Hacker News new | past | comments | ask | show | jobs | submit login

I think inventing Generative Adversarial Networks, which enable everything from Alexa to GPT-3 to DALL-E to thousands of other models we use today, is pretty visionary to the field of ML.

I’m sure you have some nitpick but I don’t think it matters




What do GANs have to do with GPT-3? Isn't GPT a Transformer-based autoregressive model?


Nothing directly. One could argue that GANs did kick-start the boom in generative model research as the first approach in some time that worked in a big way. Either way, certainly Ian is a huge name in ML.


> i'm sure you have some nitpick




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: