multipack sampler support from openchat

.github/CODE_OF_CONDUCT.md

@@ -1,129 +0,0 @@
-# Contributor Covenant Code of Conduct
-
-## Our Pledge
-
-We as members, contributors, and leaders pledge to make participation in our
-community a harassment-free experience for everyone, regardless of age, body
-size, visible or invisible disability, ethnicity, sex characteristics, gender
-identity and expression, level of experience, education, socio-economic status,
-nationality, personal appearance, race, religion, or sexual identity
-and orientation.
-
-We pledge to act and interact in ways that contribute to an open, welcoming,
-diverse, inclusive, and healthy community.
-
-## Our Standards
-
-Examples of behavior that contributes to a positive environment for our
-community include:
-
-* Demonstrating empathy and kindness toward other people
-* Being respectful of differing opinions, viewpoints, and experiences
-* Giving and gracefully accepting constructive feedback
-* Accepting responsibility and apologizing to those affected by our mistakes,
-  and learning from the experience
-* Focusing on what is best not just for us as individuals, but for the
-  overall community
-
-Examples of unacceptable behavior include:
-
-* The use of sexualized language or imagery, and sexual attention or
-  advances of any kind
-* Trolling, insulting or derogatory comments, and personal or political attacks
-* Public or private harassment
-* Publishing others' private information, such as a physical or email
-  address, without their explicit permission
-* Other conduct which could reasonably be considered inappropriate in a
-  professional setting
-
-## Enforcement Responsibilities
-
-Community leaders are responsible for clarifying and enforcing our standards of
-acceptable behavior and will take appropriate and fair corrective action in
-response to any behavior that they deem inappropriate, threatening, offensive,
-or harmful.
-
-Community leaders have the right and responsibility to remove, edit, or reject
-comments, commits, code, wiki edits, issues, and other contributions that are
-not aligned to this Code of Conduct, and will communicate reasons for moderation
-decisions when appropriate.
-
-## Scope
-
-This Code of Conduct applies within all community spaces, and also applies when
-an individual is officially representing the community in public spaces.
-Examples of representing our community include using an official e-mail address,
-posting via an official social media account, or acting as an appointed
-representative at an online or offline event.
-
-## Enforcement
-
-Instances of abusive, harassing, or otherwise unacceptable behavior may be
-reported to the community leaders responsible for enforcement on Discord
-at https://discord.gg/QYF8QrtEUm
-
-All complaints will be reviewed and investigated promptly and fairly.
-
-All community leaders are obligated to respect the privacy and security of the
-reporter of any incident.
-
-## Enforcement Guidelines
-
-Community leaders will follow these Community Impact Guidelines in determining
-the consequences for any action they deem in violation of this Code of Conduct:
-
-### 1. Correction
-
-**Community Impact**: Use of inappropriate language or other behavior deemed
-unprofessional or unwelcome in the community.
-
-**Consequence**: A private, written warning from community leaders, providing
-clarity around the nature of the violation and an explanation of why the
-behavior was inappropriate. A public apology may be requested.
-
-### 2. Warning
-
-**Community Impact**: A violation through a single incident or series
-of actions.
-
-**Consequence**: A warning with consequences for continued behavior. No
-interaction with the people involved, including unsolicited interaction with
-those enforcing the Code of Conduct, for a specified period of time. This
-includes avoiding interactions in community spaces as well as external channels
-like social media. Violating these terms may lead to a temporary or
-permanent ban.
-
-### 3. Temporary Ban
-
-**Community Impact**: A serious violation of community standards, including
-sustained inappropriate behavior.
-
-**Consequence**: A temporary ban from any sort of interaction or public
-communication with the community for a specified period of time. No public or
-private interaction with the people involved, including unsolicited interaction
-with those enforcing the Code of Conduct, is allowed during this period.
-Violating these terms may lead to a permanent ban.
-
-### 4. Permanent Ban
-
-**Community Impact**: Demonstrating a pattern of violation of community
-standards, including sustained inappropriate behavior,  harassment of an
-individual, or aggression toward or disparagement of classes of individuals.
-
-**Consequence**: A permanent ban from any sort of public interaction within
-the community.
-
-## Attribution
-
-This Code of Conduct is adapted from the [Contributor Covenant][homepage],
-version 2.0, available at
-https://www.contributor-covenant.org/version/2/0/code_of_conduct.html.
-
-Community Impact Guidelines were inspired by [Mozilla's code of conduct
-enforcement ladder](https://github.com/mozilla/diversity).
-
-[homepage]: https://www.contributor-covenant.org
-
-For answers to common questions about this code of conduct, see the FAQ at
-https://www.contributor-covenant.org/faq. Translations are available at
-https://www.contributor-covenant.org/translations.

.github/CONTRIBUTING.md

@@ -1,76 +0,0 @@
-# Contributing to axolotl
-
-First of all, thank you for your interest in contributing to axolotl! We appreciate the time and effort you're willing to invest in making our project better. This document provides guidelines and information to make the contribution process as smooth as possible.
-
-## Table of Contents
-
-- [Code of Conduct](#code-of-conduct)
-- [Getting Started](#getting-started)
-- [How to Contribute](#how-to-contribute)
-  - [Reporting Bugs](#reporting-bugs)
-  - [Suggesting Enhancements](#suggesting-enhancements)
-  - [Submitting Pull Requests](#submitting-pull-requests)
-- [Style Guidelines](#style-guidelines)
-  - [Code Style](#code-style)
-  - [Commit Messages](#commit-messages)
-- [Additional Resources](#additional-resources)
-
-## Code of Conductcode
-
-All contributors are expected to adhere to our [Code of Conduct](CODE_OF_CONDUCT.md). Please read it before participating in the axolotl community.
-
-## Getting Started
-
-Bugs? Please check for open issue else create a new [Issue](https://github.com/OpenAccess-AI-Collective/axolotl/issues/new).
-
-PRs are **greatly welcome**!
-
-1. Fork the repository and clone it to your local machine.
-2. Set up the development environment by following the instructions in the [README.md](https://github.com/OpenAccess-AI-Collective/axolotl/tree/main/README.md) file.
-3. Explore the codebase, run tests, and verify that everything works as expected.
-
-Please run below to setup env
-```bash
-pip3 install -r requirements-dev.txt -r requirements-tests.txt
-pre-commit install
-
-# test
-pytest tests/
-```
-
-## How to Contribute
-
-### Reporting Bugs
-
-If you encounter a bug or issue while using axolotl, please open a new issue on the [GitHub Issues](https://github.com/OpenAccess-AI-Collective/axolotl/issues) page. Provide a clear and concise description of the problem, steps to reproduce it, and any relevant error messages or logs.
-
-### Suggesting Enhancements
-
-We welcome ideas for improvements and new features. To suggest an enhancement, open a new issue on the [GitHub Issues](https://github.com/OpenAccess-AI-Collective/axolotl/issues) page. Describe the enhancement in detail, explain the use case, and outline the benefits it would bring to the project.
-
-### Submitting Pull Requests
-
-1. Create a new branch for your feature or bugfix. Use a descriptive name like `feature/your-feature-name` or `fix/your-bugfix-name`.
-2. Make your changes, following the [Style Guidelines](#style-guidelines) below.
-3. Test your changes and ensure that they don't introduce new issues or break existing functionality.
-4. Commit your changes, following the [commit message guidelines](#commit-messages).
-5. Push your branch to your fork on GitHub.
-6. Open a new pull request against the `main` branch of the axolotl repository. Include a clear and concise description of your changes, referencing any related issues.
-
-## Style Guidelines
-
-### Code Style
-
-axolotl uses [{codestyle}]({URLofCodestyle}) as its code style guide. Please ensure that your code follows these guidelines.
-
-### Commit Messages
-
-Write clear and concise commit messages that briefly describe the changes made in each commit. Use the imperative mood and start with a capitalized verb, e.g., "Add new feature" or "Fix bug in function".
-
-## Additional Resources
-
-- [GitHub Help](https://help.github.com/)
-- [GitHub Pull Request Documentation](https://docs.github.com/en/github/collaborating-with-issues-and-pull-requests)
-- [{codestyle}]({URLofCodestyle})
-
-Thank you once again for your interest in contributing to axolotl. We look forward to collaborating with you and creating an even better project together!

.github/FUNDING.yml

@@ -1,13 +0,0 @@
-# These are supported funding model platforms
-
-github: OpenAccess-AI-Collective # Replace with up to 4 GitHub Sponsors-enabled usernames e.g., [user1, user2]
-patreon: # Replace with a single Patreon username
-open_collective: # Replace with a single Open Collective username
-ko_fi: # Replace with a single Ko-fi username
-tidelift: # Replace with a single Tidelift platform-name/package-name e.g., npm/babel
-community_bridge: # Replace with a single Community Bridge project-name e.g., cloud-foundry
-liberapay: # Replace with a single Liberapay username
-issuehunt: # Replace with a single IssueHunt username
-otechie: # Replace with a single Otechie username
-lfx_crowdfunding: # Replace with a single LFX Crowdfunding project-name e.g., cloud-foundry
-custom: # Replace with up to 4 custom sponsorship URLs e.g., ['link1', 'link2']

.github/ISSUE_TEMPLATE/bug-report.yaml

@@ -1,105 +0,0 @@
-name: Bug Report
-description: File a bug report
-labels: ["bug", "needs triage"]
-body:
-  - type: markdown
-    attributes:
-      value: |
-        ## Before you start
-        Please **make sure you are on the latest version.**
-        If you encountered the issue after you installed, updated, or reloaded, **please try restarting before reporting the bug**.
-
-  - type: checkboxes
-    id: no-duplicate-issues
-    attributes:
-      label: "Please check that this issue hasn't been reported before."
-      description: "The **Label filters** may help make your search more focussed."
-      options:
-        - label: "I searched previous [Bug Reports](https://github.com/OpenAccess-AI-Collective/axolotl/labels/bug) didn't find any similar reports."
-          required: true
-
-  - type: textarea
-    id: expected
-    attributes:
-      label: Expected Behavior
-      description: Tell us what **should** happen.
-    validations:
-      required: true
-
-  - type: textarea
-    id: what-happened
-    attributes:
-      label: Current behaviour
-      description: |
-        Tell us what happens instead of the expected behavior.
-        Provide stacktrace and/or screenshots.
-    validations:
-      required: true
-
-  - type: textarea
-    id: reproduce
-    attributes:
-      label: Steps to reproduce
-      description: |
-        Which exact steps can a developer take to reproduce the issue?
-        The more detail you provide, the easier it will be to narrow down and fix the bug.
-        Please paste in tasks and/or queries **as text, not screenshots**.
-      placeholder: |
-        Example of the level of detail needed to reproduce any bugs efficiently and reliably.
-        1. Go to the '...' page.
-        2. Click on the '...' button.
-        3. Scroll down to '...'.
-        4. Observe the error.
-    validations:
-      required: true
-
-  - type: textarea
-    id: possible-solution
-    attributes:
-      label: Possible solution
-      description: |
-        Not obligatory, but please suggest a fix or reason for the bug, if you have an idea.
-
-
-  - type: checkboxes
-    id: operating-systems
-    attributes:
-      label: Which Operating Systems are you using?
-      description: You may select more than one.
-      options:
-        - label: Linux
-        - label: macOS
-        - label: Windows
-
-  - type: input
-    id: Python-version
-    attributes:
-      label: Python Version
-      description: Which {Programming} version are you using?
-      placeholder: 3.10 / please change accordingly
-    validations:
-      required: true
-
-  - type: input
-    id: axolotl-branch-commit
-    attributes:
-      label: axolotl branch-commit
-      description: On which branch/commit are you?
-      placeholder: main/4d6490b
-    validations:
-      required: true
-
-  - type: checkboxes
-    id: acknowledgements
-    attributes:
-      label: 'Acknowledgements'
-      description: 'Please confirm the following:'
-      options:
-        - label: 'My issue title is concise, descriptive, and in title casing.'
-          required: true
-        - label: 'I have searched the existing issues to make sure this bug has not been reported yet.'
-          required: true
-        - label: 'I am using the latest version of axolotl.'
-          required: true
-        - label: 'I have provided enough information for the maintainers to reproduce and diagnose the issue.'
-          required: true

.github/ISSUE_TEMPLATE/config.yml

@@ -1,7 +0,0 @@
-blank_issues_enabled: false
-contact_links:
-  - name: Ask a question
-    url: https://github.com/OpenAccess-AI-Collective/axolotl/discussions/categories/q-a
-    about: Ask questions and discuss with other community members
-  - name: Discuss the Project in Discord
-    url: https://discord.gg/HhrNrHJPRb

.github/ISSUE_TEMPLATE/docs.yml

@@ -1,46 +0,0 @@
-name: Documentation Improvement / Clarity
-description: Make a suggestion to improve the project documentation.
-labels: ['needs triage', 'docs']
-body:
-  - type: markdown
-    attributes:
-      value: '## :book: Documentation :book:'
-  - type: markdown
-    attributes:
-      value: |
-        * Ask questions in [Discord](https://discord.gg/HhrNrHJPRb).
-        * Before you file an issue read the [Contributing guide](./CONTRIBUTING.md).
-        * Check to make sure someone hasn't already opened a [similar issue](https://github.com/OpenAccess-AI-Collective/axolotl/issues).
-  - type: textarea
-    attributes:
-      label: What piece of documentation is affected?
-      description: Please link to the article you'd like to see updated.
-    validations:
-      required: true
-  - type: textarea
-    attributes:
-      label: What part(s) of the article would you like to see updated?
-      description: |
-        - Give as much detail as you can to help us understand the change you want to see.
-        - Why should the docs be changed? What use cases does it support?
-        - What is the expected outcome?
-    validations:
-      required: true
-  - type: textarea
-    attributes:
-      label: Additional Information
-      description: Add any other context or screenshots about the feature request here.
-    validations:
-      required: false
-  - type: checkboxes
-    id: acknowledgements
-    attributes:
-      label: 'Acknowledgements'
-      description: 'Please confirm the following:'
-      options:
-        - label: 'My issue title is concise, descriptive, and in title casing.'
-          required: true
-        - label: 'I have searched the existing issues to make sure this feature has not been requested yet.'
-          required: true
-        - label: 'I have provided enough information for the maintainers to understand and evaluate this request.'
-          required: true

.github/ISSUE_TEMPLATE/feature-request.yaml

@@ -1,63 +0,0 @@
-name: Feature Request / Enhancement
-description: Suggest a new feature or feature enhancement for the project
-labels: ["enhancement", "needs triage"]
-body:
-  - type: checkboxes
-    id: no-duplicate-issues
-    attributes:
-      label: "⚠️ Please check that this feature request hasn't been suggested before."
-      description: "There are two locations for previous feature requests. Please search in both. Thank you. The **Label filters** may help make your search more focussed."
-      options:
-        - label: "I searched previous [Ideas in Discussions](https://github.com/OpenAccess-AI-Collective/axolotl/discussions/categories/ideas) didn't find any similar feature requests."
-          required: true
-        - label: "I searched previous [Issues](https://github.com/OpenAccess-AI-Collective/axolotl/labels/enhancement) didn't find any similar feature requests."
-          required: true
-
-  - type: textarea
-    id: feature-description
-    validations:
-      required: true
-    attributes:
-      label: "🔖 Feature description"
-      description: "A clear and concise description of what the feature request is."
-      placeholder: "You should add ..."
-
-  - type: textarea
-    id: solution
-    validations:
-      required: true
-    attributes:
-      label: "✔️ Solution"
-      description: "A clear and concise description of what you want to happen, and why."
-      placeholder: "In my use-case, ..."
-
-  - type: textarea
-    id: alternatives
-    validations:
-      required: false
-    attributes:
-      label: "❓ Alternatives"
-      description: "A clear and concise description of any alternative solutions or features you've considered."
-      placeholder: "I have considered ..."
-
-  - type: textarea
-    id: additional-context
-    validations:
-      required: false
-    attributes:
-      label: "📝 Additional Context"
-      description: "Add any other context or screenshots about the feature request here."
-      placeholder: "..."
-
-  - type: checkboxes
-    id: acknowledgements
-    attributes:
-      label: 'Acknowledgements'
-      description: 'Please confirm the following:'
-      options:
-        - label: 'My issue title is concise, descriptive, and in title casing.'
-          required: true
-        - label: 'I have searched the existing issues to make sure this feature has not been requested yet.'
-          required: true
-        - label: 'I have provided enough information for the maintainers to understand and evaluate this request.'
-          required: true

.github/PULL_REQUEST_TEMPLATE/pull_request_template_simple.md

@@ -1,22 +0,0 @@
-<!--- Provide a general summary of your changes in the Title above -->
-
-# Description
-
-<!--- Describe your changes in detail -->
-
-## Motivation and Context
-
-<!--- Why is this change required? What problem does it solve? -->
-<!--- If it fixes an open issue, please link to the issue here. -->
-
-## How has this been tested?
-
-<!--- Please describe in detail how you tested your changes. -->
-<!--- Include details of your testing environment, tests ran to see how -->
-<!--- your change affects other areas of the code, etc. -->
-
-## Screenshots (if appropriate)
-
-## Types of changes
-
-<!--- What types of changes does your code introduce? Put an `x` in all the boxes that apply: -->

.github/SECURITY.md

@@ -1,9 +0,0 @@
-# Security Policy
-
-## Supported Versions
-
-Due to the nature of the fast development that is happening in this project, only the latest released version can be supported.
-
-## Reporting a Vulnerability
-
-If you find a vulnerability, please contact us on  [Discord](https://discord.gg/xcu3ECkH9a) rather than creating a GitHub issue to allow us some time to fix it before it is a known vulnerability to others.

.github/SUPPORT.md

@@ -1,10 +0,0 @@
-# Support
-
-If you need help with this project or have questions, please:
-
-1. Check the documentation.
-2. Search the existing issues and pull requests.
-3. Create a new issue if your question is not answered or your problem is not solved.
-4. Have a look in the [Discord server](https://discord.gg/HhrNrHJPRb)
-
-Please note that this project is maintained by volunteers who have limited availability. We'll do our best to address your questions and concerns in a timely manner.

.github/workflows/base.yml

@@ -18,13 +18,23 @@ jobs:
           - cuda: "118"
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.1
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 9.0+PTX"
+            pytorch: 2.0.0
+            axolotl_extras:
           - cuda: "118"
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.1
-            torch_cuda_arch_list: "7.0 7.5 8.0 8.6 9.0+PTX"
+            pytorch: 2.0.0
+            axolotl_extras:
+          - cuda: "117"
+            cuda_version: 11.7.1
+            python_version: "3.9"
+            pytorch: 1.13.1
+            axolotl_extras:
+          - cuda: "118"
+            cuda_version: 11.8.0
+            python_version: "3.9"
+            pytorch: 2.0.0
+            axolotl_extras: gptq
     steps:
       - name: Checkout
         uses: actions/checkout@v3
@@ -48,9 +58,11 @@ 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 }}
           labels: ${{ steps.metadata.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
           build-args: |
             CUDA_VERSION=${{ matrix.cuda_version }}
             CUDA=${{ matrix.cuda }}
             PYTHON_VERSION=${{ matrix.python_version }}
             PYTORCH_VERSION=${{ matrix.pytorch }}
-            TORCH_CUDA_ARCH_LIST=${{ matrix.torch_cuda_arch_list }}
+            AXOLOTL_EXTRAS=${{ matrix.axolotl_extras }}

.github/workflows/main.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - "main"
+      - "dev"
 
 jobs:
   build-axolotl:
@@ -13,17 +14,27 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - cuda: 118
+          - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.1
+            pytorch: 2.0.0
             axolotl_extras:
-          - cuda: 118
+          - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.1
+            pytorch: 2.0.0
             axolotl_extras:
-    runs-on: [self-hosted, gpu, docker]
+          - cuda: cu118
+            cuda_version: 11.8.0
+            python_version: "3.9"
+            pytorch: 2.0.0
+            axolotl_extras: gptq
+          - cuda: cu117
+            cuda_version: 11.7.1
+            python_version: "3.9"
+            pytorch: 1.13.1
+            axolotl_extras:
+    runs-on: self-hosted
     steps:
       - name: Checkout
         uses: actions/checkout@v3
@@ -44,12 +55,13 @@ jobs:
         with:
           context: .
           build-args: |
-            BASE_TAG=${{ github.ref_name }}-base-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}
-            CUDA=${{ matrix.cuda }}
+            BASE_TAG=${{ github.ref_name }}-base-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           file: ./docker/Dockerfile
           push: ${{ github.event_name != 'pull_request' }}
-          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
+          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           labels: ${{ steps.metadata.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
   build-axolotl-runpod:
     needs: build-axolotl
     if: github.repository_owner == 'OpenAccess-AI-Collective'
@@ -57,18 +69,27 @@ jobs:
     strategy:
       matrix:
         include:
-          - cuda: 118
+          - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.9"
-            pytorch: 2.0.1
+            pytorch: 2.0.0
             axolotl_extras:
-          - cuda: 118
+          - cuda: cu118
             cuda_version: 11.8.0
             python_version: "3.10"
-            pytorch: 2.0.1
+            pytorch: 2.0.0
             axolotl_extras:
-            is_latest: true
-    runs-on: [self-hosted, gpu, docker]
+          - cuda: cu118
+            cuda_version: 11.8.0
+            python_version: "3.9"
+            pytorch: 2.0.0
+            axolotl_extras: gptq
+          - cuda: cu117
+            cuda_version: 11.7.1
+            python_version: "3.9"
+            pytorch: 1.13.1
+            axolotl_extras:
+    runs-on: self-hosted
     steps:
       - name: Checkout
         uses: actions/checkout@v3
@@ -89,11 +110,10 @@ jobs:
         with:
           context: .
           build-args: |
-            BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
-            CUDA=${{ matrix.cuda }}
+            BASE_TAG=${{ github.ref_name }}-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           file: ./docker/Dockerfile-runpod
           push: ${{ github.event_name != 'pull_request' }}
-          tags: |
-             ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-cu${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
-             ${{ (matrix.is_latest) && format('{0}-latest', steps.metadata.outputs.tags) || '' }}
+          tags: ${{ steps.metadata.outputs.tags }}-py${{ matrix.python_version }}-${{ matrix.cuda }}-${{ matrix.pytorch }}${{ matrix.axolotl_extras != '' && '-' || '' }}${{ matrix.axolotl_extras }}
           labels: ${{ steps.metadata.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max

.github/workflows/pre-commit.yml

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

.github/workflows/pypi.yml

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

.github/workflows/tests.yml

@@ -1,26 +1,10 @@
-name: Tests
+name: PyTest
 on:
-  # check on push/merge to main, PRs, and manual triggers
   push:
-    branches:
-      - "main"
   pull_request:
-  workflow_dispatch:
 
 jobs:
-  pre-commit:
-    name: pre-commit
-    runs-on: ubuntu-latest
-    steps:
-      - uses: actions/checkout@v3
-      - uses: actions/setup-python@v4
-        with:
-          python-version: "3.9"
-          cache: 'pip' # caching pip dependencies
-      - uses: pre-commit/action@v3.0.0
-
-  pytest:
-    name: PyTest
+  test:
     runs-on: ubuntu-latest
     strategy:
       fail-fast: false
@@ -40,35 +24,9 @@ jobs:
 
       - name: Install dependencies
         run: |
-          pip3 install -e .
-          pip3 install -r requirements-tests.txt
+          pip install -e .
+          pip install -r requirements-tests.txt
 
       - name: Run tests
         run: |
-          pytest --ignore=tests/e2e/ tests/
-
-  e2e-test:
-    name: E2E Tests
-    runs-on: [self-hosted, gpu]
-    timeout-minutes: 20
-    needs: [pre-commit, pytest]
-
-    steps:
-      - name: Check out repository code
-        uses: actions/checkout@v3
-
-      - name: Setup Python
-        uses: actions/setup-python@v4
-        with:
-          python-version: "3.10"
-#          cache: 'pip' # caching pip dependencies
-
-      - name: Install dependencies
-        run: |
-          pip3 install -e .
-          pip3 install flash-attn
-          pip3 install -r requirements-tests.txt
-
-      - name: Run e2e tests
-        run: |
-          pytest tests/e2e/
+          pytest tests/

.mypy.ini

@@ -8,9 +8,6 @@ ignore_missing_imports = True
 [mypy-axolotl.monkeypatch.*]
 ignore_errors = True
 
-[mypy-axolotl.models.phi.*]
-ignore_errors = True
-
 [mypy-flash_attn.*]
 ignore_missing_imports = True
 
@@ -23,9 +20,6 @@ ignore_missing_imports = True
 [mypy-peft]
 ignore_missing_imports = True
 
-[mypy-wandb]
-ignore_missing_imports = True
-
 [mypy-bitsandbytes]
 ignore_missing_imports = True
 

LICENSE

@@ -1,202 +0,0 @@
-
-                                 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.

README.md

@@ -1,52 +1,10 @@
 # Axolotl
 
-Axolotl is a tool designed to streamline the fine-tuning of various AI models, offering support for multiple configurations and architectures.
-
-Features:
-- Train various Huggingface models such as llama, pythia, falcon, mpt
-- 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
-- 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
-- And more!
-
-
-<table>
-<tr>
-<td>
-
-## Table of Contents
-- [Introduction](#axolotl)
-- [Supported Features](#axolotl-supports)
-- [Quickstart](#quickstart-)
-- [Installation](#installation)
-  - [Docker Installation](#environment)
-  - [Conda/Pip venv Installation](#condapip-venv)
-  - [LambdaLabs Installation](#lambdalabs)
-- [Dataset](#dataset)
-  - [How to Add Custom Prompts](#how-to-add-custom-prompts)
-  - [How to Use Custom Pretokenized Dataset](#how-to-use-your-custom-pretokenized-dataset)
-- [Config](#config)
-  - [Train](#train)
-  - [Inference](#inference)
-  - [Merge LORA to Base](#merge-lora-to-base)
-- [Common Errors](#common-errors-)
-- [Need Help?](#need-help-)
-- [Badge](#badge-)
-- [Community Showcase](#community-showcase)
-- [Contributing](#contributing-)
-
-</td>
-<td>
-
 <div align="center">
   <img src="image/axolotl.png" alt="axolotl" width="160">
   <div>
     <p>
-      <b>Axolotl provides a unified repository for fine-tuning <br />a variety of AI models with ease</b>
+      <b>One repo to finetune them all! </b>
     </p>
     <p>
       Go ahead and axolotl questions!!
@@ -56,45 +14,36 @@ Features:
   </div>
 </div>
 
-</td>
-</tr>
-</table>
-
 ## Axolotl supports
 
-|          | fp16/fp32 | lora | qlora | gptq | gptq w/flash attn | flash attn | xformers attn |
-|----------|:----------|:-----|-------|------|-------------------|------------|--------------|
-| llama    | ✅         | ✅    | ✅     | ✅             | ✅                 | ✅          | ✅            |
-| Pythia   | ✅         | ✅    | ✅     | ❌             | ❌                 | ❌          | ❓            |
-| cerebras | ✅         | ✅    | ✅     | ❌             | ❌                 | ❌          | ❓            |
-| btlm     | ✅         | ✅    | ✅     | ❌             | ❌                 | ❌          | ❓            |
-| mpt      | ✅         | ❌    | ❓     | ❌             | ❌                 | ❌          | ❓            |
-| falcon   | ✅         | ✅    | ✅     | ❌             | ❌                 | ❌          | ❓            |
-| gpt-j    | ✅         | ✅    | ✅     | ❌             | ❌                 | ❓          | ❓            |
-| XGen     | ✅         | ❓    | ✅     | ❓             | ❓                 | ❓          | ✅            |
-| phi      | ✅         | ✅    | ✅     | ❓             | ❓                 | ❓          | ❓            |
+|          | fp16/fp32 | lora | qlora | gptq | gptq w/ lora | gptq w/flash attn | flash attn | xformers attn |
+|----------|:----------|:-----|-------|------|:-------------|-------------------|------------|---------------|
+| llama    | ✅         | ✅    | ✅     | ✅    | ✅             | ✅                 | ✅          | ✅             |
+| Pythia   | ✅         | ✅    | ✅     | ❌    | ❓            | ❌                 | ❌          | ❓             |
+| cerebras | ✅         | ✅    | ✅     | ❌    | ❓            | ❌                 | ❌          | ✅             |
+| mpt      | ✅         | ❌    | ❓     | ❌    | ❓            | ❌                 | ❌          | ❓             |
+| falcon   | ✅         | ✅    | ✅     | ❌    | ❓            | ❌                 | ❌          | ✅             |
+| gpt-j    | ✅         | ✅    | ✅     | ❌    | ❓            | ❌                 | ❓          | ✅             |
 
 
 ## Quickstart ⚡
 
-Get started with Axolotl in just a few steps! This quickstart guide will walk you through setting up and running a basic fine-tuning task.
-
-**Requirements**: Python >=3.9 and Pytorch >=2.0.
+**Requirements**: Python 3.9 and Pytorch 2.0.
 
 ```bash
 git clone https://github.com/OpenAccess-AI-Collective/axolotl
-cd axolotl
 
-pip3 install packaging
-pip3 install -e .[flash-attn]
+pip3 install -e .
 pip3 install -U git+https://github.com/huggingface/peft.git
 
+accelerate config
+
 # finetune lora
-accelerate launch -m axolotl.cli.train examples/openllama-3b/lora.yml
+accelerate launch scripts/finetune.py examples/openllama-3b/lora.yml
 
 # inference
-accelerate launch -m axolotl.cli.inference examples/openllama-3b/lora.yml \
-    --lora_model_dir="./lora-out"
+accelerate launch scripts/finetune.py examples/openllama-3b/lora.yml \
+    --inference --lora_model_dir="./lora-out"
 ```
 
 ## Installation
@@ -103,9 +52,11 @@ accelerate launch -m axolotl.cli.inference examples/openllama-3b/lora.yml \
 
 - Docker
   ```bash
-  docker run --gpus '"all"' --rm -it winglian/axolotl:main-py3.10-cu118-2.0.1
+  docker run --gpus '"all"' --rm -it winglian/axolotl:main-py3.9-cu118-2.0.0
   ```
-  - `winglian/axolotl-runpod:main-latest`: for runpod or use this [direct link](https://runpod.io/gsc?template=v2ickqhz9s&ref=6i7fkpdz)
+  - `winglian/axolotl-runpod:main-py3.9-cu118-2.0.0`: for runpod
+  - `winglian/axolotl-runpod:main-py3.9-cu118-2.0.0-gptq`: for gptq
+  - `winglian/axolotl:dev`: dev branch (not usually up to date)
 
   Or run on the current files for development:
 
@@ -114,14 +65,23 @@ accelerate launch -m axolotl.cli.inference examples/openllama-3b/lora.yml \
   ```
 
 - Conda/Pip venv
-  1. Install python >=**3.9**
+  1. Install python **3.9**
 
   2. Install pytorch stable https://pytorch.org/get-started/locally/
 
-  3. Install axolotl along with python dependencies
+  3. Install python dependencies with ONE of the following:
+      - Recommended, supports QLoRA, NO gptq/int4 support
         ```bash
-        pip3 install packaging
-        pip3 install -e .[flash-attn]
+        pip3 install -e .
+        pip3 install -U git+https://github.com/huggingface/peft.git
+        ```
+      - gptq/int4 support, NO QLoRA
+        ```bash
+        pip3 install -e .[gptq]
+        ```
+      - same as above but not recommended
+        ```bash
+        pip3 install -e .[gptq_triton]
         ```
 
 - LambdaLabs
@@ -148,7 +108,7 @@ accelerate launch -m axolotl.cli.inference examples/openllama-3b/lora.yml \
 
   3. Install torch
   ```bash
-  pip3 install -U torch --index-url https://download.pytorch.org/whl/cu118
+  pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
   ```
 
   4. Axolotl
@@ -156,10 +116,12 @@ accelerate launch -m axolotl.cli.inference examples/openllama-3b/lora.yml \
   git clone https://github.com/OpenAccess-AI-Collective/axolotl
   cd axolotl
 
-  pip3 install packaging
-  pip3 install -e .[flash-attn]
+  pip3 install -e . # change depend on needs
   pip3 install protobuf==3.20.3
-  pip3 install -U --ignore-installed requests Pillow psutil scipy
+  pip3 install -U requests
+  pip3 install -U --ignore-installed psutil
+  pip3 install -U scipy
+  pip3 install git+https://github.com/huggingface/peft.git # not for gptq
   ```
 
   5. Set path
@@ -168,18 +130,15 @@ accelerate launch -m axolotl.cli.inference examples/openllama-3b/lora.yml \
   ```
   </details>
 
-- Windows: Please use WSL or Docker!
-
 ### Dataset
 
-Axolotl supports a variety of dataset formats. Below are some of the formats you can use.
 Have dataset(s) in one of the following format (JSONL recommended):
 
 - `alpaca`: instruction; input(optional)
   ```json
   {"instruction": "...", "input": "...", "output": "..."}
   ```
-- `sharegpt:chat`: conversations where `from` is `human`/`gpt`
+- `sharegpt:chat`: conversations
   ```json
   {"conversations": [{"from": "...", "value": "..."}]}
   ```
@@ -264,18 +223,10 @@ Have dataset(s) in one of the following format (JSONL recommended):
   ```json
   {"conversations": [{"role": "...", "value": "..."}]}
   ```
-- `metharme`: instruction, adds additional eos tokens
-  ```json
-  {"prompt": "...", "generation": "..."}
-  ```
 - `sharegpt_simple.load_role`: conversations where `role` is used instead of `from`
   ```json
   {"conversations": [{"role": "...", "value": "..."}]}
   ```
-- `sharegpt_simple.load_guanaco`: conversations where `from` is `prompter`/`assistant` instead of default sharegpt
-  ```json
-  {"conversations": [{"from": "...", "value": "..."}]}
-  ```
 - `sharegpt_jokes`: creates a chat where bot is asked to tell a joke, then explain why the joke is funny
   ```json
   {"conversations": [{"title": "...", "text": "...", "explanation": "..."}]}
@@ -285,34 +236,16 @@ Have dataset(s) in one of the following format (JSONL recommended):
 
 #### How to add custom prompts
 
-Using yaml. Example:
-```yaml
-datasets:
-  - path: repo
-    type:
-      system_prompt: ""
-      no_input_format: |-
-        User: {instruction}<|end_of_turn|>
-        Assistant:
-      format: |-
-        User: {instruction}
-        {input}<|end_of_turn|>
-        Assistant:
-```
+  1. Add your method to a file in [prompt_strategies](src/axolotl/prompt_strategies). Please see other files as example.
+  2. Use your custom file name as the dataset type `<prompt_strategies_file>.load_<load_fn>`.
 
-Using file:
-1. Add your method to a file in [prompt_strategies](src/axolotl/prompt_strategies). Please see other files as example.
-2. Use your custom file name as the dataset type `<prompt_strategies_file>.load_<load_fn>`.
+Optionally, download some datasets, see [data/README.md](data/README.md)
 
-#### How to use your custom pretokenized dataset
-
-- Do not pass a `type:`
-- Dataset must contain `input_ids`, `attention_mask`, `labels` in columns
 
 
 ### Config
 
-See [examples](examples) for quick start. It is recommended to duplicate and modify to your needs. The most important options are:
+See sample configs in [configs](configs) folder or [examples](examples) for quick start. It is recommended to duplicate and modify to your needs. The most important options are:
 
 - model
   ```yaml
@@ -329,27 +262,11 @@ See [examples](examples) for quick start. It is recommended to duplicate and mod
     - path: vicgalle/alpaca-gpt4
       type: alpaca # format from earlier
 
-  # huggingface repo with specific configuration/subset
-  datasets:
-    - path: EleutherAI/pile
-      name: enron_emails
-      type: completion # format from earlier
-      field: text # Optional[str] default: text, field to use for completion data
-
-  # huggingface repo with multiple named configurations/subsets
-  datasets:
-    - path: bigcode/commitpackft
-      name:
-        - ruby
-        - python
-        - typescript
-      type: ... # unimplemented custom format
-
   # local
   datasets:
-    - path: data.jsonl # or json
-      ds_type: json # see other options below
-      type: alpaca
+    - path: json
+      data_files: data.jsonl # or json
+      type: alpaca # format from earlier
   ```
 
 - loading
@@ -401,11 +318,6 @@ tokenizer_type: AutoTokenizer
 trust_remote_code:
 # use_fast option for tokenizer loading from_pretrained, default to True
 tokenizer_use_fast:
-# Whether to use the legacy tokenizer setting, defaults to True
-tokenizer_legacy:
-# resize the model embeddings when new tokens are added to multiples of 32
-# this is reported to improve training speed on some models
-resize_token_embeddings_to_32x:
 
 # whether you are training a 4-bit GPTQ quantized model
 gptq: true
@@ -424,41 +336,14 @@ fp16: true
 # Use CUDA tf32
 tf32: true # require >=ampere
 
-# No AMP (automatic mixed precision)
-bfloat16: true # require >=ampere
-float16: true
-
 # a list of one or more datasets to finetune the model with
 datasets:
   # hf dataset repo | "json" for local dataset, make sure to fill data_files
   - path: vicgalle/alpaca-gpt4
   # The type of prompt to use for training. [alpaca, sharegpt, gpteacher, oasst, reflection]
     type: alpaca # format | format:<prompt_style> (chat/instruct) | <prompt_strategies>.load_<load_fn>
-    ds_type: # Optional[str] (json|arrow|parquet|text|csv) defines the datatype when path is a file
-    data_files: # Optional[str] path to source data files
-    shards: # Optional[int] number of shards to split data into
-    name: # Optional[str] name of dataset configuration to load
-
-  # custom user prompt
-  - path: repo
-    type:
-      # the below are defaults. only set what's needed.
-      system_prompt: ""
-      field_system: system
-      field_instruction: instruction
-      field_output: input
-
-      # customizable to be single line or multi-line
-      system_format: "{system}"
-      # 'format' can include {input}
-      format: |-
-        User: {instruction} {input}
-        Assistant:
-      # 'no_input_format' cannot include {input}
-      no_input_format: "{instruction} "
-
-      # for completions datsets, uses the provided field if not `text`
-      field:
+    data_files: # path to source data files
+    shards: # number of shards to split data into
 
 # axolotl attempts to save the dataset as an arrow after packing the data together so
 # subsequent training attempts load faster, relative path
@@ -466,14 +351,11 @@ dataset_prepared_path: data/last_run_prepared
 # push prepared dataset to hub
 push_dataset_to_hub: # repo path
 # push checkpoints to hub
-hub_model_id: # repo path to push finetuned model
-# how to push checkpoints to hub
-# https://huggingface.co/docs/transformers/v4.31.0/en/main_classes/trainer#transformers.TrainingArguments.hub_strategy
-hub_strategy:
+hub_model_id: # repo path
 # whether to use hf `use_auth_token` for loading datasets. Useful for fetching private datasets
 # required to be true when used in combination with `push_dataset_to_hub`
 hf_use_auth_token: # boolean
-# How much of the dataset to set aside as evaluation. 1 = 100%, 0.50 = 50%, etc. 0 for no eval.
+# How much of the dataset to set aside as evaluation. 1 = 100%, 0.50 = 50%, etc
 val_set_size: 0.04
 # Num shards for whole dataset
 dataset_shard_num:
@@ -483,19 +365,9 @@ dataset_shard_idx:
 # the maximum length of an input to train with, this should typically be less than 2048
 # as most models have a token/context limit of 2048
 sequence_len: 2048
-# pad inputs so each step uses constant sized buffers
-# this will reduce memory fragmentation and may prevent OOMs, by re-using memory more efficiently
-pad_to_sequence_len:
 # max sequence length to concatenate training samples together up to
 # inspired by StackLLaMA. see https://huggingface.co/blog/stackllama#supervised-fine-tuning
-# FutureWarning: This will soon be DEPRECATED
 max_packed_sequence_len: 1024
-# use efficient multi-packing with block diagonal attention and per sequence position_ids. Recommend set to 'true'
-sample_packing:
-# you can set these packing optimizations AFTER starting a training at least once.
-# The trainer will provide recommended values for these values.
-sample_packing_eff_est:
-total_num_tokens:
 
 # if you want to use 'lora' or 'qlora' or leave blank to train all parameters in original model
 adapter: lora
@@ -520,27 +392,16 @@ lora_modules_to_save:
 lora_out_dir:
 lora_fan_in_fan_out: false
 
-# ReLoRA configuration
-# must use either 'lora' or 'qlora' adapter, and does not support fsdp or deepspeed
-relora_steps: # number of steps per ReLoRA restart
-relora_warmup_steps: # number of per-restart warmup steps
-relora_cpu_offload: # true to perform lora weight merges on cpu during restarts, for modest gpu memory savings
-
 # wandb configuration if you're using it
-wandb_mode: # "offline" to save run metadata locally and not sync to the server, "disabled" to turn off wandb
-wandb_project: # your wandb project name
-wandb_entity: # a wandb Team name if using a Team
+wandb_mode:
+wandb_project:
 wandb_watch:
-wandb_run_id: # set the name of your wandb run
-wandb_log_model: # "checkpoint" to log model to wandb Artifacts every `save_steps` or "end" to log only at the end of training
+wandb_run_id:
+wandb_log_model: # 'checkpoint'
 
 # where to save the finished model to
 output_dir: ./completed-model
 
-# whether to use torch.compile and which backend to use
-torch_compile:  # bool
-torch_compile_backend:  # Optional[str]
-
 # training hyperparameters
 gradient_accumulation_steps: 1
 micro_batch_size: 2
@@ -548,25 +409,16 @@ eval_batch_size: 2
 num_epochs: 3
 warmup_steps: 100
 learning_rate: 0.00003
-lr_quadratic_warmup:
 logging_steps:
-save_strategy: # set to `no` to skip checkpoint saves
-save_steps: # leave empty to save at each epoch
-eval_steps: # leave empty to eval at each epoch
-save_total_limit: # checkpoints saved at a time
-max_steps:
-
-eval_table_size: # approximate number of predictions sent to wandb depending on batch size. Enabled above 0. Default is 0
-eval_table_max_new_tokens: # total number of tokens generated for predictions sent to wandb. Default is 128
+save_steps:
+eval_steps:
 
 # save model as safetensors (require safetensors package)
 save_safetensors:
 
 # whether to mask out or include the human's prompt from the training labels
 train_on_inputs: false
-# group similarly sized data to minimize padding
-# may be slower to start, as it must download and sort the entire dataset
-# note that training loss may have an oscillating pattern with this enabled
+# don't use this, leads to wonky training (according to someone on the internet)
 group_by_length: false
 
 # Whether to use gradient checkpointing https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing
@@ -588,30 +440,6 @@ log_sweep_min_lr:
 log_sweep_max_lr:
 
 # specify optimizer
-# Valid values are driven by the Transformers OptimizerNames class, see:
-# https://github.com/huggingface/transformers/blob/95b374952dc27d8511541d6f5a4e22c9ec11fb24/src/transformers/training_args.py#L134
-#
-# Note that not all optimizers may be available in your environment, ex: 'adamw_anyprecision' is part of
-# torchdistx, 'adamw_bnb_8bit' is part of bnb.optim.Adam8bit, etc. When in doubt, it is recommended to start with the optimizer used
-# in the examples/ for your model and fine-tuning use case.
-#
-# Valid values for 'optimizer' include:
-# - adamw_hf
-# - adamw_torch
-# - adamw_torch_fused
-# - adamw_torch_xla
-# - adamw_apex_fused
-# - adafactor
-# - adamw_anyprecision
-# - sgd
-# - adagrad
-# - adamw_bnb_8bit
-# - lion_8bit
-# - lion_32bit
-# - paged_adamw_32bit
-# - paged_adamw_8bit
-# - paged_lion_32bit
-# - paged_lion_8bit
 optimizer:
 # specify weight decay
 weight_decay:
@@ -626,8 +454,8 @@ max_grad_norm:
 flash_optimum:
 # whether to use xformers attention patch https://github.com/facebookresearch/xformers:
 xformers_attention:
-# whether to use flash attention patch https://github.com/Dao-AILab/flash-attention:
-flash_attention:
+# whether to use flash attention patch https://github.com/HazyResearch/flash-attention:
+flash_attention:  # require a100 for llama
 # whether to use scaled-dot-product attention
 # https://pytorch.org/docs/stable/generated/torch.nn.functional.scaled_dot_product_attention.html
 sdp_attention:
@@ -636,10 +464,6 @@ landmark_attention:
 # xpos RoPE see https://github.com/kaiokendev/cutoff-len-is-context-len/blob/main/util/xpos_rope_llama_monkey_patch.py
 # llama only
 xpos_rope:
-# RoPE Scaling https://github.com/huggingface/transformers/pull/24653
-rope_scaling:
-  type: # linear | dynamic
-  factor: # float
 
 # resume from a specific checkpoint dir
 resume_from_checkpoint:
@@ -662,19 +486,14 @@ tokens:
 fsdp:
 fsdp_config:
 
-# Deepspeed config path
+# Deepspeed
 deepspeed:
 
-# Advanced DDP Arguments
-ddp_timeout:
-ddp_bucket_cap_mb:
-ddp_broadcast_buffers:
-
 # Path to torch distx for optim 'adamw_anyprecision'
 torchdistx_path:
 
-# Set to HF dataset for type: 'completion' for streaming instead of pre-tokenize
-pretraining_dataset:
+# Set padding for data collator to 'longest'
+collator_pad_to_longest:
 
 # Debug mode
 debug:
@@ -688,48 +507,22 @@ strict:
 
 </details>
 
+### Accelerate
+
+Configure accelerate
+
+```bash
+accelerate config
+
+# Edit manually
+# nano ~/.cache/huggingface/accelerate/default_config.yaml
+```
+
 ### Train
 
 Run
 ```bash
-accelerate launch -m axolotl.cli.train your_config.yml
-```
-
-#### Multi-GPU
-
-You can optionally pre-tokenize dataset with the following before finetuning:
-```bash
-CUDA_VISIBLE_DEVICES="" accelerate launch -m axolotl.cli.train your_config.yml --prepare_ds_only
-```
-
-##### Config
-
-- llama FSDP
-```yaml
-fsdp:
-  - full_shard
-  - auto_wrap
-fsdp_config:
-  fsdp_offload_params: true
-  fsdp_state_dict_type: FULL_STATE_DICT
-  fsdp_transformer_layer_cls_to_wrap: LlamaDecoderLayer
-```
-
-- llama Deepspeed
-```yaml
-deepspeed: deepspeed/zero3.json
-```
-
-##### Weights & Biases Logging
-
-- wandb options
-```yaml
-wandb_mode:
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
+accelerate launch scripts/finetune.py configs/your_config.yml
 ```
 
 ### Inference
@@ -738,16 +531,16 @@ Pass the appropriate flag to the train command:
 
 - Pretrained LORA:
   ```bash
-  python -m axolotl.cli.inference examples/your_config.yml --lora_model_dir="./lora-output-dir"
+  --inference --lora_model_dir="./lora-output-dir"
   ```
 - Full weights finetune:
   ```bash
-  python -m axolotl.cli.inference examples/your_config.yml --base_model="./completed-model"
+  --inference --base_model="./completed-model"
   ```
 - Full weights finetune w/ a prompt from a text file:
   ```bash
-  cat /tmp/prompt.txt | python -m axolotl.cli.inference examples/your_config.yml \
-    --base_model="./completed-model" --prompter=None --load_in_8bit=True
+  cat /tmp/prompt.txt | python scripts/finetune.py configs/your_config.yml \
+    --base_model="./completed-model" --inference --prompter=None --load_in_8bit=True
   ```
 
 ### Merge LORA to base
@@ -755,18 +548,18 @@ Pass the appropriate flag to the train command:
 Add below flag to train command above
 
 ```bash
-python3 -m axolotl.cli.merge_lora examples/your_config.yml --lora_model_dir="./completed-model" --load_in_8bit=False --load_in_4bit=False
+--merge_lora --lora_model_dir="./completed-model" --load_in_8bit=False --load_in_4bit=False
 ```
 
 If you run out of CUDA memory, you can try to merge in system RAM with
 
 ```bash
-CUDA_VISIBLE_DEVICES="" python3 -m axolotl.cli.merge_lora ...
+CUDA_VISIBLE_DEVICES="" python3 scripts/finetune.py ...
 ```
 
 ## Common Errors 🧰
 
-> If you encounter a 'Cuda out of memory' error, it means your GPU ran out of memory during the training process. Here's how to resolve it:
+> Cuda out of memory
 
 Please reduce any below
   - `micro_batch_size`
@@ -774,12 +567,6 @@ Please reduce any below
   - `gradient_accumulation_steps`
   - `sequence_len`
 
-> `failed (exitcode: -9)`
-
-Usually means your system has run out of system memory.
-Similarly, you should consider reducing the same settings as when you run out of VRAM.
-Additionally, look into upgrading your system RAM which should be simpler than GPU upgrades.
-
 > RuntimeError: expected scalar type Float but found Half
 
 Try set `fp16: true`
@@ -788,14 +575,6 @@ Try set `fp16: true`
 
 Try to turn off xformers.
 
-> accelerate config missing
-
-It's safe to ignore it.
-
-> NCCL Timeouts during training
-
-See the [NCCL](docs/nccl.md) guide.
-
 ## Need help? 🙋♂️
 
 Join our [Discord server](https://discord.gg/HhrNrHJPRb) where we can help you
@@ -812,8 +591,6 @@ Building something cool with Axolotl? Consider adding a badge to your model card
 
 ## Community Showcase
 
-Check out some of the projects and models that have been built using Axolotl! Have a model you'd like to add to our Community Showcase? Open a PR with your model.
-
 Open Access AI Collective
 - [Minotaur 13b](https://huggingface.co/openaccess-ai-collective/minotaur-13b)
 - [Manticore 13b](https://huggingface.co/openaccess-ai-collective/manticore-13b)
@@ -824,9 +601,7 @@ PocketDoc Labs
 
 ## Contributing 🤝
 
-Please read the [contributing guide](./.github/CONTRIBUTING.md)
-
-Bugs? Please check the [open issues](https://github.com/OpenAccess-AI-Collective/axolotl/issues/bug) else create a new Issue.
+Bugs? Please check for open issue else create a new [Issue](https://github.com/OpenAccess-AI-Collective/axolotl/issues/new).
 
 PRs are **greatly welcome**!
 

data/README.md

@@ -0,0 +1,24 @@
+
+## Download some datasets
+```shell
+curl https://raw.githubusercontent.com/tloen/alpaca-lora/main/alpaca_data_gpt4.json -o data/raw/alpaca_data_gpt4.json
+curl https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json -L -o data/raw/vicuna_cleaned.json
+curl https://github.com/teknium1/GPTeacher/blob/main/Instruct/gpt4-instruct-similarity-0.6-dataset.json?raw=true -L -o data/raw/gpt4-instruct-similarity-0.6-dataset.json
+curl https://github.com/teknium1/GPTeacher/blob/main/Roleplay/roleplay-similarity_0.6-instruct-dataset.json?raw=true -L -o data/raw/roleplay-similarity_0.6-instruct-dataset.json
+```
+
+## Convert the JSON data files to JSONL.
+
+```shell
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/alpaca_data_gpt4.json --output data/alpaca_data_gpt4.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/raw/vicuna_cleaned.json --output data/vicuna_cleaned.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/raw/roleplay-similarity_0.6-instruct-dataset.json --output data/roleplay-similarity_0.6-instruct-dataset.jsonl
+python3 ./scripts/alpaca_json_to_jsonl.py --file data/raw/gpt4-instruct-similarity-0.6-dataset.json --output data/gpt4-instruct-similarity-0.6-dataset.jsonl
+```
+---
+
+Using JSONL makes it easier to subset the data if you want a smaller training set, i.e get 2000 random examples.
+
+```shell
+shuf -n2000 data/vicuna_cleaned.jsonl > data/vicuna_cleaned.subset0.jsonl
+```

data/raw/.gitignore

@@ -0,0 +1 @@
+**

deepspeed/zero1.json

@@ -1,39 +0,0 @@
-{
-    "zero_optimization": {
-      "stage": 1,
-      "overlap_comm": true
-    },
-    "bf16": {
-      "enabled": "auto"
-    },
-    "fp16": {
-      "enabled": "auto",
-      "auto_cast": false,
-      "loss_scale": 0,
-      "initial_scale_power": 32,
-      "loss_scale_window": 1000,
-      "hysteresis": 2,
-      "min_loss_scale": 1
-    },
-    "optimizer": {
-      "type": "AdamW",
-      "params": {
-        "lr": "auto",
-        "betas": "auto",
-        "eps": "auto",
-        "weight_decay": "auto"
-      }
-    },
-    "scheduler": {
-      "type": "WarmupDecayLR",
-      "params": {
-        "warmup_min_lr": "auto",
-        "warmup_max_lr": "auto",
-        "warmup_num_steps": "auto",
-        "total_num_steps": "auto"
-      }
-    },
-    "train_batch_size": "auto",
-    "train_micro_batch_size_per_gpu": "auto",
-    "wall_clock_breakdown": false
-}

deepspeed/zero2.json

@@ -1,43 +0,0 @@
-{
-    "zero_optimization": {
-      "stage": 2,
-      "offload_optimizer": {
-        "device": "cpu"
-      },
-      "contiguous_gradients": true,
-      "overlap_comm": true
-    },
-    "bf16": {
-      "enabled": "auto"
-    },
-    "fp16": {
-      "enabled": "auto",
-      "auto_cast": false,
-      "loss_scale": 0,
-      "initial_scale_power": 32,
-      "loss_scale_window": 1000,
-      "hysteresis": 2,
-      "min_loss_scale": 1
-    },
-    "optimizer": {
-      "type": "AdamW",
-      "params": {
-        "lr": "auto",
-        "betas": "auto",
-        "eps": "auto",
-        "weight_decay": "auto"
-      }
-    },
-    "scheduler": {
-      "type": "WarmupDecayLR",
-      "params": {
-        "warmup_min_lr": "auto",
-        "warmup_max_lr": "auto",
-        "warmup_num_steps": "auto",
-        "total_num_steps": "auto"
-      }
-    },
-    "train_batch_size": "auto",
-    "train_micro_batch_size_per_gpu": "auto",
-    "wall_clock_breakdown": false
-}

docker-compose.yaml

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

docker/Dockerfile

@@ -3,27 +3,24 @@ FROM winglian/axolotl-base:$BASE_TAG
 
 ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
 ARG AXOLOTL_EXTRAS=""
-ARG CUDA="118"
-ENV BNB_CUDA_VERSION=$CUDA
 
 RUN apt-get update && \
     apt-get install -y vim curl
 
 WORKDIR /workspace
 
+RUN pip3 install --force-reinstall "peft @ git+https://github.com/huggingface/peft.git@main" \
+            "accelerate @ git+https://github.com/huggingface/accelerate.git@main" \
+            "transformers @ git+https://github.com/huggingface/transformers.git@main"
+
 RUN git clone --depth=1 https://github.com/OpenAccess-AI-Collective/axolotl.git
 # If AXOLOTL_EXTRAS is set, append it in brackets
 RUN cd axolotl && \
     if [ "$AXOLOTL_EXTRAS" != "" ] ; then \
-        pip install -e .[flash-attn,$AXOLOTL_EXTRAS]; \
+        pip install -e .[$AXOLOTL_EXTRAS]; \
     else \
-        pip install -e .[flash-attn]; \
+        pip install -e .; \
     fi
 
-# fix so that git fetch/pull from remote works
-RUN cd axolotl && \
-    git config remote.origin.fetch "+refs/heads/*:refs/remotes/origin/*" && \
-    git config --get remote.origin.fetch
-
 # helper for huggingface-login cli
 RUN git config --global credential.helper store

docker/Dockerfile-base

@@ -8,7 +8,7 @@ FROM nvidia/cuda:$CUDA_VERSION-cudnn$CUDNN_VERSION-devel-ubuntu$UBUNTU_VERSION a
 ENV PATH="/root/miniconda3/bin:${PATH}"
 
 ARG PYTHON_VERSION="3.9"
-ARG PYTORCH_VERSION="2.0.1"
+ARG PYTORCH="2.0.0"
 ARG CUDA="118"
 
 ENV PYTHON_VERSION=$PYTHON_VERSION
@@ -29,17 +29,36 @@ ENV PATH="/root/miniconda3/envs/py${PYTHON_VERSION}/bin:${PATH}"
 WORKDIR /workspace
 
 RUN python3 -m pip install --upgrade pip && pip3 install packaging && \
-    python3 -m pip install --no-cache-dir -U torch==${PYTORCH_VERSION}+cu${CUDA} --extra-index-url https://download.pytorch.org/whl/cu$CUDA
+    python3 -m pip install --no-cache-dir -U torch==${PYTORCH} torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu$CUDA
+
+
+FROM base-builder AS flash-attn-builder
+
+WORKDIR /workspace
+
+ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
+
+RUN git clone https://github.com/HazyResearch/flash-attention.git && \
+    cd flash-attention && \
+    python3 setup.py bdist_wheel && \
+    cd csrc/fused_dense_lib && \
+    python3 setup.py bdist_wheel && \
+    cd ../xentropy && \
+    python3 setup.py bdist_wheel && \
+    cd ../rotary && \
+    python3 setup.py bdist_wheel && \
+    cd ../layer_norm && \
+    python3 setup.py bdist_wheel
 
 FROM base-builder AS deepspeed-builder
 
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
+ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6+PTX"
 
 WORKDIR /workspace
 
 RUN git clone https://github.com/microsoft/DeepSpeed.git && \
     cd DeepSpeed && \
-    MAX_CONCURRENCY=8 DS_BUILD_SPARSE_ATTN=0 DS_BUILD_OPS=1 DS_BUILD_EVOFORMER_ATTN=0 python3 setup.py bdist_wheel
+    MAX_CONCURRENCY=8 DS_BUILD_SPARSE_ATTN=0 DS_BUILD_OPS=1 python3 setup.py bdist_wheel
 
 FROM base-builder AS bnb-builder
 
@@ -54,9 +73,6 @@ RUN git clone https://github.com/TimDettmers/bitsandbytes.git && \
 
 FROM base-builder
 
-ARG TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 9.0+PTX"
-ENV TORCH_CUDA_ARCH_LIST=$TORCH_CUDA_ARCH_LIST
-
 # recompile apex
 RUN python3 -m pip uninstall -y apex
 RUN git clone https://github.com/NVIDIA/apex
@@ -70,8 +86,13 @@ RUN mkdir -p /workspace/wheels/bitsandbytes
 COPY --from=deepspeed-builder /workspace/DeepSpeed/dist/deepspeed-*.whl wheels
 COPY --from=bnb-builder /workspace/bitsandbytes/dist/bitsandbytes-*.whl wheels
 COPY --from=bnb-builder /workspace/bitsandbytes/bitsandbytes/libbitsandbytes*.so wheels/bitsandbytes
+COPY --from=flash-attn-builder /workspace/flash-attention/dist/flash_attn-*.whl wheels
+COPY --from=flash-attn-builder /workspace/flash-attention/csrc/fused_dense_lib/dist/fused_dense_lib-*.whl wheels
+COPY --from=flash-attn-builder /workspace/flash-attention/csrc/xentropy/dist/xentropy_cuda_lib-*.whl wheels
+COPY --from=flash-attn-builder /workspace/flash-attention/csrc/rotary/dist/rotary_emb-*.whl wheels
+COPY --from=flash-attn-builder /workspace/flash-attention/csrc/layer_norm/dist/dropout_layer_norm-*.whl wheels
 
-RUN pip3 install wheels/deepspeed-*.whl
+RUN pip3 install wheels/deepspeed-*.whl wheels/flash_attn-*.whl wheels/fused_dense_lib-*.whl wheels/xentropy_cuda_lib-*.whl wheels/rotary_emb-*.whl wheels/dropout_layer_norm-*.whl
 RUN cd /workspace/builds/bitsandbytes && python3 setup.py install
 RUN git lfs install --skip-repo
 RUN pip3 install awscli && \

docker/Dockerfile-runpod

@@ -1,10 +1,6 @@
 ARG BASE_TAG=main
 FROM winglian/axolotl:$BASE_TAG
 
-ENV HF_DATASETS_CACHE="/workspace/data/huggingface-cache/datasets"
-ENV HUGGINGFACE_HUB_CACHE="/workspace/data/huggingface-cache/hub"
-ENV TRANSFORMERS_CACHE="/workspace/data/huggingface-cache/hub"
-
 COPY scripts/runpod-entrypoint.sh /root/runpod-entrypoint.sh
 
 RUN apt install --yes --no-install-recommends openssh-server tmux && \

docs/nccl.md

@@ -1,46 +0,0 @@
-# NCCL
-
-NVIDIA NCCL is a library to facilitate and optimize multi-GPU communication operations, such as broadcast, all-gather, reduce, all-reduce, etc. Broadly, NCCL configuration is highly environment-specific and is configured via several [environment variables](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/env.html). A common NCCL-related problem occurs when a long-running operation times out causing the training process to abort:
-
-```text
-Watchdog caught collective operation timeout: WorkNCCL(SeqNum=42, OpType=ALLGATHER, Timeout(ms)=1800000) ran for 1806948 milliseconds before timing out.
-```
-
-Often, this timeout will happen after 30 minutes (the default setting) and is accompanied by below-average power consumption with near 100% GPU utilization before the error is raised. Nvidia recommends [disabling PCI access control services (ACS)](https://docs.nvidia.com/deeplearning/nccl/user-guide/docs/troubleshooting.html#pci-access-control-services-acs) as a possible solution if this is available to you.
-
-Forcing cross-GPU communication via [NVLink](https://en.wikipedia.org/wiki/NVLink) may help without increasing timeouts. To verify that your configuration is leveraging NVLink run the following command:
-
-```shell
-nvidia-smi nvlink --status
-```
-
-To force NCCL to use NVLink, simply set this in the environment:
-
-```shell
-export NCCL_P2P_LEVEL=NVL
-```
-
-If NVLink is not available in your environment there are other options for ``NCCL_P2P_LEVEL`` in the table below:
-
-| NCCL_P2P_LEVEL | Description |
-| -------------- | ----------- |
-| PIX | P2P data transfers through no more than a single PCIe bridge. Faster data transfer rates vs to paths involving multiple bridges, but slower compared to direct GPU-to-GPU communication. |
-| PXB | P2P data transfers through multiple PCIe bridges but not going through the PCIe Host Bridge; this path involves a complex routing process, potentially incurring a moderate level of latency. |
-| PHB | P2P data transfers occur over the PCIe and through a PCIe Host Bridge, typically involving the CPU, which can facilitate direct memory access but might introduce additional latency compared to more direct paths (ex PIX, NVL) |
-
-To validate that acceptable data transfer speeds exist for your training job, running [NCCL Tests](https://github.com/NVIDIA/nccl-tests/blob/master/README.md) can help pinpoint bottlenecks, for example:
-
-```shell
-./build/all_reduce_perf -b 8 -e 128M -f 2 -g 3
-```
-
-It can be useful when debugging NCCL communication timeouts to activate additional logging in both PyTorch and NCCL:
-
-```shell
-export NCCL_DEBUG=INFO
-export NCCL_DEBUG_SUBSYS=ALL
-export TORCH_DISTRIBUTED_DEBUG=INFO
-export TORCHELASTIC_ERROR_FILE=/PATH/TO/torcherror.log
-```
-
-Finally, if you believe your training job needs more time you can increase the timeout past 30 minutes by setting the ``ddp_timeout`` value in the Axolotl configuration. See [PyTorch init_process_group](https://pytorch.org/docs/stable/distributed.html#torch.distributed.init_process_group) for documentation on this value.

ds_config.json

@@ -35,17 +35,20 @@
     "type": "AdamW",
     "params": {
       "lr": "auto",
-      "betas": "auto",
-      "eps": "auto",
+      "betas": [
+        0.9,
+        0.999
+      ],
+      "eps": 1e-8,
       "weight_decay": "auto"
     }
   },
   "scheduler": {
-    "type": "WarmupLR",
+    "type": "OneCycle",
     "params": {
-      "warmup_min_lr": "auto",
-      "warmup_max_lr": "auto",
-      "warmup_num_steps": "auto"
+      "cycle_min_lr": 0.00001,
+      "cycle_max_lr": 0.00003,
+      "cycle_first_step_size": 120
     }
   },
   "train_batch_size": "auto",

examples/cerebras/btlm-ft.yml

@@ -1,90 +0,0 @@
-base_model: cerebras/btlm-3b-8k-base
-base_model_config: cerebras/btlm-3b-8k-base
-model_type: AutoModelForCausalLM
-tokenizer_type: GPT2Tokenizer
-trust_remote_code: true
-tokenizer_use_fast: true
-tokenizer_legacy: true
-
-load_in_8bit: false
-load_in_4bit: false
-strict: false
-push_dataset_to_hub:
-hf_use_auth_token: true
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_prepared_run
-val_set_size: 0.01
-
-adapter:
-lora_model_dir:
-sequence_len: 2048
-max_packed_sequence_len:
-sample_packing: false
-sample_packing_eff_est:
-sample_packing_seq_len_multiplier:
-total_num_tokens:
-
-lora_r:
-lora_alpha:
-lora_dropout:
-lora_target_modules:
-lora_target_linear:
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-output_dir: btlm-out
-gradient_accumulation_steps: 1
-micro_batch_size: 1
-num_epochs: 1
-optimizer: adamw_torch
-adam_beta2: 0.95
-adam_eps: 0.000000001
-max_grad_norm: 1.0
-
-torchdistx_path:
-lr_scheduler: cosine
-lr_quadratic_warmup: true
-learning_rate: 0.000085
-train_on_inputs: true
-group_by_length: false
-bf16: true
-fp16: false
-tf32: true
-
-gradient_checkpointing: false
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-
-xformers_attention:
-flash_attention: true
-sdp_attention:
-flash_optimum:
-
-gptq_groupsize:
-gptq_model_v1:
-
-warmup_steps: 32
-eval_steps:
-save_steps:
-save_total_limit:
-
-debug:
-deepspeed:
-weight_decay: 0.1
-special_tokens:
-  pad_token: "<|endoftext|>"
-fsdp:
-#  - full_shard
-#  - auto_wrap
-fsdp_config:
-#  fsdp_state_dict_type: FULL_STATE_DICT
-#  fsdp_transformer_layer_cls_to_wrap: BTLMBlock

examples/cerebras/qlora.yml

@@ -23,7 +23,6 @@ lora_target_modules:
 lora_target_linear:
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:
@@ -36,7 +35,7 @@ torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.0002
 train_on_inputs: false
-group_by_length: false
+group_by_length: true
 bf16: true
 fp16: false
 tf32: true

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

@@ -1,68 +0,0 @@
-base_model: codellama/CodeLlama-13b-hf
-base_model_config: codellama/CodeLlama-13b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

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

@@ -1,70 +0,0 @@
-base_model: codellama/CodeLlama-13b-hf
-base_model_config: codellama/CodeLlama-13b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: false
-load_in_4bit: true
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules:
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: paged_adamw_32bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

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

@@ -1,68 +0,0 @@
-base_model: codellama/CodeLlama-34b-hf
-base_model_config: codellama/CodeLlama-34b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

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

@@ -1,70 +0,0 @@
-base_model: codellama/CodeLlama-34b-hf
-base_model_config: codellama/CodeLlama-34b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: false
-load_in_4bit: true
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules:
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: paged_adamw_32bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

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

@@ -1,68 +0,0 @@
-base_model: codellama/CodeLlama-7b-hf
-base_model_config: codellama/CodeLlama-7b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

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

@@ -1,70 +0,0 @@
-base_model: codellama/CodeLlama-7b-hf
-base_model_config: codellama/CodeLlama-7b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: CodeLlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: false
-load_in_4bit: true
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules:
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: paged_adamw_32bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/code-llama/README.md

@@ -1,22 +0,0 @@
-# Overview
-
-This is an example of CodeLLaMA configuration for 7b, 13b and 34b.
-
-The 7b variant fits on any 24GB VRAM GPU and will take up about 17 GB of VRAM during training if using qlora and 20 GB if using lora. On a RTX 4090 it trains 3 epochs of the default dataset in about 15 minutes.
-
-The 13b variant will fit if you change these settings to these values:
-gradient_accumulation_steps: 2
-micro_batch_size: 1
-
-The 34b variant does not fit on 24GB of VRAM - you will need something with +40 gb VRAM that also supports flash attention v2 - A6000 or A100 are good choices.
-
-```shell
-accelerate launch scripts/finetune.py examples/code-llama/[MODEL_SIZE]/qlora.yml
-
-```
-or
-
-```shell
-accelerate launch scripts/finetune.py examples/code-llama/[MODEL_SIZE]/lora.yml
-
-```

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

@@ -24,7 +24,6 @@ lora_target_modules:
 lora_target_linear: true
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

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

@@ -38,7 +38,6 @@ lora_target_linear: true
 lora_fan_in_fan_out:
 
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/falcon/config-7b.yml

@@ -24,7 +24,6 @@ lora_target_modules:
 lora_target_linear: true
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/gptj/qlora.yml

@@ -20,7 +20,6 @@ lora_target_modules:
 lora_target_linear: true
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:
@@ -33,7 +32,7 @@ torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.0001
 train_on_inputs: false
-group_by_length: false
+group_by_length: true
 bf16: true
 fp16: false
 tf32: true

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

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

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

@@ -0,0 +1,62 @@
+base_model: Neko-Institute-of-Science/LLaMA-7B-4bit-128g
+base_model_config: Neko-Institute-of-Science/LLaMA-7B-4bit-128g
+model_type: LlamaForCausalLM
+tokenizer_type: LlamaTokenizer
+trust_remote_code:
+load_in_8bit: true
+gptq: true
+datasets:
+  - path: vicgalle/alpaca-gpt4
+    type: alpaca
+dataset_prepared_path: last_run_prepared
+val_set_size: 0.02
+adapter:
+lora_model_dir:
+sequence_len: 2048
+max_packed_sequence_len:
+lora_r: 8
+lora_alpha: 16
+lora_dropout: 0.05
+lora_target_modules:
+  - q_proj
+  - v_proj
+lora_fan_in_fan_out: false
+wandb_project: llama-7b-lora-int4
+wandb_watch:
+wandb_run_id:
+wandb_log_model:
+output_dir: ./llama-7b-lora-int4
+gradient_accumulation_steps: 1
+micro_batch_size: 1
+num_epochs: 3
+optimizer: adamw_bnb_8bit
+torchdistx_path:
+lr_scheduler: cosine
+learning_rate: 0.0000002
+train_on_inputs: false
+group_by_length: false
+fp16: true
+bf16: false
+tf32: true
+early_stopping_patience:
+resume_from_checkpoint:
+local_rank:
+logging_steps: 5
+xformers_attention:
+flash_attention:
+gradient_checkpointing: true
+gptq_groupsize: 128
+gptq_model_v1: false
+warmup_steps: 20
+eval_steps: 110
+save_steps: 660
+debug:
+deepspeed:
+weight_decay: 0.0001
+fsdp:
+fsdp_config:
+tokens:
+  pad_token: "[PAD]"
+  bos_token: "<s>"
+  eos_token: "</s>"
+  unk_token: "<unk>"

examples/jeopardy-bot/config.yml

@@ -18,7 +18,6 @@ lora_dropout:
 lora_target_modules:
 lora_fan_in_fan_out: false
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/llama-2/README.md

@@ -1,20 +0,0 @@
-# Overview
-
-This is an example of a llama-2 configuration for 7b and 13b. The yaml file contains configuration for the 7b variant, but you can just aswell use the same settings for 13b.
-
-The 7b variant fits on any 24GB VRAM GPU and will take up about 17 GB of VRAM during training if using qlora and 20 GB if using lora. On a RTX 4090 it trains 3 epochs of the default dataset in about 15 minutes.
-
-The 13b variant will fit if you change these settings to these values:
-gradient_accumulation_steps: 2
-micro_batch_size: 1
-
-```shell
-accelerate launch scripts/finetune.py examples/llama-2/qlora.yml
-
-```
-or
-
-```shell
-accelerate launch scripts/finetune.py examples/llama-2/lora.yml
-
-```

examples/llama-2/gptq-lora.yml

@@ -1,74 +0,0 @@
-base_model: TheBloke/Llama-2-7B-GPTQ
-base_model_config: TheBloke/Llama-2-7B-GPTQ
-is_llama_derived_model: false
-gptq: true
-gptq_disable_exllama: true
-model_type: AutoModelForCausalLM
-tokenizer_type: LlamaTokenizer
-tokenizer_use_fast: true
-tokenizer_legacy: true
-load_in_8bit: false
-load_in_4bit: false
-strict: false
-push_dataset_to_hub:
-hf_use_auth_token: true
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-adapter: lora
-lora_model_dir:
-sequence_len: 4096
-sample_packing:
-lora_r: 8
-lora_alpha: 32
-lora_dropout: 0.05
-lora_target_modules:
-  - k_proj
-  - o_proj
-  - q_proj
-  - v_proj
-lora_target_linear:
-lora_fan_in_fan_out:
-wandb_project:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-output_dir: ./model-out
-gradient_accumulation_steps: 1
-micro_batch_size: 1
-num_epochs: 3
-optimizer: adamw_torch
-adam_beta2: 0.95
-adam_eps: 0.00001
-max_grad_norm: 1.0
-torchdistx_path:
-lr_scheduler: cosine
-lr_quadratic_warmup: true
-learning_rate: 0.000017
-train_on_inputs: false
-group_by_length: false
-bf16: false
-fp16: false
-float16: true
-tf32: true
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention:
-sdp_attention:
-flash_optimum:
-warmup_steps: 100
-eval_steps:
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.1
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/llama-2/lora.yml

@@ -1,70 +0,0 @@
-base_model: meta-llama/Llama-2-7b-hf
-base_model_config: meta-llama/Llama-2-7b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-adapter: lora
-lora_model_dir:
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-eval_table_size: 5
-eval_table_max_new_tokens: 128
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/llama-2/qlora.yml

@@ -1,71 +0,0 @@
-base_model: meta-llama/Llama-2-7b-hf
-base_model_config: meta-llama/Llama-2-7b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: false
-load_in_4bit: true
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./qlora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 32
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules:
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: paged_adamw_32bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-eval_table_size: 5
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/llama-2/relora.yml

@@ -1,74 +0,0 @@
-base_model: meta-llama/Llama-2-7b-hf
-base_model_config: meta-llama/Llama-2-7b-hf
-model_type: LlamaForCausalLM
-tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: false
-load_in_4bit: true
-strict: false
-
-datasets:
-  - path: teknium/GPT4-LLM-Cleaned
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./relora-out
-
-adapter: qlora
-lora_model_dir:
-
-sequence_len: 4096
-sample_packing: true
-pad_to_sequence_len: true
-
-lora_r: 8
-lora_alpha: 16
-lora_dropout: 0.05
-lora_target_modules:
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-relora_steps: 150
-relora_warmup_steps: 10
-relora_cpu_offload: false
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 4
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-save_steps: 50
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/llama-2/tiny-llama.yml

@@ -1,69 +0,0 @@
-base_model: PY007/TinyLlama-1.1B-step-50K-105b
-base_model_config: PY007/TinyLlama-1.1B-step-50K-105b
-
-model_type: LlamaForCausalLM
-tokenizer_type: LlamaTokenizer
-is_llama_derived_model: true
-
-load_in_8bit: true
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: mhenrichsen/alpaca_2k_test
-    type: alpaca
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-output_dir: ./lora-out
-
-sequence_len: 4096
-sample_packing: 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_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 4
-micro_batch_size: 2
-num_epochs: 3
-optimizer: adamw_bnb_8bit
-lr_scheduler: cosine
-learning_rate: 0.0002
-
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-
-gradient_checkpointing: true
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention: true
-
-warmup_steps: 10
-eval_steps: 20
-eval_table_size: 5
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.0
-fsdp:
-fsdp_config:
-special_tokens:
-  bos_token: "<s>"
-  eos_token: "</s>"
-  unk_token: "<unk>"

examples/mpt-7b/config.yml

@@ -20,7 +20,6 @@ lora_target_modules:
   - v_proj
 lora_fan_in_fan_out: false
 wandb_project: mpt-alpaca-7b
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/openllama-3b/config.yml

@@ -1,5 +1,5 @@
-base_model: openlm-research/open_llama_3b_v2
-base_model_config: openlm-research/open_llama_3b_v2
+base_model: openlm-research/open_llama_3b
+base_model_config: openlm-research/open_llama_3b
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
 load_in_8bit: false
@@ -13,8 +13,8 @@ dataset_prepared_path: last_run_prepared
 val_set_size: 0.02
 adapter:
 lora_model_dir:
-sequence_len: 1024
-sample_packing: true
+sequence_len: 256
+max_packed_sequence_len:
 lora_r:
 lora_alpha:
 lora_dropout:
@@ -22,18 +22,17 @@ lora_target_modules:
 lora_target_linear:
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:
 output_dir: ./openllama-out
 gradient_accumulation_steps: 1
 micro_batch_size: 1
-num_epochs: 4
+num_epochs: 3
 optimizer: adamw_bnb_8bit
 torchdistx_path:
 lr_scheduler: cosine
-learning_rate: 0.000003
+learning_rate: 0.00001
 train_on_inputs: false
 group_by_length: false
 float16: true
@@ -45,12 +44,12 @@ early_stopping_patience:
 resume_from_checkpoint:
 local_rank:
 logging_steps: 1
-xformers_attention:
-flash_attention: true
+xformers_attention: true
+flash_attention:
 gptq_groupsize:
 gptq_model_v1:
-warmup_steps: 20
-eval_steps: 0.05
+warmup_steps: 10
+eval_steps: 50
 save_steps:
 debug:
 deepspeed:

examples/openllama-3b/lora.yml

@@ -1,5 +1,5 @@
-base_model: openlm-research/open_llama_3b_v2
-base_model_config: openlm-research/open_llama_3b_v2
+base_model: openlm-research/open_llama_3b
+base_model_config: openlm-research/open_llama_3b
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
 load_in_8bit: true
@@ -13,8 +13,8 @@ dataset_prepared_path: last_run_prepared
 val_set_size: 0.02
 adapter: lora
 lora_model_dir:
-sequence_len: 1024
-sample_packing: true
+sequence_len: 256
+max_packed_sequence_len:
 lora_r: 8
 lora_alpha: 16
 lora_dropout: 0.0
@@ -28,14 +28,13 @@ lora_target_modules:
   - o_proj
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:
 output_dir: ./lora-out
-gradient_accumulation_steps: 1
-micro_batch_size: 2
-num_epochs: 4
+batch_size: 16
+micro_batch_size: 4
+num_epochs: 3
 optimizer: adamw_bnb_8bit
 torchdistx_path:
 lr_scheduler: cosine
@@ -50,16 +49,16 @@ early_stopping_patience:
 resume_from_checkpoint:
 local_rank:
 logging_steps: 1
-xformers_attention:
-flash_attention: true
+xformers_attention: true
+flash_attention:
 gptq_groupsize:
 gptq_model_v1:
-warmup_steps: 20
-eval_steps: 0.05
+warmup_steps: 10
+eval_steps: 50
 save_steps:
 debug:
 deepspeed:
-weight_decay: 0.1
+weight_decay: 0.0
 fsdp:
 fsdp_config:
 special_tokens:

examples/openllama-3b/qlora.yml

@@ -1,5 +1,5 @@
-base_model: openlm-research/open_llama_3b_v2
-base_model_config: openlm-research/open_llama_3b_v2
+base_model: openlm-research/open_llama_3b
+base_model_config: openlm-research/open_llama_3b
 model_type: LlamaForCausalLM
 tokenizer_type: LlamaTokenizer
 load_in_8bit: false
@@ -13,8 +13,8 @@ dataset_prepared_path: last_run_prepared
 val_set_size: 0.01
 adapter: qlora
 lora_model_dir:
-sequence_len: 1024
-sample_packing: true
+sequence_len: 2048
+max_packed_sequence_len: 2048
 lora_r: 8
 lora_alpha: 32
 lora_dropout: 0.05
@@ -22,38 +22,37 @@ lora_target_modules:
 lora_target_linear: true
 lora_fan_in_fan_out:
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:
 output_dir: ./qlora-out
-gradient_accumulation_steps: 1
-micro_batch_size: 2
-num_epochs: 4
+batch_size: 4
+micro_batch_size: 4
+num_epochs: 2
 optimizer: paged_adamw_32bit
 torchdistx_path:
 lr_scheduler: cosine
 learning_rate: 0.0002
 train_on_inputs: false
-group_by_length: false
-bf16: false
-fp16: true
-tf32: false
+group_by_length: true
+bf16: true
+fp16: false
+tf32: true
 gradient_checkpointing: true
 early_stopping_patience:
 resume_from_checkpoint:
 local_rank:
 logging_steps: 1
-xformers_attention:
-flash_attention: true
+xformers_attention: true
+flash_attention:
 gptq_groupsize:
 gptq_model_v1:
-warmup_steps: 20
-eval_steps: 0.05
+warmup_steps: 10
+eval_steps: 20
 save_steps:
 debug:
 deepspeed:
-weight_decay: 0.1
+weight_decay: 0.0
 fsdp:
 fsdp_config:
 special_tokens:

examples/phi/README.md

@@ -1,11 +0,0 @@
-# Phi
-
-Due to some nuances with the phi code, please use deepspeed when training phi for full finetune.
-
-```shell
-accelerate launch -m axolotl.cli.train examples/phi/phi-ft.yml --deepspeed deepspeed/zero1.json
-
-# OR
-
-python -m axolotl.cli.train examples/phi/phi-qlora.yml
-```

examples/phi/phi-ft.yml

@@ -1,75 +0,0 @@
-base_model: microsoft/phi-1_5
-base_model_config: microsoft/phi-1_5
-model_type: MixFormerSequentialForCausalLM
-tokenizer_type: AutoTokenizer
-is_llama_derived_model: false
-trust_remote_code: true
-
-load_in_8bit: false
-load_in_4bit: false
-strict: false
-
-datasets:
-  - path: garage-bAInd/Open-Platypus
-    type: alpaca
-
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.05
-output_dir: ./phi-sft-out
-
-sequence_len: 2048
-sample_packing: true
-pad_to_sequence_len:
-
-adapter:
-lora_model_dir:
-lora_r:
-lora_alpha:
-lora_dropout:
-lora_target_linear:
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 1
-micro_batch_size: 1
-num_epochs: 4
-optimizer: adamw_torch
-adam_beta2: 0.95
-adam_epsilon: 0.00001
-max_grad_norm: 1.0
-lr_scheduler: cosine
-learning_rate: 0.000003
-
-train_on_inputs: false
-group_by_length: true
-bf16: true
-fp16: false
-tf32: true
-
-gradient_checkpointing:
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention:
-
-warmup_steps: 100
-eval_steps: 0.05
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.1
-fsdp:
-fsdp_config:
-resize_token_embeddings_to_32x: true
-special_tokens:
-  bos_token: "<|endoftext|>"
-  eos_token: "<|endoftext|>"
-  unk_token: "<|endoftext|>"
-  pad_token: "<|endoftext|>"

examples/phi/phi-qlora.yml

@@ -1,75 +0,0 @@
-base_model: microsoft/phi-1_5
-base_model_config: microsoft/phi-1_5
-model_type: AutoModelForCausalLM
-tokenizer_type: AutoTokenizer
-is_llama_derived_model: false
-trust_remote_code: true
-
-load_in_8bit: false
-load_in_4bit: true
-strict: false
-
-datasets:
-  - path: garage-bAInd/Open-Platypus
-    type: alpaca
-
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.05
-output_dir: ./phi-sft-out
-
-sequence_len: 1024
-sample_packing: false  # not CURRENTLY compatible with LoRAs
-pad_to_sequence_len:
-
-adapter: qlora
-lora_model_dir:
-lora_r: 64
-lora_alpha: 32
-lora_dropout: 0.05
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-
-gradient_accumulation_steps: 1
-micro_batch_size: 1
-num_epochs: 4
-optimizer: adamw_torch
-adam_beta2: 0.95
-adam_epsilon: 0.00001
-max_grad_norm: 1.0
-lr_scheduler: cosine
-learning_rate: 0.000003
-
-train_on_inputs: false
-group_by_length: true
-bf16: true
-fp16: false
-tf32: true
-
-gradient_checkpointing:
-early_stopping_patience:
-resume_from_checkpoint:
-local_rank:
-logging_steps: 1
-xformers_attention:
-flash_attention:
-
-warmup_steps: 100
-eval_steps: 0.05
-save_steps:
-debug:
-deepspeed:
-weight_decay: 0.1
-fsdp:
-fsdp_config:
-resize_token_embeddings_to_32x: true
-special_tokens:
-  bos_token: "<|endoftext|>"
-  eos_token: "<|endoftext|>"
-  unk_token: "<|endoftext|>"
-  pad_token: "<|endoftext|>"

examples/pythia-12b/config.yml

@@ -23,7 +23,6 @@ lora_target_modules:
 lora_target_linear: true
 lora_fan_in_fan_out: true  # pythia/GPTNeoX lora specific
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:
@@ -47,3 +46,4 @@ local_rank:
 gradient_checkpointing: true
 fsdp:
 fsdp_config:
+collator_pad_to_longest: true

examples/pythia/lora.yml

@@ -17,7 +17,6 @@ lora_target_modules:
 lora_target_linear:
 lora_fan_in_fan_out: true  # pythia/GPTNeoX lora specific
 wandb_project:
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/redpajama/config-3b.yml

@@ -21,7 +21,6 @@ lora_target_modules:
   - v_proj
 lora_fan_in_fan_out: false
 wandb_project: redpajama-alpaca-3b
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/replit-3b/config-lora.yml

@@ -20,7 +20,6 @@ lora_target_modules:
   - mlp_down
 lora_fan_in_fan_out:
 wandb_project: lora-replit
-wandb_entity:
 wandb_watch:
 wandb_run_id:
 wandb_log_model:

examples/xgen-7b/xgen-7b-8k-qlora.yml

@@ -1,91 +0,0 @@
-# An example finetuning Saleforce's XGen-7b model with 8k context using qlora
-# on Tim Dettmer's Guanaco dataset.
-base_model: Salesforce/xgen-7b-8k-base
-base_model_config: Salesforce/xgen-7b-8k-base
-trust_remote_code: true
-model_type: AutoModelForCausalLM
-tokenizer_type: AutoTokenizer
-load_in_8bit: false
-# enable 4bit for QLoRA
-load_in_4bit: true
-gptq: false
-strict: false
-push_dataset_to_hub:
-datasets:
-  - path: timdettmers/openassistant-guanaco
-    data_files:
-      - openassistant_best_replies_train.jsonl
-    type: "completion"
-dataset_prepared_path: last_run_prepared
-val_set_size: 0.01
-# enable QLoRA
-adapter: qlora
-lora_model_dir:
-sequence_len: 8192
-max_packed_sequence_len:
-
-# hyperparameters from QLoRA paper Appendix B.2
-# "We find hyperparameters to be largely robust across datasets"
-lora_r: 64
-lora_alpha: 16
-# 0.1 for models up to 13B
-# 0.05 for 33B and 65B models
-lora_dropout: 0.05
-# add LoRA modules on all linear layers of the base model
-lora_target_modules:
-lora_target_linear: true
-lora_fan_in_fan_out:
-
-wandb_project:
-wandb_entity:
-wandb_watch:
-wandb_run_id:
-wandb_log_model:
-output_dir: ./qlora-out
-
-# QLoRA paper Table 9
-# - 16 for 7b & 13b
-# - 32 for 33b, 64 for 64b
-# Max size tested on A6000
-# - 7b: 40
-# - 40b: 4
-# decrease if OOM, increase for max VRAM utilization
-micro_batch_size: 1
-gradient_accumulation_steps: 1
-num_epochs: 3
-# Optimizer for QLoRA
-optimizer: paged_adamw_32bit
-torchdistx_path:
-lr_scheduler: cosine
-# QLoRA paper Table 9
-# - 2e-4 for 7b & 13b
-# - 1e-4 for 33b & 64b
-learning_rate: 0.00002
-train_on_inputs: false
-group_by_length: false
-bf16: true
-fp16: false
-tf32: false
-gradient_checkpointing: true
-# stop training after this many evaluation losses have increased in a row
-# https://huggingface.co/transformers/v4.2.2/_modules/transformers/trainer_callback.html#EarlyStoppingCallback
-early_stopping_patience: 3
-resume_from_checkpoint:
-auto_resume_from_checkpoints: true
-local_rank:
-logging_steps: 1
-xformers_attention: true
-flash_attention:
-gptq_groupsize:
-gptq_model_v1:
-warmup_steps: 10
-eval_steps: 50
-save_steps: 50
-debug:
-deepspeed:
-weight_decay: 0.0
-special_tokens:
-  eos_token: "<|endoftext|>"
-  bos_token: "<|endoftext|>"
-  unk_token: "<|endoftext|>"
-  pad_token: "<|endoftext|>"

requirements.txt

@@ -1,32 +1,20 @@
---extra-index-url https://download.pytorch.org/whl/cu118
---extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/
-torch==2.0.1
-auto-gptq
-packaging
 peft @ git+https://github.com/huggingface/peft.git
 transformers @ git+https://github.com/huggingface/transformers.git
-bitsandbytes>=0.41.1
-accelerate @ git+https://github.com/huggingface/accelerate
+bitsandbytes>=0.39.0
 addict
-evaluate
 fire
-PyYAML>=6.0
+PyYAML==6.0
 datasets
-flash-attn>=2.2.1
+accelerate>=0.19.0
 sentencepiece
 wandb
 einops
 xformers
 optimum
-hf_transfer
-colorama
-numba
-numpy>=1.24.4
 # qlora things
 bert-score==0.3.13
 evaluate==0.4.0
 rouge-score==0.1.2
 scipy
 scikit-learn==1.2.2
-pynvml
-art
+numba

scripts/alpaca_json_to_jsonl.py

@@ -0,0 +1,49 @@
+"""Module to convert json file to jsonl"""
+
+import os
+import sys
+from pathlib import Path
+from typing import Optional, Union
+
+import fire
+
+from axolotl.convert import (
+    FileReader,
+    FileWriter,
+    JsonlSerializer,
+    JsonParser,
+    JsonToJsonlConverter,
+    StdoutWriter,
+)
+
+# add src to the pythonpath so we don't need to pip install this
+project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
+src_dir = os.path.join(project_root, "src")
+sys.path.insert(0, src_dir)
+
+
+def main(
+    file: Path,
+    output: Optional[Path] = None,
+    to_stdout: Optional[bool] = False,
+):
+    """
+    Convert a json file to jsonl
+    """
+
+    file_reader = FileReader()
+    writer: Union[StdoutWriter, FileWriter]
+    if to_stdout or output is None:
+        writer = StdoutWriter()
+    else:
+        writer = FileWriter(output)
+    json_parser = JsonParser()
+    jsonl_serializer = JsonlSerializer()
+
+    converter = JsonToJsonlConverter(file_reader, writer, json_parser, jsonl_serializer)
+
+    converter.convert(file, output)
+
+
+if __name__ == "__main__":
+    fire.Fire(main)

scripts/finetune.py

@@ -1,52 +1,352 @@
 """Prepare and train a model on a dataset. Can also infer from a model or merge lora"""
+
+import importlib
 import logging
+import os
+import random
+import signal
+import sys
 from pathlib import Path
+from typing import Any, Dict, List, Optional, Union
 
 import fire
-import transformers
+import torch
+import yaml
 
-from axolotl.cli import (
-    check_accelerate_default_config,
-    do_inference,
-    do_merge_lora,
-    load_cfg,
-    load_datasets,
-    print_axolotl_text_art,
-)
-from axolotl.cli.shard import shard
-from axolotl.common.cli import TrainerCliArgs
-from axolotl.train import train
+# add src to the pythonpath so we don't need to pip install this
+from optimum.bettertransformer import BetterTransformer
+from transformers import GenerationConfig, TextStreamer
 
-LOG = logging.getLogger("axolotl.scripts.finetune")
+from axolotl.utils.data import load_prepare_datasets, load_pretraining_dataset
+from axolotl.utils.dict import DictDefault
+from axolotl.utils.models import load_model, load_tokenizer
+from axolotl.utils.tokenization import check_dataset_labels
+from axolotl.utils.trainer import setup_trainer
+from axolotl.utils.validation import validate_config
+from axolotl.utils.wandb import setup_wandb_env_vars
+
+project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
+src_dir = os.path.join(project_root, "src")
+sys.path.insert(0, src_dir)
 
 
-def do_cli(config: Path = Path("examples/"), **kwargs):
-    print_axolotl_text_art()
-    LOG.warning(
-        str(
-            PendingDeprecationWarning(
-                "scripts/finetune.py will be replaced with calling axolotl.cli.train"
-            )
+logging.basicConfig(level=os.getenv("LOG_LEVEL", "INFO"))
+DEFAULT_DATASET_PREPARED_PATH = "last_run_prepared"
+
+
+def choose_device(cfg):
+    def get_device():
+        try:
+            if torch.cuda.is_available():
+                return f"cuda:{cfg.local_rank}"
+
+            if torch.backends.mps.is_available():
+                return "mps"
+
+            raise SystemError("No CUDA/mps device found")
+        except Exception:  # pylint: disable=broad-exception-caught
+            return "cpu"
+
+    cfg.device = get_device()
+    if cfg.device_map != "auto":
+        if cfg.device.startswith("cuda"):
+            cfg.device_map = {"": cfg.local_rank}
+        else:
+            cfg.device_map = {"": cfg.device}
+
+
+def get_multi_line_input() -> Optional[str]:
+    print("Give me an instruction (Ctrl + D to finish): ")
+    instruction = ""
+    for line in sys.stdin:
+        instruction += line  # pylint: disable=consider-using-join
+    # instruction = pathlib.Path("/proc/self/fd/0").read_text()
+    return instruction
+
+
+def do_inference(cfg, model, tokenizer, prompter: Optional[str]):
+    default_tokens = {"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>"}
+
+    for token, symbol in default_tokens.items():
+        # If the token isn't already specified in the config, add it
+        if not (cfg.special_tokens and token in cfg.special_tokens):
+            tokenizer.add_special_tokens({token: symbol})
+
+    prompter_module = None
+    if prompter:
+        prompter_module = getattr(
+            importlib.import_module("axolotl.prompters"), prompter
         )
-    )
-    parsed_cfg = load_cfg(config, **kwargs)
-    check_accelerate_default_config()
-    parser = transformers.HfArgumentParser((TrainerCliArgs))
-    parsed_cli_args, _ = parser.parse_args_into_dataclasses(
-        return_remaining_strings=True
-    )
-    if parsed_cli_args.inference:
-        do_inference(cfg=parsed_cfg, cli_args=parsed_cli_args)
-    elif parsed_cli_args.merge_lora:
-        do_merge_lora(cfg=parsed_cfg, cli_args=parsed_cli_args)
-    elif parsed_cli_args.shard:
-        shard(cfg=parsed_cfg, cli_args=parsed_cli_args)
-    else:
-        dataset_meta = load_datasets(cfg=parsed_cfg, cli_args=parsed_cli_args)
-        if parsed_cli_args.prepare_ds_only:
+
+    if cfg.landmark_attention:
+        from axolotl.monkeypatch.llama_landmark_attn import set_model_mem_id
+
+        set_model_mem_id(model, tokenizer)
+        model.set_mem_cache_args(
+            max_seq_len=255, mem_freq=50, top_k=5, max_cache_size=None
+        )
+
+    while True:
+        print("=" * 80)
+        # support for multiline inputs
+        instruction = get_multi_line_input()
+        if not instruction:
             return
-        train(cfg=parsed_cfg, cli_args=parsed_cli_args, dataset_meta=dataset_meta)
+        if prompter_module:
+            prompt: str = next(
+                prompter_module().build_prompt(instruction=instruction.strip("\n"))
+            )
+        else:
+            prompt = instruction.strip()
+        batch = tokenizer(prompt, return_tensors="pt", add_special_tokens=True)
+
+        print("=" * 40)
+        model.eval()
+        with torch.no_grad():
+            generation_config = GenerationConfig(
+                repetition_penalty=1.1,
+                max_new_tokens=1024,
+                temperature=0.9,
+                top_p=0.95,
+                top_k=40,
+                bos_token_id=tokenizer.bos_token_id,
+                eos_token_id=tokenizer.eos_token_id,
+                pad_token_id=tokenizer.pad_token_id,
+                do_sample=True,
+                use_cache=True,
+                return_dict_in_generate=True,
+                output_attentions=False,
+                output_hidden_states=False,
+                output_scores=False,
+            )
+            streamer = TextStreamer(tokenizer)
+            generated = model.generate(
+                inputs=batch["input_ids"].to(cfg.device),
+                generation_config=generation_config,
+                streamer=streamer,
+            )
+        print("=" * 40)
+        print(tokenizer.decode(generated["sequences"].cpu().tolist()[0]))
+
+
+def choose_config(path: Path):
+    yaml_files = list(path.glob("*.yml"))
+
+    if not yaml_files:
+        raise ValueError(
+            "No YAML config files found in the specified directory. Are you using a .yml extension?"
+        )
+
+    print("Choose a YAML file:")
+    for idx, file in enumerate(yaml_files):
+        print(f"{idx + 1}. {file}")
+
+    chosen_file = None
+    while chosen_file is None:
+        try:
+            choice = int(input("Enter the number of your choice: "))
+            if 1 <= choice <= len(yaml_files):
+                chosen_file = yaml_files[choice - 1]
+            else:
+                print("Invalid choice. Please choose a number from the list.")
+        except ValueError:
+            print("Invalid input. Please enter a number.")
+
+    return chosen_file
+
+
+def check_not_in(list1: List[str], list2: Union[Dict[str, Any], List[str]]) -> bool:
+    return not any(el in list2 for el in list1)
+
+
+def train(
+    config: Path = Path("configs/"),
+    prepare_ds_only: bool = False,
+    **kwargs,
+):
+    if Path(config).is_dir():
+        config = choose_config(config)
+
+    # load the config from the yaml file
+    with open(config, encoding="utf-8") as file:
+        cfg: DictDefault = DictDefault(yaml.safe_load(file))
+    # if there are any options passed in the cli, if it is something that seems valid from the yaml,
+    # then overwrite the value
+    cfg_keys = cfg.keys()
+    for k, _ in kwargs.items():
+        # if not strict, allow writing to cfg even if it's not in the yml already
+        if k in cfg_keys or not cfg.strict:
+            # handle booleans
+            if isinstance(cfg[k], bool):
+                cfg[k] = bool(kwargs[k])
+            else:
+                cfg[k] = kwargs[k]
+
+    validate_config(cfg)
+
+    # setup some derived config / hyperparams
+    cfg.gradient_accumulation_steps = cfg.gradient_accumulation_steps or (
+        cfg.batch_size // cfg.micro_batch_size
+    )
+    cfg.batch_size = (
+        cfg.batch_size or cfg.micro_batch_size * cfg.gradient_accumulation_steps
+    )
+    cfg.world_size = int(os.environ.get("WORLD_SIZE", 1))
+    cfg.local_rank = int(os.environ.get("LOCAL_RANK", 0))
+    choose_device(cfg)
+    cfg.ddp = cfg.ddp if cfg.ddp is not None else cfg.world_size != 1
+    if cfg.ddp:
+        cfg.device_map = {"": int(os.environ.get("LOCAL_RANK", 0))}
+        cfg.batch_size = cfg.batch_size * cfg.world_size
+
+    setup_wandb_env_vars(cfg)
+    if cfg.device == "mps":
+        cfg.load_in_8bit = False
+        cfg.tf32 = False
+        if cfg.bf16:
+            cfg.fp16 = True
+        cfg.bf16 = False
+
+    if cfg.tf32:
+        torch.backends.cuda.matmul.allow_tf32 = True
+
+    # load the tokenizer first
+    tokenizer_config = cfg.tokenizer_config or cfg.base_model_config
+    logging.info(f"loading tokenizer... {tokenizer_config}")
+    tokenizer = load_tokenizer(tokenizer_config, cfg.tokenizer_type, cfg)
+
+    if (
+        check_not_in(["shard", "merge_lora"], kwargs) and not cfg.inference
+    ):  # don't need to load dataset for these
+        if not cfg.pretraining_dataset:
+            train_dataset, eval_dataset = load_prepare_datasets(
+                tokenizer, cfg, DEFAULT_DATASET_PREPARED_PATH
+            )
+        else:
+            train_dataset = load_pretraining_dataset(
+                cfg.pretraining_dataset,
+                tokenizer,
+                max_tokens=cfg.sequence_len,
+                seed=cfg.seed,
+            )
+            # https://discuss.huggingface.co/t/how-to-use-huggingface-trainer-streaming-datasets-without-wrapping-it-with-torchdatas-iterablewrapper/25230
+            train_dataset = train_dataset.with_format("torch")
+            eval_dataset = None
+
+    if cfg.debug or "debug" in kwargs:
+        logging.info("check_dataset_labels...")
+        check_dataset_labels(
+            train_dataset.select(
+                [random.randrange(0, len(train_dataset) - 1) for _ in range(5)]  # nosec
+            ),
+            tokenizer,
+        )
+
+    if prepare_ds_only:
+        logging.info("Finished preparing dataset. Exiting...")
+        return
+
+    # Load the model and tokenizer
+    logging.info("loading model and peft_config...")
+    model, peft_config = load_model(
+        cfg.base_model,
+        cfg.base_model_config,
+        cfg.model_type,
+        tokenizer,
+        cfg,
+        adapter=cfg.adapter,
+    )
+
+    if "merge_lora" in kwargs and cfg.adapter is not None:
+        logging.info("running merge of LoRA with base model")
+        model = model.merge_and_unload()
+        model.to(dtype=torch.float16)
+
+        if cfg.local_rank == 0:
+            logging.info("saving merged model")
+            model.save_pretrained(str(Path(cfg.output_dir) / "merged"))
+        return
+
+    if cfg.inference:
+        logging.info("calling do_inference function")
+        prompter: Optional[str] = "AlpacaPrompter"
+        if "prompter" in kwargs:
+            if kwargs["prompter"] == "None":
+                prompter = None
+            else:
+                prompter = kwargs["prompter"]
+        do_inference(cfg, model, tokenizer, prompter=prompter)
+        return
+
+    if "shard" in kwargs:
+        model.save_pretrained(cfg.output_dir)
+        return
+
+    trainer = setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer)
+
+    model.config.use_cache = False
+
+    if torch.__version__ >= "2" and sys.platform != "win32":
+        logging.info("Compiling torch model")
+        model = torch.compile(model)
+
+    # go ahead and presave, so we have the adapter config available to inspect
+    if peft_config:
+        logging.info(f"Pre-saving adapter config to {cfg.output_dir}")
+        peft_config.save_pretrained(cfg.output_dir)
+
+    # In case we want to stop early with ctrl+c, this is a nice to have to save the pretrained model
+    if cfg.local_rank == 0:
+
+        def terminate_handler(_, __, model):
+            if cfg.flash_optimum:
+                model = BetterTransformer.reverse(model)
+            model.save_pretrained(cfg.output_dir)
+            sys.exit(0)
+
+        signal.signal(
+            signal.SIGINT, lambda signum, frame: terminate_handler(signum, frame, model)
+        )
+
+    logging.info("Starting trainer...")
+    if cfg.group_by_length:
+        logging.info("hang tight... sorting dataset for group_by_length")
+    resume_from_checkpoint = cfg.resume_from_checkpoint
+    if cfg.resume_from_checkpoint is None and cfg.auto_resume_from_checkpoints:
+        possible_checkpoints = [
+            str(cp) for cp in Path(cfg.output_dir).glob("checkpoint-*")
+        ]
+        if len(possible_checkpoints) > 0:
+            sorted_paths = sorted(
+                possible_checkpoints,
+                key=lambda path: int(path.split("-")[-1]),
+            )
+            resume_from_checkpoint = sorted_paths[-1]
+            logging.info(
+                f"Using Auto-resume functionality to start with checkpoint at {resume_from_checkpoint}"
+            )
+
+    if not Path(cfg.output_dir).is_dir():
+        os.makedirs(cfg.output_dir, exist_ok=True)
+    if cfg.flash_optimum:
+        with torch.backends.cuda.sdp_kernel(
+            enable_flash=True, enable_math=True, enable_mem_efficient=True
+        ):
+            trainer.train(resume_from_checkpoint=resume_from_checkpoint)
+    else:
+        trainer.train(resume_from_checkpoint=resume_from_checkpoint)
+
+    logging.info(f"Training Completed!!! Saving pre-trained model to {cfg.output_dir}")
+
+    # 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.local_rank == 0:
+        if cfg.flash_optimum:
+            model = BetterTransformer.reverse(model)
+        model.save_pretrained(cfg.output_dir)
+
+    # trainer.save_model(cfg.output_dir)  # TODO this may be needed for deepspeed to work? need to review another time
 
 
 if __name__ == "__main__":
-    fire.Fire(do_cli)
+    fire.Fire(train)

scripts/runpod-entrypoint.sh

@@ -1,21 +1,10 @@
 #!/bin/bash
 
-# Export specific ENV variables to /etc/rp_environment
-echo "Exporting environment variables..."
-printenv | grep -E '^RUNPOD_|^PATH=|^_=' | sed 's/^\(.*\)=\(.*\)$/export \1="\2"/' >> /etc/rp_environment
-echo 'source /etc/rp_environment' >> ~/.bashrc
+echo $PUBLIC_KEY >> ~/.ssh/authorized_keys
+chmod 700 -R ~/.ssh
 
-if [[ $PUBLIC_KEY ]]
-then
-    mkdir -p ~/.ssh
-    chmod 700 ~/.ssh
-    echo $PUBLIC_KEY >> ~/.ssh/authorized_keys
-    chmod 700 -R ~/.ssh
-    # Start the SSH service in the background
-    service ssh start
-else
-    echo "No PUBLIC_KEY ENV variable provided, not starting openSSH daemon"
-fi
+# Start the SSH service in the background
+service ssh start
 
 # Execute the passed arguments (CMD)
 exec "$@"

setup.py

@@ -2,40 +2,31 @@
 
 from setuptools import find_packages, setup
 
-
-def parse_requirements():
-    _install_requires = []
-    _dependency_links = []
-    with open("./requirements.txt", encoding="utf-8") as requirements_file:
-        lines = [r.strip() for r in requirements_file.readlines()]
-        for line in lines:
-            if line.startswith("--extra-index-url"):
-                # Handle custom index URLs
-                _, url = line.split()
-                _dependency_links.append(url)
-            elif "flash-attn" not in line and line and line[0] != "#":
-                # Handle standard packages
-                _install_requires.append(line)
-    return _install_requires, _dependency_links
-
-
-install_requires, dependency_links = parse_requirements()
-
+install_requires = []
+with open("./requirements.txt", encoding="utf-8") as requirements_file:
+    # don't include peft yet until we check the int4
+    # need to manually install peft for now...
+    reqs = [r.strip() for r in requirements_file.readlines() if "peft" not in r]
+    reqs = [r for r in reqs if r and r[0] != "#"]
+    for r in reqs:
+        install_requires.append(r)
 
 setup(
     name="axolotl",
-    version="0.3.0",
-    description="LLM Trainer",
-    long_description="Axolotl is a tool designed to streamline the fine-tuning of various AI models, offering support for multiple configurations and architectures.",
+    version="0.1",
+    description="You know you're going to axolotl questions",
     package_dir={"": "src"},
     packages=find_packages(),
     install_requires=install_requires,
-    dependency_links=dependency_links,
     extras_require={
-        "flash-attn": [
-            "flash-attn>=2.2.1",
+        "gptq": [
+            "alpaca_lora_4bit @ git+https://github.com/winglian/alpaca_lora_4bit.git@setup_pip",
+        ],
+        "gptq_triton": [
+            "alpaca_lora_4bit[triton] @ git+https://github.com/winglian/alpaca_lora_4bit.git@setup_pip",
         ],
         "extras": [
+            "flash-attn",
             "deepspeed",
         ],
     },

src/axolotl/cli/__init__.py

@@ -1,249 +0,0 @@
-"""Prepare and train a model on a dataset. Can also infer from a model or merge lora"""
-
-import importlib
-import logging
-import os
-import random
-import sys
-from pathlib import Path
-from typing import Any, Dict, List, Optional, Union
-
-import torch
-import yaml
-
-# add src to the pythonpath so we don't need to pip install this
-from accelerate.commands.config import config_args
-from art import text2art
-from transformers import GenerationConfig, TextStreamer
-
-from axolotl.common.cli import TrainerCliArgs, load_model_and_tokenizer
-from axolotl.logging_config import configure_logging
-from axolotl.train import TrainDatasetMeta
-from axolotl.utils.config import normalize_config, validate_config
-from axolotl.utils.data import prepare_dataset
-from axolotl.utils.dict import DictDefault
-from axolotl.utils.distributed import is_main_process
-from axolotl.utils.models import load_tokenizer
-from axolotl.utils.tokenization import check_dataset_labels
-from axolotl.utils.wandb_ import setup_wandb_env_vars
-
-project_root = os.path.abspath(os.path.join(os.path.dirname(__file__), ".."))
-src_dir = os.path.join(project_root, "src")
-sys.path.insert(0, src_dir)
-
-configure_logging()
-LOG = logging.getLogger("axolotl.scripts")
-
-os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
-
-
-def print_axolotl_text_art(suffix=None):
-    font = "nancyj"
-    ascii_text = "  axolotl"
-    if suffix:
-        ascii_text += f"  x  {suffix}"
-    ascii_art = text2art(" axolotl", font=font)
-
-    if is_main_process():
-        print(ascii_art)
-
-
-def get_multi_line_input() -> Optional[str]:
-    print("Give me an instruction (Ctrl + D to finish): ")
-    instruction = ""
-    for line in sys.stdin:
-        instruction += line  # pylint: disable=consider-using-join
-    # instruction = pathlib.Path("/proc/self/fd/0").read_text()
-    return instruction
-
-
-def do_merge_lora(
-    *,
-    cfg: DictDefault,
-    cli_args: TrainerCliArgs,
-):
-    model, tokenizer = load_model_and_tokenizer(cfg=cfg, cli_args=cli_args)
-    safe_serialization = cfg.save_safetensors is True
-
-    LOG.info("running merge of LoRA with base model")
-    model = model.merge_and_unload()
-    model.to(dtype=torch.float16)
-
-    if cfg.local_rank == 0:
-        LOG.info(f"saving merged model to: {str(Path(cfg.output_dir) / 'merged')}")
-        model.save_pretrained(
-            str(Path(cfg.output_dir) / "merged"),
-            safe_serialization=safe_serialization,
-        )
-        tokenizer.save_pretrained(str(Path(cfg.output_dir) / "merged"))
-
-
-def do_inference(
-    *,
-    cfg: DictDefault,
-    cli_args: TrainerCliArgs,
-):
-    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>"}
-
-    for token, symbol in default_tokens.items():
-        # If the token isn't already specified in the config, add it
-        if not (cfg.special_tokens and token in cfg.special_tokens):
-            tokenizer.add_special_tokens({token: symbol})
-
-    prompter_module = None
-    if prompter:
-        prompter_module = getattr(
-            importlib.import_module("axolotl.prompters"), prompter
-        )
-
-    if cfg.landmark_attention:
-        from axolotl.monkeypatch.llama_landmark_attn import set_model_mem_id
-
-        set_model_mem_id(model, tokenizer)
-        model.set_mem_cache_args(
-            max_seq_len=255, mem_freq=50, top_k=5, max_cache_size=None
-        )
-
-    model = model.to(cfg.device)
-
-    while True:
-        print("=" * 80)
-        # support for multiline inputs
-        instruction = get_multi_line_input()
-        if not instruction:
-            return
-        if prompter_module:
-            prompt: str = next(
-                prompter_module().build_prompt(instruction=instruction.strip("\n"))
-            )
-        else:
-            prompt = instruction.strip()
-        batch = tokenizer(prompt, return_tensors="pt", add_special_tokens=True)
-
-        print("=" * 40)
-        model.eval()
-        with torch.no_grad():
-            generation_config = GenerationConfig(
-                repetition_penalty=1.1,
-                max_new_tokens=1024,
-                temperature=0.9,
-                top_p=0.95,
-                top_k=40,
-                bos_token_id=tokenizer.bos_token_id,
-                eos_token_id=tokenizer.eos_token_id,
-                pad_token_id=tokenizer.pad_token_id,
-                do_sample=True,
-                use_cache=True,
-                return_dict_in_generate=True,
-                output_attentions=False,
-                output_hidden_states=False,
-                output_scores=False,
-            )
-            streamer = TextStreamer(tokenizer)
-            generated = model.generate(
-                inputs=batch["input_ids"].to(cfg.device),
-                generation_config=generation_config,
-                streamer=streamer,
-            )
-        print("=" * 40)
-        print(tokenizer.decode(generated["sequences"].cpu().tolist()[0]))
-
-
-def choose_config(path: Path):
-    yaml_files = list(path.glob("*.yml"))
-
-    if not yaml_files:
-        raise ValueError(
-            "No YAML config files found in the specified directory. Are you using a .yml extension?"
-        )
-
-    if len(yaml_files) == 1:
-        print(f"Using default YAML file '{yaml_files[0]}'")
-        return yaml_files[0]
-
-    print("Choose a YAML file:")
-    for idx, file in enumerate(yaml_files):
-        print(f"{idx + 1}. {file}")
-
-    chosen_file = None
-    while chosen_file is None:
-        try:
-            choice = int(input("Enter the number of your choice: "))
-            if 1 <= choice <= len(yaml_files):
-                chosen_file = yaml_files[choice - 1]
-            else:
-                print("Invalid choice. Please choose a number from the list.")
-        except ValueError:
-            print("Invalid input. Please enter a number.")
-
-    return chosen_file
-
-
-def check_not_in(list1: List[str], list2: Union[Dict[str, Any], List[str]]) -> bool:
-    return not any(el in list2 for el in list1)
-
-
-def load_cfg(config: Path = Path("examples/"), **kwargs):
-    if Path(config).is_dir():
-        config = choose_config(config)
-
-    # load the config from the yaml file
-    with open(config, encoding="utf-8") as file:
-        cfg: DictDefault = DictDefault(yaml.safe_load(file))
-    # if there are any options passed in the cli, if it is something that seems valid from the yaml,
-    # then overwrite the value
-    cfg_keys = cfg.keys()
-    for k, _ in kwargs.items():
-        # if not strict, allow writing to cfg even if it's not in the yml already
-        if k in cfg_keys or not cfg.strict:
-            # handle booleans
-            if isinstance(cfg[k], bool):
-                cfg[k] = bool(kwargs[k])
-            else:
-                cfg[k] = kwargs[k]
-
-    validate_config(cfg)
-
-    normalize_config(cfg)
-
-    setup_wandb_env_vars(cfg)
-    return cfg
-
-
-def load_datasets(
-    *,
-    cfg: DictDefault,
-    cli_args: TrainerCliArgs,
-) -> TrainDatasetMeta:
-    tokenizer = load_tokenizer(cfg)
-
-    train_dataset, eval_dataset, total_num_steps = prepare_dataset(cfg, tokenizer)
-
-    if cli_args.debug or cfg.debug:
-        LOG.info("check_dataset_labels...")
-        check_dataset_labels(
-            train_dataset.select(
-                [
-                    random.randrange(0, len(train_dataset) - 1)  # nosec
-                    for _ in range(cli_args.debug_num_examples)
-                ]
-            ),
-            tokenizer,
-            num_examples=cli_args.debug_num_examples,
-            text_only=cli_args.debug_text_only,
-        )
-
-    return TrainDatasetMeta(
-        train_dataset=train_dataset,
-        eval_dataset=eval_dataset,
-        total_num_steps=total_num_steps,
-    )
-
-
-def check_accelerate_default_config():
-    if Path(config_args.default_yaml_config_file).exists():
-        LOG.warning(
-            f"accelerate config file found at {config_args.default_yaml_config_file}. This can lead to unexpected errors"
-        )

src/axolotl/cli/inference.py

@@ -1,27 +0,0 @@
-"""
-CLI to run inference on a trained model
-"""
-from pathlib import Path
-
-import fire
-import transformers
-
-from axolotl.cli import do_inference, load_cfg, print_axolotl_text_art
-from axolotl.common.cli import TrainerCliArgs
-
-
-def do_cli(config: Path = Path("examples/"), **kwargs):
-    # pylint: disable=duplicate-code
-    print_axolotl_text_art()
-    parsed_cfg = load_cfg(config, **kwargs)
-    parser = transformers.HfArgumentParser((TrainerCliArgs))
-    parsed_cli_args, _ = parser.parse_args_into_dataclasses(
-        return_remaining_strings=True
-    )
-    parsed_cli_args.inference = True
-
-    do_inference(cfg=parsed_cfg, cli_args=parsed_cli_args)
-
-
-if __name__ == "__main__":
-    fire.Fire(do_cli)

src/axolotl/cli/merge_lora.py

@@ -1,27 +0,0 @@
-"""
-CLI to run merge a trained LoRA into a base model
-"""
-from pathlib import Path
-
-import fire
-import transformers
-
-from axolotl.cli import do_merge_lora, load_cfg, print_axolotl_text_art
-from axolotl.common.cli import TrainerCliArgs
-
-
-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, merge_lora=True, **kwargs)
-
-    do_merge_lora(cfg=parsed_cfg, cli_args=parsed_cli_args)
-
-
-if __name__ == "__main__":
-    fire.Fire(do_cli)

src/axolotl/cli/shard.py

@@ -1,42 +0,0 @@
-"""
-CLI to shard a trained model into 10GiB chunks
-"""
-import logging
-from pathlib import Path
-
-import fire
-import transformers
-
-from axolotl.cli import load_cfg, print_axolotl_text_art
-from axolotl.common.cli import TrainerCliArgs, load_model_and_tokenizer
-from axolotl.utils.dict import DictDefault
-
-LOG = logging.getLogger("axolotl.scripts")
-
-
-def shard(
-    *,
-    cfg: DictDefault,
-    cli_args: TrainerCliArgs,
-):
-    model, _ = load_model_and_tokenizer(cfg=cfg, cli_args=cli_args)
-    safe_serialization = cfg.save_safetensors is True
-    LOG.debug("Re-saving model w/ sharding")
-    model.save_pretrained(cfg.output_dir, safe_serialization=safe_serialization)
-
-
-def do_cli(config: Path = Path("examples/"), **kwargs):
-    # pylint: disable=duplicate-code
-    print_axolotl_text_art()
-    parsed_cfg = load_cfg(config, **kwargs)
-    parser = transformers.HfArgumentParser((TrainerCliArgs))
-    parsed_cli_args, _ = parser.parse_args_into_dataclasses(
-        return_remaining_strings=True
-    )
-    parsed_cli_args.shard = True
-
-    shard(cfg=parsed_cfg, cli_args=parsed_cli_args)
-
-
-if __name__ == "__main__":
-    fire.Fire(do_cli)

src/axolotl/cli/train.py

@@ -1,36 +0,0 @@
-"""
-CLI to run training on a model
-"""
-from pathlib import Path
-
-import fire
-import transformers
-
-from axolotl.cli import (
-    check_accelerate_default_config,
-    load_cfg,
-    load_datasets,
-    print_axolotl_text_art,
-)
-from axolotl.common.cli import TrainerCliArgs
-from axolotl.train import train
-
-
-def do_cli(config: Path = Path("examples/"), **kwargs):
-    # pylint: disable=duplicate-code
-    print_axolotl_text_art()
-    parsed_cfg = load_cfg(config, **kwargs)
-    check_accelerate_default_config()
-    parser = transformers.HfArgumentParser((TrainerCliArgs))
-    parsed_cli_args, _ = parser.parse_args_into_dataclasses(
-        return_remaining_strings=True
-    )
-
-    dataset_meta = load_datasets(cfg=parsed_cfg, cli_args=parsed_cli_args)
-    if parsed_cli_args.prepare_ds_only:
-        return
-    train(cfg=parsed_cfg, cli_args=parsed_cli_args, dataset_meta=dataset_meta)
-
-
-if __name__ == "__main__":
-    fire.Fire(do_cli)

src/axolotl/common/cli.py

@@ -1,43 +0,0 @@
-"""
-shared module for cli specific things
-"""
-
-import logging
-from dataclasses import dataclass, field
-from typing import Optional
-
-from axolotl.logging_config import configure_logging
-from axolotl.utils.dict import DictDefault
-from axolotl.utils.models import load_model, load_tokenizer
-
-configure_logging()
-LOG = logging.getLogger("axolotl.common.cli")
-
-
-@dataclass
-class TrainerCliArgs:
-    """
-    dataclass representing the various non-training arguments
-    """
-
-    debug: bool = field(default=False)
-    debug_text_only: bool = field(default=False)
-    debug_num_examples: int = field(default=5)
-    inference: bool = field(default=False)
-    merge_lora: bool = field(default=False)
-    prepare_ds_only: bool = field(default=False)
-    prompter: Optional[str] = field(default=None)
-    shard: bool = field(default=False)
-
-
-def load_model_and_tokenizer(
-    *,
-    cfg: DictDefault,
-    cli_args: TrainerCliArgs,
-):
-    LOG.info(f"loading tokenizer... {cfg.tokenizer_config or cfg.base_model_config}")
-    tokenizer = load_tokenizer(cfg)
-    LOG.info("loading model and (optionally) peft_config...")
-    model, _ = load_model(cfg, tokenizer, inference=cli_args.inference)
-
-    return model, tokenizer

src/axolotl/datasets.py

@@ -1,13 +1,12 @@
 """Module containing Dataset functionality"""
 
 import logging
-import os
 from typing import List
 
 import torch
-from datasets import Dataset, IterableDataset
+from datasets import IterableDataset
 
-from .prompt_tokenizers import PromptTokenizingStrategy
+from .prompt_tokenizers import InvalidDataException, PromptTokenizingStrategy
 
 # We want this to be a wrapper for an existing dataset that we have loaded
 # lets use the concept of middlewares to wrap each dataset, for example
@@ -15,12 +14,10 @@ from .prompt_tokenizers import PromptTokenizingStrategy
 # let's check to ensure we don't truncate an item in the middle, we'll use
 # the collators later on to pad the datasets
 
-LOG = logging.getLogger("axolotl")
 
-
-class TokenizedPromptDataset(Dataset):
+class TokenizedPromptDataset(IterableDataset):
     """
-    Dataset that returns tokenized prompts from a stream of text files.
+    Iterable dataset that returns tokenized prompts from a stream of text files.
         Args:
             prompt_tokenizer (PromptTokenizingStrategy): The prompt tokenizing method for proccessing the data.
             dataset (dataset.Dataset): Dataset with text files.
@@ -30,19 +27,22 @@ class TokenizedPromptDataset(Dataset):
         self,
         prompt_tokenizer: PromptTokenizingStrategy,
         dataset: IterableDataset,
-        **kwargs,
     ):
         self.prompt_tokenizer = prompt_tokenizer
-        super().__init__(self.process(dataset).data, **kwargs)
+        self.dataset = dataset
 
-    def process(self, dataset):
-        features = dataset.features.keys()
-        num_proc = min(64, os.cpu_count())
-        return dataset.map(
-            self.prompt_tokenizer.tokenize_prompt,
-            num_proc=num_proc,
-            remove_columns=features,
-        )
+    def __iter__(self):
+        iterator = iter(self.dataset)
+        count = 0
+        # Loop through the entire dataset
+        for example in iterator:
+            try:
+                yield self.prompt_tokenizer.tokenize_prompt(example)
+                count += 1
+            except InvalidDataException:
+                pass
+        if count == 0:
+            raise RuntimeError("Expected at least one datapoint in dataset.")
 
 
 # TODO this isn't the best since it can't interleave datasets
@@ -76,21 +76,14 @@ class ConstantLengthDataset(IterableDataset):
             self.tokens_dtype = torch.int64
 
     def __iter__(self):
-        buffer = {
-            "input_ids": [],
-            "attention_mask": [],
-            "labels": [],
-            "position_ids": [],
-        }
+        buffer = {"input_ids": [], "attention_mask": [], "labels": []}
         buffer_len = 0
         for dataset in self.datasets:
-            idx = 0
             iterator = iter(dataset)
             more_examples = True
             while more_examples:
                 try:
                     example = next(iterator)
-                    idx += 1
                 except StopIteration:
                     more_examples = False
                     example = None
@@ -112,9 +105,6 @@ class ConstantLengthDataset(IterableDataset):
                         attention_mask = torch.cat(buffer["attention_mask"], dim=-1)[
                             : self.seq_length
                         ]
-                        position_ids = torch.cat(buffer["position_ids"], dim=-1)[
-                            : self.seq_length
-                        ]
                         labels = torch.cat(buffer["labels"], dim=-1)[: self.seq_length]
                         if labels.size() == input_ids.size() and (
                             attention_mask.size() == input_ids.size()
@@ -123,20 +113,17 @@ class ConstantLengthDataset(IterableDataset):
                                 "input_ids": input_ids,
                                 "labels": labels,
                                 "attention_mask": attention_mask,
-                                "position_ids": position_ids,
                             }
                         else:
-                            LOG.warning(
+                            logging.warning(
                                 f"dropping batch due to tensor size mismatch input_ids: {input_ids.size()}, labels: {labels.size()}, attention_mask: {attention_mask.size()}"
                             )
                     buffer = {
                         "input_ids": [],
                         "attention_mask": [],
                         "labels": [],
-                        "position_ids": [],
                     }
                     buffer_len = 0
-                    idx = 1
 
                 if example:
                     # FIXME
@@ -145,6 +132,11 @@ class ConstantLengthDataset(IterableDataset):
                         input_ids = example["input_ids"]
                         attention_mask = example["attention_mask"]
                         labels = example["labels"]
+                        if (
+                            buffer["input_ids"]
+                            and input_ids[0] == self.tokenizer.bos_token_id
+                        ):
+                            attention_mask[0] = 0
 
                         if add_concat_token:
                             input_ids.append(self.concat_token_id)
@@ -155,17 +147,13 @@ class ConstantLengthDataset(IterableDataset):
                             input_ids, dtype=self.tokens_dtype
                         )
                         attention_mask_with_concat = torch.tensor(
-                            [idx * m for m in attention_mask], dtype=torch.int16
+                            attention_mask, dtype=self.tokens_dtype
                         )
                         labels_with_concat = torch.tensor(
                             labels, dtype=self.tokens_dtype
                         )
-                        position_ids = torch.arange(
-                            len(input_ids), dtype=self.tokens_dtype
-                        )
 
                         buffer["input_ids"].append(input_ids_with_concat)
                         buffer["attention_mask"].append(attention_mask_with_concat)
                         buffer["labels"].append(labels_with_concat)
-                        buffer["position_ids"].append(position_ids)
                         buffer_len += len(input_ids)

src/axolotl/flash_attn.py

@@ -0,0 +1,140 @@
+"""Flash attention monkey patch for llama model"""
+
+# copied from https://github.com/lm-sys/FastChat/blob/main/fastchat/train/llama_flash_attn_monkey_patch.py
+
+from typing import Optional, Tuple
+
+import torch
+import transformers
+from einops import rearrange
+from flash_attn.bert_padding import pad_input, unpad_input
+from flash_attn.flash_attn_interface import flash_attn_unpadded_qkvpacked_func
+from transformers.models.llama.modeling_llama import apply_rotary_pos_emb
+
+
+def forward(
+    self,
+    hidden_states: torch.Tensor,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.Tensor] = None,
+    past_key_value: Optional[Tuple[torch.Tensor]] = None,
+    output_attentions: bool = False,
+    use_cache: bool = False,
+) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+    """Input shape: Batch x Time x Channel
+
+    attention_mask: [bsz, q_len]
+    """
+    # pylint: disable=duplicate-code
+    bsz, q_len, _ = hidden_states.size()
+
+    query_states = (
+        self.q_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    key_states = (
+        self.k_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    value_states = (
+        self.v_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    # [bsz, q_len, nh, hd]
+    # [bsz, nh, q_len, hd]
+
+    kv_seq_len = key_states.shape[-2]
+    assert past_key_value is None, "past_key_value is not supported"
+
+    cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+    query_states, key_states = apply_rotary_pos_emb(
+        query_states, key_states, cos, sin, position_ids
+    )
+    # [bsz, nh, t, hd]
+    assert not output_attentions, "output_attentions is not supported"
+    assert not use_cache, "use_cache is not supported"
+
+    # Flash attention codes from
+    # https://github.com/HazyResearch/flash-attention/blob/main/flash_attn/flash_attention.py
+
+    # transform the data into the format required by flash attention
+    qkv = torch.stack(
+        [query_states, key_states, value_states], dim=2
+    )  # [bsz, nh, 3, q_len, hd]
+    qkv = qkv.transpose(1, 3)  # [bsz, q_len, 3, nh, hd]
+    # We have disabled _prepare_decoder_attention_mask in LlamaModel
+    # the attention_mask should be the same as the key_padding_mask
+    key_padding_mask = attention_mask
+
+    if key_padding_mask is None:
+        qkv = rearrange(qkv, "b s ... -> (b s) ...")
+        max_s = q_len
+        cu_q_lens = torch.arange(
+            0,
+            (bsz + 1) * q_len,
+            step=q_len,
+            dtype=torch.int32,
+            device=qkv.device,
+        )
+        output = flash_attn_unpadded_qkvpacked_func(
+            qkv, cu_q_lens, max_s, 0.0, softmax_scale=None, causal=True
+        )
+        output = rearrange(output, "(b s) ... -> b s ...", b=bsz)
+    else:
+        nheads = qkv.shape[-2]
+
+        # pylint: disable=invalid-name
+        x = rearrange(qkv, "b s three h d -> b s (three h d)")
+        x_unpad, indices, cu_q_lens, max_s = unpad_input(x, key_padding_mask)
+        x_unpad = rearrange(
+            x_unpad,
+            "nnz (three h d) -> nnz three h d",
+            three=3,
+            h=nheads,
+        )
+        output_unpad = flash_attn_unpadded_qkvpacked_func(
+            x_unpad,
+            cu_q_lens,
+            max_s,
+            0.0,
+            softmax_scale=None,
+            causal=True,
+        )
+        output = rearrange(
+            pad_input(
+                rearrange(output_unpad, "nnz h d -> nnz (h d)"),
+                indices,
+                bsz,
+                q_len,
+            ),
+            "b s (h d) -> b s h d",
+            h=nheads,
+        )
+    return (
+        self.o_proj(rearrange(output, "b s h d -> b s (h d)")),
+        None,
+        None,
+    )
+
+
+# Disable the transformation of the attention mask in LlamaModel as the flash attention
+# requires the attention mask to be the same as the key_padding_mask
+def _prepare_decoder_attention_mask(
+    self,
+    attention_mask,
+    input_shape,
+    inputs_embeds,
+    past_key_values_length,
+):  # pylint: disable=unused-argument
+    # [bsz, seq_len]
+    return attention_mask
+
+
+def replace_llama_attn_with_flash_attn():
+    transformers.models.llama.modeling_llama.LlamaModel._prepare_decoder_attention_mask = (  # pylint: disable=protected-access
+        _prepare_decoder_attention_mask
+    )
+    transformers.models.llama.modeling_llama.LlamaAttention.forward = forward

src/axolotl/logging_config.py

@@ -1,71 +0,0 @@
-"""
-Common logging module for axolotl
-"""
-
-import os
-import sys
-from logging import Formatter
-from logging.config import dictConfig
-from typing import Any, Dict
-
-from colorama import Fore, Style, init
-
-
-class ColorfulFormatter(Formatter):
-    """
-    Formatter to add coloring to log messages by log type
-    """
-
-    COLORS = {
-        "WARNING": Fore.YELLOW,
-        "ERROR": Fore.RED,
-        "CRITICAL": Fore.RED + Style.BRIGHT,
-    }
-
-    def format(self, record):
-        record.rank = int(os.getenv("LOCAL_RANK", "0"))
-        log_message = super().format(record)
-        return self.COLORS.get(record.levelname, "") + log_message + Fore.RESET
-
-
-DEFAULT_LOGGING_CONFIG: Dict[str, Any] = {
-    "version": 1,
-    "formatters": {
-        "simple": {
-            "format": "[%(asctime)s] [%(levelname)s] [%(name)s.%(funcName)s:%(lineno)d] [PID:%(process)d] %(message)s",
-        },
-        "colorful": {
-            "()": ColorfulFormatter,
-            "format": "[%(asctime)s] [%(levelname)s] [%(name)s.%(funcName)s:%(lineno)d] [PID:%(process)d] [RANK:%(rank)d] %(message)s",
-        },
-    },
-    "filters": {},
-    "handlers": {
-        "console": {
-            "class": "logging.StreamHandler",
-            "formatter": "simple",
-            "filters": [],
-            "stream": sys.stdout,
-        },
-        "color_console": {
-            "class": "logging.StreamHandler",
-            "formatter": "colorful",
-            "filters": [],
-            "stream": sys.stdout,
-        },
-    },
-    "root": {"handlers": ["console"], "level": os.getenv("LOG_LEVEL", "INFO")},
-    "loggers": {
-        "axolotl": {
-            "handlers": ["color_console"],
-            "level": "DEBUG",
-            "propagate": False,
-        },
-    },
-}
-
-
-def configure_logging():
-    """Configure with default logging"""
-    init()  # Initialize colorama
-    dictConfig(DEFAULT_LOGGING_CONFIG)

src/axolotl/models/phi/__init__.py

@@ -1,6 +0,0 @@
-"""
-MixFormers model architecture used for phi models
-"""
-
-from .configuration_mixformer_sequential import MixFormerSequentialConfig  # noqa
-from .modeling_mixformer_sequential import MixFormerSequentialForCausalLM  # noqa

src/axolotl/models/phi/configuration_mixformer_sequential.py

@@ -1,63 +0,0 @@
-# pylint: skip-file
-
-# Copyright (c) Microsoft Corporation.
-# Licensed under the MIT license.
-
-import math
-from typing import Any, Dict, List, Optional, Union
-
-from transformers import PretrainedConfig
-
-
-class MixFormerSequentialConfig(PretrainedConfig):
-    """MixFormer (sequential for DeepSpeed) configuration."""
-
-    model_type = "mixformer-sequential"
-
-    attribute_map = {
-        "max_position_embeddings": "n_positions",
-        "hidden_size": "n_embd",
-        "num_attention_heads": "n_head",
-        "num_hidden_layers": "n_layer",
-        "input_emb_layer": "embd_layer",  # `input_emb_layer` key is for backward compatibility
-        "blocks": "architecture",  # `blocks` key is for backward compatibility
-    }
-
-    def __init__(
-        self,
-        vocab_size: Optional[int] = 50304,
-        n_positions: Optional[int] = 2048,
-        n_embd: Optional[int] = 1024,
-        n_layer: Optional[int] = 20,
-        n_inner: Optional[int] = None,
-        n_head: Optional[int] = 16,
-        rotary_dim: Optional[int] = 32,
-        activation_function: Optional[str] = "gelu_new",
-        embd_layer: Optional[str] = "default",
-        architecture: Union[Dict[str, Any], List[Dict[str, Any]]] = None,
-        embd_pdrop: Optional[float] = 0.0,
-        resid_pdrop: Optional[float] = 0.0,
-        layer_norm_epsilon: Optional[float] = 1e-5,
-        initializer_range: Optional[float] = 0.02,
-        tie_word_embeddings: Optional[bool] = False,
-        pad_vocab_size_multiple: Optional[int] = 64,
-        **kwargs
-    ) -> None:
-        self.vocab_size = int(
-            math.ceil(vocab_size / pad_vocab_size_multiple) * pad_vocab_size_multiple
-        )
-        self.n_positions = n_positions
-        self.n_embd = n_embd
-        self.n_layer = n_layer
-        self.n_inner = n_inner
-        self.n_head = n_head
-        self.rotary_dim = min(rotary_dim, n_embd // n_head)
-        self.activation_function = activation_function
-        self.embd_layer = embd_layer
-        self.architecture = architecture
-        self.embd_pdrop = embd_pdrop
-        self.resid_pdrop = resid_pdrop
-        self.layer_norm_epsilon = layer_norm_epsilon
-        self.initializer_range = initializer_range
-
-        super().__init__(tie_word_embeddings=tie_word_embeddings, **kwargs)

src/axolotl/models/phi/modeling_mixformer_sequential.py

@@ -1,934 +0,0 @@
-# pylint: skip-file
-
-# Copyright (c) Microsoft Corporation.
-# Licensed under the MIT license.
-
-# BSD 3-Clause License
-#
-# Copyright (c) 2022, Tri Dao, trid@cs.stanford.edu.
-# All rights reserved.
-#
-# Redistribution and use in source and binary forms, with or without
-# modification, are permitted provided that the following conditions are met:
-#
-# * Redistributions of source code must retain the above copyright notice, this
-#   list of conditions and the following disclaimer.
-#
-# * Redistributions in binary form must reproduce the above copyright notice,
-#   this list of conditions and the following disclaimer in the documentation
-#   and/or other materials provided with the distribution.
-#
-# * Neither the name of the copyright holder nor the names of its
-#   contributors may be used to endorse or promote products derived from
-#   this software without specific prior written permission.
-#
-# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
-# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
-# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
-# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
-# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
-# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
-# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
-# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
-# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-
-from __future__ import annotations
-
-import copy
-import inspect
-from dataclasses import dataclass, field
-from typing import Any, Dict, Optional, Tuple
-
-import torch
-import torch.nn as nn
-from einops import rearrange
-from flash_attn.flash_attn_interface import (
-    flash_attn_kvpacked_func,
-    flash_attn_qkvpacked_func,
-    flash_attn_varlen_qkvpacked_func,
-)
-from transformers import PretrainedConfig, PreTrainedModel
-from transformers.activations import ACT2FN
-from transformers.modeling_outputs import CausalLMOutputWithPast
-
-from ...monkeypatch.utils import get_cu_seqlens_from_pos_ids
-from .configuration_mixformer_sequential import MixFormerSequentialConfig
-
-
-@dataclass
-class InferenceParams:
-    """Inference parameters that are passed to the main model in order
-    to efficienly calculate and store the context during inference.
-    Adapted from https://github.com/Dao-AILab/flash-attention."""
-
-    max_sequence_len: int
-    max_batch_size: int
-    sequence_len_offset: int = 0
-    batch_size_offset: int = 0
-    key_value_memory_dict: dict = field(default_factory=dict)
-    fused_ft_kernel: bool = False
-    lengths_per_sample: Optional[torch.Tensor] = None
-
-
-class Embedding(nn.Module):
-    """Token embedding with dropout."""
-
-    def __init__(self, config: PretrainedConfig) -> None:
-        super().__init__()
-
-        self.wte = nn.Embedding(config.vocab_size, config.n_embd)
-        self.drop = nn.Dropout(config.embd_pdrop)
-
-    def forward(self, input_ids: torch.LongTensor) -> torch.FloatTensor:
-        input_shape = input_ids.size()
-        input_ids = input_ids.view(-1, input_shape[-1])
-
-        hidden_states = self.wte(input_ids)
-        hidden_states = self.drop(hidden_states)
-
-        return hidden_states
-
-
-class RotaryEmbedding(nn.Module):
-    """PyTorch implementation of `flash-attn` RotaryEmbedding layer.
-    Adapted from https://github.com/Dao-AILab/flash-attention."""
-
-    def __init__(
-        self,
-        dim: int,
-        base: Optional[int] = 10000,
-        scale_base: Optional[float] = None,
-        device: Optional[str] = None,
-        **kwargs,
-    ) -> None:
-        super().__init__()
-
-        if scale_base is not None:
-            raise NotImplementedError
-
-        # Generate and save the inverse frequency buffer (non-trainable)
-        self.dim = dim
-        self.base = base
-        self.scale_base = scale_base
-        self.device = device
-
-        inv_freq = 1.0 / (
-            base ** (torch.arange(0, dim, 2, device=device, dtype=torch.float32) / dim)
-        )
-        self.register_buffer("inv_freq", inv_freq)
-
-        scale = (
-            (torch.arange(0, dim, 2, device=device, dtype=torch.float32) + 0.4 * dim)
-            / (1.4 * dim)
-            if scale_base is not None
-            else None
-        )
-        self.register_buffer("scale", scale)
-
-        self._seq_len_cached = 0
-        self._cos_cached = None
-        self._sin_cached = None
-        self._cos_k_cached = None
-        self._sin_k_cached = None
-
-    def _update_cos_sin_cache(
-        self, x: torch.FloatTensor, seqlen_offset: Optional[int] = 0
-    ) -> None:
-        # Reset the tables if the sequence length has changed,
-        # or if we're on a new device (possibly due to tracing for instance)
-        seqlen = x.shape[1] + seqlen_offset
-
-        # Re-generate the inverse frequency buffer if it's not fp32
-        # (for instance if model.half() was called)
-        if self.inv_freq.dtype != "torch.float32":
-            self.inv_freq = 1.0 / (
-                self.base
-                ** (
-                    torch.arange(
-                        0, self.dim, 2, device=self.device, dtype=torch.float32
-                    )
-                    / self.dim
-                )
-            )
-
-        if (
-            seqlen > self._seq_len_cached
-            or self._cos_cached.device != x.device
-            or self._cos_cached.dtype != x.dtype
-        ):
-            self._seq_len_cached = seqlen
-            t = torch.arange(seqlen, device=x.device, dtype=torch.float32)
-
-            # Don't do einsum, it converts fp32 to fp16
-            # freqs = torch.einsum("i,j->ij", t, self.inv_freq)
-            freqs = torch.outer(
-                t, self.inv_freq.to(device=t.device, dtype=torch.float32)
-            )
-            if self.scale is None:
-                self._cos_cached = torch.cos(freqs).to(x.dtype)
-                self._sin_cached = torch.sin(freqs).to(x.dtype)
-            else:
-                power = (
-                    torch.arange(
-                        seqlen, dtype=self.scale.dtype, device=self.scale.device
-                    )
-                    - seqlen // 2
-                ) / self.scale_base
-                scale = self.scale.to(device=power.device) ** rearrange(
-                    power, "s -> s 1"
-                )
-
-                # We want the multiplication by scale to happen in fp32
-                self._cos_cached = (torch.cos(freqs) * scale).to(x.dtype)
-                self._sin_cached = (torch.sin(freqs) * scale).to(x.dtype)
-                self._cos_k_cached = (torch.cos(freqs) / scale).to(x.dtype)
-                self._sin_k_cached = (torch.sin(freqs) / scale).to(x.dtype)
-
-    def apply_rotary_emb_qkv(
-        self,
-        qkv: torch.FloatTensor,
-        sin: torch.FloatTensor,
-        cos: torch.FloatTensor,
-        sin_k: Optional[torch.FloatTensor] = None,
-        cos_k: Optional[torch.FloatTensor] = None,
-    ) -> torch.FloatTensor:
-        _, seqlen, three, _, headdim = qkv.shape
-        assert three == 3
-
-        rotary_seqlen, rotary_dim = cos.shape
-        rotary_dim *= 2
-        assert rotary_dim <= headdim
-        assert seqlen <= rotary_seqlen
-
-        cos_k = cos if cos_k is None else cos_k
-        sin_k = sin if sin_k is None else sin_k
-        assert (
-            sin.shape == cos_k.shape == sin_k.shape == (rotary_seqlen, rotary_dim // 2)
-        )
-
-        q_rot = qkv[:, :, 0, :, :rotary_dim]
-        q_pass = qkv[:, :, 0, :, rotary_dim:]
-
-        k_rot = qkv[:, :, 1, :, :rotary_dim]
-        k_pass = qkv[:, :, 1, :, rotary_dim:]
-
-        # Splits the queries and keys in half
-        q1, q2 = q_rot.chunk(2, dim=-1)
-        k1, k2 = k_rot.chunk(2, dim=-1)
-        c, s = rearrange(cos[:seqlen], "s d -> s 1 d"), rearrange(
-            sin[:seqlen], "s d -> s 1 d"
-        )
-
-        # Casts to fp32 are necessary to prevent fp16 overflow issues
-        q1, q2, k1, k2, c, s = [
-            t.to(dtype=torch.float32) for t in [q1, q2, k1, k2, c, s]
-        ]
-
-        # Computes the new keys and queries, recasting to original dtype
-        q_rot = torch.cat([q1 * c - q2 * s, q1 * s + q2 * c], axis=-1).to(qkv.dtype)
-
-        k_rot = torch.cat([k1 * c - k2 * s, k1 * s + k2 * c], axis=-1).to(qkv.dtype)
-
-        return torch.cat(
-            [
-                torch.cat([q_rot, q_pass], axis=-1).unsqueeze(2),
-                torch.cat([k_rot, k_pass], axis=-1).unsqueeze(2),
-                qkv[:, :, 2:3, :, :],
-            ],
-            axis=2,
-        )
-
-    def forward(
-        self, qkv: torch.Tensor, seqlen_offset: int = 0
-    ) -> Tuple[torch.Tensor, torch.Tensor]:
-        """Perform the forward pass.
-
-        Args:
-            qkv: Query, key and value tensors of shape (batch, seqlen, nheads, headdim) or (batch, seqlen, 3, nheads, headdim).
-            seqlen_offset: Used in generation where the passed `qkv` is only the last token in the batch.
-
-        Returns:
-            New `qkv` and the cached sinusoids.
-
-        """
-
-        self._update_cos_sin_cache(qkv, seqlen_offset)
-
-        return self.apply_rotary_emb_qkv(
-            qkv, self._sin_cached[seqlen_offset:], self._cos_cached[seqlen_offset:]
-        )
-
-
-def _update_kv_cache(kv, inference_params, layer_idx):
-    """kv: (batch_size, seqlen, 2, nheads, head_dim) or (batch_size, 1, 2, nheads, head_dim)
-    Adapted from https://github.com/Dao-AILab/flash-attention."""
-    # Pre-allocate memory for key-values for inference.
-    num_heads, head_dim = kv.shape[-2:]
-    if layer_idx not in inference_params.key_value_memory_dict:
-        kv_cache = torch.empty(
-            inference_params.max_batch_size,
-            inference_params.max_sequence_len,
-            2,
-            num_heads,
-            head_dim,
-            dtype=kv.dtype,
-            device=kv.device,
-        )
-        inference_params.key_value_memory_dict[layer_idx] = kv_cache
-    else:
-        kv_cache = inference_params.key_value_memory_dict[layer_idx]
-
-    # Adjust key and value for inference
-    batch_start = inference_params.batch_size_offset
-    batch_end = batch_start + kv.shape[0]
-    sequence_start = inference_params.sequence_len_offset
-    sequence_end = sequence_start + kv.shape[1]
-    assert batch_end <= (
-        kv_cache.shape[0] if kv_cache is not None else v_cache.shape[0]  # noqa
-    )
-    assert sequence_end <= (
-        kv_cache.shape[1] if kv_cache is not None else v_cache.shape[2]  # noqa
-    )
-
-    assert kv_cache is not None
-    kv_cache[batch_start:batch_end, sequence_start:sequence_end, ...] = kv
-    kv = kv_cache[batch_start:batch_end, :sequence_end, ...]
-    return kv
-
-
-class MLP(nn.Module):
-    """Multi-Layer Perceptron.
-
-    Reference:
-        Attention Is All You Need.
-        https://arxiv.org/pdf/1706.03762.pdf.
-
-    """
-
-    def __init__(
-        self,
-        config: PretrainedConfig,
-        n_inner: Optional[int] = None,
-        act_fn: Optional[str] = None,
-    ) -> None:
-        super().__init__()
-
-        act_fn = config.activation_function if act_fn is None else act_fn
-        assert act_fn in ACT2FN.keys(), f"`act_fn` must be one of: {ACT2FN.keys()}."
-
-        n_inner = getattr(config, "n_inner", None) if n_inner is None else n_inner
-        n_inner = n_inner if n_inner is not None else 4 * config.n_embd
-
-        self.fc1 = nn.Linear(config.n_embd, n_inner)
-        self.fc2 = nn.Linear(n_inner, config.n_embd)
-        self.act = ACT2FN[act_fn]
-
-    def _load_from_state_dict(
-        self,
-        state_dict,
-        prefix,
-        local_metadata,
-        strict,
-        missing_keys,
-        unexpected_keys,
-        error_msgs,
-    ):
-        old_keys = [
-            prefix + "fc_in.weight",
-            prefix + "fc_out.weight",
-            prefix + "fc_in.bias",
-            prefix + "fc_out.bias",
-        ]
-        new_keys = [
-            prefix + "fc1.weight",
-            prefix + "fc2.weight",
-            prefix + "fc1.bias",
-            prefix + "fc2.bias",
-        ]
-
-        if all(k in state_dict for k in old_keys) and not all(
-            k in state_dict for k in new_keys
-        ):
-            # Older version of `MLP` saved with different key names.
-            for old_key, new_key in zip(old_keys, new_keys):
-                state_dict[new_key] = state_dict.pop(old_key)
-
-        return super()._load_from_state_dict(
-            state_dict,
-            prefix,
-            local_metadata,
-            strict,
-            missing_keys,
-            unexpected_keys,
-            error_msgs,
-        )
-
-    def forward(self, hidden_states: torch.FloatTensor) -> torch.FloatTensor:
-        hidden_states = self.fc1(hidden_states)
-        hidden_states = self.act(hidden_states)
-        hidden_states = self.fc2(hidden_states)
-
-        return hidden_states
-
-
-class FusedMLP(nn.Module):
-    """Fused Multi-Layer Perceptron from `flash-attn`.
-
-    Reference:
-        https://github.com/HazyResearch/flash-attention/blob/main/flash_attn/ops/fused_dense.py.
-
-    """
-
-    def __init__(
-        self,
-        config: PretrainedConfig,
-        n_inner: Optional[int] = None,
-        act_fn: Optional[str] = None,
-        raise_on_missing: bool = False,
-    ) -> None:
-        super().__init__()
-
-        act_fn = config.activation_function if act_fn is None else act_fn
-        assert act_fn in ACT2FN.keys(), f"`act_fn` must be one of: {ACT2FN.keys()}."
-
-        n_inner = getattr(config, "n_inner", None) if n_inner is None else n_inner
-        n_inner = n_inner if n_inner is not None else 4 * config.n_embd
-
-        gelu_activations = ["gelu_new", "gelu_fast", "gelu_approx"]  # noqa
-        activation = "gelu_approx" if act_fn in gelu_activations else "relu"  # noqa
-
-        self.mlp = MLP(config, n_inner=n_inner, act_fn=act_fn)
-
-    def forward(self, hidden_states: torch.FloatTensor) -> torch.FloatTensor:
-        return self.mlp(hidden_states)
-
-
-class SelfAttention(nn.Module):
-    """Implement the scaled dot product attention with softmax.
-    Adapted from https://github.com/Dao-AILab/flash-attention.
-    Arguments
-    ---------
-        softmax_scale: The temperature to use for the softmax attention.
-                      (default: 1/sqrt(d_keys) where d_keys is computed at
-                      runtime)
-        attention_dropout: The dropout rate to apply to the attention
-                           (default: 0.0)
-    """
-
-    def __init__(self, causal=False, softmax_scale=None, attention_dropout=0.0):
-        super().__init__()
-        self.causal = causal
-        self.softmax_scale = softmax_scale
-        self.drop = nn.Dropout(attention_dropout)
-
-    def forward(
-        self, qkv, causal=None, key_padding_mask=None, cu_seqlens=None, max_seqlen=None
-    ):
-        """Implements the multihead softmax attention.
-        Arguments
-        ---------
-            qkv: The tensor containing the query, key, and value. (B, S, 3, H, D)
-            causal: if passed, will override self.causal
-            key_padding_mask: boolean mask to apply to the attention weights. True means to keep,
-                False means to mask out. (B, S)
-        """
-        causal = self.causal if causal is None else causal
-        if cu_seqlens is not None:
-            return flash_attn_varlen_qkvpacked_func(
-                qkv.squeeze(0),
-                cu_seqlens,
-                max_seqlen,
-                dropout_p=self.drop.p,
-                softmax_scale=self.softmax_scale,
-                causal=causal,
-            )
-        else:
-            return flash_attn_qkvpacked_func(
-                qkv,
-                dropout_p=self.drop.p,
-                softmax_scale=self.softmax_scale,
-                causal=causal,
-            )
-
-
-class CrossAttention(nn.Module):
-    """Implement the scaled dot product attention with softmax.
-    Adapted from https://github.com/Dao-AILab/flash-attention.
-    Arguments
-    ---------
-        softmax_scale: The temperature to use for the softmax attention.
-                      (default: 1/sqrt(d_keys) where d_keys is computed at
-                      runtime)
-        attention_dropout: The dropout rate to apply to the attention
-                           (default: 0.0)
-    """
-
-    def __init__(self, causal=False, softmax_scale=None, attention_dropout=0.0):
-        super().__init__()
-        self.causal = causal
-        self.softmax_scale = softmax_scale
-        self.drop = nn.Dropout(attention_dropout)
-
-    def forward(self, q, kv, causal=None, key_padding_mask=None):
-        """Implements the multihead softmax attention.
-        Arguments
-        ---------
-            q: The tensor containing the query. (B, Sq, H, D)
-            kv: The tensor containing the key and value. (B, Sk, 2, H, D)
-            causal: if passed, will override self.causal
-            key_padding_mask: boolean mask to apply to the attention weights. True means to keep,
-                False means to mask out. (B, Sk)
-        """
-        causal = self.causal if causal is None else causal
-        return flash_attn_kvpacked_func(
-            q,
-            kv,
-            dropout_p=self.drop.p,
-            softmax_scale=self.softmax_scale,
-            causal=causal,
-        )
-
-
-def find_mha_dims(
-    config: PretrainedConfig,
-    n_head: Optional[int] = None,
-    head_dim: Optional[int] = None,
-) -> Tuple[int, int]:
-    """Validate and return the number of heads and head dimension for multi-head attention.
-
-    Args:
-        config: Model configuration.
-        n_head: Number of heads.
-        head_dim: Head dimension.
-
-    Returns:
-        Number of heads and head dimension.
-
-    """
-
-    assert all(
-        hasattr(config, attr) for attr in ["n_embd", "n_head"]
-    ), "`config` must have `n_embd` and `n_head` attributes."
-
-    if head_dim is None:
-        assert (
-            config.n_embd % config.n_head == 0
-        ), f"Hidden size ({config.n_embd}) must be divisible by the number of heads ({config.n_head})."
-
-    if n_head is None and head_dim is None:
-        head_dim = config.n_embd // config.n_head
-        n_head = config.n_head
-    elif n_head is None or head_dim is None:
-        raise ValueError("`n_head` and `head_dim` must be both specified or `None`.")
-
-    return n_head, head_dim
-
-
-class MHA(nn.Module):
-    """Multi-head attention layer.
-    Adapted from https://github.com/Dao-AILab/flash-attention."""
-
-    def __init__(
-        self,
-        config: PretrainedConfig,
-        rotary_dim: Optional[int] = None,
-        n_head: Optional[int] = None,
-        head_dim: Optional[int] = None,
-        bias: Optional[bool] = True,
-        dropout: Optional[float] = 0.0,
-        softmax_scale: Optional[float] = None,
-        causal: Optional[bool] = True,
-        layer_idx: Optional[int] = None,
-        rotary_emb_scale_base: Optional[float] = None,
-        return_residual: Optional[bool] = False,
-        checkpointing: Optional[bool] = False,
-        device: Optional[str] = None,
-        dtype: Optional[torch.dtype] = None,
-        fused_dense: Optional[bool] = True,
-        flash_attn: Optional[bool] = True,
-        cutlass_attn: Optional[bool] = False,
-        flash_rotary: Optional[bool] = True,
-        raise_on_missing: Optional[bool] = False,
-    ) -> None:
-        super().__init__()
-
-        factory_kwargs = {"device": device, "dtype": dtype}
-        n_head, head_dim = find_mha_dims(config, n_head, head_dim)
-
-        self.hidden_size = config.n_embd
-        self.n_head = n_head
-        self.head_dim = head_dim
-        self.op_size = n_head * head_dim
-
-        self.causal = causal
-        self.layer_idx = layer_idx
-        self.rotary_emb_dim = (
-            rotary_dim if rotary_dim is not None else getattr(config, "rotary_dim", 0)
-        )
-        self.fused_dense = fused_dense
-        self.flash_attn = flash_attn
-        self.cutlass_attn = cutlass_attn
-        self.flash_rotary = flash_rotary
-        self.return_residual = return_residual
-        self.checkpointing = checkpointing
-
-        if self.rotary_emb_dim > 0:
-            rotary_kwargs = {"device": device}
-            if rotary_emb_scale_base is not None and rotary_emb_scale_base > 0.0:
-                rotary_kwargs["scale_base"] = rotary_emb_scale_base
-
-            self.rotary_emb = RotaryEmbedding(self.rotary_emb_dim, **rotary_kwargs)
-        else:
-            pass
-
-        self.Wqkv = nn.Linear(
-            self.hidden_size, 3 * self.op_size, bias=bias, **factory_kwargs
-        )
-        self.out_proj = nn.Linear(
-            self.op_size, self.hidden_size, bias=bias, **factory_kwargs
-        )
-
-        self.inner_attn = SelfAttention(
-            causal=causal, softmax_scale=softmax_scale, attention_dropout=dropout
-        )
-        self.inner_cross_attn = CrossAttention(
-            causal=causal, softmax_scale=softmax_scale, attention_dropout=dropout
-        )
-
-    def _update_kv_cache(
-        self, kv: torch.FloatTensor, inference_params: InferenceParams
-    ) -> None:
-        """kv: (batch_size, seqlen, 2, nheads, head_dim) or (batch_size, 1, 2, nheads, head_dim)
-        Adapted from https://github.com/Dao-AILab/flash-attention."""
-
-        assert (
-            self.layer_idx is not None
-        ), "Generation requires layer_idx in the constructor"
-
-        return _update_kv_cache(kv, inference_params, self.layer_idx)
-
-    def forward(
-        self,
-        x: torch.FloatTensor,
-        x_kv: Optional[torch.FloatTensor] = None,
-        key_padding_mask: Optional[torch.BoolTensor] = None,
-        cu_seqlens: Optional[torch.LongTensor] = None,
-        max_seqlen: Optional[int] = None,
-        mixer_subset: Optional[torch.LongTensor] = None,
-        past_cache: Optional[InferenceParams] = None,
-        **kwargs,
-    ) -> Tuple[torch.FloatTensor, torch.FloatTensor]:
-        """Perform the forward pass.
-
-        Args:
-            x: (batch, seqlen, hidden_dim) (where hidden_dim = num heads * head dim) if
-                cu_seqlens is None and max_seqlen is None, else (total, hidden_dim) where total
-                is the is the sum of the sequence lengths in the batch.
-            x_kv: (batch, seqlen, hidden_dim), only applicable for cross-attention. If None, use x.
-            key_padding_mask: boolean mask, True means to keep, False means to mask out.
-                (batch, seqlen). Only applicable when not using FlashAttention.
-            cu_seqlens: (batch_size + 1,), dtype torch.int32. The cumulative sequence lengths
-                of the sequences in the batch, used to index into x. Only applicable when using
-                FlashAttention.
-            max_seqlen: int. Maximum sequence length in the batch.
-            mixer_subset: for cross-attention only. If not None, will take a subset of x
-                before applying the query projection. Useful for e.g., ViT where we only care
-                about the CLS token in the last layer.
-            past_cache: For generation only.
-
-        Returns:
-            (batch, seqlen, hidden_dim) if cu_seqlens is None and max_seqlen is None,
-                else (total, hidden_dim) where total is the is the sum of the sequence lengths
-                in the batch.
-
-        """
-
-        if cu_seqlens is not None:
-            assert max_seqlen is not None
-            assert key_padding_mask is None
-            assert self.flash_attn
-            # assert self.rotary_emb_dim == 0
-
-        if key_padding_mask is not None:
-            assert cu_seqlens is None
-            assert max_seqlen is None
-            assert not self.flash_attn
-
-        if past_cache is not None:
-            assert key_padding_mask is None
-            assert cu_seqlens is None and max_seqlen is None
-
-        attn_kwargs = {"key_padding_mask": key_padding_mask}
-
-        assert x_kv is None and mixer_subset is None
-
-        qkv = self.Wqkv(x)
-        qkv = rearrange(
-            qkv, "... (three h d) -> ... three h d", three=3, d=self.head_dim
-        )
-
-        if past_cache is None:
-            if self.rotary_emb_dim > 0:
-                qkv = self.rotary_emb(qkv)
-                context = self.inner_attn(
-                    qkv, cu_seqlens=cu_seqlens, max_seqlen=max_seqlen, **attn_kwargs
-                )
-
-        else:
-            if self.rotary_emb_dim > 0:
-                qkv = self.rotary_emb(qkv, seqlen_offset=past_cache.sequence_len_offset)
-            q = qkv[:, :, 0]
-            kv = self._update_kv_cache(qkv[:, :, 1:], past_cache)
-            # If we're processing the prompt, causal=None (use self.causal).
-            # If we're decoding, then causal=False.
-            causal = None if past_cache.sequence_len_offset == 0 else False
-            context = self.inner_cross_attn(q, kv, causal=causal)
-
-        out = rearrange(context, "... h d -> ... (h d)")
-        out = self.out_proj(out)
-
-        return out if not self.return_residual else (out, x)
-
-
-class ParallelBlock(nn.Module):
-    """Parallel block.
-
-    This block applies parallel mixer and MLP layers to the input (used in GPT-J and CodeGen).
-
-    """
-
-    def __init__(
-        self,
-        config: PretrainedConfig,
-        mixer: Optional[Dict[str, Any]] = None,
-        mlp: Optional[Dict[str, Any]] = None,
-        block_idx: Optional[int] = None,
-    ) -> None:
-        super().__init__()
-
-        self.ln = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
-        self.resid_dropout = nn.Dropout(config.resid_pdrop)
-        self.block_idx = block_idx
-
-        self.mixer = MHA(config=config, **mixer, layer_idx=block_idx)
-        mlp_cls = mlp.pop("mlp_cls")
-        if mlp_cls == "fused_mlp":
-            self.mlp = FusedMLP(config=config, **mlp)
-        else:
-            self.mlp = MLP(config=config, **mlp)
-
-    def forward(
-        self,
-        hidden_states: torch.FloatTensor,
-        past_cache: Optional[torch.FloatTensor] = None,
-        cu_seqlens: Optional[torch.LongTensor] = None,
-        max_seqlen: Optional[int] = None,
-    ) -> torch.FloatTensor:
-        residual = hidden_states
-        hidden_states = self.ln(hidden_states)
-
-        attn_outputs = self.mixer(
-            hidden_states,
-            past_cache=past_cache,
-            cu_seqlens=cu_seqlens,
-            max_seqlen=max_seqlen,
-        )
-        if isinstance(attn_outputs, tuple):
-            attn_outputs = attn_outputs[0]
-
-        attn_outputs = self.resid_dropout(attn_outputs)
-        feed_forward_hidden_states = self.resid_dropout(self.mlp(hidden_states))
-
-        hidden_states = attn_outputs + feed_forward_hidden_states + residual
-
-        return hidden_states
-
-
-class CausalLMHead(nn.Module):
-    """Causal Language Modeling head.
-
-    Reference:
-        Improving Language Understanding by Generative Pre-Training.
-        https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf.
-
-    """
-
-    def __init__(self, config: PretrainedConfig) -> None:
-        super().__init__()
-
-        self.ln = nn.LayerNorm(config.n_embd, eps=config.layer_norm_epsilon)
-        self.linear = nn.Linear(config.n_embd, config.vocab_size)
-
-    def forward(self, hidden_states: torch.FloatTensor) -> torch.FloatTensor:
-        hidden_states = self.ln(hidden_states)
-        logits = self.linear(hidden_states).to(torch.float32)
-
-        return logits
-
-
-class CausalLMLoss(nn.Module):
-    """Causal Language Modeling loss.
-
-    Reference:
-        Improving Language Understanding by Generative Pre-Training.
-        https://cdn.openai.com/research-covers/language-unsupervised/language_understanding_paper.pdf.
-
-    """
-
-    def __init__(self, shift_labels: Optional[bool] = True) -> None:
-        super().__init__()
-
-        self.shift_labels = shift_labels
-        self.loss_fct = nn.CrossEntropyLoss()
-
-    def forward(
-        self, logits: torch.FloatTensor, labels: torch.LongTensor
-    ) -> torch.FloatTensor:
-        if self.shift_labels:
-            logits = logits[..., :-1, :].contiguous()
-            labels = labels[..., 1:].contiguous()
-
-        loss = self.loss_fct(logits.view(-1, logits.size(-1)), labels.view(-1))
-
-        return loss
-
-
-class MixFormerSequentialPreTrainedModel(PreTrainedModel):
-    """MixFormer (sequential for DeepSpeed) pre-trained model."""
-
-    config_class = MixFormerSequentialConfig
-    base_model_prefix = "transformer"
-    supports_gradient_checkpointing = True
-
-    def __init__(self, *inputs, **kwargs) -> None:
-        super().__init__(*inputs, **kwargs)
-
-    def prepare_inputs_for_generation(
-        self, input_ids, past_key_values=None, **kwargs
-    ) -> Dict[str, Any]:
-        if "use_cache" in kwargs and not kwargs["use_cache"]:
-            return {"input_ids": input_ids}
-
-        if past_key_values is None or not (
-            isinstance(past_key_values, InferenceParams)
-        ):
-            past_key_values = InferenceParams(
-                max_batch_size=input_ids.shape[0],
-                max_sequence_len=self.config.n_positions,
-                sequence_len_offset=0,
-                batch_size_offset=0,
-                fused_ft_kernel=False,
-                key_value_memory_dict={},
-            )
-        else:
-            # assume past_key_values has cached all but last token in input_ids
-            past_key_values.sequence_len_offset = len(input_ids[0]) - 1
-            input_ids = input_ids[:, -1].unsqueeze(-1)
-
-        return {"input_ids": input_ids, "past_key_values": past_key_values, **kwargs}
-
-
-class PackedSequential(nn.Sequential):
-    def forward(
-        self,
-        input,
-        cu_seqlens: Optional[torch.LongTensor] = None,
-        max_seqlen: Optional[int] = None,
-    ):
-        for module in self:
-            sig = inspect.signature(module.forward)
-            if "cu_seqlens" in sig.parameters:
-                input = module(input, cu_seqlens=cu_seqlens, max_seqlen=max_seqlen)
-            else:
-                input = module(input)
-        return input
-
-
-class MixFormerSequentialForCausalLM(MixFormerSequentialPreTrainedModel):
-    """MixFormer (sequential for DeepSpeed) for Causal Language Modeling."""
-
-    _keys_to_ignore_on_load_missing = [""]
-    _keys_to_ignore_on_load_unexpected = [
-        r"layers\.\d+\.mlp.(fc_in|fc_out)\.(weight|bias)"
-    ]
-    _no_split_modules = ["ParallelBlock"]
-
-    def __init__(self, config: MixFormerSequentialConfig) -> None:
-        super().__init__(config)
-
-        modules = [Embedding(config)]
-        block_config = config.architecture
-
-        if not isinstance(block_config, list):
-            block_config = [block_config for _ in range(config.n_layer)]
-
-        if config.n_layer != len(block_config):
-            config.n_layer = len(block_config)
-
-        for block_idx, block in enumerate(block_config):
-            # `block_cls` with `legacy` value is for backward compatibility
-            # `path` key is for backward compatibility
-            block = copy.deepcopy(block) or {"block_cls": "parallel"}
-            block.pop("path", None) or block.pop("block_cls", None)
-
-            block["block_idx"] = block_idx
-            modules.append(ParallelBlock(config, **block))
-
-        modules.append(CausalLMHead(config))
-
-        self.layers = PackedSequential(*modules)
-        self.loss = CausalLMLoss()
-
-        self.post_init()
-
-    def get_input_embeddings(self) -> nn.Embedding:
-        return self.layers[0].wte
-
-    def set_input_embeddings(self, new_embeddings: nn.Embedding) -> None:
-        self.layers[0].wte = new_embeddings
-
-    def get_output_embeddings(self) -> nn.Linear:
-        return self.layers[-1].linear
-
-    def set_output_embeddings(self, new_embeddings: nn.Linear) -> None:
-        self.layers[-1].linear = new_embeddings
-
-    def forward(
-        self,
-        input_ids: torch.LongTensor,
-        labels: Optional[torch.LongTensor] = None,
-        past_key_values: Optional[torch.FloatTensor] = None,
-        position_ids: Optional[torch.LongTensor] = None,
-        **kwargs,
-    ) -> CausalLMOutputWithPast:
-        cu_seqlens: Optional[torch.LongTensor] = None
-        max_seqlen: Optional[int] = None
-        if position_ids is not None:
-            batch_size, seq_length = input_ids.shape
-            position_ids = position_ids.view(-1, seq_length).long()
-            cu_seqlens, max_seqlen = get_cu_seqlens_from_pos_ids(position_ids)
-            cu_seqlens = cu_seqlens.squeeze()
-
-        if not past_key_values:
-            lm_logits = self.layers(
-                input_ids, cu_seqlens=cu_seqlens, max_seqlen=max_seqlen
-            )
-        else:
-            hidden_layer = self.layers[0](input_ids)
-            for module in self.layers[1:-1]:
-                hidden_layer = module(
-                    hidden_layer,
-                    past_cache=past_key_values,
-                    cu_seqlens=cu_seqlens,
-                    max_seqlen=max_seqlen,
-                )
-            lm_logits = self.layers[-1](hidden_layer)
-
-        loss = None
-        if labels is not None:
-            loss = self.loss(lm_logits, labels)
-
-        return CausalLMOutputWithPast(
-            loss=loss, logits=lm_logits, past_key_values=past_key_values
-        )

src/axolotl/monkeypatch/btlm_attn_hijack_flash.py

@@ -1,66 +0,0 @@
-"""
-Flash attention monkey patch for cerebras btlm model
-"""
-
-import importlib
-import logging
-from typing import Optional, Tuple
-
-import accelerate
-import torch
-from flash_attn.flash_attn_interface import flash_attn_func
-from transformers import AutoConfig, AutoModelForCausalLM
-
-LOG = logging.getLogger("axolotl")
-
-
-def replace_btlm_attn_with_flash_attn(model_name="cerebras/btlm-3b-8k-base"):
-    # this is a wonky hack to get the remotely loaded module
-    model_config = AutoConfig.from_pretrained(model_name, trust_remote_code=True)
-    # we need to load the model here in order for modeling_btlm to be available
-    with accelerate.init_empty_weights():
-        AutoModelForCausalLM(model_config)
-    module_name = model_config.__class__.__module__.replace(
-        ".configuration_btlm", ".modeling_btlm"
-    )
-    modeling_btlm = importlib.import_module(module_name)
-    modeling_btlm.BTLMAttention._attn = (  # pylint: disable=protected-access
-        flashattn_attn
-    )
-
-
-def flashattn_attn(
-    self,
-    query: torch.Tensor,
-    key: Optional[torch.Tensor] = None,
-    value: Optional[torch.Tensor] = None,
-    attention_mask: Optional[torch.Tensor] = None,  # pylint: disable=unused-argument
-    head_mask: Optional[torch.Tensor] = None,
-    position_bias: Optional[torch.Tensor] = None,  # pylint: disable=unused-argument
-) -> Tuple[torch.Tensor, Optional[torch.Tensor]]:
-    softmax_scale = (
-        1 / (key.size(-1) ** self.attn_scale_power) if self.scale_attn_weights else None
-    )
-
-    query = query.permute(0, 2, 1, 3)
-    key = key.permute(0, 2, 1, 3)
-    value = value.permute(0, 2, 1, 3)
-
-    # Perform Flash attention
-    attn_output = flash_attn_func(
-        query,
-        key,
-        value,
-        dropout_p=0.0,  # Assuming you have this attribute
-        softmax_scale=softmax_scale,  # Set this if you have specific scaling in mind
-        causal=not self.is_cross_attention,  # Assuming you have this attribute
-        return_attn_probs=False,  # Set this based on your needs
-    )
-
-    # Optional: Apply head mask if it's not None
-    if head_mask is not None:
-        attn_output *= head_mask
-
-    attn_output = attn_output.permute(0, 2, 1, 3)
-
-    return attn_output, None  # We don't have explicit attn_weights in Flash attention

src/axolotl/monkeypatch/falcon_attn_hijack_flash.py

@@ -1,101 +0,0 @@
-"""
-Flash Attention monkey patch for Falcon
-
-copied from https://github.com/pacman100/DHS-LLM-Workshop/blob/main/chat_assistant/training/falcon_flash_attn_monkey_patch.py
-"""
-
-from typing import Optional, Tuple
-
-import torch
-import transformers
-from flash_attn import flash_attn_func
-
-
-def forward(
-    self,
-    hidden_states: torch.Tensor,
-    alibi: Optional[torch.Tensor],
-    attention_mask: torch.Tensor,  # pylint: disable=unused-argument
-    layer_past: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
-    head_mask: Optional[torch.Tensor] = None,  # pylint: disable=unused-argument
-    use_cache: bool = False,
-    output_attentions: bool = False,  # pylint: disable=unused-argument
-):
-    fused_qkv = self.query_key_value(
-        hidden_states
-    )  # [batch_size, seq_length, 3 x hidden_size]
-    num_kv_heads = (
-        self.num_heads if self.new_decoder_architecture else self.num_kv_heads
-    )
-    # 3 x [batch_size, seq_length, num_heads, head_dim]
-    (
-        query_layer,
-        key_layer,
-        value_layer,
-    ) = self._split_heads(  # pylint: disable=protected-access
-        fused_qkv
-    )
-
-    batch_size, query_length, _, _ = query_layer.shape
-
-    query_layer = query_layer.transpose(1, 2).reshape(
-        batch_size * self.num_heads, query_length, self.head_dim
-    )
-    key_layer = key_layer.transpose(1, 2).reshape(
-        batch_size * num_kv_heads,
-        query_length,
-        self.head_dim,
-    )
-    value_layer = value_layer.transpose(1, 2).reshape(
-        batch_size * num_kv_heads, query_length, self.head_dim
-    )
-
-    past_kv_length = 0 if layer_past is None else layer_past[0].shape[1]
-    query_layer, key_layer = self.maybe_rotary(query_layer, key_layer, past_kv_length)
-
-    if layer_past is not None:
-        past_key, past_value = layer_past
-        # concatenate along seq_length dimension:
-        #  - key: [batch_size * self.num_heads, kv_length, head_dim]
-        #  - value: [batch_size * self.num_heads, kv_length, head_dim]
-        key_layer = torch.cat((past_key, key_layer), dim=1)
-        value_layer = torch.cat((past_value, value_layer), dim=1)
-
-    # unused
-    # _, kv_length, _ = key_layer.shape
-    if use_cache:
-        present = (key_layer, value_layer)
-    else:
-        present = None
-    # unused
-    # attention_mask_float = (attention_mask * 1.0).masked_fill(attention_mask, float("-1e9")).to(query_layer.dtype)
-    query_layer_ = (
-        query_layer.reshape(batch_size, self.num_heads, -1, self.head_dim)
-        .transpose(1, 2)
-        .to(torch.bfloat16)
-    )
-    key_layer_ = (
-        key_layer.reshape(batch_size, num_kv_heads, -1, self.head_dim)
-        .transpose(1, 2)
-        .to(torch.bfloat16)
-    )
-    value_layer_ = (
-        value_layer.reshape(batch_size, num_kv_heads, -1, self.head_dim)
-        .transpose(1, 2)
-        .to(torch.bfloat16)
-    )
-
-    if alibi is not None:
-        raise ValueError("`alibi` is not supported when `use_flash_attn` is True")
-
-    # below output will have shape (batch_size, seqlen, nheads, headdim)
-    attn_output = flash_attn_func(query_layer_, key_layer_, value_layer_, causal=True)
-    attn_output = attn_output.reshape(
-        batch_size, query_length, self.num_heads * self.head_dim
-    )
-    output_tensor = self.dense(attn_output)
-    return output_tensor, present
-
-
-def replace_falcon_attn_with_flash_attn():
-    transformers.models.falcon.modeling_falcon.FalconAttention.forward = forward

src/axolotl/monkeypatch/llama_attn_hijack_flash.py

@@ -1,633 +0,0 @@
-"""Flash attention monkey patch for llama model"""
-
-# copied from https://github.com/lm-sys/FastChat/blob/main/fastchat/train/llama_flash_attn_monkey_patch.py
-
-import logging
-import warnings
-from functools import partial
-from typing import List, Optional, Tuple, Union
-
-import torch
-import torch.nn.functional as F
-import transformers
-from einops import rearrange
-from flash_attn.bert_padding import pad_input, unpad_input
-from transformers.modeling_outputs import BaseModelOutputWithPast
-from transformers.models.llama.modeling_llama import (
-    LlamaDecoderLayer as OriginalLlamaDecoderLayer,
-)
-from transformers.models.llama.modeling_llama import apply_rotary_pos_emb, repeat_kv
-
-from axolotl.monkeypatch.utils import get_cu_seqlens_from_pos_ids
-
-try:
-    from flash_attn.flash_attn_interface import (  # pylint: disable=ungrouped-imports
-        flash_attn_kvpacked_func,
-        flash_attn_varlen_kvpacked_func,
-        flash_attn_varlen_qkvpacked_func,
-    )
-except ImportError:
-    from flash_attn.flash_attn_interface import (
-        flash_attn_unpadded_kvpacked_func as flash_attn_varlen_kvpacked_func,
-    )
-    from flash_attn.flash_attn_interface import (
-        flash_attn_unpadded_qkvpacked_func as flash_attn_varlen_qkvpacked_func,
-    )
-
-
-LOG = logging.getLogger("axolotl")
-
-
-def replace_llama_attn_with_flash_attn(packed: Optional[bool] = False):
-    transformers.models.llama.modeling_llama.LlamaModel._prepare_decoder_attention_mask = (  # pylint: disable=protected-access
-        _prepare_decoder_attention_mask
-    )
-    transformers.models.llama.modeling_llama.LlamaAttention.forward = flashattn_forward
-    if packed:
-        transformers.models.llama.modeling_llama.LlamaDecoderLayer = LlamaDecoderLayer
-        transformers.models.llama.modeling_llama.LlamaModel.forward = (
-            llama_model_forward
-        )
-
-    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.info(
-            "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'`)"
-        )
-
-    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.info(
-            "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'`)"
-        )
-
-
-# Disable the transformation of the attention mask in LlamaModel as the flash attention
-# requires the attention mask to be the same as the key_padding_mask
-def _prepare_decoder_attention_mask(
-    self,
-    attention_mask,
-    input_shape,
-    inputs_embeds,
-    past_key_values_length,
-):  # pylint: disable=unused-argument
-    # [bsz, seq_len]
-    return attention_mask
-
-
-def flashattn_forward(
-    self,
-    hidden_states: torch.Tensor,
-    attention_mask: Optional[torch.Tensor] = None,
-    position_ids: Optional[torch.Tensor] = None,
-    past_key_value: Optional[Tuple[torch.Tensor]] = None,
-    output_attentions: bool = False,
-    use_cache: bool = False,
-    cu_seqlens: Optional[torch.Tensor] = None,
-    max_seqlen: Optional[torch.Tensor] = None,
-) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
-    """Input shape: Batch x Time x Channel
-
-    attention_mask: [bsz, q_len]
-    """
-    # pylint: disable=duplicate-code
-    bsz, q_len, _ = hidden_states.size()
-
-    if not hasattr(self, "pretraining_tp"):
-        self.pretraining_tp = 1
-
-    if self.pretraining_tp > 1:
-        key_value_slicing = (
-            self.num_key_value_heads * self.head_dim
-        ) // self.pretraining_tp
-        query_slices = self.q_proj.weight.split(
-            (self.num_heads * self.head_dim) // self.pretraining_tp, dim=0
-        )
-        key_slices = self.k_proj.weight.split(key_value_slicing, dim=0)
-        value_slices = self.v_proj.weight.split(key_value_slicing, dim=0)
-
-        query_states = [
-            F.linear(hidden_states, query_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        query_states = torch.cat(query_states, dim=-1)
-
-        key_states = [
-            F.linear(hidden_states, key_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        key_states = torch.cat(key_states, dim=-1)
-
-        value_states = [
-            F.linear(hidden_states, value_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        value_states = torch.cat(value_states, dim=-1)
-
-    else:
-        query_states = self.q_proj(hidden_states)
-        key_states = self.k_proj(hidden_states)
-        value_states = self.v_proj(hidden_states)
-
-    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)
-    # [bsz, q_len, nh, hd]
-    # [bsz, nh, q_len, hd]
-
-    kv_seq_len = key_states.shape[-2]
-    if past_key_value is not None:
-        kv_seq_len += past_key_value[0].shape[-2]
-
-    cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
-    query_states, key_states = apply_rotary_pos_emb(
-        query_states, key_states, cos, sin, position_ids
-    )
-    # [bsz, nh, t, hd]
-
-    if past_key_value is not None:
-        # reuse k, v, self_attention
-        key_states = torch.cat([past_key_value[0], key_states], dim=2)
-        value_states = torch.cat([past_key_value[1], value_states], dim=2)
-
-    past_key_value = (key_states, value_states) if use_cache else None
-
-    # repeat k/v heads if n_kv_heads < n_heads
-    key_states = repeat_kv(key_states, self.num_key_value_groups)
-    value_states = repeat_kv(value_states, self.num_key_value_groups)
-
-    if output_attentions:
-        warnings.warn(
-            "Output attentions is not supported for patched `LlamaAttention`, returning `None` instead."
-        )
-
-    #
-    # flash-attn v2 start
-    #
-
-    if self.training:
-        # during training q,k,v always have same seqlen
-        assert key_states.shape == query_states.shape
-        is_causal = True
-    else:
-        # turn off FA causal mask after first inference autoregressive iteration
-        # only on first autoregressive step q,k,v have same seqlen
-        is_causal = key_states.shape == query_states.shape
-
-    if cu_seqlens is not None and max_seqlen is not None and cu_seqlens.dim() == 1:
-        # special handling using sample packing
-        qkv = torch.stack(
-            [query_states, key_states, value_states], dim=2
-        )  # [bsz, nh, 3, q_len, hd]
-        qkv = qkv.transpose(1, 3)  # [bsz, q_len, 3, nh, hd]
-        qkv = rearrange(qkv, "b s ... -> (b s) ...")
-
-        output = flash_attn_varlen_qkvpacked_func(
-            qkv, cu_seqlens, max_seqlen, 0.0, softmax_scale=None, causal=True
-        )
-        output = rearrange(output, "(b s) ... -> b s ...", b=bsz)
-    elif query_states.shape == key_states.shape:
-        query_states = query_states.transpose(1, 2)
-        key_states = key_states.transpose(1, 2)
-        value_states = value_states.transpose(1, 2)
-        qkv_unpad, cu_seqlens_q, max_seqlen_q, _, output_pad_fn = generate_qkv(
-            query_states,
-            key_states,
-            value_states,
-            qkvpacked=True,
-            # We have disabled _prepare_decoder_attention_mask in LlamaModel
-            # the attention_mask should be the same as the key_padding_mask
-            key_padding_mask=attention_mask,
-            query_padding_mask=attention_mask[:, -query_states.size(1) :]
-            if attention_mask is not None
-            else None,
-        )
-        output_unpad = flash_attn_varlen_qkvpacked_func(
-            qkv_unpad,
-            cu_seqlens_q,
-            max_seqlen_q,
-            0.0,
-            softmax_scale=None,
-            causal=is_causal,
-        )
-        output = output_pad_fn(output_unpad)
-    else:
-        query_states = query_states.transpose(1, 2)
-        key_states = key_states.transpose(1, 2)
-        value_states = value_states.transpose(1, 2)
-        if attention_mask is None or attention_mask.all().item():
-            output = flash_attn_kvpacked_func(
-                query_states,
-                torch.stack([key_states, value_states], 2),
-                causal=is_causal,
-            )
-        else:
-            (  # pylint: disable=unbalanced-tuple-unpacking
-                q_unpad,
-                kv_unpad,
-                cu_seqlens_q,
-                cu_seqlens_k,
-                max_seqlen_q,
-                max_seqlen_k,
-                _,
-                _,
-                output_pad_fn,
-            ) = generate_qkv(
-                query_states,
-                key_states,
-                value_states,
-                kvpacked=True,
-                key_padding_mask=attention_mask,
-                query_padding_mask=attention_mask[:, -query_states.size(1) :]
-                if attention_mask is not None
-                else None,
-            )
-            if q_unpad.dtype != kv_unpad.dtype:
-                kv_unpad = kv_unpad.to(q_unpad.dtype)
-            output_unpad = flash_attn_varlen_kvpacked_func(
-                q_unpad,
-                kv_unpad,
-                cu_seqlens_q,
-                cu_seqlens_k,
-                max_seqlen_q,
-                max_seqlen_k,
-                0.0,
-                softmax_scale=None,
-                causal=is_causal,
-            )
-            output = output_pad_fn(output_unpad)
-
-    attn_output = output
-    if attn_output.size() != (bsz, q_len, self.num_heads, self.head_dim):
-        raise ValueError(
-            f"`attn_output` should be of size {(bsz, q_len, self.num_heads, self.head_dim)}, but is"
-            f" {attn_output.size()}"
-        )
-    attn_output = rearrange(attn_output, "b s h d -> b s (h d)")
-
-    #
-    # flash-attn v2 end
-    #
-
-    if self.pretraining_tp > 1:
-        attn_output = attn_output.split(self.hidden_size // self.pretraining_tp, dim=2)
-        o_proj_slices = self.o_proj.weight.split(
-            self.hidden_size // self.pretraining_tp, dim=1
-        )
-        attn_output = sum(
-            F.linear(attn_output[i], o_proj_slices[i])
-            for i in range(self.pretraining_tp)
-        )
-    else:
-        attn_output = self.o_proj(attn_output)
-
-    return attn_output, None, past_key_value
-
-
-# based on https://github.com/Dao-AILab/flash-attention/blob/364a5b/tests/test_flash_attn.py#L38
-def generate_qkv(
-    q,
-    k,
-    v,
-    query_padding_mask=None,
-    key_padding_mask=None,
-    kvpacked=False,
-    qkvpacked=False,
-):  # pylint: disable=invalid-name,unnecessary-lambda-assignment
-    """
-    Arguments:
-        q: (batch_size, seqlen_q, nheads, d)
-        k: (batch_size, seqlen_k, nheads_k, d)
-        v: (batch_size, seqlen_k, nheads_k, d)
-        query_padding_mask: (batch_size, seqlen), bool
-        key_padding_mask: (batch_size, seqlen), bool
-    """
-    assert not (kvpacked and qkvpacked)
-    batch_size, seqlen_q, nheads, d = q.shape
-    _, seqlen_k, nheads_k, _ = k.shape
-    assert k.shape == (batch_size, seqlen_k, nheads_k, d)
-    assert v.shape == (batch_size, seqlen_k, nheads_k, d)
-
-    if query_padding_mask is not None:
-        q_unpad, indices_q, cu_seqlens_q, max_seqlen_q = unpad_input(
-            q, query_padding_mask
-        )
-
-        output_pad_fn = lambda output_unpad: pad_input(  # noqa: E731
-            output_unpad, indices_q, batch_size, seqlen_q
-        )
-
-    else:
-        q_unpad = rearrange(q, "b s h d -> (b s) h d")
-        cu_seqlens_q = torch.arange(
-            0,
-            (batch_size + 1) * seqlen_q,
-            step=seqlen_q,
-            dtype=torch.int32,
-            device=q_unpad.device,
-        )
-        max_seqlen_q = seqlen_q
-
-        output_pad_fn = lambda output_unpad: rearrange(  # noqa: E731
-            output_unpad, "(b s) h d -> b s h d", b=batch_size
-        )
-
-    if key_padding_mask is not None:
-        k_unpad, _, cu_seqlens_k, max_seqlen_k = unpad_input(k, key_padding_mask)
-        v_unpad, _, _, _ = unpad_input(v, key_padding_mask)
-    else:
-        k_unpad = rearrange(k, "b s h d -> (b s) h d")
-        v_unpad = rearrange(v, "b s h d -> (b s) h d")
-        cu_seqlens_k = torch.arange(
-            0,
-            (batch_size + 1) * seqlen_k,
-            step=seqlen_k,
-            dtype=torch.int32,
-            device=k_unpad.device,
-        )
-        max_seqlen_k = seqlen_k
-
-    if qkvpacked:
-        assert nheads == nheads_k
-        qkv_unpad = torch.stack([q_unpad, k_unpad, v_unpad], dim=1)
-        qkv = torch.stack([q, k, v], dim=2)
-        return (qkv_unpad, cu_seqlens_q, max_seqlen_q, qkv, output_pad_fn)
-
-    if kvpacked:
-        kv_unpad = torch.stack([k_unpad, v_unpad], dim=1)
-        kv = torch.stack([k, v], dim=2)
-        return (
-            q_unpad,
-            kv_unpad,
-            cu_seqlens_q,
-            cu_seqlens_k,
-            max_seqlen_q,
-            max_seqlen_k,
-            q,
-            kv,
-            output_pad_fn,
-        )
-
-    return (
-        q_unpad,
-        k_unpad,
-        v_unpad,
-        cu_seqlens_q,
-        cu_seqlens_k,
-        max_seqlen_q,
-        max_seqlen_k,
-        q,
-        k,
-        v,
-        output_pad_fn,
-    )
-
-
-def llama_model_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,
-    use_cache: Optional[bool] = None,
-    output_attentions: Optional[bool] = None,
-    output_hidden_states: Optional[bool] = None,
-    return_dict: Optional[bool] = None,
-) -> Union[Tuple, BaseModelOutputWithPast]:
-    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
-    )
-    use_cache = use_cache if use_cache is not None else self.config.use_cache
-
-    return_dict = (
-        return_dict if return_dict is not None else self.config.use_return_dict
-    )
-
-    # retrieve input_ids and inputs_embeds
-    if input_ids is not None and inputs_embeds is not None:
-        raise ValueError(
-            "You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time"
-        )
-    if input_ids is not None:
-        batch_size, seq_length = input_ids.shape
-    elif inputs_embeds is not None:
-        batch_size, seq_length, _ = inputs_embeds.shape
-    else:
-        raise ValueError(
-            "You have to specify either decoder_input_ids or decoder_inputs_embeds"
-        )
-
-    seq_length_with_past = seq_length
-    past_key_values_length = 0
-
-    if past_key_values is not None:
-        past_key_values_length = past_key_values[0][0].shape[2]
-        seq_length_with_past = seq_length_with_past + past_key_values_length
-
-    cu_seqlens = None
-    max_seqlen = None
-    if position_ids is None:
-        device = input_ids.device if input_ids is not None else inputs_embeds.device
-        position_ids = torch.arange(
-            past_key_values_length,
-            seq_length + past_key_values_length,
-            dtype=torch.long,
-            device=device,
-        )
-        position_ids = position_ids.unsqueeze(0).view(-1, seq_length)
-    else:
-        position_ids = position_ids.view(-1, seq_length).long()
-        cu_seqlens, max_seqlen = get_cu_seqlens_from_pos_ids(position_ids)
-        cu_seqlens = cu_seqlens.squeeze()
-
-    if inputs_embeds is None:
-        inputs_embeds = self.embed_tokens(input_ids)
-    # embed positions
-    if attention_mask is None:
-        attention_mask = torch.ones(
-            (batch_size, seq_length_with_past),
-            dtype=torch.bool,
-            device=inputs_embeds.device,
-        )
-    attention_mask = (
-        self._prepare_decoder_attention_mask(  # pylint: disable=protected-access
-            attention_mask,
-            (batch_size, seq_length),
-            inputs_embeds,
-            past_key_values_length,
-        )
-    )
-
-    hidden_states = inputs_embeds
-
-    if self.gradient_checkpointing and self.training:
-        if use_cache:
-            transformers.logger.warning_once(
-                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
-            )
-            use_cache = False
-
-    # decoder layers
-    all_hidden_states = () if output_hidden_states else None
-    all_self_attns = () if output_attentions else None
-    next_decoder_cache = () if use_cache else None
-
-    for idx, decoder_layer in enumerate(self.layers):
-        if output_hidden_states:
-            all_hidden_states += (hidden_states,)
-
-        past_key_value = past_key_values[idx] if past_key_values is not None else None
-
-        if self.gradient_checkpointing and self.training:
-
-            def create_custom_forward(module):
-                def custom_forward(*inputs):
-                    # None for past_key_value
-                    return module(*inputs)
-
-                return custom_forward
-
-            layer_outputs = torch.utils.checkpoint.checkpoint(
-                create_custom_forward(decoder_layer),
-                hidden_states,
-                attention_mask,
-                position_ids,
-                None,
-                output_attentions,
-                None,
-                cu_seqlens,
-                max_seqlen,
-            )
-        else:
-            layer_outputs = decoder_layer(
-                hidden_states,
-                attention_mask=attention_mask,
-                position_ids=position_ids,
-                past_key_value=past_key_value,
-                output_attentions=output_attentions,
-                use_cache=use_cache,
-                cu_seqlens=cu_seqlens,
-                max_seqlen=max_seqlen,
-            )
-
-        hidden_states = layer_outputs[0]
-
-        if use_cache:
-            next_decoder_cache += (layer_outputs[2 if output_attentions else 1],)
-
-        if output_attentions:
-            all_self_attns += (layer_outputs[1],)
-
-    hidden_states = self.norm(hidden_states)
-
-    # add hidden states from the last decoder layer
-    if output_hidden_states:
-        all_hidden_states += (hidden_states,)
-
-    next_cache = next_decoder_cache if use_cache else None
-    if not return_dict:
-        return tuple(
-            v
-            for v in [hidden_states, next_cache, all_hidden_states, all_self_attns]
-            if v is not None
-        )
-    return BaseModelOutputWithPast(
-        last_hidden_state=hidden_states,
-        past_key_values=next_cache,
-        hidden_states=all_hidden_states,
-        attentions=all_self_attns,
-    )
-
-
-class LlamaDecoderLayer(OriginalLlamaDecoderLayer):
-    """
-    patched version of LlamaDecoderLayer to pass through the precalculated cu_seqlens
-    """
-
-    def forward(
-        self,
-        hidden_states: torch.Tensor,
-        attention_mask: Optional[torch.Tensor] = None,
-        position_ids: Optional[torch.LongTensor] = None,
-        past_key_value: Optional[Tuple[torch.Tensor]] = None,
-        output_attentions: Optional[bool] = False,
-        use_cache: Optional[bool] = False,
-        cu_seqlens: Optional[torch.Tensor] = None,
-        max_seqlen: Optional[torch.Tensor] = None,
-    ) -> Tuple[
-        torch.FloatTensor, Optional[Tuple[torch.FloatTensor, torch.FloatTensor]]
-    ]:
-        """
-        Args:
-            hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)`
-            attention_mask (`torch.FloatTensor`, *optional*): attention mask of size
-                `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values.
-            output_attentions (`bool`, *optional*):
-                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
-                returned tensors for more detail.
-            use_cache (`bool`, *optional*):
-                If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding
-                (see `past_key_values`).
-            past_key_value (`Tuple(torch.FloatTensor)`, *optional*): cached past key and value projection states
-            cu_seqlens (`torch.Tensor`, *optional*) cumulative sequence len when packing
-        """
-
-        residual = hidden_states
-
-        hidden_states = self.input_layernorm(hidden_states)
-
-        # Self Attention
-        hidden_states, self_attn_weights, present_key_value = self.self_attn(
-            hidden_states=hidden_states,
-            attention_mask=attention_mask,
-            position_ids=position_ids,
-            past_key_value=past_key_value,
-            output_attentions=output_attentions,
-            use_cache=use_cache,
-            cu_seqlens=cu_seqlens,
-            max_seqlen=max_seqlen,
-        )
-        hidden_states = residual + hidden_states
-
-        # Fully Connected
-        residual = hidden_states
-        hidden_states = self.post_attention_layernorm(hidden_states)
-        hidden_states = self.mlp(hidden_states)
-        hidden_states = residual + hidden_states
-
-        outputs = (hidden_states,)
-
-        if output_attentions:
-            outputs += (self_attn_weights,)
-
-        if use_cache:
-            outputs += (present_key_value,)
-
-        return outputs

src/axolotl/monkeypatch/llama_attn_hijack_sdp.py

@@ -1,140 +0,0 @@
-"""
-Patched LlamaAttention to use torch.nn.functional.scaled_dot_product_attention
-"""
-
-import warnings
-from typing import Optional, Tuple
-
-import torch
-import torch.nn.functional as F
-import transformers.models.llama.modeling_llama
-from transformers.models.llama.modeling_llama import apply_rotary_pos_emb, repeat_kv
-
-
-def hijack_llama_sdp_attention():
-    transformers.models.llama.modeling_llama.LlamaAttention.forward = (
-        sdp_attention_forward
-    )
-
-
-def sdp_attention_forward(
-    self,
-    hidden_states: torch.Tensor,
-    attention_mask: Optional[torch.Tensor] = None,
-    position_ids: Optional[torch.LongTensor] = None,
-    past_key_value: Optional[Tuple[torch.Tensor]] = None,
-    output_attentions: bool = False,
-    use_cache: bool = False,
-) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
-    # pylint: disable=duplicate-code
-    bsz, q_len, _ = hidden_states.size()
-
-    if not hasattr(self, "pretraining_tp"):
-        self.pretraining_tp = 1
-
-    if self.pretraining_tp > 1:
-        key_value_slicing = (
-            self.num_key_value_heads * self.head_dim
-        ) // self.pretraining_tp
-        query_slices = self.q_proj.weight.split(
-            (self.num_heads * self.head_dim) // self.pretraining_tp, dim=0
-        )
-        key_slices = self.k_proj.weight.split(key_value_slicing, dim=0)
-        value_slices = self.v_proj.weight.split(key_value_slicing, dim=0)
-
-        query_states = [
-            F.linear(hidden_states, query_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        query_states = torch.cat(query_states, dim=-1)
-
-        key_states = [
-            F.linear(hidden_states, key_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        key_states = torch.cat(key_states, dim=-1)
-
-        value_states = [
-            F.linear(hidden_states, value_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        value_states = torch.cat(value_states, dim=-1)
-
-    else:
-        query_states = self.q_proj(hidden_states)
-        key_states = self.k_proj(hidden_states)
-        value_states = self.v_proj(hidden_states)
-
-    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)
-    # [bsz, q_len, nh, hd]
-    # [bsz, nh, q_len, hd]
-
-    kv_seq_len = key_states.shape[-2]
-    if past_key_value is not None:
-        kv_seq_len += past_key_value[0].shape[-2]
-
-    cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
-    query_states, key_states = apply_rotary_pos_emb(
-        query_states, key_states, cos, sin, position_ids
-    )
-    # [bsz, nh, t, hd]
-
-    if past_key_value is not None:
-        # reuse k, v, self_attention
-        key_states = torch.cat([past_key_value[0], key_states], dim=2)
-        value_states = torch.cat([past_key_value[1], value_states], dim=2)
-
-    past_key_value = (key_states, value_states) if use_cache else None
-
-    # repeat k/v heads if n_kv_heads < n_heads
-    key_states = repeat_kv(key_states, self.num_key_value_groups)
-    value_states = repeat_kv(value_states, self.num_key_value_groups)
-
-    if output_attentions:
-        warnings.warn(
-            "Output attentions is not supported for patched `LlamaAttention`, returning `None` instead."
-        )
-
-    #
-    # sdp-attn start
-    #
-
-    with torch.backends.cuda.sdp_kernel():
-        attn_output = torch.nn.functional.scaled_dot_product_attention(
-            query_states,
-            key_states,
-            value_states,
-            attn_mask=attention_mask,
-            is_causal=False,
-        )
-
-    if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
-        raise ValueError(
-            f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
-            f" {attn_output.size()}"
-        )
-    attn_output = attn_output.transpose(1, 2)
-    attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
-
-    #
-    # sdp-attn end
-    #
-
-    if self.pretraining_tp > 1:
-        attn_output = attn_output.split(self.hidden_size // self.pretraining_tp, dim=2)
-        o_proj_slices = self.o_proj.weight.split(
-            self.hidden_size // self.pretraining_tp, dim=1
-        )
-        attn_output = sum(
-            F.linear(attn_output[i], o_proj_slices[i])
-            for i in range(self.pretraining_tp)
-        )
-    else:
-        attn_output = self.o_proj(attn_output)
-
-    return attn_output, None, past_key_value

src/axolotl/monkeypatch/llama_attn_hijack_xformers.py

@@ -3,13 +3,12 @@ Directly copied the code from https://raw.githubusercontent.com/oobabooga/text-g
 """
 
 import logging
-import warnings
+import math
 from typing import Optional, Tuple
 
 import torch
-import torch.nn.functional as F
 import transformers.models.llama.modeling_llama
-from transformers.models.llama.modeling_llama import apply_rotary_pos_emb, repeat_kv
+from torch import nn
 
 try:
     import xformers.ops
@@ -21,6 +20,12 @@ def hijack_llama_attention():
     transformers.models.llama.modeling_llama.LlamaAttention.forward = xformers_forward
 
 
+def hijack_llama_sdp_attention():
+    transformers.models.llama.modeling_llama.LlamaAttention.forward = (
+        sdp_attention_forward
+    )
+
+
 def xformers_forward(
     self,
     hidden_states: torch.Tensor,
@@ -33,57 +38,30 @@ def xformers_forward(
     # pylint: disable=duplicate-code
     bsz, q_len, _ = hidden_states.size()
 
-    if not hasattr(self, "pretraining_tp"):
-        self.pretraining_tp = 1
-
-    if self.pretraining_tp > 1:
-        key_value_slicing = (
-            self.num_key_value_heads * self.head_dim
-        ) // self.pretraining_tp
-        query_slices = self.q_proj.weight.split(
-            (self.num_heads * self.head_dim) // self.pretraining_tp, dim=0
-        )
-        key_slices = self.k_proj.weight.split(key_value_slicing, dim=0)
-        value_slices = self.v_proj.weight.split(key_value_slicing, dim=0)
-
-        query_states = [
-            F.linear(hidden_states, query_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        query_states = torch.cat(query_states, dim=-1)
-
-        key_states = [
-            F.linear(hidden_states, key_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        key_states = torch.cat(key_states, dim=-1)
-
-        value_states = [
-            F.linear(hidden_states, value_slices[i]) for i in range(self.pretraining_tp)
-        ]
-        value_states = torch.cat(value_states, dim=-1)
-
-    else:
-        query_states = self.q_proj(hidden_states)
-        key_states = self.k_proj(hidden_states)
-        value_states = self.v_proj(hidden_states)
-
-    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)
-    # [bsz, q_len, nh, hd]
-    # [bsz, nh, q_len, hd]
+    query_states = (
+        self.q_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    key_states = (
+        self.k_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    value_states = (
+        self.v_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
 
     kv_seq_len = key_states.shape[-2]
     if past_key_value is not None:
         kv_seq_len += past_key_value[0].shape[-2]
-
     cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
-    query_states, key_states = apply_rotary_pos_emb(
+    (
+        query_states,
+        key_states,
+    ) = transformers.models.llama.modeling_llama.apply_rotary_pos_emb(
         query_states, key_states, cos, sin, position_ids
     )
     # [bsz, nh, t, hd]
@@ -95,61 +73,161 @@ def xformers_forward(
 
     past_key_value = (key_states, value_states) if use_cache else None
 
-    # repeat k/v heads if n_kv_heads < n_heads
-    key_states = repeat_kv(key_states, self.num_key_value_groups)
-    value_states = repeat_kv(value_states, self.num_key_value_groups)
+    # We only apply xformers optimizations if we don't need to output the whole attention matrix
+    if not output_attentions:
+        query_states = query_states.transpose(1, 2)
+        key_states = key_states.transpose(1, 2)
+        value_states = value_states.transpose(1, 2)
 
-    if output_attentions:
-        warnings.warn(
-            "Output attentions is not supported for patched `LlamaAttention`, returning `None` instead."
-        )
-
-    #
-    # xformers-attn start
-    #
-
-    query_states = query_states.transpose(1, 2)
-    key_states = key_states.transpose(1, 2)
-    value_states = value_states.transpose(1, 2)
-
-    # This is a nasty hack. We know attention_mask in transformers is either LowerTriangular or all Zeros.
-    # We therefore check if one element in the upper triangular portion is zero. If it is, then the mask is all zeros.
-    if attention_mask is None or attention_mask[0, 0, 0, 1] == 0:
-        # input and output should be of form (bsz, q_len, num_heads, head_dim)
-        attn_output = xformers.ops.memory_efficient_attention(
-            query_states, key_states, value_states, attn_bias=None
-        )
+        # This is a nasty hack. We know attention_mask in transformers is either LowerTriangular or all Zeros.
+        # We therefore check if one element in the upper triangular portion is zero. If it is, then the mask is all zeros.
+        if attention_mask is None or attention_mask[0, 0, 0, 1] == 0:
+            # input and output should be of form (bsz, q_len, num_heads, head_dim)
+            attn_output = xformers.ops.memory_efficient_attention(
+                query_states, key_states, value_states, attn_bias=None
+            )
+        else:
+            # input and output should be of form (bsz, q_len, num_heads, head_dim)
+            attn_output = xformers.ops.memory_efficient_attention(
+                query_states,
+                key_states,
+                value_states,
+                attn_bias=xformers.ops.LowerTriangularMask(),
+            )
+        attn_weights = None
     else:
-        # input and output should be of form (bsz, q_len, num_heads, head_dim)
-        attn_output = xformers.ops.memory_efficient_attention(
+        attn_weights = torch.matmul(
+            query_states, key_states.transpose(2, 3)
+        ) / math.sqrt(self.head_dim)
+
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz * self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+            attn_weights = attn_weights + attention_mask
+            attn_weights = torch.max(
+                attn_weights, torch.tensor(torch.finfo(attn_weights.dtype).min)
+            )
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(
+            attn_weights, dim=-1, dtype=torch.float32
+        ).to(query_states.dtype)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+        attn_output = attn_output.transpose(1, 2)
+
+    attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
+    attn_output = self.o_proj(attn_output)
+    return attn_output, attn_weights, past_key_value
+
+
+def sdp_attention_forward(
+    self,
+    hidden_states: torch.Tensor,
+    attention_mask: Optional[torch.Tensor] = None,
+    position_ids: Optional[torch.LongTensor] = None,
+    past_key_value: Optional[Tuple[torch.Tensor]] = None,
+    output_attentions: bool = False,
+    use_cache: bool = False,
+) -> Tuple[torch.Tensor, Optional[torch.Tensor], Optional[Tuple[torch.Tensor]]]:
+    # pylint: disable=duplicate-code
+    bsz, q_len, _ = hidden_states.size()
+
+    query_states = (
+        self.q_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    key_states = (
+        self.k_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+    value_states = (
+        self.v_proj(hidden_states)
+        .view(bsz, q_len, self.num_heads, self.head_dim)
+        .transpose(1, 2)
+    )
+
+    kv_seq_len = key_states.shape[-2]
+    if past_key_value is not None:
+        kv_seq_len += past_key_value[0].shape[-2]
+    cos, sin = self.rotary_emb(value_states, seq_len=kv_seq_len)
+    (
+        query_states,
+        key_states,
+    ) = transformers.models.llama.modeling_llama.apply_rotary_pos_emb(
+        query_states, key_states, cos, sin, position_ids
+    )
+    # [bsz, nh, t, hd]
+
+    if past_key_value is not None:
+        # reuse k, v, self_attention
+        key_states = torch.cat([past_key_value[0], key_states], dim=2)
+        value_states = torch.cat([past_key_value[1], value_states], dim=2)
+
+    past_key_value = (key_states, value_states) if use_cache else None
+
+    # We only apply sdp attention if we don't need to output the whole attention matrix
+    if not output_attentions:
+        attn_output = torch.nn.functional.scaled_dot_product_attention(
             query_states,
             key_states,
             value_states,
-            # attn_bias=attention_mask,
-            attn_bias=xformers.ops.LowerTriangularMask(),
+            attn_mask=attention_mask,
+            is_causal=False,
         )
+        attn_weights = None
+    else:
+        attn_weights = torch.matmul(
+            query_states, key_states.transpose(2, 3)
+        ) / math.sqrt(self.head_dim)
 
-    if attn_output.size() != (bsz, q_len, self.num_heads, self.head_dim):
-        raise ValueError(
-            f"`attn_output` should be of size {(bsz, q_len, self.num_heads, self.head_dim)}, but is"
-            f" {attn_output.size()}"
-        )
+        if attn_weights.size() != (bsz, self.num_heads, q_len, kv_seq_len):
+            raise ValueError(
+                f"Attention weights should be of size {(bsz * self.num_heads, q_len, kv_seq_len)}, but is"
+                f" {attn_weights.size()}"
+            )
+
+        if attention_mask is not None:
+            if attention_mask.size() != (bsz, 1, q_len, kv_seq_len):
+                raise ValueError(
+                    f"Attention mask should be of size {(bsz, 1, q_len, kv_seq_len)}, but is {attention_mask.size()}"
+                )
+            attn_weights = attn_weights + attention_mask
+            attn_weights = torch.max(
+                attn_weights, torch.tensor(torch.finfo(attn_weights.dtype).min)
+            )
+
+        # upcast attention to fp32
+        attn_weights = nn.functional.softmax(
+            attn_weights, dim=-1, dtype=torch.float32
+        ).to(query_states.dtype)
+        attn_output = torch.matmul(attn_weights, value_states)
+
+        if attn_output.size() != (bsz, self.num_heads, q_len, self.head_dim):
+            raise ValueError(
+                f"`attn_output` should be of size {(bsz, self.num_heads, q_len, self.head_dim)}, but is"
+                f" {attn_output.size()}"
+            )
+
+    attn_output = attn_output.transpose(1, 2)
     attn_output = attn_output.reshape(bsz, q_len, self.hidden_size)
 
-    #
-    # xformers-attn end
-    #
+    attn_output = self.o_proj(attn_output)
 
-    if self.pretraining_tp > 1:
-        attn_output = attn_output.split(self.hidden_size // self.pretraining_tp, dim=2)
-        o_proj_slices = self.o_proj.weight.split(
-            self.hidden_size // self.pretraining_tp, dim=1
-        )
-        attn_output = sum(
-            F.linear(attn_output[i], o_proj_slices[i])
-            for i in range(self.pretraining_tp)
-        )
-    else:
-        attn_output = self.o_proj(attn_output)
-
-    return attn_output, None, past_key_value
+    return attn_output, attn_weights, past_key_value

src/axolotl/monkeypatch/llama_expand_mask.py

@@ -1,52 +0,0 @@
-"""
-expands the binary attention mask per 3.2.2 of https://arxiv.org/pdf/2107.02027.pdf
-"""
-from typing import Optional
-
-import torch
-
-
-def _expand_mask(mask: torch.Tensor, dtype: torch.dtype, tgt_len: Optional[int] = None):
-    """
-    Expands attention_mask from `[bsz, seq_len]` to `[bsz, 1, tgt_seq_len, src_seq_len]`.
-    This expansion handles packed sequences so that sequences share the same attention mask integer value
-    when they attend to each other within that sequence.
-    This expansion transforms the mask to lower triangular form to prevent future peeking.
-    """
-    bsz, src_len = mask.size()
-    tgt_len = tgt_len if tgt_len is not None else src_len
-
-    mask = mask.unsqueeze(1).unsqueeze(2)
-    mask = mask.expand(bsz, 1, tgt_len, src_len)
-
-    # Create a binary mask from the original mask where zeros remain zeros and all other values are set to one
-    binary_mask = torch.where(
-        mask != 0,
-        torch.tensor(1).to(dtype),
-        torch.tensor(0).to(dtype),
-    )
-
-    # Create a block-diagonal mask.
-    # we multiply by the binary mask so that 0's in the original mask are correctly excluded
-    zero_one_mask = torch.eq(mask, mask.transpose(-1, -2)).int() * binary_mask
-
-    # Now let's create a lower triangular mask of ones that will zero out the upper triangular part
-    lower_triangular_ones = torch.tril(torch.ones((tgt_len, src_len), dtype=dtype)).to(
-        mask.device
-    )
-
-    # Use the lower triangular mask to zero out the upper triangular part of the zero_one_mask
-    masked_zero_one_mask = zero_one_mask * lower_triangular_ones
-    inverted_mask = 1.0 - masked_zero_one_mask
-
-    return inverted_mask.masked_fill(
-        inverted_mask.to(torch.bool), torch.finfo(dtype).min
-    )
-
-
-def hijack_expand_mask():
-    import transformers
-
-    transformers.models.llama.modeling_llama._expand_mask = (  # pylint: disable=protected-access
-        _expand_mask
-    )

src/axolotl/monkeypatch/llama_landmark_attn.py

@@ -53,7 +53,7 @@ from transformers.utils import (
     replace_return_docstrings,
 )
 
-LOG = logging.getLogger("axolotl")
+logger = logging.get_logger(__name__)
 
 _CONFIG_FOR_DOC = "LlamaConfig"
 
@@ -862,7 +862,7 @@ class LlamaModel(LlamaPreTrainedModel):
 
         if self.gradient_checkpointing and self.training:
             if use_cache:
-                LOG.warning_once(
+                logger.warning_once(
                     "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
                 )
                 use_cache = False

src/axolotl/monkeypatch/relora.py

@@ -1,393 +0,0 @@
-"""Implements the ReLoRA training procedure from https://arxiv.org/abs/2307.05695, minus the initial full fine-tune."""
-import glob
-import json
-import logging
-import os.path
-import shutil
-from pathlib import Path
-from typing import Dict, List, Sequence
-
-import bitsandbytes as bnb
-import peft
-import safetensors.torch as st
-import torch
-from huggingface_hub import snapshot_download
-from torch.optim.lr_scheduler import LRScheduler
-from torch.optim.optimizer import Optimizer
-from transformers import (
-    TrainerCallback,
-    TrainerControl,
-    TrainerState,
-    TrainingArguments,
-)
-from transformers.trainer_utils import PREFIX_CHECKPOINT_DIR
-
-from axolotl.utils.dict import DictDefault
-from axolotl.utils.distributed import is_main_process
-
-LOG = logging.getLogger("axolotl.relora")
-
-
-def reset_optimizer(optimizer: torch.optim.Optimizer):
-    for group in optimizer.param_groups:
-        for param in group["params"]:
-            param_state = optimizer.state[param]
-            for key in param_state:
-                if "qmap" in key:
-                    continue
-
-                if key == "step" and isinstance(param_state[key], int):
-                    param_state[key] = 0
-                else:
-                    param_state[key] = torch.zeros_like(param_state[key])
-
-
-class ReLoRACallback(TrainerCallback):
-    """Callback to merge LoRA weights into the base model and save full-weight checkpoints"""
-
-    def __init__(self, cfg: DictDefault):
-        self.relora_steps = cfg.relora_steps
-        self.cpu_offload = cfg.relora_cpu_offload
-        self.quantized = cfg.load_in_4bit or cfg.load_in_8bit
-        self.last_full_model = cfg.base_model
-        self.resume_from_checkpoint = cfg.resume_from_checkpoint
-
-        if not os.path.exists(self.last_full_model):
-            self.last_full_model = str(Path(snapshot_download(cfg.base_model)))
-
-        assert os.path.exists(
-            self.last_full_model
-        ), "for ReLORA base_model must be a local path"
-
-        self.num_lora_restarts = 0
-        self.need_full_save = False
-
-    def on_train_begin(
-        self,
-        _args: TrainingArguments,
-        _state: TrainerState,
-        control: TrainerControl,
-        model: peft.LoraModel,
-        **_kwargs,
-    ):
-        if self.resume_from_checkpoint:
-            weight_path = os.path.join(self.resume_from_checkpoint, "relora")
-            if not os.path.exists(weight_path):
-                LOG.warning(
-                    "Resuming ReLoRA from checkpoint, but no full-weight save found"
-                )
-            else:
-                LOG.info(f"Loading adjusted base weights from {weight_path}")
-                load_weight_checkpoint(model, weight_path)
-        return control
-
-    def on_step_begin(
-        self,
-        args: TrainingArguments,
-        state: TrainerState,
-        control: TrainerControl,
-        model: peft.LoraModel,
-        optimizer: torch.optim.Optimizer,
-        **_kwargs,
-    ):
-        if state.global_step > 0 and state.global_step % self.relora_steps == 0:
-            checkpoint_folder = os.path.join(
-                args.output_dir,
-                f"{PREFIX_CHECKPOINT_DIR}-{state.global_step}",
-                "relora",
-            )
-
-            with torch.no_grad():
-                merge_and_save(
-                    model,
-                    self.last_full_model,
-                    checkpoint_folder,
-                    reinit=True,
-                    quantized=self.quantized,
-                    actually_save=is_main_process(),
-                    cpu_offload=self.cpu_offload,
-                )
-                reset_optimizer(optimizer)
-
-            if self.quantized:
-                self.last_full_model = checkpoint_folder
-            self.num_lora_restarts += 1
-
-        return control
-
-    def on_save(
-        self,
-        args: TrainingArguments,
-        state: TrainerState,
-        control: TrainerControl,
-        model: peft.LoraModel,
-        **_kwargs,
-    ):
-        checkpoint_folder = os.path.join(
-            args.output_dir, f"{PREFIX_CHECKPOINT_DIR}-{state.global_step}", "relora"
-        )
-        if (
-            state.global_step >= self.relora_steps
-            and state.global_step % self.relora_steps != 0
-        ):
-            if self.quantized:
-                if is_main_process() and self.last_full_model != checkpoint_folder:
-                    # ensure the latest full parameter save is in the latest checkpoint
-                    # folder, so that automatic pruning of checkpoints does not remove it
-                    LOG.info(f"moving last full parameter save to {checkpoint_folder}")
-                    os.makedirs(checkpoint_folder, exist_ok=True)
-                    chunks = glob.glob(
-                        f"{self.last_full_model}/model*.safetensors"
-                    ) + glob.glob(f"{self.last_full_model}/model*.index.json")
-                    for path in chunks:
-                        new_path = os.path.abspath(shutil.move(path, checkpoint_folder))
-                        try:
-                            os.symlink(new_path, path)
-                        except OSError:
-                            # probably on windows without permission to symlink
-                            pass
-
-                    self.last_full_model = checkpoint_folder
-            else:
-                model.model.save_pretrained(checkpoint_folder, safe_serialization=True)
-
-        return control
-
-    def on_log(
-        self,
-        _args: TrainingArguments,
-        _state: TrainerState,
-        control: TrainerControl,
-        logs: Dict[str, float],
-        **_kwargs,
-    ):
-        logs["num_lora_restarts"] = self.num_lora_restarts
-        return control
-
-    def on_train_end(
-        self,
-        args: TrainingArguments,
-        _state: TrainerState,
-        control: TrainerControl,
-        model: peft.LoraModel,
-        **_kwargs,
-    ):
-        if self.quantized:
-            # perform final merge and save
-            with torch.no_grad():
-                merge_and_save(
-                    model,
-                    self.last_full_model,
-                    args.output_dir,
-                    reinit=False,
-                    quantized=self.quantized,
-                    actually_save=is_main_process(),
-                    cpu_offload=self.cpu_offload,
-                )
-        # no need to save if unquantized, as finetune.py will call merge_and_unload()
-        return control
-
-
-class ReLoRAScheduler(LRScheduler):
-    """Wraps another scheduler to apply per-lora-restart learning rate warmups."""
-
-    def __init__(
-        self,
-        optimizer: Optimizer,
-        inner_schedule: LRScheduler,
-        relora_steps: int,
-        warmup_steps: int,
-        min_lr_scale: float = 0.001,
-    ) -> None:
-        self.inner_schedule = inner_schedule
-        self.relora_steps = relora_steps
-        self.warmup_steps = warmup_steps
-        self.min_lr_scale = min_lr_scale
-        super().__init__(optimizer, inner_schedule.last_epoch, inner_schedule.verbose)
-
-    def get_lr(self) -> float:
-        self.inner_schedule.last_epoch = self.last_epoch
-
-        original = self.inner_schedule.get_lr()
-        step = self.last_epoch
-        if step < self.relora_steps:
-            scale = 1
-        else:
-            cycle_t = min(1.0, (step % self.relora_steps) / self.warmup_steps)
-            scale = cycle_t * (1 - self.min_lr_scale) + self.min_lr_scale
-
-        if isinstance(original, Sequence):
-            return [lr * scale for lr in original]
-        return original * scale
-
-
-def sharded_paths(path: str, module_names: List[str]) -> Dict[str, str]:
-    model_name = "model.safetensors"
-    if not os.path.exists(str(Path(path) / model_name)) and not os.path.exists(
-        str(Path(path) / f"{model_name}.index.json")
-    ):
-        model_name = "pytorch_model.bin"
-
-    index_path = str(Path(path) / f"{model_name}.index.json")
-    if os.path.exists(index_path):
-        with open(index_path, "r", encoding="utf-8") as file:
-            data = json.load(file)
-        return data["weight_map"]
-    return {(module_name + ".weight"): model_name for module_name in module_names}
-
-
-def lora_delta_weight(layer: peft.tuners.lora.LoraLayer, device) -> torch.Tensor:
-    if isinstance(layer, (peft.tuners.lora.Linear8bitLt, peft.tuners.lora.Linear4bit)):
-        adapter = layer.active_adapter
-        return (
-            peft.utils.transpose(
-                layer.lora_B[adapter].weight.detach().to(device)
-                @ layer.lora_A[adapter].weight.detach().to(device),
-                getattr(layer, "fan_in_fan_out", False),
-            )
-            * layer.scaling[adapter]
-        )
-
-    return layer.get_delta_weight().to(device)
-
-
-def find_lora_modules(model: peft.LoraModel) -> Dict[str, peft.tuners.lora.LoraLayer]:
-    modules: Dict[str, peft.tuners.lora.LoraLayer] = {}
-
-    key_list = [key for key, _ in model.model.named_modules() if "lora" not in key]
-    for key in key_list:
-        try:
-            # pylint: disable=protected-access
-            _parent, target, _target_name = peft.utils._get_submodules(model.model, key)
-        except AttributeError:
-            continue
-
-        if isinstance(target, peft.tuners.lora.LoraLayer):
-            modules[key] = target
-
-    return modules
-
-
-def update_weights(
-    target: peft.tuners.lora.LoraLayer, new_weight: torch.Tensor, reinit: bool, device
-):
-    if reinit:
-        for adapter_name in target.lora_A:
-            target.reset_lora_parameters(adapter_name)
-        for adapter_name in target.lora_embedding_A:
-            target.reset_lora_parameters(adapter_name)
-
-    if isinstance(target, peft.tuners.lora.Linear4bit):
-        # This could be faster, but the quantization of Linear4bit weights occurs
-        # when the module is moved from cpu to gpu. Without meddling *too* deeply in
-        # PEFT's innards or maintaining a duplicate of that codepath, this is good
-        # enough for now.
-        target.weight.quant_state = None
-        target.weight.data = new_weight.cpu()
-        target.to(device)
-    elif isinstance(target, peft.tuners.lora.Linear8bitLt):
-        target.weight = bnb.nn.Int8Params(new_weight, requires_grad=False).to(device)
-    else:
-        target.weight.data = new_weight.to(device)
-
-
-def merge_and_save(
-    model: peft.LoraModel,
-    model_src: str,
-    model_dst: str,
-    reinit: bool = False,
-    quantized: bool = False,
-    cpu_offload: bool = False,
-    actually_save: bool = True,
-):
-    modules = find_lora_modules(model)
-
-    if not quantized:
-        for module_name, target in modules.items():
-            update = target.get_delta_weight(target.active_adapter).detach()
-            target.weight.data += update
-
-            if reinit:
-                for adapter_name in target.lora_A:
-                    target.reset_lora_parameters(adapter_name)
-                for adapter_name in target.lora_embedding_A:
-                    target.reset_lora_parameters(adapter_name)
-        return
-
-    os.makedirs(model_dst, exist_ok=True)
-    shard_paths = sharded_paths(model_src, modules.keys())
-    out_shard_paths = {}
-
-    unique_shards = list(set(shard_paths.values()))
-    for shard_path in unique_shards:
-        out_tensors = {}
-        if shard_path.endswith(".safetensors"):
-            in_tensors = st.load_file(str(Path(model_src) / shard_path))
-        else:
-            in_tensors = torch.load(Path(model_src) / shard_path)
-            if "state_dict" in in_tensors:
-                in_tensors = in_tensors["state_dict"]
-
-        for module_name, target in modules.items():
-            key = module_name + ".weight"
-            if key not in shard_paths or shard_paths[key] != shard_path:
-                continue
-
-            orig_weight = in_tensors[key]
-            old_dev = target.weight.device
-            math_dev = "cpu" if cpu_offload else old_dev
-
-            delta_weight = lora_delta_weight(target, math_dev)
-            new_weight = orig_weight.to(math_dev) + delta_weight
-            del delta_weight
-
-            if actually_save:
-                out_tensors[key] = new_weight.half().cpu()
-
-            update_weights(target, new_weight, reinit=reinit, device=old_dev)
-
-        if actually_save:
-            out_shard_name = shard_path
-            if out_shard_name.startswith("pytorch_model"):
-                out_shard_name = (
-                    out_shard_name.replace("pytorch_model", "model").rstrip(".bin")
-                    + ".safetensors"
-                )
-
-            for module_name in in_tensors:
-                if module_name not in out_tensors:
-                    out_tensors[module_name] = in_tensors[module_name].half()
-                out_shard_paths[module_name] = out_shard_name
-
-            shard_fn = str(Path(model_dst) / out_shard_name)
-            LOG.info(f"saving tensors to {shard_fn}")
-            st.save_file(out_tensors, shard_fn, metadata={"format": "pt"})
-
-        del in_tensors
-        del out_tensors
-        torch.cuda.empty_cache()
-
-    if actually_save and len(unique_shards) > 1:
-        with open(
-            str(Path(model_dst, "model.safetensors.index.json")), "w", encoding="utf-8"
-        ) as file:
-            json.dump({"metadata": {}, "weight_map": out_shard_paths}, file)
-
-
-def load_weight_checkpoint(model: peft.LoraModel, checkpoint_path: str):
-    modules = find_lora_modules(model)
-    shard_paths = sharded_paths(checkpoint_path, modules.keys())
-    unique_shards = list(set(shard_paths.values()))
-
-    for shard_path in unique_shards:
-        tensors = st.load_file(os.path.join(checkpoint_path, shard_path))
-
-        for module_name, target in modules.items():
-            key = module_name + ".weight"
-            if key not in shard_paths or shard_paths[key] != shard_path:
-                continue
-
-            new_weight = tensors[key]
-            update_weights(
-                target, new_weight, reinit=False, device=target.weight.device
-            )

src/axolotl/monkeypatch/utils.py

@@ -1,103 +0,0 @@
-"""
-Shared utils for the monkeypatches
-"""
-import torch
-
-
-def get_cu_seqlens(attn_mask):
-    """generate a cumulative sequence length mask for flash attention using attn mask"""
-    if len(attn_mask.shape) == 1:
-        attn_mask = attn_mask.unsqueeze(0)
-
-    device = attn_mask.device
-    results = []
-    max_seq_lens = []
-
-    for row in attn_mask:
-        # Exclude zeros to avoid adding their positions to the mask
-        t_non_zeros = row[row != 0]
-        # Find where the sequence number changes (including the first position)
-        seq_change = torch.cat(
-            [
-                torch.tensor([1], dtype=torch.int32, device=device),
-                t_non_zeros[1:] != t_non_zeros[:-1],
-            ]
-        )
-        # Get the indices where the sequence changes
-        change_indices = torch.cat(
-            [
-                (seq_change == 1).nonzero(as_tuple=True)[0],
-                torch.tensor([len(t_non_zeros)], dtype=torch.int32, device=device),
-            ]
-        )
-        # Calculate the sequence lengths
-        seq_lengths = change_indices[1:] - change_indices[:-1]
-        # Calculate the length of the final sequence or padding
-        final_seq_length = len(row) - change_indices[-1]
-        # Append the length of the final sequence or padding to seq_lengths
-        if final_seq_length.item():
-            seq_lengths = torch.cat(
-                [
-                    seq_lengths,
-                    torch.tensor(
-                        [final_seq_length.item()], dtype=torch.int32, device=device
-                    ),
-                ]
-            )
-        # Calculate the cumulative sequence lengths
-        cu_seqlens = torch.cat(
-            [torch.tensor([0], dtype=torch.int32, device=device), seq_lengths.cumsum(0)]
-        )
-        max_seq_len = (cu_seqlens[1:] - cu_seqlens[:-1]).max()
-        results.append(cu_seqlens)
-        max_seq_lens.append(max_seq_len)
-
-    return torch.stack(results).to(dtype=torch.int32), torch.stack(max_seq_lens)
-
-
-def get_cu_seqlens_from_pos_ids(position_ids):
-    """generate a cumulative sequence length mask for flash attention using pos ids"""
-    if len(position_ids.shape) == 1:
-        position_ids = position_ids.unsqueeze(0)
-
-    device = position_ids.device
-    results = []
-    max_seq_lens = []
-
-    for row in position_ids:
-        # Count the number of consecutive zeros from the right side
-        padding_length = (row == 0).int().flip(dims=[0]).cumprod(dim=0).sum().item()
-
-        # Adjust the row to exclude padding
-        adjusted_row = row[:-padding_length] if padding_length else row.clone()
-
-        # Find where the position resets to 0 (indicating a new sequence)
-        seq_starts = torch.cat(
-            [
-                torch.tensor([True], dtype=torch.bool, device=device),
-                adjusted_row[1:] == 0,
-            ]
-        )
-        # Get the indices where the sequence starts
-        start_indices = torch.cat(
-            [
-                (seq_starts).nonzero(as_tuple=True)[0],
-                torch.tensor([len(adjusted_row)], dtype=torch.int32, device=device),
-            ]
-        )
-        # Calculate the sequence lengths
-        seq_lengths = start_indices[1:] - start_indices[:-1]
-        # Calculate the cumulative sequence lengths
-        cu_seqlens = torch.cat(
-            [torch.tensor([0], dtype=torch.int32, device=device), seq_lengths.cumsum(0)]
-        )
-        # Append the padding length to the cumulative sequence lengths
-        if padding_length:
-            cu_seqlens = torch.cat(
-                [cu_seqlens, torch.tensor([len(row)], dtype=torch.int32, device=device)]
-            )
-        max_seq_len = (cu_seqlens[1:] - cu_seqlens[:-1]).max()
-        results.append(cu_seqlens)
-        max_seq_lens.append(max_seq_len)
-
-    return torch.stack(results).to(dtype=torch.int32), torch.stack(max_seq_lens)

src/axolotl/prompt_strategies/__init__.py

@@ -1,12 +1,9 @@
 """Module to load prompt strategies."""
 
 import importlib
-import inspect
-
-from axolotl.prompt_strategies.user_defined import UserDefinedDatasetConfig
 
 
-def load(strategy, tokenizer, cfg, ds_cfg):
+def load(strategy, tokenizer, cfg):
     try:
         load_fn = "load"
         if strategy.split(".")[-1].startswith("load_"):
@@ -14,13 +11,6 @@ def load(strategy, tokenizer, cfg, ds_cfg):
             strategy = ".".join(strategy.split(".")[:-1])
         mod = importlib.import_module(f".{strategy}", "axolotl.prompt_strategies")
         func = getattr(mod, load_fn)
-        load_kwargs = {}
-        if strategy == "user_defined":
-            load_kwargs["ds_cfg"] = UserDefinedDatasetConfig(**ds_cfg)
-        else:
-            sig = inspect.signature(func)
-            if "ds_cfg" in sig.parameters:
-                load_kwargs["ds_cfg"] = ds_cfg
-        return func(tokenizer, cfg, **load_kwargs)
+        return func(tokenizer, cfg)
     except Exception:  # pylint: disable=broad-exception-caught
         return None

src/axolotl/prompt_strategies/alpaca_w_system.py

@@ -57,8 +57,6 @@ class SystemDataPrompter(AlpacaPrompter):
     Alpaca Style Prompter that uses system prompts from the dataset
     """
 
-    system_format: str = "### System:\n{system}\n\n"
-
     def build_prompt_w_system(
         self,
         system: str,
@@ -68,47 +66,15 @@ class SystemDataPrompter(AlpacaPrompter):
     ) -> Generator[str, None, None]:
         # returns the full prompt from instruction and optional input
         # if a label (=response, =output) is provided, it's also appended.
-        formatted_sys_prompt = (
-            self.system_format.format(system=system)
-            if system and self.system_format
-            else ""
-        )
         if input:
-            res = formatted_sys_prompt + self.turn_format.format(
-                instruction=instruction, input=input
-            )
+            res = system + self.turn_format.format(instruction=instruction, input=input)
         else:
-            res = formatted_sys_prompt + self.turn_no_input_format.format(
-                instruction=instruction
-            )
+            res = system + self.turn_no_input_format.format(instruction=instruction)
         if output:
             res = f"{res}{output}"
         yield res
 
 
-class OpenOrcaSystemDataPrompter(SystemDataPrompter):
-    """
-    Alpaca Style Prompter that uses system prompts from the dataset, with OpenOrca prompts
-    """
-
-    def match_prompt_style(self):
-        # pylint: disable=duplicate-code
-        if self.prompt_style == PromptStyle.INSTRUCT.value:
-            self.turn_format = "### Human:\n{instruction}\n### Additional Context:\n{input}\n### Assistant:\n"
-            self.turn_no_input_format = "### Human:\n{instruction}\n### Assistant:\n"
-            self.system_format = "### System:\n{system}\n"
-        if self.prompt_style == PromptStyle.CHAT.value:
-            self.turn_format = "USER: {instruction}\n{input}\nASSISTANT:"
-            self.turn_no_input_format = "USER: {instruction}\nASSISTANT:"
-            self.system_format = "SYSTEM: {system}\n"
-        if self.prompt_style == PromptStyle.CHATML.value:
-            self.turn_format = "<|im_start|>user\n{instruction}\n{input}<|im_end|>\n<|im_start|>assistant\n"
-            self.turn_no_input_format = (
-                "<|im_start|>user\n{instruction}<|im_end|>\n<|im_start|>assistant\n"
-            )
-            self.system_format = "<|im_start|>system\n{system}<|im_end|>\n"
-
-
 class OpenOrcaPromptTokenizingStrategy(InstructionWSystemPromptTokenizingStrategy):
     """
     Tokenizing strategy for OpenOrca datasets
@@ -147,16 +113,7 @@ def load_chat(tokenizer, cfg):
 
 def load_open_orca(tokenizer, cfg):
     return OpenOrcaPromptTokenizingStrategy(
-        OpenOrcaSystemDataPrompter(PromptStyle.INSTRUCT.value),
-        tokenizer,
-        cfg.train_on_inputs,
-        cfg.sequence_len,
-    )
-
-
-def load_open_orca_chatml(tokenizer, cfg):
-    return OpenOrcaPromptTokenizingStrategy(
-        OpenOrcaSystemDataPrompter(PromptStyle.CHATML.value),
+        SystemDataPrompter(PromptStyle.INSTRUCT.value),
         tokenizer,
         cfg.train_on_inputs,
         cfg.sequence_len,

src/axolotl/prompt_strategies/completion.py

@@ -1,20 +0,0 @@
-"""
-Basic completion text
-"""
-from typing import Any, Dict, Optional
-
-from axolotl.prompt_tokenizers import CompletionPromptTokenizingStrategy
-from axolotl.prompters import CompletionPrompter
-
-
-def load(tokenizer, cfg, ds_cfg: Optional[Dict[str, Any]] = None):
-    strat = CompletionPromptTokenizingStrategy(
-        CompletionPrompter(),
-        tokenizer,
-        cfg.train_on_inputs,
-        cfg.sequence_len,
-    )
-    if ds_cfg and "field" in ds_cfg:
-        strat.field = ds_cfg["field"]
-
-    return strat

src/axolotl/prompt_strategies/llama2_chat.py

@@ -1,205 +0,0 @@
-"""
-Prompt Strategy for finetuning Llama2 chat models
-see also https://github.com/facebookresearch/llama/blob/6c7fe276574e78057f917549435a2554000a876d/llama/generation.py#L213 for ma reference implementation.
-
-This implementation is based on the Vicuna PR and the fastchat repo, see also:
-https://github.com/lm-sys/FastChat/blob/cdd7730686cb1bf9ae2b768ee171bdf7d1ff04f3/fastchat/conversation.py#L847
-
-Use dataset type: "llama2_chat" in conig.yml to use this prompt style.
-
-E.g. in the config.yml:
-```
-datasets:
-  - path: llama_finetune_train.jsonl
-    type: llama2_chat
-```
-
-The dataset itself should look like this:
-```
-{'conversations':[{"from": "human", "value": "Who are you?"}, {"from": "gpt", "value": "I am Vicuna"},...]}
-```
-in a jsonl file. The first message should be from the human, the second from gpt.
-For a custom system message, the first "from" can be "system" (followed by alternating "human" and "gpt" turns).
-
-Important: Don't use "special_tokens:" in your config.yml if you are not sure what you are doing!
-"""
-
-import logging
-from dataclasses import dataclass, field
-from typing import Generator, List, Sequence
-
-from axolotl.prompt_tokenizers import PromptTokenizingStrategy
-from axolotl.prompters import IGNORE_TOKEN_ID, SHAREGPT_ASSERTION_FAILED_ROLE
-
-
-@dataclass
-class Llama2ChatConversation:
-    """A class that manages prompt templates and keeps all conversation history.
-    copied from https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py"""
-
-    name: str = "llama2"
-    # The system prompt
-    system: str = (
-        "[INST] <<SYS>>\nYou are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. "
-        "Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. "
-        "Please ensure that your responses are socially unbiased and positive in nature.\n\n"
-        "If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. "
-        "If you don't know the answer to a question, please don't share false information.\n<</SYS>>\n\n"
-    )
-    roles: Sequence[str] = ("[INST]", "[/INST]")
-    messages: List[List[str]] = field(default_factory=list)
-    offset: int = 0
-    sep = " "
-    sep2 = " </s><s>"
-    stop_token_ids = [2]
-
-    def get_prompt(self) -> str:
-        """Get the prompt for generation."""
-        seps = [self.sep, self.sep2]
-        ret = ""
-        for i, (role, message) in enumerate(self.messages):
-            if (i == len(self.messages) - 1) and (role == self.roles[0]):
-                # last message is from user (due to length),
-                #  return prompt without it for training
-                return ret
-            if i == 0:
-                ret += self.system + message.strip()
-            else:
-                ret += role + " " + message.strip() + seps[i % 2]
-        return ret
-
-    def append_message(self, role: str, message: str):
-        """Append a new message."""
-        self.messages.append([role, message])
-
-
-class LLama2ChatTokenizingStrategy(PromptTokenizingStrategy):
-    """
-    Tokenizing strategy for ShareGPT prompts.
-    adapted from https://github.com/lm-sys/FastChat/blob/main/fastchat/train/train.py
-    """
-
-    def __init__(self, *args, **kwargs):
-        super().__init__(*args, **kwargs)
-        self.sequence_len = 4096
-        self.tokenizer.add_special_tokens({"pad_token": "<pad>"})
-        # https://huggingface.co/meta-llama/Llama-2-7b-chat-hf/blob/main/added_tokens.json
-
-    def tokenize_prompt(self, prompt):
-        conv = next(self.prompter.build_prompt(prompt))
-        conversation_str = conv.get_prompt()
-
-        # Tokenize conversations
-        input_ids = self.tokenizer(
-            conversation_str,
-            return_tensors="pt",
-            padding="max_length",
-            max_length=self.sequence_len,
-            truncation=True,
-        ).input_ids[0]
-        target = input_ids.clone()
-
-        # Mask targets. Only compute loss on the assistant outputs.
-        sep = conv.roles[1]
-
-        total_len = int(target.ne(self.tokenizer.pad_token_id).sum())
-
-        turns = conversation_str.split(conv.sep2)
-        cur_len = 1
-        target[:cur_len] = IGNORE_TOKEN_ID
-        for turn in turns:
-            if turn == "":
-                break
-            turn_len = len(self.tokenizer(turn).input_ids)
-
-            parts = turn.split(sep)
-            if len(parts) != 2:
-                break
-            parts[0] += sep
-            # "-1" is hardcoded for the LLaMA tokenizer to make the offset correct.
-            instruction_len = len(self.tokenizer(parts[0]).input_ids) - 1
-
-            # Ignore the user instructions
-            target[cur_len - 1 : cur_len + instruction_len] = IGNORE_TOKEN_ID
-            cur_len += turn_len + 2  # due to length of role token
-
-        target[cur_len:] = IGNORE_TOKEN_ID
-
-        if cur_len < self.sequence_len:
-            if cur_len != total_len:
-                target[:] = IGNORE_TOKEN_ID
-                logging.warning(
-                    f"WARNING: tokenization mismatch: {cur_len} vs. {total_len}."
-                    f" (ignored)"
-                )
-
-        attention_mask = input_ids.ne(self.tokenizer.pad_token_id).tolist()
-        input_ids = input_ids.tolist()
-        target = target.tolist()
-        # this is a fix for the tokenizer which tokenizes [ differently with eos tokens and
-        # follows the original llama implementation
-        for i in range(2, total_len - 2):
-            if input_ids[i] == 29961:
-                input_ids[i] = 518
-            if target[i] == 29961:
-                target[i] = 518
-        return {
-            "input_ids": input_ids,
-            "labels": target,
-            "attention_mask": attention_mask,
-        }
-
-
-class Llama2ChatPrompter:  # pylint: disable=too-few-public-methods
-    """
-    A prompter that generates prompts for Llama2 models.
-    """
-
-    system_prompt = (
-        "[INST] <<SYS>>\nYou are a helpful, respectful and honest assistant. Always answer as helpfully as possible, while being safe. "
-        "Your answers should not include any harmful, unethical, racist, sexist, toxic, dangerous, or illegal content. "
-        "Please ensure that your responses are socially unbiased and positive in nature.\n\n"
-        "If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. "
-        "If you don't know the answer to a question, please don't share false information.\n<</SYS>>\n\n"
-    )
-
-    def build_prompt(self, source) -> Generator[Llama2ChatConversation, None, None]:
-        # see https://github.com/lm-sys/FastChat/blob/da0641e567cf93756b0978ab5a6b092e96f06240/fastchat/train/train.py#L78
-        source = source["conversations"]  # fix data structure for datasets
-
-        # if system prompt provided, use it
-        if source[0]["from"] == "system":
-            system = f"[INST] <<SYS>>\n{source[0]['value']}\n<</SYS>>\n\n"
-            source = source[1:]
-        else:
-            system = self.system_prompt
-
-        conv = Llama2ChatConversation(system=system)
-
-        if len(source) < 2:
-            # If there isn't a back and forth conversation, ignore it
-            # also happens on the data splitting leaving empty conversations
-            raise IndexError
-
-        roles = {"human": conv.roles[0], "gpt": conv.roles[1]}
-
-        if roles[source[0]["from"]] != conv.roles[0]:
-            # Skip the first one if it is not from human
-            source = source[1:]
-
-        conv.messages = []  # pylint: disable=R0801
-        for j, sentence in enumerate(source):
-            role = roles[sentence["from"]]
-            assert role == conv.roles[j % 2], SHAREGPT_ASSERTION_FAILED_ROLE
-            if sentence["value"]:
-                conv.append_message(role, sentence["value"])
-        yield conv
-
-
-def load(tokenizer, cfg) -> LLama2ChatTokenizingStrategy:
-    return LLama2ChatTokenizingStrategy(
-        Llama2ChatPrompter(),
-        tokenizer,
-        cfg.train_on_inputs,
-        cfg.sequence_len,
-    )

src/axolotl/prompt_strategies/metharme.py

@@ -1,76 +0,0 @@
-"""Module containing the MetharmenPromptTokenizingStrategy and MetharmePrompter class"""
-
-import logging
-from typing import Tuple
-
-from axolotl.prompt_tokenizers import InstructionPromptTokenizingStrategy
-from axolotl.prompters import AlpacaPrompter
-
-LOG = logging.getLogger("axolotl")
-
-IGNORE_TOKEN_ID = -100
-
-# pylint: disable=duplicate-code
-
-
-class MetharmePromptTokenizingStrategy(InstructionPromptTokenizingStrategy):
-    """
-    Tokenizing strategy for the Metharme models
-    """
-
-    def parse_instruction_fields(self, prompt) -> Tuple[str, str, str]:
-        return (prompt["prompt"], "", prompt["generation"])
-
-    def _tokenize(
-        self,
-        prompt: str,
-        add_eos_token: bool = True,
-        strip_bos_token: bool = False,
-        num_eos_tokens: int = 3,
-    ):
-        result = self.tokenizer(
-            prompt,
-            truncation=True,
-            max_length=self.sequence_len,
-            padding=False,
-            return_tensors=None,
-        )
-        if len(result["input_ids"]) == 0:
-            LOG.warning("Tokenizer result is empty. You may want to audit your dataset")
-        # If there's already an EOS token there, subtract from the number added
-        if result["input_ids"][-1] == self.tokenizer.eos_token_id:
-            num_eos_tokens -= 1
-
-        if num_eos_tokens > 0 and add_eos_token and len(result["input_ids"]) > 0:
-            for _ in range(num_eos_tokens):
-                if len(result["input_ids"]) < self.sequence_len:
-                    result["input_ids"].append(self.tokenizer.eos_token_id)
-                    result["attention_mask"].append(1)
-
-        if result["input_ids"][0] == self.tokenizer.bos_token_id and strip_bos_token:
-            result["input_ids"] = result["input_ids"][1:]
-            result["attention_mask"] = result["attention_mask"][1:]
-
-        result["labels"] = result["input_ids"].copy()
-        return result
-
-
-class MetharmePrompter(AlpacaPrompter):
-    """
-    Prompter for the Metharme models.
-    """
-
-    system_prompt = ""
-    system_no_input_prompt = ""
-    system_format = ""
-    turn_format = "{instruction}"
-    turn_no_input_format = "{instruction}"
-
-    def __init__(self, *args, **kwargs):  # pylint: disable=super-init-not-called
-        pass
-
-
-def load(tokenizer, cfg):
-    return MetharmePromptTokenizingStrategy(
-        MetharmePrompter(), tokenizer, cfg.train_on_inputs, cfg.sequence_len
-    )

src/axolotl/prompt_strategies/orcamini.py

@@ -1,46 +0,0 @@
-"""
-Prompt Strategy for finetuning Orca Mini (v2) models
-see also https://huggingface.co/psmathur/orca_mini_v2_7b for more information
-
-Use dataset type: orcamini in conig.yml to use this prompt style.
-
-Compared to the alpaca_w_system.open_orca dataset type,
-this one specifies the system prompt with "### System:".
-
-Not suited/tested for multiple-turn conversations without further adjustments.
-"""
-from typing import Generator, Union
-
-from axolotl.prompt_strategies.alpaca_w_system import OpenOrcaPromptTokenizingStrategy
-from axolotl.prompters import AlpacaPrompter
-
-
-class OrcaMiniPrompter(AlpacaPrompter):
-    """Adjusted Prompter for Orca Mini (v2) datasets"""
-
-    def match_prompt_style(self):
-        self.turn_no_input_format = (
-            "### System:\n{system}\n\n### User:\n{instruction}\n\n### Response:\n"
-        )
-
-    def build_prompt_w_system(
-        self,
-        system: str,
-        instruction: str,
-        output: Union[None, str] = None,
-    ) -> Generator[str, None, None]:
-        # returns the full prompt from instruction and optional input
-        # if a label (=response, =output) is provided, it's also appended.
-        res = self.turn_no_input_format.format(system=system, instruction=instruction)
-        if output:
-            res = f"{res}{output}"
-        yield res
-
-
-def load(tokenizer, cfg):
-    return OpenOrcaPromptTokenizingStrategy(
-        OrcaMiniPrompter(),
-        tokenizer,
-        cfg.train_on_inputs,
-        cfg.sequence_len,
-    )

src/axolotl/prompt_strategies/pygmalion.py

@@ -11,8 +11,6 @@ from axolotl.prompt_tokenizers import (
     tokenize_prompt_default,
 )
 
-LOG = logging.getLogger("axolotl")
-
 IGNORE_TOKEN_ID = -100
 
 
@@ -66,7 +64,7 @@ class PygmalionPromptTokenizingStrategy(PromptTokenizingStrategy):
                     *copy.deepcopy(res["input_ids"])
                 ][len(self.bot_prefix_token_ids) :]
             else:
-                LOG.warning(f"unknown role in conversation: {role}")
+                logging.warning(f"unknown role in conversation: {role}")
                 res = defaultdict(lambda: [])
 
             # pylint: disable=duplicate-code

src/axolotl/prompt_strategies/user_defined.py

@@ -1,98 +0,0 @@
-"""
-User Defined prompts with configuration from the YML config
-"""
-
-from dataclasses import dataclass
-from functools import partial
-from typing import Optional, Tuple
-
-from axolotl.prompt_strategies.alpaca_w_system import (
-    InstructionWSystemPromptTokenizingStrategy,
-    SystemDataPrompter,
-)
-
-
-@dataclass
-class UserDefinedDatasetConfig:
-    """
-    dataclass configuration representing a userdefined dataset type
-    """
-
-    system_prompt: str = ""
-    field_system: str = "system"
-    field_instruction: str = "instruction"
-    field_input: str = "input"
-    field_output: str = "output"
-    format: str = "{instruction} {input} "
-    no_input_format: str = "{instruction} "
-    system_format: str = "{system}"
-
-    def __getitem__(self, item):
-        return getattr(self, item)
-
-
-class UserDefinedPromptTokenizationStrategy(InstructionWSystemPromptTokenizingStrategy):
-    """
-    Prompt Tokenization Strategy for user defined prompts
-    """
-
-
-def load(tokenizer, cfg, ds_cfg: Optional[UserDefinedDatasetConfig] = None):
-    if not ds_cfg:
-        raise ValueError("Missing dataset prompt configuration")
-
-    system_prompt = ""
-    if ds_cfg.system_prompt:
-        system_prompt = ds_cfg.system_prompt
-
-    def parse_instruction_fields(
-        field_instruction,
-        field_input,
-        field_output,
-        field_system,
-        system_prompt,
-        prompt,
-    ) -> Tuple[str, str, str, str]:
-        return (
-            prompt[field_instruction],
-            prompt[field_input] if field_input in prompt else "",
-            prompt[field_output] if field_output in prompt else "",
-            prompt[field_system] if field_system in prompt else system_prompt,
-        )
-
-    turn_format = ds_cfg.format
-    turn_no_input_format = ds_cfg.no_input_format
-    system_format = ds_cfg.system_format
-
-    class UserDefinedPrompter(SystemDataPrompter):
-        """
-        Prompter for user defined prompts
-        """
-
-        def match_prompt_style(self):
-            self.turn_format = turn_format
-            self.turn_no_input_format = turn_no_input_format
-            self.system_format = system_format
-
-    prompter = UserDefinedPrompter()
-
-    strat = UserDefinedPromptTokenizationStrategy(
-        prompter,
-        tokenizer,
-        cfg.train_on_inputs,
-        cfg.sequence_len,
-    )
-
-    setattr(
-        strat,
-        "parse_instruction_fields",
-        partial(
-            parse_instruction_fields,
-            ds_cfg.field_instruction,
-            ds_cfg.field_input,
-            ds_cfg.field_output,
-            ds_cfg.field_system,
-            system_prompt,
-        ),
-    )
-    return strat

src/axolotl/prompt_tokenizers.py

@@ -6,14 +6,12 @@ import functools
 import logging
 from typing import Dict, List, Tuple, Union
 
-from transformers import BatchEncoding, PreTrainedTokenizer
+from transformers import PreTrainedTokenizer
 
 from axolotl.prompters import IGNORE_TOKEN_ID
 
-LOG = logging.getLogger("axolotl")
-
 IGNORE_INDEX = -100
-LLAMA_DEFAULT_PAD_TOKEN = "<pad>"  # nosec
+LLAMA_DEFAULT_PAD_TOKEN = "[PAD]"  # nosec
 LLAMA_DEFAULT_EOS_TOKEN = "</s>"  # nosec
 LLAMA_DEFAULT_BOS_TOKEN = "<s>"  # nosec
 LLAMA_DEFAULT_UNK_TOKEN = "<unk>"  # nosec
@@ -48,55 +46,35 @@ class PromptTokenizingStrategy(abc.ABC):
 
     @functools.lru_cache(maxsize=128)
     def _get_user_token(self):
-        try:
-            id_or_ids = self.tokenizer.convert_tokens_to_ids("<|USER|>")
-            if isinstance(id_or_ids, (int,)):
-                return id_or_ids
-        except KeyError:
-            pass
+        id_or_ids = self.tokenizer.convert_tokens_to_ids("<|USER|>")
+        if isinstance(id_or_ids, (int,)):
+            return id_or_ids
         return False
 
     @functools.lru_cache(maxsize=128)
     def _get_assistant_token(self):
-        try:
-            id_or_ids = self.tokenizer.convert_tokens_to_ids("<|ASSISTANT|>")
-            if isinstance(id_or_ids, (int,)):
-                return id_or_ids
-        except KeyError:
-            pass
+        id_or_ids = self.tokenizer.convert_tokens_to_ids("<|ASSISTANT|>")
+        if isinstance(id_or_ids, (int,)):
+            return id_or_ids
         return False
 
-    def _tokenize(
-        self, prompt: str, add_eos_token: bool = True, strip_bos_token: bool = False
-    ) -> BatchEncoding:
-        result: BatchEncoding
-        if not prompt.strip():
-            LOG.warning("Empty text requested for tokenization.")
-            result = BatchEncoding(data={"input_ids": [], "attention_mask": []})
-        else:
-            result = self.tokenizer(
-                prompt,
-                truncation=True,
-                max_length=self.sequence_len,
-                padding=False,
-                return_tensors=None,
-            )
-        if len(result["input_ids"]) == 0:
-            LOG.warning("Tokenizer result is empty. You may want to audit your dataset")
+    def _tokenize(self, prompt: str, add_eos_token=True, strip_bos_token=False):
+        result = self.tokenizer(
+            prompt,
+            truncation=True,
+            max_length=self.sequence_len,
+            padding=False,
+            return_tensors=None,
+        )
         if (
-            len(result["input_ids"]) > 0
-            and result["input_ids"][-1] != self.tokenizer.eos_token_id
+            result["input_ids"][-1] != self.tokenizer.eos_token_id
             and len(result["input_ids"]) < self.sequence_len
             and add_eos_token
         ):
             result["input_ids"].append(self.tokenizer.eos_token_id)
             result["attention_mask"].append(1)
 
-        if (
-            len(result["input_ids"]) > 0
-            and result["input_ids"][0] == self.tokenizer.bos_token_id
-            and strip_bos_token
-        ):
+        if result["input_ids"][0] == self.tokenizer.bos_token_id and strip_bos_token:
             result["input_ids"] = result["input_ids"][1:]
             result["attention_mask"] = result["attention_mask"][1:]
 
@@ -252,31 +230,8 @@ class CompletionPromptTokenizingStrategy(InstructionPromptTokenizingStrategy):
     Tokenizing strategy for Completion prompts.
     """
 
-    _field: str = "text"
-
-    @property
-    def field(self) -> str:
-        return self._field
-
-    @field.setter
-    def field(self, new_field: str):
-        self._field = new_field
-
-    def parse_instruction_fields(self, prompt) -> Tuple[str, str, str]:
-        return (
-            prompt[self.field],
-            "",
-            "",
-        )
-
     def tokenize_prompt(self, prompt):
-        (
-            instruction,
-            _,
-            _,
-        ) = self.parse_instruction_fields(prompt)
-
-        full_prompt = self._build_full_prompt(instruction, None, None)
+        full_prompt = self._build_full_prompt(prompt["text"], None, None)
         tokenized_full_prompt = self._tokenize(full_prompt)
 
         return tokenized_full_prompt
@@ -429,7 +384,7 @@ class ShareGPTPromptTokenizingStrategy(PromptTokenizingStrategy):
                         # everything from this is masked out from the labels
                         labels = [IGNORE_TOKEN_ID] * len(res["input_ids"])
                     else:
-                        LOG.warning(f"unhandled role: {part[0]}")
+                        logging.warning(f"unhandled role: {part[0]}")
 
                 # pylint: disable=duplicate-code
                 result, current_len = parse_tokenized_to_result(

src/axolotl/prompters.py

@@ -5,7 +5,6 @@ import logging
 from enum import Enum, auto
 from typing import Generator, List, Optional, Tuple, Union
 
-LOG = logging.getLogger("axolotl")
 IGNORE_TOKEN_ID = -100
 
 
@@ -16,7 +15,6 @@ class PromptStyle(Enum):
 
     INSTRUCT = "instruct"
     CHAT = "chat"
-    CHATML = "chatml"
 
 
 class AlpacaPrompter:
@@ -26,7 +24,6 @@ class AlpacaPrompter:
 
     system_prompt = "Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.\n\n"
     system_no_input_prompt = "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n"
-    system_format: str = "{system}"
     turn_format: str
     turn_no_input_format: str
     prompt_style: Optional[PromptStyle] = None
@@ -36,23 +33,14 @@ class AlpacaPrompter:
         self.match_prompt_style()
 
     def match_prompt_style(self):
-        # pylint: disable=duplicate-code
         if self.prompt_style == PromptStyle.INSTRUCT.value:
             self.turn_format = "### Instruction:\n{instruction}\n\n### Input:\n{input}\n\n### Response:\n"
             self.turn_no_input_format = (
                 "### Instruction:\n{instruction}\n\n### Response:\n"
             )
-            self.system_format = "### System:\n{system}\n\n"
         if self.prompt_style == PromptStyle.CHAT.value:
             self.turn_format = "USER: {instruction}\n{input}\nASSISTANT:"
             self.turn_no_input_format = "USER: {instruction}\nASSISTANT:"
-            self.system_format = "SYSTEM: {system}\n"
-        if self.prompt_style == PromptStyle.CHATML.value:
-            self.turn_format = "<|im_start|>user\n{instruction}\n{input}<|im_end|>\n<|im_start|>assistant\n"
-            self.turn_no_input_format = (
-                "<|im_start|>user\n{instruction}<|im_end|>\n<|im_start|>assistant\n"
-            )
-            self.system_format = "<|im_start|>system\n{system}<|im_end|>\n"
 
     def build_prompt(
         self,
@@ -63,17 +51,13 @@ class AlpacaPrompter:
         # returns the full prompt from instruction and optional input
         # if a label (=response, =output) is provided, it's also appended.
         if input:
-            res = (
-                self.system_format.format(system=self.system_prompt)
-                if self.system_prompt
-                else ""
-            ) + self.turn_format.format(instruction=instruction, input=input)
+            res = self.system_prompt + self.turn_format.format(
+                instruction=instruction, input=input
+            )
         else:
-            res = (
-                self.system_format.format(system=self.system_no_input_prompt)
-                if self.system_prompt
-                else ""
-            ) + self.turn_no_input_format.format(instruction=instruction)
+            res = self.system_no_input_prompt + self.turn_no_input_format.format(
+                instruction=instruction
+            )
         if output:
             res = f"{res}{output}"
         yield res
@@ -257,7 +241,7 @@ class Conversation:
             if message:
                 yield (role + ":", " " + message)
             else:
-                LOG.warning(f"role with empty message: {role}")
+                logging.warning(f"role with empty message: {role}")
                 yield (role + ":", "")
 
     def copy(self):
@@ -275,11 +259,6 @@ class Conversation:
         self.messages.append([role, message])
 
 
-SHAREGPT_ASSERTION_FAILED_ROLE = (
-    "Role did not alternate between turns (gpt and human). Please check your data."
-)
-
-
 class ShareGPTPrompter:  # pylint: disable=too-few-public-methods
     """
     A prompter that generates prompts for the ShareGPT
@@ -309,20 +288,16 @@ class ShareGPTPrompter:  # pylint: disable=too-few-public-methods
         )
 
     def build_prompt(self, source) -> Generator[str, None, None]:
+        # ignore the system prompt if provided
+        if source[0]["from"] == "system":
+            source.pop(0)
+
         if len(source) < 2:
             # If there isn't a back and forth conversation, ignore it
             # also happens on the data splitting leaving empty conversations
-            raise IndexError(
-                f"A conversation entry has less than 2 messages :\n{source}"
-            )
+            raise IndexError
 
         conv = self._conversation.copy()
-
-        # Add the conversation system prompt if provided, otherwise use the default one
-        if source[0]["from"] == "system":
-            conv.system = source[0]["value"]
-            source.pop(0)
-
         roles = {"human": conv.roles[0], "gpt": conv.roles[1]}
 
         try:
@@ -340,7 +315,7 @@ class ShareGPTPrompter:  # pylint: disable=too-few-public-methods
         conv.messages = []
         for j, sentence in enumerate(source):
             role = roles[sentence["from"]]
-            assert role == conv.roles[j % 2], SHAREGPT_ASSERTION_FAILED_ROLE
+            assert role == conv.roles[j % 2]
             conv.append_message(role, sentence["value"])
 
         for part in conv.get_prompt():