Skip to content

[feat] reuse existing tests and remove the error config of reward models #1

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zzong2006 merged 1 commit into from
May 23, 2025
Merged

[feat] reuse existing tests and remove the error config of reward models #1

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
35 changes: 3 additions & 32 deletions .github/workflows/e2e_ppo_trainer_megatron.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,7 +65,7 @@ jobs:
- name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen) with validation and saving
run: |
ray stop --force
VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 bash tests/e2e/run_ppo_trainer_megatron.sh
ALL_OFFLOAD=True VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 bash tests/e2e/run_ppo_trainer_megatron.sh
- name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen) after resuming
run: |
ray stop --force
Expand Down Expand Up @@ -107,7 +107,7 @@ jobs:
- 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 bash tests/e2e/run_ppo_trainer_megatron.sh
ALL_OFFLOAD=True SAVE_FREQ=1 MODEL_ID=deepseek-ai/deepseek-coder-1.3b-instruct bash tests/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
Expand Down Expand Up @@ -149,7 +149,7 @@ jobs:
- name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen3) with validation and saving
run: |
ray stop --force
VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 MODEL_ID=Qwen/Qwen3-0.6B bash tests/e2e/run_ppo_trainer_megatron.sh
ALL_OFFLOAD=True VAL_BEFORE_TRAIN=True TEST_FREQ=1 SAVE_FREQ=1 MODEL_ID=Qwen/Qwen3-0.6B bash tests/e2e/run_ppo_trainer_megatron.sh
- name: Running GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron (Qwen3) after resuming
run: |
ray stop --force
Expand Down Expand Up @@ -305,34 +305,5 @@ jobs:
- name: clean up
run: |
rm -rf checkpoints
e2e_ppo_trainer_megatron-offload-qwen3:
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-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-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]
- name: Prepare GSM8K dataset
run: |
python3 examples/data_preprocess/gsm8k.py
- name: Running GRPO GSM8K E2E training tests with 3D parallelism on 8 L20 GPUs with Megatron and Offload
run: |
ray stop --force
COMMON_PARAM_OFFLOAD=True COMMON_GRAD_OFFLOAD=True COMMON_OPTIMIZER_OFFLOAD=True ADV_ESTIMATOR=grpo MODEL_ID=Qwen/Qwen3-0.6B bash tests/e2e/run_ppo_trainer_megatron.sh
- name: clean up
run: |
rm -rf checkpoints


11 changes: 4 additions & 7 deletions tests/e2e/run_ppo_trainer_megatron.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,9 +55,10 @@ RM_VPP=${RM_VPP:-$COMMON_VPP}
RM_CP=${RM_CP:-$COMMON_CP}
RM_TP=${RM_TP:-$TRAIN_TP}

COMMON_PARAM_OFFLOAD=${COMMON_PARAM_OFFLOAD:-False}
COMMON_GRAD_OFFLOAD=${COMMON_GRAD_OFFLOAD:-False}
COMMON_OPTIMIZER_OFFLOAD=${COMMON_OPTIMIZER_OFFLOAD:-False}
ALL_OFFLOAD=${ALL_OFFLOAD:-False}
COMMON_PARAM_OFFLOAD=${COMMON_PARAM_OFFLOAD:-$ALL_OFFLOAD}
COMMON_GRAD_OFFLOAD=${COMMON_GRAD_OFFLOAD:-$ALL_OFFLOAD}
COMMON_OPTIMIZER_OFFLOAD=${COMMON_OPTIMIZER_OFFLOAD:-$ALL_OFFLOAD}

ACTOR_PARAM_OFFLOAD=${ACTOR_PARAM_OFFLOAD:-$COMMON_PARAM_OFFLOAD}
ACTOR_GRAD_OFFLOAD=${ACTOR_GRAD_OFFLOAD:-$COMMON_GRAD_OFFLOAD}
Expand All @@ -66,9 +67,7 @@ REF_PARAM_OFFLOAD=${REF_PARAM_OFFLOAD:-$COMMON_PARAM_OFFLOAD}
CRITIC_PARAM_OFFLOAD=${CRITIC_PARAM_OFFLOAD:-$COMMON_PARAM_OFFLOAD}
CRITIC_GRAD_OFFLOAD=${CRITIC_GRAD_OFFLOAD:-$COMMON_GRAD_OFFLOAD}
CRITIC_OPTIMIZER_OFFLOAD=${CRITIC_OPTIMIZER_OFFLOAD:-$COMMON_OPTIMIZER_OFFLOAD}
RM_GRAD_OFFLOAD=${RM_GRAD_OFFLOAD:-$COMMON_GRAD_OFFLOAD}
RM_PARAM_OFFLOAD=${RM_PARAM_OFFLOAD:-$COMMON_PARAM_OFFLOAD}
RM_OPTIMIZER_OFFLOAD=${RM_OPTIMIZER_OFFLOAD:-$COMMON_OPTIMIZER_OFFLOAD}

CHECKPOINT_CONTENTS=['model','hf_model','optimizer','extra']
SKIP_SAVE_HF_MODEL=${SKIP_SAVE_HF_MODEL:-0}
Expand Down Expand Up @@ -134,8 +133,6 @@ python3 -m verl.trainer.main_ppo --config-path=config \
reward_model.megatron.context_parallel_size=$RM_CP \
reward_model.megatron.tensor_model_parallel_size=$RM_TP \
reward_model.megatron.param_offload=${RM_PARAM_OFFLOAD} \
reward_model.megatron.optimizer_offload=${RM_OPTIMIZER_OFFLOAD} \
reward_model.megatron.grad_offload=${RM_GRAD_OFFLOAD} \
algorithm.use_kl_in_reward=False \
algorithm.kl_penalty=kl \
algorithm.kl_ctrl.kl_coef=0.001 \
Expand Down
2 changes: 0 additions & 2 deletions verl/trainer/config/ppo_megatron_trainer.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -227,8 +227,6 @@ reward_model:
strategy: megatron
megatron:
param_offload: False
grad_offload: False
optimizer_offload: False
tensor_model_parallel_size: 1
expert_model_parallel_size: 1
expert_tensor_parallel_size: null
Expand Down