Nodes/ComfyUI-INT4-Fast/Load Diffusion Model INT4 (W4A4)
ComfyUI Node

Load Diffusion Model INT4 (W4A4)

Load and Quantize INT4 models with fast comfy-kitchen GPU execution.

By viralvfx·Created about a month ago·Updated about a month ago· 36
Load Diffusion Model INT4 (W4A4)
  • pre_lora
  • MODEL
unet_name
weight_dtype
model_type
on_the_fly_quantizationfalse
enable_convrottrue
lora_modeNone
Categoryloaders

Inputs (7)

NameTypeDefaultDescription
unet_nameCOMBO0 options:
weight_dtypeCOMBOCompute dtype. Default follows the model dtype.
model_typeCOMBOOnly used for on the fly quantization, to filter sensitive layers.
on_the_fly_quantizationBOOLEANfalseQuantize a higher precision model to INT4. If the selected model is already INT4 keep unchecked.
enable_convrotBOOLEANtrueEnable ConvRot for better quantization.
lora_modeCOMBONoneNone bakes LoRA patches with normal rounding which is the default behavior. Stochastic bakes with stochastic rounding. Dynamic applies LoRA at inference time.
pre_loraoptPRE_LORA

Outputs (1)

NameTypeDescription
MODELMODEL