cd/entity/Compressed Sparse Attention· home› entities› Compressed Sparse Attention

grep -l @compressed sparse attention /news/*.json | wc -l → 2

Compressed Sparse Attention

mentions 2 type Person feed RSS

// recent coverage 2 mentions

03:36

2026-09-14

ianbarber.blog

large-language-models

Agents love prefill

DeepSeek's V4.1 Flash technical report introduces a "Causal Encoder–Decoder" architecture that runs only the first 20 of the model's 40 layers during prefill, cutting prefill compute in half for its 5…

00:00

2026-05-11

together.ai

large-language-models

Serving DeepSeek-V4: why million-token context is an inference systems problem

DeepSeek-V4's million-token context capability stems from a hybrid attention architecture that compresses context before KV storage, reducing cache pressure. Together's early bring-up on NVIDIA HGX B2…

// co-occurs with top 8 entities

DeepSeek-V4 1 Together 1 NVIDIA HGX B200 1 Heavily Compressed Attention 1 Sliding Window Attention 1 Manifold-Constrained Hyper-Connections 1 Muon 1 DeepSeek 1

// topics top 5 topics

large language models 2 ai infrastructure 2 ai research 2 machine learning 2 artificial intelligence 1