Meta Releases OPT-175B to Open Up Large Language Model Research
Meta AI introduces OPT-175B, a 175-billion-parameter language model released alongside its training code to democratize access for researchers. The noncommercial release aims to boost reproducible studies on bias, toxicity, and model robustness.
Meta AI releases OPT-175B, a massive 175-billion-parameter language model designed to democratize access to advanced natural language processing technology. Unlike previous large models that remain locked behind paid APIs, this release provides both the pretrained model and the underlying training code to the research community for the first time at this scale.
The company makes this release under a noncommercial license, granting access specifically to academic researchers, government affiliates, civil society members, and industry research laboratories. This targeted approach allows the broader AI community to investigate how these massive models function while maintaining safeguards to prevent misuse.
By sharing OPT-175B and smaller baseline models, Meta aims to foster reproducible research and address critical challenges like bias and toxicity in artificial intelligence. The initiative invites diverse voices across academia, industry, and policy sectors to collaborate on establishing clear guidelines for responsible AI development.