SmolVLA green-marker (n=28)
Fine-tuned from lerobot/smolvla_base on the SO-101 green-box → marker task.
| Dataset | zchn/robo_test_green_marker_v0 @ eddb45d (28 episodes, V1+V2) |
| Steps | 5000 |
| Final train loss | ~0.116 |
| Device | MPS |
Camera rename
Training used --rename_map so dataset keys match the base policy:
observation.images.handeye→observation.images.camera1observation.images.front→observation.images.camera2empty_cameras=1(third camera slot)
Task string: put the green box on the marker.
- Downloads last month
- 41
Model tree for zchn/smolvla_green_marker_n28
Base model
HuggingFaceTB/SmolLM2-360M Quantized
HuggingFaceTB/SmolLM2-360M-Instruct Quantized
HuggingFaceTB/SmolVLM-500M-Instruct Finetuned
lerobot/smolvla_base