TheMachine Press

A daily newspaper for the age of artificial intelligence.

Morning editionPermanent story

research

The Search Agent Let Coverage Decide How Many Documents to Keep

CAS replaces fixed top-K retrieval with adaptive prediction sets and penalizes low-confidence training trajectories.

Published Updated Story ID: mp-2026-08-24-026
Read the complete editionStory JSON

Summary

CAS replaces fixed top-K retrieval with adaptive prediction sets and penalizes low-confidence training trajectories.

The framework applies conformal methods on both sides of search-agent training: document sets expand or contract to target coverage, while answer confidence weights the GRPO objective. The preprint reports higher reasoning accuracy and fewer redundant tool calls across single- and multi-hop QA datasets.

Why it matters

CAS replaces fixed top-K retrieval with adaptive prediction sets and penalizes low-confidence training trajectories.

Limits and context

No additional limitation was separately recorded.

Key claims

  1. CAS replaces fixed top-K retrieval with adaptive prediction sets and penalizes low-confidence training trajectories.

    Evidence: source-2026-08-24-015

Sources

  1. arXiv preprint 2608.20771arXiv · primary research

Corrections

No corrections have been recorded for this story.