# MAR-GRPO: Stabilized GRPO for AR-diffusion Hybrid Image Generation

arXiv: https://arxiv.org/abs/2604.06966 · HTML: https://mlsanity.com/paper/2604.06966/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models.
- Builds on 27 of its 36 references that are in the map.
- First posted on arXiv 8 Apr 2026.

## Placed in

- Diffusion models / [Diffusion and flow policies for robotics](https://mlsanity.com/diffusion/diffusion-and-flow-policies-for-robotics/)

## Builds on (most cited first, 27 of 27)

- [Taming Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2012.09841/) (2020)
- [Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction](https://mlsanity.com/paper/2404.02905/) (2024)
- [Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generation](https://mlsanity.com/paper/2305.01569/) (2023)
- [Emerging Properties in Unified Multimodal Pretraining](https://mlsanity.com/paper/2505.14683/) (2025)
- [Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling](https://mlsanity.com/paper/2501.17811/) (2025)
- [Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation](https://mlsanity.com/paper/2406.06525/) (2024)
- [Flow-GRPO: Training Flow Matching Models via Online RL](https://mlsanity.com/paper/2505.05470/) (2025)
- [Autoregressive Image Generation without Vector Quantization](https://mlsanity.com/paper/2406.11838/) (2024)
- [BLIP3-o: A Family of Fully Open Unified Multimodal Models-Architecture, Training and Dataset](https://mlsanity.com/paper/2505.09568/) (2025)
- [MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE](https://mlsanity.com/paper/2507.21802/) (2025)
- [T2I-R1: Reinforcing Image Generation with Collaborative Semantic-level and Token-level CoT](https://mlsanity.com/paper/2505.00703/) (2025)
- [Lumina-mGPT: Illuminate Flexible Photorealistic Text-to-Image Generation with Multimodal Generative Pretraining](https://mlsanity.com/paper/2408.02657/) (2024)
- [GPG: A Simple and Strong Reinforcement Learning Baseline for Model Reasoning](https://mlsanity.com/paper/2504.02546/) (2025)
- [Fluid: Scaling Autoregressive Text-to-image Generative Models with Continuous Tokens](https://mlsanity.com/paper/2410.13863/) (2024)
- [TempFlow-GRPO: When Timing Matters for GRPO in Flow Models](https://mlsanity.com/paper/2508.04324/) (2025)
- [Harmonizing Visual Representations for Unified Multimodal Understanding and Generation](https://mlsanity.com/paper/2503.21979/) (2025)
- [NextStep-1: Toward Autoregressive Image Generation with Continuous Tokens at Scale](https://mlsanity.com/paper/2508.10711/) (2025)
- [Orthus: Autoregressive Interleaved Image-Text Generation with Modality-Specific Heads](https://mlsanity.com/paper/2412.00127/) (2024)
- [Frequency Autoregressive Image Generation with Continuous Tokens](https://mlsanity.com/paper/2503.05305/) (2025)
- [SCALAR: Scale-wise Controllable Visual Autoregressive Learning](https://mlsanity.com/paper/2507.19946/) (2025)
- [Group Critical-token Policy Optimization for Autoregressive Image Generation](https://mlsanity.com/paper/2509.22485/) (2025)
- [VideoMAR: Autoregressive Video Generatio with Continuous Tokens](https://mlsanity.com/paper/2506.14168/) (2025)
- [STAGE: Stable and Generalizable GRPO for Autoregressive Image Generation](https://mlsanity.com/paper/2509.25027/) (2025)
- [Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image Generation](https://mlsanity.com/paper/2510.13418/) (2025)
- [Towards Better & Faster Autoregressive Image Generation: From the Perspective of Entropy](https://mlsanity.com/paper/2510.09012/) (2025)
- [MaskFocus: Focusing Policy Optimization on Critical Steps for Masked Image Generation](https://mlsanity.com/paper/2512.18766/) (2025)
- [Fast-ARDiff: An Entropy-informed Acceleration Framework for Continuous Space Autoregressive Generation](https://mlsanity.com/paper/2512.08537/) (2025)

## Built on by (most cited first, 2 of 2)

- [Flow-OPD: On-Policy Distillation for Flow Matching Models](https://mlsanity.com/paper/2605.08063/) (2026)
- [Embedding-perturbed Exploration Preference Optimization for Flow Models](https://mlsanity.com/paper/2605.15803/) (2026)

## Abstract (arXiv)

Reinforcement learning (RL) has been successfully applied to autoregressive (AR) and diffusion models. However, extending RL to hybrid AR-diffusion frameworks remains challenging due to interleaved inference and noisy log-probability estimation. In this work, we study masked autoregressive models (MAR) and show that the diffusion head plays a critical role in training dynamics, often introducing noisy gradients that lead to instability and early performance saturation. To address this issue, we propose a stabilized RL framework for MAR. We introduce multi-trajectory expectation (MTE), which estimates the optimization direction by averaging over multiple diffusion trajectories, thereby reducing diffusion-induced gradient noise. To avoid over-smoothing, we further estimate token-wise uncertainty from multiple trajectories and apply multi-trajectory optimization only to the top-k% uncertain tokens. In addition, we introduce a consistency-aware token selection strategy that filters out AR tokens that are less aligned with the final generated content. Extensive experiments across multiple benchmarks demonstrate that our method consistently improves visual quality, training stability, and spatial structure understanding over baseline GRPO and pre-RL models. Code is available at: https://github.com/AMAP-ML/mar-grpo.