Influence of Leaderboard and Trial Space on Large Language Models Popularity

Large language models (LLMs) represent advanced AI systems capable of generating human-like text. This study investigates whether the presence of a leaderboard and availability of trial space contribute to an increase in a LLM’s popularity. It uses longitudinal data on over 9,487 LLMs from the Hugging Face (HF) platform, which serves as a central hub for developers and researchers, facilitating the sharing, access, and collaboration of a wide range of LLMs. Study findings reveal that both the leaderboard and trial space on HF enhance its popularity, where the magnitude of this effect varies depending on attributes such as model maintenance and model type. This research contributes to literature on LLMs and offers guidance to platforms on optimizing model design and enhancing functions, while informing policymakers on regulation and support for the rapidly growing LLM ecosystem. Limitations and directions for future research are discussed.