Merge branch 'main' into repr-quant-config

* [training] add ds support to lora hidream

* Apply style fixes

---------

Co-authored-by: J石页 <jiangshuo9@h-partners.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

.github/workflows/benchmark.yml

@@ -23,7 +23,7 @@ jobs:
     runs-on:
       group: aws-g6-4xlarge-plus
     container:
-      image: diffusers/diffusers-pytorch-compile-cuda
+      image: diffusers/diffusers-pytorch-cuda
       options: --shm-size "16gb" --ipc host --gpus 0
     steps:
       - name: Checkout diffusers

.github/workflows/build_docker_images.yml

@@ -38,9 +38,16 @@ jobs:
           token: ${{ secrets.GITHUB_TOKEN }}
 
       - name: Build Changed Docker Images
+        env: 
+          CHANGED_FILES: ${{ steps.file_changes.outputs.all }}
         run: |
-          CHANGED_FILES="${{ steps.file_changes.outputs.all }}"
-          for FILE in $CHANGED_FILES; do
+          echo "$CHANGED_FILES"
+          for FILE in $CHANGED_FILES; do 
+            # skip anything that isn't still on disk
+            if [[ ! -f "$FILE" ]]; then
+              echo "Skipping removed file $FILE"
+              continue
+            fi           
             if [[ "$FILE" == docker/*Dockerfile ]]; then
               DOCKER_PATH="${FILE%/Dockerfile}"
               DOCKER_TAG=$(basename "$DOCKER_PATH")
@@ -65,7 +72,7 @@ jobs:
         image-name:
           - diffusers-pytorch-cpu
           - diffusers-pytorch-cuda
-          - diffusers-pytorch-compile-cuda
+          - diffusers-pytorch-cuda
           - diffusers-pytorch-xformers-cuda
           - diffusers-pytorch-minimum-cuda
           - diffusers-flax-cpu

.github/workflows/nightly_tests.yml

@@ -13,8 +13,9 @@ env:
   PYTEST_TIMEOUT: 600
   RUN_SLOW: yes
   RUN_NIGHTLY: yes
-  PIPELINE_USAGE_CUTOFF: 5000
+  PIPELINE_USAGE_CUTOFF: 0
   SLACK_API_TOKEN: ${{ secrets.SLACK_CIFEEDBACK_BOT_TOKEN }}
+  CONSOLIDATED_REPORT_PATH: consolidated_test_report.md
 
 jobs:
   setup_torch_cuda_pipeline_matrix:
@@ -99,11 +100,6 @@ jobs:
         with:
           name: pipeline_${{ matrix.module }}_test_reports
           path: reports
-      - name: Generate Report and Notify Channel
-        if: always()
-        run: |
-          pip install slack_sdk tabulate
-          python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
 
   run_nightly_tests_for_other_torch_modules:
     name: Nightly Torch CUDA Tests
@@ -174,12 +170,6 @@ jobs:
         name: torch_${{ matrix.module }}_cuda_test_reports
         path: reports
 
-    - name: Generate Report and Notify Channel
-      if: always()
-      run: |
-        pip install slack_sdk tabulate
-        python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
-
   run_torch_compile_tests:
     name: PyTorch Compile CUDA tests
 
@@ -187,7 +177,7 @@ jobs:
       group: aws-g4dn-2xlarge
 
     container:
-      image: diffusers/diffusers-pytorch-compile-cuda
+      image: diffusers/diffusers-pytorch-cuda
       options: --gpus 0 --shm-size "16gb" --ipc host
 
     steps:
@@ -223,12 +213,6 @@ jobs:
         name: torch_compile_test_reports
         path: reports
 
-    - name: Generate Report and Notify Channel
-      if: always()
-      run: |
-        pip install slack_sdk tabulate
-        python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
-  
   run_big_gpu_torch_tests:
     name: Torch tests on big GPU
     strategy:
@@ -279,12 +263,7 @@ jobs:
         with:
           name: torch_cuda_big_gpu_test_reports
           path: reports
-      - name: Generate Report and Notify Channel
-        if: always()
-        run: |
-          pip install slack_sdk tabulate
-          python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
-          
+
   torch_minimum_version_cuda_tests:
     name: Torch Minimum Version CUDA Tests
     runs-on:
@@ -341,63 +320,6 @@ jobs:
         with:
           name: torch_minimum_version_cuda_test_reports
           path: reports
- 
-  run_flax_tpu_tests:
-    name: Nightly Flax TPU Tests
-    runs-on:
-      group: gcp-ct5lp-hightpu-8t
-    if: github.event_name == 'schedule'
-
-    container:
-      image: diffusers/diffusers-flax-tpu
-      options: --shm-size "16gb" --ipc host --privileged ${{ vars.V5_LITEPOD_8_ENV}} -v /mnt/hf_cache:/mnt/hf_cache
-    defaults:
-      run:
-        shell: bash
-    steps:
-    - name: Checkout diffusers
-      uses: actions/checkout@v3
-      with:
-        fetch-depth: 2
-
-    - name: Install dependencies
-      run: |
-        python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
-        python -m uv pip install -e [quality,test]
-        pip uninstall accelerate -y && python -m uv pip install -U accelerate@git+https://github.com/huggingface/accelerate.git
-        python -m uv pip install pytest-reportlog
-
-    - name: Environment
-      run: python utils/print_env.py
-
-    - name: Run nightly Flax TPU tests
-      env:
-        HF_TOKEN: ${{ secrets.DIFFUSERS_HF_HUB_READ_TOKEN }}
-      run: |
-        python -m pytest -n 0 \
-          -s -v -k "Flax" \
-          --make-reports=tests_flax_tpu \
-          --report-log=tests_flax_tpu.log \
-          tests/
-
-    - name: Failure short reports
-      if: ${{ failure() }}
-      run: |
-        cat reports/tests_flax_tpu_stats.txt
-        cat reports/tests_flax_tpu_failures_short.txt
-
-    - name: Test suite reports artifacts
-      if: ${{ always() }}
-      uses: actions/upload-artifact@v4
-      with:
-        name: flax_tpu_test_reports
-        path: reports
-
-    - name: Generate Report and Notify Channel
-      if: always()
-      run: |
-        pip install slack_sdk tabulate
-        python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
 
   run_nightly_onnx_tests:
     name: Nightly ONNXRuntime CUDA tests on Ubuntu
@@ -448,18 +370,12 @@ jobs:
         name: tests_onnx_cuda_reports
         path: reports
 
-    - name: Generate Report and Notify Channel
-      if: always()
-      run: |
-        pip install slack_sdk tabulate
-        python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
-
   run_nightly_quantization_tests:
     name: Torch quantization nightly tests
     strategy:
       fail-fast: false
       max-parallel: 2
-      matrix: 
+      matrix:
         config:
           - backend: "bitsandbytes"
             test_location: "bnb"
@@ -519,11 +435,165 @@ jobs:
         with:
           name: torch_cuda_${{ matrix.config.backend }}_reports
           path: reports
-      - name: Generate Report and Notify Channel
-        if: always()
+          
+  run_nightly_pipeline_level_quantization_tests:
+    name: Torch quantization nightly tests
+    strategy:
+      fail-fast: false
+      max-parallel: 2
+    runs-on:
+      group: aws-g6e-xlarge-plus
+    container:
+      image: diffusers/diffusers-pytorch-cuda
+      options: --shm-size "20gb" --ipc host --gpus 0
+    steps:
+      - name: Checkout diffusers
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+      - name: NVIDIA-SMI
+        run: nvidia-smi
+      - name: Install dependencies
         run: |
+          python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
+          python -m uv pip install -e [quality,test]
+          python -m uv pip install -U bitsandbytes optimum_quanto
+          python -m uv pip install pytest-reportlog
+      - name: Environment
+        run: |
+          python utils/print_env.py
+      - name: Pipeline-level quantization tests on GPU
+        env:
+          HF_TOKEN: ${{ secrets.DIFFUSERS_HF_HUB_READ_TOKEN }}
+          # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
+          CUBLAS_WORKSPACE_CONFIG: :16:8
+          BIG_GPU_MEMORY: 40
+        run: |
+          python -m pytest -n 1 --max-worker-restart=0 --dist=loadfile \
+            --make-reports=tests_pipeline_level_quant_torch_cuda \
+            --report-log=tests_pipeline_level_quant_torch_cuda.log \
+            tests/quantization/test_pipeline_level_quantization.py
+      - name: Failure short reports
+        if: ${{ failure() }}
+        run: |
+          cat reports/tests_pipeline_level_quant_torch_cuda_stats.txt
+          cat reports/tests_pipeline_level_quant_torch_cuda_failures_short.txt
+      - name: Test suite reports artifacts
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: torch_cuda_pipeline_level_quant_reports
+          path: reports
+
+  run_flax_tpu_tests:
+    name: Nightly Flax TPU Tests
+    runs-on:
+      group: gcp-ct5lp-hightpu-8t
+    if: github.event_name == 'schedule'
+
+    container:
+      image: diffusers/diffusers-flax-tpu
+      options: --shm-size "16gb" --ipc host --privileged ${{ vars.V5_LITEPOD_8_ENV}} -v /mnt/hf_cache:/mnt/hf_cache
+    defaults:
+      run:
+        shell: bash
+    steps:
+    - name: Checkout diffusers
+      uses: actions/checkout@v3
+      with:
+        fetch-depth: 2
+
+    - name: Install dependencies
+      run: |
+        python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
+        python -m uv pip install -e [quality,test]
+        pip uninstall accelerate -y && python -m uv pip install -U accelerate@git+https://github.com/huggingface/accelerate.git
+        python -m uv pip install pytest-reportlog
+
+    - name: Environment
+      run: python utils/print_env.py
+
+    - name: Run nightly Flax TPU tests
+      env:
+        HF_TOKEN: ${{ secrets.DIFFUSERS_HF_HUB_READ_TOKEN }}
+      run: |
+        python -m pytest -n 0 \
+          -s -v -k "Flax" \
+          --make-reports=tests_flax_tpu \
+          --report-log=tests_flax_tpu.log \
+          tests/
+
+    - name: Failure short reports
+      if: ${{ failure() }}
+      run: |
+        cat reports/tests_flax_tpu_stats.txt
+        cat reports/tests_flax_tpu_failures_short.txt
+
+    - name: Test suite reports artifacts
+      if: ${{ always() }}
+      uses: actions/upload-artifact@v4
+      with:
+        name: flax_tpu_test_reports
+        path: reports
+
+  generate_consolidated_report:
+    name: Generate Consolidated Test Report
+    needs: [
+      run_nightly_tests_for_torch_pipelines,
+      run_nightly_tests_for_other_torch_modules,
+      run_torch_compile_tests,
+      run_big_gpu_torch_tests,
+      run_nightly_quantization_tests,
+      run_nightly_pipeline_level_quantization_tests,
+      run_nightly_onnx_tests,
+      torch_minimum_version_cuda_tests,
+      run_flax_tpu_tests
+    ]
+    if: always()
+    runs-on:
+      group: aws-general-8-plus
+    container:
+      image: diffusers/diffusers-pytorch-cpu
+    steps:
+      - name: Checkout diffusers
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - name: Create reports directory
+        run: mkdir -p combined_reports
+
+      - name: Download all test reports
+        uses: actions/download-artifact@v4
+        with:
+          path: artifacts
+
+      - name: Prepare reports
+        run: |
+          # Move all report files to a single directory for processing
+          find artifacts -name "*.txt" -exec cp {} combined_reports/ \;
+
+      - name: Install dependencies
+        run: |
+          pip install -e .[test]
           pip install slack_sdk tabulate
-          python utils/log_reports.py >> $GITHUB_STEP_SUMMARY
+
+      - name: Generate consolidated report
+        run: |
+          python utils/consolidated_test_report.py \
+            --reports_dir combined_reports \
+            --output_file $CONSOLIDATED_REPORT_PATH \
+            --slack_channel_name diffusers-ci-nightly
+
+      - name: Show consolidated report
+        run: |
+          cat $CONSOLIDATED_REPORT_PATH >> $GITHUB_STEP_SUMMARY
+
+      - name: Upload consolidated report
+        uses: actions/upload-artifact@v4
+        with:
+          name: consolidated_test_report
+          path: ${{ env.CONSOLIDATED_REPORT_PATH }}
 
 # M1 runner currently not well supported
 # TODO: (Dhruv) add these back when we setup better testing for Apple Silicon

.github/workflows/pr_style_bot.yml

@@ -14,4 +14,4 @@ jobs:
     with:
       python_quality_dependencies: "[quality]"
     secrets:
-      bot_token: ${{ secrets.GITHUB_TOKEN }}
+      bot_token: ${{ secrets.HF_STYLE_BOT_ACTION }}

.github/workflows/pr_tests.yml

@@ -11,6 +11,7 @@ on:
       - "tests/**.py"
       - ".github/**.yml"
       - "utils/**.py"
+      - "setup.py"
   push:
     branches:
       - ci-*
@@ -290,8 +291,8 @@ jobs:
     - name: Failure short reports
       if: ${{ failure() }}
       run: |
-        cat reports/tests_lora_failures_short.txt
-        cat reports/tests_models_lora_failures_short.txt
+        cat reports/tests_peft_main_failures_short.txt
+        cat reports/tests_models_lora_peft_main_failures_short.txt
 
     - name: Test suite reports artifacts
       if: ${{ always() }}

.github/workflows/push_tests.yml

@@ -262,7 +262,7 @@ jobs:
       group: aws-g4dn-2xlarge
 
     container:
-      image: diffusers/diffusers-pytorch-compile-cuda
+      image: diffusers/diffusers-pytorch-cuda
       options: --gpus 0 --shm-size "16gb" --ipc host
 
     steps:

.github/workflows/release_tests_fast.yml

@@ -316,7 +316,7 @@ jobs:
       group: aws-g4dn-2xlarge
 
     container:
-      image: diffusers/diffusers-pytorch-compile-cuda
+      image: diffusers/diffusers-pytorch-cuda
       options: --gpus 0 --shm-size "16gb" --ipc host
 
     steps:

docker/diffusers-pytorch-compile-cuda/Dockerfile

@@ -1,50 +0,0 @@
-FROM nvidia/cuda:12.1.0-runtime-ubuntu20.04
-LABEL maintainer="Hugging Face"
-LABEL repository="diffusers"
-
-ENV DEBIAN_FRONTEND=noninteractive
-
-RUN apt-get -y update \
-    && apt-get install -y software-properties-common \
-    && add-apt-repository ppa:deadsnakes/ppa
-
-RUN apt install -y bash \
-    build-essential \
-    git \
-    git-lfs \
-    curl \
-    ca-certificates \
-    libsndfile1-dev \
-    libgl1 \
-    python3.10 \
-    python3.10-dev \
-    python3-pip \
-    python3.10-venv && \
-    rm -rf /var/lib/apt/lists
-
-# make sure to use venv
-RUN python3.10 -m venv /opt/venv
-ENV PATH="/opt/venv/bin:$PATH"
-
-# pre-install the heavy dependencies (these can later be overridden by the deps from setup.py)
-RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
-    python3.10 -m uv pip install --no-cache-dir \
-    torch \
-    torchvision \
-    torchaudio \
-    invisible_watermark && \
-    python3.10 -m pip install --no-cache-dir \
-    accelerate \
-    datasets \
-    hf-doc-builder \
-    huggingface-hub \
-    hf_transfer \
-    Jinja2 \
-    librosa \
-    numpy==1.26.4 \
-    scipy \
-    tensorboard \
-    transformers \
-    hf_transfer
-
-CMD ["/bin/bash"]

docs/source/en/_toctree.yml

@@ -17,12 +17,6 @@
     title: AutoPipeline
   - local: tutorials/basic_training
     title: Train a diffusion model
-  - local: tutorials/using_peft_for_inference
-    title: Load LoRAs for inference
-  - local: tutorials/fast_diffusion
-    title: Accelerate inference of text-to-image diffusion models
-  - local: tutorials/inference_with_big_models
-    title: Working with big models
   title: Tutorials
 - sections:
   - local: using-diffusers/loading
@@ -33,11 +27,24 @@
     title: Load schedulers and models
   - local: using-diffusers/other-formats
     title: Model files and layouts
-  - local: using-diffusers/loading_adapters
-    title: Load adapters
   - local: using-diffusers/push_to_hub
     title: Push files to the Hub
   title: Load pipelines and adapters
+- sections:
+  - local: tutorials/using_peft_for_inference
+    title: LoRA
+  - local: using-diffusers/ip_adapter
+    title: IP-Adapter
+  - local: using-diffusers/controlnet
+    title: ControlNet
+  - local: using-diffusers/t2i_adapter
+    title: T2I-Adapter
+  - local: using-diffusers/dreambooth
+    title: DreamBooth
+  - local: using-diffusers/textual_inversion_inference
+    title: Textual inversion
+  title: Adapters
+  isExpanded: false
 - sections:
   - local: using-diffusers/unconditional_image_generation
     title: Unconditional image generation
@@ -59,8 +66,6 @@
     title: Create a server
   - local: training/distributed_inference
     title: Distributed inference
-  - local: using-diffusers/merge_loras
-    title: Merge LoRAs
   - local: using-diffusers/scheduler_features
     title: Scheduler features
   - local: using-diffusers/callback
@@ -87,8 +92,6 @@
     title: API Reference
   title: Hybrid Inference
 - sections:
-  - local: using-diffusers/cogvideox
-    title: CogVideoX
   - local: using-diffusers/consisid
     title: ConsisID
   - local: using-diffusers/sdxl
@@ -97,20 +100,12 @@
     title: SDXL Turbo
   - local: using-diffusers/kandinsky
     title: Kandinsky
-  - local: using-diffusers/ip_adapter
-    title: IP-Adapter
   - local: using-diffusers/omnigen
     title: OmniGen
   - local: using-diffusers/pag
     title: PAG
-  - local: using-diffusers/controlnet
-    title: ControlNet
-  - local: using-diffusers/t2i_adapter
-    title: T2I-Adapter
   - local: using-diffusers/inference_with_lcm
     title: Latent Consistency Model
-  - local: using-diffusers/textual_inversion_inference
-    title: Textual inversion
   - local: using-diffusers/shap-e
     title: Shap-E
   - local: using-diffusers/diffedit
@@ -180,11 +175,13 @@
   title: Quantization Methods
 - sections:
   - local: optimization/fp16
-    title: Speed up inference
+    title: Accelerate inference
+  - local: optimization/cache
+    title: Caching
   - local: optimization/memory
     title: Reduce memory usage
-  - local: optimization/torch2.0
-    title: PyTorch 2.0
+  - local: optimization/pruna
+    title: Pruna
   - local: optimization/xformers
     title: xFormers
   - local: optimization/tome
@@ -211,7 +208,7 @@
     - local: optimization/mps
       title: Metal Performance Shaders (MPS)
     - local: optimization/habana
-      title: Habana Gaudi
+      title: Intel Gaudi
     - local: optimization/neuron
       title: AWS Neuron
     title: Optimized hardware
@@ -288,6 +285,8 @@
         title: AllegroTransformer3DModel
       - local: api/models/aura_flow_transformer2d
         title: AuraFlowTransformer2DModel
+      - local: api/models/chroma_transformer
+        title: ChromaTransformer2DModel
       - local: api/models/cogvideox_transformer3d
         title: CogVideoXTransformer3DModel
       - local: api/models/cogview3plus_transformer2d
@@ -296,6 +295,8 @@
         title: CogView4Transformer2DModel
       - local: api/models/consisid_transformer3d
         title: ConsisIDTransformer3DModel
+      - local: api/models/cosmos_transformer3d
+        title: CosmosTransformer3DModel
       - local: api/models/dit_transformer2d
         title: DiTTransformer2DModel
       - local: api/models/easyanimate_transformer3d
@@ -364,6 +365,8 @@
         title: AutoencoderKLAllegro
       - local: api/models/autoencoderkl_cogvideox
         title: AutoencoderKLCogVideoX
+      - local: api/models/autoencoderkl_cosmos
+        title: AutoencoderKLCosmos
       - local: api/models/autoencoder_kl_hunyuan_video
         title: AutoencoderKLHunyuanVideo
       - local: api/models/autoencoderkl_ltx_video
@@ -406,6 +409,8 @@
       title: AutoPipeline
     - local: api/pipelines/blip_diffusion
       title: BLIP-Diffusion
+    - local: api/pipelines/chroma
+      title: Chroma
     - local: api/pipelines/cogvideox
       title: CogVideoX
     - local: api/pipelines/cogview3
@@ -434,6 +439,8 @@
       title: ControlNet-XS with Stable Diffusion XL
     - local: api/pipelines/controlnet_union
       title: ControlNetUnion
+    - local: api/pipelines/cosmos
+      title: Cosmos
     - local: api/pipelines/dance_diffusion
       title: Dance Diffusion
     - local: api/pipelines/ddim
@@ -452,6 +459,8 @@
       title: Flux
     - local: api/pipelines/control_flux_inpaint
       title: FluxControlInpaint
+    - local: api/pipelines/framepack
+      title: Framepack
     - local: api/pipelines/hidream
       title: HiDream-I1
     - local: api/pipelines/hunyuandit
@@ -568,6 +577,8 @@
       title: UniDiffuser
     - local: api/pipelines/value_guided_sampling
       title: Value-guided sampling
+    - local: api/pipelines/visualcloze
+      title: VisualCloze
     - local: api/pipelines/wan
       title: Wan
     - local: api/pipelines/wuerstchen

docs/source/en/api/cache.md

@@ -11,71 +11,19 @@ specific language governing permissions and limitations under the License. -->
 
 # Caching methods
 
-## Pyramid Attention Broadcast
+Cache methods speedup diffusion transformers by storing and reusing intermediate outputs of specific layers, such as attention and feedforward layers, instead of recalculating them at each inference step.
 
-[Pyramid Attention Broadcast](https://huggingface.co/papers/2408.12588) from Xuanlei Zhao, Xiaolong Jin, Kai Wang, Yang You.
-
-Pyramid Attention Broadcast (PAB) is a method that speeds up inference in diffusion models by systematically skipping attention computations between successive inference steps and reusing cached attention states. The attention states are not very different between successive inference steps. The most prominent difference is in the spatial attention blocks, not as much in the temporal attention blocks, and finally the least in the cross attention blocks. Therefore, many cross attention computation blocks can be skipped, followed by the temporal and spatial attention blocks. By combining other techniques like sequence parallelism and classifier-free guidance parallelism, PAB achieves near real-time video generation.
-
-Enable PAB with [`~PyramidAttentionBroadcastConfig`] on any pipeline. For some benchmarks, refer to [this](https://github.com/huggingface/diffusers/pull/9562) pull request.
-
-```python
-import torch
-from diffusers import CogVideoXPipeline, PyramidAttentionBroadcastConfig
-
-pipe = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b", torch_dtype=torch.bfloat16)
-pipe.to("cuda")
-
-# Increasing the value of `spatial_attention_timestep_skip_range[0]` or decreasing the value of
-# `spatial_attention_timestep_skip_range[1]` will decrease the interval in which pyramid attention
-# broadcast is active, leader to slower inference speeds. However, large intervals can lead to
-# poorer quality of generated videos.
-config = PyramidAttentionBroadcastConfig(
-    spatial_attention_block_skip_range=2,
-    spatial_attention_timestep_skip_range=(100, 800),
-    current_timestep_callback=lambda: pipe.current_timestep,
-)
-pipe.transformer.enable_cache(config)
-```
-
-## Faster Cache
-
-[FasterCache](https://huggingface.co/papers/2410.19355) from Zhengyao Lv, Chenyang Si, Junhao Song, Zhenyu Yang, Yu Qiao, Ziwei Liu, Kwan-Yee K. Wong.
-
-FasterCache is a method that speeds up inference in diffusion transformers by:
-- Reusing attention states between successive inference steps, due to high similarity between them
-- Skipping unconditional branch prediction used in classifier-free guidance by revealing redundancies between unconditional and conditional branch outputs for the same timestep, and therefore approximating the unconditional branch output using the conditional branch output
-
-```python
-import torch
-from diffusers import CogVideoXPipeline, FasterCacheConfig
-
-pipe = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b", torch_dtype=torch.bfloat16)
-pipe.to("cuda")
-
-config = FasterCacheConfig(
-    spatial_attention_block_skip_range=2,
-    spatial_attention_timestep_skip_range=(-1, 681),
-    current_timestep_callback=lambda: pipe.current_timestep,
-    attention_weight_callback=lambda _: 0.3,
-    unconditional_batch_skip_range=5,
-    unconditional_batch_timestep_skip_range=(-1, 781),
-    tensor_format="BFCHW",
-)
-pipe.transformer.enable_cache(config)
-```
-
-### CacheMixin
+## CacheMixin
 
 [[autodoc]] CacheMixin
 
-### PyramidAttentionBroadcastConfig
+## PyramidAttentionBroadcastConfig
 
 [[autodoc]] PyramidAttentionBroadcastConfig
 
 [[autodoc]] apply_pyramid_attention_broadcast
 
-### FasterCacheConfig
+## FasterCacheConfig
 
 [[autodoc]] FasterCacheConfig
 

docs/source/en/api/loaders/lora.md

@@ -98,4 +98,8 @@ To learn more about how to load LoRA weights, see the [LoRA](../../using-diffuse
 
 ## LoraBaseMixin
 
-[[autodoc]] loaders.lora_base.LoraBaseMixin
+[[autodoc]] loaders.lora_base.LoraBaseMixin
+
+## WanLoraLoaderMixin
+
+[[autodoc]] loaders.lora_pipeline.WanLoraLoaderMixin

docs/source/en/api/models/asymmetricautoencoderkl.md

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # AsymmetricAutoencoderKL
 
-Improved larger variational autoencoder (VAE) model with KL loss for inpainting task: [Designing a Better Asymmetric VQGAN for StableDiffusion](https://arxiv.org/abs/2306.04632) by Zixin Zhu, Xuelu Feng, Dongdong Chen, Jianmin Bao, Le Wang, Yinpeng Chen, Lu Yuan, Gang Hua.
+Improved larger variational autoencoder (VAE) model with KL loss for inpainting task: [Designing a Better Asymmetric VQGAN for StableDiffusion](https://huggingface.co/papers/2306.04632) by Zixin Zhu, Xuelu Feng, Dongdong Chen, Jianmin Bao, Le Wang, Yinpeng Chen, Lu Yuan, Gang Hua.
 
 The abstract from the paper is:
 

docs/source/en/api/models/autoencoderkl.md

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # AutoencoderKL
 
-The variational autoencoder (VAE) model with KL loss was introduced in [Auto-Encoding Variational Bayes](https://arxiv.org/abs/1312.6114v11) by Diederik P. Kingma and Max Welling. The model is used in 🤗 Diffusers to encode images into latents and to decode latent representations into images.
+The variational autoencoder (VAE) model with KL loss was introduced in [Auto-Encoding Variational Bayes](https://huggingface.co/papers/1312.6114v11) by Diederik P. Kingma and Max Welling. The model is used in 🤗 Diffusers to encode images into latents and to decode latent representations into images.
 
 The abstract from the paper is:
 

docs/source/en/api/models/autoencoderkl_cosmos.md

@@ -0,0 +1,40 @@
+<!-- Copyright 2024 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. -->
+
+# AutoencoderKLCosmos
+
+[Cosmos Tokenizers](https://github.com/NVIDIA/Cosmos-Tokenizer).
+
+Supported models:
+- [nvidia/Cosmos-1.0-Tokenizer-CV8x8x8](https://huggingface.co/nvidia/Cosmos-1.0-Tokenizer-CV8x8x8)
+
+The model can be loaded with the following code snippet.
+
+```python
+from diffusers import AutoencoderKLCosmos
+
+vae = AutoencoderKLCosmos.from_pretrained("nvidia/Cosmos-1.0-Tokenizer-CV8x8x8", subfolder="vae")
+```
+
+## AutoencoderKLCosmos
+
+[[autodoc]] AutoencoderKLCosmos
+    - decode
+    - encode
+    - all
+
+## AutoencoderKLOutput
+
+[[autodoc]] models.autoencoders.autoencoder_kl.AutoencoderKLOutput
+
+## DecoderOutput
+
+[[autodoc]] models.autoencoders.vae.DecoderOutput

docs/source/en/api/models/chroma_transformer.md

@@ -0,0 +1,19 @@
+<!--Copyright 2025 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.
+-->
+
+# ChromaTransformer2DModel
+
+A modified flux Transformer model from [Chroma](https://huggingface.co/lodestones/Chroma)
+
+## ChromaTransformer2DModel
+
+[[autodoc]] ChromaTransformer2DModel

docs/source/en/api/models/consisid_transformer3d.md

@@ -11,7 +11,7 @@ specific language governing permissions and limitations under the License. -->
 
 # ConsisIDTransformer3DModel
 
-A Diffusion Transformer model for 3D data from [ConsisID](https://github.com/PKU-YuanGroup/ConsisID) was introduced in [Identity-Preserving Text-to-Video Generation by Frequency Decomposition](https://arxiv.org/pdf/2411.17440) by Peking University & University of Rochester & etc.
+A Diffusion Transformer model for 3D data from [ConsisID](https://github.com/PKU-YuanGroup/ConsisID) was introduced in [Identity-Preserving Text-to-Video Generation by Frequency Decomposition](https://huggingface.co/papers/2411.17440) by Peking University & University of Rochester & etc.
 
 The model can be loaded with the following code snippet.
 

docs/source/en/api/models/controlnet_hunyuandit.md

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # HunyuanDiT2DControlNetModel
 
-HunyuanDiT2DControlNetModel is an implementation of ControlNet for [Hunyuan-DiT](https://arxiv.org/abs/2405.08748).
+HunyuanDiT2DControlNetModel is an implementation of ControlNet for [Hunyuan-DiT](https://huggingface.co/papers/2405.08748).
 
 ControlNet was introduced in [Adding Conditional Control to Text-to-Image Diffusion Models](https://huggingface.co/papers/2302.05543) by Lvmin Zhang, Anyi Rao, and Maneesh Agrawala.
 

docs/source/en/api/models/controlnet_sparsectrl.md

@@ -11,11 +11,11 @@ specific language governing permissions and limitations under the License. -->
 
 # SparseControlNetModel
 
-SparseControlNetModel is an implementation of ControlNet for [AnimateDiff](https://arxiv.org/abs/2307.04725).
+SparseControlNetModel is an implementation of ControlNet for [AnimateDiff](https://huggingface.co/papers/2307.04725).
 
 ControlNet was introduced in [Adding Conditional Control to Text-to-Image Diffusion Models](https://huggingface.co/papers/2302.05543) by Lvmin Zhang, Anyi Rao, and Maneesh Agrawala.
 
-The SparseCtrl version of ControlNet was introduced in [SparseCtrl: Adding Sparse Controls to Text-to-Video Diffusion Models](https://arxiv.org/abs/2311.16933) for achieving controlled generation in text-to-video diffusion models by Yuwei Guo, Ceyuan Yang, Anyi Rao, Maneesh Agrawala, Dahua Lin, and Bo Dai.
+The SparseCtrl version of ControlNet was introduced in [SparseCtrl: Adding Sparse Controls to Text-to-Video Diffusion Models](https://huggingface.co/papers/2311.16933) for achieving controlled generation in text-to-video diffusion models by Yuwei Guo, Ceyuan Yang, Anyi Rao, Maneesh Agrawala, Dahua Lin, and Bo Dai.
 
 The abstract from the paper is:
 

docs/source/en/api/models/cosmos_transformer3d.md

@@ -0,0 +1,30 @@
+<!-- Copyright 2024 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. -->
+
+# CosmosTransformer3DModel
+
+A Diffusion Transformer model for 3D video-like data was introduced in [Cosmos World Foundation Model Platform for Physical AI](https://huggingface.co/papers/2501.03575) by NVIDIA.
+
+The model can be loaded with the following code snippet.
+
+```python
+from diffusers import CosmosTransformer3DModel
+
+transformer = CosmosTransformer3DModel.from_pretrained("nvidia/Cosmos-1.0-Diffusion-7B-Text2World", subfolder="transformer", torch_dtype=torch.bfloat16)
+```
+
+## CosmosTransformer3DModel
+
+[[autodoc]] CosmosTransformer3DModel
+
+## Transformer2DModelOutput
+
+[[autodoc]] models.modeling_outputs.Transformer2DModelOutput

docs/source/en/api/models/hidream_image_transformer.md

@@ -21,6 +21,22 @@ from diffusers import HiDreamImageTransformer2DModel
 transformer = HiDreamImageTransformer2DModel.from_pretrained("HiDream-ai/HiDream-I1-Full", subfolder="transformer", torch_dtype=torch.bfloat16)
 ```
 
+## Loading GGUF quantized checkpoints for HiDream-I1
+
+GGUF checkpoints for the `HiDreamImageTransformer2DModel` can  be loaded using `~FromOriginalModelMixin.from_single_file`
+
+```python
+import torch
+from diffusers import GGUFQuantizationConfig, HiDreamImageTransformer2DModel
+
+ckpt_path = "https://huggingface.co/city96/HiDream-I1-Dev-gguf/blob/main/hidream-i1-dev-Q2_K.gguf"
+transformer = HiDreamImageTransformer2DModel.from_single_file(
+    ckpt_path,
+    quantization_config=GGUFQuantizationConfig(compute_dtype=torch.bfloat16),
+    torch_dtype=torch.bfloat16
+)
+```
+
 ## HiDreamImageTransformer2DModel
 
 [[autodoc]] HiDreamImageTransformer2DModel

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

@@ -14,7 +14,7 @@ specific language governing permissions and limitations under the License.
 
 aMUSEd was introduced in [aMUSEd: An Open MUSE Reproduction](https://huggingface.co/papers/2401.01808) by Suraj Patil, William Berman, Robin Rombach, and Patrick von Platen.
 
-Amused is a lightweight text to image model based off of the [MUSE](https://arxiv.org/abs/2301.00704) architecture. Amused is particularly useful in applications that require a lightweight and fast model such as generating many images quickly at once.
+Amused is a lightweight text to image model based off of the [MUSE](https://huggingface.co/papers/2301.00704) architecture. Amused is particularly useful in applications that require a lightweight and fast model such as generating many images quickly at once.
 
 Amused is a vqvae token based transformer that can generate an image in fewer forward passes than many diffusion models. In contrast with muse, it uses the smaller text encoder CLIP-L/14 instead of t5-xxl. Due to its small parameter count and few forward pass generation process, amused can generate many images quickly. This benefit is seen particularly at larger batch sizes.
 

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

@@ -18,7 +18,7 @@ specific language governing permissions and limitations under the License.
 
 ## Overview
 
-[AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning](https://arxiv.org/abs/2307.04725) by Yuwei Guo, Ceyuan Yang, Anyi Rao, Yaohui Wang, Yu Qiao, Dahua Lin, Bo Dai.
+[AnimateDiff: Animate Your Personalized Text-to-Image Diffusion Models without Specific Tuning](https://huggingface.co/papers/2307.04725) by Yuwei Guo, Ceyuan Yang, Anyi Rao, Yaohui Wang, Yu Qiao, Dahua Lin, Bo Dai.
 
 The abstract of the paper is the following:
 
@@ -187,7 +187,7 @@ Here are some sample outputs:
 
 ### AnimateDiffSparseControlNetPipeline
 
-[SparseCtrl: Adding Sparse Controls to Text-to-Video Diffusion Models](https://arxiv.org/abs/2311.16933) for achieving controlled generation in text-to-video diffusion models by Yuwei Guo, Ceyuan Yang, Anyi Rao, Maneesh Agrawala, Dahua Lin, and Bo Dai.
+[SparseCtrl: Adding Sparse Controls to Text-to-Video Diffusion Models](https://huggingface.co/papers/2311.16933) for achieving controlled generation in text-to-video diffusion models by Yuwei Guo, Ceyuan Yang, Anyi Rao, Maneesh Agrawala, Dahua Lin, and Bo Dai.
 
 The abstract from the paper is:
 
@@ -751,7 +751,7 @@ export_to_gif(frames, "animation.gif")
 
 ## Using FreeInit
 
-[FreeInit: Bridging Initialization Gap in Video Diffusion Models](https://arxiv.org/abs/2312.07537) by Tianxing Wu, Chenyang Si, Yuming Jiang, Ziqi Huang, Ziwei Liu.
+[FreeInit: Bridging Initialization Gap in Video Diffusion Models](https://huggingface.co/papers/2312.07537) by Tianxing Wu, Chenyang Si, Yuming Jiang, Ziqi Huang, Ziwei Liu.
 
 FreeInit is an effective method that improves temporal consistency and overall quality of videos generated using video-diffusion-models without any addition training. It can be applied to AnimateDiff, ModelScope, VideoCrafter and various other video generation models seamlessly at inference time, and works by iteratively refining the latent-initialization noise. More details can be found it the paper.
 
@@ -920,7 +920,7 @@ export_to_gif(frames, "animatelcm-motion-lora.gif")
 
 ## Using FreeNoise
 
-[FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling](https://arxiv.org/abs/2310.15169) by Haonan Qiu, Menghan Xia, Yong Zhang, Yingqing He, Xintao Wang, Ying Shan, Ziwei Liu.
+[FreeNoise: Tuning-Free Longer Video Diffusion via Noise Rescheduling](https://huggingface.co/papers/2310.15169) by Haonan Qiu, Menghan Xia, Yong Zhang, Yingqing He, Xintao Wang, Ying Shan, Ziwei Liu.
 
 FreeNoise is a sampling mechanism that can generate longer videos with short-video generation models by employing noise-rescheduling, temporal attention over sliding windows, and weighted averaging of latent frames. It also can be used with multiple prompts to allow for interpolated video generations. More details are available in the paper.
 
@@ -966,7 +966,7 @@ pipe.to("cuda")
 prompt = {
     0: "A caterpillar on a leaf, high quality, photorealistic",
     40: "A caterpillar transforming into a cocoon, on a leaf, near flowers, photorealistic",
-    80: "A cocoon on a leaf, flowers in the backgrond, photorealistic",
+    80: "A cocoon on a leaf, flowers in the background, photorealistic",
     120: "A cocoon maturing and a butterfly being born, flowers and leaves visible in the background, photorealistic",
     160: "A beautiful butterfly, vibrant colors, sitting on a leaf, flowers in the background, photorealistic",
     200: "A beautiful butterfly, flying away in a forest, photorealistic",

docs/source/en/api/pipelines/audioldm2.md

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # AudioLDM 2
 
-AudioLDM 2 was proposed in [AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining](https://arxiv.org/abs/2308.05734) by Haohe Liu et al. AudioLDM 2 takes a text prompt as input and predicts the corresponding audio. It can generate text-conditional sound effects, human speech and music.
+AudioLDM 2 was proposed in [AudioLDM 2: Learning Holistic Audio Generation with Self-supervised Pretraining](https://huggingface.co/papers/2308.05734) by Haohe Liu et al. AudioLDM 2 takes a text prompt as input and predicts the corresponding audio. It can generate text-conditional sound effects, human speech and music.
 
 Inspired by [Stable Diffusion](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion/overview), AudioLDM 2 is a text-to-audio _latent diffusion model (LDM)_ that learns continuous audio representations from text embeddings. Two text encoder models are used to compute the text embeddings from a prompt input: the text-branch of [CLAP](https://huggingface.co/docs/transformers/main/en/model_doc/clap) and the encoder of [Flan-T5](https://huggingface.co/docs/transformers/main/en/model_doc/flan-t5). These text embeddings are then projected to a shared embedding space by an [AudioLDM2ProjectionModel](https://huggingface.co/docs/diffusers/main/api/pipelines/audioldm2#diffusers.AudioLDM2ProjectionModel). A [GPT2](https://huggingface.co/docs/transformers/main/en/model_doc/gpt2) _language model (LM)_ is used to auto-regressively predict eight new embedding vectors, conditional on the projected CLAP and Flan-T5 embeddings. The generated embedding vectors and Flan-T5 text embeddings are used as cross-attention conditioning in the LDM. The [UNet](https://huggingface.co/docs/diffusers/main/en/api/pipelines/audioldm2#diffusers.AudioLDM2UNet2DConditionModel) of AudioLDM 2 is unique in the sense that it takes **two** cross-attention embeddings, as opposed to one cross-attention conditioning, as in most other LDMs.
 

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

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # BLIP-Diffusion
 
-BLIP-Diffusion was proposed in [BLIP-Diffusion: Pre-trained Subject Representation for Controllable Text-to-Image Generation and Editing](https://arxiv.org/abs/2305.14720). It enables zero-shot subject-driven generation and control-guided zero-shot generation.
+BLIP-Diffusion was proposed in [BLIP-Diffusion: Pre-trained Subject Representation for Controllable Text-to-Image Generation and Editing](https://huggingface.co/papers/2305.14720). It enables zero-shot subject-driven generation and control-guided zero-shot generation.
 
 
 The abstract from the paper is:

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

@@ -0,0 +1,71 @@
+<!--Copyright 2025 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.
+-->
+
+# Chroma
+
+<div class="flex flex-wrap space-x-1">
+  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+  <img alt="MPS" src="https://img.shields.io/badge/MPS-000000?style=flat&logo=apple&logoColor=white%22">
+</div>
+
+Chroma is a text to image generation model based on Flux.
+
+Original model checkpoints for Chroma can be found [here](https://huggingface.co/lodestones/Chroma).
+
+<Tip>
+
+Chroma can use all the same optimizations as Flux.
+
+</Tip>
+
+## Inference (Single File)
+
+The `ChromaTransformer2DModel` supports loading checkpoints in the original format. This is also useful when trying to load finetunes or quantized versions of the models that have been published by the community.
+
+The following example demonstrates how to run Chroma from a single file.
+
+Then run the following example
+
+```python
+import torch
+from diffusers import ChromaTransformer2DModel, ChromaPipeline
+from transformers import T5EncoderModel
+
+bfl_repo = "black-forest-labs/FLUX.1-dev"
+dtype = torch.bfloat16
+
+transformer = ChromaTransformer2DModel.from_single_file("https://huggingface.co/lodestones/Chroma/blob/main/chroma-unlocked-v35.safetensors", torch_dtype=dtype)
+
+text_encoder = T5EncoderModel.from_pretrained(bfl_repo, subfolder="text_encoder_2", torch_dtype=dtype)
+tokenizer = T5Tokenizer.from_pretrained(bfl_repo, subfolder="tokenizer_2", torch_dtype=dtype)
+
+pipe = ChromaPipeline.from_pretrained(bfl_repo, transformer=transformer, text_encoder=text_encoder, tokenizer=tokenizer, torch_dtype=dtype)
+
+pipe.enable_model_cpu_offload()
+
+prompt = "A cat holding a sign that says hello world"
+image = pipe(
+    prompt,
+    guidance_scale=4.0,
+    output_type="pil",
+    num_inference_steps=26,
+    generator=torch.Generator("cpu").manual_seed(0)
+).images[0]
+
+image.save("image.png")
+```
+
+## ChromaPipeline
+
+[[autodoc]] ChromaPipeline
+	- all
+	- __call__

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

@@ -13,150 +13,181 @@
 # limitations under the License.
 -->
 
-# CogVideoX
-
-<div class="flex flex-wrap space-x-1">
-  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+<div style="float: right;">
+  <div class="flex flex-wrap space-x-1">
+    <a href="https://huggingface.co/docs/diffusers/main/en/tutorials/using_peft_for_inference" target="_blank" rel="noopener">
+      <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+    </a>
+  </div>
 </div>
 
-[CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer](https://arxiv.org/abs/2408.06072) from Tsinghua University & ZhipuAI, by Zhuoyi Yang, Jiayan Teng, Wendi Zheng, Ming Ding, Shiyu Huang, Jiazheng Xu, Yuanming Yang, Wenyi Hong, Xiaohan Zhang, Guanyu Feng, Da Yin, Xiaotao Gu, Yuxuan Zhang, Weihan Wang, Yean Cheng, Ting Liu, Bin Xu, Yuxiao Dong, Jie Tang.
+# CogVideoX
 
-The abstract from the paper is:
+[CogVideoX](https://huggingface.co/papers/2408.06072) is a large diffusion transformer model - available in 2B and 5B parameters - designed to generate longer and more consistent videos from text. This model uses a 3D causal variational autoencoder to more efficiently process video data by reducing sequence length (and associated training compute) and preventing flickering in generated videos. An "expert" transformer with adaptive LayerNorm improves alignment between text and video, and 3D full attention helps accurately capture motion and time in generated videos.
 
-*We introduce CogVideoX, a large-scale diffusion transformer model designed for generating videos based on text prompts. To efficently model video data, we propose to levearge a 3D Variational Autoencoder (VAE) to compresses videos along both spatial and temporal dimensions. To improve the text-video alignment, we propose an expert transformer with the expert adaptive LayerNorm to facilitate the deep fusion between the two modalities. By employing a progressive training technique, CogVideoX is adept at producing coherent, long-duration videos characterized by significant motion. In addition, we develop an effectively text-video data processing pipeline that includes various data preprocessing strategies and a video captioning method. It significantly helps enhance the performance of CogVideoX, improving both generation quality and semantic alignment. Results show that CogVideoX demonstrates state-of-the-art performance across both multiple machine metrics and human evaluations. The model weight of CogVideoX-2B is publicly available at https://github.com/THUDM/CogVideo.*
+You can find all the original CogVideoX checkpoints under the [CogVideoX](https://huggingface.co/collections/THUDM/cogvideo-66c08e62f1685a3ade464cce) collection.
 
-<Tip>
+> [!TIP]
+> Click on the CogVideoX models in the right sidebar for more examples of other video generation tasks.
 
-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.
+The example below demonstrates how to generate a video optimized for memory or inference speed.
 
-</Tip>
+<hfoptions id="usage">
+<hfoption id="memory">
 
-This pipeline was contributed by [zRzRzRzRzRzRzR](https://github.com/zRzRzRzRzRzRzR). The original codebase can be found [here](https://huggingface.co/THUDM). The original weights can be found under [hf.co/THUDM](https://huggingface.co/THUDM).
+Refer to the [Reduce memory usage](../../optimization/memory) guide for more details about the various memory saving techniques.
 
-There are three official CogVideoX checkpoints for text-to-video and video-to-video.
-
-| checkpoints | recommended inference dtype |
-|:---:|:---:|
-| [`THUDM/CogVideoX-2b`](https://huggingface.co/THUDM/CogVideoX-2b) | torch.float16 |
-| [`THUDM/CogVideoX-5b`](https://huggingface.co/THUDM/CogVideoX-5b) | torch.bfloat16 |
-| [`THUDM/CogVideoX1.5-5b`](https://huggingface.co/THUDM/CogVideoX1.5-5b) | torch.bfloat16 |
-
-There are two official CogVideoX checkpoints available for image-to-video.
-
-| checkpoints | recommended inference dtype |
-|:---:|:---:|
-| [`THUDM/CogVideoX-5b-I2V`](https://huggingface.co/THUDM/CogVideoX-5b-I2V) | torch.bfloat16 |
-| [`THUDM/CogVideoX-1.5-5b-I2V`](https://huggingface.co/THUDM/CogVideoX-1.5-5b-I2V) | torch.bfloat16 |
-
-For the CogVideoX 1.5 series:
-- Text-to-video (T2V) works best at a resolution of 1360x768 because it was trained with that specific resolution.
-- Image-to-video (I2V) works for multiple resolutions. The width can vary from 768 to 1360, but the height must be 768. The height/width must be divisible by 16.
-- Both T2V and I2V models support generation with 81 and 161 frames and work best at this value. Exporting videos at 16 FPS is recommended.
-
-There are two official CogVideoX checkpoints that support pose controllable generation (by the [Alibaba-PAI](https://huggingface.co/alibaba-pai) team).
-
-| checkpoints | recommended inference dtype |
-|:---:|:---:|
-| [`alibaba-pai/CogVideoX-Fun-V1.1-2b-Pose`](https://huggingface.co/alibaba-pai/CogVideoX-Fun-V1.1-2b-Pose) | torch.bfloat16 |
-| [`alibaba-pai/CogVideoX-Fun-V1.1-5b-Pose`](https://huggingface.co/alibaba-pai/CogVideoX-Fun-V1.1-5b-Pose) | torch.bfloat16 |
-
-## Inference
-
-Use [`torch.compile`](https://huggingface.co/docs/diffusers/main/en/tutorials/fast_diffusion#torchcompile) to reduce the inference latency.
-
-First, load the pipeline:
-
-```python
-import torch
-from diffusers import CogVideoXPipeline, CogVideoXImageToVideoPipeline
-from diffusers.utils import export_to_video,load_image
-pipe = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b").to("cuda") # or "THUDM/CogVideoX-2b" 
-```
-
-If you are using the image-to-video pipeline, load it as follows:
-
-```python
-pipe = CogVideoXImageToVideoPipeline.from_pretrained("THUDM/CogVideoX-5b-I2V").to("cuda")
-```
-
-Then change the memory layout of the pipelines `transformer` component to `torch.channels_last`:
-
-```python
-pipe.transformer.to(memory_format=torch.channels_last)
-```
-
-Compile the components and run inference:
-
-```python
-pipe.transformer = torch.compile(pipeline.transformer, mode="max-autotune", fullgraph=True)
-
-# CogVideoX works well with long and well-described prompts
-prompt = "A panda, dressed in a small, red jacket and a tiny hat, sits on a wooden stool in a serene bamboo forest. The panda's fluffy paws strum a miniature acoustic guitar, producing soft, melodic tunes. Nearby, a few other pandas gather, watching curiously and some clapping in rhythm. Sunlight filters through the tall bamboo, casting a gentle glow on the scene. The panda's face is expressive, showing concentration and joy as it plays. The background includes a small, flowing stream and vibrant green foliage, enhancing the peaceful and magical atmosphere of this unique musical performance."
-video = pipe(prompt=prompt, guidance_scale=6, num_inference_steps=50).frames[0]
-```
-
-The [T2V benchmark](https://gist.github.com/a-r-r-o-w/5183d75e452a368fd17448fcc810bd3f) results on an 80GB A100 machine are:
-
-```
-Without torch.compile(): Average inference time: 96.89 seconds.
-With torch.compile(): Average inference time: 76.27 seconds.
-```
-
-### Memory optimization
-
-CogVideoX-2b requires about 19 GB of GPU memory to decode 49 frames (6 seconds of video at 8 FPS) with output resolution 720x480 (W x H), which makes it not possible to run on consumer GPUs or free-tier T4 Colab. The following memory optimizations could be used to reduce the memory footprint. For replication, you can refer to [this](https://gist.github.com/a-r-r-o-w/3959a03f15be5c9bd1fe545b09dfcc93) script.
-
-- `pipe.enable_model_cpu_offload()`:
-  - Without enabling cpu offloading, memory usage is `33 GB`
-  - With enabling cpu offloading, memory usage is `19 GB`
-- `pipe.enable_sequential_cpu_offload()`:
-  - Similar to `enable_model_cpu_offload` but can significantly reduce memory usage at the cost of slow inference
-  - When enabled, memory usage is under `4 GB`
-- `pipe.vae.enable_tiling()`:
-  - With enabling cpu offloading and tiling, memory usage is `11 GB`
-- `pipe.vae.enable_slicing()`
-
-## Quantization
-
-Quantization helps reduce the memory requirements of very large models by storing model weights in a lower precision data type. However, quantization may have varying impact on video quality depending on the video model.
-
-Refer to the [Quantization](../../quantization/overview) overview to learn more about supported quantization backends and selecting a quantization backend that supports your use case. The example below demonstrates how to load a quantized [`CogVideoXPipeline`] for inference with bitsandbytes.
+The quantized CogVideoX 5B model below requires ~16GB of VRAM.
 
 ```py
 import torch
-from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig, CogVideoXTransformer3DModel, CogVideoXPipeline
+from diffusers import CogVideoXPipeline, AutoModel
+from diffusers.quantizers import PipelineQuantizationConfig
+from diffusers.hooks import apply_group_offloading
 from diffusers.utils import export_to_video
-from transformers import BitsAndBytesConfig as BitsAndBytesConfig, T5EncoderModel
 
-quant_config = BitsAndBytesConfig(load_in_8bit=True)
-text_encoder_8bit = T5EncoderModel.from_pretrained(
-    "THUDM/CogVideoX-2b",
-    subfolder="text_encoder",
-    quantization_config=quant_config,
-    torch_dtype=torch.float16,
+# quantize weights to int8 with torchao
+pipeline_quant_config = PipelineQuantizationConfig(
+  quant_backend="torchao",
+  quant_kwargs={"quant_type": "int8wo"},
+  components_to_quantize=["transformer"]
 )
 
-quant_config = DiffusersBitsAndBytesConfig(load_in_8bit=True)
-transformer_8bit = CogVideoXTransformer3DModel.from_pretrained(
-    "THUDM/CogVideoX-2b",
+# fp8 layerwise weight-casting
+transformer = AutoModel.from_pretrained(
+    "THUDM/CogVideoX-5b",
     subfolder="transformer",
-    quantization_config=quant_config,
-    torch_dtype=torch.float16,
+    torch_dtype=torch.bfloat16
+)
+transformer.enable_layerwise_casting(
+    storage_dtype=torch.float8_e4m3fn, compute_dtype=torch.bfloat16
 )
 
 pipeline = CogVideoXPipeline.from_pretrained(
-    "THUDM/CogVideoX-2b",
-    text_encoder=text_encoder_8bit,
-    transformer=transformer_8bit,
-    torch_dtype=torch.float16,
-    device_map="balanced",
+    "THUDM/CogVideoX-5b",
+    transformer=transformer,
+    quantization_config=pipeline_quant_config,
+    torch_dtype=torch.bfloat16
 )
+pipeline.to("cuda")
 
-prompt = "A detailed wooden toy ship with intricately carved masts and sails is seen gliding smoothly over a plush, blue carpet that mimics the waves of the sea. The ship's hull is painted a rich brown, with tiny windows. The carpet, soft and textured, provides a perfect backdrop, resembling an oceanic expanse. Surrounding the ship are various other toys and children's items, hinting at a playful environment. The scene captures the innocence and imagination of childhood, with the toy ship's journey symbolizing endless adventures in a whimsical, indoor setting."
-video = pipeline(prompt=prompt, guidance_scale=6, num_inference_steps=50).frames[0]
-export_to_video(video, "ship.mp4", fps=8)
+# model-offloading
+pipeline.enable_model_cpu_offload()
+
+prompt = """
+A detailed wooden toy ship with intricately carved masts and sails is seen gliding smoothly over a plush, blue carpet that mimics the waves of the sea. 
+The ship's hull is painted a rich brown, with tiny windows. The carpet, soft and textured, provides a perfect backdrop, resembling an oceanic expanse. 
+Surrounding the ship are various other toys and children's items, hinting at a playful environment. The scene captures the innocence and imagination of childhood, 
+with the toy ship's journey symbolizing endless adventures in a whimsical, indoor setting.
+"""
+
+video = pipeline(
+    prompt=prompt,
+    guidance_scale=6,
+    num_inference_steps=50
+).frames[0]
+export_to_video(video, "output.mp4", fps=8)
 ```
 
+</hfoption>
+<hfoption id="inference speed">
+
+[Compilation](../../optimization/fp16#torchcompile) is slow the first time but subsequent calls to the pipeline are faster.
+
+The average inference time with torch.compile on a 80GB A100 is 76.27 seconds compared to 96.89 seconds for an uncompiled model.
+
+```py
+import torch
+from diffusers import CogVideoXPipeline
+from diffusers.utils import export_to_video
+
+pipeline = CogVideoXPipeline.from_pretrained(
+    "THUDM/CogVideoX-2b",
+    torch_dtype=torch.float16
+).to("cuda")
+
+# torch.compile
+pipeline.transformer.to(memory_format=torch.channels_last)
+pipeline.transformer = torch.compile(
+    pipeline.transformer, mode="max-autotune", fullgraph=True
+)
+
+prompt = """
+A detailed wooden toy ship with intricately carved masts and sails is seen gliding smoothly over a plush, blue carpet that mimics the waves of the sea. 
+The ship's hull is painted a rich brown, with tiny windows. The carpet, soft and textured, provides a perfect backdrop, resembling an oceanic expanse. 
+Surrounding the ship are various other toys and children's items, hinting at a playful environment. The scene captures the innocence and imagination of childhood, 
+with the toy ship's journey symbolizing endless adventures in a whimsical, indoor setting.
+"""
+
+video = pipeline(
+    prompt=prompt,
+    guidance_scale=6,
+    num_inference_steps=50
+).frames[0]
+export_to_video(video, "output.mp4", fps=8)
+```
+
+</hfoption>
+</hfoptions>
+
+## Notes
+
+- CogVideoX supports LoRAs with [`~loaders.CogVideoXLoraLoaderMixin.load_lora_weights`].
+
+  <details>
+  <summary>Show example code</summary>
+
+  ```py
+  import torch
+  from diffusers import CogVideoXPipeline
+  from diffusers.hooks import apply_group_offloading
+  from diffusers.utils import export_to_video
+
+  pipeline = CogVideoXPipeline.from_pretrained(
+      "THUDM/CogVideoX-5b",
+      torch_dtype=torch.bfloat16
+  )
+  pipeline.to("cuda")
+
+  # load LoRA weights
+  pipeline.load_lora_weights("finetrainers/CogVideoX-1.5-crush-smol-v0", adapter_name="crush-lora")
+  pipeline.set_adapters("crush-lora", 0.9)
+
+  # model-offloading
+  pipeline.enable_model_cpu_offload()
+
+  prompt = """
+  PIKA_CRUSH A large metal cylinder is seen pressing down on a pile of Oreo cookies, flattening them as if they were under a hydraulic press.
+  """
+  negative_prompt = "inconsistent motion, blurry motion, worse quality, degenerate outputs, deformed outputs"
+
+  video = pipeline(
+      prompt=prompt, 
+      negative_prompt=negative_prompt, 
+      num_frames=81, 
+      height=480,
+      width=768,
+      num_inference_steps=50
+  ).frames[0]
+  export_to_video(video, "output.mp4", fps=16)
+  ```
+
+  </details>
+
+- The text-to-video (T2V) checkpoints work best with a resolution of 1360x768 because that was the resolution it was pretrained on.
+
+- The image-to-video (I2V) checkpoints work with multiple resolutions. The width can vary from 768 to 1360, but the height must be 758. Both height and width must be divisible by 16.
+
+- Both T2V and I2V checkpoints work best with 81 and 161 frames. It is recommended to export the generated video at 16fps.
+
+- Refer to the table below to view memory usage when various memory-saving techniques are enabled.
+
+  | method | memory usage (enabled) | memory usage (disabled) |
+  |---|---|---|
+  | enable_model_cpu_offload | 19GB | 33GB |
+  | enable_sequential_cpu_offload | <4GB | ~33GB (very slow inference speed) |
+  | enable_tiling | 11GB (with enable_model_cpu_offload) | --- |
+ 
 ## CogVideoXPipeline
 
 [[autodoc]] CogVideoXPipeline

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

@@ -19,7 +19,7 @@
   <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
 </div>
 
-[Identity-Preserving Text-to-Video Generation by Frequency Decomposition](https://arxiv.org/abs/2411.17440) from Peking University & University of Rochester & etc, by Shenghai Yuan, Jinfa Huang, Xianyi He, Yunyang Ge, Yujun Shi, Liuhan Chen, Jiebo Luo, Li Yuan.
+[Identity-Preserving Text-to-Video Generation by Frequency Decomposition](https://huggingface.co/papers/2411.17440) from Peking University & University of Rochester & etc, by Shenghai Yuan, Jinfa Huang, Xianyi He, Yunyang Ge, Yujun Shi, Liuhan Chen, Jiebo Luo, Li Yuan.
 
 The abstract from the paper is:
 

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

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # ControlNet with Hunyuan-DiT
 
-HunyuanDiTControlNetPipeline is an implementation of ControlNet for [Hunyuan-DiT](https://arxiv.org/abs/2405.08748).
+HunyuanDiTControlNetPipeline is an implementation of ControlNet for [Hunyuan-DiT](https://huggingface.co/papers/2405.08748).
 
 ControlNet was introduced in [Adding Conditional Control to Text-to-Image Diffusion Models](https://huggingface.co/papers/2302.05543) by Lvmin Zhang, Anyi Rao, and Maneesh Agrawala.
 

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

@@ -0,0 +1,57 @@
+<!-- Copyright 2024 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. -->
+
+# Cosmos
+
+[Cosmos World Foundation Model Platform for Physical AI](https://huggingface.co/papers/2501.03575) by NVIDIA.
+
+*Physical AI needs to be trained digitally first. It needs a digital twin of itself, the policy model, and a digital twin of the world, the world model. In this paper, we present the Cosmos World Foundation Model Platform to help developers build customized world models for their Physical AI setups. We position a world foundation model as a general-purpose world model that can be fine-tuned into customized world models for downstream applications. Our platform covers a video curation pipeline, pre-trained world foundation models, examples of post-training of pre-trained world foundation models, and video tokenizers. To help Physical AI builders solve the most critical problems of our society, we make our platform open-source and our models open-weight with permissive licenses available via https://github.com/NVIDIA/Cosmos.*
+
+<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.
+
+</Tip>
+
+## CosmosTextToWorldPipeline
+
+[[autodoc]] CosmosTextToWorldPipeline
+  - all
+  - __call__
+
+## CosmosVideoToWorldPipeline
+
+[[autodoc]] CosmosVideoToWorldPipeline
+  - all
+  - __call__
+
+## Cosmos2TextToImagePipeline
+
+[[autodoc]] Cosmos2TextToImagePipeline
+  - all
+  - __call__
+
+## Cosmos2VideoToWorldPipeline
+
+[[autodoc]] Cosmos2VideoToWorldPipeline
+  - all
+  - __call__
+
+## CosmosPipelineOutput
+
+[[autodoc]] pipelines.cosmos.pipeline_output.CosmosPipelineOutput
+
+## CosmosImagePipelineOutput
+
+[[autodoc]] pipelines.cosmos.pipeline_output.CosmosImagePipelineOutput

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

@@ -347,7 +347,7 @@ pipe.to("cuda")
 image = pipe(image=image, prompt="<prompt>", strength=0.3).images
 ```
 
-You can also use [`torch.compile`](../../optimization/torch2.0). Note that we have not exhaustively tested `torch.compile`
+You can also use [`torch.compile`](../../optimization/fp16#torchcompile). Note that we have not exhaustively tested `torch.compile`
 with IF and it might not give expected results.
 
 ```py

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

@@ -0,0 +1,209 @@
+<!-- Copyright 2025 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. -->
+
+# Framepack
+
+<div class="flex flex-wrap space-x-1">
+  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+</div>
+
+[Packing Input Frame Context in Next-Frame Prediction Models for Video Generation](https://huggingface.co/papers/2504.12626) by Lvmin Zhang and Maneesh Agrawala.
+
+*We present a neural network structure, FramePack, to train next-frame (or next-frame-section) prediction models for video generation. The FramePack compresses input frames to make the transformer context length a fixed number regardless of the video length. As a result, we are able to process a large number of frames using video diffusion with computation bottleneck similar to image diffusion. This also makes the training video batch sizes significantly higher (batch sizes become comparable to image diffusion training). We also propose an anti-drifting sampling method that generates frames in inverted temporal order with early-established endpoints to avoid exposure bias (error accumulation over iterations). Finally, we show that existing video diffusion models can be finetuned with FramePack, and their visual quality may be improved because the next-frame prediction supports more balanced diffusion schedulers with less extreme flow shift timesteps.*
+
+<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.
+
+</Tip>
+
+## Available models
+
+| Model name | Description |
+|:---|:---|
+- [`lllyasviel/FramePackI2V_HY`](https://huggingface.co/lllyasviel/FramePackI2V_HY) | Trained with the "inverted anti-drifting" strategy as described in the paper. Inference requires setting `sampling_type="inverted_anti_drifting"` when running the pipeline. |
+- [`lllyasviel/FramePack_F1_I2V_HY_20250503`](https://huggingface.co/lllyasviel/FramePack_F1_I2V_HY_20250503) | Trained with a novel anti-drifting strategy but inference is performed in "vanilla" strategy as described in the paper. Inference requires setting `sampling_type="vanilla"` when running the pipeline. |
+
+## Usage
+
+Refer to the pipeline documentation for basic usage examples. The following section contains examples of offloading, different sampling methods, quantization, and more.
+
+### First and last frame to video
+
+The following example shows how to use Framepack with start and end image controls, using the inverted anti-drifiting sampling model.
+
+```python
+import torch
+from diffusers import HunyuanVideoFramepackPipeline, HunyuanVideoFramepackTransformer3DModel
+from diffusers.utils import export_to_video, load_image
+from transformers import SiglipImageProcessor, SiglipVisionModel
+
+transformer = HunyuanVideoFramepackTransformer3DModel.from_pretrained(
+    "lllyasviel/FramePackI2V_HY", torch_dtype=torch.bfloat16
+)
+feature_extractor = SiglipImageProcessor.from_pretrained(
+    "lllyasviel/flux_redux_bfl", subfolder="feature_extractor"
+)
+image_encoder = SiglipVisionModel.from_pretrained(
+    "lllyasviel/flux_redux_bfl", subfolder="image_encoder", torch_dtype=torch.float16
+)
+pipe = HunyuanVideoFramepackPipeline.from_pretrained(
+    "hunyuanvideo-community/HunyuanVideo",
+    transformer=transformer,
+    feature_extractor=feature_extractor,
+    image_encoder=image_encoder,
+    torch_dtype=torch.float16,
+)
+
+# Enable memory optimizations
+pipe.enable_model_cpu_offload()
+pipe.vae.enable_tiling()
+
+prompt = "CG animation style, a small blue bird takes off from the ground, flapping its wings. The bird's feathers are delicate, with a unique pattern on its chest. The background shows a blue sky with white clouds under bright sunshine. The camera follows the bird upward, capturing its flight and the vastness of the sky from a close-up, low-angle perspective."
+first_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/flf2v_input_first_frame.png"
+)
+last_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/flf2v_input_last_frame.png"
+)
+output = pipe(
+    image=first_image,
+    last_image=last_image,
+    prompt=prompt,
+    height=512,
+    width=512,
+    num_frames=91,
+    num_inference_steps=30,
+    guidance_scale=9.0,
+    generator=torch.Generator().manual_seed(0),
+    sampling_type="inverted_anti_drifting",
+).frames[0]
+export_to_video(output, "output.mp4", fps=30)
+```
+
+### Vanilla sampling
+
+The following example shows how to use Framepack with the F1 model trained with vanilla sampling but new regulation approach for anti-drifting.
+
+```python
+import torch
+from diffusers import HunyuanVideoFramepackPipeline, HunyuanVideoFramepackTransformer3DModel
+from diffusers.utils import export_to_video, load_image
+from transformers import SiglipImageProcessor, SiglipVisionModel
+
+transformer = HunyuanVideoFramepackTransformer3DModel.from_pretrained(
+    "lllyasviel/FramePack_F1_I2V_HY_20250503", torch_dtype=torch.bfloat16
+)
+feature_extractor = SiglipImageProcessor.from_pretrained(
+    "lllyasviel/flux_redux_bfl", subfolder="feature_extractor"
+)
+image_encoder = SiglipVisionModel.from_pretrained(
+    "lllyasviel/flux_redux_bfl", subfolder="image_encoder", torch_dtype=torch.float16
+)
+pipe = HunyuanVideoFramepackPipeline.from_pretrained(
+    "hunyuanvideo-community/HunyuanVideo",
+    transformer=transformer,
+    feature_extractor=feature_extractor,
+    image_encoder=image_encoder,
+    torch_dtype=torch.float16,
+)
+
+# Enable memory optimizations
+pipe.enable_model_cpu_offload()
+pipe.vae.enable_tiling()
+
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/penguin.png"
+)
+output = pipe(
+    image=image,
+    prompt="A penguin dancing in the snow",
+    height=832,
+    width=480,
+    num_frames=91,
+    num_inference_steps=30,
+    guidance_scale=9.0,
+    generator=torch.Generator().manual_seed(0),
+    sampling_type="vanilla",
+).frames[0]
+export_to_video(output, "output.mp4", fps=30)
+```
+
+### Group offloading
+
+Group offloading ([`~hooks.apply_group_offloading`]) provides aggressive memory optimizations for offloading internal parts of any model to the CPU, with possibly no additional overhead to generation time. If you have very low VRAM available, this approach may be suitable for you depending on the amount of CPU RAM available.
+
+```python
+import torch
+from diffusers import HunyuanVideoFramepackPipeline, HunyuanVideoFramepackTransformer3DModel
+from diffusers.hooks import apply_group_offloading
+from diffusers.utils import export_to_video, load_image
+from transformers import SiglipImageProcessor, SiglipVisionModel
+
+transformer = HunyuanVideoFramepackTransformer3DModel.from_pretrained(
+    "lllyasviel/FramePack_F1_I2V_HY_20250503", torch_dtype=torch.bfloat16
+)
+feature_extractor = SiglipImageProcessor.from_pretrained(
+    "lllyasviel/flux_redux_bfl", subfolder="feature_extractor"
+)
+image_encoder = SiglipVisionModel.from_pretrained(
+    "lllyasviel/flux_redux_bfl", subfolder="image_encoder", torch_dtype=torch.float16
+)
+pipe = HunyuanVideoFramepackPipeline.from_pretrained(
+    "hunyuanvideo-community/HunyuanVideo",
+    transformer=transformer,
+    feature_extractor=feature_extractor,
+    image_encoder=image_encoder,
+    torch_dtype=torch.float16,
+)
+
+# Enable group offloading
+onload_device = torch.device("cuda")
+offload_device = torch.device("cpu")
+list(map(
+    lambda x: apply_group_offloading(x, onload_device, offload_device, offload_type="leaf_level", use_stream=True, low_cpu_mem_usage=True),
+    [pipe.text_encoder, pipe.text_encoder_2, pipe.transformer]
+))
+pipe.image_encoder.to(onload_device)
+pipe.vae.to(onload_device)
+pipe.vae.enable_tiling()
+
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/penguin.png"
+)
+output = pipe(
+    image=image,
+    prompt="A penguin dancing in the snow",
+    height=832,
+    width=480,
+    num_frames=91,
+    num_inference_steps=30,
+    guidance_scale=9.0,
+    generator=torch.Generator().manual_seed(0),
+    sampling_type="vanilla",
+).frames[0]
+print(f"Max memory: {torch.cuda.max_memory_allocated() / 1024**3:.3f} GB")
+export_to_video(output, "output.mp4", fps=30)
+```
+
+## HunyuanVideoFramepackPipeline
+
+[[autodoc]] HunyuanVideoFramepackPipeline
+  - all
+  - __call__
+
+## HunyuanVideoPipelineOutput
+
+[[autodoc]] pipelines.hunyuan_video.pipeline_output.HunyuanVideoPipelineOutput
+

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

@@ -12,78 +12,171 @@
 # See the License for the specific language governing permissions and
 # limitations under the License. -->
 
-# HunyuanVideo
-
-<div class="flex flex-wrap space-x-1">
-  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+<div style="float: right;">
+  <div class="flex flex-wrap space-x-1">
+    <a href="https://huggingface.co/docs/diffusers/main/en/tutorials/using_peft_for_inference" target="_blank" rel="noopener">
+      <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+    </a>
+  </div>
 </div>
 
-[HunyuanVideo](https://www.arxiv.org/abs/2412.03603) by Tencent.
+# HunyuanVideo
 
-*Recent advancements in video generation have significantly impacted daily life for both individuals and industries. However, the leading video generation models remain closed-source, resulting in a notable performance gap between industry capabilities and those available to the public. In this report, we introduce HunyuanVideo, an innovative open-source video foundation model that demonstrates performance in video generation comparable to, or even surpassing, that of leading closed-source models. HunyuanVideo encompasses a comprehensive framework that integrates several key elements, including data curation, advanced architectural design, progressive model scaling and training, and an efficient infrastructure tailored for large-scale model training and inference. As a result, we successfully trained a video generative model with over 13 billion parameters, making it the largest among all open-source models. We conducted extensive experiments and implemented a series of targeted designs to ensure high visual quality, motion dynamics, text-video alignment, and advanced filming techniques. According to evaluations by professionals, HunyuanVideo outperforms previous state-of-the-art models, including Runway Gen-3, Luma 1.6, and three top-performing Chinese video generative models. By releasing the code for the foundation model and its applications, we aim to bridge the gap between closed-source and open-source communities. This initiative will empower individuals within the community to experiment with their ideas, fostering a more dynamic and vibrant video generation ecosystem. The code is publicly available at [this https URL](https://github.com/tencent/HunyuanVideo).*
+[HunyuanVideo](https://huggingface.co/papers/2412.03603) is a 13B parameter diffusion transformer model designed to be competitive with closed-source video foundation models and enable wider community access. This model uses a "dual-stream to single-stream" architecture to separately process the video and text tokens first, before concatenating and feeding them to the transformer to fuse the multimodal information. A pretrained multimodal large language model (MLLM) is used as the encoder because it has better image-text alignment, better image detail description and reasoning, and it can be used as a zero-shot learner if system instructions are added to user prompts. Finally, HunyuanVideo uses a 3D causal variational autoencoder to more efficiently process video data at the original resolution and frame rate.
 
-<Tip>
+You can find all the original HunyuanVideo checkpoints under the [Tencent](https://huggingface.co/tencent) organization.
 
-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.
+> [!TIP]
+> Click on the HunyuanVideo models in the right sidebar for more examples of video generation tasks.
+>
+> The examples below use a checkpoint from [hunyuanvideo-community](https://huggingface.co/hunyuanvideo-community) because the weights are stored in a layout compatible with Diffusers.
 
-</Tip>
+The example below demonstrates how to generate a video optimized for memory or inference speed.
 
-Recommendations for inference:
-- Both text encoders should be in `torch.float16`.
-- Transformer should be in `torch.bfloat16`.
-- VAE should be in `torch.float16`.
-- `num_frames` should be of the form `4 * k + 1`, for example `49` or `129`.
-- 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 images, try higher values (between `7.0` and `12.0`). The default value is `7.0` for HunyuanVideo.
-- For more information about supported resolutions and other details, please refer to the original repository [here](https://github.com/Tencent/HunyuanVideo/).
+<hfoptions id="usage">
+<hfoption id="memory">
 
-## Available models
+Refer to the [Reduce memory usage](../../optimization/memory) guide for more details about the various memory saving techniques.
 
-The following models are available for the [`HunyuanVideoPipeline`](text-to-video) pipeline:
-
-| Model name | Description |
-|:---|:---|
-| [`hunyuanvideo-community/HunyuanVideo`](https://huggingface.co/hunyuanvideo-community/HunyuanVideo) | Official HunyuanVideo (guidance-distilled). Performs best at multiple resolutions and frames. Performs best with `guidance_scale=6.0`, `true_cfg_scale=1.0` and without a negative prompt. |
-| [`https://huggingface.co/Skywork/SkyReels-V1-Hunyuan-T2V`](https://huggingface.co/Skywork/SkyReels-V1-Hunyuan-T2V) | Skywork's custom finetune of HunyuanVideo (de-distilled). Performs best with `97x544x960` resolution, `guidance_scale=1.0`, `true_cfg_scale=6.0` and a negative prompt. |
-
-The following models are available for the image-to-video pipeline:
-
-| Model name | Description |
-|:---|:---|
-| [`Skywork/SkyReels-V1-Hunyuan-I2V`](https://huggingface.co/Skywork/SkyReels-V1-Hunyuan-I2V) | Skywork's custom finetune of HunyuanVideo (de-distilled). Performs best with `97x544x960` resolution. Performs best at `97x544x960` resolution, `guidance_scale=1.0`, `true_cfg_scale=6.0` and a negative prompt. |
-| [`hunyuanvideo-community/HunyuanVideo-I2V-33ch`](https://huggingface.co/hunyuanvideo-community/HunyuanVideo-I2V) | Tecent's official HunyuanVideo 33-channel I2V model. Performs best at resolutions of 480, 720, 960, 1280. A higher `shift` value when initializing the scheduler is recommended (good values are between 7 and 20). |
-| [`hunyuanvideo-community/HunyuanVideo-I2V`](https://huggingface.co/hunyuanvideo-community/HunyuanVideo-I2V) | Tecent's official HunyuanVideo 16-channel I2V model. Performs best at resolutions of 480, 720, 960, 1280. A higher `shift` value when initializing the scheduler is recommended (good values are between 7 and 20) |
-
-## Quantization
-
-Quantization helps reduce the memory requirements of very large models by storing model weights in a lower precision data type. However, quantization may have varying impact on video quality depending on the video model.
-
-Refer to the [Quantization](../../quantization/overview) overview to learn more about supported quantization backends and selecting a quantization backend that supports your use case. The example below demonstrates how to load a quantized [`HunyuanVideoPipeline`] for inference with bitsandbytes.
+The quantized HunyuanVideo model below requires ~14GB of VRAM.
 
 ```py
 import torch
-from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig, HunyuanVideoTransformer3DModel, HunyuanVideoPipeline
+from diffusers import AutoModel, HunyuanVideoPipeline
+from diffusers.quantizers import PipelineQuantizationConfig
 from diffusers.utils import export_to_video
 
-quant_config = DiffusersBitsAndBytesConfig(load_in_8bit=True)
-transformer_8bit = HunyuanVideoTransformer3DModel.from_pretrained(
-    "hunyuanvideo-community/HunyuanVideo",
-    subfolder="transformer",
-    quantization_config=quant_config,
-    torch_dtype=torch.bfloat16,
+# quantize weights to int4 with bitsandbytes
+pipeline_quant_config = PipelineQuantizationConfig(
+    quant_backend="bitsandbytes_4bit",
+    quant_kwargs={
+      "load_in_4bit": True,
+      "bnb_4bit_quant_type": "nf4",
+      "bnb_4bit_compute_dtype": torch.bfloat16
+      },
+    components_to_quantize=["transformer"]
 )
 
 pipeline = HunyuanVideoPipeline.from_pretrained(
     "hunyuanvideo-community/HunyuanVideo",
-    transformer=transformer_8bit,
-    torch_dtype=torch.float16,
-    device_map="balanced",
+    quantization_config=pipeline_quant_config,
+    torch_dtype=torch.bfloat16,
 )
 
-prompt = "A cat walks on the grass, realistic style."
+# model-offloading and tiling
+pipeline.enable_model_cpu_offload()
+pipeline.vae.enable_tiling()
+
+prompt = "A fluffy teddy bear sits on a bed of soft pillows surrounded by children's toys."
 video = pipeline(prompt=prompt, num_frames=61, num_inference_steps=30).frames[0]
-export_to_video(video, "cat.mp4", fps=15)
+export_to_video(video, "output.mp4", fps=15)
 ```
 
+</hfoption>
+<hfoption id="inference speed">
+
+[Compilation](../../optimization/fp16#torchcompile) is slow the first time but subsequent calls to the pipeline are faster.
+
+```py
+import torch
+from diffusers import AutoModel, HunyuanVideoPipeline
+from diffusers.quantizers import PipelineQuantizationConfig
+from diffusers.utils import export_to_video
+
+# quantize weights to int4 with bitsandbytes
+pipeline_quant_config = PipelineQuantizationConfig(
+    quant_backend="bitsandbytes_4bit",
+    quant_kwargs={
+      "load_in_4bit": True,
+      "bnb_4bit_quant_type": "nf4",
+      "bnb_4bit_compute_dtype": torch.bfloat16
+      },
+    components_to_quantize=["transformer"]
+)
+
+pipeline = HunyuanVideoPipeline.from_pretrained(
+    "hunyuanvideo-community/HunyuanVideo",
+    quantization_config=pipeline_quant_config,
+    torch_dtype=torch.bfloat16,
+)
+
+# model-offloading and tiling
+pipeline.enable_model_cpu_offload()
+pipeline.vae.enable_tiling()
+
+# torch.compile
+pipeline.transformer.to(memory_format=torch.channels_last)
+pipeline.transformer = torch.compile(
+    pipeline.transformer, mode="max-autotune", fullgraph=True
+)
+
+prompt = "A fluffy teddy bear sits on a bed of soft pillows surrounded by children's toys."
+video = pipeline(prompt=prompt, num_frames=61, num_inference_steps=30).frames[0]
+export_to_video(video, "output.mp4", fps=15)
+```
+
+</hfoption>
+</hfoptions>
+
+## Notes
+
+- HunyuanVideo supports LoRAs with [`~loaders.HunyuanVideoLoraLoaderMixin.load_lora_weights`].
+
+  <details>
+  <summary>Show example code</summary>
+
+  ```py
+  import torch
+  from diffusers import AutoModel, HunyuanVideoPipeline
+  from diffusers.quantizers import PipelineQuantizationConfig
+  from diffusers.utils import export_to_video
+
+  # quantize weights to int4 with bitsandbytes
+  pipeline_quant_config = PipelineQuantizationConfig(
+      quant_backend="bitsandbytes_4bit",
+      quant_kwargs={
+        "load_in_4bit": True,
+        "bnb_4bit_quant_type": "nf4",
+        "bnb_4bit_compute_dtype": torch.bfloat16
+        },
+      components_to_quantize=["transformer"]
+  )
+
+  pipeline = HunyuanVideoPipeline.from_pretrained(
+      "hunyuanvideo-community/HunyuanVideo",
+      quantization_config=pipeline_quant_config,
+      torch_dtype=torch.bfloat16,
+  )
+
+  # load LoRA weights
+  pipeline.load_lora_weights("https://huggingface.co/lucataco/hunyuan-steamboat-willie-10", adapter_name="steamboat-willie")
+  pipeline.set_adapters("steamboat-willie", 0.9)
+
+  # model-offloading and tiling
+  pipeline.enable_model_cpu_offload()
+  pipeline.vae.enable_tiling()
+
+  # use "In the style of SWR" to trigger the LoRA
+  prompt = """
+  In the style of SWR. A black and white animated scene featuring a fluffy teddy bear sits on a bed of soft pillows surrounded by children's toys.
+  """
+  video = pipeline(prompt=prompt, num_frames=61, num_inference_steps=30).frames[0]
+  export_to_video(video, "output.mp4", fps=15)
+  ```
+
+  </details>
+
+- Refer to the table below for recommended inference values.
+
+  | parameter | recommended value |
+  |---|---|
+  | text encoder dtype | `torch.float16` |
+  | transformer dtype | `torch.bfloat16` |
+  | vae dtype | `torch.float16` |
+  | `num_frames (k)` | 4 * `k` + 1 |
+
+- Try lower `shift` values (`2.0` to `5.0`) for lower resolution videos and higher `shift` values (`7.0` to `12.0`) for higher resolution images.
+
 ## HunyuanVideoPipeline
 
 [[autodoc]] HunyuanVideoPipeline

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

@@ -13,7 +13,7 @@ specific language governing permissions and limitations under the License.
 # Hunyuan-DiT
 ![chinese elements understanding](https://github.com/gnobitab/diffusers-hunyuan/assets/1157982/39b99036-c3cb-4f16-bb1a-40ec25eda573)
 
-[Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding](https://arxiv.org/abs/2405.08748) from Tencent Hunyuan.
+[Hunyuan-DiT : A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding](https://huggingface.co/papers/2405.08748) from Tencent Hunyuan.
 
 The abstract from the paper is:
 

docs/source/en/api/pipelines/i2vgenxl.md

@@ -47,7 +47,7 @@ Sample output with I2VGenXL:
 * Unlike SVD, it additionally accepts text prompts as inputs.
 * It can generate higher resolution videos.
 * When using the [`DDIMScheduler`] (which is default for this pipeline), less than 50 steps for inference leads to bad results.
-* This implementation is 1-stage variant of I2VGenXL. The main figure in the [I2VGen-XL](https://arxiv.org/abs/2311.04145) paper shows a 2-stage variant, however, 1-stage variant works well. See [this discussion](https://github.com/huggingface/diffusers/discussions/7952) for more details.
+* This implementation is 1-stage variant of I2VGenXL. The main figure in the [I2VGen-XL](https://huggingface.co/papers/2311.04145) paper shows a 2-stage variant, however, 1-stage variant works well. See [this discussion](https://github.com/huggingface/diffusers/discussions/7952) for more details.
 
 ## I2VGenXLPipeline
 [[autodoc]] I2VGenXLPipeline

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

@@ -16,13 +16,13 @@
 
 ![latte text-to-video](https://github.com/Vchitect/Latte/blob/52bc0029899babbd6e9250384c83d8ed2670ff7a/visuals/latte.gif?raw=true)
 
-[Latte: Latent Diffusion Transformer for Video Generation](https://arxiv.org/abs/2401.03048) from Monash University, Shanghai AI Lab, Nanjing University, and Nanyang Technological University.
+[Latte: Latent Diffusion Transformer for Video Generation](https://huggingface.co/papers/2401.03048) from Monash University, Shanghai AI Lab, Nanjing University, and Nanyang Technological University.
 
 The abstract from the paper is:
 
 *We propose a novel Latent Diffusion Transformer, namely Latte, for video generation. Latte first extracts spatio-temporal tokens from input videos and then adopts a series of Transformer blocks to model video distribution in the latent space. In order to model a substantial number of tokens extracted from videos, four efficient variants are introduced from the perspective of decomposing the spatial and temporal dimensions of input videos. To improve the quality of generated videos, we determine the best practices of Latte through rigorous experimental analysis, including video clip patch embedding, model variants, timestep-class information injection, temporal positional embedding, and learning strategies. Our comprehensive evaluation demonstrates that Latte achieves state-of-the-art performance across four standard video generation datasets, i.e., FaceForensics, SkyTimelapse, UCF101, and Taichi-HD. In addition, we extend Latte to text-to-video generation (T2V) task, where Latte achieves comparable results compared to recent T2V models. We strongly believe that Latte provides valuable insights for future research on incorporating Transformers into diffusion models for video generation.*
 
-**Highlights**: Latte is a latent diffusion transformer proposed as a backbone for modeling different modalities (trained for text-to-video generation here). It achieves state-of-the-art performance across four standard video benchmarks - [FaceForensics](https://arxiv.org/abs/1803.09179), [SkyTimelapse](https://arxiv.org/abs/1709.07592), [UCF101](https://arxiv.org/abs/1212.0402) and [Taichi-HD](https://arxiv.org/abs/2003.00196). To prepare and download the datasets for evaluation, please refer to [this https URL](https://github.com/Vchitect/Latte/blob/main/docs/datasets_evaluation.md).
+**Highlights**: Latte is a latent diffusion transformer proposed as a backbone for modeling different modalities (trained for text-to-video generation here). It achieves state-of-the-art performance across four standard video benchmarks - [FaceForensics](https://huggingface.co/papers/1803.09179), [SkyTimelapse](https://huggingface.co/papers/1709.07592), [UCF101](https://huggingface.co/papers/1212.0402) and [Taichi-HD](https://huggingface.co/papers/2003.00196). To prepare and download the datasets for evaluation, please refer to [this https URL](https://github.com/Vchitect/Latte/blob/main/docs/datasets_evaluation.md).
 
 This pipeline was contributed by [maxin-cn](https://github.com/maxin-cn). The original codebase can be found [here](https://github.com/Vchitect/Latte). The original weights can be found under [hf.co/maxin-cn](https://huggingface.co/maxin-cn).
 

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

@@ -29,7 +29,7 @@ You can find additional information about LEDITS++ on the [project page](https:/
 </Tip>
 
 <Tip warning={true}>
-Due to some backward compatability issues with the current diffusers implementation of [`~schedulers.DPMSolverMultistepScheduler`] this implementation of LEdits++ can no longer guarantee perfect inversion.
+Due to some backward compatibility issues with the current diffusers implementation of [`~schedulers.DPMSolverMultistepScheduler`] this implementation of LEdits++ can no longer guarantee perfect inversion.
 This issue is unlikely to have any noticeable effects on applied use-cases. However, we provide an alternative implementation that guarantees perfect inversion in a dedicated [GitHub repo](https://github.com/ml-research/ledits_pp).
 </Tip>
 

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

@@ -12,125 +12,67 @@
 # See the License for the specific language governing permissions and
 # limitations under the License. -->
 
-# LTX Video
-
-<div class="flex flex-wrap space-x-1">
-  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
-  <img alt="MPS" src="https://img.shields.io/badge/MPS-000000?style=flat&logo=apple&logoColor=white%22">
+<div style="float: right;">
+  <div class="flex flex-wrap space-x-1">
+    <a href="https://huggingface.co/docs/diffusers/main/en/tutorials/using_peft_for_inference" target="_blank" rel="noopener">
+      <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+    </a>
+    <img alt="MPS" src="https://img.shields.io/badge/MPS-000000?style=flat&logo=apple&logoColor=white%22">
+  </div>
 </div>
 
-[LTX Video](https://huggingface.co/Lightricks/LTX-Video) is the first DiT-based video generation model capable of generating high-quality videos in real-time. It produces 24 FPS videos at a 768x512 resolution faster than they can be watched. Trained on a large-scale dataset of diverse videos, the model generates high-resolution videos with realistic and varied content. We provide a model for both text-to-video as well as image + text-to-video usecases.
+# LTX-Video
 
-<Tip>
+[LTX-Video](https://huggingface.co/Lightricks/LTX-Video) is a diffusion transformer designed for fast and real-time generation of high-resolution videos from text and images. The main feature of LTX-Video is the Video-VAE. The Video-VAE has a higher pixel to latent compression ratio (1:192) which enables more efficient video data processing and faster generation speed. To support and prevent finer details from being lost during generation, the Video-VAE decoder performs the latent to pixel conversion *and* the last denoising step.
 
-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 find all the original LTX-Video checkpoints under the [Lightricks](https://huggingface.co/Lightricks) organization.
 
-</Tip>
+> [!TIP]
+> Click on the LTX-Video models in the right sidebar for more examples of other video generation tasks.
 
-Available models:
+The example below demonstrates how to generate a video optimized for memory or inference speed.
 
-|  Model name   | Recommended dtype |
-|:-------------:|:-----------------:|
-| [`LTX Video 0.9.0`](https://huggingface.co/Lightricks/LTX-Video/blob/main/ltx-video-2b-v0.9.safetensors) | `torch.bfloat16` |
-| [`LTX Video 0.9.1`](https://huggingface.co/Lightricks/LTX-Video/blob/main/ltx-video-2b-v0.9.1.safetensors) | `torch.bfloat16` |
-| [`LTX Video 0.9.5`](https://huggingface.co/Lightricks/LTX-Video/blob/main/ltx-video-2b-v0.9.5.safetensors) | `torch.bfloat16` |
+<hfoptions id="usage">
+<hfoption id="memory">
 
-Note: The recommended dtype is for the transformer component. The VAE and text encoders can be either `torch.float32`, `torch.bfloat16` or `torch.float16` but the recommended dtype is `torch.bfloat16` as used in the original repository.
+Refer to the [Reduce memory usage](../../optimization/memory) guide for more details about the various memory saving techniques.
 
-## Loading Single Files
-
-Loading the original LTX Video checkpoints is also possible with [`~ModelMixin.from_single_file`]. We recommend using `from_single_file` for the Lightricks series of models, as they plan to release multiple models in the future in the single file format.
-
-```python
-import torch
-from diffusers import AutoencoderKLLTXVideo, LTXImageToVideoPipeline, LTXVideoTransformer3DModel
-
-# `single_file_url` could also be https://huggingface.co/Lightricks/LTX-Video/ltx-video-2b-v0.9.1.safetensors
-single_file_url = "https://huggingface.co/Lightricks/LTX-Video/ltx-video-2b-v0.9.safetensors"
-transformer = LTXVideoTransformer3DModel.from_single_file(
-  single_file_url, torch_dtype=torch.bfloat16
-)
-vae = AutoencoderKLLTXVideo.from_single_file(single_file_url, torch_dtype=torch.bfloat16)
-pipe = LTXImageToVideoPipeline.from_pretrained(
-  "Lightricks/LTX-Video", transformer=transformer, vae=vae, torch_dtype=torch.bfloat16
-)
-
-# ... inference code ...
-```
-
-Alternatively, the pipeline can be used to load the weights with [`~FromSingleFileMixin.from_single_file`].
-
-```python
-import torch
-from diffusers import LTXImageToVideoPipeline
-from transformers import T5EncoderModel, T5Tokenizer
-
-single_file_url = "https://huggingface.co/Lightricks/LTX-Video/ltx-video-2b-v0.9.safetensors"
-text_encoder = T5EncoderModel.from_pretrained(
-  "Lightricks/LTX-Video", subfolder="text_encoder", torch_dtype=torch.bfloat16
-)
-tokenizer = T5Tokenizer.from_pretrained(
-  "Lightricks/LTX-Video", subfolder="tokenizer", torch_dtype=torch.bfloat16
-)
-pipe = LTXImageToVideoPipeline.from_single_file(
-  single_file_url, text_encoder=text_encoder, tokenizer=tokenizer, torch_dtype=torch.bfloat16
-)
-```
-
-Loading [LTX GGUF checkpoints](https://huggingface.co/city96/LTX-Video-gguf) are also supported:
+The LTX-Video model below requires ~10GB of VRAM.
 
 ```py
 import torch
+from diffusers import LTXPipeline, AutoModel
+from diffusers.hooks import apply_group_offloading
 from diffusers.utils import export_to_video
-from diffusers import LTXPipeline, LTXVideoTransformer3DModel, GGUFQuantizationConfig
 
-ckpt_path = (
-    "https://huggingface.co/city96/LTX-Video-gguf/blob/main/ltx-video-2b-v0.9-Q3_K_S.gguf"
-)
-transformer = LTXVideoTransformer3DModel.from_single_file(
-    ckpt_path,
-    quantization_config=GGUFQuantizationConfig(compute_dtype=torch.bfloat16),
-    torch_dtype=torch.bfloat16,
-)
-pipe = LTXPipeline.from_pretrained(
+# fp8 layerwise weight-casting
+transformer = AutoModel.from_pretrained(
     "Lightricks/LTX-Video",
-    transformer=transformer,
-    torch_dtype=torch.bfloat16,
+    subfolder="transformer",
+    torch_dtype=torch.bfloat16
+)
+transformer.enable_layerwise_casting(
+    storage_dtype=torch.float8_e4m3fn, compute_dtype=torch.bfloat16
 )
-pipe.enable_model_cpu_offload()
 
-prompt = "A woman with long brown hair and light skin smiles at another woman with long blonde hair. The woman with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek. The camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be real-life footage"
+pipeline = LTXPipeline.from_pretrained("Lightricks/LTX-Video", transformer=transformer, torch_dtype=torch.bfloat16)
+
+# group-offloading
+onload_device = torch.device("cuda")
+offload_device = torch.device("cpu")
+pipeline.transformer.enable_group_offload(onload_device=onload_device, offload_device=offload_device, offload_type="leaf_level", use_stream=True)
+apply_group_offloading(pipeline.text_encoder, onload_device=onload_device, offload_type="block_level", num_blocks_per_group=2)
+apply_group_offloading(pipeline.vae, onload_device=onload_device, offload_type="leaf_level")
+
+prompt = """
+A woman with long brown hair and light skin smiles at another woman with long blonde hair.
+The woman with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek.
+The camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and 
+natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be real-life footage
+"""
 negative_prompt = "worst quality, inconsistent motion, blurry, jittery, distorted"
 
-video = pipe(
-    prompt=prompt,
-    negative_prompt=negative_prompt,
-    width=704,
-    height=480,
-    num_frames=161,
-    num_inference_steps=50,
-).frames[0]
-export_to_video(video, "output_gguf_ltx.mp4", fps=24)
-```
-
-Make sure to read the [documentation on GGUF](../../quantization/gguf) to learn more about our GGUF support.
-
-<!-- TODO(aryan): Update this when official weights are supported -->
-
-Loading and running inference with [LTX Video 0.9.1](https://huggingface.co/Lightricks/LTX-Video/blob/main/ltx-video-2b-v0.9.1.safetensors) weights.
-
-```python
-import torch
-from diffusers import LTXPipeline
-from diffusers.utils import export_to_video
-
-pipe = LTXPipeline.from_pretrained("a-r-r-o-w/LTX-Video-0.9.1-diffusers", torch_dtype=torch.bfloat16)
-pipe.to("cuda")
-
-prompt = "A woman with long brown hair and light skin smiles at another woman with long blonde hair. The woman with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek. The camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be real-life footage"
-negative_prompt = "worst quality, inconsistent motion, blurry, jittery, distorted"
-
-video = pipe(
+video = pipeline(
     prompt=prompt,
     negative_prompt=negative_prompt,
     width=768,
@@ -143,49 +85,306 @@ video = pipe(
 export_to_video(video, "output.mp4", fps=24)
 ```
 
-Refer to [this section](https://huggingface.co/docs/diffusers/main/en/api/pipelines/cogvideox#memory-optimization) to learn more about optimizing memory consumption.
+</hfoption>
+<hfoption id="inference speed">
 
-## Quantization
-
-Quantization helps reduce the memory requirements of very large models by storing model weights in a lower precision data type. However, quantization may have varying impact on video quality depending on the video model.
-
-Refer to the [Quantization](../../quantization/overview) overview to learn more about supported quantization backends and selecting a quantization backend that supports your use case. The example below demonstrates how to load a quantized [`LTXPipeline`] for inference with bitsandbytes.
+[Compilation](../../optimization/fp16#torchcompile) is slow the first time but subsequent calls to the pipeline are faster.
 
 ```py
 import torch
-from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig, LTXVideoTransformer3DModel, LTXPipeline
+from diffusers import LTXPipeline
 from diffusers.utils import export_to_video
-from transformers import BitsAndBytesConfig as BitsAndBytesConfig, T5EncoderModel
-
-quant_config = BitsAndBytesConfig(load_in_8bit=True)
-text_encoder_8bit = T5EncoderModel.from_pretrained(
-    "Lightricks/LTX-Video",
-    subfolder="text_encoder",
-    quantization_config=quant_config,
-    torch_dtype=torch.float16,
-)
-
-quant_config = DiffusersBitsAndBytesConfig(load_in_8bit=True)
-transformer_8bit = LTXVideoTransformer3DModel.from_pretrained(
-    "Lightricks/LTX-Video",
-    subfolder="transformer",
-    quantization_config=quant_config,
-    torch_dtype=torch.float16,
-)
 
 pipeline = LTXPipeline.from_pretrained(
-    "Lightricks/LTX-Video",
-    text_encoder=text_encoder_8bit,
-    transformer=transformer_8bit,
-    torch_dtype=torch.float16,
-    device_map="balanced",
+    "Lightricks/LTX-Video", torch_dtype=torch.bfloat16
 )
 
-prompt = "A detailed wooden toy ship with intricately carved masts and sails is seen gliding smoothly over a plush, blue carpet that mimics the waves of the sea. The ship's hull is painted a rich brown, with tiny windows. The carpet, soft and textured, provides a perfect backdrop, resembling an oceanic expanse. Surrounding the ship are various other toys and children's items, hinting at a playful environment. The scene captures the innocence and imagination of childhood, with the toy ship's journey symbolizing endless adventures in a whimsical, indoor setting."
-video = pipeline(prompt=prompt, num_frames=161, num_inference_steps=50).frames[0]
-export_to_video(video, "ship.mp4", fps=24)
+# torch.compile
+pipeline.transformer.to(memory_format=torch.channels_last)
+pipeline.transformer = torch.compile(
+    pipeline.transformer, mode="max-autotune", fullgraph=True
+)
+
+prompt = """
+A woman with long brown hair and light skin smiles at another woman with long blonde hair.
+The woman with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek.
+The camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and 
+natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be real-life footage
+"""
+negative_prompt = "worst quality, inconsistent motion, blurry, jittery, distorted"
+
+video = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    width=768,
+    height=512,
+    num_frames=161,
+    decode_timestep=0.03,
+    decode_noise_scale=0.025,
+    num_inference_steps=50,
+).frames[0]
+export_to_video(video, "output.mp4", fps=24)
 ```
 
+</hfoption>
+</hfoptions>
+
+## Notes
+
+- Refer to the following recommended settings for generation from the [LTX-Video](https://github.com/Lightricks/LTX-Video) repository.
+
+  - The recommended dtype for the transformer, VAE, and text encoder is `torch.bfloat16`. The VAE and text encoder can also be `torch.float32` or `torch.float16`.
+  - For guidance-distilled variants of LTX-Video, set `guidance_scale` to `1.0`. The `guidance_scale` for any other model should be set higher, like `5.0`, for good generation quality.
+  - For timestep-aware VAE variants (LTX-Video 0.9.1 and above), set `decode_timestep` to `0.05` and `image_cond_noise_scale` to `0.025`.
+  - For variants that support interpolation between multiple conditioning images and videos (LTX-Video 0.9.5 and above), use similar images and videos for the best results. Divergence from the conditioning inputs may lead to abrupt transitionts in the generated video.
+
+- LTX-Video 0.9.7 includes a spatial latent upscaler and a 13B parameter transformer. During inference, a low resolution video is quickly generated first and then upscaled and refined.
+
+  <details>
+  <summary>Show example code</summary>
+
+  ```py
+  import torch
+  from diffusers import LTXConditionPipeline, LTXLatentUpsamplePipeline
+  from diffusers.pipelines.ltx.pipeline_ltx_condition import LTXVideoCondition
+  from diffusers.utils import export_to_video, load_video
+
+  pipeline = LTXConditionPipeline.from_pretrained("Lightricks/LTX-Video-0.9.7-dev", torch_dtype=torch.bfloat16)
+  pipeline_upsample = LTXLatentUpsamplePipeline.from_pretrained("Lightricks/ltxv-spatial-upscaler-0.9.7", vae=pipeline.vae, torch_dtype=torch.bfloat16)
+  pipeline.to("cuda")
+  pipe_upsample.to("cuda")
+  pipeline.vae.enable_tiling()
+
+  def round_to_nearest_resolution_acceptable_by_vae(height, width):
+      height = height - (height % pipeline.vae_temporal_compression_ratio)
+      width = width - (width % pipeline.vae_temporal_compression_ratio)
+      return height, width
+
+  video = load_video(
+      "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cosmos/cosmos-video2world-input-vid.mp4"
+  )[:21]  # only use the first 21 frames as conditioning
+  condition1 = LTXVideoCondition(video=video, frame_index=0)
+
+  prompt = """
+  The video depicts a winding mountain road covered in snow, with a single vehicle 
+  traveling along it. The road is flanked by steep, rocky cliffs and sparse vegetation. 
+  The landscape is characterized by rugged terrain and a river visible in the distance. 
+  The scene captures the solitude and beauty of a winter drive through a mountainous region.
+  """
+  negative_prompt = "worst quality, inconsistent motion, blurry, jittery, distorted"
+  expected_height, expected_width = 768, 1152
+  downscale_factor = 2 / 3
+  num_frames = 161
+
+  # 1. Generate video at smaller resolution
+  # Text-only conditioning is also supported without the need to pass `conditions`
+  downscaled_height, downscaled_width = int(expected_height * downscale_factor), int(expected_width * downscale_factor)
+  downscaled_height, downscaled_width = round_to_nearest_resolution_acceptable_by_vae(downscaled_height, downscaled_width)
+  latents = pipeline(
+      conditions=[condition1],
+      prompt=prompt,
+      negative_prompt=negative_prompt,
+      width=downscaled_width,
+      height=downscaled_height,
+      num_frames=num_frames,
+      num_inference_steps=30,
+      decode_timestep=0.05,
+      decode_noise_scale=0.025,
+      image_cond_noise_scale=0.0,
+      guidance_scale=5.0,
+      guidance_rescale=0.7,
+      generator=torch.Generator().manual_seed(0),
+      output_type="latent",
+  ).frames
+
+  # 2. Upscale generated video using latent upsampler with fewer inference steps
+  # The available latent upsampler upscales the height/width by 2x
+  upscaled_height, upscaled_width = downscaled_height * 2, downscaled_width * 2
+  upscaled_latents = pipe_upsample(
+      latents=latents,
+      output_type="latent"
+  ).frames
+
+  # 3. Denoise the upscaled video with few steps to improve texture (optional, but recommended)
+  video = pipeline(
+      conditions=[condition1],
+      prompt=prompt,
+      negative_prompt=negative_prompt,
+      width=upscaled_width,
+      height=upscaled_height,
+      num_frames=num_frames,
+      denoise_strength=0.4,  # Effectively, 4 inference steps out of 10
+      num_inference_steps=10,
+      latents=upscaled_latents,
+      decode_timestep=0.05,
+      decode_noise_scale=0.025,
+      image_cond_noise_scale=0.0,
+      guidance_scale=5.0,
+      guidance_rescale=0.7,
+      generator=torch.Generator().manual_seed(0),
+      output_type="pil",
+  ).frames[0]
+
+  # 4. Downscale the video to the expected resolution
+  video = [frame.resize((expected_width, expected_height)) for frame in video]
+
+  export_to_video(video, "output.mp4", fps=24)
+  ```
+
+  </details>
+
+- LTX-Video 0.9.7 distilled model is guidance and timestep-distilled to speedup generation. It requires `guidance_scale` to be set to `1.0` and `num_inference_steps` should be set between `4` and `10` for good generation quality. You should also use the following custom timesteps for the best results.
+
+  - Base model inference to prepare for upscaling: `[1000, 993, 987, 981, 975, 909, 725, 0.03]`.
+  - Upscaling: `[1000, 909, 725, 421, 0]`.
+
+  <details>
+  <summary>Show example code</summary>
+
+  ```py
+  import torch
+  from diffusers import LTXConditionPipeline, LTXLatentUpsamplePipeline
+  from diffusers.pipelines.ltx.pipeline_ltx_condition import LTXVideoCondition
+  from diffusers.utils import export_to_video, load_video
+
+  pipeline = LTXConditionPipeline.from_pretrained("Lightricks/LTX-Video-0.9.7-distilled", torch_dtype=torch.bfloat16)
+  pipe_upsample = LTXLatentUpsamplePipeline.from_pretrained("Lightricks/ltxv-spatial-upscaler-0.9.7", vae=pipeline.vae, torch_dtype=torch.bfloat16)
+  pipeline.to("cuda")
+  pipe_upsample.to("cuda")
+  pipeline.vae.enable_tiling()
+
+  def round_to_nearest_resolution_acceptable_by_vae(height, width):
+      height = height - (height % pipeline.vae_temporal_compression_ratio)
+      width = width - (width % pipeline.vae_temporal_compression_ratio)
+      return height, width
+
+  prompt = """
+  artistic anatomical 3d render, utlra quality, human half full male body with transparent 
+  skin revealing structure instead of organs, muscular, intricate creative patterns, 
+  monochromatic with backlighting, lightning mesh, scientific concept art, blending biology 
+  with botany, surreal and ethereal quality, unreal engine 5, ray tracing, ultra realistic, 
+  16K UHD, rich details. camera zooms out in a rotating fashion
+  """
+  negative_prompt = "worst quality, inconsistent motion, blurry, jittery, distorted"
+  expected_height, expected_width = 768, 1152
+  downscale_factor = 2 / 3
+  num_frames = 161
+
+  # 1. Generate video at smaller resolution
+  downscaled_height, downscaled_width = int(expected_height * downscale_factor), int(expected_width * downscale_factor)
+  downscaled_height, downscaled_width = round_to_nearest_resolution_acceptable_by_vae(downscaled_height, downscaled_width)
+  latents = pipeline(
+      prompt=prompt,
+      negative_prompt=negative_prompt,
+      width=downscaled_width,
+      height=downscaled_height,
+      num_frames=num_frames,
+      timesteps=[1000, 993, 987, 981, 975, 909, 725, 0.03],
+      decode_timestep=0.05,
+      decode_noise_scale=0.025,
+      image_cond_noise_scale=0.0,
+      guidance_scale=1.0,
+      guidance_rescale=0.7,
+      generator=torch.Generator().manual_seed(0),
+      output_type="latent",
+  ).frames
+
+  # 2. Upscale generated video using latent upsampler with fewer inference steps
+  # The available latent upsampler upscales the height/width by 2x
+  upscaled_height, upscaled_width = downscaled_height * 2, downscaled_width * 2
+  upscaled_latents = pipe_upsample(
+      latents=latents,
+      adain_factor=1.0,
+      output_type="latent"
+  ).frames
+
+  # 3. Denoise the upscaled video with few steps to improve texture (optional, but recommended)
+  video = pipeline(
+      prompt=prompt,
+      negative_prompt=negative_prompt,
+      width=upscaled_width,
+      height=upscaled_height,
+      num_frames=num_frames,
+      denoise_strength=0.999,  # Effectively, 4 inference steps out of 5
+      timesteps=[1000, 909, 725, 421, 0],
+      latents=upscaled_latents,
+      decode_timestep=0.05,
+      decode_noise_scale=0.025,
+      image_cond_noise_scale=0.0,
+      guidance_scale=1.0,
+      guidance_rescale=0.7,
+      generator=torch.Generator().manual_seed(0),
+      output_type="pil",
+  ).frames[0]
+
+  # 4. Downscale the video to the expected resolution
+  video = [frame.resize((expected_width, expected_height)) for frame in video]
+
+  export_to_video(video, "output.mp4", fps=24)
+  ```
+
+  </details>
+
+- LTX-Video supports LoRAs with [`~loaders.LTXVideoLoraLoaderMixin.load_lora_weights`].
+
+  <details>
+  <summary>Show example code</summary>
+
+  ```py
+  import torch
+  from diffusers import LTXConditionPipeline
+  from diffusers.utils import export_to_video, load_image
+
+  pipeline = LTXConditionPipeline.from_pretrained(
+      "Lightricks/LTX-Video-0.9.5", torch_dtype=torch.bfloat16
+  )
+
+  pipeline.load_lora_weights("Lightricks/LTX-Video-Cakeify-LoRA", adapter_name="cakeify")
+  pipeline.set_adapters("cakeify")
+
+  # use "CAKEIFY" to trigger the LoRA
+  prompt = "CAKEIFY a person using a knife to cut a cake shaped like a Pikachu plushie"
+  image = load_image("https://huggingface.co/Lightricks/LTX-Video-Cakeify-LoRA/resolve/main/assets/images/pikachu.png")
+
+  video = pipeline(
+      prompt=prompt,
+      image=image,
+      width=576,
+      height=576,
+      num_frames=161,
+      decode_timestep=0.03,
+      decode_noise_scale=0.025,
+      num_inference_steps=50,
+  ).frames[0]
+  export_to_video(video, "output.mp4", fps=26)
+  ```
+
+  </details>
+
+- LTX-Video supports loading from single files, such as [GGUF checkpoints](../../quantization/gguf), with [`loaders.FromOriginalModelMixin.from_single_file`] or [`loaders.FromSingleFileMixin.from_single_file`].
+
+  <details>
+  <summary>Show example code</summary>
+
+  ```py
+  import torch
+  from diffusers.utils import export_to_video
+  from diffusers import LTXPipeline, AutoModel, GGUFQuantizationConfig
+
+  transformer = AutoModel.from_single_file(
+      "https://huggingface.co/city96/LTX-Video-gguf/blob/main/ltx-video-2b-v0.9-Q3_K_S.gguf"
+      quantization_config=GGUFQuantizationConfig(compute_dtype=torch.bfloat16),
+      torch_dtype=torch.bfloat16
+  )
+  pipeline = LTXPipeline.from_pretrained(
+      "Lightricks/LTX-Video",
+      transformer=transformer,
+      torch_dtype=torch.bfloat16
+  )
+  ```
+
+  </details>
+
 ## LTXPipeline
 
 [[autodoc]] LTXPipeline
@@ -204,6 +403,12 @@ export_to_video(video, "ship.mp4", fps=24)
   - all
   - __call__
 
+## LTXLatentUpsamplePipeline
+
+[[autodoc]] LTXLatentUpsamplePipeline
+  - all
+  - __call__
+
 ## LTXPipelineOutput
 
 [[autodoc]] pipelines.ltx.pipeline_output.LTXPipelineOutput

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

@@ -28,7 +28,7 @@ Lumina-Next has the following components:
 
 ---
 
-[Lumina-T2X: Transforming Text into Any Modality, Resolution, and Duration via Flow-based Large Diffusion Transformers](https://arxiv.org/abs/2405.05945) from Alpha-VLLM, OpenGVLab, Shanghai AI Laboratory.
+[Lumina-T2X: Transforming Text into Any Modality, Resolution, and Duration via Flow-based Large Diffusion Transformers](https://huggingface.co/papers/2405.05945) from Alpha-VLLM, OpenGVLab, Shanghai AI Laboratory.
 
 The abstract from the paper is:
 

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

@@ -15,7 +15,7 @@
 
 # OmniGen
 
-[OmniGen: Unified Image Generation](https://arxiv.org/pdf/2409.11340) from BAAI, by Shitao Xiao, Yueze Wang, Junjie Zhou, Huaying Yuan, Xingrun Xing, Ruiran Yan, Chaofan Li, Shuting Wang, Tiejun Huang, Zheng Liu.
+[OmniGen: Unified Image Generation](https://huggingface.co/papers/2409.11340) from BAAI, by Shitao Xiao, Yueze Wang, Junjie Zhou, Huaying Yuan, Xingrun Xing, Ruiran Yan, Chaofan Li, Shuting Wang, Tiejun Huang, Zheng Liu.
 
 The abstract from the paper is:
 

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

@@ -89,6 +89,7 @@ The table below lists all the pipelines currently available in 🤗 Diffusers an
 | [UniDiffuser](unidiffuser) | text2image, image2text, image variation, text variation, unconditional image generation, unconditional audio generation |
 | [Value-guided planning](value_guided_sampling) | value guided sampling |
 | [Wuerstchen](wuerstchen) | text2image |
+| [VisualCloze](visualcloze) | text2image, image2image, subject driven generation, inpainting, style transfer, image restoration, image editing, [depth,normal,edge,pose]2image, [depth,normal,edge,pose]-estimation, virtual try-on, image relighting |
 
 ## DiffusionPipeline
 

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

@@ -18,7 +18,7 @@ specific language governing permissions and limitations under the License.
 
 ## Overview
 
-[PIA: Your Personalized Image Animator via Plug-and-Play Modules in Text-to-Image Models](https://arxiv.org/abs/2312.13964) by Yiming Zhang, Zhening Xing, Yanhong Zeng, Youqing Fang, Kai Chen
+[PIA: Your Personalized Image Animator via Plug-and-Play Modules in Text-to-Image Models](https://huggingface.co/papers/2312.13964) by Yiming Zhang, Zhening Xing, Yanhong Zeng, Youqing Fang, Kai Chen
 
 Recent advancements in personalized text-to-image (T2I) models have revolutionized content creation, empowering non-experts to generate stunning images with unique styles. While promising, adding realistic motions into these personalized images by text poses significant challenges in preserving distinct styles, high-fidelity details, and achieving motion controllability by text. In this paper, we present PIA, a Personalized Image Animator that excels in aligning with condition images, achieving motion controllability by text, and the compatibility with various personalized T2I models without specific tuning. To achieve these goals, PIA builds upon a base T2I model with well-trained temporal alignment layers, allowing for the seamless transformation of any personalized T2I model into an image animation model. A key component of PIA is the introduction of the condition module, which utilizes the condition frame and inter-frame affinity as input to transfer appearance information guided by the affinity hint for individual frame synthesis in the latent space. This design mitigates the challenges of appearance-related image alignment within and allows for a stronger focus on aligning with motion-related guidance.
 
@@ -92,7 +92,7 @@ If you plan on using a scheduler that can clip samples, make sure to disable it
 
 ## Using FreeInit
 
-[FreeInit: Bridging Initialization Gap in Video Diffusion Models](https://arxiv.org/abs/2312.07537) by Tianxing Wu, Chenyang Si, Yuming Jiang, Ziqi Huang, Ziwei Liu.
+[FreeInit: Bridging Initialization Gap in Video Diffusion Models](https://huggingface.co/papers/2312.07537) by Tianxing Wu, Chenyang Si, Yuming Jiang, Ziqi Huang, Ziwei Liu.
 
 FreeInit is an effective method that improves temporal consistency and overall quality of videos generated using video-diffusion-models without any addition training. It can be applied to PIA, AnimateDiff, ModelScope, VideoCrafter and various other video generation models seamlessly at inference time, and works by iteratively refining the latent-initialization noise. More details can be found it the paper.
 

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

@@ -88,12 +88,46 @@ image.save("sana.png")
 
 Users can tweak the `max_timesteps` value for experimenting with the visual quality of the generated outputs. The default `max_timesteps` value was obtained with an inference-time search process. For more details about it, check out the paper.
 
+## Image to Image 
+
+The [`SanaSprintImg2ImgPipeline`] is a pipeline for image-to-image generation. It takes an input image and a prompt, and generates a new image based on the input image and the prompt.
+
+```py
+import torch
+from diffusers import SanaSprintImg2ImgPipeline
+from diffusers.utils.loading_utils import load_image
+
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/penguin.png"
+)
+
+pipe = SanaSprintImg2ImgPipeline.from_pretrained(
+    "Efficient-Large-Model/Sana_Sprint_1.6B_1024px_diffusers", 
+    torch_dtype=torch.bfloat16)
+pipe.to("cuda")
+
+image = pipe(
+    prompt="a cute pink bear", 
+    image=image, 
+    strength=0.5, 
+    height=832, 
+    width=480
+).images[0]
+image.save("output.png")
+```
+
 ## SanaSprintPipeline
 
 [[autodoc]] SanaSprintPipeline
   - all
   - __call__
 
+## SanaSprintImg2ImgPipeline
+
+[[autodoc]] SanaSprintImg2ImgPipeline
+  - all
+  - __call__
+
 
 ## SanaPipelineOutput
 

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

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # Stable Audio
 
-Stable Audio was proposed in [Stable Audio Open](https://arxiv.org/abs/2407.14358) by Zach Evans et al. . it takes a text prompt as input and predicts the corresponding sound or music sample.
+Stable Audio was proposed in [Stable Audio Open](https://huggingface.co/papers/2407.14358) by Zach Evans et al. . it takes a text prompt as input and predicts the corresponding sound or music sample.
 
 Stable Audio Open generates variable-length (up to 47s) stereo audio at 44.1kHz from text prompts. It comprises three components: an autoencoder that compresses waveforms into a manageable sequence length, a T5-based text embedding for text conditioning, and a transformer-based diffusion (DiT) model that operates in the latent space of the autoencoder.
 

docs/source/en/api/pipelines/stable_diffusion/adapter.md

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # T2I-Adapter
 
-[T2I-Adapter: Learning Adapters to Dig out More Controllable Ability for Text-to-Image Diffusion Models](https://arxiv.org/abs/2302.08453) by Chong Mou, Xintao Wang, Liangbin Xie, Jian Zhang, Zhongang Qi, Ying Shan, Xiaohu Qie.
+[T2I-Adapter: Learning Adapters to Dig out More Controllable Ability for Text-to-Image Diffusion Models](https://huggingface.co/papers/2302.08453) by Chong Mou, Xintao Wang, Liangbin Xie, Jian Zhang, Zhongang Qi, Ying Shan, Xiaohu Qie.
 
 Using the pretrained models we can provide control images (for example, a depth map) to control Stable Diffusion text-to-image generation so that it follows the structure of the depth image and fills in the details.
 

docs/source/en/api/pipelines/stable_diffusion/ldm3d_diffusion.md

@@ -19,7 +19,7 @@ specific language governing permissions and limitations under the License.
 LDM3D was proposed in [LDM3D: Latent Diffusion Model for 3D](https://huggingface.co/papers/2305.10853) by Gabriela Ben Melech Stan, Diana Wofk, Scottie Fox, Alex Redden, Will Saxton, Jean Yu, Estelle Aflalo, Shao-Yen Tseng, Fabio Nonato, Matthias Muller, and Vasudev Lal. LDM3D generates an image and a depth map from a given text prompt unlike the existing text-to-image diffusion models such as [Stable Diffusion](./overview) which only generates an image. With almost the same number of parameters, LDM3D achieves to create a latent space that can compress both the RGB images and the depth maps.
 
 Two checkpoints are available for use:
-- [ldm3d-original](https://huggingface.co/Intel/ldm3d). The original checkpoint used in the [paper](https://arxiv.org/pdf/2305.10853.pdf)
+- [ldm3d-original](https://huggingface.co/Intel/ldm3d). The original checkpoint used in the [paper](https://huggingface.co/papers/2305.10853)
 - [ldm3d-4c](https://huggingface.co/Intel/ldm3d-4c). The new version of LDM3D using 4 channels inputs instead of 6-channels inputs and finetuned on higher resolution images.
 
 
@@ -48,7 +48,7 @@ Make sure to check out the Stable Diffusion [Tips](overview#tips) section to lea
 
 # Upscaler
 
-[LDM3D-VR](https://arxiv.org/pdf/2311.03226.pdf) is an extended version of LDM3D.
+[LDM3D-VR](https://huggingface.co/papers/2311.03226) is an extended version of LDM3D.
 
 The abstract from the paper is:
 *Latent diffusion models have proven to be state-of-the-art in the creation and manipulation of visual outputs. However, as far as we know, the generation of depth maps jointly with RGB is still limited. We introduce LDM3D-VR, a suite of diffusion models targeting virtual reality development that includes LDM3D-pano and LDM3D-SR. These models enable the generation of panoramic RGBD based on textual prompts and the upscaling of low-resolution inputs to high-resolution RGBD, respectively. Our models are fine-tuned from existing pretrained models on datasets containing panoramic/high-resolution RGB images, depth maps and captions. Both models are evaluated in comparison to existing related methods*

docs/source/en/api/pipelines/stable_diffusion/stable_diffusion_3.md

@@ -17,7 +17,7 @@ specific language governing permissions and limitations under the License.
   <img alt="MPS" src="https://img.shields.io/badge/MPS-000000?style=flat&logo=apple&logoColor=white%22">
 </div>
 
-Stable Diffusion 3 (SD3) was proposed in [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://arxiv.org/pdf/2403.03206.pdf) by Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Muller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek, and Robin Rombach.
+Stable Diffusion 3 (SD3) was proposed in [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://huggingface.co/papers/2403.03206) by Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Muller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek, and Robin Rombach.
 
 The abstract from the paper is:
 

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

@@ -22,7 +22,7 @@ specific language governing permissions and limitations under the License.
   <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
 </div>
 
-[ModelScope Text-to-Video Technical Report](https://arxiv.org/abs/2308.06571) is by Jiuniu Wang, Hangjie Yuan, Dayou Chen, Yingya Zhang, Xiang Wang, Shiwei Zhang.
+[ModelScope Text-to-Video Technical Report](https://huggingface.co/papers/2308.06571) is by Jiuniu Wang, Hangjie Yuan, Dayou Chen, Yingya Zhang, Xiang Wang, Shiwei Zhang.
 
 The abstract from the paper is:
 

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

@@ -34,7 +34,7 @@ Our key modifications include (i) enriching the latent codes of the generated fr
 Experiments show that this leads to low overhead, yet high-quality and remarkably consistent video generation. Moreover, our approach is not limited to text-to-video synthesis but is also applicable to other tasks such as conditional and content-specialized video generation, and Video Instruct-Pix2Pix, i.e., instruction-guided video editing.
 As experiments show, our method performs comparably or sometimes better than recent approaches, despite not being trained on additional video data.*
 
-You can find additional information about Text2Video-Zero on the [project page](https://text2video-zero.github.io/), [paper](https://arxiv.org/abs/2303.13439), and [original codebase](https://github.com/Picsart-AI-Research/Text2Video-Zero).
+You can find additional information about Text2Video-Zero on the [project page](https://text2video-zero.github.io/), [paper](https://huggingface.co/papers/2303.13439), and [original codebase](https://github.com/Picsart-AI-Research/Text2Video-Zero).
 
 ## Usage example
 
@@ -55,9 +55,9 @@ result = [(r * 255).astype("uint8") for r in result]
 imageio.mimsave("video.mp4", result, fps=4)
 ```
 You can change these parameters in the pipeline call:
-* Motion field strength (see the [paper](https://arxiv.org/abs/2303.13439), Sect. 3.3.1):
+* Motion field strength (see the [paper](https://huggingface.co/papers/2303.13439), Sect. 3.3.1):
     * `motion_field_strength_x` and `motion_field_strength_y`. Default: `motion_field_strength_x=12`, `motion_field_strength_y=12`
-* `T` and `T'` (see the [paper](https://arxiv.org/abs/2303.13439), Sect. 3.3.1)
+* `T` and `T'` (see the [paper](https://huggingface.co/papers/2303.13439), Sect. 3.3.1)
     * `t0` and `t1` in the range `{0, ..., num_inference_steps}`. Default: `t0=45`, `t1=48`
 * Video length:
     * `video_length`, the number of frames video_length to be generated. Default: `video_length=8`

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

@@ -0,0 +1,300 @@
+<!--Copyright 2025 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.
+-->
+
+# VisualCloze
+
+[VisualCloze: A Universal Image Generation Framework via Visual In-Context Learning](https://huggingface.co/papers/2504.07960) is an innovative in-context learning based universal image generation framework that offers key capabilities:
+1. Support for various in-domain tasks
+2. Generalization to unseen tasks through in-context learning
+3. Unify multiple tasks into one step and generate both target image and intermediate results
+4. Support reverse-engineering conditions from target images
+
+## Overview
+
+The abstract from the paper is:
+
+*Recent progress in diffusion models significantly advances various image generation tasks. However, the current mainstream approach remains focused on building task-specific models, which have limited efficiency when supporting a wide range of different needs. While universal models attempt to address this limitation, they face critical challenges, including generalizable task instruction, appropriate task distributions, and unified architectural design. To tackle these challenges, we propose VisualCloze, a universal image generation framework, which supports a wide range of in-domain tasks, generalization to unseen ones, unseen unification of multiple tasks, and reverse generation. Unlike existing methods that rely on language-based task instruction, leading to task ambiguity and weak generalization, we integrate visual in-context learning, allowing models to identify tasks from visual demonstrations. Meanwhile, the inherent sparsity of visual task distributions hampers the learning of transferable knowledge across tasks. To this end, we introduce Graph200K, a graph-structured dataset that establishes various interrelated tasks, enhancing task density and transferable knowledge. Furthermore, we uncover that our unified image generation formulation shared a consistent objective with image infilling, enabling us to leverage the strong generative priors of pre-trained infilling models without modifying the architectures. The codes, dataset, and models are available at https://visualcloze.github.io.*
+
+## Inference
+
+### Model loading
+
+VisualCloze is a two-stage cascade pipeline, containing `VisualClozeGenerationPipeline` and `VisualClozeUpsamplingPipeline`.
+- In `VisualClozeGenerationPipeline`, each image is downsampled before concatenating images into a grid layout, avoiding excessively high resolutions. VisualCloze releases two models suitable for diffusers, i.e., [VisualClozePipeline-384](https://huggingface.co/VisualCloze/VisualClozePipeline-384) and [VisualClozePipeline-512](https://huggingface.co/VisualCloze/VisualClozePipeline-384), which downsample images to resolutions of 384 and 512, respectively. 
+- `VisualClozeUpsamplingPipeline` uses [SDEdit](https://huggingface.co/papers/2108.01073) to enable high-resolution image synthesis.
+
+The `VisualClozePipeline` integrates both stages to support convenient end-to-end sampling, while also allowing users to utilize each pipeline independently as needed.
+
+### Input Specifications
+
+#### Task and Content Prompts
+- Task prompt: Required to describe the generation task intention
+- Content prompt: Optional description or caption of the target image
+- When content prompt is not needed, pass `None`
+- For batch inference, pass `List[str|None]`
+
+#### Image Input Format
+- Format: `List[List[Image|None]]`
+- Structure:
+  - All rows except the last represent in-context examples
+  - Last row represents the current query (target image set to `None`)
+- For batch inference, pass `List[List[List[Image|None]]]`
+
+#### Resolution Control
+- Default behavior:
+  - Initial generation in the first stage: area of ${pipe.resolution}^2$
+  - Upsampling in the second stage: 3x factor
+- Custom resolution: Adjust using `upsampling_height` and `upsampling_width` parameters
+
+### Examples
+
+For comprehensive examples covering a wide range of tasks, please refer to the [Online Demo](https://huggingface.co/spaces/VisualCloze/VisualCloze) and [GitHub Repository](https://github.com/lzyhha/VisualCloze). Below are simple examples for three cases: mask-to-image conversion, edge detection, and subject-driven generation.
+
+#### Example for mask2image
+
+```python
+import torch
+from diffusers import VisualClozePipeline
+from diffusers.utils import load_image
+
+pipe = VisualClozePipeline.from_pretrained("VisualCloze/VisualClozePipeline-384", resolution=384, torch_dtype=torch.bfloat16)
+pipe.to("cuda")
+
+# Load in-context images (make sure the paths are correct and accessible)
+image_paths = [
+    # in-context examples
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_mask2image_incontext-example-1_mask.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_mask2image_incontext-example-1_image.jpg'),
+    ],
+    # query with the target image
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_mask2image_query_mask.jpg'),
+        None, # No image needed for the target image
+    ],
+]
+
+# Task and content prompt
+task_prompt = "In each row, a logical task is demonstrated to achieve [IMAGE2] an aesthetically pleasing photograph based on [IMAGE1] sam 2-generated masks with rich color coding."
+content_prompt = """Majestic photo of a golden eagle perched on a rocky outcrop in a mountainous landscape. 
+The eagle is positioned in the right foreground, facing left, with its sharp beak and keen eyes prominently visible. 
+Its plumage is a mix of dark brown and golden hues, with intricate feather details. 
+The background features a soft-focus view of snow-capped mountains under a cloudy sky, creating a serene and grandiose atmosphere. 
+The foreground includes rugged rocks and patches of green moss. Photorealistic, medium depth of field, 
+soft natural lighting, cool color palette, high contrast, sharp focus on the eagle, blurred background, 
+tranquil, majestic, wildlife photography."""
+
+# Run the pipeline
+image_result = pipe(
+    task_prompt=task_prompt,
+    content_prompt=content_prompt,
+    image=image_paths,
+    upsampling_width=1344,
+    upsampling_height=768,
+    upsampling_strength=0.4,
+    guidance_scale=30,
+    num_inference_steps=30,
+    max_sequence_length=512,
+    generator=torch.Generator("cpu").manual_seed(0)
+).images[0][0]
+
+# Save the resulting image
+image_result.save("visualcloze.png")
+```
+
+#### Example for edge-detection
+
+```python
+import torch
+from diffusers import VisualClozePipeline
+from diffusers.utils import load_image
+
+pipe = VisualClozePipeline.from_pretrained("VisualCloze/VisualClozePipeline-384", resolution=384, torch_dtype=torch.bfloat16)
+pipe.to("cuda")
+
+# Load in-context images (make sure the paths are correct and accessible)
+image_paths = [
+    # in-context examples
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_edgedetection_incontext-example-1_image.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_edgedetection_incontext-example-1_edge.jpg'),
+    ],
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_edgedetection_incontext-example-2_image.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_edgedetection_incontext-example-2_edge.jpg'),
+    ],
+    # query with the target image
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_edgedetection_query_image.jpg'),
+        None, # No image needed for the target image
+    ],
+]
+
+# Task and content prompt
+task_prompt = "Each row illustrates a pathway from [IMAGE1] a sharp and beautifully composed photograph to [IMAGE2] edge map with natural well-connected outlines using a clear logical task."
+content_prompt = ""
+
+# Run the pipeline
+image_result = pipe(
+    task_prompt=task_prompt,
+    content_prompt=content_prompt,
+    image=image_paths,
+    upsampling_width=864,
+    upsampling_height=1152,
+    upsampling_strength=0.4,
+    guidance_scale=30,
+    num_inference_steps=30,
+    max_sequence_length=512,
+    generator=torch.Generator("cpu").manual_seed(0)
+).images[0][0]
+
+# Save the resulting image
+image_result.save("visualcloze.png")
+```
+
+#### Example for subject-driven generation
+
+```python
+import torch
+from diffusers import VisualClozePipeline
+from diffusers.utils import load_image
+
+pipe = VisualClozePipeline.from_pretrained("VisualCloze/VisualClozePipeline-384", resolution=384, torch_dtype=torch.bfloat16)
+pipe.to("cuda")
+
+# Load in-context images (make sure the paths are correct and accessible)
+image_paths = [
+    # in-context examples
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_incontext-example-1_reference.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_incontext-example-1_depth.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_incontext-example-1_image.jpg'),
+    ],
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_incontext-example-2_reference.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_incontext-example-2_depth.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_incontext-example-2_image.jpg'),
+    ],
+    # query with the target image
+    [
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_query_reference.jpg'),
+        load_image('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_subjectdriven_query_depth.jpg'),
+        None, # No image needed for the target image
+    ],
+]
+
+# Task and content prompt
+task_prompt = """Each row describes a process that begins with [IMAGE1] an image containing the key object, 
+[IMAGE2] depth map revealing gray-toned spatial layers and results in 
+[IMAGE3] an image with artistic qualitya high-quality image with exceptional detail."""
+content_prompt = """A vintage porcelain collector's item. Beneath a blossoming cherry tree in early spring, 
+this treasure is photographed up close, with soft pink petals drifting through the air and vibrant blossoms framing the scene."""
+
+# Run the pipeline
+image_result = pipe(
+    task_prompt=task_prompt,
+    content_prompt=content_prompt,
+    image=image_paths,
+    upsampling_width=1024,
+    upsampling_height=1024,
+    upsampling_strength=0.2,
+    guidance_scale=30,
+    num_inference_steps=30,
+    max_sequence_length=512,
+    generator=torch.Generator("cpu").manual_seed(0)
+).images[0][0]
+
+# Save the resulting image
+image_result.save("visualcloze.png")
+```
+
+#### Utilize each pipeline independently 
+
+```python
+import torch
+from diffusers import VisualClozeGenerationPipeline, FluxFillPipeline as VisualClozeUpsamplingPipeline
+from diffusers.utils import load_image
+from PIL import Image
+
+pipe = VisualClozeGenerationPipeline.from_pretrained(
+    "VisualCloze/VisualClozePipeline-384", resolution=384, torch_dtype=torch.bfloat16
+)
+pipe.to("cuda")
+
+image_paths = [
+    # in-context examples
+    [
+        load_image(
+            "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_mask2image_incontext-example-1_mask.jpg"
+        ),
+        load_image(
+            "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_mask2image_incontext-example-1_image.jpg"
+        ),
+    ],
+    # query with the target image
+    [
+        load_image(
+            "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/visualcloze/visualcloze_mask2image_query_mask.jpg"
+        ),
+        None,  # No image needed for the target image
+    ],
+]
+task_prompt = "In each row, a logical task is demonstrated to achieve [IMAGE2] an aesthetically pleasing photograph based on [IMAGE1] sam 2-generated masks with rich color coding."
+content_prompt = "Majestic photo of a golden eagle perched on a rocky outcrop in a mountainous landscape. The eagle is positioned in the right foreground, facing left, with its sharp beak and keen eyes prominently visible. Its plumage is a mix of dark brown and golden hues, with intricate feather details. The background features a soft-focus view of snow-capped mountains under a cloudy sky, creating a serene and grandiose atmosphere. The foreground includes rugged rocks and patches of green moss. Photorealistic, medium depth of field, soft natural lighting, cool color palette, high contrast, sharp focus on the eagle, blurred background, tranquil, majestic, wildlife photography."
+
+# Stage 1: Generate initial image
+image = pipe(
+    task_prompt=task_prompt,
+    content_prompt=content_prompt,
+    image=image_paths,
+    guidance_scale=30,
+    num_inference_steps=30,
+    max_sequence_length=512,
+    generator=torch.Generator("cpu").manual_seed(0),
+).images[0][0]
+
+# Stage 2 (optional): Upsample the generated image
+pipe_upsample = VisualClozeUpsamplingPipeline.from_pipe(pipe)
+pipe_upsample.to("cuda")
+
+mask_image = Image.new("RGB", image.size, (255, 255, 255))
+
+image = pipe_upsample(
+    image=image,
+    mask_image=mask_image,
+    prompt=content_prompt,
+    width=1344,
+    height=768,
+    strength=0.4,
+    guidance_scale=30,
+    num_inference_steps=30,
+    max_sequence_length=512,
+    generator=torch.Generator("cpu").manual_seed(0),
+).images[0]
+
+image.save("visualcloze.png")
+```
+
+## VisualClozePipeline
+
+[[autodoc]] VisualClozePipeline
+  - all
+  - __call__
+
+## VisualClozeGenerationPipeline
+
+[[autodoc]] VisualClozeGenerationPipeline
+  - all
+  - __call__

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

@@ -12,128 +12,170 @@
 # See the License for the specific language governing permissions and
 # limitations under the License. -->
 
-# Wan
-
-<div class="flex flex-wrap space-x-1">
-  <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+<div style="float: right;">
+  <div class="flex flex-wrap space-x-1">
+    <a href="https://huggingface.co/docs/diffusers/main/en/tutorials/using_peft_for_inference" target="_blank" rel="noopener">
+      <img alt="LoRA" src="https://img.shields.io/badge/LoRA-d8b4fe?style=flat"/>
+    </a>
+  </div>
 </div>
 
-[Wan 2.1](https://github.com/Wan-Video/Wan2.1) by the Alibaba Wan Team.
+# Wan2.1
 
-<!-- TODO(aryan): update abstract once paper is out -->
+[Wan-2.1](https://huggingface.co/papers/2503.20314) by the Wan Team.
 
-## Generating Videos with Wan 2.1
+*This report presents Wan, a comprehensive and open suite of video foundation models designed to push the boundaries of video generation. Built upon the mainstream diffusion transformer paradigm, Wan achieves significant advancements in generative capabilities through a series of innovations, including our novel VAE, scalable pre-training strategies, large-scale data curation, and automated evaluation metrics. These contributions collectively enhance the model's performance and versatility. Specifically, Wan is characterized by four key features: Leading Performance: The 14B model of Wan, trained on a vast dataset comprising billions of images and videos, demonstrates the scaling laws of video generation with respect to both data and model size. It consistently outperforms the existing open-source models as well as state-of-the-art commercial solutions across multiple internal and external benchmarks, demonstrating a clear and significant performance superiority. Comprehensiveness: Wan offers two capable models, i.e., 1.3B and 14B parameters, for efficiency and effectiveness respectively. It also covers multiple downstream applications, including image-to-video, instruction-guided video editing, and personal video generation, encompassing up to eight tasks. Consumer-Grade Efficiency: The 1.3B model demonstrates exceptional resource efficiency, requiring only 8.19 GB VRAM, making it compatible with a wide range of consumer-grade GPUs. Openness: We open-source the entire series of Wan, including source code and all models, with the goal of fostering the growth of the video generation community. This openness seeks to significantly expand the creative possibilities of video production in the industry and provide academia with high-quality video foundation models. All the code and models are available at [this https URL](https://github.com/Wan-Video/Wan2.1).*
 
-We will first need to install some additional dependencies.
+You can find all the original Wan2.1 checkpoints under the [Wan-AI](https://huggingface.co/Wan-AI) organization.
 
-```shell
-pip install -u ftfy imageio-ffmpeg imageio
-```
+The following Wan models are supported in Diffusers:
+- [Wan 2.1 T2V 1.3B](https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B-Diffusers)
+- [Wan 2.1 T2V 14B](https://huggingface.co/Wan-AI/Wan2.1-T2V-14B-Diffusers)
+- [Wan 2.1 I2V 14B - 480P](https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-480P-Diffusers)
+- [Wan 2.1 I2V 14B - 720P](https://huggingface.co/Wan-AI/Wan2.1-I2V-14B-720P-Diffusers)
+- [Wan 2.1 FLF2V 14B - 720P](https://huggingface.co/Wan-AI/Wan2.1-FLF2V-14B-720P-diffusers)
+- [Wan 2.1 VACE 1.3B](https://huggingface.co/Wan-AI/Wan2.1-VACE-1.3B-diffusers)
+- [Wan 2.1 VACE 14B](https://huggingface.co/Wan-AI/Wan2.1-VACE-14B-diffusers)
 
-### Text to Video Generation
+> [!TIP]
+> Click on the Wan2.1 models in the right sidebar for more examples of 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.
+### Text-to-Video Generation
 
-```python
-from diffusers import WanPipeline
-from diffusers.utils import export_to_video
+The example below demonstrates how to generate a video from text optimized for memory or inference speed.
 
-# 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"
+<hfoptions id="T2V usage">
+<hfoption id="T2V memory">
 
-pipe = WanPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16)
-pipe.enable_model_cpu_offload()
+Refer to the [Reduce memory usage](../../optimization/memory) guide for more details about the various memory saving techniques.
 
-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
+The Wan2.1 text-to-video model below requires ~13GB of VRAM.
 
-frames = pipe(prompt=prompt, negative_prompt=negative_prompt, num_frames=num_frames).frames[0]
-export_to_video(frames, "wan-t2v.mp4", fps=16)
-```
-
-<Tip>
-You can improve the quality of the generated video by running the decoding step in full precision.
-</Tip>
-
-```python
-from diffusers import WanPipeline, AutoencoderKLWan
-from diffusers.utils import export_to_video
-
-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
+```py
+# pip install ftfy
 import torch
 import numpy as np
-from diffusers import AutoencoderKLWan, WanImageToVideoPipeline
+from diffusers import AutoModel, WanPipeline
+from diffusers.quantizers import PipelineQuantizationConfig
+from diffusers.hooks.group_offloading import apply_group_offloading
 from diffusers.utils import export_to_video, load_image
-from transformers import CLIPVisionModel
+from transformers import UMT5EncoderModel
 
-# 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
+text_encoder = UMT5EncoderModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", torch_dtype=torch.float32)
+transformer = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="transformer", torch_dtype=torch.bfloat16)
+
+# group-offloading
+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
 )
-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
+transformer.enable_group_offload(
+    onload_device=onload_device,
+    offload_device=offload_device,
+    offload_type="leaf_level",
+    use_stream=True
 )
 
-# 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"
+pipeline = WanPipeline.from_pretrained(
+    "Wan-AI/Wan2.1-T2V-14B-Diffusers",
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    torch_dtype=torch.bfloat16
 )
+pipeline.to("cuda")
 
-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 = """
+The camera rushes from far to near in a low-angle shot, 
+revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic 
+shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+"""
+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
+"""
 
-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,
+output = pipeline(
     prompt=prompt,
     negative_prompt=negative_prompt,
-    height=height,
-    width=width,
-    num_frames=num_frames,
+    num_frames=81,
     guidance_scale=5.0,
 ).frames[0]
-export_to_video(output, "wan-i2v.mp4", fps=16)
+export_to_video(output, "output.mp4", fps=16)
 ```
 
-### First and Last Frame Interpolation
+</hfoption>
+<hfoption id="T2V inference speed">
+
+[Compilation](../../optimization/fp16#torchcompile) is slow the first time but subsequent calls to the pipeline are faster.
+
+```py
+# pip install ftfy
+import torch
+import numpy as np
+from diffusers import AutoModel, WanPipeline
+from diffusers.hooks.group_offloading import apply_group_offloading
+from diffusers.utils import export_to_video, load_image
+from transformers import UMT5EncoderModel
+
+text_encoder = UMT5EncoderModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", torch_dtype=torch.float32)
+transformer = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="transformer", torch_dtype=torch.bfloat16)
+
+pipeline = WanPipeline.from_pretrained(
+    "Wan-AI/Wan2.1-T2V-14B-Diffusers",
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    torch_dtype=torch.bfloat16
+)
+pipeline.to("cuda")
+
+# torch.compile
+pipeline.transformer.to(memory_format=torch.channels_last)
+pipeline.transformer = torch.compile(
+    pipeline.transformer, mode="max-autotune", fullgraph=True
+)
+
+prompt = """
+The camera rushes from far to near in a low-angle shot, 
+revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic 
+shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+"""
+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
+"""
+
+output = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_frames=81,
+    guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "output.mp4", fps=16)
+```
+
+</hfoption>
+</hfoptions>
+
+### First-Last-Frame-to-Video Generation
+
+The example below demonstrates how to use the image-to-video pipeline to generate a video using a text description, a starting frame, and an ending frame.
+
+<hfoptions id="FLF2V usage">
+<hfoption id="usage">
 
 ```python
 import numpy as np
@@ -166,13 +208,13 @@ def aspect_ratio_resize(image, pipe, max_area=720 * 1280):
 def center_crop_resize(image, height, width):
     # Calculate resize ratio to match first frame dimensions
     resize_ratio = max(width / image.width, height / image.height)
-    
+
     # Resize the image
     width = round(image.width * resize_ratio)
     height = round(image.height * resize_ratio)
     size = [width, height]
     image = TF.center_crop(image, size)
-    
+
     return image, height, width
 
 first_frame, height, width = aspect_ratio_resize(first_frame, pipe)
@@ -187,320 +229,103 @@ output = pipe(
 export_to_video(output, "output.mp4", fps=16)
 ```
 
-### Video to Video Generation
+</hfoption>
+</hfoptions>
 
-```python
-import torch
-from diffusers.utils import load_video, export_to_video
-from diffusers import AutoencoderKLWan, WanVideoToVideoPipeline, UniPCMultistepScheduler
+### Any-to-Video Controllable Generation
 
-# 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()
+Wan VACE supports various generation techniques which achieve controllable video generation. Some of the capabilities include:
+- Control to Video (Depth, Pose, Sketch, Flow, Grayscale, Scribble, Layout, Boundary Box, etc.). Recommended library for preprocessing videos to obtain control videos: [huggingface/controlnet_aux]()
+- Image/Video to Video (first frame, last frame, starting clip, ending clip, random clips)
+- Inpainting and Outpainting
+- Subject to Video (faces, object, characters, etc.)
+- Composition to Video (reference anything, animate anything, swap anything, expand anything, move anything, etc.)
 
-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]
+The code snippets available in [this](https://github.com/huggingface/diffusers/pull/11582) pull request demonstrate some examples of how videos can be generated with controllability signals.
 
-export_to_video(output, "wan-v2v.mp4", fps=16)
-```
+The general rule of thumb to keep in mind when preparing inputs for the VACE pipeline is that the input images, or frames of a video that you want to use for conditioning, should have a corresponding mask that is black in color. The black mask signifies that the model will not generate new content for that area, and only use those parts for conditioning the generation process. For parts/frames that should be generated by the model, the mask should be white in color.
 
-## Memory Optimizations for Wan 2.1
+## Notes
 
-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.
+- Wan2.1 supports LoRAs with [`~loaders.WanLoraLoaderMixin.load_lora_weights`].
 
-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.
+  <details>
+  <summary>Show example code</summary>
 
-### Group Offloading the Transformer and UMT5 Text Encoder
+  ```py
+  # pip install ftfy
+  import torch
+  from diffusers import AutoModel, WanPipeline
+  from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
+  from diffusers.utils import export_to_video
 
-Find more information about group offloading [here](../optimization/memory.md)
+  vae = AutoModel.from_pretrained(
+      "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", subfolder="vae", torch_dtype=torch.float32
+  )
+  pipeline = WanPipeline.from_pretrained(
+      "Wan-AI/Wan2.1-T2V-1.3B-Diffusers", vae=vae, torch_dtype=torch.bfloat16
+  )
+  pipeline.scheduler = UniPCMultistepScheduler.from_config(
+      pipeline.scheduler.config, flow_shift=5.0
+  )
+  pipeline.to("cuda")
 
-#### Block Level Group Offloading
+  pipeline.load_lora_weights("benjamin-paine/steamboat-willie-1.3b", adapter_name="steamboat-willie")
+  pipeline.set_adapters("steamboat-willie")
 
-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`.
+  pipeline.enable_model_cpu_offload()
 
-The following example will now only require 14GB of VRAM to run, but will take approximately 30 minutes to generate a video.
+  # use "steamboat willie style" to trigger the LoRA
+  prompt = """
+  steamboat willie style, golden era animation, The camera rushes from far to near in a low-angle shot, 
+  revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+  for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+  Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic 
+  shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+  """
 
-```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
+  output = pipeline(
+      prompt=prompt,
+      num_frames=81,
+      guidance_scale=5.0,
+  ).frames[0]
+  export_to_video(output, "output.mp4", fps=16)
+  ```
 
-# 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
-)
+  </details>
 
-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)
+- [`WanTransformer3DModel`] and [`AutoencoderKLWan`] supports loading from single files with [`~loaders.FromSingleFileMixin.from_single_file`].
 
-onload_device = torch.device("cuda")
-offload_device = torch.device("cpu")
+  <details>
+  <summary>Show example code</summary>
 
-apply_group_offloading(text_encoder,
-    onload_device=onload_device,
-    offload_device=offload_device,
-    offload_type="block_level",
-    num_blocks_per_group=4
-)
+  ```py
+  # pip install ftfy
+  import torch
+  from diffusers import WanPipeline, AutoModel
 
-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")
+  vae = AutoModel.from_single_file(
+      "https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors"
+  )
+  transformer = AutoModel.from_single_file(
+      "https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/diffusion_models/wan2.1_t2v_1.3B_bf16.safetensors",
+      torch_dtype=torch.bfloat16
+  )
+  pipeline = WanPipeline.from_pretrained(
+      "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
+      vae=vae,
+      transformer=transformer,
+      torch_dtype=torch.bfloat16
+  )
+  ```
 
-image = load_image(
-    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/astronaut.jpg"
-)
+  </details>
 
-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))
+- Set the [`AutoencoderKLWan`] dtype to `torch.float32` for better decoding quality.
 
-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"
+- The number of frames per second (fps) or `k` should be calculated by `4 * k + 1`.
 
-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:
-
-```python
-from diffusers import FlowMatchEulerDiscreteScheduler, UniPCMultistepScheduler, WanPipeline
-
-scheduler_a = FlowMatchEulerDiscreteScheduler(shift=5.0)
-scheduler_b = UniPCMultistepScheduler(prediction_type="flow_prediction", use_flow_sigmas=True, flow_shift=4.0)
-
-pipe = WanPipeline.from_pretrained("Wan-AI/Wan2.1-T2V-1.3B-Diffusers", scheduler=<CUSTOM_SCHEDULER_HERE>)
-
-# or,
-pipe.scheduler = <CUSTOM_SCHEDULER_HERE>
-```
-
-## 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
-from diffusers import WanPipeline, WanTransformer3DModel
-
-ckpt_path = "https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/diffusion_models/wan2.1_t2v_1.3B_bf16.safetensors"
-transformer = WanTransformer3DModel.from_single_file(ckpt_path, torch_dtype=torch.bfloat16)
-
-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.
+- Try lower `shift` values (`2.0` to `5.0`) for lower resolution videos and higher `shift` values (`7.0` to `12.0`) for higher resolution images.
 
 ## WanPipeline
 
@@ -514,6 +339,18 @@ pipe = WanPipeline.from_pretrained("Wan-AI/Wan2.1-T2V-1.3B-Diffusers", transform
   - all
   - __call__
 
+## WanVACEPipeline
+
+[[autodoc]] WanVACEPipeline
+  - all
+  - __call__
+
+## WanVideoToVideoPipeline
+
+[[autodoc]] WanVideoToVideoPipeline
+  - all
+  - __call__
+
 ## WanPipelineOutput
 
-[[autodoc]] pipelines.wan.pipeline_output.WanPipelineOutput
+[[autodoc]] pipelines.wan.pipeline_output.WanPipelineOutput

docs/source/en/api/quantization.md

@@ -13,9 +13,7 @@ specific language governing permissions and limitations under the License.
 
 # Quantization
 
-Quantization techniques reduce memory and computational costs by representing weights and activations with lower-precision data types like 8-bit integers (int8). This enables loading larger models you normally wouldn't be able to fit into memory, and speeding up inference. Diffusers supports 8-bit and 4-bit quantization with [bitsandbytes](https://huggingface.co/docs/bitsandbytes/en/index).
-
-Quantization techniques that aren't supported in Transformers can be added with the [`DiffusersQuantizer`] class.
+Quantization techniques reduce memory and computational costs by representing weights and activations with lower-precision data types like 8-bit integers (int8). This enables loading larger models you normally wouldn't be able to fit into memory, and speeding up inference.
 
 <Tip>
 
@@ -23,6 +21,9 @@ Learn how to quantize models in the [Quantization](../quantization/overview) gui
 
 </Tip>
 
+## PipelineQuantizationConfig
+
+[[autodoc]] quantizers.PipelineQuantizationConfig
 
 ## BitsAndBytesConfig
 

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

@@ -13,7 +13,7 @@ specific language governing permissions and limitations under the License.
 # CosineDPMSolverMultistepScheduler
 
 The [`CosineDPMSolverMultistepScheduler`] is a variant of [`DPMSolverMultistepScheduler`] with cosine schedule, proposed by Nichol and Dhariwal (2021).
-It is being used in the [Stable Audio Open](https://arxiv.org/abs/2407.14358) paper and the [Stability-AI/stable-audio-tool](https://github.com/Stability-AI/stable-audio-tool) codebase.
+It is being used in the [Stable Audio Open](https://huggingface.co/papers/2407.14358) paper and the [Stability-AI/stable-audio-tool](https://github.com/Stability-AI/stable-audio-tools) codebase.
 
 This scheduler was contributed by [Yoach Lacombe](https://huggingface.co/ylacombe).
 

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

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # FlowMatchEulerDiscreteScheduler
 
-`FlowMatchEulerDiscreteScheduler` is based on the flow-matching sampling introduced in [Stable Diffusion 3](https://arxiv.org/abs/2403.03206).
+`FlowMatchEulerDiscreteScheduler` is based on the flow-matching sampling introduced in [Stable Diffusion 3](https://huggingface.co/papers/2403.03206).
 
 ## FlowMatchEulerDiscreteScheduler
 [[autodoc]] FlowMatchEulerDiscreteScheduler

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

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # FlowMatchHeunDiscreteScheduler
 
-`FlowMatchHeunDiscreteScheduler` is based on the flow-matching sampling introduced in [EDM](https://arxiv.org/abs/2403.03206).
+`FlowMatchHeunDiscreteScheduler` is based on the flow-matching sampling introduced in [EDM](https://huggingface.co/papers/2403.03206).
 
 ## FlowMatchHeunDiscreteScheduler
 [[autodoc]] FlowMatchHeunDiscreteScheduler

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

@@ -14,7 +14,7 @@ specific language governing permissions and limitations under the License.
 
 ## Overview
 
-Multistep and onestep scheduler (Algorithm 3) introduced alongside latent consistency models in the paper [Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference](https://arxiv.org/abs/2310.04378) by Simian Luo, Yiqin Tan, Longbo Huang, Jian Li, and Hang Zhao.
+Multistep and onestep scheduler (Algorithm 3) introduced alongside latent consistency models in the paper [Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference](https://huggingface.co/papers/2310.04378) by Simian Luo, Yiqin Tan, Longbo Huang, Jian Li, and Hang Zhao.
 This scheduler should be able to generate good samples from [`LatentConsistencyModelPipeline`] in 1-8 steps.
 
 ## LCMScheduler

docs/source/en/conceptual/ethical_guidelines.md

@@ -54,7 +54,7 @@ The team works daily to make the technical and non-technical tools available to
 
 - **Encouraging safety in deployment**
 
-  - [**Safe Stable Diffusion**](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_safe): It mitigates the well-known issue that models, like Stable Diffusion, that are trained on unfiltered, web-crawled datasets tend to suffer from inappropriate degeneration. Related paper: [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://arxiv.org/abs/2211.05105).
+  - [**Safe Stable Diffusion**](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_safe): It mitigates the well-known issue that models, like Stable Diffusion, that are trained on unfiltered, web-crawled datasets tend to suffer from inappropriate degeneration. Related paper: [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://huggingface.co/papers/2211.05105).
 
   - [**Safety Checker**](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py): It checks and compares the class probability of a set of hard-coded harmful concepts in the embedding space against an image after it has been generated. The harmful concepts are intentionally hidden to prevent reverse engineering of the checker.
 

docs/source/en/conceptual/evaluation.md

@@ -18,8 +18,8 @@ specific language governing permissions and limitations under the License.
 
 > [!TIP]
 > This document has now grown outdated given the emergence of existing evaluation frameworks for diffusion models for image generation. Please check
-> out works like [HEIM](https://crfm.stanford.edu/helm/heim/latest/), [T2I-Compbench](https://arxiv.org/abs/2307.06350),
-> [GenEval](https://arxiv.org/abs/2310.11513).
+> out works like [HEIM](https://crfm.stanford.edu/helm/heim/latest/), [T2I-Compbench](https://huggingface.co/papers/2307.06350),
+> [GenEval](https://huggingface.co/papers/2310.11513).
 
 Evaluation of generative models like [Stable Diffusion](https://huggingface.co/docs/diffusers/stable_diffusion) is subjective in nature. But as practitioners and researchers, we often have to make careful choices amongst many different possibilities. So, when working with different generative models (like GANs, Diffusion, etc.), how do we choose one over the other?
 
@@ -122,7 +122,7 @@ In this section, we will walk you through how to evaluate three different diffus
 
 ### Text-guided image generation
 
-[CLIP score](https://arxiv.org/abs/2104.08718) measures the compatibility of image-caption pairs. Higher CLIP scores imply higher compatibility 🔼. The CLIP score is a quantitative measurement of the qualitative concept "compatibility". Image-caption pair compatibility can also be thought of as the semantic similarity between the image and the caption. CLIP score was found to have high correlation with human judgement.
+[CLIP score](https://huggingface.co/papers/2104.08718) measures the compatibility of image-caption pairs. Higher CLIP scores imply higher compatibility 🔼. The CLIP score is a quantitative measurement of the qualitative concept "compatibility". Image-caption pair compatibility can also be thought of as the semantic similarity between the image and the caption. CLIP score was found to have high correlation with human judgement.
 
 Let's first load a [`StableDiffusionPipeline`]:
 
@@ -222,7 +222,7 @@ Here is one example:
 
 ![edit-instruction](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/evaluation_diffusion_models/edit-instruction.png)
 
-One strategy to evaluate such a model is to measure the consistency of the change between the two images (in [CLIP](https://huggingface.co/docs/transformers/model_doc/clip) space) with the change between the two image captions (as shown in [CLIP-Guided Domain Adaptation of Image Generators](https://arxiv.org/abs/2108.00946)). This is referred to as the "**CLIP directional similarity**".
+One strategy to evaluate such a model is to measure the consistency of the change between the two images (in [CLIP](https://huggingface.co/docs/transformers/model_doc/clip) space) with the change between the two image captions (as shown in [CLIP-Guided Domain Adaptation of Image Generators](https://huggingface.co/papers/2108.00946)). This is referred to as the "**CLIP directional similarity**".
 
 - Caption 1 corresponds to the input image (image 1) that is to be edited.
 - Caption 2 corresponds to the edited image (image 2). It should reflect the edit instruction.
@@ -433,7 +433,7 @@ Both CLIP score and CLIP direction similarity rely on the CLIP model, which can
 
 ### Class-conditioned image generation
 
-Class-conditioned generative models are usually pre-trained on a class-labeled dataset such as [ImageNet-1k](https://huggingface.co/datasets/imagenet-1k). Popular metrics for evaluating these models include Fréchet Inception Distance (FID), Kernel Inception Distance (KID), and Inception Score (IS). In this document, we focus on FID ([Heusel et al.](https://arxiv.org/abs/1706.08500)). We show how to compute it with the [`DiTPipeline`](https://huggingface.co/docs/diffusers/api/pipelines/dit), which uses the [DiT model](https://arxiv.org/abs/2212.09748) under the hood.
+Class-conditioned generative models are usually pre-trained on a class-labeled dataset such as [ImageNet-1k](https://huggingface.co/datasets/imagenet-1k). Popular metrics for evaluating these models include Fréchet Inception Distance (FID), Kernel Inception Distance (KID), and Inception Score (IS). In this document, we focus on FID ([Heusel et al.](https://huggingface.co/papers/1706.08500)). We show how to compute it with the [`DiTPipeline`](https://huggingface.co/docs/diffusers/api/pipelines/dit), which uses the [DiT model](https://huggingface.co/papers/2212.09748) under the hood.
 
 FID aims to measure how similar are two datasets of images. As per [this resource](https://mmgeneration.readthedocs.io/en/latest/quick_run.html#fid):
 

docs/source/en/optimization/cache.md

@@ -0,0 +1,69 @@
+<!-- Copyright 2024 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. -->
+
+# Caching
+
+Caching accelerates inference by storing and reusing intermediate outputs of different layers, such as attention and feedforward layers, instead of performing the entire computation at each inference step. It significantly improves generation speed at the expense of more memory and doesn't require additional training.
+
+This guide shows you how to use the caching methods supported in Diffusers.
+
+## Pyramid Attention Broadcast
+
+[Pyramid Attention Broadcast (PAB)](https://huggingface.co/papers/2408.12588) is based on the observation that attention outputs aren't that different between successive timesteps of the generation process. The attention differences are smallest in the cross attention layers and are generally cached over a longer timestep range. This is followed by temporal attention and spatial attention layers.
+
+> [!TIP]
+> Not all video models have three types of attention (cross, temporal, and spatial)!
+
+PAB can be combined with other techniques like sequence parallelism and classifier-free guidance parallelism (data parallelism) for near real-time video generation.
+
+Set up and pass a [`PyramidAttentionBroadcastConfig`] to a pipeline's transformer to enable it. The `spatial_attention_block_skip_range` controls how often to skip attention calculations in the spatial attention blocks and the `spatial_attention_timestep_skip_range` is the range of timesteps to skip. Take care to choose an appropriate range because a smaller interval can lead to slower inference speeds and a larger interval can result in lower generation quality.
+
+```python
+import torch
+from diffusers import CogVideoXPipeline, PyramidAttentionBroadcastConfig
+
+pipeline = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b", torch_dtype=torch.bfloat16)
+pipeline.to("cuda")
+
+config = PyramidAttentionBroadcastConfig(
+    spatial_attention_block_skip_range=2,
+    spatial_attention_timestep_skip_range=(100, 800),
+    current_timestep_callback=lambda: pipe.current_timestep,
+)
+pipeline.transformer.enable_cache(config)
+```
+
+## FasterCache
+
+[FasterCache](https://huggingface.co/papers/2410.19355) caches and reuses attention features similar to [PAB](#pyramid-attention-broadcast) since output differences are small for each successive timestep.
+
+This method may also choose to skip the unconditional branch prediction, when using classifier-free guidance for sampling (common in most base models), and estimate it from the conditional branch prediction if there is significant redundancy in the predicted latent outputs between successive timesteps.
+
+Set up and pass a [`FasterCacheConfig`] to a pipeline's transformer to enable it.
+
+```python
+import torch
+from diffusers import CogVideoXPipeline, FasterCacheConfig
+
+pipe line= CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b", torch_dtype=torch.bfloat16)
+pipeline.to("cuda")
+
+config = FasterCacheConfig(
+    spatial_attention_block_skip_range=2,
+    spatial_attention_timestep_skip_range=(-1, 681),
+    current_timestep_callback=lambda: pipe.current_timestep,
+    attention_weight_callback=lambda _: 0.3,
+    unconditional_batch_skip_range=5,
+    unconditional_batch_timestep_skip_range=(-1, 781),
+    tensor_format="BFCHW",
+)
+pipeline.transformer.enable_cache(config)
+```

docs/source/en/optimization/deepcache.md

@@ -37,7 +37,7 @@ Then load and enable the [`DeepCacheSDHelper`](https://github.com/horseee/DeepCa
 ```
 
 The `set_params` method accepts two arguments: `cache_interval` and `cache_branch_id`. `cache_interval` means the frequency of feature caching, specified as the number of steps between each cache operation. `cache_branch_id` identifies which branch of the network (ordered from the shallowest to the deepest layer) is responsible for executing the caching processes.
-Opting for a lower `cache_branch_id` or a larger `cache_interval` can lead to faster inference speed at the expense of reduced image quality (ablation experiments of these two hyperparameters can be found in the [paper](https://arxiv.org/abs/2312.00858)). Once those arguments are set, use the `enable` or `disable` methods to activate or deactivate the `DeepCacheSDHelper`.
+Opting for a lower `cache_branch_id` or a larger `cache_interval` can lead to faster inference speed at the expense of reduced image quality (ablation experiments of these two hyperparameters can be found in the [paper](https://huggingface.co/papers/2312.00858)). Once those arguments are set, use the `enable` or `disable` methods to activate or deactivate the `DeepCacheSDHelper`.
 
 <div class="flex justify-center">
     <img src="https://github.com/horseee/Diffusion_DeepCache/raw/master/static/images/example.png">

docs/source/en/optimization/fp16.md

@@ -10,120 +10,235 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# Speed up inference
+# Accelerate inference
 
-There are several ways to optimize Diffusers for inference speed, such as reducing the computational burden by lowering the data precision or using a lightweight distilled model. There are also memory-efficient attention implementations, [xFormers](xformers) and [scaled dot product attention](https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html) in PyTorch 2.0, that reduce memory usage which also indirectly speeds up inference. Different speed optimizations can be stacked together to get the fastest inference times.
+Diffusion models are slow at inference because generation is an iterative process where noise is gradually refined into an image or video over a certain number of "steps". To speedup this process, you can try experimenting with different [schedulers](../api/schedulers/overview), reduce the precision of the model weights for faster computations, use more memory-efficient attention mechanisms, and more.
 
-> [!TIP]
-> Optimizing for inference speed or reduced memory usage can lead to improved performance in the other category, so you should try to optimize for both whenever you can. This guide focuses on inference speed, but you can learn more about lowering memory usage in the [Reduce memory usage](memory) guide.
+Combine and use these techniques together to make inference faster than using any single technique on its own.
 
-The inference times below are obtained from generating a single 512x512 image from the prompt "a photo of an astronaut riding a horse on mars" with 50 DDIM steps on a NVIDIA A100.
+This guide will go over how to accelerate inference.
 
-| setup    | latency | speed-up |
-|----------|---------|----------|
-| baseline | 5.27s   | x1       |
-| tf32     | 4.14s   | x1.27    |
-| fp16     | 3.51s   | x1.50    |
-| combined | 3.41s   | x1.54    |
+## Model data type
 
-## TensorFloat-32
+The precision and data type of the model weights affect inference speed because a higher precision requires more memory to load and more time to perform the computations. PyTorch loads model weights in float32 or full precision by default, so changing the data type is a simple way to quickly get faster inference.
 
-On Ampere and later CUDA devices, matrix multiplications and convolutions can use the [TensorFloat-32 (tf32)](https://blogs.nvidia.com/blog/2020/05/14/tensorfloat-32-precision-format/) mode for faster, but slightly less accurate computations. By default, PyTorch enables tf32 mode for convolutions but not matrix multiplications. Unless your network requires full float32 precision, we recommend enabling tf32 for matrix multiplications. It can significantly speed up computations with typically negligible loss in numerical accuracy.
+<hfoptions id="dtypes">
+<hfoption id="bfloat16">
 
-```python
+bfloat16 is similar to float16 but it is more robust to numerical errors. Hardware support for bfloat16 varies, but most modern GPUs are capable of supporting bfloat16.
+
+```py
 import torch
+from diffusers import StableDiffusionXLPipeline
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
+).to("cuda")
+
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+pipeline(prompt, num_inference_steps=30).images[0]
+```
+
+</hfoption>
+<hfoption id="float16">
+
+float16 is similar to bfloat16 but may be more prone to numerical errors.
+
+```py
+import torch
+from diffusers import StableDiffusionXLPipeline
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+).to("cuda")
+
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+pipeline(prompt, num_inference_steps=30).images[0]
+```
+
+</hfoption>
+<hfoption id="TensorFloat-32">
+
+[TensorFloat-32 (tf32)](https://blogs.nvidia.com/blog/2020/05/14/tensorfloat-32-precision-format/) mode is supported on NVIDIA Ampere GPUs and it computes the convolution and matrix multiplication operations in tf32. Storage and other operations are kept in float32. This enables significantly faster computations when combined with bfloat16 or float16.
+
+PyTorch only enables tf32 mode for convolutions by default and you'll need to explicitly enable it for matrix multiplications.
+
+```py
+import torch
+from diffusers import StableDiffusionXLPipeline
 
 torch.backends.cuda.matmul.allow_tf32 = True
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
+).to("cuda")
+
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+pipeline(prompt, num_inference_steps=30).images[0]
 ```
 
-Learn more about tf32 in the [Mixed precision training](https://huggingface.co/docs/transformers/en/perf_train_gpu_one#tf32) guide.
+Refer to the [mixed precision training](https://huggingface.co/docs/transformers/en/perf_train_gpu_one#mixed-precision) docs for more details.
 
-## Half-precision weights
+</hfoption>
+</hfoptions>
 
-To save GPU memory and get more speed, set `torch_dtype=torch.float16` to load and run the model weights directly with half-precision weights.
-
-```Python
-import torch
-from diffusers import DiffusionPipeline
-
-pipe = DiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
-)
-pipe = pipe.to("cuda")
-```
-
-> [!WARNING]
-> Don't use [torch.autocast](https://pytorch.org/docs/stable/amp.html#torch.autocast) in any of the pipelines as it can lead to black images and is always slower than pure float16 precision.
-
-## Distilled model
-
-You could also use a distilled Stable Diffusion model and autoencoder to speed up inference. During distillation, many of the UNet's residual and attention blocks are shed to reduce the model size by 51% and improve latency on CPU/GPU by 43%. The distilled model is faster and uses less memory while generating images of comparable quality to the full Stable Diffusion model.
+## Scaled dot product attention
 
 > [!TIP]
-> Read the [Open-sourcing Knowledge Distillation Code and Weights of SD-Small and SD-Tiny](https://huggingface.co/blog/sd_distillation) blog post to learn more about how knowledge distillation training works to produce a faster, smaller, and cheaper generative model.
+> Memory-efficient attention optimizes for inference speed *and* [memory usage](./memory#memory-efficient-attention)!
 
-The inference times below are obtained from generating 4 images from the prompt "a photo of an astronaut riding a horse on mars" with 25 PNDM steps on a NVIDIA A100. Each generation is repeated 3 times with the distilled Stable Diffusion v1.4 model by [Nota AI](https://hf.co/nota-ai).
+[Scaled dot product attention (SDPA)](https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html) implements several attention backends, [FlashAttention](https://github.com/Dao-AILab/flash-attention), [xFormers](https://github.com/facebookresearch/xformers), and a native C++ implementation. It automatically selects the most optimal backend for your hardware.
 
-| setup                        | latency | speed-up |
-|------------------------------|---------|----------|
-| baseline                     | 6.37s   | x1       |
-| distilled                    | 4.18s   | x1.52    |
-| distilled + tiny autoencoder | 3.83s   | x1.66    |
-
-Let's load the distilled Stable Diffusion model and compare it against the original Stable Diffusion model.
+SDPA is enabled by default if you're using PyTorch >= 2.0 and no additional changes are required to your code. You could try experimenting with other attention backends though if you'd like to choose your own. The example below uses the [torch.nn.attention.sdpa_kernel](https://pytorch.org/docs/stable/generated/torch.nn.attention.sdpa_kernel.html) context manager to enable efficient attention.
 
 ```py
-from diffusers import StableDiffusionPipeline
+from torch.nn.attention import SDPBackend, sdpa_kernel
 import torch
+from diffusers import StableDiffusionXLPipeline
 
-distilled = StableDiffusionPipeline.from_pretrained(
-    "nota-ai/bk-sdm-small", torch_dtype=torch.float16, use_safetensors=True,
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
 ).to("cuda")
-prompt = "a golden vase with different flowers"
-generator = torch.manual_seed(2023)
-image = distilled("a golden vase with different flowers", num_inference_steps=25, generator=generator).images[0]
-image
+
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+
+with sdpa_kernel(SDPBackend.EFFICIENT_ATTENTION):
+  image = pipeline(prompt, num_inference_steps=30).images[0]
 ```
 
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/original_sd.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original Stable Diffusion</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/distilled_sd.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">distilled Stable Diffusion</figcaption>
-  </div>
-</div>
+## torch.compile
 
-### Tiny AutoEncoder
+[torch.compile](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) accelerates inference by compiling PyTorch code and operations into optimized kernels. Diffusers typically compiles the more compute-intensive models like the UNet, transformer, or VAE.
 
-To speed inference up even more, replace the autoencoder with a [distilled version](https://huggingface.co/sayakpaul/taesdxl-diffusers) of it.
+Enable the following compiler settings for maximum speed (refer to the [full list](https://github.com/pytorch/pytorch/blob/main/torch/_inductor/config.py) for more options).
 
 ```py
 import torch
-from diffusers import AutoencoderTiny, StableDiffusionPipeline
+from diffusers import StableDiffusionXLPipeline
 
-distilled = StableDiffusionPipeline.from_pretrained(
-    "nota-ai/bk-sdm-small", torch_dtype=torch.float16, use_safetensors=True,
-).to("cuda")
-distilled.vae = AutoencoderTiny.from_pretrained(
-    "sayakpaul/taesd-diffusers", torch_dtype=torch.float16, use_safetensors=True,
-).to("cuda")
-
-prompt = "a golden vase with different flowers"
-generator = torch.manual_seed(2023)
-image = distilled("a golden vase with different flowers", num_inference_steps=25, generator=generator).images[0]
-image
+torch._inductor.config.conv_1x1_as_mm = True
+torch._inductor.config.coordinate_descent_tuning = True
+torch._inductor.config.epilogue_fusion = False
+torch._inductor.config.coordinate_descent_check_all_directions = True
 ```
 
-<div class="flex justify-center">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/distilled_sd_vae.png" />
-    <figcaption class="mt-2 text-center text-sm text-gray-500">distilled Stable Diffusion + Tiny AutoEncoder</figcaption>
-  </div>
-</div>
+Load and compile the UNet and VAE. There are several different modes you can choose from, but `"max-autotune"` optimizes for the fastest speed by compiling to a CUDA graph. CUDA graphs effectively reduces the overhead by launching multiple GPU operations through a single CPU operation.
 
-More tiny autoencoder models for other Stable Diffusion models, like Stable Diffusion 3, are available from [madebyollin](https://huggingface.co/madebyollin).
+> [!TIP]
+> With PyTorch 2.3.1, you can control the caching behavior of torch.compile. This is particularly beneficial for compilation modes like `"max-autotune"` which performs a grid-search over several compilation flags to find the optimal configuration. Learn more in the [Compile Time Caching in torch.compile](https://pytorch.org/tutorials/recipes/torch_compile_caching_tutorial.html) tutorial.
+
+Changing the memory layout to [channels_last](./memory#torchchannels_last) also optimizes memory and inference speed.
+
+```py
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+).to("cuda")
+pipeline.unet.to(memory_format=torch.channels_last)
+pipeline.vae.to(memory_format=torch.channels_last)
+pipeline.unet = torch.compile(
+    pipeline.unet, mode="max-autotune", fullgraph=True
+)
+pipeline.vae.decode = torch.compile(
+    pipeline.vae.decode,
+    mode="max-autotune",
+    fullgraph=True
+)
+
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+pipeline(prompt, num_inference_steps=30).images[0]
+```
+
+Compilation is slow the first time, but once compiled, it is significantly faster. Try to only use the compiled pipeline on the same type of inference operations. Calling the compiled pipeline on a different image size retriggers compilation which is slow and inefficient.
+
+### Regional compilation
+
+[Regional compilation](https://docs.pytorch.org/tutorials/recipes/regional_compilation.html) reduces the cold start compilation time by only compiling a specific repeated region (or block) of the model instead of the entire model. The compiler reuses the cached and compiled code for the other blocks.
+
+[Accelerate](https://huggingface.co/docs/accelerate/index) provides the [compile_regions](https://github.com/huggingface/accelerate/blob/273799c85d849a1954a4f2e65767216eb37fa089/src/accelerate/utils/other.py#L78) method for automatically compiling the repeated blocks of a `nn.Module` sequentially. The rest of the model is compiled separately.
+
+```py
+# pip install -U accelerate
+import torch
+from diffusers import StableDiffusionXLPipeline
+from accelerate.utils import compile regions
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+).to("cuda")
+pipeline.unet = compile_regions(pipeline.unet, mode="reduce-overhead", fullgraph=True)
+```
+
+### Graph breaks
+
+It is important to specify `fullgraph=True` in torch.compile to ensure there are no graph breaks in the underlying model. This allows you to take advantage of torch.compile without any performance degradation. For the UNet and VAE, this changes how you access the return variables.
+
+```diff
+- latents = unet(
+-   latents, timestep=timestep, encoder_hidden_states=prompt_embeds
+-).sample
+
++ latents = unet(
++   latents, timestep=timestep, encoder_hidden_states=prompt_embeds, return_dict=False
++)[0]
+```
+
+### GPU sync
+
+The `step()` function is [called](https://github.com/huggingface/diffusers/blob/1d686bac8146037e97f3fd8c56e4063230f71751/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py#L1228) on the scheduler each time after the denoiser makes a prediction, and the `sigmas` variable is [indexed](https://github.com/huggingface/diffusers/blob/1d686bac8146037e97f3fd8c56e4063230f71751/src/diffusers/schedulers/scheduling_euler_discrete.py#L476). When placed on the GPU, it introduces latency because of the communication sync between the CPU and GPU. It becomes more evident when the denoiser has already been compiled.
+
+In general, the `sigmas` should [stay on the CPU](https://github.com/huggingface/diffusers/blob/35a969d297cba69110d175ee79c59312b9f49e1e/src/diffusers/schedulers/scheduling_euler_discrete.py#L240) to avoid the communication sync and latency.
+
+### Benchmarks
+
+Refer to the [diffusers/benchmarks](https://huggingface.co/datasets/diffusers/benchmarks) dataset to see inference latency and memory usage data for compiled pipelines.
+
+The [diffusers-torchao](https://github.com/sayakpaul/diffusers-torchao#benchmarking-results) repository also contains benchmarking results for compiled versions of Flux and CogVideoX.
+
+## Dynamic quantization
+
+[Dynamic quantization](https://pytorch.org/tutorials/recipes/recipes/dynamic_quantization.html) improves inference speed by reducing precision to enable faster math operations. This particular type of quantization determines how to scale the activations based on the data at runtime rather than using a fixed scaling factor. As a result, the scaling factor is more accurately aligned with the data.
+
+The example below applies [dynamic int8 quantization](https://pytorch.org/tutorials/recipes/recipes/dynamic_quantization.html) to the UNet and VAE with the [torchao](../quantization/torchao) library.
+
+> [!TIP]
+> Refer to our [torchao](../quantization/torchao) docs to learn more about how to use the Diffusers torchao integration.
+
+Configure the compiler tags for maximum speed.
+
+```py
+import torch
+from torchao import apply_dynamic_quant
+from diffusers import StableDiffusionXLPipeline
+
+torch._inductor.config.conv_1x1_as_mm = True
+torch._inductor.config.coordinate_descent_tuning = True
+torch._inductor.config.epilogue_fusion = False
+torch._inductor.config.coordinate_descent_check_all_directions = True
+torch._inductor.config.force_fuse_int_mm_with_mul = True
+torch._inductor.config.use_mixed_mm = True
+```
+
+Filter out some linear layers in the UNet and VAE which don't benefit from dynamic quantization with the [dynamic_quant_filter_fn](https://github.com/huggingface/diffusion-fast/blob/0f169640b1db106fe6a479f78c1ed3bfaeba3386/utils/pipeline_utils.py#L16).
+
+```py
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
+).to("cuda")
+
+apply_dynamic_quant(pipeline.unet, dynamic_quant_filter_fn)
+apply_dynamic_quant(pipeline.vae, dynamic_quant_filter_fn)
+
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+pipeline(prompt, num_inference_steps=30).images[0]
+```
+
+## Fused projection matrices
+
+> [!WARNING]
+> The [fuse_qkv_projections](https://github.com/huggingface/diffusers/blob/58431f102cf39c3c8a569f32d71b2ea8caa461e1/src/diffusers/pipelines/pipeline_utils.py#L2034) method is experimental and support is limited to mostly Stable Diffusion pipelines. Take a look at this [PR](https://github.com/huggingface/diffusers/pull/6179) to learn more about how to enable it for other pipelines
+
+An input is projected into three subspaces, represented by the projection matrices Q, K, and V, in an attention block. These projections are typically calculated separately, but you can horizontally combine these into a single matrix and perform the projection in a single step. It increases the size of the matrix multiplications of the input projections and also improves the impact of quantization.
+
+```py
+pipeline.fuse_qkv_projections()
+```

docs/source/en/optimization/habana.md

@@ -10,67 +10,22 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# Habana Gaudi
+# Intel Gaudi
 
-🤗 Diffusers is compatible with Habana Gaudi through 🤗 [Optimum](https://huggingface.co/docs/optimum/habana/usage_guides/stable_diffusion). Follow the [installation](https://docs.habana.ai/en/latest/Installation_Guide/index.html) guide to install the SynapseAI and Gaudi drivers, and then install Optimum Habana:
+The Intel Gaudi AI accelerator family includes [Intel Gaudi 1](https://habana.ai/products/gaudi/), [Intel Gaudi 2](https://habana.ai/products/gaudi2/), and [Intel Gaudi 3](https://habana.ai/products/gaudi3/). Each server is equipped with 8 devices, known as Habana Processing Units (HPUs), providing 128GB of memory on Gaudi 3, 96GB on Gaudi 2, and 32GB on the first-gen Gaudi. For more details on the underlying hardware architecture, check out the [Gaudi Architecture](https://docs.habana.ai/en/latest/Gaudi_Overview/Gaudi_Architecture.html) overview.
 
-```bash
-python -m pip install --upgrade-strategy eager optimum[habana]
+Diffusers pipelines can take advantage of HPU acceleration, even if a pipeline hasn't been added to [Optimum for Intel Gaudi](https://huggingface.co/docs/optimum/main/en/habana/index) yet, with the [GPU Migration Toolkit](https://docs.habana.ai/en/latest/PyTorch/PyTorch_Model_Porting/GPU_Migration_Toolkit/GPU_Migration_Toolkit.html).
+
+Call `.to("hpu")` on your pipeline to move it to a HPU device as shown below for Flux:
+```py
+import torch
+from diffusers import DiffusionPipeline
+
+pipeline = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16)
+pipeline.to("hpu")
+
+image = pipeline("An image of a squirrel in Picasso style").images[0]
 ```
 
-To generate images with Stable Diffusion 1 and 2 on Gaudi, you need to instantiate two instances:
-
-- [`~optimum.habana.diffusers.GaudiStableDiffusionPipeline`], a pipeline for text-to-image generation.
-- [`~optimum.habana.diffusers.GaudiDDIMScheduler`], a Gaudi-optimized scheduler.
-
-When you initialize the pipeline, you have to specify `use_habana=True` to deploy it on HPUs and to get the fastest possible generation, you should enable **HPU graphs** with `use_hpu_graphs=True`.
-
-Finally, specify a [`~optimum.habana.GaudiConfig`] which can be downloaded from the [Habana](https://huggingface.co/Habana) organization on the Hub.
-
-```python
-from optimum.habana import GaudiConfig
-from optimum.habana.diffusers import GaudiDDIMScheduler, GaudiStableDiffusionPipeline
-
-model_name = "stabilityai/stable-diffusion-2-base"
-scheduler = GaudiDDIMScheduler.from_pretrained(model_name, subfolder="scheduler")
-pipeline = GaudiStableDiffusionPipeline.from_pretrained(
-    model_name,
-    scheduler=scheduler,
-    use_habana=True,
-    use_hpu_graphs=True,
-    gaudi_config="Habana/stable-diffusion-2",
-)
-```
-
-Now you can call the pipeline to generate images by batches from one or several prompts:
-
-```python
-outputs = pipeline(
-    prompt=[
-        "High quality photo of an astronaut riding a horse in space",
-        "Face of a yellow cat, high resolution, sitting on a park bench",
-    ],
-    num_images_per_prompt=10,
-    batch_size=4,
-)
-```
-
-For more information, check out 🤗 Optimum Habana's [documentation](https://huggingface.co/docs/optimum/habana/usage_guides/stable_diffusion) and the [example](https://github.com/huggingface/optimum-habana/tree/main/examples/stable-diffusion) provided in the official GitHub repository.
-
-## Benchmark
-
-We benchmarked Habana's first-generation Gaudi and Gaudi2 with the [Habana/stable-diffusion](https://huggingface.co/Habana/stable-diffusion) and [Habana/stable-diffusion-2](https://huggingface.co/Habana/stable-diffusion-2) Gaudi configurations (mixed precision bf16/fp32) to demonstrate their performance.
-
-For [Stable Diffusion v1.5](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5) on 512x512 images:
-
-|                        | Latency (batch size = 1) | Throughput  |
-| ---------------------- |:------------------------:|:---------------------------:|
-| first-generation Gaudi | 3.80s                    | 0.308 images/s (batch size = 8)             |
-| Gaudi2                 | 1.33s                    | 1.081 images/s (batch size = 8)             |
-
-For [Stable Diffusion v2.1](https://huggingface.co/stabilityai/stable-diffusion-2-1) on 768x768 images:
-
-|                        | Latency (batch size = 1) | Throughput                      |
-| ---------------------- |:------------------------:|:-------------------------------:|
-| first-generation Gaudi | 10.2s                    | 0.108 images/s (batch size = 4) |
-| Gaudi2                 | 3.17s                    | 0.379 images/s (batch size = 8) |
+> [!TIP]
+> For Gaudi-optimized diffusion pipeline implementations, we recommend using [Optimum for Intel Gaudi](https://huggingface.co/docs/optimum/main/en/habana/index).

docs/source/en/optimization/memory.md

@@ -12,178 +12,258 @@ specific language governing permissions and limitations under the License.
 
 # Reduce memory usage
 
-A barrier to using diffusion models is the large amount of memory required. To overcome this challenge, there are several memory-reducing techniques you can use to run even some of the largest models on free-tier or consumer GPUs. Some of these techniques can even be combined to further reduce memory usage.
+Modern diffusion models like [Flux](../api/pipelines/flux) and [Wan](../api/pipelines/wan) have billions of parameters that take up a lot of memory on your hardware for inference. This is challenging because common GPUs often don't have sufficient memory. To overcome the memory limitations, you can use more than one GPU (if available), offload some of the pipeline components to the CPU, and more.
 
-<Tip>
+This guide will show you how to reduce your memory usage. 
 
-In many cases, optimizing for memory or speed leads to improved performance in the other, so you should try to optimize for both whenever you can. This guide focuses on minimizing memory usage, but you can also learn more about how to [Speed up inference](fp16).
+> [!TIP]
+> Keep in mind these techniques may need to be adjusted depending on the model! For example, a transformer-based diffusion model may not benefit equally from these inference speed optimizations as a UNet-based model.
 
-</Tip>
+## Multiple GPUs
 
-The results below are obtained from generating a single 512x512 image from the prompt a photo of an astronaut riding a horse on mars with 50 DDIM steps on a Nvidia Titan RTX, demonstrating the speed-up you can expect as a result of reduced memory consumption.
+If you have access to more than one GPU, there a few options for efficiently loading and distributing a large model across your hardware. These features are supported by the [Accelerate](https://huggingface.co/docs/accelerate/index) library, so make sure it is installed first.
 
-|                  | latency | speed-up |
-| ---------------- | ------- | ------- |
-| original         | 9.50s   | x1      |
-| fp16             | 3.61s   | x2.63   |
-| channels last    | 3.30s   | x2.88   |
-| traced UNet      | 3.21s   | x2.96   |
-| memory-efficient attention  | 2.63s  | x3.61   |
-
-## Sliced VAE
-
-Sliced VAE enables decoding large batches of images with limited VRAM or batches with 32 images or more by decoding the batches of latents one image at a time. You'll likely want to couple this with [`~ModelMixin.enable_xformers_memory_efficient_attention`] to reduce memory use further if you have xFormers installed.
-
-To use sliced VAE, call [`~StableDiffusionPipeline.enable_vae_slicing`] on your pipeline before inference:
-
-```python
-import torch
-from diffusers import StableDiffusionPipeline
-
-pipe = StableDiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
-)
-pipe = pipe.to("cuda")
-
-prompt = "a photo of an astronaut riding a horse on mars"
-pipe.enable_vae_slicing()
-#pipe.enable_xformers_memory_efficient_attention()
-images = pipe([prompt] * 32).images
+```bash
+pip install -U accelerate
 ```
 
-You may see a small performance boost in VAE decoding on multi-image batches, and there should be no performance impact on single-image batches.
+### Sharded checkpoints
 
-## Tiled VAE
+Loading large checkpoints in several shards in useful because the shards are loaded one at a time. This keeps memory usage low, only requiring enough memory for the model size and the largest shard size. We recommend sharding when the fp32 checkpoint is greater than 5GB. The default shard size is 5GB.
 
-Tiled VAE processing also enables working with large images on limited VRAM (for example, generating 4k images on 8GB of VRAM) by splitting the image into overlapping tiles, decoding the tiles, and then blending the outputs together to compose the final image. You should also used tiled VAE with [`~ModelMixin.enable_xformers_memory_efficient_attention`] to reduce memory use further if you have xFormers installed.
+Shard a checkpoint in [`~DiffusionPipeline.save_pretrained`] with the `max_shard_size` parameter.
 
-To use tiled VAE processing, call [`~StableDiffusionPipeline.enable_vae_tiling`] on your pipeline before inference:
+```py
+from diffusers import AutoModel
 
-```python
-import torch
-from diffusers import StableDiffusionPipeline, UniPCMultistepScheduler
-
-pipe = StableDiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
+unet = AutoModel.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", subfolder="unet"
 )
-pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
-pipe = pipe.to("cuda")
-prompt = "a beautiful landscape photograph"
-pipe.enable_vae_tiling()
-#pipe.enable_xformers_memory_efficient_attention()
-
-image = pipe([prompt], width=3840, height=2224, num_inference_steps=20).images[0]
+unet.save_pretrained("sdxl-unet-sharded", max_shard_size="5GB")
 ```
 
-The output image has some tile-to-tile tone variation because the tiles are decoded separately, but you shouldn't see any sharp and obvious seams between the tiles. Tiling is turned off for images that are 512x512 or smaller.
+Now you can use the sharded checkpoint, instead of the regular checkpoint, to save memory.
+
+```py
+import torch
+from diffusers import AutoModel, StableDiffusionXLPipeline
+
+unet = AutoModel.from_pretrained(
+    "username/sdxl-unet-sharded", torch_dtype=torch.float16
+)
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    unet=unet,
+    torch_dtype=torch.float16
+).to("cuda")
+```
+
+### Device placement
+
+> [!WARNING]
+> Device placement is an experimental feature and the API may change. Only the `balanced` strategy is supported at the moment. We plan to support additional mapping strategies in the future.
+
+The `device_map` parameter controls how the model components in a pipeline are distributed across devices. The `balanced` device placement strategy evenly splits the pipeline across all available devices.
+
+```py
+import torch
+from diffusers import AutoModel, StableDiffusionXLPipeline
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+    device_map="balanced"
+)
+```
+
+You can inspect a pipeline's device map with `hf_device_map`.
+
+```py
+print(pipeline.hf_device_map)
+{'unet': 1, 'vae': 1, 'safety_checker': 0, 'text_encoder': 0}
+```
+
+The `device_map` parameter also works on the model-level. This is useful for loading large models, such as the Flux diffusion transformer which has 12.5B parameters. Instead of `balanced`, set it to `"auto"` to automatically distribute a model across the fastest device first before moving to slower devices. Refer to the [Model sharding](../training/distributed_inference#model-sharding) docs for more details.
+
+```py
+import torch
+from diffusers import AutoModel
+
+transformer = AutoModel.from_pretrained(
+    "black-forest-labs/FLUX.1-dev", 
+    subfolder="transformer",
+    device_map="auto",
+    torch_dtype=torch.bfloat16
+)
+```
+
+For more fine-grained control, pass a dictionary to enforce the maximum GPU memory to use on each device. If a device is not in `max_memory`, it is ignored and pipeline components won't be distributed to it.
+
+```py
+import torch
+from diffusers import AutoModel, StableDiffusionXLPipeline
+
+max_memory = {0:"1GB", 1:"1GB"}
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+    device_map="balanced",
+    max_memory=max_memory
+)
+```
+
+Diffusers uses the maxmium memory of all devices by default, but if they don't fit on the GPUs, then you'll need to use a single GPU and offload to the CPU with the methods below.
+
+- [`~DiffusionPipeline.enable_model_cpu_offload`] only works on a single GPU but a very large model may not fit on it
+- [`~DiffusionPipeline.enable_sequential_cpu_offload`] may work but it is extremely slow and also limited to a single GPU
+
+Use the [`~DiffusionPipeline.reset_device_map`] method to reset the `device_map`. This is necessary if you want to use methods like `.to()`, [`~DiffusionPipeline.enable_sequential_cpu_offload`], and [`~DiffusionPipeline.enable_model_cpu_offload`] on a pipeline that was device-mapped.
+
+```py
+pipeline.reset_device_map()
+```
+
+## VAE slicing
+
+VAE slicing saves memory by splitting large batches of inputs into a single batch of data and separately processing them. This method works best when generating more than one image at a time.
+
+For example, if you're generating 4 images at once, decoding would increase peak activation memory by 4x. VAE slicing reduces this by only decoding 1 image at a time instead of all 4 images at once.
+
+Call [`~StableDiffusionPipeline.enable_vae_slicing`] to enable sliced VAE. You can expect a small increase in performance when decoding multi-image batches and no performance impact for single-image batches.
+
+```py
+import torch
+from diffusers import AutoModel, StableDiffusionXLPipeline
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+).to("cuda")
+pipeline.enable_vae_slicing()
+pipeline(["An astronaut riding a horse on Mars"]*32).images[0]
+print(f"Max memory reserved: {torch.cuda.max_memory_allocated() / 1024**3:.2f} GB")
+```
+
+> [!WARNING]
+> [`AutoencoderKLWan`] and [`AsymmetricAutoencoderKL`] don't support slicing.
+
+## VAE tiling
+
+VAE tiling saves memory by dividing an image into smaller overlapping tiles instead of processing the entire image at once. This also reduces peak memory usage because the GPU is only processing a tile at a time.
+
+Call [`~StableDiffusionPipeline.enable_vae_tiling`] to enable VAE tiling. The generated image may have some tone variation from tile-to-tile because they're decoded separately, but there shouldn't be any obvious seams between the tiles. Tiling is disabled for resolutions lower than a pre-specified (but configurable) limit. For example, this limit is 512x512 for the VAE in [`StableDiffusionPipeline`].
+
+```py
+import torch
+from diffusers import AutoPipelineForImage2Image
+from diffusers.utils import load_image
+
+pipeline = AutoPipelineForImage2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
+).to("cuda")
+pipeline.enable_vae_tiling()
+
+init_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-sdxl-init.png")
+prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
+pipeline(prompt, image=init_image, strength=0.5).images[0]
+print(f"Max memory reserved: {torch.cuda.max_memory_allocated() / 1024**3:.2f} GB")
+```
+
+> [!WARNING]
+> [`AutoencoderKLWan`] and [`AsymmetricAutoencoderKL`] don't support tiling.
 
 ## CPU offloading
 
-Offloading the weights to the CPU and only loading them on the GPU when performing the forward pass can also save memory. Often, this technique can reduce memory consumption to less than 3GB.
+CPU offloading selectively moves weights from the GPU to the CPU. When a component is required, it is transferred to the GPU and when it isn't required, it is moved to the CPU. This method works on submodules rather than whole models. It saves memory by avoiding storing the entire model on the GPU.
 
-To perform CPU offloading, call [`~StableDiffusionPipeline.enable_sequential_cpu_offload`]:
+CPU offloading dramatically reduces memory usage, but it is also **extremely slow** because submodules are passed back and forth multiple times between devices. It can often be impractical due to how slow it is.
 
-```Python
+> [!WARNING]
+> Don't move the pipeline to CUDA before calling [`~DiffusionPipeline.enable_sequential_cpu_offload`], otherwise the amount of memory saved is only minimal (refer to this [issue](https://github.com/huggingface/diffusers/issues/1934) for more details). This is a stateful operation that installs hooks on the model.
+
+Call [`~DiffusionPipeline.enable_sequential_cpu_offload`] to enable it on a pipeline.
+
+```py
 import torch
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline
 
-pipe = StableDiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
+pipeline = DiffusionPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16
 )
+pipeline.enable_sequential_cpu_offload()
 
-prompt = "a photo of an astronaut riding a horse on mars"
-pipe.enable_sequential_cpu_offload()
-image = pipe(prompt).images[0]
+pipeline(
+    prompt="An astronaut riding a horse on Mars",
+    guidance_scale=0.,
+    height=768,
+    width=1360,
+    num_inference_steps=4,
+    max_sequence_length=256,
+).images[0]
+print(f"Max memory reserved: {torch.cuda.max_memory_allocated() / 1024**3:.2f} GB")
 ```
 
-CPU offloading works on submodules rather than whole models. This is the best way to minimize memory consumption, but inference is much slower due to the iterative nature of the diffusion process. The UNet component of the pipeline runs several times (as many as `num_inference_steps`); each time, the different UNet submodules are sequentially onloaded and offloaded as needed, resulting in a large number of memory transfers.
-
-<Tip>
-
-Consider using [model offloading](#model-offloading) if you want to optimize for speed because it is much faster. The tradeoff is your memory savings won't be as large.
-
-</Tip>
-
-<Tip warning={true}>
-
-When using [`~StableDiffusionPipeline.enable_sequential_cpu_offload`], don't move the pipeline to CUDA beforehand or else the gain in memory consumption will only be minimal (see this [issue](https://github.com/huggingface/diffusers/issues/1934) for more information).
-
-[`~StableDiffusionPipeline.enable_sequential_cpu_offload`] is a stateful operation that installs hooks on the models.
-
-</Tip>
-
 ## Model offloading
 
-<Tip>
+Model offloading moves entire models to the GPU instead of selectively moving *some* layers or model components. One of the main pipeline models, usually the text encoder, UNet, and VAE, is placed on the GPU while the other components are held on the CPU. Components like the UNet that run multiple times stays on the GPU until its completely finished and no longer needed. This eliminates the communication overhead of [CPU offloading](#cpu-offloading) and makes model offloading a faster alternative. The tradeoff is memory savings won't be as large.
 
-Model offloading requires 🤗 Accelerate version 0.17.0 or higher.
+> [!WARNING]
+> Keep in mind that if models are reused outside the pipeline after hookes have been installed (see [Removing Hooks](https://huggingface.co/docs/accelerate/en/package_reference/big_modeling#accelerate.hooks.remove_hook_from_module) for more details), you need to run the entire pipeline and models in the expected order to properly offload them. This is a stateful operation that installs hooks on the model.
 
-</Tip>
+Call [`~DiffusionPipeline.enable_model_cpu_offload`] to enable it on a pipeline.
 
-[Sequential CPU offloading](#cpu-offloading) preserves a lot of memory but it makes inference slower because submodules are moved to GPU as needed, and they're immediately returned to the CPU when a new module runs.
-
-Full-model offloading is an alternative that moves whole models to the GPU, instead of handling each model's constituent *submodules*. There is a negligible impact on inference time (compared with moving the pipeline to `cuda`), and it still provides some memory savings.
-
-During model offloading, only one of the main components of the pipeline (typically the text encoder, UNet and VAE)
-is placed on the GPU while the others wait on the CPU. Components like the UNet that run for multiple iterations stay on the GPU until they're no longer needed.
-
-Enable model offloading by calling [`~StableDiffusionPipeline.enable_model_cpu_offload`] on the pipeline:
-
-```Python
+```py
 import torch
-from diffusers import StableDiffusionPipeline
+from diffusers import DiffusionPipeline
 
-pipe = StableDiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
+pipeline = DiffusionPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-schnell", torch_dtype=torch.bfloat16
 )
+pipline.enable_model_cpu_offload()
 
-prompt = "a photo of an astronaut riding a horse on mars"
-pipe.enable_model_cpu_offload()
-image = pipe(prompt).images[0]
+pipeline(
+    prompt="An astronaut riding a horse on Mars",
+    guidance_scale=0.,
+    height=768,
+    width=1360,
+    num_inference_steps=4,
+    max_sequence_length=256,
+).images[0]
+print(f"Max memory reserved: {torch.cuda.max_memory_allocated() / 1024**3:.2f} GB")
 ```
 
-<Tip warning={true}>
-
-In order to properly offload models after they're called, it is required to run the entire pipeline and models are called in the pipeline's expected order. Exercise caution if models are reused outside the context of the pipeline after hooks have been installed. See [Removing Hooks](https://huggingface.co/docs/accelerate/en/package_reference/big_modeling#accelerate.hooks.remove_hook_from_module) for more information.
-
-[`~StableDiffusionPipeline.enable_model_cpu_offload`] is a stateful operation that installs hooks on the models and state on the pipeline.
-
-</Tip>
+[`~DiffusionPipeline.enable_model_cpu_offload`] also helps when you're using the [`~StableDiffusionXLPipeline.encode_prompt`] method on its own to generate the text encoders hidden state.
 
 ## Group offloading
 
-Group offloading is the middle ground between sequential and model offloading. It works by offloading groups of internal layers (either `torch.nn.ModuleList` or `torch.nn.Sequential`), which uses less memory than model-level offloading. It is also faster than sequential-level offloading because the number of device synchronizations is reduced.
+Group offloading moves groups of internal layers ([torch.nn.ModuleList](https://pytorch.org/docs/stable/generated/torch.nn.ModuleList.html) or [torch.nn.Sequential](https://pytorch.org/docs/stable/generated/torch.nn.Sequential.html)) to the CPU. It uses less memory than [model offloading](#model-offloading) and it is faster than [CPU offloading](#cpu-offloading) because it reduces communication overhead.
 
-To enable group offloading, call the [`~ModelMixin.enable_group_offload`] method on the model if it is a Diffusers model implementation. For any other model implementation, use [`~hooks.group_offloading.apply_group_offloading`]:
+> [!WARNING]
+> Group offloading may not work with all models if the forward implementation contains weight-dependent device casting of inputs because it may clash with group offloading's device casting mechanism.
 
-```python
+Call [`~ModelMixin.enable_group_offload`] to enable it for standard Diffusers model components that inherit from [`ModelMixin`]. For other model components that don't inherit from [`ModelMixin`], such as a generic [torch.nn.Module](https://pytorch.org/docs/stable/generated/torch.nn.Module.html), use [`~hooks.apply_group_offloading`] instead.
+
+The `offload_type` parameter can be set to `block_level` or `leaf_level`.
+
+- `block_level` offloads groups of layers based on the `num_blocks_per_group` parameter. For example, if `num_blocks_per_group=2` on a model with 40 layers, 2 layers are onloaded and offloaded at a time (20 total onloads/offloads). This drastically reduces memory requirements.
+- `leaf_level` offloads individual layers at the lowest level and is equivalent to [CPU offloading](#cpu-offloading). But it can be made faster if you use streams without giving up inference speed.
+
+```py
 import torch
 from diffusers import CogVideoXPipeline
 from diffusers.hooks import apply_group_offloading
 from diffusers.utils import export_to_video
 
-# Load the pipeline
 onload_device = torch.device("cuda")
 offload_device = torch.device("cpu")
-pipe = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b", torch_dtype=torch.bfloat16)
+pipeline = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b", torch_dtype=torch.bfloat16)
 
-# We can utilize the enable_group_offload method for Diffusers model implementations
-pipe.transformer.enable_group_offload(onload_device=onload_device, offload_device=offload_device, offload_type="leaf_level", use_stream=True)
+# Use the enable_group_offload method for Diffusers model implementations
+pipeline.transformer.enable_group_offload(onload_device=onload_device, offload_device=offload_device, offload_type="leaf_level")
+pipeline.vae.enable_group_offload(onload_device=onload_device, offload_type="leaf_level")
 
-# Uncomment the following to also allow recording the current streams.
-# pipe.transformer.enable_group_offload(onload_device=onload_device, offload_device=offload_device, offload_type="leaf_level", use_stream=True, record_stream=True)
-
-# For any other model implementations, the apply_group_offloading function can be used
-apply_group_offloading(pipe.text_encoder, onload_device=onload_device, offload_type="block_level", num_blocks_per_group=2)
-apply_group_offloading(pipe.vae, onload_device=onload_device, offload_type="leaf_level")
+# Use the apply_group_offloading method for other model components
+apply_group_offloading(pipeline.text_encoder, onload_device=onload_device, offload_type="block_level", num_blocks_per_group=2)
 
 prompt = (
     "A panda, dressed in a small, red jacket and a tiny hat, sits on a wooden stool in a serene bamboo forest. "
@@ -193,48 +273,62 @@ prompt = (
     "The background includes a small, flowing stream and vibrant green foliage, enhancing the peaceful and magical "
     "atmosphere of this unique musical performance."
 )
-video = pipe(prompt=prompt, guidance_scale=6, num_inference_steps=50).frames[0]
-# This utilized about 14.79 GB. It can be further reduced by using tiling and using leaf_level offloading throughout the pipeline.
+video = pipeline(prompt=prompt, guidance_scale=6, num_inference_steps=50).frames[0]
 print(f"Max memory reserved: {torch.cuda.max_memory_allocated() / 1024**3:.2f} GB")
 export_to_video(video, "output.mp4", fps=8)
 ```
 
-Group offloading (for CUDA devices with support for asynchronous data transfer streams) overlaps data transfer and computation to reduce the overall execution time compared to sequential offloading. This is enabled using layer prefetching with CUDA streams. The next layer to be executed is loaded onto the accelerator device while the current layer is being executed - this increases the memory requirements slightly. Group offloading also supports leaf-level offloading (equivalent to sequential CPU offloading) but can be made much faster when using streams.
+### CUDA stream
+
+The `use_stream` parameter can be activated for CUDA devices that support asynchronous data transfer streams to reduce overall execution time compared to [CPU offloading](#cpu-offloading). It overlaps data transfer and computation by using layer prefetching. The next layer to be executed is loaded onto the GPU while the current layer is still being executed. It can increase CPU memory significantly so ensure you have 2x the amount of memory as the model size.
+
+Set `record_stream=True` for more of a speedup at the cost of slightly increased memory usage. Refer to the [torch.Tensor.record_stream](https://pytorch.org/docs/stable/generated/torch.Tensor.record_stream.html) docs to learn more.
+
+> [!TIP]
+> When `use_stream=True` on VAEs with tiling enabled, make sure to do a dummy forward pass (possible with dummy inputs as well) before inference to avoid device mismatch errors. This may not work on all implementations, so feel free to open an issue if you encounter any problems.
+
+If you're using `block_level` group offloading with `use_stream` enabled, the `num_blocks_per_group` parameter should be set to `1`, otherwise a warning will be raised.
+
+```py
+pipeline.transformer.enable_group_offload(onload_device=onload_device, offload_device=offload_device, offload_type="leaf_level", use_stream=True, record_stream=True)
+```
+
+The `low_cpu_mem_usage` parameter can be set to `True` to reduce CPU memory usage when using streams during group offloading. It is best for `leaf_level` offloading and when CPU memory is bottlenecked. Memory is saved by creating pinned tensors on the fly instead of pre-pinning them. However, this may increase overall execution time.
 
 <Tip>
 
-- Group offloading may not work with all models out-of-the-box. If the forward implementations of the model contain weight-dependent device-casting of inputs, it may clash with the offloading mechanism's handling of device-casting.
-- The `offload_type` parameter can be set to either `block_level` or `leaf_level`. `block_level` offloads groups of `torch::nn::ModuleList` or `torch::nn:Sequential` modules based on a configurable attribute `num_blocks_per_group`. For example, if you set `num_blocks_per_group=2` on a standard transformer model containing 40 layers, it will onload/offload 2 layers at a time for a total of 20 onload/offloads. This drastically reduces the VRAM requirements. `leaf_level` offloads individual layers at the lowest level, which is equivalent to sequential offloading. However, unlike sequential offloading, group offloading can be made much faster when using streams, with minimal compromise to end-to-end generation time.
-- The `use_stream` parameter can be used with CUDA devices to enable prefetching layers for onload. It defaults to `False`. Layer prefetching allows overlapping computation and data transfer of model weights, which drastically reduces the overall execution time compared to other offloading methods. However, it can increase the CPU RAM usage significantly. Ensure that available CPU RAM that is at least twice the size of the model when setting `use_stream=True`. You can find more information about CUDA streams [here](https://pytorch.org/docs/stable/generated/torch.cuda.Stream.html)
-- If specifying `use_stream=True` on VAEs with tiling enabled, make sure to do a dummy forward pass (possibly with dummy inputs) before the actual inference to avoid device-mismatch errors. This may not work on all implementations. Please open an issue if you encounter any problems.
-- The parameter `low_cpu_mem_usage` can be set to `True` to reduce CPU memory usage when using streams for group offloading. This is useful when the CPU memory is the bottleneck, but it may counteract the benefits of using streams and increase the overall execution time. The CPU memory savings come from creating pinned-tensors on-the-fly instead of pre-pinning them. This parameter is better suited for using `leaf_level` offloading.
-- When using `use_stream=True`, users can additionally specify `record_stream=True` to get better speedups at the expense of slightly increased memory usage. Refer to the [official PyTorch docs](https://pytorch.org/docs/stable/generated/torch.Tensor.record_stream.html) to know more about this.
-
-For more information about available parameters and an explanation of how group offloading works, refer to [`~hooks.group_offloading.apply_group_offloading`].
+The offloading strategies can be combined with [quantization](../quantization/overview.md) to enable further memory savings. For image generation, combining [quantization and model offloading](#model-offloading) can often give the best trade-off between quality, speed, and memory. However, for video generation, as the models are more
+compute-bound, [group-offloading](#group-offloading) tends to be better. Group offloading provides considerable benefits when weight transfers can be overlapped with computation (must use streams). When applying group offloading with quantization on image generation models at typical resolutions (1024x1024, for example), it is usually not possible to *fully* overlap weight transfers if the compute kernel finishes faster, making it communication bound between CPU/GPU (due to device synchronizations).
 
 </Tip>
 
-## FP8 layerwise weight-casting
+## Layerwise casting
 
-PyTorch supports `torch.float8_e4m3fn` and `torch.float8_e5m2` as weight storage dtypes, but they can't be used for computation in many different tensor operations due to unimplemented kernel support. However, you can use these dtypes to store model weights in fp8 precision and upcast them on-the-fly when the layers are used in the forward pass. This is known as layerwise weight-casting.
+Layerwise casting stores weights in a smaller data format (for example, `torch.float8_e4m3fn` and `torch.float8_e5m2`) to use less memory and upcasts those weights to a higher precision like `torch.float16` or `torch.bfloat16` for computation. Certain layers (normalization and modulation related weights) are skipped because storing them in fp8 can degrade generation quality.
 
-Typically, inference on most models is done with `torch.float16` or `torch.bfloat16` weight/computation precision. Layerwise weight-casting cuts down the memory footprint of the model weights by approximately half.
+> [!WARNING]
+> Layerwise casting may not work with all models if the forward implementation contains internal typecasting of weights. The current implementation of layerwise casting assumes the forward pass is independent of the weight precision and the input datatypes are always specified in `compute_dtype` (see [here](https://github.com/huggingface/transformers/blob/7f5077e53682ca855afc826162b204ebf809f1f9/src/transformers/models/t5/modeling_t5.py#L294-L299) for an incompatible implementation).
+>
+> Layerwise casting may also fail on custom modeling implementations with [PEFT](https://huggingface.co/docs/peft/index) layers. There are some checks available but they are not extensively tested or guaranteed to work in all cases.
 
-```python
+Call [`~ModelMixin.enable_layerwise_casting`] to set the storage and computation datatypes.
+
+```py
 import torch
 from diffusers import CogVideoXPipeline, CogVideoXTransformer3DModel
 from diffusers.utils import export_to_video
 
-model_id = "THUDM/CogVideoX-5b"
-
-# Load the model in bfloat16 and enable layerwise casting
-transformer = CogVideoXTransformer3DModel.from_pretrained(model_id, subfolder="transformer", torch_dtype=torch.bfloat16)
+transformer = CogVideoXTransformer3DModel.from_pretrained(
+    "THUDM/CogVideoX-5b",
+    subfolder="transformer",
+    torch_dtype=torch.bfloat16
+)
 transformer.enable_layerwise_casting(storage_dtype=torch.float8_e4m3fn, compute_dtype=torch.bfloat16)
 
-# Load the pipeline
-pipe = CogVideoXPipeline.from_pretrained(model_id, transformer=transformer, torch_dtype=torch.bfloat16)
-pipe.to("cuda")
-
+pipeline = CogVideoXPipeline.from_pretrained("THUDM/CogVideoX-5b",
+    transformer=transformer,
+    torch_dtype=torch.bfloat16
+).to("cuda")
 prompt = (
     "A panda, dressed in a small, red jacket and a tiny hat, sits on a wooden stool in a serene bamboo forest. "
     "The panda's fluffy paws strum a miniature acoustic guitar, producing soft, melodic tunes. Nearby, a few other "
@@ -243,43 +337,53 @@ prompt = (
     "The background includes a small, flowing stream and vibrant green foliage, enhancing the peaceful and magical "
     "atmosphere of this unique musical performance."
 )
-video = pipe(prompt=prompt, guidance_scale=6, num_inference_steps=50).frames[0]
+video = pipeline(prompt=prompt, guidance_scale=6, num_inference_steps=50).frames[0]
+print(f"Max memory reserved: {torch.cuda.max_memory_allocated() / 1024**3:.2f} GB")
 export_to_video(video, "output.mp4", fps=8)
 ```
 
-In the above example, layerwise casting is enabled on the transformer component of the pipeline. By default, certain layers are skipped from the FP8 weight casting because it can lead to significant degradation of generation quality. The normalization and modulation related weight parameters are also skipped by default.
-
-However, you gain more control and flexibility by directly utilizing the [`~hooks.layerwise_casting.apply_layerwise_casting`] function instead of [`~ModelMixin.enable_layerwise_casting`].
-
-<Tip>
-
-- Layerwise casting may not work with all models out-of-the-box. Sometimes, the forward implementations of the model might contain internal typecasting of weight values. Such implementations are not supported due to the currently simplistic implementation of layerwise casting, which assumes that the forward pass is independent of the weight precision and that the input dtypes are always in `compute_dtype`. An example of an incompatible implementation can be found [here](https://github.com/huggingface/transformers/blob/7f5077e53682ca855afc826162b204ebf809f1f9/src/transformers/models/t5/modeling_t5.py#L294-L299).
-- Layerwise casting may fail on custom modeling implementations that make use of [PEFT](https://github.com/huggingface/peft) layers. Some minimal checks to handle this case is implemented but is not extensively tested or guaranteed to work in all cases.
-- It can be also be applied partially to specific layers of a model. Partially applying layerwise casting can either be done manually by calling the `apply_layerwise_casting` function on specific internal modules, or by specifying the `skip_modules_pattern` and `skip_modules_classes` parameters for a root module. These parameters are particularly useful for layers such as normalization and modulation.
-
-</Tip>
-
-## Channels-last memory format
-
-The channels-last memory format is an alternative way of ordering NCHW tensors in memory to preserve dimension ordering. Channels-last tensors are ordered in such a way that the channels become the densest dimension (storing images pixel-per-pixel). Since not all operators currently support the channels-last format, it may result in worst performance but you should still try and see if it works for your model.
-
-For example, to set the pipeline's UNet to use the channels-last format:
+The [`~hooks.apply_layerwise_casting`] method can also be used if you need more control and flexibility. It can be partially applied to model layers by calling it on specific internal modules. Use the `skip_modules_pattern` or `skip_modules_classes` parameters to specify modules to avoid, such as the normalization and modulation layers.
 
 ```python
-print(pipe.unet.conv_out.state_dict()["weight"].stride())  # (2880, 9, 3, 1)
-pipe.unet.to(memory_format=torch.channels_last)  # in-place operation
+import torch
+from diffusers import CogVideoXTransformer3DModel
+from diffusers.hooks import apply_layerwise_casting
+
+transformer = CogVideoXTransformer3DModel.from_pretrained(
+    "THUDM/CogVideoX-5b",
+    subfolder="transformer",
+    torch_dtype=torch.bfloat16
+)
+
+# skip the normalization layer
+apply_layerwise_casting(
+    transformer,
+    storage_dtype=torch.float8_e4m3fn,
+    compute_dtype=torch.bfloat16,
+    skip_modules_classes=["norm"],
+    non_blocking=True,
+)
+```
+
+## torch.channels_last
+
+[torch.channels_last](https://pytorch.org/tutorials/intermediate/memory_format_tutorial.html) flips how tensors are stored from `(batch size, channels, height, width)` to `(batch size, heigh, width, channels)`. This aligns the tensors with how the hardware sequentially accesses the tensors stored in memory and avoids skipping around in memory to access the pixel values.
+
+Not all operators currently support the channels-last format and may result in worst performance, but it is still worth trying.
+
+```py
+print(pipeline.unet.conv_out.state_dict()["weight"].stride())  # (2880, 9, 3, 1)
+pipeline.unet.to(memory_format=torch.channels_last)  # in-place operation
 print(
-    pipe.unet.conv_out.state_dict()["weight"].stride()
+    pipeline.unet.conv_out.state_dict()["weight"].stride()
 )  # (2880, 1, 960, 320) having a stride of 1 for the 2nd dimension proves that it works
 ```
 
-## Tracing
+## torch.jit.trace
 
-Tracing runs an example input tensor through the model and captures the operations that are performed on it as that input makes its way through the model's layers. The executable or `ScriptFunction` that is returned is optimized with just-in-time compilation.
+[torch.jit.trace](https://pytorch.org/docs/stable/generated/torch.jit.trace.html) records the operations a model performs on a sample input and creates a new, optimized representation of the model based on the recorded execution path. During tracing, the model is optimized to reduce overhead from Python and dynamic control flows and operations are fused together for more efficiency. The returned executable or [ScriptFunction](https://pytorch.org/docs/stable/generated/torch.jit.ScriptFunction.html) can be compiled.
 
-To trace a UNet:
-
-```python
+```py
 import time
 import torch
 from diffusers import StableDiffusionPipeline
@@ -292,8 +396,7 @@ torch.set_grad_enabled(False)
 n_experiments = 2
 unet_runs_per_experiment = 50
 
-
-# load inputs
+# load sample inputs
 def generate_inputs():
     sample = torch.randn((2, 4, 64, 64), device="cuda", dtype=torch.float16)
     timestep = torch.rand(1, device="cuda", dtype=torch.float16) * 999
@@ -301,12 +404,12 @@ def generate_inputs():
     return sample, timestep, encoder_hidden_states
 
 
-pipe = StableDiffusionPipeline.from_pretrained(
+pipeline = StableDiffusionPipeline.from_pretrained(
     "stable-diffusion-v1-5/stable-diffusion-v1-5",
     torch_dtype=torch.float16,
     use_safetensors=True,
 ).to("cuda")
-unet = pipe.unet
+unet = pipeline.unet
 unet.eval()
 unet.to(memory_format=torch.channels_last)  # use channels_last memory format
 unet.forward = functools.partial(unet.forward, return_dict=False)  # set return_dict=False as default
@@ -323,14 +426,12 @@ unet_traced = torch.jit.trace(unet, inputs)
 unet_traced.eval()
 print("done tracing")
 
-
 # warmup and optimize graph
 for _ in range(5):
     with torch.inference_mode():
         inputs = generate_inputs()
         orig_output = unet_traced(*inputs)
 
-
 # benchmarking
 with torch.inference_mode():
     for _ in range(n_experiments):
@@ -352,20 +453,18 @@ with torch.inference_mode():
 unet_traced.save("unet_traced.pt")
 ```
 
-Replace the `unet` attribute of the pipeline with the traced model:
+Replace the pipeline's UNet with the traced version.
 
-```python
-from diffusers import StableDiffusionPipeline
+```py
 import torch
+from diffusers import StableDiffusionPipeline
 from dataclasses import dataclass
 
-
 @dataclass
 class UNet2DConditionOutput:
     sample: torch.Tensor
 
-
-pipe = StableDiffusionPipeline.from_pretrained(
+pipeline = StableDiffusionPipeline.from_pretrained(
     "stable-diffusion-v1-5/stable-diffusion-v1-5",
     torch_dtype=torch.float16,
     use_safetensors=True,
@@ -374,8 +473,7 @@ pipe = StableDiffusionPipeline.from_pretrained(
 # use jitted unet
 unet_traced = torch.jit.load("unet_traced.pt")
 
-
-# del pipe.unet
+# del pipeline.unet
 class TracedUNet(torch.nn.Module):
     def __init__(self):
         super().__init__()
@@ -386,8 +484,7 @@ class TracedUNet(torch.nn.Module):
         sample = unet_traced(latent_model_input, t, encoder_hidden_states)[0]
         return UNet2DConditionOutput(sample=sample)
 
-
-pipe.unet = TracedUNet()
+pipeline.unet = TracedUNet()
 
 with torch.inference_mode():
     image = pipe([prompt] * 1, num_inference_steps=50).images[0]
@@ -395,39 +492,31 @@ with torch.inference_mode():
 
 ## Memory-efficient attention
 
-Recent work on optimizing bandwidth in the attention block has generated huge speed-ups and reductions in GPU memory usage. The most recent type of memory-efficient attention is [Flash Attention](https://arxiv.org/abs/2205.14135) (you can check out the original code at [HazyResearch/flash-attention](https://github.com/HazyResearch/flash-attention)).
+> [!TIP]
+> Memory-efficient attention optimizes for memory usage *and* [inference speed](./fp16#scaled-dot-product-attention!
 
-<Tip>
+The Transformers attention mechanism is memory-intensive, especially for long sequences, so you can try using different and more memory-efficient attention types.
 
-If you have PyTorch >= 2.0 installed, you should not expect a speed-up for inference when enabling `xformers`.
+By default, if PyTorch >= 2.0 is installed, [scaled dot-product attention (SDPA)](https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html) is used. You don't need to make any additional changes to your code.
 
-</Tip>
+SDPA supports [FlashAttention](https://github.com/Dao-AILab/flash-attention) and [xFormers](https://github.com/facebookresearch/xformers) as well as a native C++ PyTorch implementation. It automatically selects the most optimal implementation based on your input.
 
-To use Flash Attention, install the following:
+You can explicitly use xFormers with the [`~ModelMixin.enable_xformers_memory_efficient_attention`] method.
 
-- PyTorch > 1.12
-- CUDA available
-- [xFormers](xformers)
-
-Then call [`~ModelMixin.enable_xformers_memory_efficient_attention`] on the pipeline:
-
-```python
-from diffusers import DiffusionPipeline
+```py
+# pip install xformers
 import torch
+from diffusers import StableDiffusionXLPipeline
 
-pipe = DiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
     torch_dtype=torch.float16,
-    use_safetensors=True,
 ).to("cuda")
-
-pipe.enable_xformers_memory_efficient_attention()
-
-with torch.inference_mode():
-    sample = pipe("a small cat")
-
-# optional: You can disable it via
-# pipe.disable_xformers_memory_efficient_attention()
+pipeline.enable_xformers_memory_efficient_attention()
 ```
 
-The iteration speed when using `xformers` should match the iteration speed of PyTorch 2.0 as described [here](torch2.0).
+Call [`~ModelMixin.disable_xformers_memory_efficient_attention`] to disable it.
+
+```py
+pipeline.disable_xformers_memory_efficient_attention()
+```

docs/source/en/optimization/pruna.md

@@ -0,0 +1,187 @@
+# Pruna
+
+[Pruna](https://github.com/PrunaAI/pruna) is a model optimization framework that offers various optimization methods - quantization, pruning, caching, compilation - for accelerating inference and reducing memory usage. A general overview of the optimization methods are shown below.
+
+
+| Technique    | Description                                                                                   | Speed | Memory | Quality |
+|--------------|-----------------------------------------------------------------------------------------------|:-----:|:------:|:-------:|
+| `batcher`    | Groups multiple inputs together to be processed simultaneously, improving computational efficiency and reducing processing time. | ✅    | ❌     | ➖      |
+| `cacher`     | Stores intermediate results of computations to speed up subsequent operations.               | ✅    | ➖     | ➖      |
+| `compiler`   | Optimises the model with instructions for specific hardware.                                 | ✅    | ➖     | ➖      |
+| `distiller`  | Trains a smaller, simpler model to mimic a larger, more complex model.                       | ✅    | ✅     | ❌      |
+| `quantizer`  | Reduces the precision of weights and activations, lowering memory requirements.              | ✅    | ✅     | ❌      |
+| `pruner`     | Removes less important or redundant connections and neurons, resulting in a sparser, more efficient network. | ✅    | ✅     | ❌      |
+| `recoverer`  | Restores the performance of a model after compression.                                       | ➖    | ➖     | ✅      |
+| `factorizer` | Factorization batches several small matrix multiplications into one large fused operation. | ✅ | ➖ | ➖ |
+| `enhancer`   | Enhances the model output by applying post-processing algorithms such as denoising or upscaling. | ❌ | - | ✅ |
+
+✅ (improves), ➖ (approx. the same), ❌ (worsens)
+
+Explore the full range of optimization methods in the [Pruna documentation](https://docs.pruna.ai/en/stable/docs_pruna/user_manual/configure.html#configure-algorithms).
+
+## Installation
+
+Install Pruna with the following command.
+
+```bash
+pip install pruna
+```
+
+
+## Optimize Diffusers models
+
+A broad range of optimization algorithms are supported for Diffusers models as shown below.
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/PrunaAI/documentation-images/resolve/main/diffusers/diffusers_combinations.png" alt="Overview of the supported optimization algorithms for diffusers models">
+</div>
+
+The example below optimizes [black-forest-labs/FLUX.1-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev)
+with a combination of factorizer, compiler, and cacher algorithms. This combination accelerates inference by up to 4.2x and cuts peak GPU memory usage from 34.7GB to 28.0GB, all while maintaining virtually the same output quality.
+
+> [!TIP]
+> Refer to the [Pruna optimization](https://docs.pruna.ai/en/stable/docs_pruna/user_manual/configure.html) docs to learn more about the optimization techniques used in this example.
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/PrunaAI/documentation-images/resolve/main/diffusers/flux_combination.png" alt="Optimization techniques used for FLUX.1-dev showing the combination of factorizer, compiler, and cacher algorithms">
+</div>
+
+Start by defining a `SmashConfig` with the optimization algorithms to use. To optimize the model, wrap the pipeline and the `SmashConfig` with `smash` and then use the pipeline as normal for inference.
+
+```python
+import torch
+from diffusers import FluxPipeline
+
+from pruna import PrunaModel, SmashConfig, smash
+
+# load the model
+# Try segmind/Segmind-Vega or black-forest-labs/FLUX.1-schnell with a small GPU memory
+pipe = FluxPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    torch_dtype=torch.bfloat16
+).to("cuda")
+
+# define the configuration
+smash_config = SmashConfig()
+smash_config["factorizer"] = "qkv_diffusers"
+smash_config["compiler"] = "torch_compile"
+smash_config["torch_compile_target"] = "module_list"
+smash_config["cacher"] = "fora"
+smash_config["fora_interval"] = 2
+
+# for the best results in terms of speed you can add these configs
+# however they will increase your warmup time from 1.5 min to 10 min
+# smash_config["torch_compile_mode"] = "max-autotune-no-cudagraphs"
+# smash_config["quantizer"] = "torchao"
+# smash_config["torchao_quant_type"] = "fp8dq"
+# smash_config["torchao_excluded_modules"] = "norm+embedding"
+
+# optimize the model
+smashed_pipe = smash(pipe, smash_config)
+
+# run the model
+smashed_pipe("a knitted purple prune").images[0]
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/PrunaAI/documentation-images/resolve/main/diffusers/flux_smashed_comparison.png">
+</div>
+
+After optimization, we can share and load the optimized model using the Hugging Face Hub.
+
+```python
+# save the model
+smashed_pipe.save_to_hub("<username>/FLUX.1-dev-smashed")
+
+# load the model
+smashed_pipe = PrunaModel.from_hub("<username>/FLUX.1-dev-smashed")
+```
+
+## Evaluate and benchmark Diffusers models
+
+Pruna provides the [EvaluationAgent](https://docs.pruna.ai/en/stable/docs_pruna/user_manual/evaluate.html) to evaluate the quality of your optimized models.
+
+We can metrics we care about, such as total time and throughput, and the dataset to evaluate on. We can define a model and pass it to the `EvaluationAgent`.
+
+<hfoptions id="eval">
+<hfoption id="optimized model">
+
+We can load and evaluate an optimized model by using the `EvaluationAgent` and pass it to the `Task`.
+
+```python
+import torch
+from diffusers import FluxPipeline
+
+from pruna import PrunaModel
+from pruna.data.pruna_datamodule import PrunaDataModule
+from pruna.evaluation.evaluation_agent import EvaluationAgent
+from pruna.evaluation.metrics import (
+    ThroughputMetric,
+    TorchMetricWrapper,
+    TotalTimeMetric,
+)
+from pruna.evaluation.task import Task
+
+# define the device
+device = "cuda" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu"
+
+# load the model
+# Try PrunaAI/Segmind-Vega-smashed or PrunaAI/FLUX.1-dev-smashed with a small GPU memory
+smashed_pipe = PrunaModel.from_hub("PrunaAI/FLUX.1-dev-smashed")
+
+# Define the metrics
+metrics = [
+    TotalTimeMetric(n_iterations=20, n_warmup_iterations=5),
+    ThroughputMetric(n_iterations=20, n_warmup_iterations=5),
+    TorchMetricWrapper("clip"),
+]
+
+# Define the datamodule
+datamodule = PrunaDataModule.from_string("LAION256")
+datamodule.limit_datasets(10)
+
+# Define the task and evaluation agent
+task = Task(metrics, datamodule=datamodule, device=device)
+eval_agent = EvaluationAgent(task)
+
+# Evaluate smashed model and offload it to CPU
+smashed_pipe.move_to_device(device)
+smashed_pipe_results = eval_agent.evaluate(smashed_pipe)
+smashed_pipe.move_to_device("cpu")
+```
+
+</hfoption>
+<hfoption id="standalone model">
+
+Instead of comparing the optimized model to the base model, you can also evaluate the standalone `diffusers` model. This is useful if you want to evaluate the performance of the model without the optimization. We can do so by using the `PrunaModel` wrapper and run the `EvaluationAgent` on it.
+
+```python
+import torch
+from diffusers import FluxPipeline
+
+from pruna import PrunaModel
+
+# load the model
+# Try PrunaAI/Segmind-Vega-smashed or PrunaAI/FLUX.1-dev-smashed with a small GPU memory
+pipe = FluxPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    torch_dtype=torch.bfloat16
+).to("cpu")
+wrapped_pipe = PrunaModel(model=pipe)
+```
+
+</hfoption>
+</hfoptions>
+
+Now that you have seen how to optimize and evaluate your models, you can start using Pruna to optimize your own models. Luckily, we have many examples to help you get started.
+
+> [!TIP]
+> For more details about benchmarking Flux, check out the [Announcing FLUX-Juiced: The Fastest Image Generation Endpoint (2.6 times faster)!](https://huggingface.co/blog/PrunaAI/flux-fastest-image-generation-endpoint) blog post and the [InferBench](https://huggingface.co/spaces/PrunaAI/InferBench) Space.
+
+## Reference
+
+- [Pruna](https://github.com/pruna-ai/pruna)
+- [Pruna optimization](https://docs.pruna.ai/en/stable/docs_pruna/user_manual/configure.html#configure-algorithms)
+- [Pruna evaluation](https://docs.pruna.ai/en/stable/docs_pruna/user_manual/evaluate.html)
+- [Pruna tutorials](https://docs.pruna.ai/en/stable/docs_pruna/tutorials/index.html)
+

docs/source/en/optimization/tome.md

@@ -93,4 +93,4 @@ To reproduce this benchmark, feel free to use this [script](https://gist.github.
 |          |                |              2 |         OOM |             13 |               10.78 |
 |          |                |              1 |         OOM |           6.66 |                5.54 |
 
-As seen in the tables above, the speed-up from `tomesd` becomes more pronounced for larger image resolutions. It is also interesting to note that with `tomesd`, it is possible to run the pipeline on a higher resolution like 1024x1024. You may be able to speed-up inference even more with [`torch.compile`](torch2.0).
+As seen in the tables above, the speed-up from `tomesd` becomes more pronounced for larger image resolutions. It is also interesting to note that with `tomesd`, it is possible to run the pipeline on a higher resolution like 1024x1024. You may be able to speed-up inference even more with [`torch.compile`](fp16#torchcompile).

docs/source/en/optimization/torch2.0.md

@@ -1,421 +0,0 @@
-<!--Copyright 2024 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.
--->
-
-# PyTorch 2.0
-
-🤗 Diffusers supports the latest optimizations from [PyTorch 2.0](https://pytorch.org/get-started/pytorch-2.0/) which include:
-
-1. A memory-efficient attention implementation, scaled dot product attention, without requiring any extra dependencies such as xFormers.
-2. [`torch.compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html), a just-in-time (JIT) compiler to provide an extra performance boost when individual models are compiled.
-
-Both of these optimizations require PyTorch 2.0 or later and 🤗 Diffusers > 0.13.0.
-
-```bash
-pip install --upgrade torch diffusers
-```
-
-## Scaled dot product attention
-
-[`torch.nn.functional.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention) (SDPA) is an optimized and memory-efficient attention (similar to xFormers) that automatically enables several other optimizations depending on the model inputs and GPU type. SDPA is enabled by default if you're using PyTorch 2.0 and the latest version of 🤗 Diffusers, so you don't need to add anything to your code.
-
-However, if you want to explicitly enable it, you can set a [`DiffusionPipeline`] to use [`~models.attention_processor.AttnProcessor2_0`]:
-
-```diff
-  import torch
-  from diffusers import DiffusionPipeline
-+ from diffusers.models.attention_processor import AttnProcessor2_0
-
-  pipe = DiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True).to("cuda")
-+ pipe.unet.set_attn_processor(AttnProcessor2_0())
-
-  prompt = "a photo of an astronaut riding a horse on mars"
-  image = pipe(prompt).images[0]
-```
-
-SDPA should be as fast and memory efficient as `xFormers`; check the [benchmark](#benchmark) for more details.
-
-In some cases - such as making the pipeline more deterministic or converting it to other formats - it may be helpful to use the vanilla attention processor, [`~models.attention_processor.AttnProcessor`]. To revert to [`~models.attention_processor.AttnProcessor`], call the [`~UNet2DConditionModel.set_default_attn_processor`] function on the pipeline:
-
-```diff
-  import torch
-  from diffusers import DiffusionPipeline
-
-  pipe = DiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True).to("cuda")
-+ pipe.unet.set_default_attn_processor()
-
-  prompt = "a photo of an astronaut riding a horse on mars"
-  image = pipe(prompt).images[0]
-```
-
-## torch.compile
-
-The `torch.compile` function can often provide an additional speed-up to your PyTorch code. In 🤗 Diffusers, it is usually best to wrap the UNet with `torch.compile` because it does most of the heavy lifting in the pipeline.
-
-```python
-from diffusers import DiffusionPipeline
-import torch
-
-pipe = DiffusionPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True).to("cuda")
-pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
-images = pipe(prompt, num_inference_steps=steps, num_images_per_prompt=batch_size).images[0]
-```
-
-Depending on GPU type, `torch.compile` can provide an *additional speed-up* of **5-300x** on top of SDPA! If you're using more recent GPU architectures such as Ampere (A100, 3090), Ada (4090), and Hopper (H100), `torch.compile` is able to squeeze even more performance out of these GPUs.
-
-Compilation requires some time to complete, so it is best suited for situations where you prepare your pipeline once and then perform the same type of inference operations multiple times. For example, calling the compiled pipeline on a different image size triggers compilation again which can be expensive.
-
-For more information and different options about `torch.compile`, refer to the [`torch_compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) tutorial.
-
-> [!TIP]
-> Learn more about other ways PyTorch 2.0 can help optimize your model in the [Accelerate inference of text-to-image diffusion models](../tutorials/fast_diffusion) tutorial.
-
-## Benchmark
-
-We conducted a comprehensive benchmark with PyTorch 2.0's efficient attention implementation and `torch.compile` across different GPUs and batch sizes for five of our most used pipelines. The code is benchmarked on 🤗 Diffusers v0.17.0.dev0 to optimize `torch.compile` usage (see [here](https://github.com/huggingface/diffusers/pull/3313) for more details).
-
-Expand the dropdown below to find the code used to benchmark each pipeline:
-
-<details>
-
-### Stable Diffusion text-to-image
-
-```python
-from diffusers import DiffusionPipeline
-import torch
-
-path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
-
-run_compile = True  # Set True / False
-
-pipe = DiffusionPipeline.from_pretrained(path, torch_dtype=torch.float16, use_safetensors=True)
-pipe = pipe.to("cuda")
-pipe.unet.to(memory_format=torch.channels_last)
-
-if run_compile:
-    print("Run torch compile")
-    pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
-
-prompt = "ghibli style, a fantasy landscape with castles"
-
-for _ in range(3):
-    images = pipe(prompt=prompt).images
-```
-
-### Stable Diffusion image-to-image
-
-```python
-from diffusers import StableDiffusionImg2ImgPipeline
-from diffusers.utils import load_image
-import torch
-
-url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
-
-init_image = load_image(url)
-init_image = init_image.resize((512, 512))
-
-path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
-
-run_compile = True  # Set True / False
-
-pipe = StableDiffusionImg2ImgPipeline.from_pretrained(path, torch_dtype=torch.float16, use_safetensors=True)
-pipe = pipe.to("cuda")
-pipe.unet.to(memory_format=torch.channels_last)
-
-if run_compile:
-    print("Run torch compile")
-    pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
-
-prompt = "ghibli style, a fantasy landscape with castles"
-
-for _ in range(3):
-    image = pipe(prompt=prompt, image=init_image).images[0]
-```
-
-### Stable Diffusion inpainting
-
-```python
-from diffusers import StableDiffusionInpaintPipeline
-from diffusers.utils import load_image
-import torch
-
-img_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo.png"
-mask_url = "https://raw.githubusercontent.com/CompVis/latent-diffusion/main/data/inpainting_examples/overture-creations-5sI6fQgYIuo_mask.png"
-
-init_image = load_image(img_url).resize((512, 512))
-mask_image = load_image(mask_url).resize((512, 512))
-
-path = "runwayml/stable-diffusion-inpainting"
-
-run_compile = True  # Set True / False
-
-pipe = StableDiffusionInpaintPipeline.from_pretrained(path, torch_dtype=torch.float16, use_safetensors=True)
-pipe = pipe.to("cuda")
-pipe.unet.to(memory_format=torch.channels_last)
-
-if run_compile:
-    print("Run torch compile")
-    pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
-
-prompt = "ghibli style, a fantasy landscape with castles"
-
-for _ in range(3):
-    image = pipe(prompt=prompt, image=init_image, mask_image=mask_image).images[0]
-```
-
-### ControlNet
-
-```python
-from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
-from diffusers.utils import load_image
-import torch
-
-url = "https://raw.githubusercontent.com/CompVis/stable-diffusion/main/assets/stable-samples/img2img/sketch-mountains-input.jpg"
-
-init_image = load_image(url)
-init_image = init_image.resize((512, 512))
-
-path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
-
-run_compile = True  # Set True / False
-controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionControlNetPipeline.from_pretrained(
-    path, controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True
-)
-
-pipe = pipe.to("cuda")
-pipe.unet.to(memory_format=torch.channels_last)
-pipe.controlnet.to(memory_format=torch.channels_last)
-
-if run_compile:
-    print("Run torch compile")
-    pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
-    pipe.controlnet = torch.compile(pipe.controlnet, mode="reduce-overhead", fullgraph=True)
-
-prompt = "ghibli style, a fantasy landscape with castles"
-
-for _ in range(3):
-    image = pipe(prompt=prompt, image=init_image).images[0]
-```
-
-### DeepFloyd IF text-to-image + upscaling
-
-```python
-from diffusers import DiffusionPipeline
-import torch
-
-run_compile = True  # Set True / False
-
-pipe_1 = DiffusionPipeline.from_pretrained("DeepFloyd/IF-I-M-v1.0", variant="fp16", text_encoder=None, torch_dtype=torch.float16, use_safetensors=True)
-pipe_1.to("cuda")
-pipe_2 = DiffusionPipeline.from_pretrained("DeepFloyd/IF-II-M-v1.0", variant="fp16", text_encoder=None, torch_dtype=torch.float16, use_safetensors=True)
-pipe_2.to("cuda")
-pipe_3 = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-x4-upscaler", torch_dtype=torch.float16, use_safetensors=True)
-pipe_3.to("cuda")
-
-
-pipe_1.unet.to(memory_format=torch.channels_last)
-pipe_2.unet.to(memory_format=torch.channels_last)
-pipe_3.unet.to(memory_format=torch.channels_last)
-
-if run_compile:
-    pipe_1.unet = torch.compile(pipe_1.unet, mode="reduce-overhead", fullgraph=True)
-    pipe_2.unet = torch.compile(pipe_2.unet, mode="reduce-overhead", fullgraph=True)
-    pipe_3.unet = torch.compile(pipe_3.unet, mode="reduce-overhead", fullgraph=True)
-
-prompt = "the blue hulk"
-
-prompt_embeds = torch.randn((1, 2, 4096), dtype=torch.float16)
-neg_prompt_embeds = torch.randn((1, 2, 4096), dtype=torch.float16)
-
-for _ in range(3):
-    image_1 = pipe_1(prompt_embeds=prompt_embeds, negative_prompt_embeds=neg_prompt_embeds, output_type="pt").images
-    image_2 = pipe_2(image=image_1, prompt_embeds=prompt_embeds, negative_prompt_embeds=neg_prompt_embeds, output_type="pt").images
-    image_3 = pipe_3(prompt=prompt, image=image_1, noise_level=100).images
-```
-</details>
-
-The graph below highlights the relative speed-ups for the [`StableDiffusionPipeline`] across five GPU families with PyTorch 2.0 and `torch.compile` enabled. The benchmarks for the following graphs are measured in *number of iterations/second*.
-
-![t2i_speedup](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/pt2_benchmarks/t2i_speedup.png)
-
-To give you an even better idea of how this speed-up holds for the other pipelines, consider the following
-graph for an A100 with PyTorch 2.0 and `torch.compile`:
-
-![a100_numbers](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/pt2_benchmarks/a100_numbers.png)
-
-In the following tables, we report our findings in terms of the *number of iterations/second*.
-
-### A100 (batch size: 1)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 21.66 | 23.13 | 44.03 | 49.74 |
-| SD - img2img | 21.81 | 22.40 | 43.92 | 46.32 |
-| SD - inpaint | 22.24 | 23.23 | 43.76 | 49.25 |
-| SD - controlnet | 15.02 | 15.82 | 32.13 | 36.08 |
-| IF | 20.21 / <br>13.84 / <br>24.00 | 20.12 / <br>13.70 / <br>24.03 | ❌ | 97.34 / <br>27.23 / <br>111.66 |
-| SDXL - txt2img | 8.64 | 9.9 | - | - |
-
-### A100 (batch size: 4)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 11.6 | 13.12 | 14.62 | 17.27 |
-| SD - img2img | 11.47 | 13.06 | 14.66 | 17.25 |
-| SD - inpaint | 11.67 | 13.31 | 14.88 | 17.48 |
-| SD - controlnet | 8.28 | 9.38 | 10.51 | 12.41 |
-| IF | 25.02 | 18.04 | ❌ | 48.47 |
-| SDXL - txt2img | 2.44 | 2.74 | - | - |
-
-### A100 (batch size: 16)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 3.04 | 3.6 | 3.83 | 4.68 |
-| SD - img2img | 2.98 | 3.58 | 3.83 | 4.67 |
-| SD - inpaint | 3.04 | 3.66 | 3.9 | 4.76 |
-| SD - controlnet | 2.15 | 2.58 | 2.74 | 3.35 |
-| IF | 8.78 | 9.82 | ❌ | 16.77 |
-| SDXL - txt2img | 0.64 | 0.72 | - | - |
-
-### V100 (batch size: 1)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 18.99 | 19.14 | 20.95 | 22.17 |
-| SD - img2img | 18.56 | 19.18 | 20.95 | 22.11 |
-| SD - inpaint | 19.14 | 19.06 | 21.08 | 22.20 |
-| SD - controlnet | 13.48 | 13.93 | 15.18 | 15.88 |
-| IF |  20.01 / <br>9.08 / <br>23.34 | 19.79 / <br>8.98 / <br>24.10 | ❌ | 55.75 / <br>11.57 / <br>57.67 |
-
-### V100 (batch size: 4)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 5.96 | 5.89 | 6.83 | 6.86 |
-| SD - img2img | 5.90 | 5.91 | 6.81 | 6.82 |
-| SD - inpaint | 5.99 | 6.03 | 6.93 | 6.95 |
-| SD - controlnet | 4.26 | 4.29 | 4.92 | 4.93 |
-| IF | 15.41 | 14.76 | ❌ | 22.95 |
-
-### V100 (batch size: 16)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 1.66 | 1.66 | 1.92 | 1.90 |
-| SD - img2img | 1.65 | 1.65 | 1.91 | 1.89 |
-| SD - inpaint | 1.69 | 1.69 | 1.95 | 1.93 |
-| SD - controlnet | 1.19 | 1.19 | OOM after warmup | 1.36 |
-| IF | 5.43 | 5.29 | ❌ | 7.06 |
-
-### T4 (batch size: 1)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 6.9 | 6.95 | 7.3 | 7.56 |
-| SD - img2img | 6.84 | 6.99 | 7.04 | 7.55 |
-| SD - inpaint | 6.91 | 6.7 | 7.01 | 7.37 |
-| SD - controlnet | 4.89 | 4.86 | 5.35 | 5.48 |
-| IF | 17.42 / <br>2.47 / <br>18.52 | 16.96 / <br>2.45 / <br>18.69 | ❌ | 24.63 / <br>2.47 / <br>23.39 |
-| SDXL - txt2img | 1.15 | 1.16 | - | - |
-
-### T4 (batch size: 4)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 1.79 | 1.79 | 2.03 | 1.99 |
-| SD - img2img | 1.77 | 1.77 | 2.05 | 2.04 |
-| SD - inpaint | 1.81 | 1.82 | 2.09 | 2.09 |
-| SD - controlnet | 1.34 | 1.27 | 1.47 | 1.46 |
-| IF | 5.79 |  5.61 | ❌ | 7.39 |
-| SDXL - txt2img | 0.288 | 0.289 | - | - |
-
-### T4 (batch size: 16)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 2.34s | 2.30s | OOM after 2nd iteration | 1.99s |
-| SD - img2img | 2.35s | 2.31s | OOM after warmup | 2.00s |
-| SD - inpaint | 2.30s | 2.26s | OOM after 2nd iteration | 1.95s |
-| SD - controlnet | OOM after 2nd iteration | OOM after 2nd iteration | OOM after warmup | OOM after warmup |
-| IF * | 1.44 | 1.44 | ❌ | 1.94 |
-| SDXL - txt2img | OOM | OOM | - | - |
-
-### RTX 3090 (batch size: 1)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 22.56 | 22.84 | 23.84 | 25.69 |
-| SD - img2img | 22.25 | 22.61 | 24.1 | 25.83 |
-| SD - inpaint | 22.22 | 22.54 | 24.26 | 26.02 |
-| SD - controlnet | 16.03 | 16.33 | 17.38 | 18.56 |
-| IF | 27.08 / <br>9.07 / <br>31.23 | 26.75 / <br>8.92 / <br>31.47 | ❌ | 68.08 / <br>11.16 / <br>65.29 |
-
-### RTX 3090 (batch size: 4)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 6.46 | 6.35 | 7.29 | 7.3 |
-| SD - img2img | 6.33 | 6.27 | 7.31 | 7.26 |
-| SD - inpaint | 6.47 | 6.4 | 7.44 | 7.39 |
-| SD - controlnet | 4.59 | 4.54 | 5.27 | 5.26 |
-| IF | 16.81 | 16.62 | ❌ | 21.57 |
-
-### RTX 3090 (batch size: 16)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 1.7 | 1.69 | 1.93 | 1.91 |
-| SD - img2img | 1.68 | 1.67 | 1.93 | 1.9 |
-| SD - inpaint | 1.72 | 1.71 | 1.97 | 1.94 |
-| SD - controlnet | 1.23 | 1.22 | 1.4 | 1.38 |
-| IF | 5.01 | 5.00 | ❌ | 6.33 |
-
-### RTX 4090 (batch size: 1)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 40.5 | 41.89 | 44.65 | 49.81 |
-| SD - img2img | 40.39 | 41.95 | 44.46 | 49.8 |
-| SD - inpaint | 40.51 | 41.88 | 44.58 | 49.72 |
-| SD - controlnet | 29.27 | 30.29 | 32.26 | 36.03 |
-| IF | 69.71 / <br>18.78 / <br>85.49 | 69.13 / <br>18.80 / <br>85.56 | ❌ | 124.60 / <br>26.37 / <br>138.79 |
-| SDXL - txt2img | 6.8 | 8.18 | - | - |
-
-### RTX 4090 (batch size: 4)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 12.62 | 12.84 | 15.32 | 15.59 |
-| SD - img2img | 12.61 | 12,.79 | 15.35 | 15.66 |
-| SD - inpaint | 12.65 | 12.81 | 15.3 | 15.58 |
-| SD - controlnet | 9.1 | 9.25 | 11.03 | 11.22 |
-| IF | 31.88 | 31.14 | ❌ | 43.92 |
-| SDXL - txt2img | 2.19 | 2.35 | - | - |
-
-### RTX 4090 (batch size: 16)
-
-| **Pipeline** | **torch 2.0 - <br>no compile** | **torch nightly - <br>no compile** | **torch 2.0 - <br>compile** | **torch nightly - <br>compile** |
-|:---:|:---:|:---:|:---:|:---:|
-| SD - txt2img | 3.17 | 3.2 | 3.84 | 3.85 |
-| SD - img2img | 3.16 | 3.2 | 3.84 | 3.85 |
-| SD - inpaint | 3.17 | 3.2 | 3.85 | 3.85 |
-| SD - controlnet | 2.23 | 2.3 | 2.7 | 2.75 |
-| IF | 9.26 | 9.2 | ❌ | 13.31 |
-| SDXL - txt2img | 0.52 | 0.53 | - | - |
-
-## Notes
-
-* Follow this [PR](https://github.com/huggingface/diffusers/pull/3313) for more details on the environment used for conducting the benchmarks.
-* For the DeepFloyd IF pipeline where batch sizes > 1, we only used a batch size of > 1 in the first IF pipeline for text-to-image generation and NOT for upscaling. That means the two upscaling pipelines received a batch size of 1.
-
-*Thanks to [Horace He](https://github.com/Chillee) from the PyTorch team for their support in improving our support of `torch.compile()` in Diffusers.*

docs/source/en/optimization/xdit.md

@@ -2,7 +2,7 @@
 
 [xDiT](https://github.com/xdit-project/xDiT) is an inference engine designed for the large scale parallel deployment of Diffusion Transformers (DiTs). xDiT provides a suite of efficient parallel approaches for Diffusion Models, as well as GPU kernel accelerations.
 
-There are four parallel methods supported in xDiT, including [Unified Sequence Parallelism](https://arxiv.org/abs/2405.07719), [PipeFusion](https://arxiv.org/abs/2405.14430), CFG parallelism and data parallelism. The four parallel methods in xDiT can be configured in a hybrid manner, optimizing communication patterns to best suit the underlying network hardware.
+There are four parallel methods supported in xDiT, including [Unified Sequence Parallelism](https://huggingface.co/papers/2405.07719), [PipeFusion](https://huggingface.co/papers/2405.14430), CFG parallelism and data parallelism. The four parallel methods in xDiT can be configured in a hybrid manner, optimizing communication patterns to best suit the underlying network hardware.
 
 Optimization orthogonal to parallelization focuses on accelerating single GPU performance. In addition to utilizing well-known Attention optimization libraries, we leverage compilation acceleration technologies such as torch.compile and onediff.
 
@@ -116,6 +116,6 @@ More detailed performance metric can be found on our [github page](https://githu
 
 [xDiT-project](https://github.com/xdit-project/xDiT)
 
-[USP: A Unified Sequence Parallelism Approach for Long Context Generative AI](https://arxiv.org/abs/2405.07719)
+[USP: A Unified Sequence Parallelism Approach for Long Context Generative AI](https://huggingface.co/papers/2405.07719)
 
-[PipeFusion: Displaced Patch Pipeline Parallelism for Inference of Diffusion Transformer Models](https://arxiv.org/abs/2405.14430)
+[PipeFusion: Displaced Patch Pipeline Parallelism for Inference of Diffusion Transformer Models](https://huggingface.co/papers/2405.14430)

docs/source/en/quantization/bitsandbytes.md

@@ -48,7 +48,7 @@ For Ada and higher-series GPUs. we recommend changing `torch_dtype` to `torch.bf
 ```py
 from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig
 from transformers import BitsAndBytesConfig as TransformersBitsAndBytesConfig
-
+import torch
 from diffusers import AutoModel
 from transformers import T5EncoderModel
 
@@ -88,6 +88,8 @@ Setting `device_map="auto"` automatically fills all available space on the GPU(s
 CPU, and finally, the hard drive (the absolute slowest option) if there is still not enough memory.
 
 ```py
+from diffusers import FluxPipeline
+
 pipe = FluxPipeline.from_pretrained(
     "black-forest-labs/FLUX.1-dev",
     transformer=transformer_8bit,
@@ -132,7 +134,7 @@ For Ada and higher-series GPUs. we recommend changing `torch_dtype` to `torch.bf
 ```py
 from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig
 from transformers import BitsAndBytesConfig as TransformersBitsAndBytesConfig
-
+import torch
 from diffusers import AutoModel
 from transformers import T5EncoderModel
 
@@ -171,6 +173,8 @@ Let's generate an image using our quantized models.
 Setting `device_map="auto"` automatically fills all available space on the GPU(s) first, then the CPU, and finally, the hard drive (the absolute slowest option) if there is still not enough memory.
 
 ```py
+from diffusers import FluxPipeline
+
 pipe = FluxPipeline.from_pretrained(
     "black-forest-labs/FLUX.1-dev",
     transformer=transformer_4bit,
@@ -214,6 +218,8 @@ Check your memory footprint with the `get_memory_footprint` method:
 print(model.get_memory_footprint())
 ```
 
+Note that this only tells you the memory footprint of the model params and does _not_ estimate the inference memory requirements.
+
 Quantized models can be loaded from the [`~ModelMixin.from_pretrained`] method without needing to specify the `quantization_config` parameters:
 
 ```py
@@ -410,7 +416,46 @@ text_encoder_2_4bit.dequantize()
 transformer_4bit.dequantize()
 ```
 
+## torch.compile
+
+Speed up inference with `torch.compile`. Make sure you have the latest `bitsandbytes` installed and we also recommend installing [PyTorch nightly](https://pytorch.org/get-started/locally/).
+
+<hfoptions id="bnb">
+<hfoption id="8-bit">
+```py
+torch._dynamo.config.capture_dynamic_output_shape_ops = True
+
+quant_config = DiffusersBitsAndBytesConfig(load_in_8bit=True)
+transformer_4bit = AutoModel.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    subfolder="transformer",
+    quantization_config=quant_config,
+    torch_dtype=torch.float16,
+)
+transformer_4bit.compile(fullgraph=True)
+```
+
+</hfoption>
+<hfoption id="4-bit">
+
+```py
+quant_config = DiffusersBitsAndBytesConfig(load_in_4bit=True)
+transformer_4bit = AutoModel.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    subfolder="transformer",
+    quantization_config=quant_config,
+    torch_dtype=torch.float16,
+)
+transformer_4bit.compile(fullgraph=True)
+```
+</hfoption>
+</hfoptions>
+
+On an RTX 4090 with compilation, 4-bit Flux generation completed in 25.809 seconds versus 32.570 seconds without.
+
+Check out the [benchmarking script](https://gist.github.com/sayakpaul/0db9d8eeeb3d2a0e5ed7cf0d9ca19b7d) for more details.
+
 ## Resources
 
 * [End-to-end notebook showing Flux.1 Dev inference in a free-tier Colab](https://gist.github.com/sayakpaul/c76bd845b48759e11687ac550b99d8b4)
-* [Training](https://gist.github.com/sayakpaul/05afd428bc089b47af7c016e42004527)
+* [Training](https://github.com/huggingface/diffusers/blob/8c661ea586bf11cb2440da740dd3c4cf84679b85/examples/dreambooth/README_hidream.md#using-quantization)

docs/source/en/quantization/overview.md

@@ -13,29 +13,120 @@ specific language governing permissions and limitations under the License.
 
 # Quantization
 
-Quantization techniques focus on representing data with less information while also trying to not lose too much accuracy. This often means converting a data type to represent the same information with fewer bits. For example, if your model weights are stored as 32-bit floating points and they're quantized to 16-bit floating points, this halves the model size which makes it easier to store and reduces memory-usage. Lower precision can also speedup inference because it takes less time to perform calculations with fewer bits.
+Quantization focuses on representing data with fewer bits while also trying to preserve the precision of the original data. This often means converting a data type to represent the same information with fewer bits. For example, if your model weights are stored as 32-bit floating points and they're quantized to 16-bit floating points, this halves the model size which makes it easier to store and reduces memory usage. Lower precision can also speedup inference because it takes less time to perform calculations with fewer bits.
 
-<Tip>
+Diffusers supports multiple quantization backends to make large diffusion models like [Flux](../api/pipelines/flux) more accessible. This guide shows how to use the [`~quantizers.PipelineQuantizationConfig`] class to quantize a pipeline during its initialization from a pretrained or non-quantized checkpoint.
 
-Interested in adding a new quantization method to Diffusers? Refer to the [Contribute new quantization method guide](https://huggingface.co/docs/transformers/main/en/quantization/contribute) to learn more about adding a new quantization method.
+## Pipeline-level quantization
 
-</Tip>
+There are two ways you can use [`~quantizers.PipelineQuantizationConfig`] depending on the level of control you want over the quantization specifications of each model in the pipeline.
 
-<Tip>
+- for more basic and simple use cases, you only need to define the `quant_backend`, `quant_kwargs`, and `components_to_quantize`
+- for more granular quantization control, provide a `quant_mapping` that provides the quantization specifications for the individual model components
 
-If you are new to the quantization field, we recommend you to check out these beginner-friendly courses about quantization in collaboration with DeepLearning.AI:
+### Simple quantization
 
-* [Quantization Fundamentals with Hugging Face](https://www.deeplearning.ai/short-courses/quantization-fundamentals-with-hugging-face/)
-* [Quantization in Depth](https://www.deeplearning.ai/short-courses/quantization-in-depth/)
+Initialize [`~quantizers.PipelineQuantizationConfig`] with the following parameters.
 
-</Tip>
+- `quant_backend` specifies which quantization backend to use. Currently supported backends include: `bitsandbytes_4bit`, `bitsandbytes_8bit`, `gguf`, `quanto`, and `torchao`.
+- `quant_kwargs` contains the specific quantization arguments to use.
+- `components_to_quantize` specifies which components of the pipeline to quantize. Typically, you should quantize the most compute intensive components like the transformer. The text encoder is another component to consider quantizing if a pipeline has more than one such as [`FluxPipeline`]. The example below quantizes the T5 text encoder in [`FluxPipeline`] while keeping the CLIP model intact.
 
-## When to use what?
+```py
+import torch
+from diffusers import DiffusionPipeline
+from diffusers.quantizers import PipelineQuantizationConfig
 
-Diffusers currently supports the following quantization methods.
-- [BitsandBytes](./bitsandbytes)
-- [TorchAO](./torchao)
-- [GGUF](./gguf)
-- [Quanto](./quanto.md)
+pipeline_quant_config = PipelineQuantizationConfig(
+    quant_backend="bitsandbytes_4bit",
+    quant_kwargs={"load_in_4bit": True, "bnb_4bit_quant_type": "nf4", "bnb_4bit_compute_dtype": torch.bfloat16},
+    components_to_quantize=["transformer", "text_encoder_2"],
+)
+```
 
-[This resource](https://huggingface.co/docs/transformers/main/en/quantization/overview#when-to-use-what) provides a good overview of the pros and cons of different quantization techniques.
+Pass the `pipeline_quant_config` to [`~DiffusionPipeline.from_pretrained`] to quantize the pipeline.
+
+```py
+pipe = DiffusionPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    quantization_config=pipeline_quant_config,
+    torch_dtype=torch.bfloat16,
+).to("cuda")
+
+image = pipe("photo of a cute dog").images[0]
+```
+
+### quant_mapping
+
+The `quant_mapping` argument provides more flexible options for how to quantize each individual component in a pipeline, like combining different quantization backends.
+
+Initialize [`~quantizers.PipelineQuantizationConfig`] and pass a `quant_mapping` to it. The `quant_mapping` allows you to specify the quantization options for each component in the pipeline such as the transformer and text encoder.
+
+The example below uses two quantization backends, [`~quantizers.QuantoConfig`] and [`transformers.BitsAndBytesConfig`], for the transformer and text encoder.
+
+```py
+import torch
+from diffusers import DiffusionPipeline
+from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig
+from diffusers.quantizers.quantization_config import QuantoConfig
+from diffusers.quantizers import PipelineQuantizationConfig
+from transformers import BitsAndBytesConfig as TransformersBitsAndBytesConfig
+
+pipeline_quant_config = PipelineQuantizationConfig(
+    quant_mapping={
+        "transformer": QuantoConfig(weights_dtype="int8"),
+        "text_encoder_2": TransformersBitsAndBytesConfig(
+            load_in_4bit=True, compute_dtype=torch.bfloat16
+        ),
+    }
+)
+```
+
+There is a separate bitsandbytes backend in [Transformers](https://huggingface.co/docs/transformers/main_classes/quantization#transformers.BitsAndBytesConfig). You need to import and use [`transformers.BitsAndBytesConfig`] for components that come from Transformers. For example, `text_encoder_2` in [`FluxPipeline`] is a [`~transformers.T5EncoderModel`] from Transformers so you need to use [`transformers.BitsAndBytesConfig`] instead of [`diffusers.BitsAndBytesConfig`].
+
+> [!TIP]
+> Use the [simple quantization](#simple-quantization) method above if you don't want to manage these distinct imports or aren't sure where each pipeline component comes from.
+
+```py
+import torch
+from diffusers import DiffusionPipeline
+from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig
+from diffusers.quantizers import PipelineQuantizationConfig
+from transformers import BitsAndBytesConfig as TransformersBitsAndBytesConfig
+
+pipeline_quant_config = PipelineQuantizationConfig(
+    quant_mapping={
+        "transformer": DiffusersBitsAndBytesConfig(load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16),
+        "text_encoder_2": TransformersBitsAndBytesConfig(
+            load_in_4bit=True, compute_dtype=torch.bfloat16
+        ),
+    }
+)
+```
+
+Pass the `pipeline_quant_config` to [`~DiffusionPipeline.from_pretrained`] to quantize the pipeline.
+
+```py
+pipe = DiffusionPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-dev",
+    quantization_config=pipeline_quant_config,
+    torch_dtype=torch.bfloat16,
+).to("cuda")
+
+image = pipe("photo of a cute dog").images[0]
+```
+
+## Resources
+
+Check out the resources below to learn more about quantization.
+
+- If you are new to quantization, we recommend checking out the following beginner-friendly courses in collaboration with DeepLearning.AI.
+
+    - [Quantization Fundamentals with Hugging Face](https://www.deeplearning.ai/short-courses/quantization-fundamentals-with-hugging-face/)
+    - [Quantization in Depth](https://www.deeplearning.ai/short-courses/quantization-in-depth/)
+
+- Refer to the [Contribute new quantization method guide](https://huggingface.co/docs/transformers/main/en/quantization/contribute) if you're interested in adding a new quantization method.
+
+- The Transformers quantization [Overview](https://huggingface.co/docs/transformers/quantization/overview#when-to-use-what) provides an overview of the pros and cons of different quantization backends.
+
+- Read the [Exploring Quantization Backends in Diffusers](https://huggingface.co/blog/diffusers-quantization) blog post for a brief introduction to each quantization backend, how to choose a backend, and combining quantization with other memory optimizations.

docs/source/en/quantization/torchao.md

@@ -56,7 +56,7 @@ image = pipe(
 image.save("output.png")
 ```
 
-TorchAO is fully compatible with [torch.compile](./optimization/torch2.0#torchcompile), setting it apart from other quantization methods. This makes it easy to speed up inference with just one line of code.
+TorchAO is fully compatible with [torch.compile](../optimization/fp16#torchcompile), setting it apart from other quantization methods. This makes it easy to speed up inference with just one line of code.
 
 ```python
 # In the above code, add the following after initializing the transformer
@@ -65,6 +65,9 @@ transformer = torch.compile(transformer, mode="max-autotune", fullgraph=True)
 
 For speed and memory benchmarks on Flux and CogVideoX, please refer to the table [here](https://github.com/huggingface/diffusers/pull/10009#issue-2688781450). You can also find some torchao [benchmarks](https://github.com/pytorch/ao/tree/main/torchao/quantization#benchmarks) numbers for various hardware.
 
+> [!TIP]
+> The FP8 post-training quantization schemes in torchao are effective for GPUs with compute capability of at least 8.9 (RTX-4090, Hopper, etc.). FP8 often provides the best speed, memory, and quality trade-off when generating images and videos. We recommend combining FP8 and torch.compile if your GPU is compatible.
+
 torchao also supports an automatic quantization API through [autoquant](https://github.com/pytorch/ao/blob/main/torchao/quantization/README.md#autoquantization). Autoquantization determines the best quantization strategy applicable to a model by comparing the performance of each technique on chosen input types and shapes. Currently, this can be used directly on the underlying modeling components. Diffusers will also expose an autoquant configuration option in the future.
 
 The `TorchAoConfig` class accepts three parameters:
@@ -85,13 +88,13 @@ The quantization methods supported are as follows:
 | **Category** | **Full Function Names** | **Shorthands** |
 |--------------|-------------------------|----------------|
 | **Integer quantization** | `int4_weight_only`, `int8_dynamic_activation_int4_weight`, `int8_weight_only`, `int8_dynamic_activation_int8_weight` | `int4wo`, `int4dq`, `int8wo`, `int8dq` |
-| **Floating point 8-bit quantization** | `float8_weight_only`, `float8_dynamic_activation_float8_weight`, `float8_static_activation_float8_weight` | `float8wo`, `float8wo_e5m2`, `float8wo_e4m3`, `float8dq`, `float8dq_e4m3`, `float8_e4m3_tensor`, `float8_e4m3_row` |
+| **Floating point 8-bit quantization** | `float8_weight_only`, `float8_dynamic_activation_float8_weight`, `float8_static_activation_float8_weight` | `float8wo`, `float8wo_e5m2`, `float8wo_e4m3`, `float8dq`, `float8dq_e4m3`, `float8dq_e4m3_tensor`, `float8dq_e4m3_row` |
 | **Floating point X-bit quantization** | `fpx_weight_only` | `fpX_eAwB` where `X` is the number of bits (1-7), `A` is exponent bits, and `B` is mantissa bits. Constraint: `X == A + B + 1` |
 | **Unsigned Integer quantization** | `uintx_weight_only` | `uint1wo`, `uint2wo`, `uint3wo`, `uint4wo`, `uint5wo`, `uint6wo`, `uint7wo` |
 
 Some quantization methods are aliases (for example, `int8wo` is the commonly used shorthand for `int8_weight_only`). This allows using the quantization methods described in the torchao docs as-is, while also making it convenient to remember their shorthand notations.
 
-Refer to the official torchao documentation for a better understanding of the available quantization methods and the exhaustive list of configuration options available.
+Refer to the [official torchao documentation](https://docs.pytorch.org/ao/stable/index.html) for a better understanding of the available quantization methods and the exhaustive list of configuration options available.
 
 ## Serializing and Deserializing quantized models
 
@@ -155,5 +158,5 @@ transformer.load_state_dict(state_dict, strict=True, assign=True)
 
 ## Resources
 
-- [TorchAO Quantization API](https://github.com/pytorch/ao/blob/main/torchao/quantization/README.md)
+- [TorchAO Quantization API](https://docs.pytorch.org/ao/stable/index.html)
 - [Diffusers-TorchAO examples](https://github.com/sayakpaul/diffusers-torchao)

docs/source/en/stable_diffusion.md

@@ -256,6 +256,6 @@ make_image_grid(images, 2, 2)
 
 In this tutorial, you learned how to optimize a [`DiffusionPipeline`] for computational and memory efficiency as well as improving the quality of generated outputs. If you're interested in making your pipeline even faster, take a look at the following resources:
 
-- Learn how [PyTorch 2.0](./optimization/torch2.0) and [`torch.compile`](https://pytorch.org/docs/stable/generated/torch.compile.html) can yield 5 - 300% faster inference speed. On an A100 GPU, inference can be up to 50% faster!
+- Learn how [PyTorch 2.0](./optimization/fp16) and [`torch.compile`](https://pytorch.org/docs/stable/generated/torch.compile.html) can yield 5 - 300% faster inference speed. On an A100 GPU, inference can be up to 50% faster!
 - If you can't use PyTorch 2, we recommend you install [xFormers](./optimization/xformers). Its memory-efficient attention mechanism works great with PyTorch 1.13.1 for faster speed and reduced memory consumption.
 - Other optimization techniques, such as model offloading, are covered in [this guide](./optimization/fp16).

docs/source/en/training/ddpo.md

@@ -12,6 +12,6 @@ specific language governing permissions and limitations under the License.
 
 # Reinforcement learning training with DDPO
 
-You can fine-tune Stable Diffusion on a reward function via reinforcement learning with the 🤗 TRL library and 🤗 Diffusers. This is done with the Denoising Diffusion Policy Optimization (DDPO) algorithm introduced by Black et al. in [Training Diffusion Models with Reinforcement Learning](https://arxiv.org/abs/2305.13301), which is implemented in 🤗 TRL with the [`~trl.DDPOTrainer`].
+You can fine-tune Stable Diffusion on a reward function via reinforcement learning with the 🤗 TRL library and 🤗 Diffusers. This is done with the Denoising Diffusion Policy Optimization (DDPO) algorithm introduced by Black et al. in [Training Diffusion Models with Reinforcement Learning](https://huggingface.co/papers/2305.13301), which is implemented in 🤗 TRL with the [`~trl.DDPOTrainer`].
 
 For more information, check out the [`~trl.DDPOTrainer`] API reference and the [Finetune Stable Diffusion Models with DDPO via TRL](https://huggingface.co/blog/trl-ddpo) blog post.

docs/source/en/training/lora.md

@@ -87,7 +87,7 @@ Lastly, if you want to train a model on your own dataset, take a look at the [Cr
 
 <Tip>
 
-The following sections highlight parts of the training script that are important for understanding how to modify it, but it doesn't cover every aspect of the script in detail. If you're interested in learning more, feel free to read through the [script](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/text_to_image_lora.py) and let us know if you have any questions or concerns.
+The following sections highlight parts of the training script that are important for understanding how to modify it, but it doesn't cover every aspect of the script in detail. If you're interested in learning more, feel free to read through the [script](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_lora.py) and let us know if you have any questions or concerns.
 
 </Tip>
 

docs/source/en/training/overview.md

@@ -59,5 +59,5 @@ pip install -r requirements_sdxl.txt
 
 To speedup training and reduce memory-usage, we recommend:
 
-- using PyTorch 2.0 or higher to automatically use [scaled dot product attention](../optimization/torch2.0#scaled-dot-product-attention) during training (you don't need to make any changes to the training code)
+- using PyTorch 2.0 or higher to automatically use [scaled dot product attention](../optimization/fp16#scaled-dot-product-attention) during training (you don't need to make any changes to the training code)
 - installing [xFormers](../optimization/xformers) to enable memory-efficient attention

docs/source/en/tutorials/fast_diffusion.md

@@ -1,322 +0,0 @@
-<!--Copyright 2024 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.
--->
-
-# Accelerate inference of text-to-image diffusion models
-
-Diffusion models are slower than their GAN counterparts because of the iterative and sequential reverse diffusion process. There are several techniques that can address this limitation such as progressive timestep distillation ([LCM LoRA](../using-diffusers/inference_with_lcm_lora)), model compression ([SSD-1B](https://huggingface.co/segmind/SSD-1B)), and reusing adjacent features of the denoiser ([DeepCache](../optimization/deepcache)).
-
-However, you don't necessarily need to use these techniques to speed up inference. With PyTorch 2 alone, you can accelerate the inference latency of text-to-image diffusion pipelines by up to 3x. This tutorial will show you how to progressively apply the optimizations found in PyTorch 2 to reduce inference latency. You'll use the [Stable Diffusion XL (SDXL)](../using-diffusers/sdxl) pipeline in this tutorial, but these techniques are applicable to other text-to-image diffusion pipelines too.
-
-Make sure you're using the latest version of Diffusers:
-
-```bash
-pip install -U diffusers
-```
-
-Then upgrade the other required libraries too:
-
-```bash
-pip install -U transformers accelerate peft
-```
-
-Install [PyTorch nightly](https://pytorch.org/) to benefit from the latest and fastest kernels:
-
-```bash
-pip3 install --pre torch --index-url https://download.pytorch.org/whl/nightly/cu121
-```
-
-> [!TIP]
-> The results reported below are from a 80GB 400W A100 with its clock rate set to the maximum.
-> If you're interested in the full benchmarking code, take a look at [huggingface/diffusion-fast](https://github.com/huggingface/diffusion-fast).
-
-
-## Baseline
-
-Let's start with a baseline. Disable reduced precision and the [`scaled_dot_product_attention` (SDPA)](../optimization/torch2.0#scaled-dot-product-attention) function which is automatically used by Diffusers:
-
-```python
-from diffusers import StableDiffusionXLPipeline
-
-# Load the pipeline in full-precision and place its model components on CUDA.
-pipe = StableDiffusionXLPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0"
-).to("cuda")
-
-# Run the attention ops without SDPA.
-pipe.unet.set_default_attn_processor()
-pipe.vae.set_default_attn_processor()
-
-prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
-image = pipe(prompt, num_inference_steps=30).images[0]
-```
-
-This default setup takes 7.36 seconds.
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/progressive-acceleration-sdxl/SDXL%2C_Batch_Size%3A_1%2C_Steps%3A_30_0.png" width=500>
-</div>
-
-## bfloat16
-
-Enable the first optimization, reduced precision or more specifically bfloat16. There are several benefits of using reduced precision:
-
-* Using a reduced numerical precision (such as float16 or bfloat16) for inference doesn’t affect the generation quality but significantly improves latency.
-* The benefits of using bfloat16 compared to float16 are hardware dependent, but modern GPUs tend to favor bfloat16.
-* bfloat16 is much more resilient when used with quantization compared to float16, but more recent versions of the quantization library ([torchao](https://github.com/pytorch-labs/ao)) we used don't have numerical issues with float16.
-
-```python
-from diffusers import StableDiffusionXLPipeline
-import torch
-
-pipe = StableDiffusionXLPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
-).to("cuda")
-
-# Run the attention ops without SDPA.
-pipe.unet.set_default_attn_processor()
-pipe.vae.set_default_attn_processor()
-
-prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
-image = pipe(prompt, num_inference_steps=30).images[0]
-```
-
-bfloat16 reduces the latency from 7.36 seconds to 4.63 seconds.
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/progressive-acceleration-sdxl/SDXL%2C_Batch_Size%3A_1%2C_Steps%3A_30_1.png" width=500>
-</div>
-
-<Tip>
-
-In our later experiments with float16, recent versions of torchao do not incur numerical problems from float16.
-
-</Tip>
-
-Take a look at the [Speed up inference](../optimization/fp16) guide to learn more about running inference with reduced precision.
-
-## SDPA
-
-Attention blocks are intensive to run. But with PyTorch's [`scaled_dot_product_attention`](../optimization/torch2.0#scaled-dot-product-attention) function, it is a lot more efficient. This function is used by default in Diffusers so you don't need to make any changes to the code.
-
-```python
-from diffusers import StableDiffusionXLPipeline
-import torch
-
-pipe = StableDiffusionXLPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
-).to("cuda")
-
-prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
-image = pipe(prompt, num_inference_steps=30).images[0]
-```
-
-Scaled dot product attention improves the latency from 4.63 seconds to 3.31 seconds.
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/progressive-acceleration-sdxl/SDXL%2C_Batch_Size%3A_1%2C_Steps%3A_30_2.png" width=500>
-</div>
-
-## torch.compile
-
-PyTorch 2 includes `torch.compile` which uses fast and optimized kernels. In Diffusers, the UNet and VAE are usually compiled because these are the most compute-intensive modules. First, configure a few compiler flags (refer to the [full list](https://github.com/pytorch/pytorch/blob/main/torch/_inductor/config.py) for more options):
-
-```python
-from diffusers import StableDiffusionXLPipeline
-import torch
-
-torch._inductor.config.conv_1x1_as_mm = True
-torch._inductor.config.coordinate_descent_tuning = True
-torch._inductor.config.epilogue_fusion = False
-torch._inductor.config.coordinate_descent_check_all_directions = True
-```
-
-It is also important to change the UNet and VAE's memory layout to "channels_last" when compiling them to ensure maximum speed.
-
-```python
-pipe.unet.to(memory_format=torch.channels_last)
-pipe.vae.to(memory_format=torch.channels_last)
-```
-
-Now compile and perform inference:
-
-```python
-# Compile the UNet and VAE.
-pipe.unet = torch.compile(pipe.unet, mode="max-autotune", fullgraph=True)
-pipe.vae.decode = torch.compile(pipe.vae.decode, mode="max-autotune", fullgraph=True)
-
-prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
-
-# First call to `pipe` is slow, subsequent ones are faster.
-image = pipe(prompt, num_inference_steps=30).images[0]
-```
-
-`torch.compile` offers different backends and modes. For maximum inference speed, use "max-autotune" for the inductor backend. “max-autotune” uses CUDA graphs and optimizes the compilation graph specifically for latency. CUDA graphs greatly reduces the overhead of launching GPU operations by using a mechanism to launch multiple GPU operations through a single CPU operation.
-
-Using SDPA attention and compiling both the UNet and VAE cuts the latency from 3.31 seconds to 2.54 seconds.
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/progressive-acceleration-sdxl/SDXL%2C_Batch_Size%3A_1%2C_Steps%3A_30_3.png" width=500>
-</div>
-
-> [!TIP]
-> From PyTorch 2.3.1, you can control the caching behavior of `torch.compile()`. This is particularly beneficial for compilation modes like `"max-autotune"` which performs a grid-search over several compilation flags to find the optimal configuration. Learn more in the [Compile Time Caching in torch.compile](https://pytorch.org/tutorials/recipes/torch_compile_caching_tutorial.html) tutorial.
-
-### Prevent graph breaks
-
-Specifying `fullgraph=True` ensures there are no graph breaks in the underlying model to take full advantage of `torch.compile` without any performance degradation. For the UNet and VAE, this means changing how you access the return variables.
-
-```diff
-- latents = unet(
--   latents, timestep=timestep, encoder_hidden_states=prompt_embeds
--).sample
-
-+ latents = unet(
-+   latents, timestep=timestep, encoder_hidden_states=prompt_embeds, return_dict=False
-+)[0]
-```
-
-### Remove GPU sync after compilation
-
-During the iterative reverse diffusion process, the `step()` function is [called](https://github.com/huggingface/diffusers/blob/1d686bac8146037e97f3fd8c56e4063230f71751/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py#L1228) on the scheduler each time after the denoiser predicts the less noisy latent embeddings. Inside `step()`, the `sigmas` variable is [indexed](https://github.com/huggingface/diffusers/blob/1d686bac8146037e97f3fd8c56e4063230f71751/src/diffusers/schedulers/scheduling_euler_discrete.py#L476) which when placed on the GPU, causes a communication sync between the CPU and GPU. This introduces latency and it becomes more evident when the denoiser has already been compiled.
-
-But if the `sigmas` array always [stays on the CPU](https://github.com/huggingface/diffusers/blob/35a969d297cba69110d175ee79c59312b9f49e1e/src/diffusers/schedulers/scheduling_euler_discrete.py#L240), the CPU and GPU sync doesn’t occur and you don't get any latency. In general, any CPU and GPU communication sync should be none or be kept to a bare minimum because it can impact inference latency.
-
-## Combine the attention block's projection matrices
-
-The UNet and VAE in SDXL use Transformer-like blocks which consists of attention blocks and feed-forward blocks.
-
-In an attention block, the input is projected into three sub-spaces using three different projection matrices – Q, K, and V. These projections are performed separately on the input. But we can horizontally combine the projection matrices into a single matrix and perform the projection in one step. This increases the size of the matrix multiplications of the input projections and improves the impact of quantization.
-
-You can combine the projection matrices with just a single line of code:
-
-```python
-pipe.fuse_qkv_projections()
-```
-
-This provides a minor improvement from 2.54 seconds to 2.52 seconds.
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/progressive-acceleration-sdxl/SDXL%2C_Batch_Size%3A_1%2C_Steps%3A_30_4.png" width=500>
-</div>
-
-<Tip warning={true}>
-
-Support for [`~StableDiffusionXLPipeline.fuse_qkv_projections`] is limited and experimental. It's not available for many non-Stable Diffusion pipelines such as [Kandinsky](../using-diffusers/kandinsky). You can refer to this [PR](https://github.com/huggingface/diffusers/pull/6179) to get an idea about how to enable this for the other pipelines.
-
-</Tip>
-
-## Dynamic quantization
-
-You can also use the ultra-lightweight PyTorch quantization library, [torchao](https://github.com/pytorch-labs/ao) (commit SHA `54bcd5a10d0abbe7b0c045052029257099f83fd9`), to apply [dynamic int8 quantization](https://pytorch.org/tutorials/recipes/recipes/dynamic_quantization.html) to the UNet and VAE. Quantization adds additional conversion overhead to the model that is hopefully made up for by faster matmuls (dynamic quantization). If the matmuls are too small, these techniques may degrade performance.
-
-First, configure all the compiler tags:
-
-```python
-from diffusers import StableDiffusionXLPipeline
-import torch
-
-# Notice the two new flags at the end.
-torch._inductor.config.conv_1x1_as_mm = True
-torch._inductor.config.coordinate_descent_tuning = True
-torch._inductor.config.epilogue_fusion = False
-torch._inductor.config.coordinate_descent_check_all_directions = True
-torch._inductor.config.force_fuse_int_mm_with_mul = True
-torch._inductor.config.use_mixed_mm = True
-```
-
-Certain linear layers in the UNet and VAE don’t benefit from dynamic int8 quantization. You can filter out those layers with the [`dynamic_quant_filter_fn`](https://github.com/huggingface/diffusion-fast/blob/0f169640b1db106fe6a479f78c1ed3bfaeba3386/utils/pipeline_utils.py#L16) shown below.
-
-```python
-def dynamic_quant_filter_fn(mod, *args):
-    return (
-        isinstance(mod, torch.nn.Linear)
-        and mod.in_features > 16
-        and (mod.in_features, mod.out_features)
-        not in [
-            (1280, 640),
-            (1920, 1280),
-            (1920, 640),
-            (2048, 1280),
-            (2048, 2560),
-            (2560, 1280),
-            (256, 128),
-            (2816, 1280),
-            (320, 640),
-            (512, 1536),
-            (512, 256),
-            (512, 512),
-            (640, 1280),
-            (640, 1920),
-            (640, 320),
-            (640, 5120),
-            (640, 640),
-            (960, 320),
-            (960, 640),
-        ]
-    )
-
-
-def conv_filter_fn(mod, *args):
-    return (
-        isinstance(mod, torch.nn.Conv2d) and mod.kernel_size == (1, 1) and 128 in [mod.in_channels, mod.out_channels]
-    )
-```
-
-Finally, apply all the optimizations discussed so far:
-
-```python
-# SDPA + bfloat16.
-pipe = StableDiffusionXLPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.bfloat16
-).to("cuda")
-
-# Combine attention projection matrices.
-pipe.fuse_qkv_projections()
-
-# Change the memory layout.
-pipe.unet.to(memory_format=torch.channels_last)
-pipe.vae.to(memory_format=torch.channels_last)
-```
-
-Since dynamic quantization is only limited to the linear layers, convert the appropriate pointwise convolution layers into linear layers to maximize its benefit.
-
-```python
-from torchao import swap_conv2d_1x1_to_linear
-
-swap_conv2d_1x1_to_linear(pipe.unet, conv_filter_fn)
-swap_conv2d_1x1_to_linear(pipe.vae, conv_filter_fn)
-```
-
-Apply dynamic quantization:
-
-```python
-from torchao import apply_dynamic_quant
-
-apply_dynamic_quant(pipe.unet, dynamic_quant_filter_fn)
-apply_dynamic_quant(pipe.vae, dynamic_quant_filter_fn)
-```
-
-Finally, compile and perform inference:
-
-```python
-pipe.unet = torch.compile(pipe.unet, mode="max-autotune", fullgraph=True)
-pipe.vae.decode = torch.compile(pipe.vae.decode, mode="max-autotune", fullgraph=True)
-
-prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
-image = pipe(prompt, num_inference_steps=30).images[0]
-```
-
-Applying dynamic quantization improves the latency from 2.52 seconds to 2.43 seconds.
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/sayakpaul/sample-datasets/resolve/main/progressive-acceleration-sdxl/SDXL%2C_Batch_Size%3A_1%2C_Steps%3A_30_5.png" width=500>
-</div>

docs/source/en/tutorials/inference_with_big_models.md

@@ -1,139 +0,0 @@
-<!--Copyright 2024 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.
--->
-
-# Working with big models
-
-A modern diffusion model, like [Stable Diffusion XL (SDXL)](../using-diffusers/sdxl), is not just a single model, but a collection of multiple models. SDXL has four different model-level components:
-
-* A variational autoencoder (VAE)
-* Two text encoders
-* A UNet for denoising
-
-Usually, the text encoders and the denoiser are much larger compared to the VAE.
-
-As models get bigger and better, it’s possible your model is so big that even a single copy won’t fit in memory. But that doesn’t mean it can’t be loaded. If you have more than one GPU, there is more memory available to store your model. In this case, it’s better to split your model checkpoint into several smaller *checkpoint shards*.
-
-When a text encoder checkpoint has multiple shards, like [T5-xxl for SD3](https://huggingface.co/stabilityai/stable-diffusion-3-medium-diffusers/tree/main/text_encoder_3), it is automatically handled by the [Transformers](https://huggingface.co/docs/transformers/index) library as it is a required dependency of Diffusers when using the [`StableDiffusion3Pipeline`]. More specifically, Transformers will automatically handle the loading of multiple shards within the requested model class and get it ready so that inference can be performed.
-
-The denoiser checkpoint can also have multiple shards and supports inference thanks to the [Accelerate](https://huggingface.co/docs/accelerate/index) library.
-
-> [!TIP]
-> Refer to the [Handling big models for inference](https://huggingface.co/docs/accelerate/main/en/concept_guides/big_model_inference) guide for general guidance when working with big models that are hard to fit into memory.
-
-For example, let's save a sharded checkpoint for the [SDXL UNet](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/tree/main/unet):
-
-```python
-from diffusers import AutoModel
-
-unet = AutoModel.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", subfolder="unet"
-)
-unet.save_pretrained("sdxl-unet-sharded", max_shard_size="5GB")
-```
-
-The size of the fp32 variant of the SDXL UNet checkpoint is ~10.4GB. Set the `max_shard_size` parameter to 5GB to create 3 shards. After saving, you can load them in [`StableDiffusionXLPipeline`]:
-
-```python
-from diffusers import AutoModel, StableDiffusionXLPipeline
-import torch
-
-unet = AutoModel.from_pretrained(
-    "sayakpaul/sdxl-unet-sharded", torch_dtype=torch.float16
-)
-pipeline = StableDiffusionXLPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", unet=unet, torch_dtype=torch.float16
-).to("cuda")
-
-image = pipeline("a cute dog running on the grass", num_inference_steps=30).images[0]
-image.save("dog.png")
-```
-
-If placing all the model-level components on the GPU at once is not feasible, use [`~DiffusionPipeline.enable_model_cpu_offload`] to help you:
-
-```diff
-- pipeline.to("cuda")
-+ pipeline.enable_model_cpu_offload()
-```
-
-In general, we recommend sharding when a checkpoint is more than 5GB (in fp32).
-
-## Device placement
-
-On distributed setups, you can run inference across multiple GPUs with Accelerate.
-
-> [!WARNING]
-> This feature is experimental and its APIs might change in the future.
-
-With Accelerate, you can use the `device_map` to determine how to distribute the models of a pipeline across multiple devices. This is useful in situations where you have more than one GPU.
-
-For example, if you have two 8GB GPUs, then using [`~DiffusionPipeline.enable_model_cpu_offload`] may not work so well because:
-
-* it only works on a single GPU
-* a single model might not fit on a single GPU ([`~DiffusionPipeline.enable_sequential_cpu_offload`] might work but it will be extremely slow and it is also limited to a single GPU)
-
-To make use of both GPUs, you can use the "balanced" device placement strategy which splits the models across all available GPUs.
-
-> [!WARNING]
-> Only the "balanced" strategy is supported at the moment, and we plan to support additional mapping strategies in the future.
-
-```diff
-from diffusers import DiffusionPipeline
-import torch
-
-pipeline = DiffusionPipeline.from_pretrained(
--    "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True,
-+    "stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16, use_safetensors=True, device_map="balanced"
-)
-image = pipeline("a dog").images[0]
-image
-```
-
-You can also pass a dictionary to enforce the maximum GPU memory that can be used on each device:
-
-```diff
-from diffusers import DiffusionPipeline
-import torch
-
-max_memory = {0:"1GB", 1:"1GB"}
-pipeline = DiffusionPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
-    device_map="balanced",
-+   max_memory=max_memory
-)
-image = pipeline("a dog").images[0]
-image
-```
-
-If a device is not present in `max_memory`, then it will be completely ignored and will not participate in the device placement.
-
-By default, Diffusers uses the maximum memory of all devices. If the models don't fit on the GPUs, they are offloaded to the CPU. If the CPU doesn't have enough memory, then you might see an error. In that case, you could defer to using [`~DiffusionPipeline.enable_sequential_cpu_offload`] and [`~DiffusionPipeline.enable_model_cpu_offload`].
-
-Call [`~DiffusionPipeline.reset_device_map`] to reset the `device_map` of a pipeline. This is also necessary if you want to use methods like `to()`, [`~DiffusionPipeline.enable_sequential_cpu_offload`], and [`~DiffusionPipeline.enable_model_cpu_offload`] on a pipeline that was device-mapped.
-
-```py
-pipeline.reset_device_map()
-```
-
-Once a pipeline has been device-mapped, you can also access its device map via `hf_device_map`:
-
-```py
-print(pipeline.hf_device_map)
-```
-
-An example device map would look like so:
-
-
-```bash
-{'unet': 1, 'vae': 1, 'safety_checker': 0, 'text_encoder': 0}
-```

docs/source/en/tutorials/using_peft_for_inference.md

@@ -10,218 +10,625 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-[[open-in-colab]]
+# LoRA
 
-# Load LoRAs for inference
+[LoRA (Low-Rank Adaptation)](https://huggingface.co/papers/2106.09685) is a method for quickly training a model for a new task. It works by freezing the original model weights and adding a small number of *new* trainable parameters. This means it is significantly faster and cheaper to adapt an existing model to new tasks, such as generating images in a new style.
 
-There are many adapter types (with [LoRAs](https://huggingface.co/docs/peft/conceptual_guides/adapter#low-rank-adaptation-lora) being the most popular) trained in different styles to achieve different effects. You can even combine multiple adapters to create new and unique images.
+LoRA checkpoints are typically only a couple hundred MBs in size, so they're very lightweight and easy to store. Load these smaller set of weights into an existing base model with [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] and specify the file name.
 
-In this tutorial, you'll learn how to easily load and manage adapters for inference with the 🤗 [PEFT](https://huggingface.co/docs/peft/index) integration in 🤗 Diffusers. You'll use LoRA as the main adapter technique, so you'll see the terms LoRA and adapter used interchangeably.
+<hfoptions id="usage">
+<hfoption id="text-to-image">
 
-Let's first install all the required libraries.
-
-```bash
-!pip install -q transformers accelerate peft diffusers
-```
-
-Now, load a pipeline with a [Stable Diffusion XL (SDXL)](../api/pipelines/stable_diffusion/stable_diffusion_xl) checkpoint:
-
-```python
-from diffusers import DiffusionPipeline
+```py
 import torch
+from diffusers import AutoPipelineForText2Image
 
-pipe_id = "stabilityai/stable-diffusion-xl-base-1.0"
-pipe = DiffusionPipeline.from_pretrained(pipe_id, torch_dtype=torch.float16).to("cuda")
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/super-cereal-sdxl-lora",
+    weight_name="cereal_box_sdxl_v1.safetensors",
+    adapter_name="cereal"
+)
+pipeline("bears, pizza bites").images[0]
 ```
 
-Next, load a [CiroN2022/toy-face](https://huggingface.co/CiroN2022/toy-face) adapter with the [`~diffusers.loaders.StableDiffusionXLLoraLoaderMixin.load_lora_weights`] method. With the 🤗 PEFT integration, you can assign a specific `adapter_name` to the checkpoint, which lets you easily switch between different LoRA checkpoints. Let's call this adapter `"toy"`.
+</hfoption>
+<hfoption id="text-to-video">
 
-```python
-pipe.load_lora_weights("CiroN2022/toy-face", weight_name="toy_face_sdxl.safetensors", adapter_name="toy")
+```py
+import torch
+from diffusers import LTXConditionPipeline
+from diffusers.utils import export_to_video, load_image
+
+pipeline = LTXConditionPipeline.from_pretrained(
+    "Lightricks/LTX-Video-0.9.5", torch_dtype=torch.bfloat16
+)
+
+pipeline.load_lora_weights(
+    "Lightricks/LTX-Video-Cakeify-LoRA",
+    weight_name="ltxv_095_cakeify_lora.safetensors",
+    adapter_name="cakeify"
+)
+pipeline.set_adapters("cakeify")
+
+# use "CAKEIFY" to trigger the LoRA
+prompt = "CAKEIFY a person using a knife to cut a cake shaped like a Pikachu plushie"
+image = load_image("https://huggingface.co/Lightricks/LTX-Video-Cakeify-LoRA/resolve/main/assets/images/pikachu.png")
+
+video = pipeline(
+    prompt=prompt,
+    image=image,
+    width=576,
+    height=576,
+    num_frames=161,
+    decode_timestep=0.03,
+    decode_noise_scale=0.025,
+    num_inference_steps=50,
+).frames[0]
+export_to_video(video, "output.mp4", fps=26)
 ```
 
-Make sure to include the token `toy_face` in the prompt and then you can perform inference:
+</hfoption>
+</hfoptions>
 
-```python
-prompt = "toy_face of a hacker with a hoodie"
+The [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] method is the preferred way to load LoRA weights into the UNet and text encoder because it can handle cases where:
 
-lora_scale = 0.9
-image = pipe(
-    prompt, num_inference_steps=30, cross_attention_kwargs={"scale": lora_scale}, generator=torch.manual_seed(0)
-).images[0]
-image
+- the LoRA weights don't have separate UNet and text encoder identifiers
+- the LoRA weights have separate UNet and text encoder identifiers
+
+The [`~loaders.PeftAdapterMixin.load_lora_adapter`] method is used to directly load a LoRA adapter at the *model-level*, as long as the model is a Diffusers model that is a subclass of [`PeftAdapterMixin`]. It builds and prepares the necessary model configuration for the adapter. This method also loads the LoRA adapter into the UNet.
+
+For example, if you're only loading a LoRA into the UNet, [`~loaders.PeftAdapterMixin.load_lora_adapter`] ignores the text encoder keys. Use the `prefix` parameter to filter and load the appropriate state dicts, `"unet"` to load.
+
+```py
+import torch
+from diffusers import AutoPipelineForText2Image
+
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.unet.load_lora_adapter(
+    "jbilcke-hf/sdxl-cinematic-1",
+    weight_name="pytorch_lora_weights.safetensors",
+    adapter_name="cinematic"
+    prefix="unet"
+)
+# use cnmt in the prompt to trigger the LoRA
+pipeline("A cute cnmt eating a slice of pizza, stunning color scheme, masterpiece, illustration").images[0]
 ```
 
-![toy-face](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_8_1.png)
+## torch.compile
 
-With the `adapter_name` parameter, it is really easy to use another adapter for inference! Load the [nerijs/pixel-art-xl](https://huggingface.co/nerijs/pixel-art-xl) adapter that has been fine-tuned to generate pixel art images and call it `"pixel"`.
+[torch.compile](../optimization/fp16#torchcompile) speeds up inference by compiling the PyTorch model to use optimized kernels. Before compiling, the LoRA weights need to be fused into the base model and unloaded first.
 
-The pipeline automatically sets the first loaded adapter (`"toy"`) as the active adapter, but you can activate the `"pixel"` adapter with the [`~loaders.peft.PeftAdapterMixin.set_adapters`] method:
+```py
+import torch
+from diffusers import DiffusionPipeline
 
-```python
-pipe.load_lora_weights("nerijs/pixel-art-xl", weight_name="pixel-art-xl.safetensors", adapter_name="pixel")
-pipe.set_adapters("pixel")
+# load base model and LoRA
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
+
+# activate LoRA and set adapter weight
+pipeline.set_adapters("ikea", adapter_weights=0.7)
+
+# fuse LoRAs and unload weights
+pipeline.fuse_lora(adapter_names=["ikea"], lora_scale=1.0)
+pipeline.unload_lora_weights()
 ```
 
-Make sure you include the token `pixel art` in your prompt to generate a pixel art image:
+Typically, the UNet is compiled because its the most compute intensive component of the pipeline.
 
-```python
-prompt = "a hacker with a hoodie, pixel art"
-image = pipe(
-    prompt, num_inference_steps=30, cross_attention_kwargs={"scale": lora_scale}, generator=torch.manual_seed(0)
-).images[0]
-image
+```py
+pipeline.unet.to(memory_format=torch.channels_last)
+pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
+
+pipeline("A bowl of ramen shaped like a cute kawaii bear").images[0]
 ```
 
-![pixel-art](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_12_1.png)
+Refer to the [hotswapping](#hotswapping) section to learn how to avoid recompilation when working with compiled models and multiple LoRAs.
 
-<Tip>
+## Weight scale
 
-By default, if the most up-to-date versions of PEFT and Transformers are detected, `low_cpu_mem_usage` is set to `True` to speed up the loading time of LoRA checkpoints. 
+The `scale` parameter is used to control how much of a LoRA to apply. A value of `0` is equivalent to only using the base model weights and a value of `1` is equivalent to fully using the LoRA.
 
-</Tip>
+<hfoptions id="weight-scale">
+<hfoption id="simple use case">
 
-## Merge adapters
+For simple use cases, you can pass `cross_attention_kwargs={"scale": 1.0}` to the pipeline.
 
-You can also merge different adapter checkpoints for inference to blend their styles together.
+```py
+import torch
+from diffusers import AutoPipelineForText2Image
 
-Once again, use the [`~loaders.peft.PeftAdapterMixin.set_adapters`] method to activate the `pixel` and `toy` adapters and specify the weights for how they should be merged.
-
-```python
-pipe.set_adapters(["pixel", "toy"], adapter_weights=[0.5, 1.0])
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/super-cereal-sdxl-lora",
+    weight_name="cereal_box_sdxl_v1.safetensors",
+    adapter_name="cereal"
+)
+pipeline("bears, pizza bites", cross_attention_kwargs={"scale": 1.0}).images[0]
 ```
 
-<Tip>
+</hfoption>
+<hfoption id="finer control">
 
-LoRA checkpoints in the diffusion community are almost always obtained with [DreamBooth](https://huggingface.co/docs/diffusers/main/en/training/dreambooth). DreamBooth training often relies on "trigger" words in the input text prompts in order for the generation results to look as expected. When you combine multiple LoRA checkpoints, it's important to ensure the trigger words for the corresponding LoRA checkpoints are present in the input text prompts.
+> [!WARNING]
+> The [`~loaders.PeftAdapterMixin.set_adapters`] method only scales attention weights. If a LoRA has ResNets or down and upsamplers, these components keep a scale value of `1.0`.
 
-</Tip>
+For finer control over each individual component of the UNet or text encoder, pass a dictionary instead. In the example below, the `"down"` block in the UNet is scaled by 0.9 and you can further specify in the `"up"` block the scales of the transformers in `"block_0"` and `"block_1"`. If a block like `"mid"` isn't specified, the default value `1.0` is used.
 
-Remember to use the trigger words for [CiroN2022/toy-face](https://hf.co/CiroN2022/toy-face) and [nerijs/pixel-art-xl](https://hf.co/nerijs/pixel-art-xl) (these are found in their repositories) in the prompt to generate an image.
+```py
+import torch
+from diffusers import AutoPipelineForText2Image
 
-```python
-prompt = "toy_face of a hacker with a hoodie, pixel art"
-image = pipe(
-    prompt, num_inference_steps=30, cross_attention_kwargs={"scale": 1.0}, generator=torch.manual_seed(0)
-).images[0]
-image
-```
-
-![toy-face-pixel-art](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_16_1.png)
-
-Impressive! As you can see, the model generated an image that mixed the characteristics of both adapters.
-
-> [!TIP]
-> Through its PEFT integration, Diffusers also offers more efficient merging methods which you can learn about in the [Merge LoRAs](../using-diffusers/merge_loras) guide!
-
-To return to only using one adapter, use the [`~loaders.peft.PeftAdapterMixin.set_adapters`] method to activate the `"toy"` adapter:
-
-```python
-pipe.set_adapters("toy")
-
-prompt = "toy_face of a hacker with a hoodie"
-lora_scale = 0.9
-image = pipe(
-    prompt, num_inference_steps=30, cross_attention_kwargs={"scale": lora_scale}, generator=torch.manual_seed(0)
-).images[0]
-image
-```
-
-Or to disable all adapters entirely, use the [`~loaders.peft.PeftAdapterMixin.disable_lora`] method to return the base model.
-
-```python
-pipe.disable_lora()
-
-prompt = "toy_face of a hacker with a hoodie"
-image = pipe(prompt, num_inference_steps=30, generator=torch.manual_seed(0)).images[0]
-image
-```
-
-![no-lora](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_20_1.png)
-
-### Customize adapters strength
-
-For even more customization, you can control how strongly the adapter affects each part of the pipeline. For this, pass a dictionary with the control strengths (called "scales") to [`~loaders.peft.PeftAdapterMixin.set_adapters`].
-
-For example, here's how you can turn on the adapter for the `down` parts, but turn it off for the `mid` and `up` parts:
-```python
-pipe.enable_lora()  # enable lora again, after we disabled it above
-prompt = "toy_face of a hacker with a hoodie, pixel art"
-adapter_weight_scales = { "unet": { "down": 1, "mid": 0, "up": 0} }
-pipe.set_adapters("pixel", adapter_weight_scales)
-image = pipe(prompt, num_inference_steps=30, generator=torch.manual_seed(0)).images[0]
-image
-```
-
-![block-lora-text-and-down](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_block_down.png)
-
-Let's see how turning off the `down` part and turning on the `mid` and `up` part respectively changes the image.
-```python
-adapter_weight_scales = { "unet": { "down": 0, "mid": 1, "up": 0} }
-pipe.set_adapters("pixel", adapter_weight_scales)
-image = pipe(prompt, num_inference_steps=30, generator=torch.manual_seed(0)).images[0]
-image
-```
-
-![block-lora-text-and-mid](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_block_mid.png)
-
-```python
-adapter_weight_scales = { "unet": { "down": 0, "mid": 0, "up": 1} }
-pipe.set_adapters("pixel", adapter_weight_scales)
-image = pipe(prompt, num_inference_steps=30, generator=torch.manual_seed(0)).images[0]
-image
-```
-
-![block-lora-text-and-up](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_block_up.png)
-
-Looks cool!
-
-This is a really powerful feature. You can use it to control the adapter strengths down to per-transformer level. And you can even use it for multiple adapters.
-```python
-adapter_weight_scales_toy = 0.5
-adapter_weight_scales_pixel = {
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/super-cereal-sdxl-lora",
+    weight_name="cereal_box_sdxl_v1.safetensors",
+    adapter_name="cereal"
+)
+scales = {
+    "text_encoder": 0.5,
+    "text_encoder_2": 0.5,
     "unet": {
-        "down": 0.9,  # all transformers in the down-part will use scale 0.9
-        # "mid"  # because, in this example, "mid" is not given, all transformers in the mid part will use the default scale 1.0
+        "down": 0.9,
         "up": {
-            "block_0": 0.6,  # all 3 transformers in the 0th block in the up-part will use scale 0.6
-            "block_1": [0.4, 0.8, 1.0],  # the 3 transformers in the 1st block in the up-part will use scales 0.4, 0.8 and 1.0 respectively
+            "block_0": 0.6,
+            "block_1": [0.4, 0.8, 1.0],
         }
     }
 }
-pipe.set_adapters(["toy", "pixel"], [adapter_weight_scales_toy, adapter_weight_scales_pixel])
-image = pipe(prompt, num_inference_steps=30, generator=torch.manual_seed(0)).images[0]
-image
+pipeline.set_adapters("cereal", scales)
+pipeline("bears, pizza bites").images[0]
 ```
 
-![block-lora-mixed](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/peft_integration/diffusers_peft_lora_inference_block_mixed.png)
+</hfoption>
+</hfoptions>
 
-## Manage adapters
+## Hotswapping
 
-You have attached multiple adapters in this tutorial, and if you're feeling a bit lost on what adapters have been attached to the pipeline's components, use the [`~diffusers.loaders.StableDiffusionLoraLoaderMixin.get_active_adapters`] method to check the list of active adapters:
+Hotswapping LoRAs is an efficient way to work with multiple LoRAs while avoiding accumulating memory from multiple calls to [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] and in some cases, recompilation, if a model is compiled. This workflow requires a loaded LoRA because the new LoRA weights are swapped in place for the existing loaded LoRA.
 
 ```py
-active_adapters = pipe.get_active_adapters()
-active_adapters
-["toy", "pixel"]
+import torch
+from diffusers import DiffusionPipeline
+
+# load base model and LoRAs
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
 ```
 
-You can also get the active adapters of each pipeline component with [`~diffusers.loaders.StableDiffusionLoraLoaderMixin.get_list_adapters`]:
+> [!WARNING]
+> Hotswapping is unsupported for LoRAs that target the text encoder.
+
+Set `hotswap=True` in [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] to swap the second LoRA. Use the `adapter_name` parameter to indicate which LoRA to swap (`default_0` is the default name).
 
 ```py
-list_adapters_component_wise = pipe.get_list_adapters()
-list_adapters_component_wise
-{"text_encoder": ["toy", "pixel"], "unet": ["toy", "pixel"], "text_encoder_2": ["toy", "pixel"]}
+pipeline.load_lora_weights(
+    "lordjia/by-feng-zikai",
+    hotswap=True,
+    adapter_name="ikea"
+)
 ```
 
-The [`~loaders.peft.PeftAdapterMixin.delete_adapters`] function completely removes an adapter and their LoRA layers from a model.
+### Compiled models
+
+For compiled models, use [`~loaders.lora_base.LoraBaseMixin.enable_lora_hotswap`] to avoid recompilation when hotswapping LoRAs. This method should be called *before* loading the first LoRA and `torch.compile` should be called *after* loading the first LoRA.
+
+> [!TIP]
+> The [`~loaders.lora_base.LoraBaseMixin.enable_lora_hotswap`] method isn't always necessary if the second LoRA targets the identical LoRA ranks and scales as the first LoRA.
+
+Within [`~loaders.lora_base.LoraBaseMixin.enable_lora_hotswap`], the `target_rank` parameter is important for setting the rank for all LoRA adapters. Setting it to `max_rank` sets it to the highest value. For LoRAs with different ranks, you set it to a higher rank value. The default rank value is 128.
 
 ```py
-pipe.delete_adapters("toy")
-pipe.get_active_adapters()
-["pixel"]
+import torch
+from diffusers import DiffusionPipeline
+
+# load base model and LoRAs
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+# 1. enable_lora_hotswap
+pipeline.enable_lora_hotswap(target_rank=max_rank)
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
+# 2. torch.compile
+pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
+
+# 3. hotswap
+pipeline.load_lora_weights(
+    "lordjia/by-feng-zikai",
+    hotswap=True,
+    adapter_name="ikea"
+)
 ```
 
-## PeftInputAutocastDisableHook
+> [!TIP]
+> Move your code inside the `with torch._dynamo.config.patch(error_on_recompile=True)` context manager to detect if a model was recompiled. If a model is recompiled despite following all the steps above, please open an [issue](https://github.com/huggingface/diffusers/issues) with a reproducible example.
 
-[[autodoc]] hooks.layerwise_casting.PeftInputAutocastDisableHook
+There are still scenarios where recompulation is unavoidable, such as when the hotswapped LoRA targets more layers than the initial adapter. Try to load the LoRA that targets the most layers *first*. For more details about this limitation, refer to the PEFT [hotswapping](https://huggingface.co/docs/peft/main/en/package_reference/hotswap#peft.utils.hotswap.hotswap_adapter) docs.
+
+## Merge
+
+The weights from each LoRA can be merged together to produce a blend of multiple existing styles. There are several methods for merging LoRAs, each of which differ in *how* the weights are merged (may affect generation quality).
+
+### set_adapters
+
+The [`~loaders.PeftAdapterMixin.set_adapters`] method merges LoRAs by concatenating their weighted matrices. Pass the LoRA names to [`~loaders.PeftAdapterMixin.set_adapters`] and use the `adapter_weights` parameter to control the scaling of each LoRA. For example, if `adapter_weights=[0.5, 0.5]`, the output is an average of both LoRAs.
+
+> [!TIP]
+> The `"scale"` parameter determines how much of the merged LoRA to apply. See the [Weight scale](#weight-scale) section for more details.
+
+```py
+import torch
+from diffusers import DiffusionPipeline
+
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
+pipeline.load_lora_weights(
+    "lordjia/by-feng-zikai",
+    weight_name="fengzikai_v1.0_XL.safetensors",
+    adapter_name="feng"
+)
+pipeline.set_adapters(["ikea", "feng"], adapter_weights=[0.7, 0.8])
+# use by Feng Zikai to activate the lordjia/by-feng-zikai LoRA
+pipeline("A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai", cross_attention_kwargs={"scale": 1.0}).images[0]
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lora_merge_set_adapters.png"/>
+</div>
+
+### add_weighted_adapter
+
+> [!TIP]
+> This is an experimental method and you can refer to PEFTs [Model merging](https://huggingface.co/docs/peft/developer_guides/model_merging) for more details. Take a look at this [issue](https://github.com/huggingface/diffusers/issues/6892) if you're interested in the motivation and design behind this integration.
+
+The [`~peft.LoraModel.add_weighted_adapter`] method enables more efficient merging methods like [TIES](https://huggingface.co/papers/2306.01708) or [DARE](https://huggingface.co/papers/2311.03099). These merging methods remove redundant and potentially interfering parameters from merged models. Keep in mind the LoRA ranks need to have identical ranks to be merged.
+
+Make sure the latest stable version of Diffusers and PEFT is installed.
+
+```bash
+pip install -U -q diffusers peft
+```
+
+Load a UNET that corresponds to the LoRA UNet.
+
+```py
+import copy
+import torch
+from diffusers import AutoModel, DiffusionPipeline
+from peft import get_peft_model, LoraConfig, PeftModel
+
+unet = AutoModel.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+    use_safetensors=True,
+    variant="fp16",
+    subfolder="unet",
+).to("cuda")
+```
+
+Load a pipeline, pass the UNet to it, and load a LoRA.
+
+```py
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    variant="fp16",
+    torch_dtype=torch.float16,
+    unet=unet
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
+```
+
+Create a [`~peft.PeftModel`] from the LoRA checkpoint by combining the first UNet you loaded and the LoRA UNet from the pipeline.
+
+```py
+sdxl_unet = copy.deepcopy(unet)
+ikea_peft_model = get_peft_model(
+    sdxl_unet,
+    pipeline.unet.peft_config["ikea"],
+    adapter_name="ikea"
+)
+
+original_state_dict = {f"base_model.model.{k}": v for k, v in pipeline.unet.state_dict().items()}
+ikea_peft_model.load_state_dict(original_state_dict, strict=True)
+```
+
+> [!TIP]
+> You can save and reuse the `ikea_peft_model` by pushing it to the Hub as shown below.
+> ```py
+> ikea_peft_model.push_to_hub("ikea_peft_model", token=TOKEN)
+> ```
+
+Repeat this process and create a [`~peft.PeftModel`] for the second LoRA.
+
+```py
+pipeline.delete_adapters("ikea")
+sdxl_unet.delete_adapters("ikea")
+
+pipeline.load_lora_weights(
+    "lordjia/by-feng-zikai",
+    weight_name="fengzikai_v1.0_XL.safetensors",
+    adapter_name="feng"
+)
+pipeline.set_adapters(adapter_names="feng")
+
+feng_peft_model = get_peft_model(
+    sdxl_unet,
+    pipeline.unet.peft_config["feng"],
+    adapter_name="feng"
+)
+
+original_state_dict = {f"base_model.model.{k}": v for k, v in pipe.unet.state_dict().items()}
+feng_peft_model.load_state_dict(original_state_dict, strict=True)
+```
+
+Load a base UNet model and load the adapters.
+
+```py
+base_unet = AutoModel.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+    use_safetensors=True,
+    variant="fp16",
+    subfolder="unet",
+).to("cuda")
+
+model = PeftModel.from_pretrained(
+    base_unet,
+    "stevhliu/ikea_peft_model",
+    use_safetensors=True,
+    subfolder="ikea",
+    adapter_name="ikea"
+)
+model.load_adapter(
+    "stevhliu/feng_peft_model",
+    use_safetensors=True,
+    subfolder="feng",
+    adapter_name="feng"
+)
+```
+
+Merge the LoRAs with [`~peft.LoraModel.add_weighted_adapter`] and specify how you want to merge them with `combination_type`. The example below uses the `"dare_linear"` method (refer to this [blog post](https://huggingface.co/blog/peft_merging) to learn more about these merging methods), which randomly prunes some weights and then performs a weighted sum of the tensors based on the set weightage of each LoRA in `weights`.
+
+Activate the merged LoRAs with [`~loaders.PeftAdapterMixin.set_adapters`].
+
+```py
+model.add_weighted_adapter(
+    adapters=["ikea", "feng"],
+    combination_type="dare_linear",
+    weights=[1.0, 1.0],
+    adapter_name="ikea-feng"
+)
+model.set_adapters("ikea-feng")
+
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    unet=model,
+    variant="fp16",
+    torch_dtype=torch.float16,
+).to("cuda")
+pipeline("A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai").images[0]
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ikea-feng-dare-linear.png"/>
+</div>
+
+### fuse_lora
+
+The [`~loaders.lora_base.LoraBaseMixin.fuse_lora`] method fuses the LoRA weights directly with the original UNet and text encoder weights of the underlying model. This reduces the overhead of loading the underlying model for each LoRA because it only loads the model once, which lowers memory usage and increases inference speed.
+
+```py
+import torch
+from diffusers import DiffusionPipeline
+
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
+pipeline.load_lora_weights(
+    "lordjia/by-feng-zikai",
+    weight_name="fengzikai_v1.0_XL.safetensors",
+    adapter_name="feng"
+)
+pipeline.set_adapters(["ikea", "feng"], adapter_weights=[0.7, 0.8])
+```
+
+Call [`~loaders.lora_base.LoraBaseMixin.fuse_lora`] to fuse them. The `lora_scale` parameter controls how much to scale the output by with the LoRA weights. It is important to make this adjustment now because passing `scale` to `cross_attention_kwargs` won't work in the pipeline.
+
+```py
+pipeline.fuse_lora(adapter_names=["ikea", "feng"], lora_scale=1.0)
+```
+
+Unload the LoRA weights since they're already fused with the underlying model. Save the fused pipeline with either [`~DiffusionPipeline.save_pretrained`] to save it locally or [`~PushToHubMixin.push_to_hub`] to save it to the Hub.
+
+<hfoptions id="save">
+<hfoption id="save locally">
+
+```py
+pipeline.unload_lora_weights()
+pipeline.save_pretrained("path/to/fused-pipeline")
+```
+
+</hfoption>
+<hfoption id="save to Hub">
+
+```py
+pipeline.unload_lora_weights()
+pipeline.push_to_hub("fused-ikea-feng")
+```
+
+</hfoption>
+</hfoptions>
+
+The fused pipeline can now be quickly loaded for inference without requiring each LoRA to be separately loaded.
+
+```py
+pipeline = DiffusionPipeline.from_pretrained(
+    "username/fused-ikea-feng", torch_dtype=torch.float16,
+).to("cuda")
+pipeline("A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai").images[0]
+```
+
+Use [`~loaders.LoraLoaderMixin.unfuse_lora`] to restore the underlying models weights, for example, if you want to use a different `lora_scale` value. You can only unfuse if there is a single LoRA fused. For example, it won't work with the pipeline from above because there are multiple fused LoRAs. In these cases, you'll need to reload the entire model.
+
+```py
+pipeline.unfuse_lora()
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/fuse_lora.png"/>
+</div>
+
+## Manage
+
+Diffusers provides several methods to help you manage working with LoRAs. These methods can be especially useful if you're working with multiple LoRAs.
+
+### set_adapters
+
+[`~loaders.PeftAdapterMixin.set_adapters`] also activates the current LoRA to use if there are multiple active LoRAs. This allows you to switch between different LoRAs by specifying their name.
+
+```py
+import torch
+from diffusers import DiffusionPipeline
+
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_lora_weights(
+    "ostris/ikea-instructions-lora-sdxl",
+    weight_name="ikea_instructions_xl_v1_5.safetensors",
+    adapter_name="ikea"
+)
+pipeline.load_lora_weights(
+    "lordjia/by-feng-zikai",
+    weight_name="fengzikai_v1.0_XL.safetensors",
+    adapter_name="feng"
+)
+# activates the feng LoRA instead of the ikea LoRA
+pipeline.set_adapters("feng")
+```
+
+### save_lora_adapter
+
+Save an adapter with [`~loaders.PeftAdapterMixin.save_lora_adapter`].
+
+```py
+import torch
+from diffusers import AutoPipelineForText2Image
+
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.unet.load_lora_adapter(
+    "jbilcke-hf/sdxl-cinematic-1",
+    weight_name="pytorch_lora_weights.safetensors",
+    adapter_name="cinematic"
+    prefix="unet"
+)
+pipeline.save_lora_adapter("path/to/save", adapter_name="cinematic")
+```
+
+### unload_lora_weights
+
+The [`~loaders.lora_base.LoraBaseMixin.unload_lora_weights`] method unloads any LoRA weights in the pipeline to restore the underlying model weights.
+
+```py
+pipeline.unload_lora_weights()
+```
+
+### disable_lora
+
+The [`~loaders.PeftAdapterMixin.disable_lora`] method disables all LoRAs (but they're still kept on the pipeline) and restores the pipeline to the underlying model weights.
+
+```py
+pipeline.disable_lora()
+```
+
+### get_active_adapters
+
+The [`~loaders.lora_base.LoraBaseMixin.get_active_adapters`] method returns a list of active LoRAs attached to a pipeline.
+
+```py
+pipeline.get_active_adapters()
+["cereal", "ikea"]
+```
+
+### get_list_adapters
+
+The [`~loaders.lora_base.LoraBaseMixin.get_list_adapters`] method returns the active LoRAs for each component in the pipeline.
+
+```py
+pipeline.get_list_adapters()
+{"unet": ["cereal", "ikea"], "text_encoder_2": ["cereal"]}
+```
+
+### delete_adapters
+
+The [`~loaders.PeftAdapterMixin.delete_adapters`] method completely removes a LoRA and its layers from a model.
+
+```py
+pipeline.delete_adapters("ikea")
+```
+
+## Resources
+
+Browse the [LoRA Studio](https://lorastudio.co/models) for different LoRAs to use or you can upload your favorite LoRAs from Civitai to the Hub with the Space below.
+
+<iframe
+	src="https://multimodalart-civitai-to-hf.hf.space"
+	frameborder="0"
+	width="850"
+	height="450"
+></iframe>
+
+You can find additional LoRAs in the [FLUX LoRA the Explorer](https://huggingface.co/spaces/multimodalart/flux-lora-the-explorer) and [LoRA the Explorer](https://huggingface.co/spaces/multimodalart/LoraTheExplorer) Spaces.

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

@@ -1,120 +0,0 @@
-<!--Copyright 2024 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.
--->
-# CogVideoX
-
-CogVideoX is a text-to-video generation model focused on creating more coherent videos aligned with a prompt. It achieves this using several methods.
-
-- a 3D variational autoencoder that compresses videos spatially and temporally, improving compression rate and video accuracy.
-
-- an expert transformer block to help align text and video, and a 3D full attention module for capturing and creating spatially and temporally accurate videos.
- 
-
-
-## Load model checkpoints
-Model weights may be stored in separate subfolders on the Hub or locally, in which case, you should use the [`~DiffusionPipeline.from_pretrained`] method.
-
-
-```py
-from diffusers import CogVideoXPipeline, CogVideoXImageToVideoPipeline
-pipe = CogVideoXPipeline.from_pretrained(
-    "THUDM/CogVideoX-2b",
-    torch_dtype=torch.float16
-)
-
-pipe = CogVideoXImageToVideoPipeline.from_pretrained(
-    "THUDM/CogVideoX-5b-I2V",
-    torch_dtype=torch.bfloat16
-)
-
-```
-
-## Text-to-Video
-For text-to-video, pass a text prompt. By default, CogVideoX generates a 720x480 video for the best results.
-
-```py
-import torch
-from diffusers import CogVideoXPipeline
-from diffusers.utils import export_to_video
-
-prompt = "An elderly gentleman, with a serene expression, sits at the water's edge, a steaming cup of tea by his side. He is engrossed in his artwork, brush in hand, as he renders an oil painting on a canvas that's propped up against a small, weathered table. The sea breeze whispers through his silver hair, gently billowing his loose-fitting white shirt, while the salty air adds an intangible element to his masterpiece in progress. The scene is one of tranquility and inspiration, with the artist's canvas capturing the vibrant hues of the setting sun reflecting off the tranquil sea."
-
-pipe = CogVideoXPipeline.from_pretrained(
-    "THUDM/CogVideoX-5b",
-    torch_dtype=torch.bfloat16
-)
-
-pipe.enable_model_cpu_offload()
-pipe.vae.enable_tiling()
-
-video = pipe(
-    prompt=prompt,
-    num_videos_per_prompt=1,
-    num_inference_steps=50,
-    num_frames=49,
-    guidance_scale=6,
-    generator=torch.Generator(device="cuda").manual_seed(42),
-).frames[0]
-
-export_to_video(video, "output.mp4", fps=8)
-
-```
-
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cogvideox/cogvideox_out.gif" alt="generated image of an astronaut in a jungle"/>
-</div>
-
-
-## Image-to-Video
-
-
-You'll use the [THUDM/CogVideoX-5b-I2V](https://huggingface.co/THUDM/CogVideoX-5b-I2V)  checkpoint for this guide.
-
-```py
-import torch
-from diffusers import CogVideoXImageToVideoPipeline
-from diffusers.utils import export_to_video, load_image
-
-prompt = "A vast, shimmering ocean flows gracefully under a twilight sky, its waves undulating in a mesmerizing dance of blues and greens. The surface glints with the last rays of the setting sun, casting golden highlights that ripple across the water. Seagulls soar above, their cries blending with the gentle roar of the waves. The horizon stretches infinitely, where the ocean meets the sky in a seamless blend of hues. Close-ups reveal the intricate patterns of the waves, capturing the fluidity and dynamic beauty of the sea in motion."
-image = load_image(image="cogvideox_rocket.png")
-pipe = CogVideoXImageToVideoPipeline.from_pretrained(
-    "THUDM/CogVideoX-5b-I2V",
-    torch_dtype=torch.bfloat16
-)
- 
-pipe.vae.enable_tiling()
-pipe.vae.enable_slicing()
-
-video = pipe(
-    prompt=prompt,
-    image=image,
-    num_videos_per_prompt=1,
-    num_inference_steps=50,
-    num_frames=49,
-    guidance_scale=6,
-    generator=torch.Generator(device="cuda").manual_seed(42),
-).frames[0]
-
-export_to_video(video, "output.mp4", fps=8)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cogvideox/cogvideox_rocket.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">initial image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cogvideox/cogvideox_outrocket.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">generated video</figcaption>
-  </div>
-</div>
-

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

@@ -303,7 +303,7 @@ There are many types of conditioning inputs you can use, and 🤗 Diffusers supp
 
 Diffusion models are large, and the iterative nature of denoising an image is computationally expensive and intensive. But this doesn't mean you need access to powerful - or even many - GPUs to use them. There are many optimization techniques for running diffusion models on consumer and free-tier resources. For example, you can load model weights in half-precision to save GPU memory and increase speed or offload the entire model to the GPU to save even more memory.
 
-PyTorch 2.0 also supports a more memory-efficient attention mechanism called [*scaled dot product attention*](../optimization/torch2.0#scaled-dot-product-attention) that is automatically enabled if you're using PyTorch 2.0. You can combine this with [`torch.compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) to speed your code up even more:
+PyTorch 2.0 also supports a more memory-efficient attention mechanism called [*scaled dot product attention*](../optimization/fp16#scaled-dot-product-attention) that is automatically enabled if you're using PyTorch 2.0. You can combine this with [`torch.compile`](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial.html) to speed your code up even more:
 
 ```py
 from diffusers import AutoPipelineForText2Image
@@ -313,4 +313,4 @@ pipeline = AutoPipelineForText2Image.from_pretrained("stable-diffusion-v1-5/stab
 pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
 ```
 
-For more tips on how to optimize your code to save memory and speed up inference, read the [Memory and speed](../optimization/fp16) and [Torch 2.0](../optimization/torch2.0) guides.
+For more tips on how to optimize your code to save memory and speed up inference, read the [Accelerate inference](../optimization/fp16) and [Reduce memory usage](../optimization/memory) guides.

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

@@ -65,14 +65,14 @@ For convenience, we provide a table to denote which methods are inference-only a
 |                [Fabric](#fabric)                    |         ✅         |                   ❌                    |                                                                                                 |
 ## InstructPix2Pix
 
-[Paper](https://arxiv.org/abs/2211.09800)
+[Paper](https://huggingface.co/papers/2211.09800)
 
 [InstructPix2Pix](../api/pipelines/pix2pix) is fine-tuned from Stable Diffusion to support editing input images. It takes as inputs an image and a prompt describing an edit, and it outputs the edited image.
 InstructPix2Pix has been explicitly trained to work well with [InstructGPT](https://openai.com/blog/instruction-following/)-like prompts.
 
 ## Pix2Pix Zero
 
-[Paper](https://arxiv.org/abs/2302.03027)
+[Paper](https://huggingface.co/papers/2302.03027)
 
 [Pix2Pix Zero](../api/pipelines/pix2pix_zero) allows modifying an image so that one concept or subject is translated to another one while preserving general image semantics.
 
@@ -104,7 +104,7 @@ apply Pix2Pix Zero to any of the available Stable Diffusion models.
 
 ## Attend and Excite
 
-[Paper](https://arxiv.org/abs/2301.13826)
+[Paper](https://huggingface.co/papers/2301.13826)
 
 [Attend and Excite](../api/pipelines/attend_and_excite) allows subjects in the prompt to be faithfully represented in the final image.
 
@@ -114,7 +114,7 @@ Like Pix2Pix Zero, Attend and Excite also involves a mini optimization loop (lea
 
 ## Semantic Guidance (SEGA)
 
-[Paper](https://arxiv.org/abs/2301.12247)
+[Paper](https://huggingface.co/papers/2301.12247)
 
 [SEGA](../api/pipelines/semantic_stable_diffusion) allows applying or removing one or more concepts from an image. The strength of the concept can also be controlled. I.e. the smile concept can be used to incrementally increase or decrease the smile of a portrait.
 
@@ -124,7 +124,7 @@ Unlike Pix2Pix Zero or Attend and Excite, SEGA directly interacts with the diffu
 
 ## Self-attention Guidance (SAG)
 
-[Paper](https://arxiv.org/abs/2210.00939)
+[Paper](https://huggingface.co/papers/2210.00939)
 
 [Self-attention Guidance](../api/pipelines/self_attention_guidance) improves the general quality of images.
 
@@ -140,7 +140,7 @@ It conditions on a monocular depth estimate of the original image.
 
 ## MultiDiffusion Panorama
 
-[Paper](https://arxiv.org/abs/2302.08113)
+[Paper](https://huggingface.co/papers/2302.08113)
 
 [MultiDiffusion Panorama](../api/pipelines/panorama) defines a new generation process over a pre-trained diffusion model. This process binds together multiple diffusion generation methods that can be readily applied to generate high quality and diverse images. Results adhere to user-provided controls, such as desired aspect ratio (e.g., panorama), and spatial guiding signals, ranging from tight segmentation masks to bounding boxes.
 MultiDiffusion Panorama allows to generate high-quality images at arbitrary aspect ratios (e.g., panoramas).
@@ -157,13 +157,13 @@ In addition to pre-trained models, Diffusers has training scripts for fine-tunin
 
 ## Textual Inversion
 
-[Paper](https://arxiv.org/abs/2208.01618)
+[Paper](https://huggingface.co/papers/2208.01618)
 
 [Textual Inversion](../training/text_inversion) fine-tunes a model to teach it about a new concept. I.e. a few pictures of a style of artwork can be used to generate images in that style.
 
 ## ControlNet
 
-[Paper](https://arxiv.org/abs/2302.05543)
+[Paper](https://huggingface.co/papers/2302.05543)
 
 [ControlNet](../api/pipelines/controlnet) is an auxiliary network which adds an extra condition.
 There are 8 canonical pre-trained ControlNets trained on different conditionings such as edge detection, scribbles,
@@ -176,7 +176,7 @@ input.
 
 ## Custom Diffusion
 
-[Paper](https://arxiv.org/abs/2212.04488)
+[Paper](https://huggingface.co/papers/2212.04488)
 
 [Custom Diffusion](../training/custom_diffusion) only fine-tunes the cross-attention maps of a pre-trained
 text-to-image diffusion model. It also allows for additionally performing Textual Inversion. It supports
@@ -186,7 +186,7 @@ concept(s) of interest.
 
 ## Model Editing
 
-[Paper](https://arxiv.org/abs/2303.08084)
+[Paper](https://huggingface.co/papers/2303.08084)
 
 The [text-to-image model editing pipeline](../api/pipelines/model_editing) helps you mitigate some of the incorrect implicit assumptions a pre-trained text-to-image
 diffusion model might make about the subjects present in the input prompt. For example, if you prompt Stable Diffusion to generate images for "A pack of roses", the roses in the generated images
@@ -194,14 +194,14 @@ are more likely to be red. This pipeline helps you change that assumption.
 
 ## DiffEdit
 
-[Paper](https://arxiv.org/abs/2210.11427)
+[Paper](https://huggingface.co/papers/2210.11427)
 
 [DiffEdit](../api/pipelines/diffedit) allows for semantic editing of input images along with
 input prompts while preserving the original input images as much as possible.
 
 ## T2I-Adapter
 
-[Paper](https://arxiv.org/abs/2302.08453)
+[Paper](https://huggingface.co/papers/2302.08453)
 
 [T2I-Adapter](../api/pipelines/stable_diffusion/adapter) is an auxiliary network which adds an extra condition.
 There are 8 canonical pre-trained adapters trained on different conditionings such as edge detection, sketch,
@@ -209,7 +209,7 @@ depth maps, and semantic segmentations.
 
 ## Fabric
 
-[Paper](https://arxiv.org/abs/2307.10159)
+[Paper](https://huggingface.co/papers/2307.10159)
 
 [Fabric](https://github.com/huggingface/diffusers/tree/442017ccc877279bcf24fbe92f92d3d0def191b6/examples/community#stable-diffusion-fabric-pipeline) is a training-free
 approach applicable to a wide range of popular diffusion models, which exploits

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

@@ -12,46 +12,28 @@ specific language governing permissions and limitations under the License.
 
 # ControlNet
 
-ControlNet is a type of model for controlling image diffusion models by conditioning the model with an additional input image. There are many types of conditioning inputs (canny edge, user sketching, human pose, depth, and more) you can use to control a diffusion model. This is hugely useful because it affords you greater control over image generation, making it easier to generate specific images without experimenting with different text prompts or denoising values as much.
+[ControlNet](https://huggingface.co/papers/2302.05543) is an adapter that enables controllable generation such as generating an image of a cat in a *specific pose* or following the lines in a sketch of a *specific* cat. It works by adding a smaller network of "zero convolution" layers and progressively training these to avoid disrupting with the original model. The original model parameters are frozen to avoid retraining it.
 
-<Tip>
+A ControlNet is conditioned on extra visual information or "structural controls" (canny edge, depth maps, human pose, etc.) that can be combined with text prompts to generate images that are guided by the visual input.
 
-Check out Section 3.5 of the [ControlNet](https://huggingface.co/papers/2302.05543) paper v1 for a list of ControlNet implementations on various conditioning inputs. You can find the official Stable Diffusion ControlNet conditioned models on [lllyasviel](https://huggingface.co/lllyasviel)'s Hub profile, and more [community-trained](https://huggingface.co/models?other=stable-diffusion&other=controlnet) ones on the Hub.
+> [!TIP]
+> ControlNets are available to many models such as [Flux](../api/pipelines/controlnet_flux), [Hunyuan-DiT](../api/pipelines/controlnet_hunyuandit), [Stable Diffusion 3](../api/pipelines/controlnet_sd3), and more. The examples in this guide use Flux and Stable Diffusion XL.
 
-For Stable Diffusion XL (SDXL) ControlNet models, you can find them on the 🤗 [Diffusers](https://huggingface.co/diffusers) Hub organization, or you can browse [community-trained](https://huggingface.co/models?other=stable-diffusion-xl&other=controlnet) ones on the Hub.
+Load a ControlNet conditioned on a specific control, such as canny edge, and pass it to the pipeline in [`~DiffusionPipeline.from_pretrained`].
 
-</Tip>
+<hfoptions id="usage">
+<hfoption id="text-to-image">
 
-A ControlNet model has two sets of weights (or blocks) connected by a zero-convolution layer:
-
-- a *locked copy* keeps everything a large pretrained diffusion model has learned
-- a *trainable copy* is trained on the additional conditioning input
-
-Since the locked copy preserves the pretrained model, training and implementing a ControlNet on a new conditioning input is as fast as finetuning any other model because you aren't training the model from scratch.
-
-This guide will show you how to use ControlNet for text-to-image, image-to-image, inpainting, and more! There are many types of ControlNet conditioning inputs to choose from, but in this guide we'll only focus on several of them. Feel free to experiment with other conditioning inputs!
-
-Before you begin, make sure you have the following libraries installed:
+Generate a canny image with [opencv-python](https://github.com/opencv/opencv-python).
 
 ```py
-# uncomment to install the necessary libraries in Colab
-#!pip install -q diffusers transformers accelerate opencv-python
-```
-
-## Text-to-image
-
-For text-to-image, you normally pass a text prompt to the model. But with ControlNet, you can specify an additional conditioning input. Let's condition the model with a canny image, a white outline of an image on a black background. This way, the ControlNet can use the canny image as a control to guide the model to generate an image with the same outline.
-
-Load an image and use the [opencv-python](https://github.com/opencv/opencv-python) library to extract the canny image:
-
-```py
-from diffusers.utils import load_image, make_image_grid
-from PIL import Image
 import cv2
 import numpy as np
+from PIL import Image
+from diffusers.utils import load_image
 
 original_image = load_image(
-    "https://hf.co/datasets/huggingface/documentation-images/resolve/main/diffusers/input_image_vermeer.png"
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png"
 )
 
 image = np.array(original_image)
@@ -65,523 +47,300 @@ image = np.concatenate([image, image, image], axis=2)
 canny_image = Image.fromarray(image)
 ```
 
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/input_image_vermeer.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/vermeer_canny_edged.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">canny image</figcaption>
-  </div>
-</div>
-
-Next, load a ControlNet model conditioned on canny edge detection and pass it to the [`StableDiffusionControlNetPipeline`]. Use the faster [`UniPCMultistepScheduler`] and enable model offloading to speed up inference and reduce memory usage.
+Pass the canny image to the pipeline. Use the `controlnet_conditioning_scale` parameter to determine how much weight to assign to the control.
 
 ```py
-from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, UniPCMultistepScheduler
 import torch
+from diffusers.utils import load_image
+from diffusers import FluxControlNetPipeline, FluxControlNetModel
 
-controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionControlNetPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5", controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True
+controlnet = FluxControlNetModel.from_pretrained(
+    "InstantX/FLUX.1-dev-Controlnet-Canny", torch_dtype=torch.bfloat16
 )
+pipeline = FluxControlNetPipeline.from_pretrained(
+    "black-forest-labs/FLUX.1-dev", controlnet=controlnet, torch_dtype=torch.bfloat16
+).to("cuda")
 
-pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
-pipe.enable_model_cpu_offload()
-```
+prompt = """
+A photorealistic overhead image of a cat reclining sideways in a flamingo pool floatie holding a margarita. 
+The cat is floating leisurely in the pool and completely relaxed and happy.
+"""
 
-Now pass your prompt and canny image to the pipeline:
-
-```py
-output = pipe(
-    "the mona lisa", image=canny_image
+pipeline(
+    prompt, 
+    control_image=canny_image,
+    controlnet_conditioning_scale=0.5,
+    num_inference_steps=50, 
+    guidance_scale=3.5,
 ).images[0]
-make_image_grid([original_image, canny_image, output], rows=1, cols=3)
 ```
 
-<div class="flex justify-center">
-  <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-text2img.png"/>
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png" width="300" alt="Generated image (prompt only)"/>
+    <figcaption style="text-align: center;">original image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">canny image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat-generated.png" width="300" alt="Generated image (ControlNet + prompt)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
 </div>
 
-## Image-to-image
 
-For image-to-image, you'd typically pass an initial image and a prompt to the pipeline to generate a new image. With ControlNet, you can pass an additional conditioning input to guide the model. Let's condition the model with a depth map, an image which contains spatial information. This way, the ControlNet can use the depth map as a control to guide the model to generate an image that preserves spatial information.
+</hfoption>
+<hfoption id="image-to-image">
 
-You'll use the [`StableDiffusionControlNetImg2ImgPipeline`] for this task, which is different from the [`StableDiffusionControlNetPipeline`] because it allows you to pass an initial image as the starting point for the image generation process.
-
-Load an image and use the `depth-estimation` [`~transformers.Pipeline`] from 🤗 Transformers to extract the depth map of an image:
+Generate a depth map with a depth estimation pipeline from Transformers.
 
 ```py
 import torch
 import numpy as np
+from PIL import Image
+from transformers import DPTImageProcessor, DPTForDepthEstimation
+from diffusers import ControlNetModel, StableDiffusionXLControlNetImg2ImgPipeline, AutoencoderKL
+from diffusers.utils import load_image
 
-from transformers import pipeline
-from diffusers.utils import load_image, make_image_grid
 
-image = load_image(
-    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-img2img.jpg"
-)
+depth_estimator = DPTForDepthEstimation.from_pretrained("Intel/dpt-hybrid-midas").to("cuda")
+feature_extractor = DPTImageProcessor.from_pretrained("Intel/dpt-hybrid-midas")
 
-def get_depth_map(image, depth_estimator):
-    image = depth_estimator(image)["depth"]
-    image = np.array(image)
-    image = image[:, :, None]
-    image = np.concatenate([image, image, image], axis=2)
-    detected_map = torch.from_numpy(image).float() / 255.0
-    depth_map = detected_map.permute(2, 0, 1)
-    return depth_map
+def get_depth_map(image):
+    image = feature_extractor(images=image, return_tensors="pt").pixel_values.to("cuda")
+    with torch.no_grad(), torch.autocast("cuda"):
+        depth_map = depth_estimator(image).predicted_depth
 
-depth_estimator = pipeline("depth-estimation")
-depth_map = get_depth_map(image, depth_estimator).unsqueeze(0).half().to("cuda")
-```
-
-Next, load a ControlNet model conditioned on depth maps and pass it to the [`StableDiffusionControlNetImg2ImgPipeline`]. Use the faster [`UniPCMultistepScheduler`] and enable model offloading to speed up inference and reduce memory usage.
-
-```py
-from diffusers import StableDiffusionControlNetImg2ImgPipeline, ControlNetModel, UniPCMultistepScheduler
-import torch
-
-controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11f1p_sd15_depth", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionControlNetImg2ImgPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5", controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True
-)
-
-pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
-pipe.enable_model_cpu_offload()
-```
-
-Now pass your prompt, initial image, and depth map to the pipeline:
-
-```py
-output = pipe(
-    "lego batman and robin", image=image, control_image=depth_map,
-).images[0]
-make_image_grid([image, output], rows=1, cols=2)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-img2img.jpg"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-img2img-2.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">generated image</figcaption>
-  </div>
-</div>
-
-## Inpainting
-
-For inpainting, you need an initial image, a mask image, and a prompt describing what to replace the mask with. ControlNet models allow you to add another control image to condition a model with. Let’s condition the model with an inpainting mask. This way, the ControlNet can use the inpainting mask as a control to guide the model to generate an image within the mask area.
-
-Load an initial image and a mask image:
-
-```py
-from diffusers.utils import load_image, make_image_grid
-
-init_image = load_image(
-    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-inpaint.jpg"
-)
-init_image = init_image.resize((512, 512))
-
-mask_image = load_image(
-    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-inpaint-mask.jpg"
-)
-mask_image = mask_image.resize((512, 512))
-make_image_grid([init_image, mask_image], rows=1, cols=2)
-```
-
-Create a function to prepare the control image from the initial and mask images. This'll create a tensor to mark the pixels in `init_image` as masked if the corresponding pixel in `mask_image` is over a certain threshold.
-
-```py
-import numpy as np
-import torch
-
-def make_inpaint_condition(image, image_mask):
-    image = np.array(image.convert("RGB")).astype(np.float32) / 255.0
-    image_mask = np.array(image_mask.convert("L")).astype(np.float32) / 255.0
-
-    assert image.shape[0:1] == image_mask.shape[0:1]
-    image[image_mask > 0.5] = -1.0  # set as masked pixel
-    image = np.expand_dims(image, 0).transpose(0, 3, 1, 2)
-    image = torch.from_numpy(image)
+    depth_map = torch.nn.functional.interpolate(
+        depth_map.unsqueeze(1),
+        size=(1024, 1024),
+        mode="bicubic",
+        align_corners=False,
+    )
+    depth_min = torch.amin(depth_map, dim=[1, 2, 3], keepdim=True)
+    depth_max = torch.amax(depth_map, dim=[1, 2, 3], keepdim=True)
+    depth_map = (depth_map - depth_min) / (depth_max - depth_min)
+    image = torch.cat([depth_map] * 3, dim=1)
+    image = image.permute(0, 2, 3, 1).cpu().numpy()[0]
+    image = Image.fromarray((image * 255.0).clip(0, 255).astype(np.uint8))
     return image
 
-control_image = make_inpaint_condition(init_image, mask_image)
+depth_image = get_depth_map(image)
 ```
 
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-inpaint.jpg"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-inpaint-mask.jpg"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">mask image</figcaption>
-  </div>
-</div>
-
-Load a ControlNet model conditioned on inpainting and pass it to the [`StableDiffusionControlNetInpaintPipeline`]. Use the faster [`UniPCMultistepScheduler`] and enable model offloading to speed up inference and reduce memory usage.
-
-```py
-from diffusers import StableDiffusionControlNetInpaintPipeline, ControlNetModel, UniPCMultistepScheduler
-
-controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11p_sd15_inpaint", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionControlNetInpaintPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5", controlnet=controlnet, torch_dtype=torch.float16, use_safetensors=True
-)
-
-pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
-pipe.enable_model_cpu_offload()
-```
-
-Now pass your prompt, initial image, mask image, and control image to the pipeline:
-
-```py
-output = pipe(
-    "corgi face with large ears, detailed, pixar, animated, disney",
-    num_inference_steps=20,
-    eta=1.0,
-    image=init_image,
-    mask_image=mask_image,
-    control_image=control_image,
-).images[0]
-make_image_grid([init_image, mask_image, output], rows=1, cols=3)
-```
-
-<div class="flex justify-center">
-  <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet-inpaint-result.png"/>
-</div>
-
-## Guess mode
-
-[Guess mode](https://github.com/lllyasviel/ControlNet/discussions/188) does not require supplying a prompt to a ControlNet at all! This forces the ControlNet encoder to do its best to "guess" the contents of the input control map (depth map, pose estimation, canny edge, etc.).
-
-Guess mode adjusts the scale of the output residuals from a ControlNet by a fixed ratio depending on the block depth. The shallowest `DownBlock` corresponds to 0.1, and as the blocks get deeper, the scale increases exponentially such that the scale of the `MidBlock` output becomes 1.0.
-
-<Tip>
-
-Guess mode does not have any impact on prompt conditioning and you can still provide a prompt if you want.
-
-</Tip>
-
-Set `guess_mode=True` in the pipeline, and it is [recommended](https://github.com/lllyasviel/ControlNet#guess-mode--non-prompt-mode) to set the `guidance_scale` value between 3.0 and 5.0.
-
-```py
-from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
-from diffusers.utils import load_image, make_image_grid
-import numpy as np
-import torch
-from PIL import Image
-import cv2
-
-controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", use_safetensors=True)
-pipe = StableDiffusionControlNetPipeline.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5", controlnet=controlnet, use_safetensors=True).to("cuda")
-
-original_image = load_image("https://huggingface.co/takuma104/controlnet_dev/resolve/main/bird_512x512.png")
-
-image = np.array(original_image)
-
-low_threshold = 100
-high_threshold = 200
-
-image = cv2.Canny(image, low_threshold, high_threshold)
-image = image[:, :, None]
-image = np.concatenate([image, image, image], axis=2)
-canny_image = Image.fromarray(image)
-
-image = pipe("", image=canny_image, guess_mode=True, guidance_scale=3.0).images[0]
-make_image_grid([original_image, canny_image, image], rows=1, cols=3)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_guess_mode/output_images/diffusers/output_bird_canny_0.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">regular mode with prompt</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/takuma104/controlnet_dev/resolve/main/gen_compare_guess_mode/output_images/diffusers/output_bird_canny_0_gm.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">guess mode without prompt</figcaption>
-  </div>
-</div>
-
-## ControlNet with Stable Diffusion XL
-
-There aren't too many ControlNet models compatible with Stable Diffusion XL (SDXL) at the moment, but we've trained two full-sized ControlNet models for SDXL conditioned on canny edge detection and depth maps. We're also experimenting with creating smaller versions of these SDXL-compatible ControlNet models so it is easier to run on resource-constrained hardware. You can find these checkpoints on the [🤗 Diffusers Hub organization](https://huggingface.co/diffusers)!
-
-Let's use a SDXL ControlNet conditioned on canny images to generate an image. Start by loading an image and prepare the canny image:
-
-```py
-from diffusers import StableDiffusionXLControlNetPipeline, ControlNetModel, AutoencoderKL
-from diffusers.utils import load_image, make_image_grid
-from PIL import Image
-import cv2
-import numpy as np
-import torch
-
-original_image = load_image(
-    "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/hf-logo.png"
-)
-
-image = np.array(original_image)
-
-low_threshold = 100
-high_threshold = 200
-
-image = cv2.Canny(image, low_threshold, high_threshold)
-image = image[:, :, None]
-image = np.concatenate([image, image, image], axis=2)
-canny_image = Image.fromarray(image)
-make_image_grid([original_image, canny_image], rows=1, cols=2)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/hf-logo.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/hf-logo-canny.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">canny image</figcaption>
-  </div>
-</div>
-
-Load a SDXL ControlNet model conditioned on canny edge detection and pass it to the [`StableDiffusionXLControlNetPipeline`]. You can also enable model offloading to reduce memory usage.
+Pass the depth map to the pipeline. Use the `controlnet_conditioning_scale` parameter to determine how much weight to assign to the control.
 
 ```py
 controlnet = ControlNetModel.from_pretrained(
-    "diffusers/controlnet-canny-sdxl-1.0",
+    "diffusers/controlnet-depth-sdxl-1.0-small",
     torch_dtype=torch.float16,
-    use_safetensors=True
 )
-vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
+vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
+pipeline = StableDiffusionXLControlNetImg2ImgPipeline.from_pretrained(
     "stabilityai/stable-diffusion-xl-base-1.0",
     controlnet=controlnet,
     vae=vae,
     torch_dtype=torch.float16,
-    use_safetensors=True
-)
-pipe.enable_model_cpu_offload()
-```
+).to("cuda")
 
-Now pass your prompt (and optionally a negative prompt if you're using one) and canny image to the pipeline:
-
-<Tip>
-
-The [`controlnet_conditioning_scale`](https://huggingface.co/docs/diffusers/main/en/api/pipelines/controlnet#diffusers.StableDiffusionControlNetPipeline.__call__.controlnet_conditioning_scale) parameter determines how much weight to assign to the conditioning inputs. A value of 0.5 is recommended for good generalization, but feel free to experiment with this number!
-
-</Tip>
-
-```py
-prompt = "aerial view, a futuristic research complex in a bright foggy jungle, hard lighting"
-negative_prompt = 'low quality, bad quality, sketches'
-
-image = pipe(
+prompt = """
+A photorealistic overhead image of a cat reclining sideways in a flamingo pool floatie holding a margarita. 
+The cat is floating leisurely in the pool and completely relaxed and happy.
+"""
+image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png"
+).resize((1024, 1024))
+controlnet_conditioning_scale = 0.5 
+pipeline(
     prompt,
-    negative_prompt=negative_prompt,
-    image=canny_image,
-    controlnet_conditioning_scale=0.5,
+    image=image,
+    control_image=depth_image,
+    controlnet_conditioning_scale=controlnet_conditioning_scale,
+    strength=0.99,
+    num_inference_steps=100,
 ).images[0]
-make_image_grid([original_image, canny_image, image], rows=1, cols=3)
 ```
 
-<div class="flex justify-center">
-    <img class="rounded-xl" src="https://huggingface.co/diffusers/controlnet-canny-sdxl-1.0/resolve/main/out_hug_lab_7.png"/>
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png" width="300" alt="Generated image (prompt only)"/>
+    <figcaption style="text-align: center;">original image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_depth_image.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">depth map</figcaption>
+  </figure>
+  <figure> 
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_depth_cat.png" width="300" alt="Generated image (ControlNet + prompt)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
 </div>
 
-You can use [`StableDiffusionXLControlNetPipeline`] in guess mode as well by setting the parameter to `True`:
+</hfoption>
+<hfoption id="inpainting">
+
+Generate a mask image and convert it to a tensor to mark the pixels in the original image as masked if the corresponding pixel in the mask image is over a certain threshold.
 
 ```py
-from diffusers import StableDiffusionXLControlNetPipeline, ControlNetModel, AutoencoderKL
-from diffusers.utils import load_image, make_image_grid
-import numpy as np
-import torch
 import cv2
+import torch
+import numpy as np
 from PIL import Image
+from diffusers.utils import load_image
+from diffusers import StableDiffusionXLControlNetInpaintPipeline, ControlNetModel
 
-prompt = "aerial view, a futuristic research complex in a bright foggy jungle, hard lighting"
-negative_prompt = "low quality, bad quality, sketches"
-
-original_image = load_image(
-    "https://hf.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/hf-logo.png"
+init_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png"
 )
+init_image = init_image.resize((1024, 1024))
+mask_image = load_image(
+    "/content/cat_mask.png"
+)
+mask_image = mask_image.resize((1024, 1024))
 
+def make_canny_condition(image):
+    image = np.array(image)
+    image = cv2.Canny(image, 100, 200)
+    image = image[:, :, None]
+    image = np.concatenate([image, image, image], axis=2)
+    image = Image.fromarray(image)
+    return image
+
+control_image = make_canny_condition(init_image)
+```
+
+Pass the mask and control image to the pipeline. Use the `controlnet_conditioning_scale` parameter to determine how much weight to assign to the control.
+
+```py
 controlnet = ControlNetModel.from_pretrained(
-    "diffusers/controlnet-canny-sdxl-1.0", torch_dtype=torch.float16, use_safetensors=True
+    "diffusers/controlnet-canny-sdxl-1.0", torch_dtype=torch.float16
 )
-vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", controlnet=controlnet, vae=vae, torch_dtype=torch.float16, use_safetensors=True
+pipeline = StableDiffusionXLControlNetInpaintPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", controlnet=controlnet, torch_dtype=torch.float16
 )
-pipe.enable_model_cpu_offload()
-
-image = np.array(original_image)
-image = cv2.Canny(image, 100, 200)
-image = image[:, :, None]
-image = np.concatenate([image, image, image], axis=2)
-canny_image = Image.fromarray(image)
-
-image = pipe(
-    prompt, negative_prompt=negative_prompt, controlnet_conditioning_scale=0.5, image=canny_image, guess_mode=True,
-).images[0]
-make_image_grid([original_image, canny_image, image], rows=1, cols=3)
-```
-
-<Tip>
-
-You can use a refiner model with `StableDiffusionXLControlNetPipeline` to improve image quality, just like you can with a regular `StableDiffusionXLPipeline`.
-See the [Refine image quality](./sdxl#refine-image-quality) section to learn how to use the refiner model.
-Make sure to use `StableDiffusionXLControlNetPipeline` and pass `image` and `controlnet_conditioning_scale`.
-
-```py
-base = StableDiffusionXLControlNetPipeline(...)
-image = base(
-    prompt=prompt,
+pipeline(
+    "a cute and fluffy bunny rabbit",
+    num_inference_steps=100,
+    strength=0.99,
     controlnet_conditioning_scale=0.5,
-    image=canny_image,
-    num_inference_steps=40,
-    denoising_end=0.8,
-    output_type="latent",
-).images
-# rest exactly as with StableDiffusionXLPipeline
+    image=init_image,
+    mask_image=mask_image,
+    control_image=control_image,
+).images[0]
 ```
 
-</Tip>
-
-## MultiControlNet
-
-<Tip>
-
-Replace the SDXL model with a model like [stable-diffusion-v1-5/stable-diffusion-v1-5](https://huggingface.co/stable-diffusion-v1-5/stable-diffusion-v1-5) to use multiple conditioning inputs with Stable Diffusion models.
-
-</Tip>
-
-You can compose multiple ControlNet conditionings from different image inputs to create a *MultiControlNet*. To get better results, it is often helpful to:
-
-1. mask conditionings such that they don't overlap (for example, mask the area of a canny image where the pose conditioning is located)
-2. experiment with the [`controlnet_conditioning_scale`](https://huggingface.co/docs/diffusers/main/en/api/pipelines/controlnet#diffusers.StableDiffusionControlNetPipeline.__call__.controlnet_conditioning_scale) parameter to determine how much weight to assign to each conditioning input
-
-In this example, you'll combine a canny image and a human pose estimation image to generate a new image.
-
-Prepare the canny image conditioning:
-
-```py
-from diffusers.utils import load_image, make_image_grid
-from PIL import Image
-import numpy as np
-import cv2
-
-original_image = load_image(
-    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/landscape.png"
-)
-image = np.array(original_image)
-
-low_threshold = 100
-high_threshold = 200
-
-image = cv2.Canny(image, low_threshold, high_threshold)
-
-# zero out middle columns of image where pose will be overlaid
-zero_start = image.shape[1] // 4
-zero_end = zero_start + image.shape[1] // 2
-image[:, zero_start:zero_end] = 0
-
-image = image[:, :, None]
-image = np.concatenate([image, image, image], axis=2)
-canny_image = Image.fromarray(image)
-make_image_grid([original_image, canny_image], rows=1, cols=2)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/landscape.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/controlnet/landscape_canny_masked.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">canny image</figcaption>
-  </div>
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png" width="300" alt="Generated image (prompt only)"/>
+    <figcaption style="text-align: center;">original image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat_mask.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">mask image</figcaption>
+  </figure>
+  <figure> 
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_rabbit_inpaint.png" width="300" alt="Generated image (ControlNet + prompt)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
 </div>
 
-For human pose estimation, install [controlnet_aux](https://github.com/patrickvonplaten/controlnet_aux):
+</hfoption>
+</hfoptions>
+
+## Multi-ControlNet
+
+You can compose multiple ControlNet conditionings, such as canny image and a depth map, to create a *MultiControlNet*. For the best rersults, you should mask conditionings so they don't overlap and experiment with different `controlnet_conditioning_scale` parameters to adjust how much weight is assigned to each control input.
+
+The example below composes a canny image and depth map.
+
+Pass the ControlNets as a list to the pipeline and resize the images to the expected input size.
 
 ```py
-# uncomment to install the necessary library in Colab
-#!pip install -q controlnet-aux
-```
-
-Prepare the human pose estimation conditioning:
-
-```py
-from controlnet_aux import OpenposeDetector
-
-openpose = OpenposeDetector.from_pretrained("lllyasviel/ControlNet")
-original_image = load_image(
-    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/person.png"
-)
-openpose_image = openpose(original_image)
-make_image_grid([original_image, openpose_image], rows=1, cols=2)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/person.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/blog/controlnet/person_pose.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">human pose image</figcaption>
-  </div>
-</div>
-
-Load a list of ControlNet models that correspond to each conditioning, and pass them to the [`StableDiffusionXLControlNetPipeline`]. Use the faster [`UniPCMultistepScheduler`] and enable model offloading to reduce memory usage.
-
-```py
-from diffusers import StableDiffusionXLControlNetPipeline, ControlNetModel, AutoencoderKL, UniPCMultistepScheduler
 import torch
+from diffusers import StableDiffusionXLControlNetPipeline, ControlNetModel, AutoencoderKL
 
 controlnets = [
     ControlNetModel.from_pretrained(
-        "thibaud/controlnet-openpose-sdxl-1.0", torch_dtype=torch.float16
+        "diffusers/controlnet-depth-sdxl-1.0-small", torch_dtype=torch.float16
     ),
     ControlNetModel.from_pretrained(
-        "diffusers/controlnet-canny-sdxl-1.0", torch_dtype=torch.float16, use_safetensors=True
+        "diffusers/controlnet-canny-sdxl-1.0", torch_dtype=torch.float16,
     ),
 ]
 
-vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16, use_safetensors=True)
-pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", controlnet=controlnets, vae=vae, torch_dtype=torch.float16, use_safetensors=True
-)
-pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config)
-pipe.enable_model_cpu_offload()
+vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
+pipeline = StableDiffusionXLControlNetPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", controlnet=controlnets, vae=vae, torch_dtype=torch.float16
+).to("cuda")
+
+prompt = """
+a relaxed rabbit sitting on a striped towel next to a pool with a tropical drink nearby, 
+bright sunny day, vacation scene, 35mm photograph, film, professional, 4k, highly detailed
+"""
+negative_prompt = "lowres, bad anatomy, worst quality, low quality, deformed, ugly"
+
+images = [canny_image.resize((1024, 1024)), depth_image.resize((1024, 1024))]
+
+pipeline(
+    prompt,
+    negative_prompt=negative_prompt,
+    image=images,
+    num_inference_steps=100,
+    controlnet_conditioning_scale=[0.5, 0.5],
+    strength=0.7,
+).images[0]
 ```
 
-Now you can pass your prompt (an optional negative prompt if you're using one), canny image, and pose image to the pipeline:
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png" width="300" alt="Generated image (prompt only)"/>
+    <figcaption style="text-align: center;">canny image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/multicontrolnet_depth.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">depth map</figcaption>
+  </figure>
+  <figure> 
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_multi_controlnet.png" width="300" alt="Generated image (ControlNet + prompt)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
+</div>
+
+## guess_mode
+
+[Guess mode](https://github.com/lllyasviel/ControlNet/discussions/188) generates an image from **only** the control input (canny edge, depth map, pose, etc.) and without guidance from a prompt. It adjusts the scale of the ControlNet's output residuals by a fixed ratio depending on block depth. The earlier `DownBlock` is only scaled by `0.1` and the `MidBlock` is fully scaled by `1.0`.
 
 ```py
-prompt = "a giant standing in a fantasy landscape, best quality"
-negative_prompt = "monochrome, lowres, bad anatomy, worst quality, low quality"
+import torch
+from diffusers.utils import load_iamge
+from diffusers import StableDiffusionXLControlNetPipeline, ControlNetModel
 
-generator = torch.manual_seed(1)
+controlnet = ControlNetModel.from_pretrained(
+  "diffusers/controlnet-canny-sdxl-1.0", torch_dtype=torch.float16
+)
+pipeline = StableDiffusionXLControlNetPipeline.from_pretrained(
+  "stabilityai/stable-diffusion-xl-base-1.0",
+  controlnet=controlnet,
+  torch_dtype=torch.float16
+).to("cuda")
 
-images = [openpose_image.resize((1024, 1024)), canny_image.resize((1024, 1024))]
-
-images = pipe(
-    prompt,
-    image=images,
-    num_inference_steps=25,
-    generator=generator,
-    negative_prompt=negative_prompt,
-    num_images_per_prompt=3,
-    controlnet_conditioning_scale=[1.0, 0.8],
-).images
-make_image_grid([original_image, canny_image, openpose_image,
-                images[0].resize((512, 512)), images[1].resize((512, 512)), images[2].resize((512, 512))], rows=2, cols=3)
+canny_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png")
+pipeline(
+  "",
+  image=canny_image,
+  guess_mode=True
+).images[0]
 ```
 
-<div class="flex justify-center">
-	<img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/multicontrolnet.png"/>
-</div>
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">canny image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/guess_mode.png" width="300" alt="Generated image (Guess mode)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
+</div>

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

@@ -18,7 +18,7 @@ specific language governing permissions and limitations under the License.
 
 > [!TIP] Take a look at GitHub Issue [#841](https://github.com/huggingface/diffusers/issues/841) for more context about why we're adding community pipelines to help everyone easily share their work without being slowed down.
 
-Community pipelines are any [`DiffusionPipeline`] class that are different from the original paper implementation (for example, the [`StableDiffusionControlNetPipeline`] corresponds to the [Text-to-Image Generation with ControlNet Conditioning](https://arxiv.org/abs/2302.05543) paper). They provide additional functionality or extend the original implementation of a pipeline.
+Community pipelines are any [`DiffusionPipeline`] class that are different from the original paper implementation (for example, the [`StableDiffusionControlNetPipeline`] corresponds to the [Text-to-Image Generation with ControlNet Conditioning](https://huggingface.co/papers/2302.05543) paper). They provide additional functionality or extend the original implementation of a pipeline.
 
 There are many cool community pipelines like [Marigold Depth Estimation](https://github.com/huggingface/diffusers/tree/main/examples/community#marigold-depth-estimation) or [InstantID](https://github.com/huggingface/diffusers/tree/main/examples/community#instantid-pipeline), and you can find all the official community pipelines [here](https://github.com/huggingface/diffusers/tree/main/examples/community).
 

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

@@ -0,0 +1,35 @@
+<!--Copyright 2025 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.
+-->
+
+# DreamBooth
+
+[DreamBooth](https://huggingface.co/papers/2208.12242) is a method for generating personalized images of a specific instance. It works by fine-tuning the model on 3-5 images of the subject (for example, a cat) that is associated with a unique identifier (`sks cat`). This allows you to use `sks cat` in your prompt to trigger the model to generate images of your cat in different settings, lighting, poses, and styles.
+
+DreamBooth checkpoints are typically a few GBs in size because it contains the full model weights.
+
+Load the DreamBooth checkpoint with [`~DiffusionPipeline.from_pretrained`] and include the unique identifier in the prompt to activate its generation.
+
+```py
+import torch
+from diffusers import AutoPipelineForText2Image
+
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "sd-dreambooth-library/herge-style",
+    torch_dtype=torch.float16
+).to("cuda")
+prompt = "A cute sks herge_style brown bear eating a slice of pizza, stunning color scheme, masterpiece, illustration"
+pipeline(prompt).images[0]
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_dreambooth.png" />
+</div>

docs/source/en/using-diffusers/img2img.md

@@ -35,7 +35,7 @@ pipeline.enable_xformers_memory_efficient_attention()
 
 <Tip>
 
-You'll notice throughout the guide, we use [`~DiffusionPipeline.enable_model_cpu_offload`] and [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`], to save memory and increase inference speed. If you're using PyTorch 2.0, then you don't need to call [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`] on your pipeline because it'll already be using PyTorch 2.0's native [scaled-dot product attention](../optimization/torch2.0#scaled-dot-product-attention).
+You'll notice throughout the guide, we use [`~DiffusionPipeline.enable_model_cpu_offload`] and [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`], to save memory and increase inference speed. If you're using PyTorch 2.0, then you don't need to call [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`] on your pipeline because it'll already be using PyTorch 2.0's native [scaled-dot product attention](../optimization/fp16#scaled-dot-product-attention).
 
 </Tip>
 
@@ -589,17 +589,17 @@ make_image_grid([init_image, depth_image, image_control_net, image_elden_ring],
 
 ## Optimize
 
-Running diffusion models is computationally expensive and intensive, but with a few optimization tricks, it is entirely possible to run them on consumer and free-tier GPUs. For example, you can use a more memory-efficient form of attention such as PyTorch 2.0's [scaled-dot product attention](../optimization/torch2.0#scaled-dot-product-attention) or [xFormers](../optimization/xformers) (you can use one or the other, but there's no need to use both). You can also offload the model to the GPU while the other pipeline components wait on the CPU.
+Running diffusion models is computationally expensive and intensive, but with a few optimization tricks, it is entirely possible to run them on consumer and free-tier GPUs. For example, you can use a more memory-efficient form of attention such as PyTorch 2.0's [scaled-dot product attention](../optimization/fp16#scaled-dot-product-attention) or [xFormers](../optimization/xformers) (you can use one or the other, but there's no need to use both). You can also offload the model to the GPU while the other pipeline components wait on the CPU.
 
 ```diff
 + pipeline.enable_model_cpu_offload()
 + pipeline.enable_xformers_memory_efficient_attention()
 ```
 
-With [`torch.compile`](../optimization/torch2.0#torchcompile), you can boost your inference speed even more by wrapping your UNet with it:
+With [`torch.compile`](../optimization/fp16#torchcompile), you can boost your inference speed even more by wrapping your UNet with it:
 
 ```py
 pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
 ```
 
-To learn more, take a look at the [Reduce memory usage](../optimization/memory) and [Torch 2.0](../optimization/torch2.0) guides.
+To learn more, take a look at the [Reduce memory usage](../optimization/memory) and [Accelerate inference](../optimization/fp16) guides.

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

@@ -485,7 +485,7 @@ image = image[:, :, None]
 image = np.concatenate([image, image, image], axis=2)
 canny_image = Image.fromarray(image).resize((1024, 1216))
 
-adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, varient="fp16").to("cuda")
+adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, variant="fp16").to("cuda")
 
 unet = UNet2DConditionModel.from_pretrained(
     "latent-consistency/lcm-sdxl",
@@ -551,7 +551,7 @@ image = image[:, :, None]
 image = np.concatenate([image, image, image], axis=2)
 canny_image = Image.fromarray(image).resize((1024, 1024))
 
-adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, varient="fp16").to("cuda")
+adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, variant="fp16").to("cuda")
 
 pipe = StableDiffusionXLAdapterPipeline.from_pretrained(
     "stabilityai/stable-diffusion-xl-base-1.0",

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

@@ -25,7 +25,7 @@ The major advantages of TCD are:
 - Freely change detail level: During inference, the level of detail in the image can be adjusted with a single hyperparameter, *gamma*.
 
 > [!TIP]
-> For more technical details of TCD, please refer to the [paper](https://arxiv.org/abs/2402.19159) or official [project page](https://mhh0318.github.io/tcd/)).
+> For more technical details of TCD, please refer to the [paper](https://huggingface.co/papers/2402.19159) or official [project page](https://mhh0318.github.io/tcd/).
 
 For large models like SDXL, TCD is trained with [LoRA](https://huggingface.co/docs/peft/conceptual_guides/adapter#low-rank-adaptation-lora) to reduce memory usage. This is also useful because you can reuse LoRAs between different finetuned models, as long as they share the same base model, without further training.
 

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

@@ -35,7 +35,7 @@ pipeline.enable_xformers_memory_efficient_attention()
 
 <Tip>
 
-You'll notice throughout the guide, we use [`~DiffusionPipeline.enable_model_cpu_offload`] and [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`], to save memory and increase inference speed. If you're using PyTorch 2.0, it's not necessary to call [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`] on your pipeline because it'll already be using PyTorch 2.0's native [scaled-dot product attention](../optimization/torch2.0#scaled-dot-product-attention).
+You'll notice throughout the guide, we use [`~DiffusionPipeline.enable_model_cpu_offload`] and [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`], to save memory and increase inference speed. If you're using PyTorch 2.0, it's not necessary to call [`~DiffusionPipeline.enable_xformers_memory_efficient_attention`] on your pipeline because it'll already be using PyTorch 2.0's native [scaled-dot product attention](../optimization/fp16#scaled-dot-product-attention).
 
 </Tip>
 
@@ -363,6 +363,7 @@ device = "cuda"
 pipeline = AutoPipelineForInpainting.from_pretrained(
     "runwayml/stable-diffusion-inpainting",
     torch_dtype=torch.float16,
+    variant="fp16"
 )
 pipeline = pipeline.to(device)
 
@@ -787,7 +788,7 @@ make_image_grid([init_image, mask_image, image, image_elden_ring], rows=2, cols=
 
 ## Optimize
 
-It can be difficult and slow to run diffusion models if you're resource constrained, but it doesn't have to be with a few optimization tricks. One of the biggest (and easiest) optimizations you can enable is switching to memory-efficient attention. If you're using PyTorch 2.0, [scaled-dot product attention](../optimization/torch2.0#scaled-dot-product-attention) is automatically enabled and you don't need to do anything else. For non-PyTorch 2.0 users, you can install and use [xFormers](../optimization/xformers)'s implementation of memory-efficient attention. Both options reduce memory usage and accelerate inference.
+It can be difficult and slow to run diffusion models if you're resource constrained, but it doesn't have to be with a few optimization tricks. One of the biggest (and easiest) optimizations you can enable is switching to memory-efficient attention. If you're using PyTorch 2.0, [scaled-dot product attention](../optimization/fp16#scaled-dot-product-attention) is automatically enabled and you don't need to do anything else. For non-PyTorch 2.0 users, you can install and use [xFormers](../optimization/xformers)'s implementation of memory-efficient attention. Both options reduce memory usage and accelerate inference.
 
 You can also offload the model to the CPU to save even more memory:
 
@@ -796,10 +797,10 @@ You can also offload the model to the CPU to save even more memory:
 + pipeline.enable_model_cpu_offload()
 ```
 
-To speed-up your inference code even more, use [`torch_compile`](../optimization/torch2.0#torchcompile). You should wrap `torch.compile` around the most intensive component in the pipeline which is typically the UNet:
+To speed-up your inference code even more, use [`torch_compile`](../optimization/fp16#torchcompile). You should wrap `torch.compile` around the most intensive component in the pipeline which is typically the UNet:
 
 ```py
 pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
 ```
 
-Learn more in the [Reduce memory usage](../optimization/memory) and [Torch 2.0](../optimization/torch2.0) guides.
+Learn more in the [Reduce memory usage](../optimization/memory) and [Accelerate inference](../optimization/fp16) guides.

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

@@ -1,416 +0,0 @@
-<!--Copyright 2024 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.
--->
-
-# Load adapters
-
-[[open-in-colab]]
-
-There are several [training](../training/overview) techniques for personalizing diffusion models to generate images of a specific subject or images in certain styles. Each of these training methods produces a different type of adapter. Some of the adapters generate an entirely new model, while other adapters only modify a smaller set of embeddings or weights. This means the loading process for each adapter is also different.
-
-This guide will show you how to load DreamBooth, textual inversion, and LoRA weights.
-
-<Tip>
-
-Feel free to browse the [Stable Diffusion Conceptualizer](https://huggingface.co/spaces/sd-concepts-library/stable-diffusion-conceptualizer), [LoRA the Explorer](https://huggingface.co/spaces/multimodalart/LoraTheExplorer), and the [Diffusers Models Gallery](https://huggingface.co/spaces/huggingface-projects/diffusers-gallery) for checkpoints and embeddings to use.
-
-</Tip>
-
-## DreamBooth
-
-[DreamBooth](https://dreambooth.github.io/) finetunes an *entire diffusion model* on just several images of a subject to generate images of that subject in new styles and settings. This method works by using a special word in the prompt that the model learns to associate with the subject image. Of all the training methods, DreamBooth produces the largest file size (usually a few GBs) because it is a full checkpoint model.
-
-Let's load the [herge_style](https://huggingface.co/sd-dreambooth-library/herge-style) checkpoint, which is trained on just 10 images drawn by Hergé, to generate images in that style. For it to work, you need to include the special word `herge_style` in your prompt to trigger the checkpoint:
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-
-pipeline = AutoPipelineForText2Image.from_pretrained("sd-dreambooth-library/herge-style", torch_dtype=torch.float16).to("cuda")
-prompt = "A cute herge_style brown bear eating a slice of pizza, stunning color scheme, masterpiece, illustration"
-image = pipeline(prompt).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_dreambooth.png" />
-</div>
-
-## Textual inversion
-
-[Textual inversion](https://textual-inversion.github.io/) is very similar to DreamBooth and it can also personalize a diffusion model to generate certain concepts (styles, objects) from just a few images. This method works by training and finding new embeddings that represent the images you provide with a special word in the prompt. As a result, the diffusion model weights stay the same and the training process produces a relatively tiny (a few KBs) file.
-
-Because textual inversion creates embeddings, it cannot be used on its own like DreamBooth and requires another model.
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-
-pipeline = AutoPipelineForText2Image.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda")
-```
-
-Now you can load the textual inversion embeddings with the [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] method and generate some images. Let's load the [sd-concepts-library/gta5-artwork](https://huggingface.co/sd-concepts-library/gta5-artwork) embeddings and you'll need to include the special word `<gta5-artwork>` in your prompt to trigger it:
-
-```py
-pipeline.load_textual_inversion("sd-concepts-library/gta5-artwork")
-prompt = "A cute brown bear eating a slice of pizza, stunning color scheme, masterpiece, illustration, <gta5-artwork> style"
-image = pipeline(prompt).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_txt_embed.png" />
-</div>
-
-Textual inversion can also be trained on undesirable things to create *negative embeddings* to discourage a model from generating images with those undesirable things like blurry images or extra fingers on a hand. This can be an easy way to quickly improve your prompt. You'll also load the embeddings with [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`], but this time, you'll need two more parameters:
-
-- `weight_name`: specifies the weight file to load if the file was saved in the 🤗 Diffusers format with a specific name or if the file is stored in the A1111 format
-- `token`: specifies the special word to use in the prompt to trigger the embeddings
-
-Let's load the [sayakpaul/EasyNegative-test](https://huggingface.co/sayakpaul/EasyNegative-test) embeddings:
-
-```py
-pipeline.load_textual_inversion(
-    "sayakpaul/EasyNegative-test", weight_name="EasyNegative.safetensors", token="EasyNegative"
-)
-```
-
-Now you can use the `token` to generate an image with the negative embeddings:
-
-```py
-prompt = "A cute brown bear eating a slice of pizza, stunning color scheme, masterpiece, illustration, EasyNegative"
-negative_prompt = "EasyNegative"
-
-image = pipeline(prompt, negative_prompt=negative_prompt, num_inference_steps=50).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_neg_embed.png" />
-</div>
-
-## LoRA
-
-[Low-Rank Adaptation (LoRA)](https://huggingface.co/papers/2106.09685) is a popular training technique because it is fast and generates smaller file sizes (a couple hundred MBs). Like the other methods in this guide, LoRA can train a model to learn new styles from just a few images. It works by inserting new weights into the diffusion model and then only the new weights are trained instead of the entire model. This makes LoRAs faster to train and easier to store.
-
-<Tip>
-
-LoRA is a very general training technique that can be used with other training methods. For example, it is common to train a model with DreamBooth and LoRA. It is also increasingly common to load and merge multiple LoRAs to create new and unique images. You can learn more about it in the in-depth [Merge LoRAs](merge_loras) guide since merging is outside the scope of this loading guide.
-
-</Tip>
-
-LoRAs also need to be used with another model:
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-
-pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-```
-
-Then use the [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] method to load the [ostris/super-cereal-sdxl-lora](https://huggingface.co/ostris/super-cereal-sdxl-lora) weights and specify the weights filename from the repository:
-
-```py
-pipeline.load_lora_weights("ostris/super-cereal-sdxl-lora", weight_name="cereal_box_sdxl_v1.safetensors")
-prompt = "bears, pizza bites"
-image = pipeline(prompt).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_lora.png" />
-</div>
-
-The [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] method loads LoRA weights into both the UNet and text encoder. It is the preferred way for loading LoRAs because it can handle cases where:
-
-- the LoRA weights don't have separate identifiers for the UNet and text encoder
-- the LoRA weights have separate identifiers for the UNet and text encoder
-
-To directly load (and save) a LoRA adapter at the *model-level*, use [`~loaders.PeftAdapterMixin.load_lora_adapter`], which builds and prepares the necessary model configuration for the adapter. Like [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`], [`~loaders.PeftAdapterMixin.load_lora_adapter`] can load LoRAs for both the UNet and text encoder. For example, if you're loading a LoRA for the UNet, [`~loaders.PeftAdapterMixin.load_lora_adapter`] ignores the keys for the text encoder.
-
-Use the `weight_name` parameter to specify the specific weight file and the `prefix` parameter to filter for the appropriate state dicts (`"unet"` in this case) to load.
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-
-pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-pipeline.unet.load_lora_adapter("jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", prefix="unet")
-
-# use cnmt in the prompt to trigger the LoRA
-prompt = "A cute cnmt eating a slice of pizza, stunning color scheme, masterpiece, illustration"
-image = pipeline(prompt).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_attn_proc.png" />
-</div>
-
-Save an adapter with [`~loaders.PeftAdapterMixin.save_lora_adapter`].
-
-To unload the LoRA weights, use the [`~loaders.StableDiffusionLoraLoaderMixin.unload_lora_weights`] method to discard the LoRA weights and restore the model to its original weights:
-
-```py
-pipeline.unload_lora_weights()
-```
-
-### Adjust LoRA weight scale
-
-For both [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] and [`~loaders.UNet2DConditionLoadersMixin.load_attn_procs`], you can pass the `cross_attention_kwargs={"scale": 0.5}` parameter to adjust how much of the LoRA weights to use. A value of `0` is the same as only using the base model weights, and a value of `1` is equivalent to using the fully finetuned LoRA.
-
-For more granular control on the amount of LoRA weights used per layer, you can use [`~loaders.StableDiffusionLoraLoaderMixin.set_adapters`] and pass a dictionary specifying by how much to scale the weights in each layer by.
-```python
-pipe = ... # create pipeline
-pipe.load_lora_weights(..., adapter_name="my_adapter")
-scales = {
-    "text_encoder": 0.5,
-    "text_encoder_2": 0.5,  # only usable if pipe has a 2nd text encoder
-    "unet": {
-        "down": 0.9,  # all transformers in the down-part will use scale 0.9
-        # "mid"  # in this example "mid" is not given, therefore all transformers in the mid part will use the default scale 1.0
-        "up": {
-            "block_0": 0.6,  # all 3 transformers in the 0th block in the up-part will use scale 0.6
-            "block_1": [0.4, 0.8, 1.0],  # the 3 transformers in the 1st block in the up-part will use scales 0.4, 0.8 and 1.0 respectively
-        }
-    }
-}
-pipe.set_adapters("my_adapter", scales)
-```
-
-This also works with multiple adapters - see [this guide](https://huggingface.co/docs/diffusers/tutorials/using_peft_for_inference#customize-adapters-strength) for how to do it.
-
-<Tip warning={true}>
-
-Currently, [`~loaders.StableDiffusionLoraLoaderMixin.set_adapters`] only supports scaling attention weights. If a LoRA has other parts (e.g., resnets or down-/upsamplers), they will keep a scale of 1.0.
-
-</Tip>
-
-### Hotswapping LoRA adapters
-
-A common use case when serving multiple adapters is to load one adapter first, generate images, load another adapter, generate more images, load another adapter, etc. This workflow normally requires calling [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`], [`~loaders.StableDiffusionLoraLoaderMixin.set_adapters`], and possibly [`~loaders.peft.PeftAdapterMixin.delete_adapters`] to save memory. Moreover, if the model is compiled using `torch.compile`, performing these steps requires recompilation, which takes time.
-
-To better support this common workflow, you can "hotswap" a LoRA adapter, to avoid accumulating memory and in some cases, recompilation. It requires an adapter to already be loaded, and the new adapter weights are swapped in-place for the existing adapter.
-
-Pass `hotswap=True` when loading a LoRA adapter to enable this feature. It is important to indicate the name of the existing adapter, (`default_0` is the default adapter name), to be swapped. If you loaded the first adapter with a different name, use that name instead.
-
-```python
-pipe = ...
-# load adapter 1 as normal
-pipeline.load_lora_weights(file_name_adapter_1)
-# generate some images with adapter 1
-...
-# now hot swap the 2nd adapter
-pipeline.load_lora_weights(file_name_adapter_2, hotswap=True, adapter_name="default_0")
-# generate images with adapter 2
-```
-
-
-<Tip warning={true}>
-
-Hotswapping is not currently supported for LoRA adapters that target the text encoder.
-
-</Tip>
-
-For compiled models, it is often (though not always if the second adapter targets identical LoRA ranks and scales) necessary to call [`~loaders.lora_base.LoraBaseMixin.enable_lora_hotswap`] to avoid recompilation. Use [`~loaders.lora_base.LoraBaseMixin.enable_lora_hotswap`] _before_ loading the first adapter, and `torch.compile` should be called _after_ loading the first adapter.
-
-```python
-pipe = ...
-# call this extra method
-pipe.enable_lora_hotswap(target_rank=max_rank)
-# now load adapter 1
-pipe.load_lora_weights(file_name_adapter_1)
-# now compile the unet of the pipeline
-pipe.unet = torch.compile(pipeline.unet, ...)
-# generate some images with adapter 1
-...
-# now hot swap adapter 2
-pipeline.load_lora_weights(file_name_adapter_2, hotswap=True, adapter_name="default_0")
-# generate images with adapter 2
-```
-
-The `target_rank=max_rank` argument is important for setting the maximum rank among all LoRA adapters that will be loaded. If you have one adapter with rank 8 and another with rank 16, pass `target_rank=16`. You should use a higher value if in doubt. By default, this value is 128.
-
-However, there can be situations where recompilation is unavoidable. For example, if the hotswapped adapter targets more layers than the initial adapter, then recompilation is triggered. Try to load the adapter that targets the most layers first. Refer to the PEFT docs on [hotswapping](https://huggingface.co/docs/peft/main/en/package_reference/hotswap#peft.utils.hotswap.hotswap_adapter) for more details about the limitations of this feature.
-
-<Tip>
-
-Move your code inside the `with torch._dynamo.config.patch(error_on_recompile=True)` context manager to detect if a model was recompiled. If you detect recompilation despite following all the steps above, please open an issue with [Diffusers](https://github.com/huggingface/diffusers/issues) with a reproducible example.
-
-</Tip>
-
-### Kohya and TheLastBen
-
-Other popular LoRA trainers from the community include those by [Kohya](https://github.com/kohya-ss/sd-scripts/) and [TheLastBen](https://github.com/TheLastBen/fast-stable-diffusion). These trainers create different LoRA checkpoints than those trained by 🤗 Diffusers, but they can still be loaded in the same way.
-
-<hfoptions id="other-trainers">
-<hfoption id="Kohya">
-
-To load a Kohya LoRA, let's download the [Blueprintify SD XL 1.0](https://civitai.com/models/150986/blueprintify-sd-xl-10) checkpoint from [Civitai](https://civitai.com/) as an example:
-
-```sh
-!wget https://civitai.com/api/download/models/168776 -O blueprintify-sd-xl-10.safetensors
-```
-
-Load the LoRA checkpoint with the [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] method, and specify the filename in the `weight_name` parameter:
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-
-pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-pipeline.load_lora_weights("path/to/weights", weight_name="blueprintify-sd-xl-10.safetensors")
-```
-
-Generate an image:
-
-```py
-# use bl3uprint in the prompt to trigger the LoRA
-prompt = "bl3uprint, a highly detailed blueprint of the eiffel tower, explaining how to build all parts, many txt, blueprint grid backdrop"
-image = pipeline(prompt).images[0]
-image
-```
-
-<Tip warning={true}>
-
-Some limitations of using Kohya LoRAs with 🤗 Diffusers include:
-
-- Images may not look like those generated by UIs - like ComfyUI - for multiple reasons, which are explained [here](https://github.com/huggingface/diffusers/pull/4287/#issuecomment-1655110736).
-- [LyCORIS checkpoints](https://github.com/KohakuBlueleaf/LyCORIS) aren't fully supported. The [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] method loads LyCORIS checkpoints with LoRA and LoCon modules, but Hada and LoKR are not supported.
-
-</Tip>
-
-</hfoption>
-<hfoption id="TheLastBen">
-
-Loading a checkpoint from TheLastBen is very similar. For example, to load the [TheLastBen/William_Eggleston_Style_SDXL](https://huggingface.co/TheLastBen/William_Eggleston_Style_SDXL) checkpoint:
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-
-pipeline = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-pipeline.load_lora_weights("TheLastBen/William_Eggleston_Style_SDXL", weight_name="wegg.safetensors")
-
-# use by william eggleston in the prompt to trigger the LoRA
-prompt = "a house by william eggleston, sunrays, beautiful, sunlight, sunrays, beautiful"
-image = pipeline(prompt=prompt).images[0]
-image
-```
-
-</hfoption>
-</hfoptions>
-
-## IP-Adapter
-
-[IP-Adapter](https://ip-adapter.github.io/) is a lightweight adapter that enables image prompting for any diffusion model. This adapter works by decoupling the cross-attention layers of the image and text features. All the other model components are frozen and only the embedded image features in the UNet are trained. As a result, IP-Adapter files are typically only ~100MBs.
-
-You can learn more about how to use IP-Adapter for different tasks and specific use cases in the [IP-Adapter](../using-diffusers/ip_adapter) guide.
-
-> [!TIP]
-> Diffusers currently only supports IP-Adapter for some of the most popular pipelines. Feel free to open a feature request if you have a cool use case and want to integrate IP-Adapter with an unsupported pipeline!
-> Official IP-Adapter checkpoints are available from [h94/IP-Adapter](https://huggingface.co/h94/IP-Adapter).
-
-To start, load a Stable Diffusion checkpoint.
-
-```py
-from diffusers import AutoPipelineForText2Image
-import torch
-from diffusers.utils import load_image
-
-pipeline = AutoPipelineForText2Image.from_pretrained("stable-diffusion-v1-5/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda")
-```
-
-Then load the IP-Adapter weights and add it to the pipeline with the [`~loaders.IPAdapterMixin.load_ip_adapter`] method.
-
-```py
-pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin")
-```
-
-Once loaded, you can use the pipeline with an image and text prompt to guide the image generation process.
-
-```py
-image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_neg_embed.png")
-generator = torch.Generator(device="cpu").manual_seed(33)
-images = pipeline(
-    prompt='best quality, high quality, wearing sunglasses',
-    ip_adapter_image=image,
-    negative_prompt="monochrome, lowres, bad anatomy, worst quality, low quality",
-    num_inference_steps=50,
-    generator=generator,
-).images[0]
-images
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ip-bear.png" />
-</div>
-
-### IP-Adapter Plus
-
-IP-Adapter relies on an image encoder to generate image features. If the IP-Adapter repository contains an `image_encoder` subfolder, the image encoder is automatically loaded and registered to the pipeline. Otherwise, you'll need to explicitly load the image encoder with a [`~transformers.CLIPVisionModelWithProjection`] model and pass it to the pipeline.
-
-This is the case for *IP-Adapter Plus* checkpoints which use the ViT-H image encoder.
-
-```py
-from transformers import CLIPVisionModelWithProjection
-
-image_encoder = CLIPVisionModelWithProjection.from_pretrained(
-    "h94/IP-Adapter",
-    subfolder="models/image_encoder",
-    torch_dtype=torch.float16
-)
-
-pipeline = AutoPipelineForText2Image.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    image_encoder=image_encoder,
-    torch_dtype=torch.float16
-).to("cuda")
-
-pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="sdxl_models", weight_name="ip-adapter-plus_sdxl_vit-h.safetensors")
-```
-
-### IP-Adapter Face ID models
-
-The IP-Adapter FaceID models are experimental IP Adapters that use image embeddings generated by `insightface` instead of CLIP image embeddings. Some of these models also use LoRA to improve ID consistency.
-You need to install `insightface` and all its requirements to use these models.
-
-<Tip warning={true}>
-As InsightFace pretrained models are available for non-commercial research purposes, IP-Adapter-FaceID models are released exclusively for research purposes and are not intended for commercial use.
-</Tip>
-
-```py
-pipeline = AutoPipelineForText2Image.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    torch_dtype=torch.float16
-).to("cuda")
-
-pipeline.load_ip_adapter("h94/IP-Adapter-FaceID", subfolder=None, weight_name="ip-adapter-faceid_sdxl.bin", image_encoder_folder=None)
-```
-
-If you want to use one of the two IP-Adapter FaceID Plus models, you must also load the CLIP image encoder, as this models use both `insightface` and CLIP image embeddings to achieve better photorealism.
-
-```py
-from transformers import CLIPVisionModelWithProjection
-
-image_encoder = CLIPVisionModelWithProjection.from_pretrained(
-    "laion/CLIP-ViT-H-14-laion2B-s32B-b79K",
-    torch_dtype=torch.float16,
-)
-
-pipeline = AutoPipelineForText2Image.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    image_encoder=image_encoder,
-    torch_dtype=torch.float16
-).to("cuda")
-
-pipeline.load_ip_adapter("h94/IP-Adapter-FaceID", subfolder=None, weight_name="ip-adapter-faceid-plus_sd15.bin")
-```

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

@@ -288,7 +288,7 @@ Speeding them up can be achieved by using a more efficient attention processor:
   depth = pipe(image, num_inference_steps=1)
 ```
 
-Finally, as suggested in [Optimizations](../optimization/torch2.0#torch.compile), enabling `torch.compile` can further enhance performance depending on 
+Finally, as suggested in [Optimizations](../optimization/fp16#torchcompile), enabling `torch.compile` can further enhance performance depending on 
 the target hardware.
 However, compilation incurs a significant overhead during the first pipeline invocation, making it beneficial only when 
 the same pipeline instance is called repeatedly, such as within a loop.

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

@@ -1,266 +0,0 @@
-<!--Copyright 2024 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.
--->
-
-# Merge LoRAs
-
-It can be fun and creative to use multiple [LoRAs]((https://huggingface.co/docs/peft/conceptual_guides/adapter#low-rank-adaptation-lora)) together to generate something entirely new and unique. This works by merging multiple LoRA weights together to produce images that are a blend of different styles. Diffusers provides a few methods to merge LoRAs depending on *how* you want to merge their weights, which can affect image quality.
-
-This guide will show you how to merge LoRAs using the [`~loaders.PeftAdapterMixin.set_adapters`] and [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) methods. To improve inference speed and reduce memory-usage of merged LoRAs, you'll also see how to use the [`~loaders.StableDiffusionLoraLoaderMixin.fuse_lora`] method to fuse the LoRA weights with the original weights of the underlying model.
-
-For this guide, load a Stable Diffusion XL (SDXL) checkpoint and the [KappaNeuro/studio-ghibli-style](https://huggingface.co/KappaNeuro/studio-ghibli-style) and [Norod78/sdxl-chalkboarddrawing-lora](https://huggingface.co/Norod78/sdxl-chalkboarddrawing-lora) LoRAs with the [`~loaders.StableDiffusionLoraLoaderMixin.load_lora_weights`] method. You'll need to assign each LoRA an `adapter_name` to combine them later.
-
-```py
-from diffusers import DiffusionPipeline
-import torch
-
-pipeline = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-pipeline.load_lora_weights("ostris/ikea-instructions-lora-sdxl", weight_name="ikea_instructions_xl_v1_5.safetensors", adapter_name="ikea")
-pipeline.load_lora_weights("lordjia/by-feng-zikai", weight_name="fengzikai_v1.0_XL.safetensors", adapter_name="feng")
-```
-
-## set_adapters
-
-The [`~loaders.PeftAdapterMixin.set_adapters`] method merges LoRA adapters by concatenating their weighted matrices. Use the adapter name to specify which LoRAs to merge, and the `adapter_weights` parameter to control the scaling for each LoRA. For example, if `adapter_weights=[0.5, 0.5]`, then the merged LoRA output is an average of both LoRAs. Try adjusting the adapter weights to see how it affects the generated image!
-
-```py
-pipeline.set_adapters(["ikea", "feng"], adapter_weights=[0.7, 0.8])
-
-generator = torch.manual_seed(0)
-prompt = "A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai"
-image = pipeline(prompt, generator=generator, cross_attention_kwargs={"scale": 1.0}).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lora_merge_set_adapters.png"/>
-</div>
-
-## add_weighted_adapter
-
-> [!WARNING]
-> This is an experimental method that adds PEFTs [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) method to Diffusers to enable more efficient merging methods. Check out this [issue](https://github.com/huggingface/diffusers/issues/6892) if you're interested in learning more about the motivation and design behind this integration.
-
-The [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) method provides access to more efficient merging method such as [TIES and DARE](https://huggingface.co/docs/peft/developer_guides/model_merging). To use these merging methods, make sure you have the latest stable version of Diffusers and PEFT installed.
-
-```bash
-pip install -U diffusers peft
-```
-
-There are three steps to merge LoRAs with the [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) method:
-
-1. Create a [PeftModel](https://huggingface.co/docs/peft/package_reference/peft_model#peft.PeftModel) from the underlying model and LoRA checkpoint.
-2. Load a base UNet model and the LoRA adapters.
-3. Merge the adapters using the [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) method and the merging method of your choice.
-
-Let's dive deeper into what these steps entail.
-
-1. Load a UNet that corresponds to the UNet in the LoRA checkpoint. In this case, both LoRAs use the SDXL UNet as their base model.
-
-```python
-from diffusers import AutoModel
-import torch
-
-unet = AutoModel.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
-    variant="fp16",
-    subfolder="unet",
-).to("cuda")
-```
-
-Load the SDXL pipeline and the LoRA checkpoints, starting with the [ostris/ikea-instructions-lora-sdxl](https://huggingface.co/ostris/ikea-instructions-lora-sdxl) LoRA.
-
-```python
-from diffusers import DiffusionPipeline
-
-pipeline = DiffusionPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    variant="fp16",
-    torch_dtype=torch.float16,
-    unet=unet
-).to("cuda")
-pipeline.load_lora_weights("ostris/ikea-instructions-lora-sdxl", weight_name="ikea_instructions_xl_v1_5.safetensors", adapter_name="ikea")
-```
-
-Now you'll create a [PeftModel](https://huggingface.co/docs/peft/package_reference/peft_model#peft.PeftModel) from the loaded LoRA checkpoint by combining the SDXL UNet and the LoRA UNet from the pipeline.
-
-```python
-from peft import get_peft_model, LoraConfig
-import copy
-
-sdxl_unet = copy.deepcopy(unet)
-ikea_peft_model = get_peft_model(
-    sdxl_unet,
-    pipeline.unet.peft_config["ikea"],
-    adapter_name="ikea"
-)
-
-original_state_dict = {f"base_model.model.{k}": v for k, v in pipeline.unet.state_dict().items()}
-ikea_peft_model.load_state_dict(original_state_dict, strict=True)
-```
-
-> [!TIP]
-> You can optionally push the ikea_peft_model to the Hub by calling `ikea_peft_model.push_to_hub("ikea_peft_model", token=TOKEN)`.
-
-Repeat this process to create a [PeftModel](https://huggingface.co/docs/peft/package_reference/peft_model#peft.PeftModel) from the [lordjia/by-feng-zikai](https://huggingface.co/lordjia/by-feng-zikai) LoRA.
-
-```python
-pipeline.delete_adapters("ikea")
-sdxl_unet.delete_adapters("ikea")
-
-pipeline.load_lora_weights("lordjia/by-feng-zikai", weight_name="fengzikai_v1.0_XL.safetensors", adapter_name="feng")
-pipeline.set_adapters(adapter_names="feng")
-
-feng_peft_model = get_peft_model(
-    sdxl_unet,
-    pipeline.unet.peft_config["feng"],
-    adapter_name="feng"
-)
-
-original_state_dict = {f"base_model.model.{k}": v for k, v in pipe.unet.state_dict().items()}
-feng_peft_model.load_state_dict(original_state_dict, strict=True)
-```
-
-2. Load a base UNet model and then load the adapters onto it.
-
-```python
-from peft import PeftModel
-
-base_unet = AutoModel.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    torch_dtype=torch.float16,
-    use_safetensors=True,
-    variant="fp16",
-    subfolder="unet",
-).to("cuda")
-
-model = PeftModel.from_pretrained(base_unet, "stevhliu/ikea_peft_model", use_safetensors=True, subfolder="ikea", adapter_name="ikea")
-model.load_adapter("stevhliu/feng_peft_model", use_safetensors=True, subfolder="feng", adapter_name="feng")
-```
-
-3. Merge the adapters using the [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) method and the merging method of your choice (learn more about other merging methods in this [blog post](https://huggingface.co/blog/peft_merging)). For this example, let's use the `"dare_linear"` method to merge the LoRAs.
-
-> [!WARNING]
-> Keep in mind the LoRAs need to have the same rank to be merged!
-
-```python
-model.add_weighted_adapter(
-    adapters=["ikea", "feng"],
-    weights=[1.0, 1.0],
-    combination_type="dare_linear",
-    adapter_name="ikea-feng"
-)
-model.set_adapters("ikea-feng")
-```
-
-Now you can generate an image with the merged LoRA.
-
-```python
-model = model.to(dtype=torch.float16, device="cuda")
-
-pipeline = DiffusionPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", unet=model, variant="fp16", torch_dtype=torch.float16,
-).to("cuda")
-
-image = pipeline("A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai", generator=torch.manual_seed(0)).images[0]
-image
-```
-
-<div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/ikea-feng-dare-linear.png"/>
-</div>
-
-## fuse_lora
-
-Both the [`~loaders.PeftAdapterMixin.set_adapters`] and [add_weighted_adapter](https://huggingface.co/docs/peft/package_reference/lora#peft.LoraModel.add_weighted_adapter) methods require loading the base model and the LoRA adapters separately which incurs some overhead. The [`~loaders.lora_base.LoraBaseMixin.fuse_lora`] method allows you to fuse the LoRA weights directly with the original weights of the underlying model. This way, you're only loading the model once which can increase inference and lower memory-usage.
-
-You can use PEFT to easily fuse/unfuse multiple adapters directly into the model weights (both UNet and text encoder) using the [`~loaders.lora_base.LoraBaseMixin.fuse_lora`] method, which can lead to a speed-up in inference and lower VRAM usage.
-
-For example, if you have a base model and adapters loaded and set as active with the following adapter weights:
-
-```py
-from diffusers import DiffusionPipeline
-import torch
-
-pipeline = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-pipeline.load_lora_weights("ostris/ikea-instructions-lora-sdxl", weight_name="ikea_instructions_xl_v1_5.safetensors", adapter_name="ikea")
-pipeline.load_lora_weights("lordjia/by-feng-zikai", weight_name="fengzikai_v1.0_XL.safetensors", adapter_name="feng")
-
-pipeline.set_adapters(["ikea", "feng"], adapter_weights=[0.7, 0.8])
-```
-
-Fuse these LoRAs into the UNet with the [`~loaders.lora_base.LoraBaseMixin.fuse_lora`] method. The `lora_scale` parameter controls how much to scale the output by with the LoRA weights. It is important to make the `lora_scale` adjustments in the [`~loaders.lora_base.LoraBaseMixin.fuse_lora`] method because it won’t work if you try to pass `scale` to the `cross_attention_kwargs` in the pipeline.
-
-```py
-pipeline.fuse_lora(adapter_names=["ikea", "feng"], lora_scale=1.0)
-```
-
-Then you should use [`~loaders.StableDiffusionLoraLoaderMixin.unload_lora_weights`] to unload the LoRA weights since they've already been fused with the underlying base model. Finally, call [`~DiffusionPipeline.save_pretrained`] to save the fused pipeline locally or you could call [`~DiffusionPipeline.push_to_hub`] to push the fused pipeline to the Hub.
-
-```py
-pipeline.unload_lora_weights()
-# save locally
-pipeline.save_pretrained("path/to/fused-pipeline")
-# save to the Hub
-pipeline.push_to_hub("fused-ikea-feng")
-```
-
-Now you can quickly load the fused pipeline and use it for inference without needing to separately load the LoRA adapters.
-
-```py
-pipeline = DiffusionPipeline.from_pretrained(
-    "username/fused-ikea-feng", torch_dtype=torch.float16,
-).to("cuda")
-
-image = pipeline("A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai", generator=torch.manual_seed(0)).images[0]
-image
-```
-
-You can call [`~~loaders.lora_base.LoraBaseMixin.unfuse_lora`] to restore the original model's weights (for example, if you want to use a different `lora_scale` value). However, this only works if you've only fused one LoRA adapter to the original model. If you've fused multiple LoRAs, you'll need to reload the model.
-
-```py
-pipeline.unfuse_lora()
-```
-
-### torch.compile
-
-[torch.compile](../optimization/torch2.0#torchcompile) can speed up your pipeline even more, but the LoRA weights must be fused first and then unloaded. Typically, the UNet is compiled because it is such a computationally intensive component of the pipeline.
-
-```py
-from diffusers import DiffusionPipeline
-import torch
-
-# load base model and LoRAs
-pipeline = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16).to("cuda")
-pipeline.load_lora_weights("ostris/ikea-instructions-lora-sdxl", weight_name="ikea_instructions_xl_v1_5.safetensors", adapter_name="ikea")
-pipeline.load_lora_weights("lordjia/by-feng-zikai", weight_name="fengzikai_v1.0_XL.safetensors", adapter_name="feng")
-
-# activate both LoRAs and set adapter weights
-pipeline.set_adapters(["ikea", "feng"], adapter_weights=[0.7, 0.8])
-
-# fuse LoRAs and unload weights
-pipeline.fuse_lora(adapter_names=["ikea", "feng"], lora_scale=1.0)
-pipeline.unload_lora_weights()
-
-# torch.compile
-pipeline.unet.to(memory_format=torch.channels_last)
-pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
-
-image = pipeline("A bowl of ramen shaped like a cute kawaii bear, by Feng Zikai", generator=torch.manual_seed(0)).images[0]
-```
-
-Learn more about torch.compile in the [Accelerate inference of text-to-image diffusion models](../tutorials/fast_diffusion#torchcompile) guide.
-
-## Next steps
-
-For more conceptual details about how each merging method works, take a look at the [🤗 PEFT welcomes new merging methods](https://huggingface.co/blog/peft_merging#concatenation-cat) blog post!

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

@@ -15,7 +15,7 @@ OmniGen is an image generation model. Unlike existing text-to-image models, Omni
 - Minimalist model architecture, consisting of only a VAE and a transformer module, for joint modeling of text and images.
 - Support for multimodal inputs. It can process any text-image mixed data as instructions for image generation, rather than relying solely on text.
 
-For more information, please refer to the [paper](https://arxiv.org/pdf/2409.11340).
+For more information, please refer to the [paper](https://huggingface.co/papers/2409.11340).
 This guide will walk you through using OmniGen for various tasks and use cases.
 
 ## Load model checkpoints

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

@@ -154,11 +154,11 @@ pipeline = AutoPipelineForInpainting.from_pretrained(
 pipeline.enable_model_cpu_offload()
 ```
 
-You can enable PAG on an exisiting inpainting pipeline like this
+You can enable PAG on an existing inpainting pipeline like this
 
 ```py
-pipeline_inpaint = AutoPipelineForInpaiting.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16)
-pipeline = AutoPipelineForInpaiting.from_pipe(pipeline_inpaint, enable_pag=True)
+pipeline_inpaint = AutoPipelineForInpainting.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16)
+pipeline = AutoPipelineForInpainting.from_pipe(pipeline_inpaint, enable_pag=True)
 ```
 
 This still works when your pipeline has a different task:

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

@@ -63,7 +63,7 @@ export_to_video(frames, "generated.mp4", fps=7)
 
 ## torch.compile
 
-You can gain a 20-25% speedup at the expense of slightly increased memory by [compiling](../optimization/torch2.0#torchcompile) the UNet.
+You can gain a 20-25% speedup at the expense of slightly increased memory by [compiling](../optimization/fp16#torchcompile) the UNet.
 
 ```diff
 - pipe.enable_model_cpu_offload()

docs/source/en/using-diffusers/t2i_adapter.md

@@ -12,41 +12,21 @@ specific language governing permissions and limitations under the License.
 
 # T2I-Adapter
 
-[T2I-Adapter](https://hf.co/papers/2302.08453) is a lightweight adapter for controlling and providing more accurate
-structure guidance for text-to-image models. It works by learning an alignment between the internal knowledge of the
-text-to-image model and an external control signal, such as edge detection or depth estimation.
+[T2I-Adapter](https://huggingface.co/papers/2302.08453) is an adapter that enables controllable generation like [ControlNet](./controlnet). A T2I-Adapter works by learning a *mapping* between a control signal (for example, a depth map) and a pretrained model's internal knowledge. The adapter is plugged in to the base model to provide extra guidance based on the control signal during generation.
 
-The T2I-Adapter design is simple, the condition is passed to four feature extraction blocks and three downsample
-blocks. This makes it fast and easy to train different adapters for different conditions which can be plugged into the
-text-to-image model. T2I-Adapter is similar to [ControlNet](controlnet) except it is smaller (~77M parameters) and
-faster because it only runs once during the diffusion process. The downside is that performance may be slightly worse
-than ControlNet.
-
-This guide will show you how to use T2I-Adapter with different Stable Diffusion models and how you can compose multiple
-T2I-Adapters to impose more than one condition.
-
-> [!TIP]
-> There are several T2I-Adapters available for different conditions, such as color palette, depth, sketch, pose, and
-> segmentation. Check out the [TencentARC](https://hf.co/TencentARC) repository to try them out!
-
-Before you begin, make sure you have the following libraries installed.
+Load a T2I-Adapter conditioned on a specific control, such as canny edge, and pass it to the pipeline in [`~DiffusionPipeline.from_pretrained`].
 
 ```py
-# uncomment to install the necessary libraries in Colab
-#!pip install -q diffusers accelerate controlnet-aux==0.0.7
+import torch
+from diffusers import T2IAdapter, StableDiffusionXLAdapterPipeline, AutoencoderKL
+
+t2i_adapter = T2IAdapter.from_pretrained(
+    "TencentARC/t2i-adapter-canny-sdxl-1.0",
+    torch_dtype=torch.float16,
+)
 ```
 
-## Text-to-image
-
-Text-to-image models rely on a prompt to generate an image, but sometimes, text alone may not be enough to provide more
-accurate structural guidance. T2I-Adapter allows you to provide an additional control image to guide the generation
-process. For example, you can provide a canny image (a white outline of an image on a black background) to guide the
-model to generate an image with a similar structure.
-
-<hfoptions id="stablediffusion">
-<hfoption id="Stable Diffusion 1.5">
-
-Create a canny image with the [opencv-library](https://github.com/opencv/opencv-python).
+Generate a canny image with [opencv-python](https://github.com/opencv/opencv-python).
 
 ```py
 import cv2
@@ -54,166 +34,124 @@ import numpy as np
 from PIL import Image
 from diffusers.utils import load_image
 
-image = load_image("https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/hf-logo.png")
-image = np.array(image)
+original_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png"
+)
+
+image = np.array(original_image)
 
 low_threshold = 100
 high_threshold = 200
 
 image = cv2.Canny(image, low_threshold, high_threshold)
-image = Image.fromarray(image)
+image = image[:, :, None]
+image = np.concatenate([image, image, image], axis=2)
+canny_image = Image.fromarray(image)
 ```
 
-Now load a T2I-Adapter conditioned on [canny images](https://hf.co/TencentARC/t2iadapter_canny_sd15v2) and pass it to
-the [`StableDiffusionAdapterPipeline`].
+Pass the canny image to the pipeline to generate an image.
 
 ```py
-import torch
-from diffusers import StableDiffusionAdapterPipeline, T2IAdapter
-
-adapter = T2IAdapter.from_pretrained("TencentARC/t2iadapter_canny_sd15v2", torch_dtype=torch.float16)
-pipeline = StableDiffusionAdapterPipeline.from_pretrained(
-    "stable-diffusion-v1-5/stable-diffusion-v1-5",
-    adapter=adapter,
-    torch_dtype=torch.float16,
-)
-pipeline.to("cuda")
-```
-
-Finally, pass your prompt and control image to the pipeline.
-
-```py
-generator = torch.Generator("cuda").manual_seed(0)
-
-image = pipeline(
-    prompt="cinematic photo of a plush and soft midcentury style rug on a wooden floor, 35mm photograph, film, professional, 4k, highly detailed",
-    image=image,
-    generator=generator,
-).images[0]
-image
-```
-
-<div class="flex justify-center">
-  <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/t2i-sd1.5.png"/>
-</div>
-
-</hfoption>
-<hfoption id="Stable Diffusion XL">
-
-Create a canny image with the [controlnet-aux](https://github.com/huggingface/controlnet_aux) library.
-
-```py
-from controlnet_aux.canny import CannyDetector
-from diffusers.utils import load_image
-
-canny_detector = CannyDetector()
-
-image = load_image("https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/sd_controlnet/hf-logo.png")
-image = canny_detector(image, detect_resolution=384, image_resolution=1024)
-```
-
-Now load a T2I-Adapter conditioned on [canny images](https://hf.co/TencentARC/t2i-adapter-canny-sdxl-1.0) and pass it
-to the [`StableDiffusionXLAdapterPipeline`].
-
-```py
-import torch
-from diffusers import StableDiffusionXLAdapterPipeline, T2IAdapter, EulerAncestralDiscreteScheduler, AutoencoderKL
-
-scheduler = EulerAncestralDiscreteScheduler.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", subfolder="scheduler")
 vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
-adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16)
 pipeline = StableDiffusionXLAdapterPipeline.from_pretrained(
     "stabilityai/stable-diffusion-xl-base-1.0",
-    adapter=adapter,
+    adapter=t2i_adapter,
     vae=vae,
-    scheduler=scheduler,
     torch_dtype=torch.float16,
-    variant="fp16",
-)
-pipeline.to("cuda")
-```
+).to("cuda")
 
-Finally, pass your prompt and control image to the pipeline.
+prompt = """
+A photorealistic overhead image of a cat reclining sideways in a flamingo pool floatie holding a margarita. 
+The cat is floating leisurely in the pool and completely relaxed and happy.
+"""
 
-```py
-generator = torch.Generator("cuda").manual_seed(0)
-
-image = pipeline(
-  prompt="cinematic photo of a plush and soft midcentury style rug on a wooden floor, 35mm photograph, film, professional, 4k, highly detailed",
-  image=image,
-  generator=generator,
+pipeline(
+    prompt, 
+    image=canny_image,
+    num_inference_steps=100, 
+    guidance_scale=10,
 ).images[0]
-image
 ```
 
-<div class="flex justify-center">
-  <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/t2i-sdxl.png"/>
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/non-enhanced-prompt.png" width="300" alt="Generated image (prompt only)"/>
+    <figcaption style="text-align: center;">original image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">canny image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/t2i-canny-cat-generated.png" width="300" alt="Generated image (ControlNet + prompt)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
 </div>
 
-</hfoption>
-</hfoptions>
-
 ## MultiAdapter
 
-T2I-Adapters are also composable, allowing you to use more than one adapter to impose multiple control conditions on an
-image. For example, you can use a pose map to provide structural control and a depth map for depth control. This is
-enabled by the [`MultiAdapter`] class.
+You can compose multiple controls, such as canny image and a depth map, with the [`MultiAdapter`] class.
 
-Let's condition a text-to-image model with a pose and depth adapter. Create and place your depth and pose image and in a list.
+The example below composes a canny image and depth map.
 
-```py
-from diffusers.utils import load_image
-
-pose_image = load_image(
-    "https://huggingface.co/datasets/diffusers/docs-images/resolve/main/t2i-adapter/keypose_sample_input.png"
-)
-depth_image = load_image(
-    "https://huggingface.co/datasets/diffusers/docs-images/resolve/main/t2i-adapter/depth_sample_input.png"
-)
-cond = [pose_image, depth_image]
-prompt = ["Santa Claus walking into an office room with a beautiful city view"]
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/t2i-adapter/depth_sample_input.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">depth image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/diffusers/docs-images/resolve/main/t2i-adapter/keypose_sample_input.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">pose image</figcaption>
-  </div>
-</div>
-
-Load the corresponding pose and depth adapters as a list in the [`MultiAdapter`] class.
+Load the control images and T2I-Adapters as a list.
 
 ```py
 import torch
-from diffusers import StableDiffusionAdapterPipeline, MultiAdapter, T2IAdapter
+from diffusers.utils import load_image
+from diffusers import StableDiffusionXLAdapterPipeline, AutoencoderKL, MultiAdapter, T2IAdapter
+
+canny_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png"
+)
+depth_image = load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_depth_image.png"
+)
+controls = [canny_image, depth_image]
+prompt = ["""
+a relaxed rabbit sitting on a striped towel next to a pool with a tropical drink nearby, 
+bright sunny day, vacation scene, 35mm photograph, film, professional, 4k, highly detailed
+"""]
 
 adapters = MultiAdapter(
     [
-        T2IAdapter.from_pretrained("TencentARC/t2iadapter_keypose_sd14v1"),
-        T2IAdapter.from_pretrained("TencentARC/t2iadapter_depth_sd14v1"),
+        T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16),
+        T2IAdapter.from_pretrained("TencentARC/t2i-adapter-depth-midas-sdxl-1.0", torch_dtype=torch.float16),
     ]
 )
-adapters = adapters.to(torch.float16)
 ```
 
-Finally, load a [`StableDiffusionAdapterPipeline`] with the adapters, and pass your prompt and conditioned images to
-it. Use the [`adapter_conditioning_scale`] to adjust the weight of each adapter on the image.
+Pass the adapters, prompt, and control images to [`StableDiffusionXLAdapterPipeline`]. Use the `adapter_conditioning_scale` parameter to determine how much weight to assign to each control.
 
 ```py
-pipeline = StableDiffusionAdapterPipeline.from_pretrained(
-    "CompVis/stable-diffusion-v1-4",
+vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
+pipeline = StableDiffusionXLAdapterPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
     torch_dtype=torch.float16,
+    vae=vae,
     adapter=adapters,
 ).to("cuda")
 
-image = pipeline(prompt, cond, adapter_conditioning_scale=[0.7, 0.7]).images[0]
-image
+pipeline(
+    prompt,
+    image=controls,
+    height=1024,
+    width=1024,
+    adapter_conditioning_scale=[0.7, 0.7]
+).images[0]
 ```
 
-<div class="flex justify-center">
-  <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/t2i-multi.png"/>
+<div style="display: flex; gap: 10px; justify-content: space-around; align-items: flex-end;">
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/canny-cat.png" width="300" alt="Generated image (prompt only)"/>
+    <figcaption style="text-align: center;">canny image</figcaption>
+  </figure>
+  <figure>
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/sdxl_depth_image.png" width="300" alt="Control image (Canny edges)"/>
+    <figcaption style="text-align: center;">depth map</figcaption>
+  </figure>
+  <figure> 
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/t2i-multi-rabbit.png" width="300" alt="Generated image (ControlNet + prompt)"/>
+    <figcaption style="text-align: center;">generated image</figcaption>
+  </figure>
 </div>

docs/source/en/using-diffusers/text-img2vid.md

@@ -1,4 +1,4 @@
- <!--Copyright 2024 The HuggingFace Team. All rights reserved.
+<!--Copyright 2024 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
@@ -12,551 +12,436 @@ specific language governing permissions and limitations under the License.
 
 # Video generation
 
-Video generation models include a temporal dimension to bring images, or frames, together to create a video. These models are trained on large-scale datasets of high-quality text-video pairs to learn how to combine the modalities to ensure the generated video is coherent and realistic.
+Video generation models extend image generation (can be considered a 1-frame video) to also process data related to space and time. Making sure all this data - text, space, time - remain consistent and aligned from frame-to-frame is a big challenge in generating long and high-resolution videos.
 
-[Explore](https://huggingface.co/models?other=video-generation) some of the more popular open-source video generation models available from Diffusers below.
+Modern video models tackle this challenge with the diffusion transformer (DiT) architecture. This reduces computational costs and allows more efficient scaling to larger and higher-quality image and video data.
 
-<hfoptions id="popular-models">
-<hfoption id="CogVideoX">
+Check out what some of these video models are capable of below.
 
-[CogVideoX](https://huggingface.co/collections/THUDM/cogvideo-66c08e62f1685a3ade464cce) uses a 3D causal Variational Autoencoder (VAE) to compress videos along the spatial and temporal dimensions, and it includes a stack of expert transformer blocks with a 3D full attention mechanism to better capture visual, semantic, and motion information in the data.
-
-The CogVideoX family also includes models capable of generating videos from images and videos in addition to text. The image-to-video models are indicated by **I2V** in the checkpoint name, and they should be used with the [`CogVideoXImageToVideoPipeline`]. The regular checkpoints support video-to-video through the [`CogVideoXVideoToVideoPipeline`].
-
-The example below demonstrates how to generate a video from an image and text prompt with [THUDM/CogVideoX-5b-I2V](https://huggingface.co/THUDM/CogVideoX-5b-I2V).
+<hfoptions id="popular models">
+<hfoption id="Wan2.1">
 
 ```py
+# pip install ftfy
 import torch
-from diffusers import CogVideoXImageToVideoPipeline
+import numpy as np
+from diffusers import AutoModel, WanPipeline
+from diffusers.hooks.group_offloading import apply_group_offloading
 from diffusers.utils import export_to_video, load_image
+from transformers import UMT5EncoderModel
 
-prompt = "A vast, shimmering ocean flows gracefully under a twilight sky, its waves undulating in a mesmerizing dance of blues and greens. The surface glints with the last rays of the setting sun, casting golden highlights that ripple across the water. Seagulls soar above, their cries blending with the gentle roar of the waves. The horizon stretches infinitely, where the ocean meets the sky in a seamless blend of hues. Close-ups reveal the intricate patterns of the waves, capturing the fluidity and dynamic beauty of the sea in motion."
-image = load_image(image="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cogvideox/cogvideox_rocket.png")
-pipe = CogVideoXImageToVideoPipeline.from_pretrained(
-    "THUDM/CogVideoX-5b-I2V",
-    torch_dtype=torch.bfloat16
+text_encoder = UMT5EncoderModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", torch_dtype=torch.float32)
+transformer = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="transformer", torch_dtype=torch.bfloat16)
+
+# group-offloading
+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
 )
 
-# reduce memory requirements 
-pipe.vae.enable_tiling()
-pipe.vae.enable_slicing()
+pipeline = WanPipeline.from_pretrained(
+    "Wan-AI/Wan2.1-T2V-14B-Diffusers",
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    torch_dtype=torch.bfloat16
+)
+pipeline.to("cuda")
 
-video = pipe(
+prompt = """
+The camera rushes from far to near in a low-angle shot, 
+revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic 
+shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+"""
+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
+"""
+
+output = pipeline(
     prompt=prompt,
-    image=image,
-    num_videos_per_prompt=1,
-    num_inference_steps=50,
-    num_frames=49,
-    guidance_scale=6,
-    generator=torch.Generator(device="cuda").manual_seed(42),
+    negative_prompt=negative_prompt,
+    num_frames=81,
+    guidance_scale=5.0,
 ).frames[0]
-export_to_video(video, "output.mp4", fps=8)
+export_to_video(output, "output.mp4", fps=16)
 ```
 
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cogvideox/cogvideox_rocket.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">initial image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cogvideox/cogvideox_outrocket.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">generated video</figcaption>
-  </div>
-</div>
-
 </hfoption>
 <hfoption id="HunyuanVideo">
 
-> [!TIP]
-> HunyuanVideo is a 13B parameter model and requires a lot of memory. Refer to the HunyuanVideo [Quantization](../api/pipelines/hunyuan_video#quantization) guide to learn how to quantize the model. CogVideoX and LTX-Video are more lightweight options that can still generate high-quality videos.
-
-[HunyuanVideo](https://huggingface.co/tencent/HunyuanVideo) features a dual-stream to single-stream diffusion transformer (DiT) for learning video and text tokens separately, and then subsequently concatenating the video and text tokens to combine their information. A single multimodal large language model (MLLM) serves as the text encoder, and videos are also spatio-temporally compressed with a 3D causal VAE.
-
 ```py
 import torch
-from diffusers import HunyuanVideoPipeline, HunyuanVideoTransformer3DModel
+from diffusers importAutoModel, HunyuanVideoPipeline
+from diffusers.quantizers import PipelineQuantizationConfig
 from diffusers.utils import export_to_video
 
-transformer = HunyuanVideoTransformer3DModel.from_pretrained(
-    "hunyuanvideo-community/HunyuanVideo", subfolder="transformer", torch_dtype=torch.bfloat16
-)
-pipe = HunyuanVideoPipeline.from_pretrained(
-  "hunyuanvideo-community/HunyuanVideo", transformer=transformer, torch_dtype=torch.float16
+# quantize weights to int4 with bitsandbytes
+pipeline_quant_config = PipelineQuantizationConfig(
+  quant_backend="bitsandbytes_4bit",
+  quant_kwargs={
+    "load_in_4bit": True,
+    "bnb_4bit_quant_type": "nf4",
+    "bnb_4bit_compute_dtype": torch.bfloat16
+    },
+  components_to_quantize=["transformer"]
 )
 
-# reduce memory requirements
-pipe.vae.enable_tiling()
-pipe.to("cuda")
+pipeline = HunyuanVideoPipeline.from_pretrained(
+    "hunyuanvideo-community/HunyuanVideo",
+    quantization_config=pipeline_quant_config,
+    torch_dtype=torch.bfloat16,
+)
 
-video = pipe(
-    prompt="A cat walks on the grass, realistic",
-    height=320,
-    width=512,
-    num_frames=61,
-    num_inference_steps=30,
-).frames[0]
+# model-offloading and tiling
+pipeline.enable_model_cpu_offload()
+pipeline.vae.enable_tiling()
+
+prompt = "A fluffy teddy bear sits on a bed of soft pillows surrounded by children's toys."
+video = pipeline(prompt=prompt, num_frames=61, num_inference_steps=30).frames[0]
 export_to_video(video, "output.mp4", fps=15)
 ```
 
-<div class="flex justify-center">
-  <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/hunyuan-video-output.gif"/>
-</div>
-
 </hfoption>
 <hfoption id="LTX-Video">
 
-[LTX-Video (LTXV)](https://huggingface.co/Lightricks/LTX-Video) is a diffusion transformer (DiT) with a focus on speed. It generates 768x512 resolution videos at 24 frames per second (fps), enabling near real-time generation of high-quality videos. LTXV is relatively lightweight compared to other modern video generation models, making it possible to run on consumer GPUs.
+```py
+import torch
+from diffusers import LTXPipeline, AutoModel
+from diffusers.hooks import apply_group_offloading
+from diffusers.utils import export_to_video
+
+# fp8 layerwise weight-casting
+transformer = AutoModel.from_pretrained(
+    "Lightricks/LTX-Video",
+    subfolder="transformer",
+    torch_dtype=torch.bfloat16
+)
+transformer.enable_layerwise_casting(
+    storage_dtype=torch.float8_e4m3fn, compute_dtype=torch.bfloat16
+)
+
+pipeline = LTXPipeline.from_pretrained("Lightricks/LTX-Video", transformer=transformer, torch_dtype=torch.bfloat16)
+
+# group-offloading
+onload_device = torch.device("cuda")
+offload_device = torch.device("cpu")
+pipeline.transformer.enable_group_offload(onload_device=onload_device, offload_device=offload_device, offload_type="leaf_level", use_stream=True)
+apply_group_offloading(pipeline.text_encoder, onload_device=onload_device, offload_type="block_level", num_blocks_per_group=2)
+apply_group_offloading(pipeline.vae, onload_device=onload_device, offload_type="leaf_level")
+
+prompt = """
+A woman with long brown hair and light skin smiles at another woman with long blonde hair. The woman with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek. The camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be real-life footage
+"""
+negative_prompt = "worst quality, inconsistent motion, blurry, jittery, distorted"
+
+video = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    width=768,
+    height=512,
+    num_frames=161,
+    decode_timestep=0.03,
+    decode_noise_scale=0.025,
+    num_inference_steps=50,
+).frames[0]
+export_to_video(video, "output.mp4", fps=24)
+```
+
+</hfoption>
+</hfoptions>
+
+This guide will cover video generation basics such as which parameters to configure and how to reduce their memory usage.
+
+> [!TIP]
+> If you're interested in learning more about how to use a specific model, please refer to their pipeline API model card.
+
+## Pipeline parameters
+
+There are several parameters to configure in the pipeline that'll affect video generation quality or speed. Experimenting with different parameter values is important for discovering the appropriate quality and speed tradeoff.
+
+### num_frames
+
+A frame is a still image that is played in a sequence of other frames to create motion or a video. Control the number of frames generated per second with `num_frames`. Increasing `num_frames` increases perceived motion smoothness and visual coherence, making it especially important for videos with dynamic content. A higher `num_frames` value also increases video duration.
+
+Some video models require more specific `num_frames` values for inference. For example, [`HunyuanVideoPipeline`] recommends calculating the `num_frames` with `(4 * num_frames) +1`. Always check a pipelines API model card to see if there is a recommended value.
 
 ```py
 import torch
 from diffusers import LTXPipeline
 from diffusers.utils import export_to_video
 
-pipe = LTXPipeline.from_pretrained("Lightricks/LTX-Video", torch_dtype=torch.bfloat16).to("cuda")
+pipeline = LTXPipeline.from_pretrained(
+    "Lightricks/LTX-Video", torch_dtype=torch.bfloat16
+).to("cuda")
 
-prompt = "A man walks towards a window, looks out, and then turns around. He has short, dark hair, dark skin, and is wearing a brown coat over a red and gray scarf. He walks from left to right towards a window, his gaze fixed on something outside. The camera follows him from behind at a medium distance. The room is brightly lit, with white walls and a large window covered by a white curtain. As he approaches the window, he turns his head slightly to the left, then back to the right. He then turns his entire body to the right, facing the window. The camera remains stationary as he stands in front of the window. The scene is captured in real-life footage."
-video = pipe(
+prompt = """
+A woman with long brown hair and light skin smiles at another woman with long blonde hair. The woman 
+with brown hair wears a black jacket and has a small, barely noticeable mole on her right cheek. The 
+camera angle is a close-up, focused on the woman with brown hair's face. The lighting is warm and 
+natural, likely from the setting sun, casting a soft glow on the scene. The scene appears to be 
+real-life footage
+"""
+
+video = pipeline(
     prompt=prompt,
-    width=704,
-    height=480,
+    negative_prompt=negative_prompt,
+    width=768,
+    height=512,
     num_frames=161,
+    decode_timestep=0.03,
+    decode_noise_scale=0.025,
     num_inference_steps=50,
 ).frames[0]
 export_to_video(video, "output.mp4", fps=24)
 ```
 
-<div class="flex justify-center">
-  <img src="https://huggingface.co/Lightricks/LTX-Video/resolve/main/media/ltx-video_example_00014.gif"/>
-</div>
+### guidance_scale
 
-</hfoption>
-<hfoption id="Mochi-1">
-
-> [!TIP]
-> Mochi-1 is a 10B parameter model and requires a lot of memory. Refer to the Mochi [Quantization](../api/pipelines/mochi#quantization) guide to learn how to quantize the model. CogVideoX and LTX-Video are more lightweight options that can still generate high-quality videos.
-
-[Mochi-1](https://huggingface.co/genmo/mochi-1-preview) introduces the Asymmetric Diffusion Transformer (AsymmDiT) and Asymmetric Variational Autoencoder (AsymmVAE) to reduces memory requirements. AsymmVAE causally compresses videos 128x to improve memory efficiency, and AsymmDiT jointly attends to the compressed video tokens and user text tokens. This model is noted for generating videos with high-quality motion dynamics and strong prompt adherence.
+Guidance scale or "cfg" controls how closely the generated frames adhere to the input conditioning (text, image or both). Increasing `guidance_scale` generates frames that resemble the input conditions more closely and includes finer details, but risk introducing artifacts and reducing output diversity. Lower `guidance_scale` values encourages looser prompt adherence and increased output variety, but details may not be as great. If it's too low, it may ignore your prompt entirely and generate random noise.
 
 ```py
 import torch
-from diffusers import MochiPipeline
+from diffusers import CogVideoXPipeline, CogVideoXTransformer3DModel
 from diffusers.utils import export_to_video
 
-pipe = MochiPipeline.from_pretrained("genmo/mochi-1-preview", variant="bf16", torch_dtype=torch.bfloat16)
+pipeline = CogVideoXPipeline.from_pretrained(
+  "THUDM/CogVideoX-2b",
+  torch_dtype=torch.float16
+).to("cuda")
 
-# reduce memory requirements
-pipe.enable_model_cpu_offload()
-pipe.enable_vae_tiling()
+prompt = """
+A detailed wooden toy ship with intricately carved masts and sails is seen gliding smoothly over
+a plush, blue carpet that mimics the waves of the sea. The ship's hull is painted a rich brown, 
+with tiny windows. The carpet, soft and textured, provides a perfect backdrop, resembling an 
+oceanic expanse. Surrounding the ship are various other toys and children's items, hinting at 
+a playful environment. The scene captures the innocence and imagination of childhood, 
+with the toy ship's journey symbolizing endless adventures in a whimsical, indoor setting.
+"""
 
-prompt = "Close-up of a chameleon's eye, with its scaly skin changing color. Ultra high resolution 4k."
-video = pipe(prompt, num_frames=84).frames[0]
-export_to_video(video, "output.mp4", fps=30)
-```
-
-<div class="flex justify-center">
-  <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/mochi-video-output.gif"/>
-</div>
-
-</hfoption>
-<hfoption id="StableVideoDiffusion">
-
-[StableVideoDiffusion (SVD)](https://huggingface.co/stabilityai/stable-video-diffusion-img2vid-xt) is based on the Stable Diffusion 2.1 model and it is trained on images, then low-resolution videos, and finally a smaller dataset of high-resolution videos. This model generates a short 2-4 second video from an initial image.
-
-```py
-import torch
-from diffusers import StableVideoDiffusionPipeline
-from diffusers.utils import load_image, export_to_video
-
-pipeline = StableVideoDiffusionPipeline.from_pretrained(
-    "stabilityai/stable-video-diffusion-img2vid-xt", torch_dtype=torch.float16, variant="fp16"
-)
-
-# reduce memory requirements
-pipeline.enable_model_cpu_offload()
-
-image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/svd/rocket.png")
-image = image.resize((1024, 576))
-
-generator = torch.manual_seed(42)
-frames = pipeline(image, decode_chunk_size=8, generator=generator).frames[0]
-export_to_video(frames, "generated.mp4", fps=7)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/svd/rocket.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">initial image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/svd/output_rocket.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">generated video</figcaption>
-  </div>
-</div>
-
-</hfoption>
-<hfoption id="AnimateDiff">
-
-[AnimateDiff](https://huggingface.co/guoyww/animatediff) is an adapter model that inserts a motion module into a pretrained diffusion model to animate an image. The adapter is trained on video clips to learn motion which is used to condition the generation process to create a video. It is faster and easier to only train the adapter and it can be loaded into most diffusion models, effectively turning them into “video models”.
-
-Load a `MotionAdapter` and pass it to the [`AnimateDiffPipeline`].
-
-```py
-import torch
-from diffusers import AnimateDiffPipeline, DDIMScheduler, MotionAdapter
-from diffusers.utils import export_to_gif
-
-adapter = MotionAdapter.from_pretrained("guoyww/animatediff-motion-adapter-v1-5-2", torch_dtype=torch.float16)
-pipeline = AnimateDiffPipeline.from_pretrained("emilianJR/epiCRealism", motion_adapter=adapter, torch_dtype=torch.float16)
-scheduler = DDIMScheduler.from_pretrained(
-    "emilianJR/epiCRealism",
-    subfolder="scheduler",
-    clip_sample=False,
-    timestep_spacing="linspace",
-    beta_schedule="linear",
-    steps_offset=1,
-)
-pipeline.scheduler = scheduler
-
-# reduce memory requirements
-pipeline.enable_vae_slicing()
-pipeline.enable_model_cpu_offload()
-
-output = pipeline(
-    prompt="A space rocket with trails of smoke behind it launching into space from the desert, 4k, high resolution",
-    negative_prompt="bad quality, worse quality, low resolution",
-    num_frames=16,
-    guidance_scale=7.5,
-    num_inference_steps=50,
-    generator=torch.Generator("cpu").manual_seed(49),
-)
-frames = output.frames[0]
-export_to_gif(frames, "animation.gif")
-```
-
-<div class="flex justify-center">
-  <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/animatediff.gif"/>
-</div>
-
-</hfoption>
-</hfoptions>
-
-## Configure model parameters
-
-There are a few important parameters you can configure in the pipeline that'll affect the video generation process and quality. Let's take a closer look at what these parameters do and how changing them affects the output.
-
-### Number of frames
-
-The `num_frames` parameter determines how many video frames are generated per second. A frame is an image that is played in a sequence of other frames to create motion or a video. This affects video length because the pipeline generates a certain number of frames per second (check a pipeline's API reference for the default value). To increase the video duration, you'll need to increase the `num_frames` parameter.
-
-```py
-import torch
-from diffusers import StableVideoDiffusionPipeline
-from diffusers.utils import load_image, export_to_video
-
-pipeline = StableVideoDiffusionPipeline.from_pretrained(
-    "stabilityai/stable-video-diffusion-img2vid", torch_dtype=torch.float16, variant="fp16"
-)
-pipeline.enable_model_cpu_offload()
-
-image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/svd/rocket.png")
-image = image.resize((1024, 576))
-
-generator = torch.manual_seed(42)
-frames = pipeline(image, decode_chunk_size=8, generator=generator, num_frames=25).frames[0]
-export_to_video(frames, "generated.mp4", fps=7)
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/num_frames_14.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">num_frames=14</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/num_frames_25.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">num_frames=25</figcaption>
-  </div>
-</div>
-
-### Guidance scale
-
-The `guidance_scale` parameter controls how closely aligned the generated video and text prompt or initial image is. A higher `guidance_scale` value means your generated video is more aligned with the text prompt or initial image, while a lower `guidance_scale` value means your generated video is less aligned which could give the model more "creativity" to interpret the conditioning input.
-
-<Tip>
-
-SVD uses the `min_guidance_scale` and `max_guidance_scale` parameters for applying guidance to the first and last frames respectively.
-
-</Tip>
-
-```py
-import torch
-from diffusers import I2VGenXLPipeline
-from diffusers.utils import export_to_gif, load_image
-
-pipeline = I2VGenXLPipeline.from_pretrained("ali-vilab/i2vgen-xl", torch_dtype=torch.float16, variant="fp16")
-pipeline.enable_model_cpu_offload()
-
-image_url = "https://huggingface.co/datasets/diffusers/docs-images/resolve/main/i2vgen_xl_images/img_0009.png"
-image = load_image(image_url).convert("RGB")
-
-prompt = "Papers were floating in the air on a table in the library"
-negative_prompt = "Distorted, discontinuous, Ugly, blurry, low resolution, motionless, static, disfigured, disconnected limbs, Ugly faces, incomplete arms"
-generator = torch.manual_seed(0)
-
-frames = pipeline(
-    prompt=prompt,
-    image=image,
-    num_inference_steps=50,
-    negative_prompt=negative_prompt,
-    guidance_scale=1.0,
-    generator=generator
+video = pipeline(
+  prompt=prompt,
+  guidance_scale=6,
+  num_inference_steps=50
 ).frames[0]
-export_to_gif(frames, "i2v.gif")
+export_to_video(video, "output.mp4", fps=8)
 ```
 
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/i2vgen-xl-example.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">guidance_scale=9.0</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/guidance_scale_1.0.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">guidance_scale=1.0</figcaption>
-  </div>
-</div>
+### negative_prompt
 
-### Negative prompt
-
-A negative prompt deters the model from generating things you don’t want it to. This parameter is commonly used to improve overall generation quality by removing poor or bad features such as “low resolution” or “bad details”.
+A negative prompt is useful for excluding things you don't want to see in the generated video. It is commonly used to refine the quality and alignment of the generated video by pushing the model away from undesirable elements like "blurry, distorted, ugly". This can create cleaner and more focused videos.
 
 ```py
+# pip install ftfy
 import torch
-from diffusers import AnimateDiffPipeline, DDIMScheduler, MotionAdapter
-from diffusers.utils import export_to_gif
+from diffusers import WanPipeline
+from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
+from diffusers.utils import export_to_video
 
-adapter = MotionAdapter.from_pretrained("guoyww/animatediff-motion-adapter-v1-5-2", torch_dtype=torch.float16)
-
-pipeline = AnimateDiffPipeline.from_pretrained("emilianJR/epiCRealism", motion_adapter=adapter, torch_dtype=torch.float16)
-scheduler = DDIMScheduler.from_pretrained(
-    "emilianJR/epiCRealism",
-    subfolder="scheduler",
-    clip_sample=False,
-    timestep_spacing="linspace",
-    beta_schedule="linear",
-    steps_offset=1,
+vae = AutoencoderKLWan.from_pretrained(
+  "Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", torch_dtype=torch.float32
 )
-pipeline.scheduler = scheduler
-pipeline.enable_vae_slicing()
+pipeline = WanPipeline.from_pretrained(
+  "Wan-AI/Wan2.1-T2V-14B-Diffusers", vae=vae, torch_dtype=torch.bfloat16
+)
+pipeline.scheduler = UniPCMultistepScheduler.from_config(
+  pipeline.scheduler.config, flow_shift=5.0
+)
+pipeline.to("cuda")
+
+pipeline.load_lora_weights("benjamin-paine/steamboat-willie-14b", adapter_name="steamboat-willie")
+pipeline.set_adapters("steamboat-willie")
+
 pipeline.enable_model_cpu_offload()
 
+# use "steamboat willie style" to trigger the LoRA
+prompt = """
+steamboat willie style, golden era animation, The camera rushes from far to near in a low-angle shot, 
+revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts 
+dynamic shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+"""
+
 output = pipeline(
-    prompt="360 camera shot of a sushi roll in a restaurant",
-    negative_prompt="Distorted, discontinuous, ugly, blurry, low resolution, motionless, static",
-    num_frames=16,
-    guidance_scale=7.5,
-    num_inference_steps=50,
-    generator=torch.Generator("cpu").manual_seed(0),
+  prompt=prompt,
+  num_frames=81,
+  guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "output.mp4", fps=16)
+```
+
+## Reduce memory usage
+
+Recent video models like [`HunyuanVideoPipeline`] and [`WanPipeline`], which have 10B+ parameters, require a lot of memory and it often exceeds the memory availabe on consumer hardware. Diffusers offers several techniques for reducing the memory requirements of these large models.
+
+> [!TIP]
+> Refer to the [Reduce memory usage](../optimization/memory) guide for more details about other memory saving techniques.
+
+One of these techniques is [group-offloading](../optimization/memory#group-offloading), which offloads groups of internal model layers (such as `torch.nn.Sequential`) to the CPU when it isn't being used. These layers are only loaded when they're needed for computation to avoid storing **all** the model components on the GPU. For a 14B parameter model like [`WanPipeline`], group-offloading can lower the required memory to ~13GB of VRAM.
+
+```py
+# pip install ftfy
+import torch
+import numpy as np
+from diffusers import AutoModel, WanPipeline
+from diffusers.hooks.group_offloading import apply_group_offloading
+from diffusers.utils import export_to_video, load_image
+from transformers import UMT5EncoderModel
+
+text_encoder = UMT5EncoderModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="text_encoder", torch_dtype=torch.bfloat16)
+vae = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", torch_dtype=torch.float32)
+transformer = AutoModel.from_pretrained("Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="transformer", torch_dtype=torch.bfloat16)
+
+# group-offloading
+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
 )
-frames = output.frames[0]
-export_to_gif(frames, "animation.gif")
+transformer.enable_group_offload(
+    onload_device=onload_device,
+    offload_device=offload_device,
+    offload_type="leaf_level",
+    use_stream=True
+)
+
+pipeline = WanPipeline.from_pretrained(
+    "Wan-AI/Wan2.1-T2V-14B-Diffusers",
+    vae=vae,
+    transformer=transformer,
+    text_encoder=text_encoder,
+    torch_dtype=torch.bfloat16
+)
+pipeline.to("cuda")
+
+prompt = """
+The camera rushes from far to near in a low-angle shot, 
+revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts dynamic 
+shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+"""
+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
+"""
+
+output = pipeline(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    num_frames=81,
+    guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "output.mp4", fps=16)
 ```
 
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/animatediff_no_neg.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">no negative prompt</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/animatediff_neg.gif"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">negative prompt applied</figcaption>
-  </div>
-</div>
+Another option for reducing memory is to consider quantizing a model, which stores the model weights in a lower precision data type. However, quantization may impact video quality depending on the specific video model. Refer to the quantization [Overivew](../quantization/overview) to learn more about the different supported quantization backends.
 
-### Model-specific parameters
-
-There are some pipeline parameters that are unique to each model such as adjusting the motion in a video or adding noise to the initial image.
-
-<hfoptions id="special-parameters">
-<hfoption id="Stable Video Diffusion">
-
-Stable Video Diffusion provides additional micro-conditioning for the frame rate with the `fps` parameter and for motion with the `motion_bucket_id` parameter. Together, these parameters allow for adjusting the amount of motion in the generated video.
-
-There is also a `noise_aug_strength` parameter that increases the amount of noise added to the initial image. Varying this parameter affects how similar the generated video and initial image are. A higher `noise_aug_strength` also increases the amount of motion. To learn more, read the [Micro-conditioning](../using-diffusers/svd#micro-conditioning) guide.
-
-</hfoption>
-<hfoption id="Text2Video-Zero">
-
-Text2Video-Zero computes the amount of motion to apply to each frame from randomly sampled latents. You can use the `motion_field_strength_x` and `motion_field_strength_y` parameters to control the amount of motion to apply to the x and y-axes of the video. The parameters `t0` and `t1` are the timesteps to apply motion to the latents.
-
-</hfoption>
-</hfoptions>
-
-## Control video generation
-
-Video generation can be controlled similar to how text-to-image, image-to-image, and inpainting can be controlled with a [`ControlNetModel`]. The only difference is you need to use the [`~pipelines.text_to_video_synthesis.pipeline_text_to_video_zero.CrossFrameAttnProcessor`] so each frame attends to the first frame.
-
-### Text2Video-Zero
-
-Text2Video-Zero video generation can be conditioned on pose and edge images for even greater control over a subject's motion in the generated video or to preserve the identity of a subject/object in the video. You can also use Text2Video-Zero with [InstructPix2Pix](../api/pipelines/pix2pix) for editing videos with text.
-
-<hfoptions id="t2v-zero">
-<hfoption id="pose control">
-
-Start by downloading a video and extracting the pose images from it.
+The example below uses [bitsandbytes](../quantization/bitsandbytes) to quantize a model.
 
 ```py
-from huggingface_hub import hf_hub_download
-from PIL import Image
-import imageio
+# pip install ftfy
 
-filename = "__assets__/poses_skeleton_gifs/dance1_corr.mp4"
-repo_id = "PAIR/Text2Video-Zero"
-video_path = hf_hub_download(repo_type="space", repo_id=repo_id, filename=filename)
+import torch
+from diffusers import WanPipeline
+from diffusers import AutoModel, WanPipeline
+from diffusers.quantizers import PipelineQuantizationConfig
+from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler
+from transformers import UMT5EncoderModel
+from diffusers.utils import export_to_video
 
-reader = imageio.get_reader(video_path, "ffmpeg")
-frame_count = 8
-pose_images = [Image.fromarray(reader.get_data(i)) for i in range(frame_count)]
+# quantize transformer and text encoder weights with bitsandbytes
+pipeline_quant_config = PipelineQuantizationConfig(
+  quant_backend="bitsandbytes_4bit",
+  quant_kwargs={"load_in_4bit": True},
+  components_to_quantize=["transformer", "text_encoder"]
+)
+
+vae = AutoModel.from_pretrained(
+  "Wan-AI/Wan2.1-T2V-14B-Diffusers", subfolder="vae", torch_dtype=torch.float32
+)
+pipeline = WanPipeline.from_pretrained(
+  "Wan-AI/Wan2.1-T2V-14B-Diffusers", vae=vae, quantization_config=pipeline_quant_config, torch_dtype=torch.bfloat16
+)
+pipeline.scheduler = UniPCMultistepScheduler.from_config(
+  pipeline.scheduler.config, flow_shift=5.0
+)
+pipeline.to("cuda")
+
+pipeline.load_lora_weights("benjamin-paine/steamboat-willie-14b", adapter_name="steamboat-willie")
+pipeline.set_adapters("steamboat-willie")
+
+pipeline.enable_model_cpu_offload()
+
+# use "steamboat willie style" to trigger the LoRA
+prompt = """
+steamboat willie style, golden era animation, The camera rushes from far to near in a low-angle shot, 
+revealing a white ferret on a log. It plays, leaps into the water, and emerges, as the camera zooms in 
+for a close-up. Water splashes berry bushes nearby, while moss, snow, and leaves blanket the ground. 
+Birch trees and a light blue sky frame the scene, with ferns in the foreground. Side lighting casts 
+dynamic shadows and warm highlights. Medium composition, front view, low angle, with depth of field.
+"""
+
+output = pipeline(
+  prompt=prompt,
+  num_frames=81,
+  guidance_scale=5.0,
+).frames[0]
+export_to_video(output, "output.mp4", fps=16)
 ```
 
-Load a [`ControlNetModel`] for pose estimation and a checkpoint into the [`StableDiffusionControlNetPipeline`]. Then you'll use the [`~pipelines.text_to_video_synthesis.pipeline_text_to_video_zero.CrossFrameAttnProcessor`] for the UNet and ControlNet.
+## Inference speed
+
+[torch.compile](https://pytorch.org/tutorials/intermediate/torch_compile_tutorial_.html) can speedup inference by using optimized kernels. Compilation takes longer the first time, but once compiled, it is much faster. It is best to compile the pipeline once, and then use the pipeline multiple times without changing anything. A change, such as in the image size, triggers recompilation.
+
+The example below compiles the transformer in the pipeline and uses the `"max-autotune"` mode to maximize performance.
 
 ```py
 import torch
-from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
-from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero import CrossFrameAttnProcessor
+from diffusers import CogVideoXPipeline, CogVideoXTransformer3DModel
+from diffusers.utils import export_to_video
 
-model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
-controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-openpose", torch_dtype=torch.float16)
-pipeline = StableDiffusionControlNetPipeline.from_pretrained(
-    model_id, controlnet=controlnet, torch_dtype=torch.float16
+pipeline = CogVideoXPipeline.from_pretrained(
+  "THUDM/CogVideoX-2b",
+  torch_dtype=torch.float16
 ).to("cuda")
 
-pipeline.unet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
-pipeline.controlnet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
-```
+# torch.compile
+pipeline.transformer.to(memory_format=torch.channels_last)
+pipeline.transformer = torch.compile(
+    pipeline.transformer, mode="max-autotune", fullgraph=True
+)
 
-Fix the latents for all the frames, and then pass your prompt and extracted pose images to the model to generate a video.
+prompt = """
+A detailed wooden toy ship with intricately carved masts and sails is seen gliding smoothly over a plush, blue carpet that mimics the waves of the sea. 
+The ship's hull is painted a rich brown, with tiny windows. The carpet, soft and textured, provides a perfect backdrop, resembling an oceanic expanse. 
+Surrounding the ship are various other toys and children's items, hinting at a playful environment. The scene captures the innocence and imagination of childhood, 
+with the toy ship's journey symbolizing endless adventures in a whimsical, indoor setting.
+"""
 
-```py
-latents = torch.randn((1, 4, 64, 64), device="cuda", dtype=torch.float16).repeat(len(pose_images), 1, 1, 1)
-
-prompt = "Darth Vader dancing in a desert"
-result = pipeline(prompt=[prompt] * len(pose_images), image=pose_images, latents=latents).images
-imageio.mimsave("video.mp4", result, fps=4)
-```
-
-</hfoption>
-<hfoption id="edge control">
-
-Download a video and extract the edges from it.
-
-```py
-from huggingface_hub import hf_hub_download
-from PIL import Image
-import imageio
-
-filename = "__assets__/poses_skeleton_gifs/dance1_corr.mp4"
-repo_id = "PAIR/Text2Video-Zero"
-video_path = hf_hub_download(repo_type="space", repo_id=repo_id, filename=filename)
-
-reader = imageio.get_reader(video_path, "ffmpeg")
-frame_count = 8
-pose_images = [Image.fromarray(reader.get_data(i)) for i in range(frame_count)]
-```
-
-Load a [`ControlNetModel`] for canny edge and a checkpoint into the [`StableDiffusionControlNetPipeline`]. Then you'll use the [`~pipelines.text_to_video_synthesis.pipeline_text_to_video_zero.CrossFrameAttnProcessor`] for the UNet and ControlNet.
-
-```py
-import torch
-from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
-from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero import CrossFrameAttnProcessor
-
-model_id = "stable-diffusion-v1-5/stable-diffusion-v1-5"
-controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16)
-pipeline = StableDiffusionControlNetPipeline.from_pretrained(
-    model_id, controlnet=controlnet, torch_dtype=torch.float16
-).to("cuda")
-
-pipeline.unet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
-pipeline.controlnet.set_attn_processor(CrossFrameAttnProcessor(batch_size=2))
-```
-
-Fix the latents for all the frames, and then pass your prompt and extracted edge images to the model to generate a video.
-
-```py
-latents = torch.randn((1, 4, 64, 64), device="cuda", dtype=torch.float16).repeat(len(pose_images), 1, 1, 1)
-
-prompt = "Darth Vader dancing in a desert"
-result = pipeline(prompt=[prompt] * len(pose_images), image=pose_images, latents=latents).images
-imageio.mimsave("video.mp4", result, fps=4)
-```
-
-</hfoption>
-<hfoption id="InstructPix2Pix">
-
-InstructPix2Pix allows you to use text to describe the changes you want to make to the video. Start by downloading and reading a video.
-
-```py
-from huggingface_hub import hf_hub_download
-from PIL import Image
-import imageio
-
-filename = "__assets__/pix2pix video/camel.mp4"
-repo_id = "PAIR/Text2Video-Zero"
-video_path = hf_hub_download(repo_type="space", repo_id=repo_id, filename=filename)
-
-reader = imageio.get_reader(video_path, "ffmpeg")
-frame_count = 8
-video = [Image.fromarray(reader.get_data(i)) for i in range(frame_count)]
-```
-
-Load the [`StableDiffusionInstructPix2PixPipeline`] and set the [`~pipelines.text_to_video_synthesis.pipeline_text_to_video_zero.CrossFrameAttnProcessor`] for the UNet.
-
-```py
-import torch
-from diffusers import StableDiffusionInstructPix2PixPipeline
-from diffusers.pipelines.text_to_video_synthesis.pipeline_text_to_video_zero import CrossFrameAttnProcessor
-
-pipeline = StableDiffusionInstructPix2PixPipeline.from_pretrained("timbrooks/instruct-pix2pix", torch_dtype=torch.float16).to("cuda")
-pipeline.unet.set_attn_processor(CrossFrameAttnProcessor(batch_size=3))
-```
-
-Pass a prompt describing the change you want to apply to the video.
-
-```py
-prompt = "make it Van Gogh Starry Night style"
-result = pipeline(prompt=[prompt] * len(video), image=video).images
-imageio.mimsave("edited_video.mp4", result, fps=4)
-```
-
-</hfoption>
-</hfoptions>
-
-## Optimize
-
-Video generation requires a lot of memory because you're generating many video frames at once. You can reduce your memory requirements at the expense of some inference speed. Try:
-
-1. offloading pipeline components that are no longer needed to the CPU
-2. feed-forward chunking runs the feed-forward layer in a loop instead of all at once
-3. break up the number of frames the VAE has to decode into chunks instead of decoding them all at once
-
-```diff
-- pipeline.enable_model_cpu_offload()
-- frames = pipeline(image, decode_chunk_size=8, generator=generator).frames[0]
-+ pipeline.enable_model_cpu_offload()
-+ pipeline.unet.enable_forward_chunking()
-+ frames = pipeline(image, decode_chunk_size=2, generator=generator, num_frames=25).frames[0]
-```
-
-If memory is not an issue and you want to optimize for speed, try wrapping the UNet with [`torch.compile`](../optimization/torch2.0#torchcompile).
-
-```diff
-- pipeline.enable_model_cpu_offload()
-+ pipeline.to("cuda")
-+ pipeline.unet = torch.compile(pipeline.unet, mode="reduce-overhead", fullgraph=True)
-```
-
-## Quantization
-
-Quantization helps reduce the memory requirements of very large models by storing model weights in a lower precision data type. However, quantization may have varying impact on video quality depending on the video model.
-
-Refer to the [Quantization](../../quantization/overview) to learn more about supported quantization backends (bitsandbytes, torchao, gguf) and selecting a quantization backend that supports your use case.
+video = pipeline(
+  prompt=prompt,
+  guidance_scale=6,
+  num_inference_steps=50
+).frames[0]
+export_to_video(video, "output.mp4", fps=8)
+```

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

@@ -10,109 +10,56 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# Textual inversion
+# Textual Inversion
 
-[[open-in-colab]]
+[Textual Inversion](https://huggingface.co/papers/2208.01618) is a method for generating personalized images of a concept. It works by fine-tuning a models word embeddings on 3-5 images of the concept (for example, pixel art) that is associated with a unique token (`<sks>`). This allows you to use the `<sks>` token in your prompt to trigger the model to generate pixel art images.
 
-The [`StableDiffusionPipeline`] supports textual inversion, a technique that enables a model like Stable Diffusion to learn a new concept from just a few sample images. This gives you more control over the generated images and allows you to tailor the model towards specific concepts. You can get started quickly with a collection of community created concepts in the [Stable Diffusion Conceptualizer](https://huggingface.co/spaces/sd-concepts-library/stable-diffusion-conceptualizer).
-
-This guide will show you how to run inference with textual inversion using a pre-learned concept from the Stable Diffusion Conceptualizer. If you're interested in teaching a model new concepts with textual inversion, take a look at the [Textual Inversion](../training/text_inversion) training guide.
-
-Import the necessary libraries:
+Textual Inversion weights are very lightweight and typically only a few KBs because they're only word embeddings. However, this also means the word embeddings need to be loaded after loading a model with [`~DiffusionPipeline.from_pretrained`].
 
 ```py
 import torch
-from diffusers import StableDiffusionPipeline
-from diffusers.utils import make_image_grid
-```
+from diffusers import AutoPipelineForText2Image
 
-## Stable Diffusion 1 and 2
-
-Pick a Stable Diffusion checkpoint and a pre-learned concept from the [Stable Diffusion Conceptualizer](https://huggingface.co/spaces/sd-concepts-library/stable-diffusion-conceptualizer):
-
-```py
-pretrained_model_name_or_path = "stable-diffusion-v1-5/stable-diffusion-v1-5"
-repo_id_embeds = "sd-concepts-library/cat-toy"
-```
-
-Now you can load a pipeline, and pass the pre-learned concept to it:
-
-```py
-pipeline = StableDiffusionPipeline.from_pretrained(
-    pretrained_model_name_or_path, torch_dtype=torch.float16, use_safetensors=True
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stable-diffusion-v1-5/stable-diffusion-v1-5",
+    torch_dtype=torch.float16
 ).to("cuda")
-
-pipeline.load_textual_inversion(repo_id_embeds)
 ```
 
-Create a prompt with the pre-learned concept by using the special placeholder token `<cat-toy>`, and choose the number of samples and rows of images you'd like to generate:
+Load the word embeddings with [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] and include the unique token in the prompt to activate its generation.
 
 ```py
-prompt = "a grafitti in a favela wall with a <cat-toy> on it"
-
-num_samples_per_row = 2
-num_rows = 2
-```
-
-Then run the pipeline (feel free to adjust the parameters like `num_inference_steps` and `guidance_scale` to see how they affect image quality), save the generated images and visualize them with the helper function you created at the beginning:
-
-```py
-all_images = []
-for _ in range(num_rows):
-    images = pipeline(prompt, num_images_per_prompt=num_samples_per_row, num_inference_steps=50, guidance_scale=7.5).images
-    all_images.extend(images)
-
-grid = make_image_grid(all_images, num_rows, num_samples_per_row)
-grid
+pipeline.load_textual_inversion("sd-concepts-library/gta5-artwork")
+prompt = "A cute brown bear eating a slice of pizza, stunning color scheme, masterpiece, illustration, <gta5-artwork> style"
+pipeline(prompt).images[0]
 ```
 
 <div class="flex justify-center">
-    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/textual_inversion_inference.png">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_txt_embed.png" />
 </div>
 
-## Stable Diffusion XL
+Textual Inversion can also be trained to learn *negative embeddings* to steer generation away from unwanted characteristics such as "blurry" or "ugly". It is useful for improving image quality.
 
-Stable Diffusion XL (SDXL) can also use textual inversion vectors for inference. In contrast to Stable Diffusion 1 and 2, SDXL has two text encoders so you'll need two textual inversion embeddings - one for each text encoder model.
-
-Let's download the SDXL textual inversion embeddings and have a closer look at it's structure:
+EasyNegative is a widely used negative embedding that contains multiple learned negative concepts. Load the negative embeddings and specify the file name and token associated with the negative embeddings. Pass the token to `negative_prompt` in your pipeline to activate it.
 
 ```py
-from huggingface_hub import hf_hub_download
-from safetensors.torch import load_file
-
-file = hf_hub_download("dn118/unaestheticXL", filename="unaestheticXLv31.safetensors")
-state_dict = load_file(file)
-state_dict
-```
-
-```
-{'clip_g': tensor([[ 0.0077, -0.0112,  0.0065,  ...,  0.0195,  0.0159,  0.0275],
-         ...,
-         [-0.0170,  0.0213,  0.0143,  ..., -0.0302, -0.0240, -0.0362]],
- 'clip_l': tensor([[ 0.0023,  0.0192,  0.0213,  ..., -0.0385,  0.0048, -0.0011],
-         ...,
-         [ 0.0475, -0.0508, -0.0145,  ...,  0.0070, -0.0089, -0.0163]],
-```
-
-There are two tensors, `"clip_g"` and `"clip_l"`.
-`"clip_g"` corresponds to the bigger text encoder in SDXL and refers to
-`pipe.text_encoder_2` and `"clip_l"` refers to `pipe.text_encoder`.
-
-Now you can load each tensor separately by passing them along with the correct text encoder and tokenizer
-to [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`]:
-
-```py
-from diffusers import AutoPipelineForText2Image
 import torch
+from diffusers import AutoPipelineForText2Image
 
-pipe = AutoPipelineForText2Image.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", variant="fp16", torch_dtype=torch.float16)
-pipe.to("cuda")
-
-pipe.load_textual_inversion(state_dict["clip_g"], token="unaestheticXLv31", text_encoder=pipe.text_encoder_2, tokenizer=pipe.tokenizer_2)
-pipe.load_textual_inversion(state_dict["clip_l"], token="unaestheticXLv31", text_encoder=pipe.text_encoder, tokenizer=pipe.tokenizer)
-
-# the embedding should be used as a negative embedding, so we pass it as a negative prompt
-generator = torch.Generator().manual_seed(33)
-image = pipe("a woman standing in front of a mountain", negative_prompt="unaestheticXLv31", generator=generator).images[0]
-image
+pipeline = AutoPipelineForText2Image.from_pretrained(
+    "stable-diffusion-v1-5/stable-diffusion-v1-5",
+    torch_dtype=torch.float16
+).to("cuda")
+pipeline.load_textual_inversion(
+    "EvilEngine/easynegative",
+    weight_name="easynegative.safetensors",
+    token="easynegative"
+)
+prompt = "A cute brown bear eating a slice of pizza, stunning color scheme, masterpiece, illustration"
+negative_prompt = "easynegative"
+pipeline(prompt, negative_prompt).images[0]
 ```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/load_neg_embed.png" />
+</div>

docs/source/ko/_toctree.yml

@@ -175,7 +175,7 @@
     - local: optimization/mps
       title: Metal Performance Shaders (MPS)
     - local: optimization/habana
-      title: Habana Gaudi
+      title: Intel Gaudi
     title: 최적화된 하드웨어
   title: 추론 가속화와 메모리 줄이기
 - sections:

docs/source/ko/api/pipelines/stable_diffusion/stable_diffusion_xl.md

@@ -12,7 +12,7 @@ specific language governing permissions and limitations under the License.
 
 # Stable diffusion XL
 
-Stable Diffusion XL은 Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna, Robin Rombach에 의해 [SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis](https://arxiv.org/abs/2307.01952)에서 제안되었습니다.
+Stable Diffusion XL은 Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna, Robin Rombach에 의해 [SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis](https://huggingface.co/papers/2307.01952)에서 제안되었습니다.
 
 논문 초록은 다음을 따릅니다:
 
@@ -125,7 +125,7 @@ image = pipe(prompt=prompt, image=init_image, mask_image=mask_image, num_inferen
 
 refiner를 사용할 때, 쉽게 사용할 수 있습니다
 - 1.) base 모델과 refiner을 사용하는데, 이는 *Denoisers의 앙상블*을 위한 첫 번째 제안된 [eDiff-I](https://research.nvidia.com/labs/dir/eDiff-I/)를 사용하거나
-- 2.) base 모델을 거친 후 [SDEdit](https://arxiv.org/abs/2108.01073) 방법으로 단순하게 refiner를 실행시킬 수 있습니다.
+- 2.) base 모델을 거친 후 [SDEdit](https://huggingface.co/papers/2108.01073) 방법으로 단순하게 refiner를 실행시킬 수 있습니다.
 
 **참고**: SD-XL base와 refiner를 앙상블로 사용하는 아이디어는 커뮤니티 기여자들이 처음으로 제안했으며, 이는 다음과 같은 `diffusers`를 구현하는 데도 도움을 주셨습니다.
 - [SytanSD](https://github.com/SytanSD)

docs/source/ko/conceptual/ethical_guidelines.md

@@ -55,7 +55,7 @@ Diffusers 커뮤니티는 프로젝트의 개발에 다음과 같은 윤리 지
 
 - **배포에서의 안전 유도**
 
-  - [**안전한 Stable Diffusion**](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_safe): 이는 필터되지 않은 웹 크롤링 데이터셋으로 훈련된 Stable Diffusion과 같은 모델이 부적절한 변질에 취약한 문제를 완화합니다. 관련 논문: [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://arxiv.org/abs/2211.05105).
+  - [**안전한 Stable Diffusion**](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_safe): 이는 필터되지 않은 웹 크롤링 데이터셋으로 훈련된 Stable Diffusion과 같은 모델이 부적절한 변질에 취약한 문제를 완화합니다. 관련 논문: [Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models](https://huggingface.co/papers/2211.05105).
 
   - [**안전 검사기**](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/safety_checker.py): 이미지가 생성된 후에 이미자가 임베딩 공간에서 일련의 하드코딩된 유해 개념의 클래스일 확률을 확인하고 비교합니다. 유해 개념은 역공학을 방지하기 위해 의도적으로 숨겨져 있습니다.
 

docs/source/ko/conceptual/evaluation.md

@@ -111,7 +111,7 @@ images = sd_pipeline(sample_prompts, num_images_per_prompt=1, generator=generato
 
 ### 텍스트 안내 이미지 생성[[text-guided-image-generation]]
 
-[CLIP 점수](https://arxiv.org/abs/2104.08718)는 이미지-캡션 쌍의 호환성을 측정합니다. 높은 CLIP 점수는 높은 호환성🔼을 나타냅니다. CLIP 점수는 이미지와 캡션 사이의 의미적 유사성으로 생각할 수도 있습니다. CLIP 점수는 인간 판단과 높은 상관관계를 가지고 있습니다.
+[CLIP 점수](https://huggingface.co/papers/2104.08718)는 이미지-캡션 쌍의 호환성을 측정합니다. 높은 CLIP 점수는 높은 호환성🔼을 나타냅니다. CLIP 점수는 이미지와 캡션 사이의 의미적 유사성으로 생각할 수도 있습니다. CLIP 점수는 인간 판단과 높은 상관관계를 가지고 있습니다.
 
 [`StableDiffusionPipeline`]을 일단 로드해봅시다:
 
@@ -207,7 +207,7 @@ print(f"CLIP Score with v-1-5: {sd_clip_score_1_5}")
 
 ![edit-instruction](https://huggingface.co/datasets/diffusers/docs-images/resolve/main/evaluation_diffusion_models/edit-instruction.png)
 
-모델을 평가하는 한 가지 전략은 두 이미지 캡션 간의 변경과([CLIP-Guided Domain Adaptation of Image Generators](https://arxiv.org/abs/2108.00946)에서 보여줍니다) 함께 두 이미지 사이의 변경의 일관성을 측정하는 것입니다 ([CLIP](https://huggingface.co/docs/transformers/model_doc/clip) 공간에서). 이를 "**CLIP 방향성 유사성**"이라고 합니다.
+모델을 평가하는 한 가지 전략은 두 이미지 캡션 간의 변경과([CLIP-Guided Domain Adaptation of Image Generators](https://huggingface.co/papers/2108.00946)에서 보여줍니다) 함께 두 이미지 사이의 변경의 일관성을 측정하는 것입니다 ([CLIP](https://huggingface.co/docs/transformers/model_doc/clip) 공간에서). 이를 "**CLIP 방향성 유사성**"이라고 합니다.
 
 - 캡션 1은 편집할 이미지 (이미지 1)에 해당합니다.
 - 캡션 2는 편집된 이미지 (이미지 2)에 해당합니다. 편집 지시를 반영해야 합니다.
@@ -417,7 +417,7 @@ CLIP 점수와 CLIP 방향 유사성 모두 CLIP 모델에 의존하기 때문
 
 ### 클래스 조건화 이미지 생성[[class-conditioned-image-generation]]
 
-클래스 조건화 생성 모델은 일반적으로 [ImageNet-1k](https://huggingface.co/datasets/imagenet-1k)와 같은 클래스 레이블이 지정된 데이터셋에서 사전 훈련됩니다. 이러한 모델을 평가하는 인기있는 지표에는 Fréchet Inception Distance (FID), Kernel Inception Distance (KID) 및 Inception Score (IS)가 있습니다. 이 문서에서는 FID ([Heusel et al.](https://arxiv.org/abs/1706.08500))에 초점을 맞추고 있습니다. [`DiTPipeline`](https://huggingface.co/docs/diffusers/api/pipelines/dit)을 사용하여 FID를 계산하는 방법을 보여줍니다. 이는 내부적으로 [DiT 모델](https://arxiv.org/abs/2212.09748)을 사용합니다.
+클래스 조건화 생성 모델은 일반적으로 [ImageNet-1k](https://huggingface.co/datasets/imagenet-1k)와 같은 클래스 레이블이 지정된 데이터셋에서 사전 훈련됩니다. 이러한 모델을 평가하는 인기있는 지표에는 Fréchet Inception Distance (FID), Kernel Inception Distance (KID) 및 Inception Score (IS)가 있습니다. 이 문서에서는 FID ([Heusel et al.](https://huggingface.co/papers/1706.08500))에 초점을 맞추고 있습니다. [`DiTPipeline`](https://huggingface.co/docs/diffusers/api/pipelines/dit)을 사용하여 FID를 계산하는 방법을 보여줍니다. 이는 내부적으로 [DiT 모델](https://huggingface.co/papers/2212.09748)을 사용합니다.
 
 FID는 두 개의 이미지 데이터셋이 얼마나 유사한지를 측정하는 것을 목표로 합니다. [이 자료](https://mmgeneration.readthedocs.io/en/latest/quick_run.html#fid)에 따르면: