|
--- |
|
license: apache-2.0 |
|
library_name: lerobot |
|
pipeline_tag: robotics |
|
--- |
|
|
|
## Pi0 pretrained model |
|
|
|
This repository contains the model described in [π_0: A Vision-Language-Action Flow Model for General Robot Control](https://huggingface.co/papers/2410.24164). |
|
|
|
See the [Twitter thread](https://x.com/RemiCadene/status/1886823939856589296) and [blog post](https://huggingface.co/blog/pi0) for more info regarding its integration in [LeRobot](https://github.com/huggingface/lerobot). |
|
|
|
## Usage |
|
|
|
You can download and use this model with: |
|
```python |
|
policy = Pi0Policy.from_pretrained("lerobot/pi0") |
|
action = policy.select_action(batch) |
|
``` |
|
|
|
## Fine-tuning |
|
|
|
You can easily finetune it on your dataset. For instance on @dana_55517 's [dataset](https://huggingface.co/spaces/lerobot/visualize_dataset?dataset=danaaubakirova%2Fkoch_test&episode=0): |
|
```python |
|
python lerobot/scripts/train.py \ |
|
--policy.path=lerobot/pi0 \ |
|
--dataset.repo_id=danaaubakirova/koch_test |
|
``` |
|
|
|
Take a look at the [code](https://github.com/huggingface/lerobot/blob/main/lerobot/common/policies/pi0/modeling_pi0.py) regarding the implementation. |
