Naver's open-source artificial intelligence (AI) model 'HyperCLOVA X Seed' has surpassed a cumulative download of 1 million.
According to the industry on the 20th, HyperCLOVA X Seed surpassed 1 million cumulative downloads just over three months after its public release on the global AI model-sharing platform Hugging Face. This model is a lightweight version of Naver's proprietary large-scale language model (LLM) 'HyperCLOVA X,' which has a strong understanding of the Korean language and culture.
HyperCLOVA X Seed consists of three types with 500 million parameters (0.5B), 1.5 billion parameters (1.5B), and 3 billion parameters (3B), allowing users to choose based on their intended use and resource scale. The largest 3B model has the capability to understand not only text but also images.
The 3B model has been downloaded more than 580,000 times in the past month, particularly gaining popularity. This record places it in the top 0.02% among approximately 1.87 million models registered on Hugging Face. The industry generally considers 300,000 to 400,000 downloads in the first month as a benchmark for popularity.
The background of this popularity is attributed to Naver's open policy. Naver noted, upon releasing the HyperCLOVA X Seed model, that it does so as open-source in hopes that 'other corporations become more powerful' and to create a 'sovereign AI' ecosystem. Consequently, corporations can download the model without expense and modify it to suit their respective services.