1 paper · 1 filter
Mattia Opper, N. Siddharth
We present Banyan, a model that efficiently learns semantic representations by leveraging explicit hierarchical structure. While transformers excel at scale, they struggle in low-r…