fix audio pipeline with torchcodec input

* fix

* fix

* fix

* fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

.github/workflows/check_failed_tests.yml

@@ -41,7 +41,7 @@ jobs:
   check_new_failures:
     name: " "
     runs-on:
-      group: aws-g4dn-4xlarge-cache
+      group: aws-g5-4xlarge-cache
     container:
       image: ${{ inputs.docker }}
       options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/

.github/workflows/doctest_job.yml

@@ -28,7 +28,7 @@ jobs:
       matrix:
         split_keys: ${{ fromJson(inputs.split_keys) }}
     runs-on: 
-      group: aws-g4dn-4xlarge-cache
+      group: aws-g5-4xlarge-cache
     container:
       image: huggingface/transformers-all-latest-gpu
       options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/

.github/workflows/doctests.yml

@@ -15,7 +15,7 @@ jobs:
   setup:
     name: Setup
     runs-on: 
-      group: aws-g4dn-4xlarge-cache
+      group: aws-g5-4xlarge-cache
     container:
       image: huggingface/transformers-all-latest-gpu
       options: --gpus 0 --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/

.github/workflows/get-pr-info.yml

@@ -0,0 +1,157 @@
+name: Get PR commit SHA
+on:
+  workflow_call:
+    inputs:
+      pr_number:
+        required: true
+        type: string
+    outputs:
+      PR_HEAD_REPO_FULL_NAME:
+        description: "The full name of the repository from which the pull request is created"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_REPO_FULL_NAME }}
+      PR_BASE_REPO_FULL_NAME:
+        description: "The full name of the repository to which the pull request is created"
+        value: ${{ jobs.get-pr-info.outputs.PR_BASE_REPO_FULL_NAME }}
+      PR_HEAD_REPO_OWNER:
+        description: "The owner of the repository from which the pull request is created"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_REPO_OWNER }}
+      PR_BASE_REPO_OWNER:
+        description: "The owner of the repository to which the pull request is created"
+        value: ${{ jobs.get-pr-info.outputs.PR_BASE_REPO_OWNER }}
+      PR_HEAD_REPO_NAME:
+        description: "The name of the repository from which the pull request is created"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_REPO_NAME }}
+      PR_BASE_REPO_NAME:
+        description: "The name of the repository to which the pull request is created"
+        value: ${{ jobs.get-pr-info.outputs.PR_BASE_REPO_NAME }}
+      PR_HEAD_REF:
+        description: "The branch name of the pull request in the head repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_REF }}
+      PR_BASE_REF:
+        description: "The branch name in the base repository (to merge into)"
+        value: ${{ jobs.get-pr-info.outputs.PR_BASE_REF }}
+      PR_HEAD_SHA:
+        description: "The head sha of the pull request branch in the head repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_SHA }}
+      PR_BASE_SHA:
+        description: "The head sha of the target branch in the base repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_BASE_SHA }}
+      PR_MERGE_COMMIT_SHA:
+        description: "The sha of the merge commit for the pull request (created by GitHub) in the base repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_MERGE_COMMIT_SHA }}
+      PR_HEAD_COMMIT_DATE:
+        description: "The date of the head sha of the pull request branch in the head repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_COMMIT_DATE }}
+      PR_MERGE_COMMIT_DATE:
+        description: "The date of the merge commit for the pull request (created by GitHub) in the base repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_MERGE_COMMIT_DATE }}
+      PR_HEAD_COMMIT_TIMESTAMP:
+        description: "The timestamp of the head sha of the pull request branch in the head repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_HEAD_COMMIT_TIMESTAMP }}
+      PR_MERGE_COMMIT_TIMESTAMP:
+        description: "The timestamp of the merge commit for the pull request (created by GitHub) in the base repository"
+        value: ${{ jobs.get-pr-info.outputs.PR_MERGE_COMMIT_TIMESTAMP }}
+      PR:
+        description: "The PR"
+        value: ${{ jobs.get-pr-info.outputs.PR }}
+      PR_FILES:
+        description: "The files touched in the PR"
+        value: ${{ jobs.get-pr-info.outputs.PR_FILES }}
+
+
+jobs:
+  get-pr-info:
+    runs-on: ubuntu-22.04
+    name: Get PR commit SHA better
+    outputs:
+      PR_HEAD_REPO_FULL_NAME: ${{ steps.pr_info.outputs.head_repo_full_name }}
+      PR_BASE_REPO_FULL_NAME: ${{ steps.pr_info.outputs.base_repo_full_name }}
+      PR_HEAD_REPO_OWNER: ${{ steps.pr_info.outputs.head_repo_owner }}
+      PR_BASE_REPO_OWNER: ${{ steps.pr_info.outputs.base_repo_owner }}
+      PR_HEAD_REPO_NAME: ${{ steps.pr_info.outputs.head_repo_name }}
+      PR_BASE_REPO_NAME: ${{ steps.pr_info.outputs.base_repo_name }}
+      PR_HEAD_REF: ${{ steps.pr_info.outputs.head_ref }}
+      PR_BASE_REF: ${{ steps.pr_info.outputs.base_ref }}
+      PR_HEAD_SHA: ${{ steps.pr_info.outputs.head_sha }}
+      PR_BASE_SHA: ${{ steps.pr_info.outputs.base_sha }}
+      PR_MERGE_COMMIT_SHA: ${{ steps.pr_info.outputs.merge_commit_sha }}
+      PR_HEAD_COMMIT_DATE: ${{ steps.pr_info.outputs.head_commit_date }}
+      PR_MERGE_COMMIT_DATE: ${{ steps.pr_info.outputs.merge_commit_date }}
+      PR_HEAD_COMMIT_TIMESTAMP: ${{ steps.get_timestamps.outputs.head_commit_timestamp }}
+      PR_MERGE_COMMIT_TIMESTAMP: ${{ steps.get_timestamps.outputs.merge_commit_timestamp }}
+      PR: ${{ steps.pr_info.outputs.pr }}
+      PR_FILES: ${{ steps.pr_info.outputs.files }}
+    if: ${{ inputs.pr_number != '' }}
+    steps:
+      - name: Extract PR details
+        id: pr_info
+        uses: actions/github-script@v6
+        with:
+          script: |            
+            const { data: pr } = await github.rest.pulls.get({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              pull_number: ${{ inputs.pr_number }}
+            });
+
+            const { data: head_commit }  = await github.rest.repos.getCommit({
+              owner: pr.head.repo.owner.login,
+              repo: pr.head.repo.name,
+              ref: pr.head.ref
+            });
+
+            const { data: merge_commit }  = await github.rest.repos.getCommit({
+              owner: pr.base.repo.owner.login,
+              repo: pr.base.repo.name,
+              ref: pr.merge_commit_sha,
+            });
+
+            const { data: files } = await github.rest.pulls.listFiles({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              pull_number: ${{ inputs.pr_number }}
+            });
+
+            core.setOutput('head_repo_full_name', pr.head.repo.full_name);
+            core.setOutput('base_repo_full_name', pr.base.repo.full_name);
+            core.setOutput('head_repo_owner', pr.head.repo.owner.login);
+            core.setOutput('base_repo_owner', pr.base.repo.owner.login);
+            core.setOutput('head_repo_name', pr.head.repo.name);
+            core.setOutput('base_repo_name', pr.base.repo.name);
+            core.setOutput('head_ref', pr.head.ref);
+            core.setOutput('base_ref', pr.base.ref);
+            core.setOutput('head_sha', pr.head.sha);
+            core.setOutput('base_sha', pr.base.sha);
+            core.setOutput('merge_commit_sha', pr.merge_commit_sha);
+            core.setOutput('pr', pr);
+
+            core.setOutput('head_commit_date', head_commit.commit.committer.date);
+            core.setOutput('merge_commit_date', merge_commit.commit.committer.date);
+            
+            core.setOutput('files', files);            
+            
+            console.log('PR head commit:', {
+              head_commit: head_commit,
+              commit: head_commit.commit,
+              date: head_commit.commit.committer.date
+            });
+
+            console.log('PR merge commit:', {
+              merge_commit: merge_commit,
+              commit: merge_commit.commit,
+              date: merge_commit.commit.committer.date
+            });
+
+      - name: Convert dates to timestamps
+        id: get_timestamps
+        run: |
+          head_commit_date=${{ steps.pr_info.outputs.head_commit_date }}
+          merge_commit_date=${{ steps.pr_info.outputs.merge_commit_date }}
+          echo $head_commit_date
+          echo $merge_commit_date
+          head_commit_timestamp=$(date -d "$head_commit_date" +%s)
+          merge_commit_timestamp=$(date -d "$merge_commit_date" +%s)
+          echo $head_commit_timestamp
+          echo $merge_commit_timestamp
+          echo "head_commit_timestamp=$head_commit_timestamp" >> $GITHUB_OUTPUT
+          echo "merge_commit_timestamp=$merge_commit_timestamp" >> $GITHUB_OUTPUT

.github/workflows/get-pr-number.yml

@@ -0,0 +1,36 @@
+name: Get PR number
+on:
+  workflow_call:
+    outputs:
+      PR_NUMBER:
+        description: "The extracted PR number"
+        value: ${{ jobs.get-pr-number.outputs.PR_NUMBER }}
+
+jobs:
+  get-pr-number:
+    runs-on: ubuntu-22.04
+    name: Get PR number
+    outputs:
+      PR_NUMBER: ${{ steps.set_pr_number.outputs.PR_NUMBER }}
+    steps:
+      - name: Get PR number
+        shell: bash
+        run: |
+          if [[ "${{ github.event.issue.number }}" != "" && "${{ github.event.issue.pull_request }}" != "" ]]; then
+            echo "PR_NUMBER=${{ github.event.issue.number }}" >> $GITHUB_ENV
+          elif [[ "${{ github.event.pull_request.number }}" != "" ]]; then
+            echo "PR_NUMBER=${{ github.event.pull_request.number }}" >> $GITHUB_ENV
+          elif [[ "${{ github.event.pull_request }}" != "" ]]; then
+            echo "PR_NUMBER=${{ github.event.number }}" >> $GITHUB_ENV
+          else
+            echo "PR_NUMBER=" >> $GITHUB_ENV
+          fi
+
+      - name: Check PR number
+        shell: bash
+        run: |
+          echo "${{ env.PR_NUMBER }}"
+
+      - name: Set PR number
+        id: set_pr_number
+        run: echo "PR_NUMBER=${{ env.PR_NUMBER }}" >> "$GITHUB_OUTPUT"

.github/workflows/model_jobs.yml

@@ -12,8 +12,8 @@ on:
       slice_id:
         required: true
         type: number
-      runner:
-        required: true
+      runner_map:
+        required: false
         type: string
       docker:
         required: true
@@ -45,7 +45,7 @@ jobs:
       matrix:
         folders: ${{ fromJson(inputs.folder_slices)[inputs.slice_id] }}
     runs-on:
-      group: '${{ inputs.machine_type }}'
+      group: ${{ fromJson(inputs.runner_map)[matrix.folders][inputs.machine_type] }}
     container:
       image: ${{ inputs.docker }}
       options: --gpus all --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
@@ -107,9 +107,9 @@ jobs:
         run: |
           echo "${{ inputs.machine_type }}"
 
-          if [ "${{ inputs.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ inputs.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ inputs.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ inputs.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ inputs.machine_type }}

.github/workflows/model_jobs_amd.yml

@@ -1,128 +0,0 @@
-name: model jobs
-
-on:
-  workflow_call:
-    inputs:
-      folder_slices:
-        required: true
-        type: string
-      machine_type:
-        required: true
-        type: string
-      slice_id:
-        required: true
-        type: number
-      runner:
-        required: true
-        type: string
-      docker:
-        required: true
-        type: string
-
-env:
-  HF_HOME: /mnt/cache
-  TRANSFORMERS_IS_CI: yes
-  OMP_NUM_THREADS: 8
-  MKL_NUM_THREADS: 8
-  RUN_SLOW: yes
-  # For gated repositories, we still need to agree to share information on the Hub repo. page in order to get access.
-  # This token is created under the bot `hf-transformers-bot`.
-  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
-  SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
-  TF_FORCE_GPU_ALLOW_GROWTH: true
-  CUDA_VISIBLE_DEVICES: 0,1
-
-jobs:
-  run_models_gpu:
-    name: " "
-    strategy:
-      max-parallel: 1  # For now, not to parallelize. Can change later if it works well.
-      fail-fast: false
-      matrix:
-        folders: ${{ fromJson(inputs.folder_slices)[inputs.slice_id] }}
-    runs-on: ['${{ inputs.machine_type }}', self-hosted, amd-gpu, '${{ inputs.runner }}']
-    container:
-      image: ${{ inputs.docker }}
-      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
-    steps:
-      - name: Echo input and matrix info
-        shell: bash
-        run: |
-          echo "${{ inputs.folder_slices }}"
-          echo "${{ matrix.folders }}"
-          echo "${{ toJson(fromJson(inputs.folder_slices)[inputs.slice_id]) }}"
-
-      - name: Echo folder ${{ matrix.folders }}
-        shell: bash
-        # For folders like `models/bert`, set an env. var. (`matrix_folders`) to `models_bert`, which will be used to
-        # set the artifact folder names (because the character `/` is not allowed).
-        run: |
-          echo "${{ matrix.folders }}"
-          matrix_folders=${{ matrix.folders }}
-          matrix_folders=${matrix_folders/'models/'/'models_'}
-          echo "$matrix_folders"
-          echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
-
-      - name: Update clone
-        working-directory: /transformers
-        run: git fetch && git checkout ${{ github.sha }}
-
-      - name: Reinstall transformers in edit mode (remove the one installed during docker image build)
-        working-directory: /transformers
-        run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
-
-      - name: Update / Install some packages (for Past CI)
-        if: ${{ contains(inputs.docker, '-past-') }}
-        working-directory: /transformers
-        run: |
-          python3 -m pip install -U datasets
-
-      - name: Update / Install some packages (for Past CI)
-        if: ${{ contains(inputs.docker, '-past-') && contains(inputs.docker, '-pytorch-') }}
-        working-directory: /transformers
-        run: |
-          python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
-
-      - name: ROCM-SMI
-        run: |
-          rocm-smi
-
-      - name: ROCM-INFO
-        run: |
-          rocminfo  | grep "Agent" -A 14
-
-      - name: Show ROCR environment
-        run: |
-          echo "ROCR: $ROCR_VISIBLE_DEVICES"
-
-      - name: Environment
-        working-directory: /transformers
-        run: |
-          python3 utils/print_env.py
-
-      - name: Show installed libraries and their versions
-        working-directory: /transformers
-        run: pip freeze
-
-      - name: Run all tests on GPU
-        working-directory: /transformers
-        run: python3 -m pytest -rsfE -v --make-reports=${{ inputs.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports tests/${{ matrix.folders }}  -m "not not_device_test"
-
-      - name: Failure short reports
-        if: ${{ failure() }}
-        continue-on-error: true
-        run: cat /transformers/reports/${{ inputs.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports/failures_short.txt
-
-      - name: Run test
-        shell: bash
-        run: |
-          mkdir -p /transformers/reports/${{ inputs.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports
-          echo "hello" > /transformers/reports/${{ inputs.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports/hello.txt
-          echo "${{ inputs.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports"
-
-      - name: "Test suite reports artifacts: ${{ inputs.machine_type }}_run_models_gpu_${{ env.matrix_folders }}_test_reports"
-        if: ${{ always() }}
-        uses: actions/upload-artifact@v4
-        with:
-          name: ${{ inputs.machine_type }}_run_models_gpu_${{ env.matrix_folders }}_test_reports
-          path: /transformers/reports/${{ inputs.machine_type }}_run_models_gpu_${{ matrix.folders }}_test_reports

.github/workflows/model_jobs_intel_gaudi.yml

@@ -0,0 +1,121 @@
+name: model jobs
+
+on:
+  workflow_call:
+    inputs:
+      folder_slices:
+        required: true
+        type: string
+      slice_id:
+        required: true
+        type: number
+      runner:
+        required: true
+        type: string
+      machine_type:
+        required: true
+        type: string
+      report_name_prefix:
+        required: false
+        default: run_models_gpu
+        type: string
+
+env:
+  RUN_SLOW: yes
+  PT_HPU_LAZY_MODE: 0
+  TRANSFORMERS_IS_CI: yes
+  PT_ENABLE_INT64_SUPPORT: 1
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
+  SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
+  HF_HOME: /mnt/cache/.cache/huggingface
+
+jobs:
+  run_models_gpu:
+    name: " "
+    strategy:
+      max-parallel: 8
+      fail-fast: false
+      matrix:
+        folders: ${{ fromJson(inputs.folder_slices)[inputs.slice_id] }}
+    runs-on:
+      group: ${{ inputs.runner }}
+    container:
+      image: vault.habana.ai/gaudi-docker/1.21.1/ubuntu22.04/habanalabs/pytorch-installer-2.6.0:latest
+      options: --runtime=habana
+        -v /mnt/cache/.cache/huggingface:/mnt/cache/.cache/huggingface
+        --env OMPI_MCA_btl_vader_single_copy_mechanism=none
+        --env HABANA_VISIBLE_DEVICES
+        --env HABANA_VISIBLE_MODULES
+        --cap-add=sys_nice
+        --shm-size=64G
+    steps:
+      - name: Echo input and matrix info
+        shell: bash
+        run: |
+          echo "${{ inputs.folder_slices }}"
+          echo "${{ matrix.folders }}"
+          echo "${{ toJson(fromJson(inputs.folder_slices)[inputs.slice_id]) }}"
+
+      - name: Echo folder ${{ matrix.folders }}
+        shell: bash
+        run: |
+          echo "${{ matrix.folders }}"
+          matrix_folders=${{ matrix.folders }}
+          matrix_folders=${matrix_folders/'models/'/'models_'}
+          echo "$matrix_folders"
+          echo "matrix_folders=$matrix_folders" >> $GITHUB_ENV
+
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Install dependencies
+        run: |
+          pip install -e .[testing,torch] "numpy<2.0.0" scipy scikit-learn
+
+      - name: HL-SMI
+        run: |
+          hl-smi
+          echo "HABANA_VISIBLE_DEVICES=${HABANA_VISIBLE_DEVICES}"
+          echo "HABANA_VISIBLE_MODULES=${HABANA_VISIBLE_MODULES}"
+
+      - name: Environment
+        run: python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        run: pip freeze
+
+      - name: Set `machine_type` for report and artifact names
+        shell: bash
+        run: |
+          if [ "${{ inputs.machine_type }}" = "1gaudi" ]; then
+            machine_type=single-gpu
+          elif [ "${{ inputs.machine_type }}" = "2gaudi" ]; then
+            machine_type=multi-gpu
+          else
+            machine_type=${{ inputs.machine_type }}
+          fi
+          echo "machine_type=$machine_type" >> $GITHUB_ENV
+
+      - name: Run all tests on Gaudi
+        run: python3 -m pytest -v --make-reports=${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ matrix.folders }}_test_reports tests/${{ matrix.folders }}
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: cat reports/${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ matrix.folders }}_test_reports/failures_short.txt
+
+      - name: Run test
+        shell: bash
+        run: |
+          mkdir -p reports/${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ matrix.folders }}_test_reports
+          echo "hello" > reports/${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ matrix.folders }}_test_reports/hello.txt
+          echo "${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ matrix.folders }}_test_reports"
+
+      - name: "Test suite reports artifacts: ${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ env.matrix_folders }}_test_reports"
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: ${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ env.matrix_folders }}_test_reports
+          path: reports/${{ env.machine_type }}_${{ inputs.report_name_prefix }}_${{ matrix.folders }}_test_reports

.github/workflows/pr_run_slow_ci.yml

@@ -0,0 +1,199 @@
+name: PR slow CI
+on:
+  pull_request_target:
+    types: [opened, synchronize, reopened]
+
+jobs:
+  get-pr-number:
+    name: Get PR number
+    uses: ./.github/workflows/get-pr-number.yml
+
+  get-pr-info:
+    name: Get PR commit SHA
+    needs: get-pr-number
+    if: ${{ needs.get-pr-number.outputs.PR_NUMBER != ''}}
+    uses: ./.github/workflows/get-pr-info.yml
+    with:
+      pr_number: ${{ needs.get-pr-number.outputs.PR_NUMBER }}
+
+  # We only need to verify the timestamp if the workflow is triggered by `issue_comment`.
+  verity_pr_commit:
+    name: Verity PR commit corresponds to a specific event by comparing timestamps
+    if: ${{ github.event.comment.created_at != '' }}
+    runs-on: ubuntu-22.04
+    needs: get-pr-info
+    env:
+      COMMENT_DATE: ${{ github.event.comment.created_at }}
+      PR_MERGE_COMMIT_DATE: ${{ needs.get-pr-info.outputs.PR_MERGE_COMMIT_DATE }}
+      PR_MERGE_COMMIT_TIMESTAMP: ${{ needs.get-pr-info.outputs.PR_MERGE_COMMIT_TIMESTAMP }}
+    steps:
+      - run: |
+          COMMENT_TIMESTAMP=$(date -d "${COMMENT_DATE}" +"%s")
+          echo "COMMENT_DATE: $COMMENT_DATE"
+          echo "PR_MERGE_COMMIT_DATE: $PR_MERGE_COMMIT_DATE"
+          echo "COMMENT_TIMESTAMP: $COMMENT_TIMESTAMP"
+          echo "PR_MERGE_COMMIT_TIMESTAMP: $PR_MERGE_COMMIT_TIMESTAMP"
+          if [ $COMMENT_TIMESTAMP -le $PR_MERGE_COMMIT_TIMESTAMP ]; then
+            echo "Last commit on the pull request is newer than the issue comment triggering this run! Abort!";
+            exit -1;
+          fi
+
+  get-jobs:
+    name: Get test files to run
+    runs-on: ubuntu-22.04
+    needs: [get-pr-number, get-pr-info]
+    outputs:
+      jobs: ${{ steps.get_jobs.outputs.jobs_to_run }}
+    steps:
+      - name: Get repository content
+        id: repo_content
+        uses: actions/github-script@v6
+        with:
+          script: |
+            const { data: tests_dir } = await github.rest.repos.getContent({
+              owner: '${{ needs.get-pr-info.outputs.PR_HEAD_REPO_OWNER }}',
+              repo: '${{ needs.get-pr-info.outputs.PR_HEAD_REPO_NAME }}',
+              path: 'tests',
+              ref: '${{ needs.get-pr-info.outputs.PR_HEAD_SHA }}',
+            });
+
+            const { data: tests_models_dir } = await github.rest.repos.getContent({
+              owner: '${{ needs.get-pr-info.outputs.PR_HEAD_REPO_OWNER }}',
+              repo: '${{ needs.get-pr-info.outputs.PR_HEAD_REPO_NAME }}',
+              path: 'tests/models',
+              ref: '${{ needs.get-pr-info.outputs.PR_HEAD_SHA }}',
+            });
+
+            const { data: tests_quantization_dir } = await github.rest.repos.getContent({
+              owner: '${{ needs.get-pr-info.outputs.PR_HEAD_REPO_OWNER }}',
+              repo: '${{ needs.get-pr-info.outputs.PR_HEAD_REPO_NAME }}',
+              path: 'tests/quantization',
+              ref: '${{ needs.get-pr-info.outputs.PR_HEAD_SHA }}',
+            });
+
+            core.setOutput('tests_dir', tests_dir);
+            core.setOutput('tests_models_dir', tests_models_dir);
+            core.setOutput('tests_quantization_dir', tests_quantization_dir);
+
+      # This checkout to the main branch
+      - uses: actions/checkout@v4
+        with:
+          fetch-depth: "0"
+
+      - name: Write pr_files file
+        run: |
+          cat > pr_files.txt << 'EOF'
+          ${{ needs.get-pr-info.outputs.PR_FILES }}
+          EOF
+
+      - name: Write tests_dir file
+        run: |
+          cat > tests_dir.txt << 'EOF'
+          ${{ steps.repo_content.outputs.tests_dir }}
+          EOF
+
+      - name: Write tests_models_dir file
+        run: |
+          cat > tests_models_dir.txt << 'EOF'
+          ${{ steps.repo_content.outputs.tests_models_dir }}
+          EOF
+
+      - name: Write tests_quantization_dir file
+        run: |
+          cat > tests_quantization_dir.txt << 'EOF'
+          ${{ steps.repo_content.outputs.tests_quantization_dir }}
+          EOF
+
+      - name: Run script to get jobs to run
+        id: get_jobs
+        run: |
+          python utils/get_pr_run_slow_jobs.py | tee output.txt
+          echo "jobs_to_run: $(tail -n 1 output.txt)"
+          echo "jobs_to_run=$(tail -n 1 output.txt)" >> $GITHUB_OUTPUT
+
+  send_comment:
+    # Will delete the previous comment and send a new one if:
+    #   - either the content is changed
+    #   - or the previous comment is 30 minutes or more old
+    name: Send a comment to suggest jobs to run
+    if: ${{ needs.get-jobs.outputs.jobs != '' }}
+    needs: [get-pr-number, get-jobs]
+    permissions:
+      pull-requests: write
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Check and update comment if needed
+        uses: actions/github-script@v7
+        env:
+          BODY: "\n\nrun-slow: ${{ needs.get-jobs.outputs.jobs }}"
+        with:
+          script: |
+            const prNumber = ${{ needs.get-pr-number.outputs.PR_NUMBER }};
+            const commentPrefix = "**[For maintainers]** Suggested jobs to run (before merge)";
+            const thirtyMinutesAgo = new Date(Date.now() - 30 * 60 * 1000); // 30 minutes ago
+            const newBody = `${commentPrefix}${process.env.BODY}`;
+            
+            // Get all comments on the PR
+            const { data: comments } = await github.rest.issues.listComments({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: prNumber
+            });
+            
+            // Find existing comments that start with our prefix
+            const existingComments = comments.filter(comment => 
+              comment.user.login === 'github-actions[bot]' && 
+              comment.body.startsWith(commentPrefix)
+            );
+            
+            let shouldCreateNewComment = true;
+            let commentsToDelete = [];
+            
+            if (existingComments.length > 0) {
+              // Get the most recent comment
+              const mostRecentComment = existingComments
+                .sort((a, b) => new Date(b.created_at) - new Date(a.created_at))[0];
+              
+              const commentDate = new Date(mostRecentComment.created_at);
+              const isOld = commentDate < thirtyMinutesAgo;
+              const isDifferentContent = mostRecentComment.body !== newBody;
+              
+              console.log(`Most recent comment created: ${mostRecentComment.created_at}`);
+              console.log(`Is older than 30 minutes: ${isOld}`);
+              console.log(`Has different content: ${isDifferentContent}`);
+              
+              if (isOld || isDifferentContent) {
+                // Delete all existing comments and create new one
+                commentsToDelete = existingComments;
+                console.log(`Will delete ${commentsToDelete.length} existing comment(s) and create new one`);
+              } else {
+                // Content is same and comment is recent, skip
+                shouldCreateNewComment = false;
+                console.log('Comment is recent and content unchanged, skipping update');
+              }
+            } else {
+              console.log('No existing comments found, will create new one');
+            }
+            
+            // Delete old comments if needed
+            for (const comment of commentsToDelete) {
+              console.log(`Deleting comment #${comment.id} (created: ${comment.created_at})`);
+              await github.rest.issues.deleteComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                comment_id: comment.id
+              });
+            }
+            
+            // Create new comment if needed
+            if (shouldCreateNewComment) {
+              await github.rest.issues.createComment({
+                owner: context.repo.owner,
+                repo: context.repo.repo,
+                issue_number: prNumber,
+                body: newBody
+              });
+              console.log('✅ New comment created');
+            } else {
+              console.log('ℹ️ No comment update needed');
+            }

.github/workflows/self-comment-ci.yml

@@ -29,7 +29,7 @@ jobs:
     runs-on: ubuntu-22.04
     name: Get PR number
     # For security: only allow team members to run
-    if: ${{ github.event.issue.state == 'open' && contains(fromJSON('["ydshieh", "ArthurZucker", "zucchini-nlp", "qubvel", "molbap", "gante", "LysandreJik", "Cyrilvallez", "Rocketknight1", "SunMarc", "muellerzr", "eustlb", "MekkCyber", "manueldeprada", "vasqu"]'), github.actor) && (startsWith(github.event.comment.body, 'run-slow') || startsWith(github.event.comment.body, 'run slow') || startsWith(github.event.comment.body, 'run_slow')) }}
+    if: ${{ github.event.issue.state == 'open' && contains(fromJSON('["ydshieh", "ArthurZucker", "zucchini-nlp", "qubvel", "molbap", "gante", "LysandreJik", "Cyrilvallez", "Rocketknight1", "SunMarc", "muellerzr", "eustlb", "MekkCyber", "manueldeprada", "vasqu", "ivarflakstad"]'), github.actor) && (startsWith(github.event.comment.body, 'run-slow') || startsWith(github.event.comment.body, 'run slow') || startsWith(github.event.comment.body, 'run_slow')) }}
     outputs:
       PR_NUMBER: ${{ steps.set_pr_number.outputs.PR_NUMBER }}
     steps:
@@ -185,7 +185,7 @@ jobs:
       fail-fast: false
       matrix:
         folders: ${{ fromJson(needs.get-tests.outputs.models) }}
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
        group: '${{ matrix.machine_type }}'
     container:
@@ -239,9 +239,9 @@ jobs:
         shell: bash
         run: |
           echo "${{ matrix.machine_type }}"
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -292,7 +292,7 @@ jobs:
       fail-fast: false
       matrix:
         folders: ${{ fromJson(needs.get-tests.outputs.quantizations) }}
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -338,9 +338,9 @@ jobs:
         shell: bash
         run: |
           echo "${{ matrix.machine_type }}"
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}

.github/workflows/self-push.yml

@@ -31,7 +31,7 @@ jobs:
     name: Setup
     strategy:
       matrix:
-        machine_type: [aws-g4dn-2xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -131,7 +131,7 @@ jobs:
       fail-fast: false
       matrix:
         folders: ${{ fromJson(needs.setup.outputs.matrix) }}
-        machine_type: [aws-g4dn-2xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -169,9 +169,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-2xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -244,7 +244,7 @@ jobs:
       fail-fast: false
       matrix:
         folders: ${{ fromJson(needs.setup.outputs.matrix) }}
-        machine_type: [aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -282,9 +282,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-2xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -357,7 +357,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-2xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -395,9 +395,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-2xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -467,7 +467,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -505,9 +505,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-2xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}

.github/workflows/self-scheduled-caller.yml

@@ -7,7 +7,7 @@ on:
     - cron: "17 2 * * *"
   push:
     branches:
-      - compare-test-results
+      - run_scheduled_ci*
   workflow_dispatch:
     inputs:
       prev_workflow_run_id:
@@ -22,9 +22,9 @@ on:
         default: ""
 
 
-# Used for `push` to easily modiffy the target workflow runs to compare against
+# Used for `push` to easily modify the target workflow runs to compare against
 env:
-    prev_workflow_run_id: "16064770151"
+    prev_workflow_run_id: ""
     other_workflow_run_id: ""
 
 
@@ -50,9 +50,64 @@ jobs:
     uses: ./.github/workflows/self-scheduled.yml
     with:
       job: run_models_gpu
-      slack_report_channel: "#transformers-ci-dummy"
-      runner: daily-ci
+      slack_report_channel: "#transformers-ci-daily-models"
       docker: huggingface/transformers-all-latest-gpu
       ci_event: Daily CI
       report_repo_id: hf-internal-testing/transformers_daily_ci
     secrets: inherit
+
+  torch-pipeline:
+    name: Torch pipeline CI
+    uses: ./.github/workflows/self-scheduled.yml
+    with:
+      job: run_pipelines_torch_gpu
+      slack_report_channel: "#transformers-ci-daily-pipeline-torch"
+      docker: huggingface/transformers-pytorch-gpu
+      ci_event: Daily CI
+      report_repo_id: hf-internal-testing/transformers_daily_ci
+    secrets: inherit
+
+  example-ci:
+    name: Example CI
+    uses: ./.github/workflows/self-scheduled.yml
+    with:
+      job: run_examples_gpu
+      slack_report_channel: "#transformers-ci-daily-examples"
+      docker: huggingface/transformers-all-latest-gpu
+      ci_event: Daily CI
+      report_repo_id: hf-internal-testing/transformers_daily_ci
+    secrets: inherit
+
+  trainer-fsdp-ci:
+    name: Trainer/FSDP CI
+    uses: ./.github/workflows/self-scheduled.yml
+    with:
+      job: run_trainer_and_fsdp_gpu
+      slack_report_channel: "#transformers-ci-daily-training"
+      docker: huggingface/transformers-all-latest-gpu
+      ci_event: Daily CI
+      report_repo_id: hf-internal-testing/transformers_daily_ci
+    secrets: inherit
+
+  deepspeed-ci:
+    name: DeepSpeed CI
+    uses: ./.github/workflows/self-scheduled.yml
+    with:
+      job: run_torch_cuda_extensions_gpu
+      slack_report_channel: "#transformers-ci-daily-training"
+      docker: huggingface/transformers-pytorch-deepspeed-latest-gpu
+      ci_event: Daily CI
+      working-directory-prefix: /workspace
+      report_repo_id: hf-internal-testing/transformers_daily_ci
+    secrets: inherit
+
+  quantization-ci:
+    name: Quantization CI
+    uses: ./.github/workflows/self-scheduled.yml
+    with:
+      job: run_quantization_torch_gpu
+      slack_report_channel: "#transformers-ci-daily-quantization"
+      docker: huggingface/transformers-quantization-latest-gpu
+      ci_event: Daily CI
+      report_repo_id: hf-internal-testing/transformers_daily_ci
+    secrets: inherit

.github/workflows/self-scheduled-intel-gaudi.yml

@@ -0,0 +1,342 @@
+name: Self-hosted runner (scheduled-intel-gaudi)
+
+on:
+  workflow_call:
+    inputs:
+      job:
+        required: true
+        type: string
+      slack_report_channel:
+        required: true
+        type: string
+      runner_scale_set:
+        required: true
+        type: string
+      ci_event:
+        required: true
+        type: string
+      report_repo_id:
+        required: true
+        type: string
+
+env:
+  NUM_SLICES: 2
+  RUN_SLOW: yes
+  PT_HPU_LAZY_MODE: 0
+  TRANSFORMERS_IS_CI: yes
+  PT_ENABLE_INT64_SUPPORT: 1
+  HF_HUB_READ_TOKEN: ${{ secrets.HF_HUB_READ_TOKEN }}
+  SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
+  HF_HOME: /mnt/cache/.cache/huggingface
+
+jobs:
+  setup:
+    if: contains(fromJSON('["run_models_gpu", "run_trainer_and_fsdp_gpu"]'), inputs.job)
+    name: Setup
+    runs-on: ubuntu-latest
+    outputs:
+      slice_ids: ${{ steps.set-matrix.outputs.slice_ids }}
+      folder_slices: ${{ steps.set-matrix.outputs.folder_slices }}
+      quantization_matrix: ${{ steps.set-matrix.outputs.quantization_matrix }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.10"
+
+      - id: set-matrix
+        if: contains(fromJSON('["run_models_gpu", "run_trainer_and_fsdp_gpu"]'), inputs.job)
+        name: Identify models to test
+        working-directory: tests
+        run: |
+          if [ "${{ inputs.job }}" = "run_models_gpu" ]; then
+            echo "folder_slices=$(python3 ../utils/split_model_tests.py --num_splits ${{ env.NUM_SLICES }})" >> $GITHUB_OUTPUT
+            echo "slice_ids=$(python3 -c 'd = list(range(${{ env.NUM_SLICES }})); print(d)')" >> $GITHUB_OUTPUT
+          elif [ "${{ inputs.job }}" = "run_trainer_and_fsdp_gpu" ]; then
+            echo "folder_slices=[['trainer'], ['fsdp']]" >> $GITHUB_OUTPUT
+            echo "slice_ids=[0, 1]" >> $GITHUB_OUTPUT
+          fi
+
+      - id: set-matrix-quantization
+        if: ${{ inputs.job == 'run_quantization_torch_gpu' }}
+        name: Identify quantization method to test
+        working-directory: tests
+        run: |
+          echo "quantization_matrix=$(python3 -c 'import os; tests = os.getcwd(); quantization_tests = os.listdir(os.path.join(tests, "quantization")); d = sorted(list(filter(os.path.isdir, [f"quantization/{x}" for x in quantization_tests]))) ;  print(d)')" >> $GITHUB_OUTPUT
+
+  run_models_gpu:
+    if: ${{ inputs.job == 'run_models_gpu' }}
+    name: " "
+    needs: setup
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [1gaudi, 2gaudi]
+        slice_id: ${{ fromJSON(needs.setup.outputs.slice_ids) }}
+    uses: ./.github/workflows/model_jobs_intel_gaudi.yml
+    with:
+      slice_id: ${{ matrix.slice_id }}
+      machine_type: ${{ matrix.machine_type }}
+      folder_slices: ${{ needs.setup.outputs.folder_slices }}
+      runner: ${{ inputs.runner_scale_set }}-${{ matrix.machine_type }}
+    secrets: inherit
+
+  run_trainer_and_fsdp_gpu:
+    if: ${{ inputs.job == 'run_trainer_and_fsdp_gpu' }}
+    name: " "
+    needs: setup
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [1gaudi, 2gaudi]
+        slice_id: ${{ fromJSON(needs.setup.outputs.slice_ids) }}
+    uses: ./.github/workflows/model_jobs_intel_gaudi.yml
+    with:
+      slice_id: ${{ matrix.slice_id }}
+      machine_type: ${{ matrix.machine_type }}
+      folder_slices: ${{ needs.setup.outputs.folder_slices }}
+      runner: ${{ inputs.runner_scale_set }}-${{ matrix.machine_type }}
+      report_name_prefix: run_trainer_and_fsdp_gpu
+    secrets: inherit
+
+  run_pipelines_torch_gpu:
+    if: ${{ inputs.job == 'run_pipelines_torch_gpu' }}
+    name: Pipelines
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [1gaudi, 2gaudi]
+    runs-on:
+      group: ${{ inputs.runner_scale_set }}-${{ matrix.machine_type }}
+    container:
+      image: vault.habana.ai/gaudi-docker/1.21.1/ubuntu22.04/habanalabs/pytorch-installer-2.6.0:latest
+      options: --runtime=habana
+        -v /mnt/cache/.cache/huggingface:/mnt/cache/.cache/huggingface
+        --env OMPI_MCA_btl_vader_single_copy_mechanism=none
+        --env HABANA_VISIBLE_DEVICES
+        --env HABANA_VISIBLE_MODULES
+        --cap-add=sys_nice
+        --shm-size=64G
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Install dependencies
+        run: |
+          pip install -e .[testing,torch] "numpy<2.0.0" scipy scikit-learn librosa soundfile
+
+      - name: HL-SMI
+        run: |
+          hl-smi
+          echo "HABANA_VISIBLE_DEVICES=${HABANA_VISIBLE_DEVICES}"
+          echo "HABANA_VISIBLE_MODULES=${HABANA_VISIBLE_MODULES}"
+
+      - name: Environment
+        run: python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        run: pip freeze
+
+      - name: Set `machine_type` for report and artifact names
+        shell: bash
+        run: |
+          if [ "${{ matrix.machine_type }}" = "1gaudi" ]; then
+            machine_type=single-gpu
+          elif [ "${{ matrix.machine_type }}" = "2gaudi" ]; then
+            machine_type=multi-gpu
+          else
+            machine_type=${{ matrix.machine_type }}
+          fi
+          echo "machine_type=$machine_type" >> $GITHUB_ENV
+
+      - name: Run all pipeline tests on Intel Gaudi
+        run: |
+          python3 -m pytest -v --make-reports=${{ env.machine_type }}_run_pipelines_torch_gpu_test_reports tests/pipelines -m "not not_device_test"
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: |
+          cat reports/${{ env.machine_type }}_run_pipelines_torch_gpu_test_reports/failures_short.txt
+
+      - name: "Test suite reports artifacts: ${{ env.machine_type }}_run_pipelines_torch_gpu_test_reports"
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: ${{ env.machine_type }}_run_pipelines_torch_gpu_test_reports
+          path: reports/${{ env.machine_type }}_run_pipelines_torch_gpu_test_reports
+
+  run_examples_gpu:
+    if: ${{ inputs.job == 'run_examples_gpu' }}
+    name: Examples directory
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [1gaudi]
+    runs-on:
+      group: ${{ inputs.runner_scale_set }}-${{ matrix.machine_type }}
+    container:
+      image: vault.habana.ai/gaudi-docker/1.21.1/ubuntu22.04/habanalabs/pytorch-installer-2.6.0:latest
+      options: --runtime=habana
+        -v /mnt/cache/.cache/huggingface:/mnt/cache/.cache/huggingface
+        --env OMPI_MCA_btl_vader_single_copy_mechanism=none
+        --env HABANA_VISIBLE_DEVICES
+        --env HABANA_VISIBLE_MODULES
+        --cap-add=sys_nice
+        --shm-size=64G
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Install dependencies
+        run: |
+          pip install -e .[testing,torch] "numpy<2.0.0" scipy scikit-learn librosa soundfile
+
+      - name: HL-SMI
+        run: |
+          hl-smi
+          echo "HABANA_VISIBLE_DEVICES=${HABANA_VISIBLE_DEVICES}"
+          echo "HABANA_VISIBLE_MODULES=${HABANA_VISIBLE_MODULES}"
+
+      - name: Environment
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        run: |
+          pip freeze
+
+      - name: Set `machine_type` for report and artifact names
+        shell: bash
+        run: |
+          if [ "${{ matrix.machine_type }}" = "1gaudi" ]; then
+            machine_type=single-gpu
+          elif [ "${{ matrix.machine_type }}" = "2gaudi" ]; then
+            machine_type=multi-gpu
+          else
+            machine_type=${{ matrix.machine_type }}
+          fi
+          echo "machine_type=$machine_type" >> $GITHUB_ENV
+
+      - name: Run examples tests on Intel Gaudi
+        run: |
+          pip install -r examples/pytorch/_tests_requirements.txt
+          python3 -m pytest -v --make-reports=${{ env.machine_type }}_run_examples_gpu_test_reports examples/pytorch -m "not not_device_test"
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: |
+          cat reports/${{ env.machine_type }}_run_examples_gpu_test_reports/failures_short.txt
+
+      - name: "Test suite reports artifacts: ${{ env.machine_type }}_run_examples_gpu_test_reports"
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: ${{ env.machine_type }}_run_examples_gpu_test_reports
+          path: reports/${{ env.machine_type }}_run_examples_gpu_test_reports
+
+  run_torch_cuda_extensions_gpu:
+    if: ${{ inputs.job == 'run_torch_cuda_extensions_gpu' }}
+    name: Intel Gaudi deepspeed tests
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [1gaudi, 2gaudi]
+    runs-on:
+      group: ${{ inputs.runner_scale_set }}-${{ matrix.machine_type }}
+    container:
+      image: vault.habana.ai/gaudi-docker/1.21.1/ubuntu22.04/habanalabs/pytorch-installer-2.6.0:latest
+      options: --runtime=habana
+        -v /mnt/cache/.cache/huggingface:/mnt/cache/.cache/huggingface
+        --env OMPI_MCA_btl_vader_single_copy_mechanism=none
+        --env HABANA_VISIBLE_DEVICES
+        --env HABANA_VISIBLE_MODULES
+        --cap-add=sys_nice
+        --shm-size=64G
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+
+      - name: Install dependencies
+        run: |
+          pip install -e .[testing,torch] "numpy<2.0.0" scipy scikit-learn librosa soundfile
+          pip install git+https://github.com/HabanaAI/DeepSpeed.git@1.20.0
+
+      - name: HL-SMI
+        run: |
+          hl-smi
+          echo "HABANA_VISIBLE_DEVICES=${HABANA_VISIBLE_DEVICES}"
+          echo "HABANA_VISIBLE_MODULES=${HABANA_VISIBLE_MODULES}"
+
+      - name: Environment
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        run: |
+          pip freeze
+
+      - name: Set `machine_type` for report and artifact names
+        shell: bash
+        run: |
+          if [ "${{ matrix.machine_type }}" = "1gaudi" ]; then
+            machine_type=single-gpu
+          elif [ "${{ matrix.machine_type }}" = "2gaudi" ]; then
+            machine_type=multi-gpu
+          else
+            machine_type=${{ matrix.machine_type }}
+          fi
+          echo "machine_type=$machine_type" >> $GITHUB_ENV
+
+      - name: Run all deepspeed tests on intel Gaudi
+        run: |
+          python3 -m pytest -v --make-reports=${{ env.machine_type }}_run_torch_cuda_extensions_gpu_test_reports tests/deepspeed -m "not not_device_test"
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: |
+          cat reports/${{ env.machine_type }}_run_torch_cuda_extensions_gpu_test_reports/failures_short.txt
+
+      - name: "Test suite reports artifacts: ${{ env.machine_type }}_run_torch_cuda_extensions_gpu_test_reports"
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v4
+        with:
+          name: ${{ env.machine_type }}_run_torch_cuda_extensions_gpu_test_reports
+          path: reports/${{ env.machine_type }}_run_torch_cuda_extensions_gpu_test_reports
+
+  send_results:
+    name: Slack Report
+    needs:
+      [
+        setup,
+        run_models_gpu,
+        run_examples_gpu,
+        run_torch_cuda_extensions_gpu,
+        run_pipelines_torch_gpu,
+        run_trainer_and_fsdp_gpu,
+      ]
+    if: ${{ always() }}
+    uses: ./.github/workflows/slack-report.yml
+    with:
+      job: ${{ inputs.job }}
+      setup_status: ${{ needs.setup.result }}
+      slack_report_channel: ${{ inputs.slack_report_channel }}
+      quantization_matrix: ${{ needs.setup.outputs.quantization_matrix }}
+      folder_slices: ${{ needs.setup.outputs.folder_slices }}
+      report_repo_id: ${{ inputs.report_repo_id }}
+      ci_event: ${{ inputs.ci_event }}
+
+    secrets: inherit

.github/workflows/self-scheduled-intel-gaudi3-caller.yml

@@ -0,0 +1,67 @@
+name: Self-hosted runner (Intel Gaudi3 scheduled CI caller)
+
+on:
+  repository_dispatch:
+  workflow_dispatch:
+  schedule:
+    - cron: "17 2 * * *"
+
+jobs:
+  model-ci:
+    name: Model CI
+    uses: ./.github/workflows/self-scheduled-intel-gaudi.yml
+    with:
+      job: run_models_gpu
+      ci_event: Scheduled CI (Intel) - Gaudi3
+      runner_scale_set: itac-bm-emr-gaudi3-dell
+      slack_report_channel: "#transformers-ci-daily-intel-gaudi3"
+      report_repo_id: optimum-intel/transformers_daily_ci_intel_gaudi3
+
+    secrets: inherit
+
+  pipeline-ci:
+    name: Pipeline CI
+    uses: ./.github/workflows/self-scheduled-intel-gaudi.yml
+    with:
+      job: run_pipelines_torch_gpu
+      ci_event: Scheduled CI (Intel) - Gaudi3
+      runner_scale_set: itac-bm-emr-gaudi3-dell
+      slack_report_channel: "#transformers-ci-daily-intel-gaudi3"
+      report_repo_id: optimum-intel/transformers_daily_ci_intel_gaudi3
+
+    secrets: inherit
+
+  example-ci:
+    name: Example CI
+    uses: ./.github/workflows/self-scheduled-intel-gaudi.yml
+    with:
+      job: run_examples_gpu
+      ci_event: Scheduled CI (Intel) - Gaudi3
+      runner_scale_set: itac-bm-emr-gaudi3-dell
+      slack_report_channel: "#transformers-ci-daily-intel-gaudi3"
+      report_repo_id: optimum-intel/transformers_daily_ci_intel_gaudi3
+
+    secrets: inherit
+
+  deepspeed-ci:
+    name: DeepSpeed CI
+    uses: ./.github/workflows/self-scheduled-intel-gaudi.yml
+    with:
+      job: run_torch_cuda_extensions_gpu
+      ci_event: Scheduled CI (Intel) - Gaudi3
+      runner_scale_set: itac-bm-emr-gaudi3-dell
+      slack_report_channel: "#transformers-ci-daily-intel-gaudi3"
+      report_repo_id: optimum-intel/transformers_daily_ci_intel_gaudi3
+
+    secrets: inherit
+
+  trainer-fsdp-ci:
+    name: Trainer/FSDP CI
+    uses: ./.github/workflows/self-scheduled-intel-gaudi.yml
+    with:
+      job: run_trainer_and_fsdp_gpu
+      ci_event: Scheduled CI (Intel) - Gaudi3
+      runner_scale_set: itac-bm-emr-gaudi3-dell
+      slack_report_channel: "#transformers-ci-daily-intel-gaudi3"
+      report_repo_id: optimum-intel/transformers_daily_ci_intel_gaudi3
+    secrets: inherit

.github/workflows/self-scheduled.yml

@@ -15,9 +15,6 @@ on:
       slack_report_channel:
         required: true
         type: string
-      runner:
-        required: true
-        type: string
       docker:
         required: true
         type: string
@@ -53,7 +50,7 @@ jobs:
     name: Setup
     strategy:
       matrix:
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -62,6 +59,7 @@ jobs:
     outputs:
       folder_slices: ${{ steps.set-matrix.outputs.folder_slices }}
       slice_ids: ${{ steps.set-matrix.outputs.slice_ids }}
+      runner_map: ${{ steps.set-matrix.outputs.runner_map }}
       quantization_matrix: ${{ steps.set-matrix-quantization.outputs.quantization_matrix }}
     steps:
       - name: Update clone
@@ -88,6 +86,7 @@ jobs:
           if [ "${{ inputs.job }}" = "run_models_gpu" ]; then
             echo "folder_slices=$(python3 ../utils/split_model_tests.py --num_splits ${{ env.NUM_SLICES }})" >> $GITHUB_OUTPUT
             echo "slice_ids=$(python3 -c 'd = list(range(${{ env.NUM_SLICES }})); print(d)')" >> $GITHUB_OUTPUT
+            echo "runner_map=$(python3 ../utils/get_runner_map.py)" >> $GITHUB_OUTPUT
           elif [ "${{ inputs.job }}" = "run_trainer_and_fsdp_gpu" ]; then
             echo "folder_slices=[['trainer'], ['fsdp']]" >> $GITHUB_OUTPUT
             echo "slice_ids=[0, 1]" >> $GITHUB_OUTPUT
@@ -111,14 +110,14 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-4xlarge-cache]
+        machine_type: [single-gpu, multi-gpu]
         slice_id: ${{ fromJSON(needs.setup.outputs.slice_ids) }}
     uses: ./.github/workflows/model_jobs.yml
     with:
       folder_slices: ${{ needs.setup.outputs.folder_slices }}
       machine_type: ${{ matrix.machine_type }}
       slice_id: ${{ matrix.slice_id }}
-      runner: ${{ inputs.runner }}
+      runner_map: ${{ needs.setup.outputs.runner_map }}
       docker: ${{ inputs.docker }}
     secrets: inherit
 
@@ -129,14 +128,14 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
         slice_id: [0, 1]
     uses: ./.github/workflows/model_jobs.yml
     with:
       folder_slices: ${{ needs.setup.outputs.folder_slices }}
       machine_type: ${{ matrix.machine_type }}
       slice_id: ${{ matrix.slice_id }}
-      runner: ${{ inputs.runner }}
+      runner_map: ${{ needs.setup.outputs.runner_map }}
       docker: ${{ inputs.docker }}
       report_name_prefix: run_trainer_and_fsdp_gpu
     secrets: inherit
@@ -147,7 +146,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -181,9 +180,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -215,7 +214,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-4xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -249,9 +248,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -284,7 +283,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -346,9 +345,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}
@@ -383,7 +382,7 @@ jobs:
       fail-fast: false
       matrix:
         folders: ${{ fromJson(needs.setup.outputs.quantization_matrix) }}
-        machine_type: [aws-g4dn-4xlarge-cache, aws-g4dn-12xlarge-cache]
+        machine_type: [aws-g5-4xlarge-cache, aws-g5-12xlarge-cache]
     runs-on:
       group: '${{ matrix.machine_type }}'
     container:
@@ -426,9 +425,9 @@ jobs:
         run: |
           echo "${{ matrix.machine_type }}"
 
-          if [ "${{ matrix.machine_type }}" = "aws-g4dn-4xlarge-cache" ]; then
+          if [ "${{ matrix.machine_type }}" = "aws-g5-4xlarge-cache" ]; then
             machine_type=single-gpu
-          elif [ "${{ matrix.machine_type }}" = "aws-g4dn-12xlarge-cache" ]; then
+          elif [ "${{ matrix.machine_type }}" = "aws-g5-12xlarge-cache" ]; then
             machine_type=multi-gpu
           else
             machine_type=${{ matrix.machine_type }}

Makefile

@@ -8,13 +8,19 @@ check_dirs := examples tests src utils
 exclude_folders :=  ""
 
 modified_only_fixup:
-	$(eval modified_py_files := $(shell python utils/get_modified_files.py $(check_dirs)))
-	@if test -n "$(modified_py_files)"; then \
-		echo "Checking/fixing $(modified_py_files)"; \
-		ruff check $(modified_py_files) --fix --exclude $(exclude_folders); \
-		ruff format $(modified_py_files) --exclude $(exclude_folders);\
+	@current_branch=$$(git branch --show-current); \
+	if [ "$$current_branch" = "main" ]; then \
+		echo "On main branch, running 'style' target instead..."; \
+		$(MAKE) style; \
 	else \
-		echo "No library .py files were modified"; \
+		modified_py_files=$$(python utils/get_modified_files.py $(check_dirs)); \
+		if [ -n "$$modified_py_files" ]; then \
+			echo "Checking/fixing files: $${modified_py_files}"; \
+			ruff check $${modified_py_files} --fix --exclude $(exclude_folders); \
+			ruff format $${modified_py_files} --exclude $(exclude_folders); \
+		else \
+			echo "No library .py files were modified"; \
+		fi; \
 	fi
 
 # Update src/transformers/dependency_versions_table.py

awesome-transformers.md

@@ -288,7 +288,7 @@ Keywords: Music understanding, Music generation
 
 ## [dalle-flow](https://github.com/jina-ai/dalle-flow)
 
-DALL·E Flow is an interactive workflow for generating high-definition images from a text prompt. Itt leverages DALL·E-Mega, GLID-3 XL, and Stable Diffusion to generate image candidates, and then calls CLIP-as-service to rank the candidates w.r.t. the prompt.
+DALL·E Flow is an interactive workflow for generating high-definition images from a text prompt. It leverages DALL·E-Mega, GLID-3 XL, and Stable Diffusion to generate image candidates, and then calls CLIP-as-service to rank the candidates w.r.t. the prompt.
 The preferred candidate is fed to GLID-3 XL for diffusion, which often enriches the texture and background. Finally, the candidate is upscaled to 1024x1024 via SwinIR.
 
 Keywords: High-definition image generation, Stable Diffusion, DALL-E Mega, GLID-3 XL, CLIP, SwinIR
@@ -526,7 +526,7 @@ Keywords: Model deployment, CLoud, Mobile, Edge
 
 ## [underthesea](https://github.com/undertheseanlp/underthesea)
 
-[underthesea](https://github.com/undertheseanlp/underthesea) is a Vietnamese NLP toolkit. Underthesea is a suite of open source Python modules data sets and tutorials supporting research and development in Vietnamese Natural Language Processing. We provides extremely easy API to quickly apply pretrained NLP models to your Vietnamese text, such as word segmentation, part-of-speech tagging (PoS), named entity recognition (NER), text classification and dependency parsing.
+[underthesea](https://github.com/undertheseanlp/underthesea) is a Vietnamese NLP toolkit. Underthesea is a suite of open source Python modules data sets and tutorials supporting research and development in Vietnamese Natural Language Processing. We provide extremely easy API to quickly apply pretrained NLP models to your Vietnamese text, such as word segmentation, part-of-speech tagging (PoS), named entity recognition (NER), text classification and dependency parsing.
 
 Keywords: Vietnamese, NLP
 

benchmark/benchmarks_entrypoint.py

@@ -28,7 +28,7 @@ class MetricsRecorder:
         self.commit_id = commit_id
         self.commit_msg = commit_msg
 
-    def initialise_benchmark(self, metadata: Dict[str, str]) -> int:
+    def initialise_benchmark(self, metadata: dict[str, str]) -> int:
         """
         Creates a new benchmark, returns the benchmark id
         """
@@ -55,7 +55,7 @@ class MetricsRecorder:
             f"inserted device measurements for benchmark #{benchmark_id} [CPU util: {cpu_util}, mem MBs: {mem_megabytes}, GPU util: {gpu_util}, GPU mem MBs: {gpu_mem_megabytes}]"
         )
 
-    def collect_model_measurements(self, benchmark_id: int, measurements: Dict[str, float]):
+    def collect_model_measurements(self, benchmark_id: int, measurements: dict[str, float]):
         with self.conn.cursor() as cur:
             cur.execute(
                 """
@@ -85,7 +85,7 @@ handler.setFormatter(formatter)
 logger.addHandler(handler)
 
 
-def parse_arguments() -> Tuple[str, str, str, str]:
+def parse_arguments() -> tuple[str, str, str, str]:
     """
     Parse command line arguments for the benchmarking CLI.
     """

docker/examples-torch.dockerfile

@@ -2,10 +2,10 @@ FROM python:3.9-slim
 ENV PYTHONDONTWRITEBYTECODE=1
 ARG REF=main
 USER root
-RUN apt-get update &&  apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git g++ cmake pkg-config openssh-client git
+RUN apt-get update &&  apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git g++ cmake pkg-config openssh-client git ffmpeg
 ENV UV_PYTHON=/usr/local/bin/python
 RUN pip --no-cache-dir install uv && uv venv && uv pip install --no-cache-dir -U pip setuptools
-RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' --index-url https://download.pytorch.org/whl/cpu
+RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' 'torchcodec' --index-url https://download.pytorch.org/whl/cpu
 RUN uv pip install --no-deps timm accelerate --extra-index-url https://download.pytorch.org/whl/cpu
 RUN uv pip install --no-cache-dir librosa "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[sklearn,sentencepiece,vision,testing]" seqeval albumentations jiwer
 RUN uv pip uninstall transformers

docker/pipeline-torch.dockerfile

@@ -2,10 +2,10 @@ FROM python:3.9-slim
 ENV PYTHONDONTWRITEBYTECODE=1
 ARG REF=main
 USER root
-RUN apt-get update &&  apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git pkg-config openssh-client git
+RUN apt-get update &&  apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git pkg-config openssh-client git ffmpeg
 ENV UV_PYTHON=/usr/local/bin/python
 RUN pip --no-cache-dir install uv && uv venv && uv pip install --no-cache-dir -U pip setuptools
-RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' --index-url https://download.pytorch.org/whl/cpu
+RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' 'torchcodec' --index-url https://download.pytorch.org/whl/cpu
 RUN uv pip install --no-deps timm accelerate --extra-index-url https://download.pytorch.org/whl/cpu
 RUN uv pip install --no-cache-dir librosa "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[sklearn,sentencepiece,vision,testing]"
 RUN uv pip uninstall transformers

docker/torch-light.dockerfile

@@ -2,10 +2,10 @@ FROM python:3.9-slim
 ENV PYTHONDONTWRITEBYTECODE=1
 ARG REF=main
 USER root
-RUN apt-get update &&  apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git g++ cmake pkg-config openssh-client git git-lfs
+RUN apt-get update &&  apt-get install -y --no-install-recommends libsndfile1-dev espeak-ng time git g++ cmake pkg-config openssh-client git git-lfs ffmpeg
 ENV UV_PYTHON=/usr/local/bin/python
 RUN pip --no-cache-dir install uv && uv venv && uv pip install --no-cache-dir -U pip setuptools
-RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' --index-url https://download.pytorch.org/whl/cpu
+RUN uv pip install --no-cache-dir 'torch' 'torchaudio' 'torchvision' 'torchcodec' --index-url https://download.pytorch.org/whl/cpu
 RUN uv pip install --no-deps timm accelerate --extra-index-url https://download.pytorch.org/whl/cpu
 RUN uv pip install --no-cache-dir librosa "git+https://github.com/huggingface/transformers.git@${REF}#egg=transformers[sklearn,sentencepiece,vision,testing,tiktoken,num2words,video]"
 RUN uv pip uninstall transformers

docker/transformers-all-latest-gpu/Dockerfile

@@ -26,7 +26,7 @@ RUN git clone https://github.com/huggingface/transformers && cd transformers &&
 # 1. Put several commands in a single `RUN` to avoid image/layer exporting issue. Could be revised in the future.
 # 2. Regarding `torch` part, We might need to specify proper versions for `torchvision` and `torchaudio`.
 #    Currently, let's not bother to specify their versions explicitly (so installed with their latest release versions).
-RUN python3 -m pip install --no-cache-dir -e ./transformers[dev,onnxruntime] && [ ${#PYTORCH} -gt 0 -a "$PYTORCH" != "pre" ] && VERSION='torch=='$PYTORCH'.*' ||  VERSION='torch'; echo "export VERSION='$VERSION'" >> ~/.profile && echo torch=$VERSION && [ "$PYTORCH" != "pre" ] && python3 -m pip install --no-cache-dir -U $VERSION torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/$CUDA || python3 -m pip install --no-cache-dir -U --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/$CUDA && python3 -m pip uninstall -y tensorflow tensorflow_text tensorflow_probability
+RUN python3 -m pip install --no-cache-dir -e ./transformers[dev,onnxruntime] && [ ${#PYTORCH} -gt 0 -a "$PYTORCH" != "pre" ] && VERSION='torch=='$PYTORCH'.*' ||  VERSION='torch'; echo "export VERSION='$VERSION'" >> ~/.profile && echo torch=$VERSION && [ "$PYTORCH" != "pre" ] && python3 -m pip install --no-cache-dir -U $VERSION torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/$CUDA || python3 -m pip install --no-cache-dir -U --pre torch torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/nightly/$CUDA && python3 -m pip uninstall -y tensorflow tensorflow_text tensorflow_probability
 
 RUN python3 -m pip uninstall -y flax jax
 

docker/transformers-pytorch-amd-gpu/Dockerfile

@@ -3,6 +3,9 @@ LABEL maintainer="Hugging Face"
 
 ARG DEBIAN_FRONTEND=noninteractive
 
+ARG TORCH_VISION='0.21.0'
+ARG TORCH_AUDIO='2.6.0'
+
 RUN apt update && \
     apt install -y --no-install-recommends git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-dev python3-pip python3-dev ffmpeg git-lfs && \
     apt clean && \
@@ -20,6 +23,7 @@ WORKDIR /
 ADD https://api.github.com/repos/huggingface/transformers/git/refs/heads/main version.json
 RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF
 
+RUN python3 -m pip install --no-cache-dir torchvision==$TORCH_VISION torchaudio==$TORCH_AUDIO
 RUN python3 -m pip install --no-cache-dir -e ./transformers[dev-torch,testing,video]
 
 RUN python3 -m pip uninstall -y tensorflow flax

docker/transformers-pytorch-deepspeed-latest-gpu/Dockerfile

@@ -21,7 +21,7 @@ RUN python3 -m pip install --no-cache-dir './transformers[deepspeed-testing]' 'p
 # Install latest release PyTorch
 # (PyTorch must be installed before pre-compiling any DeepSpeed c++/cuda ops.)
 # (https://www.deepspeed.ai/tutorials/advanced-install/#pre-install-deepspeed-ops)
-RUN python3 -m pip uninstall -y torch torchvision torchaudio && python3 -m pip install --no-cache-dir -U torch==$PYTORCH torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/$CUDA
+RUN python3 -m pip uninstall -y torch torchvision torchaudio && python3 -m pip install --no-cache-dir -U torch==$PYTORCH torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/$CUDA
 
 RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
 

docker/transformers-pytorch-deepspeed-nightly-gpu/Dockerfile

@@ -19,7 +19,7 @@ RUN python3 -m pip uninstall -y torch torchvision torchaudio
 # Install **nightly** release PyTorch (flag `--pre`)
 # (PyTorch must be installed before pre-compiling any DeepSpeed c++/cuda ops.)
 # (https://www.deepspeed.ai/tutorials/advanced-install/#pre-install-deepspeed-ops)
-RUN python3 -m pip install --no-cache-dir -U --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/$CUDA
+RUN python3 -m pip install --no-cache-dir -U --pre torch torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/nightly/$CUDA
 
 # `datasets` requires pandas, pandas has some modules compiled with numpy=1.x causing errors
 RUN python3 -m pip install --no-cache-dir './transformers[deepspeed-testing]' 'pandas<2' 'numpy<2'

docker/transformers-pytorch-xpu/Dockerfile

@@ -0,0 +1,93 @@
+FROM intel/deep-learning-essentials:2025.1.3-0-devel-ubuntu22.04 AS base
+LABEL maintainer="Hugging Face"
+
+SHELL ["/bin/bash", "-c"]
+
+ARG PYTHON_VER=3.11
+ENV TORCH_DEVICE_BACKEND_AUTOLOAD=0
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get remove -y python3.10 && apt-get autoremove -y
+RUN apt-get update && \
+    apt-get install -y software-properties-common && \
+    add-apt-repository -y ppa:deadsnakes/ppa && \
+    apt-get update && \
+    apt-get install -y python$PYTHON_VER python$PYTHON_VER-dev python3-pip && \
+    ln -sf /usr/bin/python$PYTHON_VER /usr/bin/python3 && \
+    ln -sf /usr/bin/python3 /usr/bin/python && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
+RUN apt-get update && \
+    apt-get -y install \
+        apt-utils \
+        build-essential \
+        ca-certificates \
+        clinfo \
+        curl \
+        git \
+        git-lfs \
+        vim \
+        numactl \
+        gnupg2 \
+        gpg-agent \
+        zlib1g-dev \
+        rsync \
+        sudo \
+        libnl-genl-3-200 \
+        xpu-smi \
+        unzip \
+        ffmpeg \
+        tesseract-ocr \
+        espeak-ng \
+        wget \
+        ncurses-term && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
+
+RUN apt-get update && \
+    apt-get install -y \
+        linux-headers-$(uname -r) \
+        linux-modules-extra-$(uname -r) \
+        flex bison \
+        intel-fw-gpu intel-i915-dkms xpu-smi \
+        intel-opencl-icd libze-intel-gpu1 libze1 \
+        intel-media-va-driver-non-free libmfx-gen1 libvpl2 \
+        libegl-mesa0 libegl1-mesa libegl1-mesa-dev libgbm1 libgl1-mesa-dev libgl1-mesa-dri \
+        libglapi-mesa libglx-mesa0 libigdgmm12 libxatracker2 mesa-va-drivers \
+        mesa-vdpau-drivers mesa-vulkan-drivers va-driver-all vainfo hwinfo clinfo intel-ocloc \
+        libigc-dev intel-igc-cm libigdfcl-dev libigfxcmrt-dev libze-dev && \
+    apt-get clean && \
+    rm -rf  /var/lib/apt/lists/*
+
+RUN pip install --upgrade pip
+RUN pip install triton==3.3.0
+
+RUN pip install torch==2.7.0 torchvision==0.22.0 torchaudio==2.7.0 --index-url https://download.pytorch.org/whl/xpu --no-cache-dir
+
+RUN pip install evaluate torchdata pyctcdecode pytesseract decord galore-torch fire scipy scikit-learn sentencepiece sacremoses nltk rouge_score librosa soundfile g2p_en mpi4py requests_mock
+RUN pip install pretty_midi essentia resampy Levenshtein av sacrebleu phonemizer invisible_watermark schedulefree
+RUN pip install gguf hqq compressed_tensors gptqmodel mergekit autoawq deepspeed torchao onnx
+RUN pip install hf_transfer huggingface-hub hf-doc-builder datasets optimum-quanto timm transformers accelerate optimum peft
+
+RUN pip install git+https://github.com/linkedin/Liger-Kernel.git --extra-index-url https://download.pytorch.org/whl/test/xpu
+
+# install bitsandbytes
+RUN pip install git+https://github.com/bitsandbytes-foundation/bitsandbytes.git
+
+ENV OCL_ICD_VENDORS=/etc/OpenCL/vendors
+ENV FI_PROVIDER_PATH=${I_MPI_ROOT}/lib/libfabric/prov:/usr/lib/x86_64-linux-gnu/libfabric
+ENV CCL_ROOT=/usr/local
+ENV CCL_ATL_TRANSPORT=ofi
+ENV I_MPI_ROOT=/usr/local
+ENV CLASSPATH=${I_MPI_ROOT}/lib/mpi.jar
+ENV PATH=${I_MPI_ROOT}/bin/libfabric:${PATH}
+ENV LD_LIBRARY_PATH=${I_MPI_ROOT}/lib/libfabric:${LD_LIBRARY_PATH}
+
+RUN touch /entrypoint.sh
+RUN chmod +x /entrypoint.sh
+RUN echo "#!/bin/bash" >> /entrypoint.sh
+RUN echo "source /opt/intel/oneapi/setvars.sh --force && /bin/bash" >> /entrypoint.sh
+
+ENTRYPOINT ["/entrypoint.sh"]

docker/transformers-quantization-latest-gpu/Dockerfile

@@ -26,7 +26,7 @@ RUN [ ${#PYTORCH} -gt 0 ] && VERSION='torch=='$PYTORCH'.*' ||  VERSION='torch';
 RUN echo torch=$VERSION
 # `torchvision` and `torchaudio` should be installed along with `torch`, especially for nightly build.
 # Currently, let's just use their latest releases (when `torch` is installed with a release version)
-RUN python3 -m pip install --no-cache-dir -U $VERSION torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/$CUDA
+RUN python3 -m pip install --no-cache-dir -U $VERSION torchvision torchaudio torchcodec --extra-index-url https://download.pytorch.org/whl/$CUDA
 
 RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/accelerate@main#egg=accelerate
 
@@ -93,6 +93,9 @@ RUN python3 -m pip install --no-cache-dir -e ./transformers[dev-torch]
 # `kernels` may give different outputs (within 1e-5 range) even with the same model (weights) and the same inputs
 RUN python3 -m pip uninstall -y kernels
 
+# Uninstall flash-attn installed by autoawq, it causes issues here : https://github.com/huggingface/transformers/actions/runs/15915442841/job/44892146131
+RUN python3 -m pip uninstall -y flash-attn
+
 # When installing in editable mode, `transformers` is not recognized as a package.
 # this line must be added in order for python to be aware of transformers.
 RUN cd transformers && python3 setup.py develop

docs/README.md

@@ -278,7 +278,7 @@ Here's an example of a single value return:
 
 ```python
     Returns:
-        `List[int]`: A list of integers in the range [0, 1] --- 1 for a special token, 0 for a sequence token.
+        `list[int]`: A list of integers in the range [0, 1] --- 1 for a special token, 0 for a sequence token.
 ```
 
 Here's an example of a tuple return, comprising several objects:

docs/source/ar/custom_models.md

@@ -30,7 +30,7 @@ class ResnetConfig(PretrainedConfig):
     def __init__(
         self,
         block_type="bottleneck",
-        layers: List[int] = [3, 4, 6, 3],
+        layers: list[int] = [3, 4, 6, 3],
         num_classes: int = 1000,
         input_channels: int = 3,
         cardinality: int = 1,

docs/source/de/testing.md

@@ -473,13 +473,6 @@ Hier ist zum Beispiel ein Test, der nur ausgeführt werden muss, wenn 2 oder meh
 def test_example_with_multi_gpu():
 ```
 
-Wenn ein Test `tensorflow` benötigt, verwenden Sie den Dekorator `require_tf`. Zum Beispiel:
-
-```python no-style
-@require_tf
-def test_tf_thing_with_tensorflow():
-```
-
 Diese Dekors können gestapelt werden. Wenn zum Beispiel ein Test langsam ist und mindestens eine GPU unter pytorch benötigt, können Sie
 wie Sie ihn einrichten können:
 
@@ -1204,9 +1197,6 @@ if torch.cuda.is_available():
 import numpy as np
 
 np.random.seed(seed)
-
-# tf RNG
-tf.random.set_seed(seed)
 ```
 
 ### Tests debuggen

docs/source/en/_toctree.yml

@@ -17,12 +17,12 @@
       title: Customizing model components
     - local: model_sharing
       title: Sharing
-    - local: add_new_model
-      title: Adding a new model to Transformers
     - local: modular_transformers
-      title: Modular Transformers
+      title: Contributing a new model to Transformers
+    - local: add_new_model
+      title: Legacy model contribution
     - local: auto_docstring
-      title: Document your models
+      title: Documenting a model
     - local: attention_interface
       title: Customizing attention function
     title: Models
@@ -97,11 +97,9 @@
     - local: perf_infer_gpu_one
       title: GPU
     - local: perf_infer_gpu_multi
-      title: Distributed GPU inference
+      title: Distributed inference
     - local: perf_infer_cpu
       title: CPU
-    - local: tf_xla
-      title: XLA
     title: Optimization
   - local: agents
     title: Agents
@@ -141,8 +139,6 @@
       title: GPU
     - local: perf_train_cpu
       title: CPU
-    - local: perf_train_tpu_tf
-      title: TPU
     - local: perf_train_special
       title: Apple Silicon
     - local: perf_train_gaudi
@@ -363,6 +359,8 @@
     - sections:
       - local: model_doc/albert
         title: ALBERT
+      - local: model_doc/arcee
+        title: Arcee
       - local: model_doc/bamba
         title: Bamba
       - local: model_doc/bart
@@ -431,6 +429,10 @@
         title: DiffLlama
       - local: model_doc/distilbert
         title: DistilBERT
+      - local: model_doc/doge
+        title: Doge
+      - local: model_doc/dots1
+        title: dots1
       - local: model_doc/dpr
         title: DPR
       - local: model_doc/electra
@@ -653,6 +655,8 @@
         title: SwitchTransformers
       - local: model_doc/t5
         title: T5
+      - local: model_doc/t5gemma
+        title: T5Gemma
       - local: model_doc/t5v1.1
         title: T5v1.1
       - local: model_doc/tapex
@@ -687,6 +691,8 @@
         title: Zamba2
       title: Text models
     - sections:
+      - local: model_doc/aimv2
+        title: Aimv2
       - local: model_doc/beit
         title: BEiT
       - local: model_doc/bit
@@ -731,6 +737,8 @@
         title: EfficientFormer
       - local: model_doc/efficientnet
         title: EfficientNet
+      - local: model_doc/eomt
+        title: EoMT
       - local: model_doc/focalnet
         title: FocalNet
       - local: model_doc/glpn
@@ -743,6 +751,8 @@
         title: ImageGPT
       - local: model_doc/levit
         title: LeViT
+      - local: model_doc/lightglue
+        title: LightGlue
       - local: model_doc/mask2former
         title: Mask2Former
       - local: model_doc/maskformer
@@ -831,6 +841,8 @@
         title: CSM
       - local: model_doc/dac
         title: dac
+      - local: model_doc/dia
+        title: Dia
       - local: model_doc/encodec
         title: EnCodec
       - local: model_doc/fastspeech2_conformer
@@ -839,6 +851,8 @@
         title: GraniteSpeech
       - local: model_doc/hubert
         title: Hubert
+      - local: model_doc/kyutai_speech_to_text
+        title: Kyutai Speech-To-Text
       - local: model_doc/mctct
         title: MCTCT
       - local: model_doc/mimi
@@ -947,8 +961,12 @@
         title: FLAVA
       - local: model_doc/gemma3
         title: Gemma3
+      - local: model_doc/gemma3n
+        title: Gemma3n
       - local: model_doc/git
         title: GIT
+      - local: model_doc/glm4v
+        title: glm4v
       - local: model_doc/got_ocr2
         title: GOT-OCR2
       - local: model_doc/granitevision
@@ -1041,6 +1059,8 @@
         title: SigLIP
       - local: model_doc/siglip2
         title: SigLIP2
+      - local: model_doc/smollm3
+        title: SmolLM3
       - local: model_doc/smolvlm
         title: SmolVLM
       - local: model_doc/speech-encoder-decoder
@@ -1124,4 +1144,3 @@
       title: Environment Variables
     title: Reference
   title: API
-

docs/source/en/add_new_model.md

@@ -13,7 +13,7 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Adding a new model to Transformers
+# Legacy model contribution
 
 > [!TIP]
 > Try adding new models with a more [modular](./modular_transformers) approach first. This makes it significantly easier to contribute a model to Transformers!
@@ -571,7 +571,7 @@ The processor should call the appropriate modality-specific processors within it
 def __call__(
     self,
     images: ImageInput = None,
-    text: Union[TextInput, PreTokenizedInput, List[TextInput], List[PreTokenizedInput]] = None,
+    text: Union[TextInput, PreTokenizedInput, list[TextInput], list[PreTokenizedInput]] = None,
     audio=None,
     videos=None,
     **kwargs: Unpack[YourModelProcessorKwargs],

docs/source/en/agents.md

@@ -14,5 +14,9 @@ rendered properly in your Markdown viewer.
 
 -->
 
+# Agents
+
+(deprecated)
+
 > [!WARNING]
 > Agents and tools were spun out into the standalone [smolagents](https://huggingface.co/docs/smolagents/index) library. They were removed from `transformers` in v4.52.

docs/source/en/attention_interface.md

@@ -92,7 +92,7 @@ def custom_attention(
     a_new_kwargs = None,  # You can now add as many kwargs as you need
     another_new_kwargs = None,  # You can now add as many kwargs as you need
     **kwargs,  # You need to accept **kwargs as models will pass other args
-) -> Tuple[torch.Tensor, Optional[torch.Tensor]]
+) -> tuple[torch.Tensor, Optional[torch.Tensor]]
     ...  # do your magic!
     return attn_output, attn_weights  # attn_weights are optional here
 

docs/source/en/auto_docstring.md

@@ -14,43 +14,26 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Utilizing the @auto_docstring Decorator
+# Documenting a model
 
-The `@auto_docstring` decorator in the Hugging Face Transformers library helps generate docstrings for model classes and their methods, which will be used to build the documentation for the library. It aims to improve consistency and reduce boilerplate by automatically including standard argument descriptions and allowing for targeted overrides and additions.
+The `@auto_docstring` decorator in Transformers generates consistent docstrings for model classes and their methods. It reduces boilerplate by automatically including standard argument descriptions while also allowing overrides to add new or custom arguments. [Contributing a new model](./modular_transformers) is easier because you don't need to manually add the standard docstrings, and only focus on documenting new arguments.
 
----
+This guide describes how to use the `@auto_docstring` decorator and how it works.
 
-## 📜 How it Works
+## @auto_docstring
 
-The `@auto_docstring` decorator constructs docstrings by:
-
-1.  **Signature Inspection:** It inspects the signature (arguments, types, defaults) of the decorated class's `__init__` method or the decorated function.
-2.  **Centralized Docstring Fetching:** It retrieves predefined docstrings for common arguments (e.g., `input_ids`, `attention_mask`) from internal library sources (like `ModelArgs` or `ImageProcessorArgs` in `utils/args_doc.py`).
-3.  **Overriding or Adding Arguments Descriptions:**
-    * **Direct Docstring Block:** It incorporates custom docstring content from an `r""" """` (or `""" """`) block below the method signature or within the `__init__` docstring. This is for documenting new arguments or overriding standard descriptions.
-    * **Decorator Arguments (`custom_args`):** A `custom_args` docstring block can be passed to the decorator to provide docstrings for specific arguments directly in the decorator call. This can be used to define the docstring block for new arguments once if they are repeated in multiple places in the modeling file.
-4.  **Adding Classes and Functions Introduction:**
-    * **`custom_intro` argument:** Allows prepending a custom introductory paragraph to a class or function docstring.
-    * **Automatic Introduction Generation:** For model classes with standard naming patterns (like `ModelForCausalLM`) or belonging to a pipeline, the decorator automatically generates an appropriate introductory paragraph using `ClassDocstring` in `utils/args_doc.py` as the source.
-5.  **Templating:** The decorator uses a templating system, allowing predefined docstrings to include dynamic information deduced from the `auto_modules` of the library, such as `{{processor_class}}` or `{{config_class}}`.
-6.  **Deducing Relevant Examples:** The decorator attempts to find appropriate usage examples based on the model's task or pipeline compatibility. It extracts checkpoint information from the model's configuration class to provide concrete examples with real model identifiers.
-7.  **Adding Return Value Documentation:** For methods like `forward`, the decorator can automatically generate the "Returns" section based on the method's return type annotation. For example, for a method returning a `ModelOutput` subclass, it will extracts field descriptions from that class's docstring to create a comprehensive return value description. A custom `Returns` section can also be manually specified in the function docstring block.
-8.  **Unrolling Kwargs Typed With Unpack Operator:** For specific methods (defined in `UNROLL_KWARGS_METHODS`) or classes (defined in `UNROLL_KWARGS_CLASSES`), the decorator processes `**kwargs` parameters that are typed with `Unpack[KwargsTypedDict]`. It extracts the documentation from the TypedDict and adds each parameter to the function's docstring. Currently, this functionality is only supported for `FastImageProcessorKwargs`.
-
-
----
-
-## 🚀 How to Use @auto_docstring
-
-### 1. Importing the Decorator
-Import the decorator into your modeling file:
+Start by importing the decorator in the modeling file (`modular_model.py` or `modeling_model.py`).
 
 ```python
 from ...utils import auto_docstring
 ```
 
-### 2. Applying to Classes
-Place `@auto_docstring` directly above the class definition. It uses the `__init__` method's signature and its docstring for parameter descriptions.
+Select whether you'd like to apply `@auto_docstring` to a class or function below to see how to use it.
+
+<hfoptions id="type">
+<hfoption id="classes">
+
+Place `@auto_docstring` directly above the class definition. The decorator derives parameter descriptions from the `__init__` method's signature and docstring.
 
 ```python
 from transformers.modeling_utils import PreTrainedModel
@@ -73,9 +56,7 @@ class MyAwesomeModel(PreTrainedModel):
     # ... other methods
 ```
 
-#### Advanced Class Decoration:
-
-Arguments can be passed directly to `@auto_docstring` for more control:
+Arguments can also be passed directly to `@auto_docstring` for more control. Use the `custom_intro` parameter to describe the argument and the `custom_args` parameter to describe the arguments.
 
 ```python
 @auto_docstring(
@@ -93,7 +74,7 @@ class MySpecialModel(PreTrainedModel):
         # ...
 ```
 
-Or:
+You can also choose to only use `custom_intro` and define the custom arguments directly in the class.
 
 ```python
 @auto_docstring(
@@ -111,8 +92,10 @@ class MySpecialModel(PreTrainedModel):
         # ...
 ```
 
-### 3. Applying to Functions (e.g., `forward` method)
-Apply the decorator above method definitions, such as the `forward` method.
+</hfoption>
+<hfoption id="functions">
+
+Place `@auto_docstring` directly above the method definition. The decorator derives parameter descriptions from the function signature.
 
 ```python
     @auto_docstring
@@ -131,9 +114,10 @@ Apply the decorator above method definitions, such as the `forward` method.
         # ...
 ```
 
-#### Advanced Function Decoration:
+Arguments can also be passed directly to `@auto_docstring` for more control. Use the `custom_intro` parameter to describe the argument and the `custom_args` parameter to describe the arguments.
+
+The `Returns` and `Examples` parts of the docstring can also be manually specified.
 
-Arguments can be passed directly to `@auto_docstring` for more control. `Returns` and `Examples` sections can also be manually specified:
 
 ```python
 MODEL_COMMON_CUSTOM_ARGS = r"""
@@ -180,100 +164,117 @@ class MyModel(PreTrainedModel):
         # ...
 ```
 
----
+</hfoption>
+</hfoptions>
 
-### ✍️ Documenting Arguments: Approach & Priority
+## Documenting arguments
 
-1.  **Standard Arguments (e.g., `input_ids`, `attention_mask`, `pixel_values`, `encoder_hidden_states` etc.):**
-    * `@auto_docstring` retrieves descriptions from a central source. Do not redefine these locally if their description and shape are the same as in `args_doc.py`.
+There are some rules for documenting different types of arguments and they're listed below.
+
+- Standard arguments (`input_ids`, `attention_mask`, `pixel_values`, etc.) are defined and retrieved from `args_doc.py`. It is the single source of truth for standard arguments and should not be redefined locally if an argument's description and shape is the same as an argument in `args_doc.py`.
+
+    If a standard argument behaves differently in your model, then you can override it locally in a `r""" """` block. This local definition has a higher priority. For example, the `labels` argument is often customized per model and typically requires overriding.
+
+
+- New or custom arguments should be documented within an `r""" """` block after the signature if it is a function or in the `__init__` method's docstring if it is a class.
+
+    ```py
+    argument_name (`type`, *optional*, defaults to `X`):
+        Description of the argument.
+        Explain its purpose, expected shape/type if complex, and default behavior.
+        This can span multiple lines.
+    ```
 
-2.  **New or Custom Arguments:**
-    * **Primary Method:** Document these within an `r""" """` docstring block following the signature (for functions) or in the `__init__` method's docstring (for class parameters).
-    * **Format:**
-        ```
-        argument_name (`type`, *optional*, defaults to `X`):
-            Description of the argument.
-            Explain its purpose, expected shape/type if complex, and default behavior.
-            This can span multiple lines.
-        ```
     * Include `type` in backticks.
-    * Add "*optional*" if the argument is not required (has a default value).
-    * Add "defaults to `X`" if it has a default value (no need to specify "defaults to `None`" if the default value is `None`).
+    * Add *optional* if the argument is not required or has a default value.
+    * Add "defaults to X" if it has a default value. You don't need to add "defaults to `None`" if the default value is `None`.
 
-3.  **Overriding Standard Arguments:**
-    * If a standard argument behaves differently (e.g., different expected shape, model-specific behavior), provide its complete description in the local `r""" """` docstring. This local definition takes precedence.
-    * The `labels` argument is often customized per model and typically requires a specific docstring.
+    These arguments can also be passed to `@auto_docstring` as a `custom_args` argument. It is used to define the docstring block for new arguments once if they are repeated in multiple places in the modeling file.
 
-4.  **Using Decorator Arguments for Overrides or New Arguments (`custom_args`):**
-    * New or custom arguments docstrings can also be passed to `@auto_docstring` as a `custom_args` argument. This can be used to define the docstring block for new arguments once if they are repeated in multiple places in the modeling file.
+    ```py
+    class MyModel(PreTrainedModel):
+    # ...
+    @auto_docstring(
+        custom_intro="""
+        This is a custom introduction for the function.
+        """
+        custom_args=r"""
+        common_arg_1 (`torch.Tensor`, *optional*, defaults to `default_value`):
+            Description of common_arg_1
+        """
+    )
+    ```
 
----
+## Checking the docstrings
 
-### Usage with [modular files](./modular_transformers)
+Transformers includes a utility script to validate the docstrings when you open a Pull Request which triggers CI (continuous integration) checks. The script checks for the following criteria.
 
-When working with modular files, follow these guidelines for applying the `@auto_docstring` decorator:
+* Ensures `@auto_docstring` is applied to relevant mode classes and public methods.
+* Ensures arguments are complete and consistent. It checks that documented arguments exist in the signature and verifies whether the types and default values in the docstring match the signature. Arguments that aren't known standard arguments or if they lack a local description are flagged.
+* Reminds you to complete placeholders like `<fill_type>` and `<fill_docstring>`.
+* Ensures docstrings are formatted according to the expected docstring style.
 
-- **For standalone models in modular files:**
-  Apply the `@auto_docstring` decorator just as you would in regular modeling files.
-
-- **For models inheriting from other library models:**
-  - When inheriting from a parent model, decorators (including `@auto_docstring`) are automatically carried over to the generated modeling file without needing to add them in your modular file.
-  - If you need to modify the `@auto_docstring` behavior, apply the customized decorator in your modular file, making sure to *include all other decorators* that were present on the original function/class.
-
-  > **Warning**: When overriding any decorator in a modular file, you must include ALL decorators that were applied to that function/class in the parent model. If you only override some decorators, the others won't be included in the generated modeling file.
-
-
-**Note**: The `check_auto_docstrings` tool doesn't check modular files directly, but it will check (and modify when using `--fix_and_overwrite`) the generated modeling files. If issues are found in the generated files, you'll need to update your modular files accordingly.
-
----
-
-## ✅ Checking Your Docstrings with `check_auto_docstrings`
-
-The library includes a utility script to validate docstrings. This check is typically run during Continuous Integration (CI).
-
-#### What it Checks:
-
-* **Decorator Presence:** Ensures `@auto_docstring` is applied to relevant model classes and public methods. (TODO)
-* **Argument Completeness & Consistency:**
-    * Flags arguments in the signature that are not known standard arguments and lack a local description.
-    * Ensures documented arguments exist in the signature. (TODO)
-    * Verifies that types and default values in the docstring match the signature. (TODO)
-* **Placeholder Detection:** Reminds you to complete placeholders like `<fill_type>` or `<fill_docstring>`.
-* **Formatting:** Adherence to the expected docstring style.
-
-#### Running the Check Locally:
-
-Run this check locally before committing. The common command is:
+You can run this check locally - before committing - by running the following command.
 
 ```bash
 make fix-copies
 ```
 
-Alternatively, to only perform docstrings and auto-docstring checks, you can use:
+`make fix-copies` runs several other checks as well. If you don't need those checks, run the command below to only perform docstring and auto-docstring checks.
 
 ```bash
 python utils/check_docstrings.py # to only check files included in the diff without fixing them
-# Or: python utils/check_docstrings.py --fix_and_overwrite # to fix and overwrite the files in the diff
-# Or: python utils/check_docstrings.py --fix_and_overwrite --check_all # to fix and overwrite all files
+# python utils/check_docstrings.py --fix_and_overwrite # to fix and overwrite the files in the diff
+# python utils/check_docstrings.py --fix_and_overwrite --check_all # to fix and overwrite all files
 ```
 
-#### Workflow with the Checker:
+## modular_model.py files
 
-1.  Add `@auto_docstring(...)` to the class or method.
-2.  For new, custom, or overridden arguments, add descriptions in an `r""" """` block.
-3.  Run `make fix-copies` (or the `check_docstrings.py` utility).
-    * For unrecognized arguments lacking documentation, the utility will create placeholder entries.
-4.  Manually edit these placeholders with accurate types and descriptions.
-5.  Re-run the check to ensure all issues are resolved.
+When working with modular files (`modular_model.py`), follow the guidelines below for applying `@auto_docstring`.
 
----
+- For standalone models in modular files, apply `@auto_docstring` like you would in a `modeling_model.py` file.
+- For models that inherit from other library models, `@auto_docstring` is automatically carried over to the generated modeling file. You don't need to add `@auto_docstring` in your modular file.
 
-## 🔑 Key Takeaways & Best Practices
+    If you need to modify the `@auto_docstring` behavior, apply the customized decorator in your modular file. Make sure to **include all other decorators** that are present in the original function or class.
 
-* Use `@auto_docstring` for new PyTorch model classes (`PreTrainedModel` subclasses) and their primary for methods (e.g., `forward`, `get_text_features` etc.).
-* For classes, the `__init__` method's docstring is the main source for parameter descriptions when using `@auto_docstring` on the class.
-* Rely on standard docstrings; do not redefine common arguments unless their behavior is different in your specific model.
+> [!WARNING]
+> When overriding any decorator in a modular file, you must include **all** decorators that were applied to that function or class in the parent model. If you only override some decorators, the others won't be included in the generated modeling file.
+
+## How it works
+
+The `@auto_docstring` decorator automatically generates docstrings by:
+
+1. Inspecting the signature (arguments, types, defaults) of the decorated class' `__init__` method or the decorated function.
+2. Retrieving the predefined docstrings for common arguments (`input_ids`, `attention_mask`, etc.) from internal library sources like [`ModelArgs`], [`ImageProcessorArgs`], and the `args_doc.py` file.
+3. Adding argument descriptions in one of two ways as shown below.
+
+    | method | description | usage |
+    |---|---|---|
+    | `r""" """` | add custom docstring content directly to a method signature or within the `__init__` docstring | document new arguments or override standard descriptions |
+    | `custom_args` | add custom docstrings for specific arguments directly in `@auto_docstring` | define docstring for new arguments once if they're repeated in multiple places in the modeling file |
+
+4. Adding class and function descriptions. For model classes with standard naming patterns, like `ModelForCausalLM`, or if it belongs to a pipeline, `@auto_docstring` automatically generates the appropriate descriptions with `ClassDocstring` from `args_doc.py`.
+
+    `@auto_docstring` also accepts the `custom_intro` argument to describe a class or function.
+
+5. Using a templating system to allow predefined docstrings to include dynamic information from Transformers' [auto_modules](https://github.com/huggingface/transformers/tree/main/src/transformers/models/auto) such as `{{processor_class}}` and `{{config_class}}`.
+
+6. Finding appropriate usage examples based on the model's task or pipeline compatibility. It extracts checkpoint information form the model's configuration class to provide concrete examples with real model identifiers.
+
+7. Adding return values to the docstring. For methods like `forward`, the decorator automatically generates the `Returns` field in the docstring based on the method's return type annotation.
+
+    For example, if a method returns a [`~transformers.utils.ModelOutput`] subclass, `@auto_docstring` extracts the field descriptions from the class' docstring to create a comprehensive return value description. You can also manually specifiy a custom `Returns` field in a functions docstring.
+
+8. Unrolling kwargs typed with the unpack operator. For specific methods (defined in `UNROLL_KWARGS_METHODS`) or classes (defined in `UNROLL_KWARGS_CLASSES`), the decorator processes `**kwargs` parameters that are typed with `Unpack[KwargsTypedDict]`. It extracts the documentations from the `TypedDict` and adds each parameter to the function's docstring.
+
+    Currently only supported for [`FastImageProcessorKwargs`].
+
+## Best practices
+
+Follow the best practices below to help maintain consistent and informative documentation for Transformers!
+
+* Use `@auto_docstring` for new PyTorch model classes ([`PreTrainedModel`] subclasses) and their primary methods like `forward` or `get_text_features`.
+* For classes, `@auto_docstring` retrieves parameter descriptions from the `__init__` method's docstring.
+* Rely on standard docstrings and do not redefine common arguments unless their behavior is different in your model.
 * Document new or custom arguments clearly.
 * Run `check_docstrings` locally and iteratively.
-
-By following these guidelines, you help maintain consistent and informative documentation for the Hugging Face Transformers library 🤗.

docs/source/en/chat_templating_multimodal.md

@@ -56,7 +56,7 @@ Create a [`ImageTextToTextPipeline`] and pass the chat to it. For large models,
 import torch
 from transformers import pipeline
 
-pipeline = pipeline("image-text-to-text", model="llava-hf/llava-onevision-qwen2-0.5b-ov-hf", device="cuda", torch_dtype=torch.float16)
+pipeline = pipeline("image-text-to-text", model="llava-hf/llava-onevision-qwen2-0.5b-ov-hf", device_map="auto", torch_dtype=torch.float16)
 pipeline(text=messages, max_new_tokens=50, return_full_text=False)
 [{'input_text': [{'role': 'system',
     'content': [{'type': 'text',
@@ -175,7 +175,7 @@ processed_chat = processor.apply_chat_template(
     add_generation_prompt=True,
     tokenize=True,
     return_dict=True,
-    video_fps=32,
+    video_fps=16,
     video_load_backend="decord",
 )
 print(processed_chat.keys())

docs/source/en/conversations.md

@@ -25,7 +25,7 @@ Check model leaderboards like [OpenLLM](https://hf.co/spaces/HuggingFaceH4/open_
 
 This guide shows you how to quickly start chatting with Transformers from the command line, how build and format a conversation, and how to chat using the [`TextGenerationPipeline`].
 
-## transformers CLI
+## chat CLI
 
 After you've [installed Transformers](./installation.md), chat with a model directly from the command line as shown below. It launches an interactive session with a model, with a few base commands listed at the start of the session.
 
@@ -49,7 +49,8 @@ For a full list of options, run the command below.
 transformers chat -h
 ```
 
-The chat is implemented on top of the [AutoClass](./model_doc/auto), using tooling from [text generation](./llm_tutorial) and [chat](./chat_templating).
+The chat is implemented on top of the [AutoClass](./model_doc/auto), using tooling from [text generation](./llm_tutorial) and [chat](./chat_templating). It uses the `transformers serve` CLI under the hood ([docs](./serving.md#serve-cli)).
+
 
 ## TextGenerationPipeline
 

docs/source/en/custom_models.md

@@ -47,7 +47,7 @@ class ResnetConfig(PretrainedConfig):
     def __init__(
         self,
         block_type="bottleneck",
-        layers: List[int] = [3, 4, 6, 3],
+        layers: list[int] = [3, 4, 6, 3],
         num_classes: int = 1000,
         input_channels: int = 3,
         cardinality: int = 1,

docs/source/en/feature_extractors.md

@@ -26,6 +26,7 @@ Pass the audio signal, typically stored in `array`, to the feature extractor and
 from transformers import AutoFeatureExtractor
 
 feature_extractor = AutoFeatureExtractor.from_pretrained("facebook/wav2vec2-base")
+dataset = load_dataset("PolyAI/minds14", name="en-US", split="train")
 processed_sample = feature_extractor(dataset[0]["audio"]["array"], sampling_rate=16000)
 processed_sample
 {'input_values': [array([ 9.4472744e-05,  3.0777880e-03, -2.8888427e-03, ...,

docs/source/en/generation_strategies.md

@@ -468,9 +468,17 @@ def generate(model, input_ids, generation_config=None, left_padding=None, **kwar
 Follow the recommended practices below to ensure your custom decoding method works as expected.
 - Feel free to reuse the logic for validation and input preparation in the original [`~GenerationMixin.generate`].
 - Pin the `transformers` version in the requirements if you use any private method/attribute in `model`.
-- You can add other files in the `custom_generate` folder, and use relative imports.
 - Consider adding model validation, input validation, or even a separate test file to help users sanity-check your code in their environment.
 
+Your custom `generate` method can relative import code from the `custom_generate` folder. For example, if you have a `utils.py` file, you can import it like this:
+
+```py
+from .utils import some_function
+```
+
+Only relative imports from the same-level `custom_generate` folder are supported. Parent/sibling folder imports are not valid. The `custom_generate` argument also works locally with any directory that contains a `custom_generate` structure. This is the recommended workflow for developing your custom decoding method.
+
+
 #### requirements.txt
 
 You can optionally specify additional Python requirements in a `requirements.txt` file inside the `custom_generate` folder. These are checked at runtime and an exception will be thrown if they're missing, nudging users to update their environment accordingly.

docs/source/en/kv_cache.md

@@ -44,7 +44,7 @@ import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
-model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16).to("cuda:0")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16, device_map="auto")
 inputs = tokenizer("I like rock music because", return_tensors="pt").to(model.device)
 
 model.generate(**inputs, do_sample=False, max_new_tokens=20, use_cache=False)
@@ -59,7 +59,7 @@ import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM, DynamicCache
 
 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
-model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16).to("cuda:0")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16, device_map="auto")
 inputs = tokenizer("I like rock music because", return_tensors="pt").to(model.device)
 
 past_key_values = DynamicCache()
@@ -142,13 +142,14 @@ Enable [`QuantizedCache`] by configuring `cache_implementation="quantized"` in [
 For [`HQQQuantizedCache`], we recommend setting the `axis-key` and `axis-value` parameters to `1`.
 
 ```py
+import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM, HQQQuantizedCache, QuantizedCacheConfig
 
 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
-model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16).to("cuda:0")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16, device_map="auto")
 inputs = tokenizer("I like rock music because", return_tensors="pt").to(model.device)
 
-out = model.generate(**inputs, do_sample=False, max_new_tokens=20, cache_implementation="quantized", cache_config={"axis-key": 1, "axis-value": 1, "backend": "hqq"})
+out = model.generate(**inputs, do_sample=False, max_new_tokens=20, cache_implementation="quantized", cache_config={"backend": "HQQ"})
 print(tokenizer.batch_decode(out, skip_special_tokens=True)[0])
 I like rock music because it's loud and energetic. It's a great way to express myself and rel
 ```
@@ -159,13 +160,14 @@ I like rock music because it's loud and energetic. It's a great way to express m
 For [`QuantoQuantizedCache`], we recommend setting the `axis-key` and `axis-value` parameters to `0`.
 
 ```py
+import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM, QuantoQuantizedCache, QuantizedCacheConfig
 
 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
-model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16).to("cuda:0")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16, device_map="auto")
 inputs = tokenizer("I like rock music because", return_tensors="pt").to(model.device)
 
-out = model.generate(**inputs, do_sample=False, max_new_tokens=20, cache_implementation="quantized", cache_config={"nbits": 4, "axis-key": 0, "axis-value": 0, "backend": "quanto"})
+out = model.generate(**inputs, do_sample=False, max_new_tokens=20, cache_implementation="quantized", cache_config={"nbits": 4, "backend": "quanto"})
 print(tokenizer.batch_decode(out, skip_special_tokens=True)[0])
 I like rock music because it's loud and energetic. It's a great way to express myself and rel
 ```
@@ -207,14 +209,14 @@ import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
 tokenizer = AutoTokenizer.from_pretrained("meta-llama/Llama-2-7b-chat-hf")
-model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16, device_map="auto")
+model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-7b-chat-hf", torch_dtype=torch.float16, device_map={"": 0})
 inputs = tokenizer("Hello, my name is", return_tensors="pt").to(model.device)
 
 out = model.generate(**inputs, do_sample=False, max_new_tokens=20, cache_implementation="offloaded_static")
 tokenizer.batch_decode(out, skip_special_tokens=True)[0]
 "Hello, my name is [Your Name], and I am a [Your Profession] with [Number of Years] of"
 ```
-Cache offloading requires a CUDA GPU.
+Cache offloading requires a CUDA GPU or Intel XPU.
 
 ### Sliding window cache
 
@@ -227,7 +229,7 @@ import torch
 from transformers import AutoTokenizer, AutoModelForCausalLM
 
 tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
-model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1", torch_dtype=torch.float16).to("cuda:0")
+model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1", torch_dtype=torch.float16, device_map="auto")
 inputs = tokenizer("Yesterday I was on a rock concert and.", return_tensors="pt").to(model.device)
 
 out = model.generate(**inputs, do_sample=False, max_new_tokens=30, cache_implementation="sliding_window")
@@ -306,15 +308,15 @@ import torch
 from transformers import AutoModelForCausalLM, AutoTokenizer, DynamicCache, StaticCache
 
 model_id = "meta-llama/Llama-2-7b-chat-hf"
-model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="cuda")
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map={"": 0})
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 
 # Init StaticCache with big enough max-length (1024 tokens for the below example)
 # You can also init a DynamicCache, if that suits you better
-prompt_cache = StaticCache(config=model.config, max_batch_size=1, max_cache_len=1024, device="cuda", dtype=torch.bfloat16)
+prompt_cache = StaticCache(config=model.config, max_batch_size=1, max_cache_len=1024, device=model.device.type, dtype=torch.bfloat16)
 
 INITIAL_PROMPT = "You are a helpful assistant. "
-inputs_initial_prompt = tokenizer(INITIAL_PROMPT, return_tensors="pt").to("cuda")
+inputs_initial_prompt = tokenizer(INITIAL_PROMPT, return_tensors="pt").to(model.device.type)
 # This is the common prompt cached, we need to run forward without grad to be able to copy
 with torch.no_grad():
      prompt_cache = model(**inputs_initial_prompt, past_key_values = prompt_cache).past_key_values
@@ -322,7 +324,7 @@ with torch.no_grad():
 prompts = ["Help me to write a blogpost about travelling.", "What is the capital of France?"]
 responses = []
 for prompt in prompts:
-    new_inputs = tokenizer(INITIAL_PROMPT + prompt, return_tensors="pt").to("cuda")
+    new_inputs = tokenizer(INITIAL_PROMPT + prompt, return_tensors="pt").to(model.device.type)
     past_key_values = copy.deepcopy(prompt_cache)
     outputs = model.generate(**new_inputs, past_key_values=past_key_values,max_new_tokens=20)
     response = tokenizer.batch_decode(outputs)[0]

docs/source/en/llm_tutorial.md

@@ -152,7 +152,7 @@ print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
 | `temperature` | `float` | How unpredictable the next selected token will be. High values (`>0.8`) are good for creative tasks, low values (e.g. `<0.4`) for tasks that require "thinking". Requires `do_sample=True`. |
 | `num_beams` | `int` | When set to `>1`, activates the beam search algorithm. Beam search is good on input-grounded tasks. Check [this guide](./generation_strategies.md) for more information. |
 | `repetition_penalty` | `float` | Set it to `>1.0` if you're seeing the model repeat itself often. Larger values apply a larger penalty. |
-| `eos_token_id` | `List[int]` | The token(s) that will cause generation to stop. The default value is usually good, but you can specify a different token. |
+| `eos_token_id` | `list[int]` | The token(s) that will cause generation to stop. The default value is usually good, but you can specify a different token. |
 
 
 ## Pitfalls

docs/source/en/model_doc/aimv2.md

@@ -0,0 +1,104 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# AIMv2
+
+## Overview
+
+The AIMv2 model was proposed in [Multimodal Autoregressive Pre-training of Large Vision Encoders](https://arxiv.org/abs/2411.14402) by Enrico Fini, Mustafa Shukor, Xiujun Li, Philipp Dufter, Michal Klein, David Haldimann, Sai Aitharaju, Victor Guilherme Turrisi da Costa, Louis Béthune, Zhe Gan, Alexander T Toshev, Marcin Eichner, Moin Nabi, Yinfei Yang, Joshua M. Susskind, Alaaeldin El-Nouby.
+
+The abstract from the paper is the following:
+
+*We introduce a novel method for pre-training of large-scale vision encoders. Building on recent advancements in autoregressive pre-training of vision models, we extend this framework to a multimodal setting, i.e., images and text. In this paper, we present AIMV2, a family of generalist vision encoders characterized by a straightforward pre-training process, scalability, and remarkable performance across a range of downstream tasks. This is achieved by pairing the vision encoder with a multimodal decoder that autoregressively generates raw image patches and text tokens. Our encoders excel not only in multimodal evaluations but also in vision benchmarks such as localization, grounding, and classification. Notably, our AIMV2-3B encoder achieves 89.5% accuracy on ImageNet-1k with a frozen trunk. Furthermore, AIMV2 consistently outperforms state-of-the-art contrastive models (e.g., CLIP, SigLIP) in multimodal image understanding across diverse settings.*
+
+
+This model was contributed by [Yaswanth Gali](https://huggingface.co/yaswanthgali).
+The original code can be found [here](https://github.com/apple/ml-aim).
+
+## Usage Example
+
+Here is an example of Image Feature Extraction using specific checkpoints on resized images and native resolution images:
+
+```python
+import requests
+from PIL import Image
+from transformers import AutoImageProcessor, AutoModel
+
+url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+image = Image.open(requests.get(url, stream=True).raw)
+
+processor = AutoImageProcessor.from_pretrained("apple/aimv2-large-patch14-native")
+model = AutoModel.from_pretrained("apple/aimv2-large-patch14-native")
+
+inputs = processor(images=image, return_tensors="pt")
+outputs = model(**inputs)
+```
+
+Here is an example of a checkpoint performing zero-shot classification:
+
+```python
+import requests
+from PIL import Image
+from transformers import AutoProcessor, AutoModel
+
+url = "http://images.cocodataset.org/val2017/000000039769.jpg"
+image = Image.open(requests.get(url, stream=True).raw)
+text = ["Picture of a dog.", "Picture of a cat.", "Picture of a horse."]
+
+processor = AutoProcessor.from_pretrained("apple/aimv2-large-patch14-224-lit")
+model = AutoModel.from_pretrained("apple/aimv2-large-patch14-224-lit")
+
+inputs = processor(
+    images=image,
+    text=text,
+    add_special_tokens=True,
+    truncation=True,
+    padding=True,
+    return_tensors="pt",
+)
+outputs = model(**inputs)
+probs = outputs.logits_per_image.softmax(dim=-1)
+```
+
+## Aimv2Config
+
+[[autodoc]] Aimv2Config
+
+## Aimv2TextConfig
+
+[[autodoc]] Aimv2TextConfig
+
+## Aimv2VisionConfig
+
+[[autodoc]] Aimv2VisionConfig
+
+## Aimv2Model
+
+[[autodoc]] Aimv2Model
+    - forward
+
+## Aimv2VisionModel
+
+[[autodoc]] Aimv2VisionModel
+    - forward
+
+## Aimv2TextModel
+
+[[autodoc]] Aimv2TextModel
+    - forward
+
+</pt>
+<tf>

docs/source/en/model_doc/arcee.md

@@ -0,0 +1,104 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
+</div>
+
+# Arcee
+
+Arcee is a decoder-only transformer model based on the Llama architecture with a key modification: it uses ReLU² (ReLU-squared) activation in the MLP blocks instead of SiLU, following recent research showing improved training efficiency with squared activations. This architecture is designed for efficient training and inference while maintaining the proven stability of the Llama design.
+
+The Arcee model is architecturally similar to Llama but uses `x * relu(x)` in MLP layers for improved gradient flow and is optimized for efficiency in both training and inference scenarios.
+
+> [!TIP]
+> The Arcee model supports extended context with RoPE scaling and all standard transformers features including Flash Attention 2, SDPA, gradient checkpointing, and quantization support.
+
+The example below demonstrates how to generate text with Arcee using [`Pipeline`] or the [`AutoModel`].
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```py
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="text-generation",
+    model="arcee-ai/AFM-4.5B",
+    torch_dtype=torch.float16,
+    device=0
+)
+
+output = pipeline("The key innovation in Arcee is")
+print(output[0]["generated_text"])
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoTokenizer, ArceeForCausalLM
+
+tokenizer = AutoTokenizer.from_pretrained("arcee-ai/AFM-4.5B")
+model = ArceeForCausalLM.from_pretrained(
+    "arcee-ai/AFM-4.5B",
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+
+inputs = tokenizer("The key innovation in Arcee is", return_tensors="pt")
+with torch.no_grad():
+    outputs = model.generate(**inputs, max_new_tokens=50)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+</hfoption>
+</hfoptions>
+
+## ArceeConfig
+
+[[autodoc]] ArceeConfig
+
+## ArceeModel
+
+[[autodoc]] ArceeModel
+    - forward
+
+## ArceeForCausalLM
+
+[[autodoc]] ArceeForCausalLM
+    - forward
+
+## ArceeForSequenceClassification
+
+[[autodoc]] ArceeForSequenceClassification
+    - forward
+
+## ArceeForQuestionAnswering
+
+[[autodoc]] ArceeForQuestionAnswering
+    - forward
+
+## ArceeForTokenClassification
+
+[[autodoc]] ArceeForTokenClassification
+    - forward

docs/source/en/model_doc/auto.md

@@ -350,6 +350,10 @@ The following auto classes are available for the following audio tasks.
 
 [[autodoc]] AutoModelForTextToWaveform
 
+### AutoModelForAudioTokenization
+
+[[autodoc]] AutoModelForAudioTokenization
+
 ## Multimodal
 
 The following auto classes are available for the following multimodal tasks.

docs/source/en/model_doc/bamba.md

@@ -14,84 +14,127 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Bamba
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-<img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
-<img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
 </div>
 
-## Overview
+# Bamba
 
-Bamba-9B is a decoder-only language model based on the [Mamba-2](https://github.com/state-spaces/mamba) architecture and is designed to handle a wide range of text generation tasks. It is trained from scratch using a two-stage training approach. In the first stage, the model is trained on 2 trillion tokens from the Dolma v1.7 dataset. In the second stage, it undergoes additional training on 200 billion tokens, leveraging a carefully curated blend of high-quality data to further refine its performance and enhance output quality.
+[Bamba](https://huggingface.co/blog/bamba) is a 9B parameter decoder-only language model built on the [Mamba-2](./mamba2) architecture. It is pretrained in two stages - it starts by training on 2T tokens from the [Dolma v1.7](https://huggingface.co/datasets/allenai/dolma) dataset and then trained on an additional 200B tokens from [FineWeb](https://huggingface.co/datasets/HuggingFaceFW/fineweb) and [Cosmopedia](https://huggingface.co/datasets/HuggingFaceTB/cosmopedia).
 
-Checkout all Bamba-9B model checkpoints [here](https://github.com/foundation-model-stack/bamba).
+You can find all the original Bamba checkpoints under the [Bamba](https://huggingface.co/collections/ibm-ai-platform/bamba-674f1388b9bbc98b413c7bab) collection.
+
+> [!TIP]
+> This model was contributed by [ani300](https://github.com/ani300) and [fabianlim](https://github.com/fabianlim).
+>
+> Click on the Bamba models in the right sidebar for more examples of how to apply Bamba to different text generation tasks.
+
+The example below demonstrates how to generate text with [`Pipeline`], [`AutoModel`], and from the command line.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```python
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="text-generation",
+    model="ibm-ai-platform/Bamba-9B-v2",
+    torch_dtype=torch.bfloat16,
+    device=0
+)
+pipeline("Plants create energy through a process known as")
+```
+
+</hfoption>
+
+<hfoption id="AutoModel">
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained("ibm-ai-platform/Bamba-9B-v2")
+model = AutoModelForCausalLM.from_pretrained("ibm-ai-platform/Bamba-9B-v2", torch_dtype=torch.bfloat16, device_map="auto", attn_implementation="sdpa")
+input_ids = tokenizer("Plants create energy through a process known as", return_tensors="pt").to("cuda")
+
+output = model.generate(**input_ids)
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+</hfoption>
+
+<hfoption id="transformers CLI">
+```bash
+echo "Plants create energy through a process known as" | transformers-cli run --task text-generation --model ibm-ai-platform/Bamba-9B-v2 --device 0
+```
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [torchao](../quantization/torchao) to only quantize the weights to int4.
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer, TorchAoConfig
+
+quantization_config = TorchAoConfig("int4_weight_only", group_size=128)
+tokenizer = AutoTokenizer.from_pretrained("ibm-ai-platform/Bamba-9B-v2")
+model = AutoModelForCausalLM.from_pretrained(
+   "ibm-ai-platform/Bamba-9B-v2",
+   quantization_config=quantization_config,
+   device_map="auto",
+   attn_implementation="sdpa"
+)
+
+inputs = tokenizer("Plants create energy through a process known as", return_tensors="pt").to("cuda")
+output = model.generate(**inputs)
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+## Notes
+
+- Bamba supports padding-free training which concatenates distinct training examples while still processing inputs as separate batches. It can significantly accelerate inference by [~2x](https://github.com/huggingface/transformers/pull/35861#issue-2807873129) (depending on model and data distribution) and reduce memory-usage if there are examples of varying lengths by avoiding unnecessary compute and memory overhead from padding tokens.
+
+  Padding-free training requires the `flash-attn`, `mamba-ssm`, and `causal-conv1d` packages and the following arguments must be passed to the model in addition to `input_ids` and `labels`.
+
+  - `position_ids: torch.LongTensor`: the position index of each token in each sequence.
+  - `seq_idx: torch.IntTensor`: the index of each sequence in the batch.
+  - Each of the [`FlashAttentionKwargs`]
+    - `cu_seq_lens_q: torch.LongTensor`: the cumulative sequence lengths of all queries.
+    - `cu_seq_lens_k: torch.LongTensor`: the cumulative sequence lengths of all keys.
+    - `max_length_q: int`: the longest query length in the batch.
+    - `max_length_k: int`: the longest key length in the batch.
+
+  The `attention_mask` inputs should not be provided. The [`DataCollatorWithFlattening`] programmatically generates the set of additional arguments above using `return_seq_idx=True` and `return_flash_attn_kwargs=True`. See the [Improving Hugging Face Training Efficiency Through Packing with Flash Attention](https://huggingface.co/blog/packing-with-FA2) blog post for additional information.
+
+  ```python
+  from transformers import DataCollatorWithFlattening
+
+  # Example of using padding-free training
+  data_collator = DataCollatorWithFlattening(
+      tokenizer=tokenizer,
+      return_seq_idx=True,
+      return_flash_attn_kwargs=True
+  )
+  ```
 
 ## BambaConfig
 
-| Model            | Params       | # Layers | Hidden Dim. | Attention Heads | GQA | KV Heads | Context Length |  Tied Embeddings |
-|-------------------|--------------|----------|-------------|-----------------|-----|----------|----------------|------------------|
-| Bamba  | 9B (9.78B)   | 32       | 4096        | 32              | Yes | 8        | 4096           | True |
-
 [[autodoc]] BambaConfig
 
-<!---
-## Usage Tips
-
-Tips:
-
-- The architecture is based on Mamba-2 models.
-
 ## BambaModel
 
 [[autodoc]] BambaModel
     - forward
--->
 
 ## BambaForCausalLM
 
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model = AutoModelForCausalLM.from_pretrained("ibm-fms/Bamba-9B")
-tokenizer = AutoTokenizer.from_pretrained("ibm-fms/Bamba-9B")
-
-message = ["Mamba is a snake with following properties  "]
-inputs = tokenizer(message, return_tensors='pt', return_token_type_ids=False)
-response = model.generate(**inputs, max_new_tokens=64)
-print(tokenizer.batch_decode(response, skip_special_tokens=True)[0])
-```
-
-
-## Padding-Free Training
-
-Bamba supports padding-free training in which distinct training examples can be concatenated
-together while nevertheless processing the inputs as though they belonged to separate batches. When
-the examples are of varying lengths, padding-free training can provide significant speed ups and
-memory savings compared to batching the examples together and using padding, as the unnecessary
-compute and memory due to padding is avoided entirely. The performance gains depend on factors such
-as the model and the data distribution, but throughput gains up to [~2x are commonly
-seen](https://github.com/huggingface/transformers/pull/35861#issue-2807873129).
-
-Using padding-free training with Bamba requires the `flash-attn`, `mamba-ssm`, and `causal-conv1d`
-packages, and the following arguments must be passed to the model in addition to `input_ids` and
-`labels`:
-* `position_ids: torch.LongTensor`: the position index of each token in each sequence.
-* `seq_idx: torch.IntTensor`: the index of each sequence in the batch.
-* Each of the [`FlashAttentionKwargs`]
-    * `cu_seq_lens_q: torch.LongTensor`: The cumulative sequence lengths of all queries.
-    * `cu_seq_lens_k: torch.LongTensor`: The cumulative sequence lengths of all keys.
-    * `max_length_q: int`: the longest query length in the batch.
-    * `max_length_k: int`: the longest key length in the batch.
-
-The `attention_mask` inputs should not be provided. The [`DataCollatorWithFlattening`] can be used
-to programmatically generate the above set of additional arguments using `return_seq_idx=True` and
-`return_flash_attn_kwargs=True`. See [this blog post](https://huggingface.co/blog/packing-with-FA2)
-for additional information.
-
-
 [[autodoc]] BambaForCausalLM
     - forward
-
-This HF implementation is contributed by [ani300](https://github.com/ani300) and [fabianlim](https://github.com/fabianlim).

docs/source/en/model_doc/bigbird_pegasus.md

@@ -14,59 +14,123 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# BigBirdPegasus
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+           <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
 </div>
 
-## Overview
+# BigBirdPegasus
 
-The BigBird model was proposed in [Big Bird: Transformers for Longer Sequences](https://huggingface.co/papers/2007.14062) by
-Zaheer, Manzil and Guruganesh, Guru and Dubey, Kumar Avinava and Ainslie, Joshua and Alberti, Chris and Ontanon,
-Santiago and Pham, Philip and Ravula, Anirudh and Wang, Qifan and Yang, Li and others. BigBird, is a sparse-attention
-based transformer which extends Transformer based models, such as BERT to much longer sequences. In addition to sparse
-attention, BigBird also applies global attention as well as random attention to the input sequence. Theoretically, it
-has been shown that applying sparse, global, and random attention approximates full attention, while being
-computationally much more efficient for longer sequences. As a consequence of the capability to handle longer context,
-BigBird has shown improved performance on various long document NLP tasks, such as question answering and
-summarization, compared to BERT or RoBERTa.
+[BigBirdPegasus](https://huggingface.co/papers/2007.14062) is an encoder-decoder (sequence-to-sequence) transformer model for long-input summarization. It extends the [BigBird](./big_bird) architecture with an additional pretraining objective borrowed from [Pegasus](./pegasus) called gap sequence generation (GSG). Whole sentences are masked and the model has to fill in the gaps in the document. BigBirdPegasus's ability to keep track of long contexts makes it effective at summarizing lengthy inputs, surpassing the performance of base Pegasus models.
 
-The abstract from the paper is the following:
+You can find all the original BigBirdPegasus checkpoints under the [Google](https://huggingface.co/google/models?search=bigbird-pegasus) organization.
 
-*Transformers-based models, such as BERT, have been one of the most successful deep learning models for NLP.
-Unfortunately, one of their core limitations is the quadratic dependency (mainly in terms of memory) on the sequence
-length due to their full attention mechanism. To remedy this, we propose, BigBird, a sparse attention mechanism that
-reduces this quadratic dependency to linear. We show that BigBird is a universal approximator of sequence functions and
-is Turing complete, thereby preserving these properties of the quadratic, full attention model. Along the way, our
-theoretical analysis reveals some of the benefits of having O(1) global tokens (such as CLS), that attend to the entire
-sequence as part of the sparse attention mechanism. The proposed sparse attention can handle sequences of length up to
-8x of what was previously possible using similar hardware. As a consequence of the capability to handle longer context,
-BigBird drastically improves performance on various NLP tasks such as question answering and summarization. We also
-propose novel applications to genomics data.*
+> [!TIP]
+> This model was contributed by [vasudevgupta](https://huggingface.co/vasudevgupta).
+>
+> Click on the BigBirdPegasus models in the right sidebar for more examples of how to apply BigBirdPegasus to different language tasks.
 
-The original code can be found [here](https://github.com/google-research/bigbird).
+The example below demonstrates how to summarize text with [`Pipeline`], [`AutoModel`], and from the command line.
 
-## Usage tips
+<hfoptions id="usage">
+<hfoption id="Pipeline">
 
-- For an in-detail explanation on how BigBird's attention works, see [this blog post](https://huggingface.co/blog/big-bird).
-- BigBird comes with 2 implementations: **original_full** & **block_sparse**. For the sequence length < 1024, using
-  **original_full** is advised as there is no benefit in using **block_sparse** attention.
-- The code currently uses window size of 3 blocks and 2 global blocks.
-- Sequence length must be divisible by block size.
-- Current implementation supports only **ITC**.
-- Current implementation doesn't support **num_random_blocks = 0**.
-- BigBirdPegasus uses the [PegasusTokenizer](https://github.com/huggingface/transformers/blob/main/src/transformers/models/pegasus/tokenization_pegasus.py).
-- BigBird is a model with absolute position embeddings so it's usually advised to pad the inputs on the right rather than
-  the left.
+```py
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="summarization",
+    model="google/bigbird-pegasus-large-arxiv",
+    torch_dtype=torch.float32,
+    device=0
+)
+pipeline("""Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle.""")
+```
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "google/bigbird-pegasus-large-arxiv"
+)
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "google/bigbird-pegasus-large-arxiv",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+)
+
+input_text = """Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle."""
+input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
+
+output = model.generate(**input_ids, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+</hfoption>
+<hfoption id="transformers-cli">
+
+```bash
+echo -e "Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet. Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts." | transformers-cli run --task summarization --model google/bigbird-pegasus-large-arxiv --device 0
+```
+
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [bitsandbytes](../quantization/bitsandbytes) to only quantize the weights to int4.
+
+```py
+import torch
+from transformers import BitsAndBytesConfig, AutoModelForSeq2SeqLM, AutoTokenizer
+
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.bfloat16,
+    bnb_4bit_quant_type="nf4"
+)
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "google/bigbird-pegasus-large-arxiv",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    quantization_config=quantization_config
+)
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "google/bigbird-pegasus-large-arxiv"
+)
+
+input_text = """Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle."""
+input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
+
+output = model.generate(**input_ids, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+## Notes
+
+- BigBirdPegasus also uses the [`PegasusTokenizer`].
+- Inputs should be padded on the right because BigBird uses absolute position embeddings.
+- BigBirdPegasus supports `original_full` and `block_sparse` attention. If the input sequence length is less than 1024, it is recommended to use `original_full` since sparse patterns don't offer much benefit for smaller inputs.
+- The current implementation uses window size of 3 blocks and 2 global blocks, only supports the ITC-implementation, and doesn't support `num_random_blocks=0`.
+- The sequence length must be divisible by the block size.
 
 ## Resources
 
-- [Text classification task guide](../tasks/sequence_classification)
-- [Question answering task guide](../tasks/question_answering)
-- [Causal language modeling task guide](../tasks/language_modeling)
-- [Translation task guide](../tasks/translation)
-- [Summarization task guide](../tasks/summarization)
+Read the [Understanding BigBird's Block Sparse Attention](https://huggingface.co/blog/big-bird) blog post for more details about how BigBird's attention works.
 
 ## BigBirdPegasusConfig
 

docs/source/en/model_doc/blip.md

@@ -14,35 +14,76 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# BLIP
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-<img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+    </div>
 </div>
 
-## Overview
+# BLIP
 
-The BLIP model was proposed in [BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation](https://huggingface.co/papers/2201.12086) by Junnan Li, Dongxu Li, Caiming Xiong, Steven Hoi.
+[BLIP](https://huggingface.co/papers/2201.12086) (Bootstrapped Language-Image Pretraining) is a vision-language pretraining (VLP) framework designed for *both* understanding and generation tasks. Most existing pretrained models are only good at one or the other. It uses a captioner to generate captions and a filter to remove the noisy captions. This increases training data quality and more effectively uses the messy web data.
 
-BLIP is a model that is able to perform various multi-modal tasks including:
-- Visual Question Answering 
-- Image-Text retrieval (Image-text matching)
-- Image Captioning
 
-The abstract from the paper is the following:
+You can find all the original BLIP checkpoints under the [BLIP](https://huggingface.co/collections/Salesforce/blip-models-65242f40f1491fbf6a9e9472) collection.
 
-*Vision-Language Pre-training (VLP) has advanced the performance for many vision-language tasks. 
-However, most existing pre-trained models only excel in either understanding-based tasks or generation-based tasks. Furthermore, performance improvement has been largely achieved by scaling up the dataset with noisy image-text pairs collected from the web, which is a suboptimal source of supervision. In this paper, we propose BLIP, a new VLP framework which transfers flexibly to both vision-language understanding and generation tasks. BLIP effectively utilizes the noisy web data by bootstrapping the captions, where a captioner generates synthetic captions and a filter removes the noisy ones. We achieve state-of-the-art results on a wide range of vision-language tasks, such as image-text retrieval (+2.7% in average recall@1), image captioning (+2.8% in CIDEr), and VQA (+1.6% in VQA score). BLIP also demonstrates strong generalization ability when directly transferred to videolanguage tasks in a zero-shot manner. Code, models, and datasets are released.*
+> [!TIP]
+> This model was contributed by [ybelkada](https://huggingface.co/ybelkada).
+> 
+> Click on the BLIP models in the right sidebar for more examples of how to apply BLIP to different vision language tasks.
 
-![BLIP.gif](https://cdn-uploads.huggingface.co/production/uploads/1670928184033-62441d1d9fdefb55a0b7d12c.gif)
+The example below demonstrates how to visual question answering with [`Pipeline`] or the [`AutoModel`] class.
 
-This model was contributed by [ybelkada](https://huggingface.co/ybelkada).
-The original code can be found [here](https://github.com/salesforce/BLIP).
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```python
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="visual-question-answering",
+    model="Salesforce/blip-vqa-base",
+    torch_dtype=torch.float16,
+    device=0
+)
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
+pipeline(question="What is the weather in this image?", image=url)
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```python
+import requests
+import torch
+from PIL import Image
+from transformers import AutoProcessor, AutoModelForVisualQuestionAnswering
+
+processor = AutoProcessor.from_pretrained("Salesforce/blip-vqa-base")
+model = AutoModelForVisualQuestionAnswering.from_pretrained(
+    "Salesforce/blip-vqa-base", 
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
+image = Image.open(requests.get(url, stream=True).raw)
+
+question = "What is the weather in this image?"
+inputs = processor(images=image, text=question, return_tensors="pt").to("cuda", torch.float16)
+
+output = model.generate(**inputs)
+processor.batch_decode(output, skip_special_tokens=True)[0]
+```
+
+</hfoption>
+</hfoptions>
 
 ## Resources
 
-- [Jupyter notebook](https://github.com/huggingface/notebooks/blob/main/examples/image_captioning_blip.ipynb) on how to fine-tune BLIP for image captioning on a custom dataset
+Refer to this [notebook](https://github.com/huggingface/notebooks/blob/main/examples/image_captioning_blip.ipynb) to learn how to fine-tune BLIP for image captioning on a custom dataset.
 
 ## BlipConfig
 

docs/source/en/model_doc/bros.md

@@ -62,11 +62,11 @@ def make_box_first_token_mask(bboxes, words, tokenizer, max_seq_length=512):
 
     box_first_token_mask = np.zeros(max_seq_length, dtype=np.bool_)
 
-    # encode(tokenize) each word from words (List[str])
-    input_ids_list: List[List[int]] = [tokenizer.encode(e, add_special_tokens=False) for e in words]
+    # encode(tokenize) each word from words (list[str])
+    input_ids_list: list[list[int]] = [tokenizer.encode(e, add_special_tokens=False) for e in words]
 
     # get the length of each box
-    tokens_length_list: List[int] = [len(l) for l in input_ids_list]
+    tokens_length_list: list[int] = [len(l) for l in input_ids_list]
 
     box_end_token_indices = np.array(list(itertools.accumulate(tokens_length_list)))
     box_start_token_indices = box_end_token_indices - np.array(tokens_length_list)

docs/source/en/model_doc/chameleon.md

@@ -191,6 +191,11 @@ model = ChameleonForConditionalGeneration.from_pretrained(
 [[autodoc]] ChameleonImageProcessor
     - preprocess
 
+## ChameleonImageProcessorFast
+
+[[autodoc]] ChameleonImageProcessorFast
+    - preprocess
+
 ## ChameleonVQVAE
 
 [[autodoc]] ChameleonVQVAE

docs/source/en/model_doc/cohere.md

@@ -3,6 +3,7 @@
         <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/cohere2.md

@@ -4,6 +4,7 @@
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/cvt.md

@@ -14,49 +14,77 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Convolutional Vision Transformer (CvT)
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-<img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+    </div>
 </div>
 
-## Overview
+# Convolutional Vision Transformer (CvT)
 
-The CvT model was proposed in [CvT: Introducing Convolutions to Vision Transformers](https://huggingface.co/papers/2103.15808) by Haiping Wu, Bin Xiao, Noel Codella, Mengchen Liu, Xiyang Dai, Lu Yuan and Lei Zhang. The Convolutional vision Transformer (CvT) improves the [Vision Transformer (ViT)](vit) in performance and efficiency by introducing convolutions into ViT to yield the best of both designs.
+Convolutional Vision Transformer (CvT) is a model that combines the strengths of convolutional neural networks (CNNs) and Vision transformers for the computer vision tasks. It introduces convolutional layers into the vision transformer architecture, allowing it to capture local patterns in images while maintaining the global context provided by self-attention mechanisms.
 
-The abstract from the paper is the following:
+You can find all the CvT checkpoints under the [Microsoft](https://huggingface.co/microsoft?search_models=cvt) organization.
 
-*We present in this paper a new architecture, named Convolutional vision Transformer (CvT), that improves Vision Transformer (ViT) 
-in performance and efficiency by introducing convolutions into ViT to yield the best of both designs. This is accomplished through 
-two primary modifications: a hierarchy of Transformers containing a new convolutional token embedding, and a convolutional Transformer 
-block leveraging a convolutional projection. These changes introduce desirable properties of convolutional neural networks (CNNs) 
-to the ViT architecture (\ie shift, scale, and distortion invariance) while maintaining the merits of Transformers (\ie dynamic attention, 
-global context, and better generalization). We validate CvT by conducting extensive experiments, showing that this approach achieves 
-state-of-the-art performance over other Vision Transformers and ResNets on ImageNet-1k, with fewer parameters and lower FLOPs. In addition, 
-performance gains are maintained when pretrained on larger datasets (\eg ImageNet-22k) and fine-tuned to downstream tasks. Pre-trained on 
-ImageNet-22k, our CvT-W24 obtains a top-1 accuracy of 87.7\% on the ImageNet-1k val set. Finally, our results show that the positional encoding, 
-a crucial component in existing Vision Transformers, can be safely removed in our model, simplifying the design for higher resolution vision tasks.*
+> [!TIP]
+> This model was contributed by [anujunj](https://huggingface.co/anugunj).
+> 
+> Click on the CvT models in the right sidebar for more examples of how to apply CvT to different computer vision tasks.
 
-This model was contributed by [anugunj](https://huggingface.co/anugunj). The original code can be found [here](https://github.com/microsoft/CvT).
+The example below demonstrates how to classify an image with [`Pipeline`] or the [`AutoModel`] class.
 
-## Usage tips
+<hfoptions id="usage">
+<hfoption id="Pipeline">
 
-- CvT models are regular Vision Transformers, but trained with convolutions. They outperform the [original model (ViT)](vit) when fine-tuned on ImageNet-1K and CIFAR-100.
-- You can check out demo notebooks regarding inference as well as fine-tuning on custom data [here](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/VisionTransformer) (you can just replace [`ViTFeatureExtractor`] by [`AutoImageProcessor`] and [`ViTForImageClassification`] by [`CvtForImageClassification`]).
-- The available checkpoints are either (1) pre-trained on [ImageNet-22k](http://www.image-net.org/) (a collection of 14 million images and 22k classes) only, (2) also fine-tuned on ImageNet-22k or (3) also fine-tuned on [ImageNet-1k](http://www.image-net.org/challenges/LSVRC/2012/) (also referred to as ILSVRC 2012, a collection of 1.3 million
-  images and 1,000 classes).
+```py
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="image-classification",
+    model="microsoft/cvt-13",
+    torch_dtype=torch.float16,
+    device=0 
+)
+pipeline(images="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg")
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+import requests
+from PIL import Image
+from transformers import AutoModelForImageClassification, AutoImageProcessor
+
+image_processor = AutoImageProcessor.from_pretrained("microsoft/cvt-13")
+model = AutoModelForImageClassification.from_pretrained(
+    "microsoft/cvt-13",
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
+image = Image.open(requests.get(url, stream=True).raw)
+inputs = image_processor(image, return_tensors="pt").to("cuda")
+
+with torch.no_grad():
+  logits = model(**inputs).logits
+predicted_class_id = logits.argmax(dim=-1).item()
+
+class_labels = model.config.id2label
+predicted_class_label = class_labels[predicted_class_id]
+print(f"The predicted class label is: {predicted_class_label}")
+```
+
+</hfoption>
+</hfoptions>
 
 ## Resources
 
-A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with CvT.
-
-<PipelineTag pipeline="image-classification"/>
-
-- [`CvtForImageClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/image-classification) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/image_classification.ipynb).
-- See also: [Image classification task guide](../tasks/image_classification)
-
-If you're interested in submitting a resource to be included here, please feel free to open a Pull Request and we'll review it! The resource should ideally demonstrate something new instead of duplicating an existing resource.
+Refer to this set of ViT [notebooks](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/VisionTransformer) for examples of inference and fine-tuning on custom datasets. Replace [`ViTFeatureExtractor`] and [`ViTForImageClassification`] in these notebooks with [`AutoImageProcessor`] and [`CvtForImageClassification`].
 
 ## CvtConfig
 

docs/source/en/model_doc/deberta-v2.md

@@ -14,66 +14,111 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# DeBERTa-v2
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-<img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+           <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white" >
+           <img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+    </div>
 </div>
 
-## Overview
 
-The DeBERTa model was proposed in [DeBERTa: Decoding-enhanced BERT with Disentangled Attention](https://huggingface.co/papers/2006.03654) by Pengcheng He, Xiaodong Liu, Jianfeng Gao, Weizhu Chen It is based on Google's
-BERT model released in 2018 and Facebook's RoBERTa model released in 2019.
+# DeBERTa-v2
 
-It builds on RoBERTa with disentangled attention and enhanced mask decoder training with half of the data used in
-RoBERTa.
+[DeBERTa-v2](https://huggingface.co/papers/2006.03654) improves on the original [DeBERTa](./deberta) architecture by using a SentencePiece-based tokenizer and a new vocabulary size of 128K. It also adds an additional convolutional layer within the first transformer layer to better learn local dependencies of input tokens. Finally, the position projection and content projection matrices are shared in the attention layer to reduce the number of parameters.
 
-The abstract from the paper is the following:
-
-*Recent progress in pre-trained neural language models has significantly improved the performance of many natural
-language processing (NLP) tasks. In this paper we propose a new model architecture DeBERTa (Decoding-enhanced BERT with
-disentangled attention) that improves the BERT and RoBERTa models using two novel techniques. The first is the
-disentangled attention mechanism, where each word is represented using two vectors that encode its content and
-position, respectively, and the attention weights among words are computed using disentangled matrices on their
-contents and relative positions. Second, an enhanced mask decoder is used to replace the output softmax layer to
-predict the masked tokens for model pretraining. We show that these two techniques significantly improve the efficiency
-of model pretraining and performance of downstream tasks. Compared to RoBERTa-Large, a DeBERTa model trained on half of
-the training data performs consistently better on a wide range of NLP tasks, achieving improvements on MNLI by +0.9%
-(90.2% vs. 91.1%), on SQuAD v2.0 by +2.3% (88.4% vs. 90.7%) and RACE by +3.6% (83.2% vs. 86.8%). The DeBERTa code and
-pre-trained models will be made publicly available at https://github.com/microsoft/DeBERTa.*
+You can find all the original [DeBERTa-v2] checkpoints under the [Microsoft](https://huggingface.co/microsoft?search_models=deberta-v2) organization.
 
 
-The following information is visible directly on the [original implementation
-repository](https://github.com/microsoft/DeBERTa). DeBERTa v2 is the second version of the DeBERTa model. It includes
-the 1.5B model used for the SuperGLUE single-model submission and achieving 89.9, versus human baseline 89.8. You can
-find more details about this submission in the authors'
-[blog](https://www.microsoft.com/en-us/research/blog/microsoft-deberta-surpasses-human-performance-on-the-superglue-benchmark/)
+> [!TIP]
+> This model was contributed by [Pengcheng He](https://huggingface.co/DeBERTa).
+>
+> Click on the DeBERTa-v2 models in the right sidebar for more examples of how to apply DeBERTa-v2 to different language tasks.
 
-New in v2:
+The example below demonstrates how to classify text with [`Pipeline`] or the [`AutoModel`] class.
 
-- **Vocabulary** In v2 the tokenizer is changed to use a new vocabulary of size 128K built from the training data.
-  Instead of a GPT2-based tokenizer, the tokenizer is now
-  [sentencepiece-based](https://github.com/google/sentencepiece) tokenizer.
-- **nGiE(nGram Induced Input Encoding)** The DeBERTa-v2 model uses an additional convolution layer aside with the first
-  transformer layer to better learn the local dependency of input tokens.
-- **Sharing position projection matrix with content projection matrix in attention layer** Based on previous
-  experiments, this can save parameters without affecting the performance.
-- **Apply bucket to encode relative positions** The DeBERTa-v2 model uses log bucket to encode relative positions
-  similar to T5.
-- **900M model & 1.5B model** Two additional model sizes are available: 900M and 1.5B, which significantly improves the
-  performance of downstream tasks.
+<hfoptions id="usage">
+<hfoption id="Pipeline">
 
-This model was contributed by [DeBERTa](https://huggingface.co/DeBERTa). This model TF 2.0 implementation was
-contributed by [kamalkraj](https://huggingface.co/kamalkraj). The original code can be found [here](https://github.com/microsoft/DeBERTa).
+```py
+import torch
+from transformers import pipeline
 
-## Resources
+pipeline = pipeline(
+    task="text-classification",
+    model="microsoft/deberta-v2-xlarge-mnli",
+    device=0,
+    torch_dtype=torch.float16
+)
+result = pipeline("DeBERTa-v2 is great at understanding context!")
+print(result)
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "microsoft/deberta-v2-xlarge-mnli"
+)
+model = AutoModelForSequenceClassification.from_pretrained(
+    "microsoft/deberta-v2-xlarge-mnli",
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+
+inputs = tokenizer("DeBERTa-v2 is great at understanding context!", return_tensors="pt").to("cuda")
+outputs = model(**inputs)
+
+logits = outputs.logits
+predicted_class_id = logits.argmax().item()
+predicted_label = model.config.id2label[predicted_class_id]
+print(f"Predicted label: {predicted_label}")
+
+```
+
+</hfoption>
+
+<hfoption id="transformers CLI">
+
+```bash
+echo -e "DeBERTa-v2 is great at understanding context!" | transformers-cli run --task fill-mask --model microsoft/deberta-v2-xlarge-mnli --device 0
+```
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [bitsandbytes quantization](../quantization/bitsandbytes) to only quantize the weights to 4-bit.
+
+```py
+from transformers import AutoModelForSequenceClassification, AutoTokenizer, BitsAndBytesConfig
+
+model_id = "microsoft/deberta-v2-xlarge-mnli"
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_compute_dtype="float16",
+    bnb_4bit_use_double_quant=True,
+)
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+model = AutoModelForSequenceClassification.from_pretrained(
+    model_id,
+    quantization_config=quantization_config,
+    torch_dtype="float16"
+)
+
+inputs = tokenizer("DeBERTa-v2 is great at understanding context!", return_tensors="pt").to("cuda")
+outputs = model(**inputs)
+logits = outputs.logits
+predicted_class_id = logits.argmax().item()
+predicted_label = model.config.id2label[predicted_class_id]
+print(f"Predicted label: {predicted_label}")
+
+```
 
-- [Text classification task guide](../tasks/sequence_classification)
-- [Token classification task guide](../tasks/token_classification)
-- [Question answering task guide](../tasks/question_answering)
-- [Masked language modeling task guide](../tasks/masked_language_modeling)
-- [Multiple choice task guide](../tasks/multiple_choice)
 
 ## DebertaV2Config
 

docs/source/en/model_doc/detr.md

@@ -149,7 +149,7 @@ As a summary, consider the following table:
 | **Description** | Predicting bounding boxes and class labels around objects in an image | Predicting masks around objects (i.e. instances) in an image | Predicting masks around both objects (i.e. instances) as well as "stuff" (i.e. background things like trees and roads) in an image |
 | **Model** | [`~transformers.DetrForObjectDetection`] | [`~transformers.DetrForSegmentation`] | [`~transformers.DetrForSegmentation`] |
 | **Example dataset** | COCO detection | COCO detection, COCO panoptic | COCO panoptic  |                                                                        |
-| **Format of annotations to provide to**  [`~transformers.DetrImageProcessor`] | {'image_id': `int`, 'annotations': `List[Dict]`} each Dict being a COCO object annotation  | {'image_id': `int`, 'annotations': `List[Dict]`}  (in case of COCO detection) or {'file_name': `str`, 'image_id': `int`, 'segments_info': `List[Dict]`} (in case of COCO panoptic) | {'file_name': `str`, 'image_id': `int`, 'segments_info': `List[Dict]`} and masks_path (path to directory containing PNG files of the masks) |
+| **Format of annotations to provide to**  [`~transformers.DetrImageProcessor`] | {'image_id': `int`, 'annotations': `list[Dict]`} each Dict being a COCO object annotation  | {'image_id': `int`, 'annotations': `list[Dict]`}  (in case of COCO detection) or {'file_name': `str`, 'image_id': `int`, 'segments_info': `list[Dict]`} (in case of COCO panoptic) | {'file_name': `str`, 'image_id': `int`, 'segments_info': `list[Dict]`} and masks_path (path to directory containing PNG files of the masks) |
 | **Postprocessing** (i.e. converting the output of the model to Pascal VOC format) | [`~transformers.DetrImageProcessor.post_process`] | [`~transformers.DetrImageProcessor.post_process_segmentation`] | [`~transformers.DetrImageProcessor.post_process_segmentation`], [`~transformers.DetrImageProcessor.post_process_panoptic`] |
 | **evaluators** | `CocoEvaluator` with `iou_types="bbox"` | `CocoEvaluator` with `iou_types="bbox"` or `"segm"` | `CocoEvaluator` with `iou_tupes="bbox"` or `"segm"`, `PanopticEvaluator` |
 

docs/source/en/model_doc/dia.md

@@ -0,0 +1,162 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Dia
+
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
+</div>
+
+## Overview
+
+Dia is an opensource text-to-speech (TTS) model (1.6B parameters) developed by [Nari Labs](https://huggingface.co/nari-labs).
+It can generate highly realistic dialogue from transcript including nonverbal communications such as laughter and coughing.
+Furthermore, emotion and tone control is also possible via audio conditioning (voice cloning).
+
+**Model Architecture:**
+Dia is an encoder-decoder transformer based on the original transformer architecture. However, some more modern features such as
+rotational positional embeddings (RoPE) are also included. For its text portion (encoder), a byte tokenizer is utilized while
+for the audio portion (decoder), a pretrained codec model [DAC](./dac.md) is used - DAC encodes speech into discrete codebook
+tokens and decodes them back into audio.
+
+## Usage Tips
+
+### Generation with Text
+
+```python
+from transformers import AutoProcessor, DiaForConditionalGeneration
+
+torch_device = "cuda"
+model_checkpoint = "nari-labs/Dia-1.6B-0626"
+
+text = ["[S1] Dia is an open weights text to dialogue model."]
+processor = AutoProcessor.from_pretrained(model_checkpoint)
+inputs = processor(text=text, padding=True, return_tensors="pt").to(torch_device)
+
+model = DiaForConditionalGeneration.from_pretrained(model_checkpoint).to(torch_device)
+outputs = model.generate(**inputs, max_new_tokens=256)  # corresponds to around ~2s
+
+# save audio to a file
+outputs = processor.batch_decode(outputs)
+processor.save_audio(outputs, "example.wav")
+
+```
+
+### Generation with Text and Audio (Voice Cloning)
+
+```python
+from datasets import load_dataset, Audio
+from transformers import AutoProcessor, DiaForConditionalGeneration
+
+torch_device = "cuda"
+model_checkpoint = "nari-labs/Dia-1.6B-0626"
+
+ds = load_dataset("hf-internal-testing/dailytalk-dummy", split="train")
+ds = ds.cast_column("audio", Audio(sampling_rate=44100))
+audio = ds[-1]["audio"]["array"]
+# text is a transcript of the audio + additional text you want as new audio
+text = ["[S1] I know. It's going to save me a lot of money, I hope. [S2] I sure hope so for you."]
+
+processor = AutoProcessor.from_pretrained(model_checkpoint)
+inputs = processor(text=text, audio=audio, padding=True, return_tensors="pt").to(torch_device)
+prompt_len = processor.get_audio_prompt_len(inputs["decoder_attention_mask"])
+
+model = DiaForConditionalGeneration.from_pretrained(model_checkpoint).to(torch_device)
+outputs = model.generate(**inputs, max_new_tokens=256)  # corresponds to around ~2s
+
+# retrieve actually generated audio and save to a file
+outputs = processor.batch_decode(outputs, audio_prompt_len=prompt_len)
+processor.save_audio(outputs, "example_with_audio.wav")
+```
+
+### Training
+
+```python
+from datasets import load_dataset, Audio
+from transformers import AutoProcessor, DiaForConditionalGeneration
+
+torch_device = "cuda"
+model_checkpoint = "nari-labs/Dia-1.6B-0626"
+
+ds = load_dataset("hf-internal-testing/dailytalk-dummy", split="train")
+ds = ds.cast_column("audio", Audio(sampling_rate=44100))
+audio = ds[-1]["audio"]["array"]
+# text is a transcript of the audio
+text = ["[S1] I know. It's going to save me a lot of money, I hope."]
+
+processor = AutoProcessor.from_pretrained(model_checkpoint)
+inputs = processor(
+    text=text,
+    audio=audio,
+    generation=False,
+    output_labels=True,
+    padding=True,
+    return_tensors="pt"
+).to(torch_device)
+
+model = DiaForConditionalGeneration.from_pretrained(model_checkpoint).to(torch_device)
+out = model(**inputs)
+out.loss.backward()
+```
+
+
+This model was contributed by [Jaeyong Sung](https://huggingface.co/buttercrab), [Arthur Zucker](https://huggingface.co/ArthurZ),
+and [Anton Vlasjuk](https://huggingface.co/AntonV). The original code can be found [here](https://github.com/nari-labs/dia/).
+
+
+## DiaConfig
+
+[[autodoc]] DiaConfig
+
+## DiaDecoderConfig
+
+[[autodoc]] DiaDecoderConfig
+
+## DiaEncoderConfig
+
+[[autodoc]] DiaEncoderConfig
+
+## DiaTokenizer
+
+[[autodoc]] DiaTokenizer
+    - __call__
+
+## DiaFeatureExtractor
+
+[[autodoc]] DiaFeatureExtractor
+    - __call__
+
+## DiaProcessor
+
+[[autodoc]] DiaProcessor
+    - __call__
+    - batch_decode
+    - decode
+
+## DiaModel
+
+[[autodoc]] DiaModel
+    - forward
+
+## DiaForConditionalGeneration
+
+[[autodoc]] DiaForConditionalGeneration
+    - forward
+    - generate

docs/source/en/model_doc/doge.md

@@ -0,0 +1,103 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Doge
+
+
+## Overview
+
+Doge is a series of small language models based on the [Doge](https://github.com/SmallDoges/small-doge) architecture, aiming to combine the advantages of state-space and self-attention algorithms, calculate dynamic masks from cached value states using the zero-order hold method, and solve the problem of existing mainstream language models getting lost in context. It uses the `wsd_scheduler` scheduler to pre-train on the `smollm-corpus`, and can continue training on new datasets or add sparse activation feedforward networks from stable stage checkpoints.
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/refs%2Fpr%2F426/transformers/model_doc/doge_architecture.png" alt="drawing" width="600"/>
+
+As shown in the figure below, the sequence transformation part of the Doge architecture uses `Dynamic Mask Attention`, which can be understood as using self-attention related to value states during training, and using state-space without past state decay during inference, to solve the problem of existing Transformers or SSMs getting lost in long text. The state transformation part of Doge uses `Cross Domain Mixture of Experts`, which consists of dense linear layers and sparse embedding layers, and can additionally increase sparse parameters to continue training from dense weight checkpoints without retraining the entire model, thereby reducing the cost of continuous iteration of the model. In addition, Doge also uses `RMSNorm` and `Residual` with learnable parameters to adapt the gradient range of deep models.
+
+Checkout all Doge model checkpoints [here](https://huggingface.co/collections/SmallDoge/doge-slm-679cc991f027c4a3abbded4a).
+
+
+## Usage
+
+<details>
+<summary>Using Doge-Base for text generation</summary>
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+tokenizer = AutoTokenizer.from_pretrained("SmallDoge/Doge-20M")
+model = AutoModelForCausalLM.from_pretrained("SmallDoge/Doge-20M")
+inputs = tokenizer("Hey how are you doing?", return_tensors="pt")
+
+outputs = model.generate(**inputs, max_new_tokens=100)
+print(tokenizer.batch_decode(outputs))
+```
+</details>
+
+<details>
+<summary>Using Doge-Instruct for question answering</summary>
+
+```python
+from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig, TextStreamer
+
+tokenizer = AutoTokenizer.from_pretrained("SmallDoge/Doge-20M-Instruct")
+model = AutoModelForCausalLM.from_pretrained("SmallDoge/Doge-20M-Instruct")
+
+generation_config = GenerationConfig(
+      max_new_tokens=100, 
+      use_cache=True, 
+      do_sample=True, 
+      temperature=0.8, 
+      top_p=0.9,
+      repetition_penalty=1.0
+)
+steamer = TextStreamer(tokenizer=tokenizer, skip_prompt=True)
+
+prompt = "Hi, how are you doing today?"
+conversation = [
+      {"role": "user", "content": prompt}
+]
+inputs = tokenizer.apply_chat_template(
+    conversation=conversation,
+    tokenize=True,
+    return_tensors="pt",
+)
+
+outputs = model.generate(
+    inputs, 
+    tokenizer=tokenizer,
+    generation_config=generation_config, 
+    streamer=steamer
+)
+```
+</details>
+
+## DogeConfig
+
+[[autodoc]] DogeConfig
+
+## DogeModel
+
+[[autodoc]] DogeModel
+    - forward
+
+## DogeForCausalLM
+
+[[autodoc]] DogeForCausalLM
+    - forward
+
+## DogeForSequenceClassification
+
+[[autodoc]] DogeForSequenceClassification
+    - forward

docs/source/en/model_doc/dots1.md

@@ -0,0 +1,40 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# dots.llm1
+
+## Overview
+
+The `dots.llm1` model was proposed in [dots.llm1 technical report](https://www.arxiv.org/pdf/2506.05767) by rednote-hilab team.
+
+The abstract from the report is the following:
+
+*Mixture of Experts (MoE) models have emerged as a promising paradigm for scaling language models efficiently by activating only a subset of parameters for each input token. In this report, we present dots.llm1, a large-scale MoE model that activates 14B parameters out of a total of 142B parameters, delivering performance on par with state-of-the-art models while reducing training and inference costs. Leveraging our meticulously crafted and efficient data processing pipeline, dots.llm1 achieves performance comparable to Qwen2.5-72B after pretraining on high-quality corpus and post-training to fully unlock its capabilities. Notably, no synthetic data is used during pretraining. To foster further research, we open-source intermediate training checkpoints spanning the entire training process, providing valuable insights into the learning dynamics of large language models.*
+
+
+## Dots1Config
+
+[[autodoc]] Dots1Config
+
+## Dots1Model
+
+[[autodoc]] Dots1Model
+    - forward
+
+## Dots1ForCausalLM
+
+[[autodoc]] Dots1ForCausalLM
+    - forward

docs/source/en/model_doc/dpt.md

@@ -78,7 +78,13 @@ If you're interested in submitting a resource to be included here, please feel f
 
 [[autodoc]] DPTImageProcessor
     - preprocess
+
+## DPTImageProcessorFast
+
+[[autodoc]] DPTImageProcessorFast
+    - preprocess
     - post_process_semantic_segmentation
+    - post_process_depth_estimation
 
 ## DPTModel
 

docs/source/en/model_doc/eomt.md

@@ -0,0 +1,210 @@
+<!--Copyright 2025 Mobile Perception Systems Lab at TU/e and The HuggingFace Inc. 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.
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+-->
+
+# EoMT
+
+<div class="flex flex-wrap space-x-1">
+<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+</div>
+
+## Overview
+
+The Encoder-only Mask Transformer (EoMT) model was introduced in the CVPR 2025 Highlight Paper [Your ViT is Secretly an Image Segmentation Model](https://www.tue-mps.org/eomt) by Tommie Kerssies, Niccolò Cavagnero, Alexander Hermans, Narges Norouzi, Giuseppe Averta, Bastian Leibe, Gijs Dubbelman, and Daan de Geus.
+EoMT reveals Vision Transformers can perform image segmentation efficiently without task-specific components.
+
+The abstract from the paper is the following:
+
+*Vision Transformers (ViTs) have shown remarkable performance and scalability across various computer vision tasks. To apply single-scale ViTs to image segmentation, existing methods adopt a convolutional adapter to generate multi-scale features, a pixel decoder to fuse these features, and a Transformer decoder that uses the fused features to make predictions. In this paper, we show that the inductive biases introduced by these task-specific components can instead be learned by the ViT itself, given sufficiently large models and extensive pre-training. Based on these findings, we introduce the Encoder-only Mask Transformer (EoMT), which repurposes the plain ViT architecture to conduct image segmentation. With large-scale models and pre-training, EoMT obtains a segmentation accuracy similar to state-of-the-art models that use task-specific components. At the same time, EoMT is significantly faster than these methods due to its architectural simplicity, e.g., up to 4x faster with ViT-L. Across a range of model sizes, EoMT demonstrates an optimal balance between segmentation accuracy and prediction speed, suggesting that compute resources are better spent on scaling the ViT itself rather than adding architectural complexity.*
+
+This model was contributed by [Yaswanth Gali](https://huggingface.co/yaswanthgali).
+The original code can be found [here](https://github.com/tue-mps/eomt).
+
+## Architecture Info
+
+The `EoMT` model uses a DINOv2-pretrained Vision Transformer with **register tokens** as its backbone. EoMT simplifies the segmentation pipeline by relying solely on the encoder, eliminating the need for task-specific decoders commonly used in prior approaches.
+
+Architecturally, EoMT introduces a small set of **learned queries** and a lightweight **mask prediction module**. These queries are injected into the final encoder blocks, enabling **joint attention** between image patches and object queries. During training, **masked attention** is applied to constrain each query to focus on its corresponding region—effectively mimicking cross-attention. This constraint is gradually phased out via a **mask annealing strategy**, allowing for **efficient, decoder-free inference** without compromising segmentation performance.
+
+<div style="text-align: center;">
+  <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/eomt_architecture.png"
+       alt="drawing" width="500"/>
+</div>
+
+
+The model supports semantic, instance, and panoptic segmentation using a unified architecture and task-specific post-processing.
+
+## Usage Examples
+
+Use the Hugging Face implementation of EoMT for inference with pre-trained models.
+
+### Semantic Segmentation
+
+The EoMT model performs semantic segmentation using sliding-window inference. The input image is resized such that the shorter side matches the target input size, then it is split into overlapping crops. Each crop is then passed through the model. After inference, the predicted logits from each crop are stitched back together and rescaled to the original image size to get the final segmentation mask.
+
+> **Note:**  
+> If you want to use a custom target size for **semantic segmentation**, specify it in the following format:  
+> `{"shortest_edge": 512}`  
+> Notice that `longest_edge` is not provided here — this is intentional. For semantic segmentation, images are typically **scaled so that the shortest edge is greater than or equal to the target size** hence longest_edge is not necessary.
+
+```python
+import matplotlib.pyplot as plt
+import requests
+import torch
+from PIL import Image
+
+from transformers import EomtForUniversalSegmentation, AutoImageProcessor
+
+
+model_id = "tue-mps/ade20k_semantic_eomt_large_512"
+processor = AutoImageProcessor.from_pretrained(model_id)
+model = EomtForUniversalSegmentation.from_pretrained(model_id)
+
+image = Image.open(requests.get("http://images.cocodataset.org/val2017/000000039769.jpg", stream=True).raw)
+
+inputs = processor(
+    images=image,
+    return_tensors="pt",
+)
+
+with torch.inference_mode():
+    outputs = model(**inputs)
+
+# Prepare the original image size in the format (height, width)
+target_sizes = [(image.height, image.width)]
+
+# Post-process the model outputs to get final segmentation prediction
+preds = processor.post_process_semantic_segmentation(
+    outputs,
+    target_sizes=target_sizes,
+)
+
+# Visualize the segmentation mask
+plt.imshow(preds[0])
+plt.axis("off")
+plt.title("Semantic Segmentation")
+plt.show()
+```
+
+### Instance Segmentation
+
+The EoMT model performs instance segmentation using padded inference. The input image is resized so that the longer side matches the target input size, and the shorter side is zero-padded to form a square. The resulting mask and class logits are combined through post-processing (adapted from Mask2Former) to produce a unified instance segmentation map, along with segment metadata like segment id, class labels and confidence scores.
+
+> **Note:**  
+> To use a custom target size, specify the size as a dictionary in the following format:  
+> `{"shortest_edge": 512, "longest_edge": 512}`  
+> For both instance and panoptic segmentation, input images will be **scaled and padded** to this target size.
+
+```python
+import matplotlib.pyplot as plt
+import requests
+import torch
+from PIL import Image
+
+from transformers import EomtForUniversalSegmentation, AutoImageProcessor
+
+
+model_id = "tue-mps/coco_instance_eomt_large_640"
+processor = AutoImageProcessor.from_pretrained(model_id)
+model = EomtForUniversalSegmentation.from_pretrained(model_id)
+
+image = Image.open(requests.get("http://images.cocodataset.org/val2017/000000039769.jpg", stream=True).raw)
+
+inputs = processor(
+    images=image,
+    return_tensors="pt",
+)
+
+with torch.inference_mode():
+    outputs = model(**inputs)
+
+# Prepare the original image size in the format (height, width)
+target_sizes = [(image.height, image.width)]
+
+# Post-process the model outputs to get final segmentation prediction
+preds = processor.post_process_instance_segmentation(
+    outputs,
+    target_sizes=target_sizes,
+)
+
+# Visualize the segmentation mask
+plt.imshow(preds[0]["segmentation"])
+plt.axis("off")
+plt.title("Instance Segmentation")
+plt.show()
+```
+
+### Panoptic Segmentation
+
+The EoMT model performs panoptic segmentation using the same padded inference strategy as in instance segmentation. After padding and normalization, the model predicts both thing (instances) and stuff (amorphous regions) classes. The resulting mask and class logits are combined through post-processing (adapted from Mask2Former) to produce a unified panoptic segmentation map, along with segment metadata like segment id, class labels and confidence scores.
+
+```python
+import matplotlib.pyplot as plt
+import requests
+import torch
+from PIL import Image
+
+from transformers import EomtForUniversalSegmentation, AutoImageProcessor
+
+
+model_id = "tue-mps/coco_panoptic_eomt_large_640"
+processor = AutoImageProcessor.from_pretrained(model_id)
+model = EomtForUniversalSegmentation.from_pretrained(model_id)
+
+image = Image.open(requests.get("http://images.cocodataset.org/val2017/000000039769.jpg", stream=True).raw)
+
+inputs = processor(
+    images=image,
+    return_tensors="pt",
+)
+
+with torch.inference_mode():
+    outputs = model(**inputs)
+
+# Prepare the original image size in the format (height, width)
+target_sizes = [(image.height, image.width)]
+
+# Post-process the model outputs to get final segmentation prediction
+preds = processor.post_process_panoptic_segmentation(
+    outputs,
+    target_sizes=target_sizes,
+)
+
+# Visualize the panoptic segmentation mask
+plt.imshow(preds[0]["segmentation"])
+plt.axis("off")
+plt.title("Panoptic Segmentation")
+plt.show()
+```
+
+## EomtImageProcessor
+
+[[autodoc]] EomtImageProcessor
+    - preprocess
+    - post_process_semantic_segmentation
+    - post_process_instance_segmentation
+    - post_process_panoptic_segmentation
+
+## EomtImageProcessorFast
+
+[[autodoc]] EomtImageProcessorFast
+    - preprocess
+    - post_process_semantic_segmentation
+    - post_process_instance_segmentation
+    - post_process_panoptic_segmentation
+
+## EomtConfig
+
+[[autodoc]] EomtConfig
+
+## EomtForUniversalSegmentation
+
+[[autodoc]] EomtForUniversalSegmentation
+    - forward

docs/source/en/model_doc/gemma.md

@@ -23,6 +23,7 @@ rendered properly in your Markdown viewer.
         ">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/gemma2.md

@@ -22,6 +22,7 @@ rendered properly in your Markdown viewer.
         ">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/gemma3n.md

@@ -0,0 +1,205 @@
+
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
+</div>
+
+# Gemma3n
+
+## Overview
+
+Gemma3n is a multimodal model with pretrained and instruction-tuned variants, available in E4B and E2B sizes. While
+large portions of the language model architecture are shared with prior Gemma releases, there are many new additions in
+this model, including [Alternating Updates][altup] (AltUp), [Learned Augmented Residual Layer][laurel] (LAuReL),
+[MatFormer][matformer], Per-Layer Embeddings (PLE), [Activation Sparsity with Statistical Top-k][spark-transformer], and KV cache sharing. The language model uses
+a similar attention pattern to [Gemma 3](./gemma3.md) with alternating 4 local sliding window self-attention layers for
+every global self-attention layer with a maximum context length of 32k tokens. Gemma 3n introduces
+[MobileNet v5][mobilenetv5] as the vision encoder, using a default resolution of 768x768 pixels, and adds a newly
+trained audio encoder based on the [Universal Speech Model][usm] (USM) architecture.
+
+The instruction-tuned variant was post-trained with knowledge distillation and reinforcement learning.
+
+You can find all the original Gemma 3n checkpoints under the [Gemma 3n][gemma3n-collection] release.
+
+> [!TIP]
+> Click on the Gemma 3n models in the right sidebar for more examples of how to apply Gemma to different vision, audio,
+> and language tasks.
+
+The example below demonstrates how to generate text based on an image with [`Pipeline`] or the [`AutoModel`] class.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```py
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="image-text-to-text",
+    model="google/gemma-3n-e4b",
+    device=0,
+    torch_dtype=torch.bfloat16
+)
+pipeline(
+    "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg",
+    text="<start_of_image> What is shown in this image?"
+)
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoProcessor, Gemma3nForConditionalGeneration
+
+model = Gemma3nForConditionalGeneration.from_pretrained(
+    "google/gemma-3n-e4b-it",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    attn_implementation="sdpa"
+)
+processor = AutoProcessor.from_pretrained(
+    "google/gemma-3n-e4b-it",
+    padding_side="left"
+)
+
+messages = [
+    {
+        "role": "system",
+        "content": [
+            {"type": "text", "text": "You are a helpful assistant."}
+        ]
+    },
+    {
+        "role": "user", "content": [
+            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"},
+            {"type": "text", "text": "What is shown in this image?"},
+        ]
+    },
+]
+inputs = processor.apply_chat_template(
+    messages,
+    tokenize=True,
+    return_dict=True,
+    return_tensors="pt",
+    add_generation_prompt=True,
+).to("cuda")
+
+output = model.generate(**inputs, max_new_tokens=50, cache_implementation="static")
+print(processor.decode(output[0], skip_special_tokens=True))
+```
+
+</hfoption>
+<hfoption id="transformers CLI">
+
+```bash
+echo -e "Plants create energy through a process known as" | transformers run --task text-generation --model google/gemma-3n-e2b --device 0
+```
+
+</hfoption>
+</hfoptions>
+
+## Notes
+
+-   Use [`Gemma3nForConditionalGeneration`] for image-audio-and-text, image-and-text, image-and-audio, audio-and-text,
+    image-only and audio-only inputs.
+-   Gemma 3n supports multiple images per input, but make sure the images are correctly batched before passing them to
+    the processor. Each batch should be a list of one or more images.
+
+    ```py
+    url_cow = "https://media.istockphoto.com/id/1192867753/photo/cow-in-berchida-beach-siniscola.jpg?s=612x612&w=0&k=20&c=v0hjjniwsMNfJSuKWZuIn8pssmD5h5bSN1peBd1CmH4="
+    url_cat = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
+
+    messages =[
+        {
+            "role": "system",
+            "content": [
+                {"type": "text", "text": "You are a helpful assistant."}
+            ]
+        },
+        {
+            "role": "user",
+            "content": [
+                {"type": "image", "url": url_cow},
+                {"type": "image", "url": url_cat},
+                {"type": "text", "text": "Which image is cuter?"},
+            ]
+        },
+    ]
+    ```
+-   Text passed to the processor should have a `<image_soft_token>` token wherever an image should be inserted.
+-   Gemma 3n accept at most one target audio clip per input, though multiple audio clips can be provided in few-shot
+    prompts, for example.
+-   Text passed to the processor should have a `<audio_soft_token>` token wherever an audio clip should be inserted.
+-   The processor has its own [`~ProcessorMixin.apply_chat_template`] method to convert chat messages to model inputs.
+
+## Gemma3nAudioFeatureExtractor
+
+[[autodoc]] Gemma3nAudioFeatureExtractor
+
+## Gemma3nProcessor
+
+[[autodoc]] Gemma3nProcessor
+
+## Gemma3nTextConfig
+
+[[autodoc]] Gemma3nTextConfig
+
+## Gemma3nVisionConfig
+
+[[autodoc]] Gemma3nVisionConfig
+
+## Gemma3nAudioConfig
+
+[[autodoc]] Gemma3nAudioConfig
+
+## Gemma3nConfig
+
+[[autodoc]] Gemma3nConfig
+
+## Gemma3nTextModel
+
+[[autodoc]] Gemma3nTextModel
+    - forward
+
+## Gemma3nModel
+
+[[autodoc]] Gemma3nModel
+    - forward
+
+## Gemma3nForCausalLM
+
+[[autodoc]] Gemma3nForCausalLM
+    - forward
+
+## Gemma3nForConditionalGeneration
+
+[[autodoc]] Gemma3nForConditionalGeneration
+    - forward
+
+[altup]: https://proceedings.neurips.cc/paper_files/paper/2023/hash/f2059277ac6ce66e7e5543001afa8bb5-Abstract-Conference.html
+[attention-mask-viz]: https://github.com/huggingface/transformers/blob/beb9b5b02246b9b7ee81ddf938f93f44cfeaad19/src/transformers/utils/attention_visualizer.py#L139
+[gemma3n-collection]: https://huggingface.co/collections/google/gemma-3n
+[laurel]: https://arxiv.org/abs/2411.07501
+[matformer]: https://arxiv.org/abs/2310.07707
+[spark-transformer]: https://arxiv.org/abs/2506.06644
+[usm]: https://arxiv.org/abs/2303.01037

docs/source/en/model_doc/glm.md

@@ -20,6 +20,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/glm4.md

@@ -18,7 +18,37 @@ rendered properly in your Markdown viewer.
 
 ## Overview
 
-To be released with the official model launch.
+The GLM family welcomes new members [GLM-4-0414](https://arxiv.org/pdf/2406.12793) series models.
+
+The **GLM-4-32B-0414** series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT
+series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features. GLM-4-32B-Base-0414
+was pre-trained on 15T of high-quality data, including substantial reasoning-type synthetic data. This lays the
+foundation for subsequent reinforcement learning extensions. In the post-training stage, we employed human preference
+alignment for dialogue scenarios. Additionally, using techniques like rejection sampling and reinforcement learning, we
+enhanced the model’s performance in instruction following, engineering code, and function calling, thus strengthening
+the atomic capabilities required for agent tasks. GLM-4-32B-0414 achieves good results in engineering code, Artifact
+generation, function calling, search-based Q&A, and report generation. In particular, on several benchmarks, such as
+code generation or specific Q&A tasks, GLM-4-32B-Base-0414 achieves comparable performance with those larger models like
+GPT-4o and DeepSeek-V3-0324 (671B).
+
+**GLM-Z1-32B-0414** is a reasoning model with deep thinking capabilities. This was developed based on GLM-4-32B-0414
+through cold start, extended reinforcement learning, and further training on tasks including mathematics, code, and
+logic. Compared to the base model, GLM-Z1-32B-0414 significantly improves mathematical abilities and the capability to
+solve complex tasks. During training, we also introduced general reinforcement learning based on pairwise ranking
+feedback, which enhances the model's general capabilities.
+
+**GLM-Z1-Rumination-32B-0414** is a deep reasoning model with rumination capabilities (against OpenAI's Deep Research).
+Unlike typical deep thinking models, the rumination model is capable of deeper and longer thinking to solve more
+open-ended and complex problems (e.g., writing a comparative analysis of AI development in two cities and their future
+development plans). Z1-Rumination is trained through scaling end-to-end reinforcement learning with responses graded by
+the ground truth answers or rubrics and can make use of search tools during its deep thinking process to handle complex
+tasks. The model shows significant improvements in research-style writing and complex tasks.
+
+Finally, **GLM-Z1-9B-0414** is a surprise. We employed all the aforementioned techniques to train a small model (9B).
+GLM-Z1-9B-0414 exhibits excellent capabilities in mathematical reasoning and general tasks. Its overall performance is
+top-ranked among all open-source models of the same size. Especially in resource-constrained scenarios, this model
+achieves an excellent balance between efficiency and effectiveness, providing a powerful option for users seeking
+lightweight deployment.
 
 ## Glm4Config
 

docs/source/en/model_doc/glm4v.md

@@ -0,0 +1,203 @@
+<!--Copyright 2025 The ZhipuAI Inc. and The HuggingFace Inc. 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.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+<img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">    </div>
+</div>
+
+# GLM-4.1V
+
+## Overview
+
+**GLM-4.1V-9B-Thinking** is a bilingual vision-language model optimized for reasoning, built on GLM-4-9B. It introduces
+a "thinking paradigm" with reinforcement learning, achieving state-of-the-art results among 10B-class models and
+rivaling 72B-scale models. It supports 64k context, 4K resolution, and arbitrary aspect ratios, with an open-source base
+model for further research. You can check our paper [here](https://huggingface.co/papers/2507.01006). and below is a abstract.
+
+*We present GLM-4.1V-Thinking, a vision-language model (VLM) designed to advance general-purpose multimodal understanding
+and reasoning. In this report, we share our key findings in the development of the reasoning-centric training framework.
+We first develop a capable vision foundation model with significant potential through large-scale pre-training, which
+arguably sets the upper bound for the final performance. We then propose Reinforcement Learning with Curriculum
+Sampling (RLCS) to unlock the full potential of the model, leading to comprehensive capability enhancement across a
+diverse range of tasks, including STEM problem solving, video understanding, content recognition, coding, grounding,
+GUI-based agents, and long document understanding. We open-source GLM-4.1V-9B-Thinking, which achieves state-of-the-art
+performance among models of comparable size. In a comprehensive evaluation across 28 public benchmarks, our model
+outperforms Qwen2.5-VL-7B on nearly all tasks and achieves comparable or even superior performance on 18 benchmarks
+relative to the significantly larger Qwen2.5-VL-72B. Notably, GLM-4.1V-9B-Thinking also demonstrates competitive or
+superior performance compared to closed-source models such as GPT-4o on challenging tasks including long document
+understanding and STEM reasoning, further underscoring its strong capabilities. Code, models and more information
+are released at https://github.com/THUDM/GLM-4.1V-Thinking.*
+
+## Usage
+
+The example below demonstrates how to generate text based on an image with [`Pipeline`] or the [`AutoModel`] class.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```py
+import torch
+from transformers import pipeline
+pipe = pipeline(
+    task="image-text-to-text",
+    model="THUDM/GLM-4.1V-9B-Thinking",
+    device=0,
+    torch_dtype=torch.bfloat16
+)
+messages = [
+    {
+        "role": "user",
+        "content": [
+            {
+                "type": "image",
+                "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg",
+            },
+            { "type": "text", "text": "Describe this image."},
+        ]
+    }
+]
+pipe(text=messages,max_new_tokens=20, return_full_text=False)
+```
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import Glm4vForConditionalGeneration, AutoProcessor
+
+model = Glm4vForConditionalGeneration.from_pretrained(
+    "THUDM/GLM-4.1V-9B-Thinking",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    attn_implementation="sdpa"
+)
+processor = AutoProcessor.from_pretrained("THUDM/GLM-4.1V-9B-Thinking")
+messages = [
+    {
+        "role":"user",
+        "content":[
+            {
+                "type":"image",
+                "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"
+            },
+            {
+                "type":"text",
+                "text":"Describe this image."
+            }
+        ]
+    }
+
+]
+
+inputs = processor.apply_chat_template(
+    messages,
+    add_generation_prompt=True,
+    tokenize=True,
+    return_dict=True,
+    return_tensors="pt"
+).to("cuda")
+
+generated_ids = model.generate(**inputs, max_new_tokens=128)
+generated_ids_trimmed = [
+            out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
+]
+output_text = processor.batch_decode(
+       generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
+)
+print(output_text)
+```
+</hfoption>
+</hfoptions>
+
+Using GLM-4.1V with video input is similar to using it with image input.
+The model can process video data and generate text based on the content of the video.
+
+```python
+from transformers import AutoProcessor, Glm4vForConditionalGeneration
+import torch
+
+processor = AutoProcessor.from_pretrained("THUDM/GLM-4.1V-9B-Thinking")
+model = Glm4vForConditionalGeneration.from_pretrained(
+    pretrained_model_name_or_path="THUDM/GLM-4.1V-9B-Thinking",
+    torch_dtype=torch.bfloat16,
+    device_map="cuda:0"
+)
+
+messages = [
+    {
+        "role": "user",
+        "content": [
+            {
+                "type": "video",
+                "url": "https://test-videos.co.uk/vids/bigbuckbunny/mp4/h264/720/Big_Buck_Bunny_720_10s_10MB.mp4",
+            },
+            {
+                "type": "text",
+                "text": "discribe this video",
+            },
+        ],
+    }
+]
+inputs = processor.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_dict=True, return_tensors="pt", padding=True).to("cuda:0")
+generated_ids = model.generate(**inputs, max_new_tokens=1024, do_sample=True, temperature=1.0)
+output_text = processor.decode(generated_ids[0][inputs["input_ids"].shape[1] :], skip_special_tokens=True)
+print(output_text)
+```
+
+## Glm4vConfig
+
+[[autodoc]] Glm4vConfig
+
+## Glm4vTextConfig
+
+[[autodoc]] Glm4vTextConfig
+
+## Glm4vImageProcessor
+
+[[autodoc]] Glm4vImageProcessor
+    - preprocess
+
+## Glm4vVideoProcessor
+
+[[autodoc]] Glm4vVideoProcessor
+    - preprocess
+
+## Glm4vImageProcessorFast
+
+[[autodoc]] Glm4vImageProcessorFast
+    - preprocess
+
+## Glm4vProcessor
+
+[[autodoc]] Glm4vProcessor
+
+## Glm4vTextModel
+
+[[autodoc]] Glm4vTextModel
+    - forward
+
+## Glm4vModel
+
+[[autodoc]] Glm4vModel
+    - forward
+
+## Glm4vForConditionalGeneration
+
+[[autodoc]] Glm4vForConditionalGeneration
+    - forward

docs/source/en/model_doc/granite.md

@@ -19,6 +19,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 # Granite

docs/source/en/model_doc/idefics2.md

@@ -162,7 +162,7 @@ To load and run a model using Flash Attention-2, simply change the code snippet
 ```diff
 model = Idefics2ForConditionalGeneration.from_pretrained(
     "HuggingFaceM4/idefics2-8b",
-+    torch_dtype=torch.float16,    
++    torch_dtype=torch.float16,
 +    attn_implementation="flash_attention_2",
 ).to(device)
 ```
@@ -184,7 +184,7 @@ Quantizing a model is as simple as passing a `quantization_config` to the model.
 + )
 model = Idefics2ForConditionalGeneration.from_pretrained(
     "HuggingFaceM4/idefics2-8b",
-+    torch_dtype=torch.float16,    
++    torch_dtype=torch.float16,
 +    quantization_config=quantization_config,
 ).to(device)
 ```
@@ -218,7 +218,10 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 [[autodoc]] Idefics2ImageProcessor
     - preprocess
 
+## Idefics2ImageProcessorFast
+[[autodoc]] Idefics2ImageProcessorFast
+    - preprocess
 
 ## Idefics2Processor
 [[autodoc]] Idefics2Processor
-    - __call__
+    - __call__

docs/source/en/model_doc/idefics3.md

@@ -80,6 +80,9 @@ This model was contributed by [amyeroberts](https://huggingface.co/amyeroberts)
 [[autodoc]] Idefics3ImageProcessor
     - preprocess
 
+## Idefics3ImageProcessorFast
+[[autodoc]] Idefics3ImageProcessorFast
+    - preprocess
 
 ## Idefics3Processor
 [[autodoc]] Idefics3Processor

docs/source/en/model_doc/kyutai_speech_to_text.md

@@ -0,0 +1,122 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# Kyutai Speech-To-Text 
+## Overview
+
+Kyutai STT is a speech-to-text model architecture based on the [Mimi codec](https://huggingface.co/docs/transformers/en/model_doc/mimi), which encodes audio into discrete tokens in a streaming fashion, and a [Moshi-like](https://huggingface.co/docs/transformers/en/model_doc/moshi) autoregressive decoder. Kyutai’s lab has released two model checkpoints:
+- [kyutai/stt-1b-en_fr](https://huggingface.co/kyutai/stt-1b-en_fr): a 1B-parameter model capable of transcribing both English and French
+- [kyutai/stt-2.6b-en](https://huggingface.co/kyutai/stt-2.6b-en): a 2.6B-parameter model focused solely on English, optimized for maximum transcription accuracy
+
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/eustlb/documentation-images/resolve/main/kyutai_stt.png"/>
+</div>
+
+## Usage Tips
+
+### Inference
+
+```python
+import torch
+from datasets import load_dataset, Audio
+from transformers import KyutaiSpeechToTextProcessor, KyutaiSpeechToTextForConditionalGeneration
+
+# 1. load the model and the processor
+torch_device = "cuda" if torch.cuda.is_available() else "cpu"
+model_id = "kyutai/stt-2.6b-en-trfs"
+
+processor = KyutaiSpeechToTextProcessor.from_pretrained(model_id)
+model = KyutaiSpeechToTextForConditionalGeneration.from_pretrained(model_id, device_map=torch_device, torch_dtype="auto")
+
+# 2. load audio samples
+ds = load_dataset(
+    "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation"
+)
+ds = ds.cast_column("audio", Audio(sampling_rate=24000))
+
+# 3. prepare the model inputs
+inputs = processor(
+    ds[0]["audio"]["array"],
+)
+inputs.to(torch_device)
+
+# 4. infer the model
+output_tokens = model.generate(**inputs)
+
+# 5. decode the generated tokens
+print(processor.batch_decode(output_tokens, skip_special_tokens=True))
+```
+
+### Batched Inference
+
+```python
+import torch
+from datasets import load_dataset, Audio
+from transformers import KyutaiSpeechToTextProcessor, KyutaiSpeechToTextForConditionalGeneration
+
+# 1. load the model and the processor
+torch_device = "cuda" if torch.cuda.is_available() else "cpu"
+model_id = "kyutai/stt-2.6b-en-trfs"
+
+processor = KyutaiSpeechToTextProcessor.from_pretrained(model_id)
+model = KyutaiSpeechToTextForConditionalGeneration.from_pretrained(model_id, device_map=torch_device, torch_dtype="auto")
+
+# 2. load audio samples
+ds = load_dataset(
+    "hf-internal-testing/librispeech_asr_dummy", "clean", split="validation"
+)
+ds = ds.cast_column("audio", Audio(sampling_rate=24000))
+
+# 3. prepare the model inputs
+audio_arrays = [ds[i]["audio"]["array"] for i in range(4)]
+inputs = processor(audio_arrays, return_tensors="pt", padding=True)
+inputs = inputs.to(torch_device)
+
+# 4. infer the model
+output_tokens = model.generate(**inputs)
+
+# 5. decode the generated tokens
+decoded_outputs = processor.batch_decode(output_tokens, skip_special_tokens=True)
+for output in decoded_outputs:
+    print(output)
+```
+
+This model was contributed by [Eustache Le Bihan](https://huggingface.co/eustlb).
+The original code can be found [here](https://github.com/kyutai-labs/moshi).
+
+
+## KyutaiSpeechToTextConfig
+
+[[autodoc]] KyutaiSpeechToTextConfig
+
+## KyutaiSpeechToTextProcessor
+
+[[autodoc]] KyutaiSpeechToTextProcessor
+    - __call__
+
+## KyutaiSpeechToTextFeatureExtractor
+
+[[autodoc]] KyutaiSpeechToTextFeatureExtractor
+
+## KyutaiSpeechToTextForConditionalGeneration
+
+[[autodoc]] KyutaiSpeechToTextForConditionalGeneration
+    - forward
+    - generate
+
+## KyutaiSpeechToTextModel
+
+[[autodoc]] KyutaiSpeechToTextModel

docs/source/en/model_doc/led.md

@@ -14,62 +14,135 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# LED
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-<img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+           <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+            <img alt="TensorFlow" src="https://img.shields.io/badge/TensorFlow-FF6F00?style=flat&logo=tensorflow&logoColor=white">
+    </div>
 </div>
 
-## Overview
+# LED
 
-The LED model was proposed in [Longformer: The Long-Document Transformer](https://huggingface.co/papers/2004.05150) by Iz
-Beltagy, Matthew E. Peters, Arman Cohan.
+[Longformer-Encoder-Decoder (LED)](https://huggingface.co/papers/2004.05150) is an encoder-decoder  transformer model for sequence-to-sequence tasks like summarization. It extends [Longformer](.longformer), an encoder-only model designed to handle long inputs, by adding a decoder layer. The decoder uses full self-attention on the encoded tokens and previously decoded locations. Because of Longformer's linear self-attention mechanism, LED is more efficient than standard encoder-decoder models when processing long sequences.
 
-The abstract from the paper is the following:
+You can find all the original [LED] checkpoints under the [Ai2](https://huggingface.co/allenai/models?search=led) organization.
 
-*Transformer-based models are unable to process long sequences due to their self-attention operation, which scales
-quadratically with the sequence length. To address this limitation, we introduce the Longformer with an attention
-mechanism that scales linearly with sequence length, making it easy to process documents of thousands of tokens or
-longer. Longformer's attention mechanism is a drop-in replacement for the standard self-attention and combines a local
-windowed attention with a task motivated global attention. Following prior work on long-sequence transformers, we
-evaluate Longformer on character-level language modeling and achieve state-of-the-art results on text8 and enwik8. In
-contrast to most prior work, we also pretrain Longformer and finetune it on a variety of downstream tasks. Our
-pretrained Longformer consistently outperforms RoBERTa on long document tasks and sets new state-of-the-art results on
-WikiHop and TriviaQA. We finally introduce the Longformer-Encoder-Decoder (LED), a Longformer variant for supporting
-long document generative sequence-to-sequence tasks, and demonstrate its effectiveness on the arXiv summarization
-dataset.*
+> [!TIP]
+> This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten).
+>
+> Click on the LED models in the right sidebar for more examples of how to apply LED to different language tasks.
 
-## Usage tips
+The example below demonstrates how to summarize text with [`Pipeline`], [`AutoModel`], and from the command line.
 
-- [`LEDForConditionalGeneration`] is an extension of
-  [`BartForConditionalGeneration`] exchanging the traditional *self-attention* layer with
-  *Longformer*'s *chunked self-attention* layer. [`LEDTokenizer`] is an alias of
-  [`BartTokenizer`].
-- LED works very well on long-range *sequence-to-sequence* tasks where the `input_ids` largely exceed a length of
-  1024 tokens.
-- LED pads the `input_ids` to be a multiple of `config.attention_window` if required. Therefore a small speed-up is
-  gained, when [`LEDTokenizer`] is used with the `pad_to_multiple_of` argument.
-- LED makes use of *global attention* by means of the `global_attention_mask` (see
-  [`LongformerModel`]). For summarization, it is advised to put *global attention* only on the first
-  `<s>` token. For question answering, it is advised to put *global attention* on all tokens of the question.
-- To fine-tune LED on all 16384, *gradient checkpointing* can be enabled in case training leads to out-of-memory (OOM)
-  errors. This can be done by executing `model.gradient_checkpointing_enable()`. 
- Moreover, the `use_cache=False`
-  flag can be used to disable the caching mechanism to save memory.
-- LED is a model with absolute position embeddings so it's usually advised to pad the inputs on the right rather than
-  the left.
+<hfoptions id="usage">
+<hfoption id="Pipeline">
 
-This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten).
+```python
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+    task="summarization",
+    model="allenai/led-base-16384",
+    torch_dtype=torch.float16,
+    device=0
+)
+pipeline("""Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle.""")
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```python
+import torch
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "allenai/led-base-16384"
+)
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "allenai/led-base-16384",
+    torch_dtype=torch.float16,
+    device_map="auto"
+)
+
+input_text = """Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle."""
+input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
+
+# Place global attention on the first token
+global_attention_mask = torch.zeros_like(input_ids.input_ids).to("cuda")
+global_attention_mask[:, 0] = 1
+
+output = model.generate(**input_ids, global_attention_mask=global_attention_mask, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+</hfoption>
+<hfoption id="transformers-cli">
+
+```bash
+!echo -e "Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet. Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts." | transformers-cli run --task summarization --model allenai/led-base-16384 --device 0
+```
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [bitsandbytes](../quantization/bitsandbytes) to only quantize the weights to int4.
+
+```python
+import torch
+from transformers import BitsAndBytesConfig, AutoModelForSeq2SeqLM, AutoTokenizer
+
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.bfloat16,
+    bnb_4bit_quant_type="nf4"
+)
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "allenai/led-large-16384",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    quantization_config=quantization_config
+)
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "allenai/led-large-16384"
+)
+
+input_text = """Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle."""
+input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
+
+# Place global attention on the first token
+global_attention_mask = torch.zeros_like(input_ids.input_ids).to("cuda")
+global_attention_mask[:, 0] = 1
+
+output = model.generate(**input_ids, global_attention_mask=global_attention_mask, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+## Notes
+
+- [`LEDForConditionalGeneration`] is an extension of [`BartForConditionalGeneration`] exchanging the traditional self-attention layer with Longformer's chunked self-attention layer. [`LEDTokenizer`] is an alias of [`BartTokenizer`].
+- LED pads the `input_ids` to be a multiple of `config.attention_window` if required. A small speedup is gained when [`LEDTokenizer`] is used with the `pad_to_multiple_of` argument.
+- LED works best on long-range sequence-to-sequence tasks where the `input_ids` are significantly longer than 1024 tokens.
+- LED uses global attention by means of the `global_attention_mask` (see [`LongformerModel`]). For summarization, it is advised to put global attention only on the first `<s>` token. For question answering, it is advised to put global attention on all tokens of the question.
+- To fine-tune LED on all 16384 parameters, gradient checkpointing can be enabled in case training leads to out-of-memory (OOM) errors. Enable gradient checkpointing by adding `model.gradient_checkpointing_enable()` and setting `use_cache=False` to disable the caching mechanism to save memory.
+- Inputs should be padded on the right because LED uses absolute position embeddings.
 
 ## Resources
 
-- [A notebook showing how to evaluate LED](https://colab.research.google.com/drive/12INTTR6n64TzS4RrXZxMSXfrOd9Xzamo?usp=sharing).
-- [A notebook showing how to fine-tune LED](https://colab.research.google.com/drive/12LjJazBl7Gam0XBPy_y0CTOJZeZ34c2v?usp=sharing).
-- [Text classification task guide](../tasks/sequence_classification)
-- [Question answering task guide](../tasks/question_answering)
-- [Translation task guide](../tasks/translation)
-- [Summarization task guide](../tasks/summarization)
+- Read the [LED on Arxiv notebook](https://colab.research.google.com/drive/12INTTR6n64TzS4RrXZxMSXfrOd9Xzamo?usp=sharing) to see how LED can achieve state-of-the-art performance on Arxiv article summarization.
+- Read the [Fine-tune LED notebook](https://colab.research.google.com/drive/12LjJazBl7Gam0XBPy_y0CTOJZeZ34c2v?usp=sharing) to learn how to fine-tune LED on PubMed articles.
 
 ## LEDConfig
 

docs/source/en/model_doc/lightglue.md

@@ -0,0 +1,104 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the MIT License; you may not use this file except in compliance with
+the License.
+
+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.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+
+-->
+
+# LightGlue
+
+## Overview
+
+The LightGlue model was proposed in [LightGlue: Local Feature Matching at Light Speed](https://arxiv.org/abs/2306.13643)
+by Philipp Lindenberger, Paul-Edouard Sarlin and Marc Pollefeys.
+
+Similar to [SuperGlue](https://huggingface.co/magic-leap-community/superglue_outdoor), this model consists of matching
+two sets of local features extracted from two images, its goal is to be faster than SuperGlue. Paired with the 
+[SuperPoint model](https://huggingface.co/magic-leap-community/superpoint), it can be used to match two images and 
+estimate the pose between them. This model is useful for tasks such as image matching, homography estimation, etc.
+
+The abstract from the paper is the following:
+
+*We introduce LightGlue, a deep neural network that learns to match local features across images. We revisit multiple
+design decisions of SuperGlue, the state of the art in sparse matching, and derive simple but effective improvements. 
+Cumulatively, they make LightGlue more efficient - in terms of both memory and computation, more accurate, and much
+easier to train. One key property is that LightGlue is adaptive to the difficulty of the problem: the inference is much
+faster on image pairs that are intuitively easy to match, for example because of a larger visual overlap or limited
+appearance change. This opens up exciting prospects for deploying deep matchers in latency-sensitive applications like
+3D reconstruction. The code and trained models are publicly available at this [https URL](https://github.com/cvg/LightGlue)*
+
+## How to use
+
+Here is a quick example of using the model. Since this model is an image matching model, it requires pairs of images to be matched. 
+The raw outputs contain the list of keypoints detected by the keypoint detector as well as the list of matches with their corresponding 
+matching scores.
+```python
+from transformers import AutoImageProcessor, AutoModel
+import torch
+from PIL import Image
+import requests
+
+url_image1 = "https://raw.githubusercontent.com/magicleap/SuperGluePretrainedNetwork/refs/heads/master/assets/phototourism_sample_images/united_states_capitol_98169888_3347710852.jpg"
+image1 = Image.open(requests.get(url_image1, stream=True).raw)
+url_image2 = "https://raw.githubusercontent.com/magicleap/SuperGluePretrainedNetwork/refs/heads/master/assets/phototourism_sample_images/united_states_capitol_26757027_6717084061.jpg"
+image2 = Image.open(requests.get(url_image2, stream=True).raw)
+
+images = [image1, image2]
+
+processor = AutoImageProcessor.from_pretrained("ETH-CVG/lightglue_superpoint")
+model = AutoModel.from_pretrained("ETH-CVG/lightglue_superpoint")
+
+inputs = processor(images, return_tensors="pt")
+with torch.no_grad():
+    outputs = model(**inputs)
+```
+
+You can use the `post_process_keypoint_matching` method from the `LightGlueImageProcessor` to get the keypoints and matches in a readable format:
+```python
+image_sizes = [[(image.height, image.width) for image in images]]
+outputs = processor.post_process_keypoint_matching(outputs, image_sizes, threshold=0.2)
+for i, output in enumerate(outputs):
+    print("For the image pair", i)
+    for keypoint0, keypoint1, matching_score in zip(
+            output["keypoints0"], output["keypoints1"], output["matching_scores"]
+    ):
+        print(
+            f"Keypoint at coordinate {keypoint0.numpy()} in the first image matches with keypoint at coordinate {keypoint1.numpy()} in the second image with a score of {matching_score}."
+        )
+```
+
+You can visualize the matches between the images by providing the original images as well as the outputs to this method:
+```python
+processor.plot_keypoint_matching(images, outputs)
+```
+
+![image/png](https://cdn-uploads.huggingface.co/production/uploads/632885ba1558dac67c440aa8/duPp09ty8NRZlMZS18ccP.png)
+
+This model was contributed by [stevenbucaille](https://huggingface.co/stevenbucaille).
+The original code can be found [here](https://github.com/cvg/LightGlue).
+
+## LightGlueConfig
+
+[[autodoc]] LightGlueConfig
+
+## LightGlueImageProcessor
+
+[[autodoc]] LightGlueImageProcessor
+
+- preprocess
+- post_process_keypoint_matching
+- plot_keypoint_matching
+
+## LightGlueForKeypointMatching
+
+[[autodoc]] LightGlueForKeypointMatching
+
+- forward

docs/source/en/model_doc/llama.md

@@ -21,6 +21,7 @@ rendered properly in your Markdown viewer.
         ">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/llama2.md

@@ -19,6 +19,7 @@ rendered properly in your Markdown viewer.
         <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
         <img alt="Flax" src="https://img.shields.io/badge/Flax-29a79b.svg?style=flat&logo=data:image/png;base64,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
         ">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/llama3.md

@@ -20,6 +20,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="Flax" src="https://img.shields.io/badge/Flax-29a79b.svg?style=flat&logo=data:image/png;base64,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
 ">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ```py3

docs/source/en/model_doc/llama4.md

@@ -21,6 +21,7 @@ rendered properly in your Markdown viewer.
     <div class="flex flex-wrap space-x-1">
         <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/mistral.md

@@ -22,6 +22,7 @@ rendered properly in your Markdown viewer.
         ">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/mixtral.md

@@ -20,6 +20,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/mobilenet_v2.md

@@ -114,6 +114,7 @@ print(f"The predicted class label is: {predicted_class_label}")
 
 [[autodoc]] MobileNetV2ImageProcessor
     - preprocess
+    - post_process_semantic_segmentation
 
 ## MobileNetV2ImageProcessorFast
 

docs/source/en/model_doc/mobilevit.md

@@ -95,6 +95,12 @@ If you're interested in submitting a resource to be included here, please feel f
     - preprocess
     - post_process_semantic_segmentation
 
+## MobileViTImageProcessorFast
+
+[[autodoc]] MobileViTImageProcessorFast
+    - preprocess
+    - post_process_semantic_segmentation
+
 <frameworkcontent>
 <pt>
 

docs/source/en/model_doc/nougat.md

@@ -107,6 +107,11 @@ The model is identical to [Donut](donut) in terms of architecture.
 [[autodoc]] NougatImageProcessor
     - preprocess
 
+## NougatImageProcessorFast
+
+[[autodoc]] NougatImageProcessorFast
+    - preprocess
+
 ## NougatTokenizerFast
 
 [[autodoc]] NougatTokenizerFast

docs/source/en/model_doc/olmo.md

@@ -20,6 +20,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/pegasus_x.md

@@ -14,35 +14,115 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# PEGASUS-X
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-<img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+    </div>
 </div>
 
-## Overview
+# PEGASUS-X
 
-The PEGASUS-X model was proposed in [Investigating Efficiently Extending Transformers for Long Input Summarization](https://huggingface.co/papers/2208.04347)  by Jason Phang, Yao Zhao and Peter J. Liu.
+[PEGASUS-X](https://huggingface.co/papers/2208.04347) is an encoder-decoder (sequence-to-sequence) transformer model for long-input summarization. It extends the [Pegasus](./pegasus) model with staggered block-local attention, global encoder tokens, and additional pretraining on long text sequences, enabling it to handle inputs of up to 16,000 tokens. PEGASUS-X matches the performance of much larger models while using fewer parameters.
 
-PEGASUS-X (PEGASUS eXtended) extends the PEGASUS models for long input summarization through additional long input pretraining and using staggered block-local attention with global tokens in the encoder.
+You can find all the original PEGASUS-X checkpoints under the [Google](https://huggingface.co/google/models?search=pegasus-x) organization.
 
-The abstract from the paper is the following:
+> [!TIP]
+> This model was contributed by [zphang](https://huggingface.co/zphang).
+>
+> Click on the PEGASUS-X models in the right sidebar for more examples of how to apply PEGASUS-X to different language tasks.
 
-*While large pretrained Transformer models have proven highly capable at tackling natural language tasks, handling long sequence inputs continues to be a significant challenge. One such task is long input summarization, where inputs are longer than the maximum input context of most pretrained models. Through an extensive set of experiments, we investigate what model architectural changes and pretraining paradigms can most efficiently adapt a pretrained Transformer for long input summarization. We find that a staggered, block-local Transformer with global encoder tokens strikes a good balance of performance and efficiency, and that an additional pretraining phase on long sequences meaningfully improves downstream summarization performance. Based on our findings, we introduce PEGASUS-X, an extension of the PEGASUS model with additional long input pretraining to handle inputs of up to 16K tokens. PEGASUS-X achieves strong performance on long input summarization tasks comparable with much larger models while adding few additional parameters and not requiring model parallelism to train.*
+The example below demonstrates how to summarize text with [`Pipeline`], [`AutoModel`], and from the command line.
 
-This model was contributed by [zphang](https://huggingface.co/zphang). The original code can be found [here](https://github.com/google-research/pegasus).
+<hfoptions id="usage">
+<hfoption id="Pipeline">
 
-## Documentation resources
+```py
+import torch
+from transformers import pipeline
 
-- [Translation task guide](../tasks/translation)
-- [Summarization task guide](../tasks/summarization)
+pipeline = pipeline(
+    task="summarization",
+    model="google/pegasus-x-large",
+    torch_dtype=torch.bfloat16,
+    device=0
+)
+pipeline("""Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle.""")
+```
+</hfoption>
+<hfoption id="AutoModel">
 
-<Tip>
+```py
+import torch
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
 
-PEGASUS-X uses the same tokenizer as [PEGASUS](pegasus).
+tokenizer = AutoTokenizer.from_pretrained(
+    "google/pegasus-x-large"
+)
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "google/pegasus-x-large",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+)
 
-</Tip>
+input_text = """Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle."""
+input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
+
+output = model.generate(**input_ids, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+</hfoption>
+<hfoption id="transformers-cli">
+
+```bash
+echo -e "Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet. Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts." | transformers-cli run --task summarization --model google/pegasus-x-large --device 0
+```
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [bitsandbytes](../quantization/bitsandbytes) to only quantize the weights to int4.
+
+```py
+import torch
+from transformers import BitsAndBytesConfig, AutoModelForSeq2SeqLM, AutoTokenizer
+
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.bfloat16,
+    bnb_4bit_quant_type="nf4"
+)
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "google/pegasus-x-large",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    quantization_config=quantization_config
+)
+
+tokenizer = AutoTokenizer.from_pretrained(
+    "google/pegasus-x-large"
+)
+
+input_text = """Plants are among the most remarkable and essential life forms on Earth, possessing a unique ability to produce their own food through a process known as photosynthesis. This complex biochemical process is fundamental not only to plant life but to virtually all life on the planet.
+Through photosynthesis, plants capture energy from sunlight using a green pigment called chlorophyll, which is located in specialized cell structures called chloroplasts. In the presence of light, plants absorb carbon dioxide from the atmosphere through small pores in their leaves called stomata, and take in water from the soil through their root systems.
+These ingredients are then transformed into glucose, a type of sugar that serves as a source of chemical energy, and oxygen, which is released as a byproduct into the atmosphere. The glucose produced during photosynthesis is not just used immediately; plants also store it as starch or convert it into other organic compounds like cellulose, which is essential for building their cellular structure.
+This energy reserve allows them to grow, develop leaves, produce flowers, bear fruit, and carry out various physiological processes throughout their lifecycle."""
+input_ids = tokenizer(input_text, return_tensors="pt").to("cuda")
+
+output = model.generate(**input_ids, cache_implementation="static")
+print(tokenizer.decode(output[0], skip_special_tokens=True))
+```
+
+## Notes
+
+- PEGASUS-X also uses the [`PegasusTokenizer`].
 
 ## PegasusXConfig
 

docs/source/en/model_doc/phi.md

@@ -18,6 +18,7 @@ rendered properly in your Markdown viewer.
         <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/phi3.md

@@ -20,6 +20,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/qwen2.md

@@ -19,6 +19,7 @@ rendered properly in your Markdown viewer.
         <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
         <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
         <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
     </div>
 </div>
 

docs/source/en/model_doc/qwen2_moe.md

@@ -18,6 +18,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 # Qwen2MoE

docs/source/en/model_doc/qwen2_vl.md

@@ -19,6 +19,7 @@ rendered properly in your Markdown viewer.
 <div class="flex flex-wrap space-x-1">
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/roc_bert.md

@@ -14,39 +14,78 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# RoCBert
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+<div style="float: right;">
+   <div class="flex flex-wrap space-x-1">
+          <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+   </div>
 </div>
 
-## Overview
+# RoCBert
 
-The RoCBert model was proposed in [RoCBert: Robust Chinese Bert with Multimodal Contrastive Pretraining](https://aclanthology.org/2022.acl-long.65.pdf)  by HuiSu, WeiweiShi, XiaoyuShen, XiaoZhou, TuoJi, JiaruiFang, JieZhou.
-It's a pretrained Chinese language model that is robust under various forms of adversarial attacks.
+[RoCBert](https://aclanthology.org/2022.acl-long.65.pdf) is a pretrained Chinese [BERT](./bert) model designed against adversarial attacks like typos and synonyms. It is pretrained with a contrastive learning objective to align normal and adversarial text examples. The examples include different semantic, phonetic, and visual features of Chinese. This makes RoCBert more robust against manipulation.
 
-The abstract from the paper is the following:
+You can find all the original RoCBert checkpoints under the [weiweishi](https://huggingface.co/weiweishi) profile.
 
-*Large-scale pretrained language models have achieved SOTA results on NLP tasks. However, they have been shown
-vulnerable to adversarial attacks especially for logographic languages like Chinese. In this work, we propose
-ROCBERT: a pretrained Chinese Bert that is robust to various forms of adversarial attacks like word perturbation,
-synonyms, typos, etc. It is pretrained with the contrastive learning objective which maximizes the label consistency
-under different synthesized adversarial examples. The model takes as input multimodal information including the
-semantic, phonetic and visual features. We show all these features are important to the model robustness since the
-attack can be performed in all the three forms. Across 5 Chinese NLU tasks, ROCBERT outperforms strong baselines under
-three blackbox adversarial algorithms without sacrificing the performance on clean testset. It also performs the best
-in the toxic content detection task under human-made attacks.*
+> [!TIP]
+> This model was contributed by [weiweishi](https://huggingface.co/weiweishi).
+> 
+> Click on the RoCBert models in the right sidebar for more examples of how to apply RoCBert to different Chinese language tasks.
 
-This model was contributed by [weiweishi](https://huggingface.co/weiweishi).
+The example below demonstrates how to predict the [MASK] token with [`Pipeline`], [`AutoModel`], and from the command line.
 
-## Resources
+<hfoptions id="usage">
+<hfoption id="Pipeline">
 
-- [Text classification task guide](../tasks/sequence_classification)
-- [Token classification task guide](../tasks/token_classification)
-- [Question answering task guide](../tasks/question_answering)
-- [Causal language modeling task guide](../tasks/language_modeling)
-- [Masked language modeling task guide](../tasks/masked_language_modeling)
-- [Multiple choice task guide](../tasks/multiple_choice)
+```py
+import torch
+from transformers import pipeline
+
+pipeline = pipeline(
+   task="fill-mask",
+   model="weiweishi/roc-bert-base-zh",
+   torch_dtype=torch.float16,
+   device=0
+)
+pipeline("這家餐廳的拉麵是我[MASK]過的最好的拉麵之")
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```py
+import torch
+from transformers import AutoModelForMaskedLM, AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained(
+   "weiweishi/roc-bert-base-zh",
+)
+model = AutoModelForMaskedLM.from_pretrained(
+   "weiweishi/roc-bert-base-zh",
+   torch_dtype=torch.float16,
+   device_map="auto",
+)
+inputs = tokenizer("這家餐廳的拉麵是我[MASK]過的最好的拉麵之", return_tensors="pt").to("cuda")
+
+with torch.no_grad():
+   outputs = model(**inputs)
+   predictions = outputs.logits
+
+masked_index = torch.where(inputs['input_ids'] == tokenizer.mask_token_id)[1]
+predicted_token_id = predictions[0, masked_index].argmax(dim=-1)
+predicted_token = tokenizer.decode(predicted_token_id)
+
+print(f"The predicted token is: {predicted_token}")
+```
+
+</hfoption>
+<hfoption id="transformers CLI">
+
+```bash
+echo -e "這家餐廳的拉麵是我[MASK]過的最好的拉麵之" | transformers-cli run --task fill-mask --model weiweishi/roc-bert-base-zh --device 0
+```
+
+</hfoption>
+</hfoptions>
 
 ## RoCBertConfig
 

docs/source/en/model_doc/seamless_m4t.md

@@ -56,7 +56,7 @@ Here is how to use the processor to process text and audio:
 ```python
 >>> # let's load an audio sample from an Arabic speech corpus
 >>> from datasets import load_dataset
->>> dataset = load_dataset("arabic_speech_corpus", split="test", streaming=True, trust_remote_code=True)
+>>> dataset = load_dataset("halabi2016/arabic_speech_corpus", split="test", streaming=True)
 >>> audio_sample = next(iter(dataset))["audio"]
 
 >>> # now, process it

docs/source/en/model_doc/seamless_m4t_v2.md

@@ -56,7 +56,7 @@ Here is how to use the processor to process text and audio:
 ```python
 >>> # let's load an audio sample from an Arabic speech corpus
 >>> from datasets import load_dataset
->>> dataset = load_dataset("arabic_speech_corpus", split="test", streaming=True, trust_remote_code=True)
+>>> dataset = load_dataset("halabi2016/arabic_speech_corpus", split="test", streaming=True)
 >>> audio_sample = next(iter(dataset))["audio"]
 
 >>> # now, process it

docs/source/en/model_doc/smollm3.md

@@ -0,0 +1,173 @@
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
+</div>
+
+# SmolLM3
+
+SmolLM3 is a fully open, compact language model designed for efficient deployment while maintaining strong performance. It uses a Transformer decoder architecture with Grouped Query Attention (GQA) to reduce the kv cache, and no RoPE, enabling improved performance on long-context tasks. It is trained using a multi-stage training approach on high-quality public datasets across web, code, and math domains. The model is multilingual and supports very large context lengths. The instruct variant is optimized for reasoning and tool use.
+
+> [!TIP]
+> Click on the SmolLM3 models in the right sidebar for more examples of how to apply SmolLM3 to different language tasks.
+
+The example below demonstrates how to generate text with [`Pipeline`], [`AutoModel`], and from the command line using the instruction-tuned models.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+```python
+import torch
+from transformers import pipeline
+
+pipe = pipeline(
+    task="text-generation",
+    model="HuggingFaceTB/SmolLM3-3B",
+    torch_dtype=torch.bfloat16,
+    device_map=0
+)
+
+messages = [
+    {"role": "system", "content": "You are a helpful assistant."},
+    {"role": "user", "content": "Tell me about yourself."},
+]
+outputs = pipe(messages, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
+print(outputs[0]["generated_text"][-1]['content'])
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```python
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model = AutoModelForCausalLM.from_pretrained(
+    "HuggingFaceTB/SmolLM3-3B",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    attn_implementation="sdpa"
+)
+tokenizer = AutoTokenizer.from_pretrained("HuggingFaceTB/SmolLM3-3B")
+
+prompt = "Give me a short introduction to large language models."
+messages = [
+    {"role": "system", "content": "You are a helpful assistant."},
+    {"role": "user", "content": prompt}
+]
+text = tokenizer.apply_chat_template(
+    messages,
+    tokenize=False,
+    add_generation_prompt=True
+)
+model_inputs = tokenizer([text], return_tensors="pt").to("cuda")
+
+generated_ids = model.generate(
+    model_inputs.input_ids,
+    cache_implementation="static",
+    max_new_tokens=512,
+    do_sample=True,
+    temperature=0.7,
+    top_k=50,
+    top_p=0.95
+)
+generated_ids = [
+    output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+]
+
+response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+print(response)
+```
+
+</hfoption>
+<hfoption id="transformers CLI">
+
+```bash
+# pip install -U flash-attn --no-build-isolation
+transformers chat HuggingFaceTB/SmolLM3-3B --torch_dtype auto --attn_implementation flash_attention_2 --device 0
+```
+
+</hfoption>
+</hfoptions>
+
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
+
+The example below uses [bitsandbytes](../quantization/bitsandbytes) to quantize the weights to 4-bits.
+
+```python
+# pip install -U flash-attn --no-build-isolation
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
+
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.bfloat16,
+    bnb_4bit_quant_type="nf4",
+    bnb_4bit_use_double_quant=True,
+)
+
+tokenizer = AutoTokenizer.from_pretrained("HuggingFaceTB/SmolLM3-3B")
+model = AutoModelForCausalLM.from_pretrained(
+    "HuggingFaceTB/SmolLM3-3B",
+    torch_dtype=torch.bfloat16,
+    device_map="auto",
+    quantization_config=quantization_config,
+    attn_implementation="flash_attention_2"
+)
+
+inputs = tokenizer("Gravity is the force", return_tensors="pt").to("cuda")
+outputs = model.generate(**inputs, max_new_tokens=100)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+
+## Notes
+
+- Ensure your Transformers library version is up-to-date. SmolLM3 requires Transformers>=4.53.0 for full support.
+
+## SmolLM3Config
+
+[[autodoc]] SmolLM3Config
+
+## SmolLM3Model
+
+[[autodoc]] SmolLM3Model
+    - forward
+
+## SmolLM3ForCausalLM
+
+[[autodoc]] SmolLM3ForCausalLM
+    - forward
+
+## SmolLM3ForSequenceClassification
+
+[[autodoc]] SmolLM3ForSequenceClassification
+    - forward
+
+## SmolLM3ForTokenClassification
+
+[[autodoc]] SmolLM3ForTokenClassification
+    - forward
+
+## SmolLM3ForQuestionAnswering
+
+[[autodoc]] SmolLM3ForQuestionAnswering
+    - forward

docs/source/en/model_doc/smolvlm.md

@@ -32,7 +32,7 @@ SmolVLM2 is an adaptation of the Idefics3 model with two main differences:
 
 Input images are processed either by upsampling (if resizing is enabled) or at their original resolution. The resizing behavior depends on two parameters: do_resize and size.
 
-Videos should not be upsampled. 
+Videos should not be upsampled.
 
 If `do_resize` is set to `True`, the model resizes images so that the longest edge is 4*512 pixels by default.
 The default resizing behavior can be customized by passing a dictionary to the `size` parameter. For example, `{"longest_edge": 4 * 512}` is the default, but you can change it to a different value if needed.
@@ -192,11 +192,14 @@ print(generated_texts[0])
 [[autodoc]] SmolVLMForConditionalGeneration
     - forward
 
-
 ## SmolVLMImageProcessor
 [[autodoc]] SmolVLMImageProcessor
     - preprocess
 
+## SmolVLMImageProcessorFast
+[[autodoc]] SmolVLMImageProcessorFast
+    - preprocess
+
 ## SmolVLMVideoProcessor
 [[autodoc]] SmolVLMVideoProcessor
     - preprocess

docs/source/en/model_doc/speech_to_text_2.md

@@ -61,19 +61,16 @@ predicted token ids.
 - Step-by-step Speech Translation
 
 ```python
->>> import torch
 >>> from transformers import Speech2Text2Processor, SpeechEncoderDecoderModel
 >>> from datasets import load_dataset
->>> import soundfile as sf
 
 >>> model = SpeechEncoderDecoderModel.from_pretrained("facebook/s2t-wav2vec2-large-en-de")
 >>> processor = Speech2Text2Processor.from_pretrained("facebook/s2t-wav2vec2-large-en-de")
 
 
->>> def map_to_array(batch):
-...     speech, _ = sf.read(batch["file"])
-...     batch["speech"] = speech
-...     return batch
+>>> def map_to_array(example):
+...     example["speech"] = example["audio"]["array"]
+...     return example
 
 
 >>> ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")

docs/source/en/model_doc/starcoder2.md

@@ -20,6 +20,7 @@ rendered properly in your Markdown viewer.
 <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
 <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
 <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+<img alt="Tensor parallelism" src="https://img.shields.io/badge/Tensor%20parallelism-06b6d4?style=flat&logoColor=white">
 </div>
 
 ## Overview

docs/source/en/model_doc/superpoint.md

@@ -10,48 +10,35 @@ specific language governing permissions and limitations under the License.
 ⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
 rendered properly in your Markdown viewer.
 
-
 -->
 
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white" >
+    </div>
+</div>
+
 # SuperPoint
 
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
-</div>
-
-## Overview
-
-The SuperPoint model was proposed
-in [SuperPoint: Self-Supervised Interest Point Detection and Description](https://huggingface.co/papers/1712.07629) by Daniel
-DeTone, Tomasz Malisiewicz and Andrew Rabinovich.
-
-This model is the result of a self-supervised training of a fully-convolutional network for interest point detection and
-description. The model is able to detect interest points that are repeatable under homographic transformations and
-provide a descriptor for each point. The use of the model in its own is limited, but it can be used as a feature
-extractor for other tasks such as homography estimation, image matching, etc.
-
-The abstract from the paper is the following:
-
-*This paper presents a self-supervised framework for training interest point detectors and descriptors suitable for a
-large number of multiple-view geometry problems in computer vision. As opposed to patch-based neural networks, our
-fully-convolutional model operates on full-sized images and jointly computes pixel-level interest point locations and
-associated descriptors in one forward pass. We introduce Homographic Adaptation, a multi-scale, multi-homography
-approach for boosting interest point detection repeatability and performing cross-domain adaptation (e.g.,
-synthetic-to-real). Our model, when trained on the MS-COCO generic image dataset using Homographic Adaptation, is able
-to repeatedly detect a much richer set of interest points than the initial pre-adapted deep model and any other
-traditional corner detector. The final system gives rise to state-of-the-art homography estimation results on HPatches
-when compared to LIFT, SIFT and ORB.*
+[SuperPoint](https://huggingface.co/papers/1712.07629) is the result of self-supervised training of a fully-convolutional network for interest point detection and description. The model is able to detect interest points that are repeatable under homographic transformations and provide a descriptor for each point. Usage on it's own is limited, but it can be used as a feature extractor for other tasks such as homography estimation and image matching.
 
 <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/superpoint_architecture.png"
 alt="drawing" width="500"/>
 
-<small> SuperPoint overview. Taken from the <a href="https://huggingface.co/papers/1712.07629v4">original paper.</a> </small>
+You can find all the original SuperPoint checkpoints under the [Magic Leap Community](https://huggingface.co/magic-leap-community) organization.
 
-## Usage tips
+> [!TIP]
+> This model was contributed by [stevenbucaille](https://huggingface.co/stevenbucaille).
+>
+> Click on the SuperPoint models in the right sidebar for more examples of how to apply SuperPoint to different computer vision tasks.
 
-Here is a quick example of using the model to detect interest points in an image:
 
-```python
+
+The example below demonstrates how to detect interest points in an image with the [`AutoModel`] class.
+<hfoptions id="usage">
+<hfoption id="AutoModel">
+
+```py
 from transformers import AutoImageProcessor, SuperPointForKeypointDetection
 import torch
 from PIL import Image
@@ -64,67 +51,76 @@ processor = AutoImageProcessor.from_pretrained("magic-leap-community/superpoint"
 model = SuperPointForKeypointDetection.from_pretrained("magic-leap-community/superpoint")
 
 inputs = processor(image, return_tensors="pt")
-outputs = model(**inputs)
+with torch.no_grad():
+    outputs = model(**inputs)
+
+# Post-process to get keypoints, scores, and descriptors
+image_size = (image.height, image.width)
+processed_outputs = processor.post_process_keypoint_detection(outputs, [image_size])
 ```
 
-The outputs contain the list of keypoint coordinates with their respective score and description (a 256-long vector).
+</hfoption>
+</hfoptions>
 
-You can also feed multiple images to the model. Due to the nature of SuperPoint, to output a dynamic number of keypoints,
-you will need to use the mask attribute to retrieve the respective information :
+## Notes
 
-```python
-from transformers import AutoImageProcessor, SuperPointForKeypointDetection
-import torch
-from PIL import Image
-import requests
+- SuperPoint outputs a dynamic number of keypoints per image, which makes it suitable for tasks requiring variable-length feature representations.
 
-url_image_1 = "http://images.cocodataset.org/val2017/000000039769.jpg"
-image_1 = Image.open(requests.get(url_image_1, stream=True).raw)
-url_image_2 = "http://images.cocodataset.org/test-stuff2017/000000000568.jpg"
-image_2 = Image.open(requests.get(url_image_2, stream=True).raw)
+    ```py
+    from transformers import AutoImageProcessor, SuperPointForKeypointDetection
+    import torch
+    from PIL import Image
+    import requests
+    processor = AutoImageProcessor.from_pretrained("magic-leap-community/superpoint")
+    model = SuperPointForKeypointDetection.from_pretrained("magic-leap-community/superpoint")
+    url_image_1 = "http://images.cocodataset.org/val2017/000000039769.jpg"
+    image_1 = Image.open(requests.get(url_image_1, stream=True).raw)
+    url_image_2 = "http://images.cocodataset.org/test-stuff2017/000000000568.jpg"
+    image_2 = Image.open(requests.get(url_image_2, stream=True).raw)
+    images = [image_1, image_2]
+    inputs = processor(images, return_tensors="pt")
+    # Example of handling dynamic keypoint output
+    outputs = model(**inputs)
+    keypoints = outputs.keypoints  # Shape varies per image
+    scores = outputs.scores        # Confidence scores for each keypoint
+    descriptors = outputs.descriptors  # 256-dimensional descriptors
+    mask = outputs.mask # Value of 1 corresponds to a keypoint detection
+    ```
 
-images = [image_1, image_2]
+- The model provides both keypoint coordinates and their corresponding descriptors (256-dimensional vectors) in a single forward pass.
+- For batch processing with multiple images, you need to use the mask attribute to retrieve the respective information for each image. You can use the `post_process_keypoint_detection` from the `SuperPointImageProcessor` to retrieve the each image information.
 
-processor = AutoImageProcessor.from_pretrained("magic-leap-community/superpoint")
-model = SuperPointForKeypointDetection.from_pretrained("magic-leap-community/superpoint")
+    ```py
+    # Batch processing example
+    images = [image1, image2, image3]
+    inputs = processor(images, return_tensors="pt")
+    outputs = model(**inputs)
+    image_sizes = [(img.height, img.width) for img in images]
+    processed_outputs = processor.post_process_keypoint_detection(outputs, image_sizes)
+    ```
 
-inputs = processor(images, return_tensors="pt")
-outputs = model(**inputs)
-image_sizes = [(image.height, image.width) for image in images]
-outputs = processor.post_process_keypoint_detection(outputs, image_sizes)
+- You can then print the keypoints on the image of your choice to visualize the result:
+    ```py
+    import matplotlib.pyplot as plt
+    plt.axis("off")
+    plt.imshow(image_1)
+    plt.scatter(
+        outputs[0]["keypoints"][:, 0],
+        outputs[0]["keypoints"][:, 1],
+        c=outputs[0]["scores"] * 100,
+        s=outputs[0]["scores"] * 50,
+        alpha=0.8
+    )
+    plt.savefig(f"output_image.png")
+    ```
 
-for output in outputs:
-    for keypoints, scores, descriptors in zip(output["keypoints"], output["scores"], output["descriptors"]):
-        print(f"Keypoints: {keypoints}")
-        print(f"Scores: {scores}")
-        print(f"Descriptors: {descriptors}")
-```
-
-You can then print the keypoints on the image of your choice to visualize the result:
-```python
-import matplotlib.pyplot as plt
-
-plt.axis("off")
-plt.imshow(image_1)
-plt.scatter(
-    outputs[0]["keypoints"][:, 0],
-    outputs[0]["keypoints"][:, 1],
-    c=outputs[0]["scores"] * 100,
-    s=outputs[0]["scores"] * 50,
-    alpha=0.8
-)
-plt.savefig(f"output_image.png")
-```
-![image/png](https://cdn-uploads.huggingface.co/production/uploads/632885ba1558dac67c440aa8/ZtFmphEhx8tcbEQqOolyE.png)
-
-This model was contributed by [stevenbucaille](https://huggingface.co/stevenbucaille).
-The original code can be found [here](https://github.com/magicleap/SuperPointPretrainedNetwork).
+<div class="flex justify-center">
+    <img src="https://cdn-uploads.huggingface.co/production/uploads/632885ba1558dac67c440aa8/ZtFmphEhx8tcbEQqOolyE.png">
+</div>
 
 ## Resources
 
-A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with SuperPoint. If you're interested in submitting a resource to be included here, please feel free to open a Pull Request and we'll review it! The resource should ideally demonstrate something new instead of duplicating an existing resource.
-
-- A notebook showcasing inference and visualization with SuperPoint can be found [here](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/SuperPoint/Inference_with_SuperPoint_to_detect_interest_points_in_an_image.ipynb). 🌎
+- Refer to this [noteboook](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/SuperPoint/Inference_with_SuperPoint_to_detect_interest_points_in_an_image.ipynb) for an inference and visualization example.
 
 ## SuperPointConfig
 
@@ -137,8 +133,12 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 - preprocess
 - post_process_keypoint_detection
 
+<frameworkcontent>
+<pt>
 ## SuperPointForKeypointDetection
 
 [[autodoc]] SuperPointForKeypointDetection
 
 - forward
+
+</pt>

docs/source/en/model_doc/t5gemma.md

@@ -0,0 +1,125 @@
+
+<!--Copyright 2025 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+<div style="float: right;">
+    <div class="flex flex-wrap space-x-1">
+        <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+        <img alt="FlashAttention" src="https://img.shields.io/badge/%E2%9A%A1%EF%B8%8E%20FlashAttention-eae0c8?style=flat">
+        <img alt="SDPA" src="https://img.shields.io/badge/SDPA-DE3412?style=flat&logo=pytorch&logoColor=white">
+    </div>
+</div>
+
+# T5Gemma
+
+T5Gemma (aka encoder-decoder Gemma) was proposed in a [research paper](https://arxiv.org/abs/2504.06225) by Google. It is a family of encoder-decoder large langauge models, developed by adapting pretrained decoder-only models into encoder-decoder. T5Gemma includes pretrained and instruction-tuned variants. The architecture is based on transformer encoder-decoder design following T5, with improvements from Gemma 2: GQA, RoPE, GeGLU activation, RMSNorm, and interleaved local/global attention.
+
+T5Gemma has two groups of model sizes: 1) [Gemma 2](https://ai.google.dev/gemma/docs/core/model_card_2) sizes (2B-2B, 9B-2B, and 9B-9B), which are based on the offical Gemma 2 models (2B and 9B); and 2) [T5](https://arxiv.org/abs/1910.10683) sizes (Small, Base, Large, and XL), where are pretrained under the Gemma 2 framework following T5 configuration. In addition, we also provide a model at ML size (medium large, ~2B in total), which is in-between T5 Large and T5 XL.
+
+The pretrained varaints are trained with two objectives: prefix language modeling with knowledge distillation (PrefixLM) and UL2, separately. We release both variants for each model size. The instruction-turned varaints was post-trained with supervised fine-tuning and reinforcement learning.
+
+> [!TIP]
+> Click on the T5Gemma models in the right sidebar for more examples of how to apply T5Gemma to different language tasks.
+
+The example below demonstrates how to chat with the model with [`Pipeline`] or the [`AutoModel`] class, and from the command line.
+
+<hfoptions id="usage">
+<hfoption id="Pipeline">
+
+
+```python
+import torch
+from transformers import pipeline
+
+pipe = pipeline(
+    "text2text-generation",
+    model="google/t5gemma-2b-2b-prefixlm-it",
+    torch_dtype=torch.bfloat16,
+    device="cuda",  # replace with "mps" to run on a Mac device
+)
+
+messages = [
+    {"role": "user", "content": "Tell me an unknown interesting biology fact about the brain."},
+]
+prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+
+pipe(prompt, max_new_tokens=32)
+```
+
+</hfoption>
+<hfoption id="AutoModel">
+
+```python
+# pip install accelerate
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+import torch
+
+tokenizer = AutoTokenizer.from_pretrained("google/t5gemma-2b-2b-prefixlm-it")
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "google/t5gemma-2b-2b-prefixlm-it",
+    device_map="auto",
+    torch_dtype=torch.bfloat16,
+)
+
+messages = [
+    {"role": "user", "content": "Tell me an unknown interesting biology fact about the brain."},
+]
+input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt", return_dict=True, add_generation_prompt=True).to("cuda")
+
+outputs = model.generate(**input_ids, max_new_tokens=32)
+print(tokenizer.decode(outputs[0]))
+```
+
+</hfoption>
+<hfoption id="transformers CLI">
+
+```
+echo -e "Write me a poem about Machine Learning. Answer:" | transformers run --task text2text-generation --model google/t5gemma-2b-2b-prefixlm --device 0
+```
+</hfoption>
+</hfoptions>
+
+## T5GemmaConfig
+
+[[autodoc]] T5GemmaConfig
+
+## T5GemmaModuleConfig
+
+[[autodoc]] T5GemmaModuleConfig
+
+## T5GemmaModel
+
+[[autodoc]] T5GemmaModel
+    - forward
+
+## T5GemmaEncoderModel
+
+[[autodoc]] T5GemmaEncoderModel
+    - forward
+
+## T5GemmaForConditionalGeneration
+
+[[autodoc]] T5GemmaForConditionalGeneration
+    - forward
+
+## T5GemmaForSequenceClassification
+
+[[autodoc]] T5GemmaForSequenceClassification
+    - forward
+
+## T5GemmaForTokenClassification
+
+[[autodoc]] T5GemmaForTokenClassification
+    - forward

docs/source/en/model_doc/trocr.md

@@ -56,6 +56,7 @@ This model was contributed by [nielsr](https://huggingface.co/nielsr). The origi
   on both printed (e.g. the [SROIE dataset](https://paperswithcode.com/dataset/sroie) and handwritten (e.g. the [IAM
   Handwriting dataset](https://fki.tic.heia-fr.ch/databases/iam-handwriting-database>) text recognition tasks. For more
   information, see the [official models](https://huggingface.co/models?other=trocr>).
+- [Fine‑tune TrOCR on your own OCR dataset](https://github.com/Ashutosh-4485/trocr-custom-fine-tune.git).
 - TrOCR is always used within the [VisionEncoderDecoder](vision-encoder-decoder) framework.
 
 ## Resources

docs/source/en/model_doc/video_llava.md

@@ -83,7 +83,7 @@ def read_video_pyav(container, indices):
     Decode the video with PyAV decoder.
     Args:
         container (`av.container.input.InputContainer`): PyAV container.
-        indices (`List[int]`): List of frame indices to decode.
+        indices (`list[int]`): List of frame indices to decode.
     Returns:
         result (np.ndarray): np array of decoded frames of shape (num_frames, height, width, 3).
     '''

docs/source/en/model_doc/vitpose.md

@@ -10,52 +10,39 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express o
 specific language governing permissions and limitations under the License.
 -->
 
-# ViTPose
-
-<div class="flex flex-wrap space-x-1">
-<img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+<div style="float: right;">
+  <div class="flex flex-wrap space-x-1">
+    <img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-DE3412?style=flat&logo=pytorch&logoColor=white">
+  </div>
 </div>
 
-## Overview
+# ViTPose
 
-The ViTPose model was proposed in [ViTPose: Simple Vision Transformer Baselines for Human Pose Estimation](https://huggingface.co/papers/2204.12484) by Yufei Xu, Jing Zhang, Qiming Zhang, Dacheng Tao. ViTPose employs a standard, non-hierarchical [Vision Transformer](vit) as backbone for the task of keypoint estimation. A simple decoder head is added on top to predict the heatmaps from a given image. Despite its simplicity, the model gets state-of-the-art results on the challenging MS COCO Keypoint Detection benchmark. The model was further improved in [ViTPose++: Vision Transformer for Generic Body Pose Estimation](https://huggingface.co/papers/2212.04246) where the authors employ
-a mixture-of-experts (MoE) module in the ViT backbone along with pre-training on more data, which further enhances the performance.
+[ViTPose](https://huggingface.co/papers/2204.12484) is a vision transformer-based model for keypoint (pose) estimation. It uses a simple, non-hierarchical [ViT](./vit) backbone and a lightweight decoder head. This architecture simplifies model design, takes advantage of transformer scalability, and can be adapted to different training strategies.
 
-The abstract from the paper is the following:
-
-*Although no specific domain knowledge is considered in the design, plain vision transformers have shown excellent performance in visual recognition tasks. However, little effort has been made to reveal the potential of such simple structures for pose estimation tasks. In this paper, we show the surprisingly good capabilities of plain vision transformers for pose estimation from various aspects, namely simplicity in model structure, scalability in model size, flexibility in training paradigm, and transferability of knowledge between models, through a simple baseline model called ViTPose. Specifically, ViTPose employs plain and non-hierarchical vision transformers as backbones to extract features for a given person instance and a lightweight decoder for pose estimation. It can be scaled up from 100M to 1B parameters by taking the advantages of the scalable model capacity and high parallelism of transformers, setting a new Pareto front between throughput and performance. Besides, ViTPose is very flexible regarding the attention type, input resolution, pre-training and finetuning strategy, as well as dealing with multiple pose tasks. We also empirically demonstrate that the knowledge of large ViTPose models can be easily transferred to small ones via a simple knowledge token. Experimental results show that our basic ViTPose model outperforms representative methods on the challenging MS COCO Keypoint Detection benchmark, while the largest model sets a new state-of-the-art.*
+[ViTPose++](https://huggingface.co/papers/2212.04246) improves on ViTPose by incorporating a mixture-of-experts (MoE) module in the backbone and using more diverse pretraining data.
 
 <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/vitpose-architecture.png"
 alt="drawing" width="600"/>
 
-<small> ViTPose architecture. Taken from the <a href="https://huggingface.co/papers/2204.12484">original paper.</a> </small>
+You can find all ViTPose and ViTPose++ checkpoints under the [ViTPose collection](https://huggingface.co/collections/usyd-community/vitpose-677fcfd0a0b2b5c8f79c4335).
 
-This model was contributed by [nielsr](https://huggingface.co/nielsr) and [sangbumchoi](https://github.com/SangbumChoi).
-The original code can be found [here](https://github.com/ViTAE-Transformer/ViTPose).
-
-## Usage Tips
-
-ViTPose is a so-called top-down keypoint detection model. This means that one first uses an object detector, like [RT-DETR](rt_detr.md), to detect people (or other instances) in an image. Next, ViTPose takes the cropped images as input and predicts the keypoints for each of them.
+The example below demonstrates pose estimation with the [`VitPoseForPoseEstimation`] class.
 
 ```py
 import torch
 import requests
 import numpy as np
-
+import supervision as sv
 from PIL import Image
-
 from transformers import AutoProcessor, RTDetrForObjectDetection, VitPoseForPoseEstimation
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
-url = "http://images.cocodataset.org/val2017/000000000139.jpg"
+url = "https://www.fcbarcelona.com/fcbarcelona/photo/2021/01/31/3c55a19f-dfc1-4451-885e-afd14e890a11/mini_2021-01-31-BARCELONA-ATHLETIC-BILBAOI-30.JPG"
 image = Image.open(requests.get(url, stream=True).raw)
 
-# ------------------------------------------------------------------------
-# Stage 1. Detect humans on the image
-# ------------------------------------------------------------------------
-
-# You can choose any detector of your choice
+# Detect humans in the image
 person_image_processor = AutoProcessor.from_pretrained("PekingU/rtdetr_r50vd_coco_o365")
 person_model = RTDetrForObjectDetection.from_pretrained("PekingU/rtdetr_r50vd_coco_o365", device_map=device)
 
@@ -67,7 +54,7 @@ with torch.no_grad():
 results = person_image_processor.post_process_object_detection(
     outputs, target_sizes=torch.tensor([(image.height, image.width)]), threshold=0.3
 )
-result = results[0]  # take first image results
+result = results[0]
 
 # Human label refers 0 index in COCO dataset
 person_boxes = result["boxes"][result["labels"] == 0]
@@ -77,10 +64,7 @@ person_boxes = person_boxes.cpu().numpy()
 person_boxes[:, 2] = person_boxes[:, 2] - person_boxes[:, 0]
 person_boxes[:, 3] = person_boxes[:, 3] - person_boxes[:, 1]
 
-# ------------------------------------------------------------------------
-# Stage 2. Detect keypoints for each person found
-# ------------------------------------------------------------------------
-
+# Detect keypoints for each person found
 image_processor = AutoProcessor.from_pretrained("usyd-community/vitpose-base-simple")
 model = VitPoseForPoseEstimation.from_pretrained("usyd-community/vitpose-base-simple", device_map=device)
 
@@ -90,54 +74,7 @@ with torch.no_grad():
     outputs = model(**inputs)
 
 pose_results = image_processor.post_process_pose_estimation(outputs, boxes=[person_boxes])
-image_pose_result = pose_results[0]  # results for first image
-```
-
-### ViTPose++ models
-
-The best [checkpoints](https://huggingface.co/collections/usyd-community/vitpose-677fcfd0a0b2b5c8f79c4335) are those of the [ViTPose++ paper](https://huggingface.co/papers/2212.04246). ViTPose++ models employ a so-called [Mixture-of-Experts (MoE)](https://huggingface.co/blog/moe) architecture for the ViT backbone, resulting in better performance.
-
-The ViTPose+ checkpoints use 6 experts, hence 6 different dataset indices can be passed. 
-An overview of the various dataset indices is provided below:
-
-- 0: [COCO validation 2017](https://cocodataset.org/#overview) dataset, using an object detector that gets 56 AP on the "person" class
-- 1: [AiC](https://github.com/fabbrimatteo/AiC-Dataset) dataset
-- 2: [MPII](https://www.mpi-inf.mpg.de/departments/computer-vision-and-machine-learning/software-and-datasets/mpii-human-pose-dataset) dataset
-- 3: [AP-10K](https://github.com/AlexTheBad/AP-10K) dataset
-- 4: [APT-36K](https://github.com/pandorgan/APT-36K) dataset
-- 5: [COCO-WholeBody](https://github.com/jin-s13/COCO-WholeBody) dataset
-
-Pass the `dataset_index` argument in the forward of the model to indicate which experts to use for each example in the batch. Example usage is shown below:
-
-```python
-image_processor = AutoProcessor.from_pretrained("usyd-community/vitpose-plus-base")
-model = VitPoseForPoseEstimation.from_pretrained("usyd-community/vitpose-plus-base", device=device)
-
-inputs = image_processor(image, boxes=[person_boxes], return_tensors="pt").to(device)
-
-dataset_index = torch.tensor([0], device=device) # must be a tensor of shape (batch_size,)
-
-with torch.no_grad():
-    outputs = model(**inputs, dataset_index=dataset_index)
-```
-
-The ViTPose+ checkpoints use 6 experts, hence 6 different dataset indices can be passed. 
-An overview of the various dataset indices is provided below:
-
-- 0: [COCO validation 2017](https://cocodataset.org/#overview) dataset, using an object detector that gets 56 AP on the "person" class
-- 1: [AiC](https://github.com/fabbrimatteo/AiC-Dataset) dataset
-- 2: [MPII](https://www.mpi-inf.mpg.de/departments/computer-vision-and-machine-learning/software-and-datasets/mpii-human-pose-dataset) dataset
-- 3: [AP-10K](https://github.com/AlexTheBad/AP-10K) dataset
-- 4: [APT-36K](https://github.com/pandorgan/APT-36K) dataset
-- 5: [COCO-WholeBody](https://github.com/jin-s13/COCO-WholeBody) dataset
-
-
-### Visualization
-
-To visualize the various keypoints, one can either leverage the `supervision` [library](https://github.com/roboflow/supervision (requires `pip install supervision`):
-
-```python
-import supervision as sv
+image_pose_result = pose_results[0]
 
 xy = torch.stack([pose_result['keypoints'] for pose_result in image_pose_result]).cpu().numpy()
 scores = torch.stack([pose_result['scores'] for pose_result in image_pose_result]).cpu().numpy()
@@ -162,119 +99,192 @@ annotated_frame = vertex_annotator.annotate(
     scene=annotated_frame,
     key_points=key_points
 )
+annotated_frame
 ```
 
-Alternatively, one can also visualize the keypoints using [OpenCV](https://opencv.org/) (requires `pip install opencv-python`):
+<div class="flex justify-center">
+    <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/vitpose.png"/>
+</div>
 
-```python
-import math
-import cv2
+Quantization reduces the memory burden of large models by representing the weights in a lower precision. Refer to the [Quantization](../quantization/overview) overview for more available quantization backends.
 
-def draw_points(image, keypoints, scores, pose_keypoint_color, keypoint_score_threshold, radius, show_keypoint_weight):
-    if pose_keypoint_color is not None:
-        assert len(pose_keypoint_color) == len(keypoints)
-    for kid, (kpt, kpt_score) in enumerate(zip(keypoints, scores)):
-        x_coord, y_coord = int(kpt[0]), int(kpt[1])
-        if kpt_score > keypoint_score_threshold:
-            color = tuple(int(c) for c in pose_keypoint_color[kid])
-            if show_keypoint_weight:
-                cv2.circle(image, (int(x_coord), int(y_coord)), radius, color, -1)
-                transparency = max(0, min(1, kpt_score))
-                cv2.addWeighted(image, transparency, image, 1 - transparency, 0, dst=image)
-            else:
-                cv2.circle(image, (int(x_coord), int(y_coord)), radius, color, -1)
+The example below uses [torchao](../quantization/torchao) to only quantize the weights to int4.
 
-def draw_links(image, keypoints, scores, keypoint_edges, link_colors, keypoint_score_threshold, thickness, show_keypoint_weight, stick_width = 2):
-    height, width, _ = image.shape
-    if keypoint_edges is not None and link_colors is not None:
-        assert len(link_colors) == len(keypoint_edges)
-        for sk_id, sk in enumerate(keypoint_edges):
-            x1, y1, score1 = (int(keypoints[sk[0], 0]), int(keypoints[sk[0], 1]), scores[sk[0]])
-            x2, y2, score2 = (int(keypoints[sk[1], 0]), int(keypoints[sk[1], 1]), scores[sk[1]])
-            if (
-                x1 > 0
-                and x1 < width
-                and y1 > 0
-                and y1 < height
-                and x2 > 0
-                and x2 < width
-                and y2 > 0
-                and y2 < height
-                and score1 > keypoint_score_threshold
-                and score2 > keypoint_score_threshold
-            ):
-                color = tuple(int(c) for c in link_colors[sk_id])
+```py
+# pip install torchao
+import torch
+import requests
+import numpy as np
+from PIL import Image
+from transformers import AutoProcessor, RTDetrForObjectDetection, VitPoseForPoseEstimation, TorchAoConfig
+
+url = "https://www.fcbarcelona.com/fcbarcelona/photo/2021/01/31/3c55a19f-dfc1-4451-885e-afd14e890a11/mini_2021-01-31-BARCELONA-ATHLETIC-BILBAOI-30.JPG"
+image = Image.open(requests.get(url, stream=True).raw)
+
+person_image_processor = AutoProcessor.from_pretrained("PekingU/rtdetr_r50vd_coco_o365")
+person_model = RTDetrForObjectDetection.from_pretrained("PekingU/rtdetr_r50vd_coco_o365", device_map=device)
+
+inputs = person_image_processor(images=image, return_tensors="pt").to(device)
+
+with torch.no_grad():
+    outputs = person_model(**inputs)
+
+results = person_image_processor.post_process_object_detection(
+    outputs, target_sizes=torch.tensor([(image.height, image.width)]), threshold=0.3
+)
+result = results[0]
+
+person_boxes = result["boxes"][result["labels"] == 0]
+person_boxes = person_boxes.cpu().numpy()
+
+person_boxes[:, 2] = person_boxes[:, 2] - person_boxes[:, 0]
+person_boxes[:, 3] = person_boxes[:, 3] - person_boxes[:, 1]
+
+quantization_config = TorchAoConfig("int4_weight_only", group_size=128)
+
+image_processor = AutoProcessor.from_pretrained("usyd-community/vitpose-plus-huge")
+model = VitPoseForPoseEstimation.from_pretrained("usyd-community/vitpose-plus-huge", device_map=device, quantization_config=quantization_config)
+
+inputs = image_processor(image, boxes=[person_boxes], return_tensors="pt").to(device)
+
+with torch.no_grad():
+    outputs = model(**inputs)
+
+pose_results = image_processor.post_process_pose_estimation(outputs, boxes=[person_boxes])
+image_pose_result = pose_results[0]
+```
+
+## Notes
+
+- Use [`AutoProcessor`] to automatically prepare bounding box and image inputs.
+- ViTPose is a top-down pose estimator. It uses a object detector to detect individuals first before keypoint prediction.
+- ViTPose++ has 6 different MoE expert heads (COCO validation `0`, AiC `1`, MPII `2`, AP-10K `3`, APT-36K `4`, COCO-WholeBody `5`) which supports 6 different datasets. Pass a specific value corresponding to the dataset to the `dataset_index` to indicate which expert to use.
+
+    ```py
+    from transformers import AutoProcessor, VitPoseForPoseEstimation
+
+    device = "cuda" if torch.cuda.is_available() else "cpu"
+
+    image_processor = AutoProcessor.from_pretrained("usyd-community/vitpose-plus-base")
+    model = VitPoseForPoseEstimation.from_pretrained("usyd-community/vitpose-plus-base", device=device)
+
+    inputs = image_processor(image, boxes=[person_boxes], return_tensors="pt").to(device)
+    dataset_index = torch.tensor([0], device=device) # must be a tensor of shape (batch_size,)
+
+    with torch.no_grad():
+        outputs = model(**inputs, dataset_index=dataset_index)
+    ```
+
+- [OpenCV](https://opencv.org/) is an alternative option for visualizing the estimated pose.
+
+    ```py
+    # pip install opencv-python
+    import math
+    import cv2
+
+    def draw_points(image, keypoints, scores, pose_keypoint_color, keypoint_score_threshold, radius, show_keypoint_weight):
+        if pose_keypoint_color is not None:
+            assert len(pose_keypoint_color) == len(keypoints)
+        for kid, (kpt, kpt_score) in enumerate(zip(keypoints, scores)):
+            x_coord, y_coord = int(kpt[0]), int(kpt[1])
+            if kpt_score > keypoint_score_threshold:
+                color = tuple(int(c) for c in pose_keypoint_color[kid])
                 if show_keypoint_weight:
-                    X = (x1, x2)
-                    Y = (y1, y2)
-                    mean_x = np.mean(X)
-                    mean_y = np.mean(Y)
-                    length = ((Y[0] - Y[1]) ** 2 + (X[0] - X[1]) ** 2) ** 0.5
-                    angle = math.degrees(math.atan2(Y[0] - Y[1], X[0] - X[1]))
-                    polygon = cv2.ellipse2Poly(
-                        (int(mean_x), int(mean_y)), (int(length / 2), int(stick_width)), int(angle), 0, 360, 1
-                    )
-                    cv2.fillConvexPoly(image, polygon, color)
-                    transparency = max(0, min(1, 0.5 * (keypoints[sk[0], 2] + keypoints[sk[1], 2])))
+                    cv2.circle(image, (int(x_coord), int(y_coord)), radius, color, -1)
+                    transparency = max(0, min(1, kpt_score))
                     cv2.addWeighted(image, transparency, image, 1 - transparency, 0, dst=image)
                 else:
-                    cv2.line(image, (x1, y1), (x2, y2), color, thickness=thickness)
+                    cv2.circle(image, (int(x_coord), int(y_coord)), radius, color, -1)
 
+    def draw_links(image, keypoints, scores, keypoint_edges, link_colors, keypoint_score_threshold, thickness, show_keypoint_weight, stick_width = 2):
+        height, width, _ = image.shape
+        if keypoint_edges is not None and link_colors is not None:
+            assert len(link_colors) == len(keypoint_edges)
+            for sk_id, sk in enumerate(keypoint_edges):
+                x1, y1, score1 = (int(keypoints[sk[0], 0]), int(keypoints[sk[0], 1]), scores[sk[0]])
+                x2, y2, score2 = (int(keypoints[sk[1], 0]), int(keypoints[sk[1], 1]), scores[sk[1]])
+                if (
+                    x1 > 0
+                    and x1 < width
+                    and y1 > 0
+                    and y1 < height
+                    and x2 > 0
+                    and x2 < width
+                    and y2 > 0
+                    and y2 < height
+                    and score1 > keypoint_score_threshold
+                    and score2 > keypoint_score_threshold
+                ):
+                    color = tuple(int(c) for c in link_colors[sk_id])
+                    if show_keypoint_weight:
+                        X = (x1, x2)
+                        Y = (y1, y2)
+                        mean_x = np.mean(X)
+                        mean_y = np.mean(Y)
+                        length = ((Y[0] - Y[1]) ** 2 + (X[0] - X[1]) ** 2) ** 0.5
+                        angle = math.degrees(math.atan2(Y[0] - Y[1], X[0] - X[1]))
+                        polygon = cv2.ellipse2Poly(
+                            (int(mean_x), int(mean_y)), (int(length / 2), int(stick_width)), int(angle), 0, 360, 1
+                        )
+                        cv2.fillConvexPoly(image, polygon, color)
+                        transparency = max(0, min(1, 0.5 * (keypoints[sk[0], 2] + keypoints[sk[1], 2])))
+                        cv2.addWeighted(image, transparency, image, 1 - transparency, 0, dst=image)
+                    else:
+                        cv2.line(image, (x1, y1), (x2, y2), color, thickness=thickness)
 
-# Note: keypoint_edges and color palette are dataset-specific
-keypoint_edges = model.config.edges
+    # Note: keypoint_edges and color palette are dataset-specific
+    keypoint_edges = model.config.edges
 
-palette = np.array(
-    [
-        [255, 128, 0],
-        [255, 153, 51],
-        [255, 178, 102],
-        [230, 230, 0],
-        [255, 153, 255],
-        [153, 204, 255],
-        [255, 102, 255],
-        [255, 51, 255],
-        [102, 178, 255],
-        [51, 153, 255],
-        [255, 153, 153],
-        [255, 102, 102],
-        [255, 51, 51],
-        [153, 255, 153],
-        [102, 255, 102],
-        [51, 255, 51],
-        [0, 255, 0],
-        [0, 0, 255],
-        [255, 0, 0],
-        [255, 255, 255],
-    ]
-)
+    palette = np.array(
+        [
+            [255, 128, 0],
+            [255, 153, 51],
+            [255, 178, 102],
+            [230, 230, 0],
+            [255, 153, 255],
+            [153, 204, 255],
+            [255, 102, 255],
+            [255, 51, 255],
+            [102, 178, 255],
+            [51, 153, 255],
+            [255, 153, 153],
+            [255, 102, 102],
+            [255, 51, 51],
+            [153, 255, 153],
+            [102, 255, 102],
+            [51, 255, 51],
+            [0, 255, 0],
+            [0, 0, 255],
+            [255, 0, 0],
+            [255, 255, 255],
+        ]
+    )
 
-link_colors = palette[[0, 0, 0, 0, 7, 7, 7, 9, 9, 9, 9, 9, 16, 16, 16, 16, 16, 16, 16]]
-keypoint_colors = palette[[16, 16, 16, 16, 16, 9, 9, 9, 9, 9, 9, 0, 0, 0, 0, 0, 0]]
+    link_colors = palette[[0, 0, 0, 0, 7, 7, 7, 9, 9, 9, 9, 9, 16, 16, 16, 16, 16, 16, 16]]
+    keypoint_colors = palette[[16, 16, 16, 16, 16, 9, 9, 9, 9, 9, 9, 0, 0, 0, 0, 0, 0]]
 
-numpy_image = np.array(image)
+    numpy_image = np.array(image)
 
-for pose_result in image_pose_result:
-    scores = np.array(pose_result["scores"])
-    keypoints = np.array(pose_result["keypoints"])
+    for pose_result in image_pose_result:
+        scores = np.array(pose_result["scores"])
+        keypoints = np.array(pose_result["keypoints"])
 
-    # draw each point on image
-    draw_points(numpy_image, keypoints, scores, keypoint_colors, keypoint_score_threshold=0.3, radius=4, show_keypoint_weight=False)
+        # draw each point on image
+        draw_points(numpy_image, keypoints, scores, keypoint_colors, keypoint_score_threshold=0.3, radius=4, show_keypoint_weight=False)
 
-    # draw links
-    draw_links(numpy_image, keypoints, scores, keypoint_edges, link_colors, keypoint_score_threshold=0.3, thickness=1, show_keypoint_weight=False)
+        # draw links
+        draw_links(numpy_image, keypoints, scores, keypoint_edges, link_colors, keypoint_score_threshold=0.3, thickness=1, show_keypoint_weight=False)
 
-pose_image = Image.fromarray(numpy_image)
-pose_image
-```
-<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/vitpose-coco.jpg" alt="drawing" width="600"/>
+    pose_image = Image.fromarray(numpy_image)
+    pose_image
+    ```
 
 ## Resources
 
-A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with ViTPose. If you're interested in submitting a resource to be included here, please feel free to open a Pull Request and we'll review it! The resource should ideally demonstrate something new instead of duplicating an existing resource.
+Refer to resources below to learn more about using ViTPose.
 
-- A demo of ViTPose on images and video can be found [here](https://huggingface.co/spaces/hysts/ViTPose-transformers).
-- A notebook illustrating inference and visualization can be found [here](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/ViTPose/Inference_with_ViTPose_for_human_pose_estimation.ipynb).
+- This [notebook](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/ViTPose/Inference_with_ViTPose_for_body_pose_estimation.ipynb) demonstrates inference and visualization.
+- This [Space](https://huggingface.co/spaces/hysts/ViTPose-transformers) demonstrates ViTPose on images and video.
 
 ## VitPoseImageProcessor