OpenAI said on Thursday it was launching its "Strawberry" series of AI models designed to spend more time processing answers to queries in order to solve hard problems. The models are capable of reasoning through complex tasks and can solve more challenging problems than previous models in science, coding and math, the AI firm said in a blog post. OpenAI used the code name Strawberry to refer to the project internally, while it dubbed the models announced on Thursday o1 and o1-mini. The o1 will be available in ChatGPT and its API starting Thursday, the company said. ChatGPT has struggled to recognize that the word "strawberry" contains three instances of the letter R. Noam Brown, a researcher at OpenAI focused on improving reasoning in the company's models, confirmed in a post on X that the models were the same as the Strawberry project.

The objective of meta-learning is to exploit the knowledge obtained from observed tasks to improve adaptation to unseen tasks. As such, meta-learners are able to generalize better when they are trained with a larger number of observed tasks and with a larger amount of data per task. Given the amount of resources that are needed, it is generally difficult to expect the tasks, their respective data, and the necessary computational capacity to be available at a single central location. It is more natural to encounter situations where these resources are spread across several agents connected by some graph topology. The formalism of meta-learning is actually well-suited to this decentralized setting, where the learner would be able to benefit from information and computational power spread across the agents. Motivated by this observation, in this work, we propose a cooperative fully-decentralized multi-agent meta-learning algorithm, referred to as Diffusion-based MAML or Dif-MAML. Decentralized optimization algorithms are superior to centralized implementations in terms of scalability, avoidance of communication bottlenecks, and privacy guarantees. The work provides a detailed theoretical analysis to show that the proposed strategy allows a collection of agents to attain agreement at a linear rate and to converge to a stationary point of the aggregate MAML objective even in non-convex environments. Simulation results illustrate the theoretical findings and the superior performance relative to the traditional non-cooperative setting.