[recipe, hardware] feat: Add GRPO with full weight updates for 1.5B models on a single GPU

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,40 +1,27 @@
 ### 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.
+> Adds a new recipe to perform GRPO with full weight-updates on 1.5B models.
 
 ### 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`
+- [X] Search for similar PRs. Paste at least one query link here: https://github.com/volcengine/verl/pull/3029
+- [X] Format the PR title as `[recipe, hardware] feat: Add GRPO with full weight updates for 1.5B models on a single GPU ` (This will be checked by the CI)
+  - `{modules}`: `recipe`, `hardware`
+  - `{type}`: `feat`
+  - No breaking API changes
 
-### 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.
+### Test
+![Training Curve 1.5b](../examples/tuning/1.5b/Qwen_math_grpo_training_curve_wandb.png)
+> Pass@1 for MATH500: 77.8
 
 ### 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).)
+- [X] Read the [Contribute Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md).
+- [X] 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`
+- [X] Add / Update [the documentation](https://github.com/volcengine/verl/tree/main/docs).
+- [X] 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: ...
+- [X] 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).)

examples/tuning/1.5b/qwen2.5math-1.5b_grpo_1_h100_fsdp_vllm.sh

@@ -0,0 +1,59 @@
+PROJECT_NAME=qwen2.5_Math_1.5b_dr_grpo
+LR=1e-5
+MODEL_NAME=Qwen/Qwen2.5-Math-1.5B
+MAX_NEW_TOKENS=3000
+MAX_PROMPT_LENGTH=1024 #1024 #512
+MODEL_ALIAS=${MODEL_NAME##*/}
+GROUP_SIZE=8
+CHOSEN_DEVICE=0 #0,1
+TOTAL_EPOCHS=2
+LOG_N_VAL_GENERATIONS=40 
+ADVANTAGE_ESTIMATOR="grpo"
+
+
+TIMESTAMP=$(date +"%Y%m%d_%H%M%S")
+RUN_NAME=${MODEL_ALIAS}_${LR}_${GROUP_SIZE}_${MAX_NEW_TOKENS}_${ADVANTAGE_ESTIMATOR}_${TIMESTAMP}
+
+CUDA_VISIBLE_DEVICES=$CHOSEN_DEVICE python3 -m verl.trainer.main_ppo \
+        algorithm.adv_estimator=$ADVANTAGE_ESTIMATOR \
+        data.train_files=data/MATH/train.parquet \
+        data.val_files=data/MATH/test.parquet \
+        data.train_batch_size=256 \
+        data.max_prompt_length=$MAX_PROMPT_LENGTH \
+        data.max_response_length=$MAX_NEW_TOKENS \
+        data.filter_overlong_prompts=True \
+        data.truncation='error' \
+        actor_rollout_ref.model.path=$MODEL_NAME \
+        actor_rollout_ref.actor.optim.lr=$LR \
+        actor_rollout_ref.actor.loss_agg_mode=seq-mean-token-sum-norm \
+        actor_rollout_ref.model.use_remove_padding=True \
+        actor_rollout_ref.actor.ppo_mini_batch_size=256 \
+        actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=4 \
+        actor_rollout_ref.actor.use_kl_loss=False \
+        actor_rollout_ref.actor.kl_loss_coef=0.001 \
+        actor_rollout_ref.actor.kl_loss_type=low_var_kl \
+        actor_rollout_ref.actor.entropy_coeff=0 \
+        actor_rollout_ref.model.enable_gradient_checkpointing=True \
+        actor_rollout_ref.actor.fsdp_config.param_offload=False \
+        actor_rollout_ref.actor.fsdp_config.optimizer_offload=False \
+        actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=4 \
+        actor_rollout_ref.rollout.tensor_model_parallel_size=1 \
+        actor_rollout_ref.rollout.name=vllm \
+        actor_rollout_ref.rollout.gpu_memory_utilization=0.4 \
+        actor_rollout_ref.rollout.n=$GROUP_SIZE \
+        actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4 \
+        actor_rollout_ref.ref.fsdp_config.param_offload=True \
+        actor_rollout_ref.rollout.temperature=1.0 \
+        actor_rollout_ref.rollout.top_p=1.0 \
+        algorithm.use_kl_in_reward=False \
+        algorithm.norm_adv_by_std_in_grpo=False\
+        trainer.critic_warmup=0 \
+        trainer.logger='["console","wandb"]' \
+        trainer.n_gpus_per_node=1 \
+        trainer.log_val_generations=$LOG_N_VAL_GENERATIONS\
+        trainer.nnodes=1 \
+        trainer.save_freq=20 \
+        trainer.test_freq=1 \
+        trainer.project_name=$PROJECT_NAME \
+        trainer.experiment_name=$RUN_NAME \
+        trainer.total_epochs=$TOTAL_EPOCHS \