update

docs/source/en/_toctree.yml

@@ -244,10 +244,14 @@
   - sections:
     - local: api/pipelines/overview
       title: Overview
+    - local: api/pipelines/alt_diffusion
+      title: AltDiffusion
     - local: api/pipelines/animatediff
       title: AnimateDiff
     - local: api/pipelines/attend_and_excite
       title: Attend-and-Excite
+    - local: api/pipelines/audio_diffusion
+      title: Audio Diffusion
     - local: api/pipelines/audioldm
       title: AudioLDM
     - local: api/pipelines/audioldm2
@@ -266,6 +270,8 @@
       title: ControlNet-XS
     - local: api/pipelines/controlnetxs_sdxl
       title: ControlNet-XS with Stable Diffusion XL
+    - local: api/pipelines/cycle_diffusion
+      title: Cycle Diffusion
     - local: api/pipelines/dance_diffusion
       title: Dance Diffusion
     - local: api/pipelines/ddim
@@ -296,14 +302,26 @@
       title: MusicLDM
     - local: api/pipelines/paint_by_example
       title: Paint by Example
+    - local: api/pipelines/paradigms
+      title: Parallel Sampling of Diffusion Models
+    - local: api/pipelines/pix2pix_zero
+      title: Pix2Pix Zero
     - local: api/pipelines/pixart
       title: PixArt-α
+    - local: api/pipelines/pndm
+      title: PNDM
+    - local: api/pipelines/repaint
+      title: RePaint
+    - local: api/pipelines/score_sde_ve
+      title: Score SDE VE
     - local: api/pipelines/self_attention_guidance
       title: Self-Attention Guidance
     - local: api/pipelines/semantic_stable_diffusion
       title: Semantic Guidance
     - local: api/pipelines/shap_e
       title: Shap-E
+    - local: api/pipelines/spectrogram_diffusion
+      title: Spectrogram Diffusion
     - sections:
       - local: api/pipelines/stable_diffusion/overview
         title: Overview
@@ -338,16 +356,26 @@
       title: Stable Diffusion
     - local: api/pipelines/stable_unclip
       title: Stable unCLIP
+    - local: api/pipelines/stochastic_karras_ve
+      title: Stochastic Karras VE
+    - local: api/pipelines/model_editing
+      title: Text-to-image model editing
     - local: api/pipelines/text_to_video
       title: Text-to-video
     - local: api/pipelines/text_to_video_zero
       title: Text2Video-Zero
     - local: api/pipelines/unclip
       title: unCLIP
+    - local: api/pipelines/latent_diffusion_uncond
+      title: Unconditional Latent Diffusion
     - local: api/pipelines/unidiffuser
       title: UniDiffuser
     - local: api/pipelines/value_guided_sampling
       title: Value-guided sampling
+    - local: api/pipelines/versatile_diffusion
+      title: Versatile Diffusion
+    - local: api/pipelines/vq_diffusion
+      title: VQ Diffusion
     - local: api/pipelines/wuerstchen
       title: Wuerstchen
     title: Pipelines

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

@@ -0,0 +1,47 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# AltDiffusion
+
+AltDiffusion was proposed in [AltCLIP: Altering the Language Encoder in CLIP for Extended Language Capabilities](https://huggingface.co/papers/2211.06679) by Zhongzhi Chen, Guang Liu, Bo-Wen Zhang, Fulong Ye, Qinghong Yang, Ledell Wu.
+
+The abstract from the paper is:
+
+*In this work, we present a conceptually simple and effective method to train a strong bilingual/multilingual multimodal representation model. Starting from the pre-trained multimodal representation model CLIP released by OpenAI, we altered its text encoder with a pre-trained multilingual text encoder XLM-R, and aligned both languages and image representations by a two-stage training schema consisting of teacher learning and contrastive learning. We validate our method through evaluations of a wide range of tasks. We set new state-of-the-art performances on a bunch of tasks including ImageNet-CN, Flicker30k-CN, COCO-CN and XTD. Further, we obtain very close performances with CLIP on almost all tasks, suggesting that one can simply alter the text encoder in CLIP for extended capabilities such as multilingual understanding. Our models and code are available at [this https URL](https://github.com/FlagAI-Open/FlagAI).*
+
+## Tips
+
+`AltDiffusion` is conceptually the same as [Stable Diffusion](./stable_diffusion/overview).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## AltDiffusionPipeline
+
+[[autodoc]] AltDiffusionPipeline
+	- all
+	- __call__
+
+## AltDiffusionImg2ImgPipeline
+
+[[autodoc]] AltDiffusionImg2ImgPipeline
+	- all
+	- __call__
+
+## AltDiffusionPipelineOutput
+
+[[autodoc]] pipelines.alt_diffusion.AltDiffusionPipelineOutput
+	- all
+	- __call__

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

@@ -0,0 +1,35 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Audio Diffusion
+
+[Audio Diffusion](https://github.com/teticio/audio-diffusion) is by Robert Dargavel Smith, and it leverages the recent advances in image generation from diffusion models by converting audio samples to and from Mel spectrogram images.
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## AudioDiffusionPipeline
+[[autodoc]] AudioDiffusionPipeline
+	- all
+	- __call__
+
+## AudioPipelineOutput
+[[autodoc]] pipelines.AudioPipelineOutput
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput
+
+## Mel
+[[autodoc]] Mel

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

@@ -0,0 +1,33 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Cycle Diffusion
+
+Cycle Diffusion is a text guided image-to-image generation model proposed in [Unifying Diffusion Models' Latent Space, with Applications to CycleDiffusion and Guidance](https://huggingface.co/papers/2210.05559) by Chen Henry Wu, Fernando De la Torre.
+
+The abstract from the paper is:
+
+*Diffusion models have achieved unprecedented performance in generative modeling. The commonly-adopted formulation of the latent code of diffusion models is a sequence of gradually denoised samples, as opposed to the simpler (e.g., Gaussian) latent space of GANs, VAEs, and normalizing flows. This paper provides an alternative, Gaussian formulation of the latent space of various diffusion models, as well as an invertible DPM-Encoder that maps images into the latent space. While our formulation is purely based on the definition of diffusion models, we demonstrate several intriguing consequences. (1) Empirically, we observe that a common latent space emerges from two diffusion models trained independently on related domains. In light of this finding, we propose CycleDiffusion, which uses DPM-Encoder for unpaired image-to-image translation. Furthermore, applying CycleDiffusion to text-to-image diffusion models, we show that large-scale text-to-image diffusion models can be used as zero-shot image-to-image editors. (2) One can guide pre-trained diffusion models and GANs by controlling the latent codes in a unified, plug-and-play formulation based on energy-based models. Using the CLIP model and a face recognition model as guidance, we demonstrate that diffusion models have better coverage of low-density sub-populations and individuals than GANs. The code is publicly available at [this https URL](https://github.com/ChenWu98/cycle-diffusion).*
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## CycleDiffusionPipeline
+[[autodoc]] CycleDiffusionPipeline
+	- all
+	- __call__
+
+## StableDiffusionPiplineOutput
+[[autodoc]] pipelines.stable_diffusion.StableDiffusionPipelineOutput

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

@@ -0,0 +1,35 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Unconditional Latent Diffusion
+
+Unconditional Latent Diffusion was proposed in [High-Resolution Image Synthesis with Latent Diffusion Models](https://huggingface.co/papers/2112.10752) by Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, Björn Ommer.
+
+The abstract from the paper is:
+
+*By decomposing the image formation process into a sequential application of denoising autoencoders, diffusion models (DMs) achieve state-of-the-art synthesis results on image data and beyond. Additionally, their formulation allows for a guiding mechanism to control the image generation process without retraining. However, since these models typically operate directly in pixel space, optimization of powerful DMs often consumes hundreds of GPU days and inference is expensive due to sequential evaluations. To enable DM training on limited computational resources while retaining their quality and flexibility, we apply them in the latent space of powerful pretrained autoencoders. In contrast to previous work, training diffusion models on such a representation allows for the first time to reach a near-optimal point between complexity reduction and detail preservation, greatly boosting visual fidelity. By introducing cross-attention layers into the model architecture, we turn diffusion models into powerful and flexible generators for general conditioning inputs such as text or bounding boxes and high-resolution synthesis becomes possible in a convolutional manner. Our latent diffusion models (LDMs) achieve a new state of the art for image inpainting and highly competitive performance on various tasks, including unconditional image generation, semantic scene synthesis, and super-resolution, while significantly reducing computational requirements compared to pixel-based DMs.*
+
+The original codebase can be found at [CompVis/latent-diffusion](https://github.com/CompVis/latent-diffusion).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## LDMPipeline
+[[autodoc]] LDMPipeline
+	- all
+	- __call__
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput

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

@@ -0,0 +1,35 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Text-to-image model editing
+
+[Editing Implicit Assumptions in Text-to-Image Diffusion Models](https://huggingface.co/papers/2303.08084) is by Hadas Orgad, Bahjat Kawar, and Yonatan Belinkov. This pipeline enables editing diffusion model weights, such that its assumptions of a given concept are changed. The resulting change is expected to take effect in all prompt generations related to the edited concept.
+
+The abstract from the paper is:
+
+*Text-to-image diffusion models often make implicit assumptions about the world when generating images. While some assumptions are useful (e.g., the sky is blue), they can also be outdated, incorrect, or reflective of social biases present in the training data. Thus, there is a need to control these assumptions without requiring explicit user input or costly re-training. In this work, we aim to edit a given implicit assumption in a pre-trained diffusion model. Our Text-to-Image Model Editing method, TIME for short, receives a pair of inputs: a "source" under-specified prompt for which the model makes an implicit assumption (e.g., "a pack of roses"), and a "destination" prompt that describes the same setting, but with a specified desired attribute (e.g., "a pack of blue roses"). TIME then updates the model's cross-attention layers, as these layers assign visual meaning to textual tokens. We edit the projection matrices in these layers such that the source prompt is projected close to the destination prompt. Our method is highly efficient, as it modifies a mere 2.2% of the model's parameters in under one second. To evaluate model editing approaches, we introduce TIMED (TIME Dataset), containing 147 source and destination prompt pairs from various domains. Our experiments (using Stable Diffusion) show that TIME is successful in model editing, generalizes well for related prompts unseen during editing, and imposes minimal effect on unrelated generations.*
+
+You can find additional information about model editing on the [project page](https://time-diffusion.github.io/), [original codebase](https://github.com/bahjat-kawar/time-diffusion), and try it out in a [demo](https://huggingface.co/spaces/bahjat-kawar/time-diffusion).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## StableDiffusionModelEditingPipeline
+[[autodoc]] StableDiffusionModelEditingPipeline
+	- __call__
+	- all
+
+## StableDiffusionPipelineOutput
+[[autodoc]] pipelines.stable_diffusion.StableDiffusionPipelineOutput

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

@@ -0,0 +1,51 @@
+<!--Copyright 2023 ParaDiGMS authors and The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Parallel Sampling of Diffusion Models
+
+[Parallel Sampling of Diffusion Models](https://huggingface.co/papers/2305.16317) is by Andy Shih, Suneel Belkhale, Stefano Ermon, Dorsa Sadigh, Nima Anari.
+
+The abstract from the paper is:
+
+*Diffusion models are powerful generative models but suffer from slow sampling, often taking 1000 sequential denoising steps for one sample. As a result, considerable efforts have been directed toward reducing the number of denoising steps, but these methods hurt sample quality. Instead of reducing the number of denoising steps (trading quality for speed), in this paper we explore an orthogonal approach: can we run the denoising steps in parallel (trading compute for speed)? In spite of the sequential nature of the denoising steps, we show that surprisingly it is possible to parallelize sampling via Picard iterations, by guessing the solution of future denoising steps and iteratively refining until convergence. With this insight, we present ParaDiGMS, a novel method to accelerate the sampling of pretrained diffusion models by denoising multiple steps in parallel. ParaDiGMS is the first diffusion sampling method that enables trading compute for speed and is even compatible with existing fast sampling techniques such as DDIM and DPMSolver. Using ParaDiGMS, we improve sampling speed by 2-4x across a range of robotics and image generation models, giving state-of-the-art sampling speeds of 0.2s on 100-step DiffusionPolicy and 14.6s on 1000-step StableDiffusion-v2 with no measurable degradation of task reward, FID score, or CLIP score.*
+
+The original codebase can be found at [AndyShih12/paradigms](https://github.com/AndyShih12/paradigms), and the pipeline was contributed by [AndyShih12](https://github.com/AndyShih12). ❤️
+
+## Tips
+
+This pipeline improves sampling speed by running denoising steps in parallel, at the cost of increased total FLOPs.
+Therefore, it is better to call this pipeline when running on multiple GPUs. Otherwise, without enough GPU bandwidth
+sampling may be even slower than sequential sampling.
+
+The two parameters to play with are `parallel` (batch size) and `tolerance`.
+- If it fits in memory, for a 1000-step DDPM you can aim for a batch size of around 100 (for example, 8 GPUs and `batch_per_device=12` to get `parallel=96`). A higher batch size may not fit in memory, and lower batch size gives less parallelism.
+- For tolerance, using a higher tolerance may get better speedups but can risk sample quality degradation. If there is quality degradation with the default tolerance, then use a lower tolerance like `0.001`.
+
+For a 1000-step DDPM on 8 A100 GPUs, you can expect around a 3x speedup from [`StableDiffusionParadigmsPipeline`] compared to the [`StableDiffusionPipeline`]
+by setting `parallel=80` and `tolerance=0.1`.
+
+🤗 Diffusers offers [distributed inference support](../../training/distributed_inference) for generating multiple prompts
+in parallel on multiple GPUs. But [`StableDiffusionParadigmsPipeline`] is designed for speeding up sampling of a single prompt by using multiple GPUs.
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## StableDiffusionParadigmsPipeline
+[[autodoc]] StableDiffusionParadigmsPipeline
+	- __call__
+	- all
+
+## StableDiffusionPipelineOutput
+[[autodoc]] pipelines.stable_diffusion.StableDiffusionPipelineOutput

docs/source/en/api/pipelines/pix2pix_zero.md

@@ -0,0 +1,289 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Pix2Pix Zero
+
+[Zero-shot Image-to-Image Translation](https://huggingface.co/papers/2302.03027) is by Gaurav Parmar, Krishna Kumar Singh, Richard Zhang, Yijun Li, Jingwan Lu, and Jun-Yan Zhu.
+
+The abstract from the paper is:
+
+*Large-scale text-to-image generative models have shown their remarkable ability to synthesize diverse and high-quality images. However, it is still challenging to directly apply these models for editing real images for two reasons. First, it is hard for users to come up with a perfect text prompt that accurately describes every visual detail in the input image. Second, while existing models can introduce desirable changes in certain regions, they often dramatically alter the input content and introduce unexpected changes in unwanted regions. In this work, we propose pix2pix-zero, an image-to-image translation method that can preserve the content of the original image without manual prompting. We first automatically discover editing directions that reflect desired edits in the text embedding space. To preserve the general content structure after editing, we further propose cross-attention guidance, which aims to retain the cross-attention maps of the input image throughout the diffusion process. In addition, our method does not need additional training for these edits and can directly use the existing pre-trained text-to-image diffusion model. We conduct extensive experiments and show that our method outperforms existing and concurrent works for both real and synthetic image editing.*
+
+You can find additional information about Pix2Pix Zero on the [project page](https://pix2pixzero.github.io/),  [original codebase](https://github.com/pix2pixzero/pix2pix-zero), and try it out in a [demo](https://huggingface.co/spaces/pix2pix-zero-library/pix2pix-zero-demo).
+
+## Tips
+
+* The pipeline can be conditioned on real input images. Check out the code examples below to know more.
+* The pipeline exposes two arguments namely `source_embeds` and `target_embeds`
+that let you control the direction of the semantic edits in the final image to be generated. Let's say,
+you wanted to translate from "cat" to "dog". In this case, the edit direction will be "cat -> dog". To reflect
+this in the pipeline, you simply have to set the embeddings related to the phrases including "cat" to
+`source_embeds` and "dog" to `target_embeds`. Refer to the code example below for more details.
+* When you're using this pipeline from a prompt, specify the _source_ concept in the prompt. Taking
+the above example, a valid input prompt would be: "a high resolution painting of a **cat** in the style of van gogh".
+* If you wanted to reverse the direction in the example above, i.e., "dog -> cat", then it's recommended to:
+    * Swap the `source_embeds` and `target_embeds`.
+    * Change the input prompt to include "dog".
+* To learn more about how the source and target embeddings are generated, refer to the [original paper](https://arxiv.org/abs/2302.03027). Below, we also provide some directions on how to generate the embeddings.
+* Note that the quality of the outputs generated with this pipeline is dependent on how good the `source_embeds` and `target_embeds` are. Please, refer to [this discussion](#generating-source-and-target-embeddings) for some suggestions on the topic.
+
+## Available Pipelines:
+
+| Pipeline | Tasks | Demo
+|---|---|:---:|
+| [StableDiffusionPix2PixZeroPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_pix2pix_zero.py) | *Text-Based Image Editing* | [🤗 Space](https://huggingface.co/spaces/pix2pix-zero-library/pix2pix-zero-demo) |
+
+<!-- TODO: add Colab -->
+
+## Usage example
+
+### Based on an image generated with the input prompt
+
+```python
+import requests
+import torch
+
+from diffusers import DDIMScheduler, StableDiffusionPix2PixZeroPipeline
+
+
+def download(embedding_url, local_filepath):
+    r = requests.get(embedding_url)
+    with open(local_filepath, "wb") as f:
+        f.write(r.content)
+
+
+model_ckpt = "CompVis/stable-diffusion-v1-4"
+pipeline = StableDiffusionPix2PixZeroPipeline.from_pretrained(
+    model_ckpt, conditions_input_image=False, torch_dtype=torch.float16
+)
+pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
+pipeline.to("cuda")
+
+prompt = "a high resolution painting of a cat in the style of van gogh"
+src_embs_url = "https://github.com/pix2pixzero/pix2pix-zero/raw/main/assets/embeddings_sd_1.4/cat.pt"
+target_embs_url = "https://github.com/pix2pixzero/pix2pix-zero/raw/main/assets/embeddings_sd_1.4/dog.pt"
+
+for url in [src_embs_url, target_embs_url]:
+    download(url, url.split("/")[-1])
+
+src_embeds = torch.load(src_embs_url.split("/")[-1])
+target_embeds = torch.load(target_embs_url.split("/")[-1])
+
+image = pipeline(
+    prompt,
+    source_embeds=src_embeds,
+    target_embeds=target_embeds,
+    num_inference_steps=50,
+    cross_attention_guidance_amount=0.15,
+).images[0]
+image
+```
+
+### Based on an input image
+
+When the pipeline is conditioned on an input image, we first obtain an inverted
+noise from it using a `DDIMInverseScheduler` with the help of a generated caption. Then the inverted noise is used to start the generation process.
+
+First, let's load our pipeline:
+
+```py
+import torch
+from transformers import BlipForConditionalGeneration, BlipProcessor
+from diffusers import DDIMScheduler, DDIMInverseScheduler, StableDiffusionPix2PixZeroPipeline
+
+captioner_id = "Salesforce/blip-image-captioning-base"
+processor = BlipProcessor.from_pretrained(captioner_id)
+model = BlipForConditionalGeneration.from_pretrained(captioner_id, torch_dtype=torch.float16, low_cpu_mem_usage=True)
+
+sd_model_ckpt = "CompVis/stable-diffusion-v1-4"
+pipeline = StableDiffusionPix2PixZeroPipeline.from_pretrained(
+    sd_model_ckpt,
+    caption_generator=model,
+    caption_processor=processor,
+    torch_dtype=torch.float16,
+    safety_checker=None,
+)
+pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
+pipeline.inverse_scheduler = DDIMInverseScheduler.from_config(pipeline.scheduler.config)
+pipeline.enable_model_cpu_offload()
+```
+
+Then, we load an input image for conditioning and obtain a suitable caption for it:
+
+```py
+from diffusers.utils import load_image
+
+img_url = "https://github.com/pix2pixzero/pix2pix-zero/raw/main/assets/test_images/cats/cat_6.png"
+raw_image = load_image(url).resize((512, 512))
+caption = pipeline.generate_caption(raw_image)
+caption
+```
+
+Then we employ the generated caption and the input image to get the inverted noise:
+
+```py
+generator = torch.manual_seed(0)
+inv_latents = pipeline.invert(caption, image=raw_image, generator=generator).latents
+```
+
+Now, generate the image with edit directions:
+
+```py
+# See the "Generating source and target embeddings" section below to
+# automate the generation of these captions with a pre-trained model like Flan-T5 as explained below.
+source_prompts = ["a cat sitting on the street", "a cat playing in the field", "a face of a cat"]
+target_prompts = ["a dog sitting on the street", "a dog playing in the field", "a face of a dog"]
+
+source_embeds = pipeline.get_embeds(source_prompts, batch_size=2)
+target_embeds = pipeline.get_embeds(target_prompts, batch_size=2)
+
+
+image = pipeline(
+    caption,
+    source_embeds=source_embeds,
+    target_embeds=target_embeds,
+    num_inference_steps=50,
+    cross_attention_guidance_amount=0.15,
+    generator=generator,
+    latents=inv_latents,
+    negative_prompt=caption,
+).images[0]
+image
+```
+
+## Generating source and target embeddings
+
+The authors originally used the [GPT-3 API](https://openai.com/api/) to generate the source and target captions for discovering
+edit directions. However, we can also leverage open source and public models for the same purpose.
+Below, we provide an end-to-end example with the [Flan-T5](https://huggingface.co/docs/transformers/model_doc/flan-t5) model
+for generating captions and [CLIP](https://huggingface.co/docs/transformers/model_doc/clip) for
+computing embeddings on the generated captions.
+
+**1. Load the generation model**:
+
+```py
+import torch
+from transformers import AutoTokenizer, T5ForConditionalGeneration
+
+tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-xl")
+model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-xl", device_map="auto", torch_dtype=torch.float16)
+```
+
+**2. Construct a starting prompt**:
+
+```py
+source_concept = "cat"
+target_concept = "dog"
+
+source_text = f"Provide a caption for images containing a {source_concept}. "
+"The captions should be in English and should be no longer than 150 characters."
+
+target_text = f"Provide a caption for images containing a {target_concept}. "
+"The captions should be in English and should be no longer than 150 characters."
+```
+
+Here, we're interested in the "cat -> dog" direction.
+
+**3. Generate captions**:
+
+We can use a utility like so for this purpose.
+
+```py
+def generate_captions(input_prompt):
+    input_ids = tokenizer(input_prompt, return_tensors="pt").input_ids.to("cuda")
+
+    outputs = model.generate(
+        input_ids, temperature=0.8, num_return_sequences=16, do_sample=True, max_new_tokens=128, top_k=10
+    )
+    return tokenizer.batch_decode(outputs, skip_special_tokens=True)
+```
+
+And then we just call it to generate our captions:
+
+```py
+source_captions = generate_captions(source_text)
+target_captions = generate_captions(target_concept)
+print(source_captions, target_captions, sep='\n')
+```
+
+We encourage you to play around with the different parameters supported by the
+`generate()` method ([documentation](https://huggingface.co/docs/transformers/main/en/main_classes/text_generation#transformers.generation_tf_utils.TFGenerationMixin.generate)) for the generation quality you are looking for.
+
+**4. Load the embedding model**:
+
+Here, we need to use the same text encoder model used by the subsequent Stable Diffusion model.
+
+```py
+from diffusers import StableDiffusionPix2PixZeroPipeline
+
+pipeline = StableDiffusionPix2PixZeroPipeline.from_pretrained(
+    "CompVis/stable-diffusion-v1-4", torch_dtype=torch.float16
+)
+pipeline = pipeline.to("cuda")
+tokenizer = pipeline.tokenizer
+text_encoder = pipeline.text_encoder
+```
+
+**5. Compute embeddings**:
+
+```py
+import torch
+
+def embed_captions(sentences, tokenizer, text_encoder, device="cuda"):
+    with torch.no_grad():
+        embeddings = []
+        for sent in sentences:
+            text_inputs = tokenizer(
+                sent,
+                padding="max_length",
+                max_length=tokenizer.model_max_length,
+                truncation=True,
+                return_tensors="pt",
+            )
+            text_input_ids = text_inputs.input_ids
+            prompt_embeds = text_encoder(text_input_ids.to(device), attention_mask=None)[0]
+            embeddings.append(prompt_embeds)
+    return torch.concatenate(embeddings, dim=0).mean(dim=0).unsqueeze(0)
+
+source_embeddings = embed_captions(source_captions, tokenizer, text_encoder)
+target_embeddings = embed_captions(target_captions, tokenizer, text_encoder)
+```
+
+And you're done! [Here](https://colab.research.google.com/drive/1tz2C1EdfZYAPlzXXbTnf-5PRBiR8_R1F?usp=sharing) is a Colab Notebook that you can use to interact with the entire process.
+
+Now, you can use these embeddings directly while calling the pipeline:
+
+```py
+from diffusers import DDIMScheduler
+
+pipeline.scheduler = DDIMScheduler.from_config(pipeline.scheduler.config)
+
+image = pipeline(
+    prompt,
+    source_embeds=source_embeddings,
+    target_embeds=target_embeddings,
+    num_inference_steps=50,
+    cross_attention_guidance_amount=0.15,
+).images[0]
+image
+```
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## StableDiffusionPix2PixZeroPipeline
+[[autodoc]] StableDiffusionPix2PixZeroPipeline
+	- __call__
+	- all

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

@@ -0,0 +1,35 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# PNDM
+
+[Pseudo Numerical Methods for Diffusion Models on Manifolds](https://huggingface.co/papers/2202.09778) (PNDM) is by Luping Liu, Yi Ren, Zhijie Lin and Zhou Zhao.
+
+The abstract from the paper is:
+
+*Denoising Diffusion Probabilistic Models (DDPMs) can generate high-quality samples such as image and audio samples. However, DDPMs require hundreds to thousands of iterations to produce final samples. Several prior works have successfully accelerated DDPMs through adjusting the variance schedule (e.g., Improved Denoising Diffusion Probabilistic Models) or the denoising equation (e.g., Denoising Diffusion Implicit Models (DDIMs)). However, these acceleration methods cannot maintain the quality of samples and even introduce new noise at a high speedup rate, which limit their practicability. To accelerate the inference process while keeping the sample quality, we provide a fresh perspective that DDPMs should be treated as solving differential equations on manifolds. Under such a perspective, we propose pseudo numerical methods for diffusion models (PNDMs). Specifically, we figure out how to solve differential equations on manifolds and show that DDIMs are simple cases of pseudo numerical methods. We change several classical numerical methods to corresponding pseudo numerical methods and find that the pseudo linear multi-step method is the best in most situations. According to our experiments, by directly using pre-trained models on Cifar10, CelebA and LSUN, PNDMs can generate higher quality synthetic images with only 50 steps compared with 1000-step DDIMs (20x speedup), significantly outperform DDIMs with 250 steps (by around 0.4 in FID) and have good generalization on different variance schedules.*
+
+The original codebase can be found at [luping-liu/PNDM](https://github.com/luping-liu/PNDM).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## PNDMPipeline
+[[autodoc]] PNDMPipeline
+	- all
+	- __call__
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput

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

@@ -0,0 +1,37 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# RePaint
+
+[RePaint: Inpainting using Denoising Diffusion Probabilistic Models](https://huggingface.co/papers/2201.09865) is by Andreas Lugmayr, Martin Danelljan, Andres Romero, Fisher Yu, Radu Timofte, Luc Van Gool.
+
+The abstract from the paper is:
+
+*Free-form inpainting is the task of adding new content to an image in the regions specified by an arbitrary binary mask. Most existing approaches train for a certain distribution of masks, which limits their generalization capabilities to unseen mask types. Furthermore, training with pixel-wise and perceptual losses often leads to simple textural extensions towards the missing areas instead of semantically meaningful generation. In this work, we propose RePaint: A Denoising Diffusion Probabilistic Model (DDPM) based inpainting approach that is applicable to even extreme masks. We employ a pretrained unconditional DDPM as the generative prior. To condition the generation process, we only alter the reverse diffusion iterations by sampling the unmasked regions using the given image information. Since this technique does not modify or condition the original DDPM network itself, the model produces high-quality and diverse output images for any inpainting form. We validate our method for both faces and general-purpose image inpainting using standard and extreme masks.
+RePaint outperforms state-of-the-art Autoregressive, and GAN approaches for at least five out of six mask distributions.*
+
+The original codebase can be found at [andreas128/RePaint](https://github.com/andreas128/RePaint).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+
+## RePaintPipeline
+[[autodoc]] RePaintPipeline
+	- all
+	- __call__
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput

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

@@ -0,0 +1,35 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Score SDE VE
+
+[Score-Based Generative Modeling through Stochastic Differential Equations](https://huggingface.co/papers/2011.13456) (Score SDE) is by Yang Song, Jascha Sohl-Dickstein, Diederik P. Kingma, Abhishek Kumar, Stefano Ermon and Ben Poole. This pipeline implements the variance expanding (VE) variant of the stochastic differential equation method.
+
+The abstract from the paper is:
+
+*Creating noise from data is easy; creating data from noise is generative modeling. We present a stochastic differential equation (SDE) that smoothly transforms a complex data distribution to a known prior distribution by slowly injecting noise, and a corresponding reverse-time SDE that transforms the prior distribution back into the data distribution by slowly removing the noise. Crucially, the reverse-time SDE depends only on the time-dependent gradient field (\aka, score) of the perturbed data distribution. By leveraging advances in score-based generative modeling, we can accurately estimate these scores with neural networks, and use numerical SDE solvers to generate samples. We show that this framework encapsulates previous approaches in score-based generative modeling and diffusion probabilistic modeling, allowing for new sampling procedures and new modeling capabilities. In particular, we introduce a predictor-corrector framework to correct errors in the evolution of the discretized reverse-time SDE. We also derive an equivalent neural ODE that samples from the same distribution as the SDE, but additionally enables exact likelihood computation, and improved sampling efficiency. In addition, we provide a new way to solve inverse problems with score-based models, as demonstrated with experiments on class-conditional generation, image inpainting, and colorization. Combined with multiple architectural improvements, we achieve record-breaking performance for unconditional image generation on CIFAR-10 with an Inception score of 9.89 and FID of 2.20, a competitive likelihood of 2.99 bits/dim, and demonstrate high fidelity generation of 1024 x 1024 images for the first time from a score-based generative model.*
+
+The original codebase can be found at [yang-song/score_sde_pytorch](https://github.com/yang-song/score_sde_pytorch).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## ScoreSdeVePipeline
+[[autodoc]] ScoreSdeVePipeline
+	- all
+	- __call__
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput

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

@@ -0,0 +1,37 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Spectrogram Diffusion
+
+[Spectrogram Diffusion](https://huggingface.co/papers/2206.05408) is by Curtis Hawthorne, Ian Simon, Adam Roberts, Neil Zeghidour, Josh Gardner, Ethan Manilow, and Jesse Engel.
+
+*An ideal music synthesizer should be both interactive and expressive, generating high-fidelity audio in realtime for arbitrary combinations of instruments and notes. Recent neural synthesizers have exhibited a tradeoff between domain-specific models that offer detailed control of only specific instruments, or raw waveform models that can train on any music but with minimal control and slow generation. In this work, we focus on a middle ground of neural synthesizers that can generate audio from MIDI sequences with arbitrary combinations of instruments in realtime. This enables training on a wide range of transcription datasets with a single model, which in turn offers note-level control of composition and instrumentation across a wide range of instruments. We use a simple two-stage process: MIDI to spectrograms with an encoder-decoder Transformer, then spectrograms to audio with a generative adversarial network (GAN) spectrogram inverter. We compare training the decoder as an autoregressive model and as a Denoising Diffusion Probabilistic Model (DDPM) and find that the DDPM approach is superior both qualitatively and as measured by audio reconstruction and Fréchet distance metrics. Given the interactivity and generality of this approach, we find this to be a promising first step towards interactive and expressive neural synthesis for arbitrary combinations of instruments and notes.*
+
+The original codebase can be found at [magenta/music-spectrogram-diffusion](https://github.com/magenta/music-spectrogram-diffusion).
+
+![img](https://storage.googleapis.com/music-synthesis-with-spectrogram-diffusion/architecture.png)
+
+As depicted above the model takes as input a MIDI file and tokenizes it into a sequence of 5 second intervals. Each tokenized interval then together with positional encodings is passed through the Note Encoder and its representation is concatenated with the previous window's generated spectrogram representation obtained via the Context Encoder. For the initial 5 second window this is set to zero. The resulting context is then used as conditioning to sample the denoised Spectrogram from the MIDI window and we concatenate this spectrogram to the final output as well as use it for the context of the next MIDI window. The process repeats till we have gone over all the MIDI inputs. Finally a MelGAN decoder converts the potentially long spectrogram to audio which is the final result of this pipeline.
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## SpectrogramDiffusionPipeline
+[[autodoc]] SpectrogramDiffusionPipeline
+	- all
+	- __call__
+
+## AudioPipelineOutput
+[[autodoc]] pipelines.AudioPipelineOutput

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

@@ -0,0 +1,33 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Stochastic Karras VE
+
+[Elucidating the Design Space of Diffusion-Based Generative Models](https://huggingface.co/papers/2206.00364) is by Tero Karras, Miika Aittala, Timo Aila and Samuli Laine. This pipeline implements the stochastic sampling tailored to variance expanding (VE) models.
+
+The abstract from the paper:
+
+*We argue that the theory and practice of diffusion-based generative models are currently unnecessarily convoluted and seek to remedy the situation by presenting a design space that clearly separates the concrete design choices. This lets us identify several changes to both the sampling and training processes, as well as preconditioning of the score networks. Together, our improvements yield new state-of-the-art FID of 1.79 for CIFAR-10 in a class-conditional setting and 1.97 in an unconditional setting, with much faster sampling (35 network evaluations per image) than prior designs. To further demonstrate their modular nature, we show that our design changes dramatically improve both the efficiency and quality obtainable with pre-trained score networks from previous work, including improving the FID of a previously trained ImageNet-64 model from 2.07 to near-SOTA 1.55, and after re-training with our proposed improvements to a new SOTA of 1.36.*
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## KarrasVePipeline
+[[autodoc]] KarrasVePipeline
+	- all
+	- __call__
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput

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

@@ -0,0 +1,54 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Versatile Diffusion
+
+Versatile Diffusion was proposed in [Versatile Diffusion: Text, Images and Variations All in One Diffusion Model](https://huggingface.co/papers/2211.08332) by Xingqian Xu, Zhangyang Wang, Eric Zhang, Kai Wang, Humphrey Shi.
+
+The abstract from the paper is:
+
+*Recent advances in diffusion models have set an impressive milestone in many generation tasks, and trending works such as DALL-E2, Imagen, and Stable Diffusion have attracted great interest. Despite the rapid landscape changes, recent new approaches focus on extensions and performance rather than capacity, thus requiring separate models for separate tasks. In this work, we expand the existing single-flow diffusion pipeline into a multi-task multimodal network, dubbed Versatile Diffusion (VD), that handles multiple flows of text-to-image, image-to-text, and variations in one unified model. The pipeline design of VD instantiates a unified multi-flow diffusion framework, consisting of sharable and swappable layer modules that enable the crossmodal generality beyond images and text. Through extensive experiments, we demonstrate that VD successfully achieves the following: a) VD outperforms the baseline approaches and handles all its base tasks with competitive quality; b) VD enables novel extensions such as disentanglement of style and semantics, dual- and multi-context blending, etc.; c) The success of our multi-flow multimodal framework over images and text may inspire further diffusion-based universal AI research.*
+
+## Tips
+
+You can load the more memory intensive "all-in-one" [`VersatileDiffusionPipeline`] that supports all the tasks or use the individual pipelines which are more memory efficient.
+
+| **Pipeline**                                         | **Supported tasks**               |
+|------------------------------------------------------|-----------------------------------|
+| [`VersatileDiffusionPipeline`]                       | all of the below                  |
+| [`VersatileDiffusionTextToImagePipeline`]            | text-to-image                     |
+| [`VersatileDiffusionImageVariationPipeline`]         | image variation                   |
+| [`VersatileDiffusionDualGuidedPipeline`]             | image-text dual guided generation |
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## VersatileDiffusionPipeline
+[[autodoc]] VersatileDiffusionPipeline
+
+## VersatileDiffusionTextToImagePipeline
+[[autodoc]] VersatileDiffusionTextToImagePipeline
+	- all
+	- __call__
+
+## VersatileDiffusionImageVariationPipeline
+[[autodoc]] VersatileDiffusionImageVariationPipeline
+	- all
+	- __call__
+
+## VersatileDiffusionDualGuidedPipeline
+[[autodoc]] VersatileDiffusionDualGuidedPipeline
+	- all
+	- __call__

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

@@ -0,0 +1,35 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# VQ Diffusion
+
+[Vector Quantized Diffusion Model for Text-to-Image Synthesis](https://huggingface.co/papers/2111.14822) is by Shuyang Gu, Dong Chen, Jianmin Bao, Fang Wen, Bo Zhang, Dongdong Chen, Lu Yuan, Baining Guo.
+
+The abstract from the paper is:
+
+*We present the vector quantized diffusion (VQ-Diffusion) model for text-to-image generation. This method is based on a vector quantized variational autoencoder (VQ-VAE) whose latent space is modeled by a conditional variant of the recently developed Denoising Diffusion Probabilistic Model (DDPM). We find that this latent-space method is well-suited for text-to-image generation tasks because it not only eliminates the unidirectional bias with existing methods but also allows us to incorporate a mask-and-replace diffusion strategy to avoid the accumulation of errors, which is a serious problem with existing methods. Our experiments show that the VQ-Diffusion produces significantly better text-to-image generation results when compared with conventional autoregressive (AR) models with similar numbers of parameters. Compared with previous GAN-based text-to-image methods, our VQ-Diffusion can handle more complex scenes and improve the synthesized image quality by a large margin. Finally, we show that the image generation computation in our method can be made highly efficient by reparameterization. With traditional AR methods, the text-to-image generation time increases linearly with the output image resolution and hence is quite time consuming even for normal size images. The VQ-Diffusion allows us to achieve a better trade-off between quality and speed. Our experiments indicate that the VQ-Diffusion model with the reparameterization is fifteen times faster than traditional AR methods while achieving a better image quality.*
+
+The original codebase can be found at [microsoft/VQ-Diffusion](https://github.com/microsoft/VQ-Diffusion).
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+
+</Tip>
+
+## VQDiffusionPipeline
+[[autodoc]] VQDiffusionPipeline
+	- all
+	- __call__
+
+## ImagePipelineOutput
+[[autodoc]] pipelines.ImagePipelineOutput

docs/source/en/training/t2i_adapters.md

@@ -224,4 +224,4 @@ image.save("./output.png")
 
 Congratulations on training a T2I-Adapter model! 🎉 To learn more:
 
-- Read the [Efficient Controllable Generation for SDXL with T2I-Adapters](https://huggingface.co/blog/t2i-sdxl-adapters) blog post to learn more details about the experimental results from the T2I-Adapter team.
+- Read the [Efficient Controllable Generation for SDXL with T2I-Adapters](https://www.cs.cmu.edu/~custom-diffusion/) blog post to learn more details about the experimental results from the T2I-Adapter team.

examples/community/README.md

@@ -41,7 +41,7 @@ If a community doesn't work as expected, please open an issue and ping the autho
 | TensorRT Stable Diffusion Inpainting Pipeline                                                                                                    | Accelerates the Stable Diffusion Inpainting Pipeline using TensorRT                                                                                                                                                                                                                                                                                                                                                                                                                                      | [TensorRT Stable Diffusion Inpainting Pipeline](#tensorrt-inpainting-stable-diffusion-pipeline)      | - |              [Asfiya Baig](https://github.com/asfiyab-nvidia) |
 |   IADB Pipeline                                                                                                    | Implementation of [Iterative α-(de)Blending: a Minimalist Deterministic Diffusion Model](https://arxiv.org/abs/2305.03486)                                                                                                                                                                                                                                                                                                                                                                                                                                      | [IADB Pipeline](#iadb-pipeline)      | - |              [Thomas Chambon](https://github.com/tchambon)
 |   Zero1to3 Pipeline                                                                                                    | Implementation of [Zero-1-to-3: Zero-shot One Image to 3D Object](https://arxiv.org/abs/2303.11328)                                                                                                                                                                                                                                                                                                                                                                                                                                      | [Zero1to3 Pipeline](#Zero1to3-pipeline)      | - |              [Xin Kong](https://github.com/kxhit) |
-| Stable Diffusion XL Long Weighted Prompt Pipeline | A pipeline support unlimited length of prompt and negative prompt, use A1111 style of prompt weighting | [Stable Diffusion XL Long Weighted Prompt Pipeline](#stable-diffusion-xl-long-weighted-prompt-pipeline) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1LsqilswLR40XLLcp6XFOl5nKb_wOe26W?usp=sharing) | [Andrew Zhu](https://xhinker.medium.com/) |
+Stable Diffusion XL Long Weighted Prompt Pipeline | A pipeline support unlimited length of prompt and negative prompt, use A1111 style of prompt weighting | [Stable Diffusion XL Long Weighted Prompt Pipeline](#stable-diffusion-xl-long-weighted-prompt-pipeline) | - | [Andrew Zhu](https://xhinker.medium.com/) |
 FABRIC - Stable Diffusion with feedback Pipeline | pipeline supports feedback from liked and disliked images | [Stable Diffusion Fabric Pipeline](#stable-diffusion-fabric-pipeline) | - | [Shauray Singh](https://shauray8.github.io/about_shauray/) |
 sketch inpaint - Inpainting with non-inpaint Stable Diffusion | sketch inpaint much like in automatic1111 | [Masked Im2Im Stable Diffusion Pipeline](#stable-diffusion-masked-im2im) | - | [Anatoly Belikov](https://github.com/noskill) |
 prompt-to-prompt | change parts of a prompt and retain image structure (see [paper page](https://prompt-to-prompt.github.io/)) | [Prompt2Prompt Pipeline](#prompt2prompt-pipeline) | - | [Umer H. Adil](https://twitter.com/UmerHAdil) |
@@ -1619,11 +1619,10 @@ This approach is using (optional) CoCa model to avoid writing image description.
 
 This SDXL pipeline support unlimited length prompt and negative prompt, compatible with A1111 prompt weighted style.
 
-You can provide both `prompt` and `prompt_2`. If only one prompt is provided, `prompt_2` will be a copy of the provided `prompt`. Here is a sample code to use this pipeline. 
+You can provide both `prompt` and `prompt_2`. if only one prompt is provided, `prompt_2` will be a copy of the provided `prompt`. Here is a sample code to use this pipeline.
 
 ```python
 from diffusers import DiffusionPipeline
-from diffusers.utils import load_image
 import torch
 
 pipe = DiffusionPipeline.from_pretrained(
@@ -1634,52 +1633,25 @@ pipe = DiffusionPipeline.from_pretrained(
     , custom_pipeline   = "lpw_stable_diffusion_xl",
 )
 
-prompt = "photo of a cute (white) cat running on the grass" * 20
-prompt2 = "chasing (birds:1.5)" * 20
+prompt = "photo of a cute (white) cat running on the grass"*20
+prompt2 = "chasing (birds:1.5)"*20
 prompt = f"{prompt},{prompt2}"
 neg_prompt = "blur, low quality, carton, animate"
 
 pipe.to("cuda")
-
-# text2img
-t2i_images = pipe(
-    prompt=prompt,
-    negative_prompt=neg_prompt,
-).images # alternatively, you can call the .text2img() function
-
-# img2img
-input_image = load_image("/path/to/local/image.png") # or URL to your input image
-i2i_images = pipe.img2img(
-  prompt=prompt,
-  negative_prompt=neg_prompt,
-  image=input_image,
-  strength=0.8, # higher strength will result in more variation compared to original image
-).images
-
-# inpaint
-input_mask = load_image("/path/to/local/mask.png") # or URL to your input inpainting mask
-inpaint_images = pipe.inpaint(
-  prompt="photo of a cute (black) cat running on the grass" * 20,
-  negative_prompt=neg_prompt,
-  image=input_image,
-  mask=input_mask,
-  strength=0.6, # higher strength will result in more variation compared to original image
-).images
+images = pipe(
+    prompt                  = prompt
+    , negative_prompt       = neg_prompt
+).images[0]
 
 pipe.to("cpu")
 torch.cuda.empty_cache()
-
-from IPython.display import display # assuming you are using this code in a notebook
-display(t2i_images[0])
-display(i2i_images[0])
-display(inpaint_images[0])
+images
 ```
 
 In the above code, the `prompt2` is appended to the `prompt`, which is more than 77 tokens. "birds" are showing up in the result.
 ![Stable Diffusion XL Long Weighted Prompt Pipeline sample](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_long_weighted_prompt.png)
 
-For more results, checkout [PR #6114](https://github.com/huggingface/diffusers/pull/6114).
-
 ## Example Images Mixing (with CoCa)
 ```python
 import requests

examples/community/lpw_stable_diffusion_xl.py

@@ -11,11 +11,10 @@ import os
 from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
 import torch
-from PIL import Image
 from transformers import CLIPTextModel, CLIPTextModelWithProjection, CLIPTokenizer
 
 from diffusers import DiffusionPipeline, StableDiffusionXLPipeline
-from diffusers.image_processor import PipelineImageInput, VaeImageProcessor
+from diffusers.image_processor import VaeImageProcessor
 from diffusers.loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
 from diffusers.models import AutoencoderKL, UNet2DConditionModel
 from diffusers.models.attention_processor import (
@@ -24,7 +23,7 @@ from diffusers.models.attention_processor import (
     LoRAXFormersAttnProcessor,
     XFormersAttnProcessor,
 )
-from diffusers.pipelines.stable_diffusion_xl.pipeline_output import StableDiffusionXLPipelineOutput
+from diffusers.pipelines.stable_diffusion_xl import StableDiffusionXLPipelineOutput
 from diffusers.schedulers import KarrasDiffusionSchedulers
 from diffusers.utils import (
     is_accelerate_available,
@@ -462,65 +461,6 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
     return noise_cfg
 
 
-# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.retrieve_latents
-def retrieve_latents(
-    encoder_output: torch.Tensor, generator: Optional[torch.Generator] = None, sample_mode: str = "sample"
-):
-    if hasattr(encoder_output, "latent_dist") and sample_mode == "sample":
-        return encoder_output.latent_dist.sample(generator)
-    elif hasattr(encoder_output, "latent_dist") and sample_mode == "argmax":
-        return encoder_output.latent_dist.mode()
-    elif hasattr(encoder_output, "latents"):
-        return encoder_output.latents
-    else:
-        raise AttributeError("Could not access latents of provided encoder_output")
-
-
-# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.retrieve_timesteps
-def retrieve_timesteps(
-    scheduler,
-    num_inference_steps: Optional[int] = None,
-    device: Optional[Union[str, torch.device]] = None,
-    timesteps: Optional[List[int]] = None,
-    **kwargs,
-):
-    """
-    Calls the scheduler's `set_timesteps` method and retrieves timesteps from the scheduler after the call. Handles
-    custom timesteps. Any kwargs will be supplied to `scheduler.set_timesteps`.
-
-    Args:
-        scheduler (`SchedulerMixin`):
-            The scheduler to get timesteps from.
-        num_inference_steps (`int`):
-            The number of diffusion steps used when generating samples with a pre-trained model. If used,
-            `timesteps` must be `None`.
-        device (`str` or `torch.device`, *optional*):
-            The device to which the timesteps should be moved to. If `None`, the timesteps are not moved.
-        timesteps (`List[int]`, *optional*):
-                Custom timesteps used to support arbitrary spacing between timesteps. If `None`, then the default
-                timestep spacing strategy of the scheduler is used. If `timesteps` is passed, `num_inference_steps`
-                must be `None`.
-
-    Returns:
-        `Tuple[torch.Tensor, int]`: A tuple where the first element is the timestep schedule from the scheduler and the
-        second element is the number of inference steps.
-    """
-    if timesteps is not None:
-        accepts_timesteps = "timesteps" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
-        if not accepts_timesteps:
-            raise ValueError(
-                f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
-                f" timestep schedules. Please check whether you are using the correct scheduler."
-            )
-        scheduler.set_timesteps(timesteps=timesteps, device=device, **kwargs)
-        timesteps = scheduler.timesteps
-        num_inference_steps = len(timesteps)
-    else:
-        scheduler.set_timesteps(num_inference_steps, device=device, **kwargs)
-        timesteps = scheduler.timesteps
-    return timesteps, num_inference_steps
-
-
 class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, LoraLoaderMixin):
     r"""
     Pipeline for text-to-image generation using Stable Diffusion XL.
@@ -586,9 +526,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
         self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
         self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
         self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
-        self.mask_processor = VaeImageProcessor(
-            vae_scale_factor=self.vae_scale_factor, do_normalize=False, do_binarize=True, do_convert_grayscale=True
-        )
         self.default_sample_size = self.unet.config.sample_size
 
         add_watermarker = add_watermarker if add_watermarker is not None else is_invisible_watermark_available()
@@ -876,7 +813,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
         prompt_2,
         height,
         width,
-        strength,
         callback_steps,
         negative_prompt=None,
         negative_prompt_2=None,
@@ -888,9 +824,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
         if height % 8 != 0 or width % 8 != 0:
             raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
 
-        if strength < 0 or strength > 1:
-            raise ValueError(f"The value of strength should in [0.0, 1.0] but is {strength}")
-
         if (callback_steps is None) or (
             callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0)
         ):
@@ -947,263 +880,23 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
                 "If `negative_prompt_embeds` are provided, `negative_pooled_prompt_embeds` also have to be passed. Make sure to generate `negative_pooled_prompt_embeds` from the same text encoder that was used to generate `negative_prompt_embeds`."
             )
 
-    def get_timesteps(self, num_inference_steps, strength, device, denoising_start=None):
-        # get the original timestep using init_timestep
-        if denoising_start is None:
-            init_timestep = min(int(num_inference_steps * strength), num_inference_steps)
-            t_start = max(num_inference_steps - init_timestep, 0)
-        else:
-            t_start = 0
-
-        timesteps = self.scheduler.timesteps[t_start * self.scheduler.order :]
-
-        # Strength is irrelevant if we directly request a timestep to start at;
-        # that is, strength is determined by the denoising_start instead.
-        if denoising_start is not None:
-            discrete_timestep_cutoff = int(
-                round(
-                    self.scheduler.config.num_train_timesteps
-                    - (denoising_start * self.scheduler.config.num_train_timesteps)
-                )
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
+    def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
+        shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
+        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_inference_steps = (timesteps < discrete_timestep_cutoff).sum().item()
-            if self.scheduler.order == 2 and num_inference_steps % 2 == 0:
-                # if the scheduler is a 2nd order scheduler we might have to do +1
-                # because `num_inference_steps` might be even given that every timestep
-                # (except the highest one) is duplicated. If `num_inference_steps` is even it would
-                # mean that we cut the timesteps in the middle of the denoising step
-                # (between 1st and 2nd devirative) which leads to incorrect results. By adding 1
-                # we ensure that the denoising process always ends after the 2nd derivate step of the scheduler
-                num_inference_steps = num_inference_steps + 1
-
-            # because t_n+1 >= t_n, we slice the timesteps starting from the end
-            timesteps = timesteps[-num_inference_steps:]
-            return timesteps, num_inference_steps
-
-        return timesteps, num_inference_steps - t_start
-
-    def prepare_latents(
-        self,
-        image,
-        mask,
-        width,
-        height,
-        num_channels_latents,
-        timestep,
-        batch_size,
-        num_images_per_prompt,
-        dtype,
-        device,
-        generator=None,
-        add_noise=True,
-        latents=None,
-        is_strength_max=True,
-        return_noise=False,
-        return_image_latents=False,
-    ):
-        batch_size *= num_images_per_prompt
-
-        if image is None:
-            shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
-            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)
-            else:
-                latents = latents.to(device)
-
-            # scale the initial noise by the standard deviation required by the scheduler
-            latents = latents * self.scheduler.init_noise_sigma
-            return latents
-
-        elif mask is None:
-            if not isinstance(image, (torch.Tensor, Image.Image, list)):
-                raise ValueError(
-                    f"`image` has to be of type `torch.Tensor`, `PIL.Image.Image` or list but is {type(image)}"
-                )
-
-            # Offload text encoder if `enable_model_cpu_offload` was enabled
-            if hasattr(self, "final_offload_hook") and self.final_offload_hook is not None:
-                self.text_encoder_2.to("cpu")
-                torch.cuda.empty_cache()
-
-            image = image.to(device=device, dtype=dtype)
-
-            if image.shape[1] == 4:
-                init_latents = image
-
-            else:
-                # make sure the VAE is in float32 mode, as it overflows in float16
-                if self.vae.config.force_upcast:
-                    image = image.float()
-                    self.vae.to(dtype=torch.float32)
-
-                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."
-                    )
-
-                elif isinstance(generator, list):
-                    init_latents = [
-                        retrieve_latents(self.vae.encode(image[i : i + 1]), generator=generator[i])
-                        for i in range(batch_size)
-                    ]
-                    init_latents = torch.cat(init_latents, dim=0)
-                else:
-                    init_latents = retrieve_latents(self.vae.encode(image), generator=generator)
-
-                if self.vae.config.force_upcast:
-                    self.vae.to(dtype)
-
-                init_latents = init_latents.to(dtype)
-                init_latents = self.vae.config.scaling_factor * init_latents
-
-            if batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] == 0:
-                # expand init_latents for batch_size
-                additional_image_per_prompt = batch_size // init_latents.shape[0]
-                init_latents = torch.cat([init_latents] * additional_image_per_prompt, dim=0)
-            elif batch_size > init_latents.shape[0] and batch_size % init_latents.shape[0] != 0:
-                raise ValueError(
-                    f"Cannot duplicate `image` of batch size {init_latents.shape[0]} to {batch_size} text prompts."
-                )
-            else:
-                init_latents = torch.cat([init_latents], dim=0)
-
-            if add_noise:
-                shape = init_latents.shape
-                noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
-                # get latents
-                init_latents = self.scheduler.add_noise(init_latents, noise, timestep)
-
-            latents = init_latents
-            return latents
-
+        if latents is None:
+            latents = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
         else:
-            shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
-            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."
-                )
+            latents = latents.to(device)
 
-            if (image is None or timestep is None) and not is_strength_max:
-                raise ValueError(
-                    "Since strength < 1. initial latents are to be initialised as a combination of Image + Noise."
-                    "However, either the image or the noise timestep has not been provided."
-                )
-
-            if image.shape[1] == 4:
-                image_latents = image.to(device=device, dtype=dtype)
-                image_latents = image_latents.repeat(batch_size // image_latents.shape[0], 1, 1, 1)
-            elif return_image_latents or (latents is None and not is_strength_max):
-                image = image.to(device=device, dtype=dtype)
-                image_latents = self._encode_vae_image(image=image, generator=generator)
-                image_latents = image_latents.repeat(batch_size // image_latents.shape[0], 1, 1, 1)
-
-            if latents is None and add_noise:
-                noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
-                # if strength is 1. then initialise the latents to noise, else initial to image + noise
-                latents = noise if is_strength_max else self.scheduler.add_noise(image_latents, noise, timestep)
-                # if pure noise then scale the initial latents by the  Scheduler's init sigma
-                latents = latents * self.scheduler.init_noise_sigma if is_strength_max else latents
-            elif add_noise:
-                noise = latents.to(device)
-                latents = noise * self.scheduler.init_noise_sigma
-            else:
-                noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
-                latents = image_latents.to(device)
-
-            outputs = (latents,)
-
-            if return_noise:
-                outputs += (noise,)
-
-            if return_image_latents:
-                outputs += (image_latents,)
-
-            return outputs
-
-    def _encode_vae_image(self, image: torch.Tensor, generator: torch.Generator):
-        dtype = image.dtype
-        if self.vae.config.force_upcast:
-            image = image.float()
-            self.vae.to(dtype=torch.float32)
-
-        if isinstance(generator, list):
-            image_latents = [
-                retrieve_latents(self.vae.encode(image[i : i + 1]), generator=generator[i])
-                for i in range(image.shape[0])
-            ]
-            image_latents = torch.cat(image_latents, dim=0)
-        else:
-            image_latents = retrieve_latents(self.vae.encode(image), generator=generator)
-
-        if self.vae.config.force_upcast:
-            self.vae.to(dtype)
-
-        image_latents = image_latents.to(dtype)
-        image_latents = self.vae.config.scaling_factor * image_latents
-
-        return image_latents
-
-    def prepare_mask_latents(
-        self, mask, masked_image, batch_size, height, width, dtype, device, generator, do_classifier_free_guidance
-    ):
-        # resize the mask to latents shape as we concatenate the mask to the latents
-        # we do that before converting to dtype to avoid breaking in case we're using cpu_offload
-        # and half precision
-        mask = torch.nn.functional.interpolate(
-            mask, size=(height // self.vae_scale_factor, width // self.vae_scale_factor)
-        )
-        mask = mask.to(device=device, dtype=dtype)
-
-        # duplicate mask and masked_image_latents for each generation per prompt, using mps friendly method
-        if mask.shape[0] < batch_size:
-            if not batch_size % mask.shape[0] == 0:
-                raise ValueError(
-                    "The passed mask and the required batch size don't match. Masks are supposed to be duplicated to"
-                    f" a total batch size of {batch_size}, but {mask.shape[0]} masks were passed. Make sure the number"
-                    " of masks that you pass is divisible by the total requested batch size."
-                )
-            mask = mask.repeat(batch_size // mask.shape[0], 1, 1, 1)
-
-        mask = torch.cat([mask] * 2) if do_classifier_free_guidance else mask
-
-        if masked_image is not None and masked_image.shape[1] == 4:
-            masked_image_latents = masked_image
-        else:
-            masked_image_latents = None
-
-        if masked_image is not None:
-            if masked_image_latents is None:
-                masked_image = masked_image.to(device=device, dtype=dtype)
-                masked_image_latents = self._encode_vae_image(masked_image, generator=generator)
-
-            if masked_image_latents.shape[0] < batch_size:
-                if not batch_size % masked_image_latents.shape[0] == 0:
-                    raise ValueError(
-                        "The passed images and the required batch size don't match. Images are supposed to be duplicated"
-                        f" to a total batch size of {batch_size}, but {masked_image_latents.shape[0]} images were passed."
-                        " Make sure the number of images that you pass is divisible by the total requested batch size."
-                    )
-                masked_image_latents = masked_image_latents.repeat(
-                    batch_size // masked_image_latents.shape[0], 1, 1, 1
-                )
-
-            masked_image_latents = (
-                torch.cat([masked_image_latents] * 2) if do_classifier_free_guidance else masked_image_latents
-            )
-
-            # aligning device to prevent device errors when concating it with the latent model input
-            masked_image_latents = masked_image_latents.to(device=device, dtype=dtype)
-
-        return mask, masked_image_latents
+        # scale the initial noise by the standard deviation required by the scheduler
+        latents = latents * self.scheduler.init_noise_sigma
+        return latents
 
     def _get_add_time_ids(self, original_size, crops_coords_top_left, target_size, dtype):
         add_time_ids = list(original_size + crops_coords_top_left + target_size)
@@ -1241,52 +934,15 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
             self.vae.decoder.conv_in.to(dtype)
             self.vae.decoder.mid_block.to(dtype)
 
-    @property
-    def guidance_scale(self):
-        return self._guidance_scale
-
-    @property
-    def guidance_rescale(self):
-        return self._guidance_rescale
-
-    # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-    # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
-    # corresponds to doing no classifier free guidance.
-    @property
-    def do_classifier_free_guidance(self):
-        return self._guidance_scale > 1 and self.unet.config.time_cond_proj_dim is None
-
-    @property
-    def cross_attention_kwargs(self):
-        return self._cross_attention_kwargs
-
-    @property
-    def denoising_end(self):
-        return self._denoising_end
-
-    @property
-    def denoising_start(self):
-        return self._denoising_start
-
-    @property
-    def num_timesteps(self):
-        return self._num_timesteps
-
     @torch.no_grad()
     @replace_example_docstring(EXAMPLE_DOC_STRING)
     def __call__(
         self,
         prompt: str = None,
         prompt_2: Optional[str] = None,
-        image: Optional[PipelineImageInput] = None,
-        mask_image: Optional[PipelineImageInput] = None,
-        masked_image_latents: Optional[torch.FloatTensor] = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
-        strength: float = 0.8,
         num_inference_steps: int = 50,
-        timesteps: List[int] = None,
-        denoising_start: Optional[float] = None,
         denoising_end: Optional[float] = None,
         guidance_scale: float = 5.0,
         negative_prompt: Optional[str] = None,
@@ -1319,46 +975,20 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
             prompt_2 (`str`):
                 The prompt to be sent to the `tokenizer_2` and `text_encoder_2`. If not defined, `prompt` is
                 used in both text-encoders
-            image (`PipelineImageInput`, *optional*):
-                `Image`, or tensor representing an image batch, that will be used as the starting point for the
-                process.
-            mask_image (`PipelineImageInput`, *optional*):
-                `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
-                replaced by noise and therefore repainted, while black pixels will be preserved. If `mask_image` is a
-                PIL image, it will be converted to a single channel (luminance) before use. If it's a tensor, it should
-                contain one color channel (L) instead of 3, so the expected shape would be `(B, H, W, 1)`.
             height (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
                 The height in pixels of the generated image.
             width (`int`, *optional*, defaults to self.unet.config.sample_size * self.vae_scale_factor):
                 The width in pixels of the generated image.
-            strength (`float`, *optional*, defaults to 0.8):
-                Conceptually, indicates how much to transform the reference `image`. Must be between 0 and 1.
-                `image` will be used as a starting point, adding more noise to it the larger the `strength`. The
-                number of denoising steps depends on the amount of noise initially added. When `strength` is 1, added
-                noise will be maximum and the denoising process will run for the full number of iterations specified in
-                `num_inference_steps`. A value of 1, therefore, essentially ignores `image`.
             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.
-            timesteps (`List[int]`, *optional*):
-                Custom timesteps to use for the denoising process with schedulers which support a `timesteps` argument
-                in their `set_timesteps` method. If not defined, the default behavior when `num_inference_steps` is
-                passed will be used. Must be in descending order.
-            denoising_start (`float`, *optional*):
-                When specified, indicates the fraction (between 0.0 and 1.0) of the total denoising process to be
-                bypassed before it is initiated. Consequently, the initial part of the denoising process is skipped and
-                it is assumed that the passed `image` is a partly denoised image. Note that when this is specified,
-                strength will be ignored. The `denoising_start` parameter is particularly beneficial when this pipeline
-                is integrated into a "Mixture of Denoisers" multi-pipeline setup, as detailed in [**Refine Image
-                Quality**](https://huggingface.co/docs/diffusers/using-diffusers/sdxl#refine-image-quality).
             denoising_end (`float`, *optional*):
                 When specified, determines the fraction (between 0.0 and 1.0) of the total denoising process to be
                 completed before it is intentionally prematurely terminated. As a result, the returned sample will
-                still retain a substantial amount of noise (ca. final 20% of timesteps still needed) and should be
-                denoised by a successor pipeline that has `denoising_start` set to 0.8 so that it only denoises the
-                final 20% of the scheduler. The denoising_end parameter should ideally be utilized when this pipeline
-                forms a part of a "Mixture of Denoisers" multi-pipeline setup, as elaborated in [**Refine Image
-                Quality**](https://huggingface.co/docs/diffusers/using-diffusers/sdxl#refine-image-quality).
+                still retain a substantial amount of noise as determined by the discrete timesteps selected by the
+                scheduler. The denoising_end parameter should ideally be utilized when this pipeline forms a part of a
+                "Mixture of Denoisers" multi-pipeline setup, as elaborated in [**Refining the Image
+                Output**](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/stable_diffusion_xl#refining-the-image-output)
             guidance_scale (`float`, *optional*, defaults to 5.0):
                 Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
                 `guidance_scale` is defined as `w` of equation 2. of [Imagen
@@ -1454,7 +1084,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
             prompt_2,
             height,
             width,
-            strength,
             callback_steps,
             negative_prompt,
             negative_prompt_2,
@@ -1464,12 +1093,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
             negative_pooled_prompt_embeds,
         )
 
-        self._guidance_scale = guidance_scale
-        self._guidance_rescale = guidance_rescale
-        self._cross_attention_kwargs = cross_attention_kwargs
-        self._denoising_end = denoising_end
-        self._denoising_start = denoising_start
-
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
@@ -1498,126 +1121,28 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
         ) = get_weighted_text_embeddings_sdxl(
             pipe=self, prompt=prompt, neg_prompt=negative_prompt, num_images_per_prompt=num_images_per_prompt
         )
-        dtype = prompt_embeds.dtype
-
-        if isinstance(image, Image.Image):
-            image = self.image_processor.preprocess(image, height=height, width=width)
-        if image is not None:
-            image = image.to(device=self.device, dtype=dtype)
-
-        if isinstance(mask_image, Image.Image):
-            mask = self.mask_processor.preprocess(mask_image, height=height, width=width)
-        else:
-            mask = mask_image
-        if mask_image is not None:
-            mask = mask.to(device=self.device, dtype=dtype)
-
-            if masked_image_latents is not None:
-                masked_image = masked_image_latents
-            elif image.shape[1] == 4:
-                # if image is in latent space, we can't mask it
-                masked_image = None
-            else:
-                masked_image = image * (mask < 0.5)
-        else:
-            mask = None
 
         # 4. Prepare timesteps
-        def denoising_value_valid(dnv):
-            return isinstance(self.denoising_end, float) and 0 < dnv < 1
+        self.scheduler.set_timesteps(num_inference_steps, device=device)
 
-        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
-        if image is not None:
-            timesteps, num_inference_steps = self.get_timesteps(
-                num_inference_steps,
-                strength,
-                device,
-                denoising_start=self.denoising_start if denoising_value_valid else None,
-            )
-
-            # check that number of inference steps is not < 1 - as this doesn't make sense
-            if num_inference_steps < 1:
-                raise ValueError(
-                    f"After adjusting the num_inference_steps by strength parameter: {strength}, the number of pipeline"
-                    f"steps is {num_inference_steps} which is < 1 and not appropriate for this pipeline."
-                )
-
-        latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt)
-        is_strength_max = strength == 1.0
-        add_noise = True if self.denoising_start is None else False
+        timesteps = self.scheduler.timesteps
 
         # 5. Prepare latent variables
-        num_channels_latents = self.vae.config.latent_channels
-        num_channels_unet = self.unet.config.in_channels
-        return_image_latents = num_channels_unet == 4
-
+        num_channels_latents = self.unet.config.in_channels
         latents = self.prepare_latents(
-            image=image,
-            mask=mask,
-            width=width,
-            height=height,
-            num_channels_latents=num_channels_unet,
-            timestep=latent_timestep,
-            batch_size=batch_size,
-            num_images_per_prompt=num_images_per_prompt,
-            dtype=prompt_embeds.dtype,
-            device=device,
-            generator=generator,
-            add_noise=add_noise,
-            latents=latents,
-            is_strength_max=is_strength_max,
-            return_noise=True,
-            return_image_latents=return_image_latents,
+            batch_size * num_images_per_prompt,
+            num_channels_latents,
+            height,
+            width,
+            prompt_embeds.dtype,
+            device,
+            generator,
+            latents,
         )
 
-        if mask is not None:
-            if return_image_latents:
-                latents, noise, image_latents = latents
-            else:
-                latents, noise = latents
-
-        # 5.1. Prepare mask latent variables
-        if mask is not None:
-            mask, masked_image_latents = self.prepare_mask_latents(
-                mask=mask,
-                masked_image=masked_image,
-                batch_size=batch_size * num_images_per_prompt,
-                height=height,
-                width=width,
-                dtype=prompt_embeds.dtype,
-                device=device,
-                generator=generator,
-                do_classifier_free_guidance=self.do_classifier_free_guidance,
-            )
-
-            # 8. Check that sizes of mask, masked image and latents match
-            if num_channels_unet == 9:
-                # default case for runwayml/stable-diffusion-inpainting
-                num_channels_mask = mask.shape[1]
-                num_channels_masked_image = masked_image_latents.shape[1]
-                if num_channels_latents + num_channels_mask + num_channels_masked_image != num_channels_unet:
-                    raise ValueError(
-                        f"Incorrect configuration settings! The config of `pipeline.unet`: {self.unet.config} expects"
-                        f" {self.unet.config.in_channels} but received `num_channels_latents`: {num_channels_latents} +"
-                        f" `num_channels_mask`: {num_channels_mask} + `num_channels_masked_image`: {num_channels_masked_image}"
-                        f" = {num_channels_latents+num_channels_masked_image+num_channels_mask}. Please verify the config of"
-                        " `pipeline.unet` or your `mask_image` or `image` input."
-                    )
-            elif num_channels_unet != 4:
-                raise ValueError(
-                    f"The unet {self.unet.__class__} should have either 4 or 9 input channels, not {self.unet.config.in_channels}."
-                )
-
         # 6. Prepare extra step kwargs. TODO: Logic should ideally just be moved out of the pipeline
         extra_step_kwargs = self.prepare_extra_step_kwargs(generator, eta)
 
-        height, width = latents.shape[-2:]
-        height = height * self.vae_scale_factor
-        width = width * self.vae_scale_factor
-
-        original_size = original_size or (height, width)
-        target_size = target_size or (height, width)
-
         # 7. Prepare added time ids & embeddings
         add_text_embeds = pooled_prompt_embeds
         add_time_ids = self._get_add_time_ids(
@@ -1633,41 +1158,20 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
         add_text_embeds = add_text_embeds.to(device)
         add_time_ids = add_time_ids.to(device).repeat(batch_size * num_images_per_prompt, 1)
 
+        # 8. Denoising loop
         num_warmup_steps = max(len(timesteps) - num_inference_steps * self.scheduler.order, 0)
 
         # 7.1 Apply denoising_end
-        if (
-            self.denoising_end is not None
-            and self.denoising_start is not None
-            and denoising_value_valid(self.denoising_end)
-            and denoising_value_valid(self.denoising_start)
-            and self.denoising_start >= self.denoising_end
-        ):
-            raise ValueError(
-                f"`denoising_start`: {self.denoising_start} cannot be larger than or equal to `denoising_end`: "
-                + f" {self.denoising_end} when using type float."
-            )
-        elif self.denoising_end is not None and denoising_value_valid(self.denoising_end):
+        if denoising_end is not None and isinstance(denoising_end, float) and denoising_end > 0 and denoising_end < 1:
             discrete_timestep_cutoff = int(
                 round(
                     self.scheduler.config.num_train_timesteps
-                    - (self.denoising_end * self.scheduler.config.num_train_timesteps)
+                    - (denoising_end * self.scheduler.config.num_train_timesteps)
                 )
             )
             num_inference_steps = len(list(filter(lambda ts: ts >= discrete_timestep_cutoff, timesteps)))
             timesteps = timesteps[:num_inference_steps]
 
-        # 8. Optionally get Guidance Scale Embedding
-        timestep_cond = None
-        if self.unet.config.time_cond_proj_dim is not None:
-            guidance_scale_tensor = torch.tensor(self.guidance_scale - 1).repeat(batch_size * num_images_per_prompt)
-            timestep_cond = self.get_guidance_scale_embedding(
-                guidance_scale_tensor, embedding_dim=self.unet.config.time_cond_proj_dim
-            ).to(device=device, dtype=latents.dtype)
-
-        self._num_timesteps = len(timesteps)
-
-        # 9. Denoising loop
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 # expand the latents if we are doing classifier free guidance
@@ -1675,17 +1179,13 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
 
                 latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
 
-                if mask is not None and num_channels_unet == 9:
-                    latent_model_input = torch.cat([latent_model_input, mask, masked_image_latents], dim=1)
-
                 # predict the noise residual
                 added_cond_kwargs = {"text_embeds": add_text_embeds, "time_ids": add_time_ids}
                 noise_pred = self.unet(
                     latent_model_input,
                     t,
                     encoder_hidden_states=prompt_embeds,
-                    timestep_cond=timestep_cond,
-                    cross_attention_kwargs=self.cross_attention_kwargs,
+                    cross_attention_kwargs=cross_attention_kwargs,
                     added_cond_kwargs=added_cond_kwargs,
                     return_dict=False,
                 )[0]
@@ -1702,22 +1202,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
                 # compute the previous noisy sample x_t -> x_t-1
                 latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
 
-                if mask is not None and num_channels_unet == 4:
-                    init_latents_proper = image_latents
-
-                    if self.do_classifier_free_guidance:
-                        init_mask, _ = mask.chunk(2)
-                    else:
-                        init_mask = mask
-
-                    if i < len(timesteps) - 1:
-                        noise_timestep = timesteps[i + 1]
-                        init_latents_proper = self.scheduler.add_noise(
-                            init_latents_proper, noise, torch.tensor([noise_timestep])
-                        )
-
-                    latents = (1 - init_mask) * init_latents_proper + init_mask * latents
-
                 # call the callback, if provided
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
@@ -1757,204 +1241,6 @@ class SDXLLongPromptWeightingPipeline(DiffusionPipeline, FromSingleFileMixin, Lo
 
         return StableDiffusionXLPipelineOutput(images=image)
 
-    def text2img(
-        self,
-        prompt: str = None,
-        prompt_2: Optional[str] = None,
-        height: Optional[int] = None,
-        width: Optional[int] = None,
-        num_inference_steps: int = 50,
-        timesteps: List[int] = None,
-        denoising_start: Optional[float] = None,
-        denoising_end: Optional[float] = None,
-        guidance_scale: float = 5.0,
-        negative_prompt: Optional[str] = None,
-        negative_prompt_2: Optional[str] = None,
-        num_images_per_prompt: Optional[int] = 1,
-        eta: float = 0.0,
-        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        output_type: Optional[str] = "pil",
-        return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
-        callback_steps: int = 1,
-        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
-        guidance_rescale: float = 0.0,
-        original_size: Optional[Tuple[int, int]] = None,
-        crops_coords_top_left: Tuple[int, int] = (0, 0),
-        target_size: Optional[Tuple[int, int]] = None,
-    ):
-        return self.__call__(
-            prompt=prompt,
-            prompt_2=prompt_2,
-            height=height,
-            width=width,
-            num_inference_steps=num_inference_steps,
-            timesteps=timesteps,
-            denoising_start=denoising_start,
-            denoising_end=denoising_end,
-            guidance_scale=guidance_scale,
-            negative_prompt=negative_prompt,
-            negative_prompt_2=negative_prompt_2,
-            num_images_per_prompt=num_images_per_prompt,
-            eta=eta,
-            generator=generator,
-            latents=latents,
-            prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
-            output_type=output_type,
-            return_dict=return_dict,
-            callback=callback,
-            callback_steps=callback_steps,
-            cross_attention_kwargs=cross_attention_kwargs,
-            guidance_rescale=guidance_rescale,
-            original_size=original_size,
-            crops_coords_top_left=crops_coords_top_left,
-            target_size=target_size,
-        )
-
-    def img2img(
-        self,
-        prompt: str = None,
-        prompt_2: Optional[str] = None,
-        image: Optional[PipelineImageInput] = None,
-        height: Optional[int] = None,
-        width: Optional[int] = None,
-        strength: float = 0.8,
-        num_inference_steps: int = 50,
-        timesteps: List[int] = None,
-        denoising_start: Optional[float] = None,
-        denoising_end: Optional[float] = None,
-        guidance_scale: float = 5.0,
-        negative_prompt: Optional[str] = None,
-        negative_prompt_2: Optional[str] = None,
-        num_images_per_prompt: Optional[int] = 1,
-        eta: float = 0.0,
-        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        output_type: Optional[str] = "pil",
-        return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
-        callback_steps: int = 1,
-        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
-        guidance_rescale: float = 0.0,
-        original_size: Optional[Tuple[int, int]] = None,
-        crops_coords_top_left: Tuple[int, int] = (0, 0),
-        target_size: Optional[Tuple[int, int]] = None,
-    ):
-        return self.__call__(
-            prompt=prompt,
-            prompt_2=prompt_2,
-            image=image,
-            height=height,
-            width=width,
-            strength=strength,
-            num_inference_steps=num_inference_steps,
-            timesteps=timesteps,
-            denoising_start=denoising_start,
-            denoising_end=denoising_end,
-            guidance_scale=guidance_scale,
-            negative_prompt=negative_prompt,
-            negative_prompt_2=negative_prompt_2,
-            num_images_per_prompt=num_images_per_prompt,
-            eta=eta,
-            generator=generator,
-            latents=latents,
-            prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
-            output_type=output_type,
-            return_dict=return_dict,
-            callback=callback,
-            callback_steps=callback_steps,
-            cross_attention_kwargs=cross_attention_kwargs,
-            guidance_rescale=guidance_rescale,
-            original_size=original_size,
-            crops_coords_top_left=crops_coords_top_left,
-            target_size=target_size,
-        )
-
-    def inpaint(
-        self,
-        prompt: str = None,
-        prompt_2: Optional[str] = None,
-        image: Optional[PipelineImageInput] = None,
-        mask_image: Optional[PipelineImageInput] = None,
-        masked_image_latents: Optional[torch.FloatTensor] = None,
-        height: Optional[int] = None,
-        width: Optional[int] = None,
-        strength: float = 0.8,
-        num_inference_steps: int = 50,
-        timesteps: List[int] = None,
-        denoising_start: Optional[float] = None,
-        denoising_end: Optional[float] = None,
-        guidance_scale: float = 5.0,
-        negative_prompt: Optional[str] = None,
-        negative_prompt_2: Optional[str] = None,
-        num_images_per_prompt: Optional[int] = 1,
-        eta: float = 0.0,
-        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        output_type: Optional[str] = "pil",
-        return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
-        callback_steps: int = 1,
-        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
-        guidance_rescale: float = 0.0,
-        original_size: Optional[Tuple[int, int]] = None,
-        crops_coords_top_left: Tuple[int, int] = (0, 0),
-        target_size: Optional[Tuple[int, int]] = None,
-    ):
-        return self.__call__(
-            prompt=prompt,
-            prompt_2=prompt_2,
-            image=image,
-            mask_image=mask_image,
-            masked_image_latents=masked_image_latents,
-            height=height,
-            width=width,
-            strength=strength,
-            num_inference_steps=num_inference_steps,
-            timesteps=timesteps,
-            denoising_start=denoising_start,
-            denoising_end=denoising_end,
-            guidance_scale=guidance_scale,
-            negative_prompt=negative_prompt,
-            negative_prompt_2=negative_prompt_2,
-            num_images_per_prompt=num_images_per_prompt,
-            eta=eta,
-            generator=generator,
-            latents=latents,
-            prompt_embeds=prompt_embeds,
-            negative_prompt_embeds=negative_prompt_embeds,
-            pooled_prompt_embeds=pooled_prompt_embeds,
-            negative_pooled_prompt_embeds=negative_pooled_prompt_embeds,
-            output_type=output_type,
-            return_dict=return_dict,
-            callback=callback,
-            callback_steps=callback_steps,
-            cross_attention_kwargs=cross_attention_kwargs,
-            guidance_rescale=guidance_rescale,
-            original_size=original_size,
-            crops_coords_top_left=crops_coords_top_left,
-            target_size=target_size,
-        )
-
     # Overrride to properly handle the loading and unloading of the additional text encoder.
     def load_lora_weights(self, pretrained_model_name_or_path_or_dict: Union[str, Dict[str, torch.Tensor]], **kwargs):
         # We could have accessed the unet config from `lora_state_dict()` too. We pass

examples/text_to_image/train_text_to_image_lora.py

@@ -799,8 +799,7 @@ def main():
                         save_path = os.path.join(args.output_dir, f"checkpoint-{global_step}")
                         accelerator.save_state(save_path)
 
-                        unwrapped_unet = accelerator.unwrap_model(unet)
-                        unet_lora_state_dict = get_peft_model_state_dict(unwrapped_unet)
+                        unet_lora_state_dict = get_peft_model_state_dict(unet)
 
                         StableDiffusionPipeline.save_lora_weights(
                             save_directory=save_path,
@@ -865,8 +864,7 @@ def main():
     if accelerator.is_main_process:
         unet = unet.to(torch.float32)
 
-        unwrapped_unet = accelerator.unwrap_model(unet)
-        unet_lora_state_dict = get_peft_model_state_dict(unwrapped_unet)
+        unet_lora_state_dict = get_peft_model_state_dict(unet)
         StableDiffusionPipeline.save_lora_weights(
             save_directory=args.output_dir,
             unet_lora_layers=unet_lora_state_dict,

examples/wuerstchen/text_to_image/README.md

@@ -77,7 +77,7 @@ First, you need to set up your development environment as explained in the [inst
 ```bash
 export DATASET_NAME="lambdalabs/pokemon-blip-captions"
 
-accelerate launch train_text_to_image_lora_prior.py \
+accelerate launch train_text_to_image_prior_lora.py \
   --mixed_precision="fp16" \
   --dataset_name=$DATASET_NAME --caption_column="text" \
   --resolution=768 \

src/diffusers/loaders/unet.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import os
-from collections import defaultdict
+from collections import OrderedDict, defaultdict
 from contextlib import nullcontext
 from typing import Callable, Dict, List, Optional, Union
 
@@ -664,80 +664,6 @@ class UNet2DConditionLoadersMixin:
             if hasattr(self, "peft_config"):
                 self.peft_config.pop(adapter_name, None)
 
-    def _convert_ip_adapter_image_proj_to_diffusers(self, state_dict):
-        updated_state_dict = {}
-        image_projection = None
-
-        if "proj.weight" in state_dict:
-            # IP-Adapter
-            num_image_text_embeds = 4
-            clip_embeddings_dim = state_dict["proj.weight"].shape[-1]
-            cross_attention_dim = state_dict["proj.weight"].shape[0] // 4
-
-            image_projection = ImageProjection(
-                cross_attention_dim=cross_attention_dim,
-                image_embed_dim=clip_embeddings_dim,
-                num_image_text_embeds=num_image_text_embeds,
-            )
-
-            for key, value in state_dict.items():
-                diffusers_name = key.replace("proj", "image_embeds")
-                updated_state_dict[diffusers_name] = value
-
-        elif "proj.3.weight" in state_dict:
-            # IP-Adapter Full
-            clip_embeddings_dim = state_dict["proj.0.weight"].shape[0]
-            cross_attention_dim = state_dict["proj.3.weight"].shape[0]
-
-            image_projection = MLPProjection(
-                cross_attention_dim=cross_attention_dim, image_embed_dim=clip_embeddings_dim
-            )
-
-            for key, value in state_dict.items():
-                diffusers_name = key.replace("proj.0", "ff.net.0.proj")
-                diffusers_name = diffusers_name.replace("proj.2", "ff.net.2")
-                diffusers_name = diffusers_name.replace("proj.3", "norm")
-                updated_state_dict[diffusers_name] = value
-
-        else:
-            # IP-Adapter Plus
-            num_image_text_embeds = state_dict["latents"].shape[1]
-            embed_dims = state_dict["proj_in.weight"].shape[1]
-            output_dims = state_dict["proj_out.weight"].shape[0]
-            hidden_dims = state_dict["latents"].shape[2]
-            heads = state_dict["layers.0.0.to_q.weight"].shape[0] // 64
-
-            image_projection = Resampler(
-                embed_dims=embed_dims,
-                output_dims=output_dims,
-                hidden_dims=hidden_dims,
-                heads=heads,
-                num_queries=num_image_text_embeds,
-            )
-
-            for key, value in state_dict.items():
-                diffusers_name = key.replace("0.to", "2.to")
-                diffusers_name = diffusers_name.replace("1.0.weight", "3.0.weight")
-                diffusers_name = diffusers_name.replace("1.0.bias", "3.0.bias")
-                diffusers_name = diffusers_name.replace("1.1.weight", "3.1.net.0.proj.weight")
-                diffusers_name = diffusers_name.replace("1.3.weight", "3.1.net.2.weight")
-
-                if "norm1" in diffusers_name:
-                    updated_state_dict[diffusers_name.replace("0.norm1", "0")] = value
-                elif "norm2" in diffusers_name:
-                    updated_state_dict[diffusers_name.replace("0.norm2", "1")] = value
-                elif "to_kv" in diffusers_name:
-                    v_chunk = value.chunk(2, dim=0)
-                    updated_state_dict[diffusers_name.replace("to_kv", "to_k")] = v_chunk[0]
-                    updated_state_dict[diffusers_name.replace("to_kv", "to_v")] = v_chunk[1]
-                elif "to_out" in diffusers_name:
-                    updated_state_dict[diffusers_name.replace("to_out", "to_out.0")] = value
-                else:
-                    updated_state_dict[diffusers_name] = value
-
-        image_projection.load_state_dict(updated_state_dict)
-        return image_projection
-
     def _load_ip_adapter_weights(self, state_dict):
         from ..models.attention_processor import (
             AttnProcessor,
@@ -798,8 +724,103 @@ class UNet2DConditionLoadersMixin:
 
         self.set_attn_processor(attn_procs)
 
-        # convert IP-Adapter Image Projection layers to diffusers
-        image_projection = self._convert_ip_adapter_image_proj_to_diffusers(state_dict["image_proj"])
+        # create image projection layers.
+        if "proj.weight" in state_dict["image_proj"]:
+            # IP-Adapter
+            clip_embeddings_dim = state_dict["image_proj"]["proj.weight"].shape[-1]
+            cross_attention_dim = state_dict["image_proj"]["proj.weight"].shape[0] // 4
+
+            image_projection = ImageProjection(
+                cross_attention_dim=cross_attention_dim,
+                image_embed_dim=clip_embeddings_dim,
+                num_image_text_embeds=num_image_text_embeds,
+            )
+            image_projection.to(dtype=self.dtype, device=self.device)
+
+            # load image projection layer weights
+            image_proj_state_dict = {}
+            image_proj_state_dict.update(
+                {
+                    "image_embeds.weight": state_dict["image_proj"]["proj.weight"],
+                    "image_embeds.bias": state_dict["image_proj"]["proj.bias"],
+                    "norm.weight": state_dict["image_proj"]["norm.weight"],
+                    "norm.bias": state_dict["image_proj"]["norm.bias"],
+                }
+            )
+            image_projection.load_state_dict(image_proj_state_dict)
+            del image_proj_state_dict
+
+        elif "proj.3.weight" in state_dict["image_proj"]:
+            clip_embeddings_dim = state_dict["image_proj"]["proj.0.weight"].shape[0]
+            cross_attention_dim = state_dict["image_proj"]["proj.3.weight"].shape[0]
+
+            image_projection = MLPProjection(
+                cross_attention_dim=cross_attention_dim, image_embed_dim=clip_embeddings_dim
+            )
+            image_projection.to(dtype=self.dtype, device=self.device)
+
+            # load image projection layer weights
+            image_proj_state_dict = {}
+            image_proj_state_dict.update(
+                {
+                    "ff.net.0.proj.weight": state_dict["image_proj"]["proj.0.weight"],
+                    "ff.net.0.proj.bias": state_dict["image_proj"]["proj.0.bias"],
+                    "ff.net.2.weight": state_dict["image_proj"]["proj.2.weight"],
+                    "ff.net.2.bias": state_dict["image_proj"]["proj.2.bias"],
+                    "norm.weight": state_dict["image_proj"]["proj.3.weight"],
+                    "norm.bias": state_dict["image_proj"]["proj.3.bias"],
+                }
+            )
+            image_projection.load_state_dict(image_proj_state_dict)
+            del image_proj_state_dict
+
+        else:
+            # IP-Adapter Plus
+            embed_dims = state_dict["image_proj"]["proj_in.weight"].shape[1]
+            output_dims = state_dict["image_proj"]["proj_out.weight"].shape[0]
+            hidden_dims = state_dict["image_proj"]["latents"].shape[2]
+            heads = state_dict["image_proj"]["layers.0.0.to_q.weight"].shape[0] // 64
+
+            image_projection = Resampler(
+                embed_dims=embed_dims,
+                output_dims=output_dims,
+                hidden_dims=hidden_dims,
+                heads=heads,
+                num_queries=num_image_text_embeds,
+            )
+
+            image_proj_state_dict = state_dict["image_proj"]
+
+            new_sd = OrderedDict()
+            for k, v in image_proj_state_dict.items():
+                if "0.to" in k:
+                    k = k.replace("0.to", "2.to")
+                elif "1.0.weight" in k:
+                    k = k.replace("1.0.weight", "3.0.weight")
+                elif "1.0.bias" in k:
+                    k = k.replace("1.0.bias", "3.0.bias")
+                elif "1.1.weight" in k:
+                    k = k.replace("1.1.weight", "3.1.net.0.proj.weight")
+                elif "1.3.weight" in k:
+                    k = k.replace("1.3.weight", "3.1.net.2.weight")
+
+                if "norm1" in k:
+                    new_sd[k.replace("0.norm1", "0")] = v
+                elif "norm2" in k:
+                    new_sd[k.replace("0.norm2", "1")] = v
+                elif "to_kv" in k:
+                    v_chunk = v.chunk(2, dim=0)
+                    new_sd[k.replace("to_kv", "to_k")] = v_chunk[0]
+                    new_sd[k.replace("to_kv", "to_v")] = v_chunk[1]
+                elif "to_out" in k:
+                    new_sd[k.replace("to_out", "to_out.0")] = v
+                else:
+                    new_sd[k] = v
+
+            image_projection.load_state_dict(new_sd)
+            del image_proj_state_dict
 
         self.encoder_hid_proj = image_projection.to(device=self.device, dtype=self.dtype)
         self.config.encoder_hid_dim_type = "ip_image_proj"
+
+    delete_adapter_layers

src/diffusers/models/controlnetxs.py

@@ -23,7 +23,9 @@ from torch.nn.modules.normalization import GroupNorm
 
 from ..configuration_utils import ConfigMixin, register_to_config
 from ..utils import BaseOutput, logging
-from .attention_processor import USE_PEFT_BACKEND, AttentionProcessor
+from .attention_processor import (
+    AttentionProcessor,
+)
 from .autoencoders import AutoencoderKL
 from .lora import LoRACompatibleConv
 from .modeling_utils import ModelMixin
@@ -815,23 +817,11 @@ def increase_block_input_in_encoder_resnet(unet: UNet2DConditionModel, block_no,
     norm_kwargs = {a: getattr(old_norm1, a) for a in norm_args}
     norm_kwargs["num_channels"] += by  # surgery done here
     # conv1
-    conv1_args = [
-        "in_channels",
-        "out_channels",
-        "kernel_size",
-        "stride",
-        "padding",
-        "dilation",
-        "groups",
-        "bias",
-        "padding_mode",
-    ]
-    if not USE_PEFT_BACKEND:
-        conv1_args.append("lora_layer")
-
+    conv1_args = (
+        "in_channels out_channels kernel_size stride padding dilation groups bias padding_mode lora_layer".split(" ")
+    )
     for a in conv1_args:
         assert hasattr(old_conv1, a)
-
     conv1_kwargs = {a: getattr(old_conv1, a) for a in conv1_args}
     conv1_kwargs["bias"] = "bias" in conv1_kwargs  # as param, bias is a boolean, but as attr, it's a tensor.
     conv1_kwargs["in_channels"] += by  # surgery done here
@@ -849,42 +839,25 @@ def increase_block_input_in_encoder_resnet(unet: UNet2DConditionModel, block_no,
     }
     # swap old with new modules
     unet.down_blocks[block_no].resnets[resnet_idx].norm1 = GroupNorm(**norm_kwargs)
-    unet.down_blocks[block_no].resnets[resnet_idx].conv1 = (
-        nn.Conv2d(**conv1_kwargs) if USE_PEFT_BACKEND else LoRACompatibleConv(**conv1_kwargs)
-    )
-    unet.down_blocks[block_no].resnets[resnet_idx].conv_shortcut = (
-        nn.Conv2d(**conv_shortcut_args_kwargs) if USE_PEFT_BACKEND else LoRACompatibleConv(**conv_shortcut_args_kwargs)
-    )
+    unet.down_blocks[block_no].resnets[resnet_idx].conv1 = LoRACompatibleConv(**conv1_kwargs)
+    unet.down_blocks[block_no].resnets[resnet_idx].conv_shortcut = LoRACompatibleConv(**conv_shortcut_args_kwargs)
     unet.down_blocks[block_no].resnets[resnet_idx].in_channels += by  # surgery done here
 
 
 def increase_block_input_in_encoder_downsampler(unet: UNet2DConditionModel, block_no, by):
     """Increase channels sizes to allow for additional concatted information from base model"""
     old_down = unet.down_blocks[block_no].downsamplers[0].conv
-
-    args = [
-        "in_channels",
-        "out_channels",
-        "kernel_size",
-        "stride",
-        "padding",
-        "dilation",
-        "groups",
-        "bias",
-        "padding_mode",
-    ]
-    if not USE_PEFT_BACKEND:
-        args.append("lora_layer")
-
+    # conv1
+    args = "in_channels out_channels kernel_size stride padding dilation groups bias padding_mode lora_layer".split(
+        " "
+    )
     for a in args:
         assert hasattr(old_down, a)
     kwargs = {a: getattr(old_down, a) for a in args}
     kwargs["bias"] = "bias" in kwargs  # as param, bias is a boolean, but as attr, it's a tensor.
     kwargs["in_channels"] += by  # surgery done here
     # swap old with new modules
-    unet.down_blocks[block_no].downsamplers[0].conv = (
-        nn.Conv2d(**kwargs) if USE_PEFT_BACKEND else LoRACompatibleConv(**kwargs)
-    )
+    unet.down_blocks[block_no].downsamplers[0].conv = LoRACompatibleConv(**kwargs)
     unet.down_blocks[block_no].downsamplers[0].channels += by  # surgery done here
 
 
@@ -898,20 +871,12 @@ def increase_block_input_in_mid_resnet(unet: UNet2DConditionModel, by):
         assert hasattr(old_norm1, a)
     norm_kwargs = {a: getattr(old_norm1, a) for a in norm_args}
     norm_kwargs["num_channels"] += by  # surgery done here
-    conv1_args = [
-        "in_channels",
-        "out_channels",
-        "kernel_size",
-        "stride",
-        "padding",
-        "dilation",
-        "groups",
-        "bias",
-        "padding_mode",
-    ]
-    if not USE_PEFT_BACKEND:
-        conv1_args.append("lora_layer")
-
+    # conv1
+    conv1_args = (
+        "in_channels out_channels kernel_size stride padding dilation groups bias padding_mode lora_layer".split(" ")
+    )
+    for a in conv1_args:
+        assert hasattr(old_conv1, a)
     conv1_kwargs = {a: getattr(old_conv1, a) for a in conv1_args}
     conv1_kwargs["bias"] = "bias" in conv1_kwargs  # as param, bias is a boolean, but as attr, it's a tensor.
     conv1_kwargs["in_channels"] += by  # surgery done here
@@ -929,12 +894,8 @@ def increase_block_input_in_mid_resnet(unet: UNet2DConditionModel, by):
     }
     # swap old with new modules
     unet.mid_block.resnets[0].norm1 = GroupNorm(**norm_kwargs)
-    unet.mid_block.resnets[0].conv1 = (
-        nn.Conv2d(**conv1_kwargs) if USE_PEFT_BACKEND else LoRACompatibleConv(**conv1_kwargs)
-    )
-    unet.mid_block.resnets[0].conv_shortcut = (
-        nn.Conv2d(**conv_shortcut_args_kwargs) if USE_PEFT_BACKEND else LoRACompatibleConv(**conv_shortcut_args_kwargs)
-    )
+    unet.mid_block.resnets[0].conv1 = LoRACompatibleConv(**conv1_kwargs)
+    unet.mid_block.resnets[0].conv_shortcut = LoRACompatibleConv(**conv_shortcut_args_kwargs)
     unet.mid_block.resnets[0].in_channels += by  # surgery done here
 
 

src/diffusers/pipelines/__init__.py

@@ -20,7 +20,6 @@ _dummy_objects = {}
 _import_structure = {
     "controlnet": [],
     "controlnet_xs": [],
-    "deprecated": [],
     "latent_diffusion": [],
     "stable_diffusion": [],
     "stable_diffusion_xl": [],
@@ -45,20 +44,16 @@ else:
     _import_structure["ddpm"] = ["DDPMPipeline"]
     _import_structure["dit"] = ["DiTPipeline"]
     _import_structure["latent_diffusion"].extend(["LDMSuperResolutionPipeline"])
+    _import_structure["latent_diffusion_uncond"] = ["LDMPipeline"]
     _import_structure["pipeline_utils"] = [
         "AudioPipelineOutput",
         "DiffusionPipeline",
         "ImagePipelineOutput",
     ]
-    _import_structure["deprecated"].extend(
-        [
-            "PNDMPipeline",
-            "LDMPipeline",
-            "RePaintPipeline",
-            "ScoreSdeVePipeline",
-            "KarrasVePipeline",
-        ]
-    )
+    _import_structure["pndm"] = ["PNDMPipeline"]
+    _import_structure["repaint"] = ["RePaintPipeline"]
+    _import_structure["score_sde_ve"] = ["ScoreSdeVePipeline"]
+    _import_structure["stochastic_karras_ve"] = ["KarrasVePipeline"]
 try:
     if not (is_torch_available() and is_librosa_available()):
         raise OptionalDependencyNotAvailable()
@@ -67,23 +62,7 @@ except OptionalDependencyNotAvailable:
 
     _dummy_objects.update(get_objects_from_module(dummy_torch_and_librosa_objects))
 else:
-    _import_structure["deprecated"].extend(["AudioDiffusionPipeline", "Mel"])
-
-try:
-    if not (is_transformers_available() and is_torch_available() and is_note_seq_available()):
-        raise OptionalDependencyNotAvailable()
-except OptionalDependencyNotAvailable:
-    from ..utils import dummy_transformers_and_torch_and_note_seq_objects  # noqa F403
-
-    _dummy_objects.update(get_objects_from_module(dummy_transformers_and_torch_and_note_seq_objects))
-else:
-    _import_structure["deprecated"].extend(
-        [
-            "MidiProcessor",
-            "SpectrogramDiffusionPipeline",
-        ]
-    )
-
+    _import_structure["audio_diffusion"] = ["AudioDiffusionPipeline", "Mel"]
 try:
     if not (is_torch_available() and is_transformers_available()):
         raise OptionalDependencyNotAvailable()
@@ -92,22 +71,10 @@ except OptionalDependencyNotAvailable:
 
     _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
 else:
-    _import_structure["deprecated"].extend(
-        [
-            "VQDiffusionPipeline",
-            "AltDiffusionPipeline",
-            "AltDiffusionImg2ImgPipeline",
-            "CycleDiffusionPipeline",
-            "StableDiffusionInpaintPipelineLegacy",
-            "StableDiffusionPix2PixZeroPipeline",
-            "StableDiffusionParadigmsPipeline",
-            "StableDiffusionModelEditingPipeline",
-            "VersatileDiffusionDualGuidedPipeline",
-            "VersatileDiffusionImageVariationPipeline",
-            "VersatileDiffusionPipeline",
-            "VersatileDiffusionTextToImagePipeline",
-        ]
-    )
+    _import_structure["alt_diffusion"] = [
+        "AltDiffusionImg2ImgPipeline",
+        "AltDiffusionPipeline",
+    ]
     _import_structure["animatediff"] = ["AnimateDiffPipeline"]
     _import_structure["audioldm"] = ["AudioLDMPipeline"]
     _import_structure["audioldm2"] = [
@@ -179,6 +146,7 @@ else:
     _import_structure["stable_diffusion"].extend(
         [
             "CLIPImageProjection",
+            "CycleDiffusionPipeline",
             "StableDiffusionAttendAndExcitePipeline",
             "StableDiffusionDepth2ImgPipeline",
             "StableDiffusionDiffEditPipeline",
@@ -188,11 +156,15 @@ else:
             "StableDiffusionImageVariationPipeline",
             "StableDiffusionImg2ImgPipeline",
             "StableDiffusionInpaintPipeline",
+            "StableDiffusionInpaintPipelineLegacy",
             "StableDiffusionInstructPix2PixPipeline",
             "StableDiffusionLatentUpscalePipeline",
             "StableDiffusionLDM3DPipeline",
+            "StableDiffusionModelEditingPipeline",
             "StableDiffusionPanoramaPipeline",
+            "StableDiffusionParadigmsPipeline",
             "StableDiffusionPipeline",
+            "StableDiffusionPix2PixZeroPipeline",
             "StableDiffusionSAGPipeline",
             "StableDiffusionUpscalePipeline",
             "StableUnCLIPImg2ImgPipeline",
@@ -226,6 +198,13 @@ else:
         "UniDiffuserPipeline",
         "UniDiffuserTextDecoder",
     ]
+    _import_structure["versatile_diffusion"] = [
+        "VersatileDiffusionDualGuidedPipeline",
+        "VersatileDiffusionImageVariationPipeline",
+        "VersatileDiffusionPipeline",
+        "VersatileDiffusionTextToImagePipeline",
+    ]
+    _import_structure["vq_diffusion"] = ["VQDiffusionPipeline"]
     _import_structure["wuerstchen"] = [
         "WuerstchenCombinedPipeline",
         "WuerstchenDecoderPipeline",
@@ -252,6 +231,7 @@ else:
         [
             "OnnxStableDiffusionImg2ImgPipeline",
             "OnnxStableDiffusionInpaintPipeline",
+            "OnnxStableDiffusionInpaintPipelineLegacy",
             "OnnxStableDiffusionPipeline",
             "OnnxStableDiffusionUpscalePipeline",
             "StableDiffusionOnnxPipeline",
@@ -299,6 +279,18 @@ else:
             "FlaxStableDiffusionXLPipeline",
         ]
     )
+try:
+    if not (is_transformers_available() and is_torch_available() and is_note_seq_available()):
+        raise OptionalDependencyNotAvailable()
+except OptionalDependencyNotAvailable:
+    from ..utils import dummy_transformers_and_torch_and_note_seq_objects  # noqa F403
+
+    _dummy_objects.update(get_objects_from_module(dummy_transformers_and_torch_and_note_seq_objects))
+else:
+    _import_structure["spectrogram_diffusion"] = [
+        "MidiProcessor",
+        "SpectrogramDiffusionPipeline",
+    ]
 
 if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
     try:
@@ -317,14 +309,18 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         from .dance_diffusion import DanceDiffusionPipeline
         from .ddim import DDIMPipeline
         from .ddpm import DDPMPipeline
-        from .deprecated import KarrasVePipeline, LDMPipeline, PNDMPipeline, RePaintPipeline, ScoreSdeVePipeline
         from .dit import DiTPipeline
         from .latent_diffusion import LDMSuperResolutionPipeline
+        from .latent_diffusion_uncond import LDMPipeline
         from .pipeline_utils import (
             AudioPipelineOutput,
             DiffusionPipeline,
             ImagePipelineOutput,
         )
+        from .pndm import PNDMPipeline
+        from .repaint import RePaintPipeline
+        from .score_sde_ve import ScoreSdeVePipeline
+        from .stochastic_karras_ve import KarrasVePipeline
 
     try:
         if not (is_torch_available() and is_librosa_available()):
@@ -332,7 +328,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
     except OptionalDependencyNotAvailable:
         from ..utils.dummy_torch_and_librosa_objects import *
     else:
-        from .deprecated import AudioDiffusionPipeline, Mel
+        from .audio_diffusion import AudioDiffusionPipeline, Mel
 
     try:
         if not (is_torch_available() and is_transformers_available()):
@@ -340,6 +336,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
     except OptionalDependencyNotAvailable:
         from ..utils.dummy_torch_and_transformers_objects import *
     else:
+        from .alt_diffusion import AltDiffusionImg2ImgPipeline, AltDiffusionPipeline
         from .animatediff import AnimateDiffPipeline
         from .audioldm import AudioLDMPipeline
         from .audioldm2 import (
@@ -369,20 +366,6 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             IFPipeline,
             IFSuperResolutionPipeline,
         )
-        from .deprecated import (
-            AltDiffusionImg2ImgPipeline,
-            AltDiffusionPipeline,
-            CycleDiffusionPipeline,
-            StableDiffusionInpaintPipelineLegacy,
-            StableDiffusionModelEditingPipeline,
-            StableDiffusionParadigmsPipeline,
-            StableDiffusionPix2PixZeroPipeline,
-            VersatileDiffusionDualGuidedPipeline,
-            VersatileDiffusionImageVariationPipeline,
-            VersatileDiffusionPipeline,
-            VersatileDiffusionTextToImagePipeline,
-            VQDiffusionPipeline,
-        )
         from .kandinsky import (
             KandinskyCombinedPipeline,
             KandinskyImg2ImgCombinedPipeline,
@@ -420,6 +403,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         from .shap_e import ShapEImg2ImgPipeline, ShapEPipeline
         from .stable_diffusion import (
             CLIPImageProjection,
+            CycleDiffusionPipeline,
             StableDiffusionAttendAndExcitePipeline,
             StableDiffusionDepth2ImgPipeline,
             StableDiffusionDiffEditPipeline,
@@ -428,11 +412,15 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             StableDiffusionImageVariationPipeline,
             StableDiffusionImg2ImgPipeline,
             StableDiffusionInpaintPipeline,
+            StableDiffusionInpaintPipelineLegacy,
             StableDiffusionInstructPix2PixPipeline,
             StableDiffusionLatentUpscalePipeline,
             StableDiffusionLDM3DPipeline,
+            StableDiffusionModelEditingPipeline,
             StableDiffusionPanoramaPipeline,
+            StableDiffusionParadigmsPipeline,
             StableDiffusionPipeline,
+            StableDiffusionPix2PixZeroPipeline,
             StableDiffusionSAGPipeline,
             StableDiffusionUpscalePipeline,
             StableUnCLIPImg2ImgPipeline,
@@ -463,6 +451,13 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             UniDiffuserPipeline,
             UniDiffuserTextDecoder,
         )
+        from .versatile_diffusion import (
+            VersatileDiffusionDualGuidedPipeline,
+            VersatileDiffusionImageVariationPipeline,
+            VersatileDiffusionPipeline,
+            VersatileDiffusionTextToImagePipeline,
+        )
+        from .vq_diffusion import VQDiffusionPipeline
         from .wuerstchen import (
             WuerstchenCombinedPipeline,
             WuerstchenDecoderPipeline,
@@ -487,6 +482,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             from .stable_diffusion import (
                 OnnxStableDiffusionImg2ImgPipeline,
                 OnnxStableDiffusionInpaintPipeline,
+                OnnxStableDiffusionInpaintPipelineLegacy,
                 OnnxStableDiffusionPipeline,
                 OnnxStableDiffusionUpscalePipeline,
                 StableDiffusionOnnxPipeline,
@@ -531,7 +527,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             from ..utils.dummy_transformers_and_torch_and_note_seq_objects import *  # noqa F403
 
         else:
-            from .deprecated import (
+            from .spectrogram_diffusion import (
                 MidiProcessor,
                 SpectrogramDiffusionPipeline,
             )

src/diffusers/pipelines/alt_diffusion/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import (
+from ...utils import (
     DIFFUSERS_SLOW_IMPORT,
     OptionalDependencyNotAvailable,
     _LazyModule,
@@ -17,7 +17,7 @@ try:
     if not (is_transformers_available() and is_torch_available()):
         raise OptionalDependencyNotAvailable()
 except OptionalDependencyNotAvailable:
-    from ....utils import dummy_torch_and_transformers_objects
+    from ...utils import dummy_torch_and_transformers_objects
 
     _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
 else:
@@ -32,7 +32,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         if not (is_transformers_available() and is_torch_available()):
             raise OptionalDependencyNotAvailable()
     except OptionalDependencyNotAvailable:
-        from ....utils.dummy_torch_and_transformers_objects import *
+        from ...utils.dummy_torch_and_transformers_objects import *
 
     else:
         from .modeling_roberta_series import RobertaSeriesModelWithTransformation

src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion.py

@@ -19,14 +19,14 @@ import torch
 from packaging import version
 from transformers import CLIPImageProcessor, CLIPVisionModelWithProjection, XLMRobertaTokenizer
 
-from ....configuration_utils import FrozenDict
-from ....image_processor import PipelineImageInput, VaeImageProcessor
-from ....loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, ImageProjection, UNet2DConditionModel
-from ....models.attention_processor import FusedAttnProcessor2_0
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import (
+from ...configuration_utils import FrozenDict
+from ...image_processor import PipelineImageInput, VaeImageProcessor
+from ...loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
+from ...models.attention_processor import FusedAttnProcessor2_0
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import (
     USE_PEFT_BACKEND,
     deprecate,
     logging,
@@ -34,9 +34,9 @@ from ....utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 from .modeling_roberta_series import RobertaSeriesModelWithTransformation
 from .pipeline_output import AltDiffusionPipelineOutput
 
@@ -119,6 +119,7 @@ def retrieve_timesteps(
     return timesteps, num_inference_steps
 
 
+# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline with Stable->Alt, CLIPTextModel->RobertaSeriesModelWithTransformation, CLIPTokenizer->XLMRobertaTokenizer, AltDiffusionSafetyChecker->StableDiffusionSafetyChecker
 class AltDiffusionPipeline(
     DiffusionPipeline, TextualInversionLoaderMixin, LoraLoaderMixin, IPAdapterMixin, FromSingleFileMixin
 ):
@@ -155,7 +156,7 @@ class AltDiffusionPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

src/diffusers/pipelines/alt_diffusion/pipeline_alt_diffusion_img2img.py

@@ -21,14 +21,14 @@ import torch
 from packaging import version
 from transformers import CLIPImageProcessor, CLIPVisionModelWithProjection, XLMRobertaTokenizer
 
-from ....configuration_utils import FrozenDict
-from ....image_processor import PipelineImageInput, VaeImageProcessor
-from ....loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, ImageProjection, UNet2DConditionModel
-from ....models.attention_processor import FusedAttnProcessor2_0
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import (
+from ...configuration_utils import FrozenDict
+from ...image_processor import PipelineImageInput, VaeImageProcessor
+from ...loaders import FromSingleFileMixin, IPAdapterMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, ImageProjection, UNet2DConditionModel
+from ...models.attention_processor import FusedAttnProcessor2_0
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import (
     PIL_INTERPOLATION,
     USE_PEFT_BACKEND,
     deprecate,
@@ -37,9 +37,9 @@ from ....utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from ..stable_diffusion.safety_checker import StableDiffusionSafetyChecker
 from .modeling_roberta_series import RobertaSeriesModelWithTransformation
 from .pipeline_output import AltDiffusionPipelineOutput
 
@@ -159,6 +159,7 @@ def retrieve_timesteps(
     return timesteps, num_inference_steps
 
 
+# Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_img2img.StableDiffusionImg2ImgPipeline with Stable->Alt, CLIPTextModel->RobertaSeriesModelWithTransformation, CLIPTokenizer->XLMRobertaTokenizer, AltDiffusionSafetyChecker->StableDiffusionSafetyChecker
 class AltDiffusionImg2ImgPipeline(
     DiffusionPipeline, TextualInversionLoaderMixin, IPAdapterMixin, LoraLoaderMixin, FromSingleFileMixin
 ):
@@ -195,7 +196,7 @@ class AltDiffusionImg2ImgPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

src/diffusers/pipelines/alt_diffusion/pipeline_output.py

@@ -4,7 +4,7 @@ from typing import List, Optional, Union
 import numpy as np
 import PIL.Image
 
-from ....utils import (
+from ...utils import (
     BaseOutput,
 )
 

src/diffusers/pipelines/animatediff/pipeline_animatediff.py

@@ -106,7 +106,7 @@ class AnimateDiffPipeline(DiffusionPipeline, TextualInversionLoaderMixin, IPAdap
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["feature_extractor", "image_encoder"]
 
     def __init__(

src/diffusers/pipelines/audio_diffusion/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
+from ...utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
 
 
 _import_structure = {

src/diffusers/pipelines/audio_diffusion/mel.py

@@ -15,8 +15,8 @@
 
 import numpy as np  # noqa: E402
 
-from ....configuration_utils import ConfigMixin, register_to_config
-from ....schedulers.scheduling_utils import SchedulerMixin
+from ...configuration_utils import ConfigMixin, register_to_config
+from ...schedulers.scheduling_utils import SchedulerMixin
 
 
 try:

src/diffusers/pipelines/audio_diffusion/pipeline_audio_diffusion.py

@@ -20,10 +20,10 @@ import numpy as np
 import torch
 from PIL import Image
 
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....schedulers import DDIMScheduler, DDPMScheduler
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import AudioPipelineOutput, BaseOutput, DiffusionPipeline, ImagePipelineOutput
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...schedulers import DDIMScheduler, DDPMScheduler
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import AudioPipelineOutput, BaseOutput, DiffusionPipeline, ImagePipelineOutput
 from .mel import Mel
 
 

src/diffusers/pipelines/controlnet/pipeline_controlnet.py

@@ -176,7 +176,7 @@ class StableDiffusionControlNetPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

src/diffusers/pipelines/controlnet/pipeline_controlnet_inpaint.py

@@ -291,7 +291,7 @@ class StableDiffusionControlNetInpaintPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py

@@ -165,7 +165,7 @@ class StableDiffusionXLControlNetPipeline(
     """
 
     # leave controlnet out on purpose because it iterates with unet
-    model_cpu_offload_seq = "text_encoder->text_encoder_2->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
     _optional_components = [
         "tokenizer",
         "tokenizer_2",

src/diffusers/pipelines/deprecated/README.md

@@ -1,3 +0,0 @@
-# Deprecated Pipelines
-
-This folder contains pipelines that have very low usage as measured by model downloads, issues and PRs. While you can still use the pipelines just as before, we will stop testing the pipelines and will not accept any changes to existing files.

src/diffusers/pipelines/deprecated/__init__.py

@@ -1,153 +0,0 @@
-from typing import TYPE_CHECKING
-
-from ...utils import (
-    DIFFUSERS_SLOW_IMPORT,
-    OptionalDependencyNotAvailable,
-    _LazyModule,
-    get_objects_from_module,
-    is_librosa_available,
-    is_note_seq_available,
-    is_torch_available,
-    is_transformers_available,
-)
-
-
-_dummy_objects = {}
-_import_structure = {}
-
-try:
-    if not is_torch_available():
-        raise OptionalDependencyNotAvailable()
-except OptionalDependencyNotAvailable:
-    from ...utils import dummy_pt_objects
-
-    _dummy_objects.update(get_objects_from_module(dummy_pt_objects))
-else:
-    _import_structure["latent_diffusion_uncond"] = ["LDMPipeline"]
-    _import_structure["pndm"] = ["PNDMPipeline"]
-    _import_structure["repaint"] = ["RePaintPipeline"]
-    _import_structure["score_sde_ve"] = ["ScoreSdeVePipeline"]
-    _import_structure["stochastic_karras_ve"] = ["KarrasVePipeline"]
-
-try:
-    if not (is_transformers_available() and is_torch_available()):
-        raise OptionalDependencyNotAvailable()
-except OptionalDependencyNotAvailable:
-    from ...utils import dummy_torch_and_transformers_objects
-
-    _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
-else:
-    _import_structure["alt_diffusion"] = [
-        "AltDiffusionImg2ImgPipeline",
-        "AltDiffusionPipeline",
-        "AltDiffusionPipelineOutput",
-    ]
-    _import_structure["versatile_diffusion"] = [
-        "VersatileDiffusionDualGuidedPipeline",
-        "VersatileDiffusionImageVariationPipeline",
-        "VersatileDiffusionPipeline",
-        "VersatileDiffusionTextToImagePipeline",
-    ]
-    _import_structure["vq_diffusion"] = ["VQDiffusionPipeline"]
-    _import_structure["stable_diffusion_variants"] = [
-        "CycleDiffusionPipeline",
-        "StableDiffusionInpaintPipelineLegacy",
-        "StableDiffusionPix2PixZeroPipeline",
-        "StableDiffusionParadigmsPipeline",
-        "StableDiffusionModelEditingPipeline",
-    ]
-
-try:
-    if not (is_torch_available() and is_librosa_available()):
-        raise OptionalDependencyNotAvailable()
-except OptionalDependencyNotAvailable:
-    from ...utils import dummy_torch_and_librosa_objects  # noqa F403
-
-    _dummy_objects.update(get_objects_from_module(dummy_torch_and_librosa_objects))
-
-else:
-    _import_structure["audio_diffusion"] = ["AudioDiffusionPipeline", "Mel"]
-
-try:
-    if not (is_transformers_available() and is_torch_available() and is_note_seq_available()):
-        raise OptionalDependencyNotAvailable()
-except OptionalDependencyNotAvailable:
-    from ...utils import dummy_transformers_and_torch_and_note_seq_objects  # noqa F403
-
-    _dummy_objects.update(get_objects_from_module(dummy_transformers_and_torch_and_note_seq_objects))
-
-else:
-    _import_structure["spectrogram_diffusion"] = ["MidiProcessor", "SpectrogramDiffusionPipeline"]
-
-
-if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
-    try:
-        if not is_torch_available():
-            raise OptionalDependencyNotAvailable()
-    except OptionalDependencyNotAvailable:
-        from ...utils.dummy_pt_objects import *
-
-    else:
-        from .latent_diffusion_uncond import LDMPipeline
-        from .pndm import PNDMPipeline
-        from .repaint import RePaintPipeline
-        from .score_sde_ve import ScoreSdeVePipeline
-        from .stochastic_karras_ve import KarrasVePipeline
-
-    try:
-        if not (is_transformers_available() and is_torch_available()):
-            raise OptionalDependencyNotAvailable()
-    except OptionalDependencyNotAvailable:
-        from ...utils.dummy_torch_and_transformers_objects import *
-
-    else:
-        from .alt_diffusion import AltDiffusionImg2ImgPipeline, AltDiffusionPipeline, AltDiffusionPipelineOutput
-        from .audio_diffusion import AudioDiffusionPipeline, Mel
-        from .spectrogram_diffusion import SpectrogramDiffusionPipeline
-        from .stable_diffusion_variants import (
-            CycleDiffusionPipeline,
-            StableDiffusionInpaintPipelineLegacy,
-            StableDiffusionModelEditingPipeline,
-            StableDiffusionParadigmsPipeline,
-            StableDiffusionPix2PixZeroPipeline,
-        )
-        from .stochastic_karras_ve import KarrasVePipeline
-        from .versatile_diffusion import (
-            VersatileDiffusionDualGuidedPipeline,
-            VersatileDiffusionImageVariationPipeline,
-            VersatileDiffusionPipeline,
-            VersatileDiffusionTextToImagePipeline,
-        )
-        from .vq_diffusion import VQDiffusionPipeline
-
-    try:
-        if not (is_torch_available() and is_librosa_available()):
-            raise OptionalDependencyNotAvailable()
-    except OptionalDependencyNotAvailable:
-        from ...utils.dummy_torch_and_librosa_objects import *
-    else:
-        from .audio_diffusion import AudioDiffusionPipeline, Mel
-
-    try:
-        if not (is_transformers_available() and is_torch_available() and is_note_seq_available()):
-            raise OptionalDependencyNotAvailable()
-    except OptionalDependencyNotAvailable:
-        from ...utils.dummy_transformers_and_torch_and_note_seq_objects import *  # noqa F403
-    else:
-        from .spectrogram_diffusion import (
-            MidiProcessor,
-            SpectrogramDiffusionPipeline,
-        )
-
-
-else:
-    import sys
-
-    sys.modules[__name__] = _LazyModule(
-        __name__,
-        globals()["__file__"],
-        _import_structure,
-        module_spec=__spec__,
-    )
-    for name, value in _dummy_objects.items():
-        setattr(sys.modules[__name__], name, value)

src/diffusers/pipelines/deprecated/stable_diffusion_variants/__init__.py

@@ -1,55 +0,0 @@
-from typing import TYPE_CHECKING
-
-from ....utils import (
-    DIFFUSERS_SLOW_IMPORT,
-    OptionalDependencyNotAvailable,
-    _LazyModule,
-    get_objects_from_module,
-    is_torch_available,
-    is_transformers_available,
-)
-
-
-_dummy_objects = {}
-_import_structure = {}
-
-try:
-    if not (is_transformers_available() and is_torch_available()):
-        raise OptionalDependencyNotAvailable()
-except OptionalDependencyNotAvailable:
-    from ....utils import dummy_torch_and_transformers_objects
-
-    _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
-else:
-    _import_structure["pipeline_cycle_diffusion"] = ["CycleDiffusionPipeline"]
-    _import_structure["pipeline_stable_diffusion_inpaint_legacy"] = ["StableDiffusionInpaintPipelineLegacy"]
-    _import_structure["pipeline_stable_diffusion_model_editing"] = ["StableDiffusionModelEditingPipeline"]
-
-    _import_structure["pipeline_stable_diffusion_paradigms"] = ["StableDiffusionParadigmsPipeline"]
-    _import_structure["pipeline_stable_diffusion_pix2pix_zero"] = ["StableDiffusionPix2PixZeroPipeline"]
-
-if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
-    try:
-        if not (is_transformers_available() and is_torch_available()):
-            raise OptionalDependencyNotAvailable()
-    except OptionalDependencyNotAvailable:
-        from ....utils.dummy_torch_and_transformers_objects import *
-
-    else:
-        from .pipeline_cycle_diffusion import CycleDiffusionPipeline
-        from .pipeline_stable_diffusion_inpaint_legacy import StableDiffusionInpaintPipelineLegacy
-        from .pipeline_stable_diffusion_model_editing import StableDiffusionModelEditingPipeline
-        from .pipeline_stable_diffusion_paradigms import StableDiffusionParadigmsPipeline
-        from .pipeline_stable_diffusion_pix2pix_zero import StableDiffusionPix2PixZeroPipeline
-
-else:
-    import sys
-
-    sys.modules[__name__] = _LazyModule(
-        __name__,
-        globals()["__file__"],
-        _import_structure,
-        module_spec=__spec__,
-    )
-    for name, value in _dummy_objects.items():
-        setattr(sys.modules[__name__], name, value)

src/diffusers/pipelines/latent_diffusion_uncond/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
+from ...utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
 
 
 _import_structure = {"pipeline_latent_diffusion_uncond": ["LDMPipeline"]}

src/diffusers/pipelines/latent_diffusion_uncond/pipeline_latent_diffusion_uncond.py

@@ -17,10 +17,10 @@ from typing import List, Optional, Tuple, Union
 
 import torch
 
-from ....models import UNet2DModel, VQModel
-from ....schedulers import DDIMScheduler
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...models import UNet2DModel, VQModel
+from ...schedulers import DDIMScheduler
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 class LDMPipeline(DiffusionPipeline):

src/diffusers/pipelines/pndm/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
+from ...utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
 
 
 _import_structure = {"pipeline_pndm": ["PNDMPipeline"]}

src/diffusers/pipelines/pndm/pipeline_pndm.py

@@ -17,10 +17,10 @@ from typing import List, Optional, Tuple, Union
 
 import torch
 
-from ....models import UNet2DModel
-from ....schedulers import PNDMScheduler
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...models import UNet2DModel
+from ...schedulers import PNDMScheduler
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 class PNDMPipeline(DiffusionPipeline):

src/diffusers/pipelines/repaint/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
+from ...utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
 
 
 _import_structure = {"pipeline_repaint": ["RePaintPipeline"]}

src/diffusers/pipelines/repaint/pipeline_repaint.py

@@ -19,11 +19,11 @@ import numpy as np
 import PIL.Image
 import torch
 
-from ....models import UNet2DModel
-from ....schedulers import RePaintScheduler
-from ....utils import PIL_INTERPOLATION, deprecate, logging
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...models import UNet2DModel
+from ...schedulers import RePaintScheduler
+from ...utils import PIL_INTERPOLATION, deprecate, logging
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/score_sde_ve/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
+from ...utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
 
 
 _import_structure = {"pipeline_score_sde_ve": ["ScoreSdeVePipeline"]}

src/diffusers/pipelines/score_sde_ve/pipeline_score_sde_ve.py

@@ -16,10 +16,10 @@ from typing import List, Optional, Tuple, Union
 
 import torch
 
-from ....models import UNet2DModel
-from ....schedulers import ScoreSdeVeScheduler
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...models import UNet2DModel
+from ...schedulers import ScoreSdeVeScheduler
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 class ScoreSdeVePipeline(DiffusionPipeline):

src/diffusers/pipelines/spectrogram_diffusion/__init__.py

@@ -1,7 +1,7 @@
 # flake8: noqa
 from typing import TYPE_CHECKING
-from ....utils import (
-    DIFFUSERS_SLOW_IMPORT,
+from ...utils import DIFFUSERS_SLOW_IMPORT
+from ...utils import (
     _LazyModule,
     is_note_seq_available,
     OptionalDependencyNotAvailable,
@@ -17,7 +17,7 @@ try:
     if not (is_transformers_available() and is_torch_available()):
         raise OptionalDependencyNotAvailable()
 except OptionalDependencyNotAvailable:
-    from ....utils import dummy_torch_and_transformers_objects  # noqa F403
+    from ...utils import dummy_torch_and_transformers_objects  # noqa F403
 
     _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
 else:
@@ -32,7 +32,7 @@ try:
     if not (is_transformers_available() and is_torch_available() and is_note_seq_available()):
         raise OptionalDependencyNotAvailable()
 except OptionalDependencyNotAvailable:
-    from ....utils import dummy_transformers_and_torch_and_note_seq_objects
+    from ...utils import dummy_transformers_and_torch_and_note_seq_objects
 
     _dummy_objects.update(get_objects_from_module(dummy_transformers_and_torch_and_note_seq_objects))
 else:
@@ -45,7 +45,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             raise OptionalDependencyNotAvailable()
 
     except OptionalDependencyNotAvailable:
-        from ....utils.dummy_torch_and_transformers_objects import *
+        from ...utils.dummy_torch_and_transformers_objects import *
     else:
         from .pipeline_spectrogram_diffusion import SpectrogramDiffusionPipeline
         from .pipeline_spectrogram_diffusion import SpectrogramContEncoder
@@ -56,7 +56,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         if not (is_transformers_available() and is_torch_available() and is_note_seq_available()):
             raise OptionalDependencyNotAvailable()
     except OptionalDependencyNotAvailable:
-        from ....utils.dummy_transformers_and_torch_and_note_seq_objects import *
+        from ...utils.dummy_transformers_and_torch_and_note_seq_objects import *
 
     else:
         from .midi_utils import MidiProcessor

src/diffusers/pipelines/spectrogram_diffusion/continuous_encoder.py

@@ -22,8 +22,8 @@ from transformers.models.t5.modeling_t5 import (
     T5LayerNorm,
 )
 
-from ....configuration_utils import ConfigMixin, register_to_config
-from ....models import ModelMixin
+from ...configuration_utils import ConfigMixin, register_to_config
+from ...models import ModelMixin
 
 
 class SpectrogramContEncoder(ModelMixin, ConfigMixin, ModuleUtilsMixin):

src/diffusers/pipelines/spectrogram_diffusion/midi_utils.py

@@ -22,7 +22,7 @@ import numpy as np
 import torch
 import torch.nn.functional as F
 
-from ....utils import is_note_seq_available
+from ...utils import is_note_seq_available
 from .pipeline_spectrogram_diffusion import TARGET_FEATURE_LENGTH
 
 

src/diffusers/pipelines/spectrogram_diffusion/notes_encoder.py

@@ -18,8 +18,8 @@ import torch.nn as nn
 from transformers.modeling_utils import ModuleUtilsMixin
 from transformers.models.t5.modeling_t5 import T5Block, T5Config, T5LayerNorm
 
-from ....configuration_utils import ConfigMixin, register_to_config
-from ....models import ModelMixin
+from ...configuration_utils import ConfigMixin, register_to_config
+from ...models import ModelMixin
 
 
 class SpectrogramNotesEncoder(ModelMixin, ConfigMixin, ModuleUtilsMixin):

src/diffusers/pipelines/spectrogram_diffusion/pipeline_spectrogram_diffusion.py

@@ -19,16 +19,16 @@ from typing import Any, Callable, List, Optional, Tuple, Union
 import numpy as np
 import torch
 
-from ....models import T5FilmDecoder
-from ....schedulers import DDPMScheduler
-from ....utils import is_onnx_available, logging
-from ....utils.torch_utils import randn_tensor
+from ...models import T5FilmDecoder
+from ...schedulers import DDPMScheduler
+from ...utils import is_onnx_available, logging
+from ...utils.torch_utils import randn_tensor
 
 
 if is_onnx_available():
-    from ...onnx_utils import OnnxRuntimeModel
+    from ..onnx_utils import OnnxRuntimeModel
 
-from ...pipeline_utils import AudioPipelineOutput, DiffusionPipeline
+from ..pipeline_utils import AudioPipelineOutput, DiffusionPipeline
 from .continuous_encoder import SpectrogramContEncoder
 from .notes_encoder import SpectrogramNotesEncoder
 

src/diffusers/pipelines/stable_diffusion/__init__.py

@@ -134,6 +134,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
 
     else:
         from .clip_image_project_model import CLIPImageProjection
+        from .pipeline_cycle_diffusion import CycleDiffusionPipeline
         from .pipeline_stable_diffusion import (
             StableDiffusionPipeline,
             StableDiffusionPipelineOutput,
@@ -148,6 +149,9 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         )
         from .pipeline_stable_diffusion_img2img import StableDiffusionImg2ImgPipeline
         from .pipeline_stable_diffusion_inpaint import StableDiffusionInpaintPipeline
+        from .pipeline_stable_diffusion_inpaint_legacy import (
+            StableDiffusionInpaintPipelineLegacy,
+        )
         from .pipeline_stable_diffusion_instruct_pix2pix import (
             StableDiffusionInstructPix2PixPipeline,
         )
@@ -155,7 +159,13 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             StableDiffusionLatentUpscalePipeline,
         )
         from .pipeline_stable_diffusion_ldm3d import StableDiffusionLDM3DPipeline
+        from .pipeline_stable_diffusion_model_editing import (
+            StableDiffusionModelEditingPipeline,
+        )
         from .pipeline_stable_diffusion_panorama import StableDiffusionPanoramaPipeline
+        from .pipeline_stable_diffusion_paradigms import (
+            StableDiffusionParadigmsPipeline,
+        )
         from .pipeline_stable_diffusion_sag import StableDiffusionSAGPipeline
         from .pipeline_stable_diffusion_upscale import StableDiffusionUpscalePipeline
         from .pipeline_stable_unclip import StableUnCLIPPipeline
@@ -189,6 +199,9 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             StableDiffusionDepth2ImgPipeline,
         )
         from .pipeline_stable_diffusion_diffedit import StableDiffusionDiffEditPipeline
+        from .pipeline_stable_diffusion_pix2pix_zero import (
+            StableDiffusionPix2PixZeroPipeline,
+        )
 
     try:
         if not (
@@ -221,6 +234,9 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         from .pipeline_onnx_stable_diffusion_inpaint import (
             OnnxStableDiffusionInpaintPipeline,
         )
+        from .pipeline_onnx_stable_diffusion_inpaint_legacy import (
+            OnnxStableDiffusionInpaintPipelineLegacy,
+        )
         from .pipeline_onnx_stable_diffusion_upscale import (
             OnnxStableDiffusionUpscalePipeline,
         )

src/diffusers/pipelines/stable_diffusion/pipeline_cycle_diffusion.py

@@ -21,17 +21,17 @@ import torch
 from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
 
-from ....configuration_utils import FrozenDict
-from ....image_processor import PipelineImageInput, VaeImageProcessor
-from ....loaders import LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import DDIMScheduler
-from ....utils import PIL_INTERPOLATION, USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...configuration_utils import FrozenDict
+from ...image_processor import PipelineImageInput, VaeImageProcessor
+from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import DDIMScheduler
+from ...utils import PIL_INTERPOLATION, USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from .pipeline_output import StableDiffusionPipelineOutput
+from .safety_checker import StableDiffusionSafetyChecker
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/stable_diffusion/pipeline_onnx_stable_diffusion_inpaint_legacy.py

@@ -6,12 +6,12 @@ import PIL.Image
 import torch
 from transformers import CLIPImageProcessor, CLIPTokenizer
 
-from ....configuration_utils import FrozenDict
-from ....schedulers import DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler
-from ....utils import deprecate, logging
-from ...onnx_utils import ORT_TO_NP_TYPE, OnnxRuntimeModel
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
+from ...configuration_utils import FrozenDict
+from ...schedulers import DDIMScheduler, LMSDiscreteScheduler, PNDMScheduler
+from ...utils import deprecate, logging
+from ..onnx_utils import ORT_TO_NP_TYPE, OnnxRuntimeModel
+from ..pipeline_utils import DiffusionPipeline
+from . import StableDiffusionPipelineOutput
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py

@@ -151,7 +151,7 @@ class StableDiffusionPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_diffedit.py

@@ -788,6 +788,7 @@ class StableDiffusionDiffEditPipeline(DiffusionPipeline, TextualInversionLoaderM
         latents = latents * self.scheduler.init_noise_sigma
         return latents
 
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_pix2pix_zero.StableDiffusionPix2PixZeroPipeline.prepare_image_latents
     def prepare_image_latents(self, image, batch_size, dtype, device, generator=None):
         if not isinstance(image, (torch.Tensor, PIL.Image.Image, list)):
             raise ValueError(

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py

@@ -191,7 +191,7 @@ class StableDiffusionImg2ImgPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py

@@ -255,7 +255,7 @@ class StableDiffusionInpaintPipeline(
             A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
     """
 
-    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->unet->vae"
     _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
     _exclude_from_cpu_offload = ["safety_checker"]
     _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds", "mask", "masked_image_latents"]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint_legacy.py

@@ -21,17 +21,17 @@ import torch
 from packaging import version
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
 
-from ....configuration_utils import FrozenDict
-from ....image_processor import VaeImageProcessor
-from ....loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import PIL_INTERPOLATION, USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion import StableDiffusionPipelineOutput
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...configuration_utils import FrozenDict
+from ...image_processor import VaeImageProcessor
+from ...loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import PIL_INTERPOLATION, USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from . import StableDiffusionPipelineOutput
+from .safety_checker import StableDiffusionSafetyChecker
 
 
 logger = logging.get_logger(__name__)

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_model_editing.py

@@ -18,17 +18,17 @@ from typing import Any, Callable, Dict, List, Optional, Union
 import torch
 from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
 
-from ....image_processor import VaeImageProcessor
-from ....loaders import LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import PNDMScheduler
-from ....schedulers.scheduling_utils import SchedulerMixin
-from ....utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...image_processor import VaeImageProcessor
+from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import PNDMScheduler
+from ...schedulers.scheduling_utils import SchedulerMixin
+from ...utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from . import StableDiffusionPipelineOutput
+from .safety_checker import StableDiffusionSafetyChecker
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_paradigms.py

@@ -18,12 +18,12 @@ from typing import Any, Callable, Dict, List, Optional, Union
 import torch
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
 
-from ....image_processor import VaeImageProcessor
-from ....loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import (
+from ...image_processor import VaeImageProcessor
+from ...loaders import FromSingleFileMixin, LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import (
     USE_PEFT_BACKEND,
     deprecate,
     logging,
@@ -31,10 +31,10 @@ from ....utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from . import StableDiffusionPipelineOutput
+from .safety_checker import StableDiffusionSafetyChecker
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_pix2pix_zero.py

@@ -28,14 +28,14 @@ from transformers import (
     CLIPTokenizer,
 )
 
-from ....image_processor import PipelineImageInput, VaeImageProcessor
-from ....loaders import LoraLoaderMixin, TextualInversionLoaderMixin
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....models.attention_processor import Attention
-from ....models.lora import adjust_lora_scale_text_encoder
-from ....schedulers import DDIMScheduler, DDPMScheduler, EulerAncestralDiscreteScheduler, LMSDiscreteScheduler
-from ....schedulers.scheduling_ddim_inverse import DDIMInverseScheduler
-from ....utils import (
+from ...image_processor import PipelineImageInput, VaeImageProcessor
+from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...models.attention_processor import Attention
+from ...models.lora import adjust_lora_scale_text_encoder
+from ...schedulers import DDIMScheduler, DDPMScheduler, EulerAncestralDiscreteScheduler, LMSDiscreteScheduler
+from ...schedulers.scheduling_ddim_inverse import DDIMInverseScheduler
+from ...utils import (
     PIL_INTERPOLATION,
     USE_PEFT_BACKEND,
     BaseOutput,
@@ -45,10 +45,10 @@ from ....utils import (
     scale_lora_layers,
     unscale_lora_layers,
 )
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline
-from ...stable_diffusion.pipeline_output import StableDiffusionPipelineOutput
-from ...stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline
+from . import StableDiffusionPipelineOutput
+from .safety_checker import StableDiffusionSafetyChecker
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py

@@ -198,7 +198,7 @@ class StableDiffusionXLPipeline(
             watermarker will be used.
     """
 
-    model_cpu_offload_seq = "text_encoder->text_encoder_2->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
     _optional_components = [
         "tokenizer",
         "tokenizer_2",

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py

@@ -219,7 +219,7 @@ class StableDiffusionXLImg2ImgPipeline(
             watermarker will be used.
     """
 
-    model_cpu_offload_seq = "text_encoder->text_encoder_2->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
     _optional_components = [
         "tokenizer",
         "tokenizer_2",

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py

@@ -364,7 +364,7 @@ class StableDiffusionXLInpaintPipeline(
             watermarker will be used.
     """
 
-    model_cpu_offload_seq = "text_encoder->text_encoder_2->image_encoder->unet->vae"
+    model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
 
     _optional_components = [
         "tokenizer",

src/diffusers/pipelines/stable_video_diffusion/pipeline_stable_video_diffusion.py

@@ -25,7 +25,7 @@ from ...image_processor import VaeImageProcessor
 from ...models import AutoencoderKLTemporalDecoder, UNetSpatioTemporalConditionModel
 from ...schedulers import EulerDiscreteScheduler
 from ...utils import BaseOutput, logging
-from ...utils.torch_utils import is_compiled_module, randn_tensor
+from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline
 
 
@@ -211,8 +211,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
 
         latents = 1 / self.vae.config.scaling_factor * latents
 
-        forward_vae_fn = self.vae._orig_mod.forward if is_compiled_module(self.vae) else self.vae.forward
-        accepts_num_frames = "num_frames" in set(inspect.signature(forward_vae_fn).parameters.keys())
+        accepts_num_frames = "num_frames" in set(inspect.signature(self.vae.forward).parameters.keys())
 
         # decode decode_chunk_size frames at a time to avoid OOM
         frames = []
@@ -291,9 +290,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
     # corresponds to doing no classifier free guidance.
     @property
     def do_classifier_free_guidance(self):
-        if isinstance(self.guidance_scale, (int, float)):
-            return self.guidance_scale
-        return self.guidance_scale.max() > 1
+        return self._guidance_scale > 1 and self.unet.config.time_cond_proj_dim is None
 
     @property
     def num_timesteps(self):
@@ -418,10 +415,10 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
         # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
         # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
         # corresponds to doing no classifier free guidance.
-        self._guidance_scale = max_guidance_scale
+        do_classifier_free_guidance = max_guidance_scale > 1.0
 
         # 3. Encode input image
-        image_embeddings = self._encode_image(image, device, num_videos_per_prompt, self.do_classifier_free_guidance)
+        image_embeddings = self._encode_image(image, device, num_videos_per_prompt, do_classifier_free_guidance)
 
         # NOTE: Stable Diffusion Video was conditioned on fps - 1, which
         # is why it is reduced here.
@@ -437,7 +434,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
         if needs_upcasting:
             self.vae.to(dtype=torch.float32)
 
-        image_latents = self._encode_vae_image(image, device, num_videos_per_prompt, self.do_classifier_free_guidance)
+        image_latents = self._encode_vae_image(image, device, num_videos_per_prompt, do_classifier_free_guidance)
         image_latents = image_latents.to(image_embeddings.dtype)
 
         # cast back to fp16 if needed
@@ -456,7 +453,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
             image_embeddings.dtype,
             batch_size,
             num_videos_per_prompt,
-            self.do_classifier_free_guidance,
+            do_classifier_free_guidance,
         )
         added_time_ids = added_time_ids.to(device)
 
@@ -492,7 +489,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 # expand the latents if we are doing classifier free guidance
-                latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
+                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
                 latent_model_input = self.scheduler.scale_model_input(latent_model_input, t)
 
                 # Concatenate image_latents over channels dimention
@@ -508,7 +505,7 @@ class StableVideoDiffusionPipeline(DiffusionPipeline):
                 )[0]
 
                 # perform guidance
-                if self.do_classifier_free_guidance:
+                if do_classifier_free_guidance:
                     noise_pred_uncond, noise_pred_cond = noise_pred.chunk(2)
                     noise_pred = noise_pred_uncond + self.guidance_scale * (noise_pred_cond - noise_pred_uncond)
 

src/diffusers/pipelines/stochastic_karras_ve/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
+from ...utils import DIFFUSERS_SLOW_IMPORT, _LazyModule
 
 
 _import_structure = {"pipeline_stochastic_karras_ve": ["KarrasVePipeline"]}

src/diffusers/pipelines/stochastic_karras_ve/pipeline_stochastic_karras_ve.py

@@ -16,10 +16,10 @@ from typing import List, Optional, Tuple, Union
 
 import torch
 
-from ....models import UNet2DModel
-from ....schedulers import KarrasVeScheduler
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...models import UNet2DModel
+from ...schedulers import KarrasVeScheduler
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 class KarrasVePipeline(DiffusionPipeline):

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero.py

@@ -1,5 +1,4 @@
 import copy
-import inspect
 from dataclasses import dataclass
 from typing import Callable, List, Optional, Union
 
@@ -10,18 +9,11 @@ import torch.nn.functional as F
 from torch.nn.functional import grid_sample
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer
 
-from ...image_processor import VaeImageProcessor
-from ...loaders import LoraLoaderMixin, TextualInversionLoaderMixin
-from ...models import AutoencoderKL, UNet2DConditionModel
-from ...models.lora import adjust_lora_scale_text_encoder
-from ...schedulers import KarrasDiffusionSchedulers
-from ...utils import USE_PEFT_BACKEND, BaseOutput, logging, scale_lora_layers, unscale_lora_layers
-from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline
-from ..stable_diffusion import StableDiffusionSafetyChecker
-
-
-logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+from diffusers.models import AutoencoderKL, UNet2DConditionModel
+from diffusers.pipelines.stable_diffusion import StableDiffusionPipeline, StableDiffusionSafetyChecker
+from diffusers.schedulers import KarrasDiffusionSchedulers
+from diffusers.utils import BaseOutput
+from diffusers.utils.torch_utils import randn_tensor
 
 
 def rearrange_0(tensor, f):
@@ -281,7 +273,7 @@ def create_motion_field_and_warp_latents(motion_field_strength_x, motion_field_s
     return warped_latents
 
 
-class TextToVideoZeroPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraLoaderMixin):
+class TextToVideoZeroPipeline(StableDiffusionPipeline):
     r"""
     Pipeline for zero-shot text-to-video generation using Stable Diffusion.
 
@@ -319,15 +311,8 @@ class TextToVideoZeroPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
         feature_extractor: CLIPImageProcessor,
         requires_safety_checker: bool = True,
     ):
-        super().__init__()
-        self.register_modules(
-            vae=vae,
-            text_encoder=text_encoder,
-            tokenizer=tokenizer,
-            unet=unet,
-            scheduler=scheduler,
-            safety_checker=safety_checker,
-            feature_extractor=feature_extractor,
+        super().__init__(
+            vae, text_encoder, tokenizer, unet, scheduler, safety_checker, feature_extractor, requires_safety_checker
         )
         processor = (
             CrossFrameAttnProcessor2_0(batch_size=2)
@@ -336,18 +321,6 @@ class TextToVideoZeroPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
         )
         self.unet.set_attn_processor(processor)
 
-        if safety_checker is None and requires_safety_checker:
-            logger.warning(
-                f"You have disabled the safety checker for {self.__class__} by passing `safety_checker=None`. Ensure"
-                " that you abide to the conditions of the Stable Diffusion license and do not expose unfiltered"
-                " results in services or applications open to the public. Both the diffusers team and Hugging Face"
-                " strongly recommend to keep the safety filter enabled in all public facing circumstances, disabling"
-                " it only for use-cases that involve analyzing network behavior or auditing its results. For more"
-                " information, please have a look at https://github.com/huggingface/diffusers/pull/254 ."
-            )
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
-        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
-
     def forward_loop(self, x_t0, t0, t1, generator):
         """
         Perform DDPM forward process from time t0 to t1. This is the same as adding noise with corresponding variance.
@@ -447,77 +420,6 @@ class TextToVideoZeroPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
                         callback(step_idx, t, latents)
         return latents.clone().detach()
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.check_inputs
-    def check_inputs(
-        self,
-        prompt,
-        height,
-        width,
-        callback_steps,
-        negative_prompt=None,
-        prompt_embeds=None,
-        negative_prompt_embeds=None,
-        callback_on_step_end_tensor_inputs=None,
-    ):
-        if height % 8 != 0 or width % 8 != 0:
-            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
-
-        if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
-            raise ValueError(
-                f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
-                f" {type(callback_steps)}."
-            )
-        if callback_on_step_end_tensor_inputs is not None and not all(
-            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
-        ):
-            raise ValueError(
-                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
-            )
-
-        if prompt is not None and prompt_embeds is not None:
-            raise ValueError(
-                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
-                " only forward one of the two."
-            )
-        elif prompt is None and prompt_embeds is None:
-            raise ValueError(
-                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
-            )
-        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
-            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
-
-        if negative_prompt is not None and negative_prompt_embeds is not None:
-            raise ValueError(
-                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:"
-                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
-            )
-
-        if prompt_embeds is not None and negative_prompt_embeds is not None:
-            if prompt_embeds.shape != negative_prompt_embeds.shape:
-                raise ValueError(
-                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
-                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
-                    f" {negative_prompt_embeds.shape}."
-                )
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
-    def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
-        shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
-        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)
-        else:
-            latents = latents.to(device)
-
-        # scale the initial noise by the standard deviation required by the scheduler
-        latents = latents * self.scheduler.init_noise_sigma
-        return latents
-
     @torch.no_grad()
     def __call__(
         self,
@@ -637,10 +539,9 @@ class TextToVideoZeroPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
         do_classifier_free_guidance = guidance_scale > 1.0
 
         # Encode input prompt
-        prompt_embeds_tuple = self.encode_prompt(
+        prompt_embeds = self._encode_prompt(
             prompt, device, num_videos_per_prompt, do_classifier_free_guidance, negative_prompt
         )
-        prompt_embeds = torch.cat([prompt_embeds_tuple[1], prompt_embeds_tuple[0]])
 
         # Prepare timesteps
         self.scheduler.set_timesteps(num_inference_steps, device=device)
@@ -744,226 +645,3 @@ class TextToVideoZeroPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lo
             return (image, has_nsfw_concept)
 
         return TextToVideoPipelineOutput(images=image, nsfw_content_detected=has_nsfw_concept)
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.run_safety_checker
-    def run_safety_checker(self, image, device, dtype):
-        if self.safety_checker is None:
-            has_nsfw_concept = None
-        else:
-            if torch.is_tensor(image):
-                feature_extractor_input = self.image_processor.postprocess(image, output_type="pil")
-            else:
-                feature_extractor_input = self.image_processor.numpy_to_pil(image)
-            safety_checker_input = self.feature_extractor(feature_extractor_input, return_tensors="pt").to(device)
-            image, has_nsfw_concept = self.safety_checker(
-                images=image, clip_input=safety_checker_input.pixel_values.to(dtype)
-            )
-        return image, has_nsfw_concept
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
-    def prepare_extra_step_kwargs(self, generator, eta):
-        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
-        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
-        # and should be between [0, 1]
-
-        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
-        extra_step_kwargs = {}
-        if accepts_eta:
-            extra_step_kwargs["eta"] = eta
-
-        # check if the scheduler accepts generator
-        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
-        if accepts_generator:
-            extra_step_kwargs["generator"] = generator
-        return extra_step_kwargs
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.encode_prompt
-    def encode_prompt(
-        self,
-        prompt,
-        device,
-        num_images_per_prompt,
-        do_classifier_free_guidance,
-        negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        lora_scale: Optional[float] = None,
-        clip_skip: Optional[int] = None,
-    ):
-        r"""
-        Encodes the prompt into text encoder hidden states.
-
-        Args:
-            prompt (`str` or `List[str]`, *optional*):
-                prompt to be encoded
-            device: (`torch.device`):
-                torch device
-            num_images_per_prompt (`int`):
-                number of images that should be generated per prompt
-            do_classifier_free_guidance (`bool`):
-                whether to use classifier free guidance or not
-            negative_prompt (`str` or `List[str]`, *optional*):
-                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`).
-            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.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
-                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
-                weighting. If not provided, negative_prompt_embeds will be generated from `negative_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.
-            clip_skip (`int`, *optional*):
-                Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
-                the output of the pre-final layer will be used for computing the prompt embeddings.
-        """
-        # 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, LoraLoaderMixin):
-            self._lora_scale = lora_scale
-
-            # dynamically adjust the LoRA scale
-            if not USE_PEFT_BACKEND:
-                adjust_lora_scale_text_encoder(self.text_encoder, lora_scale)
-            else:
-                scale_lora_layers(self.text_encoder, lora_scale)
-
-        if prompt is not None and isinstance(prompt, str):
-            batch_size = 1
-        elif prompt is not None and isinstance(prompt, list):
-            batch_size = len(prompt)
-        else:
-            batch_size = prompt_embeds.shape[0]
-
-        if prompt_embeds is None:
-            # textual inversion: procecss multi-vector tokens if necessary
-            if isinstance(self, TextualInversionLoaderMixin):
-                prompt = self.maybe_convert_prompt(prompt, self.tokenizer)
-
-            text_inputs = self.tokenizer(
-                prompt,
-                padding="max_length",
-                max_length=self.tokenizer.model_max_length,
-                truncation=True,
-                return_tensors="pt",
-            )
-            text_input_ids = text_inputs.input_ids
-            untruncated_ids = self.tokenizer(prompt, padding="longest", return_tensors="pt").input_ids
-
-            if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
-                text_input_ids, untruncated_ids
-            ):
-                removed_text = self.tokenizer.batch_decode(
-                    untruncated_ids[:, self.tokenizer.model_max_length - 1 : -1]
-                )
-                logger.warning(
-                    "The following part of your input was truncated because CLIP can only handle sequences up to"
-                    f" {self.tokenizer.model_max_length} tokens: {removed_text}"
-                )
-
-            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
-                attention_mask = text_inputs.attention_mask.to(device)
-            else:
-                attention_mask = None
-
-            if clip_skip is None:
-                prompt_embeds = self.text_encoder(text_input_ids.to(device), attention_mask=attention_mask)
-                prompt_embeds = prompt_embeds[0]
-            else:
-                prompt_embeds = self.text_encoder(
-                    text_input_ids.to(device), attention_mask=attention_mask, output_hidden_states=True
-                )
-                # Access the `hidden_states` first, that contains a tuple of
-                # all the hidden states from the encoder layers. Then index into
-                # the tuple to access the hidden states from the desired layer.
-                prompt_embeds = prompt_embeds[-1][-(clip_skip + 1)]
-                # We also need to apply the final LayerNorm here to not mess with the
-                # representations. The `last_hidden_states` that we typically use for
-                # obtaining the final prompt representations passes through the LayerNorm
-                # layer.
-                prompt_embeds = self.text_encoder.text_model.final_layer_norm(prompt_embeds)
-
-        if self.text_encoder is not None:
-            prompt_embeds_dtype = self.text_encoder.dtype
-        elif self.unet is not None:
-            prompt_embeds_dtype = self.unet.dtype
-        else:
-            prompt_embeds_dtype = prompt_embeds.dtype
-
-        prompt_embeds = prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)
-
-        bs_embed, seq_len, _ = prompt_embeds.shape
-        # duplicate text embeddings for each generation per prompt, using mps friendly method
-        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
-        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
-
-        # get unconditional embeddings for classifier free guidance
-        if do_classifier_free_guidance and negative_prompt_embeds is None:
-            uncond_tokens: List[str]
-            if negative_prompt is None:
-                uncond_tokens = [""] * batch_size
-            elif prompt is not None and type(prompt) is not type(negative_prompt):
-                raise TypeError(
-                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
-                    f" {type(prompt)}."
-                )
-            elif isinstance(negative_prompt, str):
-                uncond_tokens = [negative_prompt]
-            elif batch_size != len(negative_prompt):
-                raise ValueError(
-                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
-                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
-                    " the batch size of `prompt`."
-                )
-            else:
-                uncond_tokens = negative_prompt
-
-            # textual inversion: procecss multi-vector tokens if necessary
-            if isinstance(self, TextualInversionLoaderMixin):
-                uncond_tokens = self.maybe_convert_prompt(uncond_tokens, self.tokenizer)
-
-            max_length = prompt_embeds.shape[1]
-            uncond_input = self.tokenizer(
-                uncond_tokens,
-                padding="max_length",
-                max_length=max_length,
-                truncation=True,
-                return_tensors="pt",
-            )
-
-            if hasattr(self.text_encoder.config, "use_attention_mask") and self.text_encoder.config.use_attention_mask:
-                attention_mask = uncond_input.attention_mask.to(device)
-            else:
-                attention_mask = None
-
-            negative_prompt_embeds = self.text_encoder(
-                uncond_input.input_ids.to(device),
-                attention_mask=attention_mask,
-            )
-            negative_prompt_embeds = negative_prompt_embeds[0]
-
-        if do_classifier_free_guidance:
-            # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
-            seq_len = negative_prompt_embeds.shape[1]
-
-            negative_prompt_embeds = negative_prompt_embeds.to(dtype=prompt_embeds_dtype, device=device)
-
-            negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
-            negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
-
-        if isinstance(self, LoraLoaderMixin) and USE_PEFT_BACKEND:
-            # Retrieve the original scale by scaling back the LoRA layers
-            unscale_lora_layers(self.text_encoder, lora_scale)
-
-        return prompt_embeds, negative_prompt_embeds
-
-    def decode_latents(self, latents):
-        latents = 1 / self.vae.config.scaling_factor * latents
-        image = self.vae.decode(latents, return_dict=False)[0]
-        image = (image / 2 + 0.5).clamp(0, 1)
-        # we always cast to float32 as this does not cause significant overhead and is compatible with bfloat16
-        image = image.cpu().permute(0, 2, 3, 1).float().numpy()
-        return image

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero_sdxl.py

@@ -1,5 +1,4 @@
 import copy
-import inspect
 from dataclasses import dataclass
 from typing import Any, Callable, Dict, List, Optional, Tuple, Union
 
@@ -16,35 +15,11 @@ from transformers import (
     CLIPVisionModelWithProjection,
 )
 
-from ...image_processor import VaeImageProcessor
-from ...loaders import StableDiffusionXLLoraLoaderMixin, TextualInversionLoaderMixin
-from ...models import AutoencoderKL, UNet2DConditionModel
-from ...models.attention_processor import (
-    AttnProcessor2_0,
-    FusedAttnProcessor2_0,
-    LoRAAttnProcessor2_0,
-    LoRAXFormersAttnProcessor,
-    XFormersAttnProcessor,
-)
-from ...models.lora import adjust_lora_scale_text_encoder
-from ...schedulers import KarrasDiffusionSchedulers
-from ...utils import (
-    USE_PEFT_BACKEND,
-    BaseOutput,
-    is_invisible_watermark_available,
-    logging,
-    scale_lora_layers,
-    unscale_lora_layers,
-)
-from ...utils.torch_utils import randn_tensor
-from ..pipeline_utils import DiffusionPipeline
-
-
-if is_invisible_watermark_available():
-    from ..stable_diffusion_xl.watermark import StableDiffusionXLWatermarker
-
-
-logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+from diffusers.models import AutoencoderKL, UNet2DConditionModel
+from diffusers.pipelines.stable_diffusion_xl import StableDiffusionXLPipeline
+from diffusers.schedulers import KarrasDiffusionSchedulers
+from diffusers.utils import BaseOutput
+from diffusers.utils.torch_utils import randn_tensor
 
 
 # Copied from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero.rearrange_0
@@ -325,11 +300,7 @@ def rescale_noise_cfg(noise_cfg, noise_pred_text, guidance_rescale=0.0):
     return noise_cfg
 
 
-class TextToVideoZeroSDXLPipeline(
-    DiffusionPipeline,
-    StableDiffusionXLLoraLoaderMixin,
-    TextualInversionLoaderMixin,
-):
+class TextToVideoZeroSDXLPipeline(StableDiffusionXLPipeline):
     r"""
     Pipeline for zero-shot text-to-video generation using Stable Diffusion XL.
 
@@ -361,16 +332,6 @@ class TextToVideoZeroSDXLPipeline(
             [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
     """
 
-    model_cpu_offload_seq = "text_encoder->text_encoder_2->unet->vae"
-    _optional_components = [
-        "tokenizer",
-        "tokenizer_2",
-        "text_encoder",
-        "text_encoder_2",
-        "image_encoder",
-        "feature_extractor",
-    ]
-
     def __init__(
         self,
         vae: AutoencoderKL,
@@ -385,8 +346,7 @@ class TextToVideoZeroSDXLPipeline(
         force_zeros_for_empty_prompt: bool = True,
         add_watermarker: Optional[bool] = None,
     ):
-        super().__init__()
-        self.register_modules(
+        super().__init__(
             vae=vae,
             text_encoder=text_encoder,
             text_encoder_2=text_encoder_2,
@@ -396,435 +356,16 @@ class TextToVideoZeroSDXLPipeline(
             scheduler=scheduler,
             image_encoder=image_encoder,
             feature_extractor=feature_extractor,
+            force_zeros_for_empty_prompt=force_zeros_for_empty_prompt,
+            add_watermarker=add_watermarker,
         )
-        self.register_to_config(force_zeros_for_empty_prompt=force_zeros_for_empty_prompt)
-        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
-        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
-
-        self.default_sample_size = self.unet.config.sample_size
-
-        add_watermarker = add_watermarker if add_watermarker is not None else is_invisible_watermark_available()
-
-        if add_watermarker:
-            self.watermark = StableDiffusionXLWatermarker()
-        else:
-            self.watermark = None
-
         processor = (
             CrossFrameAttnProcessor2_0(batch_size=2)
             if hasattr(F, "scaled_dot_product_attention")
             else CrossFrameAttnProcessor(batch_size=2)
         )
-
         self.unet.set_attn_processor(processor)
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_extra_step_kwargs
-    def prepare_extra_step_kwargs(self, generator, eta):
-        # prepare extra kwargs for the scheduler step, since not all schedulers have the same signature
-        # eta (η) is only used with the DDIMScheduler, it will be ignored for other schedulers.
-        # eta corresponds to η in DDIM paper: https://arxiv.org/abs/2010.02502
-        # and should be between [0, 1]
-
-        accepts_eta = "eta" in set(inspect.signature(self.scheduler.step).parameters.keys())
-        extra_step_kwargs = {}
-        if accepts_eta:
-            extra_step_kwargs["eta"] = eta
-
-        # check if the scheduler accepts generator
-        accepts_generator = "generator" in set(inspect.signature(self.scheduler.step).parameters.keys())
-        if accepts_generator:
-            extra_step_kwargs["generator"] = generator
-        return extra_step_kwargs
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.enable_vae_slicing
-    def enable_vae_slicing(self):
-        r"""
-        Enable sliced VAE decoding. When this option is enabled, the VAE will split the input tensor in slices to
-        compute decoding in several steps. This is useful to save some memory and allow larger batch sizes.
-        """
-        self.vae.enable_slicing()
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.disable_vae_slicing
-    def disable_vae_slicing(self):
-        r"""
-        Disable sliced VAE decoding. If `enable_vae_slicing` was previously enabled, this method will go back to
-        computing decoding in one step.
-        """
-        self.vae.disable_slicing()
-
-    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.upcast_vae
-    def upcast_vae(self):
-        dtype = self.vae.dtype
-        self.vae.to(dtype=torch.float32)
-        use_torch_2_0_or_xformers = isinstance(
-            self.vae.decoder.mid_block.attentions[0].processor,
-            (
-                AttnProcessor2_0,
-                XFormersAttnProcessor,
-                LoRAXFormersAttnProcessor,
-                LoRAAttnProcessor2_0,
-                FusedAttnProcessor2_0,
-            ),
-        )
-        # if xformers or torch_2_0 is used attention block does not need
-        # to be in float32 which can save lots of memory
-        if use_torch_2_0_or_xformers:
-            self.vae.post_quant_conv.to(dtype)
-            self.vae.decoder.conv_in.to(dtype)
-            self.vae.decoder.mid_block.to(dtype)
-
-    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline._get_add_time_ids
-    def _get_add_time_ids(
-        self, original_size, crops_coords_top_left, target_size, dtype, text_encoder_projection_dim=None
-    ):
-        add_time_ids = list(original_size + crops_coords_top_left + target_size)
-
-        passed_add_embed_dim = (
-            self.unet.config.addition_time_embed_dim * len(add_time_ids) + text_encoder_projection_dim
-        )
-        expected_add_embed_dim = self.unet.add_embedding.linear_1.in_features
-
-        if expected_add_embed_dim != passed_add_embed_dim:
-            raise ValueError(
-                f"Model expects an added time embedding vector of length {expected_add_embed_dim}, but a vector of {passed_add_embed_dim} was created. The model has an incorrect config. Please check `unet.config.time_embedding_type` and `text_encoder_2.config.projection_dim`."
-            )
-
-        add_time_ids = torch.tensor([add_time_ids], dtype=dtype)
-        return add_time_ids
-
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.prepare_latents
-    def prepare_latents(self, batch_size, num_channels_latents, height, width, dtype, device, generator, latents=None):
-        shape = (batch_size, num_channels_latents, height // self.vae_scale_factor, width // self.vae_scale_factor)
-        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)
-        else:
-            latents = latents.to(device)
-
-        # scale the initial noise by the standard deviation required by the scheduler
-        latents = latents * self.scheduler.init_noise_sigma
-        return latents
-
-    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.check_inputs
-    def check_inputs(
-        self,
-        prompt,
-        prompt_2,
-        height,
-        width,
-        callback_steps,
-        negative_prompt=None,
-        negative_prompt_2=None,
-        prompt_embeds=None,
-        negative_prompt_embeds=None,
-        pooled_prompt_embeds=None,
-        negative_pooled_prompt_embeds=None,
-        callback_on_step_end_tensor_inputs=None,
-    ):
-        if height % 8 != 0 or width % 8 != 0:
-            raise ValueError(f"`height` and `width` have to be divisible by 8 but are {height} and {width}.")
-
-        if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
-            raise ValueError(
-                f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
-                f" {type(callback_steps)}."
-            )
-
-        if callback_on_step_end_tensor_inputs is not None and not all(
-            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
-        ):
-            raise ValueError(
-                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
-            )
-
-        if prompt is not None and prompt_embeds is not None:
-            raise ValueError(
-                f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
-                " only forward one of the two."
-            )
-        elif prompt_2 is not None and prompt_embeds is not None:
-            raise ValueError(
-                f"Cannot forward both `prompt_2`: {prompt_2} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
-                " only forward one of the two."
-            )
-        elif prompt is None and prompt_embeds is None:
-            raise ValueError(
-                "Provide either `prompt` or `prompt_embeds`. Cannot leave both `prompt` and `prompt_embeds` undefined."
-            )
-        elif prompt is not None and (not isinstance(prompt, str) and not isinstance(prompt, list)):
-            raise ValueError(f"`prompt` has to be of type `str` or `list` but is {type(prompt)}")
-        elif prompt_2 is not None and (not isinstance(prompt_2, str) and not isinstance(prompt_2, list)):
-            raise ValueError(f"`prompt_2` has to be of type `str` or `list` but is {type(prompt_2)}")
-
-        if negative_prompt is not None and negative_prompt_embeds is not None:
-            raise ValueError(
-                f"Cannot forward both `negative_prompt`: {negative_prompt} and `negative_prompt_embeds`:"
-                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
-            )
-        elif negative_prompt_2 is not None and negative_prompt_embeds is not None:
-            raise ValueError(
-                f"Cannot forward both `negative_prompt_2`: {negative_prompt_2} and `negative_prompt_embeds`:"
-                f" {negative_prompt_embeds}. Please make sure to only forward one of the two."
-            )
-
-        if prompt_embeds is not None and negative_prompt_embeds is not None:
-            if prompt_embeds.shape != negative_prompt_embeds.shape:
-                raise ValueError(
-                    "`prompt_embeds` and `negative_prompt_embeds` must have the same shape when passed directly, but"
-                    f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
-                    f" {negative_prompt_embeds.shape}."
-                )
-
-        if prompt_embeds is not None and pooled_prompt_embeds is None:
-            raise ValueError(
-                "If `prompt_embeds` are provided, `pooled_prompt_embeds` also have to be passed. Make sure to generate `pooled_prompt_embeds` from the same text encoder that was used to generate `prompt_embeds`."
-            )
-
-        if negative_prompt_embeds is not None and negative_pooled_prompt_embeds is None:
-            raise ValueError(
-                "If `negative_prompt_embeds` are provided, `negative_pooled_prompt_embeds` also have to be passed. Make sure to generate `negative_pooled_prompt_embeds` from the same text encoder that was used to generate `negative_prompt_embeds`."
-            )
-
-    # Copied from diffusers.pipelines.stable_diffusion_xl.pipeline_stable_diffusion_xl.StableDiffusionXLPipeline.encode_prompt
-    def encode_prompt(
-        self,
-        prompt: str,
-        prompt_2: Optional[str] = None,
-        device: Optional[torch.device] = None,
-        num_images_per_prompt: int = 1,
-        do_classifier_free_guidance: bool = True,
-        negative_prompt: Optional[str] = None,
-        negative_prompt_2: Optional[str] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        lora_scale: Optional[float] = None,
-        clip_skip: Optional[int] = None,
-    ):
-        r"""
-        Encodes the prompt into text encoder hidden states.
-
-        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 both text-encoders
-            device: (`torch.device`):
-                torch device
-            num_images_per_prompt (`int`):
-                number of images that should be generated per prompt
-            do_classifier_free_guidance (`bool`):
-                whether to use classifier free guidance or not
-            negative_prompt (`str` or `List[str]`, *optional*):
-                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`).
-            negative_prompt_2 (`str` or `List[str]`, *optional*):
-                The prompt or prompts not to guide the image generation to be sent to `tokenizer_2` and
-                `text_encoder_2`. If not defined, `negative_prompt` is used in both text-encoders
-            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.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
-                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
-                weighting. If not provided, negative_prompt_embeds will be generated from `negative_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.
-            negative_pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
-                Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
-                weighting. If not provided, pooled negative_prompt_embeds will be generated from `negative_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.
-            clip_skip (`int`, *optional*):
-                Number of layers to be skipped from CLIP while computing the prompt embeddings. A value of 1 means that
-                the output of the pre-final layer will be used for computing the prompt embeddings.
-        """
-        device = device or self._execution_device
-
-        # 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, StableDiffusionXLLoraLoaderMixin):
-            self._lora_scale = lora_scale
-
-            # dynamically adjust the LoRA scale
-            if self.text_encoder is not None:
-                if not USE_PEFT_BACKEND:
-                    adjust_lora_scale_text_encoder(self.text_encoder, lora_scale)
-                else:
-                    scale_lora_layers(self.text_encoder, lora_scale)
-
-            if self.text_encoder_2 is not None:
-                if not USE_PEFT_BACKEND:
-                    adjust_lora_scale_text_encoder(self.text_encoder_2, lora_scale)
-                else:
-                    scale_lora_layers(self.text_encoder_2, lora_scale)
-
-        prompt = [prompt] if isinstance(prompt, str) else prompt
-
-        if prompt is not None:
-            batch_size = len(prompt)
-        else:
-            batch_size = prompt_embeds.shape[0]
-
-        # Define tokenizers and text encoders
-        tokenizers = [self.tokenizer, self.tokenizer_2] if self.tokenizer is not None else [self.tokenizer_2]
-        text_encoders = (
-            [self.text_encoder, self.text_encoder_2] if self.text_encoder is not None else [self.text_encoder_2]
-        )
-
-        if prompt_embeds is None:
-            prompt_2 = prompt_2 or prompt
-            prompt_2 = [prompt_2] if isinstance(prompt_2, str) else prompt_2
-
-            # textual inversion: procecss multi-vector tokens if necessary
-            prompt_embeds_list = []
-            prompts = [prompt, prompt_2]
-            for prompt, tokenizer, text_encoder in zip(prompts, tokenizers, text_encoders):
-                if isinstance(self, TextualInversionLoaderMixin):
-                    prompt = self.maybe_convert_prompt(prompt, tokenizer)
-
-                text_inputs = tokenizer(
-                    prompt,
-                    padding="max_length",
-                    max_length=tokenizer.model_max_length,
-                    truncation=True,
-                    return_tensors="pt",
-                )
-
-                text_input_ids = text_inputs.input_ids
-                untruncated_ids = tokenizer(prompt, padding="longest", return_tensors="pt").input_ids
-
-                if untruncated_ids.shape[-1] >= text_input_ids.shape[-1] and not torch.equal(
-                    text_input_ids, untruncated_ids
-                ):
-                    removed_text = tokenizer.batch_decode(untruncated_ids[:, tokenizer.model_max_length - 1 : -1])
-                    logger.warning(
-                        "The following part of your input was truncated because CLIP can only handle sequences up to"
-                        f" {tokenizer.model_max_length} tokens: {removed_text}"
-                    )
-
-                prompt_embeds = text_encoder(text_input_ids.to(device), output_hidden_states=True)
-
-                # We are only ALWAYS interested in the pooled output of the final text encoder
-                pooled_prompt_embeds = prompt_embeds[0]
-                if clip_skip is None:
-                    prompt_embeds = prompt_embeds.hidden_states[-2]
-                else:
-                    # "2" because SDXL always indexes from the penultimate layer.
-                    prompt_embeds = prompt_embeds.hidden_states[-(clip_skip + 2)]
-
-                prompt_embeds_list.append(prompt_embeds)
-
-            prompt_embeds = torch.concat(prompt_embeds_list, dim=-1)
-
-        # get unconditional embeddings for classifier free guidance
-        zero_out_negative_prompt = negative_prompt is None and self.config.force_zeros_for_empty_prompt
-        if do_classifier_free_guidance and negative_prompt_embeds is None and zero_out_negative_prompt:
-            negative_prompt_embeds = torch.zeros_like(prompt_embeds)
-            negative_pooled_prompt_embeds = torch.zeros_like(pooled_prompt_embeds)
-        elif do_classifier_free_guidance and negative_prompt_embeds is None:
-            negative_prompt = negative_prompt or ""
-            negative_prompt_2 = negative_prompt_2 or negative_prompt
-
-            # normalize str to list
-            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt
-            negative_prompt_2 = (
-                batch_size * [negative_prompt_2] if isinstance(negative_prompt_2, str) else negative_prompt_2
-            )
-
-            uncond_tokens: List[str]
-            if prompt is not None and type(prompt) is not type(negative_prompt):
-                raise TypeError(
-                    f"`negative_prompt` should be the same type to `prompt`, but got {type(negative_prompt)} !="
-                    f" {type(prompt)}."
-                )
-            elif batch_size != len(negative_prompt):
-                raise ValueError(
-                    f"`negative_prompt`: {negative_prompt} has batch size {len(negative_prompt)}, but `prompt`:"
-                    f" {prompt} has batch size {batch_size}. Please make sure that passed `negative_prompt` matches"
-                    " the batch size of `prompt`."
-                )
-            else:
-                uncond_tokens = [negative_prompt, negative_prompt_2]
-
-            negative_prompt_embeds_list = []
-            for negative_prompt, tokenizer, text_encoder in zip(uncond_tokens, tokenizers, text_encoders):
-                if isinstance(self, TextualInversionLoaderMixin):
-                    negative_prompt = self.maybe_convert_prompt(negative_prompt, tokenizer)
-
-                max_length = prompt_embeds.shape[1]
-                uncond_input = tokenizer(
-                    negative_prompt,
-                    padding="max_length",
-                    max_length=max_length,
-                    truncation=True,
-                    return_tensors="pt",
-                )
-
-                negative_prompt_embeds = text_encoder(
-                    uncond_input.input_ids.to(device),
-                    output_hidden_states=True,
-                )
-                # We are only ALWAYS interested in the pooled output of the final text encoder
-                negative_pooled_prompt_embeds = negative_prompt_embeds[0]
-                negative_prompt_embeds = negative_prompt_embeds.hidden_states[-2]
-
-                negative_prompt_embeds_list.append(negative_prompt_embeds)
-
-            negative_prompt_embeds = torch.concat(negative_prompt_embeds_list, dim=-1)
-
-        if self.text_encoder_2 is not None:
-            prompt_embeds = prompt_embeds.to(dtype=self.text_encoder_2.dtype, device=device)
-        else:
-            prompt_embeds = prompt_embeds.to(dtype=self.unet.dtype, device=device)
-
-        bs_embed, seq_len, _ = prompt_embeds.shape
-        # duplicate text embeddings for each generation per prompt, using mps friendly method
-        prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
-        prompt_embeds = prompt_embeds.view(bs_embed * num_images_per_prompt, seq_len, -1)
-
-        if do_classifier_free_guidance:
-            # duplicate unconditional embeddings for each generation per prompt, using mps friendly method
-            seq_len = negative_prompt_embeds.shape[1]
-
-            if self.text_encoder_2 is not None:
-                negative_prompt_embeds = negative_prompt_embeds.to(dtype=self.text_encoder_2.dtype, device=device)
-            else:
-                negative_prompt_embeds = negative_prompt_embeds.to(dtype=self.unet.dtype, device=device)
-
-            negative_prompt_embeds = negative_prompt_embeds.repeat(1, num_images_per_prompt, 1)
-            negative_prompt_embeds = negative_prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
-
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt).view(
-            bs_embed * num_images_per_prompt, -1
-        )
-        if do_classifier_free_guidance:
-            negative_pooled_prompt_embeds = negative_pooled_prompt_embeds.repeat(1, num_images_per_prompt).view(
-                bs_embed * num_images_per_prompt, -1
-            )
-
-        if self.text_encoder is not None:
-            if isinstance(self, StableDiffusionXLLoraLoaderMixin) 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, StableDiffusionXLLoraLoaderMixin) and USE_PEFT_BACKEND:
-                # Retrieve the original scale by scaling back the LoRA layers
-                unscale_lora_layers(self.text_encoder_2, lora_scale)
-
-        return prompt_embeds, negative_prompt_embeds, pooled_prompt_embeds, negative_pooled_prompt_embeds
-
     # Copied from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero.TextToVideoZeroPipeline.forward_loop
     def forward_loop(self, x_t0, t0, t1, generator):
         """

src/diffusers/pipelines/versatile_diffusion/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import (
+from ...utils import (
     DIFFUSERS_SLOW_IMPORT,
     OptionalDependencyNotAvailable,
     _LazyModule,
@@ -17,7 +17,7 @@ try:
     if not (is_transformers_available() and is_torch_available() and is_transformers_version(">=", "4.25.0")):
         raise OptionalDependencyNotAvailable()
 except OptionalDependencyNotAvailable:
-    from ....utils.dummy_torch_and_transformers_objects import (
+    from ...utils.dummy_torch_and_transformers_objects import (
         VersatileDiffusionDualGuidedPipeline,
         VersatileDiffusionImageVariationPipeline,
         VersatileDiffusionPipeline,
@@ -45,7 +45,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         if not (is_transformers_available() and is_torch_available() and is_transformers_version(">=", "4.25.0")):
             raise OptionalDependencyNotAvailable()
     except OptionalDependencyNotAvailable:
-        from ....utils.dummy_torch_and_transformers_objects import (
+        from ...utils.dummy_torch_and_transformers_objects import (
             VersatileDiffusionDualGuidedPipeline,
             VersatileDiffusionImageVariationPipeline,
             VersatileDiffusionPipeline,

src/diffusers/pipelines/versatile_diffusion/modeling_text_unet.py

@@ -7,10 +7,10 @@ import torch.nn.functional as F
 
 from diffusers.utils import deprecate
 
-from ....configuration_utils import ConfigMixin, register_to_config
-from ....models import ModelMixin
-from ....models.activations import get_activation
-from ....models.attention_processor import (
+from ...configuration_utils import ConfigMixin, register_to_config
+from ...models import ModelMixin
+from ...models.activations import get_activation
+from ...models.attention_processor import (
     ADDED_KV_ATTENTION_PROCESSORS,
     CROSS_ATTENTION_PROCESSORS,
     Attention,
@@ -19,8 +19,8 @@ from ....models.attention_processor import (
     AttnAddedKVProcessor2_0,
     AttnProcessor,
 )
-from ....models.dual_transformer_2d import DualTransformer2DModel
-from ....models.embeddings import (
+from ...models.dual_transformer_2d import DualTransformer2DModel
+from ...models.embeddings import (
     GaussianFourierProjection,
     ImageHintTimeEmbedding,
     ImageProjection,
@@ -31,10 +31,10 @@ from ....models.embeddings import (
     TimestepEmbedding,
     Timesteps,
 )
-from ....models.transformer_2d import Transformer2DModel
-from ....models.unet_2d_condition import UNet2DConditionOutput
-from ....utils import USE_PEFT_BACKEND, is_torch_version, logging, scale_lora_layers, unscale_lora_layers
-from ....utils.torch_utils import apply_freeu
+from ...models.transformer_2d import Transformer2DModel
+from ...models.unet_2d_condition import UNet2DConditionOutput
+from ...utils import USE_PEFT_BACKEND, is_torch_version, logging, scale_lora_layers, unscale_lora_layers
+from ...utils.torch_utils import apply_freeu
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion.py

@@ -5,10 +5,10 @@ import PIL.Image
 import torch
 from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModel
 
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import logging
-from ...pipeline_utils import DiffusionPipeline
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import logging
+from ..pipeline_utils import DiffusionPipeline
 from .pipeline_versatile_diffusion_dual_guided import VersatileDiffusionDualGuidedPipeline
 from .pipeline_versatile_diffusion_image_variation import VersatileDiffusionImageVariationPipeline
 from .pipeline_versatile_diffusion_text_to_image import VersatileDiffusionTextToImagePipeline

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion_dual_guided.py

@@ -26,12 +26,12 @@ from transformers import (
     CLIPVisionModelWithProjection,
 )
 
-from ....image_processor import VaeImageProcessor
-from ....models import AutoencoderKL, DualTransformer2DModel, Transformer2DModel, UNet2DConditionModel
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import deprecate, logging
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...image_processor import VaeImageProcessor
+from ...models import AutoencoderKL, DualTransformer2DModel, Transformer2DModel, UNet2DConditionModel
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import deprecate, logging
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 from .modeling_text_unet import UNetFlatConditionModel
 
 

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion_image_variation.py

@@ -21,12 +21,12 @@ import torch
 import torch.utils.checkpoint
 from transformers import CLIPImageProcessor, CLIPVisionModelWithProjection
 
-from ....image_processor import VaeImageProcessor
-from ....models import AutoencoderKL, UNet2DConditionModel
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import deprecate, logging
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...image_processor import VaeImageProcessor
+from ...models import AutoencoderKL, UNet2DConditionModel
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import deprecate, logging
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

src/diffusers/pipelines/versatile_diffusion/pipeline_versatile_diffusion_text_to_image.py

@@ -19,12 +19,12 @@ import torch
 import torch.utils.checkpoint
 from transformers import CLIPImageProcessor, CLIPTextModelWithProjection, CLIPTokenizer
 
-from ....image_processor import VaeImageProcessor
-from ....models import AutoencoderKL, Transformer2DModel, UNet2DConditionModel
-from ....schedulers import KarrasDiffusionSchedulers
-from ....utils import deprecate, logging
-from ....utils.torch_utils import randn_tensor
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...image_processor import VaeImageProcessor
+from ...models import AutoencoderKL, Transformer2DModel, UNet2DConditionModel
+from ...schedulers import KarrasDiffusionSchedulers
+from ...utils import deprecate, logging
+from ...utils.torch_utils import randn_tensor
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 from .modeling_text_unet import UNetFlatConditionModel
 
 

src/diffusers/pipelines/vq_diffusion/__init__.py

@@ -1,6 +1,6 @@
 from typing import TYPE_CHECKING
 
-from ....utils import (
+from ...utils import (
     DIFFUSERS_SLOW_IMPORT,
     OptionalDependencyNotAvailable,
     _LazyModule,
@@ -16,7 +16,7 @@ try:
     if not (is_transformers_available() and is_torch_available()):
         raise OptionalDependencyNotAvailable()
 except OptionalDependencyNotAvailable:
-    from ....utils.dummy_torch_and_transformers_objects import (
+    from ...utils.dummy_torch_and_transformers_objects import (
         LearnedClassifierFreeSamplingEmbeddings,
         VQDiffusionPipeline,
     )
@@ -36,7 +36,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
         if not (is_transformers_available() and is_torch_available()):
             raise OptionalDependencyNotAvailable()
     except OptionalDependencyNotAvailable:
-        from ....utils.dummy_torch_and_transformers_objects import (
+        from ...utils.dummy_torch_and_transformers_objects import (
             LearnedClassifierFreeSamplingEmbeddings,
             VQDiffusionPipeline,
         )

src/diffusers/pipelines/vq_diffusion/pipeline_vq_diffusion.py

@@ -17,11 +17,11 @@ from typing import Callable, List, Optional, Tuple, Union
 import torch
 from transformers import CLIPTextModel, CLIPTokenizer
 
-from ....configuration_utils import ConfigMixin, register_to_config
-from ....models import ModelMixin, Transformer2DModel, VQModel
-from ....schedulers import VQDiffusionScheduler
-from ....utils import logging
-from ...pipeline_utils import DiffusionPipeline, ImagePipelineOutput
+from ...configuration_utils import ConfigMixin, register_to_config
+from ...models import ModelMixin, Transformer2DModel, VQModel
+from ...schedulers import VQDiffusionScheduler
+from ...utils import logging
+from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
 
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name

tests/pipelines/altdiffusion/test_alt_diffusion.py

@@ -0,0 +1,260 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import unittest
+
+import numpy as np
+import torch
+from transformers import CLIPTextConfig, CLIPTextModel, XLMRobertaTokenizer
+
+from diffusers import AltDiffusionPipeline, AutoencoderKL, DDIMScheduler, PNDMScheduler, UNet2DConditionModel
+from diffusers.pipelines.alt_diffusion.modeling_roberta_series import (
+    RobertaSeriesConfig,
+    RobertaSeriesModelWithTransformation,
+)
+from diffusers.utils.testing_utils import enable_full_determinism, nightly, require_torch_gpu, torch_device
+
+from ..pipeline_params import (
+    TEXT_TO_IMAGE_BATCH_PARAMS,
+    TEXT_TO_IMAGE_CALLBACK_CFG_PARAMS,
+    TEXT_TO_IMAGE_IMAGE_PARAMS,
+    TEXT_TO_IMAGE_PARAMS,
+)
+from ..test_pipelines_common import PipelineKarrasSchedulerTesterMixin, PipelineLatentTesterMixin, PipelineTesterMixin
+
+
+enable_full_determinism()
+
+
+class AltDiffusionPipelineFastTests(
+    PipelineLatentTesterMixin, PipelineKarrasSchedulerTesterMixin, PipelineTesterMixin, unittest.TestCase
+):
+    pipeline_class = AltDiffusionPipeline
+    params = TEXT_TO_IMAGE_PARAMS
+    batch_params = TEXT_TO_IMAGE_BATCH_PARAMS
+    image_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    callback_cfg_params = TEXT_TO_IMAGE_CALLBACK_CFG_PARAMS
+
+    def get_dummy_components(self):
+        torch.manual_seed(0)
+        unet = UNet2DConditionModel(
+            block_out_channels=(32, 64),
+            layers_per_block=2,
+            sample_size=32,
+            in_channels=4,
+            out_channels=4,
+            down_block_types=("DownBlock2D", "CrossAttnDownBlock2D"),
+            up_block_types=("CrossAttnUpBlock2D", "UpBlock2D"),
+            cross_attention_dim=32,
+        )
+        scheduler = DDIMScheduler(
+            beta_start=0.00085,
+            beta_end=0.012,
+            beta_schedule="scaled_linear",
+            clip_sample=False,
+            set_alpha_to_one=False,
+        )
+        torch.manual_seed(0)
+        vae = AutoencoderKL(
+            block_out_channels=[32, 64],
+            in_channels=3,
+            out_channels=3,
+            down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"],
+            up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"],
+            latent_channels=4,
+        )
+
+        # TODO: address the non-deterministic text encoder (fails for save-load tests)
+        # torch.manual_seed(0)
+        # text_encoder_config = RobertaSeriesConfig(
+        #     hidden_size=32,
+        #     project_dim=32,
+        #     intermediate_size=37,
+        #     layer_norm_eps=1e-05,
+        #     num_attention_heads=4,
+        #     num_hidden_layers=5,
+        #     vocab_size=5002,
+        # )
+        # text_encoder = RobertaSeriesModelWithTransformation(text_encoder_config)
+
+        torch.manual_seed(0)
+        text_encoder_config = CLIPTextConfig(
+            bos_token_id=0,
+            eos_token_id=2,
+            hidden_size=32,
+            projection_dim=32,
+            intermediate_size=37,
+            layer_norm_eps=1e-05,
+            num_attention_heads=4,
+            num_hidden_layers=5,
+            pad_token_id=1,
+            vocab_size=5002,
+        )
+        text_encoder = CLIPTextModel(text_encoder_config)
+
+        tokenizer = XLMRobertaTokenizer.from_pretrained("hf-internal-testing/tiny-xlm-roberta")
+        tokenizer.model_max_length = 77
+
+        components = {
+            "unet": unet,
+            "scheduler": scheduler,
+            "vae": vae,
+            "text_encoder": text_encoder,
+            "tokenizer": tokenizer,
+            "safety_checker": None,
+            "feature_extractor": None,
+            "image_encoder": None,
+        }
+        return components
+
+    def get_dummy_inputs(self, device, seed=0):
+        if str(device).startswith("mps"):
+            generator = torch.manual_seed(seed)
+        else:
+            generator = torch.Generator(device=device).manual_seed(seed)
+        inputs = {
+            "prompt": "A painting of a squirrel eating a burger",
+            "generator": generator,
+            "num_inference_steps": 2,
+            "guidance_scale": 6.0,
+            "output_type": "numpy",
+        }
+        return inputs
+
+    def test_attention_slicing_forward_pass(self):
+        super().test_attention_slicing_forward_pass(expected_max_diff=3e-3)
+
+    def test_inference_batch_single_identical(self):
+        super().test_inference_batch_single_identical(expected_max_diff=3e-3)
+
+    def test_alt_diffusion_ddim(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+
+        components = self.get_dummy_components()
+        torch.manual_seed(0)
+        text_encoder_config = RobertaSeriesConfig(
+            hidden_size=32,
+            project_dim=32,
+            intermediate_size=37,
+            layer_norm_eps=1e-05,
+            num_attention_heads=4,
+            num_hidden_layers=5,
+            vocab_size=5002,
+        )
+        # TODO: remove after fixing the non-deterministic text encoder
+        text_encoder = RobertaSeriesModelWithTransformation(text_encoder_config)
+        components["text_encoder"] = text_encoder
+
+        alt_pipe = AltDiffusionPipeline(**components)
+        alt_pipe = alt_pipe.to(device)
+        alt_pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        inputs["prompt"] = "A photo of an astronaut"
+        output = alt_pipe(**inputs)
+        image = output.images
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 64, 64, 3)
+        expected_slice = np.array(
+            [0.5748162, 0.60447145, 0.48821217, 0.50100636, 0.5431185, 0.45763683, 0.49657696, 0.48132733, 0.47573093]
+        )
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+
+    def test_alt_diffusion_pndm(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+
+        components = self.get_dummy_components()
+        components["scheduler"] = PNDMScheduler(skip_prk_steps=True)
+        torch.manual_seed(0)
+        text_encoder_config = RobertaSeriesConfig(
+            hidden_size=32,
+            project_dim=32,
+            intermediate_size=37,
+            layer_norm_eps=1e-05,
+            num_attention_heads=4,
+            num_hidden_layers=5,
+            vocab_size=5002,
+        )
+        # TODO: remove after fixing the non-deterministic text encoder
+        text_encoder = RobertaSeriesModelWithTransformation(text_encoder_config)
+        components["text_encoder"] = text_encoder
+        alt_pipe = AltDiffusionPipeline(**components)
+        alt_pipe = alt_pipe.to(device)
+        alt_pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        output = alt_pipe(**inputs)
+        image = output.images
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 64, 64, 3)
+        expected_slice = np.array(
+            [0.51605093, 0.5707241, 0.47365507, 0.50578886, 0.5633877, 0.4642503, 0.5182081, 0.48763484, 0.49084237]
+        )
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+
+
+@nightly
+@require_torch_gpu
+class AltDiffusionPipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_alt_diffusion(self):
+        # make sure here that pndm scheduler skips prk
+        alt_pipe = AltDiffusionPipeline.from_pretrained("BAAI/AltDiffusion", safety_checker=None)
+        alt_pipe = alt_pipe.to(torch_device)
+        alt_pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger"
+        generator = torch.manual_seed(0)
+        output = alt_pipe([prompt], generator=generator, guidance_scale=6.0, num_inference_steps=20, output_type="np")
+
+        image = output.images
+
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.1010, 0.0800, 0.0794, 0.0885, 0.0843, 0.0762, 0.0769, 0.0729, 0.0586])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+
+    def test_alt_diffusion_fast_ddim(self):
+        scheduler = DDIMScheduler.from_pretrained("BAAI/AltDiffusion", subfolder="scheduler")
+
+        alt_pipe = AltDiffusionPipeline.from_pretrained("BAAI/AltDiffusion", scheduler=scheduler, safety_checker=None)
+        alt_pipe = alt_pipe.to(torch_device)
+        alt_pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger"
+        generator = torch.manual_seed(0)
+
+        output = alt_pipe([prompt], generator=generator, num_inference_steps=2, output_type="numpy")
+        image = output.images
+
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.4019, 0.4052, 0.3810, 0.4119, 0.3916, 0.3982, 0.4651, 0.4195, 0.5323])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2

tests/pipelines/altdiffusion/test_alt_diffusion_img2img.py

@@ -0,0 +1,309 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import random
+import unittest
+
+import numpy as np
+import torch
+from transformers import XLMRobertaTokenizer
+
+from diffusers import (
+    AltDiffusionImg2ImgPipeline,
+    AutoencoderKL,
+    PNDMScheduler,
+    UNet2DConditionModel,
+)
+from diffusers.image_processor import VaeImageProcessor
+from diffusers.pipelines.alt_diffusion.modeling_roberta_series import (
+    RobertaSeriesConfig,
+    RobertaSeriesModelWithTransformation,
+)
+from diffusers.utils import load_image
+from diffusers.utils.testing_utils import (
+    enable_full_determinism,
+    floats_tensor,
+    load_numpy,
+    nightly,
+    require_torch_gpu,
+    torch_device,
+)
+
+
+enable_full_determinism()
+
+
+class AltDiffusionImg2ImgPipelineFastTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    @property
+    def dummy_image(self):
+        batch_size = 1
+        num_channels = 3
+        sizes = (32, 32)
+
+        image = floats_tensor((batch_size, num_channels) + sizes, rng=random.Random(0)).to(torch_device)
+        return image
+
+    @property
+    def dummy_cond_unet(self):
+        torch.manual_seed(0)
+        model = UNet2DConditionModel(
+            block_out_channels=(32, 64),
+            layers_per_block=2,
+            sample_size=32,
+            in_channels=4,
+            out_channels=4,
+            down_block_types=("DownBlock2D", "CrossAttnDownBlock2D"),
+            up_block_types=("CrossAttnUpBlock2D", "UpBlock2D"),
+            cross_attention_dim=32,
+        )
+        return model
+
+    @property
+    def dummy_vae(self):
+        torch.manual_seed(0)
+        model = AutoencoderKL(
+            block_out_channels=[32, 64],
+            in_channels=3,
+            out_channels=3,
+            down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"],
+            up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"],
+            latent_channels=4,
+        )
+        return model
+
+    @property
+    def dummy_text_encoder(self):
+        torch.manual_seed(0)
+        config = RobertaSeriesConfig(
+            hidden_size=32,
+            project_dim=32,
+            intermediate_size=37,
+            layer_norm_eps=1e-05,
+            num_attention_heads=4,
+            num_hidden_layers=5,
+            pad_token_id=1,
+            vocab_size=5006,
+        )
+        return RobertaSeriesModelWithTransformation(config)
+
+    @property
+    def dummy_extractor(self):
+        def extract(*args, **kwargs):
+            class Out:
+                def __init__(self):
+                    self.pixel_values = torch.ones([0])
+
+                def to(self, device):
+                    self.pixel_values.to(device)
+                    return self
+
+            return Out()
+
+        return extract
+
+    def test_stable_diffusion_img2img_default_case(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+        unet = self.dummy_cond_unet
+        scheduler = PNDMScheduler(skip_prk_steps=True)
+        vae = self.dummy_vae
+        bert = self.dummy_text_encoder
+        tokenizer = XLMRobertaTokenizer.from_pretrained("hf-internal-testing/tiny-xlm-roberta")
+        tokenizer.model_max_length = 77
+
+        init_image = self.dummy_image.to(device)
+        init_image = init_image / 2 + 0.5
+
+        # make sure here that pndm scheduler skips prk
+        alt_pipe = AltDiffusionImg2ImgPipeline(
+            unet=unet,
+            scheduler=scheduler,
+            vae=vae,
+            text_encoder=bert,
+            tokenizer=tokenizer,
+            safety_checker=None,
+            feature_extractor=self.dummy_extractor,
+            image_encoder=None,
+        )
+        alt_pipe.image_processor = VaeImageProcessor(vae_scale_factor=alt_pipe.vae_scale_factor, do_normalize=True)
+        alt_pipe = alt_pipe.to(device)
+        alt_pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger"
+        generator = torch.Generator(device=device).manual_seed(0)
+        output = alt_pipe(
+            [prompt],
+            generator=generator,
+            guidance_scale=6.0,
+            num_inference_steps=2,
+            output_type="np",
+            image=init_image,
+        )
+
+        image = output.images
+
+        generator = torch.Generator(device=device).manual_seed(0)
+        image_from_tuple = alt_pipe(
+            [prompt],
+            generator=generator,
+            guidance_scale=6.0,
+            num_inference_steps=2,
+            output_type="np",
+            image=init_image,
+            return_dict=False,
+        )[0]
+
+        image_slice = image[0, -3:, -3:, -1]
+        image_from_tuple_slice = image_from_tuple[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 32, 32, 3)
+        expected_slice = np.array([0.4427, 0.3731, 0.4249, 0.4941, 0.4546, 0.4148, 0.4193, 0.4666, 0.4499])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 5e-3
+        assert np.abs(image_from_tuple_slice.flatten() - expected_slice).max() < 5e-3
+
+    @unittest.skipIf(torch_device != "cuda", "This test requires a GPU")
+    def test_stable_diffusion_img2img_fp16(self):
+        """Test that stable diffusion img2img works with fp16"""
+        unet = self.dummy_cond_unet
+        scheduler = PNDMScheduler(skip_prk_steps=True)
+        vae = self.dummy_vae
+        bert = self.dummy_text_encoder
+        tokenizer = XLMRobertaTokenizer.from_pretrained("hf-internal-testing/tiny-xlm-roberta")
+        tokenizer.model_max_length = 77
+
+        init_image = self.dummy_image.to(torch_device)
+
+        # put models in fp16
+        unet = unet.half()
+        vae = vae.half()
+        bert = bert.half()
+
+        # make sure here that pndm scheduler skips prk
+        alt_pipe = AltDiffusionImg2ImgPipeline(
+            unet=unet,
+            scheduler=scheduler,
+            vae=vae,
+            text_encoder=bert,
+            tokenizer=tokenizer,
+            safety_checker=None,
+            feature_extractor=self.dummy_extractor,
+            image_encoder=None,
+        )
+        alt_pipe.image_processor = VaeImageProcessor(vae_scale_factor=alt_pipe.vae_scale_factor, do_normalize=False)
+        alt_pipe = alt_pipe.to(torch_device)
+        alt_pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger"
+        generator = torch.manual_seed(0)
+        image = alt_pipe(
+            [prompt],
+            generator=generator,
+            num_inference_steps=2,
+            output_type="np",
+            image=init_image,
+        ).images
+
+        assert image.shape == (1, 32, 32, 3)
+
+    @unittest.skipIf(torch_device != "cuda", "This test requires a GPU")
+    def test_stable_diffusion_img2img_pipeline_multiple_of_8(self):
+        init_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"
+            "/img2img/sketch-mountains-input.jpg"
+        )
+        # resize to resolution that is divisible by 8 but not 16 or 32
+        init_image = init_image.resize((760, 504))
+
+        model_id = "BAAI/AltDiffusion"
+        pipe = AltDiffusionImg2ImgPipeline.from_pretrained(
+            model_id,
+            safety_checker=None,
+        )
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+        pipe.enable_attention_slicing()
+
+        prompt = "A fantasy landscape, trending on artstation"
+
+        generator = torch.manual_seed(0)
+        output = pipe(
+            prompt=prompt,
+            image=init_image,
+            strength=0.75,
+            guidance_scale=7.5,
+            generator=generator,
+            output_type="np",
+        )
+        image = output.images[0]
+
+        image_slice = image[255:258, 383:386, -1]
+
+        assert image.shape == (504, 760, 3)
+        expected_slice = np.array([0.9358, 0.9397, 0.9599, 0.9901, 1.0000, 1.0000, 0.9882, 1.0000, 1.0000])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+
+
+@nightly
+@require_torch_gpu
+class AltDiffusionImg2ImgPipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_stable_diffusion_img2img_pipeline_default(self):
+        init_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"
+            "/img2img/sketch-mountains-input.jpg"
+        )
+        init_image = init_image.resize((768, 512))
+        expected_image = load_numpy(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/img2img/fantasy_landscape_alt.npy"
+        )
+
+        model_id = "BAAI/AltDiffusion"
+        pipe = AltDiffusionImg2ImgPipeline.from_pretrained(
+            model_id,
+            safety_checker=None,
+        )
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+        pipe.enable_attention_slicing()
+
+        prompt = "A fantasy landscape, trending on artstation"
+
+        generator = torch.manual_seed(0)
+        output = pipe(
+            prompt=prompt,
+            image=init_image,
+            strength=0.75,
+            guidance_scale=7.5,
+            generator=generator,
+            output_type="np",
+        )
+        image = output.images[0]
+
+        assert image.shape == (512, 768, 3)
+        # img2img is flaky across GPUs even in fp32, so using MAE here
+        assert np.abs(expected_image - image).max() < 1e-2

tests/pipelines/audio_diffusion/test_audio_diffusion.py

@@ -0,0 +1,203 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import (
+    AudioDiffusionPipeline,
+    AutoencoderKL,
+    DDIMScheduler,
+    DDPMScheduler,
+    DiffusionPipeline,
+    Mel,
+    UNet2DConditionModel,
+    UNet2DModel,
+)
+from diffusers.utils.testing_utils import enable_full_determinism, nightly, require_torch_gpu, torch_device
+
+
+enable_full_determinism()
+
+
+class PipelineFastTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    @property
+    def dummy_unet(self):
+        torch.manual_seed(0)
+        model = UNet2DModel(
+            sample_size=(32, 64),
+            in_channels=1,
+            out_channels=1,
+            layers_per_block=2,
+            block_out_channels=(128, 128),
+            down_block_types=("AttnDownBlock2D", "DownBlock2D"),
+            up_block_types=("UpBlock2D", "AttnUpBlock2D"),
+        )
+        return model
+
+    @property
+    def dummy_unet_condition(self):
+        torch.manual_seed(0)
+        model = UNet2DConditionModel(
+            sample_size=(64, 32),
+            in_channels=1,
+            out_channels=1,
+            layers_per_block=2,
+            block_out_channels=(128, 128),
+            down_block_types=("CrossAttnDownBlock2D", "DownBlock2D"),
+            up_block_types=("UpBlock2D", "CrossAttnUpBlock2D"),
+            cross_attention_dim=10,
+        )
+        return model
+
+    @property
+    def dummy_vqvae_and_unet(self):
+        torch.manual_seed(0)
+        vqvae = AutoencoderKL(
+            sample_size=(128, 64),
+            in_channels=1,
+            out_channels=1,
+            latent_channels=1,
+            layers_per_block=2,
+            block_out_channels=(128, 128),
+            down_block_types=("DownEncoderBlock2D", "DownEncoderBlock2D"),
+            up_block_types=("UpDecoderBlock2D", "UpDecoderBlock2D"),
+        )
+        unet = UNet2DModel(
+            sample_size=(64, 32),
+            in_channels=1,
+            out_channels=1,
+            layers_per_block=2,
+            block_out_channels=(128, 128),
+            down_block_types=("AttnDownBlock2D", "DownBlock2D"),
+            up_block_types=("UpBlock2D", "AttnUpBlock2D"),
+        )
+        return vqvae, unet
+
+    @nightly
+    def test_audio_diffusion(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+        mel = Mel(
+            x_res=self.dummy_unet.config.sample_size[1],
+            y_res=self.dummy_unet.config.sample_size[0],
+        )
+
+        scheduler = DDPMScheduler()
+        pipe = AudioDiffusionPipeline(vqvae=None, unet=self.dummy_unet, mel=mel, scheduler=scheduler)
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        generator = torch.Generator(device=device).manual_seed(42)
+        output = pipe(generator=generator, steps=4)
+        audio = output.audios[0]
+        image = output.images[0]
+
+        generator = torch.Generator(device=device).manual_seed(42)
+        output = pipe(generator=generator, steps=4, return_dict=False)
+        image_from_tuple = output[0][0]
+
+        assert audio.shape == (1, (self.dummy_unet.config.sample_size[1] - 1) * mel.hop_length)
+        assert (
+            image.height == self.dummy_unet.config.sample_size[0]
+            and image.width == self.dummy_unet.config.sample_size[1]
+        )
+        image_slice = np.frombuffer(image.tobytes(), dtype="uint8")[:10]
+        image_from_tuple_slice = np.frombuffer(image_from_tuple.tobytes(), dtype="uint8")[:10]
+        expected_slice = np.array([69, 255, 255, 255, 0, 0, 77, 181, 12, 127])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() == 0
+        assert np.abs(image_from_tuple_slice.flatten() - expected_slice).max() == 0
+
+        mel = Mel(
+            x_res=self.dummy_vqvae_and_unet[0].config.sample_size[1],
+            y_res=self.dummy_vqvae_and_unet[0].config.sample_size[0],
+        )
+
+        scheduler = DDIMScheduler()
+        dummy_vqvae_and_unet = self.dummy_vqvae_and_unet
+        pipe = AudioDiffusionPipeline(
+            vqvae=self.dummy_vqvae_and_unet[0], unet=dummy_vqvae_and_unet[1], mel=mel, scheduler=scheduler
+        )
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        np.random.seed(0)
+        raw_audio = np.random.uniform(-1, 1, ((dummy_vqvae_and_unet[0].config.sample_size[1] - 1) * mel.hop_length,))
+        generator = torch.Generator(device=device).manual_seed(42)
+        output = pipe(raw_audio=raw_audio, generator=generator, start_step=5, steps=10)
+        image = output.images[0]
+
+        assert (
+            image.height == self.dummy_vqvae_and_unet[0].config.sample_size[0]
+            and image.width == self.dummy_vqvae_and_unet[0].config.sample_size[1]
+        )
+        image_slice = np.frombuffer(image.tobytes(), dtype="uint8")[:10]
+        expected_slice = np.array([120, 117, 110, 109, 138, 167, 138, 148, 132, 121])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() == 0
+
+        dummy_unet_condition = self.dummy_unet_condition
+        pipe = AudioDiffusionPipeline(
+            vqvae=self.dummy_vqvae_and_unet[0], unet=dummy_unet_condition, mel=mel, scheduler=scheduler
+        )
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        np.random.seed(0)
+        encoding = torch.rand((1, 1, 10))
+        output = pipe(generator=generator, encoding=encoding)
+        image = output.images[0]
+        image_slice = np.frombuffer(image.tobytes(), dtype="uint8")[:10]
+        expected_slice = np.array([107, 103, 120, 127, 142, 122, 113, 122, 97, 111])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() == 0
+
+
+@nightly
+@require_torch_gpu
+class PipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_audio_diffusion(self):
+        device = torch_device
+
+        pipe = DiffusionPipeline.from_pretrained("teticio/audio-diffusion-ddim-256")
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        generator = torch.Generator(device=device).manual_seed(42)
+        output = pipe(generator=generator)
+        audio = output.audios[0]
+        image = output.images[0]
+
+        assert audio.shape == (1, (pipe.unet.config.sample_size[1] - 1) * pipe.mel.hop_length)
+        assert image.height == pipe.unet.config.sample_size[0] and image.width == pipe.unet.config.sample_size[1]
+        image_slice = np.frombuffer(image.tobytes(), dtype="uint8")[:10]
+        expected_slice = np.array([151, 167, 154, 144, 122, 134, 121, 105, 70, 26])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() == 0

tests/pipelines/controlnetxs/test_controlnetxs.py

@@ -34,7 +34,6 @@ from diffusers.utils.testing_utils import (
     enable_full_determinism,
     load_image,
     load_numpy,
-    numpy_cosine_similarity_distance,
     require_python39_or_higher,
     require_torch_2,
     require_torch_gpu,
@@ -274,9 +273,7 @@ class ControlNetXSPipelineSlowTests(unittest.TestCase):
 
         original_image = image[-3:, -3:, -1].flatten()
         expected_image = np.array([0.1274, 0.1401, 0.147, 0.1185, 0.1555, 0.1492, 0.1565, 0.1474, 0.1701])
-
-        max_diff = numpy_cosine_similarity_distance(original_image, expected_image)
-        assert max_diff < 1e-4
+        assert np.allclose(original_image, expected_image, atol=1e-04)
 
     def test_depth(self):
         controlnet = ControlNetXSModel.from_pretrained("UmerHA/ConrolNetXS-SD2.1-depth")
@@ -301,9 +298,7 @@ class ControlNetXSPipelineSlowTests(unittest.TestCase):
 
         original_image = image[-3:, -3:, -1].flatten()
         expected_image = np.array([0.1098, 0.1025, 0.1211, 0.1129, 0.1165, 0.1262, 0.1185, 0.1261, 0.1703])
-
-        max_diff = numpy_cosine_similarity_distance(original_image, expected_image)
-        assert max_diff < 1e-4
+        assert np.allclose(original_image, expected_image, atol=1e-04)
 
     @require_python39_or_higher
     @require_torch_2

tests/pipelines/ip_adapters/test_ip_adapter_stable_diffusion.py

@@ -182,33 +182,6 @@ class IPAdapterSDIntegrationTests(IPAdapterNightlyTestsMixin):
 
         assert np.allclose(image_slice, expected_slice, atol=1e-4, rtol=1e-4)
 
-    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(
-            "runwayml/stable-diffusion-v1-5", 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)
-
-        inputs = self.get_dummy_inputs()
-        output_without_offload = pipeline(**inputs).images
-
-        pipeline.enable_model_cpu_offload()
-        inputs = self.get_dummy_inputs()
-        output_with_offload = pipeline(**inputs).images
-        max_diff = np.abs(output_with_offload - output_without_offload).max()
-        self.assertLess(max_diff, 1e-3, "CPU offloading should not affect the inference results")
-
-        offloaded_modules = [
-            v
-            for k, v in pipeline.components.items()
-            if isinstance(v, torch.nn.Module) and k not in pipeline._exclude_from_cpu_offload
-        ]
-        (
-            self.assertTrue(all(v.device.type == "cpu" for v in offloaded_modules)),
-            f"Not offloaded: {[v for v in offloaded_modules if v.device.type != 'cpu']}",
-        )
-
     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(

tests/pipelines/latent_diffusion/test_latent_diffusion_uncond.py

@@ -0,0 +1,116 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import unittest
+
+import numpy as np
+import torch
+from transformers import CLIPTextConfig, CLIPTextModel
+
+from diffusers import DDIMScheduler, LDMPipeline, UNet2DModel, VQModel
+from diffusers.utils.testing_utils import enable_full_determinism, nightly, require_torch, torch_device
+
+
+enable_full_determinism()
+
+
+class LDMPipelineFastTests(unittest.TestCase):
+    @property
+    def dummy_uncond_unet(self):
+        torch.manual_seed(0)
+        model = UNet2DModel(
+            block_out_channels=(32, 64),
+            layers_per_block=2,
+            sample_size=32,
+            in_channels=3,
+            out_channels=3,
+            down_block_types=("DownBlock2D", "AttnDownBlock2D"),
+            up_block_types=("AttnUpBlock2D", "UpBlock2D"),
+        )
+        return model
+
+    @property
+    def dummy_vq_model(self):
+        torch.manual_seed(0)
+        model = VQModel(
+            block_out_channels=[32, 64],
+            in_channels=3,
+            out_channels=3,
+            down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"],
+            up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"],
+            latent_channels=3,
+        )
+        return model
+
+    @property
+    def dummy_text_encoder(self):
+        torch.manual_seed(0)
+        config = CLIPTextConfig(
+            bos_token_id=0,
+            eos_token_id=2,
+            hidden_size=32,
+            intermediate_size=37,
+            layer_norm_eps=1e-05,
+            num_attention_heads=4,
+            num_hidden_layers=5,
+            pad_token_id=1,
+            vocab_size=1000,
+        )
+        return CLIPTextModel(config)
+
+    def test_inference_uncond(self):
+        unet = self.dummy_uncond_unet
+        scheduler = DDIMScheduler()
+        vae = self.dummy_vq_model
+
+        ldm = LDMPipeline(unet=unet, vqvae=vae, scheduler=scheduler)
+        ldm.to(torch_device)
+        ldm.set_progress_bar_config(disable=None)
+
+        generator = torch.manual_seed(0)
+        image = ldm(generator=generator, num_inference_steps=2, output_type="numpy").images
+
+        generator = torch.manual_seed(0)
+        image_from_tuple = ldm(generator=generator, num_inference_steps=2, output_type="numpy", return_dict=False)[0]
+
+        image_slice = image[0, -3:, -3:, -1]
+        image_from_tuple_slice = image_from_tuple[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 64, 64, 3)
+        expected_slice = np.array([0.8512, 0.818, 0.6411, 0.6808, 0.4465, 0.5618, 0.46, 0.6231, 0.5172])
+        tolerance = 1e-2 if torch_device != "mps" else 3e-2
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < tolerance
+        assert np.abs(image_from_tuple_slice.flatten() - expected_slice).max() < tolerance
+
+
+@nightly
+@require_torch
+class LDMPipelineIntegrationTests(unittest.TestCase):
+    def test_inference_uncond(self):
+        ldm = LDMPipeline.from_pretrained("CompVis/ldm-celebahq-256")
+        ldm.to(torch_device)
+        ldm.set_progress_bar_config(disable=None)
+
+        generator = torch.manual_seed(0)
+        image = ldm(generator=generator, num_inference_steps=5, output_type="numpy").images
+
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 256, 256, 3)
+        expected_slice = np.array([0.4399, 0.44975, 0.46825, 0.474, 0.4359, 0.4581, 0.45095, 0.4341, 0.4447])
+        tolerance = 1e-2 if torch_device != "mps" else 3e-2
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < tolerance

tests/pipelines/repaint/test_repaint.py

@@ -0,0 +1,169 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import RePaintPipeline, RePaintScheduler, UNet2DModel
+from diffusers.utils.testing_utils import (
+    enable_full_determinism,
+    load_image,
+    load_numpy,
+    nightly,
+    require_torch_gpu,
+    skip_mps,
+    torch_device,
+)
+
+from ..pipeline_params import IMAGE_INPAINTING_BATCH_PARAMS, IMAGE_INPAINTING_PARAMS
+from ..test_pipelines_common import PipelineTesterMixin
+
+
+enable_full_determinism()
+
+
+class RepaintPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
+    pipeline_class = RePaintPipeline
+    params = IMAGE_INPAINTING_PARAMS - {"width", "height", "guidance_scale"}
+    required_optional_params = PipelineTesterMixin.required_optional_params - {
+        "latents",
+        "num_images_per_prompt",
+        "callback",
+        "callback_steps",
+    }
+    batch_params = IMAGE_INPAINTING_BATCH_PARAMS
+
+    def get_dummy_components(self):
+        torch.manual_seed(0)
+        torch.manual_seed(0)
+        unet = UNet2DModel(
+            block_out_channels=(32, 64),
+            layers_per_block=2,
+            sample_size=32,
+            in_channels=3,
+            out_channels=3,
+            down_block_types=("DownBlock2D", "AttnDownBlock2D"),
+            up_block_types=("AttnUpBlock2D", "UpBlock2D"),
+        )
+        scheduler = RePaintScheduler()
+        components = {"unet": unet, "scheduler": scheduler}
+        return components
+
+    def get_dummy_inputs(self, device, seed=0):
+        if str(device).startswith("mps"):
+            generator = torch.manual_seed(seed)
+        else:
+            generator = torch.Generator(device=device).manual_seed(seed)
+        image = np.random.RandomState(seed).standard_normal((1, 3, 32, 32))
+        image = torch.from_numpy(image).to(device=device, dtype=torch.float32)
+        mask = (image > 0).to(device=device, dtype=torch.float32)
+        inputs = {
+            "image": image,
+            "mask_image": mask,
+            "generator": generator,
+            "num_inference_steps": 5,
+            "eta": 0.0,
+            "jump_length": 2,
+            "jump_n_sample": 2,
+            "output_type": "numpy",
+        }
+        return inputs
+
+    def test_repaint(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+        components = self.get_dummy_components()
+        sd_pipe = RePaintPipeline(**components)
+        sd_pipe = sd_pipe.to(device)
+        sd_pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        image = sd_pipe(**inputs).images
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 32, 32, 3)
+        expected_slice = np.array([1.0000, 0.5426, 0.5497, 0.2200, 1.0000, 1.0000, 0.5623, 1.0000, 0.6274])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-3
+
+    @skip_mps
+    def test_save_load_local(self):
+        return super().test_save_load_local()
+
+    # RePaint can hardly be made deterministic since the scheduler is currently always
+    # nondeterministic
+    @unittest.skip("non-deterministic pipeline")
+    def test_inference_batch_single_identical(self):
+        return super().test_inference_batch_single_identical()
+
+    @skip_mps
+    def test_dict_tuple_outputs_equivalent(self):
+        return super().test_dict_tuple_outputs_equivalent()
+
+    @skip_mps
+    def test_save_load_optional_components(self):
+        return super().test_save_load_optional_components()
+
+    @skip_mps
+    def test_attention_slicing_forward_pass(self):
+        return super().test_attention_slicing_forward_pass()
+
+
+@nightly
+@require_torch_gpu
+class RepaintPipelineNightlyTests(unittest.TestCase):
+    def tearDown(self):
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_celebahq(self):
+        original_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/"
+            "repaint/celeba_hq_256.png"
+        )
+        mask_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/repaint/mask_256.png"
+        )
+        expected_image = load_numpy(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/"
+            "repaint/celeba_hq_256_result.npy"
+        )
+
+        model_id = "google/ddpm-ema-celebahq-256"
+        unet = UNet2DModel.from_pretrained(model_id)
+        scheduler = RePaintScheduler.from_pretrained(model_id)
+
+        repaint = RePaintPipeline(unet=unet, scheduler=scheduler).to(torch_device)
+        repaint.set_progress_bar_config(disable=None)
+        repaint.enable_attention_slicing()
+
+        generator = torch.manual_seed(0)
+        output = repaint(
+            original_image,
+            mask_image,
+            num_inference_steps=250,
+            eta=0.0,
+            jump_length=10,
+            jump_n_sample=10,
+            generator=generator,
+            output_type="np",
+        )
+        image = output.images[0]
+
+        assert image.shape == (256, 256, 3)
+        assert np.abs(expected_image - image).mean() < 1e-2

tests/pipelines/score_sde_ve/test_score_sde_ve.py

@@ -0,0 +1,91 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import ScoreSdeVePipeline, ScoreSdeVeScheduler, UNet2DModel
+from diffusers.utils.testing_utils import enable_full_determinism, nightly, require_torch, torch_device
+
+
+enable_full_determinism()
+
+
+class ScoreSdeVeipelineFastTests(unittest.TestCase):
+    @property
+    def dummy_uncond_unet(self):
+        torch.manual_seed(0)
+        model = UNet2DModel(
+            block_out_channels=(32, 64),
+            layers_per_block=2,
+            sample_size=32,
+            in_channels=3,
+            out_channels=3,
+            down_block_types=("DownBlock2D", "AttnDownBlock2D"),
+            up_block_types=("AttnUpBlock2D", "UpBlock2D"),
+        )
+        return model
+
+    def test_inference(self):
+        unet = self.dummy_uncond_unet
+        scheduler = ScoreSdeVeScheduler()
+
+        sde_ve = ScoreSdeVePipeline(unet=unet, scheduler=scheduler)
+        sde_ve.to(torch_device)
+        sde_ve.set_progress_bar_config(disable=None)
+
+        generator = torch.manual_seed(0)
+        image = sde_ve(num_inference_steps=2, output_type="numpy", generator=generator).images
+
+        generator = torch.manual_seed(0)
+        image_from_tuple = sde_ve(num_inference_steps=2, output_type="numpy", generator=generator, return_dict=False)[
+            0
+        ]
+
+        image_slice = image[0, -3:, -3:, -1]
+        image_from_tuple_slice = image_from_tuple[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 32, 32, 3)
+        expected_slice = np.array([0.0, 1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+        assert np.abs(image_from_tuple_slice.flatten() - expected_slice).max() < 1e-2
+
+
+@nightly
+@require_torch
+class ScoreSdeVePipelineIntegrationTests(unittest.TestCase):
+    def test_inference(self):
+        model_id = "google/ncsnpp-church-256"
+        model = UNet2DModel.from_pretrained(model_id)
+
+        scheduler = ScoreSdeVeScheduler.from_pretrained(model_id)
+
+        sde_ve = ScoreSdeVePipeline(unet=model, scheduler=scheduler)
+        sde_ve.to(torch_device)
+        sde_ve.set_progress_bar_config(disable=None)
+
+        generator = torch.manual_seed(0)
+        image = sde_ve(num_inference_steps=10, output_type="numpy", generator=generator).images
+
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 256, 256, 3)
+
+        expected_slice = np.array([0.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 0.0, 0.0])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2

tests/pipelines/spectrogram_diffusion/test_spectrogram_diffusion.py

@@ -0,0 +1,246 @@
+# coding=utf-8
+# Copyright 2022 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import DDPMScheduler, MidiProcessor, SpectrogramDiffusionPipeline
+from diffusers.pipelines.spectrogram_diffusion import SpectrogramContEncoder, SpectrogramNotesEncoder, T5FilmDecoder
+from diffusers.utils.testing_utils import (
+    enable_full_determinism,
+    nightly,
+    require_note_seq,
+    require_onnxruntime,
+    require_torch_gpu,
+    skip_mps,
+    torch_device,
+)
+
+from ..pipeline_params import TOKENS_TO_AUDIO_GENERATION_BATCH_PARAMS, TOKENS_TO_AUDIO_GENERATION_PARAMS
+from ..test_pipelines_common import PipelineTesterMixin
+
+
+enable_full_determinism()
+
+
+MIDI_FILE = "./tests/fixtures/elise_format0.mid"
+
+
+# The note-seq package throws an error on import because the default installed version of Ipython
+# is not compatible with python 3.8 which we run in the CI.
+# https://github.com/huggingface/diffusers/actions/runs/4830121056/jobs/8605954838#step:7:98
+@unittest.skip("The note-seq package currently throws an error on import")
+class SpectrogramDiffusionPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
+    pipeline_class = SpectrogramDiffusionPipeline
+    required_optional_params = PipelineTesterMixin.required_optional_params - {
+        "callback",
+        "latents",
+        "callback_steps",
+        "output_type",
+        "num_images_per_prompt",
+    }
+    test_attention_slicing = False
+
+    batch_params = TOKENS_TO_AUDIO_GENERATION_PARAMS
+    params = TOKENS_TO_AUDIO_GENERATION_BATCH_PARAMS
+
+    def get_dummy_components(self):
+        torch.manual_seed(0)
+        notes_encoder = SpectrogramNotesEncoder(
+            max_length=2048,
+            vocab_size=1536,
+            d_model=768,
+            dropout_rate=0.1,
+            num_layers=1,
+            num_heads=1,
+            d_kv=4,
+            d_ff=2048,
+            feed_forward_proj="gated-gelu",
+        )
+
+        continuous_encoder = SpectrogramContEncoder(
+            input_dims=128,
+            targets_context_length=256,
+            d_model=768,
+            dropout_rate=0.1,
+            num_layers=1,
+            num_heads=1,
+            d_kv=4,
+            d_ff=2048,
+            feed_forward_proj="gated-gelu",
+        )
+
+        decoder = T5FilmDecoder(
+            input_dims=128,
+            targets_length=256,
+            max_decoder_noise_time=20000.0,
+            d_model=768,
+            num_layers=1,
+            num_heads=1,
+            d_kv=4,
+            d_ff=2048,
+            dropout_rate=0.1,
+        )
+
+        scheduler = DDPMScheduler()
+
+        components = {
+            "notes_encoder": notes_encoder.eval(),
+            "continuous_encoder": continuous_encoder.eval(),
+            "decoder": decoder.eval(),
+            "scheduler": scheduler,
+            "melgan": None,
+        }
+        return components
+
+    def get_dummy_inputs(self, device, seed=0):
+        if str(device).startswith("mps"):
+            generator = torch.manual_seed(seed)
+        else:
+            generator = torch.Generator(device=device).manual_seed(seed)
+        inputs = {
+            "input_tokens": [
+                [1134, 90, 1135, 1133, 1080, 112, 1132, 1080, 1133, 1079, 133, 1132, 1079, 1133, 1] + [0] * 2033
+            ],
+            "generator": generator,
+            "num_inference_steps": 4,
+            "output_type": "mel",
+        }
+        return inputs
+
+    def test_spectrogram_diffusion(self):
+        device = "cpu"  # ensure determinism for the device-dependent torch.Generator
+        components = self.get_dummy_components()
+        pipe = SpectrogramDiffusionPipeline(**components)
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        output = pipe(**inputs)
+        mel = output.audios
+
+        mel_slice = mel[0, -3:, -3:]
+
+        assert mel_slice.shape == (3, 3)
+        expected_slice = np.array(
+            [-11.512925, -4.788215, -0.46172905, -2.051715, -10.539147, -10.970963, -9.091634, 4.0, 4.0]
+        )
+        assert np.abs(mel_slice.flatten() - expected_slice).max() < 1e-2
+
+    @skip_mps
+    def test_save_load_local(self):
+        return super().test_save_load_local()
+
+    @skip_mps
+    def test_dict_tuple_outputs_equivalent(self):
+        return super().test_dict_tuple_outputs_equivalent()
+
+    @skip_mps
+    def test_save_load_optional_components(self):
+        return super().test_save_load_optional_components()
+
+    @skip_mps
+    def test_attention_slicing_forward_pass(self):
+        return super().test_attention_slicing_forward_pass()
+
+    def test_inference_batch_single_identical(self):
+        pass
+
+    def test_inference_batch_consistent(self):
+        pass
+
+    @skip_mps
+    def test_progress_bar(self):
+        return super().test_progress_bar()
+
+
+@nightly
+@require_torch_gpu
+@require_onnxruntime
+@require_note_seq
+class PipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_callback(self):
+        # TODO - test that pipeline can decode tokens in a callback
+        # so that music can be played live
+        device = torch_device
+
+        pipe = SpectrogramDiffusionPipeline.from_pretrained("google/music-spectrogram-diffusion")
+        melgan = pipe.melgan
+        pipe.melgan = None
+
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        def callback(step, mel_output):
+            # decode mel to audio
+            audio = melgan(input_features=mel_output.astype(np.float32))[0]
+            assert len(audio[0]) == 81920 * (step + 1)
+            # simulate that audio is played
+            return audio
+
+        processor = MidiProcessor()
+        input_tokens = processor(MIDI_FILE)
+
+        input_tokens = input_tokens[:3]
+        generator = torch.manual_seed(0)
+        pipe(input_tokens, num_inference_steps=5, generator=generator, callback=callback, output_type="mel")
+
+    def test_spectrogram_fast(self):
+        device = torch_device
+
+        pipe = SpectrogramDiffusionPipeline.from_pretrained("google/music-spectrogram-diffusion")
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+        processor = MidiProcessor()
+
+        input_tokens = processor(MIDI_FILE)
+        # just run two denoising loops
+        input_tokens = input_tokens[:2]
+
+        generator = torch.manual_seed(0)
+        output = pipe(input_tokens, num_inference_steps=2, generator=generator)
+
+        audio = output.audios[0]
+
+        assert abs(np.abs(audio).sum() - 3612.841) < 1e-1
+
+    def test_spectrogram(self):
+        device = torch_device
+
+        pipe = SpectrogramDiffusionPipeline.from_pretrained("google/music-spectrogram-diffusion")
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        processor = MidiProcessor()
+
+        input_tokens = processor(MIDI_FILE)
+
+        # just run 4 denoising loops
+        input_tokens = input_tokens[:4]
+
+        generator = torch.manual_seed(0)
+        output = pipe(input_tokens, num_inference_steps=100, generator=generator)
+
+        audio = output.audios[0]
+        assert abs(np.abs(audio).sum() - 9389.1111) < 5e-2

tests/pipelines/text_to_video_synthesis/test_text_to_video_zero_sdxl.py

@@ -383,7 +383,7 @@ class TextToVideoZeroSDXLPipelineFastTests(PipelineTesterMixin, unittest.TestCas
 class TextToVideoZeroSDXLPipelineSlowTests(unittest.TestCase):
     def test_full_model(self):
         model_id = "stabilityai/stable-diffusion-xl-base-1.0"
-        pipe = TextToVideoZeroSDXLPipeline.from_pretrained(
+        pipe = self.pipeline_class.from_pretrained(
             model_id, torch_dtype=torch.float16, variant="fp16", use_safetensors=True
         )
         pipe.enable_model_cpu_offload()

tests/pipelines/versatile_diffusion/test_versatile_diffusion_dual_guided.py

@@ -0,0 +1,107 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import tempfile
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import VersatileDiffusionDualGuidedPipeline
+from diffusers.utils.testing_utils import load_image, nightly, require_torch_gpu, torch_device
+
+
+torch.backends.cuda.matmul.allow_tf32 = False
+
+
+@nightly
+@require_torch_gpu
+class VersatileDiffusionDualGuidedPipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_remove_unused_weights_save_load(self):
+        pipe = VersatileDiffusionDualGuidedPipeline.from_pretrained("shi-labs/versatile-diffusion")
+        # remove text_unet
+        pipe.remove_unused_weights()
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        second_prompt = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/versatile_diffusion/benz.jpg"
+        )
+
+        generator = torch.manual_seed(0)
+        image = pipe(
+            prompt="first prompt",
+            image=second_prompt,
+            text_to_image_strength=0.75,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=2,
+            output_type="numpy",
+        ).images
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            pipe.save_pretrained(tmpdirname)
+            pipe = VersatileDiffusionDualGuidedPipeline.from_pretrained(tmpdirname)
+
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        generator = generator.manual_seed(0)
+        new_image = pipe(
+            prompt="first prompt",
+            image=second_prompt,
+            text_to_image_strength=0.75,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=2,
+            output_type="numpy",
+        ).images
+
+        assert np.abs(image - new_image).max() < 1e-5, "Models don't have the same forward pass"
+
+    def test_inference_dual_guided(self):
+        pipe = VersatileDiffusionDualGuidedPipeline.from_pretrained("shi-labs/versatile-diffusion")
+        pipe.remove_unused_weights()
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        first_prompt = "cyberpunk 2077"
+        second_prompt = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/versatile_diffusion/benz.jpg"
+        )
+        generator = torch.manual_seed(0)
+        image = pipe(
+            prompt=first_prompt,
+            image=second_prompt,
+            text_to_image_strength=0.75,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=50,
+            output_type="numpy",
+        ).images
+
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.0787, 0.0849, 0.0826, 0.0812, 0.0807, 0.0795, 0.0818, 0.0798, 0.0779])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2

tests/pipelines/versatile_diffusion/test_versatile_diffusion_image_variation.py

@@ -0,0 +1,57 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import VersatileDiffusionImageVariationPipeline
+from diffusers.utils.testing_utils import load_image, nightly, require_torch_gpu, torch_device
+
+
+torch.backends.cuda.matmul.allow_tf32 = False
+
+
+class VersatileDiffusionImageVariationPipelineFastTests(unittest.TestCase):
+    pass
+
+
+@nightly
+@require_torch_gpu
+class VersatileDiffusionImageVariationPipelineIntegrationTests(unittest.TestCase):
+    def test_inference_image_variations(self):
+        pipe = VersatileDiffusionImageVariationPipeline.from_pretrained("shi-labs/versatile-diffusion")
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        image_prompt = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/versatile_diffusion/benz.jpg"
+        )
+        generator = torch.manual_seed(0)
+        image = pipe(
+            image=image_prompt,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=50,
+            output_type="numpy",
+        ).images
+
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.0441, 0.0469, 0.0507, 0.0575, 0.0632, 0.0650, 0.0865, 0.0909, 0.0945])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2

tests/pipelines/versatile_diffusion/test_versatile_diffusion_mega.py

@@ -0,0 +1,129 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import tempfile
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import VersatileDiffusionPipeline
+from diffusers.utils.testing_utils import load_image, nightly, require_torch_gpu, torch_device
+
+
+torch.backends.cuda.matmul.allow_tf32 = False
+
+
+class VersatileDiffusionMegaPipelineFastTests(unittest.TestCase):
+    pass
+
+
+@nightly
+@require_torch_gpu
+class VersatileDiffusionMegaPipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_from_save_pretrained(self):
+        pipe = VersatileDiffusionPipeline.from_pretrained("shi-labs/versatile-diffusion", torch_dtype=torch.float16)
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        prompt_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/versatile_diffusion/benz.jpg"
+        )
+
+        generator = torch.manual_seed(0)
+        image = pipe.dual_guided(
+            prompt="first prompt",
+            image=prompt_image,
+            text_to_image_strength=0.75,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=2,
+            output_type="numpy",
+        ).images
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            pipe.save_pretrained(tmpdirname)
+            pipe = VersatileDiffusionPipeline.from_pretrained(tmpdirname, torch_dtype=torch.float16)
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        generator = generator.manual_seed(0)
+        new_image = pipe.dual_guided(
+            prompt="first prompt",
+            image=prompt_image,
+            text_to_image_strength=0.75,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=2,
+            output_type="numpy",
+        ).images
+
+        assert np.abs(image - new_image).max() < 1e-5, "Models don't have the same forward pass"
+
+    def test_inference_dual_guided_then_text_to_image(self):
+        pipe = VersatileDiffusionPipeline.from_pretrained("shi-labs/versatile-diffusion", torch_dtype=torch.float16)
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        prompt = "cyberpunk 2077"
+        init_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/versatile_diffusion/benz.jpg"
+        )
+        generator = torch.manual_seed(0)
+        image = pipe.dual_guided(
+            prompt=prompt,
+            image=init_image,
+            text_to_image_strength=0.75,
+            generator=generator,
+            guidance_scale=7.5,
+            num_inference_steps=50,
+            output_type="numpy",
+        ).images
+
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.1448, 0.1619, 0.1741, 0.1086, 0.1147, 0.1128, 0.1199, 0.1165, 0.1001])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-1
+
+        prompt = "A painting of a squirrel eating a burger "
+        generator = torch.manual_seed(0)
+        image = pipe.text_to_image(
+            prompt=prompt, generator=generator, guidance_scale=7.5, num_inference_steps=50, output_type="numpy"
+        ).images
+
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.3367, 0.3169, 0.2656, 0.3870, 0.4790, 0.3796, 0.4009, 0.4878, 0.4778])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-1
+
+        image = pipe.image_variation(init_image, generator=generator, output_type="numpy").images
+
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.3076, 0.3123, 0.3284, 0.3782, 0.3770, 0.3894, 0.4297, 0.4331, 0.4456])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-1

tests/pipelines/versatile_diffusion/test_versatile_diffusion_text_to_image.py

@@ -0,0 +1,87 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import tempfile
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import VersatileDiffusionTextToImagePipeline
+from diffusers.utils.testing_utils import nightly, require_torch_gpu, torch_device
+
+
+torch.backends.cuda.matmul.allow_tf32 = False
+
+
+class VersatileDiffusionTextToImagePipelineFastTests(unittest.TestCase):
+    pass
+
+
+@nightly
+@require_torch_gpu
+class VersatileDiffusionTextToImagePipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_remove_unused_weights_save_load(self):
+        pipe = VersatileDiffusionTextToImagePipeline.from_pretrained("shi-labs/versatile-diffusion")
+        # remove text_unet
+        pipe.remove_unused_weights()
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger "
+        generator = torch.manual_seed(0)
+        image = pipe(
+            prompt=prompt, generator=generator, guidance_scale=7.5, num_inference_steps=2, output_type="numpy"
+        ).images
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            pipe.save_pretrained(tmpdirname)
+            pipe = VersatileDiffusionTextToImagePipeline.from_pretrained(tmpdirname)
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        generator = generator.manual_seed(0)
+        new_image = pipe(
+            prompt=prompt, generator=generator, guidance_scale=7.5, num_inference_steps=2, output_type="numpy"
+        ).images
+
+        assert np.abs(image - new_image).max() < 1e-5, "Models don't have the same forward pass"
+
+    def test_inference_text2img(self):
+        pipe = VersatileDiffusionTextToImagePipeline.from_pretrained(
+            "shi-labs/versatile-diffusion", torch_dtype=torch.float16
+        )
+        pipe.to(torch_device)
+        pipe.set_progress_bar_config(disable=None)
+
+        prompt = "A painting of a squirrel eating a burger "
+        generator = torch.manual_seed(0)
+        image = pipe(
+            prompt=prompt, generator=generator, guidance_scale=7.5, num_inference_steps=50, output_type="numpy"
+        ).images
+
+        image_slice = image[0, 253:256, 253:256, -1]
+
+        assert image.shape == (1, 512, 512, 3)
+        expected_slice = np.array([0.3367, 0.3169, 0.2656, 0.3870, 0.4790, 0.3796, 0.4009, 0.4878, 0.4778])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2

tests/pipelines/vq_diffusion/test_vq_diffusion.py

@@ -0,0 +1,227 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import unittest
+
+import numpy as np
+import torch
+from transformers import CLIPTextConfig, CLIPTextModel, CLIPTokenizer
+
+from diffusers import Transformer2DModel, VQDiffusionPipeline, VQDiffusionScheduler, VQModel
+from diffusers.pipelines.vq_diffusion.pipeline_vq_diffusion import LearnedClassifierFreeSamplingEmbeddings
+from diffusers.utils.testing_utils import load_numpy, nightly, require_torch_gpu, torch_device
+
+
+torch.backends.cuda.matmul.allow_tf32 = False
+
+
+class VQDiffusionPipelineFastTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    @property
+    def num_embed(self):
+        return 12
+
+    @property
+    def num_embeds_ada_norm(self):
+        return 12
+
+    @property
+    def text_embedder_hidden_size(self):
+        return 32
+
+    @property
+    def dummy_vqvae(self):
+        torch.manual_seed(0)
+        model = VQModel(
+            block_out_channels=[32, 64],
+            in_channels=3,
+            out_channels=3,
+            down_block_types=["DownEncoderBlock2D", "DownEncoderBlock2D"],
+            up_block_types=["UpDecoderBlock2D", "UpDecoderBlock2D"],
+            latent_channels=3,
+            num_vq_embeddings=self.num_embed,
+            vq_embed_dim=3,
+        )
+        return model
+
+    @property
+    def dummy_tokenizer(self):
+        tokenizer = CLIPTokenizer.from_pretrained("hf-internal-testing/tiny-random-clip")
+        return tokenizer
+
+    @property
+    def dummy_text_encoder(self):
+        torch.manual_seed(0)
+        config = CLIPTextConfig(
+            bos_token_id=0,
+            eos_token_id=2,
+            hidden_size=self.text_embedder_hidden_size,
+            intermediate_size=37,
+            layer_norm_eps=1e-05,
+            num_attention_heads=4,
+            num_hidden_layers=5,
+            pad_token_id=1,
+            vocab_size=1000,
+        )
+        return CLIPTextModel(config)
+
+    @property
+    def dummy_transformer(self):
+        torch.manual_seed(0)
+
+        height = 12
+        width = 12
+
+        model_kwargs = {
+            "attention_bias": True,
+            "cross_attention_dim": 32,
+            "attention_head_dim": height * width,
+            "num_attention_heads": 1,
+            "num_vector_embeds": self.num_embed,
+            "num_embeds_ada_norm": self.num_embeds_ada_norm,
+            "norm_num_groups": 32,
+            "sample_size": width,
+            "activation_fn": "geglu-approximate",
+        }
+
+        model = Transformer2DModel(**model_kwargs)
+        return model
+
+    def test_vq_diffusion(self):
+        device = "cpu"
+
+        vqvae = self.dummy_vqvae
+        text_encoder = self.dummy_text_encoder
+        tokenizer = self.dummy_tokenizer
+        transformer = self.dummy_transformer
+        scheduler = VQDiffusionScheduler(self.num_embed)
+        learned_classifier_free_sampling_embeddings = LearnedClassifierFreeSamplingEmbeddings(learnable=False)
+
+        pipe = VQDiffusionPipeline(
+            vqvae=vqvae,
+            text_encoder=text_encoder,
+            tokenizer=tokenizer,
+            transformer=transformer,
+            scheduler=scheduler,
+            learned_classifier_free_sampling_embeddings=learned_classifier_free_sampling_embeddings,
+        )
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        prompt = "teddy bear playing in the pool"
+
+        generator = torch.Generator(device=device).manual_seed(0)
+        output = pipe([prompt], generator=generator, num_inference_steps=2, output_type="np")
+        image = output.images
+
+        generator = torch.Generator(device=device).manual_seed(0)
+        image_from_tuple = pipe(
+            [prompt], generator=generator, output_type="np", return_dict=False, num_inference_steps=2
+        )[0]
+
+        image_slice = image[0, -3:, -3:, -1]
+        image_from_tuple_slice = image_from_tuple[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 24, 24, 3)
+
+        expected_slice = np.array([0.6551, 0.6168, 0.5008, 0.5676, 0.5659, 0.4295, 0.6073, 0.5599, 0.4992])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+        assert np.abs(image_from_tuple_slice.flatten() - expected_slice).max() < 1e-2
+
+    def test_vq_diffusion_classifier_free_sampling(self):
+        device = "cpu"
+
+        vqvae = self.dummy_vqvae
+        text_encoder = self.dummy_text_encoder
+        tokenizer = self.dummy_tokenizer
+        transformer = self.dummy_transformer
+        scheduler = VQDiffusionScheduler(self.num_embed)
+        learned_classifier_free_sampling_embeddings = LearnedClassifierFreeSamplingEmbeddings(
+            learnable=True, hidden_size=self.text_embedder_hidden_size, length=tokenizer.model_max_length
+        )
+
+        pipe = VQDiffusionPipeline(
+            vqvae=vqvae,
+            text_encoder=text_encoder,
+            tokenizer=tokenizer,
+            transformer=transformer,
+            scheduler=scheduler,
+            learned_classifier_free_sampling_embeddings=learned_classifier_free_sampling_embeddings,
+        )
+        pipe = pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        prompt = "teddy bear playing in the pool"
+
+        generator = torch.Generator(device=device).manual_seed(0)
+        output = pipe([prompt], generator=generator, num_inference_steps=2, output_type="np")
+        image = output.images
+
+        generator = torch.Generator(device=device).manual_seed(0)
+        image_from_tuple = pipe(
+            [prompt], generator=generator, output_type="np", return_dict=False, num_inference_steps=2
+        )[0]
+
+        image_slice = image[0, -3:, -3:, -1]
+        image_from_tuple_slice = image_from_tuple[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 24, 24, 3)
+
+        expected_slice = np.array([0.6693, 0.6075, 0.4959, 0.5701, 0.5583, 0.4333, 0.6171, 0.5684, 0.4988])
+
+        assert np.abs(image_slice.flatten() - expected_slice).max() < 2.0
+        assert np.abs(image_from_tuple_slice.flatten() - expected_slice).max() < 1e-2
+
+
+@nightly
+@require_torch_gpu
+class VQDiffusionPipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_vq_diffusion_classifier_free_sampling(self):
+        expected_image = load_numpy(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main"
+            "/vq_diffusion/teddy_bear_pool_classifier_free_sampling.npy"
+        )
+
+        pipeline = VQDiffusionPipeline.from_pretrained("microsoft/vq-diffusion-ithq")
+        pipeline = pipeline.to(torch_device)
+        pipeline.set_progress_bar_config(disable=None)
+
+        # requires GPU generator for gumbel softmax
+        # don't use GPU generator in tests though
+        generator = torch.Generator(device=torch_device).manual_seed(0)
+        output = pipeline(
+            "teddy bear playing in the pool",
+            num_images_per_prompt=1,
+            generator=generator,
+            output_type="np",
+        )
+
+        image = output.images[0]
+
+        assert image.shape == (256, 256, 3)
+        assert np.abs(expected_image - image).max() < 2.0