mirror of
https://github.com/comfyanonymous/ComfyUI.git
synced 2025-09-14 21:45:06 +00:00
Empty cache after model unloading for normal vram and lower.
This commit is contained in:
@@ -238,6 +238,8 @@ def unload_model():
|
||||
current_loaded_model.model_patches_to(current_loaded_model.offload_device)
|
||||
current_loaded_model.unpatch_model()
|
||||
current_loaded_model = None
|
||||
if vram_state != VRAMState.HIGH_VRAM:
|
||||
soft_empty_cache()
|
||||
|
||||
if vram_state != VRAMState.HIGH_VRAM:
|
||||
if len(current_gpu_controlnets) > 0:
|
||||
|
Reference in New Issue
Block a user