A dataset API for Offline Reinforcement Learning.
Minari is a Python API that hosts a collection of popular Offline Reinforcement Learning datasets. The environments from which these datasets are generated follow the Gymnasium API. The datasets are publicly available in a Farama GCP bucket and can be downloaded through the Minari CLI. Minari also provides dataset handling features such as episode sampling, filtering trajectories, as well as dataset generation utilities.
Note
This library is currently under beta development and minor/major changes are expected to come along in the near future.
// Install Minari
$ pip install minari
---> 100%
// Show remote datasets.
$ minari list remote
<i>Minari datasets in Farama server</i>
┌────────────────┬───────────┬────────────┬─────────┬───────────┐
│ │ <b>Total</b> │ <b>Total</b> │ │ │
│ <b>Name</b> │ <b>Episodes</b> │ <b>Steps</b> │ <b>Author</b> │ <b>Email</b> │
┡────────────────╇───────────╇────────────╇─────────╇───────────┩
│ <font color="#A1EFE4">door-cloned-v0</font> │ <font color="#03AC13">4356</font> │ <font color="#03AC13">1000000</font> │ <font color="#FF00FF">Farama</font> │<font color="#FF00FF">@farama.org</font>│
│ <font color="#A1EFE4">door-expert-v0</font> │ <font color="#03AC13">5000</font> │ <font color="#03AC13">1000000</font> │ <font color="#FF00FF">Farama</font> │<font color="#FF00FF">@farama.org</font>│
│ <font color="#A1EFE4">door-human-v0</font> │ <font color="#03AC13">25</font> │ <font color="#03AC13">6729</font> │ <font color="#FF00FF">Farama</font> │<font color="#FF00FF">@farama.org</font>│
└────────────────┴───────────┴────────────┴─────────┴───────────┘
// Download dataset door-cloned-v0
$ minari download door-cloned-v0
Downloading door-cloned-v0 from Farama servers...
* Downloading data file 'door-cloned-v0/data/main_data.hdf5' ...
---> 100%
Dataset door-cloned-v0 downloaded to <path-to-local-datasets>/.minari/datasets/door-cloned-v0
"見習い - learning by observation"