lerobot/README.md

69 lines
1.1 KiB
Markdown

# LeRobot
## Installation
Install dependencies using `conda`:
```
conda env create -f environment.yaml
conda activate lerobot
```
**dev**
```
python setup.py develop
```
## Usage
### Train
```
python lerobot/scripts/train.py \
--config-name=pusht hydra.job.name=pusht
```
### Visualize offline buffer
```
python lerobot/scripts/visualize_dataset.py \
--config-name=pusht hydra.run.dir=tmp/$(date +"%Y_%m_%d")
```
### Visualize online buffer / Eval
```
python lerobot/scripts/eval.py \
--config-name=pusht hydra.run.dir=tmp/$(date +"%Y_%m_%d")
```
## TODO
- [x] priority update doesnt match FOWM or original paper
- [x] self.step=100000 should be updated at every step to adjust to horizon of planner
- [ ] prefetch replay buffer to speedup training
- [ ] parallelize env to speedup eval
- [ ] clean checkpointing / loading
- [ ] clean logging
- [ ] clean config
- [ ] clean hyperparameter tuning
- [ ] add pusht
- [ ] add aloha
- [ ] add act
- [ ] add diffusion
- [ ] add aloha 2
## Contribute
**style**
```
isort lerobot
black lerobot
isort test
black test
pylint lerobot
```