Learning Multi-Agent Communication from Graph Modeling Perspective
Authors: Shengchao Hu, Li Shen, Ya Zhang, Dacheng Tao
ICLR 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on a variety of cooperative tasks substantiate the robustness of our model across diverse cooperative scenarios, where agents are able to develop more coordinated and sophisticated strategies regardless of changes in the number of agents. |
| Researcher Affiliation | Collaboration | Shengchao Hu1,2, Li Shen3 , Ya Zhang1,2, Dacheng Tao4 1 Shanghai Jiao Tong University, 2 Shanghai AI Laboratory 3 JD Explore Academy, 4 Nanyang Technological University |
| Pseudocode | Yes | The overall pseudocode is presented in Algorithm 1. |
| Open Source Code | Yes | Our code is available at: https://github.com/charleshsc/Comm Former |
| Open Datasets | Yes | We conduct a series of experiments using four environments, including Predator-Prey (PP) (Singh et al., 2018), Predator Capture-Prey (PCP) (Seraj et al., 2022), Star Craft II Multi-Agent Challenge (SMAC) (Samvelyan et al., 2019), and Google Research Football(GRF) (Kurach et al., 2020). |
| Dataset Splits | No | The paper mentions 'training Ltrain and validation Lval' as part of its bi-level optimization formulation, but it does not explicitly provide specific percentages, sample counts, or methodology for dataset splits for training, validation, or testing. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU models, CPU models, or memory specifications used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers (e.g., 'PyTorch 1.9', 'Python 3.8', or specific solver versions). |
| Experiment Setup | Yes | We provide detailed training figures (Figure 6) for various methods to substantiate our claim that our approach facilitates simultaneous optimization of the communication graph and architectural parameters in an end-to-end manner, all while preserving sample efficiency. The hyper-parameters used in the study and additional detailed results can be found in the Appendix. and Table 2: Common hyper-parameters used for our method in the experiments. and Table 3: Specific hyper-parameters used for our method in the experiments. and Table 4: Different hyper-parameters used for Comm Former in different tasks. |