MAPF-GPT and MAPF-GPT-DDG
MAPF-GPT-DDG is a specialized extension of MAPF-GPT that adds decentralized multi-agent capability and Delta D fine-tuning, making them complementary tools where the latter builds upon the former's imitation learning foundation rather than competing alternatives.
About MAPF-GPT
CognitiveAISystems/MAPF-GPT
[AAAI-2025] This repository contains MAPF-GPT, a deep learning-based model for solving MAPF problems. Trained with imitation learning on trajectories produced by LaCAM, it generates actions under partial observability without heuristics or agent communication. MAPF-GPT excels on unseen instances and outperforms learnable state-of-the-art solvers
About MAPF-GPT-DDG
Cognitive-AI-Systems/MAPF-GPT-DDG
[IROS-2025] MAPF-GPT-DDG is a scalable decentralized multi-agent pathfinding (MAPF) solver based on imitation learning. It builds upon MAPF-GPT by introducing a novel fine-tuning method called Delta Data Generation (DDG) — a reward-free active learning approach that identifies and corrects failure cases in the policy.
Scores updated daily from GitHub, PyPI, and npm data. How scores work