Why mlx-vlm not mlx-lm?
#2
by
olegshulyakov
- opened
This version presumably supports vision too, unlike the previous versions which were text only.
This is the full vision model, converted directly from the original Gemma 3n repo. The text-only model should be named differently.
Okay. Why does it has 2 model.safetensors? The size like double 4 bit