A benchmark for evaluating learning agents based on just language feedback - View it on GitHub