Search papers, labs, and topics across Lattice.
Mohamed, Zayed University of Artificial Intelligence (MBZUAI)
4
0
7
Interleaving expensive Muon spectral steps with cheap Lion sign descent not only cuts distributed training wall-clock time by 33%, but actually beats full Muon and AdamW on final perplexity.
Strong adversarial robustness can emerge without generating a single adversarial example during training, reaching 76.6% AutoAttack accuracy on CIFAR-10 purely through oscillatory neural dynamics and predictive self-supervision.
Orthogonal obfuscation slashes token recovery rates from LLM inference to just 1.3%, safeguarding sensitive data without compromising performance.
LoRA fine-tuning just got a memory-efficient upgrade: LoRSum matches or beats standard LoRA performance by reformulating optimization as a proximal problem and using diagonal K-FAC approximations, all without expensive SVD.
