XNews

OpenAI's AI models now automate training of new experimental models, including GPU kernels

According to The Information, OpenAI's internal AI models can now automate the training process for new experimental models, including writing GPU kernel code and optimizing training code. Researchers only need to provide an optimization example, and the AI spends weeks implementing and testing similar improvements. Multiple AI agents now collaborate to solve problems without human involvement, reducing experiment time from years to about a week.

Source posts