基于 Ray 使用 veRL 进行强化学习

hf download qwen/Qwen2.5-0.5B-Instruct --local-dir /path/to/Qwen2.5-0.5B-Instruct


# Copyright 2024 Bytedance Ltd. and/or its affiliates
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Preprocess the GSM8k dataset to parquet format
"""

import argparse
import os
import re

import datasets

from verl.utils.hdfs_io import copy, makedirs


def extract_solution(solution_str):
    solution = re.search("#### (\\\\-?[0-9\\\\.\\\\,]+)", solution_str)
    assert solution is not None
    final_solution = solution.group(0)
    final_solution = final_solution.split("#### ")[1].replace(",", "")
    return final_solution


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--local_dir", default="~/data/gsm8k")
    parser.add_argument("--hdfs_dir", default=None)

    args = parser.parse_args()

    data_source = "openai/gsm8k"

    dataset = datasets.load_dataset(data_source, "main")

    train_dataset = dataset["train"]
    test_dataset = dataset["test"]

    instruction_following = 'Let\\'s think step by step and output the final answer after "####".'

    # add a row to each data item that represents a unique id
    def make_map_fn(split):
        def process_fn(example, idx):
            question_raw = example.pop("question")

            question = question_raw + " " + instruction_following

            answer_raw = example.pop("answer")
            solution = extract_solution(answer_raw)
            data = {
                "data_source": data_source,
                "prompt": [
                    {
                        "role": "user",
                        "content": question,
                    }
                ],
                "ability": "math",
                "reward_model": {"style": "rule", "ground_truth": solution},
                "extra_info": {
                    "split": split,
                    "index": idx,
                    "answer": answer_raw,
                    "question": question_raw,
                },
            }
            return data

        return process_fn

    train_dataset = train_dataset.map(function=make_map_fn("train"), with_indices=True)
    test_dataset = test_dataset.map(function=make_map_fn("test"), with_indices=True)

    local_dir = args.local_dir
    hdfs_dir = args.hdfs_dir

    train_dataset.to_parquet(os.path.join(local_dir, "train.parquet"))
    test_dataset.to_parquet(os.path.join(local_dir, "test.parquet"))

    if hdfs_dir is not None:
        makedirs(hdfs_dir)

        copy(src=local_dir, dst=hdfs_dir)

python gsm8k.py --local_save_dir /path/to/gsm8k


#!/bin/bash
MODEL_PATH=/cfs_data/verl_demo/Qwen2.5-0.5B-Instruct
GSM8K_PATH=/cfs_data/verl_demo/gsm8k


set -x
python3 -m verl.trainer.main_ppo \\
   data.train_files=$GSM8K_PATH/train.parquet \\
   data.val_files=$GSM8K_PATH/test.parquet \\
   data.train_batch_size=1024 \\
   data.max_prompt_length=512 \\
   data.max_response_length=512 \\
   actor_rollout_ref.model.path=$MODEL_PATH \\
   actor_rollout_ref.actor.optim.lr=1e-6 \\
   actor_rollout_ref.model.use_remove_padding=True \\
   actor_rollout_ref.actor.ppo_mini_batch_size=256 \\
   actor_rollout_ref.actor.ppo_micro_batch_size_per_gpu=16 \\
   actor_rollout_ref.actor.fsdp_config.param_offload=False \\
   actor_rollout_ref.actor.fsdp_config.optimizer_offload=False \\
   actor_rollout_ref.model.enable_gradient_checkpointing=True \\
   actor_rollout_ref.rollout.log_prob_micro_batch_size_per_gpu=32 \\
   actor_rollout_ref.rollout.tensor_model_parallel_size=2 \\
   actor_rollout_ref.rollout.name=vllm \\
   actor_rollout_ref.rollout.gpu_memory_utilization=0.5 \\
   actor_rollout_ref.ref.log_prob_micro_batch_size_per_gpu=32 \\
   actor_rollout_ref.ref.fsdp_config.param_offload=True \\
   critic.optim.lr=1e-5 \\
   critic.model.use_remove_padding=True \\
   critic.model.path=$MODEL_PATH \\
   critic.model.enable_gradient_checkpointing=True \\
   critic.ppo_micro_batch_size_per_gpu=32 \\
   critic.model.fsdp_config.param_offload=False \\
   critic.model.fsdp_config.optimizer_offload=False \\
   algorithm.kl_ctrl.kl_coef=0.001 \\
   trainer.critic_warmup=0 \\
   trainer.logger='["console", "mlflow"]' \\
   trainer.project_name='verl_example_gsm8k' \\
   trainer.experiment_name='qwen2.5_0.5b_function_rm' \\
   trainer.n_gpus_per_node=2 \\
   trainer.nnodes=2 \\
   trainer.save_freq=-1 \\
   trainer.test_freq=1 \\
   trainer.total_epochs=15