{"total":56,"rows":[{"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":"2309.06256","title":"Mitigating the Alignment Tax of RLHF","created":"2023-09-12","cited_by_count":183,"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":"2402.01109","title":"Vaccine: Perturbation-aware Alignment for Large Language Models against Harmful Fine-tuning Attack","created":"2024-02-02","cited_by_count":115,"area":"alignment"},{"id":"2405.18641","title":"Lisa: Lazy Safety Alignment for Large Language Models against Harmful Fine-tuning Attack","created":"2024-05-28","cited_by_count":92,"area":"alignment"},{"id":"2310.14558","title":"AlpaCare:Instruction-tuned Large Language Models for Medical Application","created":"2023-10-23","cited_by_count":90,"area":"alignment"},{"id":"2310.06762","title":"TRACE: A Comprehensive Benchmark for Continual Learning in Large Language Models","created":"2023-10-10","cited_by_count":66,"area":"alignment"},{"id":"2406.16768","title":"WARP: On the Benefits of Weight Averaged Rewarded Policies","created":"2024-06-24","cited_by_count":45,"area":"alignment"},{"id":"2406.05534","title":"Online DPO: Online Direct Preference Optimization with Fast-Slow Chasing","created":"2024-06-08","cited_by_count":39,"area":"alignment"},{"id":"2312.07551","title":"Language Model Alignment with Elastic Reset","created":"2023-12-06","cited_by_count":34,"area":"alignment"},{"id":"2406.11801","title":"Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations","created":"2024-06-17","cited_by_count":32,"area":"alignment"},{"id":"2405.17931","title":"Online Merging Optimizers for Boosting Rewards and Mitigating Tax in Alignment","created":"2024-05-28","cited_by_count":28,"area":"alignment"},{"id":"2412.11041","title":"Separate the Wheat from the Chaff: A Post-Hoc Approach to Safety Re-Alignment for Fine-Tuned Language Models","created":"2024-12-15","cited_by_count":20,"area":"alignment"},{"id":"2405.13432","title":"Disperse-Then-Merge: Pushing the Limits of Instruction Tuning via Alignment Tax Reduction","created":"2024-05-22","cited_by_count":18,"area":"alignment"},{"id":"2208.14165","title":"Towards Boosting the Open-Domain Chatbot with Human Feedback","created":"2022-08-30","cited_by_count":11,"area":"alignment"},{"id":"2512.10150","title":"Unforgotten Safety: Preserving Safety Alignment of Large Language Models with Continual Learning","created":"2025-12-10","cited_by_count":10,"area":"alignment"},{"id":"2402.14228","title":"COPR: Continual Human Preference Learning via Optimal Policy Regularization","created":"2024-02-22","cited_by_count":9,"area":"alignment"},{"id":"2407.17482","title":"Reinforcement Learning from Human Feedback: Whose Culture, Whose Values, Whose Perspectives?","created":"2024-07-02","cited_by_count":8,"area":"alignment"},{"id":"2406.15178","title":"Hybrid Alignment Training for Large Language Models","created":"2024-06-21","cited_by_count":7,"area":"alignment"},{"id":"2410.11370","title":"Enhance Graph Alignment for Large Language Models","created":"2024-10-15","cited_by_count":7,"area":"alignment"},{"id":"2510.18053","title":"Adaptive Divergence Regularized Policy Optimization for Fine-tuning Generative Models","created":"2025-10-20","cited_by_count":6,"area":"alignment"},{"id":"2602.07892","title":"Safety Alignment as Continual Learning: Mitigating the Alignment Tax via Orthogonal Gradient Projection","created":"2026-02-08","cited_by_count":6,"area":"alignment"},{"id":"2503.22948","title":"SUV: Scalable Large Language Model Copyright Compliance with Regularized Selective Unlearning","created":"2025-03-29","cited_by_count":5,"area":"alignment"},{"id":"2511.18039","title":"Curvature-Aware Safety Restoration In LLMs Fine-Tuning","created":"2025-11-22","cited_by_count":5,"area":"alignment"},{"id":"2602.15823","title":"CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing","created":"2026-02-17","cited_by_count":5,"area":"alignment"},{"id":"2602.11157","title":"Response-Based Knowledge Distillation for Multilingual Jailbreak Prevention Unwittingly Compromises Safety","created":"2025-12-08","cited_by_count":4,"area":"alignment"},{"id":"2604.17215","title":"Continual Safety Alignment via Gradient-Based Sample Selection","created":"2026-04-19","cited_by_count":4,"area":"alignment"},{"id":"2412.02125","title":"Preference Goal Tuning: Post-Training as Latent Control for Frozen Policies","created":"2024-12-03","cited_by_count":3,"area":"alignment"},{"id":"2502.00136","title":"A Checks-and-Balances Framework for Context-Aware Ethical AI Alignment","created":"2025-01-31","cited_by_count":3,"area":"alignment"},{"id":"2505.09925","title":"Reinforced Interactive Continual Learning via Real-time Noisy Human Feedback","created":"2025-05-15","cited_by_count":3,"area":"alignment"},{"id":"2602.08813","title":"Robust Policy Optimization to Prevent Catastrophic Forgetting","created":"2026-02-09","cited_by_count":3,"area":"alignment"},{"id":"2310.15694","title":"COPR: Continual Learning Human Preference through Optimal Policy Regularization","created":"2023-10-24","cited_by_count":2,"area":"alignment"},{"id":"2410.00508","title":"FlipGuard: Defending Preference Alignment against Update Regression with Constrained Optimization","created":"2024-10-01","cited_by_count":2,"area":"alignment"},{"id":"2505.20359","title":"Risk-aware Direct Preference Optimization under Nested Risk Measure","created":"2025-05-26","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":"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":"2511.21101","title":"Mortgage Language Model: Domain-Adaptive Pretraining with Residual Instruction, Alignment Tuning, and Task-Specific Routing","created":"2025-11-26","cited_by_count":1,"area":"alignment"},{"id":"2601.05858","title":"CLewR: Curriculum Learning with Restarts for Machine Translation Preference Learning","created":"2026-01-09","cited_by_count":1,"area":"alignment"},{"id":"2602.03073","title":"TMS: Trajectory-Mixed Supervision for Reward-Free, On-Policy SFT","created":"2026-02-03","cited_by_count":1,"area":"alignment"},{"id":"2606.21089","title":"Repeated post-training is not Self-improving: Diagnosing Scientific Amnesia in Continual DPO Pipelines","created":"2026-06-17","cited_by_count":1,"area":"alignment"},{"id":"2607.07023","title":"Online Data Selection Is Implicit Alignment","created":"2026-07-08","cited_by_count":1,"area":"alignment"},{"id":"2601.09718","title":"StatLLaMA: Multi-Stage training for domain-optimized statistical large language models","created":"2025-12-26","cited_by_count":0,"area":"alignment"},{"id":"2601.17260","title":"The Viscosity of Logic: Phase Transitions and Hysteresis in DPO Alignment","created":"2026-01-24","cited_by_count":0,"area":"alignment"},{"id":"2603.07784","title":"ProgAgent:A Continual RL Agent with Progress-Aware Rewards","created":"2026-03-08","cited_by_count":0,"area":"alignment"},{"id":"2604.04410","title":"Relative Density Ratio Optimization for Stable and Statistically Consistent Model Alignment","created":"2026-04-06","cited_by_count":0,"area":"alignment"},{"id":"2604.17691","title":"SafeAnchor: Preventing Cumulative Safety Erosion in Continual Domain Adaptation of Large Language Models","created":"2026-04-20","cited_by_count":0,"area":"alignment"},{"id":"2605.05995","title":"Safety Anchor: Defending Harmful Fine-tuning via Geometric Bottlenecks","created":"2026-05-07","cited_by_count":0,"area":"alignment"},{"id":"2605.25629","title":"When In-Distribution Gains Fail: Evaluating Weak-to-Strong Reward Models under Preference Shift","created":"2026-05-25","cited_by_count":0,"area":"alignment"},{"id":"2606.09866","title":"Two to Tango: Coupled Task-Reference Selection for Safe LLM Fine-tuning","created":"2026-06-01","cited_by_count":0,"area":"alignment"},{"id":"2606.28998","title":"Reward-Free Code Alignment from Pretrained or Fine-Tuned LLM: Unpacking the Trade-offs for Code Generation","created":"2026-06-27","cited_by_count":0,"area":"alignment"},{"id":"2607.16242","title":"TRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment","created":"2026-06-26","cited_by_count":0,"area":"alignment"},{"id":"2608.24192","title":"Preference Data Selection for Mitigating the Alignment Tax in Large Language Models","created":"2026-08-25","cited_by_count":0,"area":"alignment"},{"id":"2608.27910","title":"AI Alignment through a Game-theoretic Lens: A Survey","created":"2026-08-28","cited_by_count":0,"area":"alignment"},{"id":"2609.06934","title":"The Geometry of Refusal: Why Post-Hoc Safety Is Fragile and Pretraining-Time Safety Persists","created":"2026-09-07","cited_by_count":0,"area":"alignment"},{"id":"2609.36748","title":"Generalizable Lifelong Model Editing via Preference Optimization","created":"2026-09-29","cited_by_count":0,"area":"alignment"},{"id":"2610.02844","title":"DNAlign: Dynamic Null-Space Safe Alignment for LLMs","created":"2026-10-02","cited_by_count":0,"area":"alignment"}],"next_cursor":null}