Search papers, labs, and topics across Lattice.
1
0
2
5
Uncovers the inner workings of transformer-based time series classifiers, revealing how specific attention heads and timesteps causally drive correct classifications.
