← back to paper

Review history

arxiv: 2608.01635 · 2 revisions

Mitigating Visual Degradation in MLLMs via Spatial-Spectral Visual Anchor Learning

1. 2026-08-15 CONDITIONAL MODERATE v1.4.0-alltime-deepseek-medium novelty 6.0
   154879 ms 17719 in 18713 out 2026-08-15T15:06:25.297736+00:00
2. 2026-08-04 CONDITIONAL MODERATE v1.3.0-daily-deepseek novelty 6.0
   130279 ms 153 in 9794 out 2026-08-04T23:49:11.384826+00:00