Pith. sign in

← back to paper

Review history

arxiv: 2510.24295 · 2 revisions

MERGE: Minimal Expression-Replacement GEneralization Test for Natural Language Inference

  1. 2026-08-15 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 6.0
    147091 ms 20123 in 17705 out 2026-08-15T15:42:14.963904+00:00
  2. 2026-08-04 CONDITIONAL MODERATE v1.3.0-alltime-deepseek novelty 6.0
    96299 ms 20069 in 10630 out 2026-08-04T07:44:58.084785+00:00