wordbyword_attention_dropout0.2.log

(root)junfeng@hadoop1:~/reasoning_attention$ python snli_reasoning_attention.py
Loading data ...
Loading train ...
550152
550149
549364
Loading dev ...
10000
10000
9842
Loading test ...
10000
10000
9824
num_epochs: 20
k: 100
batch_size: 30
display_frequency: 100
save_frequency: 1000
load previous: False
attention: True
word_by_word: True
Building network ...
unchanged_W.shape: (34283, 300)
oov_in_train_W.shape: (9166, 300)
apply dropout rate 0.2 to decoder
apply dropout mask id 139654683757200 to embedding matrix ...
dropout rate is 0.2
input var is hypo_var
apply dropout mask id 139654683757200 to embedding matrix ...
dropout rate is 0.2
input var is premise_var
Computing updates ...
Compiling functions ...
Training ...
train_df.shape: (549364, 4)
dev_df.shape: (9842, 4)
test_df.shape: (9824, 4)
Starting training...
Seen 3000 samples, time used: 70.550s
  current training loss:                1.058450
  current training accuracy:            0.454333
Seen 6000 samples, time used: 142.724s
  current training loss:                1.028129
  current training accuracy:            0.488833
Seen 9000 samples, time used: 215.414s
  current training loss:                1.012946
  current training accuracy:            0.503000
Seen 12000 samples, time used: 285.631s
  current training loss:                1.002247
  current training accuracy:            0.513500
Seen 15000 samples, time used: 358.950s
  current training loss:                0.990062
  current training accuracy:            0.527667
Seen 18000 samples, time used: 431.633s
  current training loss:                0.981677
  current training accuracy:            0.537278
Seen 21000 samples, time used: 504.303s
  current training loss:                0.973090
  current training accuracy:            0.544619
Seen 24000 samples, time used: 580.315s
  current training loss:                0.967607
  current training accuracy:            0.549458
Seen 27000 samples, time used: 654.325s
  current training loss:                0.960573
  current training accuracy:            0.555630
Seen 30000 samples, time used: 728.366s
  current training loss:                0.957582
  current training accuracy:            0.558967
saving to ..., time used 728.366s
Seen 33000 samples, time used: 801.570s
  current training loss:                0.954111
  current training accuracy:            0.560636
Seen 36000 samples, time used: 875.266s
  current training loss:                0.950953
  current training accuracy:            0.563944
Seen 39000 samples, time used: 947.427s
  current training loss:                0.948558
  current training accuracy:            0.566077
Seen 42000 samples, time used: 1021.095s
  current training loss:                0.945443
  current training accuracy:            0.567857
Seen 45000 samples, time used: 1095.418s
  current training loss:                0.941933
  current training accuracy:            0.570733
Seen 48000 samples, time used: 1169.017s
  current training loss:                0.938352
  current training accuracy:            0.573167
Seen 51000 samples, time used: 1243.517s
  current training loss:                0.935981
  current training accuracy:            0.574314
Seen 54000 samples, time used: 1317.343s
  current training loss:                0.932651
  current training accuracy:            0.576222
Seen 57000 samples, time used: 1393.935s
  current training loss:                0.929052
  current training accuracy:            0.578333
Seen 60000 samples, time used: 1469.266s
  current training loss:                0.926869
  current training accuracy:            0.580050
saving to ..., time used 740.329s
Seen 63000 samples, time used: 1543.862s
  current training loss:                0.924585
  current training accuracy:            0.581397
Seen 66000 samples, time used: 1618.405s
  current training loss:                0.923107
  current training accuracy:            0.582591
Seen 69000 samples, time used: 1694.256s
  current training loss:                0.920630
  current training accuracy:            0.583957
Seen 72000 samples, time used: 1768.998s
  current training loss:                0.917708
  current training accuracy:            0.585611
Seen 75000 samples, time used: 1845.098s
  current training loss:                0.915333
  current training accuracy:            0.587027
Seen 78000 samples, time used: 1920.224s
  current training loss:                0.913331
  current training accuracy:            0.588218
Seen 81000 samples, time used: 1996.728s
  current training loss:                0.911064
  current training accuracy:            0.589802
Seen 84000 samples, time used: 2073.960s
  current training loss:                0.909034
  current training accuracy:            0.591155
Seen 87000 samples, time used: 2149.168s
  current training loss:                0.907857
  current training accuracy:            0.591931
Seen 90000 samples, time used: 2223.754s
  current training loss:                0.906369
  current training accuracy:            0.592789
saving to ..., time used 753.934s
Seen 93000 samples, time used: 2300.643s
  current training loss:                0.904891
  current training accuracy:            0.594097
Seen 96000 samples, time used: 2374.767s
  current training loss:                0.903432
  current training accuracy:            0.595104
Seen 99000 samples, time used: 2451.684s
  current training loss:                0.902138
  current training accuracy:            0.595980
Seen 102000 samples, time used: 2527.388s
  current training loss:                0.900729
  current training accuracy:            0.597324
Seen 105000 samples, time used: 2603.019s
  current training loss:                0.898827
  current training accuracy:            0.598705
Seen 108000 samples, time used: 2678.314s
  current training loss:                0.897753
  current training accuracy:            0.599333
Seen 111000 samples, time used: 2754.624s
  current training loss:                0.896715
  current training accuracy:            0.599847
Seen 114000 samples, time used: 2830.629s
  current training loss:                0.895394
  current training accuracy:            0.600465
Seen 117000 samples, time used: 2908.376s
  current training loss:                0.893256
  current training accuracy:            0.601701
Seen 120000 samples, time used: 2985.160s
  current training loss:                0.892140
  current training accuracy:            0.602333
saving to ..., time used 760.917s
Seen 123000 samples, time used: 3061.921s
  current training loss:                0.890923
  current training accuracy:            0.603179
Seen 126000 samples, time used: 3139.882s
  current training loss:                0.889364
  current training accuracy:            0.604389
Seen 129000 samples, time used: 3216.813s
  current training loss:                0.888290
  current training accuracy:            0.605155
Seen 132000 samples, time used: 3293.700s
  current training loss:                0.887103
  current training accuracy:            0.605932
Seen 135000 samples, time used: 3370.489s
  current training loss:                0.886399
  current training accuracy:            0.606452
Seen 138000 samples, time used: 3447.364s
  current training loss:                0.885536
  current training accuracy:            0.606942
Seen 141000 samples, time used: 3523.391s
  current training loss:                0.884477
  current training accuracy:            0.607731
Seen 144000 samples, time used: 3598.818s
  current training loss:                0.883690
  current training accuracy:            0.608354
Seen 147000 samples, time used: 3675.130s
  current training loss:                0.882003
  current training accuracy:            0.609633
Seen 150000 samples, time used: 3750.648s
  current training loss:                0.880902
  current training accuracy:            0.610360
saving to ..., time used 764.986s
Seen 153000 samples, time used: 3828.365s
  current training loss:                0.879758
  current training accuracy:            0.611046
Seen 156000 samples, time used: 3904.438s
  current training loss:                0.878782
  current training accuracy:            0.611712
Seen 159000 samples, time used: 3981.737s
  current training loss:                0.877662
  current training accuracy:            0.612453
Seen 162000 samples, time used: 4056.862s
  current training loss:                0.876627
  current training accuracy:            0.613327
Seen 165000 samples, time used: 4132.630s
  current training loss:                0.875626
  current training accuracy:            0.613933
Seen 168000 samples, time used: 4210.021s
  current training loss:                0.874510
  current training accuracy:            0.614607
Seen 171000 samples, time used: 4286.664s
  current training loss:                0.873347
  current training accuracy:            0.615228
Seen 174000 samples, time used: 4363.062s
  current training loss:                0.871881
  current training accuracy:            0.616155
Seen 177000 samples, time used: 4440.542s
  current training loss:                0.870728
  current training accuracy:            0.617062
Seen 180000 samples, time used: 4516.914s
  current training loss:                0.869758
  current training accuracy:            0.617683
saving to ..., time used 765.767s
Seen 183000 samples, time used: 4592.880s
  current training loss:                0.869033
  current training accuracy:            0.618158
Seen 186000 samples, time used: 4671.211s
  current training loss:                0.868205
  current training accuracy:            0.618710
Seen 189000 samples, time used: 4748.834s
  current training loss:                0.867292
  current training accuracy:            0.619323
Seen 192000 samples, time used: 4826.439s
  current training loss:                0.866717
  current training accuracy:            0.619828
Seen 195000 samples, time used: 4904.512s
  current training loss:                0.865819
  current training accuracy:            0.620472
Seen 198000 samples, time used: 4981.086s
  current training loss:                0.864718
  current training accuracy:            0.621106
Seen 201000 samples, time used: 5057.278s
  current training loss:                0.863890
  current training accuracy:            0.621542
Seen 204000 samples, time used: 5132.647s
  current training loss:                0.863093
  current training accuracy:            0.621902
Seen 207000 samples, time used: 5209.594s
  current training loss:                0.862247
  current training accuracy:            0.622406
Seen 210000 samples, time used: 5286.110s
  current training loss:                0.861501
  current training accuracy:            0.622929
saving to ..., time used 768.582s
Seen 213000 samples, time used: 5362.617s
  current training loss:                0.860481
  current training accuracy:            0.623535
Seen 216000 samples, time used: 5440.070s
  current training loss:                0.859284
  current training accuracy:            0.624176
Seen 219000 samples, time used: 5518.801s
  current training loss:                0.858481
  current training accuracy:            0.624612
Seen 222000 samples, time used: 5595.570s
  current training loss:                0.857809
  current training accuracy:            0.625063
Seen 225000 samples, time used: 5672.717s
  current training loss:                0.856960
  current training accuracy:            0.625600
Seen 228000 samples, time used: 5750.543s
  current training loss:                0.855970
  current training accuracy:            0.626193
Seen 231000 samples, time used: 5827.890s
  current training loss:                0.855197
  current training accuracy:            0.626645
Seen 234000 samples, time used: 5906.121s
  current training loss:                0.854365
  current training accuracy:            0.627179
Seen 237000 samples, time used: 5982.841s
  current training loss:                0.853599
  current training accuracy:            0.627738
Seen 240000 samples, time used: 6059.667s
  current training loss:                0.852604
  current training accuracy:            0.628479
saving to ..., time used 773.038s
Seen 243000 samples, time used: 6137.219s
  current training loss:                0.851878
  current training accuracy:            0.628992
Seen 246000 samples, time used: 6214.856s
  current training loss:                0.851155
  current training accuracy:            0.629423
Seen 249000 samples, time used: 6292.779s
  current training loss:                0.850244
  current training accuracy:            0.630096
Seen 252000 samples, time used: 6370.032s
  current training loss:                0.849409
  current training accuracy:            0.630722
Seen 255000 samples, time used: 6448.344s
  current training loss:                0.848375
  current training accuracy:            0.631388
Seen 258000 samples, time used: 6525.827s
  current training loss:                0.847384
  current training accuracy:            0.631996
Seen 261000 samples, time used: 6602.790s
  current training loss:                0.846526
  current training accuracy:            0.632513
Seen 264000 samples, time used: 6678.194s
  current training loss:                0.845677
  current training accuracy:            0.633080
Seen 267000 samples, time used: 6754.856s
  current training loss:                0.844715
  current training accuracy:            0.633622
Seen 270000 samples, time used: 6832.567s
  current training loss:                0.843931
  current training accuracy:            0.634178
saving to ..., time used 772.379s
Seen 273000 samples, time used: 6911.022s
  current training loss:                0.843138
  current training accuracy:            0.634758
Seen 276000 samples, time used: 6989.312s
  current training loss:                0.842343
  current training accuracy:            0.635297
Seen 279000 samples, time used: 7066.756s
  current training loss:                0.841583
  current training accuracy:            0.635735
Seen 282000 samples, time used: 7144.451s
  current training loss:                0.840777
  current training accuracy:            0.636202
Seen 285000 samples, time used: 7222.840s
  current training loss:                0.839991
  current training accuracy:            0.636632
Seen 288000 samples, time used: 7300.611s
  current training loss:                0.839102
  current training accuracy:            0.637208
Seen 291000 samples, time used: 7378.999s
  current training loss:                0.838377
  current training accuracy:            0.637684
Seen 294000 samples, time used: 7455.955s
  current training loss:                0.837620
  current training accuracy:            0.638214
Seen 297000 samples, time used: 7534.368s
  current training loss:                0.836800
  current training accuracy:            0.638771
Seen 300000 samples, time used: 7612.794s
  current training loss:                0.836119
  current training accuracy:            0.639300
saving to ..., time used 779.672s
Seen 303000 samples, time used: 7690.596s
  current training loss:                0.835410
  current training accuracy:            0.639776
Seen 306000 samples, time used: 7771.854s
  current training loss:                0.834700
  current training accuracy:            0.640219
Seen 309000 samples, time used: 7850.173s
  current training loss:                0.833848
  current training accuracy:            0.640796
Seen 312000 samples, time used: 7930.185s
  current training loss:                0.833122
  current training accuracy:            0.641247
Seen 315000 samples, time used: 8010.226s
  current training loss:                0.832480
  current training accuracy:            0.641575
Seen 318000 samples, time used: 8092.514s
  current training loss:                0.831846
  current training accuracy:            0.641934
Seen 321000 samples, time used: 8171.449s
  current training loss:                0.831251
  current training accuracy:            0.642299
Seen 324000 samples, time used: 8252.354s
  current training loss:                0.830513
  current training accuracy:            0.642719
Seen 327000 samples, time used: 8331.348s
  current training loss:                0.829876
  current training accuracy:            0.643095
Seen 330000 samples, time used: 8411.129s
  current training loss:                0.829133
  current training accuracy:            0.643530
saving to ..., time used 797.739s
Seen 333000 samples, time used: 8490.465s
  current training loss:                0.828562
  current training accuracy:            0.643928
Seen 336000 samples, time used: 8571.024s
  current training loss:                0.827978
  current training accuracy:            0.644396
Seen 339000 samples, time used: 8650.506s
  current training loss:                0.827388
  current training accuracy:            0.644776
Seen 342000 samples, time used: 8729.620s
  current training loss:                0.826704
  current training accuracy:            0.645272
Seen 345000 samples, time used: 8811.936s
  current training loss:                0.825998
  current training accuracy:            0.645788
Seen 348000 samples, time used: 8891.614s
  current training loss:                0.825274
  current training accuracy:            0.646244
Seen 351000 samples, time used: 8971.370s
  current training loss:                0.824640
  current training accuracy:            0.646621
Seen 354000 samples, time used: 9050.490s
  current training loss:                0.824112
  current training accuracy:            0.646977
Seen 357000 samples, time used: 9132.854s
  current training loss:                0.823348
  current training accuracy:            0.647392
Seen 360000 samples, time used: 9210.575s
  current training loss:                0.822844
  current training accuracy:            0.647756
saving to ..., time used 798.905s
Seen 363000 samples, time used: 9288.534s
  current training loss:                0.822372
  current training accuracy:            0.648141
Seen 366000 samples, time used: 9367.652s
  current training loss:                0.821746
  current training accuracy:            0.648549
Seen 369000 samples, time used: 9446.224s
  current training loss:                0.821116
  current training accuracy:            0.648959
Seen 372000 samples, time used: 9525.181s
  current training loss:                0.820527
  current training accuracy:            0.649272
Seen 375000 samples, time used: 9603.285s
  current training loss:                0.819984
  current training accuracy:            0.649584
Seen 378000 samples, time used: 9683.882s
  current training loss:                0.819474
  current training accuracy:            0.649844
Seen 381000 samples, time used: 9762.385s
  current training loss:                0.818987
  current training accuracy:            0.650220
Seen 384000 samples, time used: 9839.191s
  current training loss:                0.818296
  current training accuracy:            0.650635
Seen 387000 samples, time used: 9918.167s
  current training loss:                0.817921
  current training accuracy:            0.650904
Seen 390000 samples, time used: 9995.818s
  current training loss:                0.817336
  current training accuracy:            0.651323
saving to ..., time used 784.710s
Seen 393000 samples, time used: 10074.369s
  current training loss:                0.816814
  current training accuracy:            0.651555
Seen 396000 samples, time used: 10154.144s
  current training loss:                0.816189
  current training accuracy:            0.651942
Seen 399000 samples, time used: 10234.721s
  current training loss:                0.815564
  current training accuracy:            0.652303
Seen 402000 samples, time used: 10313.530s
  current training loss:                0.814815
  current training accuracy:            0.652769
Seen 405000 samples, time used: 10392.349s
  current training loss:                0.814213
  current training accuracy:            0.653195
Seen 408000 samples, time used: 10471.245s
  current training loss:                0.813462
  current training accuracy:            0.653662
Seen 411000 samples, time used: 10549.904s
  current training loss:                0.812966
  current training accuracy:            0.653978
Seen 414000 samples, time used: 10626.159s
  current training loss:                0.812332
  current training accuracy:            0.654382
Seen 417000 samples, time used: 10703.324s
  current training loss:                0.811765
  current training accuracy:            0.654717
Seen 420000 samples, time used: 10781.662s
  current training loss:                0.811159
  current training accuracy:            0.655071
saving to ..., time used 785.329s
Seen 423000 samples, time used: 10862.916s
  current training loss:                0.810501
  current training accuracy:            0.655553
Seen 426000 samples, time used: 10942.944s
  current training loss:                0.810101
  current training accuracy:            0.655812
Seen 429000 samples, time used: 11024.157s
  current training loss:                0.809608
  current training accuracy:            0.656161
Seen 432000 samples, time used: 11104.469s
  current training loss:                0.809338
  current training accuracy:            0.656384
Seen 435000 samples, time used: 11185.540s
  current training loss:                0.808774
  current training accuracy:            0.656736
Seen 438000 samples, time used: 11268.074s
  current training loss:                0.808411
  current training accuracy:            0.657055
Seen 441000 samples, time used: 11348.420s
  current training loss:                0.807850
  current training accuracy:            0.657472
Seen 444000 samples, time used: 11430.465s
  current training loss:                0.807281
  current training accuracy:            0.657813
Seen 447000 samples, time used: 11506.953s
  current training loss:                0.806842
  current training accuracy:            0.658076
Seen 450000 samples, time used: 11584.854s
  current training loss:                0.806289
  current training accuracy:            0.658420
saving to ..., time used 802.709s
Seen 453000 samples, time used: 11663.175s
  current training loss:                0.805796
  current training accuracy:            0.658689
Seen 456000 samples, time used: 11741.102s
  current training loss:                0.805197
  current training accuracy:            0.659059
Seen 459000 samples, time used: 11819.311s
  current training loss:                0.804648
  current training accuracy:            0.659392
Seen 462000 samples, time used: 11896.642s
  current training loss:                0.804072
  current training accuracy:            0.659758
Seen 465000 samples, time used: 11974.515s
  current training loss:                0.803673
  current training accuracy:            0.660077
Seen 468000 samples, time used: 12053.498s
  current training loss:                0.803240
  current training accuracy:            0.660374
Seen 471000 samples, time used: 12131.072s
  current training loss:                0.802600
  current training accuracy:            0.660754
Seen 474000 samples, time used: 12208.327s
  current training loss:                0.801929
  current training accuracy:            0.661175
Seen 477000 samples, time used: 12287.774s
  current training loss:                0.801393
  current training accuracy:            0.661497
Seen 480000 samples, time used: 12365.440s
  current training loss:                0.800831
  current training accuracy:            0.661856
saving to ..., time used 780.088s
Seen 483000 samples, time used: 12444.851s
  current training loss:                0.800363
  current training accuracy:            0.662168
Seen 486000 samples, time used: 12523.230s
  current training loss:                0.799814
  current training accuracy:            0.662529
Seen 489000 samples, time used: 12602.246s
  current training loss:                0.799402
  current training accuracy:            0.662800
Seen 492000 samples, time used: 12682.109s
  current training loss:                0.798880
  current training accuracy:            0.663136
Seen 495000 samples, time used: 12763.496s
  current training loss:                0.798289
  current training accuracy:            0.663503
Seen 498000 samples, time used: 12846.349s
  current training loss:                0.797708
  current training accuracy:            0.663821
Seen 501000 samples, time used: 12923.748s
  current training loss:                0.797193
  current training accuracy:            0.664096
Seen 504000 samples, time used: 13001.622s
  current training loss:                0.796659
  current training accuracy:            0.664389
Seen 507000 samples, time used: 13081.048s
  current training loss:                0.796165
  current training accuracy:            0.664679
Seen 510000 samples, time used: 13160.527s
  current training loss:                0.795756
  current training accuracy:            0.664900
saving to ..., time used 794.591s
Seen 513000 samples, time used: 13242.257s
  current training loss:                0.795261
  current training accuracy:            0.665232
Seen 516000 samples, time used: 13320.550s
  current training loss:                0.794617
  current training accuracy:            0.665593
Seen 519000 samples, time used: 13400.905s
  current training loss:                0.794118
  current training accuracy:            0.665940
Seen 522000 samples, time used: 13480.283s
  current training loss:                0.793723
  current training accuracy:            0.666161
Seen 525000 samples, time used: 13559.521s
  current training loss:                0.793136
  current training accuracy:            0.666550
Seen 528000 samples, time used: 13638.899s
  current training loss:                0.792717
  current training accuracy:            0.666758
Seen 531000 samples, time used: 13717.249s
  current training loss:                0.792280
  current training accuracy:            0.667000
Seen 534000 samples, time used: 13796.649s
  current training loss:                0.791761
  current training accuracy:            0.667272
Seen 537000 samples, time used: 13875.270s
  current training loss:                0.791341
  current training accuracy:            0.667508
Seen 540000 samples, time used: 13956.190s
  current training loss:                0.790862
  current training accuracy:            0.667804
saving to ..., time used 795.099s
Seen 543000 samples, time used: 14035.616s
  current training loss:                0.790385
  current training accuracy:            0.668125
Seen 546000 samples, time used: 14114.374s
  current training loss:                0.790070
  current training accuracy:            0.668361
Seen 549000 samples, time used: 14193.427s
  current training loss:                0.789635
  current training accuracy:            0.668650
Epoch 1 of 20 took 14225.506s
  training loss:                0.789572
  training accuracy:            66.87 %
  validation loss:              0.656260
  validation accuracy:          72.55 %
  test loss:                    0.650373
  test accuracy:                73.16 %
saving to ./snli/word_by_word_model_epoch1.npz
Seen 3000 samples, time used: 79.568s
  current training loss:                0.679066
  current training accuracy:            0.739000
Seen 6000 samples, time used: 159.031s
  current training loss:                0.675284
  current training accuracy:            0.744333
Seen 9000 samples, time used: 236.402s
  current training loss:                0.677945
  current training accuracy:            0.741000
Seen 12000 samples, time used: 315.357s
  current training loss:                0.681474
  current training accuracy:            0.739000
Seen 15000 samples, time used: 393.435s
  current training loss:                0.682879
  current training accuracy:            0.736467
Seen 18000 samples, time used: 470.952s
  current training loss:                0.683248
  current training accuracy:            0.737444
Seen 21000 samples, time used: 550.337s
  current training loss:                0.681935
  current training accuracy:            0.736048
Seen 24000 samples, time used: 628.854s
  current training loss:                0.680633
  current training accuracy:            0.736042
Seen 27000 samples, time used: 708.434s
  current training loss:                0.682093
  current training accuracy:            0.735481
Seen 30000 samples, time used: 787.643s
  current training loss:                0.682300
  current training accuracy:            0.735133
saving to ..., time used 787.643s
Seen 33000 samples, time used: 870.331s
  current training loss:                0.681044
  current training accuracy:            0.735576
Seen 36000 samples, time used: 952.542s
  current training loss:                0.681258
  current training accuracy:            0.735472
Seen 39000 samples, time used: 1034.914s
  current training loss:                0.682185
  current training accuracy:            0.735051
Seen 42000 samples, time used: 1114.677s
  current training loss:                0.681508
  current training accuracy:            0.735381
Seen 45000 samples, time used: 1194.146s
  current training loss:                0.680435
  current training accuracy:            0.735156
Seen 48000 samples, time used: 1275.580s
  current training loss:                0.680002
  current training accuracy:            0.735271
Seen 51000 samples, time used: 1356.721s
  current training loss:                0.679898
  current training accuracy:            0.735686
Seen 54000 samples, time used: 1437.873s
  current training loss:                0.680299
  current training accuracy:            0.735222
Seen 57000 samples, time used: 1517.596s
  current training loss:                0.679055
  current training accuracy:            0.735702
Seen 60000 samples, time used: 1599.790s
  current training loss:                0.679128
  current training accuracy:            0.735517
saving to ..., time used 811.651s
Seen 63000 samples, time used: 1679.250s
  current training loss:                0.679130
  current training accuracy:            0.735333
Seen 66000 samples, time used: 1755.508s
  current training loss:                0.678999
  current training accuracy:            0.735167
Seen 69000 samples, time used: 1835.625s
  current training loss:                0.678349
  current training accuracy:            0.735290
Seen 72000 samples, time used: 1916.657s
  current training loss:                0.677087
  current training accuracy:            0.735931
Seen 75000 samples, time used: 1995.637s
  current training loss:                0.676646
  current training accuracy:            0.735973
Seen 78000 samples, time used: 2074.104s
  current training loss:                0.676273
  current training accuracy:            0.736167
Seen 81000 samples, time used: 2154.345s
  current training loss:                0.675894
  current training accuracy:            0.736136
Seen 84000 samples, time used: 2232.449s
  current training loss:                0.676019
  current training accuracy:            0.736119
Seen 87000 samples, time used: 2310.719s
  current training loss:                0.675159
  current training accuracy:            0.736391
Seen 90000 samples, time used: 2389.234s
  current training loss:                0.675206
  current training accuracy:            0.736567
saving to ..., time used 788.962s
Seen 93000 samples, time used: 2471.337s
  current training loss:                0.675138
  current training accuracy:            0.736570
Seen 96000 samples, time used: 2552.531s
  current training loss:                0.674903
  current training accuracy:            0.736563
Seen 99000 samples, time used: 2632.130s
  current training loss:                0.674511
  current training accuracy:            0.736859
Seen 102000 samples, time used: 2713.731s
  current training loss:                0.673197
  current training accuracy:            0.737343
Seen 105000 samples, time used: 2795.655s
  current training loss:                0.673563
  current training accuracy:            0.737410
Seen 108000 samples, time used: 2877.251s
  current training loss:                0.674193
  current training accuracy:            0.736991
Seen 111000 samples, time used: 2955.307s
  current training loss:                0.674506
  current training accuracy:            0.736892
Seen 114000 samples, time used: 3036.044s
  current training loss:                0.673844
  current training accuracy:            0.737211
Seen 117000 samples, time used: 3116.423s
  current training loss:                0.674136
  current training accuracy:            0.736991
Seen 120000 samples, time used: 3198.656s
  current training loss:                0.674111
  current training accuracy:            0.736917
saving to ..., time used 808.864s
Seen 123000 samples, time used: 3278.335s
  current training loss:                0.674009
  current training accuracy:            0.736894
Seen 126000 samples, time used: 3358.683s
  current training loss:                0.673946
  current training accuracy:            0.737024
Seen 129000 samples, time used: 3438.598s
  current training loss:                0.674182
  current training accuracy:            0.737240
Seen 132000 samples, time used: 3518.821s
  current training loss:                0.674068
  current training accuracy:            0.737432
Seen 135000 samples, time used: 3597.767s
  current training loss:                0.673723
  current training accuracy:            0.737585
Seen 138000 samples, time used: 3677.113s
  current training loss:                0.674091
  current training accuracy:            0.737384
Seen 141000 samples, time used: 3756.337s
  current training loss:                0.673969
  current training accuracy:            0.737113
Seen 144000 samples, time used: 3836.030s
  current training loss:                0.673851
  current training accuracy:            0.737201
Seen 147000 samples, time used: 3916.255s
  current training loss:                0.674049
  current training accuracy:            0.737000
Seen 150000 samples, time used: 3994.687s
  current training loss:                0.673459
  current training accuracy:            0.737233
saving to ..., time used 795.478s
Seen 153000 samples, time used: 4074.245s
  current training loss:                0.673184
  current training accuracy:            0.737301
Seen 156000 samples, time used: 4152.980s
  current training loss:                0.672790
  current training accuracy:            0.737462
Seen 159000 samples, time used: 4233.650s
  current training loss:                0.671863
  current training accuracy:            0.737887
Seen 162000 samples, time used: 4311.858s
  current training loss:                0.671619
  current training accuracy:            0.738117
Seen 165000 samples, time used: 4391.394s
  current training loss:                0.671677
  current training accuracy:            0.738079
Seen 168000 samples, time used: 4471.372s
  current training loss:                0.671735
  current training accuracy:            0.737946
Seen 171000 samples, time used: 4548.626s
  current training loss:                0.671749
  current training accuracy:            0.737842
Seen 174000 samples, time used: 4628.015s
  current training loss:                0.671822
  current training accuracy:            0.737805
Seen 177000 samples, time used: 4708.444s
  current training loss:                0.671405
  current training accuracy:            0.737977
Seen 180000 samples, time used: 4788.299s
  current training loss:                0.671739
  current training accuracy:            0.737872
saving to ..., time used 793.143s
Seen 183000 samples, time used: 4867.838s
  current training loss:                0.671602
  current training accuracy:            0.737831
Seen 186000 samples, time used: 4946.710s
  current training loss:                0.671485
  current training accuracy:            0.737844
Seen 189000 samples, time used: 5027.273s
  current training loss:                0.671437
  current training accuracy:            0.737952
Seen 192000 samples, time used: 5105.786s
  current training loss:                0.671488
  current training accuracy:            0.737823
Seen 195000 samples, time used: 5185.052s
  current training loss:                0.671580
  current training accuracy:            0.737728
Seen 198000 samples, time used: 5264.344s
  current training loss:                0.671214
  current training accuracy:            0.737722
Seen 201000 samples, time used: 5342.224s
  current training loss:                0.670890
  current training accuracy:            0.737985
Seen 204000 samples, time used: 5420.703s
  current training loss:                0.670605
  current training accuracy:            0.738113
Seen 207000 samples, time used: 5500.098s
  current training loss:                0.670796
  current training accuracy:            0.737884
Seen 210000 samples, time used: 5578.501s
  current training loss:                0.670460
  current training accuracy:            0.737895
saving to ..., time used 789.663s
Seen 213000 samples, time used: 5658.742s
  current training loss:                0.670216
  current training accuracy:            0.738019
Seen 216000 samples, time used: 5737.475s
  current training loss:                0.669811
  current training accuracy:            0.738306
Seen 219000 samples, time used: 5816.029s
  current training loss:                0.669517
  current training accuracy:            0.738475
Seen 222000 samples, time used: 5895.554s
  current training loss:                0.669805
  current training accuracy:            0.738383
Seen 225000 samples, time used: 5972.856s
  current training loss:                0.669620
  current training accuracy:            0.738520
Seen 228000 samples, time used: 6053.021s
  current training loss:                0.669600
  current training accuracy:            0.738583
Seen 231000 samples, time used: 6131.128s
  current training loss:                0.669327
  current training accuracy:            0.738771
Seen 234000 samples, time used: 6211.280s
  current training loss:                0.669020
  current training accuracy:            0.738868
Seen 237000 samples, time used: 6290.698s
  current training loss:                0.668841
  current training accuracy:            0.738958
Seen 240000 samples, time used: 6370.327s
  current training loss:                0.668817
  current training accuracy:            0.738900
saving to ..., time used 790.844s
Seen 243000 samples, time used: 6449.738s
  current training loss:                0.668865
  current training accuracy:            0.738819
Seen 246000 samples, time used: 6528.201s
  current training loss:                0.668542
  current training accuracy:            0.738980
Seen 249000 samples, time used: 6606.957s
  current training loss:                0.668114
  current training accuracy:            0.739221
Seen 252000 samples, time used: 6685.405s
  current training loss:                0.667811
  current training accuracy:            0.739325
Seen 255000 samples, time used: 6765.476s
  current training loss:                0.667699
  current training accuracy:            0.739451
Seen 258000 samples, time used: 6844.847s
  current training loss:                0.667585
  current training accuracy:            0.739578
Seen 261000 samples, time used: 6924.023s
  current training loss:                0.667282
  current training accuracy:            0.739670
Seen 264000 samples, time used: 7003.112s
  current training loss:                0.666964
  current training accuracy:            0.739811
Seen 267000 samples, time used: 7081.713s
  current training loss:                0.666729
  current training accuracy:            0.739798
Seen 270000 samples, time used: 7160.041s
  current training loss:                0.666610
  current training accuracy:            0.739889
saving to ..., time used 789.149s
Seen 273000 samples, time used: 7240.037s
  current training loss:                0.666388
  current training accuracy:            0.739978
Seen 276000 samples, time used: 7320.695s
  current training loss:                0.666294
  current training accuracy:            0.740029
Seen 279000 samples, time used: 7401.538s
  current training loss:                0.665855
  current training accuracy:            0.740297
Seen 282000 samples, time used: 7479.971s
  current training loss:                0.665652
  current training accuracy:            0.740383
Seen 285000 samples, time used: 7560.335s
  current training loss:                0.665551
  current training accuracy:            0.740418
Seen 288000 samples, time used: 7638.032s
  current training loss:                0.665180
  current training accuracy:            0.740604
Seen 291000 samples, time used: 7716.945s
  current training loss:                0.665218
  current training accuracy:            0.740680
Seen 294000 samples, time used: 7794.173s
  current training loss:                0.664837
  current training accuracy:            0.740918
Seen 297000 samples, time used: 7874.521s
  current training loss:                0.664435
  current training accuracy:            0.741121
Seen 300000 samples, time used: 7953.223s
  current training loss:                0.664275
  current training accuracy:            0.741173
saving to ..., time used 792.624s
Seen 303000 samples, time used: 8033.258s
  current training loss:                0.664025
  current training accuracy:            0.741363
Seen 306000 samples, time used: 8112.643s
  current training loss:                0.663911
  current training accuracy:            0.741389
Seen 309000 samples, time used: 8192.786s
  current training loss:                0.663691
  current training accuracy:            0.741476
Seen 312000 samples, time used: 8271.426s
  current training loss:                0.663433
  current training accuracy:            0.741593
Seen 315000 samples, time used: 8350.894s
  current training loss:                0.663261
  current training accuracy:            0.741781
Seen 318000 samples, time used: 8431.060s
  current training loss:                0.663058
  current training accuracy:            0.741855
Seen 321000 samples, time used: 8510.702s
  current training loss:                0.663136
  current training accuracy:            0.741798
Seen 324000 samples, time used: 8588.399s
  current training loss:                0.662876
  current training accuracy:            0.741880
Seen 327000 samples, time used: 8668.020s
  current training loss:                0.662757
  current training accuracy:            0.742034
Seen 330000 samples, time used: 8747.261s
  current training loss:                0.662413
  current training accuracy:            0.742161
saving to ..., time used 793.505s
Seen 333000 samples, time used: 8827.066s
  current training loss:                0.662232
  current training accuracy:            0.742201
Seen 336000 samples, time used: 8905.729s
  current training loss:                0.661969
  current training accuracy:            0.742345
Seen 339000 samples, time used: 8986.041s
  current training loss:                0.661742
  current training accuracy:            0.742496
Seen 342000 samples, time used: 9064.561s
  current training loss:                0.661345
  current training accuracy:            0.742643
Seen 345000 samples, time used: 9143.159s
  current training loss:                0.661188
  current training accuracy:            0.742768
Seen 348000 samples, time used: 9221.981s
  current training loss:                0.660852
  current training accuracy:            0.742836
Seen 351000 samples, time used: 9301.851s
  current training loss:                0.660750
  current training accuracy:            0.742929
Seen 354000 samples, time used: 9381.739s
  current training loss:                0.660419
  current training accuracy:            0.743048
Seen 357000 samples, time used: 9460.131s
  current training loss:                0.660370
  current training accuracy:            0.743073
Seen 360000 samples, time used: 9537.823s
  current training loss:                0.660338
  current training accuracy:            0.743108
saving to ..., time used 789.962s
Seen 363000 samples, time used: 9618.025s
  current training loss:                0.660220
  current training accuracy:            0.743248
Seen 366000 samples, time used: 9698.343s
  current training loss:                0.660106
  current training accuracy:            0.743205
Seen 369000 samples, time used: 9778.935s
  current training loss:                0.659805
  current training accuracy:            0.743336
Seen 372000 samples, time used: 9859.925s
  current training loss:                0.659593
  current training accuracy:            0.743444
Seen 375000 samples, time used: 9940.859s
  current training loss:                0.659508
  current training accuracy:            0.743515
Seen 378000 samples, time used: 10019.268s
  current training loss:                0.659220
  current training accuracy:            0.743643
Seen 381000 samples, time used: 10098.467s
  current training loss:                0.658878
  current training accuracy:            0.743801
Seen 384000 samples, time used: 10179.361s
  current training loss:                0.658621
  current training accuracy:            0.743948
Seen 387000 samples, time used: 10258.375s
  current training loss:                0.658518
  current training accuracy:            0.744028
Seen 390000 samples, time used: 10337.881s
  current training loss:                0.658467
  current training accuracy:            0.743974
saving to ..., time used 799.322s
Seen 393000 samples, time used: 10419.220s
  current training loss:                0.658329
  current training accuracy:            0.744041
Seen 396000 samples, time used: 10497.406s
  current training loss:                0.657969
  current training accuracy:            0.744222
Seen 399000 samples, time used: 10575.599s
  current training loss:                0.657871
  current training accuracy:            0.744346
Seen 402000 samples, time used: 10653.284s
  current training loss:                0.657564
  current training accuracy:            0.744525
Seen 405000 samples, time used: 10732.802s
  current training loss:                0.657571
  current training accuracy:            0.744519
Seen 408000 samples, time used: 10814.239s
  current training loss:                0.657293
  current training accuracy:            0.744713
Seen 411000 samples, time used: 10893.475s
  current training loss:                0.657085
  current training accuracy:            0.744815
Seen 414000 samples, time used: 10971.578s
  current training loss:                0.656991
  current training accuracy:            0.744935
Seen 417000 samples, time used: 11051.355s
  current training loss:                0.656879
  current training accuracy:            0.745031
Seen 420000 samples, time used: 11130.019s
  current training loss:                0.656770
  current training accuracy:            0.745079
saving to ..., time used 791.514s
Seen 423000 samples, time used: 11210.306s
  current training loss:                0.656611
  current training accuracy:            0.745191
Seen 426000 samples, time used: 11287.832s
  current training loss:                0.656447
  current training accuracy:            0.745310
Seen 429000 samples, time used: 11367.574s
  current training loss:                0.656318
  current training accuracy:            0.745406
Seen 432000 samples, time used: 11446.826s
  current training loss:                0.656205
  current training accuracy:            0.745468
Seen 435000 samples, time used: 11524.151s
  current training loss:                0.655825
  current training accuracy:            0.745676

Seen 438000 samples, time used: 11603.116s
  current training loss:                0.655765
  current training accuracy:            0.745742
Seen 441000 samples, time used: 11683.159s
  current training loss:                0.655646
  current training accuracy:            0.745794
Seen 444000 samples, time used: 11763.190s
  current training loss:                0.655589
  current training accuracy:            0.745890
Seen 447000 samples, time used: 11841.716s
  current training loss:                0.655414
  current training accuracy:            0.745957
Seen 450000 samples, time used: 11921.482s
  current training loss:                0.655191
  current training accuracy:            0.746100
saving to ..., time used 790.942s
Seen 453000 samples, time used: 12003.338s
  current training loss:                0.655024
  current training accuracy:            0.746141
Seen 456000 samples, time used: 12085.556s
  current training loss:                0.654957
  current training accuracy:            0.746237
Seen 459000 samples, time used: 12168.868s
  current training loss:                0.654808
  current training accuracy:            0.746305
Seen 462000 samples, time used: 12252.220s
  current training loss:                0.654821
  current training accuracy:            0.746316
Seen 465000 samples, time used: 12333.782s
  current training loss:                0.654885
  current training accuracy:            0.746318
Seen 468000 samples, time used: 12413.774s
  current training loss:                0.654652
  current training accuracy:            0.746470
Seen 471000 samples, time used: 12496.429s
  current training loss:                0.654512
  current training accuracy:            0.746558
Seen 474000 samples, time used: 12578.726s
  current training loss:                0.654457
  current training accuracy:            0.746580
Seen 477000 samples, time used: 12658.533s
  current training loss:                0.654224
  current training accuracy:            0.746669
Seen 480000 samples, time used: 12740.582s
  current training loss:                0.654187
  current training accuracy:            0.746671
saving to ..., time used 818.585s
Seen 483000 samples, time used: 12820.775s
  current training loss:                0.654223
  current training accuracy:            0.746660
Seen 486000 samples, time used: 12899.075s
  current training loss:                0.654229
  current training accuracy:            0.746663
Seen 489000 samples, time used: 12979.187s
  current training loss:                0.654024
  current training accuracy:            0.746744
Seen 492000 samples, time used: 13058.066s
  current training loss:                0.653740
  current training accuracy:            0.746894
Seen 495000 samples, time used: 13138.391s
  current training loss:                0.653614
  current training accuracy:            0.746867
Seen 498000 samples, time used: 13218.354s
  current training loss:                0.653548
  current training accuracy:            0.746942
Seen 501000 samples, time used: 13298.082s
  current training loss:                0.653495
  current training accuracy:            0.747002
Seen 504000 samples, time used: 13378.359s
  current training loss:                0.653252
  current training accuracy:            0.747109
Seen 507000 samples, time used: 13457.281s
  current training loss:                0.653194
  current training accuracy:            0.747150
Seen 510000 samples, time used: 13537.812s
  current training loss:                0.653133
  current training accuracy:            0.747214
saving to ..., time used 796.680s
Seen 513000 samples, time used: 13618.300s
  current training loss:                0.652940
  current training accuracy:            0.747304
Seen 516000 samples, time used: 13698.816s
  current training loss:                0.652855
  current training accuracy:            0.747337
Seen 519000 samples, time used: 13778.268s
  current training loss:                0.652837
  current training accuracy:            0.747380
Seen 522000 samples, time used: 13857.479s
  current training loss:                0.652753
  current training accuracy:            0.747374
Seen 525000 samples, time used: 13937.977s
  current training loss:                0.652573
  current training accuracy:            0.747457
Seen 528000 samples, time used: 14016.796s
  current training loss:                0.652353
  current training accuracy:            0.747591
Seen 531000 samples, time used: 14097.838s
  current training loss:                0.652286
  current training accuracy:            0.747616
Seen 534000 samples, time used: 14176.669s
  current training loss:                0.652141
  current training accuracy:            0.747663
Seen 537000 samples, time used: 14256.989s
  current training loss:                0.652008
  current training accuracy:            0.747769
Seen 540000 samples, time used: 14337.750s
  current training loss:                0.651951
  current training accuracy:            0.747804
saving to ..., time used 799.404s
Seen 543000 samples, time used: 14416.660s
  current training loss:                0.651835
  current training accuracy:            0.747877
Seen 546000 samples, time used: 14496.274s
  current training loss:                0.651733
  current training accuracy:            0.747912
Seen 549000 samples, time used: 14576.741s
  current training loss:                0.651544
  current training accuracy:            0.748011
Epoch 2 of 20 took 14609.076s
  training loss:                0.651503
  training accuracy:            74.81 %
  validation loss:              0.571310
  validation accuracy:          76.90 %
  test loss:                    0.574589
  test accuracy:                76.96 %
saving to ./snli/word_by_word_model_epoch2.npz
Seen 3000 samples, time used: 79.029s
  current training loss:                0.588799
  current training accuracy:            0.780000
Seen 6000 samples, time used: 158.764s
  current training loss:                0.583775
  current training accuracy:            0.782167
Seen 9000 samples, time used: 238.828s
  current training loss:                0.586055
  current training accuracy:            0.780222
Seen 12000 samples, time used: 318.073s
  current training loss:                0.588856
  current training accuracy:            0.781083
Seen 15000 samples, time used: 397.121s
  current training loss:                0.590528
  current training accuracy:            0.781267
Seen 18000 samples, time used: 477.280s
  current training loss:                0.591896
  current training accuracy:            0.779667
Seen 21000 samples, time used: 558.239s
  current training loss:                0.592869
  current training accuracy:            0.779714
Seen 24000 samples, time used: 637.706s
  current training loss:                0.594100
  current training accuracy:            0.778958
Seen 27000 samples, time used: 717.121s
  current training loss:                0.597962
  current training accuracy:            0.776778
Seen 30000 samples, time used: 795.150s
  current training loss:                0.598920
  current training accuracy:            0.775800
saving to ..., time used 795.151s
Seen 33000 samples, time used: 876.442s
  current training loss:                0.597972
  current training accuracy:            0.775758
Seen 36000 samples, time used: 954.184s
  current training loss:                0.598273
  current training accuracy:            0.775861
Seen 39000 samples, time used: 1031.676s
  current training loss:                0.597853
  current training accuracy:            0.775897
Seen 42000 samples, time used: 1111.134s
  current training loss:                0.597214
  current training accuracy:            0.776238
Seen 45000 samples, time used: 1190.612s
  current training loss:                0.598151
  current training accuracy:            0.775622
Seen 48000 samples, time used: 1269.492s
  current training loss:                0.598464
  current training accuracy:            0.775521
Seen 51000 samples, time used: 1349.345s
  current training loss:                0.598868
  current training accuracy:            0.775431
Seen 54000 samples, time used: 1431.694s
  current training loss:                0.598162
  current training accuracy:            0.775963
Seen 57000 samples, time used: 1510.850s
  current training loss:                0.597642
  current training accuracy:            0.776456
Seen 60000 samples, time used: 1592.490s
  current training loss:                0.598470
  current training accuracy:            0.776167
saving to ..., time used 796.826s
Seen 63000 samples, time used: 1673.624s
  current training loss:                0.598159
  current training accuracy:            0.776619
Seen 66000 samples, time used: 1754.903s
  current training loss:                0.599098
  current training accuracy:            0.776394
Seen 69000 samples, time used: 1835.211s
  current training loss:                0.599033
  current training accuracy:            0.776391
Seen 72000 samples, time used: 1913.680s
  current training loss:                0.599602
  current training accuracy:            0.776014
Seen 75000 samples, time used: 1993.112s
  current training loss:                0.599588
  current training accuracy:            0.776307
Seen 78000 samples, time used: 2071.457s
  current training loss:                0.600977
  current training accuracy:            0.775513
Seen 81000 samples, time used: 2150.273s
  current training loss:                0.600171
  current training accuracy:            0.775691
Seen 84000 samples, time used: 2230.742s
  current training loss:                0.600135
  current training accuracy:            0.775786
Seen 87000 samples, time used: 2310.958s
  current training loss:                0.600191
  current training accuracy:            0.775667
Seen 90000 samples, time used: 2389.446s
  current training loss:                0.600575
  current training accuracy:            0.775478
saving to ..., time used 796.454s
Seen 93000 samples, time used: 2468.097s
  current training loss:                0.600689
  current training accuracy:            0.775269
Seen 96000 samples, time used: 2547.844s
  current training loss:                0.600066
  current training accuracy:            0.775469
Seen 99000 samples, time used: 2628.526s
  current training loss:                0.600025
  current training accuracy:            0.775404
Seen 102000 samples, time used: 2708.228s
  current training loss:                0.599871
  current training accuracy:            0.775686
Seen 105000 samples, time used: 2787.417s
  current training loss:                0.599497
  current training accuracy:            0.775990
Seen 108000 samples, time used: 2868.594s
  current training loss:                0.599264
  current training accuracy:            0.775843
Seen 111000 samples, time used: 2949.123s
  current training loss:                0.599963
  current training accuracy:            0.775351
Seen 114000 samples, time used: 3028.265s
  current training loss:                0.600145
  current training accuracy:            0.775175
Seen 117000 samples, time used: 3108.778s
  current training loss:                0.600044
  current training accuracy:            0.775325
Seen 120000 samples, time used: 3189.174s
  current training loss:                0.599600
  current training accuracy:            0.775642
saving to ..., time used 799.247s
Seen 123000 samples, time used: 3271.499s
  current training loss:                0.600211
  current training accuracy:            0.775472
Seen 126000 samples, time used: 3348.971s
  current training loss:                0.600177
  current training accuracy:            0.775333
Seen 129000 samples, time used: 3427.537s
  current training loss:                0.600195
  current training accuracy:            0.775171
Seen 132000 samples, time used: 3506.375s
  current training loss:                0.599552
  current training accuracy:            0.775235
Seen 135000 samples, time used: 3585.693s
  current training loss:                0.600006
  current training accuracy:            0.774941
Seen 138000 samples, time used: 3664.995s
  current training loss:                0.600572
  current training accuracy:            0.774688
Seen 141000 samples, time used: 3744.300s
  current training loss:                0.600271
  current training accuracy:            0.774603
Seen 144000 samples, time used: 3825.513s
  current training loss:                0.600076
  current training accuracy:            0.774653
Seen 147000 samples, time used: 3903.888s
  current training loss:                0.600282
  current training accuracy:            0.774565
Seen 150000 samples, time used: 3983.817s
  current training loss:                0.600470
  current training accuracy:            0.774687
saving to ..., time used 794.095s
Seen 153000 samples, time used: 4064.053s
  current training loss:                0.600213
  current training accuracy:            0.774882
Seen 156000 samples, time used: 4145.541s
  current training loss:                0.599981
  current training accuracy:            0.775000
Seen 159000 samples, time used: 4224.632s
  current training loss:                0.600138
  current training accuracy:            0.775019
Seen 162000 samples, time used: 4304.741s
  current training loss:                0.599680
  current training accuracy:            0.775290
Seen 165000 samples, time used: 4382.918s
  current training loss:                0.599451
  current training accuracy:            0.775442
Seen 168000 samples, time used: 4461.755s
  current training loss:                0.599437
  current training accuracy:            0.775458
Seen 171000 samples, time used: 4541.764s
  current training loss:                0.599261
  current training accuracy:            0.775421
Seen 174000 samples, time used: 4621.149s
  current training loss:                0.599132
  current training accuracy:            0.775644
Seen 177000 samples, time used: 4699.281s
  current training loss:                0.599156
  current training accuracy:            0.775667
Seen 180000 samples, time used: 4778.260s
  current training loss:                0.599424
  current training accuracy:            0.775372
saving to ..., time used 793.955s
Seen 183000 samples, time used: 4859.127s
  current training loss:                0.599551
  current training accuracy:            0.775317
Seen 186000 samples, time used: 4938.634s
  current training loss:                0.599918
  current training accuracy:            0.775194
Seen 189000 samples, time used: 5018.912s
  current training loss:                0.599326
  current training accuracy:            0.775466
Seen 192000 samples, time used: 5100.538s
  current training loss:                0.599599
  current training accuracy:            0.775344
Seen 195000 samples, time used: 5182.800s
  current training loss:                0.599968
  current training accuracy:            0.775313
Seen 198000 samples, time used: 5263.082s
  current training loss:                0.599828
  current training accuracy:            0.775318
Seen 201000 samples, time used: 5341.417s
  current training loss:                0.599560
  current training accuracy:            0.775493
Seen 204000 samples, time used: 5421.667s
  current training loss:                0.599416
  current training accuracy:            0.775588
Seen 207000 samples, time used: 5500.690s
  current training loss:                0.599242
  current training accuracy:            0.775585
Seen 210000 samples, time used: 5579.111s
  current training loss:                0.599009
  current training accuracy:            0.775838
saving to ..., time used 800.361s
Seen 213000 samples, time used: 5658.751s
  current training loss:                0.598903
  current training accuracy:            0.775869
Seen 216000 samples, time used: 5739.046s
  current training loss:                0.598820
  current training accuracy:            0.775787
Seen 219000 samples, time used: 5818.367s
  current training loss:                0.598722
  current training accuracy:            0.775817
Seen 222000 samples, time used: 5897.846s
  current training loss:                0.598883
  current training accuracy:            0.775757
Seen 225000 samples, time used: 5976.605s
  current training loss:                0.599035
  current training accuracy:            0.775662
Seen 228000 samples, time used: 6055.656s
  current training loss:                0.599023
  current training accuracy:            0.775719
Seen 231000 samples, time used: 6133.523s
  current training loss:                0.599069
  current training accuracy:            0.775697
Seen 234000 samples, time used: 6212.087s
  current training loss:                0.598941
  current training accuracy:            0.775620
Seen 237000 samples, time used: 6290.793s
  current training loss:                0.598842
  current training accuracy:            0.775654
Seen 240000 samples, time used: 6369.782s
  current training loss:                0.598699
  current training accuracy:            0.775683
saving to ..., time used 790.178s

Seen 243000 samples, time used: 6448.134s
  current training loss:                0.598506
  current training accuracy:            0.775650
Seen 246000 samples, time used: 6528.126s
  current training loss:                0.598372
  current training accuracy:            0.775699
Seen 249000 samples, time used: 6605.333s
  current training loss:                0.598292
  current training accuracy:            0.775711
Seen 252000 samples, time used: 6683.549s
  current training loss:                0.598357
  current training accuracy:            0.775552
Seen 255000 samples, time used: 6761.873s
  current training loss:                0.598141
  current training accuracy:            0.775694
Seen 258000 samples, time used: 6841.849s
  current training loss:                0.598135
  current training accuracy:            0.775705
Seen 261000 samples, time used: 6921.897s
  current training loss:                0.598098
  current training accuracy:            0.775728
Seen 264000 samples, time used: 6999.098s
  current training loss:                0.598071
  current training accuracy:            0.775723
Seen 267000 samples, time used: 7078.883s
  current training loss:                0.597996
  current training accuracy:            0.775753
Seen 270000 samples, time used: 7157.305s
  current training loss:                0.597791
  current training accuracy:            0.775933
saving to ..., time used 787.056s
Seen 273000 samples, time used: 7236.360s
  current training loss:                0.597741
  current training accuracy:            0.775967
Seen 276000 samples, time used: 7315.281s
  current training loss:                0.597959
  current training accuracy:            0.775844
Seen 279000 samples, time used: 7394.169s
  current training loss:                0.598004
  current training accuracy:            0.775781
Seen 282000 samples, time used: 7475.436s
  current training loss:                0.598247
  current training accuracy:            0.775660
Seen 285000 samples, time used: 7555.085s
  current training loss:                0.598080
  current training accuracy:            0.775726
Seen 288000 samples, time used: 7635.248s
  current training loss:                0.597729
  current training accuracy:            0.775899
Seen 291000 samples, time used: 7713.599s
  current training loss:                0.597830
  current training accuracy:            0.775918
Seen 294000 samples, time used: 7794.203s
  current training loss:                0.597719
  current training accuracy:            0.775939
Seen 297000 samples, time used: 7871.680s
  current training loss:                0.597856
  current training accuracy:            0.775862
Seen 300000 samples, time used: 7951.878s
  current training loss:                0.597710
  current training accuracy:            0.775977
saving to ..., time used 794.083s
Seen 303000 samples, time used: 8030.715s
  current training loss:                0.597485
  current training accuracy:            0.776125
Seen 306000 samples, time used: 8109.840s
  current training loss:                0.597306
  current training accuracy:            0.776193
Seen 309000 samples, time used: 8189.620s
  current training loss:                0.597266
  current training accuracy:            0.776152
Seen 312000 samples, time used: 8268.512s
  current training loss:                0.597508
  current training accuracy:            0.775968
Seen 315000 samples, time used: 8347.011s
  current training loss:                0.597390
  current training accuracy:            0.776070
Seen 318000 samples, time used: 8426.172s
  current training loss:                0.597274
  current training accuracy:            0.776123
Seen 321000 samples, time used: 8505.752s
  current training loss:                0.597201
  current training accuracy:            0.776115
Seen 324000 samples, time used: 8586.239s
  current training loss:                0.597253
  current training accuracy:            0.776074
Seen 327000 samples, time used: 8665.927s
  current training loss:                0.597152
  current training accuracy:            0.776156
Seen 330000 samples, time used: 8744.543s
  current training loss:                0.596822
  current training accuracy:            0.776327
saving to ..., time used 792.179s
Seen 333000 samples, time used: 8825.201s
  current training loss:                0.596734
  current training accuracy:            0.776372
Seen 336000 samples, time used: 8904.318s
  current training loss:                0.596720
  current training accuracy:            0.776357
Seen 339000 samples, time used: 8983.016s
  current training loss:                0.596473
  current training accuracy:            0.776463
Seen 342000 samples, time used: 9060.974s
  current training loss:                0.596506
  current training accuracy:            0.776406
Seen 345000 samples, time used: 9139.600s
  current training loss:                0.596347
  current training accuracy:            0.776507
Seen 348000 samples, time used: 9219.134s
  current training loss:                0.596177
  current training accuracy:            0.776526
Seen 351000 samples, time used: 9297.790s
  current training loss:                0.596139
  current training accuracy:            0.776570
Seen 354000 samples, time used: 9377.361s
  current training loss:                0.596162
  current training accuracy:            0.776568
Seen 357000 samples, time used: 9457.822s
  current training loss:                0.596348
  current training accuracy:            0.776538
Seen 360000 samples, time used: 9536.635s
  current training loss:                0.596006
  current training accuracy:            0.776711
saving to ..., time used 791.605s
Seen 363000 samples, time used: 9617.801s
  current training loss:                0.596002
  current training accuracy:            0.776824
Seen 366000 samples, time used: 9697.672s
  current training loss:                0.596004
  current training accuracy:            0.776923
Seen 369000 samples, time used: 9777.569s
  current training loss:                0.595925
  current training accuracy:            0.776976
Seen 372000 samples, time used: 9856.830s
  current training loss:                0.595755
  current training accuracy:            0.777038
Seen 375000 samples, time used: 9936.816s
  current training loss:                0.595770
  current training accuracy:            0.777067
Seen 378000 samples, time used: 10015.983s
  current training loss:                0.595591
  current training accuracy:            0.777206
Seen 381000 samples, time used: 10096.601s
  current training loss:                0.595478
  current training accuracy:            0.777318
Seen 384000 samples, time used: 10175.934s
  current training loss:                0.595383
  current training accuracy:            0.777445
Seen 387000 samples, time used: 10255.391s
  current training loss:                0.595298
  current training accuracy:            0.777488
Seen 390000 samples, time used: 10336.163s
  current training loss:                0.595263
  current training accuracy:            0.777515
saving to ..., time used 799.024s
Seen 393000 samples, time used: 10416.098s
  current training loss:                0.595206
  current training accuracy:            0.777514
Seen 396000 samples, time used: 10494.780s
  current training loss:                0.595181
  current training accuracy:            0.777558
Seen 399000 samples, time used: 10573.606s
  current training loss:                0.595035
  current training accuracy:            0.777639
Seen 402000 samples, time used: 10651.604s
  current training loss:                0.594963
  current training accuracy:            0.777652
Seen 405000 samples, time used: 10730.377s
  current training loss:                0.594933
  current training accuracy:            0.777681
Seen 408000 samples, time used: 10810.997s
  current training loss:                0.594831
  current training accuracy:            0.777699
Seen 411000 samples, time used: 10889.662s
  current training loss:                0.594826
  current training accuracy:            0.777718
Seen 414000 samples, time used: 10968.100s
  current training loss:                0.594824
  current training accuracy:            0.777662
Seen 417000 samples, time used: 11047.855s
  current training loss:                0.594862
  current training accuracy:            0.777640
Seen 420000 samples, time used: 11126.195s
  current training loss:                0.594792
  current training accuracy:            0.777693
saving to ..., time used 789.477s
Seen 423000 samples, time used: 11206.603s
  current training loss:                0.594638
  current training accuracy:            0.777735
Seen 426000 samples, time used: 11286.893s
  current training loss:                0.594583
  current training accuracy:            0.777779
Seen 429000 samples, time used: 11366.771s
  current training loss:                0.594540
  current training accuracy:            0.777767
Seen 432000 samples, time used: 11445.329s
  current training loss:                0.594423
  current training accuracy:            0.777863
Seen 435000 samples, time used: 11525.699s
  current training loss:                0.594447
  current training accuracy:            0.777848
Seen 438000 samples, time used: 11604.979s
  current training loss:                0.594280
  current training accuracy:            0.777904
Seen 441000 samples, time used: 11684.418s
  current training loss:                0.594186
  current training accuracy:            0.777891
Seen 444000 samples, time used: 11761.953s
  current training loss:                0.594274
  current training accuracy:            0.777842
Seen 447000 samples, time used: 11841.524s
  current training loss:                0.594162
  current training accuracy:            0.777868
Seen 450000 samples, time used: 11920.718s
  current training loss:                0.594027
  current training accuracy:            0.777907
saving to ..., time used 793.954s
Seen 453000 samples, time used: 12000.171s
  current training loss:                0.593917
  current training accuracy:            0.777958
Seen 456000 samples, time used: 12080.672s
  current training loss:                0.593738
  current training accuracy:            0.778088
Seen 459000 samples, time used: 12163.651s
  current training loss:                0.593762
  current training accuracy:            0.778081
Seen 462000 samples, time used: 12245.604s
  current training loss:                0.593672
  current training accuracy:            0.778100
Seen 465000 samples, time used: 12326.722s
  current training loss:                0.593611
  current training accuracy:            0.778153
Seen 468000 samples, time used: 12406.507s
  current training loss:                0.593477
  current training accuracy:            0.778207
Seen 471000 samples, time used: 12489.346s
  current training loss:                0.593512
  current training accuracy:            0.778223
Seen 474000 samples, time used: 12569.970s
  current training loss:                0.593476
  current training accuracy:            0.778272
Seen 477000 samples, time used: 12651.930s
  current training loss:                0.593437
  current training accuracy:            0.778308
Seen 480000 samples, time used: 12736.027s
  current training loss:                0.593225
  current training accuracy:            0.778371
saving to ..., time used 814.796s
Seen 483000 samples, time used: 12817.020s
  current training loss:                0.593202
  current training accuracy:            0.778447
Seen 486000 samples, time used: 12896.357s
  current training loss:                0.593121
  current training accuracy:            0.778510
Seen 489000 samples, time used: 12975.757s
  current training loss:                0.593001
  current training accuracy:            0.778579
Seen 492000 samples, time used: 13055.687s
  current training loss:                0.593059
  current training accuracy:            0.778600
Seen 495000 samples, time used: 13136.130s
  current training loss:                0.592982
  current training accuracy:            0.778642
Seen 498000 samples, time used: 13216.356s
  current training loss:                0.592998
  current training accuracy:            0.778649
Seen 501000 samples, time used: 13294.564s
  current training loss:                0.592873
  current training accuracy:            0.778673
Seen 504000 samples, time used: 13372.510s
  current training loss:                0.592867
  current training accuracy:            0.778712
Seen 507000 samples, time used: 13452.353s
  current training loss:                0.592839
  current training accuracy:            0.778779
Seen 510000 samples, time used: 13532.065s
  current training loss:                0.592789
  current training accuracy:            0.778802
saving to ..., time used 795.537s
Seen 513000 samples, time used: 13613.559s
  current training loss:                0.592645
  current training accuracy:            0.778871
Seen 516000 samples, time used: 13693.810s
  current training loss:                0.592704
  current training accuracy:            0.778849
Seen 519000 samples, time used: 13772.823s
  current training loss:                0.592666
  current training accuracy:            0.778865
Seen 522000 samples, time used: 13852.399s
  current training loss:                0.592557
  current training accuracy:            0.778916
Seen 525000 samples, time used: 13933.577s
  current training loss:                0.592581
  current training accuracy:            0.778899
Seen 528000 samples, time used: 14013.921s
  current training loss:                0.592498
  current training accuracy:            0.778917
Seen 531000 samples, time used: 14092.054s
  current training loss:                0.592548
  current training accuracy:            0.778925
Seen 534000 samples, time used: 14172.685s
  current training loss:                0.592433
  current training accuracy:            0.779004
Seen 537000 samples, time used: 14251.837s
  current training loss:                0.592420
  current training accuracy:            0.779063
Seen 540000 samples, time used: 14331.350s
  current training loss:                0.592473
  current training accuracy:            0.779017
saving to ..., time used 798.787s
Seen 543000 samples, time used: 14409.934s
  current training loss:                0.592296
  current training accuracy:            0.779105
Seen 546000 samples, time used: 14491.108s
  current training loss:                0.592256
  current training accuracy:            0.779103
Seen 549000 samples, time used: 14570.131s
  current training loss:                0.592136
  current training accuracy:            0.779158
Epoch 3 of 20 took 14602.290s
  training loss:                0.592125
  training accuracy:            77.92 %
  validation loss:              0.530884
  validation accuracy:          78.88 %
  test loss:                    0.539526
  test accuracy:                78.63 %
saving to ./snli/word_by_word_model_epoch3.npz
Seen 3000 samples, time used: 78.357s
  current training loss:                0.550020
  current training accuracy:            0.790000
Seen 6000 samples, time used: 158.502s
  current training loss:                0.551978
  current training accuracy:            0.793500
Seen 9000 samples, time used: 236.963s
  current training loss:                0.554077
  current training accuracy:            0.792778
Seen 12000 samples, time used: 318.673s
  current training loss:                0.551569
  current training accuracy:            0.791917
Seen 15000 samples, time used: 397.572s
  current training loss:                0.552946
  current training accuracy:            0.793667
Seen 18000 samples, time used: 475.914s
  current training loss:                0.554686
  current training accuracy:            0.794222
Seen 21000 samples, time used: 554.946s
  current training loss:                0.555789
  current training accuracy:            0.794619
Seen 24000 samples, time used: 634.993s
  current training loss:                0.555191
  current training accuracy:            0.794708
Seen 27000 samples, time used: 714.368s
  current training loss:                0.554631
  current training accuracy:            0.794407
Seen 30000 samples, time used: 794.212s
  current training loss:                0.555280
  current training accuracy:            0.794600
saving to ..., time used 794.213s
Seen 33000 samples, time used: 873.725s
  current training loss:                0.555261
  current training accuracy:            0.794485
Seen 36000 samples, time used: 951.714s
  current training loss:                0.557247
  current training accuracy:            0.793111
Seen 39000 samples, time used: 1030.708s
  current training loss:                0.556858
  current training accuracy:            0.793590
Seen 42000 samples, time used: 1109.517s
  current training loss:                0.556750
  current training accuracy:            0.794524
Seen 45000 samples, time used: 1190.196s
  current training loss:                0.555345
  current training accuracy:            0.794933
Seen 48000 samples, time used: 1269.225s
  current training loss:                0.555834
  current training accuracy:            0.794750
Seen 51000 samples, time used: 1348.059s
  current training loss:                0.555219
  current training accuracy:            0.794882
Seen 54000 samples, time used: 1427.449s
  current training loss:                0.554880
  current training accuracy:            0.794889
Seen 57000 samples, time used: 1507.040s
  current training loss:                0.555369
  current training accuracy:            0.794947
Seen 60000 samples, time used: 1587.146s
  current training loss:                0.555993
  current training accuracy:            0.795067
saving to ..., time used 792.433s
Seen 63000 samples, time used: 1665.957s
  current training loss:                0.556433
  current training accuracy:            0.794921
Seen 66000 samples, time used: 1745.035s
  current training loss:                0.556824
  current training accuracy:            0.794955
Seen 69000 samples, time used: 1823.735s
  current training loss:                0.556838
  current training accuracy:            0.795261
Seen 72000 samples, time used: 1901.646s
  current training loss:                0.556838
  current training accuracy:            0.795514
Seen 75000 samples, time used: 1980.627s
  current training loss:                0.556271
  current training accuracy:            0.796000
Seen 78000 samples, time used: 2056.985s
  current training loss:                0.555938
  current training accuracy:            0.796064
Seen 81000 samples, time used: 2135.898s
  current training loss:                0.554671
  current training accuracy:            0.796593
Seen 84000 samples, time used: 2213.938s
  current training loss:                0.554461
  current training accuracy:            0.796762
Seen 87000 samples, time used: 2293.162s
  current training loss:                0.555089
  current training accuracy:            0.796448
Seen 90000 samples, time used: 2372.611s
  current training loss:                0.554870
  current training accuracy:            0.796856
saving to ..., time used 784.931s
Seen 93000 samples, time used: 2452.687s
  current training loss:                0.554674
  current training accuracy:            0.797065
Seen 96000 samples, time used: 2532.172s
  current training loss:                0.554811
  current training accuracy:            0.797125
Seen 99000 samples, time used: 2612.765s
  current training loss:                0.554723
  current training accuracy:            0.797313
Seen 102000 samples, time used: 2692.115s
  current training loss:                0.555257
  current training accuracy:            0.796980
Seen 105000 samples, time used: 2772.105s
  current training loss:                0.555535
  current training accuracy:            0.796971
Seen 108000 samples, time used: 2850.830s
  current training loss:                0.555711
  current training accuracy:            0.797056
Seen 111000 samples, time used: 2930.068s
  current training loss:                0.555508
  current training accuracy:            0.797054
Seen 114000 samples, time used: 3009.040s
  current training loss:                0.555822
  current training accuracy:            0.797088
Seen 117000 samples, time used: 3090.208s
  current training loss:                0.555999
  current training accuracy:            0.796838
Seen 120000 samples, time used: 3171.010s
  current training loss:                0.556636
  current training accuracy:            0.796467
saving to ..., time used 797.901s
Seen 123000 samples, time used: 3251.543s
  current training loss:                0.557033
  current training accuracy:            0.796496
Seen 126000 samples, time used: 3331.583s
  current training loss:                0.556922
  current training accuracy:            0.796635
Seen 129000 samples, time used: 3410.821s
  current training loss:                0.556969
  current training accuracy:            0.796597
Seen 132000 samples, time used: 3490.960s
  current training loss:                0.556981
  current training accuracy:            0.796720
Seen 135000 samples, time used: 3570.111s
  current training loss:                0.556882
  current training accuracy:            0.796956
Seen 138000 samples, time used: 3651.675s
  current training loss:                0.556911
  current training accuracy:            0.796862
Seen 141000 samples, time used: 3732.446s
  current training loss:                0.557098
  current training accuracy:            0.796979
Seen 144000 samples, time used: 3812.866s
  current training loss:                0.557083
  current training accuracy:            0.797104
Seen 147000 samples, time used: 3892.572s
  current training loss:                0.557256
  current training accuracy:            0.796986
Seen 150000 samples, time used: 3971.412s
  current training loss:                0.557540
  current training accuracy:            0.796820
saving to ..., time used 799.841s
Seen 153000 samples, time used: 4051.592s
  current training loss:                0.557635
  current training accuracy:            0.796915
Seen 156000 samples, time used: 4131.748s
  current training loss:                0.558002
  current training accuracy:            0.796750
Seen 159000 samples, time used: 4211.020s
  current training loss:                0.558538
  current training accuracy:            0.796535
Seen 162000 samples, time used: 4290.152s
  current training loss:                0.558442
  current training accuracy:            0.796617
Seen 165000 samples, time used: 4368.480s
  current training loss:                0.558496
  current training accuracy:            0.796679
Seen 168000 samples, time used: 4447.526s
  current training loss:                0.558772
  current training accuracy:            0.796577
Seen 171000 samples, time used: 4525.412s
  current training loss:                0.559035
  current training accuracy:            0.796368
Seen 174000 samples, time used: 4603.084s
  current training loss:                0.558904
  current training accuracy:            0.796379
Seen 177000 samples, time used: 4682.969s
  current training loss:                0.559109
  current training accuracy:            0.796418
Seen 180000 samples, time used: 4762.864s
  current training loss:                0.559067
  current training accuracy:            0.796456
saving to ..., time used 790.921s
Seen 183000 samples, time used: 4843.915s
  current training loss:                0.559476
  current training accuracy:            0.796191
Seen 186000 samples, time used: 4924.296s
  current training loss:                0.559567
  current training accuracy:            0.796043
Seen 189000 samples, time used: 5005.447s
  current training loss:                0.559261
  current training accuracy:            0.796296
Seen 192000 samples, time used: 5085.792s
  current training loss:                0.558993
  current training accuracy:            0.796370
Seen 195000 samples, time used: 5165.259s
  current training loss:                0.559259
  current training accuracy:            0.796262
Seen 198000 samples, time used: 5242.459s
  current training loss:                0.559428
  current training accuracy:            0.796141
Seen 201000 samples, time used: 5321.866s
  current training loss:                0.559559
  current training accuracy:            0.796139
Seen 204000 samples, time used: 5401.241s
  current training loss:                0.559983
  current training accuracy:            0.795961
Seen 207000 samples, time used: 5480.993s
  current training loss:                0.560103
  current training accuracy:            0.795942
Seen 210000 samples, time used: 5560.518s
  current training loss:                0.559912
  current training accuracy:            0.795957
saving to ..., time used 797.116s
Seen 213000 samples, time used: 5642.989s
  current training loss:                0.559939
  current training accuracy:            0.796000
Seen 216000 samples, time used: 5725.518s
  current training loss:                0.560205
  current training accuracy:            0.795875
Seen 219000 samples, time used: 5805.315s
  current training loss:                0.559787
  current training accuracy:            0.796023
Seen 222000 samples, time used: 5887.943s
  current training loss:                0.559711
  current training accuracy:            0.796054
Seen 225000 samples, time used: 5968.341s
  current training loss:                0.559763
  current training accuracy:            0.796004
Seen 228000 samples, time used: 6048.571s
  current training loss:                0.559621
  current training accuracy:            0.796009
Seen 231000 samples, time used: 6132.046s
  current training loss:                0.559493
  current training accuracy:            0.796139
Seen 234000 samples, time used: 6211.386s
  current training loss:                0.559265
  current training accuracy:            0.796252
Seen 237000 samples, time used: 6290.582s
  current training loss:                0.558828
  current training accuracy:            0.796570
Seen 240000 samples, time used: 6371.648s
  current training loss:                0.558355
  current training accuracy:            0.796762
saving to ..., time used 810.588s
Seen 243000 samples, time used: 6452.422s
  current training loss:                0.558178
  current training accuracy:            0.796827
Seen 246000 samples, time used: 6530.560s
  current training loss:                0.558399
  current training accuracy:            0.796622
Seen 249000 samples, time used: 6612.606s
  current training loss:                0.558281
  current training accuracy:            0.796711
Seen 252000 samples, time used: 6690.760s
  current training loss:                0.558457
  current training accuracy:            0.796639
Seen 255000 samples, time used: 6769.696s
  current training loss:                0.558470
  current training accuracy:            0.796647
Seen 258000 samples, time used: 6850.687s
  current training loss:                0.558553
  current training accuracy:            0.796643
Seen 261000 samples, time used: 6930.182s
  current training loss:                0.558849
  current training accuracy:            0.796521
Seen 264000 samples, time used: 7008.485s
  current training loss:                0.558818
  current training accuracy:            0.796583
Seen 267000 samples, time used: 7087.969s
  current training loss:                0.558796
  current training accuracy:            0.796584
Seen 270000 samples, time used: 7168.355s
  current training loss:                0.558776
  current training accuracy:            0.796578
saving to ..., time used 796.173s
Seen 273000 samples, time used: 7247.467s
  current training loss:                0.558736
  current training accuracy:            0.796634
Seen 276000 samples, time used: 7327.782s
  current training loss:                0.558761
  current training accuracy:            0.796667
Seen 279000 samples, time used: 7408.300s
  current training loss:                0.558453
  current training accuracy:            0.796853
Seen 282000 samples, time used: 7488.348s
  current training loss:                0.558332
  current training accuracy:            0.796879
Seen 285000 samples, time used: 7566.747s
  current training loss:                0.558213
  current training accuracy:            0.796884
Seen 288000 samples, time used: 7646.444s
  current training loss:                0.558188
  current training accuracy:            0.796927
Seen 291000 samples, time used: 7726.177s
  current training loss:                0.558228
  current training accuracy:            0.796897
Seen 294000 samples, time used: 7806.031s
  current training loss:                0.558101
  current training accuracy:            0.796949
Seen 297000 samples, time used: 7887.432s
  current training loss:                0.557972
  current training accuracy:            0.797007
Seen 300000 samples, time used: 7967.247s
  current training loss:                0.558063
  current training accuracy:            0.796990
saving to ..., time used 798.377s
Seen 303000 samples, time used: 8047.104s
  current training loss:                0.558075
  current training accuracy:            0.797066
Seen 306000 samples, time used: 8126.150s
  current training loss:                0.558093
  current training accuracy:            0.796951
Seen 309000 samples, time used: 8205.853s
  current training loss:                0.558132
  current training accuracy:            0.797019
Seen 312000 samples, time used: 8287.803s
  current training loss:                0.558109
  current training accuracy:            0.797042
Seen 315000 samples, time used: 8367.415s
  current training loss:                0.558101
  current training accuracy:            0.797067
Seen 318000 samples, time used: 8447.219s
  current training loss:                0.558066
  current training accuracy:            0.797101
Seen 321000 samples, time used: 8527.037s
  current training loss:                0.558022
  current training accuracy:            0.797103
Seen 324000 samples, time used: 8605.823s
  current training loss:                0.558049
  current training accuracy:            0.797056
Seen 327000 samples, time used: 8685.048s
  current training loss:                0.557982
  current training accuracy:            0.797055
Seen 330000 samples, time used: 8762.547s
  current training loss:                0.557974
  current training accuracy:            0.797052
saving to ..., time used 794.657s
Seen 333000 samples, time used: 8843.373s
  current training loss:                0.558186
  current training accuracy:            0.796919
Seen 336000 samples, time used: 8923.454s
  current training loss:                0.558368
  current training accuracy:            0.796789
Seen 339000 samples, time used: 9003.321s
  current training loss:                0.558644
  current training accuracy:            0.796649
Seen 342000 samples, time used: 9082.419s
  current training loss:                0.558506
  current training accuracy:            0.796699
Seen 345000 samples, time used: 9163.365s
  current training loss:                0.558477
  current training accuracy:            0.796672
Seen 348000 samples, time used: 9243.192s
  current training loss:                0.558473
  current training accuracy:            0.796652
Seen 351000 samples, time used: 9324.885s
  current training loss:                0.558442
  current training accuracy:            0.796672
Seen 354000 samples, time used: 9403.405s
  current training loss:                0.558401
  current training accuracy:            0.796715
Seen 357000 samples, time used: 9481.966s
  current training loss:                0.558297
  current training accuracy:            0.796784
Seen 360000 samples, time used: 9562.529s
  current training loss:                0.558266
  current training accuracy:            0.796828
saving to ..., time used 799.225s
Seen 363000 samples, time used: 9641.827s
  current training loss:                0.558260
  current training accuracy:            0.796868
Seen 366000 samples, time used: 9721.057s
  current training loss:                0.558124
  current training accuracy:            0.796891
Seen 369000 samples, time used: 9801.043s
  current training loss:                0.557991
  current training accuracy:            0.797022
Seen 372000 samples, time used: 9880.668s
  current training loss:                0.557949
  current training accuracy:            0.797008
Seen 375000 samples, time used: 9960.607s
  current training loss:                0.557897
  current training accuracy:            0.797035
Seen 378000 samples, time used: 10041.464s
  current training loss:                0.557889
  current training accuracy:            0.797048
Seen 381000 samples, time used: 10119.579s
  current training loss:                0.557882
  current training accuracy:            0.797063
Seen 384000 samples, time used: 10199.531s
  current training loss:                0.557801
  current training accuracy:            0.797120
Seen 387000 samples, time used: 10281.439s
  current training loss:                0.557851
  current training accuracy:            0.797106
Seen 390000 samples, time used: 10360.802s
  current training loss:                0.557939
  current training accuracy:            0.797041
saving to ..., time used 797.727s
Seen 393000 samples, time used: 10445.037s
  current training loss:                0.557796
  current training accuracy:            0.797084
Seen 396000 samples, time used: 10527.747s
  current training loss:                0.557568
  current training accuracy:            0.797194
Seen 399000 samples, time used: 10610.313s
  current training loss:                0.557756
  current training accuracy:            0.797110
Seen 402000 samples, time used: 10692.147s
  current training loss:                0.557808
  current training accuracy:            0.797142
Seen 405000 samples, time used: 10776.327s
  current training loss:                0.557823
  current training accuracy:            0.797160
Seen 408000 samples, time used: 10857.922s
  current training loss:                0.557854
  current training accuracy:            0.797113
Seen 411000 samples, time used: 10937.846s
  current training loss:                0.557726
  current training accuracy:            0.797197
Seen 414000 samples, time used: 11017.689s
  current training loss:                0.557569
  current training accuracy:            0.797319
Seen 417000 samples, time used: 11099.339s
  current training loss:                0.557506
  current training accuracy:            0.797300
Seen 420000 samples, time used: 11178.680s
  current training loss:                0.557660
  current training accuracy:            0.797245
saving to ..., time used 817.298s
Seen 423000 samples, time used: 11258.626s
  current training loss:                0.557617
  current training accuracy:            0.797208
Seen 426000 samples, time used: 11338.613s
  current training loss:                0.557607
  current training accuracy:            0.797261
Seen 429000 samples, time used: 11416.890s
  current training loss:                0.557744
  current training accuracy:            0.797156
Seen 432000 samples, time used: 11495.912s
  current training loss:                0.557574
  current training accuracy:            0.797201
Seen 435000 samples, time used: 11578.009s
  current training loss:                0.557409
  current training accuracy:            0.797274
Seen 438000 samples, time used: 11657.164s
  current training loss:                0.557320
  current training accuracy:            0.797297
Seen 441000 samples, time used: 11737.528s
  current training loss:                0.557243
  current training accuracy:            0.797345
Seen 444000 samples, time used: 11816.544s
  current training loss:                0.557054
  current training accuracy:            0.797455
Seen 447000 samples, time used: 11894.118s
  current training loss:                0.557229
  current training accuracy:            0.797362
Seen 450000 samples, time used: 11973.370s
  current training loss:                0.557413
  current training accuracy:            0.797238
saving to ..., time used 794.154s
Seen 453000 samples, time used: 12052.579s
  current training loss:                0.557519
  current training accuracy:            0.797150
Seen 456000 samples, time used: 12132.069s
  current training loss:                0.557456
  current training accuracy:            0.797147
Seen 459000 samples, time used: 12212.287s
  current training loss:                0.557503
  current training accuracy:            0.797131
Seen 462000 samples, time used: 12292.607s
  current training loss:                0.557470
  current training accuracy:            0.797156
Seen 465000 samples, time used: 12371.979s
  current training loss:                0.557494
  current training accuracy:            0.797108
Seen 468000 samples, time used: 12452.195s
  current training loss:                0.557430
  current training accuracy:            0.797096
Seen 471000 samples, time used: 12531.383s
  current training loss:                0.557348
  current training accuracy:            0.797091
Seen 474000 samples, time used: 12610.771s
  current training loss:                0.557439
  current training accuracy:            0.797049
Seen 477000 samples, time used: 12689.920s
  current training loss:                0.557498
  current training accuracy:            0.796998
Seen 480000 samples, time used: 12769.269s
  current training loss:                0.557315
  current training accuracy:            0.797040
saving to ..., time used 795.364s
Seen 483000 samples, time used: 12849.506s
  current training loss:                0.557242
  current training accuracy:            0.797072
Seen 486000 samples, time used: 12929.281s
  current training loss:                0.557191
  current training accuracy:            0.797080
Seen 489000 samples, time used: 13008.816s
  current training loss:                0.557106
  current training accuracy:            0.797123
Seen 492000 samples, time used: 13089.447s
  current training loss:                0.557076
  current training accuracy:            0.797098
Seen 495000 samples, time used: 13169.601s
  current training loss:                0.557001
  current training accuracy:            0.797133
Seen 498000 samples, time used: 13249.032s
  current training loss:                0.556998
  current training accuracy:            0.797090
Seen 501000 samples, time used: 13328.260s
  current training loss:                0.556871
  current training accuracy:            0.797170
Seen 504000 samples, time used: 13406.879s
  current training loss:                0.556897
  current training accuracy:            0.797177
Seen 507000 samples, time used: 13486.472s
  current training loss:                0.556840
  current training accuracy:            0.797197
Seen 510000 samples, time used: 13567.053s
  current training loss:                0.557051
  current training accuracy:            0.797110
saving to ..., time used 797.171s
Seen 513000 samples, time used: 13647.628s
  current training loss:                0.556990
  current training accuracy:            0.797123
Seen 516000 samples, time used: 13729.161s
  current training loss:                0.556784
  current training accuracy:            0.797229
Seen 519000 samples, time used: 13810.168s
  current training loss:                0.556728
  current training accuracy:            0.797216
Seen 522000 samples, time used: 13889.465s
  current training loss:                0.556734
  current training accuracy:            0.797232
Seen 525000 samples, time used: 13969.084s
  current training loss:                0.556828
  current training accuracy:            0.797210
Seen 528000 samples, time used: 14048.580s
  current training loss:                0.556885
  current training accuracy:            0.797186
Seen 531000 samples, time used: 14128.292s
  current training loss:                0.557000
  current training accuracy:            0.797139
Seen 534000 samples, time used: 14208.114s
  current training loss:                0.557002
  current training accuracy:            0.797140
Seen 537000 samples, time used: 14287.748s
  current training loss:                0.556934
  current training accuracy:            0.797209
Seen 540000 samples, time used: 14367.164s
  current training loss:                0.556794
  current training accuracy:            0.797283
saving to ..., time used 799.572s
Seen 543000 samples, time used: 14447.320s
  current training loss:                0.556832
  current training accuracy:            0.797291
Seen 546000 samples, time used: 14526.946s
  current training loss:                0.556840
  current training accuracy:            0.797293
Seen 549000 samples, time used: 14607.270s
  current training loss:                0.556867
  current training accuracy:            0.797259
Epoch 4 of 20 took 14640.336s
  training loss:                0.556858
  training accuracy:            79.73 %
  validation loss:              0.517635
  validation accuracy:          79.47 %
  test loss:                    0.523127
  test accuracy:                79.15 %
saving to ./snli/word_by_word_model_epoch4.npz
Seen 3000 samples, time used: 80.727s
  current training loss:                0.528032
  current training accuracy:            0.813333
Seen 6000 samples, time used: 159.716s
  current training loss:                0.531061
  current training accuracy:            0.814667
Seen 9000 samples, time used: 238.489s
  current training loss:                0.530689
  current training accuracy:            0.812444
Seen 12000 samples, time used: 318.568s
  current training loss:                0.525455
  current training accuracy:            0.814500
Seen 15000 samples, time used: 397.448s
  current training loss:                0.526952
  current training accuracy:            0.813000
Seen 18000 samples, time used: 476.578s
  current training loss:                0.527499
  current training accuracy:            0.812444
Seen 21000 samples, time used: 557.430s
  current training loss:                0.529573
  current training accuracy:            0.810048
Seen 24000 samples, time used: 637.332s
  current training loss:                0.529208
  current training accuracy:            0.810292
Seen 27000 samples, time used: 716.247s
  current training loss:                0.529272
  current training accuracy:            0.809926
Seen 30000 samples, time used: 797.475s
  current training loss:                0.528021
  current training accuracy:            0.810000
saving to ..., time used 797.475s
Seen 33000 samples, time used: 878.162s
  current training loss:                0.528997
  current training accuracy:            0.809545
Seen 36000 samples, time used: 959.220s
  current training loss:                0.528976
  current training accuracy:            0.809194
Seen 39000 samples, time used: 1040.166s
  current training loss:                0.528919
  current training accuracy:            0.809077
Seen 42000 samples, time used: 1120.179s
  current training loss:                0.528932
  current training accuracy:            0.808905
Seen 45000 samples, time used: 1198.149s
  current training loss:                0.529181
  current training accuracy:            0.808867
Seen 48000 samples, time used: 1277.357s
  current training loss:                0.527809
  current training accuracy:            0.809937
Seen 51000 samples, time used: 1356.363s
  current training loss:                0.528845
  current training accuracy:            0.810000
Seen 54000 samples, time used: 1436.786s
  current training loss:                0.530004
  current training accuracy:            0.809444
Seen 57000 samples, time used: 1515.603s
  current training loss:                0.530275
  current training accuracy:            0.809456
Seen 60000 samples, time used: 1595.787s
  current training loss:                0.530907
  current training accuracy:            0.809167
saving to ..., time used 797.769s
Seen 63000 samples, time used: 1675.847s
  current training loss:                0.530033
  current training accuracy:            0.809619
Seen 66000 samples, time used: 1755.852s
  current training loss:                0.530647
  current training accuracy:            0.809652
Seen 69000 samples, time used: 1836.532s
  current training loss:                0.530883
  current training accuracy:            0.809652
Seen 72000 samples, time used: 1918.607s
  current training loss:                0.530032
  current training accuracy:            0.810125
Seen 75000 samples, time used: 1997.607s
  current training loss:                0.529890
  current training accuracy:            0.810187
Seen 78000 samples, time used: 2078.373s
  current training loss:                0.531427
  current training accuracy:            0.809346
Seen 81000 samples, time used: 2158.075s
  current training loss:                0.531682
  current training accuracy:            0.809358
Seen 84000 samples, time used: 2238.840s
  current training loss:                0.531375
  current training accuracy:            0.809536
Seen 87000 samples, time used: 2318.300s
  current training loss:                0.531681
  current training accuracy:            0.809195
Seen 90000 samples, time used: 2398.286s
  current training loss:                0.531530
  current training accuracy:            0.809267
saving to ..., time used 801.955s
Seen 93000 samples, time used: 2477.937s
  current training loss:                0.531696
  current training accuracy:            0.808989
Seen 96000 samples, time used: 2556.794s
  current training loss:                0.531534
  current training accuracy:            0.809042
Seen 99000 samples, time used: 2637.678s
  current training loss:                0.531949
  current training accuracy:            0.808828
Seen 102000 samples, time used: 2717.476s
  current training loss:                0.532339
  current training accuracy:            0.808873
Seen 105000 samples, time used: 2797.883s
  current training loss:                0.532676
  current training accuracy:            0.808724
Seen 108000 samples, time used: 2878.323s
  current training loss:                0.533435
  current training accuracy:            0.808731
Seen 111000 samples, time used: 2958.757s
  current training loss:                0.533652
  current training accuracy:            0.808847
Seen 114000 samples, time used: 3038.625s
  current training loss:                0.533566
  current training accuracy:            0.808904
Seen 117000 samples, time used: 3119.581s
  current training loss:                0.533646
  current training accuracy:            0.808795
Seen 120000 samples, time used: 3200.120s
  current training loss:                0.533376
  current training accuracy:            0.809000
saving to ..., time used 801.241s
Seen 123000 samples, time used: 3280.324s
  current training loss:                0.533397
  current training accuracy:            0.808846
Seen 126000 samples, time used: 3361.019s
  current training loss:                0.533318
  current training accuracy:            0.808929
Seen 129000 samples, time used: 3439.101s
  current training loss:                0.533387
  current training accuracy:            0.808829
Seen 132000 samples, time used: 3520.195s
  current training loss:                0.533914
  current training accuracy:            0.808515
Seen 135000 samples, time used: 3600.611s
  current training loss:                0.533761
  current training accuracy:            0.808578
Seen 138000 samples, time used: 3680.840s
  current training loss:                0.533778
  current training accuracy:            0.808587
Seen 141000 samples, time used: 3760.611s
  current training loss:                0.534026
  current training accuracy:            0.808475
Seen 144000 samples, time used: 3840.447s
  current training loss:                0.534269
  current training accuracy:            0.808361
Seen 147000 samples, time used: 3919.276s
  current training loss:                0.533819
  current training accuracy:            0.808605
Seen 150000 samples, time used: 3999.590s
  current training loss:                0.533772
  current training accuracy:            0.808633
saving to ..., time used 798.903s
Seen 153000 samples, time used: 4079.857s
  current training loss:                0.533632
  current training accuracy:            0.808719
Seen 156000 samples, time used: 4159.062s
  current training loss:                0.533584
  current training accuracy:            0.808737
Seen 159000 samples, time used: 4239.923s
  current training loss:                0.533649
  current training accuracy:            0.808629
Seen 162000 samples, time used: 4317.570s
  current training loss:                0.533688
  current training accuracy:            0.808685
Seen 165000 samples, time used: 4396.281s
  current training loss:                0.533363
  current training accuracy:            0.808758
Seen 168000 samples, time used: 4480.396s
  current training loss:                0.533107
  current training accuracy:            0.808887
Seen 171000 samples, time used: 4559.915s
  current training loss:                0.533550
  current training accuracy:            0.808579
Seen 174000 samples, time used: 4640.303s
  current training loss:                0.533618
  current training accuracy:            0.808569
Seen 177000 samples, time used: 4721.291s
  current training loss:                0.533579
  current training accuracy:            0.808554
Seen 180000 samples, time used: 4801.137s
  current training loss:                0.533332
  current training accuracy:            0.808650
saving to ..., time used 800.979s
Seen 183000 samples, time used: 4880.468s
  current training loss:                0.533317
  current training accuracy:            0.808661
Seen 186000 samples, time used: 4961.726s
  current training loss:                0.533288
  current training accuracy:            0.808527
Seen 189000 samples, time used: 5042.737s
  current training loss:                0.533443
  current training accuracy:            0.808487
Seen 192000 samples, time used: 5123.820s
  current training loss:                0.533648
  current training accuracy:            0.808422
Seen 195000 samples, time used: 5204.410s
  current training loss:                0.533855
  current training accuracy:            0.808400
Seen 198000 samples, time used: 5285.665s
  current training loss:                0.534259
  current training accuracy:            0.808237
Seen 201000 samples, time used: 5366.494s
  current training loss:                0.534101
  current training accuracy:            0.808308
Seen 204000 samples, time used: 5446.775s
  current training loss:                0.534009
  current training accuracy:            0.808265
Seen 207000 samples, time used: 5527.416s
  current training loss:                0.533959
  current training accuracy:            0.808266
Seen 210000 samples, time used: 5608.873s
  current training loss:                0.534339
  current training accuracy:            0.808076
saving to ..., time used 807.203s
Seen 213000 samples, time used: 5689.092s
  current training loss:                0.534426
  current training accuracy:            0.808089
Seen 216000 samples, time used: 5769.111s
  current training loss:                0.534360
  current training accuracy:            0.808134
Seen 219000 samples, time used: 5851.836s
  current training loss:                0.534263
  current training accuracy:            0.808105
Seen 222000 samples, time used: 5931.181s
  current training loss:                0.534601
  current training accuracy:            0.808041
Seen 225000 samples, time used: 6010.280s
  current training loss:                0.534722
  current training accuracy:            0.808080
Seen 228000 samples, time used: 6091.461s
  current training loss:                0.534899
  current training accuracy:            0.808092
Seen 231000 samples, time used: 6171.401s
  current training loss:                0.534895
  current training accuracy:            0.808061
Seen 234000 samples, time used: 6252.425s
  current training loss:                0.534828
  current training accuracy:            0.808047
Seen 237000 samples, time used: 6332.658s
  current training loss:                0.534894
  current training accuracy:            0.808017
Seen 240000 samples, time used: 6413.352s
  current training loss:                0.534884
  current training accuracy:            0.807962
saving to ..., time used 803.933s
Seen 243000 samples, time used: 6494.587s
  current training loss:                0.535053
  current training accuracy:            0.807975
Seen 246000 samples, time used: 6574.541s
  current training loss:                0.535404
  current training accuracy:            0.807894
Seen 249000 samples, time used: 6657.266s
  current training loss:                0.535616
  current training accuracy:            0.807771
Seen 252000 samples, time used: 6739.183s
  current training loss:                0.535795
  current training accuracy:            0.807742
Seen 255000 samples, time used: 6818.645s
  current training loss:                0.535980
  current training accuracy:            0.807694
Seen 258000 samples, time used: 6899.667s
  current training loss:                0.535974
  current training accuracy:            0.807585
Seen 261000 samples, time used: 6979.715s
  current training loss:                0.536135
  current training accuracy:            0.807494
Seen 264000 samples, time used: 7060.095s
  current training loss:                0.535974
  current training accuracy:            0.807500
Seen 267000 samples, time used: 7141.381s
  current training loss:                0.535972
  current training accuracy:            0.807517
Seen 270000 samples, time used: 7222.567s
  current training loss:                0.535776
  current training accuracy:            0.807600
saving to ..., time used 808.635s
Seen 273000 samples, time used: 7307.366s
  current training loss:                0.535967
  current training accuracy:            0.807590
Seen 276000 samples, time used: 7388.955s
  current training loss:                0.536074
  current training accuracy:            0.807576
Seen 279000 samples, time used: 7470.412s
  current training loss:                0.536376
  current training accuracy:            0.807448
Seen 282000 samples, time used: 7551.344s
  current training loss:                0.536477
  current training accuracy:            0.807337
Seen 285000 samples, time used: 7633.924s
  current training loss:                0.536363
  current training accuracy:            0.807333
Seen 288000 samples, time used: 7715.526s
  current training loss:                0.536526
  current training accuracy:            0.807257
Seen 291000 samples, time used: 7797.493s
  current training loss:                0.536638
  current training accuracy:            0.807220
Seen 294000 samples, time used: 7877.920s
  current training loss:                0.536320
  current training accuracy:            0.807357
Seen 297000 samples, time used: 7959.223s
  current training loss:                0.536338
  current training accuracy:            0.807360
Seen 300000 samples, time used: 8040.012s
  current training loss:                0.536278
  current training accuracy:            0.807297
saving to ..., time used 816.951s
Seen 303000 samples, time used: 8121.305s
  current training loss:                0.536197
  current training accuracy:            0.807389
Seen 306000 samples, time used: 8201.348s
  current training loss:                0.536073
  current training accuracy:            0.807448
Seen 309000 samples, time used: 8282.148s
  current training loss:                0.536245
  current training accuracy:            0.807366
Seen 312000 samples, time used: 8362.303s
  current training loss:                0.536263
  current training accuracy:            0.807362
Seen 315000 samples, time used: 8442.034s
  current training loss:                0.536329
  current training accuracy:            0.807289
Seen 318000 samples, time used: 8521.018s
  current training loss:                0.536436
  current training accuracy:            0.807223
Seen 321000 samples, time used: 8602.295s
  current training loss:                0.536205
  current training accuracy:            0.807312
Seen 324000 samples, time used: 8683.209s
  current training loss:                0.536224
  current training accuracy:            0.807327
Seen 327000 samples, time used: 8762.768s
  current training loss:                0.536314
  current training accuracy:            0.807180
Seen 330000 samples, time used: 8843.424s
  current training loss:                0.536186
  current training accuracy:            0.807291
saving to ..., time used 802.934s
Seen 333000 samples, time used: 8926.347s
  current training loss:                0.536270
  current training accuracy:            0.807228
Seen 336000 samples, time used: 9006.926s
  current training loss:                0.536236
  current training accuracy:            0.807214
Seen 339000 samples, time used: 9086.657s
  current training loss:                0.536092
  current training accuracy:            0.807257
Seen 342000 samples, time used: 9167.527s
  current training loss:                0.536208
  current training accuracy:            0.807225
Seen 345000 samples, time used: 9247.381s
  current training loss:                0.536282
  current training accuracy:            0.807258
Seen 348000 samples, time used: 9327.482s
  current training loss:                0.536204
  current training accuracy:            0.807313
Seen 351000 samples, time used: 9407.959s
  current training loss:                0.536220
  current training accuracy:            0.807342
Seen 354000 samples, time used: 9488.305s
  current training loss:                0.536378
  current training accuracy:            0.807206
Seen 357000 samples, time used: 9566.655s
  current training loss:                0.536282
  current training accuracy:            0.807218
Seen 360000 samples, time used: 9648.789s
  current training loss:                0.536216
  current training accuracy:            0.807303
saving to ..., time used 804.850s
Seen 363000 samples, time used: 9728.976s
  current training loss:                0.536130
  current training accuracy:            0.807369
Seen 366000 samples, time used: 9808.621s
  current training loss:                0.535993
  current training accuracy:            0.807437
Seen 369000 samples, time used: 9889.055s
  current training loss:                0.536118
  current training accuracy:            0.807396
Seen 372000 samples, time used: 9969.861s
  current training loss:                0.535986
  current training accuracy:            0.807433
Seen 375000 samples, time used: 10051.904s
  current training loss:                0.536144
  current training accuracy:            0.807403
Seen 378000 samples, time used: 10131.608s
  current training loss:                0.536233
  current training accuracy:            0.807386
Seen 381000 samples, time used: 10213.451s
  current training loss:                0.536196
  current training accuracy:            0.807402
Seen 384000 samples, time used: 10293.697s
  current training loss:                0.536157
  current training accuracy:            0.807432
Seen 387000 samples, time used: 10375.790s
  current training loss:                0.536258
  current training accuracy:            0.807370
Seen 390000 samples, time used: 10457.610s
  current training loss:                0.536376
  current training accuracy:            0.807359
saving to ..., time used 808.335s
Seen 393000 samples, time used: 10537.227s
  current training loss:                0.536401
  current training accuracy:            0.807410
Seen 396000 samples, time used: 10618.748s
  current training loss:                0.536449
  current training accuracy:            0.807399
Seen 399000 samples, time used: 10699.550s
  current training loss:                0.536557
  current training accuracy:            0.807346
Seen 402000 samples, time used: 10777.914s
  current training loss:                0.536485
  current training accuracy:            0.807376
Seen 405000 samples, time used: 10858.917s
  current training loss:                0.536480
  current training accuracy:            0.807336
Seen 408000 samples, time used: 10939.687s
  current training loss:                0.536745
  current training accuracy:            0.807272
Seen 411000 samples, time used: 11019.378s
  current training loss:                0.536586
  current training accuracy:            0.807333
Seen 414000 samples, time used: 11096.822s
  current training loss:                0.536625
  current training accuracy:            0.807333
Seen 417000 samples, time used: 11175.523s
  current training loss:                0.536445
  current training accuracy:            0.807448
Seen 420000 samples, time used: 11256.257s
  current training loss:                0.536349
  current training accuracy:            0.807429
saving to ..., time used 798.159s
Seen 423000 samples, time used: 11336.498s
  current training loss:                0.536330
  current training accuracy:            0.807461
Seen 426000 samples, time used: 11417.521s
  current training loss:                0.536274
  current training accuracy:            0.807467
Seen 429000 samples, time used: 11497.506s
  current training loss:                0.536300
  current training accuracy:            0.807422
Seen 432000 samples, time used: 11576.663s
  current training loss:                0.536265
  current training accuracy:            0.807461
Seen 435000 samples, time used: 11659.307s
  current training loss:                0.536274
  current training accuracy:            0.807497
Seen 438000 samples, time used: 11740.048s
  current training loss:                0.536276
  current training accuracy:            0.807482
Seen 441000 samples, time used: 11819.790s
  current training loss:                0.536195
  current training accuracy:            0.807456
Seen 444000 samples, time used: 11901.238s
  current training loss:                0.536157
  current training accuracy:            0.807493
Seen 447000 samples, time used: 11983.562s
  current training loss:                0.536304
  current training accuracy:            0.807423
Seen 450000 samples, time used: 12062.856s
  current training loss:                0.536355
  current training accuracy:            0.807389
saving to ..., time used 806.103s
Seen 453000 samples, time used: 12144.927s
  current training loss:                0.536520
  current training accuracy:            0.807320
Seen 456000 samples, time used: 12226.459s
  current training loss:                0.536633
  current training accuracy:            0.807289
Seen 459000 samples, time used: 12307.510s
  current training loss:                0.536766
  current training accuracy:            0.807257
Seen 462000 samples, time used: 12391.177s
  current training loss:                0.536784
  current training accuracy:            0.807253
Seen 465000 samples, time used: 12472.141s
  current training loss:                0.536672
  current training accuracy:            0.807286
Seen 468000 samples, time used: 12552.880s
  current training loss:                0.536716
  current training accuracy:            0.807271
Seen 471000 samples, time used: 12633.664s
  current training loss:                0.536805
  current training accuracy:            0.807223
Seen 474000 samples, time used: 12713.342s
  current training loss:                0.536952
  current training accuracy:            0.807162
Seen 477000 samples, time used: 12793.965s
  current training loss:                0.536933
  current training accuracy:            0.807170
Seen 480000 samples, time used: 12874.346s
  current training loss:                0.536864
  current training accuracy:            0.807171
saving to ..., time used 811.028s
Seen 483000 samples, time used: 12955.163s
  current training loss:                0.536918
  current training accuracy:            0.807120
Seen 486000 samples, time used: 13037.300s
  current training loss:                0.536932
  current training accuracy:            0.807185
Seen 489000 samples, time used: 13118.593s
  current training loss:                0.536929
  current training accuracy:            0.807213
Seen 492000 samples, time used: 13196.270s
  current training loss:                0.536967
  current training accuracy:            0.807175
Seen 495000 samples, time used: 13277.811s
  current training loss:                0.537046
  current training accuracy:            0.807164
Seen 498000 samples, time used: 13358.074s
  current training loss:                0.536963
  current training accuracy:            0.807215
Seen 501000 samples, time used: 13438.312s
  current training loss:                0.536882
  current training accuracy:            0.807267
Seen 504000 samples, time used: 13519.116s
  current training loss:                0.536898
  current training accuracy:            0.807268
Seen 507000 samples, time used: 13598.727s
  current training loss:                0.536959
  current training accuracy:            0.807235
Seen 510000 samples, time used: 13678.349s
  current training loss:                0.536849
  current training accuracy:            0.807269
saving to ..., time used 803.508s
Seen 513000 samples, time used: 13760.103s
  current training loss:                0.536650
  current training accuracy:            0.807331
Seen 516000 samples, time used: 13842.091s
  current training loss:                0.536660
  current training accuracy:            0.807343
Seen 519000 samples, time used: 13921.689s
  current training loss:                0.536746
  current training accuracy:            0.807331
Seen 522000 samples, time used: 14001.269s
  current training loss:                0.536663
  current training accuracy:            0.807400
Seen 525000 samples, time used: 14080.728s
  current training loss:                0.536668
  current training accuracy:            0.807404
Seen 528000 samples, time used: 14160.181s
  current training loss:                0.536570
  current training accuracy:            0.807458
Seen 531000 samples, time used: 14241.252s
  current training loss:                0.536630
  current training accuracy:            0.807395
Seen 534000 samples, time used: 14321.085s
  current training loss:                0.536660
  current training accuracy:            0.807423
Seen 537000 samples, time used: 14401.430s
  current training loss:                0.536658
  current training accuracy:            0.807419
Seen 540000 samples, time used: 14479.317s
  current training loss:                0.536540
  current training accuracy:            0.807517
saving to ..., time used 800.497s
Seen 543000 samples, time used: 14561.509s
  current training loss:                0.536650
  current training accuracy:            0.807484
Seen 546000 samples, time used: 14638.853s
  current training loss:                0.536744
  current training accuracy:            0.807456
Seen 549000 samples, time used: 14717.495s
  current training loss:                0.536775
  current training accuracy:            0.807446
Epoch 5 of 20 took 14750.434s
  training loss:                0.536745
  training accuracy:            80.75 %
  validation loss:              0.499694
  validation accuracy:          80.28 %
  test loss:                    0.506387
  test accuracy:                80.11 %
saving to ./snli/word_by_word_model_epoch5.npz
Seen 3000 samples, time used: 79.906s
  current training loss:                0.519244
  current training accuracy:            0.822333
Seen 6000 samples, time used: 161.060s
  current training loss:                0.516623
  current training accuracy:            0.824500
Seen 9000 samples, time used: 241.443s
  current training loss:                0.514548
  current training accuracy:            0.822333
Seen 12000 samples, time used: 320.639s
  current training loss:                0.518933
  current training accuracy:            0.818833
Seen 15000 samples, time used: 401.048s
  current training loss:                0.517815
  current training accuracy:            0.819067
Seen 18000 samples, time used: 480.855s
  current training loss:                0.519572
  current training accuracy:            0.816889
Seen 21000 samples, time used: 560.024s
  current training loss:                0.517063
  current training accuracy:            0.817619
Seen 24000 samples, time used: 640.475s
  current training loss:                0.515957
  current training accuracy:            0.818583
Seen 27000 samples, time used: 721.092s
  current training loss:                0.514397
  current training accuracy:            0.819185
Seen 30000 samples, time used: 801.098s
  current training loss:                0.511278
  current training accuracy:            0.820267
saving to ..., time used 801.098s
Seen 33000 samples, time used: 881.157s
  current training loss:                0.509091
  current training accuracy:            0.820818
Seen 36000 samples, time used: 962.764s
  current training loss:                0.509982
  current training accuracy:            0.820361
Seen 39000 samples, time used: 1042.403s
  current training loss:                0.510671
  current training accuracy:            0.819718
Seen 42000 samples, time used: 1122.657s
  current training loss:                0.511723
  current training accuracy:            0.819548
Seen 45000 samples, time used: 1201.434s
  current training loss:                0.510975
  current training accuracy:            0.819822
Seen 48000 samples, time used: 1282.447s
  current training loss:                0.511243
  current training accuracy:            0.819604
Seen 51000 samples, time used: 1362.831s
  current training loss:                0.511188
  current training accuracy:            0.819922
Seen 54000 samples, time used: 1442.911s
  current training loss:                0.511544
  current training accuracy:            0.820426
Seen 57000 samples, time used: 1522.153s
  current training loss:                0.511684
  current training accuracy:            0.820439
Seen 60000 samples, time used: 1601.162s
  current training loss:                0.510903
  current training accuracy:            0.820483
saving to ..., time used 799.565s
Seen 63000 samples, time used: 1683.622s
  current training loss:                0.511272
  current training accuracy:            0.820270
Seen 66000 samples, time used: 1763.167s
  current training loss:                0.510780
  current training accuracy:            0.820636
Seen 69000 samples, time used: 1844.007s
  current training loss:                0.511028
  current training accuracy:            0.820217
Seen 72000 samples, time used: 1925.048s
  current training loss:                0.510830
  current training accuracy:            0.820444
Seen 75000 samples, time used: 2005.547s
  current training loss:                0.510657
  current training accuracy:            0.820573
Seen 78000 samples, time used: 2085.673s
  current training loss:                0.511246
  current training accuracy:            0.820295
Seen 81000 samples, time used: 2165.342s
  current training loss:                0.511762
  current training accuracy:            0.819951
Seen 84000 samples, time used: 2246.223s
  current training loss:                0.511843
  current training accuracy:            0.819786
Seen 87000 samples, time used: 2325.047s
  current training loss:                0.511728
  current training accuracy:            0.819954
Seen 90000 samples, time used: 2404.860s
  current training loss:                0.511839
  current training accuracy:            0.819722
saving to ..., time used 803.185s
Seen 93000 samples, time used: 2486.361s
  current training loss:                0.511933
  current training accuracy:            0.819677
Seen 96000 samples, time used: 2567.320s
  current training loss:                0.512238
  current training accuracy:            0.819510
Seen 99000 samples, time used: 2647.304s
  current training loss:                0.513017
  current training accuracy:            0.819020
Seen 102000 samples, time used: 2725.962s
  current training loss:                0.513048
  current training accuracy:            0.819039
Seen 105000 samples, time used: 2806.615s
  current training loss:                0.512946
  current training accuracy:            0.819143
Seen 108000 samples, time used: 2888.046s
  current training loss:                0.513542
  current training accuracy:            0.818759
Seen 111000 samples, time used: 2968.871s
  current training loss:                0.513902
  current training accuracy:            0.818712
Seen 114000 samples, time used: 3048.844s
  current training loss:                0.514005
  current training accuracy:            0.818518
Seen 117000 samples, time used: 3128.642s
  current training loss:                0.514404
  current training accuracy:            0.818513
Seen 120000 samples, time used: 3207.059s
  current training loss:                0.514540
  current training accuracy:            0.818450
saving to ..., time used 801.717s
Seen 123000 samples, time used: 3288.894s
  current training loss:                0.514471
  current training accuracy:            0.818407
Seen 126000 samples, time used: 3369.438s
  current training loss:                0.514598
  current training accuracy:            0.818333
Seen 129000 samples, time used: 3449.744s
  current training loss:                0.515125
  current training accuracy:            0.818000
Seen 132000 samples, time used: 3530.163s
  current training loss:                0.515353
  current training accuracy:            0.818000
Seen 135000 samples, time used: 3610.842s
  current training loss:                0.515728
  current training accuracy:            0.817852
Seen 138000 samples, time used: 3690.953s
  current training loss:                0.515426
  current training accuracy:            0.818159
Seen 141000 samples, time used: 3770.979s
  current training loss:                0.515480
  current training accuracy:            0.818177
Seen 144000 samples, time used: 3849.834s
  current training loss:                0.515495
  current training accuracy:            0.818125
Seen 147000 samples, time used: 3930.682s
  current training loss:                0.515950
  current training accuracy:            0.817912
Seen 150000 samples, time used: 4009.640s
  current training loss:                0.515583
  current training accuracy:            0.817867
saving to ..., time used 802.080s
Seen 153000 samples, time used: 4091.243s
  current training loss:                0.515757
  current training accuracy:            0.817745
Seen 156000 samples, time used: 4172.226s
  current training loss:                0.515906
  current training accuracy:            0.817750
Seen 159000 samples, time used: 4252.691s
  current training loss:                0.515637
  current training accuracy:            0.817943
Seen 162000 samples, time used: 4333.504s
  current training loss:                0.515750
  current training accuracy:            0.817821
Seen 165000 samples, time used: 4413.111s
  current training loss:                0.516089
  current training accuracy:            0.817709
Seen 168000 samples, time used: 4494.857s
  current training loss:                0.516197
  current training accuracy:            0.817744
Seen 171000 samples, time used: 4574.056s
  current training loss:                0.515899
  current training accuracy:            0.817877
Seen 174000 samples, time used: 4653.628s
  current training loss:                0.516080
  current training accuracy:            0.817655
Seen 177000 samples, time used: 4733.791s
  current training loss:                0.516112
  current training accuracy:            0.817576
Seen 180000 samples, time used: 4815.713s
  current training loss:                0.515722
  current training accuracy:            0.817756
saving to ..., time used 805.513s
Seen 183000 samples, time used: 4895.022s
  current training loss:                0.515840
  current training accuracy:            0.817683
Seen 186000 samples, time used: 4973.117s
  current training loss:                0.515688
  current training accuracy:            0.817608
Seen 189000 samples, time used: 5052.060s
  current training loss:                0.515890
  current training accuracy:            0.817534
Seen 192000 samples, time used: 5133.390s
  current training loss:                0.516483
  current training accuracy:            0.817198
Seen 195000 samples, time used: 5213.907s
  current training loss:                0.516270
  current training accuracy:            0.817323
Seen 198000 samples, time used: 5292.920s
  current training loss:                0.516212
  current training accuracy:            0.817338
Seen 201000 samples, time used: 5373.576s
  current training loss:                0.516200
  current training accuracy:            0.817428
Seen 204000 samples, time used: 5454.732s
  current training loss:                0.516529
  current training accuracy:            0.817260
Seen 207000 samples, time used: 5534.459s
  current training loss:                0.516436
  current training accuracy:            0.817213
Seen 210000 samples, time used: 5616.652s
  current training loss:                0.516398
  current training accuracy:            0.817167
saving to ..., time used 800.467s
Seen 213000 samples, time used: 5699.187s
  current training loss:                0.516485
  current training accuracy:            0.817131
Seen 216000 samples, time used: 5779.204s
  current training loss:                0.516736
  current training accuracy:            0.816991
Seen 219000 samples, time used: 5859.041s
  current training loss:                0.516765
  current training accuracy:            0.816845
Seen 222000 samples, time used: 5941.425s
  current training loss:                0.517235
  current training accuracy:            0.816626
Seen 225000 samples, time used: 6022.144s
  current training loss:                0.517275
  current training accuracy:            0.816684
Seen 228000 samples, time used: 6103.804s
  current training loss:                0.517560
  current training accuracy:            0.816535
Seen 231000 samples, time used: 6184.236s
  current training loss:                0.518072
  current training accuracy:            0.816303
Seen 234000 samples, time used: 6265.895s
  current training loss:                0.518165
  current training accuracy:            0.816303
Seen 237000 samples, time used: 6346.797s
  current training loss:                0.518428
  current training accuracy:            0.816241
Seen 240000 samples, time used: 6427.401s
  current training loss:                0.518490
  current training accuracy:            0.816229
saving to ..., time used 809.487s
Seen 243000 samples, time used: 6508.163s
  current training loss:                0.518613
  current training accuracy:            0.816091
Seen 246000 samples, time used: 6587.470s
  current training loss:                0.518811
  current training accuracy:            0.816098
Seen 249000 samples, time used: 6668.042s
  current training loss:                0.518970
  current training accuracy:            0.816088
Seen 252000 samples, time used: 6749.234s
  current training loss:                0.518857
  current training accuracy:            0.816056
Seen 255000 samples, time used: 6828.887s
  current training loss:                0.519015
  current training accuracy:            0.815933
Seen 258000 samples, time used: 6909.282s
  current training loss:                0.519129
  current training accuracy:            0.815841
Seen 261000 samples, time used: 6989.445s
  current training loss:                0.519209
  current training accuracy:            0.815839
Seen 264000 samples, time used: 7071.112s
  current training loss:                0.518906
  current training accuracy:            0.815966
Seen 267000 samples, time used: 7151.411s
  current training loss:                0.518812
  current training accuracy:            0.815895
Seen 270000 samples, time used: 7231.516s
  current training loss:                0.518734
  current training accuracy:            0.815856
saving to ..., time used 803.620s
Seen 273000 samples, time used: 7311.296s
  current training loss:                0.519029
  current training accuracy:            0.815758
Seen 276000 samples, time used: 7392.035s
  current training loss:                0.519182
  current training accuracy:            0.815688
Seen 279000 samples, time used: 7471.700s
  current training loss:                0.519463
  current training accuracy:            0.815609
Seen 282000 samples, time used: 7552.731s
  current training loss:                0.519433
  current training accuracy:            0.815652
Seen 285000 samples, time used: 7633.233s
  current training loss:                0.519472
  current training accuracy:            0.815677
Seen 288000 samples, time used: 7712.761s
  current training loss:                0.519674
  current training accuracy:            0.815545
Seen 291000 samples, time used: 7795.223s
  current training loss:                0.519793
  current training accuracy:            0.815498
Seen 294000 samples, time used: 7876.231s
  current training loss:                0.520087
  current training accuracy:            0.815367
Seen 297000 samples, time used: 7957.269s
  current training loss:                0.519930
  current training accuracy:            0.815475
Seen 300000 samples, time used: 8037.670s
  current training loss:                0.520088
  current training accuracy:            0.815407
saving to ..., time used 805.644s
Seen 303000 samples, time used: 8117.048s
  current training loss:                0.520235
  current training accuracy:            0.815323
Seen 306000 samples, time used: 8197.744s
  current training loss:                0.520095
  current training accuracy:            0.815366
Seen 309000 samples, time used: 8279.945s
  current training loss:                0.520171
  current training accuracy:            0.815359
Seen 312000 samples, time used: 8362.039s
  current training loss:                0.520427
  current training accuracy:            0.815266
Seen 315000 samples, time used: 8442.305s
  current training loss:                0.520563
  current training accuracy:            0.815235
Seen 318000 samples, time used: 8522.084s
  current training loss:                0.520562
  current training accuracy:            0.815252
Seen 321000 samples, time used: 8600.861s
  current training loss:                0.520650
  current training accuracy:            0.815293
Seen 324000 samples, time used: 8680.423s
  current training loss:                0.520686
  current training accuracy:            0.815312
Seen 327000 samples, time used: 8760.347s
  current training loss:                0.520868
  current training accuracy:            0.815300
Seen 330000 samples, time used: 8842.048s
  current training loss:                0.520945
  current training accuracy:            0.815315
saving to ..., time used 803.805s
Seen 333000 samples, time used: 8923.568s
  current training loss:                0.521017
  current training accuracy:            0.815273
Seen 336000 samples, time used: 9003.341s
  current training loss:                0.521078
  current training accuracy:            0.815202
Seen 339000 samples, time used: 9084.216s
  current training loss:                0.521101
  current training accuracy:            0.815180
Seen 342000 samples, time used: 9164.168s
  current training loss:                0.521247
  current training accuracy:            0.815120
Seen 345000 samples, time used: 9245.205s
  current training loss:                0.521156
  current training accuracy:            0.815171
Seen 348000 samples, time used: 9325.958s
  current training loss:                0.521109
  current training accuracy:            0.815147
Seen 351000 samples, time used: 9408.503s
  current training loss:                0.521075
  current training accuracy:            0.815097
Seen 354000 samples, time used: 9488.543s
  current training loss:                0.521209
  current training accuracy:            0.815059
Seen 357000 samples, time used: 9567.890s
  current training loss:                0.521147
  current training accuracy:            0.815056
Seen 360000 samples, time used: 9648.171s
  current training loss:                0.521309
  current training accuracy:            0.814997
saving to ..., time used 805.587s
Seen 363000 samples, time used: 9729.219s
  current training loss:                0.521411
  current training accuracy:            0.814865
Seen 366000 samples, time used: 9810.545s
  current training loss:                0.521366
  current training accuracy:            0.814891
Seen 369000 samples, time used: 9890.473s
  current training loss:                0.521394
  current training accuracy:            0.814883
Seen 372000 samples, time used: 9972.867s
  current training loss:                0.521342
  current training accuracy:            0.814935
Seen 375000 samples, time used: 10054.323s
  current training loss:                0.521345
  current training accuracy:            0.814933
Seen 378000 samples, time used: 10135.439s
  current training loss:                0.521442
  current training accuracy:            0.814873
Seen 381000 samples, time used: 10217.851s
  current training loss:                0.521575
  current training accuracy:            0.814843
Seen 384000 samples, time used: 10298.850s
  current training loss:                0.521432
  current training accuracy:            0.814875
Seen 387000 samples, time used: 10379.336s
  current training loss:                0.521338
  current training accuracy:            0.814902
Seen 390000 samples, time used: 10460.925s
  current training loss:                0.521302
  current training accuracy:            0.814882
saving to ..., time used 812.254s
Seen 393000 samples, time used: 10542.140s
  current training loss:                0.521292
  current training accuracy:            0.814835
Seen 396000 samples, time used: 10622.023s
  current training loss:                0.521371
  current training accuracy:            0.814823
Seen 399000 samples, time used: 10702.485s
  current training loss:                0.521461
  current training accuracy:            0.814792
Seen 402000 samples, time used: 10783.245s
  current training loss:                0.521513
  current training accuracy:            0.814759
Seen 405000 samples, time used: 10863.212s
  current training loss:                0.521574
  current training accuracy:            0.814758
Seen 408000 samples, time used: 10943.604s
  current training loss:                0.521459
  current training accuracy:            0.814816
Seen 411000 samples, time used: 11023.830s
  current training loss:                0.521632
  current training accuracy:            0.814715
Seen 414000 samples, time used: 11104.830s
  current training loss:                0.521708
  current training accuracy:            0.814671
Seen 417000 samples, time used: 11184.725s
  current training loss:                0.521734
  current training accuracy:            0.814645
Seen 420000 samples, time used: 11265.010s
  current training loss:                0.521833
  current training accuracy:            0.814600
saving to ..., time used 803.568s
Seen 423000 samples, time used: 11346.228s
  current training loss:                0.521820
  current training accuracy:            0.814631
Seen 426000 samples, time used: 11427.790s
  current training loss:                0.521735
  current training accuracy:            0.814692
Seen 429000 samples, time used: 11507.782s
  current training loss:                0.521745
  current training accuracy:            0.814702
Seen 432000 samples, time used: 11588.910s
  current training loss:                0.521633
  current training accuracy:            0.814764
Seen 435000 samples, time used: 11667.693s
  current training loss:                0.521692
  current training accuracy:            0.814715
Seen 438000 samples, time used: 11748.517s
  current training loss:                0.521482
  current training accuracy:            0.814715
Seen 441000 samples, time used: 11827.913s
  current training loss:                0.521464
  current training accuracy:            0.814671
Seen 444000 samples, time used: 11908.872s
  current training loss:                0.521578
  current training accuracy:            0.814595
Seen 447000 samples, time used: 11988.788s
  current training loss:                0.521549
  current training accuracy:            0.814642
Seen 450000 samples, time used: 12071.086s
  current training loss:                0.521718
  current training accuracy:            0.814609
saving to ..., time used 805.577s
Seen 453000 samples, time used: 12152.464s
  current training loss:                0.521825
  current training accuracy:            0.814541
Seen 456000 samples, time used: 12232.340s
  current training loss:                0.521819
  current training accuracy:            0.814577
Seen 459000 samples, time used: 12312.778s
  current training loss:                0.521899
  current training accuracy:            0.814547
Seen 462000 samples, time used: 12394.355s
  current training loss:                0.521967
  current training accuracy:            0.814517
Seen 465000 samples, time used: 12475.456s
  current training loss:                0.521907
  current training accuracy:            0.814538
Seen 468000 samples, time used: 12555.936s
  current training loss:                0.521864
  current training accuracy:            0.814556
Seen 471000 samples, time used: 12638.502s
  current training loss:                0.521789
  current training accuracy:            0.814520
Seen 474000 samples, time used: 12718.077s
  current training loss:                0.521765
  current training accuracy:            0.814515
Seen 477000 samples, time used: 12798.529s
  current training loss:                0.521873
  current training accuracy:            0.814457
Seen 480000 samples, time used: 12879.082s
  current training loss:                0.521977
  current training accuracy:            0.814479
saving to ..., time used 807.509s
Seen 483000 samples, time used: 12962.827s
  current training loss:                0.522018
  current training accuracy:            0.814451
Seen 486000 samples, time used: 13044.162s
  current training loss:                0.521940
  current training accuracy:            0.814533

Seen 489000 samples, time used: 13125.951s
  current training loss:                0.522070
  current training accuracy:            0.814532
Seen 492000 samples, time used: 13205.917s
  current training loss:                0.522000
  current training accuracy:            0.814575
Seen 495000 samples, time used: 13285.935s
  current training loss:                0.522004
  current training accuracy:            0.814596
Seen 498000 samples, time used: 13365.742s
  current training loss:                0.522155
  current training accuracy:            0.814526
Seen 501000 samples, time used: 13448.162s
  current training loss:                0.522075
  current training accuracy:            0.814575
Seen 504000 samples, time used: 13528.888s
  current training loss:                0.522059
  current training accuracy:            0.814589
Seen 507000 samples, time used: 13608.982s
  current training loss:                0.522103
  current training accuracy:            0.814580
Seen 510000 samples, time used: 13690.230s
  current training loss:                0.522366
  current training accuracy:            0.814465
saving to ..., time used 810.640s
Seen 513000 samples, time used: 13772.277s
  current training loss:                0.522264
  current training accuracy:            0.814489
Seen 516000 samples, time used: 13852.602s
  current training loss:                0.522295
  current training accuracy:            0.814430
Seen 519000 samples, time used: 13935.832s
  current training loss:                0.522323
  current training accuracy:            0.814405
Seen 522000 samples, time used: 14018.262s
  current training loss:                0.522333
  current training accuracy:            0.814414
Seen 525000 samples, time used: 14103.205s
  current training loss:                0.522434
  current training accuracy:            0.814404
Seen 528000 samples, time used: 14184.035s
  current training loss:                0.522243
  current training accuracy:            0.814515
Seen 531000 samples, time used: 14267.081s
  current training loss:                0.522152
  current training accuracy:            0.814544
Seen 534000 samples, time used: 14348.762s
  current training loss:                0.522213
  current training accuracy:            0.814536
Seen 537000 samples, time used: 14429.515s
  current training loss:                0.522377
  current training accuracy:            0.814443
Seen 540000 samples, time used: 14509.412s
  current training loss:                0.522396
  current training accuracy:            0.814443
saving to ..., time used 818.501s
Seen 543000 samples, time used: 14591.159s
  current training loss:                0.522436
  current training accuracy:            0.814457
Seen 546000 samples, time used: 14673.057s
  current training loss:                0.522490
  current training accuracy:            0.814418
Seen 549000 samples, time used: 14753.488s
  current training loss:                0.522561
  current training accuracy:            0.814423
Epoch 6 of 20 took 14785.715s
  training loss:                0.522554
  training accuracy:            81.44 %
  validation loss:              0.483081
  validation accuracy:          81.12 %
  test loss:                    0.491040
  test accuracy:                80.62 %
saving to ./snli/word_by_word_model_epoch6.npz
Seen 3000 samples, time used: 79.478s
  current training loss:                0.513726
  current training accuracy:            0.818000
Seen 6000 samples, time used: 160.671s
  current training loss:                0.510879
  current training accuracy:            0.819667
Seen 9000 samples, time used: 239.832s
  current training loss:                0.505366
  current training accuracy:            0.824222
Seen 12000 samples, time used: 320.067s
  current training loss:                0.511472
  current training accuracy:            0.823167
Seen 15000 samples, time used: 399.823s
  current training loss:                0.512133
  current training accuracy:            0.822467
Seen 18000 samples, time used: 481.270s
  current training loss:                0.507905
  current training accuracy:            0.823333
Seen 21000 samples, time used: 560.767s
  current training loss:                0.511782
  current training accuracy:            0.821762
Seen 24000 samples, time used: 641.564s
  current training loss:                0.509260
  current training accuracy:            0.822583
Seen 27000 samples, time used: 722.777s
  current training loss:                0.507959
  current training accuracy:            0.823333
Seen 30000 samples, time used: 802.422s
  current training loss:                0.508908
  current training accuracy:            0.823200
saving to ..., time used 802.422s
Seen 33000 samples, time used: 884.461s
  current training loss:                0.510654
  current training accuracy:            0.822455
Seen 36000 samples, time used: 964.982s
  current training loss:                0.511130
  current training accuracy:            0.821944
Seen 39000 samples, time used: 1045.837s
  current training loss:                0.509940
  current training accuracy:            0.821949
Seen 42000 samples, time used: 1125.945s
  current training loss:                0.509060
  current training accuracy:            0.822714
Seen 45000 samples, time used: 1207.402s
  current training loss:                0.506679
  current training accuracy:            0.823533
Seen 48000 samples, time used: 1286.199s
  current training loss:                0.506168
  current training accuracy:            0.823687
Seen 51000 samples, time used: 1366.383s
  current training loss:                0.505844
  current training accuracy:            0.824235
Seen 54000 samples, time used: 1446.941s
  current training loss:                0.505766
  current training accuracy:            0.824148
Seen 57000 samples, time used: 1528.249s
  current training loss:                0.505327
  current training accuracy:            0.824228
Seen 60000 samples, time used: 1608.203s
  current training loss:                0.504481
  current training accuracy:            0.824300
saving to ..., time used 805.288s
Seen 63000 samples, time used: 1687.973s
  current training loss:                0.505011
  current training accuracy:            0.823984
Seen 66000 samples, time used: 1767.011s
  current training loss:                0.505063
  current training accuracy:            0.823970
Seen 69000 samples, time used: 1846.389s
  current training loss:                0.504540
  current training accuracy:            0.824145
Seen 72000 samples, time used: 1924.685s
  current training loss:                0.505126
  current training accuracy:            0.824250
Seen 75000 samples, time used: 2003.991s
  current training loss:                0.505614
  current training accuracy:            0.823973
Seen 78000 samples, time used: 2082.742s
  current training loss:                0.505134
  current training accuracy:            0.824205
Seen 81000 samples, time used: 2162.114s
  current training loss:                0.504846
  current training accuracy:            0.824642
Seen 84000 samples, time used: 2241.894s
  current training loss:                0.504781
  current training accuracy:            0.824548
Seen 87000 samples, time used: 2321.512s
  current training loss:                0.504625
  current training accuracy:            0.824517
Seen 90000 samples, time used: 2399.982s
  current training loss:                0.504586
  current training accuracy:            0.824544
saving to ..., time used 791.271s
Seen 93000 samples, time used: 2479.635s
  current training loss:                0.504790
  current training accuracy:            0.824516
Seen 96000 samples, time used: 2559.174s
  current training loss:                0.504448
  current training accuracy:            0.824896
Seen 99000 samples, time used: 2639.271s
  current training loss:                0.504709
  current training accuracy:            0.824667
Seen 102000 samples, time used: 2720.068s
  current training loss:                0.504548
  current training accuracy:            0.824843
Seen 105000 samples, time used: 2801.136s
  current training loss:                0.503993
  current training accuracy:            0.825076
Seen 108000 samples, time used: 2878.279s
  current training loss:                0.504661
  current training accuracy:            0.824704
Seen 111000 samples, time used: 2960.235s
  current training loss:                0.504639
  current training accuracy:            0.824514
Seen 114000 samples, time used: 3040.924s
  current training loss:                0.504818
  current training accuracy:            0.824325
Seen 117000 samples, time used: 3118.772s
  current training loss:                0.504936
  current training accuracy:            0.824188
Seen 120000 samples, time used: 3197.324s
  current training loss:                0.504647
  current training accuracy:            0.824283
saving to ..., time used 796.837s
Seen 123000 samples, time used: 3278.662s
  current training loss:                0.505045
  current training accuracy:            0.824179
Seen 126000 samples, time used: 3359.576s
  current training loss:                0.505524
  current training accuracy:            0.823921
Seen 129000 samples, time used: 3439.038s
  current training loss:                0.505819
  current training accuracy:            0.823713
Seen 132000 samples, time used: 3519.342s
  current training loss:                0.505919
  current training accuracy:            0.823826
Seen 135000 samples, time used: 3598.403s
  current training loss:                0.505987
  current training accuracy:            0.823733
Seen 138000 samples, time used: 3678.899s
  current training loss:                0.506220
  current training accuracy:            0.823717
Seen 141000 samples, time used: 3759.848s
  current training loss:                0.506523
  current training accuracy:            0.823610
Seen 144000 samples, time used: 3839.188s
  current training loss:                0.506375
  current training accuracy:            0.823632
Seen 147000 samples, time used: 3919.632s
  current training loss:                0.506332
  current training accuracy:            0.823531
Seen 150000 samples, time used: 3998.558s
  current training loss:                0.506482
  current training accuracy:            0.823533
saving to ..., time used 800.763s
Seen 153000 samples, time used: 4078.967s
  current training loss:                0.506657
  current training accuracy:            0.823320
Seen 156000 samples, time used: 4159.366s
  current training loss:                0.506855
  current training accuracy:            0.823256
Seen 159000 samples, time used: 4238.654s
  current training loss:                0.506902
  current training accuracy:            0.823233
Seen 162000 samples, time used: 4318.101s
  current training loss:                0.507111
  current training accuracy:            0.823247
Seen 165000 samples, time used: 4397.760s
  current training loss:                0.507195
  current training accuracy:            0.823158
Seen 168000 samples, time used: 4477.324s
  current training loss:                0.507237
  current training accuracy:            0.823179
Seen 171000 samples, time used: 4557.030s
  current training loss:                0.507290
  current training accuracy:            0.823187
Seen 174000 samples, time used: 4636.522s
  current training loss:                0.506912
  current training accuracy:            0.823540

Seen 177000 samples, time used: 4716.859s
  current training loss:                0.507276
  current training accuracy:            0.823458
Seen 180000 samples, time used: 4795.238s
  current training loss:                0.507462
  current training accuracy:            0.823311
saving to ..., time used 796.181s
Seen 183000 samples, time used: 4877.173s
  current training loss:                0.507439
  current training accuracy:            0.823410
Seen 186000 samples, time used: 4957.177s
  current training loss:                0.507612
  current training accuracy:            0.823301
Seen 189000 samples, time used: 5038.331s
  current training loss:                0.507672
  current training accuracy:            0.823143
Seen 192000 samples, time used: 5117.648s
  current training loss:                0.507959
  current training accuracy:            0.822896
Seen 195000 samples, time used: 5197.588s
  current training loss:                0.507763
  current training accuracy:            0.822841
Seen 198000 samples, time used: 5275.789s
  current training loss:                0.508157
  current training accuracy:            0.822732
Seen 201000 samples, time used: 5356.630s
  current training loss:                0.508323
  current training accuracy:            0.822627
Seen 204000 samples, time used: 5437.062s
  current training loss:                0.508334
  current training accuracy:            0.822520
Seen 207000 samples, time used: 5518.746s
  current training loss:                0.508468
  current training accuracy:            0.822401
Seen 210000 samples, time used: 5600.184s
  current training loss:                0.508612
  current training accuracy:            0.822210
saving to ..., time used 804.431s
Seen 213000 samples, time used: 5679.402s
  current training loss:                0.508201
  current training accuracy:            0.822361
Seen 216000 samples, time used: 5758.094s
  current training loss:                0.508169
  current training accuracy:            0.822394
Seen 219000 samples, time used: 5839.756s
  current training loss:                0.508178
  current training accuracy:            0.822356
Seen 222000 samples, time used: 5918.096s
  current training loss:                0.508044
  current training accuracy:            0.822464
Seen 225000 samples, time used: 5996.658s
  current training loss:                0.507867
  current training accuracy:            0.822520
Seen 228000 samples, time used: 6076.968s
  current training loss:                0.507919
  current training accuracy:            0.822443
Seen 231000 samples, time used: 6158.097s
  current training loss:                0.508054
  current training accuracy:            0.822333
Seen 234000 samples, time used: 6239.127s
  current training loss:                0.507939
  current training accuracy:            0.822303
Seen 237000 samples, time used: 6319.844s
  current training loss:                0.507889
  current training accuracy:            0.822354
Seen 240000 samples, time used: 6400.458s
  current training loss:                0.507931
  current training accuracy:            0.822421
saving to ..., time used 799.795s
Seen 243000 samples, time used: 6481.393s
  current training loss:                0.508154
  current training accuracy:            0.822267
Seen 246000 samples, time used: 6561.877s
  current training loss:                0.508073
  current training accuracy:            0.822187
Seen 249000 samples, time used: 6642.162s
  current training loss:                0.508375
  current training accuracy:            0.822181
Seen 252000 samples, time used: 6719.443s
  current training loss:                0.508596
  current training accuracy:            0.822000
Seen 255000 samples, time used: 6799.821s
  current training loss:                0.508798
  current training accuracy:            0.821824
Seen 258000 samples, time used: 6879.872s
  current training loss:                0.508884
  current training accuracy:            0.821779
Seen 261000 samples, time used: 6958.869s
  current training loss:                0.508686
  current training accuracy:            0.821908
Seen 264000 samples, time used: 7041.173s
  current training loss:                0.508823
  current training accuracy:            0.821905
Seen 267000 samples, time used: 7120.817s
  current training loss:                0.508755
  current training accuracy:            0.821888
Seen 270000 samples, time used: 7202.206s
  current training loss:                0.508878
  current training accuracy:            0.821896
saving to ..., time used 801.235s
Seen 273000 samples, time used: 7283.255s
  current training loss:                0.509172
  current training accuracy:            0.821788
Seen 276000 samples, time used: 7363.497s
  current training loss:                0.509044
  current training accuracy:            0.821844
Seen 279000 samples, time used: 7446.788s
  current training loss:                0.509074
  current training accuracy:            0.821728
Seen 282000 samples, time used: 7527.161s
  current training loss:                0.509221
  current training accuracy:            0.821667
Seen 285000 samples, time used: 7607.848s
  current training loss:                0.509320
  current training accuracy:            0.821646
Seen 288000 samples, time used: 7689.959s
  current training loss:                0.509449
  current training accuracy:            0.821660
Seen 291000 samples, time used: 7771.324s
  current training loss:                0.509613
  current training accuracy:            0.821467
Seen 294000 samples, time used: 7853.607s
  current training loss:                0.509754
  current training accuracy:            0.821412
Seen 297000 samples, time used: 7936.391s
  current training loss:                0.509701
  current training accuracy:            0.821418
Seen 300000 samples, time used: 8019.810s
  current training loss:                0.509617
  current training accuracy:            0.821397
saving to ..., time used 817.116s
Seen 303000 samples, time used: 8098.315s
  current training loss:                0.509987
  current training accuracy:            0.821284
Seen 306000 samples, time used: 8178.251s
  current training loss:                0.509993
  current training accuracy:            0.821291
Seen 309000 samples, time used: 8257.589s
  current training loss:                0.510081
  current training accuracy:            0.821201
Seen 312000 samples, time used: 8338.408s
  current training loss:                0.510041
  current training accuracy:            0.821202
Seen 315000 samples, time used: 8418.045s
  current training loss:                0.509800
  current training accuracy:            0.821324
Seen 318000 samples, time used: 8497.273s
  current training loss:                0.509830
  current training accuracy:            0.821245
Seen 321000 samples, time used: 8577.628s
  current training loss:                0.509931
  current training accuracy:            0.821227
Seen 324000 samples, time used: 8658.175s
  current training loss:                0.509897
  current training accuracy:            0.821269
Seen 327000 samples, time used: 8738.472s
  current training loss:                0.509781
  current training accuracy:            0.821281
Seen 330000 samples, time used: 8817.363s
  current training loss:                0.509751
  current training accuracy:            0.821221
saving to ..., time used 797.070s
Seen 333000 samples, time used: 8896.218s
  current training loss:                0.509613
  current training accuracy:            0.821279
Seen 336000 samples, time used: 8975.353s
  current training loss:                0.509626
  current training accuracy:            0.821301
Seen 339000 samples, time used: 9055.968s
  current training loss:                0.509633
  current training accuracy:            0.821342
Seen 342000 samples, time used: 9134.994s
  current training loss:                0.509645
  current training accuracy:            0.821246
Seen 345000 samples, time used: 9216.178s
  current training loss:                0.509818
  current training accuracy:            0.821136
Seen 348000 samples, time used: 9295.921s
  current training loss:                0.509751
  current training accuracy:            0.821164
Seen 351000 samples, time used: 9375.957s
  current training loss:                0.509923
  current training accuracy:            0.821094
Seen 354000 samples, time used: 9454.511s
  current training loss:                0.509878
  current training accuracy:            0.821119
Seen 357000 samples, time used: 9533.576s
  current training loss:                0.510007
  current training accuracy:            0.821048
Seen 360000 samples, time used: 9614.106s
  current training loss:                0.510110
  current training accuracy:            0.820967
saving to ..., time used 796.234s
Seen 363000 samples, time used: 9695.202s
  current training loss:                0.510071
  current training accuracy:            0.820967
Seen 366000 samples, time used: 9776.782s
  current training loss:                0.510050
  current training accuracy:            0.820945
Seen 369000 samples, time used: 9860.233s
  current training loss:                0.509894
  current training accuracy:            0.821049
Seen 372000 samples, time used: 9940.562s
  current training loss:                0.510037
  current training accuracy:            0.821051
Seen 375000 samples, time used: 10023.190s
  current training loss:                0.510004
  current training accuracy:            0.821035
Seen 378000 samples, time used: 10102.954s
  current training loss:                0.510087
  current training accuracy:            0.821040
Seen 381000 samples, time used: 10183.949s
  current training loss:                0.509937
  current training accuracy:            0.821136
Seen 384000 samples, time used: 10265.038s
  current training loss:                0.509823
  current training accuracy:            0.821159
Seen 387000 samples, time used: 10344.655s
  current training loss:                0.509842
  current training accuracy:            0.821085
Seen 390000 samples, time used: 10425.682s
  current training loss:                0.509925
  current training accuracy:            0.821074
saving to ..., time used 811.077s


Seen 393000 samples, time used: 10506.216s
  current training loss:                0.509796
  current training accuracy:            0.821114
Seen 396000 samples, time used: 10585.390s
  current training loss:                0.509788
  current training accuracy:            0.821058
Seen 399000 samples, time used: 10666.579s
  current training loss:                0.509836
  current training accuracy:            0.821015
Seen 402000 samples, time used: 10745.643s
  current training loss:                0.509739
  current training accuracy:            0.821040
Seen 405000 samples, time used: 10826.244s
  current training loss:                0.509835
  current training accuracy:            0.821020
Seen 408000 samples, time used: 10907.919s
  current training loss:                0.509672
  current training accuracy:            0.821125
Seen 411000 samples, time used: 10988.499s
  current training loss:                0.509626
  current training accuracy:            0.821112
Seen 414000 samples, time used: 11068.492s
  current training loss:                0.509708
  current training accuracy:            0.821089
Seen 417000 samples, time used: 11149.959s
  current training loss:                0.509683
  current training accuracy:            0.821094
Seen 420000 samples, time used: 11229.290s
  current training loss:                0.509784
  current training accuracy:            0.821098
saving to ..., time used 803.106s
Seen 423000 samples, time used: 11310.833s
  current training loss:                0.509869
  current training accuracy:            0.821021
Seen 426000 samples, time used: 11391.787s
  current training loss:                0.509873
  current training accuracy:            0.821012
Seen 429000 samples, time used: 11471.184s
  current training loss:                0.509795
  current training accuracy:            0.820988
Seen 432000 samples, time used: 11549.929s
  current training loss:                0.509919
  current training accuracy:            0.820995
Seen 435000 samples, time used: 11630.216s
  current training loss:                0.509900
  current training accuracy:            0.821014
Seen 438000 samples, time used: 11712.230s
  current training loss:                0.509895
  current training accuracy:            0.821080
Seen 441000 samples, time used: 11791.321s
  current training loss:                0.509940
  current training accuracy:            0.821061
Seen 444000 samples, time used: 11871.108s
  current training loss:                0.510074
  current training accuracy:            0.821000
Seen 447000 samples, time used: 11951.997s
  current training loss:                0.510189
  current training accuracy:            0.820951
Seen 450000 samples, time used: 12029.769s
  current training loss:                0.510139
  current training accuracy:            0.821018
saving to ..., time used 799.962s
Seen 453000 samples, time used: 12108.636s
  current training loss:                0.510261
  current training accuracy:            0.820925
Seen 456000 samples, time used: 12189.728s
  current training loss:                0.510202
  current training accuracy:            0.820914
Seen 459000 samples, time used: 12271.097s
  current training loss:                0.510205
  current training accuracy:            0.820906
Seen 462000 samples, time used: 12351.095s
  current training loss:                0.510331
  current training accuracy:            0.820909
Seen 465000 samples, time used: 12432.251s
  current training loss:                0.510435
  current training accuracy:            0.820806
Seen 468000 samples, time used: 12512.443s
  current training loss:                0.510358
  current training accuracy:            0.820840
Seen 471000 samples, time used: 12594.202s
  current training loss:                0.510373
  current training accuracy:            0.820820
Seen 474000 samples, time used: 12672.919s
  current training loss:                0.510309
  current training accuracy:            0.820844
Seen 477000 samples, time used: 12751.960s
  current training loss:                0.510372
  current training accuracy:            0.820809
Seen 480000 samples, time used: 12834.122s
  current training loss:                0.510463
  current training accuracy:            0.820748
saving to ..., time used 803.863s
Seen 483000 samples, time used: 12916.335s
  current training loss:                0.510564
  current training accuracy:            0.820696
Seen 486000 samples, time used: 12999.191s
  current training loss:                0.510659
  current training accuracy:            0.820654
Seen 489000 samples, time used: 13082.316s
  current training loss:                0.510786
  current training accuracy:            0.820599
Seen 492000 samples, time used: 13164.076s
  current training loss:                0.510785
  current training accuracy:            0.820606
Seen 495000 samples, time used: 13246.983s
  current training loss:                0.510854
  current training accuracy:            0.820610
Seen 498000 samples, time used: 13329.731s
  current training loss:                0.510766
  current training accuracy:            0.820647
Seen 501000 samples, time used: 13413.056s
  current training loss:                0.510760
  current training accuracy:            0.820623
Seen 504000 samples, time used: 13494.516s
  current training loss:                0.510814
  current training accuracy:            0.820631
Seen 507000 samples, time used: 13578.356s
  current training loss:                0.510796
  current training accuracy:            0.820675
Seen 510000 samples, time used: 13658.288s
  current training loss:                0.510819
  current training accuracy:            0.820635
saving to ..., time used 823.696s
Seen 513000 samples, time used: 13739.439s
  current training loss:                0.510863
  current training accuracy:            0.820651
Seen 516000 samples, time used: 13818.959s
  current training loss:                0.510740
  current training accuracy:            0.820692
Seen 519000 samples, time used: 13898.710s
  current training loss:                0.510769
  current training accuracy:            0.820634
Seen 522000 samples, time used: 13978.309s
  current training loss:                0.510891
  current training accuracy:            0.820565
Seen 525000 samples, time used: 14058.641s
  current training loss:                0.510951
  current training accuracy:            0.820543
Seen 528000 samples, time used: 14139.139s
  current training loss:                0.510981
  current training accuracy:            0.820555
Seen 531000 samples, time used: 14217.553s
  current training loss:                0.510935
  current training accuracy:            0.820576
Seen 534000 samples, time used: 14295.937s
  current training loss:                0.510995
  current training accuracy:            0.820519
Seen 537000 samples, time used: 14375.484s
  current training loss:                0.511036
  current training accuracy:            0.820492
Seen 540000 samples, time used: 14457.403s
  current training loss:                0.510983
  current training accuracy:            0.820474
saving to ..., time used 798.621s
Seen 543000 samples, time used: 14539.107s
  current training loss:                0.511092
  current training accuracy:            0.820433
Seen 546000 samples, time used: 14618.120s
  current training loss:                0.511202
  current training accuracy:            0.820364
Seen 549000 samples, time used: 14699.330s
  current training loss:                0.511328
  current training accuracy:            0.820306
Epoch 7 of 20 took 14731.585s
  training loss:                0.511326
  training accuracy:            82.03 %
  validation loss:              0.486155
  validation accuracy:          80.61 %
  test loss:                    0.495456
  test accuracy:                80.61 %
saving to ./snli/word_by_word_model_epoch7.npz
Seen 3000 samples, time used: 79.743s
  current training loss:                0.494515
  current training accuracy:            0.831000
Seen 6000 samples, time used: 160.916s
  current training loss:                0.500306
  current training accuracy:            0.823167
Seen 9000 samples, time used: 240.705s
  current training loss:                0.499333
  current training accuracy:            0.821111
Seen 12000 samples, time used: 320.345s
  current training loss:                0.505294
  current training accuracy:            0.819833
Seen 15000 samples, time used: 400.463s
  current training loss:                0.501240
  current training accuracy:            0.821533
Seen 18000 samples, time used: 478.839s
  current training loss:                0.503297
  current training accuracy:            0.822000
Seen 21000 samples, time used: 558.061s
  current training loss:                0.501360
  current training accuracy:            0.823857
Seen 24000 samples, time used: 639.083s
  current training loss:                0.500347
  current training accuracy:            0.822958

Seen 27000 samples, time used: 719.819s
  current training loss:                0.499388
  current training accuracy:            0.822889
Seen 30000 samples, time used: 801.398s
  current training loss:                0.498031
  current training accuracy:            0.823667
saving to ..., time used 801.398s
Seen 33000 samples, time used: 881.788s
  current training loss:                0.499180
  current training accuracy:            0.823909
Seen 36000 samples, time used: 962.000s
  current training loss:                0.497671
  current training accuracy:            0.824361
Seen 39000 samples, time used: 1040.828s
  current training loss:                0.496910
  current training accuracy:            0.824846
Seen 42000 samples, time used: 1120.382s
  current training loss:                0.498000
  current training accuracy:            0.824452
Seen 45000 samples, time used: 1200.839s
  current training loss:                0.496648
  current training accuracy:            0.825467
Seen 48000 samples, time used: 1279.683s
  current training loss:                0.498275
  current training accuracy:            0.825042
Seen 51000 samples, time used: 1358.935s
  current training loss:                0.498187
  current training accuracy:            0.825235
Seen 54000 samples, time used: 1437.565s
  current training loss:                0.498342
  current training accuracy:            0.825481
Seen 57000 samples, time used: 1518.649s
  current training loss:                0.498307
  current training accuracy:            0.825912
Seen 60000 samples, time used: 1597.635s
  current training loss:                0.497773
  current training accuracy:            0.826100
saving to ..., time used 795.746s
Seen 63000 samples, time used: 1678.877s
  current training loss:                0.496781
  current training accuracy:            0.826667
Seen 66000 samples, time used: 1760.721s
  current training loss:                0.496192
  current training accuracy:            0.826636
Seen 69000 samples, time used: 1839.765s
  current training loss:                0.495492
  current training accuracy:            0.827116
Seen 72000 samples, time used: 1920.793s
  current training loss:                0.495009
  current training accuracy:            0.827347
Seen 75000 samples, time used: 2001.005s
  current training loss:                0.495452
  current training accuracy:            0.826760
Seen 78000 samples, time used: 2081.949s
  current training loss:                0.495108
  current training accuracy:            0.826936
Seen 81000 samples, time used: 2160.498s
  current training loss:                0.495309
  current training accuracy:            0.826605
Seen 84000 samples, time used: 2241.327s
  current training loss:                0.495460
  current training accuracy:            0.826655
Seen 87000 samples, time used: 2320.490s
  current training loss:                0.495038
  current training accuracy:            0.826759
Seen 90000 samples, time used: 2401.035s
  current training loss:                0.495689
  current training accuracy:            0.826389
saving to ..., time used 802.895s
Seen 93000 samples, time used: 2483.539s
  current training loss:                0.496152
  current training accuracy:            0.826237
Seen 96000 samples, time used: 2563.603s
  current training loss:                0.496676
  current training accuracy:            0.826281
Seen 99000 samples, time used: 2644.911s
  current training loss:                0.496553
  current training accuracy:            0.826535
Seen 102000 samples, time used: 2723.983s
  current training loss:                0.496319
  current training accuracy:            0.826520
Seen 105000 samples, time used: 2803.885s
  current training loss:                0.496009
  current training accuracy:            0.826800
Seen 108000 samples, time used: 2883.409s
  current training loss:                0.496079
  current training accuracy:            0.826852

Seen 111000 samples, time used: 2962.473s
  current training loss:                0.495729
  current training accuracy:            0.827207
Seen 114000 samples, time used: 3041.727s
  current training loss:                0.496269
  current training accuracy:            0.826912
Seen 117000 samples, time used: 3120.184s
  current training loss:                0.496091
  current training accuracy:            0.827085
Seen 120000 samples, time used: 3199.101s
  current training loss:                0.495632
  current training accuracy:            0.827183
saving to ..., time used 797.582s
Seen 123000 samples, time used: 3280.740s
  current training loss:                0.495483
  current training accuracy:            0.827358
Seen 126000 samples, time used: 3362.818s
  current training loss:                0.495501
  current training accuracy:            0.827333
Seen 129000 samples, time used: 3444.724s
  current training loss:                0.496177
  current training accuracy:            0.827023
Seen 132000 samples, time used: 3526.526s
  current training loss:                0.495831
  current training accuracy:            0.827189
Seen 135000 samples, time used: 3608.323s
  current training loss:                0.495714
  current training accuracy:            0.827274
Seen 138000 samples, time used: 3689.013s
  current training loss:                0.495504
  current training accuracy:            0.827232
Seen 141000 samples, time used: 3769.491s
  current training loss:                0.495300
  current training accuracy:            0.827404
Seen 144000 samples, time used: 3850.196s
  current training loss:                0.495331
  current training accuracy:            0.827361
Seen 147000 samples, time used: 3933.439s
  current training loss:                0.495516
  current training accuracy:            0.827204
Seen 150000 samples, time used: 4014.676s
  current training loss:                0.495742
  current training accuracy:            0.826960
saving to ..., time used 814.995s
Seen 153000 samples, time used: 4094.514s
  current training loss:                0.496089
  current training accuracy:            0.826876
Seen 156000 samples, time used: 4174.719s
  current training loss:                0.496369
  current training accuracy:            0.826756
Seen 159000 samples, time used: 4254.005s
  current training loss:                0.496373
  current training accuracy:            0.826830
Seen 162000 samples, time used: 4333.753s
  current training loss:                0.496395
  current training accuracy:            0.826833
Seen 165000 samples, time used: 4412.505s
  current training loss:                0.496168
  current training accuracy:            0.827042
Seen 168000 samples, time used: 4491.463s
  current training loss:                0.496583
  current training accuracy:            0.826875
Seen 171000 samples, time used: 4572.010s
  current training loss:                0.496749
  current training accuracy:            0.826836
Seen 174000 samples, time used: 4652.486s
  current training loss:                0.497030
  current training accuracy:            0.826672
Seen 177000 samples, time used: 4731.740s
  current training loss:                0.496976
  current training accuracy:            0.826785
Seen 180000 samples, time used: 4810.539s
  current training loss:                0.496973
  current training accuracy:            0.826767
saving to ..., time used 795.306s
Seen 183000 samples, time used: 4892.887s
  current training loss:                0.496992
  current training accuracy:            0.826667
Seen 186000 samples, time used: 4975.443s
  current training loss:                0.496943
  current training accuracy:            0.826661
Seen 189000 samples, time used: 5056.246s
  current training loss:                0.497042
  current training accuracy:            0.826566
Seen 192000 samples, time used: 5135.983s
  current training loss:                0.496893
  current training accuracy:            0.826687
Seen 195000 samples, time used: 5218.874s
  current training loss:                0.496701
  current training accuracy:            0.826703
Seen 198000 samples, time used: 5301.444s
  current training loss:                0.496758
  current training accuracy:            0.826626
Seen 201000 samples, time used: 5383.696s
  current training loss:                0.496859
  current training accuracy:            0.826478
Seen 204000 samples, time used: 5465.268s
  current training loss:                0.497131
  current training accuracy:            0.826382
Seen 207000 samples, time used: 5546.358s
  current training loss:                0.497077
  current training accuracy:            0.826406
Seen 210000 samples, time used: 5627.183s
  current training loss:                0.497523
  current training accuracy:            0.826095
saving to ..., time used 816.081s
Seen 213000 samples, time used: 5708.531s
  current training loss:                0.497704
  current training accuracy:            0.825986
Seen 216000 samples, time used: 5792.819s
  current training loss:                0.497585
  current training accuracy:            0.826028
Seen 219000 samples, time used: 5874.490s
  current training loss:                0.497658
  current training accuracy:            0.826018
Seen 222000 samples, time used: 5957.114s
  current training loss:                0.497802
  current training accuracy:            0.825964
Seen 225000 samples, time used: 6037.234s
  current training loss:                0.497653
  current training accuracy:            0.826027
Seen 228000 samples, time used: 6117.001s
  current training loss:                0.498155
  current training accuracy:            0.825820
Seen 231000 samples, time used: 6199.652s
  current training loss:                0.498393
  current training accuracy:            0.825723
Seen 234000 samples, time used: 6279.649s
  current training loss:                0.498494
  current training accuracy:            0.825632
Seen 237000 samples, time used: 6362.936s
  current training loss:                0.498581
  current training accuracy:            0.825561
Seen 240000 samples, time used: 6445.248s
  current training loss:                0.498701
  current training accuracy:            0.825625
saving to ..., time used 817.531s
Seen 243000 samples, time used: 6524.041s
  current training loss:                0.498983
  current training accuracy:            0.825502
Seen 246000 samples, time used: 6604.975s
  current training loss:                0.498831
  current training accuracy:            0.825561
Seen 249000 samples, time used: 6684.337s
  current training loss:                0.498921
  current training accuracy:            0.825594
Seen 252000 samples, time used: 6761.974s
  current training loss:                0.499234
  current training accuracy:            0.825488
Seen 255000 samples, time used: 6842.099s
  current training loss:                0.499408
  current training accuracy:            0.825416
Seen 258000 samples, time used: 6922.121s
  current training loss:                0.499658
  current training accuracy:            0.825248
Seen 261000 samples, time used: 7000.796s
  current training loss:                0.499426
  current training accuracy:            0.825375
Seen 264000 samples, time used: 7082.135s
  current training loss:                0.499505
  current training accuracy:            0.825386
Seen 267000 samples, time used: 7162.248s
  current training loss:                0.499578
  current training accuracy:            0.825337
Seen 270000 samples, time used: 7243.212s
  current training loss:                0.499611
  current training accuracy:            0.825322
saving to ..., time used 797.417s
Seen 273000 samples, time used: 7325.833s
  current training loss:                0.499572
  current training accuracy:            0.825344
Seen 276000 samples, time used: 7406.393s
  current training loss:                0.499356
  current training accuracy:            0.825522
Seen 279000 samples, time used: 7488.409s
  current training loss:                0.499230
  current training accuracy:            0.825491
Seen 282000 samples, time used: 7570.088s
  current training loss:                0.499412
  current training accuracy:            0.825404
Seen 285000 samples, time used: 7649.424s
  current training loss:                0.499518
  current training accuracy:            0.825358
Seen 288000 samples, time used: 7729.219s
  current training loss:                0.499600
  current training accuracy:            0.825326
Seen 291000 samples, time used: 7808.392s
  current training loss:                0.499646
  current training accuracy:            0.825354
Seen 294000 samples, time used: 7888.907s
  current training loss:                0.499808
  current training accuracy:            0.825337
Seen 297000 samples, time used: 7968.308s
  current training loss:                0.499716
  current training accuracy:            0.825377
Seen 300000 samples, time used: 8048.331s
  current training loss:                0.499736
  current training accuracy:            0.825383
saving to ..., time used 804.584s
Seen 303000 samples, time used: 8128.782s
  current training loss:                0.499839
  current training accuracy:            0.825254
Seen 306000 samples, time used: 8209.626s
  current training loss:                0.499576
  current training accuracy:            0.825359
Seen 309000 samples, time used: 8290.662s
  current training loss:                0.499743
  current training accuracy:            0.825227
Seen 312000 samples, time used: 8369.213s
  current training loss:                0.499803
  current training accuracy:            0.825247
Seen 315000 samples, time used: 8448.662s
  current training loss:                0.500058
  current training accuracy:            0.825178
Seen 318000 samples, time used: 8527.540s
  current training loss:                0.500055
  current training accuracy:            0.825220
Seen 321000 samples, time used: 8607.318s
  current training loss:                0.499979
  current training accuracy:            0.825290
Seen 324000 samples, time used: 8687.951s
  current training loss:                0.500081
  current training accuracy:            0.825275
Seen 327000 samples, time used: 8766.840s
  current training loss:                0.500063
  current training accuracy:            0.825309
Seen 330000 samples, time used: 8846.160s
  current training loss:                0.500048
  current training accuracy:            0.825370
saving to ..., time used 797.287s
Seen 333000 samples, time used: 8926.070s
  current training loss:                0.500176
  current training accuracy:            0.825342
Seen 336000 samples, time used: 9005.018s
  current training loss:                0.499910
  current training accuracy:            0.825411
Seen 339000 samples, time used: 9084.725s
  current training loss:                0.499812
  current training accuracy:            0.825481
Seen 342000 samples, time used: 9165.363s
  current training loss:                0.499819
  current training accuracy:            0.825494
Seen 345000 samples, time used: 9245.623s
  current training loss:                0.499956
  current training accuracy:            0.825478
Seen 348000 samples, time used: 9324.713s
  current training loss:                0.499909
  current training accuracy:            0.825480
Seen 351000 samples, time used: 9403.244s
  current training loss:                0.499874
  current training accuracy:            0.825473
Seen 354000 samples, time used: 9482.052s
  current training loss:                0.500093
  current training accuracy:            0.825384
Seen 357000 samples, time used: 9562.198s
  current training loss:                0.499792
  current training accuracy:            0.825529
Seen 360000 samples, time used: 9643.326s
  current training loss:                0.499505
  current training accuracy:            0.825650
saving to ..., time used 796.649s
Seen 363000 samples, time used: 9723.626s
  current training loss:                0.499662
  current training accuracy:            0.825529
Seen 366000 samples, time used: 9804.648s
  current training loss:                0.499671
  current training accuracy:            0.825492
Seen 369000 samples, time used: 9883.942s
  current training loss:                0.499660
  current training accuracy:            0.825512
Seen 372000 samples, time used: 9963.935s
  current training loss:                0.499969
  current training accuracy:            0.825387
Seen 375000 samples, time used: 10043.935s
  current training loss:                0.500018
  current training accuracy:            0.825360
Seen 378000 samples, time used: 10124.402s
  current training loss:                0.500012
  current training accuracy:            0.825362
Seen 381000 samples, time used: 10203.619s
  current training loss:                0.500078
  current training accuracy:            0.825404
Seen 384000 samples, time used: 10285.729s
  current training loss:                0.500127
  current training accuracy:            0.825432
Seen 387000 samples, time used: 10364.554s
  current training loss:                0.500104
  current training accuracy:            0.825450
Seen 390000 samples, time used: 10444.555s
  current training loss:                0.500209
  current training accuracy:            0.825356
saving to ..., time used 800.708s
Seen 393000 samples, time used: 10523.856s
  current training loss:                0.500088
  current training accuracy:            0.825346
Seen 396000 samples, time used: 10603.237s
  current training loss:                0.500229
  current training accuracy:            0.825301
Seen 399000 samples, time used: 10683.281s
  current training loss:                0.500249
  current training accuracy:            0.825303
Seen 402000 samples, time used: 10761.610s
  current training loss:                0.500201
  current training accuracy:            0.825308
Seen 405000 samples, time used: 10844.035s
  current training loss:                0.500425
  current training accuracy:            0.825200
Seen 408000 samples, time used: 10923.669s
  current training loss:                0.500457
  current training accuracy:            0.825213
Seen 411000 samples, time used: 11002.733s
  current training loss:                0.500557
  current training accuracy:            0.825146
Seen 414000 samples, time used: 11081.849s
  current training loss:                0.500702
  current training accuracy:            0.825060
Seen 417000 samples, time used: 11160.128s
  current training loss:                0.500787
  current training accuracy:            0.825031
Seen 420000 samples, time used: 11239.481s
  current training loss:                0.500967
  current training accuracy:            0.824964
saving to ..., time used 794.439s
Seen 423000 samples, time used: 11318.422s
  current training loss:                0.500901
  current training accuracy:            0.824991
Seen 426000 samples, time used: 11398.992s
  current training loss:                0.501008
  current training accuracy:            0.824937
Seen 429000 samples, time used: 11479.750s
  current training loss:                0.501099
  current training accuracy:            0.824888
Seen 432000 samples, time used: 11561.443s
  current training loss:                0.500967
  current training accuracy:            0.824921
Seen 435000 samples, time used: 11640.541s
  current training loss:                0.500993
  current training accuracy:            0.824880
Seen 438000 samples, time used: 11720.419s
  current training loss:                0.501120
  current training accuracy:            0.824747
Seen 441000 samples, time used: 11800.004s
  current training loss:                0.501205
  current training accuracy:            0.824707
Seen 444000 samples, time used: 11878.387s
  current training loss:                0.501158
  current training accuracy:            0.824718
Seen 447000 samples, time used: 11957.516s
  current training loss:                0.501251
  current training accuracy:            0.824642
Seen 450000 samples, time used: 12038.196s
  current training loss:                0.501258
  current training accuracy:            0.824642
saving to ..., time used 798.129s
Seen 453000 samples, time used: 12118.118s
  current training loss:                0.501384
  current training accuracy:            0.824600
Seen 456000 samples, time used: 12197.573s
  current training loss:                0.501377
  current training accuracy:            0.824557
Seen 459000 samples, time used: 12277.292s
  current training loss:                0.501456
  current training accuracy:            0.824508
Seen 462000 samples, time used: 12356.097s
  current training loss:                0.501447
  current training accuracy:            0.824526
Seen 465000 samples, time used: 12436.006s
  current training loss:                0.501419
  current training accuracy:            0.824557
Seen 468000 samples, time used: 12516.083s
  current training loss:                0.501521
  current training accuracy:            0.824523
Seen 471000 samples, time used: 12595.101s
  current training loss:                0.501713
  current training accuracy:            0.824452
Seen 474000 samples, time used: 12675.451s
  current training loss:                0.501802
  current training accuracy:            0.824435
Seen 477000 samples, time used: 12754.270s
  current training loss:                0.501854
  current training accuracy:            0.824440
Seen 480000 samples, time used: 12834.175s
  current training loss:                0.501798
  current training accuracy:            0.824471
saving to ..., time used 795.391s
Seen 483000 samples, time used: 12914.769s
  current training loss:                0.501895
  current training accuracy:            0.824445
Seen 486000 samples, time used: 12992.931s
  current training loss:                0.501914
  current training accuracy:            0.824401
Seen 489000 samples, time used: 13072.105s
  current training loss:                0.501912
  current training accuracy:            0.824411
Seen 492000 samples, time used: 13152.641s
  current training loss:                0.501876
  current training accuracy:            0.824425
Seen 495000 samples, time used: 13232.766s
  current training loss:                0.501831
  current training accuracy:            0.824392
Seen 498000 samples, time used: 13311.940s
  current training loss:                0.501770
  current training accuracy:            0.824418
Seen 501000 samples, time used: 13390.261s
  current training loss:                0.501738
  current training accuracy:            0.824457
Seen 504000 samples, time used: 13472.159s
  current training loss:                0.501733
  current training accuracy:            0.824466
Seen 507000 samples, time used: 13552.056s
  current training loss:                0.501747
  current training accuracy:            0.824483
Seen 510000 samples, time used: 13630.929s
  current training loss:                0.501821
  current training accuracy:            0.824455
saving to ..., time used 796.227s
Seen 513000 samples, time used: 13711.540s
  current training loss:                0.501811
  current training accuracy:            0.824421
Seen 516000 samples, time used: 13791.356s
  current training loss:                0.501832
  current training accuracy:            0.824438
Seen 519000 samples, time used: 13871.790s
  current training loss:                0.501905
  current training accuracy:            0.824437
Seen 522000 samples, time used: 13950.410s
  current training loss:                0.501921
  current training accuracy:            0.824425
Seen 525000 samples, time used: 14029.484s
  current training loss:                0.501875
  current training accuracy:            0.824381
Seen 528000 samples, time used: 14109.242s
  current training loss:                0.501802
  current training accuracy:            0.824388
Seen 531000 samples, time used: 14188.629s
  current training loss:                0.501854
  current training accuracy:            0.824371
Seen 534000 samples, time used: 14267.526s
  current training loss:                0.501803
  current training accuracy:            0.824403
Seen 537000 samples, time used: 14346.597s
  current training loss:                0.501913
  current training accuracy:            0.824330
Seen 540000 samples, time used: 14426.511s
  current training loss:                0.501939
  current training accuracy:            0.824335
saving to ..., time used 795.098s
Seen 543000 samples, time used: 14507.584s
  current training loss:                0.502122
  current training accuracy:            0.824230
Seen 546000 samples, time used: 14587.904s
  current training loss:                0.502132
  current training accuracy:            0.824205
Seen 549000 samples, time used: 14668.568s
  current training loss:                0.502117
  current training accuracy:            0.824226
Epoch 8 of 20 took 14701.725s
  training loss:                0.502091
  training accuracy:            82.42 %
  validation loss:              0.476160
  validation accuracy:          81.83 %
  test loss:                    0.483392
  test accuracy:                81.03 %
saving to ./snli/word_by_word_model_epoch8.npz
Seen 3000 samples, time used: 81.856s
  current training loss:                0.482898
  current training accuracy:            0.838667
Seen 6000 samples, time used: 161.479s
  current training loss:                0.473547
  current training accuracy:            0.839500
Seen 9000 samples, time used: 241.034s
  current training loss:                0.467655
  current training accuracy:            0.840889
Seen 12000 samples, time used: 319.683s
  current training loss:                0.472807
  current training accuracy:            0.840333
Seen 15000 samples, time used: 398.447s
  current training loss:                0.473713
  current training accuracy:            0.838067
Seen 18000 samples, time used: 478.217s
  current training loss:                0.478450
  current training accuracy:            0.836500
Seen 21000 samples, time used: 557.367s
  current training loss:                0.480071
  current training accuracy:            0.836571
Seen 24000 samples, time used: 637.952s
  current training loss:                0.481840
  current training accuracy:            0.836083
Seen 27000 samples, time used: 717.816s
  current training loss:                0.480056
  current training accuracy:            0.836704
Seen 30000 samples, time used: 797.224s
  current training loss:                0.480453
  current training accuracy:            0.837033
saving to ..., time used 797.224s
Seen 33000 samples, time used: 877.350s
  current training loss:                0.482449
  current training accuracy:            0.836121
Seen 36000 samples, time used: 958.663s
  current training loss:                0.482360
  current training accuracy:            0.836056
Seen 39000 samples, time used: 1037.590s
  current training loss:                0.483704
  current training accuracy:            0.835692
Seen 42000 samples, time used: 1118.352s
  current training loss:                0.483808
  current training accuracy:            0.835452
Seen 45000 samples, time used: 1199.030s
  current training loss:                0.482672
  current training accuracy:            0.835689
Seen 48000 samples, time used: 1277.992s
  current training loss:                0.482288
  current training accuracy:            0.835458
Seen 51000 samples, time used: 1358.289s
  current training loss:                0.481430
  current training accuracy:            0.835725
Seen 54000 samples, time used: 1437.875s
  current training loss:                0.482212
  current training accuracy:            0.835241
Seen 57000 samples, time used: 1518.173s
  current training loss:                0.481967
  current training accuracy:            0.835439
Seen 60000 samples, time used: 1597.764s
  current training loss:                0.481675
  current training accuracy:            0.835667
saving to ..., time used 800.066s
Seen 63000 samples, time used: 1677.921s
  current training loss:                0.480753
  current training accuracy:            0.835968
Seen 66000 samples, time used: 1757.780s
  current training loss:                0.481348
  current training accuracy:            0.835667
Seen 69000 samples, time used: 1836.149s
  current training loss:                0.482918
  current training accuracy:            0.834797
Seen 72000 samples, time used: 1915.244s
  current training loss:                0.482926
  current training accuracy:            0.834639
Seen 75000 samples, time used: 1997.327s
  current training loss:                0.483986
  current training accuracy:            0.834200
Seen 78000 samples, time used: 2077.503s
  current training loss:                0.483650
  current training accuracy:            0.834423
Seen 81000 samples, time used: 2158.191s
  current training loss:                0.482697
  current training accuracy:            0.834815
Seen 84000 samples, time used: 2237.758s
  current training loss:                0.483669
  current training accuracy:            0.834500
Seen 87000 samples, time used: 2317.608s
  current training loss:                0.484229
  current training accuracy:            0.834172
Seen 90000 samples, time used: 2398.991s
  current training loss:                0.484728
  current training accuracy:            0.834056
saving to ..., time used 800.731s
Seen 93000 samples, time used: 2478.221s
  current training loss:                0.484980
  current training accuracy:            0.833828
Seen 96000 samples, time used: 2555.882s
  current training loss:                0.484486
  current training accuracy:            0.833969
Seen 99000 samples, time used: 2635.207s
  current training loss:                0.484935
  current training accuracy:            0.833909
Seen 102000 samples, time used: 2713.957s
  current training loss:                0.485720
  current training accuracy:            0.833490
Seen 105000 samples, time used: 2793.191s
  current training loss:                0.486343
  current training accuracy:            0.833267
Seen 108000 samples, time used: 2873.126s
  current training loss:                0.485727
  current training accuracy:            0.833713
Seen 111000 samples, time used: 2953.760s
  current training loss:                0.486188
  current training accuracy:            0.833405
Seen 114000 samples, time used: 3033.689s
  current training loss:                0.486470
  current training accuracy:            0.833307
Seen 117000 samples, time used: 3112.941s
  current training loss:                0.486485
  current training accuracy:            0.833231
Seen 120000 samples, time used: 3193.682s
  current training loss:                0.487048
  current training accuracy:            0.832908
saving to ..., time used 794.136s
Seen 123000 samples, time used: 3274.390s
  current training loss:                0.487114
  current training accuracy:            0.832862
Seen 126000 samples, time used: 3353.716s
  current training loss:                0.487331
  current training accuracy:            0.832833
Seen 129000 samples, time used: 3432.896s
  current training loss:                0.487787
  current training accuracy:            0.832752
Seen 132000 samples, time used: 3513.153s
  current training loss:                0.487784
  current training accuracy:            0.832758
Seen 135000 samples, time used: 3592.701s
  current training loss:                0.488045
  current training accuracy:            0.832556
Seen 138000 samples, time used: 3672.729s
  current training loss:                0.488377
  current training accuracy:            0.832413
Seen 141000 samples, time used: 3752.748s
  current training loss:                0.488515
  current training accuracy:            0.832383
Seen 144000 samples, time used: 3833.246s
  current training loss:                0.488428
  current training accuracy:            0.832465
Seen 147000 samples, time used: 3913.197s
  current training loss:                0.489081
  current training accuracy:            0.832204
Seen 150000 samples, time used: 3992.861s
  current training loss:                0.488889
  current training accuracy:            0.832327
saving to ..., time used 798.664s
Seen 153000 samples, time used: 4075.785s
  current training loss:                0.489496
  current training accuracy:            0.832046
Seen 156000 samples, time used: 4153.729s
  current training loss:                0.490081
  current training accuracy:            0.831910
Seen 159000 samples, time used: 4233.267s
  current training loss:                0.490309
  current training accuracy:            0.831843
Seen 162000 samples, time used: 4314.131s
  current training loss:                0.490271
  current training accuracy:            0.831840
Seen 165000 samples, time used: 4393.630s
  current training loss:                0.490109
  current training accuracy:            0.831976
Seen 168000 samples, time used: 4472.456s
  current training loss:                0.489585
  current training accuracy:            0.832268
Seen 171000 samples, time used: 4552.992s
  current training loss:                0.489824
  current training accuracy:            0.832187
Seen 174000 samples, time used: 4633.278s
  current training loss:                0.490258
  current training accuracy:            0.831862
Seen 177000 samples, time used: 4712.421s
  current training loss:                0.490224
  current training accuracy:            0.831853
Seen 180000 samples, time used: 4791.445s
  current training loss:                0.489954
  current training accuracy:            0.831906
saving to ..., time used 798.031s
Seen 183000 samples, time used: 4871.846s
  current training loss:                0.489863
  current training accuracy:            0.831923
Seen 186000 samples, time used: 4951.953s
  current training loss:                0.490146
  current training accuracy:            0.831812
Seen 189000 samples, time used: 5031.338s
  current training loss:                0.489740
  current training accuracy:            0.832026
Seen 192000 samples, time used: 5111.080s
  current training loss:                0.489854
  current training accuracy:            0.831979
Seen 195000 samples, time used: 5190.328s
  current training loss:                0.489845
  current training accuracy:            0.832005
Seen 198000 samples, time used: 5267.269s
  current training loss:                0.490334
  current training accuracy:            0.831813
Seen 201000 samples, time used: 5345.814s
  current training loss:                0.490088
  current training accuracy:            0.831831
Seen 204000 samples, time used: 5424.888s
  current training loss:                0.490073
  current training accuracy:            0.831843
Seen 207000 samples, time used: 5503.014s
  current training loss:                0.490410
  current training accuracy:            0.831734
Seen 210000 samples, time used: 5582.567s
  current training loss:                0.490228
  current training accuracy:            0.831819
saving to ..., time used 790.541s
Seen 213000 samples, time used: 5662.338s
  current training loss:                0.490320
  current training accuracy:            0.831873
Seen 216000 samples, time used: 5742.404s
  current training loss:                0.490318
  current training accuracy:            0.831856
Seen 219000 samples, time used: 5821.943s
  current training loss:                0.490657
  current training accuracy:            0.831726
Seen 222000 samples, time used: 5902.198s
  current training loss:                0.490693
  current training accuracy:            0.831667
Seen 225000 samples, time used: 5981.200s
  current training loss:                0.490853
  current training accuracy:            0.831596
Seen 228000 samples, time used: 6060.600s
  current training loss:                0.491467
  current training accuracy:            0.831364
Seen 231000 samples, time used: 6142.998s
  current training loss:                0.491313
  current training accuracy:            0.831494
Seen 234000 samples, time used: 6222.333s
  current training loss:                0.491297
  current training accuracy:            0.831479
Seen 237000 samples, time used: 6301.550s
  current training loss:                0.491240
  current training accuracy:            0.831435
Seen 240000 samples, time used: 6381.530s
  current training loss:                0.491497
  current training accuracy:            0.831321
saving to ..., time used 798.473s
Seen 243000 samples, time used: 6463.059s
  current training loss:                0.491567
  current training accuracy:            0.831284
Seen 246000 samples, time used: 6542.371s
  current training loss:                0.491509
  current training accuracy:            0.831211
Seen 249000 samples, time used: 6621.921s
  current training loss:                0.491544
  current training accuracy:            0.831177
Seen 252000 samples, time used: 6701.513s
  current training loss:                0.491606
  current training accuracy:            0.831115
Seen 255000 samples, time used: 6782.068s
  current training loss:                0.491564
  current training accuracy:            0.831153
Seen 258000 samples, time used: 6861.067s
  current training loss:                0.491798
  current training accuracy:            0.831062
Seen 261000 samples, time used: 6939.973s
  current training loss:                0.491910
  current training accuracy:            0.831008
Seen 264000 samples, time used: 7018.834s
  current training loss:                0.492147
  current training accuracy:            0.830886
Seen 267000 samples, time used: 7097.444s
  current training loss:                0.492180
  current training accuracy:            0.830828
Seen 270000 samples, time used: 7176.486s
  current training loss:                0.492142
  current training accuracy:            0.830870
saving to ..., time used 794.455s
Seen 273000 samples, time used: 7254.571s
  current training loss:                0.492241
  current training accuracy:            0.830780
Seen 276000 samples, time used: 7332.828s
  current training loss:                0.492258
  current training accuracy:            0.830692
Seen 279000 samples, time used: 7412.548s
  current training loss:                0.492487
  current training accuracy:            0.830563
Seen 282000 samples, time used: 7491.441s
  current training loss:                0.492530
  current training accuracy:            0.830525
Seen 285000 samples, time used: 7571.421s
  current training loss:                0.492610
  current training accuracy:            0.830502
Seen 288000 samples, time used: 7652.313s
  current training loss:                0.492636
  current training accuracy:            0.830420
Seen 291000 samples, time used: 7732.180s
  current training loss:                0.492615
  current training accuracy:            0.830354
Seen 294000 samples, time used: 7810.714s
  current training loss:                0.492726
  current training accuracy:            0.830299
Seen 297000 samples, time used: 7889.668s
  current training loss:                0.492983
  current training accuracy:            0.830239
Seen 300000 samples, time used: 7969.861s
  current training loss:                0.493028
  current training accuracy:            0.830163
saving to ..., time used 792.874s
Seen 303000 samples, time used: 8049.681s
  current training loss:                0.492887
  current training accuracy:            0.830201
Seen 306000 samples, time used: 8131.136s
  current training loss:                0.493003
  current training accuracy:            0.830134
Seen 309000 samples, time used: 8212.088s
  current training loss:                0.493080
  current training accuracy:            0.830055
Seen 312000 samples, time used: 8291.947s
  current training loss:                0.492947
  current training accuracy:            0.830045
Seen 315000 samples, time used: 8371.262s
  current training loss:                0.492816
  current training accuracy:            0.830124
Seen 318000 samples, time used: 8450.802s
  current training loss:                0.492803
  current training accuracy:            0.830113
Seen 321000 samples, time used: 8530.157s
  current training loss:                0.492706
  current training accuracy:            0.830062
Seen 324000 samples, time used: 8609.495s
  current training loss:                0.492980
  current training accuracy:            0.829932
Seen 327000 samples, time used: 8688.322s
  current training loss:                0.492999
  current training accuracy:            0.829948
Seen 330000 samples, time used: 8765.424s
  current training loss:                0.493076
  current training accuracy:            0.829942
saving to ..., time used 795.100s
Seen 333000 samples, time used: 8844.093s
  current training loss:                0.493296
  current training accuracy:            0.829799
Seen 336000 samples, time used: 8924.286s
  current training loss:                0.493361
  current training accuracy:            0.829780
Seen 339000 samples, time used: 9004.273s
  current training loss:                0.493488
  current training accuracy:            0.829755
Seen 342000 samples, time used: 9084.422s
  current training loss:                0.493770
  current training accuracy:            0.829649
Seen 345000 samples, time used: 9164.199s
  current training loss:                0.493759
  current training accuracy:            0.829655
Seen 348000 samples, time used: 9243.825s
  current training loss:                0.493696
  current training accuracy:            0.829647
Seen 351000 samples, time used: 9324.774s
  current training loss:                0.493616
  current training accuracy:            0.829624
Seen 354000 samples, time used: 9404.909s
  current training loss:                0.493642
  current training accuracy:            0.829633
Seen 357000 samples, time used: 9483.783s
  current training loss:                0.493583
  current training accuracy:            0.829597
Seen 360000 samples, time used: 9562.787s
  current training loss:                0.493636
  current training accuracy:            0.829561
saving to ..., time used 796.860s
Seen 363000 samples, time used: 9643.220s
  current training loss:                0.493580
  current training accuracy:            0.829590
Seen 366000 samples, time used: 9723.728s
  current training loss:                0.493628
  current training accuracy:            0.829587
Seen 369000 samples, time used: 9804.668s
  current training loss:                0.493503
  current training accuracy:            0.829659
Seen 372000 samples, time used: 9883.694s
  current training loss:                0.493538
  current training accuracy:            0.829610
Seen 375000 samples, time used: 9962.835s
  current training loss:                0.493595
  current training accuracy:            0.829560
Seen 378000 samples, time used: 10041.133s
  current training loss:                0.493469
  current training accuracy:            0.829611
Seen 381000 samples, time used: 10119.392s
  current training loss:                0.493554
  current training accuracy:            0.829593
Seen 384000 samples, time used: 10197.230s
  current training loss:                0.493488
  current training accuracy:            0.829664
Seen 387000 samples, time used: 10276.390s
  current training loss:                0.493615
  current training accuracy:            0.829553
Seen 390000 samples, time used: 10356.172s
  current training loss:                0.493654
  current training accuracy:            0.829518
saving to ..., time used 792.888s
Seen 393000 samples, time used: 10435.512s
  current training loss:                0.493563
  current training accuracy:            0.829557
Seen 396000 samples, time used: 10515.835s
  current training loss:                0.493552
  current training accuracy:            0.829588
Seen 399000 samples, time used: 10593.505s
  current training loss:                0.493735
  current training accuracy:            0.829536
Seen 402000 samples, time used: 10673.175s
  current training loss:                0.493657
  current training accuracy:            0.829614
Seen 405000 samples, time used: 10753.714s
  current training loss:                0.493657
  current training accuracy:            0.829622
Seen 408000 samples, time used: 10834.203s
  current training loss:                0.493914
  current training accuracy:            0.829485
Seen 411000 samples, time used: 10913.236s
  current training loss:                0.493968
  current training accuracy:            0.829489
Seen 414000 samples, time used: 10992.548s
  current training loss:                0.494155
  current training accuracy:            0.829461
Seen 417000 samples, time used: 11072.697s
  current training loss:                0.494190
  current training accuracy:            0.829434
Seen 420000 samples, time used: 11150.859s
  current training loss:                0.494155
  current training accuracy:            0.829400
saving to ..., time used 794.203s
Seen 423000 samples, time used: 11230.697s
  current training loss:                0.494134
  current training accuracy:            0.829409
Seen 426000 samples, time used: 11309.230s
  current training loss:                0.494251
  current training accuracy:            0.829347
Seen 429000 samples, time used: 11388.250s
  current training loss:                0.494187
  current training accuracy:            0.829357
Seen 432000 samples, time used: 11466.950s
  current training loss:                0.494157
  current training accuracy:            0.829366
Seen 435000 samples, time used: 11544.452s
  current training loss:                0.494314
  current training accuracy:            0.829267
Seen 438000 samples, time used: 11624.082s
  current training loss:                0.494484
  current training accuracy:            0.829130
Seen 441000 samples, time used: 11702.579s
  current training loss:                0.494625
  current training accuracy:            0.829068
Seen 444000 samples, time used: 11781.822s
  current training loss:                0.494714
  current training accuracy:            0.829036
Seen 447000 samples, time used: 11860.012s
  current training loss:                0.494795
  current training accuracy:            0.828998
Seen 450000 samples, time used: 11938.597s
  current training loss:                0.494860
  current training accuracy:            0.828964
saving to ..., time used 787.176s
Seen 453000 samples, time used: 12017.290s
  current training loss:                0.494847
  current training accuracy:            0.828980
Seen 456000 samples, time used: 12097.594s
  current training loss:                0.494789
  current training accuracy:            0.828980
Seen 459000 samples, time used: 12176.799s
  current training loss:                0.494904
  current training accuracy:            0.828887
Seen 462000 samples, time used: 12257.669s
  current training loss:                0.494883
  current training accuracy:            0.828903
Seen 465000 samples, time used: 12336.057s
  current training loss:                0.495004
  current training accuracy:            0.828785
Seen 468000 samples, time used: 12414.984s
  current training loss:                0.495077
  current training accuracy:            0.828812
Seen 471000 samples, time used: 12493.820s
  current training loss:                0.495136
  current training accuracy:            0.828758
Seen 474000 samples, time used: 12572.466s
  current training loss:                0.495056
  current training accuracy:            0.828766
Seen 477000 samples, time used: 12651.668s
  current training loss:                0.495118
  current training accuracy:            0.828740
Seen 480000 samples, time used: 12730.711s
  current training loss:                0.495237
  current training accuracy:            0.828679
saving to ..., time used 791.519s
Seen 483000 samples, time used: 12810.667s
  current training loss:                0.495043
  current training accuracy:            0.828735
Seen 486000 samples, time used: 12888.883s
  current training loss:                0.494956
  current training accuracy:            0.828767
Seen 489000 samples, time used: 12966.686s
  current training loss:                0.494979
  current training accuracy:            0.828789
Seen 492000 samples, time used: 13046.187s
  current training loss:                0.495022
  current training accuracy:            0.828833
Seen 495000 samples, time used: 13125.194s
  current training loss:                0.495082
  current training accuracy:            0.828810
Seen 498000 samples, time used: 13204.334s
  current training loss:                0.495095
  current training accuracy:            0.828807
Seen 501000 samples, time used: 13284.025s
  current training loss:                0.495096
  current training accuracy:            0.828816
Seen 504000 samples, time used: 13362.931s
  current training loss:                0.495009
  current training accuracy:            0.828798
Seen 507000 samples, time used: 13442.168s
  current training loss:                0.495031
  current training accuracy:            0.828842
Seen 510000 samples, time used: 13520.188s
  current training loss:                0.495219
  current training accuracy:            0.828761
saving to ..., time used 788.959s
Seen 513000 samples, time used: 13600.259s
  current training loss:                0.495140
  current training accuracy:            0.828780
Seen 516000 samples, time used: 13679.443s
  current training loss:                0.495241
  current training accuracy:            0.828771
Seen 519000 samples, time used: 13758.734s
  current training loss:                0.495426
  current training accuracy:            0.828723
Seen 522000 samples, time used: 13837.322s
  current training loss:                0.495309
  current training accuracy:            0.828741
Seen 525000 samples, time used: 13916.098s
  current training loss:                0.495180
  current training accuracy:            0.828796
Seen 528000 samples, time used: 13994.194s
  current training loss:                0.495162
  current training accuracy:            0.828833
Seen 531000 samples, time used: 14072.601s
  current training loss:                0.495248
  current training accuracy:            0.828793
Seen 534000 samples, time used: 14150.414s
  current training loss:                0.495264
  current training accuracy:            0.828794
Seen 537000 samples, time used: 14228.512s
  current training loss:                0.495311
  current training accuracy:            0.828782
Seen 540000 samples, time used: 14307.414s
  current training loss:                0.495347
  current training accuracy:            0.828759
saving to ..., time used 786.744s
Seen 543000 samples, time used: 14387.110s
  current training loss:                0.495269
  current training accuracy:            0.828779
Seen 546000 samples, time used: 14467.092s
  current training loss:                0.495266
  current training accuracy:            0.828766
Seen 549000 samples, time used: 14548.023s
  current training loss:                0.495267
  current training accuracy:            0.828754
Epoch 9 of 20 took 14580.295s
  training loss:                0.495234
  training accuracy:            82.88 %
  validation loss:              0.468213
  validation accuracy:          82.05 %
  test loss:                    0.489718
  test accuracy:                80.87 %
saving to ./snli/word_by_word_model_epoch9.npz
Seen 3000 samples, time used: 79.210s
  current training loss:                0.451661
  current training accuracy:            0.840667
Seen 6000 samples, time used: 158.445s
  current training loss:                0.469315
  current training accuracy:            0.833000
Seen 9000 samples, time used: 238.092s
  current training loss:                0.464845
  current training accuracy:            0.835556
Seen 12000 samples, time used: 316.969s
  current training loss:                0.464105
  current training accuracy:            0.835917
Seen 15000 samples, time used: 396.335s
  current training loss:                0.463232
  current training accuracy:            0.836733
Seen 18000 samples, time used: 476.584s
  current training loss:                0.468222
  current training accuracy:            0.836167
Seen 21000 samples, time used: 556.380s
  current training loss:                0.467004
  current training accuracy:            0.837667
Seen 24000 samples, time used: 636.095s
  current training loss:                0.467543
  current training accuracy:            0.838542
Seen 27000 samples, time used: 714.759s
  current training loss:                0.468894
  current training accuracy:            0.838111
Seen 30000 samples, time used: 792.565s
  current training loss:                0.467890
  current training accuracy:            0.838700
saving to ..., time used 792.565s
Seen 33000 samples, time used: 872.666s
  current training loss:                0.469079
  current training accuracy:            0.838485
Seen 36000 samples, time used: 952.424s
  current training loss:                0.470532
  current training accuracy:            0.838250
Seen 39000 samples, time used: 1034.865s
  current training loss:                0.471098
  current training accuracy:            0.838026
Seen 42000 samples, time used: 1115.944s
  current training loss:                0.472353
  current training accuracy:            0.837381
Seen 45000 samples, time used: 1197.134s
  current training loss:                0.472803
  current training accuracy:            0.836622
Seen 48000 samples, time used: 1276.941s
  current training loss:                0.472743
  current training accuracy:            0.836396
Seen 51000 samples, time used: 1357.062s
  current training loss:                0.473381
  current training accuracy:            0.836196
Seen 54000 samples, time used: 1437.121s
  current training loss:                0.473612
  current training accuracy:            0.836796
Seen 57000 samples, time used: 1516.209s
  current training loss:                0.474200
  current training accuracy:            0.836667
Seen 60000 samples, time used: 1596.080s
  current training loss:                0.474982
  current training accuracy:            0.836283
saving to ..., time used 802.957s
Seen 63000 samples, time used: 1677.686s
  current training loss:                0.474410
  current training accuracy:            0.836540
Seen 66000 samples, time used: 1759.165s
  current training loss:                0.475060
  current training accuracy:            0.836697
Seen 69000 samples, time used: 1840.262s
  current training loss:                0.475776
  current training accuracy:            0.836246
Seen 72000 samples, time used: 1922.917s
  current training loss:                0.475449
  current training accuracy:            0.836458
Seen 75000 samples, time used: 2005.188s
  current training loss:                0.475993
  current training accuracy:            0.836373
Seen 78000 samples, time used: 2086.956s
  current training loss:                0.476570
  current training accuracy:            0.836359
Seen 81000 samples, time used: 2168.989s
  current training loss:                0.477815
  current training accuracy:            0.836025
Seen 84000 samples, time used: 2250.275s
  current training loss:                0.477612
  current training accuracy:            0.836119
Seen 87000 samples, time used: 2332.319s
  current training loss:                0.477625
  current training accuracy:            0.836299
Seen 90000 samples, time used: 2416.019s
  current training loss:                0.477503
  current training accuracy:            0.836456
saving to ..., time used 819.372s
Seen 93000 samples, time used: 2498.678s
  current training loss:                0.477413
  current training accuracy:            0.836333
Seen 96000 samples, time used: 2578.744s
  current training loss:                0.477176
  current training accuracy:            0.836448
Seen 99000 samples, time used: 2658.861s
  current training loss:                0.477609
  current training accuracy:            0.836071
Seen 102000 samples, time used: 2738.084s
  current training loss:                0.478335
  current training accuracy:            0.835922
Seen 105000 samples, time used: 2818.773s
  current training loss:                0.478553
  current training accuracy:            0.835790
Seen 108000 samples, time used: 2898.477s
  current training loss:                0.478540
  current training accuracy:            0.835574
Seen 111000 samples, time used: 2978.372s
  current training loss:                0.478333
  current training accuracy:            0.835504
Seen 114000 samples, time used: 3059.071s
  current training loss:                0.478217
  current training accuracy:            0.835596
Seen 117000 samples, time used: 3140.408s
  current training loss:                0.478729
  current training accuracy:            0.835385
Seen 120000 samples, time used: 3220.512s
  current training loss:                0.479112
  current training accuracy:            0.835108
saving to ..., time used 803.651s
Seen 123000 samples, time used: 3301.136s
  current training loss:                0.479168
  current training accuracy:            0.835130
Seen 126000 samples, time used: 3381.568s
  current training loss:                0.479221
  current training accuracy:            0.834889
Seen 129000 samples, time used: 3461.976s
  current training loss:                0.479608
  current training accuracy:            0.834961
Seen 132000 samples, time used: 3542.318s
  current training loss:                0.480061
  current training accuracy:            0.834826
Seen 135000 samples, time used: 3623.592s
  current training loss:                0.480432
  current training accuracy:            0.834474
Seen 138000 samples, time used: 3703.336s
  current training loss:                0.480270
  current training accuracy:            0.834500
Seen 141000 samples, time used: 3783.864s
  current training loss:                0.480671
  current training accuracy:            0.834355
Seen 144000 samples, time used: 3863.239s
  current training loss:                0.480931
  current training accuracy:            0.834299
Seen 147000 samples, time used: 3944.376s
  current training loss:                0.480858
  current training accuracy:            0.834374
Seen 150000 samples, time used: 4023.573s
  current training loss:                0.481166
  current training accuracy:            0.834307
saving to ..., time used 802.523s
Seen 153000 samples, time used: 4104.188s
  current training loss:                0.481298
  current training accuracy:            0.834190
Seen 156000 samples, time used: 4184.903s
  current training loss:                0.481295
  current training accuracy:            0.834250
Seen 159000 samples, time used: 4264.430s
  current training loss:                0.481485
  current training accuracy:            0.834252
Seen 162000 samples, time used: 4345.965s
  current training loss:                0.481608
  current training accuracy:            0.834296
Seen 165000 samples, time used: 4426.074s
  current training loss:                0.481095
  current training accuracy:            0.834430
Seen 168000 samples, time used: 4504.782s
  current training loss:                0.480881
  current training accuracy:            0.834458
Seen 171000 samples, time used: 4584.518s
  current training loss:                0.481084
  current training accuracy:            0.834444
Seen 174000 samples, time used: 4664.289s
  current training loss:                0.480926
  current training accuracy:            0.834420
Seen 177000 samples, time used: 4743.463s
  current training loss:                0.481191
  current training accuracy:            0.834407
Seen 180000 samples, time used: 4822.706s
  current training loss:                0.481382
  current training accuracy:            0.834344
saving to ..., time used 798.573s
Seen 183000 samples, time used: 4904.749s
  current training loss:                0.481133
  current training accuracy:            0.834437
Seen 186000 samples, time used: 4988.103s
  current training loss:                0.481291
  current training accuracy:            0.834333
Seen 189000 samples, time used: 5070.139s
  current training loss:                0.481207
  current training accuracy:            0.834354
Seen 192000 samples, time used: 5151.536s
  current training loss:                0.481521
  current training accuracy:            0.834198
Seen 195000 samples, time used: 5234.570s
  current training loss:                0.481873
  current training accuracy:            0.834010
Seen 198000 samples, time used: 5316.835s
  current training loss:                0.482183
  current training accuracy:            0.833692
Seen 201000 samples, time used: 5399.183s
  current training loss:                0.482135
  current training accuracy:            0.833711
Seen 204000 samples, time used: 5480.704s
  current training loss:                0.482075
  current training accuracy:            0.833760
Seen 207000 samples, time used: 5562.277s
  current training loss:                0.482242
  current training accuracy:            0.833657
Seen 210000 samples, time used: 5646.055s
  current training loss:                0.482210
  current training accuracy:            0.833676
saving to ..., time used 822.725s
Seen 213000 samples, time used: 5728.349s
  current training loss:                0.482331
  current training accuracy:            0.833638
Seen 216000 samples, time used: 5810.222s
  current training loss:                0.482507
  current training accuracy:            0.833606
Seen 219000 samples, time used: 5891.518s
  current training loss:                0.482724
  current training accuracy:            0.833466
Seen 222000 samples, time used: 5972.563s
  current training loss:                0.482623
  current training accuracy:            0.833572
Seen 225000 samples, time used: 6054.232s
  current training loss:                0.482699
  current training accuracy:            0.833569
Seen 228000 samples, time used: 6134.147s
  current training loss:                0.482992
  current training accuracy:            0.833443
Seen 231000 samples, time used: 6214.318s
  current training loss:                0.483156
  current training accuracy:            0.833329
Seen 234000 samples, time used: 6298.362s
  current training loss:                0.483092
  current training accuracy:            0.833282
Seen 237000 samples, time used: 6378.498s
  current training loss:                0.483080
  current training accuracy:            0.833257
Seen 240000 samples, time used: 6460.464s
  current training loss:                0.483183
  current training accuracy:            0.833304
saving to ..., time used 813.881s
Seen 243000 samples, time used: 6542.932s
  current training loss:                0.483287
  current training accuracy:            0.833300
Seen 246000 samples, time used: 6622.930s
  current training loss:                0.483303
  current training accuracy:            0.833297
Seen 249000 samples, time used: 6703.957s
  current training loss:                0.483455
  current training accuracy:            0.833201
Seen 252000 samples, time used: 6786.407s
  current training loss:                0.483591
  current training accuracy:            0.833099
Seen 255000 samples, time used: 6869.393s
  current training loss:                0.483805
  current training accuracy:            0.833055
Seen 258000 samples, time used: 6949.178s
  current training loss:                0.483524
  current training accuracy:            0.833089
Seen 261000 samples, time used: 7028.859s
  current training loss:                0.483209
  current training accuracy:            0.833245
Seen 264000 samples, time used: 7110.288s
  current training loss:                0.483530
  current training accuracy:            0.833155
Seen 267000 samples, time used: 7190.959s
  current training loss:                0.483482
  current training accuracy:            0.833232
Seen 270000 samples, time used: 7272.945s
  current training loss:                0.483701
  current training accuracy:            0.833111
saving to ..., time used 811.925s
Seen 273000 samples, time used: 7353.425s
  current training loss:                0.483944
  current training accuracy:            0.832985
Seen 276000 samples, time used: 7433.574s
  current training loss:                0.483939
  current training accuracy:            0.832946
Seen 279000 samples, time used: 7512.418s
  current training loss:                0.483988
  current training accuracy:            0.832903
Seen 282000 samples, time used: 7593.617s
  current training loss:                0.484145
  current training accuracy:            0.832812
Seen 285000 samples, time used: 7673.268s
  current training loss:                0.484209
  current training accuracy:            0.832730
Seen 288000 samples, time used: 7752.476s
  current training loss:                0.484357
  current training accuracy:            0.832604
Seen 291000 samples, time used: 7830.889s
  current training loss:                0.484355
  current training accuracy:            0.832605
Seen 294000 samples, time used: 7911.063s
  current training loss:                0.484326
  current training accuracy:            0.832653
Seen 297000 samples, time used: 7989.655s
  current training loss:                0.484221
  current training accuracy:            0.832653
Seen 300000 samples, time used: 8069.693s
  current training loss:                0.484251
  current training accuracy:            0.832590
saving to ..., time used 796.157s
Seen 303000 samples, time used: 8151.177s
  current training loss:                0.484309
  current training accuracy:            0.832554
Seen 306000 samples, time used: 8232.242s
  current training loss:                0.484302
  current training accuracy:            0.832565
Seen 309000 samples, time used: 8315.471s
  current training loss:                0.484252
  current training accuracy:            0.832605
Seen 312000 samples, time used: 8397.700s
  current training loss:                0.484384
  current training accuracy:            0.832574
Seen 315000 samples, time used: 8478.877s
  current training loss:                0.484742
  current training accuracy:            0.832416
Seen 318000 samples, time used: 8562.012s
  current training loss:                0.484856
  current training accuracy:            0.832380
Seen 321000 samples, time used: 8643.764s
  current training loss:                0.484952
  current training accuracy:            0.832346
Seen 324000 samples, time used: 8724.527s
  current training loss:                0.485094
  current training accuracy:            0.832250
Seen 327000 samples, time used: 8807.255s
  current training loss:                0.485103
  current training accuracy:            0.832235
Seen 330000 samples, time used: 8889.000s
  current training loss:                0.485430
  current training accuracy:            0.832061
saving to ..., time used 818.747s
Seen 333000 samples, time used: 8967.273s
  current training loss:                0.485155
  current training accuracy:            0.832165
Seen 336000 samples, time used: 9047.263s
  current training loss:                0.485019
  current training accuracy:            0.832247
Seen 339000 samples, time used: 9126.882s
  current training loss:                0.485111
  current training accuracy:            0.832227
Seen 342000 samples, time used: 9207.245s
  current training loss:                0.485351
  current training accuracy:            0.832158
Seen 345000 samples, time used: 9286.973s
  current training loss:                0.485538
  current training accuracy:            0.832075
Seen 348000 samples, time used: 9366.019s
  current training loss:                0.485643
  current training accuracy:            0.832129
Seen 351000 samples, time used: 9444.702s
  current training loss:                0.485730
  current training accuracy:            0.832063
Seen 354000 samples, time used: 9524.937s
  current training loss:                0.485776
  current training accuracy:            0.832079
Seen 357000 samples, time used: 9606.438s
  current training loss:                0.485851
  current training accuracy:            0.832067
Seen 360000 samples, time used: 9685.704s
  current training loss:                0.485817
  current training accuracy:            0.832072
saving to ..., time used 796.153s
Seen 363000 samples, time used: 9766.557s
  current training loss:                0.485887
  current training accuracy:            0.832107
Seen 366000 samples, time used: 9846.447s
  current training loss:                0.485898
  current training accuracy:            0.832101
Seen 369000 samples, time used: 9925.411s
  current training loss:                0.485875
  current training accuracy:            0.832179
Seen 372000 samples, time used: 10006.275s
  current training loss:                0.486167
  current training accuracy:            0.832046
Seen 375000 samples, time used: 10086.351s
  current training loss:                0.486293
  current training accuracy:            0.832011
Seen 378000 samples, time used: 10166.589s
  current training loss:                0.486258
  current training accuracy:            0.832021
Seen 381000 samples, time used: 10248.640s
  current training loss:                0.486204
  current training accuracy:            0.832013
Seen 384000 samples, time used: 10327.993s
  current training loss:                0.486307
  current training accuracy:            0.832049
Seen 387000 samples, time used: 10408.561s
  current training loss:                0.486305
  current training accuracy:            0.832034
Seen 390000 samples, time used: 10489.481s
  current training loss:                0.486283
  current training accuracy:            0.832049
saving to ..., time used 803.276s
Seen 393000 samples, time used: 10571.848s
  current training loss:                0.486596
  current training accuracy:            0.831957
Seen 396000 samples, time used: 10652.245s
  current training loss:                0.486592
  current training accuracy:            0.831972
Seen 399000 samples, time used: 10731.177s
  current training loss:                0.486599
  current training accuracy:            0.832018
Seen 402000 samples, time used: 10812.685s
  current training loss:                0.486627
  current training accuracy:            0.832012
Seen 405000 samples, time used: 10893.309s
  current training loss:                0.486591
  current training accuracy:            0.832017
Seen 408000 samples, time used: 10971.649s
  current training loss:                0.486507
  current training accuracy:            0.832064
Seen 411000 samples, time used: 11051.482s
  current training loss:                0.486463
  current training accuracy:            0.832129
Seen 414000 samples, time used: 11128.884s
  current training loss:                0.486609
  current training accuracy:            0.832063
Seen 417000 samples, time used: 11208.765s
  current training loss:                0.486680
  current training accuracy:            0.832058
Seen 420000 samples, time used: 11286.884s
  current training loss:                0.486735
  current training accuracy:            0.831988
saving to ..., time used 796.866s
Seen 423000 samples, time used: 11369.804s
  current training loss:                0.486755
  current training accuracy:            0.831950
Seen 426000 samples, time used: 11452.002s
  current training loss:                0.486598
  current training accuracy:            0.831993
Seen 429000 samples, time used: 11534.190s
  current training loss:                0.486529
  current training accuracy:            0.832054
Seen 432000 samples, time used: 11616.317s
  current training loss:                0.486538
  current training accuracy:            0.832051
Seen 435000 samples, time used: 11698.505s
  current training loss:                0.486510
  current training accuracy:            0.832103
Seen 438000 samples, time used: 11780.759s
  current training loss:                0.486521
  current training accuracy:            0.832121
Seen 441000 samples, time used: 11863.000s
  current training loss:                0.486653
  current training accuracy:            0.832068
Seen 444000 samples, time used: 11944.415s
  current training loss:                0.486656
  current training accuracy:            0.832077
Seen 447000 samples, time used: 12027.655s
  current training loss:                0.486555
  current training accuracy:            0.832116
Seen 450000 samples, time used: 12110.855s
  current training loss:                0.486564
  current training accuracy:            0.832129
saving to ..., time used 823.422s
Seen 453000 samples, time used: 12191.938s
  current training loss:                0.486555
  current training accuracy:            0.832139
Seen 456000 samples, time used: 12274.878s
  current training loss:                0.486718
  current training accuracy:            0.832042
Seen 459000 samples, time used: 12359.024s
  current training loss:                0.486892
  current training accuracy:            0.831985
Seen 462000 samples, time used: 12442.267s
  current training loss:                0.487027
  current training accuracy:            0.831972
Seen 465000 samples, time used: 12523.987s
  current training loss:                0.487116
  current training accuracy:            0.831927
Seen 468000 samples, time used: 12604.927s
  current training loss:                0.487225
  current training accuracy:            0.831906
Seen 471000 samples, time used: 12688.603s
  current training loss:                0.487158
  current training accuracy:            0.831921
Seen 474000 samples, time used: 12770.137s
  current training loss:                0.487276
  current training accuracy:            0.831821
Seen 477000 samples, time used: 12850.481s
  current training loss:                0.487316
  current training accuracy:            0.831803
Seen 480000 samples, time used: 12931.834s
  current training loss:                0.487334
  current training accuracy:            0.831781
saving to ..., time used 820.459s
Seen 483000 samples, time used: 13012.624s
  current training loss:                0.487323
  current training accuracy:            0.831787
Seen 486000 samples, time used: 13092.734s
  current training loss:                0.487466
  current training accuracy:            0.831673
Seen 489000 samples, time used: 13172.671s
  current training loss:                0.487465
  current training accuracy:            0.831679
Seen 492000 samples, time used: 13253.594s
  current training loss:                0.487370
  current training accuracy:            0.831703
Seen 495000 samples, time used: 13332.342s
  current training loss:                0.487539
  current training accuracy:            0.831630
Seen 498000 samples, time used: 13414.815s
  current training loss:                0.487648
  current training accuracy:            0.831580
Seen 501000 samples, time used: 13493.904s
  current training loss:                0.487760
  current training accuracy:            0.831547
Seen 504000 samples, time used: 13575.343s
  current training loss:                0.487810
  current training accuracy:            0.831569
Seen 507000 samples, time used: 13656.096s
  current training loss:                0.487800
  current training accuracy:            0.831556
Seen 510000 samples, time used: 13736.352s
  current training loss:                0.487807
  current training accuracy:            0.831524
saving to ..., time used 803.897s
Seen 513000 samples, time used: 13817.628s
  current training loss:                0.487829
  current training accuracy:            0.831515
Seen 516000 samples, time used: 13898.139s
  current training loss:                0.487813
  current training accuracy:            0.831490
Seen 519000 samples, time used: 13979.249s
  current training loss:                0.487805
  current training accuracy:            0.831474
Seen 522000 samples, time used: 14058.656s
  current training loss:                0.487787
  current training accuracy:            0.831481
Seen 525000 samples, time used: 14140.108s
  current training loss:                0.487823
  current training accuracy:            0.831476
Seen 528000 samples, time used: 14220.518s
  current training loss:                0.487886
  current training accuracy:            0.831485
Seen 531000 samples, time used: 14299.952s
  current training loss:                0.487862
  current training accuracy:            0.831499
Seen 534000 samples, time used: 14379.881s
  current training loss:                0.487974
  current training accuracy:            0.831403
Seen 537000 samples, time used: 14459.618s
  current training loss:                0.487901
  current training accuracy:            0.831421
Seen 540000 samples, time used: 14541.271s
  current training loss:                0.488023
  current training accuracy:            0.831365
saving to ..., time used 804.355s
Seen 543000 samples, time used: 14619.395s
  current training loss:                0.487977
  current training accuracy:            0.831398
Seen 546000 samples, time used: 14698.391s
  current training loss:                0.487959
  current training accuracy:            0.831418
Seen 549000 samples, time used: 14778.705s
  current training loss:                0.488161
  current training accuracy:            0.831286
Epoch 10 of 20 took 14810.874s
  training loss:                0.488162
  training accuracy:            83.13 %
  validation loss:              0.472761
  validation accuracy:          81.87 %
  test loss:                    0.483354
  test accuracy:                81.16 %
saving to ./snli/word_by_word_model_epoch10.npz
Seen 3000 samples, time used: 79.215s
  current training loss:                0.462566
  current training accuracy:            0.840667
Seen 6000 samples, time used: 158.946s
  current training loss:                0.475301
  current training accuracy:            0.837667
Seen 9000 samples, time used: 238.552s
  current training loss:                0.478329
  current training accuracy:            0.836556
Seen 12000 samples, time used: 319.201s
  current training loss:                0.477960
  current training accuracy:            0.836250
Seen 15000 samples, time used: 399.267s
  current training loss:                0.475953
  current training accuracy:            0.837733
Seen 18000 samples, time used: 479.318s
  current training loss:                0.476385
  current training accuracy:            0.836111
Seen 21000 samples, time used: 558.882s
  current training loss:                0.473815
  current training accuracy:            0.836286
Seen 24000 samples, time used: 639.525s
  current training loss:                0.473980
  current training accuracy:            0.836125
Seen 27000 samples, time used: 721.164s
  current training loss:                0.473166
  current training accuracy:            0.837111
Seen 30000 samples, time used: 800.430s
  current training loss:                0.472943
  current training accuracy:            0.838333
saving to ..., time used 800.430s
Seen 33000 samples, time used: 880.054s
  current training loss:                0.474130
  current training accuracy:            0.837909
Seen 36000 samples, time used: 959.601s
  current training loss:                0.474717
  current training accuracy:            0.838333
Seen 39000 samples, time used: 1039.278s
  current training loss:                0.474612
  current training accuracy:            0.838590
Seen 42000 samples, time used: 1118.445s
  current training loss:                0.473972
  current training accuracy:            0.838643
Seen 45000 samples, time used: 1199.078s
  current training loss:                0.473987
  current training accuracy:            0.838956
Seen 48000 samples, time used: 1279.653s
  current training loss:                0.475345
  current training accuracy:            0.838354
Seen 51000 samples, time used: 1362.025s
  current training loss:                0.476089
  current training accuracy:            0.838294
Seen 54000 samples, time used: 1441.086s
  current training loss:                0.476902
  current training accuracy:            0.837833
Seen 57000 samples, time used: 1521.285s
  current training loss:                0.476723
  current training accuracy:            0.838000
Seen 60000 samples, time used: 1602.264s
  current training loss:                0.475618
  current training accuracy:            0.838417
saving to ..., time used 801.290s
Seen 63000 samples, time used: 1681.014s
  current training loss:                0.475195
  current training accuracy:            0.838841
Seen 66000 samples, time used: 1760.790s
  current training loss:                0.475339
  current training accuracy:            0.838864
Seen 69000 samples, time used: 1841.605s
  current training loss:                0.475289
  current training accuracy:            0.839000
Seen 72000 samples, time used: 1920.377s
  current training loss:                0.475189
  current training accuracy:            0.838958
Seen 75000 samples, time used: 2000.652s
  current training loss:                0.475286
  current training accuracy:            0.839107
Seen 78000 samples, time used: 2082.469s
  current training loss:                0.475389
  current training accuracy:            0.839051
Seen 81000 samples, time used: 2161.996s
  current training loss:                0.474763
  current training accuracy:            0.839210
Seen 84000 samples, time used: 2240.768s
  current training loss:                0.474951
  current training accuracy:            0.839381
Seen 87000 samples, time used: 2321.135s
  current training loss:                0.474774
  current training accuracy:            0.839356
Seen 90000 samples, time used: 2401.990s
  current training loss:                0.475012
  current training accuracy:            0.839300
saving to ..., time used 799.181s
Seen 93000 samples, time used: 2482.496s
  current training loss:                0.476155
  current training accuracy:            0.838731
Seen 96000 samples, time used: 2561.281s
  current training loss:                0.475642
  current training accuracy:            0.838865
Seen 99000 samples, time used: 2642.370s
  current training loss:                0.475022
  current training accuracy:            0.839141
Seen 102000 samples, time used: 2723.419s
  current training loss:                0.476012
  current training accuracy:            0.838765
Seen 105000 samples, time used: 2802.529s
  current training loss:                0.476235
  current training accuracy:            0.838371
Seen 108000 samples, time used: 2881.349s
  current training loss:                0.476637
  current training accuracy:            0.838065
Seen 111000 samples, time used: 2962.656s
  current training loss:                0.476111
  current training accuracy:            0.838189
Seen 114000 samples, time used: 3043.284s
  current training loss:                0.476700
  current training accuracy:            0.838184
Seen 117000 samples, time used: 3122.891s
  current training loss:                0.477185
  current training accuracy:            0.838009
Seen 120000 samples, time used: 3202.282s
  current training loss:                0.477041
  current training accuracy:            0.838000
saving to ..., time used 799.753s
Seen 123000 samples, time used: 3284.815s
  current training loss:                0.477313
  current training accuracy:            0.838008
Seen 126000 samples, time used: 3365.385s
  current training loss:                0.477155
  current training accuracy:            0.837968
Seen 129000 samples, time used: 3446.919s
  current training loss:                0.477275
  current training accuracy:            0.837891
Seen 132000 samples, time used: 3527.610s
  current training loss:                0.478133
  current training accuracy:            0.837477
Seen 135000 samples, time used: 3610.538s
  current training loss:                0.477850
  current training accuracy:            0.837526
Seen 138000 samples, time used: 3693.567s
  current training loss:                0.478041
  current training accuracy:            0.837580
Seen 141000 samples, time used: 3775.971s
  current training loss:                0.478375
  current training accuracy:            0.837326
Seen 144000 samples, time used: 3859.499s
  current training loss:                0.478133
  current training accuracy:            0.837465
Seen 147000 samples, time used: 3942.733s
  current training loss:                0.478405
  current training accuracy:            0.837279
Seen 150000 samples, time used: 4025.396s
  current training loss:                0.478622
  current training accuracy:            0.837187
saving to ..., time used 822.567s
Seen 153000 samples, time used: 4107.303s
  current training loss:                0.478380
  current training accuracy:            0.837203
Seen 156000 samples, time used: 4189.866s
  current training loss:                0.478424
  current training accuracy:            0.837173
Seen 159000 samples, time used: 4271.258s
  current training loss:                0.478472
  current training accuracy:            0.837283
Seen 162000 samples, time used: 4353.118s
  current training loss:                0.478520
  current training accuracy:            0.837265
Seen 165000 samples, time used: 4435.963s
  current training loss:                0.478722
  current training accuracy:            0.837194
Seen 168000 samples, time used: 4518.085s
  current training loss:                0.478361
  current training accuracy:            0.837185
Seen 171000 samples, time used: 4600.410s
  current training loss:                0.478244
  current training accuracy:            0.837322
Seen 174000 samples, time used: 4682.935s
  current training loss:                0.478055
  current training accuracy:            0.837385
Seen 177000 samples, time used: 4762.627s
  current training loss:                0.478120
  current training accuracy:            0.837271
Seen 180000 samples, time used: 4842.310s
  current training loss:                0.478076
  current training accuracy:            0.837200
saving to ..., time used 816.405s
Seen 183000 samples, time used: 4923.329s
  current training loss:                0.478023
  current training accuracy:            0.837120
Seen 186000 samples, time used: 5004.331s
  current training loss:                0.477789
  current training accuracy:            0.837263
Seen 189000 samples, time used: 5084.552s
  current training loss:                0.477811
  current training accuracy:            0.837233
Seen 192000 samples, time used: 5164.124s
  current training loss:                0.477979
  current training accuracy:            0.837214
Seen 195000 samples, time used: 5244.486s
  current training loss:                0.478115
  current training accuracy:            0.837108
Seen 198000 samples, time used: 5324.356s
  current training loss:                0.478228
  current training accuracy:            0.837066
Seen 201000 samples, time used: 5404.551s
  current training loss:                0.478416
  current training accuracy:            0.836886
Seen 204000 samples, time used: 5485.677s
  current training loss:                0.478383
  current training accuracy:            0.836887
Seen 207000 samples, time used: 5566.243s
  current training loss:                0.478221
  current training accuracy:            0.836952
Seen 210000 samples, time used: 5646.114s
  current training loss:                0.477877
  current training accuracy:            0.837043
saving to ..., time used 803.298s
Seen 213000 samples, time used: 5725.340s
  current training loss:                0.478216
  current training accuracy:            0.837023
Seen 216000 samples, time used: 5804.668s
  current training loss:                0.478382
  current training accuracy:            0.836949
Seen 219000 samples, time used: 5884.595s
  current training loss:                0.478321
  current training accuracy:            0.837050
Seen 222000 samples, time used: 5963.057s
  current training loss:                0.478578
  current training accuracy:            0.836901
Seen 225000 samples, time used: 6043.021s
  current training loss:                0.478553
  current training accuracy:            0.837009
Seen 228000 samples, time used: 6121.974s
  current training loss:                0.478651
  current training accuracy:            0.836895
Seen 231000 samples, time used: 6202.052s
  current training loss:                0.478641
  current training accuracy:            0.836892
Seen 234000 samples, time used: 6283.252s
  current training loss:                0.478584
  current training accuracy:            0.836838
Seen 237000 samples, time used: 6364.500s
  current training loss:                0.478819
  current training accuracy:            0.836722
Seen 240000 samples, time used: 6444.473s
  current training loss:                0.478780
  current training accuracy:            0.836762
saving to ..., time used 797.867s
Seen 243000 samples, time used: 6526.513s
  current training loss:                0.478545
  current training accuracy:            0.836765
Seen 246000 samples, time used: 6609.112s
  current training loss:                0.478384
  current training accuracy:            0.836736
Seen 249000 samples, time used: 6691.024s
  current training loss:                0.478550
  current training accuracy:            0.836667
Seen 252000 samples, time used: 6774.497s
  current training loss:                0.478640
  current training accuracy:            0.836619
Seen 255000 samples, time used: 6857.159s
  current training loss:                0.478413
  current training accuracy:            0.836722
Seen 258000 samples, time used: 6938.127s
  current training loss:                0.478387
  current training accuracy:            0.836651
Seen 261000 samples, time used: 7021.250s
  current training loss:                0.478506
  current training accuracy:            0.836544
Seen 264000 samples, time used: 7102.453s
  current training loss:                0.478537
  current training accuracy:            0.836485
Seen 267000 samples, time used: 7185.839s
  current training loss:                0.478661
  current training accuracy:            0.836457
Seen 270000 samples, time used: 7269.863s
  current training loss:                0.478570
  current training accuracy:            0.836504
saving to ..., time used 824.880s
Seen 273000 samples, time used: 7354.352s
  current training loss:                0.478606
  current training accuracy:            0.836473
Seen 276000 samples, time used: 7436.641s
  current training loss:                0.478779
  current training accuracy:            0.836384
Seen 279000 samples, time used: 7517.828s
  current training loss:                0.478990
  current training accuracy:            0.836312
Seen 282000 samples, time used: 7599.212s
  current training loss:                0.478982
  current training accuracy:            0.836284
Seen 285000 samples, time used: 7682.413s
  current training loss:                0.478978
  current training accuracy:            0.836295
Seen 288000 samples, time used: 7766.070s
  current training loss:                0.479099
  current training accuracy:            0.836236
Seen 291000 samples, time used: 7848.461s
  current training loss:                0.479033
  current training accuracy:            0.836254
Seen 294000 samples, time used: 7930.063s
  current training loss:                0.479021
  current training accuracy:            0.836214
Seen 297000 samples, time used: 8012.476s
  current training loss:                0.479040
  current training accuracy:            0.836226
Seen 300000 samples, time used: 8094.372s
  current training loss:                0.479110
  current training accuracy:            0.836200
saving to ..., time used 824.015s
Seen 303000 samples, time used: 8176.772s
  current training loss:                0.479108
  current training accuracy:            0.836267
Seen 306000 samples, time used: 8255.177s
  current training loss:                0.479014
  current training accuracy:            0.836255
Seen 309000 samples, time used: 8336.002s
  current training loss:                0.479047
  current training accuracy:            0.836191
Seen 312000 samples, time used: 8414.664s
  current training loss:                0.479007
  current training accuracy:            0.836122
Seen 315000 samples, time used: 8498.188s
  current training loss:                0.479369
  current training accuracy:            0.835997
Seen 318000 samples, time used: 8577.754s
  current training loss:                0.479277
  current training accuracy:            0.836031
Seen 321000 samples, time used: 8658.580s
  current training loss:                0.479209
  current training accuracy:            0.836065
Seen 324000 samples, time used: 8739.289s
  current training loss:                0.479489
  current training accuracy:            0.835901
Seen 327000 samples, time used: 8819.222s
  current training loss:                0.479538
  current training accuracy:            0.835841
Seen 330000 samples, time used: 8899.395s
  current training loss:                0.479461
  current training accuracy:            0.835870
saving to ..., time used 804.554s
Seen 333000 samples, time used: 8981.006s
  current training loss:                0.479544
  current training accuracy:            0.835808
Seen 336000 samples, time used: 9063.021s
  current training loss:                0.479293
  current training accuracy:            0.835875
Seen 339000 samples, time used: 9146.172s
  current training loss:                0.479465
  current training accuracy:            0.835788
Seen 342000 samples, time used: 9226.466s
  current training loss:                0.479756
  current training accuracy:            0.835667
Seen 345000 samples, time used: 9308.893s
  current training loss:                0.479574
  current training accuracy:            0.835783
Seen 348000 samples, time used: 9388.726s
  current training loss:                0.479544
  current training accuracy:            0.835796
Seen 351000 samples, time used: 9468.662s
  current training loss:                0.479731
  current training accuracy:            0.835687
Seen 354000 samples, time used: 9550.269s
  current training loss:                0.479647
  current training accuracy:            0.835788
Seen 357000 samples, time used: 9630.948s
  current training loss:                0.479612
  current training accuracy:            0.835793
Seen 360000 samples, time used: 9712.228s
  current training loss:                0.479606
  current training accuracy:            0.835828
saving to ..., time used 812.329s
Seen 363000 samples, time used: 9795.913s
  current training loss:                0.479528
  current training accuracy:            0.835840
Seen 366000 samples, time used: 9878.442s
  current training loss:                0.479558
  current training accuracy:            0.835863
Seen 369000 samples, time used: 9961.227s
  current training loss:                0.479735
  current training accuracy:            0.835707
Seen 372000 samples, time used: 10044.362s
  current training loss:                0.479738
  current training accuracy:            0.835731
Seen 375000 samples, time used: 10127.420s
  current training loss:                0.480008
  current training accuracy:            0.835629
Seen 378000 samples, time used: 10210.177s
  current training loss:                0.480184
  current training accuracy:            0.835622
Seen 381000 samples, time used: 10291.327s
  current training loss:                0.480197
  current training accuracy:            0.835593
Seen 384000 samples, time used: 10373.810s
  current training loss:                0.480163
  current training accuracy:            0.835638
Seen 387000 samples, time used: 10457.620s
  current training loss:                0.480168
  current training accuracy:            0.835636
Seen 390000 samples, time used: 10539.221s
  current training loss:                0.480274
  current training accuracy:            0.835618
saving to ..., time used 826.486s
Seen 393000 samples, time used: 10619.690s
  current training loss:                0.480472
  current training accuracy:            0.835499
Seen 396000 samples, time used: 10700.496s
  current training loss:                0.480532
  current training accuracy:            0.835460
Seen 399000 samples, time used: 10781.307s
  current training loss:                0.480555
  current training accuracy:            0.835466
Seen 402000 samples, time used: 10859.518s
  current training loss:                0.480650
  current training accuracy:            0.835413
Seen 405000 samples, time used: 10937.892s
  current training loss:                0.480870
  current training accuracy:            0.835316
Seen 408000 samples, time used: 11017.169s
  current training loss:                0.480998
  current training accuracy:            0.835319
Seen 411000 samples, time used: 11097.738s
  current training loss:                0.481225
  current training accuracy:            0.835248
Seen 414000 samples, time used: 11178.095s
  current training loss:                0.481432
  current training accuracy:            0.835104
Seen 417000 samples, time used: 11258.681s
  current training loss:                0.481538
  current training accuracy:            0.835036
Seen 420000 samples, time used: 11338.564s
  current training loss:                0.481574
  current training accuracy:            0.835062
saving to ..., time used 798.806s
Seen 423000 samples, time used: 11421.687s
  current training loss:                0.481597
  current training accuracy:            0.835040
Seen 426000 samples, time used: 11504.026s
  current training loss:                0.481680
  current training accuracy:            0.834988
Seen 429000 samples, time used: 11585.707s
  current training loss:                0.481824
  current training accuracy:            0.834923
Seen 432000 samples, time used: 11667.999s
  current training loss:                0.481894
  current training accuracy:            0.834910
Seen 435000 samples, time used: 11751.609s
  current training loss:                0.481739
  current training accuracy:            0.834989
Seen 438000 samples, time used: 11833.684s
  current training loss:                0.481745
  current training accuracy:            0.834961
Seen 441000 samples, time used: 11915.187s
  current training loss:                0.481918
  current training accuracy:            0.834878
Seen 444000 samples, time used: 11997.491s
  current training loss:                0.481902
  current training accuracy:            0.834892
Seen 447000 samples, time used: 12081.019s
  current training loss:                0.481940
  current training accuracy:            0.834935
Seen 450000 samples, time used: 12163.841s
  current training loss:                0.482074
  current training accuracy:            0.834924
saving to ..., time used 824.756s
Seen 453000 samples, time used: 12246.239s
  current training loss:                0.482110
  current training accuracy:            0.834859
Seen 456000 samples, time used: 12325.668s
  current training loss:                0.482105
  current training accuracy:            0.834851
Seen 459000 samples, time used: 12407.285s
  current training loss:                0.482244
  current training accuracy:            0.834765
Seen 462000 samples, time used: 12487.151s
  current training loss:                0.482279
  current training accuracy:            0.834714
Seen 465000 samples, time used: 12568.206s
  current training loss:                0.482310
  current training accuracy:            0.834708
Seen 468000 samples, time used: 12648.281s
  current training loss:                0.482495
  current training accuracy:            0.834682
Seen 471000 samples, time used: 12728.168s
  current training loss:                0.482513
  current training accuracy:            0.834671
Seen 474000 samples, time used: 12807.977s
  current training loss:                0.482584
  current training accuracy:            0.834593
Seen 477000 samples, time used: 12889.451s
  current training loss:                0.482687
  current training accuracy:            0.834512
Seen 480000 samples, time used: 12968.904s
  current training loss:                0.482706
  current training accuracy:            0.834554
saving to ..., time used 804.564s
Seen 483000 samples, time used: 13049.754s
  current training loss:                0.482744
  current training accuracy:            0.834528
Seen 486000 samples, time used: 13129.614s
  current training loss:                0.482709
  current training accuracy:            0.834549
Seen 489000 samples, time used: 13209.144s
  current training loss:                0.482720
  current training accuracy:            0.834556
Seen 492000 samples, time used: 13289.313s
  current training loss:                0.482832
  current training accuracy:            0.834541
Seen 495000 samples, time used: 13369.972s
  current training loss:                0.482806
  current training accuracy:            0.834527
Seen 498000 samples, time used: 13450.347s
  current training loss:                0.482895
  current training accuracy:            0.834464
Seen 501000 samples, time used: 13531.324s
  current training loss:                0.482987
  current training accuracy:            0.834433
Seen 504000 samples, time used: 13610.683s
  current training loss:                0.483049
  current training accuracy:            0.834349
Seen 507000 samples, time used: 13689.934s
  current training loss:                0.483222
  current training accuracy:            0.834300
Seen 510000 samples, time used: 13772.483s
  current training loss:                0.483288
  current training accuracy:            0.834257
saving to ..., time used 803.079s
Seen 513000 samples, time used: 13852.891s
  current training loss:                0.483137
  current training accuracy:            0.834327
Seen 516000 samples, time used: 13933.917s
  current training loss:                0.483118
  current training accuracy:            0.834320
Seen 519000 samples, time used: 14014.005s
  current training loss:                0.483237
  current training accuracy:            0.834301
Seen 522000 samples, time used: 14093.310s
  current training loss:                0.483292
  current training accuracy:            0.834262
Seen 525000 samples, time used: 14174.666s
  current training loss:                0.483227
  current training accuracy:            0.834257
Seen 528000 samples, time used: 14256.702s
  current training loss:                0.483255
  current training accuracy:            0.834212
Seen 531000 samples, time used: 14335.211s
  current training loss:                0.483305
  current training accuracy:            0.834154
Seen 534000 samples, time used: 14416.012s
  current training loss:                0.483306
  current training accuracy:            0.834159
Seen 537000 samples, time used: 14498.363s
  current training loss:                0.483287
  current training accuracy:            0.834175
Seen 540000 samples, time used: 14579.744s
  current training loss:                0.483424
  current training accuracy:            0.834104
saving to ..., time used 806.790s
Seen 543000 samples, time used: 14662.000s
  current training loss:                0.483452
  current training accuracy:            0.834096
Seen 546000 samples, time used: 14744.727s
  current training loss:                0.483473
  current training accuracy:            0.834093
Seen 549000 samples, time used: 14827.124s
  current training loss:                0.483608
  current training accuracy:            0.834055
Epoch 11 of 20 took 14870.299s
  training loss:                0.483585
  training accuracy:            83.40 %
  validation loss:              0.471878
  validation accuracy:          81.70 %
  test loss:                    0.486306
  test accuracy:                81.03 %
saving to ./snli/word_by_word_model_epoch11.npz
Seen 3000 samples, time used: 86.124s
  current training loss:                0.468287
  current training accuracy:            0.842667
Seen 6000 samples, time used: 168.779s
  current training loss:                0.464830
  current training accuracy:            0.842167
Seen 9000 samples, time used: 249.747s
  current training loss:                0.465932
  current training accuracy:            0.841667
Seen 12000 samples, time used: 331.928s
  current training loss:                0.468841
  current training accuracy:            0.840833
Seen 15000 samples, time used: 413.880s
  current training loss:                0.468634
  current training accuracy:            0.839200
Seen 18000 samples, time used: 495.583s
  current training loss:                0.469053
  current training accuracy:            0.840500
Seen 21000 samples, time used: 580.370s
  current training loss:                0.468560
  current training accuracy:            0.841048
Seen 24000 samples, time used: 663.083s
  current training loss:                0.466980
  current training accuracy:            0.842208
Seen 27000 samples, time used: 745.408s
  current training loss:                0.466615
  current training accuracy:            0.842074
Seen 30000 samples, time used: 826.436s
  current training loss:                0.467738
  current training accuracy:            0.841967
saving to ..., time used 826.436s
Seen 33000 samples, time used: 906.345s
  current training loss:                0.466579
  current training accuracy:            0.842576
Seen 36000 samples, time used: 986.867s
  current training loss:                0.467719
  current training accuracy:            0.842000
Seen 39000 samples, time used: 1066.048s
  current training loss:                0.467850
  current training accuracy:            0.842308
Seen 42000 samples, time used: 1145.572s
  current training loss:                0.469146
  current training accuracy:            0.841429
Seen 45000 samples, time used: 1225.483s
  current training loss:                0.470376
  current training accuracy:            0.841222
Seen 48000 samples, time used: 1302.890s
  current training loss:                0.471084
  current training accuracy:            0.840833
Seen 51000 samples, time used: 1382.567s
  current training loss:                0.472341
  current training accuracy:            0.839922
Seen 54000 samples, time used: 1461.711s
  current training loss:                0.471906
  current training accuracy:            0.839574
Seen 57000 samples, time used: 1542.570s
  current training loss:                0.471196
  current training accuracy:            0.839491
Seen 60000 samples, time used: 1621.607s
  current training loss:                0.471858
  current training accuracy:            0.839383
saving to ..., time used 794.667s
Seen 63000 samples, time used: 1700.872s
  current training loss:                0.471989
  current training accuracy:            0.839111
Seen 66000 samples, time used: 1780.187s
  current training loss:                0.472846
  current training accuracy:            0.838621
Seen 69000 samples, time used: 1861.346s
  current training loss:                0.473208
  current training accuracy:            0.838522
Seen 72000 samples, time used: 1940.560s
  current training loss:                0.473618
  current training accuracy:            0.838361
Seen 75000 samples, time used: 2020.943s
  current training loss:                0.474244
  current training accuracy:            0.838093
Seen 78000 samples, time used: 2101.519s
  current training loss:                0.474591
  current training accuracy:            0.837782
Seen 81000 samples, time used: 2181.895s
  current training loss:                0.474442
  current training accuracy:            0.837963
Seen 84000 samples, time used: 2261.724s
  current training loss:                0.474051
  current training accuracy:            0.838095
Seen 87000 samples, time used: 2342.054s
  current training loss:                0.474094
  current training accuracy:            0.838057
Seen 90000 samples, time used: 2422.129s
  current training loss:                0.474258
  current training accuracy:            0.838089
saving to ..., time used 800.012s
Seen 93000 samples, time used: 2502.221s
  current training loss:                0.475039
  current training accuracy:            0.837677
Seen 96000 samples, time used: 2582.903s
  current training loss:                0.474339
  current training accuracy:            0.838198
Seen 99000 samples, time used: 2661.595s
  current training loss:                0.474112
  current training accuracy:            0.838182
Seen 102000 samples, time used: 2740.597s
  current training loss:                0.474034
  current training accuracy:            0.838402
Seen 105000 samples, time used: 2821.232s
  current training loss:                0.473983
  current training accuracy:            0.838448
Seen 108000 samples, time used: 2901.987s
  current training loss:                0.473584
  current training accuracy:            0.838843
Seen 111000 samples, time used: 2982.192s
  current training loss:                0.473709
  current training accuracy:            0.838721
Seen 114000 samples, time used: 3062.865s
  current training loss:                0.473459
  current training accuracy:            0.838596
Seen 117000 samples, time used: 3142.942s
  current training loss:                0.473451
  current training accuracy:            0.838880
Seen 120000 samples, time used: 3221.013s
  current training loss:                0.473554
  current training accuracy:            0.838858
saving to ..., time used 798.304s
Seen 123000 samples, time used: 3302.511s
  current training loss:                0.473308
  current training accuracy:            0.838992
Seen 126000 samples, time used: 3382.600s
  current training loss:                0.472960
  current training accuracy:            0.839175
Seen 129000 samples, time used: 3461.927s
  current training loss:                0.472565
  current training accuracy:            0.839217
Seen 132000 samples, time used: 3542.275s
  current training loss:                0.472982
  current training accuracy:            0.839167
Seen 135000 samples, time used: 3622.893s
  current training loss:                0.473090
  current training accuracy:            0.839244
Seen 138000 samples, time used: 3703.278s
  current training loss:                0.472632
  current training accuracy:            0.839464
Seen 141000 samples, time used: 3782.896s
  current training loss:                0.472860
  current training accuracy:            0.839397
Seen 144000 samples, time used: 3864.834s
  current training loss:                0.472490
  current training accuracy:            0.839465
Seen 147000 samples, time used: 3944.997s
  current training loss:                0.472342
  current training accuracy:            0.839537
Seen 150000 samples, time used: 4025.554s
  current training loss:                0.472113
  current training accuracy:            0.839533
saving to ..., time used 804.045s
Seen 153000 samples, time used: 4105.323s
  current training loss:                0.472105
  current training accuracy:            0.839536
Seen 156000 samples, time used: 4185.890s
  current training loss:                0.472388
  current training accuracy:            0.839410
Seen 159000 samples, time used: 4266.116s
  current training loss:                0.472266
  current training accuracy:            0.839333
Seen 162000 samples, time used: 4346.502s
  current training loss:                0.472307
  current training accuracy:            0.839253
Seen 165000 samples, time used: 4429.301s
  current training loss:                0.471896
  current training accuracy:            0.839491
Seen 168000 samples, time used: 4510.494s
  current training loss:                0.471732
  current training accuracy:            0.839643
Seen 171000 samples, time used: 4590.103s
  current training loss:                0.472096
  current training accuracy:            0.839491
Seen 174000 samples, time used: 4670.455s
  current training loss:                0.472157
  current training accuracy:            0.839443
Seen 177000 samples, time used: 4750.390s
  current training loss:                0.472328
  current training accuracy:            0.839418
Seen 180000 samples, time used: 4830.377s
  current training loss:                0.472614
  current training accuracy:            0.839222
saving to ..., time used 804.331s
Seen 183000 samples, time used: 4911.104s
  current training loss:                0.472707
  current training accuracy:            0.839279
Seen 186000 samples, time used: 4991.792s
  current training loss:                0.472707
  current training accuracy:            0.839194
Seen 189000 samples, time used: 5074.290s
  current training loss:                0.473013
  current training accuracy:            0.839127
Seen 192000 samples, time used: 5156.328s
  current training loss:                0.473092
  current training accuracy:            0.839083
Seen 195000 samples, time used: 5238.209s
  current training loss:                0.473078
  current training accuracy:            0.839174
Seen 198000 samples, time used: 5318.268s
  current training loss:                0.473401
  current training accuracy:            0.838985
Seen 201000 samples, time used: 5400.307s
  current training loss:                0.473257
  current training accuracy:            0.839000
Seen 204000 samples, time used: 5479.316s
  current training loss:                0.473657
  current training accuracy:            0.838926
Seen 207000 samples, time used: 5560.821s
  current training loss:                0.473851
  current training accuracy:            0.838874
Seen 210000 samples, time used: 5640.141s
  current training loss:                0.473637
  current training accuracy:            0.838952
saving to ..., time used 809.283s
Seen 213000 samples, time used: 5721.196s
  current training loss:                0.473666
  current training accuracy:            0.838887
Seen 216000 samples, time used: 5800.746s
  current training loss:                0.474087
  current training accuracy:            0.838690
Seen 219000 samples, time used: 5880.185s
  current training loss:                0.474216
  current training accuracy:            0.838626
Seen 222000 samples, time used: 5960.332s
  current training loss:                0.474043
  current training accuracy:            0.838784
Seen 225000 samples, time used: 6040.022s
  current training loss:                0.474243
  current training accuracy:            0.838716
Seen 228000 samples, time used: 6121.230s
  current training loss:                0.474390
  current training accuracy:            0.838605
Seen 231000 samples, time used: 6200.017s
  current training loss:                0.474385
  current training accuracy:            0.838649
Seen 234000 samples, time used: 6278.508s
  current training loss:                0.474362
  current training accuracy:            0.838709
Seen 237000 samples, time used: 6359.128s
  current training loss:                0.474434
  current training accuracy:            0.838688
Seen 240000 samples, time used: 6440.440s
  current training loss:                0.474602
  current training accuracy:            0.838575
saving to ..., time used 799.813s
Seen 243000 samples, time used: 6521.867s
  current training loss:                0.474590
  current training accuracy:            0.838543
Seen 246000 samples, time used: 6602.829s
  current training loss:                0.474466
  current training accuracy:            0.838606
Seen 249000 samples, time used: 6684.297s
  current training loss:                0.474654
  current training accuracy:            0.838590
Seen 252000 samples, time used: 6763.700s
  current training loss:                0.474348
  current training accuracy:            0.838710
Seen 255000 samples, time used: 6844.823s
  current training loss:                0.474289
  current training accuracy:            0.838800
Seen 258000 samples, time used: 6924.972s
  current training loss:                0.474809
  current training accuracy:            0.838523
Seen 261000 samples, time used: 7007.450s
  current training loss:                0.474855
  current training accuracy:            0.838529
Seen 264000 samples, time used: 7087.749s
  current training loss:                0.475010
  current training accuracy:            0.838466
Seen 267000 samples, time used: 7167.871s
  current training loss:                0.474959
  current training accuracy:            0.838584
Seen 270000 samples, time used: 7248.703s
  current training loss:                0.474638
  current training accuracy:            0.838607
saving to ..., time used 807.786s
Seen 273000 samples, time used: 7330.800s
  current training loss:                0.474776
  current training accuracy:            0.838564
Seen 276000 samples, time used: 7411.229s
  current training loss:                0.474839
  current training accuracy:            0.838489
Seen 279000 samples, time used: 7491.819s
  current training loss:                0.474898
  current training accuracy:            0.838466
Seen 282000 samples, time used: 7572.867s
  current training loss:                0.475081
  current training accuracy:            0.838411
Seen 285000 samples, time used: 7654.073s
  current training loss:                0.475215
  current training accuracy:            0.838442
Seen 288000 samples, time used: 7735.185s
  current training loss:                0.475271
  current training accuracy:            0.838392
Seen 291000 samples, time used: 7815.908s
  current training loss:                0.475416
  current training accuracy:            0.838333
Seen 294000 samples, time used: 7895.018s
  current training loss:                0.475543
  current training accuracy:            0.838255
Seen 297000 samples, time used: 7975.765s
  current training loss:                0.475588
  current training accuracy:            0.838202
Seen 300000 samples, time used: 8057.100s
  current training loss:                0.475571
  current training accuracy:            0.838187
saving to ..., time used 807.889s
Seen 303000 samples, time used: 8139.825s
  current training loss:                0.475541
  current training accuracy:            0.838139
Seen 306000 samples, time used: 8222.348s
  current training loss:                0.475623
  current training accuracy:            0.838042
Seen 309000 samples, time used: 8306.035s
  current training loss:                0.475504
  current training accuracy:            0.838065
Seen 312000 samples, time used: 8389.302s
  current training loss:                0.475734
  current training accuracy:            0.837946
Seen 315000 samples, time used: 8469.616s
  current training loss:                0.475603
  current training accuracy:            0.838019
Seen 318000 samples, time used: 8552.696s
  current training loss:                0.475601
  current training accuracy:            0.838075

Seen 321000 samples, time used: 8633.403s
  current training loss:                0.475626
  current training accuracy:            0.838084
Seen 324000 samples, time used: 8713.526s
  current training loss:                0.475753
  current training accuracy:            0.838003
Seen 327000 samples, time used: 8793.166s
  current training loss:                0.475888
  current training accuracy:            0.837960
Seen 330000 samples, time used: 8873.571s
  current training loss:                0.475974
  current training accuracy:            0.837903
saving to ..., time used 815.979s
Seen 333000 samples, time used: 8953.934s
  current training loss:                0.475860
  current training accuracy:            0.837952
Seen 336000 samples, time used: 9034.561s
  current training loss:                0.475856
  current training accuracy:            0.837881
Seen 339000 samples, time used: 9115.190s
  current training loss:                0.476145
  current training accuracy:            0.837732
Seen 342000 samples, time used: 9197.492s
  current training loss:                0.476151
  current training accuracy:            0.837769
Seen 345000 samples, time used: 9276.853s
  current training loss:                0.476425
  current training accuracy:            0.837696
Seen 348000 samples, time used: 9356.655s
  current training loss:                0.476481
  current training accuracy:            0.837601
Seen 351000 samples, time used: 9435.800s
  current training loss:                0.476453
  current training accuracy:            0.837627
Seen 354000 samples, time used: 9515.662s
  current training loss:                0.476601
  current training accuracy:            0.837633
Seen 357000 samples, time used: 9595.635s
  current training loss:                0.476711
  current training accuracy:            0.837580
Seen 360000 samples, time used: 9675.364s
  current training loss:                0.476721
  current training accuracy:            0.837561
saving to ..., time used 801.290s
Seen 363000 samples, time used: 9755.296s
  current training loss:                0.476596
  current training accuracy:            0.837570
Seen 366000 samples, time used: 9834.664s
  current training loss:                0.476599
  current training accuracy:            0.837601
Seen 369000 samples, time used: 9914.919s
  current training loss:                0.476675
  current training accuracy:            0.837553
Seen 372000 samples, time used: 9992.915s
  current training loss:                0.477050
  current training accuracy:            0.837417
Seen 375000 samples, time used: 10071.386s
  current training loss:                0.477094
  current training accuracy:            0.837368
Seen 378000 samples, time used: 10151.701s
  current training loss:                0.477204
  current training accuracy:            0.837331
Seen 381000 samples, time used: 10230.734s
  current training loss:                0.477174
  current training accuracy:            0.837423
Seen 384000 samples, time used: 10310.690s
  current training loss:                0.477131
  current training accuracy:            0.837430
Seen 387000 samples, time used: 10389.993s
  current training loss:                0.477065
  current training accuracy:            0.837354
Seen 390000 samples, time used: 10471.622s
  current training loss:                0.477070
  current training accuracy:            0.837372
saving to ..., time used 795.755s
Seen 393000 samples, time used: 10552.525s
  current training loss:                0.477146
  current training accuracy:            0.837303
Seen 396000 samples, time used: 10633.981s
  current training loss:                0.477305
  current training accuracy:            0.837245
Seen 399000 samples, time used: 10711.954s
  current training loss:                0.477366
  current training accuracy:            0.837246
Seen 402000 samples, time used: 10792.242s
  current training loss:                0.477346
  current training accuracy:            0.837256
Seen 405000 samples, time used: 10871.528s
  current training loss:                0.477655
  current training accuracy:            0.837202
Seen 408000 samples, time used: 10953.345s
  current training loss:                0.477730
  current training accuracy:            0.837174
Seen 411000 samples, time used: 11032.752s
  current training loss:                0.477762
  current training accuracy:            0.837124
Seen 414000 samples, time used: 11112.719s
  current training loss:                0.477840
  current training accuracy:            0.837087
Seen 417000 samples, time used: 11194.792s
  current training loss:                0.477796
  current training accuracy:            0.837129
Seen 420000 samples, time used: 11274.273s
  current training loss:                0.477837
  current training accuracy:            0.837088
saving to ..., time used 802.071s
Seen 423000 samples, time used: 11354.591s
  current training loss:                0.477848
  current training accuracy:            0.837135
Seen 426000 samples, time used: 11434.228s
  current training loss:                0.478023
  current training accuracy:            0.837115
Seen 429000 samples, time used: 11514.912s
  current training loss:                0.478168
  current training accuracy:            0.837023
Seen 432000 samples, time used: 11595.332s
  current training loss:                0.478215
  current training accuracy:            0.836944
Seen 435000 samples, time used: 11674.021s
  current training loss:                0.478296
  current training accuracy:            0.836846
Seen 438000 samples, time used: 11754.778s
  current training loss:                0.478457
  current training accuracy:            0.836735
Seen 441000 samples, time used: 11833.557s
  current training loss:                0.478507
  current training accuracy:            0.836698
Seen 444000 samples, time used: 11913.173s
  current training loss:                0.478540
  current training accuracy:            0.836651
Seen 447000 samples, time used: 11993.228s
  current training loss:                0.478595
  current training accuracy:            0.836615
Seen 450000 samples, time used: 12073.881s
  current training loss:                0.478621
  current training accuracy:            0.836580
saving to ..., time used 799.046s
Seen 453000 samples, time used: 12154.224s
  current training loss:                0.478718
  current training accuracy:            0.836519
Seen 456000 samples, time used: 12233.634s
  current training loss:                0.478675
  current training accuracy:            0.836509
Seen 459000 samples, time used: 12313.127s
  current training loss:                0.478695
  current training accuracy:            0.836471
Seen 462000 samples, time used: 12392.522s
  current training loss:                0.478726
  current training accuracy:            0.836461
Seen 465000 samples, time used: 12472.328s
  current training loss:                0.478763
  current training accuracy:            0.836458
Seen 468000 samples, time used: 12550.054s
  current training loss:                0.478737
  current training accuracy:            0.836472
Seen 471000 samples, time used: 12627.624s
  current training loss:                0.478754
  current training accuracy:            0.836423
Seen 474000 samples, time used: 12709.164s
  current training loss:                0.478586
  current training accuracy:            0.836454
Seen 477000 samples, time used: 12787.809s
  current training loss:                0.478699
  current training accuracy:            0.836365
Seen 480000 samples, time used: 12866.334s
  current training loss:                0.478843
  current training accuracy:            0.836304
saving to ..., time used 791.961s
Seen 483000 samples, time used: 12947.049s
  current training loss:                0.478925
  current training accuracy:            0.836246
Seen 486000 samples, time used: 13024.968s
  current training loss:                0.478993
  current training accuracy:            0.836212
Seen 489000 samples, time used: 13103.319s
  current training loss:                0.479019
  current training accuracy:            0.836151
Seen 492000 samples, time used: 13183.839s
  current training loss:                0.479032
  current training accuracy:            0.836106
Seen 495000 samples, time used: 13263.834s
  current training loss:                0.479032
  current training accuracy:            0.836111
Seen 498000 samples, time used: 13342.632s
  current training loss:                0.478926
  current training accuracy:            0.836151
Seen 501000 samples, time used: 13421.291s
  current training loss:                0.478863
  current training accuracy:            0.836194
Seen 504000 samples, time used: 13501.431s
  current training loss:                0.478828
  current training accuracy:            0.836141
Seen 507000 samples, time used: 13580.042s
  current training loss:                0.478813
  current training accuracy:            0.836170
Seen 510000 samples, time used: 13659.995s
  current training loss:                0.478818
  current training accuracy:            0.836182
saving to ..., time used 793.160s
Seen 513000 samples, time used: 13739.387s
  current training loss:                0.478796
  current training accuracy:            0.836216
Seen 516000 samples, time used: 13818.850s
  current training loss:                0.478771
  current training accuracy:            0.836217
Seen 519000 samples, time used: 13897.765s
  current training loss:                0.478850
  current training accuracy:            0.836204
Seen 522000 samples, time used: 13976.521s
  current training loss:                0.478837
  current training accuracy:            0.836234
Seen 525000 samples, time used: 14055.435s
  current training loss:                0.478938
  current training accuracy:            0.836210
Seen 528000 samples, time used: 14135.912s
  current training loss:                0.478972
  current training accuracy:            0.836214
Seen 531000 samples, time used: 14216.128s
  current training loss:                0.478999
  current training accuracy:            0.836228
Seen 534000 samples, time used: 14294.565s
  current training loss:                0.479072
  current training accuracy:            0.836163
Seen 537000 samples, time used: 14373.735s
  current training loss:                0.479112
  current training accuracy:            0.836089
Seen 540000 samples, time used: 14453.611s
  current training loss:                0.479105
  current training accuracy:            0.836072
saving to ..., time used 793.128s
Seen 543000 samples, time used: 14534.122s
  current training loss:                0.479169
  current training accuracy:            0.836050
Seen 546000 samples, time used: 14613.526s
  current training loss:                0.479118
  current training accuracy:            0.836093
Seen 549000 samples, time used: 14692.422s
  current training loss:                0.479164
  current training accuracy:            0.836071
Epoch 12 of 20 took 14725.010s
  training loss:                0.479178
  training accuracy:            83.61 %
  validation loss:              0.458991
  validation accuracy:          82.53 %
  test loss:                    0.476542
  test accuracy:                81.37 %
saving to ./snli/word_by_word_model_epoch12.npz
Seen 3000 samples, time used: 82.172s
  current training loss:                0.471248
  current training accuracy:            0.839667
Seen 6000 samples, time used: 163.514s
  current training loss:                0.468376
  current training accuracy:            0.840167
Seen 9000 samples, time used: 243.486s
  current training loss:                0.463228
  current training accuracy:            0.842111
Seen 12000 samples, time used: 323.659s
  current training loss:                0.466102
  current training accuracy:            0.841167
Seen 15000 samples, time used: 403.711s
  current training loss:                0.467539
  current training accuracy:            0.842267
Seen 18000 samples, time used: 483.783s
  current training loss:                0.463681
  current training accuracy:            0.843889
Seen 21000 samples, time used: 564.196s
  current training loss:                0.466471
  current training accuracy:            0.842905
Seen 24000 samples, time used: 642.947s
  current training loss:                0.465928
  current training accuracy:            0.841833
Seen 27000 samples, time used: 722.562s
  current training loss:                0.465453
  current training accuracy:            0.841667
Seen 30000 samples, time used: 801.858s
  current training loss:                0.464882
  current training accuracy:            0.841867
saving to ..., time used 801.858s
Seen 33000 samples, time used: 880.481s
  current training loss:                0.466245
  current training accuracy:            0.841485
Seen 36000 samples, time used: 960.923s
  current training loss:                0.468174
  current training accuracy:            0.840389
Seen 39000 samples, time used: 1043.288s
  current training loss:                0.468335
  current training accuracy:            0.840795
Seen 42000 samples, time used: 1124.529s
  current training loss:                0.468174
  current training accuracy:            0.841548
Seen 45000 samples, time used: 1207.029s
  current training loss:                0.467410
  current training accuracy:            0.841200
Seen 48000 samples, time used: 1288.771s
  current training loss:                0.466753
  current training accuracy:            0.841667
Seen 51000 samples, time used: 1370.092s
  current training loss:                0.468200
  current training accuracy:            0.841059
Seen 54000 samples, time used: 1453.797s
  current training loss:                0.467735
  current training accuracy:            0.841463
Seen 57000 samples, time used: 1534.798s
  current training loss:                0.468550
  current training accuracy:            0.841333
Seen 60000 samples, time used: 1617.271s
  current training loss:                0.467935
  current training accuracy:            0.841683
saving to ..., time used 814.901s
Seen 63000 samples, time used: 1696.230s
  current training loss:                0.468322
  current training accuracy:            0.841286
Seen 66000 samples, time used: 1776.788s
  current training loss:                0.468827
  current training accuracy:            0.841182
Seen 69000 samples, time used: 1858.101s
  current training loss:                0.469247
  current training accuracy:            0.840928
Seen 72000 samples, time used: 1938.355s
  current training loss:                0.468702
  current training accuracy:            0.841000
Seen 75000 samples, time used: 2018.837s
  current training loss:                0.469398
  current training accuracy:            0.840627
Seen 78000 samples, time used: 2098.297s
  current training loss:                0.469944
  current training accuracy:            0.840500
Seen 81000 samples, time used: 2177.295s
  current training loss:                0.469531
  current training accuracy:            0.840704
Seen 84000 samples, time used: 2256.970s
  current training loss:                0.470037
  current training accuracy:            0.840286
Seen 87000 samples, time used: 2336.652s
  current training loss:                0.469835
  current training accuracy:            0.840621
Seen 90000 samples, time used: 2416.350s
  current training loss:                0.469399
  current training accuracy:            0.840711
saving to ..., time used 798.534s
Seen 93000 samples, time used: 2497.571s
  current training loss:                0.470701
  current training accuracy:            0.840022
Seen 96000 samples, time used: 2577.771s
  current training loss:                0.470443
  current training accuracy:            0.840177
Seen 99000 samples, time used: 2656.692s
  current training loss:                0.470418
  current training accuracy:            0.840313
Seen 102000 samples, time used: 2737.366s
  current training loss:                0.469879
  current training accuracy:            0.840529
Seen 105000 samples, time used: 2816.217s
  current training loss:                0.470285
  current training accuracy:            0.840562
Seen 108000 samples, time used: 2893.914s
  current training loss:                0.470533
  current training accuracy:            0.840333
Seen 111000 samples, time used: 2974.244s
  current training loss:                0.470728
  current training accuracy:            0.840351
Seen 114000 samples, time used: 3053.925s
  current training loss:                0.470561
  current training accuracy:            0.840447
Seen 117000 samples, time used: 3134.635s
  current training loss:                0.470965
  current training accuracy:            0.840171
Seen 120000 samples, time used: 3214.854s
  current training loss:                0.470669
  current training accuracy:            0.840275
saving to ..., time used 797.844s
Seen 123000 samples, time used: 3294.860s
  current training loss:                0.470854
  current training accuracy:            0.840114
Seen 126000 samples, time used: 3373.301s
  current training loss:                0.470819
  current training accuracy:            0.840016
Seen 129000 samples, time used: 3454.686s
  current training loss:                0.471027
  current training accuracy:            0.839837
Seen 132000 samples, time used: 3535.643s
  current training loss:                0.471178
  current training accuracy:            0.839773
Seen 135000 samples, time used: 3616.767s
  current training loss:                0.471126
  current training accuracy:            0.839719
Seen 138000 samples, time used: 3697.297s
  current training loss:                0.471207
  current training accuracy:            0.839674
Seen 141000 samples, time used: 3776.094s
  current training loss:                0.471482
  current training accuracy:            0.839567
Seen 144000 samples, time used: 3855.995s
  current training loss:                0.471087
  current training accuracy:            0.839715
Seen 147000 samples, time used: 3936.461s
  current training loss:                0.471147
  current training accuracy:            0.839639
Seen 150000 samples, time used: 4016.610s
  current training loss:                0.471660
  current training accuracy:            0.839367
saving to ..., time used 801.221s
Seen 153000 samples, time used: 4096.240s
  current training loss:                0.471496
  current training accuracy:            0.839523
Seen 156000 samples, time used: 4176.422s
  current training loss:                0.471484
  current training accuracy:            0.839506
Seen 159000 samples, time used: 4257.200s
  current training loss:                0.471376
  current training accuracy:            0.839616
Seen 162000 samples, time used: 4337.313s
  current training loss:                0.471358
  current training accuracy:            0.839642
Seen 165000 samples, time used: 4419.081s
  current training loss:                0.471473
  current training accuracy:            0.839733
Seen 168000 samples, time used: 4499.405s
  current training loss:                0.471427
  current training accuracy:            0.839690
Seen 171000 samples, time used: 4579.828s
  current training loss:                0.471766
  current training accuracy:            0.839573
Seen 174000 samples, time used: 4662.575s
  current training loss:                0.471825
  current training accuracy:            0.839592
Seen 177000 samples, time used: 4741.419s
  current training loss:                0.471996
  current training accuracy:            0.839469
Seen 180000 samples, time used: 4821.886s
  current training loss:                0.472340
  current training accuracy:            0.839306
saving to ..., time used 804.774s
Seen 183000 samples, time used: 4905.392s
  current training loss:                0.472069
  current training accuracy:            0.839421
Seen 186000 samples, time used: 4988.589s
  current training loss:                0.471930
  current training accuracy:            0.839548
Seen 189000 samples, time used: 5071.393s
  current training loss:                0.471987
  current training accuracy:            0.839466
Seen 192000 samples, time used: 5152.349s
  current training loss:                0.472039
  current training accuracy:            0.839453
Seen 195000 samples, time used: 5237.617s
  current training loss:                0.471960
  current training accuracy:            0.839528
Seen 198000 samples, time used: 5319.794s
  current training loss:                0.471727
  current training accuracy:            0.839641
Seen 201000 samples, time used: 5403.474s
  current training loss:                0.471681
  current training accuracy:            0.839801
Seen 204000 samples, time used: 5485.596s
  current training loss:                0.471764
  current training accuracy:            0.839735
Seen 207000 samples, time used: 5566.403s
  current training loss:                0.471805
  current training accuracy:            0.839739
Seen 210000 samples, time used: 5647.288s
  current training loss:                0.471956
  current training accuracy:            0.839652
saving to ..., time used 824.921s
Seen 213000 samples, time used: 5729.424s
  current training loss:                0.472062
  current training accuracy:            0.839512
Seen 216000 samples, time used: 5811.097s
  current training loss:                0.472321
  current training accuracy:            0.839491
Seen 219000 samples, time used: 5893.170s
  current training loss:                0.472387
  current training accuracy:            0.839356
Seen 222000 samples, time used: 5975.377s
  current training loss:                0.472422
  current training accuracy:            0.839302
Seen 225000 samples, time used: 6058.709s
  current training loss:                0.472616
  current training accuracy:            0.839289
Seen 228000 samples, time used: 6141.456s
  current training loss:                0.472801
  current training accuracy:            0.839298
Seen 231000 samples, time used: 6224.248s
  current training loss:                0.472874
  current training accuracy:            0.839234
Seen 234000 samples, time used: 6305.779s
  current training loss:                0.472861
  current training accuracy:            0.839222
Seen 237000 samples, time used: 6386.625s
  current training loss:                0.472793
  current training accuracy:            0.839241
Seen 240000 samples, time used: 6468.831s
  current training loss:                0.472501
  current training accuracy:            0.839279
saving to ..., time used 820.999s
Seen 243000 samples, time used: 6549.458s
  current training loss:                0.472606
  current training accuracy:            0.839218
Seen 246000 samples, time used: 6628.229s
  current training loss:                0.472699
  current training accuracy:            0.839215
Seen 249000 samples, time used: 6707.632s
  current training loss:                0.472637
  current training accuracy:            0.839213
Seen 252000 samples, time used: 6786.238s
  current training loss:                0.472712
  current training accuracy:            0.839325
Seen 255000 samples, time used: 6864.969s
  current training loss:                0.472785
  current training accuracy:            0.839373
Seen 258000 samples, time used: 6945.346s
  current training loss:                0.472841
  current training accuracy:            0.839368
Seen 261000 samples, time used: 7024.115s
  current training loss:                0.472968
  current training accuracy:            0.839299
Seen 264000 samples, time used: 7104.335s
  current training loss:                0.473038
  current training accuracy:            0.839231
Seen 267000 samples, time used: 7184.310s
  current training loss:                0.472741
  current training accuracy:            0.839337
Seen 270000 samples, time used: 7263.505s
  current training loss:                0.472687
  current training accuracy:            0.839400
saving to ..., time used 793.366s
Seen 273000 samples, time used: 7344.219s
  current training loss:                0.472710
  current training accuracy:            0.839443
Seen 276000 samples, time used: 7424.716s
  current training loss:                0.472853
  current training accuracy:            0.839380
Seen 279000 samples, time used: 7505.282s
  current training loss:                0.472789
  current training accuracy:            0.839326
Seen 282000 samples, time used: 7587.917s
  current training loss:                0.472747
  current training accuracy:            0.839301
Seen 285000 samples, time used: 7667.769s
  current training loss:                0.472671
  current training accuracy:            0.839298
Seen 288000 samples, time used: 7747.601s
  current training loss:                0.472819
  current training accuracy:            0.839236
Seen 291000 samples, time used: 7826.056s
  current training loss:                0.472919
  current training accuracy:            0.839192
Seen 294000 samples, time used: 7906.347s
  current training loss:                0.473003
  current training accuracy:            0.839238
Seen 297000 samples, time used: 7985.523s
  current training loss:                0.473052
  current training accuracy:            0.839300
Seen 300000 samples, time used: 8065.951s
  current training loss:                0.473106
  current training accuracy:            0.839283
saving to ..., time used 801.889s
Seen 303000 samples, time used: 8147.092s
  current training loss:                0.473240
  current training accuracy:            0.839254
Seen 306000 samples, time used: 8226.606s
  current training loss:                0.473142
  current training accuracy:            0.839363
Seen 309000 samples, time used: 8307.025s
  current training loss:                0.473023
  current training accuracy:            0.839359
Seen 312000 samples, time used: 8386.585s
  current training loss:                0.472900
  current training accuracy:            0.839324
Seen 315000 samples, time used: 8466.495s
  current training loss:                0.472772
  current training accuracy:            0.839365
Seen 318000 samples, time used: 8545.405s
  current training loss:                0.472840
  current training accuracy:            0.839340
Seen 321000 samples, time used: 8626.102s
  current training loss:                0.472774
  current training accuracy:            0.839414
Seen 324000 samples, time used: 8706.297s
  current training loss:                0.473073
  current training accuracy:            0.839318
Seen 327000 samples, time used: 8786.927s
  current training loss:                0.472871
  current training accuracy:            0.839425
Seen 330000 samples, time used: 8866.696s
  current training loss:                0.472815
  current training accuracy:            0.839406
saving to ..., time used 800.257s
Seen 333000 samples, time used: 8946.168s
  current training loss:                0.472844
  current training accuracy:            0.839450
Seen 336000 samples, time used: 9024.633s
  current training loss:                0.473231
  current training accuracy:            0.839327
Seen 339000 samples, time used: 9103.974s
  current training loss:                0.473319
  current training accuracy:            0.839342
Seen 342000 samples, time used: 9182.546s
  current training loss:                0.473321
  current training accuracy:            0.839289
Seen 345000 samples, time used: 9260.904s
  current training loss:                0.473471
  current training accuracy:            0.839197
Seen 348000 samples, time used: 9341.901s
  current training loss:                0.473801
  current training accuracy:            0.839098
Seen 351000 samples, time used: 9422.131s
  current training loss:                0.473894
  current training accuracy:            0.839103
Seen 354000 samples, time used: 9502.432s
  current training loss:                0.474018
  current training accuracy:            0.839014
Seen 357000 samples, time used: 9583.685s
  current training loss:                0.474053
  current training accuracy:            0.838958
Seen 360000 samples, time used: 9665.644s
  current training loss:                0.474148
  current training accuracy:            0.838922
saving to ..., time used 798.466s
Seen 363000 samples, time used: 9745.833s
  current training loss:                0.474276
  current training accuracy:            0.838826
Seen 366000 samples, time used: 9824.829s
  current training loss:                0.474415
  current training accuracy:            0.838740
Seen 369000 samples, time used: 9905.835s
  current training loss:                0.474407
  current training accuracy:            0.838775
Seen 372000 samples, time used: 9986.929s
  current training loss:                0.474410
  current training accuracy:            0.838847
Seen 375000 samples, time used: 10066.979s
  current training loss:                0.474453
  current training accuracy:            0.838808
Seen 378000 samples, time used: 10147.412s
  current training loss:                0.474578
  current training accuracy:            0.838720
Seen 381000 samples, time used: 10227.880s
  current training loss:                0.474603
  current training accuracy:            0.838724
Seen 384000 samples, time used: 10308.116s
  current training loss:                0.474527
  current training accuracy:            0.838740
Seen 387000 samples, time used: 10388.021s
  current training loss:                0.474558
  current training accuracy:            0.838705
Seen 390000 samples, time used: 10468.075s
  current training loss:                0.474573
  current training accuracy:            0.838705
saving to ..., time used 801.918s
Seen 393000 samples, time used: 10548.051s
  current training loss:                0.474498
  current training accuracy:            0.838662
Seen 396000 samples, time used: 10628.825s
  current training loss:                0.474462
  current training accuracy:            0.838682
Seen 399000 samples, time used: 10707.991s
  current training loss:                0.474556
  current training accuracy:            0.838662
Seen 402000 samples, time used: 10787.170s
  current training loss:                0.474767
  current training accuracy:            0.838545
Seen 405000 samples, time used: 10866.413s
  current training loss:                0.474690
  current training accuracy:            0.838563
Seen 408000 samples, time used: 10945.427s
  current training loss:                0.474745
  current training accuracy:            0.838500
Seen 411000 samples, time used: 11024.948s
  current training loss:                0.474957
  current training accuracy:            0.838401
Seen 414000 samples, time used: 11106.027s
  current training loss:                0.475124
  current training accuracy:            0.838454
Seen 417000 samples, time used: 11184.919s
  current training loss:                0.475136
  current training accuracy:            0.838446
Seen 420000 samples, time used: 11264.630s
  current training loss:                0.475070
  current training accuracy:            0.838486
saving to ..., time used 796.039s
Seen 423000 samples, time used: 11346.180s
  current training loss:                0.474988
  current training accuracy:            0.838513
Seen 426000 samples, time used: 11424.667s
  current training loss:                0.475058
  current training accuracy:            0.838474
Seen 429000 samples, time used: 11504.739s
  current training loss:                0.475172
  current training accuracy:            0.838410
Seen 432000 samples, time used: 11584.537s
  current training loss:                0.475219
  current training accuracy:            0.838394
Seen 435000 samples, time used: 11665.330s
  current training loss:                0.475327
  current training accuracy:            0.838377
Seen 438000 samples, time used: 11746.851s
  current training loss:                0.475371
  current training accuracy:            0.838340
Seen 441000 samples, time used: 11827.719s
  current training loss:                0.475316
  current training accuracy:            0.838317
Seen 444000 samples, time used: 11909.492s
  current training loss:                0.475393
  current training accuracy:            0.838234
Seen 447000 samples, time used: 11991.756s
  current training loss:                0.475466
  current training accuracy:            0.838242
Seen 450000 samples, time used: 12073.593s
  current training loss:                0.475533
  current training accuracy:            0.838184
saving to ..., time used 808.480s
Seen 453000 samples, time used: 12154.358s
  current training loss:                0.475695
  current training accuracy:            0.838152
Seen 456000 samples, time used: 12235.650s
  current training loss:                0.475824
  current training accuracy:            0.838171
Seen 459000 samples, time used: 12315.547s
  current training loss:                0.475886
  current training accuracy:            0.838159
Seen 462000 samples, time used: 12394.710s
  current training loss:                0.475810
  current training accuracy:            0.838212
Seen 465000 samples, time used: 12474.411s
  current training loss:                0.475812
  current training accuracy:            0.838224
Seen 468000 samples, time used: 12554.095s
  current training loss:                0.475861
  current training accuracy:            0.838203
Seen 471000 samples, time used: 12632.597s
  current training loss:                0.475850
  current training accuracy:            0.838202
Seen 474000 samples, time used: 12713.601s
  current training loss:                0.475836
  current training accuracy:            0.838167
Seen 477000 samples, time used: 12792.464s
  current training loss:                0.475896
  current training accuracy:            0.838105
Seen 480000 samples, time used: 12870.389s
  current training loss:                0.475829
  current training accuracy:            0.838110
saving to ..., time used 796.287s
Seen 483000 samples, time used: 12950.826s
  current training loss:                0.475819
  current training accuracy:            0.838145
Seen 486000 samples, time used: 13030.809s
  current training loss:                0.475794
  current training accuracy:            0.838097
Seen 489000 samples, time used: 13111.080s
  current training loss:                0.475722
  current training accuracy:            0.838127
Seen 492000 samples, time used: 13188.134s
  current training loss:                0.475662
  current training accuracy:            0.838114
Seen 495000 samples, time used: 13268.435s
  current training loss:                0.475629
  current training accuracy:            0.838158
Seen 498000 samples, time used: 13348.537s
  current training loss:                0.475618
  current training accuracy:            0.838102
Seen 501000 samples, time used: 13428.489s
  current training loss:                0.475730
  current training accuracy:            0.838066
Seen 504000 samples, time used: 13508.589s
  current training loss:                0.475804
  current training accuracy:            0.838040
Seen 507000 samples, time used: 13587.005s
  current training loss:                0.475783
  current training accuracy:            0.838063
Seen 510000 samples, time used: 13665.473s
  current training loss:                0.475819
  current training accuracy:            0.838016
saving to ..., time used 794.610s
Seen 513000 samples, time used: 13744.263s
  current training loss:                0.475827
  current training accuracy:            0.837979
Seen 516000 samples, time used: 13823.419s
  current training loss:                0.475833
  current training accuracy:            0.837973
Seen 519000 samples, time used: 13901.628s
  current training loss:                0.475792
  current training accuracy:            0.837971
Seen 522000 samples, time used: 13980.611s
  current training loss:                0.475796
  current training accuracy:            0.837971
Seen 525000 samples, time used: 14059.638s
  current training loss:                0.475730
  current training accuracy:            0.838000
Seen 528000 samples, time used: 14138.791s
  current training loss:                0.475697
  current training accuracy:            0.838038
Seen 531000 samples, time used: 14217.061s
  current training loss:                0.475753
  current training accuracy:            0.838021
Seen 534000 samples, time used: 14295.426s
  current training loss:                0.475833
  current training accuracy:            0.837994
Seen 537000 samples, time used: 14375.490s
  current training loss:                0.475793
  current training accuracy:            0.838030
Seen 540000 samples, time used: 14456.581s
  current training loss:                0.475853
  current training accuracy:            0.838028
saving to ..., time used 790.642s
Seen 543000 samples, time used: 14537.150s
  current training loss:                0.475885
  current training accuracy:            0.837993
Seen 546000 samples, time used: 14617.467s
  current training loss:                0.475918
  current training accuracy:            0.837962
Seen 549000 samples, time used: 14700.069s
  current training loss:                0.475944
  current training accuracy:            0.837925
Epoch 13 of 20 took 14732.844s
  training loss:                0.475946
  training accuracy:            83.79 %
  validation loss:              0.458672
  validation accuracy:          82.17 %
  test loss:                    0.472106
  test accuracy:                81.59 %
saving to ./snli/word_by_word_model_epoch13.npz
Seen 3000 samples, time used: 80.743s
  current training loss:                0.472046
  current training accuracy:            0.845000
Seen 6000 samples, time used: 162.397s
  current training loss:                0.482366
  current training accuracy:            0.839667
Seen 9000 samples, time used: 243.153s
  current training loss:                0.478083
  current training accuracy:            0.839667
Seen 12000 samples, time used: 325.601s
  current training loss:                0.479074
  current training accuracy:            0.838000
Seen 15000 samples, time used: 406.269s
  current training loss:                0.474077
  current training accuracy:            0.840800
Seen 18000 samples, time used: 487.257s
  current training loss:                0.471870
  current training accuracy:            0.841389
Seen 21000 samples, time used: 566.305s
  current training loss:                0.469791
  current training accuracy:            0.841810
Seen 24000 samples, time used: 646.254s
  current training loss:                0.468456
  current training accuracy:            0.842500
Seen 27000 samples, time used: 726.738s
  current training loss:                0.469673
  current training accuracy:            0.841593
Seen 30000 samples, time used: 806.051s
  current training loss:                0.467417
  current training accuracy:            0.842167
saving to ..., time used 806.051s
Seen 33000 samples, time used: 888.958s
  current training loss:                0.468531
  current training accuracy:            0.842848
Seen 36000 samples, time used: 969.440s
  current training loss:                0.468143
  current training accuracy:            0.842889
Seen 39000 samples, time used: 1048.795s
  current training loss:                0.467911
  current training accuracy:            0.843026
Seen 42000 samples, time used: 1128.261s
  current training loss:                0.468190
  current training accuracy:            0.842786
Seen 45000 samples, time used: 1207.918s
  current training loss:                0.468379
  current training accuracy:            0.842533
Seen 48000 samples, time used: 1287.573s
  current training loss:                0.467276
  current training accuracy:            0.842958
Seen 51000 samples, time used: 1368.061s
  current training loss:                0.466944
  current training accuracy:            0.843078
Seen 54000 samples, time used: 1448.771s
  current training loss:                0.465833
  current training accuracy:            0.843444
Seen 57000 samples, time used: 1530.893s
  current training loss:                0.465356
  current training accuracy:            0.844246
Seen 60000 samples, time used: 1611.220s
  current training loss:                0.464914
  current training accuracy:            0.844667
saving to ..., time used 804.679s
Seen 63000 samples, time used: 1693.795s
  current training loss:                0.465778
  current training accuracy:            0.844333
Seen 66000 samples, time used: 1774.525s
  current training loss:                0.466179
  current training accuracy:            0.844258
Seen 69000 samples, time used: 1856.368s
  current training loss:                0.466678
  current training accuracy:            0.843855
Seen 72000 samples, time used: 1938.735s
  current training loss:                0.466770
  current training accuracy:            0.843917
Seen 75000 samples, time used: 2019.467s
  current training loss:                0.466888
  current training accuracy:            0.843987
Seen 78000 samples, time used: 2101.516s
  current training loss:                0.466169
  current training accuracy:            0.844513
Seen 81000 samples, time used: 2182.215s
  current training loss:                0.466299
  current training accuracy:            0.844148
Seen 84000 samples, time used: 2263.265s
  current training loss:                0.465867
  current training accuracy:            0.844107
Seen 87000 samples, time used: 2345.692s
  current training loss:                0.466073
  current training accuracy:            0.844046
Seen 90000 samples, time used: 2426.740s
  current training loss:                0.466247
  current training accuracy:            0.844033
saving to ..., time used 814.927s
Seen 93000 samples, time used: 2509.469s
  current training loss:                0.466313
  current training accuracy:            0.843634
Seen 96000 samples, time used: 2590.831s
  current training loss:                0.466431
  current training accuracy:            0.843573
Seen 99000 samples, time used: 2671.322s
  current training loss:                0.466250
  current training accuracy:            0.843707
Seen 102000 samples, time used: 2751.497s
  current training loss:                0.465947
  current training accuracy:            0.843922
Seen 105000 samples, time used: 2833.443s
  current training loss:                0.466013
  current training accuracy:            0.843648
Seen 108000 samples, time used: 2915.436s
  current training loss:                0.466024
  current training accuracy:            0.843500
Seen 111000 samples, time used: 2996.767s
  current training loss:                0.465942
  current training accuracy:            0.843586
Seen 114000 samples, time used: 3078.680s
  current training loss:                0.466281
  current training accuracy:            0.843404
Seen 117000 samples, time used: 3159.415s
  current training loss:                0.465858
  current training accuracy:            0.843530
Seen 120000 samples, time used: 3240.325s
  current training loss:                0.465739
  current training accuracy:            0.843342
saving to ..., time used 813.016s
Seen 123000 samples, time used: 3323.501s
  current training loss:                0.466126
  current training accuracy:            0.843276
Seen 126000 samples, time used: 3404.608s
  current training loss:                0.466074
  current training accuracy:            0.843397
Seen 129000 samples, time used: 3487.145s
  current training loss:                0.465914
  current training accuracy:            0.843388
Seen 132000 samples, time used: 3569.931s
  current training loss:                0.466163
  current training accuracy:            0.843409
Seen 135000 samples, time used: 3654.320s
  current training loss:                0.466623
  current training accuracy:            0.843156
Seen 138000 samples, time used: 3736.598s
  current training loss:                0.467188
  current training accuracy:            0.842725
Seen 141000 samples, time used: 3818.082s
  current training loss:                0.466806
  current training accuracy:            0.842894
Seen 144000 samples, time used: 3901.552s
  current training loss:                0.466397
  current training accuracy:            0.843222
Seen 147000 samples, time used: 3984.727s
  current training loss:                0.466235
  current training accuracy:            0.843347
Seen 150000 samples, time used: 4065.772s
  current training loss:                0.466273
  current training accuracy:            0.843240
saving to ..., time used 824.969s
Seen 153000 samples, time used: 4147.127s
  current training loss:                0.466777
  current training accuracy:            0.842889
Seen 156000 samples, time used: 4227.249s
  current training loss:                0.466513
  current training accuracy:            0.843058
Seen 159000 samples, time used: 4307.586s
  current training loss:                0.466779
  current training accuracy:            0.842874
Seen 162000 samples, time used: 4388.507s
  current training loss:                0.466946
  current training accuracy:            0.842827
Seen 165000 samples, time used: 4469.679s
  current training loss:                0.466937
  current training accuracy:            0.842836
Seen 168000 samples, time used: 4550.170s
  current training loss:                0.466821
  current training accuracy:            0.843000
Seen 171000 samples, time used: 4629.811s
  current training loss:                0.467617
  current training accuracy:            0.842667
Seen 174000 samples, time used: 4710.836s
  current training loss:                0.467934
  current training accuracy:            0.842437
Seen 177000 samples, time used: 4794.146s
  current training loss:                0.467999
  current training accuracy:            0.842373
Seen 180000 samples, time used: 4874.660s
  current training loss:                0.467879
  current training accuracy:            0.842344
saving to ..., time used 808.322s
Seen 183000 samples, time used: 4955.424s
  current training loss:                0.467743
  current training accuracy:            0.842317
Seen 186000 samples, time used: 5035.625s
  current training loss:                0.467311
  current training accuracy:            0.842430
Seen 189000 samples, time used: 5116.478s
  current training loss:                0.467360
  current training accuracy:            0.842312
Seen 192000 samples, time used: 5197.142s
  current training loss:                0.467731
  current training accuracy:            0.842193
Seen 195000 samples, time used: 5275.463s
  current training loss:                0.467394
  current training accuracy:            0.842221
Seen 198000 samples, time used: 5356.468s
  current training loss:                0.467497
  current training accuracy:            0.842162
Seen 201000 samples, time used: 5435.920s
  current training loss:                0.467382
  current training accuracy:            0.842164
Seen 204000 samples, time used: 5515.920s
  current training loss:                0.467314
  current training accuracy:            0.842216
Seen 207000 samples, time used: 5596.780s
  current training loss:                0.467688
  current training accuracy:            0.842087
Seen 210000 samples, time used: 5677.353s
  current training loss:                0.468152
  current training accuracy:            0.841933
saving to ..., time used 802.154s
Seen 213000 samples, time used: 5760.494s
  current training loss:                0.468136
  current training accuracy:            0.841789
Seen 216000 samples, time used: 5839.782s
  current training loss:                0.467756
  current training accuracy:            0.841838
Seen 219000 samples, time used: 5920.214s
  current training loss:                0.467881
  current training accuracy:            0.841817
Seen 222000 samples, time used: 5999.744s
  current training loss:                0.467882
  current training accuracy:            0.841829
Seen 225000 samples, time used: 6079.643s
  current training loss:                0.468172
  current training accuracy:            0.841671
Seen 228000 samples, time used: 6159.536s
  current training loss:                0.468054
  current training accuracy:            0.841689
Seen 231000 samples, time used: 6242.320s
  current training loss:                0.468053
  current training accuracy:            0.841662
Seen 234000 samples, time used: 6323.542s
  current training loss:                0.468308
  current training accuracy:            0.841585
Seen 237000 samples, time used: 6403.319s
  current training loss:                0.468326
  current training accuracy:            0.841511
Seen 240000 samples, time used: 6485.337s
  current training loss:                0.468344
  current training accuracy:            0.841400
saving to ..., time used 807.516s
Seen 243000 samples, time used: 6565.093s
  current training loss:                0.468477
  current training accuracy:            0.841337
Seen 246000 samples, time used: 6646.064s
  current training loss:                0.468431
  current training accuracy:            0.841354
Seen 249000 samples, time used: 6727.866s
  current training loss:                0.468595
  current training accuracy:            0.841329
Seen 252000 samples, time used: 6807.566s
  current training loss:                0.468557
  current training accuracy:            0.841385
Seen 255000 samples, time used: 6889.782s
  current training loss:                0.468697
  current training accuracy:            0.841329
Seen 258000 samples, time used: 6972.404s
  current training loss:                0.468813
  current training accuracy:            0.841318
Seen 261000 samples, time used: 7053.445s
  current training loss:                0.468638
  current training accuracy:            0.841414
Seen 264000 samples, time used: 7132.921s
  current training loss:                0.468709
  current training accuracy:            0.841383
Seen 267000 samples, time used: 7213.285s
  current training loss:                0.468629
  current training accuracy:            0.841438
Seen 270000 samples, time used: 7294.138s
  current training loss:                0.468583
  current training accuracy:            0.841326
saving to ..., time used 808.317s
Seen 273000 samples, time used: 7376.633s
  current training loss:                0.468591
  current training accuracy:            0.841311
Seen 276000 samples, time used: 7458.575s
  current training loss:                0.468805
  current training accuracy:            0.841250
Seen 279000 samples, time used: 7541.225s
  current training loss:                0.468798
  current training accuracy:            0.841265
Seen 282000 samples, time used: 7623.737s
  current training loss:                0.468847
  current training accuracy:            0.841291
Seen 285000 samples, time used: 7705.533s
  current training loss:                0.468818
  current training accuracy:            0.841344
Seen 288000 samples, time used: 7789.490s
  current training loss:                0.469053
  current training accuracy:            0.841267
Seen 291000 samples, time used: 7871.824s
  current training loss:                0.469172
  current training accuracy:            0.841265
Seen 294000 samples, time used: 7955.074s
  current training loss:                0.469219
  current training accuracy:            0.841279
Seen 297000 samples, time used: 8038.165s
  current training loss:                0.469162
  current training accuracy:            0.841286
Seen 300000 samples, time used: 8120.402s
  current training loss:                0.469043
  current training accuracy:            0.841263
saving to ..., time used 825.731s
Seen 303000 samples, time used: 8201.376s
  current training loss:                0.469135
  current training accuracy:            0.841290
Seen 306000 samples, time used: 8283.715s
  current training loss:                0.468959
  current training accuracy:            0.841386
Seen 309000 samples, time used: 8364.788s
  current training loss:                0.469145
  current training accuracy:            0.841324
Seen 312000 samples, time used: 8446.287s
  current training loss:                0.469098
  current training accuracy:            0.841353
Seen 315000 samples, time used: 8526.469s
  current training loss:                0.469205
  current training accuracy:            0.841241
Seen 318000 samples, time used: 8607.916s
  current training loss:                0.469391
  current training accuracy:            0.841142
Seen 321000 samples, time used: 8687.506s
  current training loss:                0.469533
  current training accuracy:            0.841140
Seen 324000 samples, time used: 8768.741s
  current training loss:                0.469333
  current training accuracy:            0.841182
Seen 327000 samples, time used: 8849.492s
  current training loss:                0.469342
  current training accuracy:            0.841162
Seen 330000 samples, time used: 8929.418s
  current training loss:                0.469270
  current training accuracy:            0.841158
saving to ..., time used 808.475s
Seen 333000 samples, time used: 9010.435s
  current training loss:                0.469288
  current training accuracy:            0.841099
Seen 336000 samples, time used: 9090.182s
  current training loss:                0.469322
  current training accuracy:            0.841024
Seen 339000 samples, time used: 9170.451s
  current training loss:                0.469607
  current training accuracy:            0.840855
Seen 342000 samples, time used: 9250.525s
  current training loss:                0.469533
  current training accuracy:            0.840833
Seen 345000 samples, time used: 9330.130s
  current training loss:                0.469701
  current training accuracy:            0.840812
Seen 348000 samples, time used: 9411.209s
  current training loss:                0.469730
  current training accuracy:            0.840761
Seen 351000 samples, time used: 9492.375s
  current training loss:                0.469901
  current training accuracy:            0.840698
Seen 354000 samples, time used: 9571.523s
  current training loss:                0.469849
  current training accuracy:            0.840715
Seen 357000 samples, time used: 9651.963s
  current training loss:                0.469787
  current training accuracy:            0.840793
Seen 360000 samples, time used: 9733.124s
  current training loss:                0.469903
  current training accuracy:            0.840775
saving to ..., time used 803.141s
Seen 363000 samples, time used: 9813.443s
  current training loss:                0.470067
  current training accuracy:            0.840741
Seen 366000 samples, time used: 9894.449s
  current training loss:                0.470383
  current training accuracy:            0.840623
Seen 369000 samples, time used: 9976.103s
  current training loss:                0.470217
  current training accuracy:            0.840672
Seen 372000 samples, time used: 10056.044s
  current training loss:                0.470473
  current training accuracy:            0.840578
Seen 375000 samples, time used: 10135.864s
  current training loss:                0.470432
  current training accuracy:            0.840589
Seen 378000 samples, time used: 10216.728s
  current training loss:                0.470352
  current training accuracy:            0.840669
Seen 381000 samples, time used: 10296.956s
  current training loss:                0.470410
  current training accuracy:            0.840711
Seen 384000 samples, time used: 10376.983s
  current training loss:                0.470379
  current training accuracy:            0.840737
Seen 387000 samples, time used: 10457.860s
  current training loss:                0.470315
  current training accuracy:            0.840674
Seen 390000 samples, time used: 10536.436s
  current training loss:                0.470364
  current training accuracy:            0.840592
saving to ..., time used 802.825s
Seen 393000 samples, time used: 10615.864s
  current training loss:                0.470508
  current training accuracy:            0.840542
Seen 396000 samples, time used: 10694.455s
  current training loss:                0.470314
  current training accuracy:            0.840591
Seen 399000 samples, time used: 10774.112s
  current training loss:                0.470288
  current training accuracy:            0.840601
Seen 402000 samples, time used: 10853.287s
  current training loss:                0.470429
  current training accuracy:            0.840532
Seen 405000 samples, time used: 10932.526s
  current training loss:                0.470347
  current training accuracy:            0.840568
Seen 408000 samples, time used: 11012.142s
  current training loss:                0.470434
  current training accuracy:            0.840510
Seen 411000 samples, time used: 11091.717s
  current training loss:                0.470562
  current training accuracy:            0.840453
Seen 414000 samples, time used: 11172.516s
  current training loss:                0.470636
  current training accuracy:            0.840432
Seen 417000 samples, time used: 11253.125s
  current training loss:                0.470973
  current training accuracy:            0.840369
Seen 420000 samples, time used: 11331.965s
  current training loss:                0.470815
  current training accuracy:            0.840440
saving to ..., time used 795.026s
Seen 423000 samples, time used: 11414.063s
  current training loss:                0.470711
  current training accuracy:            0.840513
Seen 426000 samples, time used: 11494.667s
  current training loss:                0.470791
  current training accuracy:            0.840500
Seen 429000 samples, time used: 11574.058s
  current training loss:                0.470866
  current training accuracy:            0.840429
Seen 432000 samples, time used: 11653.449s
  current training loss:                0.470819
  current training accuracy:            0.840391
Seen 435000 samples, time used: 11734.138s
  current training loss:                0.470930
  current training accuracy:            0.840292
Seen 438000 samples, time used: 11813.890s
  current training loss:                0.470929
  current training accuracy:            0.840285
Seen 441000 samples, time used: 11894.044s
  current training loss:                0.470851
  current training accuracy:            0.840336
Seen 444000 samples, time used: 11973.726s
  current training loss:                0.470824
  current training accuracy:            0.840365
Seen 447000 samples, time used: 12054.615s
  current training loss:                0.470930
  current training accuracy:            0.840338
Seen 450000 samples, time used: 12135.392s
  current training loss:                0.470955
  current training accuracy:            0.840329
saving to ..., time used 802.949s
Seen 453000 samples, time used: 12216.594s
  current training loss:                0.470984
  current training accuracy:            0.840320
Seen 456000 samples, time used: 12299.240s
  current training loss:                0.470978
  current training accuracy:            0.840283
Seen 459000 samples, time used: 12380.476s
  current training loss:                0.471095
  current training accuracy:            0.840244
Seen 462000 samples, time used: 12461.038s
  current training loss:                0.471051
  current training accuracy:            0.840260
Seen 465000 samples, time used: 12541.510s
  current training loss:                0.471182
  current training accuracy:            0.840202
Seen 468000 samples, time used: 12621.999s
  current training loss:                0.471152
  current training accuracy:            0.840239
Seen 471000 samples, time used: 12703.077s
  current training loss:                0.471278
  current training accuracy:            0.840166
Seen 474000 samples, time used: 12784.617s
  current training loss:                0.471352
  current training accuracy:            0.840135
Seen 477000 samples, time used: 12865.129s
  current training loss:                0.471532
  current training accuracy:            0.840027
Seen 480000 samples, time used: 12946.776s
  current training loss:                0.471540
  current training accuracy:            0.840023
saving to ..., time used 810.877s
Seen 483000 samples, time used: 13028.140s
  current training loss:                0.471554
  current training accuracy:            0.840012
Seen 486000 samples, time used: 13107.415s
  current training loss:                0.471612
  current training accuracy:            0.839967
Seen 489000 samples, time used: 13186.657s
  current training loss:                0.471652
  current training accuracy:            0.839957
Seen 492000 samples, time used: 13265.455s
  current training loss:                0.471659
  current training accuracy:            0.839947
Seen 495000 samples, time used: 13345.271s
  current training loss:                0.471677
  current training accuracy:            0.839958
Seen 498000 samples, time used: 13423.457s
  current training loss:                0.471653
  current training accuracy:            0.839968
Seen 501000 samples, time used: 13502.419s
  current training loss:                0.471715
  current training accuracy:            0.839934
Seen 504000 samples, time used: 13582.922s
  current training loss:                0.471876
  current training accuracy:            0.839857
Seen 507000 samples, time used: 13661.597s
  current training loss:                0.472071
  current training accuracy:            0.839761
Seen 510000 samples, time used: 13741.677s
  current training loss:                0.472093
  current training accuracy:            0.839737
saving to ..., time used 794.344s
Seen 513000 samples, time used: 13822.935s
  current training loss:                0.472213
  current training accuracy:            0.839674
Seen 516000 samples, time used: 13902.524s
  current training loss:                0.472316
  current training accuracy:            0.839624
Seen 519000 samples, time used: 13982.942s
  current training loss:                0.472377
  current training accuracy:            0.839605
Seen 522000 samples, time used: 14062.394s
  current training loss:                0.472305
  current training accuracy:            0.839703
Seen 525000 samples, time used: 14141.771s
  current training loss:                0.472176
  current training accuracy:            0.839735
Seen 528000 samples, time used: 14222.841s
  current training loss:                0.472266
  current training accuracy:            0.839680
Seen 531000 samples, time used: 14301.761s
  current training loss:                0.472279
  current training accuracy:            0.839682
Seen 534000 samples, time used: 14380.746s
  current training loss:                0.472311
  current training accuracy:            0.839618
Seen 537000 samples, time used: 14460.339s
  current training loss:                0.472356
  current training accuracy:            0.839618
Seen 540000 samples, time used: 14540.878s
  current training loss:                0.472312
  current training accuracy:            0.839615
saving to ..., time used 798.711s
Seen 543000 samples, time used: 14622.006s
  current training loss:                0.472274
  current training accuracy:            0.839626
Seen 546000 samples, time used: 14699.947s
  current training loss:                0.472235
  current training accuracy:            0.839637
Seen 549000 samples, time used: 14780.190s
  current training loss:                0.472271
  current training accuracy:            0.839614
Epoch 14 of 20 took 14812.304s
  training loss:                0.472252
  training accuracy:            83.96 %
  validation loss:              0.464758
  validation accuracy:          81.83 %
  test loss:                    0.473954
  test accuracy:                81.29 %
saving to ./snli/word_by_word_model_epoch14.npz
Seen 3000 samples, time used: 79.287s
  current training loss:                0.470361
  current training accuracy:            0.846667
Seen 6000 samples, time used: 158.503s
  current training loss:                0.469320
  current training accuracy:            0.841333
Seen 9000 samples, time used: 235.828s
  current training loss:                0.467667
  current training accuracy:            0.842444
Seen 12000 samples, time used: 316.502s
  current training loss:                0.464719
  current training accuracy:            0.842750
Seen 15000 samples, time used: 398.165s
  current training loss:                0.469426
  current training accuracy:            0.841733
Seen 18000 samples, time used: 477.749s
  current training loss:                0.468723
  current training accuracy:            0.842389
Seen 21000 samples, time used: 555.898s
  current training loss:                0.466821
  current training accuracy:            0.843714
Seen 24000 samples, time used: 634.298s
  current training loss:                0.464493
  current training accuracy:            0.844458
Seen 27000 samples, time used: 716.196s
  current training loss:                0.465809
  current training accuracy:            0.844037
Seen 30000 samples, time used: 795.028s
  current training loss:                0.465150
  current training accuracy:            0.844300
saving to ..., time used 795.028s
Seen 33000 samples, time used: 873.950s
  current training loss:                0.464338
  current training accuracy:            0.844545
Seen 36000 samples, time used: 953.841s
  current training loss:                0.464736
  current training accuracy:            0.844417
Seen 39000 samples, time used: 1033.162s
  current training loss:                0.464291
  current training accuracy:            0.844590
Seen 42000 samples, time used: 1112.915s
  current training loss:                0.463831
  current training accuracy:            0.845190
Seen 45000 samples, time used: 1193.212s
  current training loss:                0.464721
  current training accuracy:            0.844800
Seen 48000 samples, time used: 1272.494s
  current training loss:                0.463530
  current training accuracy:            0.845437
Seen 51000 samples, time used: 1351.871s
  current training loss:                0.464637
  current training accuracy:            0.845431
Seen 54000 samples, time used: 1432.061s
  current training loss:                0.464188
  current training accuracy:            0.845370
Seen 57000 samples, time used: 1511.195s
  current training loss:                0.463280
  current training accuracy:            0.845667
Seen 60000 samples, time used: 1592.480s
  current training loss:                0.464159
  current training accuracy:            0.845383
saving to ..., time used 796.916s
Seen 63000 samples, time used: 1671.720s
  current training loss:                0.464859
  current training accuracy:            0.844984
Seen 66000 samples, time used: 1751.229s
  current training loss:                0.465100
  current training accuracy:            0.844864
Seen 69000 samples, time used: 1830.269s
  current training loss:                0.465721
  current training accuracy:            0.844594
Seen 72000 samples, time used: 1908.676s
  current training loss:                0.466152
  current training accuracy:            0.844389
Seen 75000 samples, time used: 1988.783s
  current training loss:                0.465369
  current training accuracy:            0.844547
Seen 78000 samples, time used: 2069.476s
  current training loss:                0.464988
  current training accuracy:            0.844641
Seen 81000 samples, time used: 2148.872s
  current training loss:                0.464651
  current training accuracy:            0.844728
Seen 84000 samples, time used: 2229.188s
  current training loss:                0.464933
  current training accuracy:            0.844702
Seen 87000 samples, time used: 2309.007s
  current training loss:                0.464561
  current training accuracy:            0.844690
Seen 90000 samples, time used: 2388.491s
  current training loss:                0.464553
  current training accuracy:            0.844611
saving to ..., time used 795.523s
Seen 93000 samples, time used: 2469.691s
  current training loss:                0.464451
  current training accuracy:            0.844688
Seen 96000 samples, time used: 2552.538s
  current training loss:                0.464897
  current training accuracy:            0.844552
Seen 99000 samples, time used: 2635.255s
  current training loss:                0.464789
  current training accuracy:            0.844566
Seen 102000 samples, time used: 2714.607s
  current training loss:                0.465200
  current training accuracy:            0.844480
Seen 105000 samples, time used: 2796.008s
  current training loss:                0.464784
  current training accuracy:            0.844438
Seen 108000 samples, time used: 2876.813s
  current training loss:                0.465567
  current training accuracy:            0.843935
Seen 111000 samples, time used: 2959.843s
  current training loss:                0.465338
  current training accuracy:            0.843865
Seen 114000 samples, time used: 3043.230s
  current training loss:                0.465370
  current training accuracy:            0.843798
Seen 117000 samples, time used: 3126.677s
  current training loss:                0.464914
  current training accuracy:            0.843889
Seen 120000 samples, time used: 3206.783s
  current training loss:                0.464593
  current training accuracy:            0.844208
saving to ..., time used 817.733s
Seen 123000 samples, time used: 3288.147s
  current training loss:                0.464146
  current training accuracy:            0.844309
Seen 126000 samples, time used: 3371.056s
  current training loss:                0.464994
  current training accuracy:            0.843913
Seen 129000 samples, time used: 3452.334s
  current training loss:                0.465340
  current training accuracy:            0.843667
Seen 132000 samples, time used: 3534.360s
  current training loss:                0.465386
  current training accuracy:            0.843720
Seen 135000 samples, time used: 3616.258s
  current training loss:                0.465290
  current training accuracy:            0.843778
Seen 138000 samples, time used: 3698.791s
  current training loss:                0.465186
  current training accuracy:            0.843826
Seen 141000 samples, time used: 3778.427s
  current training loss:                0.465159
  current training accuracy:            0.843922
Seen 144000 samples, time used: 3859.334s
  current training loss:                0.465021
  current training accuracy:            0.843931
Seen 147000 samples, time used: 3943.681s
  current training loss:                0.464720
  current training accuracy:            0.843966
Seen 150000 samples, time used: 4026.138s
  current training loss:                0.464685
  current training accuracy:            0.844060
saving to ..., time used 818.870s
Seen 153000 samples, time used: 4108.035s
  current training loss:                0.464341
  current training accuracy:            0.844216
Seen 156000 samples, time used: 4188.809s
  current training loss:                0.464468
  current training accuracy:            0.844109
Seen 159000 samples, time used: 4270.148s
  current training loss:                0.464903
  current training accuracy:            0.843937
Seen 162000 samples, time used: 4351.190s
  current training loss:                0.464450
  current training accuracy:            0.843907
Seen 165000 samples, time used: 4433.792s
  current training loss:                0.464052
  current training accuracy:            0.844115
Seen 168000 samples, time used: 4514.510s
  current training loss:                0.464128
  current training accuracy:            0.844107
Seen 171000 samples, time used: 4592.436s
  current training loss:                0.464158
  current training accuracy:            0.843988
Seen 174000 samples, time used: 4671.134s
  current training loss:                0.464804
  current training accuracy:            0.843759
Seen 177000 samples, time used: 4751.172s
  current training loss:                0.464829
  current training accuracy:            0.843893
Seen 180000 samples, time used: 4830.591s
  current training loss:                0.465181
  current training accuracy:            0.843706
saving to ..., time used 803.969s
Seen 183000 samples, time used: 4911.249s
  current training loss:                0.465133
  current training accuracy:            0.843738
Seen 186000 samples, time used: 4992.550s
  current training loss:                0.465218
  current training accuracy:            0.843704
Seen 189000 samples, time used: 5070.737s
  current training loss:                0.465125
  current training accuracy:            0.843794
Seen 192000 samples, time used: 5150.726s
  current training loss:                0.465281
  current training accuracy:            0.843719
Seen 195000 samples, time used: 5229.962s
  current training loss:                0.465037
  current training accuracy:            0.843749
Seen 198000 samples, time used: 5310.444s
  current training loss:                0.464996
  current training accuracy:            0.843702
Seen 201000 samples, time used: 5389.597s
  current training loss:                0.464965
  current training accuracy:            0.843721
Seen 204000 samples, time used: 5468.775s
  current training loss:                0.464798
  current training accuracy:            0.843819
Seen 207000 samples, time used: 5548.290s
  current training loss:                0.464850
  current training accuracy:            0.843889
Seen 210000 samples, time used: 5626.699s
  current training loss:                0.465022
  current training accuracy:            0.843843
saving to ..., time used 795.594s
Seen 213000 samples, time used: 5706.762s
  current training loss:                0.464788
  current training accuracy:            0.843958
Seen 216000 samples, time used: 5786.481s
  current training loss:                0.464481
  current training accuracy:            0.843968
Seen 219000 samples, time used: 5865.726s
  current training loss:                0.464531
  current training accuracy:            0.843941
Seen 222000 samples, time used: 5944.216s
  current training loss:                0.464452
  current training accuracy:            0.843968
Seen 225000 samples, time used: 6023.815s
  current training loss:                0.464467
  current training accuracy:            0.843933
Seen 228000 samples, time used: 6105.704s
  current training loss:                0.464573
  current training accuracy:            0.843904
Seen 231000 samples, time used: 6184.691s
  current training loss:                0.464489
  current training accuracy:            0.843887
Seen 234000 samples, time used: 6263.474s
  current training loss:                0.464499
  current training accuracy:            0.843748
Seen 237000 samples, time used: 6343.168s
  current training loss:                0.464410
  current training accuracy:            0.843781
Seen 240000 samples, time used: 6425.170s
  current training loss:                0.464450
  current training accuracy:            0.843717
saving to ..., time used 797.961s
Seen 243000 samples, time used: 6504.310s
  current training loss:                0.464674
  current training accuracy:            0.843646
Seen 246000 samples, time used: 6582.925s
  current training loss:                0.464790
  current training accuracy:            0.843549
Seen 249000 samples, time used: 6663.695s
  current training loss:                0.464994
  current training accuracy:            0.843446
Seen 252000 samples, time used: 6742.897s
  current training loss:                0.464799
  current training accuracy:            0.843520
Seen 255000 samples, time used: 6821.823s
  current training loss:                0.464708
  current training accuracy:            0.843569
Seen 258000 samples, time used: 6900.834s
  current training loss:                0.464769
  current training accuracy:            0.843438
Seen 261000 samples, time used: 6980.098s
  current training loss:                0.465032
  current training accuracy:            0.843322
Seen 264000 samples, time used: 7058.066s
  current training loss:                0.465433
  current training accuracy:            0.843136
Seen 267000 samples, time used: 7138.809s
  current training loss:                0.465427
  current training accuracy:            0.843213
Seen 270000 samples, time used: 7218.088s
  current training loss:                0.465294
  current training accuracy:            0.843230
saving to ..., time used 792.433s
Seen 273000 samples, time used: 7298.296s
  current training loss:                0.465506
  current training accuracy:            0.843070
Seen 276000 samples, time used: 7377.755s
  current training loss:                0.465585
  current training accuracy:            0.842899
Seen 279000 samples, time used: 7458.625s
  current training loss:                0.465689
  current training accuracy:            0.842860
Seen 282000 samples, time used: 7537.971s
  current training loss:                0.465647
  current training accuracy:            0.842876
Seen 285000 samples, time used: 7617.162s
  current training loss:                0.465751
  current training accuracy:            0.842832
Seen 288000 samples, time used: 7695.739s
  current training loss:                0.465913
  current training accuracy:            0.842781
Seen 291000 samples, time used: 7775.861s
  current training loss:                0.465814
  current training accuracy:            0.842732
Seen 294000 samples, time used: 7855.727s
  current training loss:                0.465996
  current training accuracy:            0.842653
Seen 297000 samples, time used: 7935.896s
  current training loss:                0.466006
  current training accuracy:            0.842751
Seen 300000 samples, time used: 8016.579s
  current training loss:                0.466151
  current training accuracy:            0.842647
saving to ..., time used 797.993s
Seen 303000 samples, time used: 8096.321s
  current training loss:                0.466123
  current training accuracy:            0.842706
Seen 306000 samples, time used: 8176.249s
  current training loss:                0.466066
  current training accuracy:            0.842627
Seen 309000 samples, time used: 8257.890s
  current training loss:                0.466279
  current training accuracy:            0.842573
Seen 312000 samples, time used: 8336.723s
  current training loss:                0.466486
  current training accuracy:            0.842538
Seen 315000 samples, time used: 8417.845s
  current training loss:                0.466461
  current training accuracy:            0.842587
Seen 318000 samples, time used: 8496.851s
  current training loss:                0.466450
  current training accuracy:            0.842601
Seen 321000 samples, time used: 8576.064s
  current training loss:                0.466516
  current training accuracy:            0.842567
Seen 324000 samples, time used: 8655.425s
  current training loss:                0.466382
  current training accuracy:            0.842580
Seen 327000 samples, time used: 8736.423s
  current training loss:                0.466490
  current training accuracy:            0.842557
Seen 330000 samples, time used: 8816.091s
  current training loss:                0.466513
  current training accuracy:            0.842567
saving to ..., time used 799.008s
Seen 333000 samples, time used: 8896.948s
  current training loss:                0.466438
  current training accuracy:            0.842589
Seen 336000 samples, time used: 8981.287s
  current training loss:                0.466527
  current training accuracy:            0.842524
Seen 339000 samples, time used: 9062.172s
  current training loss:                0.466535
  current training accuracy:            0.842484
Seen 342000 samples, time used: 9143.841s
  current training loss:                0.466427
  current training accuracy:            0.842556
Seen 345000 samples, time used: 9226.099s
  current training loss:                0.466535
  current training accuracy:            0.842496
Seen 348000 samples, time used: 9307.218s
  current training loss:                0.466572
  current training accuracy:            0.842448
Seen 351000 samples, time used: 9389.900s
  current training loss:                0.466576
  current training accuracy:            0.842402
Seen 354000 samples, time used: 9471.701s
  current training loss:                0.466804
  current training accuracy:            0.842285
Seen 357000 samples, time used: 9553.833s
  current training loss:                0.466703
  current training accuracy:            0.842336
Seen 360000 samples, time used: 9636.768s
  current training loss:                0.466632
  current training accuracy:            0.842403
saving to ..., time used 820.209s
Seen 363000 samples, time used: 9719.215s
  current training loss:                0.466828
  current training accuracy:            0.842369
Seen 366000 samples, time used: 9800.533s
  current training loss:                0.466846
  current training accuracy:            0.842342
Seen 369000 samples, time used: 9882.989s
  current training loss:                0.466856
  current training accuracy:            0.842341
Seen 372000 samples, time used: 9964.699s
  current training loss:                0.466909
  current training accuracy:            0.842333
Seen 375000 samples, time used: 10043.630s
  current training loss:                0.466844
  current training accuracy:            0.842368
Seen 378000 samples, time used: 10123.194s
  current training loss:                0.466955
  current training accuracy:            0.842333
Seen 381000 samples, time used: 10203.796s
  current training loss:                0.467193
  current training accuracy:            0.842231
Seen 384000 samples, time used: 10283.534s
  current training loss:                0.467334
  current training accuracy:            0.842089
Seen 387000 samples, time used: 10363.521s
  current training loss:                0.467346
  current training accuracy:            0.842096
Seen 390000 samples, time used: 10442.519s
  current training loss:                0.467234
  current training accuracy:            0.842118
saving to ..., time used 805.204s
Seen 393000 samples, time used: 10526.035s
  current training loss:                0.467317
  current training accuracy:            0.842079
Seen 396000 samples, time used: 10611.283s
  current training loss:                0.467393
  current training accuracy:            0.842063
Seen 399000 samples, time used: 10690.845s
  current training loss:                0.467391
  current training accuracy:            0.842055
Seen 402000 samples, time used: 10773.260s
  current training loss:                0.467483
  current training accuracy:            0.842005
Seen 405000 samples, time used: 10856.516s
  current training loss:                0.467505
  current training accuracy:            0.841951
Seen 408000 samples, time used: 10939.409s
  current training loss:                0.467398
  current training accuracy:            0.841998
Seen 411000 samples, time used: 11019.622s
  current training loss:                0.467475
  current training accuracy:            0.841925
Seen 414000 samples, time used: 11100.405s
  current training loss:                0.467474
  current training accuracy:            0.841937
Seen 417000 samples, time used: 11183.673s
  current training loss:                0.467457
  current training accuracy:            0.841933
Seen 420000 samples, time used: 11268.242s
  current training loss:                0.467601
  current training accuracy:            0.841852
saving to ..., time used 825.220s
Seen 423000 samples, time used: 11348.361s
  current training loss:                0.467717
  current training accuracy:            0.841816
Seen 426000 samples, time used: 11428.517s
  current training loss:                0.467754
  current training accuracy:            0.841798
Seen 429000 samples, time used: 11508.206s
  current training loss:                0.467710
  current training accuracy:            0.841816
Seen 432000 samples, time used: 11588.168s
  current training loss:                0.467696
  current training accuracy:            0.841808
Seen 435000 samples, time used: 11671.563s
  current training loss:                0.467797
  current training accuracy:            0.841848
Seen 438000 samples, time used: 11750.716s
  current training loss:                0.467767
  current training accuracy:            0.841884
Seen 441000 samples, time used: 11832.503s
  current training loss:                0.467806
  current training accuracy:            0.841889
Seen 444000 samples, time used: 11913.079s
  current training loss:                0.467915
  current training accuracy:            0.841827
Seen 447000 samples, time used: 11994.493s
  current training loss:                0.467991
  current training accuracy:            0.841779
Seen 450000 samples, time used: 12074.400s
  current training loss:                0.467954
  current training accuracy:            0.841773
saving to ..., time used 805.590s
Seen 453000 samples, time used: 12156.635s
  current training loss:                0.468111
  current training accuracy:            0.841695
Seen 456000 samples, time used: 12237.667s
  current training loss:                0.468014
  current training accuracy:            0.841752
Seen 459000 samples, time used: 12316.579s
  current training loss:                0.468171
  current training accuracy:            0.841678
Seen 462000 samples, time used: 12394.632s
  current training loss:                0.468181
  current training accuracy:            0.841695
Seen 465000 samples, time used: 12474.095s
  current training loss:                0.468033
  current training accuracy:            0.841761
Seen 468000 samples, time used: 12553.772s
  current training loss:                0.468085
  current training accuracy:            0.841756
Seen 471000 samples, time used: 12634.633s
  current training loss:                0.468136
  current training accuracy:            0.841690
Seen 474000 samples, time used: 12716.496s
  current training loss:                0.468241
  current training accuracy:            0.841665
Seen 477000 samples, time used: 12796.168s
  current training loss:                0.468188
  current training accuracy:            0.841656
Seen 480000 samples, time used: 12876.820s
  current training loss:                0.468156
  current training accuracy:            0.841715
saving to ..., time used 801.860s
Seen 483000 samples, time used: 12959.739s
  current training loss:                0.468121
  current training accuracy:            0.841708
Seen 486000 samples, time used: 13040.870s
  current training loss:                0.468059
  current training accuracy:            0.841739
Seen 489000 samples, time used: 13122.409s
  current training loss:                0.468208
  current training accuracy:            0.841681
Seen 492000 samples, time used: 13204.478s
  current training loss:                0.468183
  current training accuracy:            0.841650
Seen 495000 samples, time used: 13286.577s
  current training loss:                0.468401
  current training accuracy:            0.841558
Seen 498000 samples, time used: 13368.696s
  current training loss:                0.468422
  current training accuracy:            0.841574
Seen 501000 samples, time used: 13448.714s
  current training loss:                0.468317
  current training accuracy:            0.841621
Seen 504000 samples, time used: 13529.897s
  current training loss:                0.468407
  current training accuracy:            0.841609
Seen 507000 samples, time used: 13613.050s
  current training loss:                0.468566
  current training accuracy:            0.841546
Seen 510000 samples, time used: 13694.839s
  current training loss:                0.468611
  current training accuracy:            0.841537
saving to ..., time used 817.481s
Seen 513000 samples, time used: 13774.409s
  current training loss:                0.468440
  current training accuracy:            0.841616
Seen 516000 samples, time used: 13855.068s
  current training loss:                0.468607
  current training accuracy:            0.841595
Seen 519000 samples, time used: 13933.825s
  current training loss:                0.468620
  current training accuracy:            0.841595
Seen 522000 samples, time used: 14014.187s
  current training loss:                0.468512
  current training accuracy:            0.841634
Seen 525000 samples, time used: 14094.723s
  current training loss:                0.468607
  current training accuracy:            0.841602
Seen 528000 samples, time used: 14172.898s
  current training loss:                0.468590
  current training accuracy:            0.841616
Seen 531000 samples, time used: 14251.542s
  current training loss:                0.468647
  current training accuracy:            0.841606
Seen 534000 samples, time used: 14331.201s
  current training loss:                0.468692
  current training accuracy:            0.841571
Seen 537000 samples, time used: 14411.650s
  current training loss:                0.468886
  current training accuracy:            0.841520
Seen 540000 samples, time used: 14491.021s
  current training loss:                0.468852
  current training accuracy:            0.841556
saving to ..., time used 795.632s
Seen 543000 samples, time used: 14571.417s
  current training loss:                0.468866
  current training accuracy:            0.841479
Seen 546000 samples, time used: 14651.364s
  current training loss:                0.468770
  current training accuracy:            0.841526
Seen 549000 samples, time used: 14729.285s
  current training loss:                0.468809
  current training accuracy:            0.841536
Epoch 15 of 20 took 14761.528s
  training loss:                0.468831
  training accuracy:            84.15 %
  validation loss:              0.457357
  validation accuracy:          82.66 %
  test loss:                    0.472685
  test accuracy:                81.75 %
saving to ./snli/word_by_word_model_epoch15.npz
Seen 3000 samples, time used: 79.273s
  current training loss:                0.447146
  current training accuracy:            0.848333
Seen 6000 samples, time used: 157.744s
  current training loss:                0.447412
  current training accuracy:            0.846000
Seen 9000 samples, time used: 237.890s
  current training loss:                0.451835
  current training accuracy:            0.845889
Seen 12000 samples, time used: 317.803s
  current training loss:                0.453481
  current training accuracy:            0.845583
Seen 15000 samples, time used: 400.052s
  current training loss:                0.452200
  current training accuracy:            0.847533
Seen 18000 samples, time used: 479.073s
  current training loss:                0.453590
  current training accuracy:            0.847167
Seen 21000 samples, time used: 560.732s
  current training loss:                0.457073
  current training accuracy:            0.845429
Seen 24000 samples, time used: 639.333s
  current training loss:                0.455843
  current training accuracy:            0.846000
Seen 27000 samples, time used: 719.766s
  current training loss:                0.455858
  current training accuracy:            0.846481
Seen 30000 samples, time used: 800.500s
  current training loss:                0.458644
  current training accuracy:            0.845067
saving to ..., time used 800.500s
Seen 33000 samples, time used: 880.687s
  current training loss:                0.457936
  current training accuracy:            0.845394
Seen 36000 samples, time used: 962.043s
  current training loss:                0.458493
  current training accuracy:            0.845361
Seen 39000 samples, time used: 1041.734s
  current training loss:                0.459520
  current training accuracy:            0.845410
Seen 42000 samples, time used: 1121.332s
  current training loss:                0.459294
  current training accuracy:            0.845429
Seen 45000 samples, time used: 1199.369s
  current training loss:                0.458832
  current training accuracy:            0.845733
Seen 48000 samples, time used: 1279.346s
  current training loss:                0.459354
  current training accuracy:            0.845521
Seen 51000 samples, time used: 1358.992s
  current training loss:                0.457937
  current training accuracy:            0.846314
Seen 54000 samples, time used: 1437.731s
  current training loss:                0.457564
  current training accuracy:            0.846574
Seen 57000 samples, time used: 1518.995s
  current training loss:                0.456810
  current training accuracy:            0.846491
Seen 60000 samples, time used: 1600.959s
  current training loss:                0.456999
  current training accuracy:            0.845917
saving to ..., time used 799.936s
Seen 63000 samples, time used: 1680.744s
  current training loss:                0.456977
  current training accuracy:            0.846238
Seen 66000 samples, time used: 1758.275s
  current training loss:                0.456905
  current training accuracy:            0.846197
Seen 69000 samples, time used: 1837.434s
  current training loss:                0.455404
  current training accuracy:            0.846681
Seen 72000 samples, time used: 1916.269s
  current training loss:                0.455531
  current training accuracy:            0.846667
Seen 75000 samples, time used: 1995.183s
  current training loss:                0.456747
  current training accuracy:            0.846467
Seen 78000 samples, time used: 2076.491s
  current training loss:                0.456270
  current training accuracy:            0.846705
Seen 81000 samples, time used: 2155.515s
  current training loss:                0.455548
  current training accuracy:            0.846840
Seen 84000 samples, time used: 2235.192s
  current training loss:                0.455004
  current training accuracy:            0.846869
Seen 87000 samples, time used: 2315.081s
  current training loss:                0.455634
  current training accuracy:            0.846632
Seen 90000 samples, time used: 2393.533s
  current training loss:                0.455786
  current training accuracy:            0.846844
saving to ..., time used 792.024s
Seen 93000 samples, time used: 2473.269s
  current training loss:                0.455842
  current training accuracy:            0.846753
Seen 96000 samples, time used: 2552.423s
  current training loss:                0.456209
  current training accuracy:            0.846802
Seen 99000 samples, time used: 2632.357s
  current training loss:                0.456105
  current training accuracy:            0.847010
Seen 102000 samples, time used: 2712.770s
  current training loss:                0.456318
  current training accuracy:            0.847098
Seen 105000 samples, time used: 2790.501s
  current training loss:                0.456380
  current training accuracy:            0.847029
Seen 108000 samples, time used: 2869.100s
  current training loss:                0.456836
  current training accuracy:            0.846759
Seen 111000 samples, time used: 2947.660s
  current training loss:                0.457211
  current training accuracy:            0.846757
Seen 114000 samples, time used: 3026.405s
  current training loss:                0.456543
  current training accuracy:            0.847193
Seen 117000 samples, time used: 3107.265s
  current training loss:                0.456795
  current training accuracy:            0.847179
Seen 120000 samples, time used: 3185.268s
  current training loss:                0.457233
  current training accuracy:            0.847158
saving to ..., time used 791.195s
Seen 123000 samples, time used: 3266.362s
  current training loss:                0.457401
  current training accuracy:            0.847057
Seen 126000 samples, time used: 3347.476s
  current training loss:                0.458139
  current training accuracy:            0.846833
Seen 129000 samples, time used: 3427.747s
  current training loss:                0.457896
  current training accuracy:            0.846899
Seen 132000 samples, time used: 3506.641s
  current training loss:                0.457874
  current training accuracy:            0.846886
Seen 135000 samples, time used: 3586.535s
  current training loss:                0.458082
  current training accuracy:            0.846822
Seen 138000 samples, time used: 3667.793s
  current training loss:                0.459067
  current training accuracy:            0.846558
Seen 141000 samples, time used: 3747.925s
  current training loss:                0.458899
  current training accuracy:            0.846624
Seen 144000 samples, time used: 3829.022s
  current training loss:                0.458814
  current training accuracy:            0.846667
Seen 147000 samples, time used: 3908.014s
  current training loss:                0.458539
  current training accuracy:            0.846680
Seen 150000 samples, time used: 3987.994s
  current training loss:                0.458598
  current training accuracy:            0.846407
saving to ..., time used 802.197s
Seen 153000 samples, time used: 4066.956s
  current training loss:                0.458952
  current training accuracy:            0.846242
Seen 156000 samples, time used: 4149.118s
  current training loss:                0.459262
  current training accuracy:            0.846006
Seen 159000 samples, time used: 4230.551s
  current training loss:                0.459162
  current training accuracy:            0.846006
Seen 162000 samples, time used: 4310.512s
  current training loss:                0.458992
  current training accuracy:            0.846123
Seen 165000 samples, time used: 4390.847s
  current training loss:                0.459288
  current training accuracy:            0.846036
Seen 168000 samples, time used: 4471.805s
  current training loss:                0.459123
  current training accuracy:            0.846167
Seen 171000 samples, time used: 4552.077s
  current training loss:                0.459133
  current training accuracy:            0.846035
Seen 174000 samples, time used: 4632.969s
  current training loss:                0.459344
  current training accuracy:            0.846029
Seen 177000 samples, time used: 4713.427s
  current training loss:                0.459393
  current training accuracy:            0.846090
Seen 180000 samples, time used: 4793.861s
  current training loss:                0.459262
  current training accuracy:            0.846067
saving to ..., time used 805.301s
Seen 183000 samples, time used: 4876.559s
  current training loss:                0.459578
  current training accuracy:            0.845891
Seen 186000 samples, time used: 4955.977s
  current training loss:                0.460027
  current training accuracy:            0.845640
Seen 189000 samples, time used: 5037.148s
  current training loss:                0.459915
  current training accuracy:            0.845661
Seen 192000 samples, time used: 5116.545s
  current training loss:                0.459731
  current training accuracy:            0.845724
Seen 195000 samples, time used: 5197.375s
  current training loss:                0.459754
  current training accuracy:            0.845779
Seen 198000 samples, time used: 5276.176s
  current training loss:                0.459468
  current training accuracy:            0.845899
Seen 201000 samples, time used: 5355.535s
  current training loss:                0.459786
  current training accuracy:            0.845766
Seen 204000 samples, time used: 5435.629s
  current training loss:                0.460000
  current training accuracy:            0.845637
Seen 207000 samples, time used: 5514.839s
  current training loss:                0.460263
  current training accuracy:            0.845507
Seen 210000 samples, time used: 5595.336s
  current training loss:                0.460675
  current training accuracy:            0.845400
saving to ..., time used 800.902s
Seen 213000 samples, time used: 5676.860s
  current training loss:                0.460759
  current training accuracy:            0.845343
Seen 216000 samples, time used: 5761.124s
  current training loss:                0.461178
  current training accuracy:            0.845218
Seen 219000 samples, time used: 5843.016s
  current training loss:                0.461367
  current training accuracy:            0.845100
Seen 222000 samples, time used: 5927.078s
  current training loss:                0.461297
  current training accuracy:            0.845122
Seen 225000 samples, time used: 6010.817s
  current training loss:                0.461147
  current training accuracy:            0.845200
Seen 228000 samples, time used: 6091.690s
  current training loss:                0.461608
  current training accuracy:            0.845096
Seen 231000 samples, time used: 6172.416s
  current training loss:                0.461868
  current training accuracy:            0.845009
Seen 234000 samples, time used: 6253.828s
  current training loss:                0.461846
  current training accuracy:            0.845051
Seen 237000 samples, time used: 6333.561s
  current training loss:                0.462043
  current training accuracy:            0.845017
Seen 240000 samples, time used: 6413.089s
  current training loss:                0.462026
  current training accuracy:            0.845100
saving to ..., time used 817.217s
Seen 243000 samples, time used: 6492.401s
  current training loss:                0.462320
  current training accuracy:            0.845004
Seen 246000 samples, time used: 6571.886s
  current training loss:                0.462420
  current training accuracy:            0.844963
Seen 249000 samples, time used: 6651.612s
  current training loss:                0.462551
  current training accuracy:            0.844880
Seen 252000 samples, time used: 6731.250s
  current training loss:                0.462621
  current training accuracy:            0.844869
Seen 255000 samples, time used: 6812.630s
  current training loss:                0.462693
  current training accuracy:            0.844824
Seen 258000 samples, time used: 6892.374s
  current training loss:                0.462851
  current training accuracy:            0.844771
Seen 261000 samples, time used: 6972.397s
  current training loss:                0.462943
  current training accuracy:            0.844716
Seen 264000 samples, time used: 7052.069s
  current training loss:                0.463437
  current training accuracy:            0.844561
Seen 267000 samples, time used: 7132.020s
  current training loss:                0.463324
  current training accuracy:            0.844573
Seen 270000 samples, time used: 7210.955s
  current training loss:                0.463301
  current training accuracy:            0.844556
saving to ..., time used 797.315s
Seen 273000 samples, time used: 7289.374s
  current training loss:                0.463300
  current training accuracy:            0.844498
Seen 276000 samples, time used: 7368.626s
  current training loss:                0.463366
  current training accuracy:            0.844464
Seen 279000 samples, time used: 7447.821s
  current training loss:                0.463413
  current training accuracy:            0.844412
Seen 282000 samples, time used: 7525.784s
  current training loss:                0.463334
  current training accuracy:            0.844404
Seen 285000 samples, time used: 7605.045s
  current training loss:                0.463476
  current training accuracy:            0.844400
Seen 288000 samples, time used: 7685.333s
  current training loss:                0.463328
  current training accuracy:            0.844490
Seen 291000 samples, time used: 7764.234s
  current training loss:                0.463252
  current training accuracy:            0.844540
Seen 294000 samples, time used: 7843.621s
  current training loss:                0.463470
  current training accuracy:            0.844473
Seen 297000 samples, time used: 7925.031s
  current training loss:                0.463587
  current training accuracy:            0.844451
Seen 300000 samples, time used: 8004.617s
  current training loss:                0.463907
  current training accuracy:            0.844287
saving to ..., time used 793.111s
Seen 303000 samples, time used: 8086.954s
  current training loss:                0.464033
  current training accuracy:            0.844228
Seen 306000 samples, time used: 8166.190s
  current training loss:                0.464024
  current training accuracy:            0.844190
Seen 309000 samples, time used: 8248.374s
  current training loss:                0.463902
  current training accuracy:            0.844165
Seen 312000 samples, time used: 8327.922s
  current training loss:                0.463836
  current training accuracy:            0.844141
Seen 315000 samples, time used: 8408.253s
  current training loss:                0.463791
  current training accuracy:            0.844165
Seen 318000 samples, time used: 8486.294s
  current training loss:                0.463879
  current training accuracy:            0.844135
Seen 321000 samples, time used: 8567.929s
  current training loss:                0.463943
  current training accuracy:            0.844103
Seen 324000 samples, time used: 8646.094s
  current training loss:                0.463946
  current training accuracy:            0.844046
Seen 327000 samples, time used: 8726.049s
  current training loss:                0.463856
  current training accuracy:            0.844055
Seen 330000 samples, time used: 8805.430s
  current training loss:                0.463927
  current training accuracy:            0.843979
saving to ..., time used 800.242s
Seen 333000 samples, time used: 8885.791s
  current training loss:                0.463979
  current training accuracy:            0.844036
Seen 336000 samples, time used: 8965.073s
  current training loss:                0.463848
  current training accuracy:            0.844107
Seen 339000 samples, time used: 9045.678s
  current training loss:                0.464014
  current training accuracy:            0.844015
Seen 342000 samples, time used: 9124.570s
  current training loss:                0.464183
  current training accuracy:            0.843953
Seen 345000 samples, time used: 9204.391s
  current training loss:                0.464211
  current training accuracy:            0.843899
Seen 348000 samples, time used: 9284.194s
  current training loss:                0.464265
  current training accuracy:            0.843848
Seen 351000 samples, time used: 9364.303s
  current training loss:                0.464244
  current training accuracy:            0.843903
Seen 354000 samples, time used: 9443.663s
  current training loss:                0.464197
  current training accuracy:            0.843935
Seen 357000 samples, time used: 9525.152s
  current training loss:                0.464487
  current training accuracy:            0.843826
Seen 360000 samples, time used: 9606.798s
  current training loss:                0.464347
  current training accuracy:            0.843847
saving to ..., time used 800.842s
Seen 363000 samples, time used: 9686.028s
  current training loss:                0.464390
  current training accuracy:            0.843807
Seen 366000 samples, time used: 9767.758s
  current training loss:                0.464538
  current training accuracy:            0.843776
Seen 369000 samples, time used: 9847.145s
  current training loss:                0.464667
  current training accuracy:            0.843734
Seen 372000 samples, time used: 9927.583s
  current training loss:                0.464609
  current training accuracy:            0.843702
Seen 375000 samples, time used: 10006.504s
  current training loss:                0.464704
  current training accuracy:            0.843624
Seen 378000 samples, time used: 10084.245s
  current training loss:                0.464760
  current training accuracy:            0.843566
Seen 381000 samples, time used: 10163.844s
  current training loss:                0.464783
  current training accuracy:            0.843514
Seen 384000 samples, time used: 10242.808s
  current training loss:                0.464757
  current training accuracy:            0.843536
Seen 387000 samples, time used: 10323.154s
  current training loss:                0.464843
  current training accuracy:            0.843491
Seen 390000 samples, time used: 10401.622s
  current training loss:                0.464899
  current training accuracy:            0.843397
saving to ..., time used 794.221s
Seen 393000 samples, time used: 10484.276s
  current training loss:                0.464989
  current training accuracy:            0.843361
Seen 396000 samples, time used: 10568.385s
  current training loss:                0.465023
  current training accuracy:            0.843318
Seen 399000 samples, time used: 10649.994s
  current training loss:                0.465078
  current training accuracy:            0.843306
Seen 402000 samples, time used: 10733.324s
  current training loss:                0.465186
  current training accuracy:            0.843266
Seen 405000 samples, time used: 10815.469s
  current training loss:                0.465063
  current training accuracy:            0.843237
Seen 408000 samples, time used: 10898.346s
  current training loss:                0.465278
  current training accuracy:            0.843174
Seen 411000 samples, time used: 10979.671s
  current training loss:                0.465264
  current training accuracy:            0.843182
Seen 414000 samples, time used: 11057.662s
  current training loss:                0.465311
  current training accuracy:            0.843157
Seen 417000 samples, time used: 11135.990s
  current training loss:                0.465399
  current training accuracy:            0.843175
Seen 420000 samples, time used: 11215.125s
  current training loss:                0.465428
  current training accuracy:            0.843186
saving to ..., time used 812.876s
Seen 423000 samples, time used: 11295.350s
  current training loss:                0.465555
  current training accuracy:            0.843156
Seen 426000 samples, time used: 11375.795s
  current training loss:                0.465550
  current training accuracy:            0.843195
Seen 429000 samples, time used: 11455.736s
  current training loss:                0.465550
  current training accuracy:            0.843214
Seen 432000 samples, time used: 11534.399s
  current training loss:                0.465509
  current training accuracy:            0.843206
Seen 435000 samples, time used: 11614.020s
  current training loss:                0.465530
  current training accuracy:            0.843230
Seen 438000 samples, time used: 11693.039s
  current training loss:                0.465481
  current training accuracy:            0.843237
Seen 441000 samples, time used: 11772.970s
  current training loss:                0.465599
  current training accuracy:            0.843177
Seen 444000 samples, time used: 11851.763s
  current training loss:                0.465454
  current training accuracy:            0.843209
Seen 447000 samples, time used: 11933.415s
  current training loss:                0.465546
  current training accuracy:            0.843199
Seen 450000 samples, time used: 12012.045s
  current training loss:                0.465639
  current training accuracy:            0.843176
saving to ..., time used 796.377s
Seen 453000 samples, time used: 12090.451s
  current training loss:                0.465466
  current training accuracy:            0.843221
Seen 456000 samples, time used: 12170.978s
  current training loss:                0.465606
  current training accuracy:            0.843103
Seen 459000 samples, time used: 12253.707s
  current training loss:                0.465735
  current training accuracy:            0.843081
Seen 462000 samples, time used: 12337.668s
  current training loss:                0.465755
  current training accuracy:            0.843097
Seen 465000 samples, time used: 12418.318s
  current training loss:                0.465852
  current training accuracy:            0.843065
Seen 468000 samples, time used: 12502.535s
  current training loss:                0.465841
  current training accuracy:            0.843079
Seen 471000 samples, time used: 12584.446s
  current training loss:                0.465775
  current training accuracy:            0.843204
Seen 474000 samples, time used: 12666.746s
  current training loss:                0.465709
  current training accuracy:            0.843255
Seen 477000 samples, time used: 12750.099s
  current training loss:                0.465648
  current training accuracy:            0.843310
Seen 480000 samples, time used: 12830.807s
  current training loss:                0.465806
  current training accuracy:            0.843219
saving to ..., time used 818.206s
Seen 483000 samples, time used: 12912.262s
  current training loss:                0.465725
  current training accuracy:            0.843244
Seen 486000 samples, time used: 12992.029s
  current training loss:                0.465863
  current training accuracy:            0.843189
Seen 489000 samples, time used: 13072.901s
  current training loss:                0.465924
  current training accuracy:            0.843121
Seen 492000 samples, time used: 13152.299s
  current training loss:                0.465895
  current training accuracy:            0.843067
Seen 495000 samples, time used: 13231.885s
  current training loss:                0.465950
  current training accuracy:            0.843034
Seen 498000 samples, time used: 13313.433s
  current training loss:                0.466042
  current training accuracy:            0.843010
Seen 501000 samples, time used: 13393.352s
  current training loss:                0.466168
  current training accuracy:            0.842948
Seen 504000 samples, time used: 13474.333s
  current training loss:                0.466162
  current training accuracy:            0.842954
Seen 507000 samples, time used: 13553.170s
  current training loss:                0.466159
  current training accuracy:            0.842923
Seen 510000 samples, time used: 13632.325s
  current training loss:                0.466089
  current training accuracy:            0.842982
saving to ..., time used 800.850s
Seen 513000 samples, time used: 13713.237s
  current training loss:                0.466171
  current training accuracy:            0.842938
Seen 516000 samples, time used: 13792.348s
  current training loss:                0.466278
  current training accuracy:            0.842911
Seen 519000 samples, time used: 13871.135s
  current training loss:                0.466232
  current training accuracy:            0.842896
Seen 522000 samples, time used: 13951.375s
  current training loss:                0.466153
  current training accuracy:            0.842933
Seen 525000 samples, time used: 14030.675s
  current training loss:                0.466111
  current training accuracy:            0.842890
Seen 528000 samples, time used: 14110.709s
  current training loss:                0.466203
  current training accuracy:            0.842875
Seen 531000 samples, time used: 14190.731s
  current training loss:                0.466139
  current training accuracy:            0.842895
Seen 534000 samples, time used: 14270.863s
  current training loss:                0.466190
  current training accuracy:            0.842861
Seen 537000 samples, time used: 14351.093s
  current training loss:                0.466308
  current training accuracy:            0.842836
Seen 540000 samples, time used: 14428.692s
  current training loss:                0.466468
  current training accuracy:            0.842794
saving to ..., time used 795.820s
Seen 543000 samples, time used: 14509.226s
  current training loss:                0.466408
  current training accuracy:            0.842766
Seen 546000 samples, time used: 14593.093s
  current training loss:                0.466530
  current training accuracy:            0.842766
Seen 549000 samples, time used: 14674.560s
  current training loss:                0.466487
  current training accuracy:            0.842796
Epoch 16 of 20 took 14713.960s
  training loss:                0.466481
  training accuracy:            84.28 %
  validation loss:              0.462728
  validation accuracy:          82.35 %
  test loss:                    0.470575
  test accuracy:                81.96 %
saving to ./snli/word_by_word_model_epoch16.npz
Seen 3000 samples, time used: 79.726s
  current training loss:                0.432445
  current training accuracy:            0.856000
Seen 6000 samples, time used: 160.021s
  current training loss:                0.444389
  current training accuracy:            0.854000
Seen 9000 samples, time used: 239.363s
  current training loss:                0.439668
  current training accuracy:            0.854778
Seen 12000 samples, time used: 317.743s
  current training loss:                0.439977
  current training accuracy:            0.855167
Seen 15000 samples, time used: 399.205s
  current training loss:                0.444626
  current training accuracy:            0.853467
Seen 18000 samples, time used: 479.078s
  current training loss:                0.446464
  current training accuracy:            0.853056
Seen 21000 samples, time used: 559.956s
  current training loss:                0.450183
  current training accuracy:            0.852048
Seen 24000 samples, time used: 639.735s
  current training loss:                0.450091
  current training accuracy:            0.851542
Seen 27000 samples, time used: 720.819s
  current training loss:                0.451935
  current training accuracy:            0.851037
Seen 30000 samples, time used: 800.690s
  current training loss:                0.453119
  current training accuracy:            0.850433
saving to ..., time used 800.690s
Seen 33000 samples, time used: 882.015s
  current training loss:                0.452359
  current training accuracy:            0.851273
Seen 36000 samples, time used: 961.624s
  current training loss:                0.452619
  current training accuracy:            0.851194
Seen 39000 samples, time used: 1042.061s
  current training loss:                0.453275
  current training accuracy:            0.850949
Seen 42000 samples, time used: 1122.445s
  current training loss:                0.454986
  current training accuracy:            0.850524
Seen 45000 samples, time used: 1202.152s
  current training loss:                0.454260
  current training accuracy:            0.850511
Seen 48000 samples, time used: 1282.057s
  current training loss:                0.454566
  current training accuracy:            0.849771
Seen 51000 samples, time used: 1361.941s
  current training loss:                0.454827
  current training accuracy:            0.850333
Seen 54000 samples, time used: 1441.214s
  current training loss:                0.455343
  current training accuracy:            0.850407
Seen 57000 samples, time used: 1520.812s
  current training loss:                0.456861
  current training accuracy:            0.849772
Seen 60000 samples, time used: 1600.935s
  current training loss:                0.455004
  current training accuracy:            0.850300
saving to ..., time used 799.756s
Seen 63000 samples, time used: 1682.152s
  current training loss:                0.456244
  current training accuracy:            0.849508
Seen 66000 samples, time used: 1762.015s
  current training loss:                0.457441
  current training accuracy:            0.848909
Seen 69000 samples, time used: 1839.937s
  current training loss:                0.456946
  current training accuracy:            0.848986
Seen 72000 samples, time used: 1921.769s
  current training loss:                0.457360
  current training accuracy:            0.848792
Seen 75000 samples, time used: 2001.078s
  current training loss:                0.458105
  current training accuracy:            0.848387
Seen 78000 samples, time used: 2082.070s
  current training loss:                0.457066
  current training accuracy:            0.848756
Seen 81000 samples, time used: 2162.132s
  current training loss:                0.457196
  current training accuracy:            0.848531
Seen 84000 samples, time used: 2242.728s
  current training loss:                0.457340
  current training accuracy:            0.848429
Seen 87000 samples, time used: 2322.965s
  current training loss:                0.457507
  current training accuracy:            0.848184
Seen 90000 samples, time used: 2404.787s
  current training loss:                0.458457
  current training accuracy:            0.847822
saving to ..., time used 803.357s
Seen 93000 samples, time used: 2485.462s
  current training loss:                0.458617
  current training accuracy:            0.847871
Seen 96000 samples, time used: 2563.852s
  current training loss:                0.458553
  current training accuracy:            0.847656
Seen 99000 samples, time used: 2645.279s
  current training loss:                0.458899
  current training accuracy:            0.847424
Seen 102000 samples, time used: 2725.691s
  current training loss:                0.459145
  current training accuracy:            0.847529
Seen 105000 samples, time used: 2806.705s
  current training loss:                0.459866
  current training accuracy:            0.847314
Seen 108000 samples, time used: 2886.157s
  current training loss:                0.459775
  current training accuracy:            0.847500
Seen 111000 samples, time used: 2966.253s
  current training loss:                0.459479
  current training accuracy:            0.847514
Seen 114000 samples, time used: 3045.726s
  current training loss:                0.459590
  current training accuracy:            0.847474
Seen 117000 samples, time used: 3126.081s
  current training loss:                0.459355
  current training accuracy:            0.847436
Seen 120000 samples, time used: 3205.704s
  current training loss:                0.459851
  current training accuracy:            0.847192
saving to ..., time used 800.332s
Seen 123000 samples, time used: 3285.800s
  current training loss:                0.459875
  current training accuracy:            0.846976
Seen 126000 samples, time used: 3365.657s
  current training loss:                0.459608
  current training accuracy:            0.846968
Seen 129000 samples, time used: 3445.582s
  current training loss:                0.459957
  current training accuracy:            0.846876
Seen 132000 samples, time used: 3525.025s
  current training loss:                0.460144
  current training accuracy:            0.846780
Seen 135000 samples, time used: 3604.847s
  current training loss:                0.460326
  current training accuracy:            0.846593
Seen 138000 samples, time used: 3685.312s
  current training loss:                0.460495
  current training accuracy:            0.846471
Seen 141000 samples, time used: 3765.647s
  current training loss:                0.460628
  current training accuracy:            0.846546
Seen 144000 samples, time used: 3845.787s
  current training loss:                0.460617
  current training accuracy:            0.846549
Seen 147000 samples, time used: 3926.454s
  current training loss:                0.460859
  current training accuracy:            0.846490
Seen 150000 samples, time used: 4005.656s
  current training loss:                0.461054
  current training accuracy:            0.846380
saving to ..., time used 799.465s
Seen 153000 samples, time used: 4087.766s
  current training loss:                0.460868
  current training accuracy:            0.846359
Seen 156000 samples, time used: 4168.849s
  current training loss:                0.460993
  current training accuracy:            0.846237
Seen 159000 samples, time used: 4248.406s
  current training loss:                0.460834
  current training accuracy:            0.846270
Seen 162000 samples, time used: 4330.203s
  current training loss:                0.460672
  current training accuracy:            0.846253
Seen 165000 samples, time used: 4410.137s
  current training loss:                0.460562
  current training accuracy:            0.846376
Seen 168000 samples, time used: 4488.769s
  current training loss:                0.460548
  current training accuracy:            0.846488
Seen 171000 samples, time used: 4569.040s
  current training loss:                0.460963
  current training accuracy:            0.846368
Seen 174000 samples, time used: 4648.558s
  current training loss:                0.461092
  current training accuracy:            0.846362
Seen 177000 samples, time used: 4727.745s
  current training loss:                0.461396
  current training accuracy:            0.846198
Seen 180000 samples, time used: 4808.627s
  current training loss:                0.461152
  current training accuracy:            0.846267
saving to ..., time used 802.460s
Seen 183000 samples, time used: 4890.300s
  current training loss:                0.461295
  current training accuracy:            0.846120
Seen 186000 samples, time used: 4971.934s
  current training loss:                0.461293
  current training accuracy:            0.846167
Seen 189000 samples, time used: 5055.307s
  current training loss:                0.461131
  current training accuracy:            0.846101
Seen 192000 samples, time used: 5137.945s
  current training loss:                0.460768
  current training accuracy:            0.846297
Seen 195000 samples, time used: 5222.864s
  current training loss:                0.460356
  current training accuracy:            0.846487
Seen 198000 samples, time used: 5305.802s
  current training loss:                0.460324
  current training accuracy:            0.846399
Seen 201000 samples, time used: 5390.314s
  current training loss:                0.460246
  current training accuracy:            0.846393
Seen 204000 samples, time used: 5471.460s
  current training loss:                0.460255
  current training accuracy:            0.846490
Seen 207000 samples, time used: 5554.945s
  current training loss:                0.460377
  current training accuracy:            0.846338
Seen 210000 samples, time used: 5637.652s
  current training loss:                0.460408
  current training accuracy:            0.846329
saving to ..., time used 828.513s
Seen 213000 samples, time used: 5717.590s
  current training loss:                0.460145
  current training accuracy:            0.846399
Seen 216000 samples, time used: 5798.644s
  current training loss:                0.460080
  current training accuracy:            0.846477
Seen 219000 samples, time used: 5878.888s
  current training loss:                0.459906
  current training accuracy:            0.846534
Seen 222000 samples, time used: 5958.645s
  current training loss:                0.460043
  current training accuracy:            0.846468
Seen 225000 samples, time used: 6038.901s
  current training loss:                0.460162
  current training accuracy:            0.846382
Seen 228000 samples, time used: 6117.611s
  current training loss:                0.460209
  current training accuracy:            0.846193
Seen 231000 samples, time used: 6197.949s
  current training loss:                0.460176
  current training accuracy:            0.846208
Seen 234000 samples, time used: 6277.428s
  current training loss:                0.460291
  current training accuracy:            0.846111
Seen 237000 samples, time used: 6357.732s
  current training loss:                0.460363
  current training accuracy:            0.846127
Seen 240000 samples, time used: 6437.700s
  current training loss:                0.460306
  current training accuracy:            0.846108
saving to ..., time used 799.559s
Seen 243000 samples, time used: 6517.692s
  current training loss:                0.460191
  current training accuracy:            0.846189
Seen 246000 samples, time used: 6597.376s
  current training loss:                0.460124
  current training accuracy:            0.846183
Seen 249000 samples, time used: 6676.644s
  current training loss:                0.460215
  current training accuracy:            0.846209
Seen 252000 samples, time used: 6756.621s
  current training loss:                0.460310
  current training accuracy:            0.846167
Seen 255000 samples, time used: 6836.517s
  current training loss:                0.460120
  current training accuracy:            0.846196
Seen 258000 samples, time used: 6915.321s
  current training loss:                0.460166
  current training accuracy:            0.846240
Seen 261000 samples, time used: 6995.980s
  current training loss:                0.460140
  current training accuracy:            0.846268
Seen 264000 samples, time used: 7075.284s
  current training loss:                0.460258
  current training accuracy:            0.846155
Seen 267000 samples, time used: 7157.465s
  current training loss:                0.460411
  current training accuracy:            0.846135
Seen 270000 samples, time used: 7237.021s
  current training loss:                0.460623
  current training accuracy:            0.846067
saving to ..., time used 798.814s
Seen 273000 samples, time used: 7319.165s
  current training loss:                0.460736
  current training accuracy:            0.845956
Seen 276000 samples, time used: 7402.314s
  current training loss:                0.460771
  current training accuracy:            0.846022
Seen 279000 samples, time used: 7486.532s
  current training loss:                0.460748
  current training accuracy:            0.846047
Seen 282000 samples, time used: 7571.239s
  current training loss:                0.460936
  current training accuracy:            0.845904
Seen 285000 samples, time used: 7654.028s
  current training loss:                0.460821
  current training accuracy:            0.845846
Seen 288000 samples, time used: 7735.429s
  current training loss:                0.460671
  current training accuracy:            0.845944
Seen 291000 samples, time used: 7817.648s
  current training loss:                0.460823
  current training accuracy:            0.845959
Seen 294000 samples, time used: 7899.070s
  current training loss:                0.460923
  current training accuracy:            0.845905
Seen 297000 samples, time used: 7981.124s
  current training loss:                0.460942
  current training accuracy:            0.845882
Seen 300000 samples, time used: 8065.386s
  current training loss:                0.460813
  current training accuracy:            0.845900
saving to ..., time used 827.896s
Seen 303000 samples, time used: 8145.159s
  current training loss:                0.461119
  current training accuracy:            0.845865
Seen 306000 samples, time used: 8223.932s
  current training loss:                0.461092
  current training accuracy:            0.845820
Seen 309000 samples, time used: 8305.634s
  current training loss:                0.460882
  current training accuracy:            0.845900
Seen 312000 samples, time used: 8386.785s
  current training loss:                0.460749
  current training accuracy:            0.845936
Seen 315000 samples, time used: 8467.598s
  current training loss:                0.460768
  current training accuracy:            0.845930
Seen 318000 samples, time used: 8547.897s
  current training loss:                0.460854
  current training accuracy:            0.845871
Seen 321000 samples, time used: 8627.753s
  current training loss:                0.460865
  current training accuracy:            0.845900
Seen 324000 samples, time used: 8708.271s
  current training loss:                0.460970
  current training accuracy:            0.845812
Seen 327000 samples, time used: 8788.808s
  current training loss:                0.461047
  current training accuracy:            0.845774
Seen 330000 samples, time used: 8868.177s
  current training loss:                0.460988
  current training accuracy:            0.845836
saving to ..., time used 802.307s
Seen 333000 samples, time used: 8949.974s
  current training loss:                0.461010
  current training accuracy:            0.845793
Seen 336000 samples, time used: 9030.371s
  current training loss:                0.460907
  current training accuracy:            0.845833
Seen 339000 samples, time used: 9111.016s
  current training loss:                0.461032
  current training accuracy:            0.845761
Seen 342000 samples, time used: 9194.070s
  current training loss:                0.461056
  current training accuracy:            0.845743
Seen 345000 samples, time used: 9275.525s
  current training loss:                0.460986
  current training accuracy:            0.845823
Seen 348000 samples, time used: 9356.551s
  current training loss:                0.461160
  current training accuracy:            0.845779
Seen 351000 samples, time used: 9437.612s
  current training loss:                0.461300
  current training accuracy:            0.845681
Seen 354000 samples, time used: 9518.773s
  current training loss:                0.461409
  current training accuracy:            0.845658
Seen 357000 samples, time used: 9597.392s
  current training loss:                0.461430
  current training accuracy:            0.845633
Seen 360000 samples, time used: 9677.717s
  current training loss:                0.461400
  current training accuracy:            0.845683
saving to ..., time used 809.047s
Seen 363000 samples, time used: 9761.094s
  current training loss:                0.461313
  current training accuracy:            0.845691
Seen 366000 samples, time used: 9842.943s
  current training loss:                0.461267
  current training accuracy:            0.845653
Seen 369000 samples, time used: 9926.419s
  current training loss:                0.461200
  current training accuracy:            0.845661
Seen 372000 samples, time used: 10013.336s
  current training loss:                0.461209
  current training accuracy:            0.845672
Seen 375000 samples, time used: 10094.947s
  current training loss:                0.461464
  current training accuracy:            0.845523
Seen 378000 samples, time used: 10176.910s
  current training loss:                0.461545
  current training accuracy:            0.845463
Seen 381000 samples, time used: 10260.733s
  current training loss:                0.461589
  current training accuracy:            0.845438
Seen 384000 samples, time used: 10344.066s
  current training loss:                0.461583
  current training accuracy:            0.845451
Seen 387000 samples, time used: 10426.842s
  current training loss:                0.461508
  current training accuracy:            0.845481
Seen 390000 samples, time used: 10507.665s
  current training loss:                0.461593
  current training accuracy:            0.845397
saving to ..., time used 829.474s
Seen 393000 samples, time used: 10589.279s
  current training loss:                0.461585
  current training accuracy:            0.845338
Seen 396000 samples, time used: 10669.951s
  current training loss:                0.461735
  current training accuracy:            0.845290
Seen 399000 samples, time used: 10752.125s
  current training loss:                0.461794
  current training accuracy:            0.845288
Seen 402000 samples, time used: 10833.550s
  current training loss:                0.461849
  current training accuracy:            0.845301
Seen 405000 samples, time used: 10915.464s
  current training loss:                0.461822
  current training accuracy:            0.845264
Seen 408000 samples, time used: 11000.450s
  current training loss:                0.461675
  current training accuracy:            0.845346
Seen 411000 samples, time used: 11081.529s
  current training loss:                0.461801
  current training accuracy:            0.845314
Seen 414000 samples, time used: 11165.586s
  current training loss:                0.461542
  current training accuracy:            0.845476
Seen 417000 samples, time used: 11248.695s
  current training loss:                0.461554
  current training accuracy:            0.845468
Seen 420000 samples, time used: 11333.623s
  current training loss:                0.461574
  current training accuracy:            0.845400
saving to ..., time used 825.414s
Seen 423000 samples, time used: 11414.366s
  current training loss:                0.461697
  current training accuracy:            0.845352
Seen 426000 samples, time used: 11497.095s
  current training loss:                0.461884
  current training accuracy:            0.845298
Seen 429000 samples, time used: 11578.920s
  current training loss:                0.461836
  current training accuracy:            0.845359
Seen 432000 samples, time used: 11658.696s
  current training loss:                0.461848
  current training accuracy:            0.845312
Seen 435000 samples, time used: 11739.006s
  current training loss:                0.461838
  current training accuracy:            0.845338
Seen 438000 samples, time used: 11819.479s
  current training loss:                0.461919
  current training accuracy:            0.845263
Seen 441000 samples, time used: 11899.196s
  current training loss:                0.461997
  current training accuracy:            0.845245
Seen 444000 samples, time used: 11979.706s
  current training loss:                0.462027
  current training accuracy:            0.845257
Seen 447000 samples, time used: 12060.758s
  current training loss:                0.462064
  current training accuracy:            0.845219
Seen 450000 samples, time used: 12140.083s
  current training loss:                0.462088
  current training accuracy:            0.845207
saving to ..., time used 805.960s
Seen 453000 samples, time used: 12222.190s
  current training loss:                0.461915
  current training accuracy:            0.845283
Seen 456000 samples, time used: 12303.522s
  current training loss:                0.461951
  current training accuracy:            0.845230
Seen 459000 samples, time used: 12385.831s
  current training loss:                0.461940
  current training accuracy:            0.845203
Seen 462000 samples, time used: 12467.982s
  current training loss:                0.461957
  current training accuracy:            0.845221
Seen 465000 samples, time used: 12549.007s
  current training loss:                0.462231
  current training accuracy:            0.845095
Seen 468000 samples, time used: 12628.879s
  current training loss:                0.462285
  current training accuracy:            0.845120
Seen 471000 samples, time used: 12708.544s
  current training loss:                0.462475
  current training accuracy:            0.845053
Seen 474000 samples, time used: 12789.530s
  current training loss:                0.462513
  current training accuracy:            0.845046
Seen 477000 samples, time used: 12871.178s
  current training loss:                0.462413
  current training accuracy:            0.845069
Seen 480000 samples, time used: 12952.531s
  current training loss:                0.462345
  current training accuracy:            0.845048
saving to ..., time used 811.918s
Seen 483000 samples, time used: 13034.657s
  current training loss:                0.462500
  current training accuracy:            0.845010
Seen 486000 samples, time used: 13115.988s
  current training loss:                0.462647
  current training accuracy:            0.844909
Seen 489000 samples, time used: 13196.581s
  current training loss:                0.462610
  current training accuracy:            0.844908
Seen 492000 samples, time used: 13275.676s
  current training loss:                0.462727
  current training accuracy:            0.844843
Seen 495000 samples, time used: 13356.743s
  current training loss:                0.462692
  current training accuracy:            0.844820
Seen 498000 samples, time used: 13435.983s
  current training loss:                0.462748
  current training accuracy:            0.844795
Seen 501000 samples, time used: 13516.014s
  current training loss:                0.462665
  current training accuracy:            0.844868
Seen 504000 samples, time used: 13597.429s
  current training loss:                0.462649
  current training accuracy:            0.844875
Seen 507000 samples, time used: 13676.931s
  current training loss:                0.462909
  current training accuracy:            0.844781
Seen 510000 samples, time used: 13756.761s
  current training loss:                0.462793
  current training accuracy:            0.844800
saving to ..., time used 803.730s
Seen 513000 samples, time used: 13838.990s
  current training loss:                0.462853
  current training accuracy:            0.844778
Seen 516000 samples, time used: 13919.059s
  current training loss:                0.462814
  current training accuracy:            0.844785
Seen 519000 samples, time used: 14000.790s
  current training loss:                0.462893
  current training accuracy:            0.844715
Seen 522000 samples, time used: 14081.329s
  current training loss:                0.462951
  current training accuracy:            0.844684
Seen 525000 samples, time used: 14161.077s
  current training loss:                0.462845
  current training accuracy:            0.844712
Seen 528000 samples, time used: 14240.803s
  current training loss:                0.462884
  current training accuracy:            0.844737
Seen 531000 samples, time used: 14322.208s
  current training loss:                0.462987
  current training accuracy:            0.844710
Seen 534000 samples, time used: 14402.132s
  current training loss:                0.463170
  current training accuracy:            0.844601
Seen 537000 samples, time used: 14484.013s
  current training loss:                0.463270
  current training accuracy:            0.844583
Seen 540000 samples, time used: 14564.649s
  current training loss:                0.463312
  current training accuracy:            0.844583
saving to ..., time used 807.420s
Seen 543000 samples, time used: 14645.728s
  current training loss:                0.463384
  current training accuracy:            0.844495
Seen 546000 samples, time used: 14726.269s
  current training loss:                0.463486
  current training accuracy:            0.844443
Seen 549000 samples, time used: 14806.300s
  current training loss:                0.463585
  current training accuracy:            0.844390
Epoch 17 of 20 took 14839.108s
  training loss:                0.463551
  training accuracy:            84.44 %
  validation loss:              0.458345
  validation accuracy:          82.35 %
  test loss:                    0.472640
  test accuracy:                81.66 %
saving to ./snli/word_by_word_model_epoch17.npz
Seen 3000 samples, time used: 80.652s
  current training loss:                0.446606
  current training accuracy:            0.859667
Seen 6000 samples, time used: 161.968s
  current training loss:                0.443802
  current training accuracy:            0.855500
Seen 9000 samples, time used: 242.561s
  current training loss:                0.436617
  current training accuracy:            0.856222
Seen 12000 samples, time used: 322.446s
  current training loss:                0.437387
  current training accuracy:            0.855000
Seen 15000 samples, time used: 404.581s
  current training loss:                0.440814
  current training accuracy:            0.852067
Seen 18000 samples, time used: 483.643s
  current training loss:                0.439006
  current training accuracy:            0.853833
Seen 21000 samples, time used: 562.140s
  current training loss:                0.435854
  current training accuracy:            0.855667
Seen 24000 samples, time used: 643.621s
  current training loss:                0.439681
  current training accuracy:            0.855417
Seen 27000 samples, time used: 724.605s
  current training loss:                0.439827
  current training accuracy:            0.854889
Seen 30000 samples, time used: 806.381s
  current training loss:                0.442866
  current training accuracy:            0.854033
saving to ..., time used 806.381s
Seen 33000 samples, time used: 889.968s
  current training loss:                0.444617
  current training accuracy:            0.853636
Seen 36000 samples, time used: 971.157s
  current training loss:                0.444533
  current training accuracy:            0.853000
Seen 39000 samples, time used: 1051.205s
  current training loss:                0.445254
  current training accuracy:            0.853179
Seen 42000 samples, time used: 1130.373s
  current training loss:                0.445563
  current training accuracy:            0.853357
Seen 45000 samples, time used: 1211.561s
  current training loss:                0.446872
  current training accuracy:            0.852911
Seen 48000 samples, time used: 1291.835s
  current training loss:                0.446813
  current training accuracy:            0.852812
Seen 51000 samples, time used: 1371.636s
  current training loss:                0.447168
  current training accuracy:            0.852647
Seen 54000 samples, time used: 1451.858s
  current training loss:                0.446783
  current training accuracy:            0.852833
Seen 57000 samples, time used: 1534.044s
  current training loss:                0.447534
  current training accuracy:            0.852333
Seen 60000 samples, time used: 1614.746s
  current training loss:                0.447975
  current training accuracy:            0.852233
saving to ..., time used 807.874s
Seen 63000 samples, time used: 1698.514s
  current training loss:                0.448802
  current training accuracy:            0.851857
Seen 66000 samples, time used: 1780.402s
  current training loss:                0.448145
  current training accuracy:            0.852273
Seen 69000 samples, time used: 1862.447s
  current training loss:                0.449843
  current training accuracy:            0.851580
Seen 72000 samples, time used: 1943.291s
  current training loss:                0.450335
  current training accuracy:            0.851208
Seen 75000 samples, time used: 2025.333s
  current training loss:                0.450933
  current training accuracy:            0.851040
Seen 78000 samples, time used: 2106.762s
  current training loss:                0.451667
  current training accuracy:            0.850436
Seen 81000 samples, time used: 2190.193s
  current training loss:                0.451953
  current training accuracy:            0.850321
Seen 84000 samples, time used: 2271.024s
  current training loss:                0.452378
  current training accuracy:            0.850286
Seen 87000 samples, time used: 2351.729s
  current training loss:                0.451945
  current training accuracy:            0.850115
Seen 90000 samples, time used: 2435.803s
  current training loss:                0.451950
  current training accuracy:            0.850156
saving to ..., time used 820.563s
Seen 93000 samples, time used: 2515.603s
  current training loss:                0.451829
  current training accuracy:            0.850065
Seen 96000 samples, time used: 2594.111s
  current training loss:                0.452020
  current training accuracy:            0.850115
Seen 99000 samples, time used: 2675.646s
  current training loss:                0.452329
  current training accuracy:            0.849929
Seen 102000 samples, time used: 2755.164s
  current training loss:                0.452616
  current training accuracy:            0.849873
Seen 105000 samples, time used: 2836.443s
  current training loss:                0.452905
  current training accuracy:            0.849800
Seen 108000 samples, time used: 2917.494s
  current training loss:                0.453033
  current training accuracy:            0.849583
Seen 111000 samples, time used: 2997.568s
  current training loss:                0.452743
  current training accuracy:            0.849649
Seen 114000 samples, time used: 3078.722s
  current training loss:                0.452549
  current training accuracy:            0.849658
Seen 117000 samples, time used: 3158.210s
  current training loss:                0.452600
  current training accuracy:            0.849479
Seen 120000 samples, time used: 3239.355s
  current training loss:                0.452717
  current training accuracy:            0.849550
saving to ..., time used 803.041s
Seen 123000 samples, time used: 3319.237s
  current training loss:                0.452581
  current training accuracy:            0.849431
Seen 126000 samples, time used: 3398.826s
  current training loss:                0.453026
  current training accuracy:            0.849190
Seen 129000 samples, time used: 3479.133s
  current training loss:                0.453515
  current training accuracy:            0.849357
Seen 132000 samples, time used: 3559.245s
  current training loss:                0.454194
  current training accuracy:            0.848947
Seen 135000 samples, time used: 3639.865s
  current training loss:                0.454448
  current training accuracy:            0.848756
Seen 138000 samples, time used: 3720.265s
  current training loss:                0.454501
  current training accuracy:            0.848746
Seen 141000 samples, time used: 3800.216s
  current training loss:                0.454070
  current training accuracy:            0.849021
Seen 144000 samples, time used: 3878.737s
  current training loss:                0.454057
  current training accuracy:            0.848979
Seen 147000 samples, time used: 3957.509s
  current training loss:                0.453804
  current training accuracy:            0.849020
Seen 150000 samples, time used: 4035.975s
  current training loss:                0.453842
  current training accuracy:            0.848887
saving to ..., time used 796.083s
Seen 153000 samples, time used: 4120.844s
  current training loss:                0.453735
  current training accuracy:            0.848935
Seen 156000 samples, time used: 4202.178s
  current training loss:                0.453676
  current training accuracy:            0.848936
Seen 159000 samples, time used: 4284.146s
  current training loss:                0.453673
  current training accuracy:            0.848912
Seen 162000 samples, time used: 4364.389s
  current training loss:                0.453983
  current training accuracy:            0.848858
Seen 165000 samples, time used: 4446.865s
  current training loss:                0.453965
  current training accuracy:            0.848794
Seen 168000 samples, time used: 4527.640s
  current training loss:                0.454146
  current training accuracy:            0.848690
Seen 171000 samples, time used: 4609.031s
  current training loss:                0.454767
  current training accuracy:            0.848368
Seen 174000 samples, time used: 4693.145s
  current training loss:                0.455033
  current training accuracy:            0.848241
Seen 177000 samples, time used: 4778.216s
  current training loss:                0.455368
  current training accuracy:            0.848119
Seen 180000 samples, time used: 4859.416s
  current training loss:                0.455450
  current training accuracy:            0.848078
saving to ..., time used 822.870s
Seen 183000 samples, time used: 4941.410s
  current training loss:                0.455528
  current training accuracy:            0.847940
Seen 186000 samples, time used: 5022.636s
  current training loss:                0.455851
  current training accuracy:            0.847866
Seen 189000 samples, time used: 5102.998s
  current training loss:                0.456044
  current training accuracy:            0.847698
Seen 192000 samples, time used: 5182.425s
  current training loss:                0.456136
  current training accuracy:            0.847615
Seen 195000 samples, time used: 5263.396s
  current training loss:                0.456073
  current training accuracy:            0.847538
Seen 198000 samples, time used: 5342.114s
  current training loss:                0.456128
  current training accuracy:            0.847652
Seen 201000 samples, time used: 5422.749s
  current training loss:                0.455978
  current training accuracy:            0.847692
Seen 204000 samples, time used: 5502.250s
  current training loss:                0.456257
  current training accuracy:            0.847515
Seen 207000 samples, time used: 5583.610s
  current training loss:                0.456254
  current training accuracy:            0.847633
Seen 210000 samples, time used: 5664.018s
  current training loss:                0.456388
  current training accuracy:            0.847633
saving to ..., time used 804.100s
Seen 213000 samples, time used: 5744.406s
  current training loss:                0.456258
  current training accuracy:            0.847746
Seen 216000 samples, time used: 5826.065s
  current training loss:                0.456308
  current training accuracy:            0.847778
Seen 219000 samples, time used: 5905.934s
  current training loss:                0.456380
  current training accuracy:            0.847795
Seen 222000 samples, time used: 5986.528s
  current training loss:                0.456594
  current training accuracy:            0.847658
Seen 225000 samples, time used: 6064.817s
  current training loss:                0.456659
  current training accuracy:            0.847764
Seen 228000 samples, time used: 6145.409s
  current training loss:                0.456946
  current training accuracy:            0.847732
Seen 231000 samples, time used: 6225.502s
  current training loss:                0.456779
  current training accuracy:            0.847827
Seen 234000 samples, time used: 6305.740s
  current training loss:                0.456941
  current training accuracy:            0.847765
Seen 237000 samples, time used: 6387.896s
  current training loss:                0.456789
  current training accuracy:            0.847793
Seen 240000 samples, time used: 6469.735s
  current training loss:                0.456654
  current training accuracy:            0.847800
saving to ..., time used 805.145s
Seen 243000 samples, time used: 6552.701s
  current training loss:                0.456665
  current training accuracy:            0.847807
Seen 246000 samples, time used: 6635.764s
  current training loss:                0.456415
  current training accuracy:            0.847915
Seen 249000 samples, time used: 6717.419s
  current training loss:                0.456693
  current training accuracy:            0.847795
Seen 252000 samples, time used: 6798.933s
  current training loss:                0.456714
  current training accuracy:            0.847766
Seen 255000 samples, time used: 6881.542s
  current training loss:                0.456813
  current training accuracy:            0.847686
Seen 258000 samples, time used: 6964.510s
  current training loss:                0.457017
  current training accuracy:            0.847655
Seen 261000 samples, time used: 7046.724s
  current training loss:                0.456956
  current training accuracy:            0.847705
Seen 264000 samples, time used: 7127.857s
  current training loss:                0.457006
  current training accuracy:            0.847716
Seen 267000 samples, time used: 7207.925s
  current training loss:                0.457047
  current training accuracy:            0.847704
Seen 270000 samples, time used: 7288.533s
  current training loss:                0.457131
  current training accuracy:            0.847574
saving to ..., time used 818.286s
Seen 273000 samples, time used: 7367.838s
  current training loss:                0.456954
  current training accuracy:            0.847652
Seen 276000 samples, time used: 7445.894s
  current training loss:                0.457215
  current training accuracy:            0.847558
Seen 279000 samples, time used: 7524.793s
  current training loss:                0.457389
  current training accuracy:            0.847437
Seen 282000 samples, time used: 7604.673s
  current training loss:                0.457489
  current training accuracy:            0.847376
Seen 285000 samples, time used: 7688.715s
  current training loss:                0.457415
  current training accuracy:            0.847389
Seen 288000 samples, time used: 7769.912s
  current training loss:                0.457449
  current training accuracy:            0.847281
Seen 291000 samples, time used: 7852.369s
  current training loss:                0.457217
  current training accuracy:            0.847268
Seen 294000 samples, time used: 7934.394s
  current training loss:                0.457298
  current training accuracy:            0.847160
Seen 297000 samples, time used: 8016.061s
  current training loss:                0.457519
  current training accuracy:            0.847037
Seen 300000 samples, time used: 8096.275s
  current training loss:                0.457657
  current training accuracy:            0.846937
saving to ..., time used 807.240s
Seen 303000 samples, time used: 8175.723s
  current training loss:                0.457665
  current training accuracy:            0.846891
Seen 306000 samples, time used: 8256.506s
  current training loss:                0.457758
  current training accuracy:            0.846895
Seen 309000 samples, time used: 8338.134s
  current training loss:                0.457717
  current training accuracy:            0.846890
Seen 312000 samples, time used: 8417.634s
  current training loss:                0.458020
  current training accuracy:            0.846702
Seen 315000 samples, time used: 8497.719s
  current training loss:                0.457952
  current training accuracy:            0.846679
Seen 318000 samples, time used: 8578.154s
  current training loss:                0.458142
  current training accuracy:            0.846664
Seen 321000 samples, time used: 8656.613s
  current training loss:                0.458135
  current training accuracy:            0.846723
Seen 324000 samples, time used: 8737.116s
  current training loss:                0.458416
  current training accuracy:            0.846630
Seen 327000 samples, time used: 8815.775s
  current training loss:                0.458544
  current training accuracy:            0.846535
Seen 330000 samples, time used: 8895.350s
  current training loss:                0.458704
  current training accuracy:            0.846452
saving to ..., time used 798.600s
Seen 333000 samples, time used: 8977.421s
  current training loss:                0.458923
  current training accuracy:            0.846375
Seen 336000 samples, time used: 9061.036s
  current training loss:                0.459055
  current training accuracy:            0.846292
Seen 339000 samples, time used: 9146.225s
  current training loss:                0.459167
  current training accuracy:            0.846215
Seen 342000 samples, time used: 9228.133s
  current training loss:                0.459224
  current training accuracy:            0.846149
Seen 345000 samples, time used: 9311.303s
  current training loss:                0.459320
  current training accuracy:            0.846090
Seen 348000 samples, time used: 9394.874s
  current training loss:                0.459376
  current training accuracy:            0.846152
Seen 351000 samples, time used: 9476.688s
  current training loss:                0.459414
  current training accuracy:            0.846097
Seen 354000 samples, time used: 9558.964s
  current training loss:                0.459465
  current training accuracy:            0.846068
Seen 357000 samples, time used: 9640.939s
  current training loss:                0.459342
  current training accuracy:            0.846084
Seen 360000 samples, time used: 9723.306s
  current training loss:                0.459310
  current training accuracy:            0.846064
saving to ..., time used 827.476s
Seen 363000 samples, time used: 9805.387s
  current training loss:                0.459324
  current training accuracy:            0.846019
Seen 366000 samples, time used: 9888.160s
  current training loss:                0.459196
  current training accuracy:            0.846104
Seen 369000 samples, time used: 9969.571s
  current training loss:                0.459147
  current training accuracy:            0.846122
Seen 372000 samples, time used: 10052.212s
  current training loss:                0.459312
  current training accuracy:            0.846059
Seen 375000 samples, time used: 10133.707s
  current training loss:                0.459353
  current training accuracy:            0.846029
Seen 378000 samples, time used: 10216.427s
  current training loss:                0.459258
  current training accuracy:            0.846040
Seen 381000 samples, time used: 10299.802s
  current training loss:                0.459044
  current training accuracy:            0.846097
Seen 384000 samples, time used: 10381.164s
  current training loss:                0.459016
  current training accuracy:            0.846078
Seen 387000 samples, time used: 10462.773s
  current training loss:                0.458944
  current training accuracy:            0.846090
Seen 390000 samples, time used: 10545.618s
  current training loss:                0.459005
  current training accuracy:            0.846077
saving to ..., time used 821.826s
Seen 393000 samples, time used: 10625.300s
  current training loss:                0.458962
  current training accuracy:            0.846122
Seen 396000 samples, time used: 10705.774s
  current training loss:                0.458785
  current training accuracy:            0.846194
Seen 399000 samples, time used: 10785.551s
  current training loss:                0.458915
  current training accuracy:            0.846148
Seen 402000 samples, time used: 10864.403s
  current training loss:                0.459077
  current training accuracy:            0.846119
Seen 405000 samples, time used: 10943.813s
  current training loss:                0.459213
  current training accuracy:            0.846044
Seen 408000 samples, time used: 11023.720s
  current training loss:                0.459329
  current training accuracy:            0.846010
Seen 411000 samples, time used: 11103.021s
  current training loss:                0.459464
  current training accuracy:            0.845959
Seen 414000 samples, time used: 11183.367s
  current training loss:                0.459452
  current training accuracy:            0.845952
Seen 417000 samples, time used: 11261.838s
  current training loss:                0.459543
  current training accuracy:            0.845880
Seen 420000 samples, time used: 11341.623s
  current training loss:                0.459691
  current training accuracy:            0.845829
saving to ..., time used 795.519s
Seen 423000 samples, time used: 11421.756s
  current training loss:                0.459684
  current training accuracy:            0.845842
Seen 426000 samples, time used: 11501.567s
  current training loss:                0.459775
  current training accuracy:            0.845815
Seen 429000 samples, time used: 11581.292s
  current training loss:                0.459643
  current training accuracy:            0.845872
Seen 432000 samples, time used: 11659.868s
  current training loss:                0.459779
  current training accuracy:            0.845757
Seen 435000 samples, time used: 11738.498s
  current training loss:                0.459906
  current training accuracy:            0.845740
Seen 438000 samples, time used: 11818.407s
  current training loss:                0.459937
  current training accuracy:            0.845721
Seen 441000 samples, time used: 11897.761s
  current training loss:                0.460016
  current training accuracy:            0.845717
Seen 444000 samples, time used: 11976.047s
  current training loss:                0.460103
  current training accuracy:            0.845633
Seen 447000 samples, time used: 12055.021s
  current training loss:                0.460135
  current training accuracy:            0.845608
Seen 450000 samples, time used: 12133.633s
  current training loss:                0.460331
  current training accuracy:            0.845549
saving to ..., time used 791.474s
Seen 453000 samples, time used: 12215.707s
  current training loss:                0.460463
  current training accuracy:            0.845393
Seen 456000 samples, time used: 12295.648s
  current training loss:                0.460550
  current training accuracy:            0.845357
Seen 459000 samples, time used: 12375.414s
  current training loss:                0.460462
  current training accuracy:            0.845386
Seen 462000 samples, time used: 12454.751s
  current training loss:                0.460691
  current training accuracy:            0.845307
Seen 465000 samples, time used: 12534.833s
  current training loss:                0.460707
  current training accuracy:            0.845346
Seen 468000 samples, time used: 12613.922s
  current training loss:                0.460907
  current training accuracy:            0.845220
Seen 471000 samples, time used: 12693.079s
  current training loss:                0.461004
  current training accuracy:            0.845195
Seen 474000 samples, time used: 12772.296s
  current training loss:                0.461113
  current training accuracy:            0.845086
Seen 477000 samples, time used: 12851.731s
  current training loss:                0.461188
  current training accuracy:            0.845015
Seen 480000 samples, time used: 12931.333s
  current training loss:                0.461277
  current training accuracy:            0.844981
saving to ..., time used 797.224s
Seen 483000 samples, time used: 13009.848s
  current training loss:                0.461350
  current training accuracy:            0.844919
Seen 486000 samples, time used: 13089.206s
  current training loss:                0.461601
  current training accuracy:            0.844800
Seen 489000 samples, time used: 13168.841s
  current training loss:                0.461725
  current training accuracy:            0.844730
Seen 492000 samples, time used: 13247.181s
  current training loss:                0.461605
  current training accuracy:            0.844772
Seen 495000 samples, time used: 13326.783s
  current training loss:                0.461591
  current training accuracy:            0.844786
Seen 498000 samples, time used: 13406.671s
  current training loss:                0.461577
  current training accuracy:            0.844797
Seen 501000 samples, time used: 13486.760s
  current training loss:                0.461653
  current training accuracy:            0.844816
Seen 504000 samples, time used: 13564.948s
  current training loss:                0.461736
  current training accuracy:            0.844718
Seen 507000 samples, time used: 13645.621s
  current training loss:                0.461800
  current training accuracy:            0.844716
Seen 510000 samples, time used: 13723.737s
  current training loss:                0.461944
  current training accuracy:            0.844649
saving to ..., time used 791.927s
Seen 513000 samples, time used: 13802.377s
  current training loss:                0.462137
  current training accuracy:            0.844583
Seen 516000 samples, time used: 13881.921s
  current training loss:                0.462116
  current training accuracy:            0.844612
Seen 519000 samples, time used: 13960.632s
  current training loss:                0.462279
  current training accuracy:            0.844551
Seen 522000 samples, time used: 14038.110s
  current training loss:                0.462234
  current training accuracy:            0.844557
Seen 525000 samples, time used: 14116.235s
  current training loss:                0.462207
  current training accuracy:            0.844552
Seen 528000 samples, time used: 14195.250s
  current training loss:                0.462168
  current training accuracy:            0.844606
Seen 531000 samples, time used: 14273.043s
  current training loss:                0.462198
  current training accuracy:            0.844601
Seen 534000 samples, time used: 14352.066s
  current training loss:                0.462176
  current training accuracy:            0.844586
Seen 537000 samples, time used: 14431.178s
  current training loss:                0.462086
  current training accuracy:            0.844596
Seen 540000 samples, time used: 14511.219s
  current training loss:                0.462152
  current training accuracy:            0.844600
saving to ..., time used 786.933s
Seen 543000 samples, time used: 14591.686s
  current training loss:                0.462212
  current training accuracy:            0.844589
Seen 546000 samples, time used: 14671.629s
  current training loss:                0.462244
  current training accuracy:            0.844597
Seen 549000 samples, time used: 14750.710s
  current training loss:                0.462287
  current training accuracy:            0.844576
Epoch 18 of 20 took 14783.006s
  training loss:                0.462330
  training accuracy:            84.45 %
  validation loss:              0.453374
  validation accuracy:          82.52 %
  test loss:                    0.469899
  test accuracy:                81.98 %
saving to ./snli/word_by_word_model_epoch18.npz
Seen 3000 samples, time used: 80.909s
  current training loss:                0.460079
  current training accuracy:            0.848000
Seen 6000 samples, time used: 160.874s
  current training loss:                0.441789
  current training accuracy:            0.853000
Seen 9000 samples, time used: 240.725s
  current training loss:                0.442708
  current training accuracy:            0.854556
Seen 12000 samples, time used: 320.860s
  current training loss:                0.441317
  current training accuracy:            0.854250
Seen 15000 samples, time used: 398.886s
  current training loss:                0.445436
  current training accuracy:            0.853200
Seen 18000 samples, time used: 477.754s
  current training loss:                0.445453
  current training accuracy:            0.854111
Seen 21000 samples, time used: 556.356s
  current training loss:                0.448171
  current training accuracy:            0.852952
Seen 24000 samples, time used: 634.697s
  current training loss:                0.447578
  current training accuracy:            0.853500
Seen 27000 samples, time used: 714.097s
  current training loss:                0.448984
  current training accuracy:            0.852741
Seen 30000 samples, time used: 793.537s
  current training loss:                0.448963
  current training accuracy:            0.852333
saving to ..., time used 793.537s
Seen 33000 samples, time used: 871.947s
  current training loss:                0.450725
  current training accuracy:            0.852030
Seen 36000 samples, time used: 950.359s
  current training loss:                0.450515
  current training accuracy:            0.852444
Seen 39000 samples, time used: 1028.136s
  current training loss:                0.449898
  current training accuracy:            0.852692
Seen 42000 samples, time used: 1106.572s
  current training loss:                0.451949
  current training accuracy:            0.852190
Seen 45000 samples, time used: 1187.112s
  current training loss:                0.453127
  current training accuracy:            0.851267
Seen 48000 samples, time used: 1266.491s
  current training loss:                0.452390
  current training accuracy:            0.851458
Seen 51000 samples, time used: 1344.093s
  current training loss:                0.453070
  current training accuracy:            0.850745
Seen 54000 samples, time used: 1421.773s
  current training loss:                0.452551
  current training accuracy:            0.850944
Seen 57000 samples, time used: 1500.128s
  current training loss:                0.453260
  current training accuracy:            0.850632
Seen 60000 samples, time used: 1579.241s
  current training loss:                0.453014
  current training accuracy:            0.850433
saving to ..., time used 785.189s
Seen 63000 samples, time used: 1658.972s
  current training loss:                0.452388
  current training accuracy:            0.850556
Seen 66000 samples, time used: 1738.698s
  current training loss:                0.453017
  current training accuracy:            0.850455
Seen 69000 samples, time used: 1818.541s
  current training loss:                0.452573
  current training accuracy:            0.850478
Seen 72000 samples, time used: 1896.606s
  current training loss:                0.452392
  current training accuracy:            0.850736
Seen 75000 samples, time used: 1975.319s
  current training loss:                0.452195
  current training accuracy:            0.850893
Seen 78000 samples, time used: 2055.400s
  current training loss:                0.452262
  current training accuracy:            0.850795
Seen 81000 samples, time used: 2134.274s
  current training loss:                0.452734
  current training accuracy:            0.850630
Seen 84000 samples, time used: 2213.933s
  current training loss:                0.452981
  current training accuracy:            0.850274
Seen 87000 samples, time used: 2292.626s
  current training loss:                0.453021
  current training accuracy:            0.850218
Seen 90000 samples, time used: 2371.379s
  current training loss:                0.453649
  current training accuracy:            0.849622
saving to ..., time used 791.636s
Seen 93000 samples, time used: 2452.841s
  current training loss:                0.453398
  current training accuracy:            0.849570
Seen 96000 samples, time used: 2530.654s
  current training loss:                0.453029
  current training accuracy:            0.849667
Seen 99000 samples, time used: 2609.193s
  current training loss:                0.452994
  current training accuracy:            0.849677
Seen 102000 samples, time used: 2688.794s
  current training loss:                0.453537
  current training accuracy:            0.849529
Seen 105000 samples, time used: 2765.959s
  current training loss:                0.453615
  current training accuracy:            0.849610
Seen 108000 samples, time used: 2844.344s
  current training loss:                0.453849
  current training accuracy:            0.849537
Seen 111000 samples, time used: 2922.710s
  current training loss:                0.453854
  current training accuracy:            0.849613
Seen 114000 samples, time used: 3001.079s
  current training loss:                0.453265
  current training accuracy:            0.849807
Seen 117000 samples, time used: 3079.421s
  current training loss:                0.453416
  current training accuracy:            0.849658
Seen 120000 samples, time used: 3159.422s
  current training loss:                0.453552
  current training accuracy:            0.849767
saving to ..., time used 787.501s
Seen 123000 samples, time used: 3240.203s
  current training loss:                0.453953
  current training accuracy:            0.849447
Seen 126000 samples, time used: 3317.898s
  current training loss:                0.454043
  current training accuracy:            0.849444
Seen 129000 samples, time used: 3397.937s
  current training loss:                0.454034
  current training accuracy:            0.849504
Seen 132000 samples, time used: 3477.952s
  current training loss:                0.454893
  current training accuracy:            0.849227
Seen 135000 samples, time used: 3557.427s
  current training loss:                0.454977
  current training accuracy:            0.849030
Seen 138000 samples, time used: 3637.674s
  current training loss:                0.454818
  current training accuracy:            0.849152
Seen 141000 samples, time used: 3717.587s
  current training loss:                0.454754
  current training accuracy:            0.849050
Seen 144000 samples, time used: 3798.262s
  current training loss:                0.454941
  current training accuracy:            0.848951
Seen 147000 samples, time used: 3877.091s
  current training loss:                0.454910
  current training accuracy:            0.848857
Seen 150000 samples, time used: 3957.285s
  current training loss:                0.454920
  current training accuracy:            0.848893
saving to ..., time used 797.309s
Seen 153000 samples, time used: 4036.706s
  current training loss:                0.455015
  current training accuracy:            0.848712
Seen 156000 samples, time used: 4114.377s
  current training loss:                0.455114
  current training accuracy:            0.848673
Seen 159000 samples, time used: 4193.464s
  current training loss:                0.455377
  current training accuracy:            0.848472
Seen 162000 samples, time used: 4272.214s
  current training loss:                0.455233
  current training accuracy:            0.848389
Seen 165000 samples, time used: 4350.575s
  current training loss:                0.455172
  current training accuracy:            0.848448
Seen 168000 samples, time used: 4431.612s
  current training loss:                0.455344
  current training accuracy:            0.848429
Seen 171000 samples, time used: 4509.100s
  current training loss:                0.455290
  current training accuracy:            0.848398
Seen 174000 samples, time used: 4588.982s
  current training loss:                0.455263
  current training accuracy:            0.848374
Seen 177000 samples, time used: 4666.166s
  current training loss:                0.454918
  current training accuracy:            0.848435
Seen 180000 samples, time used: 4745.465s
  current training loss:                0.454823
  current training accuracy:            0.848456
saving to ..., time used 787.642s
Seen 183000 samples, time used: 4827.753s
  current training loss:                0.454985
  current training accuracy:            0.848377
Seen 186000 samples, time used: 4911.020s
  current training loss:                0.455116
  current training accuracy:            0.848242
Seen 189000 samples, time used: 4992.726s
  current training loss:                0.455208
  current training accuracy:            0.848190
Seen 192000 samples, time used: 5074.976s
  current training loss:                0.455363
  current training accuracy:            0.848036
Seen 195000 samples, time used: 5156.258s
  current training loss:                0.455823
  current training accuracy:            0.847928
Seen 198000 samples, time used: 5237.209s
  current training loss:                0.455365
  current training accuracy:            0.848146
Seen 201000 samples, time used: 5317.865s
  current training loss:                0.455669
  current training accuracy:            0.847985
Seen 204000 samples, time used: 5399.006s
  current training loss:                0.455753
  current training accuracy:            0.847985
Seen 207000 samples, time used: 5483.179s
  current training loss:                0.455562
  current training accuracy:            0.848053
Seen 210000 samples, time used: 5564.744s
  current training loss:                0.455285
  current training accuracy:            0.848029
saving to ..., time used 818.786s
Seen 213000 samples, time used: 5646.188s
  current training loss:                0.455213
  current training accuracy:            0.848047
Seen 216000 samples, time used: 5727.172s
  current training loss:                0.455199
  current training accuracy:            0.848097
Seen 219000 samples, time used: 5808.160s
  current training loss:                0.455394
  current training accuracy:            0.848005
Seen 222000 samples, time used: 5892.398s
  current training loss:                0.455500
  current training accuracy:            0.848041
Seen 225000 samples, time used: 5972.981s
  current training loss:                0.455536
  current training accuracy:            0.848089
Seen 228000 samples, time used: 6054.626s
  current training loss:                0.455414
  current training accuracy:            0.848061
Seen 231000 samples, time used: 6139.094s
  current training loss:                0.455460
  current training accuracy:            0.848069
Seen 234000 samples, time used: 6221.275s
  current training loss:                0.455578
  current training accuracy:            0.848120
Seen 237000 samples, time used: 6304.186s
  current training loss:                0.455555
  current training accuracy:            0.848211
Seen 240000 samples, time used: 6386.401s
  current training loss:                0.455672
  current training accuracy:            0.848125
saving to ..., time used 821.155s
Seen 243000 samples, time used: 6472.048s
  current training loss:                0.456058
  current training accuracy:            0.847967
Seen 246000 samples, time used: 6553.709s
  current training loss:                0.456108
  current training accuracy:            0.847935
Seen 249000 samples, time used: 6634.624s
  current training loss:                0.456128
  current training accuracy:            0.847851
Seen 252000 samples, time used: 6717.127s
  current training loss:                0.456115
  current training accuracy:            0.847885
Seen 255000 samples, time used: 6797.295s
  current training loss:                0.456120
  current training accuracy:            0.847969
Seen 258000 samples, time used: 6879.369s
  current training loss:                0.456227
  current training accuracy:            0.847915
Seen 261000 samples, time used: 6962.118s
  current training loss:                0.456449
  current training accuracy:            0.847824
Seen 264000 samples, time used: 7042.989s
  current training loss:                0.456523
  current training accuracy:            0.847773
Seen 267000 samples, time used: 7123.284s
  current training loss:                0.456414
  current training accuracy:            0.847873
Seen 270000 samples, time used: 7203.747s
  current training loss:                0.456427
  current training accuracy:            0.847770
saving to ..., time used 816.863s
Seen 273000 samples, time used: 7283.156s
  current training loss:                0.456701
  current training accuracy:            0.847630
Seen 276000 samples, time used: 7362.861s
  current training loss:                0.456793
  current training accuracy:            0.847630
Seen 279000 samples, time used: 7441.424s
  current training loss:                0.456928
  current training accuracy:            0.847591
Seen 282000 samples, time used: 7520.347s
  current training loss:                0.456802
  current training accuracy:            0.847645
Seen 285000 samples, time used: 7600.524s
  current training loss:                0.456804
  current training accuracy:            0.847765
Seen 288000 samples, time used: 7681.008s
  current training loss:                0.457071
  current training accuracy:            0.847649
Seen 291000 samples, time used: 7759.718s
  current training loss:                0.457066
  current training accuracy:            0.847629
Seen 294000 samples, time used: 7838.477s
  current training loss:                0.457283
  current training accuracy:            0.847497
Seen 297000 samples, time used: 7919.152s
  current training loss:                0.457377
  current training accuracy:            0.847451
Seen 300000 samples, time used: 7998.725s
  current training loss:                0.457426
  current training accuracy:            0.847410
saving to ..., time used 794.507s
Seen 303000 samples, time used: 8078.191s
  current training loss:                0.457392
  current training accuracy:            0.847446
Seen 306000 samples, time used: 8158.107s
  current training loss:                0.457576
  current training accuracy:            0.847343
Seen 309000 samples, time used: 8237.653s
  current training loss:                0.457444
  current training accuracy:            0.847398
Seen 312000 samples, time used: 8317.212s
  current training loss:                0.457823
  current training accuracy:            0.847205
Seen 315000 samples, time used: 8394.913s
  current training loss:                0.458097
  current training accuracy:            0.847165
Seen 318000 samples, time used: 8475.525s
  current training loss:                0.458108
  current training accuracy:            0.847176
Seen 321000 samples, time used: 8553.844s
  current training loss:                0.458096
  current training accuracy:            0.847212
Seen 324000 samples, time used: 8632.519s
  current training loss:                0.458138
  current training accuracy:            0.847142
Seen 327000 samples, time used: 8712.202s
  current training loss:                0.458104
  current training accuracy:            0.847190
Seen 330000 samples, time used: 8793.865s
  current training loss:                0.458102
  current training accuracy:            0.847173
saving to ..., time used 794.638s
Seen 333000 samples, time used: 8873.450s
  current training loss:                0.458118
  current training accuracy:            0.847093
Seen 336000 samples, time used: 8951.796s
  current training loss:                0.458032
  current training accuracy:            0.847119
Seen 339000 samples, time used: 9032.235s
  current training loss:                0.458029
  current training accuracy:            0.847171
Seen 342000 samples, time used: 9113.827s
  current training loss:                0.457903
  current training accuracy:            0.847213
Seen 345000 samples, time used: 9192.136s
  current training loss:                0.458012
  current training accuracy:            0.847200
Seen 348000 samples, time used: 9271.556s
  current training loss:                0.457964
  current training accuracy:            0.847261
Seen 351000 samples, time used: 9349.870s
  current training loss:                0.457851
  current training accuracy:            0.847291
Seen 354000 samples, time used: 9428.876s
  current training loss:                0.458063
  current training accuracy:            0.847184
Seen 357000 samples, time used: 9509.605s
  current training loss:                0.457954
  current training accuracy:            0.847171
Seen 360000 samples, time used: 9588.373s
  current training loss:                0.457994
  current training accuracy:            0.847175
saving to ..., time used 794.029s
Seen 363000 samples, time used: 9669.493s
  current training loss:                0.458003
  current training accuracy:            0.847160
Seen 366000 samples, time used: 9749.075s
  current training loss:                0.458112
  current training accuracy:            0.847145
Seen 369000 samples, time used: 9828.680s
  current training loss:                0.458146
  current training accuracy:            0.847117
Seen 372000 samples, time used: 9908.359s
  current training loss:                0.458220
  current training accuracy:            0.847105
Seen 375000 samples, time used: 9988.790s
  current training loss:                0.458074
  current training accuracy:            0.847195
Seen 378000 samples, time used: 10070.415s
  current training loss:                0.458086
  current training accuracy:            0.847196
Seen 381000 samples, time used: 10151.037s
  current training loss:                0.458180
  current training accuracy:            0.847223
Seen 384000 samples, time used: 10231.223s
  current training loss:                0.458254
  current training accuracy:            0.847164
Seen 387000 samples, time used: 10312.054s
  current training loss:                0.458318
  current training accuracy:            0.847129
Seen 390000 samples, time used: 10392.404s
  current training loss:                0.458321
  current training accuracy:            0.847141
saving to ..., time used 803.554s
Seen 393000 samples, time used: 10471.574s
  current training loss:                0.458375
  current training accuracy:            0.847097
Seen 396000 samples, time used: 10552.077s
  current training loss:                0.458314
  current training accuracy:            0.847167
Seen 399000 samples, time used: 10632.572s
  current training loss:                0.458535
  current training accuracy:            0.847020
Seen 402000 samples, time used: 10713.115s
  current training loss:                0.458626
  current training accuracy:            0.846978
Seen 405000 samples, time used: 10793.498s
  current training loss:                0.458610
  current training accuracy:            0.846943
Seen 408000 samples, time used: 10873.640s
  current training loss:                0.458771
  current training accuracy:            0.846873
Seen 411000 samples, time used: 10953.775s
  current training loss:                0.458627
  current training accuracy:            0.846915
Seen 414000 samples, time used: 11033.321s
  current training loss:                0.458642
  current training accuracy:            0.846901
Seen 417000 samples, time used: 11111.714s
  current training loss:                0.458877
  current training accuracy:            0.846731
Seen 420000 samples, time used: 11192.201s
  current training loss:                0.459125
  current training accuracy:            0.846655
saving to ..., time used 799.284s
Seen 423000 samples, time used: 11273.409s
  current training loss:                0.459216
  current training accuracy:            0.846577
Seen 426000 samples, time used: 11354.829s
  current training loss:                0.459123
  current training accuracy:            0.846585
Seen 429000 samples, time used: 11435.803s
  current training loss:                0.459150
  current training accuracy:            0.846543
Seen 432000 samples, time used: 11517.523s
  current training loss:                0.459306
  current training accuracy:            0.846468
Seen 435000 samples, time used: 11599.482s
  current training loss:                0.459455
  current training accuracy:            0.846379
Seen 438000 samples, time used: 11679.992s
  current training loss:                0.459569
  current training accuracy:            0.846331
Seen 441000 samples, time used: 11762.868s
  current training loss:                0.459472
  current training accuracy:            0.846363
Seen 444000 samples, time used: 11843.740s
  current training loss:                0.459430
  current training accuracy:            0.846374
Seen 447000 samples, time used: 11926.104s
  current training loss:                0.459484
  current training accuracy:            0.846302
Seen 450000 samples, time used: 12008.546s
  current training loss:                0.459527
  current training accuracy:            0.846256
saving to ..., time used 815.861s
Seen 453000 samples, time used: 12088.694s
  current training loss:                0.459500
  current training accuracy:            0.846269
Seen 456000 samples, time used: 12170.794s
  current training loss:                0.459557
  current training accuracy:            0.846268
Seen 459000 samples, time used: 12250.855s
  current training loss:                0.459574
  current training accuracy:            0.846281
Seen 462000 samples, time used: 12332.852s
  current training loss:                0.459580
  current training accuracy:            0.846271
Seen 465000 samples, time used: 12414.214s
  current training loss:                0.459734
  current training accuracy:            0.846213
Seen 468000 samples, time used: 12495.537s
  current training loss:                0.459774
  current training accuracy:            0.846212
Seen 471000 samples, time used: 12577.357s
  current training loss:                0.459805
  current training accuracy:            0.846212
Seen 474000 samples, time used: 12659.076s
  current training loss:                0.459864
  current training accuracy:            0.846219
Seen 477000 samples, time used: 12740.404s
  current training loss:                0.460063
  current training accuracy:            0.846134
Seen 480000 samples, time used: 12819.956s
  current training loss:                0.460117
  current training accuracy:            0.846156
saving to ..., time used 810.913s
Seen 483000 samples, time used: 12899.992s
  current training loss:                0.460024
  current training accuracy:            0.846170
Seen 486000 samples, time used: 12977.872s
  current training loss:                0.459985
  current training accuracy:            0.846103
Seen 489000 samples, time used: 13057.239s
  current training loss:                0.459898
  current training accuracy:            0.846147
Seen 492000 samples, time used: 13136.024s
  current training loss:                0.460024
  current training accuracy:            0.846075
Seen 495000 samples, time used: 13214.250s
  current training loss:                0.460164
  current training accuracy:            0.846044
Seen 498000 samples, time used: 13293.984s
  current training loss:                0.460292
  current training accuracy:            0.845978
Seen 501000 samples, time used: 13373.266s
  current training loss:                0.460382
  current training accuracy:            0.845944
Seen 504000 samples, time used: 13451.214s
  current training loss:                0.460481
  current training accuracy:            0.845853
Seen 507000 samples, time used: 13529.654s
  current training loss:                0.460774
  current training accuracy:            0.845730
Seen 510000 samples, time used: 13607.348s
  current training loss:                0.460818
  current training accuracy:            0.845751
saving to ..., time used 786.853s
Seen 513000 samples, time used: 13688.411s
  current training loss:                0.460878
  current training accuracy:            0.845706
Seen 516000 samples, time used: 13768.436s
  current training loss:                0.460941
  current training accuracy:            0.845653
Seen 519000 samples, time used: 13847.648s
  current training loss:                0.460979
  current training accuracy:            0.845618
Seen 522000 samples, time used: 13927.995s
  current training loss:                0.461012
  current training accuracy:            0.845611
Seen 525000 samples, time used: 14008.546s
  current training loss:                0.461037
  current training accuracy:            0.845575
Seen 528000 samples, time used: 14089.561s
  current training loss:                0.461099
  current training accuracy:            0.845549
Seen 531000 samples, time used: 14169.826s
  current training loss:                0.461175
  current training accuracy:            0.845501
Seen 534000 samples, time used: 14248.865s
  current training loss:                0.461231
  current training accuracy:            0.845483
Seen 537000 samples, time used: 14330.190s
  current training loss:                0.461338
  current training accuracy:            0.845395
Seen 540000 samples, time used: 14411.442s
  current training loss:                0.461414
  current training accuracy:            0.845372
saving to ..., time used 803.583s
Seen 543000 samples, time used: 14493.798s
  current training loss:                0.461419
  current training accuracy:            0.845343
Seen 546000 samples, time used: 14573.560s
  current training loss:                0.461437
  current training accuracy:            0.845322
Seen 549000 samples, time used: 14654.704s
  current training loss:                0.461448
  current training accuracy:            0.845313
Epoch 19 of 20 took 14697.557s
  training loss:                0.461448
  training accuracy:            84.53 %
  validation loss:              0.449945
  validation accuracy:          82.52 %
  test loss:                    0.463060
  test accuracy:                82.31 %
saving to ./snli/word_by_word_model_epoch19.npz
Seen 3000 samples, time used: 78.902s
  current training loss:                0.474595
  current training accuracy:            0.844000
Seen 6000 samples, time used: 160.322s
  current training loss:                0.448436
  current training accuracy:            0.851833
Seen 9000 samples, time used: 240.645s
  current training loss:                0.442758
  current training accuracy:            0.853778
Seen 12000 samples, time used: 321.754s
  current training loss:                0.442343
  current training accuracy:            0.853167
Seen 15000 samples, time used: 403.486s
  current training loss:                0.444769
  current training accuracy:            0.851600
Seen 18000 samples, time used: 484.316s
  current training loss:                0.445861
  current training accuracy:            0.852167
Seen 21000 samples, time used: 564.456s
  current training loss:                0.446212
  current training accuracy:            0.852190
Seen 24000 samples, time used: 644.817s
  current training loss:                0.446792
  current training accuracy:            0.852125
Seen 27000 samples, time used: 726.596s
  current training loss:                0.449307
  current training accuracy:            0.852037
Seen 30000 samples, time used: 806.983s
  current training loss:                0.448525
  current training accuracy:            0.852200
saving to ..., time used 806.983s
Seen 33000 samples, time used: 887.567s
  current training loss:                0.450175
  current training accuracy:            0.851515
Seen 36000 samples, time used: 967.648s
  current training loss:                0.450685
  current training accuracy:            0.851167
Seen 39000 samples, time used: 1045.614s
  current training loss:                0.451641
  current training accuracy:            0.850359
Seen 42000 samples, time used: 1124.837s
  current training loss:                0.450421
  current training accuracy:            0.850643
Seen 45000 samples, time used: 1204.729s
  current training loss:                0.451413
  current training accuracy:            0.850067
Seen 48000 samples, time used: 1283.162s
  current training loss:                0.451061
  current training accuracy:            0.850396
Seen 51000 samples, time used: 1362.159s
  current training loss:                0.451728
  current training accuracy:            0.849765
Seen 54000 samples, time used: 1440.929s
  current training loss:                0.450960
  current training accuracy:            0.849944
Seen 57000 samples, time used: 1519.467s
  current training loss:                0.451660
  current training accuracy:            0.849579
Seen 60000 samples, time used: 1597.288s
  current training loss:                0.452446
  current training accuracy:            0.849350
saving to ..., time used 789.831s
Seen 63000 samples, time used: 1676.628s
  current training loss:                0.452512
  current training accuracy:            0.848889
Seen 66000 samples, time used: 1755.733s
  current training loss:                0.452228
  current training accuracy:            0.848803
Seen 69000 samples, time used: 1834.959s
  current training loss:                0.452671
  current training accuracy:            0.849029
Seen 72000 samples, time used: 1914.276s
  current training loss:                0.452462
  current training accuracy:            0.848944
Seen 75000 samples, time used: 1993.804s
  current training loss:                0.452508
  current training accuracy:            0.848867
Seen 78000 samples, time used: 2072.854s
  current training loss:                0.451952
  current training accuracy:            0.848731
Seen 81000 samples, time used: 2152.231s
  current training loss:                0.452106
  current training accuracy:            0.848963
Seen 84000 samples, time used: 2230.556s
  current training loss:                0.453076
  current training accuracy:            0.848583
Seen 87000 samples, time used: 2309.720s
  current training loss:                0.452374
  current training accuracy:            0.848828
Seen 90000 samples, time used: 2388.163s
  current training loss:                0.452204
  current training accuracy:            0.849044
saving to ..., time used 790.386s
Seen 93000 samples, time used: 2466.497s
  current training loss:                0.452256
  current training accuracy:            0.849075
Seen 96000 samples, time used: 2545.539s
  current training loss:                0.452733
  current training accuracy:            0.848979
Seen 99000 samples, time used: 2623.423s
  current training loss:                0.452157
  current training accuracy:            0.849212
Seen 102000 samples, time used: 2703.290s
  current training loss:                0.452616
  current training accuracy:            0.849284
Seen 105000 samples, time used: 2782.735s
  current training loss:                0.453207
  current training accuracy:            0.849286
Seen 108000 samples, time used: 2861.081s
  current training loss:                0.453264
  current training accuracy:            0.849241
Seen 111000 samples, time used: 2940.300s
  current training loss:                0.454077
  current training accuracy:            0.848910
Seen 114000 samples, time used: 3020.144s
  current training loss:                0.453438
  current training accuracy:            0.849114
Seen 117000 samples, time used: 3100.014s
  current training loss:                0.453666
  current training accuracy:            0.849017
Seen 120000 samples, time used: 3178.953s
  current training loss:                0.454097
  current training accuracy:            0.848708
saving to ..., time used 790.187s
Seen 123000 samples, time used: 3258.393s
  current training loss:                0.453644
  current training accuracy:            0.848894
Seen 126000 samples, time used: 3338.459s
  current training loss:                0.453758
  current training accuracy:            0.848984
Seen 129000 samples, time used: 3416.615s
  current training loss:                0.453496
  current training accuracy:            0.849039
Seen 132000 samples, time used: 3494.992s
  current training loss:                0.453655
  current training accuracy:            0.849114
Seen 135000 samples, time used: 3574.571s
  current training loss:                0.453637
  current training accuracy:            0.849222
Seen 138000 samples, time used: 3653.968s
  current training loss:                0.453932
  current training accuracy:            0.849036
Seen 141000 samples, time used: 3733.808s
  current training loss:                0.454404
  current training accuracy:            0.848837
Seen 144000 samples, time used: 3813.210s
  current training loss:                0.454142
  current training accuracy:            0.848944
Seen 147000 samples, time used: 3892.412s
  current training loss:                0.454671
  current training accuracy:            0.848830
Seen 150000 samples, time used: 3972.374s
  current training loss:                0.454529
  current training accuracy:            0.848873
saving to ..., time used 792.876s
Seen 153000 samples, time used: 4052.711s
  current training loss:                0.454231
  current training accuracy:            0.849033
Seen 156000 samples, time used: 4132.300s
  current training loss:                0.453694
  current training accuracy:            0.849256
Seen 159000 samples, time used: 4211.673s
  current training loss:                0.454122
  current training accuracy:            0.849050
Seen 162000 samples, time used: 4290.393s
  current training loss:                0.454326
  current training accuracy:            0.848951
Seen 165000 samples, time used: 4370.147s
  current training loss:                0.454812
  current training accuracy:            0.848703
Seen 168000 samples, time used: 4449.237s
  current training loss:                0.455135
  current training accuracy:            0.848500
Seen 171000 samples, time used: 4527.954s
  current training loss:                0.455157
  current training accuracy:            0.848257
Seen 174000 samples, time used: 4605.877s
  current training loss:                0.455517
  current training accuracy:            0.848115
Seen 177000 samples, time used: 4684.759s
  current training loss:                0.455604
  current training accuracy:            0.848141
Seen 180000 samples, time used: 4763.246s
  current training loss:                0.455636
  current training accuracy:            0.848294
saving to ..., time used 790.334s
Seen 183000 samples, time used: 4843.549s
  current training loss:                0.455538
  current training accuracy:            0.848279
Seen 186000 samples, time used: 4922.903s
  current training loss:                0.455618
  current training accuracy:            0.848269
Seen 189000 samples, time used: 5002.537s
  current training loss:                0.455700
  current training accuracy:            0.848175
Seen 192000 samples, time used: 5083.142s
  current training loss:                0.455954
  current training accuracy:            0.848052
Seen 195000 samples, time used: 5160.417s
  current training loss:                0.455884
  current training accuracy:            0.848062
Seen 198000 samples, time used: 5239.257s
  current training loss:                0.456065
  current training accuracy:            0.848050
Seen 201000 samples, time used: 5318.966s
  current training loss:                0.455917
  current training accuracy:            0.848134
Seen 204000 samples, time used: 5397.527s
  current training loss:                0.456015
  current training accuracy:            0.847995
Seen 207000 samples, time used: 5474.725s
  current training loss:                0.456257
  current training accuracy:            0.847894
Seen 210000 samples, time used: 5552.517s
  current training loss:                0.456441
  current training accuracy:            0.847843
saving to ..., time used 788.774s
Seen 213000 samples, time used: 5632.735s
  current training loss:                0.456396
  current training accuracy:            0.847887
Seen 216000 samples, time used: 5713.016s
  current training loss:                0.456507
  current training accuracy:            0.847833
Seen 219000 samples, time used: 5792.616s
  current training loss:                0.456540
  current training accuracy:            0.847813
Seen 222000 samples, time used: 5871.814s
  current training loss:                0.456681
  current training accuracy:            0.847811
Seen 225000 samples, time used: 5952.530s
  current training loss:                0.456478
  current training accuracy:            0.848004
Seen 228000 samples, time used: 6031.563s
  current training loss:                0.456603
  current training accuracy:            0.847961
Seen 231000 samples, time used: 6111.672s
  current training loss:                0.456893
  current training accuracy:            0.847797
Seen 234000 samples, time used: 6190.585s
  current training loss:                0.456953
  current training accuracy:            0.847714
Seen 237000 samples, time used: 6269.424s
  current training loss:                0.456727
  current training accuracy:            0.847810
Seen 240000 samples, time used: 6349.930s
  current training loss:                0.456960
  current training accuracy:            0.847754
saving to ..., time used 796.896s
Seen 243000 samples, time used: 6431.048s
  current training loss:                0.456962
  current training accuracy:            0.847757
Seen 246000 samples, time used: 6511.626s
  current training loss:                0.457228
  current training accuracy:            0.847618
Seen 249000 samples, time used: 6592.634s
  current training loss:                0.457032
  current training accuracy:            0.847647
Seen 252000 samples, time used: 6673.585s
  current training loss:                0.457075
  current training accuracy:            0.847560
Seen 255000 samples, time used: 6753.928s
  current training loss:                0.456826
  current training accuracy:            0.847663
Seen 258000 samples, time used: 6834.126s
  current training loss:                0.456994
  current training accuracy:            0.847570
Seen 261000 samples, time used: 6913.420s
  current training loss:                0.456851
  current training accuracy:            0.847640
Seen 264000 samples, time used: 6994.455s
  current training loss:                0.456847
  current training accuracy:            0.847625
Seen 267000 samples, time used: 7076.712s
  current training loss:                0.456861
  current training accuracy:            0.847689
Seen 270000 samples, time used: 7157.464s
  current training loss:                0.456977
  current training accuracy:            0.847626
saving to ..., time used 806.998s
Seen 273000 samples, time used: 7237.995s
  current training loss:                0.456865
  current training accuracy:            0.847696
Seen 276000 samples, time used: 7317.936s
  current training loss:                0.457170
  current training accuracy:            0.847591
Seen 279000 samples, time used: 7395.605s
  current training loss:                0.457116
  current training accuracy:            0.847595
Seen 282000 samples, time used: 7473.306s
  current training loss:                0.457115
  current training accuracy:            0.847660
Seen 285000 samples, time used: 7552.785s
  current training loss:                0.457405
  current training accuracy:            0.847565
Seen 288000 samples, time used: 7632.667s
  current training loss:                0.457484
  current training accuracy:            0.847580
Seen 291000 samples, time used: 7712.440s
  current training loss:                0.457757
  current training accuracy:            0.847498
Seen 294000 samples, time used: 7790.949s
  current training loss:                0.457791
  current training accuracy:            0.847527
Seen 297000 samples, time used: 7872.631s
  current training loss:                0.457507
  current training accuracy:            0.847646
Seen 300000 samples, time used: 7951.262s
  current training loss:                0.457455
  current training accuracy:            0.847687
saving to ..., time used 793.287s
Seen 303000 samples, time used: 8033.212s
  current training loss:                0.457522
  current training accuracy:            0.847607
Seen 306000 samples, time used: 8116.119s
  current training loss:                0.457613
  current training accuracy:            0.847595
Seen 309000 samples, time used: 8196.846s
  current training loss:                0.457527
  current training accuracy:            0.847537
Seen 312000 samples, time used: 8277.428s
  current training loss:                0.457395
  current training accuracy:            0.847651
Seen 315000 samples, time used: 8359.110s
  current training loss:                0.457383
  current training accuracy:            0.847619
Seen 318000 samples, time used: 8438.561s
  current training loss:                0.457333
  current training accuracy:            0.847626
Seen 321000 samples, time used: 8520.351s
  current training loss:                0.457313
  current training accuracy:            0.847595
Seen 324000 samples, time used: 8601.644s
  current training loss:                0.457201
  current training accuracy:            0.847713
Seen 327000 samples, time used: 8681.574s
  current training loss:                0.457023
  current training accuracy:            0.847789
Seen 330000 samples, time used: 8761.508s
  current training loss:                0.457068
  current training accuracy:            0.847764
saving to ..., time used 809.753s
Seen 333000 samples, time used: 8842.703s
  current training loss:                0.457155
  current training accuracy:            0.847733
Seen 336000 samples, time used: 8923.603s
  current training loss:                0.457399
  current training accuracy:            0.847619
Seen 339000 samples, time used: 9004.058s
  current training loss:                0.457475
  current training accuracy:            0.847555
Seen 342000 samples, time used: 9084.515s
  current training loss:                0.457434
  current training accuracy:            0.847617
Seen 345000 samples, time used: 9164.336s
  current training loss:                0.457301
  current training accuracy:            0.847612
Seen 348000 samples, time used: 9243.355s
  current training loss:                0.457359
  current training accuracy:            0.847575
Seen 351000 samples, time used: 9323.801s
  current training loss:                0.457396
  current training accuracy:            0.847644
Seen 354000 samples, time used: 9404.230s
  current training loss:                0.457508
  current training accuracy:            0.847537
Seen 357000 samples, time used: 9483.777s
  current training loss:                0.457499
  current training accuracy:            0.847538
Seen 360000 samples, time used: 9563.255s
  current training loss:                0.457565
  current training accuracy:            0.847508
saving to ..., time used 801.245s
Seen 363000 samples, time used: 9643.222s
  current training loss:                0.457504
  current training accuracy:            0.847427
Seen 366000 samples, time used: 9725.126s
  current training loss:                0.457563
  current training accuracy:            0.847429
Seen 369000 samples, time used: 9806.164s
  current training loss:                0.457474
  current training accuracy:            0.847488
Seen 372000 samples, time used: 9887.128s
  current training loss:                0.457361
  current training accuracy:            0.847538
Seen 375000 samples, time used: 9969.330s
  current training loss:                0.457469
  current training accuracy:            0.847472
Seen 378000 samples, time used: 10051.060s
  current training loss:                0.457534
  current training accuracy:            0.847460
Seen 381000 samples, time used: 10134.863s
  current training loss:                0.457496
  current training accuracy:            0.847467
Seen 384000 samples, time used: 10216.224s
  current training loss:                0.457530
  current training accuracy:            0.847432
Seen 387000 samples, time used: 10297.326s
  current training loss:                0.457561
  current training accuracy:            0.847385
Seen 390000 samples, time used: 10377.819s
  current training loss:                0.457584
  current training accuracy:            0.847400
saving to ..., time used 814.081s
Seen 393000 samples, time used: 10459.171s
  current training loss:                0.457558
  current training accuracy:            0.847422
Seen 396000 samples, time used: 10540.483s
  current training loss:                0.457477
  current training accuracy:            0.847439
Seen 399000 samples, time used: 10622.108s
  current training loss:                0.457631
  current training accuracy:            0.847439
Seen 402000 samples, time used: 10702.908s
  current training loss:                0.457542
  current training accuracy:            0.847423
Seen 405000 samples, time used: 10782.928s
  current training loss:                0.457740
  current training accuracy:            0.847363
Seen 408000 samples, time used: 10865.097s
  current training loss:                0.457805
  current training accuracy:            0.847297
Seen 411000 samples, time used: 10945.400s
  current training loss:                0.457973
  current training accuracy:            0.847219
Seen 414000 samples, time used: 11026.087s
  current training loss:                0.457907
  current training accuracy:            0.847268
Seen 417000 samples, time used: 11105.070s
  current training loss:                0.457734
  current training accuracy:            0.847254
Seen 420000 samples, time used: 11182.904s
  current training loss:                0.457696
  current training accuracy:            0.847271
saving to ..., time used 804.577s
Seen 423000 samples, time used: 11263.162s
  current training loss:                0.457777
  current training accuracy:            0.847229
Seen 426000 samples, time used: 11343.192s
  current training loss:                0.457764
  current training accuracy:            0.847235
Seen 429000 samples, time used: 11422.480s
  current training loss:                0.457890
  current training accuracy:            0.847068
Seen 432000 samples, time used: 11501.324s
  current training loss:                0.457823
  current training accuracy:            0.847069
Seen 435000 samples, time used: 11580.193s
  current training loss:                0.457916
  current training accuracy:            0.846966
Seen 438000 samples, time used: 11660.582s
  current training loss:                0.457966
  current training accuracy:            0.846918
Seen 441000 samples, time used: 11739.436s
  current training loss:                0.457917
  current training accuracy:            0.846952
Seen 444000 samples, time used: 11818.712s
  current training loss:                0.457893
  current training accuracy:            0.846926
Seen 447000 samples, time used: 11898.275s
  current training loss:                0.457924
  current training accuracy:            0.846962
Seen 450000 samples, time used: 11976.929s
  current training loss:                0.458055
  current training accuracy:            0.846964
saving to ..., time used 793.526s
Seen 453000 samples, time used: 12056.565s
  current training loss:                0.458208
  current training accuracy:            0.846921
Seen 456000 samples, time used: 12135.777s
  current training loss:                0.458440
  current training accuracy:            0.846836
Seen 459000 samples, time used: 12215.457s
  current training loss:                0.458537
  current training accuracy:            0.846797
Seen 462000 samples, time used: 12296.274s
  current training loss:                0.458466
  current training accuracy:            0.846853
Seen 465000 samples, time used: 12377.926s
  current training loss:                0.458452
  current training accuracy:            0.846903
Seen 468000 samples, time used: 12458.176s
  current training loss:                0.458487
  current training accuracy:            0.846816
Seen 471000 samples, time used: 12536.351s
  current training loss:                0.458565
  current training accuracy:            0.846815
Seen 474000 samples, time used: 12617.471s
  current training loss:                0.458476
  current training accuracy:            0.846835
Seen 477000 samples, time used: 12697.582s
  current training loss:                0.458632
  current training accuracy:            0.846786
Seen 480000 samples, time used: 12779.394s
  current training loss:                0.458629
  current training accuracy:            0.846785
saving to ..., time used 801.984s
Seen 483000 samples, time used: 12859.245s
  current training loss:                0.458718
  current training accuracy:            0.846745
Seen 486000 samples, time used: 12938.660s
  current training loss:                0.458675
  current training accuracy:            0.846765
Seen 489000 samples, time used: 13017.848s
  current training loss:                0.458654
  current training accuracy:            0.846802
Seen 492000 samples, time used: 13096.473s
  current training loss:                0.458556
  current training accuracy:            0.846856
Seen 495000 samples, time used: 13175.974s
  current training loss:                0.458571
  current training accuracy:            0.846832
Seen 498000 samples, time used: 13256.372s
  current training loss:                0.458700
  current training accuracy:            0.846745
Seen 501000 samples, time used: 13334.647s
  current training loss:                0.458743
  current training accuracy:            0.846752
Seen 504000 samples, time used: 13414.636s
  current training loss:                0.458728
  current training accuracy:            0.846788
Seen 507000 samples, time used: 13493.087s
  current training loss:                0.458633
  current training accuracy:            0.846785
Seen 510000 samples, time used: 13571.733s
  current training loss:                0.458642
  current training accuracy:            0.846780
saving to ..., time used 791.863s
Seen 513000 samples, time used: 13651.700s
  current training loss:                0.458595
  current training accuracy:            0.846797
Seen 516000 samples, time used: 13731.031s
  current training loss:                0.458605
  current training accuracy:            0.846791
Seen 519000 samples, time used: 13809.161s
  current training loss:                0.458580
  current training accuracy:            0.846827
Seen 522000 samples, time used: 13888.675s
  current training loss:                0.458676
  current training accuracy:            0.846826
Seen 525000 samples, time used: 13969.108s
  current training loss:                0.458885
  current training accuracy:            0.846709
Seen 528000 samples, time used: 14049.376s
  current training loss:                0.459064
  current training accuracy:            0.846616
Seen 531000 samples, time used: 14128.235s
  current training loss:                0.459072
  current training accuracy:            0.846597
Seen 534000 samples, time used: 14207.190s
  current training loss:                0.459206
  current training accuracy:            0.846522
Seen 537000 samples, time used: 14287.047s
  current training loss:                0.459190
  current training accuracy:            0.846529
Seen 540000 samples, time used: 14367.503s
  current training loss:                0.459237
  current training accuracy:            0.846481
saving to ..., time used 795.220s
Seen 543000 samples, time used: 14446.538s
  current training loss:                0.459174
  current training accuracy:            0.846479
Seen 546000 samples, time used: 14525.133s
  current training loss:                0.459198
  current training accuracy:            0.846416
Seen 549000 samples, time used: 14603.667s
  current training loss:                0.459283
  current training accuracy:            0.846355
Epoch 20 of 20 took 14636.017s
  training loss:                0.459243
  training accuracy:            84.64 %
  validation loss:              0.453823
  validation accuracy:          82.73 %
  test loss:                    0.467349
  test accuracy:                81.92 %
saving to ./snli/word_by_word_model_epoch20.npz