Get the latest tech news
The "it" in AI models is the dataset
’ve been at OpenAI for almost a year now. In that time, I’ve trained a lot of generative models.
As I’ve spent these hours observing the effects of tweaking various model configurations and hyperparameters, one thing that has struck me is the similarities in between all the training runs. Sufficiently large diffusion conv-unets produce the same images as ViT generators. It implies that model behavior is not determined by architecture, hyperparameters, or optimizer choices.
Or read this on Hacker News
