Search papers, labs, and topics across Lattice.
University of Sydney, San Francisco State University, IIT Madras
1
0
2
Deep transformers can encode complex grammatical structures in low-dimensional spaces, supporting the linear representation hypothesis.