← back to paper

Review history

arxiv: 2508.19277 · 2 revisions

POT: Inducing Overthinking in LLMs via Black-Box Iterative Optimization

1. 2026-08-15 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 5.0
   121630 ms 13704 in 13038 out 2026-08-15T17:07:29.095220+00:00
2. 2026-08-05 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 6.0
   133619 ms 13691 in 13610 out 2026-08-05T17:02:20.145386+00:00