Spaces:
Runtime error
Runtime error
Added cusotm models and refactor of layout
Browse files- app.py +147 -86
- finetuning.py +21 -3
- models/car.pt +3 -0
- models/vangogh.pt +3 -0
app.py
CHANGED
@@ -6,6 +6,13 @@ from finetuning import FineTunedModel
|
|
6 |
from StableDiffuser import StableDiffuser
|
7 |
from tqdm import tqdm
|
8 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
class Demo:
|
10 |
|
11 |
def __init__(self) -> None:
|
@@ -20,62 +27,49 @@ class Demo:
|
|
20 |
|
21 |
with gr.Blocks() as demo:
|
22 |
self.layout()
|
|
|
|
|
|
|
|
|
|
|
23 |
demo.queue(concurrency_count=2).launch()
|
24 |
|
25 |
def disable(self):
|
26 |
return [gr.update(interactive=False), gr.update(interactive=False)]
|
27 |
|
28 |
-
def
|
29 |
|
30 |
-
|
|
|
31 |
|
32 |
-
|
33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
34 |
|
|
|
|
|
|
|
35 |
with gr.Row():
|
36 |
-
|
37 |
-
self.prompt_input = gr.Text(
|
38 |
-
placeholder="Enter prompt...",
|
39 |
-
label="Prompt to Erase",
|
40 |
-
info="Prompt corresponding to concept to erase"
|
41 |
-
)
|
42 |
-
self.train_method_input = gr.Dropdown(
|
43 |
-
choices=['ESD-x', 'ESD-self'],
|
44 |
-
value='ESD-x',
|
45 |
-
label='Train Method',
|
46 |
-
info='Method of training'
|
47 |
-
)
|
48 |
-
|
49 |
-
self.neg_guidance_input = gr.Number(
|
50 |
-
value=1,
|
51 |
-
label="Negative Guidance",
|
52 |
-
info='Guidance of negative training used to train'
|
53 |
-
)
|
54 |
-
|
55 |
-
self.iterations_input = gr.Number(
|
56 |
-
value=150,
|
57 |
-
precision=0,
|
58 |
-
label="Iterations",
|
59 |
-
info='iterations used to train'
|
60 |
-
)
|
61 |
-
|
62 |
-
self.lr_input = gr.Number(
|
63 |
-
value=1e-5,
|
64 |
-
label="Learning Rate",
|
65 |
-
info='Learning rate used to train'
|
66 |
-
)
|
67 |
-
|
68 |
-
self.train_button = gr.Button(
|
69 |
-
value="Train",
|
70 |
-
)
|
71 |
-
|
72 |
-
self.download = gr.Files()
|
73 |
|
74 |
-
with gr.
|
75 |
|
76 |
with gr.Row():
|
77 |
|
78 |
-
|
|
|
|
|
|
|
|
|
|
|
79 |
|
80 |
self.prompt_input_infr = gr.Text(
|
81 |
placeholder="Enter prompt...",
|
@@ -83,51 +77,110 @@ class Demo:
|
|
83 |
info="Prompt to generate"
|
84 |
)
|
85 |
|
86 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
87 |
|
88 |
-
self.
|
89 |
-
|
90 |
-
|
91 |
)
|
92 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
93 |
with gr.Row():
|
94 |
|
95 |
-
self.
|
96 |
-
|
97 |
-
interactive=False
|
98 |
-
)
|
99 |
-
self.image_orig = gr.Image(
|
100 |
-
label="Orig Image",
|
101 |
-
interactive=False
|
102 |
-
)
|
103 |
|
104 |
with gr.Row():
|
105 |
|
106 |
-
|
107 |
-
|
108 |
-
|
109 |
-
|
110 |
-
|
111 |
-
|
112 |
-
|
113 |
-
|
114 |
-
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
|
121 |
-
|
122 |
-
|
123 |
-
|
124 |
-
|
125 |
-
|
126 |
-
self.iterations_input
|
127 |
-
|
128 |
-
|
129 |
-
|
130 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
|
132 |
def train(self, prompt, train_method, neg_guidance, iterations, lr, pbar = gr.Progress(track_tqdm=True)):
|
133 |
|
@@ -145,12 +198,19 @@ class Demo:
|
|
145 |
if train_method == 'ESD-x':
|
146 |
|
147 |
modules = ".*attn2$"
|
|
|
|
|
|
|
|
|
|
|
|
|
148 |
|
149 |
elif train_method == 'ESD-self':
|
150 |
|
151 |
modules = ".*attn1$"
|
|
|
152 |
|
153 |
-
finetuner = FineTunedModel(self.diffuser, modules)
|
154 |
|
155 |
optimizer = torch.optim.Adam(finetuner.parameters(), lr=lr)
|
156 |
criteria = torch.nn.MSELoss()
|
@@ -202,7 +262,7 @@ class Demo:
|
|
202 |
loss.backward()
|
203 |
optimizer.step()
|
204 |
|
205 |
-
torch.save(finetuner.state_dict(), 'ft.
|
206 |
|
207 |
self.finetuner = finetuner.eval().half()
|
208 |
|
@@ -210,19 +270,20 @@ class Demo:
|
|
210 |
|
211 |
torch.cuda.empty_cache()
|
212 |
|
213 |
-
self.training = False
|
214 |
|
215 |
-
|
216 |
|
|
|
217 |
|
218 |
-
def inference(self, prompt, seed, pbar = gr.Progress(track_tqdm=True)):
|
219 |
|
|
|
220 |
if self.generating:
|
221 |
return [None, None]
|
222 |
else:
|
223 |
self.generating = True
|
224 |
|
225 |
-
self.diffuser._seed = seed
|
226 |
|
227 |
images = self.diffuser(
|
228 |
prompt,
|
|
|
6 |
from StableDiffuser import StableDiffuser
|
7 |
from tqdm import tqdm
|
8 |
|
9 |
+
|
10 |
+
model_map = {
|
11 |
+
'Car' : 'models/car.pt',
|
12 |
+
'Van Gogh' : 'models/vangogh.pt',
|
13 |
+
}
|
14 |
+
|
15 |
+
|
16 |
class Demo:
|
17 |
|
18 |
def __init__(self) -> None:
|
|
|
27 |
|
28 |
with gr.Blocks() as demo:
|
29 |
self.layout()
|
30 |
+
self.switch_model(self.model_dropdown.value)
|
31 |
+
|
32 |
+
self.finetuner = self.finetuner.eval().half()
|
33 |
+
self.diffuser = self.diffuser.eval().half()
|
34 |
+
|
35 |
demo.queue(concurrency_count=2).launch()
|
36 |
|
37 |
def disable(self):
|
38 |
return [gr.update(interactive=False), gr.update(interactive=False)]
|
39 |
|
40 |
+
def switch_model(self, model_name):
|
41 |
|
42 |
+
if not model_name:
|
43 |
+
return
|
44 |
|
45 |
+
model_path = model_map[model_name]
|
46 |
+
|
47 |
+
checkpoint = torch.load(model_path)
|
48 |
+
|
49 |
+
del self.finetuner
|
50 |
+
|
51 |
+
torch.cuda.empty_cache()
|
52 |
+
|
53 |
+
self.finetuner = FineTunedModel.from_checkpoint(self.diffuser, checkpoint)
|
54 |
+
|
55 |
+
|
56 |
|
57 |
+
def layout(self):
|
58 |
+
|
59 |
+
|
60 |
with gr.Row():
|
61 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
|
63 |
+
with gr.Tab("Test") as inference_column:
|
64 |
|
65 |
with gr.Row():
|
66 |
|
67 |
+
self.explain_infr = gr.Markdown(interactive=False,
|
68 |
+
value='This is a demo of [Erasing Concepts from Stable Diffusion](https://erasing.baulab.info/). To try out a model where a concept has been erased, select a model and enter any prompt. For example, if you select the model "Van Gogh" you can generate images for the prompt "A portrait in the style of Van Gogh" and compare the erased and unerased models. We have also provided models with "cars" erased, and with "nudity" erased. You can also train and run your own custom model with a concept erased.')
|
69 |
+
|
70 |
+
with gr.Row():
|
71 |
+
|
72 |
+
with gr.Column(scale=1):
|
73 |
|
74 |
self.prompt_input_infr = gr.Text(
|
75 |
placeholder="Enter prompt...",
|
|
|
77 |
info="Prompt to generate"
|
78 |
)
|
79 |
|
80 |
+
with gr.Row():
|
81 |
+
|
82 |
+
self.model_dropdown = gr.Dropdown(
|
83 |
+
label="ESD Model",
|
84 |
+
choices=['Van Gogh', 'Car'],
|
85 |
+
value='Van Gogh',
|
86 |
+
interactive=True
|
87 |
+
)
|
88 |
+
|
89 |
+
self.seed_infr = gr.Number(
|
90 |
+
label="Seed",
|
91 |
+
value=42
|
92 |
+
)
|
93 |
+
|
94 |
+
with gr.Column(scale=2):
|
95 |
|
96 |
+
self.infr_button = gr.Button(
|
97 |
+
value="Generate",
|
98 |
+
interactive=True
|
99 |
)
|
100 |
|
101 |
+
with gr.Row():
|
102 |
+
|
103 |
+
self.image_new = gr.Image(
|
104 |
+
label="ESD",
|
105 |
+
interactive=False
|
106 |
+
)
|
107 |
+
self.image_orig = gr.Image(
|
108 |
+
label="SD",
|
109 |
+
interactive=False
|
110 |
+
)
|
111 |
+
|
112 |
+
with gr.Tab("Train") as training_column:
|
113 |
+
|
114 |
with gr.Row():
|
115 |
|
116 |
+
self.explain_train= gr.Markdown(interactive=False,
|
117 |
+
value='In this part you can erase any concept from Stable Diffusion. Enter a prompt for the concept or style you want to erase, and select ESD-x if you want to focus erasure on prompts that mention the concept explicitly, or ESD-u if you want to erase the concept even for prompts that do not mention the concept. With default settings, it takes about 20 minutes to fine-tune the model; then you can try inference above or download the weights. The training code used here is slightly different than the code tested in the original paper. Code and details are at [github link](https://github.com/rohitgandikota/erasing).')
|
|
|
|
|
|
|
|
|
|
|
|
|
118 |
|
119 |
with gr.Row():
|
120 |
|
121 |
+
with gr.Column(scale=3):
|
122 |
+
|
123 |
+
self.prompt_input = gr.Text(
|
124 |
+
placeholder="Enter prompt...",
|
125 |
+
label="Prompt to Erase",
|
126 |
+
info="Prompt corresponding to concept to erase"
|
127 |
+
)
|
128 |
+
self.train_method_input = gr.Dropdown(
|
129 |
+
choices=['ESD-x', 'ESD-u', 'ESD-self'],
|
130 |
+
value='ESD-x',
|
131 |
+
label='Train Method',
|
132 |
+
info='Method of training'
|
133 |
+
)
|
134 |
+
|
135 |
+
self.neg_guidance_input = gr.Number(
|
136 |
+
value=1,
|
137 |
+
label="Negative Guidance",
|
138 |
+
info='Guidance of negative training used to train'
|
139 |
+
)
|
140 |
+
|
141 |
+
self.iterations_input = gr.Number(
|
142 |
+
value=150,
|
143 |
+
precision=0,
|
144 |
+
label="Iterations",
|
145 |
+
info='iterations used to train'
|
146 |
+
)
|
147 |
+
|
148 |
+
self.lr_input = gr.Number(
|
149 |
+
value=1e-5,
|
150 |
+
label="Learning Rate",
|
151 |
+
info='Learning rate used to train'
|
152 |
+
)
|
153 |
+
|
154 |
+
with gr.Column(scale=1):
|
155 |
+
|
156 |
+
self.train_button = gr.Button(
|
157 |
+
value="Train",
|
158 |
+
)
|
159 |
+
|
160 |
+
self.download = gr.Files()
|
161 |
+
|
162 |
+
self.model_dropdown.change(self.switch_model, inputs=[self.model_dropdown])
|
163 |
+
self.infr_button.click(self.inference, inputs = [
|
164 |
+
self.prompt_input_infr,
|
165 |
+
self.seed_infr
|
166 |
+
],
|
167 |
+
outputs=[
|
168 |
+
self.image_new,
|
169 |
+
self.image_orig
|
170 |
+
]
|
171 |
+
)
|
172 |
+
self.train_button.click(self.disable,
|
173 |
+
outputs=[self.train_button, self.infr_button]
|
174 |
+
)
|
175 |
+
self.train_button.click(self.train, inputs = [
|
176 |
+
self.prompt_input,
|
177 |
+
self.train_method_input,
|
178 |
+
self.neg_guidance_input,
|
179 |
+
self.iterations_input,
|
180 |
+
self.lr_input
|
181 |
+
],
|
182 |
+
outputs=[self.train_button, self.infr_button, self.download, self.model_dropdown]
|
183 |
+
)
|
184 |
|
185 |
def train(self, prompt, train_method, neg_guidance, iterations, lr, pbar = gr.Progress(track_tqdm=True)):
|
186 |
|
|
|
198 |
if train_method == 'ESD-x':
|
199 |
|
200 |
modules = ".*attn2$"
|
201 |
+
frozen = []
|
202 |
+
|
203 |
+
elif train_method == 'ESD-u':
|
204 |
+
|
205 |
+
modules = "unet$"
|
206 |
+
frozen = [".*attn2$", "unet.time_embedding$", "unet.conv_out$"]
|
207 |
|
208 |
elif train_method == 'ESD-self':
|
209 |
|
210 |
modules = ".*attn1$"
|
211 |
+
frozen = []
|
212 |
|
213 |
+
finetuner = FineTunedModel(self.diffuser, modules, frozen_modules=frozen)
|
214 |
|
215 |
optimizer = torch.optim.Adam(finetuner.parameters(), lr=lr)
|
216 |
criteria = torch.nn.MSELoss()
|
|
|
262 |
loss.backward()
|
263 |
optimizer.step()
|
264 |
|
265 |
+
torch.save(finetuner.state_dict(), 'ft.pt')
|
266 |
|
267 |
self.finetuner = finetuner.eval().half()
|
268 |
|
|
|
270 |
|
271 |
torch.cuda.empty_cache()
|
272 |
|
273 |
+
self.training = False
|
274 |
|
275 |
+
model_map['Custom'] = 'ft.pt'
|
276 |
|
277 |
+
return [gr.update(interactive=True), gr.update(interactive=True), 'ft.pt', gr.Dropdown.update(choices=list(model_map.keys()), value='Custom')]
|
278 |
|
|
|
279 |
|
280 |
+
def inference(self, prompt, seed, pbar = gr.Progress(track_tqdm=True)):
|
281 |
if self.generating:
|
282 |
return [None, None]
|
283 |
else:
|
284 |
self.generating = True
|
285 |
|
286 |
+
self.diffuser._seed = seed or 42
|
287 |
|
288 |
images = self.diffuser(
|
289 |
prompt,
|
finetuning.py
CHANGED
@@ -38,18 +38,36 @@ class FineTunedModel(torch.nn.Module):
|
|
38 |
|
39 |
print(f"=> Finetuning {module_name}")
|
40 |
|
41 |
-
for
|
|
|
|
|
|
|
42 |
for freeze_module_name in frozen_modules:
|
43 |
|
44 |
-
match = re.search(freeze_module_name,
|
45 |
|
46 |
if match:
|
47 |
-
print(f"=> Freezing {
|
48 |
util.freeze(module)
|
49 |
|
50 |
self.ft_modules_list = torch.nn.ModuleList(self.ft_modules.values())
|
51 |
self.orig_modules_list = torch.nn.ModuleList(self.orig_modules.values())
|
52 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
53 |
def __enter__(self):
|
54 |
|
55 |
for key, ft_module in self.ft_modules.items():
|
|
|
38 |
|
39 |
print(f"=> Finetuning {module_name}")
|
40 |
|
41 |
+
for ft_module_name, module in ft_module.named_modules():
|
42 |
+
|
43 |
+
ft_module_name = f"{module_name}.{ft_module_name}"
|
44 |
+
|
45 |
for freeze_module_name in frozen_modules:
|
46 |
|
47 |
+
match = re.search(freeze_module_name, ft_module_name)
|
48 |
|
49 |
if match:
|
50 |
+
print(f"=> Freezing {ft_module_name}")
|
51 |
util.freeze(module)
|
52 |
|
53 |
self.ft_modules_list = torch.nn.ModuleList(self.ft_modules.values())
|
54 |
self.orig_modules_list = torch.nn.ModuleList(self.orig_modules.values())
|
55 |
|
56 |
+
|
57 |
+
@classmethod
|
58 |
+
def from_checkpoint(cls, model, checkpoint, frozen_modules=[]):
|
59 |
+
|
60 |
+
if isinstance(checkpoint, str):
|
61 |
+
checkpoint = torch.load(checkpoint)
|
62 |
+
|
63 |
+
modules = [f"{key}$" for key in list(checkpoint.keys())]
|
64 |
+
|
65 |
+
ftm = FineTunedModel(model, modules, frozen_modules=frozen_modules)
|
66 |
+
ftm.load_state_dict(checkpoint)
|
67 |
+
|
68 |
+
return ftm
|
69 |
+
|
70 |
+
|
71 |
def __enter__(self):
|
72 |
|
73 |
for key, ft_module in self.ft_modules.items():
|
models/car.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1a486d8417dc06dcdadfafe738ca32fb9d48f3a1a144d96cb2781e9e5f0c6f98
|
3 |
+
size 3438317621
|
models/vangogh.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:75cdb4313898f593b16f23dbceca498f3f16a749802450ab358a12c204404c27
|
3 |
+
size 175873179
|