Gradient-Guided Reward Optimization for Inference-time Alignment

• URL: http://arxiv.org/abs/2606.09635v1
• Date: Mon, 08 Jun 2026 15:33:13 GMT
• Title: Gradient-Guided Reward Optimization for Inference-time Alignment
• Abstract summary: We introduce a lightweight inference-time method that performs targeted, minimal intervention during decoding via gradient guidance.<n>Experiments show that GGRO consistently improves inference-time alignment across safety, helpfulness, and reasoning benchmarks.<n>It also increases coverage of high-quality responses and robustness to reward hacking, with minimal computational overhead.
• Score: 22.28081683932275
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Ensuring the reliability of Large Language Models (LLMs) under distribution drift requires inference-time adaptation. While inference-time alignment methods such as Best-of-$N$ and rejection sampling are widely used, they frame the task as a sampling-intensive, reward-guided search, leading to two key limitations: their performance is bounded by the base model's generation quality, and their reliance on imperfect reward models makes them vulnerable to reward hacking. To address these challenges, we introduce Gradient-Guided Reward Optimization (GGRO), a lightweight inference-time method that performs targeted, minimal intervention during decoding via gradient guidance. Specifically, GGRO monitors token-level entropy to identify high-uncertainty regions indicative of drift or misalignment. Upon detection, it responds by injecting nudging tokens, generated using gradient signals from an off-the-shelf reward model, to steer the generation trajectory rather than merely re-ranking samples. Experiments show that GGRO consistently improves inference-time alignment across safety, helpfulness, and reasoning benchmarks. It also increases coverage of high-quality responses and robustness to reward hacking, with minimal computational overhead. Code is available at https://github.com/lhk2004/GGRO.

Related papers

• Gradients Know What Outcomes Don't: Unlocking Reinforcement Learning for LLM Reasoning with Gradient-Aligned Rewards [24.636191382674184]
  Gradient Reward (GAR) operates in the policy's own gradient space.<n>C cosine similarity with an expert-anchor gradient yields a dense, reasoning-aware reward.
  arXiv  Detail & Related papers  (2026-09-03T04:00:28Z)
• Latent Reward Registers for Diffusion Preference Alignment [46.49072106092868]
  We propose Latent Reward Registers, a mechanism that estimates terminal preference directly from intermediate noisy latents.<n>This independent readout mechanism extracts latent reward evidence without altering the generator's hidden states or velocity field.<n>For training, Reward-Gradient On-Policy Distillation (RG-OPD) distills reward-guided updates along on-policy trajectories, bypassing the computationally expensive rollouts of standard policy gradients.<n>For inference, Reward-Guided Sampling (RGS) steers trajectories via magnitude-matched reward gradients without parameter updates.
  arXiv  Detail & Related papers  (2026-08-04T17:00:52Z)
• GIRL-DETR: Gradient-Isolated Reinforcement Learning for Video Moment Retrieval [10.034595422596562]
  Video Moment Retrieval (VMR) task requires accurately localizing temporal boundaries aligned with natural language queries.<n>GIRL-DETR introduces RL post-training into a lightweight temporal localization framework for the first time.
  arXiv  Detail & Related papers  (2026-05-30T15:40:00Z)
• Mitigating the ID-OOD Tradeoff in Open-Set Test-Time Adaptation [52.620992698137236]
  Open-set test-time adaptation (OSTTA) addresses the challenge of adapting models to new environments.<n>In this paper, we first analyze the limitations of entropy in OSTTA and then introduce an angular loss to regulate feature norm magnitudes.<n>Our method achieves strong OOD detection while maintaining high ID classification performance.
  arXiv  Detail & Related papers  (2026-04-02T04:00:57Z)
• Beyond Verifiable Rewards: Rubric-Based GRM for Reinforced Fine-Tuning SWE Agents [9.505140329883762]
  We introduce a Generative Reward Model (GRM) that provides richer learning signals.<n>The GRM is equipped with human-designed rubrics that indicate criteria for encouraging or discouraging specific behavioral patterns.<n>When used for Reinforced Fine-Tuning (RFT) on SWE Tasks, our approach outperforms terminal-score-only rejection sampling.
  arXiv  Detail & Related papers  (2026-03-13T02:23:49Z)
• PROMISE: Process Reward Models Unlock Test-Time Scaling Laws in Generative Recommendations [52.67948063133533]
  Generative Recommendation has emerged as a promising paradigm, reformulating recommendation as a sequence-to-sequence generation task over hierarchical Semantic IDs.<n>Existing methods suffer from a critical issue we term Semantic Drift, where errors in early, high-level tokens irreversibly divert the generation trajectory into irrelevant semantic subspaces.<n>We propose Promise, a novel framework that integrates dense, step-by-step verification into generative models.
  arXiv  Detail & Related papers  (2026-01-08T07:38:46Z)
• GDRO: Group-level Reward Post-training Suitable for Diffusion Models [55.948229011478304]
  Group-level rewards successfully align the model with the targeted reward.<n>Group-level Direct Reward Optimization (GDRO) is a new post-training paradigm for group-level reward alignment.<n>GDRO supports full offline training that saves the large time cost for image rollout sampling.<n>It is diffusion-sampler-independent, which eliminates the need for the ODE-to-SDE approximation to obtainity.
  arXiv  Detail & Related papers  (2026-01-05T11:47:18Z)
• GARDO: Reinforcing Diffusion Models without Reward Hacking [54.841464430913476]
  Fine-tuning diffusion models via online reinforcement learning (RL) has shown great potential for enhancing text-to-image alignment.<n>The mismatch often leads to reward hacking, where proxy scores increase while real image quality deteriorates and generation diversity collapses.<n>We propose Gated and Adaptive Regularization with Diversity-aware Optimization (GARDO) to address the competing demands of sample efficiency, effective exploration, and mitigation of reward hacking.
  arXiv  Detail & Related papers  (2025-12-30T10:55:45Z)
• Inference-Time Scaling of Diffusion Language Models with Particle Gibbs Sampling [70.8832906871441]
  We study how to steer generation toward desired rewards without retraining the models.<n>Prior methods typically resample or filter within a single denoising trajectory, optimizing rewards step-by-step without trajectory-level refinement.<n>We introduce particle Gibbs sampling for diffusion language models (PG-DLM), a novel inference-time algorithm enabling trajectory-level refinement while preserving generation perplexity.
  arXiv  Detail & Related papers  (2025-07-11T08:00:47Z)
• Intra-Trajectory Consistency for Reward Modeling [67.84522106537274]
  We develop an intra-trajectory consistency regularization to enforce that adjacent processes with higher next-token generation probability maintain more consistent rewards.<n>We show that the reward model trained with the proposed regularization induces better DPO-aligned policies and achieves better best-of-N (BON) inference-time verification results.
  arXiv  Detail & Related papers  (2025-06-10T12:59:14Z)
• Diffusion Sampling Path Tells More: An Efficient Plug-and-Play Strategy for Sample Filtering [18.543769006014383]
  Diffusion models often exhibit inconsistent sample quality due to variations inherent in their sampling trajectories.<n>We introduce CFG-Rejection, an efficient, plug-and-play strategy that filters low-quality samples at an early stage of the denoising process.<n>We validate the effectiveness of CFG-Rejection in image generation through extensive experiments.
  arXiv  Detail & Related papers  (2025-05-29T11:08:24Z)
• InfAlign: Inference-aware language model alignment [58.66389179049758]
  Language model alignment is a critical step in training modern generative language models.<n>We show that this train/test mismatch makes standard RLHF framework sub-optimal in view of inference-time methods.<n>We propose a framework for inference-aware alignment (InfAlign), which aims to optimize inference-time win rate of the aligned policy against the base model.
  arXiv  Detail & Related papers  (2024-12-27T18:45:36Z)

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.