NexSight: verifiable long-video tasks show top models score only 54.5%
nex-agi · hf · 2026-11-19
NexSight (Zhejiang Univ., Tongji, Shanghai Qizhi) builds verifiable tasks for agentic long-video understanding. Its SightForge pipeline yields SightLine — 310 evidence-traceable tasks from 108 long videos — where the strongest model scores only 54.52% (79.35% with annotated evidence). Using solver trajectories as step-level supervision, Sight-35B nearly doubles interactive accuracy (21.29%→42.26%) with under 6K examples, beating its family's 397B model.
More from Research
- tessera-small: a diffusion-LLM-based pixel-art text-to-image model — thejagstudio · 2026-11-19
- WorldRuler: rule-graph world models that honor user-defined game rules — AlayaLab · 2026-11-19
- PReMind: benchmark + RL framework for hazard-aware personalized VLM assistants — SeoulNatlUniv · 2026-11-19
- Late Entry in looped LLMs delivers up to 12.5x lossy and 8.4x exact speedups — NanyangTechnologicalUniversity · 2026-11-19
- Optimization × AI Forum on Nov 20: two Peking University talks on AI-driven math discovery — aaron_defazio · 2026-11-19
- SE3 Labs Releases First Open YAM Real-Robot Teleop Dataset: 500 Bimanual Episodes — paigeinsf · 2026-11-19