Unverified50% confidenceFactExact time
SwarmRSI uses a discounted UCB metric to score the two frontier sets of modifying hypotheses and modifying programs, comprehensively accounting for token cost, number of experiments and unresolved-requirement pressure.
1
Sources
50%
Confidence
Long-term
Relevance
10/20/2026
First Seen
Sources
SwarmRSI: How a Swarm of Research Agents Achieves Recursive Self-Improvement
bilibili熊二等兵10/18/2026
Related Entities
Related Claims
UnverifiedIn a two-stage design, using an upper confidence bound on the pilot variance (rather than the point estimate) can significantly reduce the problem of insufficient runs.69% similarUnverifiedDuring fine-tuning, hard negatives (e.g., positions in the same field but at a different seniority level) should be used instead of random negatives to obtain maximum gains.66% similarUnverifiedThe core of a cost-based optimizer (CBO) is estimating the costs of different execution plans and choosing the lowest, relying on statistics and a cost model; excessive estimation error may lead to choosing the wrong join order.66% similarUnverifiedIn exploratory simulations, a hybrid strategy that first screens candidate solutions is estimated to have lower loss than the currently achieved joint multi-route multi-condition search approach, but the researchers note that the difference between the two-stage strategy and the relative estimation uncertainty is small, and no definitive conclusion can yet be drawn.66% similarUnverifiedWhen randomized experiments are not feasible, causal inference methods become critical, including Difference-in-Differences (DID), Regression Discontinuity Design (RDD), Instrumental Variables (IV), and Propensity Score Matching (PSM).66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1073047API
curl https://kongchang.com/api/v1/knowledge/claims/1073047MCP
get_claim(id=1073047)