Track genomics, neuroscience and molecular biology.
Search papers, labs, and topics across Lattice.
Institute of Advanced Intelligence and Computing (IAIC), A*STAR, Singapore
2
0
5
Pre-trained Softmax ViTs can directly bootstrap linear-attention models without expensive pre-training from scratch: simply copy the operator-agnostic MLP weights and distill the attention routing behavior.
Noisy labels tank dynamic pruning performance, but AlignPrune's loss-trajectory alignment recovers up to 6.3% accuracy without architecture or training changes.