Search papers, labs, and topics across Lattice.
MIT MIT, IBM Watson AI Lab
1
0
2
Nonuniform width allocation in transformers can lead to a 22% reduction in FLOPs while enhancing language modeling performance.