Sergei Plugotarenko, CEO of ANO Tsifrovaya Ekonomia, told socialbites.ca that the Kandinsky 2.1 neural network created by Russian programmers is a global achievement of Russian technologies.
“The competitive features of domestic development are high performance and wide opportunities demanded by users. And also the fact that you can work with Kandinsky absolutely free. And this distinguishes it from foreign analogues, which limit the speed and functionality of services to users who do not have a subscription.
Plugotarenko also added that in just a few days, more than 10 million images were generated using Kandinsky 2.1 and the number of unique users reached 2 million.
“This is not just about its demand in Russia. Separately, I will draw attention to the importance of such developments for Russia’s digital economy – we know that our IT specialists want to take on more interesting first-class tasks. And Kandinsky is definitely one of them,” Plugotarenko added.
Earlier, Sber introduced a new version of the Kandinsky 2.1 generative model, which can generate images from text descriptions in natural language in a matter of seconds. The model speaks 101 languages and opens up new possibilities for creativity: it can mix several drawings, complete the image, create a painting in endless canvas mode.
The neural network took over the weights of the previous version, trained on one billion text-image pairs and also trained on 170 million high-resolution text-image pairs.