- By Ashish Singh
- Fri, 13 Sep 2024 11:41 AM (IST)
- Source:JND
The "Strawberry" series of AI models, which OpenAI formally unveiled, are made to take longer to process query responses in order to tackle challenging issues. According to a blog post Microsoft-backed AI giant, the models can reason through complicated tasks and tackle more difficult challenges in arithmetic, science, and coding than earlier models.
OpenAI labelled the models it revealed on Thursday as o1 and o1-mini, but internally referred to the project by the code name Strawberry. According to the company, the o1 will be accessible through ChatGPT and its API on Thursday.
In a post on the social media site X, OpenAI researcher Noam Brown, who works to enhance reasoning in the company's models, verified that the models were the same as the Strawberry project.
READ: OpenAI Set To Launch 'Strawberry' Model With Advanced Problem-Solving, Reflective Responses
"I'm excited to share with you all the fruit of our effort at OpenAI to create AI models capable of truly general reasoning," Brown wrote.
According to OpenAI's blog post, the o1 model performed better than its previous model, GPT-4o, with an 83% score on the International Mathematics Olympiad qualifying exam. According to the business, the model also outperformed humans in terms of accuracy on a benchmark of scientific challenges and enhanced performance on competitive programming questions.
According to Brown, the models achieved the scores by using a method called "chain-of-thought" reasoning, which divides difficult issues into more manageable logical steps.
Researchers have found that when the method is applied as a prompting strategy, AI model performance on challenging problems typically improves. Now that OpenAI has automated this feature, models are able to solve problems without assistance from the user.
READ: Ex-OpenAI Chief Ilya Sutskever’s Startup SSI Raises $1B To Pioneer Safe Superintelligence
"We educated these models to think through issues more thoroughly before responding, much like a human would. They learn to hone their thought processes, experiment with various approaches, and own up to their failures through training," according to OpenAI.
In November 2023, OpenAI's progress on the reasoning project—then known as Q*—was first covered by Reuters. In July, it was revealed that the project had adopted the name Strawberry.