File size: 48,048 Bytes
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
11ae595
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
2c2f570
 
 
 
 
 
 
 
 
b4dfea0
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
2c2f570
b4dfea0
2c2f570
b4dfea0
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
2c2f570
 
b4dfea0
 
 
 
 
 
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
 
 
 
2c2f570
 
 
 
 
 
b4dfea0
 
 
 
2c2f570
 
b4dfea0
 
 
 
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
 
2c2f570
b4dfea0
 
 
 
 
2c2f570
 
b4dfea0
 
 
 
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
 
2c2f570
b4dfea0
 
 
 
2c2f570
 
b4dfea0
 
 
2c2f570
 
 
 
 
 
 
 
 
b4dfea0
 
 
 
2c2f570
 
b4dfea0
2c2f570
b4dfea0
2c2f570
 
 
 
 
 
 
 
 
 
b4dfea0
 
 
 
2c2f570
 
b4dfea0
 
 
2c2f570
 
b4dfea0
 
 
 
2c2f570
 
b4dfea0
 
 
2c2f570
 
 
b4dfea0
 
 
 
2c2f570
 
b4dfea0
 
 
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
2c2f570
 
 
 
 
 
b4dfea0
2c2f570
 
b4dfea0
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
2c2f570
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
2c2f570
 
b4dfea0
2c2f570
b4dfea0
 
2c2f570
b4dfea0
 
 
 
2c2f570
 
 
 
 
 
 
 
b4dfea0
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
2c2f570
b4dfea0
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
2c2f570
b4dfea0
2c2f570
b4dfea0
 
 
2c2f570
 
 
 
 
 
 
 
 
 
 
 
 
 
b4dfea0
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
2c2f570
 
b4dfea0
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
2c2f570
 
 
b4dfea0
 
2c2f570
b4dfea0
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
2c2f570
b4dfea0
 
 
 
2c2f570
b4dfea0
 
 
 
 
 
 
 
 
 
 
 
11ae595
 
 
b4dfea0
11ae595
 
 
b4dfea0
 
 
 
 
 
 
 
 
 
 
11ae595
b4dfea0
 
 
 
11ae595
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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "d0b72877",
   "metadata": {},
   "source": [
    "# vqgan-jax-encoding-yfcc100m"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ba7b31e6",
   "metadata": {},
   "source": [
    "Same as `vqgan-jax-encoding-with-captions`, but for YFCC100M.\n",
    "\n",
    "This dataset was prepared by @borisdayma in Json lines format."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "3b59489e",
   "metadata": {},
   "outputs": [],
   "source": [
    "import io\n",
    "\n",
    "import requests\n",
    "from PIL import Image\n",
    "import numpy as np\n",
    "from tqdm import tqdm\n",
    "\n",
    "import torch\n",
    "import torchvision.transforms as T\n",
    "import torchvision.transforms.functional as TF\n",
    "from torchvision.transforms import InterpolationMode\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "from torchvision.datasets.folder import default_loader\n",
    "import os\n",
    "\n",
    "import jax\n",
    "from jax import pmap"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "511c3b9e",
   "metadata": {},
   "source": [
    "## VQGAN-JAX model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "2ca50dc7",
   "metadata": {},
   "outputs": [],
   "source": [
    "from vqgan_jax.modeling_flax_vqgan import VQModel"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7b60da9a",
   "metadata": {},
   "source": [
    "We'll use a VQGAN trained by using Taming Transformers and converted to a JAX model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "id": "29ce8b15",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Working with z of shape (1, 256, 16, 16) = 65536 dimensions.\n"
     ]
    }
   ],
   "source": [
    "model = VQModel.from_pretrained(\"flax-community/vqgan_f16_16384\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c7c4c1e6",
   "metadata": {},
   "source": [
    "## Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "33861477",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from pathlib import Path"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "81b19eca",
   "metadata": {},
   "outputs": [],
   "source": [
    "yfcc100m = Path('/home/khali/TPU-Test/YFCC100M_OpenAI_subset')\n",
    "# Images are 'sharded' from the following directory\n",
    "yfcc100m_images = yfcc100m/'data'/'data'/'images'\n",
    "yfcc100m_metadata = yfcc100m/'metadata_YFCC100M.jsonl'\n",
    "yfcc100m_output = yfcc100m/'metadata_encoded.tsv'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c58bb4a",
   "metadata": {},
   "source": [
    "### Cleanup"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1a14ae3d",
   "metadata": {},
   "source": [
    "We need to select entries with images that exist. Otherwise we can't build batches because `Dataloader` does not support `None` in batches. We use Huggingface Datasets, I understand they support threaded reading of jsonl files, and I was running out of memory when using pandas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "7811648c",
   "metadata": {},
   "outputs": [],
   "source": [
    "import datasets\n",
    "from datasets import Dataset, load_dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4811a230",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "tcmalloc: large alloc 1254047744 bytes == 0xb2b08000 @  0x7f9e78632680 0x7f9e78653824 0x585b92 0x504d56 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56 0x56acb6 0x5f5956 0x56aadf 0x5f5956 0x56acb6 0x568d9a 0x5f5b33 0x50b7f8 0x5f2702 0x56c332\n",
      "tcmalloc: large alloc 1254047744 bytes == 0xfd74e000 @  0x7f9e78632680 0x7f9e78653824 0x590214 0x586f90 0x56e1f3 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56 0x56acb6 0x5f5956 0x56aadf 0x5f5956 0x56acb6 0x568d9a 0x5f5b33 0x50b7f8 0x5f2702 0x56c332\n",
      "tcmalloc: large alloc 5016190976 bytes == 0x148b42000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5019099136 bytes == 0x273f12000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5019811840 bytes == 0x39f9a8000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5024571392 bytes == 0x4cb4ec000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5021097984 bytes == 0x4cb4ec000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5022818304 bytes == 0x4cb4ec000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5020794880 bytes == 0x4cb4ec000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5019451392 bytes == 0x39f9a8000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5020565504 bytes == 0x4cb4ec000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5012561920 bytes == 0x273f12000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5021835264 bytes == 0x5f6cba000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n",
      "tcmalloc: large alloc 5017436160 bytes == 0x273f12000 @  0x7f9e78632680 0x7f9e78653824 0x5b9144 0x7f9b2929127e 0x7f9b29291a19 0x7f9b29291886 0x7f9b29291cef 0x7f9b2928f204 0x5f2cc9 0x5f30ff 0x5705f6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x56acb6 0x5f5956 0x5a8cb3 0x56ae94 0x568d9a 0x68cdc7 0x5ff5d4 0x5c3cb0 0x56aadf 0x501148 0x56c422 0x501148 0x56c422 0x501148 0x504d56\n"
     ]
    }
   ],
   "source": [
    "# The metadata is too bog to load into memory at once, so chopping it into chunks\n",
    "chunk_size=1000000\n",
    "batch_no=1\n",
    "for chunk in pd.read_json(yfcc100m_metadata, orient=\"records\", lines=True,chunksize=chunk_size):\n",
    "    chunk.to_csv('./chunks/chunk'+str(batch_no)+'.tsv', sep=\"\\t\", index=False)\n",
    "    batch_no+=1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "46b2f083",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>photoid</th>\n",
       "      <th>uid</th>\n",
       "      <th>unickname</th>\n",
       "      <th>datetaken</th>\n",
       "      <th>dateuploaded</th>\n",
       "      <th>capturedevice</th>\n",
       "      <th>title</th>\n",
       "      <th>description</th>\n",
       "      <th>usertags</th>\n",
       "      <th>machinetags</th>\n",
       "      <th>...</th>\n",
       "      <th>licenseurl</th>\n",
       "      <th>serverid</th>\n",
       "      <th>farmid</th>\n",
       "      <th>secret</th>\n",
       "      <th>secretoriginal</th>\n",
       "      <th>ext</th>\n",
       "      <th>marker</th>\n",
       "      <th>key</th>\n",
       "      <th>title_clean</th>\n",
       "      <th>description_clean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>137943</td>\n",
       "      <td>48600072071@N01</td>\n",
       "      <td>doctor+paradox</td>\n",
       "      <td>2004-08-01 18:13:06.0</td>\n",
       "      <td>1091409186</td>\n",
       "      <td>NaN</td>\n",
       "      <td>A+Picture+Share%21</td>\n",
       "      <td>Antenna</td>\n",
       "      <td>cameraphone,cayugaheights,green,hydrant,ithaca...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-sa/2.0/</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1650c7cdc6</td>\n",
       "      <td>1650c7cdc6</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>d29e7c6a3028418c64eb15e3cf577c2</td>\n",
       "      <td>A Picture Share!</td>\n",
       "      <td>Antenna</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1246361</td>\n",
       "      <td>44124324682@N01</td>\n",
       "      <td>mharrsch</td>\n",
       "      <td>2004-11-03 23:04:02.0</td>\n",
       "      <td>1099523042</td>\n",
       "      <td>NaN</td>\n",
       "      <td>An+ornate+Roman+urn</td>\n",
       "      <td>Photographed+at+the+%3Ca+href%3D%22http%3A%2F%...</td>\n",
       "      <td>ancient,baltimore,burial,death,empire,funeral,...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-sa/2.0/</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>cf37054610</td>\n",
       "      <td>cf37054610</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>d29f01b149167d683f9ddde464bb3db</td>\n",
       "      <td>An ornate Roman urn</td>\n",
       "      <td>Photographed at the Walters Art Museum, Baltim...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1251599</td>\n",
       "      <td>51035803024@N01</td>\n",
       "      <td>bmitd67</td>\n",
       "      <td>2004-10-30 17:09:32.0</td>\n",
       "      <td>1099538888</td>\n",
       "      <td>Canon+PowerShot+S30</td>\n",
       "      <td>Jai+%26+Tara+on+the+Cumberland</td>\n",
       "      <td>Another+trip+for+the+happy+couple.</td>\n",
       "      <td>blue+heron,cumberland+river,jai,tara,tennessee</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-sa/2.0/</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4a4234e32c</td>\n",
       "      <td>4a4234e32c</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>d296e9e34bdae41edb6c679ff824ab2a</td>\n",
       "      <td>Jai &amp; Tara on the Cumberland</td>\n",
       "      <td>Another trip for the happy couple.</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2348587</td>\n",
       "      <td>73621375@N00</td>\n",
       "      <td>Thom+Watson</td>\n",
       "      <td>2004-12-18 21:08:09.0</td>\n",
       "      <td>1103497228</td>\n",
       "      <td>SONY+DSC-W1</td>\n",
       "      <td>Castle+gate+-+%22lite-brited%22</td>\n",
       "      <td>Taken+at+the+Miracle+of+Lights+display+in+Cent...</td>\n",
       "      <td>bullrunpark,castle,centreville,christmas,decor...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-sa/2.0/</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>7162c974c3</td>\n",
       "      <td>7162c974c3</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>d29ce96395848478b1e8396e44899</td>\n",
       "      <td>Castle gate - \"lite-brited\"</td>\n",
       "      <td>Taken at the Miracle of Lights display in Cent...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3516047</td>\n",
       "      <td>48600072071@N01</td>\n",
       "      <td>doctor+paradox</td>\n",
       "      <td>2005-01-18 16:44:18.0</td>\n",
       "      <td>1106084658</td>\n",
       "      <td>NaN</td>\n",
       "      <td>A+Picture+Share%21</td>\n",
       "      <td>Tabular</td>\n",
       "      <td>cameraphone,moblog,unfound</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-sa/2.0/</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>663e0d8b3d</td>\n",
       "      <td>663e0d8b3d</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>d29abf32c4e12ff881f975b70e0cec0</td>\n",
       "      <td>A Picture Share!</td>\n",
       "      <td>Tabular</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>999995</th>\n",
       "      <td>4648651054</td>\n",
       "      <td>24511045@N04</td>\n",
       "      <td>mtfrazier</td>\n",
       "      <td>2010-05-02 15:47:45.0</td>\n",
       "      <td>1275083371</td>\n",
       "      <td>Canon+EOS+50D</td>\n",
       "      <td>U.S.+Navy+Blue+Angels%3A+2010</td>\n",
       "      <td>2+May+2010%0ASunday%0ASt.+Joseph%2C+Missouri</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-nd/2.0/</td>\n",
       "      <td>4072</td>\n",
       "      <td>5</td>\n",
       "      <td>2d12d73fb0</td>\n",
       "      <td>dd5856ea42</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>60fa2911cb81eb25b356e9fee978aef</td>\n",
       "      <td>U.S. Navy Blue Angels: 2010</td>\n",
       "      <td>2 May 2010 Sunday St. Joseph, Missouri</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>999996</th>\n",
       "      <td>4652130996</td>\n",
       "      <td>21963865@N04</td>\n",
       "      <td>GRAB1.0</td>\n",
       "      <td>2010-05-29 19:23:10.0</td>\n",
       "      <td>1275200833</td>\n",
       "      <td>SONY+DSLR-A230</td>\n",
       "      <td>Attempts+on+Her+Life</td>\n",
       "      <td>BAPA+1+production+of+Martin+Crimp%27s+Attempts...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-nd/2.0/</td>\n",
       "      <td>4003</td>\n",
       "      <td>5</td>\n",
       "      <td>8889121579</td>\n",
       "      <td>2f46599456</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>60f5ef5ce4c2d24566226abebd67d4</td>\n",
       "      <td>Attempts on Her Life</td>\n",
       "      <td>BAPA 1 production of Martin Crimp's Attempts o...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>999997</th>\n",
       "      <td>4652568339</td>\n",
       "      <td>64025277@N00</td>\n",
       "      <td>1Sock</td>\n",
       "      <td>2010-05-13 15:38:37.0</td>\n",
       "      <td>1275234267</td>\n",
       "      <td>Canon+EOS+DIGITAL+REBEL+XT</td>\n",
       "      <td>Carlsbad+Caverns+3</td>\n",
       "      <td>%E2%99%A5%E2%99%A5%E2%99%A5%E2%99%A5%E2%99%A5%...</td>\n",
       "      <td>carlsbad,carlsbad+caverns,cave,faa,new+mexico,...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-nd/2.0/</td>\n",
       "      <td>4010</td>\n",
       "      <td>5</td>\n",
       "      <td>0a1808a69e</td>\n",
       "      <td>cf6d348e3d</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>60f029482d1d1028fda5281daf498f</td>\n",
       "      <td>Carlsbad Caverns 3</td>\n",
       "      <td>♥♥♥♥♥♥♥ Interested in purchasing this photogra...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>999998</th>\n",
       "      <td>4653110895</td>\n",
       "      <td>20483509@N00</td>\n",
       "      <td>subberculture</td>\n",
       "      <td>2010-05-30 15:37:05.0</td>\n",
       "      <td>1275245596</td>\n",
       "      <td>Canon+DIGITAL+IXUS+40</td>\n",
       "      <td>Want</td>\n",
       "      <td>Isn%27t+that+gorgeous%3F</td>\n",
       "      <td>2010,edinburgh+museum,may,phonebox,wood</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-sa/2.0/</td>\n",
       "      <td>4066</td>\n",
       "      <td>5</td>\n",
       "      <td>77c3b3a254</td>\n",
       "      <td>c4697e1511</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>60f72775f433cf8de3efaeb431866153</td>\n",
       "      <td>Want</td>\n",
       "      <td>Isn't that gorgeous?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>999999</th>\n",
       "      <td>4655503987</td>\n",
       "      <td>8457193@N07</td>\n",
       "      <td>zackojones</td>\n",
       "      <td>2010-05-30 15:34:58.0</td>\n",
       "      <td>1275310230</td>\n",
       "      <td>Canon+EOS+7D</td>\n",
       "      <td>Summertime</td>\n",
       "      <td>You+gotta+love+it%21</td>\n",
       "      <td>georgia,savannah,united+states,us</td>\n",
       "      <td>NaN</td>\n",
       "      <td>...</td>\n",
       "      <td>http://creativecommons.org/licenses/by-nc-sa/2.0/</td>\n",
       "      <td>4043</td>\n",
       "      <td>5</td>\n",
       "      <td>caff543bfe</td>\n",
       "      <td>f60952ac4d</td>\n",
       "      <td>jpg</td>\n",
       "      <td>0</td>\n",
       "      <td>60f687e11b913bce461e9525d8047e0</td>\n",
       "      <td>Summertime</td>\n",
       "      <td>You gotta love it!</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1000000 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           photoid              uid       unickname              datetaken  \\\n",
       "0           137943  48600072071@N01  doctor+paradox  2004-08-01 18:13:06.0   \n",
       "1          1246361  44124324682@N01        mharrsch  2004-11-03 23:04:02.0   \n",
       "2          1251599  51035803024@N01         bmitd67  2004-10-30 17:09:32.0   \n",
       "3          2348587     73621375@N00     Thom+Watson  2004-12-18 21:08:09.0   \n",
       "4          3516047  48600072071@N01  doctor+paradox  2005-01-18 16:44:18.0   \n",
       "...            ...              ...             ...                    ...   \n",
       "999995  4648651054     24511045@N04       mtfrazier  2010-05-02 15:47:45.0   \n",
       "999996  4652130996     21963865@N04         GRAB1.0  2010-05-29 19:23:10.0   \n",
       "999997  4652568339     64025277@N00           1Sock  2010-05-13 15:38:37.0   \n",
       "999998  4653110895     20483509@N00   subberculture  2010-05-30 15:37:05.0   \n",
       "999999  4655503987      8457193@N07      zackojones  2010-05-30 15:34:58.0   \n",
       "\n",
       "        dateuploaded               capturedevice  \\\n",
       "0         1091409186                         NaN   \n",
       "1         1099523042                         NaN   \n",
       "2         1099538888         Canon+PowerShot+S30   \n",
       "3         1103497228                 SONY+DSC-W1   \n",
       "4         1106084658                         NaN   \n",
       "...              ...                         ...   \n",
       "999995    1275083371               Canon+EOS+50D   \n",
       "999996    1275200833              SONY+DSLR-A230   \n",
       "999997    1275234267  Canon+EOS+DIGITAL+REBEL+XT   \n",
       "999998    1275245596       Canon+DIGITAL+IXUS+40   \n",
       "999999    1275310230                Canon+EOS+7D   \n",
       "\n",
       "                                  title  \\\n",
       "0                    A+Picture+Share%21   \n",
       "1                   An+ornate+Roman+urn   \n",
       "2        Jai+%26+Tara+on+the+Cumberland   \n",
       "3       Castle+gate+-+%22lite-brited%22   \n",
       "4                    A+Picture+Share%21   \n",
       "...                                 ...   \n",
       "999995    U.S.+Navy+Blue+Angels%3A+2010   \n",
       "999996             Attempts+on+Her+Life   \n",
       "999997               Carlsbad+Caverns+3   \n",
       "999998                             Want   \n",
       "999999                       Summertime   \n",
       "\n",
       "                                              description  \\\n",
       "0                                                 Antenna   \n",
       "1       Photographed+at+the+%3Ca+href%3D%22http%3A%2F%...   \n",
       "2                      Another+trip+for+the+happy+couple.   \n",
       "3       Taken+at+the+Miracle+of+Lights+display+in+Cent...   \n",
       "4                                                 Tabular   \n",
       "...                                                   ...   \n",
       "999995       2+May+2010%0ASunday%0ASt.+Joseph%2C+Missouri   \n",
       "999996  BAPA+1+production+of+Martin+Crimp%27s+Attempts...   \n",
       "999997  %E2%99%A5%E2%99%A5%E2%99%A5%E2%99%A5%E2%99%A5%...   \n",
       "999998                           Isn%27t+that+gorgeous%3F   \n",
       "999999                               You+gotta+love+it%21   \n",
       "\n",
       "                                                 usertags machinetags  ...  \\\n",
       "0       cameraphone,cayugaheights,green,hydrant,ithaca...         NaN  ...   \n",
       "1       ancient,baltimore,burial,death,empire,funeral,...         NaN  ...   \n",
       "2          blue+heron,cumberland+river,jai,tara,tennessee         NaN  ...   \n",
       "3       bullrunpark,castle,centreville,christmas,decor...         NaN  ...   \n",
       "4                              cameraphone,moblog,unfound         NaN  ...   \n",
       "...                                                   ...         ...  ...   \n",
       "999995                                                NaN         NaN  ...   \n",
       "999996                                                NaN         NaN  ...   \n",
       "999997  carlsbad,carlsbad+caverns,cave,faa,new+mexico,...         NaN  ...   \n",
       "999998            2010,edinburgh+museum,may,phonebox,wood         NaN  ...   \n",
       "999999                  georgia,savannah,united+states,us         NaN  ...   \n",
       "\n",
       "                                               licenseurl  serverid  farmid  \\\n",
       "0       http://creativecommons.org/licenses/by-nc-sa/2.0/         1       1   \n",
       "1       http://creativecommons.org/licenses/by-nc-sa/2.0/         1       1   \n",
       "2       http://creativecommons.org/licenses/by-nc-sa/2.0/         1       1   \n",
       "3       http://creativecommons.org/licenses/by-nc-sa/2.0/         2       1   \n",
       "4       http://creativecommons.org/licenses/by-nc-sa/2.0/         3       1   \n",
       "...                                                   ...       ...     ...   \n",
       "999995  http://creativecommons.org/licenses/by-nc-nd/2.0/      4072       5   \n",
       "999996  http://creativecommons.org/licenses/by-nc-nd/2.0/      4003       5   \n",
       "999997  http://creativecommons.org/licenses/by-nc-nd/2.0/      4010       5   \n",
       "999998     http://creativecommons.org/licenses/by-sa/2.0/      4066       5   \n",
       "999999  http://creativecommons.org/licenses/by-nc-sa/2.0/      4043       5   \n",
       "\n",
       "            secret secretoriginal  ext marker  \\\n",
       "0       1650c7cdc6     1650c7cdc6  jpg      0   \n",
       "1       cf37054610     cf37054610  jpg      0   \n",
       "2       4a4234e32c     4a4234e32c  jpg      0   \n",
       "3       7162c974c3     7162c974c3  jpg      0   \n",
       "4       663e0d8b3d     663e0d8b3d  jpg      0   \n",
       "...            ...            ...  ...    ...   \n",
       "999995  2d12d73fb0     dd5856ea42  jpg      0   \n",
       "999996  8889121579     2f46599456  jpg      0   \n",
       "999997  0a1808a69e     cf6d348e3d  jpg      0   \n",
       "999998  77c3b3a254     c4697e1511  jpg      0   \n",
       "999999  caff543bfe     f60952ac4d  jpg      0   \n",
       "\n",
       "                                     key                   title_clean  \\\n",
       "0        d29e7c6a3028418c64eb15e3cf577c2              A Picture Share!   \n",
       "1        d29f01b149167d683f9ddde464bb3db           An ornate Roman urn   \n",
       "2       d296e9e34bdae41edb6c679ff824ab2a  Jai & Tara on the Cumberland   \n",
       "3          d29ce96395848478b1e8396e44899   Castle gate - \"lite-brited\"   \n",
       "4        d29abf32c4e12ff881f975b70e0cec0              A Picture Share!   \n",
       "...                                  ...                           ...   \n",
       "999995   60fa2911cb81eb25b356e9fee978aef   U.S. Navy Blue Angels: 2010   \n",
       "999996    60f5ef5ce4c2d24566226abebd67d4          Attempts on Her Life   \n",
       "999997    60f029482d1d1028fda5281daf498f            Carlsbad Caverns 3   \n",
       "999998  60f72775f433cf8de3efaeb431866153                          Want   \n",
       "999999   60f687e11b913bce461e9525d8047e0                    Summertime   \n",
       "\n",
       "                                        description_clean  \n",
       "0                                                 Antenna  \n",
       "1       Photographed at the Walters Art Museum, Baltim...  \n",
       "2                      Another trip for the happy couple.  \n",
       "3       Taken at the Miracle of Lights display in Cent...  \n",
       "4                                                 Tabular  \n",
       "...                                                   ...  \n",
       "999995             2 May 2010 Sunday St. Joseph, Missouri  \n",
       "999996  BAPA 1 production of Martin Crimp's Attempts o...  \n",
       "999997  ♥♥♥♥♥♥♥ Interested in purchasing this photogra...  \n",
       "999998                               Isn't that gorgeous?  \n",
       "999999                                 You gotta love it!  \n",
       "\n",
       "[1000000 rows x 26 columns]"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# looking up at a chunk\n",
    "pd.read_csv(\"./chunks/chunk1.tsv\", sep=\"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "c51c5597",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>key</th>\n",
       "      <th>title_clean</th>\n",
       "      <th>description_clean</th>\n",
       "      <th>ext</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>d29e7c6a3028418c64eb15e3cf577c2</td>\n",
       "      <td>A Picture Share!</td>\n",
       "      <td>Antenna</td>\n",
       "      <td>jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>d29f01b149167d683f9ddde464bb3db</td>\n",
       "      <td>An ornate Roman urn</td>\n",
       "      <td>Photographed at the Walters Art Museum, Baltim...</td>\n",
       "      <td>jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>d296e9e34bdae41edb6c679ff824ab2a</td>\n",
       "      <td>Jai &amp; Tara on the Cumberland</td>\n",
       "      <td>Another trip for the happy couple.</td>\n",
       "      <td>jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>d29ce96395848478b1e8396e44899</td>\n",
       "      <td>Castle gate - \"lite-brited\"</td>\n",
       "      <td>Taken at the Miracle of Lights display in Cent...</td>\n",
       "      <td>jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>d29abf32c4e12ff881f975b70e0cec0</td>\n",
       "      <td>A Picture Share!</td>\n",
       "      <td>Tabular</td>\n",
       "      <td>jpg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                key                   title_clean  \\\n",
       "0   d29e7c6a3028418c64eb15e3cf577c2              A Picture Share!   \n",
       "1   d29f01b149167d683f9ddde464bb3db           An ornate Roman urn   \n",
       "2  d296e9e34bdae41edb6c679ff824ab2a  Jai & Tara on the Cumberland   \n",
       "3     d29ce96395848478b1e8396e44899   Castle gate - \"lite-brited\"   \n",
       "4   d29abf32c4e12ff881f975b70e0cec0              A Picture Share!   \n",
       "\n",
       "                                   description_clean  ext  \n",
       "0                                            Antenna  jpg  \n",
       "1  Photographed at the Walters Art Museum, Baltim...  jpg  \n",
       "2                 Another trip for the happy couple.  jpg  \n",
       "3  Taken at the Miracle of Lights display in Cent...  jpg  \n",
       "4                                            Tabular  jpg  "
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Looking at a chunk with only the relevant columns that we need\n",
    "df = pd.read_csv(\"./chunks/chunk1.tsv\", sep=\"\\t\")[[\"key\", \"title_clean\", \"description_clean\", \"ext\"]]\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc1668f8",
   "metadata": {},
   "source": [
    "### Grabbing each chunks from the folder, cleaning it up, only taking the entries which image exist and appending it to the global df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "abbcccf3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# the function that helps us to decide whether an image with certain id exists in storage, we only take the ones that we have the images for\n",
    "def image_exists(item):\n",
    "    name, _, _, ext, _ = item\n",
    "    root=str(yfcc100m_images)\n",
    "    image_path = (Path(root)/name[0:3]/name[3:6]/name).with_suffix(\".\"+ext)\n",
    "    if image_path.exists():\n",
    "        return True\n",
    "    else:\n",
    "        return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "44fa86ab",
   "metadata": {},
   "outputs": [],
   "source": [
    "# This cell does it all, grabs each chunk, cleans it up based on image existing condition, etc.\n",
    "global_df = pd.DataFrame()\n",
    "chunks_dir = \"./chunks\"\n",
    "for filename in os.listdir(chunks_dir):\n",
    "        df = pd.read_csv(f\"./chunks/{str(filename)}\", sep=\"\\t\")[[\"key\", \"title_clean\", \"description_clean\", \"ext\"]]\n",
    "        df['caption'] = df[\"title_clean\"]+\". \"+df['description_clean']\n",
    "        df['is_exist'] = df.apply(image_exists, axis=1)\n",
    "        df = df.dropna()[[\"key\", \"caption\"]]\n",
    "        df.columns = ['image_file', 'caption']\n",
    "        global_df = global_df.append(df, ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "45024fdc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# saving the tsv to disk\n",
    "global_df.to_csv('./chunks/YFCC_subset_clean.tsv', sep=\"\\t\", index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "id": "dca4eb73",
   "metadata": {},
   "outputs": [],
   "source": [
    "# loading the tsv from disk (for explicitness, also my electricity was gone, glad it happened after I saved to the disk :( )\n",
    "\n",
    "dataset = pd.read_csv(f\"./chunks/YFCC_subset_clean.tsv\", sep=\"\\t\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "id": "a511264a",
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Luke Melas-Kyriazi's dataset.py's modified version for YFCC\n",
    "\"\"\"\n",
    "import warnings\n",
    "from typing import Optional, Callable\n",
    "from pathlib import Path\n",
    "import numpy as np\n",
    "import torch\n",
    "import pandas as pd\n",
    "from torch.utils.data import Dataset\n",
    "from torchvision.datasets.folder import default_loader\n",
    "from PIL import ImageFile\n",
    "from PIL.Image import DecompressionBombWarning\n",
    "ImageFile.LOAD_TRUNCATED_IMAGES = True\n",
    "warnings.filterwarnings(\"ignore\", category=UserWarning)\n",
    "warnings.filterwarnings(\"ignore\", category=DecompressionBombWarning)\n",
    "\n",
    "\n",
    "class CaptionDataset(Dataset):\n",
    "    \"\"\"\n",
    "    A PyTorch Dataset class for (image, texts) tasks. Note that this dataset \n",
    "    returns the raw text rather than tokens. This is done on purpose, because\n",
    "    it's easy to tokenize a batch of text after loading it from this dataset.\n",
    "    \"\"\"\n",
    "\n",
    "    def __init__(self, *, images_root: str, captions_path: str, text_transform: Optional[Callable] = None, \n",
    "                 image_transform: Optional[Callable] = None, image_transform_type: str = 'torchvision',\n",
    "                 include_captions: bool = True):\n",
    "        \"\"\"\n",
    "        :param images_root: folder where images are stored\n",
    "        :param captions_path: path to csv that maps image filenames to captions\n",
    "        :param image_transform: image transform pipeline\n",
    "        :param text_transform: image transform pipeline\n",
    "        :param image_transform_type: image transform type, either `torchvision` or `albumentations`\n",
    "        :param include_captions: Returns a dictionary with `image`, `text` if `true`; otherwise returns just the images.\n",
    "        \"\"\"\n",
    "\n",
    "        # Base path for images\n",
    "        self.images_root = Path(images_root)\n",
    "\n",
    "        # Load captions as DataFrame\n",
    "        self.captions = pd.read_csv(f\"./chunks/YFCC_subset_clean.tsv\", sep=\"\\t\")\n",
    "        self.captions['image_file'] = self.captions['image_file'].astype(str)\n",
    "\n",
    "        # PyTorch transformation pipeline for the image (normalizing, etc.)\n",
    "        self.text_transform = text_transform\n",
    "        self.image_transform = image_transform\n",
    "        self.image_transform_type = image_transform_type.lower()\n",
    "        assert self.image_transform_type in ['torchvision', 'albumentations']\n",
    "\n",
    "        # Total number of datapoints\n",
    "        self.size = len(self.captions)\n",
    "\n",
    "        # Return image+captions or just images\n",
    "        self.include_captions = include_captions\n",
    "    \n",
    "    def image_exists(item):\n",
    "        name, caption = item\n",
    "        root=str(self.images_root)\n",
    "        image_path = (Path(root)/name[0:3]/name[3:6]/name).with_suffix(\".jpg\")\n",
    "\n",
    "        return image_path.exists()\n",
    "\n",
    "    def verify_that_all_images_exist(self):\n",
    "        for image_file in self.captions['image_file']:\n",
    "            if not image_exists:\n",
    "                print(f'file does not exist: {p}')\n",
    "\n",
    "    def _get_raw_image(self, i):\n",
    "        name = self.captions.iloc[i]['image_file']\n",
    "        image_path = (Path(self.images_root)/name[0:3]/name[3:6]/name).with_suffix(\".jpg\")\n",
    "        image = default_loader(image_path)\n",
    "        return image\n",
    "\n",
    "    def _get_raw_text(self, i):\n",
    "        return self.captions.iloc[i]['caption']\n",
    "\n",
    "    def __getitem__(self, i):\n",
    "        image = self._get_raw_image(i)\n",
    "        caption = self._get_raw_text(i)\n",
    "        if self.image_transform is not None:\n",
    "            if self.image_transform_type == 'torchvision':\n",
    "                image = self.image_transform(image)\n",
    "            elif self.image_transform_type == 'albumentations':\n",
    "                image = self.image_transform(image=np.array(image))['image']\n",
    "            else:\n",
    "                raise NotImplementedError(f\"{self.image_transform_type=}\")\n",
    "        return {'image': image, 'text': caption} if self.include_captions else image\n",
    "\n",
    "    def __len__(self):\n",
    "        return self.size\n",
    "\n",
    "\n",
    "if __name__ == \"__main__\":\n",
    "    import albumentations as A\n",
    "    from albumentations.pytorch import ToTensorV2\n",
    "    from transformers import AutoTokenizer\n",
    "    \n",
    "\n",
    "    images_root = \"/home/khali/TPU-Test/YFCC100M_OpenAI_subset/data/data/images\"\n",
    "    captions_path = './YFCC_subset_clean.tsv'\n",
    "    image_size = 256\n",
    "    \n",
    "    # Create transforms\n",
    "    def image_transform(image):\n",
    "        s = min(image.size)\n",
    "        r = image_size / s\n",
    "        s = (round(r * image.size[1]), round(r * image.size[0]))\n",
    "        image = TF.resize(image, s, interpolation=InterpolationMode.LANCZOS)\n",
    "        image = TF.center_crop(image, output_size = 2 * [image_size])\n",
    "        image = torch.unsqueeze(T.ToTensor()(image), 0)\n",
    "        image = image.permute(0, 2, 3, 1).numpy()\n",
    "        return image\n",
    "    \n",
    "    # Create dataset\n",
    "    dataset = CaptionDataset(\n",
    "        images_root=images_root,\n",
    "        captions_path=captions_path,\n",
    "        image_transform=image_transform,\n",
    "        image_transform_type='torchvision',\n",
    "        include_captions=False\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "id": "cc922704",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2483316"
      ]
     },
     "execution_count": 155,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(dataset)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "id": "6e47ba46",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataloader = DataLoader(dataset, batch_size=32, num_workers=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c8a130eb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# looking at a batch\n",
    "next(iter(dataloader))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c192fd44",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import matplotlib.pyplot as plt\n",
    "# for tensor_image, _ in dataloader:\n",
    "#     print(tensor_image)\n",
    "#     plt.imshow(tensor_image.permute(1, 2, 0))\n",
    "#     break"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62ad01c3",
   "metadata": {},
   "source": [
    "## Encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "88f36d0b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def encode(model, batch):\n",
    "#     print(\"jitting encode function\")\n",
    "    _, indices = model.encode(batch)\n",
    "    return indices"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "1f35f0cb",
   "metadata": {},
   "outputs": [],
   "source": [
    "def superbatch_generator(dataloader, num_tpus):\n",
    "    iter_loader = iter(dataloader)\n",
    "    for batch in iter_loader:\n",
    "        superbatch = [batch.squeeze(1)]\n",
    "        try:\n",
    "            for b in range(num_tpus-1):\n",
    "                batch = next(iter_loader)\n",
    "                if batch is None:\n",
    "                    break\n",
    "                # Skip incomplete last batch\n",
    "                if batch.shape[0] == dataloader.batch_size:\n",
    "                    superbatch.append(batch.squeeze(1))\n",
    "        except StopIteration:\n",
    "            pass\n",
    "        superbatch = torch.stack(superbatch, axis=0)\n",
    "        yield superbatch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "id": "2210705b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "def encode_captioned_dataset(dataset, output_tsv, batch_size=32, num_workers=16):\n",
    "    if os.path.isfile(output_tsv):\n",
    "        print(f\"Destination file {output_tsv} already exists, please move away.\")\n",
    "        return\n",
    "    \n",
    "    num_tpus = 8    \n",
    "    dataloader = DataLoader(dataset, batch_size=batch_size, num_workers=num_workers)\n",
    "    superbatches = superbatch_generator(dataloader, num_tpus=num_tpus)\n",
    "    \n",
    "    p_encoder = pmap(lambda batch: encode(model, batch))\n",
    "\n",
    "    # We save each superbatch to avoid reallocation of buffers as we process them.\n",
    "    # We keep the file open to prevent excessive file seeks.\n",
    "    with open(output_tsv, \"w\") as file:\n",
    "        iterations = len(dataset) // (batch_size * num_tpus)\n",
    "        for n in tqdm(range(iterations)):\n",
    "            superbatch = next(superbatches)\n",
    "            encoded = p_encoder(superbatch.numpy())\n",
    "            encoded = encoded.reshape(-1, encoded.shape[-1])\n",
    "\n",
    "            # Extract fields from the dataset internal `captions` property, and save to disk\n",
    "            start_index = n * batch_size * num_tpus\n",
    "            end_index = (n+1) * batch_size * num_tpus\n",
    "            paths = dataset.captions[\"image_file\"][start_index:end_index].values\n",
    "            captions = dataset.captions[\"caption\"][start_index:end_index].values\n",
    "            encoded_as_string = list(map(lambda item: np.array2string(item, separator=',', max_line_width=50000, formatter={'int':lambda x: str(x)}), encoded))\n",
    "            batch_df = pd.DataFrame.from_dict({\"image_file\": paths, \"caption\": captions, \"encoding\": encoded_as_string})\n",
    "            batch_df.to_csv(file, sep='\\t', header=(n==0), index=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "id": "7704863d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4850/4850 [2:27:51<00:00,  1.83s/it]\n"
     ]
    }
   ],
   "source": [
    "encode_captioned_dataset(dataset, yfcc100m_output, batch_size=64, num_workers=16)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8953dd84",
   "metadata": {},
   "source": [
    "----"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "db471c52d602b4f5f40ecaf278e88ccfef85c29d0a1a07185b0d51fc7acf4e26"
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}