| --- |
| library_name: gr00t |
| tags: |
| - gr00t |
| - gr00t-n1.7 |
| - robotics |
| - vla |
|
|
| base_model: nvidia/GR00T-N1.7-3B |
| --- |
| |
| # groot1.7_ring_insert_20k |
| |
| Fine-tuned [GR00T N1.7-3B](https://huggingface.co/nvidia/GR00T-N1.7-3B) for `ring_insert` on SO101 data. |
|
|
| | | | |
| |---|---| |
| | **Base model** | [nvidia/GR00T-N1.7-3B](https://huggingface.co/nvidia/GR00T-N1.7-3B) | |
| | **Dataset** | [villekuosmanen/armnetbench_ring_insert](https://huggingface.co/datasets/villekuosmanen/armnetbench_ring_insert) | |
| | **Task** | `ring_insert` | |
| | **Embodiment** | single-arm SO-101 | |
| | **Training** | 20k steps, batch 32 (8/GPU × 4 DDP), 4× RTX 5090 on Vast.ai, load_bf16=True | |
| | **Prior HF repo** | [pravsels/groot1.7_ring_insert](https://huggingface.co/pravsels/groot1.7_ring_insert) | |
| | **W&B project** | [groot1.7_ring_insert](https://wandb.ai/pravsels/groot1.7_ring_insert) | |
| | **W&B run** | [oo1rqf7q](https://wandb.ai/pravsels/groot1.7_ring_insert/runs/oo1rqf7q) | |
| |
| ## Checkpoints |
| |
| Final checkpoint (step `20000`) lives at the **repository root** for direct loading. |
| |
| ## Usage |
| |
| Load with the GR00T inference stack from this repo root (weights + processor configs). |
| |
