LingBot-VLA 2.0 — Stack-the-cubes ablations
Finetunes of LingBot-VLA 2.0 (6B, MoE action expert) on Unitree G1 Dex1 dual-arm cube stacking. All runs freeze the vision encoder; distillation teachers stay frozen, current+future depth/DINO query heads train. Muon, lr 1e-4, L1_fm, bounds_99_woclip, absolute actions, 20000 steps, 2xA100.
| Subfolder | Data | Action space | Note |
|---|---|---|---|
joint100/ |
100ep | joint (14 arm + 2 grip) | vision frozen |
eef/ |
150ep (v1+v2) | EEF pose (14 + 2 grip) | vision frozen |
eef100/ |
100ep | EEF pose (14 + 2 grip) | vision frozen |
Ablation axes: joint100 vs eef100 = joint vs EEF (same data); eef vs eef100
= mixed 150ep vs clean 100ep. Each */global_step_*/ holds deployable HF weights
(model-0000x-of-00006.safetensors + tokenizer/config). EEF models output end-effector
poses — the client must IK them back to joints (see g1-client main_eef.py).