OpenAI developing new reasoning technology named Strawberry to enhance capabilities of large language models.

From Mint: 2024-07-12 22:55:10

OpenAI is developing a new reasoning technology called Strawberry for its large language models, aiming to enhance their capabilities. Strawberry, previously known as Q*, is a closely guarded secret within the organization, showing promise in tackling complex science and math questions. The project aims to enable AI to navigate the internet autonomously and conduct in-depth research.

Strawberry represents a specialized approach to post-training AI models after initial dataset training. OpenAI plans to use Strawberry for long-horizon tasks that require extended planning and action. The goal is for AI models to autonomously browse the web and conduct research with a computer-using agent for decision-making based on findings.

OpenAI aims to push the boundaries of AI capabilities with Strawberry by improving reasoning skills and autonomy. The technology could potentially revolutionize how AI models process and utilize information, particularly in handling complex tasks over an extended period. This development highlights OpenAI’s commitment to advancing AI technology for practical applications.



Read more at Mint: ChatGPT maker OpenAI developing new breakthrough reasoning technology code-named ‘Strawberry’. Why is it important?