Experiment Results for "Towards Understanding the Impact of Plasticity Loss on Reinforcement Learning in Stochastic Environments"

Dataset with all the metrics collected and computed throughout experiment execution for the paper "Towards Understanding the Impact of Plasticity Loss on Reinforcement Learning in Stochastic Environments" published at the Nineteenth European Workshop on Reinforcement Learning, 2026. Metrics are episode returns, neural network plasticity and metrics associated with plasticity loss (e.g. representation effective rank or dormant neurons). Each run is structured into its own subdirectory, and the data for the run is organized into a single .csv file, alongside a config.yaml and run_info.yaml file required for compatibility with the data processing pipelines published on GitHub.