Release: v4.36.0

* up

* up

* test

* logits ok

* up

* up

* few fixes

* conversion script

* up

* nits

* nits

* update

* nuke

* more updates

* nites

* fix many issues

* nit

* scatter

* nit

* nuke megablocks

* nits

* fix conversion script

* nit

* remove

* nits

* nit

* update

* oupsssss

* change

* nits device

* nits

* fixup

* update

* merge

* add copied from

* fix the copy mentions

* update tests

* more fixes

* nits

* conversion script

* add parts of the readme

* Update tests/models/mixtral/test_modeling_mixtral.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* new test + conversion script

* Apply suggestions from code review

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Apply suggestions from code review

* fix

* fix copies

* fix copies

* ooops

* fix config

* Apply suggestions from code review

* fix nits

* nit

* add copies

* add batched tests

* docs

* fix flash attention

* let's add more verbose

* add correct outputs

* support router ouptus

* ignore copies where needed

* fix

* cat list if list is given for now

* nits

* Update docs/source/en/model_doc/mixtral.md

* finish router refactoring

* fix forward

* fix expected values

* nits

* fixup

* fix

* fix bug

* fix

* fix dtype mismatch

* fix

* grrr grrr I support item assignment

* fix CI

* docs

* fixup

* remove some copied form

* fix weird diff

* skip doctest fast on the config and modeling

* mark that is supports flash attention in the doc

* update

* Update src/transformers/models/mixtral/modeling_mixtral.py

Co-authored-by: Lysandre Debut <hi@lysand.re>

* Update docs/source/en/model_doc/mixtral.md

Co-authored-by: Lysandre Debut <hi@lysand.re>

* revert router logits config issue

* update doc accordingly

* Update src/transformers/models/mixtral/convert_mixtral_weights_to_hf.py

* nits

* use torch testing asssert close

* fixup

* doc nits

---------

Co-authored-by: younesbelkada <younesbelkada@gmail.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Lysandre Debut <hi@lysand.re>

.circleci/config.yml

@@ -157,11 +157,10 @@ jobs:
                 command: pip freeze | tee installed.txt
             - store_artifacts:
                   path: ~/transformers/installed.txt
-            - run: black --check examples tests src utils
-            - run: ruff examples tests src utils
+            - run: ruff check examples tests src utils
+            - run: ruff format tests src utils --check
             - run: python utils/custom_init_isort.py --check_only
             - run: python utils/sort_auto_mappings.py --check_only
-            - run: doc-builder style src/transformers docs/source --max_len 119 --check_only --path_to_docs docs/source
             - run: python utils/check_doc_toc.py
 
     check_repository_consistency:

.circleci/create_circleci_config.py

@@ -15,7 +15,6 @@
 
 import argparse
 import copy
-import glob
 import os
 import random
 from dataclasses import dataclass
@@ -239,7 +238,7 @@ class CircleCIJob:
 
         py_command = f'import os; fp = open("reports/{self.job_name}/summary_short.txt"); failed = os.linesep.join([x for x in fp.read().split(os.linesep) if x.startswith("ERROR ")]); fp.close(); fp = open("summary_short.txt", "w"); fp.write(failed); fp.close()'
         check_test_command += f"$(python3 -c '{py_command}'); "
-        check_test_command += f'cat summary_short.txt; echo ""; exit -1; '
+        check_test_command += 'cat summary_short.txt; echo ""; exit -1; '
 
         # Deeal with failed tests
         check_test_command += f'elif [ -s reports/{self.job_name}/failures_short.txt ]; '
@@ -249,7 +248,7 @@ class CircleCIJob:
 
         py_command = f'import os; fp = open("reports/{self.job_name}/summary_short.txt"); failed = os.linesep.join([x for x in fp.read().split(os.linesep) if x.startswith("FAILED ")]); fp.close(); fp = open("summary_short.txt", "w"); fp.write(failed); fp.close()'
         check_test_command += f"$(python3 -c '{py_command}'); "
-        check_test_command += f'cat summary_short.txt; echo ""; exit -1; '
+        check_test_command += 'cat summary_short.txt; echo ""; exit -1; '
 
         check_test_command += f'elif [ -s reports/{self.job_name}/stats.txt ]; then echo "All tests pass!"; '
 

.github/workflows/build-docker-images.yml

@@ -208,40 +208,41 @@ jobs:
           push: true
           tags: huggingface/transformers-pytorch-gpu
 
-  latest-pytorch-amd:
-    name: "Latest PyTorch (AMD) [dev]"
-    runs-on: [self-hosted, docker-gpu, amd-gpu, single-gpu, mi210]
-    steps:
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v3
-      - name: Check out code
-        uses: actions/checkout@v3
-      - name: Login to DockerHub
-        uses: docker/login-action@v3
-        with:
-          username: ${{ secrets.DOCKERHUB_USERNAME }}
-          password: ${{ secrets.DOCKERHUB_PASSWORD }}
-      - name: Build and push
-        uses: docker/build-push-action@v5
-        with:
-          context: ./docker/transformers-pytorch-amd-gpu
-          build-args: |
-            REF=main
-          push: true
-          tags: huggingface/transformers-pytorch-amd-gpu${{ inputs.image_postfix }}
-      # Push CI images still need to be re-built daily
-      -
-        name: Build and push (for Push CI) in a daily basis
-        # This condition allows `schedule` events, or `push` events that trigger this workflow NOT via `workflow_call`.
-        # The later case is useful for manual image building for debugging purpose. Use another tag in this case!
-        if: inputs.image_postfix != '-push-ci'
-        uses: docker/build-push-action@v5
-        with:
-          context: ./docker/transformers-pytorch-amd-gpu
-          build-args: |
-            REF=main
-          push: true
-          tags: huggingface/transformers-pytorch-amd-gpu-push-ci
+# Need to be fixed with the help from Guillaume.
+#  latest-pytorch-amd:
+#    name: "Latest PyTorch (AMD) [dev]"
+#    runs-on: [self-hosted, docker-gpu, amd-gpu, single-gpu, mi210]
+#    steps:
+#      - name: Set up Docker Buildx
+#        uses: docker/setup-buildx-action@v3
+#      - name: Check out code
+#        uses: actions/checkout@v3
+#      - name: Login to DockerHub
+#        uses: docker/login-action@v3
+#        with:
+#          username: ${{ secrets.DOCKERHUB_USERNAME }}
+#          password: ${{ secrets.DOCKERHUB_PASSWORD }}
+#      - name: Build and push
+#        uses: docker/build-push-action@v5
+#        with:
+#          context: ./docker/transformers-pytorch-amd-gpu
+#          build-args: |
+#            REF=main
+#          push: true
+#          tags: huggingface/transformers-pytorch-amd-gpu${{ inputs.image_postfix }}
+#      # Push CI images still need to be re-built daily
+#      -
+#        name: Build and push (for Push CI) in a daily basis
+#        # This condition allows `schedule` events, or `push` events that trigger this workflow NOT via `workflow_call`.
+#        # The later case is useful for manual image building for debugging purpose. Use another tag in this case!
+#        if: inputs.image_postfix != '-push-ci'
+#        uses: docker/build-push-action@v5
+#        with:
+#          context: ./docker/transformers-pytorch-amd-gpu
+#          build-args: |
+#            REF=main
+#          push: true
+#          tags: huggingface/transformers-pytorch-amd-gpu-push-ci
 
   latest-tensorflow:
     name: "Latest TensorFlow [dev]"

.github/workflows/check_runner_status.yml

@@ -1,68 +0,0 @@
-name: Self-hosted runner (check runner status)
-
-# Note that each job's dependencies go into a corresponding docker file.
-#
-# For example for `run_all_tests_torch_cuda_extensions_gpu` the docker image is
-# `huggingface/transformers-pytorch-deepspeed-latest-gpu`, which can be found at
-# `docker/transformers-pytorch-deepspeed-latest-gpu/Dockerfile`
-
-on:
-  repository_dispatch:
-  schedule:
-    # run per hour
-    - cron: "0 */1 * * *"
-
-env:
-  TRANSFORMERS_IS_CI: yes
-
-jobs:
-  check_runner_status:
-    name: Check Runner Status
-    runs-on: ubuntu-22.04
-    outputs:
-      offline_runners: ${{ steps.set-offline_runners.outputs.offline_runners }}
-    steps:
-      - name: Checkout transformers
-        uses: actions/checkout@v3
-        with:
-          fetch-depth: 2
-
-      - name: Check Runner Status
-        run: python utils/check_self_hosted_runner.py --target_runners single-gpu-ci-runner-docker,multi-gpu-ci-runner-docker,single-gpu-scheduled-ci-runner-docker,multi-scheduled-scheduled-ci-runner-docker,single-gpu-doctest-ci-runner-docker --token ${{ secrets.ACCESS_REPO_INFO_TOKEN }}
-
-      - id: set-offline_runners
-        name: Set output for offline runners
-        if: ${{ always() }}
-        run: |
-          offline_runners=$(python3 -c 'fp = open("offline_runners.txt"); failed = fp.read(); fp.close(); print(failed)')
-          echo "offline_runners=$offline_runners" >> $GITHUB_OUTPUT
-
-  send_results:
-    name: Send results to webhook
-    runs-on: ubuntu-22.04
-    needs: check_runner_status
-    if: ${{ failure() }}
-    steps:
-      - name: Preliminary job status
-        shell: bash
-        run: |
-          echo "Runner availability: ${{ needs.check_runner_status.result }}"
-
-      - uses: actions/checkout@v3
-      - uses: actions/download-artifact@v3
-      - name: Send message to Slack
-        env:
-          CI_SLACK_BOT_TOKEN: ${{ secrets.CI_SLACK_BOT_TOKEN }}
-          CI_SLACK_CHANNEL_ID: ${{ secrets.CI_SLACK_CHANNEL_ID }}
-          CI_SLACK_CHANNEL_ID_DAILY: ${{ secrets.CI_SLACK_CHANNEL_ID_DAILY }}
-          CI_SLACK_CHANNEL_DUMMY_TESTS: ${{ secrets.CI_SLACK_CHANNEL_DUMMY_TESTS }}
-          CI_SLACK_REPORT_CHANNEL_ID: ${{ secrets.CI_SLACK_CHANNEL_ID_DAILY }}
-          ACCESS_REPO_INFO_TOKEN: ${{ secrets.ACCESS_REPO_INFO_TOKEN }}
-          CI_EVENT: runner status check
-          RUNNER_STATUS: ${{ needs.check_runner_status.result }}
-          OFFLINE_RUNNERS: ${{ needs.check_runner_status.outputs.offline_runners }}
-        # We pass `needs.setup.outputs.matrix` as the argument. A processing in `notification_service.py` to change
-        # `models/bert` to `models_bert` is required, as the artifact names use `_` instead of `/`.
-        run: |
-          pip install slack_sdk
-          python utils/notification_service.py

.github/workflows/delete_doc_comment.yml

@@ -1,14 +0,0 @@
-name: Delete doc comment
-
-on:
-  workflow_run:
-    workflows: ["Delete doc comment trigger"]
-    types:
-      - completed
-
-
-jobs:
-  delete:
-    uses: huggingface/doc-builder/.github/workflows/delete_doc_comment.yml@main
-    secrets:
-      comment_bot_token: ${{ secrets.COMMENT_BOT_TOKEN }}

.github/workflows/delete_doc_comment_trigger.yml

@@ -1,12 +0,0 @@
-name: Delete doc comment trigger
-
-on:
-  pull_request:
-    types: [ closed ]
-
-
-jobs:
-  delete:
-    uses: huggingface/doc-builder/.github/workflows/delete_doc_comment_trigger.yml@main
-    with:
-      pr_number: ${{ github.event.number }}

.github/workflows/self-past.yml

@@ -88,6 +88,10 @@ jobs:
         working-directory: /transformers
         run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
 
+      - name: Update some packages
+        working-directory: /transformers
+        run: python3 -m pip install -U datasets
+
       - 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
@@ -164,6 +168,10 @@ jobs:
         working-directory: /transformers
         run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
 
+      - name: Update some packages
+        working-directory: /transformers
+        run: python3 -m pip install -U datasets
+
       - 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
@@ -240,6 +248,10 @@ jobs:
         working-directory: /transformers
         run: python3 -m pip uninstall -y transformers && python3 -m pip install -e .
 
+      - name: Update some packages
+        working-directory: /transformers
+        run: python3 -m pip install -U datasets
+
       - name: Install
         working-directory: /transformers
         run: |

.github/workflows/self-push-amd-mi210-caller.yml

@@ -18,7 +18,7 @@ on:
 jobs:
   run_amd_ci:
     name: AMD mi210
-    if: (cancelled() != true) && ((github.event_name != 'schedule') || ((github.event_name == 'push') && startsWith(github.ref_name, 'run_amd_push_ci_caller')))
+    if: (cancelled() != true) && ((github.event_name == 'push') && (github.ref_name == 'main' || startsWith(github.ref_name, 'run_amd_push_ci_caller')))
     uses: ./.github/workflows/self-push-amd.yml
     with:
       gpu_flavor: mi210

.github/workflows/self-push-amd-mi250-caller.yml

@@ -18,7 +18,7 @@ on:
 jobs:
   run_amd_ci:
     name: AMD mi250
-    if: (cancelled() != true) && ((github.event_name != 'schedule') || ((github.event_name == 'push') && startsWith(github.ref_name, 'run_amd_push_ci_caller')))
+    if: (cancelled() != true) && ((github.event_name == 'push') && (github.ref_name == 'main' || startsWith(github.ref_name, 'run_amd_push_ci_caller')))
     uses: ./.github/workflows/self-push-amd.yml
     with:
       gpu_flavor: mi250

.github/workflows/self-push-amd.yml

@@ -38,14 +38,16 @@ jobs:
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
       image: huggingface/transformers-pytorch-amd-gpu-push-ci  # <--- We test only for PyTorch for now
-      options: --device /dev/kfd --device /dev/dri --env HIP_VISIBLE_DEVICES --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+      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: ROCM-SMI
         run: |
-          rocminfo  | grep "Agent" -A 14
-      - name: Show HIP environment
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
         run: |
-          echo "HIP: $HIP_VISIBLE_DEVICES"
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
 
   setup_gpu:
@@ -57,7 +59,7 @@ jobs:
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
       image: huggingface/transformers-pytorch-amd-gpu-push-ci  # <--- We test only for PyTorch for now
-      options: --device /dev/kfd --device /dev/dri --env HIP_VISIBLE_DEVICES --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     outputs:
       matrix: ${{ steps.set-matrix.outputs.matrix }}
       test_map: ${{ steps.set-matrix.outputs.test_map }}
@@ -155,7 +157,7 @@ jobs:
     runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
     container:
       image: huggingface/transformers-pytorch-amd-gpu-push-ci  # <--- We test only for PyTorch for now
-      options: --device /dev/kfd --device /dev/dri --env HIP_VISIBLE_DEVICES --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
     steps:
       # Necessary to get the correct branch name and commit SHA for `workflow_run` event
       # We also take into account the `push` event (we might want to test some changes in a branch)
@@ -207,10 +209,12 @@ jobs:
 
       - name: ROCM-SMI
         run: |
-          rocminfo  | grep "Agent" -A 14
-      - name: Show HIP environment
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
         run: |
-          echo "HIP: $HIP_VISIBLE_DEVICES"
           echo "ROCR: $ROCR_VISIBLE_DEVICES"
 
       - name: Environment

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

@@ -0,0 +1,14 @@
+name: Self-hosted runner (AMD scheduled CI caller)
+
+on:
+  schedule:
+    - cron: "17 2 * * *"
+
+jobs:
+  run_scheduled_amd_ci:
+    name: Trigger Scheduled AMD CI
+    runs-on: ubuntu-22.04
+    if: ${{ always() }}
+    steps:
+      - name: Trigger scheduled AMD CI via workflow_run
+        run: echo "Trigger scheduled AMD CI via workflow_run"

.github/workflows/self-scheduled-amd-mi210-caller.yml

@@ -0,0 +1,19 @@
+name: Self-hosted runner (AMD mi210 scheduled CI caller)
+
+on:
+  workflow_run:
+    workflows: ["Self-hosted runner (AMD scheduled CI caller)"]
+    branches: ["main"]
+    types: [completed]
+  push:
+    branches:
+      - run_amd_scheduled_ci_caller*
+
+jobs:
+  run_amd_ci:
+    name: AMD mi210
+    if: (cancelled() != true) && ((github.event_name == 'schedule') || ((github.event_name == 'push') && startsWith(github.ref_name, 'run_amd_scheduled_ci_caller')))
+    uses: ./.github/workflows/self-scheduled-amd.yml
+    with:
+      gpu_flavor: mi210
+    secrets: inherit

.github/workflows/self-scheduled-amd-mi250-caller.yml

@@ -0,0 +1,19 @@
+name: Self-hosted runner (AMD mi250 scheduled CI caller)
+
+on:
+  workflow_run:
+    workflows: ["Self-hosted runner (AMD scheduled CI caller)"]
+    branches: ["main"]
+    types: [completed]
+  push:
+    branches:
+      - run_amd_scheduled_ci_caller*
+
+jobs:
+  run_amd_ci:
+    name: AMD mi250
+    if: (cancelled() != true) && ((github.event_name == 'schedule') || ((github.event_name == 'push') && startsWith(github.ref_name, 'run_amd_scheduled_ci_caller')))
+    uses: ./.github/workflows/self-scheduled-amd.yml
+    with:
+      gpu_flavor: mi250
+    secrets: inherit

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

@@ -0,0 +1,461 @@
+name: Self-hosted runner (scheduled-amd)
+
+# Note: For the AMD CI, we rely on a caller workflow and on the workflow_call event to trigger the
+# CI in order to run it on both MI210 and MI250, without having to use matrix here which pushes
+# us towards the limit of allowed jobs on GitHub Actions.
+on:
+  workflow_call:
+    inputs:
+      gpu_flavor:
+        required: true
+        type: string
+
+env:
+  HF_HOME: /mnt/cache
+  TRANSFORMERS_IS_CI: yes
+  OMP_NUM_THREADS: 8
+  MKL_NUM_THREADS: 8
+  RUN_SLOW: yes
+  SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
+
+
+# Important note: each job (run_tests_single_gpu, run_tests_multi_gpu, run_examples_gpu, run_pipelines_torch_gpu) requires all the previous jobs before running.
+# This is done so that we avoid parallelizing the scheduled tests, to leave available
+# runners for the push CI that is running on the same machine.
+jobs:
+  check_runner_status:
+    name: Check Runner Status
+    runs-on: ubuntu-22.04
+    steps:
+      - name: Checkout transformers
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - name: Check Runner Status
+        run: python utils/check_self_hosted_runner.py --target_runners hf-amd-mi210-ci-1gpu-1,hf-amd-mi250-ci-1gpu-1 --token ${{ secrets.ACCESS_REPO_INFO_TOKEN }}
+
+  check_runners:
+    name: Check Runners
+    needs: check_runner_status
+    strategy:
+      matrix:
+        machine_type: [single-gpu, multi-gpu]
+    runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
+    container:
+      image: huggingface/transformers-pytorch-amd-gpu
+      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: ROCM-SMI
+        run: |
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
+        run: |
+          echo "ROCR: $ROCR_VISIBLE_DEVICES"
+
+  setup:
+    name: Setup
+    needs: check_runners
+    strategy:
+      matrix:
+        machine_type: [single-gpu, multi-gpu]
+    runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
+    container:
+      image: huggingface/transformers-pytorch-amd-gpu
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    outputs:
+      matrix: ${{ steps.set-matrix.outputs.matrix }}
+    steps:
+      - name: Update clone
+        working-directory: /transformers
+        run: |
+          git fetch && git checkout ${{ github.sha }}
+
+      - name: Cleanup
+        working-directory: /transformers
+        run: |
+          rm -rf tests/__pycache__
+          rm -rf tests/models/__pycache__
+          rm -rf reports
+
+      - name: Show installed libraries and their versions
+        working-directory: /transformers
+        run: pip freeze
+
+      - id: set-matrix
+        name: Identify models to test
+        working-directory: /transformers/tests
+        run: |
+          echo "matrix=$(python3 -c 'import os; tests = os.getcwd(); model_tests = os.listdir(os.path.join(tests, "models")); d1 = sorted(list(filter(os.path.isdir, os.listdir(tests)))); d2 = sorted(list(filter(os.path.isdir, [f"models/{x}" for x in model_tests]))); d1.remove("models"); d = d2 + d1; print(d)')" >> $GITHUB_OUTPUT
+
+      - 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
+
+  run_tests_single_gpu:
+    name: Single GPU tests
+    strategy:
+      max-parallel: 1  # For now, not to parallelize. Can change later if it works well.
+      fail-fast: false
+      matrix:
+        folders: ${{ fromJson(needs.setup.outputs.matrix) }}
+        machine_type: [single-gpu]
+    runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
+    container:
+      image: huggingface/transformers-pytorch-amd-gpu
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    needs: setup
+    steps:
+      - 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: ROCM-SMI
+        run: |
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
+        run: |
+          echo "ROCR: $ROCR_VISIBLE_DEVICES"
+
+      - name: Environment
+        working-directory: /transformers
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        working-directory: /transformers
+        run: pip freeze
+
+      - name: Run all tests on GPU
+        working-directory: /transformers
+        run: python3 -m pytest -v --make-reports=${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }} tests/${{ matrix.folders }}
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: cat /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}/failures_short.txt
+
+      - name: Test suite reports artifacts
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v3
+        with:
+          name: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports
+          path: /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}
+
+  run_tests_multi_gpu:
+    name: Multi GPU tests
+    strategy:
+      max-parallel: 1
+      fail-fast: false
+      matrix:
+        folders: ${{ fromJson(needs.setup.outputs.matrix) }}
+        machine_type: [multi-gpu]
+    runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
+    container:
+      image: huggingface/transformers-pytorch-amd-gpu
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    needs: setup
+    steps:
+      - 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: ROCM-SMI
+        run: |
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
+        run: |
+          echo "ROCR: $ROCR_VISIBLE_DEVICES"
+
+      - name: Environment
+        working-directory: /transformers
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        working-directory: /transformers
+        run: pip freeze
+
+      - name: Run all tests on GPU
+        working-directory: /transformers
+        run: python3 -m pytest -v --make-reports=${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }} tests/${{ matrix.folders }}
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: cat /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}/failures_short.txt
+
+      - name: Test suite reports artifacts
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v3
+        with:
+          name: ${{ matrix.machine_type }}_run_all_tests_gpu_${{ env.matrix_folders }}_test_reports
+          path: /transformers/reports/${{ matrix.machine_type }}_tests_gpu_${{ matrix.folders }}
+
+  run_examples_gpu:
+    name: Examples tests
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [single-gpu]
+    runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
+    container:
+      image: huggingface/transformers-pytorch-amd-gpu
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    needs: setup
+    steps:
+      - 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: ROCM-SMI
+        run: |
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
+        run: |
+          echo "ROCR: $ROCR_VISIBLE_DEVICES"
+
+      - name: Environment
+        working-directory: /transformers
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        working-directory: /transformers
+        run: pip freeze
+
+      - name: Run examples tests on GPU
+        working-directory: /transformers
+        run: |
+          pip install -r examples/pytorch/_tests_requirements.txt
+          python3 -m pytest -v --make-reports=${{ matrix.machine_type }}_examples_gpu examples/pytorch
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: cat /transformers/reports/${{ matrix.machine_type }}_examples_gpu/failures_short.txt
+
+      - name: Test suite reports artifacts
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v3
+        with:
+          name: ${{ matrix.machine_type }}_run_examples_gpu
+          path: /transformers/reports/${{ matrix.machine_type }}_examples_gpu
+
+  run_pipelines_torch_gpu:
+    name: PyTorch pipelines tests
+    strategy:
+      fail-fast: false
+      matrix:
+        machine_type: [single-gpu, multi-gpu]
+    runs-on: [self-hosted, docker-gpu, amd-gpu, '${{ matrix.machine_type }}', '${{ inputs.gpu_flavor }}']
+    container:
+      image: huggingface/transformers-pytorch-amd-gpu
+      options: --device /dev/kfd --device /dev/dri --env ROCR_VISIBLE_DEVICES --shm-size "16gb" --ipc host -v /mnt/cache/.cache/huggingface:/mnt/cache/
+    needs: setup
+    steps:
+      - 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: ROCM-SMI
+        run: |
+          rocm-smi
+      - name: ROCM-INFO
+        run: |
+          rocminfo  | grep "Agent" -A 14
+      - name: Show ROCR environment
+        run: |
+          echo "ROCR: $ROCR_VISIBLE_DEVICES"
+
+      - name: Environment
+        working-directory: /transformers
+        run: |
+          python3 utils/print_env.py
+
+      - name: Show installed libraries and their versions
+        working-directory: /transformers
+        run: pip freeze
+
+      - name: Run all pipeline tests on GPU
+        working-directory: /transformers
+        run: |
+          python3 -m pytest -n 1 -v --dist=loadfile --make-reports=${{ matrix.machine_type }}_tests_torch_pipeline_gpu tests/pipelines
+
+      - name: Failure short reports
+        if: ${{ failure() }}
+        continue-on-error: true
+        run: cat /transformers/reports/${{ matrix.machine_type }}_tests_torch_pipeline_gpu/failures_short.txt
+
+      - name: Test suite reports artifacts
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v3
+        with:
+          name: ${{ matrix.machine_type }}_run_tests_torch_pipeline_gpu
+          path: /transformers/reports/${{ matrix.machine_type }}_tests_torch_pipeline_gpu
+
+  run_extract_warnings:
+    name: Extract warnings in CI artifacts
+    runs-on: ubuntu-22.04
+    if: always()
+    needs: [
+      check_runner_status,
+      check_runners,
+      setup,
+      run_tests_single_gpu,
+      run_tests_multi_gpu,
+      run_examples_gpu,
+      run_pipelines_torch_gpu,
+      # run_all_tests_torch_cuda_extensions_gpu
+    ]
+    steps:
+      - name: Checkout transformers
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 2
+
+      - name: Install transformers
+        run: pip install transformers
+
+      - name: Show installed libraries and their versions
+        run: pip freeze
+
+      - name: Create output directory
+        run: mkdir warnings_in_ci
+
+      - uses: actions/download-artifact@v3
+        with:
+          path: warnings_in_ci
+
+      - name: Show artifacts
+        run: echo "$(python3 -c 'import os; d = os.listdir(); print(d)')"
+        working-directory: warnings_in_ci
+
+      - name: Extract warnings in CI artifacts
+        run: |
+          python3 utils/extract_warnings.py --workflow_run_id ${{ github.run_id }} --output_dir warnings_in_ci --token ${{ secrets.ACCESS_REPO_INFO_TOKEN }} --from_gh
+          echo "$(python3 -c 'import os; import json; fp = open("warnings_in_ci/selected_warnings.json"); d = json.load(fp); d = "\n".join(d) ;print(d)')"
+
+      - name: Upload artifact
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v3
+        with:
+          name: warnings_in_ci
+          path: warnings_in_ci/selected_warnings.json
+
+  send_results:
+    name: Send results to webhook
+    runs-on: ubuntu-22.04
+    if: always()
+    needs: [
+      check_runner_status,
+      check_runners,
+      setup,
+      run_tests_single_gpu,
+      run_tests_multi_gpu,
+      run_examples_gpu,
+      run_pipelines_torch_gpu,
+      # run_all_tests_torch_cuda_extensions_gpu,
+      run_extract_warnings
+    ]
+    steps:
+      - name: Preliminary job status
+        shell: bash
+        # For the meaning of these environment variables, see the job `Setup`
+        run: |
+          echo "Runner availability: ${{ needs.check_runner_status.result }}"
+          echo "Runner status: ${{ needs.check_runners.result }}"
+          echo "Setup status: ${{ needs.setup.result }}"
+
+      - uses: actions/checkout@v3
+      - uses: actions/download-artifact@v3
+      - name: Send message to Slack
+        env:
+          CI_SLACK_BOT_TOKEN: ${{ secrets.CI_SLACK_BOT_TOKEN }}
+          CI_SLACK_CHANNEL_ID_DAILY_AMD: ${{ secrets.CI_SLACK_CHANNEL_ID_DAILY_AMD }}
+          CI_SLACK_CHANNEL_DUMMY_TESTS: ${{ secrets.CI_SLACK_CHANNEL_DUMMY_TESTS }}
+          CI_SLACK_REPORT_CHANNEL_ID: ${{ secrets.CI_SLACK_CHANNEL_ID_DAILY_AMD }}
+          ACCESS_REPO_INFO_TOKEN: ${{ secrets.ACCESS_REPO_INFO_TOKEN }}
+          CI_EVENT: Scheduled CI (AMD) - ${{ inputs.gpu_flavor }}
+          CI_SHA: ${{ github.sha }}
+          CI_WORKFLOW_REF: ${{ github.workflow_ref }}
+          RUNNER_STATUS: ${{ needs.check_runner_status.result }}
+          RUNNER_ENV_STATUS: ${{ needs.check_runners.result }}
+          SETUP_STATUS: ${{ needs.setup.result }}
+        # We pass `needs.setup.outputs.matrix` as the argument. A processing in `notification_service.py` to change
+        # `models/bert` to `models_bert` is required, as the artifact names use `_` instead of `/`.
+        run: |
+          sudo apt-get install -y curl
+          pip install slack_sdk
+          pip show slack_sdk
+          python utils/notification_service.py "${{ needs.setup.outputs.matrix }}"
+
+      # Upload complete failure tables, as they might be big and only truncated versions could be sent to Slack.
+      - name: Failure table artifacts
+        if: ${{ always() }}
+        uses: actions/upload-artifact@v3
+        with:
+          name: test_failure_tables
+          path: test_failure_tables

.github/workflows/self-scheduled.yml

@@ -494,5 +494,5 @@ jobs:
         if: ${{ always() }}
         uses: actions/upload-artifact@v3
         with:
-          name: test_failure_tables
-          path: test_failure_tables
+          name: prev_ci_results
+          path: prev_ci_results

ISSUES.md

@@ -152,7 +152,7 @@ You are not required to read the following guidelines before opening an issue. H
 
    ```bash
     cd examples/seq2seq
-    python -m torch.distributed.launch --nproc_per_node=2 ./finetune_trainer.py \
+    torchrun --nproc_per_node=2 ./finetune_trainer.py \
     --model_name_or_path sshleifer/distill-mbart-en-ro-12-4 --data_dir wmt_en_ro \
     --output_dir output_dir --overwrite_output_dir \
     --do_train --n_train 500 --num_train_epochs 1 \

Makefile

@@ -9,8 +9,8 @@ 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)"; \
-		black $(modified_py_files); \
-		ruff $(modified_py_files) --fix; \
+		ruff check $(modified_py_files) --fix; \
+		ruff format $(modified_py_files);\
 	else \
 		echo "No library .py files were modified"; \
 	fi
@@ -48,11 +48,10 @@ repo-consistency:
 # this target runs checks on all files
 
 quality:
-	black --check $(check_dirs) setup.py conftest.py
+	ruff check $(check_dirs) setup.py conftest.py
+	ruff format --check $(check_dirs) setup.py conftest.py
 	python utils/custom_init_isort.py --check_only
 	python utils/sort_auto_mappings.py --check_only
-	ruff $(check_dirs) setup.py conftest.py
-	doc-builder style src/transformers docs/source --max_len 119 --check_only --path_to_docs docs/source
 	python utils/check_doc_toc.py
 
 # Format source code automatically and check is there are any problems left that need manual fixing
@@ -60,14 +59,13 @@ quality:
 extra_style_checks:
 	python utils/custom_init_isort.py
 	python utils/sort_auto_mappings.py
-	doc-builder style src/transformers docs/source --max_len 119 --path_to_docs docs/source
 	python utils/check_doc_toc.py --fix_and_overwrite
 
 # this target runs checks on all files and potentially modifies some of them
 
 style:
-	black $(check_dirs) setup.py conftest.py
-	ruff $(check_dirs) setup.py conftest.py --fix
+	ruff check $(check_dirs) setup.py conftest.py --fix
+	ruff format $(check_dirs) setup.py conftest.py
 	${MAKE} autogenerate_code
 	${MAKE} extra_style_checks
 

README.md

@@ -321,7 +321,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (from LAION-AI) released with the paper [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) by Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker.
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (from MetaAI) released with the paper [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) by Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
@@ -397,12 +397,14 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (from The FAIR team of Meta AI) released with the paper [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971) by Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample.
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (from The FAIR team of Meta AI) released with the paper [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/) by Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom.
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (from Microsoft Research & University of Wisconsin-Madison) released with the paper [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485) by Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee.
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.
@@ -415,6 +417,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (from Alibaba Research) released with the paper [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) by Peng Wang, Cheng Da, and Cong Yao.
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (from Facebook) released with the paper [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) by Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
@@ -439,11 +442,13 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (from Google AI) released with the paper [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683) by Matthias Minderer, Alexey Gritsenko, Neil Houlsby.
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** (from  IBM Research) released with the paper [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf) by Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (from IBM) released with the paper [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/abs/2211.14730) by Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (from ADEPT) released in a [blog post](https://www.adept.ai/blog/persimmon-8b) by Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani.
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (from Google) released with the paper [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) by Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
@@ -464,6 +469,7 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (from Bo Peng), released on [this repo](https://github.com/BlinkDL/RWKV-LM) by Bo Peng.
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
@@ -489,10 +495,12 @@ Current number of checkpoints: ![](https://img.shields.io/endpoint?url=https://h
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim.
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (from Peking University) released with the paper [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) by Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.

README_es.md

@@ -296,7 +296,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (from LAION-AI) released with the paper [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) by Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (from MetaAI) released with the paper [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) by Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
@@ -372,12 +372,14 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (from The FAIR team of Meta AI) released with the paper [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971) by Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample.
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (from The FAIR team of Meta AI) released with the paper [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/XXX) by Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom..
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (from Microsoft Research & University of Wisconsin-Madison) released with the paper [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485) by Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee.
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.
@@ -390,6 +392,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (from Alibaba Research) released with the paper [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) by Peng Wang, Cheng Da, and Cong Yao.
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (from Facebook) released with the paper [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) by Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
@@ -414,11 +417,13 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (from Google AI) released with the paper [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683) by Matthias Minderer, Alexey Gritsenko, Neil Houlsby.
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** (from  IBM Research) released with the paper [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf) by Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (from IBM) released with the paper [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/pdf/2211.14730.pdf) by Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, and Peter J. Liu.
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (from ADEPT) released with the paper [blog post](https://www.adept.ai/blog/persimmon-8b) by Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani.
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (from Google) released with the paper [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) by Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
@@ -439,6 +444,7 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (from Bo Peng) released with the paper [this repo](https://github.com/BlinkDL/RWKV-LM) by Bo Peng.
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
@@ -464,10 +470,12 @@ Número actual de puntos de control: ![](https://img.shields.io/endpoint?url=htt
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft), released together with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (from Peking University) released with the paper [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) by Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/abs/2202.09741) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.

README_hd.md

@@ -270,7 +270,7 @@ conda install -c huggingface transformers
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (LAION-AI से) Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov. द्वाराअनुसंधान पत्र [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) के साथ जारी किया गया
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI से) साथ वाला पेपर [लर्निंग ट्रांसफरेबल विजुअल मॉडल फ्रॉम नेचुरल लैंग्वेज सुपरविजन](https://arxiv.org /abs/2103.00020) एलेक रैडफोर्ड, जोंग वूक किम, क्रिस हैलासी, आदित्य रमेश, गेब्रियल गोह, संध्या अग्रवाल, गिरीश शास्त्री, अमांडा एस्केल, पामेला मिश्किन, जैक क्लार्क, ग्रेचेन क्रुएगर, इल्या सुत्स्केवर द्वारा।
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (सेल्सफोर्स से) साथ में पेपर [प्रोग्राम सिंथेसिस के लिए एक संवादात्मक प्रतिमान](https://arxiv.org/abs/2203.13474) एरिक निजकैंप, बो पैंग, हिरोआकी हयाशी, लिफू तू, हुआन वांग, यिंगबो झोउ, सिल्वियो सावरेस, कैमिंग जिओंग रिलीज।
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (MetaAI से) Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve. द्वाराअनुसंधान पत्र [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) के साथ जारी किया गया
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (माइक्रोसॉफ्ट रिसर्च एशिया से) कागज के साथ [फास्ट ट्रेनिंग कन्वर्जेंस के लिए सशर्त डीईटीआर](https://arxiv. org/abs/2108.06152) डेपू मेंग, ज़ियाओकांग चेन, ज़ेजिया फैन, गैंग ज़ेंग, होउकियांग ली, युहुई युआन, लेई सन, जिंगडोंग वांग द्वारा।
@@ -346,12 +346,14 @@ conda install -c huggingface transformers
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (दक्षिण चीन प्रौद्योगिकी विश्वविद्यालय से) साथ में कागज [LiLT: एक सरल लेकिन प्रभावी भाषा-स्वतंत्र लेआउट ट्रांसफार्मर संरचित दस्तावेज़ समझ के लिए](https://arxiv.org/abs/2202.13669) जियापेंग वांग, लियानवेन जिन, काई डिंग द्वारा पोस्ट किया गया।
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (The FAIR team of Meta AI से) Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample. द्वाराअनुसंधान पत्र [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971) के साथ जारी किया गया
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (The FAIR team of Meta AI से) Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom.. द्वाराअनुसंधान पत्र [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/XXX) के साथ जारी किया गया
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (Microsoft Research & University of Wisconsin-Madison से) Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee. द्वाराअनुसंधान पत्र [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485) के साथ जारी किया गया
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (मैंडी गुओ, जोशुआ आइंस्ली, डेविड यूथस, सैंटियागो ओंटानन, जियानमो नि, यूं-हुआन सुंग, यिनफेई यांग द्वारा पोस्ट किया गया।
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (स्टूडियो औसिया से) साथ में पेपर [LUKE: डीप कॉन्टेक्स्टुअलाइज्ड एंटिटी रिप्रेजेंटेशन विद एंटिटी-अवेयर सेल्फ-अटेंशन](https ://arxiv.org/abs/2010.01057) Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto द्वारा।
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (UNC चैपल हिल से) साथ में पेपर [LXMERT: ओपन-डोमेन क्वेश्चन के लिए ट्रांसफॉर्मर से क्रॉस-मोडलिटी एनकोडर रिप्रेजेंटेशन सीखना Answering](https://arxiv.org/abs/1908.07490) हाओ टैन और मोहित बंसल द्वारा।
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (फेसबुक से) साथ देने वाला पेपर [बियॉन्ड इंग्लिश-सेंट्रिक मल्टीलिंगुअल मशीन ट्रांसलेशन](https://arxiv.org/ एब्स/2010.11125) एंजेला फैन, श्रुति भोसले, होल्गर श्वेन्क, झी मा, अहमद अल-किश्की, सिद्धार्थ गोयल, मनदीप बैनेस, ओनूर सेलेबी, गुइल्लाम वेन्जेक, विश्रव चौधरी, नमन गोयल, टॉम बर्च, विटाली लिपचिंस्की, सर्गेई एडुनोव, एडौर्ड द्वारा ग्रेव, माइकल औली, आर्मंड जौलिन द्वारा पोस्ट किया गया।
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Jörg द्वारा [OPUS](http://opus.nlpl.eu/) डेटा से प्रशिक्षित मशीनी अनुवाद मॉडल पोस्ट किया गया टाइडेमैन द्वारा। [मैरियन फ्रेमवर्क](https://marian-nmt.github.io/) माइक्रोसॉफ्ट ट्रांसलेटर टीम द्वारा विकसित।
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (माइक्रोसॉफ्ट रिसर्च एशिया से) साथ में पेपर [मार्कअपएलएम: विजुअली-रिच डॉक्यूमेंट अंडरस्टैंडिंग के लिए टेक्स्ट और मार्कअप लैंग्वेज का प्री-ट्रेनिंग] (https://arxiv.org/abs/2110.08518) जुनलॉन्ग ली, यिहेंग जू, लेई कुई, फुरु द्वारा वी द्वारा पोस्ट किया गया।
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (FAIR and UIUC से) Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar. द्वाराअनुसंधान पत्र [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) के साथ जारी किया गया
@@ -364,6 +366,7 @@ conda install -c huggingface transformers
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA से) साथ वाला पेपर [Megatron-LM: ट्रेनिंग मल्टी-बिलियन पैरामीटर लैंग्वेज मॉडल्स यूजिंग मॉडल पैरेललिज़्म] (https://arxiv.org/abs/1909.08053) मोहम्मद शोएबी, मोस्टोफा पटवारी, राउल पुरी, पैट्रिक लेग्रेस्ले, जेरेड कैस्पर और ब्रायन कैटानज़ारो द्वारा पोस्ट किया गया।
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (Alibaba Research से) Peng Wang, Cheng Da, and Cong Yao. द्वाराअनुसंधान पत्र [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) के साथ जारी किया गया
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (फ्रॉम Studio Ousia) साथ में पेपर [mLUKE: द पावर ऑफ एंटिटी रिप्रेजेंटेशन इन मल्टीलिंगुअल प्रीट्रेन्ड लैंग्वेज मॉडल्स](https://arxiv.org/abs/2110.08151) रयोकन री, इकुया यामाडा, और योशिमासा त्सुरोका द्वारा।
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (Facebook से) Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli. द्वाराअनुसंधान पत्र [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) के साथ जारी किया गया
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (सीएमयू/गूगल ब्रेन से) साथ में कागज [मोबाइलबर्ट: संसाधन-सीमित उपकरणों के लिए एक कॉम्पैक्ट टास्क-अज्ञेय बीईआरटी] (https://arxiv.org/abs/2004.02984) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, और Denny Zhou द्वारा पोस्ट किया गया।
@@ -388,11 +391,13 @@ conda install -c huggingface transformers
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (Google AI से) साथ में कागज [विज़न ट्रांसफॉर्मर्स के साथ सिंपल ओपन-वोकैबुलरी ऑब्जेक्ट डिटेक्शन](https:/ /arxiv.org/abs/2205.06230) मैथियास मिंडरर, एलेक्सी ग्रिट्सेंको, ऑस्टिन स्टोन, मैक्सिम न्यूमैन, डिर्क वीसेनबोर्न, एलेक्सी डोसोवित्स्की, अरविंद महेंद्रन, अनुराग अर्नब, मुस्तफा देहघानी, ज़ुओरन शेन, जिओ वांग, ज़ियाओहुआ झाई, थॉमस किफ़, और नील हॉल्सबी द्वारा पोस्ट किया गया।
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (Google AI से) Matthias Minderer, Alexey Gritsenko, Neil Houlsby. द्वाराअनुसंधान पत्र [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683) के साथ जारी किया गया
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** ( IBM Research से) Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam. द्वाराअनुसंधान पत्र [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf) के साथ जारी किया गया
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (IBM से) Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam. द्वाराअनुसंधान पत्र [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/pdf/2211.14730.pdf) के साथ जारी किया गया
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (Google की ओर से) साथ में दिया गया पेपर [लंबे इनपुट सारांश के लिए ट्रांसफ़ॉर्मरों को बेहतर तरीके से एक्सटेंड करना](https://arxiv .org/abs/2208.04347) जेसन फांग, याओ झाओ, पीटर जे लियू द्वारा।
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (दीपमाइंड से) साथ में पेपर [पर्सीवर आईओ: संरचित इनपुट और आउटपुट के लिए एक सामान्य वास्तुकला] (https://arxiv.org/abs/2107.14795) एंड्रयू जेगल, सेबेस्टियन बोरग्यूड, जीन-बैप्टिस्ट अलायराक, कार्ल डोर्श, कैटलिन इओनेस्कु, डेविड द्वारा डिंग, स्कंद कोप्पुला, डैनियल ज़ोरान, एंड्रयू ब्रॉक, इवान शेलहैमर, ओलिवियर हेनाफ, मैथ्यू एम। बोट्विनिक, एंड्रयू ज़िसरमैन, ओरिओल विनियल्स, जोआओ कैरेरा द्वारा पोस्ट किया गया।
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (ADEPT से) Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani. द्वाराअनुसंधान पत्र [blog post](https://www.adept.ai/blog/persimmon-8b) के साथ जारी किया गया
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (VinAI Research से) कागज के साथ [PhoBERT: वियतनामी के लिए पूर्व-प्रशिक्षित भाषा मॉडल](https://www .aclweb.org/anthology/2020.findings-emnlp.92/) डैट क्वोक गुयेन और अन्ह तुआन गुयेन द्वारा पोस्ट किया गया।
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (Google से) Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova. द्वाराअनुसंधान पत्र [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) के साथ जारी किया गया
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP से) साथ वाला पेपर [प्रोग्राम अंडरस्टैंडिंग एंड जेनरेशन के लिए यूनिफाइड प्री-ट्रेनिंग](https://arxiv .org/abs/2103.06333) वसी उद्दीन अहमद, सैकत चक्रवर्ती, बैशाखी रे, काई-वेई चांग द्वारा।
@@ -413,6 +418,7 @@ conda install -c huggingface transformers
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (झुईई टेक्नोलॉजी से), साथ में पेपर [रोफॉर्मर: रोटरी पोजिशन एंबेडिंग के साथ एन्हांस्ड ट्रांसफॉर्मर] (https://arxiv.org/pdf/2104.09864v1.pdf) जियानलिन सु और यू लू और शेंगफेंग पैन और बो वेन और युनफेंग लियू द्वारा प्रकाशित।
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (Bo Peng से) Bo Peng. द्वाराअनुसंधान पत्र [this repo](https://github.com/BlinkDL/RWKV-LM) के साथ जारी किया गया
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI से) Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick. द्वाराअनुसंधान पत्र [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) के साथ जारी किया गया
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP से) साथ देने वाला पेपर [भाषण पहचान के लिए अनसुपरवाइज्ड प्री-ट्रेनिंग में परफॉर्मेंस-एफिशिएंसी ट्रेड-ऑफ्स](https ://arxiv.org/abs/2109.06870) फेलिक्स वू, क्वांगयुन किम, जिंग पैन, क्यू हान, किलियन क्यू. वेनबर्गर, योव आर्टज़ी द्वारा।
@@ -438,10 +444,12 @@ conda install -c huggingface transformers
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU की ओर से) कागज के साथ [संस्करण-एक्स: एक ब्लॉग मॉडल चौकस चौक मॉडल मॉडल] (https://arxivorg/abs/1901.02860) क्वोकोक वी. ले, रुस्लैन सलाखुतदी
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research से) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. द्वाराअनुसंधान पत्र [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) के साथ जारी किया गया
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (माइक्रोसॉफ्ट रिसर्च से) साथ में दिया गया पेपर [UniSpeech: यूनिफाइड स्पीच रिप्रेजेंटेशन लर्निंग विद लेबलेड एंड अनलेबल्ड डेटा](https:/ /arxiv.org/abs/2101.07597) चेंगई वांग, यू वू, याओ कियान, केनिची कुमातानी, शुजी लियू, फुरु वेई, माइकल ज़ेंग, ज़ुएदोंग हुआंग द्वारा।
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (माइक्रोसॉफ्ट रिसर्च से) कागज के साथ [UNISPEECH-SAT: यूनिवर्सल स्पीच रिप्रेजेंटेशन लर्निंग विद स्पीकर अवेयर प्री-ट्रेनिंग ](https://arxiv.org/abs/2110.05752) सानयुआन चेन, यू वू, चेंग्यी वांग, झेंगयांग चेन, झूओ चेन, शुजी लियू, जियान वू, याओ कियान, फुरु वेई, जिन्यु ली, जियांगज़ान यू द्वारा पोस्ट किया गया।
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (from Peking University) released with the paper [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) by Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (सिंघुआ यूनिवर्सिटी और ननकाई यूनिवर्सिटी से) साथ में पेपर [विजुअल अटेंशन नेटवर्क](https://arxiv.org/ pdf/2202.09741.pdf) मेंग-हाओ गुओ, चेंग-ज़े लू, झेंग-निंग लियू, मिंग-मिंग चेंग, शि-मिन हू द्वारा।
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (मल्टीमीडिया कम्प्यूटिंग ग्रुप, नानजिंग यूनिवर्सिटी से) साथ में पेपर [वीडियोएमएई: मास्क्ड ऑटोएन्कोडर स्व-पर्यवेक्षित वीडियो प्री-ट्रेनिंग के लिए डेटा-कुशल सीखने वाले हैं] (https://arxiv.org/abs/2203.12602) ज़ान टोंग, यिबिंग सॉन्ग, जुए द्वारा वांग, लिमिन वांग द्वारा पोस्ट किया गया।

README_ja.md

@@ -330,7 +330,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (LAION-AI から) Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov. から公開された研究論文 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687)
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI から) Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever から公開された研究論文: [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020)
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (University of Göttingen から) Timo Lüddecke and Alexander Ecker から公開された研究論文: [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003)
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (Salesforce から) Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong から公開された研究論文: [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474)
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (MetaAI から) Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve. から公開された研究論文 [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/)
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (Microsoft Research Asia から) Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang から公開された研究論文: [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152)
@@ -406,12 +406,14 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (South China University of Technology から) Jiapeng Wang, Lianwen Jin, Kai Ding から公開された研究論文: [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669)
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (The FAIR team of Meta AI から) Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample. から公開された研究論文 [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971)
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (The FAIR team of Meta AI から) Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom.. から公開された研究論文 [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/XXX)
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (Microsoft Research & University of Wisconsin-Madison から) Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee. から公開された研究論文 [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485)
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (AllenAI から) Iz Beltagy, Matthew E. Peters, Arman Cohan から公開された研究論文: [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150)
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (Google AI から) Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang から公開された研究論文: [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916)
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (Studio Ousia から) Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto から公開された研究論文: [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057)
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (UNC Chapel Hill から) Hao Tan and Mohit Bansal から公開された研究論文: [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490)
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (Facebook から) Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert から公開された研究論文: [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161)
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (Facebook から) Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin から公開された研究論文: [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125)
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Jörg Tiedemann から. [OPUS](http://opus.nlpl.eu/) を使いながら学習された "Machine translation" (マシントランスレーション) モデル. [Marian Framework](https://marian-nmt.github.io/) はMicrosoft Translator Team　が現在開発中です.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (Microsoft Research Asia から) Junlong Li, Yiheng Xu, Lei Cui, Furu Wei から公開された研究論文: [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518)
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (FAIR and UIUC から) Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar. から公開された研究論文 [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527)
@@ -424,6 +426,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA から) Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro から公開された研究論文: [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053)
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (Alibaba Research から) Peng Wang, Cheng Da, and Cong Yao. から公開された研究論文 [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592)
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (Studio Ousia から) Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka から公開された研究論文: [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151)
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (Facebook から) Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli. から公開された研究論文 [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516)
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (CMU/Google Brain から) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou から公開された研究論文: [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984)
@@ -448,11 +451,13 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (Meta AI から) Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al から公開された研究論文: [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068)
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (Google AI から) Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby から公開された研究論文: [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230)
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (Google AI から) Matthias Minderer, Alexey Gritsenko, Neil Houlsby. から公開された研究論文 [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683)
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** ( IBM Research から) Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam. から公開された研究論文 [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf)
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (IBM から) Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam. から公開された研究論文 [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/pdf/2211.14730.pdf)
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (Google から) Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu から公開された研究論文: [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777)
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (Google から) Jason Phang, Yao Zhao, and Peter J. Liu から公開された研究論文: [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347)
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (Deepmind から) Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira から公開された研究論文: [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795)
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (ADEPT から) Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani. から公開された研究論文 [blog post](https://www.adept.ai/blog/persimmon-8b)
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (VinAI Research から) Dat Quoc Nguyen and Anh Tuan Nguyen から公開された研究論文: [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/)
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (Google から) Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova. から公開された研究論文 [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347)
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP から) Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang から公開された研究論文: [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333)
@@ -473,6 +478,7 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (ZhuiyiTechnology から), Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu から公開された研究論文: [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/abs/2104.09864)
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (Bo Peng から) Bo Peng. から公開された研究論文 [this repo](https://github.com/BlinkDL/RWKV-LM)
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (NVIDIA から) Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo から公開された研究論文: [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203)
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI から) Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick. から公開された研究論文 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf)
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP から) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi から公開された研究論文: [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870)
@@ -498,10 +504,12 @@ Flax、PyTorch、TensorFlowをcondaでインストールする方法は、それ
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU から) Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov から公開された研究論文: [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860)
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (Microsoft から), Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei から公開された研究論文: [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282)
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill から), Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal から公開された研究論文: [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156)
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (Intel から), Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding から公開された研究論文: [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995)
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (Google Research から) Yi Tay, Mostafa Dehghani, Vinh Q から公開された研究論文: [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research から) Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant. から公開された研究論文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research から) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang から公開された研究論文: [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597)
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (Microsoft Research から) Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu から公開された研究論文: [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752)
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (Peking University から) Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun. から公開された研究論文 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221)
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (Tsinghua University and Nankai University から) Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu から公開された研究論文: [Visual Attention Network](https://arxiv.org/abs/2202.09741)
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (Multimedia Computing Group, Nanjing University から) Zhan Tong, Yibing Song, Jue Wang, Limin Wang から公開された研究論文: [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602)

README_ko.md

@@ -245,7 +245,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (LAION-AI 에서 제공)은 Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.의 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687)논문과 함께 발표했습니다.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (OpenAI 에서) Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 의 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 논문과 함께 발표했습니다.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (University of Göttingen 에서) Timo Lüddecke and Alexander Ecker 의 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 논문과 함께 발표했습니다.
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (Salesforce 에서) Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 의 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 논문과 함께 발표했습니다.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (MetaAI 에서 제공)은 Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.의 [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/)논문과 함께 발표했습니다.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (Microsoft Research Asia 에서) Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 의 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 논문과 함께 발표했습니다.
@@ -321,12 +321,14 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (South China University of Technology 에서) Jiapeng Wang, Lianwen Jin, Kai Ding 의 [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) 논문과 함께 발표했습니다.
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (The FAIR team of Meta AI 에서 제공)은 Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample.의 [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971)논문과 함께 발표했습니다.
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (The FAIR team of Meta AI 에서 제공)은 Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom..의 [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/XXX)논문과 함께 발표했습니다.
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (Microsoft Research & University of Wisconsin-Madison 에서 제공)은 Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee.의 [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485)논문과 함께 발표했습니다.
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (AllenAI 에서) Iz Beltagy, Matthew E. Peters, Arman Cohan 의 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 논문과 함께 발표했습니다.
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (Google AI 에서) Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang 의 [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) 논문과 함께 발표했습니다.
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (Studio Ousia 에서) Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto 의 [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) 논문과 함께 발표했습니다.
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (UNC Chapel Hill 에서) Hao Tan and Mohit Bansal 의 [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) 논문과 함께 발표했습니다.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (Facebook 에서) Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert 의 [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) 논문과 함께 발표했습니다.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (Facebook 에서) Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin 의 [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) 논문과 함께 발표했습니다.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (Microsoft Research Asia 에서) Junlong Li, Yiheng Xu, Lei Cui, Furu Wei 의 [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) 논문과 함께 발표했습니다.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (FAIR and UIUC 에서 제공)은 Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.의 [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527)논문과 함께 발표했습니다.
@@ -339,6 +341,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (NVIDIA 에서) Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 의 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 논문과 함께 발표했습니다.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (Alibaba Research 에서 제공)은 Peng Wang, Cheng Da, and Cong Yao.의 [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592)논문과 함께 발표했습니다.
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (Studio Ousia 에서) Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 의 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 논문과 함께 발표했습니다.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (Facebook 에서 제공)은 Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.의 [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516)논문과 함께 발표했습니다.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (CMU/Google Brain 에서) Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 의 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 논문과 함께 발표했습니다.
@@ -363,11 +366,13 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (Meta AI 에서) Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 의 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 논문과 함께 발표했습니다.
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (Google AI 에서) Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby 의 [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) 논문과 함께 발표했습니다.
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (Google AI 에서 제공)은 Matthias Minderer, Alexey Gritsenko, Neil Houlsby.의 [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683)논문과 함께 발표했습니다.
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** ( IBM Research 에서 제공)은 Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.의 [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf)논문과 함께 발표했습니다.
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (IBM 에서 제공)은 Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.의 [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/pdf/2211.14730.pdf)논문과 함께 발표했습니다.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (Google 에서) Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 의 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 논문과 함께 발표했습니다.
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (Google 에서) Jason Phang, Yao Zhao, Peter J. Liu 의 [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) 논문과 함께 발표했습니다.
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (Deepmind 에서) Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira 의 [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) 논문과 함께 발표했습니다.
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (ADEPT 에서 제공)은 Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani.의 [blog post](https://www.adept.ai/blog/persimmon-8b)논문과 함께 발표했습니다.
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (VinAI Research 에서) Dat Quoc Nguyen and Anh Tuan Nguyen 의 [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) 논문과 함께 발표했습니다.
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (Google 에서 제공)은 Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.의 [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347)논문과 함께 발표했습니다.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (UCLA NLP 에서) Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 의 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 논문과 함께 발표했습니다.
@@ -388,6 +393,7 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (ZhuiyiTechnology 에서) Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu 의 a [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) 논문과 함께 발표했습니다.
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (Bo Peng 에서 제공)은 Bo Peng.의 [this repo](https://github.com/BlinkDL/RWKV-LM)논문과 함께 발표했습니다.
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (NVIDIA 에서) Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo 의 [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) 논문과 함께 발표했습니다.
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (Meta AI 에서 제공)은 Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.의 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf)논문과 함께 발표했습니다.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (ASAPP 에서) Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 의 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 논문과 함께 발표했습니다.
@@ -413,10 +419,12 @@ Flax, PyTorch, TensorFlow 설치 페이지에서 이들을 conda로 설치하는
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (Google/CMU 에서) Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 의 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 논문과 함께 발표했습니다.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (Microsoft 에서) Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 의 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 논문과 함께 발표했습니다.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill 에서) Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 의 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 논문과 함께 발표했습니다.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (Intel 에서) Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding 의 [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) 논문과 함께 발표했습니다.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (Google Research 에서) Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzle 의 [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) 논문과 함께 발표했습니다.
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (Google Research 에서 제공)은 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.의 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi)논문과 함께 발표했습니다.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (Microsoft Research 에서) Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 의 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 논문과 함께 발표했습니다.
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (Microsoft Research 에서) Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 의 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 논문과 함께 발표했습니다.
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (Peking University 에서 제공)은 Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.의 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221)논문과 함께 발표했습니다.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (Tsinghua University and Nankai University 에서) Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 의 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 논문과 함께 발표했습니다.
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (Multimedia Computing Group, Nanjing University 에서) Zhan Tong, Yibing Song, Jue Wang, Limin Wang 의 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 논문과 함께 발표했습니다.

README_pt-br.md

@@ -409,6 +409,7 @@ Número atual de pontos de verificação: ![](https://img.shields.io/endpoint?ur
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.

README_ru.md

@@ -399,6 +399,7 @@ conda install -c huggingface transformers
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.

README_te.md

@@ -402,6 +402,7 @@ Flax, PyTorch లేదా TensorFlow యొక్క ఇన్‌స్టా
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.

README_zh-hans.md

@@ -269,7 +269,7 @@ conda install -c huggingface transformers
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (来自 LAION-AI) 伴随论文 [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) 由 Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov 发布。
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (来自 OpenAI) 伴随论文 [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) 由 Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever 发布。
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (来自 University of Göttingen) 伴随论文 [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) 由 Timo Lüddecke and Alexander Ecker 发布。
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (来自 Salesforce) 伴随论文 [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) 由 Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong 发布。
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (来自 MetaAI) 伴随论文 [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) 由 Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve 发布。
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (来自 Microsoft Research Asia) 伴随论文 [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) 由 Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang 发布。
@@ -345,12 +345,14 @@ conda install -c huggingface transformers
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (来自 South China University of Technology) 伴随论文 [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) 由 Jiapeng Wang, Lianwen Jin, Kai Ding 发布。
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (来自 The FAIR team of Meta AI) 伴随论文 [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971) 由 Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample 发布。
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (来自 The FAIR team of Meta AI) 伴随论文 [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/XXX) 由 Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom. 发布。
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (来自 Microsoft Research & University of Wisconsin-Madison) 伴随论文 [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485) 由 Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee 发布。
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (来自 AllenAI) 伴随论文 [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) 由 Iz Beltagy, Matthew E. Peters, Arman Cohan 发布。
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (来自 Google AI) released 伴随论文 [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) 由 Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang 发布。
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (来自 Studio Ousia) 伴随论文 [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) 由 Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto 发布。
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (来自 UNC Chapel Hill) 伴随论文 [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) 由 Hao Tan and Mohit Bansal 发布。
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (来自 Facebook) 伴随论文 [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) 由 Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert 发布。
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (来自 Facebook) 伴随论文 [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) 由 Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin 发布。
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** 用 [OPUS](http://opus.nlpl.eu/) 数据训练的机器翻译模型由 Jörg Tiedemann 发布。[Marian Framework](https://marian-nmt.github.io/) 由微软翻译团队开发。
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (来自 Microsoft Research Asia) 伴随论文 [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) 由 Junlong Li, Yiheng Xu, Lei Cui, Furu Wei 发布。
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (来自 FAIR and UIUC) 伴随论文 [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) 由 Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar 发布。
@@ -363,6 +365,7 @@ conda install -c huggingface transformers
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (来自 NVIDIA) 伴随论文 [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) 由 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro 发布。
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (来自 Alibaba Research) 伴随论文 [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) 由 Peng Wang, Cheng Da, and Cong Yao 发布。
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (来自 Studio Ousia) 伴随论文 [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) 由 Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka 发布。
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (来自 Facebook) 伴随论文 [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) 由 Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli 发布。
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (来自 CMU/Google Brain) 伴随论文 [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) 由 Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou 发布。
@@ -387,11 +390,13 @@ conda install -c huggingface transformers
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (来自 Meta AI) 伴随论文 [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) 由 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al 发布。
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (来自 Google AI) 伴随论文 [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) 由 Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby 发布。
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (来自 Google AI) 伴随论文 [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683) 由 Matthias Minderer, Alexey Gritsenko, Neil Houlsby 发布。
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** (来自  IBM Research) 伴随论文 [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf) 由 Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam 发布。
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (来自 IBM) 伴随论文 [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/pdf/2211.14730.pdf) 由 Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam 发布。
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (来自 Google) 伴随论文 [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) 由 Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu 发布。
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (来自 Google) 伴随论文 [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) 由 Jason Phang, Yao Zhao, Peter J. Liu 发布。
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (来自 Deepmind) 伴随论文 [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) 由 Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira 发布。
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (来自 ADEPT) 伴随论文 [blog post](https://www.adept.ai/blog/persimmon-8b) 由 Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani 发布。
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (来自 VinAI Research) 伴随论文 [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) 由 Dat Quoc Nguyen and Anh Tuan Nguyen 发布。
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (来自 Google) 伴随论文 [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) 由 Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova 发布。
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (来自 UCLA NLP) 伴随论文 [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) 由 Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang 发布。
@@ -412,6 +417,7 @@ conda install -c huggingface transformers
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (来自 ZhuiyiTechnology), 伴随论文 [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) 由 Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu 发布。
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (来自 Bo Peng) 伴随论文 [this repo](https://github.com/BlinkDL/RWKV-LM) 由 Bo Peng 发布。
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (来自 NVIDIA) 伴随论文 [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) 由 Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo 发布。
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (来自 Meta AI) 伴随论文 [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) 由 Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick 发布。
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (来自 ASAPP) 伴随论文 [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) 由 Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi 发布。
@@ -437,10 +443,12 @@ conda install -c huggingface transformers
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (来自 Google/CMU) 伴随论文 [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) 由 Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov 发布。
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (来自 Microsoft) 伴随论文 [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) 由 Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei 发布。
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (来自 UNC Chapel Hill) 伴随论文 [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) 由 Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal 发布。
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (来自 Intel) 伴随论文 [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) 由 Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding 发布.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (来自 Google Research) 伴随论文 [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) 由 Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant 发布。
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (来自 Microsoft Research) 伴随论文 [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) 由 Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang 发布。
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (来自 Microsoft Research) 伴随论文 [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) 由 Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu 发布。
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (来自 Peking University) 伴随论文 [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) 由 Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun 发布。
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (来自 Tsinghua University and Nankai University) 伴随论文 [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) 由 Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu 发布。
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (来自 Multimedia Computing Group, Nanjing University) 伴随论文 [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) 由 Zhan Tong, Yibing Song, Jue Wang, Limin Wang 发布。

README_zh-hant.md

@@ -281,7 +281,7 @@ conda install -c huggingface transformers
 1. **[CLAP](https://huggingface.co/docs/transformers/model_doc/clap)** (from LAION-AI) released with the paper [Large-scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation](https://arxiv.org/abs/2211.06687) by Yusong Wu, Ke Chen, Tianyu Zhang, Yuchen Hui, Taylor Berg-Kirkpatrick, Shlomo Dubnov.
 1. **[CLIP](https://huggingface.co/docs/transformers/model_doc/clip)** (from OpenAI) released with the paper [Learning Transferable Visual Models From Natural Language Supervision](https://arxiv.org/abs/2103.00020) by Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, Ilya Sutskever.
 1. **[CLIPSeg](https://huggingface.co/docs/transformers/model_doc/clipseg)** (from University of Göttingen) released with the paper [Image Segmentation Using Text and Image Prompts](https://arxiv.org/abs/2112.10003) by Timo Lüddecke and Alexander Ecker.
-1. **[CLVP](https://huggingface.co/docs/transformers/main/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
+1. **[CLVP](https://huggingface.co/docs/transformers/model_doc/clvp)** released with the paper [Better speech synthesis through scaling](https://arxiv.org/abs/2305.07243) by James Betker. 
 1. **[CodeGen](https://huggingface.co/docs/transformers/model_doc/codegen)** (from Salesforce) released with the paper [A Conversational Paradigm for Program Synthesis](https://arxiv.org/abs/2203.13474) by Erik Nijkamp, Bo Pang, Hiroaki Hayashi, Lifu Tu, Huan Wang, Yingbo Zhou, Silvio Savarese, Caiming Xiong.
 1. **[CodeLlama](https://huggingface.co/docs/transformers/model_doc/llama_code)** (from MetaAI) released with the paper [Code Llama: Open Foundation Models for Code](https://ai.meta.com/research/publications/code-llama-open-foundation-models-for-code/) by Baptiste Rozière, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Jérémy Rapin, Artyom Kozhevnikov, Ivan Evtimov, Joanna Bitton, Manish Bhatt, Cristian Canton Ferrer, Aaron Grattafiori, Wenhan Xiong, Alexandre Défossez, Jade Copet, Faisal Azhar, Hugo Touvron, Louis Martin, Nicolas Usunier, Thomas Scialom, Gabriel Synnaeve.
 1. **[Conditional DETR](https://huggingface.co/docs/transformers/model_doc/conditional_detr)** (from Microsoft Research Asia) released with the paper [Conditional DETR for Fast Training Convergence](https://arxiv.org/abs/2108.06152) by Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang.
@@ -357,12 +357,14 @@ conda install -c huggingface transformers
 1. **[LiLT](https://huggingface.co/docs/transformers/model_doc/lilt)** (from South China University of Technology) released with the paper [LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding](https://arxiv.org/abs/2202.13669) by Jiapeng Wang, Lianwen Jin, Kai Ding.
 1. **[LLaMA](https://huggingface.co/docs/transformers/model_doc/llama)** (from The FAIR team of Meta AI) released with the paper [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971) by Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, Aurelien Rodriguez, Armand Joulin, Edouard Grave, Guillaume Lample.
 1. **[Llama2](https://huggingface.co/docs/transformers/model_doc/llama2)** (from The FAIR team of Meta AI) released with the paper [Llama2: Open Foundation and Fine-Tuned Chat Models](https://ai.meta.com/research/publications/llama-2-open-foundation-and-fine-tuned-chat-models/XXX) by Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton Ferrer, Moya Chen, Guillem Cucurull, David Esiobu, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez Madian Khabsa, Isabel Kloumann, Artem Korenev, Punit Singh Koura, Marie-Anne Lachaux, Thibaut Lavril, Jenya Lee, Diana Liskovich, Yinghai Lu, Yuning Mao, Xavier Martinet, Todor Mihaylov, Pushka rMishra, Igor Molybog, Yixin Nie, Andrew Poulton, Jeremy Reizenstein, Rashi Rungta, Kalyan Saladi, Alan Schelten, Ruan Silva, Eric Michael Smith, Ranjan Subramanian, Xiaoqing EllenTan, Binh Tang, Ross Taylor, Adina Williams, Jian Xiang Kuan, Puxin Xu, Zheng Yan, Iliyan Zarov, Yuchen Zhang, Angela Fan, Melanie Kambadur, Sharan Narang, Aurelien Rodriguez, Robert Stojnic, Sergey Edunov, Thomas Scialom..
+1. **[LLaVa](https://huggingface.co/docs/transformers/model_doc/llava)** (from Microsoft Research & University of Wisconsin-Madison) released with the paper [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485) by Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee.
 1. **[Longformer](https://huggingface.co/docs/transformers/model_doc/longformer)** (from AllenAI) released with the paper [Longformer: The Long-Document Transformer](https://arxiv.org/abs/2004.05150) by Iz Beltagy, Matthew E. Peters, Arman Cohan.
 1. **[LongT5](https://huggingface.co/docs/transformers/model_doc/longt5)** (from Google AI) released with the paper [LongT5: Efficient Text-To-Text Transformer for Long Sequences](https://arxiv.org/abs/2112.07916) by Mandy Guo, Joshua Ainslie, David Uthus, Santiago Ontanon, Jianmo Ni, Yun-Hsuan Sung, Yinfei Yang.
 1. **[LUKE](https://huggingface.co/docs/transformers/model_doc/luke)** (from Studio Ousia) released with the paper [LUKE: Deep Contextualized Entity Representations with Entity-aware Self-attention](https://arxiv.org/abs/2010.01057) by Ikuya Yamada, Akari Asai, Hiroyuki Shindo, Hideaki Takeda, Yuji Matsumoto.
 1. **[LXMERT](https://huggingface.co/docs/transformers/model_doc/lxmert)** (from UNC Chapel Hill) released with the paper [LXMERT: Learning Cross-Modality Encoder Representations from Transformers for Open-Domain Question Answering](https://arxiv.org/abs/1908.07490) by Hao Tan and Mohit Bansal.
 1. **[M-CTC-T](https://huggingface.co/docs/transformers/model_doc/mctct)** (from Facebook) released with the paper [Pseudo-Labeling For Massively Multilingual Speech Recognition](https://arxiv.org/abs/2111.00161) by Loren Lugosch, Tatiana Likhomanenko, Gabriel Synnaeve, and Ronan Collobert.
 1. **[M2M100](https://huggingface.co/docs/transformers/model_doc/m2m_100)** (from Facebook) released with the paper [Beyond English-Centric Multilingual Machine Translation](https://arxiv.org/abs/2010.11125) by Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines, Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky, Sergey Edunov, Edouard Grave, Michael Auli, Armand Joulin.
+1. **[MADLAD-400](https://huggingface.co/docs/transformers/model_doc/madlad-400)** (from Google) released with the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](https://arxiv.org/abs/2309.04662) by Sneha Kudugunta, Isaac Caswell, Biao Zhang, Xavier Garcia, Christopher A. Choquette-Choo, Katherine Lee, Derrick Xin, Aditya Kusupati, Romi Stella, Ankur Bapna, Orhan Firat.
 1. **[MarianMT](https://huggingface.co/docs/transformers/model_doc/marian)** Machine translation models trained using [OPUS](http://opus.nlpl.eu/) data by Jörg Tiedemann. The [Marian Framework](https://marian-nmt.github.io/) is being developed by the Microsoft Translator Team.
 1. **[MarkupLM](https://huggingface.co/docs/transformers/model_doc/markuplm)** (from Microsoft Research Asia) released with the paper [MarkupLM: Pre-training of Text and Markup Language for Visually-rich Document Understanding](https://arxiv.org/abs/2110.08518) by Junlong Li, Yiheng Xu, Lei Cui, Furu Wei.
 1. **[Mask2Former](https://huggingface.co/docs/transformers/model_doc/mask2former)** (from FAIR and UIUC) released with the paper [Masked-attention Mask Transformer for Universal Image Segmentation](https://arxiv.org/abs/2112.01527) by Bowen Cheng, Ishan Misra, Alexander G. Schwing, Alexander Kirillov, Rohit Girdhar.
@@ -375,6 +377,7 @@ conda install -c huggingface transformers
 1. **[Megatron-GPT2](https://huggingface.co/docs/transformers/model_doc/megatron_gpt2)** (from NVIDIA) released with the paper [Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism](https://arxiv.org/abs/1909.08053) by Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper and Bryan Catanzaro.
 1. **[MGP-STR](https://huggingface.co/docs/transformers/model_doc/mgp-str)** (from Alibaba Research) released with the paper [Multi-Granularity Prediction for Scene Text Recognition](https://arxiv.org/abs/2209.03592) by Peng Wang, Cheng Da, and Cong Yao.
 1. **[Mistral](https://huggingface.co/docs/transformers/model_doc/mistral)** (from Mistral AI) by The Mistral AI team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.. 
+1. **[Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral)** (from Mistral AI) by The [Mistral AI](https://mistral.ai) team: Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed. 
 1. **[mLUKE](https://huggingface.co/docs/transformers/model_doc/mluke)** (from Studio Ousia) released with the paper [mLUKE: The Power of Entity Representations in Multilingual Pretrained Language Models](https://arxiv.org/abs/2110.08151) by Ryokan Ri, Ikuya Yamada, and Yoshimasa Tsuruoka.
 1. **[MMS](https://huggingface.co/docs/transformers/model_doc/mms)** (from Facebook) released with the paper [Scaling Speech Technology to 1,000+ Languages](https://arxiv.org/abs/2305.13516) by Vineel Pratap, Andros Tjandra, Bowen Shi, Paden Tomasello, Arun Babu, Sayani Kundu, Ali Elkahky, Zhaoheng Ni, Apoorv Vyas, Maryam Fazel-Zarandi, Alexei Baevski, Yossi Adi, Xiaohui Zhang, Wei-Ning Hsu, Alexis Conneau, Michael Auli.
 1. **[MobileBERT](https://huggingface.co/docs/transformers/model_doc/mobilebert)** (from CMU/Google Brain) released with the paper [MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices](https://arxiv.org/abs/2004.02984) by Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, and Denny Zhou.
@@ -399,11 +402,13 @@ conda install -c huggingface transformers
 1. **[OPT](https://huggingface.co/docs/transformers/master/model_doc/opt)** (from Meta AI) released with the paper [OPT: Open Pre-trained Transformer Language Models](https://arxiv.org/abs/2205.01068) by Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen et al.
 1. **[OWL-ViT](https://huggingface.co/docs/transformers/model_doc/owlvit)** (from Google AI) released with the paper [Simple Open-Vocabulary Object Detection with Vision Transformers](https://arxiv.org/abs/2205.06230) by Matthias Minderer, Alexey Gritsenko, Austin Stone, Maxim Neumann, Dirk Weissenborn, Alexey Dosovitskiy, Aravindh Mahendran, Anurag Arnab, Mostafa Dehghani, Zhuoran Shen, Xiao Wang, Xiaohua Zhai, Thomas Kipf, and Neil Houlsby.
 1. **[OWLv2](https://huggingface.co/docs/transformers/model_doc/owlv2)** (from Google AI) released with the paper [Scaling Open-Vocabulary Object Detection](https://arxiv.org/abs/2306.09683) by Matthias Minderer, Alexey Gritsenko, Neil Houlsby.
+1. **[PatchTSMixer](https://huggingface.co/docs/transformers/model_doc/patchtsmixer)** (from  IBM Research) released with the paper [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf) by Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.
+1. **[PatchTST](https://huggingface.co/docs/transformers/model_doc/patchtst)** (from IBM) released with the paper [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/pdf/2211.14730.pdf) by Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong, Jayant Kalagnanam.
 1. **[Pegasus](https://huggingface.co/docs/transformers/model_doc/pegasus)** (from Google) released with the paper [PEGASUS: Pre-training with Extracted Gap-sentences for Abstractive Summarization](https://arxiv.org/abs/1912.08777) by Jingqing Zhang, Yao Zhao, Mohammad Saleh and Peter J. Liu.
 1. **[PEGASUS-X](https://huggingface.co/docs/transformers/model_doc/pegasus_x)** (from Google) released with the paper [Investigating Efficiently Extending Transformers for Long Input Summarization](https://arxiv.org/abs/2208.04347) by Jason Phang, Yao Zhao, Peter J. Liu.
 1. **[Perceiver IO](https://huggingface.co/docs/transformers/model_doc/perceiver)** (from Deepmind) released with the paper [Perceiver IO: A General Architecture for Structured Inputs & Outputs](https://arxiv.org/abs/2107.14795) by Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, João Carreira.
 1. **[Persimmon](https://huggingface.co/docs/transformers/model_doc/persimmon)** (from ADEPT) released with the paper [blog post](https://www.adept.ai/blog/persimmon-8b) by Erich Elsen, Augustus Odena, Maxwell Nye, Sağnak Taşırlar, Tri Dao, Curtis Hawthorne, Deepak Moparthi, Arushi Somani.
-1. **[Phi](https://huggingface.co/docs/transformers/main/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
+1. **[Phi](https://huggingface.co/docs/transformers/model_doc/phi)** (from Microsoft) released with the papers - [Textbooks Are All You Need](https://arxiv.org/abs/2306.11644) by Suriya Gunasekar, Yi Zhang, Jyoti Aneja, Caio César Teodoro Mendes, Allie Del Giorno, Sivakanth Gopi, Mojan Javaheripi, Piero Kauffmann, Gustavo de Rosa, Olli Saarikivi, Adil Salim, Shital Shah, Harkirat Singh Behl, Xin Wang, Sébastien Bubeck, Ronen Eldan, Adam Tauman Kalai, Yin Tat Lee and Yuanzhi Li, [Textbooks Are All You Need II: phi-1.5 technical report](https://arxiv.org/abs/2309.05463) by Yuanzhi Li, Sébastien Bubeck, Ronen Eldan, Allie Del Giorno, Suriya Gunasekar and Yin Tat Lee.
 1. **[PhoBERT](https://huggingface.co/docs/transformers/model_doc/phobert)** (from VinAI Research) released with the paper [PhoBERT: Pre-trained language models for Vietnamese](https://www.aclweb.org/anthology/2020.findings-emnlp.92/) by Dat Quoc Nguyen and Anh Tuan Nguyen.
 1. **[Pix2Struct](https://huggingface.co/docs/transformers/model_doc/pix2struct)** (from Google) released with the paper [Pix2Struct: Screenshot Parsing as Pretraining for Visual Language Understanding](https://arxiv.org/abs/2210.03347) by Kenton Lee, Mandar Joshi, Iulia Turc, Hexiang Hu, Fangyu Liu, Julian Eisenschlos, Urvashi Khandelwal, Peter Shaw, Ming-Wei Chang, Kristina Toutanova.
 1. **[PLBart](https://huggingface.co/docs/transformers/model_doc/plbart)** (from UCLA NLP) released with the paper [Unified Pre-training for Program Understanding and Generation](https://arxiv.org/abs/2103.06333) by Wasi Uddin Ahmad, Saikat Chakraborty, Baishakhi Ray, Kai-Wei Chang.
@@ -424,6 +429,7 @@ conda install -c huggingface transformers
 1. **[RoFormer](https://huggingface.co/docs/transformers/model_doc/roformer)** (from ZhuiyiTechnology), released together with the paper a [RoFormer: Enhanced Transformer with Rotary Position Embedding](https://arxiv.org/pdf/2104.09864v1.pdf) by Jianlin Su and Yu Lu and Shengfeng Pan and Bo Wen and Yunfeng Liu.
 1. **[RWKV](https://huggingface.co/docs/transformers/model_doc/rwkv)** (from Bo Peng) released with the paper [this repo](https://github.com/BlinkDL/RWKV-LM) by Bo Peng.
 1. **[SeamlessM4T](https://huggingface.co/docs/transformers/model_doc/seamless_m4t)** (from Meta AI) released with the paper [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team.
+1. **[SeamlessM4Tv2](https://huggingface.co/docs/transformers/model_doc/seamless_m4t_v2)** (from Meta AI) released with the paper [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team.
 1. **[SegFormer](https://huggingface.co/docs/transformers/model_doc/segformer)** (from NVIDIA) released with the paper [SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203) by Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo.
 1. **[Segment Anything](https://huggingface.co/docs/transformers/model_doc/sam)** (from Meta AI) released with the paper [Segment Anything](https://arxiv.org/pdf/2304.02643v1.pdf) by Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alex Berg, Wan-Yen Lo, Piotr Dollar, Ross Girshick.
 1. **[SEW](https://huggingface.co/docs/transformers/model_doc/sew)** (from ASAPP) released with the paper [Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition](https://arxiv.org/abs/2109.06870) by Felix Wu, Kwangyoun Kim, Jing Pan, Kyu Han, Kilian Q. Weinberger, Yoav Artzi.
@@ -449,10 +455,12 @@ conda install -c huggingface transformers
 1. **[Transformer-XL](https://huggingface.co/docs/transformers/model_doc/transfo-xl)** (from Google/CMU) released with the paper [Transformer-XL: Attentive Language Models Beyond a Fixed-Length Context](https://arxiv.org/abs/1901.02860) by Zihang Dai*, Zhilin Yang*, Yiming Yang, Jaime Carbonell, Quoc V. Le, Ruslan Salakhutdinov.
 1. **[TrOCR](https://huggingface.co/docs/transformers/model_doc/trocr)** (from Microsoft) released with the paper [TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models](https://arxiv.org/abs/2109.10282) by Minghao Li, Tengchao Lv, Lei Cui, Yijuan Lu, Dinei Florencio, Cha Zhang, Zhoujun Li, Furu Wei.
 1. **[TVLT](https://huggingface.co/docs/transformers/model_doc/tvlt)** (from UNC Chapel Hill) released with the paper [TVLT: Textless Vision-Language Transformer](https://arxiv.org/abs/2209.14156) by Zineng Tang, Jaemin Cho, Yixin Nie, Mohit Bansal.
+1. **[TVP](https://huggingface.co/docs/transformers/model_doc/tvp)** (from Intel) released with the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
 1. **[UL2](https://huggingface.co/docs/transformers/model_doc/ul2)** (from Google Research) released with the paper [Unifying Language Learning Paradigms](https://arxiv.org/abs/2205.05131v1) by Yi Tay, Mostafa Dehghani, Vinh Q. Tran, Xavier Garcia, Dara Bahri, Tal Schuster, Huaixiu Steven Zheng, Neil Houlsby, Donald Metzler
 1. **[UMT5](https://huggingface.co/docs/transformers/model_doc/umt5)** (from Google Research) released with the paper [UniMax: Fairer and More Effective Language Sampling for Large-Scale Multilingual Pretraining](https://openreview.net/forum?id=kXwdL1cWOAi) by Hyung Won Chung, Xavier Garcia, Adam Roberts, Yi Tay, Orhan Firat, Sharan Narang, Noah Constant.
 1. **[UniSpeech](https://huggingface.co/docs/transformers/model_doc/unispeech)** (from Microsoft Research) released with the paper [UniSpeech: Unified Speech Representation Learning with Labeled and Unlabeled Data](https://arxiv.org/abs/2101.07597) by Chengyi Wang, Yu Wu, Yao Qian, Kenichi Kumatani, Shujie Liu, Furu Wei, Michael Zeng, Xuedong Huang.
 1. **[UniSpeechSat](https://huggingface.co/docs/transformers/model_doc/unispeech-sat)** (from Microsoft Research) released with the paper [UNISPEECH-SAT: UNIVERSAL SPEECH REPRESENTATION LEARNING WITH SPEAKER AWARE PRE-TRAINING](https://arxiv.org/abs/2110.05752) by Sanyuan Chen, Yu Wu, Chengyi Wang, Zhengyang Chen, Zhuo Chen, Shujie Liu, Jian Wu, Yao Qian, Furu Wei, Jinyu Li, Xiangzhan Yu.
+1. **[UnivNet](https://huggingface.co/docs/transformers/model_doc/univnet)** (from Kakao Corporation) released with the paper [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kim, and Juntae Kim. 
 1. **[UPerNet](https://huggingface.co/docs/transformers/model_doc/upernet)** (from Peking University) released with the paper [Unified Perceptual Parsing for Scene Understanding](https://arxiv.org/abs/1807.10221) by Tete Xiao, Yingcheng Liu, Bolei Zhou, Yuning Jiang, Jian Sun.
 1. **[VAN](https://huggingface.co/docs/transformers/model_doc/van)** (from Tsinghua University and Nankai University) released with the paper [Visual Attention Network](https://arxiv.org/pdf/2202.09741.pdf) by Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu.
 1. **[VideoMAE](https://huggingface.co/docs/transformers/model_doc/videomae)** (from Multimedia Computing Group, Nanjing University) released with the paper [VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training](https://arxiv.org/abs/2203.12602) by Zhan Tong, Yibing Song, Jue Wang, Limin Wang.

docker/transformers-all-latest-gpu/Dockerfile

@@ -56,7 +56,7 @@ RUN python3 -m pip install --no-cache-dir auto-gptq --extra-index-url https://hu
 RUN python3 -m pip install --no-cache-dir einops
 
 # Add autoawq for quantization testing
-RUN python3 -m pip install --no-cache-dir https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.6/autoawq-0.1.6+cu118-cp38-cp38-linux_x86_64.whl
+RUN python3 -m pip install --no-cache-dir https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.7/autoawq-0.1.7+cu118-cp38-cp38-linux_x86_64.whl
 
 # For bettertransformer + gptq 
 RUN python3 -m pip install --no-cache-dir git+https://github.com/huggingface/optimum@main#egg=optimum

docker/transformers-pytorch-amd-gpu/Dockerfile

@@ -1,24 +1,25 @@
-FROM rocm/pytorch:rocm5.6_ubuntu20.04_py3.8_pytorch_2.0.1
+FROM rocm/dev-ubuntu-20.04:5.6
+# rocm/pytorch has no version with 2.1.0
 LABEL maintainer="Hugging Face"
 
 ARG DEBIAN_FRONTEND=noninteractive
 
+ARG PYTORCH='2.1.0'
+ARG TORCH_VISION='0.16.0'
+ARG TORCH_AUDIO='2.1.0'
+ARG ROCM='5.6'
+
 RUN apt update && \
-    apt install -y --no-install-recommends git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-pip ffmpeg && \
+    apt install -y --no-install-recommends git libsndfile1-dev tesseract-ocr espeak-ng python3 python3-dev python3-pip ffmpeg && \
     apt clean && \
     rm -rf /var/lib/apt/lists/*
 
+RUN python3 -m pip install --no-cache-dir --upgrade pip
+
+RUN python3 -m pip install torch==$PYTORCH torchvision==$TORCH_VISION torchaudio==$TORCH_AUDIO --index-url https://download.pytorch.org/whl/rocm$ROCM
+
 RUN python3 -m pip install --no-cache-dir --upgrade pip setuptools ninja git+https://github.com/facebookresearch/detectron2.git pytesseract "itsdangerous<2.1.0"
 
-# If set to nothing, will install the latest version
-ARG PYTORCH='2.0.1'
-ARG TORCH_VISION='0.15.2'
-ARG TORCH_AUDIO='2.0.2'
-ARG ROCM='5.6'
-
-RUN git clone --depth 1 --branch v$TORCH_AUDIO https://github.com/pytorch/audio.git
-RUN cd audio && USE_ROCM=1 USE_CUDA=0 python setup.py install
-
 ARG REF=main
 WORKDIR /
 RUN git clone https://github.com/huggingface/transformers && cd transformers && git checkout $REF

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

@@ -1,12 +1,12 @@
-# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-22-12.html#rel-22-12
-FROM nvcr.io/nvidia/pytorch:22.12-py3
+# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-23-11.html#rel-23-11
+FROM nvcr.io/nvidia/pytorch:23.11-py3
 LABEL maintainer="Hugging Face"
 
 ARG DEBIAN_FRONTEND=noninteractive
 
 ARG PYTORCH='2.1.0'
 # Example: `cu102`, `cu113`, etc.
-ARG CUDA='cu118'
+ARG CUDA='cu121'
 
 RUN apt -y update
 RUN apt install -y libaio-dev

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

@@ -1,11 +1,11 @@
-# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-22-12.html#rel-22-12
-FROM nvcr.io/nvidia/pytorch:22.12-py3
+# https://docs.nvidia.com/deeplearning/frameworks/pytorch-release-notes/rel-23-11.html#rel-23-11
+FROM nvcr.io/nvidia/pytorch:23.11-py3
 LABEL maintainer="Hugging Face"
 
 ARG DEBIAN_FRONTEND=noninteractive
 
 # Example: `cu102`, `cu113`, etc.
-ARG CUDA='cu118'
+ARG CUDA='cu121'
 
 RUN apt -y update
 RUN apt install -y libaio-dev

docker/transformers-pytorch-gpu/Dockerfile

@@ -1,4 +1,4 @@
-FROM nvidia/cuda:11.8.0-cudnn8-devel-ubuntu20.04
+FROM nvidia/cuda:12.1.0-cudnn8-devel-ubuntu20.04
 LABEL maintainer="Hugging Face"
 
 ARG DEBIAN_FRONTEND=noninteractive
@@ -15,7 +15,7 @@ ARG PYTORCH='2.1.0'
 ARG TORCH_VISION=''
 ARG TORCH_AUDIO=''
 # Example: `cu102`, `cu113`, etc.
-ARG CUDA='cu118'
+ARG CUDA='cu121'
 
 RUN [ ${#PYTORCH} -gt 0 ] && VERSION='torch=='$PYTORCH'.*' ||  VERSION='torch'; python3 -m pip install --no-cache-dir -U $VERSION --extra-index-url https://download.pytorch.org/whl/$CUDA
 RUN [ ${#TORCH_VISION} -gt 0 ] && VERSION='torchvision=='TORCH_VISION'.*' ||  VERSION='torchvision'; python3 -m pip install --no-cache-dir -U $VERSION --extra-index-url https://download.pytorch.org/whl/$CUDA

docs/source/_config.py

@@ -10,5 +10,5 @@ notebook_first_cells = [{"type": "code", "content": INSTALL_CONTENT}]
 black_avoid_patterns = {
     "{processor_class}": "FakeProcessorClass",
     "{model_class}": "FakeModelClass",
-    "{object_class}": "FakeObjectClass",    
+    "{object_class}": "FakeObjectClass",
 }

docs/source/de/preprocessing.md

@@ -209,7 +209,7 @@ Audioeingaben werden anders vorverarbeitet als Texteingaben, aber das Endziel bl
 pip install datasets
 ```
 
-Laden Sie den [MInDS-14](https://huggingface.co/datasets/PolyAI/minds14) Datensatz (weitere Informationen zum Laden eines Datensatzes finden Sie im 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub.html)):
+Laden Sie den [MInDS-14](https://huggingface.co/datasets/PolyAI/minds14) Datensatz (weitere Informationen zum Laden eines Datensatzes finden Sie im 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub)):
 
 ```py
 >>> from datasets import load_dataset, Audio
@@ -344,7 +344,7 @@ Laden wir den [food101](https://huggingface.co/datasets/food101) Datensatz für
 >>> dataset = load_dataset("food101", split="train[:100]")
 ```
 
-Als Nächstes sehen Sie sich das Bild mit dem Merkmal 🤗 Datensätze [Bild] (https://huggingface.co/docs/datasets/package_reference/main_classes.html?highlight=image#datasets.Image) an:
+Als Nächstes sehen Sie sich das Bild mit dem Merkmal 🤗 Datensätze [Bild] (https://huggingface.co/docs/datasets/package_reference/main_classes?highlight=image#datasets.Image) an:
 
 ```py
 >>> dataset[0]["image"]
@@ -385,7 +385,7 @@ Bei Bildverarbeitungsaufgaben ist es üblich, den Bildern als Teil der Vorverarb
 ...     return examples
 ```
 
-3. Dann verwenden Sie 🤗 Datasets [`set_transform`](https://huggingface.co/docs/datasets/process.html#format-transform), um die Transformationen im laufenden Betrieb anzuwenden:
+3. Dann verwenden Sie 🤗 Datasets [`set_transform`](https://huggingface.co/docs/datasets/process#format-transform), um die Transformationen im laufenden Betrieb anzuwenden:
 
 ```py
 >>> dataset.set_transform(transforms)

docs/source/de/quicktour.md

@@ -121,7 +121,7 @@ Erstellen wir eine [`pipeline`] mit der Aufgabe die wir lösen und dem Modell we
 >>> speech_recognizer = pipeline("automatic-speech-recognition", model="facebook/wav2vec2-base-960h")
 ```
 
-Als nächstes laden wir den Datensatz (siehe 🤗 Datasets [Quick Start](https://huggingface.co/docs/datasets/quickstart.html) für mehr Details) welches wir nutzen möchten. Zum Beispiel laden wir den [MInDS-14](https://huggingface.co/datasets/PolyAI/minds14) Datensatz:
+Als nächstes laden wir den Datensatz (siehe 🤗 Datasets [Quick Start](https://huggingface.co/docs/datasets/quickstart) für mehr Details) welches wir nutzen möchten. Zum Beispiel laden wir den [MInDS-14](https://huggingface.co/datasets/PolyAI/minds14) Datensatz:
 
 ```py
 >>> from datasets import load_dataset, Audio

docs/source/de/run_scripts.md

@@ -130,7 +130,7 @@ Der [Trainer](https://huggingface.co/docs/transformers/main_classes/trainer) unt
 - Legen Sie die Anzahl der zu verwendenden GPUs mit dem Argument `nproc_per_node` fest.
 
 ```bash
-python -m torch.distributed.launch \
+torchrun \
     --nproc_per_node 8 pytorch/summarization/run_summarization.py \
     --fp16 \
     --model_name_or_path t5-small \

docs/source/de/training.md

@@ -43,7 +43,7 @@ Laden Sie zunächst den Datensatz [Yelp Reviews](https://huggingface.co/datasets
  'text': 'My expectations for McDonalds are t rarely high. But for one to still fail so spectacularly...that takes something special!\\nThe cashier took my friends\'s order, then promptly ignored me. I had to force myself in front of a cashier who opened his register to wait on the person BEHIND me. I waited over five minutes for a gigantic order that included precisely one kid\'s meal. After watching two people who ordered after me be handed their food, I asked where mine was. The manager started yelling at the cashiers for \\"serving off their orders\\" when they didn\'t have their food. But neither cashier was anywhere near those controls, and the manager was the one serving food to customers and clearing the boards.\\nThe manager was rude when giving me my order. She didn\'t make sure that I had everything ON MY RECEIPT, and never even had the decency to apologize that I felt I was getting poor service.\\nI\'ve eaten at various McDonalds restaurants for over 30 years. I\'ve worked at more than one location. I expect bad days, bad moods, and the occasional mistake. But I have yet to have a decent experience at this store. It will remain a place I avoid unless someone in my party needs to avoid illness from low blood sugar. Perhaps I should go back to the racially biased service of Steak n Shake instead!'}
 ```
 
-Wie Sie nun wissen, benötigen Sie einen Tokenizer, um den Text zu verarbeiten und eine Auffüll- und Abschneidungsstrategie einzubauen, um mit variablen Sequenzlängen umzugehen. Um Ihren Datensatz in einem Schritt zu verarbeiten, verwenden Sie die 🤗 Methode Datasets [`map`](https://huggingface.co/docs/datasets/process.html#map), um eine Vorverarbeitungsfunktion auf den gesamten Datensatz anzuwenden:
+Wie Sie nun wissen, benötigen Sie einen Tokenizer, um den Text zu verarbeiten und eine Auffüll- und Abschneidungsstrategie einzubauen, um mit variablen Sequenzlängen umzugehen. Um Ihren Datensatz in einem Schritt zu verarbeiten, verwenden Sie die 🤗 Methode Datasets [`map`](https://huggingface.co/docs/datasets/process#map), um eine Vorverarbeitungsfunktion auf den gesamten Datensatz anzuwenden:
 
 ```py
 >>> from transformers import AutoTokenizer

docs/source/en/_config.py

@@ -10,5 +10,5 @@ notebook_first_cells = [{"type": "code", "content": INSTALL_CONTENT}]
 black_avoid_patterns = {
     "{processor_class}": "FakeProcessorClass",
     "{model_class}": "FakeModelClass",
-    "{object_class}": "FakeObjectClass",    
+    "{object_class}": "FakeObjectClass",
 }

docs/source/en/_redirects.yml

@@ -1,3 +1,3 @@
 # Optimizing inference
 
-perf_infer_gpu_many: perf_infer_gpu_one
+perf_infer_gpu_many: perf_infer_gpu_one

docs/source/en/_toctree.yml

@@ -60,7 +60,7 @@
     - local: tasks/image_classification
       title: Image classification
     - local: tasks/semantic_segmentation
-      title: Semantic segmentation
+      title: Image segmentation
     - local: tasks/video_classification
       title: Video classification
     - local: tasks/object_detection
@@ -133,6 +133,8 @@
 - sections:
   - local: performance
     title: Overview
+  - local: quantization
+    title: Quantization
   - sections:
     - local: perf_train_gpu_one
       title: Methods and tools for efficient training on a single GPU
@@ -378,6 +380,8 @@
         title: LUKE
       - local: model_doc/m2m_100
         title: M2M100
+      - local: model_doc/madlad-400
+        title: MADLAD-400
       - local: model_doc/marian
         title: MarianMT
       - local: model_doc/markuplm
@@ -392,6 +396,8 @@
         title: MegatronGPT2
       - local: model_doc/mistral
         title: Mistral
+      - local: model_doc/mixtral
+        title: Mixtral
       - local: model_doc/mluke
         title: mLUKE
       - local: model_doc/mobilebert
@@ -614,6 +620,8 @@
         title: Pop2Piano
       - local: model_doc/seamless_m4t
         title: Seamless-M4T
+      - local: model_doc/seamless_m4t_v2
+        title: SeamlessM4T-v2
       - local: model_doc/sew
         title: SEW
       - local: model_doc/sew-d
@@ -628,6 +636,8 @@
         title: UniSpeech
       - local: model_doc/unispeech-sat
         title: UniSpeech-SAT
+      - local: model_doc/univnet
+        title: UnivNet
       - local: model_doc/vits
         title: VITS
       - local: model_doc/wav2vec2
@@ -695,6 +705,8 @@
         title: LayoutXLM
       - local: model_doc/lilt
         title: LiLT
+      - local: model_doc/llava
+        title: Llava
       - local: model_doc/lxmert
         title: LXMERT
       - local: model_doc/matcha
@@ -723,6 +735,8 @@
         title: TrOCR
       - local: model_doc/tvlt
         title: TVLT
+      - local: model_doc/tvp
+        title: TVP
       - local: model_doc/vilt
         title: ViLT
       - local: model_doc/vision-encoder-decoder
@@ -747,6 +761,10 @@
         title: Autoformer
       - local: model_doc/informer
         title: Informer
+      - local: model_doc/patchtsmixer
+        title: PatchTSMixer
+      - local: model_doc/patchtst
+        title: PatchTST
       - local: model_doc/time_series_transformer
         title: Time Series Transformer
       title: Time series models

docs/source/en/chat_templating.md

@@ -376,7 +376,10 @@ input formats. Our default template for models that don't have a class-specific
 ```
 
 If you like this one, here it is in one-liner form, ready to copy into your code. The one-liner also includes
-handy support for "generation prompts" - see the next section for more!
+handy support for [generation prompts](#what-are-generation-prompts), but note that it doesn't add BOS or EOS tokens!
+If your model expects those, they won't be added automatically by `apply_chat_template` - in other words, the
+text will be tokenized with `add_special_tokens=False`. This is to avoid potential conflicts between the template and
+the `add_special_tokens` logic. If your model expects special tokens, make sure to add them to the template!
 
 ```
 tokenizer.chat_template = "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}"

docs/source/en/custom_models.md

@@ -14,7 +14,7 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Sharing custom models
+# Building custom models
 
 The 🤗 Transformers library is designed to be easily extensible. Every model is fully coded in a given subfolder
 of the repository with no abstraction, so you can easily copy a modeling file and tweak it to your needs.
@@ -22,7 +22,8 @@ of the repository with no abstraction, so you can easily copy a modeling file an
 If you are writing a brand new model, it might be easier to start from scratch. In this tutorial, we will show you
 how to write a custom model and its configuration so it can be used inside Transformers, and how you can share it
 with the community (with the code it relies on) so that anyone can use it, even if it's not present in the 🤗
-Transformers library.
+Transformers library. We'll see how to build upon transformers and extend the framework with your hooks and
+custom code.
 
 We will illustrate all of this on a ResNet model, by wrapping the ResNet class of the
 [timm library](https://github.com/rwightman/pytorch-image-models) into a [`PreTrainedModel`].
@@ -218,6 +219,27 @@ resnet50d.model.load_state_dict(pretrained_model.state_dict())
 Now let's see how to make sure that when we do [`~PreTrainedModel.save_pretrained`] or [`~PreTrainedModel.push_to_hub`], the
 code of the model is saved.
 
+## Registering a model with custom code to the auto classes
+
+If you are writing a library that extends 🤗 Transformers, you may want to extend the auto classes to include your own
+model. This is different from pushing the code to the Hub in the sense that users will need to import your library to
+get the custom models (contrarily to automatically downloading the model code from the Hub).
+
+As long as your config has a `model_type` attribute that is different from existing model types, and that your model
+classes have the right `config_class` attributes, you can just add them to the auto classes like this:
+
+```py
+from transformers import AutoConfig, AutoModel, AutoModelForImageClassification
+
+AutoConfig.register("resnet", ResnetConfig)
+AutoModel.register(ResnetConfig, ResnetModel)
+AutoModelForImageClassification.register(ResnetConfig, ResnetModelForImageClassification)
+```
+
+Note that the first argument used when registering your custom config to [`AutoConfig`] needs to match the `model_type`
+of your custom config, and the first argument used when registering your custom models to any auto model class needs
+to match the `config_class` of those models.
+
 ## Sending the code to the Hub
 
 <Tip warning={true}>
@@ -350,23 +372,3 @@ model = AutoModelForImageClassification.from_pretrained(
 Note that when browsing the commit history of the model repo on the Hub, there is a button to easily copy the commit
 hash of any commit.
 
-## Registering a model with custom code to the auto classes
-
-If you are writing a library that extends 🤗 Transformers, you may want to extend the auto classes to include your own
-model. This is different from pushing the code to the Hub in the sense that users will need to import your library to
-get the custom models (contrarily to automatically downloading the model code from the Hub).
-
-As long as your config has a `model_type` attribute that is different from existing model types, and that your model
-classes have the right `config_class` attributes, you can just add them to the auto classes like this:
-
-```py
-from transformers import AutoConfig, AutoModel, AutoModelForImageClassification
-
-AutoConfig.register("resnet", ResnetConfig)
-AutoModel.register(ResnetConfig, ResnetModel)
-AutoModelForImageClassification.register(ResnetConfig, ResnetModelForImageClassification)
-```
-
-Note that the first argument used when registering your custom config to [`AutoConfig`] needs to match the `model_type`
-of your custom config, and the first argument used when registering your custom models to any auto model class needs
-to match the `config_class` of those models.

docs/source/en/index.md

@@ -1,4 +1,4 @@
-<!--Copyright 2020 The HuggingFace Team. All rights reserved.
+        <!--Copyright 2020 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
@@ -94,7 +94,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                       [CLIPSeg](model_doc/clipseg)                       |       ✅        |         ❌         |      ❌      |
 |                          [CLVP](model_doc/clvp)                          |       ✅        |         ❌         |      ❌      |
 |                       [CodeGen](model_doc/codegen)                       |       ✅        |         ❌         |      ❌      |
-|                    [CodeLlama](model_doc/code_llama)                     |       ✅        |         ❌         |      ❌      |
+|                    [CodeLlama](model_doc/code_llama)                     |       ✅        |         ❌         |      ✅      |
 |              [Conditional DETR](model_doc/conditional_detr)              |       ✅        |         ❌         |      ❌      |
 |                      [ConvBERT](model_doc/convbert)                      |       ✅        |         ✅         |      ❌      |
 |                      [ConvNeXT](model_doc/convnext)                      |       ✅        |         ✅         |      ❌      |
@@ -167,14 +167,16 @@ Flax), PyTorch, and/or TensorFlow.
 |                           [LED](model_doc/led)                           |       ✅        |         ✅         |      ❌      |
 |                         [LeViT](model_doc/levit)                         |       ✅        |         ❌         |      ❌      |
 |                          [LiLT](model_doc/lilt)                          |       ✅        |         ❌         |      ❌      |
-|                         [LLaMA](model_doc/llama)                         |       ✅        |         ❌         |      ❌      |
-|                        [Llama2](model_doc/llama2)                        |       ✅        |         ❌         |      ❌      |
+|                         [LLaMA](model_doc/llama)                         |       ✅        |         ❌         |      ✅      |
+|                        [Llama2](model_doc/llama2)                        |       ✅        |         ❌         |      ✅      |
+|                         [LLaVa](model_doc/llava)                         |       ✅        |         ❌         |      ❌      |
 |                    [Longformer](model_doc/longformer)                    |       ✅        |         ✅         |      ❌      |
 |                        [LongT5](model_doc/longt5)                        |       ✅        |         ❌         |      ✅      |
 |                          [LUKE](model_doc/luke)                          |       ✅        |         ❌         |      ❌      |
 |                        [LXMERT](model_doc/lxmert)                        |       ✅        |         ✅         |      ❌      |
 |                        [M-CTC-T](model_doc/mctct)                        |       ✅        |         ❌         |      ❌      |
 |                       [M2M100](model_doc/m2m_100)                        |       ✅        |         ❌         |      ❌      |
+|                    [MADLAD-400](model_doc/madlad-400)                    |       ✅        |         ✅         |      ✅      |
 |                        [Marian](model_doc/marian)                        |       ✅        |         ✅         |      ✅      |
 |                      [MarkupLM](model_doc/markuplm)                      |       ✅        |         ❌         |      ❌      |
 |                   [Mask2Former](model_doc/mask2former)                   |       ✅        |         ❌         |      ❌      |
@@ -187,6 +189,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                 [Megatron-GPT2](model_doc/megatron_gpt2)                 |       ✅        |         ✅         |      ✅      |
 |                       [MGP-STR](model_doc/mgp-str)                       |       ✅        |         ❌         |      ❌      |
 |                       [Mistral](model_doc/mistral)                       |       ✅        |         ❌         |      ❌      |
+|                       [Mixtral](model_doc/mixtral)                       |       ✅        |         ❌         |      ❌      |
 |                         [mLUKE](model_doc/mluke)                         |       ✅        |         ❌         |      ❌      |
 |                           [MMS](model_doc/mms)                           |       ✅        |         ✅         |      ✅      |
 |                    [MobileBERT](model_doc/mobilebert)                    |       ✅        |         ✅         |      ❌      |
@@ -213,6 +216,8 @@ Flax), PyTorch, and/or TensorFlow.
 |                           [OPT](model_doc/opt)                           |       ✅        |         ✅         |      ✅      |
 |                       [OWL-ViT](model_doc/owlvit)                        |       ✅        |         ❌         |      ❌      |
 |                         [OWLv2](model_doc/owlv2)                         |       ✅        |         ❌         |      ❌      |
+|                  [PatchTSMixer](model_doc/patchtsmixer)                  |       ✅        |         ❌         |      ❌      |
+|                      [PatchTST](model_doc/patchtst)                      |       ✅        |         ❌         |      ❌      |
 |                       [Pegasus](model_doc/pegasus)                       |       ✅        |         ✅         |      ✅      |
 |                     [PEGASUS-X](model_doc/pegasus_x)                     |       ✅        |         ❌         |      ❌      |
 |                     [Perceiver](model_doc/perceiver)                     |       ✅        |         ❌         |      ❌      |
@@ -240,6 +245,7 @@ Flax), PyTorch, and/or TensorFlow.
 |                          [RWKV](model_doc/rwkv)                          |       ✅        |         ❌         |      ❌      |
 |                           [SAM](model_doc/sam)                           |       ✅        |         ✅         |      ❌      |
 |                  [SeamlessM4T](model_doc/seamless_m4t)                   |       ✅        |         ❌         |      ❌      |
+|                [SeamlessM4Tv2](model_doc/seamless_m4t_v2)                |       ✅        |         ❌         |      ❌      |
 |                     [SegFormer](model_doc/segformer)                     |       ✅        |         ✅         |      ❌      |
 |                           [SEW](model_doc/sew)                           |       ✅        |         ❌         |      ❌      |
 |                         [SEW-D](model_doc/sew-d)                         |       ✅        |         ❌         |      ❌      |
@@ -264,10 +270,12 @@ Flax), PyTorch, and/or TensorFlow.
 |                  [Transformer-XL](model_doc/transfo-xl)                  |       ✅        |         ✅         |      ❌      |
 |                         [TrOCR](model_doc/trocr)                         |       ✅        |         ❌         |      ❌      |
 |                          [TVLT](model_doc/tvlt)                          |       ✅        |         ❌         |      ❌      |
+|                           [TVP](model_doc/tvp)                           |       ✅        |         ❌         |      ❌      |
 |                           [UL2](model_doc/ul2)                           |       ✅        |         ✅         |      ✅      |
 |                          [UMT5](model_doc/umt5)                          |       ✅        |         ❌         |      ❌      |
 |                     [UniSpeech](model_doc/unispeech)                     |       ✅        |         ❌         |      ❌      |
 |                 [UniSpeechSat](model_doc/unispeech-sat)                  |       ✅        |         ❌         |      ❌      |
+|                       [UnivNet](model_doc/univnet)                       |       ✅        |         ❌         |      ❌      |
 |                       [UPerNet](model_doc/upernet)                       |       ✅        |         ❌         |      ❌      |
 |                           [VAN](model_doc/van)                           |       ✅        |         ❌         |      ❌      |
 |                      [VideoMAE](model_doc/videomae)                      |       ✅        |         ❌         |      ❌      |

docs/source/en/internal/generation_utils.md

@@ -368,3 +368,20 @@ A [`Constraint`] can be used to force the generation to include specific tokens
 [[autodoc]] TextStreamer
 
 [[autodoc]] TextIteratorStreamer
+
+## Caches
+
+[[autodoc]] Cache
+    - update
+
+[[autodoc]] DynamicCache
+    - update
+    - get_seq_length
+    - reorder_cache
+    - to_legacy_cache
+    - from_legacy_cache
+
+[[autodoc]] SinkCache
+    - update
+    - get_seq_length
+    - reorder_cache

docs/source/en/internal/trainer_utils.md

@@ -40,7 +40,7 @@ Most of those are only useful if you are studying the code of the Trainer in the
 
 [[autodoc]] trainer_pt_utils.DistributedTensorGatherer
 
-## Distributed Evaluation
+## Trainer Argument Parser
 
 [[autodoc]] HfArgumentParser
 

docs/source/en/llm_tutorial.md

@@ -250,7 +250,7 @@ While the autoregressive generation process is relatively straightforward, makin
 1. [Guide](generation_strategies) on how to control different generation methods, how to set up the generation configuration file, and how to stream the output;
 2. [Guide](chat_templating) on the prompt template for chat LLMs;
 3. [Guide](tasks/prompting) on to get the most of prompt design;
-4. API reference on [`~generation.GenerationConfig`], [`~generation.GenerationMixin.generate`], and [generate-related classes](internal/generation_utils).
+4. API reference on [`~generation.GenerationConfig`], [`~generation.GenerationMixin.generate`], and [generate-related classes](internal/generation_utils). Most of the classes, including the logits processors, have usage examples!
 
 ### LLM leaderboards
 

docs/source/en/llm_tutorial_optimization.md

@@ -22,7 +22,7 @@ The crux of these challenges lies in augmenting the computational and memory cap
 
 In this guide, we will go over the effective techniques for efficient LLM deployment:
 
-1.  **Lower Precision**: Research has shown that operating at reduced numerical precision, namely [8-bit and 4-bit](./main_classes/quantization.md) can achieve computational advantages without a considerable decline in model performance.
+1.  **Lower Precision:** Research has shown that operating at reduced numerical precision, namely [8-bit and 4-bit](./main_classes/quantization.md) can achieve computational advantages without a considerable decline in model performance.
 
 2.  **Flash Attention:** Flash Attention is a variation of the attention algorithm that not only provides a more memory-efficient approach but also realizes increased efficiency due to optimized GPU memory utilization.
 
@@ -58,7 +58,7 @@ As of writing this document, the largest GPU chip on the market is the A100 & H1
 🤗 Transformers does not support tensor parallelism out of the box as it requires the model architecture to be written in a specific way. If you're interested in writing models in a tensor-parallelism-friendly way, feel free to have a look at [the text-generation-inference library](https://github.com/huggingface/text-generation-inference/tree/main/server/text_generation_server/models/custom_modeling).
 
 Naive pipeline parallelism is supported out of the box. For this, simply load the model with `device="auto"` which will automatically place the different layers on the available GPUs as explained [here](https://huggingface.co/docs/accelerate/v0.22.0/en/concept_guides/big_model_inference).
-Note, however that while very effective, this naive pipeline parallelism does not tackle the issues of GPU idling. For this more advanced pipeline parallelism is required as explained [here](https://huggingface.co/docs/transformers/v4.34.0/en/perf_train_gpu_many#naive-model-parallelism-vertical-and-pipeline-parallelism).
+Note, however that while very effective, this naive pipeline parallelism does not tackle the issues of GPU idling. For this more advanced pipeline parallelism is required as explained [here](https://huggingface.co/docs/transformers/en/perf_train_gpu_many#naive-model-parallelism-vertical-and-pipeline-parallelism).
 
 If you have access to an 8 x 80GB A100 node, you could load BLOOM as follows
 
@@ -286,7 +286,7 @@ If GPU memory is not a constraint for your use case, there is often no need to l
 For more in-detail usage information, we strongly recommend taking a look at the [Transformers Quantization Docs](https://huggingface.co/docs/transformers/main_classes/quantization#general-usage).
 Next, let's look into how we can improve computational and memory efficiency by using better algorithms and an improved model architecture.
 
-# 2. Flash Attention
+## 2. Flash Attention
 
 Today's top-performing LLMs share more or less the same fundamental architecture that consists of feed-forward layers, activation layers, layer normalization layers, and most crucially, self-attention layers.
 
@@ -441,7 +441,7 @@ flush()
 ```
 
 For comparison, let's run the same function, but enable Flash Attention instead.
-To do so, we convert the model to [BetterTransformers](https://huggingface.co/docs/optimum/bettertransformer/overview) and by doing so enabling PyTorch's [SDPA self-attention](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention) which in turn is based on Flash Attention.
+To do so, we convert the model to [BetterTransformer](https://huggingface.co/docs/optimum/bettertransformer/overview) and by doing so enabling PyTorch's [SDPA self-attention](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention) which in turn is able to use Flash Attention.
 
 ```python
 model.to_bettertransformer()
@@ -484,7 +484,9 @@ We can observe that we only use roughly 100MB more GPU memory when passing a ver
 ```py
 flush()
 ```
-For more information on how to use Flash Attention, please have a look at [this doc page](https://huggingface.co/docs/transformers/v4.34.0/en/perf_infer_gpu_one#flash-attention-2).
+
+For more information on how to use Flash Attention, please have a look at [this doc page](https://huggingface.co/docs/transformers/en/perf_infer_gpu_one#flashattention-2).
+
 ## 3. Architectural Innovations
 
 So far we have looked into improving computational and memory efficiency by:
@@ -662,7 +664,15 @@ Using the key-value cache has two advantages:
 
 > One should *always* make use of the key-value cache as it leads to identical results and a significant speed-up for longer input sequences. Transformers has the key-value cache enabled by default when making use of the text pipeline or the [`generate` method](https://huggingface.co/docs/transformers/main_classes/text_generation).
 
-Note that the key-value cache is especially useful for applications such as chat where multiple passes of auto-regressive decoding are required. Let's look at an example.
+<Tip warning={true}>
+
+Note that, despite our advice to use key-value caches, your LLM output may be slightly different when you use them. This is a property of the matrix multiplication kernels themselves -- you can read more about it [here](https://github.com/huggingface/transformers/issues/25420#issuecomment-1775317535).
+
+</Tip>
+
+#### 3.2.1 Multi-round conversation
+
+The key-value cache is especially useful for applications such as chat where multiple passes of auto-regressive decoding are required. Let's look at an example.
 
 ```
 User: How many people live in France?
@@ -672,14 +682,45 @@ Assistant: Germany has ca. 81 million inhabitants
 ```
 
 In this chat, the LLM runs auto-regressive decoding twice:
-- 1. The first time, the key-value cache is empty and the input prompt is `"User: How many people live in France?"` and the model auto-regressively generates the text `"Roughly 75 million people live in France"` while increasing the key-value cache at every decoding step.
-- 2. The second time the input prompt is `"User: How many people live in France? \n Assistant: Roughly 75 million people live in France \n User: And how many in Germany?"`. Thanks to the cache, all key-value vectors for the first two sentences are already computed. Therefore the input prompt only consists of `"User: And how many in Germany?"`. While processing the shortened input prompt, it's computed key-value vectors are concatenated to the key-value cache of the first decoding. The second Assistant's answer `"Germany has ca. 81 million inhabitants"` is then auto-regressively generated with the key-value cache consisting of encoded key-value vectors of `"User: How many people live in France? \n Assistant: Roughly 75 million people live in France \n User: And how many are in Germany?"`.
+  1. The first time, the key-value cache is empty and the input prompt is `"User: How many people live in France?"` and the model auto-regressively generates the text `"Roughly 75 million people live in France"` while increasing the key-value cache at every decoding step.
+  2. The second time the input prompt is `"User: How many people live in France? \n Assistant: Roughly 75 million people live in France \n User: And how many in Germany?"`. Thanks to the cache, all key-value vectors for the first two sentences are already computed. Therefore the input prompt only consists of `"User: And how many in Germany?"`. While processing the shortened input prompt, it's computed key-value vectors are concatenated to the key-value cache of the first decoding. The second Assistant's answer `"Germany has ca. 81 million inhabitants"` is then auto-regressively generated with the key-value cache consisting of encoded key-value vectors of `"User: How many people live in France? \n Assistant: Roughly 75 million people live in France \n User: And how many are in Germany?"`.
 
 Two things should be noted here:
   1. Keeping all the context is crucial for LLMs deployed in chat so that the LLM understands all the previous context of the conversation. E.g. for the example above the LLM needs to understand that the user refers to the population when asking `"And how many are in Germany"`.
   2. The key-value cache is extremely useful for chat as it allows us to continuously grow the encoded chat history instead of having to re-encode the chat history again from scratch (as e.g. would be the case when using an encoder-decoder architecture).
 
-There is however one catch. While the required peak memory for the \\( \mathbf{QK}^T \\) matrix is significantly reduced, holding the key-value cache in memory can become very memory expensive for long input sequences or multi-turn chat. Remember that the key-value cache needs to store the key-value vectors for all previous input vectors \\( \mathbf{x}_i \text{, for } i \in \{1, \ldots, c - 1\} \\) for all self-attention layers and for all attention heads.
+In `transformers`, a `generate` call will return `past_key_values` when `return_dict_in_generate=True` is passed, in addition to the default `use_cache=True`. Note that it is not yet available through the `pipeline` interface.
+
+```python
+# Generation as usual
+prompt = system_prompt + "Question: Please write a function in Python that transforms bytes to Giga bytes.\n\nAnswer: Here"
+model_inputs = tokenizer(prompt, return_tensors='pt')
+generation_output = model.generate(**model_inputs, max_new_tokens=60, return_dict_in_generate=True)
+decoded_output = tokenizer.batch_decode(generation_output.sequences)[0]
+
+# Piping the returned `past_key_values` to speed up the next conversation round
+prompt = decoded_output + "\nQuestion: How can I modify the function above to return Mega bytes instead?\n\nAnswer: Here"
+model_inputs = tokenizer(prompt, return_tensors='pt')
+generation_output = model.generate(
+  **model_inputs,
+  past_key_values=generation_output.past_key_values,
+  max_new_tokens=60,
+  return_dict_in_generate=True
+)
+tokenizer.batch_decode(generation_output.sequences)[0][len(prompt):]
+```
+
+**Output**:
+```
+ is a modified version of the function that returns Mega bytes instead.
+
+def bytes_to_megabytes(bytes):
+   return bytes / 1024 / 1024
+
+Answer: The function takes a number of bytes as input and returns the number of
+```
+
+Great, no additional time is spent recomputing the same key and values for the attention layer! There is however one catch. While the required peak memory for the \\( \mathbf{QK}^T \\) matrix is significantly reduced, holding the key-value cache in memory can become very memory expensive for long input sequences or multi-turn chat. Remember that the key-value cache needs to store the key-value vectors for all previous input vectors \\( \mathbf{x}_i \text{, for } i \in \{1, \ldots, c - 1\} \\) for all self-attention layers and for all attention heads.
 
 Let's compute the number of float values that need to be stored in the key-value cache for the LLM `bigcode/octocoder` that we used before.
 The number of float values amounts to two times the sequence length times the number of attention heads times the attention head dimension and times the number of layers.
@@ -696,11 +737,11 @@ config = model.config
 ```
 
 Roughly 8 billion float values! Storing 8 billion float values in `float16` precision requires around 15 GB of RAM which is circa half as much as the model weights themselves!
-Researchers have proposed two methods that allow to significantly reduce the memory cost of storing the key-value cache:
+Researchers have proposed two methods that allow to significantly reduce the memory cost of storing the key-value cache, which are explored in the next subsections.
 
-  1.  [Multi-Query-Attention (MQA)](https://arxiv.org/abs/1911.02150)
+#### 3.2.2 Multi-Query-Attention (MQA)
 
-Multi-Query-Attention was proposed in Noam Shazeer's *Fast Transformer Decoding: One Write-Head is All You Need* paper. As the title says, Noam found out that instead of using `n_head` key-value projections weights, one can use a single head-value projection weight pair that is shared across all attention heads without that the model's performance significantly degrades.
+[Multi-Query-Attention](https://arxiv.org/abs/1911.02150) was proposed in Noam Shazeer's *Fast Transformer Decoding: One Write-Head is All You Need* paper. As the title says, Noam found out that instead of using `n_head` key-value projections weights, one can use a single head-value projection weight pair that is shared across all attention heads without that the model's performance significantly degrades.
 
 > By using a single head-value projection weight pair, the key value vectors \\( \mathbf{k}_i, \mathbf{v}_i \\) have to be identical across all attention heads which in turn means that we only need to store 1 key-value projection pair in the cache instead of `n_head` ones.
 
@@ -720,9 +761,9 @@ MQA has seen wide adoption by the community and is now used by many of the most
 
 Also, the checkpoint used in this notebook - `bigcode/octocoder` - makes use of MQA.
 
-  2.  [Grouped-Query-Attention (GQA)](https://arxiv.org/abs/2305.13245)
+#### 3.2.3 Grouped-Query-Attention (GQA)
 
-Grouped-Query-Attention, as proposed by Ainslie et al. from Google, found that using MQA can often lead to quality degradation compared to using vanilla multi-key-value head projections. The paper argues that more model performance can be kept by less drastically reducing the number of query head projection weights. Instead of using just a single key-value projection weight, `n < n_head` key-value projection weights should be used. By choosing `n` to a significantly smaller value than `n_head`, such as 2,4 or 8 almost all of the memory and speed gains from MQA can be kept while sacrificing less model capacity and thus arguably less performance.
+[Grouped-Query-Attention](https://arxiv.org/abs/2305.13245), as proposed by Ainslie et al. from Google, found that using MQA can often lead to quality degradation compared to using vanilla multi-key-value head projections. The paper argues that more model performance can be kept by less drastically reducing the number of query head projection weights. Instead of using just a single key-value projection weight, `n < n_head` key-value projection weights should be used. By choosing `n` to a significantly smaller value than `n_head`, such as 2,4 or 8 almost all of the memory and speed gains from MQA can be kept while sacrificing less model capacity and thus arguably less performance.
 
 Moreover, the authors of GQA found out that existing model checkpoints can be *uptrained* to have a GQA architecture with as little as 5% of the original pre-training compute. While 5% of the original pre-training compute can still be a massive amount, GQA *uptraining* allows existing checkpoints to be useful for longer input sequences.
 
@@ -731,6 +772,7 @@ The most notable application of GQA is [Llama-v2](https://huggingface.co/meta-ll
 
 > As a conclusion, it is strongly recommended to make use of either GQA or MQA if the LLM is deployed with auto-regressive decoding and is required to handle large input sequences as is the case for example for chat.
 
+
 ## Conclusion
 
 The research community is constantly coming up with new, nifty ways to speed up inference time for ever-larger LLMs. As an example, one such promising research direction is [speculative decoding](https://arxiv.org/abs/2211.17192) where "easy tokens" are generated by smaller, faster language models and only "hard tokens" are generated by the LLM itself. Going into more detail is out of the scope of this notebook, but can be read upon in this [nice blog post](https://huggingface.co/blog/assisted-generation).

docs/source/en/main_classes/deepspeed.md

@@ -287,7 +287,7 @@ The information in this section isn't not specific to the DeepSpeed integration
 
 For the duration of this section let's assume that you have 2 nodes with 8 gpus each. And you can reach the first node with `ssh hostname1` and second node with `ssh hostname2`, and both must be able to reach each other via ssh locally without a password. Of course, you will need to rename these host (node) names to the actual host names you are working with.
 
-#### The torch.distributed.run launcher
+#### The torch.distributed.run(torchrun) launcher
 
 
 For example, to use `torch.distributed.run`, you could do:

docs/source/en/main_classes/quantization.md

@@ -14,535 +14,24 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Quantize 🤗 Transformers models
+# Quantization
 
-## AWQ integration
+Quantization techniques reduces memory and computational costs by representing weights and activations with lower-precision data types like 8-bit integers (int8). This enables loading larger models you normally wouldn't be able to fit into memory, and speeding up inference. Transformers supports the AWQ and GPTQ quantization algorithms and it supports 8-bit and 4-bit quantization with bitsandbytes.
 
-AWQ method has been introduced in the [*AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration* paper](https://arxiv.org/abs/2306.00978). With AWQ you can run models in 4-bit precision, while preserving its original quality (i.e. no performance degradation) with a superior throughput that other quantization methods presented below - reaching similar throughput as pure `float16` inference.
+<Tip>
 
-We now support inference with any AWQ model, meaning anyone can load and use AWQ weights that are pushed on the Hub or saved locally. Note that using AWQ requires to have access to a NVIDIA GPU. CPU inference is not supported yet. 
+Learn how to quantize models in the [Quantization](../quantization) guide.
 
-### Quantizing a model
+</Tip>
 
-We advise users to look at different existing tools in the ecosystem to quantize their models with AWQ algorithm, such as:
-
-- [`llm-awq`](https://github.com/mit-han-lab/llm-awq) from MIT Han Lab
-- [`autoawq`](https://github.com/casper-hansen/AutoAWQ) from [`casper-hansen`](https://github.com/casper-hansen)
-- Intel neural compressor from Intel - through [`optimum-intel`](https://huggingface.co/docs/optimum/main/en/intel/optimization_inc)
-
-Many other tools might exist in the ecosystem, please feel free to open a PR to add them to the list.
-Currently the integration with 🤗 Transformers is only available for models that have been quantized using `autoawq` library and `llm-awq`. Most of the models quantized with `auto-awq` can be found under [`TheBloke`](https://huggingface.co/TheBloke) namespace of 🤗 Hub, and to quantize models with `llm-awq` please refer to the [`convert_to_hf.py`](https://github.com/mit-han-lab/llm-awq/blob/main/examples/convert_to_hf.py) script in the examples folder of [`llm-awq`](https://github.com/mit-han-lab/llm-awq/).
-
-### Load a quantized model
-
-You can load a quantized model from the Hub using the `from_pretrained` method. Make sure that the pushed weights are quantized, by checking that the attribute `quantization_config` is present in the model's configuration file (`configuration.json`). You can confirm that the model is quantized in the AWQ format by checking the field `quantization_config.quant_method` which should be set to `"awq"`. Note that loading the model will set other weights in `float16` by default for performance reasons. If you want to change that behavior, you can pass `torch_dtype` argument to `torch.float32` or `torch.bfloat16`. You can find in the sections below some example snippets and notebook.
-
-## Example usage
-
-First, you need to install [`autoawq`](https://github.com/casper-hansen/AutoAWQ) library
-
-```bash
-pip install autoawq
-```
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model_id = "TheBloke/zephyr-7B-alpha-AWQ"
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="cuda:0")
-```
-
-In case you first load your model on CPU, make sure to move it to your GPU device before using 
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model_id = "TheBloke/zephyr-7B-alpha-AWQ"
-model = AutoModelForCausalLM.from_pretrained(model_id).to("cuda:0")
-```
-
-### Combining AWQ and Flash Attention
-
-You can combine AWQ quantization with Flash Attention to get a model that is both quantized and faster. Simply load the model using `from_pretrained` and pass `use_flash_attention_2=True` argument.
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model = AutoModelForCausalLM.from_pretrained("TheBloke/zephyr-7B-alpha-AWQ", use_flash_attention_2=True, device_map="cuda:0")
-```
-
-### Benchmarks
-
-We performed some speed, throughput and latency benchmarks using [`optimum-benchmark`](https://github.com/huggingface/optimum-benchmark) library. 
-
-Note at that time of writing this documentation section, the available quantization methods were: `awq`, `gptq` and `bitsandbytes`.
-
-The benchmark was run on a NVIDIA-A100 instance and the model used was [`TheBloke/Mistral-7B-v0.1-AWQ`](https://huggingface.co/TheBloke/Mistral-7B-v0.1-AWQ) for the AWQ model, [`TheBloke/Mistral-7B-v0.1-GPTQ`](https://huggingface.co/TheBloke/Mistral-7B-v0.1-GPTQ) for the GPTQ model. We also benchmarked it against `bitsandbytes` quantization methods and native `float16` model. Some results are shown below:
-
-<div style="text-align: center">
-<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/forward_memory_plot.png">
-</div>
-
-<div style="text-align: center">
-<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/generate_memory_plot.png">
-</div>
-
-<div style="text-align: center">
-<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/generate_throughput_plot.png">
-</div>
-
-<div style="text-align: center">
-<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/forward_latency_plot.png">
-</div>
-
-You can find the full results together with packages versions in [this link](https://github.com/huggingface/optimum-benchmark/tree/main/examples/running-mistrals).
-
-From the results it appears that AWQ quantization method is the fastest quantization method for inference, text generation and among the lowest peak memory for text generation. However, AWQ seems to have the largest forward latency per batch size. 
-
-### Google colab demo
-
-Check out how to use this integration throughout this [Google Colab demo](https://colab.research.google.com/drive/1HzZH89yAXJaZgwJDhQj9LqSBux932BvY)!
-
-### AwqConfig
+## AwqConfig
 
 [[autodoc]] AwqConfig
 
-## `AutoGPTQ` Integration
-
-🤗 Transformers has integrated `optimum` API to perform GPTQ quantization on language models. You can load and quantize your model in 8, 4, 3 or even 2 bits without a big drop of performance and faster inference speed! This is supported by most GPU hardwares.
-
-To learn more about the quantization model, check out: 
-- the [GPTQ](https://arxiv.org/pdf/2210.17323.pdf) paper
-- the `optimum` [guide](https://huggingface.co/docs/optimum/llm_quantization/usage_guides/quantization) on GPTQ quantization
-- the [`AutoGPTQ`](https://github.com/PanQiWei/AutoGPTQ) library used as the backend
-
-### Requirements
-
-You need to have the following requirements installed to run the code below: 
-
-- Install latest `AutoGPTQ` library
-`pip install auto-gptq`
-
-- Install latest `optimum` from source 
-`pip install git+https://github.com/huggingface/optimum.git`
-
-- Install latest `transformers` from source 
-`pip install git+https://github.com/huggingface/transformers.git`
-
-- Install latest `accelerate` library 
-`pip install --upgrade accelerate`
-
-Note that GPTQ integration supports for now only text models and you may encounter unexpected behaviour for vision, speech or multi-modal models.
-
-### Load and quantize a model
-
-GPTQ is a quantization method that requires weights calibration before using the quantized models. If you want to quantize transformers model from scratch, it might take some time before producing the quantized model (~5 min on a Google colab for `facebook/opt-350m` model). 
-
-Hence, there are two different scenarios where you want to use GPTQ-quantized models. The first use case would be to load models that has been already quantized by other users that are available on the Hub, the second use case would be to quantize your model from scratch and save it or push it on the Hub so that other users can also use it.
-
-#### GPTQ Configuration
-
-In order to load and quantize a model, you need to create a [`GPTQConfig`]. You need to pass the number of `bits`, a `dataset` in order to calibrate the quantization and the `tokenizer` of the model in order prepare the dataset.
-
-```python 
-model_id = "facebook/opt-125m"
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-gptq_config = GPTQConfig(bits=4, dataset = "c4", tokenizer=tokenizer)
-```
-
-Note that you can pass your own dataset as a list of string. However, it is highly recommended to use the dataset from the GPTQ paper. 
-
-```python
-dataset = ["auto-gptq is an easy-to-use model quantization library with user-friendly apis, based on GPTQ algorithm."]
-quantization = GPTQConfig(bits=4, dataset = dataset, tokenizer=tokenizer)
-```
-
-#### Quantization
-
-You can quantize a model by using `from_pretrained` and setting the `quantization_config`. 
-
-```python
-from transformers import AutoModelForCausalLM
-model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=gptq_config)
-
-```
-Note that you will need a GPU to quantize a model. We will put the model in the cpu and move the modules back and forth to the gpu in order to quantize them.
-
-If you want to maximize your gpus usage while using cpu offload, you can set `device_map = "auto"`.
-
-```python
-from transformers import AutoModelForCausalLM
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", quantization_config=gptq_config)
-```
-
-Note that disk offload is not supported. Furthermore, if you are out of memory because of the dataset, you may have to pass `max_memory` in `from_pretained`. Checkout this [guide](https://huggingface.co/docs/accelerate/usage_guides/big_modeling#designing-a-device-map) to learn more about `device_map` and `max_memory`.
-
-<Tip warning={true}>
-GPTQ quantization only works for text model for now. Futhermore, the quantization process can a lot of time depending on one's hardware (175B model = 4 gpu hours using NVIDIA A100). Please check on the hub if there is not a GPTQ quantized version of the model. If not, you can submit a demand on github. 
-</Tip>
-
-### Push quantized model to 🤗 Hub
-
-You can push the quantized model like any 🤗 model to Hub with `push_to_hub`. The quantization config will be saved and pushed along the model. 
-
-```python
-quantized_model.push_to_hub("opt-125m-gptq")
-tokenizer.push_to_hub("opt-125m-gptq")
-```
-
-If you want to save your quantized model on your local machine, you can also do it with `save_pretrained`: 
-
-```python
-quantized_model.save_pretrained("opt-125m-gptq")
-tokenizer.save_pretrained("opt-125m-gptq")
-```
-
-Note that if you have quantized your model with a `device_map`, make sure to move the entire model to one of your gpus or the `cpu` before saving it.
-
-```python
-quantized_model.to("cpu")
-quantized_model.save_pretrained("opt-125m-gptq")
-```
-
-### Load a quantized model from the 🤗 Hub
-
-You can load a quantized model from the Hub by using `from_pretrained`.
-Make sure that the pushed weights are quantized, by checking that the attribute `quantization_config` is present in the model configuration object.
-
-```python
-from transformers import AutoModelForCausalLM
-model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq")
-```
-
-If you want to load a model faster and without allocating more memory than needed, the `device_map` argument also works with quantized model. Make sure that you have `accelerate` library installed.
-
-```python
-from transformers import AutoModelForCausalLM
-model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq", device_map="auto")
-```
-
-### Exllama kernels for faster inference
-
-For 4-bit model, you can use the exllama kernels in order to a faster inference speed. It is activated by default. You can change that behavior by passing `use_exllama` in [`GPTQConfig`]. This will overwrite the quantization config stored in the config. Note that you will only be able to overwrite the attributes related to the kernels. Furthermore, you need to have the entire model on gpus if you want to use exllama kernels. Also, you can perform CPU inference using Auto-GPTQ for Auto-GPTQ version > 0.4.2 by passing `device_map` = "cpu". For CPU inference, you have to pass `use_exllama = False` in the `GPTQConfig.`
-
-```py
-import torch
-gptq_config = GPTQConfig(bits=4)
-model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq", device_map="auto", quantization_config=gptq_config)
-```
-
-With the release of the exllamav2 kernels, you can get faster inference speed compared to the exllama kernels. You just need to pass `exllama_config={"version": 2}` in [`GPTQConfig`]:
-
-```py
-import torch
-gptq_config = GPTQConfig(bits=4, exllama_config={"version":2})
-model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq", device_map="auto", quantization_config = gptq_config)
-```
-
-Note that only 4-bit models are supported for now. Furthermore, it is recommended to deactivate the exllama kernels if you are finetuning a quantized model with peft. 
-
-You can find the benchmark of these kernels [here](https://github.com/huggingface/optimum/tree/main/tests/benchmark#gptq-benchmark)
-#### Fine-tune a quantized model 
-
-With the official support of adapters in the Hugging Face ecosystem, you can fine-tune models that have been quantized with GPTQ. 
-Please have a look at [`peft`](https://github.com/huggingface/peft) library for more details.
-
-### Example demo
-
-Check out the Google Colab [notebook](https://colab.research.google.com/drive/1_TIrmuKOFhuRRiTWN94iLKUFu6ZX4ceb?usp=sharing) to learn how to quantize your model with GPTQ and how finetune the quantized model with peft. 
-
-### GPTQConfig
+## GPTQConfig
 
 [[autodoc]] GPTQConfig
 
-
-## `bitsandbytes` Integration
-
-🤗 Transformers is closely integrated with most used modules on `bitsandbytes`. You can load your model in 8-bit precision with few lines of code.
-This is supported by most of the GPU hardwares since the `0.37.0` release of `bitsandbytes`.
-
-Learn more about the quantization method in the [LLM.int8()](https://arxiv.org/abs/2208.07339) paper, or the [blogpost](https://huggingface.co/blog/hf-bitsandbytes-integration) about the collaboration.
-
-Since its `0.39.0` release, you can load any model that supports `device_map` using 4-bit quantization, leveraging FP4 data type.
-
-If you want to quantize your own pytorch model, check out this [documentation](https://huggingface.co/docs/accelerate/main/en/usage_guides/quantization) from 🤗 Accelerate library. 
-
-Here are the things you can do using `bitsandbytes` integration
-
-### General usage
-
-You can quantize a model by using the `load_in_8bit` or `load_in_4bit` argument when calling the [`~PreTrainedModel.from_pretrained`] method as long as your model supports loading with 🤗 Accelerate and contains `torch.nn.Linear` layers. This should work for any modality as well.
-
-```python
-from transformers import AutoModelForCausalLM
-
-model_8bit = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", load_in_8bit=True)
-model_4bit = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", load_in_4bit=True)
-```
-
-By default all other modules (e.g. `torch.nn.LayerNorm`) will be converted in `torch.float16`, but if you want to change their `dtype` you can overwrite the `torch_dtype` argument:
-
-```python
->>> import torch
->>> from transformers import AutoModelForCausalLM
-
->>> model_8bit = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", load_in_8bit=True, torch_dtype=torch.float32)
->>> model_8bit.model.decoder.layers[-1].final_layer_norm.weight.dtype
-torch.float32
-```
-
-
-### FP4 quantization 
-
-#### Requirements
-
-Make sure that you have installed the requirements below before running any of the code snippets below.
-
-- Latest `bitsandbytes` library
-`pip install bitsandbytes>=0.39.0`
-
-- Install latest `accelerate`
-`pip install --upgrade accelerate`
-
-- Install latest `transformers`
-`pip install --upgrade transformers`
-
-#### Tips and best practices
-
-- **Advanced usage:** Refer to [this Google Colab notebook](https://colab.research.google.com/drive/1ge2F1QSK8Q7h0hn3YKuBCOAS0bK8E0wf) for advanced usage of 4-bit quantization with all the possible options.
-
-- **Faster inference with `batch_size=1` :** Since the `0.40.0` release of bitsandbytes, for `batch_size=1` you can benefit from fast inference. Check out [these release notes](https://github.com/TimDettmers/bitsandbytes/releases/tag/0.40.0) and make sure to have a version that is greater than `0.40.0` to benefit from this feature out of the box. 
-
-- **Training:** According to [QLoRA paper](https://arxiv.org/abs/2305.14314), for training 4-bit base models (e.g. using LoRA adapters) one should use `bnb_4bit_quant_type='nf4'`. 
-
-- **Inference:** For inference, `bnb_4bit_quant_type` does not have a huge impact on the performance. However for consistency with the model's weights, make sure you use the same `bnb_4bit_compute_dtype` and `torch_dtype` arguments.
-
-#### Load a large model in 4bit
-
-By using `load_in_4bit=True` when calling the `.from_pretrained` method, you can divide your memory use by 4 (roughly).
-
-```python
-# pip install transformers accelerate bitsandbytes
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model_id = "bigscience/bloom-1b7"
-
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", load_in_4bit=True)
-```
-
-<Tip warning={true}>
-
-Note that once a model has been loaded in 4-bit it is currently not possible to push the quantized weights on the Hub. Note also that you cannot train 4-bit weights as this is not supported yet. However you can use 4-bit models to train extra parameters, this will be covered in the next section.
-
-</Tip>
-
-### Load a large model in 8bit
-
-You can load a model by roughly halving the memory requirements by using `load_in_8bit=True` argument when calling `.from_pretrained` method
-
-
-```python
-# pip install transformers accelerate bitsandbytes
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model_id = "bigscience/bloom-1b7"
-
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", load_in_8bit=True)
-```
-
-Then, use your model as you would usually use a [`PreTrainedModel`].
-
-You can check the memory footprint of your model with `get_memory_footprint` method.
-
-```python
-print(model.get_memory_footprint())
-```
-
-With this integration we were able to load large models on smaller devices and run them without any issue.  
-
-<Tip warning={true}>
-
-Note that once a model has been loaded in 8-bit it is currently not possible to push the quantized weights on the Hub except if you use the latest `transformers` and `bitsandbytes`. Note also that you cannot train 8-bit weights as this is not supported yet. However you can use 8-bit models to train extra parameters, this will be covered in the next section.
-Note also that `device_map` is optional but setting `device_map = 'auto'` is prefered for inference as it will dispatch efficiently the model on the available ressources.
-
-</Tip>
-
-#### Advanced use cases
-
-Here we will cover some advanced use cases you can perform with FP4 quantization 
-
-##### Change the compute dtype
-
-The compute dtype is used to change the dtype that will be used during computation. For example, hidden states could be in `float32` but computation can be set to bf16 for speedups. By default, the compute dtype is set to `float32`.
-
-```python
-import torch
-from transformers import BitsAndBytesConfig
-
-quantization_config = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16)
-```
-
-##### Using NF4 (Normal Float 4) data type 
-
-You can also use the NF4 data type, which is a new 4bit datatype adapted for weights that have been initialized using a normal distribution. For that run:
-
-```python
-from transformers import BitsAndBytesConfig
-
-nf4_config = BitsAndBytesConfig(
-    load_in_4bit=True,
-    bnb_4bit_quant_type="nf4",
-)
-
-model_nf4 = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=nf4_config)
-```
-
-##### Use nested quantization for more memory efficient inference
-
-We also advise users to use the nested quantization technique. This saves more memory at no additional performance - from our empirical observations, this enables fine-tuning llama-13b model on an NVIDIA-T4 16GB with a sequence length of 1024, batch size of 1 and gradient accumulation steps of 4.
-
-```python
-from transformers import BitsAndBytesConfig
-
-double_quant_config = BitsAndBytesConfig(
-    load_in_4bit=True,
-    bnb_4bit_use_double_quant=True,
-)
-
-model_double_quant = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=double_quant_config)
-```
-
-
-### Push quantized models on the 🤗 Hub
-
-You can push a quantized model on the Hub by naively using `push_to_hub` method. This will first push the quantization configuration file, then push the quantized model weights.
-Make sure to use `bitsandbytes>0.37.2` (at this time of writing, we tested it on `bitsandbytes==0.38.0.post1`) to be able to use this feature. 
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model = AutoModelForCausalLM.from_pretrained("bigscience/bloom-560m", device_map="auto", load_in_8bit=True)
-tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-560m")
-
-model.push_to_hub("bloom-560m-8bit")
-```
-
-<Tip warning={true}>
-
-Pushing 8bit models on the Hub is strongely encouraged for large models. This will allow the community to benefit from the memory footprint reduction and loading for example large models on a Google Colab.
-
-</Tip>
-
-### Load a quantized model from the 🤗 Hub
-
-You can load a quantized model from the Hub by using `from_pretrained` method. Make sure that the pushed weights are quantized, by checking that the attribute `quantization_config` is present in the model configuration object.
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-model = AutoModelForCausalLM.from_pretrained("{your_username}/bloom-560m-8bit", device_map="auto")
-```
-
-Note that in this case, you don't need to specify the arguments `load_in_8bit=True`, but you need to make sure that `bitsandbytes` and `accelerate` are installed.
-Note also that `device_map` is optional but setting `device_map = 'auto'` is prefered for inference as it will dispatch efficiently the model on the available ressources.
-
-### Advanced use cases
-
-This section is intended to advanced users, that want to explore what it is possible to do beyond loading and running 8-bit models.
-
-#### Offload between `cpu` and `gpu`
-
-One of the advanced use case of this is being able to load a model and dispatch the weights between `CPU` and `GPU`. Note that the weights that will be dispatched on CPU **will not** be converted in 8-bit, thus kept in `float32`. This feature is intended for users that want to fit a very large model and dispatch the model between GPU and CPU.
-
-First, load a [`BitsAndBytesConfig`] from `transformers` and set the attribute `llm_int8_enable_fp32_cpu_offload` to `True`:
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
-
-quantization_config = BitsAndBytesConfig(llm_int8_enable_fp32_cpu_offload=True)
-```
-
-Let's say you want to load `bigscience/bloom-1b7` model, and you have just enough GPU RAM to fit the entire model except the `lm_head`. Therefore write a custom device_map as follows:
-
-```python
-device_map = {
-    "transformer.word_embeddings": 0,
-    "transformer.word_embeddings_layernorm": 0,
-    "lm_head": "cpu",
-    "transformer.h": 0,
-    "transformer.ln_f": 0,
-}
-```
-
-And load your model as follows:
-```python
-model_8bit = AutoModelForCausalLM.from_pretrained(
-    "bigscience/bloom-1b7",
-    device_map=device_map,
-    quantization_config=quantization_config,
-)
-```
-
-And that's it! Enjoy your model!
-
-#### Play with `llm_int8_threshold`
-
-You can play with the `llm_int8_threshold` argument to change the threshold of the outliers. An "outlier" is a hidden state value that is greater than a certain threshold. 
-This corresponds to the outlier threshold for outlier detection as described in `LLM.int8()` paper. Any hidden states value that is above this threshold will be considered an outlier and the operation on those values will be done in fp16. Values are usually normally distributed, that is, most values are in the range [-3.5, 3.5], but there are some exceptional systematic outliers that are very differently distributed for large models. These outliers are often in the interval [-60, -6] or [6, 60]. Int8 quantization works well for values of magnitude ~5, but beyond that, there is a significant performance penalty. A good default threshold is 6, but a lower threshold might be needed for more unstable models (small models, fine-tuning).
-This argument can impact the inference speed of the model. We suggest to play with this parameter to find which one is the best for your use case.
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
-
-model_id = "bigscience/bloom-1b7"
-
-quantization_config = BitsAndBytesConfig(
-    llm_int8_threshold=10,
-)
-
-model_8bit = AutoModelForCausalLM.from_pretrained(
-    model_id,
-    device_map=device_map,
-    quantization_config=quantization_config,
-)
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-```
-
-#### Skip the conversion of some modules
-
-Some models has several modules that needs to be not converted in 8-bit to ensure stability. For example Jukebox model has several `lm_head` modules that should be skipped. Play with `llm_int8_skip_modules` 
-
-```python
-from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
-
-model_id = "bigscience/bloom-1b7"
-
-quantization_config = BitsAndBytesConfig(
-    llm_int8_skip_modules=["lm_head"],
-)
-
-model_8bit = AutoModelForCausalLM.from_pretrained(
-    model_id,
-    device_map=device_map,
-    quantization_config=quantization_config,
-)
-tokenizer = AutoTokenizer.from_pretrained(model_id)
-```
-
-#### Fine-tune a model that has been loaded in 8-bit
-
-With the official support of adapters in the Hugging Face ecosystem, you can fine-tune models that have been loaded in 8-bit. 
-This enables fine-tuning large models such as `flan-t5-large` or `facebook/opt-6.7b` in a single google Colab. Please have a look at [`peft`](https://github.com/huggingface/peft) library for more details.
-
-Note that you don't need to pass `device_map` when loading the model for training. It will automatically load your model on your GPU. You can also set the device map to a specific device if needed (e.g. `cuda:0`, `0`, `torch.device('cuda:0')`). Please note that `device_map=auto` should be used for inference only. 
-
-### BitsAndBytesConfig
+## BitsAndBytesConfig
 
 [[autodoc]] BitsAndBytesConfig
-
-
-## Quantization with 🤗 `optimum` 
-
-Please have a look at [Optimum documentation](https://huggingface.co/docs/optimum/index) to learn more about quantization methods that are supported by `optimum` and see if these are applicable for your use case.

docs/source/en/main_classes/trainer.md

@@ -26,7 +26,7 @@ If you're looking to fine-tune a language model like Llama-2 or Mistral on a tex
 
 Before instantiating your [`Trainer`], create a [`TrainingArguments`] to access all the points of customization during training.
 
-The API supports distributed training on multiple GPUs/TPUs, mixed precision through [NVIDIA Apex](https://github.com/NVIDIA/apex) and Native AMP for PyTorch.
+The API supports distributed training on multiple GPUs/TPUs, mixed precision through [NVIDIA Apex] for NVIDIA GPUs, [ROCm APEX](https://github.com/ROCmSoftwarePlatform/apex) for AMD GPUs, and Native AMP for PyTorch.
 
 The [`Trainer`] contains the basic training loop which supports the above features. To inject custom behavior you can subclass them and override the following methods:
 
@@ -206,7 +206,7 @@ Let's discuss how you can tell your program which GPUs are to be used and in wha
 When using [`DistributedDataParallel`](https://pytorch.org/docs/stable/generated/torch.nn.parallel.DistributedDataParallel.html) to use only a subset of your GPUs, you simply specify the number of GPUs to use. For example, if you have 4 GPUs, but you wish to use the first 2 you can do:
 
 ```bash
-python -m torch.distributed.launch --nproc_per_node=2  trainer-program.py ...
+torchrun --nproc_per_node=2  trainer-program.py ...
 ```
 
 if you have either [`accelerate`](https://github.com/huggingface/accelerate) or [`deepspeed`](https://github.com/microsoft/DeepSpeed) installed you can also accomplish the same by using one of:
@@ -219,7 +219,7 @@ accelerate launch --num_processes 2 trainer-program.py ...
 deepspeed --num_gpus 2 trainer-program.py ...
 ```
 
-You don't need to use the Accelerate or [the Deepspeed integration](Deepspeed) features to use these launchers.
+You don't need to use the Accelerate or [the Deepspeed integration](deepspeed) features to use these launchers.
 
 
 Until now you were able to tell the program how many GPUs to use. Now let's discuss how to select specific GPUs and control their order.
@@ -233,7 +233,7 @@ If you have multiple GPUs and you'd like to use only 1 or a few of those GPUs, s
 For example, let's say you have 4 GPUs: 0, 1, 2 and 3. To run only on the physical GPUs 0 and 2, you can do:
 
 ```bash
-CUDA_VISIBLE_DEVICES=0,2 python -m torch.distributed.launch trainer-program.py ...
+CUDA_VISIBLE_DEVICES=0,2 torchrun trainer-program.py ...
 ```
 
 So now pytorch will see only 2 GPUs, where your physical GPUs 0 and 2 are mapped to `cuda:0` and `cuda:1` correspondingly.
@@ -241,7 +241,7 @@ So now pytorch will see only 2 GPUs, where your physical GPUs 0 and 2 are mapped
 You can even change their order:
 
 ```bash
-CUDA_VISIBLE_DEVICES=2,0 python -m torch.distributed.launch trainer-program.py ...
+CUDA_VISIBLE_DEVICES=2,0 torchrun trainer-program.py ...
 ```
 
 Here your physical GPUs 0 and 2 are mapped to `cuda:1` and `cuda:0` correspondingly.
@@ -263,7 +263,7 @@ As with any environment variable you can, of course, export those instead of add
 
 ```bash
 export CUDA_VISIBLE_DEVICES=0,2
-python -m torch.distributed.launch trainer-program.py ...
+torchrun trainer-program.py ...
 ```
 
 but this approach can be confusing since you may forget you set up the environment variable earlier and not understand why the wrong GPUs are used. Therefore, it's a common practice to set the environment variable just for a specific run on the same command line as it's shown in most examples of this section.
@@ -272,7 +272,7 @@ but this approach can be confusing since you may forget you set up the environme
 
 There is an additional environment variable `CUDA_DEVICE_ORDER` that controls how the physical devices are ordered. The two choices are:
 
-1. ordered by PCIe bus IDs (matches `nvidia-smi`'s order) - this is the default.
+1. ordered by PCIe bus IDs (matches `nvidia-smi` and `rocm-smi`'s order) - this is the default.
 
 ```bash
 export CUDA_DEVICE_ORDER=PCI_BUS_ID
@@ -284,7 +284,7 @@ export CUDA_DEVICE_ORDER=PCI_BUS_ID
 export CUDA_DEVICE_ORDER=FASTEST_FIRST
 ```
 
-Most of the time you don't need to care about this environment variable, but it's very helpful if you have a lopsided setup where you have an old and a new GPUs physically inserted in such a way so that the slow older card appears to be first. One way to fix that is to swap the cards. But if you can't swap the cards (e.g., if the cooling of the devices gets impacted) then setting `CUDA_DEVICE_ORDER=FASTEST_FIRST` will always put the newer faster card first. It'll be somewhat confusing though since `nvidia-smi` will still report them in the PCIe order.
+Most of the time you don't need to care about this environment variable, but it's very helpful if you have a lopsided setup where you have an old and a new GPUs physically inserted in such a way so that the slow older card appears to be first. One way to fix that is to swap the cards. But if you can't swap the cards (e.g., if the cooling of the devices gets impacted) then setting `CUDA_DEVICE_ORDER=FASTEST_FIRST` will always put the newer faster card first. It'll be somewhat confusing though since `nvidia-smi` (or `rocm-smi`) will still report them in the PCIe order.
 
 The other solution to swapping the order is to use:
 
@@ -426,8 +426,7 @@ To read more about it and the benefits, check out the [Fully Sharded Data Parall
 We have integrated the latest PyTorch's Fully Sharded Data Parallel (FSDP) training feature.
 All you need to do is enable it through the config.
 
-**Required PyTorch version for FSDP support**: PyTorch Nightly (or 1.12.0 if you read this after it has been released)
-as the model saving with FSDP activated is only available with recent fixes.
+**Required PyTorch version for FSDP support**: PyTorch >=2.1.0
 
 **Usage**:
 
@@ -440,6 +439,8 @@ as the model saving with FSDP activated is only available with recent fixes.
   - SHARD_GRAD_OP : Shards optimizer states + gradients across data parallel workers/GPUs.
     For this, add `--fsdp shard_grad_op` to the command line arguments.
   - NO_SHARD : No sharding. For this, add `--fsdp no_shard` to the command line arguments.
+  - HYBRID_SHARD : No sharding. For this, add `--fsdp hybrid_shard` to the command line arguments.
+  - HYBRID_SHARD_ZERO2 : No sharding. For this, add `--fsdp hybrid_shard_zero2` to the command line arguments.
 - To offload the parameters and gradients to the CPU, 
   add `--fsdp "full_shard offload"` or `--fsdp "shard_grad_op offload"` to the command line arguments.
 - To automatically recursively wrap layers with FSDP using `default_auto_wrap_policy`, 
@@ -449,18 +450,18 @@ as the model saving with FSDP activated is only available with recent fixes.
 - Remaining FSDP config is passed via `--fsdp_config <path_to_fsdp_config.json>`. It is either a location of
   FSDP json config file (e.g., `fsdp_config.json`) or an already loaded json file as `dict`. 
   - If auto wrapping is enabled, you can either use transformer based auto wrap policy or size based auto wrap policy.
-    - For transformer based auto wrap policy, it is recommended to specify `fsdp_transformer_layer_cls_to_wrap` in the config file. If not specified, the default value is `model._no_split_modules` when available.
+    - For transformer based auto wrap policy, it is recommended to specify `transformer_layer_cls_to_wrap` in the config file. If not specified, the default value is `model._no_split_modules` when available.
       This specifies the list of transformer layer class name (case-sensitive) to wrap ,e.g, [`BertLayer`], [`GPTJBlock`], [`T5Block`] ....
       This is important because submodules that share weights (e.g., embedding layer) should not end up in different FSDP wrapped units.
       Using this policy, wrapping happens for each block containing Multi-Head Attention followed by couple of MLP layers. 
       Remaining layers including the shared embeddings are conveniently wrapped in same outermost FSDP unit.
       Therefore, use this for transformer based models.
-    - For size based auto wrap policy, please add `fsdp_min_num_params` in the config file. 
+    - For size based auto wrap policy, please add `min_num_params` in the config file. 
       It specifies FSDP's minimum number of parameters for auto wrapping.
-  - `fsdp_backward_prefetch` can be specified in the config file. It controls when to prefetch next set of parameters. 
+  - `backward_prefetch` can be specified in the config file. It controls when to prefetch next set of parameters. 
     `backward_pre` and `backward_pos` are available options. 
     For more information refer `torch.distributed.fsdp.fully_sharded_data_parallel.BackwardPrefetch`
-  - `fsdp_forward_prefetch` can be specified in the config file. It controls when to prefetch next set of parameters. 
+  - `forward_prefetch` can be specified in the config file. It controls when to prefetch next set of parameters. 
     If `"True"`, FSDP explicitly prefetches the next upcoming all-gather while executing in the forward pass. 
   - `limit_all_gathers` can be specified in the config file. 
     If `"True"`, FSDP explicitly synchronizes the CPU thread to prevent too many in-flight all-gathers.
@@ -468,6 +469,20 @@ as the model saving with FSDP activated is only available with recent fixes.
     If `"True"`, FSDP activation checkpointing is a technique to reduce memory usage by clearing activations of
     certain layers and recomputing them during a backward pass. Effectively, this trades extra computation time
     for reduced memory usage.
+  - `use_orig_params` can be specified in the config file. 
+    If True, allows non-uniform `requires_grad` during init, which means support for interspersed frozen and trainable paramteres. Useful in cases such as parameter-efficient fine-tuning. This also enables to have different optimizer param groups. This should be `True` when creating optimizer object before preparing/wrapping the model with FSDP.
+    Please refer this [blog](https://dev-discuss.pytorch.org/t/rethinking-pytorch-fully-sharded-data-parallel-fsdp-from-first-principles/1019). 
+
+**Saving and loading**
+Saving entire intermediate checkpoints using `FULL_STATE_DICT` state_dict_type with CPU offloading on rank 0 takes a lot of time and often results in NCCL Timeout errors due to indefinite hanging during broadcasting. However, at the end of training, we want the whole model state dict instead of the sharded state dict which is only compatible with FSDP. Use `SHARDED_STATE_DICT` (default) state_dict_type to save the intermediate checkpoints and optimizer states in this format recommended by the PyTorch team. 
+
+Saving the final checkpoint in transformers format using default `safetensors` format requires below changes.
+```python
+if trainer.is_fsdp_enabled:
+    trainer.accelerator.state.fsdp_plugin.set_state_dict_type("FULL_STATE_DICT")
+
+trainer.save_model(script_args.output_dir)
+```
 
 **Few caveats to be aware of**
 - it is incompatible with `generate`, thus is incompatible with `--predict_with_generate` 
@@ -492,15 +507,15 @@ Pass `--fsdp "full shard"` along with following changes to be made in `--fsdp_co
   https://github.com/pytorch/xla/blob/master/torch_xla/distributed/fsdp/xla_fully_sharded_data_parallel.py).
 - `xla_fsdp_grad_ckpt`. When `True`, uses gradient checkpointing over each nested XLA FSDP wrapped layer. 
   This setting can only be used when the xla flag is set to true, and an auto wrapping policy is specified through
-  `fsdp_min_num_params` or `fsdp_transformer_layer_cls_to_wrap`. 
+  `min_num_params` or `transformer_layer_cls_to_wrap`. 
 - You can either use transformer based auto wrap policy or size based auto wrap policy.
-  - For transformer based auto wrap policy, it is recommended to specify `fsdp_transformer_layer_cls_to_wrap` in the config file. If not specified, the default value is `model._no_split_modules` when available.
+  - For transformer based auto wrap policy, it is recommended to specify `transformer_layer_cls_to_wrap` in the config file. If not specified, the default value is `model._no_split_modules` when available.
     This specifies the list of transformer layer class name (case-sensitive) to wrap ,e.g, [`BertLayer`], [`GPTJBlock`], [`T5Block`] ....
     This is important because submodules that share weights (e.g., embedding layer) should not end up in different FSDP wrapped units.
     Using this policy, wrapping happens for each block containing Multi-Head Attention followed by couple of MLP layers. 
     Remaining layers including the shared embeddings are conveniently wrapped in same outermost FSDP unit.
     Therefore, use this for transformer based models.
-  - For size based auto wrap policy, please add `fsdp_min_num_params` in the config file. 
+  - For size based auto wrap policy, please add `min_num_params` in the config file. 
     It specifies FSDP's minimum number of parameters for auto wrapping.
 
 

docs/source/en/model_doc/albert.md

@@ -59,13 +59,67 @@ This model was contributed by [lysandre](https://huggingface.co/lysandre). This
 - Layers are split in groups that share parameters (to save memory).
 Next sentence prediction is replaced by a sentence ordering prediction: in the inputs, we have two sentences A and B (that are consecutive) and we either feed A followed by B or B followed by A. The model must predict if they have been swapped or not.
 
+
+
+This model was contributed by [lysandre](https://huggingface.co/lysandre). This model jax version was contributed by
+[kamalkraj](https://huggingface.co/kamalkraj). The original code can be found [here](https://github.com/google-research/ALBERT).
+
+
 ## Resources
 
-- [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)
+
+The resources provided in the following sections consist of a list of official Hugging Face and community (indicated by 🌎) resources to help you get started with AlBERT. 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.
+
+
+<PipelineTag pipeline="text-classification"/>
+
+
+- [`AlbertForSequenceClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/text-classification).
+
+
+- [`TFAlbertForSequenceClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/text-classification).
+
+- [`FlaxAlbertForSequenceClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/flax/text-classification) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/text_classification_flax.ipynb).
+- Check the [Text classification task guide](../tasks/sequence_classification) on how to use the model.
+
+
+<PipelineTag pipeline="token-classification"/>
+
+
+- [`AlbertForTokenClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/token-classification).
+
+
+- [`TFAlbertForTokenClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/token-classification) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/token_classification-tf.ipynb).
+
+
+
+- [`FlaxAlbertForTokenClassification`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/flax/token-classification).
+- [Token classification](https://huggingface.co/course/chapter7/2?fw=pt) chapter of the 🤗 Hugging Face Course.
+- Check the [Token classification task guide](../tasks/token_classification) on how to use the model.
+
+<PipelineTag pipeline="fill-mask"/>
+
+- [`AlbertForMaskedLM`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/language-modeling#robertabertdistilbert-and-masked-language-modeling) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/language_modeling.ipynb).
+- [`TFAlbertForMaskedLM`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/language-modeling#run_mlmpy) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/language_modeling-tf.ipynb).
+- [`FlaxAlbertForMaskedLM`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/flax/language-modeling#masked-language-modeling) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/masked_language_modeling_flax.ipynb).
+- [Masked language modeling](https://huggingface.co/course/chapter7/3?fw=pt) chapter of the 🤗 Hugging Face Course.
+- Check the [Masked language modeling task guide](../tasks/masked_language_modeling) on how to use the model.
+
+<PipelineTag pipeline="question-answering"/>
+
+- [`AlbertForQuestionAnswering`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/question-answering) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/question_answering.ipynb).
+- [`TFAlbertForQuestionAnswering`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/question-answering) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/question_answering-tf.ipynb).
+- [`FlaxAlbertForQuestionAnswering`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/flax/question-answering).
+- [Question answering](https://huggingface.co/course/chapter7/7?fw=pt) chapter of the 🤗 Hugging Face Course.
+- Check the [Question answering task guide](../tasks/question_answering) on how to use the model.
+
+**Multiple choice**
+
+- [`AlbertForMultipleChoice`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/pytorch/multiple-choice) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/multiple_choice.ipynb).
+- [`TFAlbertForMultipleChoice`] is supported by this [example script](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/multiple-choice) and [notebook](https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/multiple_choice-tf.ipynb).
+
+- Check the  [Multiple choice task guide](../tasks/multiple_choice) on how to use the model.
+
 
 ## AlbertConfig
 

docs/source/en/model_doc/audio-spectrogram-transformer.md

@@ -29,7 +29,7 @@ The abstract from the paper is the following:
 <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/audio_spectogram_transformer_architecture.png"
 alt="drawing" width="600"/>
 
-<small> Audio pectrogram Transformer architecture. Taken from the <a href="https://arxiv.org/abs/2104.01778">original paper</a>.</small>
+<small> Audio Spectrogram Transformer architecture. Taken from the <a href="https://arxiv.org/abs/2104.01778">original paper</a>.</small>
 
 This model was contributed by [nielsr](https://huggingface.co/nielsr).
 The original code can be found [here](https://github.com/YuanGongND/ast).
@@ -72,4 +72,4 @@ If you're interested in submitting a resource to be included here, please feel f
 ## ASTForAudioClassification
 
 [[autodoc]] ASTForAudioClassification
-    - forward
+    - forward

docs/source/en/model_doc/auto.md

@@ -49,7 +49,7 @@ You will then be able to use the auto classes like you would usually do!
 
 <Tip warning={true}>
 
-If your `NewModelConfig` is a subclass of [`~transformer.PretrainedConfig`], make sure its
+If your `NewModelConfig` is a subclass of [`~transformers.PretrainedConfig`], make sure its
 `model_type` attribute is set to the same key you use when registering the config (here `"new-model"`).
 
 Likewise, if your `NewModel` is a subclass of [`PreTrainedModel`], make sure its

docs/source/en/model_doc/bark.md

@@ -83,10 +83,10 @@ pip install -U flash-attn --no-build-isolation
 
 ##### Usage
 
-To load a model using Flash Attention 2, we can pass the `use_flash_attention_2` flag to [`.from_pretrained`](https://huggingface.co/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel.from_pretrained). We'll also load the model in half-precision (e.g. `torch.float16`), since it results in almost no degradation to audio quality but significantly lower memory usage and faster inference:
+To load a model using Flash Attention 2, we can pass the `attn_implementation="flash_attention_2"` flag to [`.from_pretrained`](https://huggingface.co/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel.from_pretrained). We'll also load the model in half-precision (e.g. `torch.float16`), since it results in almost no degradation to audio quality but significantly lower memory usage and faster inference:
 
 ```python
-model = BarkModel.from_pretrained("suno/bark-small", torch_dtype=torch.float16, use_flash_attention_2=True).to(device)
+model = BarkModel.from_pretrained("suno/bark-small", torch_dtype=torch.float16, attn_implementation="flash_attention_2").to(device)
 ```
 
 ##### Performance comparison
@@ -114,7 +114,7 @@ import torch
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 # load in fp16 and use Flash Attention 2
-model = BarkModel.from_pretrained("suno/bark-small", torch_dtype=torch.float16, use_flash_attention_2=True).to(device)
+model = BarkModel.from_pretrained("suno/bark-small", torch_dtype=torch.float16, attn_implementation="flash_attention_2").to(device)
 
 # enable CPU offload
 model.enable_cpu_offload()

docs/source/en/model_doc/biogpt.md

@@ -18,8 +18,7 @@ rendered properly in your Markdown viewer.
 
 ## Overview
 
-The BioGPT model was proposed in [BioGPT: generative pre-trained transformer for biomedical text generation and mining
-](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu. BioGPT is a domain-specific generative pre-trained Transformer language model for biomedical text generation and mining. BioGPT follows the Transformer language model backbone, and is pre-trained on 15M PubMed abstracts from scratch.
+The BioGPT model was proposed in [BioGPT: generative pre-trained transformer for biomedical text generation and mining](https://academic.oup.com/bib/advance-article/doi/10.1093/bib/bbac409/6713511?guestAccessKey=a66d9b5d-4f83-4017-bb52-405815c907b9) by Renqian Luo, Liai Sun, Yingce Xia, Tao Qin, Sheng Zhang, Hoifung Poon and Tie-Yan Liu. BioGPT is a domain-specific generative pre-trained Transformer language model for biomedical text generation and mining. BioGPT follows the Transformer language model backbone, and is pre-trained on 15M PubMed abstracts from scratch.
 
 The abstract from the paper is the following:
 

docs/source/en/model_doc/dinov2.md

@@ -25,6 +25,49 @@ The abstract from the paper is the following:
 This model was contributed by [nielsr](https://huggingface.co/nielsr).
 The original code can be found [here](https://github.com/facebookresearch/dinov2).
 
+## Usage tips
+
+The model can be traced using `torch.jit.trace` which leverages JIT compilation to optimize the model making it faster to run. Note this still produces some mis-matched elements and the difference between the original model and the traced model is of the order of 1e-4.
+
+```python
+import torch
+from transformers import AutoImageProcessor, AutoModel
+from PIL import Image
+import requests
+
+url = 'http://images.cocodataset.org/val2017/000000039769.jpg'
+image = Image.open(requests.get(url, stream=True).raw)
+
+processor = AutoImageProcessor.from_pretrained('facebook/dinov2-base')
+model = AutoModel.from_pretrained('facebook/dinov2-base')
+
+inputs = processor(images=image, return_tensors="pt")
+outputs = model(**inputs)
+last_hidden_states = outputs[0]
+
+# We have to force return_dict=False for tracing
+model.config.return_dict = False
+
+with torch.no_grad():
+    traced_model = torch.jit.trace(model, [inputs.pixel_values])
+    traced_outputs = traced_model(inputs.pixel_values)
+
+print((last_hidden_states - traced_outputs[0]).abs().max())
+```
+
+## Resources
+
+A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with DPT.
+
+- Demo notebooks for DINOv2 can be found [here](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/DINOv2). 🌎
+
+<PipelineTag pipeline="image-classification"/>
+
+- [`Dinov2ForImageClassification`] 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.
+
 ## Dinov2Config
 
 [[autodoc]] Dinov2Config

docs/source/en/model_doc/distilbert.md

@@ -32,7 +32,7 @@ rendered properly in your Markdown viewer.
 
 The DistilBERT model was proposed in the blog post [Smaller, faster, cheaper, lighter: Introducing DistilBERT, a
 distilled version of BERT](https://medium.com/huggingface/distilbert-8cf3380435b5), and the paper [DistilBERT, a
-distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/papers/1910.01108). DistilBERT is a
+distilled version of BERT: smaller, faster, cheaper and lighter](https://arxiv.org/abs/1910.01108). DistilBERT is a
 small, fast, cheap and light Transformer model trained by distilling BERT base. It has 40% less parameters than
 *bert-base-uncased*, runs 60% faster while preserving over 95% of BERT's performances as measured on the GLUE language
 understanding benchmark.
@@ -153,7 +153,7 @@ To load and run a model using Flash Attention 2, refer to the snippet below:
 >>> device = "cuda" # the device to load the model onto
 
 >>> tokenizer = AutoTokenizer.from_pretrained('distilbert-base-uncased')
->>> model = AutoModel.from_pretrained("distilbert-base-uncased", torch_dtype=torch.float16, use_flash_attention_2=True)
+>>> model = AutoModel.from_pretrained("distilbert-base-uncased", torch_dtype=torch.float16, attn_implementation="flash_attention_2")
 
 >>> text = "Replace me by any text you'd like."
 

docs/source/en/model_doc/dpt.md

@@ -32,6 +32,21 @@ alt="drawing" width="600"/>
 
 This model was contributed by [nielsr](https://huggingface.co/nielsr). The original code can be found [here](https://github.com/isl-org/DPT).
 
+## Usage tips
+
+DPT is compatible with the [`AutoBackbone`] class. This allows to use the DPT framework with various computer vision backbones available in the library, such as [`VitDetBackbone`] or [`Dinov2Backbone`]. One can create it as follows:
+
+```python
+from transformers import Dinov2Config, DPTConfig, DPTForDepthEstimation
+
+# initialize with a Transformer-based backbone such as DINOv2
+# in that case, we also specify `reshape_hidden_states=False` to get feature maps of shape (batch_size, num_channels, height, width)
+backbone_config = Dinov2Config.from_pretrained("facebook/dinov2-base", out_features=["stage1", "stage2", "stage3", "stage4"], reshape_hidden_states=False)
+
+config = DPTConfig(backbone_config=backbone_config)
+model = DPTForDepthEstimation(config=config)
+```
+
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with DPT.

docs/source/en/model_doc/gpt_bigcode.md

@@ -59,7 +59,7 @@ To load and run a model using Flash Attention 2, refer to the snippet below:
 >>> from transformers import AutoModelForCausalLM, AutoTokenizer
 >>> device = "cuda" # the device to load the model onto
 
->>> model = AutoModelForCausalLM.from_pretrained("bigcode/gpt_bigcode-santacoder", torch_dtype=torch.float16, use_flash_attention_2=True)
+>>> model = AutoModelForCausalLM.from_pretrained("bigcode/gpt_bigcode-santacoder", torch_dtype=torch.float16, attn_implementation="flash_attention_2")
 >>> tokenizer = AutoTokenizer.from_pretrained("bigcode/gpt_bigcode-santacoder")
 
 >>> prompt = "def hello_world():"

docs/source/en/model_doc/gpt_neo.md

@@ -56,13 +56,9 @@ The `generate()` method can be used to generate text using GPT Neo model.
 
 ## Combining GPT-Neo and Flash Attention 2
 
-First, make sure to install the latest version of Flash Attention 2 to include the sliding window attention feature.
+First, make sure to install the latest version of Flash Attention 2 to include the sliding window attention feature, and make sure your hardware is compatible with Flash-Attention 2. More details are available [here](https://huggingface.co/docs/transformers/perf_infer_gpu_one#flashattention-2) concerning the installation.
 
-```bash
-pip install -U flash-attn --no-build-isolation
-```
-
-Make also sure that you have a hardware that is compatible with Flash-Attention 2. Read more about it in the official documentation of flash-attn repository. Make also sure to load your model in half-precision (e.g. `torch.float16``)
+Make sure as well to load your model in half-precision (e.g. `torch.float16`).
 
 To load and run a model using Flash Attention 2, refer to the snippet below:
 
@@ -71,7 +67,7 @@ To load and run a model using Flash Attention 2, refer to the snippet below:
 >>> from transformers import AutoModelForCausalLM, AutoTokenizer
 >>> device = "cuda" # the device to load the model onto
 
->>> model = AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neo-2.7B", torch_dtype=torch.float16, use_flash_attention_2=True)
+>>> model = AutoModelForCausalLM.from_pretrained("EleutherAI/gpt-neo-2.7B", torch_dtype=torch.float16, attn_implementation="flash_attention_2")
 >>> tokenizer = AutoTokenizer.from_pretrained("EleutherAI/gpt-neo-2.7B")
 
 >>> prompt = "def hello_world():"

docs/source/en/model_doc/gpt_neox.md

@@ -61,6 +61,40 @@ The `generate()` method can be used to generate text using GPT Neo model.
 >>> gen_text = tokenizer.batch_decode(gen_tokens)[0]
 ```
 
+## Using Flash Attention 2
+
+Flash Attention 2 is an faster, optimized version of the model.
+
+### Installation 
+
+First, check whether your hardware is compatible with Flash Attention 2. The latest list of compatible hardware can be found in the [official documentation](https://github.com/Dao-AILab/flash-attention#installation-and-features). If your hardware is not compatible with Flash Attention 2, you can still benefit from attention kernel optimisations through Better Transformer support covered [above](https://huggingface.co/docs/transformers/main/en/model_doc/bark#using-better-transformer).
+
+Next, [install](https://github.com/Dao-AILab/flash-attention#installation-and-features) the latest version of Flash Attention 2:
+
+```bash
+pip install -U flash-attn --no-build-isolation
+```
+
+### Usage
+
+To load a model using Flash Attention 2, we can pass the argument `attn_implementation="flash_attention_2"` to [`.from_pretrained`](https://huggingface.co/docs/transformers/main/en/main_classes/model#transformers.PreTrainedModel.from_pretrained). We'll also load the model in half-precision (e.g. `torch.float16`), since it results in almost no degradation to audio quality but significantly lower memory usage and faster inference:
+
+```python
+>>> from transformers import GPTNeoXForCausalLM, GPTNeoXTokenizerFast
+
+model = GPTNeoXForCausalLM.from_pretrained("EleutherAI/gpt-neox-20b", torch_dtype=torch.float16, attn_implementation="flash_attention_2").to(device)
+...
+```
+
+
+### Expected speedups
+
+Below is an expected speedup diagram that compares pure inference time between the native implementation in transformers using `stockmark/gpt-neox-japanese-1.4b` checkpoint and the Flash Attention 2 version of the model using a sequence length of 2048.
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/ybelkada/documentation-images/resolve/main/gpt-neox-1.8b-speedup.jpg">
+</div>
+
 ## Resources
 
 - [Causal language modeling task guide](../tasks/language_modeling)

docs/source/en/model_doc/llama.md

@@ -50,6 +50,9 @@ come in several checkpoints they each contain a part of each weight of the model
 
 - The LLaMA tokenizer is a BPE model based on [sentencepiece](https://github.com/google/sentencepiece). One quirk of sentencepiece is that when decoding a sequence, if the first token is the start of the word (e.g. "Banana"), the tokenizer does not prepend the prefix space to the string.
 
+This model was contributed by [zphang](https://huggingface.co/zphang) with contributions from [BlackSamorez](https://huggingface.co/BlackSamorez). The code of the implementation in Hugging Face is based on GPT-NeoX [here](https://github.com/EleutherAI/gpt-neox). The original code of the authors can be found [here](https://github.com/facebookresearch/llama). The Flax version of the implementation was contributed by [afmck](https://huggingface.co/afmck) with the code in the implementation based on Hugging Face's Flax GPT-Neo.
+
+
 Based on the original LLaMA model, Meta AI has released some follow-up works:
 
 - **Llama2**: Llama2 is an improved version of Llama with some architectural tweaks (Grouped Query Attention), and is pre-trained on 2Trillion tokens. Refer to the documentation of Llama2 which can be found [here](llama2).
@@ -112,3 +115,13 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 
 [[autodoc]] LlamaForSequenceClassification
     - forward
+
+## FlaxLlamaModel
+
+[[autodoc]] FlaxLlamaModel
+    - __call__
+
+## FlaxLlamaForCausalLM
+
+[[autodoc]] FlaxLlamaForCausalLM
+    - __call__

docs/source/en/model_doc/llava.md

@@ -0,0 +1,80 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ 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.
+
+-->
+
+# LLaVa
+
+## Overview
+
+LLaVa is an open-source chatbot trained by fine-tuning LlamA/Vicuna on GPT-generated multimodal instruction-following data. It is an auto-regressive language model, based on the transformer architecture. In other words, it is an multi-modal version of LLMs fine-tuned for chat / instructions.
+
+The LLaVa model was proposed in [Visual Instruction Tuning](https://arxiv.org/abs/2304.08485) and improved in [Improved Baselines with Visual Instruction Tuning](https://arxiv.org/pdf/2310.03744) by Haotian Liu, Chunyuan Li, Yuheng Li and Yong Jae Lee.
+
+The abstract from the paper is the following:
+
+*Large multimodal models (LMM) have recently shown encouraging progress with visual instruction tuning. In this note, we show that the fully-connected vision-language cross-modal connector in LLaVA is surprisingly powerful and data-efficient. With simple modifications to LLaVA, namely, using CLIP-ViT-L-336px with an MLP projection and adding academic-task-oriented VQA data with simple response formatting prompts, we establish stronger baselines that achieve state-of-the-art across 11 benchmarks. Our final 13B checkpoint uses merely 1.2M publicly available data, and finishes full training in ∼1 day on a single 8-A100 node. We hope this can make state-of-the-art LMM research more accessible. Code and model will be publicly available*
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/llava_architecture.jpg"
+alt="drawing" width="600"/>
+
+<small> LLaVa architecture. Taken from the <a href="https://arxiv.org/abs/2304.08485">original paper.</a> </small>
+
+This model was contributed by [ArthurZ](https://huggingface.co/ArthurZ) and [ybelkada](https://huggingface.co/ybelkada).
+The original code can be found [here](https://github.com/haotian-liu/LLaVA/tree/main/llava).
+
+## Usage tips
+
+- We advise users to use `padding_side="left"` when computing batched generation as it leads to more accurate results. Simply make sure to call `processor.tokenizer.padding_side = "left"` before generating.
+
+- Note the model has not been explicitly trained to process multiple images in the same prompt, although this is technically possible, you may experience inaccurate results.
+
+- For better results, we recommend users to prompt the model with the correct prompt format: 
+
+```bash
+"USER: <image>\n<prompt>ASSISTANT:"
+```
+
+For multiple turns conversation:
+
+```bash
+"USER: <image>\n<prompt1>ASSISTANT: <answer1>USER: <prompt2>ASSISTANT: <answer2>USER: <prompt3>ASSISTANT:"
+```
+
+### Using Flash Attention 2
+
+Flash Attention 2 is an even faster, optimized version of the previous optimization, please refer to the [Flash Attention 2 section of performance docs](https://huggingface.co/docs/transformers/perf_infer_gpu_one).
+
+## Resources
+
+A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with BEiT.
+
+<PipelineTag pipeline="image-to-text"/>
+
+- A [Google Colab demo](https://colab.research.google.com/drive/1qsl6cd2c8gGtEW1xV5io7S8NHh-Cp1TV?usp=sharing) on how to run Llava on a free-tier Google colab instance leveraging 4-bit inference.
+- A [similar notebook](https://github.com/NielsRogge/Transformers-Tutorials/blob/master/LLaVa/Inference_with_LLaVa_for_multimodal_generation.ipynb) showcasing batched inference. 🌎
+
+
+## LlavaConfig
+
+[[autodoc]] LlavaConfig
+
+## LlavaProcessor
+
+[[autodoc]] LlavaProcessor
+
+## LlavaForConditionalGeneration
+
+[[autodoc]] LlavaForConditionalGeneration
+    - forward

docs/source/en/model_doc/madlad-400.md

@@ -0,0 +1,68 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ Note that this file is in Markdown but contains specific syntax for our doc-builder (similar to MDX) that may not be
+rendered properly in your Markdown viewer.
+
+-->
+
+# MADLAD-400
+
+## Overview
+
+MADLAD-400 models were released in the paper [MADLAD-400: A Multilingual And Document-Level Large Audited Dataset](MADLAD-400: A Multilingual And Document-Level Large Audited Dataset). 
+
+The abstract from the paper is the following: 
+
+*We introduce MADLAD-400, a manually audited, general domain 3T token monolingual dataset based on CommonCrawl, spanning 419 languages. We discuss 
+the limitations revealed by self-auditing MADLAD-400, and the role data auditing
+had in the dataset creation process. We then train and release a 10.7B-parameter
+multilingual machine translation model on 250 billion tokens covering over 450
+languages using publicly available data, and find that it is competitive with models
+that are significantly larger, and report the results on different domains. In addition, we train a 8B-parameter language model, and assess the results on few-shot
+translation. We make the baseline models 1
+available to the research community.*
+
+This model was added by [Juarez Bochi](https://huggingface.co/jbochi). The original checkpoints can be found [here](https://github.com/google-research/google-research/tree/master/madlad_400). 
+
+This is a machine translation model that supports many low-resource languages, and that is competitive with models that are significantly larger.
+
+One can directly use MADLAD-400 weights without finetuning the model:
+
+```python
+>>> from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+
+>>> model = AutoModelForSeq2SeqLM.from_pretrained("google/madlad400-3b-mt")
+>>> tokenizer = AutoTokenizer.from_pretrained("google/madlad400-3b-mt")
+
+>>> inputs = tokenizer("<2pt> I love pizza!", return_tensors="pt")
+>>> outputs = model.generate(**inputs)
+>>> print(tokenizer.batch_decode(outputs, skip_special_tokens=True))
+['Eu amo pizza!']
+```
+
+Google has released the following variants:
+
+- [google/madlad400-3b-mt](https://huggingface.co/google/madlad400-3b-mt)
+
+- [google/madlad400-7b-mt](https://huggingface.co/google/madlad400-7b-mt)
+
+- [google/madlad400-7b-mt-bt](https://huggingface.co/google/madlad400-7b-mt-bt)
+
+- [google/madlad400-10b-mt](https://huggingface.co/google/madlad400-10b-mt)
+
+The original checkpoints can be found [here](https://github.com/google-research/google-research/tree/master/madlad_400).
+
+<Tip>
+
+Refer to [T5's documentation page](t5) for all API references, code examples, and notebooks. For more details regarding training and evaluation of the MADLAD-400, refer to the model card.
+
+</Tip>

docs/source/en/model_doc/mistral.md

@@ -99,7 +99,7 @@ To load and run a model using Flash Attention 2, refer to the snippet below:
 >>> from transformers import AutoModelForCausalLM, AutoTokenizer
 >>> device = "cuda" # the device to load the model onto
 
->>> model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1", torch_dtype=torch.float16, use_flash_attention_2=True)
+>>> model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1", torch_dtype=torch.float16, attn_implementation="flash_attention_2")
 >>> tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
 
 >>> prompt = "My favourite condiment is"

docs/source/en/model_doc/mixtral.md

@@ -0,0 +1,163 @@
+<!--Copyright 2023 Mistral AI and The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ 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.
+
+-->
+
+# Mixtral
+
+## Overview
+
+Mixtral-8x7B is Mistral AI's second Large Language Model (LLM). 
+
+The Mixtral model was proposed in the by the [Mistral AI](https://mistral.ai/) team.
+
+It was introduced in the [Mixtral of Experts blogpost](https://mistral.ai/news/mixtral-of-experts/) with the following introduction:
+
+*Today, the team is proud to release Mixtral 8x7B, a high-quality sparse mixture of experts models (SMoE) with open weights. Licensed under Apache 2.0. Mixtral outperforms Llama 2 70B on most benchmarks with 6x faster inference. It is the strongest open-weight model with a permissive license and the best model overall regarding cost/performance trade-offs. In particular, it matches or outperforms GPT3.5 on most standard benchmarks.*
+
+Tips:
+
+
+- The model needs to be converted using the [conversion script](https://github.com/huggingface/transformers/blob/main/src/transformers/models/mixtral/convert_mixtral_weights_to_hf.py).
+- If the model is quantized to 4bits, a single A100 is enough to fit the entire 84B model.
+
+This model was contributed by [Younes Belkada](https://huggingface.co/ybelkada) and [Arthur Zucker](https://huggingface.co/ArthurZ) .
+The original code can be found [here](https://github.com/mistralai/mistral-src).
+
+
+### Model Details
+
+Mixtral-84B is a decoder-based LM with the following architectural choices:
+
+* Mixtral is a Mixture of Expert (MOE) model with 8 experts per MLP, with a total of 85B paramateres but the compute required is the same as a 14B model. This is because even though each experts have to be loaded in RAM (70B like ram requirement) each token from the hidden states are dipatched twice (top 2 routing) and thus the compute (the operation required at each foward computation) is just 2 X sequence_length. 
+
+The following implementation details are shared with Mistral AI's first model [mistral](~models/doc/mistral):
+* Sliding Window Attention - Trained with 8k context length and fixed cache size, with a theoretical attention span of 128K tokens
+* GQA (Grouped Query Attention) - allowing faster inference and lower cache size.
+* Byte-fallback BPE tokenizer - ensures that characters are never mapped to out of vocabulary tokens.
+
+They also provide an instruction fine-tuned model: `mistralai/Mixtral-8x7B-v0.1` which can be used for chat-based inference.
+
+For more details please read our [release blog post](https://mistral.ai/news/mixtral-of-experts/)
+
+### License
+
+`Mixtral-8x7B` is released under the Apache 2.0 license.
+
+## Usage tips
+
+`Mixtral-8x7B` can be found on the [Huggingface Hub](https://huggingface.co/mistralai)
+
+These ready-to-use checkpoints can be downloaded and used via the HuggingFace Hub:
+
+```python
+>>> from transformers import AutoModelForCausalLM, AutoTokenizer
+>>> device = "cuda" # the device to load the model onto
+
+>>> model = AutoModelForCausalLM.from_pretrained("mistralai/Mixtral-8x7B-v0.1")
+>>> tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-8x7B")
+
+>>> prompt = "My favourite condiment is"
+
+>>> model_inputs = tokenizer([prompt], return_tensors="pt").to(device)
+>>> model.to(device)
+
+>>> generated_ids = model.generate(**model_inputs, max_new_tokens=100, do_sample=True)
+>>> tokenizer.batch_decode(generated_ids)[0]
+"The expected output"
+```
+
+To use the raw checkpoints with HuggingFace you can use the `convert_mixtral_weights_to_hf.py` script to convert them to the HuggingFace format:
+
+```bash
+python src/transformers/models/mixtral/convert_mixtral_weights_to_hf.py \
+    --input_dir /path/to/downloaded/mistral/weights --output_dir /output/path
+```
+
+You can then load the converted model from the `output/path`:
+
+```python
+from transformers import MixtralForCausalLM, LlamaTokenizer
+
+tokenizer = LlamaTokenizer.from_pretrained("/output/path")
+model = MixtralForCausalLM.from_pretrained("/output/path")
+```
+
+## Combining Mixtral and Flash Attention 2
+
+First, make sure to install the latest version of Flash Attention 2 to include the sliding window attention feature.
+
+```bash
+pip install -U flash-attn --no-build-isolation
+```
+
+Make also sure that you have a hardware that is compatible with Flash-Attention 2. Read more about it in the official documentation of [`flash-attn`](https://github.com/Dao-AILab/flash-attention) repository. Make also sure to load your model in half-precision (e.g. `torch.float16`)
+
+To load and run a model using Flash Attention 2, refer to the snippet below:
+
+```python
+>>> import torch
+>>> from transformers import AutoModelForCausalLM, AutoTokenizer
+>>> device = "cuda" # the device to load the model onto
+
+>>> model = AutoModelForCausalLM.from_pretrained("mistralai/Mixtral-8x7B-v0.1", torch_dtype=torch.float16, attn_implementation="flash_attention_2")
+>>> tokenizer = AutoTokenizer.from_pretrained("mistralai/Mixtral-8x7B-v0.1")
+
+>>> prompt = "My favourite condiment is"
+
+>>> model_inputs = tokenizer([prompt], return_tensors="pt").to(device)
+>>> model.to(device)
+
+>>> generated_ids = model.generate(**model_inputs, max_new_tokens=100, do_sample=True)
+>>> tokenizer.batch_decode(generated_ids)[0]
+"The expected output"
+```
+
+### Expected speedups
+
+Below is a expected speedup diagram that compares pure inference time between the native implementation in transformers using `mistralai/Mixtral-8x7B-v0.1` checkpoint and the Flash Attention 2 version of the model.
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/ybelkada/documentation-images/resolve/main/mixtral-7b-inference-large-seqlen.png">
+</div>
+
+### Sliding window Attention
+
+The current implementation supports the sliding window attention mechanism and memory efficient cache management. 
+To enable sliding window attention, just make sure to have a `flash-attn` version that is compatible with sliding window attention (`>=2.3.0`). 
+
+The Flash Attention-2 model uses also a more memory efficient cache slicing mechanism - as recommended per the official implementation of Mistral model that use rolling cache mechanism we keep the cache size fixed (`self.config.sliding_window`), support batched generation only for `padding_side="left"` and use the absolute position of the current token to compute the positional embedding.
+
+## The Mistral Team
+
+Albert Jiang, Alexandre Sablayrolles, Arthur Mensch, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Florian Bressand, Gianna Lengyel, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stock, Teven Le Scao, Thibaut Lavril, Thomas Wang, Timothée Lacroix, William El Sayed.
+
+## MixtralConfig
+
+[[autodoc]] MixtralConfig
+
+## MixtralModel
+
+[[autodoc]] MixtralModel
+    - forward
+
+## MixtralForCausalLM
+
+[[autodoc]] MixtralForCausalLM
+    - forward
+
+## MixtralForSequenceClassification
+
+[[autodoc]] MixtralForSequenceClassification
+    - forward

docs/source/en/model_doc/musicgen.md

@@ -46,6 +46,16 @@ This model was contributed by [sanchit-gandhi](https://huggingface.co/sanchit-ga
 [here](https://github.com/facebookresearch/audiocraft). The pre-trained checkpoints can be found on the
 [Hugging Face Hub](https://huggingface.co/models?sort=downloads&search=facebook%2Fmusicgen-).
 
+## Usage tips
+
+- After downloading the original checkpoints from [here](https://github.com/facebookresearch/audiocraft/blob/main/docs/MUSICGEN.md#importing--exporting-models) , you can convert them using the **conversion script** available at
+`src/transformers/models/musicgen/convert_musicgen_transformers.py` with the following command:
+
+```bash
+python src/transformers/models/musicgen/convert_musicgen_transformers.py \
+    --checkpoint small --pytorch_dump_folder /output/path --safe_serialization 
+```
+
 ## Generation
 
 MusicGen is compatible with two generation modes: greedy and sampling. In practice, sampling leads to significantly

docs/source/en/model_doc/opt.md

@@ -62,6 +62,55 @@ The resource should ideally demonstrate something new instead of duplicating an
 
 - A blog post on [How 🤗 Accelerate runs very large models thanks to PyTorch](https://huggingface.co/blog/accelerate-large-models) with OPT.
 
+
+## Combining OPT and Flash Attention 2
+
+First, make sure to install the latest version of Flash Attention 2 to include the sliding window attention feature.
+
+```bash
+pip install -U flash-attn --no-build-isolation
+```
+
+Make also sure that you have a hardware that is compatible with Flash-Attention 2. Read more about it in the official documentation of flash-attn repository. Make also sure to load your model in half-precision (e.g. `torch.float16``)
+
+To load and run a model using Flash Attention 2, refer to the snippet below:
+
+```python
+>>> import torch
+>>> from transformers import OPTForCausalLM, GPT2Tokenizer
+>>> device = "cuda" # the device to load the model onto
+
+>>> model = OPTForCausalLM.from_pretrained("facebook/opt-350m", torch_dtype=torch.float16, attn_implementation="flash_attention_2")
+>>> tokenizer = GPT2Tokenizer.from_pretrained("facebook/opt-350m")
+
+>>> prompt = ("A chat between a curious human and the Statue of Liberty.\n\nHuman: What is your name?\nStatue: I am the "
+              "Statue of Liberty.\nHuman: Where do you live?\nStatue: New York City.\nHuman: How long have you lived "
+              "there?")
+
+>>> model_inputs = tokenizer([prompt], return_tensors="pt").to(device)
+>>> model.to(device)
+
+>>> generated_ids = model.generate(**model_inputs, max_new_tokens=30, do_sample=False)
+>>> tokenizer.batch_decode(generated_ids)[0]
+'</s>A chat between a curious human and the Statue of Liberty.\n\nHuman: What is your name?\nStatue: I am the Statue of Liberty.\nHuman: Where do you live?\nStatue: New York City.\nHuman: How long have you lived there?\nStatue: I have lived here for about a year.\nHuman: What is your favorite place to eat?\nStatue: I love'
+```
+
+### Expected speedups
+
+Below is an expected speedup diagram that compares pure inference time between the native implementation in transformers using `facebook/opt-2.7b` checkpoint and the Flash Attention 2 version of the model using two different sequence lengths.
+
+<div style="text-align: center">
+<img src="https://user-images.githubusercontent.com/49240599/281101546-d2fca6d2-ee44-48f3-9534-ba8d5bee4531.png">
+</div>
+
+Below is an expected speedup diagram that compares pure inference time between the native implementation in transformers using `facebook/opt-350m` checkpoint and the Flash Attention 2 version of the model using two different sequence lengths.
+
+<div style="text-align: center">
+<img src="https://user-images.githubusercontent.com/49240599/281101682-d1144e90-0dbc-46f4-8fc8-c6206cb793c9.png">
+</div>
+
+
+
 ## OPTConfig
 
 [[autodoc]] OPTConfig

docs/source/en/model_doc/patchtsmixer.md

@@ -0,0 +1,90 @@
+<!--Copyright 2023 IBM and 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.
+
+-->
+
+# PatchTSMixer
+
+## Overview
+
+The PatchTSMixer model was proposed in [TSMixer: Lightweight MLP-Mixer Model for Multivariate Time Series Forecasting](https://arxiv.org/pdf/2306.09364.pdf) by Vijay Ekambaram, Arindam Jati, Nam Nguyen, Phanwadee Sinthong and Jayant Kalagnanam.
+
+
+PatchTSMixer is a lightweight time-series modeling approach based on the MLP-Mixer architecture. In this HuggingFace implementation, we provide PatchTSMixer's capabilities to effortlessly facilitate lightweight mixing across patches, channels, and hidden features for effective multivariate time-series modeling. It also supports various attention mechanisms starting from simple gated attention to more complex self-attention blocks that can be customized accordingly. The model can be pretrained and subsequently used for various downstream tasks such as forecasting, classification and regression.
+
+
+The abstract from the paper is the following:
+
+*TSMixer is a lightweight neural architecture exclusively composed of multi-layer perceptron (MLP) modules designed for multivariate forecasting and representation learning on patched time series. Our model draws inspiration from the success of MLP-Mixer models in computer vision. We demonstrate the challenges involved in adapting Vision MLP-Mixer for time series and introduce empirically validated components to enhance accuracy. This includes a novel design paradigm of attaching online reconciliation heads to the MLP-Mixer backbone, for explicitly modeling the time-series properties such as hierarchy and channel-correlations. We also propose a Hybrid channel modeling approach to effectively handle noisy channel interactions and generalization across diverse datasets, a common challenge in existing patch channel-mixing methods. Additionally, a simple gated attention mechanism is introduced in the backbone to prioritize important features. By incorporating these lightweight components, we significantly enhance the learning capability of simple MLP structures, outperforming complex Transformer models with minimal computing usage. Moreover, TSMixer's modular design enables compatibility with both supervised and masked self-supervised learning methods, making it a promising building block for time-series Foundation Models. TSMixer outperforms state-of-the-art MLP and Transformer models in forecasting by a considerable margin of 8-60%. It also outperforms the latest strong benchmarks of Patch-Transformer models (by 1-2%) with a significant reduction in memory and runtime (2-3X).*
+
+
+
+This model was contributed by [ajati](https://huggingface.co/ajati), [vijaye12](https://huggingface.co/vijaye12), 
+[gsinthong](https://huggingface.co/gsinthong), [namctin](https://huggingface.co/namctin),
+[wmgifford](https://huggingface.co/wmgifford), [kashif](https://huggingface.co/kashif).
+
+
+## Sample usage 
+```python
+
+from transformers import PatchTSMixerConfig, PatchTSMixerForPrediction
+from transformers import Trainer, TrainingArguments,
+
+
+config = PatchTSMixerConfig(context_length = 512, prediction_length = 96)
+model = PatchTSMixerForPrediction(config)
+trainer = Trainer(model=model, args=training_args, 
+            train_dataset=train_dataset,
+            eval_dataset=valid_dataset)
+trainer.train()
+results = trainer.evaluate(test_dataset)
+```
+
+## Usage tips
+
+The model can also be used for time series classification and time series regression. See the respective [`PatchTSMixerForTimeSeriesClassification`] and [`PatchTSMixerForRegression`] classes.
+
+## PatchTSMixerConfig
+
+[[autodoc]] PatchTSMixerConfig
+
+
+## PatchTSMixerModel
+
+[[autodoc]] PatchTSMixerModel
+    - forward
+
+
+## PatchTSMixerForPrediction
+
+[[autodoc]] PatchTSMixerForPrediction
+    - forward
+
+
+## PatchTSMixerForTimeSeriesClassification
+
+[[autodoc]] PatchTSMixerForTimeSeriesClassification
+    - forward
+
+
+## PatchTSMixerForPretraining
+
+[[autodoc]] PatchTSMixerForPretraining
+    - forward
+
+
+## PatchTSMixerForRegression
+
+[[autodoc]] PatchTSMixerForRegression
+    - forward

docs/source/en/model_doc/patchtst.md

@@ -0,0 +1,65 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ 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.
+
+-->
+
+# PatchTST
+
+## Overview
+
+The PatchTST model was proposed in [A Time Series is Worth 64 Words: Long-term Forecasting with Transformers](https://arxiv.org/abs/2211.14730) by Yuqi Nie, Nam H. Nguyen, Phanwadee Sinthong and Jayant Kalagnanam.
+
+At a high level the model vectorizes time series into patches of a given size and encodes the resulting sequence of vectors via a Transformer that then outputs the prediction length forecast via an appropriate head. The model is illustrated in the following figure:
+
+![model](https://github.com/namctin/transformers/assets/8100/150af169-29de-419a-8d98-eb78251c21fa)
+
+The abstract from the paper is the following:
+
+*We propose an efficient design of Transformer-based models for multivariate time series forecasting and self-supervised representation learning. It is based on two key components: (i) segmentation of time series into subseries-level patches which are served as input tokens to Transformer; (ii) channel-independence where each channel contains a single univariate time series that shares the same embedding and Transformer weights across all the series. Patching design naturally has three-fold benefit: local semantic information is retained in the embedding; computation and memory usage of the attention maps are quadratically reduced given the same look-back window; and the model can attend longer history. Our channel-independent patch time series Transformer (PatchTST) can improve the long-term forecasting accuracy significantly when compared with that of SOTA Transformer-based models. We also apply our model to self-supervised pre-training tasks and attain excellent fine-tuning performance, which outperforms supervised training on large datasets. Transferring of masked pre-trained representation on one dataset to others also produces SOTA forecasting accuracy.*
+
+This model was contributed by [namctin](https://huggingface.co/namctin), [gsinthong](https://huggingface.co/gsinthong), [diepi](https://huggingface.co/diepi), [vijaye12](https://huggingface.co/vijaye12), [wmgifford](https://huggingface.co/wmgifford), and [kashif](https://huggingface.co/kashif). The original code can be found [here](https://github.com/yuqinie98/PatchTST).
+
+## Usage tips
+
+The model can also be used for time series classification and time series regression. See the respective [`PatchTSTForClassification`] and [`PatchTSTForRegression`] classes.
+
+
+## PatchTSTConfig
+
+[[autodoc]] PatchTSTConfig
+
+## PatchTSTModel
+
+[[autodoc]] PatchTSTModel
+    - forward
+
+## PatchTSTForPrediction
+
+[[autodoc]] PatchTSTForPrediction
+    - forward
+
+## PatchTSTForClassification
+
+[[autodoc]] PatchTSTForClassification
+    - forward
+
+## PatchTSTForPretraining
+
+[[autodoc]] PatchTSTForPretraining
+    - forward
+
+## PatchTSTForRegression
+
+[[autodoc]] PatchTSTForRegression
+    - forward

docs/source/en/model_doc/phi.md

@@ -76,7 +76,7 @@ The original code for Phi-1 and Phi-1.5 can be found [here](https://huggingface.
 ```python
 >>> from transformers import PhiForCausalLM, AutoTokenizer
 
->>> # define the model and tokenzier.
+>>> # define the model and tokenizer.
 >>> model = PhiForCausalLM.from_pretrained("susnato/phi-1_5_dev")
 >>> tokenizer = AutoTokenizer.from_pretrained("susnato/phi-1_5_dev")
 
@@ -94,6 +94,46 @@ The original code for Phi-1 and Phi-1.5 can be found [here](https://huggingface.
 ```
 
 
+## Combining Phi and Flash Attention 2
+
+First, make sure to install the latest version of Flash Attention 2 to include the sliding window attention feature.
+
+```bash
+pip install -U flash-attn --no-build-isolation
+```
+
+Make also sure that you have a hardware that is compatible with Flash-Attention 2. Read more about it in the official documentation of flash-attn repository. Make also sure to load your model in half-precision (e.g. `torch.float16``)
+
+To load and run a model using Flash Attention 2, refer to the snippet below:
+
+```python
+>>> import torch
+>>> from transformers import PhiForCausalLM, AutoTokenizer
+
+>>> # define the model and tokenizer and push the model and tokens to the GPU.
+>>> model = PhiForCausalLM.from_pretrained("susnato/phi-1_5_dev", torch_dtype=torch.float16, attn_implementation="flash_attention_2").to("cuda")
+>>> tokenizer = AutoTokenizer.from_pretrained("susnato/phi-1_5_dev")
+
+>>> # feel free to change the prompt to your liking.
+>>> prompt = "If I were an AI that had just achieved"
+
+>>> # apply the tokenizer.
+>>> tokens = tokenizer(prompt, return_tensors="pt").to("cuda")
+
+>>> # use the model to generate new tokens.
+>>> generated_output = model.generate(**tokens, use_cache=True, max_new_tokens=10)
+
+>>> tokenizer.batch_decode(generated_output)[0]
+'If I were an AI that had just achieved a breakthrough in machine learning, I would be thrilled'
+```
+
+### Expected speedups
+Below is an expected speedup diagram that compares pure inference time between the native implementation in transformers using `susnato/phi-1_dev` checkpoint and the Flash Attention 2 version of the model using a sequence length of 2048.
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/ybelkada/documentation-images/resolve/main/phi_1_speedup_plot.jpg">
+</div>
+
+
 ## PhiConfig
 
 [[autodoc]] PhiConfig
@@ -123,4 +163,4 @@ The original code for Phi-1 and Phi-1.5 can be found [here](https://huggingface.
     - forward
 
 </pt>
-</frameworkcontent>
+</frameworkcontent>

docs/source/en/model_doc/seamless_m4t.md

@@ -15,6 +15,7 @@ specific language governing permissions and limitations under the License.
 ## Overview
 
 The SeamlessM4T model was proposed in [SeamlessM4T — Massively Multilingual & Multimodal Machine Translation](https://dl.fbaipublicfiles.com/seamless/seamless_m4t_paper.pdf) by the Seamless Communication team from Meta AI.
+This is the version 1 release of the model. For the updated version 2 release, refer to the [Seamless M4T v2 docs](./seamless_m4t_v2.md).
 
 SeamlessM4T is a collection of models designed to provide high quality translation, allowing people from different linguistic communities to communicate effortlessly through speech and text.
 

docs/source/en/model_doc/seamless_m4t_v2.md

@@ -0,0 +1,194 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+-->
+
+# SeamlessM4T-v2
+
+## Overview
+
+The SeamlessM4T-v2 model was proposed in [Seamless: Multilingual Expressive and Streaming Speech Translation](https://ai.meta.com/research/publications/seamless-multilingual-expressive-and-streaming-speech-translation/) by the Seamless Communication team from Meta AI.
+
+SeamlessM4T-v2 is a collection of models designed to provide high quality translation, allowing people from different linguistic communities to communicate effortlessly through speech and text. It is an improvement on the [previous version](./seamless_m4t.md). For more details on the differences between v1 and v2, refer to section [Difference with SeamlessM4T-v1](#difference-with-seamlessm4t-v1).
+
+SeamlessM4T-v2 enables multiple tasks without relying on separate models:
+
+- Speech-to-speech translation (S2ST)
+- Speech-to-text translation (S2TT)
+- Text-to-speech translation (T2ST)
+- Text-to-text translation (T2TT)
+- Automatic speech recognition (ASR)
+
+[`SeamlessM4Tv2Model`] can perform all the above tasks, but each task also has its own dedicated sub-model.
+
+The abstract from the paper is the following:
+
+*Recent advancements in automatic speech translation have dramatically expanded language coverage, improved multimodal capabilities, and enabled a wide range of tasks and functionalities. That said, large-scale automatic speech translation systems today lack key features that help machine-mediated communication feel seamless when compared to human-to-human dialogue. In this work, we introduce a family of models that enable end-to-end expressive and multilingual translations in a streaming fashion. First, we contribute an improved version of the massively multilingual and multimodal SeamlessM4T model—SeamlessM4T v2. This newer model, incorporating an updated UnitY2 framework, was trained on more low-resource language data. The expanded version of SeamlessAlign adds 114,800 hours of automatically aligned data for a total of 76 languages. SeamlessM4T v2 provides the foundation on which our two newest models, SeamlessExpressive and SeamlessStreaming, are initiated. SeamlessExpressive enables translation that preserves vocal styles and prosody. Compared to previous efforts in expressive speech research, our work addresses certain underexplored aspects of prosody, such as speech rate and pauses, while also preserving the style of one’s voice. As for SeamlessStreaming, our model leverages the Efficient Monotonic Multihead Attention (EMMA) mechanism to generate low-latency target translations without waiting for complete source utterances. As the first of its kind, SeamlessStreaming enables simultaneous speech-to-speech/text translation for multiple source and target languages. To understand the performance of these models, we combined novel and modified versions of existing automatic metrics to evaluate prosody, latency, and robustness. For human evaluations, we adapted existing protocols tailored for measuring the most relevant attributes in the preservation of meaning, naturalness, and expressivity. To ensure that our models can be used safely and responsibly, we implemented the first known red-teaming effort for multimodal machine translation, a system for the detection and mitigation of added toxicity, a systematic evaluation of gender bias, and an inaudible localized watermarking mechanism designed to dampen the impact of deepfakes. Consequently, we bring major components from SeamlessExpressive and SeamlessStreaming together to form Seamless, the first publicly available system that unlocks expressive cross-lingual communication in real-time. In sum, Seamless gives us a pivotal look at the technical foundation needed to turn the Universal Speech Translator from a science fiction concept into a real-world technology. Finally, contributions in this work—including models, code, and a watermark detector—are publicly released and accessible at the link below.*
+
+## Usage
+
+In the following example, we'll load an Arabic audio sample and an English text sample and convert them into Russian speech and French text.
+
+First, load the processor and a checkpoint of the model:
+
+```python
+>>> from transformers import AutoProcessor, SeamlessM4Tv2Model
+
+>>> processor = AutoProcessor.from_pretrained("facebook/seamless-m4t-v2-large")
+>>> model = SeamlessM4Tv2Model.from_pretrained("facebook/seamless-m4t-v2-large")
+```
+
+You can seamlessly use this model on text or on audio, to generated either translated text or translated audio.
+
+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)
+>>> audio_sample = next(iter(dataset))["audio"]
+
+>>> # now, process it
+>>> audio_inputs = processor(audios=audio_sample["array"], return_tensors="pt")
+
+>>> # now, process some English text as well
+>>> text_inputs = processor(text = "Hello, my dog is cute", src_lang="eng", return_tensors="pt")
+```
+
+
+### Speech
+
+[`SeamlessM4Tv2Model`] can *seamlessly* generate text or speech with few or no changes. Let's target Russian voice translation:
+
+```python
+>>> audio_array_from_text = model.generate(**text_inputs, tgt_lang="rus")[0].cpu().numpy().squeeze()
+>>> audio_array_from_audio = model.generate(**audio_inputs, tgt_lang="rus")[0].cpu().numpy().squeeze()
+```
+
+With basically the same code, I've translated English text and Arabic speech to Russian speech samples.
+
+### Text
+
+Similarly, you can generate translated text from audio files or from text with the same model. You only have to pass `generate_speech=False` to [`SeamlessM4Tv2Model.generate`].
+This time, let's translate to French.
+
+```python 
+>>> # from audio
+>>> output_tokens = model.generate(**audio_inputs, tgt_lang="fra", generate_speech=False)
+>>> translated_text_from_audio = processor.decode(output_tokens[0].tolist()[0], skip_special_tokens=True)
+
+>>> # from text
+>>> output_tokens = model.generate(**text_inputs, tgt_lang="fra", generate_speech=False)
+>>> translated_text_from_text = processor.decode(output_tokens[0].tolist()[0], skip_special_tokens=True)
+```
+
+### Tips
+
+
+#### 1. Use dedicated models
+
+[`SeamlessM4Tv2Model`] is transformers top level model to generate speech and text, but you can also use dedicated models that perform the task without additional components, thus reducing the memory footprint.
+For example, you can replace the audio-to-audio generation snippet with the model dedicated to the S2ST task, the rest is exactly the same code: 
+
+```python
+>>> from transformers import SeamlessM4Tv2ForSpeechToSpeech
+>>> model = SeamlessM4Tv2ForSpeechToSpeech.from_pretrained("facebook/seamless-m4t-v2-large")
+```
+
+Or you can replace the text-to-text generation snippet with the model dedicated to the T2TT task, you only have to remove `generate_speech=False`.
+
+```python
+>>> from transformers import SeamlessM4Tv2ForTextToText
+>>> model = SeamlessM4Tv2ForTextToText.from_pretrained("facebook/seamless-m4t-v2-large")
+```
+
+Feel free to try out [`SeamlessM4Tv2ForSpeechToText`] and [`SeamlessM4Tv2ForTextToSpeech`] as well.
+
+#### 2. Change the speaker identity
+
+You have the possibility to change the speaker used for speech synthesis with the `speaker_id` argument. Some `speaker_id` works better than other for some languages!
+
+#### 3. Change the generation strategy
+
+You can use different [generation strategies](../generation_strategies) for text generation, e.g `.generate(input_ids=input_ids, text_num_beams=4, text_do_sample=True)` which will perform multinomial beam-search decoding on the text model. Note that speech generation only supports greedy - by default - or multinomial sampling, which can be used with e.g. `.generate(..., speech_do_sample=True, speech_temperature=0.6)`.
+
+#### 4. Generate speech and text at the same time
+
+Use `return_intermediate_token_ids=True` with [`SeamlessM4Tv2Model`] to return both speech and text !
+
+## Model architecture
+
+SeamlessM4T-v2 features a versatile architecture that smoothly handles the sequential generation of text and speech. This setup comprises two sequence-to-sequence (seq2seq) models. The first model translates the input modality into translated text, while the second model generates speech tokens, known as "unit tokens," from the translated text.
+
+Each modality has its own dedicated encoder with a unique architecture. Additionally, for speech output, a vocoder inspired by the [HiFi-GAN](https://arxiv.org/abs/2010.05646) architecture is placed on top of the second seq2seq model.
+
+### Difference with SeamlessM4T-v1
+
+The architecture of this new version differs from the first in a few aspects:
+
+#### Improvements on the second-pass model
+
+The second seq2seq model, named text-to-unit model, is now non-auto regressive, meaning that it computes units in a **single forward pass**. This achievement is made possible by:
+- the use of **character-level embeddings**, meaning that each character of the predicted translated text has its own embeddings, which are then used to predict the unit tokens.
+- the use of an intermediate duration predictor, that predicts speech duration at the **character-level** on the predicted translated text.
+- the use of a new text-to-unit decoder mixing convolutions and self-attention to handle longer context.
+
+#### Difference in the speech encoder
+
+The speech encoder, which is used during the first-pass generation process to predict the translated text, differs mainly from the previous speech encoder through these mechanisms:
+- the use of chunked attention mask to prevent attention across chunks, ensuring that each position attends only to positions within its own chunk and a fixed number of previous chunks.
+- the use of relative position embeddings which only considers distance between sequence elements rather than absolute positions. Please refer to [Self-Attentionwith Relative Position Representations (Shaw et al.)](https://arxiv.org/abs/1803.02155) for more details.
+- the use of a causal depth-wise convolution instead of a non-causal one.
+
+### Generation process
+
+Here's how the generation process works:
+
+- Input text or speech is processed through its specific encoder.
+- A decoder creates text tokens in the desired language.
+- If speech generation is required, the second seq2seq model, generates unit tokens in an non auto-regressive way.
+- These unit tokens are then passed through the final vocoder to produce the actual speech.
+
+
+This model was contributed by [ylacombe](https://huggingface.co/ylacombe). The original code can be found [here](https://github.com/facebookresearch/seamless_communication).
+
+## SeamlessM4Tv2Model
+
+[[autodoc]] SeamlessM4Tv2Model
+    - generate
+
+
+## SeamlessM4Tv2ForTextToSpeech
+
+[[autodoc]] SeamlessM4Tv2ForTextToSpeech
+    - generate
+
+
+## SeamlessM4Tv2ForSpeechToSpeech
+
+[[autodoc]] SeamlessM4Tv2ForSpeechToSpeech
+    - generate
+
+
+## SeamlessM4Tv2ForTextToText
+
+[[autodoc]] transformers.SeamlessM4Tv2ForTextToText
+    - forward
+    - generate
+
+## SeamlessM4Tv2ForSpeechToText
+
+[[autodoc]] transformers.SeamlessM4Tv2ForSpeechToText
+    - forward
+    - generate
+
+## SeamlessM4Tv2Config
+
+[[autodoc]] SeamlessM4Tv2Config

docs/source/en/model_doc/t5.md

@@ -314,7 +314,7 @@ The predicted tokens will then be placed between the sentinel tokens.
 
 ## Performance
 
-If you'd like a faster training and inference performance, install [apex](https://github.com/NVIDIA/apex#quick-start) and then the model will automatically use `apex.normalization.FusedRMSNorm` instead of `T5LayerNorm`. The former uses an optimized fused kernel which is several times faster than the latter.
+If you'd like a faster training and inference performance, install [NVIDIA APEX](https://github.com/NVIDIA/apex#quick-start) for NVIDIA GPUs, or [ROCm APEX](https://github.com/ROCmSoftwarePlatform/apex) for AMD GPUs and then the model will automatically use `apex.normalization.FusedRMSNorm` instead of `T5LayerNorm`. The former uses an optimized fused kernel which is several times faster than the latter.
 
 
 ## Resources

docs/source/en/model_doc/transfo-xl.md

@@ -16,6 +16,35 @@ rendered properly in your Markdown viewer.
 
 # Transformer XL
 
+<Tip warning={true}>
+
+This model is in maintenance mode only, so we won't accept any new PRs changing its code. This model was deprecated due to security issues linked to `pickle.load`.
+
+We recommend switching to more recent models for improved security.
+
+In case you would still like to use `TransfoXL` in your experiments, we recommend using the [Hub checkpoint](https://huggingface.co/transfo-xl-wt103) with a specific revision to ensure you are downloading safe files from the Hub.
+
+You will need to set the environment variable `TRUST_REMOTE_CODE` to `True` in order to allow the
+usage of `pickle.load()`:
+
+```python
+import os
+from transformers import TransfoXLTokenizer, TransfoXLLMHeadModel
+
+os.environ["TRUST_REMOTE_CODE"] = "True"
+
+checkpoint = 'transfo-xl-wt103'
+revision = '40a186da79458c9f9de846edfaea79c412137f97'
+
+tokenizer = TransfoXLTokenizer.from_pretrained(checkpoint, revision=revision)
+model = TransfoXLLMHeadModel.from_pretrained(checkpoint, revision=revision)
+```
+
+If you run into any issues running this model, please reinstall the last version that supported this model: v4.35.0.
+You can do so by running the following command: `pip install -U transformers==4.35.0`.
+
+</Tip>
+
 <div class="flex flex-wrap space-x-1">
 <a href="https://huggingface.co/models?filter=transfo-xl">
 <img alt="Models" src="https://img.shields.io/badge/All_model_pages-transfo--xl-blueviolet">
@@ -79,13 +108,13 @@ TransformerXL does **not** work with *torch.nn.DataParallel* due to a bug in PyT
 
 ## TransfoXL specific outputs
 
-[[autodoc]] models.transfo_xl.modeling_transfo_xl.TransfoXLModelOutput
+[[autodoc]] models.deprecated.transfo_xl.modeling_transfo_xl.TransfoXLModelOutput
 
-[[autodoc]] models.transfo_xl.modeling_transfo_xl.TransfoXLLMHeadModelOutput
+[[autodoc]] models.deprecated.transfo_xl.modeling_transfo_xl.TransfoXLLMHeadModelOutput
 
-[[autodoc]] models.transfo_xl.modeling_tf_transfo_xl.TFTransfoXLModelOutput
+[[autodoc]] models.deprecated.transfo_xl.modeling_tf_transfo_xl.TFTransfoXLModelOutput
 
-[[autodoc]] models.transfo_xl.modeling_tf_transfo_xl.TFTransfoXLLMHeadModelOutput
+[[autodoc]] models.deprecated.transfo_xl.modeling_tf_transfo_xl.TFTransfoXLLMHeadModelOutput
 
 <frameworkcontent>
 <pt>

docs/source/en/model_doc/tvp.md

@@ -0,0 +1,186 @@
+<!--Copyright 2023 The Intel Team Authors and 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.
+-->
+
+# TVP
+
+## Overview
+
+The text-visual prompting (TVP) framework was proposed in the paper [Text-Visual Prompting for Efficient 2D Temporal Video Grounding](https://arxiv.org/abs/2303.04995) by Yimeng Zhang, Xin Chen, Jinghan Jia, Sijia Liu, Ke Ding.
+
+The abstract from the paper is the following:
+
+*In this paper, we study the problem of temporal video grounding (TVG), which aims to predict the starting/ending time points of moments described by a text sentence within a long untrimmed video. Benefiting from fine-grained 3D visual features, the TVG techniques have achieved remarkable progress in recent years. However, the high complexity of 3D convolutional neural networks (CNNs) makes extracting dense 3D visual features time-consuming, which calls for intensive memory and computing resources. Towards efficient TVG, we propose a novel text-visual prompting (TVP) framework, which incorporates optimized perturbation patterns (that we call ‘prompts’) into both visual inputs and textual features of a TVG model. In sharp contrast to 3D CNNs, we show that TVP allows us to effectively co-train vision encoder and language encoder in a 2D TVG model and improves the performance of cross-modal feature fusion using only low-complexity sparse 2D visual features. Further, we propose a Temporal-Distance IoU (TDIoU) loss for efficient learning of TVG. Experiments on two benchmark datasets, Charades-STA and ActivityNet Captions datasets, empirically show that the proposed TVP significantly boosts the performance of 2D TVG (e.g., 9.79% improvement on Charades-STA and 30.77% improvement on ActivityNet Captions) and achieves 5× inference acceleration over TVG using 3D visual features.*
+
+This research addresses temporal video grounding (TVG), which is the process of pinpointing the start and end times of specific events in a long video, as described by a text sentence. Text-visual prompting (TVP), is proposed to enhance TVG. TVP involves integrating specially designed patterns, known as 'prompts', into both the visual (image-based) and textual (word-based) input components of a TVG model. These prompts provide additional spatial-temporal context, improving the model's ability to accurately determine event timings in the video. The approach employs 2D visual inputs in place of 3D ones. Although 3D inputs offer more spatial-temporal detail, they are also more time-consuming to process. The use of 2D inputs with the prompting method aims to provide similar levels of context and accuracy more efficiently.
+
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/tvp_architecture.png"
+alt="drawing" width="600"/>
+
+<small> TVP architecture. Taken from the <a href="https://arxiv.org/abs/2303.04995">original paper.</a> </small>
+
+This model was contributed by [Jiqing Feng](https://huggingface.co/Jiqing). The original code can be found [here](https://github.com/intel/TVP).
+
+## Usage tips and examples
+
+Prompts are optimized perturbation patterns, which would be added to input video frames or text features. Universal set refers to using the same exact set of prompts for any input, this means that these prompts are added consistently to all video frames and text features, regardless of the input's content.
+
+TVP consists of a visual encoder and cross-modal encoder. A universal set of visual prompts and text prompts to be integrated into sampled video frames and textual features, respectively. Specially, a set of different visual prompts are applied to uniformly-sampled frames of one untrimmed video in order.
+
+The goal of this model is to incorporate trainable prompts into both visual inputs and textual features to temporal video grounding(TVG) problems.
+In principle, one can apply any visual, cross-modal encoder in the proposed architecture.
+
+The [`TvpProcessor`] wraps [`BertTokenizer`] and [`TvpImageProcessor`] into a single instance to both
+encode the text and prepare the images respectively.
+
+The following example shows how to run temporal video grounding using [`TvpProcessor`] and [`TvpForVideoGrounding`].
+```python
+import av
+import cv2
+import numpy as np
+import torch
+from huggingface_hub import hf_hub_download
+from transformers import AutoProcessor, TvpForVideoGrounding
+
+
+def pyav_decode(container, sampling_rate, num_frames, clip_idx, num_clips, target_fps):
+    '''
+    Convert the video from its original fps to the target_fps and decode the video with PyAV decoder.
+    Args:
+        container (container): pyav container.
+        sampling_rate (int): frame sampling rate (interval between two sampled frames).
+        num_frames (int): number of frames to sample.
+        clip_idx (int): if clip_idx is -1, perform random temporal sampling.
+            If clip_idx is larger than -1, uniformly split the video to num_clips
+            clips, and select the clip_idx-th video clip.
+        num_clips (int): overall number of clips to uniformly sample from the given video.
+        target_fps (int): the input video may have different fps, convert it to
+            the target video fps before frame sampling.
+    Returns:
+        frames (tensor): decoded frames from the video. Return None if the no
+            video stream was found.
+        fps (float): the number of frames per second of the video.
+    '''
+    video = container.streams.video[0]
+    fps = float(video.average_rate)
+    clip_size = sampling_rate * num_frames / target_fps * fps
+    delta = max(num_frames - clip_size, 0)
+    start_idx = delta * clip_idx / num_clips
+    end_idx = start_idx + clip_size - 1
+    timebase = video.duration / num_frames
+    video_start_pts = int(start_idx * timebase)
+    video_end_pts = int(end_idx * timebase)
+    seek_offset = max(video_start_pts - 1024, 0)
+    container.seek(seek_offset, any_frame=False, backward=True, stream=video)
+    frames = {}
+    for frame in container.decode(video=0):
+        if frame.pts < video_start_pts:
+            continue
+        frames[frame.pts] = frame
+        if frame.pts > video_end_pts:
+            break
+    frames = [frames[pts] for pts in sorted(frames)]
+    return frames, fps
+
+
+def decode(container, sampling_rate, num_frames, clip_idx, num_clips, target_fps):
+    '''
+    Decode the video and perform temporal sampling.
+    Args:
+        container (container): pyav container.
+        sampling_rate (int): frame sampling rate (interval between two sampled frames).
+        num_frames (int): number of frames to sample.
+        clip_idx (int): if clip_idx is -1, perform random temporal sampling.
+            If clip_idx is larger than -1, uniformly split the video to num_clips
+            clips, and select the clip_idx-th video clip.
+        num_clips (int): overall number of clips to uniformly sample from the given video.
+        target_fps (int): the input video may have different fps, convert it to
+            the target video fps before frame sampling.
+    Returns:
+        frames (tensor): decoded frames from the video.
+    '''
+    assert clip_idx >= -2, "Not a valied clip_idx {}".format(clip_idx)
+    frames, fps = pyav_decode(container, sampling_rate, num_frames, clip_idx, num_clips, target_fps)
+    clip_size = sampling_rate * num_frames / target_fps * fps
+    index = np.linspace(0, clip_size - 1, num_frames)
+    index = np.clip(index, 0, len(frames) - 1).astype(np.int64)
+    frames = np.array([frames[idx].to_rgb().to_ndarray() for idx in index])
+    frames = frames.transpose(0, 3, 1, 2)
+    return frames
+
+
+file = hf_hub_download(repo_id="Intel/tvp_demo", filename="AK2KG.mp4", repo_type="dataset")
+model = TvpForVideoGrounding.from_pretrained("Intel/tvp-base")
+
+decoder_kwargs = dict(
+    container=av.open(file, metadata_errors="ignore"),
+    sampling_rate=1,
+    num_frames=model.config.num_frames,
+    clip_idx=0,
+    num_clips=1,
+    target_fps=3,
+)
+raw_sampled_frms = decode(**decoder_kwargs)
+
+text = "a person is sitting on a bed."
+processor = AutoProcessor.from_pretrained("Intel/tvp-base")
+model_inputs = processor(
+    text=[text], videos=list(raw_sampled_frms), return_tensors="pt", max_text_length=100#, size=size
+)
+
+model_inputs["pixel_values"] = model_inputs["pixel_values"].to(model.dtype)
+output = model(**model_inputs)
+
+def get_video_duration(filename):
+    cap = cv2.VideoCapture(filename)
+    if cap.isOpened():
+        rate = cap.get(5)
+        frame_num = cap.get(7)
+        duration = frame_num/rate
+        return duration
+    return -1
+
+duration = get_video_duration(file)
+start, end = processor.post_process_video_grounding(output.logits, duration)
+
+print(f"The time slot of the video corresponding to the text \"{text}\" is from {start}s to {end}s")
+```
+
+Tips:
+
+- This implementation of TVP uses [`BertTokenizer`] to generate text embeddings and Resnet-50 model to compute visual embeddings.
+- Checkpoints for pre-trained [tvp-base](https://huggingface.co/Intel/tvp-base) is released.
+- Please refer to [Table 2](https://arxiv.org/pdf/2303.04995.pdf) for TVP's performance on Temporal Video Grounding task.
+
+
+## TvpConfig
+
+[[autodoc]] TvpConfig
+
+## TvpImageProcessor
+
+[[autodoc]] TvpImageProcessor
+    - preprocess
+
+## TvpProcessor
+
+[[autodoc]] TvpProcessor
+    - __call__
+
+## TvpModel
+
+[[autodoc]] TvpModel
+    - forward
+
+## TvpForVideoGrounding
+
+[[autodoc]] TvpForVideoGrounding
+    - forward

docs/source/en/model_doc/univnet.md

@@ -0,0 +1,80 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ 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.
+
+-->
+
+# UnivNet
+
+## Overview
+
+The UnivNet model was proposed in [UnivNet: A Neural Vocoder with Multi-Resolution Spectrogram Discriminators for High-Fidelity Waveform Generation](https://arxiv.org/abs/2106.07889) by Won Jang, Dan Lim, Jaesam Yoon, Bongwan Kin, and Juntae Kim.
+The UnivNet model is a generative adversarial network (GAN) trained to synthesize high fidelity speech waveforms. The UnivNet model shared in `transformers` is the *generator*, which maps a conditioning log-mel spectrogram and optional noise sequence to a speech waveform (e.g. a vocoder). Only the generator is required for inference. The *discriminator* used to train the `generator` is not implemented.
+
+The abstract from the paper is the following:
+
+*Most neural vocoders employ band-limited mel-spectrograms to generate waveforms. If full-band spectral features are used as the input, the vocoder can be provided with as much acoustic information as possible. However, in some models employing full-band mel-spectrograms, an over-smoothing problem occurs as part of which non-sharp spectrograms are generated. To address this problem, we propose UnivNet, a neural vocoder that synthesizes high-fidelity waveforms in real time. Inspired by works in the field of voice activity detection, we added a multi-resolution spectrogram discriminator that employs multiple linear spectrogram magnitudes computed using various parameter sets. Using full-band mel-spectrograms as input, we expect to generate high-resolution signals by adding a discriminator that employs spectrograms of multiple resolutions as the input. In an evaluation on a dataset containing information on hundreds of speakers, UnivNet obtained the best objective and subjective results among competing models for both seen and unseen speakers. These results, including the best subjective score for text-to-speech, demonstrate the potential for fast adaptation to new speakers without a need for training from scratch.*
+
+Tips:
+
+- The `noise_sequence` argument for [`UnivNetModel.forward`] should be standard Gaussian noise (such as from `torch.randn`) of shape `([batch_size], noise_length, model.config.model_in_channels)`, where `noise_length` should match the length dimension (dimension 1) of the `input_features` argument. If not supplied, it will be randomly generated; a `torch.Generator` can be supplied to the `generator` argument so that the forward pass can be reproduced. (Note that [`UnivNetFeatureExtractor`] will return generated noise by default, so it shouldn't be necessary to generate `noise_sequence` manually.)
+- Padding added by [`UnivNetFeatureExtractor`] can be removed from the [`UnivNetModel`] output through the [`UnivNetFeatureExtractor.batch_decode`] method, as shown in the usage example below.
+- Padding the end of each waveform with silence can reduce artifacts at the end of the generated audio sample. This can be done by supplying `pad_end = True` to [`UnivNetFeatureExtractor.__call__`]. See [this issue](https://github.com/seungwonpark/melgan/issues/8) for more details.
+
+Usage Example:
+
+```python
+import torch
+from scipy.io.wavfile import write
+from datasets import Audio, load_dataset
+
+from transformers import UnivNetFeatureExtractor, UnivNetModel
+
+model_id_or_path = "dg845/univnet-dev"
+model = UnivNetModel.from_pretrained(model_id_or_path)
+feature_extractor = UnivNetFeatureExtractor.from_pretrained(model_id_or_path)
+
+ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
+# Resample the audio to the model and feature extractor's sampling rate.
+ds = ds.cast_column("audio", Audio(sampling_rate=feature_extractor.sampling_rate))
+# Pad the end of the converted waveforms to reduce artifacts at the end of the output audio samples.
+inputs = feature_extractor(
+    ds[0]["audio"]["array"], sampling_rate=ds[0]["audio"]["sampling_rate"], pad_end=True, return_tensors="pt"
+)
+
+with torch.no_grad():
+    audio = model(**inputs)
+
+# Remove the extra padding at the end of the output.
+audio = feature_extractor.batch_decode(**audio)[0]
+# Convert to wav file
+write("sample_audio.wav", feature_extractor.sampling_rate, audio)
+```
+
+This model was contributed by [dg845](https://huggingface.co/dg845).
+To the best of my knowledge, there is no official code release, but an unofficial implementation can be found at [maum-ai/univnet](https://github.com/maum-ai/univnet) with pretrained checkpoints [here](https://github.com/maum-ai/univnet#pre-trained-model).
+
+
+## UnivNetConfig
+
+[[autodoc]] UnivNetConfig
+
+## UnivNetFeatureExtractor
+
+[[autodoc]] UnivNetFeatureExtractor
+    - __call__
+
+## UnivNetModel
+
+[[autodoc]] UnivNetModel
+    - forward

docs/source/en/model_doc/whisper.md

@@ -34,13 +34,13 @@ The original code can be found [here](https://github.com/openai/whisper).
 - Inference is currently only implemented for short-form i.e. audio is pre-segmented into <=30s segments. Long-form (including timestamps) will be implemented in a future release.
 - One can use [`WhisperProcessor`] to prepare audio for the model, and decode the predicted ID's back into text.
 
-- To convert the tokenizer, we recommend using the following:
+- To convert the model and the processor, we recommend using the following:
 
 ```bash
-python src/transformers/models/whisper/convert_openai_to_hf.py --checkpoint_path "" --pytorch_dump_folder_path "Arthur/whisper-3" --convert_tokenizer True --whisper_version 3 --multilingual True
+python src/transformers/models/whisper/convert_openai_to_hf.py --checkpoint_path "" --pytorch_dump_folder_path "Arthur/whisper-3" --convert_preprocessor True
 ```
-Here the `whisper_version` will set the number of languages to `100` to account for `cantonese` which was added in `whisper-large-v3`.
-
+The script will automatically determine all necessary parameters from the OpenAI checkpoint. A `tiktoken` library needs to be installed
+to perform the conversion of the OpenAI tokenizer to the `tokenizers` version.
 
 ## Inference
 
@@ -75,6 +75,19 @@ Here is a step-by-step guide to transcribing an audio sample using a pre-trained
 ' Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.'
 ```
 
+## Resources
+
+A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with Whisper. 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 fork with a script to [convert a Whisper model in Hugging Face format to OpenAI format](https://github.com/zuazo-forks/transformers/blob/convert_hf_to_openai/src/transformers/models/whisper/convert_hf_to_openai.py). 🌎
+Usage example:
+```bash
+pip install -U openai-whisper
+python convert_hf_to_openai.py \
+    --checkpoint openai/whisper-tiny \
+    --whisper_dump_path whisper-tiny-openai.pt
+```
+
 ## WhisperConfig
 
 [[autodoc]] WhisperConfig

docs/source/en/pad_truncation.md

@@ -54,7 +54,7 @@ The following table summarizes the recommended way to setup padding and truncati
 |                                      |                                   | `tokenizer(batch_sentences, padding='longest')`                                        |
 |                                      | padding to max model input length | `tokenizer(batch_sentences, padding='max_length')`                                     |
 |                                      | padding to specific length        | `tokenizer(batch_sentences, padding='max_length', max_length=42)`                      |
-|                                      | padding to a multiple of a value  | `tokenizer(batch_sentences, padding=True, pad_to_multiple_of=8)                        |
+|                                      | padding to a multiple of a value  | `tokenizer(batch_sentences, padding=True, pad_to_multiple_of=8)`                        |
 | truncation to max model input length | no padding                        | `tokenizer(batch_sentences, truncation=True)` or                                       |
 |                                      |                                   | `tokenizer(batch_sentences, truncation=STRATEGY)`                                      |
 |                                      | padding to max sequence in batch  | `tokenizer(batch_sentences, padding=True, truncation=True)` or                         |

docs/source/en/perf_hardware.md

@@ -134,7 +134,7 @@ Here is the full benchmark code and outputs:
 ```bash
 # DDP w/ NVLink
 
-rm -r /tmp/test-clm; CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch \
+rm -r /tmp/test-clm; CUDA_VISIBLE_DEVICES=0,1 torchrun \
 --nproc_per_node 2 examples/pytorch/language-modeling/run_clm.py --model_name_or_path gpt2 \
 --dataset_name wikitext --dataset_config_name wikitext-2-raw-v1 --do_train \
 --output_dir /tmp/test-clm --per_device_train_batch_size 4 --max_steps 200
@@ -143,7 +143,7 @@ rm -r /tmp/test-clm; CUDA_VISIBLE_DEVICES=0,1 python -m torch.distributed.launch
 
 # DDP w/o NVLink
 
-rm -r /tmp/test-clm; CUDA_VISIBLE_DEVICES=0,1 NCCL_P2P_DISABLE=1 python -m torch.distributed.launch \
+rm -r /tmp/test-clm; CUDA_VISIBLE_DEVICES=0,1 NCCL_P2P_DISABLE=1 torchrun \
 --nproc_per_node 2 examples/pytorch/language-modeling/run_clm.py --model_name_or_path gpt2 \
 --dataset_name wikitext --dataset_config_name wikitext-2-raw-v1 --do_train
 --output_dir /tmp/test-clm --per_device_train_batch_size 4 --max_steps 200

docs/source/en/perf_infer_gpu_one.md

@@ -15,7 +15,7 @@ rendered properly in your Markdown viewer.
 
 # GPU inference
 
-GPUs are the standard choice of hardware for machine learning, unlike CPUs, because they are optimized for memory bandwidth and parallelism. To keep up with the larger sizes of modern models or to run these large models on existing and older hardware, there are several optimizations you can use to speed up GPU inference. In this guide, you'll learn how to use FlashAttention-2 (a more memory-efficient attention mechanism), BetterTransformer (a PyTorch native fastpath execution), and bitsandbytes to quantize your model to a lower precision. Finally, learn how to use 🤗 Optimum to accelerate inference with ONNX Runtime on Nvidia GPUs.
+GPUs are the standard choice of hardware for machine learning, unlike CPUs, because they are optimized for memory bandwidth and parallelism. To keep up with the larger sizes of modern models or to run these large models on existing and older hardware, there are several optimizations you can use to speed up GPU inference. In this guide, you'll learn how to use FlashAttention-2 (a more memory-efficient attention mechanism), BetterTransformer (a PyTorch native fastpath execution), and bitsandbytes to quantize your model to a lower precision. Finally, learn how to use 🤗 Optimum to accelerate inference with ONNX Runtime on Nvidia and AMD GPUs.
 
 <Tip>
 
@@ -36,15 +36,30 @@ FlashAttention-2 is experimental and may change considerably in future versions.
 1. additionally parallelizing the attention computation over sequence length
 2. partitioning the work between GPU threads to reduce communication and shared memory reads/writes between them
 
-FlashAttention-2 supports inference with Llama, Mistral, Falcon and Bark models. You can request to add FlashAttention-2 support for another model by opening a GitHub Issue or Pull Request.
+FlashAttention-2 is currently supported for the following architectures:
+* [Bark](https://huggingface.co/docs/transformers/model_doc/bark#transformers.BarkModel)
+* [Bart](https://huggingface.co/docs/transformers/model_doc/bart#transformers.BartModel)
+* [DistilBert](https://huggingface.co/docs/transformers/model_doc/distilbert#transformers.DistilBertModel)
+* [GPTBigCode](https://huggingface.co/docs/transformers/model_doc/gpt_bigcode#transformers.GPTBigCodeModel)
+* [GPTNeo](https://huggingface.co/docs/transformers/model_doc/gpt_neo#transformers.GPTNeoModel)
+* [GPTNeoX](https://huggingface.co/docs/transformers/model_doc/gpt_neox#transformers.GPTNeoXModel)
+* [Falcon](https://huggingface.co/docs/transformers/model_doc/falcon#transformers.FalconModel)
+* [Llama](https://huggingface.co/docs/transformers/model_doc/llama#transformers.LlamaModel)
+* [Llava](https://huggingface.co/docs/transformers/model_doc/llava)
+* [MBart](https://huggingface.co/docs/transformers/model_doc/mbart#transformers.MBartModel)
+* [Mistral](https://huggingface.co/docs/transformers/model_doc/mistral#transformers.MistralModel)
+* [Mixtral](https://huggingface.co/docs/transformers/model_doc/mixtral#transformers.MixtralModel)
+* [OPT](https://huggingface.co/docs/transformers/model_doc/opt#transformers.OPTModel)
+* [Phi](https://huggingface.co/docs/transformers/model_doc/phi#transformers.PhiModel)
+* [Whisper](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperModel)
 
-Before you begin, make sure you have FlashAttention-2 installed (see the [installation](https://github.com/Dao-AILab/flash-attention?tab=readme-ov-file#installation-and-features) guide for more details about prerequisites):
+You can request to add FlashAttention-2 support for another model by opening a GitHub Issue or Pull Request.
 
-```bash
-pip install flash-attn --no-build-isolation
-```
+Before you begin, make sure you have FlashAttention-2 installed. For NVIDIA GPUs, the library is installable through pip: `pip install flash-attn --no-build-isolation`. We strongly suggest to refer to the [detailed installation instructions](https://github.com/Dao-AILab/flash-attention?tab=readme-ov-file#installation-and-features).
 
-To enable FlashAttention-2, add the `use_flash_attention_2` parameter to [`~AutoModelForCausalLM.from_pretrained`]:
+FlashAttention-2 is also supported on AMD GPUs, with the current support limited to **Instinct MI210 and Instinct MI250**. We strongly suggest to use the following [Dockerfile](https://github.com/huggingface/optimum-amd/tree/main/docker/transformers-pytorch-amd-gpu-flash/Dockerfile) to use FlashAttention-2 on AMD GPUs.
+
+To enable FlashAttention-2, pass the argument `attn_implementation="flash_attention_2"` to [`~AutoModelForCausalLM.from_pretrained`]:
 
 ```python
 import torch
@@ -56,13 +71,15 @@ tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(
     model_id, 
     torch_dtype=torch.bfloat16, 
-    use_flash_attention_2=True,
+    attn_implementation="flash_attention_2",
 )
 ```
 
 <Tip>
 
-FlashAttention-2 can only be used when the model's dtype is  `fp16` or `bf16`, and it only runs on Nvidia GPUs. Make sure to cast your model to the appropriate dtype and load them on a supported device before using FlashAttention-2.
+FlashAttention-2 can only be used when the model's dtype is `fp16` or `bf16`. Make sure to cast your model to the appropriate dtype and load them on a supported device before using FlashAttention-2.
+
+Note that `use_flash_attention_2=True` can also be used to enable Flash Attention 2, but is deprecated in favor of `attn_implementation="flash_attention_2"`.
   
 </Tip>
 
@@ -79,14 +96,14 @@ tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(
     model_id, 
     load_in_8bit=True,
-    use_flash_attention_2=True,
+    attn_implementation="flash_attention_2",
 )
 
 # load in 4bit
 model = AutoModelForCausalLM.from_pretrained(
     model_id, 
     load_in_4bit=True,
-    use_flash_attention_2=True,
+    attn_implementation="flash_attention_2",
 )
 ```
 
@@ -126,8 +143,58 @@ FlashAttention is more memory efficient, meaning you can train on much larger se
 <img src="https://huggingface.co/datasets/ybelkada/documentation-images/resolve/main/llama-2-large-seqlen-padding.png">
 </div>
 
+## FlashAttention and memory-efficient attention through PyTorch's scaled_dot_product_attention 
+
+PyTorch's [`torch.nn.functional.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html) (SDPA) can also call FlashAttention and memory-efficient attention kernels under the hood. SDPA support is currently being added natively in Transformers, and is used by default for `torch>=2.1.1` when an implementation is available.
+
+For now, Transformers supports inference and training through SDPA for the following architectures:
+* [Bart](https://huggingface.co/docs/transformers/model_doc/bart#transformers.BartModel)
+* [GPTBigCode](https://huggingface.co/docs/transformers/model_doc/gpt_bigcode#transformers.GPTBigCodeModel)
+* [Falcon](https://huggingface.co/docs/transformers/model_doc/falcon#transformers.FalconModel)
+* [Llama](https://huggingface.co/docs/transformers/model_doc/llama#transformers.LlamaModel)
+* [Idefics](https://huggingface.co/docs/transformers/model_doc/idefics#transformers.IdeficsModel)
+* [Whisper](https://huggingface.co/docs/transformers/model_doc/whisper#transformers.WhisperModel)
+
+Note that FlashAttention can only be used for models with the `fp16` or `bf16` torch type, so make sure to cast your model to the appropriate type before using it.
+
+By default, `torch.nn.functional.scaled_dot_product_attention` selects the most performant kernel available, but to check whether a backend is available in a given setting (hardware, problem size), you can use [`torch.backends.cuda.sdp_kernel`](https://pytorch.org/docs/master/backends.html#torch.backends.cuda.sdp_kernel) as a context manager:
+
+```diff
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained("facebook/opt-350m")
+model = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", torch_dtype=torch.float16).to("cuda")
+# convert the model to BetterTransformer
+model.to_bettertransformer()
+
+input_text = "Hello my dog is cute and"
+inputs = tokenizer(input_text, return_tensors="pt").to("cuda")
+
++ with torch.backends.cuda.sdp_kernel(enable_flash=True, enable_math=False, enable_mem_efficient=False):
+    outputs = model.generate(**inputs)
+
+print(tokenizer.decode(outputs[0], skip_special_tokens=True))
+```
+
+If you see a bug with the traceback below, try using nightly version of PyTorch which may have broader coverage for FlashAttention:
+
+```bash
+RuntimeError: No available kernel. Aborting execution.
+
+# install PyTorch nightly
+pip3 install -U --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu118
+```
+
 ## BetterTransformer
 
+<Tip warning={true}>
+
+Part of BetterTransformer features are being upstreamed in Transformers, with native `torch.nn.scaled_dot_product_attention` default support. BetterTransformer still has a wider coverage than the Transformers SDPA integration, but you can expect more and more architectures to support natively SDPA in Transformers.
+
+</Tip>
+
+
 <Tip>
 
 Check out our benchmarks with BetterTransformer and scaled dot product attention in the [Out of the box acceleration and memory savings of 🤗 decoder models with PyTorch 2.0](https://pytorch.org/blog/out-of-the-box-acceleration/) and learn more about the fastpath execution in the [BetterTransformer](https://medium.com/pytorch/bettertransformer-out-of-the-box-performance-for-huggingface-transformers-3fbe27d50ab2) blog post.
@@ -156,39 +223,6 @@ model = model.reverse_bettertransformer()
 model.save_pretrained("saved_model")
 ```
 
-### FlashAttention
-
-SDPA can also call FlashAttention kernels under the hood. FlashAttention can only be used for models using the `fp16` or `bf16` dtype, so make sure to cast your model to the appropriate dtype before using it.
-
-To enable FlashAttention or to check whether it is available in a given setting (hardware, problem size), use [`torch.backends.cuda.sdp_kernel`](https://pytorch.org/docs/master/backends.html#torch.backends.cuda.sdp_kernel) as a context manager:
-
-```diff
-import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-tokenizer = AutoTokenizer.from_pretrained("facebook/opt-350m")
-model = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", torch_dtype=torch.float16).to("cuda")
-# convert the model to BetterTransformer
-model.to_bettertransformer()
-
-input_text = "Hello my dog is cute and"
-inputs = tokenizer(input_text, return_tensors="pt").to("cuda")
-
-+ with torch.backends.cuda.sdp_kernel(enable_flash=True, enable_math=False, enable_mem_efficient=False):
-    outputs = model.generate(**inputs)
-
-print(tokenizer.decode(outputs[0], skip_special_tokens=True))
-```
-
-If you see a bug with the traceback below, try using nightly version of PyTorch which may have broader coverage for FlashAttention:
-
-```bash
-RuntimeError: No available kernel. Aborting execution.
-
-# install PyTorch nightly
-pip3 install -U --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu118
-```
-
 ## bitsandbytes
 
 bitsandbytes is a quantization library that includes support for 4-bit and 8-bit quantization. Quantization reduces your model size compared to its native full precision version, making it easier to fit large models onto GPUs with limited memory.
@@ -276,13 +310,13 @@ Feel free to try running a 11 billion parameter [T5 model](https://colab.researc
 
 <Tip>
 
-Learn more details about using ORT with 🤗 Optimum in the [Accelerated inference on NVIDIA GPUs](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/gpu#accelerated-inference-on-nvidia-gpus) guide. This section only provides a brief and simple example.
+Learn more details about using ORT with 🤗 Optimum in the [Accelerated inference on NVIDIA GPUs](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/gpu#accelerated-inference-on-nvidia-gpus) and [Accelerated inference on AMD GPUs](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/amdgpu#accelerated-inference-on-amd-gpus) guides. This section only provides a brief and simple example.
 
 </Tip>
 
-ONNX Runtime (ORT) is a model accelerator that supports accelerated inference on Nvidia GPUs. ORT uses optimization techniques like fusing common operations into a single node and constant folding to reduce the number of computations performed and speedup inference. ORT also places the most computationally intensive operations on the GPU and the rest on the CPU to intelligently distribute the workload between the two devices.
+ONNX Runtime (ORT) is a model accelerator that supports accelerated inference on Nvidia GPUs, and AMD GPUs that use [ROCm](https://www.amd.com/en/products/software/rocm.html) stack. ORT uses optimization techniques like fusing common operations into a single node and constant folding to reduce the number of computations performed and speedup inference. ORT also places the most computationally intensive operations on the GPU and the rest on the CPU to intelligently distribute the workload between the two devices.
 
-ORT is supported by 🤗 Optimum which can be used in 🤗 Transformers. You'll need to use an [`~optimum.onnxruntime.ORTModel`] for the task you're solving, and specify the `provider` parameter which can be set to either [`CUDAExecutionProvider`](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/gpu#cudaexecutionprovider) or [`TensorrtExecutionProvider`](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/gpu#tensorrtexecutionprovider). If you want to load a model that was not yet exported to ONNX, you can set `export=True` to convert your model on-the-fly to the ONNX format :
+ORT is supported by 🤗 Optimum which can be used in 🤗 Transformers. You'll need to use an [`~optimum.onnxruntime.ORTModel`] for the task you're solving, and specify the `provider` parameter which can be set to either [`CUDAExecutionProvider`](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/gpu#cudaexecutionprovider), [`ROCMExecutionProvider`](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/amdgpu) or [`TensorrtExecutionProvider`](https://huggingface.co/docs/optimum/onnxruntime/usage_guides/gpu#tensorrtexecutionprovider). If you want to load a model that was not yet exported to ONNX, you can set `export=True` to convert your model on-the-fly to the ONNX format:
 
 ```py
 from optimum.onnxruntime import ORTModelForSequenceClassification

docs/source/en/perf_train_cpu_many.md

@@ -15,7 +15,8 @@ rendered properly in your Markdown viewer.
 
 # Efficient Training on Multiple CPUs
 
-When training on a single CPU is too slow, we can use multiple CPUs. This guide focuses on PyTorch-based DDP enabling distributed CPU training efficiently.
+When training on a single CPU is too slow, we can use multiple CPUs. This guide focuses on PyTorch-based DDP enabling
+distributed CPU training efficiently on [bare metal](#usage-in-trainer) and [Kubernetes](#usage-with-kubernetes).
 
 ## Intel® oneCCL Bindings for PyTorch
 
@@ -25,7 +26,7 @@ Module `oneccl_bindings_for_pytorch` (`torch_ccl` before version 1.12)  implemen
 
 Check more detailed information for [oneccl_bind_pt](https://github.com/intel/torch-ccl).
 
-### Intel® oneCCL Bindings for PyTorch installation:
+### Intel® oneCCL Bindings for PyTorch installation
 
 Wheel files are available for the following Python versions:
 
@@ -68,9 +69,9 @@ torch_ccl_path=$(python -c "import torch; import torch_ccl; import os;  print(os
 source $torch_ccl_path/env/setvars.sh
 ```
 
-#### IPEX installation:
+#### Intel® Extension for PyTorch installation
 
-IPEX provides performance optimizations for CPU training with both Float32 and BFloat16, you could refer [single CPU section](./perf_train_cpu).
+Intel Extension for PyTorch (IPEX) provides performance optimizations for CPU training with both Float32 and BFloat16 (refer to the [single CPU section](./perf_train_cpu) to learn more).
 
 
 The following "Usage in Trainer" takes mpirun in Intel® MPI library as an example.
@@ -132,3 +133,185 @@ Now, run the following command in node0 and **4DDP** will be enabled in node0 an
  --use_ipex \
  --bf16
 ```
+
+## Usage with Kubernetes
+
+The same distributed training job from the previous section can be deployed to a Kubernetes cluster using the
+[Kubeflow PyTorchJob training operator](https://www.kubeflow.org/docs/components/training/pytorch/).
+
+### Setup
+
+This example assumes that you have:
+* Access to a Kubernetes cluster with [Kubeflow installed](https://www.kubeflow.org/docs/started/installing-kubeflow/)
+* [`kubectl`](https://kubernetes.io/docs/tasks/tools/) installed and configured to access the Kubernetes cluster
+* A [Persistent Volume Claim (PVC)](https://kubernetes.io/docs/concepts/storage/persistent-volumes/) that can be used
+  to store datasets and model files. There are multiple options for setting up the PVC including using an NFS
+  [storage class](https://kubernetes.io/docs/concepts/storage/storage-classes/) or a cloud storage bucket.
+* A Docker container that includes your model training script and all the dependencies needed to run the script. For
+  distributed CPU training jobs, this typically includes PyTorch, Transformers, Intel Extension for PyTorch, Intel
+  oneCCL Bindings for PyTorch, and OpenSSH to communicate between the containers.
+
+The snippet below is an example of a Dockerfile that uses a base image that supports distributed CPU training and then
+extracts a Transformers release to the `/workspace` directory, so that the example scripts are included in the image:
+```
+FROM intel/ai-workflows:torch-2.0.1-huggingface-multinode-py3.9
+
+WORKDIR /workspace
+
+# Download and extract the transformers code
+ARG HF_TRANSFORMERS_VER="4.35.2"
+RUN mkdir transformers && \
+    curl -sSL --retry 5 https://github.com/huggingface/transformers/archive/refs/tags/v${HF_TRANSFORMERS_VER}.tar.gz | tar -C transformers --strip-components=1 -xzf -
+```
+The image needs to be built and copied to the cluster's nodes or pushed to a container registry prior to deploying the
+PyTorchJob to the cluster.
+
+### PyTorchJob Specification File
+
+The [Kubeflow PyTorchJob](https://www.kubeflow.org/docs/components/training/pytorch/) is used to run the distributed
+training job on the cluster. The yaml file for the PyTorchJob defines parameters such as:
+ * The name of the PyTorchJob
+ * The number of replicas (workers)
+ * The python script and it's parameters that will be used to run the training job
+ * The types of resources (node selector, memory, and CPU) needed for each worker
+ * The image/tag for the Docker container to use
+ * Environment variables
+ * A volume mount for the PVC
+
+The volume mount defines a path where the PVC will be mounted in the container for each worker pod. This location can be
+used for the dataset, checkpoint files, and the saved model after training completes.
+
+The snippet below is an example of a yaml file for a PyTorchJob with 4 workers running the
+[question-answering example](https://github.com/huggingface/transformers/tree/main/examples/pytorch/question-answering).
+```yaml
+apiVersion: "kubeflow.org/v1"
+kind: PyTorchJob
+metadata:
+  name: transformers-pytorchjob
+  namespace: kubeflow
+spec:
+  elasticPolicy:
+    rdzvBackend: c10d
+    minReplicas: 1
+    maxReplicas: 4
+    maxRestarts: 10
+  pytorchReplicaSpecs:
+    Worker:
+      replicas: 4  # The number of worker pods
+      restartPolicy: OnFailure
+      template:
+        spec:
+          containers:
+            - name: pytorch
+              image: <image name>:<tag>  # Specify the docker image to use for the worker pods
+              imagePullPolicy: IfNotPresent
+              command:
+                - torchrun
+                - /workspace/transformers/examples/pytorch/question-answering/run_qa.py
+                - --model_name_or_path
+                - "bert-large-uncased"
+                - --dataset_name
+                - "squad"
+                - --do_train
+                - --do_eval
+                - --per_device_train_batch_size
+                - "12"
+                - --learning_rate
+                - "3e-5"
+                - --num_train_epochs
+                - "2"
+                - --max_seq_length
+                - "384"
+                - --doc_stride
+                - "128"
+                - --output_dir
+                - "/tmp/pvc-mount/output"
+                - --no_cuda
+                - --ddp_backend
+                - "ccl"
+                - --use_ipex
+                - --bf16  # Specify --bf16 if your hardware supports bfloat16
+              env:
+              - name: LD_PRELOAD
+                value: "/usr/lib/x86_64-linux-gnu/libtcmalloc.so.4.5.9:/usr/local/lib/libiomp5.so"
+              - name: TRANSFORMERS_CACHE
+                value: "/tmp/pvc-mount/transformers_cache"
+              - name: HF_DATASETS_CACHE
+                value: "/tmp/pvc-mount/hf_datasets_cache"
+              - name: LOGLEVEL
+                value: "INFO"
+              - name: CCL_WORKER_COUNT
+                value: "1"
+              - name: OMP_NUM_THREADS  # Can be tuned for optimal performance
+-                value: "56"
+              resources:
+                limits:
+                  cpu: 200  # Update the CPU and memory limit values based on your nodes
+                  memory: 128Gi
+                requests:
+                  cpu: 200  # Update the CPU and memory request values based on your nodes
+                  memory: 128Gi
+              volumeMounts:
+              - name: pvc-volume
+                mountPath: /tmp/pvc-mount
+              - mountPath: /dev/shm
+                name: dshm
+          restartPolicy: Never
+          nodeSelector:  #  Optionally use the node selector to specify what types of nodes to use for the workers
+            node-type: spr
+          volumes:
+          - name: pvc-volume
+            persistentVolumeClaim:
+              claimName: transformers-pvc
+          - name: dshm
+            emptyDir:
+              medium: Memory
+```
+To run this example, update the yaml based on your training script and the nodes in your cluster.
+
+<Tip>
+
+The CPU resource limits/requests in the yaml are defined in [cpu units](https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/#meaning-of-cpu)
+where 1 CPU unit is equivalent to 1 physical CPU core or 1 virtual core (depending on whether the node is a physical
+host or a VM). The amount of CPU and memory limits/requests defined in the yaml should be less than the amount of
+available CPU/memory capacity on a single machine. It is usually a good idea to not use the entire machine's capacity in
+order to leave some resources for the kubelet and OS. In order to get ["guaranteed"](https://kubernetes.io/docs/concepts/workloads/pods/pod-qos/#guaranteed)
+[quality of service](https://kubernetes.io/docs/tasks/configure-pod-container/quality-service-pod/) for the worker pods,
+set the same CPU and memory amounts for both the resource limits and requests.
+
+</Tip>
+
+### Deploy
+
+After the PyTorchJob spec has been updated with values appropriate for your cluster and training job, it can be deployed
+to the cluster using:
+```
+kubectl create -f pytorchjob.yaml
+```
+
+The `kubectl get pods -n kubeflow` command can then be used to list the pods in the `kubeflow` namespace. You should see
+the worker pods for the PyTorchJob that was just deployed. At first, they will probably have a status of "Pending" as
+the containers get pulled and created, then the status should change to "Running".
+```
+NAME                                                     READY   STATUS                  RESTARTS          AGE
+...
+transformers-pytorchjob-worker-0                         1/1     Running                 0                 7m37s
+transformers-pytorchjob-worker-1                         1/1     Running                 0                 7m37s
+transformers-pytorchjob-worker-2                         1/1     Running                 0                 7m37s
+transformers-pytorchjob-worker-3                         1/1     Running                 0                 7m37s
+...
+```
+
+The logs for worker can be viewed using `kubectl logs -n kubeflow <pod name>`. Add `-f` to stream the logs, for example:
+```
+kubectl logs -n kubeflow transformers-pytorchjob-worker-0 -f
+```
+
+After the training job completes, the trained model can be copied from the PVC or storage location. When you are done
+with the job, the PyTorchJob resource can be deleted from the cluster using `kubectl delete -f pytorchjob.yaml`.
+
+## Summary
+
+This guide covered running distributed PyTorch training jobs using multiple CPUs on bare metal and on a Kubernetes
+cluster. Both cases utilize Intel Extension for PyTorch and Intel oneCCL Bindings for PyTorch for optimal training
+performance, and can be used as a template to run your own workload on multiple nodes.

docs/source/en/perf_train_gpu_many.md

@@ -153,7 +153,7 @@ python examples/pytorch/language-modeling/run_clm.py \
 
 ```
 rm -r /tmp/test-clm; CUDA_VISIBLE_DEVICES=0,1 \
-python -m torch.distributed.launch --nproc_per_node 2 examples/pytorch/language-modeling/run_clm.py \
+torchrun --nproc_per_node 2 examples/pytorch/language-modeling/run_clm.py \
 --model_name_or_path gpt2 --dataset_name wikitext --dataset_config_name wikitext-2-raw-v1 \
 --do_train --output_dir /tmp/test-clm --per_device_train_batch_size 4 --max_steps 200
 
@@ -164,7 +164,7 @@ python -m torch.distributed.launch --nproc_per_node 2 examples/pytorch/language-
 
 ```
 rm -r /tmp/test-clm; NCCL_P2P_DISABLE=1 CUDA_VISIBLE_DEVICES=0,1 \
-python -m torch.distributed.launch --nproc_per_node 2 examples/pytorch/language-modeling/run_clm.py \
+torchrun --nproc_per_node 2 examples/pytorch/language-modeling/run_clm.py \
 --model_name_or_path gpt2 --dataset_name wikitext --dataset_config_name wikitext-2-raw-v1 \
 --do_train --output_dir /tmp/test-clm --per_device_train_batch_size 4 --max_steps 200
 

docs/source/en/perf_train_gpu_one.md

@@ -237,7 +237,7 @@ You can speedup the training throughput by using Flash Attention 2 integration i
 The most common optimizer used to train transformer models is Adam or AdamW (Adam with weight decay). Adam achieves 
 good convergence by storing the rolling average of the previous gradients; however, it adds an additional memory 
 footprint of the order of the number of model parameters. To remedy this, you can use an alternative optimizer. 
-For example if you have [NVIDIA/apex](https://github.com/NVIDIA/apex) installed, `adamw_apex_fused` will give you the 
+For example if you have [NVIDIA/apex](https://github.com/NVIDIA/apex) installed for NVIDIA GPUs, or [ROCmSoftwarePlatform/apex](https://github.com/ROCmSoftwarePlatform/apex) for AMD GPUs, `adamw_apex_fused` will give you the
 fastest training experience among all supported AdamW optimizers.
 
 [`Trainer`] integrates a variety of optimizers that can be used out of box: `adamw_hf`, `adamw_torch`, `adamw_torch_fused`, 
@@ -529,4 +529,4 @@ By default, in training mode, the BetterTransformer integration **drops the mask
 
 </Tip>
 
-Check out this [blogpost](https://pytorch.org/blog/out-of-the-box-acceleration/) to learn more about acceleration and memory-savings with SDPA.
+Check out this [blogpost](https://pytorch.org/blog/out-of-the-box-acceleration/) to learn more about acceleration and memory-savings with SDPA.

docs/source/en/preprocessing.md

@@ -220,7 +220,7 @@ array([[1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0],
 
 For audio tasks, you'll need a [feature extractor](main_classes/feature_extractor) to prepare your dataset for the model. The feature extractor is designed to extract features from raw audio data, and convert them into tensors.
 
-Load the [MInDS-14](https://huggingface.co/datasets/PolyAI/minds14) dataset (see the 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub.html) for more details on how to load a dataset) to see how you can use a feature extractor with audio datasets:
+Load the [MInDS-14](https://huggingface.co/datasets/PolyAI/minds14) dataset (see the 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub) for more details on how to load a dataset) to see how you can use a feature extractor with audio datasets:
 
 ```py
 >>> from datasets import load_dataset, Audio
@@ -340,7 +340,7 @@ You can use any library you like for image augmentation. For image preprocessing
 
 </Tip>
 
-Load the [food101](https://huggingface.co/datasets/food101) dataset (see the 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub.html) for more details on how to load a dataset) to see how you can use an image processor with computer vision datasets:
+Load the [food101](https://huggingface.co/datasets/food101) dataset (see the 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub) for more details on how to load a dataset) to see how you can use an image processor with computer vision datasets:
 
 <Tip>
 
@@ -354,7 +354,7 @@ Use 🤗 Datasets `split` parameter to only load a small sample from the trainin
 >>> dataset = load_dataset("food101", split="train[:100]")
 ```
 
-Next, take a look at the image with 🤗 Datasets [`Image`](https://huggingface.co/docs/datasets/package_reference/main_classes.html?highlight=image#datasets.Image) feature:
+Next, take a look at the image with 🤗 Datasets [`Image`](https://huggingface.co/docs/datasets/package_reference/main_classes?highlight=image#datasets.Image) feature:
 
 ```py
 >>> dataset[0]["image"]
@@ -467,7 +467,7 @@ from [`DetrImageProcessor`] and define a custom `collate_fn` to batch images tog
 
 For tasks involving multimodal inputs, you'll need a [processor](main_classes/processors) to prepare your dataset for the model. A processor couples together two processing objects such as as tokenizer and feature extractor.
 
-Load the [LJ Speech](https://huggingface.co/datasets/lj_speech) dataset (see the 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub.html) for more details on how to load a dataset) to see how you can use a processor for automatic speech recognition (ASR):
+Load the [LJ Speech](https://huggingface.co/datasets/lj_speech) dataset (see the 🤗 [Datasets tutorial](https://huggingface.co/docs/datasets/load_hub) for more details on how to load a dataset) to see how you can use a processor for automatic speech recognition (ASR):
 
 ```py
 >>> from datasets import load_dataset

docs/source/en/quantization.md

@@ -0,0 +1,612 @@
+<!--Copyright 2023 The HuggingFace Team. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
+the License. You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
+an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
+specific language governing permissions and limitations under the License.
+
+⚠️ 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.
+
+-->
+
+# Quantization
+
+Quantization techniques focus on representing data with less information while also trying to not lose too much accuracy. This often means converting a data type to represent the same information with fewer bits. For example, if your model weights are stored as 32-bit floating points and they're quantized to 16-bit floating points, this halves the model size which makes it easier to store and reduces memory-usage. Lower precision can also speedup inference because it takes less time to perform calculations with fewer bits.
+
+Transformers supports several quantization schemes to help you run inference with large language models (LLMs) and finetune adapters on quantized models. This guide will show you how to use Activation-aware Weight Quantization (AWQ), AutoGPTQ, and bitsandbytes.
+
+## AWQ
+
+<Tip>
+
+Try AWQ quantization with this [notebook](https://colab.research.google.com/drive/1HzZH89yAXJaZgwJDhQj9LqSBux932BvY)!
+
+</Tip>
+
+[Activation-aware Weight Quantization (AWQ)](https://hf.co/papers/2306.00978) doesn't quantize all the weights in a model, and instead, it preserves a small percentage of weights that are important for LLM performance. This significantly reduces quantization loss such that you can run models in 4-bit precision without experiencing any performance degradation.
+
+There are several libraries for quantizing models with the AWQ algorithm, such as [llm-awq](https://github.com/mit-han-lab/llm-awq), [autoawq](https://github.com/casper-hansen/AutoAWQ) or [optimum-intel](https://huggingface.co/docs/optimum/main/en/intel/optimization_inc). Transformers supports loading models quantized with the llm-awq and autoawq libraries. This guide will show you how to load models quantized with autoawq, but the processs is similar for llm-awq quantized models.
+
+Make sure you have autoawq installed:
+
+```bash
+pip install autoawq
+```
+
+AWQ-quantized models can be identified by checking the `quantization_config` attribute in the model's [config.json](https://huggingface.co/TheBloke/zephyr-7B-alpha-AWQ/blob/main/config.json) file:
+
+```json
+{
+  "_name_or_path": "/workspace/process/huggingfaceh4_zephyr-7b-alpha/source",
+  "architectures": [
+    "MistralForCausalLM"
+  ],
+  ...
+  ...
+  ...
+  "quantization_config": {
+    "quant_method": "awq",
+    "zero_point": true,
+    "group_size": 128,
+    "bits": 4,
+    "version": "gemm"
+  }
+}
+```
+
+A quantized model is loaded with the [`~PreTrainedModel.from_pretrained`] method. If you loaded your model on the CPU, make sure to move it to a GPU device first. Use the `device_map` parameter to specify where to place the model:
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_id = "TheBloke/zephyr-7B-alpha-AWQ"
+model = AutoModelForCausalLM.from_pretrained(model_id, device_map="cuda:0")
+```
+
+Loading an AWQ-quantized model automatically sets other weights to fp16 by default for performance reasons. If you want to load these other weights in a different format, use the `torch_dtype` parameter:
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_id = "TheBloke/zephyr-7B-alpha-AWQ"
+model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.float32)
+```
+
+AWQ quantization can also be combined with [FlashAttention-2](perf_infer_gpu_one#flashattention-2) to further accelerate inference:
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model = AutoModelForCausalLM.from_pretrained("TheBloke/zephyr-7B-alpha-AWQ", attn_implementation="flash_attention_2", device_map="cuda:0")
+```
+
+
+### Benchmarks
+
+We performed some speed, throughput and latency benchmarks using [`optimum-benchmark`](https://github.com/huggingface/optimum-benchmark) library. 
+
+Note at that time of writing this documentation section, the available quantization methods were: `awq`, `gptq` and `bitsandbytes`.
+
+The benchmark was run on a NVIDIA-A100 instance and the model used was [`TheBloke/Mistral-7B-v0.1-AWQ`](https://huggingface.co/TheBloke/Mistral-7B-v0.1-AWQ) for the AWQ model, [`TheBloke/Mistral-7B-v0.1-GPTQ`](https://huggingface.co/TheBloke/Mistral-7B-v0.1-GPTQ) for the GPTQ model. We also benchmarked it against `bitsandbytes` quantization methods and native `float16` model. Some results are shown below:
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/forward_memory_plot.png">
+</div>
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/generate_memory_plot.png">
+</div>
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/generate_throughput_plot.png">
+</div>
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/forward_latency_plot.png">
+</div>
+
+You can find the full results together with packages versions in [this link](https://github.com/huggingface/optimum-benchmark/tree/main/examples/running-mistrals).
+
+From the results it appears that AWQ quantization method is the fastest quantization method for inference, text generation and among the lowest peak memory for text generation. However, AWQ seems to have the largest forward latency per batch size. 
+
+
+### Make use of fused modules
+
+You can benefit from fused modules by passing an `AwqConfig` with `fuse_modules=True` and your expected maximum sequence length for generation to `fuse_max_seq_len`. For architectures that do not support `do_fuse=True`, you can still fuse the modules, however you need to pass a custom `fusing_mapping` to `AwqConfig()`. Let's dive into these specific usecases.
+
+Note that you cannot combine fusing modules and other optimization techniques such as Flash Attention 2.
+
+#### Fusing modules for supported architectures
+
+Currently we support out of the box AWQ module fusing for `llama` and `mistral`. 
+
+To enable this feature for supported architectures simply create an `AwqConfig` and pass the arguments `fuse_max_seq_len` and `do_fuse=True`.
+
+For example to enable module fusing for the model `TheBloke/Mistral-7B-OpenOrca-AWQ`, run:
+
+```python
+import torch
+from transformers import AwqConfig, AutoModelForCausalLM
+
+model_id = "TheBloke/Mistral-7B-OpenOrca-AWQ"
+
+quantization_config = AwqConfig(
+    bits=4,
+    fuse_max_seq_len=512,
+    do_fuse=True,
+)
+
+model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=quantization_config).to(0)
+```
+
+Note that you need to define `fuse_max_seq_len` to `AwqConfig`. That total sequence length should include the context length and the expected generation length. You can set it to a large value to be on the safe zone.
+
+You can also apply module fusing for other architectures that are not supported.
+
+#### Fusing modules for unsupported architectures
+
+For architectures that do not support out of the box module fusing, you can pass a custom fusing mapping; simply pass a dictionnary `modules_to_fuse` to `AwqConfig`, let's take an example with the Yi model:
+
+
+```python
+import torch
+from transformers import AwqConfig, AutoModelForCausalLM
+
+model_id = "TheBloke/Yi-34B-AWQ"
+
+quantization_config = AwqConfig(
+    bits=4,
+    fuse_max_seq_len=512,
+    modules_to_fuse={
+        "attention": ["q_proj", "k_proj", "v_proj", "o_proj"],
+        "layernorm": ["ln1", "ln2", "norm"],
+        "mlp": ["gate_proj", "up_proj", "down_proj"],
+        "use_alibi": False,
+        "num_attention_heads": 56,
+        "num_key_value_heads": 8,
+        "hidden_size": 7168
+    }
+)
+
+model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=quantization_config).to(0)
+```
+
+The parameter `modules_to_fuse` needs to have the following respective fields: 
+
+- `"attention"`: The names of the attention layers to fuse - in the order: query, key, value and output projection layer. In case you don't want to fuse the attention layers you can pass an empty list.
+- `"layernorm"`: The names of all the layernorm layers you want to replace with a custom fused layer norm. In case you don't want to fuse these layers you can also pass an empty list.
+- `"mlp"`: The names of the MLP layers you want to fuse into a single MLP layer in the order: (gate (dense layer post-attention) / up / down layers).
+- `"use_alibi"`: If you model uses alibi positional embedding
+- `"num_attention_heads"`: The number of attention heads
+- `"num_key_value_heads"`: This is the number of key value heads that should be used to implement Grouped Query Attention. If num_key_value_heads=num_attention_heads, the model will use Multi Head Attention (MHA), if num_key_value_heads=1 the model will use Multi Query Attention (MQA) otherwise GQA is used. 
+- `"hidden_size"`: Dimension of the hidden representations.
+
+
+#### Benchmarks
+
+We benchmarked the model with and without fused modules first using only `batch_size=1` on the `TheBloke/Mistral-7B-OpenOrca-AWQ` model and below are the results:
+
+*unfused case*
+
+|   Batch Size |   Prefill Length |   Decode Length |   Prefill tokens/s |   Decode tokens/s | Memory (VRAM)   |
+|-------------:|-----------------:|----------------:|-------------------:|------------------:|:----------------|
+|            1 |               32 |              32 |            60.0984 |           38.4537 | 4.50 GB (5.68%) |
+|            1 |               64 |              64 |          1333.67   |           31.6604 | 4.50 GB (5.68%) |
+|            1 |              128 |             128 |          2434.06   |           31.6272 | 4.50 GB (5.68%) |
+|            1 |              256 |             256 |          3072.26   |           38.1731 | 4.50 GB (5.68%) |
+|            1 |              512 |             512 |          3184.74   |           31.6819 | 4.59 GB (5.80%) |
+|            1 |             1024 |            1024 |          3148.18   |           36.8031 | 4.81 GB (6.07%) |
+|            1 |             2048 |            2048 |          2927.33   |           35.2676 | 5.73 GB (7.23%) |
+
+*fused case*
+
+|   Batch Size |   Prefill Length |   Decode Length |   Prefill tokens/s |   Decode tokens/s | Memory (VRAM)   |
+|-------------:|-----------------:|----------------:|-------------------:|------------------:|:----------------|
+|            1 |               32 |              32 |            81.4899 |           80.2569 | 4.00 GB (5.05%) |
+|            1 |               64 |              64 |          1756.1    |          106.26   | 4.00 GB (5.05%) |
+|            1 |              128 |             128 |          2479.32   |          105.631  | 4.00 GB (5.06%) |
+|            1 |              256 |             256 |          1813.6    |           85.7485 | 4.01 GB (5.06%) |
+|            1 |              512 |             512 |          2848.9    |           97.701  | 4.11 GB (5.19%) |
+|            1 |             1024 |            1024 |          3044.35   |           87.7323 | 4.41 GB (5.57%) |
+|            1 |             2048 |            2048 |          2715.11   |           89.4709 | 5.57 GB (7.04%) |
+
+We also performed benchmarks with [`optimum-benchmark`](https://github.com/huggingface/optimum-benchmark) library. And below are the results:
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/fused_forward_memory_plot.png">
+</div>
+
+<div style="text-align: center">
+<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/fused_generate_throughput_plot.png">
+</div>
+
+
+## AutoGPTQ
+
+<Tip>
+
+Try GPTQ quantization with PEFT in this [notebook](https://colab.research.google.com/drive/1_TIrmuKOFhuRRiTWN94iLKUFu6ZX4ceb?usp=sharing) and learn more about it's details in this [blog post](https://huggingface.co/blog/gptq-integration)!
+
+</Tip>
+
+The [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ) library implements the GPTQ algorithm, a post-training quantization technique where each row of the weight matrix is quantized independently to find a version of the weights that minimizes the error. These weights are quantized to int4, but they're restored to fp16 on the fly during inference. This can save your memory-usage by 4x because the int4 weights are dequantized in a fused kernel rather than a GPU's global memory, and you can also expect a speedup in inference because using a lower bitwidth takes less time to communicate.
+
+Before you begin, make sure the following libraries are installed:
+
+```bash
+pip install auto-gptq
+pip install git+https://github.com/huggingface/optimum.git
+pip install git+https://github.com/huggingface/transformers.git
+pip install --upgrade accelerate
+```
+
+To quantize a model (currently only supported for text models), you need to create a [`GPTQConfig`] class and set the number of bits to quantize to, a dataset to calibrate the weights for quantization, and a tokenizer to prepare the dataset.
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer, GPTQConfig
+
+model_id = "facebook/opt-125m"
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+gptq_config = GPTQConfig(bits=4, dataset="c4", tokenizer=tokenizer)
+```
+
+You could also pass your own dataset as a list of strings, but it is highly recommended to use the same dataset from the GPTQ paper.
+
+```py
+dataset = ["auto-gptq is an easy-to-use model quantization library with user-friendly apis, based on GPTQ algorithm."]
+gptq_config = GPTQConfig(bits=4, dataset=dataset, tokenizer=tokenizer)
+```
+
+Load a model to quantize and pass the `gptq_config` to the [`~AutoModelForCausalLM.from_pretrained`] method. Set `device_map="auto"` to automatically offload the model to a CPU to help fit the model in memory, and allow the model modules to be moved between the CPU and GPU for quantization.
+
+```py
+quantized_model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", quantization_config=gptq_config)
+```
+
+If you're running out of memory because a dataset is too large, disk offloading is not supported. If this is the case, try passing the `max_memory` parameter to allocate the amount of memory to use on your device (GPU and CPU):
+
+```py
+quantized_model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", max_memory={0: "30GiB", 1: "46GiB", "cpu": "30GiB"}, quantization_config=gptq_config)
+```
+
+<Tip warning={true}>
+
+Depending on your hardware, it can take some time to quantize a model from scratch. It can take ~5 minutes to quantize the [faceboook/opt-350m]() model on a free-tier Google Colab GPU, but it'll take ~4 hours to quantize a 175B parameter model on a NVIDIA A100. Before you quantize a model, it is a good idea to check the Hub if a GPTQ-quantized version of the model already exists.
+
+</Tip>
+
+Once your model is quantized, you can push the model and tokenizer to the Hub where it can be easily shared and accessed. Use the [`~PreTrainedModel.push_to_hub`] method to save the [`GPTQConfig`]:
+
+```py
+quantized_model.push_to_hub("opt-125m-gptq")
+tokenizer.push_to_hub("opt-125m-gptq")
+```
+
+You could also save your quantized model locally with the [`~PreTrainedModel.save_pretrained`] method. If the model was quantized with the `device_map` parameter, make sure to move the entire model to a GPU or CPU before saving it. For example, to save the model on a CPU:
+
+```py
+quantized_model.save_pretrained("opt-125m-gptq")
+tokenizer.save_pretrained("opt-125m-gptq")
+
+# if quantized with device_map set
+quantized_model.to("cpu")
+quantized_model.save_pretrained("opt-125m-gptq")
+```
+
+Reload a quantized model with the [`~PreTrainedModel.from_pretrained`] method, and set `device_map="auto"` to automatically distribute the model on all available GPUs to load the model faster without using more memory than needed.
+
+```py
+from transformers import AutoModelForCausalLM
+
+model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq", device_map="auto")
+```
+
+### ExLlama
+
+[ExLlama](https://github.com/turboderp/exllama) is a Python/C++/CUDA implementation of the [Llama](model_doc/llama) model that is designed for faster inference with 4-bit GPTQ weights (check out these [benchmarks](https://github.com/huggingface/optimum/tree/main/tests/benchmark#gptq-benchmark)). The ExLlama kernel is activated by default when you create a [`GPTQConfig`] object. To boost inference speed even further, use the [ExLlamaV2](https://github.com/turboderp/exllamav2) kernels by configuring the `exllama_config` parameter:
+
+```py
+import torch
+from transformers import AutoModelForCausalLM, GPTQConfig
+
+gptq_config = GPTQConfig(bits=4, exllama_config={"version":2})
+model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq", device_map="auto", quantization_config=gptq_config)
+```
+
+<Tip warning={true}>
+
+Only 4-bit models are supported, and we recommend deactivating the ExLlama kernels if you're finetuning a quantized model with PEFT.
+
+</Tip>
+
+The ExLlama kernels are only supported when the entire model is on the GPU. If you're doing inference on a CPU with AutoGPTQ (version > 0.4.2), then you'll need to disable the ExLlama kernel. This overwrites the attributes related to the ExLlama kernels in the quantization config of the config.json file.
+
+```py
+import torch
+from transformers import AutoModelForCausalLM, GPTQConfig
+gptq_config = GPTQConfig(bits=4, use_exllama=False)
+model = AutoModelForCausalLM.from_pretrained("{your_username}/opt-125m-gptq", device_map="cpu", quantization_config=gptq_config)
+```
+
+## bitsandbytes
+
+[bitsandbytes](https://github.com/TimDettmers/bitsandbytes) is the easiest option for quantizing a model to 8 and 4-bit. 8-bit quantization multiplies outliers in fp16 with non-outliers in int8, converts the non-outlier values back to fp16, and then adds them together to return the weights in fp16. This reduces the degradative effect outlier values have on a model's performance. 4-bit quantization compresses a model even further, and it is commonly used with [QLoRA](https://hf.co/papers/2305.14314) to finetune quantized LLMs.
+
+To use bitsandbytes, make sure you have the following libraries installed:
+
+<hfoptions id="bnb">
+<hfoption id="8-bit">
+
+```bash
+pip install transformers accelerate bitsandbytes>0.37.0
+```
+
+</hfoption>
+<hfoption id="4-bit">
+
+```bash
+pip install bitsandbytes>=0.39.0
+pip install --upgrade accelerate
+pip install --upgrade transformers
+```
+
+</hfoption>
+</hfoptions>
+
+Now you can quantize a model with the `load_in_8bit` or `load_in_4bit` parameters in the [`~PreTrainedModel.from_pretrained`] method. This works for any model in any modality, as long as it supports loading with Accelerate and contains `torch.nn.Linear` layers.
+
+<hfoptions id="bnb">
+<hfoption id="8-bit">
+
+Quantizing a model in 8-bit halves the memory-usage, and for large models, set `device_map="auto"` to efficiently use the GPUs available:
+
+```py
+from transformers import AutoModelForCausalLM
+
+model_8bit = AutoModelForCausalLM.from_pretrained("bigscience/bloom-1b7", device_map="auto", load_in_8bit=True)
+```
+
+By default, all the other modules such as `torch.nn.LayerNorm` are converted to `torch.float16`. You can change the data type of these modules with the `torch_dtype` parameter if you want:
+
+```py
+import torch
+from transformers import AutoModelForCausalLM
+
+model_8bit = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", load_in_8bit=True, torch_dtype=torch.float32)
+model_8bit.model.decoder.layers[-1].final_layer_norm.weight.dtype
+```
+
+Once a model is quantized to 8-bit, you can't push the quantized weights to the Hub unless you're using the latest version of Transformers and bitsandbytes. If you have the latest versions, then you can push the 8-bit model to the Hub with the [`~PreTrainedModel.push_to_hub`] method. The quantization config.json file is pushed first, followed by the quantized model weights.
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model = AutoModelForCausalLM.from_pretrained("bigscience/bloom-560m", device_map="auto", load_in_8bit=True)
+tokenizer = AutoTokenizer.from_pretrained("bigscience/bloom-560m")
+
+model.push_to_hub("bloom-560m-8bit")
+```
+
+</hfoption>
+<hfoption id="4-bit">
+
+Quantizing a model in 4-bit reduces your memory-usage by 4x, and for large models, set `device_map="auto"` to efficiently use the GPUs available:
+
+```py
+from transformers import AutoModelForCausalLM
+
+model_4bit = AutoModelForCausalLM.from_pretrained("bigscience/bloom-1b7", device_map="auto", load_in_4bit=True)
+```
+
+By default, all the other modules such as `torch.nn.LayerNorm` are converted to `torch.float16`. You can change the data type of these modules with the `torch_dtype` parameter if you want:
+
+```py
+import torch
+from transformers import AutoModelForCausalLM
+
+model_4bit = AutoModelForCausalLM.from_pretrained("facebook/opt-350m", load_in_4bit=True, torch_dtype=torch.float32)
+model_4bit.model.decoder.layers[-1].final_layer_norm.weight.dtype
+```
+
+Once a model is quantized to 4-bit, you can't push the quantized weights to the Hub.
+
+</hfoption>
+</hfoptions>
+
+<Tip warning={true}>
+
+Training with 8-bit and 4-bit weights are only supported for training *extra* parameters.
+
+</Tip>
+
+You can check your memory footprint with the `get_memory_footprint` method:
+
+```py
+print(model.get_memory_footprint())
+```
+
+Quantized models can be loaded from the [`~PreTrainedModel.from_pretrained`] method without needing to specify the `load_in_8bit` or `load_in_4bit` parameters:
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model = AutoModelForCausalLM.from_pretrained("{your_username}/bloom-560m-8bit", device_map="auto")
+```
+
+### 8-bit
+
+<Tip>
+
+Learn more about the details of 8-bit quantization in this [blog post](https://huggingface.co/blog/hf-bitsandbytes-integration)!
+
+</Tip>
+
+This section explores some of the specific features of 8-bit models, such as offloading, outlier thresholds, skipping module conversion, and finetuning.
+
+#### Offloading
+
+8-bit models can offload weights between the CPU and GPU to support fitting very large models into memory. The weights dispatched to the CPU are actually stored in **float32**, and aren't converted to 8-bit. For example, to enable offloading for the [bigscience/bloom-1b7](https://huggingface.co/bigscience/bloom-1b7) model, start by creating a [`BitsAndBytesConfig`]:
+
+```py
+from transformers import AutoModelForCausalLM, BitsAndBytesConfig
+
+quantization_config = BitsAndBytesConfig(llm_int8_enable_fp32_cpu_offload=True)
+```
+
+Design a custom device map to fit everything on your GPU except for the `lm_head`, which you'll dispatch to the CPU:
+
+```py
+device_map = {
+    "transformer.word_embeddings": 0,
+    "transformer.word_embeddings_layernorm": 0,
+    "lm_head": "cpu",
+    "transformer.h": 0,
+    "transformer.ln_f": 0,
+}
+```
+
+Now load your model with the custom `device_map` and `quantization_config`:
+
+```py
+model_8bit = AutoModelForCausalLM.from_pretrained(
+    "bigscience/bloom-1b7",
+    device_map=device_map,
+    quantization_config=quantization_config,
+)
+```
+
+#### Outlier threshold
+
+An "outlier" is a hidden state value greater than a certain threshold, and these values are computed in fp16. While the values are usually normally distributed ([-3.5, 3.5]), this distribution can be very different for large models ([-60, 6] or [6, 60]). 8-bit quantization works well for values ~5, but beyond that, there is a significant performance penalty. A good default threshold value is 6, but a lower threshold may be needed for more unstable models (small models or finetuning).
+
+To find the best threshold for your model, we recommend experimenting with the `llm_int8_threshold` parameter in [`BitsAndBytesConfig`]:
+
+```py
+from transformers import AutoModelForCausalLM, BitsAndBytesConfig
+
+model_id = "bigscience/bloom-1b7"
+
+quantization_config = BitsAndBytesConfig(
+    llm_int8_threshold=10,
+)
+
+model_8bit = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    device_map=device_map,
+    quantization_config=quantization_config,
+)
+```
+
+#### Skip module conversion
+
+For some models, like [Jukebox](model_doc/jukebox), you don't need to quantize every module to 8-bit which can actually cause instability. With Jukebox, there are several `lm_head` modules that should be skipped using the `llm_int8_skip_modules` parameter in [`BitsAndBytesConfig`]:
+
+```py
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+
+model_id = "bigscience/bloom-1b7"
+
+quantization_config = BitsAndBytesConfig(
+    llm_int8_skip_modules=["lm_head"],
+)
+
+model_8bit = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    device_map="auto",
+    quantization_config=quantization_config,
+)
+```
+
+#### Finetuning
+
+With the [PEFT](https://github.com/huggingface/peft) library, you can finetune large models like [flan-t5-large](https://huggingface.co/google/flan-t5-large) and [facebook/opt-6.7b](https://huggingface.co/facebook/opt-6.7b) with 8-bit quantization. You don't need to pass the `device_map` parameter for training because it'll automatically load your model on a GPU. However, you can still customize the device map with the `device_map` parameter if you want to (`device_map="auto"` should only be used for inference).
+
+### 4-bit
+
+<Tip>
+
+Try 4-bit quantization in this [notebook](https://colab.research.google.com/drive/1ge2F1QSK8Q7h0hn3YKuBCOAS0bK8E0wf) and learn more about it's details in this [blog post](https://huggingface.co/blog/4bit-transformers-bitsandbytes).
+
+</Tip>
+
+This section explores some of the specific features of 4-bit models, such as changing the compute data type, using the Normal Float 4 (NF4) data type, and using nested quantization.
+
+#### Compute data type
+
+To speedup computation, you can change the data type from float32 (the default value) to bf16 using the `bnb_4bit_compute_dtype` parameter in [`BitsAndBytesConfig`]:
+
+```py
+import torch
+from transformers import BitsAndBytesConfig
+
+quantization_config = BitsAndBytesConfig(load_in_4bit=True, bnb_4bit_compute_dtype=torch.bfloat16)
+```
+
+#### Normal Float 4 (NF4)
+
+NF4 is a 4-bit data type from the [QLoRA](https://hf.co/papers/2305.14314) paper, adapted for weights initialized from a normal distribution. You should use NF4 for training 4-bit base models. This can be configured with the `bnb_4bit_quant_type` parameter in the [`BitsAndBytesConfig`]:
+
+```py
+from transformers import BitsAndBytesConfig
+
+nf4_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_quant_type="nf4",
+)
+
+model_nf4 = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=nf4_config)
+```
+
+For inference, the `bnb_4bit_quant_type` does not have a huge impact on performance. However, to remain consistent with the model weights, you should use the `bnb_4bit_compute_dtype` and `torch_dtype` values.
+
+#### Nested quantization
+
+Nested quantization is a technique that can save additional memory at no additional performance cost. This feature performs a second quantization of the already quantized weights to save an addition 0.4 bits/parameter. For example, with nested quantization, you can finetune a [Llama-13b](https://huggingface.co/meta-llama/Llama-2-13b) model on a 16GB NVIDIA T4 GPU with a sequence length of 1024, a batch size of 1, and enabling gradient accumulation with 4 steps.
+
+```py
+from transformers import BitsAndBytesConfig
+
+double_quant_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_use_double_quant=True,
+)
+
+model_double_quant = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-2-13b", quantization_config=double_quant_config)
+```
+
+## Optimum
+
+The [Optimum](https://huggingface.co/docs/optimum/index) library supports quantization for Intel, Furiosa, ONNX Runtime, GPTQ, and lower-level PyTorch quantization functions. Consider using Optimum for quantization if you're using specific and optimized hardware like Intel CPUs, Furiosa NPUs or a model accelerator like ONNX Runtime.
+
+## Benchmarks
+
+To compare the speed, throughput, and latency of each quantization scheme, check the following benchmarks obtained from the [optimum-benchmark](https://github.com/huggingface/optimum-benchmark) library. The benchmark was run on a NVIDIA A1000 for the [TheBloke/Mistral-7B-v0.1-AWQ](https://huggingface.co/TheBloke/Mistral-7B-v0.1-AWQ) and [TheBloke/Mistral-7B-v0.1-GPTQ](https://huggingface.co/TheBloke/Mistral-7B-v0.1-GPTQ) models. These were also tested against the bitsandbytes quantization methods as well as a native fp16 model.
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/forward_memory_plot.png" alt="forward peak memory per batch size" />
+    <figcaption class="mt-2 text-center text-sm text-gray-500">forward peak memory/batch size</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/generate_memory_plot.png" alt="generate peak memory per batch size" />
+    <figcaption class="mt-2 text-center text-sm text-gray-500">generate peak memory/batch size</figcaption>
+  </div>
+</div>
+
+<div class="flex gap-4">
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/generate_throughput_plot.png" alt="generate throughput per batch size" />
+    <figcaption class="mt-2 text-center text-sm text-gray-500">generate throughput/batch size</figcaption>
+  </div>
+  <div>
+    <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/forward_latency_plot.png" alt="forward latency per batch size" />
+    <figcaption class="mt-2 text-center text-sm text-gray-500">forward latency/batch size</figcaption>
+  </div>
+</div>
+
+The benchmarks indicate AWQ quantization is the fastest for inference, text generation, and has the lowest peak memory for text generation. However, AWQ has the largest forward latency per batch size. For a more detailed discussion about the pros and cons of each quantization method, read the [Overview of natively supported quantization schemes in 🤗 Transformers](https://huggingface.co/blog/overview-quantization-transformers) blog post.

docs/source/en/run_scripts.md

@@ -130,7 +130,7 @@ The [Trainer](https://huggingface.co/docs/transformers/main_classes/trainer) sup
 - Set the number of GPUs to use with the `nproc_per_node` argument.
 
 ```bash
-python -m torch.distributed.launch \
+torchrun \
     --nproc_per_node 8 pytorch/summarization/run_summarization.py \
     --fp16 \
     --model_name_or_path t5-small \

docs/source/en/tasks/idefics.md

@@ -109,7 +109,6 @@ on the fly while loading.
 Now that you have the model loaded in one of the suggested ways, let's move on to exploring tasks that you can use IDEFICS for.
 
 ## Image captioning
-
 Image captioning is the task of predicting a caption for a given image. A common application is to aid visually impaired 
 people navigate through different situations, for instance, explore image content online. 
 
@@ -229,7 +228,7 @@ Let's get a new image for this task:
      <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/idefics-vqa.jpg" alt="Image of a couple having a picnic"/>
 </div>
 
-Photo by [Jarritos Mexican Soda](https://unsplash.com/@jarritos).
+Photo by [Jarritos Mexican Soda](https://unsplash.com/@jarritos). 
 
 You can steer the model from image captioning to visual question answering by prompting it with appropriate instructions: 
 

docs/source/en/tasks/language_modeling.md

@@ -37,7 +37,7 @@ You can finetune other architectures for causal language modeling following the
 Choose one of the following architectures:
 
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
-[BART](../model_doc/bart), [BERT](../model_doc/bert), [Bert Generation](../model_doc/bert-generation), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CodeLlama](../model_doc/code_llama), [CodeGen](../model_doc/codegen), [CPM-Ant](../model_doc/cpmant), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [Falcon](../model_doc/falcon), [Fuyu](../model_doc/fuyu), [GIT](../model_doc/git), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT NeoX Japanese](../model_doc/gpt_neox_japanese), [GPT-J](../model_doc/gptj), [LLaMA](../model_doc/llama), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [MPT](../model_doc/mpt), [MusicGen](../model_doc/musicgen), [MVP](../model_doc/mvp), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Pegasus](../model_doc/pegasus), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [QDQBert](../model_doc/qdqbert), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [RWKV](../model_doc/rwkv), [Speech2Text2](../model_doc/speech_to_text_2), [Transformer-XL](../model_doc/transfo-xl), [TrOCR](../model_doc/trocr), [Whisper](../model_doc/whisper), [XGLM](../model_doc/xglm), [XLM](../model_doc/xlm), [XLM-ProphetNet](../model_doc/xlm-prophetnet), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod)
+[BART](../model_doc/bart), [BERT](../model_doc/bert), [Bert Generation](../model_doc/bert-generation), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CodeLlama](../model_doc/code_llama), [CodeGen](../model_doc/codegen), [CPM-Ant](../model_doc/cpmant), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [Falcon](../model_doc/falcon), [Fuyu](../model_doc/fuyu), [GIT](../model_doc/git), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT NeoX Japanese](../model_doc/gpt_neox_japanese), [GPT-J](../model_doc/gptj), [LLaMA](../model_doc/llama), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [Mixtral](../model_doc/mixtral), [MPT](../model_doc/mpt), [MusicGen](../model_doc/musicgen), [MVP](../model_doc/mvp), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Pegasus](../model_doc/pegasus), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [QDQBert](../model_doc/qdqbert), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [RWKV](../model_doc/rwkv), [Speech2Text2](../model_doc/speech_to_text_2), [Transformer-XL](../model_doc/transfo-xl), [TrOCR](../model_doc/trocr), [Whisper](../model_doc/whisper), [XGLM](../model_doc/xglm), [XLM](../model_doc/xlm), [XLM-ProphetNet](../model_doc/xlm-prophetnet), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod)
 
 
 
@@ -110,7 +110,7 @@ The next step is to load a DistilGPT2 tokenizer to process the `text` subfield:
 ```
 
 You'll notice from the example above, the `text` field is actually nested inside `answers`. This means you'll need to
-extract the `text` subfield from its nested structure with the [`flatten`](https://huggingface.co/docs/datasets/process.html#flatten) method:
+extract the `text` subfield from its nested structure with the [`flatten`](https://huggingface.co/docs/datasets/process#flatten) method:
 
 ```py
 >>> eli5 = eli5.flatten()

docs/source/en/tasks/masked_language_modeling.md

@@ -105,7 +105,7 @@ For masked language modeling, the next step is to load a DistilRoBERTa tokenizer
 ```
 
 You'll notice from the example above, the `text` field is actually nested inside `answers`. This means you'll need to e
-xtract the `text` subfield from its nested structure with the [`flatten`](https://huggingface.co/docs/datasets/process.html#flatten) method:
+xtract the `text` subfield from its nested structure with the [`flatten`](https://huggingface.co/docs/datasets/process#flatten) method:
 
 ```py
 >>> eli5 = eli5.flatten()

docs/source/en/tasks/semantic_segmentation.md

@@ -14,29 +14,17 @@ rendered properly in your Markdown viewer.
 
 -->
 
-# Semantic segmentation
+# Image Segmentation
 
 [[open-in-colab]]
 
 <Youtube id="dKE8SIt9C-w"/>
 
-Semantic segmentation assigns a label or class to each individual pixel of an image. There are several types of segmentation, and in the case of semantic segmentation, no distinction is made between unique instances of the same object. Both objects are given the same label (for example, "car" instead of "car-1" and "car-2"). Common real-world applications of semantic segmentation include training self-driving cars to identify pedestrians and important traffic information, identifying cells and abnormalities in medical imagery, and monitoring environmental changes from satellite imagery.
+Image segmentation models separate areas corresponding to different areas of interest in an image. These models work by assigning a label to each pixel. There are several types of segmentation: semantic segmentation, instance segmentation, and panoptic segmentation.
 
-This guide will show you how to:
-
-1. Finetune [SegFormer](https://huggingface.co/docs/transformers/main/en/model_doc/segformer#segformer) on the [SceneParse150](https://huggingface.co/datasets/scene_parse_150) dataset.
-2. Use your finetuned model for inference.
-
-<Tip>
-The task illustrated in this tutorial is supported by the following model architectures:
-
-<!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
-
-[BEiT](../model_doc/beit), [Data2VecVision](../model_doc/data2vec-vision), [DPT](../model_doc/dpt), [MobileNetV2](../model_doc/mobilenet_v2), [MobileViT](../model_doc/mobilevit), [MobileViTV2](../model_doc/mobilevitv2), [SegFormer](../model_doc/segformer), [UPerNet](../model_doc/upernet)
-
-<!--End of the generated tip-->
-
-</Tip>
+In this guide, we will:
+1. [Take a look at different types of segmentation](#types-of-segmentation).
+2. [Have an end-to-end fine-tuning example for semantic segmentation](#fine-tuning-a-model-for-segmentation).
 
 Before you begin, make sure you have all the necessary libraries installed:
 
@@ -52,7 +40,178 @@ We encourage you to log in to your Hugging Face account so you can upload and sh
 >>> notebook_login()
 ```
 
-## Load SceneParse150 dataset
+## Types of Segmentation
+
+Semantic segmentation assigns a label or class to every single pixel in an image. Let's take a look at a semantic segmentation model output. It will assign the same class to every instance of an object it comes across in an image, for example, all cats will be labeled as "cat" instead of "cat-1", "cat-2".
+We can use transformers' image segmentation pipeline to quickly infer a semantic segmentation model. Let's take a look at the example image.
+
+```python
+from transformers import pipeline
+from PIL import Image
+import requests
+
+url = "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/segmentation_input.jpg"
+image = Image.open(requests.get(url, stream=True).raw)
+image
+```
+
+<div class="flex justify-center">
+     <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/segmentation_input.jpg" alt="Segmentation Input"/>
+</div>
+
+We will use [nvidia/segformer-b1-finetuned-cityscapes-1024-1024](https://huggingface.co/nvidia/segformer-b1-finetuned-cityscapes-1024-1024). 
+
+```python
+semantic_segmentation = pipeline("image-segmentation", "nvidia/segformer-b1-finetuned-cityscapes-1024-1024")
+results = semantic_segmentation(image)
+results
+```
+
+The segmentation pipeline output includes a mask for every predicted class. 
+```bash
+[{'score': None,
+  'label': 'road',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'sidewalk',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'building',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'wall',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'pole',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'traffic sign',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'vegetation',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'terrain',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'sky',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': None,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>}]
+```
+
+Taking a look at the mask for the car class, we can see every car is classified with the same mask.
+
+```python
+results[-1]["mask"]
+```
+<div class="flex justify-center">
+     <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/semantic_segmentation_output.png" alt="Semantic Segmentation Output"/>
+</div>
+
+In instance segmentation, the goal is not to classify every pixel, but to predict a mask for **every instance of an object** in a given image. It works very similar to object detection, where there is a bounding box for every instance, there's a segmentation mask instead. We will use [facebook/mask2former-swin-large-cityscapes-instance](https://huggingface.co/facebook/mask2former-swin-large-cityscapes-instance) for this. 
+
+```python
+instance_segmentation = pipeline("image-segmentation", "facebook/mask2former-swin-large-cityscapes-instance")
+results = instance_segmentation(Image.open(image))
+results
+```
+
+As you can see below, there are multiple cars classified, and there's no classification for pixels other than pixels that belong to car and person instances.
+
+```bash
+[{'score': 0.999944,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999945,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999652,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.903529,
+  'label': 'person',
+  'mask': <PIL.Image.Image image mode=L size=612x415>}]
+```
+Checking out one of the car masks below.
+
+```python
+results[2]["mask"]
+```
+<div class="flex justify-center">
+     <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/instance_segmentation_output.png" alt="Semantic Segmentation Output"/>
+</div>
+
+Panoptic segmentation combines semantic segmentation and instance segmentation, where every pixel is classified into a class and an instance of that class, and there are multiple masks for each instance of a class. We can use [facebook/mask2former-swin-large-cityscapes-panoptic](https://huggingface.co/facebook/mask2former-swin-large-cityscapes-panoptic) for this.
+
+```python
+panoptic_segmentation = pipeline("image-segmentation", "facebook/mask2former-swin-large-cityscapes-panoptic")
+results = panoptic_segmentation(Image.open(image))
+results
+```
+As you can see below, we have more classes. We will later illustrate to see that every pixel is classified into one of the classes.
+
+```bash
+[{'score': 0.999981,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999958,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.99997,
+  'label': 'vegetation',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999575,
+  'label': 'pole',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999958,
+  'label': 'building',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999634,
+  'label': 'road',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.996092,
+  'label': 'sidewalk',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.999221,
+  'label': 'car',
+  'mask': <PIL.Image.Image image mode=L size=612x415>},
+ {'score': 0.99987,
+  'label': 'sky',
+  'mask': <PIL.Image.Image image mode=L size=612x415>}]
+```
+
+Let's have a side by side comparison for all types of segmentation.
+
+<div class="flex justify-center">
+     <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/tasks/segmentation-comparison.png" alt="Segmentation Maps Compared"/>
+</div>
+
+Seeing all types of segmentation, let's have a deep dive on fine-tuning a model for semantic segmentation.
+
+Common real-world applications of semantic segmentation include training self-driving cars to identify pedestrians and important traffic information, identifying cells and abnormalities in medical imagery, and monitoring environmental changes from satellite imagery.
+
+## Fine-tuning a Model for Segmentation
+
+We will now:
+
+1. Finetune [SegFormer](https://huggingface.co/docs/transformers/main/en/model_doc/segformer#segformer) on the [SceneParse150](https://huggingface.co/datasets/scene_parse_150) dataset.
+2. Use your fine-tuned model for inference.
+
+<Tip>
+The task illustrated in this tutorial is supported by the following model architectures:
+
+<!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
+
+[BEiT](../model_doc/beit), [Data2VecVision](../model_doc/data2vec-vision), [DPT](../model_doc/dpt), [MobileNetV2](../model_doc/mobilenet_v2), [MobileViT](../model_doc/mobilevit), [MobileViTV2](../model_doc/mobilevitv2), [SegFormer](../model_doc/segformer), [UPerNet](../model_doc/upernet)
+
+<!--End of the generated tip-->
+
+</Tip>
+
+
+### Load SceneParse150 dataset
 
 Start by loading a smaller subset of the SceneParse150 dataset from the 🤗 Datasets library. This'll give you a chance to experiment and make sure everything works before spending more time training on the full dataset.
 
@@ -97,7 +256,60 @@ You'll also want to create a dictionary that maps a label id to a label class wh
 >>> num_labels = len(id2label)
 ```
 
-## Preprocess
+#### Custom dataset
+
+You could also create and use your own dataset if you prefer to train with the [run_semantic_segmentation.py](https://github.com/huggingface/transformers/blob/main/examples/pytorch/semantic-segmentation/run_semantic_segmentation.py) script instead of a notebook instance. The script requires:
+
+1. a [`~datasets.DatasetDict`] with two [`~datasets.Image`] columns, "image" and "label"
+
+     ```py
+     from datasets import Dataset, DatasetDict, Image
+
+     image_paths_train = ["path/to/image_1.jpg/jpg", "path/to/image_2.jpg/jpg", ..., "path/to/image_n.jpg/jpg"]
+     label_paths_train = ["path/to/annotation_1.png", "path/to/annotation_2.png", ..., "path/to/annotation_n.png"]
+
+     image_paths_validation = [...]
+     label_paths_validation = [...]
+
+     def create_dataset(image_paths, label_paths):
+         dataset = Dataset.from_dict({"image": sorted(image_paths),
+                                     "label": sorted(label_paths)})
+         dataset = dataset.cast_column("image", Image())
+         dataset = dataset.cast_column("label", Image())
+
+     return dataset
+
+     # step 1: create Dataset objects
+     train_dataset = create_dataset(image_paths_train, label_paths_train)
+     validation_dataset = create_dataset(image_paths_validation, label_paths_validation)
+
+     # step 2: create DatasetDict
+     dataset = DatasetDict({
+          "train": train_dataset,
+          "validation": validation_dataset,
+          }
+     )
+
+     # step 3: push to Hub (assumes you have ran the huggingface-cli login command in a terminal/notebook)
+     dataset.push_to_hub("your-name/dataset-repo")
+
+     # optionally, you can push to a private repo on the Hub
+     # dataset.push_to_hub("name of repo on the hub", private=True)
+     ```
+
+2. an id2label dictionary mapping the class integers to their class names
+
+     ```py
+     import json
+     # simple example
+     id2label = {0: 'cat', 1: 'dog'}
+     with open('id2label.json', 'w') as fp:
+     json.dump(id2label, fp)
+     ```
+
+As an example, take a look at this [example dataset](https://huggingface.co/datasets/nielsr/ade20k-demo) which was created with the steps shown above.
+
+### Preprocess
 
 The next step is to load a SegFormer image processor to prepare the images and annotations for the model. Some datasets, like this one, use the zero-index as the background class. However, the background class isn't actually included in the 150 classes, so you'll need to set `reduce_labels=True` to subtract one from all the labels. The zero-index is replaced by `255` so it's ignored by SegFormer's loss function:
 
@@ -204,7 +416,7 @@ The transform is applied on the fly which is faster and consumes less disk space
 </tf>
 </frameworkcontent>
 
-## Evaluate
+### Evaluate
 
 Including a metric during training is often helpful for evaluating your model's performance. You can quickly load an evaluation method with the 🤗 [Evaluate](https://huggingface.co/docs/evaluate/index) library. For this task, load the [mean Intersection over Union](https://huggingface.co/spaces/evaluate-metric/accuracy) (IoU) metric (see the 🤗 Evaluate [quick tour](https://huggingface.co/docs/evaluate/a_quick_tour) to learn more about how to load and compute a metric):
 
@@ -245,7 +457,7 @@ logits first, and then reshaped to match the size of the labels before you can c
 ...             reduce_labels=False,
 ...         )
 ...         for key, value in metrics.items():
-...             if type(value) is np.ndarray:
+...             if isinstance(value, np.ndarray):
 ...                 metrics[key] = value.tolist()
 ...         return metrics
 ```
@@ -289,7 +501,7 @@ logits first, and then reshaped to match the size of the labels before you can c
 
 Your `compute_metrics` function is ready to go now, and you'll return to it when you setup your training.
 
-## Train
+### Train
 <frameworkcontent>
 <pt>
 <Tip>
@@ -453,7 +665,7 @@ Congratulations! You have fine-tuned your model and shared it on the 🤗 Hub. Y
 </frameworkcontent>
 
 
-## Inference
+### Inference
 
 Great, now that you've finetuned a model, you can use it for inference!
 
@@ -470,43 +682,8 @@ Load an image for inference:
 
 <frameworkcontent>
 <pt>
-The simplest way to try out your finetuned model for inference is to use it in a [`pipeline`]. Instantiate a `pipeline` for image segmentation with your model, and pass your image to it:
 
-```py
->>> from transformers import pipeline
-
->>> segmenter = pipeline("image-segmentation", model="my_awesome_seg_model")
->>> segmenter(image)
-[{'score': None,
-  'label': 'wall',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062690>},
- {'score': None,
-  'label': 'sky',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062A50>},
- {'score': None,
-  'label': 'floor',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062B50>},
- {'score': None,
-  'label': 'ceiling',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062A10>},
- {'score': None,
-  'label': 'bed ',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062E90>},
- {'score': None,
-  'label': 'windowpane',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062390>},
- {'score': None,
-  'label': 'cabinet',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062550>},
- {'score': None,
-  'label': 'chair',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062D90>},
- {'score': None,
-  'label': 'armchair',
-  'mask': <PIL.Image.Image image mode=L size=640x427 at 0x7FD5B2062E10>}]
-```
-
-You can also manually replicate the results of the `pipeline` if you'd like. Process the image with an image processor and place the `pixel_values` on a GPU:
+We will now see how to infer without a pipeline. Process the image with an image processor and place the `pixel_values` on a GPU:
 
 ```py
 >>> device = torch.device("cuda" if torch.cuda.is_available() else "cpu")  # use GPU if available, otherwise use a CPU

docs/source/en/tasks/sequence_classification.md

@@ -33,7 +33,7 @@ The task illustrated in this tutorial is supported by the following model archit
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
 
 
-[ALBERT](../model_doc/albert), [BART](../model_doc/bart), [BERT](../model_doc/bert), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CANINE](../model_doc/canine), [CodeLlama](../model_doc/code_llama), [ConvBERT](../model_doc/convbert), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [DeBERTa](../model_doc/deberta), [DeBERTa-v2](../model_doc/deberta-v2), [DistilBERT](../model_doc/distilbert), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [ErnieM](../model_doc/ernie_m), [ESM](../model_doc/esm), [Falcon](../model_doc/falcon), [FlauBERT](../model_doc/flaubert), [FNet](../model_doc/fnet), [Funnel Transformer](../model_doc/funnel), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT-J](../model_doc/gptj), [I-BERT](../model_doc/ibert), [LayoutLM](../model_doc/layoutlm), [LayoutLMv2](../model_doc/layoutlmv2), [LayoutLMv3](../model_doc/layoutlmv3), [LED](../model_doc/led), [LiLT](../model_doc/lilt), [LLaMA](../model_doc/llama), [Longformer](../model_doc/longformer), [LUKE](../model_doc/luke), [MarkupLM](../model_doc/markuplm), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [MobileBERT](../model_doc/mobilebert), [MPNet](../model_doc/mpnet), [MPT](../model_doc/mpt), [MRA](../model_doc/mra), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [Nezha](../model_doc/nezha), [Nyströmformer](../model_doc/nystromformer), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Perceiver](../model_doc/perceiver), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [QDQBert](../model_doc/qdqbert), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [SqueezeBERT](../model_doc/squeezebert), [T5](../model_doc/t5), [TAPAS](../model_doc/tapas), [Transformer-XL](../model_doc/transfo-xl), [UMT5](../model_doc/umt5), [XLM](../model_doc/xlm), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod), [YOSO](../model_doc/yoso)
+[ALBERT](../model_doc/albert), [BART](../model_doc/bart), [BERT](../model_doc/bert), [BigBird](../model_doc/big_bird), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [BioGpt](../model_doc/biogpt), [BLOOM](../model_doc/bloom), [CamemBERT](../model_doc/camembert), [CANINE](../model_doc/canine), [CodeLlama](../model_doc/code_llama), [ConvBERT](../model_doc/convbert), [CTRL](../model_doc/ctrl), [Data2VecText](../model_doc/data2vec-text), [DeBERTa](../model_doc/deberta), [DeBERTa-v2](../model_doc/deberta-v2), [DistilBERT](../model_doc/distilbert), [ELECTRA](../model_doc/electra), [ERNIE](../model_doc/ernie), [ErnieM](../model_doc/ernie_m), [ESM](../model_doc/esm), [Falcon](../model_doc/falcon), [FlauBERT](../model_doc/flaubert), [FNet](../model_doc/fnet), [Funnel Transformer](../model_doc/funnel), [GPT-Sw3](../model_doc/gpt-sw3), [OpenAI GPT-2](../model_doc/gpt2), [GPTBigCode](../model_doc/gpt_bigcode), [GPT Neo](../model_doc/gpt_neo), [GPT NeoX](../model_doc/gpt_neox), [GPT-J](../model_doc/gptj), [I-BERT](../model_doc/ibert), [LayoutLM](../model_doc/layoutlm), [LayoutLMv2](../model_doc/layoutlmv2), [LayoutLMv3](../model_doc/layoutlmv3), [LED](../model_doc/led), [LiLT](../model_doc/lilt), [LLaMA](../model_doc/llama), [Longformer](../model_doc/longformer), [LUKE](../model_doc/luke), [MarkupLM](../model_doc/markuplm), [mBART](../model_doc/mbart), [MEGA](../model_doc/mega), [Megatron-BERT](../model_doc/megatron-bert), [Mistral](../model_doc/mistral), [Mixtral](../model_doc/mixtral), [MobileBERT](../model_doc/mobilebert), [MPNet](../model_doc/mpnet), [MPT](../model_doc/mpt), [MRA](../model_doc/mra), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [Nezha](../model_doc/nezha), [Nyströmformer](../model_doc/nystromformer), [OpenLlama](../model_doc/open-llama), [OpenAI GPT](../model_doc/openai-gpt), [OPT](../model_doc/opt), [Perceiver](../model_doc/perceiver), [Persimmon](../model_doc/persimmon), [Phi](../model_doc/phi), [PLBart](../model_doc/plbart), [QDQBert](../model_doc/qdqbert), [Reformer](../model_doc/reformer), [RemBERT](../model_doc/rembert), [RoBERTa](../model_doc/roberta), [RoBERTa-PreLayerNorm](../model_doc/roberta-prelayernorm), [RoCBert](../model_doc/roc_bert), [RoFormer](../model_doc/roformer), [SqueezeBERT](../model_doc/squeezebert), [T5](../model_doc/t5), [TAPAS](../model_doc/tapas), [Transformer-XL](../model_doc/transfo-xl), [UMT5](../model_doc/umt5), [XLM](../model_doc/xlm), [XLM-RoBERTa](../model_doc/xlm-roberta), [XLM-RoBERTa-XL](../model_doc/xlm-roberta-xl), [XLNet](../model_doc/xlnet), [X-MOD](../model_doc/xmod), [YOSO](../model_doc/yoso)
 
 
 

docs/source/en/tasks/summarization.md

@@ -35,7 +35,7 @@ The task illustrated in this tutorial is supported by the following model archit
 
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
 
-[BART](../model_doc/bart), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [Encoder decoder](../model_doc/encoder-decoder), [FairSeq Machine-Translation](../model_doc/fsmt), [GPTSAN-japanese](../model_doc/gptsan-japanese), [LED](../model_doc/led), [LongT5](../model_doc/longt5), [M2M100](../model_doc/m2m_100), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [NLLB](../model_doc/nllb), [NLLB-MOE](../model_doc/nllb-moe), [Pegasus](../model_doc/pegasus), [PEGASUS-X](../model_doc/pegasus_x), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [SeamlessM4T](../model_doc/seamless_m4t), [SwitchTransformers](../model_doc/switch_transformers), [T5](../model_doc/t5), [UMT5](../model_doc/umt5), [XLM-ProphetNet](../model_doc/xlm-prophetnet)
+[BART](../model_doc/bart), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [Encoder decoder](../model_doc/encoder-decoder), [FairSeq Machine-Translation](../model_doc/fsmt), [GPTSAN-japanese](../model_doc/gptsan-japanese), [LED](../model_doc/led), [LongT5](../model_doc/longt5), [M2M100](../model_doc/m2m_100), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [NLLB](../model_doc/nllb), [NLLB-MOE](../model_doc/nllb-moe), [Pegasus](../model_doc/pegasus), [PEGASUS-X](../model_doc/pegasus_x), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [SeamlessM4T](../model_doc/seamless_m4t), [SeamlessM4Tv2](../model_doc/seamless_m4t_v2), [SwitchTransformers](../model_doc/switch_transformers), [T5](../model_doc/t5), [UMT5](../model_doc/umt5), [XLM-ProphetNet](../model_doc/xlm-prophetnet)
 
 <!--End of the generated tip-->
 
@@ -126,6 +126,7 @@ Now create a batch of examples using [`DataCollatorForSeq2Seq`]. It's more effic
 
 <frameworkcontent>
 <pt>
+
 ```py
 >>> from transformers import DataCollatorForSeq2Seq
 
@@ -133,6 +134,7 @@ Now create a batch of examples using [`DataCollatorForSeq2Seq`]. It's more effic
 ```
 </pt>
 <tf>
+
 ```py
 >>> from transformers import DataCollatorForSeq2Seq
 

docs/source/en/tasks/text-to-speech.md

@@ -74,6 +74,12 @@ To follow this guide you will need a GPU. If you're working in a notebook, run t
 !nvidia-smi
 ```
 
+or alternatively for AMD GPUs:
+
+```bash
+!rocm-smi
+```
+
 </Tip>
 
 We encourage you to log in to your Hugging Face account to upload and share your model with the community. When prompted, enter your token to log in:
@@ -630,4 +636,4 @@ see if this improves the results.
 
 Finally, it is essential to consider ethical considerations. Although TTS technology has numerous useful applications, it 
 may also be used for malicious purposes, such as impersonating someone's voice without their knowledge or consent. Please 
-use TTS judiciously and responsibly.
+use TTS judiciously and responsibly.

docs/source/en/tasks/translation.md

@@ -32,7 +32,7 @@ The task illustrated in this tutorial is supported by the following model archit
 
 <!--This tip is automatically generated by `make fix-copies`, do not fill manually!-->
 
-[BART](../model_doc/bart), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [Encoder decoder](../model_doc/encoder-decoder), [FairSeq Machine-Translation](../model_doc/fsmt), [GPTSAN-japanese](../model_doc/gptsan-japanese), [LED](../model_doc/led), [LongT5](../model_doc/longt5), [M2M100](../model_doc/m2m_100), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [NLLB](../model_doc/nllb), [NLLB-MOE](../model_doc/nllb-moe), [Pegasus](../model_doc/pegasus), [PEGASUS-X](../model_doc/pegasus_x), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [SeamlessM4T](../model_doc/seamless_m4t), [SwitchTransformers](../model_doc/switch_transformers), [T5](../model_doc/t5), [UMT5](../model_doc/umt5), [XLM-ProphetNet](../model_doc/xlm-prophetnet)
+[BART](../model_doc/bart), [BigBird-Pegasus](../model_doc/bigbird_pegasus), [Blenderbot](../model_doc/blenderbot), [BlenderbotSmall](../model_doc/blenderbot-small), [Encoder decoder](../model_doc/encoder-decoder), [FairSeq Machine-Translation](../model_doc/fsmt), [GPTSAN-japanese](../model_doc/gptsan-japanese), [LED](../model_doc/led), [LongT5](../model_doc/longt5), [M2M100](../model_doc/m2m_100), [Marian](../model_doc/marian), [mBART](../model_doc/mbart), [MT5](../model_doc/mt5), [MVP](../model_doc/mvp), [NLLB](../model_doc/nllb), [NLLB-MOE](../model_doc/nllb-moe), [Pegasus](../model_doc/pegasus), [PEGASUS-X](../model_doc/pegasus_x), [PLBart](../model_doc/plbart), [ProphetNet](../model_doc/prophetnet), [SeamlessM4T](../model_doc/seamless_m4t), [SeamlessM4Tv2](../model_doc/seamless_m4t_v2), [SwitchTransformers](../model_doc/switch_transformers), [T5](../model_doc/t5), [UMT5](../model_doc/umt5), [XLM-ProphetNet](../model_doc/xlm-prophetnet)
 
 <!--End of the generated tip-->
 
@@ -232,7 +232,7 @@ At this point, only three steps remain:
 ... )
 
 >>> trainer.train()
-````
+```
 
 Once training is completed, share your model to the Hub with the [`~transformers.Trainer.push_to_hub`] method so everyone can use your model:
 

docs/source/en/training.md

@@ -43,7 +43,7 @@ Begin by loading the [Yelp Reviews](https://huggingface.co/datasets/yelp_review_
  'text': 'My expectations for McDonalds are t rarely high. But for one to still fail so spectacularly...that takes something special!\\nThe cashier took my friends\'s order, then promptly ignored me. I had to force myself in front of a cashier who opened his register to wait on the person BEHIND me. I waited over five minutes for a gigantic order that included precisely one kid\'s meal. After watching two people who ordered after me be handed their food, I asked where mine was. The manager started yelling at the cashiers for \\"serving off their orders\\" when they didn\'t have their food. But neither cashier was anywhere near those controls, and the manager was the one serving food to customers and clearing the boards.\\nThe manager was rude when giving me my order. She didn\'t make sure that I had everything ON MY RECEIPT, and never even had the decency to apologize that I felt I was getting poor service.\\nI\'ve eaten at various McDonalds restaurants for over 30 years. I\'ve worked at more than one location. I expect bad days, bad moods, and the occasional mistake. But I have yet to have a decent experience at this store. It will remain a place I avoid unless someone in my party needs to avoid illness from low blood sugar. Perhaps I should go back to the racially biased service of Steak n Shake instead!'}
 ```
 
-As you now know, you need a tokenizer to process the text and include a padding and truncation strategy to handle any variable sequence lengths. To process your dataset in one step, use 🤗 Datasets [`map`](https://huggingface.co/docs/datasets/process.html#map) method to apply a preprocessing function over the entire dataset:
+As you now know, you need a tokenizer to process the text and include a padding and truncation strategy to handle any variable sequence lengths. To process your dataset in one step, use 🤗 Datasets [`map`](https://huggingface.co/docs/datasets/process#map) method to apply a preprocessing function over the entire dataset:
 
 ```py
 >>> from transformers import AutoTokenizer
@@ -119,7 +119,7 @@ Specify where to save the checkpoints from your training:
 >>> metric = evaluate.load("accuracy")
 ```
 
-Call [`~evaluate.compute`] on `metric` to calculate the accuracy of your predictions. Before passing your predictions to `compute`, you need to convert the predictions to logits (remember all 🤗 Transformers models return logits):
+Call [`~evaluate.compute`] on `metric` to calculate the accuracy of your predictions. Before passing your predictions to `compute`, you need to convert the logits to predictions (remember all 🤗 Transformers models return logits):
 
 ```py
 >>> def compute_metrics(eval_pred):

docs/source/es/_toctree.yml

@@ -75,6 +75,10 @@
 - sections:
   - local: philosophy
     title: Filosofía
+  - local: pad_truncation
+    title: Relleno y truncamiento
   - local: bertology
     title: BERTología
+  - local: perplexity
+    title: Perplejidad de los modelos de longitud fija
   title: Guías conceptuales

docs/source/es/converting_tensorflow_models.md

@@ -96,20 +96,6 @@ transformers-cli convert --model_type gpt2 \
   [--finetuning_task_name OPENAI_GPT2_FINETUNED_TASK]
 ```
 
-## Transformer-XL
-
-Aquí hay un ejemplo del proceso para convertir un modelo Transformer-XL pre-entrenado (más información [aquí](https://github.com/kimiyoung/transformer-xl/tree/master/tf#obtain-and-evaluate-pretrained-sota-models)):
-
-```bash
-export TRANSFO_XL_CHECKPOINT_FOLDER_PATH=/path/to/transfo/xl/checkpoint
-
-transformers-cli convert --model_type transfo_xl \
-  --tf_checkpoint $TRANSFO_XL_CHECKPOINT_FOLDER_PATH \
-  --pytorch_dump_output $PYTORCH_DUMP_OUTPUT \
-  [--config TRANSFO_XL_CONFIG] \
-  [--finetuning_task_name TRANSFO_XL_FINETUNED_TASK]
-```
-
 ## XLNet
 
 Aquí hay un ejemplo del proceso para convertir un modelo XLNet pre-entrenado:

docs/source/es/pad_truncation.md

@@ -0,0 +1,69 @@
+<!--Copyright 2022 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.
+
+-->
+
+# Relleno y truncamiento
+
+Las entradas agrupadas por lotes (batched) suelen tener longitudes diferentes, por lo que no se pueden convertir en tensores de tamaño fijo. El relleno (también conocido como "Padding") y el truncamiento (conocido como "Truncation") son estrategias para abordar este problema y crear tensores rectangulares a partir de lotes de longitudes variables. El relleno agrega un **padding token** especial para garantizar que las secuencias más cortas tengan la misma longitud que la secuencia más larga en un lote o la longitud máxima aceptada por el modelo. El truncamiento funciona en la otra dirección al truncar secuencias largas.
+
+En la mayoría de los casos, es bastante eficaz rellenar el lote hasta la longitud de la secuencia más larga y truncar hasta la longitud máxima que un modelo puede aceptar. Sin embargo, la API admite más estrategias si las necesitas. Los tres argumentos que necesitas son: `padding`, `truncation` y `max_length`.
+
+El argumento `padding` controla el relleno. Puede ser un booleano o una cadena:
+
+  - `True` o `'longest'`: rellena hasta la longitud de la secuencia más larga en el lote (no se aplica relleno si solo proporcionas una única secuencia).
+  - `'max_length'`: rellena hasta una longitud especificada por el argumento `max_length` o la longitud máxima aceptada
+    por el modelo si no se proporciona `max_length` (`max_length=None`). El relleno se aplicará incluso si solo proporcionas una única secuencia.
+  - `False` o `'do_not_pad'`: no se aplica relleno. Este es el comportamiento predeterminado.
+
+El argumento `truncation` controla el truncamiento. Puede ser un booleano o una cadena:
+
+  - `True` o `'longest_first'`: trunca hasta una longitud máxima especificada por el argumento `max_length` o
+    la longitud máxima aceptada por el modelo si no se proporciona `max_length` (`max_length=None`). Esto
+    truncará token por token, eliminando un token de la secuencia más larga en el par hasta alcanzar la longitud adecuada.
+  - `'only_second'`: trunca hasta una longitud máxima especificada por el argumento `max_length` o la longitud máxima
+    aceptada por el modelo si no se proporciona `max_length` (`max_length=None`). Esto solo truncará
+    la segunda oración de un par si se proporciona un par de secuencias (o un lote de pares de secuencias).
+  - `'only_first'`: trunca hasta una longitud máxima especificada por el argumento `max_length` o la longitud máxima
+    aceptada por el modelo si no se proporciona `max_length` (`max_length=None`). Esto solo truncará
+    la primera oración de un par si se proporciona un par de secuencias (o un lote de pares de secuencias).
+  - `False` o `'do_not_truncate'`: no se aplica truncamiento. Este es el comportamiento predeterminado.
+
+El argumento `max_length` controla la longitud del relleno y del truncamiento. Puede ser un número entero o `None`, en cuyo caso se establecerá automáticamente en la longitud máxima que el modelo puede aceptar. Si el modelo no tiene una longitud máxima de entrada específica, se desactiva el truncamiento o el relleno hasta `max_length`.
+
+La siguiente tabla resume la forma recomendada de configurar el relleno y el truncamiento. Si usas pares de secuencias de entrada en alguno de los siguientes ejemplos, puedes reemplazar `truncation=True` por una `ESTRATEGIA` seleccionada en
+`['only_first', 'only_second', 'longest_first']`, es decir, `truncation='only_second'` o `truncation='longest_first'` para controlar cómo se truncan ambas secuencias en el par, como se detalló anteriormente.
+
+| Truncation                           | Padding                             | Instrucción                                                                                 |
+|-----------------------------------------|--------------------------------------|---------------------------------------------------------------------------------------------|
+| sin truncamiento                       | sin relleno                          | `tokenizer(batch_sentences)`                                                           |
+|                                         | relleno hasta la longitud máxima del lote | `tokenizer(batch_sentences, padding=True)` o                                          |
+|                                         |                                      | `tokenizer(batch_sentences, padding='longest')`                                        |
+|                                         | relleno hasta la longitud máxima del modelo | `tokenizer(batch_sentences, padding='max_length')`                                     |
+|                                         | relleno hasta una longitud específica | `tokenizer(batch_sentences, padding='max_length', max_length=42)`                      |
+|                                         | relleno hasta un múltiplo de un valor | `tokenizer(batch_sentences, padding=True, pad_to_multiple_of=8)`                        |
+| truncamiento hasta la longitud máxima del modelo | sin relleno                          | `tokenizer(batch_sentences, truncation=True)` o                                       |
+|                                         |                                      | `tokenizer(batch_sentences, truncation=ESTRATEGIA)`                                      |
+|                                         | relleno hasta la longitud máxima del lote | `tokenizer(batch_sentences, padding=True, truncation=True)` o                         |
+|                                         |                                      | `tokenizer(batch_sentences, padding=True, truncation=ESTRATEGIA)`                        |
+|                                         | relleno hasta la longitud máxima del modelo | `tokenizer(batch_sentences, padding='max_length', truncation=True)` o                 |
+|                                         |                                      | `tokenizer(batch_sentences, padding='max_length', truncation=ESTRATEGIA)`                |
+|                                         | relleno hasta una longitud específica | No es posible                                                                                |
+| truncamiento hasta una longitud específica | sin relleno                          | `tokenizer(batch_sentences, truncation=True, max_length=42)` o                        |
+|                                         |                                      | `tokenizer(batch_sentences, truncation=ESTRATEGIA, max_length=42)`                       |
+|                                         | relleno hasta la longitud máxima del lote | `tokenizer(batch_sentences, padding=True, truncation=True, max_length=42)` o          |
+|                                         |                                      | `tokenizer(batch_sentences, padding=True, truncation=ESTRATEGIA, max_length=42)`         |
+|                                         | relleno hasta la longitud máxima del modelo | No es posible                                                                                |
+|                                         | relleno hasta una longitud específica | `tokenizer(batch_sentences, padding='max_length', truncation=True, max_length=42)` o  |
+|                                         |                                      | `tokenizer(batch_sentences, padding='max_length', truncation=ESTRATEGIA, max_length=42)` |

docs/source/es/perplexity.md

@@ -0,0 +1,116 @@
+<!--Copyright 2020 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.
+
+-->
+
+# Perplejidad de los modelos de longitud fija
+
+[[open-in-colab]]
+
+La perplejidad, perplexity en inglés (PPL), es una de las métricas más comunes para evaluar modelos de lenguaje. Antes de sumergirnos, debemos tener en cuenta que esta métrica se aplica específicamente a modelos de lenguaje clásicos (a veces llamados modelos autorregresivos o causales) y no está bien definida para modelos de lenguaje enmascarados como BERT (ver [resumen del modelo](model_summary)).
+
+La perplejidad se define como la media negativa exponenciada del log-likelihood de una secuencia. Si tenemos una secuencia tokenizada \\(X = (x_0, x_1, \dots, x_t)\\), entonces la perplejidad de \\(X\\) es,
+
+$$\text{PPL}(X) = \exp \left\{ {-\frac{1}{t}\sum_i^t \log p_\theta (x_i|x_{<i}) } \right\}$$
+
+donde \\(\log p_\theta (x_i|x_{<i})\\) es el log-likelihood del token i-ésimo condicionado a los tokens precedentes \\(x_{<i}\\) según nuestro modelo. De manera intuitiva, se puede pensar en esto como una evaluación de la capacidad del modelo para predecir de manera uniforme entre el conjunto de tokens especificados en un corpus. Es importante destacar que el procedimiento de tokenización tiene un impacto directo en la perplejidad de un modelo, lo cual siempre debe tenerse en cuenta al comparar diferentes modelos.
+
+Esto también es equivalente a la exponenciación de la entropía cruzada entre los datos y las predicciones del modelo. Para obtener más intuición sobre la perplejidad y su relación con los Bits Por Carácter (BPC) y la compresión de datos, echa un vistazo a esta [fantástica publicación en el blog de "The Gradient"](https://thegradient.pub/understanding-evaluation-metrics-for-language-models/).
+
+## Cálculo de PPL con modelos de longitud fija
+
+Si no estuviéramos limitados por el tamaño del contexto de un modelo, evaluaríamos la perplejidad (PPL) del modelo auto regresivamente factorizando una secuencia y condicionándonos en toda la subsecuencia precedente en cada paso, como se muestra a continuación.
+
+<img width="600" alt="Full decomposition of a sequence with unlimited context length" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/ppl_full.gif"/>
+
+Sin embargo, al trabajar con modelos aproximados, generalmente tenemos una restricción en la cantidad de tokens que el modelo puede procesar. La versión más grande de [GPT-2](model_doc/gpt2), por ejemplo, tiene una longitud fija de 1024 tokens, por lo que no podemos calcular \\(p_\theta(x_t|x_{<t})\\) directamente cuando \\(t\\) es mayor que 1024.
+
+En cambio, la secuencia se divide típicamente en subsecuencias iguales al tamaño máximo de entrada del modelo. Si el tamaño máximo de entrada, de un modelo es \\(k\\), entonces aproximamos la probabilidad de un token \\(x_t\\) condicionándonos solo en los \\(k-1\\) tokens que lo preceden en lugar de todo el contexto. Al evaluar la perplejidad del modelo en una secuencia, un enfoque tentador pero sub óptimo es dividir la secuencia en fragmentos independientes y sumar los log-likelihood descompuestos de cada segmento de manera independiente.
+
+<img width="600" alt="Suboptimal PPL not taking advantage of full available context" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/ppl_chunked.gif"/>
+
+Esto es rápido de calcular, ya que la perplejidad de cada segmento se puede calcular en un solo pase hacia adelante, pero sirve como una aproximación pobre de la perplejidad completamente factorizada y generalmente dará como resultado una PPL más alta (peor) porque el modelo tendrá menos contexto en la mayoría de los pasos de predicción.
+
+En cambio, la PPL de modelos de longitud fija debería evaluarse con una estrategia de ventana deslizante. Esto implica deslizar repetidamente la ventana de contexto para que el modelo tenga más contexto al hacer cada predicción.
+
+<img width="600" alt="Sliding window PPL taking advantage of all available context" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/ppl_sliding.gif"/>
+
+Esta es una aproximación más cercana a la verdadera descomposición de la probabilidad de la secuencia y generalmente dará como resultado una puntuación más favorable. La desventaja es que requiere un pase hacia adelante separado para cada token en el corpus. Un buen compromiso práctico es emplear una ventana deslizante estratificada, moviendo el contexto con pasos más grandes en lugar de deslizarse de 1 token a la vez. Esto permite que la computación avance mucho más rápido, mientras le da al modelo un contexto amplio para hacer
+predicciones en cada paso.
+
+## Ejemplo: Cálculo de la perplejidad con GPT-2 en 🤗 Transformers
+
+Demostremos este proceso con GPT-2.
+
+```python
+from transformers import GPT2LMHeadModel, GPT2TokenizerFast
+
+device = "cuda"
+model_id = "gpt2-large"
+model = GPT2LMHeadModel.from_pretrained(model_id).to(device)
+tokenizer = GPT2TokenizerFast.from_pretrained(model_id)
+```
+
+Carguemos el conjunto de datos WikiText-2 y evaluemos la perplejidad utilizando algunas estrategias de ventana deslizante diferentes. Dado que este conjunto de datos es pequeño y solo estamos realizando un pase hacia adelante sobre el conjunto, podemos cargar y codificar todo el conjunto de datos en la memoria.
+
+```python
+from datasets import load_dataset
+
+test = load_dataset("wikitext", "wikitext-2-raw-v1", split="test")
+encodings = tokenizer("\n\n".join(test["text"]), return_tensors="pt")
+```
+
+Con 🤗 Transformers, simplemente podemos pasar los `input_ids` como las `labels` a nuestro modelo, y la media negativa del log-likelihood para cada token se devuelve como la pérdida. Sin embargo, con nuestro enfoque de ventana deslizante, hay superposición en los tokens que pasamos al modelo en cada iteración. No queremos que el log-likelihood de los tokens que estamos tratando solo como contexto se incluya en nuestra pérdida, por lo que podemos establecer estos objetivos en `-100` para que se ignoren. El siguiente es un ejemplo de cómo podríamos hacer esto con un paso de `512`. Esto significa que el modelo tendrá al menos `512` tokens como contexto al calcular el log-likelihood condicional de cualquier token (siempre que haya `512` tokens precedentes disponibles para condicionar).
+
+```python
+import torch
+from tqdm import tqdm
+
+max_length = model.config.n_positions
+stride = 512
+seq_len = encodings.input_ids.size(1)
+
+nlls = []
+prev_end_loc = 0
+for begin_loc in tqdm(range(0, seq_len, stride)):
+    end_loc = min(begin_loc + max_length, seq_len)
+    trg_len = end_loc - prev_end_loc  # puede ser diferente del paso en el último bucle
+    input_ids = encodings.input_ids[:, begin_loc:end_loc].to(device)
+    target_ids = input_ids.clone()
+    target_ids[:, :-trg_len] = -100
+
+    with torch.no_grad():
+        outputs = model(input_ids, labels=target_ids)
+
+        # la pérdida se calcula utilizando CrossEntropyLoss, que promedia las etiquetas válidas
+        # N.B. el modelo solo calcula la pérdida sobre trg_len - 1 etiquetas, porque desplaza las etiqueta internamente
+        # a la izquierda por 1.
+        neg_log_likelihood = outputs.loss
+
+    nlls.append(neg_log_likelihood)
+
+    prev_end_loc = end_loc
+    if end_loc == seq_len:
+        break
+
+ppl = torch.exp(torch.stack(nlls).mean())
+```
+
+Ejecuta esto con la longitud de paso igual a la longitud máxima de entrada es equivalente a la estrategia sub óptima,
+sin ventana deslizante, que discutimos anteriormente. Cuanto menor sea el paso, más contexto tendrá el modelo para
+realizar cada predicción y, por lo general, mejor será la perplejidad informada.
+
+Cuando ejecutamos lo anterior con `stride = 1024`, es decir, sin superposición, la PPL resultante es `19.44`, que es
+aproximadamente la misma que la `19.93` informada en el artículo de GPT-2. Al utilizar `stride = 512` y, por lo tanto,
+emplear nuestra estrategia de ventana deslizante, esto disminuye a `16.45`. Esto no solo es una puntuación más favorable, sino que se calcula de una manera más cercana a la verdadera descomposición autorregresiva de la probabilidad de una secuencia.