Research map / Multi-agent systems and games / Cooperative multi-agent RL
DLM: Unified Decision Language Model for Offline Multi-Agent Sequential Decision Making
Zhuohui Zhang, Bin Cheng, Bin He
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add GRPO with return-aware and executability reward after SFT | DLM-SFT · SMAC corridor | test win rate: 0.81 -> 0.92 | Improved |
| add GRPO after SFT | DLM-SFT · SMAC 3s_vs_5z | test win rate: 0.84 -> 0.94 | Improved |
| add GRPO after SFT | DLM-SFT · SMAC 5m_vs_6m | test win rate: 0.72 -> 0.80 | Improved |
| add GRPO after SFT | DLM-SFT · SMAC zero-shot zerg_5_vs_5 | win rate: 0.42 -> 0.58 | Improved |
| add GRPO after SFT | DLM-SFT · SMAC zero-shot 3m | win rate: 0.90 -> 0.93 | No clear effect |
| apply GRPO preference optimization | DLM-SFT · all SMAC tasks | OOD action rate | Improved |
| use dialogue-style inter-agent context with pre-trained LLM instead of MADT | DLM-SFT vs MADT · SMAC zero-shot | win rate: 0.00 -> 0.59 (protoss_5_vs_5) | Improved |
| use instruction-tuned LLM zero-shot (SayCan) instead of fine-tuning | SayCan · SMAC | win rate: 0.00 (corridor) | Hurt |
| use LoRA instead of full fine-tuning in SFT stage (ablation) | DLM-SFT · multi-task SMAC | representation capacity | Hurt |
Closest papers by meaning
- $\texttt{Multi}^2$: Hierarchical Multi-Agent Decision-Making with LLM-Based Agents in Interactive Environments (ICML 2026 · similarity 0.89)
- On the Modeling Capabilities of Large Language Models for Sequential Decision Making (ICLR 2025 · similarity 0.89)
- Divide and Conquer: Grounding LLMs as Efficient Decision-Making Agents via Offline Hierarchical Reinforcement Learning (ICML 2025 · similarity 0.88)
- Enhancing Decision-Making for LLM Agents via Step-Level Q-Value Models (AAAI 2025 · similarity 0.88)
- Efficient Reinforcement Learning with Large Language Model Priors (ICLR 2025 · similarity 0.88)
- From Words to Actions: Unveiling the Theoretical Underpinnings of LLM-Driven Autonomous Systems (ICML 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
