kevin510/swm-ogbench-bg-hybrid-k8

OGBench blue→green: hybrid branched verifier-MPC dataset (k=8, 5,400 episodes) Training data for Qwen-based semantic world models (SWM): candidate action trajectories scored by a VLM judge at planning time. Generated by a verifier-MPC planner (simulator + Qwen3-VL-8B judge) on the OGBench visual-cube-quadruple blue→green stacking task. At each planning step the diffusion policy proposes k=8 candidate 16-step action chunks (temperature ladder 0→2, near-duplicates resampled at… See the full description on the dataset page: https://huggingface.co/datasets/kevin510/swm-ogbench-bg-hybrid-k8.

<iframe src="https://silytics.si/embed/item/dc241e70-9355-4739-912f-6af26ca98306" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.