▲ 2 pointsDelayed Tensor Parallelism for Faster Transformer Inferenceblog.kog.aiby matt_d·2mo ago·0 comments·view on hn ↗