OpenAI Presents DALL-E Neural Network

Researchers from OpenAI presented the DALL-E neural network, which creates images from textual descriptions. DALL-E is based on the architecture of GPT-3 neural network and can generate images by processing text requests in English.

DALL-E creates images from text descriptions such as “pentagonal green clock” or “avocado-shaped armchair”.

DALL-E understands complex abstract combinations, can change the style of images, and is also capable of creating realistic photos and non-existent objects.

The neural network is able not only to generate an image but also to complete any rectangular area of ​​an existing image.

The name of the neural network comes from the name of the artist Salvador Dali and the WALL-E robot from Pixar.

Stay tuned with Software Focus!