SMAC-v-2

SMAC-v2 is a benchmark dataset used to evaluate multi-agent reinforcement learning algorithms, featuring diverse scenarios with varying numbers of agents across tasks.

Papers using SMAC-v-2 (13)

SPECTra: Scalable Multi-Agent Reinforcement Learning with Permutation-Free Networks2025 · 2 citesSTAIRS-Former: Spatio-Temporal Attention with Interleaved Recursive Structure Transformer for Offline Multi-task Multi-agent Reinforcement Learning2026Prism: Spectral Parameter Sharing for Multi-Agent Reinforcement Learning2026Bandwidth-constrained Variational Message Encoding for Cooperative Multi-agent Reinforcement Learning2025Preference-Guided Learning for Sparse-Reward Multi-Agent Reinforcement Learning2025MACTAS: Self-Attention-Based Inter-Agent Communication in Multi-Agent Reinforcement Learning with Action-Value Function Decomposition2025Fixing Incomplete Value Function Decomposition for Multi-Agent Reinforcement Learning2025R3DM: Enabling Role Discovery and Diversity Through Dynamics Models in Multi-agent Reinforcement Learning2025TACTIC: Task-Agnostic Contrastive pre-Training for Inter-Agent Communication2025Boosting Sample Efficiency and Generalization in Multi-agent Reinforcement Learning via Equivariance2024 · 2 citesQTypeMix: Enhancing Multi-Agent Cooperative Strategies through Heterogeneous and Homogeneous Value Decomposition2024Fault Tolerant Multi-Agent Learning with Adversarial Budget Constraints2025Wonder Wins Ways: Curiosity-Driven Exploration through Multi-Agent Contextual Calibration2025