Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
2
0
2
Post-training pruning of Muon-trained Transformers can be salvaged by a novel optimizer that budgets spectral geometry, yielding width-minimal representations and superior pruning criteria.
Forget about trace norms – generalization in chaotic training regimes hinges on the intricate structure of the full Hessian spectrum and its partial determinants.