---
license: apache-2.0
task_categories:
- reinforcement-learning
---
# transfer from d4rl dataset to minari dataset

transfer scripts and validation are in transfer.py

1. clone the repo
 ```
$ git clone https://huggingface.co/datasets/im-Kitsch/minari_d4rl
```

2. copy the file to minari root (default is ~/.minari)

```
mv minari_d4rl/datasets ~/.minari/datasets
```

# todo 
infos like `infos/qvel` are not saved since the interface is not stable yet and those infos cannot be read directly.