SOURCE-LINKED INTELLIGENCE
TSBPO: Tree-Structured Branching for Policy Optimization in Interactive Language Agents
Training interactive language agents from terminal outcomes poses three coupled problems: which task and state continuations to sample, how to assign delayed feedback to decisions, and how to aggregate token losses for multi-token actions. We present TSBPO, a tree-structured sampling-to-update paradigm that addresses these problems within one training loop. A likelihood-fitted Root/Branch allocator uses retained task and state evidence to target continuations likely to produce contrasting terminal outcomes under fixed rollout budgets. On the resulting replayed physical trees, trajectory advant
Read original source ↗ Open in workspace
- recordType
- paper
- region
- Global
Evidence & attribution
- arXiv · AI, language, vision and robotics · 2026-10-30T13:04:05.000Z
First collected: 2026-11-02T10:21:21.339Z. This is not the publication date.