cd/entity/DTP· home› entities› DTP
grep -l @dtp /news/*.json | wc -l → 1

@DTP

mentions 1 type Organization feed RSS
16:16
2026-05-28
blog.kog.ai
large-language-models

Delayed Tensor Parallelism for Faster Transformer Inference

Kog Team researchers introduced Delayed Tensor Parallelism (DTP), a Transformer architecture that hides communication overhead behind computation and weight streaming to accelerate batch-size-one infe…

// co-occurs with top 4 entities
AMD 1 NVIDIA 1 Delayed Tensor Parallelism 1 Transformer 1