Michal Podolinsky

GR00T-N1.7 LIBERO-X backbone features — 90-task fine-tune (LEVEL1-3) Aligned rollouts of a LIBERO-X fine-tune of GR00T-N1.7 (rohansiva/gr00t-libero-x-90task) on the LIBERO-X simulator, over the exact 90 tasks that checkpoint was fine-tuned on (30 tasks × 3 difficulty levels, LEVEL1–LEVEL3). Train and eval task sets are identical by design, so this is an in-distribution dataset for the checkpoint. 90 tasks × 20 rollouts = 1,800 episodes (600 per level). Every GR00T inference