AutoRL-Bench

Repo with the API and benchmark-generating code for AutoRL-Bench

Download the data

Get the data from HERE, download this repo and put it at the level of this repository folder.

git clone https://github.com/releaunifreiburg/AutoRL-Bench.git

cd AutoRL-Bench

wget https://figshare.com/ndownloader/files/37650989?private_link=a60c401c85ab9e8815f6

Load and query the benchmark

from benchmark_handler import BenchmarkHandler

benchmark = BenchmarkHandler(data_path = "/data_arl_bench",
                             environment = "Pendulum-v0", seed = 0,
                             search_space = "PPO", static = True)

#querying static configuration
configuration_to_query = {"lr":-6, "gamma": 0.8, "clip": 0.2}
queried_data = benchmark.get_metrics(configuration_to_query, budget=50)

#querying dynamic configuration
benchmark.static = False
configuration_to_query = {"lr":[-3,-4], 
                          "gamma": [0.8,0.99], 
                          "clip": [0.2, 0.2]}
queried_data = benchmark.get_metrics(configuration_to_query, budget=50)

Further usage

For an insightful usage description please check the file benchmark-usages-examples.ipynb

Cite Us

Please cite us when using AutoRL-Bench in your work by using the bibtex below:

@inproceedings { autorlbench-22,
  author = {Gresa Shala and Sebastian Pineda and Andr{\'e} Biedenkapp and Frank Hutter and Josif Grabocka},
  title = {AutoRL-Bench 1.0},
  booktitle = {Proceedings of the Workshop on Meta-Learning (NeurIPS)},
  year = {2022}
}