research
Research Agents Improve After Their History Is Deleted
Periodic restarts from the best candidate break an anchoring plateau without increasing the search budget.
Summary
Periodic restarts from the best candidate break an anchoring plateau without increasing the search budget.
Across heuristic programming, kernel optimization and model-training tasks, the authors report gains up to 15 percent over communicating parallel agents and 39 percent over one agent. The forgetting step discards sessions and candidate history, preserving only the best result found so far.
Why it matters
Periodic restarts from the best candidate break an anchoring plateau without increasing the search budget.
Limits and context
- The forgetting step discards sessions and candidate history, preserving only the best result found so far.
Key claims
Periodic restarts from the best candidate break an anchoring plateau without increasing the search budget.
Qualification: The forgetting step discards sessions and candidate history, preserving only the best result found so far.
Evidence: source-2026-10-15-020
Sources
- arXiv preprint 2610.20563, original v1arXiv / submitting research authors · primary research
Corrections
: Publication note: This October 15 edition was prepared before its scheduled 9:00 AM America/Detroit reveal, but publication was delayed by a production build failure. The original reporting and image-generation timestamps are preserved.