File size: 1,164 Bytes
0ca6ed0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
---
library_name: gr00t
tags:
- gr00t
- gr00t-n1.7
- robotics
- vla
- bimanual
base_model: nvidia/GR00T-N1.7-3B
---

# groot1.7_insert_candle_20k

Fine-tuned [GR00T N1.7-3B](https://huggingface.co/nvidia/GR00T-N1.7-3B) for `insert_candle` on SO101 data.

| | |
|---|---|
| **Base model** | [nvidia/GR00T-N1.7-3B](https://huggingface.co/nvidia/GR00T-N1.7-3B) |
| **Dataset** | [villekuosmanen/armnetbench_insert_candle](https://huggingface.co/datasets/villekuosmanen/armnetbench_insert_candle) |
| **Task** | `insert_candle` |
| **Embodiment** | bimanual 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_insert_candle](https://huggingface.co/pravsels/groot1.7_insert_candle) |
| **W&B project** | [groot1.7_insert_candle](https://wandb.ai/pravsels/groot1.7_insert_candle) |
| **W&B run** | [a2yr5czu](https://wandb.ai/pravsels/groot1.7_insert_candle/runs/a2yr5czu) |

## 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).