NVIDIA's MOTIVE: Precisely Filtering Video Training Clips
NVIDIAAI · x · 2026-07-08
NVIDIA Research's paper MOTIVE received an Outstanding Paper Honorable Mention at ICML 2026. Targeting video model training, the method re-weights training signals—favoring motion regions while downplaying static backgrounds. By scoring the impact of motion per clip, it filters out small batches of high-impact clips for fine-tuning to improve temporal dynamics. MOTIVE showed improvements in VBench dynamic metrics and achieved a 74.1% human preference win rate.
More from Research
- NUS builds a soft force sensor that drives actuators without electronics or power — CurieuxExplorer · 2026-07-27
- Chelsea Finn says robot RL is bottlenecked by physical rollout cost, not algorithms — ycombinator · 2026-07-27
- ICML 2026 oral paper replication scores stay middling after a stricter re-scoring — profjamesevans · 2026-07-27
- Long-running agents will need immutable event logs, this thread argues — sebpaquet · 2026-07-27
- Seed IQ navigates Doom II, prompting questions about benchmarks beyond ARC-AGI — Fit_Transition8824 · 2026-07-27
- Agentic Data Science in Practice: Agents Write Code but Answer Wrong Questions — hugobowne · 2026-07-27