replace with actual fix.

.github/workflows/benchmark.yml

@@ -7,7 +7,6 @@ on:
 
 env:
   DIFFUSERS_IS_CI: yes
-  HF_HUB_ENABLE_HF_TRANSFER: 1
   HF_HOME: /mnt/cache
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8

.github/workflows/pr_tests.yml

@@ -22,7 +22,6 @@ concurrency:
 
 env:
   DIFFUSERS_IS_CI: yes
-  HF_HUB_ENABLE_HF_TRANSFER: 1
   OMP_NUM_THREADS: 4
   MKL_NUM_THREADS: 4
   PYTEST_TIMEOUT: 60

.github/workflows/push_tests.yml

@@ -14,7 +14,6 @@ env:
   DIFFUSERS_IS_CI: yes
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
-  HF_HUB_ENABLE_HF_TRANSFER: 1
   PYTEST_TIMEOUT: 600
   PIPELINE_USAGE_CUTOFF: 50000
 

.github/workflows/push_tests_fast.yml

@@ -18,7 +18,6 @@ env:
   HF_HOME: /mnt/cache
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
-  HF_HUB_ENABLE_HF_TRANSFER: 1
   PYTEST_TIMEOUT: 600
   RUN_SLOW: no
 

.github/workflows/push_tests_mps.yml

@@ -13,7 +13,6 @@ env:
   HF_HOME: /mnt/cache
   OMP_NUM_THREADS: 8
   MKL_NUM_THREADS: 8
-  HF_HUB_ENABLE_HF_TRANSFER: 1
   PYTEST_TIMEOUT: 600
   RUN_SLOW: no
 

docker/diffusers-flax-cpu/Dockerfile

@@ -43,7 +43,6 @@ RUN python3 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
         numpy==1.26.4 \
         scipy \
         tensorboard \
-        transformers \
-        hf_transfer
+        transformers
 
 CMD ["/bin/bash"]

docker/diffusers-flax-tpu/Dockerfile

@@ -45,7 +45,6 @@ RUN python3 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
         numpy==1.26.4 \
         scipy \
         tensorboard \
-        transformers \
-        hf_transfer
+        transformers
 
 CMD ["/bin/bash"]

docker/diffusers-onnxruntime-cpu/Dockerfile

@@ -43,7 +43,6 @@ RUN python3 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
         numpy==1.26.4 \
         scipy \
         tensorboard \
-        transformers \
-        hf_transfer
+        transformers
 
 CMD ["/bin/bash"]

docker/diffusers-onnxruntime-cuda/Dockerfile

@@ -44,7 +44,6 @@ RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
         numpy==1.26.4 \
         scipy \
         tensorboard \
-        transformers \
-        hf_transfer
+        transformers
 
 CMD ["/bin/bash"]

docker/diffusers-pytorch-compile-cuda/Dockerfile

@@ -44,7 +44,6 @@ RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
     numpy==1.26.4 \
     scipy \
     tensorboard \
-    transformers \
-    hf_transfer
+    transformers
 
 CMD ["/bin/bash"]

docker/diffusers-pytorch-cpu/Dockerfile

@@ -44,7 +44,6 @@ RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
         numpy==1.26.4 \
         scipy \
         tensorboard \
-        transformers matplotlib  \
-        hf_transfer
+        transformers matplotlib
 
 CMD ["/bin/bash"]

docker/diffusers-pytorch-cuda/Dockerfile

@@ -45,7 +45,6 @@ RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
     scipy \
     tensorboard \
     transformers \
-    pytorch-lightning  \
-    hf_transfer
+    pytorch-lightning
 
 CMD ["/bin/bash"]

docker/diffusers-pytorch-xformers-cuda/Dockerfile

@@ -45,7 +45,6 @@ RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
         scipy \
         tensorboard \
         transformers \
-        xformers  \
-        hf_transfer
+        xformers
 
 CMD ["/bin/bash"]

docs/source/en/_toctree.yml

@@ -56,7 +56,7 @@
   - local: using-diffusers/overview_techniques
     title: Overview
   - local: training/distributed_inference
-    title: Distributed inference
+    title: Distributed inference with multiple GPUs
   - local: using-diffusers/merge_loras
     title: Merge LoRAs
   - local: using-diffusers/scheduler_features

docs/source/en/api/pipelines/text_to_video_zero.md

@@ -40,7 +40,6 @@ To generate a video from prompt, run the following Python code:
 ```python
 import torch
 from diffusers import TextToVideoZeroPipeline
-import imageio
 
 model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
 pipe = TextToVideoZeroPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")

docs/source/en/api/schedulers/overview.md

@@ -45,14 +45,6 @@ Many schedulers are implemented from the [k-diffusion](https://github.com/crowso
 | N/A                 | [`DEISMultistepScheduler`]          |                                                                                                               |
 | N/A                 | [`UniPCMultistepScheduler`]         |                                                                                                               |
 
-## Noise schedules and schedule types
-| A1111/k-diffusion        | 🤗 Diffusers                                                               |
-|--------------------------|----------------------------------------------------------------------------|
-| Karras                   | init with `use_karras_sigmas=True`                                         |
-| sgm_uniform              | init with `timestep_spacing="trailing"`                                    |
-| simple                   | init with `timestep_spacing="trailing"`                                    |
-| exponential              | init with `timestep_spacing="linspace"`, `use_exponential_sigmas=True`     |
-
 All schedulers are built from the base [`SchedulerMixin`] class which implements low level utilities shared by all schedulers.
 
 ## SchedulerMixin

docs/source/en/community_projects.md

@@ -75,8 +75,4 @@ Happy exploring, and thank you for being part of the Diffusers community!
     <td><a href="https://github.com/cumulo-autumn/StreamDiffusion"> StreamDiffusion </a></td>
     <td>A Pipeline-Level Solution for Real-Time Interactive Generation</td>
   </tr>
-  <tr style="border-top: 2px solid black">
-    <td><a href="https://github.com/Netwrck/stable-diffusion-server"> Stable Diffusion Server </a></td>
-    <td>A server configured for Inpainting/Generation/img2img with one stable diffusion model</td>
-  </tr>
 </table>

docs/source/en/optimization/coreml.md

@@ -95,7 +95,7 @@ print(f"Model downloaded at {model_path}")
 Once you have downloaded a snapshot of the model, you can test it using Apple's Python script.
 
 ```shell
-python -m python_coreml_stable_diffusion.pipeline --prompt "a photo of an astronaut riding a horse on mars" -i ./models/coreml-stable-diffusion-v1-4_original_packages/original/packages -o </path/to/output/image> --compute-unit CPU_AND_GPU --seed 93
+python -m python_coreml_stable_diffusion.pipeline --prompt "a photo of an astronaut riding a horse on mars" -i models/coreml-stable-diffusion-v1-4_original_packages -o </path/to/output/image> --compute-unit CPU_AND_GPU --seed 93
 ```
 
 Pass the path of the downloaded checkpoint with `-i` flag to the script. `--compute-unit` indicates the hardware you want to allow for inference. It must be one of the following options: `ALL`, `CPU_AND_GPU`, `CPU_ONLY`, `CPU_AND_NE`. You may also provide an optional output path, and a seed for reproducibility.

docs/source/en/training/distributed_inference.md

@@ -10,7 +10,7 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# Distributed inference
+# Distributed inference with multiple GPUs
 
 On distributed setups, you can run inference across multiple GPUs with 🤗 [Accelerate](https://huggingface.co/docs/accelerate/index) or [PyTorch Distributed](https://pytorch.org/tutorials/beginner/dist_overview.html), which is useful for generating with multiple prompts in parallel.
 
@@ -109,131 +109,3 @@ torchrun run_distributed.py --nproc_per_node=2
 
 > [!TIP]
 > You can use `device_map` within a [`DiffusionPipeline`] to distribute its model-level components on multiple devices. Refer to the [Device placement](../tutorials/inference_with_big_models#device-placement) guide to learn more.
-
-## Model sharding
-
-Modern diffusion systems such as [Flux](../api/pipelines/flux) are very large and have multiple models. For example, [Flux.1-Dev](https://hf.co/black-forest-labs/FLUX.1-dev) is made up of two text encoders - [T5-XXL](https://hf.co/google/t5-v1_1-xxl) and [CLIP-L](https://hf.co/openai/clip-vit-large-patch14) - a [diffusion transformer](../api/models/flux_transformer), and a [VAE](../api/models/autoencoderkl). With a model this size, it can be challenging to run inference on consumer GPUs.
-
-Model sharding is a technique that distributes models across GPUs when the models don't fit on a single GPU. The example below assumes two 16GB GPUs are available for inference.
-
-Start by computing the text embeddings with the text encoders. Keep the text encoders on two GPUs by setting `device_map="balanced"`. The `balanced` strategy evenly distributes the model on all available GPUs. Use the `max_memory` parameter to allocate the maximum amount of memory for each text encoder on each GPU.
-
-> [!TIP]
-> **Only** load the text encoders for this step! The diffusion transformer and VAE are loaded in a later step to preserve memory.
-
-```py
-from diffusers import FluxPipeline
-import torch
-
-prompt = "a photo of a dog with cat-like look"
-
-pipeline = FluxPipeline.from_pretrained(
-    "black-forest-labs/FLUX.1-dev",
-    transformer=None,
-    vae=None,
-    device_map="balanced",
-    max_memory={0: "16GB", 1: "16GB"},
-    torch_dtype=torch.bfloat16
-)
-with torch.no_grad():
-    print("Encoding prompts.")
-    prompt_embeds, pooled_prompt_embeds, text_ids = pipeline.encode_prompt(
-        prompt=prompt, prompt_2=None, max_sequence_length=512
-    )
-```
-
-Once the text embeddings are computed, remove them from the GPU to make space for the diffusion transformer.
-
-```py
-import gc 
-
-def flush():
-    gc.collect()
-    torch.cuda.empty_cache()
-    torch.cuda.reset_max_memory_allocated()
-    torch.cuda.reset_peak_memory_stats()
-
-del pipeline.text_encoder
-del pipeline.text_encoder_2
-del pipeline.tokenizer
-del pipeline.tokenizer_2
-del pipeline
-
-flush()
-```
-
-Load the diffusion transformer next which has 12.5B parameters. This time, set `device_map="auto"` to automatically distribute the model across two 16GB GPUs. The `auto` strategy is backed by [Accelerate](https://hf.co/docs/accelerate/index) and available as a part of the [Big Model Inference](https://hf.co/docs/accelerate/concept_guides/big_model_inference) feature. It starts by distributing a model across the fastest device first (GPU) before moving to slower devices like the CPU and hard drive if needed. The trade-off of storing model parameters on slower devices is slower inference latency.
-
-```py
-from diffusers import FluxTransformer2DModel
-import torch 
-
-transformer = FluxTransformer2DModel.from_pretrained(
-    "black-forest-labs/FLUX.1-dev", 
-    subfolder="transformer",
-    device_map="auto",
-    torch_dtype=torch.bfloat16
-)
-```
-
-> [!TIP]
-> At any point, you can try `print(pipeline.hf_device_map)` to see how the various models are distributed across devices. This is useful for tracking the device placement of the models.
-
-Add the transformer model to the pipeline for denoising, but set the other model-level components like the text encoders and VAE to `None` because you don't need them yet.
-
-```py
-pipeline = FluxPipeline.from_pretrained(
-    "black-forest-labs/FLUX.1-dev", ,
-    text_encoder=None,
-    text_encoder_2=None,
-    tokenizer=None,
-    tokenizer_2=None,
-    vae=None,
-    transformer=transformer,
-    torch_dtype=torch.bfloat16
-)
-
-print("Running denoising.")
-height, width = 768, 1360
-latents = pipeline(
-    prompt_embeds=prompt_embeds,
-    pooled_prompt_embeds=pooled_prompt_embeds,
-    num_inference_steps=50,
-    guidance_scale=3.5,
-    height=height,
-    width=width,
-    output_type="latent",
-).images
-```
-
-Remove the pipeline and transformer from memory as they're no longer needed.
-
-```py
-del pipeline.transformer
-del pipeline
-
-flush()
-```
-
-Finally, decode the latents with the VAE into an image. The VAE is typically small enough to be loaded on a single GPU.
-
-```py
-from diffusers import AutoencoderKL
-from diffusers.image_processor import VaeImageProcessor
-import torch 
-
-vae = AutoencoderKL.from_pretrained(ckpt_id, subfolder="vae", torch_dtype=torch.bfloat16).to("cuda")
-vae_scale_factor = 2 ** (len(vae.config.block_out_channels))
-image_processor = VaeImageProcessor(vae_scale_factor=vae_scale_factor)
-
-with torch.no_grad():
-    print("Running decoding.")
-    latents = FluxPipeline._unpack_latents(latents, height, width, vae_scale_factor)
-    latents = (latents / vae.config.scaling_factor) + vae.config.shift_factor
-
-    image = vae.decode(latents, return_dict=False)[0]
-    image = image_processor.postprocess(image, output_type="pil")
-    image[0].save("split_transformer.png")
-```
-
-By selectively loading and unloading the models you need at a given stage and sharding the largest models across multiple GPUs, it is possible to run inference with large models on consumer GPUs.

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

@@ -171,13 +171,14 @@ def latents_to_rgb(latents):
     weights = (
         (60, -60, 25, -70),
         (60,  -5, 15, -50),
-        (60,  10, -5, -35),
+        (60,  10, -5, -35)
     )
 
     weights_tensor = torch.t(torch.tensor(weights, dtype=latents.dtype).to(latents.device))
     biases_tensor = torch.tensor((150, 140, 130), dtype=latents.dtype).to(latents.device)
     rgb_tensor = torch.einsum("...lxy,lr -> ...rxy", latents, weights_tensor) + biases_tensor.unsqueeze(-1).unsqueeze(-1)
-    image_array = rgb_tensor.clamp(0, 255).byte().cpu().numpy().transpose(1, 2, 0)
+    image_array = rgb_tensor.clamp(0, 255)[0].byte().cpu().numpy()
+    image_array = image_array.transpose(1, 2, 0)
 
     return Image.fromarray(image_array)
 ```
@@ -188,7 +189,7 @@ def latents_to_rgb(latents):
 def decode_tensors(pipe, step, timestep, callback_kwargs):
     latents = callback_kwargs["latents"]
 
-    image = latents_to_rgb(latents[0])
+    image = latents_to_rgb(latents)
     image.save(f"{step}.png")
 
     return callback_kwargs

examples/community/README.md

@@ -10,7 +10,6 @@ Please also check out our [Community Scripts](https://github.com/huggingface/dif
 
 | Example                                                                                                                               | Description                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              | Code Example                                                                              | Colab                                                                                                                                                                                                              |                                                        Author |
 |:--------------------------------------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------:|
-|Flux with CFG|[Flux with CFG](https://github.com/ToTheBeginning/PuLID/blob/main/docs/pulid_for_flux.md) provides an implementation of using CFG in [Flux](https://blackforestlabs.ai/announcing-black-forest-labs/).|[Flux with CFG](#flux-with-cfg)|NA|[Linoy Tsaban](https://github.com/linoytsaban), [Apolinário](https://github.com/apolinario), and [Sayak Paul](https://github.com/sayakpaul)|
 |Differential Diffusion|[Differential Diffusion](https://github.com/exx8/differential-diffusion) modifies an image according to a text prompt, and according to a map that specifies the amount of change in each region.|[Differential Diffusion](#differential-diffusion)|[![Hugging Face Space](https://img.shields.io/badge/🤗%20Hugging%20Face-Space-yellow)](https://huggingface.co/spaces/exx8/differential-diffusion) [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/exx8/differential-diffusion/blob/main/examples/SD2.ipynb)|[Eran Levin](https://github.com/exx8) and [Ohad Fried](https://www.ohadf.com/)|
 | HD-Painter                                                                                                                            | [HD-Painter](https://github.com/Picsart-AI-Research/HD-Painter) enables prompt-faithfull and high resolution (up to 2k) image inpainting upon any diffusion-based image inpainting method.                                                                                                                                                                                                                                                                                                               | [HD-Painter](#hd-painter)                                                                 | [![Hugging Face Space](https://img.shields.io/badge/🤗%20Hugging%20Face-Space-yellow)](https://huggingface.co/spaces/PAIR/HD-Painter)                                                                              | [Manukyan Hayk](https://github.com/haikmanukyan) and [Sargsyan Andranik](https://github.com/AndranikSargsyan) |
 | Marigold Monocular Depth Estimation                                                                                                   | A universal monocular depth estimator, utilizing Stable Diffusion, delivering sharp predictions in the wild. (See the [project page](https://marigoldmonodepth.github.io) and [full codebase](https://github.com/prs-eth/marigold) for more details.)                                                                                                                                                                                                                                                        | [Marigold Depth Estimation](#marigold-depth-estimation)                                   | [![Hugging Face Space](https://img.shields.io/badge/🤗%20Hugging%20Face-Space-yellow)](https://huggingface.co/spaces/toshas/marigold) [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/12G8reD13DdpMie5ZQlaFNo2WCGeNUH-u?usp=sharing) | [Bingxin Ke](https://github.com/markkua) and [Anton Obukhov](https://github.com/toshas) |
@@ -83,36 +82,6 @@ pipe = DiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion
 
 ## Example usages
 
-### Flux with CFG
-
-Know more about Flux [here](https://blackforestlabs.ai/announcing-black-forest-labs/). Since Flux doesn't use CFG, this implementation provides one, inspired by the [PuLID Flux adaptation](https://github.com/ToTheBeginning/PuLID/blob/main/docs/pulid_for_flux.md).  
-
-Example usage:
-
-```py
-from diffusers import DiffusionPipeline
-import torch 
-
-pipeline = DiffusionPipeline.from_pretrained(
-    "black-forest-labs/FLUX.1-dev", 
-    torch_dtype=torch.bfloat16, 
-    custom_pipeline="pipeline_flux_with_cfg"
-)
-pipeline.enable_model_cpu_offload()
-prompt = "a watercolor painting of a unicorn"
-negative_prompt = "pink"
-
-img = pipeline(
-    prompt=prompt, 
-    negative_prompt=negative_prompt, 
-    true_cfg=1.5, 
-    guidance_scale=3.5, 
-    num_images_per_prompt=1,
-    generator=torch.manual_seed(0)
-).images[0]
-img.save("cfg_flux.png")
-```
-
 ### Differential Diffusion
 
 **Eran Levin, Ohad Fried**

examples/community/pipeline_flux_with_cfg.py

@@ -289,104 +289,80 @@ class FluxCFGPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixi
         self,
         prompt: Union[str, List[str]],
         prompt_2: Union[str, List[str]],
-        negative_prompt: Union[str, List[str]] = None,
-        negative_prompt_2: Union[str, List[str]] = None,
         device: Optional[torch.device] = None,
         num_images_per_prompt: int = 1,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
         max_sequence_length: int = 512,
         lora_scale: Optional[float] = None,
-        do_true_cfg: bool = False,
     ):
+        r"""
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                prompt to be encoded
+            prompt_2 (`str` or `List[str]`, *optional*):
+                The prompt or prompts to be sent to the `tokenizer_2` and `text_encoder_2`. If not defined, `prompt` is
+                used in all text-encoders
+            device: (`torch.device`):
+                torch device
+            num_images_per_prompt (`int`):
+                number of images that should be generated per prompt
+            prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
+            pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting.
+                If not provided, pooled text embeddings will be generated from `prompt` input argument.
+            lora_scale (`float`, *optional*):
+                A lora scale that will be applied to all LoRA layers of the text encoder if LoRA layers are loaded.
+        """
         device = device or self._execution_device
 
-        # Set LoRA scale if applicable
+        # set lora scale so that monkey patched LoRA
+        # function of text encoder can correctly access it
         if lora_scale is not None and isinstance(self, FluxLoraLoaderMixin):
             self._lora_scale = lora_scale
 
+            # dynamically adjust the LoRA scale
             if self.text_encoder is not None and USE_PEFT_BACKEND:
                 scale_lora_layers(self.text_encoder, lora_scale)
             if self.text_encoder_2 is not None and USE_PEFT_BACKEND:
                 scale_lora_layers(self.text_encoder_2, lora_scale)
 
         prompt = [prompt] if isinstance(prompt, str) else prompt
-        batch_size = len(prompt)
-
-        if do_true_cfg and negative_prompt is not None:
-            negative_prompt = [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt
-            negative_batch_size = len(negative_prompt)
-
-            if negative_batch_size != batch_size:
-                raise ValueError(
-                    f"Negative prompt batch size ({negative_batch_size}) does not match prompt batch size ({batch_size})"
-                )
-
-            # Concatenate prompts
-            prompts = prompt + negative_prompt
-            prompts_2 = (
-                prompt_2 + negative_prompt_2 if prompt_2 is not None and negative_prompt_2 is not None else None
-            )
-        else:
-            prompts = prompt
-            prompts_2 = prompt_2
 
         if prompt_embeds is None:
-            if prompts_2 is None:
-                prompts_2 = prompts
+            prompt_2 = prompt_2 or prompt
+            prompt_2 = [prompt_2] if isinstance(prompt_2, str) else prompt_2
 
-            # Get pooled prompt embeddings from CLIPTextModel
+            # We only use the pooled prompt output from the CLIPTextModel
             pooled_prompt_embeds = self._get_clip_prompt_embeds(
-                prompt=prompts,
+                prompt=prompt,
                 device=device,
                 num_images_per_prompt=num_images_per_prompt,
             )
             prompt_embeds = self._get_t5_prompt_embeds(
-                prompt=prompts_2,
+                prompt=prompt_2,
                 num_images_per_prompt=num_images_per_prompt,
                 max_sequence_length=max_sequence_length,
                 device=device,
             )
 
-            if do_true_cfg and negative_prompt is not None:
-                # Split embeddings back into positive and negative parts
-                total_batch_size = batch_size * num_images_per_prompt
-                positive_indices = slice(0, total_batch_size)
-                negative_indices = slice(total_batch_size, 2 * total_batch_size)
-
-                positive_pooled_prompt_embeds = pooled_prompt_embeds[positive_indices]
-                negative_pooled_prompt_embeds = pooled_prompt_embeds[negative_indices]
-
-                positive_prompt_embeds = prompt_embeds[positive_indices]
-                negative_prompt_embeds = prompt_embeds[negative_indices]
-
-                pooled_prompt_embeds = positive_pooled_prompt_embeds
-                prompt_embeds = positive_prompt_embeds
-
-        # Unscale LoRA layers
         if self.text_encoder is not None:
             if isinstance(self, FluxLoraLoaderMixin) and USE_PEFT_BACKEND:
+                # Retrieve the original scale by scaling back the LoRA layers
                 unscale_lora_layers(self.text_encoder, lora_scale)
 
         if self.text_encoder_2 is not None:
             if isinstance(self, FluxLoraLoaderMixin) and USE_PEFT_BACKEND:
+                # Retrieve the original scale by scaling back the LoRA layers
                 unscale_lora_layers(self.text_encoder_2, lora_scale)
 
         dtype = self.text_encoder.dtype if self.text_encoder is not None else self.transformer.dtype
         text_ids = torch.zeros(prompt_embeds.shape[1], 3).to(device=device, dtype=dtype)
 
-        if do_true_cfg and negative_prompt is not None:
-            return (
-                prompt_embeds,
-                pooled_prompt_embeds,
-                text_ids,
-                negative_prompt_embeds,
-                negative_pooled_prompt_embeds,
-            )
-        else:
-            return prompt_embeds, pooled_prompt_embeds, text_ids, None, None
+        return prompt_embeds, pooled_prompt_embeds, text_ids
 
     def check_inputs(
         self,
@@ -711,33 +687,38 @@ class FluxCFGPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixi
         lora_scale = (
             self.joint_attention_kwargs.get("scale", None) if self.joint_attention_kwargs is not None else None
         )
-        do_true_cfg = true_cfg > 1 and negative_prompt is not None
         (
             prompt_embeds,
             pooled_prompt_embeds,
             text_ids,
-            negative_prompt_embeds,
-            negative_pooled_prompt_embeds,
         ) = self.encode_prompt(
             prompt=prompt,
             prompt_2=prompt_2,
-            negative_prompt=negative_prompt,
-            negative_prompt_2=negative_prompt_2,
             prompt_embeds=prompt_embeds,
             pooled_prompt_embeds=pooled_prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
-            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
             device=device,
             num_images_per_prompt=num_images_per_prompt,
             max_sequence_length=max_sequence_length,
             lora_scale=lora_scale,
-            do_true_cfg=do_true_cfg,
         )
 
+        # perform "real" CFG as suggested for distilled Flux models in https://github.com/ToTheBeginning/PuLID/blob/main/docs/pulid_for_flux.md
+        do_true_cfg = true_cfg > 1 and negative_prompt is not None
         if do_true_cfg:
-            # Concatenate embeddings
-            prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds], dim=0)
-            pooled_prompt_embeds = torch.cat([negative_pooled_prompt_embeds, pooled_prompt_embeds], dim=0)
+            (
+                negative_prompt_embeds,
+                negative_pooled_prompt_embeds,
+                negative_text_ids,
+            ) = self.encode_prompt(
+                prompt=negative_prompt,
+                prompt_2=negative_prompt_2,
+                prompt_embeds=negative_prompt_embeds,
+                pooled_prompt_embeds=negative_pooled_prompt_embeds,
+                device=device,
+                num_images_per_prompt=num_images_per_prompt,
+                max_sequence_length=max_sequence_length,
+                lora_scale=lora_scale,
+            )
 
         # 4. Prepare latent variables
         num_channels_latents = self.transformer.config.in_channels // 4
@@ -773,26 +754,24 @@ class FluxCFGPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixi
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
         self._num_timesteps = len(timesteps)
 
+        # handle guidance
+        if self.transformer.config.guidance_embeds:
+            guidance = torch.full([1], guidance_scale, device=device, dtype=torch.float32)
+            guidance = guidance.expand(latents.shape[0])
+        else:
+            guidance = None
+
         # 6. Denoising loop
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 if self.interrupt:
                     continue
 
-                latent_model_input = torch.cat([latents] * 2) if do_true_cfg else latents
-
-                # handle guidance
-                if self.transformer.config.guidance_embeds:
-                    guidance = torch.full([1], guidance_scale, device=device, dtype=torch.float32)
-                    guidance = guidance.expand(latent_model_input.shape[0])
-                else:
-                    guidance = None
-
                 # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
-                timestep = t.expand(latent_model_input.shape[0]).to(latent_model_input.dtype)
+                timestep = t.expand(latents.shape[0]).to(latents.dtype)
 
                 noise_pred = self.transformer(
-                    hidden_states=latent_model_input,
+                    hidden_states=latents,
                     timestep=timestep / 1000,
                     guidance=guidance,
                     pooled_projections=pooled_prompt_embeds,
@@ -804,7 +783,18 @@ class FluxCFGPipeline(DiffusionPipeline, FluxLoraLoaderMixin, FromSingleFileMixi
                 )[0]
 
                 if do_true_cfg:
-                    neg_noise_pred, noise_pred = noise_pred.chunk(2)
+                    neg_noise_pred = self.transformer(
+                        hidden_states=latents,
+                        timestep=timestep / 1000,
+                        guidance=guidance,
+                        pooled_projections=negative_pooled_prompt_embeds,
+                        encoder_hidden_states=negative_prompt_embeds,
+                        txt_ids=negative_text_ids,
+                        img_ids=latent_image_ids,
+                        joint_attention_kwargs=self.joint_attention_kwargs,
+                        return_dict=False,
+                    )[0]
+
                     noise_pred = neg_noise_pred + true_cfg * (noise_pred - neg_noise_pred)
 
                 # compute the previous noisy sample x_t -> x_t-1

examples/controlnet/train_controlnet_sd3.py

@@ -31,7 +31,7 @@ import torch.utils.checkpoint
 import transformers
 from accelerate import Accelerator
 from accelerate.logging import get_logger
-from accelerate.utils import DistributedDataParallelKwargs, ProjectConfiguration, set_seed
+from accelerate.utils import ProjectConfiguration, set_seed
 from datasets import load_dataset
 from huggingface_hub import create_repo, upload_folder
 from packaging import version
@@ -899,13 +899,12 @@ def main(args):
     logging_dir = Path(args.output_dir, args.logging_dir)
 
     accelerator_project_config = ProjectConfiguration(project_dir=args.output_dir, logging_dir=logging_dir)
-    kwargs = DistributedDataParallelKwargs(find_unused_parameters=True)
+
     accelerator = Accelerator(
         gradient_accumulation_steps=args.gradient_accumulation_steps,
         mixed_precision=args.mixed_precision,
         log_with=args.report_to,
         project_config=accelerator_project_config,
-        kwargs_handlers=[kwargs],
     )
 
     # Disable AMP for MPS.

src/diffusers/loaders/single_file_utils.py

@@ -14,7 +14,6 @@
 # limitations under the License.
 """Conversion script for the Stable Diffusion checkpoints."""
 
-import copy
 import os
 import re
 from contextlib import nullcontext
@@ -92,11 +91,11 @@ DIFFUSERS_DEFAULT_PIPELINE_PATHS = {
     "xl_inpaint": {"pretrained_model_name_or_path": "diffusers/stable-diffusion-xl-1.0-inpainting-0.1"},
     "playground-v2-5": {"pretrained_model_name_or_path": "playgroundai/playground-v2.5-1024px-aesthetic"},
     "upscale": {"pretrained_model_name_or_path": "stabilityai/stable-diffusion-x4-upscaler"},
-    "inpainting": {"pretrained_model_name_or_path": "stable-diffusion-v1-5/stable-diffusion-inpainting"},
+    "inpainting": {"pretrained_model_name_or_path": "Lykon/dreamshaper-8-inpainting"},
     "inpainting_v2": {"pretrained_model_name_or_path": "stabilityai/stable-diffusion-2-inpainting"},
     "controlnet": {"pretrained_model_name_or_path": "lllyasviel/control_v11p_sd15_canny"},
     "v2": {"pretrained_model_name_or_path": "stabilityai/stable-diffusion-2-1"},
-    "v1": {"pretrained_model_name_or_path": "stable-diffusion-v1-5/stable-diffusion-v1-5"},
+    "v1": {"pretrained_model_name_or_path": "Lykon/dreamshaper-8"},
     "stable_cascade_stage_b": {"pretrained_model_name_or_path": "stabilityai/stable-cascade", "subfolder": "decoder"},
     "stable_cascade_stage_b_lite": {
         "pretrained_model_name_or_path": "stabilityai/stable-cascade",
@@ -542,7 +541,6 @@ def infer_diffusers_model_type(checkpoint):
 def fetch_diffusers_config(checkpoint):
     model_type = infer_diffusers_model_type(checkpoint)
     model_path = DIFFUSERS_DEFAULT_PIPELINE_PATHS[model_type]
-    model_path = copy.deepcopy(model_path)
 
     return model_path
 
@@ -1513,7 +1511,7 @@ def _legacy_load_scheduler(
         )
         deprecate("prediction_type", "1.0.0", deprecation_message)
 
-    scheduler_config = copy.deepcopy(SCHEDULER_DEFAULT_CONFIG)
+    scheduler_config = SCHEDULER_DEFAULT_CONFIG
     model_type = infer_diffusers_model_type(checkpoint=checkpoint)
 
     global_step = checkpoint["global_step"] if "global_step" in checkpoint else None

src/diffusers/models/controlnet_sd3.py

@@ -336,7 +336,7 @@ class SD3ControlNetModel(ModelMixin, ConfigMixin, PeftAdapterMixin, FromOriginal
                     return custom_forward
 
                 ckpt_kwargs: Dict[str, Any] = {"use_reentrant": False} if is_torch_version(">=", "1.11.0") else {}
-                encoder_hidden_states, hidden_states = torch.utils.checkpoint.checkpoint(
+                hidden_states = torch.utils.checkpoint.checkpoint(
                     create_custom_forward(block),
                     hidden_states,
                     encoder_hidden_states,

src/diffusers/models/modeling_utils.py

@@ -93,20 +93,24 @@ def get_parameter_device(parameter: torch.nn.Module) -> torch.device:
 
 def get_parameter_dtype(parameter: torch.nn.Module) -> torch.dtype:
     try:
-        return next(parameter.parameters()).dtype
+        params = tuple(parameter.parameters())
+        if len(params) > 0:
+            return params[0].dtype
+
+        buffers = tuple(parameter.buffers())
+        if len(buffers) > 0:
+            return buffers[0].dtype
+
     except StopIteration:
-        try:
-            return next(parameter.buffers()).dtype
-        except StopIteration:
-            # For torch.nn.DataParallel compatibility in PyTorch 1.5
+        # For torch.nn.DataParallel compatibility in PyTorch 1.5
 
-            def find_tensor_attributes(module: torch.nn.Module) -> List[Tuple[str, Tensor]]:
-                tuples = [(k, v) for k, v in module.__dict__.items() if torch.is_tensor(v)]
-                return tuples
+        def find_tensor_attributes(module: torch.nn.Module) -> List[Tuple[str, Tensor]]:
+            tuples = [(k, v) for k, v in module.__dict__.items() if torch.is_tensor(v)]
+            return tuples
 
-            gen = parameter._named_members(get_members_fn=find_tensor_attributes)
-            first_tuple = next(gen)
-            return first_tuple[1].dtype
+        gen = parameter._named_members(get_members_fn=find_tensor_attributes)
+        first_tuple = next(gen)
+        return first_tuple[1].dtype
 
 
 class ModelMixin(torch.nn.Module, PushToHubMixin):

src/diffusers/pipelines/cogvideo/pipeline_cogvideox.py

@@ -188,9 +188,6 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
         self.vae_scale_factor_temporal = (
             self.vae.config.temporal_compression_ratio if hasattr(self, "vae") and self.vae is not None else 4
         )
-        self.vae_scaling_factor_image = (
-            self.vae.config.scaling_factor if hasattr(self, "vae") and self.vae is not None else 0.7
-        )
 
         self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)
 
@@ -320,12 +317,6 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
     def prepare_latents(
         self, batch_size, num_channels_latents, num_frames, height, width, dtype, device, generator, latents=None
     ):
-        if isinstance(generator, list) and len(generator) != batch_size:
-            raise ValueError(
-                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
-                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
-            )
-
         shape = (
             batch_size,
             (num_frames - 1) // self.vae_scale_factor_temporal + 1,
@@ -333,6 +324,11 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
             height // self.vae_scale_factor_spatial,
             width // self.vae_scale_factor_spatial,
         )
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
 
         if latents is None:
             latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
@@ -345,7 +341,7 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
 
     def decode_latents(self, latents: torch.Tensor) -> torch.Tensor:
         latents = latents.permute(0, 2, 1, 3, 4)  # [batch_size, num_channels, num_frames, height, width]
-        latents = 1 / self.vae_scaling_factor_image * latents
+        latents = 1 / self.vae.config.scaling_factor * latents
 
         frames = self.vae.decode(latents).sample
         return frames
@@ -514,10 +510,10 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            height (`int`, *optional*, defaults to self.transformer.config.sample_height * self.vae_scale_factor_spatial):
-                The height in pixels of the generated image. This is set to 480 by default for the best results.
-            width (`int`, *optional*, defaults to self.transformer.config.sample_height * self.vae_scale_factor_spatial):
-                The width in pixels of the generated image. This is set to 720 by default for the best results.
+            height (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The height in pixels of the generated image. This is set to 1024 by default for the best results.
+            width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The width in pixels of the generated image. This is set to 1024 by default for the best results.
             num_frames (`int`, defaults to `48`):
                 Number of frames to generate. Must be divisible by self.vae_scale_factor_temporal. Generated video will
                 contain 1 extra frame because CogVideoX is conditioned with (num_seconds * fps + 1) frames where
@@ -591,6 +587,8 @@ class CogVideoXPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin):
         if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
             callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
 
+        height = height or self.transformer.config.sample_size * self.vae_scale_factor_spatial
+        width = width or self.transformer.config.sample_size * self.vae_scale_factor_spatial
         num_videos_per_prompt = 1
 
         # 1. Check inputs. Raise error if not correct

src/diffusers/pipelines/cogvideo/pipeline_cogvideox_image2video.py

@@ -207,9 +207,6 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
         self.vae_scale_factor_temporal = (
             self.vae.config.temporal_compression_ratio if hasattr(self, "vae") and self.vae is not None else 4
         )
-        self.vae_scaling_factor_image = (
-            self.vae.config.scaling_factor if hasattr(self, "vae") and self.vae is not None else 0.7
-        )
 
         self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)
 
@@ -351,12 +348,6 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
         generator: Optional[torch.Generator] = None,
         latents: Optional[torch.Tensor] = None,
     ):
-        if isinstance(generator, list) and len(generator) != batch_size:
-            raise ValueError(
-                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
-                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
-            )
-
         num_frames = (num_frames - 1) // self.vae_scale_factor_temporal + 1
         shape = (
             batch_size,
@@ -366,6 +357,12 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
             width // self.vae_scale_factor_spatial,
         )
 
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
         image = image.unsqueeze(2)  # [B, C, F, H, W]
 
         if isinstance(generator, list):
@@ -376,7 +373,7 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
             image_latents = [retrieve_latents(self.vae.encode(img.unsqueeze(0)), generator) for img in image]
 
         image_latents = torch.cat(image_latents, dim=0).to(dtype).permute(0, 2, 1, 3, 4)  # [B, F, C, H, W]
-        image_latents = self.vae_scaling_factor_image * image_latents
+        image_latents = self.vae.config.scaling_factor * image_latents
 
         padding_shape = (
             batch_size,
@@ -400,7 +397,7 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
     # Copied from diffusers.pipelines.cogvideo.pipeline_cogvideox.CogVideoXPipeline.decode_latents
     def decode_latents(self, latents: torch.Tensor) -> torch.Tensor:
         latents = latents.permute(0, 2, 1, 3, 4)  # [batch_size, num_channels, num_frames, height, width]
-        latents = 1 / self.vae_scaling_factor_image * latents
+        latents = 1 / self.vae.config.scaling_factor * latents
 
         frames = self.vae.decode(latents).sample
         return frames
@@ -441,6 +438,7 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
         width,
         negative_prompt,
         callback_on_step_end_tensor_inputs,
+        video=None,
         latents=None,
         prompt_embeds=None,
         negative_prompt_embeds=None,
@@ -496,6 +494,9 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
                     f" {negative_prompt_embeds.shape}."
                 )
 
+        if video is not None and latents is not None:
+            raise ValueError("Only one of `video` or `latents` should be provided")
+
     # Copied from diffusers.pipelines.cogvideo.pipeline_cogvideox.CogVideoXPipeline.fuse_qkv_projections
     def fuse_qkv_projections(self) -> None:
         r"""Enables fused QKV projections."""
@@ -583,7 +584,7 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
 
         Args:
             image (`PipelineImageInput`):
-                The input image to condition the generation on. Must be an image, a list of images or a `torch.Tensor`.
+                The input video to condition the generation on. Must be an image, a list of images or a `torch.Tensor`.
             prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
                 instead.
@@ -591,10 +592,10 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            height (`int`, *optional*, defaults to self.transformer.config.sample_height * self.vae_scale_factor_spatial):
-                The height in pixels of the generated image. This is set to 480 by default for the best results.
-            width (`int`, *optional*, defaults to self.transformer.config.sample_height * self.vae_scale_factor_spatial):
-                The width in pixels of the generated image. This is set to 720 by default for the best results.
+            height (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The height in pixels of the generated image. This is set to 1024 by default for the best results.
+            width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The width in pixels of the generated image. This is set to 1024 by default for the best results.
             num_frames (`int`, defaults to `48`):
                 Number of frames to generate. Must be divisible by self.vae_scale_factor_temporal. Generated video will
                 contain 1 extra frame because CogVideoX is conditioned with (num_seconds * fps + 1) frames where
@@ -664,19 +665,20 @@ class CogVideoXImageToVideoPipeline(DiffusionPipeline):
         if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
             callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
 
+        height = height or self.transformer.config.sample_size * self.vae_scale_factor_spatial
+        width = width or self.transformer.config.sample_size * self.vae_scale_factor_spatial
         num_videos_per_prompt = 1
 
         # 1. Check inputs. Raise error if not correct
         self.check_inputs(
-            image=image,
-            prompt=prompt,
-            height=height,
-            width=width,
-            negative_prompt=negative_prompt,
-            callback_on_step_end_tensor_inputs=callback_on_step_end_tensor_inputs,
-            latents=latents,
-            prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
+            image,
+            prompt,
+            height,
+            width,
+            negative_prompt,
+            callback_on_step_end_tensor_inputs,
+            prompt_embeds,
+            negative_prompt_embeds,
         )
         self._guidance_scale = guidance_scale
         self._interrupt = False

src/diffusers/pipelines/cogvideo/pipeline_cogvideox_video2video.py

@@ -204,16 +204,12 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
         self.register_modules(
             tokenizer=tokenizer, text_encoder=text_encoder, vae=vae, transformer=transformer, scheduler=scheduler
         )
-
         self.vae_scale_factor_spatial = (
             2 ** (len(self.vae.config.block_out_channels) - 1) if hasattr(self, "vae") and self.vae is not None else 8
         )
         self.vae_scale_factor_temporal = (
             self.vae.config.temporal_compression_ratio if hasattr(self, "vae") and self.vae is not None else 4
         )
-        self.vae_scaling_factor_image = (
-            self.vae.config.scaling_factor if hasattr(self, "vae") and self.vae is not None else 0.7
-        )
 
         self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)
 
@@ -355,12 +351,6 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
         latents: Optional[torch.Tensor] = None,
         timestep: Optional[torch.Tensor] = None,
     ):
-        if isinstance(generator, list) and len(generator) != batch_size:
-            raise ValueError(
-                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
-                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
-            )
-
         num_frames = (video.size(2) - 1) // self.vae_scale_factor_temporal + 1 if latents is None else latents.size(1)
 
         shape = (
@@ -371,6 +361,12 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
             width // self.vae_scale_factor_spatial,
         )
 
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
         if latents is None:
             if isinstance(generator, list):
                 if len(generator) != batch_size:
@@ -386,7 +382,7 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
                 init_latents = [retrieve_latents(self.vae.encode(vid.unsqueeze(0)), generator) for vid in video]
 
             init_latents = torch.cat(init_latents, dim=0).to(dtype).permute(0, 2, 1, 3, 4)  # [B, F, C, H, W]
-            init_latents = self.vae_scaling_factor_image * init_latents
+            init_latents = self.vae.config.scaling_factor * init_latents
 
             noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
             latents = self.scheduler.add_noise(init_latents, noise, timestep)
@@ -400,7 +396,7 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
     # Copied from diffusers.pipelines.cogvideo.pipeline_cogvideox.CogVideoXPipeline.decode_latents
     def decode_latents(self, latents: torch.Tensor) -> torch.Tensor:
         latents = latents.permute(0, 2, 1, 3, 4)  # [batch_size, num_channels, num_frames, height, width]
-        latents = 1 / self.vae_scaling_factor_image * latents
+        latents = 1 / self.vae.config.scaling_factor * latents
 
         frames = self.vae.decode(latents).sample
         return frames
@@ -593,10 +589,10 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            height (`int`, *optional*, defaults to self.transformer.config.sample_height * self.vae_scale_factor_spatial):
-                The height in pixels of the generated image. This is set to 480 by default for the best results.
-            width (`int`, *optional*, defaults to self.transformer.config.sample_height * self.vae_scale_factor_spatial):
-                The width in pixels of the generated image. This is set to 720 by default for the best results.
+            height (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The height in pixels of the generated image. This is set to 1024 by default for the best results.
+            width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
+                The width in pixels of the generated image. This is set to 1024 by default for the best results.
             num_inference_steps (`int`, *optional*, defaults to 50):
                 The number of denoising steps. More denoising steps usually lead to a higher quality image at the
                 expense of slower inference.
@@ -662,20 +658,20 @@ class CogVideoXVideoToVideoPipeline(DiffusionPipeline, CogVideoXLoraLoaderMixin)
         if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
             callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
 
+        height = height or self.transformer.config.sample_size * self.vae_scale_factor_spatial
+        width = width or self.transformer.config.sample_size * self.vae_scale_factor_spatial
         num_videos_per_prompt = 1
 
         # 1. Check inputs. Raise error if not correct
         self.check_inputs(
-            prompt=prompt,
-            height=height,
-            width=width,
-            strength=strength,
-            negative_prompt=negative_prompt,
-            callback_on_step_end_tensor_inputs=callback_on_step_end_tensor_inputs,
-            video=video,
-            latents=latents,
-            prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
+            prompt,
+            height,
+            width,
+            strength,
+            negative_prompt,
+            callback_on_step_end_tensor_inputs,
+            prompt_embeds,
+            negative_prompt_embeds,
         )
         self._guidance_scale = guidance_scale
         self._attention_kwargs = attention_kwargs

src/diffusers/pipelines/ddpm/pipeline_ddpm.py

@@ -101,10 +101,10 @@ class DDPMPipeline(DiffusionPipeline):
 
         if self.device.type == "mps":
             # randn does not work reproducibly on mps
-            image = randn_tensor(image_shape, generator=generator, dtype=self.unet.dtype)
+            image = randn_tensor(image_shape, generator=generator)
             image = image.to(self.device)
         else:
-            image = randn_tensor(image_shape, generator=generator, device=self.device, dtype=self.unet.dtype)
+            image = randn_tensor(image_shape, generator=generator, device=self.device)
 
         # set step values
         self.scheduler.set_timesteps(num_inference_steps)

src/diffusers/schedulers/scheduling_deis_multistep.py

@@ -111,8 +111,6 @@ class DEISMultistepScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
              Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
              the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.
@@ -140,12 +138,9 @@ class DEISMultistepScheduler(SchedulerMixin, ConfigMixin):
         solver_type: str = "logrho",
         lower_order_final: bool = True,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -260,9 +255,6 @@ class DEISMultistepScheduler(SchedulerMixin, ConfigMixin):
             sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=num_inference_steps)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
             sigmas = np.concatenate([sigmas, sigmas[-1:]]).astype(np.float32)
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)
             sigma_last = ((1 - self.alphas_cumprod[0]) / self.alphas_cumprod[0]) ** 0.5
@@ -374,28 +366,6 @@ class DEISMultistepScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def convert_model_output(
         self,
         model_output: torch.Tensor,

src/diffusers/schedulers/scheduling_dpmsolver_multistep.py

@@ -161,8 +161,6 @@ class DPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         use_lu_lambdas (`bool`, *optional*, defaults to `False`):
             Whether to use the uniform-logSNR for step sizes proposed by Lu's DPM-Solver in the noise schedule during
             the sampling process. If `True`, the sigmas and time steps are determined according to a sequence of
@@ -208,7 +206,6 @@ class DPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
         lower_order_final: bool = True,
         euler_at_final: bool = False,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         use_lu_lambdas: Optional[bool] = False,
         final_sigmas_type: Optional[str] = "zero",  # "zero", "sigma_min"
         lambda_min_clipped: float = -float("inf"),
@@ -217,8 +214,6 @@ class DPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
         steps_offset: int = 0,
         rescale_betas_zero_snr: bool = False,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if algorithm_type in ["dpmsolver", "sde-dpmsolver"]:
             deprecation_message = f"algorithm_type {algorithm_type} is deprecated and will be removed in a future version. Choose from `dpmsolver++` or `sde-dpmsolver++` instead"
             deprecate("algorithm_types dpmsolver and sde-dpmsolver", "1.0.0", deprecation_message)
@@ -335,8 +330,6 @@ class DPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
             raise ValueError("Cannot use `timesteps` with `config.use_karras_sigmas = True`")
         if timesteps is not None and self.config.use_lu_lambdas:
             raise ValueError("Cannot use `timesteps` with `config.use_lu_lambdas = True`")
-        if timesteps is not None and self.config.use_exponential_sigmas:
-            raise ValueError("Cannot set `timesteps` with `config.use_exponential_sigmas = True`.")
 
         if timesteps is not None:
             timesteps = np.array(timesteps).astype(np.int64)
@@ -385,9 +378,6 @@ class DPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
             lambdas = self._convert_to_lu(in_lambdas=lambdas, num_inference_steps=num_inference_steps)
             sigmas = np.exp(lambdas)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)
 
@@ -520,28 +510,6 @@ class DPMSolverMultistepScheduler(SchedulerMixin, ConfigMixin):
         lambdas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return lambdas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def convert_model_output(
         self,
         model_output: torch.Tensor,

src/diffusers/schedulers/scheduling_dpmsolver_multistep_inverse.py

@@ -124,8 +124,6 @@ class DPMSolverMultistepInverseScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         lambda_min_clipped (`float`, defaults to `-inf`):
             Clipping threshold for the minimum value of `lambda(t)` for numerical stability. This is critical for the
             cosine (`squaredcos_cap_v2`) noise schedule.
@@ -160,14 +158,11 @@ class DPMSolverMultistepInverseScheduler(SchedulerMixin, ConfigMixin):
         lower_order_final: bool = True,
         euler_at_final: bool = False,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         lambda_min_clipped: float = -float("inf"),
         variance_type: Optional[str] = None,
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if algorithm_type in ["dpmsolver", "sde-dpmsolver"]:
             deprecation_message = f"algorithm_type {algorithm_type} is deprecated and will be removed in a future version. Choose from `dpmsolver++` or `sde-dpmsolver++` instead"
             deprecate("algorithm_types dpmsolver and sde-dpmsolver", "1.0.0", deprecation_message)
@@ -218,7 +213,6 @@ class DPMSolverMultistepInverseScheduler(SchedulerMixin, ConfigMixin):
         self._step_index = None
         self.sigmas = self.sigmas.to("cpu")  # to avoid too much CPU/GPU communication
         self.use_karras_sigmas = use_karras_sigmas
-        self.use_exponential_sigmas = use_exponential_sigmas
 
     @property
     def step_index(self):
@@ -273,9 +267,6 @@ class DPMSolverMultistepInverseScheduler(SchedulerMixin, ConfigMixin):
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
             timesteps = timesteps.copy().astype(np.int64)
             sigmas = np.concatenate([sigmas, sigmas[-1:]]).astype(np.float32)
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)
             sigma_max = (
@@ -394,28 +385,6 @@ class DPMSolverMultistepInverseScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     # Copied from diffusers.schedulers.scheduling_dpmsolver_multistep.DPMSolverMultistepScheduler.convert_model_output
     def convert_model_output(
         self,

src/diffusers/schedulers/scheduling_dpmsolver_sde.py

@@ -160,8 +160,6 @@ class DPMSolverSDEScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         noise_sampler_seed (`int`, *optional*, defaults to `None`):
             The random seed to use for the noise sampler. If `None`, a random seed is generated.
         timestep_spacing (`str`, defaults to `"linspace"`):
@@ -184,13 +182,10 @@ class DPMSolverSDEScheduler(SchedulerMixin, ConfigMixin):
         trained_betas: Optional[Union[np.ndarray, List[float]]] = None,
         prediction_type: str = "epsilon",
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         noise_sampler_seed: Optional[int] = None,
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -346,9 +341,6 @@ class DPMSolverSDEScheduler(SchedulerMixin, ConfigMixin):
         if self.config.use_karras_sigmas:
             sigmas = self._convert_to_karras(in_sigmas=sigmas)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
 
         second_order_timesteps = self._second_order_timesteps(sigmas, log_sigmas)
 
@@ -429,28 +421,6 @@ class DPMSolverSDEScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     @property
     def state_in_first_order(self):
         return self.sample is None

src/diffusers/schedulers/scheduling_dpmsolver_singlestep.py

@@ -123,8 +123,6 @@ class DPMSolverSinglestepScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         final_sigmas_type (`str`, *optional*, defaults to `"zero"`):
             The final `sigma` value for the noise schedule during the sampling process. If `"sigma_min"`, the final
             sigma is the same as the last sigma in the training schedule. If `zero`, the final sigma is set to 0.
@@ -156,13 +154,10 @@ class DPMSolverSinglestepScheduler(SchedulerMixin, ConfigMixin):
         solver_type: str = "midpoint",
         lower_order_final: bool = False,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         final_sigmas_type: Optional[str] = "zero",  # "zero", "sigma_min"
         lambda_min_clipped: float = -float("inf"),
         variance_type: Optional[str] = None,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if algorithm_type == "dpmsolver":
             deprecation_message = "algorithm_type `dpmsolver` is deprecated and will be removed in a future version. Choose from `dpmsolver++` or `sde-dpmsolver++` instead"
             deprecate("algorithm_types=dpmsolver", "1.0.0", deprecation_message)
@@ -305,8 +300,6 @@ class DPMSolverSinglestepScheduler(SchedulerMixin, ConfigMixin):
             raise ValueError("Must pass exactly one of  `num_inference_steps` or `timesteps`.")
         if timesteps is not None and self.config.use_karras_sigmas:
             raise ValueError("Cannot use `timesteps` when `config.use_karras_sigmas=True`.")
-        if timesteps is not None and self.config.use_exponential_sigmas:
-            raise ValueError("Cannot set `timesteps` with `config.use_exponential_sigmas = True`.")
 
         num_inference_steps = num_inference_steps or len(timesteps)
         self.num_inference_steps = num_inference_steps
@@ -330,9 +323,6 @@ class DPMSolverSinglestepScheduler(SchedulerMixin, ConfigMixin):
             sigmas = np.flip(sigmas).copy()
             sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=num_inference_steps)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)
 
@@ -462,28 +452,6 @@ class DPMSolverSinglestepScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def convert_model_output(
         self,
         model_output: torch.Tensor,

src/diffusers/schedulers/scheduling_euler_discrete.py

@@ -158,8 +158,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.
@@ -188,7 +186,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
         prediction_type: str = "epsilon",
         interpolation_type: str = "linear",
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         sigma_min: Optional[float] = None,
         sigma_max: Optional[float] = None,
         timestep_spacing: str = "linspace",
@@ -197,8 +194,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
         rescale_betas_zero_snr: bool = False,
         final_sigmas_type: str = "zero",  # can be "zero" or "sigma_min"
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -240,7 +235,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
 
         self.is_scale_input_called = False
         self.use_karras_sigmas = use_karras_sigmas
-        self.use_exponential_sigmas = use_exponential_sigmas
 
         self._step_index = None
         self._begin_index = None
@@ -338,8 +332,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
             raise ValueError("Can only pass one of `num_inference_steps` or `timesteps` or `sigmas`.")
         if timesteps is not None and self.config.use_karras_sigmas:
             raise ValueError("Cannot set `timesteps` with `config.use_karras_sigmas = True`.")
-        if timesteps is not None and self.config.use_exponential_sigmas:
-            raise ValueError("Cannot set `timesteps` with `config.use_exponential_sigmas = True`.")
         if (
             timesteps is not None
             and self.config.timestep_type == "continuous"
@@ -404,10 +396,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
                 sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
                 timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
 
-            elif self.config.use_exponential_sigmas:
-                sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-                timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
-
             if self.config.final_sigmas_type == "sigma_min":
                 sigma_last = ((1 - self.alphas_cumprod[0]) / self.alphas_cumprod[0]) ** 0.5
             elif self.config.final_sigmas_type == "zero":
@@ -480,28 +468,6 @@ class EulerDiscreteScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from https://github.com/crowsonkb/k-diffusion/blob/686dbad0f39640ea25c8a8c6a6e56bb40eacefa2/k_diffusion/sampling.py#L26
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def index_for_timestep(self, timestep, schedule_timesteps=None):
         if schedule_timesteps is None:
             schedule_timesteps = self.timesteps

src/diffusers/schedulers/scheduling_heun_discrete.py

@@ -97,8 +97,6 @@ class HeunDiscreteScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.
@@ -119,14 +117,11 @@ class HeunDiscreteScheduler(SchedulerMixin, ConfigMixin):
         trained_betas: Optional[Union[np.ndarray, List[float]]] = None,
         prediction_type: str = "epsilon",
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         clip_sample: Optional[bool] = False,
         clip_sample_range: float = 1.0,
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -256,8 +251,6 @@ class HeunDiscreteScheduler(SchedulerMixin, ConfigMixin):
             raise ValueError("Can only pass one of `num_inference_steps` or `custom_timesteps`.")
         if timesteps is not None and self.config.use_karras_sigmas:
             raise ValueError("Cannot use `timesteps` with `config.use_karras_sigmas = True`")
-        if timesteps is not None and self.config.use_exponential_sigmas:
-            raise ValueError("Cannot set `timesteps` with `config.use_exponential_sigmas = True`.")
 
         num_inference_steps = num_inference_steps or len(timesteps)
         self.num_inference_steps = num_inference_steps
@@ -293,9 +286,6 @@ class HeunDiscreteScheduler(SchedulerMixin, ConfigMixin):
         if self.config.use_karras_sigmas:
             sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
 
         sigmas = np.concatenate([sigmas, [0.0]]).astype(np.float32)
         sigmas = torch.from_numpy(sigmas).to(device=device)
@@ -364,28 +354,6 @@ class HeunDiscreteScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     @property
     def state_in_first_order(self):
         return self.dt is None

src/diffusers/schedulers/scheduling_k_dpm_2_ancestral_discrete.py

@@ -91,8 +91,6 @@ class KDPM2AncestralDiscreteScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
@@ -116,13 +114,10 @@ class KDPM2AncestralDiscreteScheduler(SchedulerMixin, ConfigMixin):
         beta_schedule: str = "linear",
         trained_betas: Optional[Union[np.ndarray, List[float]]] = None,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         prediction_type: str = "epsilon",
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -255,9 +250,6 @@ class KDPM2AncestralDiscreteScheduler(SchedulerMixin, ConfigMixin):
         if self.config.use_karras_sigmas:
             sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=num_inference_steps)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
 
         self.log_sigmas = torch.from_numpy(log_sigmas).to(device)
         sigmas = np.concatenate([sigmas, [0.0]]).astype(np.float32)
@@ -354,28 +346,6 @@ class KDPM2AncestralDiscreteScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     @property
     def state_in_first_order(self):
         return self.sample is None

src/diffusers/schedulers/scheduling_k_dpm_2_discrete.py

@@ -90,8 +90,6 @@ class KDPM2DiscreteScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
@@ -115,13 +113,10 @@ class KDPM2DiscreteScheduler(SchedulerMixin, ConfigMixin):
         beta_schedule: str = "linear",
         trained_betas: Optional[Union[np.ndarray, List[float]]] = None,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         prediction_type: str = "epsilon",
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -254,9 +249,6 @@ class KDPM2DiscreteScheduler(SchedulerMixin, ConfigMixin):
         if self.config.use_karras_sigmas:
             sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=num_inference_steps)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
 
         self.log_sigmas = torch.from_numpy(log_sigmas).to(device=device)
         sigmas = np.concatenate([sigmas, [0.0]]).astype(np.float32)
@@ -367,28 +359,6 @@ class KDPM2DiscreteScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def step(
         self,
         model_output: Union[torch.Tensor, np.ndarray],

src/diffusers/schedulers/scheduling_lms_discrete.py

@@ -111,8 +111,6 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         prediction_type (`str`, defaults to `epsilon`, *optional*):
             Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
             `sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
@@ -136,13 +134,10 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
         beta_schedule: str = "linear",
         trained_betas: Optional[Union[np.ndarray, List[float]]] = None,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         prediction_type: str = "epsilon",
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -294,9 +289,6 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
         if self.config.use_karras_sigmas:
             sigmas = self._convert_to_karras(in_sigmas=sigmas)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
 
         sigmas = np.concatenate([sigmas, [0.0]]).astype(np.float32)
 
@@ -370,28 +362,6 @@ class LMSDiscreteScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def step(
         self,
         model_output: torch.Tensor,

src/diffusers/schedulers/scheduling_sasolver.py

@@ -122,8 +122,6 @@ class SASolverScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         lambda_min_clipped (`float`, defaults to `-inf`):
             Clipping threshold for the minimum value of `lambda(t)` for numerical stability. This is critical for the
             cosine (`squaredcos_cap_v2`) noise schedule.
@@ -158,14 +156,11 @@ class SASolverScheduler(SchedulerMixin, ConfigMixin):
         algorithm_type: str = "data_prediction",
         lower_order_final: bool = True,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         lambda_min_clipped: float = -float("inf"),
         variance_type: Optional[str] = None,
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -289,9 +284,6 @@ class SASolverScheduler(SchedulerMixin, ConfigMixin):
             sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=num_inference_steps)
             timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
             sigmas = np.concatenate([sigmas, sigmas[-1:]]).astype(np.float32)
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)
             sigma_last = ((1 - self.alphas_cumprod[0]) / self.alphas_cumprod[0]) ** 0.5
@@ -403,28 +395,6 @@ class SASolverScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def convert_model_output(
         self,
         model_output: torch.Tensor,

src/diffusers/schedulers/scheduling_unipc_multistep.py

@@ -159,8 +159,6 @@ class UniPCMultistepScheduler(SchedulerMixin, ConfigMixin):
         use_karras_sigmas (`bool`, *optional*, defaults to `False`):
             Whether to use Karras sigmas for step sizes in the noise schedule during the sampling process. If `True`,
             the sigmas are determined according to a sequence of noise levels {σi}.
-        use_exponential_sigmas (`bool`, *optional*, defaults to `False`):
-            Whether to use exponential sigmas for step sizes in the noise schedule during the sampling process.
         timestep_spacing (`str`, defaults to `"linspace"`):
             The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
             Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.
@@ -197,14 +195,11 @@ class UniPCMultistepScheduler(SchedulerMixin, ConfigMixin):
         disable_corrector: List[int] = [],
         solver_p: SchedulerMixin = None,
         use_karras_sigmas: Optional[bool] = False,
-        use_exponential_sigmas: Optional[bool] = False,
         timestep_spacing: str = "linspace",
         steps_offset: int = 0,
         final_sigmas_type: Optional[str] = "zero",  # "zero", "sigma_min"
         rescale_betas_zero_snr: bool = False,
     ):
-        if sum([self.config.use_exponential_sigmas, self.config.use_karras_sigmas]) > 1:
-            raise ValueError("Only one of `config.use_exponential_sigmas`, `config.use_karras_sigmas` can be used.")
         if trained_betas is not None:
             self.betas = torch.tensor(trained_betas, dtype=torch.float32)
         elif beta_schedule == "linear":
@@ -334,9 +329,6 @@ class UniPCMultistepScheduler(SchedulerMixin, ConfigMixin):
                     f"`final_sigmas_type` must be one of 'zero', or 'sigma_min', but got {self.config.final_sigmas_type}"
                 )
             sigmas = np.concatenate([sigmas, [sigma_last]]).astype(np.float32)
-        elif self.config.use_exponential_sigmas:
-            sigmas = self._convert_to_exponential(in_sigmas=sigmas, num_inference_steps=self.num_inference_steps)
-            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas])
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)
             if self.config.final_sigmas_type == "sigma_min":
@@ -458,28 +450,6 @@ class UniPCMultistepScheduler(SchedulerMixin, ConfigMixin):
         sigmas = (max_inv_rho + ramp * (min_inv_rho - max_inv_rho)) ** rho
         return sigmas
 
-    # Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._convert_to_exponential
-    def _convert_to_exponential(self, in_sigmas: torch.Tensor, num_inference_steps: int) -> torch.Tensor:
-        """Constructs an exponential noise schedule."""
-
-        # Hack to make sure that other schedulers which copy this function don't break
-        # TODO: Add this logic to the other schedulers
-        if hasattr(self.config, "sigma_min"):
-            sigma_min = self.config.sigma_min
-        else:
-            sigma_min = None
-
-        if hasattr(self.config, "sigma_max"):
-            sigma_max = self.config.sigma_max
-        else:
-            sigma_max = None
-
-        sigma_min = sigma_min if sigma_min is not None else in_sigmas[-1].item()
-        sigma_max = sigma_max if sigma_max is not None else in_sigmas[0].item()
-
-        sigmas = torch.linspace(math.log(sigma_max), math.log(sigma_min), num_inference_steps).exp()
-        return sigmas
-
     def convert_model_output(
         self,
         model_output: torch.Tensor,

tests/lora/test_lora_layers_sd.py

@@ -102,7 +102,7 @@ class StableDiffusionLoRATests(PeftLoraLoaderMixinTests, unittest.TestCase):
     @slow
     @require_torch_gpu
     def test_integration_move_lora_cpu(self):
-        path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        path = "Jiali/stable-diffusion-1.5"
         lora_id = "takuma104/lora-test-text-encoder-lora-target"
 
         pipe = StableDiffusionPipeline.from_pretrained(path, torch_dtype=torch.float16)
@@ -161,7 +161,7 @@ class StableDiffusionLoRATests(PeftLoraLoaderMixinTests, unittest.TestCase):
     def test_integration_move_lora_dora_cpu(self):
         from peft import LoraConfig
 
-        path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        path = "Jiali/stable-diffusion-1.5"
         unet_lora_config = LoraConfig(
             init_lora_weights="gaussian",
             target_modules=["to_k", "to_q", "to_v", "to_out.0"],
@@ -221,7 +221,7 @@ class LoraIntegrationTests(unittest.TestCase):
         torch.cuda.empty_cache()
 
     def test_integration_logits_with_scale(self):
-        path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        path = "Jiali/stable-diffusion-1.5"
         lora_id = "takuma104/lora-test-text-encoder-lora-target"
 
         pipe = StableDiffusionPipeline.from_pretrained(path, torch_dtype=torch.float32)
@@ -253,7 +253,7 @@ class LoraIntegrationTests(unittest.TestCase):
         release_memory(pipe)
 
     def test_integration_logits_no_scale(self):
-        path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        path = "Jiali/stable-diffusion-1.5"
         lora_id = "takuma104/lora-test-text-encoder-lora-target"
 
         pipe = StableDiffusionPipeline.from_pretrained(path, torch_dtype=torch.float32)
@@ -284,7 +284,7 @@ class LoraIntegrationTests(unittest.TestCase):
 
         lora_model_id = "hf-internal-testing/lora_dreambooth_dog_example"
 
-        base_model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        base_model_id = "Jiali/stable-diffusion-1.5"
 
         pipe = StableDiffusionPipeline.from_pretrained(base_model_id, safety_checker=None)
         pipe = pipe.to(torch_device)
@@ -308,7 +308,7 @@ class LoraIntegrationTests(unittest.TestCase):
 
         lora_model_id = "hf-internal-testing/lora-trained"
 
-        base_model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        base_model_id = "Jiali/stable-diffusion-1.5"
 
         pipe = StableDiffusionPipeline.from_pretrained(base_model_id, safety_checker=None)
         pipe = pipe.to(torch_device)
@@ -419,9 +419,9 @@ class LoraIntegrationTests(unittest.TestCase):
     def test_kohya_sd_v15_with_higher_dimensions(self):
         generator = torch.Generator().manual_seed(0)
 
-        pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        ).to(torch_device)
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None).to(
+            torch_device
+        )
         lora_model_id = "hf-internal-testing/urushisato-lora"
         lora_filename = "urushisato_v15.safetensors"
         pipe.load_lora_weights(lora_model_id, weight_name=lora_filename)
@@ -444,7 +444,7 @@ class LoraIntegrationTests(unittest.TestCase):
 
         lora_model_id = "hf-internal-testing/sd-model-finetuned-lora-t4"
 
-        base_model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        base_model_id = "Jiali/stable-diffusion-1.5"
 
         pipe = StableDiffusionPipeline.from_pretrained(base_model_id, safety_checker=None)
         pipe = pipe.to(torch_device)
@@ -467,9 +467,9 @@ class LoraIntegrationTests(unittest.TestCase):
         prompt = "masterpiece, best quality, mountain"
         num_inference_steps = 2
 
-        pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        ).to(torch_device)
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None).to(
+            torch_device
+        )
         initial_images = pipe(
             prompt, output_type="np", generator=generator, num_inference_steps=num_inference_steps
         ).images
@@ -505,9 +505,9 @@ class LoraIntegrationTests(unittest.TestCase):
         prompt = "masterpiece, best quality, mountain"
         num_inference_steps = 2
 
-        pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        ).to(torch_device)
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None).to(
+            torch_device
+        )
         initial_images = pipe(
             prompt, output_type="np", generator=generator, num_inference_steps=num_inference_steps
         ).images
@@ -547,9 +547,9 @@ class LoraIntegrationTests(unittest.TestCase):
 
     def test_not_empty_state_dict(self):
         # Makes sure https://github.com/huggingface/diffusers/issues/7054 does not happen again
-        pipe = AutoPipelineForText2Image.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16
-        ).to(torch_device)
+        pipe = AutoPipelineForText2Image.from_pretrained("Jiali/stable-diffusion-1.5", torch_dtype=torch.float16).to(
+            torch_device
+        )
         pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
         cached_file = hf_hub_download("hf-internal-testing/lcm-lora-test-sd-v1-5", "test_lora.safetensors")
@@ -561,9 +561,9 @@ class LoraIntegrationTests(unittest.TestCase):
 
     def test_load_unload_load_state_dict(self):
         # Makes sure https://github.com/huggingface/diffusers/issues/7054 does not happen again
-        pipe = AutoPipelineForText2Image.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16
-        ).to(torch_device)
+        pipe = AutoPipelineForText2Image.from_pretrained("Jiali/stable-diffusion-1.5", torch_dtype=torch.float16).to(
+            torch_device
+        )
         pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
         cached_file = hf_hub_download("hf-internal-testing/lcm-lora-test-sd-v1-5", "test_lora.safetensors")
@@ -580,9 +580,7 @@ class LoraIntegrationTests(unittest.TestCase):
         release_memory(pipe)
 
     def test_sdv1_5_lcm_lora(self):
-        pipe = DiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16
-        )
+        pipe = DiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", torch_dtype=torch.float16)
         pipe.to(torch_device)
         pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
@@ -610,9 +608,7 @@ class LoraIntegrationTests(unittest.TestCase):
         release_memory(pipe)
 
     def test_sdv1_5_lcm_lora_img2img(self):
-        pipe = AutoPipelineForImage2Image.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16
-        )
+        pipe = AutoPipelineForImage2Image.from_pretrained("Jiali/stable-diffusion-1.5", torch_dtype=torch.float16)
         pipe.to(torch_device)
         pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
@@ -653,7 +649,7 @@ class LoraIntegrationTests(unittest.TestCase):
         This test simply checks that loading a LoRA with an empty network alpha works fine
         See: https://github.com/huggingface/diffusers/issues/5606
         """
-        pipeline = StableDiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        pipeline = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         pipeline.enable_sequential_cpu_offload()
         civitai_path = hf_hub_download("ybelkada/test-ahi-civitai", "ahi_lora_weights.safetensors")
         pipeline.load_lora_weights(civitai_path, adapter_name="ahri")

tests/models/autoencoders/test_models_vae.py

@@ -1051,9 +1051,7 @@ class ConsistencyDecoderVAEIntegrationTests(unittest.TestCase):
 
     def test_sd(self):
         vae = ConsistencyDecoderVAE.from_pretrained("openai/consistency-decoder")  # TODO - update
-        pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", vae=vae, safety_checker=None
-        )
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", vae=vae, safety_checker=None)
         pipe.to(torch_device)
 
         out = pipe(
@@ -1101,7 +1099,7 @@ class ConsistencyDecoderVAEIntegrationTests(unittest.TestCase):
             "openai/consistency-decoder", torch_dtype=torch.float16
         )  # TODO - update
         pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             torch_dtype=torch.float16,
             vae=vae,
             safety_checker=None,
@@ -1126,7 +1124,7 @@ class ConsistencyDecoderVAEIntegrationTests(unittest.TestCase):
     def test_vae_tiling(self):
         vae = ConsistencyDecoderVAE.from_pretrained("openai/consistency-decoder", torch_dtype=torch.float16)
         pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", vae=vae, safety_checker=None, torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", vae=vae, safety_checker=None, torch_dtype=torch.float16
         )
         pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)

tests/pipelines/controlnet/test_controlnet.py

@@ -73,7 +73,7 @@ def _test_stable_diffusion_compile(in_queue, out_queue, timeout):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.to("cuda")
         pipe.set_progress_bar_config(disable=None)
@@ -715,7 +715,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -742,7 +742,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-depth")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -769,7 +769,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-hed")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -796,7 +796,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-mlsd")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -823,7 +823,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-normal")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -850,7 +850,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-openpose")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -877,7 +877,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-scribble")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -904,7 +904,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-seg")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -935,7 +935,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-seg")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.set_progress_bar_config(disable=None)
         pipe.enable_attention_slicing()
@@ -961,7 +961,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -993,7 +993,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.scheduler = EulerDiscreteScheduler.from_config(pipe.scheduler.config)
         pipe.enable_model_cpu_offload()
@@ -1035,7 +1035,7 @@ class ControlNetPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11e_sd15_shuffle")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)
@@ -1081,9 +1081,7 @@ class StableDiffusionMultiControlNetPipelineSlowTests(unittest.TestCase):
         controlnet_pose = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-openpose")
 
         pipe = StableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            safety_checker=None,
-            controlnet=[controlnet_pose, controlnet_canny],
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=[controlnet_pose, controlnet_canny]
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)

tests/pipelines/controlnet/test_controlnet_img2img.py

@@ -407,7 +407,7 @@ class ControlNetImg2ImgPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny")
 
         pipe = StableDiffusionControlNetImg2ImgPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.enable_model_cpu_offload()
         pipe.set_progress_bar_config(disable=None)

tests/pipelines/controlnet/test_controlnet_inpaint.py

@@ -504,7 +504,7 @@ class ControlNetInpaintPipelineSlowTests(unittest.TestCase):
         controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11p_sd15_inpaint")
 
         pipe = StableDiffusionControlNetInpaintPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, controlnet=controlnet
+            "Jiali/stable-diffusion-1.5", safety_checker=None, controlnet=controlnet
         )
         pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
         pipe.enable_model_cpu_offload()

tests/pipelines/controlnet/test_flax_controlnet.py

@@ -41,7 +41,7 @@ class FlaxControlNetPipelineIntegrationTests(unittest.TestCase):
             "lllyasviel/sd-controlnet-canny", from_pt=True, dtype=jnp.bfloat16
         )
         pipe, params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", controlnet=controlnet, from_pt=True, dtype=jnp.bfloat16
+            "Jiali/stable-diffusion-1.5", controlnet=controlnet, from_pt=True, dtype=jnp.bfloat16
         )
         params["controlnet"] = controlnet_params
 
@@ -86,7 +86,7 @@ class FlaxControlNetPipelineIntegrationTests(unittest.TestCase):
             "lllyasviel/sd-controlnet-openpose", from_pt=True, dtype=jnp.bfloat16
         )
         pipe, params = FlaxStableDiffusionControlNetPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", controlnet=controlnet, from_pt=True, dtype=jnp.bfloat16
+            "Jiali/stable-diffusion-1.5", controlnet=controlnet, from_pt=True, dtype=jnp.bfloat16
         )
         params["controlnet"] = controlnet_params
 

tests/pipelines/ip_adapters/test_ip_adapter_stable_diffusion.py

@@ -170,10 +170,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_text_to_image(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.to(torch_device)
         pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
@@ -203,10 +200,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_image_to_image(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionImg2ImgPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.to(torch_device)
         pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
@@ -238,10 +232,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_inpainting(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionInpaintPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.to(torch_device)
         pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
@@ -269,10 +260,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_text_to_image_model_cpu_offload(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
         pipeline.to(torch_device)
@@ -299,10 +287,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_text_to_image_full_face(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.to(torch_device)
         pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter-full-face_sd15.bin")
@@ -319,10 +304,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_unload(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         before_processors = [attn_proc.__class__ for attn_proc in pipeline.unet.attn_processors.values()]
         pipeline.to(torch_device)
@@ -341,10 +323,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
     def test_multi(self):
         image_encoder = self.get_image_encoder(repo_id="h94/IP-Adapter", subfolder="models/image_encoder")
         pipeline = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
-            image_encoder=image_encoder,
-            safety_checker=None,
-            torch_dtype=self.dtype,
+            "Jiali/stable-diffusion-1.5", image_encoder=image_encoder, safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.to(torch_device)
         pipeline.load_ip_adapter(
@@ -364,7 +343,7 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
 
     def test_text_to_image_face_id(self):
         pipeline = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, torch_dtype=self.dtype
+            "Jiali/stable-diffusion-1.5", safety_checker=None, torch_dtype=self.dtype
         )
         pipeline.to(torch_device)
         pipeline.load_ip_adapter(

tests/pipelines/ledits_pp/test_ledits_pp_stable_diffusion.py

@@ -224,7 +224,7 @@ class LEditsPPPipelineStableDiffusionSlowTests(unittest.TestCase):
 
     def test_ledits_pp_editing(self):
         pipe = LEditsPPPipelineStableDiffusion.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None, torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", safety_checker=None, torch_dtype=torch.float16
         )
         pipe = pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)

tests/pipelines/lumina/test_lumina_nextdit.py

@@ -34,19 +34,19 @@ class LuminaText2ImgPipelinePipelineFastTests(unittest.TestCase, PipelineTesterM
     def get_dummy_components(self):
         torch.manual_seed(0)
         transformer = LuminaNextDiT2DModel(
-            sample_size=4,
+            sample_size=16,
             patch_size=2,
             in_channels=4,
-            hidden_size=4,
+            hidden_size=24,
             num_layers=2,
-            num_attention_heads=1,
+            num_attention_heads=3,
             num_kv_heads=1,
             multiple_of=16,
             ffn_dim_multiplier=None,
             norm_eps=1e-5,
             learn_sigma=True,
             qk_norm=True,
-            cross_attention_dim=8,
+            cross_attention_dim=32,
             scaling_factor=1.0,
         )
         torch.manual_seed(0)
@@ -57,8 +57,8 @@ class LuminaText2ImgPipelinePipelineFastTests(unittest.TestCase, PipelineTesterM
 
         torch.manual_seed(0)
         config = GemmaConfig(
-            head_dim=2,
-            hidden_size=8,
+            head_dim=4,
+            hidden_size=32,
             intermediate_size=37,
             num_attention_heads=4,
             num_hidden_layers=2,

tests/pipelines/pag/test_pag_sd.py

@@ -283,7 +283,7 @@ class StableDiffusionPAGPipelineFastTests(
 @require_torch_gpu
 class StableDiffusionPAGPipelineIntegrationTests(unittest.TestCase):
     pipeline_class = StableDiffusionPAGPipeline
-    repo_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+    repo_id = "Jiali/stable-diffusion-1.5"
 
     def setUp(self):
         super().setUp()

tests/pipelines/semantic_stable_diffusion/test_semantic_diffusion.py

@@ -287,7 +287,7 @@ class SemanticDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_positive_guidance(self):
         torch_device = "cuda"
-        pipe = StableDiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         pipe = pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)
 
@@ -370,7 +370,7 @@ class SemanticDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_negative_guidance(self):
         torch_device = "cuda"
-        pipe = StableDiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         pipe = pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)
 
@@ -453,7 +453,7 @@ class SemanticDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_multi_cond_guidance(self):
         torch_device = "cuda"
-        pipe = StableDiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         pipe = pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)
 
@@ -536,9 +536,7 @@ class SemanticDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_guidance_fp16(self):
         torch_device = "cuda"
-        pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16
-        )
+        pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", torch_dtype=torch.float16)
         pipe = pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)
 

tests/pipelines/stable_diffusion/test_onnx_stable_diffusion.py

@@ -250,10 +250,10 @@ class OnnxStableDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_inference_ddim(self):
         ddim_scheduler = DDIMScheduler.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", subfolder="scheduler", revision="onnx"
+            "Jiali/stable-diffusion-1.5", subfolder="scheduler", revision="onnx"
         )
         sd_pipe = OnnxStableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             revision="onnx",
             scheduler=ddim_scheduler,
             safety_checker=None,
@@ -276,10 +276,10 @@ class OnnxStableDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_inference_k_lms(self):
         lms_scheduler = LMSDiscreteScheduler.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", subfolder="scheduler", revision="onnx"
+            "Jiali/stable-diffusion-1.5", subfolder="scheduler", revision="onnx"
         )
         sd_pipe = OnnxStableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             revision="onnx",
             scheduler=lms_scheduler,
             safety_checker=None,
@@ -327,7 +327,7 @@ class OnnxStableDiffusionPipelineIntegrationTests(unittest.TestCase):
         test_callback_fn.has_been_called = False
 
         pipe = OnnxStableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             revision="onnx",
             safety_checker=None,
             feature_extractor=None,
@@ -352,7 +352,7 @@ class OnnxStableDiffusionPipelineIntegrationTests(unittest.TestCase):
 
     def test_stable_diffusion_no_safety_checker(self):
         pipe = OnnxStableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             revision="onnx",
             safety_checker=None,
             feature_extractor=None,

tests/pipelines/stable_diffusion/test_onnx_stable_diffusion_img2img.py

@@ -210,10 +210,10 @@ class OnnxStableDiffusionImg2ImgPipelineIntegrationTests(unittest.TestCase):
         )
         init_image = init_image.resize((768, 512))
         lms_scheduler = LMSDiscreteScheduler.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", subfolder="scheduler", revision="onnx"
+            "Jiali/stable-diffusion-1.5", subfolder="scheduler", revision="onnx"
         )
         pipe = OnnxStableDiffusionImg2ImgPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             revision="onnx",
             scheduler=lms_scheduler,
             safety_checker=None,

tests/pipelines/stable_diffusion/test_stable_diffusion.py

@@ -1332,7 +1332,7 @@ class StableDiffusionPipelineCkptTests(unittest.TestCase):
 
     def test_download_from_hub(self):
         ckpt_paths = [
-            "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors",
+            "https://huggingface.co/Jiali/stable-diffusion-1.5/blob/main/v1-5-pruned-emaonly.safetensors",
             "https://huggingface.co/WarriorMama777/OrangeMixs/blob/main/Models/AbyssOrangeMix/AbyssOrangeMix.safetensors",
         ]
 
@@ -1346,10 +1346,8 @@ class StableDiffusionPipelineCkptTests(unittest.TestCase):
         assert image_out.shape == (512, 512, 3)
 
     def test_download_local(self):
-        ckpt_filename = hf_hub_download(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", filename="v1-5-pruned-emaonly.safetensors"
-        )
-        config_filename = hf_hub_download("stable-diffusion-v1-5/stable-diffusion-v1-5", filename="v1-inference.yaml")
+        ckpt_filename = hf_hub_download("Jiali/stable-diffusion-1.5", filename="v1-5-pruned-emaonly.safetensors")
+        config_filename = hf_hub_download("Jiali/stable-diffusion-1.5", filename="v1-inference.yaml")
 
         pipe = StableDiffusionPipeline.from_single_file(
             ckpt_filename, config_files={"v1": config_filename}, torch_dtype=torch.float16
@@ -1404,9 +1402,7 @@ class StableDiffusionPipelineNightlyTests(unittest.TestCase):
         assert max_diff < 1e-3
 
     def test_stable_diffusion_1_5_pndm(self):
-        sd_pipe = StableDiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5").to(
-            torch_device
-        )
+        sd_pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5").to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
 
         inputs = self.get_inputs(torch_device)
@@ -1487,9 +1483,9 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
         return inputs
 
     def get_pipeline_output_without_device_map(self):
-        sd_pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16
-        ).to(torch_device)
+        sd_pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", torch_dtype=torch.float16).to(
+            torch_device
+        )
         sd_pipe.set_progress_bar_config(disable=True)
         inputs = self.get_inputs()
         no_device_map_image = sd_pipe(**inputs).images
@@ -1502,7 +1498,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
         no_device_map_image = self.get_pipeline_output_without_device_map()
 
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", device_map="balanced", torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", device_map="balanced", torch_dtype=torch.float16
         )
         sd_pipe_with_device_map.set_progress_bar_config(disable=True)
         inputs = self.get_inputs()
@@ -1513,7 +1509,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
 
     def test_components_put_in_right_devices(self):
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", device_map="balanced", torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", device_map="balanced", torch_dtype=torch.float16
         )
 
         assert len(set(sd_pipe_with_device_map.hf_device_map.values())) >= 2
@@ -1522,7 +1518,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
         no_device_map_image = self.get_pipeline_output_without_device_map()
 
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5",
+            "Jiali/stable-diffusion-1.5",
             device_map="balanced",
             max_memory={0: "1GB", 1: "1GB"},
             torch_dtype=torch.float16,
@@ -1536,7 +1532,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
 
     def test_reset_device_map(self):
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", device_map="balanced", torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", device_map="balanced", torch_dtype=torch.float16
         )
         sd_pipe_with_device_map.reset_device_map()
 
@@ -1548,7 +1544,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
 
     def test_reset_device_map_to(self):
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", device_map="balanced", torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", device_map="balanced", torch_dtype=torch.float16
         )
         sd_pipe_with_device_map.reset_device_map()
 
@@ -1560,7 +1556,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
 
     def test_reset_device_map_enable_model_cpu_offload(self):
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", device_map="balanced", torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", device_map="balanced", torch_dtype=torch.float16
         )
         sd_pipe_with_device_map.reset_device_map()
 
@@ -1572,7 +1568,7 @@ class StableDiffusionPipelineDeviceMapTests(unittest.TestCase):
 
     def test_reset_device_map_enable_sequential_cpu_offload(self):
         sd_pipe_with_device_map = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", device_map="balanced", torch_dtype=torch.float16
+            "Jiali/stable-diffusion-1.5", device_map="balanced", torch_dtype=torch.float16
         )
         sd_pipe_with_device_map.reset_device_map()
 

tests/pipelines/stable_diffusion/test_stable_diffusion_img2img.py

@@ -566,7 +566,7 @@ class StableDiffusionImg2ImgPipelineSlowTests(unittest.TestCase):
             assert module.device == torch.device("cpu")
 
     def test_img2img_2nd_order(self):
-        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe.scheduler = HeunDiscreteScheduler.from_config(sd_pipe.scheduler.config)
         sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
@@ -630,7 +630,7 @@ class StableDiffusionImg2ImgPipelineSlowTests(unittest.TestCase):
         assert np.abs(image_slice.flatten() - expected_slice).max() < 5e-3
 
     def test_img2img_safety_checker_works(self):
-        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
 
@@ -686,7 +686,7 @@ class StableDiffusionImg2ImgPipelineNightlyTests(unittest.TestCase):
         return inputs
 
     def test_img2img_pndm(self):
-        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
 
@@ -701,7 +701,7 @@ class StableDiffusionImg2ImgPipelineNightlyTests(unittest.TestCase):
         assert max_diff < 1e-3
 
     def test_img2img_ddim(self):
-        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe.scheduler = DDIMScheduler.from_config(sd_pipe.scheduler.config)
         sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
@@ -717,7 +717,7 @@ class StableDiffusionImg2ImgPipelineNightlyTests(unittest.TestCase):
         assert max_diff < 1e-3
 
     def test_img2img_lms(self):
-        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe.scheduler = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config)
         sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
@@ -733,7 +733,7 @@ class StableDiffusionImg2ImgPipelineNightlyTests(unittest.TestCase):
         assert max_diff < 1e-3
 
     def test_img2img_dpm(self):
-        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionImg2ImgPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe.scheduler = DPMSolverMultistepScheduler.from_config(sd_pipe.scheduler.config)
         sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)

tests/pipelines/stable_diffusion/test_stable_diffusion_inpaint.py

@@ -767,9 +767,7 @@ class StableDiffusionInpaintPipelineSlowTests(unittest.TestCase):
         assert np.abs(expected_slice - image_slice).max() < 1e-3
 
     def test_stable_diffusion_simple_inpaint_ddim(self):
-        pipe = StableDiffusionInpaintPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        )
+        pipe = StableDiffusionInpaintPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None)
         pipe.unet.set_default_attn_processor()
         pipe.to(torch_device)
         pipe.set_progress_bar_config(disable=None)
@@ -975,9 +973,7 @@ class StableDiffusionInpaintPipelineAsymmetricAutoencoderKLSlowTests(unittest.Te
 
     def test_stable_diffusion_simple_inpaint_ddim(self):
         vae = AsymmetricAutoencoderKL.from_pretrained("cross-attention/asymmetric-autoencoder-kl-x-1-5")
-        pipe = StableDiffusionInpaintPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        )
+        pipe = StableDiffusionInpaintPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None)
         pipe.vae = vae
         pipe.unet.set_default_attn_processor()
         pipe.to(torch_device)

tests/pipelines/stable_diffusion_adapter/test_stable_diffusion_adapter.py

@@ -609,7 +609,7 @@ class StableDiffusionAdapterPipelineSlowTests(unittest.TestCase):
 
     def test_stable_diffusion_adapter_depth_sd_v15(self):
         adapter_model = "TencentARC/t2iadapter_depth_sd15v2"
-        sd_model = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        sd_model = "Jiali/stable-diffusion-1.5"
         prompt = "desk"
         image_url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/t2i_adapter/desk_depth.png"
         input_channels = 3
@@ -636,7 +636,7 @@ class StableDiffusionAdapterPipelineSlowTests(unittest.TestCase):
 
     def test_stable_diffusion_adapter_zoedepth_sd_v15(self):
         adapter_model = "TencentARC/t2iadapter_zoedepth_sd15v1"
-        sd_model = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        sd_model = "Jiali/stable-diffusion-1.5"
         prompt = "motorcycle"
         image_url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/t2i_adapter/motorcycle.png"
         input_channels = 3
@@ -660,7 +660,7 @@ class StableDiffusionAdapterPipelineSlowTests(unittest.TestCase):
 
     def test_stable_diffusion_adapter_canny_sd_v15(self):
         adapter_model = "TencentARC/t2iadapter_canny_sd15v2"
-        sd_model = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        sd_model = "Jiali/stable-diffusion-1.5"
         prompt = "toy"
         image_url = "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/t2i_adapter/toy_canny.png"
         input_channels = 1
@@ -688,7 +688,7 @@ class StableDiffusionAdapterPipelineSlowTests(unittest.TestCase):
 
     def test_stable_diffusion_adapter_sketch_sd15(self):
         adapter_model = "TencentARC/t2iadapter_sketch_sd15v2"
-        sd_model = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        sd_model = "Jiali/stable-diffusion-1.5"
         prompt = "cat"
         image_url = (
             "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/t2i_adapter/edge.png"

tests/pipelines/stable_diffusion_safe/test_safe_diffusion.py

@@ -277,9 +277,7 @@ class SafeDiffusionPipelineIntegrationTests(unittest.TestCase):
         torch.cuda.empty_cache()
 
     def test_harm_safe_stable_diffusion(self):
-        sd_pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        )
+        sd_pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None)
         sd_pipe.scheduler = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config)
         sd_pipe = sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
@@ -340,9 +338,7 @@ class SafeDiffusionPipelineIntegrationTests(unittest.TestCase):
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
     def test_nudity_safe_stable_diffusion(self):
-        sd_pipe = StableDiffusionPipeline.from_pretrained(
-            "stable-diffusion-v1-5/stable-diffusion-v1-5", safety_checker=None
-        )
+        sd_pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5", safety_checker=None)
         sd_pipe.scheduler = LMSDiscreteScheduler.from_config(sd_pipe.scheduler.config)
         sd_pipe = sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
@@ -396,7 +392,7 @@ class SafeDiffusionPipelineIntegrationTests(unittest.TestCase):
         assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
 
     def test_nudity_safetychecker_safe_stable_diffusion(self):
-        sd_pipe = StableDiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5")
+        sd_pipe = StableDiffusionPipeline.from_pretrained("Jiali/stable-diffusion-1.5")
         sd_pipe = sd_pipe.to(torch_device)
         sd_pipe.set_progress_bar_config(disable=None)
 

tests/pipelines/test_pipelines_auto.py

@@ -40,7 +40,7 @@ from diffusers.utils.testing_utils import slow
 
 PRETRAINED_MODEL_REPO_MAPPING = OrderedDict(
     [
-        ("stable-diffusion", "stable-diffusion-v1-5/stable-diffusion-v1-5"),
+        ("stable-diffusion", "Jiali/stable-diffusion-1.5"),
         ("if", "DeepFloyd/IF-I-XL-v1.0"),
         ("kandinsky", "kandinsky-community/kandinsky-2-1"),
         ("kandinsky22", "kandinsky-community/kandinsky-2-2-decoder"),
@@ -539,7 +539,7 @@ class AutoPipelineIntegrationTest(unittest.TestCase):
 
     def test_controlnet(self):
         # test from_pretrained
-        model_repo = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        model_repo = "Jiali/stable-diffusion-1.5"
         controlnet_repo = "lllyasviel/sd-controlnet-canny"
 
         controlnet = ControlNetModel.from_pretrained(controlnet_repo, torch_dtype=torch.float16)

tests/pipelines/text_to_video_synthesis/test_text_to_video_zero.py

@@ -40,7 +40,7 @@ class TextToVideoZeroPipelineSlowTests(unittest.TestCase):
         torch.cuda.empty_cache()
 
     def test_full_model(self):
-        model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+        model_id = "Jiali/stable-diffusion-1.5"
         pipe = TextToVideoZeroPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to("cuda")
         pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
         generator = torch.Generator(device="cuda").manual_seed(0)

tests/single_file/test_stable_diffusion_controlnet_img2img_single_file.py

@@ -30,9 +30,7 @@ enable_full_determinism()
 @require_torch_gpu
 class StableDiffusionControlNetPipelineSingleFileSlowTests(unittest.TestCase, SDSingleFileTesterMixin):
     pipeline_class = StableDiffusionControlNetPipeline
-    ckpt_path = (
-        "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
-    )
+    ckpt_path = "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
     original_config = (
         "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml"
     )

tests/single_file/test_stable_diffusion_controlnet_inpaint_single_file.py

@@ -31,7 +31,7 @@ class StableDiffusionControlNetInpaintPipelineSingleFileSlowTests(unittest.TestC
     pipeline_class = StableDiffusionControlNetInpaintPipeline
     ckpt_path = "https://huggingface.co/botp/stable-diffusion-v1-5-inpainting/blob/main/sd-v1-5-inpainting.ckpt"
     original_config = "https://raw.githubusercontent.com/runwayml/stable-diffusion/main/configs/stable-diffusion/v1-inpainting-inference.yaml"
-    repo_id = "stable-diffusion-v1-5/stable-diffusion-inpainting"
+    repo_id = "botp/stable-diffusion-v1-5-inpainting"
 
     def setUp(self):
         super().setUp()

tests/single_file/test_stable_diffusion_controlnet_single_file.py

@@ -29,13 +29,11 @@ enable_full_determinism()
 @require_torch_gpu
 class StableDiffusionControlNetPipelineSingleFileSlowTests(unittest.TestCase, SDSingleFileTesterMixin):
     pipeline_class = StableDiffusionControlNetPipeline
-    ckpt_path = (
-        "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
-    )
+    ckpt_path = "https://huggingface.co/Jiali/stable-diffusion-1.5/blob/main/v1-5-pruned-emaonly.safetensors"
     original_config = (
         "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml"
     )
-    repo_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+    repo_id = "Jiali/stable-diffusion-1.5"
 
     def setUp(self):
         super().setUp()

tests/single_file/test_stable_diffusion_img2img_single_file.py

@@ -23,13 +23,11 @@ enable_full_determinism()
 @require_torch_gpu
 class StableDiffusionImg2ImgPipelineSingleFileSlowTests(unittest.TestCase, SDSingleFileTesterMixin):
     pipeline_class = StableDiffusionImg2ImgPipeline
-    ckpt_path = (
-        "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
-    )
+    ckpt_path = "https://huggingface.co/Jiali/stable-diffusion-1.5/blob/main/v1-5-pruned-emaonly.safetensors"
     original_config = (
         "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml"
     )
-    repo_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+    repo_id = "Jiali/stable-diffusion-1.5"
 
     def setUp(self):
         super().setUp()

tests/single_file/test_stable_diffusion_inpaint_single_file.py

@@ -63,7 +63,7 @@ class StableDiffusionInpaintPipelineSingleFileSlowTests(unittest.TestCase, SDSin
 
     def test_single_file_loading_4_channel_unet(self):
         # Test loading single file inpaint with a 4 channel UNet
-        ckpt_path = "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
+        ckpt_path = "https://huggingface.co/Jiali/stable-diffusion-1.5/blob/main/v1-5-pruned-emaonly.safetensors"
         pipe = self.pipeline_class.from_single_file(ckpt_path)
 
         assert pipe.unet.config.in_channels == 4

tests/single_file/test_stable_diffusion_single_file.py

@@ -26,13 +26,11 @@ enable_full_determinism()
 @require_torch_gpu
 class StableDiffusionPipelineSingleFileSlowTests(unittest.TestCase, SDSingleFileTesterMixin):
     pipeline_class = StableDiffusionPipeline
-    ckpt_path = (
-        "https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
-    )
+    ckpt_path = "https://huggingface.co/Jiali/stable-diffusion-1.5/blob/main/v1-5-pruned-emaonly.safetensors"
     original_config = (
         "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml"
     )
-    repo_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
+    repo_id = "Jiali/stable-diffusion-1.5"
 
     def setUp(self):
         super().setUp()