--- pretty_name: Spa-Bench Physical Rollouts — VLA-0 (Partial) license: apache-2.0 task_categories: - robotics tags: - LeRobot - robotics - so-101 - spa-bench - evaluation - partial-evaluation - vla-0 configs: - config_name: default data_files: data/*/*.parquet --- # Spa-Bench physical rollouts — VLA-0 (partial) Physical SO-101 rollout recordings for the VLA-0 Spa-Bench checkpoint at the end of epoch 12. ## Scope and counts This is a partial familiar/in-distribution baseline, not a complete Spa-Bench evaluation. It contains **72 episodes**, with at least ten from each of the six task families, plus **11,051 frames, 51 unique instruction strings, two 480×640 RGB streams, and six-dimensional state/action at 10 FPS**. It contains no OOD or additional diagnostic trials. The checkpoint completed **0/72 trials (0%)**. VLA-0 received a 90-second rollout limit because autoregressive action generation was slower than the other evaluated policies. Because coverage is partial, it is excluded from the report's full OOD and diagnostic comparisons. Outcome labels and groupings are maintained in the [Spa-Bench evaluation spreadsheets](https://github.com/justintiensmith/Imperial-Thesis/tree/main/spreadsheets), not as a frame-level success field in this dataset. ## Structure and use Each trajectory stores the instruction/task index, six absolute SO-101 joint positions for state and action, and synchronized middle/wrist video. This release supports audit, integration debugging, and qualitative failure analysis. It must not be interpreted as an OOD benchmark result or a general assessment of the VLA-0 method. Do not train on these trajectories and report them as held-out performance. Robot deployment requires safeguards and human supervision. - [Evaluated model copy](https://huggingface.co/justintiensmith/Spa-Bench-VLA0-Epoch12) - [Original checkpoint repository](https://huggingface.co/mattpidden/vla0-justin-epoch12) - [Spa-Bench thesis artifact](https://github.com/justintiensmith/Imperial-Thesis) ## Citation Please cite the [completed Spa-Bench MSc report](https://github.com/justintiensmith/Imperial-Thesis/blob/main/thesis/report/spa-bench-msc-thesis.pdf), the [thesis artifact](https://github.com/justintiensmith/Imperial-Thesis), and the VLA-0 method.