claim
In certain scenarios, the performance of the 'Think-on-Graph' (ToG) approach using small large language models can exceed that of large models like GPT-4, thereby reducing the cost of LLM deployment and application.

Authors

Sources

Referenced by nodes (3)