frontier models
The Search Agent Gets Credit for What It Learned
KnowGPO aligns trajectories by acquired knowledge conditions instead of comparing noisy pages or internal states.
Summary
KnowGPO aligns trajectories by acquired knowledge conditions instead of comparing noisy pages or internal states.
Deep-search agents receive many heterogeneous observations before a single final reward, which makes it hard to identify which query or page mattered. KnowGPO decomposes the answer into atomic knowledge conditions, represents each trajectory by the conditions it has satisfied and compares trajectories within the same knowledge state. Incremental knowledge gain becomes a process reward. Across Xbench, BrowseComp-ZH and GAIA, the authors report gains of 3.8 to 4.2 points over outcome-only and process-supervision baselines; their 8-billion-parameter model rivaled or exceeded most tested 32-billion-parameter open systems. The decomposition itself remains a dependency and potential source of error.
Why it matters
KnowGPO aligns trajectories by acquired knowledge conditions instead of comparing noisy pages or internal states.
Limits and context
- Across Xbench, BrowseComp-ZH and GAIA, the authors report gains of 3.8 to 4.2 points over outcome-only and process-supervision baselines; their 8-billion-parameter model rivaled or exceeded most tested 32-billion-parameter open systems.
Key claims
KnowGPO aligns trajectories by acquired knowledge conditions instead of comparing noisy pages or internal states.
Qualification: Across Xbench, BrowseComp-ZH and GAIA, the authors report gains of 3.8 to 4.2 points over outcome-only and process-supervision baselines; their 8-billion-parameter model rivaled or exceeded most tested 32-billion-parameter open systems.
Evidence: source-2026-10-18-011
Sources
- arXiv preprint 2610.22331, original v1arXiv · primary research
Corrections
No corrections have been recorded for this story.