run_qwen3-236b_megatron.sh

#!/usr/bin/env bash
set -xeuo pipefail

# Note that we set the response length to 4k. This results in many truncations at the beginning.
# So the training dynamic acts as using RL to compress the math capabilities of QWen3 236b into 4k response instead of verbose thinking.
# We can achieve 0.5 on AIME'24 after 30 steps.

project_name='DAPO'
exp_name='DAPO-Qwen3-236b-megatron-0531a1'

adv_estimator=grpo

use_kl_in_reward=False
kl_coef=0.0
use_kl_loss=False
kl_loss_coef=0.0

clip_ratio_low=0.2
clip_ratio_high=0.28

max_prompt_length=$((1024 * 2))
max_response_length=$((1024 * 4))
enable_overlong_buffer=True
overlong_buffer_len=$((1024 * 4))
overlong_penalty_factor=0.1

loss_agg_mode="token-mean"

train_prompt_bsz=256
n_resp_per_prompt=4
train_prompt_mini_bsz=16

# H20 GPUs
NNODES=${NNODES:-32}

# Paths

RAY_DATA_HOME=${RAY_DATA_HOME:-"${HOME}/verl"}

MODEL_PATH=$RAY_DATA_HOME/models/Qwen3-235B-A22B
MCORE_MODEL_PATH=$RAY_DATA_HOME/models/Qwen3-235B-A22B_dist_ckpt_mcore/

# convert QWen3-235b-A22b to dist ckpt of mcore. Conversion process will take about 4 hours
# python scripts/converter_hf_to_mcore.py --hf_model_path $MODEL_PATH --output_path $MCORE_MODEL_PATH --use_cpu_initialization
CKPTS_DIR=$RAY_DATA_HOME/ckpt/${project_name}/${exp_name}
TRAIN_FILE=$RAY_DATA_HOME/dataset/dapo-math-17k.parquet
TEST_FILE=$RAY_DATA_HOME/dataset/aime-2024.parquet

# Algorithm
temperature=1.0
top_p=1.0
top_k=-1 # 0 for HF rollout, -1 for vLLM rollout
val_top_p=0.7

# Performance Related Parameter
offload=True
gen_tp=8
train_tp=4
train_ep=4
train_pp=8

python3 -m verl.trainer.main_ppo \
    --config-path=config \
    --config-name='ppo_megatron_trainer.yaml' \
    data.train_files="${TRAIN_FILE}" \
    data.val_files="${TEST_FILE}" \
    data.prompt_key=prompt \
    data.truncation='left' \
    data.max_prompt_length=${max_prompt_length} \
    data.max_response_length=${max_response_length} \
    data.train_batch_size=${train_prompt_bsz} \
    actor_rollout_ref.rollout.n=${n_resp_per_prompt} \
    algorithm.adv_estimator=${adv_estimator} \
    algorithm.use_kl_in_reward=${use_kl_in_reward} \
    algorithm.kl_ctrl.kl_coef=${kl_coef} \
    actor_rollout_ref.actor.use_kl_loss=${use_kl_loss} \
    actor_rollout_ref.actor.kl_loss_coef=${kl_loss_coef} \
    actor_rollout_ref.actor.clip_ratio_low=${clip_ratio_low} \
    actor_rollout_ref.actor.clip_ratio_high=${clip_ratio_high} \
    actor_rollout_ref.actor.clip_ratio_c=10.0 \
    actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=1 \
    actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=4 \
    actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=4 \
    actor_rollout_ref.model.path="${MODEL_PATH}" \
    actor_rollout_ref.actor.optim.lr=1e-6 \
    actor_rollout_ref.actor.optim.lr_warmup_steps=10 \
    actor_rollout_ref.actor.optim.weight_decay=0.1 \
    actor_rollout_ref.actor.ppo_mini_batch_size=${train_prompt_mini_bsz} \
    actor_rollout_ref.actor.megatron.param_offload=${offload} \
    actor_rollout_ref.actor.megatron.optimizer_offload=${offload} \
    actor_rollout_ref.actor.megatron.grad_offload=${offload} \
    actor_rollout_ref.actor.megatron.pipeline_model_parallel_size=${train_pp} \
    actor_rollout_ref.actor.megatron.tensor_model_parallel_size=${train_tp} \
    actor_rollout_ref.actor.megatron.expert_model_parallel_size=${train_ep} \
    actor_rollout_ref.actor.megatron.dist_checkpointing_path=${MCORE_MODEL_PATH} \
    actor_rollout_ref.actor.megatron.use_dist_checkpointing=True \
    +actor_rollout_ref.actor.megatron.override_transformer_config.num_layers_in_first_pipeline_stage=5 \
    +actor_rollout_ref.actor.megatron.override_transformer_config.num_layers_in_last_pipeline_stage=5 \
    actor_rollout_ref.actor.entropy_coeff=0 \
    actor_rollout_ref.actor.optim.clip_grad=1.0 \
    actor_rollout_ref.actor.loss_agg_mode=${loss_agg_mode} \
    actor_rollout_ref.rollout.gpu_memory_utilization=0.8 \
    actor_rollout_ref.rollout.tensor_model_parallel_size=${gen_tp} \
    actor_rollout_ref.rollout.enable_chunked_prefill=True \
    actor_rollout_ref.rollout.max_num_batched_tokens=$((max_prompt_length + max_response_length)) \
    actor_rollout_ref.rollout.temperature=${temperature} \
    actor_rollout_ref.rollout.top_p=${top_p} \
    actor_rollout_ref.rollout.top_k=${top_k} \
    actor_rollout_ref.rollout.val_kwargs.temperature=${temperature} \
    actor_rollout_ref.rollout.val_kwargs.top_p=${val_top_p} \
    actor_rollout_ref.rollout.val_kwargs.top_k=${top_k} \
    actor_rollout_ref.rollout.val_kwargs.do_sample=True \
    actor_rollout_ref.rollout.val_kwargs.n=1 \
    actor_rollout_ref.rollout.name=vllm \
    actor_rollout_ref.ref.megatron.pipeline_model_parallel_size=${train_pp} \
    actor_rollout_ref.ref.megatron.tensor_model_parallel_size=${train_tp} \
    actor_rollout_ref.ref.megatron.expert_model_parallel_size=${train_ep} \
    actor_rollout_ref.ref.megatron.param_offload=${offload} \
    actor_rollout_ref.ref.megatron.dist_checkpointing_path=${MCORE_MODEL_PATH} \
    actor_rollout_ref.ref.megatron.use_dist_checkpointing=True \
    reward_model.reward_manager=dapo \
    +reward_model.reward_kwargs.overlong_buffer_cfg.enable=${enable_overlong_buffer} \
    +reward_model.reward_kwargs.overlong_buffer_cfg.len=${overlong_buffer_len} \
    +reward_model.reward_kwargs.overlong_buffer_cfg.penalty_factor=${overlong_penalty_factor} \
    +reward_model.reward_kwargs.overlong_buffer_cfg.log=False \
    +reward_model.reward_kwargs.max_resp_len=${max_response_length} \
    trainer.logger='["console","wandb"]' \
    trainer.project_name="${project_name}" \
    trainer.experiment_name="${exp_name}" \
    trainer.n_gpus_per_node=8 \
    trainer.nnodes="${NNODES}" \
    trainer.val_before_train=False \
    trainer.test_freq=10 \
    trainer.save_freq=20 \
    trainer.total_epochs=10 \
    trainer.total_training_steps=100 \
    trainer.default_local_dir="${CKPTS_DIR}" \
    trainer.resume_mode=auto \
    trainer.log_val_generations=10