{"total":745,"rows":[{"id":"2307.15043","title":"Universal and Transferable Adversarial Attacks on Aligned Language Models","created":"2023-07-27","cited_by_count":3104,"area":"alignment"},{"id":"2409.19256","title":"HybridFlow: A Flexible and Efficient RLHF Framework","created":"2024-09-28","cited_by_count":2057,"area":"alignment"},{"id":"2403.04132","title":"Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference","created":"2024-03-07","cited_by_count":1184,"area":"alignment"},{"id":"2310.03693","title":"Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!","created":"2023-10-05","cited_by_count":1154,"area":"alignment"},{"id":"2310.04451","title":"AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models","created":"2023-10-03","cited_by_count":785,"area":"alignment"},{"id":"2307.15217","title":"Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback","created":"2023-07-27","cited_by_count":759,"area":"alignment"},{"id":"2404.13076","title":"LLM Evaluators Recognize and Favor Their Own Generations","created":"2024-04-15","cited_by_count":686,"area":"alignment"},{"id":"2401.10020","title":"Self-Rewarding Language Models","created":"2024-01-18","cited_by_count":681,"area":"alignment"},{"id":"2310.12773","title":"Safe RLHF: Safe Reinforcement Learning from Human Feedback","created":"2023-10-19","cited_by_count":656,"area":"alignment"},{"id":"2310.08491","title":"Prometheus: Inducing Fine-grained Evaluation Capability in Language Models","created":"2023-10-12","cited_by_count":575,"area":"alignment"},{"id":"2310.16944","title":"Zephyr: Direct Distillation of LM Alignment","created":"2023-10-25","cited_by_count":506,"area":"alignment"},{"id":"2404.05868","title":"Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning","created":"2024-04-08","cited_by_count":504,"area":"alignment"},{"id":"2403.13787","title":"RewardBench: Evaluating Reward Models for Language Modeling","created":"2024-03-20","cited_by_count":499,"area":"alignment"},{"id":"2406.05946","title":"Safety Alignment Should Be Made More Than Just a Few Tokens Deep","created":"2024-06-10","cited_by_count":472,"area":"alignment"},{"id":"2310.06387","title":"Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations","created":"2023-10-10","cited_by_count":431,"area":"alignment"},{"id":"2308.06463","title":"GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher","created":"2023-08-12","cited_by_count":423,"area":"alignment"},{"id":"2310.01377","title":"UltraFeedback: Boosting Language Models with Scaled AI Feedback","created":"2023-10-02","cited_by_count":406,"area":"alignment"},{"id":"2310.06452","title":"Understanding the Effects of RLHF on LLM Generalisation and Diversity","created":"2023-10-10","cited_by_count":396,"area":"alignment"},{"id":"2401.08417","title":"Contrastive Preference Optimization: Pushing the Boundaries of LLM Performance in Machine Translation","created":"2024-01-16","cited_by_count":394,"area":"alignment"},{"id":"2406.12845","title":"Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts","created":"2024-06-18","cited_by_count":376,"area":"alignment"},{"id":"2312.15685","title":"What Makes Good Data for Alignment? A Comprehensive Study of Automatic Data Selection in Instruction Tuning","created":"2023-12-25","cited_by_count":364,"area":"alignment"},{"id":"2312.11456","title":"Iterative Preference Learning from Human Feedback: Bridging Theory and Practice for RLHF under KL-Constraint","created":"2023-12-18","cited_by_count":350,"area":"alignment"},{"id":"2309.06657","title":"Statistical Rejection Sampling Improves Preference Optimization","created":"2023-09-13","cited_by_count":323,"area":"alignment"},{"id":"2310.03716","title":"A Long Way to Go: Investigating Length Correlations in RLHF","created":"2023-10-05","cited_by_count":291,"area":"alignment"},{"id":"2312.14925","title":"A Survey of Reinforcement Learning from Human Feedback","created":"2023-12-22","cited_by_count":264,"area":"alignment"},{"id":"2404.10719","title":"Is DPO Superior to PPO for LLM Alignment? A Comprehensive Study","created":"2024-04-16","cited_by_count":262,"area":"alignment"},{"id":"2406.15513","title":"PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference","created":"2024-06-20","cited_by_count":261,"area":"alignment"},{"id":"2405.20947","title":"OR-Bench: An Over-Refusal Benchmark for Large Language Models","created":"2024-05-31","cited_by_count":244,"area":"alignment"},{"id":"2309.15025","title":"Large Language Model Alignment: A Survey","created":"2023-09-26","cited_by_count":243,"area":"alignment"},{"id":"2402.13228","title":"Smaug: Fixing Failure Modes of Preference Optimisation with DPO-Positive","created":"2024-02-20","cited_by_count":241,"area":"alignment"},{"id":"2406.14598","title":"SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal","created":"2024-06-20","cited_by_count":240,"area":"alignment"},{"id":"2402.05162","title":"Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications","created":"2024-02-07","cited_by_count":231,"area":"alignment"},{"id":"2311.10702","title":"Camels in a Changing Climate: Enhancing LM Adaptation with Tulu 2","created":"2023-11-17","cited_by_count":229,"area":"alignment"},{"id":"2410.05451","title":"SecAlign: Defending Against Prompt Injection with Preference Optimization","created":"2024-10-07","cited_by_count":210,"area":"alignment"},{"id":"2307.10928","title":"FLASK: Fine-grained Language Model Evaluation based on Alignment Skill Sets","created":"2023-07-20","cited_by_count":201,"area":"alignment"},{"id":"2402.11411","title":"Aligning Modalities in Vision Large Language Models via Preference Fine-tuning","created":"2024-02-18","cited_by_count":198,"area":"alignment"},{"id":"2310.20624","title":"LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B","created":"2023-10-31","cited_by_count":195,"area":"alignment"},{"id":"2309.14348","title":"Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM","created":"2023-09-18","cited_by_count":192,"area":"alignment"},{"id":"2310.10505","title":"ReMax: A Simple, Effective, and Efficient Reinforcement Learning Method for Aligning Large Language Models","created":"2023-10-16","cited_by_count":187,"area":"alignment"},{"id":"2404.03715","title":"Direct Nash Optimization: Teaching Language Models to Self-Improve with General Preferences","created":"2024-04-04","cited_by_count":178,"area":"alignment"},{"id":"2402.10207","title":"Rewards-in-Context: Multi-objective Alignment of Foundation Models with Dynamic Preference Adjustment","created":"2024-02-15","cited_by_count":175,"area":"alignment"},{"id":"2309.16240","title":"Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints","created":"2023-09-28","cited_by_count":174,"area":"alignment"},{"id":"2311.05553","title":"Removing RLHF Protections in GPT-4 via Fine-Tuning","created":"2023-11-09","cited_by_count":171,"area":"alignment"},{"id":"2401.06080","title":"Secrets of RLHF in Large Language Models Part II: Reward Modeling","created":"2024-01-11","cited_by_count":171,"area":"alignment"},{"id":"2309.11489","title":"Text2Reward: Reward Shaping with Language Models for Reinforcement Learning","created":"2023-09-20","cited_by_count":160,"area":"alignment"},{"id":"2309.07124","title":"RAIN: Your Language Models Can Align Themselves without Finetuning","created":"2023-09-13","cited_by_count":157,"area":"alignment"},{"id":"2408.10075","title":"Personalizing Reinforcement Learning from Human Feedback with Variational Preference Learning","created":"2024-08-19","cited_by_count":155,"area":"alignment"},{"id":"2506.01937","title":"RewardBench 2: Advancing Reward Model Evaluation","created":"2025-06-02","cited_by_count":150,"area":"alignment"},{"id":"2403.16950","title":"Aligning with Human Judgement: The Role of Pairwise Preference in Large Language Model Evaluators","created":"2024-03-25","cited_by_count":149,"area":"alignment"},{"id":"2402.18571","title":"Arithmetic Control of LLMs for Diverse User Preferences: Directional Preference Alignment with Multi-Objective Rewards","created":"2024-02-28","cited_by_count":148,"area":"alignment"},{"id":"2309.10105","title":"Understanding Catastrophic Forgetting in Language Models via Implicit Inference","created":"2023-09-18","cited_by_count":144,"area":"alignment"},{"id":"2404.16019","title":"The PRISM Alignment Dataset: What Participatory, Representative and Individualised Human Feedback Reveals About the Subjective and Multicultural Alignment of Large Language Models","created":"2024-04-24","cited_by_count":144,"area":"alignment"},{"id":"2403.17710","title":"Optimization-based Prompt Injection Attack to LLM-as-a-Judge","created":"2024-03-26","cited_by_count":143,"area":"alignment"},{"id":"2406.10216","title":"Regularizing Hidden States Enables Learning Generalizable Reward Model for LLMs","created":"2024-06-14","cited_by_count":142,"area":"alignment"},{"id":"2402.07319","title":"ODIN: Disentangled Reward Mitigates Hacking in RLHF","created":"2024-02-11","cited_by_count":136,"area":"alignment"},{"id":"2404.11999","title":"Token-level Direct Preference Optimization","created":"2024-04-18","cited_by_count":135,"area":"alignment"},{"id":"2408.17003","title":"Safety Layers in Aligned Large Language Models: The Key to LLM Security","created":"2024-08-30","cited_by_count":134,"area":"alignment"},{"id":"2311.14455","title":"Universal Jailbreak Backdoors from Poisoned Human Feedback","created":"2023-11-24","cited_by_count":128,"area":"alignment"},{"id":"2404.18922","title":"DPO Meets PPO: Reinforced Token Optimization for RLHF","created":"2024-04-29","cited_by_count":127,"area":"alignment"},{"id":"2409.12822","title":"Language Models Learn to Mislead Humans via RLHF","created":"2024-09-19","cited_by_count":124,"area":"alignment"},{"id":"2311.04155","title":"Black-Box Prompt Optimization: Aligning Large Language Models without Model Training","created":"2023-11-07","cited_by_count":122,"area":"alignment"},{"id":"2402.10571","title":"Direct Preference Optimization with an Offset","created":"2024-02-16","cited_by_count":122,"area":"alignment"},{"id":"2406.00832","title":"BoNBoN Alignment for Large Language Models and the Sweetness of Best-of-n Sampling","created":"2024-06-02","cited_by_count":121,"area":"alignment"},{"id":"2406.00045","title":"Personalized Steering of Large Language Models: Versatile Steering Vectors Through Bi-directional Preference Optimization","created":"2024-05-28","cited_by_count":118,"area":"alignment"},{"id":"2406.15951","title":"Modular Pluralism: Pluralistic Alignment via Multi-LLM Collaboration","created":"2024-06-22","cited_by_count":113,"area":"alignment"},{"id":"2410.09724","title":"Taming Overconfidence in LLMs: Reward Calibration in RLHF","created":"2024-10-13","cited_by_count":112,"area":"alignment"},{"id":"2310.05344","title":"SteerLM: Attribute Conditioned SFT as an (User-Steerable) Alternative to RLHF","created":"2023-10-09","cited_by_count":111,"area":"alignment"},{"id":"2310.15140","title":"AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models","created":"2023-10-23","cited_by_count":111,"area":"alignment"},{"id":"2312.16682","title":"Some things are more CRINGE than others: Iterative Preference Optimization with the Pairwise Cringe Loss","created":"2023-12-27","cited_by_count":108,"area":"alignment"},{"id":"2402.14740","title":"Back to Basics: Revisiting REINFORCE Style Optimization for Learning from Human Feedback in LLMs","created":"2024-02-22","cited_by_count":108,"area":"alignment"},{"id":"2401.18058","title":"LongAlign: A Recipe for Long Context Alignment of Large Language Models","created":"2024-01-31","cited_by_count":106,"area":"alignment"},{"id":"2402.01694","title":"ARGS: Alignment as Reward-Guided Search","created":"2024-01-23","cited_by_count":103,"area":"alignment"},{"id":"2402.02416","title":"Aligner: Efficient Alignment by Learning to Correct","created":"2024-02-04","cited_by_count":98,"area":"alignment"},{"id":"2406.05644","title":"How Alignment and Jailbreak Work: Explain LLM Safety through Intermediate Hidden States","created":"2024-06-09","cited_by_count":98,"area":"alignment"},{"id":"2503.21878","title":"Is Best-of-N the Best of Them? Coverage, Scaling, and Optimality in Inference-Time Alignment","created":"2025-03-27","cited_by_count":97,"area":"alignment"},{"id":"2310.04373","title":"Confronting Reward Model Overoptimization with Constrained RLHF","created":"2023-10-06","cited_by_count":96,"area":"alignment"},{"id":"2403.09472","title":"Easy-to-Hard Generalization: Scalable Alignment Beyond Human Supervision","created":"2024-03-14","cited_by_count":96,"area":"alignment"},{"id":"2402.09154","title":"Attacking Large Language Models with Projected Gradient Descent","created":"2024-02-14","cited_by_count":94,"area":"alignment"},{"id":"2402.09063","title":"Soft Prompt Threats: Attacking Safety Alignment and Unlearning in Open-Source LLMs through the Embedding Space","created":"2024-02-14","cited_by_count":91,"area":"alignment"},{"id":"2403.08635","title":"Human Alignment of Large Language Models through Online Preference Optimisation","created":"2024-03-13","cited_by_count":91,"area":"alignment"},{"id":"2407.08639","title":"$\\beta$-DPO: Direct Preference Optimization with Dynamic $\\beta$","created":"2024-07-11","cited_by_count":90,"area":"alignment"},{"id":"2403.05612","title":"Unfamiliar Finetuning Examples Control How Language Models Hallucinate","created":"2024-03-08","cited_by_count":87,"area":"alignment"},{"id":"2402.04833","title":"Long Is More for Alignment: A Simple but Tough-to-Beat Baseline for Instruction Fine-Tuning","created":"2024-02-07","cited_by_count":85,"area":"alignment"},{"id":"2405.17977","title":"Aligning to Thousands of Preferences via System Message Generalization","created":"2024-05-28","cited_by_count":84,"area":"alignment"},{"id":"2309.16349","title":"Human Feedback is not Gold Standard","created":"2023-09-28","cited_by_count":83,"area":"alignment"},{"id":"2402.02030","title":"Panacea: Pareto Alignment via Preference Adaptation for LLMs","created":"2024-02-03","cited_by_count":80,"area":"alignment"},{"id":"2506.11902","title":"TreeRL: LLM Reinforcement Learning with On-Policy Tree Search","created":"2025-06-13","cited_by_count":80,"area":"alignment"},{"id":"2310.13639","title":"Contrastive Preference Learning: Learning from Human Feedback without RL","created":"2023-10-20","cited_by_count":79,"area":"alignment"},{"id":"2402.09345","title":"InfoRM: Mitigating Reward Hacking in RLHF via Information-Theoretic Reward Modeling","created":"2024-02-14","cited_by_count":79,"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":"2407.14622","title":"BOND: Aligning LLMs with Best-of-N Distillation","created":"2024-07-19","cited_by_count":76,"area":"alignment"},{"id":"2309.12053","title":"AceGPT, Localizing Large Language Models in Arabic","created":"2023-09-21","cited_by_count":75,"area":"alignment"},{"id":"2312.07000","title":"Alignment for Honesty","created":"2023-12-12","cited_by_count":74,"area":"alignment"},{"id":"2402.16192","title":"Defending Large Language Models against Jailbreak Attacks via Semantic Smoothing","created":"2024-02-25","cited_by_count":74,"area":"alignment"},{"id":"2405.19320","title":"Value-Incentivized Preference Optimization: A Unified Approach to Online and Offline RLHF","created":"2024-05-29","cited_by_count":73,"area":"alignment"},{"id":"2408.13006","title":"Systematic Evaluation of LLM-as-a-Judge in LLM Alignment Tasks: Explainable Metrics and Diverse Prompt Templates","created":"2024-08-23","cited_by_count":72,"area":"alignment"},{"id":"2404.16767","title":"REBEL: Reinforcement Learning via Regressing Relative Rewards","created":"2024-04-25","cited_by_count":69,"area":"alignment"},{"id":"2402.02992","title":"Decoding-time Realignment of Language Models","created":"2024-02-05","cited_by_count":67,"area":"alignment"},{"id":"2404.12038","title":"Uncovering Safety Risks of Large Language Models through Concept Activation Vector","created":"2024-04-18","cited_by_count":67,"area":"alignment"},{"id":"2402.15018","title":"Unintended Impacts of LLM Alignment on Global Representation","created":"2024-02-22","cited_by_count":66,"area":"alignment"}],"next_cursor":"100"}