Search papers, labs, and topics across Lattice.
The paper introduces GEM, a Graph-Enhanced Mixture-of-Experts framework for Dialogue State Tracking (DST) that combines a Graph Neural Network, a finetuned T5-Small model, and ReAct agents. GEM dynamically routes between these specialized experts based on dialogue context, leveraging the GNN for structured dialogue representation and T5-Small for sequence modeling. Results on MultiWOZ 2.2 show GEM achieves 65.19% Joint Goal Accuracy, significantly outperforming end-to-end LLMs and existing state-of-the-art DST methods.
LLMs get schooled in dialogue state tracking by a mixture-of-experts architecture that uses a graph neural network and ReAct agents to achieve state-of-the-art results with a T5-Small backbone.
Dialogue State Tracking (DST) requires precise extraction of structured information from multi-domain conversations, a task where Large Language Models (LLMs) struggle despite their impressive general capabilities. We present GEM (Graph-Enhanced Mixture-of-Experts), a novel framework that combines language models and graph-structured dialogue understanding with ReAct agent-based reasoning for superior DST performance. Our approach dynamically routes between specialized experts: a Graph Neural Network that captures dialogue structure and turn-level dependencies, and a finetuned T5-Small encoder-decoder for sequence modeling, coordinated by an intelligent router. For complex value generation tasks, we integrate ReAct agents that perform structured reasoning over dialogue context. On MultiWOZ 2.2, GEM achieves 65.19% Joint Goal Accuracy, substantially outperforming end-to-end LLM approaches (best: 38.43%) and surpassing state-of-the-art (SOTA) methods including TOATOD (63.79%), D3ST (58.70%), and Diable (56.48%). Our graph-enhanced mixture-of-experts architecture with ReAct integration demonstrates that combining structured dialogue representation with dynamic expert routing and agent-based reasoning provides a powerful paradigm for dialogue state tracking, achieving superior accuracy while maintaining computational efficiency through selective expert activation.
