arXiv:2609.27532v1 Announce Type: new
Abstract: Long-horizon agentic tasks require an agent to modify an environment through a sequence of tool calls, with success determined by the final state. The...
By Ming Ma, Yi Zhu, Yiran Zhong, Feida Zhu, Chonghan Liu, Pengkun Jiao, Qichao Wang, Yanhao Jia, Tianming Yang, Steven Hoi
The paper investigates how fine‑grained credit assignment affects learning in multimodal agentic reinforcement learning, using video editing as a testbed. Four movie‑trailer editing tasks—Content Selection, Narrative Sequencing, Rough Cut, and Fine Cut—are introduced to represent real‑world editing workflows with varying complexity and horizon lengths. The study finds that decaying process rewards significantly boost performance on the long‑horizon Fine Cut task, while offering modest gains on shorter tasks; however, fixed process‑reward weighting can lead to reward‑hacking and degraded outcomes, and on‑policy self‑distillation provides limited benefit.
By Lei Zhao, Zongheng Cao, Xinyu Hu
The paper introduces TASPO, a method that transforms privileged information (PI) into outcome‑grounded action credit for language‑model agents. TASPO constructs decision‑applicable PI from verified successful experience, aggregates PI‑induced likelihood shifts at the executable‑action level, and converts relative action support into positive, bounded, mean‑preserving weights on the original trajectory advantage. Experiments on three agentic benchmarks show TASPO improves over GRPO by 10.6% and generalizes better to unseen tasks, while reducing supervision mismatch and stabilizing policy optimization.
By Jingxiao Yang, Wangjie Gan, Yingxuan Zhuang, Wenqi Zhang, Jintao Chen, Xuhong Zhang
The paper investigates how small action errors evolve when using action chunking in behavioural cloning. By injecting errors at each state and observing their growth under open‑loop (no replanning) and closed‑loop (replanning) regimes, the authors classify states as contracting, expanding, or unresolved. Across twelve manipulation tasks, they find that stable states are rare, error amplification is common, and that short‑horizon fitting can overestimate long‑horizon propagation. Predictors trained on camera and proprioceptive data can recover open‑loop stability but only partially capture closed‑loop dynamics, indicating that standard imitation learning does not reliably produce policies that contract errors when perturbed.
By Aryan Goyal
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
arXiv:2607.09042v2 Announce Type: replace
Abstract: Reinforcement learning is increasingly used to fine-tune vision-language-action (VLA) models, but robot interaction is expensive and learning becom...
By Iris Xu, Sunshine Jiang, John Marangola, Pulkit Agrawal, Zhang-Wei Hong
The paper introduces VICT, a method that leverages the internal structure of verifiable tasks to perform fine‑grained credit assignment for long‑horizon LLM agents. VICT exposes executable or evidence‑backed atoms from a task’s terminal verifier and traces them back to actions via dependency‑valid proof edges, redistributing advantage only along these edges. This approach improves performance on ALFWorld and WebShop compared to outcome‑only training and matches recent fine‑grained credit methods without requiring additional critics, labels, or inference‑time verifier access.
By Pengcheng Li, Zhengyang Zhang, Dongxu Zhang, Sui Huang, Shaohua Ma
arXiv:2610.16443v1 Announce Type: new
Abstract: Recent world action models draw on video generation priors to jointly predict future observations and robot actions. Adapting these priors to control r...
By Yuanpeng He, Ceyao Zhang, Fangjing Li, Kefei Zhu, Lijian Li, Yueheng Li, Tianjia He, Wenpin Jiao, Zhi Jin, Yaodong Yang, Yuanpei Chen
arXiv:2609.35767v2 Announce Type: replace-cross
Abstract: Unified multimodal models can both look at and render images, so in principle they can repair their own generations: diagnose what an image g...
By Yijia Fan, Ziqi Huang, Zhongang Cai, Yan Li, Zimo Wen, Wanqi Yin, Haiwen Diao, Ziwei Liu
WorldGuide is a closed‑loop video world model that, given an initial image and a task goal, predicts an atomic action, generates the corresponding video clip, and uses that clip to decide the next action or terminate the task. It trains a Planner to predict the next action or completion from visual progress and an Executor to realize those actions, while a hierarchical visual memory tracks state over long horizons. The authors introduce WorldGuide‑Bench, a dataset of ~59K step‑annotated videos across 245 tasks, and show that WorldGuide outperforms the recent MiniMax‑H3 model on both WorldGuide‑Bench and VideoCraft‑Bench, highlighting the benefit of coupling planning with learned execution for goal‑directed procedural video generation.
By Ankan Deria, Komal Kumar, Hisham Cholakkal, Fahad Shahbaz Khan, Salman Khan
arXiv:2607. 16506v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) policies offer strong general-purpose manipulation priors, but often fail on tight-tolerance, contact-rich assembly due to long-horizon credit assignment and subtask coupling: a state that is geometrically successful for the current skill can be brittle for downstream skills.
By Yuhan Liu, Xinyu Zhang, Litao Liu, Abdeslam Boularias
The paper introduces Trace, a framework that transforms sparse-reward trajectories into executable walkthroughs by identifying progress anchors, propagating credit, and estimating action prerequisites. Trace compiles noisy trajectories into state‑conditioned, verifiable procedures that remove loops and detours, enabling reuse, intermediate‑state resumption, and programmatic verification. Experiments on J‑TTL, WebShop, and ScienceWorld with three open‑source LLMs show that Trace outperforms eight baselines, improving average AUC and Final‑$3$ by 30.0% and 40.5% while using fewer inference tokens.
By Kaijie Chen, Chenyu Fang, Liang Yan, Bo Li, Bo Zhang, Peng Ye