cloudwalk-research/psi0-g1-sneaker-199ep-v2
Psi0 G1 Sneaker-in-Box — 199 episodes (v2 training) Training subset of the 205-episode v2 canonical source, with 6 episodes held out for open-loop evaluation. Used as the input for the v2 Psi0 fine-tune (CloudWalk Research). Bimanual: roughly half LEFT-hand demos, half RIGHT-hand. Same task as v1 — "Pick up the sneaker and place it inside the box." — same Unitree G1 + Inspire RH56DFTP hardware. The held-out 6 episodes live in psi0-g1-sneaker-6ep-v2-eval. 199 + 6 = 205… See the full description on the dataset page: https://huggingface.co/datasets/cloudwalk-research/psi0-g1-sneaker-199ep-v2.
Open-source release: standardize card to LeRobot house-style; Apache-2.0 (HE pickplace keeps upstream NC); CW-RL attribution + citations.
Phase 11.5 — embed radial chart as Figure 1; renumber others (2026-05-07)
Phase 11.5 — add radial checkpoint-quality chart (2026-05-07)
Phase 11.5 — move legends out of axes; add 'lower is better' (2026-05-07)
Phase 11.5 — move legends out of axes; add 'lower is better' (2026-05-07)
Phase 11.5 — move legends out of axes; add 'lower is better' (2026-05-07)
Phase 11.5 — move legends out of axes; add 'lower is better' (2026-05-07)
Phase 11.5 — embed Figures 3 & 4 (cross-ckpt + per-ckpt grid) (2026-05-07)
Phase 11.5 — add cross-ckpt + per-ckpt grid plots (2026-05-07)
Phase 11.5 — add cross-ckpt + per-ckpt grid plots (2026-05-07)
Phase 11.5 — embed convergence + per-side asymmetry plots in README (2026-05-07)
Phase 11.5 — add open-loop eval result plots (2026-05-07)
Phase 11.5 — open-loop ID + OOD eval results (2026-05-07)
Add dataset card (README.md)
Initial: 199 training episodes (205 - 6 held-out for eval-v2). Renumbered 0..198. Stats recomputed over 199.
initial commit