SOURCE-LINKED INTELLIGENCE

TIGPO: Temporal Instance-Graph Policy Optimization for Long-Horizon LLM Agents

arXiv · AI, language, vision and robotics · article · Sep 3, 2026 · UTC

Graph-based policy optimization improves credit assignment for long-horizon LLM agents by organizing rollout trajectories into state-transition graphs. However, existing methods construct graphs independently within each policy update, discarding transitions discovered by earlier policies and limiting advantage estimation to small, batch-local rollout groups. We propose \emph{Temporal Instance-Graph Policy Optimization} (TIGPO), which extends graph-based credit assignment across policy updates. TIGPO maintains a persistent transition graph for each task, allowing valid transitions discovered b

Read original source ↗ Open in workspace

recordType

paper

region

Global