basic llama multipack

Allow in message objects the additional key `weight`, which can be set
to 0 (or 1) to cause that message to be masked out (or left unmasked)
for training (similar to [1]). This is helpful for training the model to be robust and
capable of error recovery upon a bad assistant message.
A missing `weight` key defaults to weight 1, to guarantee downward compatibility.

[1]: https://github.com/mistralai/mistral-finetune

.github/FUNDING.yml

@@ -1,13 +1,13 @@
 # These are supported funding model platforms
 
-github: OpenAccess-AI-Collective # Replace with up to 4 GitHub Sponsors-enabled usernames e.g., [user1, user2]
+github: [winglian, OpenAccess-AI-Collective] # Replace with up to 4 GitHub Sponsors-enabled usernames e.g., [user1, user2]
 patreon: # Replace with a single Patreon username
 open_collective: # Replace with a single Open Collective username
-ko_fi: # Replace with a single Ko-fi username
+ko_fi: axolotl_ai # Replace with a single Ko-fi username
 tidelift: # Replace with a single Tidelift platform-name/package-name e.g., npm/babel
 community_bridge: # Replace with a single Community Bridge project-name e.g., cloud-foundry
 liberapay: # Replace with a single Liberapay username
 issuehunt: # Replace with a single IssueHunt username
 otechie: # Replace with a single Otechie username
 lfx_crowdfunding: # Replace with a single LFX Crowdfunding project-name e.g., cloud-foundry
-custom: # Replace with up to 4 custom sponsorship URLs e.g., ['link1', 'link2']
+custom: ['https://quickchart.io/qr?text=bitcoin%3Abc1qxlgwlqwfea5s2cxm42xqsfmwjct0rj8w8ea5np&size=480&centerImageUrl=https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2F4%2F46%2FBitcoin.svg%2F64px-Bitcoin.svg.png'] # Replace with up to 4 custom sponsorship URLs e.g., ['link1', 'link2']

.github/ISSUE_TEMPLATE/bug-report.yaml

@@ -53,6 +53,14 @@ body:
     validations:
       required: true
 
+  - type: textarea
+    id: config
+    attributes:
+      label: Config yaml
+      description: |
+        Please attach the config yaml!
+      render: yaml
+
   - type: textarea
     id: possible-solution
     attributes:

.github/PULL_REQUEST_TEMPLATE.md

@@ -20,3 +20,8 @@
 ## Types of changes
 
 <!--- What types of changes does your code introduce? Put an `x` in all the boxes that apply: -->
+
+## Social Handles (Optional)
+
+<!-- Thanks for submitting a bugfix or enhancement. -->
+<!-- We'd love to show our thanks to you on Twitter & Discord if you provide your handle -->

.github/workflows/base.yml

@@ -1,30 +1,42 @@
 name: ci-cd-base
 
 on:
-  push:
-    branches:
-      - "main-base"
-      - "dev-base"
+  workflow_dispatch:
 
 jobs:
   build-base:
     if: github.repository_owner == 'OpenAccess-AI-Collective'
     # this job needs to be run on self-hosted GPU runners...
-    runs-on: self-hosted
+    runs-on: axolotl-gpu-runner
     strategy:
       fail-fast: false
       matrix:
         include:
-          - cuda: "118"
-            cuda_version: 11.8.0
-            python_version: "3.9"
-            pytorch: 2.0.1
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 9.0+PTX"
           - cuda: "118"
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.1
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 9.0+PTX"
+            pytorch: 2.1.2
+            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+          - cuda: "121"
+            cuda_version: 12.1.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+          - cuda: "121"
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.1.2
+            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+          - cuda: "121"
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.2.2
+            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+          - cuda: "121"
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            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@v3
@@ -46,7 +58,7 @@ jobs:
           context: .
           file: ./docker/Dockerfile-base
           push: ${{ github.event_name != 'pull_request' }}
-          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+          tags: ${{ steps.metadata.outputs.tags }}-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           labels: ${{ steps.metadata.outputs.labels }}
           build-args: |
             CUDA_VERSION=${{ matrix.cuda_version }}

.github/workflows/docs.yml

@@ -0,0 +1,31 @@
+name: Publish Docs
+on:
+  push:
+    branches:
+      - main
+
+permissions:
+    contents: write
+    pages: write
+
+jobs:
+    build-deploy:
+        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@v3
+          with:
+            python-version: '3.10'
+        - name: install dependencies
+          run: |
+            python3 -m pip install jupyter
+        - name: Publish to GitHub Pages (and render)
+          uses: quarto-dev/quarto-actions/publish@v2
+          with:
+            target: gh-pages
+          env:
+            GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/lint.yml

@@ -0,0 +1,23 @@
+name: lint
+on:
+  # check on PRs, and manual triggers
+  pull_request:
+      paths:
+       - '**.py'
+       - 'requirements.txt'
+       - '.github/workflows/*.yml'
+       - "*.md"
+       - "examples/**/*.y[a]?ml"
+  workflow_dispatch:
+
+jobs:
+  pre-commit:
+    name: pre-commit
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: actions/setup-python@v4
+        with:
+          python-version: "3.10"
+          cache: 'pip' # caching pip dependencies
+      - uses: pre-commit/action@v3.0.0

.github/workflows/main.yml

@@ -4,104 +4,164 @@ on:
   push:
     branches:
       - "main"
+  workflow_dispatch:
 
 jobs:
   build-axolotl:
-    if: github.repository_owner == 'OpenAccess-AI-Collective'
-    # this job needs to be run on self-hosted GPU runners...
+    if: ${{ ! contains(github.event.commits[0].message, '[skip docker]]') && github.repository_owner == 'OpenAccess-AI-Collective' }}
     strategy:
       fail-fast: false
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
-            python_version: "3.9"
-            pytorch: 2.0.1
-            axolotl_extras:
           - cuda: 118
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.1
+            pytorch: 2.1.2
             axolotl_extras:
-          - cuda: 118
-            cuda_version: 11.8.0
-            python_version: "3.9"
-            pytorch: 2.0.1
-            axolotl_extras: gptq
-    runs-on: self-hosted
+            axolotl_args: "--extra-index-url https://download.pytorch.org/whl/cu118"
+            is_latest: true
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.2.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            axolotl_extras:
+    runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
       - name: Docker metadata
         id: metadata
-        uses: docker/metadata-action@v3
+        uses: docker/metadata-action@v5
         with:
           images: winglian/axolotl
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
       - name: Login to Docker Hub
-        uses: docker/login-action@v2
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
-      - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v2
-      - name: Build
-        uses: docker/build-push-action@v4
+      # guidance for testing before pushing: https://docs.docker.com/build/ci/github-actions/test-before-push/
+      - name: Build and export to Docker
+        uses: docker/build-push-action@v5
         with:
           context: .
           build-args: |
             BASE_TAG=${{ github.ref_name }}-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}
             CUDA=${{ matrix.cuda }}
+            PYTORCH_VERSION=${{ matrix.pytorch }}
+            AXOLOTL_ARGS=${{ matrix.axolotl_args }}
           file: ./docker/Dockerfile
           push: ${{ github.event_name != 'pull_request' }}
-          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+          tags: |
+            ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+            ${{ (matrix.is_latest) && format('{0}-latest', steps.metadata.outputs.tags) || '' }}
           labels: ${{ steps.metadata.outputs.labels }}
-  build-axolotl-runpod:
+
+  build-axolotl-cloud:
     needs: build-axolotl
-    if: github.repository_owner == 'OpenAccess-AI-Collective'
+    if: ${{ ! contains(github.event.commits[0].message, '[skip docker]]') && github.repository_owner == 'OpenAccess-AI-Collective' }}
     # this job needs to be run on self-hosted GPU runners...
     strategy:
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
-            python_version: "3.9"
-            pytorch: 2.0.1
-            axolotl_extras:
           - cuda: 118
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.1
+            pytorch: 2.1.2
             axolotl_extras:
             is_latest: true
-          - cuda: 118
-            cuda_version: 11.8.0
-            python_version: "3.9"
-            pytorch: 2.0.1
-            axolotl_extras: gptq
-    runs-on: self-hosted
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.2.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            axolotl_extras:
+    runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
       - name: Docker metadata
         id: metadata
-        uses: docker/metadata-action@v3
+        uses: docker/metadata-action@v5
         with:
-          images: winglian/axolotl-runpod
+          images: winglian/axolotl-cloud
       - name: Login to Docker Hub
-        uses: docker/login-action@v2
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v2
       - name: Build
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           context: .
           build-args: |
             BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
             CUDA=${{ matrix.cuda }}
-          file: ./docker/Dockerfile-runpod
+          file: ./docker/Dockerfile-cloud
+          push: ${{ github.event_name != 'pull_request' }}
+          tags: |
+             ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+             ${{ (matrix.is_latest) && format('{0}-latest', steps.metadata.outputs.tags) || '' }}
+          labels: ${{ steps.metadata.outputs.labels }}
+
+  build-axolotl-cloud-no-tmux:
+    needs: build-axolotl
+    if: ${{ ! contains(github.event.commits[0].message, '[skip docker]]') && github.repository_owner == 'OpenAccess-AI-Collective' }}
+    # this job needs to be run on self-hosted GPU runners...
+    strategy:
+      matrix:
+        include:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            axolotl_extras:
+    runs-on: axolotl-gpu-runner
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Docker metadata
+        id: metadata
+        uses: docker/metadata-action@v5
+        with:
+          images: winglian/axolotl-cloud-term
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+      - name: Build
+        uses: docker/build-push-action@v5
+        with:
+          context: .
+          build-args: |
+            BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+            CUDA=${{ matrix.cuda }}
+          file: ./docker/Dockerfile-cloud-no-tmux
           push: ${{ github.event_name != 'pull_request' }}
           tags: |
              ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}

.github/workflows/nightlies.yml

@@ -0,0 +1,128 @@
+name: docker-nightlies
+
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: '0 0 * * *'  # Runs at 00:00 UTC every day
+
+jobs:
+  build-axolotl:
+    if: ${{ ! contains(github.event.commits[0].message, '[skip docker]]') && github.repository_owner == 'OpenAccess-AI-Collective' }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - cuda: 118
+            cuda_version: 11.8.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_extras:
+            axolotl_args: "--extra-index-url https://download.pytorch.org/whl/cu118"
+            is_latest: true
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.2.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            axolotl_extras:
+    runs-on: axolotl-gpu-runner
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Docker metadata
+        id: metadata
+        uses: docker/metadata-action@v5
+        with:
+          images: winglian/axolotl
+          tags: |
+            type=raw,value={{ branch }}-{{ date 'YYYYMMDD' }}
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+      # guidance for testing before pushing: https://docs.docker.com/build/ci/github-actions/test-before-push/
+      - name: Build and export to Docker
+        uses: docker/build-push-action@v5
+        with:
+          context: .
+          build-args: |
+            BASE_TAG=${{ github.ref_name }}-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}
+            CUDA=${{ matrix.cuda }}
+            PYTORCH_VERSION=${{ matrix.pytorch }}
+            AXOLOTL_ARGS=${{ matrix.axolotl_args }}
+          file: ./docker/Dockerfile
+          push: ${{ github.event_name != 'pull_request' }}
+          tags: |
+            ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+          labels: ${{ steps.metadata.outputs.labels }}
+
+  build-axolotl-cloud:
+    needs: build-axolotl
+    if: ${{ ! contains(github.event.commits[0].message, '[skip docker]]') && github.repository_owner == 'OpenAccess-AI-Collective' }}
+    # this job needs to be run on self-hosted GPU runners...
+    strategy:
+      matrix:
+        include:
+          - cuda: 118
+            cuda_version: 11.8.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_extras:
+            is_latest: true
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.2.2
+            axolotl_extras:
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            axolotl_extras:
+    runs-on: axolotl-gpu-runner
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Docker metadata
+        id: metadata
+        uses: docker/metadata-action@v5
+        with:
+          images: winglian/axolotl-cloud
+          tags: |
+            type=raw,value={{ branch }}-{{ date 'YYYYMMDD' }}
+      - name: Login to Docker Hub
+        uses: docker/login-action@v3
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v2
+      - name: Build
+        uses: docker/build-push-action@v5
+        with:
+          context: .
+          build-args: |
+            BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+            CUDA=${{ matrix.cuda }}
+          file: ./docker/Dockerfile-cloud
+          push: ${{ github.event_name != 'pull_request' }}
+          tags: |
+             ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+          labels: ${{ steps.metadata.outputs.labels }}

.github/workflows/pre-commit.yml

@@ -1,16 +0,0 @@
-name: pre-commit
-
-on:
-  pull_request:
-  push:
-
-jobs:
-  pre-commit:
-    runs-on: ubuntu-latest
-    steps:
-    - uses: actions/checkout@v3
-    - uses: actions/setup-python@v4
-      with:
-          python-version: "3.9"
-          cache: 'pip' # caching pip dependencies
-    - uses: pre-commit/action@v3.0.0

.github/workflows/pypi.yml

@@ -0,0 +1,45 @@
+name: publish pypi
+
+on:
+  push:
+    tags:
+      - '*'
+
+jobs:
+  pypi-publish:
+    name: Upload release to PyPI
+    runs-on: ubuntu-latest
+    environment:
+      name: pypi
+      url: https://pypi.org/p/axolotl
+    permissions:
+      id-token: write  # IMPORTANT: this permission is mandatory for trusted publishing
+    steps:
+      - name: Check out repository code
+        uses: actions/checkout@v3
+
+      - name: Setup Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.10"
+
+      - name: Install dependencies
+        run: |
+          pip3 install wheel packaging
+          pip3 install -e .
+          pip3 install -r requirements-tests.txt
+
+      - name: Extract tag name
+        id: tag
+        run: echo ::set-output name=TAG_NAME::$(echo $GITHUB_REF | cut -d / -f 3)
+
+      - name: Update version in setup.py
+        run: |
+          sed -i -E 's/version="([0-9.]+)",/version="${{ steps.tag.outputs.TAG_NAME }}",/g' setup.py
+
+      - name: Build a binary wheel
+        run: |
+          python setup.py sdist bdist_wheel
+
+      - name: Publish package distributions to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1

.github/workflows/tests.yml

@@ -1,16 +1,40 @@
-name: PyTest
+name: Tests
 on:
+  # check on push/merge to main, PRs, and manual triggers
   push:
+    branches:
+      - "main"
+    paths:
+      - '**.py'
+      - 'requirements.txt'
+      - '.github/workflows/*.yml'
   pull_request:
+      paths:
+       - '**.py'
+       - 'requirements.txt'
+       - '.github/workflows/*.yml'
+  workflow_dispatch:
 
 jobs:
-  test:
+  pre-commit:
+    name: pre-commit
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v3
+      - uses: actions/setup-python@v4
+        with:
+          python-version: "3.10"
+          cache: 'pip' # caching pip dependencies
+      - uses: pre-commit/action@v3.0.0
+
+  pytest:
+    name: PyTest
     runs-on: ubuntu-latest
     strategy:
       fail-fast: false
       matrix:
-        python_version: ["3.9", "3.10"]
-    timeout-minutes: 10
+        python_version: ["3.10", "3.11"]
+    timeout-minutes: 20
 
     steps:
       - name: Check out repository code
@@ -24,9 +48,65 @@ jobs:
 
       - name: Install dependencies
         run: |
-          pip install -e .[peft]
-          pip install -r requirements-tests.txt
+          pip3 install --upgrade pip
+          pip3 install --upgrade packaging
+          pip3 install -U -e .
+          pip3 install -r requirements-tests.txt
 
       - name: Run tests
         run: |
-          pytest tests/
+          pytest --ignore=tests/e2e/ tests/
+
+  docker-e2e-tests:
+    if: github.repository_owner == 'OpenAccess-AI-Collective'
+    # this job needs to be run on self-hosted GPU runners...
+    runs-on: [self-hosted, modal]
+    timeout-minutes: 60
+    needs: [pre-commit, pytest]
+
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - cuda: 118
+            cuda_version: 11.8.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            axolotl_args: "--extra-index-url https://download.pytorch.org/whl/cu118"
+            num_gpus: 1
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.10"
+            pytorch: 2.1.2
+            num_gpus: 1
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.2.2
+            num_gpus: 1
+          - cuda: 121
+            cuda_version: 12.1.0
+            python_version: "3.11"
+            pytorch: 2.3.0
+            num_gpus: 1
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Install Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.10"
+      - name: Install Modal
+        run: |
+          python -m pip install --upgrade pip
+          pip install modal jinja2
+      - name: Update env vars
+        run: |
+          echo "BASE_TAG=main-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}" >> $GITHUB_ENV
+          echo "PYTORCH_VERSION=${{ matrix.pytorch}}" >> $GITHUB_ENV
+          echo "AXOLOTL_ARGS=${{ matrix.axolotl_args}}" >> $GITHUB_ENV
+          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
+          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
+      - name: Run tests job on Modal
+        run: |
+          modal run cicd.tests

.gitignore

@@ -1,5 +1,8 @@
 **/axolotl.egg-info
 configs
+last_run_prepared/
+.vscode
+_site/
 
 # Byte-compiled / optimized / DLL files
 __pycache__/
@@ -130,6 +133,7 @@ venv/
 ENV/
 env.bak/
 venv.bak/
+venv3.10/
 
 # Spyder project settings
 .spyderproject
@@ -161,3 +165,20 @@ cython_debug/
 #  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/
+
+# WandB
+# wandb creates a folder to store logs for training runs
+wandb
+
+# Runs
+lora-out/*
+qlora-out/*
+mlruns/*
+
+/.quarto/
+prepared-datasets/
+submit.sh
+*.out*
+
+typings/
+out/

.isort.cfg

@@ -1,2 +1,3 @@
 [settings]
 profile=black
+known_third_party=wandb

.mypy.ini

@@ -1,5 +1,5 @@
 [mypy]
-
+plugins = pydantic.mypy
 exclude = venv
 
 [mypy-alpaca_lora_4bit.*]
@@ -8,6 +8,12 @@ ignore_missing_imports = True
 [mypy-axolotl.monkeypatch.*]
 ignore_errors = True
 
+[mypy-axolotl.models.mixtral.*]
+ignore_errors = True
+
+[mypy-axolotl.models.phi.*]
+ignore_errors = True
+
 [mypy-flash_attn.*]
 ignore_missing_imports = True
 
@@ -20,9 +26,15 @@ ignore_missing_imports = True
 [mypy-peft]
 ignore_missing_imports = True
 
+[mypy-wandb]
+ignore_missing_imports = True
+
 [mypy-bitsandbytes]
 ignore_missing_imports = True
 
+[mypy-requests]
+ignore_missing_imports = True
+
 [mypy-datasets]
 ignore_missing_imports = True
 

.pre-commit-config.yaml

@@ -31,6 +31,7 @@ repos:
       additional_dependencies:
         [
             'types-PyYAML',
+            'pydantic>=2.5.3',
         ]
 -   repo: https://github.com/PyCQA/bandit
     rev: 1.7.5

.vscode/README.md

@@ -0,0 +1 @@
+See [docs/debugging.md](../docs/debugging.md) for guidance on how to modify these files to debug axolotl with VSCode.

.vscode/launch.json

@@ -0,0 +1,34 @@
+{
+    // Use IntelliSense to learn about possible attributes.
+    // Hover to view descriptions of existing attributes.
+    // For more information, visit: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Debug axolotl prompt - sharegpt",
+            "type": "python",
+            "module": "accelerate.commands.launch",
+            "request": "launch",
+            "args": [
+                "-m", "axolotl.cli.train", "dev_sharegpt.yml",
+                // The flags below simplify debugging by overriding the axolotl config
+                // with the debugging tips above.  Modify as needed.
+                "--dataset_processes=1",      // limits data preprocessing to one process
+                "--max_steps=1",              // limits training to just one step
+                "--batch_size=1",             // minimizes batch size
+                "--micro_batch_size=1",       // minimizes batch size
+                "--val_set_size=0",           // disables validation
+                "--sample_packing=False",     // disables sample packing which is necessary for small datasets
+                "--eval_sample_packing=False",// disables sample packing on eval set
+                "--dataset_prepared_path=temp_debug/axolotl_outputs/data", // send data outputs to a temp folder
+                "--output_dir=temp_debug/axolotl_outputs/model" // send model outputs to a temp folder
+                ],
+            "console": "integratedTerminal",      // show output in the integrated terminal
+            "cwd": "${workspaceFolder}/devtools", // set working directory to devtools from the root of the project
+            "justMyCode": true,                   // step through only axolotl code
+            "env": {"CUDA_VISIBLE_DEVICES": "0",  // Since we aren't doing distributed training, we need to limit to one GPU
+                    "HF_HOME": "${workspaceFolder}/devtools/temp_debug/.hf-cache"}, // send HF cache to a temp folder
+            "preLaunchTask": "cleanup-for-dataprep", // delete temp folders (see below)
+        }
+    ]
+}

.vscode/tasks.json

@@ -0,0 +1,27 @@
+//this file is used by launch.json
+{
+  "version": "2.0.0",
+  "tasks": [
+    // this task changes into the devtools directory and deletes the temp_debug/axolotl_outputs folder
+    {
+      "label": "delete-outputs",
+      "type": "shell",
+      "command": "rm -rf temp_debug/axolotl_outputs",
+      "options":{ "cwd": "${workspaceFolder}/devtools"},
+      "problemMatcher": []
+    },
+    // this task changes into the devtools directory and deletes the `temp_debug/.hf-cache/datasets` folder
+    {
+      "label": "delete-temp-hf-dataset-cache",
+      "type": "shell",
+      "command": "rm -rf temp_debug/.hf-cache/datasets",
+      "options":{ "cwd": "${workspaceFolder}/devtools"},
+      "problemMatcher": []
+    },
+      // this task combines the two tasks above
+    {
+     "label": "cleanup-for-dataprep",
+     "dependsOn": ["delete-outputs", "delete-temp-hf-dataset-cache"],
+    }
+  ]
+}

_quarto.yml

@@ -0,0 +1,51 @@
+project:
+  type: website
+
+website:
+  title: "Axolotl"
+  description: "Fine-tuning"
+  favicon: favicon.jpg
+  navbar:
+    title: Axolotl
+    background: dark
+    pinned: false
+    collapse: false
+    tools:
+    - icon: twitter
+      href: https://twitter.com/axolotl_ai
+    - icon: github
+      href: https://github.com/OpenAccess-AI-Collective/axolotl/
+    - icon: discord
+      href: https://discord.gg/7m9sfhzaf3
+
+  sidebar:
+      pinned: true
+      collapse-level: 2
+      style: docked
+      contents:
+        - text: Home
+          href: index.qmd
+        - section: "How-To Guides"
+          contents:
+          # TODO Edit folder structure after we have more docs.
+            - docs/debugging.qmd
+            - docs/multipack.qmd
+            - docs/fsdp_qlora.qmd
+            - docs/input_output.qmd
+            - docs/rlhf.qmd
+            - docs/nccl.qmd
+            - docs/mac.qmd
+            - docs/multi-node.qmd
+        - section: "Dataset Formats"
+          contents: docs/dataset-formats/*
+        - section: "Reference"
+          contents:
+            - docs/config.qmd
+        - docs/faq.qmd
+
+
+format:
+  html:
+    theme: materia
+    css: styles.css
+    toc: true

cicd/Dockerfile.jinja

@@ -0,0 +1,40 @@
+FROM winglian/axolotl-base:{{ BASE_TAG }}
+
+ENV TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
+ENV AXOLOTL_EXTRAS="{{ AXOLOTL_EXTRAS }}"
+ENV AXOLOTL_ARGS="{{ AXOLOTL_ARGS }}"
+ENV CUDA="{{ CUDA }}"
+ENV BNB_CUDA_VERSION="{{ CUDA }}"
+ENV PYTORCH_VERSION="{{ PYTORCH_VERSION }}"
+ENV GITHUB_REF="{{ GITHUB_REF }}"
+ENV GITHUB_SHA="{{ GITHUB_SHA }}"
+
+RUN apt-get update && \
+    apt-get install -y --allow-change-held-packages vim curl nano libnccl2 libnccl-dev
+
+WORKDIR /workspace
+
+RUN git clone --depth=1 https://github.com/OpenAccess-AI-Collective/axolotl.git
+
+WORKDIR /workspace/axolotl
+
+RUN git fetch origin +$GITHUB_REF && \
+    git checkout FETCH_HEAD
+
+# If AXOLOTL_EXTRAS is set, append it in brackets
+RUN pip install causal_conv1d
+RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+    else \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore] $AXOLOTL_ARGS; \
+    fi
+
+# So we can test the Docker image
+RUN pip install pytest
+
+# fix so that git fetch/pull from remote works
+RUN git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \
+    git config --get remote.origin.fetch
+
+# helper for huggingface-login cli
+RUN git config --global credential.helper store

cicd/cicd.sh

@@ -0,0 +1,6 @@
+#!/bin/bash
+set -e
+
+pytest --ignore=tests/e2e/ /workspace/axolotl/tests/
+pytest /workspace/axolotl/tests/e2e/patched/
+pytest --ignore=tests/e2e/patched/ /workspace/axolotl/tests/e2e/

cicd/tests.py

@@ -0,0 +1,75 @@
+"""
+ modal application to run axolotl gpu tests in Modal
+ """
+import os
+import pathlib
+import tempfile
+
+import jinja2
+import modal
+from jinja2 import select_autoescape
+from modal import Image, Stub
+
+cicd_path = pathlib.Path(__file__).parent.resolve()
+
+template_loader = jinja2.FileSystemLoader(searchpath=cicd_path)
+template_env = jinja2.Environment(
+    loader=template_loader, autoescape=select_autoescape()
+)
+df_template = template_env.get_template("Dockerfile.jinja")
+
+df_args = {
+    "AXOLOTL_EXTRAS": os.environ.get("AXOLOTL_EXTRAS", ""),
+    "AXOLOTL_ARGS": os.environ.get("AXOLOTL_ARGS", ""),
+    "PYTORCH_VERSION": os.environ.get("PYTORCH_VERSION", "2.0.1"),
+    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.10-cu118-2.0.1"),
+    "CUDA": os.environ.get("CUDA", "118"),
+    "GITHUB_REF": os.environ.get("GITHUB_REF", "refs/heads/main"),
+    "GITHUB_SHA": os.environ.get("GITHUB_SHA", ""),
+}
+
+dockerfile_contents = df_template.render(**df_args)
+
+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)
+    .pip_install("fastapi==0.110.0", "pydantic==2.6.3")
+)
+
+stub = Stub("Axolotl CI/CD", secrets=[])
+
+
+N_GPUS = int(os.environ.get("N_GPUS", 1))
+GPU_CONFIG = modal.gpu.A10G(count=N_GPUS)
+
+
+def run_cmd(cmd: str, run_folder: str):
+    import subprocess  # nosec
+
+    # Propagate errors from subprocess.
+    if exit_code := subprocess.call(cmd.split(), cwd=run_folder):  # nosec
+        exit(exit_code)  # pylint: disable=consider-using-sys-exit
+
+
+@stub.function(
+    image=cicd_image,
+    gpu=GPU_CONFIG,
+    timeout=45 * 60,
+    cpu=8.0,
+    memory=131072,
+)
+def cicd_pytest():
+    run_cmd("./cicd/cicd.sh", "/workspace/axolotl")
+
+
+@stub.local_entrypoint()
+def main():
+    cicd_pytest.remote()

deepspeed/zero2.json

@@ -1,46 +0,0 @@
-{
-    "zero_optimization": {
-      "stage": 2,
-      "offload_optimizer": {
-        "device": "cpu"
-      },
-      "contiguous_gradients": true,
-      "overlap_comm": true
-    },
-    "bf16": {
-      "enabled": "auto"
-    },
-    "fp16": {
-      "enabled": "auto",
-      "auto_cast": false,
-      "loss_scale": 0,
-      "initial_scale_power": 32,
-      "loss_scale_window": 1000,
-      "hysteresis": 2,
-      "min_loss_scale": 1
-    },
-    "optimizer": {
-      "type": "AdamW",
-      "params": {
-        "lr": "auto",
-        "betas": [
-          0.9,
-          0.999
-        ],
-        "eps": 1e-8,
-        "weight_decay": "auto"
-      }
-    },
-    "scheduler": {
-      "type": "WarmupDecayLR",
-      "params": {
-        "warmup_min_lr": "auto",
-        "warmup_max_lr": "auto",
-        "warmup_num_steps": "auto",
-        "total_num_steps": "auto"
-      }
-    },
-    "train_batch_size": "auto",
-    "train_micro_batch_size_per_gpu": "auto",
-    "wall_clock_breakdown": false
-}

deepspeed_configs/zero1.json

@@ -0,0 +1,23 @@
+{
+  "zero_optimization": {
+    "stage": 1,
+    "overlap_comm": true
+  },
+  "bf16": {
+    "enabled": "auto"
+  },
+  "fp16": {
+    "enabled": "auto",
+    "auto_cast": false,
+    "loss_scale": 0,
+    "initial_scale_power": 32,
+    "loss_scale_window": 1000,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

deepspeed_configs/zero2.json

@@ -0,0 +1,27 @@
+{
+  "zero_optimization": {
+    "stage": 2,
+    "offload_optimizer": {
+      "device": "cpu"
+    },
+    "contiguous_gradients": true,
+    "overlap_comm": true
+  },
+  "bf16": {
+    "enabled": "auto"
+  },
+  "fp16": {
+    "enabled": "auto",
+    "auto_cast": false,
+    "loss_scale": 0,
+    "initial_scale_power": 32,
+    "loss_scale_window": 1000,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

deepspeed_configs/zero3.json

@@ -0,0 +1,31 @@
+{
+  "zero_optimization": {
+    "stage": 3,
+    "overlap_comm": true,
+    "contiguous_gradients": true,
+    "sub_group_size": 0,
+    "reduce_bucket_size": "auto",
+    "stage3_prefetch_bucket_size": "auto",
+    "stage3_param_persistence_threshold": "auto",
+    "stage3_max_live_parameters": 0,
+    "stage3_max_reuse_distance": 0,
+    "stage3_gather_16bit_weights_on_model_save": true
+  },
+  "bf16": {
+    "enabled": "auto"
+  },
+  "fp16": {
+    "enabled": "auto",
+    "auto_cast": false,
+    "loss_scale": 0,
+    "initial_scale_power": 32,
+    "loss_scale_window": 1000,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

deepspeed_configs/zero3_bf16.json

@@ -0,0 +1,31 @@
+{
+  "zero_optimization": {
+    "stage": 3,
+    "overlap_comm": true,
+    "contiguous_gradients": true,
+    "sub_group_size": 0,
+    "reduce_bucket_size": "auto",
+    "stage3_prefetch_bucket_size": "auto",
+    "stage3_param_persistence_threshold": "auto",
+    "stage3_max_live_parameters": 0,
+    "stage3_max_reuse_distance": 0,
+    "stage3_gather_16bit_weights_on_model_save": true
+  },
+  "bf16": {
+    "enabled": true
+  },
+  "fp16": {
+    "enabled": "auto",
+    "auto_cast": false,
+    "loss_scale": 0,
+    "initial_scale_power": 32,
+    "loss_scale_window": 1000,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

deepspeed_configs/zero3_bf16_cpuoffload_all.json

@@ -1,4 +1,6 @@
 {
+  "zero_force_ds_cpu_optimizer": false,
+  "zero_allow_untested_optimizer": true,
   "zero_optimization": {
     "stage": 3,
     "offload_optimizer": {
@@ -20,7 +22,7 @@
     "stage3_gather_16bit_weights_on_model_save": true
   },
   "bf16": {
-    "enabled": "auto"
+    "enabled": true
   },
   "fp16": {
     "enabled": "auto",
@@ -31,26 +33,8 @@
     "hysteresis": 2,
     "min_loss_scale": 1
   },
-  "optimizer": {
-    "type": "AdamW",
-    "params": {
-      "lr": "auto",
-      "betas": [
-        0.9,
-        0.95
-      ],
-      "eps": 1e-8,
-      "weight_decay": "auto"
-    }
-  },
-  "scheduler": {
-    "type": "WarmupLR",
-    "params": {
-      "warmup_min_lr": "auto",
-      "warmup_max_lr": "auto",
-      "warmup_num_steps": "auto"
-    }
-  },
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
   "train_batch_size": "auto",
   "train_micro_batch_size_per_gpu": "auto",
   "wall_clock_breakdown": false

deepspeed_configs/zero3_bf16_cpuoffload_params.json

@@ -0,0 +1,37 @@
+{
+  "zero_force_ds_cpu_optimizer": false,
+  "zero_allow_untested_optimizer": true,
+  "zero_optimization": {
+    "stage": 3,
+    "offload_param": {
+      "device": "cpu",
+      "pin_memory": true
+    },
+    "overlap_comm": true,
+    "contiguous_gradients": true,
+    "sub_group_size": 0,
+    "reduce_bucket_size": "auto",
+    "stage3_prefetch_bucket_size": "auto",
+    "stage3_param_persistence_threshold": "auto",
+    "stage3_max_live_parameters": 0,
+    "stage3_max_reuse_distance": 0,
+    "stage3_gather_16bit_weights_on_model_save": true
+  },
+  "bf16": {
+    "enabled": true
+  },
+  "fp16": {
+    "enabled": "auto",
+    "auto_cast": false,
+    "loss_scale": 0,
+    "initial_scale_power": 32,
+    "loss_scale_window": 1000,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

devtools/README.md

@@ -0,0 +1 @@
+This directory contains example config files that might be useful for debugging. Please see [docs/debugging.qmd](../docs/debugging.qmd) for more information.

devtools/dev_sharegpt.yml

@@ -0,0 +1,48 @@
+# Example config for debugging the sharegpt prompt format
+base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+load_in_8bit: true
+load_in_4bit: false
+
+datasets:
+  - path: philschmid/guanaco-sharegpt-style
+    type: sharegpt
+    shards: 10
+val_set_size: 0
+output_dir: temp_debug/axolotl_outputs/model
+dataset_prepared_path: temp_debug/axolotl_outputs/data
+dataset_processes: 1
+
+sequence_len: 4096
+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
+lora_fan_in_fan_out:
+
+micro_batch_size: 1
+num_epochs: 1
+max_steps: 10
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+train_on_inputs: false
+group_by_length: false
+bf16: false
+fp16: true
+tf32: false
+
+gradient_checkpointing: true
+logging_steps: 1
+flash_attention: true
+
+warmup_steps: 10
+weight_decay: 0.0

docker-compose.yaml

@@ -9,6 +9,11 @@ services:
       - ~/.cache/huggingface/:/root/.cache/huggingface/
     # set environment variables
     environment:
+      # Set environment variables
+      - GIT_AUTHOR_NAME=${GIT_AUTHOR_NAME}
+      - GIT_AUTHOR_EMAIL=${GIT_AUTHOR_EMAIL}
+      - GIT_COMMITTER_NAME=${GIT_COMMITTER_NAME}
+      - GIT_COMMITTER_EMAIL=${GIT_COMMITTER_EMAIL}
       - WANDB_API_KEY=${WANDB_API_KEY}
     deploy:
       resources:

docker/Dockerfile

@@ -3,27 +3,35 @@ FROM winglian/axolotl-base:$BASE_TAG
 
 ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
 ARG AXOLOTL_EXTRAS=""
+ARG AXOLOTL_ARGS=""
 ARG CUDA="118"
 ENV BNB_CUDA_VERSION=$CUDA
+ARG PYTORCH_VERSION="2.1.2"
+
+ENV PYTORCH_VERSION=$PYTORCH_VERSION
 
 RUN apt-get update && \
-    apt-get install -y vim curl
+    apt-get install -y --allow-change-held-packages vim curl nano libnccl2 libnccl-dev rsync s3fs
 
 WORKDIR /workspace
 
-RUN pip3 install "peft @ git+https://github.com/huggingface/peft.git@main"
 RUN git clone --depth=1 https://github.com/OpenAccess-AI-Collective/axolotl.git
+
+WORKDIR /workspace/axolotl
+
 # If AXOLOTL_EXTRAS is set, append it in brackets
-RUN cd axolotl && \
-    if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install -e .[flash-attn,$AXOLOTL_EXTRAS]; \
+RUN pip install causal_conv1d
+RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
     else \
-        pip install -e .[flash-attn]; \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore] $AXOLOTL_ARGS; \
     fi
 
+# So we can test the Docker image
+RUN pip install pytest
+
 # fix so that git fetch/pull from remote works
-RUN cd axolotl && \
-    git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \
+RUN git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \
     git config --get remote.origin.fetch
 
 # helper for huggingface-login cli

docker/Dockerfile-base

@@ -7,22 +7,22 @@ FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION a
 
 ENV PATH="/root/miniconda3/bin:${PATH}"
 
-ARG PYTHON_VERSION="3.9"
-ARG PYTORCH_VERSION="2.0.1"
+ARG PYTHON_VERSION="3.10"
+ARG PYTORCH_VERSION="2.1.2"
 ARG CUDA="118"
+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
-RUN apt-get install -y wget git build-essential ninja-build git-lfs libaio-dev && rm -rf /var/lib/apt/lists/*
-
-RUN wget \
+RUN apt-get update \
+    && apt-get install -y wget git build-essential ninja-build git-lfs libaio-dev && 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
-
-RUN conda create -n "py${PYTHON_VERSION}" python="${PYTHON_VERSION}"
+    && 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}"
 
@@ -31,49 +31,7 @@ WORKDIR /workspace
 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
 
-FROM base-builder AS deepspeed-builder
-
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
-
-WORKDIR /workspace
-
-RUN git clone https://github.com/microsoft/DeepSpeed.git && \
-    cd DeepSpeed && \
-    MAX_CONCURRENCY=8 DS_BUILD_SPARSE_ATTN=0 DS_BUILD_OPS=1 python3 setup.py bdist_wheel
-
-FROM base-builder AS bnb-builder
-
-WORKDIR /workspace
-ARG CUDA="118"
-ENV CUDA=$CUDA
-
-RUN git clone https://github.com/TimDettmers/bitsandbytes.git && \
-    cd bitsandbytes && \
-    CUDA_VERSION=$CUDA make cuda11x && \
-    python setup.py bdist_wheel
-
-FROM base-builder
-
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
-ENV TORCH_CUDA_ARCH_LIST=$TORCH_CUDA_ARCH_LIST
-
-# recompile apex
-RUN python3 -m pip uninstall -y apex
-RUN git clone https://github.com/NVIDIA/apex
-#  `MAX_JOBS=1` disables parallel building to avoid cpu memory OOM when building image on GitHub Action (standard) runners
-RUN cd apex && MAX_JOBS=1 python3 -m pip install -v --disable-pip-version-check --no-cache-dir --no-build-isolation --config-settings "--build-option=--cpp_ext" --config-settings "--build-option=--cuda_ext" ./
-
-RUN mkdir -p /workspace/builds
-COPY --from=bnb-builder /workspace/bitsandbytes /workspace/builds/bitsandbytes
-
-RUN mkdir -p /workspace/wheels/bitsandbytes
-COPY --from=deepspeed-builder /workspace/DeepSpeed/dist/deepspeed-*.whl wheels
-COPY --from=bnb-builder /workspace/bitsandbytes/dist/bitsandbytes-*.whl wheels
-COPY --from=bnb-builder /workspace/bitsandbytes/bitsandbytes/libbitsandbytes*.so wheels/bitsandbytes
-
-RUN pip3 install wheels/deepspeed-*.whl
-RUN cd /workspace/builds/bitsandbytes && python3 setup.py install
-RUN git lfs install --skip-repo
-RUN pip3 install awscli && \
+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

docker/Dockerfile-cloud

@@ -4,15 +4,24 @@ FROM winglian/axolotl:$BASE_TAG
 ENV HF_DATASETS_CACHE="/workspace/data/huggingface-cache/datasets"
 ENV HUGGINGFACE_HUB_CACHE="/workspace/data/huggingface-cache/hub"
 ENV TRANSFORMERS_CACHE="/workspace/data/huggingface-cache/hub"
+ENV HF_HOME="/workspace/data/huggingface-cache/hub"
+ENV HF_HUB_ENABLE_HF_TRANSFER="1"
 
-COPY scripts/runpod-entrypoint.sh /root/runpod-entrypoint.sh
+EXPOSE 8888
+EXPOSE 22
 
+COPY scripts/cloud-entrypoint.sh /root/cloud-entrypoint.sh
+COPY scripts/motd /etc/motd
+
+RUN pip install jupyterlab notebook ipywidgets && \
+    jupyter lab clean
 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 && \
-    chmod +x /workspace/axolotl/scripts/runpod-entrypoint.sh && \
-    chmod +x /root/runpod-entrypoint.sh
+    printf "[ ! -z \"\$TERM\" -a -r /etc/motd ] && cat /etc/motd\n" >> ~/.bashrc && \
+    chmod +x /workspace/axolotl/scripts/cloud-entrypoint.sh && \
+    chmod +x /root/cloud-entrypoint.sh
 
-ENTRYPOINT ["/root/runpod-entrypoint.sh"]
+ENTRYPOINT ["/root/cloud-entrypoint.sh"]
 CMD ["sleep", "infinity"]

docker/Dockerfile-cloud-no-tmux

@@ -0,0 +1,27 @@
+ARG BASE_TAG=main
+FROM winglian/axolotl:$BASE_TAG
+
+ENV HF_DATASETS_CACHE="/workspace/data/huggingface-cache/datasets"
+ENV HUGGINGFACE_HUB_CACHE="/workspace/data/huggingface-cache/hub"
+ENV TRANSFORMERS_CACHE="/workspace/data/huggingface-cache/hub"
+ENV HF_HOME="/workspace/data/huggingface-cache/hub"
+ENV HF_HUB_ENABLE_HF_TRANSFER="1"
+
+EXPOSE 8888
+EXPOSE 22
+
+COPY scripts/cloud-entrypoint-term.sh /root/cloud-entrypoint.sh
+COPY scripts/motd /etc/motd
+
+RUN pip install jupyterlab notebook ipywidgets && \
+    jupyter lab clean
+RUN apt install --yes --no-install-recommends openssh-server tmux sudo && \
+    pip3 install -U --no-cache-dir grpcio ray[default]==2.9.3 && \
+    mkdir -p ~/.ssh && \
+    chmod 700 ~/.ssh && \
+    printf "[ ! -z \"\$TERM\" -a -r /etc/motd ] && cat /etc/motd\n" >> ~/.bashrc && \
+    chmod +x /workspace/axolotl/scripts/cloud-entrypoint.sh && \
+    chmod +x /root/cloud-entrypoint.sh
+
+ENTRYPOINT ["/root/cloud-entrypoint.sh"]
+CMD ["sleep", "infinity"]

docker/Dockerfile-tests

@@ -0,0 +1,41 @@
+ARG BASE_TAG=main-base
+FROM winglian/axolotl-base:$BASE_TAG
+
+ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
+ARG AXOLOTL_EXTRAS=""
+ARG AXOLOTL_ARGS=""
+ARG CUDA="118"
+ENV BNB_CUDA_VERSION=$CUDA
+ARG PYTORCH_VERSION="2.1.2"
+ARG GITHUB_REF="main"
+
+ENV PYTORCH_VERSION=$PYTORCH_VERSION
+
+RUN apt-get update && \
+    apt-get install -y --allow-change-held-packages vim curl nano libnccl2 libnccl-dev
+
+WORKDIR /workspace
+
+RUN git clone --depth=1 https://github.com/OpenAccess-AI-Collective/axolotl.git
+
+WORKDIR /workspace/axolotl
+
+RUN git fetch origin +$GITHUB_REF && \
+    git checkout FETCH_HEAD
+
+# If AXOLOTL_EXTRAS is set, append it in brackets
+RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+    else \
+        pip install -e .[deepspeed,flash-attn,mamba-ssm] $AXOLOTL_ARGS; \
+    fi
+
+# So we can test the Docker image
+RUN pip install pytest
+
+# fix so that git fetch/pull from remote works
+RUN git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \
+    git config --get remote.origin.fetch
+
+# helper for huggingface-login cli
+RUN git config --global credential.helper store

docs/.gitignore

@@ -0,0 +1,2 @@
+/.quarto/
+_site/

docs/batch_vs_grad.qmd

@@ -0,0 +1,59 @@
+---
+title: Batch size vs Gradient accumulation
+description: Understanding of batch size and gradient accumulation steps
+---
+
+Gradient accumulation means accumulating gradients over several mini-batches and updating the model weights afterward. When the samples in each batch are diverse, this technique doesn't significantly impact learning.
+
+This method allows for effective training with larger effective batch sizes without needing proportionally larger memory. Here's why:
+
+1. **Memory Consumption with Batch Size**: The primary reason increasing the batch size impacts memory is due to the storage requirements for intermediate activations. When you forward propagate a batch through a network, you have to store the activations at each layer for each sample in the batch, because these activations are used during backpropagation to compute gradients. Therefore, larger batches mean more activations, leading to greater GPU memory consumption.
+
+2. **Gradient Accumulation**: With gradient accumulation, you're effectively simulating a larger batch size by accumulating gradients over several smaller batches (or micro-batches). However, at any given time, you're only forward and backward propagating a micro-batch. This means you only store activations for the micro-batch, not the full accumulated batch. As a result, you can simulate the effect of a larger batch size without the memory cost of storing activations for a large batch.
+
+**Example 1:**
+Micro batch size: 3
+Gradient accumulation steps: 2
+Number of GPUs: 3
+Total batch size = 3 * 2 * 3 = 18
+
+```
+| GPU 1          | GPU 2          | GPU 3          |
+|----------------|----------------|----------------|
+| S1, S2, S3     | S4, S5, S6     | S7, S8, S9     |
+| e1, e2, e3     | e4, e5, e6     | e7, e8, e9     |
+|----------------|----------------|----------------|
+| → (accumulate) | → (accumulate) | → (accumulate) |
+|----------------|----------------|----------------|
+| S10, S11, S12  | S13, S14, S15  | S16, S17, S18  |
+| e10, e11, e12  | e13, e14, e15  | e16, e17, e18  |
+|----------------|----------------|----------------|
+| → (apply)      | → (apply)      | → (apply)      |
+
+Accumulated gradient for the weight w1 after the second iteration (considering all GPUs):
+Total gradient for w1 = e1 + e2 + e3 + e4 + e5 + e6 + e7 + e8 + e9 + e10 + e11 + e12 + e13 + e14 + e15 + e16 + e17 + e18
+
+Weight update for w1:
+w1_new = w1_old - learning rate x (Total gradient for w1 / 18)
+```
+
+**Example 2:**
+Micro batch size: 2
+Gradient accumulation steps: 1
+Number of GPUs: 3
+Total batch size = 2 * 1 * 3 = 6
+
+```
+| GPU 1     | GPU 2     | GPU 3     |
+|-----------|-----------|-----------|
+| S1, S2    | S3, S4    | S5, S6    |
+| e1, e2    | e3, e4    | e5, e6    |
+|-----------|-----------|-----------|
+| → (apply) | → (apply) | → (apply) |
+
+Accumulated gradient for the weight w1 (considering all GPUs):
+Total gradient for w1 = e1 + e2 + e3 + e4 + e5 + e6
+
+Weight update for w1:
+w1_new = w1_old - learning rate × (Total gradient for w1 / 6)
+```

docs/config.qmd

@@ -0,0 +1,458 @@
+---
+title: Config options
+description: A complete list of all configuration options.
+---
+
+```yaml
+# 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
+revision_of_model:
+# Optional tokenizer configuration path 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:
+
+# (Internal use only)
+# Used to identify which the model is based on
+is_falcon_derived_model:
+is_llama_derived_model:
+is_qwen_derived_model:
+# Please note that if you set this to true, `padding_side` will be set to "left" by default
+is_mistral_derived_model:
+
+# optional overrides to the base model configuration
+overrides_of_model_config:
+  # RoPE Scaling https://github.com/huggingface/transformers/pull/24653
+  rope_scaling:
+    type: # linear | dynamic
+    factor: # float
+
+# optional overrides to the bnb 4bit quantization configuration
+# https://huggingface.co/docs/transformers/main/main_classes/quantization#transformers.BitsAndBytesConfig
+bnb_config_kwargs:
+  # These are default values
+  llm_int8_has_fp16_weight: false
+  bnb_4bit_quant_type: nf4
+  bnb_4bit_use_double_quant: true
+
+
+# Whether you are training a 4-bit GPTQ quantized model
+gptq: true
+
+# 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
+
+# Limit the memory for all available GPUs to this amount (if an integer, expressed in gigabytes); default: unset
+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
+
+# 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.
+    # Add additional keys from your dataset as input or output roles
+    roles:
+      input: # Optional[List[str]]. These will be masked based on train_on_input
+      output: # Optional[List[str]].
+
+  # Custom user instruction prompt
+  - path: repo
+    type:
+      # The below are defaults. only set what's needed if you use a different column name.
+      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
+      # Use {instruction}/{input} as key to be replaced
+      # '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:
+
+# If false, the datasets will not be shuffled and will keep their original order in `datasets`.
+# The same applies to the `test_datasets` option and the `pretraining_dataset` option. Default is true.
+shuffle_merged_datasets: true
+
+# A list of one or more datasets to eval the model with.
+# You can use either test_datasets, or val_set_size, but not both.
+test_datasets:
+  - path: /workspace/data/eval.jsonl
+    ds_type: json
+    # You need to specify a split. For "json" datasets the default split is called "train".
+    split: train
+    type: completion
+    data_files:
+      - /workspace/data/eval.jsonl
+
+# use RL training: 'dpo', 'ipo', 'kto_pair'
+rl:
+
+# Saves the desired chat template to the tokenizer_config.json for easier inferencing
+# Currently supports chatml and inst (mistral/mixtral)
+chat_template: chatml
+# 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: # 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
+# Keep dataset in memory while preprocessing
+# Only needed if cached dataset is taking too much storage
+dataset_keep_in_memory:
+# push checkpoints to hub
+hub_model_id: # private 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:
+# 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:
+# Increasing the following values helps with packing, but usually only slightly (<%1.)
+# The number of samples packed at a time.
+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
+
+# Passed through to transformers when loading the model when launched without accelerate
+# Use `sequential` when training w/ model parallelism to limit memory
+device_map:
+# Defines the max memory usage per gpu on the system. Passed through to transformers when loading the model.
+max_memory:
+
+# 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 `output_dir`.
+# Note that if you merge an adapter to the base model, a new subdirectory `merged` will be created under the `output_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 modules
+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.
+# `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
+
+lora_fan_in_fan_out: false
+
+# LoRA+ hyperparameters
+# For more details about the following options, see:
+# https://arxiv.org/abs/2402.12354  and `src/axolotl/core/train_builder.py`
+loraplus_lr_ratio: # loraplus learning rate ratio lr_B / lr_A. Recommended value is 2^4.
+loraplus_lr_embedding: #  loraplus learning rate for lora embedding layers. Default value is 1e-6.
+
+peft:
+  # Configuration options for loftq initialization for LoRA
+  # https://huggingface.co/docs/peft/developer_guides/quantization#loftq-initialization
+  loftq_config:
+    loftq_bits:  # typically 4 bits
+
+# 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_anneal_steps: # Number of anneal steps for each relora cycle
+relora_prune_ratio: # threshold for optimizer magnitude when pruning
+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
+# Make sure your `WANDB_API_KEY` environment variable is set (recommended) or you login to wandb with `wandb login`.
+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_name: # Set the name of your wandb run
+wandb_run_id: # Set the ID 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
+
+# mlflow configuration if you're using it
+mlflow_tracking_uri: # URI to mlflow
+mlflow_experiment_name: # Your experiment name
+hf_mlflow_log_artifacts:  # set to true to copy each saved checkpoint on each save to mlflow artifact registry
+
+# 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.
+# Batch size per gpu = micro_batch_size * gradient_accumulation_steps
+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:
+eval_steps: # Leave empty to eval at each epoch, integers for every N steps. decimal for fraction of total steps
+evals_per_epoch: # number of times per epoch to run evals, mutually exclusive with eval_steps
+save_strategy: # Set to `"no"` to skip checkpoint saves
+save_steps: # Leave empty to save at each epoch
+saves_per_epoch: # number of times per epoch to save a checkpoint, mutually exclusive with save_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_max_new_tokens: # Total number of tokens generated for predictions sent to wandb. Default is 128
+eval_causal_lm_metrics: # HF evaluate metrics used during evaluation. Default is ["sacrebleu", "comet", "ter", chrf]
+
+loss_watchdog_threshold: # High loss value, indicating the learning has broken down (a good estimate is ~2 times the loss at the start of training)
+loss_watchdog_patience: # Number of high-loss steps in a row before the trainer aborts (default: 3)
+
+# 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
+# additional kwargs to pass to the trainer for gradient checkpointing
+# gradient_checkpointing_kwargs:
+#   use_reentrant: true
+
+# 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:
+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)
+
+# For one_cycle optim
+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/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
+# - galore_adamw
+# - galore_adamw_8bit
+# - galore_adafactor
+# - galore_adamw_layerwise
+# - galore_adamw_8bit_layerwise
+# - galore_adafactor_layerwise
+optimizer:
+# Dictionary of arguments to pass to the optimizer
+optim_args:
+# For Galore Optimizers the following optim_args are available
+# rank:  # type: int
+# update_proj_gap  # type: int
+# scale  # type: float
+# proj_type:  # type: str, default = std
+
+# The target modules to optimize, i.e. the module names that you would like to train, right now this is used only for GaLore algorithm
+optim_target_modules:
+# - self_attn  # for llama
+# - mlp
+
+# 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
+neftune_noise_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:
+# Shifted-sparse attention (only llama) - https://arxiv.org/pdf/2309.12307.pdf
+s2_attention:
+# 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>"
+  # pad_token: "[PAD]"
+
+# Add extra tokens.
+tokens:
+
+# FSDP
+fsdp:
+fsdp_config:
+
+# Deepspeed config path. e.g., deepspeed_configs/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:
+```

docs/dataset-formats/conversation.qmd

@@ -0,0 +1,63 @@
+---
+title: Conversation
+description: Conversation format for supervised fine-tuning.
+order: 3
+---
+
+## sharegpt
+
+conversations where `from` is `human`/`gpt`. (optional: first row with role `system` to override default system prompt)
+
+```{.json filename="data.jsonl"}
+{"conversations": [{"from": "...", "value": "..."}]}
+```
+
+Note: `type: sharegpt` opens special configs:
+- `conversation`: enables conversions to many Conversation types. Refer to the 'name' [here](https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py) for options.
+- `roles`: allows you to specify the roles for input and output. This is useful for datasets with custom roles such as `tool` etc to support masking.
+- `field_human`: specify the key to use instead of `human` in the conversation.
+- `field_model`: specify the key to use instead of `gpt` in the conversation.
+
+```yaml
+datasets:
+    path: ...
+    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.
+    # Add additional keys from your dataset as input or output roles
+    roles:
+      input: # Optional[List[str]]. These will be masked based on train_on_input
+      output: # Optional[List[str]].
+```
+
+## pygmalion
+
+```{.json filename="data.jsonl"}
+{"conversations": [{"role": "...", "value": "..."}]}
+```
+
+## sharegpt.load_role
+
+conversations where `role` is used instead of `from`
+
+```{.json filename="data.jsonl"}
+{"conversations": [{"role": "...", "value": "..."}]}
+```
+
+## sharegpt.load_guanaco
+
+conversations where `from` is `prompter` `assistant` instead of default sharegpt
+
+```{.json filename="data.jsonl"}
+{"conversations": [{"from": "...", "value": "..."}]}
+```
+
+## sharegpt_jokes
+
+creates a chat where bot is asked to tell a joke, then explain why the joke is funny
+
+```{.json filename="data.jsonl"}
+{"conversations": [{"title": "...", "text": "...", "explanation": "..."}]}
+```

docs/dataset-formats/index.qmd

@@ -0,0 +1,14 @@
+---
+title: Dataset Formats
+description: Supported dataset formats.
+listing:
+  fields: [title, description]
+  type: table
+  sort-ui: false
+  filter-ui: false
+  max-description-length: 250
+---
+
+Axolotl supports a variety of dataset formats.  It is recommended to use a JSONL format.  The schema of the JSONL depends upon the task and the prompt template you wish to use. Instead of a JSONL, you can also use a HuggingFace dataset with columns for each JSONL field.
+
+Below are these various formats organized by task:

docs/dataset-formats/inst_tune.qmd

@@ -0,0 +1,189 @@
+---
+title: Instruction Tuning
+description: Instruction tuning formats for supervised fine-tuning.
+order: 2
+---
+
+## alpaca
+
+instruction; input(optional)
+
+```{.json filename="data.jsonl"}
+{"instruction": "...", "input": "...", "output": "..."}
+```
+
+## jeopardy
+
+question and answer
+
+```{.json filename="data.jsonl"}
+{"question": "...", "category": "...", "answer": "..."}
+```
+
+## oasst
+
+instruction
+
+```{.json filename="data.jsonl"}
+{"INSTRUCTION": "...", "RESPONSE": "..."}
+```
+
+## gpteacher
+
+instruction; input(optional)
+
+```{.json filename="data.jsonl"}
+{"instruction": "...", "input": "...", "response": "..."}
+```
+
+## reflection
+
+instruction with reflect; input(optional)
+
+```{.json filename="data.jsonl"}
+{"instruction": "...", "input": "...", "output": "...", "reflection": "...", "corrected": "..."}
+```
+
+## explainchoice
+
+question, choices, (solution OR explanation)
+
+```{.json filename="data.jsonl"}
+{"question": "...", "choices": ["..."], "solution": "...", "explanation": "..."}
+```
+
+## concisechoice
+
+question, choices, (solution OR explanation)
+
+```{.json filename="data.jsonl"}
+{"question": "...", "choices": ["..."], "solution": "...", "explanation": "..."}
+```
+
+## summarizetldr
+
+article and summary
+
+```{.json filename="data.jsonl"}
+{"article": "...", "summary": "..."}
+```
+
+## alpaca_chat
+
+basic instruct for alpaca chat
+
+```{.json filename="data.jsonl"}
+{"instruction": "...", "input": "...", "response": "..."}
+```
+
+## alpaca_chat.load_qa
+
+question and answer for alpaca chat
+
+```{.json filename="data.jsonl"}
+{"question": "...", "answer": "..."}
+```
+
+## alpaca_chat.load_concise
+
+question and answer for alpaca chat, for concise answers
+
+```{.json filename="data.jsonl"}
+{"instruction": "...", "input": "...", "response": "..."}
+```
+
+## alpaca_chat.load_camel_ai
+
+question and answer for alpaca chat, for load_camel_ai
+
+```{.json filename="data.jsonl"}
+{"message_1": "...", "message_2": "..."}
+```
+
+## alpaca_w_system.load_open_orca
+
+support for open orca datasets with included system prompts, instruct
+
+```{.json filename="data.jsonl"}
+{"system_prompt": "...", "question": "...", "response": "..."}
+```
+
+## context_qa
+
+in context question answering from an article
+
+```{.json filename="data.jsonl"}
+{"article": "...", "question": "...", "answer": "..."}
+```
+
+## context_qa.load_v2
+
+in context question answering (alternate)
+
+```{.json filename="data.jsonl"}
+{"context": "...", "question": "...", "answer": "..."}
+```
+
+## context_qa.load_404
+
+in context question answering from an article, with default response for no answer from context
+
+```{.json filename="data.jsonl"}
+{"article": "...", "unanswerable_question": "..."}
+```
+
+## creative_acr.load_answer
+
+instruction and revision
+
+```{.json filename="data.jsonl"}
+{"instruction": "...", "revision": "..."}
+```
+
+## creative_acr.load_critique
+
+critique
+
+```{.json filename="data.jsonl"}
+{"scores": "...", "critiques": "...", "instruction": "...", "answer": "..."}
+```
+
+## creative_acr.load_revise
+
+critique and revise
+
+```{.json filename="data.jsonl"}
+{"scores": "...", "critiques": "...", "instruction": "...", "answer": "...", "revision": "..."}
+```
+
+## metharme
+
+instruction, adds additional eos tokens
+
+```{.json filename="data.jsonl"}
+{"prompt": "...", "generation": "..."}
+```
+
+## How to add custom prompt format
+
+For a dataset that is preprocessed for instruction purposes:
+
+```{.json filename="data.jsonl"}
+{"input": "...", "output": "..."}
+```
+
+You can use this example in your YAML config:
+
+```{.yaml filename="config.yaml"}
+datasets:
+  - path: repo
+    type:
+      system_prompt: ""
+      field_system: system
+      field_instruction: input
+      field_output: output
+      format: "[INST] {instruction} [/INST]"
+      no_input_format: "[INST] {instruction} [/INST]"
+```
+
+See full config options under [here](../config.qmd).

docs/dataset-formats/pretraining.qmd

@@ -0,0 +1,26 @@
+---
+title: Pre-training
+description: Data format for a pre-training completion task.
+order: 1
+---
+
+For pretraining, there is no prompt template or roles.  The only required field is `text`:
+
+```{.json filename="data.jsonl"}
+{"text": "first row"}
+{"text": "second row"}
+...
+```
+
+:::{.callout-note}
+
+### Streaming is recommended for large datasets
+
+Axolotl usually loads the entire dataset into memory. This will be challenging for large datasets. Use the following config to enable streaming:
+
+```{.yaml filename="config.yaml"}
+pretraining_dataset: # hf path only
+...
+```
+
+:::

docs/dataset-formats/template_free.qmd

@@ -0,0 +1,7 @@
+---
+title: Template-Free
+description: Construct prompts without a template.
+order: 4
+---
+
+See [these docs](../input_output.qmd).

docs/dataset-formats/tokenized.qmd

@@ -0,0 +1,12 @@
+---
+title: Custom Pre-Tokenized Dataset
+description: How to use a custom pre-tokenized dataset.
+order: 5
+---
+
+- Do not pass a `type:` in your axolotl config.
+- Columns in Dataset must be exactly `input_ids`, `attention_mask`, `labels`
+
+```{.yaml filename="config.yml"}
+- path: ...
+```

docs/dataset_preprocessing.qmd

@@ -0,0 +1,35 @@
+---
+title: Dataset Preprocessing
+description: How datasets are processed
+---
+
+Dataset pre-processing is the step where Axolotl takes each dataset you've configured alongside
+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
+ - shuffle and merge multiple datasets together if using more than one
+
+The processing of the datasets can happen one of two ways:
+
+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
+(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.
+
+The path of the cache is controlled by `dataset_prepared_path:` and is often left blank in example
+YAMLs as this leads to a more robust solution that prevents unexpectedly reusing cached data.
+
+If `dataset_prepared_path:` is left empty, when training, the processed dataset will be cached in a
+default path of `./last_run_prepared/`, but will ignore anything already cached there. By explicitly
+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
+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
+and change your prompt templating logic, it may not pick up the changes you made and you will be
+training over the old prompt.

docs/debugging.qmd

@@ -0,0 +1,245 @@
+---
+title: Debugging
+description: How to debug Axolotl
+---
+
+
+This document provides some tips and tricks for debugging Axolotl.  It also provides an example configuration for debugging with VSCode.  A good debugging setup is essential to understanding how Axolotl code works behind the scenes.
+
+## Table of Contents
+
+- [General Tips](#general-tips)
+- [Debugging with VSCode](#debugging-with-vscode)
+    - [Background](#background)
+    - [Configuration](#configuration)
+    - [Customizing your debugger](#customizing-your-debugger)
+    - [Video Tutorial](#video-tutorial)
+- [Debugging With Docker](#debugging-with-docker)
+    - [Setup](#setup)
+    - [Attach To Container](#attach-to-container)
+    - [Video - Attaching To Docker On Remote Host](#video---attaching-to-docker-on-remote-host)
+
+## General Tips
+
+While debugging it's helpful to simplify your test scenario as much as possible.  Here are some tips for doing so:
+
+> [!Important]
+> All of these tips are incorporated into the [example configuration](#configuration) for debugging with VSCode below.
+
+1. **Make sure you are using the latest version of axolotl**:  This project changes often and bugs get fixed fast.  Check your git branch and make sure you have pulled the latest changes from `main`.
+1. **Eliminate concurrency**: Restrict the number of processes to 1 for both training and data preprocessing:
+    - 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
+    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`
+    - `max_steps: 1`
+    - `val_set_size: 0`
+5. **Clear Caches:** Axolotl caches certain steps and so does the underlying HuggingFace trainer.  You may want to clear some of these caches when debugging.
+    - Data preprocessing: When debugging data preprocessing, which includes prompt template formation, you may want to delete the directory set in `dataset_prepared_path:` in your axolotl config.  If you didn't set this value, the default is `last_run_prepared`.
+    - HF Hub: If you are debugging data preprocessing, you should clear the relevant HF cache [HuggingFace cache](https://huggingface.co/docs/datasets/cache), by deleting the appropriate `~/.cache/huggingface/datasets/...` folder(s).
+    - **The recommended approach is to redirect all outputs and caches to a temporary folder and delete selected subfolders before each run.  This is demonstrated in the example configuration below.**
+
+
+## Debugging with VSCode
+
+### Background
+
+The below example shows how to configure VSCode to debug data preprocessing of the `sharegpt` format.  This is the format used when you have the following in your axolotl config:
+
+```yaml
+datasets:
+  - path: <path to your sharegpt formatted dataset> # example on HF Hub: philschmid/guanaco-sharegpt-style
+    type: sharegpt
+```
+
+>[!Important]
+> If you are already familiar with advanced VSCode debugging, you can skip the below explanation and look at the files [.vscode/launch.json](../.vscode/launch.json) and [.vscode/tasks.json](../.vscode/tasks.json) for an example configuration.
+
+>[!Tip]
+> If you prefer to watch a video, rather than read, you can skip to the [video tutorial](#video-tutorial) below (but doing both is recommended).
+
+### Setup
+
+Make sure you have an [editable install](https://setuptools.pypa.io/en/latest/userguide/development_mode.html) of Axolotl, which ensures that changes you make to the code are reflected at runtime.  Run the following commands from the root of this project:
+
+```bash
+pip3 install packaging
+pip3 install -e '.[flash-attn,deepspeed]'
+```
+
+#### Remote Hosts
+
+If you developing on a remote host, you can easily use VSCode to debug remotely.  To do so, you will need to follow this [remote - SSH guide](https://code.visualstudio.com/docs/remote/ssh).  You can also see the video below on [Docker and Remote SSH debugging](#video---attaching-to-docker-on-remote-host).
+
+
+### Configuration
+
+The easiest way to get started is to modify the [.vscode/launch.json](../.vscode/launch.json) file in this project.  This is just an example configuration, so you may need to modify or copy it to suit your needs.
+
+For example, to mimic the command `cd devtools && CUDA_VISIBLE_DEVICES=0 accelerate launch -m axolotl.cli.train dev_sharegpt.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.
+
+```jsonc
+// .vscode/launch.json
+{
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Debug axolotl prompt - sharegpt",
+            "type": "python",
+            "module": "accelerate.commands.launch",
+            "request": "launch",
+            "args": [
+                "-m", "axolotl.cli.train", "dev_sharegpt.yml",
+                // The flags below simplify debugging by overriding the axolotl config
+                // with the debugging tips above.  Modify as needed.
+                "--dataset_processes=1",      // limits data preprocessing to one process
+                "--max_steps=1",              // limits training to just one step
+                "--batch_size=1",             // minimizes batch size
+                "--micro_batch_size=1",       // minimizes batch size
+                "--val_set_size=0",           // disables validation
+                "--sample_packing=False",     // disables sample packing which is necessary for small datasets
+                "--eval_sample_packing=False",// disables sample packing on eval set
+                "--dataset_prepared_path=temp_debug/axolotl_outputs/data", // send data outputs to a temp folder
+                "--output_dir=temp_debug/axolotl_outputs/model" // send model outputs to a temp folder
+                ],
+            "console": "integratedTerminal",      // show output in the integrated terminal
+            "cwd": "${workspaceFolder}/devtools", // set working directory to devtools from the root of the project
+            "justMyCode": true,                   // step through only axolotl code
+            "env": {"CUDA_VISIBLE_DEVICES": "0",  // Since we aren't doing distributed training, we need to limit to one GPU
+                    "HF_HOME": "${workspaceFolder}/devtools/temp_debug/.hf-cache"}, // send HF cache to a temp folder
+            "preLaunchTask": "cleanup-for-dataprep", // delete temp folders (see below)
+        }
+    ]
+}
+```
+
+**Additional notes about this configuration:**
+
+- The argument `justMyCode` is set to `true` such that you step through only the axolotl code.  If you want to step into dependencies, set this to `false`.
+- The `preLaunchTask`: `cleanup-for-dataprep` is defined in [.vscode/tasks.json](../.vscode/tasks.json) and is used to delete the following folders before debugging, which is essential to ensure that the data pre-processing code is run from scratch:
+    -  `./devtools/temp_debug/axolotl_outputs`
+    - `./devtools/temp_debug/.hf-cache/datasets`
+
+>[!Tip]
+> You may not want to delete these folders. For example, if you are debugging model training instead of data pre-processing, you may NOT want to delete the cache or output folders. You may also need to add additional tasks to the `tasks.json` file depending on your use case.
+
+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.
+
+```jsonc
+// .vscode/tasks.json
+// this file is used by launch.json
+{
+    "version": "2.0.0",
+    "tasks": [
+      // this task changes into the devtools directory and deletes the temp_debug/axolotl_outputs folder
+      {
+        "label": "delete-outputs",
+        "type": "shell",
+        "command": "rm -rf temp_debug/axolotl_outputs",
+        "options":{ "cwd": "${workspaceFolder}/devtools"},
+        "problemMatcher": []
+      },
+      // this task changes into the devtools directory and deletes the `temp_debug/.hf-cache/datasets` folder
+      {
+        "label": "delete-temp-hf-dataset-cache",
+        "type": "shell",
+        "command": "rm -rf temp_debug/.hf-cache/datasets",
+        "options":{ "cwd": "${workspaceFolder}/devtools"},
+        "problemMatcher": []
+      },
+        // this task combines the two tasks above
+      {
+       "label": "cleanup-for-dataprep",
+       "dependsOn": ["delete-outputs", "delete-temp-hf-dataset-cache"],
+      }
+    ]
+}
+```
+
+### Customizing your debugger
+
+Your debugging use case may differ from the example above.  The easiest thing to do is to put your own axolotl config in the `devtools` folder and modify the `launch.json` file to use your config.  You may also want to modify the `preLaunchTask` to delete different folders or not delete anything at all.
+
+### Video Tutorial
+
+The following video tutorial walks through the above configuration and demonstrates how to debug with VSCode, (click the image below to watch):
+
+<div style="text-align: center; line-height: 0;">
+
+<a href="https://youtu.be/xUUB11yeMmc" target="_blank"
+title="How to debug Axolotl (for fine tuning LLMs)"><img
+src="https://i.ytimg.com/vi/xUUB11yeMmc/maxresdefault.jpg"
+style="border-radius: 10px; display: block; margin: auto;" width="560" height="315" /></a>
+
+<figcaption style="font-size: smaller;"><a href="https://hamel.dev">Hamel Husain's</a> tutorial: <a href="https://www.youtube.com/watch?v=xUUB11yeMmc">Debugging Axolotl w/VSCode</a></figcaption>
+
+</div>
+<br>
+
+## Debugging With Docker
+
+Using [official Axolotl Docker images](https://hub.docker.com/r/winglian/axolotl/tags) is a great way to debug your code, and is a very popular way to use Axolotl.  Attaching VSCode to Docker takes a few more steps.
+
+### Setup
+
+On the host that is running axolotl (ex: if you are using a remote host), clone the axolotl repo and change your current directory to the root:
+
+```bash
+git clone https://github.com/OpenAccess-AI-Collective/axolotl
+cd axolotl
+```
+
+>[!Tip]
+> If you already have axolotl cloned on your host, make sure you have the latest changes and change into the root of the project.
+
+Next, run the desired docker image and mount the current directory. Below is a docker command you can run to do this:[^2]
+
+```bash
+docker run --privileged --gpus '"all"' --shm-size 10g --rm -it --name axolotl --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 --mount type=bind,src="${PWD}",target=/workspace/axolotl -v ${HOME}/.cache/huggingface:/root/.cache/huggingface winglian/axolotl:main-py3.10-cu118-2.0.1
+```
+
+>[!Tip]
+> To understand which containers are available, see the [Docker section of the README](../README.md#docker) and the [DockerHub repo](https://hub.docker.com/r/winglian/axolotl/tags).  For details of how the Docker containers are built, see axolotl's [Docker CI builds](../.github/workflows/main.yml).
+
+You will now be in the container.  Next, perform an editable install of Axolotl:
+
+```bash
+pip3 install packaging
+pip3 install -e '.[flash-attn,deepspeed]'
+```
+
+### Attach To Container
+
+Next, if you are using a remote host, [Remote into this host with VSCode](https://code.visualstudio.com/docs/remote/ssh).  If you are using a local host, you can skip this step.
+
+Next, select `Dev Containers: Attach to Running Container...` using the command palette (`CMD + SHIFT + P`) in VSCode.  You will be prompted to select a container to attach to.  Select the container you just created.  You will now be in the container with a working directory that is at the root of the project.  Any changes you make to the code will be reflected both in the container and on the host.
+
+Now you are ready to debug as described above (see [Debugging with VSCode](#debugging-with-vscode)).
+
+### Video - Attaching To Docker On Remote Host
+
+Here is a short video that demonstrates how to attach to a Docker container on a remote host:
+
+<div style="text-align: center; line-height: 0;">
+
+<a href="https://youtu.be/0AuoR7QnHR0" target="_blank"
+title="Debugging Axolotl Part 2: Attaching to Docker on a Remote Host"><img
+src="https://i.ytimg.com/vi/0AuoR7QnHR0/hqdefault.jpg"
+style="border-radius: 10px; display: block; margin: auto;" width="560" height="315" /></a>
+
+<figcaption style="font-size: smaller;"><a href="https://hamel.dev">Hamel Husain's</a> tutorial: <a href="https://youtu.be/0AuoR7QnHR0">Debugging Axolotl Part 2: Attaching to Docker on a Remote Host
+</a></figcaption>
+
+</div>
+<br>
+
+[^1]: The config actually mimics the command `CUDA_VISIBLE_DEVICES=0 python -m accelerate.commands.launch -m axolotl.cli.train devtools/sharegpt.yml`, but this is the same thing.
+
+[^2]: Many of the below flags are recommended best practices by Nvidia when using nvidia-container-toolkit.  You can read more about these flags [here](https://docs.nvidia.com/deeplearning/frameworks/user-guide/index.html).

docs/faq.qmd

@@ -0,0 +1,21 @@
+---
+title: FAQ
+description: Frequently asked questions
+---
+
+
+**Q: The trainer stopped and hasn't progressed in several minutes.**
+
+> A: Usually an issue with the GPUs communicating with each other. See the [NCCL doc](nccl.qmd)
+
+**Q: Exitcode -9**
+
+> A: This usually happens when you run out of system RAM.
+
+**Q: Exitcode -7 while using deepspeed**
+
+> A: Try upgrading deepspeed w: `pip install -U deepspeed`
+
+**Q: AttributeError: 'DummyOptim' object has no attribute 'step'**
+
+> A: You may be using deepspeed with single gpu. Please don't set `deepspeed:` in yaml or cli.

docs/fsdp_qlora.qmd

@@ -0,0 +1,43 @@
+---
+title: "FDSP + QLoRA"
+description: Use FSDP with QLoRA to fine-tune large LLMs on consumer GPUs.
+format:
+  html:
+    toc: true
+---
+
+## Background
+
+Using FSDP with QLoRA is essential for **fine-tuning larger (70b+ parameter) LLMs on consumer GPUs.**  For example, you can use FSDP + QLoRA to train a 70b model on two 24GB GPUs[^1].
+
+Below, we describe how to use this feature in Axolotl.
+
+## Usage
+
+To enable `QLoRA` with `FSDP`, you need to perform the following steps:
+
+> ![Tip]
+> See the [example config](#example-config) file in addition to reading these instructions.
+
+1. Set `adapter: qlora` in your axolotl config file.
+2. Enable FSDP in your axolotl config, as [described here](https://github.com/OpenAccess-AI-Collective/axolotl?tab=readme-ov-file#fsdp).
+3. Use one of the supported model types: `llama`, `mistral` or `mixtral`.
+
+## Example Config
+
+[examples/llama-2/qlora-fsdp.yml](../examples/llama-2/qlora-fsdp.yml) contains an example of how to enable QLoRA + FSDP in axolotl.
+
+## References
+
+- [PR #1378](https://github.com/OpenAccess-AI-Collective/axolotl/pull/1378) enabling QLoRA in FSDP in Axolotl.
+- [Blog Post](https://www.answer.ai/posts/2024-03-06-fsdp-qlora.html) from the [Answer.AI](https://www.answer.ai/) team describing the work that enabled QLoRA in FSDP.
+- Related HuggingFace PRs Enabling FDSP + QLoRA:
+    - Accelerate [PR#2544](https://github.com/huggingface/accelerate/pull/2544 )
+    - Transformers [PR#29587](https://github.com/huggingface/transformers/pull/29587)
+    - TRL [PR#1416](https://github.com/huggingface/trl/pull/1416)
+    - PEFT [PR#1550](https://github.com/huggingface/peft/pull/1550)
+
+
+
+
+[^1]: This was enabled by [this work](https://www.answer.ai/posts/2024-03-06-fsdp-qlora.html) from the Answer.AI team.

docs/input_output.qmd

@@ -0,0 +1,265 @@
+---
+title: Template-free prompt construction
+description: "Template-free prompt construction with the `input_output` format"
+---
+
+<!-- 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/OpenAccess-AI-Collective/axolotl) is
+setting the following configuration value:
+
+
+```yaml
+train_on_inputs: false
+```
+
+If you declare a [dataset formats](https://github.com/OpenAccess-AI-Collective/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 ingored 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>"
+            }
+        ]
+    }
+:::

docs/mac.qmd

@@ -0,0 +1,22 @@
+---
+title: Mac M-series
+description: Mac M-series support
+---
+
+Currently Axolotl on Mac is partially usable, many of the dependencies of Axolotl including Pytorch do not support MPS or have incomplete support.
+
+Current support:
+
+- [x] Support for all models
+- [x] Full training of models
+- [x] LoRA training
+- [x] Sample packing
+- [ ] FP16 and BF16 (awaiting AMP support for MPS in Pytorch)
+- [ ] Tri-dao's flash-attn (until it is supported use spd_attention as an alternative)
+- [ ] xformers
+- [ ] bitsandbytes (meaning no 4/8 bits loading and bnb optimizers)
+- [ ] qlora
+- [ ] DeepSpeed
+
+Untested:
+- FSDP

docs/multi-node.qmd

@@ -0,0 +1,48 @@
+---
+title: Multi Node
+description: How to use Axolotl on multiple machines
+---
+
+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:
+
+~/.cache/huggingface/accelerate/default_config.yaml
+```yaml
+compute_environment: LOCAL_MACHINE
+debug: false
+distributed_type: FSDP
+downcast_bf16: 'no'
+machine_rank: 0 # Set to 0 for the main machine, increment by one for other machines
+main_process_ip: 10.0.0.4 # Set to main machine's IP
+main_process_port: 5000
+main_training_function: main
+mixed_precision: bf16
+num_machines: 2 # Change to the number of machines
+num_processes: 4 # That's the total number of GPUs, (for example: if you have 2 machines with 4 GPU, put 8)
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false
+```
+
+Configure your model to use FSDP with for example:
+```yaml
+fsdp:
+  - full_shard
+  - auto_wrap
+fsdp_config:
+  fsdp_offload_params: true
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
+```
+
+## Machine configuration
+
+On each machine you need a copy of Axolotl, we suggest using the same commit to ensure compatibility.
+
+You will also need to have the same configuration file for your model on each machine.
+
+On the main machine only, make sure the port you set as `main_process_port` is open in TCP and reachable by other machines.
+
+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.

docs/multipack.qmd

@@ -0,0 +1,76 @@
+---
+title: Multipack (Sample Packing)
+description: Multipack is a technique to pack multiple sequences into a single batch to increase training throughput.
+---
+
+## Visualization of Multipack with Flash Attention
+
+Because Flash Attention simply drops the attention mask, we do not need to
+construct a 4d attention mask. We only need to concatenate the sequences into
+a single batch and let flash attention know where each new sequence begins.
+
+
+4k context, bsz =4,
+each character represents 256 tokens
+X represents a padding token
+
+```
+   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
+[[ A A A A A A A A A A A ]
+   B B B B B B ]
+   C C C C C C C ]
+   D D D D ]]
+
+[[ E E E E E E E E ]
+ [ F F F F ]
+ [ G G G ]
+ [ H H H H ]]
+
+[[ I I I ]
+ [ J J J ]
+ [ K K K K K]
+ [ L L L ]]
+```
+
+after padding to longest input in each step
+```
+   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
+[[ A A A A A A A A A A A ]
+   B B B B B B X X X X X X ]
+   C C C C C C C X X X X ]
+   D D D D X X X X X X X ]]
+
+[[ E E E E E E E E ]
+ [ F F F F X X X X ]
+ [ G G G X X X X X ]
+ [ H H H H X X X X ]]
+
+[[ I I I X X ]
+ [ J J J X X ]
+ [ K K K K K ]
+ [ L L L X X ]]
+```
+
+w packing ( note it's the same effective number of tokens per step, but a true bsz of 1)
+```
+   0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5
+[[ A A A A A A A A A A A B B B B B
+   B C C C C C C C D D D D E E E E
+   E E E E F F F F F G G G H H H H
+   I I I J J J J K K K K K L L L X ]]
+```
+
+cu_seqlens:
+[[ 0, 11, 17, 24, 28, 36, 41 44, 48, 51, 55, 60, 64]]
+
+
+## Multipack without Flash Attention
+
+Multipack can still be achieved without Flash attention, but with lower packing
+efficiency as we are not able to join multiple batches into a single batch due to
+context length limits without flash attention. We can use either Pytorch's Scaled
+Dot Product Attention implementation or native Pytorch attention implementation
+along with [4d attention masks](https://github.com/huggingface/transformers/pull/27539)
+to pack sequences together and avoid cross attention.
+
+<img src="./images/4d-mask.png" alt="axolotl" width="800">

docs/nccl.qmd

@@ -0,0 +1,49 @@
+---
+title: NCCL
+description: Troubleshooting NCCL issues
+---
+
+NVIDIA NCCL is a library to facilitate and optimize multi-GPU communication operations, such as broadcast, all-gather, reduce, all-reduce, etc. Broadly, NCCL configuration is highly environment-specific and is configured via several [environment variables](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/env.html). A common NCCL-related problem occurs when a long-running operation times out causing the training process to abort:
+
+```text
+Watchdog caught collective operation timeout: WorkNCCL(SeqNum=42, OpType=ALLGATHER, Timeout(ms)=1800000) ran for 1806948 milliseconds before timing out.
+```
+
+Often, this timeout will happen after 30 minutes (the default setting) and is accompanied by below-average power consumption with near 100% GPU utilization before the error is raised. Nvidia recommends [disabling PCI access control services (ACS)](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/troubleshooting.html#pci-access-control-services-acs) as a possible solution if this is available to you.
+
+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:
+
+```shell
+nvidia-smi nvlink --status
+```
+
+To force NCCL to use NVLink, simply set this in the environment:
+
+```shell
+export NCCL_P2P_LEVEL=NVL
+```
+
+If NVLink is not available in your environment there are other options for ``NCCL_P2P_LEVEL`` in the table below:
+
+| NCCL_P2P_LEVEL | Description |
+| -------------- | ----------- |
+| PIX | P2P data transfers through no more than a single PCIe bridge. Faster data transfer rates vs to paths involving multiple bridges, but slower compared to direct GPU-to-GPU communication. |
+| PXB | P2P data transfers through multiple PCIe bridges but not going through the PCIe Host Bridge; this path involves a complex routing process, potentially incurring a moderate level of latency. |
+| PHB | P2P data transfers occur over the PCIe and through a PCIe Host Bridge, typically involving the CPU, which can facilitate direct memory access but might introduce additional latency compared to more direct paths (ex PIX, NVL) |
+
+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:
+
+```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:
+
+```shell
+export NCCL_DEBUG=INFO
+export NCCL_DEBUG_SUBSYS=ALL
+export TORCH_DISTRIBUTED_DEBUG=INFO
+export TORCHELASTIC_ERROR_FILE=/PATH/TO/torcherror.log
+```
+
+Finally, if you believe your training job needs more time you can increase the timeout past 30 minutes by setting the ``ddp_timeout`` value in the Axolotl configuration. See [PyTorch init_process_group](https://pytorch.org/docs/stable/distributed.html#torch.distributed.init_process_group) for documentation on this value.

docs/rlhf.qmd

@@ -0,0 +1,72 @@
+---
+title: "RLHF (Beta)"
+description: "Reinforcement Learning from Human Feedback is a method whereby a language model is optimized from data using human feedback."
+---
+
+### 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)
+- Identity Preference Optimization (IPO)
+
+
+### RLHF using Axolotl
+
+>[!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.
+
+The various RL training methods are implemented in trl and wrapped via axolotl. Below are various examples with how you can use various preference datasets to train models that use ChatML
+
+#### DPO
+```yaml
+rl: dpo
+datasets:
+  - path: Intel/orca_dpo_pairs
+    split: train
+    type: chatml.intel
+  - path: argilla/ultrafeedback-binarized-preferences
+    split: train
+    type: chatml.argilla
+```
+
+#### IPO
+```yaml
+rl: ipo
+```
+
+#### ORPO
+
+Paper: https://arxiv.org/abs/2403.07691
+
+```yaml
+rl: orpo
+orpo_alpha: 0.1
+remove_unused_columns: false
+
+chat_template: chatml
+datasets:
+  - path: argilla/ultrafeedback-binarized-preferences-cleaned
+    type: chat_template.argilla
+```
+
+#### Using local dataset files
+```yaml
+datasets:
+  - ds_type: json
+    data_files:
+      - orca_rlhf.jsonl
+    split: train
+    type: chatml.intel
+```
+
+#### Trl autounwrap for peft
+
+Trl supports autounwrapping peft models, so that a ref model does not need to be additionally loaded, leading to less VRAM needed. This is on by default. To turn it off, pass the following config.
+
+```yaml
+# load ref model when adapter training.
+rl_adapter_ref_model: true
+```

examples/cerebras/btlm-ft.yml

@@ -0,0 +1,89 @@
+base_model: cerebras/btlm-3b-8k-base
+model_type: AutoModelForCausalLM
+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:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path: last_prepared_run
+val_set_size: 0.05
+
+adapter:
+lora_model_dir:
+sequence_len: 2048
+max_packed_sequence_len:
+sample_packing: false
+sample_packing_eff_est:
+sample_packing_seq_len_multiplier:
+total_num_tokens:
+
+lora_r:
+lora_alpha:
+lora_dropout:
+lora_target_modules:
+lora_target_linear:
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+output_dir: ./outputs/btlm-out
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 1
+optimizer: adamw_torch
+adam_beta2: 0.95
+adam_eps: 0.000000001
+max_grad_norm: 1.0
+
+torchdistx_path:
+lr_scheduler: cosine
+lr_quadratic_warmup: true
+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:
+
+gptq_groupsize:
+gptq_model_v1:
+
+warmup_steps: 32
+evals_per_epoch: 4
+saves_per_epoch: 1
+save_total_limit:
+
+debug:
+deepspeed:
+weight_decay: 0.1
+special_tokens:
+  pad_token: "<|endoftext|>"
+fsdp:
+#  - full_shard
+#  - auto_wrap
+fsdp_config:
+#  fsdp_state_dict_type: FULL_STATE_DICT
+#  fsdp_transformer_layer_cls_to_wrap: BTLMBlock

examples/cerebras/qlora.yml

@@ -1,5 +1,4 @@
 base_model: cerebras/Cerebras-GPT-1.3B
-base_model_config: cerebras/Cerebras-GPT-1.3B
 load_in_8bit: false
 load_in_4bit: true
 strict: false
@@ -7,12 +6,11 @@ push_dataset_to_hub:
 datasets:
   - path: teknium/GPT4-LLM-Cleaned
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
+dataset_prepared_path:
+val_set_size: 0.05
 adapter: qlora
 lora_model_dir:
 sequence_len: 2048
-max_packed_sequence_len: 2048
 lora_r: 16
 lora_alpha: 32
 lora_dropout: 0.05
@@ -25,9 +23,9 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
-output_dir: ./qlora-out
+output_dir: ./outputs/qlora-out
 batch_size: 4
 micro_batch_size: 4
 num_epochs: 2
@@ -37,8 +35,8 @@ lr_scheduler: cosine
 learning_rate: 0.0002
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
 early_stopping_patience:
@@ -50,8 +48,8 @@ flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.1

examples/code-llama/13b/lora.yml

@@ -1,8 +1,6 @@
 base_model: codellama/CodeLlama-13b-hf
-base_model_config: codellama/CodeLlama-13b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: true
 load_in_4bit: false
@@ -11,12 +9,13 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
 
-sequence_len: 100000
+sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 adapter: lora
 lora_model_dir:
@@ -29,20 +28,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -52,10 +51,11 @@ local_rank:
 logging_steps: 1
 xformers_attention:
 flash_attention: true
+s2_attention:
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/code-llama/13b/qlora.yml

@@ -1,8 +1,6 @@
 base_model: codellama/CodeLlama-13b-hf
-base_model_config: codellama/CodeLlama-13b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: false
 load_in_4bit: true
@@ -11,15 +9,16 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/qlora-out
 
 adapter: qlora
 lora_model_dir:
 
-sequence_len: 100000
+sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 lora_r: 32
 lora_alpha: 16
@@ -31,20 +30,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -56,8 +55,8 @@ xformers_attention:
 flash_attention: true
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/code-llama/34b/lora.yml

@@ -1,8 +1,6 @@
 base_model: codellama/CodeLlama-34b-hf
-base_model_config: codellama/CodeLlama-34b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: true
 load_in_4bit: false
@@ -11,12 +9,13 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
 
-sequence_len: 100000
+sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 adapter: lora
 lora_model_dir:
@@ -29,20 +28,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -52,10 +51,11 @@ local_rank:
 logging_steps: 1
 xformers_attention:
 flash_attention: true
+s2_attention:
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/code-llama/34b/qlora.yml

@@ -1,8 +1,6 @@
 base_model: codellama/CodeLlama-34b-hf
-base_model_config: codellama/CodeLlama-34b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: false
 load_in_4bit: true
@@ -11,15 +9,16 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/qlora-out
 
 adapter: qlora
 lora_model_dir:
 
-sequence_len: 100000
+sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 lora_r: 32
 lora_alpha: 16
@@ -31,20 +30,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -56,8 +55,8 @@ xformers_attention:
 flash_attention: true
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/code-llama/7b/lora.yml

@@ -1,8 +1,6 @@
 base_model: codellama/CodeLlama-7b-hf
-base_model_config: codellama/CodeLlama-7b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: true
 load_in_4bit: false
@@ -11,12 +9,13 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
 
-sequence_len: 100000
+sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 adapter: lora
 lora_model_dir:
@@ -29,20 +28,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -52,10 +51,11 @@ local_rank:
 logging_steps: 1
 xformers_attention:
 flash_attention: true
+s2_attention:
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/code-llama/7b/qlora.yml

@@ -1,8 +1,6 @@
 base_model: codellama/CodeLlama-7b-hf
-base_model_config: codellama/CodeLlama-7b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: false
 load_in_4bit: true
@@ -11,15 +9,16 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/qlora-out
 
 adapter: qlora
 lora_model_dir:
 
-sequence_len: 100000
+sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 lora_r: 32
 lora_alpha: 16
@@ -31,20 +30,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -56,8 +55,8 @@ xformers_attention:
 flash_attention: true
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/colab-notebooks/colab-axolotl-example.ipynb

@@ -0,0 +1,223 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "AKjdG7tbTb-n"
+   },
+   "source": [
+    "# Example notebook for running Axolotl on google colab"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "id": "RcbNpOgWRcii"
+   },
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "# Check so there is a gpu available, a T4(free tier) is enough to run this notebook\n",
+    "assert (torch.cuda.is_available()==True)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "h3nLav8oTRA5"
+   },
+   "source": [
+    "## Install Axolotl and dependencies"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "3c3yGAwnOIdi",
+    "outputId": "e3777b5a-40ef-424f-e181-62dfecd1dd01"
+   },
+   "outputs": [],
+   "source": [
+    "!pip install torch==\"2.1.2\"\n",
+    "!pip install -e git+https://github.com/OpenAccess-AI-Collective/axolotl#egg=axolotl\n",
+    "!pip install flash-attn==\"2.5.0\"\n",
+    "!pip install deepspeed==\"0.13.1\"!pip install mlflow==\"2.13.0\""
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "BW2MFr7HTjub"
+   },
+   "source": [
+    "## Create an yaml config file"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "id": "9pkF2dSoQEUN"
+   },
+   "outputs": [],
+   "source": [
+    "import yaml\n",
+    "\n",
+    "# Your YAML string\n",
+    "yaml_string = \"\"\"\n",
+    "base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T\n",
+    "model_type: LlamaForCausalLM\n",
+    "tokenizer_type: LlamaTokenizer\n",
+    "\n",
+    "load_in_8bit: false\n",
+    "load_in_4bit: true\n",
+    "strict: false\n",
+    "\n",
+    "datasets:\n",
+    "  - path: mhenrichsen/alpaca_2k_test\n",
+    "    type: alpaca\n",
+    "dataset_prepared_path:\n",
+    "val_set_size: 0.05\n",
+    "output_dir: ./outputs/qlora-out\n",
+    "\n",
+    "adapter: qlora\n",
+    "lora_model_dir:\n",
+    "\n",
+    "sequence_len: 4096\n",
+    "sample_packing: true\n",
+    "eval_sample_packing: false\n",
+    "pad_to_sequence_len: true\n",
+    "\n",
+    "lora_r: 32\n",
+    "lora_alpha: 16\n",
+    "lora_dropout: 0.05\n",
+    "lora_target_modules:\n",
+    "lora_target_linear: true\n",
+    "lora_fan_in_fan_out:\n",
+    "\n",
+    "wandb_project:\n",
+    "wandb_entity:\n",
+    "wandb_watch:\n",
+    "wandb_name:\n",
+    "wandb_log_model:\n",
+    "\n",
+    "gradient_accumulation_steps: 4\n",
+    "micro_batch_size: 2\n",
+    "num_epochs: 4\n",
+    "optimizer: paged_adamw_32bit\n",
+    "lr_scheduler: cosine\n",
+    "learning_rate: 0.0002\n",
+    "\n",
+    "train_on_inputs: false\n",
+    "group_by_length: false\n",
+    "bf16: auto\n",
+    "fp16:\n",
+    "tf32: false\n",
+    "\n",
+    "gradient_checkpointing: true\n",
+    "early_stopping_patience:\n",
+    "resume_from_checkpoint:\n",
+    "local_rank:\n",
+    "logging_steps: 1\n",
+    "xformers_attention:\n",
+    "flash_attention: true\n",
+    "\n",
+    "warmup_steps: 10\n",
+    "evals_per_epoch: 4\n",
+    "saves_per_epoch: 1\n",
+    "debug:\n",
+    "deepspeed:\n",
+    "weight_decay: 0.0\n",
+    "fsdp:\n",
+    "fsdp_config:\n",
+    "special_tokens:\n",
+    "\n",
+    "\"\"\"\n",
+    "\n",
+    "# Convert the YAML string to a Python dictionary\n",
+    "yaml_dict = yaml.safe_load(yaml_string)\n",
+    "\n",
+    "# Specify your file path\n",
+    "file_path = 'test_axolotl.yaml'\n",
+    "\n",
+    "# Write the YAML file\n",
+    "with open(file_path, 'w') as file:\n",
+    "    yaml.dump(yaml_dict, file)\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "id": "bidoj8YLTusD"
+   },
+   "source": [
+    "## Launch the training"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "colab": {
+     "base_uri": "https://localhost:8080/"
+    },
+    "id": "ydTI2Jk2RStU",
+    "outputId": "d6d0df17-4b53-439c-c802-22c0456d301b"
+   },
+   "outputs": [],
+   "source": [
+    "# Buy using the ! the comand will be executed as a bash command\n",
+    "!accelerate launch -m axolotl.cli.train /content/test_axolotl.yaml"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Play with inference"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Buy using the ! the comand will be executed as a bash command\n",
+    "!accelerate launch -m axolotl.cli.inference /content/test_axolotl.yaml \\\n",
+    "    --qlora_model_dir=\"./qlora-out\" --gradio"
+   ]
+  }
+ ],
+ "metadata": {
+  "accelerator": "GPU",
+  "colab": {
+   "gpuType": "T4",
+   "provenance": []
+  },
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.12.1"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}

examples/dbrx/16bit-lora.yaml

@@ -0,0 +1,81 @@
+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
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 512
+sample_packing: false
+pad_to_sequence_len: false
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+adapter: lora
+lora_model_dir:
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+# w1, w2, & v1 will hang the trainer
+lora_target_modules:
+  - q_proj # attn
+  - k_proj # attn
+  - v_proj # attn
+  - out_proj # attn
+  - layer # router
+#  - w1
+#  - w2
+#  - v1
+
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 1
+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
+  - auto_wrap
+fsdp_config:
+  fsdp_limit_all_gathers: true
+  fsdp_sync_module_states: true
+  fsdp_offload_params: false
+  fsdp_use_orig_params: false
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_transformer_layer_cls_to_wrap: DbrxBlock
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_activation_checkpointing: true

examples/dbrx/8bit-lora.yaml

@@ -0,0 +1,81 @@
+base_model: LnL-AI/dbrx-base-converted-v2
+trust_remote_code: true
+
+load_in_8bit: true
+load_in_4bit: false
+strict: false
+
+datasets:
+  - path: tatsu-lab/alpaca
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 512
+sample_packing: false
+pad_to_sequence_len: false
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+adapter: lora
+lora_model_dir:
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+# w1, w2, & v1 will hang the trainer
+lora_target_modules:
+  - q_proj # attn
+  - k_proj # attn
+  - v_proj # attn
+  - out_proj # attn
+  - layer # router
+#  - w1
+#  - w2
+#  - v1
+
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 1
+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
+  - auto_wrap
+fsdp_config:
+  fsdp_limit_all_gathers: true
+  fsdp_sync_module_states: true
+  fsdp_offload_params: false
+  fsdp_use_orig_params: false
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_transformer_layer_cls_to_wrap: DbrxBlock
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_activation_checkpointing: true

examples/dbrx/README.md

@@ -0,0 +1,26 @@
+# DBRX MoE
+
+Currently, for LoRA, only the `q_proj`, `k_proj`, `v_proj` `out_proj` and `layer` Linear layers are trainable.
+
+We are using the "converted" base models based on [this issue](https://huggingface.co/databricks/dbrx-instruct/discussions/10)
+where the Experts are fused as an `nn.Parameter` rather than a `nn.Linear` layer. However, the implementation
+is still a bit buggy and attempting to train a LoRA adapter over those `w1`, `w2` and `v1` layers
+results in the trainer hanging.
+
+
+### FSDP
+We've tested using the [`LnL-AI/dbrx-base-converted-v2`](https://huggingface.co/LnL-AI/dbrx-base-converted-v2) model as the base model for FSDP.
+
+The high memory usage seen w/ FSDP is due to FSDP not supporting 8bit optimizers.
+
+- 16-bit LoRA w/ FSDP
+  - ✅ w/o CPU Offload - 8x80GB uses ~80GiB/gpu
+  - ❌ w/ CPU Offload - `paged_adamw_8bit` optimizer errors from being on cpu
+- ✅ 8-bit LoRA w/ FSDP
+- ❌ 4-bit QLoRA w/ FSDP - errors w/: `Error an illegal memory access was encountered at line 90 in file /src/csrc/ops.cu`
+- ✅ bf16 full finetune w/ FSDP, freezing all but first 8 layers (8x80GB uses ~78GiB/gpu)
+
+
+### Deepspeed
+
+WIP

examples/dbrx/fft-ds-zero3.yaml

@@ -0,0 +1,56 @@
+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
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 512
+sample_packing: false
+pad_to_sequence_len: false
+
+unfrozen_parameters:
+  - transformer.blocks.[0-7].
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 1
+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

examples/falcon/config-7b-lora.yml

@@ -1,8 +1,8 @@
 base_model: tiiuae/falcon-7b
-base_model_config: tiiuae/falcon-7b
 trust_remote_code: true
 model_type: AutoModelForCausalLM
 tokenizer_type: AutoTokenizer
+
 load_in_8bit: true
 load_in_4bit: false
 gptq: false
@@ -11,8 +11,8 @@ push_dataset_to_hub:
 datasets:
   - path: teknium/GPT4-LLM-Cleaned
     type: alpaca:chat
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
+dataset_prepared_path:
+val_set_size: 0.05
 adapter: lora
 lora_model_dir:
 sequence_len: 2048
@@ -26,9 +26,9 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
-output_dir: ./falcon-7b
+output_dir: ./outputs/falcon-7b
 batch_size: 2
 micro_batch_size: 1
 num_epochs: 4
@@ -38,8 +38,8 @@ lr_scheduler: cosine
 learning_rate: 0.00003
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
 early_stopping_patience:
@@ -51,8 +51,8 @@ flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_steps: 40
-eval_steps: 5
-save_steps: 43
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0
@@ -60,5 +60,5 @@ fsdp:
 fsdp_config:
 special_tokens:
   pad_token: "<|endoftext|>"
-  bos_token: ">>ABSTRACT<<"
+  bos_token: "<|endoftext|>"
   eos_token: "<|endoftext|>"

examples/falcon/config-7b-qlora.yml

@@ -1,11 +1,11 @@
 # 1b: tiiuae/falcon-rw-1b
 # 40b: tiiuae/falcon-40b
 base_model: tiiuae/falcon-7b
-base_model_config: tiiuae/falcon-7b
 # required by falcon custom model code: https://huggingface.co/tiiuae/falcon-7b/tree/main
 trust_remote_code: true
 model_type: AutoModelForCausalLM
 tokenizer_type: AutoTokenizer
+
 load_in_8bit: false
 # enable 4bit for QLoRA
 load_in_4bit: true
@@ -17,8 +17,8 @@ datasets:
     data_files:
       - Chain-of-Thought/formatted_cot_data/gsm8k_train.json
     type: "alpaca:chat"
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
+dataset_prepared_path:
+val_set_size: 0.05
 # enable QLoRA
 adapter: qlora
 lora_model_dir:
@@ -40,9 +40,9 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
-output_dir: ./qlora-out
+output_dir: ./outputs/qlora-out
 
 # QLoRA paper Table 9
 # - 16 for 7b & 13b
@@ -53,7 +53,7 @@ output_dir: ./qlora-out
 # decrease if OOM, increase for max VRAM utilization
 micro_batch_size: 1
 gradient_accumulation_steps: 2
-num_epochs: 3
+num_epochs: 4
 # Optimizer for QLoRA
 optimizer: paged_adamw_32bit
 torchdistx_path:
@@ -64,8 +64,8 @@ lr_scheduler: cosine
 learning_rate: 0.0002
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
 # stop training after this many evaluation losses have increased in a row
@@ -80,8 +80,8 @@ flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_steps: 10
-eval_steps: 5
-save_steps: 10
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.000001
@@ -89,5 +89,5 @@ fsdp:
 fsdp_config:
 special_tokens:
   pad_token: "<|endoftext|>"
-  bos_token: ">>ABSTRACT<<"
+  bos_token: "<|endoftext|>"
   eos_token: "<|endoftext|>"

examples/falcon/config-7b.yml

@@ -1,8 +1,8 @@
 base_model: tiiuae/falcon-7b
-base_model_config: tiiuae/falcon-7b
 trust_remote_code: true
 model_type: AutoModelForCausalLM
 tokenizer_type: AutoTokenizer
+
 load_in_8bit: false
 load_in_4bit: false
 gptq: false
@@ -11,8 +11,8 @@ push_dataset_to_hub:
 datasets:
   - path: teknium/GPT4-LLM-Cleaned
     type: alpaca:chat
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
+dataset_prepared_path:
+val_set_size: 0.05
 adapter:
 lora_model_dir:
 sequence_len: 2048
@@ -26,9 +26,9 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
-output_dir: ./falcon-7b
+output_dir: ./outputs/falcon-7b
 batch_size: 2
 micro_batch_size: 1
 num_epochs: 4
@@ -38,8 +38,8 @@ lr_scheduler: cosine
 learning_rate: 0.00003
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
 early_stopping_patience:
@@ -51,8 +51,8 @@ flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_steps: 40
-eval_steps: 5
-save_steps: 43
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0
@@ -60,5 +60,5 @@ fsdp:
 fsdp_config:
 special_tokens:
   pad_token: "<|endoftext|>"
-  bos_token: ">>ABSTRACT<<"
+  bos_token: "<|endoftext|>"
   eos_token: "<|endoftext|>"

examples/gemma/qlora.yml

@@ -0,0 +1,66 @@
+# use google/gemma-7b if you have access
+base_model: mhenrichsen/gemma-7b
+model_type: AutoModelForCausalLM
+tokenizer_type: AutoTokenizer
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+# huggingface repo
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+val_set_size: 0.1
+output_dir: ./outputs/out
+
+adapter: qlora
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+
+sequence_len: 4096
+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: 3
+micro_batch_size: 2
+num_epochs: 4
+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:

examples/gptj/qlora.yml

@@ -1,5 +1,4 @@
 base_model: EleutherAI/gpt-j-6b
-base_model_config: EleutherAI/gpt-j-6b
 load_in_8bit: false
 load_in_4bit: true
 strict: false
@@ -7,8 +6,8 @@ push_dataset_to_hub:
 datasets:
   - path: teknium/GPT4-LLM-Cleaned
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
+dataset_prepared_path:
+val_set_size: 0.05
 adapter: qlora
 lora_model_dir:
 sequence_len: 2048
@@ -22,9 +21,9 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
-output_dir: ./qlora-out
+output_dir: ./outputs/qlora-out
 gradient_accumulation_steps: 2
 micro_batch_size: 2
 num_epochs: 2
@@ -34,8 +33,8 @@ lr_scheduler: cosine
 learning_rate: 0.0001
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: true
 gradient_checkpointing: true
 early_stopping_patience:
@@ -47,8 +46,8 @@ flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.1

examples/gptq-lora-7b/README.md

@@ -1,8 +0,0 @@
-# LLaMa 7B using LoRA
-
-This is a good place to start for beginners. This will run on an NVIDIA RTX4090 with no other changes needed.
-
-```shell
-accelerate launch scripts/finetune.py examples/gptq-lora-7b/config.yml
-
-```

examples/gptq-lora-7b/config.yml

@@ -1,63 +0,0 @@
-base_model: Neko-Institute-of-Science/LLaMA-7B-4bit-128g
-base_model_config: Neko-Institute-of-Science/LLaMA-7B-4bit-128g
-model_type: LlamaForCausalLM
-tokenizer_type: LlamaTokenizer
-trust_remote_code:
-load_in_8bit: true
-gptq: true
-datasets:
-  - path: vicgalle/alpaca-gpt4
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.02
-adapter:
-lora_model_dir:
-sequence_len: 2048
-max_packed_sequence_len:
-lora_r: 8
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules:
-  - q_proj
-  - v_proj
-lora_fan_in_fan_out: false
-wandb_project: llama-7b-lora-int4
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-output_dir: ./llama-7b-lora-int4
-gradient_accumulation_steps: 1
-micro_batch_size: 1
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-torchdistx_path:
-lr_scheduler: cosine
-learning_rate: 0.0000002
-train_on_inputs: false
-group_by_length: false
-fp16: true
-bf16: false
-tf32: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 5
-xformers_attention:
-flash_attention:
-gradient_checkpointing: true
-gptq_groupsize: 128
-gptq_model_v1: false
-warmup_steps: 20
-eval_steps: 110
-save_steps: 660
-debug:
-deepspeed:
-weight_decay: 0.0001
-fsdp:
-fsdp_config:
-tokens:
-  pad_token: "<pad>"
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/jamba/README.md

@@ -0,0 +1,10 @@
+# Jamba
+
+- ✅ qlora w/ deepspeed Zero-2 needs at least 2x GPUs and
+  - 35GiB VRAM per GPU w minimal context length
+  - 56GiB VRAM per GPU (w multipack enabled)
+- ✅ qlora w/ deepspeed Zero-3 needs at least 2x GPUs and 67GiB VRAM (wtf?)
+- ✅ qlora single-gpu, ~51GiB VRAM
+- ✅ multipack
+- ❓ FSDP
+- ❓ 8-bit LoRA

examples/jamba/qlora.yaml

@@ -0,0 +1,62 @@
+base_model: ai21labs/Jamba-v0.1
+trust_remote_code: true
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path:
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 4096
+sample_packing: false
+pad_to_sequence_len: false
+eval_sample_packing: false
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+adapter: qlora
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+
+low_cpu_mem_usage: true
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 2
+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:

examples/jamba/qlora_deepspeed.yaml

@@ -0,0 +1,62 @@
+base_model: ai21labs/Jamba-v0.1
+trust_remote_code: true
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path:
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 4096
+sample_packing: false
+pad_to_sequence_len: false
+eval_sample_packing: false
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+adapter: qlora
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+
+low_cpu_mem_usage: true
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 2
+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:

examples/jeopardy-bot/config.yml

@@ -1,12 +1,11 @@
 base_model: huggyllama/llama-7b
-base_model_config: huggyllama/llama-7b
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
 load_in_8bit: false
 datasets:
   - path: openaccess-ai-collective/jeopardy
     type: jeopardy
-dataset_prepared_path: last_run_prepared
+dataset_prepared_path:
 val_set_size: 0.02
 adapter:
 lora_model_dir:
@@ -20,19 +19,19 @@ lora_fan_in_fan_out: false
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
-output_dir: ./jeopardy-bot-7b
+output_dir: ./outputs/jeopardy-bot-7b
 gradient_accumulation_steps: 1
 micro_batch_size: 1
-num_epochs: 3
+num_epochs: 4
 optimizer: adamw_bnb_8bit
 torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.00003
 train_on_inputs: false
 group_by_length: false
-bf16: true
+bf16: auto
 tf32: true
 early_stopping_patience:
 resume_from_checkpoint:
@@ -43,8 +42,8 @@ flash_attention:
 gptq_groupsize:
 gptq_model_v1:
 warmup_steps: 20
-eval_steps: 110
-save_steps: 660
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.1

examples/llama-2/README.md

@@ -9,12 +9,16 @@ gradient_accumulation_steps: 2
 micro_batch_size: 1
 
 ```shell
-accelerate launch scripts/finetune.py examples/llama-2/qlora.yml
-
+accelerate launch -m axolotl.cli.train examples/llama-2/qlora.yml
 ```
 or
 
 ```shell
-accelerate launch scripts/finetune.py examples/llama-2/lora.yml
-
+accelerate launch -m axolotl.cli.train examples/llama-2/lora.yml
+```
+
+To launch a full finetuning with 16-bit precision:
+
+```shell
+accelerate launch -m axolotl.cli.train examples/llama-2/fft_optimized.yml
 ```

examples/llama-2/fft_optimized.yml

@@ -0,0 +1,68 @@
+base_model: NousResearch/Llama-2-7b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.05
+output_dir: ./outputs/out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+adapter:
+lora_model_dir:
+lora_r:
+lora_alpha:
+lora_dropout:
+lora_target_linear:
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 1
+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
+flash_attn_cross_entropy: false
+flash_attn_rms_norm: true
+flash_attn_fuse_qkv: false
+flash_attn_fuse_mlp: true
+
+warmup_steps: 100
+evals_per_epoch: 4
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed: #deepspeed_configs/zero2.json # multi-gpu only
+weight_decay: 0.1
+fsdp:
+fsdp_config:
+special_tokens:

examples/llama-2/gptq-lora.yml

@@ -0,0 +1,72 @@
+base_model: TheBloke/Llama-2-7B-GPTQ
+gptq: true
+gptq_disable_exllama: true
+model_type: AutoModelForCausalLM
+tokenizer_type: LlamaTokenizer
+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:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path:
+val_set_size: 0.05
+adapter: lora
+lora_model_dir:
+sequence_len: 4096
+sample_packing:
+lora_r: 8
+lora_alpha: 32
+lora_dropout: 0.05
+lora_target_modules:
+  - k_proj
+  - o_proj
+  - q_proj
+  - v_proj
+lora_target_linear:
+lora_fan_in_fan_out:
+wandb_project:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+output_dir: ./outputs/model-out
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 4
+optimizer: adamw_torch
+adam_beta2: 0.95
+adam_eps: 0.00001
+max_grad_norm: 1.0
+torchdistx_path:
+lr_scheduler: cosine
+lr_quadratic_warmup: true
+learning_rate: 0.000017
+train_on_inputs: false
+group_by_length: false
+bf16: false
+fp16: false
+float16: true
+tf32: true
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention:
+flash_attention:
+sdp_attention:
+flash_optimum:
+warmup_steps: 100
+evals_per_epoch: 4
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.1
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"

examples/llama-2/lisa.yml

@@ -0,0 +1,75 @@
+base_model: NousResearch/Llama-2-7b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+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: 0.05
+output_dir: ./outputs/lisa-out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+adapter:
+lora_model_dir:
+lora_r:
+lora_alpha:
+lora_dropout:
+lora_target_linear:
+lora_fan_in_fan_out:
+
+lisa_n_layers: 4
+lisa_step_interval: 20
+lisa_layers_attribute: model.layers
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 2
+micro_batch_size: 1
+num_epochs: 1
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 5e-5 # recommendation from lisa paper for 7b
+
+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
+flash_attn_cross_entropy: false
+flash_attn_rms_norm: true
+flash_attn_fuse_qkv: false
+flash_attn_fuse_mlp: true
+
+warmup_steps: 100
+evals_per_epoch: 4
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.1
+fsdp:
+fsdp_config:
+special_tokens:
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"

examples/llama-2/loftq.yml

@@ -0,0 +1,69 @@
+base_model: NousResearch/Llama-2-7b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+adapter: lora
+lora_model_dir:
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+lora_fan_in_fan_out:
+peft:
+  loftq_config:
+    loftq_bits: 4
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 4
+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
+eval_table_size:
+eval_max_new_tokens: 128
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:

examples/llama-2/lora.yml

@@ -1,8 +1,6 @@
-base_model: meta-llama/Llama-2-7b-hf
-base_model_config: meta-llama/Llama-2-7b-hf
+base_model: NousResearch/Llama-2-7b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: true
 load_in_4bit: false
@@ -11,12 +9,13 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
 
 sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 adapter: lora
 lora_model_dir:
@@ -29,20 +28,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -52,16 +51,16 @@ local_rank:
 logging_steps: 1
 xformers_attention:
 flash_attention: true
+s2_attention:
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+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:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/llama-2/qlora-fsdp.yml

@@ -0,0 +1,78 @@
+base_model: NousResearch/Llama-2-7b-hf
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: yahma/alpaca-cleaned
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.05
+output_dir: ./outputs/qlora-out
+
+adapter: qlora
+lora_model_dir:
+
+sequence_len: 512
+sample_packing: false
+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:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 4
+num_epochs: 4
+optimizer: adamw_torch
+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: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention:
+flash_attention: true
+
+warmup_steps: 10
+evals_per_epoch: 4
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+  - full_shard
+  - auto_wrap
+fsdp_config:
+  fsdp_limit_all_gathers: true
+  fsdp_sync_module_states: true
+  fsdp_offload_params: true
+  fsdp_use_orig_params: false
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
+  fsdp_state_dict_type: FULL_STATE_DICT
+special_tokens:

examples/llama-2/qlora.yml

@@ -1,8 +1,6 @@
-base_model: meta-llama/Llama-2-7b-hf
-base_model_config: meta-llama/Llama-2-7b-hf
+base_model: NousResearch/Llama-2-7b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
 
 load_in_8bit: false
 load_in_4bit: true
@@ -11,15 +9,16 @@ strict: false
 datasets:
   - path: mhenrichsen/alpaca_2k_test
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/qlora-out
 
 adapter: qlora
 lora_model_dir:
 
 sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 lora_r: 32
 lora_alpha: 16
@@ -31,20 +30,20 @@ lora_fan_in_fan_out:
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 2
-num_epochs: 3
+num_epochs: 4
 optimizer: paged_adamw_32bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -56,14 +55,12 @@ xformers_attention:
 flash_attention: true
 
 warmup_steps: 10
-eval_steps: 20
-save_steps:
+evals_per_epoch: 4
+eval_table_size:
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0
 fsdp:
 fsdp_config:
 special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/llama-2/relora.yml

@@ -1,8 +1,7 @@
-base_model: meta-llama/Llama-2-7b-hf
-base_model_config: meta-llama/Llama-2-7b-hf
+base_model: NousResearch/Llama-2-7b-hf
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
+
 
 load_in_8bit: false
 load_in_4bit: true
@@ -11,15 +10,16 @@ strict: false
 datasets:
   - path: teknium/GPT4-LLM-Cleaned
     type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./relora-out
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/relora-out
 
 adapter: qlora
 lora_model_dir:
 
 sequence_len: 4096
 sample_packing: true
+pad_to_sequence_len: true
 
 lora_r: 8
 lora_alpha: 16
@@ -35,20 +35,20 @@ relora_cpu_offload: false
 wandb_project:
 wandb_entity:
 wandb_watch:
-wandb_run_id:
+wandb_name:
 wandb_log_model:
 
 gradient_accumulation_steps: 4
 micro_batch_size: 4
-num_epochs: 3
+num_epochs: 4
 optimizer: adamw_bnb_8bit
 lr_scheduler: cosine
 learning_rate: 0.0002
 
 train_on_inputs: false
 group_by_length: false
-bf16: true
-fp16: false
+bf16: auto
+fp16:
 tf32: false
 
 gradient_checkpointing: true
@@ -60,8 +60,8 @@ xformers_attention:
 flash_attention: true
 
 warmup_steps: 10
-eval_steps: 20
-save_steps: 50
+evals_per_epoch: 4
+saves_per_epoch: 1
 debug:
 deepspeed:
 weight_decay: 0.0

examples/llama-3/README.md

@@ -0,0 +1,13 @@
+# Llama-3
+
+https://llama.meta.com/llama3/
+
+[8B Base Model](https://huggingface.co/meta-llama/Meta-Llama-3-8B)
+ - [Full Fine Tune](./fft-8b.yaml)
+   - Single GPU @ 48GB VRAM
+ - [LoRA](./lora-8b.yml)
+   - Single GPU @ 11GB VRAM
+
+[70B Base Model](https://huggingface.co/meta-llama/Meta-Llama-3-70B)
+ - [QLORA+FSDP](./qlora-fsdp-70b.yaml)
+   - Dual GPU @ 21GB VRAM

examples/llama-3/fft-8b.yaml

@@ -0,0 +1,58 @@
+base_model: meta-llama/Meta-Llama-3-8B
+model_type: LlamaForCausalLM
+tokenizer_type: AutoTokenizer
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
+datasets:
+  - path: tatsu-lab/alpaca
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.05
+output_dir: ./outputs/out
+
+sequence_len: 8192
+sample_packing: true
+pad_to_sequence_len: true
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 8
+micro_batch_size: 1
+num_epochs: 1
+optimizer: paged_adamw_8bit
+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
+fsdp:
+fsdp_config:
+special_tokens:
+  pad_token: <|end_of_text|>

examples/llama-3/instruct-lora-8b.yml

@@ -0,0 +1,76 @@
+base_model: meta-llama/Meta-Llama-3-8B-Instruct
+model_type: LlamaForCausalLM
+tokenizer_type: AutoTokenizer
+
+load_in_8bit: true
+load_in_4bit: false
+strict: false
+
+chat_template: llama3
+datasets:
+  - path: fozziethebeat/alpaca_messages_2k_test
+    type: chat_template
+    chat_template: llama3
+    field_messages: messages
+    message_field_role: role
+    message_field_content: content
+    roles:
+      user:
+        - user
+      assistant:
+        - assistant
+
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
+
+sequence_len: 4096
+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
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 4
+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
+eval_table_size:
+eval_max_new_tokens: 128
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:

examples/llama-3/lora-8b.yml

@@ -0,0 +1,70 @@
+base_model: meta-llama/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
+dataset_prepared_path:
+val_set_size: 0.05
+output_dir: ./outputs/lora-out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+adapter: lora
+lora_model_dir:
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_linear: true
+lora_fan_in_fan_out:
+lora_modules_to_save:
+  - embed_tokens
+  - lm_head
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 4
+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
+eval_table_size:
+eval_max_new_tokens: 128
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+   pad_token: <|end_of_text|>

examples/llama-3/qlora-fsdp-70b.yaml

@@ -0,0 +1,80 @@
+base_model: casperhansen/llama-3-70b-fp16
+model_type: LlamaForCausalLM
+tokenizer_type: AutoTokenizer  # PreTrainedTokenizerFast
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: tatsu-lab/alpaca
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.05
+output_dir: ./outputs/out/qlora-llama3-70b
+
+adapter: qlora
+lora_model_dir:
+
+sequence_len: 512
+sample_packing: false
+pad_to_sequence_len: true
+
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules:
+lora_target_linear: true
+lora_fan_in_fan_out:
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 4
+optimizer: adamw_torch
+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: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention:
+flash_attention: true
+
+warmup_steps: 10
+evals_per_epoch: 4
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+  - full_shard
+  - auto_wrap
+fsdp_config:
+  fsdp_limit_all_gathers: true
+  fsdp_sync_module_states: true
+  fsdp_offload_params: true
+  fsdp_use_orig_params: false
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sharding_strategy: FULL_SHARD
+special_tokens:
+  pad_token: <|end_of_text|>

examples/llama-3/qlora.yml

@@ -0,0 +1,67 @@
+base_model: meta-llama/Meta-Llama-3-8B
+model_type: AutoModelForCausalLM
+tokenizer_type: AutoTokenizer
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: aaditya/alpaca_subset_1
+    type: alpaca
+dataset_prepared_path:
+val_set_size: 0
+output_dir: ./outputs/qlora-out
+
+adapter: qlora
+lora_model_dir:
+
+sequence_len: 4096
+sample_packing: true
+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:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 4
+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
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+  pad_token: "<|end_of_text|>"

examples/mamba/config.yml

@@ -0,0 +1,61 @@
+base_model: state-spaces/mamba-2.8b
+model_type: MambaLMHeadModel
+tokenizer_type: AutoTokenizer
+tokenizer_config: EleutherAI/gpt-neox-20b
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
+datasets:
+  - path: mhenrichsen/alpaca_2k_test
+    type: alpaca
+dataset_prepared_path:
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 2048
+sample_packing: false
+pad_to_sequence_len: false
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 2
+optimizer: paged_adamw_8bit
+lr_scheduler: cosine
+learning_rate: 5e-5
+
+train_on_inputs: false
+group_by_length: true
+
+bf16: auto
+fp16:
+tf32: true
+
+gradient_checkpointing: false
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention:
+flash_attention:
+
+warmup_steps: 10
+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:
+tokens:
+save_safetensors: False

examples/mistral/README.md

@@ -0,0 +1,12 @@
+**Mistral 7B** is a language model with a total of 7.3 billion parameters, showcasing a notable performance across a variety of benchmarks.
+
+Fine Tune:
+```shell
+accelerate launch -m axolotl.cli.train examples/mistral/config.yml
+
+```
+
+If you run into CUDA OOM, use deepspeed with config zero2.json:
+```shell
+accelerate launch -m axolotl.cli.train examples/mistral/config.yml --deepspeed deepspeed_configs/zero2.json
+```

examples/mistral/bigstral-ds-zero3.yaml

@@ -0,0 +1,63 @@
+base_model: mistral-community/Mixtral-8x22B-v0.1
+model_type: AutoModelForCausalLM
+tokenizer_type: LlamaTokenizer
+trust_remote_code: true
+
+load_in_8bit: false
+load_in_4bit: false
+strict: false
+
+unfrozen_parameters:
+  - ^lm_head.weight$
+  - ^model.embed_tokens.weight$
+  - model.layers.4[4-9]+.block_sparse_moe.gate
+  - model.layers.4[4-9]+.block_sparse_moe.experts
+  - model.layers.5[0-5]+.block_sparse_moe.gate
+  - model.layers.5[0-5]+.block_sparse_moe.experts
+
+model_config:
+  output_router_logits: true
+
+datasets:
+  - path: tatsu-lab/alpaca
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.05
+output_dir: ./outputs/out
+
+sequence_len: 2048
+sample_packing: true
+pad_to_sequence_len: true
+
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 3
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0001
+
+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
+
+save_total_limit: 1
+save_steps:
+debug:
+deepspeed: deepspeed_configs/zero3_bf16_cpuoffload_params.json
+weight_decay: 0.0
+fsdp:
+fsdp_config:
+special_tokens:
+  eos_token: "<|im_end|>"
+tokens:
+  - "<|im_start|>"