Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Make --novram completely offload weights.

comfyanonymouse44d0ac7f778
diff --git a/comfy/ldm/pixart/blocks.py b/comfy/ldm/pixart/blocks.py
index 40b0663e5..2225076e5 100644
--- a/comfy/ldm/pixart/blocks.py
+++ b/comfy/ldm/pixart/blocks.py
@@ -6,7 +6,6 @@ import torch.nn as nn
import torch.nn.functional as F
from einops import rearrange
-from comfy import model_management
from comfy.ldm.modules.diffusionmodules.mmdit import TimestepEmbedder, Mlp, timestep_embedding
from comfy.ldm.modules.attention import optimized_attention
diff --git a/comfy/model_management.py b/comfy/model_management.py
index b480aaaa4..d77ae8c06 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -521,7 +521,7 @@ def load_models_gpu(models, memory_required=0, force_patch_weights=False, minimu
lowvram_model_memory = 0
if vram_set_state == VRAMState.NO_VRAM:
- lowvram_model_memory = 64 * 1024 * 1024
+ lowvram_model_memory = 0.1
loaded_model.model_load(lowvram_model_memory, force_patch_weights=force_patch_weights)
current_loaded_models.insert(0, loaded_model)