update

* Hunyuan Team: add HunyuanDiT related updates


---------

Co-authored-by: XCLiu <liuxc1996@gmail.com>
Co-authored-by: yiyixuxu <yixu310@gmail.com>

.github/workflows/benchmark.yml

@@ -39,7 +39,7 @@ jobs:
           python utils/print_env.py
       - name: Diffusers Benchmarking
         env:
-            HUGGING_FACE_HUB_TOKEN: ${{ secrets.DIFFUSERS_BOT_TOKEN }}
+            HF_TOKEN: ${{ secrets.DIFFUSERS_BOT_TOKEN }}
             BASE_PATH: benchmark_outputs
         run: |
           export TOTAL_GPU_MEMORY=$(python -c "import torch; print(torch.cuda.get_device_properties(0).total_memory / (1024**3))")

.github/workflows/build_docker_images.yml

@@ -25,17 +25,17 @@ jobs:
     steps:
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v1
-      
+
       - name: Check out code
         uses: actions/checkout@v3
-      
+
       - name: Find Changed Dockerfiles
         id: file_changes
         uses: jitterbit/get-changed-files@v1
         with:
           format: 'space-delimited'
           token: ${{ secrets.GITHUB_TOKEN }}
-      
+
       - name: Build Changed Docker Images
         run: |
           CHANGED_FILES="${{ steps.file_changes.outputs.all }}"
@@ -52,7 +52,7 @@ jobs:
   build-and-push-docker-images:
     runs-on: [ self-hosted, intel-cpu, 8-cpu, ci ]
     if: github.event_name != 'pull_request'
-    
+
     permissions:
       contents: read
       packages: write
@@ -69,6 +69,7 @@ jobs:
           - diffusers-flax-tpu
           - diffusers-onnxruntime-cpu
           - diffusers-onnxruntime-cuda
+          - diffusers-doc-builder
 
     steps:
       - name: Checkout repository
@@ -90,24 +91,11 @@ jobs:
 
       - name: Post to a Slack channel
         id: slack
-        uses: slackapi/slack-github-action@6c661ce58804a1a20f6dc5fbee7f0381b469e001
+        uses: huggingface/hf-workflows/.github/actions/post-slack@main
         with:
           # Slack channel id, channel name, or user id to post message.
           # See also: https://api.slack.com/methods/chat.postMessage#channels
-          channel-id: ${{ env.CI_SLACK_CHANNEL }}
-          # For posting a rich message using Block Kit
-          payload: |
-            {
-              "text": "${{ matrix.image-name }} Docker Image build result: ${{ job.status }}\n${{ github.event.head_commit.url }}",
-              "blocks": [
-                {
-                  "type": "section",
-                  "text": {
-                    "type": "mrkdwn",
-                    "text": "${{ matrix.image-name }} Docker Image build result: ${{ job.status }}\n${{ github.event.head_commit.url }}"
-                  }
-                }
-              ]
-            }
-        env:
-          SLACK_BOT_TOKEN: ${{ secrets.SLACK_CIFEEDBACK_BOT_TOKEN }}
+          slack_channel: ${{ env.CI_SLACK_CHANNEL }}
+          title: "🤗 Results of the ${{ matrix.image-name }} Docker Image build"
+          status: ${{ job.status }}
+          slack_token: ${{ secrets.SLACK_CIFEEDBACK_BOT_TOKEN }}

.github/workflows/build_documentation.yml

@@ -21,7 +21,7 @@ jobs:
       package: diffusers
       notebook_folder: diffusers_doc
       languages: en ko zh ja pt
-
+      custom_container: diffusers/diffusers-doc-builder
     secrets:
       token: ${{ secrets.HUGGINGFACE_PUSH }}
       hf_token: ${{ secrets.HF_DOC_BUILD_PUSH }}

.github/workflows/build_pr_documentation.yml

@@ -20,3 +20,4 @@ jobs:
       install_libgl1: true
       package: diffusers
       languages: en ko zh ja pt
+      custom_container: diffusers/diffusers-doc-builder

.github/workflows/nightly_tests.yml

@@ -59,7 +59,7 @@ jobs:
     runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-pytorch-cuda
-      options: --shm-size "16gb" --ipc host -v /mnt/hf_cache:/mnt/cache/ --gpus 0
+      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface/diffusers:/mnt/cache/ --gpus 0
     steps:
       - name: Checkout diffusers
         uses: actions/checkout@v3
@@ -81,7 +81,7 @@ jobs:
 
       - name: Nightly PyTorch CUDA checkpoint (pipelines) tests
         env:
-          HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
           # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
           CUBLAS_WORKSPACE_CONFIG: :16:8
         run: |
@@ -112,7 +112,7 @@ jobs:
 
   run_nightly_tests_for_other_torch_modules:
     name: Torch Non-Pipelines CUDA Nightly Tests
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-pytorch-cuda
       options: --shm-size "16gb" --ipc host -v /mnt/hf_cache:/mnt/cache/ --gpus 0
@@ -141,7 +141,7 @@ jobs:
     - name: Run nightly PyTorch CUDA tests for non-pipeline modules
       if: ${{ matrix.module != 'examples'}}
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
         # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
         CUBLAS_WORKSPACE_CONFIG: :16:8
       run: |
@@ -154,7 +154,7 @@ jobs:
     - name: Run nightly example tests with Torch
       if: ${{ matrix.module == 'examples' }}
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
         # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
         CUBLAS_WORKSPACE_CONFIG: :16:8
       run: |
@@ -185,7 +185,7 @@ jobs:
 
   run_lora_nightly_tests:
     name: Nightly LoRA Tests with PEFT and TORCH
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-pytorch-cuda
       options: --shm-size "16gb" --ipc host -v /mnt/hf_cache:/mnt/cache/ --gpus 0
@@ -211,7 +211,7 @@ jobs:
 
     - name: Run nightly LoRA tests with PEFT and Torch
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
         # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
         CUBLAS_WORKSPACE_CONFIG: :16:8
       run: |
@@ -269,7 +269,7 @@ jobs:
 
     - name: Run nightly Flax TPU tests
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m pytest -n 0 \
           -s -v -k "Flax" \
@@ -298,7 +298,7 @@ jobs:
 
   run_nightly_onnx_tests:
     name: Nightly ONNXRuntime CUDA tests on Ubuntu
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-onnxruntime-cuda
       options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/hf_cache:/mnt/cache/
@@ -324,7 +324,7 @@ jobs:
 
     - name: Run nightly ONNXRuntime CUDA tests
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m pytest -n 1 --max-worker-restart=0 --dist=loadfile \
           -s -v -k "Onnx" \
@@ -390,7 +390,7 @@ jobs:
         shell: arch -arch arm64 bash {0}
         env:
           HF_HOME: /System/Volumes/Data/mnt/cache
-          HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
         run: |
           ${CONDA_RUN} python -m pytest -n 1 -s -v --make-reports=tests_torch_mps \
             --report-log=tests_torch_mps.log \

.github/workflows/pr_test_fetcher.yml

@@ -15,7 +15,7 @@ concurrency:
 jobs:
   setup_pr_tests:
     name: Setup PR Tests
-    runs-on: docker-cpu
+    runs-on: [ self-hosted, intel-cpu, 8-cpu, ci ]
     container:
       image: diffusers/diffusers-pytorch-cpu
       options: --shm-size "16gb" --ipc host -v /mnt/hf_cache:/mnt/cache/
@@ -73,7 +73,7 @@ jobs:
       max-parallel: 2
       matrix:
         modules: ${{ fromJson(needs.setup_pr_tests.outputs.matrix) }}
-    runs-on: docker-cpu
+    runs-on: [ self-hosted, intel-cpu, 8-cpu, ci ]
     container:
       image: diffusers/diffusers-pytorch-cpu
       options: --shm-size "16gb" --ipc host -v /mnt/hf_cache:/mnt/cache/
@@ -123,7 +123,7 @@ jobs:
         config:
           - name: Hub tests for models, schedulers, and pipelines
             framework: hub_tests_pytorch
-            runner: docker-cpu
+            runner: [ self-hosted, intel-cpu, 8-cpu, ci ]
             image: diffusers/diffusers-pytorch-cpu
             report: torch_hub
 

.github/workflows/pr_tests.yml

@@ -156,7 +156,7 @@ jobs:
       if: ${{ matrix.config.framework == 'pytorch_examples' }}
       run: |
         python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
-        python -m uv pip install peft
+        python -m uv pip install peft timm
         python -m pytest -n 4 --max-worker-restart=0 --dist=loadfile \
           --make-reports=tests_${{ matrix.config.report }} \
           examples

.github/workflows/push_tests.yml

@@ -31,14 +31,13 @@ jobs:
         uses: actions/checkout@v3
         with:
           fetch-depth: 2
-      - name: Set up Python
-        uses: actions/setup-python@v4
-        with:
-          python-version: "3.8"
       - name: Install dependencies
         run: |
           python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
           python -m uv pip install -e [quality,test]
+      - name: Environment
+        run: |
+          python utils/print_env.py
       - name: Fetch Pipeline Matrix
         id: fetch_pipeline_matrix
         run: |
@@ -57,12 +56,13 @@ jobs:
     needs: setup_torch_cuda_pipeline_matrix
     strategy:
       fail-fast: false
+      max-parallel: 8
       matrix:
         module: ${{ fromJson(needs.setup_torch_cuda_pipeline_matrix.outputs.pipeline_test_matrix) }}
     runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-pytorch-cuda
-      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/ --gpus 0 --privileged
+      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface/diffusers:/mnt/cache/ --gpus 0
     steps:
       - name: Checkout diffusers
         uses: actions/checkout@v3
@@ -71,12 +71,6 @@ jobs:
       - name: NVIDIA-SMI
         run: |
           nvidia-smi
-      - name: Tailscale
-        uses: huggingface/tailscale-action@v1
-        with:
-          authkey: ${{ secrets.TAILSCALE_SSH_AUTHKEY }}
-          slackChannel: ${{ secrets.SLACK_CIFEEDBACK_CHANNEL }}
-          slackToken: ${{ secrets.SLACK_CIFEEDBACK_BOT_TOKEN }}
       - name: Install dependencies
         run: |
           python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
@@ -87,7 +81,7 @@ jobs:
           python utils/print_env.py
       - name: Slow PyTorch CUDA checkpoint tests on Ubuntu
         env:
-          HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
           # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
           CUBLAS_WORKSPACE_CONFIG: :16:8
         run: |
@@ -95,18 +89,11 @@ jobs:
             -s -v -k "not Flax and not Onnx" \
             --make-reports=tests_pipeline_${{ matrix.module }}_cuda \
             tests/pipelines/${{ matrix.module }}
-      - name: Tailscale Wait
-        if: ${{ failure() || runner.debug == '1' }}
-        uses: huggingface/tailscale-action@v1
-        with:
-           waitForSSH: true
-           authkey: ${{ secrets.TAILSCALE_SSH_AUTHKEY }}
       - name: Failure short reports
         if: ${{ failure() }}
         run: |
           cat reports/tests_pipeline_${{ matrix.module }}_cuda_stats.txt
           cat reports/tests_pipeline_${{ matrix.module }}_cuda_failures_short.txt
-
       - name: Test suite reports artifacts
         if: ${{ always() }}
         uses: actions/upload-artifact@v2
@@ -116,16 +103,16 @@ jobs:
 
   torch_cuda_tests:
     name: Torch CUDA Tests
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-pytorch-cuda
-      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/ --gpus 0
+      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface/diffusers:/mnt/cache/ --gpus 0
     defaults:
       run:
         shell: bash
     strategy:
       matrix:
-        module: [models, schedulers, lora, others]
+        module: [models, schedulers, lora, others, single_file]
     steps:
     - name: Checkout diffusers
       uses: actions/checkout@v3
@@ -144,7 +131,7 @@ jobs:
 
     - name: Run slow PyTorch CUDA tests
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
         # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
         CUBLAS_WORKSPACE_CONFIG: :16:8
       run: |
@@ -168,10 +155,10 @@ jobs:
 
   peft_cuda_tests:
     name: PEFT CUDA Tests
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-pytorch-cuda
-      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/ --gpus 0
+      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface/diffusers:/mnt/cache/ --gpus 0
     defaults:
       run:
         shell: bash
@@ -194,7 +181,7 @@ jobs:
 
     - name: Run slow PEFT CUDA tests
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
         # https://pytorch.org/docs/stable/notes/randomness.html#avoiding-nondeterministic-algorithms
         CUBLAS_WORKSPACE_CONFIG: :16:8
       run: |
@@ -243,7 +230,7 @@ jobs:
 
     - name: Run slow Flax TPU tests
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m pytest -n 0 \
           -s -v -k "Flax" \
@@ -265,7 +252,7 @@ jobs:
 
   onnx_cuda_tests:
     name: ONNX CUDA Tests
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
     container:
       image: diffusers/diffusers-onnxruntime-cuda
       options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/ --gpus 0
@@ -290,7 +277,7 @@ jobs:
 
     - name: Run slow ONNXRuntime CUDA tests
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m pytest -n 1 --max-worker-restart=0 --dist=loadfile \
           -s -v -k "Onnx" \
@@ -313,7 +300,7 @@ jobs:
   run_torch_compile_tests:
     name: PyTorch Compile CUDA tests
 
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
 
     container:
       image: diffusers/diffusers-pytorch-compile-cuda
@@ -337,7 +324,7 @@ jobs:
         python utils/print_env.py
     - name: Run example tests on GPU
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m pytest -n 1 --max-worker-restart=0 --dist=loadfile -s -v -k "compile" --make-reports=tests_torch_compile_cuda tests/
     - name: Failure short reports
@@ -354,7 +341,7 @@ jobs:
   run_xformers_tests:
     name: PyTorch xformers CUDA tests
 
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
 
     container:
       image: diffusers/diffusers-pytorch-xformers-cuda
@@ -378,7 +365,7 @@ jobs:
         python utils/print_env.py
     - name: Run example tests on GPU
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m pytest -n 1 --max-worker-restart=0 --dist=loadfile -s -v -k "xformers" --make-reports=tests_torch_xformers_cuda tests/
     - name: Failure short reports
@@ -395,7 +382,7 @@ jobs:
   run_examples_tests:
     name: Examples PyTorch CUDA tests on Ubuntu
 
-    runs-on: docker-gpu
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
 
     container:
       image: diffusers/diffusers-pytorch-cuda
@@ -423,9 +410,10 @@ jobs:
 
     - name: Run example tests on GPU
       env:
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
+        python -m uv pip install timm
         python -m pytest -n 1 --max-worker-restart=0 --dist=loadfile -s -v --make-reports=examples_torch_cuda examples/
 
     - name: Failure short reports

.github/workflows/push_tests_fast.yml

@@ -107,7 +107,7 @@ jobs:
       if: ${{ matrix.config.framework == 'pytorch_examples' }}
       run: |
         python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
-        python -m uv pip install peft
+        python -m uv pip install peft timm
         python -m pytest -n 4 --max-worker-restart=0 --dist=loadfile \
           --make-reports=tests_${{ matrix.config.report }} \
           examples

.github/workflows/push_tests_mps.yml

@@ -23,7 +23,7 @@ concurrency:
 jobs:
   run_fast_tests_apple_m1:
     name: Fast PyTorch MPS tests on MacOS
-    runs-on: [ self-hosted, apple-m1 ]
+    runs-on: macos-13-xlarge
 
     steps:
     - name: Checkout diffusers
@@ -59,7 +59,7 @@ jobs:
       shell: arch -arch arm64 bash {0}
       env:
         HF_HOME: /System/Volumes/Data/mnt/cache
-        HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}
+        HF_TOKEN: ${{ secrets.HF_TOKEN }}
       run: |
         ${CONDA_RUN} python -m pytest -n 0 -s -v --make-reports=tests_torch_mps tests/
 

.github/workflows/run_tests_from_a_pr.yml

@@ -0,0 +1,73 @@
+name: Check running SLOW tests from a PR (only GPU)
+
+on:
+  workflow_dispatch:
+    inputs:
+      docker_image:
+        default: 'diffusers/diffusers-pytorch-cuda'
+        description: 'Name of the Docker image'
+        required: true
+      branch: 
+        description: 'PR Branch to test on'
+        required: true
+      test:
+        description: 'Tests to run (e.g.: `tests/models`).'
+        required: true
+
+env:
+  DIFFUSERS_IS_CI: yes
+  IS_GITHUB_CI: "1"
+  HF_HOME: /mnt/cache
+  OMP_NUM_THREADS: 8
+  MKL_NUM_THREADS: 8
+  PYTEST_TIMEOUT: 600
+  RUN_SLOW: yes
+
+jobs:
+  run_tests:
+    name: "Run a test on our runner from a PR"
+    runs-on: [single-gpu, nvidia-gpu, t4, ci]
+    container:
+      image: ${{ github.event.inputs.docker_image }}
+      options: --gpus 0 --privileged --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+
+    steps:
+      - name: Validate test files input
+        id: validate_test_files
+        env: 
+          PY_TEST: ${{ github.event.inputs.test }}
+        run: |
+          if [[ ! "$PY_TEST" =~ ^tests/ ]]; then
+            echo "Error: The input string must start with 'tests/'."
+            exit 1
+          fi
+          
+          if [[ ! "$PY_TEST" =~ ^tests/(models|pipelines) ]]; then
+            echo "Error: The input string must contain either 'models' or 'pipelines' after 'tests/'."
+            exit 1
+          fi
+          
+          if [[ "$PY_TEST" == *";"* ]]; then
+            echo "Error: The input string must not contain ';'."
+            exit 1
+          fi
+          echo "$PY_TEST"
+
+      - name: Checkout PR branch
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ github.event.inputs.branch }}
+          repository: ${{ github.event.pull_request.head.repo.full_name }}
+
+
+      - name: Install pytest 
+        run: | 
+          python -m venv /opt/venv && export PATH="/opt/venv/bin:$PATH"
+          python -m uv pip install -e [quality,test]
+          python -m uv pip install peft
+      
+      - name: Run tests
+        env: 
+            PY_TEST: ${{ github.event.inputs.test }}
+        run: |
+          pytest "$PY_TEST"

.github/workflows/ssh-runner.yml

@@ -25,7 +25,7 @@ jobs:
     runs-on: [single-gpu, nvidia-gpu, "${{ github.event.inputs.runner_type }}", ci]
     container:
       image: ${{ github.event.inputs.docker_image }}
-      options: --gpus all --privileged --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+      options: --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface/diffusers:/mnt/cache/ --gpus 0 --privileged
 
     steps:
       - name: Checkout diffusers

.github/workflows/update_metadata.yml

@@ -25,6 +25,6 @@ jobs:
 
       - name: Update metadata
         env:
-          HUGGING_FACE_HUB_TOKEN: ${{ secrets.SAYAK_HF_TOKEN }}
+          HF_TOKEN: ${{ secrets.SAYAK_HF_TOKEN }}
         run: |
           python utils/update_metadata.py --commit_sha ${{ github.sha }}

CONTRIBUTING.md

@@ -355,7 +355,7 @@ You will need basic `git` proficiency to be able to contribute to
 manual. Type `git --help` in a shell and enjoy. If you prefer books, [Pro
 Git](https://git-scm.com/book/en/v2) is a very good reference.
 
-Follow these steps to start contributing ([supported Python versions](https://github.com/huggingface/diffusers/blob/main/setup.py#L265)):
+Follow these steps to start contributing ([supported Python versions](https://github.com/huggingface/diffusers/blob/42f25d601a910dceadaee6c44345896b4cfa9928/setup.py#L270)):
 
 1. Fork the [repository](https://github.com/huggingface/diffusers) by
 clicking on the 'Fork' button on the repository's page. This creates a copy of the code

README.md

@@ -77,7 +77,7 @@ Please refer to the [How to use Stable Diffusion in Apple Silicon](https://huggi
 
 ## Quickstart
 
-Generating outputs is super easy with 🤗 Diffusers. To generate an image from text, use the `from_pretrained` method to load any pretrained diffusion model (browse the [Hub](https://huggingface.co/models?library=diffusers&sort=downloads) for 22000+ checkpoints):
+Generating outputs is super easy with 🤗 Diffusers. To generate an image from text, use the `from_pretrained` method to load any pretrained diffusion model (browse the [Hub](https://huggingface.co/models?library=diffusers&sort=downloads) for 25.000+ checkpoints):
 
 ```python
 from diffusers import DiffusionPipeline
@@ -219,7 +219,7 @@ Also, say 👋 in our public Discord channel <a href="https://discord.gg/G7tWnz9
 - https://github.com/deep-floyd/IF
 - https://github.com/bentoml/BentoML
 - https://github.com/bmaltais/kohya_ss
-- +9000 other amazing GitHub repositories 💪
+- +11.000 other amazing GitHub repositories 💪
 
 Thank you for using us ❤️.
 

docker/diffusers-doc-builder/Dockerfile

@@ -0,0 +1,52 @@
+FROM ubuntu:20.04
+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 \
+                   python3.10 \
+                   python3-pip \
+                   libgl1 \
+                   zip \
+                   wget \
+                   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 \
+        --extra-index-url https://download.pytorch.org/whl/cpu && \
+    python3.10 -m uv pip install --no-cache-dir \
+        accelerate \
+        datasets \
+        hf-doc-builder \
+        huggingface-hub \
+        Jinja2 \
+        librosa \
+        numpy \
+        scipy \
+        tensorboard \
+        transformers \
+        matplotlib \
+        setuptools==69.5.1
+
+CMD ["/bin/bash"]

docker/diffusers-flax-cpu/Dockerfile

@@ -4,22 +4,25 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
-                   build-essential \
-                   git \
-                   git-lfs \
-                   curl \
-                   ca-certificates \
-                   libsndfile1-dev \
-                   libgl1 \
-                   python3.8 \
-                   python3-pip \
-                   python3.8-venv && \
+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-pip \
+        python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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)

docker/diffusers-flax-tpu/Dockerfile

@@ -4,8 +4,11 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
@@ -13,13 +16,13 @@ RUN apt update && \
                    ca-certificates \
                    libsndfile1-dev \
                    libgl1 \
-                   python3.8 \
+                   python3.10 \
                    python3-pip \
-                   python3.8-venv && \
+                   python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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)

docker/diffusers-onnxruntime-cpu/Dockerfile

@@ -4,8 +4,11 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
@@ -13,13 +16,13 @@ RUN apt update && \
                    ca-certificates \
                    libsndfile1-dev \
                    libgl1 \
-                   python3.8 \
+                   python3.10 \
                    python3-pip \
-                   python3.8-venv && \
+                   python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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)

docker/diffusers-onnxruntime-cuda/Dockerfile

@@ -4,8 +4,11 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
@@ -13,24 +16,24 @@ RUN apt update && \
                    ca-certificates \
                    libsndfile1-dev \
                    libgl1 \
-                   python3.8 \
+                   python3.10 \
                    python3-pip \
-                   python3.8-venv && \
+                   python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
-    python3 -m uv pip install --no-cache-dir \
+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 \
         "onnxruntime-gpu>=1.13.1" \
         --extra-index-url https://download.pytorch.org/whl/cu117 && \
-    python3 -m uv pip install --no-cache-dir \
+    python3.10 -m uv pip install --no-cache-dir \
         accelerate \
         datasets \
         hf-doc-builder \

docker/diffusers-pytorch-compile-cuda/Dockerfile

@@ -4,8 +4,11 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
@@ -13,24 +16,23 @@ RUN apt update && \
     ca-certificates \
     libsndfile1-dev \
     libgl1 \
-    python3.9 \
-    python3.9-dev \
+    python3.10 \
     python3-pip \
-    python3.9-venv && \
+    python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3.9 -m venv /opt/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 python3.9 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
-    python3.9 -m uv pip install --no-cache-dir \
+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 && \
-    python3.9 -m pip install --no-cache-dir \
+    python3.10 -m pip install --no-cache-dir \
     accelerate \
     datasets \
     hf-doc-builder \

docker/diffusers-pytorch-cpu/Dockerfile

@@ -4,33 +4,36 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
-                   python3.8 \
+                   python3.10 \
                    python3-pip \
                    libgl1 \
-                   python3.8-venv && \
+                   python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
-    python3 -m uv pip install --no-cache-dir \
+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 \
         --extra-index-url https://download.pytorch.org/whl/cpu && \
-    python3 -m uv pip install --no-cache-dir \
+    python3.10 -m uv pip install --no-cache-dir \
         accelerate \
         datasets \
         hf-doc-builder \

docker/diffusers-pytorch-cuda/Dockerfile

@@ -4,8 +4,11 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
@@ -13,23 +16,23 @@ RUN apt update && \
     ca-certificates \
     libsndfile1-dev \
     libgl1 \
-    python3.8 \
+    python3.10 \
     python3-pip \
-    python3.8-venv && \
+    python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
-    python3 -m uv pip install --no-cache-dir \
+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 -m pip install --no-cache-dir \
+    python3.10 -m pip install --no-cache-dir \
     accelerate \
     datasets \
     hf-doc-builder \

docker/diffusers-pytorch-xformers-cuda/Dockerfile

@@ -4,8 +4,11 @@ LABEL repository="diffusers"
 
 ENV DEBIAN_FRONTEND=noninteractive
 
-RUN apt update && \
-    apt install -y bash \
+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 \
@@ -13,23 +16,23 @@ RUN apt update && \
                    ca-certificates \
                    libsndfile1-dev \
                    libgl1 \
-                   python3.8 \
+                   python3.10 \
                    python3-pip \
-                   python3.8-venv && \
+                   python3.10-venv && \
     rm -rf /var/lib/apt/lists
 
 # make sure to use venv
-RUN python3 -m venv /opt/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 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
-    python3 -m pip install --no-cache-dir \
+RUN python3.10 -m pip install --no-cache-dir --upgrade pip uv==0.1.11 && \
+    python3.10 -m pip install --no-cache-dir \
         torch \
         torchvision \
         torchaudio \
         invisible_watermark && \
-    python3 -m uv pip install --no-cache-dir \
+    python3.10 -m uv pip install --no-cache-dir \
         accelerate \
         datasets \
         hf-doc-builder \

docs/README.md

@@ -242,10 +242,10 @@ Here's an example of a tuple return, comprising several objects:
 
 ```
     Returns:
-        `tuple(torch.FloatTensor)` comprising various elements depending on the configuration ([`BertConfig`]) and inputs:
-        - ** loss** (*optional*, returned when `masked_lm_labels` is provided) `torch.FloatTensor` of shape `(1,)` --
+        `tuple(torch.Tensor)` comprising various elements depending on the configuration ([`BertConfig`]) and inputs:
+        - ** loss** (*optional*, returned when `masked_lm_labels` is provided) `torch.Tensor` of shape `(1,)` --
           Total loss is the sum of the masked language modeling loss and the next sequence prediction (classification) loss.
-        - **prediction_scores** (`torch.FloatTensor` of shape `(batch_size, sequence_length, config.vocab_size)`) --
+        - **prediction_scores** (`torch.Tensor` of shape `(batch_size, sequence_length, config.vocab_size)`) --
           Prediction scores of the language modeling head (scores for each vocabulary token before SoftMax).
 ```
 

docs/source/en/_toctree.yml

@@ -29,10 +29,8 @@
     title: Load community pipelines and components
   - local: using-diffusers/schedulers
     title: Load schedulers and models
-  - local: using-diffusers/using_safetensors
-    title: Load safetensors
   - local: using-diffusers/other-formats
-    title: Load different Stable Diffusion formats
+    title: Model files and layouts
   - local: using-diffusers/loading_adapters
     title: Load adapters
   - local: using-diffusers/push_to_hub
@@ -59,6 +57,8 @@
     title: Distributed inference with multiple GPUs
   - local: using-diffusers/merge_loras
     title: Merge LoRAs
+  - local: using-diffusers/scheduler_features
+    title: Scheduler features
   - local: using-diffusers/callback
     title: Pipeline callbacks
   - local: using-diffusers/reusing_seeds
@@ -68,6 +68,10 @@
   - local: using-diffusers/weighted_prompts
     title: Prompt techniques
   title: Inference techniques
+- sections:
+  - local: advanced_inference/outpaint
+    title: Outpainting
+  title: Advanced inference
 - sections:
   - local: using-diffusers/sdxl
     title: Stable Diffusion XL
@@ -81,20 +85,20 @@
     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
     title: DiffEdit
-  - local: using-diffusers/inference_with_lcm_lora
-    title: Latent Consistency Model-LoRA
-  - local: using-diffusers/inference_with_lcm
-    title: Latent Consistency Model
   - local: using-diffusers/inference_with_tcd_lora
     title: Trajectory Consistency Distillation-LoRA
   - local: using-diffusers/svd
     title: Stable Video Diffusion
+  - local: using-diffusers/marigold_usage
+    title: Marigold Computer Vision
   title: Specific pipeline examples
 - sections:
   - local: training/overview
@@ -141,8 +145,6 @@
 - sections:
   - local: optimization/fp16
     title: Speed up inference
-  - local: using-diffusers/distilled_sd
-    title: Distilled Stable Diffusion inference
   - local: optimization/memory
     title: Reduce memory usage
   - local: optimization/torch2.0
@@ -236,6 +238,10 @@
       title: ConsistencyDecoderVAE
     - local: api/models/transformer2d
       title: Transformer2D
+    - local: api/models/pixart_transformer2d
+      title: PixArtTransformer2D
+    - local: api/models/dit_transformer2d
+      title: DiTTransformer2D
     - local: api/models/transformer_temporal
       title: Transformer Temporal
     - local: api/models/prior_transformer
@@ -299,6 +305,8 @@
       title: Latent Diffusion
     - local: api/pipelines/ledits_pp
       title: LEDITS++
+    - local: api/pipelines/marigold
+      title: Marigold
     - local: api/pipelines/panorama
       title: MultiDiffusion
     - local: api/pipelines/musicldm
@@ -309,6 +317,8 @@
       title: Personalized Image Animator (PIA)
     - local: api/pipelines/pixart
       title: PixArt-α
+    - local: api/pipelines/pixart_sigma
+      title: PixArt-Σ
     - local: api/pipelines/self_attention_guidance
       title: Self-Attention Guidance
     - local: api/pipelines/semantic_stable_diffusion
@@ -443,6 +453,8 @@
       title: Utilities
     - local: api/image_processor
       title: VAE Image Processor
+    - local: api/video_processor
+      title: Video Processor
     title: Internal classes
     isExpanded: false
-  title: API
+  title: API

docs/source/en/advanced_inference/outpaint.md

@@ -0,0 +1,231 @@
+<!--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.
+-->
+
+# Outpainting
+
+Outpainting extends an image beyond its original boundaries, allowing you to add, replace, or modify visual elements in an image while preserving the original image. Like [inpainting](../using-diffusers/inpaint), you want to fill the white area (in this case, the area outside of the original image) with new visual elements while keeping the original image (represented by a mask of black pixels). There are a couple of ways to outpaint, such as with a [ControlNet](https://hf.co/blog/OzzyGT/outpainting-controlnet) or with [Differential Diffusion](https://hf.co/blog/OzzyGT/outpainting-differential-diffusion).
+
+This guide will show you how to outpaint with an inpainting model, ControlNet, and a ZoeDepth estimator.
+
+Before you begin, make sure you have the [controlnet_aux](https://github.com/huggingface/controlnet_aux) library installed so you can use the ZoeDepth estimator.
+
+```py
+!pip install -q controlnet_aux
+```
+
+## Image preparation
+
+Start by picking an image to outpaint with and remove the background with a Space like [BRIA-RMBG-1.4](https://hf.co/spaces/briaai/BRIA-RMBG-1.4).
+
+<iframe
+	src="https://briaai-bria-rmbg-1-4.hf.space"
+	frameborder="0"
+	width="850"
+	height="450"
+></iframe>
+
+For example, remove the background from this image of a pair of shoes.
+
+<div class="flex flex-row gap-4">
+  <div class="flex-1">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/original-jordan.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">original image</figcaption>
+  </div>
+  <div class="flex-1">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/no-background-jordan.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">background removed</figcaption>
+  </div>
+</div>
+
+[Stable Diffusion XL (SDXL)](../using-diffusers/sdxl) models work best with 1024x1024 images, but you can resize the image to any size as long as your hardware has enough memory to support it. The transparent background in the image should also be replaced with a white background. Create a function (like the one below) that scales and pastes the image onto a white background.
+
+```py
+import random
+
+import requests
+import torch
+from controlnet_aux import ZoeDetector
+from PIL import Image, ImageOps
+
+from diffusers import (
+    AutoencoderKL,
+    ControlNetModel,
+    StableDiffusionXLControlNetPipeline,
+    StableDiffusionXLInpaintPipeline,
+)
+
+def scale_and_paste(original_image):
+    aspect_ratio = original_image.width / original_image.height
+
+    if original_image.width > original_image.height:
+        new_width = 1024
+        new_height = round(new_width / aspect_ratio)
+    else:
+        new_height = 1024
+        new_width = round(new_height * aspect_ratio)
+
+    resized_original = original_image.resize((new_width, new_height), Image.LANCZOS)
+    white_background = Image.new("RGBA", (1024, 1024), "white")
+    x = (1024 - new_width) // 2
+    y = (1024 - new_height) // 2
+    white_background.paste(resized_original, (x, y), resized_original)
+
+    return resized_original, white_background
+
+original_image = Image.open(
+    requests.get(
+        "https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/no-background-jordan.png",
+        stream=True,
+    ).raw
+).convert("RGBA")
+resized_img, white_bg_image = scale_and_paste(original_image)
+```
+
+To avoid adding unwanted extra details, use the ZoeDepth estimator to provide additional guidance during generation and to ensure the shoes remain consistent with the original image.
+
+```py
+zoe = ZoeDetector.from_pretrained("lllyasviel/Annotators")
+image_zoe = zoe(white_bg_image, detect_resolution=512, image_resolution=1024)
+image_zoe
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/zoedepth-jordan.png"/>
+</div>
+
+## Outpaint
+
+Once your image is ready, you can generate content in the white area around the shoes with [controlnet-inpaint-dreamer-sdxl](https://hf.co/destitech/controlnet-inpaint-dreamer-sdxl), a SDXL ControlNet trained for inpainting.
+
+Load the inpainting ControlNet, ZoeDepth model, VAE and pass them to the [`StableDiffusionXLControlNetPipeline`]. Then you can create an optional `generate_image` function (for convenience) to outpaint an initial image.
+
+```py
+controlnets = [
+    ControlNetModel.from_pretrained(
+        "destitech/controlnet-inpaint-dreamer-sdxl", torch_dtype=torch.float16, variant="fp16"
+    ),
+    ControlNetModel.from_pretrained(
+        "diffusers/controlnet-zoe-depth-sdxl-1.0", torch_dtype=torch.float16
+    ),
+]
+vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16).to("cuda")
+pipeline = StableDiffusionXLControlNetPipeline.from_pretrained(
+    "SG161222/RealVisXL_V4.0", torch_dtype=torch.float16, variant="fp16", controlnet=controlnets, vae=vae
+).to("cuda")
+
+def generate_image(prompt, negative_prompt, inpaint_image, zoe_image, seed: int = None):
+    if seed is None:
+        seed = random.randint(0, 2**32 - 1)
+
+    generator = torch.Generator(device="cpu").manual_seed(seed)
+
+    image = pipeline(
+        prompt,
+        negative_prompt=negative_prompt,
+        image=[inpaint_image, zoe_image],
+        guidance_scale=6.5,
+        num_inference_steps=25,
+        generator=generator,
+        controlnet_conditioning_scale=[0.5, 0.8],
+        control_guidance_end=[0.9, 0.6],
+    ).images[0]
+
+    return image
+
+prompt = "nike air jordans on a basketball court"
+negative_prompt = ""
+
+temp_image = generate_image(prompt, negative_prompt, white_bg_image, image_zoe, 908097)
+```
+
+Paste the original image over the initial outpainted image. You'll improve the outpainted background in a later step.
+
+```py
+x = (1024 - resized_img.width) // 2
+y = (1024 - resized_img.height) // 2
+temp_image.paste(resized_img, (x, y), resized_img)
+temp_image
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/initial-outpaint.png"/>
+</div>
+
+> [!TIP]
+> Now is a good time to free up some memory if you're running low!
+>
+> ```py
+> pipeline=None
+> torch.cuda.empty_cache()
+> ```
+
+Now that you have an initial outpainted image, load the [`StableDiffusionXLInpaintPipeline`] with the [RealVisXL](https://hf.co/SG161222/RealVisXL_V4.0) model to generate the final outpainted image with better quality.
+
+```py
+pipeline = StableDiffusionXLInpaintPipeline.from_pretrained(
+    "OzzyGT/RealVisXL_V4.0_inpainting",
+    torch_dtype=torch.float16,
+    variant="fp16",
+    vae=vae,
+).to("cuda")
+```
+
+Prepare a mask for the final outpainted image. To create a more natural transition between the original image and the outpainted background, blur the mask to help it blend better.
+
+```py
+mask = Image.new("L", temp_image.size)
+mask.paste(resized_img.split()[3], (x, y))
+mask = ImageOps.invert(mask)
+final_mask = mask.point(lambda p: p > 128 and 255)
+mask_blurred = pipeline.mask_processor.blur(final_mask, blur_factor=20)
+mask_blurred
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/blurred-mask.png"/>
+</div>
+
+Create a better prompt and pass it to the `generate_outpaint` function to generate the final outpainted image. Again, paste the original image over the final outpainted background.
+
+```py
+def generate_outpaint(prompt, negative_prompt, image, mask, seed: int = None):
+    if seed is None:
+        seed = random.randint(0, 2**32 - 1)
+
+    generator = torch.Generator(device="cpu").manual_seed(seed)
+
+    image = pipeline(
+        prompt,
+        negative_prompt=negative_prompt,
+        image=image,
+        mask_image=mask,
+        guidance_scale=10.0,
+        strength=0.8,
+        num_inference_steps=30,
+        generator=generator,
+    ).images[0]
+
+    return image
+
+prompt = "high quality photo of nike air jordans on a basketball court, highly detailed"
+negative_prompt = ""
+
+final_image = generate_outpaint(prompt, negative_prompt, temp_image, mask_blurred, 7688778)
+x = (1024 - resized_img.width) // 2
+y = (1024 - resized_img.height) // 2
+final_image.paste(resized_img, (x, y), resized_img)
+final_image
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/final-outpaint.png"/>
+</div>

docs/source/en/api/attnprocessor.md

@@ -55,3 +55,6 @@ An attention processor is a class for applying different types of attention mech
 
 ## XFormersAttnProcessor
 [[autodoc]] models.attention_processor.XFormersAttnProcessor
+
+## AttnProcessorNPU
+[[autodoc]] models.attention_processor.AttnProcessorNPU

docs/source/en/api/image_processor.md

@@ -25,3 +25,11 @@ All pipelines with [`VaeImageProcessor`] accept PIL Image, PyTorch tensor, or Nu
 The [`VaeImageProcessorLDM3D`] accepts RGB and depth inputs and returns RGB and depth outputs.
 
 [[autodoc]] image_processor.VaeImageProcessorLDM3D
+
+## PixArtImageProcessor
+
+[[autodoc]] image_processor.PixArtImageProcessor
+
+## IPAdapterMaskProcessor
+
+[[autodoc]] image_processor.IPAdapterMaskProcessor

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

@@ -10,13 +10,134 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# Single files
+# Loading Pipelines and Models via `from_single_file`
 
-Diffusers supports loading pretrained pipeline (or model) weights stored in a single file, such as a `ckpt` or `safetensors` file. These single file types are typically produced from community trained models. There are three classes for loading single file weights:
+The `from_single_file` method allows you to load supported pipelines using a single checkpoint file as opposed to Diffusers' multiple folders format. This is useful if you are working with Stable Diffusion Web UI's (such as A1111) that rely on a single file format to distribute all the components of a model.
 
-- [`FromSingleFileMixin`] supports loading pretrained pipeline weights stored in a single file, which can either be a `ckpt` or `safetensors` file.
-- [`FromOriginalVAEMixin`] supports loading a pretrained [`AutoencoderKL`] from pretrained ControlNet weights stored in a single file, which can either be a `ckpt` or `safetensors` file.
-- [`FromOriginalControlnetMixin`] supports loading pretrained ControlNet weights stored in a single file, which can either be a `ckpt` or `safetensors` file.
+The `from_single_file` method also supports loading models in their originally distributed format. This means that supported models that have been finetuned with other services can be loaded directly into Diffusers model objects and pipelines.
+
+## Pipelines that currently support `from_single_file` loading
+
+- [`StableDiffusionPipeline`]
+- [`StableDiffusionImg2ImgPipeline`]
+- [`StableDiffusionInpaintPipeline`]
+- [`StableDiffusionControlNetPipeline`]
+- [`StableDiffusionControlNetImg2ImgPipeline`]
+- [`StableDiffusionControlNetInpaintPipeline`]
+- [`StableDiffusionUpscalePipeline`]
+- [`StableDiffusionXLPipeline`]
+- [`StableDiffusionXLImg2ImgPipeline`]
+- [`StableDiffusionXLInpaintPipeline`]
+- [`StableDiffusionXLInstructPix2PixPipeline`]
+- [`StableDiffusionXLControlNetPipeline`]
+- [`StableDiffusionXLKDiffusionPipeline`]
+- [`LatentConsistencyModelPipeline`]
+- [`LatentConsistencyModelImg2ImgPipeline`]
+- [`StableDiffusionControlNetXSPipeline`]
+- [`StableDiffusionXLControlNetXSPipeline`]
+- [`LEditsPPPipelineStableDiffusion`]
+- [`LEditsPPPipelineStableDiffusionXL`]
+- [`PIAPipeline`]
+
+## Models that currently support `from_single_file` loading
+
+- [`UNet2DConditionModel`]
+- [`StableCascadeUNet`]
+- [`AutoencoderKL`]
+- [`ControlNetModel`]
+
+## Usage Examples
+
+## Loading a Pipeline using `from_single_file`
+
+```python
+from diffusers import StableDiffusionXLPipeline
+
+ckpt_path = "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0_0.9vae.safetensors"
+pipe = StableDiffusionXLPipeline.from_single_file(ckpt_path)
+```
+
+## Setting components in a Pipeline using `from_single_file`
+
+Set components of a pipeline by passing them directly to the `from_single_file` method. For example, here we are swapping out the pipeline's default scheduler with the `DDIMScheduler`.
+
+```python
+from diffusers import StableDiffusionXLPipeline, DDIMScheduler
+
+ckpt_path = "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0_0.9vae.safetensors"
+
+scheduler = DDIMScheduler()
+pipe = StableDiffusionXLPipeline.from_single_file(ckpt_path, scheduler=scheduler)
+
+```
+
+Here we are passing in a ControlNet model to the `StableDiffusionControlNetPipeline`.
+
+```python
+from diffusers import StableDiffusionControlNetPipeline, ControlNetModel
+
+ckpt_path = "https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/v1-5-pruned-emaonly.safetensors"
+
+controlnet = ControlNetModel.from_pretrained("lllyasviel/control_v11p_sd15_canny")
+pipe = StableDiffusionControlNetPipeline.from_single_file(ckpt_path, controlnet=controlnet)
+
+```
+
+## Loading a Model using `from_single_file`
+
+```python
+from diffusers import StableCascadeUNet
+
+ckpt_path = "https://huggingface.co/stabilityai/stable-cascade/blob/main/stage_b_lite.safetensors"
+model = StableCascadeUNet.from_single_file(ckpt_path)
+
+```
+
+## Using a Diffusers model repository to configure single file loading
+
+Under the hood, `from_single_file` will try to automatically determine a model repository to use to configure the components of a pipeline. You can also explicitly set the model repository to configure the pipeline with the `config` argument.
+
+```python
+from diffusers import StableDiffusionXLPipeline
+
+ckpt_path = "https://huggingface.co/segmind/SSD-1B/blob/main/SSD-1B.safetensors"
+repo_id = "segmind/SSD-1B"
+
+pipe = StableDiffusionXLPipeline.from_single_file(ckpt_path, config=repo_id)
+
+```
+
+In the example above, since we explicitly passed `repo_id="segmind/SSD-1B"` to the `config` argument, it will use this [configuration file](https://huggingface.co/segmind/SSD-1B/blob/main/unet/config.json) from the `unet` subfolder in `"segmind/SSD-1B"` to configure the `unet` component of the pipeline; Similarly, it will use the `config.json` file from `vae` subfolder to configure the `vae` model, `config.json` file from `text_encoder` folder to configure `text_encoder` and so on.
+
+<Tip>
+
+Most of the time you do not need to explicitly set a `config` argument. `from_single_file` will automatically map the checkpoint to the appropriate model repository. However, this option can be useful in cases where model components in the checkpoint might have been changed from what was originally distributed, or in cases where a checkpoint file might not have the necessary metadata to correctly determine the configuration to use for the pipeline.
+
+</Tip>
+
+## Override configuration options when using single file loading
+
+Override the default model or pipeline configuration options by providing the relevant arguments directly to the `from_single_file` method. Any argument supported by the model or pipeline class can be configured in this way:
+
+### Setting a pipeline configuration option
+
+```python
+from diffusers import StableDiffusionXLInstructPix2PixPipeline
+
+ckpt_path = "https://huggingface.co/stabilityai/cosxl/blob/main/cosxl_edit.safetensors"
+pipe = StableDiffusionXLInstructPix2PixPipeline.from_single_file(ckpt_path, config="diffusers/sdxl-instructpix2pix-768", is_cosxl_edit=True)
+
+```
+
+### Setting a model configuration option
+
+```python
+from diffusers import UNet2DConditionModel
+
+ckpt_path = "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0_0.9vae.safetensors"
+model = UNet2DConditionModel.from_single_file(ckpt_path, upcast_attention=True)
+
+```
 
 <Tip>
 
@@ -24,14 +145,116 @@ To learn more about how to load single file weights, see the [Load different Sta
 
 </Tip>
 
+## Working with local files
+
+As of `diffusers>=0.28.0` the `from_single_file` method will attempt to configure a pipeline or model by first inferring the model type from the keys in the checkpoint file. This inferred model type is then used to determine the appropriate model repository on the Hugging Face Hub to configure the model or pipeline.
+
+For example, any single file checkpoint based on the Stable Diffusion XL base model will use the [`stabilityai/stable-diffusion-xl-base-1.0`](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0) model repository to configure the pipeline.
+
+If you are working in an environment with restricted internet access, it is recommended that you download the config files and checkpoints for the model to your preferred directory and pass the local paths to the `pretrained_model_link_or_path` and `config` arguments of the `from_single_file` method.
+
+```python
+from huggingface_hub import hf_hub_download, snapshot_download
+
+my_local_checkpoint_path = hf_hub_download(
+    repo_id="segmind/SSD-1B",
+    filename="SSD-1B.safetensors"
+)
+
+my_local_config_path = snapshot_download(
+    repo_id="segmind/SSD-1B",
+    allowed_patterns=["*.json", "**/*.json", "*.txt", "**/*.txt"]
+)
+
+pipe = StableDiffusionXLPipeline.from_single_file(my_local_checkpoint_path, config=my_local_config_path, local_files_only=True)
+
+```
+
+By default this will download the checkpoints and config files to the [Hugging Face Hub cache directory](https://huggingface.co/docs/huggingface_hub/en/guides/manage-cache). You can also specify a local directory to download the files to by passing the `local_dir` argument to the `hf_hub_download` and `snapshot_download` functions.
+
+```python
+from huggingface_hub import hf_hub_download, snapshot_download
+
+my_local_checkpoint_path = hf_hub_download(
+    repo_id="segmind/SSD-1B",
+    filename="SSD-1B.safetensors"
+    local_dir="my_local_checkpoints"
+)
+
+my_local_config_path = snapshot_download(
+    repo_id="segmind/SSD-1B",
+    allowed_patterns=["*.json", "**/*.json", "*.txt", "**/*.txt"]
+    local_dir="my_local_config"
+)
+
+pipe = StableDiffusionXLPipeline.from_single_file(my_local_checkpoint_path, config=my_local_config_path, local_files_only=True)
+
+```
+
+## Working with local files on file systems that do not support symlinking
+
+By default the `from_single_file` method relies on the `huggingface_hub` caching mechanism to fetch and store checkpoints and config files for models and pipelines. If you are working with a file system that does not support symlinking, it is recommended that you first download the checkpoint file to a local directory and disable symlinking by passing the `local_dir_use_symlink=False` argument to the `hf_hub_download` and `snapshot_download` functions.
+
+```python
+from huggingface_hub import hf_hub_download, snapshot_download
+
+my_local_checkpoint_path = hf_hub_download(
+    repo_id="segmind/SSD-1B",
+    filename="SSD-1B.safetensors"
+    local_dir="my_local_checkpoints",
+    local_dir_use_symlinks=False
+)
+print("My local checkpoint: ", my_local_checkpoint_path)
+
+my_local_config_path = snapshot_download(
+    repo_id="segmind/SSD-1B",
+    allowed_patterns=["*.json", "**/*.json", "*.txt", "**/*.txt"]
+    local_dir_use_symlinks=False,
+)
+print("My local config: ", my_local_config_path)
+
+```
+
+Then pass the local paths to the `pretrained_model_link_or_path` and `config` arguments of the `from_single_file` method.
+
+```python
+pipe = StableDiffusionXLPipeline.from_single_file(my_local_checkpoint_path, config=my_local_config_path, local_files_only=True)
+
+```
+
+<Tip>
+
+As of `huggingface_hub>=0.23.0` the `local_dir_use_symlinks` argument isn't necessary for the `hf_hub_download` and `snapshot_download` functions.
+
+</Tip>
+
+## Using the original configuration file of a model
+
+If you would like to configure the model components in a pipeline using the orignal YAML configuration file, you can pass a local path or url to the original configuration file via the `original_config` argument.
+
+```python
+from diffusers import StableDiffusionXLPipeline
+
+ckpt_path = "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0_0.9vae.safetensors"
+repo_id = "stabilityai/stable-diffusion-xl-base-1.0"
+original_config = "https://raw.githubusercontent.com/Stability-AI/generative-models/main/configs/inference/sd_xl_base.yaml"
+
+pipe = StableDiffusionXLPipeline.from_single_file(ckpt_path, original_config=original_config)
+```
+
+<Tip>
+
+When using `original_config` with `local_files_only=True`, Diffusers will attempt to infer the components of the pipeline based on the type signatures of pipeline class, rather than attempting to fetch the configuration files from a model repository on the Hugging Face Hub. This is to prevent backward breaking changes in existing code that might not be able to connect to the internet to fetch the necessary configuration files.
+
+This is not as reliable as providing a path to a local model repository using the `config` argument and might lead to errors when configuring the pipeline. To avoid this, please run the pipeline with `local_files_only=False` once to download the appropriate pipeline configuration files to the local cache.
+
+</Tip>
+
+
 ## FromSingleFileMixin
 
 [[autodoc]] loaders.single_file.FromSingleFileMixin
 
-## FromOriginalVAEMixin
+## FromOriginalModelMixin
 
-[[autodoc]] loaders.autoencoder.FromOriginalVAEMixin
-
-## FromOriginalControlnetMixin
-
-[[autodoc]] loaders.controlnet.FromOriginalControlNetMixin
+[[autodoc]] loaders.single_file_model.FromOriginalModelMixin

docs/source/en/api/models/dit_transformer2d.md

@@ -0,0 +1,19 @@
+<!--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.
+-->
+
+# DiTTransformer2D
+
+A Transformer model for image-like data from [DiT](https://huggingface.co/papers/2212.09748).
+
+## DiTTransformer2DModel
+
+[[autodoc]] DiTTransformer2DModel

docs/source/en/api/models/pixart_transformer2d.md

@@ -0,0 +1,19 @@
+<!--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.
+-->
+
+# PixArtTransformer2D
+
+A Transformer model for image-like data from [PixArt-Alpha](https://huggingface.co/papers/2310.00426) and [PixArt-Sigma](https://huggingface.co/papers/2403.04692). 
+
+## PixArtTransformer2DModel
+
+[[autodoc]] PixArtTransformer2DModel

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

@@ -24,4 +24,4 @@ The abstract from the paper is:
 
 ## VQEncoderOutput
 
-[[autodoc]] models.vq_model.VQEncoderOutput
+[[autodoc]] models.autoencoders.vq_model.VQEncoderOutput

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

@@ -101,6 +101,53 @@ AnimateDiff tends to work better with finetuned Stable Diffusion models. If you
 
 </Tip>
 
+### AnimateDiffSDXLPipeline
+
+AnimateDiff can also be used with SDXL models. This is currently an experimental feature as only a beta release of the motion adapter checkpoint is available.
+
+```python
+import torch
+from diffusers.models import MotionAdapter
+from diffusers import AnimateDiffSDXLPipeline, DDIMScheduler
+from diffusers.utils import export_to_gif
+
+adapter = MotionAdapter.from_pretrained("guoyww/animatediff-motion-adapter-sdxl-beta", torch_dtype=torch.float16)
+
+model_id = "stabilityai/stable-diffusion-xl-base-1.0"
+scheduler = DDIMScheduler.from_pretrained(
+    model_id,
+    subfolder="scheduler",
+    clip_sample=False,
+    timestep_spacing="linspace",
+    beta_schedule="linear",
+    steps_offset=1,
+)
+pipe = AnimateDiffSDXLPipeline.from_pretrained(
+    model_id,
+    motion_adapter=adapter,
+    scheduler=scheduler,
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
+
+# enable memory savings
+pipe.enable_vae_slicing()
+pipe.enable_vae_tiling()
+
+output = pipe(
+    prompt="a panda surfing in the ocean, realistic, high quality",
+    negative_prompt="low quality, worst quality",
+    num_inference_steps=20,
+    guidance_scale=8,
+    width=1024,
+    height=1024,
+    num_frames=16,
+)
+
+frames = output.frames[0]
+export_to_gif(frames, "animation.gif")
+```
+
 ### AnimateDiffVideoToVideoPipeline
 
 AnimateDiff can also be used to generate visually similar videos or enable style/character/background or other edits starting from an initial video, allowing you to seamlessly explore creative possibilities.
@@ -522,6 +569,12 @@ export_to_gif(frames, "animatelcm-motion-lora.gif")
   - all
   - __call__
 
+## AnimateDiffSDXLPipeline
+
+[[autodoc]] AnimateDiffSDXLPipeline
+  - all
+  - __call__
+
 ## AnimateDiffVideoToVideoPipeline
 
 [[autodoc]] AnimateDiffVideoToVideoPipeline

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

@@ -47,6 +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.
 
 ## I2VGenXLPipeline
 [[autodoc]] I2VGenXLPipeline

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

@@ -0,0 +1,76 @@
+<!--Copyright 2024 Marigold authors and The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Marigold Pipelines for Computer Vision Tasks
+
+![marigold](https://marigoldmonodepth.github.io/images/teaser_collage_compressed.jpg)
+
+Marigold was proposed in [Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation](https://huggingface.co/papers/2312.02145), a CVPR 2024 Oral paper by [Bingxin Ke](http://www.kebingxin.com/), [Anton Obukhov](https://www.obukhov.ai/), [Shengyu Huang](https://shengyuh.github.io/), [Nando Metzger](https://nandometzger.github.io/), [Rodrigo Caye Daudt](https://rcdaudt.github.io/), and [Konrad Schindler](https://scholar.google.com/citations?user=FZuNgqIAAAAJ&hl=en). 
+The idea is to repurpose the rich generative prior of Text-to-Image Latent Diffusion Models (LDMs) for traditional computer vision tasks. 
+Initially, this idea was explored to fine-tune Stable Diffusion for Monocular Depth Estimation, as shown in the teaser above. 
+Later, 
+- [Tianfu Wang](https://tianfwang.github.io/) trained the first Latent Consistency Model (LCM) of Marigold, which unlocked fast single-step inference;
+- [Kevin Qu](https://www.linkedin.com/in/kevin-qu-b3417621b/?locale=en_US) extended the approach to Surface Normals Estimation;
+- [Anton Obukhov](https://www.obukhov.ai/) contributed the pipelines and documentation into diffusers (enabled and supported by [YiYi Xu](https://yiyixuxu.github.io/) and [Sayak Paul](https://sayak.dev/)).
+
+The abstract from the paper is:
+
+*Monocular depth estimation is a fundamental computer vision task. Recovering 3D depth from a single image is geometrically ill-posed and requires scene understanding, so it is not surprising that the rise of deep learning has led to a breakthrough. The impressive progress of monocular depth estimators has mirrored the growth in model capacity, from relatively modest CNNs to large Transformer architectures. Still, monocular depth estimators tend to struggle when presented with images with unfamiliar content and layout, since their knowledge of the visual world is restricted by the data seen during training, and challenged by zero-shot generalization to new domains. This motivates us to explore whether the extensive priors captured in recent generative diffusion models can enable better, more generalizable depth estimation. We introduce Marigold, a method for affine-invariant monocular depth estimation that is derived from Stable Diffusion and retains its rich prior knowledge. The estimator can be fine-tuned in a couple of days on a single GPU using only synthetic training data. It delivers state-of-the-art performance across a wide range of datasets, including over 20% performance gains in specific cases. Project page: https://marigoldmonodepth.github.io.*
+
+## Available Pipelines
+
+Each pipeline supports one Computer Vision task, which takes an input RGB image as input and produces a *prediction* of the modality of interest, such as a depth map of the input image. 
+Currently, the following tasks are implemented:
+
+| Pipeline                                                                                                                                    | Predicted Modalities                                                                                             |                                                                       Demos                                                                        |
+|---------------------------------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------:|
+| [MarigoldDepthPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/marigold/pipeline_marigold_depth.py)     | [Depth](https://en.wikipedia.org/wiki/Depth_map), [Disparity](https://en.wikipedia.org/wiki/Binocular_disparity) | [Fast Demo (LCM)](https://huggingface.co/spaces/prs-eth/marigold-lcm), [Slow Original Demo (DDIM)](https://huggingface.co/spaces/prs-eth/marigold) |
+| [MarigoldNormalsPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/marigold/pipeline_marigold_normals.py) | [Surface normals](https://en.wikipedia.org/wiki/Normal_mapping)                                                  |                                   [Fast Demo (LCM)](https://huggingface.co/spaces/prs-eth/marigold-normals-lcm)                                    |
+
+
+## Available Checkpoints
+
+The original checkpoints can be found under the [PRS-ETH](https://huggingface.co/prs-eth/) Hugging Face organization. 
+
+<Tip>
+
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines. Also, to know more about reducing the memory usage of this pipeline, refer to the ["Reduce memory usage"] section [here](../../using-diffusers/svd#reduce-memory-usage).
+
+</Tip>
+
+<Tip warning={true}>
+
+Marigold pipelines were designed and tested only with `DDIMScheduler` and `LCMScheduler`. 
+Depending on the scheduler, the number of inference steps required to get reliable predictions varies, and there is no universal value that works best across schedulers.
+Because of that, the default value of `num_inference_steps` in the `__call__` method of the pipeline is set to `None` (see the API reference). 
+Unless set explicitly, its value will be taken from the checkpoint configuration `model_index.json`. 
+This is done to ensure high-quality predictions when calling the pipeline with just the `image` argument. 
+
+</Tip>
+
+See also Marigold [usage examples](marigold_usage).
+
+## MarigoldDepthPipeline
+[[autodoc]] MarigoldDepthPipeline
+	- all
+	- __call__
+
+## MarigoldNormalsPipeline
+[[autodoc]] MarigoldNormalsPipeline
+	- all
+	- __call__
+
+## MarigoldDepthOutput
+[[autodoc]] pipelines.marigold.pipeline_marigold_depth.MarigoldDepthOutput
+
+## MarigoldNormalsOutput
+[[autodoc]] pipelines.marigold.pipeline_marigold_normals.MarigoldNormalsOutput

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

@@ -31,7 +31,7 @@ Some notes about this pipeline:
 
 <Tip>
 
-Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading#reuse-components-across-pipelines) section to learn how to efficiently load the same components into multiple pipelines.
+Make sure to check out the Schedulers [guide](../../using-diffusers/schedulers.md) to learn how to explore the tradeoff between scheduler speed and quality, and see the [reuse components across pipelines](../../using-diffusers/loading.md#reuse-a-pipeline) section to learn how to efficiently load the same components into multiple pipelines.
 
 </Tip>
 

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

@@ -0,0 +1,151 @@
+<!--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.
+-->
+
+# PixArt-Σ
+
+![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/pixart/header_collage_sigma.jpg)
+
+[PixArt-Σ: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation](https://huggingface.co/papers/2403.04692) is Junsong Chen, Jincheng Yu, Chongjian Ge, Lewei Yao, Enze Xie, Yue Wu, Zhongdao Wang, James Kwok, Ping Luo, Huchuan Lu, and Zhenguo Li.
+
+The abstract from the paper is:
+
+*In this paper, we introduce PixArt-Σ, a Diffusion Transformer model (DiT) capable of directly generating images at 4K resolution. PixArt-Σ represents a significant advancement over its predecessor, PixArt-α, offering images of markedly higher fidelity and improved alignment with text prompts. A key feature of PixArt-Σ is its training efficiency. Leveraging the foundational pre-training of PixArt-α, it evolves from the ‘weaker’ baseline to a ‘stronger’ model via incorporating higher quality data, a process we term “weak-to-strong training”. The advancements in PixArt-Σ are twofold: (1) High-Quality Training Data: PixArt-Σ incorporates superior-quality image data, paired with more precise and detailed image captions. (2) Efficient Token Compression: we propose a novel attention module within the DiT framework that compresses both keys and values, significantly improving efficiency and facilitating ultra-high-resolution image generation. Thanks to these improvements, PixArt-Σ achieves superior image quality and user prompt adherence capabilities with significantly smaller model size (0.6B parameters) than existing text-to-image diffusion models, such as SDXL (2.6B parameters) and SD Cascade (5.1B parameters). Moreover, PixArt-Σ’s capability to generate 4K images supports the creation of high-resolution posters and wallpapers, efficiently bolstering the production of highquality visual content in industries such as film and gaming.*
+
+You can find the original codebase at [PixArt-alpha/PixArt-sigma](https://github.com/PixArt-alpha/PixArt-sigma) and all the available checkpoints at [PixArt-alpha](https://huggingface.co/PixArt-alpha).
+
+Some notes about this pipeline:
+
+* It uses a Transformer backbone (instead of a UNet) for denoising. As such it has a similar architecture as [DiT](https://hf.co/docs/transformers/model_doc/dit).
+* It was trained using text conditions computed from T5. This aspect makes the pipeline better at following complex text prompts with intricate details.
+* It is good at producing high-resolution images at different aspect ratios. To get the best results, the authors recommend some size brackets which can be found [here](https://github.com/PixArt-alpha/PixArt-sigma/blob/master/diffusion/data/datasets/utils.py).
+* It rivals the quality of state-of-the-art text-to-image generation systems (as of this writing) such as PixArt-α, Stable Diffusion XL, Playground V2.0 and DALL-E 3, while being more efficient than them.
+* It shows the ability of generating super high resolution images, such as 2048px or even 4K.
+* It shows that text-to-image models can grow from a weak model to a stronger one through several improvements (VAEs, datasets, and so on.)
+
+<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>
+
+## Inference with under 8GB GPU VRAM
+
+Run the [`PixArtSigmaPipeline`] with under 8GB GPU VRAM by loading the text encoder in 8-bit precision. Let's walk through a full-fledged example. 
+
+First, install the [bitsandbytes](https://github.com/TimDettmers/bitsandbytes) library:
+
+```bash
+pip install -U bitsandbytes
+```
+
+Then load the text encoder in 8-bit:
+
+```python
+from transformers import T5EncoderModel
+from diffusers import PixArtSigmaPipeline
+import torch
+
+text_encoder = T5EncoderModel.from_pretrained(
+    "PixArt-alpha/PixArt-Sigma-XL-2-1024-MS",
+    subfolder="text_encoder",
+    load_in_8bit=True,
+    device_map="auto",
+
+)
+pipe = PixArtSigmaPipeline.from_pretrained(
+    "PixArt-alpha/PixArt-Sigma-XL-2-1024-MS",
+    text_encoder=text_encoder,
+    transformer=None,
+    device_map="balanced"
+)
+```
+
+Now, use the `pipe` to encode a prompt:
+
+```python
+with torch.no_grad():
+    prompt = "cute cat"
+    prompt_embeds, prompt_attention_mask, negative_embeds, negative_prompt_attention_mask = pipe.encode_prompt(prompt)
+```
+
+Since text embeddings have been computed, remove the `text_encoder` and `pipe` from the memory, and free up som GPU VRAM:
+
+```python
+import gc 
+
+def flush():
+    gc.collect()
+    torch.cuda.empty_cache()
+
+del text_encoder
+del pipe
+flush()
+```
+
+Then compute the latents with the prompt embeddings as inputs:
+
+```python
+pipe = PixArtSigmaPipeline.from_pretrained(
+    "PixArt-alpha/PixArt-Sigma-XL-2-1024-MS",
+    text_encoder=None,
+    torch_dtype=torch.float16,
+).to("cuda")
+
+latents = pipe(
+    negative_prompt=None, 
+    prompt_embeds=prompt_embeds,
+    negative_prompt_embeds=negative_embeds,
+    prompt_attention_mask=prompt_attention_mask,
+    negative_prompt_attention_mask=negative_prompt_attention_mask,
+    num_images_per_prompt=1,
+    output_type="latent",
+).images
+
+del pipe.transformer
+flush()
+```
+
+<Tip>
+
+Notice that while initializing `pipe`, you're setting `text_encoder` to `None` so that it's not loaded.
+
+</Tip>
+
+Once the latents are computed, pass it off to the VAE to decode into a real image:
+
+```python
+with torch.no_grad():
+    image = pipe.vae.decode(latents / pipe.vae.config.scaling_factor, return_dict=False)[0]
+image = pipe.image_processor.postprocess(image, output_type="pil")[0]
+image.save("cat.png")
+```
+
+By deleting components you aren't using and flushing the GPU VRAM, you should be able to run [`PixArtSigmaPipeline`] with under 8GB GPU VRAM.
+
+![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/pixart/8bits_cat.png)
+
+If you want a report of your memory-usage, run this [script](https://gist.github.com/sayakpaul/3ae0f847001d342af27018a96f467e4e).
+
+<Tip warning={true}>
+
+Text embeddings computed in 8-bit can impact the quality of the generated images because of the information loss in the representation space caused by the reduced precision. It's recommended to compare the outputs with and without 8-bit.
+
+</Tip>
+
+While loading the `text_encoder`, you set `load_in_8bit` to `True`. You could also specify `load_in_4bit` to bring your memory requirements down even further to under 7GB.
+
+## PixArtSigmaPipeline
+
+[[autodoc]] PixArtSigmaPipeline
+	- all
+	- __call__
+	

docs/source/en/api/video_processor.md

@@ -0,0 +1,21 @@
+<!--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.
+-->
+
+# Video Processor
+
+The [`VideoProcessor`] provides a unified API for video pipelines to prepare inputs for VAE encoding and post-processing outputs once they're decoded. The class inherits [`VaeImageProcessor`] so it includes transformations such as resizing, normalization, and conversion between PIL Image, PyTorch, and NumPy arrays.
+
+## VideoProcessor
+
+[[autodoc]] video_processor.VideoProcessor.preprocess_video
+
+[[autodoc]] video_processor.VideoProcessor.postprocess_video

docs/source/en/conceptual/philosophy.md

@@ -70,7 +70,7 @@ The following design principles are followed:
 - Pipelines should be used **only** for inference.
 - Pipelines should be very readable, self-explanatory, and easy to tweak.
 - Pipelines should be designed to build on top of each other and be easy to integrate into higher-level APIs.
-- Pipelines are **not** intended to be feature-complete user interfaces. For future complete user interfaces one should rather have a look at [InvokeAI](https://github.com/invoke-ai/InvokeAI), [Diffuzers](https://github.com/abhishekkrthakur/diffuzers), and [lama-cleaner](https://github.com/Sanster/lama-cleaner).
+- Pipelines are **not** intended to be feature-complete user interfaces. For feature-complete user interfaces one should rather have a look at [InvokeAI](https://github.com/invoke-ai/InvokeAI), [Diffuzers](https://github.com/abhishekkrthakur/diffuzers), and [lama-cleaner](https://github.com/Sanster/lama-cleaner).
 - Every pipeline should have one and only one way to run it via a `__call__` method. The naming of the `__call__` arguments should be shared across all pipelines.
 - Pipelines should be named after the task they are intended to solve.
 - In almost all cases, novel diffusion pipelines shall be implemented in a new pipeline folder/file.

docs/source/en/installation.md

@@ -112,7 +112,7 @@ pip install -e ".[flax]"
 
 These commands will link the folder you cloned the repository to and your Python library paths.
 Python will now look inside the folder you cloned to in addition to the normal library paths.
-For example, if your Python packages are typically installed in `~/anaconda3/envs/main/lib/python3.8/site-packages/`, Python will also search the `~/diffusers/` folder you cloned to.
+For example, if your Python packages are typically installed in `~/anaconda3/envs/main/lib/python3.10/site-packages/`, Python will also search the `~/diffusers/` folder you cloned to.
 
 <Tip warning={true}>
 

docs/source/en/optimization/fp16.md

@@ -12,27 +12,23 @@ specific language governing permissions and limitations under the License.
 
 # Speed up inference
 
-There are several ways to optimize 🤗 Diffusers for inference speed. As a general rule of thumb, we recommend using either [xFormers](xformers) or `torch.nn.functional.scaled_dot_product_attention` in PyTorch 2.0 for their memory-efficient attention.
+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.
 
-<Tip>
+> [!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.
 
-In many cases, optimizing for speed or memory leads to improved performance in the other, so you should try to optimize for both whenever you can. This guide focuses on inference speed, but you can learn more about preserving memory in the [Reduce memory usage](memory) guide.
+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.
 
-</Tip>
+| setup    | latency | speed-up |
+|----------|---------|----------|
+| baseline | 5.27s   | x1       |
+| tf32     | 4.14s   | x1.27    |
+| fp16     | 3.51s   | x1.50    |
+| combined | 3.41s   | x1.54    |
 
-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.
+## TensorFloat-32
 
-|                  | 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   |
-
-## Use TensorFloat-32
-
-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 speeds up computations with typically negligible loss in numerical accuracy.
+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.
 
 ```python
 import torch
@@ -40,11 +36,11 @@ import torch
 torch.backends.cuda.matmul.allow_tf32 = True
 ```
 
-You can learn more about TF32 in the [Mixed precision training](https://huggingface.co/docs/transformers/en/perf_train_gpu_one#tf32) guide.
+Learn more about tf32 in the [Mixed precision training](https://huggingface.co/docs/transformers/en/perf_train_gpu_one#tf32) guide.
 
 ## Half-precision weights
 
-To save GPU memory and get more speed, try loading and running the model weights directly in half-precision or float16:
+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
@@ -56,19 +52,76 @@ pipe = DiffusionPipeline.from_pretrained(
     use_safetensors=True,
 )
 pipe = pipe.to("cuda")
-
-prompt = "a photo of an astronaut riding a horse on mars"
-image = pipe(prompt).images[0]
 ```
 
-<Tip warning={true}>
-
-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.
-
-</Tip>
+> [!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. The distilled model is faster and uses less memory while generating images of comparable quality to the full Stable Diffusion 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.
 
-Learn more about in the [Distilled Stable Diffusion inference](../using-diffusers/distilled_sd) guide!
+> [!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.
+
+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).
+
+| 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.
+
+```py
+from diffusers import StableDiffusionPipeline
+import torch
+
+distilled = StableDiffusionPipeline.from_pretrained(
+    "nota-ai/bk-sdm-small", 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
+```
+
+<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>
+
+### Tiny AutoEncoder
+
+To speed inference up even more, replace the autoencoder with a [distilled version](https://huggingface.co/sayakpaul/taesdxl-diffusers) of it.
+
+```py
+import torch
+from diffusers import AutoencoderTiny, StableDiffusionPipeline
+
+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
+```
+
+<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>

docs/source/en/optimization/memory.md

@@ -261,7 +261,7 @@ from dataclasses import dataclass
 
 @dataclass
 class UNet2DConditionOutput:
-    sample: torch.FloatTensor
+    sample: torch.Tensor
 
 
 pipe = StableDiffusionPipeline.from_pretrained(

docs/source/en/optimization/tgate.md

@@ -6,7 +6,7 @@ Before you begin, make sure you install T-GATE.
 
 ```bash
 pip install tgate
-pip install -U pytorch diffusers transformers accelerate DeepCache
+pip install -U torch diffusers transformers accelerate DeepCache
 ```
 
 
@@ -46,12 +46,12 @@ pipe = TgatePixArtLoader(
 
 image = pipe.tgate(
        "An alpaca made of colorful building blocks, cyberpunk.",
-        gate_step=gate_step,
+       gate_step=gate_step,
        num_inference_steps=inference_step,
 ).images[0]
 ```
 </hfoption>
-<hfoption id="Stable Diffusion XL"> 
+<hfoption id="Stable Diffusion XL">
 
 Accelerate `StableDiffusionXLPipeline` with T-GATE:
 
@@ -78,9 +78,9 @@ pipe = TgateSDXLLoader(
 ).to("cuda")
 
 image = pipe.tgate(
-        "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k.",
-        gate_step=gate_step,
-        num_inference_steps=inference_step
+       "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k.",
+       gate_step=gate_step,
+       num_inference_steps=inference_step
 ).images[0]
 ```
 </hfoption>
@@ -111,9 +111,9 @@ pipe = TgateSDXLDeepCacheLoader(
 ).to("cuda")
 
 image = pipe.tgate(
-        "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k.",
-        gate_step=gate_step,
-        num_inference_steps=inference_step
+       "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k.",
+       gate_step=gate_step,
+       num_inference_steps=inference_step
 ).images[0]
 ```
 </hfoption>
@@ -151,9 +151,9 @@ pipe = TgateSDXLLoader(
 ).to("cuda")
 
 image = pipe.tgate(
-        "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k.",
-        gate_step=gate_step,
-        num_inference_steps=inference_step
+       "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k.",
+       gate_step=gate_step,
+       num_inference_steps=inference_step
 ).images[0]
 ```
 </hfoption>

docs/source/en/training/dreambooth.md

@@ -440,6 +440,198 @@ Stable Diffusion XL (SDXL) is a powerful text-to-image model that generates high
 
 The SDXL training script is discussed in more detail in the [SDXL training](sdxl) guide.
 
+## DeepFloyd IF
+
+DeepFloyd IF is a cascading pixel diffusion model with three stages. The first stage generates a base image and the second and third stages progressively upscales the base image into a high-resolution 1024x1024 image. Use the [train_dreambooth_lora.py](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora.py) or [train_dreambooth.py](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth.py) scripts to train a DeepFloyd IF model with LoRA or the full model.
+
+DeepFloyd IF uses predicted variance, but the Diffusers training scripts uses predicted error so the trained DeepFloyd IF models are switched to a fixed variance schedule. The training scripts will update the scheduler config of the fully trained model for you. However, when you load the saved LoRA weights you must also update the pipeline's scheduler config.
+
+```py
+from diffusers import DiffusionPipeline
+
+pipe = DiffusionPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0", use_safetensors=True)
+
+pipe.load_lora_weights("<lora weights path>")
+
+# Update scheduler config to fixed variance schedule
+pipe.scheduler = pipe.scheduler.__class__.from_config(pipe.scheduler.config, variance_type="fixed_small")
+```
+
+The stage 2 model requires additional validation images to upscale. You can download and use a downsized version of the training images for this.
+
+```py
+from huggingface_hub import snapshot_download
+
+local_dir = "./dog_downsized"
+snapshot_download(
+    "diffusers/dog-example-downsized",
+    local_dir=local_dir,
+    repo_type="dataset",
+    ignore_patterns=".gitattributes",
+)
+```
+
+The code samples below provide a brief overview of how to train a DeepFloyd IF model with a combination of DreamBooth and LoRA. Some important parameters to note are:
+
+* `--resolution=64`, a much smaller resolution is required because DeepFloyd IF is a pixel diffusion model and to work on uncompressed pixels, the input images must be smaller
+* `--pre_compute_text_embeddings`, compute the text embeddings ahead of time to save memory because the [`~transformers.T5Model`] can take up a lot of memory
+* `--tokenizer_max_length=77`, you can use a longer default text length with T5 as the text encoder but the default model encoding procedure uses a shorter text length
+* `--text_encoder_use_attention_mask`, to pass the attention mask to the text encoder
+
+<hfoptions id="IF-DreamBooth">
+<hfoption id="Stage 1 LoRA DreamBooth">
+
+Training stage 1 of DeepFloyd IF with LoRA and DreamBooth requires ~28GB of memory.
+
+```bash
+export MODEL_NAME="DeepFloyd/IF-I-XL-v1.0"
+export INSTANCE_DIR="dog"
+export OUTPUT_DIR="dreambooth_dog_lora"
+
+accelerate launch train_dreambooth_lora.py \
+  --report_to wandb \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --instance_prompt="a sks dog" \
+  --resolution=64 \
+  --train_batch_size=4 \
+  --gradient_accumulation_steps=1 \
+  --learning_rate=5e-6 \
+  --scale_lr \
+  --max_train_steps=1200 \
+  --validation_prompt="a sks dog" \
+  --validation_epochs=25 \
+  --checkpointing_steps=100 \
+  --pre_compute_text_embeddings \
+  --tokenizer_max_length=77 \
+  --text_encoder_use_attention_mask
+```
+
+</hfoption>
+<hfoption id="Stage 2 LoRA DreamBooth">
+
+For stage 2 of DeepFloyd IF with LoRA and DreamBooth, pay attention to these parameters:
+
+* `--validation_images`, the images to upscale during validation
+* `--class_labels_conditioning=timesteps`, to additionally conditional the UNet as needed in stage 2
+* `--learning_rate=1e-6`, a lower learning rate is used compared to stage 1
+* `--resolution=256`, the expected resolution for the upscaler
+
+```bash
+export MODEL_NAME="DeepFloyd/IF-II-L-v1.0"
+export INSTANCE_DIR="dog"
+export OUTPUT_DIR="dreambooth_dog_upscale"
+export VALIDATION_IMAGES="dog_downsized/image_1.png dog_downsized/image_2.png dog_downsized/image_3.png dog_downsized/image_4.png"
+
+python train_dreambooth_lora.py \
+    --report_to wandb \
+    --pretrained_model_name_or_path=$MODEL_NAME \
+    --instance_data_dir=$INSTANCE_DIR \
+    --output_dir=$OUTPUT_DIR \
+    --instance_prompt="a sks dog" \
+    --resolution=256 \
+    --train_batch_size=4 \
+    --gradient_accumulation_steps=1 \
+    --learning_rate=1e-6 \ 
+    --max_train_steps=2000 \
+    --validation_prompt="a sks dog" \
+    --validation_epochs=100 \
+    --checkpointing_steps=500 \
+    --pre_compute_text_embeddings \
+    --tokenizer_max_length=77 \
+    --text_encoder_use_attention_mask \
+    --validation_images $VALIDATION_IMAGES \
+    --class_labels_conditioning=timesteps
+```
+
+</hfoption>
+<hfoption id="Stage 1 DreamBooth">
+
+For stage 1 of DeepFloyd IF with DreamBooth, pay attention to these parameters:
+
+* `--skip_save_text_encoder`, to skip saving the full T5 text encoder with the finetuned model
+* `--use_8bit_adam`, to use 8-bit Adam optimizer to save memory due to the size of the optimizer state when training the full model
+* `--learning_rate=1e-7`, a really low learning rate should be used for full model training otherwise the model quality is degraded (you can use a higher learning rate with a larger batch size)
+
+Training with 8-bit Adam and a batch size of 4, the full model can be trained with ~48GB of memory.
+
+```bash
+export MODEL_NAME="DeepFloyd/IF-I-XL-v1.0"
+export INSTANCE_DIR="dog"
+export OUTPUT_DIR="dreambooth_if"
+
+accelerate launch train_dreambooth.py \
+  --pretrained_model_name_or_path=$MODEL_NAME  \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --instance_prompt="a photo of sks dog" \
+  --resolution=64 \
+  --train_batch_size=4 \
+  --gradient_accumulation_steps=1 \
+  --learning_rate=1e-7 \
+  --max_train_steps=150 \
+  --validation_prompt "a photo of sks dog" \
+  --validation_steps 25 \
+  --text_encoder_use_attention_mask \
+  --tokenizer_max_length 77 \
+  --pre_compute_text_embeddings \
+  --use_8bit_adam \
+  --set_grads_to_none \
+  --skip_save_text_encoder \
+  --push_to_hub
+```
+
+</hfoption>
+<hfoption id="Stage 2 DreamBooth">
+
+For stage 2 of DeepFloyd IF with DreamBooth, pay attention to these parameters:
+
+* `--learning_rate=5e-6`, use a lower learning rate with a smaller effective batch size
+* `--resolution=256`, the expected resolution for the upscaler
+* `--train_batch_size=2` and `--gradient_accumulation_steps=6`, to effectively train on images wiht faces requires larger batch sizes
+
+```bash
+export MODEL_NAME="DeepFloyd/IF-II-L-v1.0"
+export INSTANCE_DIR="dog"
+export OUTPUT_DIR="dreambooth_dog_upscale"
+export VALIDATION_IMAGES="dog_downsized/image_1.png dog_downsized/image_2.png dog_downsized/image_3.png dog_downsized/image_4.png"
+
+accelerate launch train_dreambooth.py \
+  --report_to wandb \
+  --pretrained_model_name_or_path=$MODEL_NAME \
+  --instance_data_dir=$INSTANCE_DIR \
+  --output_dir=$OUTPUT_DIR \
+  --instance_prompt="a sks dog" \
+  --resolution=256 \
+  --train_batch_size=2 \
+  --gradient_accumulation_steps=6 \
+  --learning_rate=5e-6 \
+  --max_train_steps=2000 \
+  --validation_prompt="a sks dog" \
+  --validation_steps=150 \
+  --checkpointing_steps=500 \
+  --pre_compute_text_embeddings \
+  --tokenizer_max_length=77 \
+  --text_encoder_use_attention_mask \
+  --validation_images $VALIDATION_IMAGES \
+  --class_labels_conditioning timesteps \
+  --push_to_hub
+```
+
+</hfoption>
+</hfoptions>
+
+### Training tips
+
+Training the DeepFloyd IF model can be challenging, but here are some tips that we've found helpful:
+
+- LoRA is sufficient for training the stage 1 model because the model's low resolution makes representing finer details difficult regardless.
+- For common or simple objects, you don't necessarily need to finetune the upscaler. Make sure the prompt passed to the upscaler is adjusted to remove the new token from the instance prompt. For example, if your stage 1 prompt is "a sks dog" then your stage 2 prompt should be "a dog".
+- For finer details like faces, fully training the stage 2 upscaler is better than training the stage 2 model with LoRA. It also helps to use lower learning rates with larger batch sizes.
+- Lower learning rates should be used to train the stage 2 model.
+- The [`DDPMScheduler`] works better than the DPMSolver used in the training scripts.
+
 ## Next steps
 
 Congratulations on training your DreamBooth model! To learn more about how to use your new model, the following guide may be helpful:

docs/source/en/training/kandinsky.md

@@ -205,7 +205,7 @@ model_pred = unet(noisy_latents, timesteps, None, added_cond_kwargs=added_cond_k
 
 Once you’ve made all your changes or you’re okay with the default configuration, you’re ready to launch the training script! 🚀
 
-You'll train on the [Pokémon BLIP captions](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) dataset to generate your own Pokémon, but you can also create and train on your own dataset by following the [Create a dataset for training](create_dataset) guide. Set the environment variable `DATASET_NAME` to the name of the dataset on the Hub or if you're training on your own files, set the environment variable `TRAIN_DIR` to a path to your dataset.
+You'll train on the [Naruto BLIP captions](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) dataset to generate your own Naruto characters, but you can also create and train on your own dataset by following the [Create a dataset for training](create_dataset) guide. Set the environment variable `DATASET_NAME` to the name of the dataset on the Hub or if you're training on your own files, set the environment variable `TRAIN_DIR` to a path to your dataset.
 
 If you’re training on more than one GPU, add the `--multi_gpu` parameter to the `accelerate launch` command.
 
@@ -219,7 +219,7 @@ To monitor training progress with Weights & Biases, add the `--report_to=wandb`
 <hfoption id="prior model">
 
 ```bash
-export DATASET_NAME="lambdalabs/pokemon-blip-captions"
+export DATASET_NAME="lambdalabs/naruto-blip-captions"
 
 accelerate launch --mixed_precision="fp16"  train_text_to_image_prior.py \
   --dataset_name=$DATASET_NAME \
@@ -232,17 +232,17 @@ accelerate launch --mixed_precision="fp16"  train_text_to_image_prior.py \
   --checkpoints_total_limit=3 \
   --lr_scheduler="constant" \
   --lr_warmup_steps=0 \
-  --validation_prompts="A robot pokemon, 4k photo" \
+  --validation_prompts="A robot naruto, 4k photo" \
   --report_to="wandb" \
   --push_to_hub \
-  --output_dir="kandi2-prior-pokemon-model"
+  --output_dir="kandi2-prior-naruto-model"
 ```
 
 </hfoption>
 <hfoption id="decoder model">
 
 ```bash
-export DATASET_NAME="lambdalabs/pokemon-blip-captions"
+export DATASET_NAME="lambdalabs/naruto-blip-captions"
 
 accelerate launch --mixed_precision="fp16"  train_text_to_image_decoder.py \
   --dataset_name=$DATASET_NAME \
@@ -256,10 +256,10 @@ accelerate launch --mixed_precision="fp16"  train_text_to_image_decoder.py \
   --checkpoints_total_limit=3 \
   --lr_scheduler="constant" \
   --lr_warmup_steps=0 \
-  --validation_prompts="A robot pokemon, 4k photo" \
+  --validation_prompts="A robot naruto, 4k photo" \
   --report_to="wandb" \
   --push_to_hub \
-  --output_dir="kandi2-decoder-pokemon-model"
+  --output_dir="kandi2-decoder-naruto-model"
 ```
 
 </hfoption>
@@ -279,7 +279,7 @@ prior_components = {"prior_" + k: v for k,v in prior_pipeline.components.items()
 pipeline = AutoPipelineForText2Image.from_pretrained("kandinsky-community/kandinsky-2-2-decoder", **prior_components, torch_dtype=torch.float16)
 
 pipe.enable_model_cpu_offload()
-prompt="A robot pokemon, 4k photo"
+prompt="A robot naruto, 4k photo"
 image = pipeline(prompt=prompt, negative_prompt=negative_prompt).images[0]
 ```
 
@@ -299,7 +299,7 @@ import torch
 pipeline = AutoPipelineForText2Image.from_pretrained("path/to/saved/model", torch_dtype=torch.float16)
 pipeline.enable_model_cpu_offload()
 
-prompt="A robot pokemon, 4k photo"
+prompt="A robot naruto, 4k photo"
 image = pipeline(prompt=prompt).images[0]
 ```
 
@@ -313,7 +313,7 @@ unet = UNet2DConditionModel.from_pretrained("path/to/saved/model" + "/checkpoint
 pipeline = AutoPipelineForText2Image.from_pretrained("kandinsky-community/kandinsky-2-2-decoder", unet=unet, torch_dtype=torch.float16)
 pipeline.enable_model_cpu_offload()
 
-image = pipeline(prompt="A robot pokemon, 4k photo").images[0]
+image = pipeline(prompt="A robot naruto, 4k photo").images[0]
 ```
 
 </hfoption>

docs/source/en/training/lora.md

@@ -170,7 +170,7 @@ Aside from setting up the LoRA layers, the training script is more or less the s
 
 Once you've made all your changes or you're okay with the default configuration, you're ready to launch the training script! 🚀
 
-Let's train on the [Pokémon BLIP captions](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) dataset to generate our own Pokémon. Set the environment variables `MODEL_NAME` and `DATASET_NAME` to the model and dataset respectively. You should also specify where to save the model in `OUTPUT_DIR`, and the name of the model to save to on the Hub with `HUB_MODEL_ID`. The script creates and saves the following files to your repository:
+Let's train on the [Naruto BLIP captions](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) dataset to generate your own Naruto characters. Set the environment variables `MODEL_NAME` and `DATASET_NAME` to the model and dataset respectively. You should also specify where to save the model in `OUTPUT_DIR`, and the name of the model to save to on the Hub with `HUB_MODEL_ID`. The script creates and saves the following files to your repository:
 
 - saved model checkpoints
 - `pytorch_lora_weights.safetensors` (the trained LoRA weights)
@@ -185,9 +185,9 @@ A full training run takes ~5 hours on a 2080 Ti GPU with 11GB of VRAM.
 
 ```bash
 export MODEL_NAME="runwayml/stable-diffusion-v1-5"
-export OUTPUT_DIR="/sddata/finetune/lora/pokemon"
-export HUB_MODEL_ID="pokemon-lora"
-export DATASET_NAME="lambdalabs/pokemon-blip-captions"
+export OUTPUT_DIR="/sddata/finetune/lora/naruto"
+export HUB_MODEL_ID="naruto-lora"
+export DATASET_NAME="lambdalabs/naruto-blip-captions"
 
 accelerate launch --mixed_precision="fp16"  train_text_to_image_lora.py \
   --pretrained_model_name_or_path=$MODEL_NAME \
@@ -208,7 +208,7 @@ accelerate launch --mixed_precision="fp16"  train_text_to_image_lora.py \
   --hub_model_id=${HUB_MODEL_ID} \
   --report_to=wandb \
   --checkpointing_steps=500 \
-  --validation_prompt="A pokemon with blue eyes." \
+  --validation_prompt="A naruto with blue eyes." \
   --seed=1337
 ```
 
@@ -220,7 +220,7 @@ import torch
 
 pipeline = AutoPipelineForText2Image.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16).to("cuda")
 pipeline.load_lora_weights("path/to/lora/model", weight_name="pytorch_lora_weights.safetensors")
-image = pipeline("A pokemon with blue eyes").images[0]
+image = pipeline("A naruto with blue eyes").images[0]
 ```
 
 ## Next steps

docs/source/en/training/sdxl.md

@@ -176,7 +176,7 @@ If you want to learn more about how the training loop works, check out the [Unde
 
 Once you’ve made all your changes or you’re okay with the default configuration, you’re ready to launch the training script! 🚀
 
-Let’s train on the [Pokémon BLIP captions](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) dataset to generate your own Pokémon. Set the environment variables `MODEL_NAME` and `DATASET_NAME` to the model and the dataset (either from the Hub or a local path). You should also specify a VAE other than the SDXL VAE (either from the Hub or a local path) with `VAE_NAME` to avoid numerical instabilities.
+Let’s train on the [Naruto BLIP captions](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) dataset to generate your own Naruto characters. Set the environment variables `MODEL_NAME` and `DATASET_NAME` to the model and the dataset (either from the Hub or a local path). You should also specify a VAE other than the SDXL VAE (either from the Hub or a local path) with `VAE_NAME` to avoid numerical instabilities.
 
 <Tip>
 
@@ -187,7 +187,7 @@ To monitor training progress with Weights & Biases, add the `--report_to=wandb`
 ```bash
 export MODEL_NAME="stabilityai/stable-diffusion-xl-base-1.0"
 export VAE_NAME="madebyollin/sdxl-vae-fp16-fix"
-export DATASET_NAME="lambdalabs/pokemon-blip-captions"
+export DATASET_NAME="lambdalabs/naruto-blip-captions"
 
 accelerate launch train_text_to_image_sdxl.py \
   --pretrained_model_name_or_path=$MODEL_NAME \
@@ -211,7 +211,7 @@ accelerate launch train_text_to_image_sdxl.py \
   --validation_prompt="a cute Sundar Pichai creature" \
   --validation_epochs 5 \
   --checkpointing_steps=5000 \
-  --output_dir="sdxl-pokemon-model" \
+  --output_dir="sdxl-naruto-model" \
   --push_to_hub
 ```
 
@@ -226,9 +226,9 @@ import torch
 
 pipeline = DiffusionPipeline.from_pretrained("path/to/your/model", torch_dtype=torch.float16).to("cuda")
 
-prompt = "A pokemon with green eyes and red legs."
+prompt = "A naruto with green eyes and red legs."
 image = pipeline(prompt, num_inference_steps=30, guidance_scale=7.5).images[0]
-image.save("pokemon.png")
+image.save("naruto.png")
 ```
 
 </hfoption>
@@ -244,11 +244,11 @@ import torch_xla.core.xla_model as xm
 device = xm.xla_device()
 pipeline = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0").to(device)
 
-prompt = "A pokemon with green eyes and red legs."
+prompt = "A naruto with green eyes and red legs."
 start = time()
 image = pipeline(prompt, num_inference_steps=inference_steps).images[0]
 print(f'Compilation time is {time()-start} sec')
-image.save("pokemon.png")
+image.save("naruto.png")
 
 start = time()
 image = pipeline(prompt, num_inference_steps=inference_steps).images[0]

docs/source/en/training/text2image.md

@@ -158,7 +158,7 @@ Once you've made all your changes or you're okay with the default configuration,
 <hfoptions id="training-inference">
 <hfoption id="PyTorch">
 
-Let's train on the [Pokémon BLIP captions](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) dataset to generate your own Pokémon. Set the environment variables `MODEL_NAME` and `dataset_name` to the model and the dataset (either from the Hub or a local path). If you're training on more than one GPU, add the `--multi_gpu` parameter to the `accelerate launch` command.
+Let's train on the [Naruto BLIP captions](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) dataset to generate your own Naruto characters. Set the environment variables `MODEL_NAME` and `dataset_name` to the model and the dataset (either from the Hub or a local path). If you're training on more than one GPU, add the `--multi_gpu` parameter to the `accelerate launch` command.
 
 <Tip>
 
@@ -168,7 +168,7 @@ To train on a local dataset, set the `TRAIN_DIR` and `OUTPUT_DIR` environment va
 
 ```bash
 export MODEL_NAME="runwayml/stable-diffusion-v1-5"
-export dataset_name="lambdalabs/pokemon-blip-captions"
+export dataset_name="lambdalabs/naruto-blip-captions"
 
 accelerate launch --mixed_precision="fp16"  train_text_to_image.py \
   --pretrained_model_name_or_path=$MODEL_NAME \
@@ -183,7 +183,7 @@ accelerate launch --mixed_precision="fp16"  train_text_to_image.py \
   --max_grad_norm=1 \
   --enable_xformers_memory_efficient_attention
   --lr_scheduler="constant" --lr_warmup_steps=0 \
-  --output_dir="sd-pokemon-model" \
+  --output_dir="sd-naruto-model" \
   --push_to_hub
 ```
 
@@ -202,7 +202,7 @@ To train on a local dataset, set the `TRAIN_DIR` and `OUTPUT_DIR` environment va
 
 ```bash
 export MODEL_NAME="runwayml/stable-diffusion-v1-5"
-export dataset_name="lambdalabs/pokemon-blip-captions"
+export dataset_name="lambdalabs/naruto-blip-captions"
 
 python train_text_to_image_flax.py \
   --pretrained_model_name_or_path=$MODEL_NAME \
@@ -212,7 +212,7 @@ python train_text_to_image_flax.py \
   --max_train_steps=15000 \
   --learning_rate=1e-05 \
   --max_grad_norm=1 \
-  --output_dir="sd-pokemon-model" \
+  --output_dir="sd-naruto-model" \
   --push_to_hub
 ```
 
@@ -231,7 +231,7 @@ import torch
 pipeline = StableDiffusionPipeline.from_pretrained("path/to/saved_model", torch_dtype=torch.float16, use_safetensors=True).to("cuda")
 
 image = pipeline(prompt="yoda").images[0]
-image.save("yoda-pokemon.png")
+image.save("yoda-naruto.png")
 ```
 
 </hfoption>
@@ -246,7 +246,7 @@ from diffusers import FlaxStableDiffusionPipeline
 
 pipeline, params = FlaxStableDiffusionPipeline.from_pretrained("path/to/saved_model", dtype=jax.numpy.bfloat16)
 
-prompt = "yoda pokemon"
+prompt = "yoda naruto"
 prng_seed = jax.random.PRNGKey(0)
 num_inference_steps = 50
 
@@ -261,7 +261,7 @@ prompt_ids = shard(prompt_ids)
 
 images = pipeline(prompt_ids, params, prng_seed, num_inference_steps, jit=True).images
 images = pipeline.numpy_to_pil(np.asarray(images.reshape((num_samples,) + images.shape[-3:])))
-image.save("yoda-pokemon.png")
+image.save("yoda-naruto.png")
 ```
 
 </hfoption>

docs/source/en/training/wuerstchen.md

@@ -131,7 +131,7 @@ If you want to learn more about how the training loop works, check out the [Unde
 
 Once you’ve made all your changes or you’re okay with the default configuration, you’re ready to launch the training script! 🚀
 
-Set the `DATASET_NAME` environment variable to the dataset name from the Hub. This guide uses the [Pokémon BLIP captions](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) dataset, but you can create and train on your own datasets as well (see the [Create a dataset for training](create_dataset) guide).
+Set the `DATASET_NAME` environment variable to the dataset name from the Hub. This guide uses the [Naruto BLIP captions](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) dataset, but you can create and train on your own datasets as well (see the [Create a dataset for training](create_dataset) guide).
 
 <Tip>
 
@@ -140,7 +140,7 @@ To monitor training progress with Weights & Biases, add the `--report_to=wandb`
 </Tip>
 
 ```bash
-export DATASET_NAME="lambdalabs/pokemon-blip-captions"
+export DATASET_NAME="lambdalabs/naruto-blip-captions"
 
 accelerate launch  train_text_to_image_prior.py \
   --mixed_precision="fp16" \
@@ -156,10 +156,10 @@ accelerate launch  train_text_to_image_prior.py \
   --checkpoints_total_limit=3 \
   --lr_scheduler="constant" \
   --lr_warmup_steps=0 \
-  --validation_prompts="A robot pokemon, 4k photo" \
+  --validation_prompts="A robot naruto, 4k photo" \
   --report_to="wandb" \
   --push_to_hub \
-  --output_dir="wuerstchen-prior-pokemon-model"
+  --output_dir="wuerstchen-prior-naruto-model"
 ```
 
 Once training is complete, you can use your newly trained model for inference!
@@ -171,7 +171,7 @@ from diffusers.pipelines.wuerstchen import DEFAULT_STAGE_C_TIMESTEPS
 
 pipeline = AutoPipelineForText2Image.from_pretrained("path/to/saved/model", torch_dtype=torch.float16).to("cuda")
 
-caption = "A cute bird pokemon holding a shield"
+caption = "A cute bird naruto holding a shield"
 images = pipeline(
     caption,
     width=1024,

docs/source/en/tutorials/basic_training.md

@@ -260,7 +260,7 @@ Then, you'll need a way to evaluate the model. For evaluation, you can use the [
 ...     # The default pipeline output type is `List[PIL.Image]`
 ...     images = pipeline(
 ...         batch_size=config.eval_batch_size,
-...         generator=torch.manual_seed(config.seed),
+...         generator=torch.Generator(device='cpu').manual_seed(config.seed), # Use a separate torch generator to avoid rewinding the random state of the main training loop
 ...     ).images
 
 ...     # Make a grid out of the images

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

@@ -19,13 +19,74 @@ The denoising loop of a pipeline can be modified with custom defined functions u
 
 This guide will demonstrate how callbacks work by a few features you can implement with them.
 
+## Official callbacks
+
+We provide a list of callbacks you can plug into an existing pipeline and modify the denoising loop. This is the current list of official callbacks:
+
+- `SDCFGCutoffCallback`: Disables the CFG after a certain number of steps for all SD 1.5 pipelines, including text-to-image, image-to-image, inpaint, and controlnet.
+- `SDXLCFGCutoffCallback`: Disables the CFG after a certain number of steps for all SDXL pipelines, including text-to-image, image-to-image, inpaint, and controlnet.
+- `IPAdapterScaleCutoffCallback`: Disables the IP Adapter after a certain number of steps for all pipelines supporting IP-Adapter.
+
+> [!TIP]
+> If you want to add a new official callback, feel free to open a [feature request](https://github.com/huggingface/diffusers/issues/new/choose) or [submit a PR](https://huggingface.co/docs/diffusers/main/en/conceptual/contribution#how-to-open-a-pr).
+
+To set up a callback, you need to specify the number of denoising steps after which the callback comes into effect. You can do so by using either one of these two arguments
+
+- `cutoff_step_ratio`: Float number with the ratio of the steps.
+- `cutoff_step_index`: Integer number with the exact number of the step.
+
+```python
+import torch
+
+from diffusers import DPMSolverMultistepScheduler, StableDiffusionXLPipeline
+from diffusers.callbacks import SDXLCFGCutoffCallback
+
+
+callback = SDXLCFGCutoffCallback(cutoff_step_ratio=0.4)
+# can also be used with cutoff_step_index
+# callback = SDXLCFGCutoffCallback(cutoff_step_ratio=None, cutoff_step_index=10)
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
+pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config, use_karras_sigmas=True)
+
+prompt = "a sports car at the road, best quality, high quality, high detail, 8k resolution"
+
+generator = torch.Generator(device="cpu").manual_seed(2628670641)
+
+out = pipeline(
+    prompt=prompt,
+    negative_prompt="",
+    guidance_scale=6.5,
+    num_inference_steps=25,
+    generator=generator,
+    callback_on_step_end=callback,
+)
+
+out.images[0].save("official_callback.png")
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/without_cfg_callback.png" alt="generated image of a sports car at the road" />
+    <figcaption class="mt-2 text-center text-sm text-gray-500">without SDXLCFGCutoffCallback</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/with_cfg_callback.png" alt="generated image of a a sports car at the road with cfg callback" />
+    <figcaption class="mt-2 text-center text-sm text-gray-500">with SDXLCFGCutoffCallback</figcaption>
+  </div>
+</div>
+
 ## Dynamic classifier-free guidance
 
 Dynamic classifier-free guidance (CFG) is a feature that allows you to disable CFG after a certain number of inference steps which can help you save compute with minimal cost to performance. The callback function for this should have the following arguments:
 
-* `pipeline` (or the pipeline instance) provides access to important properties such as `num_timesteps` and `guidance_scale`. You can modify these properties by updating the underlying attributes. For this example, you'll disable CFG by setting `pipeline._guidance_scale=0.0`.
-* `step_index` and `timestep` tell you where you are in the denoising loop. Use `step_index` to turn off CFG after reaching 40% of `num_timesteps`.
-* `callback_kwargs` is a dict that contains tensor variables you can modify during the denoising loop. It only includes variables specified in the `callback_on_step_end_tensor_inputs` argument, which is passed to the pipeline's `__call__` method. Different pipelines may use different sets of variables, so please check a pipeline's `_callback_tensor_inputs` attribute for the list of variables you can modify. Some common variables include `latents` and `prompt_embeds`. For this function, change the batch size of `prompt_embeds` after setting `guidance_scale=0.0` in order for it to work properly.
+- `pipeline` (or the pipeline instance) provides access to important properties such as `num_timesteps` and `guidance_scale`. You can modify these properties by updating the underlying attributes. For this example, you'll disable CFG by setting `pipeline._guidance_scale=0.0`.
+- `step_index` and `timestep` tell you where you are in the denoising loop. Use `step_index` to turn off CFG after reaching 40% of `num_timesteps`.
+- `callback_kwargs` is a dict that contains tensor variables you can modify during the denoising loop. It only includes variables specified in the `callback_on_step_end_tensor_inputs` argument, which is passed to the pipeline's `__call__` method. Different pipelines may use different sets of variables, so please check a pipeline's `_callback_tensor_inputs` attribute for the list of variables you can modify. Some common variables include `latents` and `prompt_embeds`. For this function, change the batch size of `prompt_embeds` after setting `guidance_scale=0.0` in order for it to work properly.
 
 Your callback function should look something like this:
 

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

@@ -1,133 +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.
--->
-
-# Distilled Stable Diffusion inference
-
-[[open-in-colab]]
-
-Stable Diffusion inference can be a computationally intensive process because it must iteratively denoise the latents to generate an image. To reduce the computational burden, you can use a *distilled* version of the Stable Diffusion model from [Nota AI](https://huggingface.co/nota-ai). The distilled version of their Stable Diffusion model eliminates some of the residual and attention blocks from the UNet, reducing the model size by 51% and improving latency on CPU/GPU by 43%.
-
-<Tip>
-
-Read this [blog post](https://huggingface.co/blog/sd_distillation) to learn more about how knowledge distillation training works to produce a faster, smaller, and cheaper generative model.
-
-</Tip>
-
-Let's load the distilled Stable Diffusion model and compare it against the original Stable Diffusion model:
-
-```py
-from diffusers import StableDiffusionPipeline
-import torch
-
-distilled = StableDiffusionPipeline.from_pretrained(
-    "nota-ai/bk-sdm-small", torch_dtype=torch.float16, use_safetensors=True,
-).to("cuda")
-
-original = StableDiffusionPipeline.from_pretrained(
-    "CompVis/stable-diffusion-v1-4", torch_dtype=torch.float16, use_safetensors=True,
-).to("cuda")
-```
-
-Given a prompt, get the inference time for the original model:
-
-```py
-import time
-
-seed = 2023
-generator = torch.manual_seed(seed)
-
-NUM_ITERS_TO_RUN = 3
-NUM_INFERENCE_STEPS = 25
-NUM_IMAGES_PER_PROMPT = 4
-
-prompt = "a golden vase with different flowers"
-
-start = time.time_ns()
-for _ in range(NUM_ITERS_TO_RUN):
-    images = original(
-        prompt,
-        num_inference_steps=NUM_INFERENCE_STEPS,
-        generator=generator,
-        num_images_per_prompt=NUM_IMAGES_PER_PROMPT
-    ).images
-end = time.time_ns()
-original_sd = f"{(end - start) / 1e6:.1f}"
-
-print(f"Execution time -- {original_sd} ms\n")
-"Execution time -- 45781.5 ms"
-```
-
-Time the distilled model inference:
-
-```py
-start = time.time_ns()
-for _ in range(NUM_ITERS_TO_RUN):
-    images = distilled(
-        prompt,
-        num_inference_steps=NUM_INFERENCE_STEPS,
-        generator=generator,
-        num_images_per_prompt=NUM_IMAGES_PER_PROMPT
-    ).images
-end = time.time_ns()
-
-distilled_sd = f"{(end - start) / 1e6:.1f}"
-print(f"Execution time -- {distilled_sd} ms\n")
-"Execution time -- 29884.2 ms"
-```
-
-<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 (45781.5 ms)</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 (29884.2 ms)</figcaption>
-  </div>
-</div>
-
-## Tiny AutoEncoder
-
-To speed inference up even more, use a tiny distilled version of the [Stable Diffusion VAE](https://huggingface.co/sayakpaul/taesdxl-diffusers) to denoise the latents into images. Replace the VAE in the distilled Stable Diffusion model with the tiny VAE:
-
-```py
-from diffusers import AutoencoderTiny
-
-distilled.vae = AutoencoderTiny.from_pretrained(
-    "sayakpaul/taesd-diffusers", torch_dtype=torch.float16, use_safetensors=True,
-).to("cuda")
-```
-
-Time the distilled model and distilled VAE inference:
-
-```py
-start = time.time_ns()
-for _ in range(NUM_ITERS_TO_RUN):
-    images = distilled(
-        prompt,
-        num_inference_steps=NUM_INFERENCE_STEPS,
-        generator=generator,
-        num_images_per_prompt=NUM_IMAGES_PER_PROMPT
-    ).images
-end = time.time_ns()
-
-distilled_tiny_sd = f"{(end - start) / 1e6:.1f}"
-print(f"Execution time -- {distilled_tiny_sd} ms\n")
-"Execution time -- 27165.7 ms"
-```
-
-<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 (27165.7 ms)</figcaption>
-  </div>
-</div>

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

@@ -12,54 +12,10 @@ specific language governing permissions and limitations under the License.
 
 # Controlling image quality
 
-The components of a diffusion model, like the UNet and scheduler, can be optimized to improve the quality of generated images leading to better image lighting and details. These techniques are especially useful if you don't have the resources to simply use a larger model for inference. You can enable these techniques during inference without any additional training.
+The components of a diffusion model, like the UNet and scheduler, can be optimized to improve the quality of generated images leading to better details. These techniques are especially useful if you don't have the resources to simply use a larger model for inference. You can enable these techniques during inference without any additional training.
 
 This guide will show you how to turn these techniques on in your pipeline and how to configure them to improve the quality of your generated images.
 
-## Lighting
-
-The Stable Diffusion models aren't very good at generating images that are very bright or dark because the scheduler doesn't start sampling from the last timestep and it doesn't enforce a zero signal-to-noise ratio (SNR). The [Common Diffusion Noise Schedules and Sample Steps are Flawed](https://hf.co/papers/2305.08891) paper fixes these issues which are now available in some Diffusers schedulers.
-
-> [!TIP]
-> For inference, you need a model that has been trained with *v_prediction*. To train your own model with *v_prediction*, add the following flag to the [train_text_to_image.py](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py) or [train_text_to_image_lora.py](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_lora.py) scripts.
->
-> ```bash
-> --prediction_type="v_prediction"
-> ```
-
-For example, load the [ptx0/pseudo-journey-v2](https://hf.co/ptx0/pseudo-journey-v2) checkpoint which was trained with `v_prediction` and the [`DDIMScheduler`]. Now you should configure the following parameters in the [`DDIMScheduler`].
-
-* `rescale_betas_zero_snr=True` to rescale the noise schedule to zero SNR
-* `timestep_spacing="trailing"` to start sampling from the last timestep
-
-Set `guidance_rescale` in the pipeline to prevent over-exposure. A lower value increases brightness but some of the details may appear washed out.
-
-```py
-from diffusers import DiffusionPipeline, DDIMScheduler
-
-pipeline = DiffusionPipeline.from_pretrained("ptx0/pseudo-journey-v2", use_safetensors=True)
-
-pipeline.scheduler = DDIMScheduler.from_config(
-    pipeline.scheduler.config, rescale_betas_zero_snr=True, timestep_spacing="trailing"
-)
-pipeline.to("cuda")
-prompt = "cinematic photo of a snowy mountain at night with the northern lights aurora borealis overhead, 35mm photograph, film, professional, 4k, highly detailed"
-generator = torch.Generator(device="cpu").manual_seed(23)
-image = pipeline(prompt, guidance_rescale=0.7, generator=generator).images[0]
-image
-```
-
-<div class="flex gap-4">
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/no-zero-snr.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">default Stable Diffusion v2-1 image</figcaption>
-  </div>
-  <div>
-    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/zero-snr.png"/>
-    <figcaption class="mt-2 text-center text-sm text-gray-500">image with zero SNR and trailing timestep spacing enabled</figcaption>
-  </div>
-</div>
-
 ## Details
 
 [FreeU](https://hf.co/papers/2309.11497) improves image details by rebalancing the UNet's backbone and skip connection weights. The skip connections can cause the model to overlook some of the backbone semantics which may lead to unnatural image details in the generated image. This technique does not require any additional training and can be applied on the fly during inference for tasks like image-to-image and text-to-video.

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

@@ -10,29 +10,30 @@ 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]]
-
 # Latent Consistency Model
 
-Latent Consistency Models (LCM) enable quality image generation in typically 2-4 steps making it possible to use diffusion models in almost real-time settings. 
+[[open-in-colab]]
 
-From the [official website](https://latent-consistency-models.github.io/):
+[Latent Consistency Models (LCMs)](https://hf.co/papers/2310.04378) enable fast high-quality image generation by directly predicting the reverse diffusion process in the latent rather than pixel space. In other words, LCMs try to predict the noiseless image from the noisy image in contrast to typical diffusion models that iteratively remove noise from the noisy image. By avoiding the iterative sampling process, LCMs are able to generate high-quality images in 2-4 steps instead of 20-30 steps.
 
-> LCMs can be distilled from any pre-trained Stable Diffusion (SD) in only 4,000 training steps (~32 A100 GPU Hours) for generating high quality 768 x 768 resolution images in 2~4 steps or even one step, significantly accelerating text-to-image generation. We employ LCM to distill the Dreamshaper-V7 version of SD in just 4,000 training iterations.
+LCMs are distilled from pretrained models which requires ~32 hours of A100 compute. To speed this up, [LCM-LoRAs](https://hf.co/papers/2311.05556) train a [LoRA adapter](https://huggingface.co/docs/peft/conceptual_guides/adapter#low-rank-adaptation-lora) which have much fewer parameters to train compared to the full model. The LCM-LoRA can be plugged into a diffusion model once it has been trained.
 
-For a more technical overview of LCMs, refer to [the paper](https://huggingface.co/papers/2310.04378).
+This guide will show you how to use LCMs and LCM-LoRAs for fast inference on tasks and how to use them with other adapters like ControlNet or T2I-Adapter.
 
-LCM distilled models are available for [stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5), [stable-diffusion-xl-base-1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0), and the [SSD-1B](https://huggingface.co/segmind/SSD-1B) model. All the checkpoints can be found in this [collection](https://huggingface.co/collections/latent-consistency/latent-consistency-models-weights-654ce61a95edd6dffccef6a8).
-
-This guide shows how to perform inference with LCMs for 
-- text-to-image
-- image-to-image
-- combined with style LoRAs
-- ControlNet/T2I-Adapter
+> [!TIP]
+> LCMs and LCM-LoRAs are available for Stable Diffusion v1.5, Stable Diffusion XL, and the SSD-1B model. You can find their checkpoints on the [Latent Consistency](https://hf.co/collections/latent-consistency/latent-consistency-models-weights-654ce61a95edd6dffccef6a8) Collections.
 
 ## Text-to-image
 
-You'll use the [`StableDiffusionXLPipeline`] pipeline with the [`LCMScheduler`] and then load the LCM-LoRA. Together with the LCM-LoRA and the scheduler, the pipeline enables a fast inference workflow, overcoming the slow iterative nature of diffusion models.
+<hfoptions id="lcm-text2img">
+<hfoption id="LCM">
+
+To use LCMs, you need to load the LCM checkpoint for your supported model into [`UNet2DConditionModel`] and replace the scheduler with the [`LCMScheduler`]. Then you can use the pipeline as usual, and pass a text prompt to generate an image in just 4 steps.
+
+A couple of notes to keep in mind when using LCMs are:
+
+* Typically, batch size is doubled inside the pipeline for classifier-free guidance. But LCM applies guidance with guidance embeddings and doesn't need to double the batch size, which leads to faster inference. The downside is that negative prompts don't work with LCM because they don't have any effect on the denoising process.
+* The ideal range for `guidance_scale` is [3., 13.] because that is what the UNet was trained with. However, disabling `guidance_scale` with a value of 1.0 is also effective in most cases.
 
 ```python
 from diffusers import StableDiffusionXLPipeline, UNet2DConditionModel, LCMScheduler
@@ -49,31 +50,69 @@ pipe = StableDiffusionXLPipeline.from_pretrained(
 pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
 prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"
-
 generator = torch.manual_seed(0)
 image = pipe(
     prompt=prompt, num_inference_steps=4, generator=generator, guidance_scale=8.0
 ).images[0]
+image
 ```
 
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdxl_t2i.png)
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdxl_t2i.png"/>
+</div>
 
-Notice that we use only 4 steps for generation which is way less than what's typically used for standard SDXL.
+</hfoption>
+<hfoption id="LCM-LoRA">
 
-Some details to keep in mind:
+To use LCM-LoRAs, you need to replace the scheduler with the [`LCMScheduler`] and load the LCM-LoRA weights with the [`~loaders.LoraLoaderMixin.load_lora_weights`] method. Then you can use the pipeline as usual, and pass a text prompt to generate an image in just 4 steps.
 
-* To perform classifier-free guidance, batch size is usually doubled inside the pipeline. LCM, however, applies guidance using guidance embeddings, so the batch size does not have to be doubled in this case. This leads to a faster inference time, with the drawback that negative prompts don't have any effect on the denoising process.
-* The UNet was trained using the [3., 13.] guidance scale range. So, that is the ideal range for `guidance_scale`. However, disabling `guidance_scale` using a value of 1.0 is also effective in most cases.
+A couple of notes to keep in mind when using LCM-LoRAs are:
 
+* Typically, batch size is doubled inside the pipeline for classifier-free guidance. But LCM applies guidance with guidance embeddings and doesn't need to double the batch size, which leads to faster inference. The downside is that negative prompts don't work with LCM because they don't have any effect on the denoising process.
+* You could use guidance with LCM-LoRAs, but it is very sensitive to high `guidance_scale` values and can lead to artifacts in the generated image. The best values we've found are between [1.0, 2.0].
+* Replace [stabilityai/stable-diffusion-xl-base-1.0](https://hf.co/stabilityai/stable-diffusion-xl-base-1.0) with any finetuned model. For example, try using the [animagine-xl](https://huggingface.co/Linaqruf/animagine-xl) checkpoint to generate anime images with SDXL.
+
+```py
+import torch
+from diffusers import DiffusionPipeline, LCMScheduler
+
+pipe = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    variant="fp16",
+    torch_dtype=torch.float16
+).to("cuda")
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl")
+
+prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"
+generator = torch.manual_seed(42)
+image = pipe(
+    prompt=prompt, num_inference_steps=4, generator=generator, guidance_scale=1.0
+).images[0]
+image
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdxl_t2i.png"/>
+</div>
+
+</hfoption>
+</hfoptions>
 
 ## Image-to-image
 
-LCMs can be applied to image-to-image tasks too. For this example, we'll use the [LCM_Dreamshaper_v7](https://huggingface.co/SimianLuo/LCM_Dreamshaper_v7) model, but the same steps can be applied to other LCM models as well.
+<hfoptions id="lcm-img2img">
+<hfoption id="LCM">
+
+To use LCMs for image-to-image, you need to load the LCM checkpoint for your supported model into [`UNet2DConditionModel`] and replace the scheduler with the [`LCMScheduler`]. Then you can use the pipeline as usual, and pass a text prompt and initial image to generate an image in just 4 steps.
+
+> [!TIP]
+> Experiment with different values for `num_inference_steps`, `strength`, and `guidance_scale` to get the best results.
 
 ```python
 import torch
 from diffusers import AutoPipelineForImage2Image, UNet2DConditionModel, LCMScheduler
-from diffusers.utils import make_image_grid, load_image
+from diffusers.utils import load_image
 
 unet = UNet2DConditionModel.from_pretrained(
     "SimianLuo/LCM_Dreamshaper_v7",
@@ -89,12 +128,8 @@ pipe = AutoPipelineForImage2Image.from_pretrained(
 ).to("cuda")
 pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
-# prepare image
-url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-init.png"
-init_image = load_image(url)
+init_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-init.png")
 prompt = "Astronauts in a jungle, cold color palette, muted colors, detailed, 8k"
-
-# pass prompt and image to pipeline
 generator = torch.manual_seed(0)
 image = pipe(
     prompt,
@@ -104,22 +139,130 @@ image = pipe(
     strength=0.5,
     generator=generator
 ).images[0]
-make_image_grid([init_image, image], rows=1, cols=2)
+image
 ```
 
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdv1-5_i2i.png)
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-init.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/lcm-img2img.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">generated image</figcaption>
+  </div>
+</div>
 
+</hfoption>
+<hfoption id="LCM-LoRA">
 
-<Tip>
+To use LCM-LoRAs for image-to-image, you need to replace the scheduler with the [`LCMScheduler`] and load the LCM-LoRA weights with the [`~loaders.LoraLoaderMixin.load_lora_weights`] method. Then you can use the pipeline as usual, and pass a text prompt and initial image to generate an image in just 4 steps.
 
-You can get different results based on your prompt and the image you provide. To get the best results, we recommend trying different values for `num_inference_steps`, `strength`, and `guidance_scale` parameters and choose the best one.
+> [!TIP]
+> Experiment with different values for `num_inference_steps`, `strength`, and `guidance_scale` to get the best results.
 
-</Tip>
+```py
+import torch
+from diffusers import AutoPipelineForImage2Image, LCMScheduler
+from diffusers.utils import make_image_grid, load_image
 
+pipe = AutoPipelineForImage2Image.from_pretrained(
+    "Lykon/dreamshaper-7",
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
 
-## Combine with style LoRAs
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
-LCMs can be used with other styled LoRAs to generate styled-images in very few steps (4-8). In the following example, we'll use the [papercut LoRA](TheLastBen/Papercut_SDXL). 
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
+
+init_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-init.png")
+prompt = "Astronauts in a jungle, cold color palette, muted colors, detailed, 8k"
+
+generator = torch.manual_seed(0)
+image = pipe(
+    prompt,
+    image=init_image,
+    num_inference_steps=4,
+    guidance_scale=1,
+    strength=0.6,
+    generator=generator
+).images[0]
+image
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-init.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/lcm-lora-img2img.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">generated image</figcaption>
+  </div>
+</div>
+
+</hfoption>
+</hfoptions>
+
+## Inpainting
+
+To use LCM-LoRAs for inpainting, you need to replace the scheduler with the [`LCMScheduler`] and load the LCM-LoRA weights with the [`~loaders.LoraLoaderMixin.load_lora_weights`] method. Then you can use the pipeline as usual, and pass a text prompt, initial image, and mask image to generate an image in just 4 steps.
+
+```py
+import torch
+from diffusers import AutoPipelineForInpainting, LCMScheduler
+from diffusers.utils import load_image, make_image_grid
+
+pipe = AutoPipelineForInpainting.from_pretrained(
+    "runwayml/stable-diffusion-inpainting",
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
+
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
+
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
+
+init_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/inpaint.png")
+mask_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/inpaint_mask.png")
+
+prompt = "concept art digital painting of an elven castle, inspired by lord of the rings, highly detailed, 8k"
+generator = torch.manual_seed(0)
+image = pipe(
+    prompt=prompt,
+    image=init_image,
+    mask_image=mask_image,
+    generator=generator,
+    num_inference_steps=4,
+    guidance_scale=4, 
+).images[0]
+image
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/inpaint.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/lcm-lora-inpaint.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">generated image</figcaption>
+  </div>
+</div>
+
+## Adapters
+
+LCMs are compatible with adapters like LoRA, ControlNet, T2I-Adapter, and AnimateDiff. You can bring the speed of LCMs to these adapters to generate images in a certain style or condition the model on another input like a canny image.
+
+### LoRA
+
+[LoRA](../using-diffusers/loading_adapters#lora) adapters can be rapidly finetuned to learn a new style from just a few images and plugged into a pretrained model to generate images in that style.
+
+<hfoptions id="lcm-lora">
+<hfoption id="LCM">
+
+Load the LCM checkpoint for your supported model into [`UNet2DConditionModel`] and replace the scheduler with the [`LCMScheduler`]. Then you can use the [`~loaders.LoraLoaderMixin.load_lora_weights`] method to load the LoRA weights into the LCM and generate a styled image in a few steps.
 
 ```python
 from diffusers import StableDiffusionXLPipeline, UNet2DConditionModel, LCMScheduler
@@ -134,11 +277,9 @@ pipe = StableDiffusionXLPipeline.from_pretrained(
     "stabilityai/stable-diffusion-xl-base-1.0", unet=unet, torch_dtype=torch.float16, variant="fp16",
 ).to("cuda")
 pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
 pipe.load_lora_weights("TheLastBen/Papercut_SDXL", weight_name="papercut.safetensors", adapter_name="papercut")
 
 prompt = "papercut, a cute fox"
-
 generator = torch.manual_seed(0)
 image = pipe(
     prompt=prompt, num_inference_steps=4, generator=generator, guidance_scale=8.0
@@ -146,15 +287,58 @@ image = pipe(
 image
 ```
 
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdx_lora_mix.png)
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdx_lora_mix.png"/>
+</div>
 
+</hfoption>
+<hfoption id="LCM-LoRA">
 
-## ControlNet/T2I-Adapter
+Replace the scheduler with the [`LCMScheduler`]. Then you can use the [`~loaders.LoraLoaderMixin.load_lora_weights`] method to load the LCM-LoRA weights and the style LoRA you want to use. Combine both LoRA adapters with the [`~loaders.UNet2DConditionLoadersMixin.set_adapters`] method and generate a styled image in a few steps.
 
-Let's look at how we can perform inference with ControlNet/T2I-Adapter and a LCM. 
+```py
+import torch
+from diffusers import DiffusionPipeline, LCMScheduler
+
+pipe = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    variant="fp16",
+    torch_dtype=torch.float16
+).to("cuda")
+
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
+
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl", adapter_name="lcm")
+pipe.load_lora_weights("TheLastBen/Papercut_SDXL", weight_name="papercut.safetensors", adapter_name="papercut")
+
+pipe.set_adapters(["lcm", "papercut"], adapter_weights=[1.0, 0.8])
+
+prompt = "papercut, a cute fox"
+generator = torch.manual_seed(0)
+image = pipe(prompt, num_inference_steps=4, guidance_scale=1, generator=generator).images[0]
+image
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdx_lora_mix.png"/>
+</div>
+
+</hfoption>
+</hfoptions>
 
 ### ControlNet
-For this example, we'll use the [LCM_Dreamshaper_v7](https://huggingface.co/SimianLuo/LCM_Dreamshaper_v7) model with canny ControlNet, but the same steps can be applied to other LCM models as well.
+
+[ControlNet](./controlnet) are adapters that can be trained on a variety of inputs like canny edge, pose estimation, or depth. The ControlNet can be inserted into the pipeline to provide additional conditioning and control to the model for more accurate generation.
+
+You can find additional ControlNet models trained on other inputs in [lllyasviel's](https://hf.co/lllyasviel) repository.
+
+<hfoptions id="lcm-controlnet">
+<hfoption id="LCM">
+
+Load a ControlNet model trained on canny images and pass it to the [`ControlNetModel`]. Then you can load a LCM model into [`StableDiffusionControlNetPipeline`] and replace the scheduler with the [`LCMScheduler`]. Now pass the canny image to the pipeline and generate an image.
+
+> [!TIP]
+> Experiment with different values for `num_inference_steps`, `controlnet_conditioning_scale`, `cross_attention_kwargs`, and `guidance_scale` to get the best results.
 
 ```python
 import torch
@@ -186,8 +370,6 @@ pipe = StableDiffusionControlNetPipeline.from_pretrained(
     torch_dtype=torch.float16,
     safety_checker=None,
 ).to("cuda")
-
-# set scheduler
 pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
 generator = torch.manual_seed(0)
@@ -200,16 +382,84 @@ image = pipe(
 make_image_grid([canny_image, image], rows=1, cols=2)
 ```
 
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdv1-5_controlnet.png)
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdv1-5_controlnet.png"/>
+</div>
 
+</hfoption>
+<hfoption id="LCM-LoRA">
 
-<Tip>
-The inference parameters in this example might not work for all examples, so we recommend trying different values for the `num_inference_steps`, `guidance_scale`, `controlnet_conditioning_scale`, and `cross_attention_kwargs` parameters and choosing the best one. 
-</Tip>
+Load a ControlNet model trained on canny images and pass it to the [`ControlNetModel`]. Then you can load a Stable Diffusion v1.5 model into [`StableDiffusionControlNetPipeline`] and replace the scheduler with the [`LCMScheduler`]. Use the [`~loaders.LoraLoaderMixin.load_lora_weights`] method to load the LCM-LoRA weights, and pass the canny image to the pipeline and generate an image.
+
+> [!TIP]
+> Experiment with different values for `num_inference_steps`, `controlnet_conditioning_scale`, `cross_attention_kwargs`, and `guidance_scale` to get the best results.
+
+```py
+import torch
+import cv2
+import numpy as np
+from PIL import Image
+
+from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, LCMScheduler
+from diffusers.utils import load_image
+
+image = load_image(
+    "https://hf.co/datasets/huggingface/documentation-images/resolve/main/diffusers/input_image_vermeer.png"
+).resize((512, 512))
+
+image = np.array(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)
+
+controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16)
+pipe = StableDiffusionControlNetPipeline.from_pretrained(
+    "runwayml/stable-diffusion-v1-5",
+    controlnet=controlnet,
+    torch_dtype=torch.float16,
+    safety_checker=None,
+    variant="fp16"
+).to("cuda")
+
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
+
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
+
+generator = torch.manual_seed(0)
+image = pipe(
+    "the mona lisa",
+    image=canny_image,
+    num_inference_steps=4,
+    guidance_scale=1.5,
+    controlnet_conditioning_scale=0.8,
+    cross_attention_kwargs={"scale": 1},
+    generator=generator,
+).images[0]
+image
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdv1-5_controlnet.png"/>
+</div>
+
+</hfoption>
+</hfoptions>
 
 ### T2I-Adapter
 
-This example shows how to use the `lcm-sdxl` with the [Canny T2I-Adapter](TencentARC/t2i-adapter-canny-sdxl-1.0).
+[T2I-Adapter](./t2i_adapter) is an even more lightweight adapter than ControlNet, that provides an additional input to condition a pretrained model with. It is faster than ControlNet but the results may be slightly worse.
+
+You can find additional T2I-Adapter checkpoints trained on other inputs in [TencentArc's](https://hf.co/TencentARC) repository.
+
+<hfoptions id="lcm-t2i">
+<hfoption id="LCM">
+
+Load a T2IAdapter trained on canny images and pass it to the [`StableDiffusionXLAdapterPipeline`]. Then load a LCM checkpoint into [`UNet2DConditionModel`] and replace the scheduler with the [`LCMScheduler`]. Now pass the canny image to the pipeline and generate an image.
 
 ```python
 import torch
@@ -220,10 +470,9 @@ from PIL import Image
 from diffusers import StableDiffusionXLAdapterPipeline, UNet2DConditionModel, T2IAdapter, LCMScheduler
 from diffusers.utils import load_image, make_image_grid
 
-# Prepare image
-# Detect the canny map in low resolution to avoid high-frequency details
+# detect the canny map in low resolution to avoid high-frequency details
 image = load_image(
-    "https://huggingface.co/Adapter/t2iadapter/resolve/main/figs_SDXLV1.0/org_canny.jpg"
+    "https://hf.co/datasets/huggingface/documentation-images/resolve/main/diffusers/input_image_vermeer.png"
 ).resize((384, 384))
 
 image = np.array(image)
@@ -236,7 +485,6 @@ image = image[:, :, None]
 image = np.concatenate([image, image, image], axis=2)
 canny_image = Image.fromarray(image).resize((1024, 1216))
 
-# load adapter
 adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, varient="fp16").to("cuda")
 
 unet = UNet2DConditionModel.from_pretrained(
@@ -254,7 +502,7 @@ pipe = StableDiffusionXLAdapterPipeline.from_pretrained(
 
 pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
 
-prompt = "Mystical fairy in real, magic, 4k picture, high quality"
+prompt = "the mona lisa, 4k picture, high quality"
 negative_prompt = "extra digit, fewer digits, cropped, worst quality, low quality, glitch, deformed, mutated, ugly, disfigured"
 
 generator = torch.manual_seed(0)
@@ -268,7 +516,116 @@ image = pipe(
     adapter_conditioning_factor=1,
     generator=generator,
 ).images[0]
-grid = make_image_grid([canny_image, image], rows=1, cols=2)
 ```
 
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_full_sdxl_t2iadapter.png)
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm-t2i.png"/>
+</div>
+
+</hfoption>
+<hfoption id="LCM-LoRA">
+
+Load a T2IAdapter trained on canny images and pass it to the [`StableDiffusionXLAdapterPipeline`]. Replace the scheduler with the [`LCMScheduler`], and use the [`~loaders.LoraLoaderMixin.load_lora_weights`] method to load the LCM-LoRA weights. Pass the canny image to the pipeline and generate an image.
+
+```py
+import torch
+import cv2
+import numpy as np
+from PIL import Image
+
+from diffusers import StableDiffusionXLAdapterPipeline, UNet2DConditionModel, T2IAdapter, LCMScheduler
+from diffusers.utils import load_image, make_image_grid
+
+# detect the canny map in low resolution to avoid high-frequency details
+image = load_image(
+    "https://hf.co/datasets/huggingface/documentation-images/resolve/main/diffusers/input_image_vermeer.png"
+).resize((384, 384))
+
+image = np.array(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).resize((1024, 1024))
+
+adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, varient="fp16").to("cuda")
+
+pipe = StableDiffusionXLAdapterPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0", 
+    adapter=adapter,
+    torch_dtype=torch.float16,
+    variant="fp16", 
+).to("cuda")
+
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
+
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl")
+
+prompt = "the mona lisa, 4k picture, high quality"
+negative_prompt = "extra digit, fewer digits, cropped, worst quality, low quality, glitch, deformed, mutated, ugly, disfigured"
+
+generator = torch.manual_seed(0)
+image = pipe(
+    prompt=prompt,
+    negative_prompt=negative_prompt,
+    image=canny_image,
+    num_inference_steps=4,
+    guidance_scale=1.5, 
+    adapter_conditioning_scale=0.8, 
+    adapter_conditioning_factor=1,
+    generator=generator,
+).images[0]
+```
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm-lora-t2i.png"/>
+</div>
+
+</hfoption>
+</hfoptions>
+
+### AnimateDiff
+
+[AnimateDiff](../api/pipelines/animatediff) is an adapter that adds motion to an image. It can be used with most Stable Diffusion models, effectively turning them into "video generation" models. Generating good results with a video model usually requires generating multiple frames (16-24), which can be very slow with a regular Stable Diffusion model. LCM-LoRA can speed up this process by only taking 4-8 steps for each frame.
+
+Load a [`AnimateDiffPipeline`] and pass a [`MotionAdapter`] to it. Then replace the scheduler with the [`LCMScheduler`], and combine both LoRA adapters with the [`~loaders.UNet2DConditionLoadersMixin.set_adapters`] method. Now you can pass a prompt to the pipeline and generate an animated image.
+
+```py
+import torch
+from diffusers import MotionAdapter, AnimateDiffPipeline, DDIMScheduler, LCMScheduler
+from diffusers.utils import export_to_gif
+
+adapter = MotionAdapter.from_pretrained("guoyww/animatediff-motion-adapter-v1-5")
+pipe = AnimateDiffPipeline.from_pretrained(
+    "frankjoshua/toonyou_beta6",
+    motion_adapter=adapter,
+).to("cuda")
+
+# set scheduler
+pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
+
+# load LCM-LoRA
+pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5", adapter_name="lcm")
+pipe.load_lora_weights("guoyww/animatediff-motion-lora-zoom-in", weight_name="diffusion_pytorch_model.safetensors", adapter_name="motion-lora")
+
+pipe.set_adapters(["lcm", "motion-lora"], adapter_weights=[0.55, 1.2])
+
+prompt = "best quality, masterpiece, 1girl, looking at viewer, blurry background, upper body, contemporary, dress"
+generator = torch.manual_seed(0)
+frames = pipe(
+    prompt=prompt,
+    num_inference_steps=5,
+    guidance_scale=1.25,
+    cross_attention_kwargs={"scale": 1},
+    num_frames=24,
+    generator=generator
+).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/lcm-lora-animatediff.gif"/>
+</div>

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

@@ -1,422 +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.
--->
-
-[[open-in-colab]]
-
-# Performing inference with LCM-LoRA
-
-Latent Consistency Models (LCM) enable quality image generation in typically 2-4 steps making it possible to use diffusion models in almost real-time settings. 
-
-From the [official website](https://latent-consistency-models.github.io/):
-
-> LCMs can be distilled from any pre-trained Stable Diffusion (SD) in only 4,000 training steps (~32 A100 GPU Hours) for generating high quality 768 x 768 resolution images in 2~4 steps or even one step, significantly accelerating text-to-image generation. We employ LCM to distill the Dreamshaper-V7 version of SD in just 4,000 training iterations.
-
-For a more technical overview of LCMs, refer to [the paper](https://huggingface.co/papers/2310.04378).
-
-However, each model needs to be distilled separately for latent consistency distillation. The core idea with LCM-LoRA is to train just a few adapter layers, the adapter being LoRA in this case. 
-This way, we don't have to train the full model and keep the number of trainable parameters manageable. The resulting LoRAs can then be applied to any fine-tuned version of the model without distilling them separately.
-Additionally, the LoRAs can be applied to image-to-image, ControlNet/T2I-Adapter, inpainting, AnimateDiff etc. 
-The LCM-LoRA can also be combined with other LoRAs to generate styled images in very few steps (4-8).
-
-LCM-LoRAs are available for [stable-diffusion-v1-5](https://huggingface.co/runwayml/stable-diffusion-v1-5), [stable-diffusion-xl-base-1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0), and the [SSD-1B](https://huggingface.co/segmind/SSD-1B) model. All the checkpoints can be found in this [collection](https://huggingface.co/collections/latent-consistency/latent-consistency-models-loras-654cdd24e111e16f0865fba6).
-
-For more details about LCM-LoRA, refer to [the technical report](https://huggingface.co/papers/2311.05556).
-
-This guide shows how to perform inference with LCM-LoRAs for 
-- text-to-image
-- image-to-image
-- combined with styled LoRAs
-- ControlNet/T2I-Adapter
-- inpainting
-- AnimateDiff
-
-Before going through this guide, we'll take a look at the general workflow for performing inference with LCM-LoRAs.
-LCM-LoRAs are similar to other Stable Diffusion LoRAs so they can be used with any [`DiffusionPipeline`] that supports LoRAs.
-
-- Load the task specific pipeline and model.
-- Set the scheduler to [`LCMScheduler`].
-- Load the LCM-LoRA weights for the model.
-- Reduce the `guidance_scale` between `[1.0, 2.0]` and set the `num_inference_steps` between [4, 8].
-- Perform inference with the pipeline with the usual parameters.
-
-Let's look at how we can perform inference with LCM-LoRAs for different tasks.
-
-First, make sure you have [peft](https://github.com/huggingface/peft) installed, for better LoRA support.
-
-```bash
-pip install -U peft
-```
-
-## Text-to-image
-
-You'll use the [`StableDiffusionXLPipeline`] with the scheduler: [`LCMScheduler`] and then load the LCM-LoRA. Together with the LCM-LoRA and the scheduler, the pipeline enables a fast inference workflow overcoming the slow iterative nature of diffusion models.
-
-```python
-import torch
-from diffusers import DiffusionPipeline, LCMScheduler
-
-pipe = DiffusionPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    variant="fp16",
-    torch_dtype=torch.float16
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl")
-
-prompt = "Self-portrait oil painting, a beautiful cyborg with golden hair, 8k"
-
-generator = torch.manual_seed(42)
-image = pipe(
-    prompt=prompt, num_inference_steps=4, generator=generator, guidance_scale=1.0
-).images[0]
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdxl_t2i.png)
-
-Notice that we use only 4 steps for generation which is way less than what's typically used for standard SDXL.
-
-<Tip>
-
-You may have noticed that we set `guidance_scale=1.0`, which disables classifer-free-guidance. This is because the LCM-LoRA is trained with guidance, so the batch size does not have to be doubled in this case. This leads to a faster inference time, with the drawback that negative prompts don't have any effect on the denoising process.
-
-You can also use guidance with LCM-LoRA, but due to the nature of training the model is very sensitve to the `guidance_scale` values, high values can lead to artifacts in the generated images. In our experiments, we found that the best values are in the range of [1.0, 2.0].
-
-</Tip>
-
-### Inference with a fine-tuned model
-
-As mentioned above, the LCM-LoRA can be applied to any fine-tuned version of the model without having to distill them separately. Let's look at how we can perform inference with a fine-tuned model. In this example, we'll use the [animagine-xl](https://huggingface.co/Linaqruf/animagine-xl) model, which is a fine-tuned version of the SDXL model for generating anime.
-
-```python
-from diffusers import DiffusionPipeline, LCMScheduler
-
-pipe = DiffusionPipeline.from_pretrained(
-    "Linaqruf/animagine-xl",
-    variant="fp16",
-    torch_dtype=torch.float16
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl")
-
-prompt = "face focus, cute, masterpiece, best quality, 1girl, green hair, sweater, looking at viewer, upper body, beanie, outdoors, night, turtleneck"
-
-generator = torch.manual_seed(0)
-image = pipe(
-    prompt=prompt, num_inference_steps=4, generator=generator, guidance_scale=1.0
-).images[0]
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdxl_t2i_finetuned.png)
-
-
-## Image-to-image
-
-LCM-LoRA can be applied to image-to-image tasks too. Let's look at how we can perform image-to-image generation with LCMs. For this example we'll use the [dreamshaper-7](https://huggingface.co/Lykon/dreamshaper-7) model and the LCM-LoRA for `stable-diffusion-v1-5 `.
-
-```python
-import torch
-from diffusers import AutoPipelineForImage2Image, LCMScheduler
-from diffusers.utils import make_image_grid, load_image
-
-pipe = AutoPipelineForImage2Image.from_pretrained(
-    "Lykon/dreamshaper-7",
-    torch_dtype=torch.float16,
-    variant="fp16",
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
-
-# prepare image
-url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/img2img-init.png"
-init_image = load_image(url)
-prompt = "Astronauts in a jungle, cold color palette, muted colors, detailed, 8k"
-
-# pass prompt and image to pipeline
-generator = torch.manual_seed(0)
-image = pipe(
-    prompt,
-    image=init_image,
-    num_inference_steps=4,
-    guidance_scale=1,
-    strength=0.6,
-    generator=generator
-).images[0]
-make_image_grid([init_image, image], rows=1, cols=2)
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdv1-5_i2i.png)
-
-
-<Tip>
-
-You can get different results based on your prompt and the image you provide. To get the best results, we recommend trying different values for `num_inference_steps`, `strength`, and `guidance_scale` parameters and choose the best one.
-
-</Tip>
-
-
-## Combine with styled LoRAs
-
-LCM-LoRA can be combined with other LoRAs to generate styled-images in very few steps (4-8). In the following example, we'll use the LCM-LoRA with the [papercut LoRA](TheLastBen/Papercut_SDXL). 
-To learn more about how to combine LoRAs, refer to [this guide](https://huggingface.co/docs/diffusers/tutorials/using_peft_for_inference#combine-multiple-adapters).
-
-```python
-import torch
-from diffusers import DiffusionPipeline, LCMScheduler
-
-pipe = DiffusionPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0",
-    variant="fp16",
-    torch_dtype=torch.float16
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LoRAs
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl", adapter_name="lcm")
-pipe.load_lora_weights("TheLastBen/Papercut_SDXL", weight_name="papercut.safetensors", adapter_name="papercut")
-
-# Combine LoRAs
-pipe.set_adapters(["lcm", "papercut"], adapter_weights=[1.0, 0.8])
-
-prompt = "papercut, a cute fox"
-generator = torch.manual_seed(0)
-image = pipe(prompt, num_inference_steps=4, guidance_scale=1, generator=generator).images[0]
-image
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdx_lora_mix.png)
-
-
-## ControlNet/T2I-Adapter
-
-Let's look at how we can perform inference with ControlNet/T2I-Adapter and LCM-LoRA. 
-
-### ControlNet
-For this example, we'll use the SD-v1-5 model and the LCM-LoRA for SD-v1-5 with canny ControlNet.
-
-```python
-import torch
-import cv2
-import numpy as np
-from PIL import Image
-
-from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, LCMScheduler
-from diffusers.utils import load_image
-
-image = load_image(
-    "https://hf.co/datasets/huggingface/documentation-images/resolve/main/diffusers/input_image_vermeer.png"
-).resize((512, 512))
-
-image = np.array(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)
-
-controlnet = ControlNetModel.from_pretrained("lllyasviel/sd-controlnet-canny", torch_dtype=torch.float16)
-pipe = StableDiffusionControlNetPipeline.from_pretrained(
-    "runwayml/stable-diffusion-v1-5",
-    controlnet=controlnet,
-    torch_dtype=torch.float16,
-    safety_checker=None,
-    variant="fp16"
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
-
-generator = torch.manual_seed(0)
-image = pipe(
-    "the mona lisa",
-    image=canny_image,
-    num_inference_steps=4,
-    guidance_scale=1.5,
-    controlnet_conditioning_scale=0.8,
-    cross_attention_kwargs={"scale": 1},
-    generator=generator,
-).images[0]
-make_image_grid([canny_image, image], rows=1, cols=2)
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdv1-5_controlnet.png)
-
-
-<Tip>
-The inference parameters in this example might not work for all examples, so we recommend you to try different values for `num_inference_steps`, `guidance_scale`, `controlnet_conditioning_scale` and `cross_attention_kwargs` parameters and choose the best one. 
-</Tip>
-
-### T2I-Adapter
-
-This example shows how to use the LCM-LoRA with the [Canny T2I-Adapter](TencentARC/t2i-adapter-canny-sdxl-1.0) and SDXL.
-
-```python
-import torch
-import cv2
-import numpy as np
-from PIL import Image
-
-from diffusers import StableDiffusionXLAdapterPipeline, T2IAdapter, LCMScheduler
-from diffusers.utils import load_image, make_image_grid
-
-# Prepare image
-# Detect the canny map in low resolution to avoid high-frequency details
-image = load_image(
-    "https://huggingface.co/Adapter/t2iadapter/resolve/main/figs_SDXLV1.0/org_canny.jpg"
-).resize((384, 384))
-
-image = np.array(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).resize((1024, 1024))
-
-# load adapter
-adapter = T2IAdapter.from_pretrained("TencentARC/t2i-adapter-canny-sdxl-1.0", torch_dtype=torch.float16, varient="fp16").to("cuda")
-
-pipe = StableDiffusionXLAdapterPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-xl-base-1.0", 
-    adapter=adapter,
-    torch_dtype=torch.float16,
-    variant="fp16", 
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdxl")
-
-prompt = "Mystical fairy in real, magic, 4k picture, high quality"
-negative_prompt = "extra digit, fewer digits, cropped, worst quality, low quality, glitch, deformed, mutated, ugly, disfigured"
-
-generator = torch.manual_seed(0)
-image = pipe(
-    prompt=prompt,
-    negative_prompt=negative_prompt,
-    image=canny_image,
-    num_inference_steps=4,
-    guidance_scale=1.5, 
-    adapter_conditioning_scale=0.8, 
-    adapter_conditioning_factor=1,
-    generator=generator,
-).images[0]
-make_image_grid([canny_image, image], rows=1, cols=2)
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdxl_t2iadapter.png)
-
-
-## Inpainting
-
-LCM-LoRA can be used for inpainting as well. 
-
-```python
-import torch
-from diffusers import AutoPipelineForInpainting, LCMScheduler
-from diffusers.utils import load_image, make_image_grid
-
-pipe = AutoPipelineForInpainting.from_pretrained(
-    "runwayml/stable-diffusion-inpainting",
-    torch_dtype=torch.float16,
-    variant="fp16",
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5")
-
-# load base and mask image
-init_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/inpaint.png")
-mask_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/inpaint_mask.png")
-
-# generator = torch.Generator("cuda").manual_seed(92)
-prompt = "concept art digital painting of an elven castle, inspired by lord of the rings, highly detailed, 8k"
-generator = torch.manual_seed(0)
-image = pipe(
-    prompt=prompt,
-    image=init_image,
-    mask_image=mask_image,
-    generator=generator,
-    num_inference_steps=4,
-    guidance_scale=4, 
-).images[0]
-make_image_grid([init_image, mask_image, image], rows=1, cols=3)
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdv1-5_inpainting.png)
-
-
-## AnimateDiff
-
-[`AnimateDiff`] allows you to animate images using Stable Diffusion models. To get good results, we need to generate multiple frames (16-24), and doing this with standard SD models can be very slow. 
-LCM-LoRA can be used to speed up the process significantly, as you just need to do 4-8 steps for each frame. Let's look at how we can perform animation with LCM-LoRA and AnimateDiff.
-
-```python
-import torch
-from diffusers import MotionAdapter, AnimateDiffPipeline, DDIMScheduler, LCMScheduler
-from diffusers.utils import export_to_gif
-
-adapter = MotionAdapter.from_pretrained("diffusers/animatediff-motion-adapter-v1-5")
-pipe = AnimateDiffPipeline.from_pretrained(
-    "frankjoshua/toonyou_beta6",
-    motion_adapter=adapter,
-).to("cuda")
-
-# set scheduler
-pipe.scheduler = LCMScheduler.from_config(pipe.scheduler.config)
-
-# load LCM-LoRA
-pipe.load_lora_weights("latent-consistency/lcm-lora-sdv1-5", adapter_name="lcm")
-pipe.load_lora_weights("guoyww/animatediff-motion-lora-zoom-in", weight_name="diffusion_pytorch_model.safetensors", adapter_name="motion-lora")
-
-pipe.set_adapters(["lcm", "motion-lora"], adapter_weights=[0.55, 1.2])
-
-prompt = "best quality, masterpiece, 1girl, looking at viewer, blurry background, upper body, contemporary, dress"
-generator = torch.manual_seed(0)
-frames = pipe(
-    prompt=prompt,
-    num_inference_steps=5,
-    guidance_scale=1.25,
-    cross_attention_kwargs={"scale": 1},
-    num_frames=24,
-    generator=generator
-).frames[0]
-export_to_gif(frames, "animation.gif")
-```
-
-![](https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/lcm/lcm_sdv1-5_animatediff.gif)

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

@@ -78,7 +78,7 @@ image = pipe(
     prompt=prompt,
     num_inference_steps=4,
     guidance_scale=0,
-    eta=0.3, 
+    eta=0.3,
     generator=torch.Generator(device=device).manual_seed(0),
 ).images[0]
 ```
@@ -156,14 +156,14 @@ image = pipe(
     prompt=prompt,
     num_inference_steps=8,
     guidance_scale=0,
-    eta=0.3, 
+    eta=0.3,
     generator=torch.Generator(device=device).manual_seed(0),
 ).images[0]
 ```
 
 ![](https://github.com/jabir-zheng/TCD/raw/main/assets/animagine_xl.png)
 
-TCD-LoRA also supports other LoRAs trained on different styles. For example, let's load the [TheLastBen/Papercut_SDXL](https://huggingface.co/TheLastBen/Papercut_SDXL) LoRA and fuse it with the TCD-LoRA with the [`~loaders.UNet2DConditionLoadersMixin.set_adapters`] method. 
+TCD-LoRA also supports other LoRAs trained on different styles. For example, let's load the [TheLastBen/Papercut_SDXL](https://huggingface.co/TheLastBen/Papercut_SDXL) LoRA and fuse it with the TCD-LoRA with the [`~loaders.UNet2DConditionLoadersMixin.set_adapters`] method.
 
 > [!TIP]
 > Check out the [Merge LoRAs](merge_loras) guide to learn more about efficient merging methods.
@@ -171,7 +171,7 @@ TCD-LoRA also supports other LoRAs trained on different styles. For example, let
 ```python
 import torch
 from diffusers import StableDiffusionXLPipeline
-from scheduling_tcd import TCDScheduler 
+from scheduling_tcd import TCDScheduler
 
 device = "cuda"
 base_model_id = "stabilityai/stable-diffusion-xl-base-1.0"
@@ -191,7 +191,7 @@ image = pipe(
     prompt=prompt,
     num_inference_steps=4,
     guidance_scale=0,
-    eta=0.3, 
+    eta=0.3,
     generator=torch.Generator(device=device).manual_seed(0),
 ).images[0]
 ```
@@ -215,7 +215,7 @@ from PIL import Image
 from transformers import DPTFeatureExtractor, DPTForDepthEstimation
 from diffusers import ControlNetModel, StableDiffusionXLControlNetPipeline
 from diffusers.utils import load_image, make_image_grid
-from scheduling_tcd import TCDScheduler 
+from scheduling_tcd import TCDScheduler
 
 device = "cuda"
 depth_estimator = DPTForDepthEstimation.from_pretrained("Intel/dpt-hybrid-midas").to(device)
@@ -249,13 +249,13 @@ controlnet = ControlNetModel.from_pretrained(
     controlnet_id,
     torch_dtype=torch.float16,
     variant="fp16",
-).to(device)
+)
 pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
     base_model_id,
     controlnet=controlnet,
     torch_dtype=torch.float16,
     variant="fp16",
-).to(device)
+)
 pipe.enable_model_cpu_offload()
 
 pipe.scheduler = TCDScheduler.from_config(pipe.scheduler.config)
@@ -271,9 +271,9 @@ depth_image = get_depth_map(image)
 controlnet_conditioning_scale = 0.5  # recommended for good generalization
 
 image = pipe(
-    prompt, 
-    image=depth_image, 
-    num_inference_steps=4, 
+    prompt,
+    image=depth_image,
+    num_inference_steps=4,
     guidance_scale=0,
     eta=0.3,
     controlnet_conditioning_scale=controlnet_conditioning_scale,
@@ -290,7 +290,7 @@ grid_image = make_image_grid([depth_image, image], rows=1, cols=2)
 import torch
 from diffusers import ControlNetModel, StableDiffusionXLControlNetPipeline
 from diffusers.utils import load_image, make_image_grid
-from scheduling_tcd import TCDScheduler 
+from scheduling_tcd import TCDScheduler
 
 device = "cuda"
 base_model_id = "stabilityai/stable-diffusion-xl-base-1.0"
@@ -301,13 +301,13 @@ controlnet = ControlNetModel.from_pretrained(
     controlnet_id,
     torch_dtype=torch.float16,
     variant="fp16",
-).to(device)
+)
 pipe = StableDiffusionXLControlNetPipeline.from_pretrained(
     base_model_id,
     controlnet=controlnet,
     torch_dtype=torch.float16,
     variant="fp16",
-).to(device)
+)
 pipe.enable_model_cpu_offload()
 
 pipe.scheduler = TCDScheduler.from_config(pipe.scheduler.config)
@@ -322,9 +322,9 @@ canny_image = load_image("https://huggingface.co/datasets/hf-internal-testing/di
 controlnet_conditioning_scale = 0.5  # recommended for good generalization
 
 image = pipe(
-    prompt, 
-    image=canny_image, 
-    num_inference_steps=4, 
+    prompt,
+    image=canny_image,
+    num_inference_steps=4,
     guidance_scale=0,
     eta=0.3,
     controlnet_conditioning_scale=controlnet_conditioning_scale,
@@ -336,7 +336,7 @@ grid_image = make_image_grid([canny_image, image], rows=1, cols=2)
 ![](https://github.com/jabir-zheng/TCD/raw/main/assets/controlnet_canny_tcd.png)
 
 <Tip>
-The inference parameters in this example might not work for all examples, so we recommend you to try different values for `num_inference_steps`, `guidance_scale`, `controlnet_conditioning_scale` and `cross_attention_kwargs` parameters and choose the best one. 
+The inference parameters in this example might not work for all examples, so we recommend you to try different values for `num_inference_steps`, `guidance_scale`, `controlnet_conditioning_scale` and `cross_attention_kwargs` parameters and choose the best one.
 </Tip>
 
 </hfoption>
@@ -350,7 +350,7 @@ from diffusers import StableDiffusionXLPipeline
 from diffusers.utils import load_image, make_image_grid
 
 from ip_adapter import IPAdapterXL
-from scheduling_tcd import TCDScheduler 
+from scheduling_tcd import TCDScheduler
 
 device = "cuda"
 base_model_path = "stabilityai/stable-diffusion-xl-base-1.0"
@@ -359,8 +359,8 @@ ip_ckpt = "sdxl_models/ip-adapter_sdxl.bin"
 tcd_lora_id = "h1t/TCD-SDXL-LoRA"
 
 pipe = StableDiffusionXLPipeline.from_pretrained(
-    base_model_path, 
-    torch_dtype=torch.float16, 
+    base_model_path,
+    torch_dtype=torch.float16,
     variant="fp16"
 )
 pipe.scheduler = TCDScheduler.from_config(pipe.scheduler.config)
@@ -375,13 +375,13 @@ ref_image = load_image("https://raw.githubusercontent.com/tencent-ailab/IP-Adapt
 prompt = "best quality, high quality, wearing sunglasses"
 
 image = ip_model.generate(
-    pil_image=ref_image, 
+    pil_image=ref_image,
     prompt=prompt,
     scale=0.5,
-    num_samples=1, 
-    num_inference_steps=4, 
+    num_samples=1,
+    num_inference_steps=4,
     guidance_scale=0,
-    eta=0.3, 
+    eta=0.3,
     seed=0,
 )[0]
 

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

@@ -230,7 +230,7 @@ from diffusers.utils import load_image, make_image_grid
 
 pipeline = AutoPipelineForInpainting.from_pretrained(
     "runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16, variant="fp16"
-).to("cuda")
+)
 pipeline.enable_model_cpu_offload()
 # remove following line if xFormers is not installed or you have PyTorch 2.0 or higher installed
 pipeline.enable_xformers_memory_efficient_attention()
@@ -255,7 +255,7 @@ from diffusers.utils import load_image, make_image_grid
 
 pipeline = AutoPipelineForInpainting.from_pretrained(
     "runwayml/stable-diffusion-inpainting", torch_dtype=torch.float16, variant="fp16"
-).to("cuda")
+)
 pipeline.enable_model_cpu_offload()
 # remove following line if xFormers is not installed or you have PyTorch 2.0 or higher installed
 pipeline.enable_xformers_memory_efficient_attention()
@@ -296,7 +296,7 @@ from diffusers.utils import load_image, make_image_grid
 
 pipeline = AutoPipelineForInpainting.from_pretrained(
     "runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16, variant="fp16"
-).to("cuda")
+)
 pipeline.enable_model_cpu_offload()
 # remove following line if xFormers is not installed or you have PyTorch 2.0 or higher installed
 pipeline.enable_xformers_memory_efficient_attention()
@@ -319,7 +319,7 @@ from diffusers.utils import load_image, make_image_grid
 
 pipeline = AutoPipelineForInpainting.from_pretrained(
     "runwayml/stable-diffusion-inpainting", torch_dtype=torch.float16, variant="fp16"
-).to("cuda")
+)
 pipeline.enable_model_cpu_offload()
 # remove following line if xFormers is not installed or you have PyTorch 2.0 or higher installed
 pipeline.enable_xformers_memory_efficient_attention()

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

@@ -0,0 +1,466 @@
+<!--Copyright 2024 Marigold authors and The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# Marigold Pipelines for Computer Vision Tasks
+
+[Marigold](../api/pipelines/marigold) is a novel diffusion-based dense prediction approach, and a set of pipelines for various computer vision tasks, such as monocular depth estimation.
+
+This guide will show you how to use Marigold to obtain fast and high-quality predictions for images and videos.
+
+Each pipeline supports one Computer Vision task, which takes an input RGB image as input and produces a *prediction* of the modality of interest, such as a depth map of the input image.
+Currently, the following tasks are implemented:
+
+| Pipeline                                                                                                                                    | Predicted Modalities                                                                                             |                                                                       Demos                                                                        |
+|---------------------------------------------------------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------:|
+| [MarigoldDepthPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/marigold/pipeline_marigold_depth.py)     | [Depth](https://en.wikipedia.org/wiki/Depth_map), [Disparity](https://en.wikipedia.org/wiki/Binocular_disparity) | [Fast Demo (LCM)](https://huggingface.co/spaces/prs-eth/marigold-lcm), [Slow Original Demo (DDIM)](https://huggingface.co/spaces/prs-eth/marigold) |
+| [MarigoldNormalsPipeline](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/marigold/pipeline_marigold_normals.py) | [Surface normals](https://en.wikipedia.org/wiki/Normal_mapping)                                                  |                                   [Fast Demo (LCM)](https://huggingface.co/spaces/prs-eth/marigold-normals-lcm)                                    |
+
+The original checkpoints can be found under the [PRS-ETH](https://huggingface.co/prs-eth/) Hugging Face organization.
+These checkpoints are meant to work with diffusers pipelines and the [original codebase](https://github.com/prs-eth/marigold).
+The original code can also be used to train new checkpoints.
+
+| Checkpoint                                                                                    | Modality | Comment                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                |
+|-----------------------------------------------------------------------------------------------|----------|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| [prs-eth/marigold-v1-0](https://huggingface.co/prs-eth/marigold-v1-0)                         | Depth    | The first Marigold Depth checkpoint, which predicts *affine-invariant depth* maps. The performance of this checkpoint in benchmarks was studied in the original [paper](https://huggingface.co/papers/2312.02145). Designed to be used with the `DDIMScheduler` at inference, it requires at least 10 steps to get reliable predictions. Affine-invariant depth prediction has a range of values in each pixel between 0 (near plane) and 1 (far plane); both planes are chosen by the model as part of the inference process. See the `MarigoldImageProcessor` reference for visualization utilities. |
+| [prs-eth/marigold-depth-lcm-v1-0](https://huggingface.co/prs-eth/marigold-depth-lcm-v1-0)     | Depth    | The fast Marigold Depth checkpoint, fine-tuned from `prs-eth/marigold-v1-0`. Designed to be used with the `LCMScheduler` at inference, it requires as little as 1 step to get reliable predictions. The prediction reliability saturates at 4 steps and declines after that.                                                                                                                                                                                                                                                                                                                           |
+| [prs-eth/marigold-normals-v0-1](https://huggingface.co/prs-eth/marigold-normals-v0-1)         | Normals  | A preview checkpoint for the Marigold Normals pipeline. Designed to be used with the `DDIMScheduler` at inference, it requires at least 10 steps to get reliable predictions. The surface normals predictions are unit-length 3D vectors with values in the range from -1 to 1. *This checkpoint will be phased out after the release of `v1-0` version.*                                                                                                                                                                                                                                              |
+| [prs-eth/marigold-normals-lcm-v0-1](https://huggingface.co/prs-eth/marigold-normals-lcm-v0-1) | Normals  | The fast Marigold Normals checkpoint, fine-tuned from `prs-eth/marigold-normals-v0-1`. Designed to be used with the `LCMScheduler` at inference, it requires as little as 1 step to get reliable predictions. The prediction reliability saturates at 4 steps and declines after that. *This checkpoint will be phased out after the release of `v1-0` version.*                                                                                                                                                                                                                                       |
+The examples below are mostly given for depth prediction, but they can be universally applied with other supported modalities.
+We showcase the predictions using the same input image of Albert Einstein generated by Midjourney.
+This makes it easier to compare visualizations of the predictions across various modalities and checkpoints.
+
+<div class="flex gap-4" style="justify-content: center; width: 100%;">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://marigoldmonodepth.github.io/images/einstein.jpg"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Example input image for all Marigold pipelines
+    </figcaption>
+  </div>
+</div>
+
+### Depth Prediction Quick Start
+
+To get the first depth prediction, load `prs-eth/marigold-depth-lcm-v1-0` checkpoint into `MarigoldDepthPipeline` pipeline, put the image through the pipeline, and save the predictions:
+
+```python
+import diffusers
+import torch
+
+pipe = diffusers.MarigoldDepthPipeline.from_pretrained(
+    "prs-eth/marigold-depth-lcm-v1-0", variant="fp16", torch_dtype=torch.float16
+).to("cuda")
+
+image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+depth = pipe(image)
+
+vis = pipe.image_processor.visualize_depth(depth.prediction)
+vis[0].save("einstein_depth.png")
+
+depth_16bit = pipe.image_processor.export_depth_to_16bit_png(depth.prediction)
+depth_16bit[0].save("einstein_depth_16bit.png")
+```
+
+The visualization function for depth [`~pipelines.marigold.marigold_image_processing.MarigoldImageProcessor.visualize_depth`] applies one of [matplotlib's colormaps](https://matplotlib.org/stable/users/explain/colors/colormaps.html) (`Spectral` by default) to map the predicted pixel values from a single-channel `[0, 1]` depth range into an RGB image.
+With the `Spectral` colormap, pixels with near depth are painted red, and far pixels are assigned blue color.
+The 16-bit PNG file stores the single channel values mapped linearly from the `[0, 1]` range into `[0, 65535]`.
+Below are the raw and the visualized predictions; as can be seen, dark areas (mustache) are easier to distinguish in the visualization:
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_lcm_depth_16bit.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Predicted depth (16-bit PNG)
+    </figcaption>
+  </div>
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_lcm_depth.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Predicted depth visualization (Spectral)
+    </figcaption>
+  </div>
+</div>
+
+### Surface Normals Prediction Quick Start
+
+Load `prs-eth/marigold-normals-lcm-v0-1` checkpoint into `MarigoldNormalsPipeline` pipeline, put the image through the pipeline, and save the predictions:
+
+```python
+import diffusers
+import torch
+
+pipe = diffusers.MarigoldNormalsPipeline.from_pretrained(
+    "prs-eth/marigold-normals-lcm-v0-1", variant="fp16", torch_dtype=torch.float16
+).to("cuda")
+
+image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+normals = pipe(image)
+
+vis = pipe.image_processor.visualize_normals(normals.prediction)
+vis[0].save("einstein_normals.png")
+```
+
+The visualization function for normals [`~pipelines.marigold.marigold_image_processing.MarigoldImageProcessor.visualize_normals`] maps the three-dimensional prediction with pixel values in the range `[-1, 1]` into an RGB image.
+The visualization function supports flipping surface normals axes to make the visualization compatible with other choices of the frame of reference.
+Conceptually, each pixel is painted according to the surface normal vector in the frame of reference, where `X` axis points right, `Y` axis points up, and `Z` axis points at the viewer.
+Below is the visualized prediction:
+
+<div class="flex gap-4" style="justify-content: center; width: 100%;">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_lcm_normals.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Predicted surface normals visualization
+    </figcaption>
+  </div>
+</div>
+
+In this example, the nose tip almost certainly has a point on the surface, in which the surface normal vector points straight at the viewer, meaning that its coordinates are `[0, 0, 1]`.
+This vector maps to the RGB `[128, 128, 255]`, which corresponds to the violet-blue color.
+Similarly, a surface normal on the cheek in the right part of the image has a large `X` component, which increases the red hue.
+Points on the shoulders pointing up with a large `Y` promote green color.
+
+### Speeding up inference
+
+The above quick start snippets are already optimized for speed: they load the LCM checkpoint, use the `fp16` variant of weights and computation, and perform just one denoising diffusion step.
+The `pipe(image)` call completes in 280ms on RTX 3090 GPU.
+Internally, the input image is encoded with the Stable Diffusion VAE encoder, then the U-Net performs one denoising step, and finally, the prediction latent is decoded with the VAE decoder into pixel space.
+In this case, two out of three module calls are dedicated to converting between pixel and latent space of LDM.
+Because Marigold's latent space is compatible with the base Stable Diffusion, it is possible to speed up the pipeline call by more than 3x (85ms on RTX 3090) by using a [lightweight replacement of the SD VAE](../api/models/autoencoder_tiny):
+
+```diff
+  import diffusers
+  import torch
+  
+  pipe = diffusers.MarigoldDepthPipeline.from_pretrained(
+      "prs-eth/marigold-depth-lcm-v1-0", variant="fp16", torch_dtype=torch.float16
+  ).to("cuda")
+  
++ pipe.vae = diffusers.AutoencoderTiny.from_pretrained(
++     "madebyollin/taesd", torch_dtype=torch.float16
++ ).cuda()
+  
+  image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+  depth = pipe(image)
+```
+
+As suggested in [Optimizations](../optimization/torch2.0#torch.compile), adding `torch.compile` may squeeze extra performance depending on the target hardware:
+
+```diff
+  import diffusers
+  import torch
+  
+  pipe = diffusers.MarigoldDepthPipeline.from_pretrained(
+      "prs-eth/marigold-depth-lcm-v1-0", variant="fp16", torch_dtype=torch.float16
+  ).to("cuda")
+  
++ pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
+  
+  image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+  depth = pipe(image)
+```
+
+## Qualitative Comparison with Depth Anything
+
+With the above speed optimizations, Marigold delivers predictions with more details and faster than [Depth Anything](https://huggingface.co/docs/transformers/main/en/model_doc/depth_anything) with the largest checkpoint [LiheYoung/depth-anything-large-hf](https://huggingface.co/LiheYoung/depth-anything-large-hf):
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_lcm_depth.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Marigold LCM fp16 with Tiny AutoEncoder
+    </figcaption>
+  </div>
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/einstein_depthanything_large.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Depth Anything Large
+    </figcaption>
+  </div>
+</div>
+
+## Maximizing Precision and Ensembling
+
+Marigold pipelines have a built-in ensembling mechanism combining multiple predictions from different random latents.
+This is a brute-force way of improving the precision of predictions, capitalizing on the generative nature of diffusion.
+The ensembling path is activated automatically when the `ensemble_size` argument is set greater than `1`.
+When aiming for maximum precision, it makes sense to adjust `num_inference_steps` simultaneously with `ensemble_size`.
+The recommended values vary across checkpoints but primarily depend on the scheduler type.
+The effect of ensembling is particularly well-seen with surface normals:
+
+```python
+import diffusers
+
+model_path = "prs-eth/marigold-normals-v1-0"
+
+model_paper_kwargs = {
+	diffusers.schedulers.DDIMScheduler: {
+		"num_inference_steps": 10,
+		"ensemble_size": 10,
+	},
+	diffusers.schedulers.LCMScheduler: {
+		"num_inference_steps": 4,
+		"ensemble_size": 5,
+	},	
+}
+
+image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+
+pipe = diffusers.MarigoldNormalsPipeline.from_pretrained(model_path).to("cuda")
+pipe_kwargs = model_paper_kwargs[type(pipe.scheduler)]
+
+depth = pipe(image, **pipe_kwargs)
+
+vis = pipe.image_processor.visualize_normals(depth.prediction)
+vis[0].save("einstein_normals.png")
+```
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_lcm_normals.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Surface normals, no ensembling
+    </figcaption>
+  </div>
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_normals.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Surface normals, with ensembling
+    </figcaption>
+  </div>
+</div>
+
+As can be seen, all areas with fine-grained structurers, such as hair, got more conservative and on average more correct predictions.
+Such a result is more suitable for precision-sensitive downstream tasks, such as 3D reconstruction.
+
+## Quantitative Evaluation
+
+To evaluate Marigold quantitatively in standard leaderboards and benchmarks (such as NYU, KITTI, and other datasets), follow the evaluation protocol outlined in the paper: load the full precision fp32 model and use appropriate values for `num_inference_steps` and `ensemble_size`.
+Optionally seed randomness to ensure reproducibility. Maximizing `batch_size` will deliver maximum device utilization.
+
+```python
+import diffusers
+import torch
+
+device = "cuda"
+seed = 2024
+model_path = "prs-eth/marigold-v1-0"
+
+model_paper_kwargs = {
+	diffusers.schedulers.DDIMScheduler: {
+		"num_inference_steps": 50,
+		"ensemble_size": 10,
+	},
+	diffusers.schedulers.LCMScheduler: {
+		"num_inference_steps": 4,
+		"ensemble_size": 10,
+	},	
+}
+
+image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+
+generator = torch.Generator(device=device).manual_seed(seed)
+pipe = diffusers.MarigoldDepthPipeline.from_pretrained(model_path).to(device)
+pipe_kwargs = model_paper_kwargs[type(pipe.scheduler)]
+
+depth = pipe(image, generator=generator, **pipe_kwargs)
+
+# evaluate metrics
+```
+
+## Using Predictive Uncertainty
+
+The ensembling mechanism built into Marigold pipelines combines multiple predictions obtained from different random latents.
+As a side effect, it can be used to quantify epistemic (model) uncertainty; simply specify `ensemble_size` greater than 1 and set `output_uncertainty=True`.
+The resulting uncertainty will be available in the `uncertainty` field of the output.
+It can be visualized as follows:
+
+```python
+import diffusers
+import torch
+
+pipe = diffusers.MarigoldDepthPipeline.from_pretrained(
+    "prs-eth/marigold-depth-lcm-v1-0", variant="fp16", torch_dtype=torch.float16
+).to("cuda")
+
+image = diffusers.utils.load_image("https://marigoldmonodepth.github.io/images/einstein.jpg")
+depth = pipe(
+	image,
+	ensemble_size=10,  # any number greater than 1; higher values yield higher precision
+	output_uncertainty=True,
+)
+
+uncertainty = pipe.image_processor.visualize_uncertainty(depth.uncertainty)
+uncertainty[0].save("einstein_depth_uncertainty.png")
+```
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_depth_uncertainty.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Depth uncertainty
+    </figcaption>
+  </div>
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_einstein_normals_uncertainty.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Surface normals uncertainty
+    </figcaption>
+  </div>
+</div>
+
+The interpretation of uncertainty is easy: higher values (white) correspond to pixels, where the model struggles to make consistent predictions.
+Evidently, the depth model is the least confident around edges with discontinuity, where the object depth changes drastically.
+The surface normals model is the least confident in fine-grained structures, such as hair, and dark areas, such as the collar.
+
+## Frame-by-frame Video Processing with Temporal Consistency
+
+Due to Marigold's generative nature, each prediction is unique and defined by the random noise sampled for the latent initialization.
+This becomes an obvious drawback compared to traditional end-to-end dense regression networks, as exemplified in the following videos:
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_obama.gif"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">Input video</figcaption>
+  </div>
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_obama_depth_independent.gif"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">Marigold Depth applied to input video frames independently</figcaption>
+  </div>
+</div>
+
+To address this issue, it is possible to pass `latents` argument to the pipelines, which defines the starting point of diffusion.
+Empirically, we found that a convex combination of the very same starting point noise latent and the latent corresponding to the previous frame prediction give sufficiently smooth results, as implemented in the snippet below:
+
+```python
+import imageio
+from PIL import Image
+from tqdm import tqdm
+import diffusers
+import torch
+
+device = "cuda"
+path_in = "obama.mp4"
+path_out = "obama_depth.gif"
+
+pipe = diffusers.MarigoldDepthPipeline.from_pretrained(
+    "prs-eth/marigold-depth-lcm-v1-0", variant="fp16", torch_dtype=torch.float16
+).to(device)
+pipe.vae = diffusers.AutoencoderTiny.from_pretrained(
+    "madebyollin/taesd", torch_dtype=torch.float16
+).to(device)
+pipe.set_progress_bar_config(disable=True)
+
+with imageio.get_reader(path_in) as reader:
+    size = reader.get_meta_data()['size']
+    last_frame_latent = None
+    latent_common = torch.randn(
+        (1, 4, 768 * size[1] // (8 * max(size)), 768 * size[0] // (8 * max(size)))
+    ).to(device=device, dtype=torch.float16)
+
+    out = []
+    for frame_id, frame in tqdm(enumerate(reader), desc="Processing Video"):
+        frame = Image.fromarray(frame)
+        latents = latent_common
+        if last_frame_latent is not None:
+            latents = 0.9 * latents + 0.1 * last_frame_latent
+
+        depth = pipe(
+			frame, match_input_resolution=False, latents=latents, output_latent=True
+        )
+        last_frame_latent = depth.latent
+        out.append(pipe.image_processor.visualize_depth(depth.prediction)[0])
+
+    diffusers.utils.export_to_gif(out, path_out, fps=reader.get_meta_data()['fps'])
+```
+
+Here, the diffusion process starts from the given computed latent.
+The pipeline sets `output_latent=True` to access `out.latent` and computes its contribution to the next frame's latent initialization.
+The result is much more stable now:
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_obama_depth_independent.gif"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">Marigold Depth applied to input video frames independently</figcaption>
+  </div>
+  <div style="flex: 1 1 50%; max-width: 50%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/marigold_obama_depth_consistent.gif"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">Marigold Depth with forced latents initialization</figcaption>
+  </div>
+</div>
+
+## Marigold for ControlNet
+
+A very common application for depth prediction with diffusion models comes in conjunction with ControlNet.
+Depth crispness plays a crucial role in obtaining high-quality results from ControlNet.
+As seen in comparisons with other methods above, Marigold excels at that task.
+The snippet below demonstrates how to load an image, compute depth, and pass it into ControlNet in a compatible format:
+
+```python
+import torch
+import diffusers
+
+device = "cuda"
+generator = torch.Generator(device=device).manual_seed(2024)
+image = diffusers.utils.load_image(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_depth_source.png"
+)
+
+pipe = diffusers.MarigoldDepthPipeline.from_pretrained(
+    "prs-eth/marigold-depth-lcm-v1-0", torch_dtype=torch.float16, variant="fp16"
+).to("cuda")
+
+depth_image = pipe(image, generator=generator).prediction
+depth_image = pipe.image_processor.visualize_depth(depth_image, color_map="binary")
+depth_image[0].save("motorcycle_controlnet_depth.png")
+
+controlnet = diffusers.ControlNetModel.from_pretrained(
+    "diffusers/controlnet-depth-sdxl-1.0", torch_dtype=torch.float16, variant="fp16"
+).to("cuda")
+pipe = diffusers.StableDiffusionXLControlNetPipeline.from_pretrained(
+    "SG161222/RealVisXL_V4.0", torch_dtype=torch.float16, variant="fp16", controlnet=controlnet
+).to("cuda")
+pipe.scheduler = diffusers.DPMSolverMultistepScheduler.from_config(pipe.scheduler.config, use_karras_sigmas=True)
+
+controlnet_out = pipe(
+    prompt="high quality photo of a sports bike, city",
+    negative_prompt="",
+    guidance_scale=6.5,
+    num_inference_steps=25,
+    image=depth_image,
+    controlnet_conditioning_scale=0.7,
+    control_guidance_end=0.7,
+    generator=generator,
+).images
+controlnet_out[0].save("motorcycle_controlnet_out.png")
+```
+
+<div class="flex gap-4">
+  <div style="flex: 1 1 33%; max-width: 33%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/controlnet_depth_source.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Input image
+    </figcaption>
+  </div>
+  <div style="flex: 1 1 33%; max-width: 33%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/motorcycle_controlnet_depth.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      Depth in the format compatible with ControlNet
+    </figcaption>
+  </div>
+  <div style="flex: 1 1 33%; max-width: 33%;">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/marigold/motorcycle_controlnet_out.png"/>
+    <figcaption class="mt-1 text-center text-sm text-gray-500">
+      ControlNet generation, conditioned on depth and prompt: "high quality photo of a sports bike, city"
+    </figcaption>
+  </div>
+</div>
+
+Hopefully, you will find Marigold useful for solving your downstream tasks, be it a part of a more broad generative workflow, or a perception task, such as 3D reconstruction.

docs/source/en/using-diffusers/other-formats.md

@@ -10,156 +10,86 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# Load different Stable Diffusion formats
+# Model files and layouts
 
 [[open-in-colab]]
 
-Stable Diffusion models are available in different formats depending on the framework they're trained and saved with, and where you download them from. Converting these formats for use in 🤗 Diffusers allows you to use all the features supported by the library, such as [using different schedulers](schedulers) for inference, [building your custom pipeline](write_own_pipeline), and a variety of techniques and methods for [optimizing inference speed](../optimization/opt_overview).
+Diffusion models are saved in various file types and organized in different layouts. Diffusers stores model weights as safetensors files in *Diffusers-multifolder* layout and it also supports loading files (like safetensors and ckpt files) from a *single-file* layout which is commonly used in the diffusion ecosystem.
 
-<Tip>
+Each layout has its own benefits and use cases, and this guide will show you how to load the different files and layouts, and how to convert them.
 
-We highly recommend using the `.safetensors` format because it is more secure than traditional pickled files which are vulnerable and can be exploited to execute any code on your machine (learn more in the [Load safetensors](using_safetensors) guide).
+## Files
 
-</Tip>
+PyTorch model weights are typically saved with Python's [pickle](https://docs.python.org/3/library/pickle.html) utility as ckpt or bin files. However, pickle is not secure and pickled files may contain malicious code that can be executed. This vulnerability is a serious concern given the popularity of model sharing. To address this security issue, the [Safetensors](https://hf.co/docs/safetensors) library was developed as a secure alternative to pickle, which saves models as safetensors files.
 
-This guide will show you how to convert other Stable Diffusion formats to be compatible with 🤗 Diffusers.
+### safetensors
 
-## PyTorch .ckpt
+> [!TIP]
+> Learn more about the design decisions and why safetensor files are preferred for saving and loading model weights in the [Safetensors audited as really safe and becoming the default](https://blog.eleuther.ai/safetensors-security-audit/) blog post.
 
-The checkpoint - or `.ckpt` - format is commonly used to store and save models. The `.ckpt` file contains the entire model and is typically several GBs in size. While you can load and use a `.ckpt` file directly with the [`~StableDiffusionPipeline.from_single_file`] method, it is generally better to convert the `.ckpt` file to 🤗 Diffusers so both formats are available.
+[Safetensors](https://hf.co/docs/safetensors) is a safe and fast file format for securely storing and loading tensors. Safetensors restricts the header size to limit certain types of attacks, supports lazy loading (useful for distributed setups), and has generally faster loading speeds.
 
-There are two options for converting a `.ckpt` file: use a Space to convert the checkpoint or convert the `.ckpt` file with a script.
+Make sure you have the [Safetensors](https://hf.co/docs/safetensors) library installed.
 
-### Convert with a Space
-
-The easiest and most convenient way to convert a `.ckpt` file is to use the [SD to Diffusers](https://huggingface.co/spaces/diffusers/sd-to-diffusers) Space. You can follow the instructions on the Space to convert the `.ckpt` file.
-
-This approach works well for basic models, but it may struggle with more customized models. You'll know the Space failed if it returns an empty pull request or error. In this case, you can try converting the `.ckpt` file with a script.
-
-### Convert with a script
-
-🤗 Diffusers provides a [conversion script](https://github.com/huggingface/diffusers/blob/main/scripts/convert_original_stable_diffusion_to_diffusers.py) for converting `.ckpt` files. This approach is more reliable than the Space above.
-
-Before you start, make sure you have a local clone of 🤗 Diffusers to run the script and log in to your Hugging Face account so you can open pull requests and push your converted model to the Hub.
-
-```bash
-huggingface-cli login
+```py
+!pip install safetensors
 ```
 
-To use the script:
+Safetensors stores weights in a safetensors file. Diffusers loads safetensors files by default if they're available and the Safetensors library is installed. There are two ways safetensors files can be organized:
 
-1. Git clone the repository containing the `.ckpt` file you want to convert. For this example, let's convert this [TemporalNet](https://huggingface.co/CiaraRowles/TemporalNet) `.ckpt` file:
+1. Diffusers-multifolder layout: there may be several separate safetensors files, one for each pipeline component (text encoder, UNet, VAE), organized in subfolders (check out the [runwayml/stable-diffusion-v1-5](https://hf.co/runwayml/stable-diffusion-v1-5/tree/main) repository as an example)
+2. single-file layout: all the model weights may be saved in a single file (check out the [WarriorMama777/OrangeMixs](https://hf.co/WarriorMama777/OrangeMixs/tree/main/Models/AbyssOrangeMix) repository as an example)
 
-```bash
-git lfs install
-git clone https://huggingface.co/CiaraRowles/TemporalNet
-```
+<hfoptions id="safetensors">
+<hfoption id="multifolder">
 
-2. Open a pull request on the repository where you're converting the checkpoint from:
-
-```bash
-cd TemporalNet && git fetch origin refs/pr/13:pr/13
-git checkout pr/13
-```
-
-3. There are several input arguments to configure in the conversion script, but the most important ones are:
-
-    - `checkpoint_path`: the path to the `.ckpt` file to convert.
-    - `original_config_file`: a YAML file defining the configuration of the original architecture. If you can't find this file, try searching for the YAML file in the GitHub repository where you found the `.ckpt` file.
-    - `dump_path`: the path to the converted model.
-
-        For example, you can take the `cldm_v15.yaml` file from the [ControlNet](https://github.com/lllyasviel/ControlNet/tree/main/models) repository because the TemporalNet model is a Stable Diffusion v1.5 and ControlNet model.
-
-4. Now you can run the script to convert the `.ckpt` file:
-
-```bash
-python ../diffusers/scripts/convert_original_stable_diffusion_to_diffusers.py --checkpoint_path temporalnetv3.ckpt --original_config_file cldm_v15.yaml --dump_path ./ --controlnet
-```
-
-5. Once the conversion is done, upload your converted model and test out the resulting [pull request](https://huggingface.co/CiaraRowles/TemporalNet/discussions/13)!
-
-```bash
-git push origin pr/13:refs/pr/13
-```
-
-## Keras .pb or .h5
-
-<Tip warning={true}>
-
-🧪 This is an experimental feature. Only Stable Diffusion v1 checkpoints are supported by the Convert KerasCV Space at the moment.
-
-</Tip>
-
-[KerasCV](https://keras.io/keras_cv/) supports training for [Stable Diffusion](https://github.com/keras-team/keras-cv/blob/master/keras_cv/models/stable_diffusion) v1 and v2. However, it offers limited support for experimenting with Stable Diffusion models for inference and deployment whereas 🤗 Diffusers has a more complete set of features for this purpose, such as different [noise schedulers](https://huggingface.co/docs/diffusers/using-diffusers/schedulers), [flash attention](https://huggingface.co/docs/diffusers/optimization/xformers), and [other
-optimization techniques](https://huggingface.co/docs/diffusers/optimization/fp16).
-
-The [Convert KerasCV](https://huggingface.co/spaces/sayakpaul/convert-kerascv-sd-diffusers) Space converts `.pb` or `.h5` files to PyTorch, and then wraps them in a [`StableDiffusionPipeline`] so it is ready for inference. The converted checkpoint is stored in a repository on the Hugging Face Hub.
-
-For this example, let's convert the [`sayakpaul/textual-inversion-kerasio`](https://huggingface.co/sayakpaul/textual-inversion-kerasio/tree/main) checkpoint which was trained with Textual Inversion. It uses the special token `<my-funny-cat>` to personalize images with cats.
-
-The Convert KerasCV Space allows you to input the following:
-
-* Your Hugging Face token.
-* Paths to download the UNet and text encoder weights from. Depending on how the model was trained, you don't necessarily need to provide the paths to both the UNet and text encoder. For example, Textual Inversion only requires the embeddings from the text encoder and a text-to-image model only requires the UNet weights.
-* Placeholder token is only applicable for textual inversion models.
-* The `output_repo_prefix` is the name of the repository where the converted model is stored.
-
-Click the **Submit** button to automatically convert the KerasCV checkpoint! Once the checkpoint is successfully converted, you'll see a link to the new repository containing the converted checkpoint. Follow the link to the new repository, and you'll see the Convert KerasCV Space generated a model card with an inference widget to try out the converted model.
-
-If you prefer to run inference with code, click on the **Use in Diffusers** button in the upper right corner of the model card to copy and paste the code snippet:
+Use the [`~DiffusionPipeline.from_pretrained`] method to load a model with safetensors files stored in multiple folders.
 
 ```py
 from diffusers import DiffusionPipeline
 
 pipeline = DiffusionPipeline.from_pretrained(
-    "sayakpaul/textual-inversion-cat-kerascv_sd_diffusers_pipeline", use_safetensors=True
+    "runwayml/stable-diffusion-v1-5",
+    use_safetensors=True
 )
 ```
 
-Then, you can generate an image like:
+</hfoption>
+<hfoption id="single file">
+
+Use the [`~loaders.FromSingleFileMixin.from_single_file`] method to load a model with all the weights stored in a single safetensors file.
 
 ```py
-from diffusers import DiffusionPipeline
+from diffusers import StableDiffusionPipeline
 
-pipeline = DiffusionPipeline.from_pretrained(
-    "sayakpaul/textual-inversion-cat-kerascv_sd_diffusers_pipeline", use_safetensors=True
+pipeline = StableDiffusionPipeline.from_single_file(
+    "https://huggingface.co/WarriorMama777/OrangeMixs/blob/main/Models/AbyssOrangeMix/AbyssOrangeMix.safetensors"
 )
-pipeline.to("cuda")
-
-placeholder_token = "<my-funny-cat-token>"
-prompt = f"two {placeholder_token} getting married, photorealistic, high quality"
-image = pipeline(prompt, num_inference_steps=50).images[0]
 ```
 
-## A1111 LoRA files
+</hfoption>
+</hfoptions>
 
-[Automatic1111](https://github.com/AUTOMATIC1111/stable-diffusion-webui) (A1111) is a popular web UI for Stable Diffusion that supports model sharing platforms like [Civitai](https://civitai.com/). Models trained with the Low-Rank Adaptation (LoRA) technique are especially popular because they're fast to train and have a much smaller file size than a fully finetuned model. 🤗 Diffusers supports loading A1111 LoRA checkpoints with [`~loaders.LoraLoaderMixin.load_lora_weights`]:
+#### LoRA files
+
+[LoRA](https://hf.co/docs/peft/conceptual_guides/adapter#low-rank-adaptation-lora) is a lightweight adapter that is fast and easy to train, making them especially popular for generating images in a certain way or style. These adapters are commonly stored in a safetensors file, and are widely popular on model sharing platforms like [civitai](https://civitai.com/).
+
+LoRAs are loaded into a base model with the [`~loaders.LoraLoaderMixin.load_lora_weights`] method.
 
 ```py
 from diffusers import StableDiffusionXLPipeline
 import torch
 
+# base model
 pipeline = StableDiffusionXLPipeline.from_pretrained(
     "Lykon/dreamshaper-xl-1-0", torch_dtype=torch.float16, variant="fp16"
 ).to("cuda")
-```
 
-Download a LoRA checkpoint from Civitai; this example uses the [Blueprintify SD XL 1.0](https://civitai.com/models/150986/blueprintify-sd-xl-10) checkpoint, but feel free to try out any LoRA checkpoint!
+# download LoRA weights
+!wget https://civitai.com/api/download/models/168776 -O blueprintify.safetensors
 
-```py
-# uncomment to download the safetensor weights
-#!wget https://civitai.com/api/download/models/168776 -O blueprintify.safetensors
-```
-
-Load the LoRA checkpoint into the pipeline with the [`~loaders.LoraLoaderMixin.load_lora_weights`] method:
-
-```py
+# load LoRA weights
 pipeline.load_lora_weights(".", weight_name="blueprintify.safetensors")
-```
-
-Now you can use the pipeline to generate images:
-
-```py
 prompt = "bl3uprint, a highly detailed blueprint of the empire state building, explaining how to build all parts, many txt, blueprint grid backdrop"
 negative_prompt = "lowres, cropped, worst quality, low quality, normal quality, artifacts, signature, watermark, username, blurry, more than one bridge, bad architecture"
 
@@ -174,3 +104,166 @@ image
 <div class="flex justify-center">
     <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/blueprint-lora.png"/>
 </div>
+
+### ckpt
+
+> [!WARNING]
+> Pickled files may be unsafe because they can be exploited to execute malicious code. It is recommended to use safetensors files instead where possible, or convert the weights to safetensors files.
+
+PyTorch's [torch.save](https://pytorch.org/docs/stable/generated/torch.save.html) function uses Python's [pickle](https://docs.python.org/3/library/pickle.html) utility to serialize and save models. These files are saved as a ckpt file and they contain the entire model's weights.
+
+Use the [`~loaders.FromSingleFileMixin.from_single_file`] method to directly load a ckpt file.
+
+```py
+from diffusers import StableDiffusionPipeline
+
+pipeline = StableDiffusionPipeline.from_single_file(
+    "https://huggingface.co/runwayml/stable-diffusion-v1-5/blob/main/v1-5-pruned.ckpt"
+)
+```
+
+## Storage layout
+
+There are two ways model files are organized, either in a Diffusers-multifolder layout or in a single-file layout. The Diffusers-multifolder layout is the default, and each component file (text encoder, UNet, VAE) is stored in a separate subfolder. Diffusers also supports loading models from a single-file layout where all the components are bundled together.
+
+### Diffusers-multifolder
+
+The Diffusers-multifolder layout is the default storage layout for Diffusers. Each component's (text encoder, UNet, VAE) weights are stored in a separate subfolder. The weights can be stored as safetensors or ckpt files.
+
+<div class="flex flex-row gap-4">
+  <div class="flex-1">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/multifolder-layout.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">multifolder layout</figcaption>
+  </div>
+  <div class="flex-1">
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/multifolder-unet.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">UNet subfolder</figcaption>
+  </div>
+</div>
+
+To load from Diffusers-multifolder layout, use the [`~DiffusionPipeline.from_pretrained`] method.
+
+```py
+from diffusers import DiffusionPipeline
+
+pipeline = DiffusionPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-xl-base-1.0",
+    torch_dtype=torch.float16,
+    variant="fp16",
+    use_safetensors=True,
+).to("cuda")
+```
+
+Benefits of using the Diffusers-multifolder layout include:
+
+1. Faster to load each component file individually or in parallel.
+2. Reduced memory usage because you only load the components you need. For example, models like [SDXL Turbo](https://hf.co/stabilityai/sdxl-turbo), [SDXL Lightning](https://hf.co/ByteDance/SDXL-Lightning), and [Hyper-SD](https://hf.co/ByteDance/Hyper-SD) have the same components except for the UNet. You can reuse their shared components with the [`~DiffusionPipeline.from_pipe`] method without consuming any additional memory (take a look at the [Reuse a pipeline](./loading#reuse-a-pipeline) guide) and only load the UNet. This way, you don't need to download redundant components and unnecessarily use more memory.
+
+    ```py
+    import torch
+    from diffusers import StableDiffusionXLPipeline, UNet2DConditionModel, EulerDiscreteScheduler
+
+    # download one model
+    sdxl_pipeline = StableDiffusionXLPipeline.from_pretrained(
+        "stabilityai/stable-diffusion-xl-base-1.0",
+        torch_dtype=torch.float16,
+        variant="fp16",
+        use_safetensors=True,
+    ).to("cuda")
+
+    # switch UNet for another model
+    unet = UNet2DConditionModel.from_pretrained(
+        "stabilityai/sdxl-turbo",
+        subfolder="unet",
+        torch_dtype=torch.float16,
+        variant="fp16",
+        use_safetensors=True
+    )
+    # reuse all the same components in new model except for the UNet
+    turbo_pipeline = StableDiffusionXLPipeline.from_pipe(
+        sdxl_pipeline, unet=unet,
+    ).to("cuda")
+    turbo_pipeline.scheduler = EulerDiscreteScheduler.from_config(
+        turbo_pipeline.scheduler.config,
+        timestep+spacing="trailing"
+    )
+    image = turbo_pipeline(
+        "an astronaut riding a unicorn on mars",
+        num_inference_steps=1,
+        guidance_scale=0.0,
+    ).images[0]
+    image
+    ```
+
+3. Reduced storage requirements because if a component, such as the SDXL [VAE](https://hf.co/madebyollin/sdxl-vae-fp16-fix), is shared across multiple models, you only need to download and store a single copy of it instead of downloading and storing it multiple times. For 10 SDXL models, this can save ~3.5GB of storage. The storage savings is even greater for newer models like PixArt Sigma, where the [text encoder](https://hf.co/PixArt-alpha/PixArt-Sigma-XL-2-1024-MS/tree/main/text_encoder) alone is ~19GB!
+4. Flexibility to replace a component in the model with a newer or better version.
+
+    ```py
+    from diffusers import DiffusionPipeline, AutoencoderKL
+
+    vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16, use_safetensors=True)
+    pipeline = DiffusionPipeline.from_pretrained(
+        "stabilityai/stable-diffusion-xl-base-1.0",
+        vae=vae,
+        torch_dtype=torch.float16,
+        variant="fp16",
+        use_safetensors=True,
+    ).to("cuda")
+    ```
+
+5. More visibility and information about a model's components, which are stored in a [config.json](https://hf.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/unet/config.json) file in each component subfolder.
+
+### Single-file
+
+The single-file layout stores all the model weights in a single file. All the model components (text encoder, UNet, VAE) weights are kept together instead of separately in subfolders. This can be a safetensors or ckpt file.
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/single-file-layout.png"/>
+</div>
+
+To load from a single-file layout, use the [`~loaders.FromSingleFileMixin.from_single_file`] method.
+
+```py
+import torch
+from diffusers import StableDiffusionXLPipeline
+
+pipeline = StableDiffusionXLPipeline.from_single_file(
+    "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0.safetensors",
+    torch_dtype=torch.float16,
+    variant="fp16",
+    use_safetensors=True,
+).to("cuda")
+```
+
+Benefits of using a single-file layout include:
+
+1. Easy compatibility with diffusion interfaces such as [ComfyUI](https://github.com/comfyanonymous/ComfyUI) or [Automatic1111](https://github.com/AUTOMATIC1111/stable-diffusion-webui) which commonly use a single-file layout.
+2. Easier to manage (download and share) a single file.
+
+## Convert layout and files
+
+Diffusers provides many scripts and methods to convert storage layouts and file formats to enable broader support across the diffusion ecosystem.
+
+Take a look at the [diffusers/scripts](https://github.com/huggingface/diffusers/tree/main/scripts) collection to find a script that fits your conversion needs.
+
+> [!TIP]
+> Scripts that have "`to_diffusers`" appended at the end mean they convert a model to the Diffusers-multifolder layout. Each script has their own specific set of arguments for configuring the conversion, so make sure you check what arguments are available!
+
+For example, to convert a Stable Diffusion XL model stored in Diffusers-multifolder layout to a single-file layout, run the [convert_diffusers_to_original_sdxl.py](https://github.com/huggingface/diffusers/blob/main/scripts/convert_diffusers_to_original_sdxl.py) script. Provide the path to the model to convert, and the path to save the converted model to. You can optionally specify whether you want to save the model as a safetensors file and whether to save the model in half-precision.
+
+```bash
+python convert_diffusers_to_original_sdxl.py --model_path path/to/model/to/convert --checkpoint_path path/to/save/model/to --use_safetensors
+```
+
+You can also save a model to Diffusers-multifolder layout with the [`~DiffusionPipeline.save_pretrained`] method. This creates a directory for you if it doesn't already exist, and it also saves the files as a safetensors file by default.
+
+```py
+from diffusers import StableDiffusionXLPipeline
+
+pipeline = StableDiffusionXLPipeline.from_single_file(
+    "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/sd_xl_base_1.0.safetensors",
+)
+pipeline.save_pretrained()
+```
+
+Lastly, there are also Spaces, such as [SD To Diffusers](https://hf.co/spaces/diffusers/sd-to-diffusers) and [SD-XL To Diffusers](https://hf.co/spaces/diffusers/sdxl-to-diffusers), that provide a more user-friendly interface for converting models to Diffusers-multifolder layout. This is the easiest and most convenient option for converting layouts, and it'll open a PR on your model repository with the converted files. However, this option is not as reliable as running a script, and the Space may fail for more complicated models.

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

@@ -0,0 +1,235 @@
+<!--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.
+-->
+
+# Scheduler features
+
+The scheduler is an important component of any diffusion model because it controls the entire denoising (or sampling) process. There are many types of schedulers, some are optimized for speed and some for quality. With Diffusers, you can modify the scheduler configuration to use custom noise schedules, sigmas, and rescale the noise schedule. Changing these parameters can have profound effects on inference quality and speed.
+
+This guide will demonstrate how to use these features to improve inference quality.
+
+> [!TIP]
+> Diffusers currently only supports the `timesteps` and `sigmas` parameters for a select list of schedulers and pipelines. Feel free to open a [feature request](https://github.com/huggingface/diffusers/issues/new/choose) if you want to extend these parameters to a scheduler and pipeline that does not currently support it!
+
+## Timestep schedules
+
+The timestep or noise schedule determines the amount of noise at each sampling step. The scheduler uses this to generate an image with the corresponding amount of noise at each step. The timestep schedule is generated from the scheduler's default configuration, but you can customize the scheduler to use new and optimized sampling schedules that aren't in Diffusers yet.
+
+For example, [Align Your Steps (AYS)](https://research.nvidia.com/labs/toronto-ai/AlignYourSteps/) is a method for optimizing a sampling schedule to generate a high-quality image in as little as 10 steps. The optimal [10-step schedule](https://github.com/huggingface/diffusers/blob/a7bf77fc284810483f1e60afe34d1d27ad91ce2e/src/diffusers/schedulers/scheduling_utils.py#L51) for Stable Diffusion XL is:
+
+```py
+from diffusers.schedulers import AysSchedules
+
+sampling_schedule = AysSchedules["StableDiffusionXLTimesteps"]
+print(sampling_schedule)
+"[999, 845, 730, 587, 443, 310, 193, 116, 53, 13]"
+```
+
+You can use the AYS sampling schedule in a pipeline by passing it to the `timesteps` parameter.
+
+```py
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "SG161222/RealVisXL_V4.0",
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
+pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config, algorithm_type="sde-dpmsolver++")
+
+prompt = "A cinematic shot of a cute little rabbit wearing a jacket and doing a thumbs up"
+generator = torch.Generator(device="cpu").manual_seed(2487854446)
+image = pipeline(
+    prompt=prompt,
+    negative_prompt="",
+    generator=generator,
+    timesteps=sampling_schedule,
+).images[0]
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/ays.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">AYS timestep schedule 10 steps</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/10.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">Linearly-spaced timestep schedule 10 steps</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/YiYiXu/testing-images/resolve/main/25.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">Linearly-spaced timestep schedule 25 steps</figcaption>
+  </div>
+</div>
+
+## Timestep spacing
+
+The way sample steps are selected in the schedule can affect the quality of the generated image, especially with respect to [rescaling the noise schedule](#rescale-noise-schedule), which can enable a model to generate much brighter or darker images. Diffusers provides three timestep spacing methods:
+
+- `leading` creates evenly spaced steps
+- `linspace` includes the first and last steps and evenly selects the remaining intermediate steps
+- `trailing` only includes the last step and evenly selects the remaining intermediate steps starting from the end
+
+It is recommended to use the `trailing` spacing method because it generates higher quality images with more details when there are fewer sample steps. But the difference in quality is not as obvious for more standard sample step values.
+
+```py
+import torch
+from diffusers import StableDiffusionXLPipeline, DPMSolverMultistepScheduler
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "SG161222/RealVisXL_V4.0",
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
+pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config, timestep_spacing="trailing")
+
+prompt = "A cinematic shot of a cute little black cat sitting on a pumpkin at night"
+generator = torch.Generator(device="cpu").manual_seed(2487854446)
+image = pipeline(
+    prompt=prompt,
+    negative_prompt="",
+    generator=generator,
+    num_inference_steps=5,
+).images[0]
+image
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/trailing_spacing.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">trailing spacing after 5 steps</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/leading_spacing.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">leading spacing after 5 steps</figcaption>
+  </div>
+</div>
+
+## Sigmas
+
+The `sigmas` parameter is the amount of noise added at each timestep according to the timestep schedule. Like the `timesteps` parameter, you can customize the `sigmas` parameter to control how much noise is added at each step. When you use a custom `sigmas` value, the `timesteps` are calculated from the custom `sigmas` value and the default scheduler configuration is ignored.
+
+For example, you can manually pass the [sigmas](https://github.com/huggingface/diffusers/blob/6529ee67ec02fcf58d2fd9242164ea002b351d75/src/diffusers/schedulers/scheduling_utils.py#L55) for something like the 10-step AYS schedule from before to the pipeline.
+
+```py
+import torch
+
+from diffusers import DiffusionPipeline, EulerDiscreteScheduler
+
+model_id = "stabilityai/stable-diffusion-xl-base-1.0"
+pipeline = DiffusionPipeline.from_pretrained(
+  "stabilityai/stable-diffusion-xl-base-1.0",
+  torch_dtype=torch.float16,
+  variant="fp16",
+).to("cuda")
+pipeline.scheduler = EulerDiscreteScheduler.from_config(pipeline.scheduler.config)
+
+sigmas = [14.615, 6.315, 3.771, 2.181, 1.342, 0.862, 0.555, 0.380, 0.234, 0.113, 0.0]
+prompt = "anthropomorphic capybara wearing a suit and working with a computer"
+generator = torch.Generator(device='cuda').manual_seed(123)
+image = pipeline(
+    prompt=prompt, 
+    num_inference_steps=10,
+    sigmas=sigmas,
+    generator=generator
+).images[0]
+```
+
+When you take a look at the scheduler's `timesteps` parameter, you'll see that it is the same as the AYS timestep schedule because the `timestep` schedule is calculated from the `sigmas`.
+
+```py
+print(f" timesteps: {pipe.scheduler.timesteps}")
+"timesteps: tensor([999., 845., 730., 587., 443., 310., 193., 116.,  53.,  13.], device='cuda:0')"
+```
+
+### Karras sigmas
+
+> [!TIP]
+> Refer to the scheduler API [overview](../api/schedulers/overview) for a list of schedulers that support Karras sigmas.
+>
+> Karras sigmas should not be used for models that weren't trained with them. For example, the base Stable Diffusion XL model shouldn't use Karras sigmas but the [DreamShaperXL](https://hf.co/Lykon/dreamshaper-xl-1-0) model can since they are trained with Karras sigmas.
+
+Karras scheduler's use the timestep schedule and sigmas from the [Elucidating the Design Space of Diffusion-Based Generative Models](https://hf.co/papers/2206.00364) paper. This scheduler variant applies a smaller amount of noise per step as it approaches the end of the sampling process compared to other schedulers, and can increase the level of details in the generated image.
+
+Enable Karras sigmas by setting `use_karras_sigmas=True` in the scheduler.
+
+```py
+import torch
+from diffusers import StableDiffusionXLPipeline, DPMSolverMultistepScheduler
+
+pipeline = StableDiffusionXLPipeline.from_pretrained(
+    "SG161222/RealVisXL_V4.0",
+    torch_dtype=torch.float16,
+    variant="fp16",
+).to("cuda")
+pipeline.scheduler = DPMSolverMultistepScheduler.from_config(pipeline.scheduler.config, algorithm_type="sde-dpmsolver++", use_karras_sigmas=True)
+
+prompt = "A cinematic shot of a cute little rabbit wearing a jacket and doing a thumbs up"
+generator = torch.Generator(device="cpu").manual_seed(2487854446)
+image = pipeline(
+    prompt=prompt,
+    negative_prompt="",
+    generator=generator,
+).images[0]
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/karras_sigmas_true.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">Karras sigmas enabled</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/stevhliu/testing-images/resolve/main/karras_sigmas_false.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">Karras sigmas disabled</figcaption>
+  </div>
+</div>
+
+## Rescale noise schedule
+
+In the [Common Diffusion Noise Schedules and Sample Steps are Flawed](https://hf.co/papers/2305.08891) paper, the authors discovered that common noise schedules allowed some signal to leak into the last timestep. This signal leakage at inference can cause models to only generate images with medium brightness. By enforcing a zero signal-to-noise ratio (SNR) for the timstep schedule and sampling from the last timestep, the model can be improved to generate very bright or dark images.
+
+> [!TIP]
+> For inference, you need a model that has been trained with *v_prediction*. To train your own model with *v_prediction*, add the following flag to the [train_text_to_image.py](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py) or [train_text_to_image_lora.py](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_lora.py) scripts.
+>
+> ```bash
+> --prediction_type="v_prediction"
+> ```
+
+For example, load the [ptx0/pseudo-journey-v2](https://hf.co/ptx0/pseudo-journey-v2) checkpoint which was trained with `v_prediction` and the [`DDIMScheduler`]. Configure the following parameters in the [`DDIMScheduler`]:
+
+* `rescale_betas_zero_snr=True` to rescale the noise schedule to zero SNR
+* `timestep_spacing="trailing"` to start sampling from the last timestep
+
+Set `guidance_rescale` in the pipeline to prevent over-exposure. A lower value increases brightness but some of the details may appear washed out.
+
+```py
+from diffusers import DiffusionPipeline, DDIMScheduler
+
+pipeline = DiffusionPipeline.from_pretrained("ptx0/pseudo-journey-v2", use_safetensors=True)
+
+pipeline.scheduler = DDIMScheduler.from_config(
+    pipeline.scheduler.config, rescale_betas_zero_snr=True, timestep_spacing="trailing"
+)
+pipeline.to("cuda")
+prompt = "cinematic photo of a snowy mountain at night with the northern lights aurora borealis overhead, 35mm photograph, film, professional, 4k, highly detailed"
+generator = torch.Generator(device="cpu").manual_seed(23)
+image = pipeline(prompt, guidance_rescale=0.7, generator=generator).images[0]
+image
+```
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/no-zero-snr.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">default Stable Diffusion v2-1 image</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/zero-snr.png"/>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">image with zero SNR and trailing timestep spacing enabled</figcaption>
+  </div>
+</div>

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

@@ -1,84 +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 safetensors
-
-[[open-in-colab]]
-
-[safetensors](https://github.com/huggingface/safetensors) is a safe and fast file format for storing and loading tensors. Typically, PyTorch model weights are saved or *pickled* into a `.bin` file with Python's [`pickle`](https://docs.python.org/3/library/pickle.html) utility. However, `pickle` is not secure and pickled files may contain malicious code that can be executed. safetensors is a secure alternative to `pickle`, making it ideal for sharing model weights.
-
-This guide will show you how you load `.safetensor` files, and how to convert Stable Diffusion model weights stored in other formats to `.safetensor`. Before you start, make sure you have safetensors installed:
-
-```py
-# uncomment to install the necessary libraries in Colab
-#!pip install safetensors
-```
-
-If you look at the [`runwayml/stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5/tree/main) repository, you'll see weights inside the `text_encoder`, `unet` and `vae` subfolders are stored in the `.safetensors` format. By default, 🤗 Diffusers automatically loads these `.safetensors` files from their subfolders if they're available in the model repository.
-
-For more explicit control, you can optionally set `use_safetensors=True` (if `safetensors` is not installed, you'll get an error message asking you to install it):
-
-```py
-from diffusers import DiffusionPipeline
-
-pipeline = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", use_safetensors=True)
-```
-
-However, model weights are not necessarily stored in separate subfolders like in the example above. Sometimes, all the weights are stored in a single `.safetensors` file. In this case, if the weights are Stable Diffusion weights, you can load the file directly with the [`~diffusers.loaders.FromSingleFileMixin.from_single_file`] method:
-
-```py
-from diffusers import StableDiffusionPipeline
-
-pipeline = StableDiffusionPipeline.from_single_file(
-    "https://huggingface.co/WarriorMama777/OrangeMixs/blob/main/Models/AbyssOrangeMix/AbyssOrangeMix.safetensors"
-)
-```
-
-## Convert to safetensors
-
-Not all weights on the Hub are available in the `.safetensors` format, and you may encounter weights stored as `.bin`. In this case, use the [Convert Space](https://huggingface.co/spaces/diffusers/convert) to convert the weights to `.safetensors`. The Convert Space downloads the pickled weights, converts them, and opens a Pull Request to upload the newly converted `.safetensors` file on the Hub. This way, if there is any malicious code contained in the pickled files, they're uploaded to the Hub - which has a [security scanner](https://huggingface.co/docs/hub/security-pickle#hubs-security-scanner) to detect unsafe files and suspicious pickle imports - instead of your computer.
-
-You can use the model with the new `.safetensors` weights by specifying the reference to the Pull Request in the `revision` parameter (you can also test it in this [Check PR](https://huggingface.co/spaces/diffusers/check_pr) Space on the Hub), for example `refs/pr/22`:
-
-```py
-from diffusers import DiffusionPipeline
-
-pipeline = DiffusionPipeline.from_pretrained(
-    "stabilityai/stable-diffusion-2-1", revision="refs/pr/22", use_safetensors=True
-)
-```
-
-## Why use safetensors?
-
-There are several reasons for using safetensors:
-
-- Safety is the number one reason for using safetensors. As open-source and model distribution grows, it is important to be able to trust the model weights you downloaded don't contain any malicious code. The current size of the header in safetensors prevents parsing extremely large JSON files.
-- Loading speed between switching models is another reason to use safetensors, which performs zero-copy of the tensors. It is especially fast compared to `pickle` if you're loading the weights to CPU (the default case), and just as fast if not faster when directly loading the weights to GPU. You'll only notice the performance difference if the model is already loaded, and not if you're downloading the weights or loading the model for the first time.
-
-	The time it takes to load the entire pipeline:
-
-	```py
- 	from diffusers import StableDiffusionPipeline
-
- 	pipeline = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1", use_safetensors=True)
- 	"Loaded in safetensors 0:00:02.033658"
- 	"Loaded in PyTorch 0:00:02.663379"
-	```
-
-	But the actual time it takes to load 500MB of the model weights is only:
-
-	```bash
-	safetensors: 3.4873ms
-	PyTorch: 172.7537ms
-	```
-
-- Lazy loading is also supported in safetensors, which is useful in distributed settings to only load some of the tensors. This format allowed the [BLOOM](https://huggingface.co/bigscience/bloom) model to be loaded in 45 seconds on 8 GPUs instead of 10 minutes with regular PyTorch weights.

docs/source/ja/installation.md

@@ -106,7 +106,7 @@ pip install -e ".[flax]"
 
 これらのコマンドは、リポジトリをクローンしたフォルダと Python のライブラリパスをリンクします。
 Python は通常のライブラリパスに加えて、クローンしたフォルダの中を探すようになります。
-例えば、Python パッケージが通常 `~/anaconda3/envs/main/lib/python3.8/site-packages/` にインストールされている場合、Python はクローンした `~/diffusers/` フォルダも同様に参照します。
+例えば、Python パッケージが通常 `~/anaconda3/envs/main/lib/python3.10/site-packages/` にインストールされている場合、Python はクローンした `~/diffusers/` フォルダも同様に参照します。
 
 <Tip warning={true}>
 

docs/source/ko/installation.md

@@ -105,7 +105,7 @@ pip install -e ".[flax]"
 
 이러한 명령어들은 저장소를 복제한 폴더와 Python 라이브러리 경로를 연결합니다.
 Python은 이제 일반 라이브러리 경로에 더하여 복제한 폴더 내부를 살펴봅니다.
-예를들어 Python 패키지가 `~/anaconda3/envs/main/lib/python3.8/site-packages/`에 설치되어 있는 경우 Python은 복제한 폴더인 `~/diffusers/`도 검색합니다.
+예를들어 Python 패키지가 `~/anaconda3/envs/main/lib/python3.10/site-packages/`에 설치되어 있는 경우 Python은 복제한 폴더인 `~/diffusers/`도 검색합니다.
 
 <Tip warning={true}>
 

docs/source/ko/optimization/fp16.md

@@ -339,7 +339,7 @@ from dataclasses import dataclass
 
 @dataclass
 class UNet2DConditionOutput:
-    sample: torch.FloatTensor
+    sample: torch.Tensor
 
 
 pipe = StableDiffusionPipeline.from_pretrained(

docs/source/ko/training/lora.md

@@ -49,15 +49,15 @@ huggingface-cli login
 
 ### 학습[[dreambooth-training]]
 
-[Pokémon BLIP 캡션](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) 데이터셋으로 [`stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5)를 파인튜닝해 나만의 포켓몬을 생성해 보겠습니다.
+[Naruto BLIP 캡션](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) 데이터셋으로 [`stable-diffusion-v1-5`](https://huggingface.co/runwayml/stable-diffusion-v1-5)를 파인튜닝해 나만의 포켓몬을 생성해 보겠습니다.
 
 시작하려면 `MODEL_NAME` 및 `DATASET_NAME` 환경 변수가 설정되어 있는지 확인하십시오. `OUTPUT_DIR` 및 `HUB_MODEL_ID` 변수는 선택 사항이며 허브에서 모델을 저장할 위치를 지정합니다.
 
 ```bash
 export MODEL_NAME="runwayml/stable-diffusion-v1-5"
-export OUTPUT_DIR="/sddata/finetune/lora/pokemon"
-export HUB_MODEL_ID="pokemon-lora"
-export DATASET_NAME="lambdalabs/pokemon-blip-captions"
+export OUTPUT_DIR="/sddata/finetune/lora/naruto"
+export HUB_MODEL_ID="naruto-lora"
+export DATASET_NAME="lambdalabs/naruto-blip-captions"
 ```
 
 학습을 시작하기 전에 알아야 할 몇 가지 플래그가 있습니다.

docs/source/ko/training/text2image.md

@@ -73,12 +73,12 @@ xFormers는 Flax에 사용할 수 없습니다.
 
 <frameworkcontent>
 <pt>
-다음과 같이 [Pokémon BLIP 캡션](https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions) 데이터셋에서 파인튜닝 실행을 위해 [PyTorch 학습 스크립트](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py)를 실행합니다:
+다음과 같이 [Naruto BLIP 캡션](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) 데이터셋에서 파인튜닝 실행을 위해 [PyTorch 학습 스크립트](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py)를 실행합니다:
 
 
 ```bash
 export MODEL_NAME="CompVis/stable-diffusion-v1-4"
-export dataset_name="lambdalabs/pokemon-blip-captions"
+export dataset_name="lambdalabs/naruto-blip-captions"
 
 accelerate launch train_text_to_image.py \
   --pretrained_model_name_or_path=$MODEL_NAME \
@@ -93,7 +93,7 @@ accelerate launch train_text_to_image.py \
   --learning_rate=1e-05 \
   --max_grad_norm=1 \
   --lr_scheduler="constant" --lr_warmup_steps=0 \
-  --output_dir="sd-pokemon-model" 
+  --output_dir="sd-naruto-model" 
 ```
 
 자체 데이터셋으로 파인튜닝하려면 🤗 [Datasets](https://huggingface.co/docs/datasets/index)에서 요구하는 형식에 따라 데이터셋을 준비하세요. [데이터셋을 허브에 업로드](https://huggingface.co/docs/datasets/image_dataset#upload-dataset-to-the-hub)하거나 [파일들이 있는 로컬 폴더를 준비](https ://huggingface.co/docs/datasets/image_dataset#imagefolder)할 수 있습니다.
@@ -136,7 +136,7 @@ pip install -U -r requirements_flax.txt
 
 ```bash
 export MODEL_NAME="runwayml/stable-diffusion-v1-5"
-export dataset_name="lambdalabs/pokemon-blip-captions"
+export dataset_name="lambdalabs/naruto-blip-captions"
 
 python train_text_to_image_flax.py \
   --pretrained_model_name_or_path=$MODEL_NAME \
@@ -146,7 +146,7 @@ python train_text_to_image_flax.py \
   --max_train_steps=15000 \
   --learning_rate=1e-05 \
   --max_grad_norm=1 \
-  --output_dir="sd-pokemon-model" 
+  --output_dir="sd-naruto-model" 
 ```
 
 자체 데이터셋으로 파인튜닝하려면 🤗 [Datasets](https://huggingface.co/docs/datasets/index)에서 요구하는 형식에 따라 데이터셋을 준비하세요. [데이터셋을 허브에 업로드](https://huggingface.co/docs/datasets/image_dataset#upload-dataset-to-the-hub)하거나 [파일들이 있는 로컬 폴더를 준비](https ://huggingface.co/docs/datasets/image_dataset#imagefolder)할 수 있습니다.
@@ -166,7 +166,7 @@ python train_text_to_image_flax.py \
   --max_train_steps=15000 \
   --learning_rate=1e-05 \
   --max_grad_norm=1 \
-  --output_dir="sd-pokemon-model"
+  --output_dir="sd-naruto-model"
 ```
 </jax>
 </frameworkcontent>
@@ -189,7 +189,7 @@ pipe = StableDiffusionPipeline.from_pretrained(model_path, torch_dtype=torch.flo
 pipe.to("cuda")
 
 image = pipe(prompt="yoda").images[0]
-image.save("yoda-pokemon.png")
+image.save("yoda-naruto.png")
 ```
 </pt>
 <jax>
@@ -203,7 +203,7 @@ from diffusers import FlaxStableDiffusionPipeline
 model_path = "path_to_saved_model"
 pipe, params = FlaxStableDiffusionPipeline.from_pretrained(model_path, dtype=jax.numpy.bfloat16)
 
-prompt = "yoda pokemon"
+prompt = "yoda naruto"
 prng_seed = jax.random.PRNGKey(0)
 num_inference_steps = 50
 
@@ -218,7 +218,7 @@ prompt_ids = shard(prompt_ids)
 
 images = pipeline(prompt_ids, params, prng_seed, num_inference_steps, jit=True).images
 images = pipeline.numpy_to_pil(np.asarray(images.reshape((num_samples,) + images.shape[-3:])))
-image.save("yoda-pokemon.png")
+image.save("yoda-naruto.png")
 ```
 </jax>
 </frameworkcontent>

docs/source/ko/training/unconditional_training.md

@@ -103,13 +103,13 @@ accelerate launch train_unconditional.py \
 <div class="flex justify-center">
     <img src="https://user-images.githubusercontent.com/26864830/180248660-a0b143d0-b89a-42c5-8656-2ebf6ece7e52.png"/>
 </div>
-[Pokemon](https://huggingface.co/datasets/huggan/pokemon) 데이터셋을 사용할 경우:
+[Naruto](https://huggingface.co/datasets/lambdalabs/naruto-blip-captions) 데이터셋을 사용할 경우:
 
 ```bash
 accelerate launch train_unconditional.py \
-  --dataset_name="huggan/pokemon" \
+  --dataset_name="lambdalabs/naruto-blip-captions" \
   --resolution=64 \
-  --output_dir="ddpm-ema-pokemon-64" \
+  --output_dir="ddpm-ema-naruto-64" \
   --train_batch_size=16 \
   --num_epochs=100 \
   --gradient_accumulation_steps=1 \
@@ -129,9 +129,9 @@ accelerate launch train_unconditional.py \
 
 ```bash
 accelerate launch --mixed_precision="fp16" --multi_gpu train_unconditional.py \
-  --dataset_name="huggan/pokemon" \
+  --dataset_name="lambdalabs/naruto-blip-captions" \
   --resolution=64 --center_crop --random_flip \
-  --output_dir="ddpm-ema-pokemon-64" \
+  --output_dir="ddpm-ema-naruto-64" \
   --train_batch_size=16 \
   --num_epochs=100 \
   --gradient_accumulation_steps=1 \

docs/source/pt/installation.md

@@ -102,7 +102,7 @@ pip install -e ".[flax]"
 
 Esses comandos irá linkar a pasta que você clonou o repositório e os caminhos das suas bibliotecas Python.
 Python então irá procurar dentro da pasta que você clonou além dos caminhos normais das bibliotecas.
-Por exemplo, se o pacote python for tipicamente instalado no `~/anaconda3/envs/main/lib/python3.8/site-packages/`, o Python também irá procurar na pasta `~/diffusers/` que você clonou.
+Por exemplo, se o pacote python for tipicamente instalado no `~/anaconda3/envs/main/lib/python3.10/site-packages/`, o Python também irá procurar na pasta `~/diffusers/` que você clonou.
 
 <Tip warning={true}>
 

docs/source/zh/installation.md

@@ -107,7 +107,7 @@ pip install -e ".[flax]"
 
 这些命令将连接到你克隆的版本库和你的 Python 库路径。
 现在，不只是在通常的库路径，Python 还会在你克隆的文件夹内寻找包。
-例如，如果你的 Python 包通常安装在 `~/anaconda3/envs/main/lib/python3.8/Site-packages/`，Python 也会搜索你克隆到的文件夹。`~/diffusers/`。
+例如，如果你的 Python 包通常安装在 `~/anaconda3/envs/main/lib/python3.10/Site-packages/`，Python 也会搜索你克隆到的文件夹。`~/diffusers/`。
 
 <Tip warning={true}>
 

examples/advanced_diffusion_training/train_dreambooth_lora_sd15_advanced.py

@@ -71,7 +71,7 @@ from diffusers.utils.import_utils import is_xformers_available
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.28.0.dev0")
+check_min_version("0.29.0.dev0")
 
 logger = get_logger(__name__)
 
@@ -758,7 +758,7 @@ class TokenEmbeddingsHandler:
 
             idx += 1
 
-    # copied from train_dreambooth_lora_sdxl_advanced.py
+    # Copied from train_dreambooth_lora_sdxl_advanced.py
     def save_embeddings(self, file_path: str):
         assert self.train_ids is not None, "Initialize new tokens before saving embeddings."
         tensors = {}
@@ -981,7 +981,7 @@ def collate_fn(examples, with_prior_preservation=False):
 
 
 class PromptDataset(Dataset):
-    "A simple dataset to prepare the prompts to generate class images on multiple GPUs."
+    """A simple dataset to prepare the prompts to generate class images on multiple GPUs."""
 
     def __init__(self, prompt, num_samples):
         self.prompt = prompt

examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py

@@ -78,7 +78,7 @@ from diffusers.utils.torch_utils import is_compiled_module
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.28.0.dev0")
+check_min_version("0.29.0.dev0")
 
 logger = get_logger(__name__)
 
@@ -1136,7 +1136,7 @@ def collate_fn(examples, with_prior_preservation=False):
 
 
 class PromptDataset(Dataset):
-    "A simple dataset to prepare the prompts to generate class images on multiple GPUs."
+    """A simple dataset to prepare the prompts to generate class images on multiple GPUs."""
 
     def __init__(self, prompt, num_samples):
         self.prompt = prompt

examples/community/README.md

@@ -68,6 +68,8 @@ Please also check out our [Community Scripts](https://github.com/huggingface/dif
 |   InstantID Pipeline                                                                                               | Stable Diffusion XL Pipeline that supports InstantID                                                                                                                                                                                                                                                                                                                                                 |  [InstantID Pipeline](#instantid-pipeline) | [![Hugging Face Space](https://img.shields.io/badge/🤗%20Hugging%20Face-Space-yellow)](https://huggingface.co/spaces/InstantX/InstantID) | [Haofan Wang](https://github.com/haofanwang) |
 |   UFOGen Scheduler                                                                                               | Scheduler for UFOGen Model (compatible with Stable Diffusion pipelines)                                                                                                                                                                                                                                                                                                                                                 |  [UFOGen Scheduler](#ufogen-scheduler) | - | [dg845](https://github.com/dg845) |
 | Stable Diffusion XL IPEX Pipeline | Accelerate Stable Diffusion XL inference pipeline with BF16/FP32 precision on Intel Xeon CPUs with [IPEX](https://github.com/intel/intel-extension-for-pytorch) | [Stable Diffusion XL on IPEX](#stable-diffusion-xl-on-ipex) | - | [Dan Li](https://github.com/ustcuna/) |
+| Stable Diffusion BoxDiff Pipeline | Training-free controlled generation with bounding boxes using [BoxDiff](https://github.com/showlab/BoxDiff) | [Stable Diffusion BoxDiff Pipeline](#stable-diffusion-boxdiff) | - | [Jingyang Zhang](https://github.com/zjysteven/) |
+|   FRESCO V2V Pipeline                                                                                                    | Implementation of [[CVPR 2024] FRESCO: Spatial-Temporal Correspondence for Zero-Shot Video Translation](https://arxiv.org/abs/2403.12962)                                                                                                                                                                                                                                                                                                                                                                                                                                      | [FRESCO V2V Pipeline](#fresco)      | - |              [Yifan Zhou](https://github.com/SingleZombie) |
 
 To load a custom pipeline you just need to pass the `custom_pipeline` argument to `DiffusionPipeline`, as one of the files in `diffusers/examples/community`. Feel free to send a PR with your own pipelines, we will merge them quickly.
 
@@ -238,12 +240,12 @@ pipeline_output = pipe(
     # denoising_steps=10,     # (optional) Number of denoising steps of each inference pass. Default: 10.
     # ensemble_size=10,       # (optional) Number of inference passes in the ensemble. Default: 10.
     # ------------------------------------------------
-    
+
     # ----- recommended setting for LCM version ------
     # denoising_steps=4,
     # ensemble_size=5,
     # -------------------------------------------------
-    
+
     # processing_res=768,     # (optional) Maximum resolution of processing. If set to 0: will not resize at all. Defaults to 768.
     # match_input_res=True,   # (optional) Resize depth prediction to match input resolution.
     # batch_size=0,           # (optional) Inference batch size, no bigger than `num_ensemble`. If set to 0, the script will automatically decide the proper batch size. Defaults to 0.
@@ -1030,7 +1032,7 @@ image = pipe().images[0]
 
 Make sure you have @crowsonkb's <https://github.com/crowsonkb/k-diffusion> installed:
 
-```
+```sh
 pip install k-diffusion
 ```
 
@@ -1676,6 +1678,68 @@ image = pipe(prompt, image=input_image, strength=0.75,).images[0]
 image.save('tensorrt_img2img_new_zealand_hills.png')
 ```
 
+### Stable Diffusion BoxDiff
+BoxDiff is a training-free method for controlled generation with bounding box coordinates. It shoud work with any Stable Diffusion model. Below shows an example with `stable-diffusion-2-1-base`.
+```py
+import torch
+from PIL import Image, ImageDraw
+from copy import deepcopy
+
+from examples.community.pipeline_stable_diffusion_boxdiff import StableDiffusionBoxDiffPipeline
+
+def draw_box_with_text(img, boxes, names):
+    colors = ["red", "olive", "blue", "green", "orange", "brown", "cyan", "purple"]
+    img_new = deepcopy(img)
+    draw = ImageDraw.Draw(img_new)
+
+    W, H = img.size
+    for bid, box in enumerate(boxes):
+        draw.rectangle([box[0] * W, box[1] * H, box[2] * W, box[3] * H], outline=colors[bid % len(colors)], width=4)
+        draw.text((box[0] * W, box[1] * H), names[bid], fill=colors[bid % len(colors)])
+    return img_new
+
+pipe = StableDiffusionBoxDiffPipeline.from_pretrained(
+    "stabilityai/stable-diffusion-2-1-base",
+    torch_dtype=torch.float16,
+)
+pipe.to("cuda")
+
+# example 1
+prompt = "as the aurora lights up the sky, a herd of reindeer leisurely wanders on the grassy meadow, admiring the breathtaking view, a serene lake quietly reflects the magnificent display, and in the distance, a snow-capped mountain stands majestically, fantasy, 8k, highly detailed"
+phrases = [
+    "aurora",
+    "reindeer",
+    "meadow",
+    "lake",
+    "mountain"
+]
+boxes = [[1,3,512,202], [75,344,421,495], [1,327,508,507], [2,217,507,341], [1,135,509,242]]
+
+# example 2
+# prompt = "A rabbit wearing sunglasses looks very proud"
+# phrases = ["rabbit", "sunglasses"]
+# boxes = [[67,87,366,512], [66,130,364,262]]
+
+boxes = [[x / 512 for x in box] for box in boxes]
+
+images = pipe(
+    prompt,
+    boxdiff_phrases=phrases,
+    boxdiff_boxes=boxes,
+    boxdiff_kwargs={
+        "attention_res": 16,
+        "normalize_eot": True
+    },
+    num_inference_steps=50,
+    guidance_scale=7.5,
+    generator=torch.manual_seed(42),
+    safety_checker=None
+).images
+
+draw_box_with_text(images[0], boxes, phrases).save("output.png")
+```
+
+
 ### Stable Diffusion Reference
 
 This pipeline uses the Reference Control. Refer to the [sd-webui-controlnet discussion: Reference-only Control](https://github.com/Mikubill/sd-webui-controlnet/discussions/1236)[sd-webui-controlnet discussion: Reference-adain Control](https://github.com/Mikubill/sd-webui-controlnet/discussions/1280).
@@ -1790,13 +1854,13 @@ To use this pipeline, you need to:
 
 You can simply use pip to install IPEX with the latest version.
 
-```python
+```sh
 python -m pip install intel_extension_for_pytorch
 ```
 
 **Note:** To install a specific version, run with the following command:
 
-```
+```sh
 python -m pip install intel_extension_for_pytorch==<version_name> -f https://developer.intel.com/ipex-whl-stable-cpu
 ```
 
@@ -1894,13 +1958,13 @@ To use this pipeline, you need to:
 
 You can simply use pip to install IPEX with the latest version.
 
-```python
+```sh
 python -m pip install intel_extension_for_pytorch
 ```
 
 **Note:** To install a specific version, run with the following command:
 
-```
+```sh
 python -m pip install intel_extension_for_pytorch==<version_name> -f https://developer.intel.com/ipex-whl-stable-cpu
 ```
 
@@ -2946,8 +3010,8 @@ This code implements a pipeline for the Stable Diffusion model, enabling the div
 
 ### Sample Code
 
-```
-from from examples.community.regional_prompting_stable_diffusion import RegionalPromptingStableDiffusionPipeline
+```py
+from examples.community.regional_prompting_stable_diffusion import RegionalPromptingStableDiffusionPipeline
 pipe = RegionalPromptingStableDiffusionPipeline.from_single_file(model_path, vae=vae)
 
 rp_args = {
@@ -3972,6 +4036,93 @@ onestep_image = pipe(prompt, num_inference_steps=1).images[0]
 multistep_image = pipe(prompt, num_inference_steps=4).images[0]
 ```
 
+### FRESCO
+
+This is the Diffusers implementation of zero-shot video-to-video translation pipeline [FRESCO](https://github.com/williamyang1991/FRESCO) (without Ebsynth postprocessing and background smooth). To run the code, please install gmflow. Then modify the path in `gmflow_dir`. After that, you can run the pipeline with:
+
+```py
+from PIL import Image
+import cv2
+import torch
+import numpy as np
+
+from diffusers import ControlNetModel,DDIMScheduler, DiffusionPipeline
+import sys
+gmflow_dir = "/path/to/gmflow"
+sys.path.insert(0, gmflow_dir)
+
+def video_to_frame(video_path: str, interval: int):
+    vidcap = cv2.VideoCapture(video_path)
+    success = True
+
+    count = 0
+    res = []
+    while success:
+        count += 1
+        success, image = vidcap.read()
+        if count % interval != 1:
+            continue
+        if image is not None:
+            image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+            res.append(image)
+            if len(res) >= 8:
+                break
+
+    vidcap.release()
+    return res
+
+
+input_video_path = 'https://github.com/williamyang1991/FRESCO/raw/main/data/car-turn.mp4'
+output_video_path = 'car.gif'
+
+# You can use any fintuned SD here
+model_path = 'SG161222/Realistic_Vision_V2.0'
+
+prompt = 'a red car turns in the winter'
+a_prompt = ', RAW photo, subject, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3, '
+n_prompt = '(deformed iris, deformed pupils, semi-realistic, cgi, 3d, render, sketch, cartoon, drawing, anime, mutated hands and fingers:1.4), (deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, wrong anatomy, extra limb, missing limb, floating limbs, disconnected limbs, mutation, mutated, ugly, disgusting, amputation'
+
+input_interval = 5
+frames = video_to_frame(
+    input_video_path, input_interval)
+
+control_frames = []
+# get canny image
+for frame in frames:
+    image = cv2.Canny(frame, 50, 100)
+    np_image = np.array(image)
+    np_image = np_image[:, :, None]
+    np_image = np.concatenate([np_image, np_image, np_image], axis=2)
+    canny_image = Image.fromarray(np_image)
+    control_frames.append(canny_image)
+
+# You can use any ControlNet here
+controlnet = ControlNetModel.from_pretrained(
+    "lllyasviel/sd-controlnet-canny").to('cuda')
+
+pipe = DiffusionPipeline.from_pretrained(
+    model_path, controlnet=controlnet, custom_pipeline='fresco_v2v').to('cuda')
+pipe.scheduler = DDIMScheduler.from_config(pipe.scheduler.config)
+
+generator = torch.manual_seed(0)
+frames = [Image.fromarray(frame) for frame in frames]
+
+output_frames = pipe(
+    prompt + a_prompt,
+    frames,
+    control_frames,
+    num_inference_steps=20,
+    strength=0.75,
+    controlnet_conditioning_scale=0.7,
+    generator=generator,
+    negative_prompt=n_prompt
+).images
+
+output_frames[0].save(output_video_path, save_all=True,
+                 append_images=output_frames[1:], duration=100, loop=0)
+
+```
+
 # Perturbed-Attention Guidance
 
 [Project](https://ku-cvlab.github.io/Perturbed-Attention-Guidance/) / [arXiv](https://arxiv.org/abs/2403.17377) / [GitHub](https://github.com/KU-CVLAB/Perturbed-Attention-Guidance)
@@ -3980,7 +4131,7 @@ This implementation is based on [Diffusers](https://huggingface.co/docs/diffuser
 
 ## Example Usage
 
-```
+```py
 import os
 import torch
 

examples/community/bit_diffusion.py

@@ -44,9 +44,9 @@ def bits_to_decimal(x, bits=BITS):
 # modified scheduler step functions for clamping the predicted x_0 between -bit_scale and +bit_scale
 def ddim_bit_scheduler_step(
     self,
-    model_output: torch.FloatTensor,
+    model_output: torch.Tensor,
     timestep: int,
-    sample: torch.FloatTensor,
+    sample: torch.Tensor,
     eta: float = 0.0,
     use_clipped_model_output: bool = True,
     generator=None,
@@ -56,9 +56,9 @@ def ddim_bit_scheduler_step(
     Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
     process from the learned model outputs (most often the predicted noise).
     Args:
-        model_output (`torch.FloatTensor`): direct output from learned diffusion model.
+        model_output (`torch.Tensor`): direct output from learned diffusion model.
         timestep (`int`): current discrete timestep in the diffusion chain.
-        sample (`torch.FloatTensor`):
+        sample (`torch.Tensor`):
             current instance of sample being created by diffusion process.
         eta (`float`): weight of noise for added noise in diffusion step.
         use_clipped_model_output (`bool`): TODO
@@ -134,9 +134,9 @@ def ddim_bit_scheduler_step(
 
 def ddpm_bit_scheduler_step(
     self,
-    model_output: torch.FloatTensor,
+    model_output: torch.Tensor,
     timestep: int,
-    sample: torch.FloatTensor,
+    sample: torch.Tensor,
     prediction_type="epsilon",
     generator=None,
     return_dict: bool = True,
@@ -145,9 +145,9 @@ def ddpm_bit_scheduler_step(
     Predict the sample at the previous timestep by reversing the SDE. Core function to propagate the diffusion
     process from the learned model outputs (most often the predicted noise).
     Args:
-        model_output (`torch.FloatTensor`): direct output from learned diffusion model.
+        model_output (`torch.Tensor`): direct output from learned diffusion model.
         timestep (`int`): current discrete timestep in the diffusion chain.
-        sample (`torch.FloatTensor`):
+        sample (`torch.Tensor`):
             current instance of sample being created by diffusion process.
         prediction_type (`str`, default `epsilon`):
             indicates whether the model predicts the noise (epsilon), or the samples (`sample`).

examples/community/checkpoint_merger.py

@@ -138,7 +138,6 @@ class CheckpointMergerPipeline(DiffusionPipeline):
             comparison_result &= self._compare_model_configs(config_dicts[idx - 1], config_dicts[idx])
             if not force and comparison_result is False:
                 raise ValueError("Incompatible checkpoints. Please check model_index.json for the models.")
-                print(config_dicts[0], config_dicts[1])
         print("Compatible model_index.json files found")
         # Step 2: Basic Validation has succeeded. Let's download the models and save them into our local files.
         cached_folders = []

examples/community/clip_guided_images_mixing_stable_diffusion.py

@@ -233,8 +233,8 @@ class CLIPGuidedImagesMixingStableDiffusion(DiffusionPipeline, StableDiffusionMi
     @torch.no_grad()
     def __call__(
         self,
-        style_image: Union[torch.FloatTensor, PIL.Image.Image],
-        content_image: Union[torch.FloatTensor, PIL.Image.Image],
+        style_image: Union[torch.Tensor, PIL.Image.Image],
+        content_image: Union[torch.Tensor, PIL.Image.Image],
         style_prompt: Optional[str] = None,
         content_prompt: Optional[str] = None,
         height: Optional[int] = 512,

examples/community/clip_guided_stable_diffusion.py

@@ -180,7 +180,7 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline, StableDiffusionMixin):
         num_cutouts: Optional[int] = 4,
         use_cutouts: Optional[bool] = True,
         generator: Optional[torch.Generator] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
     ):

examples/community/clip_guided_stable_diffusion_img2img.py

@@ -306,7 +306,7 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline, StableDiffusionMixin):
         prompt: Union[str, List[str]],
         height: Optional[int] = 512,
         width: Optional[int] = 512,
-        image: Union[torch.FloatTensor, PIL.Image.Image] = None,
+        image: Union[torch.Tensor, PIL.Image.Image] = None,
         strength: float = 0.8,
         num_inference_steps: Optional[int] = 50,
         guidance_scale: Optional[float] = 7.5,
@@ -317,7 +317,7 @@ class CLIPGuidedStableDiffusion(DiffusionPipeline, StableDiffusionMixin):
         num_cutouts: Optional[int] = 4,
         use_cutouts: Optional[bool] = True,
         generator: Optional[torch.Generator] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
     ):

examples/community/composable_stable_diffusion.py

@@ -354,10 +354,10 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin)
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[torch.Generator] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
         weights: Optional[str] = "",
     ):
@@ -391,7 +391,7 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin)
             generator (`torch.Generator`, *optional*):
                 A [torch generator](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation
                 deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
@@ -403,7 +403,7 @@ class ComposableStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin)
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function will be called. If not specified, the callback will be
                 called at every step.

examples/community/ddim_noise_comparative_analysis.py

@@ -103,7 +103,7 @@ class DDIMNoiseComparativeAnalysisPipeline(DiffusionPipeline):
     @torch.no_grad()
     def __call__(
         self,
-        image: Union[torch.FloatTensor, PIL.Image.Image] = None,
+        image: Union[torch.Tensor, PIL.Image.Image] = None,
         strength: float = 0.8,
         batch_size: int = 1,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
@@ -115,7 +115,7 @@ class DDIMNoiseComparativeAnalysisPipeline(DiffusionPipeline):
     ) -> Union[ImagePipelineOutput, Tuple]:
         r"""
         Args:
-            image (`torch.FloatTensor` or `PIL.Image.Image`):
+            image (`torch.Tensor` or `PIL.Image.Image`):
                 `Image`, or tensor representing an image batch, that will be used as the starting point for the
                 process.
             strength (`float`, *optional*, defaults to 0.8):

examples/community/gluegen.py

@@ -205,7 +205,7 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
         safety_checker: StableDiffusionSafetyChecker,
         feature_extractor: CLIPImageProcessor,
         language_adapter: TranslatorNoLN = None,
-        tensor_norm: torch.FloatTensor = None,
+        tensor_norm: torch.Tensor = None,
         requires_safety_checker: bool = True,
     ):
         super().__init__()
@@ -231,7 +231,7 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
         num_token: int,
         dim: int,
         dim_out: int,
-        tensor_norm: torch.FloatTensor,
+        tensor_norm: torch.Tensor,
         mult: int = 2,
         depth: int = 5,
     ):
@@ -242,7 +242,7 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
         )
         self.language_adapter.load_state_dict(torch.load(model_path))
 
-    def _adapt_language(self, prompt_embeds: torch.FloatTensor):
+    def _adapt_language(self, prompt_embeds: torch.Tensor):
         prompt_embeds = prompt_embeds / 3
         prompt_embeds = self.language_adapter(prompt_embeds) * (self.tensor_norm / 2)
         return prompt_embeds
@@ -254,8 +254,8 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
         clip_skip: Optional[int] = None,
     ):
@@ -275,10 +275,10 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -535,7 +535,7 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
                 data type of the generated embeddings
 
         Returns:
-            `torch.FloatTensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
+            `torch.Tensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
         """
         assert len(w.shape) == 1
         w = w * 1000.0
@@ -594,9 +594,9 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -635,14 +635,14 @@ class GlueGenStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin, Lo
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor is generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
             ip_adapter_image: (`PipelineImageInput`, *optional*): Optional image input to work with IP Adapters.

examples/community/hd_painter.py

@@ -28,10 +28,10 @@ class RASGAttnProcessor:
     def __call__(
         self,
         attn: Attention,
-        hidden_states: torch.FloatTensor,
-        encoder_hidden_states: Optional[torch.FloatTensor] = None,
-        attention_mask: Optional[torch.FloatTensor] = None,
-        temb: Optional[torch.FloatTensor] = None,
+        hidden_states: torch.Tensor,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        temb: Optional[torch.Tensor] = None,
         scale: float = 1.0,
     ) -> torch.Tensor:
         # Same as the default AttnProcessor up untill the part where similarity matrix gets saved
@@ -111,10 +111,10 @@ class PAIntAAttnProcessor:
     def __call__(
         self,
         attn: Attention,
-        hidden_states: torch.FloatTensor,
-        encoder_hidden_states: Optional[torch.FloatTensor] = None,
-        attention_mask: Optional[torch.FloatTensor] = None,
-        temb: Optional[torch.FloatTensor] = None,
+        hidden_states: torch.Tensor,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        temb: Optional[torch.Tensor] = None,
         scale: float = 1.0,
     ) -> torch.Tensor:
         # Automatically recognize the resolution of the current attention layer and resize the masks accordingly
@@ -454,7 +454,7 @@ class StableDiffusionHDPainterPipeline(StableDiffusionInpaintPipeline):
         prompt: Union[str, List[str]] = None,
         image: PipelineImageInput = None,
         mask_image: PipelineImageInput = None,
-        masked_image_latents: torch.FloatTensor = None,
+        masked_image_latents: torch.Tensor = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
         padding_mask_crop: Optional[int] = None,
@@ -467,9 +467,9 @@ class StableDiffusionHDPainterPipeline(StableDiffusionInpaintPipeline):
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.01,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,

examples/community/iadb.py

@@ -17,21 +17,21 @@ class IADBScheduler(SchedulerMixin, ConfigMixin):
 
     def step(
         self,
-        model_output: torch.FloatTensor,
+        model_output: torch.Tensor,
         timestep: int,
-        x_alpha: torch.FloatTensor,
-    ) -> torch.FloatTensor:
+        x_alpha: torch.Tensor,
+    ) -> torch.Tensor:
         """
         Predict the sample at the previous timestep by reversing the ODE. Core function to propagate the diffusion
         process from the learned model outputs (most often the predicted noise).
 
         Args:
-            model_output (`torch.FloatTensor`): direct output from learned diffusion model. It is the direction from x0 to x1.
+            model_output (`torch.Tensor`): direct output from learned diffusion model. It is the direction from x0 to x1.
             timestep (`float`): current timestep in the diffusion chain.
-            x_alpha (`torch.FloatTensor`): x_alpha sample for the current timestep
+            x_alpha (`torch.Tensor`): x_alpha sample for the current timestep
 
         Returns:
-            `torch.FloatTensor`: the sample at the previous timestep
+            `torch.Tensor`: the sample at the previous timestep
 
         """
         if self.num_inference_steps is None:
@@ -53,10 +53,10 @@ class IADBScheduler(SchedulerMixin, ConfigMixin):
 
     def add_noise(
         self,
-        original_samples: torch.FloatTensor,
-        noise: torch.FloatTensor,
-        alpha: torch.FloatTensor,
-    ) -> torch.FloatTensor:
+        original_samples: torch.Tensor,
+        noise: torch.Tensor,
+        alpha: torch.Tensor,
+    ) -> torch.Tensor:
         return original_samples * alpha + noise * (1 - alpha)
 
     def __len__(self):

examples/community/imagic_stable_diffusion.py

@@ -110,7 +110,7 @@ class ImagicStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin):
     def train(
         self,
         prompt: Union[str, List[str]],
-        image: Union[torch.FloatTensor, PIL.Image.Image],
+        image: Union[torch.Tensor, PIL.Image.Image],
         height: Optional[int] = 512,
         width: Optional[int] = 512,
         generator: Optional[torch.Generator] = None,
@@ -144,7 +144,7 @@ class ImagicStableDiffusionPipeline(DiffusionPipeline, StableDiffusionMixin):
             generator (`torch.Generator`, *optional*):
                 A [torch generator](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation
                 deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.

examples/community/img2img_inpainting.py

@@ -133,9 +133,9 @@ class ImageToImageInpaintingPipeline(DiffusionPipeline):
     def __call__(
         self,
         prompt: Union[str, List[str]],
-        image: Union[torch.FloatTensor, PIL.Image.Image],
-        inner_image: Union[torch.FloatTensor, PIL.Image.Image],
-        mask_image: Union[torch.FloatTensor, PIL.Image.Image],
+        image: Union[torch.Tensor, PIL.Image.Image],
+        inner_image: Union[torch.Tensor, PIL.Image.Image],
+        mask_image: Union[torch.Tensor, PIL.Image.Image],
         height: int = 512,
         width: int = 512,
         num_inference_steps: int = 50,
@@ -144,10 +144,10 @@ class ImageToImageInpaintingPipeline(DiffusionPipeline):
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[torch.Generator] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
         **kwargs,
     ):
@@ -194,7 +194,7 @@ class ImageToImageInpaintingPipeline(DiffusionPipeline):
             generator (`torch.Generator`, *optional*):
                 A [torch generator](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation
                 deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
@@ -206,7 +206,7 @@ class ImageToImageInpaintingPipeline(DiffusionPipeline):
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function will be called. If not specified, the callback will be
                 called at every step.

examples/community/instaflow_one_step.py

@@ -189,8 +189,8 @@ class InstaFlowPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
     ):
         deprecation_message = "`_encode_prompt()` is deprecated and it will be removed in a future version. Use `encode_prompt()` instead. Also, be aware that the output format changed from a concatenated tensor to a tuple."
@@ -219,8 +219,8 @@ class InstaFlowPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
     ):
         r"""
@@ -239,10 +239,10 @@ class InstaFlowPipeline(
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -501,12 +501,12 @@ class InstaFlowPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         guidance_rescale: float = 0.0,
@@ -538,14 +538,14 @@ class InstaFlowPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor is generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
             output_type (`str`, *optional*, defaults to `"pil"`):
@@ -555,7 +555,7 @@ class InstaFlowPipeline(
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that calls every `callback_steps` steps during inference. The function is called with the
-                following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function is called. If not specified, the callback is called at
                 every step.

examples/community/interpolate_stable_diffusion.py

@@ -132,12 +132,12 @@ class StableDiffusionWalkPipeline(DiffusionPipeline, StableDiffusionMixin):
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[torch.Generator] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
-        text_embeddings: Optional[torch.FloatTensor] = None,
+        text_embeddings: Optional[torch.Tensor] = None,
         **kwargs,
     ):
         r"""
@@ -170,7 +170,7 @@ class StableDiffusionWalkPipeline(DiffusionPipeline, StableDiffusionMixin):
             generator (`torch.Generator`, *optional*):
                 A [torch generator](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation
                 deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
@@ -182,11 +182,11 @@ class StableDiffusionWalkPipeline(DiffusionPipeline, StableDiffusionMixin):
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function will be called. If not specified, the callback will be
                 called at every step.
-            text_embeddings (`torch.FloatTensor`, *optional*, defaults to `None`):
+            text_embeddings (`torch.Tensor`, *optional*, defaults to `None`):
                 Pre-generated text embeddings to be used as inputs for image generation. Can be used in place of
                 `prompt` to avoid re-computing the embeddings. If not provided, the embeddings will be generated from
                 the supplied `prompt`.

examples/community/ip_adapter_face_id.py

@@ -62,7 +62,7 @@ class IPAdapterFullImageProjection(nn.Module):
         self.ff = FeedForward(image_embed_dim, cross_attention_dim * num_tokens, mult=mult, activation_fn="gelu")
         self.norm = nn.LayerNorm(cross_attention_dim)
 
-    def forward(self, image_embeds: torch.FloatTensor):
+    def forward(self, image_embeds: torch.Tensor):
         x = self.ff(image_embeds)
         x = x.reshape(-1, self.num_tokens, self.cross_attention_dim)
         return self.norm(x)
@@ -452,8 +452,8 @@ class IPAdapterFaceIDStableDiffusionPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
         **kwargs,
     ):
@@ -484,8 +484,8 @@ class IPAdapterFaceIDStableDiffusionPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
         clip_skip: Optional[int] = None,
     ):
@@ -505,10 +505,10 @@ class IPAdapterFaceIDStableDiffusionPipeline(
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -788,7 +788,7 @@ class IPAdapterFaceIDStableDiffusionPipeline(
                 data type of the generated embeddings
 
         Returns:
-            `torch.FloatTensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
+            `torch.Tensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
         """
         assert len(w.shape) == 1
         w = w * 1000.0
@@ -847,10 +847,10 @@ class IPAdapterFaceIDStableDiffusionPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        image_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        image_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -891,17 +891,17 @@ class IPAdapterFaceIDStableDiffusionPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor is generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
-            image_embeds (`torch.FloatTensor`, *optional*):
+            image_embeds (`torch.Tensor`, *optional*):
                 Pre-generated image embeddings.
             output_type (`str`, *optional*, defaults to `"pil"`):
                 The output format of the generated image. Choose between `PIL.Image` or `np.array`.

examples/community/kohya_hires_fix.py

@@ -0,0 +1,468 @@
+# 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.
+
+from typing import Any, Dict, List, Optional, Tuple, Union
+
+import torch
+import torch.nn as nn
+import torch.utils.checkpoint
+from transformers import CLIPImageProcessor, CLIPTextModel, CLIPTokenizer, CLIPVisionModelWithProjection
+
+from diffusers.configuration_utils import register_to_config
+from diffusers.image_processor import VaeImageProcessor
+from diffusers.models.autoencoders import AutoencoderKL
+from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel, UNet2DConditionOutput
+from diffusers.pipelines.stable_diffusion import StableDiffusionPipeline
+from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker
+from diffusers.schedulers import KarrasDiffusionSchedulers
+from diffusers.utils import USE_PEFT_BACKEND, deprecate, logging, scale_lora_layers, unscale_lora_layers
+
+
+logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
+
+
+class UNet2DConditionModelHighResFix(UNet2DConditionModel):
+    r"""
+    A conditional 2D UNet model that applies Kohya fix proposed for high resolution image generation.
+
+    This model inherits from [`UNet2DConditionModel`]. Check the superclass documentation for learning about all the parameters.
+
+    Parameters:
+        high_res_fix (`List[Dict]`, *optional*, defaults to `[{'timestep': 600, 'scale_factor': 0.5, 'block_num': 1}]`):
+            Enables Kohya fix for high resolution generation. The activation maps are scaled based on the scale_factor up to the timestep at specified block_num.
+    """
+
+    _supports_gradient_checkpointing = True
+
+    @register_to_config
+    def __init__(self, high_res_fix: List[Dict] = [{"timestep": 600, "scale_factor": 0.5, "block_num": 1}], **kwargs):
+        super().__init__(**kwargs)
+        if high_res_fix:
+            self.config.high_res_fix = sorted(high_res_fix, key=lambda x: x["timestep"], reverse=True)
+
+    @classmethod
+    def _resize(cls, sample, target=None, scale_factor=1, mode="bicubic"):
+        dtype = sample.dtype
+        if dtype == torch.bfloat16:
+            sample = sample.to(torch.float32)
+
+        if target is not None:
+            if sample.shape[-2:] != target.shape[-2:]:
+                sample = nn.functional.interpolate(sample, size=target.shape[-2:], mode=mode, align_corners=False)
+        elif scale_factor != 1:
+            sample = nn.functional.interpolate(sample, scale_factor=scale_factor, mode=mode, align_corners=False)
+
+        return sample.to(dtype)
+
+    def forward(
+        self,
+        sample: torch.FloatTensor,
+        timestep: Union[torch.Tensor, float, int],
+        encoder_hidden_states: torch.Tensor,
+        class_labels: Optional[torch.Tensor] = None,
+        timestep_cond: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        cross_attention_kwargs: Optional[Dict[str, Any]] = None,
+        added_cond_kwargs: Optional[Dict[str, torch.Tensor]] = None,
+        down_block_additional_residuals: Optional[Tuple[torch.Tensor]] = None,
+        mid_block_additional_residual: Optional[torch.Tensor] = None,
+        down_intrablock_additional_residuals: Optional[Tuple[torch.Tensor]] = None,
+        encoder_attention_mask: Optional[torch.Tensor] = None,
+        return_dict: bool = True,
+    ) -> Union[UNet2DConditionOutput, Tuple]:
+        r"""
+        The [`UNet2DConditionModel`] forward method.
+
+        Args:
+            sample (`torch.FloatTensor`):
+                The noisy input tensor with the following shape `(batch, channel, height, width)`.
+            timestep (`torch.FloatTensor` or `float` or `int`): The number of timesteps to denoise an input.
+            encoder_hidden_states (`torch.FloatTensor`):
+                The encoder hidden states with shape `(batch, sequence_length, feature_dim)`.
+            class_labels (`torch.Tensor`, *optional*, defaults to `None`):
+                Optional class labels for conditioning. Their embeddings will be summed with the timestep embeddings.
+            timestep_cond: (`torch.Tensor`, *optional*, defaults to `None`):
+                Conditional embeddings for timestep. If provided, the embeddings will be summed with the samples passed
+                through the `self.time_embedding` layer to obtain the timestep embeddings.
+            attention_mask (`torch.Tensor`, *optional*, defaults to `None`):
+                An attention mask of shape `(batch, key_tokens)` is applied to `encoder_hidden_states`. If `1` the mask
+                is kept, otherwise if `0` it is discarded. Mask will be converted into a bias, which adds large
+                negative values to the attention scores corresponding to "discard" tokens.
+            cross_attention_kwargs (`dict`, *optional*):
+                A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
+                `self.processor` in
+                [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
+            added_cond_kwargs: (`dict`, *optional*):
+                A kwargs dictionary containing additional embeddings that if specified are added to the embeddings that
+                are passed along to the UNet blocks.
+            down_block_additional_residuals: (`tuple` of `torch.Tensor`, *optional*):
+                A tuple of tensors that if specified are added to the residuals of down unet blocks.
+            mid_block_additional_residual: (`torch.Tensor`, *optional*):
+                A tensor that if specified is added to the residual of the middle unet block.
+            down_intrablock_additional_residuals (`tuple` of `torch.Tensor`, *optional*):
+                additional residuals to be added within UNet down blocks, for example from T2I-Adapter side model(s)
+            encoder_attention_mask (`torch.Tensor`):
+                A cross-attention mask of shape `(batch, sequence_length)` is applied to `encoder_hidden_states`. If
+                `True` the mask is kept, otherwise if `False` it is discarded. Mask will be converted into a bias,
+                which adds large negative values to the attention scores corresponding to "discard" tokens.
+            return_dict (`bool`, *optional*, defaults to `True`):
+                Whether or not to return a [`~models.unets.unet_2d_condition.UNet2DConditionOutput`] instead of a plain
+                tuple.
+
+        Returns:
+            [`~models.unets.unet_2d_condition.UNet2DConditionOutput`] or `tuple`:
+                If `return_dict` is True, an [`~models.unets.unet_2d_condition.UNet2DConditionOutput`] is returned,
+                otherwise a `tuple` is returned where the first element is the sample tensor.
+        """
+        # By default samples have to be AT least a multiple of the overall upsampling factor.
+        # The overall upsampling factor is equal to 2 ** (# num of upsampling layers).
+        # However, the upsampling interpolation output size can be forced to fit any upsampling size
+        # on the fly if necessary.
+        default_overall_up_factor = 2**self.num_upsamplers
+
+        # upsample size should be forwarded when sample is not a multiple of `default_overall_up_factor`
+        forward_upsample_size = False
+        upsample_size = None
+
+        for dim in sample.shape[-2:]:
+            if dim % default_overall_up_factor != 0:
+                # Forward upsample size to force interpolation output size.
+                forward_upsample_size = True
+                break
+
+        # ensure attention_mask is a bias, and give it a singleton query_tokens dimension
+        # expects mask of shape:
+        #   [batch, key_tokens]
+        # adds singleton query_tokens dimension:
+        #   [batch,                    1, key_tokens]
+        # this helps to broadcast it as a bias over attention scores, which will be in one of the following shapes:
+        #   [batch,  heads, query_tokens, key_tokens] (e.g. torch sdp attn)
+        #   [batch * heads, query_tokens, key_tokens] (e.g. xformers or classic attn)
+        if attention_mask is not None:
+            # assume that mask is expressed as:
+            #   (1 = keep,      0 = discard)
+            # convert mask into a bias that can be added to attention scores:
+            #       (keep = +0,     discard = -10000.0)
+            attention_mask = (1 - attention_mask.to(sample.dtype)) * -10000.0
+            attention_mask = attention_mask.unsqueeze(1)
+
+        # convert encoder_attention_mask to a bias the same way we do for attention_mask
+        if encoder_attention_mask is not None:
+            encoder_attention_mask = (1 - encoder_attention_mask.to(sample.dtype)) * -10000.0
+            encoder_attention_mask = encoder_attention_mask.unsqueeze(1)
+
+        # 0. center input if necessary
+        if self.config.center_input_sample:
+            sample = 2 * sample - 1.0
+
+        # 1. time
+        t_emb = self.get_time_embed(sample=sample, timestep=timestep)
+        emb = self.time_embedding(t_emb, timestep_cond)
+        aug_emb = None
+
+        class_emb = self.get_class_embed(sample=sample, class_labels=class_labels)
+        if class_emb is not None:
+            if self.config.class_embeddings_concat:
+                emb = torch.cat([emb, class_emb], dim=-1)
+            else:
+                emb = emb + class_emb
+
+        aug_emb = self.get_aug_embed(
+            emb=emb, encoder_hidden_states=encoder_hidden_states, added_cond_kwargs=added_cond_kwargs
+        )
+        if self.config.addition_embed_type == "image_hint":
+            aug_emb, hint = aug_emb
+            sample = torch.cat([sample, hint], dim=1)
+
+        emb = emb + aug_emb if aug_emb is not None else emb
+
+        if self.time_embed_act is not None:
+            emb = self.time_embed_act(emb)
+
+        encoder_hidden_states = self.process_encoder_hidden_states(
+            encoder_hidden_states=encoder_hidden_states, added_cond_kwargs=added_cond_kwargs
+        )
+
+        # 2. pre-process
+        sample = self.conv_in(sample)
+
+        # 2.5 GLIGEN position net
+        if cross_attention_kwargs is not None and cross_attention_kwargs.get("gligen", None) is not None:
+            cross_attention_kwargs = cross_attention_kwargs.copy()
+            gligen_args = cross_attention_kwargs.pop("gligen")
+            cross_attention_kwargs["gligen"] = {"objs": self.position_net(**gligen_args)}
+
+        # 3. down
+        # we're popping the `scale` instead of getting it because otherwise `scale` will be propagated
+        # to the internal blocks and will raise deprecation warnings. this will be confusing for our users.
+        if cross_attention_kwargs is not None:
+            cross_attention_kwargs = cross_attention_kwargs.copy()
+            lora_scale = cross_attention_kwargs.pop("scale", 1.0)
+        else:
+            lora_scale = 1.0
+
+        if USE_PEFT_BACKEND:
+            # weight the lora layers by setting `lora_scale` for each PEFT layer
+            scale_lora_layers(self, lora_scale)
+
+        is_controlnet = mid_block_additional_residual is not None and down_block_additional_residuals is not None
+        # using new arg down_intrablock_additional_residuals for T2I-Adapters, to distinguish from controlnets
+        is_adapter = down_intrablock_additional_residuals is not None
+        # maintain backward compatibility for legacy usage, where
+        #       T2I-Adapter and ControlNet both use down_block_additional_residuals arg
+        #       but can only use one or the other
+        if not is_adapter and mid_block_additional_residual is None and down_block_additional_residuals is not None:
+            deprecate(
+                "T2I should not use down_block_additional_residuals",
+                "1.3.0",
+                "Passing intrablock residual connections with `down_block_additional_residuals` is deprecated \
+                       and will be removed in diffusers 1.3.0.  `down_block_additional_residuals` should only be used \
+                       for ControlNet. Please make sure use `down_intrablock_additional_residuals` instead. ",
+                standard_warn=False,
+            )
+            down_intrablock_additional_residuals = down_block_additional_residuals
+            is_adapter = True
+
+        down_block_res_samples = (sample,)
+        for down_i, downsample_block in enumerate(self.down_blocks):
+            if hasattr(downsample_block, "has_cross_attention") and downsample_block.has_cross_attention:
+                # For t2i-adapter CrossAttnDownBlock2D
+                additional_residuals = {}
+                if is_adapter and len(down_intrablock_additional_residuals) > 0:
+                    additional_residuals["additional_residuals"] = down_intrablock_additional_residuals.pop(0)
+
+                sample, res_samples = downsample_block(
+                    hidden_states=sample,
+                    temb=emb,
+                    encoder_hidden_states=encoder_hidden_states,
+                    attention_mask=attention_mask,
+                    cross_attention_kwargs=cross_attention_kwargs,
+                    encoder_attention_mask=encoder_attention_mask,
+                    **additional_residuals,
+                )
+
+            else:
+                sample, res_samples = downsample_block(hidden_states=sample, temb=emb)
+                if is_adapter and len(down_intrablock_additional_residuals) > 0:
+                    sample += down_intrablock_additional_residuals.pop(0)
+
+            down_block_res_samples += res_samples
+
+            # kohya high res fix
+            if self.config.high_res_fix:
+                for high_res_fix in self.config.high_res_fix:
+                    if timestep > high_res_fix["timestep"] and down_i == high_res_fix["block_num"]:
+                        sample = self.__class__._resize(sample, scale_factor=high_res_fix["scale_factor"])
+                        break
+
+        if is_controlnet:
+            new_down_block_res_samples = ()
+
+            for down_block_res_sample, down_block_additional_residual in zip(
+                down_block_res_samples, down_block_additional_residuals
+            ):
+                down_block_res_sample = down_block_res_sample + down_block_additional_residual
+                new_down_block_res_samples = new_down_block_res_samples + (down_block_res_sample,)
+
+            down_block_res_samples = new_down_block_res_samples
+
+        # 4. mid
+        if self.mid_block is not None:
+            if hasattr(self.mid_block, "has_cross_attention") and self.mid_block.has_cross_attention:
+                sample = self.mid_block(
+                    sample,
+                    emb,
+                    encoder_hidden_states=encoder_hidden_states,
+                    attention_mask=attention_mask,
+                    cross_attention_kwargs=cross_attention_kwargs,
+                    encoder_attention_mask=encoder_attention_mask,
+                )
+            else:
+                sample = self.mid_block(sample, emb)
+
+            # To support T2I-Adapter-XL
+            if (
+                is_adapter
+                and len(down_intrablock_additional_residuals) > 0
+                and sample.shape == down_intrablock_additional_residuals[0].shape
+            ):
+                sample += down_intrablock_additional_residuals.pop(0)
+
+        if is_controlnet:
+            sample = sample + mid_block_additional_residual
+
+        # 5. up
+        for i, upsample_block in enumerate(self.up_blocks):
+            is_final_block = i == len(self.up_blocks) - 1
+
+            res_samples = down_block_res_samples[-len(upsample_block.resnets) :]
+            down_block_res_samples = down_block_res_samples[: -len(upsample_block.resnets)]
+
+            # up scaling of kohya high res fix
+            if self.config.high_res_fix is not None:
+                if res_samples[0].shape[-2:] != sample.shape[-2:]:
+                    sample = self.__class__._resize(sample, target=res_samples[0])
+                    res_samples_up_sampled = (res_samples[0],)
+                    for res_sample in res_samples[1:]:
+                        res_samples_up_sampled += (self.__class__._resize(res_sample, target=res_samples[0]),)
+                    res_samples = res_samples_up_sampled
+
+            # if we have not reached the final block and need to forward the
+            # upsample size, we do it here
+            if not is_final_block and forward_upsample_size:
+                upsample_size = down_block_res_samples[-1].shape[2:]
+
+            if hasattr(upsample_block, "has_cross_attention") and upsample_block.has_cross_attention:
+                sample = upsample_block(
+                    hidden_states=sample,
+                    temb=emb,
+                    res_hidden_states_tuple=res_samples,
+                    encoder_hidden_states=encoder_hidden_states,
+                    cross_attention_kwargs=cross_attention_kwargs,
+                    upsample_size=upsample_size,
+                    attention_mask=attention_mask,
+                    encoder_attention_mask=encoder_attention_mask,
+                )
+            else:
+                sample = upsample_block(
+                    hidden_states=sample,
+                    temb=emb,
+                    res_hidden_states_tuple=res_samples,
+                    upsample_size=upsample_size,
+                )
+
+        # 6. post-process
+        if self.conv_norm_out:
+            sample = self.conv_norm_out(sample)
+            sample = self.conv_act(sample)
+        sample = self.conv_out(sample)
+
+        if USE_PEFT_BACKEND:
+            # remove `lora_scale` from each PEFT layer
+            unscale_lora_layers(self, lora_scale)
+
+        if not return_dict:
+            return (sample,)
+
+        return UNet2DConditionOutput(sample=sample)
+
+    @classmethod
+    def from_unet(cls, unet: UNet2DConditionModel, high_res_fix: list):
+        config = dict((unet.config))
+        config["high_res_fix"] = high_res_fix
+        unet_high_res = cls(**config)
+        unet_high_res.load_state_dict(unet.state_dict())
+        unet_high_res.to(unet.dtype)
+        return unet_high_res
+
+
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```py
+        >>> import torch
+        >>> from diffusers import DiffusionPipeline
+
+        >>> pipe = DiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4",
+                                         custom_pipeline="kohya_hires_fix",
+                                         torch_dtype=torch.float16,
+                                         high_res_fix=[{'timestep': 600,
+                                                        'scale_factor': 0.5,
+                                                        'block_num': 1}])
+        >>> pipe = pipe.to("cuda")
+
+        >>> prompt = "a photo of an astronaut riding a horse on mars"
+        >>> image = pipe(prompt, height=1000, width=1600).images[0]
+        ```
+"""
+
+
+class StableDiffusionHighResFixPipeline(StableDiffusionPipeline):
+    r"""
+    Pipeline for text-to-image generation using Stable Diffusion with Kohya fix for high resolution generation.
+
+    This model inherits from [`StableDiffusionPipeline`]. Check the superclass documentation for the generic methods.
+
+    The pipeline also inherits the following loading methods:
+        - [`~loaders.TextualInversionLoaderMixin.load_textual_inversion`] for loading textual inversion embeddings
+        - [`~loaders.LoraLoaderMixin.load_lora_weights`] for loading LoRA weights
+        - [`~loaders.LoraLoaderMixin.save_lora_weights`] for saving LoRA weights
+        - [`~loaders.FromSingleFileMixin.from_single_file`] for loading `.ckpt` files
+        - [`~loaders.IPAdapterMixin.load_ip_adapter`] for loading IP Adapters
+
+    Args:
+        vae ([`AutoencoderKL`]):
+            Variational Auto-Encoder (VAE) model to encode and decode images to and from latent representations.
+        text_encoder ([`~transformers.CLIPTextModel`]):
+            Frozen text-encoder ([clip-vit-large-patch14](https://huggingface.co/openai/clip-vit-large-patch14)).
+        tokenizer ([`~transformers.CLIPTokenizer`]):
+            A `CLIPTokenizer` to tokenize text.
+        unet ([`UNet2DConditionModel`]):
+            A `UNet2DConditionModel` to denoise the encoded image latents.
+        scheduler ([`SchedulerMixin`]):
+            A scheduler to be used in combination with `unet` to denoise the encoded image latents. Can be one of
+            [`DDIMScheduler`], [`LMSDiscreteScheduler`], or [`PNDMScheduler`].
+        safety_checker ([`StableDiffusionSafetyChecker`]):
+            Classification module that estimates whether generated images could be considered offensive or harmful.
+            Please refer to the [model card](https://huggingface.co/runwayml/stable-diffusion-v1-5) for more details
+            about a model's potential harms.
+        feature_extractor ([`~transformers.CLIPImageProcessor`]):
+            A `CLIPImageProcessor` to extract features from generated images; used as inputs to the `safety_checker`.
+        high_res_fix (`List[Dict]`, *optional*, defaults to `[{'timestep': 600, 'scale_factor': 0.5, 'block_num': 1}]`):
+            Enables Kohya fix for high resolution generation. The activation maps are scaled based on the scale_factor up to the timestep at specified block_num.
+    """
+
+    model_cpu_offload_seq = "text_encoder->image_encoder->unet->vae"
+    _optional_components = ["safety_checker", "feature_extractor", "image_encoder"]
+    _exclude_from_cpu_offload = ["safety_checker"]
+    _callback_tensor_inputs = ["latents", "prompt_embeds", "negative_prompt_embeds"]
+
+    def __init__(
+        self,
+        vae: AutoencoderKL,
+        text_encoder: CLIPTextModel,
+        tokenizer: CLIPTokenizer,
+        unet: UNet2DConditionModel,
+        scheduler: KarrasDiffusionSchedulers,
+        safety_checker: StableDiffusionSafetyChecker,
+        feature_extractor: CLIPImageProcessor,
+        image_encoder: CLIPVisionModelWithProjection = None,
+        requires_safety_checker: bool = True,
+        high_res_fix: List[Dict] = [{"timestep": 600, "scale_factor": 0.5, "block_num": 1}],
+    ):
+        super().__init__(
+            vae=vae,
+            text_encoder=text_encoder,
+            tokenizer=tokenizer,
+            unet=unet,
+            scheduler=scheduler,
+            safety_checker=safety_checker,
+            feature_extractor=feature_extractor,
+            image_encoder=image_encoder,
+            requires_safety_checker=requires_safety_checker,
+        )
+
+        unet = UNet2DConditionModelHighResFix.from_unet(unet=unet, high_res_fix=high_res_fix)
+        self.register_modules(
+            vae=vae,
+            text_encoder=text_encoder,
+            tokenizer=tokenizer,
+            unet=unet,
+            scheduler=scheduler,
+            safety_checker=safety_checker,
+            feature_extractor=feature_extractor,
+            image_encoder=image_encoder,
+        )
+        self.vae_scale_factor = 2 ** (len(self.vae.config.block_out_channels) - 1)
+        self.image_processor = VaeImageProcessor(vae_scale_factor=self.vae_scale_factor)
+        self.register_to_config(requires_safety_checker=requires_safety_checker)

examples/community/latent_consistency_img2img.py

@@ -88,7 +88,7 @@ class LatentConsistencyModelImg2ImgPipeline(DiffusionPipeline):
                 torch device
             num_images_per_prompt (`int`):
                 number of images that should be generated per prompt
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
         """
@@ -240,14 +240,6 @@ class LatentConsistencyModelImg2ImgPipeline(DiffusionPipeline):
 
         return latents
 
-        if latents is None:
-            latents = torch.randn(shape, dtype=dtype).to(device)
-        else:
-            latents = latents.to(device)
-        # scale the initial noise by the standard deviation required by the scheduler
-        latents = latents * self.scheduler.init_noise_sigma
-        return latents
-
     def get_w_embedding(self, w, embedding_dim=512, dtype=torch.float32):
         """
         see https://github.com/google-research/vdm/blob/dc27b98a554f65cdc654b800da5aa1846545d41b/model_vdm.py#L298
@@ -290,10 +282,10 @@ class LatentConsistencyModelImg2ImgPipeline(DiffusionPipeline):
         width: Optional[int] = 768,
         guidance_scale: float = 7.5,
         num_images_per_prompt: Optional[int] = 1,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         num_inference_steps: int = 4,
         lcm_origin_steps: int = 50,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -403,16 +395,16 @@ class LCMSchedulerOutput(BaseOutput):
     """
     Output class for the scheduler's `step` function output.
     Args:
-        prev_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):
+        prev_sample (`torch.Tensor` of shape `(batch_size, num_channels, height, width)` for images):
             Computed sample `(x_{t-1})` of previous timestep. `prev_sample` should be used as next model input in the
             denoising loop.
-        pred_original_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):
+        pred_original_sample (`torch.Tensor` of shape `(batch_size, num_channels, height, width)` for images):
             The predicted denoised sample `(x_{0})` based on the model output from the current timestep.
             `pred_original_sample` can be used to preview progress or for guidance.
     """
 
-    prev_sample: torch.FloatTensor
-    denoised: Optional[torch.FloatTensor] = None
+    prev_sample: torch.Tensor
+    denoised: Optional[torch.Tensor] = None
 
 
 # Copied from diffusers.schedulers.scheduling_ddpm.betas_for_alpha_bar
@@ -460,10 +452,10 @@ def rescale_zero_terminal_snr(betas):
     """
     Rescales betas to have zero terminal SNR Based on https://arxiv.org/pdf/2305.08891.pdf (Algorithm 1)
     Args:
-        betas (`torch.FloatTensor`):
+        betas (`torch.Tensor`):
             the betas that the scheduler is being initialized with.
     Returns:
-        `torch.FloatTensor`: rescaled betas with zero terminal SNR
+        `torch.Tensor`: rescaled betas with zero terminal SNR
     """
     # Convert betas to alphas_bar_sqrt
     alphas = 1.0 - betas
@@ -573,7 +565,7 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
             # Glide cosine schedule
             self.betas = betas_for_alpha_bar(num_train_timesteps)
         else:
-            raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
+            raise NotImplementedError(f"{beta_schedule} is not implemented for {self.__class__}")
 
         # Rescale for zero SNR
         if rescale_betas_zero_snr:
@@ -595,17 +587,17 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
         self.num_inference_steps = None
         self.timesteps = torch.from_numpy(np.arange(0, num_train_timesteps)[::-1].copy().astype(np.int64))
 
-    def scale_model_input(self, sample: torch.FloatTensor, timestep: Optional[int] = None) -> torch.FloatTensor:
+    def scale_model_input(self, sample: torch.Tensor, timestep: Optional[int] = None) -> torch.Tensor:
         """
         Ensures interchangeability with schedulers that need to scale the denoising model input depending on the
         current timestep.
         Args:
-            sample (`torch.FloatTensor`):
+            sample (`torch.Tensor`):
                 The input sample.
             timestep (`int`, *optional*):
                 The current timestep in the diffusion chain.
         Returns:
-            `torch.FloatTensor`:
+            `torch.Tensor`:
                 A scaled input sample.
         """
         return sample
@@ -621,7 +613,7 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
         return variance
 
     # Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler._threshold_sample
-    def _threshold_sample(self, sample: torch.FloatTensor) -> torch.FloatTensor:
+    def _threshold_sample(self, sample: torch.Tensor) -> torch.Tensor:
         """
         "Dynamic thresholding: At each sampling step we set s to a certain percentile absolute pixel value in xt0 (the
         prediction of x_0 at timestep t), and if s > 1, then we threshold xt0 to the range [-s, s] and then divide by
@@ -693,25 +685,25 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
 
     def step(
         self,
-        model_output: torch.FloatTensor,
+        model_output: torch.Tensor,
         timeindex: int,
         timestep: int,
-        sample: torch.FloatTensor,
+        sample: torch.Tensor,
         eta: float = 0.0,
         use_clipped_model_output: bool = False,
         generator=None,
-        variance_noise: Optional[torch.FloatTensor] = None,
+        variance_noise: Optional[torch.Tensor] = None,
         return_dict: bool = True,
     ) -> Union[LCMSchedulerOutput, Tuple]:
         """
         Predict the sample from the previous timestep by reversing the SDE. This function propagates the diffusion
         process from the learned model outputs (most often the predicted noise).
         Args:
-            model_output (`torch.FloatTensor`):
+            model_output (`torch.Tensor`):
                 The direct output from learned diffusion model.
             timestep (`float`):
                 The current discrete timestep in the diffusion chain.
-            sample (`torch.FloatTensor`):
+            sample (`torch.Tensor`):
                 A current instance of a sample created by the diffusion process.
             eta (`float`):
                 The weight of noise for added noise in diffusion step.
@@ -722,7 +714,7 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
                 `use_clipped_model_output` has no effect.
             generator (`torch.Generator`, *optional*):
                 A random number generator.
-            variance_noise (`torch.FloatTensor`):
+            variance_noise (`torch.Tensor`):
                 Alternative to generating noise with `generator` by directly providing the noise for the variance
                 itself. Useful for methods such as [`CycleDiffusion`].
             return_dict (`bool`, *optional*, defaults to `True`):
@@ -785,10 +777,10 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
     # Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.add_noise
     def add_noise(
         self,
-        original_samples: torch.FloatTensor,
-        noise: torch.FloatTensor,
+        original_samples: torch.Tensor,
+        noise: torch.Tensor,
         timesteps: torch.IntTensor,
-    ) -> torch.FloatTensor:
+    ) -> torch.Tensor:
         # Make sure alphas_cumprod and timestep have same device and dtype as original_samples
         alphas_cumprod = self.alphas_cumprod.to(device=original_samples.device, dtype=original_samples.dtype)
         timesteps = timesteps.to(original_samples.device)
@@ -807,9 +799,7 @@ class LCMSchedulerWithTimestamp(SchedulerMixin, ConfigMixin):
         return noisy_samples
 
     # Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.get_velocity
-    def get_velocity(
-        self, sample: torch.FloatTensor, noise: torch.FloatTensor, timesteps: torch.IntTensor
-    ) -> torch.FloatTensor:
+    def get_velocity(self, sample: torch.Tensor, noise: torch.Tensor, timesteps: torch.IntTensor) -> torch.Tensor:
         # Make sure alphas_cumprod and timestep have same device and dtype as sample
         alphas_cumprod = self.alphas_cumprod.to(device=sample.device, dtype=sample.dtype)
         timesteps = timesteps.to(sample.device)

examples/community/latent_consistency_interpolate.py

@@ -281,8 +281,8 @@ class LatentConsistencyModelWalkPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
         clip_skip: Optional[int] = None,
     ):
@@ -302,10 +302,10 @@ class LatentConsistencyModelWalkPipeline(
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -506,7 +506,7 @@ class LatentConsistencyModelWalkPipeline(
                 data type of the generated embeddings
 
         Returns:
-            `torch.FloatTensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
+            `torch.Tensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
         """
         assert len(w.shape) == 1
         w = w * 1000.0
@@ -546,7 +546,7 @@ class LatentConsistencyModelWalkPipeline(
         height: int,
         width: int,
         callback_steps: int,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
         callback_on_step_end_tensor_inputs=None,
     ):
         if height % 8 != 0 or width % 8 != 0:
@@ -580,11 +580,11 @@ class LatentConsistencyModelWalkPipeline(
     @torch.no_grad()
     def interpolate_embedding(
         self,
-        start_embedding: torch.FloatTensor,
-        end_embedding: torch.FloatTensor,
+        start_embedding: torch.Tensor,
+        end_embedding: torch.Tensor,
         num_interpolation_steps: Union[int, List[int]],
         interpolation_type: str,
-    ) -> torch.FloatTensor:
+    ) -> torch.Tensor:
         if interpolation_type == "lerp":
             interpolation_fn = lerp
         elif interpolation_type == "slerp":
@@ -611,11 +611,11 @@ class LatentConsistencyModelWalkPipeline(
     @torch.no_grad()
     def interpolate_latent(
         self,
-        start_latent: torch.FloatTensor,
-        end_latent: torch.FloatTensor,
+        start_latent: torch.Tensor,
+        end_latent: torch.Tensor,
         num_interpolation_steps: Union[int, List[int]],
         interpolation_type: str,
-    ) -> torch.FloatTensor:
+    ) -> torch.Tensor:
         if interpolation_type == "lerp":
             interpolation_fn = lerp
         elif interpolation_type == "slerp":
@@ -663,8 +663,8 @@ class LatentConsistencyModelWalkPipeline(
         guidance_scale: float = 8.5,
         num_images_per_prompt: Optional[int] = 1,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -705,11 +705,11 @@ class LatentConsistencyModelWalkPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor is generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
             output_type (`str`, *optional*, defaults to `"pil"`):

examples/community/latent_consistency_txt2img.py

@@ -86,7 +86,7 @@ class LatentConsistencyModelPipeline(DiffusionPipeline):
                 torch device
             num_images_per_prompt (`int`):
                 number of images that should be generated per prompt
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
         """
@@ -208,10 +208,10 @@ class LatentConsistencyModelPipeline(DiffusionPipeline):
         width: Optional[int] = 768,
         guidance_scale: float = 7.5,
         num_images_per_prompt: Optional[int] = 1,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         num_inference_steps: int = 4,
         lcm_origin_steps: int = 50,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -310,16 +310,16 @@ class LCMSchedulerOutput(BaseOutput):
     """
     Output class for the scheduler's `step` function output.
     Args:
-        prev_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):
+        prev_sample (`torch.Tensor` of shape `(batch_size, num_channels, height, width)` for images):
             Computed sample `(x_{t-1})` of previous timestep. `prev_sample` should be used as next model input in the
             denoising loop.
-        pred_original_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):
+        pred_original_sample (`torch.Tensor` of shape `(batch_size, num_channels, height, width)` for images):
             The predicted denoised sample `(x_{0})` based on the model output from the current timestep.
             `pred_original_sample` can be used to preview progress or for guidance.
     """
 
-    prev_sample: torch.FloatTensor
-    denoised: Optional[torch.FloatTensor] = None
+    prev_sample: torch.Tensor
+    denoised: Optional[torch.Tensor] = None
 
 
 # Copied from diffusers.schedulers.scheduling_ddpm.betas_for_alpha_bar
@@ -367,10 +367,10 @@ def rescale_zero_terminal_snr(betas):
     """
     Rescales betas to have zero terminal SNR Based on https://arxiv.org/pdf/2305.08891.pdf (Algorithm 1)
     Args:
-        betas (`torch.FloatTensor`):
+        betas (`torch.Tensor`):
             the betas that the scheduler is being initialized with.
     Returns:
-        `torch.FloatTensor`: rescaled betas with zero terminal SNR
+        `torch.Tensor`: rescaled betas with zero terminal SNR
     """
     # Convert betas to alphas_bar_sqrt
     alphas = 1.0 - betas
@@ -477,7 +477,7 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
             # Glide cosine schedule
             self.betas = betas_for_alpha_bar(num_train_timesteps)
         else:
-            raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
+            raise NotImplementedError(f"{beta_schedule} is not implemented for {self.__class__}")
 
         # Rescale for zero SNR
         if rescale_betas_zero_snr:
@@ -499,17 +499,17 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
         self.num_inference_steps = None
         self.timesteps = torch.from_numpy(np.arange(0, num_train_timesteps)[::-1].copy().astype(np.int64))
 
-    def scale_model_input(self, sample: torch.FloatTensor, timestep: Optional[int] = None) -> torch.FloatTensor:
+    def scale_model_input(self, sample: torch.Tensor, timestep: Optional[int] = None) -> torch.Tensor:
         """
         Ensures interchangeability with schedulers that need to scale the denoising model input depending on the
         current timestep.
         Args:
-            sample (`torch.FloatTensor`):
+            sample (`torch.Tensor`):
                 The input sample.
             timestep (`int`, *optional*):
                 The current timestep in the diffusion chain.
         Returns:
-            `torch.FloatTensor`:
+            `torch.Tensor`:
                 A scaled input sample.
         """
         return sample
@@ -525,7 +525,7 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
         return variance
 
     # Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler._threshold_sample
-    def _threshold_sample(self, sample: torch.FloatTensor) -> torch.FloatTensor:
+    def _threshold_sample(self, sample: torch.Tensor) -> torch.Tensor:
         """
         "Dynamic thresholding: At each sampling step we set s to a certain percentile absolute pixel value in xt0 (the
         prediction of x_0 at timestep t), and if s > 1, then we threshold xt0 to the range [-s, s] and then divide by
@@ -593,25 +593,25 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
 
     def step(
         self,
-        model_output: torch.FloatTensor,
+        model_output: torch.Tensor,
         timeindex: int,
         timestep: int,
-        sample: torch.FloatTensor,
+        sample: torch.Tensor,
         eta: float = 0.0,
         use_clipped_model_output: bool = False,
         generator=None,
-        variance_noise: Optional[torch.FloatTensor] = None,
+        variance_noise: Optional[torch.Tensor] = None,
         return_dict: bool = True,
     ) -> Union[LCMSchedulerOutput, Tuple]:
         """
         Predict the sample from the previous timestep by reversing the SDE. This function propagates the diffusion
         process from the learned model outputs (most often the predicted noise).
         Args:
-            model_output (`torch.FloatTensor`):
+            model_output (`torch.Tensor`):
                 The direct output from learned diffusion model.
             timestep (`float`):
                 The current discrete timestep in the diffusion chain.
-            sample (`torch.FloatTensor`):
+            sample (`torch.Tensor`):
                 A current instance of a sample created by the diffusion process.
             eta (`float`):
                 The weight of noise for added noise in diffusion step.
@@ -622,7 +622,7 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
                 `use_clipped_model_output` has no effect.
             generator (`torch.Generator`, *optional*):
                 A random number generator.
-            variance_noise (`torch.FloatTensor`):
+            variance_noise (`torch.Tensor`):
                 Alternative to generating noise with `generator` by directly providing the noise for the variance
                 itself. Useful for methods such as [`CycleDiffusion`].
             return_dict (`bool`, *optional*, defaults to `True`):
@@ -685,10 +685,10 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
     # Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.add_noise
     def add_noise(
         self,
-        original_samples: torch.FloatTensor,
-        noise: torch.FloatTensor,
+        original_samples: torch.Tensor,
+        noise: torch.Tensor,
         timesteps: torch.IntTensor,
-    ) -> torch.FloatTensor:
+    ) -> torch.Tensor:
         # Make sure alphas_cumprod and timestep have same device and dtype as original_samples
         alphas_cumprod = self.alphas_cumprod.to(device=original_samples.device, dtype=original_samples.dtype)
         timesteps = timesteps.to(original_samples.device)
@@ -707,9 +707,7 @@ class LCMScheduler(SchedulerMixin, ConfigMixin):
         return noisy_samples
 
     # Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.get_velocity
-    def get_velocity(
-        self, sample: torch.FloatTensor, noise: torch.FloatTensor, timesteps: torch.IntTensor
-    ) -> torch.FloatTensor:
+    def get_velocity(self, sample: torch.Tensor, noise: torch.Tensor, timesteps: torch.IntTensor) -> torch.Tensor:
         # Make sure alphas_cumprod and timestep have same device and dtype as sample
         alphas_cumprod = self.alphas_cumprod.to(device=sample.device, dtype=sample.dtype)
         timesteps = timesteps.to(sample.device)

examples/community/llm_grounded_diffusion.py

@@ -756,13 +756,13 @@ class LLMGroundedDiffusionPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         clip_skip: Optional[int] = None,
@@ -807,14 +807,14 @@ class LLMGroundedDiffusionPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor is generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
             ip_adapter_image: (`PipelineImageInput`, *optional*): Optional image input to work with IP Adapters.
@@ -825,7 +825,7 @@ class LLMGroundedDiffusionPipeline(
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that calls every `callback_steps` steps during inference. The function is called with the
-                following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function is called. If not specified, the callback is called at
                 every step.
@@ -1194,8 +1194,8 @@ class LLMGroundedDiffusionPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
         **kwargs,
     ):
@@ -1227,8 +1227,8 @@ class LLMGroundedDiffusionPipeline(
         num_images_per_prompt,
         do_classifier_free_guidance,
         negative_prompt=None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
         clip_skip: Optional[int] = None,
     ):
@@ -1248,10 +1248,10 @@ class LLMGroundedDiffusionPipeline(
                 The prompt or prompts not to guide the image generation. If not defined, one has to pass
                 `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
                 less than `1`).
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -1509,7 +1509,7 @@ class LLMGroundedDiffusionPipeline(
                 data type of the generated embeddings
 
         Returns:
-            `torch.FloatTensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
+            `torch.Tensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
         """
         assert len(w.shape) == 1
         w = w * 1000.0
@@ -1524,35 +1524,35 @@ class LLMGroundedDiffusionPipeline(
         assert emb.shape == (w.shape[0], embedding_dim)
         return emb
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.guidance_scale
     @property
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.guidance_scale
     def guidance_scale(self):
         return self._guidance_scale
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.guidance_rescale
     @property
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.guidance_rescale
     def guidance_rescale(self):
         return self._guidance_rescale
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.clip_skip
     @property
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.clip_skip
     def clip_skip(self):
         return self._clip_skip
 
     # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
     # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
     # corresponds to doing no classifier free guidance.
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.do_classifier_free_guidance
     @property
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.do_classifier_free_guidance
     def do_classifier_free_guidance(self):
         return self._guidance_scale > 1 and self.unet.config.time_cond_proj_dim is None
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.cross_attention_kwargs
     @property
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.cross_attention_kwargs
     def cross_attention_kwargs(self):
         return self._cross_attention_kwargs
 
-    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.num_timesteps
     @property
+    # Copied from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion.StableDiffusionPipeline.num_timesteps
     def num_timesteps(self):
         return self._num_timesteps

examples/community/lpw_stable_diffusion.py

@@ -378,7 +378,7 @@ def preprocess_image(image, batch_size):
 
 
 def preprocess_mask(mask, batch_size, scale_factor=8):
-    if not isinstance(mask, torch.FloatTensor):
+    if not isinstance(mask, torch.Tensor):
         mask = mask.convert("L")
         w, h = mask.size
         w, h = (x - x % 8 for x in (w, h))  # resize to integer multiple of 8
@@ -543,8 +543,8 @@ class StableDiffusionLongPromptWeightingPipeline(
         do_classifier_free_guidance,
         negative_prompt=None,
         max_embeddings_multiples=3,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
     ):
         r"""
         Encodes the prompt into text encoder hidden states.
@@ -767,8 +767,8 @@ class StableDiffusionLongPromptWeightingPipeline(
         self,
         prompt: Union[str, List[str]],
         negative_prompt: Optional[Union[str, List[str]]] = None,
-        image: Union[torch.FloatTensor, PIL.Image.Image] = None,
-        mask_image: Union[torch.FloatTensor, PIL.Image.Image] = None,
+        image: Union[torch.Tensor, PIL.Image.Image] = None,
+        mask_image: Union[torch.Tensor, PIL.Image.Image] = None,
         height: int = 512,
         width: int = 512,
         num_inference_steps: int = 50,
@@ -778,13 +778,13 @@ class StableDiffusionLongPromptWeightingPipeline(
         add_predicted_noise: Optional[bool] = False,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         max_embeddings_multiples: Optional[int] = 3,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         is_cancelled_callback: Optional[Callable[[], bool]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -798,10 +798,10 @@ class StableDiffusionLongPromptWeightingPipeline(
             negative_prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation. Ignored when not using guidance (i.e., ignored
                 if `guidance_scale` is less than `1`).
-            image (`torch.FloatTensor` or `PIL.Image.Image`):
+            image (`torch.Tensor` or `PIL.Image.Image`):
                 `Image`, or tensor representing an image batch, that will be used as the starting point for the
                 process.
-            mask_image (`torch.FloatTensor` or `PIL.Image.Image`):
+            mask_image (`torch.Tensor` or `PIL.Image.Image`):
                 `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
                 replaced by noise and therefore repainted, while black pixels will be preserved. If `mask_image` is a
                 PIL image, it will be converted to a single channel (luminance) before use. If it's a tensor, it should
@@ -836,14 +836,14 @@ class StableDiffusionLongPromptWeightingPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                 to make generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -857,7 +857,7 @@ class StableDiffusionLongPromptWeightingPipeline(
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             is_cancelled_callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. If the function returns
                 `True`, the inference will be cancelled.
@@ -1032,13 +1032,13 @@ class StableDiffusionLongPromptWeightingPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         max_embeddings_multiples: Optional[int] = 3,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         is_cancelled_callback: Optional[Callable[[], bool]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -1072,14 +1072,14 @@ class StableDiffusionLongPromptWeightingPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                 to make generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -1093,7 +1093,7 @@ class StableDiffusionLongPromptWeightingPipeline(
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             is_cancelled_callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. If the function returns
                 `True`, the inference will be cancelled.
@@ -1137,7 +1137,7 @@ class StableDiffusionLongPromptWeightingPipeline(
 
     def img2img(
         self,
-        image: Union[torch.FloatTensor, PIL.Image.Image],
+        image: Union[torch.Tensor, PIL.Image.Image],
         prompt: Union[str, List[str]],
         negative_prompt: Optional[Union[str, List[str]]] = None,
         strength: float = 0.8,
@@ -1146,12 +1146,12 @@ class StableDiffusionLongPromptWeightingPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: Optional[float] = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         max_embeddings_multiples: Optional[int] = 3,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         is_cancelled_callback: Optional[Callable[[], bool]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -1159,7 +1159,7 @@ class StableDiffusionLongPromptWeightingPipeline(
         r"""
         Function for image-to-image generation.
         Args:
-            image (`torch.FloatTensor` or `PIL.Image.Image`):
+            image (`torch.Tensor` or `PIL.Image.Image`):
                 `Image`, or tensor representing an image batch, that will be used as the starting point for the
                 process.
             prompt (`str` or `List[str]`):
@@ -1190,10 +1190,10 @@ class StableDiffusionLongPromptWeightingPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                 to make generation deterministic.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -1207,7 +1207,7 @@ class StableDiffusionLongPromptWeightingPipeline(
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             is_cancelled_callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. If the function returns
                 `True`, the inference will be cancelled.
@@ -1249,8 +1249,8 @@ class StableDiffusionLongPromptWeightingPipeline(
 
     def inpaint(
         self,
-        image: Union[torch.FloatTensor, PIL.Image.Image],
-        mask_image: Union[torch.FloatTensor, PIL.Image.Image],
+        image: Union[torch.Tensor, PIL.Image.Image],
+        mask_image: Union[torch.Tensor, PIL.Image.Image],
         prompt: Union[str, List[str]],
         negative_prompt: Optional[Union[str, List[str]]] = None,
         strength: float = 0.8,
@@ -1260,12 +1260,12 @@ class StableDiffusionLongPromptWeightingPipeline(
         add_predicted_noise: Optional[bool] = False,
         eta: Optional[float] = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         max_embeddings_multiples: Optional[int] = 3,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         is_cancelled_callback: Optional[Callable[[], bool]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -1273,10 +1273,10 @@ class StableDiffusionLongPromptWeightingPipeline(
         r"""
         Function for inpaint.
         Args:
-            image (`torch.FloatTensor` or `PIL.Image.Image`):
+            image (`torch.Tensor` or `PIL.Image.Image`):
                 `Image`, or tensor representing an image batch, that will be used as the starting point for the
                 process. This is the image whose masked region will be inpainted.
-            mask_image (`torch.FloatTensor` or `PIL.Image.Image`):
+            mask_image (`torch.Tensor` or `PIL.Image.Image`):
                 `Image`, or tensor representing an image batch, to mask `image`. White pixels in the mask will be
                 replaced by noise and therefore repainted, while black pixels will be preserved. If `mask_image` is a
                 PIL image, it will be converted to a single channel (luminance) before use. If it's a tensor, it should
@@ -1311,10 +1311,10 @@ class StableDiffusionLongPromptWeightingPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                 to make generation deterministic.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
@@ -1328,7 +1328,7 @@ class StableDiffusionLongPromptWeightingPipeline(
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             is_cancelled_callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. If the function returns
                 `True`, the inference will be cancelled.

examples/community/lpw_stable_diffusion_xl.py

@@ -694,10 +694,10 @@ class SDXLLongPromptWeightingPipeline(
         do_classifier_free_guidance: bool = True,
         negative_prompt: Optional[str] = None,
         negative_prompt_2: Optional[str] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        pooled_prompt_embeds: Optional[torch.Tensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.Tensor] = None,
         lora_scale: Optional[float] = None,
     ):
         r"""
@@ -722,17 +722,17 @@ class SDXLLongPromptWeightingPipeline(
             negative_prompt_2 (`str` or `List[str]`, *optional*):
                 The prompt or prompts not to guide the image generation to be sent to `tokenizer_2` and
                 `text_encoder_2`. If not defined, `negative_prompt` is used in both text-encoders
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
-            pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+            pooled_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting.
                 If not provided, pooled text embeddings will be generated from `prompt` input argument.
-            negative_pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_pooled_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, pooled negative_prompt_embeds will be generated from `negative_prompt`
                 input argument.
@@ -1320,7 +1320,7 @@ class SDXLLongPromptWeightingPipeline(
                 data type of the generated embeddings
 
         Returns:
-            `torch.FloatTensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
+            `torch.Tensor`: Embedding vectors with shape `(len(timesteps), embedding_dim)`
         """
         assert len(w.shape) == 1
         w = w * 1000.0
@@ -1378,7 +1378,7 @@ class SDXLLongPromptWeightingPipeline(
         prompt_2: Optional[str] = None,
         image: Optional[PipelineImageInput] = None,
         mask_image: Optional[PipelineImageInput] = None,
-        masked_image_latents: Optional[torch.FloatTensor] = None,
+        masked_image_latents: Optional[torch.Tensor] = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
         strength: float = 0.8,
@@ -1392,12 +1392,12 @@ class SDXLLongPromptWeightingPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        pooled_prompt_embeds: Optional[torch.Tensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -1481,23 +1481,23 @@ class SDXLLongPromptWeightingPipeline(
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
                 to make generation deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
             ip_adapter_image: (`PipelineImageInput`, *optional*):
                 Optional image input to work with IP Adapters.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
                 provided, text embeddings will be generated from `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
-            pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+            pooled_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting.
                 If not provided, pooled text embeddings will be generated from `prompt` input argument.
-            negative_pooled_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_pooled_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative pooled text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, pooled negative_prompt_embeds will be generated from `negative_prompt`
                 input argument.
@@ -1926,12 +1926,12 @@ class SDXLLongPromptWeightingPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        pooled_prompt_embeds: Optional[torch.Tensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -2001,12 +2001,12 @@ class SDXLLongPromptWeightingPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        pooled_prompt_embeds: Optional[torch.Tensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
@@ -2066,7 +2066,7 @@ class SDXLLongPromptWeightingPipeline(
         prompt_2: Optional[str] = None,
         image: Optional[PipelineImageInput] = None,
         mask_image: Optional[PipelineImageInput] = None,
-        masked_image_latents: Optional[torch.FloatTensor] = None,
+        masked_image_latents: Optional[torch.Tensor] = None,
         height: Optional[int] = None,
         width: Optional[int] = None,
         strength: float = 0.8,
@@ -2080,12 +2080,12 @@ class SDXLLongPromptWeightingPipeline(
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         ip_adapter_image: Optional[PipelineImageInput] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
-        pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_pooled_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
+        pooled_prompt_embeds: Optional[torch.Tensor] = None,
+        negative_pooled_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,

examples/community/marigold_depth_estimation.py

@@ -43,7 +43,7 @@ from diffusers.utils import BaseOutput, check_min_version
 
 
 # Will error if the minimal version of diffusers is not installed. Remove at your own risks.
-check_min_version("0.25.0")
+check_min_version("0.29.0.dev0")
 
 
 class MarigoldDepthOutput(BaseOutput):

examples/community/masked_stable_diffusion_img2img.py

@@ -16,10 +16,10 @@ class MaskedStableDiffusionImg2ImgPipeline(StableDiffusionImg2ImgPipeline):
         self,
         prompt: Union[str, List[str]] = None,
         image: Union[
-            torch.FloatTensor,
+            torch.Tensor,
             PIL.Image.Image,
             np.ndarray,
-            List[torch.FloatTensor],
+            List[torch.Tensor],
             List[PIL.Image.Image],
             List[np.ndarray],
         ] = None,
@@ -30,18 +30,18 @@ class MaskedStableDiffusionImg2ImgPipeline(StableDiffusionImg2ImgPipeline):
         num_images_per_prompt: Optional[int] = 1,
         eta: Optional[float] = 0.0,
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
-        prompt_embeds: Optional[torch.FloatTensor] = None,
-        negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        prompt_embeds: Optional[torch.Tensor] = None,
+        negative_prompt_embeds: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
         cross_attention_kwargs: Optional[Dict[str, Any]] = None,
         mask: Union[
-            torch.FloatTensor,
+            torch.Tensor,
             PIL.Image.Image,
             np.ndarray,
-            List[torch.FloatTensor],
+            List[torch.Tensor],
             List[PIL.Image.Image],
             List[np.ndarray],
         ] = None,
@@ -52,7 +52,7 @@ class MaskedStableDiffusionImg2ImgPipeline(StableDiffusionImg2ImgPipeline):
         Args:
             prompt (`str` or `List[str]`, *optional*):
                 The prompt or prompts to guide image generation. If not defined, you need to pass `prompt_embeds`.
-            image (`torch.FloatTensor`, `PIL.Image.Image`, `np.ndarray`, `List[torch.FloatTensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`):
+            image (`torch.Tensor`, `PIL.Image.Image`, `np.ndarray`, `List[torch.Tensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`):
                 `Image` or tensor representing an image batch to be used as the starting point. Can also accept image
                 latents as `image`, but if passing latents directly it is not encoded again.
             strength (`float`, *optional*, defaults to 0.8):
@@ -78,10 +78,10 @@ class MaskedStableDiffusionImg2ImgPipeline(StableDiffusionImg2ImgPipeline):
             generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
                 A [`torch.Generator`](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make
                 generation deterministic.
-            prompt_embeds (`torch.FloatTensor`, *optional*):
+            prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated text embeddings. Can be used to easily tweak text inputs (prompt weighting). If not
                 provided, text embeddings are generated from the `prompt` input argument.
-            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+            negative_prompt_embeds (`torch.Tensor`, *optional*):
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs (prompt weighting). If
                 not provided, `negative_prompt_embeds` are generated from the `negative_prompt` input argument.
             output_type (`str`, *optional*, defaults to `"pil"`):
@@ -91,14 +91,14 @@ class MaskedStableDiffusionImg2ImgPipeline(StableDiffusionImg2ImgPipeline):
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that calls every `callback_steps` steps during inference. The function is called with the
-                following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function is called. If not specified, the callback is called at
                 every step.
             cross_attention_kwargs (`dict`, *optional*):
                 A kwargs dictionary that if specified is passed along to the [`AttentionProcessor`] as defined in
                 [`self.processor`](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
-            mask (`torch.FloatTensor`, `PIL.Image.Image`, `np.ndarray`, `List[torch.FloatTensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`, *optional*):
+            mask (`torch.Tensor`, `PIL.Image.Image`, `np.ndarray`, `List[torch.Tensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`, *optional*):
                 A mask with non-zero elements for the area to be inpainted. If not specified, no mask is applied.
         Examples:
 

examples/community/mixture_canvas.py

@@ -154,7 +154,7 @@ class Text2ImageRegion(DiffusionRegion):
 class Image2ImageRegion(DiffusionRegion):
     """Class defining a region where an image guided diffusion process is acting"""
 
-    reference_image: torch.FloatTensor = None
+    reference_image: torch.Tensor = None
     strength: float = 0.8  # Strength of the image
 
     def __post_init__(self):

examples/community/multilingual_stable_diffusion.py

@@ -147,10 +147,10 @@ class MultilingualStableDiffusion(DiffusionPipeline, StableDiffusionMixin):
         num_images_per_prompt: Optional[int] = 1,
         eta: float = 0.0,
         generator: Optional[torch.Generator] = None,
-        latents: Optional[torch.FloatTensor] = None,
+        latents: Optional[torch.Tensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
+        callback: Optional[Callable[[int, int, torch.Tensor], None]] = None,
         callback_steps: int = 1,
         **kwargs,
     ):
@@ -184,7 +184,7 @@ class MultilingualStableDiffusion(DiffusionPipeline, StableDiffusionMixin):
             generator (`torch.Generator`, *optional*):
                 A [torch generator](https://pytorch.org/docs/stable/generated/torch.Generator.html) to make generation
                 deterministic.
-            latents (`torch.FloatTensor`, *optional*):
+            latents (`torch.Tensor`, *optional*):
                 Pre-generated noisy latents, sampled from a Gaussian distribution, to be used as inputs for image
                 generation. Can be used to tweak the same generation with different prompts. If not provided, a latents
                 tensor will ge generated by sampling using the supplied random `generator`.
@@ -196,7 +196,7 @@ class MultilingualStableDiffusion(DiffusionPipeline, StableDiffusionMixin):
                 plain tuple.
             callback (`Callable`, *optional*):
                 A function that will be called every `callback_steps` steps during inference. The function will be
-                called with the following arguments: `callback(step: int, timestep: int, latents: torch.FloatTensor)`.
+                called with the following arguments: `callback(step: int, timestep: int, latents: torch.Tensor)`.
             callback_steps (`int`, *optional*, defaults to 1):
                 The frequency at which the `callback` function will be called. If not specified, the callback will be
                 called at every step.