Search papers, labs, and topics across Lattice.
Affiliation:
1
0
3
2
Achieving the accuracy of a 12-layer GPT-2 Small with just a 3-layer Gated Recurrent Transformer showcases a groundbreaking efficiency in model design.
