BLOOM: BigScience Large Open-Science Open-Access Multilingual large language model

Large language models (LLMs) have made a significant impact on AI research. These powerful, general models can take on a wide variety of new language tasks from a user’s instructions. However, academia, nonprofits and smaller companies’ research labs find it difficult to create, study, or even use LLMs as only a few industrial labs with the necessary resources and exclusive rights can fully access them. BLOOM is the first multilingual LLM trained in complete transparency — the result of a large collaboration of AI researchers.

With its 176 billion parameters, BLOOM is able to generate text in 46 natural languages and 13 programming languages. For almost all of them, such as Spanish, French and Arabic, BLOOM will be the first language model with over 100B parameters ever created. This has the potential to improve people’s wellbeing in countries where these languages are spoken.

BLOOM is the culmination of a year of work involving over 1000 researchers from 70+ countries and 250+ institutions, leading to a final run of 117 days (March 11 – July 6) training the BLOOM model on the Jean Zay supercomputer in the south of Paris, France thanks to a compute grant worth an estimated €3M from French research agencies CNRS and GENCI.

Researchers can now download, run and study BLOOM to investigate the performance and behavior of recently developed large language models down to their deepest internal operations. More generally, any individual or institution who agrees to the terms of the model’s Responsible AI License (developed during the BigScience project itself) can use and build upon the model on a local machine or on a cloud provider – since it’s embedded in the Hugging Face ecosystem, it’s as easy as importing it with transformers and running it with accelerate. In this spirit of collaboration and continuous improvement, the intermediary checkpoints and optimizer states of the training were also released.