---
library_name: lerobot
tags:
- smolvla
- robotics
- lerobot
- bimanual
- so101
- armnetbench
base_model: lerobot/smolvla_base
datasets:
- villekuosmanen/armnetbench_transfer_cube
---

# smolvla_transfer_cube

Fine-tuned [SmolVLA](https://huggingface.co/lerobot/smolvla_base) for bimanual SO101 (12-dim action/state).

| | |
|---|---|
| **Base model** | [lerobot/smolvla_base](https://huggingface.co/lerobot/smolvla_base) |
| **Dataset** | [villekuosmanen/armnetbench_transfer_cube](https://huggingface.co/datasets/villekuosmanen/armnetbench_transfer_cube) |
| **Checkpoints** | 5,000 and 10,000 steps |
| **Training** | Vast.ai 4xH100, batch 128, ~8.5h total |
| **Action dim** | 12 (bimanual SO101, 6 per arm) |
| **Cameras** | top, left_wrist, right_wrist |
| **W&B** | [smolvla_transfer_cube](https://wandb.ai/pravsels/smolvla_transfer_cube/runs/w2j060oy) |

## Checkpoints

- `checkpoints/005000/pretrained_model/`
- `checkpoints/010000/pretrained_model/`

## Usage

```python
from lerobot.policies.smolvla.modeling_smolvla import SmolVLAPolicy

policy = SmolVLAPolicy.from_pretrained("pravsels/smolvla_transfer_cube", subfolder="checkpoints/010000/pretrained_model")
```

## Training Logs

Full training curves and metrics: [https://wandb.ai/pravsels/smolvla_transfer_cube/runs/w2j060oy](https://wandb.ai/pravsels/smolvla_transfer_cube/runs/w2j060oy)