MAVEN: Multi-Agent Variational Exploration

Authors: Anuj Mahajan, Tabish Rashid, Mikayel Samvelyan, Shimon Whiteson

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental results show that MAVEN achieves significant performance improvements on the challenging SMAC domain [43].
Researcher Affiliation Academia Dept. of Computer Science, University of Oxford Russian-Armenian University
Pseudocode Yes Algorithm 1 MAVEN
Open Source Code No The paper does not provide any explicit statement or link to open-source code for the described methodology.
Open Datasets Yes Star Craft Multi-Agent Challenge We consider a challenging set of cooperative Star Craft II maps from the SMAC benchmark [43] which Samvelyan et al. have classified as Easy, Hard and Super Hard.
Dataset Splits No The paper mentions '32 evaluation episodes' and '100k training steps' but does not provide specific dataset split percentages, sample counts, or references to predefined train/validation/test splits.
Hardware Specification No The paper mentions 'generous equipment grant by NVIDIA and cloud credit grant from Oracle Cloud Innovation Accelerator' but does not provide specific GPU models, CPU models, or detailed cloud instance types used for experiments.
Software Dependencies No The paper does not list specific software components with their version numbers (e.g., Python 3.x, PyTorch x.x, CUDA x.x).
Experiment Setup Yes We use grid search to tune hyperparameters. Appendix C.1 contains additional experimental details.