SURGE Reasoning Tasks Keep Improving Through 14 Labeled Videos; CVoS Resists Pretraining
maier_ak · x · 2026-10-14
Continuing the SURGE surgical foundation model label-efficiency study: unlike perception tasks that saturate after one labeled video, reasoning keeps benefiting from more data — instrument-action-target triplets climb from 20.30 to about 38 mAP across 14 labeled videos without saturating.
The Critical View of Safety also resists pretraining, suggesting high-level safety reasoning cannot be acquired via large-scale unlabeled pretraining.
Related event: Surgical foundation model SURGE maps the limits of annotation gains(3 posts)→
More from Research
- Neuro for AI & AI for Neuro workshop at AAAI 2027 opens call for papers in Montreal — naturecomputes · 2026-10-14
- BVB Benchmark Tests Agent Video Understanding by Rebuilding 288 Real Videos in Blender — SonglinYang4 · 2026-10-14
- NSF awards $20M to AI Institute in Dynamic Systems for robotic manufacturing — eigensteve · 2026-10-14
- Should ICLR slash its acceptance rate to ~10% as submissions surge? — boknilev · 2026-10-14
- Is 'Attention' the Essence of Intelligence? Quanta examines the idea behind Transformers — TMWNN · 2026-10-14
- A Full QC Pipeline for RL Data: Intake Review, Active Testing, Packaging — joecole · 2026-10-14