# verl-project/verl

verl/HybridFlow: A Flexible and Efficient RL Post-Training Framework

Repository: https://github.com/verl-project/verl
Canonical: https://ross.abutalabs.com/products/verl
Homepage: https://verl-project.github.io
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-08-26T22:36:08+00:00

## Health v2 (maintenance only)
Score: 84/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 86, longevity 47
- inputs: {"age_days": 671, "days_push": 7, "days_rel": 19, "gap_med": 46.0, "n_releases_24m": 15}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 23145, forks 4448 (observed 2026-08-28T04:11:34.546121+00:00)

## What it is
verl (Volcano Engine Reinforcement Learning) is a flexible, production-ready RL post-training library for large language models, open-sourced from ByteDance Seed's HybridFlow paper. It implements a hybrid-controller programming model that lets users express RLHF dataflows like PPO and GRPO in a few lines while integrating with FSDP, Megatron-LM, vLLM, and SGLang for state-of-the-art training and rollout throughput.

## Use cases
- train LLMs with RLHF using PPO or GRPO
- run RL post-training on my own GPUs with vLLM rollouts
- implement a custom RL algorithm for language model fine-tuning
- scale RLHF training across a multi-GPU cluster
- fine-tune a HuggingFace model with reinforcement learning
- train a draft model for speculative decoding
- run RL post-training on AMD ROCm GPUs

## When to choose
- you need production-grade RLHF/RL post-training for LLMs with high throughput
- you want flexible algorithm experimentation with PPO, GRPO, or custom RL dataflows
- you need to combine existing training (FSDP, Megatron-LM) and inference (vLLM, SGLang) engines efficiently
- you need flexible GPU placement across different cluster sizes or heterogeneous hardware

## When to avoid
- you only need supervised fine-tuning (SFT) without RL
- you want a simple one-command fine-tuning script rather than a configurable RL framework
- you are not working with large language models or GPU clusters
- you need a lightweight library without distributed training infrastructure

## Facets
- artifact type: library
- maturity: active
- function: llm-training, reinforcement-learning, machine-learning, gpu-computing
- domain: large-language-models, reinforcement-learning, machine-learning, gpu-computing
- platform: python
- tags: rlhf, post-training, ppo, grpo, vllm, sglang, fsdp, megatron-lm, bytedance, hybridflow, gpu, linux, docker

## Member repositories
- verl-project/verl (main) score 84

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:34.546121+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:56:46.704447+00:00, confidence not recorded.
  - readme: https://github.com/verl-project/verl (fetched 2026-08-28T04:11:34.546121+00:00, sha fc526d88864e)
  - homepage: https://verl-project.github.io (fetched 2026-08-29T07:54:37.888000+00:00, sha 1fbfa6592f3d)
  - registry_pypi: https://pypi.org/pypi/verl/json (fetched 2026-08-29T07:54:37.903516+00:00, sha fe58e14f8a4a)
- Data as of 2026-08-30T08:39:29.467469+00:00.