Search papers, labs, and topics across Lattice.
University of Liverpool
1
0
2
Jointly optimizing action predictions and state transitions leads to significant performance gains in GUI agent training, outperforming conventional methods.
