[model] fix vlm utils (#8388)

Co-authored-by: viyer <vivek_iyer2@apple.com>

.dockerignore

@@ -3,9 +3,13 @@
 .github
 .venv
 cache
-data
-examples
+docker
+saves
+hf_cache
+ms_cache
+om_cache
+shared_data
+output
 .dockerignore
 .gitattributes
 .gitignore
-Dockerfile

.env.local

@@ -0,0 +1,42 @@
+# Note: actually we do not support .env, just for reference
+# api
+API_HOST=
+API_PORT=
+API_KEY=
+API_MODEL_NAME=
+API_VERBOSE=
+FASTAPI_ROOT_PATH=
+MAX_CONCURRENT=
+# general
+DISABLE_VERSION_CHECK=
+FORCE_CHECK_IMPORTS=
+ALLOW_EXTRA_ARGS=
+LLAMAFACTORY_VERBOSITY=
+USE_MODELSCOPE_HUB=
+USE_OPENMIND_HUB=
+USE_RAY=
+RECORD_VRAM=
+OPTIM_TORCH=
+NPU_JIT_COMPILE=
+# torchrun
+FORCE_TORCHRUN=
+MASTER_ADDR=
+MASTER_PORT=
+NNODES=
+NODE_RANK=
+NPROC_PER_NODE=
+# wandb
+WANDB_DISABLED=
+WANDB_PROJECT=
+WANDB_API_KEY=
+# gradio ui
+GRADIO_SHARE=
+GRADIO_SERVER_NAME=
+GRADIO_SERVER_PORT=
+GRADIO_ROOT_PATH=
+GRADIO_IPV6=
+# setup
+ENABLE_SHORT_CONSOLE=
+# reserved (do not use)
+LLAMABOARD_ENABLED=
+LLAMABOARD_WORKDIR=

.github/CONTRIBUTING.md

@@ -19,3 +19,49 @@ There are several ways you can contribute to LLaMA Factory:
 ### Style guide
 
 LLaMA Factory follows the [Google Python Style Guide](https://google.github.io/styleguide/pyguide.html), check it for details.
+
+### Create a Pull Request
+
+1. Fork the [repository](https://github.com/hiyouga/LLaMA-Factory) by clicking on the [Fork](https://github.com/hiyouga/LLaMA-Factory/fork) button on the repository's page. This creates a copy of the code under your GitHub user account.
+
+2. Clone your fork to your local disk, and add the base repository as a remote:
+
+```bash
+git clone git@github.com:[username]/LLaMA-Factory.git
+cd LLaMA-Factory
+git remote add upstream https://github.com/hiyouga/LLaMA-Factory.git
+```
+
+3. Create a new branch to hold your development changes:
+
+```bash
+git checkout -b dev_your_branch
+```
+
+4. Set up a development environment by running the following command in a virtual environment:
+
+```bash
+pip install -e ".[dev]"
+```
+
+If LLaMA Factory was already installed in the virtual environment, remove it with `pip uninstall llamafactory` before reinstalling it in editable mode with the -e flag.
+
+5. Check code before commit:
+
+```bash
+make commit
+make style && make quality
+make test
+```
+
+6. Submit changes:
+
+```bash
+git add .
+git commit -m "commit message"
+git fetch upstream
+git rebase upstream/main
+git push -u origin dev_your_branch
+```
+
+7. Create a merge request from your branch `dev_your_branch` at [origin repo](https://github.com/hiyouga/LLaMA-Factory).

.github/ISSUE_TEMPLATE/1-bug-report.yml

@@ -0,0 +1,61 @@
+name: "\U0001F41B Bug / help"
+description: Create a report to help us improve the LLaMA Factory
+labels: ["bug", "pending"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Issues included in **[FAQs](https://github.com/hiyouga/LLaMA-Factory/issues/4614)** or those with **insufficient** information may be closed without a response.
+        已经包含在 **[常见问题](https://github.com/hiyouga/LLaMA-Factory/issues/4614)** 内或提供信息**不完整**的 issues 可能不会被回复。
+
+  - type: markdown
+    attributes:
+      value: |
+        Please do not create issues that are not related to framework bugs under this category, use **[Discussions](https://github.com/hiyouga/LLaMA-Factory/discussions/categories/q-a)** instead.
+        请勿在此分类下创建和框架 bug 无关的 issues，训练问题求助请使用 **[讨论区](https://github.com/hiyouga/LLaMA-Factory/discussions/categories/q-a)**。
+
+  - type: checkboxes
+    id: reminder
+    attributes:
+      label: Reminder
+      description: |
+        Please ensure you have read the above rules carefully and searched the existing issues (including FAQs).
+        请确保您已经认真阅读了上述规则并且搜索过现有的 issues（包括常见问题）。
+
+      options:
+        - label: I have read the above rules and searched the existing issues.
+          required: true
+
+  - type: textarea
+    id: system-info
+    validations:
+      required: true
+    attributes:
+      label: System Info
+      description: |
+        Please share your system info with us. You can run the command **llamafactory-cli env** and copy-paste its output below.
+        请提供您的系统信息。您可以在命令行运行 **llamafactory-cli env** 并将其输出复制到该文本框中。
+
+      placeholder: llamafactory version, platform, python version, ...
+
+  - type: textarea
+    id: reproduction
+    validations:
+      required: true
+    attributes:
+      label: Reproduction
+      description: |
+        Please provide entry arguments, error messages and stack traces that reproduces the problem.
+        请提供入口参数，错误日志以及异常堆栈以便于我们复现问题。
+
+      value: |
+        ```text
+        Put your message here.
+        ```
+
+  - type: textarea
+    id: others
+    validations:
+      required: false
+    attributes:
+      label: Others

.github/ISSUE_TEMPLATE/2-feature-request.yml

@@ -0,0 +1,41 @@
+name: "\U0001F680 Feature request"
+description: Submit a request for a new feature
+labels: ["enhancement", "pending"]
+body:
+  - type: markdown
+    attributes:
+      value: |
+        Please do not create issues that are not related to new features under this category.
+        请勿在此分类下创建和新特性无关的 issues。
+
+  - type: checkboxes
+    id: reminder
+    attributes:
+      label: Reminder
+      description: |
+        Please ensure you have read the above rules carefully and searched the existing issues.
+        请确保您已经认真阅读了上述规则并且搜索过现有的 issues。
+
+      options:
+        - label: I have read the above rules and searched the existing issues.
+          required: true
+
+  - type: textarea
+    id: description
+    validations:
+      required: true
+    attributes:
+      label: Description
+      description: |
+        A clear and concise description of the feature proposal.
+        请详细描述您希望加入的新功能特性。
+
+  - type: textarea
+    id: contribution
+    validations:
+      required: false
+    attributes:
+      label: Pull Request
+      description: |
+        Have you already created the relevant PR and submitted the code?
+        您是否已经创建了相关 PR 并提交了代码？

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -1,58 +0,0 @@
-name: "\U0001F41B Bug / Help"
-description: Create a report to help us improve the LLaMA Factory
-body:
-  - type: checkboxes
-    id: reminder
-    attributes:
-      label: Reminder
-      description: |
-        Please ensure you have read the README carefully and searched the existing issues.
-        请确保您已经认真阅读了 README 并且搜索过现有的 Issue。
-
-      options:
-        - label: I have read the README and searched the existing issues.
-          required: true
-
-  - type: textarea
-    id: reproduction
-    validations:
-      required: true
-    attributes:
-      label: Reproduction
-      description: |
-        Please provide code snippets, error messages and stack traces that reproduces the problem.
-        请提供运行参数，错误信息以及异常堆栈以便于我们复现该问题。
-        Remember to use Markdown tags to correctly format your code.
-        请合理使用 Markdown 标签来格式化您的文本。
-
-      placeholder: |
-        python src/train_bash.py ...
-
-  - type: textarea
-    id: expected-behavior
-    validations:
-      required: false
-    attributes:
-      label: Expected behavior
-      description: |
-        Please provide a clear and concise description of what you would expect to happen.
-        请提供您原本的目的，即这段代码的期望行为。
-
-  - type: textarea
-    id: system-info
-    validations:
-      required: false
-    attributes:
-      label: System Info
-      description: |
-        Please share your system info with us. You can run the command **transformers-cli env** and copy-paste its output below.
-        请提供您的系统信息。您可以在命令行运行 **transformers-cli env** 并将其输出复制到该文本框中。
-
-      placeholder: transformers version, platform, python version, ...
-
-  - type: textarea
-    id: others
-    validations:
-      required: false
-    attributes:
-      label: Others

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1 @@
+blank_issues_enabled: false

.github/PULL_REQUEST_TEMPLATE.md

@@ -5,3 +5,4 @@ Fixes # (issue)
 ## Before submitting
 
 - [ ] Did you read the [contributor guideline](https://github.com/hiyouga/LLaMA-Factory/blob/main/.github/CONTRIBUTING.md)?
+- [ ] Did you write any new necessary tests?

.github/workflows/docker.yml

@@ -0,0 +1,66 @@
+name: docker
+
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - "main"
+    paths:
+      - "**/*.py"
+      - "requirements.txt"
+      - "docker/**"
+      - ".github/workflows/*.yml"
+  pull_request:
+    branches:
+      - "main"
+    paths:
+      - "**/*.py"
+      - "requirements.txt"
+      - "docker/**"
+      - ".github/workflows/*.yml"
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+
+    concurrency:
+      group: ${{ github.workflow }}-${{ github.ref }}
+      cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+    environment:
+      name: docker
+      url: https://hub.docker.com/r/hiyouga/llamafactory
+
+    steps:
+      - name: Free up disk space
+        run: |
+          df -h
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf /opt/ghc
+          sudo rm -rf /opt/hostedtoolcache
+          df -h
+
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      - name: Login to Docker Hub
+        if: github.event_name != 'pull_request'
+        uses: docker/login-action@v3
+        with:
+          username: ${{ vars.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+
+      - name: Build and push Docker image
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          file: ./docker/docker-cuda/Dockerfile
+          build-args: |
+            EXTRAS=metrics,deepspeed,liger-kernel
+          push: ${{ github.event_name != 'pull_request' }}
+          tags: docker.io/hiyouga/llamafactory:latest
+          cache-from: type=gha
+          cache-to: type=gha,mode=max

.github/workflows/label_issue.yml

@@ -0,0 +1,32 @@
+name: label_issue
+
+on:
+  issues:
+    types:
+      - opened
+
+jobs:
+  label_issue:
+    runs-on: ubuntu-latest
+
+    permissions:
+      issues: write
+
+    steps:
+      - env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          ISSUE_URL: ${{ github.event.issue.html_url }}
+          ISSUE_TITLE: ${{ github.event.issue.title }}
+        run: |
+          LABEL=""
+          NPU_KEYWORDS=(npu huawei ascend 华为 昇腾)
+          ISSUE_TITLE_LOWER=$(echo $ISSUE_TITLE | tr '[:upper:]' '[:lower:]')
+          for KEYWORD in ${NPU_KEYWORDS[@]}; do
+            if [[ $ISSUE_TITLE_LOWER == *$KEYWORD* ]] && [[ $ISSUE_TITLE_LOWER != *input* ]]; then
+              LABEL="npu"
+              break
+            fi
+          done
+          if [ -n "$LABEL" ]; then
+            gh issue edit $ISSUE_URL --add-label $LABEL
+          fi

.github/workflows/publish.yml

@@ -0,0 +1,36 @@
+name: publish
+
+on:
+  workflow_dispatch:
+  release:
+    types:
+      - published
+
+jobs:
+  publish:
+    name: Upload release to PyPI
+
+    runs-on: ubuntu-latest
+
+    environment:
+      name: release
+      url: https://pypi.org/p/llamafactory
+
+    permissions:
+      id-token: write
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.9"
+
+      - name: Build package
+        run: |
+          make build
+
+      - name: Publish package
+        uses: pypa/gh-action-pypi-publish@release/v1

.github/workflows/tests.yml

@@ -1,29 +1,102 @@
 name: tests
 
 on:
+  workflow_dispatch:
   push:
-    branches: [ "main" ]
+    branches:
+      - "main"
+    paths:
+      - "**.py"
+      - "requirements.txt"
+      - ".github/workflows/*.yml"
   pull_request:
-    branches: [ "main" ]
+    branches:
+      - "main"
+    paths:
+      - "**.py"
+      - "requirements.txt"
+      - ".github/workflows/*.yml"
 
 jobs:
-  check_code_quality:
+  tests:
+    strategy:
+      fail-fast: false
+      matrix:
+        python:
+          - "3.9"
+          - "3.10"
+          - "3.11"
+          - "3.12"
+        os:
+          - "ubuntu-latest"
+          - "windows-latest"
+          - "macos-13"
+        transformers:
+          - null
+        include:  # test backward compatibility
+          - python: "3.9"
+            os: "ubuntu-latest"
+            transformers: "4.45.0"
+          - python: "3.9"
+            os: "ubuntu-latest"
+            transformers: "4.49.0"
+          - python: "3.9"
+            os: "ubuntu-latest"
+            transformers: "4.51.0"
 
-    runs-on: ubuntu-latest
+    runs-on: ${{ matrix.os }}
+
+    concurrency:
+      group: ${{ github.workflow }}-${{ github.ref }}-${{ matrix.os }}-${{ matrix.python }}-${{ matrix.transformers }}
+      cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+    env:
+      HF_TOKEN: ${{ secrets.HF_TOKEN }}
+      OS_NAME: ${{ matrix.os }}
 
     steps:
-      - uses: actions/checkout@v4
+      - name: Checkout
+        uses: actions/checkout@v4
 
       - name: Set up Python
         uses: actions/setup-python@v5
         with:
-          python-version: "3.8"
+          python-version: ${{ matrix.python }}
+          cache: "pip"
+          cache-dependency-path: "**/requirements*.txt"
 
       - name: Install dependencies
         run: |
           python -m pip install --upgrade pip
-          python -m pip install ruff
+          python -m pip install ".[torch,dev]"
+
+      - name: Install transformers
+        if: ${{ matrix.transformers }}
+        run: |
+          python -m pip install "transformers==${{ matrix.transformers }}"
+
+      - name: Cache files
+        id: hf-hub-cache
+        uses: actions/cache@v4
+        with:
+          path: ${{ runner.temp }}/huggingface
+          key: huggingface-${{ matrix.os }}-${{ matrix.python }}-${{ matrix.transformers }}-${{ hashFiles('tests/version.txt') }}
 
       - name: Check quality
         run: |
-            make style && make quality
+          make style && make quality
+
+      - name: Check license
+        run: |
+          make license
+
+      - name: Check build
+        run: |
+          make build
+
+      - name: Test with pytest
+        run: |
+          make test
+        env:
+          HF_HOME: ${{ runner.temp }}/huggingface
+          HF_HUB_OFFLINE: "${{ steps.hf-hub-cache.outputs.cache-hit == 'true' && '1' || '0' }}"

.gitignore

@@ -159,7 +159,21 @@ cython_debug/
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
 .idea/
 
+# vscode
+.vscode/
+
+# uv
+uv.lock
+
 # custom .gitignore
-user.config
-saves/
+hf_cache/
+ms_cache/
+om_cache/
 cache/
+config/
+saves/
+output/
+wandb/
+swanlog/
+generated_predictions.jsonl
+predictions_score.json

.pre-commit-config.yaml

@@ -0,0 +1,28 @@
+repos:
+-   repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v5.0.0
+    hooks:
+    -   id: check-ast
+    -   id: check-added-large-files
+        args: ['--maxkb=25000']
+    -   id: check-merge-conflict
+    -   id: check-yaml
+    -   id: debug-statements
+    -   id: end-of-file-fixer
+    -   id: trailing-whitespace
+        args: [--markdown-linebreak-ext=md]
+    -   id: no-commit-to-branch
+        args: ['--branch', 'main']
+
+-   repo: https://github.com/asottile/pyupgrade
+    rev: v3.17.0
+    hooks:
+    -   id: pyupgrade
+        args: [--py38-plus]
+
+-   repo: https://github.com/astral-sh/ruff-pre-commit
+    rev: v0.6.9
+    hooks:
+    -   id: ruff
+        args: [--fix]
+    -   id: ruff-format

CITATION.cff

@@ -12,12 +12,16 @@ authors:
   given-names: "Yanhan"
 - family-names: "Luo"
   given-names: "Zheyan"
+- family-names: "Feng"
+  given-names: "Zhangchi"
 - family-names: "Ma"
   given-names: "Yongqiang"
 title: "LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models"
 url: "https://arxiv.org/abs/2403.13372"
 preferred-citation:
-  type: article
+  type: conference-paper
+  conference:
+    name: "Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations)"
   authors:
     - family-names: "Zheng"
       given-names: "Yaowei"
@@ -29,9 +33,12 @@ preferred-citation:
       given-names: "Yanhan"
     - family-names: "Luo"
       given-names: "Zheyan"
+    - family-names: "Feng"
+      given-names: "Zhangchi"
     - family-names: "Ma"
       given-names: "Yongqiang"
-  journal: "arXiv preprint arXiv:2403.13372"
   title: "LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models"
   url: "https://arxiv.org/abs/2403.13372"
   year: 2024
+  publisher: "Association for Computational Linguistics"
+  address: "Bangkok, Thailand"

Dockerfile

@@ -1,14 +0,0 @@
-FROM nvcr.io/nvidia/pytorch:24.01-py3
-
-WORKDIR /app
-
-COPY requirements.txt /app/
-RUN pip install -r requirements.txt
-
-COPY . /app/
-RUN pip install -e .[deepspeed,metrics,bitsandbytes,qwen]
-
-VOLUME [ "/root/.cache/huggingface/", "/app/data", "/app/output" ]
-EXPOSE 7860
-
-CMD [ "python", "src/train_web.py" ]

MANIFEST.in

@@ -0,0 +1 @@
+include LICENSE requirements.txt

Makefile

@@ -1,6 +1,16 @@
-.PHONY: quality style
+.PHONY: build commit license quality style test
 
-check_dirs := scripts src tests
+check_dirs := scripts src tests setup.py
+
+build:
+	pip3 install build && python3 -m build
+
+commit:
+	pre-commit install
+	pre-commit run --all-files
+
+license:
+	python3 tests/check_license.py $(check_dirs)
 
 quality:
 	ruff check $(check_dirs)
@@ -9,3 +19,6 @@ quality:
 style:
 	ruff check $(check_dirs) --fix
 	ruff format $(check_dirs)
+
+test:
+	CUDA_VISIBLE_DEVICES= WANDB_DISABLED=true pytest -vv tests/

data/README.md

@@ -1,16 +1,22 @@
-If you are using a custom dataset, please provide your dataset definition in the following format in `dataset_info.json`.
+The [dataset_info.json](dataset_info.json) contains all available datasets. If you are using a custom dataset, please **make sure** to add a *dataset description* in `dataset_info.json` and specify `dataset: dataset_name` before training to use it.
+
+The `dataset_info.json` file should be put in the `dataset_dir` directory. You can change `dataset_dir` to use another directory. The default value is `./data`.
+
+Currently we support datasets in **alpaca** and **sharegpt** format. Allowed file types include json, jsonl, csv, parquet, arrow.
 
 ```json
 "dataset_name": {
-  "hf_hub_url": "the name of the dataset repository on the Hugging Face hub. (if specified, ignore script_url and file_name)",
-  "ms_hub_url": "the name of the dataset repository on the ModelScope hub. (if specified, ignore script_url and file_name)",
-  "script_url": "the name of the directory containing a dataset loading script. (if specified, ignore file_name)",
-  "file_name": "the name of the dataset file in this directory. (required if above are not specified)",
-  "file_sha1": "the SHA-1 hash value of the dataset file. (optional, does not affect training)",
-  "subset": "the name of the subset. (optional, default: None)",
-  "folder": "the name of the folder of the dataset repository on the Hugging Face hub. (optional, default: None)",
-  "ranking": "whether the dataset is a preference dataset or not. (default: false)",
+  "hf_hub_url": "the name of the dataset repository on the Hugging Face hub. (if specified, ignore script_url, file_name and cloud_file_name)",
+  "ms_hub_url": "the name of the dataset repository on the Model Scope hub. (if specified, ignore script_url, file_name and cloud_file_name)",
+  "script_url": "the name of the directory containing a dataset loading script. (if specified, ignore file_name and cloud_file_name)",
+  "cloud_file_name": "the name of the dataset file in s3/gcs cloud storage. (if specified, ignore file_name)",
+  "file_name": "the name of the dataset folder or dataset file in this directory. (required if above are not specified)",
   "formatting": "the format of the dataset. (optional, default: alpaca, can be chosen from {alpaca, sharegpt})",
+  "ranking": "whether the dataset is a preference dataset or not. (default: False)",
+  "subset": "the name of the subset. (optional, default: None)",
+  "split": "the name of dataset split to be used. (optional, default: train)",
+  "folder": "the name of the folder of the dataset repository on the Hugging Face hub. (optional, default: None)",
+  "num_samples": "the number of samples in the dataset to be used. (optional, default: None)",
   "columns (optional)": {
     "prompt": "the column name in the dataset containing the prompts. (default: instruction)",
     "query": "the column name in the dataset containing the queries. (default: input)",
@@ -18,7 +24,13 @@ If you are using a custom dataset, please provide your dataset definition in the
     "history": "the column name in the dataset containing the histories. (default: None)",
     "messages": "the column name in the dataset containing the messages. (default: conversations)",
     "system": "the column name in the dataset containing the system prompts. (default: None)",
-    "tools": "the column name in the dataset containing the tool description. (default: None)"
+    "tools": "the column name in the dataset containing the tool description. (default: None)",
+    "images": "the column name in the dataset containing the image inputs. (default: None)",
+    "videos": "the column name in the dataset containing the videos inputs. (default: None)",
+    "audios": "the column name in the dataset containing the audios inputs. (default: None)",
+    "chosen": "the column name in the dataset containing the chosen answers. (default: None)",
+    "rejected": "the column name in the dataset containing the rejected answers. (default: None)",
+    "kto_tag": "the column name in the dataset containing the kto tags. (default: None)"
   },
   "tags (optional, used for the sharegpt format)": {
     "role_tag": "the key in the message represents the identity. (default: from)",
@@ -32,11 +44,19 @@ If you are using a custom dataset, please provide your dataset definition in the
 }
 ```
 
-Given above, you can use the custom dataset via specifying `--dataset dataset_name`.
+## Alpaca Format
 
-----
+### Supervised Fine-Tuning Dataset
 
-Currently we support dataset in **alpaca** or **sharegpt** format, the dataset in alpaca format should follow the below format:
+* [Example dataset](alpaca_en_demo.json)
+
+In supervised fine-tuning, the `instruction` column will be concatenated with the `input` column and used as the user prompt, then the user prompt would be `instruction\ninput`. The `output` column represents the model response.
+
+For reasoning models, if the dataset contains chain-of-thought (CoT), the CoT needs to be placed in the model responses, such as `<think>cot</think>output`.
+
+The `system` column will be used as the system prompt if specified.
+
+The `history` column is a list consisting of string tuples representing prompt-response pairs in the history messages. Note that the responses in the history **will also be learned by the model** in supervised fine-tuning.
 
 ```json
 [
@@ -53,10 +73,11 @@ Currently we support dataset in **alpaca** or **sharegpt** format, the dataset i
 ]
 ```
 
-Regarding the above dataset, the `columns` in `dataset_info.json` should be:
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
 
 ```json
 "dataset_name": {
+  "file_name": "data.json",
   "columns": {
     "prompt": "instruction",
     "query": "input",
@@ -67,30 +88,193 @@ Regarding the above dataset, the `columns` in `dataset_info.json` should be:
 }
 ```
 
-The `query` column will be concatenated with the `prompt` column and used as the user prompt, then the user prompt would be `prompt\nquery`. The `response` column represents the model response.
+> [!TIP]  
+> If the model has reasoning capabilities (e.g. Qwen3) but the dataset does not contain chain-of-thought (CoT), LLaMA-Factory will automatically add empty CoT to the data. When `enable_thinking` is `True` (slow thinking, by default), the empty CoT will be added to the model responses and loss computation will be considered; otherwise (fast thinking), it will be added to the user prompts and loss computation will be ignored. Please keep the `enable_thinking` parameter consistent during training and inference.
+>
+> If you want to train data containing CoT with slow thinking and data without CoT with fast thinking, you can set `enable_thinking` to `None`. However, this feature is relatively complicated and should be used with caution.
 
-The `system` column will be used as the system prompt. The `history` column is a list consisting string tuples representing prompt-response pairs in the history. Note that the responses in the history **will also be used for training**.
+### Pre-training Dataset
 
-For the pre-training datasets, only the `prompt` column will be used for training.
+- [Example dataset](c4_demo.jsonl)
 
-For the preference datasets, the `response` column should be a string list whose length is 2, with the preferred answers appearing first, for example:
+In pre-training, only the `text` column will be used for model learning.
 
 ```json
-{
-  "instruction": "user instruction",
-  "input": "user input",
-  "output": [
-    "chosen answer",
-    "rejected answer"
-  ]
+[
+  {"text": "document"},
+  {"text": "document"}
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "columns": {
+    "prompt": "text"
+  }
 }
 ```
 
-Remember to set `"ranking": true` for the preference datasets.
+### Preference Dataset
 
-----
+Preference datasets are used for reward modeling, DPO training, ORPO and SimPO training.
 
-The dataset in sharegpt format should follow the below format:
+It requires a better response in `chosen` column and a worse response in `rejected` column.
+
+```json
+[
+  {
+    "instruction": "user instruction (required)",
+    "input": "user input (optional)",
+    "chosen": "chosen answer (required)",
+    "rejected": "rejected answer (required)"
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "ranking": true,
+  "columns": {
+    "prompt": "instruction",
+    "query": "input",
+    "chosen": "chosen",
+    "rejected": "rejected"
+  }
+}
+```
+
+### KTO Dataset
+
+An additional column `kto_tag` is required. Please refer to the [sharegpt](#sharegpt-format) format for details.
+
+### Multimodal Image Dataset
+
+An additional column `images` is required. Please refer to the [sharegpt](#sharegpt-format) format for details.
+
+### Multimodal Video Dataset
+
+An additional column `videos` is required. Please refer to the [sharegpt](#sharegpt-format) format for details.
+
+### Multimodal Audio Dataset
+
+An additional column `audios` is required. Please refer to the [sharegpt](#sharegpt-format) format for details.
+
+## Sharegpt Format
+
+### Supervised Fine-Tuning Dataset
+
+- [Example dataset](glaive_toolcall_en_demo.json)
+
+Compared to the alpaca format, the sharegpt format allows the datasets have **more roles**, such as human, gpt, observation and function. They are presented in a list of objects in the `conversations` column.
+
+Note that the human and observation should appear in odd positions, while gpt and function should appear in even positions.
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "user instruction"
+      },
+      {
+        "from": "function_call",
+        "value": "tool arguments"
+      },
+      {
+        "from": "observation",
+        "value": "tool result"
+      },
+      {
+        "from": "gpt",
+        "value": "model response"
+      }
+    ],
+    "system": "system prompt (optional)",
+    "tools": "tool description (optional)"
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "system": "system",
+    "tools": "tools"
+  }
+}
+```
+
+### Pre-training Dataset
+
+Not yet supported, please use the [alpaca](#alpaca-format) format.
+
+### Preference Dataset
+
+- [Example dataset](dpo_en_demo.json)
+
+Preference datasets in sharegpt format also require a better message in `chosen` column and a worse message in `rejected` column.
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "user instruction"
+      },
+      {
+        "from": "gpt",
+        "value": "model response"
+      },
+      {
+        "from": "human",
+        "value": "user instruction"
+      }
+    ],
+    "chosen": {
+      "from": "gpt",
+      "value": "chosen answer (required)"
+    },
+    "rejected": {
+      "from": "gpt",
+      "value": "rejected answer (required)"
+    }
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "ranking": true,
+  "columns": {
+    "messages": "conversations",
+    "chosen": "chosen",
+    "rejected": "rejected"
+  }
+}
+```
+
+### KTO Dataset
+
+- [Example dataset](kto_en_demo.json)
+
+KTO datasets require a extra `kto_tag` column containing the boolean human feedback.
 
 ```json
 [
@@ -105,30 +289,187 @@ The dataset in sharegpt format should follow the below format:
         "value": "model response"
       }
     ],
-    "system": "system prompt (optional)",
-    "tools": "tool description (optional)"
+    "kto_tag": "human feedback [true/false] (required)"
   }
 ]
 ```
 
-Regarding the above dataset, the `columns` in `dataset_info.json` should be:
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
 
 ```json
 "dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
   "columns": {
     "messages": "conversations",
-    "system": "system",
-    "tools": "tools"
-  },
-  "tags": {
-    "role_tag": "from",
-    "content_tag": "value",
-    "user_tag": "human",
-    "assistant_tag": "gpt"
+    "kto_tag": "kto_tag"
   }
 }
 ```
 
-where the `messages` column should be a list following the `u/a/u/a/u/a` order.
+### Multimodal Image Dataset
 
-Pre-training datasets and preference datasets are incompatible with the sharegpt format yet.
+- [Example dataset](mllm_demo.json)
+
+Multimodal image datasets require an `images` column containing the paths to the input images.
+
+The number of images should be identical to the `<image>` tokens in the conversations.
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "<image>user instruction"
+      },
+      {
+        "from": "gpt",
+        "value": "model response"
+      }
+    ],
+    "images": [
+      "image path (required)"
+    ]
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "images": "images"
+  }
+}
+```
+
+### Multimodal Video Dataset
+
+- [Example dataset](mllm_video_demo.json)
+
+Multimodal video datasets require a `videos` column containing the paths to the input videos.
+
+The number of videos should be identical to the `<video>` tokens in the conversations.
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "<video>user instruction"
+      },
+      {
+        "from": "gpt",
+        "value": "model response"
+      }
+    ],
+    "videos": [
+      "video path (required)"
+    ]
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "videos": "videos"
+  }
+}
+```
+
+### Multimodal Audio Dataset
+
+- [Example dataset](mllm_audio_demo.json)
+
+Multimodal audio datasets require an `audios` column containing the paths to the input audios.
+
+The number of audios should be identical to the `<audio>` tokens in the conversations.
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "<audio>user instruction"
+      },
+      {
+        "from": "gpt",
+        "value": "model response"
+      }
+    ],
+    "audios": [
+      "audio path (required)"
+    ]
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "audios": "audios"
+  }
+}
+```
+
+### OpenAI Format
+
+The openai format is simply a special case of the sharegpt format, where the first message may be a system prompt.
+
+```json
+[
+  {
+    "messages": [
+      {
+        "role": "system",
+        "content": "system prompt (optional)"
+      },
+      {
+        "role": "user",
+        "content": "user instruction"
+      },
+      {
+        "role": "assistant",
+        "content": "model response"
+      }
+    ]
+  }
+]
+```
+
+Regarding the above dataset, the *dataset description* in `dataset_info.json` should be:
+
+```json
+"dataset_name": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "messages"
+  },
+  "tags": {
+    "role_tag": "role",
+    "content_tag": "content",
+    "user_tag": "user",
+    "assistant_tag": "assistant",
+    "system_tag": "system"
+  }
+}
+```

data/README_zh.md

@@ -1,16 +1,21 @@
-如果您使用自定义数据集，请务必在 `dataset_info.json` 文件中按照以下格式提供数据集定义。
+[dataset_info.json](dataset_info.json) 包含了所有可用的数据集。如果您希望使用自定义数据集，请**务必**在 `dataset_info.json` 文件中添加*数据集描述*，并通过修改 `dataset: 数据集名称` 配置来使用数据集。
+
+其中 `dataset_info.json` 文件应放置在 `dataset_dir` 目录下。您可以通过修改 `dataset_dir` 参数来使用其他目录。默认值为 `./data`。
+
+目前我们支持 **alpaca** 格式和 **sharegpt** 格式的数据集。允许的文件类型包括 json、jsonl、csv、parquet 和 arrow。
 
 ```json
 "数据集名称": {
   "hf_hub_url": "Hugging Face 的数据集仓库地址（若指定，则忽略 script_url 和 file_name）",
   "ms_hub_url": "ModelScope 的数据集仓库地址（若指定，则忽略 script_url 和 file_name）",
   "script_url": "包含数据加载脚本的本地文件夹名称（若指定，则忽略 file_name）",
-  "file_name": "该目录下数据集文件的名称（若上述参数未指定，则此项必需）",
-  "file_sha1": "数据集文件的 SHA-1 哈希值（可选，留空不影响训练）",
-  "subset": "数据集子集的名称（可选，默认：None）",
-  "folder": "Hugging Face 仓库的文件夹名称（可选，默认：None）",
-  "ranking": "是否为偏好数据集（可选，默认：False）",
+  "file_name": "该目录下数据集文件夹或文件的名称（若上述参数未指定，则此项必需）",
   "formatting": "数据集格式（可选，默认：alpaca，可以为 alpaca 或 sharegpt）",
+  "ranking": "是否为偏好数据集（可选，默认：False）",
+  "subset": "数据集子集的名称（可选，默认：None）",
+  "split": "所使用的数据集切分（可选，默认：train）",
+  "folder": "Hugging Face 仓库的文件夹名称（可选，默认：None）",
+  "num_samples": "该数据集所使用的样本数量。（可选，默认：None）",
   "columns（可选）": {
     "prompt": "数据集代表提示词的表头名称（默认：instruction）",
     "query": "数据集代表请求的表头名称（默认：input）",
@@ -18,7 +23,13 @@
     "history": "数据集代表历史对话的表头名称（默认：None）",
     "messages": "数据集代表消息列表的表头名称（默认：conversations）",
     "system": "数据集代表系统提示的表头名称（默认：None）",
-    "tools": "数据集代表工具描述的表头名称（默认：None）"
+    "tools": "数据集代表工具描述的表头名称（默认：None）",
+    "images": "数据集代表图像输入的表头名称（默认：None）",
+    "videos": "数据集代表视频输入的表头名称（默认：None）",
+    "audios": "数据集代表音频输入的表头名称（默认：None）",
+    "chosen": "数据集代表更优回答的表头名称（默认：None）",
+    "rejected": "数据集代表更差回答的表头名称（默认：None）",
+    "kto_tag": "数据集代表 KTO 标签的表头名称（默认：None）"
   },
   "tags（可选，用于 sharegpt 格式）": {
     "role_tag": "消息中代表发送者身份的键名（默认：from）",
@@ -27,16 +38,24 @@
     "assistant_tag": "消息中代表助手的 role_tag（默认：gpt）",
     "observation_tag": "消息中代表工具返回结果的 role_tag（默认：observation）",
     "function_tag": "消息中代表工具调用的 role_tag（默认：function_call）",
-    "system_tag": "消息中代表系统提示的 role_tag（默认：system，会覆盖 system 列）"
+    "system_tag": "消息中代表系统提示的 role_tag（默认：system，会覆盖 system column）"
   }
 }
 ```
 
-添加后可通过指定 `--dataset 数据集名称` 参数使用自定义数据集。
+## Alpaca 格式
 
-----
+### 指令监督微调数据集
 
-该项目目前支持两种格式的数据集：**alpaca** 和 **sharegpt**，其中 alpaca 格式的数据集按照以下方式组织：
+- [样例数据集](alpaca_zh_demo.json)
+
+在指令监督微调时，`instruction` 列对应的内容会与 `input` 列对应的内容拼接后作为提示词，即提示词为 `instruction\ninput`。而 `output` 列对应的内容为模型回答。
+
+对于推理类模型的微调，如果数据集包含思维链，则需要把思维链放在模型回答中，例如 `<think>cot</think>output`。
+
+如果指定，`system` 列对应的内容将被作为系统提示词。
+
+`history` 列是由多个字符串二元组构成的列表，分别代表历史消息中每轮对话的指令和回答。注意在指令监督微调时，历史消息中的回答内容**也会被用于模型学习**。
 
 ```json
 [
@@ -53,10 +72,11 @@
 ]
 ```
 
-对于上述格式的数据，`dataset_info.json` 中的 `columns` 应为：
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
 
 ```json
 "数据集名称": {
+  "file_name": "data.json",
   "columns": {
     "prompt": "instruction",
     "query": "input",
@@ -67,30 +87,193 @@
 }
 ```
 
-其中 `query` 列对应的内容会与 `prompt` 列对应的内容拼接后作为用户指令，即用户指令为 `prompt\nquery`。`response` 列对应的内容为模型回答。
+> [!TIP]
+> 如果模型本身具备推理能力（如 Qwen3）而数据集不包含思维链，LLaMA-Factory 会自动为数据添加空思维链。当 `enable_thinking` 为 `True` 时（慢思考，默认），空思维链会添加到模型回答中并且计算损失，否则会添加到用户指令中并且不计算损失（快思考）。请在训练和推理时保持 `enable_thinking` 参数一致。
+>
+> 如果您希望训练包含思维链的数据时使用慢思考，训练不包含思维链的数据时使用快思考，可以设置 `enable_thinking` 为 `None`。但该功能较为复杂，请谨慎使用。
 
-`system` 列对应的内容将被作为系统提示词。`history` 列是由多个字符串二元组构成的列表，分别代表历史消息中每轮的指令和回答。注意历史消息中的回答**也会被用于训练**。
+### 预训练数据集
 
-对于预训练数据集，仅 `prompt` 列中的内容会用于模型训练。
+- [样例数据集](c4_demo.jsonl)
 
-对于偏好数据集，`response` 列应当是一个长度为 2 的字符串列表，排在前面的代表更优的回答，例如：
+在预训练时，只有 `text` 列中的内容会用于模型学习。
 
 ```json
-{
-  "instruction": "用户指令",
-  "input": "用户输入",
-  "output": [
-    "优质回答",
-    "劣质回答"
-  ]
+[
+  {"text": "document"},
+  {"text": "document"}
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "columns": {
+    "prompt": "text"
+  }
 }
 ```
 
-添加偏好数据集需要额外指定 `"ranking": true`。
+### 偏好数据集
 
-----
+偏好数据集用于奖励模型训练、DPO 训练、ORPO 训练和 SimPO 训练。
 
-而 sharegpt 格式的数据集按照以下方式组织：
+它需要在 `chosen` 列中提供更优的回答，并在 `rejected` 列中提供更差的回答。
+
+```json
+[
+  {
+    "instruction": "用户指令（必填）",
+    "input": "用户输入（选填）",
+    "chosen": "优质回答（必填）",
+    "rejected": "劣质回答（必填）"
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "ranking": true,
+  "columns": {
+    "prompt": "instruction",
+    "query": "input",
+    "chosen": "chosen",
+    "rejected": "rejected"
+  }
+}
+```
+
+### KTO 数据集
+
+KTO 数据集需要提供额外的 `kto_tag` 列。详情请参阅 [sharegpt](#sharegpt-格式)。
+
+### 多模态图像数据集
+
+多模态图像数据集需要提供额外的 `images` 列。详情请参阅 [sharegpt](#sharegpt-格式)。
+
+### 多模态视频数据集
+
+多模态视频数据集需要提供额外的 `videos` 列。详情请参阅 [sharegpt](#sharegpt-格式)。
+
+### 多模态音频数据集
+
+多模态音频数据集需要提供额外的 `audios` 列。详情请参阅 [sharegpt](#sharegpt-格式)。
+
+## Sharegpt 格式
+
+### 指令监督微调数据集
+
+- [样例数据集](glaive_toolcall_zh_demo.json)
+
+相比 alpaca 格式的数据集，sharegpt 格式支持**更多的角色种类**，例如 human、gpt、observation、function 等等。它们构成一个对象列表呈现在 `conversations` 列中。
+
+注意其中 human 和 observation 必须出现在奇数位置，gpt 和 function 必须出现在偶数位置。
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "用户指令"
+      },
+      {
+        "from": "function_call",
+        "value": "工具参数"
+      },
+      {
+        "from": "observation",
+        "value": "工具结果"
+      },
+      {
+        "from": "gpt",
+        "value": "模型回答"
+      }
+    ],
+    "system": "系统提示词（选填）",
+    "tools": "工具描述（选填）"
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "system": "system",
+    "tools": "tools"
+  }
+}
+```
+
+### 预训练数据集
+
+尚不支持，请使用 [alpaca](#alpaca-格式) 格式。
+
+### 偏好数据集
+
+- [样例数据集](dpo_zh_demo.json)
+
+Sharegpt 格式的偏好数据集同样需要在 `chosen` 列中提供更优的消息，并在 `rejected` 列中提供更差的消息。
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "用户指令"
+      },
+      {
+        "from": "gpt",
+        "value": "模型回答"
+      },
+      {
+        "from": "human",
+        "value": "用户指令"
+      }
+    ],
+    "chosen": {
+      "from": "gpt",
+      "value": "优质回答"
+    },
+    "rejected": {
+      "from": "gpt",
+      "value": "劣质回答"
+    }
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "ranking": true,
+  "columns": {
+    "messages": "conversations",
+    "chosen": "chosen",
+    "rejected": "rejected"
+  }
+}
+```
+
+### KTO 数据集
+
+- [样例数据集](kto_en_demo.json)
+
+KTO 数据集需要额外添加一个 `kto_tag` 列，包含 bool 类型的人类反馈。
 
 ```json
 [
@@ -105,30 +288,191 @@
         "value": "模型回答"
       }
     ],
-    "system": "系统提示词（选填）",
-    "tools": "工具描述（选填）"
+    "kto_tag": "人类反馈 [true/false]（必填）"
   }
 ]
 ```
 
-对于上述格式的数据，`dataset_info.json` 中的 `columns` 应为：
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
 
 ```json
 "数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
   "columns": {
     "messages": "conversations",
-    "system": "system",
-    "tools": "tools"
-  },
-  "tags": {
-    "role_tag": "from",
-    "content_tag": "value",
-    "user_tag": "human",
-    "assistant_tag": "gpt"
+    "kto_tag": "kto_tag"
   }
 }
 ```
 
-其中 `messages` 列应当是一个列表，且符合 `用户/模型/用户/模型/用户/模型` 的顺序。
+### 多模态图像数据集
 
-预训练数据集和偏好数据集尚不支持 sharegpt 格式。
+- [样例数据集](mllm_demo.json)
+
+多模态图像数据集需要额外添加一个 `images` 列，包含输入图像的路径。
+
+注意图片的数量必须与文本中所有 `<image>` 标记的数量严格一致。
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "<image><image>用户指令"
+      },
+      {
+        "from": "gpt",
+        "value": "模型回答"
+      }
+    ],
+    "images": [
+      "图像路径（必填）",
+      "图像路径（必填）"
+    ]
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "images": "images"
+  }
+}
+```
+
+### 多模态视频数据集
+
+- [样例数据集](mllm_video_demo.json)
+
+多模态视频数据集需要额外添加一个 `videos` 列，包含输入视频的路径。
+
+注意视频的数量必须与文本中所有 `<video>` 标记的数量严格一致。
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "<video><video>用户指令"
+      },
+      {
+        "from": "gpt",
+        "value": "模型回答"
+      }
+    ],
+    "videos": [
+      "视频路径（必填）",
+      "视频路径（必填）"
+    ]
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "videos": "videos"
+  }
+}
+```
+
+### 多模态音频数据集
+
+- [样例数据集](mllm_audio_demo.json)
+
+多模态音频数据集需要额外添加一个 `audios` 列，包含输入音频的路径。
+
+注意音频的数量必须与文本中所有 `<audio>` 标记的数量严格一致。
+
+```json
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "<audio><audio>用户指令"
+      },
+      {
+        "from": "gpt",
+        "value": "模型回答"
+      }
+    ],
+    "audios": [
+      "音频路径（必填）",
+      "音频路径（必填）"
+    ]
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "conversations",
+    "audios": "audios"
+  }
+}
+```
+
+
+### OpenAI 格式
+
+OpenAI 格式仅仅是 sharegpt 格式的一种特殊情况，其中第一条消息可能是系统提示词。
+
+```json
+[
+  {
+    "messages": [
+      {
+        "role": "system",
+        "content": "系统提示词（选填）"
+      },
+      {
+        "role": "user",
+        "content": "用户指令"
+      },
+      {
+        "role": "assistant",
+        "content": "模型回答"
+      }
+    ]
+  }
+]
+```
+
+对于上述格式的数据，`dataset_info.json` 中的*数据集描述*应为：
+
+```json
+"数据集名称": {
+  "file_name": "data.json",
+  "formatting": "sharegpt",
+  "columns": {
+    "messages": "messages"
+  },
+  "tags": {
+    "role_tag": "role",
+    "content_tag": "content",
+    "user_tag": "user",
+    "assistant_tag": "assistant",
+    "system_tag": "system"
+  }
+}
+```

data/alpaca_data_en_52k.json.REMOVED.git-id

@@ -1 +0,0 @@
-3779ddbc040543ab1834ef216c983d6fcc06cc9a

data/alpaca_data_zh_51k.json.REMOVED.git-id

@@ -1 +0,0 @@
-a97cf9475291591843976554878568e046d8a46d

data/alpaca_gpt4_data_en.json.REMOVED.git-id

@@ -1 +0,0 @@
-25508714b7879a1e5a6764ba7f979a980f549f1a

data/alpaca_gpt4_data_zh.json.REMOVED.git-id

@@ -1 +0,0 @@
-7cb6a7d11455bddc3d495750a2392683d775b184

data/belle_multiturn/belle_multiturn.py

@@ -1,3 +1,18 @@
+# Copyright 2025 the LlamaFactory team.
+# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
 import json
 import os
 
@@ -10,16 +25,16 @@ _DESCRIPTION = "BELLE multiturn chat dataset."
 
 _CITATION = """\
 @article{belle2023exploring,
-  title={Exploring the Impact of Instruction Data Scaling on Large Language Models: An Empirical Study on Real-World Use Cases},
+  title={Exploring the Impact of Instruction Data Scaling on Large Language Models},
   author={Yunjie Ji, Yong Deng, Yan Gong, Yiping Peng, Qiang Niu, Lei Zhang, Baochang Ma, Xiangang Li},
   journal={arXiv preprint arXiv:2303.14742},
   year={2023}
 }
 """
 
-_HOMEPAGE = "{}/datasets/BelleGroup/multiturn_chat_0.8M".format(_HF_ENDPOINT)
+_HOMEPAGE = f"{_HF_ENDPOINT}/datasets/BelleGroup/multiturn_chat_0.8M"
 _LICENSE = "gpl-3.0"
-_URL = "{}/datasets/BelleGroup/multiturn_chat_0.8M/resolve/main/multiturn_chat_0.8M.json".format(_HF_ENDPOINT)
+_URL = f"{_HF_ENDPOINT}/datasets/BelleGroup/multiturn_chat_0.8M/resolve/main/multiturn_chat_0.8M.json"
 
 
 class BelleMultiturn(datasets.GeneratorBasedBuilder):
@@ -38,7 +53,7 @@ class BelleMultiturn(datasets.GeneratorBasedBuilder):
         return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": file_path})]
 
     def _generate_examples(self, filepath: str):
-        with open(filepath, "r", encoding="utf-8") as f:
+        with open(filepath, encoding="utf-8") as f:
             for key, row in enumerate(f):
                 data = json.loads(row)
                 conversations = []

data/comparison_gpt4_data_en.json.REMOVED.git-id

@@ -1 +0,0 @@
-f5cb08305ff5dc9c17a09809c54c8c8834aadc70

data/comparison_gpt4_data_zh.json.REMOVED.git-id

@@ -1 +0,0 @@
-aee47b7b443496e37808d7f34ef10403ff99bcc3

data/example_dataset/example_dataset.py

@@ -1,37 +0,0 @@
-import json
-from typing import Any, Dict, Generator, List, Tuple
-
-import datasets
-
-
-_DESCRIPTION = "An example of dataset."
-_CITATION = ""
-_HOMEPAGE = ""
-_LICENSE = ""
-_URL = "examples.json"
-
-
-class ExampleDataset(datasets.GeneratorBasedBuilder):
-    VERSION = datasets.Version("0.0.0")
-
-    def _info(self) -> datasets.DatasetInfo:
-        features = datasets.Features(
-            {
-                "instruction": datasets.Value("string"),
-                "input": datasets.Value("string"),
-                "output": datasets.Value("string"),
-                "history": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
-            }
-        )
-        return datasets.DatasetInfo(
-            description=_DESCRIPTION, features=features, homepage=_HOMEPAGE, license=_LICENSE, citation=_CITATION
-        )
-
-    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
-        file_path = dl_manager.download(_URL)
-        return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": file_path})]
-
-    def _generate_examples(self, filepath: str) -> Generator[Tuple[int, Dict[str, Any]], None, None]:
-        example_dataset = json.load(open(filepath, "r", encoding="utf-8"))
-        for key, example in enumerate(example_dataset):
-            yield key, example

data/glaive_toolcall_10k.json.REMOVED.git-id

@@ -1 +0,0 @@
-4748dff00d1dc42768a5b6cc772143c313017812

data/hh_rlhf_en/hh_rlhf_en.py

@@ -1,6 +1,20 @@
+# Copyright 2025 the LlamaFactory team.
+# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
 import json
 import os
-from typing import List
 
 import datasets
 
@@ -8,9 +22,9 @@ import datasets
 _HF_ENDPOINT = os.getenv("HF_ENDPOINT", "https://huggingface.co")
 _DESCRIPTION = "Human preference data about helpfulness and harmlessness."
 _CITATION = ""
-_HOMEPAGE = "{}/datasets/Anthropic/hh-rlhf".format(_HF_ENDPOINT)
+_HOMEPAGE = f"{_HF_ENDPOINT}/datasets/Anthropic/hh-rlhf"
 _LICENSE = "mit"
-_URL = "{}/datasets/Anthropic/hh-rlhf/resolve/main/".format(_HF_ENDPOINT)
+_URL = f"{_HF_ENDPOINT}/datasets/Anthropic/hh-rlhf/resolve/main/"
 _URLS = {
     "train": [
         _URL + "harmless-base/train.jsonl.gz",
@@ -34,7 +48,8 @@ class HhRlhfEn(datasets.GeneratorBasedBuilder):
         features = datasets.Features(
             {
                 "instruction": datasets.Value("string"),
-                "output": datasets.Sequence(datasets.Value("string")),
+                "chosen": datasets.Value("string"),
+                "rejected": datasets.Value("string"),
                 "history": datasets.Sequence(datasets.Sequence(datasets.Value("string"))),
             }
         )
@@ -49,10 +64,10 @@ class HhRlhfEn(datasets.GeneratorBasedBuilder):
             datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepaths": file_path["test"]}),
         ]
 
-    def _generate_examples(self, filepaths: List[str]):
+    def _generate_examples(self, filepaths: list[str]):
         key = 0
         for filepath in filepaths:
-            with open(filepath, "r", encoding="utf-8") as f:
+            with open(filepath, encoding="utf-8") as f:
                 for row in f:
                     data = json.loads(row)
                     chosen = data["chosen"]
@@ -79,5 +94,5 @@ class HhRlhfEn(datasets.GeneratorBasedBuilder):
                             break
                         prompt = prompt[:human_idx]
 
-                    yield key, {"instruction": query, "output": [r_accept, r_reject], "history": history}
+                    yield key, {"instruction": query, "chosen": r_accept, "rejected": r_reject, "history": history}
                     key += 1

data/oaast_rm.json.REMOVED.git-id

@@ -1 +0,0 @@
-274079ea921762be356de85b18f13fa60b7ba8cb

data/oaast_sft.json.REMOVED.git-id

@@ -1 +0,0 @@
-57fd080be5bffe4153fe3ee26a175e3d56da30f3

data/orca_rlhf.json.REMOVED.git-id

@@ -1 +0,0 @@
-736bcedea2b24a1414765c6d69cbdafaea839f3c

data/ultra_chat/ultra_chat.py

@@ -1,6 +1,20 @@
+# Copyright 2025 the LlamaFactory team.
+# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
 import json
 import os
-from typing import List
 
 import datasets
 
@@ -11,7 +25,7 @@ _DESCRIPTION = "UltraChat: Large-scale, Informative, and Diverse Multi-round Dia
 
 _CITATION = """\
 @misc{UltraChat,
-  author = {Ding, Ning and Chen, Yulin and Xu, Bokai and Hu, Shengding and Qin, Yujia and Liu, Zhiyuan and Sun, Maosong and Zhou, Bowen},
+  author = {Ding, Ning and Chen, Yulin and Xu, Bokai and Hu, Shengding and others},
   title = {UltraChat: A Large-scale Auto-generated Multi-round Dialogue Data},
   year = {2023},
   publisher = {GitHub},
@@ -20,9 +34,9 @@ _CITATION = """\
 }
 """
 
-_HOMEPAGE = "{}/datasets/stingning/ultrachat".format(_HF_ENDPOINT)
+_HOMEPAGE = f"{_HF_ENDPOINT}/datasets/stingning/ultrachat"
 _LICENSE = "cc-by-nc-4.0"
-_BASE_DATA_URL = "{}/datasets/stingning/ultrachat/resolve/main/train_{{idx}}.jsonl".format(_HF_ENDPOINT)
+_BASE_DATA_URL = f"{_HF_ENDPOINT}/datasets/stingning/ultrachat/resolve/main/train_{{idx}}.jsonl"
 
 
 class UltraChat(datasets.GeneratorBasedBuilder):
@@ -40,16 +54,16 @@ class UltraChat(datasets.GeneratorBasedBuilder):
         file_paths = [dl_manager.download(_BASE_DATA_URL.format(idx=idx)) for idx in range(10)]  # multiple shards
         return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepaths": file_paths})]
 
-    def _generate_examples(self, filepaths: List[str]):
+    def _generate_examples(self, filepaths: list[str]):
         for filepath in filepaths:
-            with open(filepath, "r", encoding="utf-8") as f:
+            with open(filepath, encoding="utf-8") as f:
                 for row in f:
                     try:
                         data = json.loads(row)
                     except Exception:
                         continue
                     key: int = data["id"]
-                    content: List[str] = data["data"]
+                    content: list[str] = data["data"]
                     if len(content) % 2 == 1:
                         content.pop(-1)
                     if len(content) < 2:

data/wiki_demo.txt.REMOVED.git-id

@@ -1 +0,0 @@
-c9cf509b7fdac5490cfd6dae72c2d7b8a60af6cb

docker-compose.yml

@@ -1,25 +0,0 @@
-version: '3.8'
-
-services:
-  llama-factory:
-    build:
-      dockerfile: Dockerfile
-      context: .
-    container_name: llama_factory
-    volumes:
-      - ./hf_cache:/root/.cache/huggingface/
-      - ./data:/app/data
-      - ./output:/app/output
-    environment:
-      - CUDA_VISIBLE_DEVICES=0
-    ports:
-      - "7860:7860"
-    ipc: host
-    deploy:
-      resources:
-        reservations:
-          devices:
-          - driver: nvidia
-            count: "all"
-            capabilities: [gpu]
-    restart: unless-stopped

docker/docker-cuda/Dockerfile

@@ -0,0 +1,66 @@
+# https://hub.docker.com/r/hiyouga/pytorch/tags
+ARG BASE_IMAGE=hiyouga/pytorch:th2.6.0-cu124-flashattn2.7.4-cxx11abi0-devel
+FROM ${BASE_IMAGE}
+
+# Installation arguments
+ARG PIP_INDEX=https://pypi.org/simple
+ARG EXTRAS=metrics
+ARG INSTALL_FLASHATTN=false
+ARG HTTP_PROXY=""
+
+# Define environments
+ENV MAX_JOBS=16
+ENV FLASH_ATTENTION_FORCE_BUILD=TRUE
+ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
+ENV DEBIAN_FRONTEND=noninteractive
+ENV NODE_OPTIONS=""
+ENV PIP_ROOT_USER_ACTION=ignore
+ENV http_proxy="${HTTP_PROXY}"
+ENV https_proxy="${HTTP_PROXY}"
+
+# Use Bash instead of default /bin/sh
+SHELL ["/bin/bash", "-c"]
+
+# Set the working directory
+WORKDIR /app
+
+# Change pip source
+RUN pip config set global.index-url "${PIP_INDEX}" && \
+    pip config set global.extra-index-url "${PIP_INDEX}" && \
+    pip install --no-cache-dir --upgrade pip packaging wheel setuptools
+
+# Install the requirements
+COPY requirements.txt /app
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the rest of the application into the image
+COPY . /app
+
+# Install LLaMA Factory
+RUN pip install --no-cache-dir -e ".[${EXTRAS}]" --no-build-isolation
+
+# Rebuild flash attention
+RUN if [ "${INSTALL_FLASHATTN}" == "true" ]; then \
+        pip uninstall -y ninja && \
+        pip install --no-cache-dir ninja && \
+        pip install --no-cache-dir flash-attn --no-build-isolation; \
+    fi
+
+# Set up volumes
+# VOLUME [ "/root/.cache/huggingface", "/app/shared_data", "/app/output" ]
+
+# Expose port 7860 for LLaMA Board
+ENV GRADIO_SERVER_PORT=7860
+EXPOSE 7860
+
+# Expose port 8000 for API service
+ENV API_PORT=8000
+EXPOSE 8000
+
+# unset proxy
+ENV http_proxy=
+ENV https_proxy=
+
+# Reset pip config
+RUN pip config unset global.index-url && \
+    pip config unset global.extra-index-url

docker/docker-cuda/Dockerfile.base

@@ -0,0 +1,55 @@
+# Start from the pytorch official image (ubuntu-22.04 + cuda-12.4.1 + python-3.11)
+# https://hub.docker.com/r/pytorch/pytorch/tags
+FROM pytorch/pytorch:2.6.0-cuda12.4-cudnn9-devel
+
+# Define environments
+ENV MAX_JOBS=16
+ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
+ENV DEBIAN_FRONTEND=noninteractive
+ENV NODE_OPTIONS=""
+ENV PIP_ROOT_USER_ACTION=ignore
+
+# Define installation arguments
+ARG APT_SOURCE=https://mirrors.tuna.tsinghua.edu.cn/ubuntu/
+ARG PIP_INDEX=https://mirrors.tuna.tsinghua.edu.cn/pypi/web/simple
+
+# Set apt source
+RUN cp /etc/apt/sources.list /etc/apt/sources.list.bak && \
+    { \
+    echo "deb ${APT_SOURCE} jammy main restricted universe multiverse"; \
+    echo "deb ${APT_SOURCE} jammy-updates main restricted universe multiverse"; \
+    echo "deb ${APT_SOURCE} jammy-backports main restricted universe multiverse"; \
+    echo "deb ${APT_SOURCE} jammy-security main restricted universe multiverse"; \
+    } > /etc/apt/sources.list
+
+# Install systemctl and wget
+RUN apt-get update && \
+    apt-get install -y -o Dpkg::Options::="--force-confdef" systemd wget && \
+    apt-get clean
+
+# Install git and vim
+RUN apt-get update && \
+    apt-get install -y git vim && \
+    apt-get clean
+
+# Install gcc and g++
+RUN apt-get update && \
+    apt-get install -y gcc g++ && \
+    apt-get clean
+
+# Change pip source
+RUN pip config set global.index-url "${PIP_INDEX}" && \
+    pip config set global.extra-index-url "${PIP_INDEX}" && \
+    pip install --no-cache-dir --upgrade pip packaging wheel setuptools
+
+# Install flash-attn-2.7.4.post1 (cxx11abi=False)
+RUN wget -nv https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl && \
+    pip install --no-cache-dir flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp311-cp311-linux_x86_64.whl
+
+# Install flashinfer-0.2.2.post1+cu124 (cxx11abi=False)
+RUN wget -nv https://github.com/flashinfer-ai/flashinfer/releases/download/v0.2.2.post1/flashinfer_python-0.2.2.post1+cu124torch2.6-cp38-abi3-linux_x86_64.whl && \
+    pip install --no-cache-dir flashinfer_python-0.2.2.post1+cu124torch2.6-cp38-abi3-linux_x86_64.whl
+
+# Reset pip config
+RUN pip config unset global.index-url && \
+    pip config unset global.extra-index-url

docker/docker-cuda/docker-compose.yml

@@ -0,0 +1,25 @@
+services:
+  llamafactory:
+    build:
+      dockerfile: ./docker/docker-cuda/Dockerfile
+      context: ../..
+      args:
+        PIP_INDEX: https://pypi.org/simple
+        EXTRAS: metrics
+    container_name: llamafactory
+    ports:
+      - "7860:7860"
+      - "8000:8000"
+    ipc: host
+    tty: true
+    # shm_size: "16gb"  # ipc: host is set
+    stdin_open: true
+    command: bash
+    deploy:
+      resources:
+        reservations:
+          devices:
+          - driver: nvidia
+            count: "all"
+            capabilities: [ gpu ]
+    restart: unless-stopped

docker/docker-npu/Dockerfile

@@ -0,0 +1,58 @@
+# https://hub.docker.com/r/ascendai/cann/tags
+ARG BASE_IMAGE=ascendai/cann:8.0.0-910b-ubuntu22.04-py3.11
+FROM ${BASE_IMAGE}
+
+# Installation arguments
+ARG PIP_INDEX=https://pypi.org/simple
+ARG EXTRAS=torch-npu,metrics
+ARG HTTP_PROXY=""
+
+# Define environments
+ENV MAX_JOBS=16
+ENV FLASH_ATTENTION_FORCE_BUILD=TRUE
+ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
+ENV DEBIAN_FRONTEND=noninteractive
+ENV NODE_OPTIONS=""
+ENV PIP_ROOT_USER_ACTION=ignore
+ENV http_proxy="${HTTP_PROXY}"
+ENV https_proxy="${HTTP_PROXY}"
+
+# Use Bash instead of default /bin/sh
+SHELL ["/bin/bash", "-c"]
+
+# Set the working directory
+WORKDIR /app
+
+# Change pip source
+RUN pip config set global.index-url "${PIP_INDEX}" && \
+    pip config set global.extra-index-url "${PIP_INDEX}" && \
+    pip install --no-cache-dir --upgrade pip packaging wheel setuptools
+
+# Install the requirements
+COPY requirements.txt /app
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the rest of the application into the image
+COPY . /app
+
+# Install LLaMA Factory
+RUN pip install --no-cache-dir -e ".[${EXTRAS}]" --no-build-isolation
+
+# Set up volumes
+# VOLUME [ "/root/.cache/huggingface", "/app/shared_data", "/app/output" ]
+
+# Expose port 7860 for LLaMA Board
+ENV GRADIO_SERVER_PORT=7860
+EXPOSE 7860
+
+# Expose port 8000 for API service
+ENV API_PORT=8000
+EXPOSE 8000
+
+# unset proxy
+ENV http_proxy=
+ENV https_proxy=
+
+# Reset pip config
+RUN pip config unset global.index-url && \
+    pip config unset global.extra-index-url

docker/docker-npu/docker-compose.yml

@@ -0,0 +1,28 @@
+services:
+  llamafactory:
+    build:
+      dockerfile: ./docker/docker-npu/Dockerfile
+      context: ../..
+      args:
+        PIP_INDEX: https://pypi.org/simple
+        EXTRAS: torch-npu,metrics
+    container_name: llamafactory
+    volumes:
+      - /usr/local/dcmi:/usr/local/dcmi
+      - /usr/local/bin/npu-smi:/usr/local/bin/npu-smi
+      - /usr/local/Ascend/driver:/usr/local/Ascend/driver
+      - /etc/ascend_install.info:/etc/ascend_install.info
+    ports:
+      - "7860:7860"
+      - "8000:8000"
+    ipc: host
+    tty: true
+    # shm_size: "16gb"  # ipc: host is set
+    stdin_open: true
+    command: bash
+    devices:
+      - /dev/davinci0
+      - /dev/davinci_manager
+      - /dev/devmm_svm
+      - /dev/hisi_hdc
+    restart: unless-stopped

docker/docker-rocm/Dockerfile

@@ -0,0 +1,71 @@
+# https://hub.docker.com/r/rocm/pytorch/tags
+ARG BASE_IMAGE=rocm/pytorch:rocm6.4.1_ubuntu22.04_py3.10_pytorch_release_2.6.0
+FROM ${BASE_IMAGE}
+
+# Installation arguments
+ARG PIP_INDEX=https://pypi.org/simple
+ARG EXTRAS=metrics
+ARG INSTALL_FLASHATTN=false
+ARG HTTP_PROXY=""
+ARG PYTORCH_INDEX=https://download.pytorch.org/whl/rocm6.3
+
+# Define environments
+ENV MAX_JOBS=16
+ENV FLASH_ATTENTION_FORCE_BUILD=TRUE
+ENV VLLM_WORKER_MULTIPROC_METHOD=spawn
+ENV DEBIAN_FRONTEND=noninteractive
+ENV NODE_OPTIONS=""
+ENV PIP_ROOT_USER_ACTION=ignore
+ENV http_proxy="${HTTP_PROXY}"
+ENV https_proxy="${HTTP_PROXY}"
+
+# Use Bash instead of default /bin/sh
+SHELL ["/bin/bash", "-c"]
+
+# Set the working directory
+WORKDIR /app
+
+# Change pip source
+RUN pip config set global.index-url "${PIP_INDEX}" && \
+    pip config set global.extra-index-url "${PIP_INDEX}" && \
+    pip install --no-cache-dir --upgrade pip packaging wheel setuptools
+
+# Reinstall pytorch rocm
+RUN pip uninstall -y torch torchvision torchaudio && \
+    pip install --no-cache-dir --pre torch torchvision torchaudio --index-url "${PYTORCH_INDEX}"
+
+# Install the requirements
+COPY requirements.txt /app
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the rest of the application into the image
+COPY . /app
+
+# Install LLaMA Factory
+RUN pip install --no-cache-dir -e ".[${EXTRAS}]" --no-build-isolation
+
+# Rebuild flash attention
+RUN if [ "${INSTALL_FLASHATTN}" == "true" ]; then \
+        pip uninstall -y ninja && \
+        pip install --no-cache-dir ninja && \
+        pip install --no-cache-dir flash-attn --no-build-isolation; \
+    fi
+
+# Set up volumes
+# VOLUME [ "/root/.cache/huggingface", "/app/shared_data", "/app/output" ]
+
+# Expose port 7860 for LLaMA Board
+ENV GRADIO_SERVER_PORT=7860
+EXPOSE 7860
+
+# Expose port 8000 for API service
+ENV API_PORT=8000
+EXPOSE 8000
+
+# unset proxy
+ENV http_proxy=
+ENV https_proxy=
+
+# Reset pip config
+RUN pip config unset global.index-url && \
+    pip config unset global.extra-index-url

docker/docker-rocm/docker-compose.yml

@@ -0,0 +1,21 @@
+services:
+  llamafactory:
+    build:
+      dockerfile: ./docker/docker-rocm/Dockerfile
+      context: ../..
+      args:
+        PIP_INDEX: https://pypi.org/simple
+        EXTRAS: metrics
+    container_name: llamafactory
+    ports:
+      - "7860:7860"
+      - "8000:8000"
+    ipc: host
+    tty: true
+    # shm_size: "16gb"  # ipc: host is set
+    stdin_open: true
+    command: bash
+    devices:
+      - /dev/kfd:/dev/kfd
+      - /dev/dri:/dev/dri
+    restart: unless-stopped

evaluation/ceval/ceval.py

@@ -1,3 +1,4 @@
+# Copyright 2025 the LlamaFactory team.
 # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -11,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import os
 
 import datasets
@@ -20,14 +22,15 @@ import pandas as pd
 _CITATION = """\
 @article{huang2023ceval,
   title={C-Eval: A Multi-Level Multi-Discipline Chinese Evaluation Suite for Foundation Models},
-  author={Huang, Yuzhen and Bai, Yuzhuo and Zhu, Zhihao and Zhang, Junlei and Zhang, Jinghan and Su, Tangjun and Liu, Junteng and Lv, Chuancheng and Zhang, Yikai and Lei, Jiayi and Fu, Yao and Sun, Maosong and He, Junxian},
+  author={Huang, Yuzhen and Bai, Yuzhuo and Zhu, Zhihao and others},
   journal={arXiv preprint arXiv:2305.08322},
   year={2023}
 }
 """
 
 _DESCRIPTION = """\
-C-Eval is a comprehensive Chinese evaluation suite for foundation models. It consists of 13948 multi-choice questions spanning 52 diverse disciplines and four difficulty levels.
+C-Eval is a comprehensive Chinese evaluation suite for foundation models.
+It consists of 13948 multi-choice questions spanning 52 diverse disciplines and four difficulty levels.
 """
 
 _HOMEPAGE = "https://cevalbenchmark.com"
@@ -133,25 +136,19 @@ class Ceval(datasets.GeneratorBasedBuilder):
             datasets.SplitGenerator(
                 name=datasets.Split.TEST,
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir, "test", f"{task_name}_test.csv"
-                    ),
+                    "filepath": os.path.join(data_dir, "test", f"{task_name}_test.csv"),
                 },
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.VALIDATION,
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir, "val", f"{task_name}_val.csv"
-                    ),
+                    "filepath": os.path.join(data_dir, "val", f"{task_name}_val.csv"),
                 },
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir, "dev", f"{task_name}_dev.csv"
-                    ),
+                    "filepath": os.path.join(data_dir, "dev", f"{task_name}_dev.csv"),
                 },
             ),
         ]

evaluation/cmmlu/cmmlu.py

@@ -1,3 +1,4 @@
+# Copyright 2025 the LlamaFactory team.
 # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -11,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import os
 
 import datasets
@@ -20,14 +22,15 @@ import pandas as pd
 _CITATION = """\
 @article{li2023cmmlu,
   title={CMMLU: Measuring massive multitask language understanding in Chinese},
-  author={Haonan Li and Yixuan Zhang and Fajri Koto and Yifei Yang and Hai Zhao and Yeyun Gong and Nan Duan and Timothy Baldwin},
+  author={Haonan Li and Yixuan Zhang and Fajri Koto and Yifei Yang and others,
   journal={arXiv preprint arXiv:2306.09212},
   year={2023}
 }
 """
 
 _DESCRIPTION = """\
-CMMLU is a comprehensive Chinese assessment suite specifically designed to evaluate the advanced knowledge and reasoning abilities of LLMs within the Chinese language and cultural context.
+CMMLU is a comprehensive Chinese assessment suite specifically designed to evaluate the advanced knowledge
+and reasoning abilities of LLMs within the Chinese language and cultural context.
 """
 
 _HOMEPAGE = "https://github.com/haonan-li/CMMLU"
@@ -37,73 +40,73 @@ _LICENSE = "Creative Commons Attribution-NonCommercial-ShareAlike 4.0 Internatio
 _URL = "cmmlu.zip"
 
 task_list = [
-     'agronomy',
-     'anatomy',
-     'ancient_chinese',
-     'arts',
-     'astronomy',
-     'business_ethics',
-     'chinese_civil_service_exam',
-     'chinese_driving_rule',
-     'chinese_food_culture',
-     'chinese_foreign_policy',
-     'chinese_history',
-     'chinese_literature',
-     'chinese_teacher_qualification',
-     'clinical_knowledge',
-     'college_actuarial_science',
-     'college_education',
-     'college_engineering_hydrology',
-     'college_law',
-     'college_mathematics',
-     'college_medical_statistics',
-     'college_medicine',
-     'computer_science',
-     'computer_security',
-     'conceptual_physics',
-     'construction_project_management',
-     'economics',
-     'education',
-     'electrical_engineering',
-     'elementary_chinese',
-     'elementary_commonsense',
-     'elementary_information_and_technology',
-     'elementary_mathematics',
-     'ethnology',
-     'food_science',
-     'genetics',
-     'global_facts',
-     'high_school_biology',
-     'high_school_chemistry',
-     'high_school_geography',
-     'high_school_mathematics',
-     'high_school_physics',
-     'high_school_politics',
-     'human_sexuality',
-     'international_law',
-     'journalism',
-     'jurisprudence',
-     'legal_and_moral_basis',
-     'logical',
-     'machine_learning',
-     'management',
-     'marketing',
-     'marxist_theory',
-     'modern_chinese',
-     'nutrition',
-     'philosophy',
-     'professional_accounting',
-     'professional_law',
-     'professional_medicine',
-     'professional_psychology',
-     'public_relations',
-     'security_study',
-     'sociology',
-     'sports_science',
-     'traditional_chinese_medicine',
-     'virology',
-     'world_history',
-     'world_religions',
+    "agronomy",
+    "anatomy",
+    "ancient_chinese",
+    "arts",
+    "astronomy",
+    "business_ethics",
+    "chinese_civil_service_exam",
+    "chinese_driving_rule",
+    "chinese_food_culture",
+    "chinese_foreign_policy",
+    "chinese_history",
+    "chinese_literature",
+    "chinese_teacher_qualification",
+    "clinical_knowledge",
+    "college_actuarial_science",
+    "college_education",
+    "college_engineering_hydrology",
+    "college_law",
+    "college_mathematics",
+    "college_medical_statistics",
+    "college_medicine",
+    "computer_science",
+    "computer_security",
+    "conceptual_physics",
+    "construction_project_management",
+    "economics",
+    "education",
+    "electrical_engineering",
+    "elementary_chinese",
+    "elementary_commonsense",
+    "elementary_information_and_technology",
+    "elementary_mathematics",
+    "ethnology",
+    "food_science",
+    "genetics",
+    "global_facts",
+    "high_school_biology",
+    "high_school_chemistry",
+    "high_school_geography",
+    "high_school_mathematics",
+    "high_school_physics",
+    "high_school_politics",
+    "human_sexuality",
+    "international_law",
+    "journalism",
+    "jurisprudence",
+    "legal_and_moral_basis",
+    "logical",
+    "machine_learning",
+    "management",
+    "marketing",
+    "marxist_theory",
+    "modern_chinese",
+    "nutrition",
+    "philosophy",
+    "professional_accounting",
+    "professional_law",
+    "professional_medicine",
+    "professional_psychology",
+    "public_relations",
+    "security_study",
+    "sociology",
+    "sports_science",
+    "traditional_chinese_medicine",
+    "virology",
+    "world_history",
+    "world_religions",
 ]
 
 

evaluation/mmlu/mmlu.py

@@ -1,3 +1,4 @@
+# Copyright 2025 the LlamaFactory team.
 # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
@@ -11,6 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+
 import os
 
 import datasets
@@ -20,14 +22,15 @@ import pandas as pd
 _CITATION = """\
 @article{hendryckstest2021,
   title={Measuring Massive Multitask Language Understanding},
-  author={Dan Hendrycks and Collin Burns and Steven Basart and Andy Zou and Mantas Mazeika and Dawn Song and Jacob Steinhardt},
+  author={Dan Hendrycks and Collin Burns and others},
   journal={Proceedings of the International Conference on Learning Representations (ICLR)},
   year={2021}
 }
 """
 
 _DESCRIPTION = """\
-Measuring Massive Multitask Language Understanding by Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt (ICLR 2021).
+Measuring Massive Multitask Language Understanding by Dan Hendrycks, Collin Burns, Steven Basart,
+Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt (ICLR 2021).
 """
 
 _HOMEPAGE = "https://github.com/hendrycks/test"
@@ -136,32 +139,25 @@ class MMLU(datasets.GeneratorBasedBuilder):
             datasets.SplitGenerator(
                 name=datasets.Split.TEST,
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir, "data", "test", f"{task_name}_test.csv"
-                    ),
+                    "filepath": os.path.join(data_dir, "data", "test", f"{task_name}_test.csv"),
                 },
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.VALIDATION,
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir, "data", "val", f"{task_name}_val.csv"
-                    ),
+                    "filepath": os.path.join(data_dir, "data", "val", f"{task_name}_val.csv"),
                 },
             ),
             datasets.SplitGenerator(
                 name=datasets.Split.TRAIN,
                 gen_kwargs={
-                    "filepath": os.path.join(
-                        data_dir, "data", "dev", f"{task_name}_dev.csv"
-                    ),
+                    "filepath": os.path.join(data_dir, "data", "dev", f"{task_name}_dev.csv"),
                 },
             ),
         ]
 
     def _generate_examples(self, filepath):
-        df = pd.read_csv(filepath)
+        df = pd.read_csv(filepath, header=None)
         df.columns = ["question", "A", "B", "C", "D", "answer"]
 
-        for i, instance in enumerate(df.to_dict(orient="records")):
-            yield i, instance
+        yield from enumerate(df.to_dict(orient="records"))

examples/README.md

@@ -1,48 +1,292 @@
 We provide diverse examples about fine-tuning LLMs.
 
+Make sure to execute these commands in the `LLaMA-Factory` directory.
+
+## Table of Contents
+
+- [LoRA Fine-Tuning](#lora-fine-tuning)
+- [QLoRA Fine-Tuning](#qlora-fine-tuning)
+- [Full-Parameter Fine-Tuning](#full-parameter-fine-tuning)
+- [Merging LoRA Adapters and Quantization](#merging-lora-adapters-and-quantization)
+- [Inferring LoRA Fine-Tuned Models](#inferring-lora-fine-tuned-models)
+- [Extras](#extras)
+
+Use `CUDA_VISIBLE_DEVICES` (GPU) or `ASCEND_RT_VISIBLE_DEVICES` (NPU) to choose computing devices.
+
+By default, LLaMA-Factory uses all visible computing devices.
+
+Basic usage:
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
 ```
-examples/
-├── lora_single_gpu/
-│   ├── pretrain.sh: Do continuous pre-training using LoRA
-│   ├── sft.sh: Do supervised fine-tuning using LoRA
-│   ├── reward.sh: Do reward modeling using LoRA
-│   ├── ppo.sh: Do PPO training using LoRA
-│   ├── dpo.sh: Do DPO training using LoRA
-│   ├── orpo.sh: Do ORPO training using LoRA
-│   ├── prepare.sh: Save tokenized dataset
-│   └── predict.sh: Do batch predict and compute BLEU and ROUGE scores after LoRA tuning
-├── qlora_single_gpu/
-│   ├── bitsandbytes.sh: Fine-tune 4/8-bit BNB models using QLoRA
-│   ├── gptq.sh: Fine-tune 4/8-bit GPTQ models using QLoRA
-│   ├── awq.sh: Fine-tune 4-bit AWQ models using QLoRA
-│   └── aqlm.sh: Fine-tune 2-bit AQLM models using QLoRA
-├── lora_multi_gpu/
-│   ├── single_node.sh: Fine-tune model with Accelerate on single node using LoRA
-│   └── multi_node.sh: Fine-tune model with Accelerate on multiple nodes using LoRA
-├── full_multi_gpu/
-│   ├── single_node.sh: Full fine-tune model with DeepSpeed on single node
-│   ├── multi_node.sh: Full fine-tune model with DeepSpeed on multiple nodes
-│   └── predict.sh: Do batch predict and compute BLEU and ROUGE scores after full tuning
-├── merge_lora/
-│   ├── merge.sh: Merge LoRA weights into the pre-trained models
-│   └── quantize.sh: Quantize the fine-tuned model with AutoGPTQ
-├── inference/
-│   ├── cli_demo.sh: Launch a command line interface with LoRA adapters
-│   ├── api_demo.sh: Launch an OpenAI-style API with LoRA adapters
-│   ├── web_demo.sh: Launch a web interface with LoRA adapters
-│   └── evaluate.sh: Evaluate model on the MMLU/CMMLU/C-Eval benchmarks with LoRA adapters
-└── extras/
-    ├── galore/
-    │   └── sft.sh: Fine-tune model with GaLore
-    ├── badam/
-    │   └── sft.sh: Fine-tune model with BAdam
-    ├── loraplus/
-    │   └── sft.sh: Fine-tune model using LoRA+
-    ├── mod/
-    │   └── sft.sh: Fine-tune model using Mixture-of-Depths
-    ├── llama_pro/
-    │   ├── expand.sh: Expand layers in the model
-    │   └── sft.sh: Fine-tune the expanded model
-    └── fsdp_qlora/
-        └── sft.sh: Fine-tune quantized model with FSDP+QLoRA
+
+Advanced usage:
+
+```bash
+CUDA_VISIBLE_DEVICES=0,1 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml \
+    learning_rate=1e-5 \
+    logging_steps=1
+```
+
+```bash
+bash examples/train_lora/llama3_lora_sft.sh
+```
+
+## Examples
+
+### LoRA Fine-Tuning
+
+#### (Continuous) Pre-Training
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_pretrain.yaml
+```
+
+#### Supervised Fine-Tuning
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+```
+
+#### Multimodal Supervised Fine-Tuning
+
+```bash
+llamafactory-cli train examples/train_lora/qwen2_5vl_lora_sft.yaml
+```
+
+#### DPO/ORPO/SimPO Training
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_dpo.yaml
+```
+
+#### Multimodal DPO/ORPO/SimPO Training
+
+```bash
+llamafactory-cli train examples/train_lora/qwen2_5vl_lora_dpo.yaml
+```
+
+#### Reward Modeling
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_reward.yaml
+```
+
+#### PPO Training
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_ppo.yaml
+```
+
+#### KTO Training
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_kto.yaml
+```
+
+#### Preprocess Dataset
+
+It is useful for large dataset, use `tokenized_path` in config to load the preprocessed dataset.
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_preprocess.yaml
+```
+
+#### Evaluating on MMLU/CMMLU/C-Eval Benchmarks
+
+```bash
+llamafactory-cli eval examples/train_lora/llama3_lora_eval.yaml
+```
+
+#### Supervised Fine-Tuning on Multiple Nodes
+
+```bash
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+```
+
+#### Supervised Fine-Tuning with DeepSpeed ZeRO-3 (Weight Sharding)
+
+```bash
+FORCE_TORCHRUN=1 llamafactory-cli train examples/train_lora/llama3_lora_sft_ds3.yaml
+```
+
+#### Supervised Fine-Tuning with Ray on 4 GPUs
+
+```bash
+USE_RAY=1 llamafactory-cli train examples/train_lora/llama3_lora_sft_ray.yaml
+```
+
+### QLoRA Fine-Tuning
+
+#### Supervised Fine-Tuning with 4/8-bit Bitsandbytes/HQQ/EETQ Quantization (Recommended)
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_otfq.yaml
+```
+
+#### Supervised Fine-Tuning with 4-bit Bitsandbytes Quantization on Ascend NPU
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_bnb_npu.yaml
+```
+
+#### Supervised Fine-Tuning with 4/8-bit GPTQ Quantization
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_gptq.yaml
+```
+
+#### Supervised Fine-Tuning with 4-bit AWQ Quantization
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_awq.yaml
+```
+
+#### Supervised Fine-Tuning with 2-bit AQLM Quantization
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_aqlm.yaml
+```
+
+### Full-Parameter Fine-Tuning
+
+#### Supervised Fine-Tuning on Single Node
+
+```bash
+FORCE_TORCHRUN=1 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+```
+
+#### Supervised Fine-Tuning on Multiple Nodes
+
+```bash
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+```
+
+### Elastic and Fault-Tolerant Supervised Fine-Tuning on Multiple Nodes
+
+To launch an elastic job with `MAX_RESTARTS` failures retries, run the following on at least `MIN_NNODES` nodes and at most `MAX_NNODES` nodes. `RDZV_ID` should be set as a unique job id (shared by all nodes participating in the job). See also [torchrun](https://docs.pytorch.org/docs/stable/elastic/run.html).
+
+```bash
+FORCE_TORCHRUN=1 MIN_NNODES=1 MAX_NNODES=3 MAX_RESTARTS=3 RDZV_ID=llamafactory MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+```
+
+#### Multimodal Supervised Fine-Tuning
+
+```bash
+FORCE_TORCHRUN=1 llamafactory-cli train examples/train_full/qwen2_5vl_full_sft.yaml
+```
+
+### Merging LoRA Adapters and Quantization
+
+#### Merge LoRA Adapters
+
+Note: DO NOT use quantized model or `quantization_bit` when merging LoRA adapters.
+
+```bash
+llamafactory-cli export examples/merge_lora/llama3_lora_sft.yaml
+```
+
+#### Quantizing Model using AutoGPTQ
+
+```bash
+llamafactory-cli export examples/merge_lora/llama3_gptq.yaml
+```
+
+### Save Ollama modelfile
+
+```bash
+llamafactory-cli export examples/merge_lora/llama3_full_sft.yaml
+```
+
+### Inferring LoRA Fine-Tuned Models
+
+#### Evaluation using vLLM's Multi-GPU Inference
+
+```
+python scripts/vllm_infer.py --model_name_or_path meta-llama/Meta-Llama-3-8B-Instruct --template llama3 --dataset alpaca_en_demo
+python scripts/eval_bleu_rouge.py generated_predictions.jsonl
+```
+
+#### Use CLI ChatBox
+
+```bash
+llamafactory-cli chat examples/inference/llama3_lora_sft.yaml
+```
+
+#### Use Web UI ChatBox
+
+```bash
+llamafactory-cli webchat examples/inference/llama3_lora_sft.yaml
+```
+
+#### Launch OpenAI-style API
+
+```bash
+llamafactory-cli api examples/inference/llama3_lora_sft.yaml
+```
+
+### Extras
+
+#### Full-Parameter Fine-Tuning using GaLore
+
+```bash
+llamafactory-cli train examples/extras/galore/llama3_full_sft.yaml
+```
+
+#### Full-Parameter Fine-Tuning using APOLLO
+
+```bash
+llamafactory-cli train examples/extras/apollo/llama3_full_sft.yaml
+```
+
+#### Full-Parameter Fine-Tuning using BAdam
+
+```bash
+llamafactory-cli train examples/extras/badam/llama3_full_sft.yaml
+```
+
+#### Full-Parameter Fine-Tuning using Adam-mini
+
+```bash
+llamafactory-cli train examples/extras/adam_mini/qwen2_full_sft.yaml
+```
+
+#### Full-Parameter Fine-Tuning using Muon
+
+```bash
+llamafactory-cli train examples/extras/muon/qwen2_full_sft.yaml
+```
+
+#### LoRA+ Fine-Tuning
+
+```bash
+llamafactory-cli train examples/extras/loraplus/llama3_lora_sft.yaml
+```
+
+#### PiSSA Fine-Tuning
+
+```bash
+llamafactory-cli train examples/extras/pissa/llama3_lora_sft.yaml
+```
+
+#### Mixture-of-Depths Fine-Tuning
+
+```bash
+llamafactory-cli train examples/extras/mod/llama3_full_sft.yaml
+```
+
+#### LLaMA-Pro Fine-Tuning
+
+```bash
+bash examples/extras/llama_pro/expand.sh
+llamafactory-cli train examples/extras/llama_pro/llama3_freeze_sft.yaml
+```
+
+#### FSDP+QLoRA Fine-Tuning
+
+```bash
+bash examples/extras/fsdp_qlora/train.sh
 ```

examples/README_zh.md

@@ -1,48 +1,292 @@
 我们提供了多样化的大模型微调示例脚本。
 
+请确保在 `LLaMA-Factory` 目录下执行下述命令。
+
+## 目录
+
+- [LoRA 微调](#lora-微调)
+- [QLoRA 微调](#qlora-微调)
+- [全参数微调](#全参数微调)
+- [合并 LoRA 适配器与模型量化](#合并-lora-适配器与模型量化)
+- [推理 LoRA 模型](#推理-lora-模型)
+- [杂项](#杂项)
+
+使用 `CUDA_VISIBLE_DEVICES`（GPU）或 `ASCEND_RT_VISIBLE_DEVICES`（NPU）选择计算设备。
+
+LLaMA-Factory 默认使用所有可见的计算设备。
+
+基础用法：
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
 ```
-examples/
-├── lora_single_gpu/
-│   ├── pretrain.sh: 基于 LoRA 进行增量预训练
-│   ├── sft.sh: 基于 LoRA 进行指令监督微调
-│   ├── reward.sh: 基于 LoRA 进行奖励模型训练
-│   ├── ppo.sh: 基于 LoRA 进行 PPO 训练
-│   ├── dpo.sh: 基于 LoRA 进行 DPO 训练
-│   ├── orpo.sh: 基于 LoRA 进行 ORPO 训练
-│   ├── prepare.sh: 保存预处理后的数据集
-│   └── predict.sh: 基于 LoRA 进行批量预测并计算 BLEU 和 ROUGE 分数
-├── qlora_single_gpu/
-│   ├── bitsandbytes.sh: 基于 QLoRA 微调 4/8 比特 BNB 模型
-│   ├── gptq.sh: 基于 QLoRA 微调 4/8 比特 GPTQ 模型
-│   ├── awq.sh: 基于 QLoRA 微调 4 比特 AWQ 模型
-│   └── aqlm.sh: 基于 QLoRA 微调 2 比特 AQLM 模型
-├── lora_multi_gpu/
-│   ├── single_node.sh: 使用 Accelerate 进行单节点 LoRA 训练
-│   └── multi_node.sh: 使用 Accelerate 进行多节点 LoRA 训练
-├── full_multi_gpu/
-│   ├── single_node.sh: 使用 DeepSpeed 进行单节点全量训练
-│   ├── multi_node.sh: 使用 DeepSpeed 进行多节点全量训练
-│   └── predict.sh: 基于全量训练进行批量预测并计算 BLEU 和 ROUGE 分数
-├── merge_lora/
-│   ├── merge.sh: 将 LoRA 权重合并到预训练模型中
-│   └── quantize.sh: 使用 AutoGPTQ 量化微调后的模型
-├── inference/
-│   ├── cli_demo.sh: 启动 LoRA 模型的命令行推理接口
-│   ├── api_demo.sh: 启动 LoRA 模型的 OpenAI 风格 API
-│   ├── web_demo.sh: 启动 LoRA 模型的浏览器推理接口
-│   └── evaluate.sh: 在 MMLU/CMMLU/C-Eval 数据集上评测 LoRA 模型
-└── extras/
-    ├── galore/
-    │   └── sft.sh: 使用 GaLore 训练模型
-    ├── badam/
-    │   └── sft.sh: 使用 BAdam 训练模型
-    ├── loraplus/
-    │   └── sft.sh: 使用 LoRA+ 训练模型
-    ├── mod/
-    │   └── sft.sh: 使用深度混合训练模型
-    ├── llama_pro/
-    │   ├── expand.sh: 扩展模型中的层
-    │   └── sft.sh: 训练扩展后的模型
-    └── fsdp_qlora/
-        └── sft.sh: 使用 FSDP+QLoRA 微调量化模型
+
+高级用法：
+
+```bash
+CUDA_VISIBLE_DEVICES=0,1 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml \
+    learning_rate=1e-5 \
+    logging_steps=1
+```
+
+```bash
+bash examples/train_lora/llama3_lora_sft.sh
+```
+
+## 示例
+
+### LoRA 微调
+
+#### （增量）预训练
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_pretrain.yaml
+```
+
+#### 指令监督微调
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+```
+
+#### 多模态指令监督微调
+
+```bash
+llamafactory-cli train examples/train_lora/qwen2_5vl_lora_sft.yaml
+```
+
+#### DPO/ORPO/SimPO 训练
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_dpo.yaml
+```
+
+#### 多模态 DPO/ORPO/SimPO 训练
+
+```bash
+llamafactory-cli train examples/train_lora/qwen2_5vl_lora_dpo.yaml
+```
+
+#### 奖励模型训练
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_reward.yaml
+```
+
+#### PPO 训练
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_ppo.yaml
+```
+
+#### KTO 训练
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_kto.yaml
+```
+
+#### 预处理数据集
+
+对于大数据集有帮助，在配置中使用 `tokenized_path` 以加载预处理后的数据集。
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_preprocess.yaml
+```
+
+#### 在 MMLU/CMMLU/C-Eval 上评估
+
+```bash
+llamafactory-cli eval examples/train_lora/llama3_lora_eval.yaml
+```
+
+#### 多机指令监督微调
+
+```bash
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_lora/llama3_lora_sft.yaml
+```
+
+### 支持弹性和容错的多机指令监督微调
+
+要启动一个支持弹性节点和容错的多机指令微调，在每个节点上执行以下命令。弹性节点数量范围为 `MIN_NNODES:MAX_NNODES`，每个节点最多允许因为错误重启 `MAX_RESTARTS` 次。`RDZV_ID` 应设置为一个唯一的作业 ID（由参与该作业的所有节点共享）。更多新可以参考官方文档 [torchrun](https://docs.pytorch.org/docs/stable/elastic/run.html)。
+
+```bash
+FORCE_TORCHRUN=1 MIN_NNODES=1 MAX_NNODES=3 MAX_RESTARTS=3 RDZV_ID=llamafactory MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+```
+
+#### 使用 DeepSpeed ZeRO-3 平均分配显存
+
+```bash
+FORCE_TORCHRUN=1 llamafactory-cli train examples/train_lora/llama3_lora_sft_ds3.yaml
+```
+
+#### 使用 Ray 在 4 张 GPU 上微调
+
+```bash
+USE_RAY=1 llamafactory-cli train examples/train_lora/llama3_lora_sft_ray.yaml
+```
+
+### QLoRA 微调
+
+#### 基于 4/8 比特 Bitsandbytes/HQQ/EETQ 量化进行指令监督微调（推荐）
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_otfq.yaml
+```
+
+#### 在 NPU 上基于 4 比特 Bitsandbytes 量化进行指令监督微调
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_bnb_npu.yaml
+```
+
+#### 基于 4/8 比特 GPTQ 量化进行指令监督微调
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_gptq.yaml
+```
+
+#### 基于 4 比特 AWQ 量化进行指令监督微调
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_awq.yaml
+```
+
+#### 基于 2 比特 AQLM 量化进行指令监督微调
+
+```bash
+llamafactory-cli train examples/train_qlora/llama3_lora_sft_aqlm.yaml
+```
+
+### 全参数微调
+
+#### 在单机上进行指令监督微调
+
+```bash
+FORCE_TORCHRUN=1 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+```
+
+#### 在多机上进行指令监督微调
+
+```bash
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=0 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+FORCE_TORCHRUN=1 NNODES=2 NODE_RANK=1 MASTER_ADDR=192.168.0.1 MASTER_PORT=29500 llamafactory-cli train examples/train_full/llama3_full_sft.yaml
+```
+
+#### 多模态指令监督微调
+
+```bash
+FORCE_TORCHRUN=1 llamafactory-cli train examples/train_full/qwen2_5vl_full_sft.yaml
+```
+
+### 合并 LoRA 适配器与模型量化
+
+#### 合并 LoRA 适配器
+
+注：请勿使用量化后的模型或 `quantization_bit` 参数来合并 LoRA 适配器。
+
+```bash
+llamafactory-cli export examples/merge_lora/llama3_lora_sft.yaml
+```
+
+#### 使用 AutoGPTQ 量化模型
+
+```bash
+llamafactory-cli export examples/merge_lora/llama3_gptq.yaml
+```
+
+### 保存 Ollama 配置文件
+
+```bash
+llamafactory-cli export examples/merge_lora/llama3_full_sft.yaml
+```
+
+### 推理 LoRA 模型
+
+#### 使用 vLLM 多卡推理评估
+
+```
+python scripts/vllm_infer.py --model_name_or_path meta-llama/Meta-Llama-3-8B-Instruct --template llama3 --dataset alpaca_en_demo
+python scripts/eval_bleu_rouge.py generated_predictions.jsonl
+```
+
+#### 使用命令行对话框
+
+```bash
+llamafactory-cli chat examples/inference/llama3_lora_sft.yaml
+```
+
+#### 使用浏览器对话框
+
+```bash
+llamafactory-cli webchat examples/inference/llama3_lora_sft.yaml
+```
+
+#### 启动 OpenAI 风格 API
+
+```bash
+llamafactory-cli api examples/inference/llama3_lora_sft.yaml
+```
+
+### 杂项
+
+#### 使用 GaLore 进行全参数训练
+
+```bash
+llamafactory-cli train examples/extras/galore/llama3_full_sft.yaml
+```
+
+#### 使用 APOLLO 进行全参数训练
+
+```bash
+llamafactory-cli train examples/extras/apollo/llama3_full_sft.yaml
+```
+
+#### 使用 BAdam 进行全参数训练
+
+```bash
+llamafactory-cli train examples/extras/badam/llama3_full_sft.yaml
+```
+
+#### 使用 Adam-mini 进行全参数训练
+
+```bash
+llamafactory-cli train examples/extras/adam_mini/qwen2_full_sft.yaml
+```
+
+#### 使用 Muon 进行全参数训练
+
+```bash
+llamafactory-cli train examples/extras/muon/qwen2_full_sft.yaml
+```
+
+#### LoRA+ 微调
+
+```bash
+llamafactory-cli train examples/extras/loraplus/llama3_lora_sft.yaml
+```
+
+#### PiSSA 微调
+
+```bash
+llamafactory-cli train examples/extras/pissa/llama3_lora_sft.yaml
+```
+
+#### 深度混合微调
+
+```bash
+llamafactory-cli train examples/extras/mod/llama3_full_sft.yaml
+```
+
+#### LLaMA-Pro 微调
+
+```bash
+bash examples/extras/llama_pro/expand.sh
+llamafactory-cli train examples/extras/llama_pro/llama3_freeze_sft.yaml
+```
+
+#### FSDP+QLoRA 微调
+
+```bash
+bash examples/extras/fsdp_qlora/train.sh
 ```

examples/accelerate/fsdp_config.yaml

@@ -5,18 +5,18 @@ downcast_bf16: 'no'
 fsdp_config:
   fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
   fsdp_backward_prefetch: BACKWARD_PRE
-  fsdp_cpu_ram_efficient_loading: true
   fsdp_forward_prefetch: false
-  fsdp_offload_params: true
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_offload_params: false
   fsdp_sharding_strategy: FULL_SHARD
   fsdp_state_dict_type: FULL_STATE_DICT
   fsdp_sync_module_states: true
-  fsdp_use_orig_params: false
+  fsdp_use_orig_params: true
 machine_rank: 0
 main_training_function: main
-mixed_precision: fp16
-num_machines: 1 # the number of nodes
-num_processes: 2 # the number of GPUs in all nodes
+mixed_precision: bf16  # or fp16
+num_machines: 1  # the number of nodes
+num_processes: 2  # the number of GPUs in all nodes
 rdzv_backend: static
 same_network: true
 tpu_env: []

examples/accelerate/fsdp_config_offload.yaml

@@ -0,0 +1,25 @@
+compute_environment: LOCAL_MACHINE
+debug: false
+distributed_type: FSDP
+downcast_bf16: 'no'
+fsdp_config:
+  fsdp_auto_wrap_policy: TRANSFORMER_BASED_WRAP
+  fsdp_backward_prefetch: BACKWARD_PRE
+  fsdp_forward_prefetch: false
+  fsdp_cpu_ram_efficient_loading: true
+  fsdp_offload_params: true  # offload may affect training speed
+  fsdp_sharding_strategy: FULL_SHARD
+  fsdp_state_dict_type: FULL_STATE_DICT
+  fsdp_sync_module_states: true
+  fsdp_use_orig_params: true
+machine_rank: 0
+main_training_function: main
+mixed_precision: bf16  # or fp16
+num_machines: 1  # the number of nodes
+num_processes: 2  # the number of GPUs in all nodes
+rdzv_backend: static
+same_network: true
+tpu_env: []
+tpu_use_cluster: false
+tpu_use_sudo: false
+use_cpu: false

examples/accelerate/master_config.yaml

@@ -1,18 +0,0 @@
-compute_environment: LOCAL_MACHINE
-debug: false
-distributed_type: MULTI_GPU
-downcast_bf16: 'no'
-gpu_ids: all
-machine_rank: 0
-main_process_ip: 192.168.0.1
-main_process_port: 29555
-main_training_function: main
-mixed_precision: fp16
-num_machines: 2 # the number of nodes
-num_processes: 16 # the number of GPUs in all nodes
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false

examples/accelerate/single_config.yaml

@@ -1,16 +0,0 @@
-compute_environment: LOCAL_MACHINE
-debug: false
-distributed_type: MULTI_GPU
-downcast_bf16: 'no'
-gpu_ids: all
-machine_rank: 0
-main_training_function: main
-mixed_precision: fp16
-num_machines: 1 # the number of nodes
-num_processes: 4 # the number of GPUs in all nodes
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false

examples/accelerate/slave_config.yaml

@@ -1,18 +0,0 @@
-compute_environment: LOCAL_MACHINE
-debug: false
-distributed_type: MULTI_GPU
-downcast_bf16: 'no'
-gpu_ids: all
-machine_rank: 1
-main_process_ip: 192.168.0.1
-main_process_port: 29555
-main_training_function: main
-mixed_precision: fp16
-num_machines: 2 # the number of nodes
-num_processes: 16 # the number of GPUs in all nodes
-rdzv_backend: static
-same_network: true
-tpu_env: []
-tpu_use_cluster: false
-tpu_use_sudo: false
-use_cpu: false

examples/extras/MoD/sft.sh

@@ -1,33 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../../data \
-    --template default \
-    --finetuning_type full \
-    --mixture_of_depths convert \
-    --output_dir ../../../saves/LLaMA2-7B/mod/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 8 \
-    --optim paged_adamw_8bit \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --plot_loss \
-    --pure_bf16

examples/extras/adam_mini/qwen2_full_sft.yaml

@@ -0,0 +1,43 @@
+### model
+model_name_or_path: Qwen/Qwen2-1.5B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+use_adam_mini: true
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: qwen
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/qwen2-1_5b/full/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-5
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/apollo/llama3_full_sft.yaml

@@ -0,0 +1,48 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+use_apollo: true
+apollo_layerwise: true  # choices: [true, false], use false for DDP training
+apollo_target: all
+apollo_rank: 128
+apollo_scale: 32.0
+apollo_scale_type: channel
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/full/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 1  # use 1 for layerwise apollo
+learning_rate: 1.0e-5
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+pure_bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/badam/llama3_full_sft.yaml

@@ -0,0 +1,46 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+use_badam: true
+badam_mode: layer
+badam_switch_mode: ascending
+badam_switch_interval: 50
+badam_verbose: 2
+# deepspeed: examples/deepspeed/ds_z3_config.json
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/full/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-5
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/badam/sft.sh

@@ -1,35 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../../data \
-    --template default \
-    --finetuning_type full \
-    --use_badam \
-    --badam_switch_mode descending \
-    --badam_switch_block_every 50 \
-    --badam_verbose 2 \
-    --output_dir ../../../saves/LLaMA2-7B/badam/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 8 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --plot_loss \
-    --pure_bf16

examples/extras/fsdp_qlora/llama3_lora_sft.yaml

@@ -0,0 +1,45 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+quantization_bit: 4
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: lora
+lora_rank: 8
+lora_target: all
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/lora/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-4
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/fsdp_qlora/sft.sh

@@ -1,40 +0,0 @@
-#!/bin/bash
-
-pip install "transformers>=4.39.1"
-pip install "accelerate>=0.28.0"
-pip install "bitsandbytes>=0.43.0"
-
-CUDA_VISIBLE_DEVICES=0,1 accelerate launch \
-    --config_file ../../accelerate/fsdp_config.yaml \
-    ../../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-70b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../../data \
-    --template default \
-    --finetuning_type lora \
-    --lora_target q_proj,v_proj \
-    --output_dir ../../../saves/LLaMA2-70B/lora/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 4 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --ddp_timeout 180000000 \
-    --quantization_bit 4 \
-    --plot_loss \
-    --fp16

examples/extras/fsdp_qlora/train.sh

@@ -0,0 +1,6 @@
+#!/bin/bash
+# DO NOT use GPTQ/AWQ model in FSDP+QLoRA
+
+CUDA_VISIBLE_DEVICES=0,1 accelerate launch \
+    --config_file examples/accelerate/fsdp_config.yaml \
+    src/train.py examples/extras/fsdp_qlora/llama3_lora_sft.yaml

examples/extras/galore/llama3_full_sft.yaml

@@ -0,0 +1,47 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+use_galore: true
+galore_layerwise: true  # choices: [true, false], use false for DDP training
+galore_target: all
+galore_rank: 128
+galore_scale: 2.0
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/full/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 1  # use 1 for layerwise galore
+learning_rate: 1.0e-5
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+pure_bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/galore/sft.sh

@@ -1,36 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../../data \
-    --template default \
-    --finetuning_type full \
-    --use_galore \
-    --galore_layerwise \
-    --galore_target mlp,self_attn \
-    --galore_rank 128 \
-    --galore_scale 2.0 \
-    --output_dir ../../../saves/LLaMA2-7B/galore/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 1 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --plot_loss \
-    --pure_bf16

examples/extras/llama_pro/expand.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 
-python ../../../scripts/llama_pro.py \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --output_dir ../../../models/llama2-7b-pro \
+python scripts/llama_pro.py \
+    --model_name_or_path meta-llama/Meta-Llama-3-8B-Instruct \
+    --output_dir models/llama3-8b-pro \
     --num_expand 8

examples/extras/llama_pro/llama3_freeze_sft.yaml

@@ -0,0 +1,45 @@
+### model
+model_name_or_path: models/llama3-8b-pro
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: freeze
+freeze_trainable_layers: 8
+freeze_trainable_modules: all
+use_llama_pro: true
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b-pro/freeze/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-4
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/llama_pro/sft.sh

@@ -1,34 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path ../../../models/llama2-7b-pro \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../../data \
-    --template default \
-    --finetuning_type freeze \
-    --name_module_trainable all \
-    --num_layer_trainable 8 \
-    --use_llama_pro \
-    --output_dir ../../../saves/LLaMA2-7B-Pro/lora/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 8 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --plot_loss \
-    --fp16

examples/extras/loraplus/llama3_lora_sft.yaml

@@ -0,0 +1,45 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: lora
+lora_rank: 8
+lora_target: all
+loraplus_lr_ratio: 16.0
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/lora/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-4
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/loraplus/sft.sh

@@ -1,33 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../data \
-    --template default \
-    --finetuning_type lora \
-    --lora_target q_proj,v_proj \
-    --loraplus_lr_ratio 16.0 \
-    --output_dir ../../saves/LLaMA2-7B/loraplus/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 8 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --plot_loss \
-    --fp16

examples/extras/mod/llama3_full_sft.yaml

@@ -0,0 +1,44 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+mixture_of_depths: convert
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b-mod/full/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+optim: paged_adamw_8bit
+learning_rate: 1.0e-5
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+pure_bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/muon/qwen2_full_sft.yaml

@@ -0,0 +1,43 @@
+### model
+model_name_or_path: Qwen/Qwen2-1.5B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: full
+use_muon: true
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: qwen
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/qwen2-1_5b/full/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-5
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/extras/nlg_eval/llama3_lora_predict.yaml

@@ -0,0 +1,31 @@
+# The batch generation can be SLOW using this config.
+# For faster inference, we recommend to use `scripts/vllm_infer.py`.
+
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+adapter_name_or_path: saves/llama3-8b/lora/sft
+trust_remote_code: true
+
+### method
+stage: sft
+do_predict: true
+finetuning_type: lora
+
+### dataset
+eval_dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 50
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/lora/predict
+overwrite_output_dir: true
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### eval
+per_device_eval_batch_size: 1
+predict_with_generate: true
+ddp_timeout: 180000000

examples/extras/pissa/init.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+python scripts/pissa_init.py \
+    --model_name_or_path meta-llama/Meta-Llama-3-8B-Instruct \
+    --output_dir models/llama3-8b-pissa

examples/extras/pissa/llama3_lora_sft.yaml

@@ -0,0 +1,47 @@
+### model
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+trust_remote_code: true
+
+### method
+stage: sft
+do_train: true
+finetuning_type: lora
+lora_rank: 8
+lora_target: all
+pissa_init: true
+pissa_iter: 16
+pissa_convert: true
+
+### dataset
+dataset: identity,alpaca_en_demo
+template: llama3
+cutoff_len: 2048
+max_samples: 1000
+overwrite_cache: true
+preprocessing_num_workers: 16
+dataloader_num_workers: 4
+
+### output
+output_dir: saves/llama3-8b/lora/sft
+logging_steps: 10
+save_steps: 500
+plot_loss: true
+overwrite_output_dir: true
+save_only_model: false
+report_to: none  # choices: [none, wandb, tensorboard, swanlab, mlflow]
+
+### train
+per_device_train_batch_size: 1
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-4
+num_train_epochs: 3.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+# val_size: 0.1
+# per_device_eval_batch_size: 1
+# eval_strategy: steps
+# eval_steps: 500

examples/full_multi_gpu/multi_node.sh

@@ -1,38 +0,0 @@
-#!/bin/bash
-
-python -m torch.distributed.run \
-    --nproc_per_node $NPROC_PER_NODE \
-    --nnodes $NNODES \
-    --node_rank $RANK \
-    --master_addr $MASTER_ADDR \
-    --master_port $MASTER_PORT \
-    ../../src/train_bash.py \
-    --deepspeed ../deepspeed/ds_z3_config.json \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../data \
-    --template default \
-    --finetuning_type full \
-    --output_dir ../../saves/LLaMA2-7B/full/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 2 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --ddp_timeout 180000000 \
-    --plot_loss \
-    --fp16

examples/full_multi_gpu/predict.sh

@@ -1,18 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../src/train_bash.py \
-    --stage sft \
-    --do_predict \
-    --model_name_or_path ../../saves/LLaMA2-7B/full/sft \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../data \
-    --template default \
-    --finetuning_type full \
-    --output_dir ../../saves/LLaMA2-7B/full/predict \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_eval_batch_size 1 \
-    --max_samples 20 \
-    --predict_with_generate

examples/full_multi_gpu/single_node.sh

@@ -1,32 +0,0 @@
-#!/bin/bash
-
-deepspeed --num_gpus 4 ../../src/train_bash.py \
-    --deepspeed ../deepspeed/ds_z3_config.json \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../data \
-    --template default \
-    --finetuning_type full \
-    --output_dir ../../saves/LLaMA2-7B/full/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 2 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --ddp_timeout 180000000 \
-    --plot_loss \
-    --fp16

examples/inference/api_demo.sh

@@ -1,7 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 API_PORT=8000 python ../../src/api_demo.py \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --adapter_name_or_path ../../saves/LLaMA2-7B/lora/sft \
-    --template default \
-    --finetuning_type lora

examples/inference/cli_demo.sh

@@ -1,7 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../src/cli_demo.py \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --adapter_name_or_path ../../saves/LLaMA2-7B/lora/sft \
-    --template default \
-    --finetuning_type lora

examples/inference/evaluate.sh

@@ -1,12 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../src/evaluate.py \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --adapter_name_or_path ../../saves/LLaMA2-7B/lora/sft \
-    --template fewshot \
-    --finetuning_type lora \
-    --task mmlu \
-    --split test \
-    --lang en \
-    --n_shot 5 \
-    --batch_size 4

examples/inference/llama3.yaml

@@ -0,0 +1,4 @@
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+template: llama3
+infer_backend: huggingface  # choices: [huggingface, vllm, sglang]
+trust_remote_code: true

examples/inference/llama3_full_sft.yaml

@@ -0,0 +1,4 @@
+model_name_or_path: saves/llama3-8b/full/sft
+template: llama3
+infer_backend: huggingface  # choices: [huggingface, vllm, sglang]
+trust_remote_code: true

examples/inference/llama3_lora_sft.yaml

@@ -0,0 +1,5 @@
+model_name_or_path: meta-llama/Meta-Llama-3-8B-Instruct
+adapter_name_or_path: saves/llama3-8b/lora/sft
+template: llama3
+infer_backend: huggingface  # choices: [huggingface, vllm, sglang]
+trust_remote_code: true

examples/inference/qwen2_5vl.yaml

@@ -0,0 +1,4 @@
+model_name_or_path: Qwen/Qwen2.5-VL-7B-Instruct
+template: qwen2_vl
+infer_backend: huggingface  # choices: [huggingface, vllm, sglang]
+trust_remote_code: true

examples/inference/web_demo.sh

@@ -1,7 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0 python ../../src/web_demo.py \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --adapter_name_or_path ../../saves/LLaMA2-7B/lora/sft \
-    --template default \
-    --finetuning_type lora

examples/lora_multi_gpu/multi_node.sh

@@ -1,35 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0,1,2,3 accelerate launch \
-    --config_file ../accelerate/master_config.yaml \
-    ../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../data \
-    --template default \
-    --finetuning_type lora \
-    --lora_target q_proj,v_proj \
-    --output_dir ../../saves/LLaMA2-7B/lora/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 2 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --ddp_timeout 180000000 \
-    --plot_loss \
-    --fp16

examples/lora_multi_gpu/single_node.sh

@@ -1,35 +0,0 @@
-#!/bin/bash
-
-CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 accelerate launch \
-    --config_file ../accelerate/single_config.yaml \
-    ../../src/train_bash.py \
-    --stage sft \
-    --do_train \
-    --model_name_or_path meta-llama/Llama-2-7b-hf \
-    --dataset alpaca_gpt4_en,glaive_toolcall \
-    --dataset_dir ../../data \
-    --template default \
-    --finetuning_type lora \
-    --lora_target q_proj,v_proj \
-    --output_dir ../../saves/LLaMA2-7B/lora/sft \
-    --overwrite_cache \
-    --overwrite_output_dir \
-    --cutoff_len 1024 \
-    --preprocessing_num_workers 16 \
-    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 1 \
-    --gradient_accumulation_steps 2 \
-    --lr_scheduler_type cosine \
-    --logging_steps 10 \
-    --warmup_steps 20 \
-    --save_steps 100 \
-    --eval_steps 100 \
-    --evaluation_strategy steps \
-    --load_best_model_at_end \
-    --learning_rate 5e-5 \
-    --num_train_epochs 3.0 \
-    --max_samples 3000 \
-    --val_size 0.1 \
-    --ddp_timeout 180000000 \
-    --plot_loss \
-    --fp16