Rethinking Value Function Learning for Generalization in Reinforcement Learning

Authors: Seungyong Moon, JunYeong Lee, Hyun Oh Song

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our proposed algorithms significantly improve observational generalization performance and sample efficiency on the Procgen Benchmark.
Researcher Affiliation Collaboration 1Seoul National University, 2Neural Processing Research Center, 3Deep Metrics
Pseudocode Yes Algorithm 1 Dynamics-aware Delayed-Critic Policy Gradient (DDCPG)
Open Source Code Yes The code can be found at https://github.com/snu-mllab/DCPG.
Open Datasets Yes In this paper, we utilize the Procgen benchmark as a testbed for observational generalization [10].
Dataset Splits No The paper describes training and testing on the Procgen benchmark but does not explicitly mention a validation dataset or split details for it.
Hardware Specification No The paper states: 'We describe the computational resource in the supplementary material.', implying these details are not in the main body of the paper.
Software Dependencies No The paper does not provide specific software dependencies with version numbers in the main text.
Experiment Setup No For the implementation details and hyperparameters, please refer to Appendix D.