Remix.run Logo
rimeice 2 days ago

Basically yes, the useful models need a modernish GPU to get inference running at a usable speed. You can get smaller parameter models 3b/7b running on older laptops, it just won’t produce output at a useful speed.