Hacker News new | past | comments | ask | show | jobs | submit
ycui7 | on: Qwen 3.8 27B
if you have the VRAM, use offical release. quantized model lose focus after long context and can do damages or thinking loop