Datasets and evaluation results for Mulligan: Performance-Guided Data Collection for Efficient On-Robot Learning.
Performance-Guided Data Collection for Efficient On-Robot Learning
Project and paper · Policy Arena · Full dataset catalog and mapping
release release-1: 241 verified trajectory repositories plus two simulation evaluation bundles. The Arena selects mainline results and required training ancestry. Derived views share episodes with their raw parents.
cNN: collection increment · rNN: model round · bNN: evaluation block. Consult the card and training recipe before mixing views or treating a historical evaluation as held out.
Use the release-1 tag or its commit for reproducibility. Historical standalone D1/pilot campaigns are excluded; D1-named Routing sources are retained only as required D2 ancestry.
596 checkpoints in 180 model repositories that produced or scored the released episodes. Each card lists training and validation datasets and evaluation results.
