15 papers
OrthoSkillVLA: Continual Skill Learning via Gradient-Informed Skill Subspace Adaptation
Jiaqi Wang, Zhou Fang, Qiongfeng Shi +1
Pretrained Vision-Language-Action models provide a strong foundation for robot learning, but sequentially adapting them to diverse skills can perturb the representations and veloci…
FailForge: Distilling Procedural Competence from Persistent Failures into Code Agents
Dongyi Lv, Fushun E, Aichen Cai +8
Rejection sampling fine-tuning (RFT) is widely used to train code agents by generating trajectories on verifiable software engineering tasks, retaining those that pass the tests, a…
SearchAuditor: Auditing and Attributing Failures in Long-Horizon Search Agents
Zhixiang Liang, Yifei Liu, Yidan Huang +5
Deep search agents tackle challenging questions through long-horizon web interactions, a process that is both complex and fragile: small reasoning errors may propagate through long…
SearchMaster: Grounded and Regulated Self-Play for Search Agents
Wentao Tan, Qiong Cao, Jiaqi Wang +1
Training LLM-based search agents requires high-quality search data: tasks that demand genuine multi-hop retrieval and trajectories that use search tools effectively. Existing pipel…
DOPD: Dual On-policy Distillation
Xinlei Yu, Gen Li, Qingyi Si +13
On-policy distillation (OPD) offers superior capacity transfer by supervising student-sampled trajectories with dense token-level signals. To furnish high-quality supervision sourc…
Leveraging Error Diversity in Group Rollouts for Reinforcement Learning
Wenpu Liu, Yuqi Xu, Weichu Xie +8
Reinforcement Learning from Verifiable Rewards (RLVR) typically samples multiple responses per prompt and assigns binary rewards based on individual correctness, yet the collective…