Skip to content
Synced · WeChat

Recent LLM Architecture Changes: From Gemma 4 to DeepSeek V4

LLM近期重大架构进化一览:从Gemma 4到DeepSeek V4

Jiqizhixin translated Sebastian Raschka’s blog on recent LLM architecture changes, covering long-context cost reductions in Gemma 4, Laguna XS.2, and ZAYA1-8B; the article states that Gemma 4 E2B saves about 2.7GB of KV cache at 128K context with bfloat16 precision.

Why it matters: Raschka walks through architecture changes in Gemma 4 through DeepSeek V4, all aimed at cutting long-context memory cost.

Read the original ↗Export Markdown