ServiceNow (NYSE: NOW), the leading digital workflow company making the world work better for everyone, today announced the release of one of the world’s most responsibly developed and strongest-performing open-access large language model (LLM) for code generation. Led by ServiceNow Research and Hugging Face, the open-access, open-science, and open-governance 15 billion parameter StarCoder LLM makes generative AI more transparent and accessible to enable responsible innovation at scale.
The StarCoder model is designed to level the playing field so developers from organizations of all sizes can harness the power of generative AI and maximize the business impact of automation with the proper governance, safety, and compliance protocols. This new LLM marks the next major milestone in the BigCode Project, an ambitious initiative to develop state-of-the-art AI systems for code in an open and responsible manner with the support of the open-scientific AI research community.
“ServiceNow’s collaboration with Hugging Face expands our longstanding commitment to AI excellence,” said Harm de Vries, lead of the Large Language Model Lab at ServiceNow Research and co-lead of BigCode. “New, responsible AI practices to train and share large language models are vital to ensuring the right protocols, safeguards, and permissive licenses are in place for our customers, and StarCoder is making this possible.”
“The joint efforts led by Hugging Face and ServiceNow enable the release of powerful base models that empower the community to build a wide range of applications more efficiently than a single company could come up with,” said Leandro von Werra, machine learning engineer at Hugging Face and co-lead of BigCode. “This endeavor is a testament to the potential of open-source as we work toward democratizing AI.”
Trained with a trillion tokens of permissively licensed source code covering over 80 programming languages from BigCode’s The Stack v1.2 dataset, StarCoder can be deployed to bring pair-programing like generative AI to applications with capabilities like text-to-code and text-to-workflow. With this, StarCoder gives professional software engineers the power to tackle the most complex programming challenges and empowers citizen developers to build new software regardless of technical ability—accelerating AI innovation at scale. The model will be released with open-access on the Code Open RAIL-M license to permit royalty-free distribution. Unlike traditional open-source software released without use case restrictions, BigCode releases the model with a responsible AI model license that includes use case restrictions that apply to modifications of the model, and applications using the model - for example, to restrict the models from being used to generate or distribute malicious code to harm electronic systems. Supporting code has been open sourced on the BigCode project’s GitHub.