Search papers, labs, and topics across Lattice.
2
2
3
4
ELDR slashes median latency by up to 13.9% for MoE models by intelligently routing requests based on expert activation signatures.
Train massive models on unproven hardware with confidence: SIGMA achieves 94% accelerator utilization and trains a 200B MoE with near-perfect stability on early-life AI accelerators.
