← back to paper

Review history

arxiv: 2608.06243 · 2 revisions

DASH: Divergence-Adaptive Supervision Horizons for On-Policy Self-Distillation of Reasoning Models

1. 2026-08-15 CONDITIONAL HIGH v1.4.0-alltime-deepseek-medium novelty 6.0
   111399 ms 20429 in 14810 out 2026-08-15T14:33:53.948334+00:00
2. 2026-08-07 CONDITIONAL MODERATE v1.4.0-daily-deepseek-medium novelty 6.0
   123899 ms 20481 in 12496 out 2026-08-07T11:36:22.717937+00:00