{"total":124,"rows":[{"id":"2411.02306","title":"On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback","created":"2024-11-04","cited_by_count":47,"area":"alignment"},{"id":"2503.11701","title":"A Survey of Direct Preference Optimization","created":"2025-03-12","cited_by_count":36,"area":"alignment"},{"id":"2507.09650","title":"Cultivating Pluralism In Algorithmic Monoculture: The Community Alignment Dataset","created":"2025-07-13","cited_by_count":35,"area":"alignment"},{"id":"2505.17147","title":"MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming","created":"2025-05-22","cited_by_count":32,"area":"alignment"},{"id":"2503.15463","title":"From 1,000,000 Users to Every User: Scaling Up Personalized Preference for User-level Alignment","created":"2025-03-19","cited_by_count":30,"area":"alignment"},{"id":"2505.06274","title":"PARM: Multi-Objective Test-Time Alignment via Preference-Aware Autoregressive Reward Model","created":"2025-05-06","cited_by_count":28,"area":"alignment"},{"id":"2506.03637","title":"RewardAnything: Generalizable Principle-Following Reward Models","created":"2025-06-04","cited_by_count":28,"area":"alignment"},{"id":"2510.26202","title":"What's In My Human Feedback? Learning Interpretable Descriptions of Preference Data","created":"2025-10-30","cited_by_count":26,"area":"alignment"},{"id":"2507.01915","title":"Gradient-Adaptive Policy Optimization: Towards Multi-Objective Alignment of Large Language Models","created":"2025-07-02","cited_by_count":23,"area":"alignment"},{"id":"2506.09996","title":"From Judgment to Interference: Early Stopping LLM Harmful Outputs via Streaming Content Monitoring","created":"2025-06-11","cited_by_count":22,"area":"alignment"},{"id":"2509.01909","title":"Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Models","created":"2025-09-02","cited_by_count":19,"area":"alignment"},{"id":"2505.16869","title":"MPO: Multilingual Safety Alignment via Reward Gap Optimization","created":"2025-05-22","cited_by_count":18,"area":"alignment"},{"id":"2503.08942","title":"Extragradient Preference Optimization (EGPO): Beyond Last-Iterate Convergence for Nash Learning from Human Feedback","created":"2025-03-11","cited_by_count":16,"area":"alignment"},{"id":"2505.20259","title":"Lifelong Safety Alignment for Language Models","created":"2025-05-26","cited_by_count":15,"area":"alignment"},{"id":"2410.15483","title":"Understanding Forgetting in LLM Supervised Fine-Tuning and Preference Learning -- A Convex Optimization Perspective","created":"2024-10-20","cited_by_count":12,"area":"alignment"},{"id":"2411.12762","title":"Playing Language Game with LLMs Leads to Jailbreaking","created":"2024-11-16","cited_by_count":11,"area":"alignment"},{"id":"2505.24846","title":"MiCRo: Mixture Modeling and Context-aware Routing for Personalized Preference Learning","created":"2025-05-30","cited_by_count":11,"area":"alignment"},{"id":"2512.11391","title":"Mitigating the Safety Alignment Tax with Null-Space Constrained Policy Optimization","created":"2025-12-12","cited_by_count":11,"area":"alignment"},{"id":"2411.08302","title":"RED: Unleashing Token-Level Rewards from Holistic Feedback via Reward Redistribution","created":"2024-11-13","cited_by_count":10,"area":"alignment"},{"id":"2409.00162","title":"Sequence to Sequence Reward Modeling: Improving RLHF by Language Feedback","created":"2024-08-30","cited_by_count":9,"area":"alignment"},{"id":"2505.12225","title":"Mining Intrinsic Rewards from LLM Hidden States for Efficient Best-of-N Sampling","created":"2025-05-18","cited_by_count":9,"area":"alignment"},{"id":"2505.23729","title":"Bounded Rationality for LLMs: Satisficing Alignment at Inference-Time","created":"2025-05-29","cited_by_count":9,"area":"alignment"},{"id":"2506.06636","title":"SafeLawBench: Towards Safe Alignment of Large Language Models","created":"2025-06-07","cited_by_count":9,"area":"alignment"},{"id":"2602.16660","title":"Align Once, Benefit Multilingually: Enforcing Multilingual Consistency for LLM Safety Alignment","created":"2026-02-18","cited_by_count":9,"area":"alignment"},{"id":"2601.01887","title":"Safety at One Shot: Patching Fine-Tuned LLMs with A Single Instance","created":"2026-01-05","cited_by_count":8,"area":"alignment"},{"id":"2602.00038","title":"LSSF: Safety Alignment for Large Language Models through Low-Rank Safety Subspace Fusion","created":"2026-01-19","cited_by_count":8,"area":"alignment"},{"id":"2509.24610","title":"OrthAlign: Orthogonal Subspace Decomposition for Non-Interfering Multi-Objective Alignment","created":"2025-09-29","cited_by_count":7,"area":"alignment"},{"id":"2603.18736","title":"CausalRM: Causal-Theoretic Reward Modeling for RLHF from Observational User Feedbacks","created":"2026-03-19","cited_by_count":7,"area":"alignment"},{"id":"2501.15453","title":"Data-adaptive Safety Rules for Training Reward Models","created":"2025-01-26","cited_by_count":6,"area":"alignment"},{"id":"2504.20106","title":"Adaptive Helpfulness-Harmlessness Alignment with Preference Vectors","created":"2025-04-27","cited_by_count":6,"area":"alignment"},{"id":"2505.14185","title":"Safety Subspaces are Not Linearly Distinct: A Fine-Tuning Case Study","created":"2025-05-20","cited_by_count":6,"area":"alignment"},{"id":"2505.23473","title":"EVOREFUSE: Evolutionary Prompt Optimization for Evaluation and Mitigation of LLM Over-Refusal to Pseudo-Malicious Instructions","created":"2025-05-29","cited_by_count":6,"area":"alignment"},{"id":"2506.16502","title":"Relic: Enhancing Reward Model Generalization for Low-Resource Indic Languages with Few-Shot Examples","created":"2025-06-19","cited_by_count":6,"area":"alignment"},{"id":"2606.19818","title":"Uncertainty-Aware Reward Modeling for Stable RLHF","created":"2026-06-18","cited_by_count":6,"area":"alignment"},{"id":"2501.13264","title":"OpenGenAlign: A Preference Dataset and Benchmark for Trustworthy Reward Modeling in Open-Ended, Long-Context Generation","created":"2025-01-22","cited_by_count":5,"area":"alignment"},{"id":"2512.23816","title":"Improved Bounds for Private and Robust Alignment","created":"2025-12-29","cited_by_count":5,"area":"alignment"},{"id":"2503.20995","title":"ENCORE: Entropy-guided Reward Composition for Multi-head Safety Reward Models","created":"2025-03-26","cited_by_count":4,"area":"alignment"},{"id":"2602.13576","title":"Rubrics as an Attack Surface: Stealthy Preference Drift in LLM Judges","created":"2026-02-14","cited_by_count":4,"area":"alignment"},{"id":"2603.07445","title":"Few Tokens, Big Leverage: Preserving Safety Alignment by Constraining Safety Tokens during Fine-tuning","created":"2026-03-08","cited_by_count":4,"area":"alignment"},{"id":"2502.02153","title":"Vulnerability Mitigation for Safety-Aligned Language Models via Debiasing","created":"2025-02-04","cited_by_count":3,"area":"alignment"},{"id":"2506.02460","title":"MidPO: Dual Preference Optimization for Safety and Helpfulness in Large Language Models via a Mixture of Experts Framework","created":"2025-06-03","cited_by_count":3,"area":"alignment"},{"id":"2510.19479","title":"Graph Unlearning Meets Influence-aware Negative Preference Optimization","created":"2025-10-22","cited_by_count":3,"area":"alignment"},{"id":"2511.06222","title":"SPA: Achieving Consensus in LLM Alignment via Self-Priority Optimization","created":"2025-11-09","cited_by_count":3,"area":"alignment"},{"id":"2511.12464","title":"Probing Preference Representations: A Multi-Dimensional Evaluation and Analysis Method for Reward Models","created":"2025-11-16","cited_by_count":3,"area":"alignment"},{"id":"2511.17937","title":"Alignment Faking - the Train -> Deploy Asymmetry: Through a Game-Theoretic Lens with Bayesian-Stackelberg Equilibria","created":"2025-11-22","cited_by_count":3,"area":"alignment"},{"id":"2602.09538","title":"UniARM: Towards a Unified Autoregressive Reward Model for Multi-Objective Test-Time Alignment","created":"2026-02-10","cited_by_count":3,"area":"alignment"},{"id":"2602.10286","title":"What Does Preference Learning Recover from Pairwise Comparison Data?","created":"2026-02-10","cited_by_count":3,"area":"alignment"},{"id":"2603.23184","title":"ImplicitRM: Unbiased Reward Modeling from Implicit Preference Data for LLM alignment","created":"2026-03-24","cited_by_count":3,"area":"alignment"},{"id":"2604.12710","title":"LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety","created":"2026-04-13","cited_by_count":3,"area":"alignment"},{"id":"2605.02971","title":"Multilingual Safety Alignment via Self-Distillation","created":"2026-05-03","cited_by_count":3,"area":"alignment"},{"id":"2605.06036","title":"Optimal Transport for LLM Reward Modeling from Noisy Preference","created":"2026-05-07","cited_by_count":3,"area":"alignment"},{"id":"2606.16771","title":"GD$^2$PO: Mitigating Multi-Reward Conflicts via Group-Dynamic reward-Decoupled Policy Optimization","created":"2026-06-15","cited_by_count":3,"area":"alignment"},{"id":"2501.05336","title":"Stream Aligner: Efficient Sentence-Level Alignment via Distribution Induction","created":"2025-01-09","cited_by_count":2,"area":"alignment"},{"id":"2502.14400","title":"HPS: Hard Preference Sampling for Human Preference Alignment","created":"2025-02-20","cited_by_count":2,"area":"alignment"},{"id":"2502.17173","title":"Cheems: A Practical Guidance for Building and Evaluating Chinese Reward Models from Scratch","created":"2025-02-24","cited_by_count":2,"area":"alignment"},{"id":"2505.23114","title":"Alignment Data Map for Efficient Preference Data Selection and Diagnosis","created":"2025-05-29","cited_by_count":2,"area":"alignment"},{"id":"2509.23564","title":"Clean First, Align Later: Benchmarking Preference Data Cleaning for Reliable LLM Alignment","created":"2025-09-28","cited_by_count":2,"area":"alignment"},{"id":"2601.05882","title":"An Empirical Study on Preference Tuning Generalization and Diversity Under Domain Shift","created":"2026-01-09","cited_by_count":2,"area":"alignment"},{"id":"2602.06869","title":"Uncovering Cross-Objective Interference in Multi-Objective Alignment","created":"2026-02-06","cited_by_count":2,"area":"alignment"},{"id":"2603.10243","title":"GR-SAP: Generative Replay for Safety Alignment Preservation during Fine-Tuning","created":"2026-03-10","cited_by_count":2,"area":"alignment"},{"id":"2603.18113","title":"VC-Soup: Value-Consistency Guided Multi-Value Alignment for Large Language Models","created":"2026-03-18","cited_by_count":2,"area":"alignment"},{"id":"2604.10673","title":"Principles Do Not Apply Themselves: A Hermeneutic Perspective on AI Alignment","created":"2026-04-12","cited_by_count":2,"area":"alignment"},{"id":"2411.02481","title":"Dr. SoW: Density Ratio of Strong-over-weak LLMs for Reducing the Cost of Human Annotation in Preference Tuning","created":"2024-11-04","cited_by_count":1,"area":"alignment"},{"id":"2505.19743","title":"Token-level Accept or Reject: A Micro Alignment Approach for Large Language Models","created":"2025-05-26","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":"2508.04626","title":"P-Aligner: Enabling Pre-Alignment of Language Models via Principled Instruction Synthesis","created":"2025-08-06","cited_by_count":1,"area":"alignment"},{"id":"2508.20038","title":"Forewarned is Forearmed: Pre-Synthesizing Jailbreak-like Instructions to Enhance LLM Safety Guardrail to Potential Attacks","created":"2025-08-27","cited_by_count":1,"area":"alignment"},{"id":"2512.04210","title":"Balancing Safety and Helpfulness in Healthcare AI Assistants through Iterative Preference Alignment","created":"2025-12-03","cited_by_count":1,"area":"alignment"},{"id":"2512.06515","title":"ProSocialAlign: Preference Conditioned Test Time Alignment in Language Models","created":"2025-12-06","cited_by_count":1,"area":"alignment"},{"id":"2512.13837","title":"Explainable reinforcement learning from human feedback to improve alignment","created":"2025-12-15","cited_by_count":1,"area":"alignment"},{"id":"2603.02225","title":"Scaling Reward Modeling without Human Supervision","created":"2026-02-11","cited_by_count":1,"area":"alignment"},{"id":"2604.17299","title":"Cat-DPO: Category-Adaptive Safety Alignment","created":"2026-04-19","cited_by_count":1,"area":"alignment"},{"id":"2605.01899","title":"Disentangling Intent from Role: Adversarial Self-Play for Persona-Invariant Safety Alignment","created":"2026-05-03","cited_by_count":1,"area":"alignment"},{"id":"2605.05003","title":"Misaligned by Reward: Socially Undesirable Preferences in LLMs","created":"2026-05-06","cited_by_count":1,"area":"alignment"},{"id":"2605.11679","title":"Explaining and Breaking the Safety-Helpfulness Ceiling via Preference Dimensional Expansion","created":"2026-05-12","cited_by_count":1,"area":"alignment"},{"id":"2605.29396","title":"Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization","created":"2026-05-28","cited_by_count":1,"area":"alignment"},{"id":"2605.30619","title":"Reward Learning from Best-of-$N$ Preference Data: Targets, Tradeoffs, and Design Principles","created":"2026-05-28","cited_by_count":1,"area":"alignment"},{"id":"2605.30808","title":"Differentially Private Preference Data Synthesis for Large Language Model Alignment","created":"2026-05-29","cited_by_count":1,"area":"alignment"},{"id":"2606.19744","title":"Beyond Uniform Forgetting: A Study of Sequential Direct Preference Optimization Across Preference Settings","created":"2026-06-18","cited_by_count":1,"area":"alignment"},{"id":"2606.31524","title":"On the Convergence of Self-Improving Online LLM Alignment","created":"2026-06-30","cited_by_count":1,"area":"alignment"},{"id":"2607.08034","title":"PLURAL: A Global Dataset for Value Alignment","created":"2026-07-09","cited_by_count":1,"area":"alignment"},{"id":"2608.16553","title":"STAGE: Controlled Objective Admission for Multi-Preference LLM Alignment","created":"2026-08-17","cited_by_count":1,"area":"alignment"},{"id":"2502.12189","title":"Self-supervised Attribute-aware Dynamic Preference Ranking Alignment","created":"2025-02-15","cited_by_count":0,"area":"alignment"},{"id":"2505.16003","title":"SLMEval: Entropy-Based Calibration for Human-Aligned Evaluation of Large Language Models","created":"2025-05-21","cited_by_count":0,"area":"alignment"},{"id":"2505.19134","title":"Incentivizing High-Quality Human Annotations with Golden Questions","created":"2025-05-25","cited_by_count":0,"area":"alignment"},{"id":"2506.08965","title":"GFRIEND: Generative Few-shot Reward Inference through EfficieNt DPO","created":"2025-06-10","cited_by_count":0,"area":"alignment"},{"id":"2507.09973","title":"Tiny Reward Models","created":"2025-07-14","cited_by_count":0,"area":"alignment"},{"id":"2508.13415","title":"MAVIS: Multi-Objective Alignment via Inference-Time Value-Guided Selection","created":"2025-08-19","cited_by_count":0,"area":"alignment"},{"id":"2510.09004","title":"Decoupling Safety into Orthogonal Subspace: Cost-Efficient and Performance-Preserving Alignment for Large Language Models","created":"2025-10-10","cited_by_count":0,"area":"alignment"},{"id":"2511.02623","title":"The Realignment Problem: When Right becomes Wrong in LLMs","created":"2025-11-04","cited_by_count":0,"area":"alignment"},{"id":"2511.13007","title":"GEM: Generative Entropy-Guided Preference Modeling for Few-shot Alignment of LLMs","created":"2025-11-17","cited_by_count":0,"area":"alignment"},{"id":"2512.05518","title":"Matching Ranks Over Probability Yields Truly Deep Safety Alignment","created":"2025-12-05","cited_by_count":0,"area":"alignment"},{"id":"2601.06157","title":"ECLIPTICA -- A Framework for Switchable LLM Alignment via CITA - Contrastive Instruction-Tuned Alignment","created":"2026-01-06","cited_by_count":0,"area":"alignment"},{"id":"2603.03536","title":"SafeCRS: Personalized Safety Alignment for LLM-Based Conversational Recommender Systems","created":"2026-03-03","cited_by_count":0,"area":"alignment"},{"id":"2603.04974","title":"VRM: Teaching Reward Models to Understand Authentic Human Preferences","created":"2026-03-05","cited_by_count":0,"area":"alignment"},{"id":"2603.06727","title":"Safe Transformer: An Explicit Safety Bit For Interpretable And Controllable Alignment","created":"2026-03-06","cited_by_count":0,"area":"alignment"},{"id":"2603.15647","title":"Steering Frozen LLMs: Adaptive Social Alignment via Online Prompt Routing","created":"2026-03-04","cited_by_count":0,"area":"alignment"},{"id":"2604.02574","title":"Understanding the Effects of Safety Unalignment on Large Language Models","created":"2026-04-02","cited_by_count":0,"area":"alignment"},{"id":"2604.06833","title":"FedDetox: Robust Federated SLM Alignment via On-Device Data Sanitization","created":"2026-04-08","cited_by_count":0,"area":"alignment"},{"id":"2604.10966","title":"You Only Judge Once: Multi-response Reward Modeling in a Single Forward Pass","created":"2026-04-13","cited_by_count":0,"area":"alignment"}],"next_cursor":"100"}