How do Scaling Laws Apply to Knowledge Graph Engineering Tasks? The Impact of Model Size on Large Language Model Performance

TL;DR

This paper investigates how the performance of Large Language Models on Knowledge Graph Engineering tasks scales with model size, revealing general applicability of scaling laws but also identifying cases of performance plateaus and cost-effective smaller models.

Contribution

It introduces the LLM-KG-Bench framework for benchmarking LLMs on KGE tasks and analyzes the relationship between model size and performance across 26 models.

Findings

Scaling laws generally apply to KGE tasks.

Performance plateaus occur, favoring smaller models.

Larger models sometimes perform worse within the same family.

Abstract

When using Large Language Models (LLMs) to support Knowledge Graph Engineering (KGE), one of the first indications when searching for an appropriate model is its size. According to the scaling laws, larger models typically show higher capabilities. However, in practice, resource costs are also an important factor and thus it makes sense to consider the ratio between model performance and costs. The LLM-KG-Bench framework enables the comparison of LLMs in the context of KGE tasks and assesses their capabilities of understanding and producing KGs and KG queries. Based on a dataset created in an LLM-KG-Bench run covering 26 open state-of-the-art LLMs, we explore the model size scaling laws specific to KGE tasks. In our analyses, we assess how benchmark scores evolve between different model size categories. Additionally, we inspect how the general score development of single models and families of models correlates to their size. Our analyses revealed that, with a few exceptions, the model size scaling laws generally also apply to the selected KGE tasks. However, in some cases, plateau or ceiling effects occurred, i.e., the task performance did not change much between a model and the next larger model. In these cases, smaller models could be considered to achieve high cost-effectiveness. Regarding models of the same family, sometimes larger models performed worse than smaller models of the same family. These effects occurred only locally. Hence it is advisable to additionally test the next smallest and largest model of the same family.