← back to paper

Review history

arxiv: 2605.12309 · 2 revisions

G$^2$TR: Generation-Guided Visual Token Reduction for Separate-Encoder Unified Multimodal Models

1. 2026-05-19 CONDITIONAL LOW v0.9.0 novelty 6.0
   36160 ms 5826 in 1319 out 2026-05-19T16:49:05.384592+00:00
2. 2026-05-13 UNVERDICTED LOW v0.9.0 novelty 7.0
   126399 ms 5578 in 1340 out 2026-05-13T05:28:37.832351+00:00