wip on multimodal packing support

---------

Co-authored-by: Sunny <sunny@Sunnys-MacBook-Air.local>
Co-authored-by: sunny <sunnyliu19981005@gmail.com>

.github/workflows/base.yml

@@ -12,36 +12,24 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: "118"
-            cuda_version: 11.8.0
+          - cuda: "121"
+            cuda_version: 12.1.1
+            cudnn_version: 8
             python_version: "3.10"
-            pytorch: 2.1.2
+            pytorch: 2.3.1
             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"
-          - cuda: "121"
-            cuda_version: 12.1.0
+            cuda_version: 12.1.1
+            cudnn_version: 8
             python_version: "3.11"
             pytorch: 2.3.1
             torch_cuda_arch_list: "7.0 7.5 8.0 8.6 8.7 8.9 9.0+PTX"
+          - cuda: "124"
+            cuda_version: 12.4.1
+            cudnn_version: ""
+            python_version: "3.11"
+            pytorch: 2.4.0
+            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
@@ -67,6 +55,7 @@ jobs:
           labels: ${{ steps.metadata.outputs.labels }}
           build-args: |
             CUDA_VERSION=${{ matrix.cuda_version }}
+            CUDNN_VERSION=${{ matrix.cudnn_version }}
             CUDA=${{ matrix.cuda }}
             PYTHON_VERSION=${{ matrix.python_version }}
             PYTORCH_VERSION=${{ matrix.pytorch }}

.github/workflows/lint.yml

@@ -6,7 +6,7 @@ on:
        - '**.py'
        - 'requirements.txt'
        - '.github/workflows/*.yml'
-       - "*.md"
+       - "*.[q]md"
        - "examples/**/*.y[a]?ml"
   workflow_dispatch:
 

.github/workflows/main.yml

@@ -13,28 +13,22 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
+          - cuda: 121
+            cuda_version: 12.1.1
             python_version: "3.10"
-            pytorch: 2.1.2
-            axolotl_extras:
-            axolotl_args: "--extra-index-url https://download.pytorch.org/whl/cu118"
+            pytorch: 2.3.1
+            axolotl_extras: mamba-ssm
           - 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
+            cuda_version: 12.1.1
             python_version: "3.11"
             pytorch: 2.3.1
-            axolotl_extras:
+            axolotl_extras: mamba-ssm
             is_latest: true
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.0
+            axolotl_extras:
     runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout
@@ -65,6 +59,7 @@ jobs:
           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 }}
+            ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}
             ${{ (matrix.is_latest) && format('{0}-latest', steps.metadata.outputs.tags) || '' }}
           labels: ${{ steps.metadata.outputs.labels }}
 
@@ -75,27 +70,22 @@ jobs:
     strategy:
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
+          - cuda: 121
+            cuda_version: 12.1.1
             python_version: "3.10"
-            pytorch: 2.1.2
+            pytorch: 2.3.1
             axolotl_extras:
           - 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
+            cuda_version: 12.1.1
             python_version: "3.11"
             pytorch: 2.3.1
             axolotl_extras:
             is_latest: true
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.0
+            axolotl_extras:
     runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout
@@ -134,7 +124,7 @@ jobs:
       matrix:
         include:
           - cuda: 121
-            cuda_version: 12.1.0
+            cuda_version: 12.1.1
             python_version: "3.11"
             pytorch: 2.3.1
             axolotl_extras:

.github/workflows/multi-gpu-e2e.yml

@@ -0,0 +1,55 @@
+name: docker-multigpu-tests-biweekly
+
+on:
+  pull_request:
+    paths:
+      - 'tests/e2e/multigpu/*.py'
+  workflow_dispatch:
+  schedule:
+    - cron: '0 0 * * 1,4'  # Runs at 00:00 UTC every monday & thursday
+
+jobs:
+  test-axolotl-multigpu:
+    if: ${{ ! contains(github.event.commits[0].message, '[skip docker]]') && github.repository_owner == 'axolotl-ai-cloud' }}
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - cuda: 121
+            cuda_version: 12.1.1
+            python_version: "3.11"
+            pytorch: 2.3.1
+            axolotl_extras:
+            num_gpus: 2
+          - cuda: 121
+            cuda_version: 12.1.1
+            python_version: "3.11"
+            pytorch: 2.3.1
+            axolotl_extras:
+            num_gpus: 2
+            nightly_build: "true"
+    runs-on: [self-hosted, modal]
+    timeout-minutes: 120
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Install Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.10"
+      - name: Install Modal
+        run: |
+          python -m pip install --upgrade pip
+          pip install modal==0.63.64 jinja2
+      - name: Update env vars
+        run: |
+          echo "BASE_TAG=main-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}" >> $GITHUB_ENV
+          echo "PYTORCH_VERSION=${{ matrix.pytorch}}" >> $GITHUB_ENV
+          echo "AXOLOTL_ARGS=${{ matrix.axolotl_args}}" >> $GITHUB_ENV
+          echo "AXOLOTL_EXTRAS=${{ matrix.axolotl_extras}}" >> $GITHUB_ENV
+          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
+          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
+          echo "NIGHTLY_BUILD=${{ matrix.nightly_build }}" >> $GITHUB_ENV
+      - name: Run tests job on Modal
+        run: |
+          modal run cicd.multigpu

.github/workflows/nightlies.yml

@@ -12,28 +12,22 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
+          - cuda: 121
+            cuda_version: 12.1.1
             python_version: "3.10"
-            pytorch: 2.1.2
-            axolotl_extras:
-            axolotl_args: "--extra-index-url https://download.pytorch.org/whl/cu118"
-          - cuda: 121
-            cuda_version: 12.1.0
-            python_version: "3.10"
-            pytorch: 2.1.2
+            pytorch: 2.3.1
             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
+            cuda_version: 12.1.1
             python_version: "3.11"
             pytorch: 2.3.1
             axolotl_extras:
             is_latest: true
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.0
+            axolotl_extras:
     runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout
@@ -75,27 +69,22 @@ jobs:
     strategy:
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
+          - cuda: 121
+            cuda_version: 12.1.1
             python_version: "3.10"
-            pytorch: 2.1.2
+            pytorch: 2.3.1
             axolotl_extras:
           - 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
+            cuda_version: 12.1.1
             python_version: "3.11"
             pytorch: 2.3.1
             axolotl_extras:
             is_latest: true
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.0
+            axolotl_extras:
     runs-on: axolotl-gpu-runner
     steps:
       - name: Checkout

.github/workflows/tests-nightly.yml

@@ -0,0 +1,120 @@
+name: Tests Nightly against upstream main
+on:
+  workflow_dispatch:
+  schedule:
+    - cron: '0 0 * * *'  # Runs at 00:00 UTC every day
+
+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
+        env:
+          SKIP: no-commit-to-branch
+
+  pytest:
+    name: PyTest
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        python_version: ["3.10", "3.11"]
+        pytorch_version: ["2.3.1", "2.4.0"]
+    timeout-minutes: 20
+
+    steps:
+      - name: Check out repository code
+        uses: actions/checkout@v3
+
+      - name: Setup Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: ${{ matrix.python_version }}
+          cache: 'pip' # caching pip dependencies
+
+      - name: Install PyTorch
+        run: |
+          pip3 install torch==${{ matrix.pytorch_version }} --index-url https://download.pytorch.org/whl/cpu
+
+      - name: Update requirements.txt
+        run: |
+          sed -i 's#^transformers.*#transformers @ git+https://github.com/huggingface/transformers.git@main#' requirements.txt
+          sed -i 's#^peft.*#peft @ git+https://github.com/huggingface/peft.git@main#' requirements.txt
+          sed -i 's#^accelerate.*#accelerate @ git+https://github.com/huggingface/accelerate.git@main#' requirements.txt
+
+      - name: Install dependencies
+        run: |
+          pip3 install --upgrade pip
+          pip3 install --upgrade packaging
+          pip3 install -U -e .
+          pip3 install -r requirements-tests.txt
+
+      - name: Run tests
+        run: |
+          pytest --ignore=tests/e2e/ tests/
+
+      - name: cleanup pip cache
+        run: |
+          find "$(pip cache dir)/http-v2" -type f -mtime +14 -exec rm {} \;
+
+  docker-e2e-tests:
+    if: github.repository_owner == 'axolotl-ai-cloud'
+    # this job needs to be run on self-hosted GPU runners...
+    runs-on: [self-hosted, modal]
+    timeout-minutes: 60
+    needs: [pre-commit, pytest]
+
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - cuda: 121
+            cuda_version: 12.1.1
+            python_version: "3.10"
+            pytorch: 2.3.1
+            num_gpus: 1
+            axolotl_extras: mamba-ssm
+            nightly_build: "true"
+          - cuda: 121
+            cuda_version: 12.1.1
+            python_version: "3.11"
+            pytorch: 2.3.1
+            num_gpus: 1
+            axolotl_extras: mamba-ssm
+            nightly_build: "true"
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.0
+            num_gpus: 1
+            axolotl_extras:
+            nightly_build: "true"
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Install Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.10"
+      - name: Install Modal
+        run: |
+          python -m pip install --upgrade pip
+          pip install modal==0.63.64 jinja2
+      - name: Update env vars
+        run: |
+          echo "BASE_TAG=main-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}" >> $GITHUB_ENV
+          echo "PYTORCH_VERSION=${{ matrix.pytorch}}" >> $GITHUB_ENV
+          echo "AXOLOTL_ARGS=${{ matrix.axolotl_args}}" >> $GITHUB_ENV
+          echo "AXOLOTL_EXTRAS=${{ matrix.axolotl_extras}}" >> $GITHUB_ENV
+          echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
+          echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
+          echo "NIGHTLY_BUILD=${{ matrix.nightly_build }}" >> $GITHUB_ENV
+      - name: Run tests job on Modal
+        run: |
+          modal run cicd.tests

.github/workflows/tests.yml

@@ -26,6 +26,8 @@ jobs:
           python-version: "3.10"
           cache: 'pip' # caching pip dependencies
       - uses: pre-commit/action@v3.0.0
+        env:
+          SKIP: no-commit-to-branch
 
   pytest:
     name: PyTest
@@ -34,6 +36,7 @@ jobs:
       fail-fast: false
       matrix:
         python_version: ["3.10", "3.11"]
+        pytorch_version: ["2.3.1", "2.4.0"]
     timeout-minutes: 20
 
     steps:
@@ -46,6 +49,10 @@ jobs:
           python-version: ${{ matrix.python_version }}
           cache: 'pip' # caching pip dependencies
 
+      - name: Install PyTorch
+        run: |
+          pip3 install torch==${{ matrix.pytorch_version }} --index-url https://download.pytorch.org/whl/cpu
+
       - name: Install dependencies
         run: |
           pip3 install --upgrade pip
@@ -57,6 +64,10 @@ jobs:
         run: |
           pytest --ignore=tests/e2e/ tests/
 
+      - name: cleanup pip cache
+        run: |
+          find "$(pip cache dir)/http-v2" -type f -mtime +14 -exec rm {} \;
+
   docker-e2e-tests:
     if: github.repository_owner == 'axolotl-ai-cloud'
     # this job needs to be run on self-hosted GPU runners...
@@ -68,27 +79,24 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: 118
-            cuda_version: 11.8.0
+          - cuda: 121
+            cuda_version: 12.1.1
             python_version: "3.10"
-            pytorch: 2.1.2
-            axolotl_args: "--extra-index-url https://download.pytorch.org/whl/cu118"
+            pytorch: 2.3.1
             num_gpus: 1
+            axolotl_extras: mamba-ssm
           - cuda: 121
-            cuda_version: 12.1.0
-            python_version: "3.10"
-            pytorch: 2.1.2
-            num_gpus: 1
-          - cuda: 121
-            cuda_version: 12.1.0
-            python_version: "3.11"
-            pytorch: 2.2.2
-            num_gpus: 1
-          - cuda: 121
-            cuda_version: 12.1.0
+            cuda_version: 12.1.1
             python_version: "3.11"
             pytorch: 2.3.1
             num_gpus: 1
+            axolotl_extras: mamba-ssm
+          - cuda: 124
+            cuda_version: 12.4.1
+            python_version: "3.11"
+            pytorch: 2.4.0
+            num_gpus: 1
+            axolotl_extras:
     steps:
       - name: Checkout
         uses: actions/checkout@v4
@@ -99,12 +107,13 @@ jobs:
       - name: Install Modal
         run: |
           python -m pip install --upgrade pip
-          pip install modal jinja2
+          pip install modal==0.63.64 jinja2
       - name: Update env vars
         run: |
           echo "BASE_TAG=main-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}" >> $GITHUB_ENV
           echo "PYTORCH_VERSION=${{ matrix.pytorch}}" >> $GITHUB_ENV
           echo "AXOLOTL_ARGS=${{ matrix.axolotl_args}}" >> $GITHUB_ENV
+          echo "AXOLOTL_EXTRAS=${{ matrix.axolotl_extras}}" >> $GITHUB_ENV
           echo "CUDA=${{ matrix.cuda }}" >> $GITHUB_ENV
           echo "N_GPUS=${{ matrix.num_gpus }}" >> $GITHUB_ENV
       - name: Run tests job on Modal

.mypy.ini

@@ -11,6 +11,9 @@ ignore_errors = True
 [mypy-axolotl.models.mixtral.*]
 ignore_errors = True
 
+[mypy-axolotl.integrations.liger.models.*]
+ignore_errors = True
+
 [mypy-axolotl.models.phi.*]
 ignore_errors = True
 

.pre-commit-config.yaml

@@ -8,6 +8,8 @@ repos:
     -   id: check-yaml
     -   id: end-of-file-fixer
     -   id: trailing-whitespace
+    -   id: no-commit-to-branch
+        args: ['--branch', 'main']
 -   repo: https://github.com/psf/black
     rev: 23.3.0
     hooks:

README.md

@@ -1,5 +1,9 @@
 # Axolotl
 
+![tests](https://github.com/axolotl-ai-cloud/axolotl/actions/workflows/tests.yml/badge.svg)
+![tests-nightly](https://github.com/axolotl-ai-cloud/axolotl/actions/workflows/tests-nightly.yml/badge.svg)
+![multigpu-semi-weekly tests](https://github.com/axolotl-ai-cloud/axolotl/actions/workflows/multi-gpu-e2e.yml/badge.svg)
+
 Axolotl is a tool designed to streamline the fine-tuning of various AI models, offering support for multiple configurations and architectures.
 
 Features:
@@ -7,7 +11,7 @@ Features:
 - Supports fullfinetune, lora, qlora, relora, and gptq
 - Customize configurations using a simple yaml file or CLI overwrite
 - Load different dataset formats, use custom formats, or bring your own tokenized datasets
-- Integrated with xformer, flash attention, rope scaling, and multipacking
+- Integrated with xformer, flash attention, [liger kernel](https://github.com/linkedin/Liger-Kernel), rope scaling, and multipacking
 - Works with single GPU or multiple GPUs via FSDP or Deepspeed
 - Easily run with Docker locally or on the cloud
 - Log results and optionally checkpoints to wandb or mlflow
@@ -22,38 +26,50 @@ Features:
 <td>
 
 ## Table of Contents
-- [Introduction](#axolotl)
-- [Supported Features](#axolotl-supports)
-- [Quickstart](#quickstart-)
-- [Environment](#environment)
-  - [Docker](#docker)
-  - [Conda/Pip venv](#condapip-venv)
-  - [Cloud GPU](#cloud-gpu) - Latitude.sh, JarvisLabs, RunPod
-  - [Bare Metal Cloud GPU](#bare-metal-cloud-gpu)
-  - [Windows](#windows)
-  - [Mac](#mac)
-  - [Google Colab](#google-colab)
-  - [Launching on public clouds via SkyPilot](#launching-on-public-clouds-via-skypilot)
-  - [Launching on public clouds via dstack](#launching-on-public-clouds-via-dstack)
-- [Dataset](#dataset)
-- [Config](#config)
-  - [Train](#train)
-  - [Inference](#inference-playground)
-  - [Merge LORA to Base](#merge-lora-to-base)
-  - [Special Tokens](#special-tokens)
-  - [All Config Options](#all-config-options)
-- Advanced Topics
-  - [Multipack](./docs/multipack.qmd)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
-  - [RLHF & DPO](./docs/rlhf.qmd)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
-  - [Dataset Pre-Processing](./docs/dataset_preprocessing.qmd)<svg width="24" height="24" viewBox="0 0 24 24" xmlns="http://www.w3.org/2000/svg"><path d="M17 13.5v6H5v-12h6m3-3h6v6m0-6-9 9" class="icon_svg-stroke" stroke="#666" stroke-width="1.5" fill="none" fill-rule="evenodd" stroke-linecap="round" stroke-linejoin="round"></path></svg>
-- [Common Errors](#common-errors-)
-  - [Tokenization Mismatch b/w Training & Inference](#tokenization-mismatch-bw-inference--training)
-- [Debugging Axolotl](#debugging-axolotl)
-- [Need Help?](#need-help-)
-- [Badge](#badge-)
-- [Community Showcase](#community-showcase)
-- [Contributing](#contributing-)
-- [Sponsors](#sponsors-)
+- [Axolotl](#axolotl)
+  - [Table of Contents](#table-of-contents)
+  - [Axolotl supports](#axolotl-supports)
+  - [Quickstart ⚡](#quickstart-)
+    - [Usage](#usage)
+  - [Advanced Setup](#advanced-setup)
+    - [Environment](#environment)
+      - [Docker](#docker)
+      - [Conda/Pip venv](#condapip-venv)
+      - [Cloud GPU](#cloud-gpu)
+      - [Bare Metal Cloud GPU](#bare-metal-cloud-gpu)
+        - [LambdaLabs](#lambdalabs)
+        - [GCP](#gcp)
+      - [Windows](#windows)
+      - [Mac](#mac)
+      - [Google Colab](#google-colab)
+      - [Launching on public clouds via SkyPilot](#launching-on-public-clouds-via-skypilot)
+      - [Launching on public clouds via dstack](#launching-on-public-clouds-via-dstack)
+    - [Dataset](#dataset)
+    - [Config](#config)
+      - [All Config Options](#all-config-options)
+    - [Train](#train)
+      - [Preprocess dataset](#preprocess-dataset)
+      - [Multi-GPU](#multi-gpu)
+        - [DeepSpeed](#deepspeed)
+        - [FSDP](#fsdp)
+        - [FSDP + QLoRA](#fsdp--qlora)
+        - [Weights \& Biases Logging](#weights--biases-logging)
+        - [Special Tokens](#special-tokens)
+      - [Liger Kernel](#liger-kernel)
+    - [Inference Playground](#inference-playground)
+    - [Merge LORA to base](#merge-lora-to-base)
+  - [Common Errors 🧰](#common-errors-)
+    - [Tokenization Mismatch b/w Inference \& Training](#tokenization-mismatch-bw-inference--training)
+  - [Debugging Axolotl](#debugging-axolotl)
+  - [Need help? 🙋](#need-help-)
+  - [Badge ❤🏷️](#badge-️)
+  - [Community Showcase](#community-showcase)
+  - [Contributing 🤝](#contributing-)
+  - [Sponsors 🤝❤](#sponsors-)
+      - [💎 Diamond Sponsors - Contact directly](#-diamond-sponsors---contact-directly)
+      - [🥇 Gold Sponsors - $5000/mo](#-gold-sponsors---5000mo)
+      - [🥈 Silver Sponsors - $1000/mo](#-silver-sponsors---1000mo)
+      - [🥉 Bronze Sponsors - $500/mo](#-bronze-sponsors---500mo)
 
 </td>
 <td>
@@ -95,6 +111,7 @@ Features:
 | RWKV        | ✅         | ❓    | ❓     | ❓             | ❓                 | ❓          | ❓            |
 | Qwen        | ✅         | ✅    | ✅     | ❓             | ❓                 | ❓          | ❓            |
 | Gemma       | ✅         | ✅    | ✅     | ❓             | ❓                 | ✅          | ❓            |
+| Jamba       | ✅         | ✅    | ✅     | ❓             | ❓                 | ✅          | ❓            |
 
 ✅: supported
 ❌: not supported
@@ -333,7 +350,7 @@ For further and fine-grained use cases, please refer to the official [dstack doc
 
 Axolotl supports a variety of dataset formats.  It is recommended to use a JSONL.  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.
 
-See [these docs](https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/) for more information on how to use different dataset formats.
+See [the documentation](https://axolotl-ai-cloud.github.io/axolotl/docs/dataset-formats/) for more information on how to use different dataset formats.
 
 ### Config
 
@@ -514,6 +531,25 @@ tokens: # these are delimiters
 
 When you include these tokens in your axolotl config, axolotl adds these tokens to the tokenizer's vocabulary.
 
+##### Liger Kernel
+
+Liger Kernel: Efficient Triton Kernels for LLM Training
+
+https://github.com/linkedin/Liger-Kernel
+
+Liger (LinkedIn GPU Efficient Runtime) Kernel is a collection of Triton kernels designed specifically for LLM training.
+It can effectively increase multi-GPU training throughput by 20% and reduces memory usage by 60%. The Liger Kernel
+composes well and is compatible with both FSDP and Deepspeed.
+
+```yaml
+plugins:
+  - axolotl.integrations.liger.LigerPlugin
+liger_rope: true
+liger_rms_norm: true
+liger_swiglu: true
+liger_fused_linear_cross_entropy: true
+```
+
 ### Inference Playground
 
 Axolotl allows you to load your model in an interactive terminal playground for quick experimentation.

_quarto.yml

@@ -36,6 +36,8 @@ website:
             - docs/nccl.qmd
             - docs/mac.qmd
             - docs/multi-node.qmd
+            - docs/unsloth.qmd
+            - docs/amd_hpc.qmd
         - section: "Dataset Formats"
           contents: docs/dataset-formats/*
         - section: "Reference"

cicd/Dockerfile.jinja

@@ -8,6 +8,7 @@ ENV BNB_CUDA_VERSION="{{ CUDA }}"
 ENV PYTORCH_VERSION="{{ PYTORCH_VERSION }}"
 ENV GITHUB_REF="{{ GITHUB_REF }}"
 ENV GITHUB_SHA="{{ GITHUB_SHA }}"
+ENV NIGHTLY_BUILD="{{ NIGHTLY_BUILD }}"
 
 RUN apt-get update && \
     apt-get install -y --allow-change-held-packages vim curl nano libnccl2 libnccl-dev
@@ -23,14 +24,20 @@ RUN git fetch origin +$GITHUB_REF && \
 
 # If AXOLOTL_EXTRAS is set, append it in brackets
 RUN pip install causal_conv1d
+RUN if [ "$NIGHTLY_BUILD" = "true" ] ; then \
+        sed -i 's#^transformers.*#transformers @ git+https://github.com/huggingface/transformers.git@main#' requirements.txt; \
+        sed -i 's#^peft.*#peft @ git+https://github.com/huggingface/peft.git@main#' requirements.txt; \
+        sed -i 's#^accelerate.*#accelerate @ git+https://github.com/huggingface/accelerate.git@main#' requirements.txt; \
+    fi
+
 RUN if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,optimizers,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
     else \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,optimizers] $AXOLOTL_ARGS; \
     fi
 
 # So we can test the Docker image
-RUN pip install pytest
+RUN pip install -r requirements-tests.txt
 
 # fix so that git fetch/pull from remote works
 RUN git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \

cicd/cicd.sh

@@ -2,5 +2,5 @@
 set -e
 
 pytest --ignore=tests/e2e/ /workspace/axolotl/tests/
-pytest /workspace/axolotl/tests/e2e/patched/
-pytest --ignore=tests/e2e/patched/ /workspace/axolotl/tests/e2e/
+pytest -n1 --dist loadfile -v /workspace/axolotl/tests/e2e/patched/ /workspace/axolotl/tests/e2e/integrations/
+pytest --ignore=tests/e2e/patched/ --ignore=tests/e2e/multigpu/ --ignore=tests/e2e/integrations/ /workspace/axolotl/tests/e2e/

cicd/multigpu.py

@@ -0,0 +1,77 @@
+"""
+ modal application to run axolotl gpu tests in Modal
+ """
+# pylint: disable=duplicate-code
+
+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", "2.3.1"),
+    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.11-cu121-2.3.1"),
+    "CUDA": os.environ.get("CUDA", "121"),
+    "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", 2))
+GPU_CONFIG = modal.gpu.H100(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 * N_GPUS,
+)
+def cicd_pytest():
+    run_cmd("./cicd/multigpu.sh", "/workspace/axolotl")
+
+
+@stub.local_entrypoint()
+def main():
+    cicd_pytest.remote()

cicd/multigpu.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+set -e
+
+# only run one test at a time so as not to OOM the GPU
+pytest -n1 /workspace/axolotl/tests/e2e/multigpu/

cicd/tests.py

@@ -1,6 +1,8 @@
 """
  modal application to run axolotl gpu tests in Modal
  """
+# pylint: disable=duplicate-code
+
 import os
 import pathlib
 import tempfile
@@ -21,11 +23,12 @@ 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"),
+    "PYTORCH_VERSION": os.environ.get("PYTORCH_VERSION", "2.3.1"),
+    "BASE_TAG": os.environ.get("BASE_TAG", "main-base-py3.11-cu121-2.3.1"),
+    "CUDA": os.environ.get("CUDA", "121"),
     "GITHUB_REF": os.environ.get("GITHUB_REF", "refs/heads/main"),
     "GITHUB_SHA": os.environ.get("GITHUB_SHA", ""),
+    "NIGHTLY_BUILD": os.environ.get("NIGHTLY_BUILD", ""),
 }
 
 dockerfile_contents = df_template.render(**df_args)

docker/Dockerfile

@@ -22,9 +22,9 @@ WORKDIR /workspace/axolotl
 # 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] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,optimizers,$AXOLOTL_EXTRAS] $AXOLOTL_ARGS; \
     else \
-        pip install -e .[deepspeed,flash-attn,mamba-ssm,galore] $AXOLOTL_ARGS; \
+        pip install -e .[deepspeed,flash-attn,optimizers] $AXOLOTL_ARGS; \
     fi
 
 # So we can test the Docker image

docker/Dockerfile-base

@@ -3,7 +3,7 @@ ARG CUDNN_VERSION="8"
 ARG UBUNTU_VERSION="22.04"
 ARG MAX_JOBS=4
 
-FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION as base-builder
+FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION AS base-builder
 
 ENV PATH="/root/miniconda3/bin:${PATH}"
 

docker/Dockerfile-cloud

@@ -3,7 +3,6 @@ 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"
 

docker/Dockerfile-cloud-no-tmux

@@ -3,7 +3,6 @@ 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"
 

docs/amd_hpc.qmd

@@ -0,0 +1,108 @@
+---
+title: Training with AMD GPUs on HPC Systems
+description: A comprehensive guide for using Axolotl on distributed systems with AMD GPUs
+---
+
+This guide provides step-by-step instructions for installing and configuring Axolotl on a High-Performance Computing (HPC) environment equipped with AMD GPUs.
+
+## Setup
+
+### 1. Install Python
+
+We recommend using Miniforge, a minimal conda-based Python distribution:
+
+```bash
+curl -L -O "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh"
+bash Miniforge3-$(uname)-$(uname -m).sh
+```
+
+### 2. Configure Python Environment
+Add Python to your PATH and ensure it's available at login:
+
+```bash
+echo 'export PATH=~/miniforge3/bin:$PATH' >> ~/.bashrc
+echo 'if [ -f ~/.bashrc ]; then . ~/.bashrc; fi' >> ~/.bash_profile
+```
+
+### 3. Load AMD GPU Software
+
+Load the ROCm module:
+
+```bash
+module load rocm/5.7.1
+```
+
+Note: The specific module name and version may vary depending on your HPC system. Consult your system documentation for the correct module name.
+
+### 4. Install PyTorch
+
+Install PyTorch with ROCm support:
+
+```bash
+pip install -U torch torchvision torchaudio --index-url https://download.pytorch.org/whl/rocm5.7 --force-reinstall
+```
+
+### 5. Install Flash Attention
+
+Clone and install the Flash Attention repository:
+
+```bash
+git clone --recursive https://github.com/ROCmSoftwarePlatform/flash-attention.git
+export GPU_ARCHS="gfx90a"
+cd flash-attention
+export PYTHON_SITE_PACKAGES=$(python -c 'import site; print(site.getsitepackages()[0])')
+patch "${PYTHON_SITE_PACKAGES}/torch/utils/hipify/hipify_python.py" hipify_patch.patch
+pip install .
+```
+
+### 6. Install Axolotl
+
+Clone and install Axolotl:
+
+```bash
+git clone https://github.com/axolotl-ai-cloud/axolotl
+cd axolotl
+pip install packaging ninja
+pip install -e .
+```
+
+### 7. Apply xformers Workaround
+
+xformers appears to be incompatible with ROCm. Apply the following workarounds:
+ - Edit $HOME/packages/axolotl/src/axolotl/monkeypatch/llama_attn_hijack_flash.py modifying the code to always return `False` for SwiGLU availability from xformers.
+ - Edit $HOME/miniforge3/lib/python3.10/site-packages/xformers/ops/swiglu_op.py replacing the "SwiGLU" function with a pass statement.
+
+### 8. Prepare Job Submission Script
+
+Create a script for job submission using your HPC's particular software (e.g. Slurm, PBS). Include necessary environment setup and the command to run Axolotl training. If the compute node(s) do(es) not have internet access, it is recommended to include
+
+```bash
+export TRANSFORMERS_OFFLINE=1
+export HF_DATASETS_OFFLINE=1
+```
+
+### 9. Download Base Model
+
+Download a base model using the Hugging Face CLI:
+
+```bash
+huggingface-cli download meta-llama/Meta-Llama-3.1-8B --local-dir ~/hfdata/llama3.1-8B
+```
+
+### 10. Create Axolotl Configuration
+
+Create an Axolotl configuration file (YAML format) tailored to your specific training requirements and dataset. Use FSDP for multi-node training.
+
+Note: Deepspeed did not work at the time of testing. However, if anyone managed to get it working, please let us know.
+
+### 11. Preprocess Data
+
+Run preprocessing on the login node:
+
+```bash
+CUDA_VISIBLE_DEVICES="" python -m axolotl.cli.preprocess /path/to/your/config.yaml
+```
+
+### 12. Train
+
+You are now ready to submit your previously prepared job script. 🚂

docs/dataset-formats/conversation.qmd

@@ -54,6 +54,14 @@ conversations where `from` is `prompter` `assistant` instead of default sharegpt
 {"conversations": [{"from": "...", "value": "..."}]}
 ```
 
+## sharegpt.load_ultrachat
+
+conversations where the turns field is 'messages', human is 'user' and gpt is 'assistant'.
+
+```{.json filename="data.jsonl"}
+{"messages": [{"user": "...", "assistant": "..."}]}
+```
+
 ## sharegpt_jokes
 
 creates a chat where bot is asked to tell a joke, then explain why the joke is funny

docs/dataset-formats/tokenized.qmd

@@ -7,7 +7,7 @@ order: 5
 - Pass an empty `type:` in your axolotl config.
 - Columns in Dataset must be exactly `input_ids`, `attention_mask`, `labels`
 - To indicate that a token should be ignored during training, set its corresponding label to `-100`.
-- Do not add BOS/EOS. Axolotl will add them for you based on the default tokenizer for the model you're using.
+- You must add BOS and EOS, and make sure that you are training on EOS by not setting its label to -100.
 - For pretraining, do not truncate/pad documents to the context window length.
 - For instruction training, documents must be truncated/padded as desired.
 

docs/input_output.qmd

@@ -205,7 +205,7 @@ ds = load_from_disk(f'last_run_prepared/{directory[0]}/')
     hi there!.  goodbye  farewell</s>
 ```
 
-We can check that the right tokens are ingored by comparing the labels
+We can check that the right tokens are ignored by comparing the labels
 to each token:
 
 ```python

docs/multimodal.qmd

@@ -0,0 +1,28 @@
+# MultiModal / Vision Language Models (BETA)
+
+### Supported Models
+
+- Mllama, i.e. llama with vision models
+
+### Usage
+
+Currently multimodal support is limited and doesn't have full feature parity. To finetune a multimodal Llama w/ LoRA,
+you'll need to use the following in YAML in combination with the rest of the required hyperparams.
+
+```yaml
+base_model: alpindale/Llama-3.2-11B-Vision-Instruct
+processor_type: AutoProcessor
+skip_prepare_dataset: true
+
+chat_template: llama3_2_vision
+datasets:
+  - path: HuggingFaceH4/llava-instruct-mix-vsft
+    type: chat_template
+    split: train[:1%]
+    field_messages: messages
+remove_unused_columns: false
+sample_packing: false
+
+# only finetune the Language model, leave the vision model and vision tower frozen
+lora_target_modules: 'language_model.model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'
+```

docs/torchao.qmd

@@ -0,0 +1,19 @@
+---
+title: "PyTorch ao"
+description: "Custom data types and layouts for training and inference"
+---
+
+### Installation
+
+Stable Release from the PyTorch index
+
+```bash
+pip install torchao --extra-index-url https://download.pytorch.org/whl/cu121 # full options are cpu/cu118/cu121/cu124
+```
+
+
+Nightly release
+
+```bash
+pip install --pre torchao-nightly --index-url https://download.pytorch.org/whl/nightly/cu121 # full options are cpu/cu118/cu121/cu124
+```

docs/unsloth.qmd

@@ -0,0 +1,49 @@
+---
+title: "Unsloth"
+description: "Hyper-optimized QLoRA finetuning for single GPUs"
+---
+
+### Overview
+
+Unsloth provides hand-written optimized kernels for LLM finetuning that slightly improve speed and VRAM over
+standard industry baselines.
+
+
+### Installation
+
+The following will install unsloth from source and downgrade xformers as unsloth is incompatible with the most up
+to date libraries.
+
+```bash
+pip install --no-deps "unsloth @ git+https://github.com/unslothai/unsloth.git"
+pip install --no-deps --force-reinstall xformers==0.0.26.post1
+```
+
+### Using unsloth w Axolotl
+
+Axolotl exposes a few configuration options to try out unsloth and get most of the performance gains.
+
+Our unsloth integration is currently limited to the following model architectures:
+ - llama
+
+These options are specific to LoRA finetuning and cannot be used for multi-GPU finetuning
+```yaml
+unsloth_lora_mlp: true
+unsloth_lora_qkv: true
+unsloth_lora_o: true
+```
+
+These options are composable and can be used with multi-gpu finetuning
+```yaml
+unsloth_cross_entropy_loss: true
+unsloth_rms_norm: true
+unsloth_rope: true
+```
+
+### Limitations
+
+- Single GPU only; e.g. no multi-gpu support
+- No deepspeed or FSDP support (requires multi-gpu)
+- LoRA + QLoRA support only. No full fine tunes or fp8 support.
+- Limited model architecture support. Llama, Phi, Gemma, Mistral only
+- No MoE support.

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

@@ -43,7 +43,6 @@
    },
    "outputs": [],
    "source": [
-    "!pip install torch==\"2.1.2\"\n",
     "!pip install -e git+https://github.com/axolotl-ai-cloud/axolotl#egg=axolotl\n",
     "!pip install flash-attn==\"2.5.0\"\n",
     "!pip install deepspeed==\"0.13.1\"!pip install mlflow==\"2.13.0\""

examples/deepseek-v2/fft-fsdp-16b.yaml

@@ -0,0 +1,67 @@
+base_model: deepseek-ai/DeepSeek-V2-Lite
+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: 2048
+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: adamw_torch
+lr_scheduler: cosine
+learning_rate: 2e-5
+
+train_on_inputs: false
+group_by_length: false
+bf16: auto
+fp16:
+tf32: false
+
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: false
+early_stopping_patience:
+resume_from_checkpoint:
+logging_steps: 1
+xformers_attention:
+flash_attention: true
+
+warmup_steps: 100
+evals_per_epoch: 2
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+special_tokens:
+fsdp:
+  - 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: DeepseekV2DecoderLayer
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sharding_strategy: FULL_SHARD

examples/deepseek-v2/qlora-fsdp-2_5.yaml

@@ -0,0 +1,83 @@
+base_model: axolotl-quants/DeepSeek-V2.5-bnb-nf4-bf16
+trust_remote_code: true
+
+load_in_8bit: false
+load_in_4bit: true
+strict: false
+
+
+plugins:
+  - axolotl.integrations.liger.LigerPlugin
+liger_rms_norm: true
+liger_swiglu: true
+liger_fused_linear_cross_entropy: true
+
+chat_template: deepseek_v2
+datasets:
+  - path: mlabonne/FineTome-100k
+    type: chat_template
+    split: train
+
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+adapter: qlora
+lora_r: 256
+lora_alpha: 256
+lora_target_linear: true
+peft_use_rslora: true
+
+gradient_accumulation_steps: 1
+micro_batch_size: 8
+num_epochs: 1
+optimizer: adamw_torch
+lr_scheduler: cosine
+learning_rate: 2e-5
+
+train_on_inputs: false
+group_by_length: false
+bf16: auto
+fp16:
+tf32: false
+
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: false
+early_stopping_patience:
+resume_from_checkpoint:
+logging_steps: 1
+xformers_attention:
+flash_attention: true
+
+warmup_steps: 100
+evals_per_epoch: 2
+eval_table_size:
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+special_tokens:
+fsdp:
+  - 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: DeepseekV2DecoderLayer
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sharding_strategy: FULL_SHARD

examples/jamba/README.md

@@ -6,5 +6,5 @@
 - ✅ qlora w/ deepspeed Zero-3 needs at least 2x GPUs and 67GiB VRAM (wtf?)
 - ✅ qlora single-gpu, ~51GiB VRAM
 - ✅ multipack
-- ❓ FSDP
+- ✅ FSDP
 - ❓ 8-bit LoRA

examples/jamba/qlora_fsdp_large.yaml

@@ -0,0 +1,61 @@
+base_model: ai21labs/AI21-Jamba-1.5-Large
+tokenizer_type: AutoTokenizer
+
+load_in_4bit: true
+strict: false
+use_tensorboard: true
+datasets:
+  - path: cgato/SlimOrcaDedupCleaned
+    type: chat_template
+    chat_template: jamba
+    drop_system_message: true
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: jamba-large-fsdp-qlora-ft
+save_safetensors: true
+adapter: qlora
+sequence_len: 2048
+sample_packing: true
+pad_to_sequence_len: true
+
+lora_r: 16
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules: [down_proj,gate_proj,in_proj,k_proj,o_proj,out_proj,q_proj,up_proj,v_proj,x_proj]
+lora_target_linear: false
+
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 2
+optimizer: adamw_torch
+lr_scheduler: cosine
+learning_rate: 0.00001
+
+train_on_inputs: false
+group_by_length: false
+bf16: true
+tf32: true
+
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: true
+logging_steps: 1
+flash_attention: true
+
+warmup_steps: 10
+evals_per_epoch: 1
+saves_per_epoch: 1
+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: JambaAttentionDecoderLayer,JambaMambaDecoderLayer
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sharding_strategy: FULL_SHARD

examples/llama-3-vision/lora-11b.yaml

@@ -0,0 +1,63 @@
+base_model: alpindale/Llama-3.2-11B-Vision-Instruct
+processor_type: AutoProcessor
+strict: false
+
+# these 3 lines are needed for now to handle vision chat templates w images
+skip_prepare_dataset: true
+remove_unused_columns: false
+sample_packing: false
+
+chat_template: llama3_2_vision
+datasets:
+  - path: HuggingFaceH4/llava-instruct-mix-vsft
+    type: chat_template
+    split: train[:1%]
+    field_messages: messages
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: ./outputs/out
+
+adapter: lora
+lora_model_dir:
+
+sequence_len: 8192
+pad_to_sequence_len: false
+
+lora_r: 32
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules: 'language_model.model.layers.[\d]+.(mlp|cross_attn|self_attn).(up|down|gate|q|k|v|o)_proj'
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 1
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+train_on_inputs: false
+group_by_length: false
+bf16: true
+fp16:
+tf32: true
+
+gradient_checkpointing: true
+local_rank:
+logging_steps: 1
+flash_attention: true
+eager_attention:
+
+warmup_ratio: 0.1
+evals_per_epoch: 1
+saves_per_epoch: 1
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:

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

@@ -0,0 +1,76 @@
+base_model: NousResearch/Meta-Llama-3.1-8B
+
+plugins:
+  - axolotl.integrations.liger.LigerPlugin
+liger_rope: true
+liger_rms_norm: true
+liger_swiglu: true
+liger_fused_linear_cross_entropy: true
+
+strict: false
+
+chat_template: llama3
+datasets:
+  - path: mlabonne/FineTome-100k
+    type: chat_template
+    split: train[:20%]
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.02
+output_dir: ./outputs/out
+
+sequence_len: 4096
+sample_packing: true
+pad_to_sequence_len: true
+
+wandb_project:
+wandb_entity:
+wandb_watch:
+wandb_name:
+wandb_log_model:
+
+gradient_accumulation_steps: 4
+micro_batch_size: 2
+num_epochs: 1
+optimizer: adamw_torch
+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:
+  - 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
+  fsdp_backward_prefetch: BACKWARD_PRE
+special_tokens:
+  pad_token: <|finetune_right_pad_id|>
+  eos_token: <|eot_id|>

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

@@ -1,6 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B
-model_type: LlamaForCausalLM
-tokenizer_type: AutoTokenizer
+base_model: NousResearch/Meta-Llama-3.1-8B
 
 load_in_8bit: false
 load_in_4bit: false

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

@@ -0,0 +1,81 @@
+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
+rl: dpo
+datasets:
+  - path: fozziethebeat/alpaca_messages_2k_dpo_test
+    type: chat_template.default
+    chat_template: llama3
+    field_messages: conversation
+    field_chosen: chosen
+    field_rejected: rejected
+    message_field_role: role
+    message_field_content: content
+    roles:
+      system:
+        - system
+      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/instruct-lora-8b.yml

@@ -1,4 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B-Instruct
+base_model: NousResearch/Meta-Llama-3-8B-Instruct
 model_type: LlamaForCausalLM
 tokenizer_type: AutoTokenizer
 
@@ -74,3 +74,5 @@ deepspeed:
 weight_decay: 0.0
 fsdp:
 fsdp_config:
+special_tokens:
+   pad_token: <|end_of_text|>

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

@@ -1,4 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B
+base_model: NousResearch/Meta-Llama-3-8B
 model_type: LlamaForCausalLM
 tokenizer_type: AutoTokenizer
 

examples/llama-3/qlora-fsdp-405b.yaml

@@ -0,0 +1,63 @@
+base_model: hugging-quants/Meta-Llama-3.1-405B-BNB-NF4-BF16
+tokenizer_type: AutoTokenizer
+
+load_in_4bit: true
+strict: false
+
+datasets:
+  - path: tatsu-lab/alpaca
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.0
+output_dir: ./outputs/out/qlora-llama3_1-405b
+save_safetensors: true
+
+adapter: qlora
+
+sequence_len: 2048
+sample_packing: true
+pad_to_sequence_len: true
+
+lora_r: 16
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules:
+lora_target_linear: true
+
+gradient_accumulation_steps: 4
+micro_batch_size: 1
+num_epochs: 2
+optimizer: adamw_torch
+lr_scheduler: cosine
+learning_rate: 0.00001
+
+train_on_inputs: false
+group_by_length: false
+bf16: true
+tf32: true
+
+gradient_checkpointing: true
+gradient_checkpointing_kwargs:
+  use_reentrant: true
+logging_steps: 1
+flash_attention: true
+
+warmup_steps: 10
+evals_per_epoch: 4
+saves_per_epoch: 1
+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: <|finetune_right_pad_id|>

examples/llama-3/qlora.yml

@@ -1,4 +1,4 @@
-base_model: meta-llama/Meta-Llama-3-8B
+base_model: NousResearch/Meta-Llama-3-8B
 model_type: AutoModelForCausalLM
 tokenizer_type: AutoTokenizer
 

examples/phi/lora-3.5.yaml

@@ -0,0 +1,76 @@
+base_model: microsoft/Phi-3.5-mini-instruct
+model_type: AutoModelForCausalLM
+tokenizer_type: AutoTokenizer
+
+load_in_8bit: true
+load_in_4bit: false
+strict: false
+
+chat_template: phi_3
+datasets:
+  - path: fozziethebeat/alpaca_messages_2k_test
+    type: chat_template
+    chat_template: phi_3
+    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: 4
+num_epochs: 2
+optimizer: adamw_bnb_8bit
+lr_scheduler: cosine
+learning_rate: 0.0002
+
+train_on_inputs: false
+group_by_length: false
+bfloat16: true
+bf16: true
+fp16:
+tf32: false
+
+gradient_checkpointing: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 1
+xformers_attention:
+s2_attention:
+
+warmup_steps: 10
+evals_per_epoch: 4
+eval_table_size:
+eval_max_new_tokens: 128
+saves_per_epoch: 4
+debug:
+deepspeed:
+weight_decay: 0.0
+fsdp:
+fsdp_config:

examples/qwen2/qlora-fsdp.yaml

@@ -72,4 +72,5 @@ fsdp_config:
   fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
   fsdp_transformer_layer_cls_to_wrap: Qwen2DecoderLayer
   fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sharding_strategy: FULL_SHARD
 special_tokens:

examples/tiny-llama/lora-mps.yml

@@ -1,4 +1,4 @@
-base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
+base_model: TinyLlama/TinyLlama_v1.1
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
 

examples/tiny-llama/lora.yml

@@ -1,6 +1,5 @@
-base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
-model_type: LlamaForCausalLM
-tokenizer_type: LlamaTokenizer
+base_model: TinyLlama/TinyLlama_v1.1
+tokenizer_type: AutoTokenizer
 
 load_in_8bit: true
 load_in_4bit: false

examples/tiny-llama/pretrain.yml

@@ -9,9 +9,9 @@ strict: false
 
 max_steps: 200
 pretraining_dataset:
-  path: c4
-  name: en
-  type: pretrain
+  - path: allenai/c4
+    name: en
+    type: pretrain
 dataset_prepared_path:
 val_set_size: 0.0
 output_dir: ./outputs/model-out

examples/tiny-llama/qlora.yml

@@ -1,4 +1,4 @@
-base_model: TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
+base_model: TinyLlama/TinyLlama_v1.1
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
 

requirements-tests.txt

@@ -1 +1,2 @@
 pytest
+pytest-xdist

requirements.txt

@@ -1,18 +1,18 @@
 --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/
 packaging==23.2
-peft==0.11.1
-transformers==4.42.3
-tokenizers==0.19.1
-bitsandbytes==0.43.1
-accelerate==0.32.0
-deepspeed @ git+https://github.com/microsoft/DeepSpeed.git@bc48371c5e1fb8fd70fc79285e66201dbb65679b
+peft==0.13.0
+transformers==4.45.1
+tokenizers>=0.19.1
+bitsandbytes==0.44.0
+accelerate==0.34.2
+datasets==2.21.0
+deepspeed==0.14.4
 pydantic==2.6.3
 addict
 fire
 PyYAML>=6.0
 requests
-datasets==2.19.1
-flash-attn==2.5.8
+flash-attn==2.6.3
 sentencepiece
 wandb
 einops
@@ -21,23 +21,26 @@ optimum==1.16.2
 hf_transfer
 colorama
 numba
-numpy>=1.24.4
+numpy>=1.24.4,<=2.0.1
 # qlora things
 evaluate==0.4.1
 scipy
-scikit-learn==1.2.2
+scikit-learn==1.4.2
 pynvml
 art
 fschat @ git+https://github.com/lm-sys/FastChat.git@27a05b04a35510afb1d767ae7e5990cbd278f8fe
 gradio==3.50.2
 tensorboard
 python-dotenv==1.0.1
+autoawq>=0.2.5
+triton>=2.3.0
+liger-kernel==0.3.0
 
 mamba-ssm==1.2.0.post1
 
 # remote filesystems
-s3fs
-gcsfs
+s3fs>=2024.5.0
+gcsfs>=2024.5.0
 # adlfs
 
 trl==0.9.6

setup.py

@@ -80,13 +80,13 @@ setup(
     dependency_links=dependency_links,
     extras_require={
         "flash-attn": [
-            "flash-attn==2.5.8",
+            "flash-attn==2.6.3",
         ],
         "fused-dense-lib": [
-            "fused-dense-lib  @ git+https://github.com/Dao-AILab/flash-attention@v2.5.8#subdirectory=csrc/fused_dense_lib",
+            "fused-dense-lib  @ git+https://github.com/Dao-AILab/flash-attention@v2.6.2#subdirectory=csrc/fused_dense_lib",
         ],
         "deepspeed": [
-            "deepspeed @ git+https://github.com/microsoft/DeepSpeed.git@bc48371c5e1fb8fd70fc79285e66201dbb65679b",
+            "deepspeed==0.14.4",
             "deepspeed-kernels",
         ],
         "mamba-ssm": [
@@ -104,5 +104,11 @@ setup(
         "galore": [
             "galore_torch",
         ],
+        "optimizers": [
+            "galore_torch",
+            "lion-pytorch==0.1.2",
+            "lomo-optim==0.1.1",
+            "torch-optimi==0.2.1",
+        ],
     },
 )

src/axolotl/cli/__init__.py

@@ -27,8 +27,10 @@ from transformers.utils import is_torch_bf16_gpu_available
 from transformers.utils.import_utils import _is_package_available
 
 from axolotl.common.cli import TrainerCliArgs, load_model_and_tokenizer
+from axolotl.integrations.base import PluginManager
 from axolotl.logging_config import configure_logging
 from axolotl.train import TrainDatasetMeta
+from axolotl.utils.chat_templates import chat_templates
 from axolotl.utils.config import (
     normalize_cfg_datasets,
     normalize_config,
@@ -38,9 +40,9 @@ from axolotl.utils.data import load_prepare_dpo_datasets, prepare_dataset
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.distributed import is_main_process
 from axolotl.utils.mlflow_ import setup_mlflow_env_vars
-from axolotl.utils.models import load_tokenizer
+from axolotl.utils.models import load_processor, load_tokenizer
 from axolotl.utils.tokenization import check_dataset_labels
-from axolotl.utils.trainer import prepare_optim_env
+from axolotl.utils.trainer import prepare_opinionated_env, prepare_optim_env
 from axolotl.utils.wandb_ import setup_wandb_env_vars
 
 project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
@@ -233,7 +235,8 @@ def do_inference_gradio(
 
     model, tokenizer = load_model_and_tokenizer(cfg=cfg, cli_args=cli_args)
     prompter = cli_args.prompter
-    default_tokens = {"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>"}
+    # default_tokens = {"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>"}
+    default_tokens: Dict[str, str] = {}
 
     for token, symbol in default_tokens.items():
         # If the token isn't already specified in the config, add it
@@ -241,10 +244,13 @@ def do_inference_gradio(
             tokenizer.add_special_tokens({token: symbol})
 
     prompter_module = None
+    chat_template_str = None
     if prompter:
         prompter_module = getattr(
             importlib.import_module("axolotl.prompters"), prompter
         )
+    elif cfg.chat_template:
+        chat_template_str = chat_templates(cfg.chat_template)
 
     model = model.to(cfg.device, dtype=cfg.torch_dtype)
 
@@ -258,7 +264,24 @@ def do_inference_gradio(
             )
         else:
             prompt = instruction.strip()
-        batch = tokenizer(prompt, return_tensors="pt", add_special_tokens=True)
+
+        if chat_template_str:
+            batch = tokenizer.apply_chat_template(
+                [
+                    {
+                        "role": "user",
+                        "content": prompt,
+                    }
+                ],
+                return_tensors="pt",
+                add_special_tokens=True,
+                add_generation_prompt=True,
+                chat_template=chat_template_str,
+                tokenize=True,
+                return_dict=True,
+            )
+        else:
+            batch = tokenizer(prompt, return_tensors="pt", add_special_tokens=True)
 
         model.eval()
         with torch.no_grad():
@@ -281,6 +304,7 @@ def do_inference_gradio(
             streamer = TextIteratorStreamer(tokenizer)
             generation_kwargs = {
                 "inputs": batch["input_ids"].to(cfg.device),
+                "attention_mask": batch["attention_mask"].to(cfg.device),
                 "generation_config": generation_config,
                 "streamer": streamer,
             }
@@ -365,6 +389,11 @@ def load_cfg(config: Union[str, Path] = Path("examples/"), **kwargs):
 
     cfg.axolotl_config_path = config
 
+    if cfg.get("plugins"):
+        plugin_manager = PluginManager.get_instance()
+        for plugin_name in cfg["plugins"]:
+            plugin_manager.register(plugin_name)
+
     try:
         device_props = torch.cuda.get_device_properties("cuda")
         gpu_version = "sm_" + str(device_props.major) + str(device_props.minor)
@@ -375,13 +404,15 @@ def load_cfg(config: Union[str, Path] = Path("examples/"), **kwargs):
         cfg,
         capabilities={
             "bf16": is_torch_bf16_gpu_available(),
-            "n_gpu": os.environ.get("WORLD_SIZE", 1),
+            "n_gpu": int(os.environ.get("WORLD_SIZE", 1)),
             "compute_capability": gpu_version,
         },
     )
 
     prepare_optim_env(cfg)
 
+    prepare_opinionated_env(cfg)
+
     normalize_config(cfg)
 
     normalize_cfg_datasets(cfg)
@@ -399,9 +430,12 @@ def load_datasets(
     cli_args: TrainerCliArgs,
 ) -> TrainDatasetMeta:
     tokenizer = load_tokenizer(cfg)
+    processor = load_processor(cfg, tokenizer=tokenizer) if cfg.processor_type else None
 
     train_dataset, eval_dataset, total_num_steps, prompters = prepare_dataset(
-        cfg, tokenizer
+        cfg,
+        tokenizer,
+        processor=processor,
     )
 
     if cli_args.debug or cfg.debug:

src/axolotl/cli/merge_sharded_fsdp_weights.py

@@ -0,0 +1,204 @@
+"""
+This module provides a CLI to merge sharded FSDP model checkpoints into a single combined checkpoint
+"""
+import json
+import logging
+import os
+import shutil
+from pathlib import Path
+from typing import Dict, Union
+
+import fire
+import torch
+import torch.distributed.checkpoint as dist_cp
+import torch.distributed.checkpoint.format_utils as dist_cp_format_utils
+import transformers
+from accelerate.utils import (
+    SAFE_WEIGHTS_INDEX_NAME,
+    SAFE_WEIGHTS_NAME,
+    WEIGHTS_INDEX_NAME,
+    WEIGHTS_NAME,
+    is_torch_version,
+)
+from dotenv import load_dotenv
+from huggingface_hub import split_torch_state_dict_into_shards
+from safetensors.torch import save_file as safe_save_file
+from torch.distributed.checkpoint.format_utils import _EmptyStateDictLoadPlanner
+
+from axolotl.cli import load_cfg, print_axolotl_text_art
+from axolotl.common.cli import TrainerCliArgs
+
+LOG = logging.getLogger("axolotl.cli.merge_sharded_fsdp_weights")
+
+
+class BFloat16CastPlanner(_EmptyStateDictLoadPlanner):
+    """
+    A custom planner to cast tensors to bfloat16 on the fly during loading.
+    """
+
+    def commit_tensor(self, read_item, tensor):  # pylint: disable=unused-argument
+        tensor.copy_(tensor.to(torch.bfloat16))
+
+
+def _distributed_checkpoint_to_merged_weights(
+    checkpoint_dir: Union[str, Path],
+    save_path: str,
+    safe_serialization: bool = False,
+    max_shard_size: str = "5GB",
+):
+    """
+    Passthrough to `torch.distributed.checkpoint.format_utils.dcp_to_torch_save`
+
+    Will save under `save_path` as either `model.safetensors` or `pytorch_model.bin`.
+    """
+
+    state_dict: Dict = {}
+    save_path_ = Path(save_path)
+    save_path_.mkdir(exist_ok=True)
+    dist_cp_format_utils._load_state_dict(  # pylint: disable=protected-access
+        state_dict,
+        storage_reader=dist_cp.FileSystemReader(checkpoint_dir),
+        planner=BFloat16CastPlanner(),  # pylint: disable=protected-access
+        no_dist=True,
+    )
+
+    # To handle if state is a dict like {model: {...}}
+    if len(state_dict.keys()) == 1:
+        state_dict = state_dict[list(state_dict)[0]]
+
+    # Ensure all tensors are in bfloat16
+    for key, value in state_dict.items():
+        if isinstance(value, torch.Tensor) and value.dtype != torch.bfloat16:
+            state_dict[key] = value.to(torch.bfloat16)
+
+    weights_name = SAFE_WEIGHTS_NAME if safe_serialization else WEIGHTS_NAME
+
+    filename_pattern = weights_name.replace(".bin", "{suffix}.bin").replace(
+        ".safetensors", "{suffix}.safetensors"
+    )
+    state_dict_split = split_torch_state_dict_into_shards(
+        state_dict, filename_pattern=filename_pattern, max_shard_size=max_shard_size
+    )
+    # Save index if sharded
+    index = None
+    if state_dict_split.is_sharded:
+        index = {
+            "metadata": state_dict_split.metadata,
+            "weight_map": state_dict_split.tensor_to_filename,
+        }
+
+    # Save the model
+    filename_to_tensors = state_dict_split.filename_to_tensors.items()
+
+    for shard_file, tensors in filename_to_tensors:
+        shard = {tensor: state_dict[tensor] for tensor in tensors}
+
+        if safe_serialization:
+            safe_save_file(
+                shard, os.path.join(save_path_, shard_file), metadata={"format": "pt"}
+            )
+        else:
+            torch.save(shard, os.path.join(save_path_, shard_file))
+
+    if index is not None:
+        save_index_file = (
+            SAFE_WEIGHTS_INDEX_NAME if safe_serialization else WEIGHTS_INDEX_NAME
+        )
+        save_index_file = os.path.join(save_path_, save_index_file)
+        # Save the index as well
+        with open(save_index_file, "w", encoding="utf-8") as fout:
+            content = json.dumps(index, indent=2, sort_keys=True) + "\n"
+            fout.write(content)
+
+    return save_path_
+
+
+def merge_fsdp_weights(
+    checkpoint_dir: str,
+    output_path: str,
+    safe_serialization: bool = False,
+    remove_checkpoint_dir: bool = False,
+):
+    """
+    Merge the weights from sharded FSDP model checkpoints into a single combined checkpoint. Should be used if
+    `SHARDED_STATE_DICT` was used for the model. Weights will be saved to `{output_path}/model.safetensors` if
+    `safe_serialization` else `pytorch_model.bin`.
+
+    Note: this is a CPU-bound process.
+
+    Args:
+        checkpoint_dir (`str`):
+            The directory containing the FSDP checkpoints (can be either the model or optimizer).
+        output_path (`str`):
+            The path to save the merged checkpoint.
+        safe_serialization (`bool`, *optional*, defaults to `True`):
+            Whether to save the merged weights with safetensors (recommended).
+        remove_checkpoint_dir (`bool`, *optional*, defaults to `False`):
+            Whether to remove the checkpoint directory after merging.
+    """
+    checkpoint_dir_ = Path(checkpoint_dir)
+    from accelerate.state import PartialState
+
+    if not is_torch_version(">=", "2.3.0"):
+        raise ValueError("`merge_fsdp_weights` requires PyTorch >= 2.3.0`")
+
+    # Verify that the checkpoint directory exists
+    if not checkpoint_dir_.exists():
+        model_path_exists = (checkpoint_dir_ / "pytorch_model_fsdp_0").exists()
+        optimizer_path_exists = (checkpoint_dir_ / "optimizer_0").exists()
+        err = f"Tried to load from {checkpoint_dir_} but couldn't find a valid metadata file."
+        if model_path_exists and optimizer_path_exists:
+            err += (
+                " However, potential model and optimizer checkpoint directories exist."
+            )
+            err += f"Please pass in either {checkpoint_dir_}/pytorch_model_fsdp_0 or {checkpoint_dir_}/optimizer_0"
+            err += "instead."
+        elif model_path_exists:
+            err += " However, a potential model checkpoint directory exists."
+            err += (
+                f"Please try passing in {checkpoint_dir_}/pytorch_model_fsdp_0 instead."
+            )
+        elif optimizer_path_exists:
+            err += " However, a potential optimizer checkpoint directory exists."
+            err += f"Please try passing in {checkpoint_dir_}/optimizer_0 instead."
+        raise ValueError(err)
+
+    # To setup `save` to work
+    state = PartialState()
+    if state.is_main_process:
+        LOG.info(f"Merging FSDP weights from {checkpoint_dir_}")
+        save_path = _distributed_checkpoint_to_merged_weights(
+            checkpoint_dir_, output_path, safe_serialization
+        )
+        LOG.info(f"Successfully merged FSDP weights and saved to {save_path}")
+        if remove_checkpoint_dir:
+            LOG.info(f"Removing old checkpoint directory {checkpoint_dir_}")
+            shutil.rmtree(checkpoint_dir_)
+    state.wait_for_everyone()
+
+
+def do_cli(config: Path = Path("examples/"), **kwargs):
+    # pylint: disable=duplicate-code
+    print_axolotl_text_art()
+    parser = transformers.HfArgumentParser((TrainerCliArgs))
+    parsed_cli_args, _ = parser.parse_args_into_dataclasses(
+        return_remaining_strings=True
+    )
+    parsed_cli_args.merge_lora = True
+
+    parsed_cfg = load_cfg(
+        config,
+        **kwargs,
+    )
+
+    fsdp_dir = Path(parsed_cfg.output_dir) / "pytorch_model_fsdp_0"
+    merge_fsdp_weights(
+        checkpoint_dir=str(fsdp_dir),
+        output_path=str(Path(parsed_cfg.output_dir) / "merged"),
+        safe_serialization=True,
+    )
+
+
+if __name__ == "__main__":
+    load_dotenv()
+    fire.Fire(do_cli)

src/axolotl/cli/preprocess.py

@@ -2,6 +2,7 @@
 CLI to run training on a model
 """
 import logging
+import warnings
 from pathlib import Path
 from typing import Union
 
@@ -76,8 +77,19 @@ def do_cli(config: Union[Path, str] = Path("examples/"), **kwargs):
 
     if parsed_cli_args.download:
         model_name = parsed_cfg.base_model
-        with init_empty_weights():
-            AutoModelForCausalLM.from_pretrained(model_name, trust_remote_code=True)
+        with warnings.catch_warnings():
+            # there are a bunch of useless UserWarnings about
+            # "copying from a non-meta parameter in the checkpoint to a meta parameter in the current model"
+            warnings.simplefilter("ignore")
+            with init_empty_weights(include_buffers=True):
+                # fmt: off
+                try:
+                    AutoModelForCausalLM.from_pretrained(
+                        model_name, trust_remote_code=True
+                    )
+                except Exception as exc:  # pylint: disable=broad-exception-caught,unused-variable  # nosec B110  # noqa F841
+                    pass
+                # fmt: on
 
     LOG.info(
         Fore.GREEN

src/axolotl/common/architectures.py

@@ -0,0 +1,15 @@
+"""
+Common architecture specific constants
+"""
+
+MOE_ARCH_BLOCK = {
+    "dbrx": "DbrxFFN",
+    "jamba": "JambaSparseMoeBlock",
+    "jetmoe": [
+        "JetMoeMoA",
+        "JetMoeMoE",
+    ],
+    "mixtral": "MixtralSparseMoeBlock",
+    "qwen2_moe": "Qwen2MoeSparseMoeBlock",
+    "deepseek_v2": "DeepseekV2MoE",
+}

src/axolotl/core/tokenizer_utils.py

@@ -0,0 +1,150 @@
+"""
+helper functions for fixing the embeddings/tokenizer
+"""
+
+# Copyright 2023-present Daniel Han-Chen & the Unsloth team. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import gc
+import itertools
+
+import numpy as np
+import torch
+
+
+@torch.inference_mode
+def fix_untrained_tokens(model, tokenizer, train_dataset, eps=1e-16):
+    """
+    Many of the newer models have reserved tokens that are not trained.
+    """
+    embedding_matrix = model.get_input_embeddings().weight
+    lm_head_matrix = model.get_output_embeddings().weight
+
+    # Get untrained tokens
+    indicator_untrained = torch.amax(embedding_matrix, axis=1) <= eps
+    where_untrained = torch.where(indicator_untrained)[0]
+    n_untrained = where_untrained.shape[0]
+    n_trained = embedding_matrix.shape[0] - n_untrained
+
+    # Get set and actual tokens
+    where_untrained = where_untrained.tolist()
+    if len(where_untrained) == 0:
+        return False
+
+    # Remove untrained indices where it's longer
+
+    where_untrained_set = frozenset(where_untrained)
+    actual_bad_tokens = tokenizer.convert_ids_to_tokens(where_untrained)
+    # Remove None items in actual_bad_tokens
+    actual_bad_tokens = [x for x in actual_bad_tokens if x is not None]
+
+    # Check if tokenizer and training datasets have bad tokens
+    if_bad_first = False
+    if_bad_second = False
+    # Check tokenizer's chat template for any untrained tokens
+    chat_template = getattr(tokenizer, "chat_template", None)
+    if chat_template is not None:
+        if_bad_first = any(x in chat_template for x in actual_bad_tokens)
+
+    # Check the first 250, last 250 input_ids
+    size_dataset = len(train_dataset)
+    size = min(size_dataset, 250)
+    for j in range(size):
+        input_ids = train_dataset[j]
+        if "input_ids" in input_ids:
+            input_ids = input_ids["input_ids"]
+            if_bad = any(item in where_untrained_set for item in input_ids)
+            if if_bad:
+                if_bad_second = True
+                break
+
+    # Check last 250
+    if not if_bad_second:
+        left = max(size_dataset - 250, 0)
+        for j in range(left, size_dataset):
+            input_ids = train_dataset[j]
+            if "input_ids" in input_ids:
+                input_ids = input_ids["input_ids"]
+                if_bad = any(item in where_untrained_set for item in input_ids)
+                if if_bad:
+                    if_bad_second = True
+                    break
+
+    # Check if bad tokens exists!
+    if not if_bad_first and not if_bad_second:
+        return False
+
+    # Count all the possible bad tokens
+    final_counts = np.zeros(
+        max(len(tokenizer), embedding_matrix.shape[0]), dtype=np.int64
+    )
+
+    def mapping(examples):
+        input_ids = examples["input_ids"]
+        counter = np.fromiter(itertools.chain.from_iterable(input_ids), dtype=np.int32)
+        np.add.at(final_counts, counter, 1)
+
+    train_dataset.map(mapping, batched=True, desc="Counting untrained tokens")
+
+    # Get sum of all items
+    sum_embedding = torch.sum(embedding_matrix, dtype=torch.float32, axis=0)
+    sum_lm_head = torch.sum(lm_head_matrix, dtype=torch.float32, axis=0)
+
+    # Remove bad tokens
+    sum_embedding -= torch.sum(
+        embedding_matrix[where_untrained], dtype=torch.float32, axis=0
+    )
+    sum_lm_head -= torch.sum(
+        lm_head_matrix[where_untrained], dtype=torch.float32, axis=0
+    )
+
+    # Find correct average by dividing by sum of trained tokens
+    mean_embedding = sum_embedding / n_trained
+    mean_lm_head = sum_lm_head / n_trained
+
+    # Scale each to be equal to 1/max_frequency. Also set some to 0 if none seen
+    scaling = final_counts[where_untrained] / max(final_counts.max(), 1)
+    scaling = torch.tensor(scaling, device=mean_embedding.device).unsqueeze(1)
+    mean_embedding = (
+        mean_embedding.repeat(
+            (
+                n_untrained,
+                1,
+            )
+        )
+        * scaling
+    )
+    mean_lm_head = (
+        mean_lm_head.repeat(
+            (
+                n_untrained,
+                1,
+            )
+        )
+        * scaling
+    )
+    where_null = scaling.ravel() == 0
+    mean_embedding[where_null] = 0
+    mean_lm_head[where_null] = 0
+
+    # Set them to the mean
+    embedding_matrix[where_untrained] = mean_embedding.to(embedding_matrix.dtype)
+    lm_head_matrix[where_untrained] = mean_lm_head.to(lm_head_matrix.dtype)
+
+    # Clean up
+    for _ in range(3):
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    return True

src/axolotl/core/trainer_builder.py

@@ -4,21 +4,25 @@ Builder for the training args and trainer
 """
 
 import abc
+import gc
 import importlib
 import importlib.util
 import logging
 import math
+import os
 import sys
 from abc import abstractmethod
 from collections import defaultdict
 from dataclasses import dataclass, field
 from functools import wraps
 from pathlib import Path
-from typing import Dict, List, Literal, Optional, Type, Union
+from typing import Any, Dict, List, Literal, Optional, Type, Union
 
 import torch
 import transformers
 from datasets import Dataset
+from peft.optimizers import create_loraplus_optimizer
+from torch import nn
 from torch.optim.lr_scheduler import OneCycleLR
 from torch.utils.data import BatchSampler, DataLoader, RandomSampler, SequentialSampler
 from transformers import (
@@ -28,12 +32,20 @@ from transformers import (
     TrainerCallback,
     TrainingArguments,
 )
-from transformers.trainer_utils import seed_worker
+from transformers.trainer_utils import PREFIX_CHECKPOINT_DIR, seed_worker
 from transformers.utils import is_sagemaker_mp_enabled
-from trl import DPOConfig, DPOTrainer, KTOConfig, KTOTrainer, ORPOConfig, ORPOTrainer
+from trl import (
+    CPOConfig,
+    CPOTrainer,
+    DPOConfig,
+    DPOTrainer,
+    KTOConfig,
+    KTOTrainer,
+    ORPOConfig,
+    ORPOTrainer,
+)
 from trl.trainer.utils import pad_to_length
 
-from axolotl.loraplus import create_loraplus_optimizer
 from axolotl.monkeypatch.multipack import SUPPORTED_MULTIPACK_MODEL_TYPES
 from axolotl.monkeypatch.relora import ReLoRACallback, ReLoRAScheduler
 from axolotl.utils import is_mlflow_available
@@ -49,12 +61,14 @@ from axolotl.utils.callbacks import (
     log_prediction_callback_factory,
 )
 from axolotl.utils.callbacks.lisa import lisa_callback_factory
+from axolotl.utils.chat_templates import chat_templates
 from axolotl.utils.collators import (
     BatchSamplerDataCollatorForSeq2Seq,
     DataCollatorForSeq2Seq,
     MambaDataCollator,
     V2BatchSamplerDataCollatorForSeq2Seq,
 )
+from axolotl.utils.collators.mm_chat import MultiModalChatDataCollator
 from axolotl.utils.models import ensure_dtype
 from axolotl.utils.samplers import MultipackBatchSampler, get_dataset_lengths
 from axolotl.utils.schedulers import (
@@ -226,6 +240,22 @@ class AxolotlTrainingMixins:
         default=None,
         metadata={"help": "whether to use sequential sampling for curriculum learning"},
     )
+    alternate_optimizer: Optional[str] = field(
+        default=None,
+        metadata={
+            "help": "workaround to pass an alternate optimizer to the HF trainer"
+        },
+    )
+    alternate_lr_scheduler_type: Optional[str] = field(
+        default=None,
+        metadata={
+            "help": "workaround to pass an alternate lr scheduler to the HF trainer"
+        },
+    )
+    chat_template: Optional[str] = field(
+        default=None,
+        metadata={"help": "Chat template converting chat messages to text"},
+    )
 
 
 @dataclass
@@ -259,58 +289,24 @@ class AxolotlKTOConfig(AxolotlTrainingMixins, KTOConfig):
     """
 
 
-class AxolotlTrainer(Trainer):
+@dataclass
+class AxolotlCPOConfig(AxolotlTrainingMixins, CPOConfig):
     """
-    Extend the base Trainer for axolotl helpers
+    CPO config for CPO training
+    """
+
+    simpo_gamma: Optional[float] = field(
+        default=None,
+        metadata={"help": "simpo gamma parameter"},
+    )
+
+
+class SchedulerMixin(Trainer):
+    """
+    Mixin class for scheduler setup in CausalTrainer.
     """
 
     args = None  # type: AxolotlTrainingArguments
-    tag_names = ["axolotl"]
-
-    def __init__(
-        self,
-        *_args,
-        num_epochs=1,
-        bench_data_collator=None,
-        eval_data_collator=None,
-        **kwargs,
-    ):
-        self.num_epochs = num_epochs
-        self.bench_data_collator = bench_data_collator
-        self.eval_data_collator = eval_data_collator
-        super().__init__(*_args, **kwargs)
-        self.train_data_collator = self.data_collator
-        self._stored_metrics = defaultdict(lambda: defaultdict(list))
-        if self.args.orpo_alpha:
-            self.loss_fct = torch.nn.CrossEntropyLoss(reduction="none")
-
-    def create_optimizer(self):
-        if self.args.loraplus_lr_ratio is None:
-            return super().create_optimizer()
-
-        opt_model = self.model_wrapped if is_sagemaker_mp_enabled() else self.model
-        if self.optimizer is None:  # pylint: disable=access-member-before-definition
-            optimizer_cls, optimizer_kwargs = Trainer.get_optimizer_cls_and_kwargs(
-                self.args,
-                opt_model,
-            )
-
-            loraplus_lr_ratio = getattr(self.args, "loraplus_lr_ratio", None)
-            loraplus_lr_embedding = getattr(self.args, "loraplus_lr_embedding", None)
-            self.optimizer = create_loraplus_optimizer(  # pylint: disable=attribute-defined-outside-init
-                opt_model,
-                optimizer_cls,
-                optimizer_kwargs,
-                loraplus_lr_ratio,
-                loraplus_lr_embedding,
-            )
-
-        if is_sagemaker_mp_enabled():
-            self.optimizer = smp.DistributedOptimizer(  # pylint: disable=attribute-defined-outside-init
-                self.optimizer
-            )
-
-        return self.optimizer
 
     def create_scheduler(
         self, num_training_steps: int, optimizer: torch.optim.Optimizer = None
@@ -336,7 +332,23 @@ class AxolotlTrainer(Trainer):
         # fmt: off
         if self.lr_scheduler is None:  # type: ignore  # pylint: disable=access-member-before-definition
             # fmt: on
-            if use_cosine_quadratic:
+            if self.args.alternate_lr_scheduler_type == "one_cycle":
+                num_warmup_steps = self.args.get_warmup_steps(num_training_steps)
+                pct_start = num_warmup_steps / num_training_steps
+                extra_lr_kwargs = {}
+                if "pct_start" not in self.args.lr_scheduler_kwargs:
+                    extra_lr_kwargs["pct_start"] = pct_start
+                if "anneal_strategy" not in self.args.lr_scheduler_kwargs:
+                    extra_lr_kwargs["anneal_strategy"] = "cos"
+
+                self.lr_scheduler = OneCycleLR(
+                    optimizer,
+                    max_lr=self.args.learning_rate,
+                    total_steps=num_training_steps,
+                    **extra_lr_kwargs,
+                    **self.args.lr_scheduler_kwargs,
+                )
+            elif use_cosine_quadratic:
                 if use_cosine_min_lr:
                     LOG.warning("Both cosine quadratic warmup and min lr detected. Using quadratic warmup.")
 
@@ -374,6 +386,125 @@ class AxolotlTrainer(Trainer):
 
         return self.lr_scheduler
 
+
+class AxolotlTrainer(SchedulerMixin, Trainer):
+    """
+    Extend the base Trainer for axolotl helpers
+    """
+
+    args = None  # type: AxolotlTrainingArguments
+    tag_names = ["axolotl"]
+
+    def __init__(
+        self,
+        *_args,
+        num_epochs=1,
+        bench_data_collator=None,
+        eval_data_collator=None,
+        **kwargs,
+    ):
+        self.num_epochs = num_epochs
+        self.bench_data_collator = bench_data_collator
+        self.eval_data_collator = eval_data_collator
+        super().__init__(*_args, **kwargs)
+        self.train_data_collator = self.data_collator
+        self._stored_metrics = defaultdict(lambda: defaultdict(list))
+        if self.args.orpo_alpha:
+            self.loss_fct = torch.nn.CrossEntropyLoss(reduction="none")
+
+    def _wrap_model(self, model, training=True, dataloader=None):
+        if self.args.torch_compile:
+            torch._dynamo.config.accumulated_cache_size_limit = (  # pylint: disable=protected-access
+                256
+            )
+            model = torch.compile(
+                model,
+                backend=self.args.torch_compile_backend,
+                mode=self.args.torch_compile_mode,
+            )
+        return super()._wrap_model(model, training=training, dataloader=dataloader)
+
+    def create_optimizer(self):
+        if (
+            self.args.loraplus_lr_ratio is None
+            and self.args.alternate_optimizer
+            not in ["optimi_adamw", "ao_adamw_8bit", "ao_adamw_4bit", "ao_adamw_fp8"]
+        ):
+            return super().create_optimizer()
+
+        opt_model = self.model_wrapped if is_sagemaker_mp_enabled() else self.model
+        if self.optimizer is None:  # pylint: disable=access-member-before-definition
+            decay_parameters = self.get_decay_parameter_names(opt_model)
+            optimizer_grouped_parameters = [
+                {
+                    "params": [
+                        p
+                        for n, p in opt_model.named_parameters()
+                        if (n in decay_parameters and p.requires_grad)
+                    ],
+                    "weight_decay": self.args.weight_decay,
+                },
+                {
+                    "params": [
+                        p
+                        for n, p in opt_model.named_parameters()
+                        if (n not in decay_parameters and p.requires_grad)
+                    ],
+                    "weight_decay": 0.0,
+                },
+            ]
+
+            optimizer_cls, optimizer_kwargs = Trainer.get_optimizer_cls_and_kwargs(
+                self.args,
+                opt_model,
+            )
+
+            if self.args.loraplus_lr_ratio is not None:
+                loraplus_lr_ratio = getattr(self.args, "loraplus_lr_ratio", None)
+                loraplus_lr_embedding = getattr(
+                    self.args, "loraplus_lr_embedding", 1e-6
+                )
+                self.optimizer = create_loraplus_optimizer(  # pylint: disable=attribute-defined-outside-init
+                    opt_model,
+                    optimizer_cls,
+                    loraplus_lr_ratio=loraplus_lr_ratio,
+                    loraplus_lr_embedding=loraplus_lr_embedding,
+                    **optimizer_kwargs,
+                )
+            elif self.args.alternate_optimizer == "optimi_adamw":
+                from optimi import AdamW
+
+                self.optimizer = (  # pylint: disable=attribute-defined-outside-init
+                    AdamW(
+                        optimizer_grouped_parameters, foreach=False, **optimizer_kwargs
+                    )
+                )
+            elif self.args.alternate_optimizer == "ao_adamw_4bit":
+                from torchao.prototype.low_bit_optim import AdamW4bit
+
+                self.optimizer = (  # pylint: disable=attribute-defined-outside-init
+                    AdamW4bit(optimizer_grouped_parameters, **optimizer_kwargs)
+                )
+            elif self.args.alternate_optimizer == "ao_adamw_8bit":
+                from torchao.prototype.low_bit_optim import AdamW8bit
+
+                self.optimizer = (  # pylint: disable=attribute-defined-outside-init
+                    AdamW8bit(optimizer_grouped_parameters, **optimizer_kwargs)
+                )
+            elif self.args.alternate_optimizer == "ao_adamw_fp8":
+                from torchao.prototype.low_bit_optim import AdamWFp8
+
+                self.optimizer = (  # pylint: disable=attribute-defined-outside-init
+                    AdamWFp8(optimizer_grouped_parameters, **optimizer_kwargs)
+                )
+
+        if is_sagemaker_mp_enabled():
+            self.optimizer = smp.DistributedOptimizer(  # pylint: disable=attribute-defined-outside-init
+                self.optimizer
+            )
+
+        return self.optimizer
+
     def _get_train_sampler(self) -> Optional[torch.utils.data.Sampler]:
         if self.args.sample_packing and not self.args.pretraining:
             if self.args.multipack_real_batches:
@@ -381,9 +512,10 @@ class AxolotlTrainer(Trainer):
                 batch_max_len = self.args.max_seq_length
             else:
                 batch_size = 1
-                batch_max_len = (
-                    self.args.per_device_train_batch_size * self.args.max_seq_length
+                train_batch_size = (
+                    self.state.train_batch_size or self.args.per_device_train_batch_size
                 )
+                batch_max_len = train_batch_size * self.args.max_seq_length
             return MultipackBatchSampler(
                 RandomSampler(self.train_dataset),
                 lengths=get_dataset_lengths(self.train_dataset),
@@ -738,6 +870,14 @@ class AxolotlTrainer(Trainer):
         for key, value in metrics.items():
             self._stored_metrics[train_eval][key].append(value)
 
+    def _save_checkpoint(self, model, trial, metrics=None):
+        # make sure the checkpoint dir exists, since trainer is flakey
+        checkpoint_folder = f"{PREFIX_CHECKPOINT_DIR}-{self.state.global_step}"
+        run_dir = self._get_output_dir(trial=trial)
+        output_dir = os.path.join(run_dir, checkpoint_folder)
+        os.makedirs(output_dir, exist_ok=True)
+        return super()._save_checkpoint(model, trial, metrics=metrics)
+
 
 class AxolotlMambaTrainer(AxolotlTrainer):
     """
@@ -767,37 +907,6 @@ class AxolotlMambaTrainer(AxolotlTrainer):
         return lm_loss
 
 
-class OneCycleLRSchedulerTrainer(AxolotlTrainer):
-    """
-    Trainer subclass that uses the OneCycleLR scheduler
-    """
-
-    tag_names = ["axolotl", "onecycle"]
-
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
-        self.lr_scheduler = None
-
-    def create_scheduler(
-        self,
-        num_training_steps: int,
-        optimizer: Optional[torch.optim.Optimizer] = None,
-    ):
-        optimizer = self.optimizer if optimizer is None else optimizer
-        num_warmup_steps = self.args.get_warmup_steps(num_training_steps)
-        pct_start = num_warmup_steps / num_training_steps
-
-        self.lr_scheduler = OneCycleLR(
-            optimizer,
-            max_lr=self.args.learning_rate,
-            total_steps=num_training_steps,
-            pct_start=pct_start,
-            div_factor=6,
-        )
-
-        return self.lr_scheduler
-
-
 class ReLoRATrainer(AxolotlTrainer):
     """
     Trainer subclass that uses the OneCycleLR scheduler
@@ -837,7 +946,7 @@ class ReLoRATrainer(AxolotlTrainer):
         return self.lr_scheduler
 
 
-class AxolotlDPOTrainer(DPOTrainer):
+class AxolotlDPOTrainer(SchedulerMixin, DPOTrainer):
     """
     Extend the base DPOTrainer for axolotl helpers
     """
@@ -866,9 +975,9 @@ class AxolotlDPOTrainer(DPOTrainer):
             self.optimizer = create_loraplus_optimizer(  # pylint: disable=attribute-defined-outside-init
                 opt_model,
                 optimizer_cls,
-                optimizer_kwargs,
-                loraplus_lr_ratio,
-                loraplus_lr_embedding,
+                loraplus_lr_ratio=loraplus_lr_ratio,
+                loraplus_lr_embedding=loraplus_lr_embedding,
+                **optimizer_kwargs,
             )
 
         if is_sagemaker_mp_enabled():
@@ -897,8 +1006,16 @@ class AxolotlDPOTrainer(DPOTrainer):
                 res[key] = res[key][1:]
         return res
 
+    def training_step(
+        self, model: nn.Module, inputs: Dict[str, Union[torch.Tensor, Any]]
+    ) -> torch.Tensor:
+        loss: torch.Tensor = super().training_step(model, inputs)
+        gc.collect()
+        torch.cuda.empty_cache()
+        return loss
 
-class AxolotlORPOTrainer(ORPOTrainer):
+
+class AxolotlORPOTrainer(SchedulerMixin, ORPOTrainer):
     """
     Extend the base ORPOTrainer for axolotl helpers
     """
@@ -906,7 +1023,7 @@ class AxolotlORPOTrainer(ORPOTrainer):
     tag_names = ["axolotl", "orpo"]
 
 
-class AxolotlKTOTrainer(KTOTrainer):
+class AxolotlKTOTrainer(SchedulerMixin, KTOTrainer):
     """
     Extend the base KTOTrainer for axolotl helpers
     """
@@ -914,6 +1031,14 @@ class AxolotlKTOTrainer(KTOTrainer):
     tag_names = ["axolotl", "kto"]
 
 
+class AxolotlCPOTrainer(SchedulerMixin, CPOTrainer):
+    """
+    Extend the base CPOTrainer for axolotl helpers
+    """
+
+    tag_names = ["axolotl", "cpo"]
+
+
 class TrainerBuilderBase(abc.ABC):
     """
     Base class for trainer builder
@@ -924,10 +1049,11 @@ class TrainerBuilderBase(abc.ABC):
     _model_ref = None
     _peft_config = None
 
-    def __init__(self, cfg, model, tokenizer):
+    def __init__(self, cfg, model, tokenizer, processor=None):
         self.cfg = cfg
         self.model = model
         self.tokenizer = tokenizer
+        self.processor = processor
 
         # in case the model supports tagging, add the axolotl tag.
         # This makes sure the tag is correctly pushed even if a user calls
@@ -1073,10 +1199,6 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         return callbacks
 
     def _get_trainer_cls(self):
-        if self.cfg.lr_scheduler == "one_cycle" and (
-            self.cfg.fsdp or self.cfg.adapter == "qlora"
-        ):
-            return OneCycleLRSchedulerTrainer
         if self.cfg.relora_steps:
             return ReLoRATrainer
         if self.cfg.model_config_type == "mamba":
@@ -1126,7 +1248,9 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         if self.cfg.fsdp:
             training_arguments_kwargs["fsdp"] = self.cfg.fsdp
             if self.cfg.fsdp_config:
-                training_arguments_kwargs["fsdp_config"] = dict(self.cfg.fsdp_config)
+                training_arguments_kwargs["fsdp_config"] = {
+                    k.lstrip("fsdp_"): v for k, v in dict(self.cfg.fsdp_config).items()
+                }
 
         if self.cfg.adapter == "qlora":
             training_arguments_kwargs["qlora"] = True
@@ -1235,6 +1359,10 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
                     training_arguments_kwargs[
                         "torch_compile_backend"
                     ] = self.cfg.torch_compile_backend
+                if self.cfg.torch_compile_mode:
+                    training_arguments_kwargs[
+                        "torch_compile_mode"
+                    ] = self.cfg.torch_compile_mode
 
         # DDP Config
         if self.cfg.ddp_timeout:
@@ -1259,6 +1387,10 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
             training_arguments_kwargs[
                 "per_device_eval_batch_size"
             ] = self.cfg.eval_batch_size
+        if self.cfg.auto_find_batch_size is not None:
+            training_arguments_kwargs[
+                "auto_find_batch_size"
+            ] = self.cfg.auto_find_batch_size
         training_arguments_kwargs[
             "gradient_accumulation_steps"
         ] = self.cfg.gradient_accumulation_steps
@@ -1292,6 +1424,8 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         report_to = []
         if self.cfg.use_wandb:
             report_to.append("wandb")
+            if self.cfg.wandb_name:
+                training_arguments_kwargs["run_name"] = self.cfg.wandb_name
         if self.cfg.use_mlflow:
             report_to.append("mlflow")
         if self.cfg.use_tensorboard:
@@ -1320,12 +1454,15 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         training_arguments_kwargs[
             "loraplus_lr_embedding"
         ] = self.cfg.loraplus_lr_embedding
-        training_arguments_kwargs["lr_scheduler_type"] = (
-            self.cfg.lr_scheduler
-            if self.cfg.lr_scheduler
-            and self.cfg.lr_scheduler not in ("one_cycle", "log_sweep")
-            else "cosine"
-        )
+        if self.cfg.lr_scheduler in ["one_cycle", "log_sweep"]:
+            training_arguments_kwargs["lr_scheduler_type"] = "cosine"
+            training_arguments_kwargs[
+                "alternate_lr_scheduler_type"
+            ] = self.cfg.lr_scheduler
+        else:
+            training_arguments_kwargs["lr_scheduler_type"] = (
+                self.cfg.lr_scheduler if self.cfg.lr_scheduler else "cosine"
+            )
         training_arguments_kwargs["lr_scheduler_kwargs"] = (
             self.cfg.lr_scheduler_kwargs if self.cfg.lr_scheduler_kwargs else {}
         )
@@ -1338,9 +1475,9 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         )
 
         training_arguments_kwargs["sample_packing"] = bool(self.cfg.sample_packing)
-        training_arguments_kwargs[
-            "multipack_real_batches"
-        ] = not self.cfg.flash_attention
+        training_arguments_kwargs["multipack_real_batches"] = (
+            not self.cfg.flash_attention or self.cfg.multipack_real_batches
+        )
         training_arguments_kwargs["eval_sample_packing"] = bool(
             self.cfg.eval_sample_packing
         )
@@ -1385,6 +1522,10 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         )
         training_arguments_kwargs["model_type"] = self.cfg.model_config_type
         training_arguments_kwargs["pretraining"] = bool(self.cfg.pretraining_dataset)
+        if self.cfg.chat_template:
+            training_arguments_kwargs["chat_template"] = chat_templates(
+                self.cfg.chat_template
+            )
 
         if self.cfg.rl == "orpo":
             training_arguments_kwargs["orpo_alpha"] = self.cfg.orpo_alpha
@@ -1396,6 +1537,16 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
 
         trainer_kwargs = {}
 
+        if self.cfg.optimizer in [
+            "optimi_adamw",
+            "ao_adamw_4bit",
+            "ao_adamw_8bit",
+            "ao_adamw_fp8",
+        ]:
+            # Set default so transformers doesn't throw
+            training_arguments_kwargs["optim"] = "adamw_hf"
+            training_arguments_kwargs["alternate_optimizer"] = self.cfg.optimizer
+
         if self.cfg.optimizer == "lion_pytorch":
             from lion_pytorch import Lion
 
@@ -1424,6 +1575,11 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
                 sys.path.append(self.cfg.torchdistx_path)
                 importlib.import_module("torchdistx")
 
+        if self.cfg.accelerator_config:
+            training_arguments_kwargs[
+                "accelerator_config"
+            ] = self.cfg.accelerator_config
+
         training_args = (
             AxolotlTrainingArguments(  # pylint: disable=unexpected-keyword-arg
                 **training_arguments_kwargs,
@@ -1431,6 +1587,12 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
         )
         training_args = self.hook_post_create_training_args(training_args)
 
+        # unset run_name so wandb sets up experiment names
+        if self.cfg.use_wandb and training_args.run_name == training_args.output_dir:
+            training_args.run_name = (  # pylint: disable=attribute-defined-outside-init
+                None
+            )
+
         data_collator_kwargs = {
             "padding": True,  # True/"longest" is the default
         }
@@ -1510,7 +1672,12 @@ class HFCausalTrainerBuilder(TrainerBuilderBase):
             else:
                 collator = BatchSamplerDataCollatorForSeq2Seq
         else:
-            collator = DataCollatorForSeq2Seq
+            if self.cfg.processor_type and self.processor:
+                collator = MultiModalChatDataCollator
+                kwargs["processor"] = self.processor
+                kwargs["chat_template"] = training_args.chat_template
+            else:
+                collator = DataCollatorForSeq2Seq
 
         return collator(
             self.tokenizer,
@@ -1617,16 +1784,27 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             # default to saving each epoch if not defined
             training_args_kwargs["save_strategy"] = "epoch"
 
+        if self.cfg.rl_beta:
+            training_args_kwargs["beta"] = self.cfg.rl_beta
         if self.cfg.orpo_alpha:
             # trl does some odd mapping of alpha to beta to reuse the beta parameter ???
             training_args_kwargs["beta"] = self.cfg.orpo_alpha
 
+        training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
         training_args_cls = AxolotlDPOConfig
         if self.cfg.rpo_alpha is not None:
             training_args_kwargs["rpo_alpha"] = self.cfg.rpo_alpha
+
+        if self.cfg.rl == "simpo":
+            training_args_cls = AxolotlCPOConfig
+            training_args_kwargs["loss_type"] = "simpo"
+            training_args_kwargs["max_length"] = self.cfg.sequence_len
+            training_args_kwargs["simpo_gamma"] = self.cfg.simpo_gamma
+            if self.cfg.cpo_alpha is not None:
+                training_args_kwargs["cpo_alpha"] = self.cfg.cpo_alpha
+
         if self.cfg.rl == "orpo":
             training_args_cls = AxolotlORPOConfig
-            training_args_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
             training_args_kwargs["max_length"] = self.cfg.sequence_len
             if self.cfg.max_prompt_len:
                 training_args_kwargs["max_prompt_length"] = self.cfg.max_prompt_len
@@ -1634,7 +1812,6 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
         if self.cfg.rl == "kto":
             training_args_cls = AxolotlKTOConfig
 
-            training_args_kwargs["beta"] = self.cfg.rl_beta or 0.1
             training_args_kwargs["desirable_weight"] = (
                 self.cfg.kto_desirable_weight or 1.0
             )
@@ -1680,7 +1857,6 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             ] = self.cfg.precompute_ref_log_probs
         if self.cfg.rl in ["dpo", "ipo"]:
             trainer_cls = AxolotlDPOTrainer
-            dpo_trainer_kwargs["beta"] = self.cfg.rl_beta or 0.1
             trainer_cls_args = [self.model, self.model_ref]
 
             # these aren't used for the ORPO trainer
@@ -1688,14 +1864,15 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
             dpo_trainer_kwargs["max_target_length"] = None
             dpo_trainer_kwargs["max_prompt_length"] = self.cfg.sequence_len
             dpo_trainer_kwargs["generate_during_eval"] = True
-            if self.cfg.rl == "dpo":
-                dpo_trainer_kwargs["dataset_num_proc"] = self.cfg.dataset_processes
         elif self.cfg.rl == "orpo":
             trainer_cls = AxolotlORPOTrainer
             trainer_cls_args = [self.model]
         elif self.cfg.rl in ["kto"]:
             trainer_cls = AxolotlKTOTrainer
             trainer_cls_args = [self.model]
+        elif self.cfg.rl in ["simpo"]:
+            trainer_cls = AxolotlCPOTrainer
+            trainer_cls_args = [self.model]
         else:
             raise ValueError(f"Unsupported RL: {self.cfg.rl}")
         dpo_trainer = trainer_cls(
@@ -1708,6 +1885,8 @@ class HFRLTrainerBuilder(TrainerBuilderBase):
         )
         if self.cfg.fsdp:
             ensure_dtype(dpo_trainer.model, dtype=self.cfg.torch_dtype)
+            if self.cfg.rl in ["dpo", "ipo"] and dpo_trainer.ref_model:
+                ensure_dtype(dpo_trainer.ref_model, dtype=self.cfg.torch_dtype)
 
         dpo_trainer = self.hook_post_create_trainer(dpo_trainer)
         for callback in self.get_post_trainer_create_callbacks(dpo_trainer):

src/axolotl/integrations/LICENSE.md

@@ -0,0 +1,58 @@
+### AXOLOTL COMMUNITY LICENSE AGREEMENT
+
+This Axolotl Community License Agreement (“Agreement”) is entered into by and between Axolotl AI Corp. (“Axolotl”) and
+any individual or entity (“Licensee”) who wishes to use the Software (as defined below) in accordance with the terms
+and conditions set forth in this Agreement.
+
+1.  Definitions
+    1.1 “Licensee” refers to any individual or entity who has obtained a copy of the Software under this Agreement.
+    1.2 “Plugin Integration” means independent integration software modules which may or may not be offered by Axolotl,
+        which may be licensed separately by their respective  authors and/or licensors.
+    1.3 “Software” refers to the specific sub-directory of the Axolotl, Inc. software located at
+        https://github.com/axolotl-ai-cloud/axolotl/tree/main/src/axolotl/integrations and its subdirectories which
+        permits Plugin Integrations to integrate with the Axolotl service.
+2.  Grant of License
+    2.1	Axolotl hereby grants Licensee a worldwide, non-exclusive, royalty-free, license to use, copy, modify, merge,
+        publish, distribute, sublicense, and/or otherwise exploit the Software, subject to the following conditions:
+        - Licensee must comply with all the terms and conditions of this Agreement.
+        - Licensee must include the original copyright notice and disclaimer of warranty in all copies or substantial
+          portions of the Software.
+    2.2 Licensee may use the Software for any lawful purpose, except as restricted in Section 3.
+3.  Restrictions
+    3.1 Licensee shall not use the Software for any activity that constitutes a commercial activity of offering for
+        free or for sale any services, platform, or equivalent  to third parties for the purposes of allowing such
+        third parties to fine-tune artificial intelligence models.
+    3.2 Licensee shall not:
+        - Use the Software for any illegal or unauthorized purpose.
+        - Reverse engineer, decompile, or disassemble the Software.
+        - Remove or modify any copyright, trademark, or other proprietary notices contained in the Software.
+        - Use the Software in a way that could damage, disable, overburden, or impair the functionality of the
+          Software or interfere with any third-party use of the Software.
+    3.3 Axolotl reserves the right to restrict certain Plugin Integrations for use with the Software. To the extent Licensee integrates a permitted, applicable Plugin Integration with the Software, Licensee shall comply with any additional terms and conditions imposed by the licensors of such Plugin Integration for use of such Plugin Integrations. Licensee shall contact Axolotl if it has questions about whether its use of the Software falls beyond the scope of this Agreement.
+4.  Intellectual Property Rights
+    4.1 Axolotl and its contributors retain all intellectual property rights in and to the Software. Licensee
+        acknowledges that this Agreement does not transfer any ownership rights or intellectual property rights to
+        Licensee.
+5.  Disclaimer of Warranty
+    5.1 THE SOFTWARE IS PROVIDED “AS IS,” WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
+        TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. IN NO EVENT SHALL
+        THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES, OR OTHER LIABILITY, WHETHER IN AN ACTION OF
+        CONTRACT, TORT, OR OTHERWISE, ARISING FROM, OUT OF, OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+        DEALINGS IN THE SOFTWARE.
+6.  Termination
+    6.1 Axolotl may terminate this Agreement at any time if Licensee fails to comply with any of the terms and
+        conditions set forth herein. Upon termination, Licensee shall cease all use of the Software and destroy any
+        copies in its possession.
+7.  Governing Law
+    7.1 This Agreement shall be governed by and construed in accordance with the laws of the State of California,
+        without regards to conflicts of laws provisions thereof.
+8.  Entire Agreement
+    8.1 This Agreement constitutes the entire agreement between Axolotl and Licensee with respect to the subject matter
+        hereof and supersedes all prior or contemporaneous understandings or agreements between the parties concerning
+        the Software, whether written or oral. Axolotl may update the terms of this Agreement from time to time, and
+        Licensee’s continued use of the Software after any such updates shall constitute acceptance of updated terms
+        on a go-forward basis.  Axolotl will use commercially reasonable efforts to provide Licensee notice of any
+        material updates. By using the Software, Licensee acknowledges that it has read, understood, and agrees to be
+        bound by the terms and conditions of this Agreement.
+
+This Agreement was last updated on August 23, 2024.

src/axolotl/integrations/base.py

@@ -0,0 +1,383 @@
+# Copyright 2024 Axolotl AI. All rights reserved.
+#
+# This software may be used and distributed according to
+# the terms of the Axolotl Community License Agreement (the "License");
+# you may not use this file except in compliance with the License.
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+# License for the specific language governing permissions and limitations under
+# the License.
+
+"""
+Base class for all plugins.
+
+A plugin is a reusable, modular, and self-contained piece of code that extends the functionality of Axolotl.
+Plugins can be used to integrate third-party models, modify the training process, or add new features.
+
+To create a new plugin, you need to inherit from the BasePlugin class and implement the required methods.
+"""
+import importlib
+import logging
+from typing import List
+
+
+class BasePlugin:
+    """
+    Base class for all plugins. Defines the interface for plugin methods.
+
+    Attributes:
+    None
+
+    Methods:
+    register(cfg): Registers the plugin with the given configuration.
+    pre_model_load(cfg): Performs actions before the model is loaded.
+    post_model_load(cfg, model): Performs actions after the model is loaded.
+    pre_lora_load(cfg, model): Performs actions before LoRA weights are loaded.
+    post_lora_load(cfg, model): Performs actions after LoRA weights are loaded.
+    create_optimizer(cfg, trainer): Creates and returns an optimizer for training.
+    create_lr_scheduler(cfg, trainer, optimizer): Creates and returns a learning rate scheduler.
+    add_callbacks_pre_trainer(cfg, model): Adds callbacks to the trainer before training.
+    add_callbacks_post_trainer(cfg, trainer): Adds callbacks to the trainer after training.
+    """
+
+    def __init__(self):
+        """
+        Initializes the BasePlugin.
+        """
+
+    def register(self, cfg):
+        """
+        Registers the plugin with the given configuration.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+
+        Returns:
+        None
+        """
+
+    def get_input_args(self):
+        """
+        Returns a pydantic model for the plugin's input arguments.
+        """
+
+    def pre_model_load(self, cfg):
+        """
+        Performs actions before the model is loaded.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+
+        Returns:
+        None
+        """
+
+    def post_model_load(self, cfg, model):
+        """
+        Performs actions after the model is loaded.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        model (object): The loaded model.
+
+        Returns:
+        None
+        """
+
+    def pre_lora_load(self, cfg, model):
+        """
+        Performs actions before LoRA weights are loaded.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        model (object): The loaded model.
+
+        Returns:
+        None
+        """
+
+    def post_lora_load(self, cfg, model):
+        """
+        Performs actions after LoRA weights are loaded.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        model (object): The loaded model.
+
+        Returns:
+        None
+        """
+
+    def create_optimizer(self, cfg, trainer):
+        """
+        Creates and returns an optimizer for training.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        trainer (object): The trainer object for training.
+
+        Returns:
+        object: The created optimizer.
+        """
+
+    def create_lr_scheduler(self, cfg, trainer, optimizer):
+        """
+        Creates and returns a learning rate scheduler.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        trainer (object): The trainer object for training.
+        optimizer (object): The optimizer for training.
+
+        Returns:
+        object: The created learning rate scheduler.
+        """
+
+    def add_callbacks_pre_trainer(self, cfg, model):
+        """
+        Adds callbacks to the trainer before training.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        model (object): The loaded model.
+
+        Returns:
+        List[callable]: A list of callback functions to be added to the TrainingArgs
+        """
+
+    def add_callbacks_post_trainer(self, cfg, trainer):
+        """
+        Adds callbacks to the trainer after training.
+
+        Parameters:
+        cfg (dict): The configuration for the plugin.
+        trainer (object): The trainer object for training.
+
+        Returns:
+        List[callable]: A list of callback functions to be added to the TrainingArgs
+        """
+
+
+def load_plugin(plugin_name: str) -> BasePlugin:
+    """
+    Loads a plugin based on the given plugin name.
+
+    The plugin name should be in the format "module_name.class_name".
+    This function splits the plugin name into module and class, imports the module,
+    retrieves the class from the module, and creates an instance of the class.
+
+    Parameters:
+    plugin_name (str): The name of the plugin to be loaded. The name should be in the format "module_name.class_name".
+
+    Returns:
+    BasePlugin: An instance of the loaded plugin.
+
+    Raises:
+    ImportError: If the plugin module cannot be imported.
+    """
+    # split the plugin name into module and class
+    module_name, class_name = plugin_name.rsplit(".", 1)
+
+    # import the module
+    module = importlib.import_module(module_name)
+    # instantiate the class
+    plugin_class = getattr(module, class_name)
+    # create an instance of the class
+    plugin = plugin_class()
+
+    return plugin
+
+
+class PluginManager:
+    """
+    The PluginManager class is responsible for loading and managing plugins.
+    It should be a singleton so it can be accessed from anywhere in the codebase.
+
+    Attributes:
+    plugins (List[BasePlugin]): A list of loaded plugins.
+
+    Methods:
+    get_instance(): Static method to get the singleton instance of PluginManager.
+    register(plugin_name: str): Registers a new plugin by its name.
+    pre_model_load(cfg): Calls the pre_model_load method of all registered plugins.
+    """
+
+    plugins: List[BasePlugin] = []
+
+    _instance = None
+
+    def __new__(cls):
+        """
+        Creates a new instance of PluginManager if it doesn't exist yet.
+        """
+        if cls._instance is None:
+            cls._instance = super(PluginManager, cls).__new__(cls)
+            cls._instance.plugins: List[BasePlugin] = []
+        return cls._instance
+
+    @staticmethod
+    def get_instance() -> "PluginManager":
+        """
+        Returns the singleton instance of PluginManager.
+        If the instance doesn't exist, it creates a new one.
+        """
+        if PluginManager._instance is None:
+            PluginManager()
+        return PluginManager._instance  # type: ignore
+
+    def register(self, plugin_name: str):
+        """
+        Registers a new plugin by its name.
+
+        Parameters:
+        plugin_name (str): The name of the plugin to be registered.
+
+        Returns:
+        None
+
+        Raises:
+        ImportError: If the plugin module cannot be imported.
+        """
+        try:
+            plugin = load_plugin(plugin_name)
+            self.plugins.append(plugin)
+        except ImportError:
+            logging.error(f"Failed to load plugin: {plugin_name}")
+
+    def get_input_args(self):
+        """
+        Returns a list of Pydantic classes for all registered plugins' input arguments.'
+
+        Returns:
+        list[str]: A list of Pydantic classes for all registered plugins' input arguments.'
+        """
+        input_args = []
+        for plugin in self.plugins:
+            input_args_from_plugin = plugin.get_input_args()
+            if input_args_from_plugin is not None:
+                input_args.append(input_args_from_plugin)
+        return input_args
+
+    def pre_model_load(self, cfg):
+        """
+        Calls the pre_model_load method of all registered plugins.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+
+        Returns:
+        None
+        """
+        for plugin in self.plugins:
+            plugin.pre_model_load(cfg)
+
+    def post_model_load(self, cfg, model):
+        """
+        Calls the post_model_load method of all registered plugins.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        model (object): The loaded model.
+
+        Returns:
+        None
+        """
+        for plugin in self.plugins:
+            plugin.post_model_load(cfg, model)
+
+    def pre_lora_load(self, cfg, model):
+        """
+        Calls the pre_lora_load method of all registered plugins.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        model (object): The loaded model.
+
+        Returns:
+        None
+        """
+        for plugin in self.plugins:
+            plugin.pre_lora_load(cfg, model)
+
+    def post_lora_load(self, cfg, model):
+        """
+        Calls the post_lora_load method of all registered plugins.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        model (object): The loaded model.
+
+        Returns:
+        None
+        """
+        for plugin in self.plugins:
+            plugin.post_lora_load(cfg, model)
+
+    def create_optimizer(self, cfg, trainer):
+        """
+        Calls the create_optimizer method of all registered plugins and returns the first non-None optimizer.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        trainer (object): The trainer object for training.
+
+        Returns:
+        object: The created optimizer, or None if none was found.
+        """
+        for plugin in self.plugins:
+            optimizer = plugin.create_optimizer(cfg, trainer)
+            if optimizer is not None:
+                return optimizer
+        return None
+
+    def create_lr_scheduler(self, cfg, trainer, optimizer):
+        """
+        Calls the create_lr_scheduler method of all registered plugins and returns the first non-None scheduler.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        trainer (object): The trainer object for training.
+        optimizer (object): The optimizer for training.
+
+        Returns:
+        object: The created learning rate scheduler, or None if none was found.
+        """
+        for plugin in self.plugins:
+            scheduler = plugin.create_lr_scheduler(cfg, trainer, optimizer)
+            if scheduler is not None:
+                return scheduler
+        return None
+
+    def add_callbacks_pre_trainer(self, cfg, model):
+        """
+        Calls the add_callbacks_pre_trainer method of all registered plugins.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        model (object): The loaded model.
+
+        Returns:
+        List[callable]: A list of callback functions to be added to the TrainingArgs.
+        """
+        callbacks = []
+        for plugin in self.plugins:
+            callbacks.extend(plugin.add_callbacks_pre_trainer(cfg, model))
+        return callbacks
+
+    def add_callbacks_post_trainer(self, cfg, trainer):
+        """
+        Calls the add_callbacks_post_trainer method of all registered plugins.
+
+        Parameters:
+        cfg (dict): The configuration for the plugins.
+        trainer (object): The trainer object for training.
+
+        Returns:
+        List[callable]: A list of callback functions to be added to the TrainingArgs.
+        """
+        callbacks = []
+        for plugin in self.plugins:
+            callbacks.extend(plugin.add_callbacks_post_trainer(cfg, trainer))
+        return callbacks

src/axolotl/integrations/config.py

@@ -0,0 +1,65 @@
+# Copyright 2024 Axolotl AI. All rights reserved.
+#
+# This software may be used and distributed according to
+# the terms of the Axolotl Community License Agreement (the "License");
+# you may not use this file except in compliance with the License.
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
+# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
+# License for the specific language governing permissions and limitations under
+# the License.
+
+"""
+module to handle merging the plugins' input arguments with the base configurations.
+
+this was moved here to prevent circular imports
+"""
+
+from typing import Any, Dict, List
+
+from axolotl.utils.config.models.input.v0_4_1 import (
+    AxolotlConfigWCapabilities as AxolotlConfigWCapabilitiesBase,
+)
+from axolotl.utils.config.models.input.v0_4_1 import (
+    AxolotlInputConfig as AxolotlInputConfigBase,
+)
+
+
+def merge_input_args():
+    """
+    Merges input arguments from registered plugins with the base configurations.
+
+    This function retrieves the input arguments from registered plugins using the PluginManager.
+    It then dynamically creates new classes, AxolotlConfigWCapabilities and AxolotlInputConfig,
+    that inherit from the base configurations and include the input arguments from the plugins.
+
+    Returns:
+    tuple: A tuple containing the newly created classes, AxolotlConfigWCapabilities and AxolotlInputConfig.
+    """
+    from axolotl.integrations.base import PluginManager
+
+    plugin_manager = PluginManager.get_instance()
+    input_args: List[str] = plugin_manager.get_input_args()
+    plugin_classes = []
+    dynamic_input = ""
+    for plugin_args in input_args:
+        plugin_module, plugin_cls = plugin_args.rsplit(".", 1)
+        dynamic_input += f"from {plugin_module} import {plugin_cls}\n"
+        plugin_classes.append(plugin_cls)
+    if dynamic_input:
+        dynamic_input += f"class AxolotlConfigWCapabilities(AxolotlConfigWCapabilitiesBase, {', '.join(plugin_classes)}):\n    pass\n"
+        dynamic_input += f"class AxolotlInputConfig(AxolotlInputConfigBase, {', '.join(plugin_classes)}):\n    pass\n"
+
+        namespace: Dict[Any, Any] = {}
+        exec(  # pylint: disable=exec-used  # nosec B102
+            dynamic_input, globals(), namespace
+        )
+        AxolotlInputConfig = namespace[  # pylint: disable=invalid-name
+            "AxolotlInputConfig"
+        ]
+        AxolotlConfigWCapabilities = namespace[  # pylint: disable=invalid-name
+            "AxolotlConfigWCapabilities"
+        ]
+        return AxolotlConfigWCapabilities, AxolotlInputConfig
+    return AxolotlConfigWCapabilitiesBase, AxolotlInputConfigBase

src/axolotl/integrations/liger/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

src/axolotl/integrations/liger/__init__.py

@@ -0,0 +1,189 @@
+# Copyright 2024 Axolotl AI. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""
+Module for the Plugin for LIGER integraton with Axolotl.
+
+Liger Kernel is the collection of Triton-native kernels for LLM Training.
+It is designed to be performant, correct, and light-weight.
+"""
+import logging
+import sys
+from functools import partial
+
+from liger_kernel.transformers.cross_entropy import LigerCrossEntropyLoss
+from liger_kernel.transformers.geglu import LigerGEGLUMLP
+from liger_kernel.transformers.rms_norm import LigerRMSNorm
+from liger_kernel.transformers.rope import liger_rotary_pos_emb
+from liger_kernel.transformers.swiglu import LigerSwiGLUMLP
+
+from axolotl.integrations.base import BasePlugin
+
+from .args import LigerArgs  # pylint: disable=unused-import. # noqa: F401
+
+
+class LigerPlugin(BasePlugin):
+    """
+    Plugin for LIGER integraton with Axolotl.
+    """
+
+    def get_input_args(self):
+        return "axolotl.integrations.liger.LigerArgs"
+
+    def pre_model_load(self, cfg):
+        if cfg.model_config_type == "llama":
+            from liger_kernel.transformers.model.llama import (
+                lce_forward as llama_lce_forward,
+            )
+            from transformers.models.llama import modeling_llama
+
+            if cfg.liger_rope:
+                modeling_llama.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_llama.LlamaRMSNorm = LigerRMSNorm
+            if cfg.liger_swiglu:
+                modeling_llama.LlamaMLP = LigerSwiGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_llama.CrossEntropyLoss = LigerCrossEntropyLoss
+            elif cfg.liger_fused_linear_cross_entropy:
+                modeling_llama.LlamaForCausalLM.forward = llama_lce_forward
+
+        elif cfg.model_config_type == "mistral":
+            from liger_kernel.transformers.model.mistral import (
+                lce_forward as mistral_lce_forward,
+            )
+            from transformers.models.mistral import modeling_mistral
+
+            if cfg.liger_rope:
+                modeling_mistral.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_mistral.MistralRMSNorm = LigerRMSNorm
+            if cfg.liger_swiglu:
+                modeling_mistral.MistralMLP = LigerSwiGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_mistral.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                modeling_mistral.MistralForCausalLM.forward = mistral_lce_forward
+
+        elif cfg.model_config_type == "gemma":
+            from liger_kernel.transformers.model.gemma import (
+                lce_forward as gemma_lce_forward,
+            )
+            from transformers.models.gemma import modeling_gemma
+
+            if cfg.liger_rope:
+                modeling_gemma.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_gemma.GemmaRMSNorm = partial(
+                    LigerRMSNorm, offset=1.0, init_fn="zeros", casting_mode="gemma"
+                )
+            if cfg.liger_swiglu:
+                modeling_gemma.GemmaMLP = LigerGEGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_gemma.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                modeling_gemma.GemmaForCausalLM.forward = gemma_lce_forward
+
+        elif cfg.model_config_type == "jamba":
+            from transformers.models.jamba import modeling_jamba
+
+            from .models.jamba import lce_forward as jamba_lce_forward
+
+            if cfg.liger_rope:
+                modeling_jamba.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_jamba.JambaRMSNorm = LigerRMSNorm
+            if cfg.liger_swiglu:
+                modeling_jamba.JambaMLP = LigerSwiGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_jamba.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                modeling_jamba.JambaForCausalLM.forward = jamba_lce_forward
+
+        elif cfg.model_config_type == "qwen2":
+            from liger_kernel.transformers.model.qwen2 import (
+                lce_forward as qwen2_lce_forward,
+            )
+            from transformers.models.qwen2 import modeling_qwen2
+
+            if cfg.liger_rope:
+                modeling_qwen2.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_qwen2.Qwen2RMSNorm = LigerRMSNorm
+            if cfg.liger_swiglu:
+                modeling_qwen2.Qwen2MLP = LigerSwiGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_qwen2.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                modeling_qwen2.Qwen2ForCausalLM.forward = qwen2_lce_forward
+
+        elif cfg.model_config_type == "deepseek_v2":
+            from accelerate import init_empty_weights
+            from transformers import AutoModelForCausalLM
+
+            with init_empty_weights():
+                model = AutoModelForCausalLM.from_pretrained(
+                    cfg.base_model, trust_remote_code=cfg.trust_remote_code or False
+                )
+                modeling_mod = sys.modules[model.__class__.__module__]
+
+            from .models.deepseekv2 import lce_forward as deepseekv2_lce_forward
+
+            if cfg.liger_rope:
+                # The DeepseekV2 version of RoPE is different than upstream LLaMA.
+                # See https://github.com/linkedin/Liger-Kernel/issues/129#issuecomment-2313763528
+                logging.warning("Fused liger_rope is not supported for DeepseekV2.")
+            if cfg.liger_rms_norm:
+                modeling_mod.DeepseekV2RMSNorm = LigerRMSNorm
+            if cfg.liger_swiglu:
+                modeling_mod.DeepseekV2MLP.forward = LigerSwiGLUMLP.forward
+            if cfg.liger_cross_entropy:
+                modeling_mod.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                modeling_mod.DeepseekV2ForCausalLM.forward = deepseekv2_lce_forward
+
+        elif cfg.model_config_type == "gemma2":
+            from transformers.models.gemma2 import modeling_gemma2
+
+            if cfg.liger_rope:
+                modeling_gemma2.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_gemma2.Gemma2RMSNorm = partial(
+                    LigerRMSNorm, offset=1.0, init_fn="zeros", casting_mode="gemma"
+                )
+            if cfg.liger_swiglu:
+                modeling_gemma2.Gemma2MLP = LigerGEGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_gemma2.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                logging.warning(
+                    "Fused linear cross entropy is not supported for Gemma 2."
+                )
+
+        elif cfg.model_config_type == "phi3":
+            from liger_kernel.transformers.model.phi3 import (
+                lce_forward as phi3_lce_forward,
+            )
+            from transformers.models.phi3 import modeling_phi3
+
+            if cfg.liger_rope:
+                modeling_phi3.apply_rotary_pos_emb = liger_rotary_pos_emb
+            if cfg.liger_rms_norm:
+                modeling_phi3.Phi3RMSNorm = LigerRMSNorm
+            if cfg.liger_swiglu:
+                modeling_phi3.Phi3MLP = LigerSwiGLUMLP
+            if cfg.liger_cross_entropy:
+                modeling_phi3.CrossEntropyLoss = LigerCrossEntropyLoss
+            if cfg.liger_fused_linear_cross_entropy:
+                modeling_phi3.Phi3ForCausalLM.forward = phi3_lce_forward

src/axolotl/integrations/liger/args.py

@@ -0,0 +1,32 @@
+# Copyright 2024 Axolotl AI. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""
+Module for handling LIGER input arguments.
+"""
+from typing import Optional
+
+from pydantic import BaseModel
+
+
+class LigerArgs(BaseModel):
+    """
+    Input args for LIGER.
+    """
+
+    liger_rope: Optional[bool] = None
+    liger_rms_norm: Optional[bool] = None
+    liger_swiglu: Optional[bool] = None
+    liger_cross_entropy: Optional[bool] = None
+    liger_fused_linear_cross_entropy: Optional[bool] = None

src/axolotl/integrations/liger/models/deepseekv2.py

@@ -0,0 +1,127 @@
+"""
+DeepseekV2 model with LigerFusedLinearCrossEntropyLoss
+"""
+# pylint: disable=duplicate-code
+
+from typing import List, Optional, Tuple, Union
+
+import torch
+from liger_kernel.transformers.fused_linear_cross_entropy import (
+    LigerFusedLinearCrossEntropyLoss,
+)
+from torch.nn import CrossEntropyLoss
+from transformers.modeling_outputs import CausalLMOutputWithPast
+
+
+# @add_start_docstrings_to_model_forward(DeepseekV2_INPUTS_DOCSTRING)
+# @replace_return_docstrings(
+#    output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC
+# )
+def lce_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[List[torch.FloatTensor]] = None,
+    inputs_embeds: Optional[torch.FloatTensor] = None,
+    labels: Optional[torch.LongTensor] = None,
+    use_cache: Optional[bool] = None,
+    output_attentions: Optional[bool] = None,
+    output_hidden_states: Optional[bool] = None,
+    return_dict: Optional[bool] = None,
+) -> Union[Tuple, CausalLMOutputWithPast]:
+    r"""
+    Args:
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, transformers.,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, transformers., config.vocab_size]`.
+
+    Returns:
+
+    Example:
+
+    ```python
+    >>> from transformers import AutoTokenizer, DeepseekV2ForCausalLM
+
+    >>> model = DeepseekV2ForCausalLM.from_pretrained(PATH_TO_CONVERTED_WEIGHTS)
+    >>> tokenizer = AutoTokenizer.from_pretrained(PATH_TO_CONVERTED_TOKENIZER)
+
+    >>> prompt = "Hey, are you conscious? Can you talk to me?"
+    >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+    >>> # Generate
+    >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+    >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+    "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+    ```"""
+    output_attentions = (
+        output_attentions
+        if output_attentions is not None
+        else self.config.output_attentions
+    )
+    output_hidden_states = (
+        output_hidden_states
+        if output_hidden_states is not None
+        else self.config.output_hidden_states
+    )
+    return_dict = (
+        return_dict if return_dict is not None else self.config.use_return_dict
+    )
+
+    # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+    outputs = self.model(
+        input_ids=input_ids,
+        attention_mask=attention_mask,
+        position_ids=position_ids,
+        past_key_values=past_key_values,
+        inputs_embeds=inputs_embeds,
+        use_cache=use_cache,
+        output_attentions=output_attentions,
+        output_hidden_states=output_hidden_states,
+        return_dict=return_dict,
+    )
+
+    hidden_states = outputs[0]
+
+    loss = None
+    logits = None
+
+    if self.training:
+        shift_hidden_states = hidden_states[..., :-1, :].contiguous()
+        shift_labels = labels[..., 1:].contiguous()
+
+        # flatten tokens
+        shift_hidden_states = shift_hidden_states.view(-1, self.config.hidden_size)
+        shift_labels = shift_labels.view(-1)
+
+        lce = LigerFusedLinearCrossEntropyLoss()
+        loss = lce(self.lm_head.weight, shift_hidden_states, shift_labels)
+    else:
+        logits = self.lm_head(hidden_states)
+        logits = logits.float()
+
+        loss = None
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+    if not return_dict:
+        output = (logits,) + outputs[1:]
+        return (loss,) + output if loss is not None else output
+
+    return CausalLMOutputWithPast(
+        loss=loss,
+        logits=logits,
+        past_key_values=outputs.past_key_values,
+        hidden_states=outputs.hidden_states,
+        attentions=outputs.attentions,
+    )

src/axolotl/integrations/liger/models/jamba.py

@@ -0,0 +1,173 @@
+"""
+Jamba model with LigerFusedLinearCrossEntropyLoss
+"""
+# pylint: disable=duplicate-code
+
+from typing import Optional, Tuple, Union
+
+import torch
+from liger_kernel.transformers.fused_linear_cross_entropy import (
+    LigerFusedLinearCrossEntropyLoss,
+)
+from torch.nn import CrossEntropyLoss
+from transformers.modeling_outputs import MoeCausalLMOutputWithPast
+from transformers.models.jamba.modeling_jamba import (
+    _CONFIG_FOR_DOC,
+    JAMBA_INPUTS_DOCSTRING,
+    HybridMambaAttentionDynamicCache,
+    load_balancing_loss_func,
+)
+from transformers.utils import (
+    add_start_docstrings_to_model_forward,
+    replace_return_docstrings,
+)
+
+
+@add_start_docstrings_to_model_forward(JAMBA_INPUTS_DOCSTRING)
+@replace_return_docstrings(
+    output_type=MoeCausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC
+)
+def lce_forward(
+    self,
+    input_ids: torch.LongTensor = None,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_values: Optional[HybridMambaAttentionDynamicCache] = None,
+    inputs_embeds: Optional[torch.FloatTensor] = None,
+    labels: Optional[torch.LongTensor] = None,
+    use_cache: Optional[bool] = None,
+    output_attentions: Optional[bool] = None,
+    output_hidden_states: Optional[bool] = None,
+    output_router_logits: Optional[bool] = None,
+    return_dict: Optional[bool] = None,
+    cache_position: Optional[torch.LongTensor] = None,
+    num_logits_to_keep: Optional[Union[int, None]] = None,
+) -> Union[Tuple, MoeCausalLMOutputWithPast]:
+    r"""
+    Args:
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should either be in `[0, ...,
+            config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored
+            (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`.
+
+        num_logits_to_keep (`int` or `None`, *optional*):
+            Calculate logits for the last `num_logits_to_keep` tokens. If `None`, calculate logits for all
+            `input_ids`. Only last token logits are needed for generation, and calculating them only for that token
+            can save memory, which becomes pretty significant for long sequences.
+
+    Returns:
+
+    Example:
+
+    ```python
+    >>> from transformers import AutoTokenizer, JambaForCausalLM
+
+    >>> model = JambaForCausalLM.from_pretrained("ai21labs/Jamba-v0.1")
+    >>> tokenizer = AutoTokenizer.from_pretrained("ai21labs/Jamba-v0.1")
+
+    >>> prompt = "Hey, are you conscious? Can you talk to me?"
+    >>> inputs = tokenizer(prompt, return_tensors="pt")
+
+    >>> # Generate
+    >>> generate_ids = model.generate(inputs.input_ids, max_length=30)
+    >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]
+    "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you."
+    ```"""
+
+    output_attentions = (
+        output_attentions
+        if output_attentions is not None
+        else self.config.output_attentions
+    )
+    output_router_logits = (
+        output_router_logits
+        if output_router_logits is not None
+        else self.config.output_router_logits
+    )
+
+    output_hidden_states = (
+        output_hidden_states
+        if output_hidden_states is not None
+        else self.config.output_hidden_states
+    )
+    return_dict = (
+        return_dict if return_dict is not None else self.config.use_return_dict
+    )
+
+    # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+    outputs = self.model(
+        input_ids=input_ids,
+        attention_mask=attention_mask,
+        position_ids=position_ids,
+        past_key_values=past_key_values,
+        inputs_embeds=inputs_embeds,
+        use_cache=use_cache,
+        output_attentions=output_attentions,
+        output_hidden_states=output_hidden_states,
+        output_router_logits=output_router_logits,
+        cache_position=cache_position,
+        return_dict=return_dict,
+    )
+
+    hidden_states = outputs[0]
+
+    loss = None
+    logits = None
+
+    if self.training:
+        shift_hidden_states = hidden_states[..., :-1, :].contiguous()
+        shift_labels = labels[..., 1:].contiguous()
+
+        # flatten tokens
+        shift_hidden_states = shift_hidden_states.view(-1, self.config.hidden_size)
+        shift_labels = shift_labels.view(-1)
+
+        lce = LigerFusedLinearCrossEntropyLoss()
+        loss = lce(self.lm_head.weight, shift_hidden_states, shift_labels)
+    else:
+        if num_logits_to_keep is None:
+            logits = self.lm_head(hidden_states)
+        else:
+            logits = self.lm_head(hidden_states[..., -num_logits_to_keep:, :])
+        logits = logits.float()
+
+        if labels is not None:
+            # Shift so that tokens < n predict n
+            shift_logits = logits[..., :-1, :].contiguous()
+            shift_labels = labels[..., 1:].contiguous()
+            # Flatten the tokens
+            loss_fct = CrossEntropyLoss()
+            shift_logits = shift_logits.view(-1, self.config.vocab_size)
+            shift_labels = shift_labels.view(-1)
+            # Enable model parallelism
+            shift_labels = shift_labels.to(shift_logits.device)
+            loss = loss_fct(shift_logits, shift_labels)
+
+    aux_loss = None
+    if output_router_logits:
+        aux_loss = load_balancing_loss_func(
+            outputs.router_logits if return_dict else outputs[-1],
+            self.num_experts,
+            self.num_experts_per_tok,
+            attention_mask,
+        )
+        if labels is not None:
+            loss += self.router_aux_loss_coef * aux_loss.to(
+                loss.device
+            )  # make sure to reside in the same device
+
+    if not return_dict:
+        output = (logits,) + outputs[1:]
+        if output_router_logits:
+            output = (aux_loss,) + output
+        return (loss,) + output if loss is not None else output
+
+    return MoeCausalLMOutputWithPast(
+        loss=loss,
+        aux_loss=aux_loss,
+        logits=logits,
+        past_key_values=outputs.past_key_values,
+        hidden_states=outputs.hidden_states,
+        attentions=outputs.attentions,
+        router_logits=outputs.router_logits,
+    )

src/axolotl/integrations/spectrum/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

src/axolotl/integrations/spectrum/README.md

@@ -0,0 +1,21 @@
+## Spectrum: Targeted Training on Signal to Noise Ratio
+
+by Eric Hartford, Lucas Atkins, Fernando Fernandes, David Golchinfar
+
+This plugin contains code to freeze the bottom fraction of modules in a model, based on the Signal-to-Noise Ratio (SNR).
+
+### Overview
+
+Spectrum is a tool for scanning and evaluating the Signal-to-Noise Ratio (SNR) of layers in large language models.
+By identifying the top n% of layers with the highest SNR, you can optimize training efficiency.
+
+### Usage
+
+```yaml
+plugins:
+  - axolotl.integrations.spectrum.SpectrumPlugin
+
+spectrum_top_fraction: 0.5
+# Optional if using a pre-scanned model as your base_model. Useful if using a model mirror
+spectrum_model_name: meta-llama/Meta-Llama-3.1-8B
+```

src/axolotl/integrations/spectrum/__init__.py

@@ -0,0 +1,102 @@
+# Copyright 2024 Axolotl AI. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""
+Spectrum Plugin to automatically generate unfrozen parameters based on SNR data.
+"""
+
+import json
+import logging
+
+import requests
+
+from axolotl.integrations.base import BasePlugin
+
+from .args import SpectrumArgs  # pylint: disable=unused-import. # noqa: F401
+
+
+def _generate_unfrozen_params_yaml(snr_data, top_fraction=0.5):
+    unfrozen_parameters = {}
+    for layer_name, info in snr_data.items():
+        layer_type = info["type"]
+        if layer_type not in unfrozen_parameters:
+            unfrozen_parameters[layer_type] = []
+        unfrozen_parameters[layer_type].append((layer_name, info["snr"]))
+    top_layers_by_type = {}
+    for layer_type, layers in unfrozen_parameters.items():
+        layers_sorted = sorted(layers, key=lambda x: x[1], reverse=True)
+        num_top_layers = int(len(layers) * top_fraction)
+        top_layers_by_type[layer_type] = [
+            layer[0] for layer in layers_sorted[:num_top_layers]
+        ]
+    unfrozen_parameters = [
+        "^lm_head.weight$",
+        "^model.embed_tokens.weight$",
+    ]
+    for layer_type, layer_names in top_layers_by_type.items():
+        for layer_name in layer_names:
+            unfrozen_parameters.append(layer_name)
+    return unfrozen_parameters
+
+
+class SpectrumPlugin(BasePlugin):
+    """
+    Spectrum Plugin to automatically generate unfrozen parameters based on SNR data.
+    """
+
+    base_url = "https://raw.githubusercontent.com/cognitivecomputations/spectrum/main/model_snr_results/"
+    base_path = "./model_snr_results/"
+    snr_file_template = "snr_results_{model_name_slug}.json"
+
+    def get_input_args(self):
+        return "axolotl.integrations.spectrum.SpectrumArgs"
+
+    def pre_model_load(self, cfg):
+        if cfg.get("spectrum_model_name"):
+            model_name = cfg["spectrum_model_name"]
+        else:
+            model_name = cfg["base_model"]
+        top_fraction = cfg.get("spectrum_top_fraction", 50)
+        model_slug = model_name.replace("/", "-").replace("_", "-")
+        snr_url = self.base_url + self.snr_file_template.format(
+            model_name_slug=model_slug
+        )
+        snr_path = self.base_path + self.snr_file_template.format(
+            model_name_slug=model_slug
+        )
+        # first check if the files exist locally and read the json
+        snr_data = None
+        try:
+            with open(snr_path, "r", encoding="utf-8") as fin:
+                snr_data = json.load(fin)
+        except FileNotFoundError:
+            pass
+        except Exception as exc:  # pylint: disable=broad-exception-caught
+            logging.warning(f"Failed to read SNR data from {snr_path}: {exc}")
+
+        if not snr_data:
+            try:
+                snr_data = requests.get(snr_url, timeout=60).json()
+            except requests.exceptions.RequestException as exc:
+                logging.warning(f"Failed to fetch SNR data from {snr_url}: {exc}")
+                return
+            # also catch json parsing errors
+            except json.JSONDecodeError as exc:
+                logging.warning(f"Failed to parse SNR data from {snr_url}: {exc}")
+                return
+
+        unfrozen_parameters = _generate_unfrozen_params_yaml(
+            snr_data, top_fraction=top_fraction
+        )
+        cfg["unfrozen_parameters"] = unfrozen_parameters

src/axolotl/integrations/spectrum/args.py

@@ -0,0 +1,29 @@
+# Copyright 2024 Axolotl AI. All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+"""
+Module for handling Spectrum input arguments.
+"""
+from typing import Optional
+
+from pydantic import BaseModel
+
+
+class SpectrumArgs(BaseModel):
+    """
+    Input args for Spectrum.
+    """
+
+    spectrum_top_fraction: Optional[float] = 0.5
+    spectrum_model_name: Optional[str] = None

src/axolotl/loraplus.py

@@ -1,133 +0,0 @@
-"""Module for LoRA+"""
-
-# MIT License
-#
-# Copyright (c) 2024 nikhil-ghosh-berkeley
-# https://github.com/nikhil-ghosh-berkeley/loraplus
-
-import logging
-from functools import reduce
-
-from peft.tuners import lora
-from torch import nn
-from transformers.pytorch_utils import ALL_LAYERNORM_LAYERS
-from transformers.trainer_pt_utils import get_parameter_names
-
-LOG = logging.getLogger("axolotl.loraplus")
-
-
-def get_module(name, opt_model):
-    """
-    Retrieve a module from a model using its parameter name.
-    Args:
-        name (str): Full name of the parameter, typically including module path.
-        opt_model (torch.nn.Module): The model from which to retrieve the module.
-
-    Returns:
-        Module corresponding to the given name.
-    """
-    parent_idx = 2 if "lora" in name else 1
-    module_names = name.split(sep=".")[:-parent_idx]
-    module = reduce(getattr, module_names, opt_model)
-    return module
-
-
-def create_loraplus_optimizer(
-    opt_model,
-    optimizer_cls,
-    optimizer_kwargs,
-    loraplus_lr_ratio,
-    loraplus_lr_embedding=None,
-):
-    """
-    Creates an optimizer for the given model, applying LoRA-specific learning rate adjustments to different parameter groups.
-
-    Args:
-        opt_model (torch.nn.Module): The model for which the optimizer is being created.
-        optimizer_cls (class): The class of the optimizer to be used (e.g., torch.optim.Adam).
-        optimizer_kwargs (dict): A dictionary of keyword arguments for the optimizer's initialization.
-        loraplus_lr_ratio (float): The learning rate ratio to be applied to LoRA parameters.
-        loraplus_lr_embedding (float, optional): A specific learning rate for embedding parameters, with a default value if not provided.
-
-    Returns:
-        An instance of the specified optimizer class configured with the model's parameters organized into groups with custom learning rates.
-    """
-
-    assert loraplus_lr_ratio is not None, "loraplus_lr_ratio must be provided."
-
-    if loraplus_lr_embedding is None:
-        loraplus_lr_embedding = 1e-6
-
-    decay_parameters = get_parameter_names(opt_model, ALL_LAYERNORM_LAYERS)
-    decay_parameters = [name for name in decay_parameters if "bias" not in name]
-    param_groups = {
-        "groupA": {},
-        "groupB": {},
-        "groupB_no_decay": {},
-        "embedding": {},
-    }
-
-    for name, param in opt_model.named_parameters():
-        if not param.requires_grad:
-            continue
-
-        module = get_module(name, opt_model)
-        if isinstance(module, lora.Embedding):
-            param_groups["embedding"][name] = param
-        elif "lora_B" in name or param.ndim == 1:
-            if name in decay_parameters:
-                param_groups["groupB"][name] = param
-            else:
-                param_groups["groupB_no_decay"][name] = param
-        else:
-            param_groups["groupA"][name] = param
-
-    assigned_param_groups = ""
-    for group, group_params in param_groups.items():
-        assigned_param_groups += f"{group}\n {list(group_params.keys())}\n\n"
-    LOG.info(assigned_param_groups)
-
-    lr = optimizer_kwargs["lr"]  # pylint: disable=invalid-name
-    weight_decay = optimizer_kwargs.get("weight_decay", 0.0)
-
-    optimizer_grouped_parameters = [
-        {
-            "params": list(param_groups["groupA"].values()),
-            "weight_decay": weight_decay,
-            "lr": lr,
-        },
-        {
-            "params": list(param_groups["embedding"].values()),
-            "weight_decay": weight_decay,
-            "lr": loraplus_lr_embedding,
-        },
-        {
-            "params": list(param_groups["groupB"].values()),
-            "weight_decay": weight_decay,
-            "lr": lr * loraplus_lr_ratio,
-        },
-        {
-            "params": list(param_groups["groupB_no_decay"].values()),
-            "weight_decay": 0.0,
-            "lr": lr * loraplus_lr_ratio,
-        },
-    ]
-
-    optimizer = optimizer_cls(optimizer_grouped_parameters, **optimizer_kwargs)
-    if optimizer_cls.__name__ == "Adam8bit":
-        import bitsandbytes
-
-        manager = bitsandbytes.optim.GlobalOptimManager.get_instance()
-
-        skipped = 0
-        for module in opt_model.modules():
-            if isinstance(module, nn.Embedding):
-                skipped += sum(
-                    {p.data_ptr(): p.numel() for p in module.parameters()}.values()
-                )
-                LOG.info(f"skipped {module}: {skipped/2**20}M params")
-                manager.register_module_override(module, "weight", {"optim_bits": 32})
-                LOG.debug(f"bitsandbytes: will optimize {module} in fp32")
-        LOG.info(f"skipped: {skipped/2**20}M params")
-
-    return optimizer

src/axolotl/monkeypatch/attention/mllama.py

@@ -0,0 +1,229 @@
+"""
+Monkeypatch for Vision Llama for FA2 support
+"""
+# pylint: disable=duplicate-code
+
+from typing import Optional, Tuple
+
+import torch
+from flash_attn.flash_attn_interface import flash_attn_func
+from transformers.cache_utils import Cache
+from transformers.modeling_flash_attention_utils import _flash_attention_forward
+from transformers.models.mllama.configuration_mllama import MllamaTextConfig
+from transformers.models.mllama.modeling_mllama import (
+    MllamaTextCrossAttention,
+    MllamaTextSelfAttention,
+    apply_rotary_pos_emb,
+    repeat_kv,
+)
+from transformers.utils import is_flash_attn_greater_or_equal_2_10
+
+
+class MllamaTextCrossFlashAttention2(MllamaTextCrossAttention):
+    """
+    Mllama flash cross-attention module. This module inherits from `MllamaTextCrossAttention` and
+    implements the forward pass using Flash Attention for improved performance.
+    """
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        # Check if flash attention version is greater or equal to 2.1
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        cross_attention_states: Optional[torch.Tensor] = None,
+        past_key_value: Optional[Cache] = None,
+        attention_mask: Optional[  # pylint: disable=unused-argument
+            torch.Tensor
+        ] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,  # pylint: disable=unused-argument
+        cache_position: Optional[torch.LongTensor] = None,
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        query_states = query_states.view(
+            bsz, q_len, self.num_heads, self.head_dim
+        ).transpose(1, 2)
+        query_states = self.q_norm(query_states)
+
+        if cross_attention_states is not None:
+            key_states = self.k_proj(cross_attention_states)
+            value_states = self.v_proj(cross_attention_states)
+            key_states = key_states.view(
+                bsz, -1, self.num_key_value_heads, self.head_dim
+            ).transpose(1, 2)
+            value_states = value_states.view(
+                bsz, -1, self.num_key_value_heads, self.head_dim
+            ).transpose(1, 2)
+            key_states = repeat_kv(key_states, self.num_key_value_groups)
+            value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+            key_states = self.k_norm(key_states)
+            if past_key_value is not None:
+                key_states, value_states = past_key_value.update(
+                    key_states,
+                    value_states,
+                    self.layer_idx,
+                    {"cache_position": cache_position},
+                )
+        elif cache_position[0] != 0:
+            key_states, value_states = (
+                past_key_value.key_cache[self.layer_idx],
+                past_key_value.value_cache[self.layer_idx],
+            )
+        else:
+            raise ValueError(
+                "Cross attention layer can't find neither `cross_attn_states` nor cached values for key/values!"
+            )
+
+        # Transpose to get the expected layout for flash attention
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        # Apply Flash Attention
+        dropout_rate = self.dropout if self.training else 0.0
+        output = flash_attn_func(
+            query_states,
+            key_states,
+            value_states,
+            dropout_p=dropout_rate,
+            softmax_scale=None,
+            causal=False,
+            return_attn_probs=output_attentions,
+        )
+
+        attn_output = output.contiguous().view(bsz, q_len, -1)
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+class MllamaTextSelfFlashAttention2(MllamaTextSelfAttention):
+    """
+    Mllama flash self-attention module. This module inherits from `MllamaTextSelfAttention` and
+    implements the forward pass using Flash Attention for improved performance.
+    """
+
+    def __init__(self, config: MllamaTextConfig, layer_idx: int, *args, **kwargs):
+        super().__init__(config, layer_idx, *args, **kwargs)
+
+        # Check if flash attention version is greater or equal to 2.1
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.Tensor] = None,
+        position_embeddings: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        output_attentions: bool = False,
+        use_cache: bool = False,  # pylint: disable=unused-argument
+        past_key_value=None,
+        cache_position: Optional[torch.LongTensor] = None,
+        **kwargs,  # pylint: disable=unused-argument
+    ) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+        output_attentions = False
+
+        bsz, q_len, _ = hidden_states.size()
+
+        query_states = self.q_proj(hidden_states)
+        key_states = self.k_proj(hidden_states)
+        value_states = self.v_proj(hidden_states)
+
+        # Flash attention requires the input to have the shape
+        # batch_size x seq_length x num_heads x head_dim
+        query_states = query_states.view(
+            bsz, q_len, self.num_heads, self.head_dim
+        ).transpose(1, 2)
+        key_states = key_states.view(
+            bsz, q_len, self.num_key_value_heads, self.head_dim
+        ).transpose(1, 2)
+        value_states = value_states.view(
+            bsz, q_len, self.num_key_value_heads, self.head_dim
+        ).transpose(1, 2)
+
+        cos, sin = position_embeddings
+        query_states, key_states = apply_rotary_pos_emb(
+            query_states, key_states, cos, sin
+        )
+
+        if past_key_value is not None:
+            # sin and cos are specific to RoPE models; cache_position needed for the static cache
+            cache_kwargs = {"sin": sin, "cos": cos, "cache_position": cache_position}
+            key_states, value_states = past_key_value.update(
+                key_states, value_states, self.layer_idx, cache_kwargs
+            )
+
+        key_states = repeat_kv(key_states, self.num_key_value_groups)
+        value_states = repeat_kv(value_states, self.num_key_value_groups)
+
+        # Transpose to get the expected layout for flash attention
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
+
+        dropout_rate = self.dropout if self.training else 0.0
+
+        # Handle potential silent casting to float32
+        input_dtype = query_states.dtype
+        if input_dtype == torch.float32:
+            if torch.is_autocast_enabled():
+                target_dtype = torch.get_autocast_gpu_dtype()
+            elif hasattr(self.config, "_pre_quantization_dtype"):
+                target_dtype = (
+                    self.config._pre_quantization_dtype  # pylint: disable=protected-access
+                )
+            else:
+                target_dtype = self.q_proj.weight.dtype
+
+            query_states = query_states.to(target_dtype)
+            key_states = key_states.to(target_dtype)
+            value_states = value_states.to(target_dtype)
+
+        attn_output = _flash_attention_forward(
+            query_states,
+            key_states,
+            value_states,
+            attention_mask,
+            q_len,
+            dropout=dropout_rate,
+            use_top_left_mask=self._flash_attn_uses_top_left_mask,
+            is_causal=True,
+        )
+
+        attn_output = attn_output.reshape(bsz, q_len, -1).contiguous()
+        attn_output = self.o_proj(attn_output)
+
+        if not output_attentions:
+            attn_weights = None
+
+        return attn_output, attn_weights, past_key_value
+
+
+def patch_mllama():
+    from transformers.models.mllama.modeling_mllama import (
+        MLLAMA_TEXT_ATTENTION_CLASSES,
+        MLLAMA_TEXT_CROSS_ATTENTION_CLASSES,
+        MLLAMA_VISION_ATTENTION_CLASSES,
+        MllamaPreTrainedModel,
+    )
+
+    MllamaPreTrainedModel._supports_flash_attn_2 = (  # pylint: disable=protected-access
+        True
+    )
+    MLLAMA_TEXT_ATTENTION_CLASSES["flash_attention_2"] = MllamaTextSelfFlashAttention2
+    MLLAMA_TEXT_CROSS_ATTENTION_CLASSES[
+        "flash_attention_2"
+    ] = MllamaTextCrossFlashAttention2
+    # fallback to SDPA
+    MLLAMA_VISION_ATTENTION_CLASSES[
+        "flash_attention_2"
+    ] = MLLAMA_VISION_ATTENTION_CLASSES["sdpa"]

src/axolotl/monkeypatch/llama_attn_hijack_flash.py

@@ -78,6 +78,33 @@ def replace_llama_qkv_with_fused(model):
             set_module_name(model, name, qkv)
 
 
+def patch_llama_cross_entropy():
+    from flash_attn.losses.cross_entropy import CrossEntropyLoss
+
+    LOG.info("patching with flash_attn.losses.cross_entropy")
+    transformers.models.llama.modeling_llama.CrossEntropyLoss = partial(
+        CrossEntropyLoss, inplace_backward=True
+    )
+
+
+def patch_llama_rms_norm():
+    try:
+        from flash_attn.ops.rms_norm import RMSNorm
+
+        class LlamaRMSNorm(RMSNorm):
+            """Patched LLamaRMSNorm"""
+
+            def __init__(self, hidden_size, eps=1e-6):
+                super().__init__(hidden_size, eps=eps)
+
+        LOG.info("patching with flash_attn.ops.rms_norm")
+        transformers.models.llama.modeling_llama.LlamaRMSNorm = LlamaRMSNorm
+    except ImportError:
+        LOG.warning(
+            "optimized flash-attention RMSNorm not found (run `pip install 'git+https://github.com/Dao-AILab/flash-attention.git#egg=dropout_layer_norm&subdirectory=csrc/layer_norm'`)"
+        )
+
+
 def replace_llama_attn_with_flash_attn(
     packed: Optional[bool] = False,
     cross_entropy: Optional[bool] = False,
@@ -104,35 +131,11 @@ def replace_llama_attn_with_flash_attn(
 
     # skip only if explicitly disabled
     if cross_entropy:
-        try:
-            from flash_attn.losses.cross_entropy import CrossEntropyLoss
-
-            LOG.info("patching with flash_attn.losses.cross_entropy")
-            transformers.models.llama.modeling_llama.CrossEntropyLoss = partial(
-                CrossEntropyLoss, inplace_backward=True
-            )
-        except ImportError:
-            LOG.warning(
-                "optimized flash-attention CrossEntropyLoss not found (run `pip install 'git+https://github.com/Dao-AILab/flash-attention.git#egg=xentropy_cuda_lib&subdirectory=csrc/xentropy'`)"
-            )
+        patch_llama_cross_entropy()
 
     # skip only if explicitly disabled
     if rms_norm:
-        try:
-            from flash_attn.ops.rms_norm import RMSNorm
-
-            class LlamaRMSNorm(RMSNorm):
-                """Patched LLamaRMSNorm"""
-
-                def __init__(self, hidden_size, eps=1e-6):
-                    super().__init__(hidden_size, eps=eps)
-
-            LOG.info("patching with flash_attn.ops.rms_norm")
-            transformers.models.llama.modeling_llama.LlamaRMSNorm = LlamaRMSNorm
-        except ImportError:
-            LOG.warning(
-                "optimized flash-attention RMSNorm not found (run `pip install 'git+https://github.com/Dao-AILab/flash-attention.git#egg=dropout_layer_norm&subdirectory=csrc/layer_norm'`)"
-            )
+        patch_llama_rms_norm()
 
 
 class FusedAttention(LlamaAttention):

src/axolotl/monkeypatch/llama_patch_multipack.py

@@ -9,18 +9,18 @@ from axolotl.monkeypatch.utils import (
 
 
 def hijack_llama_prepare_4d_mask():
-    import transformers.modeling_attn_mask_utils
-    import transformers.models.llama.modeling_llama
+    from transformers import modeling_attn_mask_utils
+    from transformers.models.llama import modeling_llama
 
-    transformers.models.llama.modeling_llama._prepare_4d_causal_attention_mask_for_sdpa = (  # pylint: disable=protected-access
+    modeling_llama._prepare_4d_causal_attention_mask_for_sdpa = (  # pylint: disable=protected-access
         patched_prepare_4d_causal_attention_mask_for_sdpa
     )
-    transformers.modeling_attn_mask_utils._prepare_4d_causal_attention_mask_for_sdpa = (  # pylint: disable=protected-access
+    modeling_attn_mask_utils._prepare_4d_causal_attention_mask_for_sdpa = (  # pylint: disable=protected-access
         patched_prepare_4d_causal_attention_mask_for_sdpa
     )
-    transformers.models.llama.modeling_llama._prepare_4d_causal_attention_mask = (  # pylint: disable=protected-access
+    modeling_llama._prepare_4d_causal_attention_mask = (  # pylint: disable=protected-access
         patched_prepare_4d_causal_attention_mask
     )
-    transformers.modeling_attn_mask_utils._prepare_4d_causal_attention_mask = (  # pylint: disable=protected-access
+    modeling_attn_mask_utils._prepare_4d_causal_attention_mask = (  # pylint: disable=protected-access
         patched_prepare_4d_causal_attention_mask
     )

src/axolotl/monkeypatch/mistral_attn_hijack_flash.py

@@ -2,6 +2,7 @@
 # pylint: disable=duplicate-code
 
 import logging
+from functools import partial
 from typing import List, Optional, Tuple, Union
 
 import torch
@@ -45,6 +46,15 @@ def replace_mistral_attn_with_flash_attn(
         )
 
 
+def patch_mistral_cross_entropy():
+    from flash_attn.losses.cross_entropy import CrossEntropyLoss
+
+    LOG.info("patching with flash_attn.losses.cross_entropy")
+    transformers.models.mistral.modeling_mistral.CrossEntropyLoss = partial(
+        CrossEntropyLoss, inplace_backward=True
+    )
+
+
 @torch.jit.script
 def _make_sliding_window_causal_mask(
     bsz: int,

src/axolotl/monkeypatch/multipack.py

@@ -10,11 +10,15 @@ from axolotl.monkeypatch.mixtral import patch_mixtral_moe_forward_zero3
 from axolotl.monkeypatch.utils import get_unpad_data
 
 SUPPORTED_MULTIPACK_MODEL_TYPES = [
+    "mllama_text_model",
+    "llama",
+    "mistral",
     "mixtral",
     "qwen2",
     "qwen2_moe",
     "falcon",
     "phi",
+    "phi3",
     "gemma",
     "gemma2",
     "gemmoe",
@@ -23,13 +27,36 @@ SUPPORTED_MULTIPACK_MODEL_TYPES = [
 ]
 
 
-def patch_for_multipack(model_type, model_name=None):
+def patch_for_multipack(model_type, model_name=None, is_remote_code=False):
+    if model_type == "gemmoe":
+        patch_remote(model_name, ".configuration_gemmoe", ".modeling_gemmoe")
+    elif model_type == "deepseek_v2":
+        patch_remote(model_name, ".configuration_deepseek", ".modeling_deepseek")
+    elif hasattr(transformers, "modeling_flash_attention_utils") and not is_remote_code:
+        transformers.modeling_flash_attention_utils._get_unpad_data = (  # pylint: disable=protected-access
+            get_unpad_data
+        )
+        if model_type == "mixtral" and is_deepspeed_zero3_enabled():
+            patch_mixtral_moe_forward_zero3()
+        return
+
+    # retain for legacy
     if model_type == "mixtral":
         transformers.models.mixtral.modeling_mixtral._get_unpad_data = (  # pylint: disable=protected-access
             get_unpad_data
         )
         if is_deepspeed_zero3_enabled():
             patch_mixtral_moe_forward_zero3()
+    elif model_type == "llama":
+        if hasattr(transformers.models.llama.modeling_llama, "_get_unpad_data"):
+            transformers.models.llama.modeling_llama._get_unpad_data = (  # pylint: disable=protected-access
+                get_unpad_data
+            )
+    elif model_type == "mistral":
+        if hasattr(transformers.models.mistral.modeling_mistral, "_get_unpad_data"):
+            transformers.models.llama.modeling_llama._get_unpad_data = (  # pylint: disable=protected-access
+                get_unpad_data
+            )
     elif model_type == "qwen2":
         transformers.models.qwen2.modeling_qwen2._get_unpad_data = (  # pylint: disable=protected-access
             get_unpad_data
@@ -58,12 +85,6 @@ def patch_for_multipack(model_type, model_name=None):
         transformers.models.starcoder2.modeling_starcoder2._get_unpad_data = (  # pylint: disable=protected-access
             get_unpad_data
         )
-    elif model_type == "gemmoe":
-        patch_remote(model_name, ".configuration_gemmoe", ".modeling_gemmoe")
-    elif model_type == "jamba":
-        patch_remote(model_name, ".configuration_jamba", ".modeling_jamba")
-    elif model_type == "deepseek_v2":
-        patch_remote(model_name, ".configuration_deepseek", ".modeling_deepseek")
 
 
 def patch_remote(model_name, config_name, modeling_name):

src/axolotl/monkeypatch/stablelm_attn_hijack_flash.py

@@ -16,6 +16,7 @@
 # This code is based off the following work:
 # https://github.com/huggingface/transformers/blob/main/src/transformers/models/llama/modeling_llama.py
 # https://github.com/huggingface/transformers/blob/main/src/transformers/models/gpt_neox/modeling_gpt_neox.py
+# pylint: disable=duplicate-code
 """ PyTorch StableLM Epoch model. """
 import importlib
 import math

src/axolotl/monkeypatch/unsloth_.py

@@ -1,21 +1,22 @@
 """module for patching with unsloth optimizations"""
 
 import inspect
-import logging
 import re
 import types
 from typing import Tuple
 
+import torch
+from accelerate.logging import get_logger
 from peft import PeftModelForCausalLM
+from torch import nn
 from transformers.models.llama.modeling_llama import (
     LlamaFlashAttention2,
     LlamaForCausalLM,
 )
 
-LOG = logging.getLogger("axolotl.monkeypatch.unsloth")
+LOG = get_logger("axolotl.monkeypatch.unsloth")
 
-ORIGINAL_CEL_CODE = """    if labels is not None:
-        # Shift so that tokens < n predict n
+ORIGINAL_CEL_CODE = """# Shift so that tokens < n predict n
         shift_logits = logits[..., :-1, :].contiguous()
         shift_labels = labels[..., 1:].contiguous()
         # Flatten the tokens
@@ -27,8 +28,7 @@ ORIGINAL_CEL_CODE = """    if labels is not None:
         loss = loss_fct(shift_logits, shift_labels)
 """
 
-PATCHED_CEL_CODE = """    if labels is not None:
-        shift_logits = logits[..., :-1, :].contiguous()
+PATCHED_CEL_CODE = """shift_logits = logits[..., :-1, :].contiguous()
         shift_labels = labels[..., 1:].contiguous()
         loss = fast_cross_entropy_loss(
             logits = shift_logits,
@@ -97,48 +97,51 @@ def check_self_attn_is_patchable() -> bool:
     return ORIGINAL_QKV_CODE in qkv and ORIGINAL_O_CODE in qkv
 
 
-def integrate_cross_entropy_loss_patch():
-    forward = get_forward_code()
-    LlamaForCausalLM._original_forward = forward  # pylint: disable=protected-access
-    forward, _ = detab_code(forward)
-    assert ORIGINAL_CEL_CODE in forward, "Original forward code not found"
+def integrate_cross_entropy_loss_patch(model_type: str = "llama") -> None:
+    if model_type == "llama":
+        forward = get_forward_code()
+        LlamaForCausalLM._original_forward = forward  # pylint: disable=protected-access
+        forward, _ = detab_code(forward)
+        assert ORIGINAL_CEL_CODE in forward, "Original forward code not found"
 
-    forward = forward.replace(
-        "@add_start_docstrings_to_model_forward(LLAMA_INPUTS_DOCSTRING)", ""
-    )
-    forward = forward.replace(
-        "@replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)",
-        "",
-    )
-    forward = forward.replace(ORIGINAL_CEL_CODE, PATCHED_CEL_CODE)
-    forward = forward.replace(
-        "def forward(",
-        "def fast_cross_entropy_loss_forward(",
-        1,
-    )
+        forward = forward.replace(
+            "@add_start_docstrings_to_model_forward(LLAMA_INPUTS_DOCSTRING)", ""
+        )
+        forward = forward.replace(
+            "@replace_return_docstrings(output_type=CausalLMOutputWithPast, config_class=_CONFIG_FOR_DOC)",
+            "",
+        )
+        forward = forward.replace(ORIGINAL_CEL_CODE, PATCHED_CEL_CODE)
+        forward = forward.replace(
+            "def forward(",
+            "def fast_cross_entropy_loss_forward(",
+            1,
+        )
 
-    # load imports necessary
-    import transformers.models.llama.modeling_llama
+        # load imports necessary
+        import transformers.models.llama.modeling_llama
 
-    items_to_import = []
-    for item in dir(transformers.models.llama.modeling_llama):
-        if item in forward:
-            items_to_import.append(item)
+        items_to_import = []
+        for item in dir(transformers.models.llama.modeling_llama):
+            if item in forward:
+                items_to_import.append(item)
 
-    exec(  # pylint: disable=exec-used  # nosec B102
-        "from unsloth.kernels.cross_entropy_loss import fast_cross_entropy_loss",
-        globals(),
-    )
+        exec(  # pylint: disable=exec-used  # nosec B102
+            "from unsloth.kernels.cross_entropy_loss import fast_cross_entropy_loss",
+            globals(),
+        )
 
-    exec(  # pylint: disable=exec-used  # nosec B102
-        "from transformers.models.llama.modeling_llama import ("
-        + ", ".join(x for x in items_to_import)
-        + ")",
-        globals(),
-    )
-    exec(forward, globals())  # pylint: disable=exec-used  # nosec B102
-    print("patching unsloth fast_cross_entropy_loss")
-    LlamaForCausalLM.forward = fast_cross_entropy_loss_forward  # pylint: disable=undefined-variable  # noqa: F821
+        exec(  # pylint: disable=exec-used  # nosec B102
+            "from transformers.models.llama.modeling_llama import ("
+            + ", ".join(x for x in items_to_import)
+            + ")",
+            globals(),
+        )
+        exec(forward, globals())  # pylint: disable=exec-used  # nosec B102
+        LOG.info("patching unsloth fast_cross_entropy_loss", main_process_only=True)
+        LlamaForCausalLM.forward = fast_cross_entropy_loss_forward  # pylint: disable=undefined-variable  # noqa: F821
+    else:
+        raise ValueError("Unsupported model type")
 
 
 def detab_code(code: str) -> Tuple[str, str]:
@@ -179,12 +182,30 @@ def patch_self_attn_lora():
         globals(),
     )
     exec(self_attn_forward, globals())  # pylint: disable=exec-used  # nosec B102
-    print("patching unsloth attn lora")
+    LOG.info("patching unsloth attn lora", main_process_only=True)
     LlamaFlashAttention2.forward = (
         unsloth_attn_forward  # pylint: disable=undefined-variable  # noqa: F821
     )
 
 
+def integrate_rope_embeddings():
+    import transformers.models.llama.modeling_llama
+    from unsloth.kernels.rope_embedding import fast_rope_embedding
+
+    def apply_rotary_pos_emb(  # pylint: disable=unused-argument
+        q,  # pylint: disable=invalid-name
+        k,  # pylint: disable=invalid-name
+        cos,
+        sin,
+        position_ids=None,
+        unsqueeze_dim=1,
+    ):
+        return fast_rope_embedding(q, k, cos, sin)
+
+    LOG.info("patching unsloth RoPE embeddings", main_process_only=True)
+    transformers.models.llama.modeling_llama.apply_rotary_pos_emb = apply_rotary_pos_emb
+
+
 def integrate_lora_mlp_patch(peft_model: PeftModelForCausalLM):
     if peft_model.base_model.config.model_type in ["llama", "mistral"]:
         from unsloth.kernels import apply_lora_mlp_swiglu
@@ -217,7 +238,7 @@ def integrate_lora_mlp_patch(peft_model: PeftModelForCausalLM):
         if is_mlp_lora and mlp_no_bias and mlp_not_dora:
             layer.mlp.forward = types.MethodType(apply_lora_mlp, layer.mlp)
         else:
-            logging.warning("unable to apply unsloth lora mlp patch to layer %d", idx)
+            LOG.warning("unable to apply unsloth lora mlp patch to layer %d", idx)
 
 
 def integrate_lora_patch(peft_model: PeftModelForCausalLM, cfg):
@@ -243,9 +264,7 @@ def integrate_lora_patch(peft_model: PeftModelForCausalLM, cfg):
                 layer.self_attn.apply_qkv = apply_lora_qkv
             else:
                 layer.self_attn.apply_qkv = original_apply_qkv
-                logging.warning(
-                    "unable to apply unsloth lora qkv patch to layer %d", idx
-                )
+                LOG.warning("unable to apply unsloth lora qkv patch to layer %d", idx)
         if cfg.unsloth_lora_o:
             layer_modules = [
                 getattr(layer.self_attn, linear_proj) for linear_proj in ["o_proj"]
@@ -264,6 +283,33 @@ def integrate_lora_patch(peft_model: PeftModelForCausalLM, cfg):
                 layer.self_attn.apply_o = apply_lora_o
             else:
                 layer.self_attn.apply_o = original_apply_o
-                logging.warning(
+                LOG.warning(
                     "unable to apply unsloth lora o_proj patch to layer %d", idx
                 )
+
+
+def patch_unsloth_layernorm():
+    try:
+        import transformers.models.llama.modeling_llama
+        from unsloth.kernels.rms_layernorm import Fast_RMS_Layernorm
+
+        class LlamaRMSNorm(nn.Module):
+            """LlamaRMSNorm"""
+
+            def __init__(self, hidden_size, eps=1e-6):
+                """
+                LlamaRMSNorm is equivalent to T5LayerNorm
+                """
+                super().__init__()
+                self.weight = nn.Parameter(torch.ones(hidden_size))
+                self.variance_epsilon = eps
+
+            def forward(self, hidden_states):
+                return Fast_RMS_Layernorm.apply(
+                    hidden_states, self.weight, self.variance_epsilon, False
+                )
+
+        LOG.info("patching with unsloth.kernels.rms_layernorm")
+        transformers.models.llama.modeling_llama.LlamaRMSNorm = LlamaRMSNorm
+    except ImportError:
+        LOG.warning("missing unsloth library")

src/axolotl/monkeypatch/utils.py

@@ -17,11 +17,9 @@ def get_max_seqlen_in_batch(attention_mask: torch.Tensor) -> torch.Tensor:
     max_num = int(torch.max(attention_mask).item())
     batch_size, _ = attention_mask.shape
     counts = torch.zeros((batch_size, max_num), dtype=torch.int32)
-
     for i in range(1, max_num + 1):
         mask = attention_mask == i
         counts[:, i - 1] = torch.sum(mask, dim=-1).to(dtype=torch.int32)
-
     result = counts.flatten()
     nonzero_indices = torch.nonzero(result).squeeze(-1)
     return result[nonzero_indices]

src/axolotl/prompt_strategies/__init__.py

@@ -9,7 +9,7 @@ from axolotl.prompt_strategies.user_defined import UserDefinedDatasetConfig
 LOG = logging.getLogger("axolotl.prompt_strategies")
 
 
-def load(strategy, tokenizer, cfg, ds_cfg):
+def load(strategy, tokenizer, cfg, ds_cfg, processor=None):
     try:
         load_fn = "load"
         if strategy.split(".")[-1].startswith("load_"):
@@ -24,6 +24,8 @@ def load(strategy, tokenizer, cfg, ds_cfg):
             sig = inspect.signature(func)
             if "ds_cfg" in sig.parameters:
                 load_kwargs["ds_cfg"] = ds_cfg
+            if "processor" in sig.parameters:
+                load_kwargs["processor"] = processor
         return func(tokenizer, cfg, **load_kwargs)
     except ModuleNotFoundError:
         return None

src/axolotl/prompt_strategies/chat_template.py

@@ -5,23 +5,30 @@ HF Chat Templates prompt strategy
 import logging
 from typing import Any, Dict, List, Optional
 
+from transformers import ProcessorMixin
+
 from axolotl.prompt_tokenizers import PromptTokenizingStrategy
-from axolotl.prompters import Prompter
+from axolotl.prompters import IGNORE_TOKEN_ID, Prompter
 from axolotl.utils.chat_templates import chat_templates
 
+# Configure the logger
 LOG = logging.getLogger("axolotl")
+LOG.setLevel(logging.INFO)
 
 
 class ChatTemplatePrompter(Prompter):
-    """prompter for HF chat templates"""
+    """Prompter for HF chat templates"""
 
     def __init__(
         self,
         tokenizer,
+        processor=None,
         chat_template=None,
         max_length=2048,
         message_field_role: str = "from",
         message_field_content: str = "value",
+        message_field_training: Optional[str] = None,
+        message_field_training_detail: Optional[str] = None,
         roles: Optional[Dict[str, List[str]]] = None,
         drop_system_message: bool = False,
     ):
@@ -37,16 +44,20 @@ class ChatTemplatePrompter(Prompter):
             }
         self.message_field_role = message_field_role
         self.message_field_content = message_field_content
+        self.message_field_training = message_field_training
+        self.message_field_training_detail = message_field_training_detail
         self.tokenizer = tokenizer
+        self.processor: ProcessorMixin = processor
         self.chat_template = chat_template
         self.max_length = max_length
         self.drop_system_message = drop_system_message
 
-    def build_prompt(self, conversation, add_generation_prompt=False):
+    def build_prompt(self, conversation, add_generation_prompt=False, images=None):
         turns = [
             {
                 "role": self.roles[t[self.message_field_role]],
                 "content": t[self.message_field_content],
+                "training": t.get(self.message_field_training, None),
             }
             for t in conversation
         ]
@@ -54,6 +65,28 @@ class ChatTemplatePrompter(Prompter):
         if self.drop_system_message and turns[0]["role"] == "system":
             turns = turns[1:]
 
+        if self.processor:
+            text = self.processor.apply_chat_template(
+                turns,
+                chat_template=self.chat_template,
+                tokenize=False,
+                add_generation_prompt=add_generation_prompt,
+            )
+            batch = self.processor(
+                text=text,
+                images=images,
+                return_tensors="pt",
+                truncation=True,
+                max_length=self.max_length,
+            )
+            # workaround since processor works in batches instead of single examples
+            for k, val in batch.items():
+                if k in ["pixel_values"]:
+                    batch[k] = val.tolist()
+                else:
+                    batch[k] = val.squeeze().tolist()
+            return batch
+
         return self.tokenizer.apply_chat_template(
             turns,
             truncation=True,
@@ -62,6 +95,108 @@ class ChatTemplatePrompter(Prompter):
             chat_template=self.chat_template,
         )
 
+    def get_offsets_for_train_detail(
+        self, text: str, train_details: List[Dict], mask_untrainable: bool = True
+    ) -> List[int]:
+        tokenized_output = self.tokenizer(
+            text, return_offsets_mapping=True, add_special_tokens=False
+        )
+        tokens = tokenized_output.tokens()
+        token_offsets = tokenized_output["offset_mapping"]
+
+        LOG.debug(f"Tokenizing text: {text}")
+        LOG.debug(f"Tokens: {tokens}")
+        # Adjust the end offsets. For some reason by default they are set to the same value as the start offsets.
+        for i in range(len(token_offsets) - 1):
+            token_offsets[i] = (token_offsets[i][0], token_offsets[i + 1][0] - 1)
+        # Ensure the last token's end offset is set correctly
+        token_offsets[-1] = (token_offsets[-1][0], len(text) - 1)
+        LOG.debug(f"Token offsets: {token_offsets}")
+
+        # Initialize all offsets as IGNORE_TOKEN_ID (not trained)
+        result = [IGNORE_TOKEN_ID] * len(token_offsets)
+
+        # Adjust train_details to align with token boundaries
+        adjusted_train_details = self.adjust_train_details(train_details, token_offsets)
+
+        for idx, (start, end) in enumerate(token_offsets):
+            for detail in adjusted_train_details:
+                # Check if the token is completely within the detail's range
+                if start >= detail["begin_offset"] and end <= detail["end_offset"]:
+                    if detail["train"] or not mask_untrainable:
+                        result[idx] = start
+                        LOG.debug(f"Token {idx} ({tokens[idx]}) marked for training")
+                    else:
+                        LOG.debug(
+                            f"Token {idx} ({tokens[idx]}) marked as non-trainable"
+                        )
+                elif start < detail["end_offset"] and end > detail["begin_offset"]:
+                    # Token partially overlaps with detail, always mark as non-trainable
+                    LOG.debug(
+                        f"Token {idx} ({tokens[idx]}) partially overlaps detail, marked as non-trainable"
+                    )
+
+        LOG.debug(f"Final result: {result}")
+        return result
+
+    def adjust_train_details(
+        self, train_details: List[Dict], token_offsets: List[tuple]
+    ) -> List[Dict]:
+        adjusted_details = []
+        for detail in train_details:
+            begin_offset = detail["begin_offset"]
+            end_offset = detail["end_offset"]
+
+            # Find the first token that starts after or at the begin_offset
+            begin_token = next(
+                (
+                    i
+                    for i, (t_start, t_end) in enumerate(token_offsets)
+                    if t_start >= begin_offset
+                ),
+                len(token_offsets),
+            )
+            if begin_token > 0 and token_offsets[begin_token - 1][1] > begin_offset:
+                begin_token -= 1
+
+            # Find the last token that ends before or at the end_offset
+            end_token = next(
+                (
+                    i
+                    for i in range(len(token_offsets) - 1, -1, -1)
+                    if token_offsets[i][1] <= end_offset
+                ),
+                -1,
+            )
+            if (
+                end_token < len(token_offsets) - 1
+                and token_offsets[end_token + 1][0] < end_offset
+            ):
+                end_token += 1
+
+            if begin_token <= end_token:
+                adjusted_begin = token_offsets[begin_token][0]
+                adjusted_end = token_offsets[end_token][1]
+
+                if adjusted_begin != begin_offset or adjusted_end != end_offset:
+                    LOG.warning(
+                        f"Adjusting detail offsets: ({begin_offset}, {end_offset}) -> ({adjusted_begin}, {adjusted_end})"
+                    )
+
+                adjusted_details.append(
+                    {
+                        "begin_offset": adjusted_begin,
+                        "end_offset": adjusted_end,
+                        "train": detail["train"],
+                    }
+                )
+            else:
+                LOG.warning(
+                    f"Could not adjust detail offsets: ({begin_offset}, {end_offset}). Skipping this detail."
+                )
+
+        return adjusted_details
+
 
 class ChatTemplateStrategy(PromptTokenizingStrategy):
     """
@@ -70,6 +205,20 @@ class ChatTemplateStrategy(PromptTokenizingStrategy):
 
     _messages = "conversations"
 
+    def __init__(
+        self,
+        prompter,
+        tokenizer,
+        train_on_inputs,
+        sequence_len,
+        roles_to_train=None,
+        train_on_eos=None,
+    ):
+        super().__init__(prompter, tokenizer, train_on_inputs, sequence_len)
+        self.roles_to_train = roles_to_train if roles_to_train is not None else []
+        self.train_on_eos = train_on_eos
+        self.images = "images"
+
     @property
     def messages(self):
         return self._messages
@@ -79,62 +228,212 @@ class ChatTemplateStrategy(PromptTokenizingStrategy):
         self._messages = messages
 
     def tokenize_prompt(self, prompt):
-        turns = self.get_conversation_thread(prompt)
-        prompt_ids = self.prompter.build_prompt(turns[:-1], add_generation_prompt=True)
+        # Old simple legacy behavior that works reliably.
+        if (
+            not self.roles_to_train
+            and not self.train_on_eos
+            and not self.prompter.message_field_training
+            and not self.prompter.message_field_training_detail
+        ):
+            turns = self.get_conversation_thread(prompt)
+            images = self.get_images(prompt)
+            prompt_ids = self.prompter.build_prompt(
+                turns[:-1],
+                add_generation_prompt=True,
+                images=images,
+            )
+            tokenized_res = self.prompter.build_prompt(turns, images=images)
+            tokenized_prompt = {}
+            if isinstance(tokenized_res, list):
+                input_ids = prompt_ids + tokenized_res[len(prompt_ids) :]
+                tokenized_prompt["input_ids"] = input_ids
+                tokenized_prompt["attention_mask"] = [1] * len(input_ids)
+            else:
+                input_ids = tokenized_res["input_ids"]
+                tokenized_prompt = tokenized_res
+
+            if not self.train_on_inputs:
+                user_prompt_len = len(prompt_ids)
+                labels = [-100] * user_prompt_len + input_ids[user_prompt_len:]
+            else:
+                labels = input_ids
+
+            tokenized_prompt["labels"] = labels
+
+            return tokenized_prompt
+
+        turns = prompt[self.messages]
         input_ids = self.prompter.build_prompt(turns)
+        labels = [IGNORE_TOKEN_ID] * len(input_ids)
 
-        if not self.train_on_inputs:
-            user_prompt_len = len(prompt_ids)
-            labels = [-100] * user_prompt_len + input_ids[user_prompt_len:]
-        else:
-            labels = input_ids
+        last_eos_idx = -1
+        for index, turn in enumerate(turns):
+            role = turn.get(self.prompter.message_field_role)
+            content = turn.get(self.prompter.message_field_content)
+            train_turn = turn.get(self.prompter.message_field_training)
+            train_detail = turn.get(self.prompter.message_field_training_detail)
 
-        tokenized_prompt = {
+            LOG.debug(
+                f"Processing turn {index}: role={role}, content={content}, train_turn={train_turn}, train_detail={train_detail}"
+            )
+
+            should_train = (
+                train_turn
+                if train_turn is not None
+                else (
+                    bool(train_detail is not None)
+                    if train_detail is not None
+                    else self.train_on_inputs or role in self.roles_to_train
+                )
+            )
+
+            LOG.debug(f"Should train: {should_train}")
+
+            turn_start_idx, turn_end_idx = self.find_turn(
+                conversation_ids=input_ids, turn=index, turn_content=turn
+            )
+
+            LOG.debug(f"Turn indices: start={turn_start_idx}, end={turn_end_idx}")
+
+            if should_train and turn_start_idx != -1 and turn_end_idx != -1:
+                if train_detail:
+                    token_offsets = self.prompter.get_offsets_for_train_detail(
+                        content, train_detail
+                    )
+                    LOG.debug(f"Token offsets: {token_offsets}")
+                    for i, offset in enumerate(token_offsets):
+                        if offset != IGNORE_TOKEN_ID and turn_start_idx + i < len(
+                            input_ids
+                        ):
+                            labels[turn_start_idx + i] = input_ids[turn_start_idx + i]
+                            LOG.debug(
+                                f"Label set at index {turn_start_idx + i}: {input_ids[turn_start_idx + i]}"
+                            )
+                else:
+                    labels[turn_start_idx:turn_end_idx] = input_ids[
+                        turn_start_idx:turn_end_idx
+                    ]
+                    LOG.debug(f"Labels set for range {turn_start_idx}:{turn_end_idx}")
+
+                LOG.debug(f"Labels after processing turn {index}: {labels}")
+
+            # Handle EOS token
+            eos_idx = self.find_eos_token(input_ids, turn_end_idx)
+            if eos_idx == turn_end_idx:
+                last_eos_idx = eos_idx
+                if self.train_on_eos == "all" or (
+                    self.train_on_eos == "turn" and should_train
+                ):
+                    labels[eos_idx] = input_ids[eos_idx]
+                    LOG.debug(f"EOS token set for training at index {eos_idx}")
+            else:
+                LOG.debug(
+                    f"EOS token missing after turn {turn}. eos_idx: {eos_idx}, turn_end_idx: {turn_end_idx}"
+                )
+
+        # Handle 'last' option for train_on_eos
+        if self.train_on_eos == "last" and last_eos_idx != -1:
+            labels[last_eos_idx] = input_ids[last_eos_idx]
+            LOG.debug(f"Last EOS token set for training at index {last_eos_idx}")
+
+        LOG.debug(f"Final labels: {labels}")
+
+        return {
             "input_ids": input_ids,
             "labels": labels,
             "attention_mask": [1] * len(input_ids),
         }
 
-        return tokenized_prompt
+    def find_eos_token(self, input_ids, start_idx):
+        eos_token_id = self.tokenizer.eos_token_id
+        for i in range(start_idx, len(input_ids)):
+            if input_ids[i] == eos_token_id:
+                return i
+        return -1
+
+    def find_turn(self, conversation_ids, turn, turn_content):
+        """
+        Locate the starting and ending indices of the specified turn in a conversation.
+
+        Args:
+            conversation_ids (list[int]): Token IDs representing the conversation.
+            turn (int): The turn number to locate (based on EOS tokens).
+            turn_content (str): String containing the content of the turn.
+
+        Returns:
+            tuple: (start_idx, end_idx) indices of the start and end of the turn content.
+                   Returns (-1, -1) if the turn content is not found.
+        """
+        content = turn_content.get(self.prompter.message_field_content, "")
+        content_ids = self.tokenizer.encode(content, add_special_tokens=False)
+
+        eos_token_id = self.tokenizer.eos_token_id
+        eos_count = 0
+        start_search_idx = 0
+
+        # Locate the starting index after the specified number of EOS tokens
+        for i, token_id in enumerate(conversation_ids):
+            if token_id == eos_token_id:
+                eos_count += 1
+                if eos_count == turn:
+                    start_search_idx = (
+                        i + 1
+                    )  # Start searching after the specified turn's EOS token
+                    break
+
+        # Find the start index of the content within the conversation
+        start_idx = -1
+        for i in range(start_search_idx, len(conversation_ids) - len(content_ids) + 1):
+            if conversation_ids[i : i + len(content_ids)] == content_ids:
+                start_idx = i
+                break
+
+        if start_idx != -1:
+            end_idx = start_idx + len(content_ids)
+        else:
+            end_idx = -1
+
+        return start_idx, end_idx
 
     def get_conversation_thread(self, prompt):
         return prompt[self.messages]
 
+    def get_images(self, prompt):
+        return prompt.get(self.images, None)
 
-def load(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
-    chat_template = (
-        ds_cfg["chat_template"] if ds_cfg and "chat_template" in ds_cfg else "chatml"
-    )
-    message_field_role = (
-        ds_cfg["message_field_role"]
-        if ds_cfg and "message_field_role" in ds_cfg
-        else "from"
-    )
-    message_field_content = (
-        ds_cfg["message_field_content"]
-        if ds_cfg and "message_field_content" in ds_cfg
-        else "value"
-    )
-    roles = ds_cfg["roles"] if ds_cfg and "roles" in ds_cfg else None
-    drop_system_message = (
-        ds_cfg["drop_system_message"]
-        if ds_cfg and "drop_system_message" in ds_cfg
-        else False
-    )
+
+def load(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None, processor=None):
+    ds_cfg = ds_cfg or {}
+
+    prompter_params = {
+        "tokenizer": tokenizer,
+        "chat_template": chat_templates(ds_cfg.get("chat_template", "chatml")),
+        "message_field_role": ds_cfg.get("message_field_role", "role"),
+        "message_field_content": ds_cfg.get("message_field_content", "content"),
+        "message_field_training": ds_cfg.get("message_field_training", None),
+        "message_field_training_detail": ds_cfg.get(
+            "message_field_training_detail",
+            None,
+        ),
+        "roles": ds_cfg.get("roles"),
+        "drop_system_message": ds_cfg.get("drop_system_message", False),
+        # we need to add one for detecting sequences with exceeding the `sequence_len` limit.
+        "max_length": cfg.sequence_len + 1,
+        "processor": processor,
+    }
+
+    strategy_params = {
+        "train_on_inputs": cfg.train_on_inputs,
+        "sequence_len": cfg.sequence_len,
+        "roles_to_train": ds_cfg.get("roles_to_train", []),
+        "train_on_eos": ds_cfg.get("train_on_eos", None),
+    }
 
     strategy = ChatTemplateStrategy(
-        ChatTemplatePrompter(
-            tokenizer,
-            chat_templates(chat_template),
-            message_field_role=message_field_role,
-            message_field_content=message_field_content,
-            roles=roles,
-            drop_system_message=drop_system_message,
-        ),
-        tokenizer,
-        cfg.train_on_inputs,
-        cfg.sequence_len,
+        ChatTemplatePrompter(**prompter_params), tokenizer=tokenizer, **strategy_params
     )
-    if ds_cfg and "field_messages" in ds_cfg and hasattr(strategy, "messages"):
+
+    if "field_messages" in ds_cfg and hasattr(strategy, "messages"):
         strategy.messages = ds_cfg["field_messages"]
+
     return strategy

src/axolotl/prompt_strategies/dpo/chat_template.py

@@ -0,0 +1,78 @@
+"""
+DPO prompt strategies for using tokenizer chat templates.
+"""
+
+from axolotl.utils.chat_templates import chat_templates
+
+
+def default(
+    cfg, dataset_idx=0, **kwargs
+):  # pylint: disable=possibly-unused-variable,unused-argument
+    ds_cfg = cfg["datasets"][dataset_idx]
+    chat_template_str = chat_templates(cfg.chat_template)
+
+    field_messages = ds_cfg.get("field_messages", "messages")
+    field_chosen = ds_cfg.get("field_chosen", "chosen")
+    field_rejected = ds_cfg.get("field_rejected", "rejected")
+    field_message_role = ds_cfg.get("message_field_role", "role")
+    field_message_content = ds_cfg.get("message_field_content", "content")
+    role_map_inv = ds_cfg.get(
+        "roles",
+        {
+            "user": ["user"],
+            "assistant": ["assistant"],
+            "system": ["system"],
+        },
+    )
+    role_map = {}
+    for target, sources in role_map_inv.items():
+        for source in sources:
+            role_map[source] = target
+
+    def transform_fn(sample, tokenizer=None):
+        messages = sample[field_messages]
+        messages = [
+            {
+                "role": role_map[m[field_message_role]],
+                "content": m[field_message_content],
+            }
+            for m in messages
+        ]
+        chosen = {
+            "role": role_map[sample[field_chosen][field_message_role]],
+            "content": sample[field_chosen][field_message_content],
+        }
+        rejected = {
+            "role": role_map[sample[field_rejected][field_message_role]],
+            "content": sample[field_rejected][field_message_content],
+        }
+
+        result = {}
+        result["prompt"] = tokenizer.apply_chat_template(
+            messages,
+            add_generation_prompt=True,
+            chat_template=chat_template_str,
+            tokenize=False,
+        )
+
+        result["chosen"] = tokenizer.apply_chat_template(
+            [chosen],
+            add_generation_prompt=False,
+            chat_template=chat_template_str,
+            tokenize=False,
+        )
+        chosen_strip_index = result["chosen"].find(chosen["content"])
+        result["chosen"] = result["chosen"][chosen_strip_index:].rstrip()
+
+        result["rejected"] = tokenizer.apply_chat_template(
+            [rejected],
+            add_generation_prompt=False,
+            chat_template=chat_template_str,
+            tokenize=False,
+        )
+        rejected_strip_index = result["rejected"].find(rejected["content"])
+        result["rejected"] = result["rejected"][rejected_strip_index:].rstrip()
+
+        return result
+
+    return transform_fn

src/axolotl/prompters.py

@@ -65,8 +65,10 @@ class AlpacaPrompter(Prompter):
             self.system_format = "<|im_start|>system\n{system}<|im_end|>\n"
         elif self.prompt_style == PromptStyle.PHI.value:
             self.turn_format = "<|user|>\n{instruction}<|end|>{input}<|assistant|>"
-            self.turn_no_input_format = "<|user|>\n{instruction}<|end|><|assistant|>"
-            self.system_format = "<|system|>{system}\n"
+            self.turn_no_input_format = (
+                "<|user|>\n{instruction}<|end|>\n<|assistant|>\n"
+            )
+            self.system_format = "<|system|>\n{system}<|end|>\n"
 
     def _build_result(self, instruction, input_text, output):
         # returns the full prompt from instruction and optional input
@@ -350,9 +352,12 @@ class ShareGPTPrompter(Prompter):  # pylint: disable=too-few-public-methods
                         "Please help us by creating an Issue to add support for this conversation type."
                     )
 
-                role = CONVERSATION_ROLE_FORMAT[self._conversation.name].format(
-                    ROLE=from_role
-                )
+                if self._conversation.name in ["llama3"]:
+                    role = from_role
+                else:
+                    role = CONVERSATION_ROLE_FORMAT[self._conversation.name].format(
+                        ROLE=from_role
+                    )
 
             if len(conv.messages) > 0 and ((role == conv.messages[-1][0])):
                 if (

src/axolotl/train.py

@@ -12,6 +12,7 @@ import torch
 import transformers.modelcard
 from accelerate import Accelerator
 from accelerate.logging import get_logger
+from accelerate.utils import save_fsdp_model
 from datasets import Dataset
 from peft import PeftModel
 from pkg_resources import get_distribution  # type: ignore
@@ -19,10 +20,11 @@ from transformers import PreTrainedModel, PreTrainedTokenizer
 from transformers.integrations.deepspeed import is_deepspeed_zero3_enabled
 
 from axolotl.common.cli import TrainerCliArgs
+from axolotl.core.tokenizer_utils import fix_untrained_tokens
 from axolotl.logging_config import configure_logging
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.freeze import freeze_layers_except
-from axolotl.utils.models import load_model, load_tokenizer
+from axolotl.utils.models import load_model, load_processor, load_tokenizer
 from axolotl.utils.trainer import setup_trainer
 
 try:
@@ -52,12 +54,24 @@ class TrainDatasetMeta:
 def train(
     *, cfg: DictDefault, cli_args: TrainerCliArgs, dataset_meta: TrainDatasetMeta
 ) -> Tuple[Union[PeftModel, PreTrainedModel], PreTrainedTokenizer]:
+    # enable expandable segments for cuda allocation to improve VRAM usage
+    torch_version = torch.__version__.split(".")
+    torch_major, torch_minor = int(torch_version[0]), int(torch_version[1])
+    if torch_major == 2 and torch_minor >= 2:
+        if os.getenv("PYTORCH_CUDA_ALLOC_CONF") is None:
+            os.environ[
+                "PYTORCH_CUDA_ALLOC_CONF"
+            ] = "expandable_segments:True,roundup_power2_divisions:16"
+
     # load the tokenizer first
     LOG.debug(
         f"loading tokenizer... {cfg.tokenizer_config or cfg.base_model_config}",
         main_process_only=True,
     )
     tokenizer = load_tokenizer(cfg)
+    processor = None
+    if cfg.is_multimodal:
+        processor = load_processor(cfg, tokenizer)
 
     train_dataset = dataset_meta.train_dataset
     eval_dataset = dataset_meta.eval_dataset
@@ -85,7 +99,9 @@ def train(
     LOG.debug(msg)
     # we wait unitl the last possible moment to setup Accelerator
     Accelerator()
-    model, peft_config = load_model(cfg, tokenizer, inference=cli_args.inference)
+    model, peft_config = load_model(
+        cfg, tokenizer, processor=processor, inference=cli_args.inference
+    )
     model.generation_config.do_sample = True
 
     model_ref = None
@@ -111,9 +127,17 @@ def train(
         eval_dataset,
         (model, model_ref, peft_config),
         tokenizer,
+        processor,
         total_num_steps,
     )
 
+    if cfg.fix_untrained_tokens:
+        fix_untrained_tokens(model, tokenizer, train_dataset)
+        if cfg.local_rank == 0:
+            model.save_pretrained(
+                str(Path(cfg.output_dir)), safe_serialization=safe_serialization
+            )
+
     # go ahead and presave, so we have the adapter config available to inspect
     if peft_config:
         LOG.info(f"Pre-saving adapter config to {cfg.output_dir}")
@@ -177,9 +201,12 @@ def train(
         if hasattr(module, "_post_training"):
             module._post_training(model, name)  # pylint: disable=protected-access
 
+    state_dict_type = "FULL_STATE_DICT"
     if trainer.is_fsdp_enabled:
-        trainer.accelerator.state.fsdp_plugin.set_state_dict_type("FULL_STATE_DICT")
-        LOG.info("Set FSDP state dict type to FULL_STATE_DICT for saving.")
+        if cfg.fsdp_final_state_dict_type:
+            state_dict_type = cfg.fsdp_final_state_dict_type
+        trainer.accelerator.state.fsdp_plugin.set_state_dict_type(state_dict_type)
+        LOG.info(f"Set FSDP state dict type to {state_dict_type} for saving.")
 
     if cfg.relora_steps:
         if cfg.adapter == "lora" and not (cfg.load_in_4bit or cfg.load_in_8bit):
@@ -191,30 +218,38 @@ def train(
     # TODO do we need this fix? https://huggingface.co/docs/accelerate/usage_guides/fsdp#saving-and-loading
     # only save on rank 0, otherwise it corrupts output on multi-GPU when multiple processes attempt to write the same file
     if cfg.fsdp:
-        trainer.save_model(cfg.output_dir)
+        if (
+            state_dict_type == "SHARDED_STATE_DICT"
+            and cfg.fsdp_config.fsdp_state_dict_type == "SHARDED_STATE_DICT"
+        ):
+            save_fsdp_model(
+                trainer.accelerator.state.fsdp_plugin,
+                trainer.accelerator,
+                trainer.model,
+                cfg.output_dir,
+            )
+        elif state_dict_type == "FULL_STATE_DICT":
+            trainer.save_model(cfg.output_dir)
     elif cfg.deepspeed and is_deepspeed_zero3_enabled():
         # Copied over from: https://github.com/huggingface/accelerate/blob/5ae611118057232f441055f7ef9ba0b0f2b8d533/docs/source/usage_guides/deepspeed.md#saving-and-loading
         trainer.accelerator.wait_for_everyone()
-        unwrapped_model = trainer.accelerator.unwrap_model(trainer.model_wrapped)
+        trainer.save_model(cfg.output_dir)
 
         # the trainer saved a model.safetensors file in the output directory,
-        # but it is a proxy model and should be deleted
-        if os.path.exists(os.path.join(cfg.output_dir, "model.safetensors")):
+        # but it is most likely a proxy model and if so, should be deleted
+        maybe_proxy = os.path.exists(os.path.join(cfg.output_dir, "model.safetensors"))
+        maybe_sharded = os.path.exists(
+            os.path.join(cfg.output_dir, "model.safetensors.index.json")
+        )
+
+        if maybe_proxy and maybe_sharded:
             LOG.info(f"Deleting {os.path.join(cfg.output_dir, 'model.safetensors')}")
             LOG.info("This is a proxy model and should be deleted")
-            os.remove(os.path.join(cfg.output_dir, "model.safetensors"))
+            try:
+                os.remove(os.path.join(cfg.output_dir, "model.safetensors"))
+            except FileNotFoundError:
+                pass
 
-        # Saves the whole/unpartitioned fp16 model when in ZeRO Stage-3 to the output directory if
-        # `stage3_gather_16bit_weights_on_model_save` is True in DeepSpeed Config file or
-        # `zero3_save_16bit_model` is True in DeepSpeed Plugin.
-        # For Zero Stages 1 and 2, models are saved as usual in the output directory.
-        # The model name saved is `pytorch_model.bin`
-        unwrapped_model.save_pretrained(
-            cfg.output_dir,
-            is_main_process=trainer.accelerator.is_main_process,
-            save_function=trainer.accelerator.save,
-            state_dict=trainer.accelerator.get_state_dict(trainer.model_wrapped),
-        )
     elif cfg.local_rank == 0:
         if cfg.flash_optimum and BetterTransformer:
             model = BetterTransformer.reverse(model)

src/axolotl/utils/collators/__init__.py

@@ -0,0 +1,10 @@
+"""
+shared axolotl collators for multipack, mamba, multimodal
+"""
+from .batching import (  # noqa: F401
+    BatchSamplerDataCollatorForSeq2Seq,
+    DataCollatorForSeq2Seq,
+    PretrainingBatchSamplerDataCollatorForSeq2Seq,
+    V2BatchSamplerDataCollatorForSeq2Seq,
+)
+from .mamba import MambaDataCollator  # noqa: F401

src/axolotl/utils/collators/batching.py

@@ -1,17 +1,14 @@
 """
 DataCollator for axolotl to pad labels and position_ids for packed sequences
 """
+
 from dataclasses import dataclass
-from typing import Any, Dict, Optional, Sequence, Union
+from typing import Any, Optional, Union
 
 import numpy as np
-import torch
-import transformers
 from transformers import PreTrainedTokenizerBase
 from transformers.utils import PaddingStrategy
 
-IGNORE_INDEX = -100
-
 
 @dataclass
 class DataCollatorForSeq2Seq:
@@ -183,34 +180,6 @@ class V2BatchSamplerDataCollatorForSeq2Seq(DataCollatorForSeq2Seq):
         return super().__call__(out_features, return_tensors=return_tensors)
 
 
-@dataclass
-class MambaDataCollator:
-    """
-    Collator for State Space Models (Mamba)
-    """
-
-    tokenizer: transformers.PreTrainedTokenizer
-
-    def __call__(self, instances: Sequence[Dict]) -> Dict[str, torch.Tensor]:
-        input_ids, labels = tuple(
-            [torch.LongTensor(instance[key]) for instance in instances]
-            for key in ("input_ids", "labels")
-        )
-        input_ids = torch.nn.utils.rnn.pad_sequence(
-            input_ids,
-            batch_first=True,
-            padding_value=self.tokenizer.pad_token_id,
-        )
-        labels = torch.nn.utils.rnn.pad_sequence(
-            labels, batch_first=True, padding_value=IGNORE_INDEX
-        )
-
-        return {
-            "input_ids": input_ids,
-            "labels": labels,
-        }
-
-
 @dataclass
 class PretrainingBatchSamplerDataCollatorForSeq2Seq(DataCollatorForSeq2Seq):
     """

src/axolotl/utils/collators/core.py

@@ -0,0 +1,4 @@
+"""
+basic shared collator constants
+"""
+IGNORE_INDEX = -100

src/axolotl/utils/collators/mamba.py

@@ -0,0 +1,38 @@
+"""
+collators for Mamba
+"""
+from dataclasses import dataclass
+from typing import Dict, Sequence
+
+import torch
+import transformers
+
+from axolotl.utils.collators.core import IGNORE_INDEX
+
+
+@dataclass
+class MambaDataCollator:
+    """
+    Collator for State Space Models (Mamba)
+    """
+
+    tokenizer: transformers.PreTrainedTokenizer
+
+    def __call__(self, instances: Sequence[Dict]) -> Dict[str, torch.Tensor]:
+        input_ids, labels = tuple(
+            [torch.LongTensor(instance[key]) for instance in instances]
+            for key in ("input_ids", "labels")
+        )
+        input_ids = torch.nn.utils.rnn.pad_sequence(
+            input_ids,
+            batch_first=True,
+            padding_value=self.tokenizer.pad_token_id,
+        )
+        labels = torch.nn.utils.rnn.pad_sequence(
+            labels, batch_first=True, padding_value=IGNORE_INDEX
+        )
+
+        return {
+            "input_ids": input_ids,
+            "labels": labels,
+        }

src/axolotl/utils/collators/mm_chat.py

@@ -0,0 +1,179 @@
+"""
+Collators for multi-modal chat messages and packing
+"""
+from dataclasses import dataclass
+from typing import Any, Dict, List, Optional, Union
+
+from transformers import PreTrainedTokenizerBase, ProcessorMixin
+from transformers.data.data_collator import DataCollatorMixin
+from transformers.utils import PaddingStrategy
+
+
+@dataclass
+class MultiModalChatDataCollator(DataCollatorMixin):
+    """
+    Collator for multi-modal chat messages
+    """
+
+    tokenizer: PreTrainedTokenizerBase
+    processor: ProcessorMixin
+    return_tensors: str = "pt"
+    chat_template: Optional[str] = None
+    packing: bool = False
+    sequence_length: Optional[int] = None
+    max_images: int = -1
+    padding: Union[bool, str, PaddingStrategy] = True
+    pad_to_multiple_of: Optional[int] = None
+
+    def __post_init__(self):
+        if self.packing:
+            raise ValueError("Packing is currently not supported.")
+
+    def torch_call(
+        self, examples: List[Union[List[int], Any, Dict[str, Any]]]
+    ) -> Dict[str, Any]:
+        # Handle dict or lists with proper padding and conversion to tensor.
+        if self.packing:
+            return self.__class__.process_rows_packing(
+                examples,
+                self.processor,
+                self.chat_template,
+                self.max_images,
+                self.sequence_length,
+            )
+
+        return self.__class__.process_rows(
+            examples, self.processor, self.chat_template, self.max_images
+        )
+
+    @staticmethod
+    def process_rows_packing(
+        examples,
+        processor,
+        chat_template,
+        max_images,
+        sequence_length,
+        length_only=False,
+    ):
+        import torch
+
+        # Perform sample packing within a batch
+
+        if processor.tokenizer.sep_token is None:
+            sep_token = "[SEP]"
+            processor.tokenizer.add_tokens([sep_token])
+            processor.tokenizer.sep_token = sep_token
+        sep_token_id = processor.tokenizer.convert_tokens_to_ids(
+            processor.tokenizer.sep_token
+        )
+        pad_token_id = processor.tokenizer.pad_token_id
+
+        texts = [
+            processor.apply_chat_template(
+                example["messages"], chat_template=chat_template, tokenize=False
+            )
+            for example in examples
+        ]
+        images = [example["images"] for example in examples]
+
+        if max_images > 0:
+            images = [img_batch[:max_images] for img_batch in images]
+
+        batch = processor(text=texts, images=images, padding=False)
+
+        n_sequence = len(examples)
+        n_seq_in_batch = 0
+        pack_len = 0
+        features_pack = {}
+        packed = {}
+        features = list[batch.keys()]
+        for feature in features:
+            features_pack[feature] = []
+            packed[feature] = []
+        features.remove("input_ids")
+
+        for seq_in_batch_id in range(n_sequence):
+            next_seq_len = len(batch["input_ids"][seq_in_batch_id])
+            if not pack_len + next_seq_len + 1 < sequence_length:
+                n_seq_in_batch += 1
+                pack_len += next_seq_len + 1
+                features_pack["input_ids"] += batch["input_ids"][seq_in_batch_id] + [
+                    sep_token_id
+                ]
+
+                """
+                Do something with attention mask and cross-attention
+                """
+
+                for feature in features:
+                    features_pack[feature] += batch[feature][seq_in_batch_id]
+
+            else:
+                for _ in range(sequence_length - pack_len):
+                    features_pack["input_ids"] += [pad_token_id]
+
+                packed["input_ids"].append(
+                    torch.tensor(features_pack["input_ids"].copy())
+                )
+
+                for feature in features:
+                    packed[feature].append(torch.tensor(features_pack[feature].copy()))
+                    features_pack[feature] = []
+
+                pack_len = 0
+
+        image_token_id = processor.tokenizer.convert_tokens_to_ids(
+            processor.image_token
+        )
+        labels = [pack.clone() for pack in packed["input_ids"]]
+        for label_id, label in enumerate(labels):
+            labels[label_id][label == processor.tokenizer.pad_token_id] = -100  #
+            # Ignore the image token index in the loss computation (model specific)
+
+            labels[label_id][label == image_token_id] = -100
+        packed["labels"] = labels
+
+        if length_only:
+            return {
+                "length": [len(sample["input_ids"]) for sample in batch["input_ids"]]
+            }
+        return packed
+
+    @staticmethod
+    def process_rows(examples, processor, chat_template, max_images, length_only=False):
+        # HINT: use `_torch_collate_batch` to stack and pad tensors
+        # see also DataCollatorWithFlattening and DefaultDataCollator
+
+        # *** This is COPIED from the trl example sft_vlm.py code ***
+        # use this as a starting point
+
+        # Get the texts and images, and apply the chat template
+        texts = [
+            processor.apply_chat_template(
+                example["messages"], chat_template=chat_template, tokenize=False
+            )
+            for example in examples
+        ]
+        images = [example["images"] for example in examples]
+
+        if max_images > 0:
+            images = [img_batch[:max_images] for img_batch in images]
+
+        # Tokenize the texts and process the images
+        batch = processor(text=texts, images=images, return_tensors="pt", padding=True)
+
+        # The labels are the input_ids, and we mask the padding tokens in the loss computation
+        labels = batch["input_ids"].clone()
+        labels[labels == processor.tokenizer.pad_token_id] = -100  #
+        # Ignore the image token index in the loss computation (model specific)
+        image_token_id = processor.tokenizer.convert_tokens_to_ids(
+            processor.image_token
+        )
+        labels[labels == image_token_id] = -100
+        batch["labels"] = labels
+
+        if length_only:
+            return {
+                "length": [len(sample["input_ids"]) for sample in batch["input_ids"]]
+            }
+        return batch

src/axolotl/utils/config/__init__.py

@@ -8,11 +8,14 @@ from typing import Optional
 import torch
 from transformers.utils import is_torch_bf16_gpu_available
 
+from axolotl.integrations.config import merge_input_args
 from axolotl.utils.bench import log_gpu_memory_usage
+from axolotl.utils.config.models.input.v0_4_1 import SUPPORTED_METRICS
 from axolotl.utils.config.models.input.v0_4_1 import (
-    SUPPORTED_METRICS,
-    AxolotlConfigWCapabilities,
-    AxolotlInputConfig,
+    AxolotlConfigWCapabilities as AxolotlConfigWCapabilitiesBase,
+)
+from axolotl.utils.config.models.input.v0_4_1 import (
+    AxolotlInputConfig as AxolotlInputConfigBase,
 )
 from axolotl.utils.dict import DictDefault
 from axolotl.utils.models import load_model_config
@@ -118,15 +121,36 @@ def normalize_config(cfg):
         cfg.base_model_config = cfg.base_model
 
     model_config = load_model_config(cfg)
-    cfg.model_config_type = model_config.model_type
 
     cfg.tokenizer_config = (
         cfg.tokenizer_config or cfg.base_model_config or cfg.base_model
     )
 
+    cfg.is_multimodal = (
+        hasattr(model_config, "model_type")
+        and model_config.model_type in ["llava", "mllama"]
+        or any(
+            multimodal_name in cfg.base_model.lower()
+            for multimodal_name in [
+                "pixtral",
+            ]
+        )
+        or cfg.is_multimodal
+    )
+    if cfg.is_multimodal:
+        cfg.processor_config = (
+            cfg.processor_config or cfg.base_model_config or cfg.base_model
+        )
+        model_config = model_config.text_config
+
+    cfg.model_config_type = model_config.model_type
+
     # figure out if the model is llama
     cfg.is_llama_derived_model = (
-        (hasattr(model_config, "model_type") and model_config.model_type == "llama")
+        (
+            hasattr(model_config, "model_type")
+            and model_config.model_type == ["llama", "mllama_text_model"]
+        )
         or cfg.is_llama_derived_model
         or "llama" in cfg.base_model.lower()
         or (cfg.type_of_model and "llama" in cfg.type_of_model.lower())
@@ -207,6 +231,15 @@ def normalize_cfg_datasets(cfg):
 
 
 def validate_config(cfg: DictDefault, capabilities: Optional[dict] = None):
+    AxolotlConfigWCapabilities = AxolotlConfigWCapabilitiesBase
+    AxolotlInputConfig = AxolotlInputConfigBase
+
+    if cfg.plugins:
+        (
+            AxolotlConfigWCapabilities,  # pylint: disable=invalid-name
+            AxolotlInputConfig,  # pylint: disable=invalid-name
+        ) = merge_input_args()
+
     if capabilities:
         return DictDefault(
             dict(

src/axolotl/utils/config/models/input/v0_4_1/__init__.py

@@ -7,6 +7,7 @@ Module for pydantic models for configuration
 import logging
 import os
 from enum import Enum
+from importlib.metadata import version
 from typing import Any, Dict, List, Literal, Optional, Tuple, Union
 
 from pydantic import BaseModel, Field, conlist, field_validator, model_validator
@@ -77,6 +78,7 @@ class PretrainingDataset(BaseModel):
     split: Optional[str] = "train"
     text_column: Optional[str] = "text"
     type: Optional[str] = "pretrain"
+    trust_remote_code: Optional[bool] = False
 
 
 class UserDefinedPrompterType(BaseModel):
@@ -114,10 +116,16 @@ class SFTDataset(BaseModel):
     field_messages: Optional[str] = None
     message_field_role: Optional[str] = None
     message_field_content: Optional[str] = None
+    message_field_training: Optional[str] = None
+    message_field_training_detail: Optional[str] = None
+    roles_to_train: Optional[List[str]] = None
+    train_on_eos: Optional[str] = None
 
     roles: Optional[Dict[str, List[str]]] = None
     drop_system_message: Optional[bool] = None
 
+    trust_remote_code: Optional[bool] = False
+
 
 class UserDefinedDPOType(BaseModel):
     """User defined typing for DPO"""
@@ -158,6 +166,7 @@ class KTODataset(BaseModel):
     split: Optional[str] = None
     type: Optional[Union[UserDefinedKTOType, str]] = None
     data_files: Optional[List[str]] = None
+    trust_remote_code: Optional[bool] = False
 
 
 class RLType(str, Enum):
@@ -167,6 +176,7 @@ class RLType(str, Enum):
     ipo = "ipo"  # pylint: disable=invalid-name
     orpo = "orpo"  # pylint: disable=invalid-name
     kto = "kto"  # pylint: disable=invalid-name
+    simpo = "simpo"  # pylint: disable=invalid-name
 
 
 class ChatTemplate(str, Enum):
@@ -178,7 +188,11 @@ class ChatTemplate(str, Enum):
     gemma = "gemma"  # pylint: disable=invalid-name
     cohere = "cohere"  # pylint: disable=invalid-name
     llama3 = "llama3"  # pylint: disable=invalid-name
+    llama3_2_vision = "llama3_2_vision"  # pylint: disable=invalid-name
     phi_3 = "phi_3"  # pylint: disable=invalid-name
+    phi_35 = "phi_35"  # pylint: disable=invalid-name
+    deepseek_v2 = "deepseek_v2"  # pylint: disable=invalid-name
+    jamba = "jamba"  # pylint: disable=invalid-name
 
 
 class LoftQConfig(BaseModel):
@@ -215,16 +229,23 @@ class LoraConfig(BaseModel):
     lora_r: Optional[int] = None
     lora_alpha: Optional[int] = None
     lora_fan_in_fan_out: Optional[bool] = None
-    lora_target_modules: Optional[List[str]] = None
+    lora_target_modules: Optional[Union[str, List[str]]] = None
     lora_target_linear: Optional[bool] = None
     lora_modules_to_save: Optional[List[str]] = None
     lora_dropout: Optional[float] = 0.0
     peft_layers_to_transform: Optional[List[int]] = None
+    peft_layers_pattern: Optional[List[str]] = None
     peft: Optional[PeftConfig] = None
     peft_use_dora: Optional[bool] = None
     peft_use_rslora: Optional[bool] = None
     peft_layer_replication: Optional[List[Tuple[int, int]]] = None
 
+    qlora_sharded_model_loading: Optional[bool] = Field(
+        default=False,
+        metadata={
+            "help": "load qlora model in sharded format for FSDP using answer.ai technique."
+        },
+    )
     lora_on_cpu: Optional[bool] = None
     gptq: Optional[bool] = None
     bnb_config_kwargs: Optional[Dict[str, Any]] = None
@@ -279,6 +300,13 @@ class LoraConfig(BaseModel):
                     raise ValueError("Require cfg.load_in_4bit to be True for qlora")
         return self
 
+    @field_validator("loraplus_lr_embedding")
+    @classmethod
+    def convert_loraplus_lr_embedding(cls, loraplus_lr_embedding):
+        if loraplus_lr_embedding and isinstance(loraplus_lr_embedding, str):
+            loraplus_lr_embedding = float(loraplus_lr_embedding)
+        return loraplus_lr_embedding
+
 
 class ReLoRAConfig(BaseModel):
     """ReLoRA configuration subset"""
@@ -302,8 +330,13 @@ class ModelInputConfig(BaseModel):
     tokenizer_type: Optional[str] = Field(
         default=None, metadata={"help": "transformers tokenizer class"}
     )
+    processor_type: Optional[str] = Field(
+        default=None, metadata={"help": "transformers processor class"}
+    )
     trust_remote_code: Optional[bool] = None
 
+    model_kwargs: Optional[Dict[str, Any]] = None
+
     @field_validator("trust_remote_code")
     @classmethod
     def hint_trust_remote_code(cls, trust_remote_code):
@@ -335,13 +368,24 @@ class HyperparametersConfig(BaseModel):
         },
     )
 
+    auto_find_batch_size: Optional[bool] = None
+
     train_on_inputs: Optional[bool] = False
     group_by_length: Optional[bool] = None
 
     learning_rate: Union[str, float]
     weight_decay: Optional[float] = 0.0
     optimizer: Optional[
-        Union[OptimizerNames, Literal["lion_pytorch"]]
+        Union[
+            OptimizerNames,
+            Literal[
+                "lion_pytorch",
+                "optimi_adamw",
+                "ao_adamw_4bit",
+                "ao_adamw_8bit",
+                "ao_adamw_fp8",
+            ],
+        ]
     ] = OptimizerNames.ADAMW_HF.value
     optim_args: Optional[Union[str, Dict[str, Any]]] = Field(
         default=None, metadata={"help": "Optional arguments to supply to optimizer."}
@@ -353,7 +397,7 @@ class HyperparametersConfig(BaseModel):
         },
     )
     torchdistx_path: Optional[str] = None
-    lr_scheduler: Optional[SchedulerType] = "cosine"
+    lr_scheduler: Optional[Union[SchedulerType, Literal["one_cycle"]]] = "cosine"
     lr_scheduler_kwargs: Optional[Dict[str, Any]] = None
     lr_quadratic_warmup: Optional[bool] = None
     cosine_min_lr_ratio: Optional[float] = None
@@ -491,6 +535,7 @@ class AxolotlInputConfig(
     dataset_prepared_path: Optional[str] = None
     dataset_shard_num: Optional[int] = None
     dataset_shard_idx: Optional[int] = None
+    skip_prepare_dataset: Optional[bool] = False
 
     pretraining_dataset: Optional[  # type: ignore
         conlist(Union[PretrainingDataset, SFTDataset], min_length=1)
@@ -504,6 +549,8 @@ class AxolotlInputConfig(
     dataloader_prefetch_factor: Optional[int] = None
     dataloader_drop_last: Optional[bool] = None
 
+    accelerator_config: Optional[Dict[str, Any]] = None
+
     remove_unused_columns: Optional[bool] = None
 
     push_dataset_to_hub: Optional[str] = None
@@ -561,6 +608,7 @@ class AxolotlInputConfig(
     eval_sample_packing: Optional[bool] = None
     pad_to_sequence_len: Optional[bool] = None
     curriculum_sampling: Optional[bool] = None
+    multipack_real_batches: Optional[bool] = None
 
     # for PoSE context length extension
     use_pose: Optional[bool] = None
@@ -586,14 +634,21 @@ class AxolotlInputConfig(
     flash_attn_fuse_mlp: Optional[bool] = None
     flash_optimum: Optional[bool] = None
 
+    eager_attention: Optional[bool] = None
+
     unsloth_cross_entropy_loss: Optional[bool] = None
     unsloth_lora_mlp: Optional[bool] = None
     unsloth_lora_qkv: Optional[bool] = None
     unsloth_lora_o: Optional[bool] = None
+    unsloth_rms_norm: Optional[bool] = None
+    unsloth_rope: Optional[bool] = None
 
     deepspeed: Optional[Union[str, Dict[str, Any]]] = None
     fsdp: Optional[List[str]] = None
     fsdp_config: Optional[Dict[str, Any]] = None
+    fsdp_final_state_dict_type: Optional[
+        Literal["FULL_STATE_DICT", "LOCAL_STATE_DICT", "SHARDED_STATE_DICT"]
+    ] = None
 
     val_set_size: Optional[float] = Field(default=0.0)
 
@@ -602,6 +657,9 @@ class AxolotlInputConfig(
 
     torch_compile: Optional[bool] = None
     torch_compile_backend: Optional[str] = None
+    torch_compile_mode: Optional[
+        Literal["default", "reduce-overhead", "max-autotune"]
+    ] = None
 
     max_steps: Optional[int] = None
     warmup_steps: Optional[int] = None
@@ -623,6 +681,8 @@ class AxolotlInputConfig(
 
     orpo_alpha: Optional[float] = None
     rpo_alpha: Optional[float] = None
+    simpo_gamma: Optional[float] = None
+    cpo_alpha: Optional[float] = None
 
     kto_desirable_weight: Optional[float] = None
     kto_undesirable_weight: Optional[float] = None
@@ -637,6 +697,8 @@ class AxolotlInputConfig(
     chat_template: Optional[ChatTemplate] = None
     default_system_message: Optional[str] = None
 
+    fix_untrained_tokens: Optional[bool] = None
+
     # INTERNALS - document for now, generally not set externally
     is_preprocess: Optional[bool] = None
 
@@ -702,6 +764,24 @@ class AxolotlInputConfig(
             )
         return data
 
+    @model_validator(mode="before")
+    @classmethod
+    def check_pretraining_split_batches_accelerate(cls, data):
+        # alternatively set ACCELERATE_SPLIT_BATCHES=False
+        if data.get("pretraining_dataset"):
+            accelerator_config = data.get("accelerator_config", {})
+            if not accelerator_config:
+                data["accelerator_config"] = {
+                    "split_batches": False,
+                    "dispatch_batches": False,
+                }
+            else:
+                if accelerator_config.get("split_batches") is None:
+                    data["accelerator_config"]["split_batches"] = False
+                if accelerator_config.get("dispatch_batches") is None:
+                    data["accelerator_config"]["dispatch_batches"] = False
+        return data
+
     @model_validator(mode="before")
     @classmethod
     def check_gptq_w_revision(cls, data):
@@ -820,7 +900,7 @@ class AxolotlInputConfig(
     @model_validator(mode="after")
     def check_adamw_optimizer_params(self):
         if any([self.adam_beta1, self.adam_beta2, self.adam_epsilon]) and (
-            not self.optimizer or "adamw" not in self.optimizer.value
+            not self.optimizer or "adamw" not in str(self.optimizer).lower()
         ):
             LOG.warning("adamw hyperparameters found, but no adamw optimizer set")
         return self
@@ -891,6 +971,8 @@ class AxolotlInputConfig(
     @model_validator(mode="before")
     @classmethod
     def check_eval_packing(cls, data):
+        # TODO also should check test_datasets and val_set_size as we can skip
+        # if there are no eval datasets/splits
         if (
             data.get("sample_packing")
             and data.get("eval_table_size")
@@ -921,6 +1003,18 @@ class AxolotlInputConfig(
 
         return data
 
+    @model_validator(mode="before")
+    @classmethod
+    def check_mm_prepare(cls, data):
+        if data.get("skip_prepare_dataset"):
+            if data.get("remove_unused_columns") is None:
+                LOG.info(
+                    "setting `remove_unused_columns: false` for skip_prepare_dataset"
+                )
+                data["remove_unused_columns"] = False
+
+        return data
+
     @model_validator(mode="before")
     @classmethod
     def check_warmup(cls, data):
@@ -948,12 +1042,20 @@ class AxolotlInputConfig(
         return neftune_noise_alpha
 
     @model_validator(mode="after")
-    def check(self):
+    def check_rl_beta(self):
         if self.dpo_beta and not self.rl_beta:
             self.rl_beta = self.dpo_beta
             del self.dpo_beta
         return self
 
+    @model_validator(mode="after")
+    def check_simpo_warmup(self):
+        if self.rl == "simpo" and self.warmup_ratio:
+            raise ValueError(
+                "warmup_ratio is not supported with the simpo trainer. Please use `warmup_steps` instead"
+            )
+        return self
+
     @model_validator(mode="before")
     @classmethod
     def check_frozen(cls, data):
@@ -968,6 +1070,15 @@ class AxolotlInputConfig(
 
         return data
 
+    @model_validator(mode="before")
+    @classmethod
+    def check_peft_layers_pattern(cls, data):
+        if data.get("peft_layers_pattern") and not data.get("peft_layers_to_transform"):
+            raise ValueError(
+                "peft_layers_pattern requires peft_layers_to_transform to be set"
+            )
+        return data
+
     @model_validator(mode="after")
     def check_fft_possible_bad_config(self):
         if (
@@ -1087,6 +1198,20 @@ class AxolotlInputConfig(
             )
         return data
 
+    @model_validator(mode="before")
+    @classmethod
+    def check_fsdp_sharded_state_dict_w_safetensors(cls, data):
+        if (
+            data.get("fsdp")
+            and data.get("save_safetensors")
+            and data.get("fsdp_config")
+            and data["fsdp_config"].get("fsdp_state_dict_type") == "SHARDED_STATE_DICT"
+        ):
+            raise ValueError(
+                "FSDP SHARDED_STATE_DICT not compatible with save_safetensors"
+            )
+        return data
+
     @model_validator(mode="before")
     @classmethod
     def check_causal_lm_evals(cls, data):
@@ -1112,6 +1237,55 @@ class AxolotlInputConfig(
             raise ValueError("either datasets or pretraining_dataset is required")
         return data
 
+    @model_validator(mode="before")
+    @classmethod
+    def check_xentropy_patch_conflicts(cls, data):
+        if data.get("flash_attn_cross_entropy") and data.get(
+            "unsloth_cross_entropy_loss"
+        ):
+            raise ValueError(
+                "flash_attn_cross_entropy and unsloth_cross_entropy_loss cannot be both enabled"
+            )
+        return data
+
+    @model_validator(mode="before")
+    @classmethod
+    def check_qlora_unsloth(cls, data):
+        if (
+            data.get("unsloth_lora_mlp")
+            or data.get("unsloth_lora_qkv")
+            or data.get("unsloth_lora_o")
+        ):
+            if data.get("adapter") == "lora" or data.get("load_in_8bit"):
+                raise ValueError(
+                    "unsloth_lora_mlp, unsloth_lora_qkv, and unsloth_lora_o are not compatible with 8-bit LoRA"
+                )
+        return data
+
+    @model_validator(mode="before")
+    @classmethod
+    def check_unsloth_xformers_version(cls, data):
+        if (
+            data.get("unsloth_lora_mlp")
+            or data.get("unsloth_lora_qkv")
+            or data.get("unsloth_lora_o")
+        ):
+            xformers_version = version("xformers")
+            if xformers_version == "0.0.27":
+                raise ValueError(
+                    "xformers version 0.0.27 is not supported with unsloth. Please downgrade to 0.0.26.post1"
+                )
+        return data
+
+    @model_validator(mode="before")
+    @classmethod
+    def check_torch_compile_deepspeed(cls, data):
+        if data.get("deepspeed") and data.get("torch_compile"):
+            raise ValueError(
+                "torch_compile should be set within your deepspeed config file"
+            )
+        return data
+
 
 class AxolotlConfigWCapabilities(AxolotlInputConfig):
     """wrapper to valdiate gpu capabilities with the configured options"""
@@ -1157,9 +1331,37 @@ class AxolotlConfigWCapabilities(AxolotlInputConfig):
 
         return data
 
+    @model_validator(mode="before")
+    @classmethod
+    def check_hopper_8bit_lora(cls, data):
+        is_sm_90: bool = (
+            data["capabilities"]
+            and data["capabilities"].get("compute_capability") == "sm_90"
+        )
+        if data.get("adapter") and data.get("load_in_8bit") and is_sm_90:
+            # see https://github.com/bitsandbytes-foundation/bitsandbytes/issues/538#issuecomment-2262945464
+            raise ValueError("8-bit LoRA is not supported on Hopper GPUs")
+
+        return data
+
     @model_validator(mode="before")
     @classmethod
     def check_fsdp_deepspeed(cls, data):
         if data.get("deepspeed") and data.get("fsdp"):
             raise ValueError("deepspeed and fsdp cannot be used together.")
         return data
+
+    @model_validator(mode="before")
+    @classmethod
+    def check_multigpu_unsloth(cls, data):
+        if (
+            data.get("unsloth_lora_mlp")
+            or data.get("unsloth_lora_qkv")
+            or data.get("unsloth_lora_o")
+        ):
+            capabilities = data.get("capabilities")
+            if capabilities and capabilities.get("n_gpu", 0) > 1:
+                raise ValueError(
+                    "unsloth_lora_mlp, unsloth_lora_qkv, and unsloth_lora_o are not compatible with multi-GPU training."
+                )
+        return data

src/axolotl/utils/data/pretraining.py

@@ -18,10 +18,10 @@ LOG = logging.getLogger("axolotl")
 
 
 def encode_pretraining(
-    tokenizer: PreTrainedTokenizerBase, max_tokens: int, examples: List[str]
+    tokenizer: PreTrainedTokenizerBase, max_tokens: int, examples: Dict[str, List]
 ) -> Dict[str, List]:
     res = tokenizer(
-        examples,
+        examples["text"],
         truncation=True,
         max_length=max_tokens - 2,
         add_special_tokens=True,

src/axolotl/utils/data/rl.py

@@ -1,4 +1,5 @@
 """data handling specific to DPO"""
+
 import inspect
 import logging
 from functools import partial

src/axolotl/utils/data/sft.py

@@ -42,7 +42,7 @@ from axolotl.prompters import (
 from axolotl.utils.data.pretraining import wrap_pretraining_dataset
 from axolotl.utils.data.utils import md5
 from axolotl.utils.dict import DictDefault
-from axolotl.utils.distributed import is_main_process, zero_first
+from axolotl.utils.distributed import is_local_main_process, zero_first
 from axolotl.utils.trainer import (
     calculate_total_num_steps,
     process_datasets_for_packing,
@@ -51,20 +51,31 @@ from axolotl.utils.trainer import (
 LOG = logging.getLogger("axolotl")
 
 
-def prepare_dataset(cfg, tokenizer):
+def prepare_dataset(cfg, tokenizer, processor=None):
     prompters = []
     if not cfg.pretraining_dataset:
-        with zero_first(is_main_process()):
+        with zero_first(is_local_main_process()):
             if cfg.test_datasets:
                 train_dataset, _, prompters = load_prepare_datasets(
-                    tokenizer, cfg, DEFAULT_DATASET_PREPARED_PATH, split="train"
+                    tokenizer,
+                    cfg,
+                    DEFAULT_DATASET_PREPARED_PATH,
+                    split="train",
+                    processor=processor,
                 )
                 _, eval_dataset, _ = load_prepare_datasets(
-                    tokenizer, cfg, DEFAULT_DATASET_PREPARED_PATH, split="test"
+                    tokenizer,
+                    cfg,
+                    DEFAULT_DATASET_PREPARED_PATH,
+                    split="test",
+                    processor=processor,
                 )
             else:
                 train_dataset, eval_dataset, prompters = load_prepare_datasets(
-                    tokenizer, cfg, DEFAULT_DATASET_PREPARED_PATH
+                    tokenizer,
+                    cfg,
+                    DEFAULT_DATASET_PREPARED_PATH,
+                    processor=processor,
                 )
     else:
         path = cfg.pretraining_dataset
@@ -123,6 +134,7 @@ def load_tokenized_prepared_datasets(
     cfg,
     default_dataset_prepared_path,
     split="train",
+    processor=None,
 ) -> Tuple[DatasetDict, List[Prompter]]:
     cfg_datasets = cfg.test_datasets if split == "test" else cfg.datasets
     tokenizer_name = cfg.tokenizer_config
@@ -160,8 +172,12 @@ def load_tokenized_prepared_datasets(
     use_auth_token = cfg.hf_use_auth_token
     try:
         if cfg.push_dataset_to_hub:
+            LOG.info(
+                f"Attempting to load prepared dataset from Huggingface hub at {cfg.push_dataset_to_hub} (version {ds_hash})..."
+            )
             dataset = load_dataset(
-                f"{cfg.push_dataset_to_hub}/{ds_hash}",
+                cfg.push_dataset_to_hub,
+                ds_hash,
                 token=use_auth_token,
             )
             dataset = dataset[split]
@@ -170,17 +186,26 @@ def load_tokenized_prepared_datasets(
 
     # pylint: disable=duplicate-code
     if dataset:
+        # This is for the case where we already loaded a pretokenized dataset from the hub
         ...
     elif (
         cfg.dataset_prepared_path
         and any(prepared_ds_path.glob("*"))
         and not cfg.is_preprocess
+        and not cfg.skip_prepare_dataset
     ):
         LOG.info(f"Loading prepared dataset from disk at {prepared_ds_path}...")
         dataset = load_from_disk(str(prepared_ds_path))
         LOG.info("Prepared dataset loaded from disk...")
     else:
-        LOG.info(f"Unable to find prepared dataset in {prepared_ds_path}")
+        if cfg.push_dataset_to_hub:
+            LOG.info("Unable to find prepared dataset in Huggingface hub")
+        if cfg.is_preprocess:
+            LOG.info(
+                f"Skipping prepared dataset in {prepared_ds_path} for pre-processing..."
+            )
+        else:
+            LOG.info(f"Unable to find prepared dataset in {prepared_ds_path}")
         LOG.info("Loading raw datasets...")
         if not cfg.is_preprocess:
             LOG.warning(
@@ -198,6 +223,8 @@ def load_tokenized_prepared_datasets(
         def for_d_in_datasets(dataset_configs):
             for dataset in dataset_configs:
                 if dataset.name and isinstance(dataset.name, list):
+                    # load_dataset doesn't properly handle multiple named configurations
+                    # at the same time for a given dataset
                     for name in dataset.name:
                         yield DictDefault({**dataset, "name": name})
                 else:
@@ -208,6 +235,8 @@ def load_tokenized_prepared_datasets(
             ds: Optional[Union[Dataset, DatasetDict]] = None
             ds_from_hub = False
             try:
+                # this is just a basic check to see if the path is a
+                # valid HF dataset that's loadable
                 load_dataset(
                     config_dataset.path,
                     name=config_dataset.name,
@@ -407,12 +436,16 @@ def load_tokenized_prepared_datasets(
                 dataset=ds,
                 d_base_type=d_base_type,
                 d_prompt_style=d_prompt_style,
+                processor=processor,
             )
             datasets.append(dataset_wrapper)
             prompters.append(dataset_prompter)
 
-        LOG.info("merging datasets")
-        dataset = concatenate_datasets(datasets)
+        if len(datasets) == 1:
+            dataset = datasets[0]
+        else:
+            LOG.info("merging datasets")
+            dataset = concatenate_datasets(datasets)
 
         if len(datasets) > 1:
             if cfg.shuffle_merged_datasets:
@@ -421,17 +454,20 @@ def load_tokenized_prepared_datasets(
             else:
                 LOG.debug("NOT shuffling merged datasets")
 
-        dataset, _ = process_datasets_for_packing(cfg, dataset, None)
+        if not cfg.skip_prepare_dataset:
+            dataset, _ = process_datasets_for_packing(cfg, dataset, None)
 
-        if cfg.local_rank == 0:
+        if cfg.local_rank == 0 and not cfg.skip_prepare_dataset:
             LOG.info(f"Saving merged prepared dataset to disk... {prepared_ds_path}")
             dataset.save_to_disk(str(prepared_ds_path))
             if cfg.push_dataset_to_hub:
                 LOG.info(
-                    f"Saving merged prepared dataset with push_to_hub... {cfg.push_dataset_to_hub}/{ds_hash}"
+                    f"Pushing merged prepared dataset to Huggingface hub at {cfg.push_dataset_to_hub} (version {ds_hash})..."
                 )
                 dataset.push_to_hub(
-                    f"{cfg.push_dataset_to_hub}/{ds_hash}", private=True
+                    cfg.push_dataset_to_hub,
+                    ds_hash,
+                    private=True,
                 )
 
     return dataset, prompters
@@ -460,9 +496,14 @@ def load_prepare_datasets(
     cfg,
     default_dataset_prepared_path,
     split="train",
+    processor=None,
 ) -> Tuple[Dataset, Dataset, List[Prompter]]:
     dataset, prompters = load_tokenized_prepared_datasets(
-        tokenizer, cfg, default_dataset_prepared_path, split=split
+        tokenizer,
+        cfg,
+        default_dataset_prepared_path,
+        split=split,
+        processor=processor,
     )
 
     if cfg.dataset_shard_num and cfg.dataset_shard_idx is not None:
@@ -528,6 +569,7 @@ def get_dataset_wrapper(
     d_base_type,
     dataset,
     d_prompt_style=None,
+    processor=None,
 ):
     dataset_wrapper = None
     dataset_prompter = None
@@ -560,7 +602,11 @@ def get_dataset_wrapper(
             dataset,
             **ds_kwargs,
         )
-    elif ds_strategy := load(config_dataset.type, tokenizer, cfg, config_dataset):
+    elif cfg.skip_prepare_dataset:
+        dataset_wrapper = dataset
+    elif ds_strategy := load(
+        config_dataset.type, tokenizer, cfg, config_dataset, processor=processor
+    ):
         dataset_prompter = UnsupportedPrompter()
         dataset_wrapper = TokenizedPromptDataset(
             ds_strategy,

src/axolotl/utils/distributed.py

@@ -44,6 +44,10 @@ def is_main_process():
     return dist.get_rank() == 0
 
 
+def is_local_main_process():
+    return PartialState().is_main_process
+
+
 def get_world_size():
     return int(os.getenv("WORLD_SIZE", "1"))
 
@@ -149,11 +153,11 @@ def compute_and_broadcast(fn):  # pylint: disable=invalid-name
     if is_main_process():
         value_scalar = fn()
         value_tensor = torch.tensor(
-            value_scalar, device=torch.cuda.current_device()
-        ).float()
+            value_scalar, device=torch.cuda.current_device(), dtype=torch.float32
+        )
     else:
         value_tensor = torch.tensor(
-            0.0, device=torch.cuda.current_device()
+            0.0, device=torch.cuda.current_device(), dtype=torch.float32
         )  # Placeholder tensor
 
     # Broadcast the tensor to all processes.

src/axolotl/utils/model_shard_quant.py

@@ -13,6 +13,7 @@ from fastcore.parallel import parallel
 from torch import Tensor, nn
 from tqdm import tqdm
 from transformers import AutoModelForCausalLM
+from transformers.quantizers import AutoHfQuantizer
 from transformers.utils import SAFE_WEIGHTS_INDEX_NAME, SAFE_WEIGHTS_NAME, hub
 
 
@@ -173,6 +174,7 @@ def load_sharded_model_quant(
     low_memory=True,
     verbose=False,
     loading_workers=2,
+    quantization_config=None,
 ):
     with init_empty_weights():
         model = AutoModelForCausalLM.from_config(
@@ -186,15 +188,26 @@ def load_sharded_model_quant(
                 compute_dtype=compute_dtype,
                 quant_type="nf4",
                 quant_storage=quant_storage,
+                compress_statistics=True,  # bnb_4bit_use_double_quant
+                skip_modules=[
+                    "lm_head",
+                    "embed_out",
+                ],
             )
         else:
             # this is the more common case with HF transformers
+            # TODO can we detect the model arch and dynamically set skip_modules
             model.model = _replace_linear(
                 model.model,
                 Linear4bit,
                 compute_dtype=compute_dtype,
                 quant_type="nf4",
                 quant_storage=quant_storage,
+                compress_statistics=True,  # bnb_4bit_use_double_quant
+                skip_modules=[
+                    "lm_head",
+                    "embed_out",
+                ],
             )
     model.is_loaded_in_4bit = True
 
@@ -251,6 +264,11 @@ def load_sharded_model_quant(
             quant_method=quant_method,
         )
 
+    # these attributes are needed to inform transformers/peft of the quantization
+    model.is_quantized = True
+    model.quantization_method = "bitsandbytes"
+    model.hf_quantizer = AutoHfQuantizer.from_config(quantization_config)
+
     if cfg.local_rank == 0 and verbose:
         print(f"Loaded model weights in {time.time()-start:.3f} seconds")
     # cleanup any extra memory usage from parallel loading

src/axolotl/utils/models.py

@@ -1,7 +1,7 @@
 """Module for models and model loading"""
 
 # pylint: disable=too-many-lines
-
+import gc
 import logging
 import math
 import os
@@ -28,14 +28,21 @@ from transformers import (  # noqa: F401
     AddedToken,
     AutoConfig,
     AutoModelForCausalLM,
+    AutoModelForVision2Seq,
+    AutoProcessor,
     AutoTokenizer,
+    AwqConfig,
     BitsAndBytesConfig,
     GPTQConfig,
+    LlavaForConditionalGeneration,
+    MllamaForConditionalGeneration,
     PreTrainedModel,
     PreTrainedTokenizerBase,
+    ProcessorMixin,
 )
 from transformers.integrations.deepspeed import is_deepspeed_zero3_enabled
 
+from axolotl.common.architectures import MOE_ARCH_BLOCK
 from axolotl.models.mamba import fix_mamba_attn_for_loss
 from axolotl.monkeypatch.multipack import (
     SUPPORTED_MULTIPACK_MODEL_TYPES,
@@ -78,6 +85,9 @@ def get_module_class_from_name(module, name):
 
 
 def check_model_config(cfg: DictDefault, model_config: Union[AutoConfig, DictDefault]):
+    if cfg.is_multimodal:
+        model_config = model_config.text_config
+
     quant_config_exists = (
         hasattr(model_config, "quantization_config")
         and model_config.quantization_config
@@ -94,7 +104,7 @@ def check_model_config(cfg: DictDefault, model_config: Union[AutoConfig, DictDef
             "Please make sure to point to a GPTQ model."
         )
 
-    if not cfg.gptq and quant_config_exists:
+    if not cfg.gptq and quant_config_exists and not cfg.load_in_4bit:
         raise ValueError(
             "model_config.quantization_config is set but `gptq` flag is not. "
             "Please use the `gptq` flag to train quantized model or point to a non-quantized model."
@@ -297,25 +307,63 @@ def load_tokenizer(cfg):
     return tokenizer
 
 
+def load_processor(cfg: DictDefault, tokenizer: PreTrainedTokenizerBase):
+    processor_kwargs: Dict[str, Any] = {}  # do we actually need this?
+
+    processor_cls = AutoProcessor
+    if cfg.processor_type:
+        processor_cls = getattr(transformers, cfg.processor_type)
+
+    processor = processor_cls.from_pretrained(
+        cfg.processor_config,
+        trust_remote_code=cfg.trust_remote_code or False,
+        tokenizer=tokenizer,
+        **processor_kwargs,
+    )
+
+    return processor
+
+
 def load_model(
     cfg: DictDefault,
     tokenizer: PreTrainedTokenizerBase,
+    *,
+    processor: ProcessorMixin = None,  # pylint: disable=unused-argument
     inference: bool = False,
     reference_model: bool = False,
+    **kwargs,  # pylint: disable=unused-argument
 ) -> Tuple[PreTrainedModel, Optional[PeftConfig]]:
     """
     Load a model for a given configuration and tokenizer.
     """
+
     base_model = cfg.base_model
     model_type = cfg.type_of_model
     model_config = load_model_config(cfg)
 
+    # load any patches from plugins
+    from axolotl.integrations.base import PluginManager
+
+    plugin_manager = PluginManager.get_instance()
+    plugin_manager.pre_model_load(cfg)
+
+    if cfg.is_multimodal:
+        text_model_config = model_config.text_config
+    else:
+        text_model_config = model_config
+
     # TODO refactor as a kwarg
     load_in_8bit = cfg.load_in_8bit
 
     if cfg.gradient_checkpointing == "unsloth":
         transformers.modeling_utils.checkpoint = hf_grad_checkpoint_unsloth_wrapper
 
+    if hasattr(model_config, "model_type") and model_config.model_type == "mllama":
+        if cfg.flash_attention:
+            from axolotl.monkeypatch.attention.mllama import patch_mllama
+
+            patch_mllama()
+
     if hasattr(model_config, "model_type") and model_config.model_type == "btlm":
         if cfg.flash_attention:
             from axolotl.monkeypatch.btlm_attn_hijack_flash import (
@@ -346,7 +394,36 @@ def load_model(
         and cfg.flash_attention
         and cfg.sample_packing
     ):
-        patch_for_multipack(cfg.model_config_type, model_name=cfg.base_model)
+        patch_for_multipack(
+            cfg.model_config_type,
+            model_name=cfg.base_model,
+            is_remote_code=cfg.trust_remote_code,
+        )
+
+        if cfg.is_llama_derived_model:
+            from axolotl.monkeypatch.llama_attn_hijack_flash import (
+                patch_llama_cross_entropy,
+                patch_llama_rms_norm,
+            )
+
+            if cfg.flash_attn_cross_entropy:
+                patch_llama_cross_entropy()
+            if cfg.flash_attn_rms_norm:
+                patch_llama_rms_norm()
+            elif cfg.unsloth_rms_norm:
+                from axolotl.monkeypatch.unsloth_ import patch_unsloth_layernorm
+
+                patch_unsloth_layernorm()
+            if cfg.unsloth_cross_entropy_loss:
+                from axolotl.monkeypatch.unsloth_ import (
+                    integrate_cross_entropy_loss_patch,
+                )
+
+                integrate_cross_entropy_loss_patch(model_type="llama")
+            if cfg.unsloth_lora_qkv or cfg.unsloth_lora_o:
+                from axolotl.monkeypatch.unsloth_ import patch_self_attn_lora
+
+                patch_self_attn_lora()
     elif cfg.is_llama_derived_model:
         # Modify all llama derived models in one block
 
@@ -371,6 +448,12 @@ def load_model(
                     rms_norm=cfg.flash_attn_rms_norm,
                     use_shifted_sparse_attn=True,
                 )
+            elif cfg.flash_attn_cross_entropy or cfg.flash_attn_rms_norm:
+                replace_llama_attn_with_flash_attn(
+                    packed=False,
+                    cross_entropy=cfg.flash_attn_cross_entropy,
+                    rms_norm=cfg.flash_attn_rms_norm,
+                )
         elif cfg.xformers_attention:
             from axolotl.monkeypatch.llama_attn_hijack_xformers import (
                 hijack_llama_attention,
@@ -393,7 +476,7 @@ def load_model(
         if cfg.unsloth_cross_entropy_loss:
             from axolotl.monkeypatch.unsloth_ import integrate_cross_entropy_loss_patch
 
-            integrate_cross_entropy_loss_patch()
+            integrate_cross_entropy_loss_patch(model_type="llama")
 
         if cfg.unsloth_lora_qkv or cfg.unsloth_lora_o:
             from axolotl.monkeypatch.unsloth_ import patch_self_attn_lora
@@ -401,23 +484,12 @@ def load_model(
             patch_self_attn_lora()
 
     # Modify mistral derived models
-    if (
-        cfg.model_config_type == "mistral"
-        and cfg.flash_attention
-        and cfg.sample_packing
-    ):
+    if cfg.model_config_type == "mistral" and cfg.flash_attn_cross_entropy_loss:
         from axolotl.monkeypatch.mistral_attn_hijack_flash import (
-            replace_mistral_attn_with_flash_attn,
+            patch_mistral_cross_entropy,
         )
 
-        LOG.info("patching mistral with flash attention")
-        replace_mistral_attn_with_flash_attn(packed=cfg.sample_packing)
-
-    if cfg.is_llama_derived_model and cfg.sample_packing and not inference:
-        from axolotl.monkeypatch.llama_expand_mask import hijack_expand_mask
-
-        LOG.info("patching _expand_mask")
-        hijack_expand_mask()
+        patch_mistral_cross_entropy()
 
     model_kwargs: Dict[str, Any] = {}
 
@@ -428,6 +500,19 @@ def load_model(
     max_memory = cfg.max_memory
     device_map = cfg.device_map
 
+    AutoModelLoader = AutoModelForCausalLM  # pylint: disable=invalid-name
+    if cfg.is_multimodal:
+        if model_config.model_type == "llava":
+            AutoModelLoader = (  # pylint: disable=invalid-name
+                LlavaForConditionalGeneration
+            )
+        elif model_config.model_type == "mllama":
+            AutoModelLoader = (  # pylint: disable=invalid-name
+                MllamaForConditionalGeneration
+            )
+        else:
+            AutoModelLoader = AutoModelForVision2Seq  # pylint: disable=invalid-name
+
     if cfg.gpu_memory_limit:
         gpu_memory_limit = (
             str(cfg.gpu_memory_limit) + "GiB"
@@ -445,7 +530,7 @@ def load_model(
         from accelerate import infer_auto_device_map
 
         with init_empty_weights():
-            model_canvas = AutoModelForCausalLM.from_config(
+            model_canvas = AutoModelLoader.from_config(
                 model_config, trust_remote_code=cfg.trust_remote_code or False
             )
         model_canvas.tie_weights()
@@ -490,7 +575,25 @@ def load_model(
             model_kwargs["quantization_config"] = GPTQConfig(
                 **model_config.quantization_config
             )
-    if cfg.adapter == "qlora" and cfg.load_in_4bit:
+    if (
+        cfg.adapter in ["qlora", "lora"]
+        and hasattr(model_config, "quantization_config")
+        and model_config.quantization_config["quant_method"]
+        in ["gptq", "awq", "bitsandbytes"]
+    ):
+        if model_config.quantization_config["quant_method"] == "gptq":
+            model_kwargs["quantization_config"] = GPTQConfig(
+                **model_config.quantization_config
+            )
+        elif model_config.quantization_config["quant_method"] == "awq":
+            model_kwargs["quantization_config"] = AwqConfig(
+                **model_config.quantization_config
+            )
+        elif model_config.quantization_config["quant_method"] == "bitsandbytes":
+            model_kwargs["quantization_config"] = BitsAndBytesConfig(
+                **model_config.quantization_config
+            )
+    elif cfg.adapter == "qlora" and cfg.load_in_4bit:
         bnb_config = {
             "load_in_4bit": True,
             "llm_int8_threshold": 6.0,
@@ -500,7 +603,9 @@ def load_model(
             "bnb_4bit_quant_type": "nf4",
             "bnb_4bit_quant_storage": torch.bfloat16,
         }
-        if cfg.model_config_type in ["jamba", "qwen2_moe"] and not cfg.deepspeed:
+        if cfg.model_config_type in ["jamba", "qwen2_moe"] and not (
+            cfg.deepspeed or cfg.fsdp
+        ):
             # for some reason, this causes the loss to be off by an order of magnitude
             # but deepspeed needs this still in bfloat16
             bnb_config["bnb_4bit_quant_storage"] = torch.float32
@@ -536,25 +641,12 @@ def load_model(
 
     # sample packing uses custom FA2 patch
     if cfg.flash_attention:
-        if not cfg.sample_packing:
-            if cfg.s2_attention:
-                pass
-            # most other models support flash attention, we can define exceptions as they come up
-            model_kwargs["attn_implementation"] = "flash_attention_2"
-            model_config._attn_implementation = (  # pylint: disable=protected-access
-                "flash_attention_2"
-            )
-        else:
-            if model_config.model_type in SUPPORTED_MULTIPACK_MODEL_TYPES:
-                model_kwargs["attn_implementation"] = "flash_attention_2"
-                model_config._attn_implementation = (  # pylint: disable=protected-access
-                    "flash_attention_2"
-                )
-            else:
-                model_kwargs["attn_implementation"] = "eager"
-                model_config._attn_implementation = (  # pylint: disable=protected-access
-                    "eager"
-                )
+        if not cfg.sample_packing and cfg.s2_attention:
+            pass
+        model_kwargs["attn_implementation"] = "flash_attention_2"
+        model_config._attn_implementation = (  # pylint: disable=protected-access
+            "flash_attention_2"
+        )
     elif cfg.sdp_attention:
         model_kwargs["attn_implementation"] = "sdpa"
         model_config._attn_implementation = "sdpa"  # pylint: disable=protected-access
@@ -584,14 +676,23 @@ def load_model(
         elif (
             qlora_fsdp
             and cfg.fsdp_config.fsdp_cpu_ram_efficient_loading
-            and cfg.model_config_type == "dbrx"
+            and (cfg.model_config_type == "dbrx" or cfg.qlora_sharded_model_loading)
         ):
             quant_storage = cfg.torch_dtype
+            quantization_config = hasattr(
+                model_config, "quantization_config"
+            ) and getattr(model_config, "quantization_config")
+            quantization_config = (
+                quantization_config or model_kwargs["quantization_config"]
+            )
+            if cfg.is_multimodal:
+                model_config.text_config = text_model_config
             model = load_sharded_model_quant(
                 base_model,
                 model_config,
                 cfg,
                 quant_storage=quant_storage,
+                quantization_config=quantization_config,
             )
             skip_move_to_device = True
         elif (
@@ -599,12 +700,14 @@ def load_model(
             and not cfg.trust_remote_code
             and not cfg.gptq
         ):
-            if qlora_fsdp and cfg.fsdp_config.fsdp_cpu_ram_efficient_loading:
+            if cfg.fsdp and cfg.fsdp_config.fsdp_cpu_ram_efficient_loading:
                 skip_move_to_device = True
                 if "device_map" in model_kwargs:
                     del model_kwargs["device_map"]
 
-            model = AutoModelForCausalLM.from_pretrained(
+            if cfg.is_multimodal:
+                model_config.text_config = text_model_config
+            model = AutoModelLoader.from_pretrained(
                 base_model,
                 config=model_config,
                 **model_kwargs,
@@ -643,13 +746,17 @@ def load_model(
             and not cfg.trust_remote_code
         ):
             if cfg.gptq:
-                model = AutoModelForCausalLM.from_pretrained(
+                if cfg.is_multimodal:
+                    model_config.text_config = text_model_config
+                model = AutoModelLoader.from_pretrained(
                     base_model,
                     config=model_config,
                     trust_remote_code=cfg.trust_remote_code or False,
                     **model_kwargs,
                 )
             else:
+                if cfg.is_multimodal:
+                    model_config.text_config = text_model_config
                 model = getattr(transformers, model_type).from_pretrained(
                     base_model,
                     config=model_config,
@@ -660,34 +767,38 @@ def load_model(
             # Shouldn't be a problem most of the time. will obviously error if the model doesn't support this
             # when training starts
             if (
-                hasattr(model_config, "max_seq_len")
-                and model_config.max_seq_len
+                hasattr(text_model_config, "max_seq_len")
+                and text_model_config.max_seq_len
                 and cfg.sequence_len > model_config.max_seq_len
             ):
-                model_config.max_seq_len = cfg.sequence_len
+                text_model_config.max_seq_len = cfg.sequence_len
                 LOG.warning(f"increasing context length to {cfg.sequence_len}")
             elif (
-                hasattr(model_config, "max_sequence_length")
-                and model_config.max_sequence_length
-                and cfg.sequence_len > model_config.max_sequence_length
+                hasattr(text_model_config, "max_sequence_length")
+                and text_model_config.max_sequence_length
+                and cfg.sequence_len > text_model_config.max_sequence_length
             ):
-                model_config.max_sequence_length = cfg.sequence_len
+                text_model_config.max_sequence_length = cfg.sequence_len
                 LOG.warning(f"increasing context length to {cfg.sequence_len}")
             if cfg.gptq:
-                model = AutoModelForCausalLM.from_pretrained(
+                if cfg.is_multimodal:
+                    model_config.text_config = text_model_config
+                model = AutoModelLoader.from_pretrained(
                     base_model,
                     config=model_config,
                     trust_remote_code=cfg.trust_remote_code or False,
                     **model_kwargs,
                 )
             else:
-                if qlora_fsdp and cfg.fsdp_config.fsdp_cpu_ram_efficient_loading:
+                if cfg.fsdp and cfg.fsdp_config.fsdp_cpu_ram_efficient_loading:
                     # disabling either of these two still leads to VRAM spike before setting back down
                     skip_move_to_device = True
                     if "device_map" in model_kwargs:
                         del model_kwargs["device_map"]
 
-                model = AutoModelForCausalLM.from_pretrained(
+                if cfg.is_multimodal:
+                    model_config.text_config = text_model_config
+                model = AutoModelLoader.from_pretrained(
                     base_model,
                     config=model_config,
                     trust_remote_code=cfg.trust_remote_code or False,
@@ -765,12 +876,16 @@ def load_model(
             set_z3_leaf_modules,
         )
 
-        if cfg.model_config_type == "mixtral":
-            moe_block = get_module_class_from_name(model, "MixtralSparseMoeBlock")
-            set_z3_leaf_modules(model, [moe_block])
-        elif cfg.model_config_type == "dbrx":
-            moe_block = get_module_class_from_name(model, "DbrxFFN")
-            set_z3_leaf_modules(model, [moe_block])
+        if cfg.model_config_type in MOE_ARCH_BLOCK:
+            moe_blocks = MOE_ARCH_BLOCK[cfg.model_config_type]
+            moe_blocks = [moe_blocks] if isinstance(moe_blocks, str) else moe_blocks
+            set_z3_leaf_modules(
+                model,
+                [
+                    get_module_class_from_name(model, module_name)
+                    for module_name in moe_blocks
+                ],
+            )
 
     if cfg.model_config_type == "qwen" and cfg.adapter == "lora":
         # Qwen doesn't play nicely with LoRA if this is enabled
@@ -784,6 +899,9 @@ def load_model(
         # make sure everything is in the same dtype
         skip_prepare_model_for_kbit_training = True
 
+    if is_deepspeed_zero3_enabled():
+        skip_prepare_model_for_kbit_training = True
+
     if cfg.adapter in ["lora", "qlora"]:
         if cfg.gradient_checkpointing:
             model.gradient_checkpointing_enable(
@@ -818,6 +936,9 @@ def load_model(
         else:
             model, lora_config = load_adapter(model, cfg, cfg.adapter)
 
+    if is_deepspeed_zero3_enabled():
+        skip_move_to_device = True
+
     if (
         cfg.ddp
         and not load_in_8bit
@@ -857,6 +978,15 @@ def load_model(
 
         integrate_lora_patch(model, cfg)
 
+    if cfg.unsloth_rope:
+        from axolotl.monkeypatch.unsloth_ import integrate_rope_embeddings
+
+        integrate_rope_embeddings()
+
+    for _ in range(3):
+        gc.collect()
+        torch.cuda.empty_cache()
+
     # TODO resume_from_checkpoint handling
     return model, lora_config
 
@@ -950,12 +1080,17 @@ def load_lora(model, cfg, inference=False, config_only=False):
 
     from peft import LoraConfig, get_peft_model
 
-    lora_target_modules = list(cfg.lora_target_modules or [])
+    lora_target_modules = cfg.lora_target_modules or []
 
     if cfg.lora_target_linear:
         linear_names = find_all_linear_names(model)
-        LOG.info(f"found linear modules: {repr(linear_names)}")
-        lora_target_modules = list(set(lora_target_modules + linear_names))
+        LOG.info(f"found linear modules: {repr(sorted(linear_names))}")
+        lora_target_modules_as_list = (
+            lora_target_modules
+            if isinstance(lora_target_modules, list)
+            else [lora_target_modules]
+        )
+        lora_target_modules = list(set(lora_target_modules_as_list + linear_names))
 
     lora_config_kwargs = {}
     loftq_bits = cfg.peft and cfg.peft.loftq_config and cfg.peft.loftq_config.loftq_bits
@@ -974,6 +1109,7 @@ def load_lora(model, cfg, inference=False, config_only=False):
         lora_alpha=cfg.lora_alpha,
         target_modules=lora_target_modules,
         layers_to_transform=cfg.peft_layers_to_transform,
+        layers_pattern=cfg.peft_layers_pattern,
         lora_dropout=cfg.lora_dropout,
         fan_in_fan_out=cfg.lora_fan_in_fan_out,
         modules_to_save=cfg.lora_modules_to_save if cfg.lora_modules_to_save else None,
@@ -1030,9 +1166,20 @@ def load_lora(model, cfg, inference=False, config_only=False):
 
 def ensure_dtype(model, dtype=torch.bfloat16):
     for name, module in model.named_modules():
+        weight_mismatch = False
+        bias_mismatch = False
         try:
-            if module.weight.dtype != dtype:
-                print(f"Converting module {name}: {module.weight.dtype} -> {dtype}")
-                module.to(dtype)
+            weight_mismatch = module.weight.dtype != dtype
         except AttributeError:
             pass
+        try:
+            bias_mismatch = module.bias.dtype != dtype
+        except AttributeError:
+            pass
+
+        if weight_mismatch:
+            print(f"Converting module {name}.weight: {module.weight.dtype} -> {dtype}")
+        if bias_mismatch:
+            print(f"Converting module {name}.bias: {module.bias.dtype} -> {dtype}")
+        if weight_mismatch or bias_mismatch:
+            module.to(dtype)

src/axolotl/utils/samplers/multipack.py

@@ -11,6 +11,8 @@ import numba
 import numpy as np
 from torch.utils.data import BatchSampler, Sampler
 
+from axolotl.utils.distributed import reduce_and_broadcast
+
 LOG = logging.getLogger("axolotl.utils.samplers.multipack")
 
 
@@ -174,16 +176,46 @@ class MultipackBatchSampler(BatchSampler):
     def efficiency(self):
         return self.eff_total_used / self.eff_total_slots
 
+    def gather_efficiency(self):
+        def calc_sample_packing_eff_est(estimates: List[float]):
+            LOG.debug(f"sample_packing_eff_est across ranks: {repr(estimates)}")
+            return math.floor(0.997 * max(estimates))
+
+        sample_packing_actual_eff_all = reduce_and_broadcast(
+            lambda: self.efficiency(),  # pylint: disable=unnecessary-lambda
+            calc_sample_packing_eff_est,
+        )
+        sample_packing_eff_est = (
+            math.ceil(sample_packing_actual_eff_all * 200.0) / 200.0
+        )
+        return sample_packing_eff_est
+
+    def gather_len_batches(self, num):
+        def calc_min_len(estimates: list[(int, float)]):
+            LOG.info(f"gather_len_batches: {repr(estimates)}")
+            return math.floor(0.998 * min(estimates))
+
+        min_len_batches = reduce_and_broadcast(
+            lambda: num,
+            calc_min_len,
+        )
+        return min_len_batches
+
     def __len__(self):
-        self.num_batches()
-        return self._len_est()
+        len_batches = self.num_batches()
+        return self.gather_len_batches(len_batches)
 
     def _len_est(self):
+        efficiency = (
+            self.packing_efficiency_estimate
+            if self.packing_efficiency_estimate
+            else self.gather_efficiency()
+        )
         world_size = int(os.getenv("WORLD_SIZE", "1"))
         lengths_sum = np.sum(self.lengths)
         lengths_sum_per_device = lengths_sum // world_size
         LOG.info(
-            f"packing_efficiency_estimate: {self.packing_efficiency_estimate} "
+            f"packing_efficiency_estimate: {efficiency} "
             f"total_num_tokens per device: {lengths_sum_per_device}"
         )
 
@@ -195,7 +227,7 @@ class MultipackBatchSampler(BatchSampler):
                 * math.floor(
                     0.99
                     * lengths_sum_per_device
-                    / self.packing_efficiency_estimate
+                    / efficiency
                     // (self.batch_max_len * self.batch_size)
                 )
                 - 1

src/axolotl/utils/tokenization.py

@@ -62,7 +62,7 @@ def process_tokens_for_rl_debug(tokens, color, tokenizer, text_only):
     """Helper function to process and color tokens."""
     colored_tokens = [
         color_token_for_rl_debug(tokenizer.decode(token), token, color, text_only)
-        for token in tokenizer.encode(tokens)
+        for token in tokenizer.encode(tokens, add_special_tokens=False)
     ]
     return colored_tokens
 

src/axolotl/utils/trainer.py

@@ -1,4 +1,5 @@
 """Module containing the Trainer class and related functions"""
+import json
 import math
 import os
 import random
@@ -15,7 +16,7 @@ from torch.utils.data import DataLoader, RandomSampler
 from transformers.utils import is_torch_bf16_gpu_available
 
 from axolotl.core.trainer_builder import HFCausalTrainerBuilder, HFRLTrainerBuilder
-from axolotl.utils.distributed import is_main_process, reduce_and_broadcast, zero_first
+from axolotl.utils.distributed import reduce_and_broadcast
 from axolotl.utils.samplers import MultipackBatchSampler, get_dataset_lengths
 
 LOG = get_logger("axolotl")
@@ -182,90 +183,106 @@ def process_datasets_for_packing(cfg, train_dataset, eval_dataset):
         sequence_len=cfg.sequence_len,
         min_sequence_len=cfg.min_sample_len or 2,
     )
-    with zero_first(is_main_process()):
-        if cfg.is_preprocess:
-            min_input_len = np.min(get_dataset_lengths(train_dataset))
-            LOG.debug(f"min_input_len: {min_input_len}", main_process_only=True)
-            max_input_len = np.max(get_dataset_lengths(train_dataset))
-            LOG.debug(f"max_input_len: {max_input_len}", main_process_only=True)
 
-        if (
-            cfg.is_mistral_derived_model and cfg.flash_attention
-        ) or cfg.model_config_type == "mamba":
-            LOG.info("dropping attention_mask column")
-            train_dataset = train_dataset.remove_columns("attention_mask")
-            if eval_dataset:
-                eval_dataset = eval_dataset.remove_columns("attention_mask")
+    if cfg.is_preprocess:
+        min_input_len = np.min(get_dataset_lengths(train_dataset))
+        LOG.debug(f"min_input_len: {min_input_len}", main_process_only=True)
+        max_input_len = np.max(get_dataset_lengths(train_dataset))
+        LOG.debug(f"max_input_len: {max_input_len}", main_process_only=True)
 
-        if cfg.model_config_type == "falcon":
-            LOG.info("dropping token_type_ids column if it exists")
-            if "token_type_ids" in train_dataset.column_names:
-                train_dataset = train_dataset.remove_columns("token_type_ids")
-            if eval_dataset and "token_type_ids" in eval_dataset.column_names:
-                eval_dataset = eval_dataset.remove_columns("token_type_ids")
+    if cfg.model_config_type == "mamba":
+        LOG.info("dropping attention_mask column")
+        train_dataset = train_dataset.remove_columns("attention_mask")
+        if eval_dataset:
+            eval_dataset = eval_dataset.remove_columns("attention_mask")
 
-        train_dataset = train_dataset.filter(
+    if cfg.model_config_type == "falcon":
+        LOG.info("dropping token_type_ids column if it exists")
+        if "token_type_ids" in train_dataset.column_names:
+            train_dataset = train_dataset.remove_columns("token_type_ids")
+        if eval_dataset and "token_type_ids" in eval_dataset.column_names:
+            eval_dataset = eval_dataset.remove_columns("token_type_ids")
+
+    train_dataset = train_dataset.filter(
+        drop_long,
+        num_proc=cfg.dataset_processes,
+        load_from_cache_file=not cfg.is_preprocess,
+        desc="Dropping Long Sequences",
+    )
+    if eval_dataset:
+        eval_dataset = eval_dataset.filter(
             drop_long,
             num_proc=cfg.dataset_processes,
             load_from_cache_file=not cfg.is_preprocess,
             desc="Dropping Long Sequences",
         )
-        if eval_dataset:
-            eval_dataset = eval_dataset.filter(
-                drop_long,
-                num_proc=cfg.dataset_processes,
-                load_from_cache_file=not cfg.is_preprocess,
-                desc="Dropping Long Sequences",
-            )
 
-        if cfg.group_by_length:
-            train_dataset = train_dataset.map(
-                add_length,
-                num_proc=cfg.dataset_processes,
-                load_from_cache_file=not cfg.is_preprocess,
-                desc="Group By Length",
-            )
+    # drop samples with where the number of elements with labels not equal to -100 is zero
+    def drop_no_trainable_tokens(sample):
+        return np.sum(np.array(sample["labels"]) != -100) > 0
 
-        if cfg.use_pose:
-            pose_kwargs = {}
-            if cfg.pose_num_chunks is not None:
-                pose_kwargs["chunks"] = cfg.pose_num_chunks
-            pose_fn = partial(
-                add_pose_position_ids,
-                max_context_len=cfg.pose_max_context_len,
-                split_on_token_ids=cfg.pose_split_on_token_ids,
-                **pose_kwargs,
-            )
-            train_dataset = train_dataset.map(
-                pose_fn,
-                num_proc=cfg.dataset_processes,
-                load_from_cache_file=not cfg.is_preprocess,
-                desc="Add position_id column (PoSE)",
-            )
-            train_dataset = train_dataset.sort("sequence_len")
-            if cfg.eval_sample_packing is not False:
-                if eval_dataset:
-                    eval_dataset = eval_dataset.map(
-                        pose_fn,
-                        num_proc=cfg.dataset_processes,
-                        load_from_cache_file=not cfg.is_preprocess,
-                        desc="Add position_id column (PoSE)",
-                    )
-        elif cfg.sample_packing:
-            train_dataset = train_dataset.map(
-                add_position_ids,
-                num_proc=cfg.dataset_processes,
-                load_from_cache_file=not cfg.is_preprocess,
-                desc="Add position_id column (Sample Packing)",
-            )
-            if cfg.eval_sample_packing is not False:
-                if eval_dataset:
-                    eval_dataset = eval_dataset.map(
-                        add_position_ids,
-                        num_proc=cfg.dataset_processes,
-                        load_from_cache_file=not cfg.is_preprocess,
-                        desc="Add position_id column (Sample Packing)",
-                    )
+    train_dataset = train_dataset.filter(
+        drop_no_trainable_tokens,
+        num_proc=cfg.dataset_processes,
+        load_from_cache_file=not cfg.is_preprocess,
+        desc="Drop Samples with Zero Trainable Tokens",
+    )
+    if eval_dataset:
+        eval_dataset = eval_dataset.filter(
+            drop_no_trainable_tokens,
+            num_proc=cfg.dataset_processes,
+            load_from_cache_file=not cfg.is_preprocess,
+            desc="Drop Samples with Zero Trainable Tokens",
+        )
+
+    if cfg.group_by_length:
+        train_dataset = train_dataset.map(
+            add_length,
+            num_proc=cfg.dataset_processes,
+            load_from_cache_file=not cfg.is_preprocess,
+            desc="Group By Length",
+        )
+
+    if cfg.use_pose:
+        pose_kwargs = {}
+        if cfg.pose_num_chunks is not None:
+            pose_kwargs["chunks"] = cfg.pose_num_chunks
+        pose_fn = partial(
+            add_pose_position_ids,
+            max_context_len=cfg.pose_max_context_len,
+            split_on_token_ids=cfg.pose_split_on_token_ids,
+            **pose_kwargs,
+        )
+        train_dataset = train_dataset.map(
+            pose_fn,
+            num_proc=cfg.dataset_processes,
+            load_from_cache_file=not cfg.is_preprocess,
+            desc="Add position_id column (PoSE)",
+        )
+        train_dataset = train_dataset.sort("sequence_len")
+        if cfg.eval_sample_packing is not False:
+            if eval_dataset:
+                eval_dataset = eval_dataset.map(
+                    pose_fn,
+                    num_proc=cfg.dataset_processes,
+                    load_from_cache_file=not cfg.is_preprocess,
+                    desc="Add position_id column (PoSE)",
+                )
+    elif cfg.sample_packing:
+        train_dataset = train_dataset.map(
+            add_position_ids,
+            num_proc=cfg.dataset_processes,
+            load_from_cache_file=not cfg.is_preprocess,
+            desc="Add position_id column (Sample Packing)",
+        )
+        if cfg.eval_sample_packing is not False:
+            if eval_dataset:
+                eval_dataset = eval_dataset.map(
+                    add_position_ids,
+                    num_proc=cfg.dataset_processes,
+                    load_from_cache_file=not cfg.is_preprocess,
+                    desc="Add position_id column (Sample Packing)",
+                )
 
     return train_dataset, eval_dataset
 
@@ -289,7 +306,7 @@ def process_pretraining_datasets_for_packing(
 
 
 def calculate_total_num_steps(cfg, train_dataset, update=True):
-    if not cfg.total_num_tokens:
+    if not cfg.total_num_tokens and not cfg.skip_prepare_dataset:
         total_num_tokens = np.sum(
             train_dataset.data.column("input_ids")
             .to_pandas()
@@ -302,7 +319,11 @@ def calculate_total_num_steps(cfg, train_dataset, update=True):
 
     skip_estimates = cfg.model_config_type == "mamba"
 
-    if not skip_estimates and not cfg.total_supervised_tokens:
+    if (
+        not skip_estimates
+        and not cfg.total_supervised_tokens
+        and not cfg.skip_prepare_dataset
+    ):
         total_supervised_tokens = (
             train_dataset.data.column("labels")
             .to_pandas()
@@ -340,7 +361,7 @@ def calculate_total_num_steps(cfg, train_dataset, update=True):
                 main_process_only=True,
             )
         else:
-            if cfg.flash_attention:
+            if cfg.flash_attention and not cfg.multipack_real_batches:
                 sampler_batch_size = 1
                 batch_max_len = cfg.micro_batch_size * cfg.sequence_len
             else:
@@ -391,6 +412,27 @@ def calculate_total_num_steps(cfg, train_dataset, update=True):
     return total_num_steps
 
 
+def setup_torch_compile_env(cfg):
+    if cfg.torch_compile:
+        if not cfg.torch_compile_backend:
+            os.environ["ACCELERATE_DYNAMO_BACKEND"] = "INDUCTOR"
+        else:
+            os.environ["ACCELERATE_DYNAMO_BACKEND"] = cfg.torch_compile_backend.upper()
+
+
+def setup_deepspeed_env(cfg, stage=None):
+    from transformers.integrations.deepspeed import HfTrainerDeepSpeedConfig
+
+    os.environ["ACCELERATE_USE_DEEPSPEED"] = "true"
+    os.environ["ACCELERATE_DEEPSPEED_CONFIG_FILE"] = cfg.deepspeed
+    if stage:
+        os.environ["ACCELERATE_DEEPSPEED_ZERO_STAGE"] = str(stage)
+        if stage == 3:
+            os.environ["ACCELERATE_DEEPSPEED_ZERO3_INIT"] = "true"
+    # If we don't assign this, it doesn't actually get set in the accelerate weakref
+    _ = HfTrainerDeepSpeedConfig(cfg.deepspeed)
+
+
 def setup_fsdp_envs(cfg):
     os.environ["ACCELERATE_USE_FSDP"] = "true"
     if cfg.fsdp_config.fsdp_activation_checkpointing:
@@ -417,8 +459,16 @@ def prepare_optim_env(cfg):
     if cfg.fsdp:
         setup_fsdp_envs(cfg)
     elif cfg.deepspeed:
-        os.environ["ACCELERATE_USE_DEEPSPEED"] = "true"
-        os.environ["ACCELERATE_DEEPSPEED_CONFIG_FILE"] = cfg.deepspeed
+        stage = None
+        # check if the cfg.deepspeed is a file
+        if os.path.isfile(cfg.deepspeed):
+            # parse with json
+            with open(cfg.deepspeed, "r", encoding="utf-8") as fin:
+                deepspeed_config = json.load(fin)
+            stage = deepspeed_config.get("zero_optimization", {}).get("stage", None)
+        setup_deepspeed_env(cfg, stage=stage)
+
+    setup_torch_compile_env(cfg)
 
     if (cfg.bf16 == "auto" and is_torch_bf16_gpu_available()) or cfg.bf16 is True:
         os.environ["ACCELERATE_MIXED_PRECISION"] = "bf16"
@@ -426,13 +476,21 @@ def prepare_optim_env(cfg):
         os.environ["ACCELERATE_MIXED_PRECISION"] = "fp16"
 
 
-def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer, total_num_steps):
-    if cfg.rl in ["dpo", "ipo", "orpo", "kto"]:
-        trainer_builder = HFRLTrainerBuilder(cfg, model[0], tokenizer)
+def prepare_opinionated_env(cfg):
+    if cfg.qlora_sharded_model_loading:
+        # model loading is forked after the tokenizer
+        os.environ["TOKENIZERS_PARALLELISM"] = "false"
+
+
+def setup_trainer(
+    cfg, train_dataset, eval_dataset, model, tokenizer, processor, total_num_steps
+):
+    if cfg.rl in ["dpo", "ipo", "orpo", "kto", "simpo"]:
+        trainer_builder = HFRLTrainerBuilder(cfg, model[0], tokenizer, processor)
         trainer_builder.model_ref = model[1]
         trainer_builder.peft_config = model[2]
     else:
-        trainer_builder = HFCausalTrainerBuilder(cfg, model[0], tokenizer)
+        trainer_builder = HFCausalTrainerBuilder(cfg, model[0], tokenizer, processor)
 
     trainer_builder.train_dataset = train_dataset
     trainer_builder.eval_dataset = eval_dataset