Search papers, labs, and topics across Lattice.
1
0
2
8
Gaze-only models can be significantly improved by injecting lightweight language signals, achieving up to a 2.9 percentage point gain in reading comprehension prediction accuracy.
