# ReNFT: Repairing Mode Collapse in Reward Post-Training via Internal Probability-Mass Recalibration

arXiv: https://arxiv.org/abs/2609.00061 · HTML: https://mlsanity.com/paper/2609.00061/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 31 references that are in the map.
- First posted on arXiv 30 Aug 2026.

## Builds on (most cited first, 24 of 24)

- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [Scaling Rectified Flow Transformers for High-Resolution Image Synthesis](https://mlsanity.com/paper/2403.03206/) (2024)
- [LAION-5B: An open large-scale dataset for training next generation image-text models](https://mlsanity.com/paper/2210.08402/) (2022)
- [GenEval: An Object-Focused Framework for Evaluating Text-to-Image Alignment](https://mlsanity.com/paper/2310.11513/) (2023)
- [Flow-GRPO: Training Flow Matching Models via Online RL](https://mlsanity.com/paper/2505.05470/) (2025)
- [Pref-GRPO: Pairwise Preference Reward-based GRPO for Stable Text-to-Image Reinforcement Learning](https://mlsanity.com/paper/2508.20751/) (2025)
- [Uniworld-V2: Reinforce Image Editing with Diffusion Negative-aware Finetuning and MLLM Implicit Feedback](https://mlsanity.com/paper/2510.16888/) (2025)
- [Advantage Weighted Matching: Aligning RL with Pretraining in Diffusion Models](https://mlsanity.com/paper/2509.25050/) (2025)
- [DiffusionOPD: A Unified Perspective of On-Policy Distillation in Diffusion Models](https://mlsanity.com/paper/2605.15055/) (2026)
- [Taming Preference Mode Collapse via Directional Decoupling Alignment in Diffusion Reinforcement Learning](https://mlsanity.com/paper/2512.24146/) (2025)
- [DiverseGRPO: Mitigating Mode Collapse in Image Generation via Diversity-Aware GRPO](https://mlsanity.com/paper/2512.21514/) (2025)
- [Stage-wise Dynamics of Classifier-Free Guidance in Diffusion Models](https://mlsanity.com/paper/2509.22007/) (2025)
- [The Image as Its Own Reward: Reinforcement Learning with Adversarial Reward for Image Generation](https://mlsanity.com/paper/2511.20256/) (2025)
- [Alleviating Sparse Rewards by Modeling Step-Wise and Long-Term Sampling Effects in Flow-Based GRPO](https://mlsanity.com/paper/2602.06422/) (2026)
- [FeRA: Frequency-Energy Constrained Routing for Effective Diffusion Adaptation Fine-Tuning](https://mlsanity.com/paper/2511.17979/) (2025)
- [SOAR: Self-Correction for Optimal Alignment and Refinement in Diffusion Models](https://mlsanity.com/paper/2604.12617/) (2026)
- [Flow-DPPO: Divergence Proximal Policy Optimization for Flow Matching Models](https://mlsanity.com/paper/2606.11025/) (2026)
- [Beyond the Dirac Delta: Mitigating Diversity Collapse in Reinforcement Fine-Tuning for Versatile Image Generation](https://mlsanity.com/paper/2601.12401/) (2026)
- [Embedding-perturbed Exploration Preference Optimization for Flow Models](https://mlsanity.com/paper/2605.15803/) (2026)
- [TMPO: Trajectory Matching Policy Optimization for Diverse and Efficient Diffusion Alignment](https://mlsanity.com/paper/2605.10983/) (2026)
- [When Policy Entropy Constraint Fails: Preserving Diversity in Flow-based RLHF via Perceptual Entropy](https://mlsanity.com/paper/2605.12112/) (2026)
- [Advances in GRPO for Generation Models: A Survey](https://mlsanity.com/paper/2603.06623/) (2026)
- [Stitched Value Model for Diffusion Alignment](https://mlsanity.com/paper/2605.19804/) (2026)
- [Optimizing Visual Generative Models via Distribution-wise Rewards](https://mlsanity.com/paper/2607.02291/) (2026)

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

- [MEND: RL For Flow Models via Proximal Velocity Matching](https://mlsanity.com/paper/2610.05954/) (2026)
- [GuidanceNFT: NFT-Based Reward Fine-Tuning of Guided Diffusion Models](https://mlsanity.com/paper/2610.15856/) (2026)

## Abstract (arXiv)

Reward post-training of diffusion generators inevitably concentrates probability mass on a few reward-favored modes, a mode collapse that erases within-prompt diversity. Existing methods for mitigating collapse rely on external signals or interfaces, augmenting the reward with perceptual objectives, adjusting reference regularization, or modifying the text encoder, but none repairs an adapter that has already collapsed while preserving the acquired reward. We observe that online post-training primarily reallocates probability mass over capabilities inherited from pretraining rather than learning new visual content. Collapse is therefore suppression, not deletion, and can be reversed from within the generator. We propose ReNFT, which repairs a high-reward, low-diversity adapter through internal probability-mass recalibration. Unconditional probes first prioritize "anti-hub" prompts where the prompt-independent bias is easiest to expose. Two policy-dominated mixed routes then generate matched counterfactual proposals from the same prompt and initial noise, one probing the frozen base direction for suppressed alternatives and the other exposing the post-trained unconditional tendency. Reward ranking with an adaptive flipping guard assigns pull and push roles, and a joint-and-paired NFT update realizes the repair. On PickScore and GenEval, ReNFT retains 98.9% and 99.0% of NFT's reward while improving DreamSim-Div by 58.8% and 55.0%, respectively, offering a complementary alternative to external interventions.