Search papers, labs, and topics across Lattice.
1
0
2
RTRL's computational bottleneck can be slashed by >90% without sacrificing performance, thanks to massive redundancy in the recurrent Jacobian.
