Search papers, labs, and topics across Lattice.
Affiliation:
2
0
4
Training data attribution on 32B-parameter models no longer requires intractable gradient storage: extreme 1-bit eigenbasis projection achieves competitive influence estimation accuracy at over 100脳 lower per-example footprint than standard baselines.
Optimal transport provides a surprisingly tight and efficiently computable bound on transductive generalization in graph node classification, revealing how GNN depth impacts representation geometry.
