Shapelearn Runs Qwen 3.8 27B in Just 13.1 GB of VRAM
syntaxing · hn · 2026-09-18
A Byteshape blog post presents Shapelearn, a setup that runs the Qwen 3.8 27B model within only 13.1 GB of VRAM, an exercise in quantization and memory optimization for local deployment.
More from Infra
- Google engineers: LLM benchmark harnesses silently drop requests — 200 QPS in, 38 out — AI Engineer · 2026-09-20
- The rig built to run Emacs and doomscroll X is now worth more than its owner's car — tetsuoai · 2026-09-19
- Apple M4 sustains 10 instructions per cycle, beating most rivals; M5 speedup explained — lemire · 2026-09-19
- Apple M6 bumps cores to 12 with two super cores; CPUs keep improving fast — lemire · 2026-09-19
- Apple M-series chips gained ~50% Geekbench 6 performance over three years — lemire · 2026-09-19
- Inside OpenAI's inference routing: why the proportional controller had to go — AI Engineer · 2026-09-19