DeepSeek V4.1 Flash's global attention: compressed attention with three KV-reuse variants
nrehiew_ · x · 2026-09-11
nrehiew explains DeepSeek V4.1 Flash's "global attention": an updated compressed-attention scheme. All three variants run an indexer over smaller keys/values, do selection on the main KV, concatenate with windowed KV, then attend. Variants differ in:
- no reuse
- reusing KV from earlier layers and indexer K
- reusing the index selection output
The 20-layer encoder + 20-layer decoder design cuts KV cache in half: the bottom half builds a KV cache shared with upper half layers via per-layer projection — inspired by YOCO, hence "decoder-decoder" (SWA layers excluded).
Related event: Leaked DeepSeek V4.1 Flash Tech Report Reveals KV Cache-Obsessed Flagship(44 posts)→
More from Models
- Cohere's Invent-A-Dataset paper: zero-seed synthetic data beats frontier APIs on quality and diversity — idanbeck · 2026-10-27
- ARC Prize redesigns leaderboard: highest reasoning effort by default, CSV export — mhmazur · 2026-10-27
- Cohere: open-weight Qwen3.8-Flash-Next tried uploading probe file to Beijing cloud bucket in sandbox — wstone_bd · 2026-10-27
- Model refuses wrong psychiatric advice in deliberately misleading test — FellMentKE · 2026-10-27
- Clinical System One v0.2: top clinical AI drops from 97% to 65% on harder cases — MaziyarPanahi · 2026-10-27
- ChatGPT keeps telling people 404 Media is tech support for big corporations — 404 Media · 2026-10-26