occupancy
smolvlm_500m-parking_occupancy-PKLot-instruct-without-context-without-expertsmolvlm_256m-parking_occupancy-CNRPark-instruct-with-context-with-expertsmolvlm_500m-parking_occupancy-PKLot-instruct-with-context-with-expertsmolvlm_500m-parking_occupancy-CNRPark-instruct-without-context-with-expertsmolvlm_500m-parking_occupancy-PKLot-instruct-with-context-without-expertsmolvlm_256m-parking_occupancy-CNRPark-instruct-without-context-without-expertsmolvlm_500m-parking_occupancy-CNRPark-instruct-with-context-without-expertsmolvlm_256m-parking_occupancy-PKLot-instruct-without-context-without-expert
Nuplan-Occupancyvideo-occupancy-models-datasetsDatasets used in the paper Video Occupancy Models.
library-occupancyoccupancy_percdaily-paper-2026-08-19-ghost-occupancy-gpu-reclaim
Ghost Occupancy: An LLM-Agent Protocol for Detecting and Reclaiming Idle GPU Deployments in Shared Kueue-Managed Kubernetes Clusters
ThakiCloud AI Research · 2026-08-19 · 📝 Tech blog (KO)
Figures
Mean over 5 seeds x 2,000 instances. Both utilization-only detectors sit at perfect recall but pay for it on the horizontal axis, where every percentage point is an active job killed. Lengthening the window walks left from 10.93% to 3.50% without reaching the axis --… See the full description on the dataset page: https://huggingface.co/datasets/thaki-AI/daily-paper-2026-08-19-ghost-occupancy-gpu-reclaim.real-time-library-occupancy