Search papers, labs, and topics across Lattice.
INRAE, Univ Montpellier
1
0
3
Increased overcompleteness in sparse autoencoders can actually reduce their interpretability, challenging assumptions about model complexity.
