Search papers, labs, and topics across Lattice.
2
0
4
11
LLM training bottlenecks? ZipCCL achieves up to 1.18x end-to-end speedups by losslessly compressing communication collectives, without sacrificing model quality.
Lossless compression can actually *speed up* LLM inference on GPUs, not just shrink model size, thanks to ZipServ's hardware-aware design.
