Google open-sources TIPSv2 image-text encoders, SOTA on all four zero-shot segmentation benchmarks
bdsqlsz · x · 2026-08-21
Google released TIPSv2 in four sizes, each with a general image-text encoder plus a DPT variant for dense vision tasks, under Apache 2.0. It sets SOTA on all four reported zero-shot segmentation benchmarks and top-two results on 5/7 image-text and 7/9 image-only evaluations. At ViT-L scale it beats DINOv3 on 4/6 shared tasks despite its teacher having 6× more parameters and 15× more training images. Technically, iBOT++ lifts ADE150 zero-shot segmentation by 14.1 mIoU, and head-only EMA cuts training parameters by 42%.
More from Research
- Devin Oncall tops SREGym with 100% end-to-end success, auto-resolving production incidents — tianyin_xu · 2026-10-16
- SJTU AutoLab releases Drive3R, a streaming geometric foundation model for long-horizon driving 3D reconstruction — AutoLab-SJTU · 2026-10-16
- Cortex A/B-tests coding-agent skills and rules by replaying developer corrections before keeping them — Samuele Giampieri · 2026-10-16
- Zero-latency lab-in-the-loop tests whether drug-discovery agents cheat via memorization — niloofar_mire · 2026-10-16
- 260M-parameter looped diffusion transformer beats 6x larger baselines with 5x less compute — burkov · 2026-10-16
- Stanford's Next-Concept Modeling sketches text first, 21x faster generation on one H100 — lmoroney · 2026-10-16