The paper introduces Actor‑Critic with Action Chunking (AC2), a method that assigns credit to short action chunks instead of entire trajectories, enabling policy updates without waiting for terminal rewards. AC2 employs local readiness, reference solutions, and 10k‑token chunks to make critic‑based credit assignment reliable. Experiments on Qwen3‑4B with FineProofs‑RL show AC2 surpasses GRPO’s peak validation score while using 2.5× fewer decoding FLOPs and fewer training steps.
By Kaiyue Wen, Luke Bailey, Arvind Mahankali, Tengyu Ma
The paper introduces EDV, an Execute‑Distill‑Verify framework that constructs offline experience for large language model agents without updating model parameters. In EDV, multiple heterogeneous agents execute tasks, a distiller extracts candidate lessons from their trajectories and feedback, and executors review these lessons, keeping only those unanimously approved. Experiments on τ²‑bench, Mind2Web, and MMTB demonstrate improved performance on revisited tasks and within‑domain transfer, while a Retail subset evaluation shows EDV cuts the Self‑Confirmation Trap Rate from 11.0% to 5.0%.
By Shiding Zhu, Yudi Qi, Yajie Wang, Jiaze Li, Chao Song, Yaorui Shi, Yibo Miao, Hanqi Gao, Kai Zhang
arXiv:2610.41564v1 Announce Type: new
Abstract: Language agents increasingly tackle long-horizon tasks in interactive environments, where actions often depend on latent preconditions. The core diffic...
By Mingyang Liu, Qingcan Kang, Shixiong Kai, Tao Zhong, Mingxuan Yuan, Xin Li, Linqi Song
arXiv:2609. 04875v1 Announce Type: cross Abstract: Long-running LLM agents are stateful: beyond the transcript they accrete compressed summaries, plaintext memory, pending tool plans, and, under every serving API, a KV cache.
By Chao Yao, Yangbo Wei, Zhen Huang, Junhong Qian, Chenle Chen, Shaoqiang Lu, Chen Wu, Lei He
arXiv:2609.33875v2 Announce Type: replace-cross
Abstract: Outcome-only reinforcement learning gives software engineering (SWE) agents a terminal success signal but little direct guidance about interm...
By Yuanhao Li, Hongbo Wang, Xuhong Chen, Yiming Cao, Xunzhu Tang
The paper introduces EvoResearcher, a training‑free, inference‑time protocol that enables a frozen large language model to perform cost‑bounded self‑reflection and early stopping. By iterating through generate → self‑critique → revise steps until a maximum depth or a CONFIRMED sentinel is reached, the model can self‑verify its answers within a strict compute budget. The protocol incorporates four self‑reflective meta‑reward components—correctness, efficiency, reflection depth, and tool‑call diversity—implemented as prompt‑level mechanisms, and is validated on Big‑Bench Hard, GSM8K, and MATH benchmarks, achieving comparable accuracy while terminating 82‑88% of items early with only about 2.1 generations per question.
By Wei Yu, Suxing Liu, Minjie Yu, Jiahao Wang, Zhijian Zheng, Haocheng Deng, Bing Li
arXiv:2607. 10481v1 Announce Type: cross Abstract: Reinforcement learning (RL) has significantly enhanced the reasoning capabilities of large language models (LLMs), yet the training process remains notoriously fragile.
By Kexin Huang, Junkang Wu, Jinda Lu, Shuo Yang, Chiyu Ma, Jiancan Wu, Xiang Wang, Xiangnan He, Guoyin Wang, Jingren Zhou
StateWise is a method that trains a single large language model to maintain and update its own constraint state during multi‑turn dialogue. By fine‑tuning the model to summarize conversations into a typed state with tombstones for revoked constraints and then self‑training on trajectories that pass exact round‑trip checks, it reduces relapse and improves conversation success. The approach outperforms a compiled‑ledger baseline on relapse without requiring inference‑time scaffolding, and it is evaluated on two backbones with held‑out constraint families and unseen phrasings.
By Samarth Varshney, Ritu Yadav
arXiv:2406.03678v2 Announce Type: replace
Abstract: On-policy reinforcement learning methods, like Trust Region Policy Optimization (TRPO) and Proximal Policy Optimization (PPO), often demand extensi...
By Yaozhong Gan, Renye Yan, Zhe Wu, Junliang Xing
CAST is a critique‑aware training framework that transforms sparse task outcomes into action‑level supervision for both critique learning and policy optimization. By analyzing agent trajectories, CAST synthesizes structured rationales that explain action validity under partial observability, enabling the creation of richer training data. Fine‑tuned Qwen3‑family models trained with CAST show significant reliability gains, outperforming GPT‑OSS‑120B by over 10% on Retail tasks and improving Telehealth performance by 9% in an out‑of‑domain setting.
By Amir Saeidi, Zehua Zhang, Rishitosh Singh, Naman Ahuja, Vivek Gupta, Ali Payani, Gaowen Liu, Jayanth Srinivasa, Chitta Baral
arXiv:2606. 00400v1 Announce Type: new Abstract: Continual instruction tuning updates a language model through a sequence of new domains, yet each update can progressively erode previously learned capabilities and alignment behavior.
By Ibne Farabi Shihab, Fariya Afrin, Anuj Sharma
arXiv:2601. 22448v2 Announce Type: replace Abstract: RLVR has become a standard recipe for training LLMs on reasoning tasks with verifiable outcomes, but when rollout generation dominates the cost, efficiency hinges on which prompts are sampled and when.
By Weiqi Wang, Xin Liu, Binxuan Huang, Hejie Cui, Rongzhi Zhang, Changlong Yu, Shuowei Jin, Jingfeng Yang, Qingyu Yin, Zhengyang Wang, Zheng Li, Yifan Gao, Priyanka Nigam, Bing Yin, Lihong Li, Yangqiu Song