update style bot workflow

* model card gen code

* push modelcard creation

* remove optional from params

* add import

* add use_dora check

* correct lora var use in tags

* make style && make quality

---------

Co-authored-by: Aryan <aryan@huggingface.co>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

.github/workflows/pr_style_bot.yml

@@ -13,39 +13,5 @@ jobs:
     uses: huggingface/huggingface_hub/.github/workflows/style-bot-action.yml@main
     with:
       python_quality_dependencies: "[quality]"
-      pre_commit_script_name: "Download and Compare files from the main branch"
-      pre_commit_script: |
-        echo "Downloading the files from the main branch"
-
-        curl -o main_Makefile https://raw.githubusercontent.com/huggingface/diffusers/main/Makefile
-        curl -o main_setup.py https://raw.githubusercontent.com/huggingface/diffusers/refs/heads/main/setup.py
-        curl -o main_check_doc_toc.py https://raw.githubusercontent.com/huggingface/diffusers/refs/heads/main/utils/check_doc_toc.py
-
-        echo "Compare the files and raise error if needed"
-
-        diff_failed=0
-        if ! diff -q main_Makefile Makefile; then
-          echo "Error: The Makefile has changed. Please ensure it matches the main branch."
-          diff_failed=1
-        fi
-
-        if ! diff -q main_setup.py setup.py; then
-          echo "Error: The setup.py has changed. Please ensure it matches the main branch."
-          diff_failed=1
-        fi
-
-        if ! diff -q main_check_doc_toc.py utils/check_doc_toc.py; then
-          echo "Error: The utils/check_doc_toc.py has changed. Please ensure it matches the main branch."
-          diff_failed=1
-        fi
-
-        if [ $diff_failed -eq 1 ]; then
-          echo "❌ Error happened as we detected changes in the files that should not be changed ❌"
-          exit 1
-        fi
-
-        echo "No changes in the files. Proceeding..."
-        rm -rf main_Makefile main_setup.py main_check_doc_toc.py
-      style_command: "make style && make quality"
     secrets:
       bot_token: ${{ secrets.GITHUB_TOKEN }}

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

@@ -22,18 +22,397 @@
 
 <!-- TODO(aryan): update abstract once paper is out -->
 
+## Generating Videos with Wan 2.1
+
+We will first need to install some addtional dependencies.
+
+```shell
+pip install -u ftfy imageio-ffmpeg imageio
+```
+
+### Text to Video Generation
+
+The following example requires 11GB VRAM to run and uses the smaller `Wan-AI/Wan2.1-T2V-1.3B-Diffusers` model. You can switch it out
+for the larger `Wan2.1-I2V-14B-720P-Diffusers` or `Wan-AI/Wan2.1-I2V-14B-480P-Diffusers` if you have at least 35GB VRAM available.
+
+```python
+from diffusers import WanPipeline
+from diffusers.utils import export_to_video
+
+# Available models: Wan-AI/Wan2.1-I2V-14B-720P-Diffusers or Wan-AI/Wan2.1-I2V-14B-480P-Diffusers
+model_id = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+
+pipe = WanPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
+pipe.enable_model_cpu_offload()
+
+prompt = "A cat and a dog baking a cake together in a kitchen. The cat is carefully measuring flour, while the dog is stirring the batter with a wooden spoon. The kitchen is cozy, with sunlight streaming through the window."
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+num_frames = 33
+
+frames = pipe(prompt=prompt, negative_prompt=negative_prompt, num_frames=num_frames).frames[0]
+export_to_video(frames, "wan-t2v.mp4", fps=16)
+```
+
 <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-a-pipeline) section to learn how to efficiently load the same components into multiple pipelines.
-
+You can improve the quality of the generated video by running the decoding step in full precision.
 </Tip>
 
-Recommendations for inference:
-- VAE in `torch.float32` for better decoding quality.
-- `num_frames` should be of the form `4 * k + 1`, for example `49` or `81`.
-- For smaller resolution videos, try lower values of `shift` (between `2.0` to `5.0`) in the [Scheduler](https://huggingface.co/docs/diffusers/main/en/api/schedulers/flow_match_euler_discrete#diffusers.FlowMatchEulerDiscreteScheduler.shift). For larger resolution videos, try higher values (between `7.0` and `12.0`). The default value is `3.0` for Wan.
+```python
+from diffusers import WanPipeline, AutoencoderKLWan
+from diffusers.utils import export_to_video
 
-### Using a custom scheduler
+model_id = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+
+vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+pipe = WanPipeline.from_pretrained(model_id, vae=vae, torch_dtype=torch.bfloat16)
+
+# replace this with pipe.to("cuda") if you have sufficient VRAM
+pipe.enable_model_cpu_offload()
+
+prompt = "A cat and a dog baking a cake together in a kitchen. The cat is carefully measuring flour, while the dog is stirring the batter with a wooden spoon. The kitchen is cozy, with sunlight streaming through the window."
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+num_frames = 33
+
+frames = pipe(prompt=prompt, num_frames=num_frames).frames[0]
+export_to_video(frames, "wan-t2v.mp4", fps=16)
+```
+
+### Image to Video Generation
+
+The Image to Video pipeline requires loading the `AutoencoderKLWan` and the `CLIPVisionModel` components in full precision. The following example will need at least
+35GB of VRAM to run.
+
+```python
+import torch
+import numpy as np
+from diffusers import AutoencoderKLWan, WanImageToVideoPipeline
+from diffusers.utils import export_to_video, load_image
+from transformers import CLIPVisionModel
+
+# Available models: Wan-AI/Wan2.1-I2V-14B-480P-Diffusers, Wan-AI/Wan2.1-I2V-14B-720P-Diffusers
+model_id = "Wan-AI/Wan2.1-I2V-14B-480P-Diffusers"
+image_encoder = CLIPVisionModel.from_pretrained(
+    model_id, subfolder="image_encoder", torch_dtype=torch.float32
+)
+vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+pipe = WanImageToVideoPipeline.from_pretrained(
+    model_id, vae=vae, image_encoder=image_encoder, torch_dtype=torch.bfloat16
+)
+
+# replace this with pipe.to("cuda") if you have sufficient VRAM
+pipe.enable_model_cpu_offload()
+
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg"
+)
+
+max_area = 480 * 832
+aspect_ratio = image.height / image.width
+mod_value = pipe.vae_scale_factor_spatial * pipe.transformer.config.patch_size[1]
+height = round(np.sqrt(max_area * aspect_ratio)) // mod_value * mod_value
+width = round(np.sqrt(max_area / aspect_ratio)) // mod_value * mod_value
+image = image.resize((width, height))
+
+prompt = (
+    "An astronaut hatching from an egg, on the surface of the moon, the darkness and depth of space realised in "
+    "the background. High quality, ultrarealistic detail and breath-taking movie-like camera shot."
+)
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+
+num_frames = 33
+
+output = pipe(
+    image=image,
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    height=height,
+    width=width,
+    num_frames=num_frames,
+    guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "wan-i2v.mp4", fps=16)
+```
+
+### Video to Video Generation
+
+```python
+import torch
+from diffusers.utils import load_video, export_to_video
+from diffusers import AutoencoderKLWan, WanVideoToVideoPipeline, UniPCMultistepScheduler
+
+# Available models: Wan-AI/Wan2.1-T2V-14B-Diffusers, Wan-AI/Wan2.1-T2V-1.3B-Diffusers
+model_id = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+vae = AutoencoderKLWan.from_pretrained(
+    model_id, subfolder="vae", torch_dtype=torch.float32
+)
+pipe = WanVideoToVideoPipeline.from_pretrained(
+    model_id, vae=vae, torch_dtype=torch.bfloat16
+)
+flow_shift = 3.0  # 5.0 for 720P, 3.0 for 480P
+pipe.scheduler = UniPCMultistepScheduler.from_config(
+    pipe.scheduler.config, flow_shift=flow_shift
+)
+# change to pipe.to("cuda") if you have sufficient VRAM
+pipe.enable_model_cpu_offload()
+
+prompt = "A robot standing on a mountain top. The sun is setting in the background"
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+video = load_video(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/hiker.mp4"
+)
+output = pipe(
+    video=video,
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    height=480,
+    width=512,
+    guidance_scale=7.0,
+    strength=0.7,
+).frames[0]
+
+export_to_video(output, "wan-v2v.mp4", fps=16)
+```
+
+## Memory Optimizations for Wan 2.1
+
+Base inference with the large 14B Wan 2.1 models can take up to 35GB of VRAM when generating videos at 720p resolution. We'll outline a few memory optimizations we can apply to reduce the VRAM required to run the model.
+
+We'll use `Wan-AI/Wan2.1-I2V-14B-720P-Diffusers` model in these examples to demonstrate the memory savings, but the techniques are applicable to all model checkpoints.
+
+### Group Offloading the Transformer and UMT5 Text Encoder
+
+Find more information about group offloading [here](../optimization/memory.md)
+
+#### Block Level Group Offloading
+
+We can reduce our VRAM requirements by applying group offloading to the larger model components of the pipeline; the `WanTransformer3DModel` and `UMT5EncoderModel`. Group offloading will break up the individual modules of a model and offload/onload them onto your GPU as needed during inference. In this example, we'll apply `block_level` offloading, which will group the modules in a model into blocks of size `num_blocks_per_group` and offload/onload them to GPU. Moving to between CPU and GPU does add latency to the inference process. You can trade off between latency and memory savings by increasing or decreasing the `num_blocks_per_group`.
+
+The following example will now only require 14GB of VRAM to run, but will take approximately 30 minutes to generate a video.
+
+```python
+import torch
+import numpy as np
+from diffusers import AutoencoderKLWan, WanTransformer3DModel, WanImageToVideoPipeline
+from diffusers.hooks.group_offloading import apply_group_offloading
+from diffusers.utils import export_to_video, load_image
+from transformers import UMT5EncoderModel, CLIPVisionModel
+
+# Available models: Wan-AI/Wan2.1-I2V-14B-480P-Diffusers, Wan-AI/Wan2.1-I2V-14B-720P-Diffusers
+model_id = "Wan-AI/Wan2.1-I2V-14B-720P-Diffusers"
+image_encoder = CLIPVisionModel.from_pretrained(
+    model_id, subfolder="image_encoder", torch_dtype=torch.float32
+)
+
+text_encoder = UMT5EncoderModel.from_pretrained(model_id, subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+transformer = WanTransformer3DModel.from_pretrained(model_id, subfolder="transformer", torch_dtype=torch.bfloat16)
+
+onload_device = torch.device("cuda")
+offload_device = torch.device("cpu")
+
+apply_group_offloading(text_encoder,
+    onload_device=onload_device,
+    offload_device=offload_device,
+    offload_type="block_level",
+    num_blocks_per_group=4
+)
+
+transformer.enable_group_offload(
+    onload_device=onload_device,
+    offload_device=offload_device,
+    offload_type="block_level",
+    num_blocks_per_group=4,
+)
+pipe = WanImageToVideoPipeline.from_pretrained(
+    model_id,
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    image_encoder=image_encoder,
+    torch_dtype=torch.bfloat16
+)
+# Since we've offloaded the larger models alrady, we can move the rest of the model components to GPU
+pipe.to("cuda")
+
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg"
+)
+
+max_area = 720 * 832
+aspect_ratio = image.height / image.width
+mod_value = pipe.vae_scale_factor_spatial * pipe.transformer.config.patch_size[1]
+height = round(np.sqrt(max_area * aspect_ratio)) // mod_value * mod_value
+width = round(np.sqrt(max_area / aspect_ratio)) // mod_value * mod_value
+image = image.resize((width, height))
+
+prompt = (
+    "An astronaut hatching from an egg, on the surface of the moon, the darkness and depth of space realised in "
+    "the background. High quality, ultrarealistic detail and breath-taking movie-like camera shot."
+)
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+
+num_frames = 33
+
+output = pipe(
+    image=image,
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    height=height,
+    width=width,
+    num_frames=num_frames,
+    guidance_scale=5.0,
+).frames[0]
+
+export_to_video(output, "wan-i2v.mp4", fps=16)
+```
+
+#### Block Level Group Offloading with CUDA Streams
+
+We can speed up group offloading inference, by enabling the use of [CUDA streams](https://pytorch.org/docs/stable/generated/torch.cuda.Stream.html). However, using CUDA streams requires moving the model parameters into pinned memory. This allocation is handled by Pytorch under the hood, and can result in a significant spike in CPU RAM usage. Please consider this option if your CPU RAM is atleast 2X the size of the model you are group offloading.
+
+In the following example we will use CUDA streams when group offloading the `WanTransformer3DModel`. When testing on an A100, this example will require 14GB of VRAM, 52GB of CPU RAM, but will generate a video in approximately 9 minutes.
+
+```python
+import torch
+import numpy as np
+from diffusers import AutoencoderKLWan, WanTransformer3DModel, WanImageToVideoPipeline
+from diffusers.hooks.group_offloading import apply_group_offloading
+from diffusers.utils import export_to_video, load_image
+from transformers import UMT5EncoderModel, CLIPVisionModel
+
+# Available models: Wan-AI/Wan2.1-I2V-14B-480P-Diffusers, Wan-AI/Wan2.1-I2V-14B-720P-Diffusers
+model_id = "Wan-AI/Wan2.1-I2V-14B-720P-Diffusers"
+image_encoder = CLIPVisionModel.from_pretrained(
+    model_id, subfolder="image_encoder", torch_dtype=torch.float32
+)
+
+text_encoder = UMT5EncoderModel.from_pretrained(model_id, subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+transformer = WanTransformer3DModel.from_pretrained(model_id, subfolder="transformer", torch_dtype=torch.bfloat16)
+
+onload_device = torch.device("cuda")
+offload_device = torch.device("cpu")
+
+apply_group_offloading(text_encoder,
+    onload_device=onload_device,
+    offload_device=offload_device,
+    offload_type="block_level",
+    num_blocks_per_group=4
+)
+
+transformer.enable_group_offload(
+    onload_device=onload_device,
+    offload_device=offload_device,
+    offload_type="leaf_level",
+    use_stream=True
+)
+pipe = WanImageToVideoPipeline.from_pretrained(
+    model_id,
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    image_encoder=image_encoder,
+    torch_dtype=torch.bfloat16
+)
+# Since we've offloaded the larger models alrady, we can move the rest of the model components to GPU
+pipe.to("cuda")
+
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg"
+)
+
+max_area = 720 * 832
+aspect_ratio = image.height / image.width
+mod_value = pipe.vae_scale_factor_spatial * pipe.transformer.config.patch_size[1]
+height = round(np.sqrt(max_area * aspect_ratio)) // mod_value * mod_value
+width = round(np.sqrt(max_area / aspect_ratio)) // mod_value * mod_value
+image = image.resize((width, height))
+
+prompt = (
+    "An astronaut hatching from an egg, on the surface of the moon, the darkness and depth of space realised in "
+    "the background. High quality, ultrarealistic detail and breath-taking movie-like camera shot."
+)
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+
+num_frames = 33
+
+output = pipe(
+    image=image,
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    height=height,
+    width=width,
+    num_frames=num_frames,
+    guidance_scale=5.0,
+).frames[0]
+
+export_to_video(output, "wan-i2v.mp4", fps=16)
+```
+
+### Applying Layerwise Casting to the Transformer
+
+Find more information about layerwise casting [here](../optimization/memory.md)
+
+In this example, we will model offloading with layerwise casting. Layerwise casting will downcast each layer's weights to `torch.float8_e4m3fn`, temporarily upcast to `torch.bfloat16` during the forward pass of the layer, then revert to `torch.float8_e4m3fn` afterward. This approach reduces memory requirements by approximately 50% while introducing a minor quality reduction in the generated video due to the precision trade-off.
+
+This example will require 20GB of VRAM.
+
+```python
+import torch
+import numpy as np
+from diffusers import AutoencoderKLWan, WanTransformer3DModel, WanImageToVideoPipeline
+from diffusers.hooks.group_offloading import apply_group_offloading
+from diffusers.utils import export_to_video, load_image
+from transformers import UMT5EncoderModel, CLIPVisionModel
+
+model_id = "Wan-AI/Wan2.1-I2V-14B-720P-Diffusers"
+image_encoder = CLIPVisionModel.from_pretrained(
+    model_id, subfolder="image_encoder", torch_dtype=torch.float32
+)
+text_encoder = UMT5EncoderModel.from_pretrained(model_id, subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+
+transformer = WanTransformer3DModel.from_pretrained(model_id, subfolder="transformer", torch_dtype=torch.bfloat16)
+transformer.enable_layerwise_casting(storage_dtype=torch.float8_e4m3fn, compute_dtype=torch.bfloat16)
+
+pipe = WanImageToVideoPipeline.from_pretrained(
+    model_id,
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    image_encoder=image_encoder,
+    torch_dtype=torch.bfloat16
+)
+pipe.enable_model_cpu_offload()
+image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg")
+
+max_area = 720 * 832
+aspect_ratio = image.height / image.width
+mod_value = pipe.vae_scale_factor_spatial * pipe.transformer.config.patch_size[1]
+height = round(np.sqrt(max_area * aspect_ratio)) // mod_value * mod_value
+width = round(np.sqrt(max_area / aspect_ratio)) // mod_value * mod_value
+image = image.resize((width, height))
+prompt = (
+    "An astronaut hatching from an egg, on the surface of the moon, the darkness and depth of space realised in "
+    "the background. High quality, ultrarealistic detail and breath-taking movie-like camera shot."
+)
+negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+num_frames = 33
+
+output = pipe(
+    image=image,
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    height=height,
+    width=width,
+    num_frames=num_frames,
+    num_inference_steps=50,
+    guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "wan-i2v.mp4", fps=16)
+```
+
+## Using a Custom Scheduler
 
 Wan can be used with many different schedulers, each with their own benefits regarding speed and generation quality. By default, Wan uses the `UniPCMultistepScheduler(prediction_type="flow_prediction", use_flow_sigmas=True, flow_shift=3.0)` scheduler. You can use a different scheduler as follows:
 
@@ -49,11 +428,10 @@ pipe = WanPipeline.from_pretrained("Wan-AI/Wan2.1-T2V-1.3B-Diffusers", scheduler
 pipe.scheduler = <CUSTOM_SCHEDULER_HERE>
 ```
 
-### Using single file loading with Wan
-
-The `WanTransformer3DModel` and `AutoencoderKLWan` models support loading checkpoints in their original format via the `from_single_file` loading 
-method. 
+## Using Single File Loading with Wan 2.1
 
+The `WanTransformer3DModel` and `AutoencoderKLWan` models support loading checkpoints in their original format via the `from_single_file` loading
+method.
 
 ```python
 import torch
@@ -65,6 +443,11 @@ transformer = WanTransformer3DModel.from_single_file(ckpt_path, torch_dtype=torc
 pipe = WanPipeline.from_pretrained("Wan-AI/Wan2.1-T2V-1.3B-Diffusers", transformer=transformer)
 ```
 
+## Recommendations for Inference
+- Keep `AutencoderKLWan` in `torch.float32` for better decoding quality.
+- `num_frames` should satisfy the following constraint: `(num_frames - 1) % 4 == 0`
+- For smaller resolution videos, try lower values of `shift` (between `2.0` to `5.0`) in the [Scheduler](https://huggingface.co/docs/diffusers/main/en/api/schedulers/flow_match_euler_discrete#diffusers.FlowMatchEulerDiscreteScheduler.shift). For larger resolution videos, try higher values (between `7.0` and `12.0`). The default value is `3.0` for Wan.
+
 ## WanPipeline
 
 [[autodoc]] WanPipeline

docs/source/en/installation.md

@@ -161,10 +161,10 @@ Your Python environment will find the `main` version of 🤗 Diffusers on the ne
 
 Model weights and files are downloaded from the Hub to a cache which is usually your home directory. You can change the cache location by specifying the `HF_HOME` or `HUGGINFACE_HUB_CACHE` environment variables or configuring the `cache_dir` parameter in methods like [`~DiffusionPipeline.from_pretrained`].
 
-Cached files allow you to run 🤗 Diffusers offline. To prevent 🤗 Diffusers from connecting to the internet, set the `HF_HUB_OFFLINE` environment variable to `True` and 🤗 Diffusers will only load previously downloaded files in the cache.
+Cached files allow you to run 🤗 Diffusers offline. To prevent 🤗 Diffusers from connecting to the internet, set the `HF_HUB_OFFLINE` environment variable to `1` and 🤗 Diffusers will only load previously downloaded files in the cache.
 
 ```shell
-export HF_HUB_OFFLINE=True
+export HF_HUB_OFFLINE=1
 ```
 
 For more details about managing and cleaning the cache, take a look at the [caching](https://huggingface.co/docs/huggingface_hub/guides/manage-cache) guide.
@@ -179,14 +179,16 @@ Telemetry is only sent when loading models and pipelines from the Hub,
 and it is not collected if you're loading local files.
 
 We understand that not everyone wants to share additional information,and we respect your privacy.
-You can disable telemetry collection by setting the `DISABLE_TELEMETRY` environment variable from your terminal:
+You can disable telemetry collection by setting the `HF_HUB_DISABLE_TELEMETRY` environment variable from your terminal:
 
 On Linux/MacOS:
+
 ```bash
-export DISABLE_TELEMETRY=YES
+export HF_HUB_DISABLE_TELEMETRY=1
 ```
 
 On Windows:
+
 ```bash
-set DISABLE_TELEMETRY=YES
+set HF_HUB_DISABLE_TELEMETRY=1
 ```

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

@@ -95,6 +95,23 @@ Use the Space below to gauge a pipeline's memory requirements before you downloa
     ></iframe>
 </div>
 
+### Specifying Component-Specific Data Types
+
+You can customize the data types for individual sub-models by passing a dictionary to the `torch_dtype` parameter. This allows you to load different components of a pipeline in different floating point precisions. For instance, if you want to load the transformer with `torch.bfloat16` and all other components with `torch.float16`, you can pass a dictionary mapping:
+
+```python
+from diffusers import HunyuanVideoPipeline
+import torch
+
+pipe = HunyuanVideoPipeline.from_pretrained(
+    "hunyuanvideo-community/HunyuanVideo",
+    torch_dtype={'transformer': torch.bfloat16, 'default': torch.float16},
+)
+print(pipe.transformer.dtype, pipe.vae.dtype)  # (torch.bfloat16, torch.float16)
+```
+
+If a component is not explicitly specified in the dictionary and no `default` is provided, it will be loaded with `torch.float32`.
+
 ### Local pipeline
 
 To load a pipeline locally, use [git-lfs](https://git-lfs.github.com/) to manually download a checkpoint to your local disk.

examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py

@@ -71,6 +71,7 @@ from diffusers.utils import (
     convert_unet_state_dict_to_peft,
     is_wandb_available,
 )
+from diffusers.utils.hub_utils import load_or_create_model_card, populate_model_card
 from diffusers.utils.import_utils import is_xformers_available
 from diffusers.utils.torch_utils import is_compiled_module
 
@@ -101,7 +102,7 @@ def determine_scheduler_type(pretrained_model_name_or_path, revision):
 def save_model_card(
     repo_id: str,
     use_dora: bool,
-    images=None,
+    images: list = None,
     base_model: str = None,
     train_text_encoder=False,
     train_text_encoder_ti=False,
@@ -111,20 +112,17 @@ def save_model_card(
     repo_folder=None,
     vae_path=None,
 ):
-    img_str = "widget:\n"
     lora = "lora" if not use_dora else "dora"
-    for i, image in enumerate(images):
-        image.save(os.path.join(repo_folder, f"image_{i}.png"))
-        img_str += f"""
-        - text: '{validation_prompt if validation_prompt else ' ' }'
-          output:
-            url:
-                "image_{i}.png"
-        """
-    if not images:
-        img_str += f"""
-        - text: '{instance_prompt}'
-        """
+
+    widget_dict = []
+    if images is not None:
+        for i, image in enumerate(images):
+            image.save(os.path.join(repo_folder, f"image_{i}.png"))
+            widget_dict.append(
+                {"text": validation_prompt if validation_prompt else " ", "output": {"url": f"image_{i}.png"}}
+            )
+    else:
+        widget_dict.append({"text": instance_prompt})
     embeddings_filename = f"{repo_folder}_emb"
     instance_prompt_webui = re.sub(r"<s\d+>", "", re.sub(r"<s\d+>", embeddings_filename, instance_prompt, count=1))
     ti_keys = ", ".join(f'"{match}"' for match in re.findall(r"<s\d+>", instance_prompt))
@@ -169,23 +167,7 @@ pipeline.load_textual_inversion(state_dict["clip_g"], token=[{ti_keys}], text_en
 to trigger concept `{key}` → use `{tokens}` in your prompt \n
 """
 
-    yaml = f"""---
-tags:
-- stable-diffusion-xl
-- stable-diffusion-xl-diffusers
-- diffusers-training
-- text-to-image
-- diffusers
-- {lora}
-- template:sd-lora
-{img_str}
-base_model: {base_model}
-instance_prompt: {instance_prompt}
-license: openrail++
----
-"""
-
-    model_card = f"""
+    model_description = f"""
 # SDXL LoRA DreamBooth - {repo_id}
 
 <Gallery />
@@ -234,8 +216,25 @@ Special VAE used for training: {vae_path}.
 
 {license}
 """
-    with open(os.path.join(repo_folder, "README.md"), "w") as f:
-        f.write(yaml + model_card)
+    model_card = load_or_create_model_card(
+        repo_id_or_path=repo_id,
+        from_training=True,
+        license="openrail++",
+        base_model=base_model,
+        prompt=instance_prompt,
+        model_description=model_description,
+        widget=widget_dict,
+    )
+    tags = [
+        "text-to-image",
+        "stable-diffusion-xl",
+        "stable-diffusion-xl-diffusers",
+        "text-to-image",
+        "diffusers",
+        lora,
+        "template:sd-lora",
+    ]
+    model_card = populate_model_card(model_card, tags=tags)
 
 
 def log_validation(

examples/community/README.md

@@ -85,7 +85,7 @@ PIXART-α Controlnet pipeline | Implementation of the controlnet model for pixar
 | Stable Diffusion XL Attentive Eraser Pipeline |[[AAAI2025 Oral] Attentive Eraser](https://github.com/Anonym0u3/AttentiveEraser) is a novel tuning-free method that enhances object removal capabilities in pre-trained diffusion models.|[Stable Diffusion XL Attentive Eraser Pipeline](#stable-diffusion-xl-attentive-eraser-pipeline)|-|[Wenhao Sun](https://github.com/Anonym0u3) and [Benlei Cui](https://github.com/Benny079)|
 | Perturbed-Attention Guidance |StableDiffusionPAGPipeline is a modification of StableDiffusionPipeline to support Perturbed-Attention Guidance (PAG).|[Perturbed-Attention Guidance](#perturbed-attention-guidance)|[Notebook](https://github.com/huggingface/notebooks/blob/main/diffusers/perturbed_attention_guidance.ipynb)|[Hyoungwon Cho](https://github.com/HyoungwonCho)|
 | CogVideoX DDIM Inversion Pipeline | Implementation of DDIM inversion and guided attention-based editing denoising process on CogVideoX. | [CogVideoX DDIM Inversion Pipeline](#cogvideox-ddim-inversion-pipeline) | - | [LittleNyima](https://github.com/LittleNyima) |
-
+| FaithDiff Stable Diffusion XL Pipeline | Implementation of [(CVPR 2025) FaithDiff: Unleashing Diffusion Priors for Faithful Image Super-resolutionUnleashing Diffusion Priors for Faithful Image Super-resolution](https://arxiv.org/abs/2411.18824) - FaithDiff is a faithful image super-resolution method that leverages latent diffusion models by actively adapting the diffusion prior and jointly fine-tuning its components (encoder and diffusion model) with an alignment module to ensure high fidelity and structural consistency. | [FaithDiff Stable Diffusion XL Pipeline](#faithdiff-stable-diffusion-xl-pipeline) | [![Hugging Face Models](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Models-blue)](https://huggingface.co/jychen9811/FaithDiff) | [Junyang Chen, Jinshan Pan, Jiangxin Dong, IMAG Lab, (Adapted by Eliseu Silva)](https://github.com/JyChen9811/FaithDiff) |
 To load a custom pipeline you just need to pass the `custom_pipeline` argument to `DiffusionPipeline`, as one of the files in `diffusers/examples/community`. Feel free to send a PR with your own pipelines, we will merge them quickly.
 
 ```py
@@ -5334,3 +5334,103 @@ output = pipeline_for_inversion(
 pipeline.export_latents_to_video(output.inverse_latents[-1], "path/to/inverse_video.mp4", fps=8)
 pipeline.export_latents_to_video(output.recon_latents[-1], "path/to/recon_video.mp4", fps=8)
 ```
+# FaithDiff Stable Diffusion XL Pipeline
+
+[Project](https://jychen9811.github.io/FaithDiff_page/) / [GitHub](https://github.com/JyChen9811/FaithDiff/)
+
+This the implementation of the FaithDiff pipeline for SDXL, adapted to use the HuggingFace Diffusers.
+
+For more details see the project links above.
+
+## Example Usage
+
+This example upscale and restores a low-quality image. The input image has a resolution of 512x512 and will be upscaled at a scale of 2x, to a final resolution of 1024x1024. It is possible to upscale to a larger scale, but it is recommended that the input image be at least 1024x1024 in these cases. To upscale this image by 4x, for example, it would be recommended to re-input the result into a new 2x processing, thus performing progressive scaling.
+
+````py
+import random
+import numpy as np
+import torch
+from diffusers import DiffusionPipeline, AutoencoderKL, UniPCMultistepScheduler
+from huggingface_hub import hf_hub_download
+from diffusers.utils import load_image
+from PIL import Image
+
+device = "cuda"
+dtype = torch.float16
+MAX_SEED = np.iinfo(np.int32).max
+
+# Download weights for additional unet layers
+model_file = hf_hub_download(
+    "jychen9811/FaithDiff",
+    filename="FaithDiff.bin", local_dir="./proc_data/faithdiff", local_dir_use_symlinks=False
+)
+
+# Initialize the models and pipeline
+vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=dtype)
+
+model_id = "SG161222/RealVisXL_V4.0"
+pipe = DiffusionPipeline.from_pretrained(
+    model_id,
+    torch_dtype=dtype,
+    vae=vae,
+    unet=None, #<- Do not load with original model.
+    custom_pipeline="pipeline_faithdiff_stable_diffusion_xl",    
+    use_safetensors=True,
+    variant="fp16",
+).to(device)
+
+# Here we need use pipeline internal unet model
+pipe.unet = pipe.unet_model.from_pretrained(model_id, subfolder="unet", variant="fp16", use_safetensors=True)
+
+# Load aditional layers to the model
+pipe.unet.load_additional_layers(weight_path="proc_data/faithdiff/FaithDiff.bin", dtype=dtype)
+
+# Enable vae tiling
+pipe.set_encoder_tile_settings()
+pipe.enable_vae_tiling()
+
+# Optimization
+pipe.enable_model_cpu_offload()
+
+# Set selected scheduler
+pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
+
+#input params
+prompt = "The image features a woman in her 55s with blonde hair and a white shirt, smiling at the camera. She appears to be in a good mood and is wearing a white scarf around her neck. "
+upscale = 2 # scale here
+start_point = "lr" # or "noise"
+latent_tiled_overlap = 0.5
+latent_tiled_size = 1024
+
+# Load image
+lq_image = load_image("https://huggingface.co/datasets/DEVAIEXP/assets/resolve/main/woman.png")
+original_height = lq_image.height
+original_width = lq_image.width
+print(f"Current resolution: H:{original_height} x W:{original_width}")
+
+width = original_width * int(upscale)
+height = original_height * int(upscale)
+print(f"Final resolution: H:{height} x W:{width}")
+
+# Restoration
+image = lq_image.resize((width, height), Image.LANCZOS)
+input_image, width_init, height_init, width_now, height_now = pipe.check_image_size(image)
+
+generator = torch.Generator(device=device).manual_seed(random.randint(0, MAX_SEED))
+gen_image = pipe(lr_img=input_image, 
+                 prompt = prompt,                  
+                 num_inference_steps=20, 
+                 guidance_scale=5, 
+                 generator=generator, 
+                 start_point=start_point, 
+                 height = height_now, 
+                 width=width_now, 
+                 overlap=latent_tiled_overlap, 
+                 target_size=(latent_tiled_size, latent_tiled_size)
+                ).images[0]
+
+cropped_image = gen_image.crop((0, 0, width_init, height_init))
+cropped_image.save("data/result.png")
+````
+### Result
+[<img src="https://huggingface.co/datasets/DEVAIEXP/assets/resolve/main/faithdiff_restored.PNG" width="512px" height="512px"/>](https://imgsli.com/MzY1NzE2)

examples/community/lpw_stable_diffusion_xl.py

@@ -1773,7 +1773,7 @@ class SDXLLongPromptWeightingPipeline(
                         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"
+                        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:
@@ -1924,7 +1924,22 @@ class SDXLLongPromptWeightingPipeline(
                 self.upcast_vae()
                 latents = latents.to(next(iter(self.vae.post_quant_conv.parameters())).dtype)
 
-            image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]
+            # unscale/denormalize the latents
+            # denormalize with the mean and std if available and not None
+            has_latents_mean = hasattr(self.vae.config, "latents_mean") and self.vae.config.latents_mean is not None
+            has_latents_std = hasattr(self.vae.config, "latents_std") and self.vae.config.latents_std is not None
+            if has_latents_mean and has_latents_std:
+                latents_mean = (
+                    torch.tensor(self.vae.config.latents_mean).view(1, 4, 1, 1).to(latents.device, latents.dtype)
+                )
+                latents_std = (
+                    torch.tensor(self.vae.config.latents_std).view(1, 4, 1, 1).to(latents.device, latents.dtype)
+                )
+                latents = latents * latents_std / self.vae.config.scaling_factor + latents_mean
+            else:
+                latents = latents / self.vae.config.scaling_factor
+
+            image = self.vae.decode(latents, return_dict=False)[0]
 
             # cast back to fp16 if needed
             if needs_upcasting:

src/diffusers/__init__.py

@@ -509,6 +509,7 @@ else:
             "VQDiffusionPipeline",
             "WanImageToVideoPipeline",
             "WanPipeline",
+            "WanVideoToVideoPipeline",
             "WuerstchenCombinedPipeline",
             "WuerstchenDecoderPipeline",
             "WuerstchenPriorPipeline",
@@ -1062,6 +1063,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             VQDiffusionPipeline,
             WanImageToVideoPipeline,
             WanPipeline,
+            WanVideoToVideoPipeline,
             WuerstchenCombinedPipeline,
             WuerstchenDecoderPipeline,
             WuerstchenPriorPipeline,

src/diffusers/configuration_utils.py

@@ -35,6 +35,7 @@ from huggingface_hub.utils import (
     validate_hf_hub_args,
 )
 from requests import HTTPError
+from typing_extensions import Self
 
 from . import __version__
 from .utils import (
@@ -185,7 +186,9 @@ class ConfigMixin:
             )
 
     @classmethod
-    def from_config(cls, config: Union[FrozenDict, Dict[str, Any]] = None, return_unused_kwargs=False, **kwargs):
+    def from_config(
+        cls, config: Union[FrozenDict, Dict[str, Any]] = None, return_unused_kwargs=False, **kwargs
+    ) -> Union[Self, Tuple[Self, Dict[str, Any]]]:
         r"""
         Instantiate a Python class from a config dictionary.
 

src/diffusers/loaders/peft.py

@@ -307,6 +307,9 @@ class PeftAdapterMixin:
             try:
                 inject_adapter_in_model(lora_config, self, adapter_name=adapter_name, **peft_kwargs)
                 incompatible_keys = set_peft_model_state_dict(self, state_dict, adapter_name, **peft_kwargs)
+                # Set peft config loaded flag to True if module has been successfully injected and incompatible keys retrieved
+                if not self._hf_peft_config_loaded:
+                    self._hf_peft_config_loaded = True
             except Exception as e:
                 # In case `inject_adapter_in_model()` was unsuccessful even before injecting the `peft_config`.
                 if hasattr(self, "peft_config"):

src/diffusers/models/autoencoders/autoencoder_kl_cogvideox.py

@@ -105,6 +105,7 @@ class CogVideoXCausalConv3d(nn.Module):
         self.width_pad = width_pad
         self.time_pad = time_pad
         self.time_causal_padding = (width_pad, width_pad, height_pad, height_pad, time_pad, 0)
+        self.const_padding_conv3d = (0, self.width_pad, self.height_pad)
 
         self.temporal_dim = 2
         self.time_kernel_size = time_kernel_size
@@ -117,6 +118,8 @@ class CogVideoXCausalConv3d(nn.Module):
             kernel_size=kernel_size,
             stride=stride,
             dilation=dilation,
+            padding=0 if self.pad_mode == "replicate" else self.const_padding_conv3d,
+            padding_mode="zeros",
         )
 
     def fake_context_parallel_forward(
@@ -137,9 +140,7 @@ class CogVideoXCausalConv3d(nn.Module):
         if self.pad_mode == "replicate":
             conv_cache = None
         else:
-            padding_2d = (self.width_pad, self.width_pad, self.height_pad, self.height_pad)
             conv_cache = inputs[:, :, -self.time_kernel_size + 1 :].clone()
-            inputs = F.pad(inputs, padding_2d, mode="constant", value=0)
 
         output = self.conv(inputs)
         return output, conv_cache

src/diffusers/models/transformers/latte_transformer_3d.py

@@ -273,7 +273,7 @@ class LatteTransformer3DModel(ModelMixin, ConfigMixin, CacheMixin):
                 hidden_states = hidden_states.reshape(-1, hidden_states.shape[-2], hidden_states.shape[-1])
 
                 if i == 0 and num_frame > 1:
-                    hidden_states = hidden_states + self.temp_pos_embed
+                    hidden_states = hidden_states + self.temp_pos_embed.to(hidden_states.dtype)
 
                 if torch.is_grad_enabled() and self.gradient_checkpointing:
                     hidden_states = self._gradient_checkpointing_func(

src/diffusers/models/transformers/transformer_ltx.py

@@ -26,6 +26,7 @@ from ...utils import USE_PEFT_BACKEND, logging, scale_lora_layers, unscale_lora_
 from ...utils.torch_utils import maybe_allow_in_graph
 from ..attention import FeedForward
 from ..attention_processor import Attention
+from ..cache_utils import CacheMixin
 from ..embeddings import PixArtAlphaTextProjection
 from ..modeling_outputs import Transformer2DModelOutput
 from ..modeling_utils import ModelMixin
@@ -298,7 +299,7 @@ class LTXVideoTransformerBlock(nn.Module):
 
 
 @maybe_allow_in_graph
-class LTXVideoTransformer3DModel(ModelMixin, ConfigMixin, FromOriginalModelMixin, PeftAdapterMixin):
+class LTXVideoTransformer3DModel(ModelMixin, ConfigMixin, FromOriginalModelMixin, PeftAdapterMixin, CacheMixin):
     r"""
     A Transformer model for video-like data used in [LTX](https://huggingface.co/Lightricks/LTX-Video).
 

src/diffusers/models/transformers/transformer_wan.py

@@ -24,6 +24,7 @@ from ...loaders import FromOriginalModelMixin, PeftAdapterMixin
 from ...utils import USE_PEFT_BACKEND, logging, scale_lora_layers, unscale_lora_layers
 from ..attention import FeedForward
 from ..attention_processor import Attention
+from ..cache_utils import CacheMixin
 from ..embeddings import PixArtAlphaTextProjection, TimestepEmbedding, Timesteps, get_1d_rotary_pos_embed
 from ..modeling_outputs import Transformer2DModelOutput
 from ..modeling_utils import ModelMixin
@@ -288,7 +289,7 @@ class WanTransformerBlock(nn.Module):
         return hidden_states
 
 
-class WanTransformer3DModel(ModelMixin, ConfigMixin, PeftAdapterMixin, FromOriginalModelMixin):
+class WanTransformer3DModel(ModelMixin, ConfigMixin, PeftAdapterMixin, FromOriginalModelMixin, CacheMixin):
     r"""
     A Transformer model for video-like data used in the Wan model.
 

src/diffusers/pipelines/__init__.py

@@ -356,7 +356,7 @@ else:
         "WuerstchenDecoderPipeline",
         "WuerstchenPriorPipeline",
     ]
-    _import_structure["wan"] = ["WanPipeline", "WanImageToVideoPipeline"]
+    _import_structure["wan"] = ["WanPipeline", "WanImageToVideoPipeline", "WanVideoToVideoPipeline"]
 try:
     if not is_onnx_available():
         raise OptionalDependencyNotAvailable()
@@ -709,7 +709,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
             UniDiffuserPipeline,
             UniDiffuserTextDecoder,
         )
-        from .wan import WanImageToVideoPipeline, WanPipeline
+        from .wan import WanImageToVideoPipeline, WanPipeline, WanVideoToVideoPipeline
         from .wuerstchen import (
             WuerstchenCombinedPipeline,
             WuerstchenDecoderPipeline,

src/diffusers/pipelines/cogview4/pipeline_cogview4.py

@@ -213,9 +213,7 @@ class CogView4Pipeline(DiffusionPipeline, CogView4LoraLoaderMixin):
                 device=text_input_ids.device,
             )
             text_input_ids = torch.cat([pad_ids, text_input_ids], dim=1)
-        prompt_embeds = self.text_encoder(
-            text_input_ids.to(self.text_encoder.device), output_hidden_states=True
-        ).hidden_states[-2]
+        prompt_embeds = self.text_encoder(text_input_ids.to(device), output_hidden_states=True).hidden_states[-2]
 
         prompt_embeds = prompt_embeds.to(dtype=dtype, device=device)
         return prompt_embeds

src/diffusers/pipelines/cogview4/pipeline_cogview4_control.py

@@ -216,9 +216,7 @@ class CogView4ControlPipeline(DiffusionPipeline):
                 device=text_input_ids.device,
             )
             text_input_ids = torch.cat([pad_ids, text_input_ids], dim=1)
-        prompt_embeds = self.text_encoder(
-            text_input_ids.to(self.text_encoder.device), output_hidden_states=True
-        ).hidden_states[-2]
+        prompt_embeds = self.text_encoder(text_input_ids.to(device), output_hidden_states=True).hidden_states[-2]
 
         prompt_embeds = prompt_embeds.to(dtype=dtype, device=device)
         return prompt_embeds

src/diffusers/pipelines/ltx/pipeline_ltx.py

@@ -489,6 +489,10 @@ class LTXPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraLoaderMixi
     def num_timesteps(self):
         return self._num_timesteps
 
+    @property
+    def current_timestep(self):
+        return self._current_timestep
+
     @property
     def attention_kwargs(self):
         return self._attention_kwargs
@@ -622,6 +626,7 @@ class LTXPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraLoaderMixi
         self._guidance_scale = guidance_scale
         self._attention_kwargs = attention_kwargs
         self._interrupt = False
+        self._current_timestep = None
 
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
@@ -706,6 +711,8 @@ class LTXPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraLoaderMixi
                 if self.interrupt:
                     continue
 
+                self._current_timestep = t
+
                 latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
                 latent_model_input = latent_model_input.to(prompt_embeds.dtype)
 

src/diffusers/pipelines/ltx/pipeline_ltx_condition.py

@@ -774,6 +774,10 @@ class LTXConditionPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraL
     def num_timesteps(self):
         return self._num_timesteps
 
+    @property
+    def current_timestep(self):
+        return self._current_timestep
+
     @property
     def attention_kwargs(self):
         return self._attention_kwargs
@@ -933,6 +937,7 @@ class LTXConditionPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraL
         self._guidance_scale = guidance_scale
         self._attention_kwargs = attention_kwargs
         self._interrupt = False
+        self._current_timestep = None
 
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
@@ -1066,6 +1071,8 @@ class LTXConditionPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLoraL
                 if self.interrupt:
                     continue
 
+                self._current_timestep = t
+
                 if image_cond_noise_scale > 0:
                     # Add timestep-dependent noise to the hard-conditioning latents
                     # This helps with motion continuity, especially when conditioned on a single frame

src/diffusers/pipelines/ltx/pipeline_ltx_image2video.py

@@ -487,19 +487,21 @@ class LTXImageToVideoPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLo
     ) -> torch.Tensor:
         height = height // self.vae_spatial_compression_ratio
         width = width // self.vae_spatial_compression_ratio
-        num_frames = (
-            (num_frames - 1) // self.vae_temporal_compression_ratio + 1 if latents is None else latents.size(2)
-        )
+        num_frames = (num_frames - 1) // self.vae_temporal_compression_ratio + 1
 
         shape = (batch_size, num_channels_latents, num_frames, height, width)
         mask_shape = (batch_size, 1, num_frames, height, width)
 
         if latents is not None:
-            conditioning_mask = latents.new_zeros(shape)
+            conditioning_mask = latents.new_zeros(mask_shape)
             conditioning_mask[:, :, 0] = 1.0
             conditioning_mask = self._pack_latents(
                 conditioning_mask, self.transformer_spatial_patch_size, self.transformer_temporal_patch_size
-            )
+            ).squeeze(-1)
+            if latents.ndim != 3 or latents.shape[:2] != conditioning_mask.shape:
+                raise ValueError(
+                    f"Provided `latents` tensor has shape {latents.shape}, but the expected shape is {conditioning_mask.shape + (num_channels_latents,)}."
+                )
             return latents.to(device=device, dtype=dtype), conditioning_mask
 
         if isinstance(generator, list):
@@ -548,6 +550,10 @@ class LTXImageToVideoPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLo
     def num_timesteps(self):
         return self._num_timesteps
 
+    @property
+    def current_timestep(self):
+        return self._current_timestep
+
     @property
     def attention_kwargs(self):
         return self._attention_kwargs
@@ -684,6 +690,7 @@ class LTXImageToVideoPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLo
         self._guidance_scale = guidance_scale
         self._attention_kwargs = attention_kwargs
         self._interrupt = False
+        self._current_timestep = None
 
         # 2. Define call parameters
         if prompt is not None and isinstance(prompt, str):
@@ -776,6 +783,8 @@ class LTXImageToVideoPipeline(DiffusionPipeline, FromSingleFileMixin, LTXVideoLo
                 if self.interrupt:
                     continue
 
+                self._current_timestep = t
+
                 latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
                 latent_model_input = latent_model_input.to(prompt_embeds.dtype)
 

src/diffusers/pipelines/pipeline_loading_utils.py

@@ -592,6 +592,11 @@ def _get_final_device_map(device_map, pipeline_class, passed_class_obj, init_dic
                 loaded_sub_model = passed_class_obj[name]
 
         else:
+            sub_model_dtype = (
+                torch_dtype.get(name, torch_dtype.get("default", torch.float32))
+                if isinstance(torch_dtype, dict)
+                else torch_dtype
+            )
             loaded_sub_model = _load_empty_model(
                 library_name=library_name,
                 class_name=class_name,
@@ -600,7 +605,7 @@ def _get_final_device_map(device_map, pipeline_class, passed_class_obj, init_dic
                 is_pipeline_module=is_pipeline_module,
                 pipeline_class=pipeline_class,
                 name=name,
-                torch_dtype=torch_dtype,
+                torch_dtype=sub_model_dtype,
                 cached_folder=kwargs.get("cached_folder", None),
                 force_download=kwargs.get("force_download", None),
                 proxies=kwargs.get("proxies", None),
@@ -616,7 +621,12 @@ def _get_final_device_map(device_map, pipeline_class, passed_class_obj, init_dic
     # Obtain a sorted dictionary for mapping the model-level components
     # to their sizes.
     module_sizes = {
-        module_name: compute_module_sizes(module, dtype=torch_dtype)[""]
+        module_name: compute_module_sizes(
+            module,
+            dtype=torch_dtype.get(module_name, torch_dtype.get("default", torch.float32))
+            if isinstance(torch_dtype, dict)
+            else torch_dtype,
+        )[""]
         for module_name, module in init_empty_modules.items()
         if isinstance(module, torch.nn.Module)
     }

src/diffusers/pipelines/pipeline_utils.py

@@ -552,9 +552,12 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing a dduf file
-            torch_dtype (`str` or `torch.dtype`, *optional*):
+            torch_dtype (`str` or `torch.dtype` or `dict[str, Union[str, torch.dtype]]`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype. If "auto" is passed, the
-                dtype is automatically derived from the model's weights.
+                dtype is automatically derived from the model's weights. To load submodels with different dtype pass a
+                `dict` (for example `{'transformer': torch.bfloat16, 'vae': torch.float16}`). Set the default dtype for
+                unspecified components with `default` (for example `{'transformer': torch.bfloat16, 'default':
+                torch.float16}`). If a component is not specified and no default is set, `torch.float32` is used.
             custom_pipeline (`str`, *optional*):
 
                 <Tip warning={true}>
@@ -703,7 +706,7 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
         use_onnx = kwargs.pop("use_onnx", None)
         load_connected_pipeline = kwargs.pop("load_connected_pipeline", False)
 
-        if torch_dtype is not None and not isinstance(torch_dtype, torch.dtype):
+        if torch_dtype is not None and not isinstance(torch_dtype, dict) and not isinstance(torch_dtype, torch.dtype):
             torch_dtype = torch.float32
             logger.warning(
                 f"Passed `torch_dtype` {torch_dtype} is not a `torch.dtype`. Defaulting to `torch.float32`."
@@ -950,6 +953,11 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
                 loaded_sub_model = passed_class_obj[name]
             else:
                 # load sub model
+                sub_model_dtype = (
+                    torch_dtype.get(name, torch_dtype.get("default", torch.float32))
+                    if isinstance(torch_dtype, dict)
+                    else torch_dtype
+                )
                 loaded_sub_model = load_sub_model(
                     library_name=library_name,
                     class_name=class_name,
@@ -957,7 +965,7 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
                     pipelines=pipelines,
                     is_pipeline_module=is_pipeline_module,
                     pipeline_class=pipeline_class,
-                    torch_dtype=torch_dtype,
+                    torch_dtype=sub_model_dtype,
                     provider=provider,
                     sess_options=sess_options,
                     device_map=current_device_map,
@@ -998,7 +1006,7 @@ class DiffusionPipeline(ConfigMixin, PushToHubMixin):
             for module in missing_modules:
                 init_kwargs[module] = passed_class_obj.get(module, None)
         elif len(missing_modules) > 0:
-            passed_modules = set(list(init_kwargs.keys()) + list(passed_class_obj.keys())) - optional_kwargs
+            passed_modules = set(list(init_kwargs.keys()) + list(passed_class_obj.keys())) - set(optional_kwargs)
             raise ValueError(
                 f"Pipeline {pipeline_class} expected {expected_modules}, but only {passed_modules} were passed."
             )

src/diffusers/pipelines/wan/__init__.py

@@ -24,7 +24,7 @@ except OptionalDependencyNotAvailable:
 else:
     _import_structure["pipeline_wan"] = ["WanPipeline"]
     _import_structure["pipeline_wan_i2v"] = ["WanImageToVideoPipeline"]
-
+    _import_structure["pipeline_wan_video2video"] = ["WanVideoToVideoPipeline"]
 if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
     try:
         if not (is_transformers_available() and is_torch_available()):
@@ -35,6 +35,7 @@ if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
     else:
         from .pipeline_wan import WanPipeline
         from .pipeline_wan_i2v import WanImageToVideoPipeline
+        from .pipeline_wan_video2video import WanVideoToVideoPipeline
 
 else:
     import sys

src/diffusers/pipelines/wan/pipeline_wan.py

@@ -458,6 +458,13 @@ class WanPipeline(DiffusionPipeline, WanLoraLoaderMixin):
             callback_on_step_end_tensor_inputs,
         )
 
+        if num_frames % self.vae_scale_factor_temporal != 1:
+            logger.warning(
+                f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. Rounding to the nearest number."
+            )
+            num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
+        num_frames = max(num_frames, 1)
+
         self._guidance_scale = guidance_scale
         self._attention_kwargs = attention_kwargs
         self._current_timestep = None

src/diffusers/pipelines/wan/pipeline_wan_i2v.py

@@ -220,8 +220,13 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
 
         return prompt_embeds
 
-    def encode_image(self, image: PipelineImageInput):
-        image = self.image_processor(images=image, return_tensors="pt").to(self.device)
+    def encode_image(
+        self,
+        image: PipelineImageInput,
+        device: Optional[torch.device] = None,
+    ):
+        device = device or self._execution_device
+        image = self.image_processor(images=image, return_tensors="pt").to(device)
         image_embeds = self.image_encoder(**image, output_hidden_states=True)
         return image_embeds.hidden_states[-2]
 
@@ -554,6 +559,13 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
             callback_on_step_end_tensor_inputs,
         )
 
+        if num_frames % self.vae_scale_factor_temporal != 1:
+            logger.warning(
+                f"`num_frames - 1` has to be divisible by {self.vae_scale_factor_temporal}. Rounding to the nearest number."
+            )
+            num_frames = num_frames // self.vae_scale_factor_temporal * self.vae_scale_factor_temporal + 1
+        num_frames = max(num_frames, 1)
+
         self._guidance_scale = guidance_scale
         self._attention_kwargs = attention_kwargs
         self._current_timestep = None
@@ -587,7 +599,7 @@ class WanImageToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
         if negative_prompt_embeds is not None:
             negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)
 
-        image_embeds = self.encode_image(image)
+        image_embeds = self.encode_image(image, device)
         image_embeds = image_embeds.repeat(batch_size, 1, 1)
         image_embeds = image_embeds.to(transformer_dtype)
 

src/diffusers/pipelines/wan/pipeline_wan_video2video.py

@@ -0,0 +1,725 @@
+# Copyright 2025 The Wan Team 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.
+
+import html
+import inspect
+from typing import Any, Callable, Dict, List, Optional, Union
+
+import ftfy
+import regex as re
+import torch
+from PIL import Image
+from transformers import AutoTokenizer, UMT5EncoderModel
+
+from ...callbacks import MultiPipelineCallbacks, PipelineCallback
+from ...loaders import WanLoraLoaderMixin
+from ...models import AutoencoderKLWan, WanTransformer3DModel
+from ...schedulers import FlowMatchEulerDiscreteScheduler
+from ...utils import is_torch_xla_available, logging, replace_example_docstring
+from ...utils.torch_utils import randn_tensor
+from ...video_processor import VideoProcessor
+from ..pipeline_utils import DiffusionPipeline
+from .pipeline_output import WanPipelineOutput
+
+
+if is_torch_xla_available():
+    import torch_xla.core.xla_model as xm
+
+    XLA_AVAILABLE = True
+else:
+    XLA_AVAILABLE = False
+
+logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+
+
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```python
+        >>> import torch
+        >>> from diffusers.utils import export_to_video
+        >>> from diffusers import AutoencoderKLWan, WanVideoToVideoPipeline
+        >>> from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
+
+        >>> # Available models: Wan-AI/Wan2.1-T2V-14B-Diffusers, Wan-AI/Wan2.1-T2V-1.3B-Diffusers
+        >>> model_id = "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+        >>> vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
+        >>> pipe = WanVideoToVideoPipeline.from_pretrained(model_id, vae=vae, torch_dtype=torch.bfloat16)
+        >>> flow_shift = 3.0  # 5.0 for 720P, 3.0 for 480P
+        >>> pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config, flow_shift=flow_shift)
+        >>> pipe.to("cuda")
+
+        >>> prompt = "A robot standing on a mountain top. The sun is setting in the background"
+        >>> negative_prompt = "Bright tones, overexposed, static, blurred details, subtitles, style, works, paintings, images, static, overall gray, worst quality, low quality, JPEG compression residue, ugly, incomplete, extra fingers, poorly drawn hands, poorly drawn faces, deformed, disfigured, misshapen limbs, fused fingers, still picture, messy background, three legs, many people in the background, walking backwards"
+        >>> video = load_video(
+        ...     "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/hiker.mp4"
+        ... )
+        >>> output = pipe(
+        ...     video=video,
+        ...     prompt=prompt,
+        ...     negative_prompt=negative_prompt,
+        ...     height=480,
+        ...     width=720,
+        ...     guidance_scale=5.0,
+        ...     strength=0.7,
+        ... ).frames[0]
+        >>> export_to_video(output, "output.mp4", fps=16)
+        ```
+"""
+
+
+def basic_clean(text):
+    text = ftfy.fix_text(text)
+    text = html.unescape(html.unescape(text))
+    return text.strip()
+
+
+def whitespace_clean(text):
+    text = re.sub(r"\s+", " ", text)
+    text = text.strip()
+    return text
+
+
+def prompt_clean(text):
+    text = whitespace_clean(basic_clean(text))
+    return text
+
+
+# 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,
+    sigmas: Optional[List[float]] = None,
+    **kwargs,
+):
+    r"""
+    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 override the timestep spacing strategy of the scheduler. If `timesteps` is passed,
+            `num_inference_steps` and `sigmas` must be `None`.
+        sigmas (`List[float]`, *optional*):
+            Custom sigmas used to override the timestep spacing strategy of the scheduler. If `sigmas` is passed,
+            `num_inference_steps` and `timesteps` 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 and sigmas is not None:
+        raise ValueError("Only one of `timesteps` or `sigmas` can be passed. Please choose one to set custom values")
+    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)
+    elif sigmas is not None:
+        accept_sigmas = "sigmas" in set(inspect.signature(scheduler.set_timesteps).parameters.keys())
+        if not accept_sigmas:
+            raise ValueError(
+                f"The current scheduler class {scheduler.__class__}'s `set_timesteps` does not support custom"
+                f" sigmas schedules. Please check whether you are using the correct scheduler."
+            )
+        scheduler.set_timesteps(sigmas=sigmas, 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
+
+
+# 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")
+
+
+class WanVideoToVideoPipeline(DiffusionPipeline, WanLoraLoaderMixin):
+    r"""
+    Pipeline for video-to-video generation using Wan.
+
+    This model inherits from [`DiffusionPipeline`]. Check the superclass documentation for the generic methods
+    implemented for all pipelines (downloading, saving, running on a particular device, etc.).
+
+    Args:
+        tokenizer ([`T5Tokenizer`]):
+            Tokenizer from [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5Tokenizer),
+            specifically the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
+        text_encoder ([`T5EncoderModel`]):
+            [T5](https://huggingface.co/docs/transformers/en/model_doc/t5#transformers.T5EncoderModel), specifically
+            the [google/umt5-xxl](https://huggingface.co/google/umt5-xxl) variant.
+        transformer ([`WanTransformer3DModel`]):
+            Conditional Transformer to denoise the input latents.
+        scheduler ([`UniPCMultistepScheduler`]):
+            A scheduler to be used in combination with `transformer` to denoise the encoded image latents.
+        vae ([`AutoencoderKLWan`]):
+            Variational Auto-Encoder (VAE) Model to encode and decode videos to and from latent representations.
+    """
+
+    model_cpu_offload_seq = "text_encoder->transformer->vae"
+    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]
+
+    def __init__(
+        self,
+        tokenizer: AutoTokenizer,
+        text_encoder: UMT5EncoderModel,
+        transformer: WanTransformer3DModel,
+        vae: AutoencoderKLWan,
+        scheduler: FlowMatchEulerDiscreteScheduler,
+    ):
+        super().__init__()
+
+        self.register_modules(
+            vae=vae,
+            text_encoder=text_encoder,
+            tokenizer=tokenizer,
+            transformer=transformer,
+            scheduler=scheduler,
+        )
+
+        self.vae_scale_factor_temporal = 2 ** sum(self.vae.temperal_downsample) if getattr(self, "vae", None) else 4
+        self.vae_scale_factor_spatial = 2 ** len(self.vae.temperal_downsample) if getattr(self, "vae", None) else 8
+        self.video_processor = VideoProcessor(vae_scale_factor=self.vae_scale_factor_spatial)
+
+    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline._get_t5_prompt_embeds
+    def _get_t5_prompt_embeds(
+        self,
+        prompt: Union[str, List[str]] = None,
+        num_videos_per_prompt: int = 1,
+        max_sequence_length: int = 226,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ):
+        device = device or self._execution_device
+        dtype = dtype or self.text_encoder.dtype
+
+        prompt = [prompt] if isinstance(prompt, str) else prompt
+        prompt = [prompt_clean(u) for u in prompt]
+        batch_size = len(prompt)
+
+        text_inputs = self.tokenizer(
+            prompt,
+            padding="max_length",
+            max_length=max_sequence_length,
+            truncation=True,
+            add_special_tokens=True,
+            return_attention_mask=True,
+            return_tensors="pt",
+        )
+        text_input_ids, mask = text_inputs.input_ids, text_inputs.attention_mask
+        seq_lens = mask.gt(0).sum(dim=1).long()
+
+        prompt_embeds = self.text_encoder(text_input_ids.to(device), mask.to(device)).last_hidden_state
+        prompt_embeds = prompt_embeds.to(dtype=dtype, device=device)
+        prompt_embeds = [u[:v] for u, v in zip(prompt_embeds, seq_lens)]
+        prompt_embeds = torch.stack(
+            [torch.cat([u, u.new_zeros(max_sequence_length - u.size(0), u.size(1))]) for u in prompt_embeds], dim=0
+        )
+
+        # duplicate text embeddings for each generation per prompt, using mps friendly method
+        _, seq_len, _ = prompt_embeds.shape
+        prompt_embeds = prompt_embeds.repeat(1, num_videos_per_prompt, 1)
+        prompt_embeds = prompt_embeds.view(batch_size * num_videos_per_prompt, seq_len, -1)
+
+        return prompt_embeds
+
+    # Copied from diffusers.pipelines.wan.pipeline_wan.WanPipeline.encode_prompt
+    def encode_prompt(
+        self,
+        prompt: Union[str, List[str]],
+        negative_prompt: Optional[Union[str, List[str]]] = None,
+        do_classifier_free_guidance: bool = True,
+        num_videos_per_prompt: int = 1,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        max_sequence_length: int = 226,
+        device: Optional[torch.device] = None,
+        dtype: Optional[torch.dtype] = None,
+    ):
+        r"""
+        Encodes the prompt into text encoder hidden states.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                prompt to be encoded
+            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`).
+            do_classifier_free_guidance (`bool`, *optional*, defaults to `True`):
+                Whether to use classifier free guidance or not.
+            num_videos_per_prompt (`int`, *optional*, defaults to 1):
+                Number of videos that should be generated per prompt. torch device to place the resulting embeddings on
+            prompt_embeds (`torch.Tensor`, *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.Tensor`, *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.
+            device: (`torch.device`, *optional*):
+                torch device
+            dtype: (`torch.dtype`, *optional*):
+                torch dtype
+        """
+        device = device or self._execution_device
+
+        prompt = [prompt] if isinstance(prompt, str) else prompt
+        if prompt is not None:
+            batch_size = len(prompt)
+        else:
+            batch_size = prompt_embeds.shape[0]
+
+        if prompt_embeds is None:
+            prompt_embeds = self._get_t5_prompt_embeds(
+                prompt=prompt,
+                num_videos_per_prompt=num_videos_per_prompt,
+                max_sequence_length=max_sequence_length,
+                device=device,
+                dtype=dtype,
+            )
+
+        if do_classifier_free_guidance and negative_prompt_embeds is None:
+            negative_prompt = negative_prompt or ""
+            negative_prompt = batch_size * [negative_prompt] if isinstance(negative_prompt, str) else negative_prompt
+
+            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`."
+                )
+
+            negative_prompt_embeds = self._get_t5_prompt_embeds(
+                prompt=negative_prompt,
+                num_videos_per_prompt=num_videos_per_prompt,
+                max_sequence_length=max_sequence_length,
+                device=device,
+                dtype=dtype,
+            )
+
+        return prompt_embeds, negative_prompt_embeds
+
+    def check_inputs(
+        self,
+        prompt,
+        negative_prompt,
+        height,
+        width,
+        video=None,
+        latents=None,
+        prompt_embeds=None,
+        negative_prompt_embeds=None,
+        callback_on_step_end_tensor_inputs=None,
+    ):
+        if height % 16 != 0 or width % 16 != 0:
+            raise ValueError(f"`height` and `width` have to be divisible by 16 but are {height} and {width}.")
+
+        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 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`: {negative_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 negative_prompt is not None and (
+            not isinstance(negative_prompt, str) and not isinstance(negative_prompt, list)
+        ):
+            raise ValueError(f"`negative_prompt` has to be of type `str` or `list` but is {type(negative_prompt)}")
+
+        if video is not None and latents is not None:
+            raise ValueError("Only one of `video` or `latents` should be provided")
+
+    def prepare_latents(
+        self,
+        video: Optional[torch.Tensor] = None,
+        batch_size: int = 1,
+        num_channels_latents: int = 16,
+        height: int = 480,
+        width: int = 832,
+        dtype: Optional[torch.dtype] = None,
+        device: Optional[torch.device] = None,
+        generator: Optional[torch.Generator] = None,
+        latents: Optional[torch.Tensor] = None,
+        timestep: Optional[torch.Tensor] = None,
+    ):
+        if isinstance(generator, list) and len(generator) != batch_size:
+            raise ValueError(
+                f"You have passed a list of generators of length {len(generator)}, but requested an effective batch"
+                f" size of {batch_size}. Make sure the batch size matches the length of the generators."
+            )
+
+        num_latent_frames = (
+            (video.size(2) - 1) // self.vae_scale_factor_temporal + 1 if latents is None else latents.size(1)
+        )
+        shape = (
+            batch_size,
+            num_channels_latents,
+            num_latent_frames,
+            height // self.vae_scale_factor_spatial,
+            width // self.vae_scale_factor_spatial,
+        )
+
+        if latents is None:
+            if isinstance(generator, list):
+                init_latents = [
+                    retrieve_latents(self.vae.encode(video[i].unsqueeze(0)), generator[i]) for i in range(batch_size)
+                ]
+            else:
+                init_latents = [retrieve_latents(self.vae.encode(vid.unsqueeze(0)), generator) for vid in video]
+
+            init_latents = torch.cat(init_latents, dim=0).to(dtype)
+
+            latents_mean = (
+                torch.tensor(self.vae.config.latents_mean).view(1, self.vae.config.z_dim, 1, 1, 1).to(device, dtype)
+            )
+            latents_std = 1.0 / torch.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
+                device, dtype
+            )
+
+            init_latents = (init_latents - latents_mean) * latents_std
+
+            noise = randn_tensor(shape, generator=generator, device=device, dtype=dtype)
+            if hasattr(self.scheduler, "add_noise"):
+                latents = self.scheduler.add_noise(init_latents, noise, timestep)
+            else:
+                latents = self.scheduelr.scale_noise(init_latents, timestep, noise)
+        else:
+            latents = latents.to(device)
+
+        return latents
+
+    # Copied from diffusers.pipelines.animatediff.pipeline_animatediff_video2video.AnimateDiffVideoToVideoPipeline.get_timesteps
+    def get_timesteps(self, num_inference_steps, timesteps, strength, device):
+        # get the original timestep using init_timestep
+        init_timestep = min(int(num_inference_steps * strength), num_inference_steps)
+
+        t_start = max(num_inference_steps - init_timestep, 0)
+        timesteps = timesteps[t_start * self.scheduler.order :]
+
+        return timesteps, num_inference_steps - t_start
+
+    @property
+    def guidance_scale(self):
+        return self._guidance_scale
+
+    @property
+    def do_classifier_free_guidance(self):
+        return self._guidance_scale > 1.0
+
+    @property
+    def num_timesteps(self):
+        return self._num_timesteps
+
+    @property
+    def current_timestep(self):
+        return self._current_timestep
+
+    @property
+    def interrupt(self):
+        return self._interrupt
+
+    @property
+    def attention_kwargs(self):
+        return self._attention_kwargs
+
+    @torch.no_grad()
+    @replace_example_docstring(EXAMPLE_DOC_STRING)
+    def __call__(
+        self,
+        video: List[Image.Image] = None,
+        prompt: Union[str, List[str]] = None,
+        negative_prompt: Union[str, List[str]] = None,
+        height: int = 480,
+        width: int = 832,
+        num_inference_steps: int = 50,
+        timesteps: Optional[List[int]] = None,
+        guidance_scale: float = 5.0,
+        strength: float = 0.8,
+        num_videos_per_prompt: Optional[int] = 1,
+        generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        output_type: Optional[str] = "np",
+        return_dict: bool = True,
+        attention_kwargs: Optional[Dict[str, Any]] = None,
+        callback_on_step_end: Optional[
+            Union[Callable[[int, int, Dict], None], PipelineCallback, MultiPipelineCallbacks]
+        ] = None,
+        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
+        max_sequence_length: int = 512,
+    ):
+        r"""
+        The call function to the pipeline for generation.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
+                instead.
+            height (`int`, defaults to `480`):
+                The height in pixels of the generated image.
+            width (`int`, defaults to `832`):
+                The width in pixels of the generated image.
+            num_frames (`int`, defaults to `81`):
+                The number of frames in the generated video.
+            num_inference_steps (`int`, defaults to `50`):
+                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
+                expense of slower inference.
+            guidance_scale (`float`, 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
+                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
+                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
+                usually at the expense of lower image quality.
+            num_videos_per_prompt (`int`, *optional*, defaults to 1):
+                The number of images to generate per prompt.
+            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
+                A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
+                generation deterministic.
+            latents (`torch.Tensor`, *optional*):
+                Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
+                generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
+                tensor is generated by sampling using the supplied random `generator`.
+            prompt_embeds (`torch.Tensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
+                provided, text embeddings are generated from the `prompt` input argument.
+            output_type (`str`, *optional*, defaults to `"pil"`):
+                The output format of the generated image. Choose between `PIL.Image` or `np.array`.
+            return_dict (`bool`, *optional*, defaults to `True`):
+                Whether or not to return a [`WanPipelineOutput`] instead of a plain tuple.
+            attention_kwargs (`dict`, *optional*):
+                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
+                `self.processor` in
+                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
+            callback_on_step_end (`Callable`, `PipelineCallback`, `MultiPipelineCallbacks`, *optional*):
+                A function or a subclass of `PipelineCallback` or `MultiPipelineCallbacks` that is called at the end of
+                each denoising step during the inference. with the following arguments: `callback_on_step_end(self:
+                DiffusionPipeline, step: int, timestep: int, callback_kwargs: Dict)`. `callback_kwargs` will include a
+                list of all tensors as specified by `callback_on_step_end_tensor_inputs`.
+            callback_on_step_end_tensor_inputs (`List`, *optional*):
+                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
+                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
+                `._callback_tensor_inputs` attribute of your pipeline class.
+            autocast_dtype (`torch.dtype`, *optional*, defaults to `torch.bfloat16`):
+                The dtype to use for the torch.amp.autocast.
+
+        Examples:
+
+        Returns:
+            [`~WanPipelineOutput`] or `tuple`:
+                If `return_dict` is `True`, [`WanPipelineOutput`] is returned, otherwise a `tuple` is returned where
+                the first element is a list with the generated images and the second element is a list of `bool`s
+                indicating whether the corresponding generated image contains "not-safe-for-work" (nsfw) content.
+        """
+
+        if isinstance(callback_on_step_end, (PipelineCallback, MultiPipelineCallbacks)):
+            callback_on_step_end_tensor_inputs = callback_on_step_end.tensor_inputs
+
+        height = height or self.transformer.config.sample_height * self.vae_scale_factor_spatial
+        width = width or self.transformer.config.sample_width * self.vae_scale_factor_spatial
+        num_videos_per_prompt = 1
+
+        # 1. Check inputs. Raise error if not correct
+        self.check_inputs(
+            prompt,
+            negative_prompt,
+            height,
+            width,
+            video,
+            latents,
+            prompt_embeds,
+            negative_prompt_embeds,
+            callback_on_step_end_tensor_inputs,
+        )
+
+        self._guidance_scale = guidance_scale
+        self._attention_kwargs = attention_kwargs
+        self._current_timestep = None
+        self._interrupt = False
+
+        device = self._execution_device
+
+        # 2. Define call parameters
+        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]
+
+        # 3. Encode input prompt
+        prompt_embeds, negative_prompt_embeds = self.encode_prompt(
+            prompt=prompt,
+            negative_prompt=negative_prompt,
+            do_classifier_free_guidance=self.do_classifier_free_guidance,
+            num_videos_per_prompt=num_videos_per_prompt,
+            prompt_embeds=prompt_embeds,
+            negative_prompt_embeds=negative_prompt_embeds,
+            max_sequence_length=max_sequence_length,
+            device=device,
+        )
+
+        transformer_dtype = self.transformer.dtype
+        prompt_embeds = prompt_embeds.to(transformer_dtype)
+        if negative_prompt_embeds is not None:
+            negative_prompt_embeds = negative_prompt_embeds.to(transformer_dtype)
+
+        # 4. Prepare timesteps
+        timesteps, num_inference_steps = retrieve_timesteps(self.scheduler, num_inference_steps, device, timesteps)
+        timesteps, num_inference_steps = self.get_timesteps(num_inference_steps, timesteps, strength, device)
+        latent_timestep = timesteps[:1].repeat(batch_size * num_videos_per_prompt)
+        self._num_timesteps = len(timesteps)
+
+        if latents is None:
+            video = self.video_processor.preprocess_video(video, height=height, width=width).to(
+                device, dtype=torch.float32
+            )
+
+        # 5. Prepare latent variables
+        num_channels_latents = self.transformer.config.in_channels
+        latents = self.prepare_latents(
+            video,
+            batch_size * num_videos_per_prompt,
+            num_channels_latents,
+            height,
+            width,
+            torch.float32,
+            device,
+            generator,
+            latents,
+            latent_timestep,
+        )
+
+        # 6. Denoising loop
+        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
+        self._num_timesteps = len(timesteps)
+
+        with self.progress_bar(total=num_inference_steps) as progress_bar:
+            for i, t in enumerate(timesteps):
+                if self.interrupt:
+                    continue
+
+                self._current_timestep = t
+                latent_model_input = latents.to(transformer_dtype)
+                timestep = t.expand(latents.shape[0])
+
+                noise_pred = self.transformer(
+                    hidden_states=latent_model_input,
+                    timestep=timestep,
+                    encoder_hidden_states=prompt_embeds,
+                    attention_kwargs=attention_kwargs,
+                    return_dict=False,
+                )[0]
+
+                if self.do_classifier_free_guidance:
+                    noise_uncond = self.transformer(
+                        hidden_states=latent_model_input,
+                        timestep=timestep,
+                        encoder_hidden_states=negative_prompt_embeds,
+                        attention_kwargs=attention_kwargs,
+                        return_dict=False,
+                    )[0]
+                    noise_pred = noise_uncond + guidance_scale * (noise_pred - noise_uncond)
+
+                # compute the previous noisy sample x_t -> x_t-1
+                latents = self.scheduler.step(noise_pred, t, latents, return_dict=False)[0]
+
+                if callback_on_step_end is not None:
+                    callback_kwargs = {}
+                    for k in callback_on_step_end_tensor_inputs:
+                        callback_kwargs[k] = locals()[k]
+                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
+
+                    latents = callback_outputs.pop("latents", latents)
+                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
+                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)
+
+                # 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()
+
+                if XLA_AVAILABLE:
+                    xm.mark_step()
+
+        self._current_timestep = None
+
+        if not output_type == "latent":
+            latents = latents.to(self.vae.dtype)
+            latents_mean = (
+                torch.tensor(self.vae.config.latents_mean)
+                .view(1, self.vae.config.z_dim, 1, 1, 1)
+                .to(latents.device, latents.dtype)
+            )
+            latents_std = 1.0 / torch.tensor(self.vae.config.latents_std).view(1, self.vae.config.z_dim, 1, 1, 1).to(
+                latents.device, latents.dtype
+            )
+            latents = latents / latents_std + latents_mean
+            video = self.vae.decode(latents, return_dict=False)[0]
+            video = self.video_processor.postprocess_video(video, output_type=output_type)
+        else:
+            video = latents
+
+        # Offload all models
+        self.maybe_free_model_hooks()
+
+        if not return_dict:
+            return (video,)
+
+        return WanPipelineOutput(frames=video)

src/diffusers/schedulers/scheduling_utils.py

@@ -19,6 +19,7 @@ from typing import Optional, Union
 
 import torch
 from huggingface_hub.utils import validate_hf_hub_args
+from typing_extensions import Self
 
 from ..utils import BaseOutput, PushToHubMixin
 
@@ -99,7 +100,7 @@ class SchedulerMixin(PushToHubMixin):
         subfolder: Optional[str] = None,
         return_unused_kwargs=False,
         **kwargs,
-    ):
+    ) -> Self:
         r"""
         Instantiate a scheduler from a pre-defined JSON configuration file in a local directory or Hub repository.
 

src/diffusers/utils/dummy_torch_and_transformers_objects.py

@@ -2762,6 +2762,21 @@ class WanPipeline(metaclass=DummyObject):
         requires_backends(cls, ["torch", "transformers"])
 
 
+class WanVideoToVideoPipeline(metaclass=DummyObject):
+    _backends = ["torch", "transformers"]
+
+    def __init__(self, *args, **kwargs):
+        requires_backends(self, ["torch", "transformers"])
+
+    @classmethod
+    def from_config(cls, *args, **kwargs):
+        requires_backends(cls, ["torch", "transformers"])
+
+    @classmethod
+    def from_pretrained(cls, *args, **kwargs):
+        requires_backends(cls, ["torch", "transformers"])
+
+
 class WuerstchenCombinedPipeline(metaclass=DummyObject):
     _backends = ["torch", "transformers"]
 

src/diffusers/utils/import_utils.py

@@ -109,6 +109,7 @@ if _onnx_available:
         "onnxruntime-rocm",
         "onnxruntime-migraphx",
         "onnxruntime-training",
+        "onnxruntime-vitisai",
     )
     _onnxruntime_version = None
     # For the metadata, we have to look for both onnxruntime and onnxruntime-gpu

src/diffusers/utils/testing_utils.py

@@ -1161,7 +1161,7 @@ if is_torch_available():
     }
     BACKEND_RESET_MAX_MEMORY_ALLOCATED = {
         "cuda": torch.cuda.reset_max_memory_allocated,
-        "xpu": None,
+        "xpu": getattr(torch.xpu, "reset_peak_memory_stats", None),
         "cpu": None,
         "mps": None,
         "default": None,

tests/pipelines/controlnet_hunyuandit/test_controlnet_hunyuandit.py

@@ -153,9 +153,14 @@ class HunyuanDiTControlNetPipelineFastTests(unittest.TestCase, PipelineTesterMix
         image_slice = image[0, -3:, -3:, -1]
         assert image.shape == (1, 16, 16, 3)
 
-        expected_slice = np.array(
-            [0.6953125, 0.89208984, 0.59375, 0.5078125, 0.5786133, 0.6035156, 0.5839844, 0.53564453, 0.52246094]
-        )
+        if torch_device == "xpu":
+            expected_slice = np.array(
+                [0.6376953, 0.84375, 0.58691406, 0.48046875, 0.43652344, 0.5517578, 0.54248047, 0.5644531, 0.48217773]
+            )
+        else:
+            expected_slice = np.array(
+                [0.6953125, 0.89208984, 0.59375, 0.5078125, 0.5786133, 0.6035156, 0.5839844, 0.53564453, 0.52246094]
+            )
 
         assert (
             np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
@@ -351,6 +356,7 @@ class HunyuanDiTControlNetPipelineSlowTests(unittest.TestCase):
         assert image.shape == (1024, 1024, 3)
 
         original_image = image[-3:, -3:, -1].flatten()
+
         expected_image = np.array(
             [0.43652344, 0.44018555, 0.4494629, 0.44995117, 0.45654297, 0.44848633, 0.43603516, 0.4404297, 0.42626953]
         )

tests/pipelines/test_pipelines_common.py

@@ -2283,6 +2283,29 @@ class PipelineTesterMixin:
         self.assertTrue(np.allclose(output_without_group_offloading, output_with_group_offloading1, atol=1e-4))
         self.assertTrue(np.allclose(output_without_group_offloading, output_with_group_offloading2, atol=1e-4))
 
+    def test_torch_dtype_dict(self):
+        components = self.get_dummy_components()
+        if not components:
+            self.skipTest("No dummy components defined.")
+
+        pipe = self.pipeline_class(**components)
+
+        specified_key = next(iter(components.keys()))
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            torch_dtype_dict = {specified_key: torch.bfloat16, "default": torch.float16}
+            loaded_pipe = self.pipeline_class.from_pretrained(tmpdirname, torch_dtype=torch_dtype_dict)
+
+        for name, component in loaded_pipe.components.items():
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                expected_dtype = torch_dtype_dict.get(name, torch_dtype_dict.get("default", torch.float32))
+                self.assertEqual(
+                    component.dtype,
+                    expected_dtype,
+                    f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
+                )
+
 
 @is_staging_test
 class PipelinePushToHubTester(unittest.TestCase):

tests/pipelines/wan/test_wan_video_to_video.py

@@ -0,0 +1,146 @@
+# Copyright 2025 The HuggingFace Team.
+#
+# 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 PIL import Image
+from transformers import AutoTokenizer, T5EncoderModel
+
+from diffusers import AutoencoderKLWan, UniPCMultistepScheduler, WanTransformer3DModel, WanVideoToVideoPipeline
+from diffusers.utils.testing_utils import (
+    enable_full_determinism,
+)
+
+from ..pipeline_params import TEXT_TO_IMAGE_IMAGE_PARAMS, TEXT_TO_IMAGE_PARAMS
+from ..test_pipelines_common import (
+    PipelineTesterMixin,
+)
+
+
+enable_full_determinism()
+
+
+class WanVideoToVideoPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
+    pipeline_class = WanVideoToVideoPipeline
+    params = TEXT_TO_IMAGE_PARAMS - {"cross_attention_kwargs"}
+    batch_params = frozenset(["video", "prompt", "negative_prompt"])
+    image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    required_optional_params = frozenset(
+        [
+            "num_inference_steps",
+            "generator",
+            "latents",
+            "return_dict",
+            "callback_on_step_end",
+            "callback_on_step_end_tensor_inputs",
+        ]
+    )
+    test_xformers_attention = False
+    supports_dduf = False
+
+    def get_dummy_components(self):
+        torch.manual_seed(0)
+        vae = AutoencoderKLWan(
+            base_dim=3,
+            z_dim=16,
+            dim_mult=[1, 1, 1, 1],
+            num_res_blocks=1,
+            temperal_downsample=[False, True, True],
+        )
+
+        torch.manual_seed(0)
+        scheduler = UniPCMultistepScheduler(flow_shift=3.0)
+        text_encoder = T5EncoderModel.from_pretrained("hf-internal-testing/tiny-random-t5")
+        tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-t5")
+
+        torch.manual_seed(0)
+        transformer = WanTransformer3DModel(
+            patch_size=(1, 2, 2),
+            num_attention_heads=2,
+            attention_head_dim=12,
+            in_channels=16,
+            out_channels=16,
+            text_dim=32,
+            freq_dim=256,
+            ffn_dim=32,
+            num_layers=2,
+            cross_attn_norm=True,
+            qk_norm="rms_norm_across_heads",
+            rope_max_seq_len=32,
+        )
+
+        components = {
+            "transformer": transformer,
+            "vae": vae,
+            "scheduler": scheduler,
+            "text_encoder": text_encoder,
+            "tokenizer": tokenizer,
+        }
+        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)
+
+        video = [Image.new("RGB", (16, 16))] * 17
+        inputs = {
+            "video": video,
+            "prompt": "dance monkey",
+            "negative_prompt": "negative",  # TODO
+            "generator": generator,
+            "num_inference_steps": 4,
+            "guidance_scale": 6.0,
+            "height": 16,
+            "width": 16,
+            "max_sequence_length": 16,
+            "output_type": "pt",
+        }
+        return inputs
+
+    def test_inference(self):
+        device = "cpu"
+
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+        pipe.to(device)
+        pipe.set_progress_bar_config(disable=None)
+
+        inputs = self.get_dummy_inputs(device)
+        video = pipe(**inputs).frames
+        generated_video = video[0]
+
+        self.assertEqual(generated_video.shape, (17, 3, 16, 16))
+        expected_video = torch.randn(17, 3, 16, 16)
+        max_diff = np.abs(generated_video - expected_video).max()
+        self.assertLessEqual(max_diff, 1e10)
+
+    @unittest.skip("Test not supported")
+    def test_attention_slicing_forward_pass(self):
+        pass
+
+    @unittest.skip(
+        "WanVideoToVideoPipeline has to run in mixed precision. Casting the entire pipeline will result in errors"
+    )
+    def test_float16_inference(self):
+        pass
+
+    @unittest.skip(
+        "WanVideoToVideoPipeline has to run in mixed precision. Save/Load the entire pipeline in FP16 will result in errors"
+    )
+    def test_save_load_float16(self):
+        pass

tests/quantization/bnb/test_mixed_int8.py

@@ -221,7 +221,7 @@ class BnB8bitBasicTests(Base8bitTests):
                     self.assertTrue(module.weight.dtype == torch.int8)
 
         # test if inference works.
-        with torch.no_grad() and torch.amp.autocast("cuda", dtype=torch.float16):
+        with torch.no_grad() and torch.autocast(model.device.type, dtype=torch.float16):
             input_dict_for_transformer = self.get_dummy_inputs()
             model_inputs = {
                 k: v.to(device=torch_device) for k, v in input_dict_for_transformer.items() if not isinstance(v, bool)
@@ -315,7 +315,7 @@ class BnB8bitBasicTests(Base8bitTests):
         _ = self.model_fp16.float()
 
         # Check that this does not throw an error
-        _ = self.model_fp16.cuda()
+        _ = self.model_fp16.to(torch_device)
 
 
 class Bnb8bitDeviceTests(Base8bitTests):