Mac mini connected to SSD freezer and DRAM fridge, illustrating the layered architecture of LLM in a Flash

Cramming a 400B Model into 48GB: The Magic Behind LLM in a Flash

An Apple paper from 2023 made it possible to run a 400 billion parameter model on an ordinary MacBook. The core technologies—MoE and quantization—hide an engineering philosophy built around on-demand loading.

March 24, 2026 · 5 min · 857 words · Dream Beast Programming
Why Go 1.26 didn't include JSON v2

Go 1.26 Shipped Without JSON v2: 18 of 44 Tasks Still Open

Go 1.26 is out, but JSON v2 remains experimental. With 18 of 44 subtasks still open, unresolved performance regressions, and pending Union type support—this isn’t a delay, it’s the Go team building a solid foundation for the standard library.

February 16, 2026 · 6 min · 1242 words · Mengshou Programming
High‑value AI Toolkit Less than a coffee/month →
扫码关注公众号
微信公众号二维码

Weekly Rust / AI tips · Community Q&A · Exclusive perks