YOCO explained: one shared KV cache reused across the model's second half
stochasticchasm · x · 2026-09-11
On the YOCO (You Only Cache Once) architecture, the author clarifies the design is quite simple:
- The first half of the model is standard;
- Hidden states are projected once more to K/Vs, and that identical KV cache is reused by every layer in the second half, slashing cache overhead.
The author also notes the work does acknowledge YoCo, and quips the naming should be YOCO, not YoCo.
Related event: Inside KV-Cache Sharing: How DeepSeek CED and GLM 5.2 Differ(4 posts)→
More from Research
- Researchers identify "Forking": train loss falls while validation loss turns up — ZimingLiu11 · 2026-10-26
- Google DeepMind pilots world's first double-blind evaluations for frontier models — scychan_brains · 2026-10-26
- Rewriting scaling law study: optimal synthetic data share rises from 10% to 32% with compute — heghbalz · 2026-10-26
- FLASH: GPU-native deformable sim learns cloth folding in minutes on one RTX 5090 — ChongZzZhang · 2026-10-26
- LEMON: zero-shot LLM misbehavior monitor flags hallucinations from attention maps — yoavartzi · 2026-10-26
- Behind the EMNLP keynote: LMs as simulated learners to generate child language acquisition hypotheses — najoungkim · 2026-10-26