AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Is There 10x More Data to Train On?
There's 10x more data to train on than when GPT-3 came out. I don't think the 48 million papers that Galactica was trained on, was something GPT- 3 was trained on. The other company too, I would say again, continue taking a look at is hugging face. One of the key insights is like EMNOB, which is the stock conference in NLP,. And they highlight sort of not research papers, but products, sort of demos that are relevant for research.