↻
Mike Frank reposted
@mitpress.bsky.social
"[Babies] only splash about in the shallows of a fathomless ocean of language. And yet, somehow, that’s enough. From a drop, they infer the depths." To understand how children learn human language, @mcxfrank.bsky.social and colleagues studied the experience from a child's pers…
AI Weekly's analysis
→
- A pre-teen has heard about 100 million words; Meta's Llama 3.1 trained on 15 trillion tokens, roughly 150,000 times more.
- The 2024 BabyLM champion, GPT-BERT, beat Meta's Llama 2 70B on some benchmarks despite training on 15,000 times less data.
- Princeton's Brenden Lake trained a model on 61 hours of SAYCam egocentric video and got object recognition without innate biases.
Read full analysis →
View on Bluesky →