Initial commit

.gemini/config.yaml

+have_fun: false
+code_review:
+  disable: false
+  comment_severity_threshold: HIGH
+  max_review_comments: -1
+  pull_request_opened:
+    help: false
+    summary: false
+    code_review: true
+ignore_patterns: []

.github/CODEOWNERS

+/docs @eric-haibin-lin @zhaochenyang20 @hongpeng-guo
+/docs/amd_tutorial @yushengsu-thu
+/docs/slang_multiturn @zhaochenyang20 @SwordFaith
+
+/recipe/dapo @tongyx361 @PeterSH6
+/recipe/spin @zhaochenyang20
+/recipe/sppo @zhaochenyang20
+
+/third_party/sglang @zhaochenyang20 @SwordFaith
+/third_party/vllm @PeterSH6 @wuxibin89
+/verl/single_controller @zw0610 @wuxibin89 @hongpeng-guo
+/verl/trainer @eric-haibin-lin @vermouth1992 @tongyx361 @PeterSH6
+/verl/workers/engine @eric-haibin-lin @vermouth1992 @ZihengJiang
+/verl/workers/roles @eric-haibin-lin @vermouth1992 @ZihengJiang
+/verl/workers/engine/fsdp @eric-haibin-lin @vermouth1992 @ZihengJiang
+/verl/workers/rollout/vllm_rollout @wuxibin89 @PeterSH6 @chenhaiq
+/verl/workers/rollout/sglang_rollout @zhaochenyang20 @SwordFaith @chenhaiq
+
+/tests/single_controller @zw0610 @wuxibin89
+/tests/trainer @eric-haibin-lin @vermouth1992 @tongyx361 @PeterSH6
+/tests/workers/rollout/vllm_rollout @wuxibin89 @PeterSH6 @chenhaiq

.github/PULL_REQUEST_TEMPLATE.md

+### What does this PR do?
+
+> Add **concise** overview of what this PR aims to achieve or accomplish. Reference related GitHub issues and PRs that help with the review.
+
+### Checklist Before Starting
+
+- [ ] Search for similar PRs. Paste at least one query link here: ...
+- [ ] Format the PR title as `[{modules}] {type}: {description}` (This will be checked by the CI)
+  - `{modules}` include `fsdp`, `megatron`, `sglang`, `vllm`, `rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`, `deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`, `model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data`
+  - If this PR involves multiple modules, separate them with `,` like `[megatron, fsdp, doc]`
+  - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test`
+  - If this PR breaks any API (CLI arguments, config, function signature, etc.), add `[BREAKING]` to the beginning of the title.
+  - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching`
+
+### Test
+
+> For changes that can not be tested by CI (e.g., algorithm implementation, new model support), validate by experiment(s) and show results like training curve plots, evaluation results, etc.
+
+### API and Usage Example
+
+> Demonstrate how the API changes if any, and provide usage example(s) if possible.
+
+```python
+# Add code snippet or script demonstrating how to use this
+```
+
+### Design & Code Changes
+
+> Demonstrate the high-level design if this PR is complex, and list the specific changes.
+
+### Checklist Before Submitting
+
+> [!IMPORTANT]
+> Please check all the following items before requesting a review, otherwise the reviewer might deprioritize this PR for review.
+
+- [ ] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md).
+- [ ] Apply [pre-commit checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting): `pre-commit install && pre-commit run --all-files --show-diff-on-failure --color=always`
+- [ ] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs).
+- [ ] Add unit or end-to-end test(s) to [the CI workflow](https://github.com/volcengine/verl/tree/main/.github/workflows) to cover all the code. If not feasible, explain why: ...
+- [ ] Once your PR is ready for CI, send a message in [the `ci-request` channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the `verl` Slack workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ). (If not accessible, please try [the Feishu group (飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).)

.github/dependabot.yml

+## Enabled the dependabot to check the dependencies of the project
+## Dependabot will open pull requests to update dependencies automatically
+
+version: 2
+updates:
+  - package-ecosystem: pip
+    directory: "/"
+    schedule:
+      interval: weekly
\ No newline at end of file

.github/workflows/README.md

+### Adding a New Workflow
+
+When adding a new workflow for continuous integration (CI), you have two runner options: a fixed runner or a machine from the vemlp.
+
+- **Fixed Runner**: To use a fixed runner, specify it in your workflow using the `runs-on` keyword, like `runs-on: [L20x8]`. 
+- **Vemlp Runner**: Opting for a Vemlp machine allows you to launch tasks elastically. 
+
+Here is a template to assist you. This template is designed for using Vemlp machines. Currently, for each workflow, you need to create a `setup` and a `cleanup` job. When using this template, the main parts you need to modify are the `IMAGE` environment variable and the specific `job steps`.
+
+```yaml
+name: Your Default Workflow
+
+on:
+  push:
+    branches:
+      - main
+      - v0.*
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      - ".github/workflows/template.yml"
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+permissions:
+  contents: read
+
+env:
+  IMAGE: "your vemlp image" # e.g. "verl-ci-cn-beijing.cr.volces.com/verlai/verl:app-verl0.4-vllm0.8.5-mcore0.12.2"
+  DYNAMIC_RUNNER_URL: "https://sd10g3clalm04ug7alq90.apigateway-cn-beijing.volceapi.com/runner" # public veFaas api
+
+jobs:
+  setup:
+    if: github.repository_owner == 'volcengine'
+    runs-on: ubuntu-latest
+    outputs:
+      runner-label: ${{ steps.create-runner.outputs.runner-label }}
+      task-id: ${{ steps.create-runner.outputs.task-id }}
+    steps:
+      - uses: actions/checkout@v4
+      - id: create-runner
+        uses: volcengine/vemlp-github-runner@v1 
+        with:
+          mode: "create"
+          faas-url: "${{ env.DYNAMIC_RUNNER_URL }}"
+          image: "${{ env.DEFAULT_IMAGE }}"
+
+  your_job:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'default-runner' }}"]
+    steps:
+      xxxx # your jobs
+
+  cleanup:
+    runs-on: ubuntu-latest
+    needs: [setup, your_job]
+    if: always()
+    steps:
+      - id: destroy-runner
+        uses: volcengine/vemlp-github-runner@v1
+        with:
+          mode: "destroy"
+          faas-url: "${{ env.DYNAMIC_RUNNER_URL }}"
+          task-id: "${{ needs.setup.outputs.task-id }}"
\ No newline at end of file

.github/workflows/check-pr-title.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+on:
+  pull_request:
+    types: [opened, edited, synchronize]
+
+jobs:
+  check-title:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Set up Python
+        uses: actions/setup-python@v5
+        with:
+          python-version: '3.11'
+
+      - name: Run PR title checker
+        run: python3 tests/special_sanity/check_pr_title.py
+        env:
+          PR_TITLE: ${{ github.event.pull_request.title }}
+
+      - name: Run PR description checker
+        run: python3 tests/special_sanity/check_pr_description.py
+        env:
+          PR_TITLE: ${{ github.event.pull_request.title }}
+          GITHUB_EVENT_PATH: ${{ github.event_path }}

.github/workflows/checkpoint_converter.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+name: checkpoint_converter
+# latest version: Megatron-LM core_r0.11.0 https://github.com/NVIDIA/Megatron-LM/tree/core_r0.11.0
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # FSDP
+      - "!verl/workers/**/*dp_*.py"
+      # Entrypoints
+      - ".github/workflows/checkpoint_converter.yml"
+      - ".github/workflows/e2e_ppo_trainer_megatron.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_ppo_trainer_megatron.sh"
+      - "verl/trainer/main_ppo.py"
+      - "verl/trainer/config/ppo_megatron_trainer.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  checkpoint_converter:
+    runs-on: [L20x8]
+    timeout-minutes: 20 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test]
+      - name: Download Model to Use
+        run: |
+          huggingface-cli download Qwen/Qwen2.5-0.5B --local-dir ${HOME}/models/Qwen/Qwen2.5-0.5B
+          huggingface-cli download deepseek-ai/deepseek-coder-1.3b-instruct --local-dir ${HOME}/models/deepseek-ai/deepseek-coder-1.3b-instruct
+          export HF_HUB_OFFLINE=1
+      - name: Running Huggingface to Megatron dist_ckpt converter (Qwen/Qwen2.5-0.5B)
+        run: |
+          ray stop --force
+          python scripts/converter_hf_to_mcore.py --hf_model_path=${HOME}/models/Qwen/Qwen2.5-0.5B --output_path checkpoints/Qwen/Qwen2.5-0.5B --test
+      - name: Running Huggingface to Megatron dist_ckpt converter (deepseek-ai/deepseek-coder-1.3b-instruct)
+        run: |
+          ray stop --force
+          python scripts/converter_hf_to_mcore.py --hf_model_path=${HOME}/models/deepseek-ai/deepseek-coder-1.3b-instruct --output_path checkpoints/deepseek-ai/deepseek-coder-1.3b-instruct --test
+      - name: Clean up
+        run: |
+          rm -rf checkpoints
+  checkpoint_converter_large_moe_models:
+    runs-on: [L20x8]
+    timeout-minutes: 30 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+      HF_ENDPOINT: "https://hf-mirror.com"
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test]
+      - name: Download Model to Use
+        run: |
+          huggingface-cli download Qwen/Qwen1.5-MoE-A2.7B-Chat --local-dir ${HOME}/models/Qwen/Qwen1.5-MoE-A2.7B-Chat
+          export HF_HUB_OFFLINE=1
+      - name: Running Huggingface to Megatron dist_ckpt CPU converter (Qwen/Qwen1.5-MoE-A2.7B-Chat)
+        run: |
+          ray stop --force
+          python scripts/converter_hf_to_mcore.py --hf_model_path=${HOME}/models/Qwen/Qwen1.5-MoE-A2.7B-Chat --output_path checkpoints/Qwen/Qwen1.5-MoE-A2.7B-Chat --use_cpu_initialization
+      - name: Running distributed Huggingface to Megatron dist_ckpt CPU converter (Qwen/Qwen1.5-MoE-A2.7B-Chat)
+        run: |
+          ray stop --force
+          torchrun --nproc_per_node 8 --nnodes 1 scripts/converter_hf_to_mcore.py --hf_model_path=${HOME}/models/Qwen/Qwen1.5-MoE-A2.7B-Chat --output_path checkpoints/Qwen/Qwen1.5-MoE-A2.7B-Chat_dist --use_cpu_initialization
+      - name: clean up
+        run: |
+          rm -rf checkpoints

.github/workflows/cpu_unit_tests.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: cpu_unit_tests
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      - .github/workflows/cpu_unit_tests.yml
+      - "!recipe/**/*.py"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  cpu_unit_tests:
+    runs-on: [L20x8]
+    timeout-minutes: 20 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-vllm0.9.1-mcore0.12.2-te2.2
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip install -e .[test,prime,geo]
+          pip install --upgrade "ray>=2.40.0" pillow
+      - name: Running CPU unit tests
+        run: |
+          [ ! -d "$HOME/verl-data" ] && huggingface-cli download verl-team/gsm8k-v0.4.1 --repo-type dataset --local-dir ~/verl-data/gsm8k
+          python3 examples/data_preprocess/geo3k.py
+          echo '[pytest]' > pytest.ini
+          echo 'python_files = *_on_cpu.py' >> pytest.ini
+          pytest -s -x --asyncio-mode=auto tests/
\ No newline at end of file

.github/workflows/disabled/e2e_prime.yml

+name: e2e_prime
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - disabled_ci
+  pull_request:
+    branches:
+      - disabled_ci
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Other recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Home
+      - "recipe/prime"
+      # Entrypoints
+      - ".github/workflows/e2e_prime.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_prime.sh"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  e2e_prime:
+    runs-on: [L20x8]
+    timeout-minutes: 50 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: whatcanyousee/verl:ngc-cu124-vllm0.8.5-sglang0.4.6.post5-mcore0.12.0-te2.3
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu]
+      - name: Prepare gsm8k dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E with prime alg
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_prime.sh

.github/workflows/doc.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: doc_test
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      - "docs/**"
+      - .github/workflows/doc.yml
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read      # for checkout
+  pages: write        # for deploy-pages
+  id-token: write     # for deploy-pages
+
+jobs:
+  doc_test:
+    runs-on: ubuntu-latest
+    timeout-minutes: 5 # Increase this timeout value as needed
+    strategy:
+      matrix:
+        python-version: ["3.10"]
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # v5.3.0
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install the current repository
+        run: |
+          pip install -e .[test]
+          pip install -r docs/requirements-docs.txt
+
+      - name: Run doc make html
+        run: |
+          cd docs 
+          make clean
+          make html SPHINXOPTS="--keep-going -w _build/sphinx.log"
+          if grep -q ": ERROR:" _build/sphinx.log; then
+            echo "🚨 Sphinx doc build contained ERRORs - see _build/sphinx.log"
+            exit 1
+          fi
+          if grep -q "WARNING: document isn't included in any toctree" _build/sphinx.log; then
+            echo "🚨 Sphinx doc build contained WARNING. Please include newly added docs in index.rst. See _build/sphinx.log for details"
+            exit 1
+          fi
+          if grep -q "WARNING: Inline emphasis" _build/sphinx.log; then
+            echo "🚨 Sphinx doc build contained WARNING. Please check inline emphasis is correct. See _build/sphinx.log for details"
+            exit 1
+          fi
+          if grep -q "WARNING: Definition list ends without a blank line" _build/sphinx.log; then
+            echo "🚨 Sphinx doc build contained WARNING. Please check if the indentation is correct. See _build/sphinx.log for details"
+            exit 1
+          fi

.github/workflows/e2e_ascend.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: e2e_ascend
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+  pull_request:
+    branches:
+      - main
+    paths:
+      - "**/*.py"
+      - "requirements-npu.txt"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # Entrypoints
+      - ".github/workflows/e2e_ascend.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "examples/data_preprocess/geo3k.py"
+      - "tests/special_e2e/ppo_trainer"
+      - "verl/trainer/main_ppo.py"
+      - "verl/trainer/config/ppo_trainer.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    name: verl Ascend test (self-host)
+    runs-on: [self-hosted, npu-0]
+    timeout-minutes: 40 # Increase this timeout value as needed
+    container:
+      image: crispig/verl_npu:cann8.1rc1-py3.10-torch2.5.1-vllm-ascend0.7.3.post1-250616
+      volumes:
+        - /usr/local/dcmi:/usr/local/dcmi
+        - /usr/local/bin/npu-smi:/usr/local/bin/npu-smi
+        - /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/
+        - /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info
+        - /etc/ascend_install.info:/etc/ascend_install.info
+        # Use self-host cache speed up pip and model download
+        # - /home/action/actions-runner/_work/cache:/github/home/.cache/
+      options: >-
+        --device /dev/davinci0
+        --device /dev/davinci_manager
+        --device /dev/devmm_svm
+        --device /dev/hisi_hdc
+        --network host
+        --privileged
+        --shm-size 16g
+    env: 
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - name: Check npu and CANN info
+        run: |
+          cat /usr/local/Ascend/ascend-toolkit/latest/"$(uname -i)"-linux/ascend_toolkit_install.info
+          npu-smi info
+      - name: Checkout volcengine/verl repo
+        uses: actions/checkout@v4
+      - name: Install the current repository
+        run: |
+          pip3 install hf_transfer peft
+          pip3 install -r requirements-npu.txt
+          pip install -e .
+      - name: Install torchviison
+        run: |
+          pip install torchvision==0.20.1+cpu --index-url https://download.pytorch.org/whl/cpu
+      - name: Prepare gsm8k dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Prepare geo3k dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/geo3k.py
+      - name: Running gsm8k e2e training tests with peft sft on ASCEND NPU
+        run: |
+          ray stop --force
+          bash tests/special_npu/run_qwen2_5_05b_sft_peft_sp2.sh
+          rm -rf $HOME/ckpts
+      - name: Running gsm8k e2e training tests with GRPO on ASCEND NPU
+        run: |
+          ray stop --force
+          bash tests/special_npu/run_qwen2_5_05b_grpo.sh
+          rm -rf $HOME/ckpts
+      - name: Running geo3k e2e training tests with GRPO on ASCEND NPU
+        run: |
+          ray stop --force
+          bash tests/special_npu/run_qwen2_5_vl_3b_npu.sh
+          rm -rf $HOME/ckpts
+      - name: Running gsm8k e2e training tests with DAPO on ASCEND NPU
+        run: |
+          ray stop --force
+          bash tests/special_npu/run_qwen2_5_05b_dapo.sh
+          rm -rf $HOME/ckpts

.github/workflows/e2e_dapo.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: e2e_dapo
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  # For push, for now only anti-patterns are specified so it is more conservative
+  # and achieves higher coverage.
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "verl/*.py"
+      # Other entrypoints
+      - "!examples/*trainer*"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      - "!recipe/**"
+      - "recipe/dapo"
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Other recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Home
+      - "recipe/dapo"
+      # Entrypoints
+      - ".github/workflows/e2e_dapo.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_dapo.sh"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  e2e_dapo:
+    runs-on: [L20x8]
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-vllm0.9.1-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running the E2E test with the DAPO algorithm
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_dapo.sh

.github/workflows/e2e_eval_aime24.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: e2e_eval_aime24
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  # For push, for now only anti-patterns are specified so it is more conservative
+  # and achieves higher coverage.
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!*.md"
+      - "!docker/**"
+      - "!docs/**"
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      - "!recipe/**"
+      - "recipe/r1"
+      - "!recipe/r1/README.md"
+  pull_request:
+    branches:
+      - main
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!*.md"
+      - "!docker/**"
+      - "!docs/**"
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Home
+      - "recipe/r1"
+      - "!recipe/r1/README.md"
+      # Other recipes
+      - "!recipe/**"
+      # Entrypoints
+      - ".github/workflows/e2e_eval_aime24.yml"
+      - "tests/special_e2e/run_r1_distill_qwen_aime24_eval.sh"
+      - "verl/trainer/main_generation.py"
+      - "verl/trainer/config/generation.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+env:
+  IMAGE: "verl-ci-cn-beijing.cr.volces.com/verlai/verl:app-verl0.5-vllm0.9.1-mcore0.12.2-te2.2"
+  DYNAMIC_RUNNER_ENDPOINT: "https://sd10g3clalm04ug7alq90.apigateway-cn-beijing.volceapi.com/runner"
+
+jobs:
+  setup:
+    if: github.repository_owner == 'volcengine'
+    runs-on: ubuntu-latest
+    outputs:
+      runner-label: ${{ steps.create-runner.outputs.runner-label }}
+      mlp-task-id: ${{ steps.create-runner.outputs.mlp-task-id }}
+    steps:
+      - uses: actions/checkout@v4
+      - id: create-runner
+        uses: volcengine/vemlp-github-runner@v1 
+        with:
+          mode: "create"
+          faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+          mlp-image: "${{ env.IMAGE }}"
+  
+  e2e_eval_aime24:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu,math]
+          pip3 install math-verify
+      - name: Prepare aime24 dataset
+        run: |
+          ray stop --force
+          python3 recipe/r1/data_process.py --task aime2024
+      - name: Running generation and evaluation in AIME 2024
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_r1_distill_qwen_aime24_eval.sh
+
+  cleanup:
+      runs-on: ubuntu-latest
+      needs: [setup, e2e_eval_aime24]
+      if: always()
+      steps:
+        - id: destroy-runner
+          uses: volcengine/vemlp-github-runner@v1
+          with:
+            mode: "destroy"
+            faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+            mlp-task-id: "${{ needs.setup.outputs.mlp-task-id }}"
\ No newline at end of file

.github/workflows/e2e_genrm_remote.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: e2e_genrm_remote
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      - "tests/**"
+      - "!recipe/**"
+      - "recipe/genrm_remote"
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Other recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Home
+      - "recipe/genrm_remote"
+      - "!recipe/genrm_remote/README.md"
+      # Entrypoints
+      - ".github/workflows/e2e_genrm_remote.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_genrm_remote.sh"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  e2e_genrm_remote:
+    runs-on: [L20x8]
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: whatcanyousee/verl:ngc-cu124-vllm0.8.5-sglang0.4.6.post5-mcore0.12.0-te2.3
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running the E2E test with the Generative Reward Model
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_genrm_remote.sh

.github/workflows/e2e_one_step_off_policy.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+
+name: e2e_one_step_off_policy
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  # For push, for now only anti-patterns are specified so it is more conservative
+  # and achieves higher coverage.
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      - "!**/*.md"
+      - "!**/*.sh"
+      # Other entrypoints
+      - "!examples/*trainer*"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      - "!recipe/**"
+      - "recipe/one_step_off_policy"
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      - "!**/*.md"
+      - "!**/*.sh"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Other recipes
+      - "!recipe/**"
+      # Home
+      - "recipe/one_step_off_policy"
+      # Entrypoints
+      - ".github/workflows/e2e_one_step_off_policy.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_one_step_off_policy.sh"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  # Test FSDP2 strategy
+  e2e_one_step_off_policy_fsdp2:
+    runs-on: [L20x8]
+    timeout-minutes: 50 # Increase timeout for async training
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+      ACTOR_STRATEGY: "fsdp2"
+    container:
+      image: verlai/verl:app-verl0.4-vllm0.8.5-mcore0.12.1
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running the E2E test with one_step_off_policy algorithm (FSDP2)
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_one_step_off_policy.sh
+
+  # Test Megatron strategy
+  e2e_one_step_off_policy_megatron:
+    runs-on: [L20x8]
+    timeout-minutes: 50 # Increase timeout for async training
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+      ACTOR_STRATEGY: "megatron"
+    container:
+      image: verlai/verl:app-verl0.4-vllm0.8.5-mcore0.12.1
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running the E2E test with one_step_off_policy algorithm (Megatron)
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_one_step_off_policy.sh
+

.github/workflows/e2e_ppo_trainer.yml

+name: e2e_ppo_trainer
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  # For push, for now only anti-patterns are specified so it is more conservative
+  # and achieves higher coverage.
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!**/*.md"
+      - "!docker/**"
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Docs
+      - "!docs/**"
+      # Recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Entrypoints
+      - ".github/workflows/e2e_ppo_trainer.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "examples/data_preprocess/geo3k.py"
+      - "tests/special_e2e/ppo_trainer"
+      - "verl/trainer/main_ppo.py"
+      - "verl/trainer/config/ppo_trainer.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+jobs:
+  pre_commit_for_ppo:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.12"]
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # v5.3.0
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install the current repository
+        run: |
+          pip install -e .
+      - name: Set ruff --output-format=github
+        run: |
+          sed -i 's/--output-format=full/--output-format=github/' .pre-commit-config.yaml
+          git add .pre-commit-config.yaml
+      - uses: pre-commit/action@v3.0.1
+        with:
+          extra_args: "" # Overriding default "--all-files"
+
+  e2e_ppo_trainer_vllm:
+    runs-on: [L20x8]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-vllm0.9.1-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,vllm]
+      - name: Prepare GSM8K dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/gsm8k.py
+      # HF sanity
+      - name: Running GSM8K E2E training tests on 1 L20 GPU with hf for sanity
+        run: |
+          ray stop --force
+          bash tests/special_e2e/ppo_trainer/run_single_gpu.sh
+      # HF sanity
+      - name: Running GSM8K E2E training tests on 1 L20 GPU with engine interface for sanity.
+        run: |
+          ray stop --force
+          bash tests/special_e2e/ppo_trainer/run_single_gpu_with_engine.sh
+      # Function RM
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm with validation and saving (FSDP_SIZE=8)
+        run: |
+          ray stop --force
+          VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 SAVE_HF_MODEL=True VERL_EXP_NAME="qwen2.5-0.5b-function-reward-minimal-fsdp-size8" bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm after resuming
+        run: |
+          ray stop --force
+          RESUME_MODE=auto VERL_EXP_NAME="qwen2.5-0.5b-function-reward-minimal-fsdp-size8" bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Test merging FSDP checkpoints (Qwen Actor) 
+        run: |
+          exp_name="qwen2.5-0.5b-function-reward-minimal-fsdp-size8"
+          python -m verl.model_merger test --backend fsdp --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm with validation and saving (DDP_SIZE=2, FSDP_SIZE=4)
+        run: |
+          ray stop --force
+          VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 SAVE_HF_MODEL=True FSDP_SIZE=4 VERL_EXP_NAME="qwen2.5-0.5b-function-reward-minimal-ddp-size2-fsdp-size4" bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Test merging DDP+FSDP checkpoints (Qwen Actor) 
+        run: |
+          exp_name="qwen2.5-0.5b-function-reward-minimal-ddp-size2-fsdp-size4"
+          python -m verl.model_merger test --backend fsdp --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm with validation and saving (FSDP2)
+        run: |
+          ray stop --force
+          VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 SAVE_HF_MODEL=True VERL_EXP_NAME="qwen2.5-0.5b-function-reward-minimal-fsdp2-size8" STRATEGY=fsdp2 bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Test merging FSDP2 checkpoints (Qwen Actor) 
+        run: |
+          exp_name="qwen2.5-0.5b-function-reward-minimal-fsdp2-size8"
+          python -m verl.model_merger test --backend fsdp --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+      - name: Running GSM8K E2E without rmpad using function rm
+        run: |
+          ray stop --force
+          RM_PAD=False bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm (GRPO)
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_KL=True bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm (ReMax)
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=remax USE_KL=True bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using customized reward function
+        run: |
+          ray stop --force
+          CUSTOM_REWARD_FN=True bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm with in-reward kl and kl loss
+        run: |
+          ray stop --force
+          USE_KL=True bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      # LoRA tests
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with grpo lora using function rm with use_shm
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_SHM=True LORA_RANK=32 LOAD_FORMAT=safetensors bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with grpo lora using function rm with use_shm and layered_summon
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_SHM=True LORA_RANK=32 LOAD_FORMAT=safetensors LAYERED_SUMMON=True TOTAL_TRAIN_STEPS=1 SAVE_FREQ=1 FSDP_SIZE=4 VERL_EXP_NAME="qwen2.5-0.5b-function-reward-minimal" bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Test GRPO LoRA checkpoints merging function
+        run: |
+          export EXP_NAME="qwen2.5-0.5b-function-reward-minimal"
+          ls checkpoints/verl-test/${EXP_NAME}/global_step_1/actor
+          cat checkpoints/verl-test/${EXP_NAME}/global_step_1/actor/huggingface/config.json
+          python3 -m verl.model_merger merge --backend fsdp --local_dir checkpoints/verl-test/${EXP_NAME}/global_step_1/actor/ --target_dir checkpoints/verl-test/${EXP_NAME}/global_step_1/actor/huggingface
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with grpo lora using function rm with use_shm and layered_summon with fsdp2
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_SHM=True LORA_RANK=32 LOAD_FORMAT=safetensors LAYERED_SUMMON=True STRATEGY=fsdp2 bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      # Model RM
+      - name: Running GRPO GSM8K E2E training tests with FSDP on 8 L20 GPUs (DeepSeek)
+        run: |
+          ray stop --force
+          MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E with rmpad using model rm
+        run: |
+          ray stop --force
+          bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E without rmpad using model rm
+        run: |
+          ray stop --force
+          RM_PAD=False bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E with rmpad using model rm and ulysses sp=2
+        run: |
+          ray stop --force
+          SP_SIZE=2 bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E with rmpad using model rm and dynamic batch size
+        run: |
+          ray stop --force
+          SEQ_BALANCE=True bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E with rmpad using model rm with Liger Kernel enabled
+        run: |
+          ray stop --force
+          LIGER=True bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E with rmpad using model rm with Fused Kernel enabled
+        run: |
+          ray stop --force
+          FUSED_KERNELS=True bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E with rmpad using model rm with Fused Kernel enabled
+        run: |
+          ray stop --force
+          FUSED_KERNEL=True FUSED_KERNEL_BACKEND=triton bash tests/special_e2e/ppo_trainer/run_model_reward.sh
+      - name: Running GSM8K E2E training tests on vllm async
+        run: |
+          ray stop --force
+          export VLLM_USE_V1=1
+          ray start --head
+          TOTAL_TRAIN_STEPS=2 ENGINE=vllm ROLLOUT_MODE=async bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+
+  e2e_ppo_trainer_vllm_vlm:
+    runs-on: [L20x8]
+    needs: pre_commit_for_ppo
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-vllm0.9.1-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=50g # Visual dataloader requires large memory
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test,gpu,vllm,geo,trl]
+          pip install "transformers[hf_xet]<4.53.0" # Fix for transformers 4.53.0
+      # Geo3k
+      - name: Prepare GEO3K dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/geo3k.py
+      - name: Running GEO3K VLM GRPO E2E training tests on 8 L20 GPUs with rmpad using function rm
+        run: |
+          ray stop --force
+          TRAIN_FILES=$HOME/data/geo3k/train.parquet VAL_FILES=$HOME/data/geo3k/test.parquet \
+            MAX_PROMPT_LEN=1536 MAX_RESPONSE_LEN=1536 \
+            MODEL_ID=Qwen/Qwen2-VL-2B-Instruct \
+            ADV_ESTIMATOR=grpo RM_PAD=True USE_KL=True ENABLE_CHUNKED_PREFILL=False \
+            SP_SIZE=2 \
+            bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+
+      - name: Running GEO3K VLM PPO E2E training tests on 8 L20 GPUs with rmpad using function rm
+        run: |
+          ray stop --force
+          TRAIN_FILES=$HOME/data/geo3k/train.parquet VAL_FILES=$HOME/data/geo3k/test.parquet \
+            MAX_PROMPT_LEN=1536 MAX_RESPONSE_LEN=1536 \
+            MODEL_ID=Qwen/Qwen2-VL-2B-Instruct \
+            ADV_ESTIMATOR=gae RM_PAD=True USE_KL=True ENABLE_CHUNKED_PREFILL=False \
+            SP_SIZE=2 \
+            bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GEO3K VLM GRPO E2E lora training tests on 8 L20 GPUs with rmpad using function rm
+        run: |
+          ray stop --force
+          TRAIN_FILES=$HOME/data/geo3k/train.parquet VAL_FILES=$HOME/data/geo3k/test.parquet \
+            MAX_PROMPT_LEN=1536 MAX_RESPONSE_LEN=1536 \
+            MODEL_ID=Qwen/Qwen2-VL-2B-Instruct \
+            ADV_ESTIMATOR=grpo RM_PAD=True USE_KL=True ENABLE_CHUNKED_PREFILL=False \
+            SP_SIZE=2 \
+            LORA_RANK=32 LORA_EXCLUDE=".*visual.*" \
+            bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+          
+
+  e2e_ppo_trainer_sglang:
+    runs-on: [L20x8]
+    needs: pre_commit_for_ppo
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test,gpu,sglang] --no-deps
+      - name: Prepare gsm8k dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm and save ckpt
+        run: |
+          ray stop --force
+          ENGINE=sglang bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GSM8K E2E training tests on sglang async
+        run: |
+          ray stop --force
+          TOTAL_TRAIN_STEPS=2 ENGINE=sglang ROLLOUT_MODE=async bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+
+  e2e_ppo_trainer_sglang_multiturn_with_tool:
+    runs-on: [L20x8]
+    needs: pre_commit_for_ppo
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test,gpu,sglang] --no-deps
+      - name: Prepare gsm8k dataset with tool
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/gsm8k_multiturn_w_tool.py --local_dir $HOME/data/gsm8k_verl_sgl_multi_turn_preprocessed
+      - name: Running GSM8K with tool E2E training tests on 8 L20 GPUs with rmpad using function rm and save ckpt with sglang
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_gsm8k_fsdp_sgl_multiturn_w_tool.sh
+      - name: Running GSM8K with tool E2E training tests with FSDP2
+        run: |
+          ray stop --force
+          FSDP_STRATEGY=fsdp2 bash tests/special_e2e/run_gsm8k_fsdp_sgl_multiturn_w_tool.sh
+
+  e2e_ppo_trainer_sglang_vlm:
+    runs-on: [L20x8]
+    needs: pre_commit_for_ppo
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=50g # Visual dataloader requires large memory
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test,geo,gpu,sglang]
+      # Geo3k
+      - name: Prepare GEO3K dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/geo3k.py
+      - name: Running GEO3K VLM E2E training tests on 8 L20 GPUs with rmpad using function rm
+        run: |
+          ray stop --force
+          TRAIN_FILES=$HOME/data/geo3k/train.parquet VAL_FILES=$HOME/data/geo3k/test.parquet \
+            MAX_PROMPT_LEN=1536 MAX_RESPONSE_LEN=1536 \
+            MODEL_ID=Qwen/Qwen2-VL-2B-Instruct \
+            ADV_ESTIMATOR=grpo RM_PAD=True USE_KL=True ENABLE_CHUNKED_PREFILL=False \
+            ENGINE=sglang GPU_MEMORY_UTILIZATION=0.6 ACTOR_FSDP_PARAM_OFFLOAD=True \
+            ACTOR_FSDP_OPTIMIZER_OFFLOAD=True REF_FSDP_PARAM_OFFLOAD=True \
+            bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GEO3K VLM E2E with rmpad using torch fused kernel (Qwen2.5-VL)
+        run: |
+          ray stop --force
+          FUSED_KERNELS=True TRAIN_FILES=$HOME/data/geo3k/train.parquet VAL_FILES=$HOME/data/geo3k/test.parquet \
+            MAX_PROMPT_LEN=1536 MAX_RESPONSE_LEN=1536 \
+            MODEL_ID=Qwen/Qwen2.5-VL-3B-Instruct \
+            ADV_ESTIMATOR=grpo RM_PAD=True USE_KL=True ENABLE_CHUNKED_PREFILL=False \
+            ENGINE=sglang GPU_MEMORY_UTILIZATION=0.6 ACTOR_FSDP_PARAM_OFFLOAD=True \
+            ACTOR_FSDP_OPTIMIZER_OFFLOAD=True REF_FSDP_PARAM_OFFLOAD=True \
+            bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+      - name: Running GEO3K VLM E2E with rmpad using triton fused kernel (Qwen2.5-VL)
+        run: |
+          ray stop --force
+          FUSED_KERNELS=True FUSED_KERNEL_BACKEND=triton \
+            TRAIN_FILES=$HOME/data/geo3k/train.parquet VAL_FILES=$HOME/data/geo3k/test.parquet \
+            MAX_PROMPT_LEN=1536 MAX_RESPONSE_LEN=1536 \
+            MODEL_ID=Qwen/Qwen2.5-VL-3B-Instruct \
+            ADV_ESTIMATOR=grpo RM_PAD=True USE_KL=True ENABLE_CHUNKED_PREFILL=False \
+            ENGINE=sglang GPU_MEMORY_UTILIZATION=0.6 ACTOR_FSDP_PARAM_OFFLOAD=True \
+            ACTOR_FSDP_OPTIMIZER_OFFLOAD=True REF_FSDP_PARAM_OFFLOAD=True \
+            bash tests/special_e2e/ppo_trainer/run_function_reward.sh
+
+  e2e_ppo_trainer_sglang_vlm_multiturn_with_tool:
+    runs-on: [L20x8]
+    needs: pre_commit_for_ppo
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test,geo,gpu,sglang]
+      - name: Prepare geo3k dataset with tool
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/geo3k_multiturn_w_tool.py --local_dir $HOME/data/geo3k_verl_sgl_multi_turn_preprocessed
+      - name: Running GEO3K with tool E2E training tests on 8 L20 GPUs with rmpad using function rm and save ckpt with sglang
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_geo3k_fsdp_sgl_multiturn_w_tool.sh
+      - name: Running GEO3K with tool E2E training tests with FSDP2
+        run: |
+          ray stop --force
+          FSDP_STRATEGY=fsdp2 bash tests/special_e2e/run_geo3k_fsdp_sgl_multiturn_w_tool.sh

.github/workflows/e2e_ppo_trainer_megatron_sglang.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+name: e2e_ppo_trainer_megatron_sglang
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch.
+  # For push, for now only anti-patterns are specified so it is more conservative
+  # and achieves higher coverage.
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # FSDP
+      - "!verl/workers/**/*dp_*.py"
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!docker/**"
+      # Docs
+      - "!**/*.md"
+      - "!docs/**"
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # FSDP
+      - "!verl/workers/**/*dp_*.py"
+      # Entrypoints
+      - ".github/workflows/e2e_ppo_trainer_megatron_sglang.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "examples/data_preprocess/geo3k.py"
+      - "tests/special_e2e/run_ppo_trainer_megatron.sh"
+      - "verl/trainer/main_ppo.py"
+      - "verl/trainer/config/ppo_megatron_trainer.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+env:
+  IMAGE: "verl-ci-cn-beijing.cr.volces.com/verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2"
+  DYNAMIC_RUNNER_ENDPOINT: "https://sd10g3clalm04ug7alq90.apigateway-cn-beijing.volceapi.com/runner"
+
+jobs:
+  setup:
+    if: github.repository_owner == 'volcengine'
+    runs-on: ubuntu-latest
+    outputs:
+      runner-label: ${{ steps.create-runner.outputs.runner-label }}
+      mlp-task-id: ${{ steps.create-runner.outputs.mlp-task-id }}
+    steps:
+      - uses: actions/checkout@v4
+      - id: create-runner
+        uses: volcengine/vemlp-github-runner@v1
+        with:
+          mode: "create"
+          faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+          mlp-image: "${{ env.IMAGE }}"
+
+  e2e_ppo_trainer_megatron-deepseek:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          ENGINE=sglang ALL_OFFLOAD=True SAVE_FREQ=1 MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          export VLLM_USE_V1=1
+          ray start --head
+          ENGINE=sglang MODE=async RESUME_MODE=auto MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct TOTAL_TRAIN_STEPS=2 bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Test Megatron checkpoints merging function (DeepSeek Actor and Critic)
+        run: |
+          exp_name="deepseek-coder-1.3b-instruct-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: Running GRPO GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Deepseek)
+        run: |
+          ray stop --force
+          ENGINE=sglang ADV_ESTIMATOR=grpo USE_DYNAMIC_BSZ=False MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-qwen3:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen3) with validation and saving
+        run: |
+          ray stop --force
+          ENGINE=sglang ALL_OFFLOAD=True VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 MODEL_ID=Qwen/Qwen3-0.6B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen3) testing learning rate scheduler
+        run: |
+          ray stop --force
+          ENGINE=sglang LR_WARMUP_STEPS=1 TOTAL_TRAIN_STEPS=2 MODEL_ID=Qwen/Qwen3-0.6B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+
+      - name: Test Megatron checkpoints merging function (Qwen3 Actor and Critic)
+        run: |
+          exp_name="qwen3-0.6b-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --tie-word-embedding --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-different-train-infer-tp-qwen-tie-embedding:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with tie-embedding Megatron (Qwen) with train tp > infer tp
+        run: |
+          ray stop --force
+          ENGINE=sglang VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 TRAIN_TP=2 INFER_TP=1 MODEL_ID=Qwen/Qwen2.5-1.5B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen) with  train tp < infer tp
+        run: |
+          ray stop --force
+          ENGINE=sglang VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 TRAIN_TP=1 INFER_TP=2 MODEL_ID=Qwen/Qwen2.5-1.5B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-qwen-override-transformer-config:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Download Model to Use
+        run: |
+          huggingface-cli download Qwen/Qwen2.5-0.5B --local-dir ${HOME}/models/Qwen/Qwen2.5-0.5B
+          export HF_HUB_OFFLINE=1
+      - name: Prepare dist_ckpt of Qwen2.5-0.5B, uneven layer distribution only supports dist_ckpt
+        run: |
+          python3 scripts/converter_hf_to_mcore.py --hf_model_path ${HOME}/models/Qwen/Qwen2.5-0.5B --output_path checkpoints/verl-test/qwen2.5-0.5b-megatron
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen)
+        run: |
+          ray stop --force
+          ENGINE=sglang SAVE_FREQ=1 COMMON_PP=4 COMMON_VPP=null COMMON_CP=1 SKIP_SAVE_HF_MODEL=1 bash tests/special_e2e/run_ppo_trainer_megatron.sh +actor_rollout_ref.actor.megatron.override_transformer_config.num_layers_in_first_pipeline_stage=8 +actor_rollout_ref.actor.megatron.override_transformer_config.num_layers_in_last_pipeline_stage=4 actor_rollout_ref.actor.megatron.use_dist_checkpointing=true actor_rollout_ref.actor.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron actor_rollout_ref.ref.megatron.use_dist_checkpointing=true actor_rollout_ref.ref.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron critic.megatron.use_dist_checkpointing=true critic.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron reward_model.megatron.use_dist_checkpointing=true reward_model.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron
+          cp -r checkpoints checkpoints-dut
+          ENGINE=sglang SAVE_FREQ=1 COMMON_PP=4 COMMON_VPP=null COMMON_CP=1 bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Test Megatron checkpoints merging function (Qwen Actor and Critic)
+        run: |
+          exp_name="qwen2.5-0.5b-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --tie-word-embedding --local_dir checkpoints-dut/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints-dut/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-deepseek-override-transformer-config:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          ENGINE=sglang SAVE_FREQ=1 MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct COMMON_PP=2 COMMON_VPP=null bash tests/special_e2e/run_ppo_trainer_megatron.sh +actor_rollout_ref.actor.megatron.override_transformer_config.account_for_embedding_in_pipeline_split=true +actor_rollout_ref.actor.megatron.override_transformer_config.account_for_loss_in_pipeline_split=true
+      - name: Test Megatron checkpoints merging function (DeepSeek Actor and Critic)
+        run: |
+          exp_name="deepseek-coder-1.3b-instruct-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-moe-expert-parallel:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_DUMMY_MODEL=True DUMMY_MODEL_CONFIG_PATH=tests/special_e2e/ppo_trainer/expert_parallel/qwen2moe_minimal.json \
+          PPO_MAX_TOKEN_LEN=512 FWD_MAX_TOKEN_LEN=512 \
+          MAX_PROMPT_LENGTH=256 MAX_RESPONSE_LENGTH=256 \
+          MODEL_ID=Qwen/Qwen1.5-MoE-A2.7B-Chat \
+          ENGINE=sglang COMMON_PP=2 COMMON_VPP=null COMMON_CP=1 COMMON_TP=4 COMMON_EP=4 COMMON_ETP=1 INFER_TP=8 \
+          USE_DIST_CKPT=True ALL_OFFLOAD=True SKIP_SAVE_HF_MODEL=1 bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-qwen2_5vl-3b:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare Geo3k dataset
+        run: |
+          python3 examples/data_preprocess/geo3k.py
+      - name: Prepare dist_ckpt of Qwen2.5-VL-3B, only supports dist_ckpt
+        run: |
+          python3 scripts/converter_hf_to_mcore.py --hf_model_path ${HOME}/models/Qwen/Qwen2.5-VL-3B-Instruct --output_path checkpoints/verl-test/qwen2.5-vl-3b-megatron
+      - name: Running Geo3k E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen)
+        run: |
+          ray stop --force
+          ENGINE=sglang TRAIN_FILES=${HOME}/data/geo3k/train.parquet VAL_FILES=${HOME}/data/geo3k/test.parquet MAX_PROMPT_LENGTH=1024 MAX_RESPONSE_LENGTH=2048  MODEL_ID=Qwen/Qwen2.5-VL-3B-Instruct ADV_ESTIMATOR=grpo USE_DYNAMIC_BSZ=False SKIP_SAVE_HF_MODEL=1 COMMON_PP=4 COMMON_VPP=null COMMON_CP=1 COMMON_TP=2 USE_DIST_CKPT=true DIST_CKPT_PATH=checkpoints/verl-test/qwen2.5-vl-3b-megatron bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+
+  cleanup:
+    runs-on: ubuntu-latest
+    needs:
+      [
+        setup,
+        e2e_ppo_trainer_megatron-deepseek,
+        e2e_ppo_trainer_megatron-qwen3,
+        e2e_ppo_trainer_megatron-different-train-infer-tp-qwen-tie-embedding,
+        e2e_ppo_trainer_megatron-qwen-override-transformer-config,
+        e2e_ppo_trainer_megatron-deepseek-override-transformer-config,
+        e2e_ppo_trainer_megatron-moe-expert-parallel,
+        e2e_ppo_trainer_megatron-qwen2_5vl-3b,
+      ]
+    if: always()
+    steps:
+      - id: destroy-runner
+        uses: volcengine/vemlp-github-runner@v1
+        with:
+          mode: "destroy"
+          faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+          mlp-task-id: "${{ needs.setup.outputs.mlp-task-id }}"

.github/workflows/e2e_ppo_trainer_megatron_vllm.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+name: e2e_ppo_trainer_megatron_vllm
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch.
+  # For push, for now only anti-patterns are specified so it is more conservative
+  # and achieves higher coverage.
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # FSDP
+      - "!verl/workers/**/*dp_*.py"
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!docker/**"
+      # Docs
+      - "!**/*.md"
+      - "!docs/**"
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # FSDP
+      - "!verl/workers/**/*dp_*.py"
+      # Entrypoints
+      - ".github/workflows/e2e_ppo_trainer_megatron_vllm.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "examples/data_preprocess/geo3k.py"
+      - "tests/special_e2e/run_ppo_trainer_megatron.sh"
+      - "verl/trainer/main_ppo.py"
+      - "verl/trainer/config/ppo_megatron_trainer.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+env:
+  IMAGE: "verl-ci-cn-beijing.cr.volces.com/verlai/verl:app-verl0.5-vllm0.9.1-mcore0.12.2-te2.2"
+  DYNAMIC_RUNNER_ENDPOINT: "https://sd10g3clalm04ug7alq90.apigateway-cn-beijing.volceapi.com/runner"
+
+jobs:
+  setup:
+    if: github.repository_owner == 'volcengine'
+    runs-on: ubuntu-latest
+    outputs:
+      runner-label: ${{ steps.create-runner.outputs.runner-label }}
+      mlp-task-id: ${{ steps.create-runner.outputs.mlp-task-id }}
+    steps:
+      - uses: actions/checkout@v4
+      - id: create-runner
+        uses: volcengine/vemlp-github-runner@v1
+        with:
+          mode: "create"
+          faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+          mlp-image: "${{ env.IMAGE }}"
+
+  e2e_ppo_trainer_megatron-deepseek:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          ALL_OFFLOAD=True SAVE_FREQ=1 MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          export VLLM_USE_V1=1
+          ray start --head
+          MODE=async USE_FUSED_KERNELS=True RESUME_MODE=auto MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct TOTAL_TRAIN_STEPS=2 bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Test Megatron checkpoints merging function (DeepSeek Actor and Critic)
+        run: |
+          exp_name="deepseek-coder-1.3b-instruct-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: Test Megatron distributed checkpoints merging function (DeepSeek)
+        run: |
+          exp_name="deepseek-coder-1.3b-instruct-megatron-gsm8k-minimal"
+          torchrun --nproc_per_node 4 --nnodes 1  -m verl.model_merger merge --backend megatron --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --target_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/hf_model
+      - name: Running GRPO GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Deepseek)
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_DYNAMIC_BSZ=False MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-qwen3:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen3) with validation and saving
+        run: |
+          ray stop --force
+          ALL_OFFLOAD=True VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 MODEL_ID=Qwen/Qwen3-0.6B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen3) testing learning rate scheduler
+        run: |
+          ray stop --force
+          LR_WARMUP_STEPS=1 TOTAL_TRAIN_STEPS=2 MODEL_ID=Qwen/Qwen3-0.6B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+
+      - name: Test Megatron checkpoints merging function (Qwen3 Actor and Critic)
+        run: |
+          exp_name="qwen3-0.6b-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --tie-word-embedding --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-different-train-infer-tp-qwen-tie-embedding:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with tie-embedding Megatron (Qwen) with train tp > infer tp
+        run: |
+          ray stop --force
+          VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 TRAIN_TP=2 INFER_TP=1 MODEL_ID=Qwen/Qwen2.5-1.5B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen) with  train tp < infer tp
+        run: |
+          ray stop --force
+          VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 TRAIN_TP=1 INFER_TP=2 ALL_OFFLOAD=True MODEL_ID=Qwen/Qwen2.5-1.5B bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-qwen-override-transformer-config:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Download Model to Use
+        run: |
+          huggingface-cli download Qwen/Qwen2.5-0.5B --local-dir ${HOME}/models/Qwen/Qwen2.5-0.5B
+          export HF_HUB_OFFLINE=1
+      - name: Prepare dist_ckpt of Qwen2.5-0.5B, uneven layer distribution only supports dist_ckpt
+        run: |
+          python3 scripts/converter_hf_to_mcore.py --hf_model_path ${HOME}/models/Qwen/Qwen2.5-0.5B --output_path checkpoints/verl-test/qwen2.5-0.5b-megatron
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen)
+        run: |
+          ray stop --force
+          SAVE_FREQ=1 COMMON_PP=4 COMMON_VPP=null COMMON_CP=1 SKIP_SAVE_HF_MODEL=1 bash tests/special_e2e/run_ppo_trainer_megatron.sh +actor_rollout_ref.actor.megatron.override_transformer_config.num_layers_in_first_pipeline_stage=8 +actor_rollout_ref.actor.megatron.override_transformer_config.num_layers_in_last_pipeline_stage=4 actor_rollout_ref.actor.megatron.use_dist_checkpointing=true actor_rollout_ref.actor.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron actor_rollout_ref.ref.megatron.use_dist_checkpointing=true actor_rollout_ref.ref.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron critic.megatron.use_dist_checkpointing=true critic.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron reward_model.megatron.use_dist_checkpointing=true reward_model.megatron.dist_checkpointing_path=checkpoints/verl-test/qwen2.5-0.5b-megatron
+          cp -r checkpoints checkpoints-dut
+          SAVE_FREQ=1 COMMON_PP=4 COMMON_VPP=null COMMON_CP=1 bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: Test Megatron checkpoints merging function (Qwen Actor and Critic)
+        run: |
+          exp_name="qwen2.5-0.5b-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --tie-word-embedding --local_dir checkpoints-dut/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints-dut/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-deepseek-override-transformer-config:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          SAVE_FREQ=1 MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct COMMON_PP=2 COMMON_VPP=null bash tests/special_e2e/run_ppo_trainer_megatron.sh +actor_rollout_ref.actor.megatron.override_transformer_config.account_for_embedding_in_pipeline_split=true +actor_rollout_ref.actor.megatron.override_transformer_config.account_for_loss_in_pipeline_split=true
+      - name: Test Megatron checkpoints merging function (DeepSeek Actor and Critic)
+        run: |
+          exp_name="deepseek-coder-1.3b-instruct-megatron-gsm8k-minimal"
+          python -m verl.model_merger test --backend megatron --local_dir checkpoints/verl-test/${exp_name}/global_step_1/actor --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/actor/huggingface
+          python -m verl.model_merger test --backend megatron --is-value-model --local_dir checkpoints/verl-test/${exp_name}/global_step_1/critic --test_hf_dir checkpoints/verl-test/${exp_name}/global_step_1/critic/huggingface
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-moe-expert-parallel:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+          pip3 install mbridge
+      - name: Prepare GSM8K dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (DeepSeek)
+        run: |
+          ray stop --force
+          ADV_ESTIMATOR=grpo USE_DUMMY_MODEL=True DUMMY_MODEL_CONFIG_PATH=tests/special_e2e/ppo_trainer/expert_parallel/qwen2moe_minimal.json \
+          PPO_MAX_TOKEN_LEN=512 FWD_MAX_TOKEN_LEN=512 \
+          MAX_PROMPT_LENGTH=256 MAX_RESPONSE_LENGTH=256 \
+          MODEL_ID=Qwen/Qwen1.5-MoE-A2.7B-Chat USE_MBRIDGE=True \
+          COMMON_PP=2 COMMON_VPP=null COMMON_CP=1 COMMON_TP=4 COMMON_EP=4 COMMON_ETP=1 INFER_TP=8 \
+          USE_DIST_CKPT=True ALL_OFFLOAD=True SKIP_SAVE_HF_MODEL=1 bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+  e2e_ppo_trainer_megatron-qwen2_5vl-3b:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 60 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install --no-deps -e .[test]
+          pip3 install "transformers[hf_xet]<4.52.0"
+      - name: Prepare Geo3k dataset
+        run: |
+          python3 examples/data_preprocess/geo3k.py
+      - name: Prepare dist_ckpt of Qwen2.5-VL-3B, only supports dist_ckpt
+        run: |
+          python3 scripts/converter_hf_to_mcore.py --hf_model_path ${HOME}/models/Qwen/Qwen2.5-VL-3B-Instruct --output_path checkpoints/verl-test/qwen2.5-vl-3b-megatron
+      - name: Running Geo3k E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen)
+        run: |
+          ray stop --force
+          TRAIN_FILES=${HOME}/data/geo3k/train.parquet VAL_FILES=${HOME}/data/geo3k/test.parquet MAX_PROMPT_LENGTH=1024 MAX_RESPONSE_LENGTH=2048  MODEL_ID=Qwen/Qwen2.5-VL-3B-Instruct ADV_ESTIMATOR=grpo USE_DYNAMIC_BSZ=False SKIP_SAVE_HF_MODEL=1 COMMON_PP=4 COMMON_VPP=null COMMON_CP=1 COMMON_TP=2 USE_DIST_CKPT=true DIST_CKPT_PATH=checkpoints/verl-test/qwen2.5-vl-3b-megatron bash tests/special_e2e/run_ppo_trainer_megatron.sh
+      - name: clean up
+        run: |
+          rm -rf checkpoints
+
+  cleanup:
+    runs-on: ubuntu-latest
+    needs:
+      [
+        setup,
+        e2e_ppo_trainer_megatron-deepseek,
+        e2e_ppo_trainer_megatron-qwen3,
+        e2e_ppo_trainer_megatron-different-train-infer-tp-qwen-tie-embedding,
+        e2e_ppo_trainer_megatron-qwen-override-transformer-config,
+        e2e_ppo_trainer_megatron-deepseek-override-transformer-config,
+        e2e_ppo_trainer_megatron-moe-expert-parallel,
+        e2e_ppo_trainer_megatron-qwen2_5vl-3b,
+      ]
+    if: always()
+    steps:
+      - id: destroy-runner
+        uses: volcengine/vemlp-github-runner@v1
+        with:
+          mode: "destroy"
+          faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+          mlp-task-id: "${{ needs.setup.outputs.mlp-task-id }}"

.github/workflows/e2e_sft.yml

+# # Tests layout
+
+# Each folder under tests/ corresponds to a test category for a sub-namespace in verl. For instance:
+# - `tests/trainer` for testing functionality related to `verl/trainer`
+# - `tests/models` for testing functionality related to `verl/models`
+# - ...
+
+# There are a few folders with `special_` prefix, created for special purposes:
+# - `special_distributed`: unit tests that must run with multiple GPUs
+# - `special_e2e`: end-to-end tests with training/generation scripts
+# - `special_npu`: tests for NPUs
+# - `special_sanity`: a suite of quick sanity tests
+# - `special_standalone`: a set of test that are designed to run in dedicated environments
+
+# Accelerators for tests 
+# - By default tests are run with GPU available, except for the ones under `special_npu`, and any test script whose name ends with `on_cpu.py`.
+# - For test scripts with `on_cpu.py` name suffix would be tested on CPU resources in linux environment.
+
+# # Workflow layout
+
+# All CI tests are configured by yaml files in `.github/workflows/`. Here's an overview of all test configs:
+# 1. A list of always triggered CPU sanity tests: `check-pr-title.yml`, `secrets_scan.yml`, `check-pr-title,yml`, `pre-commit.yml`, `doc.yml`
+# 2. Some heavy multi-GPU unit tests, such as `model.yml`, `vllm.yml`, `sgl.yml`
+# 3. End-to-end tests: `e2e_*.yml`
+# 4. Unit tests
+#   - `cpu_unit_tests.yml`, run pytest on all scripts with file name pattern `tests/**/test_*_on_cpu.py`
+#   - `gpu_unit_tests.yml`, run pytest on all scripts with file without the `on_cpu.py` suffix.
+#   - Since cpu/gpu unit tests by default runs all tests under `tests`, please make sure tests are manually excluded in them when
+#     - new workflow yaml is added to `.github/workflows`
+#     - new tests are added to workflow mentioned in 2.
+
+name: e2e_sft
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Entrypoints
+      - ".github/workflows/e2e_sft.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/sft"
+      - "verl/trainer/fsdp_sft_trainer.py"
+      - "verl/trainer/config/sft_trainer.yaml"
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+env:
+  IMAGE: "verl-ci-cn-beijing.cr.volces.com/verlai/verl:base-verl0.5-cu126-cudnn9.8-torch2.7.1-fa2.7.4"
+  DYNAMIC_RUNNER_ENDPOINT: "https://sd10g3clalm04ug7alq90.apigateway-cn-beijing.volceapi.com/runner"
+
+jobs:
+  setup:
+      if: github.repository_owner == 'volcengine'
+      runs-on: ubuntu-latest
+      outputs:
+        runner-label: ${{ steps.create-runner.outputs.runner-label }}
+        mlp-task-id: ${{ steps.create-runner.outputs.mlp-task-id }}
+      steps:
+        - uses: actions/checkout@v4
+        - id: create-runner
+          uses: volcengine/vemlp-github-runner@v1 
+          with:
+            mode: "create"
+            faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+            mlp-image: "${{ env.IMAGE }}"
+  e2e_sft:
+    needs: setup
+    runs-on: ["${{ needs.setup.outputs.runner-label || 'L20x8' }}"]
+    timeout-minutes: 20 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install peft
+          pip3 install --no-deps -e .[test,gpu]
+      - name: Prepare gsm8k dataset
+        run: |
+          ray stop --force
+          python3 examples/data_preprocess/gsm8k.py
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with rmpad using function rm
+        run: |
+          ray stop --force
+          bash tests/special_e2e/sft/run_sft.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs w/o rmpad using function rm
+        run: |
+          ray stop --force
+          RM_PAD=False bash tests/special_e2e/sft/run_sft.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with sequence parallism
+        run: |
+          ray stop --force
+          SP_SIZE=2 bash tests/special_e2e/sft/run_sft.sh
+      - name: Check loss difference between sequence parallel vs. default implementation
+        run: |
+          ray stop --force
+          ENTRYPOINT="tests/special_e2e/sft/test_sp_loss_match.py" SP_SIZE=2 bash tests/special_e2e/sft/run_sft.sh
+      - name: Running GSM8K E2E training tests on 8 L20 GPUs with sequence parallism and liger
+        run: |
+          ray stop --force
+          SP_SIZE=2 LIGER=True bash tests/special_e2e/sft/run_sft.sh
+      - name: Running GSM8K E2E training tests with LoRA
+        run: |
+          ray stop --force
+          LORA_RANK=32 bash tests/special_e2e/sft/run_sft.sh
+      - name: Run GSM8K E2E training and resume tests resuming from the checkpoint manager
+        run: |
+          ray stop --force
+          LORA_RANK=32 RESUME_MODE=auto TOTAL_TRAIN_STEP=2 bash tests/special_e2e/sft/run_sft.sh
+      # TODO: multiturn
+  
+  cleanup:
+    runs-on: ubuntu-latest
+    needs: [setup, e2e_sft]
+    if: always()
+    steps:
+      - id: destroy-runner
+        uses: volcengine/vemlp-github-runner@v1
+        with:
+          mode: "destroy"
+          faas-url: "${{ env.DYNAMIC_RUNNER_ENDPOINT }}"
+          mlp-task-id: "${{ needs.setup.outputs.mlp-task-id }}"

.github/workflows/e2e_spin.yml

+name: e2e_spin
+
+on:
+  # Trigger the workflow on push or pull request,
+  # but only for the main branch
+  push:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Other recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Home
+      - "recipe/spin"
+      # Entrypoints
+      - ".github/workflows/e2e_spin.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_spin.sh"
+      - "!examples"
+  pull_request:
+    branches:
+      - main
+      - v0.*
+    paths:
+      - "**/*.py"
+      # Other entrypoints
+      - "!examples/**"
+      - "!tests/**"
+      - "!verl/trainer/main_*.py"
+      - "!verl/trainer/fsdp_sft_trainer.py"
+      # Other recipes
+      - "!recipe/**"
+      # Megatron
+      - "!verl/workers/**/megatron_*.py"
+      # Home
+      - "recipe/spin"
+      # Entrypoints
+      - ".github/workflows/e2e_spin.yml"
+      - "examples/data_preprocess/gsm8k.py"
+      - "tests/special_e2e/run_spin.sh"
+      - "!examples"
+
+# Declare permissions just read content.
+permissions:
+  contents: read
+
+# Cancel jobs on the same ref if a new one is triggered
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+jobs:
+  e2e_spin:
+    runs-on: [L20x8]
+    timeout-minutes: 40 # Increase this timeout value as needed
+    env:
+      HTTP_PROXY: ${{ secrets.PROXY_HTTP }}
+      HTTPS_PROXY: ${{ secrets.PROXY_HTTPS }}
+      NO_PROXY: "localhost,127.0.0.1,hf-mirror.com"
+      HF_ENDPOINT: "https://hf-mirror.com"
+      HF_HUB_ENABLE_HF_TRANSFER: "0" # This is more stable
+    container:
+      image: verlai/verl:app-verl0.5-sglang0.4.8-mcore0.12.2-te2.2
+      options: --gpus all --shm-size=10g
+    steps:
+      - uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+      - name: Install the current repository
+        run: |
+          pip3 install -e .[test,gpu,sglang] --no-deps
+      - name: Prepare gsm8k dataset
+        run: |
+          python3 examples/data_preprocess/gsm8k.py --local_dir ./data/gsm8k
+      - name: Prepare Model checkpoint
+        run: |
+          huggingface-cli download Qwen/Qwen2.5-0.5B-Instruct --local-dir ./models/Qwen2.5-0.5B-Instruct
+      - name: Running the E2E test with the spin algorithm
+        run: |
+          ray stop --force
+          bash tests/special_e2e/run_spin.sh