smolvla_busybox_push_green_button
Full-component fine-tune of SmolVLA for busybox_push_green_button on SO101 data.
| Policy | SmolVLA (freeze_vision_encoder=false, train_expert_only=false) |
| Init checkpoint | lerobot/smolvla_base |
| Dataset | villekuosmanen/busybox_push_green_button |
| Task | busybox_push_green_button |
| Action dim | 6 (single-arm) |
| Cameras | native top / wrist / front remapped to camera1 / camera2 / camera3 |
| Training | Vast.ai 4× A100-80GB DDP, full-component, 30k steps, global batch 32 (8×4), chunk/action horizon 50, 720x1280 cameras |
| W&B project | smolvla_busybox_push_green_button |
| W&B run | e0cq4jze |
Checkpoints
main is step 30,000. Intermediate steps are git tags:
| Step | Revision |
|---|---|
| 30000 | step-30000 (main) |
| 25000 | step-25000 |
| 20000 | step-20000 |
| 15000 | step-15000 |
| 10000 | step-10000 |
| 5000 | step-5000 |
Usage
from lerobot.policies.smolvla.modeling_smolvla import SmolVLAPolicy
policy = SmolVLAPolicy.from_pretrained("pravsels/smolvla_busybox_push_green_button")
# earlier step: revision="step-25000"
- Downloads last month
- -
Model tree for pravsels/smolvla_busybox_push_green_button
Base model
lerobot/smolvla_base