lawam_so101_block_stack_sim

LaWAM fine-tuned to stack one 30 mm block on another with an SO-101, in simulation. Trained by train/lawam_so101.yaml in vla-test on drwahl/so101_block_stack_sim.

LoRA (r=32, alpha 64) on the language backbone of jialei02/lawam-pretrain-lerobot, with the flow head, the VLM-to-LAM projection and the LAM decoder dense, vision frozen: 20k steps at batch 4, the protocol drwahl/pi05_so101_block_stack_sim was trained on. Normalization, learning rate and chunk length are the base checkpoint's own. adapter/ holds the adapter and the processors, which carry the normalization and the camera renaming; lerobot loads it onto the base weights.

hf download drwahl/lawam_so101_block_stack_sim --local-dir lawam_so101_block_stack_sim
uv run python sim/scripts/eval.py lawam_so101_block_stack_sim/adapter --no-smooth --horizon 10

The eval rolls the policy out on the 150 held-out layouts and scores it by the gates that screened the oracle. Executing 10 steps of each 24-step chunk, the setting that stacks most on one seed, it stacks 110.7 of 150 and lifts 123.3, averaged over three further seeds. Executed whole, its chunk stacks 108.3.