… Finally, we optimize ToolCUA with Online Agentic RL in a high-fidelity GUI-Tool environment, guided by a Tool-Efficient Path Reward that encourages appropriate tool use and shorter execution paths. …
…Instead, it treats an agent as an AgentProcess : a schedulable execution subject with process identity, parent-child lineage, lifecycle state, a tool table derived from an AgentImage , typed Object Memory , explicit capabilities…
Papers arxiv:2604.25135 FAMA: Failure-Aware Meta-Agentic Framework for Open-Source LLMs in Interactive Tool Use Environments Published on Apr 28 Submitted by Amir on Apr 30 Authors: Amir Saeidi , , , , , , Abstract Failure-Aware Meta-Agentic framework improves open-source LLM performance in convers… …
…Moreover, the standard approach, in which scenarios are first written in natural language and then mapped to tool sequences, captures only a narrow subset of the tool-use patterns agents exercise. In…
…Evaluation of LLM Agents in Dynamic, Interdependent, and Large-Scale Tool Sandbox (2026) EnvFactory: Scaling Tool-Use Agents via Executable Environments Synthesis and Robust RL (2026) UniToolCall: Unifying Tool-Use Representation, Data…
Papers arxiv:2605.09252 LLM Agents Already Know When to Call Tools -- Even Without Reasoning Published on May 10 Submitted by Chung-En, Sun on May 13 Authors: Chung-En Sun , , , , Abstract…
…AI-generated summary Multimodal deep search requires an agent to solve open-world problems by chaining search, tool use, and visual reasoning over evolving textual and visual context. Two bottlenecks limit current…
… We release the tasks, code, and containerized tooling to support reproducible evaluation. …
… The following papers were recommended by the Semantic Scholar API Small Model as Master Orchestrator: Learning Unified Agent-Tool Orchestration with Parallel Subtask Decomposition 2026 Utility-Guided Agent Orchestration for Efficient LLM Tool Use 2026 SWE-Shepherd: Advancing PRMs for Reinforcing Co… …
… View arXiv page View PDF GitHub 2.32k Add to collection Community We present GLM-5V-Turbo, a native multimodal foundation model for agentic tasks in real digital environments. …