Observation record / arXiv · 人工智能

Recursive Reward Improvement via Adversarial Meta-Prompt Evolution

First seen 10/17, 04:20 · Last observed 10/18, 10:21 · Source data is current

Title-based outlineRecursive Reward Improvement via Adversarial Meta-Prompt Evolution

Why it mattersResearch metadata links to papers and authors. Read the original for conclusions; publication order is not impact.

Published 10/17, 12:00

30 minutesRank movement—

Building history

1 hourRank movement—

Building history

24 hoursRank movement—

Building history

Publication position history

6 observations

Publication feed in source order. Position changes do not count as heat or momentum.

arXiv metadata CC0; papers retain individual licenses· Author Wei Dai, Qianchu Liu, Guanghui Qin, Xing Han, Tristan Naumann, Paul Pu Liang, Sheng Zhang

View original observation times and ranks

Observation time (Beijing)	Source position
10/18, 10:21	29
10/18, 04:21	29
10/17, 22:21	29
10/17, 16:21	29
10/17, 10:20	29
10/17, 04:20	29

Daily aggregate history

日期	Best daily position	Last daily position	Observations
2026-10-17	29	29	4
2026-10-18	29	29	2