Hugging Face's logo

trevorbchen
/
diffusion_memorization

Model card Files
xet
 Community
trevorbchen commited on
Commit
a6d1ad4
·
verified ·
1 Parent(s): 9bd1e58

Add files using upload-large-folder tool

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d100/runner.log +502 -0
  2. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120/config.json +26 -0
  3. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120/metrics.jsonl +0 -0
  4. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120/runner.log +502 -0
  5. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140/config.json +26 -0
  6. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140/metrics.jsonl +0 -0
  7. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140/runner.log +502 -0
  8. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160/config.json +26 -0
  9. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160/metrics.jsonl +0 -0
  10. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160/runner.log +502 -0
  11. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180/config.json +26 -0
  12. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180/metrics.jsonl +0 -0
  13. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180/runner.log +502 -0
  14. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20/config.json +26 -0
  15. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20/metrics.jsonl +0 -0
  16. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20/runner.log +502 -0
  17. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200/config.json +26 -0
  18. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200/metrics.jsonl +0 -0
  19. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200/runner.log +502 -0
  20. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30/config.json +26 -0
  21. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30/metrics.jsonl +0 -0
  22. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30/runner.log +502 -0
  23. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40/config.json +26 -0
  24. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40/metrics.jsonl +0 -0
  25. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40/runner.log +502 -0
  26. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50/config.json +26 -0
  27. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50/metrics.jsonl +0 -0
  28. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50/runner.log +502 -0
  29. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60/config.json +26 -0
  30. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60/metrics.jsonl +0 -0
  31. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60/runner.log +502 -0
  32. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70/config.json +26 -0
  33. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70/metrics.jsonl +0 -0
  34. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70/runner.log +502 -0
  35. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80/config.json +26 -0
  36. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80/metrics.jsonl +0 -0
  37. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80/runner.log +502 -0
  38. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90/config.json +26 -0
  39. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90/metrics.jsonl +0 -0
  40. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90/runner.log +502 -0
  41. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10/config.json +26 -0
  42. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10/metrics.jsonl +0 -0
  43. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10/runner.log +502 -0
  44. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100/config.json +26 -0
  45. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100/metrics.jsonl +0 -0
  46. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100/runner.log +502 -0
  47. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120/config.json +26 -0
  48. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120/metrics.jsonl +0 -0
  49. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120/runner.log +502 -0
  50. results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d140/config.json +26 -0
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d100/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 73s | train_loss=0.9964 | fid=289.3122
3
+ step 10000 | 168s | train_loss=0.8072
4
+ step 20000 | 264s | train_loss=0.6833
5
+ step 30000 | 360s | train_loss=0.5477
6
+ step 40000 | 456s | train_loss=0.4074
7
+ step 50000 | 551s | train_loss=0.2949
8
+ step 60000 | 647s | train_loss=0.2196
9
+ step 70000 | 743s | train_loss=0.1773
10
+ step 80000 | 839s | train_loss=0.1533
11
+ step 90000 | 934s | train_loss=0.1390
12
+ step 100000 | 1097s | train_loss=0.1271 | fid=57.8864
13
+ step 110000 | 1193s | train_loss=0.1220
14
+ step 120000 | 1289s | train_loss=0.1127
15
+ step 130000 | 1385s | train_loss=0.1088
16
+ step 140000 | 1481s | train_loss=0.1053
17
+ step 150000 | 1577s | train_loss=0.1000
18
+ step 160000 | 1672s | train_loss=0.0960
19
+ step 170000 | 1768s | train_loss=0.0922
20
+ step 180000 | 1864s | train_loss=0.0898
21
+ step 190000 | 1959s | train_loss=0.0885
22
+ step 200000 | 2123s | train_loss=0.0864 | fid=57.2715
23
+ step 210000 | 2219s | train_loss=0.0821
24
+ step 220000 | 2315s | train_loss=0.0815
25
+ step 230000 | 2410s | train_loss=0.0789
26
+ step 240000 | 2506s | train_loss=0.0752
27
+ step 250000 | 2602s | train_loss=0.0755
28
+ step 260000 | 2698s | train_loss=0.0730
29
+ step 270000 | 2793s | train_loss=0.0722
30
+ step 280000 | 2889s | train_loss=0.0681
31
+ step 290000 | 2982s | train_loss=0.0687
32
+ step 300000 | 3149s | train_loss=0.0658 | fid=56.1550
33
+ step 310000 | 3245s | train_loss=0.0655
34
+ step 320000 | 3341s | train_loss=0.0656
35
+ step 330000 | 3437s | train_loss=0.0630
36
+ step 340000 | 3533s | train_loss=0.0641
37
+ step 350000 | 3629s | train_loss=0.0624
38
+ step 360000 | 3724s | train_loss=0.0613
39
+ step 370000 | 3820s | train_loss=0.0606
40
+ step 380000 | 3913s | train_loss=0.0579
41
+ step 390000 | 4009s | train_loss=0.0590
42
+ step 400000 | 4175s | train_loss=0.0569 | fid=55.3632
43
+ step 410000 | 4271s | train_loss=0.0569
44
+ step 420000 | 4367s | train_loss=0.0561
45
+ step 430000 | 4463s | train_loss=0.0553
46
+ step 440000 | 4559s | train_loss=0.0546
47
+ step 450000 | 4654s | train_loss=0.0548
48
+ step 460000 | 4750s | train_loss=0.0550
49
+ step 470000 | 4843s | train_loss=0.0528
50
+ step 480000 | 4939s | train_loss=0.0535
51
+ step 490000 | 5035s | train_loss=0.0525
52
+ step 500000 | 5201s | train_loss=0.0522 | fid=53.2239
53
+ step 510000 | 5297s | train_loss=0.0528
54
+ step 520000 | 5392s | train_loss=0.0516
55
+ step 530000 | 5488s | train_loss=0.0504
56
+ step 540000 | 5583s | train_loss=0.0529
57
+ step 550000 | 5679s | train_loss=0.0511
58
+ step 560000 | 5775s | train_loss=0.0481
59
+ step 570000 | 5868s | train_loss=0.0481
60
+ step 580000 | 5963s | train_loss=0.0474
61
+ step 590000 | 6059s | train_loss=0.0476
62
+ step 600000 | 6225s | train_loss=0.0465 | fid=51.1204
63
+ step 610000 | 6320s | train_loss=0.0487
64
+ step 620000 | 6416s | train_loss=0.0464
65
+ step 630000 | 6512s | train_loss=0.0470
66
+ step 640000 | 6607s | train_loss=0.0471
67
+ step 650000 | 6703s | train_loss=0.0454
68
+ step 660000 | 6796s | train_loss=0.0451
69
+ step 670000 | 6891s | train_loss=0.0450
70
+ step 680000 | 6987s | train_loss=0.0453
71
+ step 690000 | 7083s | train_loss=0.0464
72
+ step 700000 | 7249s | train_loss=0.0448 | fid=49.3855
73
+ step 710000 | 7344s | train_loss=0.0439
74
+ step 720000 | 7440s | train_loss=0.0436
75
+ step 730000 | 7536s | train_loss=0.0429
76
+ step 740000 | 7632s | train_loss=0.0443
77
+ step 750000 | 7725s | train_loss=0.0448
78
+ step 760000 | 7821s | train_loss=0.0444
79
+ step 770000 | 7916s | train_loss=0.0412
80
+ step 780000 | 8012s | train_loss=0.0430
81
+ step 790000 | 8108s | train_loss=0.0433
82
+ step 800000 | 8274s | train_loss=0.0406 | fid=50.3185
83
+ step 810000 | 8370s | train_loss=0.0415
84
+ step 820000 | 8465s | train_loss=0.0412
85
+ step 830000 | 8561s | train_loss=0.0409
86
+ step 840000 | 8656s | train_loss=0.0416
87
+ step 850000 | 8750s | train_loss=0.0404
88
+ step 860000 | 8846s | train_loss=0.0404
89
+ step 870000 | 8942s | train_loss=0.0395
90
+ step 880000 | 9038s | train_loss=0.0391
91
+ step 890000 | 9133s | train_loss=0.0400
92
+ step 900000 | 9298s | train_loss=0.0388 | fid=49.9891
93
+ step 910000 | 9394s | train_loss=0.0393
94
+ step 920000 | 9490s | train_loss=0.0401
95
+ step 930000 | 9586s | train_loss=0.0389
96
+ step 940000 | 9679s | train_loss=0.0392
97
+ step 950000 | 9774s | train_loss=0.0387
98
+ step 960000 | 9870s | train_loss=0.0385
99
+ step 970000 | 9966s | train_loss=0.0375
100
+ step 980000 | 10061s | train_loss=0.0371
101
+ step 990000 | 10157s | train_loss=0.0375
102
+ step 1000000 | 10324s | train_loss=0.0371 | fid=50.5376
103
+ step 1010000 | 10419s | train_loss=0.0372
104
+ step 1020000 | 10515s | train_loss=0.0377
105
+ step 1030000 | 10608s | train_loss=0.0371
106
+ step 1040000 | 10704s | train_loss=0.0359
107
+ step 1050000 | 10799s | train_loss=0.0375
108
+ step 1060000 | 10895s | train_loss=0.0364
109
+ step 1070000 | 10990s | train_loss=0.0373
110
+ step 1080000 | 11086s | train_loss=0.0360
111
+ step 1090000 | 11183s | train_loss=0.0373
112
+ step 1100000 | 11350s | train_loss=0.0359 | fid=50.6930
113
+ step 1110000 | 11446s | train_loss=0.0350
114
+ step 1120000 | 11538s | train_loss=0.0360
115
+ step 1130000 | 11634s | train_loss=0.0355
116
+ step 1140000 | 11729s | train_loss=0.0341
117
+ step 1150000 | 11825s | train_loss=0.0360
118
+ step 1160000 | 11920s | train_loss=0.0354
119
+ step 1170000 | 12016s | train_loss=0.0364
120
+ step 1180000 | 12112s | train_loss=0.0335
121
+ step 1190000 | 12209s | train_loss=0.0334
122
+ step 1200000 | 12374s | train_loss=0.0362 | fid=52.7425
123
+ step 1210000 | 12470s | train_loss=0.0333
124
+ step 1220000 | 12563s | train_loss=0.0339
125
+ step 1230000 | 12659s | train_loss=0.0336
126
+ step 1240000 | 12754s | train_loss=0.0332
127
+ step 1250000 | 12850s | train_loss=0.0363
128
+ step 1260000 | 12946s | train_loss=0.0328
129
+ step 1270000 | 13042s | train_loss=0.0324
130
+ step 1280000 | 13137s | train_loss=0.0324
131
+ step 1290000 | 13234s | train_loss=0.0315
132
+ step 1300000 | 13399s | train_loss=0.0330 | fid=56.8269
133
+ step 1310000 | 13492s | train_loss=0.0330
134
+ step 1320000 | 13587s | train_loss=0.0317
135
+ step 1330000 | 13683s | train_loss=0.0319
136
+ step 1340000 | 13779s | train_loss=0.0329
137
+ step 1350000 | 13875s | train_loss=0.0319
138
+ step 1360000 | 13970s | train_loss=0.0303
139
+ step 1370000 | 14066s | train_loss=0.0328
140
+ step 1380000 | 14162s | train_loss=0.0311
141
+ step 1390000 | 14257s | train_loss=0.0320
142
+ step 1400000 | 14420s | train_loss=0.0312 | fid=60.0658
143
+ step 1410000 | 14516s | train_loss=0.0310
144
+ step 1420000 | 14611s | train_loss=0.0308
145
+ step 1430000 | 14707s | train_loss=0.0309
146
+ step 1440000 | 14803s | train_loss=0.0299
147
+ step 1450000 | 14899s | train_loss=0.0302
148
+ step 1460000 | 14995s | train_loss=0.0304
149
+ step 1470000 | 15090s | train_loss=0.0310
150
+ step 1480000 | 15186s | train_loss=0.0312
151
+ step 1490000 | 15280s | train_loss=0.0299
152
+ step 1500000 | 15436s | train_loss=0.0305 | fid=60.2249
153
+ step 1510000 | 15531s | train_loss=0.0299
154
+ step 1520000 | 15625s | train_loss=0.0300
155
+ step 1530000 | 15719s | train_loss=0.0299
156
+ step 1540000 | 15815s | train_loss=0.0289
157
+ step 1550000 | 15911s | train_loss=0.0290
158
+ step 1560000 | 16007s | train_loss=0.0300
159
+ step 1570000 | 16102s | train_loss=0.0290
160
+ step 1580000 | 16198s | train_loss=0.0299
161
+ step 1590000 | 16289s | train_loss=0.0288
162
+ step 1600000 | 16456s | train_loss=0.0284 | fid=61.6971
163
+ step 1610000 | 16552s | train_loss=0.0283
164
+ step 1620000 | 16647s | train_loss=0.0291
165
+ step 1630000 | 16743s | train_loss=0.0285
166
+ step 1640000 | 16839s | train_loss=0.0300
167
+ step 1650000 | 16935s | train_loss=0.0287
168
+ step 1660000 | 17031s | train_loss=0.0279
169
+ step 1670000 | 17126s | train_loss=0.0277
170
+ step 1680000 | 17222s | train_loss=0.0279
171
+ step 1690000 | 17308s | train_loss=0.0287
172
+ step 1700000 | 17477s | train_loss=0.0280 | fid=57.4990
173
+ step 1710000 | 17573s | train_loss=0.0276
174
+ step 1720000 | 17669s | train_loss=0.0272
175
+ step 1730000 | 17764s | train_loss=0.0280
176
+ step 1740000 | 17860s | train_loss=0.0277
177
+ step 1750000 | 17955s | train_loss=0.0273
178
+ step 1760000 | 18051s | train_loss=0.0270
179
+ step 1770000 | 18146s | train_loss=0.0276
180
+ step 1780000 | 18243s | train_loss=0.0273
181
+ step 1790000 | 18330s | train_loss=0.0276
182
+ step 1800000 | 18501s | train_loss=0.0274 | fid=56.3094
183
+ step 1810000 | 18597s | train_loss=0.0263
184
+ step 1820000 | 18693s | train_loss=0.0263
185
+ step 1830000 | 18788s | train_loss=0.0273
186
+ step 1840000 | 18884s | train_loss=0.0272
187
+ step 1850000 | 18980s | train_loss=0.0264
188
+ step 1860000 | 19076s | train_loss=0.0273
189
+ step 1870000 | 19170s | train_loss=0.0264
190
+ step 1880000 | 19263s | train_loss=0.0258
191
+ step 1890000 | 19353s | train_loss=0.0285
192
+ step 1900000 | 19521s | train_loss=0.0259 | fid=52.2663
193
+ step 1910000 | 19616s | train_loss=0.0252
194
+ step 1920000 | 19710s | train_loss=0.0259
195
+ step 1930000 | 19805s | train_loss=0.0258
196
+ step 1940000 | 19899s | train_loss=0.0254
197
+ step 1950000 | 19994s | train_loss=0.0258
198
+ step 1960000 | 20090s | train_loss=0.0256
199
+ step 1970000 | 20183s | train_loss=0.0257
200
+ step 1980000 | 20278s | train_loss=0.0260
201
+ step 1990000 | 20371s | train_loss=0.0254
202
+ step 2000000 | 20540s | train_loss=0.0244 | fid=51.8307
203
+ step 2010000 | 20636s | train_loss=0.0257
204
+ step 2020000 | 20731s | train_loss=0.0255
205
+ step 2030000 | 20826s | train_loss=0.0246
206
+ step 2040000 | 20921s | train_loss=0.0240
207
+ step 2050000 | 21015s | train_loss=0.0241
208
+ step 2060000 | 21110s | train_loss=0.0251
209
+ step 2070000 | 21202s | train_loss=0.0243
210
+ step 2080000 | 21296s | train_loss=0.0238
211
+ step 2090000 | 21387s | train_loss=0.0236
212
+ step 2100000 | 21555s | train_loss=0.0238 | fid=50.8899
213
+ step 2110000 | 21650s | train_loss=0.0232
214
+ step 2120000 | 21745s | train_loss=0.0252
215
+ step 2130000 | 21840s | train_loss=0.0231
216
+ step 2140000 | 21936s | train_loss=0.0234
217
+ step 2150000 | 22032s | train_loss=0.0243
218
+ step 2160000 | 22124s | train_loss=0.0243
219
+ step 2170000 | 22220s | train_loss=0.0235
220
+ step 2180000 | 22316s | train_loss=0.0242
221
+ step 2190000 | 22407s | train_loss=0.0233
222
+ step 2200000 | 22577s | train_loss=0.0227 | fid=48.7237
223
+ step 2210000 | 22673s | train_loss=0.0225
224
+ step 2220000 | 22768s | train_loss=0.0225
225
+ step 2230000 | 22864s | train_loss=0.0232
226
+ step 2240000 | 22960s | train_loss=0.0235
227
+ step 2250000 | 23049s | train_loss=0.0224
228
+ step 2260000 | 23136s | train_loss=0.0233
229
+ step 2270000 | 23230s | train_loss=0.0228
230
+ step 2280000 | 23321s | train_loss=0.0224
231
+ step 2290000 | 23414s | train_loss=0.0216
232
+ step 2300000 | 23581s | train_loss=0.0223 | fid=48.4139
233
+ step 2310000 | 23675s | train_loss=0.0225
234
+ step 2320000 | 23769s | train_loss=0.0230
235
+ step 2330000 | 23864s | train_loss=0.0224
236
+ step 2340000 | 23959s | train_loss=0.0215
237
+ step 2350000 | 24050s | train_loss=0.0218
238
+ step 2360000 | 24145s | train_loss=0.0226
239
+ step 2370000 | 24240s | train_loss=0.0227
240
+ step 2380000 | 24330s | train_loss=0.0220
241
+ step 2390000 | 24425s | train_loss=0.0212
242
+ step 2400000 | 24593s | train_loss=0.0210 | fid=46.2694
243
+ step 2410000 | 24687s | train_loss=0.0218
244
+ step 2420000 | 24782s | train_loss=0.0225
245
+ step 2430000 | 24877s | train_loss=0.0219
246
+ step 2440000 | 24971s | train_loss=0.0213
247
+ step 2450000 | 25063s | train_loss=0.0220
248
+ step 2460000 | 25158s | train_loss=0.0219
249
+ step 2470000 | 25253s | train_loss=0.0219
250
+ step 2480000 | 25342s | train_loss=0.0218
251
+ step 2490000 | 25437s | train_loss=0.0211
252
+ step 2500000 | 25605s | train_loss=0.0220 | fid=46.9994
253
+ step 2510000 | 25699s | train_loss=0.0217
254
+ step 2520000 | 25793s | train_loss=0.0209
255
+ step 2530000 | 25888s | train_loss=0.0212
256
+ step 2540000 | 25979s | train_loss=0.0208
257
+ step 2550000 | 26074s | train_loss=0.0211
258
+ step 2560000 | 26169s | train_loss=0.0205
259
+ step 2570000 | 26263s | train_loss=0.0203
260
+ step 2580000 | 26354s | train_loss=0.0211
261
+ step 2590000 | 26449s | train_loss=0.0207
262
+ step 2600000 | 26618s | train_loss=0.0203 | fid=45.7667
263
+ step 2610000 | 26712s | train_loss=0.0205
264
+ step 2620000 | 26807s | train_loss=0.0203
265
+ step 2630000 | 26898s | train_loss=0.0200
266
+ step 2640000 | 26993s | train_loss=0.0205
267
+ step 2650000 | 27088s | train_loss=0.0202
268
+ step 2660000 | 27182s | train_loss=0.0227
269
+ step 2670000 | 27276s | train_loss=0.0202
270
+ step 2680000 | 27368s | train_loss=0.0201
271
+ step 2690000 | 27462s | train_loss=0.0198
272
+ step 2700000 | 27630s | train_loss=0.0209 | fid=46.8937
273
+ step 2710000 | 27725s | train_loss=0.0198
274
+ step 2720000 | 27819s | train_loss=0.0204
275
+ step 2730000 | 27912s | train_loss=0.0195
276
+ step 2740000 | 28007s | train_loss=0.0203
277
+ step 2750000 | 28103s | train_loss=0.0196
278
+ step 2760000 | 28198s | train_loss=0.0198
279
+ step 2770000 | 28292s | train_loss=0.0195
280
+ step 2780000 | 28384s | train_loss=0.0192
281
+ step 2790000 | 28474s | train_loss=0.0196
282
+ step 2800000 | 28634s | train_loss=0.0204 | fid=44.9500
283
+ step 2810000 | 28723s | train_loss=0.0188
284
+ step 2820000 | 28812s | train_loss=0.0195
285
+ step 2830000 | 28908s | train_loss=0.0194
286
+ step 2840000 | 29003s | train_loss=0.0193
287
+ step 2850000 | 29098s | train_loss=0.0209
288
+ step 2860000 | 29195s | train_loss=0.0188
289
+ step 2870000 | 29290s | train_loss=0.0186
290
+ step 2880000 | 29382s | train_loss=0.0186
291
+ step 2890000 | 29472s | train_loss=0.0193
292
+ step 2900000 | 29631s | train_loss=0.0207 | fid=46.5905
293
+ step 2910000 | 29720s | train_loss=0.0187
294
+ step 2920000 | 29815s | train_loss=0.0191
295
+ step 2930000 | 29911s | train_loss=0.0193
296
+ step 2940000 | 30006s | train_loss=0.0199
297
+ step 2950000 | 30102s | train_loss=0.0188
298
+ step 2960000 | 30198s | train_loss=0.0182
299
+ step 2970000 | 30293s | train_loss=0.0191
300
+ step 2980000 | 30386s | train_loss=0.0195
301
+ step 2990000 | 30481s | train_loss=0.0190
302
+ step 3000000 | 30650s | train_loss=0.0188 | fid=43.8188
303
+ step 3010000 | 30743s | train_loss=0.0189
304
+ step 3020000 | 30839s | train_loss=0.0181
305
+ step 3030000 | 30934s | train_loss=0.0179
306
+ step 3040000 | 31029s | train_loss=0.0191
307
+ step 3050000 | 31125s | train_loss=0.0188
308
+ step 3060000 | 31222s | train_loss=0.0180
309
+ step 3070000 | 31314s | train_loss=0.0189
310
+ step 3080000 | 31408s | train_loss=0.0190
311
+ step 3090000 | 31503s | train_loss=0.0185
312
+ step 3100000 | 31665s | train_loss=0.0186 | fid=44.2686
313
+ step 3110000 | 31755s | train_loss=0.0191
314
+ step 3120000 | 31849s | train_loss=0.0175
315
+ step 3130000 | 31944s | train_loss=0.0175
316
+ step 3140000 | 32039s | train_loss=0.0188
317
+ step 3150000 | 32134s | train_loss=0.0178
318
+ step 3160000 | 32228s | train_loss=0.0180
319
+ step 3170000 | 32319s | train_loss=0.0179
320
+ step 3180000 | 32415s | train_loss=0.0200
321
+ step 3190000 | 32510s | train_loss=0.0182
322
+ step 3200000 | 32675s | train_loss=0.0194 | fid=43.0592
323
+ step 3210000 | 32771s | train_loss=0.0176
324
+ step 3220000 | 32866s | train_loss=0.0179
325
+ step 3230000 | 32961s | train_loss=0.0185
326
+ step 3240000 | 33055s | train_loss=0.0175
327
+ step 3250000 | 33150s | train_loss=0.0186
328
+ step 3260000 | 33243s | train_loss=0.0177
329
+ step 3270000 | 33335s | train_loss=0.0170
330
+ step 3280000 | 33429s | train_loss=0.0178
331
+ step 3290000 | 33525s | train_loss=0.0183
332
+ step 3300000 | 33692s | train_loss=0.0175 | fid=44.0758
333
+ step 3310000 | 33788s | train_loss=0.0176
334
+ step 3320000 | 33883s | train_loss=0.0173
335
+ step 3330000 | 33977s | train_loss=0.0175
336
+ step 3340000 | 34073s | train_loss=0.0173
337
+ step 3350000 | 34169s | train_loss=0.0176
338
+ step 3360000 | 34256s | train_loss=0.0165
339
+ step 3370000 | 34349s | train_loss=0.0177
340
+ step 3380000 | 34445s | train_loss=0.0174
341
+ step 3390000 | 34537s | train_loss=0.0172
342
+ step 3400000 | 34700s | train_loss=0.0175 | fid=42.7610
343
+ step 3410000 | 34795s | train_loss=0.0168
344
+ step 3420000 | 34891s | train_loss=0.0168
345
+ step 3430000 | 34986s | train_loss=0.0183
346
+ step 3440000 | 35082s | train_loss=0.0173
347
+ step 3450000 | 35178s | train_loss=0.0177
348
+ step 3460000 | 35269s | train_loss=0.0166
349
+ step 3470000 | 35352s | train_loss=0.0175
350
+ step 3480000 | 35448s | train_loss=0.0167
351
+ step 3490000 | 35536s | train_loss=0.0162
352
+ step 3500000 | 35700s | train_loss=0.0167 | fid=42.9117
353
+ step 3510000 | 35795s | train_loss=0.0181
354
+ step 3520000 | 35891s | train_loss=0.0171
355
+ step 3530000 | 35987s | train_loss=0.0164
356
+ step 3540000 | 36083s | train_loss=0.0166
357
+ step 3550000 | 36178s | train_loss=0.0160
358
+ step 3560000 | 36275s | train_loss=0.0162
359
+ step 3570000 | 36362s | train_loss=0.0174
360
+ step 3580000 | 36457s | train_loss=0.0166
361
+ step 3590000 | 36542s | train_loss=0.0165
362
+ step 3600000 | 36710s | train_loss=0.0169 | fid=42.9397
363
+ step 3610000 | 36806s | train_loss=0.0179
364
+ step 3620000 | 36902s | train_loss=0.0155
365
+ step 3630000 | 36997s | train_loss=0.0157
366
+ step 3640000 | 37093s | train_loss=0.0171
367
+ step 3650000 | 37189s | train_loss=0.0167
368
+ step 3660000 | 37286s | train_loss=0.0177
369
+ step 3670000 | 37376s | train_loss=0.0163
370
+ step 3680000 | 37472s | train_loss=0.0157
371
+ step 3690000 | 37556s | train_loss=0.0165
372
+ step 3700000 | 37726s | train_loss=0.0175 | fid=42.5667
373
+ step 3710000 | 37822s | train_loss=0.0159
374
+ step 3720000 | 37918s | train_loss=0.0165
375
+ step 3730000 | 38014s | train_loss=0.0165
376
+ step 3740000 | 38110s | train_loss=0.0162
377
+ step 3750000 | 38206s | train_loss=0.0159
378
+ step 3760000 | 38301s | train_loss=0.0160
379
+ step 3770000 | 38392s | train_loss=0.0160
380
+ step 3780000 | 38488s | train_loss=0.0172
381
+ step 3790000 | 38572s | train_loss=0.0158
382
+ step 3800000 | 38742s | train_loss=0.0157 | fid=42.7331
383
+ step 3810000 | 38837s | train_loss=0.0168
384
+ step 3820000 | 38933s | train_loss=0.0161
385
+ step 3830000 | 39029s | train_loss=0.0173
386
+ step 3840000 | 39125s | train_loss=0.0166
387
+ step 3850000 | 39220s | train_loss=0.0154
388
+ step 3860000 | 39314s | train_loss=0.0160
389
+ step 3870000 | 39408s | train_loss=0.0163
390
+ step 3880000 | 39499s | train_loss=0.0162
391
+ step 3890000 | 39587s | train_loss=0.0158
392
+ step 3900000 | 39758s | train_loss=0.0163 | fid=42.5941
393
+ step 3910000 | 39853s | train_loss=0.0165
394
+ step 3920000 | 39949s | train_loss=0.0169
395
+ step 3930000 | 40045s | train_loss=0.0148
396
+ step 3940000 | 40141s | train_loss=0.0158
397
+ step 3950000 | 40236s | train_loss=0.0158
398
+ step 3960000 | 40328s | train_loss=0.0154
399
+ step 3970000 | 40423s | train_loss=0.0154
400
+ step 3980000 | 40510s | train_loss=0.0164
401
+ step 3990000 | 40602s | train_loss=0.0158
402
+ step 4000000 | 40772s | train_loss=0.0152 | fid=43.1700
403
+ step 4010000 | 40868s | train_loss=0.0158
404
+ step 4020000 | 40964s | train_loss=0.0165
405
+ step 4030000 | 41059s | train_loss=0.0159
406
+ step 4040000 | 41154s | train_loss=0.0151
407
+ step 4050000 | 41247s | train_loss=0.0150
408
+ step 4060000 | 41335s | train_loss=0.0148
409
+ step 4070000 | 41426s | train_loss=0.0158
410
+ step 4080000 | 41511s | train_loss=0.0152
411
+ step 4090000 | 41603s | train_loss=0.0163
412
+ step 4100000 | 41766s | train_loss=0.0154 | fid=42.1916
413
+ step 4110000 | 41859s | train_loss=0.0148
414
+ step 4120000 | 41951s | train_loss=0.0155
415
+ step 4130000 | 42043s | train_loss=0.0157
416
+ step 4140000 | 42136s | train_loss=0.0146
417
+ step 4150000 | 42229s | train_loss=0.0156
418
+ step 4160000 | 42316s | train_loss=0.0150
419
+ step 4170000 | 42409s | train_loss=0.0158
420
+ step 4180000 | 42497s | train_loss=0.0149
421
+ step 4190000 | 42589s | train_loss=0.0156
422
+ step 4200000 | 42753s | train_loss=0.0152 | fid=41.2330
423
+ step 4210000 | 42846s | train_loss=0.0166
424
+ step 4220000 | 42938s | train_loss=0.0150
425
+ step 4230000 | 43030s | train_loss=0.0155
426
+ step 4240000 | 43122s | train_loss=0.0152
427
+ step 4250000 | 43215s | train_loss=0.0148
428
+ step 4260000 | 43301s | train_loss=0.0145
429
+ step 4270000 | 43389s | train_loss=0.0147
430
+ step 4280000 | 43481s | train_loss=0.0152
431
+ step 4290000 | 43574s | train_loss=0.0149
432
+ step 4300000 | 43738s | train_loss=0.0146 | fid=40.6116
433
+ step 4310000 | 43831s | train_loss=0.0145
434
+ step 4320000 | 43923s | train_loss=0.0148
435
+ step 4330000 | 44015s | train_loss=0.0159
436
+ step 4340000 | 44107s | train_loss=0.0143
437
+ step 4350000 | 44200s | train_loss=0.0144
438
+ step 4360000 | 44285s | train_loss=0.0141
439
+ step 4370000 | 44374s | train_loss=0.0145
440
+ step 4380000 | 44465s | train_loss=0.0153
441
+ step 4390000 | 44557s | train_loss=0.0152
442
+ step 4400000 | 44722s | train_loss=0.0140 | fid=40.8679
443
+ step 4410000 | 44814s | train_loss=0.0145
444
+ step 4420000 | 44907s | train_loss=0.0156
445
+ step 4430000 | 44999s | train_loss=0.0144
446
+ step 4440000 | 45092s | train_loss=0.0151
447
+ step 4450000 | 45184s | train_loss=0.0137
448
+ step 4460000 | 45269s | train_loss=0.0136
449
+ step 4470000 | 45361s | train_loss=0.0159
450
+ step 4480000 | 45452s | train_loss=0.0151
451
+ step 4490000 | 45544s | train_loss=0.0140
452
+ step 4500000 | 45707s | train_loss=0.0140 | fid=40.9575
453
+ step 4510000 | 45800s | train_loss=0.0140
454
+ step 4520000 | 45892s | train_loss=0.0151
455
+ step 4530000 | 45985s | train_loss=0.0141
456
+ step 4540000 | 46078s | train_loss=0.0137
457
+ step 4550000 | 46170s | train_loss=0.0147
458
+ step 4560000 | 46261s | train_loss=0.0144
459
+ step 4570000 | 46356s | train_loss=0.0139
460
+ step 4580000 | 46451s | train_loss=0.0141
461
+ step 4590000 | 46547s | train_loss=0.0143
462
+ step 4600000 | 46720s | train_loss=0.0136 | fid=41.2447
463
+ step 4610000 | 46816s | train_loss=0.0141
464
+ step 4620000 | 46912s | train_loss=0.0141
465
+ step 4630000 | 47008s | train_loss=0.0139
466
+ step 4640000 | 47103s | train_loss=0.0144
467
+ step 4650000 | 47193s | train_loss=0.0132
468
+ step 4660000 | 47288s | train_loss=0.0156
469
+ step 4670000 | 47384s | train_loss=0.0134
470
+ step 4680000 | 47479s | train_loss=0.0138
471
+ step 4690000 | 47576s | train_loss=0.0149
472
+ step 4700000 | 47747s | train_loss=0.0140 | fid=39.8078
473
+ step 4710000 | 47843s | train_loss=0.0131
474
+ step 4720000 | 47939s | train_loss=0.0146
475
+ step 4730000 | 48035s | train_loss=0.0143
476
+ step 4740000 | 48124s | train_loss=0.0136
477
+ step 4750000 | 48197s | train_loss=0.0134
478
+ step 4760000 | 48274s | train_loss=0.0141
479
+ step 4770000 | 48351s | train_loss=0.0138
480
+ step 4780000 | 48427s | train_loss=0.0133
481
+ step 4790000 | 48504s | train_loss=0.0147
482
+ step 4800000 | 48643s | train_loss=0.0136 | fid=40.5253
483
+ step 4810000 | 48720s | train_loss=0.0133
484
+ step 4820000 | 48796s | train_loss=0.0138
485
+ step 4830000 | 48872s | train_loss=0.0136
486
+ step 4840000 | 48948s | train_loss=0.0143
487
+ step 4850000 | 49018s | train_loss=0.0132
488
+ step 4860000 | 49095s | train_loss=0.0140
489
+ step 4870000 | 49173s | train_loss=0.0131
490
+ step 4880000 | 49251s | train_loss=0.0135
491
+ step 4890000 | 49328s | train_loss=0.0135
492
+ step 4900000 | 49468s | train_loss=0.0137 | fid=39.9001
493
+ step 4910000 | 49545s | train_loss=0.0133
494
+ step 4920000 | 49622s | train_loss=0.0133
495
+ step 4930000 | 49699s | train_loss=0.0129
496
+ step 4940000 | 49777s | train_loss=0.0139
497
+ step 4950000 | 49823s | train_loss=0.0140
498
+ step 4960000 | 49845s | train_loss=0.0136
499
+ step 4970000 | 49867s | train_loss=0.0131
500
+ step 4980000 | 49889s | train_loss=0.0136
501
+ step 4990000 | 49911s | train_loss=0.0127
502
+ step 5000000 | 49951s | train_loss=0.0131 | fid=39.6069
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d120/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d120/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 67s | train_loss=0.9960 | fid=266.4189
3
+ step 10000 | 133s | train_loss=0.7891
4
+ step 20000 | 221s | train_loss=0.6532
5
+ step 30000 | 309s | train_loss=0.5202
6
+ step 40000 | 397s | train_loss=0.4014
7
+ step 50000 | 485s | train_loss=0.3037
8
+ step 60000 | 573s | train_loss=0.2347
9
+ step 70000 | 661s | train_loss=0.1922
10
+ step 80000 | 749s | train_loss=0.1649
11
+ step 90000 | 837s | train_loss=0.1450
12
+ step 100000 | 992s | train_loss=0.1341 | fid=64.4376
13
+ step 110000 | 1079s | train_loss=0.1282
14
+ step 120000 | 1155s | train_loss=0.1181
15
+ step 130000 | 1243s | train_loss=0.1107
16
+ step 140000 | 1331s | train_loss=0.1077
17
+ step 150000 | 1419s | train_loss=0.1025
18
+ step 160000 | 1507s | train_loss=0.1009
19
+ step 170000 | 1595s | train_loss=0.0951
20
+ step 180000 | 1683s | train_loss=0.0909
21
+ step 190000 | 1771s | train_loss=0.0900
22
+ step 200000 | 1927s | train_loss=0.0866 | fid=65.3789
23
+ step 210000 | 2015s | train_loss=0.0835
24
+ step 220000 | 2101s | train_loss=0.0828
25
+ step 230000 | 2182s | train_loss=0.0798
26
+ step 240000 | 2270s | train_loss=0.0789
27
+ step 250000 | 2358s | train_loss=0.0780
28
+ step 260000 | 2446s | train_loss=0.0751
29
+ step 270000 | 2534s | train_loss=0.0725
30
+ step 280000 | 2622s | train_loss=0.0715
31
+ step 290000 | 2710s | train_loss=0.0695
32
+ step 300000 | 2866s | train_loss=0.0728 | fid=63.1521
33
+ step 310000 | 2953s | train_loss=0.0670
34
+ step 320000 | 3039s | train_loss=0.0671
35
+ step 330000 | 3128s | train_loss=0.0654
36
+ step 340000 | 3211s | train_loss=0.0655
37
+ step 350000 | 3299s | train_loss=0.0645
38
+ step 360000 | 3387s | train_loss=0.0620
39
+ step 370000 | 3475s | train_loss=0.0620
40
+ step 380000 | 3563s | train_loss=0.0616
41
+ step 390000 | 3651s | train_loss=0.0626
42
+ step 400000 | 3807s | train_loss=0.0588 | fid=61.7543
43
+ step 410000 | 3895s | train_loss=0.0588
44
+ step 420000 | 3982s | train_loss=0.0591
45
+ step 430000 | 4069s | train_loss=0.0579
46
+ step 440000 | 4157s | train_loss=0.0562
47
+ step 450000 | 4240s | train_loss=0.0580
48
+ step 460000 | 4327s | train_loss=0.0568
49
+ step 470000 | 4415s | train_loss=0.0559
50
+ step 480000 | 4503s | train_loss=0.0548
51
+ step 490000 | 4591s | train_loss=0.0552
52
+ step 500000 | 4747s | train_loss=0.0553 | fid=59.3120
53
+ step 510000 | 4835s | train_loss=0.0539
54
+ step 520000 | 4923s | train_loss=0.0539
55
+ step 530000 | 5009s | train_loss=0.0538
56
+ step 540000 | 5097s | train_loss=0.0548
57
+ step 550000 | 5185s | train_loss=0.0532
58
+ step 560000 | 5269s | train_loss=0.0497
59
+ step 570000 | 5356s | train_loss=0.0519
60
+ step 580000 | 5444s | train_loss=0.0529
61
+ step 590000 | 5532s | train_loss=0.0509
62
+ step 600000 | 5689s | train_loss=0.0489 | fid=57.9690
63
+ step 610000 | 5776s | train_loss=0.0505
64
+ step 620000 | 5864s | train_loss=0.0483
65
+ step 630000 | 5952s | train_loss=0.0491
66
+ step 640000 | 6039s | train_loss=0.0484
67
+ step 650000 | 6126s | train_loss=0.0478
68
+ step 660000 | 6214s | train_loss=0.0488
69
+ step 670000 | 6299s | train_loss=0.0471
70
+ step 680000 | 6386s | train_loss=0.0472
71
+ step 690000 | 6474s | train_loss=0.0467
72
+ step 700000 | 6630s | train_loss=0.0473 | fid=58.1777
73
+ step 710000 | 6717s | train_loss=0.0454
74
+ step 720000 | 6805s | train_loss=0.0472
75
+ step 730000 | 6893s | train_loss=0.0451
76
+ step 740000 | 6980s | train_loss=0.0462
77
+ step 750000 | 7068s | train_loss=0.0450
78
+ step 760000 | 7156s | train_loss=0.0451
79
+ step 770000 | 7244s | train_loss=0.0439
80
+ step 780000 | 7329s | train_loss=0.0454
81
+ step 790000 | 7415s | train_loss=0.0451
82
+ step 800000 | 7570s | train_loss=0.0437 | fid=56.3452
83
+ step 810000 | 7658s | train_loss=0.0454
84
+ step 820000 | 7746s | train_loss=0.0434
85
+ step 830000 | 7834s | train_loss=0.0440
86
+ step 840000 | 7922s | train_loss=0.0446
87
+ step 850000 | 8008s | train_loss=0.0436
88
+ step 860000 | 8096s | train_loss=0.0435
89
+ step 870000 | 8184s | train_loss=0.0435
90
+ step 880000 | 8272s | train_loss=0.0421
91
+ step 890000 | 8358s | train_loss=0.0451
92
+ step 900000 | 8511s | train_loss=0.0416 | fid=56.6608
93
+ step 910000 | 8599s | train_loss=0.0412
94
+ step 920000 | 8687s | train_loss=0.0424
95
+ step 930000 | 8775s | train_loss=0.0421
96
+ step 940000 | 8863s | train_loss=0.0433
97
+ step 950000 | 8949s | train_loss=0.0395
98
+ step 960000 | 9037s | train_loss=0.0407
99
+ step 970000 | 9125s | train_loss=0.0398
100
+ step 980000 | 9213s | train_loss=0.0399
101
+ step 990000 | 9301s | train_loss=0.0393
102
+ step 1000000 | 9452s | train_loss=0.0408 | fid=59.3875
103
+ step 1010000 | 9540s | train_loss=0.0392
104
+ step 1020000 | 9628s | train_loss=0.0402
105
+ step 1030000 | 9716s | train_loss=0.0388
106
+ step 1040000 | 9804s | train_loss=0.0384
107
+ step 1050000 | 9892s | train_loss=0.0401
108
+ step 1060000 | 9978s | train_loss=0.0383
109
+ step 1070000 | 10066s | train_loss=0.0395
110
+ step 1080000 | 10154s | train_loss=0.0406
111
+ step 1090000 | 10242s | train_loss=0.0388
112
+ step 1100000 | 10400s | train_loss=0.0395 | fid=57.6837
113
+ step 1110000 | 10483s | train_loss=0.0377
114
+ step 1120000 | 10571s | train_loss=0.0392
115
+ step 1130000 | 10659s | train_loss=0.0375
116
+ step 1140000 | 10747s | train_loss=0.0361
117
+ step 1150000 | 10835s | train_loss=0.0375
118
+ step 1160000 | 10922s | train_loss=0.0386
119
+ step 1170000 | 11009s | train_loss=0.0376
120
+ step 1180000 | 11097s | train_loss=0.0370
121
+ step 1190000 | 11185s | train_loss=0.0357
122
+ step 1200000 | 11341s | train_loss=0.0375 | fid=59.4218
123
+ step 1210000 | 11429s | train_loss=0.0358
124
+ step 1220000 | 11512s | train_loss=0.0360
125
+ step 1230000 | 11600s | train_loss=0.0367
126
+ step 1240000 | 11688s | train_loss=0.0362
127
+ step 1250000 | 11776s | train_loss=0.0381
128
+ step 1260000 | 11864s | train_loss=0.0361
129
+ step 1270000 | 11950s | train_loss=0.0363
130
+ step 1280000 | 12038s | train_loss=0.0350
131
+ step 1290000 | 12126s | train_loss=0.0363
132
+ step 1300000 | 12282s | train_loss=0.0347 | fid=59.3276
133
+ step 1310000 | 12370s | train_loss=0.0353
134
+ step 1320000 | 12459s | train_loss=0.0341
135
+ step 1330000 | 12542s | train_loss=0.0332
136
+ step 1340000 | 12630s | train_loss=0.0348
137
+ step 1350000 | 12718s | train_loss=0.0357
138
+ step 1360000 | 12806s | train_loss=0.0355
139
+ step 1370000 | 12894s | train_loss=0.0361
140
+ step 1380000 | 12980s | train_loss=0.0343
141
+ step 1390000 | 13068s | train_loss=0.0345
142
+ step 1400000 | 13223s | train_loss=0.0343 | fid=66.7886
143
+ step 1410000 | 13311s | train_loss=0.0338
144
+ step 1420000 | 13399s | train_loss=0.0338
145
+ step 1430000 | 13488s | train_loss=0.0342
146
+ step 1440000 | 13571s | train_loss=0.0336
147
+ step 1450000 | 13659s | train_loss=0.0343
148
+ step 1460000 | 13747s | train_loss=0.0342
149
+ step 1470000 | 13835s | train_loss=0.0338
150
+ step 1480000 | 13921s | train_loss=0.0336
151
+ step 1490000 | 14009s | train_loss=0.0329
152
+ step 1500000 | 14164s | train_loss=0.0343 | fid=64.5321
153
+ step 1510000 | 14252s | train_loss=0.0325
154
+ step 1520000 | 14340s | train_loss=0.0350
155
+ step 1530000 | 14428s | train_loss=0.0337
156
+ step 1540000 | 14516s | train_loss=0.0324
157
+ step 1550000 | 14600s | train_loss=0.0317
158
+ step 1560000 | 14687s | train_loss=0.0323
159
+ step 1570000 | 14776s | train_loss=0.0323
160
+ step 1580000 | 14863s | train_loss=0.0327
161
+ step 1590000 | 14950s | train_loss=0.0319
162
+ step 1600000 | 15106s | train_loss=0.0328 | fid=68.2866
163
+ step 1610000 | 15194s | train_loss=0.0317
164
+ step 1620000 | 15282s | train_loss=0.0312
165
+ step 1630000 | 15370s | train_loss=0.0302
166
+ step 1640000 | 15458s | train_loss=0.0319
167
+ step 1650000 | 15546s | train_loss=0.0316
168
+ step 1660000 | 15630s | train_loss=0.0297
169
+ step 1670000 | 15717s | train_loss=0.0304
170
+ step 1680000 | 15805s | train_loss=0.0298
171
+ step 1690000 | 15891s | train_loss=0.0317
172
+ step 1700000 | 16047s | train_loss=0.0311 | fid=66.0988
173
+ step 1710000 | 16135s | train_loss=0.0314
174
+ step 1720000 | 16222s | train_loss=0.0302
175
+ step 1730000 | 16311s | train_loss=0.0314
176
+ step 1740000 | 16398s | train_loss=0.0302
177
+ step 1750000 | 16487s | train_loss=0.0292
178
+ step 1760000 | 16575s | train_loss=0.0307
179
+ step 1770000 | 16660s | train_loss=0.0304
180
+ step 1780000 | 16746s | train_loss=0.0313
181
+ step 1790000 | 16834s | train_loss=0.0287
182
+ step 1800000 | 16988s | train_loss=0.0286 | fid=63.0022
183
+ step 1810000 | 17076s | train_loss=0.0287
184
+ step 1820000 | 17164s | train_loss=0.0291
185
+ step 1830000 | 17252s | train_loss=0.0298
186
+ step 1840000 | 17340s | train_loss=0.0296
187
+ step 1850000 | 17428s | train_loss=0.0280
188
+ step 1860000 | 17516s | train_loss=0.0286
189
+ step 1870000 | 17604s | train_loss=0.0280
190
+ step 1880000 | 17689s | train_loss=0.0289
191
+ step 1890000 | 17775s | train_loss=0.0290
192
+ step 1900000 | 17930s | train_loss=0.0274 | fid=61.3866
193
+ step 1910000 | 18018s | train_loss=0.0280
194
+ step 1920000 | 18106s | train_loss=0.0273
195
+ step 1930000 | 18194s | train_loss=0.0281
196
+ step 1940000 | 18282s | train_loss=0.0272
197
+ step 1950000 | 18370s | train_loss=0.0275
198
+ step 1960000 | 18458s | train_loss=0.0280
199
+ step 1970000 | 18545s | train_loss=0.0291
200
+ step 1980000 | 18634s | train_loss=0.0294
201
+ step 1990000 | 18719s | train_loss=0.0289
202
+ step 2000000 | 18872s | train_loss=0.0268 | fid=59.5174
203
+ step 2010000 | 18960s | train_loss=0.0281
204
+ step 2020000 | 19048s | train_loss=0.0282
205
+ step 2030000 | 19136s | train_loss=0.0279
206
+ step 2040000 | 19224s | train_loss=0.0263
207
+ step 2050000 | 19312s | train_loss=0.0273
208
+ step 2060000 | 19400s | train_loss=0.0286
209
+ step 2070000 | 19488s | train_loss=0.0274
210
+ step 2080000 | 19576s | train_loss=0.0257
211
+ step 2090000 | 19664s | train_loss=0.0262
212
+ step 2100000 | 19816s | train_loss=0.0268 | fid=58.7810
213
+ step 2110000 | 19903s | train_loss=0.0274
214
+ step 2120000 | 19991s | train_loss=0.0271
215
+ step 2130000 | 20079s | train_loss=0.0261
216
+ step 2140000 | 20167s | train_loss=0.0266
217
+ step 2150000 | 20255s | train_loss=0.0263
218
+ step 2160000 | 20343s | train_loss=0.0277
219
+ step 2170000 | 20431s | train_loss=0.0273
220
+ step 2180000 | 20519s | train_loss=0.0268
221
+ step 2190000 | 20607s | train_loss=0.0254
222
+ step 2200000 | 20763s | train_loss=0.0259 | fid=59.2992
223
+ step 2210000 | 20846s | train_loss=0.0262
224
+ step 2220000 | 20934s | train_loss=0.0247
225
+ step 2230000 | 21022s | train_loss=0.0267
226
+ step 2240000 | 21110s | train_loss=0.0256
227
+ step 2250000 | 21198s | train_loss=0.0256
228
+ step 2260000 | 21286s | train_loss=0.0248
229
+ step 2270000 | 21374s | train_loss=0.0252
230
+ step 2280000 | 21462s | train_loss=0.0244
231
+ step 2290000 | 21550s | train_loss=0.0241
232
+ step 2300000 | 21706s | train_loss=0.0249 | fid=57.5907
233
+ step 2310000 | 21793s | train_loss=0.0253
234
+ step 2320000 | 21875s | train_loss=0.0242
235
+ step 2330000 | 21963s | train_loss=0.0252
236
+ step 2340000 | 22051s | train_loss=0.0241
237
+ step 2350000 | 22139s | train_loss=0.0237
238
+ step 2360000 | 22227s | train_loss=0.0260
239
+ step 2370000 | 22315s | train_loss=0.0251
240
+ step 2380000 | 22403s | train_loss=0.0263
241
+ step 2390000 | 22491s | train_loss=0.0252
242
+ step 2400000 | 22647s | train_loss=0.0233 | fid=59.0513
243
+ step 2410000 | 22735s | train_loss=0.0258
244
+ step 2420000 | 22821s | train_loss=0.0242
245
+ step 2430000 | 22905s | train_loss=0.0247
246
+ step 2440000 | 22994s | train_loss=0.0245
247
+ step 2450000 | 23082s | train_loss=0.0238
248
+ step 2460000 | 23169s | train_loss=0.0243
249
+ step 2470000 | 23257s | train_loss=0.0237
250
+ step 2480000 | 23346s | train_loss=0.0247
251
+ step 2490000 | 23433s | train_loss=0.0243
252
+ step 2500000 | 23589s | train_loss=0.0233 | fid=58.2873
253
+ step 2510000 | 23678s | train_loss=0.0242
254
+ step 2520000 | 23765s | train_loss=0.0236
255
+ step 2530000 | 23851s | train_loss=0.0228
256
+ step 2540000 | 23936s | train_loss=0.0233
257
+ step 2550000 | 24023s | train_loss=0.0238
258
+ step 2560000 | 24111s | train_loss=0.0225
259
+ step 2570000 | 24199s | train_loss=0.0223
260
+ step 2580000 | 24287s | train_loss=0.0225
261
+ step 2590000 | 24375s | train_loss=0.0232
262
+ step 2600000 | 24532s | train_loss=0.0236 | fid=58.3120
263
+ step 2610000 | 24620s | train_loss=0.0225
264
+ step 2620000 | 24708s | train_loss=0.0237
265
+ step 2630000 | 24794s | train_loss=0.0223
266
+ step 2640000 | 24881s | train_loss=0.0219
267
+ step 2650000 | 24967s | train_loss=0.0231
268
+ step 2660000 | 25054s | train_loss=0.0227
269
+ step 2670000 | 25142s | train_loss=0.0227
270
+ step 2680000 | 25230s | train_loss=0.0215
271
+ step 2690000 | 25318s | train_loss=0.0231
272
+ step 2700000 | 25473s | train_loss=0.0227 | fid=55.9309
273
+ step 2710000 | 25561s | train_loss=0.0213
274
+ step 2720000 | 25649s | train_loss=0.0224
275
+ step 2730000 | 25737s | train_loss=0.0217
276
+ step 2740000 | 25823s | train_loss=0.0226
277
+ step 2750000 | 25910s | train_loss=0.0226
278
+ step 2760000 | 25997s | train_loss=0.0213
279
+ step 2770000 | 26083s | train_loss=0.0222
280
+ step 2780000 | 26171s | train_loss=0.0216
281
+ step 2790000 | 26259s | train_loss=0.0214
282
+ step 2800000 | 26414s | train_loss=0.0232 | fid=53.7460
283
+ step 2810000 | 26502s | train_loss=0.0217
284
+ step 2820000 | 26590s | train_loss=0.0223
285
+ step 2830000 | 26678s | train_loss=0.0218
286
+ step 2840000 | 26764s | train_loss=0.0217
287
+ step 2850000 | 26852s | train_loss=0.0225
288
+ step 2860000 | 26941s | train_loss=0.0219
289
+ step 2870000 | 27026s | train_loss=0.0207
290
+ step 2880000 | 27112s | train_loss=0.0216
291
+ step 2890000 | 27201s | train_loss=0.0209
292
+ step 2900000 | 27357s | train_loss=0.0212 | fid=53.1943
293
+ step 2910000 | 27445s | train_loss=0.0215
294
+ step 2920000 | 27533s | train_loss=0.0209
295
+ step 2930000 | 27620s | train_loss=0.0222
296
+ step 2940000 | 27709s | train_loss=0.0220
297
+ step 2950000 | 27795s | train_loss=0.0210
298
+ step 2960000 | 27883s | train_loss=0.0198
299
+ step 2970000 | 27972s | train_loss=0.0209
300
+ step 2980000 | 28056s | train_loss=0.0211
301
+ step 2990000 | 28143s | train_loss=0.0224
302
+ step 3000000 | 28299s | train_loss=0.0223 | fid=53.2871
303
+ step 3010000 | 28387s | train_loss=0.0220
304
+ step 3020000 | 28475s | train_loss=0.0196
305
+ step 3030000 | 28563s | train_loss=0.0198
306
+ step 3040000 | 28651s | train_loss=0.0213
307
+ step 3050000 | 28737s | train_loss=0.0213
308
+ step 3060000 | 28825s | train_loss=0.0214
309
+ step 3070000 | 28913s | train_loss=0.0208
310
+ step 3080000 | 29001s | train_loss=0.0205
311
+ step 3090000 | 29086s | train_loss=0.0210
312
+ step 3100000 | 29241s | train_loss=0.0202 | fid=50.1183
313
+ step 3110000 | 29329s | train_loss=0.0212
314
+ step 3120000 | 29417s | train_loss=0.0197
315
+ step 3130000 | 29505s | train_loss=0.0202
316
+ step 3140000 | 29593s | train_loss=0.0207
317
+ step 3150000 | 29681s | train_loss=0.0204
318
+ step 3160000 | 29767s | train_loss=0.0195
319
+ step 3170000 | 29855s | train_loss=0.0200
320
+ step 3180000 | 29943s | train_loss=0.0205
321
+ step 3190000 | 30032s | train_loss=0.0214
322
+ step 3200000 | 30183s | train_loss=0.0204 | fid=49.3341
323
+ step 3210000 | 30271s | train_loss=0.0217
324
+ step 3220000 | 30359s | train_loss=0.0200
325
+ step 3230000 | 30447s | train_loss=0.0207
326
+ step 3240000 | 30535s | train_loss=0.0196
327
+ step 3250000 | 30623s | train_loss=0.0209
328
+ step 3260000 | 30709s | train_loss=0.0206
329
+ step 3270000 | 30797s | train_loss=0.0199
330
+ step 3280000 | 30885s | train_loss=0.0193
331
+ step 3290000 | 30973s | train_loss=0.0189
332
+ step 3300000 | 31129s | train_loss=0.0192 | fid=48.2404
333
+ step 3310000 | 31214s | train_loss=0.0201
334
+ step 3320000 | 31302s | train_loss=0.0192
335
+ step 3330000 | 31390s | train_loss=0.0190
336
+ step 3340000 | 31478s | train_loss=0.0202
337
+ step 3350000 | 31566s | train_loss=0.0188
338
+ step 3360000 | 31654s | train_loss=0.0193
339
+ step 3370000 | 31741s | train_loss=0.0191
340
+ step 3380000 | 31829s | train_loss=0.0203
341
+ step 3390000 | 31917s | train_loss=0.0192
342
+ step 3400000 | 32073s | train_loss=0.0207 | fid=46.3206
343
+ step 3410000 | 32160s | train_loss=0.0186
344
+ step 3420000 | 32245s | train_loss=0.0193
345
+ step 3430000 | 32332s | train_loss=0.0193
346
+ step 3440000 | 32420s | train_loss=0.0200
347
+ step 3450000 | 32508s | train_loss=0.0192
348
+ step 3460000 | 32596s | train_loss=0.0195
349
+ step 3470000 | 32682s | train_loss=0.0190
350
+ step 3480000 | 32770s | train_loss=0.0193
351
+ step 3490000 | 32857s | train_loss=0.0185
352
+ step 3500000 | 33013s | train_loss=0.0185 | fid=45.4876
353
+ step 3510000 | 33102s | train_loss=0.0195
354
+ step 3520000 | 33189s | train_loss=0.0193
355
+ step 3530000 | 33275s | train_loss=0.0182
356
+ step 3540000 | 33361s | train_loss=0.0190
357
+ step 3550000 | 33449s | train_loss=0.0189
358
+ step 3560000 | 33537s | train_loss=0.0184
359
+ step 3570000 | 33625s | train_loss=0.0196
360
+ step 3580000 | 33711s | train_loss=0.0188
361
+ step 3590000 | 33799s | train_loss=0.0187
362
+ step 3600000 | 33955s | train_loss=0.0184 | fid=45.8988
363
+ step 3610000 | 34043s | train_loss=0.0197
364
+ step 3620000 | 34132s | train_loss=0.0184
365
+ step 3630000 | 34218s | train_loss=0.0179
366
+ step 3640000 | 34304s | train_loss=0.0194
367
+ step 3650000 | 34390s | train_loss=0.0182
368
+ step 3660000 | 34478s | train_loss=0.0195
369
+ step 3670000 | 34566s | train_loss=0.0191
370
+ step 3680000 | 34652s | train_loss=0.0178
371
+ step 3690000 | 34740s | train_loss=0.0183
372
+ step 3700000 | 34896s | train_loss=0.0201 | fid=47.8671
373
+ step 3710000 | 34984s | train_loss=0.0182
374
+ step 3720000 | 35072s | train_loss=0.0179
375
+ step 3730000 | 35160s | train_loss=0.0185
376
+ step 3740000 | 35247s | train_loss=0.0186
377
+ step 3750000 | 35332s | train_loss=0.0181
378
+ step 3760000 | 35418s | train_loss=0.0185
379
+ step 3770000 | 35506s | train_loss=0.0178
380
+ step 3780000 | 35594s | train_loss=0.0192
381
+ step 3790000 | 35680s | train_loss=0.0181
382
+ step 3800000 | 35835s | train_loss=0.0185 | fid=46.3106
383
+ step 3810000 | 35924s | train_loss=0.0191
384
+ step 3820000 | 36012s | train_loss=0.0186
385
+ step 3830000 | 36099s | train_loss=0.0192
386
+ step 3840000 | 36187s | train_loss=0.0184
387
+ step 3850000 | 36273s | train_loss=0.0176
388
+ step 3860000 | 36359s | train_loss=0.0177
389
+ step 3870000 | 36445s | train_loss=0.0181
390
+ step 3880000 | 36533s | train_loss=0.0185
391
+ step 3890000 | 36619s | train_loss=0.0191
392
+ step 3900000 | 36775s | train_loss=0.0174 | fid=46.0681
393
+ step 3910000 | 36863s | train_loss=0.0185
394
+ step 3920000 | 36951s | train_loss=0.0194
395
+ step 3930000 | 37039s | train_loss=0.0168
396
+ step 3940000 | 37127s | train_loss=0.0176
397
+ step 3950000 | 37215s | train_loss=0.0175
398
+ step 3960000 | 37301s | train_loss=0.0179
399
+ step 3970000 | 37386s | train_loss=0.0181
400
+ step 3980000 | 37473s | train_loss=0.0188
401
+ step 3990000 | 37561s | train_loss=0.0171
402
+ step 4000000 | 37714s | train_loss=0.0176 | fid=44.4851
403
+ step 4010000 | 37802s | train_loss=0.0178
404
+ step 4020000 | 37890s | train_loss=0.0178
405
+ step 4030000 | 37978s | train_loss=0.0173
406
+ step 4040000 | 38065s | train_loss=0.0175
407
+ step 4050000 | 38153s | train_loss=0.0178
408
+ step 4060000 | 38241s | train_loss=0.0168
409
+ step 4070000 | 38328s | train_loss=0.0183
410
+ step 4080000 | 38413s | train_loss=0.0166
411
+ step 4090000 | 38499s | train_loss=0.0185
412
+ step 4100000 | 38654s | train_loss=0.0173 | fid=43.8388
413
+ step 4110000 | 38742s | train_loss=0.0182
414
+ step 4120000 | 38830s | train_loss=0.0174
415
+ step 4130000 | 38918s | train_loss=0.0181
416
+ step 4140000 | 39005s | train_loss=0.0171
417
+ step 4150000 | 39093s | train_loss=0.0176
418
+ step 4160000 | 39181s | train_loss=0.0177
419
+ step 4170000 | 39269s | train_loss=0.0176
420
+ step 4180000 | 39357s | train_loss=0.0165
421
+ step 4190000 | 39441s | train_loss=0.0181
422
+ step 4200000 | 39593s | train_loss=0.0170 | fid=44.2248
423
+ step 4210000 | 39681s | train_loss=0.0189
424
+ step 4220000 | 39769s | train_loss=0.0170
425
+ step 4230000 | 39856s | train_loss=0.0190
426
+ step 4240000 | 39944s | train_loss=0.0175
427
+ step 4250000 | 40032s | train_loss=0.0167
428
+ step 4260000 | 40120s | train_loss=0.0176
429
+ step 4270000 | 40208s | train_loss=0.0169
430
+ step 4280000 | 40296s | train_loss=0.0177
431
+ step 4290000 | 40385s | train_loss=0.0169
432
+ step 4300000 | 40536s | train_loss=0.0174 | fid=43.2344
433
+ step 4310000 | 40623s | train_loss=0.0172
434
+ step 4320000 | 40711s | train_loss=0.0168
435
+ step 4330000 | 40798s | train_loss=0.0182
436
+ step 4340000 | 40886s | train_loss=0.0166
437
+ step 4350000 | 40974s | train_loss=0.0164
438
+ step 4360000 | 41062s | train_loss=0.0164
439
+ step 4370000 | 41150s | train_loss=0.0168
440
+ step 4380000 | 41238s | train_loss=0.0177
441
+ step 4390000 | 41326s | train_loss=0.0184
442
+ step 4400000 | 41480s | train_loss=0.0156 | fid=42.2085
443
+ step 4410000 | 41564s | train_loss=0.0166
444
+ step 4420000 | 41651s | train_loss=0.0176
445
+ step 4430000 | 41738s | train_loss=0.0179
446
+ step 4440000 | 41827s | train_loss=0.0167
447
+ step 4450000 | 41914s | train_loss=0.0158
448
+ step 4460000 | 42002s | train_loss=0.0162
449
+ step 4470000 | 42090s | train_loss=0.0168
450
+ step 4480000 | 42178s | train_loss=0.0170
451
+ step 4490000 | 42266s | train_loss=0.0162
452
+ step 4500000 | 42421s | train_loss=0.0157 | fid=43.9327
453
+ step 4510000 | 42508s | train_loss=0.0164
454
+ step 4520000 | 42592s | train_loss=0.0174
455
+ step 4530000 | 42678s | train_loss=0.0167
456
+ step 4540000 | 42766s | train_loss=0.0165
457
+ step 4550000 | 42854s | train_loss=0.0168
458
+ step 4560000 | 42942s | train_loss=0.0166
459
+ step 4570000 | 43029s | train_loss=0.0160
460
+ step 4580000 | 43118s | train_loss=0.0163
461
+ step 4590000 | 43206s | train_loss=0.0154
462
+ step 4600000 | 43361s | train_loss=0.0169 | fid=41.2007
463
+ step 4610000 | 43449s | train_loss=0.0167
464
+ step 4620000 | 43534s | train_loss=0.0168
465
+ step 4630000 | 43619s | train_loss=0.0157
466
+ step 4640000 | 43705s | train_loss=0.0173
467
+ step 4650000 | 43793s | train_loss=0.0172
468
+ step 4660000 | 43881s | train_loss=0.0172
469
+ step 4670000 | 43969s | train_loss=0.0155
470
+ step 4680000 | 44057s | train_loss=0.0160
471
+ step 4690000 | 44145s | train_loss=0.0178
472
+ step 4700000 | 44301s | train_loss=0.0172 | fid=41.5687
473
+ step 4710000 | 44389s | train_loss=0.0159
474
+ step 4720000 | 44477s | train_loss=0.0174
475
+ step 4730000 | 44563s | train_loss=0.0164
476
+ step 4740000 | 44649s | train_loss=0.0166
477
+ step 4750000 | 44735s | train_loss=0.0163
478
+ step 4760000 | 44823s | train_loss=0.0167
479
+ step 4770000 | 44911s | train_loss=0.0158
480
+ step 4780000 | 44999s | train_loss=0.0159
481
+ step 4790000 | 45087s | train_loss=0.0170
482
+ step 4800000 | 45244s | train_loss=0.0157 | fid=42.2849
483
+ step 4810000 | 45332s | train_loss=0.0156
484
+ step 4820000 | 45419s | train_loss=0.0163
485
+ step 4830000 | 45505s | train_loss=0.0154
486
+ step 4840000 | 45593s | train_loss=0.0167
487
+ step 4850000 | 45679s | train_loss=0.0153
488
+ step 4860000 | 45765s | train_loss=0.0160
489
+ step 4870000 | 45853s | train_loss=0.0159
490
+ step 4880000 | 45941s | train_loss=0.0155
491
+ step 4890000 | 46029s | train_loss=0.0154
492
+ step 4900000 | 46186s | train_loss=0.0159 | fid=41.7271
493
+ step 4910000 | 46274s | train_loss=0.0154
494
+ step 4920000 | 46362s | train_loss=0.0161
495
+ step 4930000 | 46450s | train_loss=0.0157
496
+ step 4940000 | 46536s | train_loss=0.0157
497
+ step 4950000 | 46623s | train_loss=0.0158
498
+ step 4960000 | 46709s | train_loss=0.0160
499
+ step 4970000 | 46796s | train_loss=0.0155
500
+ step 4980000 | 46884s | train_loss=0.0152
501
+ step 4990000 | 46972s | train_loss=0.0159
502
+ step 5000000 | 47127s | train_loss=0.0163 | fid=41.6628
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d140/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d140/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 69s | train_loss=0.9971 | fid=310.9130
3
+ step 10000 | 165s | train_loss=0.7554
4
+ step 20000 | 261s | train_loss=0.6284
5
+ step 30000 | 357s | train_loss=0.5155
6
+ step 40000 | 454s | train_loss=0.4168
7
+ step 50000 | 551s | train_loss=0.3339
8
+ step 60000 | 648s | train_loss=0.2690
9
+ step 70000 | 745s | train_loss=0.2206
10
+ step 80000 | 841s | train_loss=0.1907
11
+ step 90000 | 938s | train_loss=0.1625
12
+ step 100000 | 1106s | train_loss=0.1488 | fid=65.5886
13
+ step 110000 | 1201s | train_loss=0.1429
14
+ step 120000 | 1296s | train_loss=0.1276
15
+ step 130000 | 1393s | train_loss=0.1198
16
+ step 140000 | 1490s | train_loss=0.1166
17
+ step 150000 | 1587s | train_loss=0.1077
18
+ step 160000 | 1684s | train_loss=0.1037
19
+ step 170000 | 1781s | train_loss=0.1036
20
+ step 180000 | 1877s | train_loss=0.0939
21
+ step 190000 | 1971s | train_loss=0.0936
22
+ step 200000 | 2140s | train_loss=0.0912 | fid=64.3814
23
+ step 210000 | 2231s | train_loss=0.0856
24
+ step 220000 | 2327s | train_loss=0.0850
25
+ step 230000 | 2423s | train_loss=0.0841
26
+ step 240000 | 2520s | train_loss=0.0788
27
+ step 250000 | 2617s | train_loss=0.0817
28
+ step 260000 | 2713s | train_loss=0.0753
29
+ step 270000 | 2810s | train_loss=0.0719
30
+ step 280000 | 2904s | train_loss=0.0718
31
+ step 290000 | 3000s | train_loss=0.0732
32
+ step 300000 | 3172s | train_loss=0.0740 | fid=66.0594
33
+ step 310000 | 3263s | train_loss=0.0694
34
+ step 320000 | 3359s | train_loss=0.0701
35
+ step 330000 | 3456s | train_loss=0.0658
36
+ step 340000 | 3553s | train_loss=0.0653
37
+ step 350000 | 3649s | train_loss=0.0644
38
+ step 360000 | 3746s | train_loss=0.0650
39
+ step 370000 | 3841s | train_loss=0.0604
40
+ step 380000 | 3936s | train_loss=0.0615
41
+ step 390000 | 4031s | train_loss=0.0614
42
+ step 400000 | 4204s | train_loss=0.0589 | fid=67.0732
43
+ step 410000 | 4294s | train_loss=0.0598
44
+ step 420000 | 4390s | train_loss=0.0585
45
+ step 430000 | 4487s | train_loss=0.0580
46
+ step 440000 | 4584s | train_loss=0.0572
47
+ step 450000 | 4681s | train_loss=0.0568
48
+ step 460000 | 4776s | train_loss=0.0559
49
+ step 470000 | 4872s | train_loss=0.0582
50
+ step 480000 | 4968s | train_loss=0.0535
51
+ step 490000 | 5063s | train_loss=0.0536
52
+ step 500000 | 5235s | train_loss=0.0515 | fid=65.3153
53
+ step 510000 | 5327s | train_loss=0.0539
54
+ step 520000 | 5423s | train_loss=0.0532
55
+ step 530000 | 5520s | train_loss=0.0523
56
+ step 540000 | 5617s | train_loss=0.0528
57
+ step 550000 | 5711s | train_loss=0.0533
58
+ step 560000 | 5807s | train_loss=0.0476
59
+ step 570000 | 5903s | train_loss=0.0522
60
+ step 580000 | 5998s | train_loss=0.0523
61
+ step 590000 | 6095s | train_loss=0.0488
62
+ step 600000 | 6266s | train_loss=0.0502 | fid=63.9979
63
+ step 610000 | 6358s | train_loss=0.0481
64
+ step 620000 | 6455s | train_loss=0.0478
65
+ step 630000 | 6551s | train_loss=0.0476
66
+ step 640000 | 6646s | train_loss=0.0477
67
+ step 650000 | 6742s | train_loss=0.0481
68
+ step 660000 | 6839s | train_loss=0.0464
69
+ step 670000 | 6935s | train_loss=0.0456
70
+ step 680000 | 7029s | train_loss=0.0457
71
+ step 690000 | 7126s | train_loss=0.0458
72
+ step 700000 | 7299s | train_loss=0.0462 | fid=63.0765
73
+ step 710000 | 7391s | train_loss=0.0455
74
+ step 720000 | 7487s | train_loss=0.0458
75
+ step 730000 | 7582s | train_loss=0.0428
76
+ step 740000 | 7679s | train_loss=0.0455
77
+ step 750000 | 7775s | train_loss=0.0453
78
+ step 760000 | 7871s | train_loss=0.0436
79
+ step 770000 | 7966s | train_loss=0.0444
80
+ step 780000 | 8062s | train_loss=0.0430
81
+ step 790000 | 8159s | train_loss=0.0421
82
+ step 800000 | 8331s | train_loss=0.0423 | fid=61.6797
83
+ step 810000 | 8423s | train_loss=0.0431
84
+ step 820000 | 8518s | train_loss=0.0413
85
+ step 830000 | 8614s | train_loss=0.0432
86
+ step 840000 | 8711s | train_loss=0.0446
87
+ step 850000 | 8806s | train_loss=0.0405
88
+ step 860000 | 8903s | train_loss=0.0422
89
+ step 870000 | 8998s | train_loss=0.0432
90
+ step 880000 | 9095s | train_loss=0.0389
91
+ step 890000 | 9191s | train_loss=0.0409
92
+ step 900000 | 9364s | train_loss=0.0416 | fid=64.8150
93
+ step 910000 | 9454s | train_loss=0.0413
94
+ step 920000 | 9550s | train_loss=0.0388
95
+ step 930000 | 9646s | train_loss=0.0415
96
+ step 940000 | 9743s | train_loss=0.0419
97
+ step 950000 | 9839s | train_loss=0.0394
98
+ step 960000 | 9934s | train_loss=0.0396
99
+ step 970000 | 10031s | train_loss=0.0395
100
+ step 980000 | 10128s | train_loss=0.0400
101
+ step 990000 | 10225s | train_loss=0.0389
102
+ step 1000000 | 10396s | train_loss=0.0402 | fid=62.3086
103
+ step 1010000 | 10487s | train_loss=0.0365
104
+ step 1020000 | 10584s | train_loss=0.0362
105
+ step 1030000 | 10681s | train_loss=0.0376
106
+ step 1040000 | 10777s | train_loss=0.0359
107
+ step 1050000 | 10873s | train_loss=0.0371
108
+ step 1060000 | 10968s | train_loss=0.0375
109
+ step 1070000 | 11065s | train_loss=0.0394
110
+ step 1080000 | 11162s | train_loss=0.0392
111
+ step 1090000 | 11258s | train_loss=0.0363
112
+ step 1100000 | 11428s | train_loss=0.0373 | fid=58.2602
113
+ step 1110000 | 11520s | train_loss=0.0398
114
+ step 1120000 | 11617s | train_loss=0.0370
115
+ step 1130000 | 11714s | train_loss=0.0355
116
+ step 1140000 | 11810s | train_loss=0.0346
117
+ step 1150000 | 11906s | train_loss=0.0347
118
+ step 1160000 | 12002s | train_loss=0.0351
119
+ step 1170000 | 12099s | train_loss=0.0358
120
+ step 1180000 | 12196s | train_loss=0.0347
121
+ step 1190000 | 12291s | train_loss=0.0351
122
+ step 1200000 | 12461s | train_loss=0.0341 | fid=58.6994
123
+ step 1210000 | 12554s | train_loss=0.0337
124
+ step 1220000 | 12651s | train_loss=0.0342
125
+ step 1230000 | 12747s | train_loss=0.0344
126
+ step 1240000 | 12844s | train_loss=0.0343
127
+ step 1250000 | 12939s | train_loss=0.0368
128
+ step 1260000 | 13035s | train_loss=0.0342
129
+ step 1270000 | 13132s | train_loss=0.0332
130
+ step 1280000 | 13227s | train_loss=0.0324
131
+ step 1290000 | 13323s | train_loss=0.0335
132
+ step 1300000 | 13495s | train_loss=0.0314 | fid=62.0691
133
+ step 1310000 | 13587s | train_loss=0.0347
134
+ step 1320000 | 13684s | train_loss=0.0325
135
+ step 1330000 | 13781s | train_loss=0.0312
136
+ step 1340000 | 13877s | train_loss=0.0322
137
+ step 1350000 | 13973s | train_loss=0.0333
138
+ step 1360000 | 14070s | train_loss=0.0313
139
+ step 1370000 | 14165s | train_loss=0.0336
140
+ step 1380000 | 14262s | train_loss=0.0328
141
+ step 1390000 | 14358s | train_loss=0.0349
142
+ step 1400000 | 14530s | train_loss=0.0326 | fid=59.5210
143
+ step 1410000 | 14622s | train_loss=0.0311
144
+ step 1420000 | 14719s | train_loss=0.0318
145
+ step 1430000 | 14816s | train_loss=0.0325
146
+ step 1440000 | 14910s | train_loss=0.0325
147
+ step 1450000 | 15007s | train_loss=0.0323
148
+ step 1460000 | 15102s | train_loss=0.0312
149
+ step 1470000 | 15199s | train_loss=0.0312
150
+ step 1480000 | 15295s | train_loss=0.0327
151
+ step 1490000 | 15392s | train_loss=0.0297
152
+ step 1500000 | 15564s | train_loss=0.0316 | fid=64.8010
153
+ step 1510000 | 15657s | train_loss=0.0303
154
+ step 1520000 | 15754s | train_loss=0.0307
155
+ step 1530000 | 15851s | train_loss=0.0307
156
+ step 1540000 | 15946s | train_loss=0.0295
157
+ step 1550000 | 16041s | train_loss=0.0298
158
+ step 1560000 | 16138s | train_loss=0.0294
159
+ step 1570000 | 16234s | train_loss=0.0292
160
+ step 1580000 | 16331s | train_loss=0.0304
161
+ step 1590000 | 16428s | train_loss=0.0295
162
+ step 1600000 | 16599s | train_loss=0.0296 | fid=61.5597
163
+ step 1610000 | 16691s | train_loss=0.0298
164
+ step 1620000 | 16788s | train_loss=0.0277
165
+ step 1630000 | 16883s | train_loss=0.0295
166
+ step 1640000 | 16979s | train_loss=0.0282
167
+ step 1650000 | 17075s | train_loss=0.0281
168
+ step 1660000 | 17171s | train_loss=0.0282
169
+ step 1670000 | 17268s | train_loss=0.0285
170
+ step 1680000 | 17365s | train_loss=0.0271
171
+ step 1690000 | 17462s | train_loss=0.0293
172
+ step 1700000 | 17633s | train_loss=0.0284 | fid=63.5395
173
+ step 1710000 | 17726s | train_loss=0.0296
174
+ step 1720000 | 17822s | train_loss=0.0271
175
+ step 1730000 | 17917s | train_loss=0.0291
176
+ step 1740000 | 18012s | train_loss=0.0271
177
+ step 1750000 | 18108s | train_loss=0.0267
178
+ step 1760000 | 18205s | train_loss=0.0288
179
+ step 1770000 | 18302s | train_loss=0.0275
180
+ step 1780000 | 18399s | train_loss=0.0281
181
+ step 1790000 | 18495s | train_loss=0.0272
182
+ step 1800000 | 18667s | train_loss=0.0295 | fid=61.8386
183
+ step 1810000 | 18760s | train_loss=0.0275
184
+ step 1820000 | 18855s | train_loss=0.0253
185
+ step 1830000 | 18950s | train_loss=0.0263
186
+ step 1840000 | 19047s | train_loss=0.0282
187
+ step 1850000 | 19143s | train_loss=0.0253
188
+ step 1860000 | 19240s | train_loss=0.0270
189
+ step 1870000 | 19337s | train_loss=0.0266
190
+ step 1880000 | 19434s | train_loss=0.0272
191
+ step 1890000 | 19530s | train_loss=0.0295
192
+ step 1900000 | 19702s | train_loss=0.0255 | fid=60.2368
193
+ step 1910000 | 19795s | train_loss=0.0266
194
+ step 1920000 | 19887s | train_loss=0.0250
195
+ step 1930000 | 19984s | train_loss=0.0266
196
+ step 1940000 | 20080s | train_loss=0.0253
197
+ step 1950000 | 20177s | train_loss=0.0271
198
+ step 1960000 | 20273s | train_loss=0.0252
199
+ step 1970000 | 20370s | train_loss=0.0258
200
+ step 1980000 | 20467s | train_loss=0.0272
201
+ step 1990000 | 20563s | train_loss=0.0263
202
+ step 2000000 | 20734s | train_loss=0.0268 | fid=58.7355
203
+ step 2010000 | 20824s | train_loss=0.0270
204
+ step 2020000 | 20920s | train_loss=0.0262
205
+ step 2030000 | 21017s | train_loss=0.0260
206
+ step 2040000 | 21114s | train_loss=0.0231
207
+ step 2050000 | 21211s | train_loss=0.0247
208
+ step 2060000 | 21308s | train_loss=0.0256
209
+ step 2070000 | 21404s | train_loss=0.0242
210
+ step 2080000 | 21501s | train_loss=0.0249
211
+ step 2090000 | 21597s | train_loss=0.0246
212
+ step 2100000 | 21767s | train_loss=0.0245 | fid=55.6680
213
+ step 2110000 | 21857s | train_loss=0.0240
214
+ step 2120000 | 21953s | train_loss=0.0254
215
+ step 2130000 | 22050s | train_loss=0.0253
216
+ step 2140000 | 22147s | train_loss=0.0249
217
+ step 2150000 | 22243s | train_loss=0.0256
218
+ step 2160000 | 22340s | train_loss=0.0252
219
+ step 2170000 | 22437s | train_loss=0.0243
220
+ step 2180000 | 22534s | train_loss=0.0246
221
+ step 2190000 | 22631s | train_loss=0.0271
222
+ step 2200000 | 22798s | train_loss=0.0238 | fid=57.9561
223
+ step 2210000 | 22891s | train_loss=0.0239
224
+ step 2220000 | 22988s | train_loss=0.0228
225
+ step 2230000 | 23085s | train_loss=0.0237
226
+ step 2240000 | 23181s | train_loss=0.0236
227
+ step 2250000 | 23278s | train_loss=0.0232
228
+ step 2260000 | 23375s | train_loss=0.0242
229
+ step 2270000 | 23471s | train_loss=0.0245
230
+ step 2280000 | 23568s | train_loss=0.0236
231
+ step 2290000 | 23663s | train_loss=0.0231
232
+ step 2300000 | 23832s | train_loss=0.0228 | fid=56.2438
233
+ step 2310000 | 23926s | train_loss=0.0250
234
+ step 2320000 | 24021s | train_loss=0.0229
235
+ step 2330000 | 24118s | train_loss=0.0242
236
+ step 2340000 | 24215s | train_loss=0.0227
237
+ step 2350000 | 24312s | train_loss=0.0220
238
+ step 2360000 | 24409s | train_loss=0.0250
239
+ step 2370000 | 24506s | train_loss=0.0237
240
+ step 2380000 | 24601s | train_loss=0.0247
241
+ step 2390000 | 24698s | train_loss=0.0225
242
+ step 2400000 | 24867s | train_loss=0.0219 | fid=56.9743
243
+ step 2410000 | 24962s | train_loss=0.0244
244
+ step 2420000 | 25057s | train_loss=0.0226
245
+ step 2430000 | 25154s | train_loss=0.0231
246
+ step 2440000 | 25251s | train_loss=0.0231
247
+ step 2450000 | 25347s | train_loss=0.0228
248
+ step 2460000 | 25444s | train_loss=0.0231
249
+ step 2470000 | 25539s | train_loss=0.0232
250
+ step 2480000 | 25636s | train_loss=0.0253
251
+ step 2490000 | 25732s | train_loss=0.0223
252
+ step 2500000 | 25903s | train_loss=0.0224 | fid=53.9406
253
+ step 2510000 | 25998s | train_loss=0.0227
254
+ step 2520000 | 26093s | train_loss=0.0219
255
+ step 2530000 | 26190s | train_loss=0.0219
256
+ step 2540000 | 26287s | train_loss=0.0220
257
+ step 2550000 | 26384s | train_loss=0.0223
258
+ step 2560000 | 26479s | train_loss=0.0219
259
+ step 2570000 | 26576s | train_loss=0.0211
260
+ step 2580000 | 26672s | train_loss=0.0217
261
+ step 2590000 | 26767s | train_loss=0.0223
262
+ step 2600000 | 26939s | train_loss=0.0218 | fid=52.9492
263
+ step 2610000 | 27034s | train_loss=0.0215
264
+ step 2620000 | 27129s | train_loss=0.0210
265
+ step 2630000 | 27226s | train_loss=0.0209
266
+ step 2640000 | 27323s | train_loss=0.0215
267
+ step 2650000 | 27418s | train_loss=0.0215
268
+ step 2660000 | 27515s | train_loss=0.0217
269
+ step 2670000 | 27611s | train_loss=0.0207
270
+ step 2680000 | 27708s | train_loss=0.0198
271
+ step 2690000 | 27803s | train_loss=0.0220
272
+ step 2700000 | 27975s | train_loss=0.0206 | fid=53.1374
273
+ step 2710000 | 28070s | train_loss=0.0199
274
+ step 2720000 | 28165s | train_loss=0.0203
275
+ step 2730000 | 28262s | train_loss=0.0200
276
+ step 2740000 | 28357s | train_loss=0.0212
277
+ step 2750000 | 28454s | train_loss=0.0214
278
+ step 2760000 | 28550s | train_loss=0.0219
279
+ step 2770000 | 28647s | train_loss=0.0199
280
+ step 2780000 | 28741s | train_loss=0.0201
281
+ step 2790000 | 28838s | train_loss=0.0203
282
+ step 2800000 | 29009s | train_loss=0.0219 | fid=54.2534
283
+ step 2810000 | 29104s | train_loss=0.0198
284
+ step 2820000 | 29199s | train_loss=0.0196
285
+ step 2830000 | 29294s | train_loss=0.0205
286
+ step 2840000 | 29390s | train_loss=0.0198
287
+ step 2850000 | 29487s | train_loss=0.0222
288
+ step 2860000 | 29584s | train_loss=0.0197
289
+ step 2870000 | 29680s | train_loss=0.0194
290
+ step 2880000 | 29775s | train_loss=0.0203
291
+ step 2890000 | 29872s | train_loss=0.0203
292
+ step 2900000 | 30044s | train_loss=0.0208 | fid=53.1247
293
+ step 2910000 | 30139s | train_loss=0.0209
294
+ step 2920000 | 30232s | train_loss=0.0200
295
+ step 2930000 | 30329s | train_loss=0.0205
296
+ step 2940000 | 30425s | train_loss=0.0202
297
+ step 2950000 | 30522s | train_loss=0.0189
298
+ step 2960000 | 30618s | train_loss=0.0183
299
+ step 2970000 | 30713s | train_loss=0.0189
300
+ step 2980000 | 30810s | train_loss=0.0204
301
+ step 2990000 | 30907s | train_loss=0.0204
302
+ step 3000000 | 31079s | train_loss=0.0198 | fid=50.2843
303
+ step 3010000 | 31170s | train_loss=0.0210
304
+ step 3020000 | 31266s | train_loss=0.0190
305
+ step 3030000 | 31362s | train_loss=0.0186
306
+ step 3040000 | 31459s | train_loss=0.0205
307
+ step 3050000 | 31556s | train_loss=0.0208
308
+ step 3060000 | 31652s | train_loss=0.0189
309
+ step 3070000 | 31748s | train_loss=0.0188
310
+ step 3080000 | 31844s | train_loss=0.0192
311
+ step 3090000 | 31941s | train_loss=0.0199
312
+ step 3100000 | 32112s | train_loss=0.0195 | fid=51.1955
313
+ step 3110000 | 32207s | train_loss=0.0204
314
+ step 3120000 | 32302s | train_loss=0.0178
315
+ step 3130000 | 32399s | train_loss=0.0190
316
+ step 3140000 | 32495s | train_loss=0.0201
317
+ step 3150000 | 32592s | train_loss=0.0193
318
+ step 3160000 | 32686s | train_loss=0.0182
319
+ step 3170000 | 32783s | train_loss=0.0182
320
+ step 3180000 | 32880s | train_loss=0.0208
321
+ step 3190000 | 32976s | train_loss=0.0193
322
+ step 3200000 | 33146s | train_loss=0.0190 | fid=47.8470
323
+ step 3210000 | 33241s | train_loss=0.0200
324
+ step 3220000 | 33336s | train_loss=0.0192
325
+ step 3230000 | 33432s | train_loss=0.0197
326
+ step 3240000 | 33529s | train_loss=0.0188
327
+ step 3250000 | 33625s | train_loss=0.0179
328
+ step 3260000 | 33720s | train_loss=0.0196
329
+ step 3270000 | 33817s | train_loss=0.0179
330
+ step 3280000 | 33914s | train_loss=0.0183
331
+ step 3290000 | 34009s | train_loss=0.0186
332
+ step 3300000 | 34182s | train_loss=0.0189 | fid=47.0677
333
+ step 3310000 | 34276s | train_loss=0.0180
334
+ step 3320000 | 34371s | train_loss=0.0181
335
+ step 3330000 | 34468s | train_loss=0.0187
336
+ step 3340000 | 34565s | train_loss=0.0186
337
+ step 3350000 | 34660s | train_loss=0.0181
338
+ step 3360000 | 34756s | train_loss=0.0179
339
+ step 3370000 | 34853s | train_loss=0.0176
340
+ step 3380000 | 34948s | train_loss=0.0183
341
+ step 3390000 | 35045s | train_loss=0.0188
342
+ step 3400000 | 35218s | train_loss=0.0181 | fid=46.1300
343
+ step 3410000 | 35312s | train_loss=0.0172
344
+ step 3420000 | 35406s | train_loss=0.0173
345
+ step 3430000 | 35503s | train_loss=0.0189
346
+ step 3440000 | 35599s | train_loss=0.0178
347
+ step 3450000 | 35694s | train_loss=0.0183
348
+ step 3460000 | 35790s | train_loss=0.0175
349
+ step 3470000 | 35886s | train_loss=0.0184
350
+ step 3480000 | 35982s | train_loss=0.0173
351
+ step 3490000 | 36079s | train_loss=0.0168
352
+ step 3500000 | 36253s | train_loss=0.0180 | fid=47.0167
353
+ step 3510000 | 36346s | train_loss=0.0203
354
+ step 3520000 | 36441s | train_loss=0.0179
355
+ step 3530000 | 36538s | train_loss=0.0181
356
+ step 3540000 | 36633s | train_loss=0.0183
357
+ step 3550000 | 36729s | train_loss=0.0176
358
+ step 3560000 | 36824s | train_loss=0.0177
359
+ step 3570000 | 36921s | train_loss=0.0181
360
+ step 3580000 | 37018s | train_loss=0.0172
361
+ step 3590000 | 37114s | train_loss=0.0177
362
+ step 3600000 | 37288s | train_loss=0.0173 | fid=42.7347
363
+ step 3610000 | 37382s | train_loss=0.0191
364
+ step 3620000 | 37478s | train_loss=0.0164
365
+ step 3630000 | 37574s | train_loss=0.0171
366
+ step 3640000 | 37669s | train_loss=0.0184
367
+ step 3650000 | 37763s | train_loss=0.0182
368
+ step 3660000 | 37860s | train_loss=0.0180
369
+ step 3670000 | 37957s | train_loss=0.0169
370
+ step 3680000 | 38052s | train_loss=0.0169
371
+ step 3690000 | 38149s | train_loss=0.0176
372
+ step 3700000 | 38322s | train_loss=0.0183 | fid=45.6584
373
+ step 3710000 | 38415s | train_loss=0.0170
374
+ step 3720000 | 38510s | train_loss=0.0175
375
+ step 3730000 | 38605s | train_loss=0.0171
376
+ step 3740000 | 38700s | train_loss=0.0182
377
+ step 3750000 | 38796s | train_loss=0.0178
378
+ step 3760000 | 38893s | train_loss=0.0169
379
+ step 3770000 | 38989s | train_loss=0.0166
380
+ step 3780000 | 39086s | train_loss=0.0170
381
+ step 3790000 | 39182s | train_loss=0.0184
382
+ step 3800000 | 39356s | train_loss=0.0181 | fid=44.3522
383
+ step 3810000 | 39449s | train_loss=0.0186
384
+ step 3820000 | 39544s | train_loss=0.0176
385
+ step 3830000 | 39637s | train_loss=0.0179
386
+ step 3840000 | 39734s | train_loss=0.0178
387
+ step 3850000 | 39830s | train_loss=0.0161
388
+ step 3860000 | 39927s | train_loss=0.0171
389
+ step 3870000 | 40024s | train_loss=0.0188
390
+ step 3880000 | 40120s | train_loss=0.0166
391
+ step 3890000 | 40216s | train_loss=0.0175
392
+ step 3900000 | 40390s | train_loss=0.0173 | fid=43.8234
393
+ step 3910000 | 40483s | train_loss=0.0172
394
+ step 3920000 | 40574s | train_loss=0.0169
395
+ step 3930000 | 40671s | train_loss=0.0157
396
+ step 3940000 | 40767s | train_loss=0.0174
397
+ step 3950000 | 40863s | train_loss=0.0174
398
+ step 3960000 | 40960s | train_loss=0.0182
399
+ step 3970000 | 41057s | train_loss=0.0171
400
+ step 3980000 | 41154s | train_loss=0.0187
401
+ step 3990000 | 41250s | train_loss=0.0166
402
+ step 4000000 | 41426s | train_loss=0.0181 | fid=42.8758
403
+ step 4010000 | 41517s | train_loss=0.0167
404
+ step 4020000 | 41611s | train_loss=0.0165
405
+ step 4030000 | 41707s | train_loss=0.0165
406
+ step 4040000 | 41803s | train_loss=0.0162
407
+ step 4050000 | 41900s | train_loss=0.0158
408
+ step 4060000 | 41997s | train_loss=0.0162
409
+ step 4070000 | 42093s | train_loss=0.0168
410
+ step 4080000 | 42190s | train_loss=0.0167
411
+ step 4090000 | 42287s | train_loss=0.0172
412
+ step 4100000 | 42460s | train_loss=0.0179 | fid=41.4978
413
+ step 4110000 | 42552s | train_loss=0.0167
414
+ step 4120000 | 42647s | train_loss=0.0158
415
+ step 4130000 | 42744s | train_loss=0.0172
416
+ step 4140000 | 42841s | train_loss=0.0155
417
+ step 4150000 | 42938s | train_loss=0.0164
418
+ step 4160000 | 43034s | train_loss=0.0165
419
+ step 4170000 | 43131s | train_loss=0.0173
420
+ step 4180000 | 43228s | train_loss=0.0163
421
+ step 4190000 | 43324s | train_loss=0.0160
422
+ step 4200000 | 43496s | train_loss=0.0158 | fid=41.4712
423
+ step 4210000 | 43588s | train_loss=0.0177
424
+ step 4220000 | 43683s | train_loss=0.0167
425
+ step 4230000 | 43780s | train_loss=0.0171
426
+ step 4240000 | 43877s | train_loss=0.0171
427
+ step 4250000 | 43973s | train_loss=0.0162
428
+ step 4260000 | 44070s | train_loss=0.0166
429
+ step 4270000 | 44167s | train_loss=0.0155
430
+ step 4280000 | 44263s | train_loss=0.0163
431
+ step 4290000 | 44358s | train_loss=0.0165
432
+ step 4300000 | 44527s | train_loss=0.0158 | fid=43.4034
433
+ step 4310000 | 44622s | train_loss=0.0157
434
+ step 4320000 | 44718s | train_loss=0.0152
435
+ step 4330000 | 44814s | train_loss=0.0175
436
+ step 4340000 | 44911s | train_loss=0.0163
437
+ step 4350000 | 45008s | train_loss=0.0152
438
+ step 4360000 | 45104s | train_loss=0.0158
439
+ step 4370000 | 45201s | train_loss=0.0158
440
+ step 4380000 | 45296s | train_loss=0.0171
441
+ step 4390000 | 45393s | train_loss=0.0180
442
+ step 4400000 | 45560s | train_loss=0.0152 | fid=42.0224
443
+ step 4410000 | 45655s | train_loss=0.0157
444
+ step 4420000 | 45750s | train_loss=0.0166
445
+ step 4430000 | 45847s | train_loss=0.0162
446
+ step 4440000 | 45944s | train_loss=0.0157
447
+ step 4450000 | 46041s | train_loss=0.0156
448
+ step 4460000 | 46137s | train_loss=0.0151
449
+ step 4470000 | 46232s | train_loss=0.0152
450
+ step 4480000 | 46329s | train_loss=0.0166
451
+ step 4490000 | 46426s | train_loss=0.0150
452
+ step 4500000 | 46595s | train_loss=0.0153 | fid=41.5596
453
+ step 4510000 | 46690s | train_loss=0.0153
454
+ step 4520000 | 46785s | train_loss=0.0181
455
+ step 4530000 | 46881s | train_loss=0.0156
456
+ step 4540000 | 46978s | train_loss=0.0156
457
+ step 4550000 | 47075s | train_loss=0.0160
458
+ step 4560000 | 47161s | train_loss=0.0153
459
+ step 4570000 | 47239s | train_loss=0.0162
460
+ step 4580000 | 47317s | train_loss=0.0168
461
+ step 4590000 | 47395s | train_loss=0.0153
462
+ step 4600000 | 47532s | train_loss=0.0141 | fid=42.7286
463
+ step 4610000 | 47608s | train_loss=0.0159
464
+ step 4620000 | 47684s | train_loss=0.0153
465
+ step 4630000 | 47762s | train_loss=0.0149
466
+ step 4640000 | 47840s | train_loss=0.0166
467
+ step 4650000 | 47918s | train_loss=0.0164
468
+ step 4660000 | 47996s | train_loss=0.0177
469
+ step 4670000 | 48075s | train_loss=0.0148
470
+ step 4680000 | 48152s | train_loss=0.0152
471
+ step 4690000 | 48228s | train_loss=0.0175
472
+ step 4700000 | 48368s | train_loss=0.0167 | fid=40.2410
473
+ step 4710000 | 48443s | train_loss=0.0148
474
+ step 4720000 | 48519s | train_loss=0.0162
475
+ step 4730000 | 48596s | train_loss=0.0153
476
+ step 4740000 | 48674s | train_loss=0.0164
477
+ step 4750000 | 48752s | train_loss=0.0144
478
+ step 4760000 | 48831s | train_loss=0.0152
479
+ step 4770000 | 48909s | train_loss=0.0145
480
+ step 4780000 | 48986s | train_loss=0.0152
481
+ step 4790000 | 49049s | train_loss=0.0156
482
+ step 4800000 | 49156s | train_loss=0.0153 | fid=39.1865
483
+ step 4810000 | 49210s | train_loss=0.0146
484
+ step 4820000 | 49266s | train_loss=0.0163
485
+ step 4830000 | 49326s | train_loss=0.0144
486
+ step 4840000 | 49386s | train_loss=0.0167
487
+ step 4850000 | 49445s | train_loss=0.0150
488
+ step 4860000 | 49505s | train_loss=0.0148
489
+ step 4870000 | 49565s | train_loss=0.0146
490
+ step 4880000 | 49624s | train_loss=0.0153
491
+ step 4890000 | 49684s | train_loss=0.0148
492
+ step 4900000 | 49793s | train_loss=0.0152 | fid=39.4070
493
+ step 4910000 | 49848s | train_loss=0.0154
494
+ step 4920000 | 49904s | train_loss=0.0155
495
+ step 4930000 | 49964s | train_loss=0.0150
496
+ step 4940000 | 50024s | train_loss=0.0149
497
+ step 4950000 | 50083s | train_loss=0.0154
498
+ step 4960000 | 50142s | train_loss=0.0156
499
+ step 4970000 | 50200s | train_loss=0.0143
500
+ step 4980000 | 50260s | train_loss=0.0147
501
+ step 4990000 | 50320s | train_loss=0.0146
502
+ step 5000000 | 50429s | train_loss=0.0154 | fid=40.5859
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d160/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d160/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 72s | train_loss=0.9963 | fid=291.0109
3
+ step 10000 | 167s | train_loss=0.7181
4
+ step 20000 | 264s | train_loss=0.6080
5
+ step 30000 | 360s | train_loss=0.5233
6
+ step 40000 | 458s | train_loss=0.4460
7
+ step 50000 | 555s | train_loss=0.3738
8
+ step 60000 | 652s | train_loss=0.3135
9
+ step 70000 | 750s | train_loss=0.2578
10
+ step 80000 | 847s | train_loss=0.2229
11
+ step 90000 | 944s | train_loss=0.1903
12
+ step 100000 | 1113s | train_loss=0.1688 | fid=62.2098
13
+ step 110000 | 1210s | train_loss=0.1580
14
+ step 120000 | 1306s | train_loss=0.1430
15
+ step 130000 | 1404s | train_loss=0.1320
16
+ step 140000 | 1501s | train_loss=0.1223
17
+ step 150000 | 1598s | train_loss=0.1215
18
+ step 160000 | 1696s | train_loss=0.1112
19
+ step 170000 | 1793s | train_loss=0.1113
20
+ step 180000 | 1890s | train_loss=0.0978
21
+ step 190000 | 1984s | train_loss=0.0973
22
+ step 200000 | 2153s | train_loss=0.1024 | fid=79.9509
23
+ step 210000 | 2250s | train_loss=0.0894
24
+ step 220000 | 2346s | train_loss=0.0865
25
+ step 230000 | 2443s | train_loss=0.0880
26
+ step 240000 | 2541s | train_loss=0.0830
27
+ step 250000 | 2637s | train_loss=0.0808
28
+ step 260000 | 2735s | train_loss=0.0777
29
+ step 270000 | 2832s | train_loss=0.0749
30
+ step 280000 | 2926s | train_loss=0.0744
31
+ step 290000 | 3022s | train_loss=0.0761
32
+ step 300000 | 3190s | train_loss=0.0765 | fid=85.6549
33
+ step 310000 | 3287s | train_loss=0.0709
34
+ step 320000 | 3383s | train_loss=0.0780
35
+ step 330000 | 3480s | train_loss=0.0673
36
+ step 340000 | 3577s | train_loss=0.0651
37
+ step 350000 | 3675s | train_loss=0.0675
38
+ step 360000 | 3772s | train_loss=0.0657
39
+ step 370000 | 3866s | train_loss=0.0612
40
+ step 380000 | 3963s | train_loss=0.0635
41
+ step 390000 | 4058s | train_loss=0.0645
42
+ step 400000 | 4228s | train_loss=0.0574 | fid=92.5522
43
+ step 410000 | 4324s | train_loss=0.0631
44
+ step 420000 | 4421s | train_loss=0.0581
45
+ step 430000 | 4518s | train_loss=0.0600
46
+ step 440000 | 4616s | train_loss=0.0569
47
+ step 450000 | 4713s | train_loss=0.0574
48
+ step 460000 | 4807s | train_loss=0.0560
49
+ step 470000 | 4904s | train_loss=0.0647
50
+ step 480000 | 5000s | train_loss=0.0552
51
+ step 490000 | 5097s | train_loss=0.0559
52
+ step 500000 | 5264s | train_loss=0.0527 | fid=93.2812
53
+ step 510000 | 5360s | train_loss=0.0537
54
+ step 520000 | 5458s | train_loss=0.0541
55
+ step 530000 | 5555s | train_loss=0.0521
56
+ step 540000 | 5652s | train_loss=0.0553
57
+ step 550000 | 5747s | train_loss=0.0514
58
+ step 560000 | 5844s | train_loss=0.0501
59
+ step 570000 | 5940s | train_loss=0.0524
60
+ step 580000 | 6036s | train_loss=0.0528
61
+ step 590000 | 6133s | train_loss=0.0530
62
+ step 600000 | 6300s | train_loss=0.0492 | fid=89.6598
63
+ step 610000 | 6397s | train_loss=0.0478
64
+ step 620000 | 6494s | train_loss=0.0463
65
+ step 630000 | 6591s | train_loss=0.0474
66
+ step 640000 | 6686s | train_loss=0.0487
67
+ step 650000 | 6783s | train_loss=0.0490
68
+ step 660000 | 6880s | train_loss=0.0455
69
+ step 670000 | 6975s | train_loss=0.0472
70
+ step 680000 | 7072s | train_loss=0.0469
71
+ step 690000 | 7169s | train_loss=0.0442
72
+ step 700000 | 7336s | train_loss=0.0444 | fid=79.1466
73
+ step 710000 | 7433s | train_loss=0.0439
74
+ step 720000 | 7530s | train_loss=0.0435
75
+ step 730000 | 7626s | train_loss=0.0418
76
+ step 740000 | 7722s | train_loss=0.0458
77
+ step 750000 | 7819s | train_loss=0.0430
78
+ step 760000 | 7915s | train_loss=0.0438
79
+ step 770000 | 8011s | train_loss=0.0424
80
+ step 780000 | 8108s | train_loss=0.0451
81
+ step 790000 | 8205s | train_loss=0.0424
82
+ step 800000 | 8373s | train_loss=0.0413 | fid=86.1900
83
+ step 810000 | 8469s | train_loss=0.0465
84
+ step 820000 | 8565s | train_loss=0.0410
85
+ step 830000 | 8662s | train_loss=0.0410
86
+ step 840000 | 8759s | train_loss=0.0431
87
+ step 850000 | 8855s | train_loss=0.0408
88
+ step 860000 | 8951s | train_loss=0.0410
89
+ step 870000 | 9048s | train_loss=0.0435
90
+ step 880000 | 9145s | train_loss=0.0390
91
+ step 890000 | 9242s | train_loss=0.0457
92
+ step 900000 | 9409s | train_loss=0.0402 | fid=81.5234
93
+ step 910000 | 9504s | train_loss=0.0394
94
+ step 920000 | 9601s | train_loss=0.0396
95
+ step 930000 | 9699s | train_loss=0.0388
96
+ step 940000 | 9795s | train_loss=0.0419
97
+ step 950000 | 9893s | train_loss=0.0394
98
+ step 960000 | 9988s | train_loss=0.0396
99
+ step 970000 | 10085s | train_loss=0.0374
100
+ step 980000 | 10182s | train_loss=0.0377
101
+ step 990000 | 10279s | train_loss=0.0372
102
+ step 1000000 | 10445s | train_loss=0.0386 | fid=74.6000
103
+ step 1010000 | 10542s | train_loss=0.0369
104
+ step 1020000 | 10639s | train_loss=0.0373
105
+ step 1030000 | 10736s | train_loss=0.0362
106
+ step 1040000 | 10832s | train_loss=0.0353
107
+ step 1050000 | 10927s | train_loss=0.0383
108
+ step 1060000 | 11024s | train_loss=0.0367
109
+ step 1070000 | 11121s | train_loss=0.0385
110
+ step 1080000 | 11218s | train_loss=0.0378
111
+ step 1090000 | 11315s | train_loss=0.0397
112
+ step 1100000 | 11481s | train_loss=0.0349 | fid=72.8011
113
+ step 1110000 | 11578s | train_loss=0.0369
114
+ step 1120000 | 11676s | train_loss=0.0356
115
+ step 1130000 | 11772s | train_loss=0.0365
116
+ step 1140000 | 11870s | train_loss=0.0333
117
+ step 1150000 | 11965s | train_loss=0.0354
118
+ step 1160000 | 12062s | train_loss=0.0359
119
+ step 1170000 | 12159s | train_loss=0.0378
120
+ step 1180000 | 12256s | train_loss=0.0348
121
+ step 1190000 | 12352s | train_loss=0.0346
122
+ step 1200000 | 12520s | train_loss=0.0339 | fid=70.6478
123
+ step 1210000 | 12617s | train_loss=0.0327
124
+ step 1220000 | 12714s | train_loss=0.0331
125
+ step 1230000 | 12811s | train_loss=0.0332
126
+ step 1240000 | 12906s | train_loss=0.0332
127
+ step 1250000 | 13003s | train_loss=0.0380
128
+ step 1260000 | 13101s | train_loss=0.0342
129
+ step 1270000 | 13198s | train_loss=0.0328
130
+ step 1280000 | 13293s | train_loss=0.0335
131
+ step 1290000 | 13390s | train_loss=0.0363
132
+ step 1300000 | 13558s | train_loss=0.0324 | fid=63.4103
133
+ step 1310000 | 13656s | train_loss=0.0344
134
+ step 1320000 | 13752s | train_loss=0.0313
135
+ step 1330000 | 13849s | train_loss=0.0309
136
+ step 1340000 | 13945s | train_loss=0.0319
137
+ step 1350000 | 14042s | train_loss=0.0332
138
+ step 1360000 | 14139s | train_loss=0.0302
139
+ step 1370000 | 14235s | train_loss=0.0334
140
+ step 1380000 | 14332s | train_loss=0.0310
141
+ step 1390000 | 14429s | train_loss=0.0334
142
+ step 1400000 | 14596s | train_loss=0.0312 | fid=65.7176
143
+ step 1410000 | 14693s | train_loss=0.0323
144
+ step 1420000 | 14791s | train_loss=0.0301
145
+ step 1430000 | 14886s | train_loss=0.0301
146
+ step 1440000 | 14983s | train_loss=0.0306
147
+ step 1450000 | 15081s | train_loss=0.0332
148
+ step 1460000 | 15176s | train_loss=0.0306
149
+ step 1470000 | 15273s | train_loss=0.0325
150
+ step 1480000 | 15370s | train_loss=0.0315
151
+ step 1490000 | 15467s | train_loss=0.0298
152
+ step 1500000 | 15635s | train_loss=0.0316 | fid=65.7271
153
+ step 1510000 | 15732s | train_loss=0.0299
154
+ step 1520000 | 15829s | train_loss=0.0316
155
+ step 1530000 | 15925s | train_loss=0.0312
156
+ step 1540000 | 16022s | train_loss=0.0295
157
+ step 1550000 | 16117s | train_loss=0.0289
158
+ step 1560000 | 16214s | train_loss=0.0299
159
+ step 1570000 | 16312s | train_loss=0.0287
160
+ step 1580000 | 16409s | train_loss=0.0307
161
+ step 1590000 | 16506s | train_loss=0.0303
162
+ step 1600000 | 16674s | train_loss=0.0323 | fid=63.5859
163
+ step 1610000 | 16771s | train_loss=0.0296
164
+ step 1620000 | 16866s | train_loss=0.0288
165
+ step 1630000 | 16964s | train_loss=0.0297
166
+ step 1640000 | 17060s | train_loss=0.0297
167
+ step 1650000 | 17156s | train_loss=0.0294
168
+ step 1660000 | 17254s | train_loss=0.0277
169
+ step 1670000 | 17350s | train_loss=0.0285
170
+ step 1680000 | 17448s | train_loss=0.0266
171
+ step 1690000 | 17544s | train_loss=0.0294
172
+ step 1700000 | 17713s | train_loss=0.0275 | fid=64.5899
173
+ step 1710000 | 17810s | train_loss=0.0283
174
+ step 1720000 | 17905s | train_loss=0.0283
175
+ step 1730000 | 18000s | train_loss=0.0289
176
+ step 1740000 | 18097s | train_loss=0.0274
177
+ step 1750000 | 18194s | train_loss=0.0264
178
+ step 1760000 | 18291s | train_loss=0.0295
179
+ step 1770000 | 18388s | train_loss=0.0264
180
+ step 1780000 | 18485s | train_loss=0.0274
181
+ step 1790000 | 18582s | train_loss=0.0280
182
+ step 1800000 | 18749s | train_loss=0.0280 | fid=69.6064
183
+ step 1810000 | 18844s | train_loss=0.0283
184
+ step 1820000 | 18940s | train_loss=0.0270
185
+ step 1830000 | 19037s | train_loss=0.0270
186
+ step 1840000 | 19134s | train_loss=0.0277
187
+ step 1850000 | 19231s | train_loss=0.0263
188
+ step 1860000 | 19328s | train_loss=0.0266
189
+ step 1870000 | 19426s | train_loss=0.0279
190
+ step 1880000 | 19523s | train_loss=0.0267
191
+ step 1890000 | 19620s | train_loss=0.0300
192
+ step 1900000 | 19789s | train_loss=0.0249 | fid=61.9937
193
+ step 1910000 | 19882s | train_loss=0.0257
194
+ step 1920000 | 19978s | train_loss=0.0272
195
+ step 1930000 | 20075s | train_loss=0.0254
196
+ step 1940000 | 20173s | train_loss=0.0250
197
+ step 1950000 | 20270s | train_loss=0.0270
198
+ step 1960000 | 20367s | train_loss=0.0257
199
+ step 1970000 | 20464s | train_loss=0.0286
200
+ step 1980000 | 20561s | train_loss=0.0273
201
+ step 1990000 | 20658s | train_loss=0.0266
202
+ step 2000000 | 20822s | train_loss=0.0258 | fid=62.5504
203
+ step 2010000 | 20919s | train_loss=0.0275
204
+ step 2020000 | 21016s | train_loss=0.0281
205
+ step 2030000 | 21113s | train_loss=0.0255
206
+ step 2040000 | 21211s | train_loss=0.0231
207
+ step 2050000 | 21308s | train_loss=0.0268
208
+ step 2060000 | 21405s | train_loss=0.0270
209
+ step 2070000 | 21502s | train_loss=0.0249
210
+ step 2080000 | 21599s | train_loss=0.0253
211
+ step 2090000 | 21697s | train_loss=0.0247
212
+ step 2100000 | 21861s | train_loss=0.0250 | fid=64.8525
213
+ step 2110000 | 21958s | train_loss=0.0258
214
+ step 2120000 | 22055s | train_loss=0.0263
215
+ step 2130000 | 22152s | train_loss=0.0252
216
+ step 2140000 | 22250s | train_loss=0.0259
217
+ step 2150000 | 22347s | train_loss=0.0266
218
+ step 2160000 | 22444s | train_loss=0.0254
219
+ step 2170000 | 22541s | train_loss=0.0248
220
+ step 2180000 | 22639s | train_loss=0.0256
221
+ step 2190000 | 22734s | train_loss=0.0245
222
+ step 2200000 | 22899s | train_loss=0.0241 | fid=61.0309
223
+ step 2210000 | 22996s | train_loss=0.0242
224
+ step 2220000 | 23093s | train_loss=0.0220
225
+ step 2230000 | 23191s | train_loss=0.0277
226
+ step 2240000 | 23288s | train_loss=0.0264
227
+ step 2250000 | 23385s | train_loss=0.0233
228
+ step 2260000 | 23482s | train_loss=0.0262
229
+ step 2270000 | 23579s | train_loss=0.0251
230
+ step 2280000 | 23674s | train_loss=0.0228
231
+ step 2290000 | 23771s | train_loss=0.0231
232
+ step 2300000 | 23938s | train_loss=0.0236 | fid=61.2967
233
+ step 2310000 | 24035s | train_loss=0.0247
234
+ step 2320000 | 24132s | train_loss=0.0245
235
+ step 2330000 | 24230s | train_loss=0.0260
236
+ step 2340000 | 24327s | train_loss=0.0223
237
+ step 2350000 | 24424s | train_loss=0.0234
238
+ step 2360000 | 24522s | train_loss=0.0252
239
+ step 2370000 | 24617s | train_loss=0.0251
240
+ step 2380000 | 24714s | train_loss=0.0260
241
+ step 2390000 | 24809s | train_loss=0.0221
242
+ step 2400000 | 24976s | train_loss=0.0224 | fid=62.5187
243
+ step 2410000 | 25074s | train_loss=0.0261
244
+ step 2420000 | 25171s | train_loss=0.0247
245
+ step 2430000 | 25268s | train_loss=0.0262
246
+ step 2440000 | 25366s | train_loss=0.0241
247
+ step 2450000 | 25463s | train_loss=0.0229
248
+ step 2460000 | 25559s | train_loss=0.0235
249
+ step 2470000 | 25656s | train_loss=0.0226
250
+ step 2480000 | 25751s | train_loss=0.0251
251
+ step 2490000 | 25848s | train_loss=0.0245
252
+ step 2500000 | 26017s | train_loss=0.0259 | fid=62.7560
253
+ step 2510000 | 26114s | train_loss=0.0240
254
+ step 2520000 | 26212s | train_loss=0.0227
255
+ step 2530000 | 26309s | train_loss=0.0229
256
+ step 2540000 | 26406s | train_loss=0.0225
257
+ step 2550000 | 26502s | train_loss=0.0237
258
+ step 2560000 | 26599s | train_loss=0.0216
259
+ step 2570000 | 26696s | train_loss=0.0221
260
+ step 2580000 | 26791s | train_loss=0.0229
261
+ step 2590000 | 26889s | train_loss=0.0242
262
+ step 2600000 | 27058s | train_loss=0.0217 | fid=60.4777
263
+ step 2610000 | 27155s | train_loss=0.0211
264
+ step 2620000 | 27253s | train_loss=0.0218
265
+ step 2630000 | 27350s | train_loss=0.0213
266
+ step 2640000 | 27445s | train_loss=0.0220
267
+ step 2650000 | 27543s | train_loss=0.0234
268
+ step 2660000 | 27639s | train_loss=0.0231
269
+ step 2670000 | 27735s | train_loss=0.0225
270
+ step 2680000 | 27832s | train_loss=0.0220
271
+ step 2690000 | 27930s | train_loss=0.0225
272
+ step 2700000 | 28098s | train_loss=0.0219 | fid=60.6492
273
+ step 2710000 | 28195s | train_loss=0.0211
274
+ step 2720000 | 28292s | train_loss=0.0221
275
+ step 2730000 | 28388s | train_loss=0.0213
276
+ step 2740000 | 28485s | train_loss=0.0236
277
+ step 2750000 | 28581s | train_loss=0.0233
278
+ step 2760000 | 28679s | train_loss=0.0209
279
+ step 2770000 | 28774s | train_loss=0.0229
280
+ step 2780000 | 28871s | train_loss=0.0207
281
+ step 2790000 | 28968s | train_loss=0.0204
282
+ step 2800000 | 29135s | train_loss=0.0234 | fid=59.2857
283
+ step 2810000 | 29232s | train_loss=0.0216
284
+ step 2820000 | 29328s | train_loss=0.0214
285
+ step 2830000 | 29425s | train_loss=0.0215
286
+ step 2840000 | 29522s | train_loss=0.0201
287
+ step 2850000 | 29619s | train_loss=0.0262
288
+ step 2860000 | 29714s | train_loss=0.0222
289
+ step 2870000 | 29812s | train_loss=0.0215
290
+ step 2880000 | 29909s | train_loss=0.0207
291
+ step 2890000 | 30006s | train_loss=0.0208
292
+ step 2900000 | 30175s | train_loss=0.0222 | fid=58.7565
293
+ step 2910000 | 30270s | train_loss=0.0209
294
+ step 2920000 | 30367s | train_loss=0.0194
295
+ step 2930000 | 30464s | train_loss=0.0230
296
+ step 2940000 | 30562s | train_loss=0.0217
297
+ step 2950000 | 30659s | train_loss=0.0201
298
+ step 2960000 | 30754s | train_loss=0.0189
299
+ step 2970000 | 30852s | train_loss=0.0204
300
+ step 2980000 | 30949s | train_loss=0.0218
301
+ step 2990000 | 31046s | train_loss=0.0225
302
+ step 3000000 | 31210s | train_loss=0.0209 | fid=59.3503
303
+ step 3010000 | 31307s | train_loss=0.0222
304
+ step 3020000 | 31404s | train_loss=0.0194
305
+ step 3030000 | 31501s | train_loss=0.0192
306
+ step 3040000 | 31598s | train_loss=0.0256
307
+ step 3050000 | 31694s | train_loss=0.0220
308
+ step 3060000 | 31791s | train_loss=0.0205
309
+ step 3070000 | 31888s | train_loss=0.0198
310
+ step 3080000 | 31985s | train_loss=0.0212
311
+ step 3090000 | 32081s | train_loss=0.0211
312
+ step 3100000 | 32249s | train_loss=0.0209 | fid=61.3206
313
+ step 3110000 | 32345s | train_loss=0.0212
314
+ step 3120000 | 32443s | train_loss=0.0194
315
+ step 3130000 | 32540s | train_loss=0.0205
316
+ step 3140000 | 32637s | train_loss=0.0213
317
+ step 3150000 | 32732s | train_loss=0.0210
318
+ step 3160000 | 32829s | train_loss=0.0192
319
+ step 3170000 | 32926s | train_loss=0.0193
320
+ step 3180000 | 33021s | train_loss=0.0234
321
+ step 3190000 | 33119s | train_loss=0.0215
322
+ step 3200000 | 33285s | train_loss=0.0210 | fid=58.3603
323
+ step 3210000 | 33382s | train_loss=0.0211
324
+ step 3220000 | 33479s | train_loss=0.0203
325
+ step 3230000 | 33576s | train_loss=0.0231
326
+ step 3240000 | 33671s | train_loss=0.0191
327
+ step 3250000 | 33768s | train_loss=0.0195
328
+ step 3260000 | 33866s | train_loss=0.0219
329
+ step 3270000 | 33961s | train_loss=0.0196
330
+ step 3280000 | 34058s | train_loss=0.0189
331
+ step 3290000 | 34155s | train_loss=0.0197
332
+ step 3300000 | 34322s | train_loss=0.0221 | fid=55.1089
333
+ step 3310000 | 34418s | train_loss=0.0190
334
+ step 3320000 | 34515s | train_loss=0.0192
335
+ step 3330000 | 34613s | train_loss=0.0202
336
+ step 3340000 | 34708s | train_loss=0.0191
337
+ step 3350000 | 34805s | train_loss=0.0199
338
+ step 3360000 | 34900s | train_loss=0.0190
339
+ step 3370000 | 34998s | train_loss=0.0201
340
+ step 3380000 | 35095s | train_loss=0.0212
341
+ step 3390000 | 35192s | train_loss=0.0203
342
+ step 3400000 | 35358s | train_loss=0.0193 | fid=55.8787
343
+ step 3410000 | 35455s | train_loss=0.0185
344
+ step 3420000 | 35552s | train_loss=0.0181
345
+ step 3430000 | 35647s | train_loss=0.0210
346
+ step 3440000 | 35744s | train_loss=0.0183
347
+ step 3450000 | 35840s | train_loss=0.0199
348
+ step 3460000 | 35937s | train_loss=0.0183
349
+ step 3470000 | 36034s | train_loss=0.0192
350
+ step 3480000 | 36131s | train_loss=0.0183
351
+ step 3490000 | 36228s | train_loss=0.0187
352
+ step 3500000 | 36393s | train_loss=0.0199 | fid=55.0119
353
+ step 3510000 | 36490s | train_loss=0.0214
354
+ step 3520000 | 36588s | train_loss=0.0184
355
+ step 3530000 | 36683s | train_loss=0.0191
356
+ step 3540000 | 36778s | train_loss=0.0204
357
+ step 3550000 | 36875s | train_loss=0.0185
358
+ step 3560000 | 36972s | train_loss=0.0189
359
+ step 3570000 | 37069s | train_loss=0.0199
360
+ step 3580000 | 37165s | train_loss=0.0189
361
+ step 3590000 | 37263s | train_loss=0.0202
362
+ step 3600000 | 37429s | train_loss=0.0194 | fid=58.1675
363
+ step 3610000 | 37526s | train_loss=0.0201
364
+ step 3620000 | 37621s | train_loss=0.0176
365
+ step 3630000 | 37716s | train_loss=0.0175
366
+ step 3640000 | 37813s | train_loss=0.0198
367
+ step 3650000 | 37911s | train_loss=0.0193
368
+ step 3660000 | 38007s | train_loss=0.0217
369
+ step 3670000 | 38103s | train_loss=0.0189
370
+ step 3680000 | 38200s | train_loss=0.0199
371
+ step 3690000 | 38298s | train_loss=0.0198
372
+ step 3700000 | 38464s | train_loss=0.0200 | fid=59.3654
373
+ step 3710000 | 38561s | train_loss=0.0175
374
+ step 3720000 | 38655s | train_loss=0.0187
375
+ step 3730000 | 38752s | train_loss=0.0182
376
+ step 3740000 | 38849s | train_loss=0.0199
377
+ step 3750000 | 38946s | train_loss=0.0202
378
+ step 3760000 | 39043s | train_loss=0.0184
379
+ step 3770000 | 39140s | train_loss=0.0181
380
+ step 3780000 | 39237s | train_loss=0.0193
381
+ step 3790000 | 39335s | train_loss=0.0193
382
+ step 3800000 | 39500s | train_loss=0.0199 | fid=56.4994
383
+ step 3810000 | 39594s | train_loss=0.0194
384
+ step 3820000 | 39692s | train_loss=0.0193
385
+ step 3830000 | 39788s | train_loss=0.0198
386
+ step 3840000 | 39885s | train_loss=0.0191
387
+ step 3850000 | 39982s | train_loss=0.0186
388
+ step 3860000 | 40079s | train_loss=0.0195
389
+ step 3870000 | 40176s | train_loss=0.0202
390
+ step 3880000 | 40273s | train_loss=0.0181
391
+ step 3890000 | 40371s | train_loss=0.0188
392
+ step 3900000 | 40536s | train_loss=0.0190 | fid=53.7992
393
+ step 3910000 | 40631s | train_loss=0.0181
394
+ step 3920000 | 40727s | train_loss=0.0199
395
+ step 3930000 | 40824s | train_loss=0.0178
396
+ step 3940000 | 40921s | train_loss=0.0188
397
+ step 3950000 | 41018s | train_loss=0.0184
398
+ step 3960000 | 41115s | train_loss=0.0218
399
+ step 3970000 | 41212s | train_loss=0.0182
400
+ step 3980000 | 41310s | train_loss=0.0222
401
+ step 3990000 | 41407s | train_loss=0.0194
402
+ step 4000000 | 41569s | train_loss=0.0191 | fid=55.1953
403
+ step 4010000 | 41665s | train_loss=0.0190
404
+ step 4020000 | 41762s | train_loss=0.0179
405
+ step 4030000 | 41859s | train_loss=0.0177
406
+ step 4040000 | 41956s | train_loss=0.0190
407
+ step 4050000 | 42054s | train_loss=0.0178
408
+ step 4060000 | 42150s | train_loss=0.0173
409
+ step 4070000 | 42248s | train_loss=0.0180
410
+ step 4080000 | 42345s | train_loss=0.0176
411
+ step 4090000 | 42441s | train_loss=0.0192
412
+ step 4100000 | 42605s | train_loss=0.0190 | fid=54.6558
413
+ step 4110000 | 42703s | train_loss=0.0198
414
+ step 4120000 | 42800s | train_loss=0.0178
415
+ step 4130000 | 42897s | train_loss=0.0180
416
+ step 4140000 | 42994s | train_loss=0.0170
417
+ step 4150000 | 43091s | train_loss=0.0177
418
+ step 4160000 | 43188s | train_loss=0.0182
419
+ step 4170000 | 43286s | train_loss=0.0195
420
+ step 4180000 | 43381s | train_loss=0.0178
421
+ step 4190000 | 43479s | train_loss=0.0184
422
+ step 4200000 | 43643s | train_loss=0.0169 | fid=53.5106
423
+ step 4210000 | 43740s | train_loss=0.0190
424
+ step 4220000 | 43837s | train_loss=0.0186
425
+ step 4230000 | 43934s | train_loss=0.0193
426
+ step 4240000 | 44031s | train_loss=0.0182
427
+ step 4250000 | 44129s | train_loss=0.0170
428
+ step 4260000 | 44226s | train_loss=0.0176
429
+ step 4270000 | 44321s | train_loss=0.0170
430
+ step 4280000 | 44418s | train_loss=0.0189
431
+ step 4290000 | 44514s | train_loss=0.0189
432
+ step 4300000 | 44683s | train_loss=0.0181 | fid=50.9260
433
+ step 4310000 | 44780s | train_loss=0.0181
434
+ step 4320000 | 44877s | train_loss=0.0180
435
+ step 4330000 | 44974s | train_loss=0.0183
436
+ step 4340000 | 45071s | train_loss=0.0188
437
+ step 4350000 | 45169s | train_loss=0.0172
438
+ step 4360000 | 45264s | train_loss=0.0172
439
+ step 4370000 | 45361s | train_loss=0.0174
440
+ step 4380000 | 45458s | train_loss=0.0194
441
+ step 4390000 | 45553s | train_loss=0.0199
442
+ step 4400000 | 45722s | train_loss=0.0179 | fid=50.6713
443
+ step 4410000 | 45819s | train_loss=0.0177
444
+ step 4420000 | 45916s | train_loss=0.0185
445
+ step 4430000 | 46013s | train_loss=0.0190
446
+ step 4440000 | 46111s | train_loss=0.0176
447
+ step 4450000 | 46206s | train_loss=0.0168
448
+ step 4460000 | 46303s | train_loss=0.0167
449
+ step 4470000 | 46401s | train_loss=0.0180
450
+ step 4480000 | 46496s | train_loss=0.0181
451
+ step 4490000 | 46593s | train_loss=0.0173
452
+ step 4500000 | 46761s | train_loss=0.0159 | fid=46.9822
453
+ step 4510000 | 46858s | train_loss=0.0171
454
+ step 4520000 | 46956s | train_loss=0.0200
455
+ step 4530000 | 47053s | train_loss=0.0187
456
+ step 4540000 | 47144s | train_loss=0.0172
457
+ step 4550000 | 47223s | train_loss=0.0178
458
+ step 4560000 | 47301s | train_loss=0.0169
459
+ step 4570000 | 47380s | train_loss=0.0183
460
+ step 4580000 | 47456s | train_loss=0.0180
461
+ step 4590000 | 47534s | train_loss=0.0165
462
+ step 4600000 | 47669s | train_loss=0.0163 | fid=50.7119
463
+ step 4610000 | 47748s | train_loss=0.0174
464
+ step 4620000 | 47826s | train_loss=0.0167
465
+ step 4630000 | 47905s | train_loss=0.0175
466
+ step 4640000 | 47983s | train_loss=0.0185
467
+ step 4650000 | 48062s | train_loss=0.0197
468
+ step 4660000 | 48140s | train_loss=0.0192
469
+ step 4670000 | 48216s | train_loss=0.0173
470
+ step 4680000 | 48294s | train_loss=0.0166
471
+ step 4690000 | 48372s | train_loss=0.0183
472
+ step 4700000 | 48508s | train_loss=0.0214 | fid=50.1210
473
+ step 4710000 | 48586s | train_loss=0.0169
474
+ step 4720000 | 48664s | train_loss=0.0197
475
+ step 4730000 | 48743s | train_loss=0.0178
476
+ step 4740000 | 48821s | train_loss=0.0169
477
+ step 4750000 | 48899s | train_loss=0.0158
478
+ step 4760000 | 48978s | train_loss=0.0170
479
+ step 4770000 | 49043s | train_loss=0.0158
480
+ step 4780000 | 49102s | train_loss=0.0156
481
+ step 4790000 | 49160s | train_loss=0.0176
482
+ step 4800000 | 49262s | train_loss=0.0168 | fid=51.1449
483
+ step 4810000 | 49322s | train_loss=0.0161
484
+ step 4820000 | 49382s | train_loss=0.0183
485
+ step 4830000 | 49442s | train_loss=0.0172
486
+ step 4840000 | 49501s | train_loss=0.0196
487
+ step 4850000 | 49561s | train_loss=0.0165
488
+ step 4860000 | 49621s | train_loss=0.0170
489
+ step 4870000 | 49681s | train_loss=0.0170
490
+ step 4880000 | 49741s | train_loss=0.0184
491
+ step 4890000 | 49798s | train_loss=0.0168
492
+ step 4900000 | 49902s | train_loss=0.0174 | fid=48.4835
493
+ step 4910000 | 49962s | train_loss=0.0169
494
+ step 4920000 | 50022s | train_loss=0.0173
495
+ step 4930000 | 50082s | train_loss=0.0168
496
+ step 4940000 | 50141s | train_loss=0.0167
497
+ step 4950000 | 50200s | train_loss=0.0170
498
+ step 4960000 | 50260s | train_loss=0.0175
499
+ step 4970000 | 50319s | train_loss=0.0161
500
+ step 4980000 | 50379s | train_loss=0.0164
501
+ step 4990000 | 50435s | train_loss=0.0169
502
+ step 5000000 | 50484s | train_loss=0.0174 | fid=49.0778
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d180/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d180/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 70s | train_loss=0.9951 | fid=345.3157
3
+ step 10000 | 167s | train_loss=0.6678
4
+ step 20000 | 263s | train_loss=0.5607
5
+ step 30000 | 359s | train_loss=0.4965
6
+ step 40000 | 456s | train_loss=0.4303
7
+ step 50000 | 553s | train_loss=0.3583
8
+ step 60000 | 649s | train_loss=0.2921
9
+ step 70000 | 746s | train_loss=0.2499
10
+ step 80000 | 843s | train_loss=0.2065
11
+ step 90000 | 939s | train_loss=0.1884
12
+ step 100000 | 1109s | train_loss=0.1647 | fid=65.6769
13
+ step 110000 | 1206s | train_loss=0.1607
14
+ step 120000 | 1301s | train_loss=0.1400
15
+ step 130000 | 1398s | train_loss=0.1360
16
+ step 140000 | 1494s | train_loss=0.1286
17
+ step 150000 | 1591s | train_loss=0.1205
18
+ step 160000 | 1688s | train_loss=0.1144
19
+ step 170000 | 1785s | train_loss=0.1114
20
+ step 180000 | 1882s | train_loss=0.1109
21
+ step 190000 | 1975s | train_loss=0.1014
22
+ step 200000 | 2146s | train_loss=0.1008 | fid=83.3151
23
+ step 210000 | 2241s | train_loss=0.0968
24
+ step 220000 | 2337s | train_loss=0.0940
25
+ step 230000 | 2434s | train_loss=0.0906
26
+ step 240000 | 2530s | train_loss=0.0924
27
+ step 250000 | 2627s | train_loss=0.0908
28
+ step 260000 | 2723s | train_loss=0.0902
29
+ step 270000 | 2820s | train_loss=0.0819
30
+ step 280000 | 2913s | train_loss=0.0883
31
+ step 290000 | 3010s | train_loss=0.0805
32
+ step 300000 | 3180s | train_loss=0.0840 | fid=93.6419
33
+ step 310000 | 3276s | train_loss=0.0792
34
+ step 320000 | 3372s | train_loss=0.0792
35
+ step 330000 | 3468s | train_loss=0.0757
36
+ step 340000 | 3565s | train_loss=0.0768
37
+ step 350000 | 3662s | train_loss=0.0762
38
+ step 360000 | 3759s | train_loss=0.0739
39
+ step 370000 | 3853s | train_loss=0.0733
40
+ step 380000 | 3948s | train_loss=0.0728
41
+ step 390000 | 4043s | train_loss=0.0700
42
+ step 400000 | 4215s | train_loss=0.0695 | fid=99.0523
43
+ step 410000 | 4309s | train_loss=0.0671
44
+ step 420000 | 4405s | train_loss=0.0683
45
+ step 430000 | 4502s | train_loss=0.0666
46
+ step 440000 | 4599s | train_loss=0.0658
47
+ step 450000 | 4695s | train_loss=0.0671
48
+ step 460000 | 4790s | train_loss=0.0635
49
+ step 470000 | 4886s | train_loss=0.0629
50
+ step 480000 | 4982s | train_loss=0.0633
51
+ step 490000 | 5077s | train_loss=0.0659
52
+ step 500000 | 5249s | train_loss=0.0605 | fid=99.3261
53
+ step 510000 | 5343s | train_loss=0.0631
54
+ step 520000 | 5439s | train_loss=0.0587
55
+ step 530000 | 5536s | train_loss=0.0598
56
+ step 540000 | 5633s | train_loss=0.0611
57
+ step 550000 | 5727s | train_loss=0.0598
58
+ step 560000 | 5823s | train_loss=0.0566
59
+ step 570000 | 5919s | train_loss=0.0579
60
+ step 580000 | 6014s | train_loss=0.0639
61
+ step 590000 | 6110s | train_loss=0.0573
62
+ step 600000 | 6282s | train_loss=0.0578 | fid=104.3311
63
+ step 610000 | 6376s | train_loss=0.0537
64
+ step 620000 | 6472s | train_loss=0.0550
65
+ step 630000 | 6569s | train_loss=0.0572
66
+ step 640000 | 6663s | train_loss=0.0521
67
+ step 650000 | 6759s | train_loss=0.0604
68
+ step 660000 | 6856s | train_loss=0.0566
69
+ step 670000 | 6952s | train_loss=0.0536
70
+ step 680000 | 7047s | train_loss=0.0543
71
+ step 690000 | 7143s | train_loss=0.0550
72
+ step 700000 | 7315s | train_loss=0.0535 | fid=99.2349
73
+ step 710000 | 7409s | train_loss=0.0500
74
+ step 720000 | 7506s | train_loss=0.0504
75
+ step 730000 | 7600s | train_loss=0.0511
76
+ step 740000 | 7696s | train_loss=0.0532
77
+ step 750000 | 7793s | train_loss=0.0500
78
+ step 760000 | 7889s | train_loss=0.0496
79
+ step 770000 | 7983s | train_loss=0.0519
80
+ step 780000 | 8080s | train_loss=0.0518
81
+ step 790000 | 8176s | train_loss=0.0495
82
+ step 800000 | 8349s | train_loss=0.0487 | fid=95.0502
83
+ step 810000 | 8443s | train_loss=0.0499
84
+ step 820000 | 8538s | train_loss=0.0519
85
+ step 830000 | 8635s | train_loss=0.0483
86
+ step 840000 | 8731s | train_loss=0.0484
87
+ step 850000 | 8827s | train_loss=0.0499
88
+ step 860000 | 8924s | train_loss=0.0454
89
+ step 870000 | 9018s | train_loss=0.0450
90
+ step 880000 | 9115s | train_loss=0.0471
91
+ step 890000 | 9212s | train_loss=0.0460
92
+ step 900000 | 9384s | train_loss=0.0457 | fid=93.3052
93
+ step 910000 | 9476s | train_loss=0.0435
94
+ step 920000 | 9572s | train_loss=0.0451
95
+ step 930000 | 9669s | train_loss=0.0472
96
+ step 940000 | 9765s | train_loss=0.0507
97
+ step 950000 | 9861s | train_loss=0.0464
98
+ step 960000 | 9956s | train_loss=0.0439
99
+ step 970000 | 10053s | train_loss=0.0426
100
+ step 980000 | 10149s | train_loss=0.0577
101
+ step 990000 | 10246s | train_loss=0.0418
102
+ step 1000000 | 10415s | train_loss=0.0440 | fid=86.0909
103
+ step 1010000 | 10509s | train_loss=0.0418
104
+ step 1020000 | 10606s | train_loss=0.0438
105
+ step 1030000 | 10702s | train_loss=0.0445
106
+ step 1040000 | 10798s | train_loss=0.0444
107
+ step 1050000 | 10895s | train_loss=0.0450
108
+ step 1060000 | 10989s | train_loss=0.0409
109
+ step 1070000 | 11086s | train_loss=0.0416
110
+ step 1080000 | 11182s | train_loss=0.0412
111
+ step 1090000 | 11279s | train_loss=0.0405
112
+ step 1100000 | 11448s | train_loss=0.0431 | fid=81.6066
113
+ step 1110000 | 11543s | train_loss=0.0411
114
+ step 1120000 | 11639s | train_loss=0.0431
115
+ step 1130000 | 11736s | train_loss=0.0395
116
+ step 1140000 | 11832s | train_loss=0.0403
117
+ step 1150000 | 11927s | train_loss=0.0389
118
+ step 1160000 | 12023s | train_loss=0.0405
119
+ step 1170000 | 12120s | train_loss=0.0415
120
+ step 1180000 | 12216s | train_loss=0.0400
121
+ step 1190000 | 12311s | train_loss=0.0398
122
+ step 1200000 | 12483s | train_loss=0.0421 | fid=75.5364
123
+ step 1210000 | 12578s | train_loss=0.0384
124
+ step 1220000 | 12674s | train_loss=0.0380
125
+ step 1230000 | 12770s | train_loss=0.0390
126
+ step 1240000 | 12866s | train_loss=0.0406
127
+ step 1250000 | 12961s | train_loss=0.0390
128
+ step 1260000 | 13058s | train_loss=0.0385
129
+ step 1270000 | 13154s | train_loss=0.0399
130
+ step 1280000 | 13249s | train_loss=0.0392
131
+ step 1290000 | 13345s | train_loss=0.0396
132
+ step 1300000 | 13517s | train_loss=0.0431 | fid=66.3190
133
+ step 1310000 | 13612s | train_loss=0.0373
134
+ step 1320000 | 13709s | train_loss=0.0418
135
+ step 1330000 | 13805s | train_loss=0.0365
136
+ step 1340000 | 13900s | train_loss=0.0378
137
+ step 1350000 | 13997s | train_loss=0.0381
138
+ step 1360000 | 14093s | train_loss=0.0386
139
+ step 1370000 | 14188s | train_loss=0.0382
140
+ step 1380000 | 14284s | train_loss=0.0354
141
+ step 1390000 | 14381s | train_loss=0.0384
142
+ step 1400000 | 14552s | train_loss=0.0373 | fid=64.8511
143
+ step 1410000 | 14647s | train_loss=0.0365
144
+ step 1420000 | 14744s | train_loss=0.0372
145
+ step 1430000 | 14841s | train_loss=0.0377
146
+ step 1440000 | 14935s | train_loss=0.0378
147
+ step 1450000 | 15032s | train_loss=0.0354
148
+ step 1460000 | 15127s | train_loss=0.0367
149
+ step 1470000 | 15223s | train_loss=0.0349
150
+ step 1480000 | 15320s | train_loss=0.0365
151
+ step 1490000 | 15416s | train_loss=0.0376
152
+ step 1500000 | 15588s | train_loss=0.0369 | fid=58.6195
153
+ step 1510000 | 15683s | train_loss=0.0355
154
+ step 1520000 | 15779s | train_loss=0.0387
155
+ step 1530000 | 15874s | train_loss=0.0416
156
+ step 1540000 | 15971s | train_loss=0.0342
157
+ step 1550000 | 16066s | train_loss=0.0341
158
+ step 1560000 | 16162s | train_loss=0.0354
159
+ step 1570000 | 16259s | train_loss=0.0337
160
+ step 1580000 | 16355s | train_loss=0.0326
161
+ step 1590000 | 16452s | train_loss=0.0323
162
+ step 1600000 | 16623s | train_loss=0.0358 | fid=56.4349
163
+ step 1610000 | 16718s | train_loss=0.0342
164
+ step 1620000 | 16815s | train_loss=0.0364
165
+ step 1630000 | 16910s | train_loss=0.0348
166
+ step 1640000 | 17005s | train_loss=0.0343
167
+ step 1650000 | 17101s | train_loss=0.0344
168
+ step 1660000 | 17197s | train_loss=0.0350
169
+ step 1670000 | 17294s | train_loss=0.0327
170
+ step 1680000 | 17391s | train_loss=0.0360
171
+ step 1690000 | 17487s | train_loss=0.0336
172
+ step 1700000 | 17659s | train_loss=0.0328 | fid=54.8841
173
+ step 1710000 | 17754s | train_loss=0.0320
174
+ step 1720000 | 17849s | train_loss=0.0351
175
+ step 1730000 | 17943s | train_loss=0.0337
176
+ step 1740000 | 18040s | train_loss=0.0376
177
+ step 1750000 | 18136s | train_loss=0.0321
178
+ step 1760000 | 18233s | train_loss=0.0303
179
+ step 1770000 | 18329s | train_loss=0.0317
180
+ step 1780000 | 18426s | train_loss=0.0324
181
+ step 1790000 | 18522s | train_loss=0.0381
182
+ step 1800000 | 18694s | train_loss=0.0327 | fid=57.4873
183
+ step 1810000 | 18790s | train_loss=0.0370
184
+ step 1820000 | 18883s | train_loss=0.0315
185
+ step 1830000 | 18979s | train_loss=0.0313
186
+ step 1840000 | 19076s | train_loss=0.0311
187
+ step 1850000 | 19173s | train_loss=0.0293
188
+ step 1860000 | 19269s | train_loss=0.0306
189
+ step 1870000 | 19366s | train_loss=0.0309
190
+ step 1880000 | 19463s | train_loss=0.0304
191
+ step 1890000 | 19559s | train_loss=0.0320
192
+ step 1900000 | 19731s | train_loss=0.0299 | fid=58.9902
193
+ step 1910000 | 19822s | train_loss=0.0324
194
+ step 1920000 | 19919s | train_loss=0.0315
195
+ step 1930000 | 20015s | train_loss=0.0300
196
+ step 1940000 | 20111s | train_loss=0.0313
197
+ step 1950000 | 20208s | train_loss=0.0325
198
+ step 1960000 | 20304s | train_loss=0.0295
199
+ step 1970000 | 20401s | train_loss=0.0316
200
+ step 1980000 | 20497s | train_loss=0.0335
201
+ step 1990000 | 20593s | train_loss=0.0297
202
+ step 2000000 | 20764s | train_loss=0.0303 | fid=58.0105
203
+ step 2010000 | 20858s | train_loss=0.0316
204
+ step 2020000 | 20954s | train_loss=0.0313
205
+ step 2030000 | 21051s | train_loss=0.0315
206
+ step 2040000 | 21148s | train_loss=0.0284
207
+ step 2050000 | 21244s | train_loss=0.0329
208
+ step 2060000 | 21341s | train_loss=0.0294
209
+ step 2070000 | 21438s | train_loss=0.0322
210
+ step 2080000 | 21534s | train_loss=0.0290
211
+ step 2090000 | 21630s | train_loss=0.0289
212
+ step 2100000 | 21800s | train_loss=0.0299 | fid=59.5889
213
+ step 2110000 | 21895s | train_loss=0.0290
214
+ step 2120000 | 21991s | train_loss=0.0296
215
+ step 2130000 | 22088s | train_loss=0.0288
216
+ step 2140000 | 22184s | train_loss=0.0291
217
+ step 2150000 | 22281s | train_loss=0.0276
218
+ step 2160000 | 22378s | train_loss=0.0302
219
+ step 2170000 | 22474s | train_loss=0.0292
220
+ step 2180000 | 22571s | train_loss=0.0281
221
+ step 2190000 | 22666s | train_loss=0.0290
222
+ step 2200000 | 22836s | train_loss=0.0282 | fid=60.3940
223
+ step 2210000 | 22931s | train_loss=0.0321
224
+ step 2220000 | 23028s | train_loss=0.0295
225
+ step 2230000 | 23125s | train_loss=0.0293
226
+ step 2240000 | 23221s | train_loss=0.0302
227
+ step 2250000 | 23318s | train_loss=0.0276
228
+ step 2260000 | 23414s | train_loss=0.0263
229
+ step 2270000 | 23511s | train_loss=0.0341
230
+ step 2280000 | 23605s | train_loss=0.0286
231
+ step 2290000 | 23702s | train_loss=0.0271
232
+ step 2300000 | 23873s | train_loss=0.0288 | fid=63.7584
233
+ step 2310000 | 23968s | train_loss=0.0268
234
+ step 2320000 | 24064s | train_loss=0.0300
235
+ step 2330000 | 24161s | train_loss=0.0276
236
+ step 2340000 | 24258s | train_loss=0.0263
237
+ step 2350000 | 24354s | train_loss=0.0291
238
+ step 2360000 | 24451s | train_loss=0.0285
239
+ step 2370000 | 24546s | train_loss=0.0296
240
+ step 2380000 | 24643s | train_loss=0.0285
241
+ step 2390000 | 24739s | train_loss=0.0266
242
+ step 2400000 | 24909s | train_loss=0.0261 | fid=63.5887
243
+ step 2410000 | 25004s | train_loss=0.0282
244
+ step 2420000 | 25101s | train_loss=0.0268
245
+ step 2430000 | 25198s | train_loss=0.0274
246
+ step 2440000 | 25294s | train_loss=0.0279
247
+ step 2450000 | 25391s | train_loss=0.0251
248
+ step 2460000 | 25486s | train_loss=0.0270
249
+ step 2470000 | 25582s | train_loss=0.0262
250
+ step 2480000 | 25678s | train_loss=0.0263
251
+ step 2490000 | 25773s | train_loss=0.0285
252
+ step 2500000 | 25946s | train_loss=0.0274 | fid=54.0249
253
+ step 2510000 | 26041s | train_loss=0.0258
254
+ step 2520000 | 26137s | train_loss=0.0263
255
+ step 2530000 | 26234s | train_loss=0.0277
256
+ step 2540000 | 26331s | train_loss=0.0244
257
+ step 2550000 | 26426s | train_loss=0.0254
258
+ step 2560000 | 26522s | train_loss=0.0260
259
+ step 2570000 | 26619s | train_loss=0.0272
260
+ step 2580000 | 26715s | train_loss=0.0245
261
+ step 2590000 | 26810s | train_loss=0.0289
262
+ step 2600000 | 26983s | train_loss=0.0269 | fid=57.8750
263
+ step 2610000 | 27078s | train_loss=0.0276
264
+ step 2620000 | 27175s | train_loss=0.0268
265
+ step 2630000 | 27272s | train_loss=0.0246
266
+ step 2640000 | 27367s | train_loss=0.0270
267
+ step 2650000 | 27463s | train_loss=0.0261
268
+ step 2660000 | 27559s | train_loss=0.0254
269
+ step 2670000 | 27656s | train_loss=0.0254
270
+ step 2680000 | 27751s | train_loss=0.0247
271
+ step 2690000 | 27848s | train_loss=0.0310
272
+ step 2700000 | 28020s | train_loss=0.0262 | fid=57.6780
273
+ step 2710000 | 28115s | train_loss=0.0268
274
+ step 2720000 | 28212s | train_loss=0.0258
275
+ step 2730000 | 28306s | train_loss=0.0244
276
+ step 2740000 | 28403s | train_loss=0.0241
277
+ step 2750000 | 28499s | train_loss=0.0255
278
+ step 2760000 | 28596s | train_loss=0.0252
279
+ step 2770000 | 28692s | train_loss=0.0256
280
+ step 2780000 | 28787s | train_loss=0.0265
281
+ step 2790000 | 28883s | train_loss=0.0250
282
+ step 2800000 | 29056s | train_loss=0.0242 | fid=55.3447
283
+ step 2810000 | 29151s | train_loss=0.0253
284
+ step 2820000 | 29246s | train_loss=0.0243
285
+ step 2830000 | 29343s | train_loss=0.0265
286
+ step 2840000 | 29439s | train_loss=0.0234
287
+ step 2850000 | 29535s | train_loss=0.0241
288
+ step 2860000 | 29632s | train_loss=0.0228
289
+ step 2870000 | 29726s | train_loss=0.0239
290
+ step 2880000 | 29823s | train_loss=0.0258
291
+ step 2890000 | 29920s | train_loss=0.0233
292
+ step 2900000 | 30092s | train_loss=0.0286 | fid=53.9513
293
+ step 2910000 | 30186s | train_loss=0.0261
294
+ step 2920000 | 30282s | train_loss=0.0251
295
+ step 2930000 | 30379s | train_loss=0.0247
296
+ step 2940000 | 30475s | train_loss=0.0267
297
+ step 2950000 | 30572s | train_loss=0.0236
298
+ step 2960000 | 30668s | train_loss=0.0231
299
+ step 2970000 | 30763s | train_loss=0.0276
300
+ step 2980000 | 30860s | train_loss=0.0244
301
+ step 2990000 | 30957s | train_loss=0.0229
302
+ step 3000000 | 31128s | train_loss=0.0246 | fid=55.5251
303
+ step 3010000 | 31224s | train_loss=0.0233
304
+ step 3020000 | 31320s | train_loss=0.0244
305
+ step 3030000 | 31417s | train_loss=0.0213
306
+ step 3040000 | 31513s | train_loss=0.0270
307
+ step 3050000 | 31610s | train_loss=0.0225
308
+ step 3060000 | 31705s | train_loss=0.0249
309
+ step 3070000 | 31802s | train_loss=0.0256
310
+ step 3080000 | 31899s | train_loss=0.0227
311
+ step 3090000 | 31995s | train_loss=0.0227
312
+ step 3100000 | 32166s | train_loss=0.0213 | fid=53.8581
313
+ step 3110000 | 32261s | train_loss=0.0254
314
+ step 3120000 | 32357s | train_loss=0.0227
315
+ step 3130000 | 32453s | train_loss=0.0229
316
+ step 3140000 | 32550s | train_loss=0.0245
317
+ step 3150000 | 32646s | train_loss=0.0223
318
+ step 3160000 | 32741s | train_loss=0.0222
319
+ step 3170000 | 32837s | train_loss=0.0237
320
+ step 3180000 | 32933s | train_loss=0.0237
321
+ step 3190000 | 33027s | train_loss=0.0219
322
+ step 3200000 | 33200s | train_loss=0.0246 | fid=54.4545
323
+ step 3210000 | 33295s | train_loss=0.0233
324
+ step 3220000 | 33392s | train_loss=0.0256
325
+ step 3230000 | 33489s | train_loss=0.0256
326
+ step 3240000 | 33585s | train_loss=0.0259
327
+ step 3250000 | 33680s | train_loss=0.0246
328
+ step 3260000 | 33777s | train_loss=0.0245
329
+ step 3270000 | 33873s | train_loss=0.0261
330
+ step 3280000 | 33968s | train_loss=0.0211
331
+ step 3290000 | 34064s | train_loss=0.0210
332
+ step 3300000 | 34236s | train_loss=0.0238 | fid=51.8879
333
+ step 3310000 | 34332s | train_loss=0.0247
334
+ step 3320000 | 34428s | train_loss=0.0260
335
+ step 3330000 | 34524s | train_loss=0.0222
336
+ step 3340000 | 34620s | train_loss=0.0215
337
+ step 3350000 | 34715s | train_loss=0.0232
338
+ step 3360000 | 34812s | train_loss=0.0227
339
+ step 3370000 | 34906s | train_loss=0.0242
340
+ step 3380000 | 35003s | train_loss=0.0217
341
+ step 3390000 | 35099s | train_loss=0.0232
342
+ step 3400000 | 35272s | train_loss=0.0242 | fid=51.6108
343
+ step 3410000 | 35366s | train_loss=0.0231
344
+ step 3420000 | 35463s | train_loss=0.0226
345
+ step 3430000 | 35559s | train_loss=0.0244
346
+ step 3440000 | 35654s | train_loss=0.0223
347
+ step 3450000 | 35750s | train_loss=0.0223
348
+ step 3460000 | 35845s | train_loss=0.0251
349
+ step 3470000 | 35942s | train_loss=0.0229
350
+ step 3480000 | 36038s | train_loss=0.0213
351
+ step 3490000 | 36135s | train_loss=0.0230
352
+ step 3500000 | 36306s | train_loss=0.0249 | fid=50.1647
353
+ step 3510000 | 36401s | train_loss=0.0219
354
+ step 3520000 | 36498s | train_loss=0.0212
355
+ step 3530000 | 36595s | train_loss=0.0209
356
+ step 3540000 | 36689s | train_loss=0.0224
357
+ step 3550000 | 36784s | train_loss=0.0218
358
+ step 3560000 | 36880s | train_loss=0.0247
359
+ step 3570000 | 36977s | train_loss=0.0207
360
+ step 3580000 | 37073s | train_loss=0.0233
361
+ step 3590000 | 37169s | train_loss=0.0213
362
+ step 3600000 | 37342s | train_loss=0.0220 | fid=48.7682
363
+ step 3610000 | 37437s | train_loss=0.0227
364
+ step 3620000 | 37534s | train_loss=0.0220
365
+ step 3630000 | 37628s | train_loss=0.0230
366
+ step 3640000 | 37723s | train_loss=0.0213
367
+ step 3650000 | 37820s | train_loss=0.0219
368
+ step 3660000 | 37916s | train_loss=0.0232
369
+ step 3670000 | 38012s | train_loss=0.0210
370
+ step 3680000 | 38108s | train_loss=0.0249
371
+ step 3690000 | 38204s | train_loss=0.0253
372
+ step 3700000 | 38377s | train_loss=0.0209 | fid=50.7961
373
+ step 3710000 | 38472s | train_loss=0.0223
374
+ step 3720000 | 38568s | train_loss=0.0226
375
+ step 3730000 | 38662s | train_loss=0.0209
376
+ step 3740000 | 38758s | train_loss=0.0211
377
+ step 3750000 | 38855s | train_loss=0.0206
378
+ step 3760000 | 38950s | train_loss=0.0203
379
+ step 3770000 | 39047s | train_loss=0.0223
380
+ step 3780000 | 39143s | train_loss=0.0203
381
+ step 3790000 | 39240s | train_loss=0.0236
382
+ step 3800000 | 39413s | train_loss=0.0210 | fid=47.6852
383
+ step 3810000 | 39509s | train_loss=0.0210
384
+ step 3820000 | 39602s | train_loss=0.0202
385
+ step 3830000 | 39698s | train_loss=0.0201
386
+ step 3840000 | 39794s | train_loss=0.0210
387
+ step 3850000 | 39890s | train_loss=0.0219
388
+ step 3860000 | 39987s | train_loss=0.0222
389
+ step 3870000 | 40083s | train_loss=0.0205
390
+ step 3880000 | 40179s | train_loss=0.0217
391
+ step 3890000 | 40276s | train_loss=0.0210
392
+ step 3900000 | 40450s | train_loss=0.0217 | fid=48.9140
393
+ step 3910000 | 40542s | train_loss=0.0196
394
+ step 3920000 | 40637s | train_loss=0.0210
395
+ step 3930000 | 40733s | train_loss=0.0216
396
+ step 3940000 | 40829s | train_loss=0.0206
397
+ step 3950000 | 40925s | train_loss=0.0200
398
+ step 3960000 | 41022s | train_loss=0.0202
399
+ step 3970000 | 41118s | train_loss=0.0209
400
+ step 3980000 | 41215s | train_loss=0.0214
401
+ step 3990000 | 41311s | train_loss=0.0248
402
+ step 4000000 | 41483s | train_loss=0.0216 | fid=48.0519
403
+ step 4010000 | 41577s | train_loss=0.0198
404
+ step 4020000 | 41673s | train_loss=0.0221
405
+ step 4030000 | 41769s | train_loss=0.0196
406
+ step 4040000 | 41866s | train_loss=0.0199
407
+ step 4050000 | 41962s | train_loss=0.0191
408
+ step 4060000 | 42059s | train_loss=0.0185
409
+ step 4070000 | 42155s | train_loss=0.0213
410
+ step 4080000 | 42252s | train_loss=0.0190
411
+ step 4090000 | 42348s | train_loss=0.0212
412
+ step 4100000 | 42519s | train_loss=0.0213 | fid=47.7898
413
+ step 4110000 | 42612s | train_loss=0.0200
414
+ step 4120000 | 42709s | train_loss=0.0193
415
+ step 4130000 | 42805s | train_loss=0.0206
416
+ step 4140000 | 42902s | train_loss=0.0210
417
+ step 4150000 | 42998s | train_loss=0.0227
418
+ step 4160000 | 43094s | train_loss=0.0200
419
+ step 4170000 | 43191s | train_loss=0.0220
420
+ step 4180000 | 43288s | train_loss=0.0198
421
+ step 4190000 | 43383s | train_loss=0.0213
422
+ step 4200000 | 43554s | train_loss=0.0191 | fid=42.5568
423
+ step 4210000 | 43649s | train_loss=0.0205
424
+ step 4220000 | 43745s | train_loss=0.0196
425
+ step 4230000 | 43841s | train_loss=0.0250
426
+ step 4240000 | 43937s | train_loss=0.0212
427
+ step 4250000 | 44034s | train_loss=0.0196
428
+ step 4260000 | 44130s | train_loss=0.0197
429
+ step 4270000 | 44227s | train_loss=0.0194
430
+ step 4280000 | 44322s | train_loss=0.0200
431
+ step 4290000 | 44418s | train_loss=0.0199
432
+ step 4300000 | 44589s | train_loss=0.0205 | fid=44.4774
433
+ step 4310000 | 44684s | train_loss=0.0197
434
+ step 4320000 | 44781s | train_loss=0.0196
435
+ step 4330000 | 44877s | train_loss=0.0190
436
+ step 4340000 | 44974s | train_loss=0.0205
437
+ step 4350000 | 45070s | train_loss=0.0210
438
+ step 4360000 | 45167s | train_loss=0.0197
439
+ step 4370000 | 45261s | train_loss=0.0201
440
+ step 4380000 | 45358s | train_loss=0.0195
441
+ step 4390000 | 45454s | train_loss=0.0198
442
+ step 4400000 | 45624s | train_loss=0.0213 | fid=45.3550
443
+ step 4410000 | 45719s | train_loss=0.0198
444
+ step 4420000 | 45816s | train_loss=0.0207
445
+ step 4430000 | 45912s | train_loss=0.0180
446
+ step 4440000 | 46009s | train_loss=0.0188
447
+ step 4450000 | 46106s | train_loss=0.0197
448
+ step 4460000 | 46200s | train_loss=0.0189
449
+ step 4470000 | 46297s | train_loss=0.0193
450
+ step 4480000 | 46394s | train_loss=0.0195
451
+ step 4490000 | 46488s | train_loss=0.0198
452
+ step 4500000 | 46660s | train_loss=0.0210 | fid=42.3287
453
+ step 4510000 | 46756s | train_loss=0.0190
454
+ step 4520000 | 46852s | train_loss=0.0220
455
+ step 4530000 | 46949s | train_loss=0.0187
456
+ step 4540000 | 47045s | train_loss=0.0209
457
+ step 4550000 | 47137s | train_loss=0.0191
458
+ step 4560000 | 47215s | train_loss=0.0213
459
+ step 4570000 | 47293s | train_loss=0.0198
460
+ step 4580000 | 47371s | train_loss=0.0193
461
+ step 4590000 | 47446s | train_loss=0.0182
462
+ step 4600000 | 47586s | train_loss=0.0190 | fid=44.1092
463
+ step 4610000 | 47663s | train_loss=0.0195
464
+ step 4620000 | 47740s | train_loss=0.0204
465
+ step 4630000 | 47818s | train_loss=0.0198
466
+ step 4640000 | 47896s | train_loss=0.0195
467
+ step 4650000 | 47974s | train_loss=0.0202
468
+ step 4660000 | 48053s | train_loss=0.0198
469
+ step 4670000 | 48130s | train_loss=0.0190
470
+ step 4680000 | 48207s | train_loss=0.0229
471
+ step 4690000 | 48284s | train_loss=0.0180
472
+ step 4700000 | 48424s | train_loss=0.0200 | fid=45.8360
473
+ step 4710000 | 48501s | train_loss=0.0193
474
+ step 4720000 | 48577s | train_loss=0.0191
475
+ step 4730000 | 48655s | train_loss=0.0189
476
+ step 4740000 | 48733s | train_loss=0.0199
477
+ step 4750000 | 48811s | train_loss=0.0190
478
+ step 4760000 | 48889s | train_loss=0.0186
479
+ step 4770000 | 48967s | train_loss=0.0180
480
+ step 4780000 | 49034s | train_loss=0.0178
481
+ step 4790000 | 49093s | train_loss=0.0192
482
+ step 4800000 | 49198s | train_loss=0.0194 | fid=44.8237
483
+ step 4810000 | 49255s | train_loss=0.0238
484
+ step 4820000 | 49314s | train_loss=0.0200
485
+ step 4830000 | 49374s | train_loss=0.0186
486
+ step 4840000 | 49433s | train_loss=0.0180
487
+ step 4850000 | 49493s | train_loss=0.0173
488
+ step 4860000 | 49552s | train_loss=0.0175
489
+ step 4870000 | 49612s | train_loss=0.0197
490
+ step 4880000 | 49672s | train_loss=0.0185
491
+ step 4890000 | 49731s | train_loss=0.0197
492
+ step 4900000 | 49838s | train_loss=0.0177 | fid=46.1419
493
+ step 4910000 | 49895s | train_loss=0.0175
494
+ step 4920000 | 49954s | train_loss=0.0185
495
+ step 4930000 | 50013s | train_loss=0.0181
496
+ step 4940000 | 50073s | train_loss=0.0185
497
+ step 4950000 | 50131s | train_loss=0.0190
498
+ step 4960000 | 50190s | train_loss=0.0216
499
+ step 4970000 | 50250s | train_loss=0.0190
500
+ step 4980000 | 50309s | train_loss=0.0190
501
+ step 4990000 | 50368s | train_loss=0.0191
502
+ step 5000000 | 50463s | train_loss=0.0190 | fid=47.1872
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d20/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d20/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 65s | train_loss=1.0089 | fid=224.6709
3
+ step 10000 | 146s | train_loss=0.8834
4
+ step 20000 | 231s | train_loss=0.8775
5
+ step 30000 | 315s | train_loss=0.8727
6
+ step 40000 | 399s | train_loss=0.8670
7
+ step 50000 | 483s | train_loss=0.8607
8
+ step 60000 | 567s | train_loss=0.8537
9
+ step 70000 | 651s | train_loss=0.8432
10
+ step 80000 | 735s | train_loss=0.8242
11
+ step 90000 | 819s | train_loss=0.7929
12
+ step 100000 | 966s | train_loss=0.7508 | fid=72.0185
13
+ step 110000 | 1045s | train_loss=0.6937
14
+ step 120000 | 1129s | train_loss=0.6179
15
+ step 130000 | 1213s | train_loss=0.5296
16
+ step 140000 | 1297s | train_loss=0.4479
17
+ step 150000 | 1382s | train_loss=0.3742
18
+ step 160000 | 1466s | train_loss=0.3164
19
+ step 170000 | 1550s | train_loss=0.2757
20
+ step 180000 | 1634s | train_loss=0.2467
21
+ step 190000 | 1718s | train_loss=0.2240
22
+ step 200000 | 1865s | train_loss=0.2090 | fid=72.4469
23
+ step 210000 | 1945s | train_loss=0.1960
24
+ step 220000 | 2027s | train_loss=0.1875
25
+ step 230000 | 2111s | train_loss=0.1764
26
+ step 240000 | 2195s | train_loss=0.1669
27
+ step 250000 | 2279s | train_loss=0.1610
28
+ step 260000 | 2363s | train_loss=0.1534
29
+ step 270000 | 2447s | train_loss=0.1470
30
+ step 280000 | 2532s | train_loss=0.1420
31
+ step 290000 | 2616s | train_loss=0.1367
32
+ step 300000 | 2762s | train_loss=0.1314 | fid=70.6538
33
+ step 310000 | 2844s | train_loss=0.1283
34
+ step 320000 | 2926s | train_loss=0.1248
35
+ step 330000 | 3009s | train_loss=0.1210
36
+ step 340000 | 3093s | train_loss=0.1161
37
+ step 350000 | 3177s | train_loss=0.1157
38
+ step 360000 | 3261s | train_loss=0.1108
39
+ step 370000 | 3345s | train_loss=0.1091
40
+ step 380000 | 3429s | train_loss=0.1052
41
+ step 390000 | 3513s | train_loss=0.1024
42
+ step 400000 | 3662s | train_loss=0.1005 | fid=70.1288
43
+ step 410000 | 3743s | train_loss=0.0975
44
+ step 420000 | 3825s | train_loss=0.0963
45
+ step 430000 | 3909s | train_loss=0.0935
46
+ step 440000 | 3992s | train_loss=0.0928
47
+ step 450000 | 4076s | train_loss=0.0901
48
+ step 460000 | 4160s | train_loss=0.0892
49
+ step 470000 | 4244s | train_loss=0.0873
50
+ step 480000 | 4328s | train_loss=0.0850
51
+ step 490000 | 4412s | train_loss=0.0845
52
+ step 500000 | 4561s | train_loss=0.0825 | fid=68.6364
53
+ step 510000 | 4643s | train_loss=0.0832
54
+ step 520000 | 4727s | train_loss=0.0800
55
+ step 530000 | 4808s | train_loss=0.0791
56
+ step 540000 | 4892s | train_loss=0.0773
57
+ step 550000 | 4975s | train_loss=0.0759
58
+ step 560000 | 5059s | train_loss=0.0756
59
+ step 570000 | 5143s | train_loss=0.0747
60
+ step 580000 | 5227s | train_loss=0.0733
61
+ step 590000 | 5311s | train_loss=0.0727
62
+ step 600000 | 5462s | train_loss=0.0716 | fid=69.6382
63
+ step 610000 | 5543s | train_loss=0.0703
64
+ step 620000 | 5628s | train_loss=0.0688
65
+ step 630000 | 5709s | train_loss=0.0687
66
+ step 640000 | 5793s | train_loss=0.0672
67
+ step 650000 | 5876s | train_loss=0.0671
68
+ step 660000 | 5960s | train_loss=0.0657
69
+ step 670000 | 6044s | train_loss=0.0654
70
+ step 680000 | 6128s | train_loss=0.0656
71
+ step 690000 | 6212s | train_loss=0.0634
72
+ step 700000 | 6363s | train_loss=0.0630 | fid=71.0725
73
+ step 710000 | 6444s | train_loss=0.0625
74
+ step 720000 | 6528s | train_loss=0.0611
75
+ step 730000 | 6610s | train_loss=0.0607
76
+ step 740000 | 6694s | train_loss=0.0611
77
+ step 750000 | 6778s | train_loss=0.0600
78
+ step 760000 | 6860s | train_loss=0.0591
79
+ step 770000 | 6945s | train_loss=0.0589
80
+ step 780000 | 7029s | train_loss=0.0577
81
+ step 790000 | 7113s | train_loss=0.0580
82
+ step 800000 | 7263s | train_loss=0.0570 | fid=68.9953
83
+ step 810000 | 7344s | train_loss=0.0576
84
+ step 820000 | 7428s | train_loss=0.0564
85
+ step 830000 | 7512s | train_loss=0.0548
86
+ step 840000 | 7593s | train_loss=0.0548
87
+ step 850000 | 7677s | train_loss=0.0546
88
+ step 860000 | 7762s | train_loss=0.0534
89
+ step 870000 | 7844s | train_loss=0.0534
90
+ step 880000 | 7928s | train_loss=0.0524
91
+ step 890000 | 8012s | train_loss=0.0527
92
+ step 900000 | 8160s | train_loss=0.0516 | fid=68.7992
93
+ step 910000 | 8241s | train_loss=0.0507
94
+ step 920000 | 8325s | train_loss=0.0507
95
+ step 930000 | 8409s | train_loss=0.0507
96
+ step 940000 | 8490s | train_loss=0.0496
97
+ step 950000 | 8574s | train_loss=0.0496
98
+ step 960000 | 8658s | train_loss=0.0495
99
+ step 970000 | 8743s | train_loss=0.0487
100
+ step 980000 | 8825s | train_loss=0.0484
101
+ step 990000 | 8909s | train_loss=0.0490
102
+ step 1000000 | 9058s | train_loss=0.0477 | fid=67.4563
103
+ step 1010000 | 9139s | train_loss=0.0477
104
+ step 1020000 | 9223s | train_loss=0.0476
105
+ step 1030000 | 9307s | train_loss=0.0464
106
+ step 1040000 | 9391s | train_loss=0.0460
107
+ step 1050000 | 9472s | train_loss=0.0463
108
+ step 1060000 | 9556s | train_loss=0.0462
109
+ step 1070000 | 9640s | train_loss=0.0458
110
+ step 1080000 | 9723s | train_loss=0.0447
111
+ step 1090000 | 9807s | train_loss=0.0449
112
+ step 1100000 | 9958s | train_loss=0.0443 | fid=68.7496
113
+ step 1110000 | 10041s | train_loss=0.0443
114
+ step 1120000 | 10123s | train_loss=0.0442
115
+ step 1130000 | 10207s | train_loss=0.0434
116
+ step 1140000 | 10291s | train_loss=0.0437
117
+ step 1150000 | 10372s | train_loss=0.0429
118
+ step 1160000 | 10456s | train_loss=0.0430
119
+ step 1170000 | 10540s | train_loss=0.0426
120
+ step 1180000 | 10624s | train_loss=0.0424
121
+ step 1190000 | 10707s | train_loss=0.0420
122
+ step 1200000 | 10855s | train_loss=0.0416 | fid=69.3845
123
+ step 1210000 | 10939s | train_loss=0.0414
124
+ step 1220000 | 11020s | train_loss=0.0413
125
+ step 1230000 | 11104s | train_loss=0.0417
126
+ step 1240000 | 11188s | train_loss=0.0408
127
+ step 1250000 | 11272s | train_loss=0.0401
128
+ step 1260000 | 11353s | train_loss=0.0401
129
+ step 1270000 | 11437s | train_loss=0.0396
130
+ step 1280000 | 11521s | train_loss=0.0399
131
+ step 1290000 | 11605s | train_loss=0.0392
132
+ step 1300000 | 11753s | train_loss=0.0391 | fid=68.4728
133
+ step 1310000 | 11837s | train_loss=0.0388
134
+ step 1320000 | 11919s | train_loss=0.0387
135
+ step 1330000 | 12003s | train_loss=0.0385
136
+ step 1340000 | 12087s | train_loss=0.0384
137
+ step 1350000 | 12171s | train_loss=0.0383
138
+ step 1360000 | 12252s | train_loss=0.0387
139
+ step 1370000 | 12336s | train_loss=0.0378
140
+ step 1380000 | 12420s | train_loss=0.0370
141
+ step 1390000 | 12504s | train_loss=0.0369
142
+ step 1400000 | 12654s | train_loss=0.0375 | fid=68.9206
143
+ step 1410000 | 12738s | train_loss=0.0370
144
+ step 1420000 | 12820s | train_loss=0.0365
145
+ step 1430000 | 12904s | train_loss=0.0364
146
+ step 1440000 | 12988s | train_loss=0.0366
147
+ step 1450000 | 13072s | train_loss=0.0359
148
+ step 1460000 | 13155s | train_loss=0.0360
149
+ step 1470000 | 13237s | train_loss=0.0359
150
+ step 1480000 | 13321s | train_loss=0.0355
151
+ step 1490000 | 13404s | train_loss=0.0358
152
+ step 1500000 | 13556s | train_loss=0.0354 | fid=68.2307
153
+ step 1510000 | 13639s | train_loss=0.0350
154
+ step 1520000 | 13720s | train_loss=0.0353
155
+ step 1530000 | 13804s | train_loss=0.0348
156
+ step 1540000 | 13888s | train_loss=0.0345
157
+ step 1550000 | 13972s | train_loss=0.0343
158
+ step 1560000 | 14056s | train_loss=0.0340
159
+ step 1570000 | 14137s | train_loss=0.0340
160
+ step 1580000 | 14221s | train_loss=0.0340
161
+ step 1590000 | 14302s | train_loss=0.0338
162
+ step 1600000 | 14455s | train_loss=0.0341 | fid=67.0740
163
+ step 1610000 | 14538s | train_loss=0.0338
164
+ step 1620000 | 14619s | train_loss=0.0335
165
+ step 1630000 | 14704s | train_loss=0.0336
166
+ step 1640000 | 14788s | train_loss=0.0332
167
+ step 1650000 | 14872s | train_loss=0.0335
168
+ step 1660000 | 14956s | train_loss=0.0330
169
+ step 1670000 | 15037s | train_loss=0.0330
170
+ step 1680000 | 15121s | train_loss=0.0324
171
+ step 1690000 | 15200s | train_loss=0.0327
172
+ step 1700000 | 15380s | train_loss=0.0322 | fid=67.5261
173
+ step 1710000 | 15465s | train_loss=0.0322
174
+ step 1720000 | 15531s | train_loss=0.0327
175
+ step 1730000 | 15615s | train_loss=0.0320
176
+ step 1740000 | 15699s | train_loss=0.0317
177
+ step 1750000 | 15783s | train_loss=0.0316
178
+ step 1760000 | 15867s | train_loss=0.0314
179
+ step 1770000 | 15947s | train_loss=0.0310
180
+ step 1780000 | 16022s | train_loss=0.0315
181
+ step 1790000 | 16099s | train_loss=0.0312
182
+ step 1800000 | 16253s | train_loss=0.0311 | fid=67.6975
183
+ step 1810000 | 16337s | train_loss=0.0309
184
+ step 1820000 | 16415s | train_loss=0.0309
185
+ step 1830000 | 16499s | train_loss=0.0310
186
+ step 1840000 | 16583s | train_loss=0.0305
187
+ step 1850000 | 16667s | train_loss=0.0308
188
+ step 1860000 | 16751s | train_loss=0.0310
189
+ step 1870000 | 16835s | train_loss=0.0306
190
+ step 1880000 | 16916s | train_loss=0.0303
191
+ step 1890000 | 16997s | train_loss=0.0300
192
+ step 1900000 | 17151s | train_loss=0.0299 | fid=67.6106
193
+ step 1910000 | 17235s | train_loss=0.0302
194
+ step 1920000 | 17318s | train_loss=0.0296
195
+ step 1930000 | 17399s | train_loss=0.0292
196
+ step 1940000 | 17483s | train_loss=0.0298
197
+ step 1950000 | 17566s | train_loss=0.0302
198
+ step 1960000 | 17650s | train_loss=0.0291
199
+ step 1970000 | 17734s | train_loss=0.0290
200
+ step 1980000 | 17818s | train_loss=0.0292
201
+ step 1990000 | 17895s | train_loss=0.0291
202
+ step 2000000 | 18049s | train_loss=0.0286 | fid=67.5222
203
+ step 2010000 | 18133s | train_loss=0.0287
204
+ step 2020000 | 18217s | train_loss=0.0287
205
+ step 2030000 | 18298s | train_loss=0.0282
206
+ step 2040000 | 18380s | train_loss=0.0284
207
+ step 2050000 | 18464s | train_loss=0.0286
208
+ step 2060000 | 18548s | train_loss=0.0283
209
+ step 2070000 | 18632s | train_loss=0.0279
210
+ step 2080000 | 18716s | train_loss=0.0281
211
+ step 2090000 | 18794s | train_loss=0.0279
212
+ step 2100000 | 18945s | train_loss=0.0277 | fid=67.9847
213
+ step 2110000 | 19029s | train_loss=0.0279
214
+ step 2120000 | 19113s | train_loss=0.0280
215
+ step 2130000 | 19194s | train_loss=0.0274
216
+ step 2140000 | 19279s | train_loss=0.0274
217
+ step 2150000 | 19361s | train_loss=0.0272
218
+ step 2160000 | 19445s | train_loss=0.0271
219
+ step 2170000 | 19528s | train_loss=0.0269
220
+ step 2180000 | 19612s | train_loss=0.0275
221
+ step 2190000 | 19690s | train_loss=0.0271
222
+ step 2200000 | 19841s | train_loss=0.0269 | fid=68.1450
223
+ step 2210000 | 19925s | train_loss=0.0266
224
+ step 2220000 | 20009s | train_loss=0.0266
225
+ step 2230000 | 20090s | train_loss=0.0268
226
+ step 2240000 | 20174s | train_loss=0.0266
227
+ step 2250000 | 20256s | train_loss=0.0265
228
+ step 2260000 | 20340s | train_loss=0.0263
229
+ step 2270000 | 20424s | train_loss=0.0262
230
+ step 2280000 | 20508s | train_loss=0.0265
231
+ step 2290000 | 20589s | train_loss=0.0264
232
+ step 2300000 | 20740s | train_loss=0.0262 | fid=67.1282
233
+ step 2310000 | 20824s | train_loss=0.0259
234
+ step 2320000 | 20908s | train_loss=0.0261
235
+ step 2330000 | 20989s | train_loss=0.0257
236
+ step 2340000 | 21073s | train_loss=0.0258
237
+ step 2350000 | 21157s | train_loss=0.0255
238
+ step 2360000 | 21239s | train_loss=0.0259
239
+ step 2370000 | 21323s | train_loss=0.0251
240
+ step 2380000 | 21407s | train_loss=0.0251
241
+ step 2390000 | 21487s | train_loss=0.0252
242
+ step 2400000 | 21636s | train_loss=0.0253 | fid=67.7781
243
+ step 2410000 | 21720s | train_loss=0.0249
244
+ step 2420000 | 21804s | train_loss=0.0251
245
+ step 2430000 | 21885s | train_loss=0.0250
246
+ step 2440000 | 21969s | train_loss=0.0246
247
+ step 2450000 | 22053s | train_loss=0.0247
248
+ step 2460000 | 22137s | train_loss=0.0244
249
+ step 2470000 | 22219s | train_loss=0.0243
250
+ step 2480000 | 22303s | train_loss=0.0246
251
+ step 2490000 | 22384s | train_loss=0.0243
252
+ step 2500000 | 22534s | train_loss=0.0243 | fid=66.5222
253
+ step 2510000 | 22618s | train_loss=0.0243
254
+ step 2520000 | 22702s | train_loss=0.0241
255
+ step 2530000 | 22783s | train_loss=0.0241
256
+ step 2540000 | 22867s | train_loss=0.0243
257
+ step 2550000 | 22951s | train_loss=0.0241
258
+ step 2560000 | 23035s | train_loss=0.0242
259
+ step 2570000 | 23118s | train_loss=0.0239
260
+ step 2580000 | 23200s | train_loss=0.0241
261
+ step 2590000 | 23284s | train_loss=0.0238
262
+ step 2600000 | 23439s | train_loss=0.0240 | fid=67.9495
263
+ step 2610000 | 23520s | train_loss=0.0238
264
+ step 2620000 | 23603s | train_loss=0.0237
265
+ step 2630000 | 23685s | train_loss=0.0236
266
+ step 2640000 | 23769s | train_loss=0.0232
267
+ step 2650000 | 23852s | train_loss=0.0234
268
+ step 2660000 | 23936s | train_loss=0.0232
269
+ step 2670000 | 24020s | train_loss=0.0230
270
+ step 2680000 | 24100s | train_loss=0.0229
271
+ step 2690000 | 24184s | train_loss=0.0230
272
+ step 2700000 | 24338s | train_loss=0.0232 | fid=68.3081
273
+ step 2710000 | 24419s | train_loss=0.0231
274
+ step 2720000 | 24503s | train_loss=0.0229
275
+ step 2730000 | 24583s | train_loss=0.0228
276
+ step 2740000 | 24667s | train_loss=0.0231
277
+ step 2750000 | 24751s | train_loss=0.0228
278
+ step 2760000 | 24835s | train_loss=0.0224
279
+ step 2770000 | 24919s | train_loss=0.0226
280
+ step 2780000 | 25001s | train_loss=0.0226
281
+ step 2790000 | 25083s | train_loss=0.0223
282
+ step 2800000 | 25236s | train_loss=0.0224 | fid=67.1816
283
+ step 2810000 | 25318s | train_loss=0.0222
284
+ step 2820000 | 25401s | train_loss=0.0220
285
+ step 2830000 | 25482s | train_loss=0.0223
286
+ step 2840000 | 25566s | train_loss=0.0218
287
+ step 2850000 | 25649s | train_loss=0.0219
288
+ step 2860000 | 25733s | train_loss=0.0218
289
+ step 2870000 | 25817s | train_loss=0.0223
290
+ step 2880000 | 25899s | train_loss=0.0217
291
+ step 2890000 | 25981s | train_loss=0.0217
292
+ step 2900000 | 26135s | train_loss=0.0216 | fid=66.6862
293
+ step 2910000 | 26219s | train_loss=0.0220
294
+ step 2920000 | 26299s | train_loss=0.0212
295
+ step 2930000 | 26381s | train_loss=0.0215
296
+ step 2940000 | 26464s | train_loss=0.0211
297
+ step 2950000 | 26548s | train_loss=0.0208
298
+ step 2960000 | 26632s | train_loss=0.0211
299
+ step 2970000 | 26716s | train_loss=0.0216
300
+ step 2980000 | 26798s | train_loss=0.0212
301
+ step 2990000 | 26882s | train_loss=0.0214
302
+ step 3000000 | 27033s | train_loss=0.0212 | fid=67.0902
303
+ step 3010000 | 27117s | train_loss=0.0211
304
+ step 3020000 | 27198s | train_loss=0.0209
305
+ step 3030000 | 27280s | train_loss=0.0213
306
+ step 3040000 | 27363s | train_loss=0.0209
307
+ step 3050000 | 27446s | train_loss=0.0213
308
+ step 3060000 | 27530s | train_loss=0.0213
309
+ step 3070000 | 27613s | train_loss=0.0208
310
+ step 3080000 | 27696s | train_loss=0.0210
311
+ step 3090000 | 27780s | train_loss=0.0209
312
+ step 3100000 | 27931s | train_loss=0.0206 | fid=68.7723
313
+ step 3110000 | 28015s | train_loss=0.0209
314
+ step 3120000 | 28099s | train_loss=0.0205
315
+ step 3130000 | 28181s | train_loss=0.0203
316
+ step 3140000 | 28262s | train_loss=0.0205
317
+ step 3150000 | 28346s | train_loss=0.0204
318
+ step 3160000 | 28430s | train_loss=0.0204
319
+ step 3170000 | 28511s | train_loss=0.0203
320
+ step 3180000 | 28594s | train_loss=0.0203
321
+ step 3190000 | 28677s | train_loss=0.0201
322
+ step 3200000 | 28827s | train_loss=0.0197 | fid=67.4118
323
+ step 3210000 | 28911s | train_loss=0.0200
324
+ step 3220000 | 28995s | train_loss=0.0201
325
+ step 3230000 | 29077s | train_loss=0.0200
326
+ step 3240000 | 29158s | train_loss=0.0199
327
+ step 3250000 | 29242s | train_loss=0.0203
328
+ step 3260000 | 29326s | train_loss=0.0198
329
+ step 3270000 | 29408s | train_loss=0.0196
330
+ step 3280000 | 29491s | train_loss=0.0198
331
+ step 3290000 | 29574s | train_loss=0.0196
332
+ step 3300000 | 29726s | train_loss=0.0200 | fid=66.7757
333
+ step 3310000 | 29809s | train_loss=0.0198
334
+ step 3320000 | 29893s | train_loss=0.0198
335
+ step 3330000 | 29977s | train_loss=0.0196
336
+ step 3340000 | 30056s | train_loss=0.0197
337
+ step 3350000 | 30140s | train_loss=0.0197
338
+ step 3360000 | 30224s | train_loss=0.0197
339
+ step 3370000 | 30306s | train_loss=0.0197
340
+ step 3380000 | 30390s | train_loss=0.0192
341
+ step 3390000 | 30474s | train_loss=0.0193
342
+ step 3400000 | 30625s | train_loss=0.0194 | fid=67.0931
343
+ step 3410000 | 30709s | train_loss=0.0191
344
+ step 3420000 | 30792s | train_loss=0.0193
345
+ step 3430000 | 30876s | train_loss=0.0194
346
+ step 3440000 | 30939s | train_loss=0.0193
347
+ step 3450000 | 31019s | train_loss=0.0194
348
+ step 3460000 | 31103s | train_loss=0.0188
349
+ step 3470000 | 31186s | train_loss=0.0192
350
+ step 3480000 | 31270s | train_loss=0.0187
351
+ step 3490000 | 31354s | train_loss=0.0192
352
+ step 3500000 | 31552s | train_loss=0.0187 | fid=67.3352
353
+ step 3510000 | 31637s | train_loss=0.0190
354
+ step 3520000 | 31719s | train_loss=0.0193
355
+ step 3530000 | 31803s | train_loss=0.0188
356
+ step 3540000 | 31875s | train_loss=0.0189
357
+ step 3550000 | 31957s | train_loss=0.0184
358
+ step 3560000 | 32040s | train_loss=0.0186
359
+ step 3570000 | 32124s | train_loss=0.0184
360
+ step 3580000 | 32208s | train_loss=0.0185
361
+ step 3590000 | 32291s | train_loss=0.0186
362
+ step 3600000 | 32445s | train_loss=0.0185 | fid=67.9619
363
+ step 3610000 | 32529s | train_loss=0.0185
364
+ step 3620000 | 32612s | train_loss=0.0185
365
+ step 3630000 | 32696s | train_loss=0.0184
366
+ step 3640000 | 32778s | train_loss=0.0182
367
+ step 3650000 | 32859s | train_loss=0.0184
368
+ step 3660000 | 32941s | train_loss=0.0187
369
+ step 3670000 | 33025s | train_loss=0.0182
370
+ step 3680000 | 33108s | train_loss=0.0182
371
+ step 3690000 | 33192s | train_loss=0.0179
372
+ step 3700000 | 33343s | train_loss=0.0180 | fid=67.3683
373
+ step 3710000 | 33427s | train_loss=0.0180
374
+ step 3720000 | 33511s | train_loss=0.0180
375
+ step 3730000 | 33594s | train_loss=0.0180
376
+ step 3740000 | 33675s | train_loss=0.0180
377
+ step 3750000 | 33756s | train_loss=0.0181
378
+ step 3760000 | 33830s | train_loss=0.0180
379
+ step 3770000 | 33914s | train_loss=0.0181
380
+ step 3780000 | 33998s | train_loss=0.0176
381
+ step 3790000 | 34082s | train_loss=0.0178
382
+ step 3800000 | 34255s | train_loss=0.0178 | fid=68.2178
383
+ step 3810000 | 34339s | train_loss=0.0177
384
+ step 3820000 | 34423s | train_loss=0.0173
385
+ step 3830000 | 34508s | train_loss=0.0177
386
+ step 3840000 | 34580s | train_loss=0.0173
387
+ step 3850000 | 34661s | train_loss=0.0176
388
+ step 3860000 | 34735s | train_loss=0.0175
389
+ step 3870000 | 34814s | train_loss=0.0172
390
+ step 3880000 | 34898s | train_loss=0.0176
391
+ step 3890000 | 34982s | train_loss=0.0176
392
+ step 3900000 | 35181s | train_loss=0.0174 | fid=67.5382
393
+ step 3910000 | 35265s | train_loss=0.0171
394
+ step 3920000 | 35349s | train_loss=0.0175
395
+ step 3930000 | 35433s | train_loss=0.0171
396
+ step 3940000 | 35505s | train_loss=0.0175
397
+ step 3950000 | 35589s | train_loss=0.0175
398
+ step 3960000 | 35671s | train_loss=0.0171
399
+ step 3970000 | 35746s | train_loss=0.0172
400
+ step 3980000 | 35830s | train_loss=0.0170
401
+ step 3990000 | 35914s | train_loss=0.0167
402
+ step 4000000 | 36106s | train_loss=0.0170 | fid=67.4431
403
+ step 4010000 | 36190s | train_loss=0.0169
404
+ step 4020000 | 36274s | train_loss=0.0171
405
+ step 4030000 | 36359s | train_loss=0.0168
406
+ step 4040000 | 36425s | train_loss=0.0168
407
+ step 4050000 | 36508s | train_loss=0.0170
408
+ step 4060000 | 36592s | train_loss=0.0167
409
+ step 4070000 | 36663s | train_loss=0.0169
410
+ step 4080000 | 36746s | train_loss=0.0168
411
+ step 4090000 | 36830s | train_loss=0.0168
412
+ step 4100000 | 37026s | train_loss=0.0167 | fid=66.8059
413
+ step 4110000 | 37110s | train_loss=0.0169
414
+ step 4120000 | 37194s | train_loss=0.0165
415
+ step 4130000 | 37276s | train_loss=0.0165
416
+ step 4140000 | 37347s | train_loss=0.0164
417
+ step 4150000 | 37428s | train_loss=0.0165
418
+ step 4160000 | 37512s | train_loss=0.0167
419
+ step 4170000 | 37591s | train_loss=0.0163
420
+ step 4180000 | 37666s | train_loss=0.0165
421
+ step 4190000 | 37750s | train_loss=0.0166
422
+ step 4200000 | 37945s | train_loss=0.0163 | fid=67.4547
423
+ step 4210000 | 38030s | train_loss=0.0162
424
+ step 4220000 | 38112s | train_loss=0.0164
425
+ step 4230000 | 38196s | train_loss=0.0161
426
+ step 4240000 | 38268s | train_loss=0.0161
427
+ step 4250000 | 38346s | train_loss=0.0163
428
+ step 4260000 | 38430s | train_loss=0.0160
429
+ step 4270000 | 38515s | train_loss=0.0161
430
+ step 4280000 | 38588s | train_loss=0.0164
431
+ step 4290000 | 38672s | train_loss=0.0160
432
+ step 4300000 | 38873s | train_loss=0.0163 | fid=67.9438
433
+ step 4310000 | 38956s | train_loss=0.0161
434
+ step 4320000 | 39040s | train_loss=0.0162
435
+ step 4330000 | 39124s | train_loss=0.0160
436
+ step 4340000 | 39199s | train_loss=0.0161
437
+ step 4350000 | 39275s | train_loss=0.0158
438
+ step 4360000 | 39359s | train_loss=0.0159
439
+ step 4370000 | 39443s | train_loss=0.0162
440
+ step 4380000 | 39519s | train_loss=0.0161
441
+ step 4390000 | 39603s | train_loss=0.0159
442
+ step 4400000 | 39785s | train_loss=0.0161 | fid=67.3423
443
+ step 4410000 | 39867s | train_loss=0.0157
444
+ step 4420000 | 39951s | train_loss=0.0157
445
+ step 4430000 | 40035s | train_loss=0.0159
446
+ step 4440000 | 40119s | train_loss=0.0156
447
+ step 4450000 | 40187s | train_loss=0.0159
448
+ step 4460000 | 40271s | train_loss=0.0158
449
+ step 4470000 | 40355s | train_loss=0.0154
450
+ step 4480000 | 40434s | train_loss=0.0156
451
+ step 4490000 | 40513s | train_loss=0.0157
452
+ step 4500000 | 40694s | train_loss=0.0159 | fid=66.4418
453
+ step 4510000 | 40778s | train_loss=0.0158
454
+ step 4520000 | 40862s | train_loss=0.0156
455
+ step 4530000 | 40946s | train_loss=0.0156
456
+ step 4540000 | 41030s | train_loss=0.0154
457
+ step 4550000 | 41100s | train_loss=0.0155
458
+ step 4560000 | 41182s | train_loss=0.0157
459
+ step 4570000 | 41266s | train_loss=0.0155
460
+ step 4580000 | 41349s | train_loss=0.0154
461
+ step 4590000 | 41426s | train_loss=0.0152
462
+ step 4600000 | 41588s | train_loss=0.0154 | fid=67.2044
463
+ step 4610000 | 41672s | train_loss=0.0156
464
+ step 4620000 | 41756s | train_loss=0.0157
465
+ step 4630000 | 41839s | train_loss=0.0153
466
+ step 4640000 | 41923s | train_loss=0.0152
467
+ step 4650000 | 42001s | train_loss=0.0155
468
+ step 4660000 | 42085s | train_loss=0.0152
469
+ step 4670000 | 42168s | train_loss=0.0151
470
+ step 4680000 | 42252s | train_loss=0.0150
471
+ step 4690000 | 42331s | train_loss=0.0150
472
+ step 4700000 | 42483s | train_loss=0.0150 | fid=67.4073
473
+ step 4710000 | 42566s | train_loss=0.0153
474
+ step 4720000 | 42649s | train_loss=0.0155
475
+ step 4730000 | 42733s | train_loss=0.0153
476
+ step 4740000 | 42816s | train_loss=0.0152
477
+ step 4750000 | 42898s | train_loss=0.0152
478
+ step 4760000 | 42980s | train_loss=0.0150
479
+ step 4770000 | 43064s | train_loss=0.0148
480
+ step 4780000 | 43147s | train_loss=0.0152
481
+ step 4790000 | 43231s | train_loss=0.0148
482
+ step 4800000 | 43379s | train_loss=0.0153 | fid=67.8355
483
+ step 4810000 | 43463s | train_loss=0.0149
484
+ step 4820000 | 43546s | train_loss=0.0148
485
+ step 4830000 | 43630s | train_loss=0.0146
486
+ step 4840000 | 43713s | train_loss=0.0149
487
+ step 4850000 | 43795s | train_loss=0.0149
488
+ step 4860000 | 43877s | train_loss=0.0152
489
+ step 4870000 | 43960s | train_loss=0.0147
490
+ step 4880000 | 44044s | train_loss=0.0146
491
+ step 4890000 | 44127s | train_loss=0.0147
492
+ step 4900000 | 44253s | train_loss=0.0147 | fid=67.4834
493
+ step 4910000 | 44320s | train_loss=0.0147
494
+ step 4920000 | 44387s | train_loss=0.0146
495
+ step 4930000 | 44453s | train_loss=0.0147
496
+ step 4940000 | 44520s | train_loss=0.0146
497
+ step 4950000 | 44585s | train_loss=0.0147
498
+ step 4960000 | 44652s | train_loss=0.0148
499
+ step 4970000 | 44717s | train_loss=0.0143
500
+ step 4980000 | 44784s | train_loss=0.0144
501
+ step 4990000 | 44851s | train_loss=0.0145
502
+ step 5000000 | 44969s | train_loss=0.0144 | fid=67.6126
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d200/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d200/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 73s | train_loss=0.9954 | fid=305.3385
3
+ step 10000 | 162s | train_loss=0.6266
4
+ step 20000 | 255s | train_loss=0.5169
5
+ step 30000 | 347s | train_loss=0.4636
6
+ step 40000 | 439s | train_loss=0.4202
7
+ step 50000 | 532s | train_loss=0.3687
8
+ step 60000 | 624s | train_loss=0.3167
9
+ step 70000 | 717s | train_loss=0.2722
10
+ step 80000 | 809s | train_loss=0.2351
11
+ step 90000 | 902s | train_loss=0.2073
12
+ step 100000 | 1061s | train_loss=0.1861 | fid=65.6381
13
+ step 110000 | 1143s | train_loss=0.1713
14
+ step 120000 | 1235s | train_loss=0.1553
15
+ step 130000 | 1328s | train_loss=0.1455
16
+ step 140000 | 1420s | train_loss=0.1370
17
+ step 150000 | 1513s | train_loss=0.1302
18
+ step 160000 | 1605s | train_loss=0.1268
19
+ step 170000 | 1698s | train_loss=0.1201
20
+ step 180000 | 1790s | train_loss=0.1143
21
+ step 190000 | 1883s | train_loss=0.1093
22
+ step 200000 | 2042s | train_loss=0.1058 | fid=73.5089
23
+ step 210000 | 2136s | train_loss=0.1051
24
+ step 220000 | 2221s | train_loss=0.1014
25
+ step 230000 | 2313s | train_loss=0.0981
26
+ step 240000 | 2405s | train_loss=0.1003
27
+ step 250000 | 2498s | train_loss=0.0981
28
+ step 260000 | 2590s | train_loss=0.0921
29
+ step 270000 | 2683s | train_loss=0.0879
30
+ step 280000 | 2775s | train_loss=0.0874
31
+ step 290000 | 2866s | train_loss=0.0878
32
+ step 300000 | 3028s | train_loss=0.0893 | fid=82.7076
33
+ step 310000 | 3120s | train_loss=0.0830
34
+ step 320000 | 3208s | train_loss=0.0809
35
+ step 330000 | 3300s | train_loss=0.0806
36
+ step 340000 | 3393s | train_loss=0.0782
37
+ step 350000 | 3485s | train_loss=0.0852
38
+ step 360000 | 3578s | train_loss=0.0771
39
+ step 370000 | 3670s | train_loss=0.0776
40
+ step 380000 | 3763s | train_loss=0.0785
41
+ step 390000 | 3853s | train_loss=0.0735
42
+ step 400000 | 4015s | train_loss=0.0726 | fid=87.7442
43
+ step 410000 | 4108s | train_loss=0.0721
44
+ step 420000 | 4202s | train_loss=0.0706
45
+ step 430000 | 4288s | train_loss=0.0725
46
+ step 440000 | 4380s | train_loss=0.0687
47
+ step 450000 | 4473s | train_loss=0.0688
48
+ step 460000 | 4565s | train_loss=0.0704
49
+ step 470000 | 4658s | train_loss=0.0673
50
+ step 480000 | 4748s | train_loss=0.0669
51
+ step 490000 | 4841s | train_loss=0.0682
52
+ step 500000 | 5001s | train_loss=0.0642 | fid=92.0288
53
+ step 510000 | 5094s | train_loss=0.0660
54
+ step 520000 | 5187s | train_loss=0.0637
55
+ step 530000 | 5274s | train_loss=0.0659
56
+ step 540000 | 5367s | train_loss=0.0618
57
+ step 550000 | 5459s | train_loss=0.0629
58
+ step 560000 | 5552s | train_loss=0.0622
59
+ step 570000 | 5644s | train_loss=0.0616
60
+ step 580000 | 5735s | train_loss=0.0649
61
+ step 590000 | 5827s | train_loss=0.0605
62
+ step 600000 | 5988s | train_loss=0.0614 | fid=102.2590
63
+ step 610000 | 6081s | train_loss=0.0593
64
+ step 620000 | 6173s | train_loss=0.0587
65
+ step 630000 | 6267s | train_loss=0.0614
66
+ step 640000 | 6353s | train_loss=0.0558
67
+ step 650000 | 6446s | train_loss=0.0635
68
+ step 660000 | 6538s | train_loss=0.0616
69
+ step 670000 | 6629s | train_loss=0.0553
70
+ step 680000 | 6721s | train_loss=0.0561
71
+ step 690000 | 6814s | train_loss=0.0576
72
+ step 700000 | 6975s | train_loss=0.0558 | fid=95.9475
73
+ step 710000 | 7067s | train_loss=0.0515
74
+ step 720000 | 7160s | train_loss=0.0552
75
+ step 730000 | 7252s | train_loss=0.0554
76
+ step 740000 | 7341s | train_loss=0.0573
77
+ step 750000 | 7433s | train_loss=0.0541
78
+ step 760000 | 7525s | train_loss=0.0526
79
+ step 770000 | 7616s | train_loss=0.0564
80
+ step 780000 | 7709s | train_loss=0.0539
81
+ step 790000 | 7801s | train_loss=0.0520
82
+ step 800000 | 7963s | train_loss=0.0503 | fid=95.8307
83
+ step 810000 | 8055s | train_loss=0.0552
84
+ step 820000 | 8148s | train_loss=0.0540
85
+ step 830000 | 8240s | train_loss=0.0508
86
+ step 840000 | 8334s | train_loss=0.0520
87
+ step 850000 | 8422s | train_loss=0.0515
88
+ step 860000 | 8512s | train_loss=0.0475
89
+ step 870000 | 8605s | train_loss=0.0491
90
+ step 880000 | 8697s | train_loss=0.0479
91
+ step 890000 | 8790s | train_loss=0.0479
92
+ step 900000 | 8952s | train_loss=0.0479 | fid=99.8701
93
+ step 910000 | 9044s | train_loss=0.0480
94
+ step 920000 | 9137s | train_loss=0.0473
95
+ step 930000 | 9229s | train_loss=0.0504
96
+ step 940000 | 9322s | train_loss=0.0518
97
+ step 950000 | 9410s | train_loss=0.0499
98
+ step 960000 | 9500s | train_loss=0.0454
99
+ step 970000 | 9593s | train_loss=0.0442
100
+ step 980000 | 9685s | train_loss=0.0693
101
+ step 990000 | 9778s | train_loss=0.0457
102
+ step 1000000 | 9940s | train_loss=0.0455 | fid=92.7281
103
+ step 1010000 | 10032s | train_loss=0.0445
104
+ step 1020000 | 10125s | train_loss=0.0460
105
+ step 1030000 | 10217s | train_loss=0.0468
106
+ step 1040000 | 10310s | train_loss=0.0462
107
+ step 1050000 | 10400s | train_loss=0.0455
108
+ step 1060000 | 10488s | train_loss=0.0437
109
+ step 1070000 | 10580s | train_loss=0.0455
110
+ step 1080000 | 10673s | train_loss=0.0448
111
+ step 1090000 | 10765s | train_loss=0.0455
112
+ step 1100000 | 10927s | train_loss=0.0446 | fid=89.3890
113
+ step 1110000 | 11020s | train_loss=0.0415
114
+ step 1120000 | 11112s | train_loss=0.0441
115
+ step 1130000 | 11205s | train_loss=0.0412
116
+ step 1140000 | 11297s | train_loss=0.0422
117
+ step 1150000 | 11388s | train_loss=0.0419
118
+ step 1160000 | 11477s | train_loss=0.0426
119
+ step 1170000 | 11568s | train_loss=0.0421
120
+ step 1180000 | 11661s | train_loss=0.0432
121
+ step 1190000 | 11753s | train_loss=0.0398
122
+ step 1200000 | 11915s | train_loss=0.0426 | fid=85.9814
123
+ step 1210000 | 12007s | train_loss=0.0396
124
+ step 1220000 | 12100s | train_loss=0.0403
125
+ step 1230000 | 12192s | train_loss=0.0410
126
+ step 1240000 | 12283s | train_loss=0.0412
127
+ step 1250000 | 12376s | train_loss=0.0397
128
+ step 1260000 | 12468s | train_loss=0.0402
129
+ step 1270000 | 12556s | train_loss=0.0436
130
+ step 1280000 | 12649s | train_loss=0.0409
131
+ step 1290000 | 12741s | train_loss=0.0429
132
+ step 1300000 | 12902s | train_loss=0.0463 | fid=83.9533
133
+ step 1310000 | 12995s | train_loss=0.0402
134
+ step 1320000 | 13087s | train_loss=0.0447
135
+ step 1330000 | 13180s | train_loss=0.0390
136
+ step 1340000 | 13270s | train_loss=0.0384
137
+ step 1350000 | 13363s | train_loss=0.0406
138
+ step 1360000 | 13455s | train_loss=0.0419
139
+ step 1370000 | 13545s | train_loss=0.0397
140
+ step 1380000 | 13636s | train_loss=0.0377
141
+ step 1390000 | 13728s | train_loss=0.0431
142
+ step 1400000 | 13890s | train_loss=0.0402 | fid=77.5910
143
+ step 1410000 | 13983s | train_loss=0.0381
144
+ step 1420000 | 14076s | train_loss=0.0393
145
+ step 1430000 | 14166s | train_loss=0.0415
146
+ step 1440000 | 14259s | train_loss=0.0409
147
+ step 1450000 | 14351s | train_loss=0.0372
148
+ step 1460000 | 14444s | train_loss=0.0384
149
+ step 1470000 | 14536s | train_loss=0.0375
150
+ step 1480000 | 14625s | train_loss=0.0368
151
+ step 1490000 | 14717s | train_loss=0.0381
152
+ step 1500000 | 14879s | train_loss=0.0380 | fid=76.7936
153
+ step 1510000 | 14972s | train_loss=0.0372
154
+ step 1520000 | 15064s | train_loss=0.0415
155
+ step 1530000 | 15155s | train_loss=0.0408
156
+ step 1540000 | 15247s | train_loss=0.0367
157
+ step 1550000 | 15340s | train_loss=0.0339
158
+ step 1560000 | 15432s | train_loss=0.0357
159
+ step 1570000 | 15525s | train_loss=0.0360
160
+ step 1580000 | 15615s | train_loss=0.0349
161
+ step 1590000 | 15705s | train_loss=0.0349
162
+ step 1600000 | 15867s | train_loss=0.0386 | fid=68.7086
163
+ step 1610000 | 15960s | train_loss=0.0368
164
+ step 1620000 | 16051s | train_loss=0.0352
165
+ step 1630000 | 16143s | train_loss=0.0388
166
+ step 1640000 | 16236s | train_loss=0.0362
167
+ step 1650000 | 16328s | train_loss=0.0371
168
+ step 1660000 | 16421s | train_loss=0.0373
169
+ step 1670000 | 16513s | train_loss=0.0333
170
+ step 1680000 | 16605s | train_loss=0.0368
171
+ step 1690000 | 16694s | train_loss=0.0343
172
+ step 1700000 | 16856s | train_loss=0.0340 | fid=67.3148
173
+ step 1710000 | 16949s | train_loss=0.0354
174
+ step 1720000 | 17039s | train_loss=0.0392
175
+ step 1730000 | 17132s | train_loss=0.0360
176
+ step 1740000 | 17224s | train_loss=0.0358
177
+ step 1750000 | 17317s | train_loss=0.0331
178
+ step 1760000 | 17409s | train_loss=0.0317
179
+ step 1770000 | 17502s | train_loss=0.0344
180
+ step 1780000 | 17594s | train_loss=0.0348
181
+ step 1790000 | 17684s | train_loss=0.0426
182
+ step 1800000 | 17843s | train_loss=0.0334 | fid=64.3976
183
+ step 1810000 | 17934s | train_loss=0.0397
184
+ step 1820000 | 18026s | train_loss=0.0334
185
+ step 1830000 | 18119s | train_loss=0.0350
186
+ step 1840000 | 18211s | train_loss=0.0331
187
+ step 1850000 | 18304s | train_loss=0.0311
188
+ step 1860000 | 18396s | train_loss=0.0324
189
+ step 1870000 | 18489s | train_loss=0.0322
190
+ step 1880000 | 18581s | train_loss=0.0329
191
+ step 1890000 | 18673s | train_loss=0.0345
192
+ step 1900000 | 18832s | train_loss=0.0324 | fid=62.2532
193
+ step 1910000 | 18922s | train_loss=0.0329
194
+ step 1920000 | 19015s | train_loss=0.0346
195
+ step 1930000 | 19107s | train_loss=0.0315
196
+ step 1940000 | 19200s | train_loss=0.0333
197
+ step 1950000 | 19292s | train_loss=0.0341
198
+ step 1960000 | 19385s | train_loss=0.0321
199
+ step 1970000 | 19477s | train_loss=0.0347
200
+ step 1980000 | 19570s | train_loss=0.0359
201
+ step 1990000 | 19663s | train_loss=0.0328
202
+ step 2000000 | 19820s | train_loss=0.0330 | fid=61.0235
203
+ step 2010000 | 19912s | train_loss=0.0341
204
+ step 2020000 | 20005s | train_loss=0.0327
205
+ step 2030000 | 20097s | train_loss=0.0299
206
+ step 2040000 | 20190s | train_loss=0.0315
207
+ step 2050000 | 20282s | train_loss=0.0328
208
+ step 2060000 | 20375s | train_loss=0.0297
209
+ step 2070000 | 20467s | train_loss=0.0339
210
+ step 2080000 | 20559s | train_loss=0.0308
211
+ step 2090000 | 20652s | train_loss=0.0295
212
+ step 2100000 | 20810s | train_loss=0.0315 | fid=62.6059
213
+ step 2110000 | 20901s | train_loss=0.0305
214
+ step 2120000 | 20993s | train_loss=0.0310
215
+ step 2130000 | 21086s | train_loss=0.0321
216
+ step 2140000 | 21178s | train_loss=0.0301
217
+ step 2150000 | 21271s | train_loss=0.0279
218
+ step 2160000 | 21363s | train_loss=0.0335
219
+ step 2170000 | 21456s | train_loss=0.0312
220
+ step 2180000 | 21548s | train_loss=0.0299
221
+ step 2190000 | 21641s | train_loss=0.0312
222
+ step 2200000 | 21801s | train_loss=0.0305 | fid=59.6343
223
+ step 2210000 | 21890s | train_loss=0.0340
224
+ step 2220000 | 21983s | train_loss=0.0287
225
+ step 2230000 | 22075s | train_loss=0.0307
226
+ step 2240000 | 22168s | train_loss=0.0312
227
+ step 2250000 | 22260s | train_loss=0.0299
228
+ step 2260000 | 22353s | train_loss=0.0289
229
+ step 2270000 | 22445s | train_loss=0.0319
230
+ step 2280000 | 22538s | train_loss=0.0298
231
+ step 2290000 | 22630s | train_loss=0.0287
232
+ step 2300000 | 22790s | train_loss=0.0301 | fid=58.1248
233
+ step 2310000 | 22879s | train_loss=0.0292
234
+ step 2320000 | 22970s | train_loss=0.0324
235
+ step 2330000 | 23063s | train_loss=0.0284
236
+ step 2340000 | 23155s | train_loss=0.0268
237
+ step 2350000 | 23248s | train_loss=0.0295
238
+ step 2360000 | 23341s | train_loss=0.0298
239
+ step 2370000 | 23433s | train_loss=0.0312
240
+ step 2380000 | 23525s | train_loss=0.0314
241
+ step 2390000 | 23616s | train_loss=0.0277
242
+ step 2400000 | 23777s | train_loss=0.0270 | fid=58.5284
243
+ step 2410000 | 23869s | train_loss=0.0288
244
+ step 2420000 | 23958s | train_loss=0.0278
245
+ step 2430000 | 24050s | train_loss=0.0297
246
+ step 2440000 | 24143s | train_loss=0.0290
247
+ step 2450000 | 24235s | train_loss=0.0264
248
+ step 2460000 | 24328s | train_loss=0.0282
249
+ step 2470000 | 24420s | train_loss=0.0279
250
+ step 2480000 | 24513s | train_loss=0.0276
251
+ step 2490000 | 24604s | train_loss=0.0304
252
+ step 2500000 | 24765s | train_loss=0.0295 | fid=55.1360
253
+ step 2510000 | 24858s | train_loss=0.0261
254
+ step 2520000 | 24947s | train_loss=0.0288
255
+ step 2530000 | 25038s | train_loss=0.0294
256
+ step 2540000 | 25131s | train_loss=0.0257
257
+ step 2550000 | 25223s | train_loss=0.0277
258
+ step 2560000 | 25316s | train_loss=0.0274
259
+ step 2570000 | 25408s | train_loss=0.0272
260
+ step 2580000 | 25499s | train_loss=0.0261
261
+ step 2590000 | 25592s | train_loss=0.0290
262
+ step 2600000 | 25753s | train_loss=0.0275 | fid=56.3788
263
+ step 2610000 | 25846s | train_loss=0.0287
264
+ step 2620000 | 25938s | train_loss=0.0287
265
+ step 2630000 | 26027s | train_loss=0.0248
266
+ step 2640000 | 26120s | train_loss=0.0295
267
+ step 2650000 | 26212s | train_loss=0.0289
268
+ step 2660000 | 26305s | train_loss=0.0280
269
+ step 2670000 | 26397s | train_loss=0.0269
270
+ step 2680000 | 26488s | train_loss=0.0260
271
+ step 2690000 | 26581s | train_loss=0.0306
272
+ step 2700000 | 26742s | train_loss=0.0267 | fid=55.5792
273
+ step 2710000 | 26834s | train_loss=0.0278
274
+ step 2720000 | 26927s | train_loss=0.0267
275
+ step 2730000 | 27017s | train_loss=0.0259
276
+ step 2740000 | 27108s | train_loss=0.0251
277
+ step 2750000 | 27200s | train_loss=0.0261
278
+ step 2760000 | 27293s | train_loss=0.0267
279
+ step 2770000 | 27384s | train_loss=0.0270
280
+ step 2780000 | 27476s | train_loss=0.0265
281
+ step 2790000 | 27568s | train_loss=0.0260
282
+ step 2800000 | 27731s | train_loss=0.0264 | fid=52.1749
283
+ step 2810000 | 27823s | train_loss=0.0264
284
+ step 2820000 | 27916s | train_loss=0.0248
285
+ step 2830000 | 28008s | train_loss=0.0267
286
+ step 2840000 | 28097s | train_loss=0.0256
287
+ step 2850000 | 28189s | train_loss=0.0253
288
+ step 2860000 | 28281s | train_loss=0.0261
289
+ step 2870000 | 28372s | train_loss=0.0262
290
+ step 2880000 | 28465s | train_loss=0.0295
291
+ step 2890000 | 28557s | train_loss=0.0253
292
+ step 2900000 | 28719s | train_loss=0.0316 | fid=49.1377
293
+ step 2910000 | 28812s | train_loss=0.0270
294
+ step 2920000 | 28904s | train_loss=0.0241
295
+ step 2930000 | 28997s | train_loss=0.0261
296
+ step 2940000 | 29087s | train_loss=0.0291
297
+ step 2950000 | 29178s | train_loss=0.0250
298
+ step 2960000 | 29268s | train_loss=0.0251
299
+ step 2970000 | 29361s | train_loss=0.0270
300
+ step 2980000 | 29453s | train_loss=0.0261
301
+ step 2990000 | 29546s | train_loss=0.0235
302
+ step 3000000 | 29708s | train_loss=0.0254 | fid=51.5600
303
+ step 3010000 | 29801s | train_loss=0.0252
304
+ step 3020000 | 29893s | train_loss=0.0256
305
+ step 3030000 | 29986s | train_loss=0.0233
306
+ step 3040000 | 30078s | train_loss=0.0265
307
+ step 3050000 | 30168s | train_loss=0.0239
308
+ step 3060000 | 30257s | train_loss=0.0259
309
+ step 3070000 | 30350s | train_loss=0.0259
310
+ step 3080000 | 30442s | train_loss=0.0239
311
+ step 3090000 | 30534s | train_loss=0.0243
312
+ step 3100000 | 30696s | train_loss=0.0233 | fid=54.0866
313
+ step 3110000 | 30789s | train_loss=0.0260
314
+ step 3120000 | 30882s | train_loss=0.0238
315
+ step 3130000 | 30974s | train_loss=0.0244
316
+ step 3140000 | 31067s | train_loss=0.0267
317
+ step 3150000 | 31154s | train_loss=0.0236
318
+ step 3160000 | 31245s | train_loss=0.0236
319
+ step 3170000 | 31337s | train_loss=0.0259
320
+ step 3180000 | 31430s | train_loss=0.0254
321
+ step 3190000 | 31522s | train_loss=0.0243
322
+ step 3200000 | 31684s | train_loss=0.0273 | fid=50.2156
323
+ step 3210000 | 31777s | train_loss=0.0251
324
+ step 3220000 | 31870s | train_loss=0.0261
325
+ step 3230000 | 31962s | train_loss=0.0288
326
+ step 3240000 | 32055s | train_loss=0.0247
327
+ step 3250000 | 32144s | train_loss=0.0252
328
+ step 3260000 | 32235s | train_loss=0.0246
329
+ step 3270000 | 32326s | train_loss=0.0279
330
+ step 3280000 | 32418s | train_loss=0.0221
331
+ step 3290000 | 32510s | train_loss=0.0227
332
+ step 3300000 | 32672s | train_loss=0.0256 | fid=48.4332
333
+ step 3310000 | 32765s | train_loss=0.0270
334
+ step 3320000 | 32857s | train_loss=0.0263
335
+ step 3330000 | 32950s | train_loss=0.0225
336
+ step 3340000 | 33040s | train_loss=0.0230
337
+ step 3350000 | 33133s | train_loss=0.0260
338
+ step 3360000 | 33222s | train_loss=0.0235
339
+ step 3370000 | 33313s | train_loss=0.0257
340
+ step 3380000 | 33406s | train_loss=0.0230
341
+ step 3390000 | 33498s | train_loss=0.0246
342
+ step 3400000 | 33660s | train_loss=0.0248 | fid=50.5835
343
+ step 3410000 | 33753s | train_loss=0.0224
344
+ step 3420000 | 33845s | train_loss=0.0236
345
+ step 3430000 | 33938s | train_loss=0.0245
346
+ step 3440000 | 34029s | train_loss=0.0238
347
+ step 3450000 | 34121s | train_loss=0.0228
348
+ step 3460000 | 34212s | train_loss=0.0243
349
+ step 3470000 | 34303s | train_loss=0.0245
350
+ step 3480000 | 34393s | train_loss=0.0223
351
+ step 3490000 | 34486s | train_loss=0.0239
352
+ step 3500000 | 34647s | train_loss=0.0248 | fid=50.8862
353
+ step 3510000 | 34739s | train_loss=0.0234
354
+ step 3520000 | 34832s | train_loss=0.0226
355
+ step 3530000 | 34922s | train_loss=0.0227
356
+ step 3540000 | 35015s | train_loss=0.0240
357
+ step 3550000 | 35107s | train_loss=0.0237
358
+ step 3560000 | 35200s | train_loss=0.0275
359
+ step 3570000 | 35289s | train_loss=0.0214
360
+ step 3580000 | 35379s | train_loss=0.0253
361
+ step 3590000 | 35472s | train_loss=0.0233
362
+ step 3600000 | 35634s | train_loss=0.0232 | fid=48.4399
363
+ step 3610000 | 35726s | train_loss=0.0236
364
+ step 3620000 | 35819s | train_loss=0.0234
365
+ step 3630000 | 35910s | train_loss=0.0234
366
+ step 3640000 | 36002s | train_loss=0.0220
367
+ step 3650000 | 36095s | train_loss=0.0233
368
+ step 3660000 | 36187s | train_loss=0.0224
369
+ step 3670000 | 36277s | train_loss=0.0233
370
+ step 3680000 | 36368s | train_loss=0.0261
371
+ step 3690000 | 36458s | train_loss=0.0253
372
+ step 3700000 | 36619s | train_loss=0.0228 | fid=48.9145
373
+ step 3710000 | 36712s | train_loss=0.0237
374
+ step 3720000 | 36802s | train_loss=0.0240
375
+ step 3730000 | 36895s | train_loss=0.0224
376
+ step 3740000 | 36987s | train_loss=0.0219
377
+ step 3750000 | 37080s | train_loss=0.0218
378
+ step 3760000 | 37172s | train_loss=0.0216
379
+ step 3770000 | 37265s | train_loss=0.0236
380
+ step 3780000 | 37352s | train_loss=0.0207
381
+ step 3790000 | 37443s | train_loss=0.0227
382
+ step 3800000 | 37605s | train_loss=0.0230 | fid=46.8959
383
+ step 3810000 | 37698s | train_loss=0.0216
384
+ step 3820000 | 37789s | train_loss=0.0216
385
+ step 3830000 | 37881s | train_loss=0.0212
386
+ step 3840000 | 37974s | train_loss=0.0213
387
+ step 3850000 | 38066s | train_loss=0.0233
388
+ step 3860000 | 38158s | train_loss=0.0239
389
+ step 3870000 | 38251s | train_loss=0.0221
390
+ step 3880000 | 38342s | train_loss=0.0216
391
+ step 3890000 | 38432s | train_loss=0.0220
392
+ step 3900000 | 38592s | train_loss=0.0215 | fid=46.7090
393
+ step 3910000 | 38683s | train_loss=0.0207
394
+ step 3920000 | 38775s | train_loss=0.0217
395
+ step 3930000 | 38868s | train_loss=0.0230
396
+ step 3940000 | 38960s | train_loss=0.0215
397
+ step 3950000 | 39052s | train_loss=0.0216
398
+ step 3960000 | 39145s | train_loss=0.0213
399
+ step 3970000 | 39237s | train_loss=0.0235
400
+ step 3980000 | 39330s | train_loss=0.0215
401
+ step 3990000 | 39419s | train_loss=0.0281
402
+ step 4000000 | 39579s | train_loss=0.0220 | fid=49.7388
403
+ step 4010000 | 39670s | train_loss=0.0205
404
+ step 4020000 | 39762s | train_loss=0.0238
405
+ step 4030000 | 39854s | train_loss=0.0212
406
+ step 4040000 | 39947s | train_loss=0.0213
407
+ step 4050000 | 40040s | train_loss=0.0197
408
+ step 4060000 | 40132s | train_loss=0.0214
409
+ step 4070000 | 40224s | train_loss=0.0221
410
+ step 4080000 | 40317s | train_loss=0.0203
411
+ step 4090000 | 40408s | train_loss=0.0222
412
+ step 4100000 | 40564s | train_loss=0.0215 | fid=47.8117
413
+ step 4110000 | 40656s | train_loss=0.0219
414
+ step 4120000 | 40748s | train_loss=0.0207
415
+ step 4130000 | 40841s | train_loss=0.0227
416
+ step 4140000 | 40933s | train_loss=0.0210
417
+ step 4150000 | 41026s | train_loss=0.0232
418
+ step 4160000 | 41118s | train_loss=0.0204
419
+ step 4170000 | 41211s | train_loss=0.0221
420
+ step 4180000 | 41303s | train_loss=0.0201
421
+ step 4190000 | 41396s | train_loss=0.0229
422
+ step 4200000 | 41551s | train_loss=0.0196 | fid=50.3229
423
+ step 4210000 | 41641s | train_loss=0.0209
424
+ step 4220000 | 41734s | train_loss=0.0204
425
+ step 4230000 | 41826s | train_loss=0.0243
426
+ step 4240000 | 41919s | train_loss=0.0214
427
+ step 4250000 | 42011s | train_loss=0.0207
428
+ step 4260000 | 42104s | train_loss=0.0202
429
+ step 4270000 | 42196s | train_loss=0.0209
430
+ step 4280000 | 42289s | train_loss=0.0200
431
+ step 4290000 | 42381s | train_loss=0.0210
432
+ step 4300000 | 42537s | train_loss=0.0213 | fid=48.7110
433
+ step 4310000 | 42628s | train_loss=0.0217
434
+ step 4320000 | 42721s | train_loss=0.0212
435
+ step 4330000 | 42813s | train_loss=0.0200
436
+ step 4340000 | 42906s | train_loss=0.0208
437
+ step 4350000 | 42998s | train_loss=0.0223
438
+ step 4360000 | 43091s | train_loss=0.0210
439
+ step 4370000 | 43183s | train_loss=0.0203
440
+ step 4380000 | 43276s | train_loss=0.0202
441
+ step 4390000 | 43367s | train_loss=0.0208
442
+ step 4400000 | 43527s | train_loss=0.0212 | fid=49.0618
443
+ step 4410000 | 43618s | train_loss=0.0199
444
+ step 4420000 | 43708s | train_loss=0.0216
445
+ step 4430000 | 43801s | train_loss=0.0192
446
+ step 4440000 | 43893s | train_loss=0.0199
447
+ step 4450000 | 43986s | train_loss=0.0215
448
+ step 4460000 | 44078s | train_loss=0.0196
449
+ step 4470000 | 44171s | train_loss=0.0204
450
+ step 4480000 | 44263s | train_loss=0.0215
451
+ step 4490000 | 44354s | train_loss=0.0203
452
+ step 4500000 | 44516s | train_loss=0.0222 | fid=48.6554
453
+ step 4510000 | 44606s | train_loss=0.0206
454
+ step 4520000 | 44697s | train_loss=0.0221
455
+ step 4530000 | 44790s | train_loss=0.0194
456
+ step 4540000 | 44882s | train_loss=0.0223
457
+ step 4550000 | 44975s | train_loss=0.0202
458
+ step 4560000 | 45067s | train_loss=0.0224
459
+ step 4570000 | 45160s | train_loss=0.0208
460
+ step 4580000 | 45250s | train_loss=0.0206
461
+ step 4590000 | 45343s | train_loss=0.0189
462
+ step 4600000 | 45504s | train_loss=0.0207 | fid=48.1524
463
+ step 4610000 | 45596s | train_loss=0.0211
464
+ step 4620000 | 45687s | train_loss=0.0208
465
+ step 4630000 | 45778s | train_loss=0.0204
466
+ step 4640000 | 45871s | train_loss=0.0200
467
+ step 4650000 | 45963s | train_loss=0.0215
468
+ step 4660000 | 46056s | train_loss=0.0215
469
+ step 4670000 | 46148s | train_loss=0.0195
470
+ step 4680000 | 46239s | train_loss=0.0242
471
+ step 4690000 | 46331s | train_loss=0.0192
472
+ step 4700000 | 46494s | train_loss=0.0208 | fid=48.5394
473
+ step 4710000 | 46587s | train_loss=0.0194
474
+ step 4720000 | 46676s | train_loss=0.0207
475
+ step 4730000 | 46767s | train_loss=0.0196
476
+ step 4740000 | 46860s | train_loss=0.0215
477
+ step 4750000 | 46952s | train_loss=0.0204
478
+ step 4760000 | 47045s | train_loss=0.0190
479
+ step 4770000 | 47134s | train_loss=0.0194
480
+ step 4780000 | 47209s | train_loss=0.0192
481
+ step 4790000 | 47284s | train_loss=0.0196
482
+ step 4800000 | 47416s | train_loss=0.0208 | fid=52.1318
483
+ step 4810000 | 47491s | train_loss=0.0232
484
+ step 4820000 | 47564s | train_loss=0.0217
485
+ step 4830000 | 47636s | train_loss=0.0202
486
+ step 4840000 | 47709s | train_loss=0.0200
487
+ step 4850000 | 47784s | train_loss=0.0181
488
+ step 4860000 | 47859s | train_loss=0.0186
489
+ step 4870000 | 47934s | train_loss=0.0203
490
+ step 4880000 | 48008s | train_loss=0.0201
491
+ step 4890000 | 48083s | train_loss=0.0199
492
+ step 4900000 | 48215s | train_loss=0.0183 | fid=48.8937
493
+ step 4910000 | 48290s | train_loss=0.0184
494
+ step 4920000 | 48365s | train_loss=0.0196
495
+ step 4930000 | 48435s | train_loss=0.0182
496
+ step 4940000 | 48508s | train_loss=0.0186
497
+ step 4950000 | 48582s | train_loss=0.0198
498
+ step 4960000 | 48657s | train_loss=0.0241
499
+ step 4970000 | 48732s | train_loss=0.0203
500
+ step 4980000 | 48807s | train_loss=0.0197
501
+ step 4990000 | 48881s | train_loss=0.0195
502
+ step 5000000 | 49013s | train_loss=0.0183 | fid=51.7889
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d30/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d30/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 67s | train_loss=0.9997 | fid=246.3696
3
+ step 10000 | 155s | train_loss=0.8678
4
+ step 20000 | 243s | train_loss=0.8565
5
+ step 30000 | 330s | train_loss=0.8390
6
+ step 40000 | 418s | train_loss=0.8040
7
+ step 50000 | 505s | train_loss=0.7455
8
+ step 60000 | 593s | train_loss=0.6627
9
+ step 70000 | 681s | train_loss=0.5480
10
+ step 80000 | 768s | train_loss=0.4254
11
+ step 90000 | 856s | train_loss=0.3313
12
+ step 100000 | 1003s | train_loss=0.2685 | fid=62.6922
13
+ step 110000 | 1089s | train_loss=0.2283
14
+ step 120000 | 1177s | train_loss=0.1992
15
+ step 130000 | 1264s | train_loss=0.1830
16
+ step 140000 | 1352s | train_loss=0.1672
17
+ step 150000 | 1439s | train_loss=0.1565
18
+ step 160000 | 1527s | train_loss=0.1449
19
+ step 170000 | 1615s | train_loss=0.1367
20
+ step 180000 | 1702s | train_loss=0.1308
21
+ step 190000 | 1790s | train_loss=0.1225
22
+ step 200000 | 1937s | train_loss=0.1176 | fid=62.5761
23
+ step 210000 | 2023s | train_loss=0.1118
24
+ step 220000 | 2110s | train_loss=0.1078
25
+ step 230000 | 2198s | train_loss=0.1028
26
+ step 240000 | 2286s | train_loss=0.0993
27
+ step 250000 | 2373s | train_loss=0.0956
28
+ step 260000 | 2461s | train_loss=0.0936
29
+ step 270000 | 2549s | train_loss=0.0898
30
+ step 280000 | 2636s | train_loss=0.0870
31
+ step 290000 | 2724s | train_loss=0.0844
32
+ step 300000 | 2872s | train_loss=0.0822 | fid=60.9021
33
+ step 310000 | 2959s | train_loss=0.0809
34
+ step 320000 | 3045s | train_loss=0.0788
35
+ step 330000 | 3133s | train_loss=0.0759
36
+ step 340000 | 3221s | train_loss=0.0743
37
+ step 350000 | 3308s | train_loss=0.0738
38
+ step 360000 | 3396s | train_loss=0.0706
39
+ step 370000 | 3484s | train_loss=0.0699
40
+ step 380000 | 3571s | train_loss=0.0680
41
+ step 390000 | 3655s | train_loss=0.0671
42
+ step 400000 | 3806s | train_loss=0.0647 | fid=59.5025
43
+ step 410000 | 3894s | train_loss=0.0631
44
+ step 420000 | 3980s | train_loss=0.0624
45
+ step 430000 | 4068s | train_loss=0.0614
46
+ step 440000 | 4155s | train_loss=0.0608
47
+ step 450000 | 4243s | train_loss=0.0598
48
+ step 460000 | 4331s | train_loss=0.0581
49
+ step 470000 | 4418s | train_loss=0.0564
50
+ step 480000 | 4506s | train_loss=0.0564
51
+ step 490000 | 4588s | train_loss=0.0554
52
+ step 500000 | 4742s | train_loss=0.0556 | fid=59.0599
53
+ step 510000 | 4830s | train_loss=0.0540
54
+ step 520000 | 4917s | train_loss=0.0537
55
+ step 530000 | 5004s | train_loss=0.0518
56
+ step 540000 | 5092s | train_loss=0.0513
57
+ step 550000 | 5179s | train_loss=0.0514
58
+ step 560000 | 5267s | train_loss=0.0498
59
+ step 570000 | 5355s | train_loss=0.0493
60
+ step 580000 | 5440s | train_loss=0.0488
61
+ step 590000 | 5522s | train_loss=0.0477
62
+ step 600000 | 5676s | train_loss=0.0470 | fid=58.1116
63
+ step 610000 | 5764s | train_loss=0.0466
64
+ step 620000 | 5852s | train_loss=0.0461
65
+ step 630000 | 5938s | train_loss=0.0459
66
+ step 640000 | 6025s | train_loss=0.0458
67
+ step 650000 | 6113s | train_loss=0.0450
68
+ step 660000 | 6200s | train_loss=0.0441
69
+ step 670000 | 6288s | train_loss=0.0436
70
+ step 680000 | 6370s | train_loss=0.0440
71
+ step 690000 | 6455s | train_loss=0.0428
72
+ step 700000 | 6610s | train_loss=0.0427 | fid=58.6299
73
+ step 710000 | 6697s | train_loss=0.0419
74
+ step 720000 | 6785s | train_loss=0.0414
75
+ step 730000 | 6871s | train_loss=0.0411
76
+ step 740000 | 6959s | train_loss=0.0407
77
+ step 750000 | 7046s | train_loss=0.0408
78
+ step 760000 | 7134s | train_loss=0.0399
79
+ step 770000 | 7218s | train_loss=0.0392
80
+ step 780000 | 7303s | train_loss=0.0389
81
+ step 790000 | 7388s | train_loss=0.0390
82
+ step 800000 | 7545s | train_loss=0.0391 | fid=57.6503
83
+ step 810000 | 7633s | train_loss=0.0379
84
+ step 820000 | 7720s | train_loss=0.0376
85
+ step 830000 | 7807s | train_loss=0.0378
86
+ step 840000 | 7894s | train_loss=0.0370
87
+ step 850000 | 7982s | train_loss=0.0370
88
+ step 860000 | 8069s | train_loss=0.0368
89
+ step 870000 | 8153s | train_loss=0.0362
90
+ step 880000 | 8236s | train_loss=0.0361
91
+ step 890000 | 8324s | train_loss=0.0359
92
+ step 900000 | 8481s | train_loss=0.0354 | fid=58.1807
93
+ step 910000 | 8569s | train_loss=0.0354
94
+ step 920000 | 8657s | train_loss=0.0345
95
+ step 930000 | 8745s | train_loss=0.0351
96
+ step 940000 | 8830s | train_loss=0.0345
97
+ step 950000 | 8918s | train_loss=0.0342
98
+ step 960000 | 9003s | train_loss=0.0340
99
+ step 970000 | 9088s | train_loss=0.0333
100
+ step 980000 | 9173s | train_loss=0.0335
101
+ step 990000 | 9261s | train_loss=0.0341
102
+ step 1000000 | 9418s | train_loss=0.0329 | fid=57.0327
103
+ step 1010000 | 9505s | train_loss=0.0328
104
+ step 1020000 | 9593s | train_loss=0.0328
105
+ step 1030000 | 9681s | train_loss=0.0324
106
+ step 1040000 | 9767s | train_loss=0.0320
107
+ step 1050000 | 9852s | train_loss=0.0324
108
+ step 1060000 | 9939s | train_loss=0.0317
109
+ step 1070000 | 10024s | train_loss=0.0316
110
+ step 1080000 | 10109s | train_loss=0.0317
111
+ step 1090000 | 10196s | train_loss=0.0310
112
+ step 1100000 | 10354s | train_loss=0.0312 | fid=57.1399
113
+ step 1110000 | 10441s | train_loss=0.0313
114
+ step 1120000 | 10529s | train_loss=0.0305
115
+ step 1130000 | 10617s | train_loss=0.0305
116
+ step 1140000 | 10703s | train_loss=0.0305
117
+ step 1150000 | 10787s | train_loss=0.0298
118
+ step 1160000 | 10872s | train_loss=0.0297
119
+ step 1170000 | 10957s | train_loss=0.0295
120
+ step 1180000 | 11045s | train_loss=0.0296
121
+ step 1190000 | 11132s | train_loss=0.0290
122
+ step 1200000 | 11290s | train_loss=0.0286 | fid=57.1631
123
+ step 1210000 | 11377s | train_loss=0.0289
124
+ step 1220000 | 11465s | train_loss=0.0286
125
+ step 1230000 | 11553s | train_loss=0.0281
126
+ step 1240000 | 11637s | train_loss=0.0286
127
+ step 1250000 | 11723s | train_loss=0.0280
128
+ step 1260000 | 11808s | train_loss=0.0280
129
+ step 1270000 | 11893s | train_loss=0.0284
130
+ step 1280000 | 11981s | train_loss=0.0278
131
+ step 1290000 | 12069s | train_loss=0.0279
132
+ step 1300000 | 12224s | train_loss=0.0275 | fid=56.6978
133
+ step 1310000 | 12312s | train_loss=0.0268
134
+ step 1320000 | 12399s | train_loss=0.0278
135
+ step 1330000 | 12487s | train_loss=0.0271
136
+ step 1340000 | 12572s | train_loss=0.0265
137
+ step 1350000 | 12655s | train_loss=0.0268
138
+ step 1360000 | 12743s | train_loss=0.0264
139
+ step 1370000 | 12828s | train_loss=0.0267
140
+ step 1380000 | 12916s | train_loss=0.0263
141
+ step 1390000 | 13003s | train_loss=0.0261
142
+ step 1400000 | 13161s | train_loss=0.0260 | fid=57.1048
143
+ step 1410000 | 13248s | train_loss=0.0261
144
+ step 1420000 | 13336s | train_loss=0.0258
145
+ step 1430000 | 13420s | train_loss=0.0252
146
+ step 1440000 | 13508s | train_loss=0.0253
147
+ step 1450000 | 13591s | train_loss=0.0253
148
+ step 1460000 | 13676s | train_loss=0.0249
149
+ step 1470000 | 13763s | train_loss=0.0252
150
+ step 1480000 | 13851s | train_loss=0.0255
151
+ step 1490000 | 13939s | train_loss=0.0254
152
+ step 1500000 | 14096s | train_loss=0.0249 | fid=57.5126
153
+ step 1510000 | 14184s | train_loss=0.0247
154
+ step 1520000 | 14271s | train_loss=0.0244
155
+ step 1530000 | 14356s | train_loss=0.0244
156
+ step 1540000 | 14443s | train_loss=0.0250
157
+ step 1550000 | 14527s | train_loss=0.0240
158
+ step 1560000 | 14611s | train_loss=0.0243
159
+ step 1570000 | 14699s | train_loss=0.0237
160
+ step 1580000 | 14787s | train_loss=0.0238
161
+ step 1590000 | 14874s | train_loss=0.0233
162
+ step 1600000 | 15032s | train_loss=0.0241 | fid=57.1424
163
+ step 1610000 | 15119s | train_loss=0.0240
164
+ step 1620000 | 15201s | train_loss=0.0232
165
+ step 1630000 | 15289s | train_loss=0.0233
166
+ step 1640000 | 15369s | train_loss=0.0232
167
+ step 1650000 | 15455s | train_loss=0.0231
168
+ step 1660000 | 15525s | train_loss=0.0232
169
+ step 1670000 | 15612s | train_loss=0.0231
170
+ step 1680000 | 15700s | train_loss=0.0228
171
+ step 1690000 | 15787s | train_loss=0.0226
172
+ step 1700000 | 15986s | train_loss=0.0230 | fid=57.0478
173
+ step 1710000 | 16068s | train_loss=0.0222
174
+ step 1720000 | 16154s | train_loss=0.0228
175
+ step 1730000 | 16242s | train_loss=0.0222
176
+ step 1740000 | 16326s | train_loss=0.0222
177
+ step 1750000 | 16410s | train_loss=0.0220
178
+ step 1760000 | 16496s | train_loss=0.0227
179
+ step 1770000 | 16584s | train_loss=0.0220
180
+ step 1780000 | 16671s | train_loss=0.0218
181
+ step 1790000 | 16759s | train_loss=0.0213
182
+ step 1800000 | 16914s | train_loss=0.0219 | fid=56.1358
183
+ step 1810000 | 16998s | train_loss=0.0214
184
+ step 1820000 | 17086s | train_loss=0.0215
185
+ step 1830000 | 17170s | train_loss=0.0213
186
+ step 1840000 | 17258s | train_loss=0.0213
187
+ step 1850000 | 17343s | train_loss=0.0215
188
+ step 1860000 | 17429s | train_loss=0.0211
189
+ step 1870000 | 17516s | train_loss=0.0210
190
+ step 1880000 | 17604s | train_loss=0.0215
191
+ step 1890000 | 17691s | train_loss=0.0212
192
+ step 1900000 | 17849s | train_loss=0.0216 | fid=56.0707
193
+ step 1910000 | 17935s | train_loss=0.0206
194
+ step 1920000 | 18023s | train_loss=0.0207
195
+ step 1930000 | 18107s | train_loss=0.0212
196
+ step 1940000 | 18195s | train_loss=0.0208
197
+ step 1950000 | 18279s | train_loss=0.0204
198
+ step 1960000 | 18365s | train_loss=0.0207
199
+ step 1970000 | 18453s | train_loss=0.0200
200
+ step 1980000 | 18541s | train_loss=0.0200
201
+ step 1990000 | 18628s | train_loss=0.0208
202
+ step 2000000 | 18782s | train_loss=0.0201 | fid=56.7878
203
+ step 2010000 | 18870s | train_loss=0.0198
204
+ step 2020000 | 18955s | train_loss=0.0202
205
+ step 2030000 | 19043s | train_loss=0.0194
206
+ step 2040000 | 19129s | train_loss=0.0195
207
+ step 2050000 | 19215s | train_loss=0.0204
208
+ step 2060000 | 19301s | train_loss=0.0198
209
+ step 2070000 | 19389s | train_loss=0.0201
210
+ step 2080000 | 19476s | train_loss=0.0199
211
+ step 2090000 | 19564s | train_loss=0.0196
212
+ step 2100000 | 19716s | train_loss=0.0197 | fid=56.5434
213
+ step 2110000 | 19803s | train_loss=0.0199
214
+ step 2120000 | 19888s | train_loss=0.0196
215
+ step 2130000 | 19975s | train_loss=0.0196
216
+ step 2140000 | 20060s | train_loss=0.0196
217
+ step 2150000 | 20147s | train_loss=0.0198
218
+ step 2160000 | 20235s | train_loss=0.0192
219
+ step 2170000 | 20321s | train_loss=0.0188
220
+ step 2180000 | 20409s | train_loss=0.0193
221
+ step 2190000 | 20497s | train_loss=0.0191
222
+ step 2200000 | 20649s | train_loss=0.0186 | fid=56.6299
223
+ step 2210000 | 20734s | train_loss=0.0187
224
+ step 2220000 | 20821s | train_loss=0.0188
225
+ step 2230000 | 20908s | train_loss=0.0186
226
+ step 2240000 | 20993s | train_loss=0.0188
227
+ step 2250000 | 21080s | train_loss=0.0186
228
+ step 2260000 | 21168s | train_loss=0.0184
229
+ step 2270000 | 21254s | train_loss=0.0185
230
+ step 2280000 | 21341s | train_loss=0.0182
231
+ step 2290000 | 21426s | train_loss=0.0183
232
+ step 2300000 | 21580s | train_loss=0.0186 | fid=56.5733
233
+ step 2310000 | 21666s | train_loss=0.0183
234
+ step 2320000 | 21753s | train_loss=0.0182
235
+ step 2330000 | 21840s | train_loss=0.0184
236
+ step 2340000 | 21925s | train_loss=0.0186
237
+ step 2350000 | 22013s | train_loss=0.0178
238
+ step 2360000 | 22100s | train_loss=0.0183
239
+ step 2370000 | 22187s | train_loss=0.0179
240
+ step 2380000 | 22274s | train_loss=0.0177
241
+ step 2390000 | 22359s | train_loss=0.0177
242
+ step 2400000 | 22516s | train_loss=0.0180 | fid=57.2717
243
+ step 2410000 | 22601s | train_loss=0.0172
244
+ step 2420000 | 22688s | train_loss=0.0178
245
+ step 2430000 | 22773s | train_loss=0.0177
246
+ step 2440000 | 22861s | train_loss=0.0181
247
+ step 2450000 | 22948s | train_loss=0.0175
248
+ step 2460000 | 23036s | train_loss=0.0176
249
+ step 2470000 | 23122s | train_loss=0.0184
250
+ step 2480000 | 23208s | train_loss=0.0171
251
+ step 2490000 | 23295s | train_loss=0.0171
252
+ step 2500000 | 23449s | train_loss=0.0169 | fid=57.1755
253
+ step 2510000 | 23537s | train_loss=0.0172
254
+ step 2520000 | 23624s | train_loss=0.0172
255
+ step 2530000 | 23709s | train_loss=0.0172
256
+ step 2540000 | 23797s | train_loss=0.0174
257
+ step 2550000 | 23884s | train_loss=0.0167
258
+ step 2560000 | 23972s | train_loss=0.0172
259
+ step 2570000 | 24059s | train_loss=0.0173
260
+ step 2580000 | 24143s | train_loss=0.0170
261
+ step 2590000 | 24231s | train_loss=0.0167
262
+ step 2600000 | 24385s | train_loss=0.0166 | fid=56.0969
263
+ step 2610000 | 24472s | train_loss=0.0166
264
+ step 2620000 | 24559s | train_loss=0.0166
265
+ step 2630000 | 24645s | train_loss=0.0167
266
+ step 2640000 | 24732s | train_loss=0.0169
267
+ step 2650000 | 24820s | train_loss=0.0167
268
+ step 2660000 | 24907s | train_loss=0.0161
269
+ step 2670000 | 24993s | train_loss=0.0163
270
+ step 2680000 | 25078s | train_loss=0.0169
271
+ step 2690000 | 25166s | train_loss=0.0164
272
+ step 2700000 | 25320s | train_loss=0.0162 | fid=57.3769
273
+ step 2710000 | 25407s | train_loss=0.0169
274
+ step 2720000 | 25492s | train_loss=0.0164
275
+ step 2730000 | 25579s | train_loss=0.0162
276
+ step 2740000 | 25667s | train_loss=0.0161
277
+ step 2750000 | 25754s | train_loss=0.0167
278
+ step 2760000 | 25841s | train_loss=0.0157
279
+ step 2770000 | 25927s | train_loss=0.0157
280
+ step 2780000 | 26013s | train_loss=0.0161
281
+ step 2790000 | 26101s | train_loss=0.0163
282
+ step 2800000 | 26255s | train_loss=0.0158 | fid=56.3607
283
+ step 2810000 | 26342s | train_loss=0.0159
284
+ step 2820000 | 26427s | train_loss=0.0157
285
+ step 2830000 | 26515s | train_loss=0.0157
286
+ step 2840000 | 26602s | train_loss=0.0154
287
+ step 2850000 | 26690s | train_loss=0.0154
288
+ step 2860000 | 26775s | train_loss=0.0153
289
+ step 2870000 | 26863s | train_loss=0.0156
290
+ step 2880000 | 26949s | train_loss=0.0152
291
+ step 2890000 | 27033s | train_loss=0.0151
292
+ step 2900000 | 27190s | train_loss=0.0155 | fid=56.8216
293
+ step 2910000 | 27276s | train_loss=0.0153
294
+ step 2920000 | 27362s | train_loss=0.0155
295
+ step 2930000 | 27450s | train_loss=0.0153
296
+ step 2940000 | 27537s | train_loss=0.0149
297
+ step 2950000 | 27623s | train_loss=0.0153
298
+ step 2960000 | 27711s | train_loss=0.0148
299
+ step 2970000 | 27798s | train_loss=0.0150
300
+ step 2980000 | 27884s | train_loss=0.0148
301
+ step 2990000 | 27969s | train_loss=0.0151
302
+ step 3000000 | 28124s | train_loss=0.0146 | fid=57.0132
303
+ step 3010000 | 28209s | train_loss=0.0146
304
+ step 3020000 | 28296s | train_loss=0.0146
305
+ step 3030000 | 28384s | train_loss=0.0147
306
+ step 3040000 | 28471s | train_loss=0.0151
307
+ step 3050000 | 28555s | train_loss=0.0151
308
+ step 3060000 | 28642s | train_loss=0.0150
309
+ step 3070000 | 28729s | train_loss=0.0151
310
+ step 3080000 | 28814s | train_loss=0.0150
311
+ step 3090000 | 28899s | train_loss=0.0144
312
+ step 3100000 | 29052s | train_loss=0.0143 | fid=57.0907
313
+ step 3110000 | 29137s | train_loss=0.0150
314
+ step 3120000 | 29225s | train_loss=0.0146
315
+ step 3130000 | 29313s | train_loss=0.0146
316
+ step 3140000 | 29399s | train_loss=0.0146
317
+ step 3150000 | 29485s | train_loss=0.0144
318
+ step 3160000 | 29572s | train_loss=0.0144
319
+ step 3170000 | 29659s | train_loss=0.0146
320
+ step 3180000 | 29743s | train_loss=0.0141
321
+ step 3190000 | 29829s | train_loss=0.0141
322
+ step 3200000 | 29984s | train_loss=0.0145 | fid=56.3519
323
+ step 3210000 | 30069s | train_loss=0.0144
324
+ step 3220000 | 30156s | train_loss=0.0142
325
+ step 3230000 | 30244s | train_loss=0.0144
326
+ step 3240000 | 30330s | train_loss=0.0142
327
+ step 3250000 | 30417s | train_loss=0.0142
328
+ step 3260000 | 30505s | train_loss=0.0141
329
+ step 3270000 | 30592s | train_loss=0.0138
330
+ step 3280000 | 30678s | train_loss=0.0142
331
+ step 3290000 | 30764s | train_loss=0.0141
332
+ step 3300000 | 30955s | train_loss=0.0140 | fid=56.7912
333
+ step 3310000 | 31043s | train_loss=0.0138
334
+ step 3320000 | 31130s | train_loss=0.0141
335
+ step 3330000 | 31216s | train_loss=0.0137
336
+ step 3340000 | 31304s | train_loss=0.0139
337
+ step 3350000 | 31391s | train_loss=0.0138
338
+ step 3360000 | 31479s | train_loss=0.0139
339
+ step 3370000 | 31552s | train_loss=0.0134
340
+ step 3380000 | 31640s | train_loss=0.0140
341
+ step 3390000 | 31726s | train_loss=0.0139
342
+ step 3400000 | 31881s | train_loss=0.0136 | fid=56.3661
343
+ step 3410000 | 31969s | train_loss=0.0137
344
+ step 3420000 | 32055s | train_loss=0.0144
345
+ step 3430000 | 32142s | train_loss=0.0137
346
+ step 3440000 | 32230s | train_loss=0.0132
347
+ step 3450000 | 32317s | train_loss=0.0135
348
+ step 3460000 | 32405s | train_loss=0.0138
349
+ step 3470000 | 32489s | train_loss=0.0132
350
+ step 3480000 | 32577s | train_loss=0.0132
351
+ step 3490000 | 32663s | train_loss=0.0137
352
+ step 3500000 | 32817s | train_loss=0.0137 | fid=57.4654
353
+ step 3510000 | 32903s | train_loss=0.0132
354
+ step 3520000 | 32991s | train_loss=0.0130
355
+ step 3530000 | 33078s | train_loss=0.0133
356
+ step 3540000 | 33166s | train_loss=0.0133
357
+ step 3550000 | 33253s | train_loss=0.0130
358
+ step 3560000 | 33339s | train_loss=0.0134
359
+ step 3570000 | 33425s | train_loss=0.0130
360
+ step 3580000 | 33513s | train_loss=0.0128
361
+ step 3590000 | 33599s | train_loss=0.0130
362
+ step 3600000 | 33783s | train_loss=0.0129 | fid=56.8537
363
+ step 3610000 | 33871s | train_loss=0.0131
364
+ step 3620000 | 33958s | train_loss=0.0131
365
+ step 3630000 | 34046s | train_loss=0.0126
366
+ step 3640000 | 34134s | train_loss=0.0125
367
+ step 3650000 | 34221s | train_loss=0.0131
368
+ step 3660000 | 34301s | train_loss=0.0126
369
+ step 3670000 | 34389s | train_loss=0.0129
370
+ step 3680000 | 34476s | train_loss=0.0130
371
+ step 3690000 | 34558s | train_loss=0.0127
372
+ step 3700000 | 34758s | train_loss=0.0126 | fid=56.9296
373
+ step 3710000 | 34845s | train_loss=0.0131
374
+ step 3720000 | 34933s | train_loss=0.0128
375
+ step 3730000 | 35020s | train_loss=0.0125
376
+ step 3740000 | 35108s | train_loss=0.0125
377
+ step 3750000 | 35181s | train_loss=0.0129
378
+ step 3760000 | 35269s | train_loss=0.0125
379
+ step 3770000 | 35357s | train_loss=0.0126
380
+ step 3780000 | 35444s | train_loss=0.0124
381
+ step 3790000 | 35520s | train_loss=0.0128
382
+ step 3800000 | 35708s | train_loss=0.0128 | fid=56.6692
383
+ step 3810000 | 35796s | train_loss=0.0127
384
+ step 3820000 | 35884s | train_loss=0.0128
385
+ step 3830000 | 35971s | train_loss=0.0126
386
+ step 3840000 | 36058s | train_loss=0.0124
387
+ step 3850000 | 36134s | train_loss=0.0120
388
+ step 3860000 | 36222s | train_loss=0.0124
389
+ step 3870000 | 36309s | train_loss=0.0125
390
+ step 3880000 | 36388s | train_loss=0.0121
391
+ step 3890000 | 36466s | train_loss=0.0119
392
+ step 3900000 | 36669s | train_loss=0.0123 | fid=56.5404
393
+ step 3910000 | 36756s | train_loss=0.0125
394
+ step 3920000 | 36844s | train_loss=0.0124
395
+ step 3930000 | 36932s | train_loss=0.0121
396
+ step 3940000 | 37010s | train_loss=0.0121
397
+ step 3950000 | 37093s | train_loss=0.0123
398
+ step 3960000 | 37181s | train_loss=0.0121
399
+ step 3970000 | 37267s | train_loss=0.0118
400
+ step 3980000 | 37342s | train_loss=0.0121
401
+ step 3990000 | 37426s | train_loss=0.0117
402
+ step 4000000 | 37628s | train_loss=0.0118 | fid=57.0633
403
+ step 4010000 | 37716s | train_loss=0.0118
404
+ step 4020000 | 37803s | train_loss=0.0118
405
+ step 4030000 | 37891s | train_loss=0.0121
406
+ step 4040000 | 37965s | train_loss=0.0117
407
+ step 4050000 | 38053s | train_loss=0.0117
408
+ step 4060000 | 38138s | train_loss=0.0121
409
+ step 4070000 | 38226s | train_loss=0.0117
410
+ step 4080000 | 38296s | train_loss=0.0116
411
+ step 4090000 | 38384s | train_loss=0.0116
412
+ step 4100000 | 38570s | train_loss=0.0116 | fid=56.5137
413
+ step 4110000 | 38658s | train_loss=0.0115
414
+ step 4120000 | 38745s | train_loss=0.0117
415
+ step 4130000 | 38829s | train_loss=0.0116
416
+ step 4140000 | 38906s | train_loss=0.0117
417
+ step 4150000 | 38991s | train_loss=0.0119
418
+ step 4160000 | 39079s | train_loss=0.0118
419
+ step 4170000 | 39167s | train_loss=0.0115
420
+ step 4180000 | 39238s | train_loss=0.0115
421
+ step 4190000 | 39325s | train_loss=0.0115
422
+ step 4200000 | 39506s | train_loss=0.0113 | fid=57.2520
423
+ step 4210000 | 39593s | train_loss=0.0116
424
+ step 4220000 | 39681s | train_loss=0.0114
425
+ step 4230000 | 39763s | train_loss=0.0112
426
+ step 4240000 | 39844s | train_loss=0.0119
427
+ step 4250000 | 39932s | train_loss=0.0111
428
+ step 4260000 | 40019s | train_loss=0.0113
429
+ step 4270000 | 40107s | train_loss=0.0112
430
+ step 4280000 | 40178s | train_loss=0.0114
431
+ step 4290000 | 40266s | train_loss=0.0114
432
+ step 4300000 | 40453s | train_loss=0.0114 | fid=57.3825
433
+ step 4310000 | 40540s | train_loss=0.0111
434
+ step 4320000 | 40628s | train_loss=0.0110
435
+ step 4330000 | 40703s | train_loss=0.0114
436
+ step 4340000 | 40791s | train_loss=0.0111
437
+ step 4350000 | 40878s | train_loss=0.0113
438
+ step 4360000 | 40966s | train_loss=0.0114
439
+ step 4370000 | 41053s | train_loss=0.0111
440
+ step 4380000 | 41125s | train_loss=0.0112
441
+ step 4390000 | 41213s | train_loss=0.0110
442
+ step 4400000 | 41382s | train_loss=0.0114 | fid=56.8287
443
+ step 4410000 | 41469s | train_loss=0.0112
444
+ step 4420000 | 41554s | train_loss=0.0108
445
+ step 4430000 | 41636s | train_loss=0.0108
446
+ step 4440000 | 41723s | train_loss=0.0109
447
+ step 4450000 | 41810s | train_loss=0.0108
448
+ step 4460000 | 41897s | train_loss=0.0109
449
+ step 4470000 | 41985s | train_loss=0.0109
450
+ step 4480000 | 42067s | train_loss=0.0110
451
+ step 4490000 | 42154s | train_loss=0.0111
452
+ step 4500000 | 42314s | train_loss=0.0109 | fid=56.7372
453
+ step 4510000 | 42400s | train_loss=0.0108
454
+ step 4520000 | 42483s | train_loss=0.0114
455
+ step 4530000 | 42570s | train_loss=0.0111
456
+ step 4540000 | 42657s | train_loss=0.0109
457
+ step 4550000 | 42744s | train_loss=0.0107
458
+ step 4560000 | 42831s | train_loss=0.0108
459
+ step 4570000 | 42917s | train_loss=0.0105
460
+ step 4580000 | 43002s | train_loss=0.0105
461
+ step 4590000 | 43090s | train_loss=0.0111
462
+ step 4600000 | 43247s | train_loss=0.0103 | fid=56.6307
463
+ step 4610000 | 43332s | train_loss=0.0108
464
+ step 4620000 | 43416s | train_loss=0.0106
465
+ step 4630000 | 43503s | train_loss=0.0106
466
+ step 4640000 | 43590s | train_loss=0.0108
467
+ step 4650000 | 43677s | train_loss=0.0106
468
+ step 4660000 | 43765s | train_loss=0.0105
469
+ step 4670000 | 43850s | train_loss=0.0106
470
+ step 4680000 | 43935s | train_loss=0.0104
471
+ step 4690000 | 44023s | train_loss=0.0106
472
+ step 4700000 | 44178s | train_loss=0.0111 | fid=56.6334
473
+ step 4710000 | 44246s | train_loss=0.0102
474
+ step 4720000 | 44314s | train_loss=0.0103
475
+ step 4730000 | 44384s | train_loss=0.0106
476
+ step 4740000 | 44454s | train_loss=0.0107
477
+ step 4750000 | 44524s | train_loss=0.0104
478
+ step 4760000 | 44591s | train_loss=0.0105
479
+ step 4770000 | 44662s | train_loss=0.0101
480
+ step 4780000 | 44729s | train_loss=0.0110
481
+ step 4790000 | 44799s | train_loss=0.0103
482
+ step 4800000 | 44927s | train_loss=0.0104 | fid=56.5863
483
+ step 4810000 | 44987s | train_loss=0.0101
484
+ step 4820000 | 45040s | train_loss=0.0104
485
+ step 4830000 | 45093s | train_loss=0.0103
486
+ step 4840000 | 45145s | train_loss=0.0104
487
+ step 4850000 | 45198s | train_loss=0.0103
488
+ step 4860000 | 45239s | train_loss=0.0106
489
+ step 4870000 | 45275s | train_loss=0.0105
490
+ step 4880000 | 45304s | train_loss=0.0103
491
+ step 4890000 | 45339s | train_loss=0.0100
492
+ step 4900000 | 45405s | train_loss=0.0103 | fid=56.5028
493
+ step 4910000 | 45441s | train_loss=0.0101
494
+ step 4920000 | 45476s | train_loss=0.0099
495
+ step 4930000 | 45511s | train_loss=0.0100
496
+ step 4940000 | 45547s | train_loss=0.0101
497
+ step 4950000 | 45582s | train_loss=0.0103
498
+ step 4960000 | 45617s | train_loss=0.0102
499
+ step 4970000 | 45653s | train_loss=0.0101
500
+ step 4980000 | 45681s | train_loss=0.0099
501
+ step 4990000 | 45716s | train_loss=0.0099
502
+ step 5000000 | 45783s | train_loss=0.0103 | fid=56.2486
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d40/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d40/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 67s | train_loss=0.9934 | fid=276.0154
3
+ step 10000 | 156s | train_loss=0.8531
4
+ step 20000 | 246s | train_loss=0.8206
5
+ step 30000 | 337s | train_loss=0.7603
6
+ step 40000 | 427s | train_loss=0.6698
7
+ step 50000 | 518s | train_loss=0.5341
8
+ step 60000 | 608s | train_loss=0.3887
9
+ step 70000 | 699s | train_loss=0.2841
10
+ step 80000 | 790s | train_loss=0.2228
11
+ step 90000 | 880s | train_loss=0.1887
12
+ step 100000 | 1031s | train_loss=0.1681 | fid=60.5336
13
+ step 110000 | 1121s | train_loss=0.1528
14
+ step 120000 | 1212s | train_loss=0.1406
15
+ step 130000 | 1302s | train_loss=0.1302
16
+ step 140000 | 1393s | train_loss=0.1230
17
+ step 150000 | 1484s | train_loss=0.1162
18
+ step 160000 | 1574s | train_loss=0.1111
19
+ step 170000 | 1665s | train_loss=0.1048
20
+ step 180000 | 1755s | train_loss=0.0999
21
+ step 190000 | 1845s | train_loss=0.0972
22
+ step 200000 | 1998s | train_loss=0.0926 | fid=59.2439
23
+ step 210000 | 2088s | train_loss=0.0886
24
+ step 220000 | 2178s | train_loss=0.0874
25
+ step 230000 | 2269s | train_loss=0.0837
26
+ step 240000 | 2359s | train_loss=0.0798
27
+ step 250000 | 2449s | train_loss=0.0778
28
+ step 260000 | 2540s | train_loss=0.0760
29
+ step 270000 | 2630s | train_loss=0.0732
30
+ step 280000 | 2721s | train_loss=0.0716
31
+ step 290000 | 2805s | train_loss=0.0698
32
+ step 300000 | 2964s | train_loss=0.0686 | fid=56.5992
33
+ step 310000 | 3055s | train_loss=0.0664
34
+ step 320000 | 3145s | train_loss=0.0649
35
+ step 330000 | 3236s | train_loss=0.0630
36
+ step 340000 | 3327s | train_loss=0.0624
37
+ step 350000 | 3417s | train_loss=0.0607
38
+ step 360000 | 3508s | train_loss=0.0593
39
+ step 370000 | 3598s | train_loss=0.0585
40
+ step 380000 | 3683s | train_loss=0.0576
41
+ step 390000 | 3771s | train_loss=0.0562
42
+ step 400000 | 3930s | train_loss=0.0552 | fid=55.7165
43
+ step 410000 | 4021s | train_loss=0.0547
44
+ step 420000 | 4111s | train_loss=0.0529
45
+ step 430000 | 4202s | train_loss=0.0527
46
+ step 440000 | 4292s | train_loss=0.0519
47
+ step 450000 | 4383s | train_loss=0.0509
48
+ step 460000 | 4473s | train_loss=0.0509
49
+ step 470000 | 4558s | train_loss=0.0499
50
+ step 480000 | 4645s | train_loss=0.0497
51
+ step 490000 | 4735s | train_loss=0.0482
52
+ step 500000 | 4894s | train_loss=0.0479 | fid=54.3212
53
+ step 510000 | 4984s | train_loss=0.0470
54
+ step 520000 | 5075s | train_loss=0.0464
55
+ step 530000 | 5165s | train_loss=0.0453
56
+ step 540000 | 5256s | train_loss=0.0458
57
+ step 550000 | 5346s | train_loss=0.0450
58
+ step 560000 | 5434s | train_loss=0.0438
59
+ step 570000 | 5519s | train_loss=0.0442
60
+ step 580000 | 5609s | train_loss=0.0440
61
+ step 590000 | 5697s | train_loss=0.0431
62
+ step 600000 | 5859s | train_loss=0.0421 | fid=54.6626
63
+ step 610000 | 5949s | train_loss=0.0418
64
+ step 620000 | 6040s | train_loss=0.0414
65
+ step 630000 | 6130s | train_loss=0.0428
66
+ step 640000 | 6221s | train_loss=0.0407
67
+ step 650000 | 6309s | train_loss=0.0402
68
+ step 660000 | 6396s | train_loss=0.0398
69
+ step 670000 | 6484s | train_loss=0.0401
70
+ step 680000 | 6574s | train_loss=0.0393
71
+ step 690000 | 6662s | train_loss=0.0394
72
+ step 700000 | 6824s | train_loss=0.0386 | fid=54.3034
73
+ step 710000 | 6915s | train_loss=0.0379
74
+ step 720000 | 7005s | train_loss=0.0380
75
+ step 730000 | 7096s | train_loss=0.0373
76
+ step 740000 | 7184s | train_loss=0.0366
77
+ step 750000 | 7270s | train_loss=0.0368
78
+ step 760000 | 7358s | train_loss=0.0365
79
+ step 770000 | 7448s | train_loss=0.0368
80
+ step 780000 | 7538s | train_loss=0.0354
81
+ step 790000 | 7626s | train_loss=0.0353
82
+ step 800000 | 7788s | train_loss=0.0351 | fid=52.6865
83
+ step 810000 | 7879s | train_loss=0.0349
84
+ step 820000 | 7969s | train_loss=0.0348
85
+ step 830000 | 8060s | train_loss=0.0345
86
+ step 840000 | 8147s | train_loss=0.0338
87
+ step 850000 | 8231s | train_loss=0.0343
88
+ step 860000 | 8322s | train_loss=0.0335
89
+ step 870000 | 8412s | train_loss=0.0333
90
+ step 880000 | 8500s | train_loss=0.0336
91
+ step 890000 | 8590s | train_loss=0.0339
92
+ step 900000 | 8752s | train_loss=0.0328 | fid=52.3237
93
+ step 910000 | 8843s | train_loss=0.0323
94
+ step 920000 | 8934s | train_loss=0.0323
95
+ step 930000 | 9021s | train_loss=0.0322
96
+ step 940000 | 9109s | train_loss=0.0317
97
+ step 950000 | 9196s | train_loss=0.0321
98
+ step 960000 | 9286s | train_loss=0.0313
99
+ step 970000 | 9377s | train_loss=0.0311
100
+ step 980000 | 9464s | train_loss=0.0309
101
+ step 990000 | 9555s | train_loss=0.0305
102
+ step 1000000 | 9717s | train_loss=0.0301 | fid=52.8634
103
+ step 1010000 | 9807s | train_loss=0.0302
104
+ step 1020000 | 9895s | train_loss=0.0299
105
+ step 1030000 | 9982s | train_loss=0.0298
106
+ step 1040000 | 10069s | train_loss=0.0296
107
+ step 1050000 | 10160s | train_loss=0.0305
108
+ step 1060000 | 10250s | train_loss=0.0296
109
+ step 1070000 | 10340s | train_loss=0.0289
110
+ step 1080000 | 10427s | train_loss=0.0293
111
+ step 1090000 | 10518s | train_loss=0.0289
112
+ step 1100000 | 10680s | train_loss=0.0286 | fid=52.7596
113
+ step 1110000 | 10767s | train_loss=0.0287
114
+ step 1120000 | 10855s | train_loss=0.0289
115
+ step 1130000 | 10945s | train_loss=0.0276
116
+ step 1140000 | 11032s | train_loss=0.0282
117
+ step 1150000 | 11123s | train_loss=0.0285
118
+ step 1160000 | 11213s | train_loss=0.0272
119
+ step 1170000 | 11300s | train_loss=0.0272
120
+ step 1180000 | 11391s | train_loss=0.0271
121
+ step 1190000 | 11482s | train_loss=0.0276
122
+ step 1200000 | 11641s | train_loss=0.0272 | fid=52.2709
123
+ step 1210000 | 11731s | train_loss=0.0270
124
+ step 1220000 | 11819s | train_loss=0.0269
125
+ step 1230000 | 11906s | train_loss=0.0266
126
+ step 1240000 | 11997s | train_loss=0.0267
127
+ step 1250000 | 12087s | train_loss=0.0268
128
+ step 1260000 | 12178s | train_loss=0.0267
129
+ step 1270000 | 12266s | train_loss=0.0262
130
+ step 1280000 | 12356s | train_loss=0.0261
131
+ step 1290000 | 12447s | train_loss=0.0259
132
+ step 1300000 | 12604s | train_loss=0.0259 | fid=52.1968
133
+ step 1310000 | 12692s | train_loss=0.0259
134
+ step 1320000 | 12780s | train_loss=0.0253
135
+ step 1330000 | 12870s | train_loss=0.0253
136
+ step 1340000 | 12961s | train_loss=0.0250
137
+ step 1350000 | 13051s | train_loss=0.0251
138
+ step 1360000 | 13141s | train_loss=0.0252
139
+ step 1370000 | 13229s | train_loss=0.0249
140
+ step 1380000 | 13319s | train_loss=0.0252
141
+ step 1390000 | 13407s | train_loss=0.0245
142
+ step 1400000 | 13564s | train_loss=0.0241 | fid=52.1689
143
+ step 1410000 | 13654s | train_loss=0.0243
144
+ step 1420000 | 13742s | train_loss=0.0242
145
+ step 1430000 | 13833s | train_loss=0.0243
146
+ step 1440000 | 13923s | train_loss=0.0237
147
+ step 1450000 | 14013s | train_loss=0.0249
148
+ step 1460000 | 14100s | train_loss=0.0239
149
+ step 1470000 | 14191s | train_loss=0.0243
150
+ step 1480000 | 14280s | train_loss=0.0235
151
+ step 1490000 | 14367s | train_loss=0.0237
152
+ step 1500000 | 14524s | train_loss=0.0233 | fid=52.3043
153
+ step 1510000 | 14611s | train_loss=0.0236
154
+ step 1520000 | 14702s | train_loss=0.0231
155
+ step 1530000 | 14792s | train_loss=0.0230
156
+ step 1540000 | 14883s | train_loss=0.0230
157
+ step 1550000 | 14973s | train_loss=0.0227
158
+ step 1560000 | 15060s | train_loss=0.0229
159
+ step 1570000 | 15151s | train_loss=0.0224
160
+ step 1580000 | 15235s | train_loss=0.0225
161
+ step 1590000 | 15325s | train_loss=0.0222
162
+ step 1600000 | 15475s | train_loss=0.0222 | fid=52.5750
163
+ step 1610000 | 15553s | train_loss=0.0218
164
+ step 1620000 | 15642s | train_loss=0.0215
165
+ step 1630000 | 15732s | train_loss=0.0220
166
+ step 1640000 | 15822s | train_loss=0.0216
167
+ step 1650000 | 15912s | train_loss=0.0215
168
+ step 1660000 | 15989s | train_loss=0.0223
169
+ step 1670000 | 16073s | train_loss=0.0221
170
+ step 1680000 | 16163s | train_loss=0.0215
171
+ step 1690000 | 16250s | train_loss=0.0213
172
+ step 1700000 | 16411s | train_loss=0.0215 | fid=51.6409
173
+ step 1710000 | 16499s | train_loss=0.0213
174
+ step 1720000 | 16589s | train_loss=0.0215
175
+ step 1730000 | 16679s | train_loss=0.0211
176
+ step 1740000 | 16770s | train_loss=0.0216
177
+ step 1750000 | 16860s | train_loss=0.0213
178
+ step 1760000 | 16947s | train_loss=0.0206
179
+ step 1770000 | 17034s | train_loss=0.0210
180
+ step 1780000 | 17124s | train_loss=0.0210
181
+ step 1790000 | 17211s | train_loss=0.0206
182
+ step 1800000 | 17369s | train_loss=0.0206 | fid=51.9642
183
+ step 1810000 | 17460s | train_loss=0.0212
184
+ step 1820000 | 17549s | train_loss=0.0201
185
+ step 1830000 | 17640s | train_loss=0.0202
186
+ step 1840000 | 17730s | train_loss=0.0208
187
+ step 1850000 | 17820s | train_loss=0.0205
188
+ step 1860000 | 17903s | train_loss=0.0204
189
+ step 1870000 | 17993s | train_loss=0.0202
190
+ step 1880000 | 18080s | train_loss=0.0200
191
+ step 1890000 | 18171s | train_loss=0.0207
192
+ step 1900000 | 18328s | train_loss=0.0212 | fid=51.8678
193
+ step 1910000 | 18419s | train_loss=0.0196
194
+ step 1920000 | 18509s | train_loss=0.0197
195
+ step 1930000 | 18600s | train_loss=0.0195
196
+ step 1940000 | 18690s | train_loss=0.0194
197
+ step 1950000 | 18775s | train_loss=0.0192
198
+ step 1960000 | 18863s | train_loss=0.0195
199
+ step 1970000 | 18951s | train_loss=0.0193
200
+ step 1980000 | 19041s | train_loss=0.0197
201
+ step 1990000 | 19130s | train_loss=0.0193
202
+ step 2000000 | 19288s | train_loss=0.0187 | fid=52.2860
203
+ step 2010000 | 19377s | train_loss=0.0194
204
+ step 2020000 | 19467s | train_loss=0.0189
205
+ step 2030000 | 19556s | train_loss=0.0186
206
+ step 2040000 | 19642s | train_loss=0.0189
207
+ step 2050000 | 19726s | train_loss=0.0187
208
+ step 2060000 | 19816s | train_loss=0.0189
209
+ step 2070000 | 19901s | train_loss=0.0184
210
+ step 2080000 | 19991s | train_loss=0.0190
211
+ step 2090000 | 20078s | train_loss=0.0186
212
+ step 2100000 | 20239s | train_loss=0.0190 | fid=51.9741
213
+ step 2110000 | 20329s | train_loss=0.0188
214
+ step 2120000 | 20419s | train_loss=0.0180
215
+ step 2130000 | 20510s | train_loss=0.0196
216
+ step 2140000 | 20597s | train_loss=0.0186
217
+ step 2150000 | 20683s | train_loss=0.0177
218
+ step 2160000 | 20770s | train_loss=0.0180
219
+ step 2170000 | 20860s | train_loss=0.0184
220
+ step 2180000 | 20946s | train_loss=0.0185
221
+ step 2190000 | 21036s | train_loss=0.0186
222
+ step 2200000 | 21196s | train_loss=0.0175 | fid=51.9769
223
+ step 2210000 | 21285s | train_loss=0.0179
224
+ step 2220000 | 21375s | train_loss=0.0176
225
+ step 2230000 | 21461s | train_loss=0.0175
226
+ step 2240000 | 21550s | train_loss=0.0174
227
+ step 2250000 | 21635s | train_loss=0.0182
228
+ step 2260000 | 21725s | train_loss=0.0172
229
+ step 2270000 | 21814s | train_loss=0.0173
230
+ step 2280000 | 21901s | train_loss=0.0184
231
+ step 2290000 | 21992s | train_loss=0.0174
232
+ step 2300000 | 22153s | train_loss=0.0173 | fid=51.7212
233
+ step 2310000 | 22244s | train_loss=0.0176
234
+ step 2320000 | 22331s | train_loss=0.0173
235
+ step 2330000 | 22421s | train_loss=0.0167
236
+ step 2340000 | 22510s | train_loss=0.0172
237
+ step 2350000 | 22595s | train_loss=0.0166
238
+ step 2360000 | 22685s | train_loss=0.0174
239
+ step 2370000 | 22772s | train_loss=0.0169
240
+ step 2380000 | 22863s | train_loss=0.0168
241
+ step 2390000 | 22954s | train_loss=0.0167
242
+ step 2400000 | 23112s | train_loss=0.0162 | fid=52.5185
243
+ step 2410000 | 23200s | train_loss=0.0163
244
+ step 2420000 | 23289s | train_loss=0.0167
245
+ step 2430000 | 23378s | train_loss=0.0164
246
+ step 2440000 | 23460s | train_loss=0.0169
247
+ step 2450000 | 23549s | train_loss=0.0166
248
+ step 2460000 | 23639s | train_loss=0.0168
249
+ step 2470000 | 23725s | train_loss=0.0175
250
+ step 2480000 | 23814s | train_loss=0.0163
251
+ step 2490000 | 23904s | train_loss=0.0161
252
+ step 2500000 | 24063s | train_loss=0.0164 | fid=51.7748
253
+ step 2510000 | 24151s | train_loss=0.0161
254
+ step 2520000 | 24241s | train_loss=0.0165
255
+ step 2530000 | 24329s | train_loss=0.0160
256
+ step 2540000 | 24413s | train_loss=0.0168
257
+ step 2550000 | 24503s | train_loss=0.0159
258
+ step 2560000 | 24589s | train_loss=0.0158
259
+ step 2570000 | 24679s | train_loss=0.0165
260
+ step 2580000 | 24768s | train_loss=0.0158
261
+ step 2590000 | 24857s | train_loss=0.0156
262
+ step 2600000 | 25015s | train_loss=0.0156 | fid=52.3944
263
+ step 2610000 | 25105s | train_loss=0.0152
264
+ step 2620000 | 25194s | train_loss=0.0159
265
+ step 2630000 | 25281s | train_loss=0.0156
266
+ step 2640000 | 25367s | train_loss=0.0160
267
+ step 2650000 | 25456s | train_loss=0.0156
268
+ step 2660000 | 25542s | train_loss=0.0156
269
+ step 2670000 | 25632s | train_loss=0.0152
270
+ step 2680000 | 25720s | train_loss=0.0159
271
+ step 2690000 | 25810s | train_loss=0.0152
272
+ step 2700000 | 25968s | train_loss=0.0155 | fid=52.2704
273
+ step 2710000 | 26057s | train_loss=0.0160
274
+ step 2720000 | 26144s | train_loss=0.0160
275
+ step 2730000 | 26233s | train_loss=0.0154
276
+ step 2740000 | 26319s | train_loss=0.0147
277
+ step 2750000 | 26406s | train_loss=0.0152
278
+ step 2760000 | 26495s | train_loss=0.0149
279
+ step 2770000 | 26584s | train_loss=0.0151
280
+ step 2780000 | 26673s | train_loss=0.0147
281
+ step 2790000 | 26761s | train_loss=0.0150
282
+ step 2800000 | 26921s | train_loss=0.0145 | fid=51.9537
283
+ step 2810000 | 27010s | train_loss=0.0145
284
+ step 2820000 | 27097s | train_loss=0.0153
285
+ step 2830000 | 27184s | train_loss=0.0144
286
+ step 2840000 | 27272s | train_loss=0.0144
287
+ step 2850000 | 27358s | train_loss=0.0149
288
+ step 2860000 | 27447s | train_loss=0.0156
289
+ step 2870000 | 27537s | train_loss=0.0146
290
+ step 2880000 | 27625s | train_loss=0.0145
291
+ step 2890000 | 27714s | train_loss=0.0143
292
+ step 2900000 | 27874s | train_loss=0.0141 | fid=51.8377
293
+ step 2910000 | 27961s | train_loss=0.0143
294
+ step 2920000 | 28052s | train_loss=0.0139
295
+ step 2930000 | 28139s | train_loss=0.0142
296
+ step 2940000 | 28227s | train_loss=0.0141
297
+ step 2950000 | 28317s | train_loss=0.0136
298
+ step 2960000 | 28406s | train_loss=0.0137
299
+ step 2970000 | 28492s | train_loss=0.0142
300
+ step 2980000 | 28575s | train_loss=0.0138
301
+ step 2990000 | 28661s | train_loss=0.0136
302
+ step 3000000 | 28816s | train_loss=0.0133 | fid=51.4702
303
+ step 3010000 | 28904s | train_loss=0.0134
304
+ step 3020000 | 28994s | train_loss=0.0133
305
+ step 3030000 | 29082s | train_loss=0.0138
306
+ step 3040000 | 29170s | train_loss=0.0143
307
+ step 3050000 | 29261s | train_loss=0.0142
308
+ step 3060000 | 29351s | train_loss=0.0138
309
+ step 3070000 | 29436s | train_loss=0.0136
310
+ step 3080000 | 29522s | train_loss=0.0134
311
+ step 3090000 | 29608s | train_loss=0.0134
312
+ step 3100000 | 29763s | train_loss=0.0140 | fid=52.0210
313
+ step 3110000 | 29853s | train_loss=0.0140
314
+ step 3120000 | 29944s | train_loss=0.0131
315
+ step 3130000 | 30028s | train_loss=0.0137
316
+ step 3140000 | 30119s | train_loss=0.0133
317
+ step 3150000 | 30210s | train_loss=0.0139
318
+ step 3160000 | 30298s | train_loss=0.0133
319
+ step 3170000 | 30389s | train_loss=0.0130
320
+ step 3180000 | 30479s | train_loss=0.0132
321
+ step 3190000 | 30570s | train_loss=0.0134
322
+ step 3200000 | 30729s | train_loss=0.0130 | fid=51.8826
323
+ step 3210000 | 30819s | train_loss=0.0130
324
+ step 3220000 | 30908s | train_loss=0.0129
325
+ step 3230000 | 30975s | train_loss=0.0132
326
+ step 3240000 | 31066s | train_loss=0.0132
327
+ step 3250000 | 31155s | train_loss=0.0139
328
+ step 3260000 | 31245s | train_loss=0.0127
329
+ step 3270000 | 31335s | train_loss=0.0129
330
+ step 3280000 | 31425s | train_loss=0.0139
331
+ step 3290000 | 31511s | train_loss=0.0125
332
+ step 3300000 | 31661s | train_loss=0.0131 | fid=52.2213
333
+ step 3310000 | 31751s | train_loss=0.0124
334
+ step 3320000 | 31835s | train_loss=0.0129
335
+ step 3330000 | 31916s | train_loss=0.0126
336
+ step 3340000 | 32005s | train_loss=0.0123
337
+ step 3350000 | 32095s | train_loss=0.0132
338
+ step 3360000 | 32184s | train_loss=0.0124
339
+ step 3370000 | 32274s | train_loss=0.0130
340
+ step 3380000 | 32363s | train_loss=0.0127
341
+ step 3390000 | 32450s | train_loss=0.0124
342
+ step 3400000 | 32612s | train_loss=0.0125 | fid=52.1386
343
+ step 3410000 | 32702s | train_loss=0.0128
344
+ step 3420000 | 32789s | train_loss=0.0124
345
+ step 3430000 | 32876s | train_loss=0.0120
346
+ step 3440000 | 32964s | train_loss=0.0122
347
+ step 3450000 | 33053s | train_loss=0.0121
348
+ step 3460000 | 33142s | train_loss=0.0120
349
+ step 3470000 | 33232s | train_loss=0.0120
350
+ step 3480000 | 33321s | train_loss=0.0124
351
+ step 3490000 | 33408s | train_loss=0.0124
352
+ step 3500000 | 33569s | train_loss=0.0121 | fid=51.5588
353
+ step 3510000 | 33657s | train_loss=0.0122
354
+ step 3520000 | 33745s | train_loss=0.0121
355
+ step 3530000 | 33823s | train_loss=0.0123
356
+ step 3540000 | 33913s | train_loss=0.0120
357
+ step 3550000 | 34002s | train_loss=0.0116
358
+ step 3560000 | 34093s | train_loss=0.0127
359
+ step 3570000 | 34184s | train_loss=0.0120
360
+ step 3580000 | 34266s | train_loss=0.0123
361
+ step 3590000 | 34356s | train_loss=0.0125
362
+ step 3600000 | 34518s | train_loss=0.0118 | fid=51.3932
363
+ step 3610000 | 34597s | train_loss=0.0117
364
+ step 3620000 | 34686s | train_loss=0.0121
365
+ step 3630000 | 34761s | train_loss=0.0114
366
+ step 3640000 | 34851s | train_loss=0.0121
367
+ step 3650000 | 34942s | train_loss=0.0121
368
+ step 3660000 | 35033s | train_loss=0.0116
369
+ step 3670000 | 35123s | train_loss=0.0114
370
+ step 3680000 | 35199s | train_loss=0.0116
371
+ step 3690000 | 35289s | train_loss=0.0117
372
+ step 3700000 | 35451s | train_loss=0.0112 | fid=52.1657
373
+ step 3710000 | 35530s | train_loss=0.0121
374
+ step 3720000 | 35621s | train_loss=0.0114
375
+ step 3730000 | 35702s | train_loss=0.0113
376
+ step 3740000 | 35791s | train_loss=0.0119
377
+ step 3750000 | 35881s | train_loss=0.0129
378
+ step 3760000 | 35972s | train_loss=0.0115
379
+ step 3770000 | 36061s | train_loss=0.0109
380
+ step 3780000 | 36140s | train_loss=0.0114
381
+ step 3790000 | 36231s | train_loss=0.0117
382
+ step 3800000 | 36387s | train_loss=0.0109 | fid=52.0835
383
+ step 3810000 | 36467s | train_loss=0.0111
384
+ step 3820000 | 36557s | train_loss=0.0113
385
+ step 3830000 | 36640s | train_loss=0.0114
386
+ step 3840000 | 36724s | train_loss=0.0114
387
+ step 3850000 | 36814s | train_loss=0.0110
388
+ step 3860000 | 36905s | train_loss=0.0112
389
+ step 3870000 | 36991s | train_loss=0.0110
390
+ step 3880000 | 37072s | train_loss=0.0114
391
+ step 3890000 | 37163s | train_loss=0.0113
392
+ step 3900000 | 37320s | train_loss=0.0109 | fid=52.1548
393
+ step 3910000 | 37401s | train_loss=0.0109
394
+ step 3920000 | 37492s | train_loss=0.0111
395
+ step 3930000 | 37580s | train_loss=0.0114
396
+ step 3940000 | 37658s | train_loss=0.0109
397
+ step 3950000 | 37748s | train_loss=0.0108
398
+ step 3960000 | 37839s | train_loss=0.0107
399
+ step 3970000 | 37922s | train_loss=0.0111
400
+ step 3980000 | 38006s | train_loss=0.0105
401
+ step 3990000 | 38095s | train_loss=0.0109
402
+ step 4000000 | 38253s | train_loss=0.0107 | fid=51.4698
403
+ step 4010000 | 38334s | train_loss=0.0106
404
+ step 4020000 | 38424s | train_loss=0.0109
405
+ step 4030000 | 38515s | train_loss=0.0106
406
+ step 4040000 | 38594s | train_loss=0.0106
407
+ step 4050000 | 38685s | train_loss=0.0109
408
+ step 4060000 | 38775s | train_loss=0.0107
409
+ step 4070000 | 38856s | train_loss=0.0103
410
+ step 4080000 | 38939s | train_loss=0.0109
411
+ step 4090000 | 39030s | train_loss=0.0104
412
+ step 4100000 | 39190s | train_loss=0.0104 | fid=51.9780
413
+ step 4110000 | 39271s | train_loss=0.0107
414
+ step 4120000 | 39361s | train_loss=0.0108
415
+ step 4130000 | 39452s | train_loss=0.0106
416
+ step 4140000 | 39533s | train_loss=0.0105
417
+ step 4150000 | 39624s | train_loss=0.0106
418
+ step 4160000 | 39714s | train_loss=0.0101
419
+ step 4170000 | 39793s | train_loss=0.0102
420
+ step 4180000 | 39883s | train_loss=0.0104
421
+ step 4190000 | 39973s | train_loss=0.0106
422
+ step 4200000 | 40133s | train_loss=0.0106 | fid=52.7970
423
+ step 4210000 | 40214s | train_loss=0.0102
424
+ step 4220000 | 40304s | train_loss=0.0107
425
+ step 4230000 | 40395s | train_loss=0.0107
426
+ step 4240000 | 40474s | train_loss=0.0102
427
+ step 4250000 | 40565s | train_loss=0.0100
428
+ step 4260000 | 40655s | train_loss=0.0104
429
+ step 4270000 | 40732s | train_loss=0.0100
430
+ step 4280000 | 40823s | train_loss=0.0108
431
+ step 4290000 | 40913s | train_loss=0.0104
432
+ step 4300000 | 41072s | train_loss=0.0100 | fid=51.8571
433
+ step 4310000 | 41154s | train_loss=0.0097
434
+ step 4320000 | 41242s | train_loss=0.0101
435
+ step 4330000 | 41329s | train_loss=0.0100
436
+ step 4340000 | 41411s | train_loss=0.0102
437
+ step 4350000 | 41498s | train_loss=0.0100
438
+ step 4360000 | 41580s | train_loss=0.0098
439
+ step 4370000 | 41665s | train_loss=0.0098
440
+ step 4380000 | 41753s | train_loss=0.0100
441
+ step 4390000 | 41841s | train_loss=0.0100
442
+ step 4400000 | 41998s | train_loss=0.0097 | fid=51.9293
443
+ step 4410000 | 42085s | train_loss=0.0104
444
+ step 4420000 | 42173s | train_loss=0.0097
445
+ step 4430000 | 42260s | train_loss=0.0097
446
+ step 4440000 | 42344s | train_loss=0.0100
447
+ step 4450000 | 42429s | train_loss=0.0098
448
+ step 4460000 | 42513s | train_loss=0.0104
449
+ step 4470000 | 42601s | train_loss=0.0098
450
+ step 4480000 | 42688s | train_loss=0.0099
451
+ step 4490000 | 42776s | train_loss=0.0097
452
+ step 4500000 | 42929s | train_loss=0.0096 | fid=51.9711
453
+ step 4510000 | 43017s | train_loss=0.0097
454
+ step 4520000 | 43104s | train_loss=0.0101
455
+ step 4530000 | 43192s | train_loss=0.0097
456
+ step 4540000 | 43276s | train_loss=0.0094
457
+ step 4550000 | 43362s | train_loss=0.0101
458
+ step 4560000 | 43446s | train_loss=0.0096
459
+ step 4570000 | 43533s | train_loss=0.0095
460
+ step 4580000 | 43621s | train_loss=0.0095
461
+ step 4590000 | 43709s | train_loss=0.0099
462
+ step 4600000 | 43864s | train_loss=0.0100 | fid=51.8202
463
+ step 4610000 | 43952s | train_loss=0.0097
464
+ step 4620000 | 44039s | train_loss=0.0092
465
+ step 4630000 | 44126s | train_loss=0.0093
466
+ step 4640000 | 44198s | train_loss=0.0096
467
+ step 4650000 | 44265s | train_loss=0.0096
468
+ step 4660000 | 44335s | train_loss=0.0096
469
+ step 4670000 | 44405s | train_loss=0.0094
470
+ step 4680000 | 44475s | train_loss=0.0095
471
+ step 4690000 | 44545s | train_loss=0.0093
472
+ step 4700000 | 44670s | train_loss=0.0096 | fid=52.1596
473
+ step 4710000 | 44740s | train_loss=0.0095
474
+ step 4720000 | 44811s | train_loss=0.0094
475
+ step 4730000 | 44881s | train_loss=0.0091
476
+ step 4740000 | 44947s | train_loss=0.0092
477
+ step 4750000 | 45003s | train_loss=0.0090
478
+ step 4760000 | 45056s | train_loss=0.0095
479
+ step 4770000 | 45109s | train_loss=0.0091
480
+ step 4780000 | 45162s | train_loss=0.0091
481
+ step 4790000 | 45214s | train_loss=0.0094
482
+ step 4800000 | 45283s | train_loss=0.0089 | fid=51.4009
483
+ step 4810000 | 45318s | train_loss=0.0092
484
+ step 4820000 | 45354s | train_loss=0.0093
485
+ step 4830000 | 45389s | train_loss=0.0095
486
+ step 4840000 | 45417s | train_loss=0.0091
487
+ step 4850000 | 45452s | train_loss=0.0090
488
+ step 4860000 | 45488s | train_loss=0.0094
489
+ step 4870000 | 45523s | train_loss=0.0090
490
+ step 4880000 | 45559s | train_loss=0.0087
491
+ step 4890000 | 45594s | train_loss=0.0088
492
+ step 4900000 | 45662s | train_loss=0.0093 | fid=51.8569
493
+ step 4910000 | 45698s | train_loss=0.0088
494
+ step 4920000 | 45733s | train_loss=0.0092
495
+ step 4930000 | 45769s | train_loss=0.0087
496
+ step 4940000 | 45790s | train_loss=0.0089
497
+ step 4950000 | 45808s | train_loss=0.0090
498
+ step 4960000 | 45826s | train_loss=0.0086
499
+ step 4970000 | 45844s | train_loss=0.0090
500
+ step 4980000 | 45861s | train_loss=0.0088
501
+ step 4990000 | 45879s | train_loss=0.0091
502
+ step 5000000 | 45912s | train_loss=0.0090 | fid=51.4152
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d50/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d50/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 68s | train_loss=0.9927 | fid=244.7512
3
+ step 10000 | 153s | train_loss=0.8471
4
+ step 20000 | 238s | train_loss=0.7901
5
+ step 30000 | 322s | train_loss=0.7044
6
+ step 40000 | 407s | train_loss=0.5757
7
+ step 50000 | 492s | train_loss=0.4133
8
+ step 60000 | 577s | train_loss=0.2892
9
+ step 70000 | 661s | train_loss=0.2186
10
+ step 80000 | 746s | train_loss=0.1796
11
+ step 90000 | 831s | train_loss=0.1594
12
+ step 100000 | 977s | train_loss=0.1447 | fid=58.4990
13
+ step 110000 | 1059s | train_loss=0.1332
14
+ step 120000 | 1143s | train_loss=0.1252
15
+ step 130000 | 1228s | train_loss=0.1163
16
+ step 140000 | 1313s | train_loss=0.1094
17
+ step 150000 | 1398s | train_loss=0.1061
18
+ step 160000 | 1482s | train_loss=0.1010
19
+ step 170000 | 1567s | train_loss=0.0958
20
+ step 180000 | 1652s | train_loss=0.0916
21
+ step 190000 | 1737s | train_loss=0.0877
22
+ step 200000 | 1883s | train_loss=0.0843 | fid=56.8484
23
+ step 210000 | 1966s | train_loss=0.0824
24
+ step 220000 | 2049s | train_loss=0.0787
25
+ step 230000 | 2133s | train_loss=0.0769
26
+ step 240000 | 2218s | train_loss=0.0746
27
+ step 250000 | 2303s | train_loss=0.0729
28
+ step 260000 | 2387s | train_loss=0.0698
29
+ step 270000 | 2472s | train_loss=0.0687
30
+ step 280000 | 2557s | train_loss=0.0669
31
+ step 290000 | 2642s | train_loss=0.0655
32
+ step 300000 | 2789s | train_loss=0.0645 | fid=54.4505
33
+ step 310000 | 2872s | train_loss=0.0629
34
+ step 320000 | 2957s | train_loss=0.0613
35
+ step 330000 | 3040s | train_loss=0.0597
36
+ step 340000 | 3124s | train_loss=0.0587
37
+ step 350000 | 3209s | train_loss=0.0574
38
+ step 360000 | 3294s | train_loss=0.0569
39
+ step 370000 | 3379s | train_loss=0.0559
40
+ step 380000 | 3463s | train_loss=0.0548
41
+ step 390000 | 3548s | train_loss=0.0538
42
+ step 400000 | 3695s | train_loss=0.0524 | fid=53.2654
43
+ step 410000 | 3780s | train_loss=0.0519
44
+ step 420000 | 3863s | train_loss=0.0510
45
+ step 430000 | 3946s | train_loss=0.0505
46
+ step 440000 | 4031s | train_loss=0.0497
47
+ step 450000 | 4115s | train_loss=0.0488
48
+ step 460000 | 4200s | train_loss=0.0492
49
+ step 470000 | 4285s | train_loss=0.0472
50
+ step 480000 | 4369s | train_loss=0.0477
51
+ step 490000 | 4454s | train_loss=0.0467
52
+ step 500000 | 4600s | train_loss=0.0454 | fid=50.9748
53
+ step 510000 | 4685s | train_loss=0.0457
54
+ step 520000 | 4766s | train_loss=0.0450
55
+ step 530000 | 4851s | train_loss=0.0445
56
+ step 540000 | 4935s | train_loss=0.0443
57
+ step 550000 | 5020s | train_loss=0.0434
58
+ step 560000 | 5104s | train_loss=0.0443
59
+ step 570000 | 5189s | train_loss=0.0433
60
+ step 580000 | 5274s | train_loss=0.0432
61
+ step 590000 | 5359s | train_loss=0.0420
62
+ step 600000 | 5507s | train_loss=0.0415 | fid=51.2633
63
+ step 610000 | 5592s | train_loss=0.0426
64
+ step 620000 | 5675s | train_loss=0.0411
65
+ step 630000 | 5759s | train_loss=0.0406
66
+ step 640000 | 5845s | train_loss=0.0404
67
+ step 650000 | 5927s | train_loss=0.0402
68
+ step 660000 | 6012s | train_loss=0.0394
69
+ step 670000 | 6097s | train_loss=0.0385
70
+ step 680000 | 6181s | train_loss=0.0388
71
+ step 690000 | 6266s | train_loss=0.0386
72
+ step 700000 | 6415s | train_loss=0.0377 | fid=50.9390
73
+ step 710000 | 6500s | train_loss=0.0376
74
+ step 720000 | 6585s | train_loss=0.0378
75
+ step 730000 | 6667s | train_loss=0.0373
76
+ step 740000 | 6752s | train_loss=0.0366
77
+ step 750000 | 6835s | train_loss=0.0366
78
+ step 760000 | 6920s | train_loss=0.0371
79
+ step 770000 | 7005s | train_loss=0.0355
80
+ step 780000 | 7090s | train_loss=0.0356
81
+ step 790000 | 7174s | train_loss=0.0352
82
+ step 800000 | 7323s | train_loss=0.0357 | fid=49.8402
83
+ step 810000 | 7408s | train_loss=0.0350
84
+ step 820000 | 7493s | train_loss=0.0347
85
+ step 830000 | 7574s | train_loss=0.0345
86
+ step 840000 | 7659s | train_loss=0.0339
87
+ step 850000 | 7744s | train_loss=0.0337
88
+ step 860000 | 7827s | train_loss=0.0340
89
+ step 870000 | 7912s | train_loss=0.0330
90
+ step 880000 | 7997s | train_loss=0.0333
91
+ step 890000 | 8078s | train_loss=0.0336
92
+ step 900000 | 8231s | train_loss=0.0325 | fid=48.3976
93
+ step 910000 | 8315s | train_loss=0.0323
94
+ step 920000 | 8400s | train_loss=0.0322
95
+ step 930000 | 8482s | train_loss=0.0328
96
+ step 940000 | 8567s | train_loss=0.0319
97
+ step 950000 | 8651s | train_loss=0.0319
98
+ step 960000 | 8737s | train_loss=0.0309
99
+ step 970000 | 8820s | train_loss=0.0311
100
+ step 980000 | 8904s | train_loss=0.0313
101
+ step 990000 | 8986s | train_loss=0.0308
102
+ step 1000000 | 9139s | train_loss=0.0309 | fid=48.6847
103
+ step 1010000 | 9224s | train_loss=0.0303
104
+ step 1020000 | 9308s | train_loss=0.0317
105
+ step 1030000 | 9393s | train_loss=0.0301
106
+ step 1040000 | 9475s | train_loss=0.0300
107
+ step 1050000 | 9560s | train_loss=0.0301
108
+ step 1060000 | 9644s | train_loss=0.0301
109
+ step 1070000 | 9728s | train_loss=0.0299
110
+ step 1080000 | 9812s | train_loss=0.0293
111
+ step 1090000 | 9894s | train_loss=0.0288
112
+ step 1100000 | 10046s | train_loss=0.0293 | fid=48.8198
113
+ step 1110000 | 10131s | train_loss=0.0289
114
+ step 1120000 | 10216s | train_loss=0.0290
115
+ step 1130000 | 10301s | train_loss=0.0282
116
+ step 1140000 | 10382s | train_loss=0.0284
117
+ step 1150000 | 10467s | train_loss=0.0298
118
+ step 1160000 | 10552s | train_loss=0.0280
119
+ step 1170000 | 10637s | train_loss=0.0277
120
+ step 1180000 | 10720s | train_loss=0.0274
121
+ step 1190000 | 10801s | train_loss=0.0278
122
+ step 1200000 | 10954s | train_loss=0.0275 | fid=47.7317
123
+ step 1210000 | 11039s | train_loss=0.0275
124
+ step 1220000 | 11123s | train_loss=0.0270
125
+ step 1230000 | 11208s | train_loss=0.0271
126
+ step 1240000 | 11290s | train_loss=0.0274
127
+ step 1250000 | 11375s | train_loss=0.0267
128
+ step 1260000 | 11459s | train_loss=0.0267
129
+ step 1270000 | 11544s | train_loss=0.0269
130
+ step 1280000 | 11628s | train_loss=0.0265
131
+ step 1290000 | 11709s | train_loss=0.0261
132
+ step 1300000 | 11859s | train_loss=0.0275 | fid=47.2035
133
+ step 1310000 | 11944s | train_loss=0.0262
134
+ step 1320000 | 12029s | train_loss=0.0260
135
+ step 1330000 | 12113s | train_loss=0.0263
136
+ step 1340000 | 12198s | train_loss=0.0258
137
+ step 1350000 | 12280s | train_loss=0.0261
138
+ step 1360000 | 12365s | train_loss=0.0259
139
+ step 1370000 | 12450s | train_loss=0.0253
140
+ step 1380000 | 12532s | train_loss=0.0257
141
+ step 1390000 | 12616s | train_loss=0.0252
142
+ step 1400000 | 12765s | train_loss=0.0249 | fid=46.7165
143
+ step 1410000 | 12849s | train_loss=0.0249
144
+ step 1420000 | 12934s | train_loss=0.0252
145
+ step 1430000 | 13019s | train_loss=0.0246
146
+ step 1440000 | 13104s | train_loss=0.0248
147
+ step 1450000 | 13185s | train_loss=0.0246
148
+ step 1460000 | 13270s | train_loss=0.0236
149
+ step 1470000 | 13355s | train_loss=0.0247
150
+ step 1480000 | 13436s | train_loss=0.0243
151
+ step 1490000 | 13521s | train_loss=0.0236
152
+ step 1500000 | 13671s | train_loss=0.0240 | fid=47.1075
153
+ step 1510000 | 13756s | train_loss=0.0240
154
+ step 1520000 | 13841s | train_loss=0.0239
155
+ step 1530000 | 13925s | train_loss=0.0242
156
+ step 1540000 | 14010s | train_loss=0.0239
157
+ step 1550000 | 14093s | train_loss=0.0232
158
+ step 1560000 | 14176s | train_loss=0.0241
159
+ step 1570000 | 14261s | train_loss=0.0232
160
+ step 1580000 | 14342s | train_loss=0.0228
161
+ step 1590000 | 14427s | train_loss=0.0227
162
+ step 1600000 | 14578s | train_loss=0.0229 | fid=47.1521
163
+ step 1610000 | 14663s | train_loss=0.0229
164
+ step 1620000 | 14747s | train_loss=0.0225
165
+ step 1630000 | 14832s | train_loss=0.0225
166
+ step 1640000 | 14917s | train_loss=0.0221
167
+ step 1650000 | 15002s | train_loss=0.0231
168
+ step 1660000 | 15083s | train_loss=0.0224
169
+ step 1670000 | 15168s | train_loss=0.0225
170
+ step 1680000 | 15247s | train_loss=0.0222
171
+ step 1690000 | 15332s | train_loss=0.0225
172
+ step 1700000 | 15522s | train_loss=0.0224 | fid=47.6420
173
+ step 1710000 | 15607s | train_loss=0.0220
174
+ step 1720000 | 15691s | train_loss=0.0228
175
+ step 1730000 | 15776s | train_loss=0.0217
176
+ step 1740000 | 15861s | train_loss=0.0219
177
+ step 1750000 | 15943s | train_loss=0.0216
178
+ step 1760000 | 16017s | train_loss=0.0209
179
+ step 1770000 | 16095s | train_loss=0.0212
180
+ step 1780000 | 16180s | train_loss=0.0215
181
+ step 1790000 | 16261s | train_loss=0.0209
182
+ step 1800000 | 16417s | train_loss=0.0208 | fid=46.9522
183
+ step 1810000 | 16501s | train_loss=0.0212
184
+ step 1820000 | 16586s | train_loss=0.0207
185
+ step 1830000 | 16671s | train_loss=0.0206
186
+ step 1840000 | 16755s | train_loss=0.0207
187
+ step 1850000 | 16840s | train_loss=0.0206
188
+ step 1860000 | 16922s | train_loss=0.0202
189
+ step 1870000 | 17003s | train_loss=0.0203
190
+ step 1880000 | 17088s | train_loss=0.0205
191
+ step 1890000 | 17170s | train_loss=0.0204
192
+ step 1900000 | 17325s | train_loss=0.0225 | fid=46.8620
193
+ step 1910000 | 17408s | train_loss=0.0202
194
+ step 1920000 | 17493s | train_loss=0.0197
195
+ step 1930000 | 17577s | train_loss=0.0194
196
+ step 1940000 | 17662s | train_loss=0.0194
197
+ step 1950000 | 17747s | train_loss=0.0196
198
+ step 1960000 | 17831s | train_loss=0.0193
199
+ step 1970000 | 17909s | train_loss=0.0200
200
+ step 1980000 | 17994s | train_loss=0.0195
201
+ step 1990000 | 18075s | train_loss=0.0192
202
+ step 2000000 | 18230s | train_loss=0.0200 | fid=46.4135
203
+ step 2010000 | 18315s | train_loss=0.0201
204
+ step 2020000 | 18398s | train_loss=0.0190
205
+ step 2030000 | 18483s | train_loss=0.0187
206
+ step 2040000 | 18568s | train_loss=0.0193
207
+ step 2050000 | 18652s | train_loss=0.0192
208
+ step 2060000 | 18737s | train_loss=0.0191
209
+ step 2070000 | 18815s | train_loss=0.0188
210
+ step 2080000 | 18900s | train_loss=0.0187
211
+ step 2090000 | 18982s | train_loss=0.0188
212
+ step 2100000 | 19135s | train_loss=0.0185 | fid=46.7264
213
+ step 2110000 | 19220s | train_loss=0.0186
214
+ step 2120000 | 19303s | train_loss=0.0182
215
+ step 2130000 | 19388s | train_loss=0.0192
216
+ step 2140000 | 19472s | train_loss=0.0183
217
+ step 2150000 | 19557s | train_loss=0.0182
218
+ step 2160000 | 19638s | train_loss=0.0183
219
+ step 2170000 | 19717s | train_loss=0.0180
220
+ step 2180000 | 19801s | train_loss=0.0179
221
+ step 2190000 | 19883s | train_loss=0.0177
222
+ step 2200000 | 20038s | train_loss=0.0174 | fid=46.1692
223
+ step 2210000 | 20123s | train_loss=0.0176
224
+ step 2220000 | 20208s | train_loss=0.0181
225
+ step 2230000 | 20291s | train_loss=0.0179
226
+ step 2240000 | 20376s | train_loss=0.0177
227
+ step 2250000 | 20460s | train_loss=0.0180
228
+ step 2260000 | 20542s | train_loss=0.0182
229
+ step 2270000 | 20627s | train_loss=0.0170
230
+ step 2280000 | 20709s | train_loss=0.0176
231
+ step 2290000 | 20790s | train_loss=0.0169
232
+ step 2300000 | 20945s | train_loss=0.0172 | fid=47.0484
233
+ step 2310000 | 21029s | train_loss=0.0171
234
+ step 2320000 | 21114s | train_loss=0.0177
235
+ step 2330000 | 21197s | train_loss=0.0168
236
+ step 2340000 | 21282s | train_loss=0.0172
237
+ step 2350000 | 21366s | train_loss=0.0166
238
+ step 2360000 | 21448s | train_loss=0.0175
239
+ step 2370000 | 21532s | train_loss=0.0171
240
+ step 2380000 | 21614s | train_loss=0.0169
241
+ step 2390000 | 21696s | train_loss=0.0167
242
+ step 2400000 | 21851s | train_loss=0.0164 | fid=47.9304
243
+ step 2410000 | 21936s | train_loss=0.0162
244
+ step 2420000 | 22021s | train_loss=0.0160
245
+ step 2430000 | 22106s | train_loss=0.0164
246
+ step 2440000 | 22189s | train_loss=0.0171
247
+ step 2450000 | 22274s | train_loss=0.0157
248
+ step 2460000 | 22355s | train_loss=0.0167
249
+ step 2470000 | 22440s | train_loss=0.0177
250
+ step 2480000 | 22521s | train_loss=0.0163
251
+ step 2490000 | 22603s | train_loss=0.0161
252
+ step 2500000 | 22758s | train_loss=0.0157 | fid=46.6554
253
+ step 2510000 | 22843s | train_loss=0.0151
254
+ step 2520000 | 22927s | train_loss=0.0159
255
+ step 2530000 | 23012s | train_loss=0.0159
256
+ step 2540000 | 23097s | train_loss=0.0159
257
+ step 2550000 | 23180s | train_loss=0.0159
258
+ step 2560000 | 23262s | train_loss=0.0158
259
+ step 2570000 | 23347s | train_loss=0.0152
260
+ step 2580000 | 23430s | train_loss=0.0158
261
+ step 2590000 | 23509s | train_loss=0.0151
262
+ step 2600000 | 23662s | train_loss=0.0154 | fid=47.2406
263
+ step 2610000 | 23746s | train_loss=0.0150
264
+ step 2620000 | 23831s | train_loss=0.0154
265
+ step 2630000 | 23915s | train_loss=0.0146
266
+ step 2640000 | 24000s | train_loss=0.0153
267
+ step 2650000 | 24081s | train_loss=0.0155
268
+ step 2660000 | 24166s | train_loss=0.0147
269
+ step 2670000 | 24250s | train_loss=0.0147
270
+ step 2680000 | 24333s | train_loss=0.0149
271
+ step 2690000 | 24413s | train_loss=0.0148
272
+ step 2700000 | 24568s | train_loss=0.0148 | fid=47.0615
273
+ step 2710000 | 24653s | train_loss=0.0155
274
+ step 2720000 | 24737s | train_loss=0.0146
275
+ step 2730000 | 24822s | train_loss=0.0152
276
+ step 2740000 | 24906s | train_loss=0.0141
277
+ step 2750000 | 24989s | train_loss=0.0146
278
+ step 2760000 | 25072s | train_loss=0.0146
279
+ step 2770000 | 25156s | train_loss=0.0149
280
+ step 2780000 | 25238s | train_loss=0.0138
281
+ step 2790000 | 25320s | train_loss=0.0147
282
+ step 2800000 | 25474s | train_loss=0.0142 | fid=46.6341
283
+ step 2810000 | 25559s | train_loss=0.0149
284
+ step 2820000 | 25643s | train_loss=0.0145
285
+ step 2830000 | 25728s | train_loss=0.0139
286
+ step 2840000 | 25812s | train_loss=0.0139
287
+ step 2850000 | 25895s | train_loss=0.0150
288
+ step 2860000 | 25978s | train_loss=0.0148
289
+ step 2870000 | 26062s | train_loss=0.0138
290
+ step 2880000 | 26144s | train_loss=0.0137
291
+ step 2890000 | 26228s | train_loss=0.0134
292
+ step 2900000 | 26378s | train_loss=0.0133 | fid=47.0993
293
+ step 2910000 | 26462s | train_loss=0.0136
294
+ step 2920000 | 26547s | train_loss=0.0143
295
+ step 2930000 | 26631s | train_loss=0.0133
296
+ step 2940000 | 26716s | train_loss=0.0131
297
+ step 2950000 | 26798s | train_loss=0.0129
298
+ step 2960000 | 26883s | train_loss=0.0142
299
+ step 2970000 | 26966s | train_loss=0.0139
300
+ step 2980000 | 27047s | train_loss=0.0131
301
+ step 2990000 | 27132s | train_loss=0.0136
302
+ step 3000000 | 27284s | train_loss=0.0134 | fid=47.4228
303
+ step 3010000 | 27368s | train_loss=0.0135
304
+ step 3020000 | 27453s | train_loss=0.0132
305
+ step 3030000 | 27537s | train_loss=0.0131
306
+ step 3040000 | 27620s | train_loss=0.0141
307
+ step 3050000 | 27705s | train_loss=0.0133
308
+ step 3060000 | 27789s | train_loss=0.0137
309
+ step 3070000 | 27873s | train_loss=0.0133
310
+ step 3080000 | 27954s | train_loss=0.0135
311
+ step 3090000 | 28039s | train_loss=0.0126
312
+ step 3100000 | 28191s | train_loss=0.0136 | fid=47.4502
313
+ step 3110000 | 28276s | train_loss=0.0139
314
+ step 3120000 | 28360s | train_loss=0.0130
315
+ step 3130000 | 28445s | train_loss=0.0126
316
+ step 3140000 | 28526s | train_loss=0.0127
317
+ step 3150000 | 28610s | train_loss=0.0129
318
+ step 3160000 | 28694s | train_loss=0.0127
319
+ step 3170000 | 28778s | train_loss=0.0124
320
+ step 3180000 | 28858s | train_loss=0.0125
321
+ step 3190000 | 28943s | train_loss=0.0130
322
+ step 3200000 | 29095s | train_loss=0.0129 | fid=47.3526
323
+ step 3210000 | 29180s | train_loss=0.0125
324
+ step 3220000 | 29265s | train_loss=0.0123
325
+ step 3230000 | 29349s | train_loss=0.0129
326
+ step 3240000 | 29432s | train_loss=0.0123
327
+ step 3250000 | 29516s | train_loss=0.0126
328
+ step 3260000 | 29599s | train_loss=0.0122
329
+ step 3270000 | 29683s | train_loss=0.0123
330
+ step 3280000 | 29763s | train_loss=0.0130
331
+ step 3290000 | 29848s | train_loss=0.0120
332
+ step 3300000 | 30000s | train_loss=0.0123 | fid=47.2246
333
+ step 3310000 | 30085s | train_loss=0.0120
334
+ step 3320000 | 30170s | train_loss=0.0123
335
+ step 3330000 | 30254s | train_loss=0.0118
336
+ step 3340000 | 30337s | train_loss=0.0124
337
+ step 3350000 | 30422s | train_loss=0.0131
338
+ step 3360000 | 30507s | train_loss=0.0124
339
+ step 3370000 | 30591s | train_loss=0.0119
340
+ step 3380000 | 30674s | train_loss=0.0123
341
+ step 3390000 | 30757s | train_loss=0.0123
342
+ step 3400000 | 30933s | train_loss=0.0120 | fid=46.9517
343
+ step 3410000 | 31013s | train_loss=0.0115
344
+ step 3420000 | 31098s | train_loss=0.0121
345
+ step 3430000 | 31181s | train_loss=0.0120
346
+ step 3440000 | 31265s | train_loss=0.0114
347
+ step 3450000 | 31350s | train_loss=0.0116
348
+ step 3460000 | 31435s | train_loss=0.0118
349
+ step 3470000 | 31514s | train_loss=0.0112
350
+ step 3480000 | 31590s | train_loss=0.0118
351
+ step 3490000 | 31673s | train_loss=0.0114
352
+ step 3500000 | 31850s | train_loss=0.0111 | fid=46.2775
353
+ step 3510000 | 31930s | train_loss=0.0117
354
+ step 3520000 | 32013s | train_loss=0.0116
355
+ step 3530000 | 32098s | train_loss=0.0116
356
+ step 3540000 | 32182s | train_loss=0.0112
357
+ step 3550000 | 32267s | train_loss=0.0113
358
+ step 3560000 | 32351s | train_loss=0.0113
359
+ step 3570000 | 32435s | train_loss=0.0115
360
+ step 3580000 | 32518s | train_loss=0.0113
361
+ step 3590000 | 32603s | train_loss=0.0112
362
+ step 3600000 | 32753s | train_loss=0.0114 | fid=46.9106
363
+ step 3610000 | 32834s | train_loss=0.0112
364
+ step 3620000 | 32917s | train_loss=0.0112
365
+ step 3630000 | 33002s | train_loss=0.0107
366
+ step 3640000 | 33086s | train_loss=0.0110
367
+ step 3650000 | 33171s | train_loss=0.0113
368
+ step 3660000 | 33255s | train_loss=0.0111
369
+ step 3670000 | 33339s | train_loss=0.0110
370
+ step 3680000 | 33422s | train_loss=0.0114
371
+ step 3690000 | 33507s | train_loss=0.0108
372
+ step 3700000 | 33659s | train_loss=0.0108 | fid=46.5416
373
+ step 3710000 | 33744s | train_loss=0.0111
374
+ step 3720000 | 33815s | train_loss=0.0107
375
+ step 3730000 | 33900s | train_loss=0.0106
376
+ step 3740000 | 33985s | train_loss=0.0112
377
+ step 3750000 | 34069s | train_loss=0.0116
378
+ step 3760000 | 34154s | train_loss=0.0115
379
+ step 3770000 | 34236s | train_loss=0.0111
380
+ step 3780000 | 34316s | train_loss=0.0105
381
+ step 3790000 | 34401s | train_loss=0.0104
382
+ step 3800000 | 34584s | train_loss=0.0105 | fid=47.8120
383
+ step 3810000 | 34667s | train_loss=0.0106
384
+ step 3820000 | 34741s | train_loss=0.0104
385
+ step 3830000 | 34821s | train_loss=0.0110
386
+ step 3840000 | 34906s | train_loss=0.0108
387
+ step 3850000 | 34991s | train_loss=0.0104
388
+ step 3860000 | 35075s | train_loss=0.0105
389
+ step 3870000 | 35153s | train_loss=0.0103
390
+ step 3880000 | 35231s | train_loss=0.0103
391
+ step 3890000 | 35315s | train_loss=0.0104
392
+ step 3900000 | 35492s | train_loss=0.0106 | fid=46.6278
393
+ step 3910000 | 35575s | train_loss=0.0105
394
+ step 3920000 | 35659s | train_loss=0.0106
395
+ step 3930000 | 35733s | train_loss=0.0106
396
+ step 3940000 | 35818s | train_loss=0.0102
397
+ step 3950000 | 35903s | train_loss=0.0104
398
+ step 3960000 | 35988s | train_loss=0.0101
399
+ step 3970000 | 36069s | train_loss=0.0109
400
+ step 3980000 | 36145s | train_loss=0.0105
401
+ step 3990000 | 36230s | train_loss=0.0100
402
+ step 4000000 | 36427s | train_loss=0.0102 | fid=46.6349
403
+ step 4010000 | 36512s | train_loss=0.0103
404
+ step 4020000 | 36597s | train_loss=0.0100
405
+ step 4030000 | 36668s | train_loss=0.0103
406
+ step 4040000 | 36752s | train_loss=0.0105
407
+ step 4050000 | 36837s | train_loss=0.0105
408
+ step 4060000 | 36922s | train_loss=0.0099
409
+ step 4070000 | 36999s | train_loss=0.0098
410
+ step 4080000 | 37078s | train_loss=0.0102
411
+ step 4090000 | 37162s | train_loss=0.0102
412
+ step 4100000 | 37357s | train_loss=0.0098 | fid=46.4307
413
+ step 4110000 | 37442s | train_loss=0.0096
414
+ step 4120000 | 37526s | train_loss=0.0100
415
+ step 4130000 | 37603s | train_loss=0.0096
416
+ step 4140000 | 37682s | train_loss=0.0099
417
+ step 4150000 | 37767s | train_loss=0.0098
418
+ step 4160000 | 37851s | train_loss=0.0098
419
+ step 4170000 | 37927s | train_loss=0.0095
420
+ step 4180000 | 38008s | train_loss=0.0096
421
+ step 4190000 | 38090s | train_loss=0.0101
422
+ step 4200000 | 38293s | train_loss=0.0099 | fid=47.6920
423
+ step 4210000 | 38378s | train_loss=0.0098
424
+ step 4220000 | 38462s | train_loss=0.0099
425
+ step 4230000 | 38544s | train_loss=0.0099
426
+ step 4240000 | 38622s | train_loss=0.0100
427
+ step 4250000 | 38706s | train_loss=0.0099
428
+ step 4260000 | 38791s | train_loss=0.0096
429
+ step 4270000 | 38864s | train_loss=0.0094
430
+ step 4280000 | 38944s | train_loss=0.0100
431
+ step 4290000 | 39029s | train_loss=0.0095
432
+ step 4300000 | 39229s | train_loss=0.0098 | fid=47.2215
433
+ step 4310000 | 39314s | train_loss=0.0094
434
+ step 4320000 | 39398s | train_loss=0.0101
435
+ step 4330000 | 39481s | train_loss=0.0100
436
+ step 4340000 | 39559s | train_loss=0.0095
437
+ step 4350000 | 39644s | train_loss=0.0095
438
+ step 4360000 | 39729s | train_loss=0.0098
439
+ step 4370000 | 39801s | train_loss=0.0092
440
+ step 4380000 | 39886s | train_loss=0.0093
441
+ step 4390000 | 39971s | train_loss=0.0093
442
+ step 4400000 | 40169s | train_loss=0.0090 | fid=47.6412
443
+ step 4410000 | 40254s | train_loss=0.0097
444
+ step 4420000 | 40339s | train_loss=0.0090
445
+ step 4430000 | 40421s | train_loss=0.0091
446
+ step 4440000 | 40498s | train_loss=0.0091
447
+ step 4450000 | 40582s | train_loss=0.0089
448
+ step 4460000 | 40662s | train_loss=0.0099
449
+ step 4470000 | 40739s | train_loss=0.0092
450
+ step 4480000 | 40824s | train_loss=0.0091
451
+ step 4490000 | 40909s | train_loss=0.0095
452
+ step 4500000 | 41105s | train_loss=0.0090 | fid=47.2975
453
+ step 4510000 | 41189s | train_loss=0.0089
454
+ step 4520000 | 41274s | train_loss=0.0094
455
+ step 4530000 | 41358s | train_loss=0.0093
456
+ step 4540000 | 41436s | train_loss=0.0090
457
+ step 4550000 | 41520s | train_loss=0.0092
458
+ step 4560000 | 41596s | train_loss=0.0090
459
+ step 4570000 | 41680s | train_loss=0.0089
460
+ step 4580000 | 41764s | train_loss=0.0088
461
+ step 4590000 | 41849s | train_loss=0.0091
462
+ step 4600000 | 42000s | train_loss=0.0089 | fid=46.4624
463
+ step 4610000 | 42084s | train_loss=0.0088
464
+ step 4620000 | 42168s | train_loss=0.0087
465
+ step 4630000 | 42252s | train_loss=0.0090
466
+ step 4640000 | 42333s | train_loss=0.0088
467
+ step 4650000 | 42415s | train_loss=0.0088
468
+ step 4660000 | 42496s | train_loss=0.0087
469
+ step 4670000 | 42580s | train_loss=0.0090
470
+ step 4680000 | 42664s | train_loss=0.0086
471
+ step 4690000 | 42748s | train_loss=0.0091
472
+ step 4700000 | 42898s | train_loss=0.0093 | fid=46.7983
473
+ step 4710000 | 42981s | train_loss=0.0086
474
+ step 4720000 | 43065s | train_loss=0.0088
475
+ step 4730000 | 43149s | train_loss=0.0088
476
+ step 4740000 | 43234s | train_loss=0.0090
477
+ step 4750000 | 43313s | train_loss=0.0090
478
+ step 4760000 | 43393s | train_loss=0.0088
479
+ step 4770000 | 43478s | train_loss=0.0083
480
+ step 4780000 | 43562s | train_loss=0.0087
481
+ step 4790000 | 43646s | train_loss=0.0089
482
+ step 4800000 | 43796s | train_loss=0.0085 | fid=47.2846
483
+ step 4810000 | 43878s | train_loss=0.0086
484
+ step 4820000 | 43963s | train_loss=0.0088
485
+ step 4830000 | 44047s | train_loss=0.0092
486
+ step 4840000 | 44131s | train_loss=0.0085
487
+ step 4850000 | 44200s | train_loss=0.0089
488
+ step 4860000 | 44264s | train_loss=0.0084
489
+ step 4870000 | 44331s | train_loss=0.0085
490
+ step 4880000 | 44398s | train_loss=0.0084
491
+ step 4890000 | 44466s | train_loss=0.0087
492
+ step 4900000 | 44586s | train_loss=0.0087 | fid=46.3799
493
+ step 4910000 | 44654s | train_loss=0.0083
494
+ step 4920000 | 44719s | train_loss=0.0085
495
+ step 4930000 | 44787s | train_loss=0.0085
496
+ step 4940000 | 44854s | train_loss=0.0085
497
+ step 4950000 | 44920s | train_loss=0.0083
498
+ step 4960000 | 44979s | train_loss=0.0089
499
+ step 4970000 | 45030s | train_loss=0.0088
500
+ step 4980000 | 45081s | train_loss=0.0086
501
+ step 4990000 | 45131s | train_loss=0.0088
502
+ step 5000000 | 45225s | train_loss=0.0083 | fid=46.8706
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d60/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d60/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 74s | train_loss=0.9916 | fid=298.4622
3
+ step 10000 | 168s | train_loss=0.8359
4
+ step 20000 | 263s | train_loss=0.7608
5
+ step 30000 | 358s | train_loss=0.6558
6
+ step 40000 | 453s | train_loss=0.5013
7
+ step 50000 | 548s | train_loss=0.3444
8
+ step 60000 | 642s | train_loss=0.2440
9
+ step 70000 | 737s | train_loss=0.1916
10
+ step 80000 | 832s | train_loss=0.1630
11
+ step 90000 | 927s | train_loss=0.1470
12
+ step 100000 | 1093s | train_loss=0.1337 | fid=56.4878
13
+ step 110000 | 1186s | train_loss=0.1256
14
+ step 120000 | 1281s | train_loss=0.1165
15
+ step 130000 | 1376s | train_loss=0.1119
16
+ step 140000 | 1472s | train_loss=0.1058
17
+ step 150000 | 1566s | train_loss=0.1008
18
+ step 160000 | 1661s | train_loss=0.0958
19
+ step 170000 | 1756s | train_loss=0.0936
20
+ step 180000 | 1851s | train_loss=0.0903
21
+ step 190000 | 1944s | train_loss=0.0862
22
+ step 200000 | 2110s | train_loss=0.0842 | fid=54.2992
23
+ step 210000 | 2203s | train_loss=0.0802
24
+ step 220000 | 2298s | train_loss=0.0781
25
+ step 230000 | 2393s | train_loss=0.0754
26
+ step 240000 | 2487s | train_loss=0.0741
27
+ step 250000 | 2582s | train_loss=0.0721
28
+ step 260000 | 2677s | train_loss=0.0691
29
+ step 270000 | 2771s | train_loss=0.0680
30
+ step 280000 | 2866s | train_loss=0.0670
31
+ step 290000 | 2958s | train_loss=0.0658
32
+ step 300000 | 3124s | train_loss=0.0653 | fid=52.7089
33
+ step 310000 | 3217s | train_loss=0.0641
34
+ step 320000 | 3312s | train_loss=0.0615
35
+ step 330000 | 3407s | train_loss=0.0602
36
+ step 340000 | 3501s | train_loss=0.0586
37
+ step 350000 | 3596s | train_loss=0.0581
38
+ step 360000 | 3690s | train_loss=0.0575
39
+ step 370000 | 3785s | train_loss=0.0564
40
+ step 380000 | 3878s | train_loss=0.0550
41
+ step 390000 | 3972s | train_loss=0.0544
42
+ step 400000 | 4138s | train_loss=0.0538 | fid=51.7992
43
+ step 410000 | 4232s | train_loss=0.0525
44
+ step 420000 | 4327s | train_loss=0.0517
45
+ step 430000 | 4421s | train_loss=0.0512
46
+ step 440000 | 4516s | train_loss=0.0506
47
+ step 450000 | 4610s | train_loss=0.0502
48
+ step 460000 | 4705s | train_loss=0.0499
49
+ step 470000 | 4800s | train_loss=0.0493
50
+ step 480000 | 4892s | train_loss=0.0482
51
+ step 490000 | 4987s | train_loss=0.0471
52
+ step 500000 | 5153s | train_loss=0.0468 | fid=50.6037
53
+ step 510000 | 5247s | train_loss=0.0474
54
+ step 520000 | 5342s | train_loss=0.0465
55
+ step 530000 | 5436s | train_loss=0.0461
56
+ step 540000 | 5530s | train_loss=0.0456
57
+ step 550000 | 5625s | train_loss=0.0447
58
+ step 560000 | 5719s | train_loss=0.0445
59
+ step 570000 | 5811s | train_loss=0.0438
60
+ step 580000 | 5906s | train_loss=0.0433
61
+ step 590000 | 6001s | train_loss=0.0429
62
+ step 600000 | 6167s | train_loss=0.0424 | fid=48.9136
63
+ step 610000 | 6260s | train_loss=0.0421
64
+ step 620000 | 6355s | train_loss=0.0423
65
+ step 630000 | 6449s | train_loss=0.0416
66
+ step 640000 | 6544s | train_loss=0.0405
67
+ step 650000 | 6638s | train_loss=0.0418
68
+ step 660000 | 6733s | train_loss=0.0405
69
+ step 670000 | 6825s | train_loss=0.0398
70
+ step 680000 | 6920s | train_loss=0.0400
71
+ step 690000 | 7015s | train_loss=0.0391
72
+ step 700000 | 7180s | train_loss=0.0392 | fid=48.2915
73
+ step 710000 | 7273s | train_loss=0.0386
74
+ step 720000 | 7367s | train_loss=0.0385
75
+ step 730000 | 7462s | train_loss=0.0385
76
+ step 740000 | 7557s | train_loss=0.0374
77
+ step 750000 | 7652s | train_loss=0.0377
78
+ step 760000 | 7743s | train_loss=0.0380
79
+ step 770000 | 7838s | train_loss=0.0372
80
+ step 780000 | 7933s | train_loss=0.0381
81
+ step 790000 | 8028s | train_loss=0.0360
82
+ step 800000 | 8194s | train_loss=0.0363 | fid=47.3080
83
+ step 810000 | 8287s | train_loss=0.0356
84
+ step 820000 | 8382s | train_loss=0.0354
85
+ step 830000 | 8476s | train_loss=0.0365
86
+ step 840000 | 8571s | train_loss=0.0352
87
+ step 850000 | 8663s | train_loss=0.0355
88
+ step 860000 | 8758s | train_loss=0.0344
89
+ step 870000 | 8853s | train_loss=0.0339
90
+ step 880000 | 8948s | train_loss=0.0340
91
+ step 890000 | 9043s | train_loss=0.0349
92
+ step 900000 | 9209s | train_loss=0.0340 | fid=47.0582
93
+ step 910000 | 9303s | train_loss=0.0331
94
+ step 920000 | 9398s | train_loss=0.0331
95
+ step 930000 | 9492s | train_loss=0.0334
96
+ step 940000 | 9587s | train_loss=0.0328
97
+ step 950000 | 9679s | train_loss=0.0330
98
+ step 960000 | 9774s | train_loss=0.0323
99
+ step 970000 | 9868s | train_loss=0.0321
100
+ step 980000 | 9962s | train_loss=0.0323
101
+ step 990000 | 10057s | train_loss=0.0316
102
+ step 1000000 | 10223s | train_loss=0.0318 | fid=46.4680
103
+ step 1010000 | 10319s | train_loss=0.0314
104
+ step 1020000 | 10412s | train_loss=0.0322
105
+ step 1030000 | 10506s | train_loss=0.0309
106
+ step 1040000 | 10598s | train_loss=0.0302
107
+ step 1050000 | 10693s | train_loss=0.0309
108
+ step 1060000 | 10787s | train_loss=0.0320
109
+ step 1070000 | 10882s | train_loss=0.0303
110
+ step 1080000 | 10976s | train_loss=0.0303
111
+ step 1090000 | 11071s | train_loss=0.0294
112
+ step 1100000 | 11238s | train_loss=0.0303 | fid=47.4125
113
+ step 1110000 | 11334s | train_loss=0.0295
114
+ step 1120000 | 11427s | train_loss=0.0299
115
+ step 1130000 | 11521s | train_loss=0.0303
116
+ step 1140000 | 11613s | train_loss=0.0304
117
+ step 1150000 | 11708s | train_loss=0.0298
118
+ step 1160000 | 11802s | train_loss=0.0290
119
+ step 1170000 | 11896s | train_loss=0.0292
120
+ step 1180000 | 11991s | train_loss=0.0289
121
+ step 1190000 | 12086s | train_loss=0.0281
122
+ step 1200000 | 12252s | train_loss=0.0288 | fid=45.1623
123
+ step 1210000 | 12347s | train_loss=0.0285
124
+ step 1220000 | 12441s | train_loss=0.0281
125
+ step 1230000 | 12532s | train_loss=0.0281
126
+ step 1240000 | 12627s | train_loss=0.0279
127
+ step 1250000 | 12721s | train_loss=0.0279
128
+ step 1260000 | 12815s | train_loss=0.0282
129
+ step 1270000 | 12910s | train_loss=0.0278
130
+ step 1280000 | 13005s | train_loss=0.0275
131
+ step 1290000 | 13099s | train_loss=0.0269
132
+ step 1300000 | 13265s | train_loss=0.0266 | fid=45.1683
133
+ step 1310000 | 13360s | train_loss=0.0263
134
+ step 1320000 | 13451s | train_loss=0.0268
135
+ step 1330000 | 13546s | train_loss=0.0272
136
+ step 1340000 | 13640s | train_loss=0.0258
137
+ step 1350000 | 13735s | train_loss=0.0267
138
+ step 1360000 | 13830s | train_loss=0.0263
139
+ step 1370000 | 13924s | train_loss=0.0266
140
+ step 1380000 | 14019s | train_loss=0.0258
141
+ step 1390000 | 14113s | train_loss=0.0262
142
+ step 1400000 | 14279s | train_loss=0.0258 | fid=45.3896
143
+ step 1410000 | 14373s | train_loss=0.0263
144
+ step 1420000 | 14464s | train_loss=0.0253
145
+ step 1430000 | 14558s | train_loss=0.0252
146
+ step 1440000 | 14653s | train_loss=0.0256
147
+ step 1450000 | 14747s | train_loss=0.0250
148
+ step 1460000 | 14841s | train_loss=0.0244
149
+ step 1470000 | 14936s | train_loss=0.0250
150
+ step 1480000 | 15031s | train_loss=0.0244
151
+ step 1490000 | 15126s | train_loss=0.0241
152
+ step 1500000 | 15291s | train_loss=0.0246 | fid=44.8532
153
+ step 1510000 | 15377s | train_loss=0.0250
154
+ step 1520000 | 15467s | train_loss=0.0240
155
+ step 1530000 | 15559s | train_loss=0.0242
156
+ step 1540000 | 15652s | train_loss=0.0245
157
+ step 1550000 | 15746s | train_loss=0.0239
158
+ step 1560000 | 15840s | train_loss=0.0246
159
+ step 1570000 | 15934s | train_loss=0.0242
160
+ step 1580000 | 16028s | train_loss=0.0231
161
+ step 1590000 | 16122s | train_loss=0.0232
162
+ step 1600000 | 16288s | train_loss=0.0229 | fid=44.8364
163
+ step 1610000 | 16379s | train_loss=0.0228
164
+ step 1620000 | 16472s | train_loss=0.0236
165
+ step 1630000 | 16566s | train_loss=0.0231
166
+ step 1640000 | 16660s | train_loss=0.0228
167
+ step 1650000 | 16755s | train_loss=0.0231
168
+ step 1660000 | 16849s | train_loss=0.0231
169
+ step 1670000 | 16943s | train_loss=0.0223
170
+ step 1680000 | 17038s | train_loss=0.0226
171
+ step 1690000 | 17132s | train_loss=0.0227
172
+ step 1700000 | 17298s | train_loss=0.0231 | fid=44.4551
173
+ step 1710000 | 17391s | train_loss=0.0230
174
+ step 1720000 | 17484s | train_loss=0.0225
175
+ step 1730000 | 17578s | train_loss=0.0223
176
+ step 1740000 | 17672s | train_loss=0.0217
177
+ step 1750000 | 17767s | train_loss=0.0221
178
+ step 1760000 | 17861s | train_loss=0.0215
179
+ step 1770000 | 17955s | train_loss=0.0222
180
+ step 1780000 | 18049s | train_loss=0.0218
181
+ step 1790000 | 18143s | train_loss=0.0214
182
+ step 1800000 | 18306s | train_loss=0.0215 | fid=44.6249
183
+ step 1810000 | 18400s | train_loss=0.0214
184
+ step 1820000 | 18495s | train_loss=0.0216
185
+ step 1830000 | 18588s | train_loss=0.0210
186
+ step 1840000 | 18683s | train_loss=0.0216
187
+ step 1850000 | 18777s | train_loss=0.0209
188
+ step 1860000 | 18871s | train_loss=0.0207
189
+ step 1870000 | 18965s | train_loss=0.0211
190
+ step 1880000 | 19060s | train_loss=0.0203
191
+ step 1890000 | 19153s | train_loss=0.0224
192
+ step 1900000 | 19313s | train_loss=0.0223 | fid=44.5650
193
+ step 1910000 | 19407s | train_loss=0.0203
194
+ step 1920000 | 19501s | train_loss=0.0205
195
+ step 1930000 | 19593s | train_loss=0.0200
196
+ step 1940000 | 19686s | train_loss=0.0197
197
+ step 1950000 | 19779s | train_loss=0.0197
198
+ step 1960000 | 19872s | train_loss=0.0199
199
+ step 1970000 | 19965s | train_loss=0.0200
200
+ step 1980000 | 20060s | train_loss=0.0200
201
+ step 1990000 | 20154s | train_loss=0.0198
202
+ step 2000000 | 20317s | train_loss=0.0198 | fid=45.0652
203
+ step 2010000 | 20411s | train_loss=0.0194
204
+ step 2020000 | 20506s | train_loss=0.0196
205
+ step 2030000 | 20599s | train_loss=0.0190
206
+ step 2040000 | 20692s | train_loss=0.0197
207
+ step 2050000 | 20786s | train_loss=0.0192
208
+ step 2060000 | 20880s | train_loss=0.0192
209
+ step 2070000 | 20972s | train_loss=0.0193
210
+ step 2080000 | 21066s | train_loss=0.0185
211
+ step 2090000 | 21156s | train_loss=0.0195
212
+ step 2100000 | 21321s | train_loss=0.0188 | fid=44.3627
213
+ step 2110000 | 21414s | train_loss=0.0192
214
+ step 2120000 | 21507s | train_loss=0.0187
215
+ step 2130000 | 21599s | train_loss=0.0195
216
+ step 2140000 | 21693s | train_loss=0.0189
217
+ step 2150000 | 21786s | train_loss=0.0188
218
+ step 2160000 | 21880s | train_loss=0.0185
219
+ step 2170000 | 21975s | train_loss=0.0182
220
+ step 2180000 | 22069s | train_loss=0.0182
221
+ step 2190000 | 22160s | train_loss=0.0182
222
+ step 2200000 | 22328s | train_loss=0.0178 | fid=44.2017
223
+ step 2210000 | 22422s | train_loss=0.0179
224
+ step 2220000 | 22516s | train_loss=0.0179
225
+ step 2230000 | 22609s | train_loss=0.0178
226
+ step 2240000 | 22704s | train_loss=0.0178
227
+ step 2250000 | 22797s | train_loss=0.0182
228
+ step 2260000 | 22892s | train_loss=0.0178
229
+ step 2270000 | 22986s | train_loss=0.0175
230
+ step 2280000 | 23069s | train_loss=0.0180
231
+ step 2290000 | 23159s | train_loss=0.0173
232
+ step 2300000 | 23323s | train_loss=0.0178 | fid=44.6670
233
+ step 2310000 | 23416s | train_loss=0.0175
234
+ step 2320000 | 23508s | train_loss=0.0182
235
+ step 2330000 | 23600s | train_loss=0.0169
236
+ step 2340000 | 23693s | train_loss=0.0174
237
+ step 2350000 | 23786s | train_loss=0.0169
238
+ step 2360000 | 23879s | train_loss=0.0178
239
+ step 2370000 | 23973s | train_loss=0.0171
240
+ step 2380000 | 24063s | train_loss=0.0170
241
+ step 2390000 | 24156s | train_loss=0.0161
242
+ step 2400000 | 24322s | train_loss=0.0173 | fid=44.5812
243
+ step 2410000 | 24414s | train_loss=0.0167
244
+ step 2420000 | 24508s | train_loss=0.0166
245
+ step 2430000 | 24599s | train_loss=0.0167
246
+ step 2440000 | 24693s | train_loss=0.0174
247
+ step 2450000 | 24786s | train_loss=0.0160
248
+ step 2460000 | 24880s | train_loss=0.0171
249
+ step 2470000 | 24972s | train_loss=0.0176
250
+ step 2480000 | 25063s | train_loss=0.0164
251
+ step 2490000 | 25157s | train_loss=0.0166
252
+ step 2500000 | 25322s | train_loss=0.0157 | fid=44.5078
253
+ step 2510000 | 25415s | train_loss=0.0155
254
+ step 2520000 | 25507s | train_loss=0.0162
255
+ step 2530000 | 25600s | train_loss=0.0162
256
+ step 2540000 | 25691s | train_loss=0.0165
257
+ step 2550000 | 25784s | train_loss=0.0161
258
+ step 2560000 | 25877s | train_loss=0.0159
259
+ step 2570000 | 25968s | train_loss=0.0161
260
+ step 2580000 | 26061s | train_loss=0.0156
261
+ step 2590000 | 26154s | train_loss=0.0156
262
+ step 2600000 | 26320s | train_loss=0.0156 | fid=44.5514
263
+ step 2610000 | 26411s | train_loss=0.0158
264
+ step 2620000 | 26505s | train_loss=0.0152
265
+ step 2630000 | 26598s | train_loss=0.0152
266
+ step 2640000 | 26690s | train_loss=0.0155
267
+ step 2650000 | 26783s | train_loss=0.0158
268
+ step 2660000 | 26875s | train_loss=0.0151
269
+ step 2670000 | 26966s | train_loss=0.0153
270
+ step 2680000 | 27060s | train_loss=0.0154
271
+ step 2690000 | 27153s | train_loss=0.0159
272
+ step 2700000 | 27319s | train_loss=0.0157 | fid=44.6510
273
+ step 2710000 | 27410s | train_loss=0.0155
274
+ step 2720000 | 27503s | train_loss=0.0155
275
+ step 2730000 | 27597s | train_loss=0.0147
276
+ step 2740000 | 27689s | train_loss=0.0149
277
+ step 2750000 | 27782s | train_loss=0.0154
278
+ step 2760000 | 27872s | train_loss=0.0152
279
+ step 2770000 | 27966s | train_loss=0.0148
280
+ step 2780000 | 28061s | train_loss=0.0142
281
+ step 2790000 | 28155s | train_loss=0.0143
282
+ step 2800000 | 28322s | train_loss=0.0146 | fid=44.2750
283
+ step 2810000 | 28413s | train_loss=0.0149
284
+ step 2820000 | 28501s | train_loss=0.0152
285
+ step 2830000 | 28589s | train_loss=0.0142
286
+ step 2840000 | 28675s | train_loss=0.0147
287
+ step 2850000 | 28762s | train_loss=0.0155
288
+ step 2860000 | 28853s | train_loss=0.0156
289
+ step 2870000 | 28948s | train_loss=0.0142
290
+ step 2880000 | 29043s | train_loss=0.0142
291
+ step 2890000 | 29137s | train_loss=0.0139
292
+ step 2900000 | 29305s | train_loss=0.0138 | fid=43.5885
293
+ step 2910000 | 29396s | train_loss=0.0138
294
+ step 2920000 | 29485s | train_loss=0.0145
295
+ step 2930000 | 29573s | train_loss=0.0135
296
+ step 2940000 | 29660s | train_loss=0.0135
297
+ step 2950000 | 29750s | train_loss=0.0137
298
+ step 2960000 | 29844s | train_loss=0.0139
299
+ step 2970000 | 29940s | train_loss=0.0138
300
+ step 2980000 | 30034s | train_loss=0.0134
301
+ step 2990000 | 30129s | train_loss=0.0139
302
+ step 3000000 | 30297s | train_loss=0.0136 | fid=44.1270
303
+ step 3010000 | 30391s | train_loss=0.0134
304
+ step 3020000 | 30486s | train_loss=0.0133
305
+ step 3030000 | 30580s | train_loss=0.0133
306
+ step 3040000 | 30672s | train_loss=0.0153
307
+ step 3050000 | 30765s | train_loss=0.0139
308
+ step 3060000 | 30860s | train_loss=0.0137
309
+ step 3070000 | 30954s | train_loss=0.0132
310
+ step 3080000 | 31048s | train_loss=0.0134
311
+ step 3090000 | 31143s | train_loss=0.0129
312
+ step 3100000 | 31309s | train_loss=0.0136 | fid=44.3625
313
+ step 3110000 | 31402s | train_loss=0.0135
314
+ step 3120000 | 31496s | train_loss=0.0130
315
+ step 3130000 | 31589s | train_loss=0.0127
316
+ step 3140000 | 31670s | train_loss=0.0132
317
+ step 3150000 | 31760s | train_loss=0.0136
318
+ step 3160000 | 31853s | train_loss=0.0129
319
+ step 3170000 | 31946s | train_loss=0.0132
320
+ step 3180000 | 32040s | train_loss=0.0127
321
+ step 3190000 | 32134s | train_loss=0.0128
322
+ step 3200000 | 32299s | train_loss=0.0128 | fid=44.2475
323
+ step 3210000 | 32391s | train_loss=0.0129
324
+ step 3220000 | 32486s | train_loss=0.0124
325
+ step 3230000 | 32580s | train_loss=0.0133
326
+ step 3240000 | 32671s | train_loss=0.0127
327
+ step 3250000 | 32764s | train_loss=0.0127
328
+ step 3260000 | 32858s | train_loss=0.0125
329
+ step 3270000 | 32951s | train_loss=0.0130
330
+ step 3280000 | 33044s | train_loss=0.0129
331
+ step 3290000 | 33138s | train_loss=0.0120
332
+ step 3300000 | 33303s | train_loss=0.0124 | fid=44.0520
333
+ step 3310000 | 33394s | train_loss=0.0125
334
+ step 3320000 | 33488s | train_loss=0.0124
335
+ step 3330000 | 33582s | train_loss=0.0120
336
+ step 3340000 | 33675s | train_loss=0.0121
337
+ step 3350000 | 33767s | train_loss=0.0127
338
+ step 3360000 | 33861s | train_loss=0.0125
339
+ step 3370000 | 33954s | train_loss=0.0120
340
+ step 3380000 | 34049s | train_loss=0.0126
341
+ step 3390000 | 34144s | train_loss=0.0122
342
+ step 3400000 | 34302s | train_loss=0.0121 | fid=45.0200
343
+ step 3410000 | 34395s | train_loss=0.0120
344
+ step 3420000 | 34490s | train_loss=0.0125
345
+ step 3430000 | 34573s | train_loss=0.0120
346
+ step 3440000 | 34668s | train_loss=0.0115
347
+ step 3450000 | 34761s | train_loss=0.0119
348
+ step 3460000 | 34855s | train_loss=0.0120
349
+ step 3470000 | 34950s | train_loss=0.0116
350
+ step 3480000 | 35045s | train_loss=0.0122
351
+ step 3490000 | 35140s | train_loss=0.0119
352
+ step 3500000 | 35296s | train_loss=0.0114 | fid=44.3487
353
+ step 3510000 | 35389s | train_loss=0.0120
354
+ step 3520000 | 35483s | train_loss=0.0117
355
+ step 3530000 | 35565s | train_loss=0.0121
356
+ step 3540000 | 35660s | train_loss=0.0112
357
+ step 3550000 | 35752s | train_loss=0.0113
358
+ step 3560000 | 35847s | train_loss=0.0122
359
+ step 3570000 | 35942s | train_loss=0.0114
360
+ step 3580000 | 36037s | train_loss=0.0116
361
+ step 3590000 | 36131s | train_loss=0.0113
362
+ step 3600000 | 36300s | train_loss=0.0111 | fid=44.1442
363
+ step 3610000 | 36390s | train_loss=0.0114
364
+ step 3620000 | 36484s | train_loss=0.0123
365
+ step 3630000 | 36564s | train_loss=0.0112
366
+ step 3640000 | 36659s | train_loss=0.0113
367
+ step 3650000 | 36752s | train_loss=0.0113
368
+ step 3660000 | 36847s | train_loss=0.0114
369
+ step 3670000 | 36942s | train_loss=0.0112
370
+ step 3680000 | 37035s | train_loss=0.0112
371
+ step 3690000 | 37131s | train_loss=0.0113
372
+ step 3700000 | 37300s | train_loss=0.0108 | fid=45.0354
373
+ step 3710000 | 37390s | train_loss=0.0117
374
+ step 3720000 | 37485s | train_loss=0.0116
375
+ step 3730000 | 37568s | train_loss=0.0109
376
+ step 3740000 | 37662s | train_loss=0.0116
377
+ step 3750000 | 37755s | train_loss=0.0117
378
+ step 3760000 | 37850s | train_loss=0.0112
379
+ step 3770000 | 37944s | train_loss=0.0107
380
+ step 3780000 | 38039s | train_loss=0.0107
381
+ step 3790000 | 38134s | train_loss=0.0108
382
+ step 3800000 | 38303s | train_loss=0.0106 | fid=44.9143
383
+ step 3810000 | 38394s | train_loss=0.0110
384
+ step 3820000 | 38489s | train_loss=0.0109
385
+ step 3830000 | 38571s | train_loss=0.0111
386
+ step 3840000 | 38666s | train_loss=0.0115
387
+ step 3850000 | 38760s | train_loss=0.0106
388
+ step 3860000 | 38854s | train_loss=0.0110
389
+ step 3870000 | 38949s | train_loss=0.0102
390
+ step 3880000 | 39044s | train_loss=0.0107
391
+ step 3890000 | 39139s | train_loss=0.0106
392
+ step 3900000 | 39308s | train_loss=0.0104 | fid=44.4503
393
+ step 3910000 | 39399s | train_loss=0.0107
394
+ step 3920000 | 39491s | train_loss=0.0107
395
+ step 3930000 | 39576s | train_loss=0.0108
396
+ step 3940000 | 39671s | train_loss=0.0105
397
+ step 3950000 | 39764s | train_loss=0.0104
398
+ step 3960000 | 39858s | train_loss=0.0107
399
+ step 3970000 | 39953s | train_loss=0.0114
400
+ step 3980000 | 40048s | train_loss=0.0108
401
+ step 3990000 | 40142s | train_loss=0.0101
402
+ step 4000000 | 40311s | train_loss=0.0105 | fid=43.9596
403
+ step 4010000 | 40403s | train_loss=0.0101
404
+ step 4020000 | 40492s | train_loss=0.0106
405
+ step 4030000 | 40578s | train_loss=0.0110
406
+ step 4040000 | 40673s | train_loss=0.0104
407
+ step 4050000 | 40767s | train_loss=0.0109
408
+ step 4060000 | 40861s | train_loss=0.0108
409
+ step 4070000 | 40956s | train_loss=0.0099
410
+ step 4080000 | 41051s | train_loss=0.0106
411
+ step 4090000 | 41145s | train_loss=0.0103
412
+ step 4100000 | 41308s | train_loss=0.0103 | fid=44.7875
413
+ step 4110000 | 41395s | train_loss=0.0098
414
+ step 4120000 | 41482s | train_loss=0.0103
415
+ step 4130000 | 41569s | train_loss=0.0097
416
+ step 4140000 | 41659s | train_loss=0.0101
417
+ step 4150000 | 41751s | train_loss=0.0106
418
+ step 4160000 | 41840s | train_loss=0.0097
419
+ step 4170000 | 41931s | train_loss=0.0099
420
+ step 4180000 | 42022s | train_loss=0.0101
421
+ step 4190000 | 42113s | train_loss=0.0103
422
+ step 4200000 | 42276s | train_loss=0.0102 | fid=44.4682
423
+ step 4210000 | 42364s | train_loss=0.0097
424
+ step 4220000 | 42451s | train_loss=0.0103
425
+ step 4230000 | 42542s | train_loss=0.0100
426
+ step 4240000 | 42633s | train_loss=0.0097
427
+ step 4250000 | 42723s | train_loss=0.0098
428
+ step 4260000 | 42813s | train_loss=0.0097
429
+ step 4270000 | 42903s | train_loss=0.0098
430
+ step 4280000 | 42994s | train_loss=0.0107
431
+ step 4290000 | 43084s | train_loss=0.0099
432
+ step 4300000 | 43247s | train_loss=0.0096 | fid=44.4887
433
+ step 4310000 | 43336s | train_loss=0.0095
434
+ step 4320000 | 43423s | train_loss=0.0097
435
+ step 4330000 | 43514s | train_loss=0.0099
436
+ step 4340000 | 43605s | train_loss=0.0094
437
+ step 4350000 | 43696s | train_loss=0.0098
438
+ step 4360000 | 43785s | train_loss=0.0100
439
+ step 4370000 | 43876s | train_loss=0.0095
440
+ step 4380000 | 43967s | train_loss=0.0097
441
+ step 4390000 | 44058s | train_loss=0.0093
442
+ step 4400000 | 44220s | train_loss=0.0094 | fid=44.6314
443
+ step 4410000 | 44308s | train_loss=0.0094
444
+ step 4420000 | 44395s | train_loss=0.0094
445
+ step 4430000 | 44486s | train_loss=0.0095
446
+ step 4440000 | 44576s | train_loss=0.0095
447
+ step 4450000 | 44667s | train_loss=0.0096
448
+ step 4460000 | 44756s | train_loss=0.0099
449
+ step 4470000 | 44847s | train_loss=0.0098
450
+ step 4480000 | 44938s | train_loss=0.0097
451
+ step 4490000 | 45029s | train_loss=0.0096
452
+ step 4500000 | 45191s | train_loss=0.0094 | fid=44.6845
453
+ step 4510000 | 45278s | train_loss=0.0092
454
+ step 4520000 | 45368s | train_loss=0.0097
455
+ step 4530000 | 45458s | train_loss=0.0092
456
+ step 4540000 | 45549s | train_loss=0.0096
457
+ step 4550000 | 45639s | train_loss=0.0092
458
+ step 4560000 | 45728s | train_loss=0.0095
459
+ step 4570000 | 45819s | train_loss=0.0092
460
+ step 4580000 | 45910s | train_loss=0.0091
461
+ step 4590000 | 46002s | train_loss=0.0091
462
+ step 4600000 | 46167s | train_loss=0.0091 | fid=44.5098
463
+ step 4610000 | 46256s | train_loss=0.0092
464
+ step 4620000 | 46350s | train_loss=0.0090
465
+ step 4630000 | 46445s | train_loss=0.0093
466
+ step 4640000 | 46540s | train_loss=0.0094
467
+ step 4650000 | 46635s | train_loss=0.0090
468
+ step 4660000 | 46729s | train_loss=0.0094
469
+ step 4670000 | 46824s | train_loss=0.0093
470
+ step 4680000 | 46920s | train_loss=0.0096
471
+ step 4690000 | 47015s | train_loss=0.0091
472
+ step 4700000 | 47180s | train_loss=0.0096 | fid=44.8634
473
+ step 4710000 | 47272s | train_loss=0.0087
474
+ step 4720000 | 47368s | train_loss=0.0091
475
+ step 4730000 | 47463s | train_loss=0.0090
476
+ step 4740000 | 47559s | train_loss=0.0095
477
+ step 4750000 | 47654s | train_loss=0.0091
478
+ step 4760000 | 47749s | train_loss=0.0092
479
+ step 4770000 | 47844s | train_loss=0.0089
480
+ step 4780000 | 47940s | train_loss=0.0090
481
+ step 4790000 | 48036s | train_loss=0.0088
482
+ step 4800000 | 48184s | train_loss=0.0088 | fid=44.7807
483
+ step 4810000 | 48259s | train_loss=0.0088
484
+ step 4820000 | 48336s | train_loss=0.0090
485
+ step 4830000 | 48411s | train_loss=0.0094
486
+ step 4840000 | 48487s | train_loss=0.0088
487
+ step 4850000 | 48563s | train_loss=0.0097
488
+ step 4860000 | 48638s | train_loss=0.0090
489
+ step 4870000 | 48713s | train_loss=0.0088
490
+ step 4880000 | 48788s | train_loss=0.0089
491
+ step 4890000 | 48862s | train_loss=0.0090
492
+ step 4900000 | 48997s | train_loss=0.0091 | fid=44.4087
493
+ step 4910000 | 49069s | train_loss=0.0084
494
+ step 4920000 | 49146s | train_loss=0.0089
495
+ step 4930000 | 49224s | train_loss=0.0088
496
+ step 4940000 | 49302s | train_loss=0.0089
497
+ step 4950000 | 49379s | train_loss=0.0084
498
+ step 4960000 | 49455s | train_loss=0.0089
499
+ step 4970000 | 49528s | train_loss=0.0091
500
+ step 4980000 | 49604s | train_loss=0.0094
501
+ step 4990000 | 49680s | train_loss=0.0094
502
+ step 5000000 | 49815s | train_loss=0.0084 | fid=44.9422
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d70/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d70/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 72s | train_loss=0.9937 | fid=281.7382
3
+ step 10000 | 164s | train_loss=0.8266
4
+ step 20000 | 259s | train_loss=0.7339
5
+ step 30000 | 353s | train_loss=0.6103
6
+ step 40000 | 448s | train_loss=0.4497
7
+ step 50000 | 543s | train_loss=0.3074
8
+ step 60000 | 637s | train_loss=0.2225
9
+ step 70000 | 732s | train_loss=0.1782
10
+ step 80000 | 827s | train_loss=0.1544
11
+ step 90000 | 921s | train_loss=0.1394
12
+ step 100000 | 1088s | train_loss=0.1269 | fid=57.5316
13
+ step 110000 | 1178s | train_loss=0.1216
14
+ step 120000 | 1272s | train_loss=0.1134
15
+ step 130000 | 1367s | train_loss=0.1085
16
+ step 140000 | 1462s | train_loss=0.1036
17
+ step 150000 | 1557s | train_loss=0.1001
18
+ step 160000 | 1651s | train_loss=0.0950
19
+ step 170000 | 1746s | train_loss=0.0928
20
+ step 180000 | 1841s | train_loss=0.0896
21
+ step 190000 | 1934s | train_loss=0.0857
22
+ step 200000 | 2102s | train_loss=0.0835 | fid=54.9400
23
+ step 210000 | 2191s | train_loss=0.0804
24
+ step 220000 | 2286s | train_loss=0.0775
25
+ step 230000 | 2380s | train_loss=0.0762
26
+ step 240000 | 2475s | train_loss=0.0735
27
+ step 250000 | 2570s | train_loss=0.0730
28
+ step 260000 | 2664s | train_loss=0.0707
29
+ step 270000 | 2759s | train_loss=0.0686
30
+ step 280000 | 2853s | train_loss=0.0665
31
+ step 290000 | 2945s | train_loss=0.0651
32
+ step 300000 | 3116s | train_loss=0.0650 | fid=53.0602
33
+ step 310000 | 3206s | train_loss=0.0636
34
+ step 320000 | 3301s | train_loss=0.0615
35
+ step 330000 | 3395s | train_loss=0.0621
36
+ step 340000 | 3490s | train_loss=0.0607
37
+ step 350000 | 3584s | train_loss=0.0585
38
+ step 360000 | 3679s | train_loss=0.0584
39
+ step 370000 | 3773s | train_loss=0.0574
40
+ step 380000 | 3868s | train_loss=0.0569
41
+ step 390000 | 3960s | train_loss=0.0547
42
+ step 400000 | 4131s | train_loss=0.0547 | fid=50.0583
43
+ step 410000 | 4221s | train_loss=0.0542
44
+ step 420000 | 4315s | train_loss=0.0533
45
+ step 430000 | 4410s | train_loss=0.0541
46
+ step 440000 | 4505s | train_loss=0.0512
47
+ step 450000 | 4599s | train_loss=0.0527
48
+ step 460000 | 4693s | train_loss=0.0509
49
+ step 470000 | 4788s | train_loss=0.0493
50
+ step 480000 | 4880s | train_loss=0.0503
51
+ step 490000 | 4975s | train_loss=0.0484
52
+ step 500000 | 5145s | train_loss=0.0489 | fid=49.9181
53
+ step 510000 | 5236s | train_loss=0.0484
54
+ step 520000 | 5331s | train_loss=0.0477
55
+ step 530000 | 5425s | train_loss=0.0470
56
+ step 540000 | 5519s | train_loss=0.0470
57
+ step 550000 | 5614s | train_loss=0.0468
58
+ step 560000 | 5708s | train_loss=0.0450
59
+ step 570000 | 5800s | train_loss=0.0448
60
+ step 580000 | 5895s | train_loss=0.0450
61
+ step 590000 | 5989s | train_loss=0.0438
62
+ step 600000 | 6158s | train_loss=0.0442 | fid=48.4160
63
+ step 610000 | 6250s | train_loss=0.0436
64
+ step 620000 | 6344s | train_loss=0.0431
65
+ step 630000 | 6439s | train_loss=0.0437
66
+ step 640000 | 6534s | train_loss=0.0429
67
+ step 650000 | 6628s | train_loss=0.0428
68
+ step 660000 | 6723s | train_loss=0.0419
69
+ step 670000 | 6814s | train_loss=0.0425
70
+ step 680000 | 6909s | train_loss=0.0414
71
+ step 690000 | 7004s | train_loss=0.0415
72
+ step 700000 | 7174s | train_loss=0.0408 | fid=47.4201
73
+ step 710000 | 7266s | train_loss=0.0403
74
+ step 720000 | 7361s | train_loss=0.0404
75
+ step 730000 | 7455s | train_loss=0.0396
76
+ step 740000 | 7550s | train_loss=0.0391
77
+ step 750000 | 7645s | train_loss=0.0397
78
+ step 760000 | 7736s | train_loss=0.0391
79
+ step 770000 | 7831s | train_loss=0.0387
80
+ step 780000 | 7926s | train_loss=0.0383
81
+ step 790000 | 8020s | train_loss=0.0387
82
+ step 800000 | 8189s | train_loss=0.0380 | fid=46.2294
83
+ step 810000 | 8281s | train_loss=0.0375
84
+ step 820000 | 8376s | train_loss=0.0367
85
+ step 830000 | 8470s | train_loss=0.0381
86
+ step 840000 | 8565s | train_loss=0.0374
87
+ step 850000 | 8658s | train_loss=0.0359
88
+ step 860000 | 8752s | train_loss=0.0367
89
+ step 870000 | 8847s | train_loss=0.0355
90
+ step 880000 | 8941s | train_loss=0.0355
91
+ step 890000 | 9036s | train_loss=0.0361
92
+ step 900000 | 9205s | train_loss=0.0347 | fid=45.2889
93
+ step 910000 | 9297s | train_loss=0.0359
94
+ step 920000 | 9392s | train_loss=0.0344
95
+ step 930000 | 9486s | train_loss=0.0349
96
+ step 940000 | 9581s | train_loss=0.0343
97
+ step 950000 | 9672s | train_loss=0.0341
98
+ step 960000 | 9767s | train_loss=0.0346
99
+ step 970000 | 9861s | train_loss=0.0332
100
+ step 980000 | 9956s | train_loss=0.0337
101
+ step 990000 | 10050s | train_loss=0.0326
102
+ step 1000000 | 10219s | train_loss=0.0328 | fid=45.6167
103
+ step 1010000 | 10313s | train_loss=0.0336
104
+ step 1020000 | 10405s | train_loss=0.0331
105
+ step 1030000 | 10500s | train_loss=0.0329
106
+ step 1040000 | 10592s | train_loss=0.0318
107
+ step 1050000 | 10687s | train_loss=0.0335
108
+ step 1060000 | 10781s | train_loss=0.0325
109
+ step 1070000 | 10876s | train_loss=0.0317
110
+ step 1080000 | 10970s | train_loss=0.0322
111
+ step 1090000 | 11065s | train_loss=0.0307
112
+ step 1100000 | 11232s | train_loss=0.0329 | fid=44.2804
113
+ step 1110000 | 11325s | train_loss=0.0310
114
+ step 1120000 | 11418s | train_loss=0.0312
115
+ step 1130000 | 11513s | train_loss=0.0308
116
+ step 1140000 | 11604s | train_loss=0.0316
117
+ step 1150000 | 11699s | train_loss=0.0308
118
+ step 1160000 | 11793s | train_loss=0.0306
119
+ step 1170000 | 11888s | train_loss=0.0301
120
+ step 1180000 | 11983s | train_loss=0.0297
121
+ step 1190000 | 12077s | train_loss=0.0298
122
+ step 1200000 | 12245s | train_loss=0.0300 | fid=43.6437
123
+ step 1210000 | 12338s | train_loss=0.0297
124
+ step 1220000 | 12432s | train_loss=0.0290
125
+ step 1230000 | 12523s | train_loss=0.0299
126
+ step 1240000 | 12618s | train_loss=0.0294
127
+ step 1250000 | 12712s | train_loss=0.0290
128
+ step 1260000 | 12807s | train_loss=0.0297
129
+ step 1270000 | 12901s | train_loss=0.0290
130
+ step 1280000 | 12996s | train_loss=0.0293
131
+ step 1290000 | 13090s | train_loss=0.0282
132
+ step 1300000 | 13258s | train_loss=0.0281 | fid=43.4329
133
+ step 1310000 | 13351s | train_loss=0.0284
134
+ step 1320000 | 13442s | train_loss=0.0281
135
+ step 1330000 | 13536s | train_loss=0.0274
136
+ step 1340000 | 13631s | train_loss=0.0278
137
+ step 1350000 | 13725s | train_loss=0.0281
138
+ step 1360000 | 13820s | train_loss=0.0280
139
+ step 1370000 | 13915s | train_loss=0.0280
140
+ step 1380000 | 14009s | train_loss=0.0280
141
+ step 1390000 | 14104s | train_loss=0.0267
142
+ step 1400000 | 14272s | train_loss=0.0268 | fid=42.9891
143
+ step 1410000 | 14363s | train_loss=0.0261
144
+ step 1420000 | 14454s | train_loss=0.0264
145
+ step 1430000 | 14548s | train_loss=0.0268
146
+ step 1440000 | 14643s | train_loss=0.0263
147
+ step 1450000 | 14738s | train_loss=0.0263
148
+ step 1460000 | 14832s | train_loss=0.0257
149
+ step 1470000 | 14927s | train_loss=0.0265
150
+ step 1480000 | 15022s | train_loss=0.0262
151
+ step 1490000 | 15116s | train_loss=0.0256
152
+ step 1500000 | 15284s | train_loss=0.0250 | fid=42.1649
153
+ step 1510000 | 15370s | train_loss=0.0256
154
+ step 1520000 | 15458s | train_loss=0.0263
155
+ step 1530000 | 15551s | train_loss=0.0257
156
+ step 1540000 | 15644s | train_loss=0.0258
157
+ step 1550000 | 15738s | train_loss=0.0249
158
+ step 1560000 | 15832s | train_loss=0.0266
159
+ step 1570000 | 15927s | train_loss=0.0258
160
+ step 1580000 | 16021s | train_loss=0.0249
161
+ step 1590000 | 16114s | train_loss=0.0242
162
+ step 1600000 | 16282s | train_loss=0.0248 | fid=42.4073
163
+ step 1610000 | 16372s | train_loss=0.0250
164
+ step 1620000 | 16465s | train_loss=0.0245
165
+ step 1630000 | 16559s | train_loss=0.0237
166
+ step 1640000 | 16653s | train_loss=0.0239
167
+ step 1650000 | 16748s | train_loss=0.0249
168
+ step 1660000 | 16842s | train_loss=0.0244
169
+ step 1670000 | 16936s | train_loss=0.0242
170
+ step 1680000 | 17031s | train_loss=0.0241
171
+ step 1690000 | 17125s | train_loss=0.0238
172
+ step 1700000 | 17293s | train_loss=0.0235 | fid=41.9175
173
+ step 1710000 | 17383s | train_loss=0.0237
174
+ step 1720000 | 17477s | train_loss=0.0228
175
+ step 1730000 | 17571s | train_loss=0.0231
176
+ step 1740000 | 17665s | train_loss=0.0232
177
+ step 1750000 | 17760s | train_loss=0.0236
178
+ step 1760000 | 17854s | train_loss=0.0225
179
+ step 1770000 | 17948s | train_loss=0.0232
180
+ step 1780000 | 18043s | train_loss=0.0225
181
+ step 1790000 | 18137s | train_loss=0.0224
182
+ step 1800000 | 18300s | train_loss=0.0224 | fid=41.6803
183
+ step 1810000 | 18392s | train_loss=0.0223
184
+ step 1820000 | 18488s | train_loss=0.0219
185
+ step 1830000 | 18581s | train_loss=0.0220
186
+ step 1840000 | 18675s | train_loss=0.0221
187
+ step 1850000 | 18769s | train_loss=0.0216
188
+ step 1860000 | 18863s | train_loss=0.0214
189
+ step 1870000 | 18958s | train_loss=0.0220
190
+ step 1880000 | 19052s | train_loss=0.0219
191
+ step 1890000 | 19146s | train_loss=0.0221
192
+ step 1900000 | 19309s | train_loss=0.0238 | fid=42.4386
193
+ step 1910000 | 19401s | train_loss=0.0208
194
+ step 1920000 | 19495s | train_loss=0.0204
195
+ step 1930000 | 19587s | train_loss=0.0215
196
+ step 1940000 | 19681s | train_loss=0.0208
197
+ step 1950000 | 19773s | train_loss=0.0211
198
+ step 1960000 | 19867s | train_loss=0.0210
199
+ step 1970000 | 19960s | train_loss=0.0216
200
+ step 1980000 | 20055s | train_loss=0.0214
201
+ step 1990000 | 20149s | train_loss=0.0208
202
+ step 2000000 | 20314s | train_loss=0.0206 | fid=42.2453
203
+ step 2010000 | 20408s | train_loss=0.0202
204
+ step 2020000 | 20503s | train_loss=0.0203
205
+ step 2030000 | 20596s | train_loss=0.0201
206
+ step 2040000 | 20690s | train_loss=0.0204
207
+ step 2050000 | 20783s | train_loss=0.0206
208
+ step 2060000 | 20877s | train_loss=0.0210
209
+ step 2070000 | 20970s | train_loss=0.0204
210
+ step 2080000 | 21064s | train_loss=0.0198
211
+ step 2090000 | 21154s | train_loss=0.0200
212
+ step 2100000 | 21320s | train_loss=0.0196 | fid=41.8879
213
+ step 2110000 | 21413s | train_loss=0.0198
214
+ step 2120000 | 21506s | train_loss=0.0198
215
+ step 2130000 | 21598s | train_loss=0.0206
216
+ step 2140000 | 21692s | train_loss=0.0195
217
+ step 2150000 | 21785s | train_loss=0.0200
218
+ step 2160000 | 21879s | train_loss=0.0195
219
+ step 2170000 | 21974s | train_loss=0.0195
220
+ step 2180000 | 22068s | train_loss=0.0195
221
+ step 2190000 | 22159s | train_loss=0.0196
222
+ step 2200000 | 22327s | train_loss=0.0186 | fid=41.9945
223
+ step 2210000 | 22421s | train_loss=0.0189
224
+ step 2220000 | 22515s | train_loss=0.0187
225
+ step 2230000 | 22608s | train_loss=0.0190
226
+ step 2240000 | 22702s | train_loss=0.0183
227
+ step 2250000 | 22796s | train_loss=0.0193
228
+ step 2260000 | 22891s | train_loss=0.0188
229
+ step 2270000 | 22985s | train_loss=0.0184
230
+ step 2280000 | 23068s | train_loss=0.0184
231
+ step 2290000 | 23159s | train_loss=0.0186
232
+ step 2300000 | 23324s | train_loss=0.0189 | fid=42.3059
233
+ step 2310000 | 23418s | train_loss=0.0185
234
+ step 2320000 | 23510s | train_loss=0.0188
235
+ step 2330000 | 23602s | train_loss=0.0182
236
+ step 2340000 | 23695s | train_loss=0.0177
237
+ step 2350000 | 23789s | train_loss=0.0182
238
+ step 2360000 | 23882s | train_loss=0.0184
239
+ step 2370000 | 23976s | train_loss=0.0186
240
+ step 2380000 | 24066s | train_loss=0.0176
241
+ step 2390000 | 24159s | train_loss=0.0179
242
+ step 2400000 | 24325s | train_loss=0.0178 | fid=42.2154
243
+ step 2410000 | 24418s | train_loss=0.0179
244
+ step 2420000 | 24511s | train_loss=0.0174
245
+ step 2430000 | 24603s | train_loss=0.0177
246
+ step 2440000 | 24697s | train_loss=0.0177
247
+ step 2450000 | 24790s | train_loss=0.0169
248
+ step 2460000 | 24884s | train_loss=0.0185
249
+ step 2470000 | 24975s | train_loss=0.0176
250
+ step 2480000 | 25067s | train_loss=0.0170
251
+ step 2490000 | 25161s | train_loss=0.0175
252
+ step 2500000 | 25326s | train_loss=0.0172 | fid=41.5502
253
+ step 2510000 | 25420s | train_loss=0.0168
254
+ step 2520000 | 25513s | train_loss=0.0178
255
+ step 2530000 | 25605s | train_loss=0.0166
256
+ step 2540000 | 25697s | train_loss=0.0173
257
+ step 2550000 | 25790s | train_loss=0.0174
258
+ step 2560000 | 25884s | train_loss=0.0164
259
+ step 2570000 | 25974s | train_loss=0.0166
260
+ step 2580000 | 26068s | train_loss=0.0169
261
+ step 2590000 | 26161s | train_loss=0.0164
262
+ step 2600000 | 26326s | train_loss=0.0166 | fid=42.2077
263
+ step 2610000 | 26419s | train_loss=0.0167
264
+ step 2620000 | 26513s | train_loss=0.0169
265
+ step 2630000 | 26607s | train_loss=0.0157
266
+ step 2640000 | 26698s | train_loss=0.0168
267
+ step 2650000 | 26791s | train_loss=0.0175
268
+ step 2660000 | 26882s | train_loss=0.0158
269
+ step 2670000 | 26975s | train_loss=0.0161
270
+ step 2680000 | 27069s | train_loss=0.0166
271
+ step 2690000 | 27162s | train_loss=0.0161
272
+ step 2700000 | 27326s | train_loss=0.0161 | fid=41.8389
273
+ step 2710000 | 27420s | train_loss=0.0160
274
+ step 2720000 | 27513s | train_loss=0.0164
275
+ step 2730000 | 27607s | train_loss=0.0156
276
+ step 2740000 | 27699s | train_loss=0.0155
277
+ step 2750000 | 27792s | train_loss=0.0159
278
+ step 2760000 | 27882s | train_loss=0.0157
279
+ step 2770000 | 27976s | train_loss=0.0155
280
+ step 2780000 | 28071s | train_loss=0.0155
281
+ step 2790000 | 28165s | train_loss=0.0157
282
+ step 2800000 | 28331s | train_loss=0.0155 | fid=41.1992
283
+ step 2810000 | 28423s | train_loss=0.0156
284
+ step 2820000 | 28511s | train_loss=0.0158
285
+ step 2830000 | 28600s | train_loss=0.0149
286
+ step 2840000 | 28686s | train_loss=0.0153
287
+ step 2850000 | 28772s | train_loss=0.0160
288
+ step 2860000 | 28866s | train_loss=0.0162
289
+ step 2870000 | 28961s | train_loss=0.0151
290
+ step 2880000 | 29055s | train_loss=0.0147
291
+ step 2890000 | 29149s | train_loss=0.0143
292
+ step 2900000 | 29315s | train_loss=0.0145 | fid=41.5969
293
+ step 2910000 | 29407s | train_loss=0.0146
294
+ step 2920000 | 29496s | train_loss=0.0146
295
+ step 2930000 | 29585s | train_loss=0.0143
296
+ step 2940000 | 29673s | train_loss=0.0148
297
+ step 2950000 | 29763s | train_loss=0.0143
298
+ step 2960000 | 29858s | train_loss=0.0152
299
+ step 2970000 | 29952s | train_loss=0.0148
300
+ step 2980000 | 30046s | train_loss=0.0140
301
+ step 2990000 | 30141s | train_loss=0.0146
302
+ step 3000000 | 30307s | train_loss=0.0147 | fid=41.5932
303
+ step 3010000 | 30402s | train_loss=0.0145
304
+ step 3020000 | 30497s | train_loss=0.0139
305
+ step 3030000 | 30591s | train_loss=0.0143
306
+ step 3040000 | 30681s | train_loss=0.0162
307
+ step 3050000 | 30776s | train_loss=0.0149
308
+ step 3060000 | 30870s | train_loss=0.0145
309
+ step 3070000 | 30964s | train_loss=0.0142
310
+ step 3080000 | 31059s | train_loss=0.0143
311
+ step 3090000 | 31154s | train_loss=0.0142
312
+ step 3100000 | 31320s | train_loss=0.0146 | fid=41.9533
313
+ step 3110000 | 31415s | train_loss=0.0141
314
+ step 3120000 | 31508s | train_loss=0.0136
315
+ step 3130000 | 31602s | train_loss=0.0137
316
+ step 3140000 | 31681s | train_loss=0.0135
317
+ step 3150000 | 31773s | train_loss=0.0143
318
+ step 3160000 | 31866s | train_loss=0.0132
319
+ step 3170000 | 31959s | train_loss=0.0144
320
+ step 3180000 | 32054s | train_loss=0.0131
321
+ step 3190000 | 32147s | train_loss=0.0134
322
+ step 3200000 | 32311s | train_loss=0.0140 | fid=41.5354
323
+ step 3210000 | 32405s | train_loss=0.0137
324
+ step 3220000 | 32500s | train_loss=0.0132
325
+ step 3230000 | 32594s | train_loss=0.0143
326
+ step 3240000 | 32684s | train_loss=0.0133
327
+ step 3250000 | 32778s | train_loss=0.0142
328
+ step 3260000 | 32872s | train_loss=0.0132
329
+ step 3270000 | 32966s | train_loss=0.0137
330
+ step 3280000 | 33059s | train_loss=0.0135
331
+ step 3290000 | 33152s | train_loss=0.0130
332
+ step 3300000 | 33316s | train_loss=0.0133 | fid=41.7007
333
+ step 3310000 | 33410s | train_loss=0.0128
334
+ step 3320000 | 33504s | train_loss=0.0137
335
+ step 3330000 | 33596s | train_loss=0.0135
336
+ step 3340000 | 33690s | train_loss=0.0132
337
+ step 3350000 | 33783s | train_loss=0.0134
338
+ step 3360000 | 33877s | train_loss=0.0136
339
+ step 3370000 | 33970s | train_loss=0.0126
340
+ step 3380000 | 34065s | train_loss=0.0130
341
+ step 3390000 | 34160s | train_loss=0.0133
342
+ step 3400000 | 34318s | train_loss=0.0129 | fid=41.3274
343
+ step 3410000 | 34413s | train_loss=0.0124
344
+ step 3420000 | 34507s | train_loss=0.0128
345
+ step 3430000 | 34591s | train_loss=0.0126
346
+ step 3440000 | 34686s | train_loss=0.0125
347
+ step 3450000 | 34778s | train_loss=0.0127
348
+ step 3460000 | 34873s | train_loss=0.0127
349
+ step 3470000 | 34968s | train_loss=0.0125
350
+ step 3480000 | 35062s | train_loss=0.0129
351
+ step 3490000 | 35156s | train_loss=0.0126
352
+ step 3500000 | 35310s | train_loss=0.0122 | fid=41.2492
353
+ step 3510000 | 35405s | train_loss=0.0124
354
+ step 3520000 | 35499s | train_loss=0.0122
355
+ step 3530000 | 35581s | train_loss=0.0130
356
+ step 3540000 | 35676s | train_loss=0.0121
357
+ step 3550000 | 35769s | train_loss=0.0126
358
+ step 3560000 | 35863s | train_loss=0.0129
359
+ step 3570000 | 35958s | train_loss=0.0126
360
+ step 3580000 | 36052s | train_loss=0.0127
361
+ step 3590000 | 36147s | train_loss=0.0125
362
+ step 3600000 | 36311s | train_loss=0.0118 | fid=40.9639
363
+ step 3610000 | 36405s | train_loss=0.0123
364
+ step 3620000 | 36499s | train_loss=0.0129
365
+ step 3630000 | 36581s | train_loss=0.0118
366
+ step 3640000 | 36675s | train_loss=0.0119
367
+ step 3650000 | 36768s | train_loss=0.0122
368
+ step 3660000 | 36863s | train_loss=0.0119
369
+ step 3670000 | 36958s | train_loss=0.0121
370
+ step 3680000 | 37052s | train_loss=0.0123
371
+ step 3690000 | 37147s | train_loss=0.0119
372
+ step 3700000 | 37314s | train_loss=0.0120 | fid=41.4676
373
+ step 3710000 | 37408s | train_loss=0.0121
374
+ step 3720000 | 37502s | train_loss=0.0120
375
+ step 3730000 | 37585s | train_loss=0.0119
376
+ step 3740000 | 37680s | train_loss=0.0120
377
+ step 3750000 | 37773s | train_loss=0.0130
378
+ step 3760000 | 37868s | train_loss=0.0123
379
+ step 3770000 | 37962s | train_loss=0.0116
380
+ step 3780000 | 38057s | train_loss=0.0114
381
+ step 3790000 | 38151s | train_loss=0.0119
382
+ step 3800000 | 38318s | train_loss=0.0115 | fid=41.3386
383
+ step 3810000 | 38413s | train_loss=0.0118
384
+ step 3820000 | 38504s | train_loss=0.0115
385
+ step 3830000 | 38590s | train_loss=0.0120
386
+ step 3840000 | 38684s | train_loss=0.0124
387
+ step 3850000 | 38778s | train_loss=0.0114
388
+ step 3860000 | 38872s | train_loss=0.0120
389
+ step 3870000 | 38967s | train_loss=0.0111
390
+ step 3880000 | 39061s | train_loss=0.0116
391
+ step 3890000 | 39156s | train_loss=0.0116
392
+ step 3900000 | 39322s | train_loss=0.0119 | fid=41.9942
393
+ step 3910000 | 39417s | train_loss=0.0117
394
+ step 3920000 | 39506s | train_loss=0.0115
395
+ step 3930000 | 39594s | train_loss=0.0117
396
+ step 3940000 | 39689s | train_loss=0.0108
397
+ step 3950000 | 39782s | train_loss=0.0115
398
+ step 3960000 | 39877s | train_loss=0.0110
399
+ step 3970000 | 39971s | train_loss=0.0119
400
+ step 3980000 | 40066s | train_loss=0.0114
401
+ step 3990000 | 40160s | train_loss=0.0109
402
+ step 4000000 | 40326s | train_loss=0.0111 | fid=41.0059
403
+ step 4010000 | 40421s | train_loss=0.0109
404
+ step 4020000 | 40507s | train_loss=0.0117
405
+ step 4030000 | 40597s | train_loss=0.0107
406
+ step 4040000 | 40691s | train_loss=0.0109
407
+ step 4050000 | 40785s | train_loss=0.0116
408
+ step 4060000 | 40880s | train_loss=0.0112
409
+ step 4070000 | 40974s | train_loss=0.0105
410
+ step 4080000 | 41069s | train_loss=0.0112
411
+ step 4090000 | 41163s | train_loss=0.0110
412
+ step 4100000 | 41323s | train_loss=0.0110 | fid=42.2207
413
+ step 4110000 | 41414s | train_loss=0.0108
414
+ step 4120000 | 41497s | train_loss=0.0112
415
+ step 4130000 | 41588s | train_loss=0.0103
416
+ step 4140000 | 41679s | train_loss=0.0111
417
+ step 4150000 | 41769s | train_loss=0.0109
418
+ step 4160000 | 41860s | train_loss=0.0106
419
+ step 4170000 | 41951s | train_loss=0.0106
420
+ step 4180000 | 42042s | train_loss=0.0110
421
+ step 4190000 | 42134s | train_loss=0.0111
422
+ step 4200000 | 42293s | train_loss=0.0109 | fid=41.0943
423
+ step 4210000 | 42384s | train_loss=0.0106
424
+ step 4220000 | 42471s | train_loss=0.0109
425
+ step 4230000 | 42562s | train_loss=0.0110
426
+ step 4240000 | 42653s | train_loss=0.0103
427
+ step 4250000 | 42744s | train_loss=0.0105
428
+ step 4260000 | 42834s | train_loss=0.0105
429
+ step 4270000 | 42924s | train_loss=0.0103
430
+ step 4280000 | 43016s | train_loss=0.0110
431
+ step 4290000 | 43106s | train_loss=0.0106
432
+ step 4300000 | 43264s | train_loss=0.0101 | fid=42.1437
433
+ step 4310000 | 43355s | train_loss=0.0101
434
+ step 4320000 | 43443s | train_loss=0.0104
435
+ step 4330000 | 43534s | train_loss=0.0106
436
+ step 4340000 | 43625s | train_loss=0.0105
437
+ step 4350000 | 43717s | train_loss=0.0107
438
+ step 4360000 | 43806s | train_loss=0.0106
439
+ step 4370000 | 43897s | train_loss=0.0102
440
+ step 4380000 | 43988s | train_loss=0.0107
441
+ step 4390000 | 44079s | train_loss=0.0100
442
+ step 4400000 | 44237s | train_loss=0.0105 | fid=41.3410
443
+ step 4410000 | 44324s | train_loss=0.0106
444
+ step 4420000 | 44415s | train_loss=0.0104
445
+ step 4430000 | 44506s | train_loss=0.0102
446
+ step 4440000 | 44597s | train_loss=0.0100
447
+ step 4450000 | 44688s | train_loss=0.0104
448
+ step 4460000 | 44778s | train_loss=0.0103
449
+ step 4470000 | 44869s | train_loss=0.0108
450
+ step 4480000 | 44960s | train_loss=0.0100
451
+ step 4490000 | 45052s | train_loss=0.0103
452
+ step 4500000 | 45210s | train_loss=0.0102 | fid=41.2370
453
+ step 4510000 | 45299s | train_loss=0.0099
454
+ step 4520000 | 45389s | train_loss=0.0104
455
+ step 4530000 | 45480s | train_loss=0.0100
456
+ step 4540000 | 45570s | train_loss=0.0105
457
+ step 4550000 | 45661s | train_loss=0.0104
458
+ step 4560000 | 45751s | train_loss=0.0100
459
+ step 4570000 | 45842s | train_loss=0.0102
460
+ step 4580000 | 45934s | train_loss=0.0100
461
+ step 4590000 | 46025s | train_loss=0.0100
462
+ step 4600000 | 46186s | train_loss=0.0098 | fid=41.3752
463
+ step 4610000 | 46278s | train_loss=0.0099
464
+ step 4620000 | 46372s | train_loss=0.0096
465
+ step 4630000 | 46467s | train_loss=0.0100
466
+ step 4640000 | 46561s | train_loss=0.0100
467
+ step 4650000 | 46656s | train_loss=0.0097
468
+ step 4660000 | 46749s | train_loss=0.0097
469
+ step 4670000 | 46844s | train_loss=0.0100
470
+ step 4680000 | 46939s | train_loss=0.0100
471
+ step 4690000 | 47034s | train_loss=0.0097
472
+ step 4700000 | 47197s | train_loss=0.0103 | fid=41.1172
473
+ step 4710000 | 47292s | train_loss=0.0097
474
+ step 4720000 | 47386s | train_loss=0.0101
475
+ step 4730000 | 47481s | train_loss=0.0097
476
+ step 4740000 | 47576s | train_loss=0.0101
477
+ step 4750000 | 47671s | train_loss=0.0096
478
+ step 4760000 | 47765s | train_loss=0.0098
479
+ step 4770000 | 47860s | train_loss=0.0097
480
+ step 4780000 | 47955s | train_loss=0.0098
481
+ step 4790000 | 48050s | train_loss=0.0102
482
+ step 4800000 | 48194s | train_loss=0.0095 | fid=41.7027
483
+ step 4810000 | 48271s | train_loss=0.0094
484
+ step 4820000 | 48347s | train_loss=0.0097
485
+ step 4830000 | 48422s | train_loss=0.0101
486
+ step 4840000 | 48498s | train_loss=0.0096
487
+ step 4850000 | 48574s | train_loss=0.0093
488
+ step 4860000 | 48648s | train_loss=0.0097
489
+ step 4870000 | 48723s | train_loss=0.0095
490
+ step 4880000 | 48799s | train_loss=0.0096
491
+ step 4890000 | 48873s | train_loss=0.0093
492
+ step 4900000 | 49006s | train_loss=0.0096 | fid=41.5412
493
+ step 4910000 | 49081s | train_loss=0.0092
494
+ step 4920000 | 49158s | train_loss=0.0096
495
+ step 4930000 | 49235s | train_loss=0.0095
496
+ step 4940000 | 49312s | train_loss=0.0093
497
+ step 4950000 | 49389s | train_loss=0.0096
498
+ step 4960000 | 49464s | train_loss=0.0098
499
+ step 4970000 | 49539s | train_loss=0.0100
500
+ step 4980000 | 49615s | train_loss=0.0099
501
+ step 4990000 | 49692s | train_loss=0.0101
502
+ step 5000000 | 49821s | train_loss=0.0090 | fid=41.4839
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d80/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d80/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 74s | train_loss=0.9962 | fid=276.0305
3
+ step 10000 | 168s | train_loss=0.8176
4
+ step 20000 | 263s | train_loss=0.7137
5
+ step 30000 | 357s | train_loss=0.5845
6
+ step 40000 | 452s | train_loss=0.4294
7
+ step 50000 | 546s | train_loss=0.2965
8
+ step 60000 | 641s | train_loss=0.2179
9
+ step 70000 | 736s | train_loss=0.1743
10
+ step 80000 | 830s | train_loss=0.1513
11
+ step 90000 | 924s | train_loss=0.1374
12
+ step 100000 | 1091s | train_loss=0.1277 | fid=56.0463
13
+ step 110000 | 1183s | train_loss=0.1205
14
+ step 120000 | 1278s | train_loss=0.1133
15
+ step 130000 | 1373s | train_loss=0.1070
16
+ step 140000 | 1468s | train_loss=0.1035
17
+ step 150000 | 1562s | train_loss=0.1010
18
+ step 160000 | 1657s | train_loss=0.0953
19
+ step 170000 | 1752s | train_loss=0.0916
20
+ step 180000 | 1846s | train_loss=0.0901
21
+ step 190000 | 1939s | train_loss=0.0858
22
+ step 200000 | 2106s | train_loss=0.0840 | fid=55.3489
23
+ step 210000 | 2199s | train_loss=0.0807
24
+ step 220000 | 2293s | train_loss=0.0789
25
+ step 230000 | 2388s | train_loss=0.0768
26
+ step 240000 | 2482s | train_loss=0.0747
27
+ step 250000 | 2576s | train_loss=0.0741
28
+ step 260000 | 2671s | train_loss=0.0714
29
+ step 270000 | 2765s | train_loss=0.0697
30
+ step 280000 | 2860s | train_loss=0.0675
31
+ step 290000 | 2952s | train_loss=0.0674
32
+ step 300000 | 3120s | train_loss=0.0650 | fid=53.5774
33
+ step 310000 | 3213s | train_loss=0.0661
34
+ step 320000 | 3308s | train_loss=0.0636
35
+ step 330000 | 3402s | train_loss=0.0623
36
+ step 340000 | 3497s | train_loss=0.0612
37
+ step 350000 | 3591s | train_loss=0.0607
38
+ step 360000 | 3685s | train_loss=0.0603
39
+ step 370000 | 3779s | train_loss=0.0591
40
+ step 380000 | 3873s | train_loss=0.0577
41
+ step 390000 | 3965s | train_loss=0.0562
42
+ step 400000 | 4134s | train_loss=0.0553 | fid=52.5275
43
+ step 410000 | 4226s | train_loss=0.0559
44
+ step 420000 | 4321s | train_loss=0.0542
45
+ step 430000 | 4415s | train_loss=0.0547
46
+ step 440000 | 4510s | train_loss=0.0544
47
+ step 450000 | 4604s | train_loss=0.0522
48
+ step 460000 | 4698s | train_loss=0.0523
49
+ step 470000 | 4793s | train_loss=0.0517
50
+ step 480000 | 4885s | train_loss=0.0522
51
+ step 490000 | 4980s | train_loss=0.0498
52
+ step 500000 | 5148s | train_loss=0.0503 | fid=49.8012
53
+ step 510000 | 5240s | train_loss=0.0499
54
+ step 520000 | 5335s | train_loss=0.0488
55
+ step 530000 | 5429s | train_loss=0.0493
56
+ step 540000 | 5523s | train_loss=0.0477
57
+ step 550000 | 5617s | train_loss=0.0472
58
+ step 560000 | 5712s | train_loss=0.0471
59
+ step 570000 | 5803s | train_loss=0.0464
60
+ step 580000 | 5897s | train_loss=0.0458
61
+ step 590000 | 5992s | train_loss=0.0459
62
+ step 600000 | 6161s | train_loss=0.0456 | fid=48.0231
63
+ step 610000 | 6253s | train_loss=0.0462
64
+ step 620000 | 6347s | train_loss=0.0448
65
+ step 630000 | 6441s | train_loss=0.0440
66
+ step 640000 | 6536s | train_loss=0.0440
67
+ step 650000 | 6630s | train_loss=0.0441
68
+ step 660000 | 6725s | train_loss=0.0434
69
+ step 670000 | 6816s | train_loss=0.0421
70
+ step 680000 | 6911s | train_loss=0.0426
71
+ step 690000 | 7006s | train_loss=0.0427
72
+ step 700000 | 7174s | train_loss=0.0430 | fid=48.5343
73
+ step 710000 | 7266s | train_loss=0.0416
74
+ step 720000 | 7360s | train_loss=0.0418
75
+ step 730000 | 7454s | train_loss=0.0426
76
+ step 740000 | 7549s | train_loss=0.0415
77
+ step 750000 | 7643s | train_loss=0.0406
78
+ step 760000 | 7735s | train_loss=0.0402
79
+ step 770000 | 7829s | train_loss=0.0397
80
+ step 780000 | 7924s | train_loss=0.0402
81
+ step 790000 | 8018s | train_loss=0.0405
82
+ step 800000 | 8186s | train_loss=0.0387 | fid=46.3511
83
+ step 810000 | 8276s | train_loss=0.0385
84
+ step 820000 | 8371s | train_loss=0.0380
85
+ step 830000 | 8466s | train_loss=0.0398
86
+ step 840000 | 8560s | train_loss=0.0385
87
+ step 850000 | 8654s | train_loss=0.0372
88
+ step 860000 | 8746s | train_loss=0.0390
89
+ step 870000 | 8841s | train_loss=0.0363
90
+ step 880000 | 8935s | train_loss=0.0362
91
+ step 890000 | 9030s | train_loss=0.0373
92
+ step 900000 | 9198s | train_loss=0.0362 | fid=44.7119
93
+ step 910000 | 9289s | train_loss=0.0361
94
+ step 920000 | 9382s | train_loss=0.0360
95
+ step 930000 | 9476s | train_loss=0.0357
96
+ step 940000 | 9571s | train_loss=0.0355
97
+ step 950000 | 9662s | train_loss=0.0361
98
+ step 960000 | 9757s | train_loss=0.0360
99
+ step 970000 | 9851s | train_loss=0.0347
100
+ step 980000 | 9945s | train_loss=0.0356
101
+ step 990000 | 10040s | train_loss=0.0341
102
+ step 1000000 | 10209s | train_loss=0.0348 | fid=44.7196
103
+ step 1010000 | 10300s | train_loss=0.0341
104
+ step 1020000 | 10393s | train_loss=0.0348
105
+ step 1030000 | 10487s | train_loss=0.0338
106
+ step 1040000 | 10579s | train_loss=0.0336
107
+ step 1050000 | 10673s | train_loss=0.0340
108
+ step 1060000 | 10768s | train_loss=0.0346
109
+ step 1070000 | 10862s | train_loss=0.0335
110
+ step 1080000 | 10956s | train_loss=0.0336
111
+ step 1090000 | 11051s | train_loss=0.0327
112
+ step 1100000 | 11219s | train_loss=0.0329 | fid=43.4767
113
+ step 1110000 | 11311s | train_loss=0.0323
114
+ step 1120000 | 11404s | train_loss=0.0326
115
+ step 1130000 | 11499s | train_loss=0.0329
116
+ step 1140000 | 11590s | train_loss=0.0318
117
+ step 1150000 | 11684s | train_loss=0.0314
118
+ step 1160000 | 11778s | train_loss=0.0323
119
+ step 1170000 | 11872s | train_loss=0.0318
120
+ step 1180000 | 11967s | train_loss=0.0307
121
+ step 1190000 | 12061s | train_loss=0.0313
122
+ step 1200000 | 12231s | train_loss=0.0304 | fid=43.5890
123
+ step 1210000 | 12322s | train_loss=0.0303
124
+ step 1220000 | 12415s | train_loss=0.0304
125
+ step 1230000 | 12506s | train_loss=0.0306
126
+ step 1240000 | 12601s | train_loss=0.0310
127
+ step 1250000 | 12695s | train_loss=0.0311
128
+ step 1260000 | 12789s | train_loss=0.0313
129
+ step 1270000 | 12884s | train_loss=0.0301
130
+ step 1280000 | 12978s | train_loss=0.0297
131
+ step 1290000 | 13072s | train_loss=0.0297
132
+ step 1300000 | 13240s | train_loss=0.0293 | fid=42.1323
133
+ step 1310000 | 13332s | train_loss=0.0291
134
+ step 1320000 | 13426s | train_loss=0.0305
135
+ step 1330000 | 13517s | train_loss=0.0298
136
+ step 1340000 | 13611s | train_loss=0.0290
137
+ step 1350000 | 13705s | train_loss=0.0285
138
+ step 1360000 | 13800s | train_loss=0.0285
139
+ step 1370000 | 13894s | train_loss=0.0288
140
+ step 1380000 | 13989s | train_loss=0.0286
141
+ step 1390000 | 14084s | train_loss=0.0285
142
+ step 1400000 | 14251s | train_loss=0.0277 | fid=42.4876
143
+ step 1410000 | 14341s | train_loss=0.0278
144
+ step 1420000 | 14431s | train_loss=0.0281
145
+ step 1430000 | 14526s | train_loss=0.0275
146
+ step 1440000 | 14620s | train_loss=0.0272
147
+ step 1450000 | 14714s | train_loss=0.0275
148
+ step 1460000 | 14809s | train_loss=0.0271
149
+ step 1470000 | 14903s | train_loss=0.0269
150
+ step 1480000 | 14998s | train_loss=0.0265
151
+ step 1490000 | 15092s | train_loss=0.0264
152
+ step 1500000 | 15261s | train_loss=0.0267 | fid=41.7123
153
+ step 1510000 | 15350s | train_loss=0.0269
154
+ step 1520000 | 15433s | train_loss=0.0260
155
+ step 1530000 | 15526s | train_loss=0.0269
156
+ step 1540000 | 15619s | train_loss=0.0266
157
+ step 1550000 | 15712s | train_loss=0.0259
158
+ step 1560000 | 15806s | train_loss=0.0263
159
+ step 1570000 | 15900s | train_loss=0.0271
160
+ step 1580000 | 15994s | train_loss=0.0260
161
+ step 1590000 | 16088s | train_loss=0.0250
162
+ step 1600000 | 16256s | train_loss=0.0261 | fid=42.0618
163
+ step 1610000 | 16343s | train_loss=0.0255
164
+ step 1620000 | 16438s | train_loss=0.0250
165
+ step 1630000 | 16530s | train_loss=0.0253
166
+ step 1640000 | 16624s | train_loss=0.0249
167
+ step 1650000 | 16718s | train_loss=0.0255
168
+ step 1660000 | 16813s | train_loss=0.0258
169
+ step 1670000 | 16907s | train_loss=0.0252
170
+ step 1680000 | 17001s | train_loss=0.0255
171
+ step 1690000 | 17095s | train_loss=0.0248
172
+ step 1700000 | 17263s | train_loss=0.0248 | fid=40.7455
173
+ step 1710000 | 17348s | train_loss=0.0245
174
+ step 1720000 | 17443s | train_loss=0.0242
175
+ step 1730000 | 17535s | train_loss=0.0245
176
+ step 1740000 | 17629s | train_loss=0.0241
177
+ step 1750000 | 17724s | train_loss=0.0251
178
+ step 1760000 | 17818s | train_loss=0.0234
179
+ step 1770000 | 17911s | train_loss=0.0232
180
+ step 1780000 | 18006s | train_loss=0.0241
181
+ step 1790000 | 18100s | train_loss=0.0236
182
+ step 1800000 | 18266s | train_loss=0.0235 | fid=40.2259
183
+ step 1810000 | 18357s | train_loss=0.0242
184
+ step 1820000 | 18451s | train_loss=0.0236
185
+ step 1830000 | 18544s | train_loss=0.0232
186
+ step 1840000 | 18639s | train_loss=0.0230
187
+ step 1850000 | 18733s | train_loss=0.0228
188
+ step 1860000 | 18827s | train_loss=0.0222
189
+ step 1870000 | 18921s | train_loss=0.0229
190
+ step 1880000 | 19015s | train_loss=0.0224
191
+ step 1890000 | 19110s | train_loss=0.0239
192
+ step 1900000 | 19273s | train_loss=0.0253 | fid=40.2822
193
+ step 1910000 | 19363s | train_loss=0.0221
194
+ step 1920000 | 19456s | train_loss=0.0216
195
+ step 1930000 | 19548s | train_loss=0.0217
196
+ step 1940000 | 19642s | train_loss=0.0220
197
+ step 1950000 | 19734s | train_loss=0.0220
198
+ step 1960000 | 19828s | train_loss=0.0220
199
+ step 1970000 | 19920s | train_loss=0.0232
200
+ step 1980000 | 20014s | train_loss=0.0226
201
+ step 1990000 | 20108s | train_loss=0.0222
202
+ step 2000000 | 20274s | train_loss=0.0218 | fid=40.6751
203
+ step 2010000 | 20365s | train_loss=0.0212
204
+ step 2020000 | 20459s | train_loss=0.0213
205
+ step 2030000 | 20552s | train_loss=0.0215
206
+ step 2040000 | 20646s | train_loss=0.0216
207
+ step 2050000 | 20739s | train_loss=0.0216
208
+ step 2060000 | 20833s | train_loss=0.0215
209
+ step 2070000 | 20926s | train_loss=0.0210
210
+ step 2080000 | 21019s | train_loss=0.0209
211
+ step 2090000 | 21112s | train_loss=0.0212
212
+ step 2100000 | 21275s | train_loss=0.0210 | fid=40.5968
213
+ step 2110000 | 21364s | train_loss=0.0206
214
+ step 2120000 | 21458s | train_loss=0.0205
215
+ step 2130000 | 21550s | train_loss=0.0219
216
+ step 2140000 | 21642s | train_loss=0.0207
217
+ step 2150000 | 21736s | train_loss=0.0217
218
+ step 2160000 | 21829s | train_loss=0.0205
219
+ step 2170000 | 21923s | train_loss=0.0203
220
+ step 2180000 | 22018s | train_loss=0.0206
221
+ step 2190000 | 22108s | train_loss=0.0202
222
+ step 2200000 | 22278s | train_loss=0.0198 | fid=40.1776
223
+ step 2210000 | 22367s | train_loss=0.0198
224
+ step 2220000 | 22462s | train_loss=0.0198
225
+ step 2230000 | 22556s | train_loss=0.0206
226
+ step 2240000 | 22648s | train_loss=0.0195
227
+ step 2250000 | 22743s | train_loss=0.0204
228
+ step 2260000 | 22837s | train_loss=0.0202
229
+ step 2270000 | 22931s | train_loss=0.0199
230
+ step 2280000 | 23024s | train_loss=0.0193
231
+ step 2290000 | 23104s | train_loss=0.0199
232
+ step 2300000 | 23270s | train_loss=0.0199 | fid=40.0106
233
+ step 2310000 | 23359s | train_loss=0.0191
234
+ step 2320000 | 23451s | train_loss=0.0197
235
+ step 2330000 | 23544s | train_loss=0.0195
236
+ step 2340000 | 23636s | train_loss=0.0191
237
+ step 2350000 | 23728s | train_loss=0.0189
238
+ step 2360000 | 23821s | train_loss=0.0193
239
+ step 2370000 | 23914s | train_loss=0.0201
240
+ step 2380000 | 24007s | train_loss=0.0187
241
+ step 2390000 | 24098s | train_loss=0.0189
242
+ step 2400000 | 24265s | train_loss=0.0189 | fid=40.3519
243
+ step 2410000 | 24354s | train_loss=0.0187
244
+ step 2420000 | 24447s | train_loss=0.0186
245
+ step 2430000 | 24540s | train_loss=0.0184
246
+ step 2440000 | 24631s | train_loss=0.0189
247
+ step 2450000 | 24724s | train_loss=0.0181
248
+ step 2460000 | 24817s | train_loss=0.0188
249
+ step 2470000 | 24910s | train_loss=0.0192
250
+ step 2480000 | 25000s | train_loss=0.0181
251
+ step 2490000 | 25094s | train_loss=0.0181
252
+ step 2500000 | 25260s | train_loss=0.0181 | fid=39.6046
253
+ step 2510000 | 25349s | train_loss=0.0181
254
+ step 2520000 | 25442s | train_loss=0.0177
255
+ step 2530000 | 25535s | train_loss=0.0178
256
+ step 2540000 | 25626s | train_loss=0.0187
257
+ step 2550000 | 25718s | train_loss=0.0180
258
+ step 2560000 | 25811s | train_loss=0.0174
259
+ step 2570000 | 25904s | train_loss=0.0181
260
+ step 2580000 | 25995s | train_loss=0.0179
261
+ step 2590000 | 26087s | train_loss=0.0179
262
+ step 2600000 | 26254s | train_loss=0.0174 | fid=40.7129
263
+ step 2610000 | 26344s | train_loss=0.0175
264
+ step 2620000 | 26437s | train_loss=0.0186
265
+ step 2630000 | 26530s | train_loss=0.0171
266
+ step 2640000 | 26622s | train_loss=0.0185
267
+ step 2650000 | 26714s | train_loss=0.0184
268
+ step 2660000 | 26807s | train_loss=0.0178
269
+ step 2670000 | 26898s | train_loss=0.0169
270
+ step 2680000 | 26991s | train_loss=0.0175
271
+ step 2690000 | 27084s | train_loss=0.0178
272
+ step 2700000 | 27250s | train_loss=0.0178 | fid=39.8427
273
+ step 2710000 | 27340s | train_loss=0.0181
274
+ step 2720000 | 27433s | train_loss=0.0172
275
+ step 2730000 | 27527s | train_loss=0.0172
276
+ step 2740000 | 27620s | train_loss=0.0165
277
+ step 2750000 | 27712s | train_loss=0.0169
278
+ step 2760000 | 27804s | train_loss=0.0174
279
+ step 2770000 | 27895s | train_loss=0.0164
280
+ step 2780000 | 27988s | train_loss=0.0166
281
+ step 2790000 | 28083s | train_loss=0.0166
282
+ step 2800000 | 28251s | train_loss=0.0167 | fid=40.1233
283
+ step 2810000 | 28341s | train_loss=0.0174
284
+ step 2820000 | 28432s | train_loss=0.0178
285
+ step 2830000 | 28520s | train_loss=0.0167
286
+ step 2840000 | 28608s | train_loss=0.0173
287
+ step 2850000 | 28693s | train_loss=0.0169
288
+ step 2860000 | 28779s | train_loss=0.0174
289
+ step 2870000 | 28872s | train_loss=0.0165
290
+ step 2880000 | 28967s | train_loss=0.0164
291
+ step 2890000 | 29061s | train_loss=0.0164
292
+ step 2900000 | 29228s | train_loss=0.0155 | fid=39.9649
293
+ step 2910000 | 29320s | train_loss=0.0159
294
+ step 2920000 | 29412s | train_loss=0.0163
295
+ step 2930000 | 29500s | train_loss=0.0161
296
+ step 2940000 | 29588s | train_loss=0.0156
297
+ step 2950000 | 29675s | train_loss=0.0155
298
+ step 2960000 | 29765s | train_loss=0.0166
299
+ step 2970000 | 29860s | train_loss=0.0157
300
+ step 2980000 | 29954s | train_loss=0.0152
301
+ step 2990000 | 30048s | train_loss=0.0164
302
+ step 3000000 | 30217s | train_loss=0.0158 | fid=40.1291
303
+ step 3010000 | 30309s | train_loss=0.0160
304
+ step 3020000 | 30404s | train_loss=0.0158
305
+ step 3030000 | 30498s | train_loss=0.0161
306
+ step 3040000 | 30593s | train_loss=0.0174
307
+ step 3050000 | 30683s | train_loss=0.0165
308
+ step 3060000 | 30777s | train_loss=0.0161
309
+ step 3070000 | 30872s | train_loss=0.0151
310
+ step 3080000 | 30965s | train_loss=0.0161
311
+ step 3090000 | 31060s | train_loss=0.0151
312
+ step 3100000 | 31229s | train_loss=0.0163 | fid=39.6787
313
+ step 3110000 | 31320s | train_loss=0.0161
314
+ step 3120000 | 31414s | train_loss=0.0150
315
+ step 3130000 | 31508s | train_loss=0.0148
316
+ step 3140000 | 31601s | train_loss=0.0152
317
+ step 3150000 | 31680s | train_loss=0.0159
318
+ step 3160000 | 31772s | train_loss=0.0149
319
+ step 3170000 | 31864s | train_loss=0.0158
320
+ step 3180000 | 31957s | train_loss=0.0147
321
+ step 3190000 | 32051s | train_loss=0.0144
322
+ step 3200000 | 32218s | train_loss=0.0147 | fid=39.6359
323
+ step 3210000 | 32308s | train_loss=0.0152
324
+ step 3220000 | 32401s | train_loss=0.0144
325
+ step 3230000 | 32495s | train_loss=0.0154
326
+ step 3240000 | 32589s | train_loss=0.0146
327
+ step 3250000 | 32679s | train_loss=0.0151
328
+ step 3260000 | 32773s | train_loss=0.0146
329
+ step 3270000 | 32867s | train_loss=0.0157
330
+ step 3280000 | 32960s | train_loss=0.0147
331
+ step 3290000 | 33053s | train_loss=0.0140
332
+ step 3300000 | 33219s | train_loss=0.0143 | fid=39.9308
333
+ step 3310000 | 33310s | train_loss=0.0143
334
+ step 3320000 | 33402s | train_loss=0.0140
335
+ step 3330000 | 33496s | train_loss=0.0149
336
+ step 3340000 | 33587s | train_loss=0.0144
337
+ step 3350000 | 33682s | train_loss=0.0149
338
+ step 3360000 | 33774s | train_loss=0.0149
339
+ step 3370000 | 33867s | train_loss=0.0139
340
+ step 3380000 | 33960s | train_loss=0.0152
341
+ step 3390000 | 34055s | train_loss=0.0146
342
+ step 3400000 | 34253s | train_loss=0.0147 | fid=39.2905
343
+ step 3410000 | 34345s | train_loss=0.0144
344
+ step 3420000 | 34440s | train_loss=0.0147
345
+ step 3430000 | 34532s | train_loss=0.0142
346
+ step 3440000 | 34617s | train_loss=0.0138
347
+ step 3450000 | 34710s | train_loss=0.0140
348
+ step 3460000 | 34803s | train_loss=0.0140
349
+ step 3470000 | 34898s | train_loss=0.0138
350
+ step 3480000 | 34992s | train_loss=0.0140
351
+ step 3490000 | 35087s | train_loss=0.0137
352
+ step 3500000 | 35299s | train_loss=0.0137 | fid=39.6349
353
+ step 3510000 | 35391s | train_loss=0.0138
354
+ step 3520000 | 35486s | train_loss=0.0141
355
+ step 3530000 | 35567s | train_loss=0.0137
356
+ step 3540000 | 35662s | train_loss=0.0132
357
+ step 3550000 | 35754s | train_loss=0.0135
358
+ step 3560000 | 35849s | train_loss=0.0140
359
+ step 3570000 | 35943s | train_loss=0.0133
360
+ step 3580000 | 36038s | train_loss=0.0135
361
+ step 3590000 | 36132s | train_loss=0.0136
362
+ step 3600000 | 36309s | train_loss=0.0137 | fid=40.4410
363
+ step 3610000 | 36402s | train_loss=0.0138
364
+ step 3620000 | 36496s | train_loss=0.0139
365
+ step 3630000 | 36576s | train_loss=0.0131
366
+ step 3640000 | 36671s | train_loss=0.0135
367
+ step 3650000 | 36764s | train_loss=0.0134
368
+ step 3660000 | 36858s | train_loss=0.0143
369
+ step 3670000 | 36953s | train_loss=0.0135
370
+ step 3680000 | 37047s | train_loss=0.0133
371
+ step 3690000 | 37141s | train_loss=0.0134
372
+ step 3700000 | 37309s | train_loss=0.0133 | fid=39.2224
373
+ step 3710000 | 37402s | train_loss=0.0139
374
+ step 3720000 | 37496s | train_loss=0.0128
375
+ step 3730000 | 37579s | train_loss=0.0134
376
+ step 3740000 | 37673s | train_loss=0.0137
377
+ step 3750000 | 37766s | train_loss=0.0146
378
+ step 3760000 | 37861s | train_loss=0.0135
379
+ step 3770000 | 37955s | train_loss=0.0125
380
+ step 3780000 | 38050s | train_loss=0.0130
381
+ step 3790000 | 38144s | train_loss=0.0130
382
+ step 3800000 | 38311s | train_loss=0.0128 | fid=40.1168
383
+ step 3810000 | 38405s | train_loss=0.0128
384
+ step 3820000 | 38497s | train_loss=0.0131
385
+ step 3830000 | 38581s | train_loss=0.0131
386
+ step 3840000 | 38676s | train_loss=0.0129
387
+ step 3850000 | 38770s | train_loss=0.0128
388
+ step 3860000 | 38863s | train_loss=0.0135
389
+ step 3870000 | 38958s | train_loss=0.0128
390
+ step 3880000 | 39053s | train_loss=0.0134
391
+ step 3890000 | 39147s | train_loss=0.0129
392
+ step 3900000 | 39314s | train_loss=0.0125 | fid=40.0467
393
+ step 3910000 | 39406s | train_loss=0.0131
394
+ step 3920000 | 39497s | train_loss=0.0128
395
+ step 3930000 | 39583s | train_loss=0.0129
396
+ step 3940000 | 39678s | train_loss=0.0125
397
+ step 3950000 | 39771s | train_loss=0.0130
398
+ step 3960000 | 39865s | train_loss=0.0125
399
+ step 3970000 | 39960s | train_loss=0.0130
400
+ step 3980000 | 40054s | train_loss=0.0127
401
+ step 3990000 | 40148s | train_loss=0.0123
402
+ step 4000000 | 40316s | train_loss=0.0121 | fid=39.6458
403
+ step 4010000 | 40408s | train_loss=0.0121
404
+ step 4020000 | 40497s | train_loss=0.0129
405
+ step 4030000 | 40584s | train_loss=0.0121
406
+ step 4040000 | 40678s | train_loss=0.0121
407
+ step 4050000 | 40772s | train_loss=0.0132
408
+ step 4060000 | 40867s | train_loss=0.0127
409
+ step 4070000 | 40962s | train_loss=0.0117
410
+ step 4080000 | 41056s | train_loss=0.0126
411
+ step 4090000 | 41150s | train_loss=0.0125
412
+ step 4100000 | 41312s | train_loss=0.0123 | fid=39.7908
413
+ step 4110000 | 41401s | train_loss=0.0121
414
+ step 4120000 | 41486s | train_loss=0.0127
415
+ step 4130000 | 41574s | train_loss=0.0117
416
+ step 4140000 | 41664s | train_loss=0.0122
417
+ step 4150000 | 41755s | train_loss=0.0128
418
+ step 4160000 | 41845s | train_loss=0.0122
419
+ step 4170000 | 41936s | train_loss=0.0123
420
+ step 4180000 | 42026s | train_loss=0.0123
421
+ step 4190000 | 42117s | train_loss=0.0129
422
+ step 4200000 | 42278s | train_loss=0.0125 | fid=40.1447
423
+ step 4210000 | 42367s | train_loss=0.0117
424
+ step 4220000 | 42454s | train_loss=0.0121
425
+ step 4230000 | 42544s | train_loss=0.0125
426
+ step 4240000 | 42635s | train_loss=0.0119
427
+ step 4250000 | 42725s | train_loss=0.0120
428
+ step 4260000 | 42814s | train_loss=0.0117
429
+ step 4270000 | 42905s | train_loss=0.0119
430
+ step 4280000 | 42995s | train_loss=0.0125
431
+ step 4290000 | 43085s | train_loss=0.0117
432
+ step 4300000 | 43248s | train_loss=0.0122 | fid=40.2323
433
+ step 4310000 | 43337s | train_loss=0.0113
434
+ step 4320000 | 43424s | train_loss=0.0117
435
+ step 4330000 | 43515s | train_loss=0.0118
436
+ step 4340000 | 43605s | train_loss=0.0120
437
+ step 4350000 | 43696s | train_loss=0.0118
438
+ step 4360000 | 43785s | train_loss=0.0120
439
+ step 4370000 | 43876s | train_loss=0.0121
440
+ step 4380000 | 43966s | train_loss=0.0120
441
+ step 4390000 | 44057s | train_loss=0.0111
442
+ step 4400000 | 44219s | train_loss=0.0117 | fid=40.4408
443
+ step 4410000 | 44307s | train_loss=0.0119
444
+ step 4420000 | 44394s | train_loss=0.0115
445
+ step 4430000 | 44484s | train_loss=0.0116
446
+ step 4440000 | 44574s | train_loss=0.0111
447
+ step 4450000 | 44665s | train_loss=0.0114
448
+ step 4460000 | 44754s | train_loss=0.0113
449
+ step 4470000 | 44845s | train_loss=0.0124
450
+ step 4480000 | 44936s | train_loss=0.0113
451
+ step 4490000 | 45026s | train_loss=0.0116
452
+ step 4500000 | 45189s | train_loss=0.0116 | fid=39.4711
453
+ step 4510000 | 45276s | train_loss=0.0115
454
+ step 4520000 | 45366s | train_loss=0.0119
455
+ step 4530000 | 45456s | train_loss=0.0115
456
+ step 4540000 | 45546s | train_loss=0.0115
457
+ step 4550000 | 45636s | train_loss=0.0117
458
+ step 4560000 | 45725s | train_loss=0.0115
459
+ step 4570000 | 45816s | train_loss=0.0112
460
+ step 4580000 | 45907s | train_loss=0.0112
461
+ step 4590000 | 45998s | train_loss=0.0110
462
+ step 4600000 | 46164s | train_loss=0.0111 | fid=39.5546
463
+ step 4610000 | 46252s | train_loss=0.0112
464
+ step 4620000 | 46346s | train_loss=0.0109
465
+ step 4630000 | 46440s | train_loss=0.0113
466
+ step 4640000 | 46535s | train_loss=0.0111
467
+ step 4650000 | 46630s | train_loss=0.0110
468
+ step 4660000 | 46723s | train_loss=0.0112
469
+ step 4670000 | 46818s | train_loss=0.0109
470
+ step 4680000 | 46913s | train_loss=0.0114
471
+ step 4690000 | 47008s | train_loss=0.0110
472
+ step 4700000 | 47174s | train_loss=0.0116 | fid=39.9821
473
+ step 4710000 | 47265s | train_loss=0.0109
474
+ step 4720000 | 47360s | train_loss=0.0109
475
+ step 4730000 | 47455s | train_loss=0.0110
476
+ step 4740000 | 47550s | train_loss=0.0113
477
+ step 4750000 | 47645s | train_loss=0.0116
478
+ step 4760000 | 47740s | train_loss=0.0113
479
+ step 4770000 | 47834s | train_loss=0.0106
480
+ step 4780000 | 47929s | train_loss=0.0108
481
+ step 4790000 | 48024s | train_loss=0.0112
482
+ step 4800000 | 48177s | train_loss=0.0107 | fid=39.3969
483
+ step 4810000 | 48249s | train_loss=0.0111
484
+ step 4820000 | 48326s | train_loss=0.0113
485
+ step 4830000 | 48401s | train_loss=0.0117
486
+ step 4840000 | 48476s | train_loss=0.0111
487
+ step 4850000 | 48552s | train_loss=0.0107
488
+ step 4860000 | 48628s | train_loss=0.0108
489
+ step 4870000 | 48702s | train_loss=0.0104
490
+ step 4880000 | 48777s | train_loss=0.0103
491
+ step 4890000 | 48851s | train_loss=0.0112
492
+ step 4900000 | 48988s | train_loss=0.0111 | fid=39.9446
493
+ step 4910000 | 49057s | train_loss=0.0107
494
+ step 4920000 | 49134s | train_loss=0.0115
495
+ step 4930000 | 49211s | train_loss=0.0101
496
+ step 4940000 | 49288s | train_loss=0.0106
497
+ step 4950000 | 49365s | train_loss=0.0109
498
+ step 4960000 | 49441s | train_loss=0.0108
499
+ step 4970000 | 49515s | train_loss=0.0107
500
+ step 4980000 | 49590s | train_loss=0.0109
501
+ step 4990000 | 49666s | train_loss=0.0119
502
+ step 5000000 | 49804s | train_loss=0.0105 | fid=40.0152
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d90/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 43,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed43/d90/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 71s | train_loss=0.9971 | fid=304.9606
3
+ step 10000 | 155s | train_loss=0.8111
4
+ step 20000 | 244s | train_loss=0.6950
5
+ step 30000 | 333s | train_loss=0.5601
6
+ step 40000 | 422s | train_loss=0.4138
7
+ step 50000 | 510s | train_loss=0.2936
8
+ step 60000 | 599s | train_loss=0.2199
9
+ step 70000 | 688s | train_loss=0.1773
10
+ step 80000 | 777s | train_loss=0.1524
11
+ step 90000 | 866s | train_loss=0.1393
12
+ step 100000 | 1024s | train_loss=0.1285 | fid=58.0634
13
+ step 110000 | 1104s | train_loss=0.1203
14
+ step 120000 | 1193s | train_loss=0.1142
15
+ step 130000 | 1282s | train_loss=0.1094
16
+ step 140000 | 1371s | train_loss=0.1052
17
+ step 150000 | 1460s | train_loss=0.0990
18
+ step 160000 | 1549s | train_loss=0.0957
19
+ step 170000 | 1638s | train_loss=0.0919
20
+ step 180000 | 1727s | train_loss=0.0894
21
+ step 190000 | 1816s | train_loss=0.0892
22
+ step 200000 | 1973s | train_loss=0.0830 | fid=58.0931
23
+ step 210000 | 2063s | train_loss=0.0834
24
+ step 220000 | 2146s | train_loss=0.0807
25
+ step 230000 | 2235s | train_loss=0.0772
26
+ step 240000 | 2323s | train_loss=0.0744
27
+ step 250000 | 2412s | train_loss=0.0756
28
+ step 260000 | 2501s | train_loss=0.0732
29
+ step 270000 | 2590s | train_loss=0.0702
30
+ step 280000 | 2679s | train_loss=0.0695
31
+ step 290000 | 2768s | train_loss=0.0697
32
+ step 300000 | 2927s | train_loss=0.0661 | fid=56.6222
33
+ step 310000 | 3016s | train_loss=0.0656
34
+ step 320000 | 3107s | train_loss=0.0671
35
+ step 330000 | 3188s | train_loss=0.0624
36
+ step 340000 | 3277s | train_loss=0.0635
37
+ step 350000 | 3366s | train_loss=0.0609
38
+ step 360000 | 3455s | train_loss=0.0612
39
+ step 370000 | 3544s | train_loss=0.0588
40
+ step 380000 | 3633s | train_loss=0.0580
41
+ step 390000 | 3721s | train_loss=0.0582
42
+ step 400000 | 3881s | train_loss=0.0567 | fid=54.3975
43
+ step 410000 | 3970s | train_loss=0.0559
44
+ step 420000 | 4059s | train_loss=0.0556
45
+ step 430000 | 4143s | train_loss=0.0557
46
+ step 440000 | 4231s | train_loss=0.0554
47
+ step 450000 | 4319s | train_loss=0.0552
48
+ step 460000 | 4408s | train_loss=0.0538
49
+ step 470000 | 4497s | train_loss=0.0528
50
+ step 480000 | 4586s | train_loss=0.0524
51
+ step 490000 | 4675s | train_loss=0.0518
52
+ step 500000 | 4836s | train_loss=0.0512 | fid=52.3860
53
+ step 510000 | 4925s | train_loss=0.0519
54
+ step 520000 | 5014s | train_loss=0.0494
55
+ step 530000 | 5104s | train_loss=0.0502
56
+ step 540000 | 5189s | train_loss=0.0494
57
+ step 550000 | 5276s | train_loss=0.0491
58
+ step 560000 | 5365s | train_loss=0.0483
59
+ step 570000 | 5454s | train_loss=0.0495
60
+ step 580000 | 5543s | train_loss=0.0484
61
+ step 590000 | 5632s | train_loss=0.0486
62
+ step 600000 | 5793s | train_loss=0.0472 | fid=50.6112
63
+ step 610000 | 5882s | train_loss=0.0474
64
+ step 620000 | 5971s | train_loss=0.0465
65
+ step 630000 | 6060s | train_loss=0.0451
66
+ step 640000 | 6150s | train_loss=0.0457
67
+ step 650000 | 6233s | train_loss=0.0442
68
+ step 660000 | 6321s | train_loss=0.0457
69
+ step 670000 | 6410s | train_loss=0.0439
70
+ step 680000 | 6499s | train_loss=0.0443
71
+ step 690000 | 6588s | train_loss=0.0447
72
+ step 700000 | 6750s | train_loss=0.0444 | fid=50.0330
73
+ step 710000 | 6838s | train_loss=0.0445
74
+ step 720000 | 6927s | train_loss=0.0439
75
+ step 730000 | 7016s | train_loss=0.0426
76
+ step 740000 | 7105s | train_loss=0.0433
77
+ step 750000 | 7190s | train_loss=0.0419
78
+ step 760000 | 7277s | train_loss=0.0416
79
+ step 770000 | 7366s | train_loss=0.0412
80
+ step 780000 | 7455s | train_loss=0.0412
81
+ step 790000 | 7544s | train_loss=0.0414
82
+ step 800000 | 7706s | train_loss=0.0405 | fid=48.8353
83
+ step 810000 | 7795s | train_loss=0.0402
84
+ step 820000 | 7884s | train_loss=0.0399
85
+ step 830000 | 7973s | train_loss=0.0409
86
+ step 840000 | 8062s | train_loss=0.0409
87
+ step 850000 | 8151s | train_loss=0.0398
88
+ step 860000 | 8234s | train_loss=0.0409
89
+ step 870000 | 8322s | train_loss=0.0395
90
+ step 880000 | 8411s | train_loss=0.0402
91
+ step 890000 | 8500s | train_loss=0.0396
92
+ step 900000 | 8662s | train_loss=0.0388 | fid=48.7647
93
+ step 910000 | 8751s | train_loss=0.0379
94
+ step 920000 | 8840s | train_loss=0.0381
95
+ step 930000 | 8929s | train_loss=0.0387
96
+ step 940000 | 9018s | train_loss=0.0371
97
+ step 950000 | 9107s | train_loss=0.0386
98
+ step 960000 | 9196s | train_loss=0.0398
99
+ step 970000 | 9281s | train_loss=0.0367
100
+ step 980000 | 9368s | train_loss=0.0372
101
+ step 990000 | 9457s | train_loss=0.0357
102
+ step 1000000 | 9617s | train_loss=0.0373 | fid=47.5269
103
+ step 1010000 | 9706s | train_loss=0.0375
104
+ step 1020000 | 9795s | train_loss=0.0370
105
+ step 1030000 | 9883s | train_loss=0.0368
106
+ step 1040000 | 9972s | train_loss=0.0354
107
+ step 1050000 | 10061s | train_loss=0.0362
108
+ step 1060000 | 10150s | train_loss=0.0369
109
+ step 1070000 | 10235s | train_loss=0.0356
110
+ step 1080000 | 10323s | train_loss=0.0367
111
+ step 1090000 | 10411s | train_loss=0.0368
112
+ step 1100000 | 10574s | train_loss=0.0349 | fid=47.3856
113
+ step 1110000 | 10663s | train_loss=0.0339
114
+ step 1120000 | 10751s | train_loss=0.0358
115
+ step 1130000 | 10840s | train_loss=0.0356
116
+ step 1140000 | 10929s | train_loss=0.0335
117
+ step 1150000 | 11018s | train_loss=0.0339
118
+ step 1160000 | 11107s | train_loss=0.0342
119
+ step 1170000 | 11197s | train_loss=0.0348
120
+ step 1180000 | 11281s | train_loss=0.0330
121
+ step 1190000 | 11369s | train_loss=0.0333
122
+ step 1200000 | 11532s | train_loss=0.0332 | fid=46.7593
123
+ step 1210000 | 11620s | train_loss=0.0320
124
+ step 1220000 | 11709s | train_loss=0.0324
125
+ step 1230000 | 11798s | train_loss=0.0330
126
+ step 1240000 | 11887s | train_loss=0.0322
127
+ step 1250000 | 11976s | train_loss=0.0344
128
+ step 1260000 | 12065s | train_loss=0.0327
129
+ step 1270000 | 12154s | train_loss=0.0321
130
+ step 1280000 | 12241s | train_loss=0.0320
131
+ step 1290000 | 12328s | train_loss=0.0314
132
+ step 1300000 | 12489s | train_loss=0.0317 | fid=46.6235
133
+ step 1310000 | 12577s | train_loss=0.0315
134
+ step 1320000 | 12666s | train_loss=0.0307
135
+ step 1330000 | 12755s | train_loss=0.0316
136
+ step 1340000 | 12844s | train_loss=0.0332
137
+ step 1350000 | 12933s | train_loss=0.0306
138
+ step 1360000 | 13022s | train_loss=0.0301
139
+ step 1370000 | 13111s | train_loss=0.0313
140
+ step 1380000 | 13200s | train_loss=0.0314
141
+ step 1390000 | 13284s | train_loss=0.0312
142
+ step 1400000 | 13443s | train_loss=0.0300 | fid=47.3859
143
+ step 1410000 | 13532s | train_loss=0.0303
144
+ step 1420000 | 13621s | train_loss=0.0288
145
+ step 1430000 | 13709s | train_loss=0.0301
146
+ step 1440000 | 13798s | train_loss=0.0302
147
+ step 1450000 | 13887s | train_loss=0.0293
148
+ step 1460000 | 13976s | train_loss=0.0295
149
+ step 1470000 | 14065s | train_loss=0.0295
150
+ step 1480000 | 14154s | train_loss=0.0292
151
+ step 1490000 | 14244s | train_loss=0.0286
152
+ step 1500000 | 14401s | train_loss=0.0286 | fid=46.2391
153
+ step 1510000 | 14489s | train_loss=0.0296
154
+ step 1520000 | 14577s | train_loss=0.0293
155
+ step 1530000 | 14666s | train_loss=0.0282
156
+ step 1540000 | 14755s | train_loss=0.0276
157
+ step 1550000 | 14844s | train_loss=0.0276
158
+ step 1560000 | 14933s | train_loss=0.0283
159
+ step 1570000 | 15022s | train_loss=0.0288
160
+ step 1580000 | 15111s | train_loss=0.0281
161
+ step 1590000 | 15199s | train_loss=0.0275
162
+ step 1600000 | 15390s | train_loss=0.0276 | fid=45.4969
163
+ step 1610000 | 15477s | train_loss=0.0272
164
+ step 1620000 | 15565s | train_loss=0.0269
165
+ step 1630000 | 15654s | train_loss=0.0271
166
+ step 1640000 | 15742s | train_loss=0.0280
167
+ step 1650000 | 15831s | train_loss=0.0268
168
+ step 1660000 | 15920s | train_loss=0.0271
169
+ step 1670000 | 16009s | train_loss=0.0267
170
+ step 1680000 | 16097s | train_loss=0.0257
171
+ step 1690000 | 16186s | train_loss=0.0271
172
+ step 1700000 | 16342s | train_loss=0.0260 | fid=45.8536
173
+ step 1710000 | 16431s | train_loss=0.0259
174
+ step 1720000 | 16519s | train_loss=0.0255
175
+ step 1730000 | 16607s | train_loss=0.0266
176
+ step 1740000 | 16696s | train_loss=0.0259
177
+ step 1750000 | 16785s | train_loss=0.0258
178
+ step 1760000 | 16874s | train_loss=0.0277
179
+ step 1770000 | 16963s | train_loss=0.0258
180
+ step 1780000 | 17052s | train_loss=0.0249
181
+ step 1790000 | 17141s | train_loss=0.0257
182
+ step 1800000 | 17302s | train_loss=0.0248 | fid=43.9300
183
+ step 1810000 | 17391s | train_loss=0.0240
184
+ step 1820000 | 17479s | train_loss=0.0241
185
+ step 1830000 | 17568s | train_loss=0.0240
186
+ step 1840000 | 17657s | train_loss=0.0247
187
+ step 1850000 | 17746s | train_loss=0.0250
188
+ step 1860000 | 17835s | train_loss=0.0247
189
+ step 1870000 | 17923s | train_loss=0.0237
190
+ step 1880000 | 18012s | train_loss=0.0238
191
+ step 1890000 | 18101s | train_loss=0.0259
192
+ step 1900000 | 18263s | train_loss=0.0239 | fid=44.2214
193
+ step 1910000 | 18348s | train_loss=0.0235
194
+ step 1920000 | 18437s | train_loss=0.0242
195
+ step 1930000 | 18525s | train_loss=0.0240
196
+ step 1940000 | 18614s | train_loss=0.0236
197
+ step 1950000 | 18703s | train_loss=0.0240
198
+ step 1960000 | 18791s | train_loss=0.0231
199
+ step 1970000 | 18880s | train_loss=0.0228
200
+ step 1980000 | 18969s | train_loss=0.0239
201
+ step 1990000 | 19058s | train_loss=0.0236
202
+ step 2000000 | 19219s | train_loss=0.0232 | fid=43.4767
203
+ step 2010000 | 19306s | train_loss=0.0235
204
+ step 2020000 | 19392s | train_loss=0.0228
205
+ step 2030000 | 19482s | train_loss=0.0227
206
+ step 2040000 | 19569s | train_loss=0.0221
207
+ step 2050000 | 19657s | train_loss=0.0220
208
+ step 2060000 | 19746s | train_loss=0.0231
209
+ step 2070000 | 19834s | train_loss=0.0225
210
+ step 2080000 | 19922s | train_loss=0.0223
211
+ step 2090000 | 20011s | train_loss=0.0219
212
+ step 2100000 | 20174s | train_loss=0.0231 | fid=43.7887
213
+ step 2110000 | 20264s | train_loss=0.0222
214
+ step 2120000 | 20348s | train_loss=0.0240
215
+ step 2130000 | 20437s | train_loss=0.0222
216
+ step 2140000 | 20526s | train_loss=0.0220
217
+ step 2150000 | 20613s | train_loss=0.0222
218
+ step 2160000 | 20702s | train_loss=0.0230
219
+ step 2170000 | 20791s | train_loss=0.0222
220
+ step 2180000 | 20879s | train_loss=0.0220
221
+ step 2190000 | 20967s | train_loss=0.0217
222
+ step 2200000 | 21129s | train_loss=0.0217 | fid=42.9387
223
+ step 2210000 | 21218s | train_loss=0.0223
224
+ step 2220000 | 21306s | train_loss=0.0209
225
+ step 2230000 | 21390s | train_loss=0.0213
226
+ step 2240000 | 21478s | train_loss=0.0216
227
+ step 2250000 | 21565s | train_loss=0.0217
228
+ step 2260000 | 21654s | train_loss=0.0217
229
+ step 2270000 | 21742s | train_loss=0.0220
230
+ step 2280000 | 21831s | train_loss=0.0203
231
+ step 2290000 | 21920s | train_loss=0.0209
232
+ step 2300000 | 22082s | train_loss=0.0210 | fid=41.9826
233
+ step 2310000 | 22171s | train_loss=0.0221
234
+ step 2320000 | 22261s | train_loss=0.0217
235
+ step 2330000 | 22344s | train_loss=0.0207
236
+ step 2340000 | 22433s | train_loss=0.0199
237
+ step 2350000 | 22522s | train_loss=0.0203
238
+ step 2360000 | 22609s | train_loss=0.0217
239
+ step 2370000 | 22698s | train_loss=0.0205
240
+ step 2380000 | 22787s | train_loss=0.0195
241
+ step 2390000 | 22876s | train_loss=0.0200
242
+ step 2400000 | 23080s | train_loss=0.0202 | fid=41.6671
243
+ step 2410000 | 23168s | train_loss=0.0203
244
+ step 2420000 | 23257s | train_loss=0.0200
245
+ step 2430000 | 23340s | train_loss=0.0199
246
+ step 2440000 | 23428s | train_loss=0.0198
247
+ step 2450000 | 23517s | train_loss=0.0204
248
+ step 2460000 | 23604s | train_loss=0.0197
249
+ step 2470000 | 23692s | train_loss=0.0206
250
+ step 2480000 | 23781s | train_loss=0.0202
251
+ step 2490000 | 23869s | train_loss=0.0191
252
+ step 2500000 | 24028s | train_loss=0.0198 | fid=41.7906
253
+ step 2510000 | 24117s | train_loss=0.0202
254
+ step 2520000 | 24206s | train_loss=0.0195
255
+ step 2530000 | 24294s | train_loss=0.0193
256
+ step 2540000 | 24378s | train_loss=0.0191
257
+ step 2550000 | 24466s | train_loss=0.0197
258
+ step 2560000 | 24555s | train_loss=0.0193
259
+ step 2570000 | 24642s | train_loss=0.0191
260
+ step 2580000 | 24730s | train_loss=0.0190
261
+ step 2590000 | 24819s | train_loss=0.0198
262
+ step 2600000 | 24980s | train_loss=0.0192 | fid=40.9371
263
+ step 2610000 | 25069s | train_loss=0.0185
264
+ step 2620000 | 25157s | train_loss=0.0187
265
+ step 2630000 | 25246s | train_loss=0.0183
266
+ step 2640000 | 25330s | train_loss=0.0186
267
+ step 2650000 | 25418s | train_loss=0.0183
268
+ step 2660000 | 25506s | train_loss=0.0191
269
+ step 2670000 | 25595s | train_loss=0.0186
270
+ step 2680000 | 25682s | train_loss=0.0181
271
+ step 2690000 | 25770s | train_loss=0.0185
272
+ step 2700000 | 25929s | train_loss=0.0192 | fid=41.0889
273
+ step 2710000 | 26017s | train_loss=0.0185
274
+ step 2720000 | 26106s | train_loss=0.0182
275
+ step 2730000 | 26194s | train_loss=0.0188
276
+ step 2740000 | 26282s | train_loss=0.0180
277
+ step 2750000 | 26367s | train_loss=0.0183
278
+ step 2760000 | 26456s | train_loss=0.0184
279
+ step 2770000 | 26544s | train_loss=0.0174
280
+ step 2780000 | 26631s | train_loss=0.0181
281
+ step 2790000 | 26720s | train_loss=0.0179
282
+ step 2800000 | 26880s | train_loss=0.0179 | fid=40.2600
283
+ step 2810000 | 26969s | train_loss=0.0170
284
+ step 2820000 | 27057s | train_loss=0.0179
285
+ step 2830000 | 27146s | train_loss=0.0173
286
+ step 2840000 | 27235s | train_loss=0.0171
287
+ step 2850000 | 27320s | train_loss=0.0186
288
+ step 2860000 | 27407s | train_loss=0.0175
289
+ step 2870000 | 27496s | train_loss=0.0172
290
+ step 2880000 | 27584s | train_loss=0.0169
291
+ step 2890000 | 27671s | train_loss=0.0174
292
+ step 2900000 | 27830s | train_loss=0.0176 | fid=39.8393
293
+ step 2910000 | 27919s | train_loss=0.0170
294
+ step 2920000 | 28007s | train_loss=0.0177
295
+ step 2930000 | 28096s | train_loss=0.0171
296
+ step 2940000 | 28185s | train_loss=0.0187
297
+ step 2950000 | 28273s | train_loss=0.0173
298
+ step 2960000 | 28358s | train_loss=0.0169
299
+ step 2970000 | 28445s | train_loss=0.0172
300
+ step 2980000 | 28532s | train_loss=0.0184
301
+ step 2990000 | 28618s | train_loss=0.0172
302
+ step 3000000 | 28773s | train_loss=0.0165 | fid=39.7965
303
+ step 3010000 | 28862s | train_loss=0.0172
304
+ step 3020000 | 28951s | train_loss=0.0169
305
+ step 3030000 | 29040s | train_loss=0.0164
306
+ step 3040000 | 29128s | train_loss=0.0172
307
+ step 3050000 | 29218s | train_loss=0.0172
308
+ step 3060000 | 29304s | train_loss=0.0166
309
+ step 3070000 | 29391s | train_loss=0.0167
310
+ step 3080000 | 29478s | train_loss=0.0174
311
+ step 3090000 | 29564s | train_loss=0.0167
312
+ step 3100000 | 29721s | train_loss=0.0167 | fid=39.4477
313
+ step 3110000 | 29810s | train_loss=0.0171
314
+ step 3120000 | 29899s | train_loss=0.0159
315
+ step 3130000 | 29988s | train_loss=0.0161
316
+ step 3140000 | 30077s | train_loss=0.0171
317
+ step 3150000 | 30166s | train_loss=0.0168
318
+ step 3160000 | 30254s | train_loss=0.0160
319
+ step 3170000 | 30340s | train_loss=0.0168
320
+ step 3180000 | 30429s | train_loss=0.0182
321
+ step 3190000 | 30518s | train_loss=0.0165
322
+ step 3200000 | 30678s | train_loss=0.0170 | fid=41.2066
323
+ step 3210000 | 30767s | train_loss=0.0165
324
+ step 3220000 | 30856s | train_loss=0.0165
325
+ step 3230000 | 30944s | train_loss=0.0167
326
+ step 3240000 | 31033s | train_loss=0.0164
327
+ step 3250000 | 31122s | train_loss=0.0168
328
+ step 3260000 | 31212s | train_loss=0.0162
329
+ step 3270000 | 31300s | train_loss=0.0159
330
+ step 3280000 | 31385s | train_loss=0.0163
331
+ step 3290000 | 31474s | train_loss=0.0156
332
+ step 3300000 | 31683s | train_loss=0.0161 | fid=39.4391
333
+ step 3310000 | 31772s | train_loss=0.0159
334
+ step 3320000 | 31860s | train_loss=0.0155
335
+ step 3330000 | 31948s | train_loss=0.0151
336
+ step 3340000 | 32037s | train_loss=0.0155
337
+ step 3350000 | 32126s | train_loss=0.0162
338
+ step 3360000 | 32214s | train_loss=0.0150
339
+ step 3370000 | 32300s | train_loss=0.0158
340
+ step 3380000 | 32387s | train_loss=0.0159
341
+ step 3390000 | 32476s | train_loss=0.0154
342
+ step 3400000 | 32637s | train_loss=0.0154 | fid=39.7822
343
+ step 3410000 | 32725s | train_loss=0.0151
344
+ step 3420000 | 32813s | train_loss=0.0153
345
+ step 3430000 | 32902s | train_loss=0.0162
346
+ step 3440000 | 32991s | train_loss=0.0151
347
+ step 3450000 | 33079s | train_loss=0.0157
348
+ step 3460000 | 33167s | train_loss=0.0154
349
+ step 3470000 | 33255s | train_loss=0.0157
350
+ step 3480000 | 33340s | train_loss=0.0152
351
+ step 3490000 | 33429s | train_loss=0.0146
352
+ step 3500000 | 33588s | train_loss=0.0153 | fid=39.6846
353
+ step 3510000 | 33677s | train_loss=0.0156
354
+ step 3520000 | 33764s | train_loss=0.0154
355
+ step 3530000 | 33853s | train_loss=0.0149
356
+ step 3540000 | 33941s | train_loss=0.0156
357
+ step 3550000 | 34030s | train_loss=0.0146
358
+ step 3560000 | 34119s | train_loss=0.0143
359
+ step 3570000 | 34208s | train_loss=0.0153
360
+ step 3580000 | 34289s | train_loss=0.0142
361
+ step 3590000 | 34374s | train_loss=0.0146
362
+ step 3600000 | 34571s | train_loss=0.0148 | fid=40.1983
363
+ step 3610000 | 34660s | train_loss=0.0152
364
+ step 3620000 | 34747s | train_loss=0.0145
365
+ step 3630000 | 34836s | train_loss=0.0147
366
+ step 3640000 | 34925s | train_loss=0.0148
367
+ step 3650000 | 35014s | train_loss=0.0151
368
+ step 3660000 | 35103s | train_loss=0.0151
369
+ step 3670000 | 35191s | train_loss=0.0149
370
+ step 3680000 | 35275s | train_loss=0.0145
371
+ step 3690000 | 35353s | train_loss=0.0143
372
+ step 3700000 | 35555s | train_loss=0.0154 | fid=39.9048
373
+ step 3710000 | 35644s | train_loss=0.0143
374
+ step 3720000 | 35731s | train_loss=0.0148
375
+ step 3730000 | 35820s | train_loss=0.0140
376
+ step 3740000 | 35909s | train_loss=0.0140
377
+ step 3750000 | 35998s | train_loss=0.0143
378
+ step 3760000 | 36087s | train_loss=0.0143
379
+ step 3770000 | 36176s | train_loss=0.0144
380
+ step 3780000 | 36266s | train_loss=0.0152
381
+ step 3790000 | 36346s | train_loss=0.0136
382
+ step 3800000 | 36553s | train_loss=0.0148 | fid=39.8880
383
+ step 3810000 | 36641s | train_loss=0.0143
384
+ step 3820000 | 36729s | train_loss=0.0142
385
+ step 3830000 | 36818s | train_loss=0.0152
386
+ step 3840000 | 36907s | train_loss=0.0145
387
+ step 3850000 | 36996s | train_loss=0.0132
388
+ step 3860000 | 37085s | train_loss=0.0146
389
+ step 3870000 | 37173s | train_loss=0.0149
390
+ step 3880000 | 37263s | train_loss=0.0138
391
+ step 3890000 | 37347s | train_loss=0.0143
392
+ step 3900000 | 37545s | train_loss=0.0145 | fid=39.7396
393
+ step 3910000 | 37634s | train_loss=0.0150
394
+ step 3920000 | 37722s | train_loss=0.0143
395
+ step 3930000 | 37811s | train_loss=0.0133
396
+ step 3940000 | 37899s | train_loss=0.0137
397
+ step 3950000 | 37988s | train_loss=0.0147
398
+ step 3960000 | 38077s | train_loss=0.0137
399
+ step 3970000 | 38166s | train_loss=0.0141
400
+ step 3980000 | 38255s | train_loss=0.0152
401
+ step 3990000 | 38340s | train_loss=0.0142
402
+ step 4000000 | 38536s | train_loss=0.0138 | fid=39.2601
403
+ step 4010000 | 38625s | train_loss=0.0139
404
+ step 4020000 | 38714s | train_loss=0.0135
405
+ step 4030000 | 38802s | train_loss=0.0141
406
+ step 4040000 | 38890s | train_loss=0.0136
407
+ step 4050000 | 38979s | train_loss=0.0129
408
+ step 4060000 | 39068s | train_loss=0.0139
409
+ step 4070000 | 39157s | train_loss=0.0148
410
+ step 4080000 | 39246s | train_loss=0.0132
411
+ step 4090000 | 39331s | train_loss=0.0132
412
+ step 4100000 | 39530s | train_loss=0.0130 | fid=39.6340
413
+ step 4110000 | 39619s | train_loss=0.0134
414
+ step 4120000 | 39708s | train_loss=0.0132
415
+ step 4130000 | 39795s | train_loss=0.0140
416
+ step 4140000 | 39884s | train_loss=0.0129
417
+ step 4150000 | 39973s | train_loss=0.0137
418
+ step 4160000 | 40062s | train_loss=0.0133
419
+ step 4170000 | 40150s | train_loss=0.0137
420
+ step 4180000 | 40239s | train_loss=0.0126
421
+ step 4190000 | 40324s | train_loss=0.0137
422
+ step 4200000 | 40526s | train_loss=0.0130 | fid=40.2813
423
+ step 4210000 | 40615s | train_loss=0.0146
424
+ step 4220000 | 40704s | train_loss=0.0132
425
+ step 4230000 | 40792s | train_loss=0.0138
426
+ step 4240000 | 40881s | train_loss=0.0133
427
+ step 4250000 | 40970s | train_loss=0.0124
428
+ step 4260000 | 41059s | train_loss=0.0133
429
+ step 4270000 | 41148s | train_loss=0.0132
430
+ step 4280000 | 41235s | train_loss=0.0136
431
+ step 4290000 | 41319s | train_loss=0.0135
432
+ step 4300000 | 41498s | train_loss=0.0129 | fid=40.1062
433
+ step 4310000 | 41585s | train_loss=0.0131
434
+ step 4320000 | 41673s | train_loss=0.0131
435
+ step 4330000 | 41760s | train_loss=0.0142
436
+ step 4340000 | 41847s | train_loss=0.0126
437
+ step 4350000 | 41934s | train_loss=0.0126
438
+ step 4360000 | 42022s | train_loss=0.0126
439
+ step 4370000 | 42109s | train_loss=0.0130
440
+ step 4380000 | 42197s | train_loss=0.0143
441
+ step 4390000 | 42282s | train_loss=0.0132
442
+ step 4400000 | 42440s | train_loss=0.0123 | fid=39.5950
443
+ step 4410000 | 42527s | train_loss=0.0130
444
+ step 4420000 | 42615s | train_loss=0.0136
445
+ step 4430000 | 42702s | train_loss=0.0128
446
+ step 4440000 | 42788s | train_loss=0.0136
447
+ step 4450000 | 42875s | train_loss=0.0124
448
+ step 4460000 | 42963s | train_loss=0.0126
449
+ step 4470000 | 43050s | train_loss=0.0129
450
+ step 4480000 | 43138s | train_loss=0.0133
451
+ step 4490000 | 43226s | train_loss=0.0124
452
+ step 4500000 | 43380s | train_loss=0.0125 | fid=39.4576
453
+ step 4510000 | 43468s | train_loss=0.0126
454
+ step 4520000 | 43555s | train_loss=0.0131
455
+ step 4530000 | 43643s | train_loss=0.0127
456
+ step 4540000 | 43731s | train_loss=0.0130
457
+ step 4550000 | 43816s | train_loss=0.0135
458
+ step 4560000 | 43904s | train_loss=0.0129
459
+ step 4570000 | 43991s | train_loss=0.0120
460
+ step 4580000 | 44079s | train_loss=0.0123
461
+ step 4590000 | 44166s | train_loss=0.0128
462
+ step 4600000 | 44319s | train_loss=0.0118 | fid=39.3031
463
+ step 4610000 | 44407s | train_loss=0.0127
464
+ step 4620000 | 44494s | train_loss=0.0125
465
+ step 4630000 | 44581s | train_loss=0.0121
466
+ step 4640000 | 44669s | train_loss=0.0124
467
+ step 4650000 | 44755s | train_loss=0.0124
468
+ step 4660000 | 44842s | train_loss=0.0139
469
+ step 4670000 | 44930s | train_loss=0.0117
470
+ step 4680000 | 45017s | train_loss=0.0122
471
+ step 4690000 | 45105s | train_loss=0.0125
472
+ step 4700000 | 45256s | train_loss=0.0122 | fid=38.8237
473
+ step 4710000 | 45343s | train_loss=0.0119
474
+ step 4720000 | 45430s | train_loss=0.0125
475
+ step 4730000 | 45518s | train_loss=0.0125
476
+ step 4740000 | 45605s | train_loss=0.0126
477
+ step 4750000 | 45693s | train_loss=0.0118
478
+ step 4760000 | 45779s | train_loss=0.0121
479
+ step 4770000 | 45866s | train_loss=0.0122
480
+ step 4780000 | 45954s | train_loss=0.0117
481
+ step 4790000 | 46042s | train_loss=0.0128
482
+ step 4800000 | 46194s | train_loss=0.0122 | fid=39.5528
483
+ step 4810000 | 46282s | train_loss=0.0116
484
+ step 4820000 | 46371s | train_loss=0.0128
485
+ step 4830000 | 46460s | train_loss=0.0117
486
+ step 4840000 | 46549s | train_loss=0.0125
487
+ step 4850000 | 46638s | train_loss=0.0115
488
+ step 4860000 | 46725s | train_loss=0.0122
489
+ step 4870000 | 46814s | train_loss=0.0112
490
+ step 4880000 | 46903s | train_loss=0.0121
491
+ step 4890000 | 46992s | train_loss=0.0115
492
+ step 4900000 | 47150s | train_loss=0.0121 | fid=40.0329
493
+ step 4910000 | 47234s | train_loss=0.0119
494
+ step 4920000 | 47323s | train_loss=0.0116
495
+ step 4930000 | 47412s | train_loss=0.0114
496
+ step 4940000 | 47501s | train_loss=0.0121
497
+ step 4950000 | 47590s | train_loss=0.0124
498
+ step 4960000 | 47679s | train_loss=0.0118
499
+ step 4970000 | 47767s | train_loss=0.0118
500
+ step 4980000 | 47856s | train_loss=0.0121
501
+ step 4990000 | 47945s | train_loss=0.0116
502
+ step 5000000 | 48103s | train_loss=0.0116 | fid=39.5702
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d10/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 44,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d10/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 63s | train_loss=0.9700 | fid=193.3753
3
+ step 10000 | 133s | train_loss=0.8926
4
+ step 20000 | 218s | train_loss=0.8908
5
+ step 30000 | 302s | train_loss=0.8897
6
+ step 40000 | 387s | train_loss=0.8889
7
+ step 50000 | 471s | train_loss=0.8881
8
+ step 60000 | 556s | train_loss=0.8873
9
+ step 70000 | 640s | train_loss=0.8866
10
+ step 80000 | 725s | train_loss=0.8858
11
+ step 90000 | 809s | train_loss=0.8854
12
+ step 100000 | 959s | train_loss=0.8847 | fid=87.2855
13
+ step 110000 | 1040s | train_loss=0.8841
14
+ step 120000 | 1120s | train_loss=0.8834
15
+ step 130000 | 1204s | train_loss=0.8828
16
+ step 140000 | 1288s | train_loss=0.8823
17
+ step 150000 | 1373s | train_loss=0.8815
18
+ step 160000 | 1457s | train_loss=0.8809
19
+ step 170000 | 1541s | train_loss=0.8800
20
+ step 180000 | 1625s | train_loss=0.8792
21
+ step 190000 | 1709s | train_loss=0.8784
22
+ step 200000 | 1859s | train_loss=0.8776 | fid=86.9470
23
+ step 210000 | 1939s | train_loss=0.8764
24
+ step 220000 | 2023s | train_loss=0.8751
25
+ step 230000 | 2104s | train_loss=0.8739
26
+ step 240000 | 2188s | train_loss=0.8725
27
+ step 250000 | 2272s | train_loss=0.8711
28
+ step 260000 | 2357s | train_loss=0.8692
29
+ step 270000 | 2441s | train_loss=0.8670
30
+ step 280000 | 2525s | train_loss=0.8644
31
+ step 290000 | 2609s | train_loss=0.8611
32
+ step 300000 | 2759s | train_loss=0.8568 | fid=87.0508
33
+ step 310000 | 2839s | train_loss=0.8508
34
+ step 320000 | 2923s | train_loss=0.8413
35
+ step 330000 | 3006s | train_loss=0.8280
36
+ step 340000 | 3090s | train_loss=0.8115
37
+ step 350000 | 3172s | train_loss=0.7899
38
+ step 360000 | 3256s | train_loss=0.7625
39
+ step 370000 | 3341s | train_loss=0.7302
40
+ step 380000 | 3425s | train_loss=0.6904
41
+ step 390000 | 3509s | train_loss=0.6493
42
+ step 400000 | 3658s | train_loss=0.6081 | fid=87.3304
43
+ step 410000 | 3738s | train_loss=0.5590
44
+ step 420000 | 3822s | train_loss=0.5310
45
+ step 430000 | 3906s | train_loss=0.4833
46
+ step 440000 | 3989s | train_loss=0.4559
47
+ step 450000 | 4074s | train_loss=0.4315
48
+ step 460000 | 4156s | train_loss=0.3930
49
+ step 470000 | 4240s | train_loss=0.3751
50
+ step 480000 | 4324s | train_loss=0.3444
51
+ step 490000 | 4408s | train_loss=0.3342
52
+ step 500000 | 4558s | train_loss=0.3173 | fid=86.2262
53
+ step 510000 | 4639s | train_loss=0.2992
54
+ step 520000 | 4723s | train_loss=0.2889
55
+ step 530000 | 4807s | train_loss=0.2756
56
+ step 540000 | 4891s | train_loss=0.2624
57
+ step 550000 | 4974s | train_loss=0.2586
58
+ step 560000 | 5058s | train_loss=0.2470
59
+ step 570000 | 5140s | train_loss=0.2412
60
+ step 580000 | 5225s | train_loss=0.2338
61
+ step 590000 | 5309s | train_loss=0.2257
62
+ step 600000 | 5458s | train_loss=0.2222 | fid=85.3701
63
+ step 610000 | 5539s | train_loss=0.2179
64
+ step 620000 | 5623s | train_loss=0.2109
65
+ step 630000 | 5707s | train_loss=0.2057
66
+ step 640000 | 5791s | train_loss=0.2004
67
+ step 650000 | 5876s | train_loss=0.1957
68
+ step 660000 | 5958s | train_loss=0.1920
69
+ step 670000 | 6043s | train_loss=0.1904
70
+ step 680000 | 6127s | train_loss=0.1834
71
+ step 690000 | 6209s | train_loss=0.1817
72
+ step 700000 | 6359s | train_loss=0.1820 | fid=87.8508
73
+ step 710000 | 6441s | train_loss=0.1764
74
+ step 720000 | 6523s | train_loss=0.1719
75
+ step 730000 | 6607s | train_loss=0.1757
76
+ step 740000 | 6691s | train_loss=0.1692
77
+ step 750000 | 6776s | train_loss=0.1652
78
+ step 760000 | 6860s | train_loss=0.1597
79
+ step 770000 | 6943s | train_loss=0.1612
80
+ step 780000 | 7027s | train_loss=0.1558
81
+ step 790000 | 7111s | train_loss=0.1540
82
+ step 800000 | 7259s | train_loss=0.1545 | fid=86.4788
83
+ step 810000 | 7341s | train_loss=0.1551
84
+ step 820000 | 7424s | train_loss=0.1486
85
+ step 830000 | 7508s | train_loss=0.1473
86
+ step 840000 | 7592s | train_loss=0.1478
87
+ step 850000 | 7676s | train_loss=0.1445
88
+ step 860000 | 7760s | train_loss=0.1440
89
+ step 870000 | 7845s | train_loss=0.1418
90
+ step 880000 | 7927s | train_loss=0.1437
91
+ step 890000 | 8012s | train_loss=0.1398
92
+ step 900000 | 8162s | train_loss=0.1392 | fid=86.6292
93
+ step 910000 | 8242s | train_loss=0.1366
94
+ step 920000 | 8324s | train_loss=0.1328
95
+ step 930000 | 8408s | train_loss=0.1345
96
+ step 940000 | 8492s | train_loss=0.1336
97
+ step 950000 | 8577s | train_loss=0.1302
98
+ step 960000 | 8661s | train_loss=0.1299
99
+ step 970000 | 8745s | train_loss=0.1285
100
+ step 980000 | 8829s | train_loss=0.1241
101
+ step 990000 | 8912s | train_loss=0.1278
102
+ step 1000000 | 9061s | train_loss=0.1266 | fid=85.5726
103
+ step 1010000 | 9143s | train_loss=0.1229
104
+ step 1020000 | 9223s | train_loss=0.1237
105
+ step 1030000 | 9307s | train_loss=0.1237
106
+ step 1040000 | 9392s | train_loss=0.1217
107
+ step 1050000 | 9476s | train_loss=0.1192
108
+ step 1060000 | 9560s | train_loss=0.1235
109
+ step 1070000 | 9644s | train_loss=0.1166
110
+ step 1080000 | 9728s | train_loss=0.1169
111
+ step 1090000 | 9813s | train_loss=0.1140
112
+ step 1100000 | 9961s | train_loss=0.1161 | fid=87.9468
113
+ step 1110000 | 10043s | train_loss=0.1139
114
+ step 1120000 | 10125s | train_loss=0.1114
115
+ step 1130000 | 10207s | train_loss=0.1123
116
+ step 1140000 | 10291s | train_loss=0.1111
117
+ step 1150000 | 10376s | train_loss=0.1143
118
+ step 1160000 | 10460s | train_loss=0.1099
119
+ step 1170000 | 10544s | train_loss=0.1114
120
+ step 1180000 | 10628s | train_loss=0.1084
121
+ step 1190000 | 10712s | train_loss=0.1061
122
+ step 1200000 | 10860s | train_loss=0.1053 | fid=86.8908
123
+ step 1210000 | 10942s | train_loss=0.1050
124
+ step 1220000 | 11024s | train_loss=0.1074
125
+ step 1230000 | 11108s | train_loss=0.1040
126
+ step 1240000 | 11193s | train_loss=0.1035
127
+ step 1250000 | 11275s | train_loss=0.1038
128
+ step 1260000 | 11359s | train_loss=0.1027
129
+ step 1270000 | 11443s | train_loss=0.1028
130
+ step 1280000 | 11528s | train_loss=0.1016
131
+ step 1290000 | 11612s | train_loss=0.0991
132
+ step 1300000 | 11761s | train_loss=0.0989 | fid=88.1766
133
+ step 1310000 | 11842s | train_loss=0.0996
134
+ step 1320000 | 11925s | train_loss=0.1006
135
+ step 1330000 | 12009s | train_loss=0.0971
136
+ step 1340000 | 12093s | train_loss=0.0974
137
+ step 1350000 | 12177s | train_loss=0.0977
138
+ step 1360000 | 12259s | train_loss=0.0966
139
+ step 1370000 | 12344s | train_loss=0.0969
140
+ step 1380000 | 12428s | train_loss=0.0944
141
+ step 1390000 | 12512s | train_loss=0.0954
142
+ step 1400000 | 12662s | train_loss=0.0942 | fid=87.7817
143
+ step 1410000 | 12744s | train_loss=0.0928
144
+ step 1420000 | 12825s | train_loss=0.0946
145
+ step 1430000 | 12909s | train_loss=0.0932
146
+ step 1440000 | 12993s | train_loss=0.0924
147
+ step 1450000 | 13077s | train_loss=0.0905
148
+ step 1460000 | 13162s | train_loss=0.0903
149
+ step 1470000 | 13245s | train_loss=0.0899
150
+ step 1480000 | 13328s | train_loss=0.0905
151
+ step 1490000 | 13412s | train_loss=0.0905
152
+ step 1500000 | 13561s | train_loss=0.0896 | fid=87.7798
153
+ step 1510000 | 13643s | train_loss=0.0872
154
+ step 1520000 | 13728s | train_loss=0.0877
155
+ step 1530000 | 13809s | train_loss=0.0893
156
+ step 1540000 | 13893s | train_loss=0.0878
157
+ step 1550000 | 13977s | train_loss=0.0871
158
+ step 1560000 | 14061s | train_loss=0.0876
159
+ step 1570000 | 14145s | train_loss=0.0877
160
+ step 1580000 | 14230s | train_loss=0.0866
161
+ step 1590000 | 14312s | train_loss=0.0857
162
+ step 1600000 | 14462s | train_loss=0.0868 | fid=88.9997
163
+ step 1610000 | 14544s | train_loss=0.0845
164
+ step 1620000 | 14628s | train_loss=0.0846
165
+ step 1630000 | 14711s | train_loss=0.0838
166
+ step 1640000 | 14793s | train_loss=0.0847
167
+ step 1650000 | 14877s | train_loss=0.0838
168
+ step 1660000 | 14962s | train_loss=0.0823
169
+ step 1670000 | 15046s | train_loss=0.0830
170
+ step 1680000 | 15130s | train_loss=0.0816
171
+ step 1690000 | 15214s | train_loss=0.0834
172
+ step 1700000 | 15362s | train_loss=0.0827 | fid=88.3021
173
+ step 1710000 | 15444s | train_loss=0.0808
174
+ step 1720000 | 15529s | train_loss=0.0790
175
+ step 1730000 | 15611s | train_loss=0.0818
176
+ step 1740000 | 15696s | train_loss=0.0789
177
+ step 1750000 | 15778s | train_loss=0.0789
178
+ step 1760000 | 15862s | train_loss=0.0780
179
+ step 1770000 | 15946s | train_loss=0.0783
180
+ step 1780000 | 16030s | train_loss=0.0793
181
+ step 1790000 | 16115s | train_loss=0.0772
182
+ step 1800000 | 16265s | train_loss=0.0767 | fid=86.5601
183
+ step 1810000 | 16345s | train_loss=0.0787
184
+ step 1820000 | 16429s | train_loss=0.0769
185
+ step 1830000 | 16511s | train_loss=0.0770
186
+ step 1840000 | 16596s | train_loss=0.0769
187
+ step 1850000 | 16680s | train_loss=0.0754
188
+ step 1860000 | 16763s | train_loss=0.0769
189
+ step 1870000 | 16847s | train_loss=0.0748
190
+ step 1880000 | 16931s | train_loss=0.0754
191
+ step 1890000 | 17015s | train_loss=0.0743
192
+ step 1900000 | 17165s | train_loss=0.0745 | fid=86.9884
193
+ step 1910000 | 17247s | train_loss=0.0756
194
+ step 1920000 | 17330s | train_loss=0.0722
195
+ step 1930000 | 17412s | train_loss=0.0729
196
+ step 1940000 | 17496s | train_loss=0.0730
197
+ step 1950000 | 17580s | train_loss=0.0731
198
+ step 1960000 | 17664s | train_loss=0.0730
199
+ step 1970000 | 17748s | train_loss=0.0717
200
+ step 1980000 | 17832s | train_loss=0.0735
201
+ step 1990000 | 17916s | train_loss=0.0714
202
+ step 2000000 | 18066s | train_loss=0.0731 | fid=87.2536
203
+ step 2010000 | 18148s | train_loss=0.0699
204
+ step 2020000 | 18232s | train_loss=0.0710
205
+ step 2030000 | 18313s | train_loss=0.0702
206
+ step 2040000 | 18397s | train_loss=0.0716
207
+ step 2050000 | 18481s | train_loss=0.0700
208
+ step 2060000 | 18565s | train_loss=0.0707
209
+ step 2070000 | 18648s | train_loss=0.0703
210
+ step 2080000 | 18732s | train_loss=0.0701
211
+ step 2090000 | 18817s | train_loss=0.0695
212
+ step 2100000 | 18966s | train_loss=0.0695 | fid=87.1157
213
+ step 2110000 | 19048s | train_loss=0.0683
214
+ step 2120000 | 19133s | train_loss=0.0695
215
+ step 2130000 | 19215s | train_loss=0.0693
216
+ step 2140000 | 19300s | train_loss=0.0686
217
+ step 2150000 | 19382s | train_loss=0.0687
218
+ step 2160000 | 19467s | train_loss=0.0682
219
+ step 2170000 | 19551s | train_loss=0.0683
220
+ step 2180000 | 19634s | train_loss=0.0679
221
+ step 2190000 | 19718s | train_loss=0.0666
222
+ step 2200000 | 19867s | train_loss=0.0680 | fid=87.2321
223
+ step 2210000 | 19950s | train_loss=0.0674
224
+ step 2220000 | 20034s | train_loss=0.0669
225
+ step 2230000 | 20117s | train_loss=0.0665
226
+ step 2240000 | 20201s | train_loss=0.0654
227
+ step 2250000 | 20285s | train_loss=0.0665
228
+ step 2260000 | 20368s | train_loss=0.0655
229
+ step 2270000 | 20452s | train_loss=0.0657
230
+ step 2280000 | 20536s | train_loss=0.0651
231
+ step 2290000 | 20619s | train_loss=0.0673
232
+ step 2300000 | 20769s | train_loss=0.0639 | fid=87.4887
233
+ step 2310000 | 20851s | train_loss=0.0644
234
+ step 2320000 | 20935s | train_loss=0.0642
235
+ step 2330000 | 21019s | train_loss=0.0655
236
+ step 2340000 | 21102s | train_loss=0.0630
237
+ step 2350000 | 21186s | train_loss=0.0628
238
+ step 2360000 | 21270s | train_loss=0.0635
239
+ step 2370000 | 21353s | train_loss=0.0628
240
+ step 2380000 | 21437s | train_loss=0.0636
241
+ step 2390000 | 21522s | train_loss=0.0624
242
+ step 2400000 | 21670s | train_loss=0.0621 | fid=87.1673
243
+ step 2410000 | 21752s | train_loss=0.0636
244
+ step 2420000 | 21836s | train_loss=0.0614
245
+ step 2430000 | 21921s | train_loss=0.0621
246
+ step 2440000 | 22003s | train_loss=0.0632
247
+ step 2450000 | 22087s | train_loss=0.0618
248
+ step 2460000 | 22171s | train_loss=0.0620
249
+ step 2470000 | 22256s | train_loss=0.0608
250
+ step 2480000 | 22340s | train_loss=0.0634
251
+ step 2490000 | 22423s | train_loss=0.0622
252
+ step 2500000 | 22571s | train_loss=0.0610 | fid=86.0623
253
+ step 2510000 | 22653s | train_loss=0.0630
254
+ step 2520000 | 22737s | train_loss=0.0600
255
+ step 2530000 | 22821s | train_loss=0.0610
256
+ step 2540000 | 22903s | train_loss=0.0609
257
+ step 2550000 | 22988s | train_loss=0.0602
258
+ step 2560000 | 23072s | train_loss=0.0613
259
+ step 2570000 | 23156s | train_loss=0.0602
260
+ step 2580000 | 23240s | train_loss=0.0591
261
+ step 2590000 | 23325s | train_loss=0.0618
262
+ step 2600000 | 23472s | train_loss=0.0598 | fid=88.1269
263
+ step 2610000 | 23553s | train_loss=0.0601
264
+ step 2620000 | 23637s | train_loss=0.0585
265
+ step 2630000 | 23721s | train_loss=0.0598
266
+ step 2640000 | 23803s | train_loss=0.0603
267
+ step 2650000 | 23888s | train_loss=0.0601
268
+ step 2660000 | 23972s | train_loss=0.0597
269
+ step 2670000 | 24056s | train_loss=0.0594
270
+ step 2680000 | 24140s | train_loss=0.0598
271
+ step 2690000 | 24224s | train_loss=0.0587
272
+ step 2700000 | 24374s | train_loss=0.0570 | fid=87.8086
273
+ step 2710000 | 24454s | train_loss=0.0585
274
+ step 2720000 | 24537s | train_loss=0.0597
275
+ step 2730000 | 24621s | train_loss=0.0596
276
+ step 2740000 | 24703s | train_loss=0.0577
277
+ step 2750000 | 24787s | train_loss=0.0574
278
+ step 2760000 | 24871s | train_loss=0.0561
279
+ step 2770000 | 24956s | train_loss=0.0556
280
+ step 2780000 | 25040s | train_loss=0.0566
281
+ step 2790000 | 25124s | train_loss=0.0569
282
+ step 2800000 | 25273s | train_loss=0.0573 | fid=87.2051
283
+ step 2810000 | 25355s | train_loss=0.0568
284
+ step 2820000 | 25438s | train_loss=0.0566
285
+ step 2830000 | 25521s | train_loss=0.0556
286
+ step 2840000 | 25603s | train_loss=0.0563
287
+ step 2850000 | 25687s | train_loss=0.0564
288
+ step 2860000 | 25771s | train_loss=0.0560
289
+ step 2870000 | 25856s | train_loss=0.0550
290
+ step 2880000 | 25940s | train_loss=0.0553
291
+ step 2890000 | 26024s | train_loss=0.0566
292
+ step 2900000 | 26173s | train_loss=0.0559 | fid=87.2551
293
+ step 2910000 | 26255s | train_loss=0.0566
294
+ step 2920000 | 26339s | train_loss=0.0555
295
+ step 2930000 | 26422s | train_loss=0.0560
296
+ step 2940000 | 26503s | train_loss=0.0558
297
+ step 2950000 | 26587s | train_loss=0.0549
298
+ step 2960000 | 26671s | train_loss=0.0550
299
+ step 2970000 | 26755s | train_loss=0.0545
300
+ step 2980000 | 26840s | train_loss=0.0556
301
+ step 2990000 | 26924s | train_loss=0.0551
302
+ step 3000000 | 27074s | train_loss=0.0543 | fid=86.8879
303
+ step 3010000 | 27156s | train_loss=0.0541
304
+ step 3020000 | 27240s | train_loss=0.0549
305
+ step 3030000 | 27324s | train_loss=0.0542
306
+ step 3040000 | 27409s | train_loss=0.0540
307
+ step 3050000 | 27488s | train_loss=0.0546
308
+ step 3060000 | 27572s | train_loss=0.0533
309
+ step 3070000 | 27656s | train_loss=0.0519
310
+ step 3080000 | 27740s | train_loss=0.0527
311
+ step 3090000 | 27824s | train_loss=0.0542
312
+ step 3100000 | 27974s | train_loss=0.0525 | fid=86.5691
313
+ step 3110000 | 28056s | train_loss=0.0532
314
+ step 3120000 | 28141s | train_loss=0.0535
315
+ step 3130000 | 28225s | train_loss=0.0541
316
+ step 3140000 | 28309s | train_loss=0.0524
317
+ step 3150000 | 28392s | train_loss=0.0514
318
+ step 3160000 | 28471s | train_loss=0.0527
319
+ step 3170000 | 28556s | train_loss=0.0528
320
+ step 3180000 | 28640s | train_loss=0.0523
321
+ step 3190000 | 28724s | train_loss=0.0530
322
+ step 3200000 | 28874s | train_loss=0.0530 | fid=87.2951
323
+ step 3210000 | 28956s | train_loss=0.0530
324
+ step 3220000 | 29040s | train_loss=0.0521
325
+ step 3230000 | 29124s | train_loss=0.0520
326
+ step 3240000 | 29208s | train_loss=0.0513
327
+ step 3250000 | 29291s | train_loss=0.0513
328
+ step 3260000 | 29375s | train_loss=0.0520
329
+ step 3270000 | 29457s | train_loss=0.0525
330
+ step 3280000 | 29541s | train_loss=0.0509
331
+ step 3290000 | 29625s | train_loss=0.0507
332
+ step 3300000 | 29775s | train_loss=0.0525 | fid=87.0834
333
+ step 3310000 | 29857s | train_loss=0.0512
334
+ step 3320000 | 29941s | train_loss=0.0517
335
+ step 3330000 | 30025s | train_loss=0.0513
336
+ step 3340000 | 30110s | train_loss=0.0515
337
+ step 3350000 | 30192s | train_loss=0.0514
338
+ step 3360000 | 30276s | train_loss=0.0513
339
+ step 3370000 | 30361s | train_loss=0.0498
340
+ step 3380000 | 30444s | train_loss=0.0509
341
+ step 3390000 | 30526s | train_loss=0.0504
342
+ step 3400000 | 30676s | train_loss=0.0511 | fid=87.4795
343
+ step 3410000 | 30758s | train_loss=0.0501
344
+ step 3420000 | 30842s | train_loss=0.0503
345
+ step 3430000 | 30926s | train_loss=0.0504
346
+ step 3440000 | 31011s | train_loss=0.0506
347
+ step 3450000 | 31093s | train_loss=0.0491
348
+ step 3460000 | 31177s | train_loss=0.0489
349
+ step 3470000 | 31261s | train_loss=0.0498
350
+ step 3480000 | 31346s | train_loss=0.0509
351
+ step 3490000 | 31429s | train_loss=0.0505
352
+ step 3500000 | 31577s | train_loss=0.0510 | fid=87.1327
353
+ step 3510000 | 31659s | train_loss=0.0495
354
+ step 3520000 | 31743s | train_loss=0.0495
355
+ step 3530000 | 31828s | train_loss=0.0497
356
+ step 3540000 | 31912s | train_loss=0.0498
357
+ step 3550000 | 31994s | train_loss=0.0503
358
+ step 3560000 | 32078s | train_loss=0.0480
359
+ step 3570000 | 32162s | train_loss=0.0487
360
+ step 3580000 | 32246s | train_loss=0.0485
361
+ step 3590000 | 32331s | train_loss=0.0489
362
+ step 3600000 | 32478s | train_loss=0.0487 | fid=87.2199
363
+ step 3610000 | 32559s | train_loss=0.0494
364
+ step 3620000 | 32643s | train_loss=0.0484
365
+ step 3630000 | 32728s | train_loss=0.0479
366
+ step 3640000 | 32812s | train_loss=0.0486
367
+ step 3650000 | 32894s | train_loss=0.0486
368
+ step 3660000 | 32978s | train_loss=0.0481
369
+ step 3670000 | 33062s | train_loss=0.0482
370
+ step 3680000 | 33146s | train_loss=0.0487
371
+ step 3690000 | 33231s | train_loss=0.0470
372
+ step 3700000 | 33378s | train_loss=0.0486 | fid=86.7139
373
+ step 3710000 | 33460s | train_loss=0.0488
374
+ step 3720000 | 33543s | train_loss=0.0476
375
+ step 3730000 | 33627s | train_loss=0.0484
376
+ step 3740000 | 33711s | train_loss=0.0477
377
+ step 3750000 | 33794s | train_loss=0.0491
378
+ step 3760000 | 33878s | train_loss=0.0476
379
+ step 3770000 | 33962s | train_loss=0.0474
380
+ step 3780000 | 34046s | train_loss=0.0477
381
+ step 3790000 | 34130s | train_loss=0.0479
382
+ step 3800000 | 34279s | train_loss=0.0465 | fid=87.7611
383
+ step 3810000 | 34360s | train_loss=0.0481
384
+ step 3820000 | 34444s | train_loss=0.0454
385
+ step 3830000 | 34527s | train_loss=0.0463
386
+ step 3840000 | 34611s | train_loss=0.0466
387
+ step 3850000 | 34695s | train_loss=0.0479
388
+ step 3860000 | 34777s | train_loss=0.0465
389
+ step 3870000 | 34862s | train_loss=0.0478
390
+ step 3880000 | 34946s | train_loss=0.0468
391
+ step 3890000 | 35030s | train_loss=0.0462
392
+ step 3900000 | 35180s | train_loss=0.0455 | fid=87.0507
393
+ step 3910000 | 35262s | train_loss=0.0462
394
+ step 3920000 | 35345s | train_loss=0.0471
395
+ step 3930000 | 35429s | train_loss=0.0470
396
+ step 3940000 | 35514s | train_loss=0.0460
397
+ step 3950000 | 35596s | train_loss=0.0460
398
+ step 3960000 | 35678s | train_loss=0.0457
399
+ step 3970000 | 35762s | train_loss=0.0466
400
+ step 3980000 | 35846s | train_loss=0.0461
401
+ step 3990000 | 35930s | train_loss=0.0466
402
+ step 4000000 | 36083s | train_loss=0.0466 | fid=86.5076
403
+ step 4010000 | 36164s | train_loss=0.0460
404
+ step 4020000 | 36249s | train_loss=0.0467
405
+ step 4030000 | 36331s | train_loss=0.0456
406
+ step 4040000 | 36415s | train_loss=0.0470
407
+ step 4050000 | 36500s | train_loss=0.0454
408
+ step 4060000 | 36578s | train_loss=0.0461
409
+ step 4070000 | 36663s | train_loss=0.0448
410
+ step 4080000 | 36747s | train_loss=0.0448
411
+ step 4090000 | 36831s | train_loss=0.0451
412
+ step 4100000 | 36984s | train_loss=0.0448 | fid=86.4818
413
+ step 4110000 | 37065s | train_loss=0.0452
414
+ step 4120000 | 37149s | train_loss=0.0453
415
+ step 4130000 | 37234s | train_loss=0.0453
416
+ step 4140000 | 37316s | train_loss=0.0452
417
+ step 4150000 | 37400s | train_loss=0.0449
418
+ step 4160000 | 37482s | train_loss=0.0453
419
+ step 4170000 | 37565s | train_loss=0.0449
420
+ step 4180000 | 37649s | train_loss=0.0442
421
+ step 4190000 | 37733s | train_loss=0.0454
422
+ step 4200000 | 37886s | train_loss=0.0438 | fid=86.7038
423
+ step 4210000 | 37968s | train_loss=0.0445
424
+ step 4220000 | 38052s | train_loss=0.0450
425
+ step 4230000 | 38136s | train_loss=0.0438
426
+ step 4240000 | 38219s | train_loss=0.0438
427
+ step 4250000 | 38303s | train_loss=0.0450
428
+ step 4260000 | 38385s | train_loss=0.0442
429
+ step 4270000 | 38469s | train_loss=0.0438
430
+ step 4280000 | 38553s | train_loss=0.0442
431
+ step 4290000 | 38636s | train_loss=0.0430
432
+ step 4300000 | 38789s | train_loss=0.0433 | fid=87.9144
433
+ step 4310000 | 38871s | train_loss=0.0442
434
+ step 4320000 | 38955s | train_loss=0.0446
435
+ step 4330000 | 39039s | train_loss=0.0438
436
+ step 4340000 | 39124s | train_loss=0.0441
437
+ step 4350000 | 39206s | train_loss=0.0437
438
+ step 4360000 | 39287s | train_loss=0.0446
439
+ step 4370000 | 39371s | train_loss=0.0441
440
+ step 4380000 | 39456s | train_loss=0.0438
441
+ step 4390000 | 39540s | train_loss=0.0435
442
+ step 4400000 | 39691s | train_loss=0.0434 | fid=86.4724
443
+ step 4410000 | 39773s | train_loss=0.0434
444
+ step 4420000 | 39857s | train_loss=0.0440
445
+ step 4430000 | 39941s | train_loss=0.0427
446
+ step 4440000 | 40025s | train_loss=0.0437
447
+ step 4450000 | 40110s | train_loss=0.0424
448
+ step 4460000 | 40189s | train_loss=0.0429
449
+ step 4470000 | 40273s | train_loss=0.0432
450
+ step 4480000 | 40358s | train_loss=0.0424
451
+ step 4490000 | 40442s | train_loss=0.0420
452
+ step 4500000 | 40591s | train_loss=0.0428 | fid=87.8980
453
+ step 4510000 | 40673s | train_loss=0.0423
454
+ step 4520000 | 40757s | train_loss=0.0427
455
+ step 4530000 | 40841s | train_loss=0.0427
456
+ step 4540000 | 40925s | train_loss=0.0426
457
+ step 4550000 | 41009s | train_loss=0.0425
458
+ step 4560000 | 41093s | train_loss=0.0419
459
+ step 4570000 | 41174s | train_loss=0.0412
460
+ step 4580000 | 41258s | train_loss=0.0420
461
+ step 4590000 | 41342s | train_loss=0.0428
462
+ step 4600000 | 41495s | train_loss=0.0431 | fid=88.2130
463
+ step 4610000 | 41575s | train_loss=0.0418
464
+ step 4620000 | 41659s | train_loss=0.0419
465
+ step 4630000 | 41743s | train_loss=0.0414
466
+ step 4640000 | 41827s | train_loss=0.0412
467
+ step 4650000 | 41912s | train_loss=0.0412
468
+ step 4660000 | 41996s | train_loss=0.0415
469
+ step 4670000 | 42078s | train_loss=0.0415
470
+ step 4680000 | 42160s | train_loss=0.0418
471
+ step 4690000 | 42245s | train_loss=0.0413
472
+ step 4700000 | 42398s | train_loss=0.0408 | fid=87.2102
473
+ step 4710000 | 42479s | train_loss=0.0419
474
+ step 4720000 | 42564s | train_loss=0.0416
475
+ step 4730000 | 42646s | train_loss=0.0418
476
+ step 4740000 | 42730s | train_loss=0.0412
477
+ step 4750000 | 42815s | train_loss=0.0408
478
+ step 4760000 | 42899s | train_loss=0.0415
479
+ step 4770000 | 42981s | train_loss=0.0419
480
+ step 4780000 | 43065s | train_loss=0.0419
481
+ step 4790000 | 43148s | train_loss=0.0417
482
+ step 4800000 | 43299s | train_loss=0.0417 | fid=87.7933
483
+ step 4810000 | 43381s | train_loss=0.0426
484
+ step 4820000 | 43465s | train_loss=0.0416
485
+ step 4830000 | 43549s | train_loss=0.0411
486
+ step 4840000 | 43632s | train_loss=0.0408
487
+ step 4850000 | 43716s | train_loss=0.0409
488
+ step 4860000 | 43800s | train_loss=0.0398
489
+ step 4870000 | 43882s | train_loss=0.0409
490
+ step 4880000 | 43966s | train_loss=0.0413
491
+ step 4890000 | 44051s | train_loss=0.0407
492
+ step 4900000 | 44199s | train_loss=0.0409 | fid=86.0890
493
+ step 4910000 | 44281s | train_loss=0.0403
494
+ step 4920000 | 44365s | train_loss=0.0408
495
+ step 4930000 | 44449s | train_loss=0.0400
496
+ step 4940000 | 44534s | train_loss=0.0402
497
+ step 4950000 | 44616s | train_loss=0.0412
498
+ step 4960000 | 44701s | train_loss=0.0406
499
+ step 4970000 | 44783s | train_loss=0.0413
500
+ step 4980000 | 44867s | train_loss=0.0405
501
+ step 4990000 | 44951s | train_loss=0.0405
502
+ step 5000000 | 45100s | train_loss=0.0399 | fid=88.2123
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d100/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 44,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d100/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 70s | train_loss=0.9957 | fid=287.1331
3
+ step 10000 | 154s | train_loss=0.8020
4
+ step 20000 | 243s | train_loss=0.6792
5
+ step 30000 | 331s | train_loss=0.5434
6
+ step 40000 | 420s | train_loss=0.4046
7
+ step 50000 | 509s | train_loss=0.2932
8
+ step 60000 | 598s | train_loss=0.2222
9
+ step 70000 | 687s | train_loss=0.1812
10
+ step 80000 | 775s | train_loss=0.1555
11
+ step 90000 | 864s | train_loss=0.1400
12
+ step 100000 | 1021s | train_loss=0.1296 | fid=61.5573
13
+ step 110000 | 1098s | train_loss=0.1250
14
+ step 120000 | 1187s | train_loss=0.1172
15
+ step 130000 | 1275s | train_loss=0.1096
16
+ step 140000 | 1364s | train_loss=0.1065
17
+ step 150000 | 1453s | train_loss=0.1030
18
+ step 160000 | 1541s | train_loss=0.0968
19
+ step 170000 | 1630s | train_loss=0.0936
20
+ step 180000 | 1719s | train_loss=0.0884
21
+ step 190000 | 1808s | train_loss=0.0878
22
+ step 200000 | 1963s | train_loss=0.0866 | fid=59.8282
23
+ step 210000 | 2053s | train_loss=0.0816
24
+ step 220000 | 2132s | train_loss=0.0802
25
+ step 230000 | 2221s | train_loss=0.0779
26
+ step 240000 | 2310s | train_loss=0.0762
27
+ step 250000 | 2398s | train_loss=0.0733
28
+ step 260000 | 2487s | train_loss=0.0733
29
+ step 270000 | 2576s | train_loss=0.0720
30
+ step 280000 | 2665s | train_loss=0.0717
31
+ step 290000 | 2753s | train_loss=0.0687
32
+ step 300000 | 2910s | train_loss=0.0667 | fid=56.4418
33
+ step 310000 | 2998s | train_loss=0.0645
34
+ step 320000 | 3087s | train_loss=0.0649
35
+ step 330000 | 3170s | train_loss=0.0639
36
+ step 340000 | 3258s | train_loss=0.0627
37
+ step 350000 | 3347s | train_loss=0.0623
38
+ step 360000 | 3436s | train_loss=0.0603
39
+ step 370000 | 3525s | train_loss=0.0606
40
+ step 380000 | 3613s | train_loss=0.0581
41
+ step 390000 | 3702s | train_loss=0.0593
42
+ step 400000 | 3858s | train_loss=0.0590 | fid=56.1448
43
+ step 410000 | 3947s | train_loss=0.0576
44
+ step 420000 | 4036s | train_loss=0.0572
45
+ step 430000 | 4118s | train_loss=0.0572
46
+ step 440000 | 4207s | train_loss=0.0557
47
+ step 450000 | 4295s | train_loss=0.0567
48
+ step 460000 | 4384s | train_loss=0.0543
49
+ step 470000 | 4473s | train_loss=0.0550
50
+ step 480000 | 4562s | train_loss=0.0531
51
+ step 490000 | 4650s | train_loss=0.0532
52
+ step 500000 | 4807s | train_loss=0.0509 | fid=52.7436
53
+ step 510000 | 4896s | train_loss=0.0518
54
+ step 520000 | 4985s | train_loss=0.0514
55
+ step 530000 | 5072s | train_loss=0.0503
56
+ step 540000 | 5155s | train_loss=0.0492
57
+ step 550000 | 5244s | train_loss=0.0498
58
+ step 560000 | 5333s | train_loss=0.0495
59
+ step 570000 | 5422s | train_loss=0.0502
60
+ step 580000 | 5510s | train_loss=0.0494
61
+ step 590000 | 5599s | train_loss=0.0479
62
+ step 600000 | 5756s | train_loss=0.0487 | fid=54.3750
63
+ step 610000 | 5845s | train_loss=0.0478
64
+ step 620000 | 5934s | train_loss=0.0487
65
+ step 630000 | 6023s | train_loss=0.0471
66
+ step 640000 | 6109s | train_loss=0.0457
67
+ step 650000 | 6193s | train_loss=0.0462
68
+ step 660000 | 6282s | train_loss=0.0450
69
+ step 670000 | 6371s | train_loss=0.0463
70
+ step 680000 | 6460s | train_loss=0.0449
71
+ step 690000 | 6548s | train_loss=0.0443
72
+ step 700000 | 6705s | train_loss=0.0436 | fid=52.7829
73
+ step 710000 | 6794s | train_loss=0.0443
74
+ step 720000 | 6883s | train_loss=0.0435
75
+ step 730000 | 6971s | train_loss=0.0434
76
+ step 740000 | 7060s | train_loss=0.0436
77
+ step 750000 | 7142s | train_loss=0.0427
78
+ step 760000 | 7231s | train_loss=0.0411
79
+ step 770000 | 7320s | train_loss=0.0422
80
+ step 780000 | 7409s | train_loss=0.0422
81
+ step 790000 | 7497s | train_loss=0.0414
82
+ step 800000 | 7654s | train_loss=0.0418 | fid=52.3147
83
+ step 810000 | 7742s | train_loss=0.0414
84
+ step 820000 | 7831s | train_loss=0.0412
85
+ step 830000 | 7920s | train_loss=0.0404
86
+ step 840000 | 8009s | train_loss=0.0399
87
+ step 850000 | 8097s | train_loss=0.0398
88
+ step 860000 | 8180s | train_loss=0.0405
89
+ step 870000 | 8269s | train_loss=0.0418
90
+ step 880000 | 8358s | train_loss=0.0395
91
+ step 890000 | 8447s | train_loss=0.0395
92
+ step 900000 | 8604s | train_loss=0.0390 | fid=51.1110
93
+ step 910000 | 8692s | train_loss=0.0398
94
+ step 920000 | 8781s | train_loss=0.0383
95
+ step 930000 | 8870s | train_loss=0.0378
96
+ step 940000 | 8959s | train_loss=0.0380
97
+ step 950000 | 9048s | train_loss=0.0373
98
+ step 960000 | 9135s | train_loss=0.0374
99
+ step 970000 | 9219s | train_loss=0.0385
100
+ step 980000 | 9308s | train_loss=0.0363
101
+ step 990000 | 9396s | train_loss=0.0373
102
+ step 1000000 | 9554s | train_loss=0.0365 | fid=51.7354
103
+ step 1010000 | 9642s | train_loss=0.0363
104
+ step 1020000 | 9731s | train_loss=0.0381
105
+ step 1030000 | 9820s | train_loss=0.0363
106
+ step 1040000 | 9909s | train_loss=0.0378
107
+ step 1050000 | 9998s | train_loss=0.0386
108
+ step 1060000 | 10086s | train_loss=0.0359
109
+ step 1070000 | 10173s | train_loss=0.0350
110
+ step 1080000 | 10258s | train_loss=0.0345
111
+ step 1090000 | 10346s | train_loss=0.0359
112
+ step 1100000 | 10504s | train_loss=0.0362 | fid=51.7029
113
+ step 1110000 | 10592s | train_loss=0.0347
114
+ step 1120000 | 10681s | train_loss=0.0350
115
+ step 1130000 | 10770s | train_loss=0.0365
116
+ step 1140000 | 10859s | train_loss=0.0359
117
+ step 1150000 | 10948s | train_loss=0.0339
118
+ step 1160000 | 11036s | train_loss=0.0347
119
+ step 1170000 | 11123s | train_loss=0.0351
120
+ step 1180000 | 11208s | train_loss=0.0340
121
+ step 1190000 | 11296s | train_loss=0.0332
122
+ step 1200000 | 11453s | train_loss=0.0343 | fid=52.7815
123
+ step 1210000 | 11542s | train_loss=0.0331
124
+ step 1220000 | 11631s | train_loss=0.0333
125
+ step 1230000 | 11719s | train_loss=0.0336
126
+ step 1240000 | 11808s | train_loss=0.0342
127
+ step 1250000 | 11897s | train_loss=0.0332
128
+ step 1260000 | 11985s | train_loss=0.0339
129
+ step 1270000 | 12075s | train_loss=0.0366
130
+ step 1280000 | 12161s | train_loss=0.0332
131
+ step 1290000 | 12245s | train_loss=0.0339
132
+ step 1300000 | 12402s | train_loss=0.0320 | fid=54.0813
133
+ step 1310000 | 12491s | train_loss=0.0311
134
+ step 1320000 | 12579s | train_loss=0.0318
135
+ step 1330000 | 12668s | train_loss=0.0319
136
+ step 1340000 | 12757s | train_loss=0.0322
137
+ step 1350000 | 12846s | train_loss=0.0314
138
+ step 1360000 | 12934s | train_loss=0.0318
139
+ step 1370000 | 13023s | train_loss=0.0322
140
+ step 1380000 | 13112s | train_loss=0.0312
141
+ step 1390000 | 13200s | train_loss=0.0316
142
+ step 1400000 | 13351s | train_loss=0.0312 | fid=55.0241
143
+ step 1410000 | 13440s | train_loss=0.0306
144
+ step 1420000 | 13529s | train_loss=0.0306
145
+ step 1430000 | 13617s | train_loss=0.0310
146
+ step 1440000 | 13706s | train_loss=0.0324
147
+ step 1450000 | 13795s | train_loss=0.0306
148
+ step 1460000 | 13883s | train_loss=0.0304
149
+ step 1470000 | 13972s | train_loss=0.0320
150
+ step 1480000 | 14061s | train_loss=0.0310
151
+ step 1490000 | 14147s | train_loss=0.0317
152
+ step 1500000 | 14301s | train_loss=0.0304 | fid=55.8929
153
+ step 1510000 | 14389s | train_loss=0.0310
154
+ step 1520000 | 14478s | train_loss=0.0305
155
+ step 1530000 | 14566s | train_loss=0.0305
156
+ step 1540000 | 14655s | train_loss=0.0291
157
+ step 1550000 | 14744s | train_loss=0.0280
158
+ step 1560000 | 14833s | train_loss=0.0301
159
+ step 1570000 | 14922s | train_loss=0.0302
160
+ step 1580000 | 15010s | train_loss=0.0293
161
+ step 1590000 | 15100s | train_loss=0.0295
162
+ step 1600000 | 15250s | train_loss=0.0283 | fid=56.6696
163
+ step 1610000 | 15338s | train_loss=0.0300
164
+ step 1620000 | 15427s | train_loss=0.0294
165
+ step 1630000 | 15516s | train_loss=0.0280
166
+ step 1640000 | 15604s | train_loss=0.0291
167
+ step 1650000 | 15693s | train_loss=0.0290
168
+ step 1660000 | 15782s | train_loss=0.0289
169
+ step 1670000 | 15871s | train_loss=0.0278
170
+ step 1680000 | 15959s | train_loss=0.0284
171
+ step 1690000 | 16048s | train_loss=0.0310
172
+ step 1700000 | 16203s | train_loss=0.0275 | fid=55.8361
173
+ step 1710000 | 16288s | train_loss=0.0278
174
+ step 1720000 | 16377s | train_loss=0.0277
175
+ step 1730000 | 16466s | train_loss=0.0281
176
+ step 1740000 | 16554s | train_loss=0.0274
177
+ step 1750000 | 16643s | train_loss=0.0277
178
+ step 1760000 | 16732s | train_loss=0.0283
179
+ step 1770000 | 16821s | train_loss=0.0289
180
+ step 1780000 | 16910s | train_loss=0.0286
181
+ step 1790000 | 16998s | train_loss=0.0274
182
+ step 1800000 | 17154s | train_loss=0.0267 | fid=52.6903
183
+ step 1810000 | 17244s | train_loss=0.0266
184
+ step 1820000 | 17329s | train_loss=0.0269
185
+ step 1830000 | 17418s | train_loss=0.0264
186
+ step 1840000 | 17506s | train_loss=0.0262
187
+ step 1850000 | 17595s | train_loss=0.0266
188
+ step 1860000 | 17684s | train_loss=0.0261
189
+ step 1870000 | 17773s | train_loss=0.0273
190
+ step 1880000 | 17862s | train_loss=0.0263
191
+ step 1890000 | 17950s | train_loss=0.0265
192
+ step 1900000 | 18108s | train_loss=0.0255 | fid=52.0008
193
+ step 1910000 | 18195s | train_loss=0.0263
194
+ step 1920000 | 18281s | train_loss=0.0253
195
+ step 1930000 | 18369s | train_loss=0.0257
196
+ step 1940000 | 18458s | train_loss=0.0250
197
+ step 1950000 | 18546s | train_loss=0.0262
198
+ step 1960000 | 18635s | train_loss=0.0245
199
+ step 1970000 | 18724s | train_loss=0.0248
200
+ step 1980000 | 18813s | train_loss=0.0251
201
+ step 1990000 | 18901s | train_loss=0.0256
202
+ step 2000000 | 19059s | train_loss=0.0237 | fid=49.2406
203
+ step 2010000 | 19147s | train_loss=0.0259
204
+ step 2020000 | 19235s | train_loss=0.0252
205
+ step 2030000 | 19320s | train_loss=0.0248
206
+ step 2040000 | 19409s | train_loss=0.0245
207
+ step 2050000 | 19497s | train_loss=0.0247
208
+ step 2060000 | 19586s | train_loss=0.0242
209
+ step 2070000 | 19675s | train_loss=0.0248
210
+ step 2080000 | 19764s | train_loss=0.0241
211
+ step 2090000 | 19853s | train_loss=0.0261
212
+ step 2100000 | 20009s | train_loss=0.0237 | fid=48.5596
213
+ step 2110000 | 20098s | train_loss=0.0226
214
+ step 2120000 | 20184s | train_loss=0.0224
215
+ step 2130000 | 20274s | train_loss=0.0232
216
+ step 2140000 | 20358s | train_loss=0.0245
217
+ step 2150000 | 20447s | train_loss=0.0240
218
+ step 2160000 | 20536s | train_loss=0.0252
219
+ step 2170000 | 20625s | train_loss=0.0236
220
+ step 2180000 | 20713s | train_loss=0.0240
221
+ step 2190000 | 20802s | train_loss=0.0223
222
+ step 2200000 | 20958s | train_loss=0.0234 | fid=47.8792
223
+ step 2210000 | 21047s | train_loss=0.0232
224
+ step 2220000 | 21135s | train_loss=0.0223
225
+ step 2230000 | 21222s | train_loss=0.0230
226
+ step 2240000 | 21309s | train_loss=0.0242
227
+ step 2250000 | 21396s | train_loss=0.0224
228
+ step 2260000 | 21485s | train_loss=0.0234
229
+ step 2270000 | 21574s | train_loss=0.0226
230
+ step 2280000 | 21663s | train_loss=0.0224
231
+ step 2290000 | 21751s | train_loss=0.0223
232
+ step 2300000 | 21908s | train_loss=0.0220 | fid=46.2454
233
+ step 2310000 | 21997s | train_loss=0.0241
234
+ step 2320000 | 22086s | train_loss=0.0220
235
+ step 2330000 | 22173s | train_loss=0.0221
236
+ step 2340000 | 22261s | train_loss=0.0213
237
+ step 2350000 | 22347s | train_loss=0.0223
238
+ step 2360000 | 22435s | train_loss=0.0220
239
+ step 2370000 | 22524s | train_loss=0.0215
240
+ step 2380000 | 22613s | train_loss=0.0223
241
+ step 2390000 | 22701s | train_loss=0.0217
242
+ step 2400000 | 22858s | train_loss=0.0211 | fid=45.5099
243
+ step 2410000 | 22947s | train_loss=0.0223
244
+ step 2420000 | 23036s | train_loss=0.0227
245
+ step 2430000 | 23125s | train_loss=0.0223
246
+ step 2440000 | 23211s | train_loss=0.0213
247
+ step 2450000 | 23300s | train_loss=0.0214
248
+ step 2460000 | 23386s | train_loss=0.0227
249
+ step 2470000 | 23474s | train_loss=0.0213
250
+ step 2480000 | 23563s | train_loss=0.0212
251
+ step 2490000 | 23651s | train_loss=0.0211
252
+ step 2500000 | 23809s | train_loss=0.0212 | fid=45.5290
253
+ step 2510000 | 23897s | train_loss=0.0214
254
+ step 2520000 | 23986s | train_loss=0.0209
255
+ step 2530000 | 24075s | train_loss=0.0215
256
+ step 2540000 | 24163s | train_loss=0.0204
257
+ step 2550000 | 24250s | train_loss=0.0203
258
+ step 2560000 | 24339s | train_loss=0.0219
259
+ step 2570000 | 24425s | train_loss=0.0206
260
+ step 2580000 | 24513s | train_loss=0.0199
261
+ step 2590000 | 24602s | train_loss=0.0218
262
+ step 2600000 | 24759s | train_loss=0.0202 | fid=45.3749
263
+ step 2610000 | 24848s | train_loss=0.0207
264
+ step 2620000 | 24937s | train_loss=0.0210
265
+ step 2630000 | 25025s | train_loss=0.0201
266
+ step 2640000 | 25115s | train_loss=0.0203
267
+ step 2650000 | 25201s | train_loss=0.0199
268
+ step 2660000 | 25289s | train_loss=0.0204
269
+ step 2670000 | 25375s | train_loss=0.0197
270
+ step 2680000 | 25464s | train_loss=0.0200
271
+ step 2690000 | 25552s | train_loss=0.0209
272
+ step 2700000 | 25709s | train_loss=0.0196 | fid=42.9271
273
+ step 2710000 | 25798s | train_loss=0.0199
274
+ step 2720000 | 25887s | train_loss=0.0204
275
+ step 2730000 | 25975s | train_loss=0.0198
276
+ step 2740000 | 26064s | train_loss=0.0196
277
+ step 2750000 | 26153s | train_loss=0.0195
278
+ step 2760000 | 26239s | train_loss=0.0202
279
+ step 2770000 | 26328s | train_loss=0.0202
280
+ step 2780000 | 26414s | train_loss=0.0203
281
+ step 2790000 | 26503s | train_loss=0.0196
282
+ step 2800000 | 26660s | train_loss=0.0202 | fid=43.3720
283
+ step 2810000 | 26749s | train_loss=0.0194
284
+ step 2820000 | 26837s | train_loss=0.0195
285
+ step 2830000 | 26926s | train_loss=0.0199
286
+ step 2840000 | 27015s | train_loss=0.0197
287
+ step 2850000 | 27104s | train_loss=0.0189
288
+ step 2860000 | 27192s | train_loss=0.0198
289
+ step 2870000 | 27279s | train_loss=0.0188
290
+ step 2880000 | 27368s | train_loss=0.0189
291
+ step 2890000 | 27454s | train_loss=0.0196
292
+ step 2900000 | 27610s | train_loss=0.0196 | fid=42.8767
293
+ step 2910000 | 27699s | train_loss=0.0184
294
+ step 2920000 | 27787s | train_loss=0.0186
295
+ step 2930000 | 27876s | train_loss=0.0187
296
+ step 2940000 | 27965s | train_loss=0.0188
297
+ step 2950000 | 28053s | train_loss=0.0189
298
+ step 2960000 | 28143s | train_loss=0.0198
299
+ step 2970000 | 28228s | train_loss=0.0186
300
+ step 2980000 | 28317s | train_loss=0.0189
301
+ step 2990000 | 28403s | train_loss=0.0187
302
+ step 3000000 | 28559s | train_loss=0.0187 | fid=42.6937
303
+ step 3010000 | 28648s | train_loss=0.0184
304
+ step 3020000 | 28737s | train_loss=0.0184
305
+ step 3030000 | 28826s | train_loss=0.0188
306
+ step 3040000 | 28914s | train_loss=0.0179
307
+ step 3050000 | 29003s | train_loss=0.0178
308
+ step 3060000 | 29092s | train_loss=0.0193
309
+ step 3070000 | 29180s | train_loss=0.0185
310
+ step 3080000 | 29267s | train_loss=0.0181
311
+ step 3090000 | 29356s | train_loss=0.0177
312
+ step 3100000 | 29510s | train_loss=0.0199 | fid=43.1121
313
+ step 3110000 | 29598s | train_loss=0.0180
314
+ step 3120000 | 29687s | train_loss=0.0188
315
+ step 3130000 | 29776s | train_loss=0.0189
316
+ step 3140000 | 29864s | train_loss=0.0187
317
+ step 3150000 | 29953s | train_loss=0.0176
318
+ step 3160000 | 30042s | train_loss=0.0183
319
+ step 3170000 | 30131s | train_loss=0.0186
320
+ step 3180000 | 30217s | train_loss=0.0173
321
+ step 3190000 | 30305s | train_loss=0.0179
322
+ step 3200000 | 30459s | train_loss=0.0187 | fid=43.8235
323
+ step 3210000 | 30548s | train_loss=0.0194
324
+ step 3220000 | 30636s | train_loss=0.0185
325
+ step 3230000 | 30725s | train_loss=0.0182
326
+ step 3240000 | 30814s | train_loss=0.0178
327
+ step 3250000 | 30903s | train_loss=0.0174
328
+ step 3260000 | 30991s | train_loss=0.0178
329
+ step 3270000 | 31080s | train_loss=0.0185
330
+ step 3280000 | 31169s | train_loss=0.0180
331
+ step 3290000 | 31255s | train_loss=0.0178
332
+ step 3300000 | 31413s | train_loss=0.0169 | fid=41.8306
333
+ step 3310000 | 31499s | train_loss=0.0174
334
+ step 3320000 | 31588s | train_loss=0.0178
335
+ step 3330000 | 31677s | train_loss=0.0181
336
+ step 3340000 | 31766s | train_loss=0.0189
337
+ step 3350000 | 31854s | train_loss=0.0178
338
+ step 3360000 | 31943s | train_loss=0.0176
339
+ step 3370000 | 32031s | train_loss=0.0181
340
+ step 3380000 | 32120s | train_loss=0.0169
341
+ step 3390000 | 32208s | train_loss=0.0172
342
+ step 3400000 | 32363s | train_loss=0.0171 | fid=41.0435
343
+ step 3410000 | 32449s | train_loss=0.0183
344
+ step 3420000 | 32538s | train_loss=0.0188
345
+ step 3430000 | 32627s | train_loss=0.0172
346
+ step 3440000 | 32716s | train_loss=0.0176
347
+ step 3450000 | 32804s | train_loss=0.0163
348
+ step 3460000 | 32893s | train_loss=0.0178
349
+ step 3470000 | 32981s | train_loss=0.0170
350
+ step 3480000 | 33070s | train_loss=0.0185
351
+ step 3490000 | 33159s | train_loss=0.0175
352
+ step 3500000 | 33315s | train_loss=0.0176 | fid=41.9942
353
+ step 3510000 | 33404s | train_loss=0.0171
354
+ step 3520000 | 33489s | train_loss=0.0174
355
+ step 3530000 | 33577s | train_loss=0.0166
356
+ step 3540000 | 33666s | train_loss=0.0169
357
+ step 3550000 | 33755s | train_loss=0.0161
358
+ step 3560000 | 33844s | train_loss=0.0166
359
+ step 3570000 | 33932s | train_loss=0.0164
360
+ step 3580000 | 34021s | train_loss=0.0174
361
+ step 3590000 | 34110s | train_loss=0.0166
362
+ step 3600000 | 34265s | train_loss=0.0178 | fid=41.8423
363
+ step 3610000 | 34354s | train_loss=0.0175
364
+ step 3620000 | 34442s | train_loss=0.0164
365
+ step 3630000 | 34528s | train_loss=0.0172
366
+ step 3640000 | 34617s | train_loss=0.0159
367
+ step 3650000 | 34706s | train_loss=0.0166
368
+ step 3660000 | 34795s | train_loss=0.0170
369
+ step 3670000 | 34883s | train_loss=0.0155
370
+ step 3680000 | 34972s | train_loss=0.0166
371
+ step 3690000 | 35060s | train_loss=0.0169
372
+ step 3700000 | 35218s | train_loss=0.0159 | fid=40.8183
373
+ step 3710000 | 35305s | train_loss=0.0161
374
+ step 3720000 | 35394s | train_loss=0.0164
375
+ step 3730000 | 35480s | train_loss=0.0161
376
+ step 3740000 | 35568s | train_loss=0.0166
377
+ step 3750000 | 35657s | train_loss=0.0162
378
+ step 3760000 | 35746s | train_loss=0.0161
379
+ step 3770000 | 35835s | train_loss=0.0163
380
+ step 3780000 | 35923s | train_loss=0.0156
381
+ step 3790000 | 36012s | train_loss=0.0170
382
+ step 3800000 | 36172s | train_loss=0.0159 | fid=41.5565
383
+ step 3810000 | 36258s | train_loss=0.0160
384
+ step 3820000 | 36346s | train_loss=0.0173
385
+ step 3830000 | 36436s | train_loss=0.0163
386
+ step 3840000 | 36521s | train_loss=0.0173
387
+ step 3850000 | 36610s | train_loss=0.0156
388
+ step 3860000 | 36699s | train_loss=0.0163
389
+ step 3870000 | 36787s | train_loss=0.0158
390
+ step 3880000 | 36876s | train_loss=0.0166
391
+ step 3890000 | 36964s | train_loss=0.0154
392
+ step 3900000 | 37124s | train_loss=0.0158 | fid=41.2400
393
+ step 3910000 | 37213s | train_loss=0.0161
394
+ step 3920000 | 37300s | train_loss=0.0166
395
+ step 3930000 | 37388s | train_loss=0.0154
396
+ step 3940000 | 37476s | train_loss=0.0153
397
+ step 3950000 | 37563s | train_loss=0.0162
398
+ step 3960000 | 37651s | train_loss=0.0156
399
+ step 3970000 | 37740s | train_loss=0.0162
400
+ step 3980000 | 37829s | train_loss=0.0163
401
+ step 3990000 | 37917s | train_loss=0.0157
402
+ step 4000000 | 38078s | train_loss=0.0151 | fid=40.8581
403
+ step 4010000 | 38167s | train_loss=0.0152
404
+ step 4020000 | 38253s | train_loss=0.0159
405
+ step 4030000 | 38342s | train_loss=0.0156
406
+ step 4040000 | 38431s | train_loss=0.0152
407
+ step 4050000 | 38516s | train_loss=0.0154
408
+ step 4060000 | 38605s | train_loss=0.0154
409
+ step 4070000 | 38693s | train_loss=0.0150
410
+ step 4080000 | 38782s | train_loss=0.0154
411
+ step 4090000 | 38871s | train_loss=0.0148
412
+ step 4100000 | 39032s | train_loss=0.0161 | fid=41.0416
413
+ step 4110000 | 39120s | train_loss=0.0145
414
+ step 4120000 | 39208s | train_loss=0.0148
415
+ step 4130000 | 39295s | train_loss=0.0146
416
+ step 4140000 | 39384s | train_loss=0.0147
417
+ step 4150000 | 39473s | train_loss=0.0154
418
+ step 4160000 | 39558s | train_loss=0.0162
419
+ step 4170000 | 39646s | train_loss=0.0147
420
+ step 4180000 | 39735s | train_loss=0.0150
421
+ step 4190000 | 39823s | train_loss=0.0148
422
+ step 4200000 | 39982s | train_loss=0.0151 | fid=40.6318
423
+ step 4210000 | 40071s | train_loss=0.0148
424
+ step 4220000 | 40160s | train_loss=0.0152
425
+ step 4230000 | 40245s | train_loss=0.0147
426
+ step 4240000 | 40334s | train_loss=0.0154
427
+ step 4250000 | 40423s | train_loss=0.0144
428
+ step 4260000 | 40511s | train_loss=0.0159
429
+ step 4270000 | 40597s | train_loss=0.0151
430
+ step 4280000 | 40685s | train_loss=0.0146
431
+ step 4290000 | 40774s | train_loss=0.0145
432
+ step 4300000 | 40935s | train_loss=0.0149 | fid=40.3678
433
+ step 4310000 | 41024s | train_loss=0.0142
434
+ step 4320000 | 41113s | train_loss=0.0158
435
+ step 4330000 | 41202s | train_loss=0.0147
436
+ step 4340000 | 41287s | train_loss=0.0142
437
+ step 4350000 | 41376s | train_loss=0.0151
438
+ step 4360000 | 41465s | train_loss=0.0145
439
+ step 4370000 | 41550s | train_loss=0.0150
440
+ step 4380000 | 41638s | train_loss=0.0151
441
+ step 4390000 | 41727s | train_loss=0.0141
442
+ step 4400000 | 41886s | train_loss=0.0148 | fid=40.3021
443
+ step 4410000 | 41975s | train_loss=0.0144
444
+ step 4420000 | 42064s | train_loss=0.0147
445
+ step 4430000 | 42153s | train_loss=0.0152
446
+ step 4440000 | 42241s | train_loss=0.0143
447
+ step 4450000 | 42328s | train_loss=0.0144
448
+ step 4460000 | 42417s | train_loss=0.0154
449
+ step 4470000 | 42504s | train_loss=0.0156
450
+ step 4480000 | 42591s | train_loss=0.0136
451
+ step 4490000 | 42679s | train_loss=0.0144
452
+ step 4500000 | 42837s | train_loss=0.0141 | fid=41.1638
453
+ step 4510000 | 42926s | train_loss=0.0135
454
+ step 4520000 | 43014s | train_loss=0.0145
455
+ step 4530000 | 43103s | train_loss=0.0153
456
+ step 4540000 | 43193s | train_loss=0.0145
457
+ step 4550000 | 43278s | train_loss=0.0133
458
+ step 4560000 | 43366s | train_loss=0.0144
459
+ step 4570000 | 43455s | train_loss=0.0140
460
+ step 4580000 | 43543s | train_loss=0.0142
461
+ step 4590000 | 43629s | train_loss=0.0146
462
+ step 4600000 | 43788s | train_loss=0.0141 | fid=39.8811
463
+ step 4610000 | 43876s | train_loss=0.0142
464
+ step 4620000 | 43965s | train_loss=0.0151
465
+ step 4630000 | 44054s | train_loss=0.0149
466
+ step 4640000 | 44143s | train_loss=0.0139
467
+ step 4650000 | 44230s | train_loss=0.0134
468
+ step 4660000 | 44317s | train_loss=0.0140
469
+ step 4670000 | 44406s | train_loss=0.0144
470
+ step 4680000 | 44495s | train_loss=0.0147
471
+ step 4690000 | 44580s | train_loss=0.0162
472
+ step 4700000 | 44740s | train_loss=0.0149 | fid=39.9527
473
+ step 4710000 | 44829s | train_loss=0.0146
474
+ step 4720000 | 44918s | train_loss=0.0140
475
+ step 4730000 | 45007s | train_loss=0.0140
476
+ step 4740000 | 45095s | train_loss=0.0139
477
+ step 4750000 | 45184s | train_loss=0.0141
478
+ step 4760000 | 45270s | train_loss=0.0140
479
+ step 4770000 | 45359s | train_loss=0.0140
480
+ step 4780000 | 45448s | train_loss=0.0138
481
+ step 4790000 | 45535s | train_loss=0.0141
482
+ step 4800000 | 45691s | train_loss=0.0142 | fid=39.4007
483
+ step 4810000 | 45780s | train_loss=0.0141
484
+ step 4820000 | 45869s | train_loss=0.0140
485
+ step 4830000 | 45958s | train_loss=0.0137
486
+ step 4840000 | 46046s | train_loss=0.0128
487
+ step 4850000 | 46135s | train_loss=0.0142
488
+ step 4860000 | 46224s | train_loss=0.0141
489
+ step 4870000 | 46310s | train_loss=0.0138
490
+ step 4880000 | 46398s | train_loss=0.0137
491
+ step 4890000 | 46487s | train_loss=0.0144
492
+ step 4900000 | 46640s | train_loss=0.0137 | fid=39.8510
493
+ step 4910000 | 46729s | train_loss=0.0137
494
+ step 4920000 | 46818s | train_loss=0.0132
495
+ step 4930000 | 46907s | train_loss=0.0146
496
+ step 4940000 | 46996s | train_loss=0.0140
497
+ step 4950000 | 47085s | train_loss=0.0134
498
+ step 4960000 | 47174s | train_loss=0.0126
499
+ step 4970000 | 47262s | train_loss=0.0132
500
+ step 4980000 | 47349s | train_loss=0.0126
501
+ step 4990000 | 47438s | train_loss=0.0134
502
+ step 5000000 | 47597s | train_loss=0.0143 | fid=40.2444
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d120/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 44,
25
+ "device": null
26
+ }
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120/metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d120/runner.log ADDED
@@ -0,0 +1,502 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ FID real-side stats cached over 1000 diverse train images
2
+ step 1 | 81s | train_loss=1.0001 | fid=264.9698
3
+ step 10000 | 178s | train_loss=0.7904
4
+ step 20000 | 274s | train_loss=0.6536
5
+ step 30000 | 371s | train_loss=0.5207
6
+ step 40000 | 468s | train_loss=0.4009
7
+ step 50000 | 564s | train_loss=0.3059
8
+ step 60000 | 661s | train_loss=0.2379
9
+ step 70000 | 758s | train_loss=0.1964
10
+ step 80000 | 855s | train_loss=0.1641
11
+ step 90000 | 952s | train_loss=0.1488
12
+ step 100000 | 1127s | train_loss=0.1344 | fid=64.2460
13
+ step 110000 | 1217s | train_loss=0.1236
14
+ step 120000 | 1314s | train_loss=0.1204
15
+ step 130000 | 1411s | train_loss=0.1111
16
+ step 140000 | 1508s | train_loss=0.1087
17
+ step 150000 | 1606s | train_loss=0.1059
18
+ step 160000 | 1703s | train_loss=0.0995
19
+ step 170000 | 1800s | train_loss=0.0975
20
+ step 180000 | 1882s | train_loss=0.0911
21
+ step 190000 | 1963s | train_loss=0.0898
22
+ step 200000 | 2110s | train_loss=0.0880 | fid=66.4076
23
+ step 210000 | 2184s | train_loss=0.0842
24
+ step 220000 | 2266s | train_loss=0.0847
25
+ step 230000 | 2348s | train_loss=0.0794
26
+ step 240000 | 2429s | train_loss=0.0796
27
+ step 250000 | 2510s | train_loss=0.0761
28
+ step 260000 | 2575s | train_loss=0.0761
29
+ step 270000 | 2641s | train_loss=0.0730
30
+ step 280000 | 2707s | train_loss=0.0723
31
+ step 290000 | 2773s | train_loss=0.0710
32
+ step 300000 | 2872s | train_loss=0.0691 | fid=64.0767
33
+ step 310000 | 2913s | train_loss=0.0669
34
+ step 320000 | 2964s | train_loss=0.0662
35
+ step 330000 | 3048s | train_loss=0.0667
36
+ step 340000 | 3179s | train_loss=0.0653
37
+ step 350000 | 3319s | train_loss=0.0638
38
+ step 360000 | 3459s | train_loss=0.0641
39
+ step 370000 | 3600s | train_loss=0.0617
40
+ step 380000 | 3740s | train_loss=0.0623
41
+ step 390000 | 3880s | train_loss=0.0613
42
+ step 400000 | 4125s | train_loss=0.0591 | fid=60.2693
43
+ step 410000 | 4262s | train_loss=0.0611
44
+ step 420000 | 4402s | train_loss=0.0592
45
+ step 430000 | 4541s | train_loss=0.0584
46
+ step 440000 | 4678s | train_loss=0.0581
47
+ step 450000 | 4817s | train_loss=0.0577
48
+ step 460000 | 4957s | train_loss=0.0571
49
+ step 470000 | 5097s | train_loss=0.0563
50
+ step 480000 | 5236s | train_loss=0.0554
51
+ step 490000 | 5376s | train_loss=0.0565
52
+ step 500000 | 5624s | train_loss=0.0536 | fid=62.0595
53
+ step 510000 | 5761s | train_loss=0.0542
54
+ step 520000 | 5901s | train_loss=0.0536
55
+ step 530000 | 6038s | train_loss=0.0533
56
+ step 540000 | 6178s | train_loss=0.0515
57
+ step 550000 | 6317s | train_loss=0.0523
58
+ step 560000 | 6457s | train_loss=0.0532
59
+ step 570000 | 6597s | train_loss=0.0527
60
+ step 580000 | 6737s | train_loss=0.0507
61
+ step 590000 | 6877s | train_loss=0.0504
62
+ step 600000 | 7123s | train_loss=0.0521 | fid=60.2587
63
+ step 610000 | 7262s | train_loss=0.0501
64
+ step 620000 | 7401s | train_loss=0.0502
65
+ step 630000 | 7537s | train_loss=0.0495
66
+ step 640000 | 7677s | train_loss=0.0480
67
+ step 650000 | 7816s | train_loss=0.0474
68
+ step 660000 | 7957s | train_loss=0.0484
69
+ step 670000 | 8096s | train_loss=0.0468
70
+ step 680000 | 8236s | train_loss=0.0467
71
+ step 690000 | 8376s | train_loss=0.0459
72
+ step 700000 | 8622s | train_loss=0.0475 | fid=58.0118
73
+ step 710000 | 8761s | train_loss=0.0472
74
+ step 720000 | 8898s | train_loss=0.0454
75
+ step 730000 | 9036s | train_loss=0.0461
76
+ step 740000 | 9176s | train_loss=0.0455
77
+ step 750000 | 9317s | train_loss=0.0450
78
+ step 760000 | 9454s | train_loss=0.0438
79
+ step 770000 | 9594s | train_loss=0.0450
80
+ step 780000 | 9733s | train_loss=0.0441
81
+ step 790000 | 9873s | train_loss=0.0441
82
+ step 800000 | 10121s | train_loss=0.0456 | fid=60.9981
83
+ step 810000 | 10261s | train_loss=0.0441
84
+ step 820000 | 10397s | train_loss=0.0442
85
+ step 830000 | 10536s | train_loss=0.0432
86
+ step 840000 | 10676s | train_loss=0.0442
87
+ step 850000 | 10816s | train_loss=0.0438
88
+ step 860000 | 10955s | train_loss=0.0429
89
+ step 870000 | 11031s | train_loss=0.0430
90
+ step 880000 | 11082s | train_loss=0.0414
91
+ step 890000 | 11133s | train_loss=0.0423
92
+ step 900000 | 11229s | train_loss=0.0431 | fid=61.3530
93
+ step 910000 | 11277s | train_loss=0.0415
94
+ step 920000 | 11323s | train_loss=0.0423
95
+ step 930000 | 11374s | train_loss=0.0419
96
+ step 940000 | 11425s | train_loss=0.0409
97
+ step 950000 | 11476s | train_loss=0.0411
98
+ step 960000 | 11527s | train_loss=0.0411
99
+ step 970000 | 11578s | train_loss=0.0416
100
+ step 980000 | 11629s | train_loss=0.0392
101
+ step 990000 | 11680s | train_loss=0.0403
102
+ step 1000000 | 11775s | train_loss=0.0403 | fid=61.8227
103
+ step 1010000 | 11825s | train_loss=0.0390
104
+ step 1020000 | 11869s | train_loss=0.0399
105
+ step 1030000 | 11920s | train_loss=0.0390
106
+ step 1040000 | 11971s | train_loss=0.0397
107
+ step 1050000 | 12022s | train_loss=0.0399
108
+ step 1060000 | 12073s | train_loss=0.0393
109
+ step 1070000 | 12124s | train_loss=0.0384
110
+ step 1080000 | 12176s | train_loss=0.0402
111
+ step 1090000 | 12227s | train_loss=0.0376
112
+ step 1100000 | 12321s | train_loss=0.0380 | fid=64.6304
113
+ step 1110000 | 12372s | train_loss=0.0379
114
+ step 1120000 | 12416s | train_loss=0.0369
115
+ step 1130000 | 12467s | train_loss=0.0384
116
+ step 1140000 | 12518s | train_loss=0.0387
117
+ step 1150000 | 12569s | train_loss=0.0378
118
+ step 1160000 | 12620s | train_loss=0.0370
119
+ step 1170000 | 12671s | train_loss=0.0369
120
+ step 1180000 | 12722s | train_loss=0.0365
121
+ step 1190000 | 12773s | train_loss=0.0373
122
+ step 1200000 | 12870s | train_loss=0.0375 | fid=63.3420
123
+ step 1210000 | 12921s | train_loss=0.0353
124
+ step 1220000 | 12965s | train_loss=0.0353
125
+ step 1230000 | 13016s | train_loss=0.0368
126
+ step 1240000 | 13067s | train_loss=0.0353
127
+ step 1250000 | 13118s | train_loss=0.0363
128
+ step 1260000 | 13169s | train_loss=0.0365
129
+ step 1270000 | 13220s | train_loss=0.0365
130
+ step 1280000 | 13271s | train_loss=0.0342
131
+ step 1290000 | 13322s | train_loss=0.0345
132
+ step 1300000 | 13418s | train_loss=0.0344 | fid=67.4713
133
+ step 1310000 | 13469s | train_loss=0.0338
134
+ step 1320000 | 13512s | train_loss=0.0357
135
+ step 1330000 | 13563s | train_loss=0.0349
136
+ step 1340000 | 13614s | train_loss=0.0349
137
+ step 1350000 | 13665s | train_loss=0.0350
138
+ step 1360000 | 13716s | train_loss=0.0346
139
+ step 1370000 | 13767s | train_loss=0.0341
140
+ step 1380000 | 13818s | train_loss=0.0339
141
+ step 1390000 | 13869s | train_loss=0.0352
142
+ step 1400000 | 13965s | train_loss=0.0343 | fid=68.7317
143
+ step 1410000 | 14016s | train_loss=0.0349
144
+ step 1420000 | 14059s | train_loss=0.0343
145
+ step 1430000 | 14110s | train_loss=0.0335
146
+ step 1440000 | 14161s | train_loss=0.0333
147
+ step 1450000 | 14212s | train_loss=0.0339
148
+ step 1460000 | 14263s | train_loss=0.0330
149
+ step 1470000 | 14314s | train_loss=0.0346
150
+ step 1480000 | 14365s | train_loss=0.0337
151
+ step 1490000 | 14415s | train_loss=0.0331
152
+ step 1500000 | 14512s | train_loss=0.0336 | fid=68.3534
153
+ step 1510000 | 14562s | train_loss=0.0326
154
+ step 1520000 | 14606s | train_loss=0.0338
155
+ step 1530000 | 14657s | train_loss=0.0350
156
+ step 1540000 | 14708s | train_loss=0.0314
157
+ step 1550000 | 14759s | train_loss=0.0331
158
+ step 1560000 | 14810s | train_loss=0.0318
159
+ step 1570000 | 14861s | train_loss=0.0314
160
+ step 1580000 | 14912s | train_loss=0.0341
161
+ step 1590000 | 14963s | train_loss=0.0312
162
+ step 1600000 | 15060s | train_loss=0.0307 | fid=67.2073
163
+ step 1610000 | 15111s | train_loss=0.0309
164
+ step 1620000 | 15155s | train_loss=0.0320
165
+ step 1630000 | 15206s | train_loss=0.0309
166
+ step 1640000 | 15257s | train_loss=0.0311
167
+ step 1650000 | 15308s | train_loss=0.0317
168
+ step 1660000 | 15359s | train_loss=0.0323
169
+ step 1670000 | 15410s | train_loss=0.0302
170
+ step 1680000 | 15461s | train_loss=0.0305
171
+ step 1690000 | 15512s | train_loss=0.0297
172
+ step 1700000 | 15610s | train_loss=0.0298 | fid=65.4129
173
+ step 1710000 | 15661s | train_loss=0.0297
174
+ step 1720000 | 15706s | train_loss=0.0308
175
+ step 1730000 | 15755s | train_loss=0.0309
176
+ step 1740000 | 15806s | train_loss=0.0311
177
+ step 1750000 | 15857s | train_loss=0.0307
178
+ step 1760000 | 15908s | train_loss=0.0299
179
+ step 1770000 | 15959s | train_loss=0.0299
180
+ step 1780000 | 16011s | train_loss=0.0303
181
+ step 1790000 | 16062s | train_loss=0.0291
182
+ step 1800000 | 16159s | train_loss=0.0294 | fid=62.9730
183
+ step 1810000 | 16210s | train_loss=0.0292
184
+ step 1820000 | 16256s | train_loss=0.0287
185
+ step 1830000 | 16305s | train_loss=0.0281
186
+ step 1840000 | 16356s | train_loss=0.0293
187
+ step 1850000 | 16407s | train_loss=0.0285
188
+ step 1860000 | 16458s | train_loss=0.0287
189
+ step 1870000 | 16509s | train_loss=0.0285
190
+ step 1880000 | 16560s | train_loss=0.0290
191
+ step 1890000 | 16611s | train_loss=0.0278
192
+ step 1900000 | 16708s | train_loss=0.0283 | fid=59.0636
193
+ step 1910000 | 16759s | train_loss=0.0290
194
+ step 1920000 | 16806s | train_loss=0.0291
195
+ step 1930000 | 16854s | train_loss=0.0278
196
+ step 1940000 | 16905s | train_loss=0.0281
197
+ step 1950000 | 16956s | train_loss=0.0290
198
+ step 1960000 | 17007s | train_loss=0.0279
199
+ step 1970000 | 17058s | train_loss=0.0268
200
+ step 1980000 | 17109s | train_loss=0.0285
201
+ step 1990000 | 17160s | train_loss=0.0280
202
+ step 2000000 | 17258s | train_loss=0.0280 | fid=59.4066
203
+ step 2010000 | 17309s | train_loss=0.0282
204
+ step 2020000 | 17356s | train_loss=0.0275
205
+ step 2030000 | 17404s | train_loss=0.0268
206
+ step 2040000 | 17455s | train_loss=0.0265
207
+ step 2050000 | 17506s | train_loss=0.0265
208
+ step 2060000 | 17557s | train_loss=0.0274
209
+ step 2070000 | 17608s | train_loss=0.0269
210
+ step 2080000 | 17659s | train_loss=0.0268
211
+ step 2090000 | 17710s | train_loss=0.0278
212
+ step 2100000 | 17808s | train_loss=0.0271 | fid=56.0580
213
+ step 2110000 | 17859s | train_loss=0.0257
214
+ step 2120000 | 17906s | train_loss=0.0262
215
+ step 2130000 | 17954s | train_loss=0.0254
216
+ step 2140000 | 18005s | train_loss=0.0274
217
+ step 2150000 | 18056s | train_loss=0.0260
218
+ step 2160000 | 18107s | train_loss=0.0254
219
+ step 2170000 | 18158s | train_loss=0.0289
220
+ step 2180000 | 18209s | train_loss=0.0271
221
+ step 2190000 | 18260s | train_loss=0.0264
222
+ step 2200000 | 18358s | train_loss=0.0255 | fid=55.2576
223
+ step 2210000 | 18409s | train_loss=0.0245
224
+ step 2220000 | 18456s | train_loss=0.0245
225
+ step 2230000 | 18503s | train_loss=0.0241
226
+ step 2240000 | 18555s | train_loss=0.0260
227
+ step 2250000 | 18606s | train_loss=0.0245
228
+ step 2260000 | 18657s | train_loss=0.0249
229
+ step 2270000 | 18708s | train_loss=0.0245
230
+ step 2280000 | 18759s | train_loss=0.0240
231
+ step 2290000 | 18810s | train_loss=0.0248
232
+ step 2300000 | 18952s | train_loss=0.0263 | fid=56.2251
233
+ step 2310000 | 18997s | train_loss=0.0243
234
+ step 2320000 | 19048s | train_loss=0.0243
235
+ step 2330000 | 19099s | train_loss=0.0254
236
+ step 2340000 | 19150s | train_loss=0.0230
237
+ step 2350000 | 19201s | train_loss=0.0236
238
+ step 2360000 | 19253s | train_loss=0.0244
239
+ step 2370000 | 19304s | train_loss=0.0249
240
+ step 2380000 | 19355s | train_loss=0.0240
241
+ step 2390000 | 19406s | train_loss=0.0236
242
+ step 2400000 | 19545s | train_loss=0.0237 | fid=56.5549
243
+ step 2410000 | 19596s | train_loss=0.0245
244
+ step 2420000 | 19647s | train_loss=0.0261
245
+ step 2430000 | 19698s | train_loss=0.0246
246
+ step 2440000 | 19750s | train_loss=0.0238
247
+ step 2450000 | 19801s | train_loss=0.0242
248
+ step 2460000 | 19852s | train_loss=0.0241
249
+ step 2470000 | 19903s | train_loss=0.0236
250
+ step 2480000 | 19954s | train_loss=0.0238
251
+ step 2490000 | 20006s | train_loss=0.0235
252
+ step 2500000 | 20096s | train_loss=0.0234 | fid=53.9069
253
+ step 2510000 | 20148s | train_loss=0.0245
254
+ step 2520000 | 20199s | train_loss=0.0240
255
+ step 2530000 | 20250s | train_loss=0.0233
256
+ step 2540000 | 20301s | train_loss=0.0229
257
+ step 2550000 | 20352s | train_loss=0.0226
258
+ step 2560000 | 20403s | train_loss=0.0248
259
+ step 2570000 | 20454s | train_loss=0.0225
260
+ step 2580000 | 20505s | train_loss=0.0225
261
+ step 2590000 | 20556s | train_loss=0.0235
262
+ step 2600000 | 20646s | train_loss=0.0232 | fid=53.7149
263
+ step 2610000 | 20697s | train_loss=0.0229
264
+ step 2620000 | 20749s | train_loss=0.0236
265
+ step 2630000 | 20800s | train_loss=0.0217
266
+ step 2640000 | 20851s | train_loss=0.0222
267
+ step 2650000 | 20902s | train_loss=0.0231
268
+ step 2660000 | 20953s | train_loss=0.0227
269
+ step 2670000 | 21004s | train_loss=0.0222
270
+ step 2680000 | 21055s | train_loss=0.0232
271
+ step 2690000 | 21106s | train_loss=0.0232
272
+ step 2700000 | 21197s | train_loss=0.0218 | fid=50.2741
273
+ step 2710000 | 21248s | train_loss=0.0223
274
+ step 2720000 | 21299s | train_loss=0.0216
275
+ step 2730000 | 21350s | train_loss=0.0218
276
+ step 2740000 | 21401s | train_loss=0.0221
277
+ step 2750000 | 21452s | train_loss=0.0218
278
+ step 2760000 | 21503s | train_loss=0.0230
279
+ step 2770000 | 21554s | train_loss=0.0217
280
+ step 2780000 | 21605s | train_loss=0.0220
281
+ step 2790000 | 21656s | train_loss=0.0224
282
+ step 2800000 | 21798s | train_loss=0.0225 | fid=49.8990
283
+ step 2810000 | 21849s | train_loss=0.0213
284
+ step 2820000 | 21900s | train_loss=0.0229
285
+ step 2830000 | 21952s | train_loss=0.0212
286
+ step 2840000 | 22003s | train_loss=0.0218
287
+ step 2850000 | 22054s | train_loss=0.0220
288
+ step 2860000 | 22105s | train_loss=0.0222
289
+ step 2870000 | 22156s | train_loss=0.0208
290
+ step 2880000 | 22207s | train_loss=0.0212
291
+ step 2890000 | 22254s | train_loss=0.0218
292
+ step 2900000 | 22398s | train_loss=0.0222 | fid=49.5496
293
+ step 2910000 | 22449s | train_loss=0.0207
294
+ step 2920000 | 22500s | train_loss=0.0210
295
+ step 2930000 | 22551s | train_loss=0.0207
296
+ step 2940000 | 22602s | train_loss=0.0212
297
+ step 2950000 | 22653s | train_loss=0.0214
298
+ step 2960000 | 22704s | train_loss=0.0219
299
+ step 2970000 | 22752s | train_loss=0.0206
300
+ step 2980000 | 22799s | train_loss=0.0211
301
+ step 2990000 | 22850s | train_loss=0.0213
302
+ step 3000000 | 22951s | train_loss=0.0207 | fid=47.5800
303
+ step 3010000 | 23002s | train_loss=0.0200
304
+ step 3020000 | 23053s | train_loss=0.0203
305
+ step 3030000 | 23104s | train_loss=0.0197
306
+ step 3040000 | 23155s | train_loss=0.0200
307
+ step 3050000 | 23206s | train_loss=0.0199
308
+ step 3060000 | 23257s | train_loss=0.0218
309
+ step 3070000 | 23305s | train_loss=0.0201
310
+ step 3080000 | 23352s | train_loss=0.0211
311
+ step 3090000 | 23403s | train_loss=0.0204
312
+ step 3100000 | 23500s | train_loss=0.0212 | fid=45.6669
313
+ step 3110000 | 23552s | train_loss=0.0201
314
+ step 3120000 | 23603s | train_loss=0.0207
315
+ step 3130000 | 23654s | train_loss=0.0199
316
+ step 3140000 | 23705s | train_loss=0.0198
317
+ step 3150000 | 23756s | train_loss=0.0194
318
+ step 3160000 | 23807s | train_loss=0.0204
319
+ step 3170000 | 23854s | train_loss=0.0203
320
+ step 3180000 | 23901s | train_loss=0.0194
321
+ step 3190000 | 23952s | train_loss=0.0198
322
+ step 3200000 | 24069s | train_loss=0.0212 | fid=45.3629
323
+ step 3210000 | 24120s | train_loss=0.0221
324
+ step 3220000 | 24171s | train_loss=0.0204
325
+ step 3230000 | 24222s | train_loss=0.0213
326
+ step 3240000 | 24274s | train_loss=0.0191
327
+ step 3250000 | 24325s | train_loss=0.0188
328
+ step 3260000 | 24374s | train_loss=0.0204
329
+ step 3270000 | 24423s | train_loss=0.0192
330
+ step 3280000 | 24470s | train_loss=0.0192
331
+ step 3290000 | 24521s | train_loss=0.0194
332
+ step 3300000 | 24636s | train_loss=0.0195 | fid=42.8405
333
+ step 3310000 | 24687s | train_loss=0.0195
334
+ step 3320000 | 24738s | train_loss=0.0193
335
+ step 3330000 | 24790s | train_loss=0.0185
336
+ step 3340000 | 24841s | train_loss=0.0206
337
+ step 3350000 | 24892s | train_loss=0.0196
338
+ step 3360000 | 24939s | train_loss=0.0206
339
+ step 3370000 | 24986s | train_loss=0.0192
340
+ step 3380000 | 25037s | train_loss=0.0190
341
+ step 3390000 | 25088s | train_loss=0.0189
342
+ step 3400000 | 25214s | train_loss=0.0203 | fid=43.8401
343
+ step 3410000 | 25265s | train_loss=0.0200
344
+ step 3420000 | 25316s | train_loss=0.0214
345
+ step 3430000 | 25367s | train_loss=0.0187
346
+ step 3440000 | 25418s | train_loss=0.0197
347
+ step 3450000 | 25465s | train_loss=0.0182
348
+ step 3460000 | 25513s | train_loss=0.0189
349
+ step 3470000 | 25564s | train_loss=0.0187
350
+ step 3480000 | 25615s | train_loss=0.0194
351
+ step 3490000 | 25666s | train_loss=0.0184
352
+ step 3500000 | 25800s | train_loss=0.0189 | fid=43.7605
353
+ step 3510000 | 25851s | train_loss=0.0196
354
+ step 3520000 | 25902s | train_loss=0.0183
355
+ step 3530000 | 25953s | train_loss=0.0181
356
+ step 3540000 | 26001s | train_loss=0.0186
357
+ step 3550000 | 26048s | train_loss=0.0179
358
+ step 3560000 | 26099s | train_loss=0.0190
359
+ step 3570000 | 26150s | train_loss=0.0187
360
+ step 3580000 | 26201s | train_loss=0.0189
361
+ step 3590000 | 26252s | train_loss=0.0180
362
+ step 3600000 | 26363s | train_loss=0.0192 | fid=42.7516
363
+ step 3610000 | 26414s | train_loss=0.0194
364
+ step 3620000 | 26465s | train_loss=0.0177
365
+ step 3630000 | 26516s | train_loss=0.0206
366
+ step 3640000 | 26564s | train_loss=0.0179
367
+ step 3650000 | 26611s | train_loss=0.0179
368
+ step 3660000 | 26662s | train_loss=0.0190
369
+ step 3670000 | 26713s | train_loss=0.0175
370
+ step 3680000 | 26764s | train_loss=0.0178
371
+ step 3690000 | 26815s | train_loss=0.0183
372
+ step 3700000 | 26910s | train_loss=0.0170 | fid=43.0350
373
+ step 3710000 | 26962s | train_loss=0.0175
374
+ step 3720000 | 27013s | train_loss=0.0187
375
+ step 3730000 | 27064s | train_loss=0.0183
376
+ step 3740000 | 27111s | train_loss=0.0174
377
+ step 3750000 | 27158s | train_loss=0.0178
378
+ step 3760000 | 27210s | train_loss=0.0184
379
+ step 3770000 | 27261s | train_loss=0.0179
380
+ step 3780000 | 27312s | train_loss=0.0169
381
+ step 3790000 | 27363s | train_loss=0.0187
382
+ step 3800000 | 27457s | train_loss=0.0177 | fid=41.9639
383
+ step 3810000 | 27508s | train_loss=0.0179
384
+ step 3820000 | 27559s | train_loss=0.0188
385
+ step 3830000 | 27610s | train_loss=0.0186
386
+ step 3840000 | 27657s | train_loss=0.0182
387
+ step 3850000 | 27705s | train_loss=0.0176
388
+ step 3860000 | 27756s | train_loss=0.0173
389
+ step 3870000 | 27807s | train_loss=0.0171
390
+ step 3880000 | 27858s | train_loss=0.0185
391
+ step 3890000 | 27909s | train_loss=0.0176
392
+ step 3900000 | 28003s | train_loss=0.0176 | fid=41.6389
393
+ step 3910000 | 28054s | train_loss=0.0182
394
+ step 3920000 | 28105s | train_loss=0.0172
395
+ step 3930000 | 28156s | train_loss=0.0168
396
+ step 3940000 | 28203s | train_loss=0.0172
397
+ step 3950000 | 28251s | train_loss=0.0171
398
+ step 3960000 | 28302s | train_loss=0.0171
399
+ step 3970000 | 28353s | train_loss=0.0186
400
+ step 3980000 | 28404s | train_loss=0.0169
401
+ step 3990000 | 28455s | train_loss=0.0169
402
+ step 4000000 | 28549s | train_loss=0.0172 | fid=41.7202
403
+ step 4010000 | 28600s | train_loss=0.0172
404
+ step 4020000 | 28651s | train_loss=0.0176
405
+ step 4030000 | 28702s | train_loss=0.0171
406
+ step 4040000 | 28750s | train_loss=0.0169
407
+ step 4050000 | 28799s | train_loss=0.0170
408
+ step 4060000 | 28848s | train_loss=0.0174
409
+ step 4070000 | 28899s | train_loss=0.0174
410
+ step 4080000 | 28950s | train_loss=0.0165
411
+ step 4090000 | 29001s | train_loss=0.0170
412
+ step 4100000 | 29095s | train_loss=0.0166 | fid=41.8958
413
+ step 4110000 | 29147s | train_loss=0.0167
414
+ step 4120000 | 29198s | train_loss=0.0166
415
+ step 4130000 | 29249s | train_loss=0.0162
416
+ step 4140000 | 29296s | train_loss=0.0161
417
+ step 4150000 | 29347s | train_loss=0.0165
418
+ step 4160000 | 29394s | train_loss=0.0177
419
+ step 4170000 | 29446s | train_loss=0.0157
420
+ step 4180000 | 29497s | train_loss=0.0171
421
+ step 4190000 | 29548s | train_loss=0.0164
422
+ step 4200000 | 29642s | train_loss=0.0170 | fid=41.8837
423
+ step 4210000 | 29693s | train_loss=0.0159
424
+ step 4220000 | 29744s | train_loss=0.0169
425
+ step 4230000 | 29795s | train_loss=0.0169
426
+ step 4240000 | 29843s | train_loss=0.0169
427
+ step 4250000 | 29894s | train_loss=0.0160
428
+ step 4260000 | 29941s | train_loss=0.0181
429
+ step 4270000 | 29992s | train_loss=0.0172
430
+ step 4280000 | 30043s | train_loss=0.0157
431
+ step 4290000 | 30094s | train_loss=0.0161
432
+ step 4300000 | 30188s | train_loss=0.0165 | fid=40.9179
433
+ step 4310000 | 30240s | train_loss=0.0157
434
+ step 4320000 | 30291s | train_loss=0.0175
435
+ step 4330000 | 30342s | train_loss=0.0162
436
+ step 4340000 | 30390s | train_loss=0.0159
437
+ step 4350000 | 30440s | train_loss=0.0158
438
+ step 4360000 | 30488s | train_loss=0.0161
439
+ step 4370000 | 30539s | train_loss=0.0166
440
+ step 4380000 | 30590s | train_loss=0.0169
441
+ step 4390000 | 30641s | train_loss=0.0165
442
+ step 4400000 | 30735s | train_loss=0.0162 | fid=40.5435
443
+ step 4410000 | 30787s | train_loss=0.0160
444
+ step 4420000 | 30838s | train_loss=0.0163
445
+ step 4430000 | 30889s | train_loss=0.0164
446
+ step 4440000 | 30939s | train_loss=0.0163
447
+ step 4450000 | 30987s | train_loss=0.0157
448
+ step 4460000 | 31034s | train_loss=0.0163
449
+ step 4470000 | 31086s | train_loss=0.0170
450
+ step 4480000 | 31137s | train_loss=0.0153
451
+ step 4490000 | 31188s | train_loss=0.0167
452
+ step 4500000 | 31282s | train_loss=0.0160 | fid=39.7900
453
+ step 4510000 | 31333s | train_loss=0.0148
454
+ step 4520000 | 31385s | train_loss=0.0164
455
+ step 4530000 | 31436s | train_loss=0.0160
456
+ step 4540000 | 31487s | train_loss=0.0159
457
+ step 4550000 | 31534s | train_loss=0.0150
458
+ step 4560000 | 31581s | train_loss=0.0172
459
+ step 4570000 | 31632s | train_loss=0.0156
460
+ step 4580000 | 31683s | train_loss=0.0155
461
+ step 4590000 | 31735s | train_loss=0.0158
462
+ step 4600000 | 31829s | train_loss=0.0156 | fid=38.7069
463
+ step 4610000 | 31880s | train_loss=0.0159
464
+ step 4620000 | 31932s | train_loss=0.0175
465
+ step 4630000 | 31983s | train_loss=0.0155
466
+ step 4640000 | 32034s | train_loss=0.0152
467
+ step 4650000 | 32081s | train_loss=0.0152
468
+ step 4660000 | 32128s | train_loss=0.0157
469
+ step 4670000 | 32179s | train_loss=0.0150
470
+ step 4680000 | 32230s | train_loss=0.0164
471
+ step 4690000 | 32281s | train_loss=0.0183
472
+ step 4700000 | 32379s | train_loss=0.0173 | fid=40.5607
473
+ step 4710000 | 32430s | train_loss=0.0162
474
+ step 4720000 | 32481s | train_loss=0.0151
475
+ step 4730000 | 32532s | train_loss=0.0155
476
+ step 4740000 | 32583s | train_loss=0.0154
477
+ step 4750000 | 32630s | train_loss=0.0161
478
+ step 4760000 | 32677s | train_loss=0.0153
479
+ step 4770000 | 32728s | train_loss=0.0148
480
+ step 4780000 | 32779s | train_loss=0.0149
481
+ step 4790000 | 32830s | train_loss=0.0147
482
+ step 4800000 | 32926s | train_loss=0.0159 | fid=38.7715
483
+ step 4810000 | 32978s | train_loss=0.0163
484
+ step 4820000 | 33029s | train_loss=0.0155
485
+ step 4830000 | 33080s | train_loss=0.0145
486
+ step 4840000 | 33131s | train_loss=0.0140
487
+ step 4850000 | 33178s | train_loss=0.0150
488
+ step 4860000 | 33226s | train_loss=0.0151
489
+ step 4870000 | 33276s | train_loss=0.0149
490
+ step 4880000 | 33327s | train_loss=0.0156
491
+ step 4890000 | 33378s | train_loss=0.0164
492
+ step 4900000 | 33476s | train_loss=0.0154 | fid=39.3493
493
+ step 4910000 | 33527s | train_loss=0.0143
494
+ step 4920000 | 33578s | train_loss=0.0150
495
+ step 4930000 | 33629s | train_loss=0.0159
496
+ step 4940000 | 33680s | train_loss=0.0152
497
+ step 4950000 | 33718s | train_loss=0.0151
498
+ step 4960000 | 33750s | train_loss=0.0141
499
+ step 4970000 | 33767s | train_loss=0.0154
500
+ step 4980000 | 33786s | train_loss=0.0145
501
+ step 4990000 | 33804s | train_loss=0.0151
502
+ step 5000000 | 33843s | train_loss=0.0159 | fid=39.1780
results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d140/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "vae_checkpoint": "vae_checkpoints/celeba_resnet_modernloss_d140/vae.pt",
3
+ "out": "results/celeba_diverse1k_bigmlp_sgd_lr001_m08_10k_5m/seed44/d140",
4
+ "subset_path": "diagnostics/celeba_diverse_1k_mean_plus_sd/subset_indices.json",
5
+ "train_tar": "/overflow/data/cerberus/celeba_hq_256_partitioned_ws/train/celeba_hq_256-train-0000.tar",
6
+ "n_train": 1000,
7
+ "total_steps": 5000000,
8
+ "eval_interval": 10000,
9
+ "mem_interval": 100000,
10
+ "fid_interval": 100000,
11
+ "fid_n_real": 1000,
12
+ "fid_n_gen": 10000,
13
+ "n_gen_samples": 10000,
14
+ "n_sde_steps": 500,
15
+ "batch_size": 512,
16
+ "hidden": 1024,
17
+ "depth": 5,
18
+ "optimizer": "sgd",
19
+ "lr": 0.001,
20
+ "momentum": 0.8,
21
+ "t_min": 0.01,
22
+ "t_max": 3.0,
23
+ "t_eval": 0.1,
24
+ "seed": 44,
25
+ "device": null
26
+ }