-
Notifications
You must be signed in to change notification settings - Fork 0
/
genai_model_[nxtwave].py
1098 lines (885 loc) · 424 KB
/
genai_model_[nxtwave].py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# -*- coding: utf-8 -*-
"""GenAI Model [NxtWave].ipynb
Automatically generated by Colaboratory.
Original file is located at
https://colab.research.google.com/drive/1j4fHUUbNBtFA3VrNJqcz36L9R7YKUely
# Mounting Your Google Drive
"""
from google.colab import drive
drive.mount('/content/gdrive')
"""# Dependencies Installation"""
# Commented out IPython magic to ensure Python compatibility.
#@markdown #Dependencies
from IPython.utils import capture
import time
import os
print('[1;32mInstalling dependencies...')
with capture.capture_output() as cap:
# %cd /content/
!pip install -qq --no-deps accelerate==0.12.0
!wget -q -i https://raw.githubusercontent.com/TheLastBen/fast-stable-diffusion/main/Dependencies/dbdeps.txt
!dpkg -i *.deb
!tar -C / --zstd -xf gcolabdeps.tar.zst
!rm *.deb | rm *.zst | rm *.txt
!git clone -q --depth 1 --branch main https://github.com/TheLastBen/diffusers
!pip install gradio==3.16.2 --no-deps -qq
# %env LD_PRELOAD=libtcmalloc.so
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
os.environ['PYTHONWARNINGS'] = 'ignore'
print('[1;32mDone, proceed')
"""# Base Model Download"""
# Commented out IPython magic to ensure Python compatibility.
import os
import time
from IPython.utils import capture
from IPython.display import clear_output
import wget
from subprocess import check_output
import urllib.request
import base64
#@markdown - Skip this cell if you are loading a previous session that contains a trained model.
#@markdown ---
Model_Version = "1.5" #@param [ "1.5", "V2.1-512px", "V2.1-768px"]
#@markdown - Choose which version to finetune.
with capture.capture_output() as cap:
# %cd /content/
modelnm="model.ckpt"
if os.path.exists('/content/gdrive/MyDrive/Fast-Dreambooth/token.txt'):
with open("/content/gdrive/MyDrive/Fast-Dreambooth/token.txt") as f:
token = f.read()
authe=f'https://USER:{token}@'
else:
authe="https://"
def downloadmodel():
if os.path.exists('/content/stable-diffusion-v1-5'):
!rm -r /content/stable-diffusion-v1-5
clear_output()
# %cd /content/
clear_output()
!mkdir /content/stable-diffusion-v1-5
# %cd /content/stable-diffusion-v1-5
!git init
!git lfs install --system --skip-repo
!git remote add -f origin "https://huggingface.co/runwayml/stable-diffusion-v1-5"
!git config core.sparsecheckout true
!echo -e "scheduler\ntext_encoder\ntokenizer\nunet\nvae\nmodel_index.json\n!vae/diffusion_pytorch_model.bin\n!*.safetensors" > .git/info/sparse-checkout
!git pull origin main
if os.path.exists('/content/stable-diffusion-v1-5/unet/diffusion_pytorch_model.bin'):
!wget -q -O vae/diffusion_pytorch_model.bin https://huggingface.co/stabilityai/sd-vae-ft-mse/resolve/main/diffusion_pytorch_model.bin
!rm -r .git
!rm model_index.json
time.sleep(1)
wget.download('https://raw.githubusercontent.com/TheLastBen/fast-stable-diffusion/main/Dreambooth/model_index.json')
# %cd /content/
clear_output()
print('[1;32mDONE !')
else:
while not os.path.exists('/content/stable-diffusion-v1-5/unet/diffusion_pytorch_model.bin'):
print('[1;31mSomething went wrong')
time.sleep(5)
def newdownloadmodel():
# %cd /content/
clear_output()
!mkdir /content/stable-diffusion-v2-768
# %cd /content/stable-diffusion-v2-768
!git init
!git lfs install --system --skip-repo
!git remote add -f origin "https://huggingface.co/stabilityai/stable-diffusion-2-1"
!git config core.sparsecheckout true
!echo -e "scheduler\ntext_encoder\ntokenizer\nunet\nvae\nfeature_extractor\nmodel_index.json\n!*.safetensors" > .git/info/sparse-checkout
!git pull origin main
!rm -r /content/stable-diffusion-v2-768/.git
# %cd /content/
clear_output()
print('[1;32mDONE !')
def newdownloadmodelb():
# %cd /content/
clear_output()
!mkdir /content/stable-diffusion-v2-512
# %cd /content/stable-diffusion-v2-512
!git init
!git lfs install --system --skip-repo
!git remote add -f origin "https://huggingface.co/stabilityai/stable-diffusion-2-1-base"
!git config core.sparsecheckout true
!echo -e "scheduler\ntext_encoder\ntokenizer\nunet\nvae\nfeature_extractor\nmodel_index.json\n!*.safetensors" > .git/info/sparse-checkout
!git pull origin main
!rm -r /content/stable-diffusion-v2-512/.git
# %cd /content/
clear_output()
print('[1;32mDONE !')
if Model_Version=="1.5":
if not os.path.exists('/content/stable-diffusion-v1-5'):
downloadmodel()
MODEL_NAME="/content/stable-diffusion-v1-5"
else:
MODEL_NAME="/content/stable-diffusion-v1-5"
print("[1;32mThe v1.5 model already exists, using this model.")
elif Model_Version=="V2.1-512px":
if not os.path.exists('/content/stable-diffusion-v2-512'):
newdownloadmodelb()
MODEL_NAME="/content/stable-diffusion-v2-512"
else:
MODEL_NAME="/content/stable-diffusion-v2-512"
print("[1;32mThe v2-512px model already exists, using this model.")
elif Model_Version=="V2.1-768px":
if not os.path.exists('/content/stable-diffusion-v2-768'):
newdownloadmodel()
MODEL_NAME="/content/stable-diffusion-v2-768"
else:
MODEL_NAME="/content/stable-diffusion-v2-768"
print("[1;32mThe v2-768px model already exists, using this model.")
"""# Dreambooth Fine Tuning Setup
To learn more about dreambooth - https://dreambooth.github.io/
"""
# Commented out IPython magic to ensure Python compatibility.
import os
from IPython.display import clear_output
from IPython.utils import capture
from os import listdir
from os.path import isfile
from subprocess import check_output
import wget
import time
#@markdown #Create/Load a Session
try:
MODEL_NAME
pass
except:
MODEL_NAME=""
PT=""
Session_Name = "nxw" #@param{type: 'string'}
while Session_Name=="":
print('[1;31mInput the Session Name:')
Session_Name=input('')
Session_Name=Session_Name.replace(" ","_")
#@markdown - Enter the session name, it if it exists, it will load it, otherwise it'll create an new session.
WORKSPACE='/content/gdrive/MyDrive/Fast-Dreambooth'
INSTANCE_NAME=Session_Name
OUTPUT_DIR="/content/models/"+Session_Name
SESSION_DIR=WORKSPACE+'/Sessions/'+Session_Name
INSTANCE_DIR=SESSION_DIR+'/instance_images'
CAPTIONS_DIR=SESSION_DIR+'/captions'
MDLPTH=str(SESSION_DIR+"/"+Session_Name+'.ckpt')
if os.path.exists(str(SESSION_DIR)):
mdls=[ckpt for ckpt in listdir(SESSION_DIR) if ckpt.split(".")[-1]=="ckpt"]
if not os.path.exists(MDLPTH) and '.ckpt' in str(mdls):
def f(n):
k=0
for i in mdls:
if k==n:
!mv "$SESSION_DIR/$i" $MDLPTH
k=k+1
k=0
print('[1;33mNo final checkpoint model found, select which intermediary checkpoint to use, enter only the number, (000 to skip):\n[1;34m')
for i in mdls:
print(str(k)+'- '+i)
k=k+1
n=input()
while int(n)>k-1:
n=input()
if n!="000":
f(int(n))
print('[1;32mUsing the model '+ mdls[int(n)]+" ...")
time.sleep(2)
else:
print('[1;32mSkipping the intermediary checkpoints.')
del n
with capture.capture_output() as cap:
# %cd /content
resume=False
if os.path.exists(str(SESSION_DIR)) and not os.path.exists(MDLPTH):
print('[1;32mLoading session with no previous model, using the original model or the custom downloaded model')
if MODEL_NAME=="":
print('[1;31mNo model found, use the "Model Download" cell to download a model.')
else:
print('[1;32mSession Loaded, proceed to uploading instance images')
elif os.path.exists(MDLPTH):
print('[1;32mSession found, loading the trained model ...')
wget.download('https://github.com/TheLastBen/fast-stable-diffusion/raw/main/Dreambooth/det.py')
print('[1;33mDetecting model version...')
Model_Version=check_output('python det.py --MODEL_PATH '+MDLPTH, shell=True).decode('utf-8').replace('\n', '')
clear_output()
print('[1;32m'+Model_Version+' Detected')
!rm det.py
if Model_Version=='1.5':
!wget -q -O config.yaml https://github.com/CompVis/stable-diffusion/raw/main/configs/stable-diffusion/v1-inference.yaml
print('[1;32mSession found, loading the trained model ...')
!python /content/diffusers/scripts/convert_original_stable_diffusion_to_diffusers.py --checkpoint_path $MDLPTH --dump_path "$OUTPUT_DIR" --original_config_file config.yaml
!rm /content/config.yaml
elif Model_Version=='V2.1-512px':
!wget -q -O convertodiff.py https://raw.githubusercontent.com/TheLastBen/fast-stable-diffusion/main/Dreambooth/convertodiffv2.py
print('[1;32mSession found, loading the trained model ...')
!python /content/convertodiff.py "$MDLPTH" "$OUTPUT_DIR" --v2 --reference_model stabilityai/stable-diffusion-2-1-base
!rm /content/convertodiff.py
elif Model_Version=='V2.1-768px':
!wget -q -O convertodiff.py https://github.com/TheLastBen/fast-stable-diffusion/raw/main/Dreambooth/convertodiffv2-768.py
print('[1;32mSession found, loading the trained model ...')
!python /content/convertodiff.py "$MDLPTH" "$OUTPUT_DIR" --v2 --reference_model stabilityai/stable-diffusion-2-1
!rm /content/convertodiff.py
if os.path.exists(OUTPUT_DIR+'/unet/diffusion_pytorch_model.bin'):
resume=True
clear_output()
print('[1;32mSession loaded.')
else:
if not os.path.exists(OUTPUT_DIR+'/unet/diffusion_pytorch_model.bin'):
print('[1;31mConversion error, if the error persists, remove the CKPT file from the current session folder')
elif not os.path.exists(str(SESSION_DIR)):
# %mkdir -p "$INSTANCE_DIR"
print('[1;32mCreating session...')
if MODEL_NAME=="":
print('[1;31mNo model found, use the "Model Download" cell to download a model.')
else:
print('[1;32mSession created, proceed to uploading instance images')
#@markdown
#@markdown # The most important step is to rename the instance pictures of each subject to a unique unknown identifier, example :
#@markdown - If you have 10 pictures of your dog, simply select them all and rename only one to the chosen identifier for example : xzg, the files would be : xzg (1).jpg, xzg (2).png ....etc then upload them, do the same for other people or objects with a different identifier, and that's it.
#@markdown - If you are not able to collect good quality images on your own for finetuning, you can use the samples collected by dreambooth project creators from this link - https://github.com/google/dreambooth/tree/main/dataset. (You can download the images in any one of the folders from the link)
# Commented out IPython magic to ensure Python compatibility.
import shutil
from google.colab import files
import time
from PIL import Image
from tqdm import tqdm
import ipywidgets as widgets
from io import BytesIO
import wget
with capture.capture_output() as cap:
# %cd /content
if not os.path.exists("/content/smart_crop.py"):
wget.download('https://raw.githubusercontent.com/TheLastBen/fast-stable-diffusion/main/Dreambooth/smart_crop.py')
from smart_crop import *
#@markdown #Instance Images
#@markdown ----
#@markdown
#@markdown - Run the cell to upload the instance pictures.
#@markdown - You can add `external captions` in txt files by simply giving each txt file the same name as the instance image, for example dikgur (1).jpg and dikgur (1).txt, and upload them here, to use the external captions, check the box "external_captions" in the training cell. `All the images must have one same extension` jpg or png or....etc
Remove_existing_instance_images= True #@param{type: 'boolean'}
#@markdown - Uncheck the box to keep the existing instance images.
if Remove_existing_instance_images:
if os.path.exists(str(INSTANCE_DIR)):
!rm -r "$INSTANCE_DIR"
if os.path.exists(str(CAPTIONS_DIR)):
!rm -r "$CAPTIONS_DIR"
if not os.path.exists(str(INSTANCE_DIR)):
# %mkdir -p "$INSTANCE_DIR"
if not os.path.exists(str(CAPTIONS_DIR)):
# %mkdir -p "$CAPTIONS_DIR"
if os.path.exists(INSTANCE_DIR+"/.ipynb_checkpoints"):
# %rm -r $INSTANCE_DIR"/.ipynb_checkpoints"
IMAGES_FOLDER_OPTIONAL="" #@param{type: 'string'}
#@markdown - If you prefer to specify directly the folder of the pictures instead of uploading, this will add the pictures to the existing (if any) instance images. Leave EMPTY to upload.
Smart_Crop_images= False #@param{type: 'boolean'}
Crop_size = 512 #@param ["512", "576", "640", "704", "768", "832", "896", "960", "1024"] {type:"raw"}
#@markdown - Smart crop the images without manual intervention.
while IMAGES_FOLDER_OPTIONAL !="" and not os.path.exists(str(IMAGES_FOLDER_OPTIONAL)):
print('[1;31mThe image folder specified does not exist, use the colab file explorer to copy the path :')
IMAGES_FOLDER_OPTIONAL=input('')
if IMAGES_FOLDER_OPTIONAL!="":
if os.path.exists(IMAGES_FOLDER_OPTIONAL+"/.ipynb_checkpoints"):
# %rm -r "$IMAGES_FOLDER_OPTIONAL""/.ipynb_checkpoints"
with capture.capture_output() as cap:
!mv $IMAGES_FOLDER_OPTIONAL/*.txt $CAPTIONS_DIR
if Smart_Crop_images:
for filename in tqdm(os.listdir(IMAGES_FOLDER_OPTIONAL), bar_format=' |{bar:15}| {n_fmt}/{total_fmt} Uploaded'):
extension = filename.split(".")[-1]
identifier=filename.split(".")[0]
new_path_with_file = os.path.join(INSTANCE_DIR, filename)
file = Image.open(IMAGES_FOLDER_OPTIONAL+"/"+filename)
width, height = file.size
if file.size !=(Crop_size, Crop_size):
image=crop_image(file, Crop_size)
if extension.upper()=="JPG" or extension.upper()=="jpg":
image[0] = image[0].convert("RGB")
image[0].save(new_path_with_file, format="JPEG", quality = 100)
else:
image[0].save(new_path_with_file, format=extension.upper())
else:
!cp "$IMAGES_FOLDER_OPTIONAL/$filename" "$INSTANCE_DIR"
else:
for filename in tqdm(os.listdir(IMAGES_FOLDER_OPTIONAL), bar_format=' |{bar:15}| {n_fmt}/{total_fmt} Uploaded'):
# %cp -r "$IMAGES_FOLDER_OPTIONAL/$filename" "$INSTANCE_DIR"
print('\n[1;32mDone, proceed to the next cell')
elif IMAGES_FOLDER_OPTIONAL =="":
up=""
uploaded = files.upload()
for filename in uploaded.keys():
if filename.split(".")[-1]=="txt":
shutil.move(filename, CAPTIONS_DIR)
up=[filename for filename in uploaded.keys() if filename.split(".")[-1]!="txt"]
if Smart_Crop_images:
for filename in tqdm(up, bar_format=' |{bar:15}| {n_fmt}/{total_fmt} Uploaded'):
shutil.move(filename, INSTANCE_DIR)
extension = filename.split(".")[-1]
identifier=filename.split(".")[0]
new_path_with_file = os.path.join(INSTANCE_DIR, filename)
file = Image.open(new_path_with_file)
width, height = file.size
if file.size !=(Crop_size, Crop_size):
image=crop_image(file, Crop_size)
if extension.upper()=="JPG" or extension.upper()=="jpg":
image[0] = image[0].convert("RGB")
image[0].save(new_path_with_file, format="JPEG", quality = 100)
else:
image[0].save(new_path_with_file, format=extension.upper())
clear_output()
else:
for filename in tqdm(uploaded.keys(), bar_format=' |{bar:15}| {n_fmt}/{total_fmt} Uploaded'):
shutil.move(filename, INSTANCE_DIR)
clear_output()
print('\n[1;32mDone, proceed to the next cell')
with capture.capture_output() as cap:
# %cd "$INSTANCE_DIR"
!find . -name "* *" -type f | rename 's/ /-/g'
# %cd "$CAPTIONS_DIR"
!find . -name "* *" -type f | rename 's/ /-/g'
# %cd $SESSION_DIR
!rm instance_images.zip captions.zip
!zip -r instance_images instance_images
!zip -r captions captions
# %cd /content
import ipywidgets as widgets
from io import BytesIO
#@markdown #Captions (optional)
#@markdown - Open a tool to manually `create` captions or edit existing captions of the instance images, do not use captions when training on a face.
paths=""
out=""
widgets_l=""
clear_output()
def Caption(path):
if path!="Select an instance image to caption":
name = os.path.splitext(os.path.basename(path))[0]
ext=os.path.splitext(os.path.basename(path))[-1][1:]
if ext=="jpg" or "JPG":
ext="JPEG"
if os.path.exists(CAPTIONS_DIR+"/"+name + '.txt'):
with open(CAPTIONS_DIR+"/"+name + '.txt', 'r') as f:
text = f.read()
else:
with open(CAPTIONS_DIR+"/"+name + '.txt', 'w') as f:
f.write("")
with open(CAPTIONS_DIR+"/"+name + '.txt', 'r') as f:
text = f.read()
img=Image.open(os.path.join(INSTANCE_DIR,path))
img=img.convert("RGB")
img=img.resize((420, 420))
image_bytes = BytesIO()
img.save(image_bytes, format=ext, qualiy=10)
image_bytes.seek(0)
image_data = image_bytes.read()
img= image_data
image = widgets.Image(
value=img,
width=420,
height=420
)
text_area = widgets.Textarea(value=text, description='', disabled=False, layout={'width': '300px', 'height': '120px'})
def update_text(text):
with open(CAPTIONS_DIR+"/"+name + '.txt', 'w') as f:
f.write(text)
button = widgets.Button(description='Save', button_style='success')
button.on_click(lambda b: update_text(text_area.value))
return widgets.VBox([widgets.HBox([image, text_area, button])])
paths = os.listdir(INSTANCE_DIR)
widgets_l = widgets.Select(options=["Select an instance image to caption"]+paths, rows=25)
out = widgets.Output()
def click(change):
with out:
out.clear_output()
display(Caption(change.new))
widgets_l.observe(click, names='value')
display(widgets.HBox([widgets_l, out]))
"""# Training"""
# Commented out IPython magic to ensure Python compatibility.
#@markdown ---
#@markdown #Start DreamBooth
#@markdown ---
import os
from IPython.display import clear_output
from google.colab import runtime
from subprocess import getoutput
import time
import random
if os.path.exists(INSTANCE_DIR+"/.ipynb_checkpoints"):
# %rm -r $INSTANCE_DIR"/.ipynb_checkpoints"
if os.path.exists(CAPTIONS_DIR+"/.ipynb_checkpoints"):
# %rm -r $CAPTIONS_DIR"/.ipynb_checkpoints"
Resume_Training = False #@param {type:"boolean"}
if resume and not Resume_Training:
print('[1;31mOverwrite your previously trained model ? answering "yes" will train a new model, answering "no" will resume the training of the previous model? yes or no ?[0m')
while True:
ansres=input('')
if ansres=='no':
Resume_Training = True
break
elif ansres=='yes':
Resume_Training = False
resume= False
break
while not Resume_Training and MODEL_NAME=="":
print('[1;31mNo model found, use the "Model Download" cell to download a model.')
time.sleep(5)
#@markdown - If you're not satisfied with the result, check this box, run again the cell and it will continue training the current model.
MODELT_NAME=MODEL_NAME
UNet_Training_Steps=1200 #@param{type: 'number'}
UNet_Learning_Rate = 2e-6 #@param ["2e-5","1e-5","9e-6","8e-6","7e-6","6e-6","5e-6", "4e-6", "3e-6", "2e-6"] {type:"raw"}
untlr=UNet_Learning_Rate
#@markdown - These default settings are for a dataset of 10 pictures which is enough for training a face, start with 1500 or lower, test the model, if not enough, resume training for 200 steps, keep testing until you get the desired output, `set it to 0 to train only the text_encoder`.
Text_Encoder_Training_Steps=350 #@param{type: 'number'}
#@markdown - 200-450 steps is enough for a small dataset, keep this number small to avoid overfitting, set to 0 to disable, `set it to 0 before resuming training if it is already trained`.
Text_Encoder_Learning_Rate = 1e-6 #@param ["2e-6", "1e-6","8e-7","6e-7","5e-7","4e-7"] {type:"raw"}
txlr=Text_Encoder_Learning_Rate
#@markdown - Learning rate for the text_encoder, keep it low to avoid overfitting (1e-6 is higher than 4e-7)
trnonltxt=""
if UNet_Training_Steps==0:
trnonltxt="--train_only_text_encoder"
Seed=''
ofstnse=""
Offset_Noise = False #@param {type:"boolean"}
#@markdown - Always use it for style training.
if Offset_Noise:
ofstnse="--offset_noise"
External_Captions = False #@param {type:"boolean"}
#@markdown - Get the captions from a text file for each instance image.
extrnlcptn=""
if External_Captions:
extrnlcptn="--external_captions"
Resolution = "512" #@param ["512", "576", "640", "704", "768", "832", "896", "960", "1024"]
Res=int(Resolution)
#@markdown - Higher resolution = Higher quality, make sure the instance images are cropped to this selected size (or larger).
fp16 = True
if Seed =='' or Seed=='0':
Seed=random.randint(1, 999999)
else:
Seed=int(Seed)
if fp16:
prec="fp16"
else:
prec="no"
precision=prec
resuming=""
if Resume_Training and os.path.exists(OUTPUT_DIR+'/unet/diffusion_pytorch_model.bin'):
MODELT_NAME=OUTPUT_DIR
print('[1;32mResuming Training...[0m')
resuming="Yes"
elif Resume_Training and not os.path.exists(OUTPUT_DIR+'/unet/diffusion_pytorch_model.bin'):
print('[1;31mPrevious model not found, training a new model...[0m')
MODELT_NAME=MODEL_NAME
while MODEL_NAME=="":
print('[1;31mNo model found, use the "Model Download" cell to download a model.')
time.sleep(5)
V2=False
if os.path.getsize(MODELT_NAME+"/text_encoder/pytorch_model.bin") > 670901463:
V2=True
s = getoutput('nvidia-smi')
GCUNET="--gradient_checkpointing"
TexRes=Res
if Res<=768:
GCUNET=""
if V2:
if Res>704:
GCUNET="--gradient_checkpointing"
if Res>576:
TexRes=576
if 'A100' in s :
GCUNET=""
TexRes=Res
Enable_text_encoder_training= True
if Text_Encoder_Training_Steps==0 :
Enable_text_encoder_training= False
else:
stptxt=Text_Encoder_Training_Steps
#@markdown ---------------------------
Save_Checkpoint_Every_n_Steps = False #@param {type:"boolean"}
Save_Checkpoint_Every=500 #@param{type: 'number'}
if Save_Checkpoint_Every==None:
Save_Checkpoint_Every=1
#@markdown - Minimum 200 steps between each save.
stp=0
Start_saving_from_the_step=500 #@param{type: 'number'}
if Start_saving_from_the_step==None:
Start_saving_from_the_step=0
if (Start_saving_from_the_step < 200):
Start_saving_from_the_step=Save_Checkpoint_Every
stpsv=Start_saving_from_the_step
if Save_Checkpoint_Every_n_Steps:
stp=Save_Checkpoint_Every
#@markdown - Start saving intermediary checkpoints from this step.
Disconnect_after_training=False #@param {type:"boolean"}
#@markdown - Auto-disconnect from google colab after the training to avoid wasting compute units.
def dump_only_textenc(trnonltxt, MODELT_NAME, INSTANCE_DIR, OUTPUT_DIR, PT, Seed, precision, Training_Steps):
!accelerate launch /content/diffusers/examples/dreambooth/train_dreambooth.py \
$trnonltxt \
$extrnlcptn \
$ofstnse \
--image_captions_filename \
--train_text_encoder \
--dump_only_text_encoder \
--pretrained_model_name_or_path="$MODELT_NAME" \
--instance_data_dir="$INSTANCE_DIR" \
--output_dir="$OUTPUT_DIR" \
--captions_dir="$CAPTIONS_DIR" \
--instance_prompt="$PT" \
--seed=$Seed \
--resolution=$TexRes \
--mixed_precision=$precision \
--train_batch_size=1 \
--gradient_accumulation_steps=1 --gradient_checkpointing \
--use_8bit_adam \
--learning_rate=$txlr \
--lr_scheduler="linear" \
--lr_warmup_steps=0 \
--max_train_steps=$Training_Steps
def train_only_unet(stpsv, stp, SESSION_DIR, MODELT_NAME, INSTANCE_DIR, OUTPUT_DIR, PT, Seed, Res, precision, Training_Steps):
clear_output()
if resuming=="Yes":
print('[1;32mResuming Training...[0m')
print('[1;33mTraining the UNet...[0m')
!accelerate launch /content/diffusers/examples/dreambooth/train_dreambooth.py \
$extrnlcptn \
$ofstnse \
--image_captions_filename \
--train_only_unet \
--save_starting_step=$stpsv \
--save_n_steps=$stp \
--Session_dir=$SESSION_DIR \
--pretrained_model_name_or_path="$MODELT_NAME" \
--instance_data_dir="$INSTANCE_DIR" \
--output_dir="$OUTPUT_DIR" \
--captions_dir="$CAPTIONS_DIR" \
--instance_prompt="$PT" \
--seed=$Seed \
--resolution=$Res \
--mixed_precision=$precision \
--train_batch_size=1 \
--gradient_accumulation_steps=1 $GCUNET \
--use_8bit_adam \
--learning_rate=$untlr \
--lr_scheduler="linear" \
--lr_warmup_steps=0 \
--max_train_steps=$Training_Steps
if Enable_text_encoder_training :
print('[1;33mTraining the text encoder...[0m')
if os.path.exists(OUTPUT_DIR+'/'+'text_encoder_trained'):
# %rm -r $OUTPUT_DIR"/text_encoder_trained"
dump_only_textenc(trnonltxt, MODELT_NAME, INSTANCE_DIR, OUTPUT_DIR, PT, Seed, precision, Training_Steps=stptxt)
if UNet_Training_Steps!=0:
train_only_unet(stpsv, stp, SESSION_DIR, MODELT_NAME, INSTANCE_DIR, OUTPUT_DIR, PT, Seed, Res, precision, Training_Steps=UNet_Training_Steps)
if UNet_Training_Steps==0 and Text_Encoder_Training_Steps==0 :
print('[1;32mNothing to do')
else:
if os.path.exists('/content/models/'+INSTANCE_NAME+'/unet/diffusion_pytorch_model.bin'):
prc="--fp16" if precision=="fp16" else ""
!python /content/diffusers/scripts/convertosdv2.py $prc $OUTPUT_DIR $SESSION_DIR/$Session_Name".ckpt"
clear_output()
if os.path.exists(SESSION_DIR+"/"+INSTANCE_NAME+'.ckpt'):
clear_output()
print("[1;32mDONE, the CKPT model is in your Gdrive in the sessions folder")
if Disconnect_after_training :
time.sleep(20)
runtime.unassign()
else:
print("[1;31mSomething went wrong")
else:
print("[1;31mSomething went wrong")
"""* Reference on Finetuning Parameters: [Training Stable Diffusion with Dreambooth](https://wandb.ai/psuraj/dreambooth/reports/Training-Stable-Diffusion-with-Dreambooth--VmlldzoyNzk0NDc3)
# Test Your Model
"""
!curl ipv4.icanhazip.com
# Commented out IPython magic to ensure Python compatibility.
import os
import time
import sys
import fileinput
from IPython.display import clear_output
from subprocess import getoutput
from IPython.utils import capture
from pyngrok import ngrok, conf
import base64
blasphemy=base64.b64decode(("d2VidWk=")).decode('ascii')
Previous_Session="" #@param{type: 'string'}
#@markdown - Leave empty if you want to use the current trained model.
Use_Custom_Path = False #@param {type:"boolean"}
try:
INSTANCE_NAME
INSTANCET=INSTANCE_NAME
except:
pass
#@markdown - if checked, an input box will ask the full path to a desired model.
if Previous_Session!="":
INSTANCET=Previous_Session
INSTANCET=INSTANCET.replace(" ","_")
if Use_Custom_Path:
try:
INSTANCET
del INSTANCET
except:
pass
try:
INSTANCET
if Previous_Session!="":
path_to_trained_model='/content/gdrive/MyDrive/Fast-Dreambooth/Sessions/'+Previous_Session+"/"+Previous_Session+'.ckpt'
else:
path_to_trained_model=SESSION_DIR+"/"+INSTANCET+'.ckpt'
except:
print('[1;31mIt seems that you did not perform training during this session [1;32mor you chose to use a custom path,\nprovide the full path to the model (including the name of the model):\n')
path_to_trained_model=input()
while not os.path.exists(path_to_trained_model):
print("[1;31mThe model doesn't exist on you Gdrive, use the file explorer to get the path : ")
path_to_trained_model=input()
fgitclone = "git clone --depth 1"
with capture.capture_output() as cap:
if not os.path.exists('/content/gdrive/MyDrive'):
!mkdir -p /content/gdrive/MyDrive
if not os.path.exists('/content/gdrive/MyDrive/sd/stablediffusion'):
!wget -q -O /content/sd_rep.tar.zst https://huggingface.co/TheLastBen/dependencies/resolve/main/sd_rep.tar.zst
!tar -C /content/gdrive/MyDrive --zstd -xf /content/sd_rep.tar.zst
!rm /content/sd_rep.tar.zst
clear_output()
with capture.capture_output() as cap:
# %cd /content/gdrive/MyDrive/sd
!git clone -q --branch master https://github.com/AUTOMATIC1111/stable-diffusion-$blasphemy
# %cd stable-diffusion-$blasphemy
!mkdir cache
!sed -i 's@~/.cache@/content/gdrive/MyDrive/sd/stable-diffusion-{blasphemy}/cache@' /usr/local/lib/python3.10/dist-packages/transformers/utils/hub.py
clear_output()
!git reset --hard
time.sleep(1)
!rm webui.sh
!git pull
!git fetch --unshallow
!git checkout a9eab236d7e8afa4d6205127904a385b2c43bb24
with capture.capture_output() as cap:
if not os.path.exists('/tools/node/bin/lt'):
!npm install -g localtunnel
Use_localtunnel = True #@param {type:"boolean"}
auth=""
with capture.capture_output() as cap:
# %cd modules
!wget -q -O paths.py https://github.com/TheLastBen/fast-stable-diffusion/raw/5632d2ef7fffd940976538d270854ec4faf26855/AUTOMATIC1111_files/paths.py
!wget -q -O extras.py https://github.com/AUTOMATIC1111/stable-diffusion-$blasphemy/raw/a9eab236d7e8afa4d6205127904a385b2c43bb24/modules/extras.py
!wget -q -O sd_models.py https://github.com/AUTOMATIC1111/stable-diffusion-$blasphemy/raw/a9eab236d7e8afa4d6205127904a385b2c43bb24/modules/sd_models.py
!wget -q -O /usr/local/lib/python3.10/dist-packages/gradio/blocks.py https://github.com/TheLastBen/fast-stable-diffusion/raw/7ff88eaa1fb4997bacd9845bd487f9a14335d625/AUTOMATIC1111_files/blocks.py
# %cd /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/
!sed -i "s@os.path.splitext(checkpoint_file)@os.path.splitext(checkpoint_file); map_location='cuda'@" /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/modules/sd_models.py
!sed -i 's@ui.create_ui().*@ui.create_ui();shared.demo.queue(concurrency_count=999999,status_update_rate=0.1)@' /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/webui.py
!sed -i "s@map_location='cpu'@map_location='cuda'@" /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/modules/extras.py
!sed -i 's@print(\"No module.*@@' /content/gdrive/MyDrive/sd/stablediffusion/ldm/modules/diffusionmodules/model.py
!sed -i 's@\"quicksettings\": OptionInfo(.*@"quicksettings": OptionInfo("sd_model_checkpoint, sd_vae, CLIP_stop_at_last_layers, inpainting_mask_weight, initial_noise_multiplier", "Quicksettings list"),@' /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/modules/shared.py
share=''
if Use_localtunnel:
with capture.capture_output() as cap:
share=''
# %cd /content
!nohup lt --port 7860 > srv.txt 2>&1 &
time.sleep(2)
!grep -o 'https[^ ]*' /content/srv.txt >srvr.txt
time.sleep(2)
srv= getoutput('cat /content/srvr.txt')
for line in fileinput.input('/usr/local/lib/python3.10/dist-packages/gradio/blocks.py', inplace=True):
if line.strip().startswith('self.server_name ='):
line = f' self.server_name = "{srv[8:]}"\n'
if line.strip().startswith('self.protocol = "https"'):
line = ' self.protocol = "https"\n'
if line.strip().startswith('if self.local_url.startswith("https") or self.is_colab'):
line = ''
if line.strip().startswith('else "http"'):
line = ''
sys.stdout.write(line)
!rm /content/srv.txt /content/srvr.txt
# %cd /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy
else:
share='--share'
configf="--api --disable-safe-unpickle --enable-insecure-extension-access --no-half-vae --opt-sdp-attention --no-download-sd-model --disable-console-progressbars"
clear_output()
if os.path.isfile(path_to_trained_model):
!python /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/webui.py $share --ckpt "$path_to_trained_model" $auth $configf
else:
!python /content/gdrive/MyDrive/sd/stable-diffusion-$blasphemy/webui.py $share --ckpt-dir "$path_to_trained_model" $auth $configf
"""# Upload Trained Model to Hugging Face"""
# Commented out IPython magic to ensure Python compatibility.
from slugify import slugify
from huggingface_hub import HfApi, HfFolder, CommitOperationAdd
from huggingface_hub import create_repo
from IPython.display import display_markdown
from IPython.display import clear_output
from IPython.utils import capture
from google.colab import files
import shutil
import time
import os
Upload_sample_images = True #@param {type:"boolean"}
#@markdown - Upload showcase images of your trained model
Name_of_your_concept = "Sports Car" #@param {type:"string"}
if(Name_of_your_concept == ""):
Name_of_your_concept = Session_Name
Name_of_your_concept=Name_of_your_concept.replace(" ","-")
Your_Unique_Project_Submission_Code = "20CS07" #@param {type:"string"}
if(Your_Unique_Project_Submission_Code == ""):
Your_Unique_Project_Submission_Code = "NXTWave Build Your Own GenAI Model Session"
Your_Unique_Project_Submission_Code=Your_Unique_Project_Submission_Code.replace(" ","-")
#@markdown - [Create a write access token](https://huggingface.co/settings/tokens) , go to "New token" -> Role : Write. A regular read token won't work here.
hf_token_write = "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" #@param {type:"string"}
if hf_token_write =="":
print('[1;32mYour Hugging Face write access token : ')
hf_token_write=input()
hf_token = hf_token_write
api = HfApi()
your_username = api.whoami(token=hf_token)["name"]
repo_id = f"{your_username}/{slugify(Name_of_your_concept)}"
output_dir = f'/content/models/'+INSTANCE_NAME
def bar(prg):
br="[1;33mUploading to HuggingFace : " '[0m|'+'█' * prg + ' ' * (25-prg)+'| ' +str(prg*4)+ "%"
return br
print("[1;32mLoading...")
NM="False"
if os.path.getsize(OUTPUT_DIR+"/text_encoder/pytorch_model.bin") > 670901463:
NM="True"
with capture.capture_output() as cap:
if NM=="False":
# %cd $OUTPUT_DIR
!rm -r safety_checker feature_extractor .git
!rm model_index.json
!git init
!git lfs install --system --skip-repo
!git remote add -f origin "https://USER:{hf_token}@huggingface.co/runwayml/stable-diffusion-v1-5"
!git config core.sparsecheckout true
!echo -e "feature_extractor\nsafety_checker\nmodel_index.json" > .git/info/sparse-checkout
!git pull origin main
!rm -r .git
# %cd /content
else:
# %cd $OUTPUT_DIR
!rm -r feature_extractor .git
!git init
!git lfs install --system --skip-repo
!git remote add -f origin "https://USER:{hf_token}@huggingface.co/stabilityai/stable-diffusion-2-1"
!git config core.sparsecheckout true
!echo -e "feature_extractor" > .git/info/sparse-checkout
!git pull origin main
!rm -r .git
# %cd /content
image_string = ""
if os.path.exists('/content/sample_images'):
!rm -r /content/sample_images
Samples="/content/sample_images"
!mkdir $Samples
clear_output()
if Upload_sample_images:
print("[1;32mUpload Sample images of the model")
uploaded = files.upload()
for filename in uploaded.keys():
shutil.move(filename, Samples)
# %cd $Samples
!find . -name "* *" -type f | rename 's/ /_/g'
# %cd /content
clear_output()
print(bar(1))
images_upload = os.listdir(Samples)
instance_prompt_list = []
for i, image in enumerate(images_upload):
image_string = f'''
{image_string}
'''
readme_text = f'''---
license: creativeml-openrail-m
tags:
- NxtWave-GenAI-Webinar
- text-to-image
- stable-diffusion
---
### {Name_of_your_concept} Dreambooth model trained by {api.whoami(token=hf_token)["name"]} following the "Build your own Gen AI model" session by NxtWave.
Project Submission Code: {Your_Unique_Project_Submission_Code}
Sample pictures of this concept:
{image_string}
'''
#Save the readme to a file
readme_file = open("README.md", "w")
readme_file.write(readme_text)
readme_file.close()
operations = [
CommitOperationAdd(path_in_repo="README.md", path_or_fileobj="README.md"),
CommitOperationAdd(path_in_repo=f"{Session_Name}.ckpt",path_or_fileobj=MDLPTH)
]
create_repo(repo_id,private=False, token=hf_token)
api.create_commit(
repo_id=repo_id,
operations=operations,