Search papers, labs, and topics across Lattice.
Argonne National Laboratory
3
0
6
4
Cerebras CS-3 can deliver 100x speedups over CPU for sparse matrix multiplication at 90% sparsity, but surprisingly, becomes *slower* than CPU beyond 99% sparsity.
Fine-tuning LLMs in federated settings just got easier: SplitFT lets clients adapt their cut layers and LoRA ranks, boosting performance and slashing communication costs.
Achieve orders-of-magnitude speedups in topology-aware lossy compression for scientific data with TopoSZp, without sacrificing compression ratios or introducing false positives in critical point detection.
