Spaces:
Runtime error
Runtime error
1lint
commited on
Commit
Β·
919fef8
1
Parent(s):
a4b1962
refactor code and fix cpu support
Browse files- app.py +58 -286
- footer.html β html/footer.html +0 -0
- header.html β html/header.html +0 -0
- style.css β html/style.css +0 -0
- model_ids.txt +6 -0
- test.ipynb +73 -0
- utils/__init__.py +0 -0
- utils/functions.py +273 -0
- inpaint_pipeline.py β utils/inpaint_pipeline.py +63 -22
- utils/shared.py +16 -0
- textual_inversion.py β utils/textual_inversion.py +216 -78
app.py
CHANGED
@@ -1,283 +1,19 @@
|
|
1 |
-
|
2 |
-
# inpaint pipeline with fix to avoid noise added to latents during final iteration of denoising loop
|
3 |
-
from inpaint_pipeline import SDInpaintPipeline as StableDiffusionInpaintPipelineLegacy
|
4 |
-
|
5 |
-
from diffusers import (
|
6 |
-
StableDiffusionPipeline,
|
7 |
-
StableDiffusionImg2ImgPipeline,
|
8 |
-
)
|
9 |
-
|
10 |
-
import diffusers.schedulers
|
11 |
import gradio as gr
|
12 |
-
import torch
|
13 |
-
import random
|
14 |
from multiprocessing import cpu_count
|
15 |
-
import
|
16 |
-
from
|
17 |
-
import os
|
18 |
-
import argparse
|
19 |
-
import shutil
|
20 |
-
import gc
|
21 |
-
|
22 |
-
import importlib
|
23 |
-
|
24 |
-
from textual_inversion import main as run_textual_inversion
|
25 |
-
|
26 |
-
def pad_image(image):
|
27 |
-
w, h = image.size
|
28 |
-
if w == h:
|
29 |
-
return image
|
30 |
-
elif w > h:
|
31 |
-
new_image = Image.new(image.mode, (w, w), (0, 0, 0))
|
32 |
-
new_image.paste(image, (0, (w - h) // 2))
|
33 |
-
return new_image
|
34 |
-
else:
|
35 |
-
new_image = Image.new(image.mode, (h, h), (0, 0, 0))
|
36 |
-
new_image.paste(image, ((h - w) // 2, 0))
|
37 |
-
return new_image
|
38 |
-
|
39 |
-
_xformers_available = importlib.util.find_spec("xformers") is not None
|
40 |
-
device = "cuda" if torch.cuda.is_available() else "cpu"
|
41 |
-
low_vram_mode = False
|
42 |
-
|
43 |
-
# scheduler dict includes superclass SchedulerMixin (it still generates reasonable images)
|
44 |
-
scheduler_dict = {
|
45 |
-
k: v
|
46 |
-
for k, v in diffusers.schedulers.__dict__.items()
|
47 |
-
if "Scheduler" in k and "Flax" not in k
|
48 |
-
}
|
49 |
-
scheduler_dict.pop(
|
50 |
-
"VQDiffusionScheduler"
|
51 |
-
) # requires unique parameter, unlike other schedulers
|
52 |
-
scheduler_names = list(scheduler_dict.keys())
|
53 |
-
default_scheduler = scheduler_names[3] # expected to be DPM Multistep
|
54 |
-
|
55 |
-
model_ids = [
|
56 |
-
"andite/anything-v4.0",
|
57 |
-
"hakurei/waifu-diffusion",
|
58 |
-
"prompthero/openjourney-v2",
|
59 |
-
"runwayml/stable-diffusion-v1-5",
|
60 |
-
"johnslegers/epic-diffusion",
|
61 |
-
"stabilityai/stable-diffusion-2-1",
|
62 |
-
]
|
63 |
-
|
64 |
-
loaded_model_id = ""
|
65 |
-
|
66 |
-
|
67 |
-
def load_pipe(
|
68 |
-
model_id, scheduler_name, pipe_class=StableDiffusionPipeline, pipe_kwargs="{}"
|
69 |
-
):
|
70 |
-
global pipe, loaded_model_id
|
71 |
-
|
72 |
-
scheduler = scheduler_dict[scheduler_name]
|
73 |
-
|
74 |
-
# load new weights from disk only when changing model_id
|
75 |
-
if model_id != loaded_model_id:
|
76 |
-
pipe = pipe_class.from_pretrained(
|
77 |
-
model_id,
|
78 |
-
torch_dtype=torch.float16,
|
79 |
-
safety_checker=None,
|
80 |
-
requires_safety_checker=False,
|
81 |
-
scheduler=scheduler.from_pretrained(model_id, subfolder="scheduler"),
|
82 |
-
**json.loads(pipe_kwargs),
|
83 |
-
)
|
84 |
-
loaded_model_id = model_id
|
85 |
-
|
86 |
-
# if same model_id, instantiate new pipeline with same underlying pytorch objects to avoid reloading weights from disk
|
87 |
-
elif pipe_class != pipe.__class__ or not isinstance(pipe.scheduler, scheduler):
|
88 |
-
pipe.components["scheduler"] = scheduler.from_pretrained(
|
89 |
-
model_id, subfolder="scheduler"
|
90 |
-
)
|
91 |
-
pipe = pipe_class(**pipe.components)
|
92 |
-
|
93 |
-
if device == 'cuda':
|
94 |
-
pipe = pipe.to(device)
|
95 |
-
if _xformers_available:
|
96 |
-
pipe.enable_xformers_memory_efficient_attention()
|
97 |
-
print("using xformers")
|
98 |
-
if low_vram_mode:
|
99 |
-
pipe.enable_attention_slicing()
|
100 |
-
print("using attention slicing to lower VRAM")
|
101 |
-
|
102 |
-
return pipe
|
103 |
-
|
104 |
-
|
105 |
-
pipe = None
|
106 |
-
pipe = load_pipe(model_ids[0], default_scheduler)
|
107 |
-
|
108 |
-
@torch.autocast(device)
|
109 |
-
@torch.no_grad()
|
110 |
-
def generate(
|
111 |
-
model_name,
|
112 |
-
scheduler_name,
|
113 |
-
prompt,
|
114 |
-
guidance,
|
115 |
-
steps,
|
116 |
-
n_images=1,
|
117 |
-
width=512,
|
118 |
-
height=512,
|
119 |
-
seed=0,
|
120 |
-
image=None,
|
121 |
-
strength=0.5,
|
122 |
-
inpaint_image=None,
|
123 |
-
inpaint_strength=0.5,
|
124 |
-
inpaint_radio='',
|
125 |
-
neg_prompt="",
|
126 |
-
pipe_class=StableDiffusionPipeline,
|
127 |
-
pipe_kwargs="{}",
|
128 |
-
):
|
129 |
-
|
130 |
-
if seed == -1:
|
131 |
-
seed = random.randint(0, 2147483647)
|
132 |
-
|
133 |
-
generator = torch.Generator("cuda").manual_seed(seed)
|
134 |
-
|
135 |
-
pipe = load_pipe(
|
136 |
-
model_id=model_name,
|
137 |
-
scheduler_name=scheduler_name,
|
138 |
-
pipe_class=pipe_class,
|
139 |
-
pipe_kwargs=pipe_kwargs,
|
140 |
-
)
|
141 |
-
|
142 |
-
status_message = (
|
143 |
-
f"Prompt: '{prompt}' | Seed: {seed} | Guidance: {guidance} | Scheduler: {scheduler_name} | Steps: {steps}"
|
144 |
-
)
|
145 |
-
|
146 |
-
if pipe_class == StableDiffusionPipeline:
|
147 |
-
status_message = "Text to Image " + status_message
|
148 |
-
|
149 |
-
result = pipe(
|
150 |
-
prompt,
|
151 |
-
negative_prompt=neg_prompt,
|
152 |
-
num_images_per_prompt=n_images,
|
153 |
-
num_inference_steps=int(steps),
|
154 |
-
guidance_scale=guidance,
|
155 |
-
width=width,
|
156 |
-
height=height,
|
157 |
-
generator=generator,
|
158 |
-
)
|
159 |
-
|
160 |
-
elif pipe_class == StableDiffusionImg2ImgPipeline:
|
161 |
-
|
162 |
-
status_message = "Image to Image " + status_message
|
163 |
-
print(image.size)
|
164 |
-
image = image.resize((width, height))
|
165 |
-
print(image.size)
|
166 |
-
|
167 |
-
result = pipe(
|
168 |
-
prompt,
|
169 |
-
negative_prompt=neg_prompt,
|
170 |
-
num_images_per_prompt=n_images,
|
171 |
-
image=image,
|
172 |
-
num_inference_steps=int(steps),
|
173 |
-
strength=strength,
|
174 |
-
guidance_scale=guidance,
|
175 |
-
generator=generator,
|
176 |
-
)
|
177 |
-
|
178 |
-
elif pipe_class == StableDiffusionInpaintPipelineLegacy:
|
179 |
-
status_message = "Inpainting " + status_message
|
180 |
-
|
181 |
-
init_image = inpaint_image["image"].resize((width, height))
|
182 |
-
mask = inpaint_image["mask"].resize((width, height))
|
183 |
-
|
184 |
-
|
185 |
-
result = pipe(
|
186 |
-
prompt,
|
187 |
-
negative_prompt=neg_prompt,
|
188 |
-
num_images_per_prompt=n_images,
|
189 |
-
image=init_image,
|
190 |
-
mask_image=mask,
|
191 |
-
num_inference_steps=int(steps),
|
192 |
-
strength=inpaint_strength,
|
193 |
-
preserve_unmasked_image=(inpaint_radio == inpaint_options[0]),
|
194 |
-
guidance_scale=guidance,
|
195 |
-
generator=generator,
|
196 |
-
)
|
197 |
-
|
198 |
-
else:
|
199 |
-
return None, f"Unhandled pipeline class: {pipe_class}", -1
|
200 |
-
|
201 |
-
return result.images, status_message, seed
|
202 |
-
|
203 |
-
|
204 |
-
# based on lvkaokao/textual-inversion-training
|
205 |
-
def train_textual_inversion(model_name, scheduler_name, type_of_thing, files, concept_word, init_word, text_train_steps, text_train_bsz, text_learning_rate, progress=gr.Progress(track_tqdm=True)):
|
206 |
-
|
207 |
-
pipe = load_pipe(
|
208 |
-
model_id=model_name,
|
209 |
-
scheduler_name=scheduler_name,
|
210 |
-
pipe_class=StableDiffusionPipeline,
|
211 |
-
)
|
212 |
-
|
213 |
-
pipe.disable_xformers_memory_efficient_attention() # xformers handled by textual inversion script
|
214 |
-
|
215 |
-
concept_dir = 'concept_images'
|
216 |
-
output_dir = 'output_model'
|
217 |
-
training_resolution = 512
|
218 |
-
|
219 |
-
if os.path.exists(output_dir): shutil.rmtree('output_model')
|
220 |
-
if os.path.exists(concept_dir): shutil.rmtree('concept_images')
|
221 |
-
|
222 |
-
os.makedirs(concept_dir, exist_ok=True)
|
223 |
-
os.makedirs(output_dir, exist_ok=True)
|
224 |
-
|
225 |
-
gc.collect()
|
226 |
-
torch.cuda.empty_cache()
|
227 |
-
|
228 |
-
if(prompt == "" or prompt == None):
|
229 |
-
raise gr.Error("You forgot to define your concept prompt")
|
230 |
-
|
231 |
-
for j, file_temp in enumerate(files):
|
232 |
-
file = Image.open(file_temp.name)
|
233 |
-
image = pad_image(file)
|
234 |
-
image = image.resize((training_resolution, training_resolution))
|
235 |
-
extension = file_temp.name.split(".")[1]
|
236 |
-
image = image.convert('RGB')
|
237 |
-
image.save(f'{concept_dir}/{j+1}.{extension}', quality=100)
|
238 |
-
|
239 |
-
|
240 |
-
args_general = argparse.Namespace(
|
241 |
-
train_data_dir=concept_dir,
|
242 |
-
learnable_property=type_of_thing,
|
243 |
-
placeholder_token=concept_word,
|
244 |
-
initializer_token=init_word,
|
245 |
-
resolution=training_resolution,
|
246 |
-
train_batch_size=text_train_bsz,
|
247 |
-
gradient_accumulation_steps=1,
|
248 |
-
gradient_checkpointing=True,
|
249 |
-
mixed_precision='fp16',
|
250 |
-
use_bf16=False,
|
251 |
-
max_train_steps=int(text_train_steps),
|
252 |
-
learning_rate=text_learning_rate,
|
253 |
-
scale_lr=True,
|
254 |
-
lr_scheduler="constant",
|
255 |
-
lr_warmup_steps=0,
|
256 |
-
output_dir=output_dir,
|
257 |
-
)
|
258 |
-
|
259 |
-
try:
|
260 |
-
final_result = run_textual_inversion(pipe, args_general)
|
261 |
-
except Exception as e:
|
262 |
-
raise gr.Error(e)
|
263 |
-
|
264 |
-
gc.collect()
|
265 |
-
torch.cuda.empty_cache()
|
266 |
-
|
267 |
-
return f'Finished training! Check the {output_dir} directory for saved model weights'
|
268 |
-
|
269 |
|
270 |
default_img_size = 512
|
271 |
|
272 |
-
with open("header.html") as fp:
|
273 |
header = fp.read()
|
274 |
|
275 |
-
with open("footer.html") as fp:
|
276 |
footer = fp.read()
|
277 |
|
278 |
-
with gr.Blocks(css="style.css") as demo:
|
279 |
|
280 |
-
pipe_state = gr.State(lambda:
|
281 |
|
282 |
gr.HTML(header)
|
283 |
|
@@ -293,7 +29,7 @@ with gr.Blocks(css="style.css") as demo:
|
|
293 |
|
294 |
with gr.Column(scale=30):
|
295 |
model_name = gr.Dropdown(
|
296 |
-
label="Model", choices=model_ids, value=
|
297 |
)
|
298 |
scheduler_name = gr.Dropdown(
|
299 |
label="Scheduler", choices=scheduler_names, value=default_scheduler
|
@@ -305,10 +41,10 @@ with gr.Blocks(css="style.css") as demo:
|
|
305 |
with gr.Column():
|
306 |
|
307 |
with gr.Tab("Text to Image") as tab:
|
308 |
-
tab.select(lambda:
|
309 |
|
310 |
with gr.Tab("Image to image") as tab:
|
311 |
-
tab.select(lambda:
|
312 |
|
313 |
image = gr.Image(
|
314 |
label="Image to Image",
|
@@ -326,7 +62,7 @@ with gr.Blocks(css="style.css") as demo:
|
|
326 |
)
|
327 |
|
328 |
with gr.Tab("Inpainting") as tab:
|
329 |
-
tab.select(lambda:
|
330 |
|
331 |
inpaint_image = gr.Image(
|
332 |
label="Inpainting",
|
@@ -342,13 +78,26 @@ with gr.Blocks(css="style.css") as demo:
|
|
342 |
step=0.02,
|
343 |
value=0.8,
|
344 |
)
|
345 |
-
inpaint_options = [
|
346 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
347 |
|
348 |
with gr.Tab("Textual Inversion") as tab:
|
349 |
-
tab.select(lambda:
|
350 |
|
351 |
-
type_of_thing = gr.Dropdown(
|
|
|
|
|
|
|
|
|
|
|
352 |
|
353 |
text_train_bsz = gr.Slider(
|
354 |
label="Training Batch Size",
|
@@ -358,14 +107,23 @@ with gr.Blocks(css="style.css") as demo:
|
|
358 |
value=1,
|
359 |
)
|
360 |
|
361 |
-
files = gr.File(
|
|
|
|
|
|
|
|
|
|
|
362 |
|
363 |
text_train_steps = gr.Number(label="How many steps", value=1000)
|
364 |
|
365 |
-
text_learning_rate = gr.Number(label="Learning Rate", value=5.
|
366 |
|
367 |
-
concept_word = gr.Textbox(
|
368 |
-
|
|
|
|
|
|
|
|
|
369 |
|
370 |
textual_inversion_button = gr.Button(value="Train Textual Inversion")
|
371 |
|
@@ -436,17 +194,31 @@ with gr.Blocks(css="style.css") as demo:
|
|
436 |
pipe_state,
|
437 |
pipe_kwargs,
|
438 |
]
|
439 |
-
outputs = [gallery, generation_details
|
440 |
|
441 |
prompt.submit(generate, inputs=inputs, outputs=outputs)
|
442 |
generate_button.click(generate, inputs=inputs, outputs=outputs)
|
443 |
|
444 |
-
textual_inversion_inputs = [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
445 |
|
446 |
-
textual_inversion_button.click(
|
|
|
|
|
|
|
|
|
447 |
|
448 |
|
449 |
-
#demo = gr.TabbedInterface([demo, dreambooth_tab], ["Main", "Dreambooth"])
|
450 |
|
451 |
demo.queue(concurrency_count=cpu_count())
|
452 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
|
|
|
|
2 |
from multiprocessing import cpu_count
|
3 |
+
from utils.functions import generate, train_textual_inversion
|
4 |
+
from utils.shared import model_ids, scheduler_names, default_scheduler
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
|
6 |
default_img_size = 512
|
7 |
|
8 |
+
with open("html/header.html") as fp:
|
9 |
header = fp.read()
|
10 |
|
11 |
+
with open("html/footer.html") as fp:
|
12 |
footer = fp.read()
|
13 |
|
14 |
+
with gr.Blocks(css="html/style.css") as demo:
|
15 |
|
16 |
+
pipe_state = gr.State(lambda: 1)
|
17 |
|
18 |
gr.HTML(header)
|
19 |
|
|
|
29 |
|
30 |
with gr.Column(scale=30):
|
31 |
model_name = gr.Dropdown(
|
32 |
+
label="Model", choices=model_ids, value=model_ids[0]
|
33 |
)
|
34 |
scheduler_name = gr.Dropdown(
|
35 |
label="Scheduler", choices=scheduler_names, value=default_scheduler
|
|
|
41 |
with gr.Column():
|
42 |
|
43 |
with gr.Tab("Text to Image") as tab:
|
44 |
+
tab.select(lambda: 1, [], pipe_state)
|
45 |
|
46 |
with gr.Tab("Image to image") as tab:
|
47 |
+
tab.select(lambda: 2, [], pipe_state)
|
48 |
|
49 |
image = gr.Image(
|
50 |
label="Image to Image",
|
|
|
62 |
)
|
63 |
|
64 |
with gr.Tab("Inpainting") as tab:
|
65 |
+
tab.select(lambda: 3, [], pipe_state)
|
66 |
|
67 |
inpaint_image = gr.Image(
|
68 |
label="Inpainting",
|
|
|
78 |
step=0.02,
|
79 |
value=0.8,
|
80 |
)
|
81 |
+
inpaint_options = [
|
82 |
+
"preserve non-masked portions of image",
|
83 |
+
"output entire inpainted image",
|
84 |
+
]
|
85 |
+
inpaint_radio = gr.Radio(
|
86 |
+
inpaint_options,
|
87 |
+
value=inpaint_options[0],
|
88 |
+
show_label=False,
|
89 |
+
interactive=True,
|
90 |
+
)
|
91 |
|
92 |
with gr.Tab("Textual Inversion") as tab:
|
93 |
+
tab.select(lambda: 4, [], pipe_state)
|
94 |
|
95 |
+
type_of_thing = gr.Dropdown(
|
96 |
+
label="What would you like to train?",
|
97 |
+
choices=["object", "person", "style"],
|
98 |
+
value="object",
|
99 |
+
interactive=True,
|
100 |
+
)
|
101 |
|
102 |
text_train_bsz = gr.Slider(
|
103 |
label="Training Batch Size",
|
|
|
107 |
value=1,
|
108 |
)
|
109 |
|
110 |
+
files = gr.File(
|
111 |
+
label=f"""Upload the images for your concept""",
|
112 |
+
file_count="multiple",
|
113 |
+
interactive=True,
|
114 |
+
visible=True,
|
115 |
+
)
|
116 |
|
117 |
text_train_steps = gr.Number(label="How many steps", value=1000)
|
118 |
|
119 |
+
text_learning_rate = gr.Number(label="Learning Rate", value=5.0e-4)
|
120 |
|
121 |
+
concept_word = gr.Textbox(
|
122 |
+
label=f"""concept word - use a unique, made up word to avoid collisions"""
|
123 |
+
)
|
124 |
+
init_word = gr.Textbox(
|
125 |
+
label=f"""initial word - to init the concept embedding"""
|
126 |
+
)
|
127 |
|
128 |
textual_inversion_button = gr.Button(value="Train Textual Inversion")
|
129 |
|
|
|
194 |
pipe_state,
|
195 |
pipe_kwargs,
|
196 |
]
|
197 |
+
outputs = [gallery, generation_details]
|
198 |
|
199 |
prompt.submit(generate, inputs=inputs, outputs=outputs)
|
200 |
generate_button.click(generate, inputs=inputs, outputs=outputs)
|
201 |
|
202 |
+
textual_inversion_inputs = [
|
203 |
+
model_name,
|
204 |
+
scheduler_name,
|
205 |
+
type_of_thing,
|
206 |
+
files,
|
207 |
+
concept_word,
|
208 |
+
init_word,
|
209 |
+
text_train_steps,
|
210 |
+
text_train_bsz,
|
211 |
+
text_learning_rate,
|
212 |
+
]
|
213 |
|
214 |
+
textual_inversion_button.click(
|
215 |
+
train_textual_inversion,
|
216 |
+
inputs=textual_inversion_inputs,
|
217 |
+
outputs=[training_status],
|
218 |
+
)
|
219 |
|
220 |
|
221 |
+
# demo = gr.TabbedInterface([demo, dreambooth_tab], ["Main", "Dreambooth"])
|
222 |
|
223 |
demo.queue(concurrency_count=cpu_count())
|
224 |
|
footer.html β html/footer.html
RENAMED
File without changes
|
header.html β html/header.html
RENAMED
File without changes
|
style.css β html/style.css
RENAMED
File without changes
|
model_ids.txt
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
andite/anything-v4.0
|
2 |
+
hakurei/waifu-diffusion
|
3 |
+
prompthero/openjourney-v2
|
4 |
+
runwayml/stable-diffusion-v1-5
|
5 |
+
johnslegers/epic-diffusion
|
6 |
+
stabilityai/stable-diffusion-2-1
|
test.ipynb
ADDED
@@ -0,0 +1,73 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": 3,
|
6 |
+
"metadata": {},
|
7 |
+
"outputs": [],
|
8 |
+
"source": [
|
9 |
+
"with open('model_ids.txt', 'r') as fp:\n",
|
10 |
+
" model_ids = fp.read().splitlines() "
|
11 |
+
]
|
12 |
+
},
|
13 |
+
{
|
14 |
+
"cell_type": "code",
|
15 |
+
"execution_count": 4,
|
16 |
+
"metadata": {},
|
17 |
+
"outputs": [
|
18 |
+
{
|
19 |
+
"data": {
|
20 |
+
"text/plain": [
|
21 |
+
"['andite/anything-v4.0',\n",
|
22 |
+
" 'hakurei/waifu-diffusion',\n",
|
23 |
+
" 'prompthero/openjourney-v2',\n",
|
24 |
+
" 'runwayml/stable-diffusion-v1-5',\n",
|
25 |
+
" 'johnslegers/epic-diffusion',\n",
|
26 |
+
" 'stabilityai/stable-diffusion-2-1']"
|
27 |
+
]
|
28 |
+
},
|
29 |
+
"execution_count": 4,
|
30 |
+
"metadata": {},
|
31 |
+
"output_type": "execute_result"
|
32 |
+
}
|
33 |
+
],
|
34 |
+
"source": [
|
35 |
+
"model_ids"
|
36 |
+
]
|
37 |
+
},
|
38 |
+
{
|
39 |
+
"cell_type": "code",
|
40 |
+
"execution_count": null,
|
41 |
+
"metadata": {},
|
42 |
+
"outputs": [],
|
43 |
+
"source": []
|
44 |
+
}
|
45 |
+
],
|
46 |
+
"metadata": {
|
47 |
+
"kernelspec": {
|
48 |
+
"display_name": "ml",
|
49 |
+
"language": "python",
|
50 |
+
"name": "python3"
|
51 |
+
},
|
52 |
+
"language_info": {
|
53 |
+
"codemirror_mode": {
|
54 |
+
"name": "ipython",
|
55 |
+
"version": 3
|
56 |
+
},
|
57 |
+
"file_extension": ".py",
|
58 |
+
"mimetype": "text/x-python",
|
59 |
+
"name": "python",
|
60 |
+
"nbconvert_exporter": "python",
|
61 |
+
"pygments_lexer": "ipython3",
|
62 |
+
"version": "3.10.8"
|
63 |
+
},
|
64 |
+
"orig_nbformat": 4,
|
65 |
+
"vscode": {
|
66 |
+
"interpreter": {
|
67 |
+
"hash": "cbbcdde725e9a65f1cb734ac4223fed46e03daf1eb62d8ccb3c48face3871521"
|
68 |
+
}
|
69 |
+
}
|
70 |
+
},
|
71 |
+
"nbformat": 4,
|
72 |
+
"nbformat_minor": 2
|
73 |
+
}
|
utils/__init__.py
ADDED
File without changes
|
utils/functions.py
ADDED
@@ -0,0 +1,273 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
import random
|
4 |
+
from PIL import Image
|
5 |
+
import os
|
6 |
+
import argparse
|
7 |
+
import shutil
|
8 |
+
import gc
|
9 |
+
import importlib
|
10 |
+
import json
|
11 |
+
|
12 |
+
from diffusers import (
|
13 |
+
StableDiffusionPipeline,
|
14 |
+
StableDiffusionImg2ImgPipeline,
|
15 |
+
)
|
16 |
+
|
17 |
+
|
18 |
+
from .inpaint_pipeline import SDInpaintPipeline as StableDiffusionInpaintPipelineLegacy
|
19 |
+
|
20 |
+
from .textual_inversion import main as run_textual_inversion
|
21 |
+
from .shared import default_scheduler, scheduler_dict, model_ids
|
22 |
+
|
23 |
+
|
24 |
+
_xformers_available = importlib.util.find_spec("xformers") is not None
|
25 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
26 |
+
# device = 'cpu'
|
27 |
+
dtype = torch.float16 if device == "cuda" else torch.float32
|
28 |
+
low_vram_mode = False
|
29 |
+
|
30 |
+
|
31 |
+
tab_to_pipeline = {
|
32 |
+
1: StableDiffusionPipeline,
|
33 |
+
2: StableDiffusionImg2ImgPipeline,
|
34 |
+
3: StableDiffusionInpaintPipelineLegacy,
|
35 |
+
}
|
36 |
+
|
37 |
+
|
38 |
+
def load_pipe(model_id, scheduler_name, tab_index=1, pipe_kwargs="{}"):
|
39 |
+
global pipe, loaded_model_id
|
40 |
+
|
41 |
+
scheduler = scheduler_dict[scheduler_name]
|
42 |
+
|
43 |
+
pipe_class = tab_to_pipeline[tab_index]
|
44 |
+
|
45 |
+
# load new weights from disk only when changing model_id
|
46 |
+
if model_id != loaded_model_id:
|
47 |
+
pipe = pipe_class.from_pretrained(
|
48 |
+
model_id,
|
49 |
+
torch_dtype=dtype,
|
50 |
+
safety_checker=None,
|
51 |
+
requires_safety_checker=False,
|
52 |
+
scheduler=scheduler.from_pretrained(model_id, subfolder="scheduler"),
|
53 |
+
**json.loads(pipe_kwargs),
|
54 |
+
)
|
55 |
+
loaded_model_id = model_id
|
56 |
+
|
57 |
+
# if same model_id, instantiate new pipeline with same underlying pytorch objects to avoid reloading weights from disk
|
58 |
+
elif pipe_class != pipe.__class__ or not isinstance(pipe.scheduler, scheduler):
|
59 |
+
pipe.components["scheduler"] = scheduler.from_pretrained(
|
60 |
+
model_id, subfolder="scheduler"
|
61 |
+
)
|
62 |
+
pipe = pipe_class(**pipe.components)
|
63 |
+
|
64 |
+
if device == "cuda":
|
65 |
+
pipe = pipe.to(device)
|
66 |
+
if _xformers_available:
|
67 |
+
pipe.enable_xformers_memory_efficient_attention()
|
68 |
+
print("using xformers")
|
69 |
+
if low_vram_mode:
|
70 |
+
pipe.enable_attention_slicing()
|
71 |
+
print("using attention slicing to lower VRAM")
|
72 |
+
|
73 |
+
return pipe
|
74 |
+
|
75 |
+
|
76 |
+
pipe = None
|
77 |
+
loaded_model_id = ""
|
78 |
+
pipe = load_pipe(model_ids[0], default_scheduler)
|
79 |
+
|
80 |
+
|
81 |
+
def pad_image(image):
|
82 |
+
w, h = image.size
|
83 |
+
if w == h:
|
84 |
+
return image
|
85 |
+
elif w > h:
|
86 |
+
new_image = Image.new(image.mode, (w, w), (0, 0, 0))
|
87 |
+
new_image.paste(image, (0, (w - h) // 2))
|
88 |
+
return new_image
|
89 |
+
else:
|
90 |
+
new_image = Image.new(image.mode, (h, h), (0, 0, 0))
|
91 |
+
new_image.paste(image, ((h - w) // 2, 0))
|
92 |
+
return new_image
|
93 |
+
|
94 |
+
|
95 |
+
@torch.no_grad()
|
96 |
+
def generate(
|
97 |
+
model_name,
|
98 |
+
scheduler_name,
|
99 |
+
prompt,
|
100 |
+
guidance,
|
101 |
+
steps,
|
102 |
+
n_images=1,
|
103 |
+
width=512,
|
104 |
+
height=512,
|
105 |
+
seed=0,
|
106 |
+
image=None,
|
107 |
+
strength=0.5,
|
108 |
+
inpaint_image=None,
|
109 |
+
inpaint_strength=0.5,
|
110 |
+
inpaint_radio="",
|
111 |
+
neg_prompt="",
|
112 |
+
tab_index=1,
|
113 |
+
pipe_kwargs="{}",
|
114 |
+
progress=gr.Progress(track_tqdm=True),
|
115 |
+
):
|
116 |
+
|
117 |
+
if seed == -1:
|
118 |
+
seed = random.randint(0, 2147483647)
|
119 |
+
|
120 |
+
generator = torch.Generator(device).manual_seed(seed)
|
121 |
+
|
122 |
+
pipe = load_pipe(
|
123 |
+
model_id=model_name,
|
124 |
+
scheduler_name=scheduler_name,
|
125 |
+
tab_index=tab_index,
|
126 |
+
pipe_kwargs=pipe_kwargs,
|
127 |
+
)
|
128 |
+
|
129 |
+
status_message = f"Prompt: '{prompt}' | Seed: {seed} | Guidance: {guidance} | Scheduler: {scheduler_name} | Steps: {steps}"
|
130 |
+
|
131 |
+
if tab_index == 1:
|
132 |
+
status_message = "Text to Image " + status_message
|
133 |
+
|
134 |
+
result = pipe(
|
135 |
+
prompt,
|
136 |
+
negative_prompt=neg_prompt,
|
137 |
+
num_images_per_prompt=n_images,
|
138 |
+
num_inference_steps=int(steps),
|
139 |
+
guidance_scale=guidance,
|
140 |
+
width=width,
|
141 |
+
height=height,
|
142 |
+
generator=generator,
|
143 |
+
)
|
144 |
+
|
145 |
+
elif tab_index == 2:
|
146 |
+
|
147 |
+
status_message = "Image to Image " + status_message
|
148 |
+
print(image.size)
|
149 |
+
image = image.resize((width, height))
|
150 |
+
print(image.size)
|
151 |
+
|
152 |
+
result = pipe(
|
153 |
+
prompt,
|
154 |
+
negative_prompt=neg_prompt,
|
155 |
+
num_images_per_prompt=n_images,
|
156 |
+
image=image,
|
157 |
+
num_inference_steps=int(steps),
|
158 |
+
strength=strength,
|
159 |
+
guidance_scale=guidance,
|
160 |
+
generator=generator,
|
161 |
+
)
|
162 |
+
|
163 |
+
elif tab_index == 3:
|
164 |
+
status_message = "Inpainting " + status_message
|
165 |
+
|
166 |
+
init_image = inpaint_image["image"].resize((width, height))
|
167 |
+
mask = inpaint_image["mask"].resize((width, height))
|
168 |
+
|
169 |
+
result = pipe(
|
170 |
+
prompt,
|
171 |
+
negative_prompt=neg_prompt,
|
172 |
+
num_images_per_prompt=n_images,
|
173 |
+
image=init_image,
|
174 |
+
mask_image=mask,
|
175 |
+
num_inference_steps=int(steps),
|
176 |
+
strength=inpaint_strength,
|
177 |
+
preserve_unmasked_image=(
|
178 |
+
inpaint_radio == "preserve non-masked portions of image"
|
179 |
+
),
|
180 |
+
guidance_scale=guidance,
|
181 |
+
generator=generator,
|
182 |
+
)
|
183 |
+
|
184 |
+
else:
|
185 |
+
return None, f"Unhandled tab index: {tab_index}"
|
186 |
+
|
187 |
+
return result.images, status_message
|
188 |
+
|
189 |
+
|
190 |
+
# based on lvkaokao/textual-inversion-training
|
191 |
+
def train_textual_inversion(
|
192 |
+
model_name,
|
193 |
+
scheduler_name,
|
194 |
+
type_of_thing,
|
195 |
+
files,
|
196 |
+
concept_word,
|
197 |
+
init_word,
|
198 |
+
text_train_steps,
|
199 |
+
text_train_bsz,
|
200 |
+
text_learning_rate,
|
201 |
+
progress=gr.Progress(track_tqdm=True),
|
202 |
+
):
|
203 |
+
|
204 |
+
if device == "cpu":
|
205 |
+
raise gr.Error("Textual inversion training not supported on CPU")
|
206 |
+
|
207 |
+
pipe = load_pipe(
|
208 |
+
model_id=model_name,
|
209 |
+
scheduler_name=scheduler_name,
|
210 |
+
tab_index=1,
|
211 |
+
)
|
212 |
+
|
213 |
+
pipe.disable_xformers_memory_efficient_attention() # xformers handled by textual inversion script
|
214 |
+
|
215 |
+
concept_dir = "concept_images"
|
216 |
+
output_dir = "output_model"
|
217 |
+
training_resolution = 512
|
218 |
+
|
219 |
+
if os.path.exists(output_dir):
|
220 |
+
shutil.rmtree("output_model")
|
221 |
+
if os.path.exists(concept_dir):
|
222 |
+
shutil.rmtree("concept_images")
|
223 |
+
|
224 |
+
os.makedirs(concept_dir, exist_ok=True)
|
225 |
+
os.makedirs(output_dir, exist_ok=True)
|
226 |
+
|
227 |
+
gc.collect()
|
228 |
+
torch.cuda.empty_cache()
|
229 |
+
|
230 |
+
if concept_word == "" or concept_word == None:
|
231 |
+
raise gr.Error("You forgot to define your concept prompt")
|
232 |
+
|
233 |
+
for j, file_temp in enumerate(files):
|
234 |
+
file = Image.open(file_temp.name)
|
235 |
+
image = pad_image(file)
|
236 |
+
image = image.resize((training_resolution, training_resolution))
|
237 |
+
extension = file_temp.name.split(".")[1]
|
238 |
+
image = image.convert("RGB")
|
239 |
+
image.save(f"{concept_dir}/{j+1}.{extension}", quality=100)
|
240 |
+
|
241 |
+
args_general = argparse.Namespace(
|
242 |
+
train_data_dir=concept_dir,
|
243 |
+
learnable_property=type_of_thing,
|
244 |
+
placeholder_token=concept_word,
|
245 |
+
initializer_token=init_word,
|
246 |
+
resolution=training_resolution,
|
247 |
+
train_batch_size=text_train_bsz,
|
248 |
+
gradient_accumulation_steps=1,
|
249 |
+
gradient_checkpointing=True,
|
250 |
+
mixed_precision="fp16",
|
251 |
+
use_bf16=False,
|
252 |
+
max_train_steps=int(text_train_steps),
|
253 |
+
learning_rate=text_learning_rate,
|
254 |
+
scale_lr=True,
|
255 |
+
lr_scheduler="constant",
|
256 |
+
lr_warmup_steps=0,
|
257 |
+
output_dir=output_dir,
|
258 |
+
)
|
259 |
+
|
260 |
+
try:
|
261 |
+
final_result = run_textual_inversion(pipe, args_general)
|
262 |
+
except Exception as e:
|
263 |
+
raise gr.Error(e)
|
264 |
+
|
265 |
+
pipe.text_encoder = pipe.text_encoder.eval().to(device, dtype=dtype)
|
266 |
+
pipe.unet = pipe.unet.eval().to(device, dtype=dtype)
|
267 |
+
|
268 |
+
gc.collect()
|
269 |
+
torch.cuda.empty_cache()
|
270 |
+
|
271 |
+
return (
|
272 |
+
f"Finished training! Check the {output_dir} directory for saved model weights"
|
273 |
+
)
|
inpaint_pipeline.py β utils/inpaint_pipeline.py
RENAMED
@@ -1,4 +1,3 @@
|
|
1 |
-
|
2 |
# Licensed under the Apache License, Version 2.0 (the "License");
|
3 |
# you may not use this file except in compliance with the License.
|
4 |
# You may obtain a copy of the License at
|
@@ -16,21 +15,30 @@ from typing import Optional, Union, List, Callable
|
|
16 |
import PIL
|
17 |
import numpy as np
|
18 |
|
19 |
-
from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_inpaint_legacy import
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
|
21 |
def preprocess_mask(mask, scale_factor=8):
|
22 |
mask = mask.convert("L")
|
23 |
w, h = mask.size
|
24 |
w, h = map(lambda x: x - x % 32, (w, h)) # resize to integer multiple of 32
|
25 |
|
26 |
-
#input_mask = mask.resize((w, h), resample=PIL_INTERPOLATION["nearest"])
|
27 |
input_mask = np.array(mask).astype(np.float32) / 255.0
|
28 |
input_mask = np.tile(input_mask, (3, 1, 1))
|
29 |
input_mask = input_mask[None].transpose(0, 1, 2, 3) # add batch dimension
|
30 |
input_mask = 1 - input_mask # repaint white, keep black
|
31 |
input_mask = torch.round(torch.from_numpy(input_mask))
|
32 |
|
33 |
-
mask = mask.resize(
|
|
|
|
|
34 |
mask = np.array(mask).astype(np.float32) / 255.0
|
35 |
mask = np.tile(mask, (4, 1, 1))
|
36 |
mask = mask[None].transpose(0, 1, 2, 3) # add batch dimension
|
@@ -40,7 +48,6 @@ def preprocess_mask(mask, scale_factor=8):
|
|
40 |
return mask, input_mask
|
41 |
|
42 |
|
43 |
-
|
44 |
class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
45 |
|
46 |
# forward call is same as StableDiffusionInpaintPipelineLegacy, but with line added to avoid noise added to final latents right before decoding step
|
@@ -120,8 +127,8 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
120 |
The frequency at which the `callback` function will be called. If not specified, the callback will be
|
121 |
called at every step.
|
122 |
preserve_unmasked_image (`bool`, *optional*, defaults to `True`):
|
123 |
-
Whether or not to preserve the unmasked portions of the original image in the inpainted output. If False,
|
124 |
-
inpainting of the masked latents may produce noticeable distortion of unmasked portions of the decoded
|
125 |
image.
|
126 |
|
127 |
Returns:
|
@@ -148,7 +155,11 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
148 |
|
149 |
# 3. Encode input prompt
|
150 |
text_embeddings = self._encode_prompt(
|
151 |
-
prompt,
|
|
|
|
|
|
|
|
|
152 |
)
|
153 |
|
154 |
# 4. Preprocess image and mask
|
@@ -157,17 +168,27 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
157 |
|
158 |
# get mask corresponding to input latents as well as image
|
159 |
if not isinstance(mask_image, torch.FloatTensor):
|
160 |
-
mask_image, input_mask_image = preprocess_mask(
|
|
|
|
|
161 |
|
162 |
# 5. set timesteps
|
163 |
self.scheduler.set_timesteps(num_inference_steps, device=device)
|
164 |
-
timesteps, num_inference_steps = self.get_timesteps(
|
|
|
|
|
165 |
latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt)
|
166 |
|
167 |
# 6. Prepare latent variables
|
168 |
# encode the init image into latents and scale the latents
|
169 |
latents, init_latents_orig, noise = self.prepare_latents(
|
170 |
-
image,
|
|
|
|
|
|
|
|
|
|
|
|
|
171 |
)
|
172 |
|
173 |
# 7. Prepare mask latent
|
@@ -181,33 +202,47 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
181 |
num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
|
182 |
with self.progress_bar(total=num_inference_steps) as progress_bar:
|
183 |
for i, t in enumerate(timesteps):
|
184 |
-
|
185 |
# expand the latents if we are doing classifier free guidance
|
186 |
-
latent_model_input =
|
187 |
-
|
|
|
|
|
|
|
|
|
188 |
|
189 |
# predict the noise residual
|
190 |
-
noise_pred = self.unet(
|
|
|
|
|
191 |
|
192 |
# perform guidance
|
193 |
if do_classifier_free_guidance:
|
194 |
noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
|
195 |
-
noise_pred = noise_pred_uncond + guidance_scale * (
|
|
|
|
|
196 |
|
197 |
# compute the previous noisy sample x_t -> x_t-1
|
198 |
-
latents = self.scheduler.step(
|
|
|
|
|
199 |
# masking
|
200 |
if add_predicted_noise:
|
201 |
init_latents_proper = self.scheduler.add_noise(
|
202 |
init_latents_orig, noise_pred_uncond, torch.tensor([t])
|
203 |
)
|
204 |
else:
|
205 |
-
init_latents_proper = self.scheduler.add_noise(
|
|
|
|
|
206 |
|
207 |
latents = (init_latents_proper * mask) + (latents * (1 - mask))
|
208 |
|
209 |
# call the callback, if provided
|
210 |
-
if i == len(timesteps) - 1 or (
|
|
|
|
|
211 |
progress_bar.update()
|
212 |
if callback is not None and i % callback_steps == 0:
|
213 |
callback(i, t, latents)
|
@@ -225,7 +260,9 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
225 |
# restore unmasked parts of image with original image
|
226 |
input_mask_image = input_mask_image.to(inpaint_image)
|
227 |
image = image.to(inpaint_image)
|
228 |
-
image = (image * input_mask_image) + (
|
|
|
|
|
229 |
|
230 |
# post-processing of image
|
231 |
image = (image / 2 + 0.5).clamp(0, 1)
|
@@ -235,7 +272,9 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
235 |
image = self.decode_latents(latents)
|
236 |
|
237 |
# 11. Run safety checker
|
238 |
-
image, has_nsfw_concept = self.run_safety_checker(
|
|
|
|
|
239 |
|
240 |
# 12. Convert to PIL
|
241 |
if output_type == "pil":
|
@@ -244,4 +283,6 @@ class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
|
244 |
if not return_dict:
|
245 |
return (image, has_nsfw_concept)
|
246 |
|
247 |
-
return StableDiffusionPipelineOutput(
|
|
|
|
|
|
|
|
1 |
# Licensed under the Apache License, Version 2.0 (the "License");
|
2 |
# you may not use this file except in compliance with the License.
|
3 |
# You may obtain a copy of the License at
|
|
|
15 |
import PIL
|
16 |
import numpy as np
|
17 |
|
18 |
+
from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_inpaint_legacy import (
|
19 |
+
preprocess_image,
|
20 |
+
deprecate,
|
21 |
+
StableDiffusionInpaintPipelineLegacy,
|
22 |
+
StableDiffusionPipelineOutput,
|
23 |
+
PIL_INTERPOLATION,
|
24 |
+
)
|
25 |
+
|
26 |
|
27 |
def preprocess_mask(mask, scale_factor=8):
|
28 |
mask = mask.convert("L")
|
29 |
w, h = mask.size
|
30 |
w, h = map(lambda x: x - x % 32, (w, h)) # resize to integer multiple of 32
|
31 |
|
32 |
+
# input_mask = mask.resize((w, h), resample=PIL_INTERPOLATION["nearest"])
|
33 |
input_mask = np.array(mask).astype(np.float32) / 255.0
|
34 |
input_mask = np.tile(input_mask, (3, 1, 1))
|
35 |
input_mask = input_mask[None].transpose(0, 1, 2, 3) # add batch dimension
|
36 |
input_mask = 1 - input_mask # repaint white, keep black
|
37 |
input_mask = torch.round(torch.from_numpy(input_mask))
|
38 |
|
39 |
+
mask = mask.resize(
|
40 |
+
(w // scale_factor, h // scale_factor), resample=PIL_INTERPOLATION["nearest"]
|
41 |
+
)
|
42 |
mask = np.array(mask).astype(np.float32) / 255.0
|
43 |
mask = np.tile(mask, (4, 1, 1))
|
44 |
mask = mask[None].transpose(0, 1, 2, 3) # add batch dimension
|
|
|
48 |
return mask, input_mask
|
49 |
|
50 |
|
|
|
51 |
class SDInpaintPipeline(StableDiffusionInpaintPipelineLegacy):
|
52 |
|
53 |
# forward call is same as StableDiffusionInpaintPipelineLegacy, but with line added to avoid noise added to final latents right before decoding step
|
|
|
127 |
The frequency at which the `callback` function will be called. If not specified, the callback will be
|
128 |
called at every step.
|
129 |
preserve_unmasked_image (`bool`, *optional*, defaults to `True`):
|
130 |
+
Whether or not to preserve the unmasked portions of the original image in the inpainted output. If False,
|
131 |
+
inpainting of the masked latents may produce noticeable distortion of unmasked portions of the decoded
|
132 |
image.
|
133 |
|
134 |
Returns:
|
|
|
155 |
|
156 |
# 3. Encode input prompt
|
157 |
text_embeddings = self._encode_prompt(
|
158 |
+
prompt,
|
159 |
+
device,
|
160 |
+
num_images_per_prompt,
|
161 |
+
do_classifier_free_guidance,
|
162 |
+
negative_prompt,
|
163 |
)
|
164 |
|
165 |
# 4. Preprocess image and mask
|
|
|
168 |
|
169 |
# get mask corresponding to input latents as well as image
|
170 |
if not isinstance(mask_image, torch.FloatTensor):
|
171 |
+
mask_image, input_mask_image = preprocess_mask(
|
172 |
+
mask_image, self.vae_scale_factor
|
173 |
+
)
|
174 |
|
175 |
# 5. set timesteps
|
176 |
self.scheduler.set_timesteps(num_inference_steps, device=device)
|
177 |
+
timesteps, num_inference_steps = self.get_timesteps(
|
178 |
+
num_inference_steps, strength, device
|
179 |
+
)
|
180 |
latent_timestep = timesteps[:1].repeat(batch_size * num_images_per_prompt)
|
181 |
|
182 |
# 6. Prepare latent variables
|
183 |
# encode the init image into latents and scale the latents
|
184 |
latents, init_latents_orig, noise = self.prepare_latents(
|
185 |
+
image,
|
186 |
+
latent_timestep,
|
187 |
+
batch_size,
|
188 |
+
num_images_per_prompt,
|
189 |
+
text_embeddings.dtype,
|
190 |
+
device,
|
191 |
+
generator,
|
192 |
)
|
193 |
|
194 |
# 7. Prepare mask latent
|
|
|
202 |
num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
|
203 |
with self.progress_bar(total=num_inference_steps) as progress_bar:
|
204 |
for i, t in enumerate(timesteps):
|
205 |
+
|
206 |
# expand the latents if we are doing classifier free guidance
|
207 |
+
latent_model_input = (
|
208 |
+
torch.cat([latents] * 2) if do_classifier_free_guidance else latents
|
209 |
+
)
|
210 |
+
latent_model_input = self.scheduler.scale_model_input(
|
211 |
+
latent_model_input, t
|
212 |
+
)
|
213 |
|
214 |
# predict the noise residual
|
215 |
+
noise_pred = self.unet(
|
216 |
+
latent_model_input, t, encoder_hidden_states=text_embeddings
|
217 |
+
).sample
|
218 |
|
219 |
# perform guidance
|
220 |
if do_classifier_free_guidance:
|
221 |
noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
|
222 |
+
noise_pred = noise_pred_uncond + guidance_scale * (
|
223 |
+
noise_pred_text - noise_pred_uncond
|
224 |
+
)
|
225 |
|
226 |
# compute the previous noisy sample x_t -> x_t-1
|
227 |
+
latents = self.scheduler.step(
|
228 |
+
noise_pred, t, latents, **extra_step_kwargs
|
229 |
+
).prev_sample
|
230 |
# masking
|
231 |
if add_predicted_noise:
|
232 |
init_latents_proper = self.scheduler.add_noise(
|
233 |
init_latents_orig, noise_pred_uncond, torch.tensor([t])
|
234 |
)
|
235 |
else:
|
236 |
+
init_latents_proper = self.scheduler.add_noise(
|
237 |
+
init_latents_orig, noise, torch.tensor([t])
|
238 |
+
)
|
239 |
|
240 |
latents = (init_latents_proper * mask) + (latents * (1 - mask))
|
241 |
|
242 |
# call the callback, if provided
|
243 |
+
if i == len(timesteps) - 1 or (
|
244 |
+
(i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0
|
245 |
+
):
|
246 |
progress_bar.update()
|
247 |
if callback is not None and i % callback_steps == 0:
|
248 |
callback(i, t, latents)
|
|
|
260 |
# restore unmasked parts of image with original image
|
261 |
input_mask_image = input_mask_image.to(inpaint_image)
|
262 |
image = image.to(inpaint_image)
|
263 |
+
image = (image * input_mask_image) + (
|
264 |
+
inpaint_image * (1 - input_mask_image)
|
265 |
+
) # use original unmasked portions of image to avoid degradation
|
266 |
|
267 |
# post-processing of image
|
268 |
image = (image / 2 + 0.5).clamp(0, 1)
|
|
|
272 |
image = self.decode_latents(latents)
|
273 |
|
274 |
# 11. Run safety checker
|
275 |
+
image, has_nsfw_concept = self.run_safety_checker(
|
276 |
+
image, device, text_embeddings.dtype
|
277 |
+
)
|
278 |
|
279 |
# 12. Convert to PIL
|
280 |
if output_type == "pil":
|
|
|
283 |
if not return_dict:
|
284 |
return (image, has_nsfw_concept)
|
285 |
|
286 |
+
return StableDiffusionPipelineOutput(
|
287 |
+
images=image, nsfw_content_detected=has_nsfw_concept
|
288 |
+
)
|
utils/shared.py
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import diffusers.schedulers
|
2 |
+
|
3 |
+
# scheduler dict includes superclass SchedulerMixin (it still generates reasonable images)
|
4 |
+
scheduler_dict = {
|
5 |
+
k: v
|
6 |
+
for k, v in diffusers.schedulers.__dict__.items()
|
7 |
+
if "Scheduler" in k and "Flax" not in k
|
8 |
+
}
|
9 |
+
scheduler_dict.pop(
|
10 |
+
"VQDiffusionScheduler"
|
11 |
+
) # requires unique parameter, unlike other schedulers
|
12 |
+
scheduler_names = list(scheduler_dict.keys())
|
13 |
+
default_scheduler = scheduler_names[3] # expected to be DPM Multistep
|
14 |
+
|
15 |
+
with open("model_ids.txt", "r") as fp:
|
16 |
+
model_ids = fp.read().splitlines()
|
textual_inversion.py β utils/textual_inversion.py
RENAMED
@@ -34,7 +34,12 @@ import transformers
|
|
34 |
from accelerate import Accelerator
|
35 |
from accelerate.logging import get_logger
|
36 |
from accelerate.utils import set_seed
|
37 |
-
from diffusers import
|
|
|
|
|
|
|
|
|
|
|
38 |
from diffusers.optimization import get_scheduler
|
39 |
from diffusers.utils import check_min_version
|
40 |
from diffusers.utils.import_utils import is_xformers_available
|
@@ -76,7 +81,11 @@ logger = get_logger(__name__)
|
|
76 |
|
77 |
def save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path):
|
78 |
logger.info("Saving embeddings")
|
79 |
-
learned_embeds =
|
|
|
|
|
|
|
|
|
80 |
learned_embeds_dict = {args.placeholder_token: learned_embeds.detach().cpu()}
|
81 |
torch.save(learned_embeds_dict, save_path)
|
82 |
|
@@ -114,7 +123,10 @@ def parse_args():
|
|
114 |
help="Pretrained tokenizer name or path if not the same as model_name",
|
115 |
)
|
116 |
parser.add_argument(
|
117 |
-
"--train_data_dir",
|
|
|
|
|
|
|
118 |
)
|
119 |
parser.add_argument(
|
120 |
"--placeholder_token",
|
@@ -123,18 +135,33 @@ def parse_args():
|
|
123 |
help="A token to use as a placeholder for the concept.",
|
124 |
)
|
125 |
parser.add_argument(
|
126 |
-
"--initializer_token",
|
|
|
|
|
|
|
127 |
)
|
128 |
|
129 |
-
parser.add_argument(
|
130 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
131 |
parser.add_argument(
|
132 |
"--output_dir",
|
133 |
type=str,
|
134 |
default="text-inversion-model",
|
135 |
help="The output directory where the model predictions and checkpoints will be written.",
|
136 |
)
|
137 |
-
parser.add_argument(
|
|
|
|
|
138 |
parser.add_argument(
|
139 |
"--resolution",
|
140 |
type=int,
|
@@ -145,10 +172,15 @@ def parse_args():
|
|
145 |
),
|
146 |
)
|
147 |
parser.add_argument(
|
148 |
-
"--center_crop",
|
|
|
|
|
149 |
)
|
150 |
parser.add_argument(
|
151 |
-
"--train_batch_size",
|
|
|
|
|
|
|
152 |
)
|
153 |
parser.add_argument("--num_train_epochs", type=int, default=100)
|
154 |
parser.add_argument(
|
@@ -190,14 +222,43 @@ def parse_args():
|
|
190 |
),
|
191 |
)
|
192 |
parser.add_argument(
|
193 |
-
"--lr_warmup_steps",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
194 |
)
|
195 |
-
parser.add_argument("--adam_beta1", type=float, default=0.9, help="The beta1 parameter for the Adam optimizer.")
|
196 |
-
parser.add_argument("--adam_beta2", type=float, default=0.999, help="The beta2 parameter for the Adam optimizer.")
|
197 |
-
parser.add_argument("--adam_weight_decay", type=float, default=1e-2, help="Weight decay to use.")
|
198 |
-
parser.add_argument("--adam_epsilon", type=float, default=1e-08, help="Epsilon value for the Adam optimizer")
|
199 |
-
parser.add_argument("--push_to_hub", action="store_true", help="Whether or not to push the model to the Hub.")
|
200 |
-
parser.add_argument("--hub_token", type=str, default=None, help="The token to use to push to the Model Hub.")
|
201 |
parser.add_argument(
|
202 |
"--hub_model_id",
|
203 |
type=str,
|
@@ -241,7 +302,12 @@ def parse_args():
|
|
241 |
' (default), `"wandb"` and `"comet_ml"`. Use `"all"` to report to all integrations.'
|
242 |
),
|
243 |
)
|
244 |
-
parser.add_argument(
|
|
|
|
|
|
|
|
|
|
|
245 |
parser.add_argument(
|
246 |
"--checkpointing_steps",
|
247 |
type=int,
|
@@ -261,7 +327,9 @@ def parse_args():
|
|
261 |
),
|
262 |
)
|
263 |
parser.add_argument(
|
264 |
-
"--enable_xformers_memory_efficient_attention",
|
|
|
|
|
265 |
)
|
266 |
|
267 |
args = parser.parse_args()
|
@@ -269,7 +337,7 @@ def parse_args():
|
|
269 |
if env_local_rank != -1 and env_local_rank != args.local_rank:
|
270 |
args.local_rank = env_local_rank
|
271 |
|
272 |
-
#if args.train_data_dir is None:
|
273 |
# raise ValueError("You must specify a train data directory.")
|
274 |
|
275 |
return args
|
@@ -350,7 +418,10 @@ class TextualInversionDataset(Dataset):
|
|
350 |
self.center_crop = center_crop
|
351 |
self.flip_p = flip_p
|
352 |
|
353 |
-
self.image_paths = [
|
|
|
|
|
|
|
354 |
|
355 |
self.num_images = len(self.image_paths)
|
356 |
self._length = self.num_images
|
@@ -365,7 +436,11 @@ class TextualInversionDataset(Dataset):
|
|
365 |
"lanczos": PIL_INTERPOLATION["lanczos"],
|
366 |
}[interpolation]
|
367 |
|
368 |
-
self.templates =
|
|
|
|
|
|
|
|
|
369 |
self.flip_transform = transforms.RandomHorizontalFlip(p=self.flip_p)
|
370 |
|
371 |
def __len__(self):
|
@@ -394,14 +469,13 @@ class TextualInversionDataset(Dataset):
|
|
394 |
|
395 |
if self.center_crop:
|
396 |
crop = min(img.shape[0], img.shape[1])
|
397 |
-
(
|
398 |
-
h,
|
399 |
-
w,
|
400 |
-
) = (
|
401 |
img.shape[0],
|
402 |
img.shape[1],
|
403 |
)
|
404 |
-
img = img[
|
|
|
|
|
405 |
|
406 |
image = Image.fromarray(img)
|
407 |
image = image.resize((self.size, self.size), resample=self.interpolation)
|
@@ -414,7 +488,9 @@ class TextualInversionDataset(Dataset):
|
|
414 |
return example
|
415 |
|
416 |
|
417 |
-
def get_full_repo_name(
|
|
|
|
|
418 |
if token is None:
|
419 |
token = HfFolder.get_token()
|
420 |
if organization is None:
|
@@ -424,7 +500,6 @@ def get_full_repo_name(model_id: str, organization: Optional[str] = None, token:
|
|
424 |
return f"{organization}/{model_id}"
|
425 |
|
426 |
|
427 |
-
|
428 |
def main(pipe, args_imported):
|
429 |
|
430 |
args = parse_args()
|
@@ -464,11 +539,15 @@ def main(pipe, args_imported):
|
|
464 |
if accelerator.is_main_process:
|
465 |
if args.push_to_hub:
|
466 |
if args.hub_model_id is None:
|
467 |
-
repo_name = get_full_repo_name(
|
|
|
|
|
468 |
else:
|
469 |
repo_name = args.hub_model_id
|
470 |
create_repo(repo_name, exist_ok=True, token=args.hub_token)
|
471 |
-
repo = Repository(
|
|
|
|
|
472 |
|
473 |
with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
|
474 |
if "step_*" not in gitignore:
|
@@ -530,7 +609,9 @@ def main(pipe, args_imported):
|
|
530 |
if is_xformers_available():
|
531 |
unet.enable_xformers_memory_efficient_attention()
|
532 |
else:
|
533 |
-
raise ValueError(
|
|
|
|
|
534 |
|
535 |
# Enable TF32 for faster training on Ampere GPUs,
|
536 |
# cf https://pytorch.org/docs/stable/notes/cuda.html#tensorfloat-32-tf32-on-ampere-devices
|
@@ -539,7 +620,10 @@ def main(pipe, args_imported):
|
|
539 |
|
540 |
if args.scale_lr:
|
541 |
args.learning_rate = (
|
542 |
-
args.learning_rate
|
|
|
|
|
|
|
543 |
)
|
544 |
|
545 |
# Initialize the optimizer
|
@@ -562,11 +646,15 @@ def main(pipe, args_imported):
|
|
562 |
center_crop=args.center_crop,
|
563 |
set="train",
|
564 |
)
|
565 |
-
train_dataloader = torch.utils.data.DataLoader(
|
|
|
|
|
566 |
|
567 |
# Scheduler and math around the number of training steps.
|
568 |
overrode_max_train_steps = False
|
569 |
-
num_update_steps_per_epoch = math.ceil(
|
|
|
|
|
570 |
if args.max_train_steps is None:
|
571 |
args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
|
572 |
overrode_max_train_steps = True
|
@@ -597,7 +685,9 @@ def main(pipe, args_imported):
|
|
597 |
text_encoder.to(accelerator.device, dtype=torch.float32)
|
598 |
|
599 |
# We need to recalculate our total training steps as the size of the training dataloader may have changed.
|
600 |
-
num_update_steps_per_epoch = math.ceil(
|
|
|
|
|
601 |
if overrode_max_train_steps:
|
602 |
args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
|
603 |
# Afterwards we recalculate our number of training epochs
|
@@ -609,13 +699,19 @@ def main(pipe, args_imported):
|
|
609 |
accelerator.init_trackers("textual_inversion", config=vars(args))
|
610 |
|
611 |
# Train!
|
612 |
-
total_batch_size =
|
|
|
|
|
|
|
|
|
613 |
|
614 |
logger.info("***** Running training *****")
|
615 |
logger.info(f" Num examples = {len(train_dataset)}")
|
616 |
logger.info(f" Num Epochs = {args.num_train_epochs}")
|
617 |
logger.info(f" Instantaneous batch size per device = {args.train_batch_size}")
|
618 |
-
logger.info(
|
|
|
|
|
619 |
logger.info(f" Gradient Accumulation steps = {args.gradient_accumulation_steps}")
|
620 |
logger.info(f" Total optimization steps = {args.max_train_steps}")
|
621 |
global_step = 0
|
@@ -640,31 +736,51 @@ def main(pipe, args_imported):
|
|
640 |
resume_step = resume_global_step % num_update_steps_per_epoch
|
641 |
|
642 |
# Only show the progress bar once on each machine.
|
643 |
-
progress_bar = tqdm(
|
|
|
|
|
|
|
644 |
progress_bar.set_description("Steps")
|
645 |
|
646 |
# keep original embeddings as reference
|
647 |
-
orig_embeds_params =
|
|
|
|
|
|
|
|
|
648 |
|
649 |
-
for epoch in
|
650 |
text_encoder.train()
|
651 |
for step, batch in enumerate(train_dataloader):
|
652 |
# Skip steps until we reach the resumed step
|
653 |
-
if
|
|
|
|
|
|
|
|
|
654 |
if step % args.gradient_accumulation_steps == 0:
|
655 |
progress_bar.update(1)
|
656 |
continue
|
657 |
|
658 |
with accelerator.accumulate(text_encoder):
|
659 |
# Convert images to latent space
|
660 |
-
latents =
|
|
|
|
|
|
|
|
|
661 |
latents = latents * 0.18215
|
662 |
|
663 |
# Sample noise that we'll add to the latents
|
664 |
noise = torch.randn_like(latents)
|
665 |
bsz = latents.shape[0]
|
666 |
# Sample a random timestep for each image
|
667 |
-
timesteps = torch.randint(
|
|
|
|
|
|
|
|
|
|
|
668 |
timesteps = timesteps.long()
|
669 |
|
670 |
# Add noise to the latents according to the noise magnitude at each timestep
|
@@ -672,10 +788,14 @@ def main(pipe, args_imported):
|
|
672 |
noisy_latents = noise_scheduler.add_noise(latents, noise, timesteps)
|
673 |
|
674 |
# Get the text embedding for conditioning
|
675 |
-
encoder_hidden_states = text_encoder(batch["input_ids"])[0].to(
|
|
|
|
|
676 |
|
677 |
# Predict the noise residual
|
678 |
-
model_pred = unet(
|
|
|
|
|
679 |
|
680 |
# Get the target for loss depending on the prediction type
|
681 |
if noise_scheduler.config.prediction_type == "epsilon":
|
@@ -683,7 +803,9 @@ def main(pipe, args_imported):
|
|
683 |
elif noise_scheduler.config.prediction_type == "v_prediction":
|
684 |
target = noise_scheduler.get_velocity(latents, noise, timesteps)
|
685 |
else:
|
686 |
-
raise ValueError(
|
|
|
|
|
687 |
|
688 |
loss = F.mse_loss(model_pred.float(), target.float(), reduction="mean")
|
689 |
|
@@ -694,8 +816,12 @@ def main(pipe, args_imported):
|
|
694 |
else:
|
695 |
grads = text_encoder.get_input_embeddings().weight.grad
|
696 |
# Get the index for tokens that we want to zero the grads for
|
697 |
-
index_grads_to_zero =
|
698 |
-
|
|
|
|
|
|
|
|
|
699 |
|
700 |
optimizer.step()
|
701 |
lr_scheduler.step()
|
@@ -704,21 +830,31 @@ def main(pipe, args_imported):
|
|
704 |
# Let's make sure we don't update any embedding weights besides the newly added token
|
705 |
index_no_updates = torch.arange(len(tokenizer)) != placeholder_token_id
|
706 |
with torch.no_grad():
|
707 |
-
accelerator.unwrap_model(
|
|
|
|
|
|
|
|
|
708 |
index_no_updates
|
709 |
-
]
|
710 |
|
711 |
# Checks if the accelerator has performed an optimization step behind the scenes
|
712 |
if accelerator.sync_gradients:
|
713 |
progress_bar.update(1)
|
714 |
global_step += 1
|
715 |
if global_step % args.save_steps == 0:
|
716 |
-
save_path = os.path.join(
|
717 |
-
|
|
|
|
|
|
|
|
|
718 |
|
719 |
if global_step % args.checkpointing_steps == 0:
|
720 |
if accelerator.is_main_process:
|
721 |
-
save_path = os.path.join(
|
|
|
|
|
722 |
accelerator.save_state(save_path)
|
723 |
logger.info(f"Saved state to {save_path}")
|
724 |
|
@@ -733,7 +869,9 @@ def main(pipe, args_imported):
|
|
733 |
accelerator.wait_for_everyone()
|
734 |
if accelerator.is_main_process:
|
735 |
if args.push_to_hub and args.only_save_embeds:
|
736 |
-
logger.warn(
|
|
|
|
|
737 |
save_full_model = True
|
738 |
else:
|
739 |
save_full_model = not args.only_save_embeds
|
@@ -744,35 +882,35 @@ def main(pipe, args_imported):
|
|
744 |
save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path)
|
745 |
|
746 |
if args.push_to_hub:
|
747 |
-
repo.push_to_hub(
|
|
|
|
|
748 |
|
749 |
accelerator.end_training()
|
750 |
-
text_encoder.eval()
|
751 |
-
unet.eval()
|
752 |
-
vae.eval()
|
753 |
|
754 |
|
755 |
if __name__ == "__main__":
|
756 |
-
pipeline = StableDiffusionPipeline.from_pretrained(
|
757 |
-
|
758 |
-
|
759 |
-
train_data_dir="concept_images",
|
760 |
-
learnable_property='object',
|
761 |
-
placeholder_token='redeyegirl',
|
762 |
-
initializer_token='girl',
|
763 |
-
resolution=512,
|
764 |
-
train_batch_size=1,
|
765 |
-
gradient_accumulation_steps=1,
|
766 |
-
gradient_checkpointing=True,
|
767 |
-
mixed_precision='fp16',
|
768 |
-
use_bf16=False,
|
769 |
-
max_train_steps=1000,
|
770 |
-
learning_rate=5.0e-4,
|
771 |
-
scale_lr=False,
|
772 |
-
lr_scheduler="constant",
|
773 |
-
lr_warmup_steps=0,
|
774 |
-
output_dir="output_model",
|
775 |
-
)
|
776 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
777 |
|
778 |
main(pipeline, imported_args)
|
|
|
34 |
from accelerate import Accelerator
|
35 |
from accelerate.logging import get_logger
|
36 |
from accelerate.utils import set_seed
|
37 |
+
from diffusers import (
|
38 |
+
AutoencoderKL,
|
39 |
+
DDPMScheduler,
|
40 |
+
StableDiffusionPipeline,
|
41 |
+
UNet2DConditionModel,
|
42 |
+
)
|
43 |
from diffusers.optimization import get_scheduler
|
44 |
from diffusers.utils import check_min_version
|
45 |
from diffusers.utils.import_utils import is_xformers_available
|
|
|
81 |
|
82 |
def save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path):
|
83 |
logger.info("Saving embeddings")
|
84 |
+
learned_embeds = (
|
85 |
+
accelerator.unwrap_model(text_encoder)
|
86 |
+
.get_input_embeddings()
|
87 |
+
.weight[placeholder_token_id]
|
88 |
+
)
|
89 |
learned_embeds_dict = {args.placeholder_token: learned_embeds.detach().cpu()}
|
90 |
torch.save(learned_embeds_dict, save_path)
|
91 |
|
|
|
123 |
help="Pretrained tokenizer name or path if not the same as model_name",
|
124 |
)
|
125 |
parser.add_argument(
|
126 |
+
"--train_data_dir",
|
127 |
+
type=str,
|
128 |
+
default=None,
|
129 |
+
help="A folder containing the training data.",
|
130 |
)
|
131 |
parser.add_argument(
|
132 |
"--placeholder_token",
|
|
|
135 |
help="A token to use as a placeholder for the concept.",
|
136 |
)
|
137 |
parser.add_argument(
|
138 |
+
"--initializer_token",
|
139 |
+
type=str,
|
140 |
+
default=None,
|
141 |
+
help="A token to use as initializer word.",
|
142 |
)
|
143 |
|
144 |
+
parser.add_argument(
|
145 |
+
"--learnable_property",
|
146 |
+
type=str,
|
147 |
+
default="object",
|
148 |
+
help="Choose between 'object' and 'style'",
|
149 |
+
)
|
150 |
+
parser.add_argument(
|
151 |
+
"--repeats",
|
152 |
+
type=int,
|
153 |
+
default=100,
|
154 |
+
help="How many times to repeat the training data.",
|
155 |
+
)
|
156 |
parser.add_argument(
|
157 |
"--output_dir",
|
158 |
type=str,
|
159 |
default="text-inversion-model",
|
160 |
help="The output directory where the model predictions and checkpoints will be written.",
|
161 |
)
|
162 |
+
parser.add_argument(
|
163 |
+
"--seed", type=int, default=None, help="A seed for reproducible training."
|
164 |
+
)
|
165 |
parser.add_argument(
|
166 |
"--resolution",
|
167 |
type=int,
|
|
|
172 |
),
|
173 |
)
|
174 |
parser.add_argument(
|
175 |
+
"--center_crop",
|
176 |
+
action="store_true",
|
177 |
+
help="Whether to center crop images before resizing to resolution",
|
178 |
)
|
179 |
parser.add_argument(
|
180 |
+
"--train_batch_size",
|
181 |
+
type=int,
|
182 |
+
default=16,
|
183 |
+
help="Batch size (per device) for the training dataloader.",
|
184 |
)
|
185 |
parser.add_argument("--num_train_epochs", type=int, default=100)
|
186 |
parser.add_argument(
|
|
|
222 |
),
|
223 |
)
|
224 |
parser.add_argument(
|
225 |
+
"--lr_warmup_steps",
|
226 |
+
type=int,
|
227 |
+
default=500,
|
228 |
+
help="Number of steps for the warmup in the lr scheduler.",
|
229 |
+
)
|
230 |
+
parser.add_argument(
|
231 |
+
"--adam_beta1",
|
232 |
+
type=float,
|
233 |
+
default=0.9,
|
234 |
+
help="The beta1 parameter for the Adam optimizer.",
|
235 |
+
)
|
236 |
+
parser.add_argument(
|
237 |
+
"--adam_beta2",
|
238 |
+
type=float,
|
239 |
+
default=0.999,
|
240 |
+
help="The beta2 parameter for the Adam optimizer.",
|
241 |
+
)
|
242 |
+
parser.add_argument(
|
243 |
+
"--adam_weight_decay", type=float, default=1e-2, help="Weight decay to use."
|
244 |
+
)
|
245 |
+
parser.add_argument(
|
246 |
+
"--adam_epsilon",
|
247 |
+
type=float,
|
248 |
+
default=1e-08,
|
249 |
+
help="Epsilon value for the Adam optimizer",
|
250 |
+
)
|
251 |
+
parser.add_argument(
|
252 |
+
"--push_to_hub",
|
253 |
+
action="store_true",
|
254 |
+
help="Whether or not to push the model to the Hub.",
|
255 |
+
)
|
256 |
+
parser.add_argument(
|
257 |
+
"--hub_token",
|
258 |
+
type=str,
|
259 |
+
default=None,
|
260 |
+
help="The token to use to push to the Model Hub.",
|
261 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
262 |
parser.add_argument(
|
263 |
"--hub_model_id",
|
264 |
type=str,
|
|
|
302 |
' (default), `"wandb"` and `"comet_ml"`. Use `"all"` to report to all integrations.'
|
303 |
),
|
304 |
)
|
305 |
+
parser.add_argument(
|
306 |
+
"--local_rank",
|
307 |
+
type=int,
|
308 |
+
default=-1,
|
309 |
+
help="For distributed training: local_rank",
|
310 |
+
)
|
311 |
parser.add_argument(
|
312 |
"--checkpointing_steps",
|
313 |
type=int,
|
|
|
327 |
),
|
328 |
)
|
329 |
parser.add_argument(
|
330 |
+
"--enable_xformers_memory_efficient_attention",
|
331 |
+
action="store_true",
|
332 |
+
help="Whether or not to use xformers.",
|
333 |
)
|
334 |
|
335 |
args = parser.parse_args()
|
|
|
337 |
if env_local_rank != -1 and env_local_rank != args.local_rank:
|
338 |
args.local_rank = env_local_rank
|
339 |
|
340 |
+
# if args.train_data_dir is None:
|
341 |
# raise ValueError("You must specify a train data directory.")
|
342 |
|
343 |
return args
|
|
|
418 |
self.center_crop = center_crop
|
419 |
self.flip_p = flip_p
|
420 |
|
421 |
+
self.image_paths = [
|
422 |
+
os.path.join(self.data_root, file_path)
|
423 |
+
for file_path in os.listdir(self.data_root)
|
424 |
+
]
|
425 |
|
426 |
self.num_images = len(self.image_paths)
|
427 |
self._length = self.num_images
|
|
|
436 |
"lanczos": PIL_INTERPOLATION["lanczos"],
|
437 |
}[interpolation]
|
438 |
|
439 |
+
self.templates = (
|
440 |
+
imagenet_style_templates_small
|
441 |
+
if learnable_property == "style"
|
442 |
+
else imagenet_templates_small
|
443 |
+
)
|
444 |
self.flip_transform = transforms.RandomHorizontalFlip(p=self.flip_p)
|
445 |
|
446 |
def __len__(self):
|
|
|
469 |
|
470 |
if self.center_crop:
|
471 |
crop = min(img.shape[0], img.shape[1])
|
472 |
+
(h, w,) = (
|
|
|
|
|
|
|
473 |
img.shape[0],
|
474 |
img.shape[1],
|
475 |
)
|
476 |
+
img = img[
|
477 |
+
(h - crop) // 2 : (h + crop) // 2, (w - crop) // 2 : (w + crop) // 2
|
478 |
+
]
|
479 |
|
480 |
image = Image.fromarray(img)
|
481 |
image = image.resize((self.size, self.size), resample=self.interpolation)
|
|
|
488 |
return example
|
489 |
|
490 |
|
491 |
+
def get_full_repo_name(
|
492 |
+
model_id: str, organization: Optional[str] = None, token: Optional[str] = None
|
493 |
+
):
|
494 |
if token is None:
|
495 |
token = HfFolder.get_token()
|
496 |
if organization is None:
|
|
|
500 |
return f"{organization}/{model_id}"
|
501 |
|
502 |
|
|
|
503 |
def main(pipe, args_imported):
|
504 |
|
505 |
args = parse_args()
|
|
|
539 |
if accelerator.is_main_process:
|
540 |
if args.push_to_hub:
|
541 |
if args.hub_model_id is None:
|
542 |
+
repo_name = get_full_repo_name(
|
543 |
+
Path(args.output_dir).name, token=args.hub_token
|
544 |
+
)
|
545 |
else:
|
546 |
repo_name = args.hub_model_id
|
547 |
create_repo(repo_name, exist_ok=True, token=args.hub_token)
|
548 |
+
repo = Repository(
|
549 |
+
args.output_dir, clone_from=repo_name, token=args.hub_token
|
550 |
+
)
|
551 |
|
552 |
with open(os.path.join(args.output_dir, ".gitignore"), "w+") as gitignore:
|
553 |
if "step_*" not in gitignore:
|
|
|
609 |
if is_xformers_available():
|
610 |
unet.enable_xformers_memory_efficient_attention()
|
611 |
else:
|
612 |
+
raise ValueError(
|
613 |
+
"xformers is not available. Make sure it is installed correctly"
|
614 |
+
)
|
615 |
|
616 |
# Enable TF32 for faster training on Ampere GPUs,
|
617 |
# cf https://pytorch.org/docs/stable/notes/cuda.html#tensorfloat-32-tf32-on-ampere-devices
|
|
|
620 |
|
621 |
if args.scale_lr:
|
622 |
args.learning_rate = (
|
623 |
+
args.learning_rate
|
624 |
+
* args.gradient_accumulation_steps
|
625 |
+
* args.train_batch_size
|
626 |
+
* accelerator.num_processes
|
627 |
)
|
628 |
|
629 |
# Initialize the optimizer
|
|
|
646 |
center_crop=args.center_crop,
|
647 |
set="train",
|
648 |
)
|
649 |
+
train_dataloader = torch.utils.data.DataLoader(
|
650 |
+
train_dataset, batch_size=args.train_batch_size, shuffle=True
|
651 |
+
)
|
652 |
|
653 |
# Scheduler and math around the number of training steps.
|
654 |
overrode_max_train_steps = False
|
655 |
+
num_update_steps_per_epoch = math.ceil(
|
656 |
+
len(train_dataloader) / args.gradient_accumulation_steps
|
657 |
+
)
|
658 |
if args.max_train_steps is None:
|
659 |
args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
|
660 |
overrode_max_train_steps = True
|
|
|
685 |
text_encoder.to(accelerator.device, dtype=torch.float32)
|
686 |
|
687 |
# We need to recalculate our total training steps as the size of the training dataloader may have changed.
|
688 |
+
num_update_steps_per_epoch = math.ceil(
|
689 |
+
len(train_dataloader) / args.gradient_accumulation_steps
|
690 |
+
)
|
691 |
if overrode_max_train_steps:
|
692 |
args.max_train_steps = args.num_train_epochs * num_update_steps_per_epoch
|
693 |
# Afterwards we recalculate our number of training epochs
|
|
|
699 |
accelerator.init_trackers("textual_inversion", config=vars(args))
|
700 |
|
701 |
# Train!
|
702 |
+
total_batch_size = (
|
703 |
+
args.train_batch_size
|
704 |
+
* accelerator.num_processes
|
705 |
+
* args.gradient_accumulation_steps
|
706 |
+
)
|
707 |
|
708 |
logger.info("***** Running training *****")
|
709 |
logger.info(f" Num examples = {len(train_dataset)}")
|
710 |
logger.info(f" Num Epochs = {args.num_train_epochs}")
|
711 |
logger.info(f" Instantaneous batch size per device = {args.train_batch_size}")
|
712 |
+
logger.info(
|
713 |
+
f" Total train batch size (w. parallel, distributed & accumulation) = {total_batch_size}"
|
714 |
+
)
|
715 |
logger.info(f" Gradient Accumulation steps = {args.gradient_accumulation_steps}")
|
716 |
logger.info(f" Total optimization steps = {args.max_train_steps}")
|
717 |
global_step = 0
|
|
|
736 |
resume_step = resume_global_step % num_update_steps_per_epoch
|
737 |
|
738 |
# Only show the progress bar once on each machine.
|
739 |
+
progress_bar = tqdm(
|
740 |
+
range(global_step, args.max_train_steps),
|
741 |
+
disable=not accelerator.is_local_main_process,
|
742 |
+
)
|
743 |
progress_bar.set_description("Steps")
|
744 |
|
745 |
# keep original embeddings as reference
|
746 |
+
orig_embeds_params = (
|
747 |
+
accelerator.unwrap_model(text_encoder)
|
748 |
+
.get_input_embeddings()
|
749 |
+
.weight.data.clone()
|
750 |
+
)
|
751 |
|
752 |
+
for epoch in range(first_epoch, args.num_train_epochs):
|
753 |
text_encoder.train()
|
754 |
for step, batch in enumerate(train_dataloader):
|
755 |
# Skip steps until we reach the resumed step
|
756 |
+
if (
|
757 |
+
args.resume_from_checkpoint
|
758 |
+
and epoch == first_epoch
|
759 |
+
and step < resume_step
|
760 |
+
):
|
761 |
if step % args.gradient_accumulation_steps == 0:
|
762 |
progress_bar.update(1)
|
763 |
continue
|
764 |
|
765 |
with accelerator.accumulate(text_encoder):
|
766 |
# Convert images to latent space
|
767 |
+
latents = (
|
768 |
+
vae.encode(batch["pixel_values"].to(dtype=weight_dtype))
|
769 |
+
.latent_dist.sample()
|
770 |
+
.detach()
|
771 |
+
)
|
772 |
latents = latents * 0.18215
|
773 |
|
774 |
# Sample noise that we'll add to the latents
|
775 |
noise = torch.randn_like(latents)
|
776 |
bsz = latents.shape[0]
|
777 |
# Sample a random timestep for each image
|
778 |
+
timesteps = torch.randint(
|
779 |
+
0,
|
780 |
+
noise_scheduler.config.num_train_timesteps,
|
781 |
+
(bsz,),
|
782 |
+
device=latents.device,
|
783 |
+
)
|
784 |
timesteps = timesteps.long()
|
785 |
|
786 |
# Add noise to the latents according to the noise magnitude at each timestep
|
|
|
788 |
noisy_latents = noise_scheduler.add_noise(latents, noise, timesteps)
|
789 |
|
790 |
# Get the text embedding for conditioning
|
791 |
+
encoder_hidden_states = text_encoder(batch["input_ids"])[0].to(
|
792 |
+
dtype=weight_dtype
|
793 |
+
)
|
794 |
|
795 |
# Predict the noise residual
|
796 |
+
model_pred = unet(
|
797 |
+
noisy_latents, timesteps, encoder_hidden_states
|
798 |
+
).sample
|
799 |
|
800 |
# Get the target for loss depending on the prediction type
|
801 |
if noise_scheduler.config.prediction_type == "epsilon":
|
|
|
803 |
elif noise_scheduler.config.prediction_type == "v_prediction":
|
804 |
target = noise_scheduler.get_velocity(latents, noise, timesteps)
|
805 |
else:
|
806 |
+
raise ValueError(
|
807 |
+
f"Unknown prediction type {noise_scheduler.config.prediction_type}"
|
808 |
+
)
|
809 |
|
810 |
loss = F.mse_loss(model_pred.float(), target.float(), reduction="mean")
|
811 |
|
|
|
816 |
else:
|
817 |
grads = text_encoder.get_input_embeddings().weight.grad
|
818 |
# Get the index for tokens that we want to zero the grads for
|
819 |
+
index_grads_to_zero = (
|
820 |
+
torch.arange(len(tokenizer)) != placeholder_token_id
|
821 |
+
)
|
822 |
+
grads.data[index_grads_to_zero, :] = grads.data[
|
823 |
+
index_grads_to_zero, :
|
824 |
+
].fill_(0)
|
825 |
|
826 |
optimizer.step()
|
827 |
lr_scheduler.step()
|
|
|
830 |
# Let's make sure we don't update any embedding weights besides the newly added token
|
831 |
index_no_updates = torch.arange(len(tokenizer)) != placeholder_token_id
|
832 |
with torch.no_grad():
|
833 |
+
accelerator.unwrap_model(
|
834 |
+
text_encoder
|
835 |
+
).get_input_embeddings().weight[
|
836 |
+
index_no_updates
|
837 |
+
] = orig_embeds_params[
|
838 |
index_no_updates
|
839 |
+
]
|
840 |
|
841 |
# Checks if the accelerator has performed an optimization step behind the scenes
|
842 |
if accelerator.sync_gradients:
|
843 |
progress_bar.update(1)
|
844 |
global_step += 1
|
845 |
if global_step % args.save_steps == 0:
|
846 |
+
save_path = os.path.join(
|
847 |
+
args.output_dir, f"{args.placeholder_token}-{global_step}.bin"
|
848 |
+
)
|
849 |
+
save_progress(
|
850 |
+
text_encoder, placeholder_token_id, accelerator, args, save_path
|
851 |
+
)
|
852 |
|
853 |
if global_step % args.checkpointing_steps == 0:
|
854 |
if accelerator.is_main_process:
|
855 |
+
save_path = os.path.join(
|
856 |
+
args.output_dir, f"checkpoint-{global_step}"
|
857 |
+
)
|
858 |
accelerator.save_state(save_path)
|
859 |
logger.info(f"Saved state to {save_path}")
|
860 |
|
|
|
869 |
accelerator.wait_for_everyone()
|
870 |
if accelerator.is_main_process:
|
871 |
if args.push_to_hub and args.only_save_embeds:
|
872 |
+
logger.warn(
|
873 |
+
"Enabling full model saving because --push_to_hub=True was specified."
|
874 |
+
)
|
875 |
save_full_model = True
|
876 |
else:
|
877 |
save_full_model = not args.only_save_embeds
|
|
|
882 |
save_progress(text_encoder, placeholder_token_id, accelerator, args, save_path)
|
883 |
|
884 |
if args.push_to_hub:
|
885 |
+
repo.push_to_hub(
|
886 |
+
commit_message="End of training", blocking=False, auto_lfs_prune=True
|
887 |
+
)
|
888 |
|
889 |
accelerator.end_training()
|
|
|
|
|
|
|
890 |
|
891 |
|
892 |
if __name__ == "__main__":
|
893 |
+
pipeline = StableDiffusionPipeline.from_pretrained(
|
894 |
+
"andite/anything-v4.0", torch_dtype=torch.float16
|
895 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
896 |
|
897 |
+
imported_args = argparse.Namespace(
|
898 |
+
train_data_dir="concept_images",
|
899 |
+
learnable_property="object",
|
900 |
+
placeholder_token="redeyegirl",
|
901 |
+
initializer_token="girl",
|
902 |
+
resolution=512,
|
903 |
+
train_batch_size=1,
|
904 |
+
gradient_accumulation_steps=1,
|
905 |
+
gradient_checkpointing=True,
|
906 |
+
mixed_precision="fp16",
|
907 |
+
use_bf16=False,
|
908 |
+
max_train_steps=1000,
|
909 |
+
learning_rate=5.0e-4,
|
910 |
+
scale_lr=False,
|
911 |
+
lr_scheduler="constant",
|
912 |
+
lr_warmup_steps=0,
|
913 |
+
output_dir="output_model",
|
914 |
+
)
|
915 |
|
916 |
main(pipeline, imported_args)
|