# Expand and Prune: Maximizing Trajectory Diversity for Effective GRPO in Generative Models

arXiv: https://arxiv.org/abs/2512.15347 · HTML: https://mlsanity.com/paper/2512.15347/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Reinforcement learning for reasoning, one of 17 directions in LLM reasoning.
- Builds on 14 of its 49 references that are in the map.
- First posted on arXiv 17 Dec 2025.

## Placed in

- LLM reasoning / [Reinforcement learning for reasoning](https://mlsanity.com/lr/reinforcement-learning-for-reasoning/)

## Builds on (most cited first, 14 of 14)

- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow](https://mlsanity.com/paper/2209.03003/) (2022)
- [Training Diffusion Models with Reinforcement Learning](https://mlsanity.com/paper/2305.13301/) (2023)
- [Stochastic Interpolants: A Unifying Framework for Flows and Diffusions](https://mlsanity.com/paper/2303.08797/) (2023)
- [Flow-GRPO: Training Flow Matching Models via Online RL](https://mlsanity.com/paper/2505.05470/) (2025)
- [DanceGRPO: Unleashing GRPO on Visual Generation](https://mlsanity.com/paper/2505.07818/) (2025)
- [Human Preference Score: Better Aligning Text-to-Image Models with Human Preference](https://mlsanity.com/paper/2303.14420/) (2023)
- [MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE](https://mlsanity.com/paper/2507.21802/) (2025)
- [Can We Generate Images with CoT? Let's Verify and Reinforce Image Generation Step by Step](https://mlsanity.com/paper/2501.13926/) (2025)
- [SimpleAR: Pushing the Frontier of Autoregressive Visual Generation through Pretraining, SFT, and RL](https://mlsanity.com/paper/2504.11455/) (2025)
- [ReasonGen-R1: CoT for Autoregressive Image generation models through SFT and RL](https://mlsanity.com/paper/2505.24875/) (2025)
- [RePrompt: Reasoning-Augmented Reprompting for Text-to-Image Generation via Reinforcement Learning](https://mlsanity.com/paper/2505.17540/) (2025)
- [AR-GRPO: Training Autoregressive Image Generation Models via Reinforcement Learning](https://mlsanity.com/paper/2508.06924/) (2025)
- [Reinforcement Learning for Generative AI: A Survey](https://mlsanity.com/paper/2308.14328/) (2023)

## Built on by (most cited first, 6 of 6)

- [FP4 Explore, BF16 Train: Diffusion Reinforcement Learning via Efficient Rollout Scaling](https://mlsanity.com/paper/2604.06916/) (2026)
- [Advances in GRPO for Generation Models: A Survey](https://mlsanity.com/paper/2603.06623/) (2026)
- [Learning to Credit the Right Steps: Objective-aware Process Optimization for Visual Generation](https://mlsanity.com/paper/2604.19234/) (2026)
- [Can the Environment Speak for Itself? T2-GRPO: A Turn-Trajectory Group Relative Policy Optimization for Caregiver Agents](https://mlsanity.com/paper/2606.08875/) (2026)
- [Multi-History-Step SDE Inversion for Image Editing with Superior Regional Awareness](https://mlsanity.com/paper/2609.06602/) (2026)
- [DSPO: Diversity-aware Subjective Policy Optimization for Robust Emotional Reasoning](https://mlsanity.com/paper/2609.36775/) (2026)

## Abstract (arXiv)

Group Relative Policy Optimization (GRPO) is a powerful technique for aligning generative models, but its effectiveness is bottlenecked by the conflict between large group sizes and prohibitive computational costs. In this work, we investigate the trade-off through empirical studies, yielding two key observations. First, we discover the reward clustering phenomenon in which many trajectories collapse toward the group-mean reward, offering limited optimization value. Second, we design a heuristic strategy named Optimal Variance Filtering (OVF), and verify that a high-variance subset of trajectories, selected by OVF can outperform the larger, unfiltered group. However, this static, post-sampling OVF approach still necessitates critical computational overhead, as it performs unnecessary sampling for trajectories that are ultimately discarded. To resolve this, we propose Pro-GRPO (Proactive GRPO), a novel dynamic framework that integrates latent feature-based trajectory pruning into the sampling process. Through the early termination of reward-clustered trajectories, Pro-GRPO reduces computational overhead. Leveraging its efficiency, Pro-GRPO employs an "Expand-and-Prune" strategy. This strategy first expands the size of initial sampling group to maximize trajectory diversity, then it applies multi-step OVF to the latents, avoiding prohibitive computational costs. Extensive experiments on both diffusion-based and flow-based models demonstrate the generality and effectiveness of our Pro-GRPO framework.