Unlocking Generative AI: ServiceNow, Hugging Face, and NVIDIA Launch New Open-Access LLMs for Enterprise Application Development

 

ServiceNow (NYSE: NOW), Hugging Face, and NVIDIA have unveiled StarCoder2, a groundbreaking series of open-access large language models designed to revolutionize code generation. Developed in collaboration with the BigCode Community, managed by ServiceNow, and Hugging Face, StarCoder2 promises unparalleled performance, transparency, and cost-effectiveness.

Trained across a staggering 619 programming languages, StarCoder2 empowers developers with an array of capabilities, including application source code generation, workflow creation, text summarization, and more. With its advanced features like code completion and snippets retrieval, StarCoder2 accelerates innovation and enhances productivity.

Offering three model sizes – a 3-billion-parameter model by ServiceNow, a 7-billion-parameter model by Hugging Face, and a 15-billion-parameter model by NVIDIA – StarCoder2 caters to diverse computing needs while delivering exceptional performance. Remarkably, the smaller variants match the performance of the original 15-billion-parameter model, while significantly reducing compute costs.

Harm de Vries, lead of ServiceNow’s StarCoder2 team, highlighted the model's contribution to open scientific collaboration and responsible AI practices, emphasizing its potential to elevate developer productivity and facilitate organizational growth.

Leandro von Werra, machine learning engineer at Hugging Face, emphasized the significance of open-source initiatives in democratizing responsible AI, while Jonathan Cohen, vice president of applied research at NVIDIA, underscored the transformative impact of code LLMs in driving efficiency and innovation across industries.

Through this collaborative effort, StarCoder2 aims to empower developers worldwide with accessible and accountable generative AI, fueling breakthroughs and advancing the global community.