Google DeepMind, Unsplash
ServiceNow and Hugging Face announced the release of StarCoder, a 15 billion parameter open access large language model (LLM) for code generation.
This project is the next milestone in the BigCode project, an initiative led by ServiceNow Research and Hugging Face, with the support of the open scientific AI research community to develop AI systems for code in an open and responsible manner. develop.
Launched on the Hugging Face platform, StarCoder LLM makes generative AI more transparent and accessible to enable responsible innovation at scale, empowering developers in organizations of all sizes to ensure proper governance, safety, and compliance. The goal is to enable you to take advantage of generative AI using protocols.
Covering over 80 programming languages in BigCode’s The Stack v1.2 dataset, StarCoder can be deployed to bring generative AI to your applications with text-to-code, text-to-workflow capabilities, and more. Professional software engineers can tackle complex programming challenges and Citizen he developers can build new software regardless of technical ability.
This launch follows a growing number of AI capabilities announced for the Now Platform, including a recent release in Utah.
“ServiceNow’s collaboration with Hugging Face extends our long-standing commitment to AI excellence,” said Harm de Bruce, head of the Large Language Models Lab at ServiceNow Research and co-head of BigCode Vries said. “New responsible AI practices for training and sharing language models at scale are essential to providing our customers with the right protocols, safeguards and permissive licenses, and StarCoder makes this possible. .”
Leandro von Werra, Machine Learning Engineer at Hugging, said: Face and co-leader of BigCode. “This initiative is a testament to the potential of open source as we work towards democratizing AI.”
