Kakao's Applied AI Model team has open-sourced "Kanana-2-v-decide," a model that understands images and text together to make decisions. Researchers Jung Seong-hyo and Lee Ju-young posted an introduction on Kakao's tech blog.
The team says real services often need a clear conclusion more than a long explanation. Classifying the type of a customer inquiry, examining a product photo, or choosing which tool an agent should use next are examples.
An agent repeats choices like these many times while handling a single request. So the model's goal is to quickly get the conclusion needed at each step and hand it on to the service's next action.
The blog post doesn't give many details, such as performance figures or the license. Specifics will have to be confirmed in materials Kakao releases separately.
Source: Kakao Tech Blog
This article was newly written by Raon Community based on the facts in the original source. Copyright of the original belongs to its original author; please see the source for details.