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.

Fix lowvram model merging.

comfyanonymousa57b0c797b06
diff --git a/comfy/controlnet.py b/comfy/controlnet.py
index 5279307c0..4e20100c3 100644
--- a/comfy/controlnet.py
+++ b/comfy/controlnet.py
@@ -257,12 +257,7 @@ class ControlLora(ControlNet):
cm = self.control_model.state_dict()
for k in sd:
- weight = sd[k]
- if weight.device == torch.device("meta"): #lowvram NOTE: this depends on the inner working of the accelerate library so it might break.
- key_split = k.split('.') # I have no idea why they don't just leave the weight there instead of using the meta device.
- op = comfy.utils.get_attr(diffusion_model, '.'.join(key_split[:-1]))
- weight = op._hf_hook.weights_map[key_split[-1]]
-
+ weight = comfy.model_management.resolve_lowvram_weight(sd[k], diffusion_model, k)
try:
comfy.utils.set_attr(self.control_model, k, weight)
except:
diff --git a/comfy/model_base.py b/comfy/model_base.py
index 979e2c65e..d654f56f6 100644
--- a/comfy/model_base.py
+++ b/comfy/model_base.py
@@ -3,6 +3,7 @@ from comfy.ldm.modules.diffusionmodules.openaimodel import UNetModel
from comfy.ldm.modules.encoders.noise_aug_modules import CLIPEmbeddingNoiseAugmentation
from comfy.ldm.modules.diffusionmodules.util import make_beta_schedule
from comfy.ldm.modules.diffusionmodules.openaimodel import Timestep
+import comfy.model_management
import numpy as np
from enum import Enum
from . import utils
@@ -93,7 +94,11 @@ class BaseModel(torch.nn.Module):
def state_dict_for_saving(self, clip_state_dict, vae_state_dict):
clip_state_dict = self.model_config.process_clip_state_dict_for_saving(clip_state_dict)
- unet_state_dict = self.diffusion_model.state_dict()
+ unet_sd = self.diffusion_model.state_dict()
+ unet_state_dict = {}
+ for k in unet_sd:
+ unet_state_dict[k] = comfy.model_management.resolve_lowvram_weight(unet_sd[k], self.diffusion_model, k)
+
unet_state_dict = self.model_config.process_unet_state_dict_for_saving(unet_state_dict)
vae_state_dict = self.model_config.process_vae_state_dict_for_saving(vae_state_dict)
if self.get_dtype() == torch.float16:
diff --git a/comfy/model_management.py b/comfy/model_management.py
index 016434492..e5c80bf61 100644
--- a/comfy/model_management.py
+++ b/comfy/model_management.py
@@ -1,6 +1,7 @@
import psutil
from enum import Enum
from comfy.cli_args import args
+import comfy.utils
import torch
import sys
@@ -637,6 +638,13 @@ def soft_empty_cache():
torch.cuda.empty_cache()
torch.cuda.ipc_collect()
+def resolve_lowvram_weight(weight, model, key):
+ if weight.device == torch.device("meta"): #lowvram NOTE: this depends on the inner working of the accelerate library so it might break.
+ key_split = key.split('.') # I have no idea why they don't just leave the weight there instead of using the meta device.
+ op = comfy.utils.get_attr(model, '.'.join(key_split[:-1]))
+ weight = op._hf_hook.weights_map[key_split[-1]]
+ return weight
+
#TODO: might be cleaner to put this somewhere else
import threading