The AI paper, “Visual Autoregressive Modelling: Scalable Image Generation via Next-Scale Prediction”, examined a unique approach to generating images at a faster speed with better quality and scalability. The award is expected to be given at the conference in Vancouver, Canada, from December 10 to 15.
It was one of the more than 50 research works from TikTok owner ByteDance that was accepted at this year’s NeurIPS. The other studies from the Beijing-based firm covered topics ranging from natural language processing, computer vision, machine learning and other subjects related to AI models.
The paper’s lead author is Tian Keyu, who bears the same surname as the ex-intern – identified only as “Tian” – who was dismissed by ByteDance in August. In its lawsuit before the Haidian District Court in Beijing, ByteDance is demanding 8 million yuan (US$1.1 million) in compensation as well as a public apology.
Neither ByteDance nor the paper’s lead author replied to a request for comment on Wednesday.
