verl

verl: Volcano Engine Reinforcement Learning for LLMs

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-controller programming model enables flexible representation and efficient execution of complex post-training dataflows. Build RL dataflows such as GRPO, PPO 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 FSDP, Megatron-LM, vLLM, SGLang, etc - **Flexible device mapping**: Supports various placement of models onto different sets of GPUs for efficient resource utilization and scalability across different cluster sizes. - Ready integration with popular HuggingFace models verl is fast with: - **State-of-the-art throughput**: SOTA LLM training and inference engine integrations and SOTA RL throughput. - **Efficient actor model resharding with 3D-HybridEngine**: Eliminates memory redundancy and significantly reduces communication overhead during transitions between training and generation phases.

## News - [2026/08] [verl-vla](https://github.com/verl-project/verl-vla) `v0.1.0` is released: a unified VLA post-training framework for human-in-the-loop data collection, supervised fine-tuning, and reinforcement learning across simulators, real robots, and distributed cloud-edge resources, built on top of verl. - [2026/08] [VeRL-Tinker](https://github.com/verl-project/verl-recipe/tree/main/verl_tinker) is released: keep the Tinker Cookbook loop you know, and run SFT, RL, and distillation on verl-managed GPU workers you control; read the blog [here](https://verl-project.github.io/posts/2026-08-11-verl-tinker/). - [2026/08] [VeRL-Omni](https://github.com/verl-project/verl-omni) `v0.2.0` is released: faster diffusion RL, rebuilt Qwen3-Omni multimodal training (DPO & GSPO), plus LTX-2.3, Qwen-Image-Edit support and more. - [2026/07] [RL-Insight](https://github.com/verl-project/rl-insight) is released: online observability for reinforcement learning training. RL-Insight connects training-side metrics, RL state traces, and service dashboards across distributed rollout and optimization workloads. - [2026/06] [verl-SpeCo](https://github.com/verl-project/verl-SpeCo) is pre-released: a co-training framework for speculative decoding across RL training and inference, keeping draft models aligned during training and reusable for accelerated serving, built on top of verl. - [2026/05] [uni-agent](https://github.com/verl-project/uni-agent) is released: a unified agent framework to build, run, and train LLM agents at scale, built on top of verl. - [2026/05] [VeRL-Omni](https://github.com/verl-project/verl-omni) is pre-released: a unified RL stack for diffusion and omni-modal model post-training built on top of verl. Read the [blog post](https://vllm.ai/blog/2026-05-14-verl-omni) for details. - [2026/05] verl's zero-mismatch HuggingFace rollout [vexact](https://github.com/verl-project/vexact) is released: with batch-invariant kernels, shared model definition with FSDP, and out-of-box examples compatible with VeOmni. - [2026/04] verl's Megatron backend LoRA and router replay support is showcased at [PyTorch Conference Europe 2026](https://pytorchconferenceeu2026.sched.com/event/2Juce/optimizing-reinforcement-learning-at-trillion-parameter-scale-songlin-jiang-aalto-university-mind-lab). - [2026/03] verl is presented at NVIDIA GTC26: [session#1](https://www.nvidia.com/en-us/on-demand/session/gtc26-S81829/), [session#2](https://www.nvidia.com/en-us/on-demand/session/gtc26-S81620/) - [2026/01] verl has been migrated to the [verl-project](https://github.com/verl-project) - [2026/01] verl first meetup was successfully held in Shanghai on 01/10, hosted by Volcengine and NVIDIA, the slides has been uploaded to [verl-data](https://github.com/verl-project/verl-data). - [2026/01] The `recipe` directory has been migrated to a dedicated repository: [verl-recipe](https://github.com/verl-project/verl-recipe) and added as a submodule. See https://github.com/verl-project/verl/pull/4795. It can be used as it was after `git submodule update --init --recursive recipe`. Note that [`transfer_queue`](verl/experimental/transfer_queue), [`fully_async_policy`](verl/experimental/fully_async_policy), [`one_step_off_policy`](verl/experimental/one_step_off_policy) and [`vla`](verl/experimental/vla) are kept under [`verl/experimental`](verl/experimental) since they are planned to be merged into the main library. Use them through `verl.experimental.{module}`. - [2025/12] [Mind Lab](https://macaron.im/mindlab) successfully used [verl](https://github.com/verl-project/verl) and [Megatron-bridge](https://github.com/NVIDIA-NeMo/Megatron-Bridge) to train GRPO Lora for Trillion-parameter model on 64 H800 - See their [techblog](https://macaron.im/mindlab/research/building-trillion-parameter-reasoning-rl-with-10-gpus). - [2025/10] verl is presented in the [PyTorch Conference 2025](https://pytorch.org/event/pytorch-conference-2025/). - [2025/08] verl is presented in the [PyTorch Expert Exchange Webinar](https://www.youtube.com/watch?v=Vd79NmmqY3Q&t=2s). [Slides](https://github.com/eric-haibin-lin/verl-community/blob/main/slides/verl_talk_pytorch_2025_08.pdf) available. - [2025/07] The [ReTool](https://arxiv.org/pdf/2504.11536) recipe is fully open sourced. [Blog](https://www.notion.so/verl-reTool-recipe-Using-multi-round-conversations-and-code-sandboxing-to-improve-the-math-of-large-23a8b5b7feba80b386b2e5b5e3c1cde0) - [2025/07] The first verl meetup will be held at ICML Vancouver on July 16th! Please [join us](https://lu.ma/0ek2nyao) if you are at ICML! (onsite only) - [2025/06] verl with Megatron backend enables large MoE models such as [DeepSeek-671B and Qwen3-235B](https://verl.readthedocs.io/en/latest/perf/dpsk.html). - [2025/03] [DAPO](https://dapo-sia.github.io/) is the open-sourced SOTA RL algorithm that achieves 50 points on AIME 2024 based on the Qwen2.5-32B pre-trained model, surpassing the previous SOTA achieved by DeepSeek's GRPO (DeepSeek-R1-Zero-Qwen-32B). DAPO's training is fully powered by verl and the reproduction code is available in `recipe/dapo` now. more...

• [2025/04] [Seed-Thinking-v1.5](https://github.com/ByteDance-Seed/Seed-Thinking-v1.5/blob/main/seed-thinking-v1.5.pdf) tech report is released! Trained with verl, Seed-Thinking-v1.5 achieves 86.7 on AIME 2024, 55.0 on Codeforces and 77.3 on GPQA, demonstrating excellent reasoning abilities in STEM and coding. Beyond reasoning tasks, the method demonstrates notable generalization across diverse domains.
• [2025/07] verl keynote at [AWS AI Hours Singapore](https://pages.awscloud.com/aws-ai-hours-sg.html#agenda) on 7/8, verl & verl-agent project updates at [Agent for SWE meetup](https://lu.ma/e498qhsi) by LF AI & Data Singapore on 7/11.
• [2025/06] verl team will provide latest project updates at [PyTorch Day China](https://www.lfasiallc.com/pytorch-day-china/) on June 7th. Meet our dev team in Beijing!
• [2025/04] [VAPO](https://arxiv.org/pdf/2504.05118) (value-based augmented PPO) paper covers our latest RL method for reasoning models. Trained from Qwen-32B-base model, VAPO achieves 60.4 on AIME 2024, outperforming DAPO-32B.
• [2025/05] [PF-PPO](https://arxiv.org/abs/2409.06957), accepted to ICML 2025, is now supported in verl! PF-PPO enhances policy learning efficiency and robustness by filtering potentially noisy reward signals and reusing high-quality experiences via a replay buffer.
• [2025/04] We will give a tutorial about latest post-training techniques and programming guide for verl at [ICLR 2025 Expo](https://iclr.cc/virtual/2025/calendar?filter_events=Expo+Talk+Panel&filter_rooms=), [SCI-FM workshop](https://open-foundation-model.github.io/) and [LMSys afterparty](https://lu.ma/d23nyynm). Talk materials available [here](https://github.com/eric-haibin-lin/verl-community/tree/main/iclr25).
• [2025/03] verl v0.3.0.post1 is released! See [release note](https://github.com/verl-project/verl/releases/) for details. It achieves [~1.4x speedup](https://tongyx361.github.io/blogs/posts/verl-intro/#/verl-flexible-and-efficient-rl-for-llms) compared to prev versions.
• [2025/05] verl will be presented at [A2M Shanghai](https://a2m.msup.com.cn/home/?aid=4488&city=shanghai) on 5/16 - 5/17.
• [2025/05] verl will be presented at [GOSIM x PyTorch Day 2025](https://paris2025.gosim.org/). See you in Paris!
• [2025/03] We introduced the programming model of verl at the [vLLM Beijing Meetup](https://mp.weixin.qq.com/s/n77GibL2corAtQHtVEAzfg) and [verl intro and updates](https://github.com/eric-haibin-lin/verl-community/blob/main/slides/verl-lmsys-meetup.pdf) at the [SGLang-LMSYS Org Meetup](https://lu.ma/ntjrr7ig) in Sunnyvale mid-March.
• [2025/03] We will present verl(HybridFlow) at EuroSys 2025. See you in Rotterdam!
• [2025/02] verl v0.2.0.post2 is released!
• [2025/02] We presented verl in the Bytedance/NVIDIA/Anyscale Ray Meetup. See you in San Jose!
• [2025/01] [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] verl is presented at Ray Forward 2024. Slides available here
• [2024/12] The team presented Post-training LLMs: From Algorithms to Infrastructure at NeurIPS 2024. Slides and video available.
• [2024/10] verl is presented at Ray Summit. Youtube video available.
• [2024/08] HybridFlow (verl) is accepted to EuroSys 2025.

## Key Features - **FSDP**, **FSDP2** and **Megatron-LM** for training. - **vLLM**, **SGLang** and **HF Transformers** for rollout generation. - Compatible with Hugging Face Transformers and Modelscope Hub: Qwen3.5, Qwen3, Qwen-2.5, Llama3.1, Gemma2, DeepSeek-LLM, etc - Supervised fine-tuning. - Reinforcement learning with [PPO](examples/ppo_trainer/), [GRPO](examples/grpo_trainer/), [GSPO](https://github.com/verl-project/verl-recipe/tree/main/gspo/), [ReMax](examples/remax_trainer/), [REINFORCE++](https://verl.readthedocs.io/en/latest/examples/config.html#algorithm), [RLOO](examples/rloo_trainer/), [PRIME](https://github.com/verl-project/verl-recipe/tree/main/prime/), [DAPO](https://github.com/verl-project/verl-recipe/tree/main/dapo/), [DrGRPO](https://github.com/verl-project/verl-recipe/tree/main/drgrpo), [KL_Cov & Clip_Cov](https://github.com/verl-project/verl-recipe/tree/main/entropy) etc. - Support model-based reward and function-based reward (verifiable reward) for math, [coding](https://github.com/verl-project/verl-recipe/tree/main/dapo), etc - Support vision-language models (VLMs) and [multi-modal RL](examples/grpo_trainer/run_qwen2_5_vl_7b_fsdp.sh) with Qwen2.5-vl, Kimi-VL - [Multi-turn with tool calling](examples/tutorial/agent_loop_get_started/) - LLM alignment recipes such as [Self-play preference optimization (SPPO)](https://github.com/verl-