Search papers, labs, and topics across Lattice.
The paper introduces TreeSeeker, an inference-time framework that enhances deep search by organizing it as a branch-and-return process over tree-structured states. This method addresses the challenge of navigating multiple plausible search directions by employing a controlled trial-and-error approach that balances exploitation and exploration based on textual Upper Confidence Bound (UCB) signals. Experimental results on multiple benchmarks demonstrate that TreeSeeker significantly outperforms existing open-source baselines, highlighting the effectiveness of structured control in improving search outcomes.
TreeSeeker's branch-and-return mechanism enables agents to navigate complex search spaces more effectively, outperforming traditional methods by leveraging structured uncertainty management.
Deep search requires agents to answer complex questions through multi-step web search, browsing, evidence comparison, and synthesis. A central challenge is deciding how to search when several directions look plausible but only some will later lead to reliable evidence. If an agent greedily follows the current best-looking direction, it may keep extending a weak continuation. If it explores without discipline, it may waste budget on disconnected trials. We propose TreeSeeker, an inference-time framework for controlled trial-and-error in deep search. TreeSeeker organizes search as branch-and-return search over tree-structured states, where each branch is a tentative direction for a sub-goal. At each round, TreeSearch reads all sub-goal trees, identifies active goals, and uses textual UCB signals of value, uncertainty, and risk to select among exploiting a promising branch, exploring an uncertain alternative, or pruning an unproductive continuation and returning to an earlier branch point. TreeMem supports this control loop by keeping evidence, uncertainty, conflicts, progress, and failure cues attached to the branches that produced them, so trial outcomes can guide later decisions. Experiments on XBench-DeepSearch, BrowseComp, and BrowseComp-ZH show that TreeSeeker consistently outperforms strong open-source baselines, suggesting that explicit branch-and-return control complements stronger reasoning and tool execution.
Microsoft Research