← back to paper

Review history

arxiv: 2607.27902 · 2 revisions

One Patch Is Enough: Reinforcement-Optimized Visual Token Grounding for MLLM-Based Scene Text Spotting

1. 2026-08-04 CONDITIONAL MODERATE v1.3.0-alltime-deepseek novelty 6.0
   673489 ms 22221 in 14826 out 2026-08-04T03:15:58.901799+00:00
2. 2026-07-31 CONDITIONAL MODERATE v1.3.0-daily-deepseek novelty 5.0
   174093 ms 22198 in 18427 out 2026-07-31T23:19:53.790843+00:00