Hyperparameter Tuning in Reinforcement Learning is Easy, Actually Permalink
TL;DR: Hyperparameter Optimization tools perform well on Reinforcement Learning, outperforming Grid Searches with less than 10% of the budget. If not reporte...
Recent Posts
Contextualize Me – The Case for Context in Reinforcement Learning Permalink
TL;DR: We can model and investigate generalization in RL with contextual RL and our benchmark library CARL. In theory, without adding context we cannot achie...
Understanding AutoRL Hyperparameter Landscapes Permalink
TL;DR: We investigate hyperparameters in RL by building landscapes of algorithm performance for different hyperparameter values at different stages of traini...
TempoRL - Learning When to Act Permalink
TL;DR: Jointly learning when and how to act improves sample efficiency of RL agents through better exploration and improved exploitation.
Theory-Inspired Parameter Control Benchmarks for DAC Permalink
TL;DR: New toy benchmarks enable better study of RL agents performance and allows us to compare against ground truth optimal policies.