Policy-based Tuning of Autoregressive Image Models with Instance- and Distribution-Level Rewards

• URL: http://arxiv.org/abs/2603.23086v1
• Date: Tue, 24 Mar 2026 11:28:36 GMT
• Title: Policy-based Tuning of Autoregressive Image Models with Instance- and Distribution-Level Rewards
• Abstract summary: Autoregressive (AR) models are highly effective for image generation, yet their standard maximum-likelihood estimation training lacks direct optimization for sample quality and diversity.<n>We propose a lightweight RL framework that casts token-based AR inference as a Markov Decision Process, optimized via Group Relative Policy Optimization.<n>Our core contribution is the introduction of a novel distribution-level Leave-One-Out FID (LOO-FID) reward.
• Score: 16.135177543347773
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Autoregressive (AR) models are highly effective for image generation, yet their standard maximum-likelihood estimation training lacks direct optimization for sample quality and diversity. While reinforcement learning (RL) has been used to align diffusion models, these methods typically suffer from output diversity collapse. Similarly, concurrent RL methods for AR models rely strictly on instance-level rewards, often trading off distributional coverage for quality. To address these limitations, we propose a lightweight RL framework that casts token-based AR synthesis as a Markov Decision Process, optimized via Group Relative Policy Optimization (GRPO). Our core contribution is the introduction of a novel distribution-level Leave-One-Out FID (LOO-FID) reward; by leveraging an exponential moving average of feature moments, it explicitly encourages sample diversity and prevents mode collapse during policy updates. We integrate this with composite instance-level rewards (CLIP and HPSv2) for strict semantic and perceptual fidelity, and stabilize the multi-objective learning with an adaptive entropy regularization term. Extensive experiments on LlamaGen and VQGAN architectures demonstrate clear improvements across standard quality and diversity metrics within only a few hundred tuning iterations. The results also show that the model can be updated to produce competitive samples even without Classifier-Free Guidance, and bypass its 2x inference cost.

Related papers

• Optimizing Visual Generative Models via Distribution-wise Rewards [57.46109819595665]
  We present a novel framework that finetunes generative models using distribution-wise rewards, ensuring better alignment with real-world data distributions.<n>Our approach significantly improves FID-50K across various base models, from 5.77 for SiT and from 3.74 to 3.52 for EDM2.
  arXiv  Detail & Related papers  (2026-07-02T15:08:56Z)
• Too Correct to Learn: Reinforcement Learning on Saturated Reasoning Data [55.84428098924793]
  We propose Constrained Uniform Top-K Sampling (CUTS), a parameter-free decoding strategy enforcing structure-preserving exploration.<n>We integrate this into Mixed-CUTS, a training framework synergizing exploitative and exploratory rollouts to amplify intra-group advantage variance.<n> Notably, Mixed-CUTS improves Pass@1 accuracy on the challenging AIME25 benchmark by up to 15.1% over standard GRPO.
  arXiv  Detail & Related papers  (2026-04-20T16:43:28Z)
• RS-Prune: Training-Free Data Pruning at High Ratios for Efficient Remote Sensing Diffusion Foundation Models [14.093802378976315]
  Diffusion-based remote sensing (RS) generative foundation models rely on large amounts of globally representative data.<n>We propose a training-free, two-stage data pruning approach that quickly select a high-quality subset under high pruning ratios.<n> Experiments show that, even after pruning 85% of the training data, our method significantly improves convergence and generation quality.
  arXiv  Detail & Related papers  (2025-12-29T06:44:06Z)
• Enhancing Diffusion-based Restoration Models via Difficulty-Adaptive Reinforcement Learning with IQA Reward [93.04811239892852]
  Reinforcement Learning (RL) has recently been incorporated into diffusion models.<n>In this paper, we investigate how to effectively integrate RL into diffusion-based restoration models.
  arXiv  Detail & Related papers  (2025-11-03T14:57:57Z)
• G$^2$RPO: Granular GRPO for Precise Reward in Flow Models [74.21206048155669]
  We propose a novel Granular-GRPO (G$2$RPO) framework that achieves precise and comprehensive reward assessments of sampling directions.<n>We introduce a Multi-Granularity Advantage Integration module that aggregates advantages computed at multiple diffusion scales.<n>Our G$2$RPO significantly outperforms existing flow-based GRPO baselines.
  arXiv  Detail & Related papers  (2025-10-02T12:57:12Z)
• MMR1: Enhancing Multimodal Reasoning with Variance-Aware Sampling and Open Resources [113.33902847941941]
  Variance-Aware Sampling (VAS) is a data selection strategy guided by Variance Promotion Score (VPS)<n>We release large-scale, carefully curated resources containing 1.6M long CoT cold-start data and 15k RL QA pairs.<n> Experiments across mathematical reasoning benchmarks demonstrate the effectiveness of both the curated data and the proposed VAS.
  arXiv  Detail & Related papers  (2025-09-25T14:58:29Z)
• Flow-GRPO: Training Flow Matching Models via Online RL [80.62659379624867]
  We propose Flow-GRPO, the first method to integrate online policy reinforcement learning into flow matching models.<n>Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary Differential Equation into an equivalent Differential Equation (SDE) that matches the original model's marginal distribution at all timesteps; and (2) a Denoising Reduction strategy that reduces training denoising steps while retaining the original number of inference steps.
  arXiv  Detail & Related papers  (2025-05-08T17:58:45Z)
• Adaptive Multi-Fidelity Reinforcement Learning for Variance Reduction in Engineering Design Optimization [0.0]
  Multi-fidelity Reinforcement Learning (RL) frameworks efficiently utilize computational resources by integrating analysis models of varying accuracy and costs.<n>This work proposes a novel adaptive multi-fidelity RL framework, in which multiple heterogeneous, non-hierarchical low-fidelity models are dynamically leveraged alongside a high-fidelity model.<n>The effectiveness of the approach is demonstrated in an octocopter design optimization problem, utilizing two low-fidelity models alongside a high-fidelity simulator.
  arXiv  Detail & Related papers  (2025-03-23T22:29:08Z)
• ROCM: RLHF on consistency models [8.905375742101707]
  We propose a reward optimization framework for applying RLHF to consistency models.<n>We investigate various $f$-divergences as regularization strategies, striking a balance between reward and model consistency.
  arXiv  Detail & Related papers  (2025-03-08T11:19:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.