-
Notifications
You must be signed in to change notification settings - Fork 4
/
Copy pathREADME.md.verl
142 lines (109 loc) · 10.1 KB
/
README.md.verl
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
<h1 style="text-align: center;">verl: Volcano Engine Reinforcement Learning for LLM</h1>
verl is a flexible, efficient and production-ready RL training library for large language models (LLMs).
verl is the open-source version of **[HybridFlow: A Flexible and Efficient RLHF Framework](https://arxiv.org/abs/2409.19256v2)** paper.
verl is flexible and easy to use with:
- **Easy extension of diverse RL algorithms**: The Hybrid programming model combines the strengths of single-controller and multi-controller paradigms to enable flexible representation and efficient execution of complex Post-Training dataflows. Allowing users to build RL dataflows in a few lines of code.
- **Seamless integration of existing LLM infra with modular APIs**: Decouples computation and data dependencies, enabling seamless integration with existing LLM frameworks, such as PyTorch FSDP, Megatron-LM and vLLM. Moreover, users can easily extend to other LLM training and inference frameworks.
- **Flexible device mapping**: Supports various placement of models onto different sets of GPUs for efficient resource utilization and scalability across different cluster sizes.
- Readily integration with popular HuggingFace models
verl is fast with:
- **State-of-the-art throughput**: By seamlessly integrating existing SOTA LLM training and inference frameworks, verl achieves high generation and training throughput.
- **Efficient actor model resharding with 3D-HybridEngine**: Eliminates memory redundancy and significantly reduces communication overhead during transitions between training and generation phases.
<p align="center">
| <a href="https://verl.readthedocs.io/en/latest/index.html"><b>Documentation</b></a> | <a href="https://arxiv.org/abs/2409.19256v2"><b>Paper</b></a> | <a href="https://join.slack.com/t/verlgroup/shared_invite/zt-2w5p9o4c3-yy0x2Q56s_VlGLsJ93A6vA"><b>Slack</b></a> | <a href="https://raw.githubusercontent.com/eric-haibin-lin/verl-community/refs/heads/main/WeChat.JPG"><b>Wechat</b></a> | <a href="https://x.com/verl_project"><b>Twitter</b></a>
<!-- <a href=""><b>Slides</b></a> | -->
</p>
## News
- [2025/3] We will present verl(HybridFlow) at [EuroSys 2025](https://2025.eurosys.org/). See you in in Rotterdam!
- [2025/2] verl v0.2.0.post1 is released! See [release note](https://github.com/volcengine/verl/releases/) for details.
- [2025/2] We presented verl in the [Bytedance/NVIDIA/Anyscale Ray Meetup](https://lu.ma/ji7atxux). See you in San Jose!
- [2025/1] [Doubao-1.5-pro](https://team.doubao.com/zh/special/doubao_1_5_pro) is released with SOTA-level performance on LLM & VLM. The RL scaling preview model is trained using verl, reaching OpenAI O1-level performance on math benchmarks (70.0 pass@1 on AIME).
- [2024/12] The team presented <a href="https://neurips.cc/Expo/Conferences/2024/workshop/100677">Post-training LLMs: From Algorithms to Infrastructure</a> at NeurIPS 2024. [Slides](https://github.com/eric-haibin-lin/verl-data/tree/neurips) and [video](https://neurips.cc/Expo/Conferences/2024/workshop/100677) available.
- [2024/12] verl is presented at Ray Forward 2024. Slides available [here](https://github.com/eric-haibin-lin/verl-community/blob/main/slides/Ray_Forward_2024_%E5%B7%AB%E9%94%A1%E6%96%8C.pdf).
- [2024/10] verl is presented at Ray Summit. [Youtube video](https://www.youtube.com/watch?v=MrhMcXkXvJU&list=PLzTswPQNepXntmT8jr9WaNfqQ60QwW7-U&index=37) available.
- [2024/08] HybridFlow (verl) is accepted to EuroSys 2025.
## Key Features
- **FSDP** and **Megatron-LM** for training.
- **vLLM** and **TGI** for rollout generation, **SGLang** support coming soon.
- huggingface models support
- Supervised fine-tuning
- Reinforcement learning from human feedback with [PPO](https://github.com/volcengine/verl/tree/main/examples/ppo_trainer), [GRPO](https://github.com/volcengine/verl/tree/main/examples/grpo_trainer), [ReMax](https://github.com/volcengine/verl/tree/main/examples/remax_trainer), Reinforce++, [RLOO](https://github.com/volcengine/verl/tree/main/examples/rloo_trainer/run_qwen2-7b.sh), etc
- Support model-based reward and function-based reward (verifiable reward)
- flash-attention, [sequence packing](examples/ppo_trainer/run_qwen2-7b_seq_balance.sh), [long context](examples/ppo_trainer/run_deepseek7b_llm_sp2.sh) support via DeepSpeed Ulysses, [LoRA](examples/sft/gsm8k/run_qwen_05_peft.sh), [Liger-kernel](examples/sft/gsm8k/run_qwen_05_sp2_liger.sh)
- scales up to 70B models and hundreds of GPUs
- experiment tracking with wandb, swanlab and mlflow
## Upcoming Features
- Reward model training
- DPO training
- DeepSeek integration with Megatron v0.11
- SGLang integration
- vision language model RL
## Getting Started
**Quickstart:**
- [Installation](https://verl.readthedocs.io/en/latest/start/install.html)
- [Quickstart](https://verl.readthedocs.io/en/latest/start/quickstart.html)
- [Programming Guide](https://verl.readthedocs.io/en/latest/hybrid_flow.html)
**Running a PPO example step-by-step:**
- Data and Reward Preparation
- [Prepare Data for Post-Training](https://verl.readthedocs.io/en/latest/preparation/prepare_data.html)
- [Implement Reward Function for Dataset](https://verl.readthedocs.io/en/latest/preparation/reward_function.html)
- Understanding the PPO Example
- [PPO Example Architecture](https://verl.readthedocs.io/en/latest/examples/ppo_code_architecture.html)
- [Config Explanation](https://verl.readthedocs.io/en/latest/examples/config.html)
- [Run GSM8K Example](https://verl.readthedocs.io/en/latest/examples/gsm8k_example.html)
**Reproducible algorithm baselines:**
- [PPO, GRPO, ReMax](https://verl.readthedocs.io/en/latest/experiment/ppo.html)
**For code explanation and advance usage (extension):**
- PPO Trainer and Workers
- [PPO Ray Trainer](https://verl.readthedocs.io/en/latest/workers/ray_trainer.html)
- [PyTorch FSDP Backend](https://verl.readthedocs.io/en/latest/workers/fsdp_workers.html)
- [Megatron-LM Backend](https://verl.readthedocs.io/en/latest/index.html)
- Advance Usage and Extension
- [Ray API design tutorial](https://verl.readthedocs.io/en/latest/advance/placement.html)
- [Extend to Other RL(HF) algorithms](https://verl.readthedocs.io/en/latest/advance/dpo_extension.html)
- [Add Models with the FSDP Backend](https://verl.readthedocs.io/en/latest/advance/fsdp_extension.html)
- [Add Models with the Megatron-LM Backend](https://verl.readthedocs.io/en/latest/advance/megatron_extension.html)
- [Deployment using Separate GPU Resources](https://github.com/volcengine/verl/tree/main/examples/split_placement)
**Blogs from the community**
- [使用verl进行GRPO分布式强化学习训练最佳实践](https://www.volcengine.com/docs/6459/1463942)
- [HybridFlow veRL 原文浅析](https://github.com/zhaochenyang20/Awesome-ML-SYS-Tutorial/blob/main/rlhf/verl/readme.md)
- [最高提升20倍吞吐量！豆包大模型团队发布全新 RLHF 框架，现已开源！](https://team.doubao.com/en/blog/%E6%9C%80%E9%AB%98%E6%8F%90%E5%8D%8720%E5%80%8D%E5%90%9E%E5%90%90%E9%87%8F-%E8%B1%86%E5%8C%85%E5%A4%A7%E6%A8%A1%E5%9E%8B%E5%9B%A2%E9%98%9F%E5%8F%91%E5%B8%83%E5%85%A8%E6%96%B0-rlhf-%E6%A1%86%E6%9E%B6-%E7%8E%B0%E5%B7%B2%E5%BC%80%E6%BA%90)
Checkout this [Jupyter Notebook](https://github.com/volcengine/verl/tree/main/examples/ppo_trainer/verl_getting_started.ipynb) to get started with PPO training with a single 24GB L4 GPU (**FREE** GPU quota provided by [Lighting Studio](https://lightning.ai/hlin-verl/studios/verl-getting-started))!
## Performance Tuning Guide
The performance is essential for on-policy RL algorithm. We write a detailed performance tuning guide to allow people tune the performance. See [here](https://verl.readthedocs.io/en/latest/perf/perf_tuning.html) for more details.
## vLLM v0.7 testing version
We have released a testing version of veRL that supports vLLM>=0.7.0. Please refer to [this document](https://github.com/volcengine/verl/blob/main/docs/README_vllm0.7.md) for installation guide and more information.
## Citation and acknowledgement
If you find the project helpful, please cite:
- [HybridFlow: A Flexible and Efficient RLHF Framework](https://arxiv.org/abs/2409.19256v2)
- [A Framework for Training Large Language Models for Code Generation via Proximal Policy Optimization](https://i.cs.hku.hk/~cwu/papers/gmsheng-NL2Code24.pdf)
```tex
@article{sheng2024hybridflow,
title = {HybridFlow: A Flexible and Efficient RLHF Framework},
author = {Guangming Sheng and Chi Zhang and Zilingfeng Ye and Xibin Wu and Wang Zhang and Ru Zhang and Yanghua Peng and Haibin Lin and Chuan Wu},
year = {2024},
journal = {arXiv preprint arXiv: 2409.19256}
}
```
verl is inspired by the design of Nemo-Aligner, Deepspeed-chat and OpenRLHF. The project is adopted and supported by Anyscale, Bytedance, LMSys.org, Shanghai AI Lab, Tsinghua University, UC Berkeley, UCLA, UIUC, and University of Hong Kong.
## Awesome work using verl
- [Enhancing Multi-Step Reasoning Abilities of Language Models through Direct Q-Function Optimization](https://arxiv.org/abs/2410.09302)
- [Flaming-hot Initiation with Regular Execution Sampling for Large Language Models](https://arxiv.org/abs/2410.21236)
- [Process Reinforcement Through Implicit Rewards](https://github.com/PRIME-RL/PRIME/)
- [TinyZero](https://github.com/Jiayi-Pan/TinyZero): a reproduction of DeepSeek R1 Zero recipe for reasoning tasks
- [RAGEN](https://github.com/ZihanWang314/ragen): a general-purpose reasoning agent training framework
- [Logic R1](https://github.com/Unakar/Logic-RL): a reproduced DeepSeek R1 Zero on 2K Tiny Logic Puzzle Dataset.
- [deepscaler](https://github.com/agentica-project/deepscaler): iterative context scaling with GRPO
- [critic-rl](https://github.com/HKUNLP/critic-rl): Teaching Language Models to Critique via Reinforcement Learning
## Contribution Guide
Contributions from the community are welcome!
### Code formatting
We use yapf (Google style) to enforce strict code formatting when reviewing PRs. To reformat you code locally, make sure you installed **latest** `yapf`
```bash
pip3 install yapf --upgrade
```
Then, make sure you are at top level of verl repo and run
```bash
bash scripts/format.sh
```
We are HIRING! Send us an [email](mailto:[email protected]) if you are interested in internship/FTE opportunities in MLSys/LLM reasoning/multimodal alignment.