Generation Domain
Image generation is a rapidly advancing field focused on creating realistic and diverse images from various inputs, such as text descriptions or other images. Current research emphasizes scaling up model architectures like autoregressive models and transformers, exploring efficient training methods to overcome limitations in video generation and address issues like model collapse from self-consuming training loops. This work is significant for its potential to improve image editing, create new forms of visual content, and raise awareness of privacy concerns related to the large datasets used in training these powerful models.
Papers
November 14, 2024
October 11, 2024
June 10, 2024
February 22, 2024
November 28, 2023
June 25, 2023
December 8, 2022
October 3, 2022
April 18, 2022