{"total":81,"rows":[{"id":"2410.12832","title":"Generative Reward Models","created":"2024-10-02","cited_by_count":148,"area":"alignment"},{"id":"2407.13399","title":"Correcting the Mythos of KL-Regularization: Direct Alignment without Overoptimization via Chi-Squared Preference Optimization","created":"2024-07-18","cited_by_count":78,"area":"alignment"},{"id":"2407.17387","title":"PERSONA: A Reproducible Testbed for Pluralistic Alignment","created":"2024-07-24","cited_by_count":73,"area":"alignment"},{"id":"2410.08847","title":"Unintentional Unalignment: Likelihood Displacement in Direct Preference Optimization","created":"2024-10-11","cited_by_count":67,"area":"alignment"},{"id":"2406.11827","title":"WPO: Enhancing RLHF with Weighted Preference Optimization","created":"2024-06-17","cited_by_count":54,"area":"alignment"},{"id":"2408.06266","title":"Anchored Preference Optimization and Contrastive Revisions: Addressing Underspecification in Alignment","created":"2024-08-12","cited_by_count":46,"area":"alignment"},{"id":"2409.20370","title":"The Perfect Blend: Redefining RLHF with Mixture of Judges","created":"2024-09-30","cited_by_count":43,"area":"alignment"},{"id":"2505.20065","title":"SafeDPO: A Simple Approach to Direct Preference Optimization with Enhanced Safety","created":"2025-05-26","cited_by_count":37,"area":"alignment"},{"id":"2604.13602","title":"Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges","created":"2026-04-15","cited_by_count":33,"area":"alignment"},{"id":"2406.10957","title":"Eliminating Biased Length Reliance of Direct Preference Optimization via Down-Sampled KL Divergence","created":"2024-06-16","cited_by_count":31,"area":"alignment"},{"id":"2410.10148","title":"AlphaDPO: Adaptive Reward Margin for Direct Preference Optimization","created":"2024-10-14","cited_by_count":29,"area":"alignment"},{"id":"2501.12735","title":"Online Preference Alignment for Language Models via Count-based Exploration","created":"2025-01-22","cited_by_count":26,"area":"alignment"},{"id":"2412.06000","title":"Does RLHF Scale? Exploring the Impacts From Data, Model, and Method","created":"2024-12-08","cited_by_count":23,"area":"alignment"},{"id":"2503.00295","title":"Robust Multi-Objective Preference Alignment with Online DPO","created":"2025-03-01","cited_by_count":22,"area":"alignment"},{"id":"2501.19358","title":"The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward Hacking","created":"2025-01-31","cited_by_count":18,"area":"alignment"},{"id":"2410.10093","title":"How to Leverage Demonstration Data in Alignment for Large Language Model? A Self-Imitation Learning Perspective","created":"2024-10-14","cited_by_count":14,"area":"alignment"},{"id":"2501.03884","title":"AlphaPO: Reward Shape Matters for LLM Alignment","created":"2025-01-07","cited_by_count":14,"area":"alignment"},{"id":"2412.09544","title":"Sail into the Headwind: Alignment via Robust Rewards and Dynamic Labels against Reward Hacking","created":"2024-12-12","cited_by_count":13,"area":"alignment"},{"id":"2505.05465","title":"ComPO: Preference Alignment via Comparison Oracles","created":"2025-05-08","cited_by_count":13,"area":"alignment"},{"id":"2505.07271","title":"On the Robustness of Reward Models for Language Model Alignment","created":"2025-05-12","cited_by_count":13,"area":"alignment"},{"id":"2409.17401","title":"Zeroth-Order Policy Gradient for Reinforcement Learning from Human Feedback without Reward Inference","created":"2024-09-25","cited_by_count":12,"area":"alignment"},{"id":"2503.00539","title":"Distributionally Robust Reinforcement Learning with Human Feedback","created":"2025-03-01","cited_by_count":12,"area":"alignment"},{"id":"2505.18126","title":"Reward Model Overoptimisation in Iterated RLHF","created":"2025-05-23","cited_by_count":12,"area":"alignment"},{"id":"2505.12763","title":"Rethinking Reward Model Evaluation Through the Lens of Reward Overoptimization","created":"2025-05-19","cited_by_count":11,"area":"alignment"},{"id":"2506.21463","title":"Aligning Spoken Dialogue Models from User Interactions","created":"2025-06-26","cited_by_count":11,"area":"alignment"},{"id":"2502.19731","title":"Preference Learning Unlocks LLMs' Psycho-Counseling Skills","created":"2025-02-27","cited_by_count":10,"area":"alignment"},{"id":"2506.08712","title":"ConfPO: Exploiting Policy Model Confidence for Critical Token Selection in Preference Optimization","created":"2025-06-10","cited_by_count":10,"area":"alignment"},{"id":"2607.05904","title":"More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges","created":"2026-07-07","cited_by_count":10,"area":"alignment"},{"id":"2407.06542","title":"LIONs: An Empirically Optimized Approach to Align Language Models","created":"2024-07-09","cited_by_count":7,"area":"alignment"},{"id":"2410.01735","title":"LASeR: Learning to Adaptively Select Reward Models with Multi-Armed Bandits","created":"2024-10-02","cited_by_count":7,"area":"alignment"},{"id":"2410.11677","title":"Understanding Likelihood Over-optimisation in Direct Alignment Algorithms","created":"2024-10-15","cited_by_count":7,"area":"alignment"},{"id":"2510.19050","title":"Rectifying Shortcut Behaviors in Preference-based Reward Learning","created":"2025-10-21","cited_by_count":6,"area":"alignment"},{"id":"2502.13177","title":"KL Penalty Control via Perturbation for Direct Preference Optimization","created":"2025-02-18","cited_by_count":5,"area":"alignment"},{"id":"2505.20556","title":"Learning a Pessimistic Reward Model in RLHF","created":"2025-05-26","cited_by_count":5,"area":"alignment"},{"id":"2510.13694","title":"Information-Theoretic Reward Modeling for Stable RLHF: Detecting and Mitigating Reward Hacking","created":"2025-10-15","cited_by_count":5,"area":"alignment"},{"id":"2410.02451","title":"Strong Preferences Affect the Robustness of Preference Models and Value Alignment","created":"2024-10-03","cited_by_count":4,"area":"alignment"},{"id":"2502.06861","title":"Design Considerations in Offline Preference-based RL","created":"2025-02-08","cited_by_count":4,"area":"alignment"},{"id":"2502.13603","title":"Efficient Safety Retrofitting Against Jailbreaking for LLMs","created":"2025-02-19","cited_by_count":4,"area":"alignment"},{"id":"2502.17721","title":"Aligning Compound AI Systems via System-level DPO","created":"2025-02-24","cited_by_count":4,"area":"alignment"},{"id":"2503.07426","title":"RePO: Understanding Preference Learning Through ReLU-Based Optimization","created":"2025-03-10","cited_by_count":4,"area":"alignment"},{"id":"2506.08266","title":"Reinforcement Learning from Human Feedback with High-Confidence Safety Constraints","created":"2025-06-09","cited_by_count":4,"area":"alignment"},{"id":"2602.11523","title":"Unifying Stable Optimization and Reference Regularization in RLHF","created":"2026-02-12","cited_by_count":4,"area":"alignment"},{"id":"2605.04266","title":"Explaining and Preventing Alignment Collapse in Iterative RLHF","created":"2026-05-05","cited_by_count":4,"area":"alignment"},{"id":"2606.09076","title":"Z-Reward: Beyond Scalar Rewards by Internalizing Reasoning into Score Distributions","created":"2026-06-08","cited_by_count":4,"area":"alignment"},{"id":"2410.07025","title":"CheXalign: Preference fine-tuning in chest X-ray interpretation models without human feedback","created":"2024-10-09","cited_by_count":3,"area":"alignment"},{"id":"2502.01237","title":"The Differences Between Direct Alignment Algorithms are a Blur","created":"2025-02-03","cited_by_count":3,"area":"alignment"},{"id":"2603.04851","title":"Why Is RLHF Alignment Shallow? A Gradient Analysis","created":"2026-03-05","cited_by_count":2,"area":"alignment"},{"id":"2603.10938","title":"Safe RLHF Beyond Expectation: Stochastic Dominance for Universal Spectral Risk Control","created":"2026-03-11","cited_by_count":2,"area":"alignment"},{"id":"2605.06977","title":"$f$-Divergence Regularized RLHF: Two Tales of Sampling and Unified Analyses","created":"2026-05-07","cited_by_count":2,"area":"alignment"},{"id":"2606.03962","title":"Using Reward Uncertainty to Induce Diverse Behaviour in Reinforcement Learning","created":"2026-06-02","cited_by_count":2,"area":"alignment"},{"id":"2411.10545","title":"Efficient Alignment of Large Language Models via Data Sampling","created":"2024-11-15","cited_by_count":1,"area":"alignment"},{"id":"2412.16475","title":"When Can Proxies Improve the Sample Complexity of Preference Learning?","created":"2024-12-21","cited_by_count":1,"area":"alignment"},{"id":"2506.00152","title":"Aligning Language Models with Observational Data: Opportunities and Risks from a Causal Perspective","created":"2025-05-30","cited_by_count":1,"area":"alignment"},{"id":"2506.08681","title":"Mitigating Reward Over-optimization in Direct Alignment Algorithms with Importance Sampling","created":"2025-06-10","cited_by_count":1,"area":"alignment"},{"id":"2506.09457","title":"Towards Bridging the Reward-Generation Gap in Direct Alignment Algorithms","created":"2025-06-11","cited_by_count":1,"area":"alignment"},{"id":"2509.22851","title":"Adaptive Margin RLHF via Preference over Preferences","created":"2025-09-26","cited_by_count":1,"area":"alignment"},{"id":"2509.26169","title":"Alignment-Aware Decoding","created":"2025-09-30","cited_by_count":1,"area":"alignment"},{"id":"2511.12036","title":"Preference Learning from Physics-Based Feedback: Tuning Language Models to Design BCC/B2 Superalloys","created":"2025-11-15","cited_by_count":1,"area":"alignment"},{"id":"2512.15765","title":"Shapley-based Data Valuation for LLM Alignment via Sequential Preference Optimization","created":"2025-12-12","cited_by_count":1,"area":"alignment"},{"id":"2601.21350","title":"Factored Causal Representation Learning for Robust Reward Modeling in RLHF","created":"2026-01-29","cited_by_count":1,"area":"alignment"},{"id":"2602.11902","title":"Mitigating Mismatch within Reference-based Preference Optimization","created":"2026-02-12","cited_by_count":1,"area":"alignment"},{"id":"2606.00291","title":"The Representation-Rationalizability Tradeoff in Reward Learning","created":"2026-05-29","cited_by_count":1,"area":"alignment"},{"id":"2606.04145","title":"EvalStop: Using World Feedback to Detect and Correct Reward Overoptimization in Multi-Tenant RLHF Platforms","created":"2026-06-02","cited_by_count":1,"area":"alignment"},{"id":"2607.02781","title":"Safe Inference-Time Alignment via Lagrangian Reward Augmentation","created":"2026-07-02","cited_by_count":1,"area":"alignment"},{"id":"2607.16240","title":"Normalized Rewards for Preference Optimization","created":"2026-06-26","cited_by_count":1,"area":"alignment"},{"id":"2509.20265","title":"Failure Modes of Maximum Entropy RLHF","created":"2025-09-24","cited_by_count":0,"area":"alignment"},{"id":"2601.06108","title":"From RLHF to Direct Alignment: A Theoretical Unification of Preference Learning for Large Language Models","created":"2026-01-03","cited_by_count":0,"area":"alignment"},{"id":"2603.08145","title":"DARC: Disagreement-Aware Alignment via Risk-Constrained Decoding","created":"2026-03-09","cited_by_count":0,"area":"alignment"},{"id":"2603.10279","title":"Robust Post-Training for Generative Recommenders: Why Exponential Reward-Weighted SFT Outperforms RLHF","created":"2026-03-10","cited_by_count":0,"area":"alignment"},{"id":"2603.17378","title":"Efficient Exploration at Scale","created":"2026-03-18","cited_by_count":0,"area":"alignment"},{"id":"2605.04356","title":"Efficiently Aligning Language Models with Online Natural Language Feedback","created":"2026-05-05","cited_by_count":0,"area":"alignment"},{"id":"2605.10981","title":"$\\xi$-DPO: Direct Preference Optimization via Ratio Reward Margin","created":"2026-05-09","cited_by_count":0,"area":"alignment"},{"id":"2605.20994","title":"Towards Context-Invariant Safety Alignment for Large Language Models","created":"2026-05-20","cited_by_count":0,"area":"alignment"},{"id":"2607.16195","title":"Rater State Bias in RLHF Preference Data: An Audit Framework","created":"2026-04-14","cited_by_count":0,"area":"alignment"},{"id":"2608.08002","title":"Evaluator Ensembles Under Reward Hacking: Covariance Geometry and Finite-Search Guarantees","created":"2026-08-08","cited_by_count":0,"area":"alignment"},{"id":"2609.16532","title":"Style-Debiased DPO: Updating LLM Knowledge with Factuality-Aware Synthetic Preference Data","created":"2026-09-15","cited_by_count":0,"area":"alignment"},{"id":"2609.19144","title":"A Zeroth-Order Paradigm for LLM Preference Alignment","created":"2026-09-16","cited_by_count":0,"area":"alignment"},{"id":"2609.32720","title":"BiasReducer: Adaptive Bias Mitigation for Reward Models","created":"2026-09-26","cited_by_count":0,"area":"alignment"},{"id":"2609.38526","title":"Revisiting scaling laws for reward optimization","created":"2026-09-29","cited_by_count":0,"area":"alignment"},{"id":"2609.38847","title":"Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics","created":"2026-09-30","cited_by_count":0,"area":"alignment"},{"id":"2610.02568","title":"Mitigating Social Sycophancy via Pluralistic Preference Optimization","created":"2026-10-01","cited_by_count":0,"area":"alignment"}],"next_cursor":null}