Spaces:
Runtime error
Runtime error
Gpu mem stuff
Browse files
app.py
CHANGED
@@ -6,6 +6,7 @@ from convertModels import convert_ldm_unet_checkpoint, create_unet_diffusers_con
|
|
6 |
from omegaconf import OmegaConf
|
7 |
from StableDiffuser import StableDiffuser
|
8 |
from diffusers import UNet2DConditionModel
|
|
|
9 |
|
10 |
ckpt_path = "stable_diffusion/models/ldm/sd-v1-4-full-ema.ckpt"
|
11 |
config_path = "stable_diffusion/configs/stable-diffusion/v1-inference.yaml"
|
@@ -18,10 +19,16 @@ class Demo:
|
|
18 |
|
19 |
self.training = False
|
20 |
self.generating = False
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
|
22 |
with gr.Blocks() as demo:
|
23 |
self.layout()
|
24 |
-
demo.queue(concurrency_count=
|
25 |
|
26 |
def disable(self):
|
27 |
return [gr.update(interactive=False), gr.update(interactive=False)]
|
@@ -131,6 +138,8 @@ class Demo:
|
|
131 |
else:
|
132 |
self.training = True
|
133 |
|
|
|
|
|
134 |
model_orig, model_edited = train_esd(prompt,
|
135 |
train_method,
|
136 |
3,
|
@@ -146,8 +155,16 @@ class Demo:
|
|
146 |
original_config = OmegaConf.load(config_path)
|
147 |
original_config["model"]["params"]["unet_config"]["params"]["in_channels"] = 4
|
148 |
unet_config = create_unet_diffusers_config(original_config, image_size=512)
|
149 |
-
|
150 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
151 |
|
152 |
self.init_inference(model_edited_sd, model_orig_sd, unet_config)
|
153 |
|
@@ -155,16 +172,17 @@ class Demo:
|
|
155 |
|
156 |
def init_inference(self, model_edited_sd, model_orig_sd, unet_config):
|
157 |
|
|
|
|
|
|
|
|
|
158 |
self.model_edited_sd = model_edited_sd
|
159 |
self.model_orig_sd = model_orig_sd
|
160 |
|
161 |
-
self.diffuser = StableDiffuser(42)
|
162 |
-
|
163 |
-
self.diffuser.unet = UNet2DConditionModel(**unet_config)
|
164 |
self.diffuser.to('cuda')
|
165 |
|
166 |
self.training = False
|
167 |
-
|
168 |
|
169 |
def inference(self, prompt, seed, pbar = gr.Progress(track_tqdm=True)):
|
170 |
|
@@ -185,6 +203,8 @@ class Demo:
|
|
185 |
|
186 |
orig_image = images[0][0]
|
187 |
|
|
|
|
|
188 |
self.diffuser.unet.load_state_dict(self.model_edited_sd)
|
189 |
|
190 |
images = self.diffuser(
|
@@ -197,6 +217,8 @@ class Demo:
|
|
197 |
|
198 |
self.generating = False
|
199 |
|
|
|
|
|
200 |
return edited_image, orig_image
|
201 |
|
202 |
|
|
|
6 |
from omegaconf import OmegaConf
|
7 |
from StableDiffuser import StableDiffuser
|
8 |
from diffusers import UNet2DConditionModel
|
9 |
+
import torch
|
10 |
|
11 |
ckpt_path = "stable_diffusion/models/ldm/sd-v1-4-full-ema.ckpt"
|
12 |
config_path = "stable_diffusion/configs/stable-diffusion/v1-inference.yaml"
|
|
|
19 |
|
20 |
self.training = False
|
21 |
self.generating = False
|
22 |
+
self.model_edited_sd = None
|
23 |
+
self.model_orig_sd = None
|
24 |
+
|
25 |
+
self.diffuser = StableDiffuser(42)
|
26 |
+
self.diffuser.to('cpu')
|
27 |
+
self.diffuser = self.diffuser.half()
|
28 |
|
29 |
with gr.Blocks() as demo:
|
30 |
self.layout()
|
31 |
+
demo.queue(concurrency_count=1).launch()
|
32 |
|
33 |
def disable(self):
|
34 |
return [gr.update(interactive=False), gr.update(interactive=False)]
|
|
|
138 |
else:
|
139 |
self.training = True
|
140 |
|
141 |
+
self.diffuser.to('cpu')
|
142 |
+
|
143 |
model_orig, model_edited = train_esd(prompt,
|
144 |
train_method,
|
145 |
3,
|
|
|
155 |
original_config = OmegaConf.load(config_path)
|
156 |
original_config["model"]["params"]["unet_config"]["params"]["in_channels"] = 4
|
157 |
unet_config = create_unet_diffusers_config(original_config, image_size=512)
|
158 |
+
_model_edited_sd = convert_ldm_unet_checkpoint(model_edited.state_dict(), unet_config)
|
159 |
+
_model_orig_sd = convert_ldm_unet_checkpoint(model_orig.state_dict(), unet_config)
|
160 |
+
|
161 |
+
model_edited_sd = {key: value.cpu() for key, value in _model_edited_sd.items()}
|
162 |
+
model_orig_sd = {key: value.cpu() for key, value in _model_orig_sd.items()}
|
163 |
+
|
164 |
+
del model_orig, _model_orig_sd
|
165 |
+
del model_edited, _model_edited_sd
|
166 |
+
|
167 |
+
torch.cuda.empty_cache()
|
168 |
|
169 |
self.init_inference(model_edited_sd, model_orig_sd, unet_config)
|
170 |
|
|
|
172 |
|
173 |
def init_inference(self, model_edited_sd, model_orig_sd, unet_config):
|
174 |
|
175 |
+
del self.model_edited_sd, self.model_orig_sd
|
176 |
+
|
177 |
+
torch.cuda.empty_cache()
|
178 |
+
|
179 |
self.model_edited_sd = model_edited_sd
|
180 |
self.model_orig_sd = model_orig_sd
|
181 |
|
|
|
|
|
|
|
182 |
self.diffuser.to('cuda')
|
183 |
|
184 |
self.training = False
|
185 |
+
|
186 |
|
187 |
def inference(self, prompt, seed, pbar = gr.Progress(track_tqdm=True)):
|
188 |
|
|
|
203 |
|
204 |
orig_image = images[0][0]
|
205 |
|
206 |
+
torch.cuda.empty_cache()
|
207 |
+
|
208 |
self.diffuser.unet.load_state_dict(self.model_edited_sd)
|
209 |
|
210 |
images = self.diffuser(
|
|
|
217 |
|
218 |
self.generating = False
|
219 |
|
220 |
+
torch.cuda.empty_cache()
|
221 |
+
|
222 |
return edited_image, orig_image
|
223 |
|
224 |
|