Shapelearn Qwen 3.8 27B (13.1 GB VRAM)

Beyond a Single Number: Evaluating Quantized Models for Deployment

A 30B Qwen model walks into a Raspberry Pi and runs in real time