Lune

学者档案2024-2026

Yankai Lin

Renmin University of China

13论文数
7末位作者
15合作者人数
0一作
0独著

一作与末位作者按论文署名顺序统计,部分顶会采用姓氏字母顺序署名。

近期论文13

  1. LaSeR: Reinforcement Learning with Last-Token Self-Rewarding第 7 位,共 7 位作者ICLR 2026 · 12 次引用
  2. CURE: Critique-Driven Unified Reinforcement Learning for Test-Time Self-Improvement第 6 位,共 6 位作者ACL 2026
  3. Less Noise, More Voice: Reinforcement Learning for Reasoning via Instruction Purification第 3 位,共 3 位作者ACL Findings 2026
  4. Scaling Spatial Reasoning in MLLMs through Programmatic Data Synthesis第 12 位,共 13 位作者CVPR Findings 2026
  5. Towards Preference Following in Tool Calling Language Agents第 5 位,共 5 位作者ACL Findings 2026
  6. Towards Thinking-Optimal Scaling of Test-Time Compute for LLM Reasoning第 3 位,共 4 位作者NeurIPS 2025 · 141 次引用
  7. Learning to Focus: Causal Attention Distillation via Gradient-Guided Token Pruning第 6 位,共 6 位作者NeurIPS 2025 · 14 次引用
  8. Beyond the Surface: Measuring Self-Preference in LLM Judgments第 5 位,共 5 位作者EMNLP 2025
  9. Rational Decision-Making Agent with Learning Internal Utility Judgment第 6 位,共 8 位作者ICLR 2025
  10. Super(ficial)-alignment: Strong Models May Deceive Weak Models in Weak-to-Strong Generalization第 7 位,共 8 位作者ICLR 2025
  11. Towards Codable Watermarking for Injecting Multi-Bits Information to LLMs第 5 位,共 8 位作者ICLR 2024 · 55 次引用
  12. Large Language Model-based Human-Agent Collaboration for Complex Task Solving第 4 位,共 7 位作者EMNLP Findings 2024 · 24 次引用
  13. Towards Tool Use Alignment of Large Language Models第 6 位,共 6 位作者EMNLP 2024 · 5 次引用
黄昏的海面,两侧是细线勾勒的悬崖