search

MinusZoneAI / ComfyUI-CogVideoX-MZ

CogVideoX-5B 4-bit quantization model
GNU General Public License v3.0
91 stars 3 forks source link

weight_dtype as GGUF? #17

Open clemarjonlaguardia opened 1 month ago

clemarjonlaguardia commented 1 month ago

How can I set the weight_dtype as GGFU? Because it doesn't appeared in the list. image

wailovet commented 1 month ago

No need to set, it will be used automatically, because only some of the layers use Q4 quantization, and the rest of the layers use the data type in weight_dtype