https://github.com/kevinzakka/robopianist-rl
RL code for training piano-playing policies for RoboPianist.
https://github.com/kevinzakka/robopianist-rl
jax mujoco robopianist
Last synced: 7 months ago
JSON representation
RL code for training piano-playing policies for RoboPianist.
- Host: GitHub
- URL: https://github.com/kevinzakka/robopianist-rl
- Owner: kevinzakka
- License: apache-2.0
- Created: 2023-08-11T17:16:10.000Z (about 2 years ago)
- Default Branch: main
- Last Pushed: 2023-09-21T21:28:16.000Z (about 2 years ago)
- Last Synced: 2025-03-19T01:11:22.020Z (7 months ago)
- Topics: jax, mujoco, robopianist
- Language: Python
- Homepage: https://kzakka.com/robopianist/
- Size: 13.7 KB
- Stars: 48
- Watchers: 2
- Forks: 8
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# robopianist-rl
Reinforcement learning code for [RoboPianist](https://github.com/google-research/robopianist).
## Installation
Note: Make sure you are using the same conda environment you created for RoboPianist (see [here](https://github.com/google-research/robopianist/blob/main/README.md#installation)).
1. Install [JAX](https://github.com/google/jax#installation)
2. Run `pip install -r requirements.txt`
## Usage
We provide an example bash script to train a policy to play Twinkle Twinkle Little Star with the task parameters used in the paper.
```bash
bash run.sh
```
To look at all the possible command-line flags, run:
```bash
python train.py --help
```
## Citation
If you use this code, please cite our paper:
```bibtex
@article{zakka2023robopianist,
author = {Zakka, Kevin and Smith, Laura and Gileadi, Nimrod and Howell, Taylor and Peng, Xue Bin and Singh, Sumeet and Tassa, Yuval and Florence, Pete and Zeng, Andy and Abbeel, Pieter},
title = {{RoboPianist: A Benchmark for High-Dimensional Robot Control}},
journal = {arXiv preprint arXiv:2304.04150},
year = {2023},
}
```