OpenAI’s AI System Can Predict How to Complete Half an Image
OpenAI’s AI system can accept half an image and predict how to complete it. After GPT-2, a San Francisco–based research lab’s AI system, could write convincing passages of English after being fed the beginning of a sentence or paragraph, the lab experimented with imagery.
By training GPT-2 on pixels instead of words, the model can now accept half an image and predict how to complete it and is called iGPT. The new system uses one-dimensional data and views the images through a single sequence of pixels.
Lead the technologies of the future. Click to know more about the MIT Technology Leadership Program (TLP).