Search papers, labs, and topics across Lattice.
This study investigates how Graph Language Models (GLMs) process graph tokens and their effectiveness in conveying graph structure within Large Language Models (LLMs). The authors reveal that while graph sink tokens exhibit high activation levels, they do not serve as the primary carriers of graph information, indicating a disconnect between token saliency and semantic utility. Through various interventions, the research highlights critical limitations in current graph-token mechanisms, suggesting that existing methods may not yield optimal topology-aware representations for downstream tasks.
Graph sink tokens may light up the neural network, but they don't carry the graph's true meaning鈥攔evealing a critical flaw in how GLMs interpret graph information.
Graph Language Models (GLMs) have become a promising direction for adapting Large Language Models (LLMs) to graph learning tasks. By transforming graph topology and node information into graph tokens, GLMs allow LLMs to jointly process structured graph inputs and textual instructions. Yet, it remains unclear how LLMs internally interpret these graph tokens and whether graph tokens act as meaningful carriers of graph structure. In this work, we analyze how LLMs process graph information through graph-token behavior in representative GLM architectures. Findings. We find that the internal saliency of graph tokens in GLMs is not equivalent to graph information utilization. Graph sink tokens consistently emerge as activation-level outliers: they can be identified by massive activation values along a small set of hidden-state dimensions and are biased toward early graph-token positions. However, this activation-level saliency does not imply that these tokens are the main carriers of graph information. Unlike classical attention sinks in language and vision-language models, graph sink tokens do not necessarily attract the largest attention weights from query tokens. Through pruning, repositioning, and swapping interventions, we show that graph sink tokens are not the most important semantic or structural tokens for downstream prediction. Implications. Together, these results suggest that after current GLMs map graph structure into the LLM token space, the resulting graph-token representations do not naturally form a fully usable topology-aware internal representation; instead, they exhibit a decoupling between activation-level saliency and graph-semantic utility. This decoupling points to limitations in existing graph-token construction, placement, and alignment mechanisms.
