Merge branch 'main' into flx_attn_support

remove unnecessary components
test v2batch w/ flex attn
2025-02-21 11:27:25 -05:00 · 2025-02-21 11:23:21 -05:00 · 2025-02-13 00:11:45 -05:00 · 2025-02-11 23:31:56 -05:00 · 2025-02-07 14:44:52 -05:00 · 2025-02-04 21:57:21 -05:00
464 changed files with 5588 additions and 20572 deletions
--- a/.coveragerc
+++ b/.coveragerc
@@ -1,14 +0,0 @@
-[run]
-source = axolotl
-omit =
-    */tests/*
-    setup.py
-
-[report]
-exclude_lines =
-    pragma: no cover
-    def __repr__
-    raise NotImplementedError
-    if __name__ == .__main__.:
-    pass
-    raise ImportError
--- a/.github/workflows/base.yml
+++ b/.github/workflows/base.yml
@@ -40,36 +40,6 @@ jobs:
            python_version: "3.11"
            pytorch: 2.6.0
            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
-          - cuda: "126"
-            cuda_version: 12.6.3
-            cudnn_version: ""
-            python_version: "3.11"
-            pytorch: 2.6.0
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
-          - cuda: "126"
-            cuda_version: 12.6.3
-            cudnn_version: ""
-            python_version: "3.11"
-            pytorch: 2.7.0
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
-          - cuda: "128"
-            cuda_version: 12.6.3
-            cudnn_version: ""
-            python_version: "3.11"
-            pytorch: 2.7.0
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
-          - cuda: "128"
-            cuda_version: 12.8.1
-            cudnn_version: ""
-            python_version: "3.11"
-            pytorch: nightly
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
-          - cuda: "128"
-            cuda_version: 12.8.1
-            cudnn_version: ""
-            python_version: "3.11"
-            pytorch: next
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
    steps:
      - name: Checkout
        uses: actions/checkout@v4
@@ -91,7 +61,7 @@ jobs:
        uses: docker/build-push-action@v4
        with:
          context: .
-          file: ${{ matrix.pytorch == 'nightly' && './docker/Dockerfile-base-nightly' || matrix.pytorch == 'next' && './docker/Dockerfile-base-next' || './docker/Dockerfile-base' }}
+          file: ./docker/Dockerfile-base
          push: ${{ github.event_name != 'pull_request' }}
          tags: ${{ steps.metadata.outputs.tags }}-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
          labels: ${{ steps.metadata.outputs.labels }}
--- a/.github/workflows/docs.yml
+++ b/.github/workflows/docs.yml
@@ -20,12 +20,9 @@ jobs:
          uses: actions/setup-python@v5
          with:
            python-version: '3.11'
-        - name: Install dependencies
+        - name: install dependencies
          run: |
-            python3 -m pip install jupyter quartodoc
-            python3 -m pip install -e . --no-deps
-        - name: Build autodoc
-          run: quartodoc build
+            python3 -m pip install jupyter
        - name: Publish to GitHub Pages (and render)
          uses: quarto-dev/quarto-actions/publish@v2
          with:
--- a/.github/workflows/main.yml
+++ b/.github/workflows/main.yml
@@ -24,18 +24,13 @@ jobs:
            cuda_version: 12.4.1
            python_version: "3.11"
            pytorch: 2.5.1
-            axolotl_extras:
+            axolotl_extras: vllm
+            is_latest: true
          - cuda: 124
            cuda_version: 12.4.1
            python_version: "3.11"
            pytorch: 2.6.0
-            axolotl_extras: vllm
-            is_latest: true
-          - cuda: 126
-            cuda_version: 12.6.3
-            python_version: "3.11"
-            pytorch: 2.7.0
-            axolotl_extras: vllm
+            axolotl_extras:
    runs-on: axolotl-gpu-runner
    steps:
      - name: Checkout
@@ -92,17 +87,7 @@ jobs:
            python_version: "3.11"
            pytorch: 2.5.1
            axolotl_extras:
-          - cuda: 124
-            cuda_version: 12.4.1
-            python_version: "3.11"
-            pytorch: 2.6.0
-            axolotl_extras:
            is_latest: true
-          - cuda: 126
-            cuda_version: 12.6.3
-            python_version: "3.11"
-            pytorch: 2.7.0
-            axolotl_extras:
    runs-on: axolotl-gpu-runner
    steps:
      - name: Checkout
@@ -148,7 +133,7 @@ jobs:
          - cuda: 124
            cuda_version: 12.4.1
            python_version: "3.11"
-            pytorch: 2.6.0
+            pytorch: 2.4.1
            axolotl_extras:
    runs-on: axolotl-gpu-runner
    steps:
--- a/.github/workflows/multi-gpu-e2e.yml
+++ b/.github/workflows/multi-gpu-e2e.yml
@@ -4,11 +4,6 @@ on:
  pull_request:
    paths:
      - 'tests/e2e/multigpu/*.py'
-      - 'requirements.txt'
-      - 'setup.py'
-      - 'pyproject.toml'
-      - '.github/workflows/multi-gpu-e2e.yml'
-      - 'src/axolotl/core/trainers/mixins/sequence_parallel.py'
  workflow_dispatch:
  schedule:
    - cron: '0 0 * * 1,4'  # Runs at 00:00 UTC every monday & thursday
@@ -25,13 +20,6 @@ jobs:
      fail-fast: false
      matrix:
        include:
-          - cuda: 124
-            cuda_version: 12.4.1
-            python_version: "3.11"
-            pytorch: 2.6.0
-            axolotl_extras: vllm
-            num_gpus: 2
-            nightly_build: "true"
          - cuda: 124
            cuda_version: 12.4.1
            python_version: "3.11"
@@ -43,13 +31,14 @@ jobs:
            cuda_version: 12.4.1
            python_version: "3.11"
            pytorch: 2.5.1
-            axolotl_extras:
+            axolotl_extras: vllm
            num_gpus: 2
            nightly_build: "true"
-          - cuda: 126
-            cuda_version: 12.6.3
+          - cuda: 124
+            cuda_version: 12.4.1
            python_version: "3.11"
-            pytorch: 2.7.0
+            pytorch: 2.6.0
+            # awaiting vllm#12721
            axolotl_extras:
            num_gpus: 2
            nightly_build: "true"
@@ -75,7 +64,6 @@ jobs:
          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
          echo "NIGHTLY_BUILD=${{ matrix.nightly_build }}" >> $GITHUB_ENV
-          echo "CODECOV_TOKEN=${{ secrets.CODECOV_TOKEN }}" >> $GITHUB_ENV
      - name: Run tests job on Modal
        run: |
          modal run cicd.multigpu
--- a/.github/workflows/nightlies.yml
+++ b/.github/workflows/nightlies.yml
@@ -80,11 +80,6 @@ jobs:
            python_version: "3.11"
            pytorch: 2.5.1
            axolotl_extras:
-          - cuda: 124
-            cuda_version: 12.4.1
-            python_version: "3.11"
-            pytorch: 2.6.0
-            axolotl_extras:
    runs-on: axolotl-gpu-runner
    steps:
      - name: Checkout
--- a/.github/workflows/precommit-autoupdate.yml
+++ b/.github/workflows/precommit-autoupdate.yml
@@ -1,49 +0,0 @@
-name: Pre-commit auto-update
-
-on:
-  schedule:
-    - cron: '0 0 * * 0'  # Run weekly
-  workflow_dispatch:  # Manual kickoff
-
-jobs:
-  auto-update:
-    runs-on: ubuntu-latest
-    permissions:
-      contents: write
-      pull-requests: write
-    steps:
-      - uses: actions/checkout@v4
-
-      - uses: actions/setup-python@v5
-        with:
-          python-version: '3.11'
-
-      - name: Update pre-commit hooks
-        id: update
-        run: |
-          pip install pre-commit
-          pre-commit autoupdate
-          if [[ -n $(git status --porcelain) ]]; then
-            echo "changes=true" >> $GITHUB_OUTPUT
-            git diff .pre-commit-config.yaml > pre-commit-update.diff
-          fi
-
-      - name: Create Pull Request
-        if: steps.update.outputs.changes == 'true'
-        uses: peter-evans/create-pull-request@v6
-        with:
-          token: ${{ secrets.GITHUB_TOKEN }}
-          branch: update/pre-commit-hooks
-          delete-branch: true
-          title: "chore: update pre-commit hooks"
-          commit-message: "chore: update pre-commit hooks"
-          body: |
-            Automated PR to update pre-commit hooks to their latest versions.
-
-            <details>
-            <summary>Changes:</summary>
-
-            ```diff
-            ${{ steps.update.outputs.diff }}
-            ```
-            </details>
--- a/.github/workflows/preview-docs.yml
+++ b/.github/workflows/preview-docs.yml
@@ -1,55 +0,0 @@
-name: Preview
-on:
-  workflow_dispatch:
-  pull_request:
-    types: [opened, synchronize, reopened]
-
-permissions:
-  checks: write
-  contents: write
-  deployments: write
-  issues: write
-  discussions: write
-  pages: write
-  pull-requests: write
-  statuses: write
-
-jobs:
-  preview:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Check out repository
-        uses: actions/checkout@v4
-
-      - name: Set up Quarto
-        uses: quarto-dev/quarto-actions/setup@v2
-
-      - name: Setup Python
-        uses: actions/setup-python@v5
-        with:
-          python-version: '3.11'
-
-      - name: Install dependencies
-        run: |
-          python3 -m pip install jupyter quartodoc
-          python3 -m pip install -e . --no-deps
-
-      - name: Build autodoc
-        run: quartodoc build
-
-      - name: Quarto render
-        run: quarto render
-
-      - name: Netlify Publish
-        uses: nwtgck/actions-netlify@v3.0
-        with:
-          publish-dir: './_site'
-          enable-pull-request-comment: true
-          enable-github-deployment: true
-          github-token: ${{ secrets.GITHUB_TOKEN }}
-          deploy-message: "Deployed On Netlify"
-          github-deployment-environment: 'preview'
-          github-deployment-description: 'Preview Deployment'
-        env:
-          NETLIFY_AUTH_TOKEN: ${{ secrets.NETLIFY_AUTH_TOKEN }}
-          NETLIFY_SITE_ID: ${{ secrets.NETLIFY_SITE_ID }}
--- a/.github/workflows/pypi.yml
+++ b/.github/workflows/pypi.yml
@@ -40,7 +40,7 @@ jobs:

      - name: Install dependencies
        run: |
-          pip3 install wheel packaging==23.2
+          pip3 install wheel packaging
          pip3 install --no-build-isolation -e .
          pip3 install -r requirements-dev.txt -r requirements-tests.txt

--- a/.github/workflows/tests-nightly.yml
+++ b/.github/workflows/tests-nightly.yml
@@ -33,15 +33,6 @@ jobs:
      - name: Check out repository code
        uses: actions/checkout@v4

-      - name: Restore HF cache
-        id: hf-cache-restore
-        uses: actions/cache/restore@v4
-        with:
-          path: |
-            /home/runner/.cache/huggingface/hub/datasets--*
-            /home/runner/.cache/huggingface/hub/models--*
-          key: ${{ runner.os }}-hf-hub-cache-v2
-
      - name: Setup Python
        uses: actions/setup-python@v5
        with:
@@ -51,11 +42,11 @@ jobs:
      - name: upgrade pip
        run: |
          pip3 install --upgrade pip
-          pip3 install --upgrade packaging==23.2 setuptools==75.8.0 wheel
+          pip3 install --upgrade packaging setuptools wheel

      - name: Install PyTorch
        run: |
-          pip3 install torch==${{ matrix.pytorch_version }}
+          pip3 install torch==${{ matrix.pytorch_version }} --index-url https://download.pytorch.org/whl/cpu

      - name: Update requirements.txt
        run: |
@@ -67,7 +58,8 @@ jobs:

      - name: Install dependencies
        run: |
-          pip3 show torch
+          pip3 install --upgrade pip
+          pip3 install --upgrade packaging
          pip3 install --no-build-isolation -U -e .
          python scripts/unsloth_install.py | sh
          python scripts/cutcrossentropy_install.py | sh
@@ -81,15 +73,10 @@ jobs:
        run: |
          axolotl --help

-      - name: Pre-Download dataset fixture
-        run: |
-          huggingface-cli download --repo-type=dataset axolotl-ai-internal/axolotl-oss-dataset-fixtures
-
      - name: Run tests
        run: |
-          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/
-          pytest -v tests/patched/
-          pytest -v tests/cli/
+          pytest -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ tests/
+          pytest tests/patched/

      - name: cleanup pip cache
        run: |
@@ -147,7 +134,6 @@ jobs:
          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
          echo "NIGHTLY_BUILD=${{ matrix.nightly_build }}" >> $GITHUB_ENV
-          echo "CODECOV_TOKEN=${{ secrets.CODECOV_TOKEN }}" >> $GITHUB_ENV
      - name: Run tests job on Modal
        run: |
-          modal run cicd.e2e_tests
+          modal run cicd.tests
--- a/.github/workflows/tests.yml
+++ b/.github/workflows/tests.yml
@@ -49,7 +49,7 @@ jobs:
      max-parallel: 2
      matrix:
        python_version: ["3.11"]
-        pytorch_version: ["2.4.1", "2.5.1", "2.6.0", "2.7.0"]
+        pytorch_version: ["2.4.1", "2.5.1", "2.6.0"]
    timeout-minutes: 20

    steps:
@@ -63,7 +63,7 @@ jobs:
          path: |
            /home/runner/.cache/huggingface/hub/datasets--*
            /home/runner/.cache/huggingface/hub/models--*
-          key: ${{ runner.os }}-hf-hub-cache-v2
+          key: ${{ runner.os }}-hf-hub-cache-${{ hashFiles('**/conftest.py') }}

      - name: Setup Python
        uses: actions/setup-python@v5
@@ -74,7 +74,7 @@ jobs:
      - name: upgrade pip
        run: |
          pip3 install --upgrade pip
-          pip3 install --upgrade packaging==23.2 setuptools==75.8.0 wheel
+          pip3 install --upgrade packaging setuptools wheel

      - name: Install PyTorch
        run: |
@@ -96,23 +96,10 @@ jobs:
        run: |
          axolotl --help

-      - name: Pre-Download dataset fixture
-        run: |
-          huggingface-cli download --repo-type=dataset axolotl-ai-internal/axolotl-oss-dataset-fixtures
-
      - name: Run tests
        run: |
-          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/ --cov=axolotl --cov-report=xml
-          pytest -v tests/patched/ --cov=axolotl --cov-append --cov-report=xml
-          pytest -v tests/cli/ --cov=axolotl --cov-append --cov-report=xml
-
-      - name: Upload coverage to Codecov
-        uses: codecov/codecov-action@v5
-        with:
-          token: ${{ secrets.CODECOV_TOKEN }}
-          files: ./coverage.xml
-          flags: unittests,pytorch-${{ matrix.pytorch_version }}
-          fail_ci_if_error: false
+          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ tests/
+          pytest -v tests/patched/

      - name: cleanup pip cache
        run: |
@@ -149,7 +136,7 @@ jobs:
          path: |
            /home/runner/.cache/huggingface/hub/datasets--*
            /home/runner/.cache/huggingface/hub/models--*
-          key: ${{ runner.os }}-hf-hub-cache-v2
+          key: ${{ runner.os }}-hf-hub-cache-${{ hashFiles('**/conftest.py') }}

      - name: Setup Python
        uses: actions/setup-python@v5
@@ -160,7 +147,7 @@ jobs:
      - name: upgrade pip
        run: |
          pip3 install --upgrade pip
-          pip3 install --upgrade packaging==23.2 setuptools==75.8.0 setuptools_scm build wheel
+          pip3 install --upgrade packaging setuptools setuptools_scm build wheel

      - name: Install PyTorch
        run: |
@@ -183,14 +170,10 @@ jobs:
        run: |
          axolotl --help

-      - name: Show HF cache
-        run: huggingface-cli scan-cache
-
      - name: Run tests
        run: |
-          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ --ignore=tests/cli/ tests/
+          pytest -v -n8 --dist loadfile --ignore=tests/e2e/ --ignore=tests/patched/ tests/
          pytest -v tests/patched/
-          pytest -v tests/cli/

      - name: cleanup pip cache
        run: |
@@ -219,7 +202,7 @@ jobs:
          - cuda: 124
            cuda_version: 12.4.1
            python_version: "3.11"
-            pytorch: 2.6.0
+            pytorch: 2.5.1
            num_gpus: 1
            axolotl_extras: vllm
    steps:
@@ -242,10 +225,9 @@ jobs:
          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
          echo "MODAL_IMAGE_BUILDER_VERSION=2024.10" >> $GITHUB_ENV
          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
-          echo "CODECOV_TOKEN=${{ secrets.CODECOV_TOKEN }}" >> $GITHUB_ENV
      - name: Run tests job on Modal
        run: |
-          modal run cicd.e2e_tests
+          modal run cicd.tests

  docker-e2e-tests:
    if: github.repository_owner == 'axolotl-ai-cloud'
@@ -267,13 +249,7 @@ jobs:
          - cuda: 124
            cuda_version: 12.4.1
            python_version: "3.11"
-            pytorch: 2.5.1
-            num_gpus: 1
-            axolotl_extras:
-          - cuda: 126
-            cuda_version: 12.6.3
-            python_version: "3.11"
-            pytorch: 2.7.0
+            pytorch: 2.6.0
            num_gpus: 1
            axolotl_extras:
    steps:
@@ -296,7 +272,6 @@ jobs:
          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
          echo "MODAL_IMAGE_BUILDER_VERSION=2024.10" >> $GITHUB_ENV
          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
-          echo "CODECOV_TOKEN=${{ secrets.CODECOV_TOKEN }}" >> $GITHUB_ENV
      - name: Run tests job on Modal
        run: |
-          modal run cicd.e2e_tests
+          modal run cicd.tests
--- a/.gitignore
+++ b/.gitignore
@@ -181,10 +181,6 @@ prepared-datasets/
 submit.sh
 *.out*

-# Quartodoc generated files
-objects.json
-site_libs/
-
 typings/
 out/

--- a/.isort.cfg
+++ b/.isort.cfg
@@ -1,4 +1,3 @@
 [settings]
 profile=black
 known_third_party=wandb,comet_ml
-known_local_folder=src,tests
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -3,7 +3,7 @@ default_language_version:

 repos:
 -   repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v5.0.0
+    rev: v4.4.0
    hooks:
    -   id: check-yaml
    -   id: end-of-file-fixer
@@ -11,23 +11,23 @@ repos:
    -   id: no-commit-to-branch
        args: ['--branch', 'main']
 -   repo: https://github.com/psf/black
-    rev: 25.1.0
+    rev: 23.3.0
    hooks:
    -   id: black
 -   repo: https://github.com/pycqa/isort
-    rev: 6.0.1
+    rev: 5.12.0
    hooks:
      - id: isort
 -   repo: https://github.com/PyCQA/flake8
-    rev: 7.1.2
+    rev: 6.1.0
    hooks:
    - id: flake8
-   repo: https://github.com/pylint-dev/pylint
-    rev: v3.3.6
+-   repo: https://github.com/PyCQA/pylint
+    rev: v3.3.0
    hooks:
    - id: pylint
 -   repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.15.0
+    rev: v1.3.0
    hooks:
    - id: mypy
      additional_dependencies:
@@ -36,7 +36,7 @@ repos:
            'pydantic>=2.5.3',
        ]
 -   repo: https://github.com/PyCQA/bandit
-    rev: 1.8.3
+    rev: 1.7.5
    hooks:
    -   id: bandit
        args: [
--- a/.runpod/.gitignore
+++ b/.runpod/.gitignore
@@ -1,161 +0,0 @@
-# Byte-compiled / optimized / DLL files
-__pycache__/
-*.py[cod]
-*$py.class
-
-# C extensions
-*.so
-
-# Distribution / packaging
-.Python
-build/
-develop-eggs/
-dist/
-downloads/
-eggs/
-.eggs/
-lib/
-lib64/
-parts/
-sdist/
-var/
-wheels/
-share/python-wheels/
-*.egg-info/
-.installed.cfg
-*.egg
-MANIFEST
-
-# PyInstaller
-#  Usually these files are written by a python script from a template
-#  before PyInstaller builds the exe, so as to inject date/other infos into it.
-*.manifest
-*.spec
-
-# Installer logs
-pip-log.txt
-pip-delete-this-directory.txt
-
-# Unit test / coverage reports
-htmlcov/
-.tox/
-.nox/
-.coverage
-.coverage.*
-.cache
-nosetests.xml
-coverage.xml
-*.cover
-*.py,cover
-.hypothesis/
-.pytest_cache/
-cover/
-
-# Translations
-*.mo
-*.pot
-
-# Django stuff:
-*.log
-local_settings.py
-db.sqlite3
-db.sqlite3-journal
-
-# Flask stuff:
-instance/
-.webassets-cache
-
-# Scrapy stuff:
-.scrapy
-
-# Sphinx documentation
-docs/_build/
-
-# PyBuilder
-.pybuilder/
-target/
-
-# Jupyter Notebook
-.ipynb_checkpoints
-
-# IPython
-profile_default/
-ipython_config.py
-
-# pyenv
-#   For a library or package, you might want to ignore these files since the code is
-#   intended to run in multiple environments; otherwise, check them in:
-# .python-version
-
-# pipenv
-#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
-#   However, in case of collaboration, if having platform-specific dependencies or dependencies
-#   having no cross-platform support, pipenv may install dependencies that don't work, or not
-#   install all needed dependencies.
-#Pipfile.lock
-
-# poetry
-#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
-#   This is especially recommended for binary packages to ensure reproducibility, and is more
-#   commonly ignored for libraries.
-#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
-#poetry.lock
-
-# pdm
-#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
-#pdm.lock
-#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
-#   in version control.
-#   https://pdm.fming.dev/#use-with-ide
-.pdm.toml
-
-# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
-__pypackages__/
-
-# Celery stuff
-celerybeat-schedule
-celerybeat.pid
-
-# SageMath parsed files
-*.sage.py
-
-# Environments
-.env
-.venv
-env/
-venv/
-ENV/
-env.bak/
-venv.bak/
-
-# Spyder project settings
-.spyderproject
-.spyproject
-
-# Rope project settings
-.ropeproject
-
-# mkdocs documentation
-/site
-
-# mypy
-.mypy_cache/
-.dmypy.json
-dmypy.json
-
-# Pyre type checker
-.pyre/
-
-# pytype static type analyzer
-.pytype/
-
-# Cython debug symbols
-cython_debug/
-
-# PyCharm
-#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
-#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
-#  and can be added to the global gitignore or merged into this file.  For a more nuclear
-#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
-pod/scripts/config.yaml
--- a/.runpod/Dockerfile
+++ b/.runpod/Dockerfile
@@ -1,18 +0,0 @@
-FROM axolotlai/axolotl-cloud:main-py3.11-cu124-2.6.0
-
-COPY .runpod/requirements.txt /requirements.txt
-RUN --mount=type=cache,target=/root/.cache/pip \
-    python3 -m pip install --upgrade pip && \
-    python3 -m pip install --upgrade -r /requirements.txt
-
-# Environment settings
-ARG BASE_VOLUME="/runpod-volume"
-ENV BASE_VOLUME=$BASE_VOLUME
-ENV HF_DATASETS_CACHE="${BASE_VOLUME}/huggingface-cache/datasets"
-ENV HUGGINGFACE_HUB_CACHE="${BASE_VOLUME}/huggingface-cache/hub"
-ENV TRANSFORMERS_CACHE="${BASE_VOLUME}/huggingface-cache/hub"
-
-COPY .runpod/src /src
-
-WORKDIR /src
-CMD ["python3", "/src/handler.py"]
--- a/.runpod/README.md
+++ b/.runpod/README.md
@@ -1,335 +0,0 @@
-<h1>LLM Post Training- Full fine-tune, LoRA, QLoRa etc. Llama/Mistral/Gemma and more</h1>
-
-# Configuration Options
-
-This document outlines all available configuration options for training models. The configuration can be provided as a JSON request.
-
-## Usage
-
-You can use these configuration Options:
-
-1. As a JSON request body:
-
-```json
-{
-  "input": {
-    "user_id": "user",
-    "model_id": "model-name",
-    "run_id": "run-id",
-    "credentials": {
-      "wandb_api_key": "", # add your Weights & biases key. TODO:  you will be able to set this in Enviornment variables.
-      "hf_token": "", # add your HF_token. TODO:  you will be able to set this in Enviornment variables.
-    },
-    "args": {
-      "base_model": "NousResearch/Llama-3.2-1B",
-      // ... other options
-    }
-  }
-}
-```
-
-## Configuration Options
-
-### Model Configuration
-
-| Option              | Description                                                                                   | Default              |
-| ------------------- | --------------------------------------------------------------------------------------------- | -------------------- |
-| `base_model`        | Path to the base model (local or HuggingFace)                                                 | Required             |
-| `base_model_config` | Configuration path for the base model                                                         | Same as base_model   |
-| `revision_of_model` | Specific model revision from HuggingFace hub                                                  | Latest               |
-| `tokenizer_config`  | Custom tokenizer configuration path                                                           | Optional             |
-| `model_type`        | Type of model to load                                                                         | AutoModelForCausalLM |
-| `tokenizer_type`    | Type of tokenizer to use                                                                      | AutoTokenizer        |
-| `hub_model_id`      | Repository ID where the model will be pushed on Hugging Face Hub (format: username/repo-name) | Optional             |
-
-## Model Family Identification
-
-| Option                     | Default | Description                    |
-| -------------------------- | ------- | ------------------------------ |
-| `is_falcon_derived_model`  | `false` | Whether model is Falcon-based  |
-| `is_llama_derived_model`   | `false` | Whether model is LLaMA-based   |
-| `is_qwen_derived_model`    | `false` | Whether model is Qwen-based    |
-| `is_mistral_derived_model` | `false` | Whether model is Mistral-based |
-
-## Model Configuration Overrides
-
-| Option                                          | Default    | Description                        |
-| ----------------------------------------------- | ---------- | ---------------------------------- |
-| `overrides_of_model_config.rope_scaling.type`   | `"linear"` | RoPE scaling type (linear/dynamic) |
-| `overrides_of_model_config.rope_scaling.factor` | `1.0`      | RoPE scaling factor                |
-
-### Model Loading Options
-
-| Option         | Description                   | Default |
-| -------------- | ----------------------------- | ------- |
-| `load_in_8bit` | Load model in 8-bit precision | false   |
-| `load_in_4bit` | Load model in 4-bit precision | false   |
-| `bf16`         | Use bfloat16 precision        | false   |
-| `fp16`         | Use float16 precision         | false   |
-| `tf32`         | Use tensor float 32 precision | false   |
-
-## Memory and Device Settings
-
-| Option             | Default   | Description             |
-| ------------------ | --------- | ----------------------- |
-| `gpu_memory_limit` | `"20GiB"` | GPU memory limit        |
-| `lora_on_cpu`      | `false`   | Load LoRA on CPU        |
-| `device_map`       | `"auto"`  | Device mapping strategy |
-| `max_memory`       | `null`    | Max memory per device   |
-
-## Training Hyperparameters
-
-| Option                        | Default   | Description                 |
-| ----------------------------- | --------- | --------------------------- |
-| `gradient_accumulation_steps` | `1`       | Gradient accumulation steps |
-| `micro_batch_size`            | `2`       | Batch size per GPU          |
-| `eval_batch_size`             | `null`    | Evaluation batch size       |
-| `num_epochs`                  | `4`       | Number of training epochs   |
-| `warmup_steps`                | `100`     | Warmup steps                |
-| `warmup_ratio`                | `0.05`    | Warmup ratio                |
-| `learning_rate`               | `0.00003` | Learning rate               |
-| `lr_quadratic_warmup`         | `false`   | Quadratic warmup            |
-| `logging_steps`               | `null`    | Logging frequency           |
-| `eval_steps`                  | `null`    | Evaluation frequency        |
-| `evals_per_epoch`             | `null`    | Evaluations per epoch       |
-| `save_strategy`               | `"epoch"` | Checkpoint saving strategy  |
-| `save_steps`                  | `null`    | Saving frequency            |
-| `saves_per_epoch`             | `null`    | Saves per epoch             |
-| `save_total_limit`            | `null`    | Maximum checkpoints to keep |
-| `max_steps`                   | `null`    | Maximum training steps      |
-
-### Dataset Configuration
-
-```yaml
-datasets:
-  - path: vicgalle/alpaca-gpt4 # HuggingFace dataset or TODO: You will be able to add the local path.
-    type: alpaca # Format type (alpaca, gpteacher, oasst, etc.)
-    ds_type: json # Dataset type
-    data_files: path/to/data # Source data files
-    train_on_split: train # Dataset split to use
-```
-
-## Chat Template Settings
-
-| Option                   | Default                          | Description            |
-| ------------------------ | -------------------------------- | ---------------------- |
-| `chat_template`          | `"tokenizer_default"`            | Chat template type     |
-| `chat_template_jinja`    | `null`                           | Custom Jinja template  |
-| `default_system_message` | `"You are a helpful assistant."` | Default system message |
-
-## Dataset Processing
-
-| Option                        | Default                    | Description                       |
-| ----------------------------- | -------------------------- | --------------------------------- |
-| `dataset_prepared_path`       | `"data/last_run_prepared"` | Path for prepared dataset         |
-| `push_dataset_to_hub`         | `""`                       | Push dataset to HF hub            |
-| `dataset_processes`           | `4`                        | Number of preprocessing processes |
-| `dataset_keep_in_memory`      | `false`                    | Keep dataset in memory            |
-| `shuffle_merged_datasets`     | `true`                     | Shuffle merged datasets           |
-| `dataset_exact_deduplication` | `true`                     | Deduplicate datasets              |
-
-## LoRA Configuration
-
-| Option                     | Default                | Description                    |
-| -------------------------- | ---------------------- | ------------------------------ |
-| `adapter`                  | `"lora"`               | Adapter type (lora/qlora)      |
-| `lora_model_dir`           | `""`                   | Directory with pretrained LoRA |
-| `lora_r`                   | `8`                    | LoRA attention dimension       |
-| `lora_alpha`               | `16`                   | LoRA alpha parameter           |
-| `lora_dropout`             | `0.05`                 | LoRA dropout                   |
-| `lora_target_modules`      | `["q_proj", "v_proj"]` | Modules to apply LoRA          |
-| `lora_target_linear`       | `false`                | Target all linear modules      |
-| `peft_layers_to_transform` | `[]`                   | Layers to transform            |
-| `lora_modules_to_save`     | `[]`                   | Modules to save                |
-| `lora_fan_in_fan_out`      | `false`                | Fan in/out structure           |
-
-## Optimization Settings
-
-| Option                    | Default | Description                |
-| ------------------------- | ------- | -------------------------- |
-| `train_on_inputs`         | `false` | Train on input prompts     |
-| `group_by_length`         | `false` | Group by sequence length   |
-| `gradient_checkpointing`  | `false` | Use gradient checkpointing |
-| `early_stopping_patience` | `3`     | Early stopping patience    |
-
-## Learning Rate Scheduling
-
-| Option                     | Default    | Description          |
-| -------------------------- | ---------- | -------------------- |
-| `lr_scheduler`             | `"cosine"` | Scheduler type       |
-| `lr_scheduler_kwargs`      | `{}`       | Scheduler parameters |
-| `cosine_min_lr_ratio`      | `null`     | Minimum LR ratio     |
-| `cosine_constant_lr_ratio` | `null`     | Constant LR ratio    |
-| `lr_div_factor`            | `null`     | LR division factor   |
-
-## Optimizer Settings
-
-| Option                 | Default      | Description         |
-| ---------------------- | ------------ | ------------------- |
-| `optimizer`            | `"adamw_hf"` | Optimizer choice    |
-| `optim_args`           | `{}`         | Optimizer arguments |
-| `optim_target_modules` | `[]`         | Target modules      |
-| `weight_decay`         | `null`       | Weight decay        |
-| `adam_beta1`           | `null`       | Adam beta1          |
-| `adam_beta2`           | `null`       | Adam beta2          |
-| `adam_epsilon`         | `null`       | Adam epsilon        |
-| `max_grad_norm`        | `null`       | Gradient clipping   |
-
-## Attention Implementations
-
-| Option                     | Default | Description                   |
-| -------------------------- | ------- | ----------------------------- |
-| `flash_optimum`            | `false` | Use better transformers       |
-| `xformers_attention`       | `false` | Use xformers                  |
-| `flash_attention`          | `false` | Use flash attention           |
-| `flash_attn_cross_entropy` | `false` | Flash attention cross entropy |
-| `flash_attn_rms_norm`      | `false` | Flash attention RMS norm      |
-| `flash_attn_fuse_qkv`      | `false` | Fuse QKV operations           |
-| `flash_attn_fuse_mlp`      | `false` | Fuse MLP operations           |
-| `sdp_attention`            | `false` | Use scaled dot product        |
-| `s2_attention`             | `false` | Use shifted sparse attention  |
-
-## Tokenizer Modifications
-
-| Option           | Default | Description                  |
-| ---------------- | ------- | ---------------------------- |
-| `special_tokens` | -       | Special tokens to add/modify |
-| `tokens`         | `[]`    | Additional tokens            |
-
-## Distributed Training
-
-| Option                  | Default | Description           |
-| ----------------------- | ------- | --------------------- |
-| `fsdp`                  | `null`  | FSDP configuration    |
-| `fsdp_config`           | `null`  | FSDP config options   |
-| `deepspeed`             | `null`  | Deepspeed config path |
-| `ddp_timeout`           | `null`  | DDP timeout           |
-| `ddp_bucket_cap_mb`     | `null`  | DDP bucket capacity   |
-| `ddp_broadcast_buffers` | `null`  | DDP broadcast buffers |
-
-<details>
-<summary><h3>Example Configuration Request:</h3></summary>
-
-Here's a complete example for fine-tuning a LLaMA model using LoRA:
-
-```json
-{
-  "input": {
-    "user_id": "user",
-    "model_id": "llama-test",
-    "run_id": "test-run",
-    "credentials": {
-      "wandb_api_key": "",
-      "hf_token": ""
-    },
-    "args": {
-      "base_model": "NousResearch/Llama-3.2-1B",
-      "load_in_8bit": false,
-      "load_in_4bit": false,
-      "strict": false,
-      "datasets": [
-        {
-          "path": "teknium/GPT4-LLM-Cleaned",
-          "type": "alpaca"
-        }
-      ],
-      "dataset_prepared_path": "last_run_prepared",
-      "val_set_size": 0.1,
-      "output_dir": "./outputs/lora-out",
-      "adapter": "lora",
-      "sequence_len": 2048,
-      "sample_packing": true,
-      "eval_sample_packing": true,
-      "pad_to_sequence_len": true,
-      "lora_r": 16,
-      "lora_alpha": 32,
-      "lora_dropout": 0.05,
-      "lora_target_modules": [
-        "gate_proj",
-        "down_proj",
-        "up_proj",
-        "q_proj",
-        "v_proj",
-        "k_proj",
-        "o_proj"
-      ],
-      "gradient_accumulation_steps": 2,
-      "micro_batch_size": 2,
-      "num_epochs": 1,
-      "optimizer": "adamw_8bit",
-      "lr_scheduler": "cosine",
-      "learning_rate": 0.0002,
-      "train_on_inputs": false,
-      "group_by_length": false,
-      "bf16": "auto",
-      "tf32": false,
-      "gradient_checkpointing": true,
-      "logging_steps": 1,
-      "flash_attention": true,
-      "loss_watchdog_threshold": 5,
-      "loss_watchdog_patience": 3,
-      "warmup_steps": 10,
-      "evals_per_epoch": 4,
-      "saves_per_epoch": 1,
-      "weight_decay": 0,
-      "hub_model_id": "runpod/llama-fr-lora",
-      "wandb_name": "test-run-1",
-      "wandb_project": "test-run-1",
-      "wandb_entity": "axo-test",
-      "special_tokens": {
-        "pad_token": "<|end_of_text|>"
-      }
-    }
-  }
-}
-```
-
-</details>
-
-### Advanced Features
-
-#### Wandb Integration
-
- `wandb_project`: Project name for Weights & Biases
- `wandb_entity`: Team name in W&B
- `wandb_watch`: Monitor model with W&B
- `wandb_name`: Name of the W&B run
- `wandb_run_id`: ID for the W&B run
-
-#### Performance Optimization
-
- `sample_packing`: Enable efficient sequence packing
- `eval_sample_packing`: Use sequence packing during evaluation
- `torch_compile`: Enable PyTorch 2.0 compilation
- `flash_attention`: Use Flash Attention implementation
- `xformers_attention`: Use xFormers attention implementation
-
-### Available Optimizers
-
-The following optimizers are supported:
-
- `adamw_hf`: HuggingFace's AdamW implementation
- `adamw_torch`: PyTorch's AdamW
- `adamw_torch_fused`: Fused AdamW implementation
- `adamw_torch_xla`: XLA-optimized AdamW
- `adamw_apex_fused`: NVIDIA Apex fused AdamW
- `adafactor`: Adafactor optimizer
- `adamw_anyprecision`: Anyprecision AdamW
- `adamw_bnb_8bit`: 8-bit AdamW from bitsandbytes
- `lion_8bit`: 8-bit Lion optimizer
- `lion_32bit`: 32-bit Lion optimizer
- `sgd`: Stochastic Gradient Descent
- `adagrad`: Adagrad optimizer
-
-## Notes
-
- Set `load_in_8bit: true` or `load_in_4bit: true` for memory-efficient training
- Enable `flash_attention: true` for faster training on modern GPUs
- Use `gradient_checkpointing: true` to reduce memory usage
- Adjust `micro_batch_size` and `gradient_accumulation_steps` based on your GPU memory
-
-For more detailed information, please refer to the [documentation](https://axolotl-ai-cloud.github.io/axolotl/docs/config.html).
-
-### Errors:
-
- if you face any issues with the Flash Attention-2, Delete yoor worker and Re-start.
--- a/.runpod/hub.json
+++ b/.runpod/hub.json
@@ -1,93 +0,0 @@
-{
-  "title": "Axolotl Fine-Tuning",
-  "description": "Serverless fine-tuning of open-source LLMs with Axolotl. Supports LoRA, QLoRA, DPO, and more using Hugging Face models and datasets.",
-  "type": "serverless",
-  "category": "language",
-  "iconUrl": "https://avatars.githubusercontent.com/u/167502477",
-  "config": {
-    "runsOn": "GPU",
-    "containerDiskInGb": 200,
-    "gpuCount": 1,
-    "allowedCudaVersions": [
-      "12.8",
-      "12.7",
-      "12.6",
-      "12.5",
-      "12.4"
-    ],
-    "presets": [],
-    "env": [
-      {
-        "key": "TOKENIZER",
-        "input": {
-          "name": "Tokenizer",
-          "type": "string",
-          "description": "Name or path of the Hugging Face tokenizer to use.",
-          "default": "",
-          "advanced": true
-        }
-      },
-      {
-        "key": "MAX_NUM_SEQS",
-        "input": {
-          "name": "Max Num Seqs",
-          "type": "number",
-          "description": "Maximum number of sequences per iteration.",
-          "default": 256,
-          "advanced": true
-        }
-      },
-      {
-        "key": "DISABLE_LOG_STATS",
-        "input": {
-          "name": "Disable Log Stats",
-          "type": "boolean",
-          "description": "Disable logging statistics.",
-          "default": false,
-          "trueValue": "true",
-          "falseValue": "false"
-        }
-      },
-      {
-        "key": "LOAD_FORMAT",
-        "input": {
-          "name": "Load Format",
-          "type": "string",
-          "description": "The format of the model weights to load.",
-          "default": "auto",
-          "options": [
-            {
-              "label": "auto",
-              "value": "auto"
-            },
-            {
-              "label": "pt",
-              "value": "pt"
-            },
-            {
-              "label": "safetensors",
-              "value": "safetensors"
-            },
-            {
-              "label": "npcache",
-              "value": "npcache"
-            },
-            {
-              "label": "dummy",
-              "value": "dummy"
-            },
-            {
-              "label": "tensorizer",
-              "value": "tensorizer"
-            },
-            {
-              "label": "bitsandbytes",
-              "value": "bitsandbytes"
-            }
-          ],
-          "advanced": true
-        }
-      }
-    ]
-  }
-}
--- a/.runpod/requirements.txt
+++ b/.runpod/requirements.txt
@@ -1,7 +0,0 @@
-# Required Python packages get listed here, one per line.
-# Reccomended to lock the version number to avoid unexpected changes.
-
-# You can also install packages from a git repository, e.g.:
-# git+https://github.com/runpod/runpod-python.git
-# To learn more, see https://pip.pypa.io/en/stable/reference/requirements-file-format/
-runpod~=1.7.0
--- a/.runpod/src/config/config.yaml
+++ b/.runpod/src/config/config.yaml
@@ -1,577 +0,0 @@
-# # This is the huggingface model that contains *.pt, *.safetensors, or *.bin files
-# # This can also be a relative path to a model on disk
-# base_model: ./llama-7b-hf
-# # You can specify an ignore pattern if the model repo contains more than 1 model type (*.pt, etc)
-# base_model_ignore_patterns:
-# # If the base_model repo on hf hub doesn't include configuration .json files,
-# # You can set that here, or leave this empty to default to base_model
-# base_model_config: ./llama-7b-hf
-# # You can specify to choose a specific model revision from huggingface hub
-# model_revision:
-# # Optional tokenizer configuration override in case you want to use a different tokenizer
-# # than the one defined in the base model
-# tokenizer_config:
-# # If you want to specify the type of model to load, AutoModelForCausalLM is a good choice too
-# model_type: AutoModelForCausalLM
-# # Corresponding tokenizer for the model AutoTokenizer is a good choice
-# tokenizer_type: AutoTokenizer
-# # Trust remote code for untrusted source
-# trust_remote_code:
-# # use_fast option for tokenizer loading from_pretrained, default to True
-# tokenizer_use_fast:
-# # Whether to use the legacy tokenizer setting, defaults to True
-# tokenizer_legacy:
-# # Resize the model embeddings when new tokens are added to multiples of 32
-# # This is reported to improve training speed on some models
-# resize_token_embeddings_to_32x:
-
-# # Used to identify which the model is based on
-# is_falcon_derived_model:
-# is_llama_derived_model:
-# # Please note that if you set this to true, `padding_side` will be set to "left" by default
-# is_mistral_derived_model:
-# is_qwen_derived_model:
-
-# # optional overrides to the base model configuration
-# model_config:
-#   # RoPE Scaling https://github.com/huggingface/transformers/pull/24653
-#   rope_scaling:
-#     type: # linear | dynamic
-#     factor: # float
-
-
-# # Whether you are training a 4-bit GPTQ quantized model
-# gptq: true
-# gptq_groupsize: 128 # group size
-# gptq_model_v1: false # v1 or v2
-
-# # This will attempt to quantize the model down to 8 bits and use adam 8 bit optimizer
-# load_in_8bit: true
-# # Use bitsandbytes 4 bit
-# load_in_4bit:
-
-# # Use CUDA bf16
-# bf16: true # bool or 'full' for `bf16_full_eval`. require >=ampere
-# # Use CUDA fp16
-# fp16: true
-# # Use CUDA tf32
-# tf32: true # require >=ampere
-
-# # No AMP (automatic mixed precision)
-# bfloat16: true # require >=ampere
-# float16: true
-
-# # A list of one or more datasets to finetune the model with
-# datasets:
-#   # HuggingFace dataset repo | s3://,gs:// path | "json" for local dataset, make sure to fill data_files
-#   - path: vicgalle/alpaca-gpt4
-#   # The type of prompt to use for training. [alpaca, sharegpt, gpteacher, oasst, reflection]
-#     type: alpaca # format | format:<prompt_style> (chat/instruct) | <prompt_strategies>.load_<load_fn>
-#     ds_type: # Optional[str] (json|arrow|parquet|text|csv) defines the datatype when path is a file
-#     data_files: # Optional[str] path to source data files
-#     shards: # Optional[int] number of shards to split data into
-#     name: # Optional[str] name of dataset configuration to load
-#     train_on_split: train # Optional[str] name of dataset split to load from
-
-#     # Optional[str] fastchat conversation type, only used with type: sharegpt
-#     conversation:  # Options (see Conversation 'name'): https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py
-#     field_human: # Optional[str]. Human key to use for conversation.
-#     field_model: # Optional[str]. Assistant key to use for conversation.
-
-#   # Custom user prompt
-#   - path: repo
-#     type:
-#       # The below are defaults. only set what's needed.
-#       system_prompt: ""
-#       system_format: "{system}"
-#       field_system: system
-#       field_instruction: instruction
-#       field_input: input
-#       field_output: output
-
-#       # Customizable to be single line or multi-line
-#       # 'format' can include {input}
-#       format: |-
-#         User: {instruction} {input}
-#         Assistant:
-#       # 'no_input_format' cannot include {input}
-#       no_input_format: "{instruction} "
-
-#       # For `completion` datsets only, uses the provided field instead of `text` column
-#       field:
-
-# # Axolotl attempts to save the dataset as an arrow after packing the data together so
-# # subsequent training attempts load faster, relative path
-# dataset_prepared_path: data/last_run_prepared
-# # Push prepared dataset to hub
-# push_dataset_to_hub: # repo path
-# # The maximum number of processes to use while preprocessing your input dataset. This defaults to `os.cpu_count()`
-# # if not set.
-# dataset_processes: # defaults to os.cpu_count() if not set
-# # push checkpoints to hub
-# hub_model_id: # repo path to push finetuned model
-# # how to push checkpoints to hub
-# # https://huggingface.co/docs/transformers/v4.31.0/en/main_classes/trainer#transformers.TrainingArguments.hub_strategy
-# hub_strategy:
-# # Whether to use hf `use_auth_token` for loading datasets. Useful for fetching private datasets
-# # Required to be true when used in combination with `push_dataset_to_hub`
-# hf_use_auth_token: # boolean
-# # How much of the dataset to set aside as evaluation. 1 = 100%, 0.50 = 50%, etc. 0 for no eval.
-# val_set_size: 0.04
-# # Num shards for whole dataset
-# dataset_shard_num:
-# # Index of shard to use for whole dataset
-# dataset_shard_idx:
-
-# # The maximum length of an input to train with, this should typically be less than 2048
-# # as most models have a token/context limit of 2048
-# sequence_len: 2048
-# # Pad inputs so each step uses constant sized buffers
-# # This will reduce memory fragmentation and may prevent OOMs, by re-using memory more efficiently
-# pad_to_sequence_len:
-# # Max sequence length to concatenate training samples together up to
-# # Inspired by StackLLaMA. see https://huggingface.co/blog/stackllama#supervised-fine-tuning
-# # FutureWarning: This will soon be DEPRECATED
-# max_packed_sequence_len: 1024
-# # Use efficient multi-packing with block diagonal attention and per sequence position_ids. Recommend set to 'true'
-# sample_packing:
-# # Set to 'false' if getting errors during eval with sample_packing on.
-# eval_sample_packing:
-# # You can set these packing optimizations AFTER starting a training at least once.
-# # The trainer will provide recommended values for these values.
-# sample_packing_eff_est:
-# total_num_tokens:
-
-# # If you want to use 'lora' or 'qlora' or leave blank to train all parameters in original model
-# adapter: lora
-# # If you already have a lora model trained that you want to load, put that here.
-# # This means after training, if you want to test the model, you should set this to the value of `lora_out_dir`.
-# lora_model_dir:
-
-# # LoRA hyperparameters
-# # For more details about the following options, see:
-# # https://www.anyscale.com/blog/fine-tuning-llms-lora-or-full-parameter-an-in-depth-analysis-with-llama-2
-# lora_r: 8
-# lora_alpha: 16
-# lora_dropout: 0.05
-# lora_target_modules:
-#   - q_proj
-#   - v_proj
-# #  - k_proj
-# #  - o_proj
-# #  - gate_proj
-# #  - down_proj
-# #  - up_proj
-# lora_target_linear: # If true, will target all linear layers
-
-# # If you added new tokens to the tokenizer, you may need to save some LoRA modules because they need to know the new tokens.
-# # For LLaMA and Mistral, you need to save `embed_tokens` and `lm_head`. It may vary for other models.
-# # `embed_tokens` converts tokens to embeddings, and `lm_head` converts embeddings to token probabilities.
-# # https://github.com/huggingface/peft/issues/334#issuecomment-1561727994
-# lora_modules_to_save:
-# #  - embed_tokens
-# #  - lm_head
-
-# # Once you complete training, the model will be saved to the following directory.
-# # If you merge the adapter to the base model, a subdirectory `merged` will be created under this directory.
-# # Make sure `lora_model_dir` points to this directory if you want to use the trained model.
-# lora_out_dir:
-# lora_fan_in_fan_out: false
-
-# # ReLoRA configuration
-# # Must use either 'lora' or 'qlora' adapter, and does not support fsdp or deepspeed
-# relora_steps: # Number of steps per ReLoRA restart
-# relora_warmup_steps: # Number of per-restart warmup steps
-# relora_cpu_offload: # True to perform lora weight merges on cpu during restarts, for modest gpu memory savings
-
-# # wandb configuration if you're using it
-# wandb_mode: # "offline" to save run metadata locally and not sync to the server, "disabled" to turn off wandb
-# wandb_project: # Your wandb project name
-# wandb_entity: # A wandb Team name if using a Team
-# wandb_watch:
-# wandb_run_id: # Set the name of your wandb run
-# wandb_log_model: # "checkpoint" to log model to wandb Artifacts every `save_steps` or "end" to log only at the end of training
-
-# # Where to save the full-finetuned model to
-# output_dir: ./completed-model
-
-# # Whether to use torch.compile and which backend to use
-# torch_compile:  # bool
-# torch_compile_backend:  # Optional[str]
-
-# # Training hyperparameters
-
-# # If greater than 1, backpropagation will be skipped and the gradients will be accumulated for the given number of steps.
-# gradient_accumulation_steps: 1
-# # The number of samples to include in each batch. This is the number of samples sent to each GPU.
-# micro_batch_size: 2
-# eval_batch_size:
-# num_epochs: 4
-# warmup_steps: 100  # cannot use with warmup_ratio
-# warmup_ratio: 0.05  # cannot use with warmup_steps
-# learning_rate: 0.00003
-# lr_quadratic_warmup:
-# logging_steps:
-# save_strategy: # Set to `no` to skip checkpoint saves
-# save_steps: # Leave empty to save at each epoch
-# eval_steps: # Leave empty to eval at each epoch, integers for every N steps. decimal for fraction of total steps
-# save_total_limit: # Checkpoints saved at a time
-# # Maximum number of iterations to train for. It precedes num_epochs which means that
-# # if both are set, num_epochs will not be guaranteed.
-# # e.g., when 1 epoch is 1000 steps => `num_epochs: 2` and `max_steps: 100` will train for 100 steps
-# max_steps:
-
-# eval_table_size: # Approximate number of predictions sent to wandb depending on batch size. Enabled above 0. Default is 0
-# eval_table_max_new_tokens: # Total number of tokens generated for predictions sent to wandb. Default is 128
-
-# # Save model as safetensors (require safetensors package)
-# save_safetensors:
-
-# # Whether to mask out or include the human's prompt from the training labels
-# train_on_inputs: false
-# # Group similarly sized data to minimize padding.
-# # May be slower to start, as it must download and sort the entire dataset.
-# # Note that training loss may have an oscillating pattern with this enabled.
-# group_by_length: false
-
-# # Whether to use gradient checkpointing https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing
-# gradient_checkpointing: false
-
-# # Stop training after this many evaluation losses have increased in a row
-# # https://huggingface.co/transformers/v4.2.2/_modules/transformers/trainer_callback.html#EarlyStoppingCallback
-# early_stopping_patience: 3
-
-# # Specify a scheduler and kwargs to use with the optimizer
-# lr_scheduler: # 'one_cycle' | 'log_sweep' | empty for cosine
-# lr_scheduler_kwargs:
-
-# # For one_cycle optim
-# lr_div_factor: # Learning rate div factor
-
-# # For log_sweep optim
-# log_sweep_min_lr:
-# log_sweep_max_lr:
-
-# # Specify optimizer
-# # Valid values are driven by the Transformers OptimizerNames class, see:
-# # https://github.com/huggingface/transformers/blob/95b374952dc27d8511541d6f5a4e22c9ec11fb24/src/transformers/training_args.py#L134
-# #
-# # Note that not all optimizers may be available in your environment, ex: 'adamw_anyprecision' is part of
-# # torchdistx, 'adamw_bnb_8bit' is part of bnb.optim.Adam8bit, etc. When in doubt, it is recommended to start with the optimizer used
-# # in the examples/ for your model and fine-tuning use case.
-# #
-# # Valid values for 'optimizer' include:
-# # - adamw_hf
-# # - adamw_torch
-# # - adamw_torch_fused
-# # - adamw_torch_xla
-# # - adamw_apex_fused
-# # - adafactor
-# # - adamw_anyprecision
-# # - sgd
-# # - adagrad
-# # - adamw_bnb_8bit
-# # - lion_8bit
-# # - lion_32bit
-# # - paged_adamw_32bit
-# # - paged_adamw_8bit
-# # - paged_lion_32bit
-# # - paged_lion_8bit
-# optimizer:
-# # Specify weight decay
-# weight_decay:
-# # adamw hyperparams
-# adam_beta1:
-# adam_beta2:
-# adam_epsilon:
-# # Gradient clipping max norm
-# max_grad_norm:
-
-# # Augmentation techniques
-# # NEFT https://arxiv.org/abs/2310.05914, set this to a number (paper default is 5) to add noise to embeddings
-# # currently only supported on Llama and Mistral
-# noisy_embedding_alpha:
-
-# # Whether to bettertransformers
-# flash_optimum:
-# # Whether to use xformers attention patch https://github.com/facebookresearch/xformers:
-# xformers_attention:
-# # Whether to use flash attention patch https://github.com/Dao-AILab/flash-attention:
-# flash_attention:
-# flash_attn_cross_entropy:  # Whether to use flash-attention cross entropy implementation - advanced use only
-# flash_attn_rms_norm:  # Whether to use flash-attention rms norm implementation - advanced use only
-# flash_attn_fuse_qkv: # Whether to fuse QKV into a single operation
-# flash_attn_fuse_mlp: # Whether to fuse part of the MLP into a single operation
-# # Whether to use scaled-dot-product attention
-# # https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html
-# sdp_attention:
-# # Landmark attention (only llama)
-# landmark_attention:
-# # xpos RoPE see https://github.com/kaiokendev/cutoff-len-is-context-len/blob/main/util/xpos_rope_llama_monkey_patch.py
-# # LLaMA only
-# xpos_rope:
-
-# # Resume from a specific checkpoint dir
-# resume_from_checkpoint:
-# # If resume_from_checkpoint isn't set and you simply want it to start where it left off.
-# # Be careful with this being turned on between different models.
-# auto_resume_from_checkpoints: false
-
-# # Don't mess with this, it's here for accelerate and torchrun
-# local_rank:
-
-# # Add or change special tokens.
-# # If you add tokens here, you don't need to add them to the `tokens` list.
-# special_tokens:
-#   # bos_token: "<s>"
-#   # eos_token: "</s>"
-#   # unk_token: "<unk>"
-
-# # Add extra tokens.
-# tokens:
-
-# # FSDP
-# fsdp:
-# fsdp_config:
-
-# # Deepspeed config path. e.g., deepspeed/zero3.json
-# deepspeed:
-
-# # Advanced DDP Arguments
-# ddp_timeout:
-# ddp_bucket_cap_mb:
-# ddp_broadcast_buffers:
-
-# # Path to torch distx for optim 'adamw_anyprecision'
-# torchdistx_path:
-
-# # Set to HF dataset for type: 'completion' for streaming instead of pre-tokenize
-# pretraining_dataset:
-
-# # Debug mode
-# debug:
-
-# # Seed
-# seed:
-
-# # Allow overwrite yml config using from cli
-# strict:
-
-
-
-base_model: ${BASE_MODEL}
-base_model_ignore_patterns: ${BASE_MODEL_IGNORE_PATTERNS}
-base_model_config: ${BASE_MODEL_CONFIG}
-revision_of_model: ${REVISION_OF_MODEL}
-tokenizer_config: ${TOKENIZER_CONFIG}
-model_type: ${MODEL_TYPE}
-tokenizer_type: ${TOKENIZER_TYPE}
-trust_remote_code: ${TRUST_REMOTE_CODE}
-tokenizer_use_fast: ${TOKENIZER_USE_FAST}
-tokenizer_legacy: ${TOKENIZER_LEGACY}
-resize_token_embeddings_to_32x: ${RESIZE_TOKEN_EMBEDDINGS_TO_32X}
-
-is_falcon_derived_model: ${IS_FALCON_DERIVED_MODEL}
-is_llama_derived_model: ${IS_LLAMA_DERIVED_MODEL}
-is_qwen_derived_model: ${IS_QWEN_DERIVED_MODEL}
-is_mistral_derived_model: ${IS_MISTRAL_DERIVED_MODEL}
-
-overrides_of_model_config:
-  rope_scaling:
-    type: ${ROPE_SCALING_TYPE}
-    factor: ${ROPE_SCALING_FACTOR}
-
-bnb_config_kwargs:
-  llm_int8_has_fp16_weight: ${BNB_LLM_INT8_HAS_FP16_WEIGHT}
-  bnb_4bit_quant_type: ${BNB_4BIT_QUANT_TYPE}
-  bnb_4bit_use_double_quant: ${BNB_4BIT_USE_DOUBLE_QUANT}
-
-gptq: ${GPTQ}
-load_in_8bit: ${LOAD_IN_8BIT}
-load_in_4bit: ${LOAD_IN_4BIT}
-bf16: ${BF16}
-fp16: ${FP16}
-tf32: ${TF32}
-bfloat16: ${BFLOAT16}
-float16: ${FLOAT16}
-
-gpu_memory_limit: ${GPU_MEMORY_LIMIT}
-lora_on_cpu: ${LORA_ON_CPU}
-
-datasets:
-  - path: ${DATASET_PATH}
-    type: ${DATASET_TYPE}
-    ds_type: ${DATASET_DS_TYPE}
-    data_files: ${DATASET_DATA_FILES}
-    shards: ${DATASET_SHARDS}
-    name: ${DATASET_NAME}
-    train_on_split: ${DATASET_TRAIN_ON_SPLIT}
-    revision: ${DATASET_REVISION}
-    trust_remote_code: ${DATASET_TRUST_REMOTE_CODE}
-
-rl: ${RL}
-dpo_use_weighting: ${DPO_USE_WEIGHTING}
-
-chat_template: ${CHAT_TEMPLATE}
-chat_template_jinja: ${CHAT_TEMPLATE_JINJA}
-default_system_message: ${DEFAULT_SYSTEM_MESSAGE}
-dataset_prepared_path: ${DATASET_PREPARED_PATH}
-push_dataset_to_hub: ${PUSH_DATASET_TO_HUB}
-dataset_processes: ${DATASET_PROCESSES}
-dataset_keep_in_memory: ${DATASET_KEEP_IN_MEMORY}
-hub_model_id: ${HUB_MODEL_ID}
-hub_strategy: ${HUB_STRATEGY}
-hf_use_auth_token: ${HF_USE_AUTH_TOKEN}
-val_set_size: ${VAL_SET_SIZE}
-dataset_shard_num: ${DATASET_SHARD_NUM}
-dataset_shard_idx: ${DATASET_SHARD_IDX}
-
-sequence_len: ${SEQUENCE_LEN}
-pad_to_sequence_len: ${PAD_TO_SEQUENCE_LEN}
-sample_packing: ${SAMPLE_PACKING}
-eval_sample_packing: ${EVAL_SAMPLE_PACKING}
-sample_packing_eff_est: ${SAMPLE_PACKING_EFF_EST}
-total_num_tokens: ${TOTAL_NUM_TOKENS}
-sample_packing_group_size: ${SAMPLE_PACKING_GROUP_SIZE}
-sample_packing_bin_size: ${SAMPLE_PACKING_BIN_SIZE}
-
-batch_flattening: ${BATCH_FLATTENING}
-device_map: ${DEVICE_MAP}
-max_memory: ${MAX_MEMORY}
-
-adapter: ${ADAPTER}
-lora_model_dir: ${LORA_MODEL_DIR}
-
-lora_r: ${LORA_R}
-lora_alpha: ${LORA_ALPHA}
-lora_dropout: ${LORA_DROPOUT}
-lora_target_modules:
-  - ${LORA_TARGET_MODULES}
-lora_target_linear: ${LORA_TARGET_LINEAR}
-peft_layers_to_transform: ${PEFT_LAYERS_TO_TRANSFORM}
-lora_modules_to_save: ${LORA_MODULES_TO_SAVE}
-lora_fan_in_fan_out: ${LORA_FAN_IN_FAN_OUT}
-
-loraplus_lr_ratio: ${LORAPLUS_LR_RATIO}
-loraplus_lr_embedding: ${LORAPLUS_LR_EMBEDDING}
-
-peft:
-  loftq_config:
-    loftq_bits: ${LOFTQ_BITS}
-
-relora_steps: ${RELORA_STEPS}
-relora_warmup_steps: ${RELORA_WARMUP_STEPS}
-relora_anneal_steps: ${RELORA_ANNEAL_STEPS}
-relora_prune_ratio: ${RELORA_PRUNE_RATIO}
-relora_cpu_offload: ${RELORA_CPU_OFFLOAD}
-
-wandb_mode: ${WANDB_MODE}
-wandb_project: ${WANDB_PROJECT}
-wandb_entity: ${WANDB_ENTITY}
-wandb_watch: ${WANDB_WATCH}
-wandb_name: ${WANDB_NAME}
-wandb_run_id: ${WANDB_RUN_ID}
-wandb_log_model: ${WANDB_LOG_MODEL}
-
-mlflow_tracking_uri: ${MLFLOW_TRACKING_URI}
-mlflow_experiment_name: ${MLFLOW_EXPERIMENT_NAME}
-mlflow_run_name: ${MLFLOW_RUN_NAME}
-hf_mlflow_log_artifacts: ${HF_MLFLOW_LOG_ARTIFACTS}
-
-use_comet: ${USE_COMET}
-comet_api_key: ${COMET_API_KEY}
-comet_workspace: ${COMET_WORKSPACE}
-comet_project_name: ${COMET_PROJECT_NAME}
-comet_experiment_key: ${COMET_EXPERIMENT_KEY}
-comet_mode: ${COMET_MODE}
-comet_online: ${COMET_ONLINE}
-comet_experiment_config: ${COMET_EXPERIMENT_CONFIG}
-
-output_dir: ${OUTPUT_DIR}
-
-torch_compile: ${TORCH_COMPILE}
-torch_compile_backend: ${TORCH_COMPILE_BACKEND}
-
-gradient_accumulation_steps: ${GRADIENT_ACCUMULATION_STEPS}
-micro_batch_size: ${MICRO_BATCH_SIZE}
-eval_batch_size: ${EVAL_BATCH_SIZE}
-num_epochs: ${NUM_EPOCHS}
-warmup_steps: ${WARMUP_STEPS}
-warmup_ratio: ${WARMUP_RATIO}
-learning_rate: ${LEARNING_RATE}
-lr_quadratic_warmup: ${LR_QUADRATIC_WARMUP}
-logging_steps: ${LOGGING_STEPS}
-eval_steps: ${EVAL_STEPS}
-evals_per_epoch: ${EVALS_PER_EPOCH}
-save_strategy: ${SAVE_STRATEGY}
-save_steps: ${SAVE_STEPS}
-saves_per_epoch: ${SAVES_PER_EPOCH}
-save_total_limit: ${SAVE_TOTAL_LIMIT}
-max_steps: ${MAX_STEPS}
-
-eval_table_size: ${EVAL_TABLE_SIZE}
-eval_max_new_tokens: ${EVAL_MAX_NEW_TOKENS}
-eval_causal_lm_metrics: ${EVAL_CAUSAL_LM_METRICS}
-
-profiler_steps: ${PROFILER_STEPS}
-loss_watchdog_threshold: ${LOSS_WATCHDOG_THRESHOLD}
-loss_watchdog_patience: ${LOSS_WATCHDOG_PATIENCE}
-
-save_safetensors: ${SAVE_SAFETENSORS}
-train_on_inputs: ${TRAIN_ON_INPUTS}
-group_by_length: ${GROUP_BY_LENGTH}
-gradient_checkpointing: ${GRADIENT_CHECKPOINTING}
-early_stopping_patience: ${EARLY_STOPPING_PATIENCE}
-
-lr_scheduler: ${LR_SCHEDULER}
-lr_scheduler_kwargs: ${LR_SCHEDULER_KWARGS}
-cosine_min_lr_ratio: ${COSINE_MIN_LR_RATIO}
-cosine_constant_lr_ratio: ${COSINE_CONSTANT_LR_RATIO}
-lr_div_factor: ${LR_DIV_FACTOR}
-
-optimizer: ${OPTIMIZER}
-optim_args: ${OPTIM_ARGS}
-optim_target_modules: ${OPTIM_TARGET_MODULES}
-weight_decay: ${WEIGHT_DECAY}
-adam_beta1: ${ADAM_BETA1}
-adam_beta2: ${ADAM_BETA2}
-adam_epsilon: ${ADAM_EPSILON}
-max_grad_norm: ${MAX_GRAD_NORM}
-
-neftune_noise_alpha: ${NEFTUNE_NOISE_ALPHA}
-
-flash_optimum: ${FLASH_OPTIMUM}
-xformers_attention: ${XFORMERS_ATTENTION}
-flash_attention: ${FLASH_ATTENTION}
-flash_attn_cross_entropy: ${FLASH_ATTN_CROSS_ENTROPY}
-flash_attn_rms_norm: ${FLASH_ATTN_RMS_NORM}
-flash_attn_fuse_qkv: ${FLASH_ATTN_FUSE_QKV}
-flash_attn_fuse_mlp: ${FLASH_ATTN_FUSE_MLP}
-sdp_attention: ${SDP_ATTENTION}
-s2_attention: ${S2_ATTENTION}
-resume_from_checkpoint: ${RESUME_FROM_CHECKPOINT}
-auto_resume_from_checkpoints: ${AUTO_RESUME_FROM_CHECKPOINTS}
-
-local_rank: ${LOCAL_RANK}
-
-special_tokens:
-  bos_token: ${SPECIAL_TOKEN_BOS}
-  eos_token: ${SPECIAL_TOKEN_EOS}
-  unk_token: ${SPECIAL_TOKEN_UNK}
-  pad_token: ${SPECIAL_TOKEN_PAD}
-
-tokens: ${TOKENS}
-
-fsdp: ${FSDP}
-fsdp_config: ${FSDP_CONFIG}
-deepspeed: ${DEEPSPEED}
-
-ddp_timeout: ${DDP_TIMEOUT}
-ddp_bucket_cap_mb: ${DDP_BUCKET_CAP_MB}
-ddp_broadcast_buffers: ${DDP_BROADCAST_BUFFERS}
-
-torchdistx_path: ${TORCHDISTX_PATH}
-pretraining_dataset: ${PRETRAINING_DATASET}
-debug: ${DEBUG}
-seed: ${SEED}
-strict: ${STRICT}
--- a/.runpod/src/handler.py
+++ b/.runpod/src/handler.py
@@ -1,64 +0,0 @@
-"""
-Runpod serverless entrypoint handler
-"""
-
-import os
-
-import runpod
-import yaml
-from huggingface_hub._login import login
-from train import train
-from utils import get_output_dir
-
-BASE_VOLUME = os.environ.get("BASE_VOLUME", "/runpod-volume")
-if not os.path.exists(BASE_VOLUME):
-    os.makedirs(BASE_VOLUME)
-
-logger = runpod.RunPodLogger()
-
-
-async def handler(job):
-    runpod_job_id = job["id"]
-    inputs = job["input"]
-    run_id = inputs.get("run_id", "default_run_id")
-    args = inputs.get("args", {})
-
-    # Set output directory
-    output_dir = os.path.join(BASE_VOLUME, get_output_dir(run_id))
-    args["output_dir"] = output_dir
-
-    # First save args to a temporary config file
-    config_path = "/workspace/test_config.yaml"
-
-    # Add run_name and job_id to args before saving
-    args["run_name"] = run_id
-    args["runpod_job_id"] = runpod_job_id
-
-    yaml_data = yaml.dump(args, default_flow_style=False)
-    with open(config_path, "w", encoding="utf-8") as file:
-        file.write(yaml_data)
-
-    # Handle credentials
-    credentials = inputs.get("credentials", {})
-
-    if "wandb_api_key" in credentials:
-        os.environ["WANDB_API_KEY"] = credentials["wandb_api_key"]
-    if "hf_token" in credentials:
-        os.environ["HF_TOKEN"] = credentials["hf_token"]
-
-    if os.environ.get("HF_TOKEN"):
-        login(token=os.environ["HF_TOKEN"])
-    else:
-        logger.info("No HF_TOKEN provided. Skipping login.")
-
-    logger.info("Starting Training.")
-    async for result in train(config_path):  # Pass the config path instead of args
-        logger.info(result)
-    logger.info("Training Complete.")
-
-    # Cleanup
-    del os.environ["WANDB_API_KEY"]
-    del os.environ["HF_TOKEN"]
-
-
-runpod.serverless.start({"handler": handler, "return_aggregate_stream": True})
--- a/.runpod/src/test_input.json
+++ b/.runpod/src/test_input.json
@@ -1,61 +0,0 @@
-{
-  "input": {
-    "user_id": "user",
-    "model_id": "llama-test",
-    "run_id": "llama-test",
-    "credentials": {
-      "wandb_api_key": "",
-      "hf_token": ""
-    },
-    "args": {
-      "base_model": "NousResearch/Meta-Llama-3-8B",
-      "model_type": "LlamaForCausalLM",
-      "tokenizer_type": "AutoTokenizer",
-      "load_in_8bit": true,
-      "load_in_4bit": false,
-      "strict": false,
-      "datasets": [
-        {
-          "path": "mhenrichsen/alpaca_2k_test",
-          "type": "alpaca"
-        }
-      ],
-      "val_set_size": 0.05,
-      "output_dir": "./outputs/lora-out",
-      "sequence_len": 4096,
-      "sample_packing": true,
-      "eval_sample_packing": false,
-      "pad_to_sequence_len": true,
-      "adapter": "lora",
-      "lora_r": 32,
-      "lora_alpha": 16,
-      "lora_dropout": 0.05,
-      "lora_target_linear": true,
-      "lora_modules_to_save": [
-        "embed_tokens",
-        "lm_head"
-      ],
-      "gradient_accumulation_steps": 4,
-      "micro_batch_size": 2,
-      "num_epochs": 1,
-      "optimizer": "adamw_bnb_8bit",
-      "lr_scheduler": "cosine",
-      "learning_rate": 0.0002,
-      "train_on_inputs": false,
-      "group_by_length": false,
-      "bf16": "auto",
-      "tf32": false,
-      "gradient_checkpointing": true,
-      "logging_steps": 1,
-      "flash_attention": true,
-      "warmup_steps": 1,
-      "evals_per_epoch": 1,
-      "eval_max_new_tokens": 128,
-      "saves_per_epoch": 1,
-      "weight_decay": 0.0,
-      "special_tokens": {
-        "pad_token": "<|end_of_text|>"
-      }
-    }
-  }
-}
--- a/.runpod/src/train.py
+++ b/.runpod/src/train.py
@@ -1,45 +0,0 @@
-"""
-Runpod train entrypoint
-"""
-
-import asyncio
-
-
-async def train(config_path: str, gpu_id: str = "0", preprocess: bool = True):
-    """
-    Run preprocessing (if enabled) and training with the given config file
-    :param config_path: Path to the YAML config file
-    :param gpu_id: GPU ID to use (default: "0")
-    :param preprocess: Whether to run preprocessing (default: True)
-
-    """
-    # First check if preprocessing is needed
-    if preprocess:
-        # Preprocess command
-        preprocess_cmd = (
-            f"CUDA_VISIBLE_DEVICES={gpu_id} axolotl preprocess {config_path}"
-        )
-        process = await asyncio.create_subprocess_shell(
-            preprocess_cmd,
-            stdout=asyncio.subprocess.PIPE,
-            stderr=asyncio.subprocess.STDOUT,
-        )
-
-        if process.stdout is not None:
-            async for line in process.stdout:
-                yield f"Preprocessing: {line.decode().strip()}"
-        await process.wait()
-        yield "Preprocessing completed."
-    else:
-        yield "Skipping preprocessing step."
-
-    # Training command
-    train_cmd = f"axolotl train {config_path}"
-    process = await asyncio.create_subprocess_shell(
-        train_cmd, stdout=asyncio.subprocess.PIPE, stderr=asyncio.subprocess.STDOUT
-    )
-
-    if process.stdout is not None:
-        async for line in process.stdout:
-            yield f"Training: {line.decode().strip()}"
-    await process.wait()
--- a/.runpod/src/utils.py
+++ b/.runpod/src/utils.py
@@ -1,89 +0,0 @@
-"""
-Runpod launcher utils
-"""
-
-import os
-
-import yaml
-
-
-def get_output_dir(run_id):
-    path = f"fine-tuning/{run_id}"
-    return path
-
-
-def make_valid_config(input_args):
-    """
-    Creates and saves updated config file, returns the path to the new config
-    :param input_args: dict of input args
-    :return: str, path to the updated config file
-    """
-    # Load default config
-    with open("config/config.yaml", "r", encoding="utf-8") as fin:
-        all_args = yaml.safe_load(fin)
-
-    if not input_args:
-        print("No args provided, using defaults")
-    else:
-        all_args.update(input_args)
-
-    # Create updated config path
-    updated_config_path = "config/updated_config.yaml"
-
-    # Save updated config to new file
-    with open(updated_config_path, "w", encoding="utf-8") as f:
-        yaml.dump(all_args, f)
-
-    return updated_config_path
-
-
-def set_config_env_vars(args: dict):
-    """
-    Convert API arguments into environment variables.
-    Handles nested dictionaries, lists, and special values.
-
-    Args:
-        args (dict): The arguments dictionary from the API request
-    """
-
-    def process_value(value):
-        """Convert Python values to string format for environment variables"""
-        if value is None:
-            return ""
-        if isinstance(value, bool):
-            return str(value).lower()
-        if isinstance(value, (list, dict)):
-            return str(value)
-        return str(value)
-
-    def set_env_vars(data, prefix=""):
-        """Recursively set environment variables from nested dictionary"""
-        for key, value in data.items():
-            env_key = prefix + key.upper()
-
-            # Handle special cases
-            if isinstance(value, dict):
-                # For nested dictionaries (like special_tokens)
-                set_env_vars(value, f"{env_key}_")
-            elif isinstance(value, list):
-                # Handle list of dictionaries (like datasets)
-                if value and isinstance(value[0], dict):
-                    for i, item in enumerate(value):
-                        set_env_vars(item, f"{env_key}_{i}_")
-                else:
-                    # For simple lists (like lora_target_modules)
-                    os.environ[env_key] = process_value(value)
-            else:
-                # Handle all other cases
-                os.environ[env_key] = process_value(value)
-
-    # Clear any existing related environment variables
-    # This prevents old values from persisting
-    for key in list(os.environ.keys()):
-        if key.startswith(
-            ("BASE_MODEL", "MODEL_TYPE", "TOKENIZER_TYPE", "DATASET", "LORA_", "WANDB_")
-        ):
-            del os.environ[key]
-
-    # Set new environment variables
-    set_env_vars(args)
--- a/.runpod/tests.json
+++ b/.runpod/tests.json
@@ -1,85 +0,0 @@
-{
-  "input": {
-    "name": "quick_smoke_test_sft",
-    "user_id": "user",
-    "model_id": "llama-test",
-    "run_id": "llama-test",
-    "credentials": {
-      "wandb_api_key": "",
-      "hf_token": ""
-    },
-    "args": {
-      "base_model": "HuggingFaceTB/SmolLM2-135M",
-      "model_type": "AutoModelForCausalLM",
-      "tokenizer_type": "AutoTokenizer",
-      "load_in_8bit": true,
-      "load_in_4bit": false,
-      "strict": false,
-      "datasets": [
-        {
-          "path": "mhenrichsen/alpaca_2k_test",
-          "type": "alpaca"
-        }
-      ],
-      "val_set_size": 0.05,
-      "output_dir": "./outputs/lora-out",
-      "sequence_len": 4096,
-      "sample_packing": true,
-      "eval_sample_packing": false,
-      "pad_to_sequence_len": true,
-      "adapter": "lora",
-      "lora_r": 32,
-      "lora_alpha": 64,
-      "lora_dropout": 0.05,
-      "lora_target_linear": true,
-      "lora_modules_to_save": [
-        "embed_tokens",
-        "lm_head"
-      ],
-      "gradient_accumulation_steps": 4,
-      "micro_batch_size": 2,
-      "num_epochs": 1,
-      "optimizer": "adamw_torch_fused",
-      "lr_scheduler": "cosine",
-      "learning_rate": 0.0002,
-      "train_on_inputs": false,
-      "group_by_length": false,
-      "bf16": "auto",
-      "tf32": true,
-      "gradient_checkpointing": true,
-      "logging_steps": 1,
-      "flash_attention": true,
-      "warmup_steps": 1,
-      "evals_per_epoch": 1,
-      "eval_max_new_tokens": 128,
-      "saves_per_epoch": 1,
-      "weight_decay": 0.0,
-      "special_tokens": {
-        "pad_token": "<|endoftext|>"
-      }
-    },
-    "timeout": 100000
-  },
-  "config": {
-    "gpuTypeId": "NVIDIA GeForce RTX 4090",
-    "gpuCount": 1,
-    "containerDiskInGb": 200,
-    "env": [
-      {
-        "key": "TOKENIZER",
-        "value": ""
-      },
-      {
-        "key": "DISABLE_LOG_STATS",
-        "value": "true"
-      }
-    ],
-    "allowedCudaVersions": [
-      "12.8",
-      "12.7",
-      "12.6",
-      "12.5",
-      "12.4"
-    ]
-  }
-}
--- a/1
+++ b/1
@@ -1 +0,0 @@
-docs.axolotl.ai
--- a/README.md
+++ b/README.md
@@ -9,7 +9,6 @@
 <p align="center">
    <img src="https://img.shields.io/github/license/axolotl-ai-cloud/axolotl.svg?color=blue" alt="GitHub License">
    <img src="https://github.com/axolotl-ai-cloud/axolotl/actions/workflows/tests.yml/badge.svg" alt="tests">
-    <a href="https://codecov.io/gh/axolotl-ai-cloud/axolotl"><img src="https://codecov.io/gh/axolotl-ai-cloud/axolotl/branch/main/graph/badge.svg" alt="codecov"></a>
    <a href="https://github.com/axolotl-ai-cloud/axolotl/releases"><img src="https://img.shields.io/github/release/axolotl-ai-cloud/axolotl.svg" alt="Releases"></a>
    <br/>
    <a href="https://github.com/axolotl-ai-cloud/axolotl/graphs/contributors"><img src="https://img.shields.io/github/contributors-anon/axolotl-ai-cloud/axolotl?color=yellow&style=flat-square" alt="contributors" style="height: 20px;"></a>
@@ -20,6 +19,9 @@
    <br/>
    <img src="https://github.com/axolotl-ai-cloud/axolotl/actions/workflows/tests-nightly.yml/badge.svg" alt="tests-nightly">
    <img src="https://github.com/axolotl-ai-cloud/axolotl/actions/workflows/multi-gpu-e2e.yml/badge.svg" alt="multigpu-semi-weekly tests">
+    <a href="https://www.phorm.ai/query?projectId=e315ba4a-4e14-421f-ab05-38a1f9076f25">
+    <img alt="phorm.ai" src="https://img.shields.io/badge/Phorm-Ask_AI-%23F2777A.svg?&logo=data:image/svg+xml;base64,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">
+  </a>
 </p>

 Axolotl is a tool designed to streamline post-training for various AI models.
@@ -48,15 +50,13 @@ Features:
 ## 🚀 Quick Start

 **Requirements**:
-
 - NVIDIA GPU (Ampere or newer for `bf16` and Flash Attention) or AMD GPU
 - Python 3.11
 - PyTorch ≥2.4.1

 ### Installation

-```bash
-pip3 install -U packaging==23.2 setuptools==75.8.0 wheel ninja
+```shell
 pip3 install --no-build-isolation axolotl[flash-attn,deepspeed]

 # Download example axolotl configs, deepspeed configs
@@ -64,11 +64,11 @@ axolotl fetch examples
 axolotl fetch deepspeed_configs  # OPTIONAL
 ```

-Other installation approaches are described [here](https://docs.axolotl.ai/docs/installation.html).
+Other installation approaches are described [here](https://axolotl-ai-cloud.github.io/axolotl/docs/installation.html).

 ### Your First Fine-tune

-```bash
+```shell
 # Fetch axolotl examples
 axolotl fetch examples

@@ -79,7 +79,7 @@ axolotl fetch examples --dest path/to/folder
 axolotl train examples/llama-3/lora-1b.yml
 ```

-That's it! Check out our [Getting Started Guide](https://docs.axolotl.ai/docs/getting-started.html) for a more detailed walkthrough.
+That's it! Check out our [Getting Started Guide](https://axolotl-ai-cloud.github.io/axolotl/docs/getting-started.html) for a more detailed walkthrough.

 ## ✨ Key Features

@@ -92,20 +92,19 @@ That's it! Check out our [Getting Started Guide](https://docs.axolotl.ai/docs/ge

 ## 📚 Documentation

- [Installation Options](https://docs.axolotl.ai/docs/installation.html) - Detailed setup instructions for different environments
- [Configuration Guide](https://docs.axolotl.ai/docs/config.html) - Full configuration options and examples
- [Dataset Guide](https://docs.axolotl.ai/docs/dataset-formats/) - Supported formats and how to use them
- [Multi-GPU Training](https://docs.axolotl.ai/docs/multi-gpu.html)
- [Multi-Node Training](https://docs.axolotl.ai/docs/multi-node.html)
- [Multipacking](https://docs.axolotl.ai/docs/multipack.html)
- [API Reference](https://docs.axolotl.ai/docs/api/) - Auto-generated code documentation
- [FAQ](https://docs.axolotl.ai/docs/faq.html) - Frequently asked questions
+- [Installation Options](https://axolotl-ai-cloud.github.io/axolotl/docs/installation.html) - Detailed setup instructions for different environments
+- [Configuration Guide](https://axolotl-ai-cloud.github.io/axolotl/docs/config.html) - Full configuration options and examples
+- [Dataset Guide](https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/) - Supported formats and how to use them
+- [Multi-GPU Training](https://axolotl-ai-cloud.github.io/axolotl/docs/multi-gpu.html)
+- [Multi-Node Training](https://axolotl-ai-cloud.github.io/axolotl/docs/multi-node.html)
+- [Multipacking](https://axolotl-ai-cloud.github.io/axolotl/docs/multipack.html)
+- [FAQ](https://axolotl-ai-cloud.github.io/axolotl/docs/faq.html) - Frequently asked questions

 ## 🤝 Getting Help

 - Join our [Discord community](https://discord.gg/HhrNrHJPRb) for support
 - Check out our [Examples](https://github.com/axolotl-ai-cloud/axolotl/tree/main/examples/) directory
- Read our [Debugging Guide](https://docs.axolotl.ai/docs/debugging.html)
+- Read our [Debugging Guide](https://axolotl-ai-cloud.github.io/axolotl/docs/debugging.html)
 - Need dedicated support? Please contact [✉️wing@axolotl.ai](mailto:wing@axolotl.ai) for options

 ## 🌟 Contributing
--- a/_quarto.yml
+++ b/_quarto.yml
@@ -1,188 +1,12 @@
 project:
  type: website

-quartodoc:
-  dir: docs/api
-  package: axolotl
-  title: API Reference
-  parser: google
-
-  sections:
-    - title: Core
-      desc: Core functionality for training
-      contents:
-        - train
-        - evaluate
-        - datasets
-        - convert
-        - prompt_tokenizers
-        - logging_config
-        - core.trainer_builder
-        - core.training_args
-        - core.chat.messages
-        - core.chat.format.chatml
-        - core.chat.format.llama3x
-        - core.chat.format.shared
-        - core.datasets.chat
-        - core.datasets.transforms.chat_builder
-    - title: CLI
-      desc: Command-line interface
-      contents:
-        - cli.main
-        - cli.train
-        - cli.evaluate
-        - cli.args
-        - cli.checks
-        - cli.config
-        - cli.inference
-        - cli.merge_lora
-        - cli.merge_sharded_fsdp_weights
-        - cli.preprocess
-        - cli.sweeps
-        - cli.utils
-        - cli.vllm_serve
-        - cli.cloud.base
-        - cli.cloud.modal_
-    - title: Trainers
-      desc: Training implementations
-      contents:
-        - core.trainers.base
-        - core.trainers.trl
-        - core.trainers.dpo.trainer
-        - core.trainers.grpo.trainer
-    - title: Prompt Strategies
-      desc: Prompt formatting strategies
-      contents:
-        - prompt_strategies.base
-        - prompt_strategies.chat_template
-        - prompt_strategies.alpaca_chat
-        - prompt_strategies.alpaca_instruct
-        - prompt_strategies.alpaca_w_system
-        - prompt_strategies.user_defined
-        - prompt_strategies.llama2_chat
-        - prompt_strategies.completion
-        - prompt_strategies.input_output
-        - prompt_strategies.stepwise_supervised
-        - prompt_strategies.metharme
-        - prompt_strategies.orcamini
-        - prompt_strategies.pygmalion
-        - prompt_strategies.messages.chat
-        - prompt_strategies.dpo.chat_template
-        - prompt_strategies.dpo.llama3
-        - prompt_strategies.dpo.chatml
-        - prompt_strategies.dpo.zephyr
-        - prompt_strategies.dpo.user_defined
-        - prompt_strategies.dpo.passthrough
-        - prompt_strategies.kto.llama3
-        - prompt_strategies.kto.chatml
-        - prompt_strategies.kto.user_defined
-        - prompt_strategies.orpo.chat_template
-        - prompt_strategies.bradley_terry.llama3
-    - title: Kernels
-      desc: Low-level performance optimizations
-      contents:
-        - kernels.lora
-        - kernels.geglu
-        - kernels.swiglu
-        - kernels.quantize
-        - kernels.utils
-    - title: MonkeyPatches
-      desc: Runtime patches for model optimizations
-      contents:
-        - monkeypatch.llama_attn_hijack_flash
-        - monkeypatch.llama_attn_hijack_xformers
-        - monkeypatch.mistral_attn_hijack_flash
-        - monkeypatch.multipack
-        - monkeypatch.relora
-        - monkeypatch.llama_expand_mask
-        - monkeypatch.lora_kernels
-        - monkeypatch.utils
-        - monkeypatch.btlm_attn_hijack_flash
-        - monkeypatch.llama_patch_multipack
-        - monkeypatch.stablelm_attn_hijack_flash
-        - monkeypatch.trainer_fsdp_optim
-        - monkeypatch.transformers_fa_utils
-        - monkeypatch.unsloth_
-        - monkeypatch.attention.mllama
-        - monkeypatch.data.batch_dataset_fetcher
-        - monkeypatch.mixtral
-    - title: Utils
-      desc: Utility functions
-      contents:
-        - utils.models
-        - utils.tokenization
-        - utils.chat_templates
-        - utils.lora
-        - utils.lora_embeddings
-        - utils.model_shard_quant
-        - utils.bench
-        - utils.freeze
-        - utils.trainer
-        - utils.schedulers
-        - utils.distributed
-        - utils.dict
-        - utils.optimizers.adopt
-        - utils.data.pretraining
-        - utils.data.sft
-        - utils.gradient_checkpointing.unsloth
-    - title: Schemas
-      desc: Pydantic data models for Axolotl config
-      contents:
-        - utils.schemas.config
-        - utils.schemas.model
-        - utils.schemas.training
-        - utils.schemas.datasets
-        - utils.schemas.peft
-        - utils.schemas.trl
-        - utils.schemas.multimodal
-        - utils.schemas.integrations
-        - utils.schemas.enums
-        - utils.schemas.utils
-    - title: Integrations
-      desc: Third-party integrations and extensions
-      contents:
-        - integrations.base
-        - integrations.cut_cross_entropy.args
-        - integrations.grokfast.optimizer
-        - integrations.kd.trainer
-        - integrations.liger.args
-        - integrations.lm_eval.args
-        - integrations.spectrum.args
-    - title: Common
-      desc: Common utilities and shared functionality
-      contents:
-        - common.architectures
-        - common.const
-        - common.datasets
-    - title: Models
-      desc: Custom model implementations
-      contents:
-        - models.mamba.modeling_mamba
-    - title: Data Processing
-      desc: Data processing utilities
-      contents:
-        - utils.collators.core
-        - utils.collators.batching
-        - utils.collators.mamba
-        - utils.collators.mm_chat
-        - utils.samplers.multipack
-    - title: Callbacks
-      desc: Training callbacks
-      contents:
-        - utils.callbacks.perplexity
-        - utils.callbacks.profiler
-        - utils.callbacks.lisa
-        - utils.callbacks.mlflow_
-        - utils.callbacks.comet_
-
 website:
  title: "Axolotl"
-  description: "We make fine-tuning accessible, scalable, and fun"
+  description: "Fine-tuning"
  favicon: favicon.jpg
-
  navbar:
-    logo: image/axolotl_logo_digital_white.svg
-    title: false
+    title: Axolotl
    background: dark
    pinned: false
    collapse: false
@@ -201,79 +25,33 @@ website:
      contents:
        - text: Home
          href: index.qmd
-
-        - section: "Getting Started"
+        - section: "How-To Guides"
          contents:
+          # TODO Edit folder structure after we have more docs.
            - docs/getting-started.qmd
            - docs/installation.qmd
+            - docs/debugging.qmd
            - docs/inference.qmd
-            - docs/cli.qmd
-            - docs/config.qmd
-            - text: "API Reference"
-              href: docs/api
-
-        - section: "Dataset Formats"
-          contents: docs/dataset-formats/*
-
-        - section: "Deployments"
-          contents:
-            - docs/docker.qmd
+            - docs/multipack.qmd
+            - docs/fsdp_qlora.qmd
+            - docs/input_output.qmd
+            - docs/rlhf.qmd
+            - docs/nccl.qmd
+            - docs/mac.qmd
            - docs/multi-gpu.qmd
            - docs/multi-node.qmd
-            - docs/ray-integration.qmd
-            - docs/amd_hpc.qmd
-            - docs/mac.qmd
-
-        - section: "How To Guides"
-          contents:
-            - docs/multimodal.qmd
-            - docs/rlhf.qmd
-            - docs/reward_modelling.qmd
-            - docs/lr_groups.qmd
-            - docs/lora_optims.qmd
-            - docs/dataset_loading.qmd
-
-        - section: "Core Concepts"
-          contents:
-            - docs/batch_vs_grad.qmd
-            - docs/dataset_preprocessing.qmd
-            - docs/multipack.qmd
-
-        - section: "Advanced Features"
-          contents:
-            - docs/fsdp_qlora.qmd
            - docs/unsloth.qmd
-            - docs/torchao.qmd
-            - docs/custom_integrations.qmd
-            - docs/sequence_parallelism.qmd
-
-        - section: "Troubleshooting"
+            - docs/amd_hpc.qmd
+            - docs/ray-integration.qmd
+        - section: "Dataset Formats"
+          contents: docs/dataset-formats/*
+        - section: "Reference"
          contents:
-            - docs/faq.qmd
-            - docs/debugging.qmd
-            - docs/nccl.qmd
+            - docs/config.qmd
+        - docs/faq.qmd

 format:
  html:
-    theme: darkly
+    theme: materia
    css: styles.css
    toc: true
-    # Enable better handling of line breaks in markdown
-    preserve-tabs: true
-    html-math-method: mathjax
-    # Improved markdown processing options
-    md-extensions:
-      - markdown_it
-      - def_list
-      - attr_list
-      - fenced_divs
-      - tables
-      - html_admonition
-      - lineblocks
-      - fancy_lists
-    # Control whitespace handling
-    whitespace: preserve
-    # Process newlines in paragraphs
-    wrap: preserve
-    # Better line break handling
-    preserve-linebreaks: true
--- a/cicd/Dockerfile.jinja
+++ b/cicd/Dockerfile.jinja
@@ -31,11 +31,10 @@ RUN if [ "$NIGHTLY_BUILD" = "true" ] ; then \
        sed -i 's#^datasets.*#datasets @ git+https://github.com/huggingface/datasets.git@main#' requirements.txt; \
    fi

-RUN pip install packaging==23.2 setuptools==75.8.0
 RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install --no-build-isolation -e .[deepspeed,flash-attn,ring-flash-attn,optimizers,ray,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+        pip install --no-build-isolation -e .[deepspeed,flash-attn,optimizers,ray,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
    else \
-        pip install --no-build-isolation -e .[deepspeed,flash-attn,ring-flash-attn,optimizers,ray] $AXOLOTL_ARGS; \
+        pip install --no-build-isolation -e .[deepspeed,flash-attn,optimizers,ray] $AXOLOTL_ARGS; \
    fi

 RUN python scripts/unsloth_install.py | sh
--- a/cicd/cicd.sh
+++ b/cicd/cicd.sh
@@ -3,53 +3,9 @@ set -e

 python -c "import torch; assert '$PYTORCH_VERSION' in torch.__version__"

-# Run unit tests with initial coverage report
-pytest -v --durations=10 -n8 \
-  --ignore=tests/e2e/ \
-  --ignore=tests/patched/ \
-  --ignore=tests/cli \
-  /workspace/axolotl/tests/ \
-  --cov=axolotl
-
-# Run lora kernels tests with coverage append
-pytest -v --durations=10 \
-  /workspace/axolotl/tests/e2e/patched/lora_kernels \
-  --cov=axolotl \
-  --cov-append
-
-# Run patched tests excluding lora kernels with coverage append
-pytest -v --durations=10 \
-  --ignore=tests/e2e/patched/lora_kernels \
-  /workspace/axolotl/tests/e2e/patched \
-  --cov=axolotl \
-  --cov-append
-
-# Run solo tests with coverage append
-pytest -v --durations=10 -n1 \
-  /workspace/axolotl/tests/e2e/solo/ \
-  --cov=axolotl \
-  --cov-append
-
-# Run integration tests with coverage append
-pytest -v --durations=10 \
-  /workspace/axolotl/tests/e2e/integrations/ \
-  --cov=axolotl \
-  --cov-append
-
-pytest -v --durations=10 /workspace/axolotl/tests/cli \
-  --cov=axolotl \
-  --cov-append
-
-# Run remaining e2e tests with coverage append and final report
-pytest -v --durations=10 \
-  --ignore=tests/e2e/solo/ \
-  --ignore=tests/e2e/patched/ \
-  --ignore=tests/e2e/multigpu/ \
-  --ignore=tests/e2e/integrations/ \
-  --ignore=tests/cli \
-  /workspace/axolotl/tests/e2e/ \
-  --cov=axolotl \
-  --cov-append \
-  --cov-report=xml:e2e-coverage.xml
-
-codecov upload-process -t $CODECOV_TOKEN -f e2e-coverage.xml -F e2e,pytorch-${PYTORCH_VERSION} || true
+pytest -v --durations=10 -n8 --ignore=tests/e2e/ --ignore=tests/patched/ /workspace/axolotl/tests/
+pytest -v --durations=10 /workspace/axolotl/tests/e2e/patched/lora_kernels  # running these with the other patches causes a failure
+pytest -v --durations=10 --ignore=tests/e2e/patched/lora_kernels /workspace/axolotl/tests/e2e/patched
+pytest -v --durations=10 -n1 /workspace/axolotl/tests/e2e/solo/
+pytest -v --durations=10 /workspace/axolotl/tests/e2e/integrations/
+pytest -v --durations=10 --ignore=tests/e2e/solo/ --ignore=tests/e2e/patched/ --ignore=tests/e2e/multigpu/ --ignore=tests/e2e/integrations/ /workspace/axolotl/tests/e2e/
--- a/cicd/multigpu.py
+++ b/cicd/multigpu.py
@@ -1,7 +1,6 @@
 """
-modal application to run axolotl gpu tests in Modal
-"""
-
+ modal application to run axolotl gpu tests in Modal
+ """
 # pylint: disable=duplicate-code

 import os
@@ -29,7 +28,6 @@ df_args = {
    "CUDA": os.environ.get("CUDA", "121"),
    "GITHUB_REF": os.environ.get("GITHUB_REF", "refs/heads/main"),
    "GITHUB_SHA": os.environ.get("GITHUB_SHA", ""),
-    "CODECOV_TOKEN": os.environ.get("CODECOV_TOKEN", ""),
    "HF_HOME": "/workspace/data/huggingface-cache/hub",
 }

@@ -39,11 +37,15 @@ temp_dir = tempfile.mkdtemp()
 with open(pathlib.Path(temp_dir) / "Dockerfile", "w", encoding="utf-8") as f:
    f.write(dockerfile_contents)

-cicd_image = Image.from_dockerfile(
-    pathlib.Path(temp_dir) / "Dockerfile",
-    force_build=True,
-    gpu="A10G",
-).env(df_args)
+cicd_image = (
+    Image.from_dockerfile(
+        pathlib.Path(temp_dir) / "Dockerfile",
+        force_build=True,
+        gpu="A10G",
+    )
+    .env(df_args)
+    .pip_install("fastapi==0.110.0", "pydantic==2.6.3")
+)

 app = App("Axolotl CI/CD", secrets=[])

@@ -69,7 +71,7 @@ def run_cmd(cmd: str, run_folder: str):
@app.function(
    image=cicd_image,
    gpu=GPU_CONFIG,
-    timeout=90 * 60,
+    timeout=60 * 60,
    cpu=8.0,
    memory=131072 * N_GPUS,
    volumes=VOLUME_CONFIG,
--- a/cicd/multigpu.sh
+++ b/cicd/multigpu.sh
@@ -1,23 +1,5 @@
 #!/bin/bash
 set -e

-# Only run two tests at a time to avoid OOM on GPU (with coverage collection)
-pytest -v -n2 \
-  --ignore=/workspace/axolotl/tests/e2e/multigpu/solo/ \
-  --ignore=/workspace/axolotl/tests/e2e/multigpu/patched/ \
-  /workspace/axolotl/tests/e2e/multigpu/ \
-  --cov=axolotl
-
-# Run solo tests with coverage append
-pytest -v --durations=10 -n1 \
-  /workspace/axolotl/tests/e2e/multigpu/solo/ \
-  --cov=axolotl \
-  --cov-append
-
-pytest -v  --durations=10 -n1 /workspace/axolotl/tests/e2e/multigpu/patched/ \
-  --cov=axolotl \
-  --cov-append \
-  --cov-report=xml:multigpu-coverage.xml
-
-# Upload coverage to Codecov
-codecov upload-process -t "${CODECOV_TOKEN}" -f multigpu-coverage.xml -F multigpu,docker-tests,pytorch-${PYTORCH_VERSION} || true
+# only run one test at a time so as not to OOM the GPU
+pytest -v -n2 /workspace/axolotl/tests/e2e/multigpu/
--- a/cicd/e2e_tests.py
+++ b/cicd/e2e_tests.py
@@ -1,5 +1,4 @@
 """Modal app to run axolotl GPU tests"""
-
 # pylint: disable=duplicate-code

 import os
@@ -28,7 +27,6 @@ df_args = {
    "GITHUB_REF": os.environ.get("GITHUB_REF", "refs/heads/main"),
    "GITHUB_SHA": os.environ.get("GITHUB_SHA", ""),
    "NIGHTLY_BUILD": os.environ.get("NIGHTLY_BUILD", ""),
-    "CODECOV_TOKEN": os.environ.get("CODECOV_TOKEN", ""),
    "HF_HOME": "/workspace/data/huggingface-cache/hub",
 }

--- a/codecov.yml
+++ b/codecov.yml
@@ -1,56 +0,0 @@
-codecov:
-  require_ci_to_pass: yes
-  notify:
-    wait_for_ci: true
-
-coverage:
-  precision: 2
-  round: down
-  range: "70...100"
-  status:
-    project:
-      default:
-        # basic
-        target: auto
-        threshold: 0%
-        base: auto
-        # advanced
-        branches: null
-        if_no_uploads: error
-        if_not_found: success
-        if_ci_failed: error
-        only_pulls: false
-        flags: null
-        paths: null
-    patch:
-      default:
-        # basic
-        target: auto
-        threshold: 0%
-        base: auto
-        # advanced
-        branches: null
-        if_no_uploads: error
-        if_not_found: success
-        if_ci_failed: error
-        only_pulls: false
-        flags: null
-        paths: null
-
-parsers:
-  gcov:
-    branch_detection:
-      conditional: yes
-      loop: yes
-      method: no
-      macro: no
-
-comment:
-  layout: "reach,diff,flags,files,footer"
-  behavior: default
-  require_changes: no
-  require_base: no
-  require_head: yes
-
-github_checks:
-  annotations: false
--- a/docker/Dockerfile
+++ b/docker/Dockerfile
@@ -20,9 +20,9 @@ WORKDIR /workspace/axolotl

 # If AXOLOTL_EXTRAS is set, append it in brackets
 RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install --no-build-isolation -e .[deepspeed,flash-attn,ring-flash-attn,optimizers,ray,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+        pip install --no-build-isolation -e .[deepspeed,flash-attn,optimizers,ray,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
    else \
-        pip install --no-build-isolation -e .[deepspeed,flash-attn,ring-flash-attn,optimizers,ray] $AXOLOTL_ARGS; \
+        pip install --no-build-isolation -e .[deepspeed,flash-attn,optimizers,ray] $AXOLOTL_ARGS; \
    fi

 RUN python scripts/unsloth_install.py | sh
--- a/docker/Dockerfile-base
+++ b/docker/Dockerfile-base
@@ -28,8 +28,8 @@ ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"

 WORKDIR /workspace

-RUN python3 -m pip install --upgrade pip && pip3 install -U packaging==23.2 setuptools==75.8.0 wheel && \
-    python3 -m pip install --no-cache-dir -U torch==${PYTORCH_VERSION}+cu${CUDA} torchvision --extra-index-url https://download.pytorch.org/whl/cu$CUDA && \
+RUN python3 -m pip install --upgrade pip && pip3 install packaging && \
+    python3 -m pip install --no-cache-dir -U torch==${PYTORCH_VERSION}+cu${CUDA} --extra-index-url https://download.pytorch.org/whl/cu$CUDA && \
    python3 -m pip install --no-cache-dir "causal_conv1d @ git+https://github.com/Dao-AILab/causal-conv1d.git@main" && \
    python3 -m pip install --no-cache-dir "mamba_ssm @ git+https://github.com/state-spaces/mamba.git@main"

@@ -37,7 +37,3 @@ RUN git lfs install --skip-repo && \
    pip3 install awscli && \
    # The base image ships with `pydantic==1.8.2` which is not working
    pip3 install -U --no-cache-dir pydantic==1.10.10
-
-RUN if [ "$PYTORCH_VERSION" = "2.7.0" ] ; then \
-        pip3 install flash-attn==2.7.4.post1; \
-    fi
--- a/docker/Dockerfile-base-next
+++ b/docker/Dockerfile-base-next
@@ -1,38 +0,0 @@
-ARG CUDA_VERSION="12.8.1"
-ARG CUDNN_VERSION="8"
-ARG UBUNTU_VERSION="22.04"
-ARG MAX_JOBS=4
-
-FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION AS base-builder
-
-ENV PATH="/root/miniconda3/bin:${PATH}"
-
-ARG PYTHON_VERSION="3.11"
-ARG PYTORCH_VERSION="next"
-ARG CUDA="128"
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
-
-ENV PYTHON_VERSION=$PYTHON_VERSION
-ENV TORCH_CUDA_ARCH_LIST=$TORCH_CUDA_ARCH_LIST
-
-RUN apt-get update \
-    && apt-get install -y wget git build-essential ninja-build git-lfs libaio-dev pkg-config && rm -rf /var/lib/apt/lists/* \
-    && wget \
-    https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
-    && mkdir /root/.conda \
-    && bash Miniconda3-latest-Linux-x86_64.sh -b \
-    && rm -f Miniconda3-latest-Linux-x86_64.sh \
-    && conda create -n "py${PYTHON_VERSION}" python="${PYTHON_VERSION}"
-
-ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
-
-WORKDIR /workspace
-
-RUN python3 -m pip install --upgrade pip && pip3 install packaging && \
-    python3 -m pip install --no-cache-dir -U torch==2.7.0 --extra-index-url https://download.pytorch.org/whl/test/cu$CUDA && \
-    python3 -m pip install --no-cache-dir "causal_conv1d @ git+https://github.com/Dao-AILab/causal-conv1d.git@main" && \
-    python3 -m pip install --no-cache-dir "mamba_ssm @ git+https://github.com/state-spaces/mamba.git@main"
-
-RUN git lfs install --skip-repo && \
-    pip3 install awscli && \
-    pip3 install -U --no-cache-dir pydantic==2.10.6
--- a/docker/Dockerfile-base-nightly
+++ b/docker/Dockerfile-base-nightly
@@ -1,39 +0,0 @@
-ARG CUDA_VERSION="12.8.1"
-ARG CUDNN_VERSION="8"
-ARG UBUNTU_VERSION="22.04"
-ARG MAX_JOBS=4
-
-FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION AS base-builder
-
-ENV PATH="/root/miniconda3/bin:${PATH}"
-
-ARG PYTHON_VERSION="3.11"
-ARG PYTORCH_VERSION="nightly"
-ARG CUDA="128"
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
-
-ENV PYTHON_VERSION=$PYTHON_VERSION
-ENV TORCH_CUDA_ARCH_LIST=$TORCH_CUDA_ARCH_LIST
-
-RUN apt-get update \
-    && apt-get install -y wget git build-essential ninja-build git-lfs libaio-dev pkg-config && rm -rf /var/lib/apt/lists/* \
-    && wget \
-    https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh \
-    && mkdir /root/.conda \
-    && bash Miniconda3-latest-Linux-x86_64.sh -b \
-    && rm -f Miniconda3-latest-Linux-x86_64.sh \
-    && conda create -n "py${PYTHON_VERSION}" python="${PYTHON_VERSION}"
-
-ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
-
-WORKDIR /workspace
-
-RUN python3 -m pip install --upgrade pip && pip3 install packaging && \
-    python3 -m pip install --no-cache-dir -U torch --extra-index-url https://download.pytorch.org/whl/nightly/cu$CUDA && \
-    python3 -m pip install --no-cache-dir "causal_conv1d @ git+https://github.com/Dao-AILab/causal-conv1d.git@main" && \
-    python3 -m pip install --no-cache-dir "mamba_ssm @ git+https://github.com/state-spaces/mamba.git@main"
-
-RUN git lfs install --skip-repo && \
-    pip3 install awscli && \
-    # The base image ships with `pydantic==1.8.2` which is not working
-    pip3 install -U --no-cache-dir pydantic==1.10.10
--- a/docker/Dockerfile-cloud
+++ b/docker/Dockerfile-cloud
@@ -14,7 +14,7 @@ COPY scripts/motd /etc/motd

 RUN pip install jupyterlab notebook ipywidgets && \
    jupyter lab clean
-RUN apt install --yes --no-install-recommends openssh-server tmux iproute2 nvtop && \
+RUN apt install --yes --no-install-recommends openssh-server tmux && \
    mkdir -p ~/.ssh && \
    chmod 700 ~/.ssh && \
    printf "\n[[ -z \"\$TMUX\"  ]] && { tmux attach-session -t ssh_tmux || tmux new-session -s ssh_tmux; exit; }\n" >> ~/.bashrc && \
--- a/docs/.gitignore
+++ b/docs/.gitignore
@@ -1,4 +1,2 @@
 /.quarto/
 _site/
-/api/*.qmd
-/api/*.html
--- a/docs/amd_hpc.qmd
+++ b/docs/amd_hpc.qmd
@@ -1,5 +1,5 @@
 ---
-title: AMD GPUs on HPC Systems
+title: Training with AMD GPUs on HPC Systems
 description: A comprehensive guide for using Axolotl on distributed systems with AMD GPUs
 ---

--- a/docs/cli.qmd
+++ b/docs/cli.qmd
@@ -1,19 +1,28 @@
---
-title: "Command Line Interface (CLI)"
-format:
-  html:
-    toc: true
-    toc-expand: 2
-    toc-depth: 3
-execute:
-  enabled: false
---
+# Axolotl CLI Documentation

 The Axolotl CLI provides a streamlined interface for training and fine-tuning large language models. This guide covers
 the CLI commands, their usage, and common examples.

+### Table of Contents

-## Basic Commands
+- Basic Commands
+- Command Reference
+  - fetch
+  - preprocess
+  - train
+  - inference
+  - merge-lora
+  - merge-sharded-fsdp-weights
+  - evaluate
+  - lm-eval
+- Legacy CLI Usage
+- Remote Compute with Modal Cloud
+  - Cloud Configuration
+  - Running on Modal Cloud
+  - Cloud Configuration Options
+
+
+### Basic Commands

 All Axolotl commands follow this general structure:

@@ -23,9 +32,9 @@ axolotl <command> [config.yml] [options]

 The config file can be local or a URL to a raw YAML file.

-## Command Reference
+### Command Reference

-### fetch
+#### fetch

 Downloads example configurations and deepspeed configs to your local machine.

@@ -40,7 +49,7 @@ axolotl fetch deepspeed_configs
 axolotl fetch examples --dest path/to/folder
 ```

-### preprocess
+#### preprocess

 Preprocesses and tokenizes your dataset before training. This is recommended for large datasets.

@@ -65,7 +74,7 @@ dataset_prepared_path: Local folder for saving preprocessed data
 push_dataset_to_hub: HuggingFace repo to push preprocessed data (optional)
 ```

-### train
+#### train

 Trains or fine-tunes a model using the configuration specified in your YAML file.

@@ -86,38 +95,7 @@ axolotl train config.yml --no-accelerate
 axolotl train config.yml --resume-from-checkpoint path/to/checkpoint
 ```

-It is possible to run sweeps over multiple hyperparameters by passing in a sweeps config.
-
-```bash
-# Basic training with sweeps
-axolotl train config.yml --sweep path/to/sweep.yaml
-```
-
-Example sweep config:
-```yaml
-_:
-  # This section is for dependent variables we need to fix
-  - load_in_8bit: false
-    load_in_4bit: false
-    adapter: lora
-  - load_in_8bit: true
-    load_in_4bit: false
-    adapter: lora
-
-# These are independent variables
-learning_rate: [0.0003, 0.0006]
-lora_r:
-  - 16
-  - 32
-lora_alpha:
-  - 16
-  - 32
-  - 64
-```
-
-
-
-### inference
+#### inference

 Runs inference using your trained model in either CLI or Gradio interface mode.

@@ -137,7 +115,7 @@ cat prompt.txt | axolotl inference config.yml \
    --base-model="./completed-model"
 ```

-### merge-lora
+#### merge-lora

 Merges trained LoRA adapters into the base model.

@@ -159,7 +137,7 @@ gpu_memory_limit: Limit GPU memory usage
 lora_on_cpu: Load LoRA weights on CPU
 ```

-### merge-sharded-fsdp-weights
+#### merge-sharded-fsdp-weights

 Merges sharded FSDP model checkpoints into a single combined checkpoint.

@@ -168,49 +146,36 @@ Merges sharded FSDP model checkpoints into a single combined checkpoint.
 axolotl merge-sharded-fsdp-weights config.yml
 ```

-### evaluate
+#### evaluate

-Evaluates a model's performance (loss etc) on the train and eval datasets.
+Evaluates a model's performance using metrics specified in the config.

 ```bash
 # Basic evaluation
 axolotl evaluate config.yml
 ```

-### lm-eval
+#### lm-eval

 Runs LM Evaluation Harness on your model.

 ```bash
 # Basic evaluation
 axolotl lm-eval config.yml
+
+# Evaluate specific tasks
+axolotl lm-eval config.yml --tasks arc_challenge,hellaswag
 ```

 Configuration options:

 ```yaml
-# List of tasks to evaluate
-lm_eval_tasks:
-  - arc_challenge
-  - hellaswag
-lm_eval_batch_size: # Batch size for evaluation
-output_dir: # Directory to save evaluation results
+lm_eval_tasks: List of tasks to evaluate
+lm_eval_batch_size: Batch size for evaluation
+output_dir: Directory to save evaluation results
 ```

-See [LM Eval Harness](https://github.com/EleutherAI/lm-evaluation-harness) for more details.
-
-### delinearize-llama4
-
-Delinearizes a Llama 4 linearized model into a regular HuggingFace Llama 4 model. This only works with the non-quantized linearized model.
-
-```bash
-axolotl delinearize-llama4 --model path/to/model_dir --output path/to/output_dir
-```
-
-This would be necessary to use with other frameworks. If you have an adapter, merge it with the non-quantized linearized model before delinearizing.
-
-
-## Legacy CLI Usage
+### Legacy CLI Usage

 While the new Click-based CLI is preferred, Axolotl still supports the legacy module-based CLI:

@@ -230,25 +195,19 @@ accelerate launch -m axolotl.cli.inference config.yml \
    --lora_model_dir="./outputs/lora-out" --gradio
 ```

-::: {.callout-important}
-When overriding CLI parameters in the legacy CLI, use same notation as in yaml file (e.g., `--lora_model_dir`).
-
-**Note:** This differs from the new Click-based CLI, which uses dash notation (e.g., `--lora-model-dir`). Keep this in mind if you're referencing newer documentation or switching between CLI versions.
-:::
-
-## Remote Compute with Modal Cloud
+### Remote Compute with Modal Cloud

 Axolotl supports running training and inference workloads on Modal cloud infrastructure. This is configured using a
 cloud YAML file alongside your regular Axolotl config.

-### Cloud Configuration
+#### Cloud Configuration

 Create a cloud config YAML with your Modal settings:

 ```yaml
 # cloud_config.yml
 provider: modal
-gpu: a100       # Supported: l40s, a100-40gb, a100-80gb, a10g, h100, t4, l4
+gpu: a100  # Supported: l40s, a100-40gb, a100-80gb, a10g, h100, t4, l4
 gpu_count: 1    # Number of GPUs to use
 timeout: 86400  # Maximum runtime in seconds (24 hours)
 branch: main    # Git branch to use (optional)
@@ -256,17 +215,13 @@ branch: main    # Git branch to use (optional)
 volumes:        # Persistent storage volumes
  - name: axolotl-cache
    mount: /workspace/cache
-  - name: axolotl-data
-    mount: /workspace/data
-  - name: axolotl-artifacts
-    mount: /workspace/artifacts

-secrets:        # Secrets to inject
+env:            # Environment variables
  - WANDB_API_KEY
  - HF_TOKEN
 ```

-### Running on Modal Cloud
+#### Running on Modal Cloud

 Commands that support the --cloud flag:

@@ -284,30 +239,18 @@ axolotl train config.yml --cloud cloud_config.yml --no-accelerate
 axolotl lm-eval config.yml --cloud cloud_config.yml
 ```

-### Cloud Configuration Options
+#### Cloud Configuration Options

 ```yaml
-provider:    # compute provider, currently only `modal` is supported
-gpu:         # GPU type to use
-gpu_count:   # Number of GPUs (default: 1)
-memory:      # RAM in GB (default: 128)
-timeout:     # Maximum runtime in seconds
-timeout_preprocess: # Preprocessing timeout
-branch:      # Git branch to use
-docker_tag:  # Custom Docker image tag
-volumes:     # List of persistent storage volumes
-
-# Environment variables to pass. Can be specified in two ways:
-# 1. As a string: Will load the value from the host computer's environment variables
-# 2. As a key-value pair: Will use the specified value directly
-# Example:
-# env:
-#   - CUSTOM_VAR  # Loads from host's $CUSTOM_VAR
-#   - {CUSTOM_VAR: "value"}  # Uses "value" directly
-env:
-
-# Secrets to inject. Same input format as `env` but for sensitive data.
-secrets:
-  # - HF_TOKEN
-  # - WANDB_API_KEY
+provider: compute provider, currently only `modal` is supported
+gpu: GPU type to use
+gpu_count: Number of GPUs (default: 1)
+memory: RAM in GB (default: 128)
+timeout: Maximum runtime in seconds
+timeout_preprocess: Preprocessing timeout
+branch: Git branch to use
+docker_tag: Custom Docker image tag
+volumes: List of persistent storage volumes
+env: Environment variables to pass
+secrets: Secrets to inject
 ```
--- a/docs/config.qmd
+++ b/docs/config.qmd
@@ -1,5 +1,5 @@
 ---
-title: Config Reference
+title: Config options
 description: A complete list of all configuration options.
 ---

@@ -30,11 +30,6 @@ tokenizer_legacy:
 # Resize the model embeddings when new tokens are added to multiples of 32
 # This is reported to improve training speed on some models
 resize_token_embeddings_to_32x:
-# Optional[bool] Whether to shrink the embeddings to len(tokenizer). By default, we won't shrink.
-shrink_embeddings:
-# Whether to load the model with randomly initialized weights. Useful for
-# pre-training a model from scratch or debugging purposes.
-random_init_weights:

 # (Internal use only)
 # Used to identify which the model is based on
@@ -88,12 +83,6 @@ gpu_memory_limit: 20GiB
 # Do the LoRA/PEFT loading on CPU -- this is required if the base model is so large it takes up most or all of the available GPU VRAM, e.g. during a model and LoRA merge
 lora_on_cpu: true

-# List[str]. Add plugins to extend the pipeline.
-# See `src/axolotl/integrations` for the available plugins or doc below for more details.
-# https://docs.axolotl.ai/docs/custom_integrations.html
-plugins:
-  # - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
-
 # A list of one or more datasets to finetune the model with
 datasets:
  # HuggingFace dataset repo | s3://,gs:// path | "json" for local dataset, make sure to fill data_files
@@ -109,7 +98,7 @@ datasets:
    preprocess_shards: # Optional[int] process dataset in N sequential chunks for memory efficiency (exclusive with `shards`)

    name: # Optional[str] name of dataset configuration to load
-    split: train # Optional[str] name of dataset split to load from
+    train_on_split: train # Optional[str] name of dataset split to load from
    revision: # Optional[str] The specific revision of the dataset to use when loading from the Hugging Face Hub. This can be a commit hash, tag, or branch name. If not specified, the latest version will be used. This parameter is ignored for local datasets.
    trust_remote_code: # Optional[bool] Trust remote code for untrusted source

@@ -154,10 +143,6 @@ datasets:
    # Key containing the messages (default: "messages")
    field_messages: messages

-    # Key containing the system message (default: "system")
-    # If the system message is not present in the dataset sample, it will be loaded from the field_system property.
-    field_system: system
-
    # Mapping of properties from the input dataset to the chat template.
    # (default: message_property_mappings={'role':'role', 'content':'content'})
    # If a property exists in the template but not in this mapping, the system will attempt
@@ -169,25 +154,19 @@ datasets:
      content: value
      # ...

-    # Optional[Dict[str, List]]. Roles mapping in the messages.
-    # The format is {target_role: [source_roles]}. All source roles will be mapped to the target role.
-    # The default is:
+    message_property_mappings:
+
+    # Optional[Dict[str, List]]. Roles mapping in the messages. The default is:
    roles:
      user: ["human", "user"]
      assistant: ["gpt", "assistant"]
      system: ["system"]
      tool: ["tool"]

-    # Optional[bool]. Whether to drop the system turn from the dataset. Only works with chat_template.
-    # This does not drop the default system message from chat_template if it exists. If you wish to,
-    # we recommend using a custom jinja template with the default system message removed or
-    # adding a system turn with empty content.
-    drop_system_message:
-
    # IMPORTANT: The following fields determine which parts of the conversation to train on.
    # Priority order: message_field_training > message_field_training_detail > train_on_inputs or role in roles_to_train
    # See examples at `docs/dataset-formats/conversation.qmd`
-    # Note: If the below 5 fields are empty, defaults to training only on the last message.
+    # Note: If the below 4 fields are empty, defaults to training only on the last message.

    # Optional[List[str]]. Roles to train on. The tokens from these roles will be considered for the loss.
    roles_to_train: ["assistant"]  # default
@@ -195,14 +174,7 @@ datasets:
    # - all: train on all EOS tokens
    # - turn (default): train on the EOS token at the end of each trainable turn
    # - last: train on the last EOS token in the conversation
-    # TIP: Please make sure that your `tokenizer.eos_token` is same as EOS/EOT token in template. Otherwise, set `eos_token` under `special_tokens`.
-    train_on_eos: turn
-    # Optional[str]. Which EOT (End-of-Turn) tokens to train on in the conversation. Possible values are:
-    # - all: train on all EOT tokens
-    # - turn: train on the EOT token at the end of each trainable turn
-    # - last: train on the last EOT token in the conversation
-    # If not specified, defaults to the value of train_on_eos for backward compatibility.
-    train_on_eot:
+    train_on_eos: last
    # The key in the message turn that indicates via boolean whether tokens of a turn should be considered for training. Useful to selectively train on certain turns besides the `roles_to_train`.
    message_field_training: training
    # The key in the message turn that contains the training details. Useful to selectively train on certain tokens in a turn.
@@ -228,46 +200,10 @@ test_datasets:
    data_files:
      - /workspace/data/eval.jsonl

-# use RL training: 'dpo', 'ipo', 'kto', 'simpo', 'orpo', 'grpo'
+# use RL training: 'dpo', 'ipo', 'kto'
 rl:
-rl_beta:  # Optional[float]. The beta parameter for the RL training.
-
-# dpo
-dpo_use_weighting:  # Optional[bool]. Whether to perform weighting.
-rpo_alpha: # Optional[float]. Weighting of NLL term in loss from RPO paper.
-
-# orpo
-orpo_alpha: 0.1  # Parameter controlling the relative ratio loss weight in the ORPO loss. Passed to `beta` in `ORPOConfig` due to trl mapping.
-
-# kto
-kto_desirable_weight: # Optional[float]. Factor for desirable loss term in KTO loss.
-kto_undesirable_weight: # Optional[float]. Factor for undesirable loss term in KTO loss.
-
-# simpo
-cpo_alpha: 1.0  # Weight of the BC regularizer
-simpo_gamma: 0.5  # Target reward margin for the SimPO loss
-
-# grpo
-trl:
-  use_vllm: # Optional[bool]. Whether to use VLLM for RL training.
-  vllm_server_host: # Optional[str]. Host of the vLLM server to connect to.
-  vllm_server_port: # Optional[int]. Port of the vLLM server to connect to.
-  vllm_server_timeout: # Optional[int]. Total timeout (in seconds) to wait for the vLLM server to respond.
-  vllm_guided_decoding_regex: # Optional[str]. Regex for vLLM guided decoding.
-
-  beta: # Optional[float]. Beta parameter for the RL training. Same as `rl_beta`. Use
-  max_completion_length: # Optional[int]. Maximum length of the completion for RL training.
-
-  reward_funcs: # Optional[list[str]]. List of reward functions to load. Paths must be importable from current dir.
-  reward_weights: # Optional[list[float]]. List of reward weights for the reward functions.
-
-  num_generations: # Optional[int]. Number of generations to sample.
-  log_completions: # Optional[bool]. Whether to log completions.
-
-  sync_ref_model: # Optional[bool]. Whether to sync the reference model.
-  ref_model_mixup_alpha: # Optional[float]. Mixup alpha for the reference model.
-  ref_model_sync_steps: # Optional[int]. Sync steps for the reference model.
-
+# whether to perform weighting if doing DPO training. Boolean.
+dpo_use_weighting:

 # reward modelling: `True` or `False`
 reward_model:
@@ -285,22 +221,13 @@ process_reward_model:
 chat_template: tokenizer_default
 # custom jinja template for chat template. This will be only used if chat_template is set to `jinja` or `null` (in which case chat_template is automatically set to `jinja`). Default is null.
 chat_template_jinja: null
-# Optional[List[str]]. Custom EOT (End-of-Turn) tokens to mask/unmask during training.
-# These tokens mark the boundaries between conversation turns.
-# For example: ["/INST", "</s>", "[/SYSTEM_PROMPT]"]
-# If not specified, defaults to just the model's eos_token.
-# This is useful for templates that use multiple delimiter tokens.
-eot_tokens:
-  # - "</s>"
-  # - "[/INST]"
-  # - "[/SYSTEM_PROMPT]"
 # Changes the default system message
 default_system_message: You are a helpful assistant. Please give a long and detailed answer. # Currently only supports chatml.
 # Axolotl attempts to save the dataset as an arrow after packing the data together so
 # subsequent training attempts load faster, relative path
 dataset_prepared_path: data/last_run_prepared
 # Push prepared dataset to hub
-push_dataset_to_hub: # Optional[str] repo_org/repo_name
+push_dataset_to_hub: # repo path
 # The maximum number of processes to use while preprocessing your input dataset. This defaults to `os.cpu_count()`
 # if not set.
 dataset_processes: # defaults to os.cpu_count() if not set
@@ -341,13 +268,9 @@ total_num_tokens:
 sample_packing_group_size: 100000
 # The number of samples which can be packed into one sequence. Increase if using a large sequence_len with many short samples.
 sample_packing_bin_size: 200
-sample_pack_sequentially: # Optional[bool]. Whether to pack samples sequentially.
-
 # whether to concatenate samples during pretraining
 pretraining_sample_concatenation:

-curriculum_sampling: # Optional[bool]. Whether to use sequential sampling for curriculum learning
-
 # Use batch flattening for speedups when not using sample_packing
 batch_flattening:

@@ -379,27 +302,7 @@ lora_target_modules:
 #  - down_proj
 #  - up_proj
 lora_target_linear: # If true, will target all linear modules
-
-# List[int] | int. # The layer indices to transform, otherwise, apply to all layers
-# https://huggingface.co/docs/peft/v0.15.0/en/package_reference/lora#peft.LoraConfig.layers_to_transform
-peft_layers_to_transform:
-
-# Optional[bool]. Whether to use DoRA.
-# https://huggingface.co/docs/peft/v0.15.0/en/developer_guides/lora#weight-decomposed-low-rank-adaptation-dora
-peft_use_dora:
-
-# Optional[bool]. Whether to use RSLoRA.
-# https://huggingface.co/docs/peft/v0.15.0/en/developer_guides/lora#rank-stabilized-lora
-peft_use_rslora:
-
-# Optional[list[tuple[int, int]]]. List of layer indices to replicate.
-# https://huggingface.co/docs/peft/v0.15.0/en/developer_guides/lora#memory-efficient-layer-replication-with-lora
-peft_layer_replication:
-
-# bool | Literal["gaussian", "eva", "olora", "pissa", "pissa_niter_[number of iters]", "corda", "loftq"]
-# How to initialize LoRA weights. Default to True which is MS original implementation.
-# https://huggingface.co/docs/peft/v0.15.0/en/developer_guides/lora#initialization
-peft_init_lora_weights:
+peft_layers_to_transform: # The layer indices to transform, otherwise, apply to all layers

 # If you added new tokens to the tokenizer, you may need to save some LoRA modules because they need to know the new tokens.
 # For LLaMA and Mistral, you need to save `embed_tokens` and `lm_head`. It may vary for other models.
@@ -413,7 +316,7 @@ lora_fan_in_fan_out: false

 # Apply custom LoRA autograd functions and activation function Triton kernels for
 # speed and memory savings
-# See: https://docs.axolotl.ai/docs/lora_optims.html
+# See: https://axolotl-ai-cloud.github.io/axolotl/docs/lora_optims.html
 lora_mlp_kernel: true
 lora_qkv_kernel: true
 lora_o_kernel: true
@@ -511,7 +414,6 @@ auto_find_batch_size: # Optional[bool]

 eval_table_size: # Approximate number of predictions sent to wandb depending on batch size. Enabled above 0. Default is 0
 eval_max_new_tokens: # Total number of tokens generated for predictions sent to wandb. Default is 128
-do_causal_lm_eval: # Whether to run causal language model evaluation for metrics in `eval_causal_lm_metrics`.
 eval_causal_lm_metrics: # HF evaluate metrics used during evaluation. Default is ["sacrebleu", "comet", "ter", "chrf", "perplexity"]

 profiler_steps: # enable the pytorch profiler to capture the first N steps of training to the output_dir.
@@ -531,8 +433,7 @@ train_on_inputs: false
 # Note that training loss may have an oscillating pattern with this enabled.
 group_by_length: false

-# Whether to use gradient checkpointing. Available options are: true, false, "offload".
-# https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing
+# Whether to use gradient checkpointing https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing
 gradient_checkpointing: false
 # additional kwargs to pass to the trainer for gradient checkpointing
 # gradient_checkpointing_kwargs:
@@ -543,7 +444,7 @@ gradient_checkpointing: false
 early_stopping_patience: 3

 # Specify a scheduler and kwargs to use with the optimizer
-lr_scheduler: # 'one_cycle' | 'rex' | 'log_sweep' | empty for cosine
+lr_scheduler: # 'one_cycle' | 'log_sweep' | empty for cosine
 lr_scheduler_kwargs:
 cosine_min_lr_ratio: # decay lr to some percentage of the peak lr, e.g. cosine_min_lr_ratio=0.1 for 10% of peak lr
 cosine_constant_lr_ratio: # freeze lr at some percentage of the step, e.g. cosine_constant_lr_ratio=0.8 means start cosine_min_lr at 80% of training step (https://arxiv.org/pdf/2308.04014.pdf)
@@ -553,58 +454,36 @@ lr_div_factor: # Learning rate div factor

 # Specify optimizer
 # Valid values are driven by the Transformers OptimizerNames class, see:
-# https://github.com/huggingface/transformers/blob/cbf924b76c03828101a34069a96d209314114fd5/src/transformers/training_args.py#L144-L189
+# https://github.com/huggingface/transformers/blob/95b374952dc27d8511541d6f5a4e22c9ec11fb24/src/transformers/training_args.py#L134
 #
 # Note that not all optimizers may be available in your environment, ex: 'adamw_anyprecision' is part of
 # torchdistx, 'adamw_bnb_8bit' is part of bnb.optim.Adam8bit, etc. When in doubt, it is recommended to start with the optimizer used
 # in the examples/ for your model and fine-tuning use case.
 #
 # Valid values for 'optimizer' include:
+# - adamw_hf
 # - adamw_torch
 # - adamw_torch_fused
 # - adamw_torch_xla
-# - adamw_torch_npu_fused
 # - adamw_apex_fused
-# - adopt_adamw  (an EXPERIMENTAL optimizer, only for torch version >= 2.5.1)
+# - adopt_adamw (an EXPERIMENTAL optimizer, only for torch version >= 2.5.1)
 # - adafactor
 # - adamw_anyprecision
-# - adamw_torch_4bit
-# - ademamix
 # - sgd
 # - adagrad
 # - adamw_bnb_8bit
-# - adamw_8bit   # alias for adamw_bnb_8bit
-# - ademamix_8bit
 # - lion_8bit
 # - lion_32bit
 # - paged_adamw_32bit
 # - paged_adamw_8bit
-# - paged_ademamix_32bit
-# - paged_ademamix_8bit
 # - paged_lion_32bit
 # - paged_lion_8bit
-# - rmsprop
-# - rmsprop_bnb
-# - rmsprop_bnb_8bit
-# - rmsprop_bnb_32bit
 # - galore_adamw
 # - galore_adamw_8bit
 # - galore_adafactor
 # - galore_adamw_layerwise
 # - galore_adamw_8bit_layerwise
 # - galore_adafactor_layerwise
-# - lomo
-# - adalomo
-# - grokadamw
-# - schedule_free_adamw
-# - schedule_free_sgd
-# - apollo_adamw
-# - apollo_adamw_layerwise
-#
-# Additional custom optimizers include:
-# - optimi_adamw
-# - ao_adamw_8bit
-# - ao_adamw_fp8
 optimizer:
 # Dictionary of arguments to pass to the optimizer
 optim_args:
@@ -633,42 +512,27 @@ max_grad_norm:
 # currently only supported on Llama and Mistral
 neftune_noise_alpha:

-# Optional[bool]. Whether to bettertransformers
+# Whether to bettertransformers
 flash_optimum:
-
-# Note: Only one of the following attention patches can be used at a time.
-# For example, if you set `xformers_attention` to `true`, do not set `flash_attention` to `true`.
-
-# Optional[bool]. Whether to use xformers attention patch https://github.com/facebookresearch/xformers:
+# Whether to use xformers attention patch https://github.com/facebookresearch/xformers:
 xformers_attention:
-# Optional[bool]. Whether to use flash attention patch https://github.com/Dao-AILab/flash-attention:
+# Whether to use flash attention patch https://github.com/Dao-AILab/flash-attention:
 flash_attention:
-flash_attn_cross_entropy:  # Optional[bool]. Whether to use flash-attention cross entropy implementation - advanced use only
-flash_attn_rms_norm:  # Optional[bool]. Whether to use flash-attention rms norm implementation - advanced use only
-flash_attn_fuse_qkv: # Optional[bool]. Whether to fuse QKV into a single operation
-flash_attn_fuse_mlp: # Optional[bool]. Whether to fuse part of the MLP into a single operation
-# Optional[bool]. Whether to use scaled-dot-product attention
+flash_attn_cross_entropy:  # Whether to use flash-attention cross entropy implementation - advanced use only
+flash_attn_rms_norm:  # Whether to use flash-attention rms norm implementation - advanced use only
+flash_attn_fuse_qkv: # Whether to fuse QKV into a single operation
+flash_attn_fuse_mlp: # Whether to fuse part of the MLP into a single operation
+# Whether to use scaled-dot-product attention
 # https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html
 sdp_attention:
-# Optional[bool]. Shifted-sparse attention (only llama) - https://arxiv.org/pdf/2309.12307.pdf
+# Shifted-sparse attention (only llama) - https://arxiv.org/pdf/2309.12307.pdf
 s2_attention:
-
-# Optional[bool]. Whether to use low_cpu_mem_usage
-low_cpu_mem_usage:
-# Optional[str]. Resume from a specific checkpoint dir
+# Resume from a specific checkpoint dir
 resume_from_checkpoint:
-# Optional[bool]. If resume_from_checkpoint isn't set and you simply want it to start where it left off.
+# If resume_from_checkpoint isn't set and you simply want it to start where it left off.
 # Be careful with this being turned on between different models.
 auto_resume_from_checkpoints: false

-## Multimodal section
-# int | tuple[int, int] | None . Size to resize images to, width x height.
-# Will read from model/processor config if not set.
-image_size:
-# str. Algorithm to use for image resizing. "bilinear", "bicubic", "lanczos". Default is "bilinear".
-image_resize_algorithm: 'bilinear'
-## End of multimodal section
-
 # Don't mess with this, it's here for accelerate and torchrun
 local_rank:

@@ -680,17 +544,8 @@ special_tokens:
  # unk_token: "<unk>"
  # pad_token: "[PAD]"

-# Optional[list[str]]. Add extra tokens to the tokenizer.
+# Add extra tokens.
 tokens:
-  # - "<|startoftext|>"
-  # - "<|endoftext|>"
-
-# Mapping token_id to new_token_string to override reserved added_tokens in the tokenizer.
-# Only works for tokens that are not part of the base vocab (aka are added_tokens).
-# Can be checked if they exist in tokenizer.json added_tokens.
-added_tokens_overrides:  # Dict[int, str]
-#  128041: "<|im_start|>"
-#  128042: "<|im_end|>"

 # FSDP
 fsdp:
@@ -704,20 +559,6 @@ ddp_timeout:
 ddp_bucket_cap_mb:
 ddp_broadcast_buffers:

-# Sequence parallelism
-# Set to a divisor of the number of GPUs available to split sequences into chunks of equal size.
-# Use in long context training to prevent OOM when sequences cannot fit into a single GPU's VRAM.
-# E.g., if 4 GPUs are available, set this value to 2 to split each sequence into two equal-sized
-# subsequences, or set to 4 to split into four equal-sized subsequences.
-# See https://docs.axolotl.ai/docs/sequence_parallelism.html for more details.
-sequence_parallel_degree:
-# Optional; strides across the key dimension. Larger values use more memory but should make training faster.
-# Must evenly divide the number of KV heads in your model.
-heads_k_stride: 1
-# One of "varlen_llama3", "batch_ring", "batch_zigzag", "batch_stripe". Defaults to "varlen_llama3"
-# in the sample packing case, and "batch_ring" in the non-sample packing case.
-ring_attn_func:
-
 # Path to torch distx for optim 'adamw_anyprecision'
 torchdistx_path:

--- a/docs/custom_integrations.qmd
+++ b/docs/custom_integrations.qmd
@@ -1,101 +0,0 @@
---
-title: Custom Integrations
-toc: true
-toc-depth: 3
---
-
-```{python}
-#| echo: false
-
-import re
-
-def process_readme(integration_name):
-    try:
-        path = f'../src/axolotl/integrations/{integration_name}/README.md'
-        with open(path, 'r') as f:
-            txt = f.read()
-            # Remove h1 headings
-            txt = re.sub(r'^# .*\n?', '', txt, flags=re.MULTILINE)
-            # Convert h2 to h3
-            txt = re.sub(r'^## ', '### ', txt, flags=re.MULTILINE)
-            return txt
-    except FileNotFoundError:
-        return None
-
-def print_section(name, folder_name):
-    output = f"\n## {name}\n"
-    content = process_readme(folder_name)
-    if content:
-        output += content
-    output += f"\nPlease see reference [here](https://github.com/axolotl-ai-cloud/axolotl/tree/main/src/axolotl/integrations/{folder_name})\n"
-    return output
-```
-
-```{python}
-#| output: asis
-#| echo: false
-
-# Introduction text
-print("""
-Axolotl adds custom features through `integrations`. They are located within the `src/axolotl/integrations` directory.
-
-To enable them, please check the respective documentations.
-""")
-
-# Sections
-sections = [
-    ("Cut Cross Entropy", "cut_cross_entropy"),
-    ("Grokfast", "grokfast"),
-    ("Knowledge Distillation (KD)", "kd"),
-    ("Liger Kernels", "liger"),
-    ("Language Model Evaluation Harness (LM Eval)", "lm_eval"),
-    ("Spectrum", "spectrum")
-]
-
-for section_name, folder_name in sections:
-    print(print_section(section_name, folder_name))
-```
-
-## Adding a new integration
-
-Plugins can be used to customize the behavior of the training pipeline through [hooks](https://en.wikipedia.org/wiki/Hooking). See [`axolotl.integrations.BasePlugin`](https://github.com/axolotl-ai-cloud/axolotl/blob/main/src/axolotl/integrations/base.py) for the possible hooks.
-
-To add a new integration, please follow these steps:
-
-1. Create a new folder in the `src/axolotl/integrations` directory.
-2. Add any relevant files (`LICENSE`, `README.md`, `ACKNOWLEDGEMENTS.md`, etc.) to the new folder.
-3. Add `__init__.py` and `args.py` files to the new folder.
-  - `__init__.py` should import the integration and hook into the appropriate functions.
-  - `args.py` should define the arguments for the integration.
-4. (If applicable) Add CPU tests under `tests/integrations` or GPU tests under `tests/e2e/integrations`.
-
-::: {.callout-tip}
-
-See [src/axolotl/integrations/cut_cross_entropy](https://github.com/axolotl-ai-cloud/axolotl/tree/main/src/axolotl/integrations/cut_cross_entropy) for a minimal integration example.
-
-:::
-
-::: {.callout-warning}
-
-If you could not load your integration, please ensure you are pip installing in editable mode.
-
-```bash
-pip install -e .
-```
-
-and correctly spelled the integration name in the config file.
-
-```yaml
-plugins:
-  - axolotl.integrations.your_integration_name.YourIntegrationPlugin
-```
-
-:::
-
-::: {.callout-note}
-
-It is not necessary to place your integration in the `integrations` folder. It can be in any location, so long as it's installed in a package in your python env.
-
-See this repo for an example: [https://github.com/axolotl-ai-cloud/diff-transformer](https://github.com/axolotl-ai-cloud/diff-transformer)
-
-:::
--- a/docs/dataset-formats/conversation.qmd
+++ b/docs/dataset-formats/conversation.qmd
@@ -4,6 +4,16 @@ description: Conversation format for supervised fine-tuning.
 order: 3
 ---

+## sharegpt
+
+IMPORTANT: ShareGPT is deprecated!. Please see [chat_template](#chat_template) section below.
+
+## pygmalion
+
+```{.json filename="data.jsonl"}
+{"conversations": [{"role": "...", "value": "..."}]}
+```
+
 ## chat_template

 Chat Template strategy uses a jinja2 template that converts a list of messages into a prompt. Support using tokenizer's template, a supported template, or custom jinja2.
@@ -52,7 +62,7 @@ We recommend checking the below examples for other usecases.

 ### Examples

-1. (Legacy) Using the default chat template in the tokenizer_config.json on OpenAI messages format, training on only last message.
+1. Using the default chat template in the tokenizer_config.json on OpenAI messages format, training on only last message.

 ```yaml
 datasets:
@@ -62,10 +72,6 @@ datasets:
    train_on_eos:
 ```

-::: {.callout-tip}
-If you receive an error like "`chat_template` choice is `tokenizer_default` but tokenizer's `chat_template` is null.", it means the tokenizer does not have a default `chat_template`. Follow the examples below instead to set a custom `chat_template`.
-:::
-
 2. Using the `gemma` chat template to override the tokenizer_config.json's chat template on OpenAI messages format, training on all assistant messages.

 ```yaml
@@ -96,56 +102,7 @@ datasets:
    type: chat_template
 ```

-::: {.callout-important}
-Please make sure that your `tokenizer.eos_token` is same as EOS (End-of-Sequence) token in template. Otherwise, set `eos_token` under `special_tokens: `.
-:::
-
-5. If you are using a template that has a different EOT (End-of-Turn) token from EOS token or multiple EOT tokens (like Mistral V7 Tekken), set the `eot_tokens: ` config. The handling of EOT tokens follows `train_on_eos: ` which defaults to turn.
-
-```yaml
-eot_tokens:
-  - "[/INST]"
-  # - "[/SYSTEM_PROMPT]"
-
-datasets:
-  - path: ...
-    type: chat_template
-
-    # optional
-    train_on_eot: turn  # defaults read from train_on_eos (which defaults to turn)
-```
-
-::: {.callout-tip}
-See [config documentation](../config.qmd) for detailed explanations of "turn", "last", and "all" options for training on tokens.
-:::
-
-::: {.callout-note}
-Using `eot_tokens` requires each token that exists in `chat_template` to be a single token in the tokenizer. Otherwise, the tokenizer will split the token and cause unexpected behavior.
-
-You can add those tokens as new tokens under `tokens: ` or (recommended) override unused added_tokens via `added_tokens_overrides: `. See [config](../config.qmd) for more details.
-:::
-
-6. Continuing from the previous example, if you want to train on all EOT token trainable turns but only last EOS token, set `train_on_eos: last`.
-
-```yaml
-eot_tokens:
-  - "[/INST]"
-  # ...
-
-datasets:
-  - path: ...
-    type: chat_template
-
-    train_on_eos: last
-    train_on_eot: turn
-```
-
-::: {.callout-tip}
-If EOS token only appears at the end of a prompt, `train_on_eos: last` is equivalent to `train_on_eos: turn`. Therefore, generally, you can leave them to their defaults and omit them.
-:::
-
-
-7. (Advanced) Using fine-grained control over tokens and turns to train in a conversation
+5. (Advanced) Using fine-grained control over tokens and turns to train in a conversation

 For a data sample that looks like:

@@ -192,18 +149,4 @@ datasets:
    message_field_training_detail: train_detail
 ```

-::: {.callout-tip}
-It is not necessary to set both `message_field_training` and `message_field_training_detail` at once.
-:::
-
-## sharegpt
-
-::: {.callout-important}
-ShareGPT is deprecated!. Please see [chat_template](#chat_template) section.
-:::
-
-## pygmalion
-
-```{.json filename="data.jsonl"}
-{"conversations": [{"role": "...", "value": "..."}]}
-```
+Tip: It is not necessary to use both `message_field_training` and `message_field_training_detail` at a time.
--- a/docs/dataset-formats/index.qmd
+++ b/docs/dataset-formats/index.qmd
@@ -13,14 +13,7 @@ As there are a lot of available options in Axolotl, this guide aims to provide a

 Axolotl supports 3 kinds of training methods: pre-training, supervised fine-tuning, and preference-based post-training (e.g. DPO, ORPO, PRMs). Each method has their own dataset format which are described below.

-::: {.callout-tip}
-
-This guide will mainly use JSONL as an introduction. Please refer to the [dataset loading docs](../dataset_loading.qmd) to understand how to load datasets from other sources.
-
-For `pretraining_dataset:` specifically, please refer to the [Pre-training section](#pre-training).
-:::
-
-## Pre-training
+## [Pre-training](pretraining.qmd)

 When aiming to train on large corpora of text datasets, pre-training is your go-to choice. Due to the size of these datasets, downloading the entire-datasets before beginning training would be prohibitively time-consuming. Axolotl supports [streaming](https://huggingface.co/docs/datasets/en/stream) to only load batches into memory at a time.

@@ -103,10 +96,6 @@ One step is equal to `sequence_len * micro_batch_size * gradient_accumulation_st

 It is recommended to leave this off if downloading from Hugging Face hub as it would download the entire dataset which can be very large.

-### Reference
-
-Please see docs [here](pretraining.qmd).
-
 ## Supervised fine-tuning (SFT)

 Supervised fine-tuning is the process of training models to respond to an instruction or chat input.
@@ -131,12 +120,11 @@ If you went through the flow chart and did not find one that matches, it is reco
 You can mix and match within each approach or across approaches to train a model on a variety of datasets.
 :::

-### Pre-Tokenized Dataset
+### [Pre-Tokenized Dataset](tokenized.qmd)

 We suggest this approach when you want to bring your own tokenized dataset.

 Axolotl expects the dataset to have three keys:
-
 - `input_ids`: from tokenizing formatted prompt
 - `attention_mask`: for masking padding. If you don't add padding, it would be equal to `len(input_ids) * [1]`
 - `labels`: this is the same as `input_ids`, however, if you want to mask certain tokens, you would set those indices to `-100`.
@@ -157,9 +145,7 @@ datasets:
 `type: ` is empty!
 :::

-Reference: [Pre-Tokenized Dataset Documentation](tokenized.qmd).
-
-### Template Free Dataset
+### [Template Free Dataset](template_free.qmd)

 We reccomend this approach when you want granular control over the prompt formatting, special tokens, and masking, whilst letting Axolotl handle the tokenization. This is very useful if your dataset has unique prompts that differ across samples and where one single general template wouldn't suffice.

@@ -196,9 +182,7 @@ datasets:
    type: input_output
 ```

-Reference: [Template Free Documentation](template_free.qmd).
-
-### Conversation Dataset
+### [Conversation Dataset](conversation.qmd)

 `conversation` messages are a list of messages which usually contain a `role` and `content` key.

@@ -274,7 +258,7 @@ Newer conversation datasets usually follow the OpenAI format.

 Axolotl supports both as well as allowing customization of any kind of key.

-#### Chat Template Usage
+#### [Chat Template Usage](conversation.qmd#chat_template)

 To properly use this method, it is important to identify three things:

@@ -356,19 +340,9 @@ datasets:
      narrator: ["narrator"]
 ```

-::: {.callout-tip}
-As chat_templates may use hardcoded EOS/EOT tokens that are different from the tokenizer's EOS, it is highly recommended to set them. For example, `ChatML` uses `<|im_end|>` to end turns.
+#### Applying `chat_template`

-```yaml
-special_tokens:
-  eos_token: <|im_end|>
-```
-
-:::
-
-##### Applying `chat_template`
-
-Once all the above steps are completed, you could combine all these configs together to form a bespoke configuration for your custom dataset.
+Once all the above steps are completed, you could combine all these configs together to form a bespoke configuration for your  custom dataset. The final step would be to correctly set the EOS token in your config:

 ```yaml
 datasets:
@@ -417,17 +391,7 @@ If this config were to be applied to the sample dataset above, the output would

 The first number refers to the label, the second refers to the `token_id`. For example, `-100` labels appear on non-assistant portions, meaning that they are masked during. For assistant portions, the label is the same as the `token_id`.

-::: {.callout-note}
-
-If during `preprocess`, there are a lot of warnings of `Could not find content __ boundary`, please check the FAQ section for [chat_templates](../faq.qmd#chat-templates).
-
-:::
-
-#### Reference
-
-Please see docs [here](conversation.qmd).
-
-### Instruction Dataset
+### [Instruction Dataset](inst_tune.qmd)

 Instruction datasets are used to train instruction-following models and comprise a prompt, containing an instruction, and a single response. In contrast to chat datasets which may be multi-turn, instruct datasets are typically single-turn.

@@ -457,7 +421,7 @@ datasets:
    type: alpaca
 ```

-Axolotl supports many kinds of instruction dataset. All of them can be found in the [Instruction Dataset Documentation](inst_tune.qmd) with their respective type and sample row format.
+Axolotl supports many kinds of instruction dataset. All of them can be found here (https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/inst_tune.html) with their respective type and sample row format.

 #### Custom Instruct Prompt Format

@@ -489,8 +453,6 @@ datasets:

 The config sets that the `field_instruction` is actually named `input`, and the `field_input` is empty as we don't have an `input` in this sample. Generally, `instruction` can be thought as the question to the model, and `input` as the additional information with `output` being the response. It is not necessary to have an `input` nor `system`. In the end, the most important part is to understand what format you want it to look like and how you can customize this to your use case.

-Reference: [Custom Instruct Prompt Format Documentation](inst_tune.qmd#how-to-add-custom-prompt-format).
-
 ## Reinforcement Learning from Human Feedback (RLHF)

-As there are multiple RLHF methods with their own dataset requirements. Please see [RLHF documentation](../rlhf.qmd) for more detail.
+As there are multiple RLHF methods with their own dataset requirements. Please see [RLHF datasets](../rlhf.qmd) documentation for more detail.
--- a/docs/dataset-formats/pretraining.qmd
+++ b/docs/dataset-formats/pretraining.qmd
@@ -27,6 +27,7 @@ pretraining_dataset:
    type: pretrain
    trust_remote_code:
    skip: # number of rows of data to skip over from the beginning
+...
 ```

 :::
--- a/docs/dataset-formats/template_free.qmd
+++ b/docs/dataset-formats/template_free.qmd
@@ -1,239 +1,7 @@
 ---
 title: Template-Free
 description: Construct prompts without a template.
-toc: true
-toc-depth: 3
 order: 4
 ---

-## Background {#sec-background}
-
-### Masking Inputs {#masking-inputs}
-
-One of the most popular features of
-[axolotl](https://github.com/axolotl-ai-cloud/axolotl) is
-setting the following configuration value:
-
-
-```yaml
-train_on_inputs: false
-```
-
-If you declare a [dataset formats](https://github.com/axolotl-ai-cloud/axolotl?tab=readme-ov-file#dataset)
-such as `alpaca` or `chatml`, axolotl knows what is an input
-(i.e. human) vs. an output (i.e. the assistant) and masks the input
-labels so that your model can focus on predicting the outputs only.
-
-### You may not want prompt templates {#sec-you-may-not-want-prompt-templates}
-
-However, there are many situations where you don't want to use one of
-these formats or templates. This is because they can:
-
-   Add unnecessary boilerplate to your prompts.
-   Create artifacts like special delimiters `<|im_start|>` that can
-    quickly become footguns if you don't include them correctly at
-    inference time.
-   Enforce a *chat* interface when you do not want one. Sometimes you
-    just want to fine-tune a model to a very specific task and do NOT
-    want multi-turn conversations, roles, etc.
-   Limit you to only certain roles that the template allows.
-
-### The `input_output` format {#sec-the-inputoutput-format}
-
-You can construct your prompts without a template by using the
-`input_output` format, by setting `type: input_output` in your
-configuration file like this:
-
-**config.yml**
-
-```yaml
-train_on_inputs: false # Mask segments of your data
-datasets:
-  - path: output.jsonl
-    type: input_output  # use template free prompt construction
-```
-
-Unlike `type: completion`, which is also template-free,
-`type: input_output` allows you to mask segments of your text. More
-details on how this works are described below.
-
-## Usage {#sec-usage}
-
-This is how you can use the `input_output` format:
-
-### 1. Prepare Data {#sec-1-prepare-data}
-
-To use the `input_output` format, collect your data in the following
-format into a jsonl file (below is the first row from the file
-`output`.jsonl` pretty printed):
-
-```bash
-$ head -n1 output.jsonl | python -m json.tool
-```
-
-:::{.cell-output .cell-output-stdout}
-    {
-        "segments": [
-            {
-                "label": true,
-                "text": "<s>Hello\n"
-            },
-            {
-                "label": true,
-                "text": "hi there!. "
-            },
-            {
-                "label": false,
-                "text": "goodbye "
-            },
-            {
-                "label": true,
-                "text": "farewell</s>"
-            }
-        ]
-    }
-:::
-
-Set `label:false` when you want to mask a segment of text so that the
-model isn't trained on it. Some things to keep in mind:
-
-> [!IMPORTANT]
-> 1.  **EOS, BOS, spaces, newlines etc. are entirely up to you. Axolotl
-    concatenates all the segments as-is.** The tokenizer doesn't add
-    anything additional. Notice how I added spaces, newlines, `<s>`
-    (BOS), and `</s>` (EOS) myself.
-> 2.  Make sure you check the materialized output to validate that the
-    prompt is getting assembled how you like.
-
-### 2. Use `type: input_output` {#sec-2-use-type-inputoutput}
-
-Let's materialize data with our `output.jsonl` file by setting
-`type: input_output` in our axolotl config:
-
-```yaml
-# training_config.yaml
-base_model: mistralai/Mistral-7B-v0.1
-data_seed: 49
-seed: 49
-
-datasets:
-  - path: output.jsonl
-    type: input_output
-val_set_size: 0.1
-
-sequence_len: 896
-sample_packing: false
-
-micro_batch_size: 2
-gradient_accumulation_steps: 3
-eval_batch_size: 2
-num_epochs: 1
-learning_rate: 0.0002
-
-train_on_inputs: false
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"
-```
-
-You can use the following command to materialize your data. The
-`--debug` flag will print the tokens, along with the labels so you can
-verify that the correct items are being ignored:
-
-```bash
-axolotl preprocess training_config.yaml --debug
-
-...
-[2024-03-05 23:36:46,969] [INFO] [axolotl.check_example_labels:35] [PID:607731] [RANK:0] <s>(1, 1) Hello(22557, 22557)
-(13, 13) hi(12014, 12014) there(736, 736) !(28808, 28808) .(28723, 28723) (28705, 28705) good(-100, 1179) bye(-100, 17664) (-100, 28705) fare(19111, 19111) well(5458, 5458) </s>(2, 2)
-
-```
-
-The format is `decoded_token`(`label`, `token_id`), for example,
-`<s>(1, 1)` means that the token is `<s>`, the label is `1` and the
-token_id is `1`. When the label is `-100` then that token is ignored for
-training.
-
-### 3. Check the prompts {#sec-3-check-the-prompts}
-
-Here is another way to check the materialized output:
-
-```python
-from transformers import AutoTokenizer
-from datasets import load_from_disk
-import yaml
-
-directory = !ls last_run_prepared/
-with open('training_config.yaml', 'r') as f:
-    cfg = yaml.safe_load(f)
-model_id = cfg['base_model']
-tok = AutoTokenizer.from_pretrained(model_id)
-ds = load_from_disk(f'last_run_prepared/{directory[0]}/')
-```
-
-```python
->>> row = ds[0]
->>> print(tok.decode(row['input_ids']))
-<s> Hello
-    hi there!.  goodbye  farewell</s>
-```
-
-We can check that the right tokens are ignored by comparing the labels
-to each token:
-
-```python
-import pandas as pd
-pd.DataFrame([{'token': tok.decode(i), 'label': l, 'id':i} for i,l in
-              zip(row['input_ids'], row['labels'])])
-```
-
-| token | label | id    |
-|-------|-------|-------|
-| 0     | \<s\> | 1     |
-| 1     | Hello | 22557 |
-| 2     | \\n   | 13    |
-| 3     | hi    | 12014 |
-| 4     | there | 736   |
-| 5     | !     | 28808 |
-| 6     | .     | 28723 |
-| 7     |       | 28705 |
-| 8     | good  | -100  |
-| 9     | bye   | -100  |
-| 10    |       | -100  |
-| 11    | fare  | 19111 |
-| 12    | well  | 5458  |
-| 13    | \</s\>| 2     |
-
-
-
-If we look at the input data, the above table seems correct! (The jsonl
-version is repeated below for reference):
-
-
-```bash
-$ head -n1 output.jsonl | python -m json.tool
-```
-
-:::{.cell-output .cell-output-stdout}
-    {
-        "segments": [
-            {
-                "label": true,
-                "text": "<s>Hello\n"
-            },
-            {
-                "label": true,
-                "text": "hi there!. "
-            },
-            {
-                "label": false,
-                "text": "goodbye "
-            },
-            {
-                "label": true,
-                "text": "farewell</s>"
-            }
-        ]
-    }
-:::
+See [these docs](../input_output.qmd).
--- a/docs/dataset_loading.qmd
+++ b/docs/dataset_loading.qmd
@@ -1,276 +0,0 @@
---
-title: Dataset Loading
-description: Understanding how to load datasets from different sources
-back-to-top-navigation: true
-toc: true
-toc-depth: 5
---
-
-## Overview
-
-Datasets can be loaded in a number of different ways depending on the how it is saved (the extension of the file) and where it is stored.
-
-## Loading Datasets
-
-We use the `datasets` library to load datasets and a mix of `load_dataset` and `load_from_disk` to load them.
-
-You may recognize the similar named configs between `load_dataset` and the `datasets` section of the config file.
-
-```yaml
-datasets:
-  - path:
-    name:
-    data_files:
-    split:
-    revision:
-    trust_remote_code:
-```
-
-::: {.callout-tip}
-
-Do not feel overwhelmed by the number of options here. A lot of them are optional. In fact, the most common config to use would be `path` and sometimes `data_files`.
-
-:::
-
-This matches the API of [`datasets.load_dataset`](https://github.com/huggingface/datasets/blob/0b5998ac62f08e358f8dcc17ec6e2f2a5e9450b6/src/datasets/load.py#L1838-L1858), so if you're familiar with that, you will feel right at home.
-
-For HuggingFace's guide to load different dataset types, see [here](https://huggingface.co/docs/datasets/loading).
-
-For full details on the config, see [config.qmd](config.qmd).
-
-::: {.callout-note}
-
-You can set multiple datasets in the config file by more than one entry under `datasets`.
-
-```yaml
-datasets:
-  - path: /path/to/your/dataset
-  - path: /path/to/your/other/dataset
-```
-
-:::
-
-### Local dataset
-
-#### Files
-
-Usually, to load a JSON file, you would do something like this:
-
-```python
-from datasets import load_dataset
-
-dataset = load_dataset("json", data_files="data.json")
-```
-
-Which translates to the following config:
-
-```yaml
-datasets:
-  - path: json
-    data_files: /path/to/your/file.jsonl
-```
-
-However, to make things easier, we have added a few shortcuts for loading local dataset files.
-
-You can just point the `path` to the file or directory along with the `ds_type` to load the dataset. The below example shows for a JSON file:
-
-```yaml
-datasets:
-  - path: /path/to/your/file.jsonl
-    ds_type: json
-```
-
-This works for CSV, JSON, Parquet, and Arrow files.
-
-::: {.callout-tip}
-
-If `path` points to a file and `ds_type` is not specified, we will automatically infer the dataset type from the file extension, so you could omit `ds_type` if you'd like.
-
-:::
-
-#### Directory
-
-If you're loading a directory, you can point the `path` to the directory.
-
-Then, you have two options:
-
-##### Loading entire directory
-
-You do not need any additional configs.
-
-We will attempt to load in the following order:
- datasets saved with `datasets.save_to_disk`
- loading entire directory of files (such as with parquet/arrow files)
-
-```yaml
-datasets:
-  - path: /path/to/your/directory
-```
-
-##### Loading specific files in directory
-
-Provide `data_files` with a list of files to load.
-
-```yaml
-datasets:
-    # single file
-  - path: /path/to/your/directory
-    ds_type: csv
-    data_files: file1.csv
-
-    # multiple files
-  - path: /path/to/your/directory
-    ds_type: json
-    data_files:
-      - file1.jsonl
-      - file2.jsonl
-
-    # multiple files for parquet
-  - path: /path/to/your/directory
-    ds_type: parquet
-    data_files:
-      - file1.parquet
-      - file2.parquet
-
-```
-
-### HuggingFace Hub
-
-The method you use to load the dataset depends on how the dataset was created, whether a folder was uploaded directly or a HuggingFace Dataset was pushed.
-
-::: {.callout-note}
-
-If you're using a private dataset, you will need to enable the `hf_use_auth_token` flag in the root-level of the config file.
-
-:::
-
-#### Folder uploaded
-
-This would mean that the dataset is a single file or file(s) uploaded to the Hub.
-
-```yaml
-datasets:
-  - path: org/dataset-name
-    data_files:
-      - file1.jsonl
-      - file2.jsonl
-```
-
-#### HuggingFace Dataset
-
-This means that the dataset is created as a HuggingFace Dataset and pushed to the Hub via `datasets.push_to_hub`.
-
-```yaml
-datasets:
-  - path: org/dataset-name
-```
-
-::: {.callout-note}
-
-There are some other configs which may be required like `name`, `split`, `revision`, `trust_remote_code`, etc depending on the dataset.
-
-:::
-
-### Remote Filesystems
-
-Via the `storage_options` config under `load_dataset`, you can load datasets from remote filesystems like S3, GCS, Azure, and OCI.
-
-::: {.callout-warning}
-
-This is currently experimental. Please let us know if you run into any issues!
-
-:::
-
-The only difference between the providers is that you need to prepend the path with the respective protocols.
-
-```yaml
-datasets:
-    # Single file
-  - path: s3://bucket-name/path/to/your/file.jsonl
-
-    # Directory
-  - path: s3://bucket-name/path/to/your/directory
-```
-
-For directory, we load via `load_from_disk`.
-
-#### S3
-
-Prepend the path with `s3://`.
-
-The credentials are pulled in the following order:
-
- `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY`, and `AWS_SESSION_TOKEN` environment variables
- from the `~/.aws/credentials` file
- for nodes on EC2, the IAM metadata provider
-
-::: {.callout-note}
-
-We assume you have credentials setup and not using anonymous access. If you want to use anonymous access, let us know! We may have to open a config option for this.
-
-:::
-
-Other environment variables that can be set can be found in [boto3 docs](https://boto3.amazonaws.com/v1/documentation/api/latest/guide/configuration.html#using-environment-variables)
-
-#### GCS
-
-Prepend the path with `gs://` or `gcs://`.
-
-The credentials are loaded in the following order:
-
- gcloud credentials
- for nodes on GCP, the google metadata service
- anonymous access
-
-#### Azure
-
-##### Gen 1
-
-Prepend the path with `adl://`.
-
-Ensure you have the following environment variables set:
-
- `AZURE_STORAGE_TENANT_ID`
- `AZURE_STORAGE_CLIENT_ID`
- `AZURE_STORAGE_CLIENT_SECRET`
-
-##### Gen 2
-
-Prepend the path with `abfs://` or `az://`.
-
-Ensure you have the following environment variables set:
-
- `AZURE_STORAGE_ACCOUNT_NAME`
- `AZURE_STORAGE_ACCOUNT_KEY`
-
-Other environment variables that can be set can be found in [adlfs docs](https://github.com/fsspec/adlfs?tab=readme-ov-file#setting-credentials)
-
-#### OCI
-
-Prepend the path with `oci://`.
-
-It would attempt to read in the following order:
-
- `OCIFS_IAM_TYPE`, `OCIFS_CONFIG_LOCATION`, and `OCIFS_CONFIG_PROFILE` environment variables
- when on OCI resource, resource principal
-
-Other environment variables:
-
- `OCI_REGION_METADATA`
-
-Please see the [ocifs docs](https://ocifs.readthedocs.io/en/latest/getting-connected.html#Using-Environment-Variables).
-
-### HTTPS
-
-The path should start with `https://`.
-
-```yaml
-datasets:
-  - path: https://path/to/your/dataset/file.jsonl
-```
-
-This must be publically accessible.
-
-## Next steps
-
-Now that you know how to load datasets, you can learn more on how to load your specific dataset format into your target output format [dataset formats docs](dataset-formats).
--- a/docs/dataset_preprocessing.qmd
+++ b/docs/dataset_preprocessing.qmd
@@ -3,11 +3,8 @@ title: Dataset Preprocessing
 description: How datasets are processed
 ---

-## Overview
-
 Dataset pre-processing is the step where Axolotl takes each dataset you've configured alongside
-the [dataset format](dataset-formats) and prompt strategies to:
-
+the (dataset format)[../dataset-formats/] and prompt strategies to:
 - parse the dataset based on the *dataset format*
 - transform the dataset to how you would interact with the model based on the *prompt strategy*
 - tokenize the dataset based on the configured model & tokenizer
@@ -15,12 +12,10 @@ the [dataset format](dataset-formats) and prompt strategies to:

 The processing of the datasets can happen one of two ways:

-1. Before kicking off training by calling `axolotl preprocess config.yaml --debug`
+1. Before kicking off training by calling `python -m axolotl.cli.preprocess /path/to/your.yaml --debug`
 2. When training is started

-### What are the benefits of pre-processing?
-
-When training interactively or for sweeps
+What are the benefits of pre-processing? When training interactively or for sweeps
 (e.g. you are restarting the trainer often), processing the datasets can oftentimes be frustratingly
 slow. Pre-processing will cache the tokenized/formatted datasets according to a hash of dependent
 training parameters so that it will intelligently pull from its cache when possible.
@@ -33,12 +28,8 @@ default path of `./last_run_prepared/`, but will ignore anything already cached
 setting `dataset_prepared_path: ./last_run_prepared`, the trainer will use whatever pre-processed
 data is in the cache.

-### What are the edge cases?
-
-Let's say you are writing a custom prompt strategy or using a user-defined
+What are the edge cases? Let's say you are writing a custom prompt strategy or using a user-defined
 prompt template. Because the trainer cannot readily detect these changes, we cannot change the
-calculated hash value for the pre-processed dataset.
-
-If you have `dataset_prepared_path: ...` set
+calculated hash value for the pre-processed dataset. If you have `dataset_prepared_path: ...` set
 and change your prompt templating logic, it may not pick up the changes you made and you will be
 training over the old prompt.
--- a/docs/debugging.qmd
+++ b/docs/debugging.qmd
@@ -31,13 +31,11 @@ While debugging it's helpful to simplify your test scenario as much as possible.
    - Set `CUDA_VISIBLE_DEVICES` to a single GPU, ex: `export CUDA_VISIBLE_DEVICES=0`.
    - Set `dataset_processes: 1` in your axolotl config or run the training command with `--dataset_processes=1`.
 2. **Use a small dataset**: Construct or use a small dataset from HF Hub. When using a small dataset, you will often have to make sure `sample_packing: False` and `eval_sample_packing: False` to avoid errors.  If you are in a pinch and don't have time to construct a small dataset but want to use from the HF Hub, you can shard the data (this will still tokenize the entire dataset, but will only use a fraction of the data for training.  For example, to shard the dataset into 20 pieces, add the following to your axolotl config):
-
    ```yaml
-    datasets:
+    dataset:
        ...
        shards: 20
    ```
-
 3. **Use a small model**: A good example of a small model is [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0).
 4. **Minimize iteration time**: Make sure the training loop finishes as fast as possible, with these settings.
    - `micro_batch_size: 1`
@@ -87,7 +85,7 @@ The easiest way to get started is to modify the [.vscode/launch.json](../.vscode

 For example, to mimic the command `cd devtools && CUDA_VISIBLE_DEVICES=0 accelerate launch -m axolotl.cli.train dev_chat_template.yml`, you would use the below configuration[^1].  Note that we add additional flags that override the axolotl config and incorporate the tips above (see the comments). We also set the working directory to `devtools` and set the `env` variable `HF_HOME` to a temporary folder that is later partially deleted.  This is because we want to delete the HF dataset cache before each run in order to ensure that the data preprocessing code is run from scratch.

-```json
+```jsonc
 // .vscode/launch.json
 {
    "version": "0.2.0",
@@ -134,7 +132,7 @@ For example, to mimic the command `cd devtools && CUDA_VISIBLE_DEVICES=0 acceler

 Below is the [./vscode/tasks.json](../.vscode/tasks.json) file that defines the `cleanup-for-dataprep` task.  This task is run before each debugging session when you use the above configuration.  Note how there are two tasks that delete the two folders mentioned above.  The third task `cleanup-for-dataprep` is a composite task that combines the two tasks.  A composite task is necessary because VSCode does not allow you to specify multiple tasks in the `preLaunchTask` argument of the `launch.json` file.

-```json
+```jsonc
 // .vscode/tasks.json
 // this file is used by launch.json
 {
--- a/docs/docker.qmd
+++ b/docs/docker.qmd
@@ -1,142 +0,0 @@
---
-title: "Docker"
-format:
-  html:
-    toc: true
-    toc-depth: 4
---
-
-This section describes the different Docker images that are released by AxolotlAI at [Docker Hub](https://hub.docker.com/u/axolotlai).
-
-## Base
-
-The base image is the most minimal image that can install Axolotl. It is based on the `nvidia/cuda` image. It includes python, torch, git, git-lfs, awscli, pydantic, and more.
-
-#### Image
-
-```
-axolotlai/axolotl-base
-```
-
-Link: [Docker Hub](https://hub.docker.com/r/axolotlai/axolotl-base)
-
-#### Tags format
-
-```bash
-main-base-py{python_version}-cu{cuda_version}-{pytorch_version}
-```
-
-Tags examples:
-
- `main-base-py3.11-cu128-2.7.0`
- `main-base-py3.11-cu126-2.7.0`
- `main-base-py3.11-cu124-2.6.0`
- `main-base-py3.11-cu124-2.5.1`
- `main-base-py3.11-cu124-2.4.1`
-
-## Main
-
-The main image is the image that is used to run Axolotl. It is based on the `axolotlai/axolotl-base` image and includes the Axolotl codebase, dependencies, and more.
-
-#### Image
-
-```
-axolotlai/axolotl
-```
-
-Link: [Docker Hub](https://hub.docker.com/r/axolotlai/axolotl)
-
-#### Tags format {#sec-main-tags}
-
-```bash
-# on push to main
-main-py{python_version}-cu{cuda_version}-{pytorch_version}
-
-# latest main (currently torch 2.6.0, python 3.11, cuda 12.4)
-main-latest
-
-# nightly build
-{branch}-{date_in_YYYYMMDD}-py{python_version}-cu{cuda_version}-{pytorch_version}
-
-# tagged release
-{version}
-```
-
-:::{.callout-tip}
-
-There may be some extra tags appended to the image, like `-vllm` which installs those packages.
-
-:::
-
-Tags examples:
-
- `main-py3.11-cu126-2.7.0`
- `main-py3.11-cu124-2.6.0`
- `main-py3.11-cu124-2.5.1`
- `main-py3.11-cu124-2.4.1`
- `main-latest`
- `main-20250303-py3.11-cu124-2.6.0`
- `main-20250303-py3.11-cu124-2.5.1`
- `main-20250303-py3.11-cu124-2.4.1`
- `0.7.1`
-
-## Cloud
-
-The cloud image is the image that is used to run Axolotl in the cloud. It is based on the `axolotlai/axolotl` image and sets ENV variables like HuggingFace cache directories for volume mounts, tmux, and more for different cloud providers.
-
-:::{.callout-tip}
-
-Jupyter lab is run by default. Set `JUPYTER_DISABLE=1` in the environment variables to disable it.
-
-:::
-
-#### Image
-
-```
-axolotlai/axolotl-cloud
-```
-
-Link: [Docker Hub](https://hub.docker.com/r/axolotlai/axolotl-cloud)
-
-#### Tags format
-
-This uses the same tags as the [`main` image](#sec-main-tags).
-
-#### Environment variables
-
- `JUPYTER_DISABLE`: Disable Jupyter lab.
- `JUPYTER_PASSWORD`: Set a password for the Jupyter lab.
- `PUBLIC_KEY` / `SSH_KEY`: Add a public key for the SSH service.
-
-#### Volume mounts
-
-:::{.callout-tip}
-
-We recommend mounting volumes to `/workspace/data` for data persistence. `/workspace/axolotl` contains the source code and is ephemeral.
-
-:::
-
- `/workspace/data/axolotl-artifacts`: Directory to store Axolotl artifacts.
- `/workspace/data/huggingface-cache`: Directory to store HuggingFace cache.
-
-## Cloud-no-tmux
-
-This is the same as the [`cloud` image](#sec-cloud) but without tmux.
-
-#### Image
-
-```
-axolotlai/axolotl-cloud-term
-```
-
-Link: [Docker Hub](https://hub.docker.com/r/axolotlai/axolotl-cloud-term)
-
-:::{.callout-note}
-
-The naming may be a bit confusing as it has `-term` appended to the end.
-
-:::
-
-#### Tags format
-
-This uses the same tags as the [`cloud` image](#sec-cloud-tags).
--- a/docs/faq.qmd
+++ b/docs/faq.qmd
@@ -3,7 +3,6 @@ title: FAQ
 description: Frequently asked questions
 ---

-### General

 **Q: The trainer stopped and hasn't progressed in several minutes.**

@@ -19,94 +18,12 @@ description: Frequently asked questions

 **Q: AttributeError: 'DummyOptim' object has no attribute 'step'**

-**Q: ModuleNotFoundError: No module named 'mpi4py' using single GPU with deepspeed**
-
-> A: You may be using deepspeed with single gpu. Please remove the `deepspeed:` section in the yaml file or `--deepspeed` CLI flag.
+> A: You may be using deepspeed with single gpu. Please don't set `deepspeed:` in yaml or cli.

 **Q: The codes is stuck on saving preprocessed datasets.**

 > A: This is usually an issue with the GPU. This can be resolved through setting the os environment variable `CUDA_VISIBLE_DEVICES=0`. If you are on runpod, this is usually a pod issue. Starting a new pod should take care of it.

-**Q: Received mismatch error on merge adapters / loading adapters between torch.Size of checkpoint and model.**
-
-> A: This is likely due to vocab size mismatch. By default, Axolotl expands the model's embeddings if the tokenizer has more tokens than the model. Please use the `axolotl merge-lora` command to merge the adapters instead of using your own scripts.
-
-> On the other hand, if the model has more tokens than the tokenizer, Axolotl does not shrink the model's embeddings unless `shrink_embeddings: true` is set in the config.
-
-**Q: How to call Axolotl via custom python scripts?**
-
-> A: Since Axolotl is just Python, please see `src/axolotl/cli/main.py` on how each command is called.
-
-**Q: How to know the value to use for `fsdp_transformer_layer_cls_to_wrap`?**
-
-> A: This is the class name of the transformer layer to wrap with FSDP. For example, for `LlamaForCausalLM`, the value is `LlamaDecoderLayer`. To find this for a specific model, check the model's `PreTrainedModel` definition and look for `_no_split_modules` variable in the `modeling_<model_name>.py` file within `transformers` library.
-
-**Q: ValueError: Asking to pad but the tokenizer does not have a padding token. Please select a token to use as pad_token**
-
-> A: This is because the tokenizer does not have a padding token. Please add a padding token to the tokenizer via:
-
-> ```yaml
-> special_tokens:
->   # str. If you're not sure, set to same as `eos_token`.
->   pad_token: "..."
-> ```
-
-### Chat templates
-
 **Q: `jinja2.exceptions.UndefinedError: 'dict object' has no attribute 'content' / 'role' / ____`**

 > A: This means that the property mapping for the stated attribute does not exist when building `chat_template` prompt. For example, if `no attribute 'content'`, please check you have added the correct mapping for `content` under `message_property_mappings`.
-
-**Q: `Empty template generated for turn ___`**
-
-> A: The `content` is empty for that turn.
-
-**Q: `Could not find content start/end boundary for turn __`**
-
-> A: The specific turn's start/end could not be detected. Please ensure you have set the `eos_token` following your `chat_template`. Otherwise, this could be a `chat_template` which doesn't use proper boundaries for each turn (like system). On the rare occurrence, make sure your content is not `[[dummy_message]]`. Please let us know about this.
-
-**Q: `Content end boundary is before start boundary for turn ___`**
-
-> A: This is an edge case which should not occur. Please create an Issue if this happens.
-
-**Q: `Content end boundary is the same as start boundary for turn ___. This is likely an empty turn.`**
-
-> A: This is likely an empty turn.
-
-**Q: The EOS token is incorrectly being masked or not being masked / `EOS token __ not found in chat template`.**
-
-> A: There can be two reasons:
-
-> 1. This is because of the mismatch between `tokenizer.eos_token` and EOS token in template. Please make sure to set `eos_token: ` under `special_tokens: ` to the same EOS token as in template.
-
-> 2. The EOS token is not in the template. Please check if your template is correct. As an example, `phi_35` template does not use its dedicated EOS token `<|endoftext|>` at the end.
-
-**Q: "`chat_template` choice is `tokenizer_default` but tokenizer's `chat_template` is null. Please add a `chat_template` in tokenizer config"**
-
-> A: This is because the tokenizer does not have a chat template. Please add a chat template in the tokenizer config. See [chat_template](dataset-formats/conversation.qmd#chat-template) for more details.
-
-**Q: The EOT token(s) are incorrectly being masked or not being masked / `EOT token __ not found in chat template`.**
-
-> A: There can be two reasons:
-
-> 1. The EOT token is different from the EOS token and was not specified under `eot_tokens: `. Please set `eot_tokens: ` to the same EOT token(s) as in template.
-
-> 2. There is more than one EOT token per turn in the template. Please raise an issue with examples as we recognize this as an edge case.
-
-**Q: `EOT token encoding failed. Please check if the token is valid and can be encoded.`**
-
-> A: There could be some issue with the tokenizer or unicode encoding. Please raise an issue with examples with the EOT token & tokenizer causing the issue.
-
-**Q: `EOT token __ is encoded as multiple tokens.`**
-
-> A: This is because the EOT token is encoded as multiple tokens which can cause unexpected behavior. Please add it under `tokens: ` or (recommended) override unused added_tokens via `added_tokens_overrides: `.
-
-**Q: `Conflict between train_on_eos and train_on_eot. eos_token is in eot_tokens and train_on_eos != train_on_eot`**
-
-> A: This is because the EOS token is in the `eot_tokens: ` while mismatch between `train_on_eos: ` and `train_on_eot: `. This will cause one to override the other. Please ensure that `train_on_eos: ` and `train_on_eot: ` are the same or remove the EOS token from `eot_tokens: `.
-
-**Q: If `eot_tokens: ` is not provided, what happens?**
-
-> A: If `eot_tokens: ` is not provided, the default behavior is the same as before. EOS tokens used to delimit turns are masked/unmasked depending on whether the turn is trainable.
-
-> Internally, `eot_tokens: tokenizer.eos_token` and `train_on_eot: train_on_eos` (which defaults to `turn`). This transition helps clarify the naming and behavior of EOT/EOS tokens.
--- a/docs/getting-started.qmd
+++ b/docs/getting-started.qmd
@@ -1,5 +1,5 @@
 ---
-title: "Quickstart"
+title: "Getting Started with Axolotl"
 format:
  html:
    toc: true
@@ -17,12 +17,12 @@ Let's start by fine-tuning a small language model using LoRA. This example uses
 Assuming `axolotl` is installed (if not, see our [Installation Guide](installation.qmd))

 1. Download example configs:
-```bash
+```shell
 axolotl fetch examples
 ```

 2. Run the training:
-```bash
+```shell
 axolotl train examples/llama-3/lora-1b.yml
 ```

@@ -36,9 +36,7 @@ The YAML configuration file controls everything about your training. Here's what

 ```yaml
 base_model: NousResearch/Llama-3.2-1B
-
-load_in_8bit: true
-adapter: lora
+# hub_model_id: username/custom_model_name

 datasets:
  - path: teknium/GPT4-LLM-Cleaned
@@ -46,15 +44,11 @@ datasets:
 dataset_prepared_path: last_run_prepared
 val_set_size: 0.1
 output_dir: ./outputs/lora-out
+
+adapter: lora
+lora_model_dir:
 ```

-::: {.callout-tip}
-`load_in_8bit: true` and `adapter: lora` enables LoRA adapter finetuning.
-
- To perform Full finetuning, remove these two lines.
- To perform QLoRA finetuning, replace with `load_in_4bit: true` and `adapter: qlora`.
-:::
-
 See our [Config options](config.qmd) for more details.

 ### Training {#sec-training}
@@ -62,7 +56,7 @@ See our [Config options](config.qmd) for more details.
 When you run `axolotl train`, Axolotl:

 1. Downloads the base model
-2. (If specified) applies QLoRA/LoRA adapter layers
+2. (If specified) applies LoRA adapter layers
 3. Loads and processes the dataset
 4. Runs the training loop
 5. Saves the trained model and / or LoRA weights
@@ -75,8 +69,6 @@ Let's modify the example for your own data:

 ```yaml
 base_model: NousResearch/Nous-Hermes-llama-1b-v1
-
-load_in_8bit: true
 adapter: lora

 # Training settings
@@ -112,9 +104,11 @@ format):
 {"instruction": "Classify this text", "input": "Not good at all", "output": "negative"}
 ```

+Please consult the supported [Dataset Formats](dataset-formats/) for more details.
+
 3. Run the training:

-```bash
+```shell
 axolotl train my_training.yml
 ```

@@ -124,7 +118,7 @@ axolotl train my_training.yml

 After training, test your model:

-```bash
+```shell
 axolotl inference my_training.yml --lora-model-dir="./outputs/lora-out"
 ```

@@ -132,7 +126,7 @@ axolotl inference my_training.yml --lora-model-dir="./outputs/lora-out"

 For large datasets, preprocess first:

-```bash
+```shell
 axolotl preprocess my_training.yml
 ```

@@ -140,7 +134,7 @@ axolotl preprocess my_training.yml

 Launch a Gradio interface:

-```bash
+```shell
 axolotl inference my_training.yml --lora-model-dir="./outputs/lora-out" --gradio
 ```

--- a/docs/inference.qmd
+++ b/docs/inference.qmd
@@ -1,22 +1,19 @@
 ---
-title: "Inference and Merging"
+title: "Inference Guide"
 format:
  html:
    toc: true
    toc-depth: 3
    number-sections: true
+    code-tools: true
 execute:
  enabled: false
 ---

-This guide covers how to use your trained models for inference, including model loading, interactive testing, merging adapters, and common troubleshooting steps.
+This guide covers how to use your trained models for inference, including model loading, interactive testing, and common troubleshooting steps.

 ## Quick Start {#sec-quickstart}

-::: {.callout-tip}
-Use the same config used for training on inference/merging.
-:::
-
 ### Basic Inference {#sec-basic}

 ::: {.panel-tabset}
--- a/docs/input_output.qmd
+++ b/docs/input_output.qmd
@@ -3,4 +3,263 @@ title: Template-free prompt construction
 description: "Template-free prompt construction with the `input_output` format"
 ---

-The documentation moved to [here](dataset-formats/template_free.qmd).
+<!-- TOC -->
+
+- [Background](#background)
+    - [Masking Inputs](#masking-inputs)
+    - [You may not want prompt templates](#you-may-not-want-prompt-templates)
+    - [The `input_output` format](#the-input_output-format)
+- [Usage](#usage)
+    - [1. Prepare Data](#1-prepare-data)
+    - [2. Use `type: input_output`](#2-use-type-input_output)
+    - [3. Check the prompts](#3-check-the-prompts)
+
+<!-- /TOC -->
+
+<a id="markdown-background" name="background"></a>
+
+## Background
+
+<a id="markdown-masking-inputs" name="masking-inputs"></a>
+
+### Masking Inputs
+
+One of the most popular features of
+[axolotl](https://github.com/axolotl-ai-cloud/axolotl) is
+setting the following configuration value:
+
+
+```yaml
+train_on_inputs: false
+```
+
+If you declare a [dataset formats](https://github.com/axolotl-ai-cloud/axolotl?tab=readme-ov-file#dataset)
+such as `alpaca` or `chatml`, axolotl knows what is an input
+(i.e. human) vs. an output (i.e. the assistant) and masks the input
+labels so that your model can focus on predicting the outputs only.
+
+<a id="markdown-you-may-not-want-prompt-templates" name="you-may-not-want-prompt-templates"></a>
+
+### You may not want prompt templates
+
+However, there are many situations where you don't want to use one of
+these formats or templates. This is because they can:
+
+-   Add unnecessary boilerplate to your prompts.
+-   Create artifacts like special delimiters `<|im_start|>` that can
+    quickly become footguns if you don't include them correctly at
+    inference time.
+-   Enforce a *chat* interface when you do not want one. Sometimes you
+    just want to fine-tune a model to a very specific task and do NOT
+    want multi-turn conversations, roles, etc.
+-   Limit you to only certain roles that the template allows.
+
+<a id="markdown-the-inputoutput-format" name="the-inputoutput-format"></a>
+
+### The `input_output` format
+
+You can construct your prompts without a template by using the
+`input_output` format, by setting `type: input_output` in your
+configuration file like this:
+
+**config.yml**
+
+```yaml
+train_on_inputs: false # Mask segments of your data
+datasets:
+  - path: output.jsonl
+    type: input_output  # use template free prompt construction
+```
+
+Unlike `type: completion`, which is also template-free,
+`type: input_output` allows you to mask segments of your text. More
+details on how this works are described below.
+
+<a id="markdown-usage" name="usage"></a>
+
+## Usage
+
+This is how you can use the `input_output` format:
+
+<a id="markdown-1-prepare-data" name="1-prepare-data"></a>
+
+### 1. Prepare Data
+
+To use the `input_output` format, collect your data in the following
+format into a jsonl file (below is the first row from the file
+`output`.jsonl` pretty printed):
+
+```bash
+$ head -n1 output.jsonl | python -m json.tool
+```
+
+:::{.cell-output .cell-output-stdout}
+    {
+        "segments": [
+            {
+                "label": true,
+                "text": "<s>Hello\n"
+            },
+            {
+                "label": true,
+                "text": "hi there!. "
+            },
+            {
+                "label": false,
+                "text": "goodbye "
+            },
+            {
+                "label": true,
+                "text": "farewell</s>"
+            }
+        ]
+    }
+:::
+
+Set `label:false` when you want to mask a segment of text so that the
+model isn't trained on it. Some things to keep in mind:
+
+> [!IMPORTANT]
+> 1.  **EOS, BOS, spaces, newlines etc. are entirely up to you. Axolotl
+    concatenates all the segments as-is.** The tokenizer doesn't add
+    anything additional. Notice how I added spaces, newlines, `<s>`
+    (BOS), and `</s>` (EOS) myself.
+> 2.  Make sure you check the materialized output to validate that the
+    prompt is getting assembled how you like.
+
+<a id="markdown-2-use-type-inputoutput" name="2-use-type-inputoutput"></a>
+
+### 2. Use `type: input_output`
+
+Let's materialize data with our `output.jsonl` file by setting
+`type: input_output` in our axolotl config:
+
+```yaml
+# training_config.yaml
+base_model: mistralai/Mistral-7B-v0.1
+data_seed: 49
+seed: 49
+
+datasets:
+  - path: output.jsonl
+    type: input_output
+val_set_size: 0.1
+
+sequence_len: 896
+sample_packing: false
+
+micro_batch_size: 2
+gradient_accumulation_steps: 3
+eval_batch_size: 2
+num_epochs: 1
+learning_rate: 0.0002
+
+train_on_inputs: false
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"
+```
+
+You can use the following command to materialize your data. The
+`--debug` flag will print the tokens, along with the labels so you can
+verify that the correct items are being ignored:
+
+```bash
+$ python -m axolotl.cli.preprocess training_config.yaml --debug
+
+...
+[2024-03-05 23:36:46,969] [INFO] [axolotl.check_example_labels:35] [PID:607731] [RANK:0] <s>(1, 1) Hello(22557, 22557)
+(13, 13) hi(12014, 12014) there(736, 736) !(28808, 28808) .(28723, 28723) (28705, 28705) good(-100, 1179) bye(-100, 17664) (-100, 28705) fare(19111, 19111) well(5458, 5458) </s>(2, 2)
+
+```
+
+The format is `decoded_token`(`label`, `token_id`), for example,
+`<s>(1, 1)` means that the token is `<s>`, the label is `1` and the
+token_id is `1`. When the label is `-100` then that token is ignored for
+training.
+
+<a id="markdown-3-check-the-prompts" name="3-check-the-prompts"></a>
+
+### 3. Check the prompts
+
+Here is another way to check the materialized output:
+
+```python
+from transformers import AutoTokenizer
+from datasets import load_from_disk
+import yaml
+
+directory = !ls last_run_prepared/
+with open('training_config.yaml', 'r') as f:
+    cfg = yaml.safe_load(f)
+model_id = cfg['base_model']
+tok = AutoTokenizer.from_pretrained(model_id)
+ds = load_from_disk(f'last_run_prepared/{directory[0]}/')
+```
+
+```python
+>>> row = ds[0]
+>>> print(tok.decode(row['input_ids']))
+<s> Hello
+    hi there!.  goodbye  farewell</s>
+```
+
+We can check that the right tokens are ignored by comparing the labels
+to each token:
+
+```python
+import pandas as pd
+pd.DataFrame([{'token': tok.decode(i), 'label': l, 'id':i} for i,l in
+              zip(row['input_ids'], row['labels'])])
+```
+
+| token | label | id    |
+|-------|-------|-------|
+| 0     | \<s\> | 1     |
+| 1     | Hello | 22557 |
+| 2     | \\n   | 13    |
+| 3     | hi    | 12014 |
+| 4     | there | 736   |
+| 5     | !     | 28808 |
+| 6     | .     | 28723 |
+| 7     |       | 28705 |
+| 8     | good  | -100  |
+| 9     | bye   | -100  |
+| 10    |       | -100  |
+| 11    | fare  | 19111 |
+| 12    | well  | 5458  |
+| 13    | \</s\>| 2     |
+
+
+
+If we look at the input data, the above table seems correct! (The jsonl
+version is repeated below for reference):
+
+
+```bash
+$ head -n1 output.jsonl | python -m json.tool
+```
+
+:::{.cell-output .cell-output-stdout}
+    {
+        "segments": [
+            {
+                "label": true,
+                "text": "<s>Hello\n"
+            },
+            {
+                "label": true,
+                "text": "hi there!. "
+            },
+            {
+                "label": false,
+                "text": "goodbye "
+            },
+            {
+                "label": true,
+                "text": "farewell</s>"
+            }
+        ]
+    }
+:::
--- a/docs/installation.qmd
+++ b/docs/installation.qmd
@@ -1,10 +1,11 @@
 ---
-title: "Installation"
+title: "Installation Guide"
 format:
  html:
    toc: true
    toc-depth: 3
    number-sections: true
+    code-tools: true
 execute:
  enabled: false
 ---
@@ -19,16 +20,9 @@ This guide covers all the ways you can install and set up Axolotl for your envir

 ## Installation Methods {#sec-installation-methods}

-::: {.callout-important}
-Please make sure to have Pytorch installed before installing Axolotl in your local environment.
-
-Follow the instructions at: [https://pytorch.org/get-started/locally/](https://pytorch.org/get-started/locally/)
-:::
-
 ### PyPI Installation (Recommended) {#sec-pypi}

 ```{.bash}
-pip3 install -U packaging setuptools wheel ninja
 pip3 install --no-build-isolation axolotl[flash-attn,deepspeed]
 ```

@@ -44,7 +38,7 @@ For the latest features between releases:
 ```{.bash}
 git clone https://github.com/axolotl-ai-cloud/axolotl.git
 cd axolotl
-pip3 install -U packaging setuptools wheel ninja
+pip3 install packaging ninja
 pip3 install --no-build-isolation -e '.[flash-attn,deepspeed]'
 ```

@@ -72,8 +66,6 @@ docker run --privileged --gpus '"all"' --shm-size 10g --rm -it \
 ```
 :::

-Please refer to the [Docker documentation](docker.qmd) for more information on the different Docker images that are available.
-
 ## Cloud Environments {#sec-cloud}

 ### Cloud GPU Providers {#sec-cloud-gpu}
@@ -85,7 +77,6 @@ For providers supporting Docker:
  - [Latitude.sh](https://latitude.sh/blueprint/989e0e79-3bf6-41ea-a46b-1f246e309d5c)
  - [JarvisLabs.ai](https://jarvislabs.ai/templates/axolotl)
  - [RunPod](https://runpod.io/gsc?template=v2ickqhz9s&ref=6i7fkpdz)
-  - [Novita](https://novita.ai/gpus-console?templateId=311)

 ### Google Colab {#sec-colab}

@@ -115,7 +106,7 @@ We recommend using WSL2 (Windows Subsystem for Linux) or Docker.
 2. Install PyTorch: https://pytorch.org/get-started/locally/
 3. Install Axolotl:
   ```{.bash}
-   pip3 install -U packaging setuptools wheel ninja
+   pip3 install packaging
   pip3 install --no-build-isolation -e '.[flash-attn,deepspeed]'
   ```
 4. (Optional) Login to Hugging Face:
--- a/docs/lora_optims.qmd
+++ b/docs/lora_optims.qmd
@@ -1,6 +1,7 @@
 ---
 title: "LoRA Optimizations"
-description: "Custom autograd functions and Triton kernels in Axolotl for optimized LoRA fine-tuning"
+description: "Custom autograd functions and Triton kernels in Axolotl for optimized
+LoRA fine-tuning"
 ---

 Inspired by [Unsloth](https://github.com/unslothai/unsloth), we've implemented two
@@ -17,7 +18,6 @@ We currently support several common model architectures, including (but not limi
 - `qwen2`
 - `gemma`
 - `gemma2`
- `gemma3`

 <details>

@@ -67,10 +67,6 @@ logic to be compatible with more of them.

 </details>

-::: {.callout-tip}
-Check out our [LoRA optimizations blog](https://axolotlai.substack.com/p/accelerating-lora-fine-tuning-with).
-:::
-
 ## Usage

 These optimizations can be enabled in your Axolotl config YAML file. The
--- a/docs/mac.qmd
+++ b/docs/mac.qmd
@@ -19,5 +19,4 @@ Current support:
 - [ ] DeepSpeed

 Untested:
-
 - FSDP
--- a/docs/multi-gpu.qmd
+++ b/docs/multi-gpu.qmd
@@ -1,5 +1,5 @@
 ---
-title: "Multi-GPU"
+title: "Multi-GPU Training Guide"
 format:
  html:
    toc: true
@@ -18,7 +18,6 @@ Axolotl supports several methods for multi-GPU training:

 - DeepSpeed (recommended)
 - FSDP (Fully Sharded Data Parallel)
- Sequence parallelism
 - FSDP + QLoRA

 ## DeepSpeed {#sec-deepspeed}
@@ -36,14 +35,7 @@ deepspeed: deepspeed_configs/zero1.json
 ### Usage {#sec-deepspeed-usage}

 ```{.bash}
-# Fetch deepspeed configs (if not already present)
-axolotl fetch deepspeed_configs
-
-# Passing arg via config
-axolotl train config.yml
-
-# Passing arg via cli
-axolotl train config.yml --deepspeed deepspeed_configs/zero1.json
+accelerate launch -m axolotl.cli.train examples/llama-2/config.yml --deepspeed deepspeed_configs/zero1.json
 ```

 ### ZeRO Stages {#sec-zero-stages}
@@ -51,20 +43,10 @@ axolotl train config.yml --deepspeed deepspeed_configs/zero1.json
 We provide default configurations for:

 - ZeRO Stage 1 (`zero1.json`)
- ZeRO Stage 1 with torch compile (`zero1_torch_compile.json`)
 - ZeRO Stage 2 (`zero2.json`)
 - ZeRO Stage 3 (`zero3.json`)
- ZeRO Stage 3 with bf16 (`zero3_bf16.json`)
- ZeRO Stage 3 with bf16 and CPU offload params(`zero3_bf16_cpuoffload_params.json`)
- ZeRO Stage 3 with bf16 and CPU offload params and optimizer (`zero3_bf16_cpuoffload_all.json`)

-::: {.callout-tip}
-
-Choose the configuration that offloads the least amount to memory while still being able to fit on VRAM for best performance.
-
-Start from Stage 1 -> Stage 2 -> Stage 3.
-
-:::
+Choose based on your memory requirements and performance needs.

 ## FSDP {#sec-fsdp}

@@ -80,28 +62,6 @@ fsdp_config:
  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
 ```

-## Sequence parallelism {#sec-sequence-parallelism}
-
-We support sequence parallelism (SP) via the
-[ring-flash-attention](https://github.com/zhuzilin/ring-flash-attention) project. This
-allows one to split up sequences across GPUs, which is useful in the event that a
-single sequence causes OOM errors during model training.
-
-First, install `ring-flash-attn`, recommended via `pip install axolotl[ring-flash-attn]`,
-or from source with `pip install .[ring-flash-attn]`.
-
-Your Axolotl YAML config should contain the following lines:
-
-```{.yaml}
-sequence_parallel_degree: 4  # Split each sequence into 4 parts, one per GPU
-flash_attention: true  # Required with sequence parallelism
-
-# Optional; strides across the key dimension. Larger values use more memory but will make training faster.
-heads_k_stride: 1
-```
-
-See our [dedicated guide](sequence_parallelism.qmd) for more details.
-
 ### FSDP + QLoRA {#sec-fsdp-qlora}

 For combining FSDP with QLoRA, see our [dedicated guide](fsdp_qlora.qmd).
@@ -110,7 +70,25 @@ For combining FSDP with QLoRA, see our [dedicated guide](fsdp_qlora.qmd).

 ### Liger Kernel Integration {#sec-liger}

-Please see [docs](custom_integrations.qmd#liger) for more info.
+::: {.callout-note}
+Liger Kernel provides efficient Triton kernels for LLM training, offering:
+
+- 20% increase in multi-GPU training throughput
+- 60% reduction in memory usage
+- Compatibility with both FSDP and DeepSpeed
+:::
+
+Configuration:
+
+```{.yaml}
+plugins:
+  - axolotl.integrations.liger.LigerPlugin
+liger_rope: true
+liger_rms_norm: true
+liger_glu_activation: true
+liger_layer_norm: true
+liger_fused_linear_cross_entropy: true
+```

 ## Troubleshooting {#sec-troubleshooting}

--- a/docs/multi-node.qmd
+++ b/docs/multi-node.qmd
@@ -13,7 +13,7 @@ You will also need to have the same configuration file for your model on each ma
 Make sure the main machine is reachable by other machines.
 :::

-## Accelerate
+# Accelerate

 You will need to create a configuration for accelerate, either by using `accelerate config` and follow the instructions or you can use one of the preset below:

@@ -51,17 +51,17 @@ fsdp_config:

 All you have to do now is launch using accelerate as you would usually do on each machine and voila, the processes will start once you have launched accelerate on every machine.

-## Raytrain
+# Raytrain

 Please see ray train doc [here](ray-integration.qmd).

-## Torchrun
+# Torchrun

 If you are using Infiniband, we recommend torchrun to utilize the full bandwidth.

 Set the following env (change buffersize/socketname depending on your system):

-```bash
+```yaml
 export NCCL_IB_DISABLE=0
 export NCCL_SOCKET_IFNAME="eth0,en,eth,em,bond"
 export NCCL_BUFFSIZE=2097152
--- a/docs/multimodal.qmd
+++ b/docs/multimodal.qmd
@@ -1,180 +1,28 @@
---
-title: MultiModal / Vision Language Models (BETA)
-format:
-  html:
-    toc: true
-    toc-depth: 3
---
+# MultiModal / Vision Language Models (BETA)

-## Supported Models
+### Supported Models

- [Mllama](#sec-mllama)
- [Llama4](#sec-llama4)
- [Pixtral](#sec-pixtral)
- [Llava-1.5](#sec-llava-15)
- [Mistral-Small-3.1](#sec-mistral-small-31)
- [Gemma-3](#sec-gemma-3)
- [Qwen2-VL](#sec-qwen2-vl)
- [Qwen2.5-VL](#sec-qwen25-vl)
+- Mllama, i.e. llama with vision models

-## Usage
+### Usage

-Multimodal support is limited and doesn't have full feature parity.
-
-Here are the hyperparams you'll need to use to finetune a multimodal model.
+Currently multimodal support is limited and doesn't have full feature parity. To finetune a multimodal Llama w/ LoRA,
+you'll need to use the following in YAML in combination with the rest of the required hyperparams.

 ```yaml
+base_model: alpindale/Llama-3.2-11B-Vision-Instruct
 processor_type: AutoProcessor
-
 skip_prepare_dataset: true
-remove_unused_columns: false  # leave columns in place as they are needed to handle image embeddings during training
-sample_packing: false  # not yet supported with multimodal

-chat_template:  # see in next section
-
-# example dataset
+chat_template: llama3_2_vision
 datasets:
  - path: HuggingFaceH4/llava-instruct-mix-vsft
    type: chat_template
    split: train[:1%]
    field_messages: messages
+remove_unused_columns: false
+sample_packing: false

-# (optional) if doing lora, only finetune the Language model,
-# leave the vision model and vision tower frozen
-# load_in_8bit: true
-adapter: lora
+# only finetune the Language model, leave the vision model and vision tower frozen
 lora_target_modules: 'language_model.model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'
-
-# (optional) if you want to resize images to a set size
-image_size: 512
-image_resize_algorithm: bilinear
-```
-
-Please see [examples](https://github.com/axolotl-ai/axolotl/tree/main/examples) folder for full configs.
-
-::: {.callout-warning}
-Some of our chat_templates have been extended to support broader dataset types. This should not break any existing configs.
-:::
-
-### Mllama {#sec-mllama}
-
-```yaml
-base_model: meta-llama/Llama-3.2-11B-Vision-Instruct
-
-chat_template: llama3_2_vision
-```
-
-### Llama4 {#sec-llama4}
-
-```yaml
-base_model: meta-llama/Llama-4-Scout-17B-16E-Instruct
-
-chat_template: llama4
-```
-
-### Pixtral {#sec-pixtral}
-
-```yaml
-base_model: mistralai/Pixtral-12B-2409
-
-chat_template: pixtral
-```
-
-### Llava-1.5 {#sec-llava-15}
-
-```yaml
-base_model: llava-hf/llava-1.5-7b-hf
-
-chat_template: llava
-```
-
-### Mistral-Small-3.1 {#sec-mistral-small-31}
-
-```yaml
-base_model: mistralai/Mistral-Small-3.1-24B-Instruct-2503
-
-chat_template: mistral_v7_tekken
-```
-
-### Gemma-3 {#sec-gemma-3}
-
-::: {.callout-tip}
-The Gemma3-1B model is a text-only model, so please train as regular text model.
-:::
-
-For multi-modal 4B/12B/27B models, use the following config:
-
-```yaml
-base_model: google/gemma-3-4b-it
-
-chat_template: gemma3
-```
-
-### Qwen2-VL {#sec-qwen2-vl}
-
-```yaml
-base_model: Qwen/Qwen2-VL-7B-Instruct
-
-chat_template: qwen2_vl
-```
-
-### Qwen2.5-VL {#sec-qwen25-vl}
-
-```yaml
-base_model: Qwen/Qwen2.5-VL-7B-Instruct
-
-chat_template: qwen2_vl  # same as qwen2-vl
-```
-
-## Dataset Format
-
-For multi-modal datasets, we adopt an extended `chat_template` format similar to OpenAI's Message format.
-
- A message is a list of `role` and `content`.
- `role` can be `system`, `user`, `assistant`, etc.
- `content` is a list of `type` and (`text` or `image` or `path` or `url` or `base64`).
-
-::: {.callout-note}
-For backwards compatibility:
-
- If the dataset has a `images` or `image` column of `list[Image]`, it will be appended to the first `content` list as `{"type": "image", "image": ...}`. However, if the content already has a `{"type": "image"}` but no `image` key, it will be set the `image` key.
- If `content` is a string, it will be converted to a list with `type` as `text`.
-:::
-
-::: {.callout-tip}
-For image loading, you can use the following keys within `content` alongside `"type": "image"`:
-
- `"path": "/path/to/image.jpg"`
- `"url": "https://example.com/image.jpg"`
- `"base64": "..."`
- `"image": PIL.Image`
-:::
-
-Here is an example of a multi-modal dataset:
-```json
-[
-  {
-    "messages": [
-        {
-            "role": "system",
-            "content": [
-              {"type": "text", "text": "You are a helpful assistant."}
-              ]
-        },
-        {
-            "role": "user",
-            "content": [
-                {"type": "image", "image": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
-                {"type": "text", "text": "Describe this image in detail."}
-            ]
-        },
-        {
-            "role": "assistant",
-            "content": [
-              {"type": "text", "text": "The image is a bee."}
-            ]
-        }
-    ]
-  }
-]
 ```
--- a/docs/nccl.qmd
+++ b/docs/nccl.qmd
@@ -13,13 +13,13 @@ Often, this timeout will happen after 30 minutes (the default setting) and is ac

 Forcing cross-GPU communication via [NVLink](https://en.wikipedia.org/wiki/NVLink) may help without increasing timeouts. To verify that your configuration is leveraging NVLink run the following command:

-```bash
+```shell
 nvidia-smi nvlink --status
 ```

 To force NCCL to use NVLink, simply set this in the environment:

-```bash
+```shell
 export NCCL_P2P_LEVEL=NVL
 ```

@@ -33,13 +33,13 @@ If NVLink is not available in your environment there are other options for ``NCC

 To validate that acceptable data transfer speeds exist for your training job, running [NCCL Tests](https://github.com/NVIDIA/nccl-tests/blob/master/README.md) can help pinpoint bottlenecks, for example:

-```bash
+```shell
 ./build/all_reduce_perf -b 8 -e 128M -f 2 -g 3
 ```

 It can be useful when debugging NCCL communication timeouts to activate additional logging in both PyTorch and NCCL:

-```bash
+```shell
 export NCCL_DEBUG=INFO
 export NCCL_DEBUG_SUBSYS=ALL
 export TORCH_DISTRIBUTED_DEBUG=INFO
--- a/docs/ray-integration.qmd
+++ b/docs/ray-integration.qmd
@@ -1,5 +1,5 @@
 ---
-title: Ray Train
+title: Ray Train integration
 description: How to use Axolotl with Ray Train
 ---

@@ -9,7 +9,7 @@ With the `--use-ray` CLI flag, Axolotl will use Ray Train's [`TorchTrainer`](htt

 ## Ray cluster setup

-A prerequisite using the Ray Train integration is to setup a Ray cluster on your desired node(s). For a detailed guide on how you can get started with ray clusters, check the official Ray docs [here](https://docs.ray.io/en/latest/cluster/getting-started.html).
+A prerequisite using the Ray Train integration is to setup a Ray cluster on your desired node(s). For a detailed guide on how you can get started with ray clusters, check the official Ray docs here: https://docs.ray.io/en/latest/cluster/getting-started.html

 Every Ray cluster has one _head_ node and a set of worker nodes. The head node is just like any other worker node, but it also runs certain special processes related to scheduling and orchestration. Ray-enabled scripts are run on the head node and depending on the resources (number of CPUs, GPUs, etc) they request, will be scheduled to run certain tasks on the worker nodes. For more on key concepts behind a Ray cluster, you can refer this [doc](https://docs.ray.io/en/latest/cluster/key-concepts.html#cluster-key-concepts).

@@ -58,11 +58,13 @@ You can find an example configuration at `configs/llama-3/lora-1b-ray.yaml`.
 The key parameters to note here are:

 ```yaml
+...
 use_ray: true
 ray_num_workers: 4
 # optional
 resources_per_worker:
    GPU: 1
+...
 ```

 - `use_ray`: This is the flag that enables the Ray Train integration. You can either use the corresponding `--use-ray` flag in the CLI or set `use_ray` in the config file.
--- a/docs/reward_modelling.qmd
+++ b/docs/reward_modelling.qmd
@@ -28,23 +28,8 @@ val_set_size: 0.1
 eval_steps: 100
 ```

-Bradley-Terry chat templates expect single-turn conversations in the following format:
-
-```json
-{
-    "system": "...", // optional
-    "input": "...",
-    "chosen": "...",
-    "rejected": "..."
-}
-```
-
 ### Process Reward Models (PRM)

-::: {.callout-tip}
-Check out our [PRM blog](https://axolotlai.substack.com/p/process-reward-models).
-:::
-
 Process reward models are trained using data which contains preference annotations for each step in a series of interactions. Typically, PRMs are trained to provide reward signals over each step of a reasoning trace and are used for downstream reinforcement learning.
 ```yaml
 base_model: Qwen/Qwen2.5-3B
@@ -60,5 +45,3 @@ datasets:
 val_set_size: 0.1
 eval_steps: 100
 ```
-
-Please see [stepwise_supervised](dataset-formats/stepwise_supervised.qmd) for more details on the dataset format.
--- a/docs/rlhf.qmd
+++ b/docs/rlhf.qmd
@@ -3,23 +3,22 @@ title: "RLHF (Beta)"
 description: "Reinforcement Learning from Human Feedback is a method whereby a language model is optimized from data using human feedback."
 back-to-top-navigation: true
 toc: true
-toc-expand: 2
-toc-depth: 4
+toc-depth: 3
 ---

-## Overview
+# Overview

 Reinforcement Learning from Human Feedback is a method whereby a language model is optimized from data using human
 feedback. Various methods include, but not limited to:

+- Proximal Policy Optimization (PPO) (not yet supported in axolotl)
 - [Direct Preference Optimization (DPO)](#dpo)
 - [Identity Preference Optimization (IPO)](#ipo)
 - [Kahneman-Tversky Optimization (KTO)](#kto)
 - [Odds Ratio Preference Optimization (ORPO)](#orpo)
- Proximal Policy Optimization (PPO) (not yet supported in axolotl)


-## RLHF using Axolotl
+# RLHF using Axolotl

 ::: {.callout-important}
 This is a BETA feature and many features are not fully implemented. You are encouraged to open new PRs to improve the integration and functionality.
@@ -31,7 +30,7 @@ We rely on the [TRL](https://github.com/huggingface/trl) library for implementat
 You can find what each method supports by going into `src/axolotl/prompt_strategies/{method}` where `{method}` is one of our supported methods. The `type: ` can be retrieved from `{method}.{function_name}`.
 :::

-### DPO
+## DPO

 Example config:

@@ -48,7 +47,7 @@ datasets:

 DPO supports the following types with the following dataset format:

-#### chatml.argilla
+### chatml.argilla

 ```json
 {
@@ -59,7 +58,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### chatml.argilla_chat
+### chatml.argilla_chat

 ```json
 {
@@ -74,7 +73,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### chatml.icr
+### chatml.icr

 ```json
 {
@@ -85,7 +84,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### chatml.intel
+### chatml.intel

 ```json
 {
@@ -96,7 +95,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### chatml.prompt_pairs
+### chatml.prompt_pairs

 ```json
 {
@@ -107,7 +106,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### chatml.ultra
+### chatml.ultra

 ```json
 {
@@ -124,7 +123,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### llama3.argilla
+### llama3.argilla

 ```json
 {
@@ -135,7 +134,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### llama3.argilla_chat
+### llama3.argilla_chat

 ```json
 {
@@ -150,7 +149,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### llama3.icr
+### llama3.icr

 ```json
 {
@@ -161,7 +160,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### llama3.intel
+### llama3.intel

 ```json
 {
@@ -172,7 +171,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### llama3.prompt_pairs
+### llama3.prompt_pairs

 ```json
 {
@@ -183,7 +182,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### llama3.ultra
+### llama3.ultra

 ```json
 {
@@ -200,7 +199,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### zephyr.nectar
+### zephyr.nectar

 ```json
 {
@@ -219,7 +218,7 @@ DPO supports the following types with the following dataset format:
 }
 ```

-#### chat_template.default
+### chat_template.default

 ```yaml
 rl: dpo
@@ -265,7 +264,7 @@ Sample input format:
 }
 ```

-#### user_defined.default
+### user_defined.default

 For custom behaviors,

@@ -296,15 +295,15 @@ The input format is a simple JSON input with customizable fields based on the ab
 }
 ```

-### IPO
+## IPO

-As IPO is just DPO with a different loss function, all supported dataset formats for [DPO](#dpo) are also supported for IPO.
+As IPO is just DPO with a different loss function, all supported options for DPO works here.

 ```yaml
 rl: ipo
 ```

-### ORPO
+## ORPO

 Paper: https://arxiv.org/abs/2403.07691

@@ -321,7 +320,7 @@ datasets:

 ORPO supports the following types with the following dataset format:

-#### chat_template.argilla
+### chat_template.argilla

 ```json
 {
@@ -340,13 +339,12 @@ ORPO supports the following types with the following dataset format:
 }
 ```

-### KTO
+## KTO

 ```yaml
 rl: kto
-rl_beta: 0.1  # default
-kto_desirable_weight: 1.0  # default
-kto_undesirable_weight: 1.0  # default
+rl_beta: 0.5
+kto_desirable_weight: 0.2

 remove_unused_columns: false

@@ -362,7 +360,7 @@ gradient_checkpointing_kwargs:

 KTO supports the following types with the following dataset format:

-#### chatml.argilla
+### chatml.argilla

 ```json
 {
@@ -372,7 +370,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### chatml.argilla_chat
+### chatml.argilla_chat

 ```json
 {
@@ -385,7 +383,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### chatml.intel
+### chatml.intel

 ```json
 {
@@ -395,7 +393,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### chatml.prompt_pairs
+### chatml.prompt_pairs

 ```json
 {
@@ -405,7 +403,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### chatml.ultra
+### chatml.ultra

 ```json
 {
@@ -415,7 +413,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### llama3.argilla
+### llama3.argilla

 ```json
 {
@@ -425,7 +423,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### llama3.argilla_chat
+### llama3.argilla_chat

 ```json
 {
@@ -436,7 +434,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### llama3.intel
+### llama3.intel

 ```json
 {
@@ -446,7 +444,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### llama3.prompt_pairs
+### llama3.prompt_pairs

 ```json
 {
@@ -456,7 +454,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### llama3.ultra
+### llama3.ultra

 ```json
 {
@@ -466,7 +464,7 @@ KTO supports the following types with the following dataset format:
 }
 ```

-#### user_defined.default
+### user_defined.default

 For custom behaviors,

@@ -496,108 +494,7 @@ The input format is a simple JSON input with customizable fields based on the ab
 }
 ```

-### GRPO
-
-::: {.callout-tip}
-Check out our [GRPO cookbook](https://github.com/axolotl-ai-cloud/axolotl-cookbook/tree/main/grpo#training-an-r1-style-large-language-model-using-grpo).
-:::
-
-In the latest GRPO implementation, `vLLM` is used to significantly speedup trajectory generation during training. In this example, we're using 4 GPUs - 2 for training, and 2 for vLLM:
-
-::: {.callout-important}
-Make sure you've installed the correct version of vLLM by including it as an extra when installing axolotl, e.g. `pip install axolotl[vllm]`.
-:::
-
-```yaml
-base_model: Qwen/Qwen2.5-1.5B-Instruct
-
-vllm:
-    host: 0.0.0.0
-    port: 8000
-    tensor_parallel_size: 2
-    gpu_memory_utilization: 0.85
-    dtype: auto
-    # max_model_len: # you may find it useful to set the vLLM model context length if you know this beforehand
-
-rl: grpo
-trl:
-    use_vllm: true
-    vllm_server_host: 0.0.0.0
-    vllm_server_port: 8000
-    vllm_server_timeout: 300
-```
-
-```bash
-CUDA_VISIBLE_DEVICES=2,3 axolotl vllm-serve grpo.yaml
-```
-
-Your `vLLM` instance will now attempt to spin up, and it's time to kick off training utilizing our remaining two GPUs. In another terminal, execute:
-
-```bash
-CUDA_VISIBLE_DEVICES=0,1 axolotl train grpo.yaml --num-processes 2
-```
-
-::: {.callout-note}
-Due to TRL's implementation with vLLM, the vLLM instance must use the last N GPUs instead of the first N GPUs. This is why in the example above, we use `CUDA_VISIBLE_DEVICES=2,3` for the vLLM instance.
-:::
-
-#### Reward functions
-
-GRPO uses custom reward functions and transformations. Please have them ready locally.
-
-For example, to load OpenAI's GSM8K and use a random reward for completions:
-
-```python
-# rewards.py
-import random
-
-def rand_reward_func(completions, **kwargs) -> list[float]:
-    return [random.uniform(0, 1) for _ in completions]
-
-def oai_gsm8k_transform(cfg, *args, **kwargs):
-    def transform_fn(example, tokenizer=None):
-        label = example["answer"].split("####")[-1].strip().replace(",", "")
-        return {
-            "prompt": [{"role": "user", "content": example["question"]},],
-            "answer": label,
-        }
-    return transform_fn, {"remove_columns": ["question"]}
-```
-
-```yaml
-rl: grpo
-
-trl:
-    beta: 0.001
-    max_completion_length: 256
-    use_vllm: True
-    num_generations: 4
-    reward_funcs: ["rewards.rand_reward_func"]    # format: '{file_name}.{fn_name}'
-    reward_weights: [1.0]
-datasets:
-  - path: openai/gsm8k
-    name: main
-    type: rewards.oai_gsm8k_transform  # format: '{file_name}.{fn_name}'
-```
-
-To see other examples of custom reward functions, please see [TRL GRPO Docs](https://github.com/huggingface/trl/blob/main/docs/source/grpo_trainer.md#using-a-custom-reward-function).
-
-To see description of the configs, please see [TRLConfig](https://github.com/axolotl-ai-cloud/axolotl/blob/main/src/axolotl/utils/config/models/input/v0_4_1/trl.py).
-
-### SimPO
-
-SimPO uses [CPOTrainer](https://huggingface.co/docs/trl/main/en/cpo_trainer) but with alternative loss function.
-
-```yaml
-rl: simpo
-rl_beta: 0.1  # default in CPOTrainer
-cpo_alpha: 1.0  # default in CPOTrainer
-simpo_gamma: 0.5  # default in CPOTrainer
-```
-
-This method uses the same dataset format as [DPO](#dpo).
-
-### Using local dataset files
+## Using local dataset files

 ```yaml
 datasets:
@@ -608,7 +505,7 @@ datasets:
    type: chatml.intel
 ```

-### TRL auto-unwrapping for PEFT
+## TRL auto-unwrapping for PEFT

 TRL supports auto-unwrapping PEFT models for RL training paradigms which rely on a reference model. This significantly reduces memory pressure as an additional refreference model does not need to be loaded, and reference model log-probabilities can be obtained by disabling PEFT adapters. This is enabled by default. To turn it off, pass the following config:

--- a/docs/sequence_parallelism.qmd
+++ b/docs/sequence_parallelism.qmd
@@ -1,100 +0,0 @@
---
-title: Sequence Parallelism
-description: Train with long sequences split across multiple GPUs.
---
-
-# Sequence Parallelism
-
-Sequence parallelism is a technique that splits sequences across multiple GPUs,
-allowing you to train with very long sequences that wouldn't fit on a single GPU. Each
-GPU processes a different portion of the sequence, and the results are aggregated
-through a ring communication pattern.
-
-## When to Use Sequence Parallelism
-
-Use sequence parallelism when:
-
- You need to train with sequence lengths that don't fit into a single GPU's memory
- You have multiple GPUs available
- You're experiencing OOM (Out Of Memory) errors with long sequences
-
-## Configuration
-
-To enable sequence parallelism, add the following to your configuration file:
-
-```yaml
-# Set to a divisor (> 1) of the number of GPUs available
-sequence_parallel_degree: 4  # Split sequences across 4 GPUs
-# Optional; strides across the key dimension. Larger values use more memory but should make training faster.
-heads_k_stride: 1
-# Optional; one of "varlen_llama3", "batch_ring", "batch_zigzag", "batch_stripe". Defaults to
-# "varlen_llama3" when `sample_packing: true`, and "batch_ring" otherwise.
-ring_attn_func:
-```
-
-The `sequence_parallel_degree` should be a divisor of the total number of GPUs. For example:
-
- With 8 GPUs, valid values would be 2, 4, or 8
- With 4 GPUs, valid values would be 2 or 4
-
-## Implementation Details
-
-When sequence parallelism is enabled:
-
-1. Each sequence is divided into equal chunks across the GPUs in a sequence parallel group
-2. The data collator handles the chunking of input_ids, attention_mask, labels, and position_ids
-3. Position IDs are adjusted to maintain proper relative positions, especially for packed sequences
-4. The trainer uses special ring communication patterns for attention operations
-
-## Requirements
-
-To use sequence parallelism, you need:
-
- Multiple GPUs (at least 2)
- The `ring-flash-attn` package. Install with:
-  - `pip install axolotl[ring-flash-attn]` (preferred)
-  - `pip install ring-flash-attn>=0.1.4`
-
-## Limitations
-
- Flash attention must be enabled for this to work (`flash_attention: true` in config YAML)
- May have a small performance overhead due to communication between GPUs
-
-## Example
-
-```yaml
-base_model: meta-llama/Llama-3-8B-Instruct
-sequence_len: 8192
-
-...
-
-sequence_parallel_degree: 4  # Split each sequence into 4 parts, one per GPU
-flash_attention: true  # Required with sequence parallelism
-# Optional; strides across the key dimension. Larger values use more memory but should make training faster.
-heads_k_stride: 1
-
-...
-```
-
-This will train the Llama 3 8B model with 8K context length, with each sequence split
-into 2 subsequences of length 4096 across 2 GPUs.
-
-## Sample Packing with Sequence Parallelism
-
-Sequence parallelism is compatible with Axolotl's sample packing functionality. When using both features together:
-
-1. Samples are first packed together
-2. The packed sequences are then divided across GPUs in the sequence parallel group
-3. Position IDs are automatically adjusted to maintain proper relative positions
-
-## Effect on Batch Size
-
-When using sequence parallelism, your effective global batch size is **divided** by the `sequence_parallel_degree`. This happens because:
-
- Each group of `sequence_parallel_degree` GPUs works on the same batch (just different parts of each sequence)
- The number of batches processed per step decreases
-
-For example:
- With 8 GPUs and no sequence parallelism: 8 different batches processed per step
- With 8 GPUs and `sequence_parallel_degree=4`: Only 2 different batches processed per step (each split across 4 GPUs)
- If your per-GPU `micro_batch_size` is 2, the global batch size decreases from 16 to 4
--- a/docs/torchao.qmd
+++ b/docs/torchao.qmd
@@ -3,12 +3,6 @@ title: "PyTorch ao"
 description: "Custom data types and layouts for training and inference"
 ---

-To use experimental optimizers (`AdamWFp8`, `AdamW4bit`, `AdamW8bit`) from Pytorch Ao, please install the package as shown below.
-
-::: {.callout-tip}
-Some experimental optimizers are already present in regular Pytorch, so please re-check if you actually need this package!
-:::
-
 ### Installation

 Stable Release from the PyTorch index
--- a/docs/unsloth.qmd
+++ b/docs/unsloth.qmd
@@ -8,12 +8,6 @@ description: "Hyper-optimized QLoRA finetuning for single GPUs"
 Unsloth provides hand-written optimized kernels for LLM finetuning that slightly improve speed and VRAM over
 standard industry baselines.

-::: {.callout-important}
-Due to breaking changes in transformers `v4.48.0`, users will need to downgrade to `<=v4.47.1` to use this patch.
-
-This will later be deprecated in favor of [LoRA Optimizations](lora_optims.qmd).
-:::
-

 ### Installation

@@ -23,7 +17,7 @@ The following will install the correct unsloth and extras from source.
 python scripts/unsloth_install.py | sh
 ```

-### Usage
+### Using unsloth w Axolotl

 Axolotl exposes a few configuration options to try out unsloth and get most of the performance gains.

--- a/examples/cerebras/btlm-ft.yml
+++ b/examples/cerebras/btlm-ft.yml
@@ -8,6 +8,10 @@ tokenizer_type: GPT2Tokenizer
 trust_remote_code: true
 tokenizer_use_fast: true
 tokenizer_legacy: true
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
 push_dataset_to_hub:
 hf_use_auth_token: true
 datasets:
@@ -30,6 +34,7 @@ lora_alpha:
 lora_dropout:
 lora_target_modules:
 lora_target_linear:
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -53,12 +58,16 @@ learning_rate: 0.000085
 train_on_inputs: true
 group_by_length: false
 bf16: auto
+fp16:
 tf32: true

 gradient_checkpointing: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1

+xformers_attention:
 flash_attention: true
 sdp_attention:
 flash_optimum:
@@ -71,6 +80,8 @@ evals_per_epoch: 4
 saves_per_epoch: 1
 save_total_limit:

+debug:
+deepspeed:
 weight_decay: 0.1
 special_tokens:
  pad_token: "<|endoftext|>"
--- a/examples/cerebras/qlora.yml
+++ b/examples/cerebras/qlora.yml
@@ -4,6 +4,7 @@ base_model: cerebras/Cerebras-GPT-1.3B

 load_in_8bit: false
 load_in_4bit: true
+strict: false
 push_dataset_to_hub:
 datasets:
  - path: teknium/GPT4-LLM-Cleaned
@@ -21,6 +22,7 @@ lora_target_modules:
  - c_attn
  - c_proj
 lora_target_linear:
+lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
@@ -34,10 +36,15 @@ optimizer: paged_adamw_8bit
 torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.0002
+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
 xformers_attention: true
 flash_attention:
@@ -46,6 +53,10 @@ gptq_model_v1:
 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.1
+fsdp:
+fsdp_config:
 special_tokens:
  pad_token: "<|endoftext|>"
--- a/examples/code-llama/13b/lora.yml
+++ b/examples/code-llama/13b/lora.yml
@@ -7,6 +7,7 @@ tokenizer_type: CodeLlamaTokenizer

 load_in_8bit: true
 load_in_4bit: false
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -25,6 +26,7 @@ lora_r: 32
 lora_alpha: 16
 lora_dropout: 0.05
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -39,18 +41,29 @@ optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true
+s2_attention:

 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
--- a/examples/code-llama/13b/qlora.yml
+++ b/examples/code-llama/13b/qlora.yml
@@ -7,6 +7,7 @@ tokenizer_type: CodeLlamaTokenizer

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -25,7 +26,9 @@ pad_to_sequence_len: true
 lora_r: 32
 lora_alpha: 16
 lora_dropout: 0.05
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -40,18 +43,28 @@ optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
--- a/examples/code-llama/34b/lora.yml
+++ b/examples/code-llama/34b/lora.yml
@@ -7,6 +7,7 @@ tokenizer_type: CodeLlamaTokenizer

 load_in_8bit: true
 load_in_4bit: false
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -25,6 +26,7 @@ lora_r: 32
 lora_alpha: 16
 lora_dropout: 0.05
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -39,18 +41,29 @@ optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true
+s2_attention:

 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
--- a/examples/code-llama/34b/qlora.yml
+++ b/examples/code-llama/34b/qlora.yml
@@ -7,6 +7,7 @@ tokenizer_type: CodeLlamaTokenizer

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -25,7 +26,9 @@ pad_to_sequence_len: true
 lora_r: 32
 lora_alpha: 16
 lora_dropout: 0.05
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -40,18 +43,28 @@ optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
--- a/examples/code-llama/7b/lora.yml
+++ b/examples/code-llama/7b/lora.yml
@@ -7,6 +7,7 @@ tokenizer_type: CodeLlamaTokenizer

 load_in_8bit: true
 load_in_4bit: false
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -25,6 +26,7 @@ lora_r: 32
 lora_alpha: 16
 lora_dropout: 0.05
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -39,18 +41,29 @@ optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true
+s2_attention:

 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
--- a/examples/code-llama/7b/qlora.yml
+++ b/examples/code-llama/7b/qlora.yml
@@ -7,6 +7,7 @@ tokenizer_type: CodeLlamaTokenizer

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -25,7 +26,9 @@ pad_to_sequence_len: true
 lora_r: 32
 lora_alpha: 16
 lora_dropout: 0.05
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -40,18 +43,28 @@ optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  bos_token: "<s>"
  eos_token: "</s>"
--- a/examples/cohere/command-r-7b-qlora.yml
+++ b/examples/cohere/command-r-7b-qlora.yml
@@ -1,58 +0,0 @@
-base_model: CohereForAI/c4ai-command-r7b-12-2024
-model_type: AutoModelForCausalLM
-tokenizer_type: AutoTokenizer
-
-load_in_8bit: false
-load_in_4bit: true
-
-# huggingface repo
-chat_template: cohere
-datasets:
-  - path: cgato/SlimOrcaDedupCleaned
-    type: chat_template
-    field_messages: conversations
-    message_property_mappings:
-      role: from
-      content: value
-
-val_set_size: 0.0
-output_dir: ./outputs/out
-
-adapter: qlora
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-
-sequence_len: 2048
-sample_packing: true
-eval_sample_packing: false
-pad_to_sequence_len: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-
-gradient_accumulation_steps: 4
-micro_batch_size: 1
-num_epochs: 4
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: true
-
-gradient_checkpointing: true
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-warmup_ratio: 0.1
-evals_per_epoch:
-saves_per_epoch: 1
-weight_decay: 0.0
-special_tokens:
--- a/examples/dbrx/16bit-lora.yaml
+++ b/examples/dbrx/16bit-lora.yaml
@@ -4,6 +4,10 @@ base_model: LnL-AI/dbrx-base-converted-v2

 trust_remote_code: true

+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
 datasets:
  - path: tatsu-lab/alpaca
    type: alpaca
@@ -44,20 +48,26 @@ optimizer: paged_adamw_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: false  # don't use with fsdp_activation_checkpointing
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch:
 saves_per_epoch: 1
-
+debug:
 weight_decay: 0.0
 fsdp:
  - full_shard
--- a/examples/dbrx/8bit-lora.yaml
+++ b/examples/dbrx/8bit-lora.yaml
@@ -6,6 +6,7 @@ trust_remote_code: true

 load_in_8bit: true
 load_in_4bit: false
+strict: false

 datasets:
  - path: tatsu-lab/alpaca
@@ -47,20 +48,26 @@ optimizer: paged_adamw_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: false  # don't use with fsdp_activation_checkpointing
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch:
 saves_per_epoch: 1
-
+debug:
 weight_decay: 0.0
 fsdp:
  - full_shard
--- a/examples/dbrx/fft-ds-zero3.yaml
+++ b/examples/dbrx/fft-ds-zero3.yaml
@@ -4,6 +4,10 @@ base_model: LnL-AI/dbrx-base-converted-v2

 trust_remote_code: true

+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
 datasets:
  - path: tatsu-lab/alpaca
    type: alpaca
@@ -31,19 +35,25 @@ optimizer: paged_adamw_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch:
 saves_per_epoch: 1
-
+debug:
 weight_decay: 0.0
 deepspeed: deepspeed_configs/zero3_bf16.json
--- a/examples/deepcoder/deepcoder-14B-preview-lora.yml
+++ b/examples/deepcoder/deepcoder-14B-preview-lora.yml
@@ -1,58 +0,0 @@
-base_model: agentica-org/DeepCoder-14B-Preview
-# Automatically upload checkpoint and final model to HF
-# hub_model_id: username/custom_model_name
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: fozziethebeat/alpaca_messages_2k_test
-    type: chat_template
-    field_messages: messages
-    message_property_mappings:
-      role: role
-      content: content
-
-dataset_prepared_path:
-val_set_size: 0.05
-output_dir: ./outputs/lora-out
-
-sequence_len: 4096
-sample_packing: true
-eval_sample_packing: false
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 2
-micro_batch_size: 2
-num_epochs: 4
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: true
-
-gradient_checkpointing: true
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-warmup_steps: 10
-evals_per_epoch: 1
-saves_per_epoch: 1
-weight_decay: 0.0
-special_tokens:
--- a/examples/deepcogito/cogito-v1-preview-llama-3B-lora.yml
+++ b/examples/deepcogito/cogito-v1-preview-llama-3B-lora.yml
@@ -1,58 +0,0 @@
-base_model: deepcogito/cogito-v1-preview-llama-3B
-# Automatically upload checkpoint and final model to HF
-# hub_model_id: username/custom_model_name
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: fozziethebeat/alpaca_messages_2k_test
-    type: chat_template
-    field_messages: messages
-    message_property_mappings:
-      role: role
-      content: content
-
-dataset_prepared_path:
-val_set_size: 0.05
-output_dir: ./outputs/lora-out
-
-sequence_len: 4096
-sample_packing: true
-eval_sample_packing: false
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 2
-micro_batch_size: 2
-num_epochs: 1
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: true
-
-gradient_checkpointing: true
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-warmup_steps: 10
-evals_per_epoch: 1
-saves_per_epoch: 1
-weight_decay: 0.0
-special_tokens:
--- a/examples/deepcogito/cogito-v1-preview-qwen-14B-lora.yml
+++ b/examples/deepcogito/cogito-v1-preview-qwen-14B-lora.yml
@@ -1,58 +0,0 @@
-base_model: deepcogito/cogito-v1-preview-qwen-14B
-# Automatically upload checkpoint and final model to HF
-# hub_model_id: username/custom_model_name
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: fozziethebeat/alpaca_messages_2k_test
-    type: chat_template
-    field_messages: messages
-    message_property_mappings:
-      role: role
-      content: content
-
-dataset_prepared_path:
-val_set_size: 0.05
-output_dir: ./outputs/lora-out
-
-sequence_len: 4096
-sample_packing: true
-eval_sample_packing: false
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 2
-micro_batch_size: 2
-num_epochs: 1
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: true
-
-gradient_checkpointing: true
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-warmup_steps: 10
-evals_per_epoch: 1
-saves_per_epoch: 1
-weight_decay: 0.0
-special_tokens:
--- a/examples/deepseek-v2/fft-fsdp-16b.yaml
+++ b/examples/deepseek-v2/fft-fsdp-16b.yaml
@@ -3,6 +3,10 @@ base_model: deepseek-ai/DeepSeek-V2-Lite
 # hub_model_id: username/custom_model_name
 trust_remote_code: true

+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
 datasets:
  - path: tatsu-lab/alpaca
    type: alpaca
@@ -27,19 +31,27 @@ optimizer: adamw_torch_fused
 lr_scheduler: cosine
 learning_rate: 2e-5

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 100
 evals_per_epoch: 2
+eval_table_size:
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
 special_tokens:
 fsdp:
--- a/examples/deepseek-v2/qlora-fsdp-2_5.yaml
+++ b/examples/deepseek-v2/qlora-fsdp-2_5.yaml
@@ -6,6 +6,7 @@ trust_remote_code: true

 load_in_8bit: false
 load_in_4bit: true
+strict: false


 plugins:
@@ -51,19 +52,27 @@ optimizer: adamw_torch_fused
 lr_scheduler: cosine
 learning_rate: 2e-5

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 100
 evals_per_epoch: 2
+eval_table_size:
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
 special_tokens:
 fsdp:
--- a/examples/falcon/config-7b-lora.yml
+++ b/examples/falcon/config-7b-lora.yml
@@ -11,6 +11,7 @@ trust_remote_code: true
 load_in_8bit: true
 load_in_4bit: false
 gptq: false
+strict: false
 push_dataset_to_hub:
 datasets:
  - path: teknium/GPT4-LLM-Cleaned
@@ -24,7 +25,9 @@ max_packed_sequence_len:
 lora_r: 16
 lora_alpha: 32
 lora_dropout: 0.0
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
@@ -38,10 +41,15 @@ optimizer: adamw_bnb_8bit
 torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.00003
+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
 xformers_attention: true
 flash_attention:
@@ -50,7 +58,11 @@ gptq_model_v1:
 warmup_steps: 40
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  pad_token: "<|endoftext|>"
  bos_token: "<|endoftext|>"
--- a/examples/falcon/config-7b-qlora.yml
+++ b/examples/falcon/config-7b-qlora.yml
@@ -15,6 +15,7 @@ load_in_8bit: false
 # enable 4bit for QLoRA
 load_in_4bit: true
 gptq: false
+strict: false
 push_dataset_to_hub:
 datasets:
  - path: QingyiSi/Alpaca-CoT
@@ -37,7 +38,9 @@ lora_alpha: 16
 # 0.05 for 33B and 65B models
 lora_dropout: 0.05
 # add LoRA modules on all linear layers of the base model
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:

 wandb_project:
 wandb_entity:
@@ -64,7 +67,10 @@ lr_scheduler: cosine
 # - 2e-4 for 7b & 13b
 # - 1e-4 for 33b & 64b
 learning_rate: 0.0002
+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
 # stop training after this many evaluation losses have increased in a row
@@ -72,6 +78,7 @@ gradient_checkpointing: true
 early_stopping_patience: 3
 resume_from_checkpoint:
 auto_resume_from_checkpoints: true
+local_rank:
 logging_steps: 1
 xformers_attention: true
 flash_attention:
@@ -80,7 +87,11 @@ gptq_model_v1:
 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.000001
+fsdp:
+fsdp_config:
 special_tokens:
  pad_token: "<|endoftext|>"
  bos_token: "<|endoftext|>"
--- a/examples/falcon/config-7b.yml
+++ b/examples/falcon/config-7b.yml
@@ -7,7 +7,11 @@ tokenizer_type: AutoTokenizer

 # required by falcon custom model code: https://huggingface.co/tiiuae/falcon-7b/tree/main
 trust_remote_code: true
+
+load_in_8bit: false
+load_in_4bit: false
 gptq: false
+strict: false
 push_dataset_to_hub:
 datasets:
  - path: teknium/GPT4-LLM-Cleaned
@@ -21,7 +25,9 @@ max_packed_sequence_len:
 lora_r: 64
 lora_alpha: 32
 lora_dropout: 0.0
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
@@ -35,10 +41,15 @@ optimizer: adamw_bnb_8bit
 torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.00003
+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
 xformers_attention: true
 flash_attention:
@@ -47,7 +58,11 @@ gptq_model_v1:
 warmup_steps: 40
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
  pad_token: "<|endoftext|>"
  bos_token: "<|endoftext|>"
--- a/examples/gemma/qlora.yml
+++ b/examples/gemma/qlora.yml
@@ -8,6 +8,7 @@ tokenizer_type: AutoTokenizer

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 # huggingface repo
 datasets:
@@ -41,16 +42,28 @@ optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_ratio: 0.1
 evals_per_epoch: 4
+eval_table_size:
+eval_max_new_tokens: 128
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
--- a/examples/gemma2/qlora.yml
+++ b/examples/gemma2/qlora.yml
@@ -7,6 +7,7 @@ tokenizer_type: AutoTokenizer

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 # huggingface repo
 chat_template: gemma
@@ -47,16 +48,28 @@ optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: true

 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_ratio: 0.1
 evals_per_epoch:
+eval_table_size:
+eval_max_new_tokens: 128
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
--- a/examples/gemma2/reward-model.yaml
+++ b/examples/gemma2/reward-model.yaml
@@ -6,6 +6,10 @@ tokenizer_type: AutoTokenizer
 # Automatically upload checkpoint and final model to HF
 # hub_model_id: username/custom_model_name

+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
 reward_model: true
 chat_template: gemma
 datasets:
@@ -34,6 +38,8 @@ optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002

+train_on_inputs: false
+group_by_length: false
 bf16: true
 fp16:
 tf32: true
@@ -41,12 +47,21 @@ tf32: true
 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_ratio: 0.1
 evals_per_epoch:
+eval_table_size:
+eval_max_new_tokens: 128
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
+fsdp:
+fsdp_config:
 special_tokens:
--- a/examples/gemma3/gemma-3-1b-qlora.yml
+++ b/examples/gemma3/gemma-3-1b-qlora.yml
@@ -1,66 +0,0 @@
-base_model: google/gemma-3-1b-it
-# optionally might have model_type or tokenizer_type
-model_type: AutoModelForCausalLM
-tokenizer_type: AutoTokenizer
-# Automatically upload checkpoint and final model to HF
-# hub_model_id: username/custom_model_name
-
-# gemma3 doesn't seem to play nice with ddp
-ddp_find_unused_parameters: true
-
-load_in_8bit: false
-load_in_4bit: true
-
-# huggingface repo
-chat_template: gemma3
-datasets:
-  - path: cgato/SlimOrcaDedupCleaned
-    type: chat_template
-    field_messages: conversations
-    message_property_mappings:
-      role: from
-      content: value
-
-val_set_size: 0.0
-output_dir: ./outputs/out
-
-adapter: qlora
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-
-sequence_len: 2048
-sample_packing: true
-eval_sample_packing: false
-pad_to_sequence_len: true
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-
-gradient_accumulation_steps: 4
-micro_batch_size: 1
-num_epochs: 4
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: true
-
-gradient_checkpointing: true
-gradient_checkpointing_kwargs:
-  use_reentrant: false
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-warmup_ratio: 0.1
-evals_per_epoch:
-saves_per_epoch: 1
-weight_decay: 0.0
-special_tokens:
--- a/examples/gemma3/gemma-3-4b-qlora.yml
+++ b/examples/gemma3/gemma-3-4b-qlora.yml
@@ -1,60 +0,0 @@
-base_model: google/gemma-3-4b-it
-
-load_in_4bit: true
-
-# gemma3 doesn't seem to play nice with ddp
-ddp_find_unused_parameters: true
-
-chat_template: gemma3
-datasets:
-  - path: cgato/SlimOrcaDedupCleaned
-    type: chat_template
-    field_messages: conversations
-    message_property_mappings:
-      role: from
-      content: value
-
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./outputs/out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 2048
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules: 'language_model.model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 1
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: true
-fp16:
-tf32: true
-
-gradient_checkpointing: true
-gradient_checkpointing_kwargs:
-  use_reentrant: false
-logging_steps: 1
-flash_attention: true
-eager_attention:
-
-warmup_ratio: 0.1
-evals_per_epoch: 1
-saves_per_epoch: 1
-weight_decay: 0.0
--- a/examples/gemma3/gemma-3-4b-vision-qlora.yml
+++ b/examples/gemma3/gemma-3-4b-vision-qlora.yml
@@ -1,62 +0,0 @@
-base_model: google/gemma-3-4b-it
-processor_type: AutoProcessor
-
-load_in_4bit: true
-
-# these 3 lines are needed for now to handle vision chat templates w images
-skip_prepare_dataset: true
-remove_unused_columns: false
-sample_packing: false
-
-# gemma3 doesn't seem to play nice with ddp
-ddp_find_unused_parameters: true
-
-chat_template: gemma3
-datasets:
-  - path: HuggingFaceH4/llava-instruct-mix-vsft
-    type: chat_template
-    split: train[:1%]
-    field_messages: messages
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./outputs/out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 2048
-pad_to_sequence_len: false
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules: 'language_model.model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 1
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: true
-fp16:
-tf32: true
-
-gradient_checkpointing: true
-gradient_checkpointing_kwargs:
-  use_reentrant: false
-logging_steps: 1
-flash_attention: true
-eager_attention:
-
-warmup_ratio: 0.1
-evals_per_epoch: 1
-saves_per_epoch: 1
-weight_decay: 0.0
--- a/examples/glm4/qlora-32b.yaml
+++ b/examples/glm4/qlora-32b.yaml
@@ -1,62 +0,0 @@
-base_model: THUDM/GLM-4-32B-0414
-# Automatically upload checkpoint and final model to HF
-# hub_model_id: username/custom_model_name
-
-load_in_4bit: true
-
-datasets:
-  - path: teknium/GPT4-LLM-Cleaned
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0
-output_dir: ./outputs/qlora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 2048
-sample_packing: true
-eval_sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 16
-lora_alpha: 32
-lora_dropout: 0.05
-lora_target_modules:
-  - gate_proj
-  - down_proj
-  - up_proj
-  - q_proj
-  - v_proj
-  - k_proj
-  - o_proj
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_name:
-wandb_log_model:
-
-gradient_accumulation_steps: 2
-micro_batch_size: 2
-num_epochs: 1
-optimizer: adamw_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-bf16: auto
-tf32: false
-
-gradient_checkpointing: true
-resume_from_checkpoint:
-logging_steps: 1
-flash_attention: true
-
-loss_watchdog_threshold: 5.0
-loss_watchdog_patience: 3
-
-warmup_steps: 10
-evals_per_epoch: 1
-saves_per_epoch: 1
-weight_decay: 0.0
-special_tokens:
--- a/examples/gptj/qlora.yml
+++ b/examples/gptj/qlora.yml
@@ -4,6 +4,7 @@ base_model: EleutherAI/gpt-j-6b

 load_in_8bit: false
 load_in_4bit: true
+strict: false
 push_dataset_to_hub:
 datasets:
  - path: teknium/GPT4-LLM-Cleaned
@@ -17,7 +18,9 @@ max_packed_sequence_len:
 lora_r: 8
 lora_alpha: 32
 lora_dropout: 0.05
+lora_target_modules:
 lora_target_linear: true
+lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
@@ -31,10 +34,15 @@ optimizer: paged_adamw_8bit
 torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.0001
+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
 xformers_attention: true
 flash_attention:
@@ -43,6 +51,10 @@ gptq_model_v1:
 warmup_steps: 10
 evals_per_epoch: 4
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.1
+fsdp:
+fsdp_config:
 special_tokens:
  pad_token: "<|endoftext|>"
--- a/examples/jamba/qlora.yaml
+++ b/examples/jamba/qlora.yaml
@@ -6,6 +6,7 @@ trust_remote_code: true

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -39,18 +40,26 @@ optimizer: paged_adamw_8bit
 lr_scheduler: cosine
 learning_rate: 0.00001

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch:
 saves_per_epoch: 1
+debug:
+deepspeed:
 weight_decay: 0.0
 special_tokens:
--- a/examples/jamba/qlora_deepspeed.yaml
+++ b/examples/jamba/qlora_deepspeed.yaml
@@ -5,6 +5,7 @@ trust_remote_code: true

 load_in_8bit: false
 load_in_4bit: true
+strict: false

 datasets:
  - path: mhenrichsen/alpaca_2k_test
@@ -38,20 +39,26 @@ optimizer: paged_adamw_8bit
 lr_scheduler: cosine
 learning_rate: 0.00001

+train_on_inputs: false
+group_by_length: false
 bf16: auto
+fp16:
 tf32: false

 gradient_checkpointing: true
 gradient_checkpointing_kwargs:
  use_reentrant: false
+early_stopping_patience:
 resume_from_checkpoint:
+local_rank:
 logging_steps: 1
+xformers_attention:
 flash_attention: true

 warmup_steps: 10
 evals_per_epoch:
 saves_per_epoch: 1
-
+debug:
 deepspeed: deepspeed_configs/zero2.json
 weight_decay: 0.0
 special_tokens:
--- a/examples/jamba/qlora_fsdp_large.yaml
+++ b/examples/jamba/qlora_fsdp_large.yaml
@@ -5,6 +5,7 @@ tokenizer_type: AutoTokenizer
 # hub_model_id: username/custom_model_name

 load_in_4bit: true
+strict: false
 use_tensorboard: true
 chat_template: jamba
 datasets:
@@ -38,6 +39,8 @@ optimizer: adamw_torch_fused
 lr_scheduler: cosine
 learning_rate: 0.00001

+train_on_inputs: false
+group_by_length: false
 bf16: true
 tf32: true

--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Sung Ching Liu	328bb0466b	Merge branch 'main' into flx_attn_support	2025-02-21 11:27:25 -05:00
Sunny Liu	e792b54bab	remove unnecessary components	2025-02-21 11:23:21 -05:00
bursteratom	82d04ea060	test v2batch w/ flex attn	2025-02-13 00:11:45 -05:00
Sung Ching Liu	0ef1f011fe	Merge branch 'main' into flx_attn_support	2025-02-11 23:31:56 -05:00
Sunny Liu	c0a1d205c7	packed doc mask starts at 1, 0 means masked out	2025-02-07 14:44:52 -05:00
Sunny Liu	d0e739da24	attempt at getting around bf16 error	2025-02-04 21:57:21 -05:00
Sunny Liu	3f6be519d5	stack	2025-02-04 21:25:13 -05:00
Sunny Liu	adcbc7459b	misc	2025-02-04 21:17:50 -05:00
Sunny Liu	470ba65c44	make doc mask instead of the whole block mask in collator	2025-02-04 20:27:39 -05:00
Sunny Liu	8e1adc154d	stuff	2025-02-02 20:36:14 -05:00
Sunny Liu	e5b36900e4	misc	2025-02-02 20:32:03 -05:00
Sunny Liu	9f6c89b12b	undo my stupidity	2025-02-02 20:25:53 -05:00
Sunny Liu	b0871c8d3b	attempt - mask padding	2025-02-02 20:18:49 -05:00
bursteratom	d3ea379a23	figure out slight diff from flash result	2025-02-02 01:45:54 -05:00
bursteratom	0ebab63309	test	2025-02-02 01:27:15 -05:00
bursteratom	e98581f6f5	BLOCK SIZE	2025-02-02 01:22:23 -05:00
bursteratom	b832b11c8f	stuff	2025-02-02 00:51:43 -05:00
bursteratom	b692d394b1	more test	2025-02-02 00:48:57 -05:00
bursteratom	2319e5276d	more test	2025-02-02 00:48:15 -05:00
bursteratom	9a43a0925d	more test	2025-02-02 00:45:30 -05:00
bursteratom	10de67e8ea	more test	2025-02-02 00:43:41 -05:00
bursteratom	fa7355404c	test	2025-02-02 00:38:35 -05:00
bursteratom	907424a2e8	stuff	2025-02-02 00:29:09 -05:00
Sunny Liu	3f4fd3c1eb	remove padding self attention	2025-02-01 22:47:10 -05:00
Sunny Liu	48c3c47071	vanills mask	2025-02-01 14:23:37 -05:00
Sunny Liu	3ed9c117fb	try vanilla mask	2025-02-01 14:09:13 -05:00
Sunny Liu	84960003ed	reset llama_patch_multipack.py	2025-01-30 14:40:18 -05:00
Sunny Liu	93a268e43d	--no-verify fixes silly mistake	2025-01-30 14:08:26 -05:00
Sunny Liu	065f6d477e	flex batching WIP	2025-01-30 14:04:59 -05:00
Sunny Liu	96ad741cd5	flex batching WIP	2025-01-30 12:35:25 -05:00
bursteratom	ba88bc7840	wip flex block mask creation	2025-01-29 00:25:25 -05:00
Sung Ching Liu	b31796a681	Merge branch 'main' into flx_attn_support	2025-01-28 14:20:43 -05:00
Sunny Liu	5ca57cb55a	undo bool conversion	2025-01-23 17:56:13 -05:00
Sunny Liu	0149de7fb0	mask to bool	2025-01-23 15:30:08 -05:00
Sunny Liu	8c34c65181	dummy	2025-01-23 14:56:26 -05:00
Sunny Liu	555aa5772a	skip mask conversion if already 4d	2025-01-23 14:01:53 -05:00
Sunny Liu	e8b2789086	revert mask expand	2025-01-23 11:20:38 -05:00
Sunny Liu	85752cdfc9	mask expansion	2025-01-22 21:33:38 -05:00
Sunny Liu	f2f23c8041	mask expansion	2025-01-22 21:31:42 -05:00
Sunny Liu	8b3eec7f6e	mask expansion	2025-01-22 21:29:52 -05:00
Sunny Liu	bb9bea3110	mask expansion	2025-01-22 21:27:25 -05:00
Sunny Liu	0dd18a3681	llama sdpa patching WIP - static class function import	2025-01-22 21:10:05 -05:00
Sunny Liu	152e988d3c	llama sdpa patching WIP - static class function import	2025-01-22 21:02:26 -05:00
Sunny Liu	27532825a9	llama sdpa patching WIP - static class function import	2025-01-22 21:00:34 -05:00
Sunny Liu	06f83a54a5	llama sdpa patching WIP - static class function import	2025-01-22 20:45:44 -05:00
Sunny Liu	d7b133dc1f	llama sdpa patching WIP - static class function import	2025-01-22 20:33:13 -05:00
Sunny Liu	f3bec17917	llama sdpa patching WIP - static class function import	2025-01-22 20:25:26 -05:00
Sunny Liu	b7deb5241c	llama sdpa patching WIP	2025-01-22 20:16:27 -05:00
Sunny Liu	cee310dcfa	llama sdpa patching WIP	2025-01-22 20:15:23 -05:00
Sunny Liu	d1be6e228d	llama sdpa patching WIP	2025-01-22 20:14:20 -05:00
Sunny Liu	5f9f77f384	llama patch	2025-01-22 11:29:28 -05:00
bursteratom	b2a34380b3	sample packing doc mask creation WIP	2025-01-21 09:18:38 -05:00
Sunny Liu	80bfc50d1f	get seqlens from position ids for foc masking	2025-01-17 17:22:04 -05:00
Sunny Liu	a5360c172c	llama hijacking	2025-01-17 15:54:03 -05:00
Sunny Liu	013a9b73fc	fix transformers version for testing	2025-01-16 15:32:57 -05:00
Sunny	aad62428e0	not sure if this is necessary actually	2025-01-16 15:08:34 -05:00
Sunny	a6f2c5d583	flex sample packing WIP	2025-01-15 21:12:33 -05:00
Sunny	dbcd11e533	revert seq len in multipack sampler	2025-01-14 11:45:35 -05:00
Sunny	c06a6be915	flex_attn sample packing WIP	2025-01-14 00:22:05 -05:00
bursteratom	d3a0cb5edb	transformers version	2025-01-13 10:33:00 -05:00
bursteratom	8b47e456b0	revert to transformers 4.47.1	2025-01-13 10:29:27 -05:00
Sunny Liu	2319ac729c	Merge branch 'main' into flx_attn_support	2025-01-13 09:42:58 -05:00
Sunny	f99cae0e7b	llama test	2025-01-12 17:30:19 -05:00
Wing Lian	888cd9407f	use 2.5.1 docker images as latest tag as it seems stable (#2198 )	2025-01-12 13:34:17 -05:00
Wing Lian	bd62d6e10a	rename liger test so it properly runs in ci (#2246 )	2025-01-12 13:34:17 -05:00
NanoCode012	5eae134110	feat: add support for data_files in pretraining (#2238 )	2025-01-12 13:34:17 -05:00
Wing Lian	b7d27bdfa4	update upstream HF deps (#2239 ) * bump axolotl contribs for upstream main conflicts: * bump datasets, tokenizer, trl * remove log workarounds in trl * bump lm-eval * remove unsloth_ import from critical path * remove llama fa2 from conftest * unsloth breaks with latest upstream	2025-01-12 13:34:17 -05:00
Vincenzo di Cicco	da97a21bdc	Use SequentialSampler if curriculum_sampling is enabled with sample_packing (#2235 )	2025-01-12 13:34:17 -05:00
Wing Lian	e0d4b88598	update modal version for ci (#2242 )	2025-01-12 13:34:17 -05:00
NanoCode012	fac059a209	fix: mistral nemo does not recognize token_type_ids in forward (#2233 )	2025-01-12 13:34:17 -05:00
Wing Lian	9c9ac1cf0b	add hf cache caching for GHA (#2247 ) * add hf cache caching for GHA * use modal volume to cache hf data * make sure to update the cache as we add new fixtures in conftest	2025-01-12 13:34:17 -05:00
Wing Lian	2346f21b2b	Merge group queue (#2248 ) * add support for merge groups * also lint merge groups	2025-01-12 13:34:17 -05:00
salman	0b47281f51	Fixing OSX installation (#2231 ) * bumping version, removing non-osx compatible deps * updating pylintrc * fixing linters * reverting changes	2025-01-12 13:34:17 -05:00
Sunny	543daaf46f	llama test	2025-01-09 16:08:24 -05:00
Sunny	bcd9ad44e0	flex attention support	2025-01-06 19:54:11 -05:00
bursteratom	61ad375bf4	config validation for flex attention	2025-01-05 23:27:49 -05:00