xArm tactile: points9_arrow_len0 ablation
Collection
Zero-length-arrow tactile overlay ablation: paired LeRobot dataset + pi0.5 LoRA checkpoint for 4 xArm tasks. • 8 items • Updated
pi05_xarm_dishwasher_points9_arrow_len0_lora
pi0.5 (openpi) LoRA fine-tune for the xArm dishwasher task ("pull the basket outside of the dishwasher and pick up the mug and put it into the basket"), trained on
camera frames with the points9_arrow_len0 tactile overlay: 9 tactile-pad dots per
finger at FK-projected positions, force arrows at zero length — the force-information
ablation of points9_arrow.
Training
| Base | pi05_droid (knowledge-insulated pi0.5, DROID pre-training) |
| Recipe | LoRA r16/a16 on Gemma-2B backbone + r32/a32 on 300M action expert, attn+ffn |
| Batch / LR | 8 / cosine 1e-4 -> 1e-5, 500 warmup, AdamW, grad-clip 1.0 |
| Data | EdwardoSunny/xarm_dishwasher_points9_arrow_len0 (100 eps, 28096 frames) |
| Early stop | step 7800 (rolling-1k-window <0.5% rel. improvement, 2 consecutive checks) |
| Final train loss | 0.0287 (baseline no-overlay: 0.029; full points9_arrow: 0.024) |
Training-loss comparison across this task's variants (identical hyperparameters/protocol, only the burned-in overlay differs) — see the collection for all four tasks.
Contents
params/— merged base+LoRA weights (orbax), self-contained for inferenceassets/local/xarm_dishwasher_points9_arrow_len0/norm_stats.json— state/action normalization- Optimizer state stripped (inference-ready). Config name:
pi05_xarm_dishwasher_points9_arrow_len0_lora
Deployment note
At inference, render the live camera frames with the SAME overlay
(SensorDrawer mode points9_arrow, arrow_length_scale=0, thickness 8, dot size 22)
or the policy will be out-of-distribution. See tactile-data-collection/scripts/render_arrowlen0.py.