Rather than relying on traditional mathematical optimizers, OPRO uses “meta problems” defined in human language to form the basis of the optimization process. LLM then generates candidate solutions based on the problem definition and previous solutions and tests them using quality scores.
Interestingly, specific encouraging phrases such as “Take a deep breath and try to solve this problem step by step” significantly improve results. When tested using the Google PaLM 2 language model, this phrase achieved the highest accuracy of 80.2% in solving school math problems.
This approach allows you to use the LLM’s extensive language database, resulting in more useful and accurate results. OPRO could lead to improvements in how people interact with LLMs and get help from them in a variety of applications.
Source: Ferra
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
