This work presents CoworkEnv, a capability-guided framework for synthesizing executable training environments for AI coworkers, and compares RL training dynamics across configurations that vary reward processing and credit assignment.
Zhong-Wei Xie, Yi-Ming Du, Jiaxin Bai et al.· 0 citations
Full-pipeline data analysis over real-world data lakes requires LLM agents to locate and integrate information across large collections of tables with different schemas. Existing data agents follow task-specific exploration, which exhausts the context window in massive lakes. As the context fills, the agent's reasoning...
Bo-Yi Zhang, Hong Ting Tsang, Yi-Sen Gao et al.· 0 citations
Omni-modal assistants must remember not only what users say, but also what they show and convey. Existing long-term memory benchmarks are mostly text-and-image or report a single end-to-end score, so they cannot locate whether evidence is lost when it is represented, retrieved, delivered, or read. We introduce MOSAIC,...
Hong Ting Tsang, Bo-Yi Zhang, Haoyu Huang et al.· 0 citations
Repeated event forecasting requires extracting predictive signals from noisy evidence and evaluating their value across successive decisions. We introduce JockeyBench, a benchmark and evaluation protocol that studies this problem through historical Japanese horse racing. The protocol combines time-gated evidence, price...
Zhong-Wei Xie, Xing-Su Li, Jing Yu et al.· 0 citations
Prefix reuse improves LLM inference by sharing the KV cache of common prompt prefixes. This helps only if the scheduler can execute a matched branch in the current worker round. Schedulers assume that a matched prefix is ready, which holds for full attention but fails for hybrid LLMs. There, a branch (the consumer) may...
Hui-Hao Jing, Wen-Bin Hu, Shao-Jin Chen et al.· 0 citations
A benchmark that contrasts authorization evidence for the same operation in its language track is introduced, which combines expert-annotated natural-language scenarios with executed and verified operations, plus externally sourced native-policy requests.
Hui-Hao Jing, Shao-Jin Chen, Wen-Bin Hu et al.· 0 citations
Long-horizon agents must complete workflows in which later activities depend on earlier outputs and changes to shared state. A key challenge for evaluation is to expand task requirements and their dependencies while keeping tasks executable and outcomes verifiable. We introduce SkillHorizon, a benchmark that constructs...
Yuxuan Liu, Zhao-Chen Su, Hang-Yu Guo et al.· 0 citations
This survey focuses on methods that augment LLMs with such parametric memory at deployment: a memory-bearing parameter object is plugged into the forward pass during inference, whether it is acquired before or during deployment.
Haoyu Huang, Zhong-Wei Xie, Jiaxin Bai et al.· 1 citation
This work proposes soft-target fine-tuning (SoFT) to balance learning from teacher demonstrations with retaining the Base model's existing capabilities, with improvements in both in-distribution capability acquisition and out-of-distribution generalization.
Evaluated across three main benchmarks, two domain-specific studies, and six LLMs, SkillRevise substantially outperforms one-shot baselines, and the revised skills transfer across both executors and task environments, suggesting that SkillRevise captures reusable procedural knowledge beyond any single executor.
Yuxuan Liu, Zhao-Chen Su, Lin Xie et al.· arXiv.org· 17 citations
Overall, persistent skill self-evolution is better understood as sparse, validation-filtered search with model- and benchmark-dependent returns, rather than steady improvement from additional rounds.
Yuxuan Liu, Zhaochen Su, Yuhao Zhang et al.· 2 citations
This survey treats isolation as a first-class principle for LLM-agent system safety, and organizes the literature with a boundary-centric taxonomy of five boundaries: user-agent, agent-tool, agent-execution, agent-agent, and system-environment.