10 +36,10 @@ from diffusers import (
     LMSDiscreteScheduler,
     UniPCMultistepScheduler,
     VQDiffusionScheduler,
+    logging,
 )
 from diffusers.configuration_utils import ConfigMixin, register_to_config
 from diffusers.schedulers.scheduling_utils import SchedulerMixin
-from diffusers.utils import logging
 from diffusers.utils.testing_utils import CaptureLogger, torch_device
 
 from ..others.test_utils import TOKEN, USER, is_staging_test
@@ -48,9 +48,6 @@ from ..others.test_utils import TOKEN, USER, is_staging_test
 torch.backends.cuda.matmul.allow_tf32 = False
 
 
-logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
-
-
 class SchedulerObject(SchedulerMixin, ConfigMixin):
     config_name = "config.json"
 
@@ -256,60 +253,6 @@ class SchedulerCommonTest(unittest.TestCase):
     scheduler_classes = ()
     forward_default_kwargs = ()
 
-    @property
-    def default_num_inference_steps(self):
-        return 50
-
-    @property
-    def default_timestep(self):
-        kwargs = dict(self.forward_default_kwargs)
-        num_inference_steps = kwargs.get("num_inference_steps", self.default_num_inference_steps)
-
-        try:
-            scheduler_config = self.get_scheduler_config()
-            scheduler = self.scheduler_classes[0](**scheduler_config)
-
-            scheduler.set_timesteps(num_inference_steps)
-            timestep = scheduler.timesteps[0]
-        except NotImplementedError:
-            logger.warning(
-                f"The scheduler {self.__class__.__name__} does not implement a `get_scheduler_config` method."
-                f" `default_timestep` will be set to the default value of 1."
-            )
-            timestep = 1
-
-        return timestep
-
-    # NOTE: currently taking the convention that default_timestep > default_timestep_2 (alternatively,
-    # default_timestep comes earlier in the timestep schedule than default_timestep_2)
-    @property
-    def default_timestep_2(self):
-        kwargs = dict(self.forward_default_kwargs)
-        num_inference_steps = kwargs.get("num_inference_steps", self.default_num_inference_steps)
-
-        try:
-            scheduler_config = self.get_scheduler_config()
-            scheduler = self.scheduler_classes[0](**scheduler_config)
-
-            scheduler.set_timesteps(num_inference_steps)
-            if len(scheduler.timesteps) >= 2:
-                timestep_2 = scheduler.timesteps[1]
-            else:
-                logger.warning(
-                    f"Using num_inference_steps from the scheduler testing class's default config leads to a timestep"
-                    f" scheduler of length {len(scheduler.timesteps)} < 2. The default `default_timestep_2` value of 0"
-                    f" will be used."
-                )
-                timestep_2 = 0
-        except NotImplementedError:
-            logger.warning(
-                f"The scheduler {self.__class__.__name__} does not implement a `get_scheduler_config` method."
-                f" `default_timestep_2` will be set to the default value of 0."
-            )
-            timestep_2 = 0
-
-        return timestep_2
-
     @property
     def dummy_sample(self):
         batch_size = 4
@@ -370,7 +313,6 @@ class SchedulerCommonTest(unittest.TestCase):
         kwargs = dict(self.forward_default_kwargs)
 
         num_inference_steps = kwargs.pop("num_inference_steps", None)
-        time_step = time_step if time_step is not None else self.default_timestep
 
         for scheduler_class in self.scheduler_classes:
             # TODO(Suraj) - delete the following two lines once DDPM, DDIM, and PNDM have timesteps casted to float by default
@@ -429,7 +371,6 @@ class SchedulerCommonTest(unittest.TestCase):
         kwargs.update(forward_kwargs)
 
         num_inference_steps = kwargs.pop("num_inference_steps", None)
-        time_step = time_step if time_step is not None else self.default_timestep
 
         for scheduler_class in self.scheduler_classes:
             if scheduler_class in (EulerAncestralDiscreteScheduler, EulerDiscreteScheduler, LMSDiscreteScheduler):
@@ -470,10 +411,10 @@ class SchedulerCommonTest(unittest.TestCase):
     def test_from_save_pretrained(self):
         kwargs = dict(self.forward_default_kwargs)
 
-        num_inference_steps = kwargs.pop("num_inference_steps", self.default_num_inference_steps)
+        num_inference_steps = kwargs.pop("num_inference_steps", None)
 
         for scheduler_class in self.scheduler_classes:
-            timestep = self.default_timestep
+            timestep = 1
             if scheduler_class in (EulerAncestralDiscreteScheduler, EulerDiscreteScheduler, LMSDiscreteScheduler):
                 timestep = float(timestep)
 
@@ -556,10 +497,10 @@ class SchedulerCommonTest(unittest.TestCase):
     def test_step_shape(self):
         kwargs = dict(self.forward_default_kwargs)
 
-        num_inference_steps = kwargs.pop("num_inference_steps", self.default_num_inference_steps)
+        num_inference_steps = kwargs.pop("num_inference_steps", None)
 
-        timestep_0 = self.default_timestep
-        timestep_1 = self.default_timestep_2
+        timestep_0 = 1
+        timestep_1 = 0
 
         for scheduler_class in self.scheduler_classes:
             if scheduler_class in (EulerAncestralDiscreteScheduler, EulerDiscreteScheduler, LMSDiscreteScheduler):
@@ -617,9 +558,9 @@ class SchedulerCommonTest(unittest.TestCase):
                 )
 
         kwargs = dict(self.forward_default_kwargs)
-        num_inference_steps = kwargs.pop("num_inference_steps", self.default_num_inference_steps)
+        num_inference_steps = kwargs.pop("num_inference_steps", 50)
 
-        timestep = self.default_timestep
+        timestep = 0
         if len(self.scheduler_classes) > 0 and self.scheduler_classes[0] == IPNDMScheduler:
             timestep = 1
 
@@ -703,7 +644,7 @@ class SchedulerCommonTest(unittest.TestCase):
                 continue
             scheduler_config = self.get_scheduler_config()
             scheduler = scheduler_class(**scheduler_config)
-            scheduler.set_timesteps(self.default_num_inference_steps)
+            scheduler.set_timesteps(100)
 
             sample = self.dummy_sample.to(torch_device)
             if scheduler_class == CMStochasticIterativeScheduler: