Does it possible to create a version without MTP layer to save some VRAM
#3
by
adonishong
- opened
Appreciate for your work, does it possible to create a version without MTP layer to save some VRAM as described in title?
I have done it with a script generated by Gemini 2.5 Pro. There is no VRAM saving, because vLLM already skips the MTP layer.