crossvalidate.sh log

(env_spreadingvectors_gpu) matthijs@devfair0144:~/src/spreadingvectors$ bash crossvalidate.sh
+ lambdas='0.001 0.002 0.005 0.01 0.02 0.05'
+ dout=24
+ db=deep1b
+ quant=zn_79
+ best_lambda=-1
+ best_perf=0.0000
+ for lambda in '$lambdas'
+ mkdir -p test_ckpt/0.001
+ python -u train.py --dout 24 --save_best_criterion zn_79,rank=10 --database deep1b --lambda_uniform 0.001 --checkpoint_dir test_ckpt/0.001
+ tee test_ckpt/0.001.stdout
Namespace(batch_size=64, checkpoint_dir='test_ckpt/0.001', database='deep1b', device='cuda', dint=1024, dout=24, epochs=160, init_name='', lambda_uniform=0.00
1, lr_schedule='0.1,0.1,0.05,0.01', momentum=0.9, num_learn=500000, quantizer_train='', rank_negative=50, rank_positive=10, save_best_criterion='zn_79,rank=10
', seed=1234, size_base=1000000, val_freq=10, validation_quantizers=['zn_3', 'zn_10', 'zn_79'])
load dataset deep1b
keeping 500000/357380000 training vectors
computing training ground truth
build network
Lr schedule [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1,
0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1
, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0
.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0
5, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01,
 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01]
  Forward pass
  Distances
  Train
epoch 0, times: [hn 4.04 s epoch 50.09 s val 0.00 s] lr = 0.100000 loss = -0.0022452 = 0.00103775 + lam * -3.28295, offending 17513
  Forward pass
  Distances
  Train
epoch 1, times: [hn 3.99 s epoch 48.94 s val 0.00 s] lr = 0.100000 loss = -0.000939117 = 0.00147518 + lam * -2.41429, offending 54846
  Forward pass
  Distances
  Train
epoch 2, times: [hn 4.03 s epoch 44.67 s val 0.00 s] lr = 0.100000 loss = -0.00117097 = 0.00112393 + lam * -2.2949, offending 57558
  Forward pass
  Distances
  Train
epoch 3, times: [hn 3.97 s epoch 44.40 s val 0.00 s] lr = 0.100000 loss = -0.00104841 = 0.00117305 + lam * -2.22146, offending 68163
  Forward pass
  Distances
  Train
epoch 4, times: [hn 4.03 s epoch 42.09 s val 0.00 s] lr = 0.100000 loss = -0.0011024 = 0.00108996 + lam * -2.19237, offending 69710
  Forward pass
  Distances
  Train
epoch 5, times: [hn 3.96 s epoch 43.40 s val 0.00 s] lr = 0.100000 loss = -0.0010927 = 0.00109198 + lam * -2.18467, offending 74032
  Forward pass
  Distances
  Train
epoch 6, times: [hn 4.03 s epoch 42.71 s val 0.00 s] lr = 0.100000 loss = -0.00109144 = 0.00108422 + lam * -2.17566, offending 76919
  Forward pass
  Distances
  Train
epoch 7, times: [hn 4.04 s epoch 42.12 s val 0.00 s] lr = 0.100000 loss = -0.00112373 = 0.00105503 + lam * -2.17876, offending 77216
  Forward pass
  Distances
  Train
epoch 8, times: [hn 4.03 s epoch 43.55 s val 0.00 s] lr = 0.100000 loss = -0.00113067 = 0.00105519 + lam * -2.18586, offending 78370
  Forward pass
  Distances
  Train
Valiation at epoch 9
zn_3     nbit= 14:  0.0000 0.0001 0.0013
zn_10    nbit= 32:  0.0000 0.0004 0.0035
zn_79    nbit= 64:  0.0027 0.0180 0.0890
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.01797 > 0), keeping as best
epoch 9, times: [hn 4.04 s epoch 42.38 s val 10.66 s] lr = 0.100000 loss = -0.00118187 = 0.00102711 + lam * -2.20898, offending 76706
  Forward pass
  Distances
  Train
epoch 10, times: [hn 4.11 s epoch 42.44 s val 0.00 s] lr = 0.100000 loss = -0.00119132 = 0.0010369 + lam * -2.22823, offending 77115
  Forward pass
  Distances
  Train
epoch 11, times: [hn 4.05 s epoch 40.52 s val 0.00 s] lr = 0.100000 loss = -0.00121622 = 0.00102839 + lam * -2.24461, offending 76411
  Forward pass
  Distances
  Train
epoch 12, times: [hn 4.03 s epoch 41.32 s val 0.00 s] lr = 0.100000 loss = -0.00126156 = 0.00100754 + lam * -2.2691, offending 74924
  Forward pass
  Distances
  Train
epoch 13, times: [hn 4.04 s epoch 40.89 s val 0.00 s] lr = 0.100000 loss = -0.00127385 = 0.00101914 + lam * -2.293, offending 74374
  Forward pass
  Distances
  Train
epoch 14, times: [hn 4.02 s epoch 40.30 s val 0.00 s] lr = 0.100000 loss = -0.00129166 = 0.00101886 + lam * -2.31052, offending 73626
  Forward pass
  Distances
  Train
epoch 15, times: [hn 4.03 s epoch 38.85 s val 0.00 s] lr = 0.100000 loss = -0.00132132 = 0.00100816 + lam * -2.32948, offending 71916
  Forward pass
  Distances
  Train
epoch 16, times: [hn 4.03 s epoch 40.09 s val 0.00 s] lr = 0.100000 loss = -0.00135717 = 0.00099723 + lam * -2.3544, offending 70597
  Forward pass
  Distances
  Train
epoch 17, times: [hn 4.07 s epoch 40.71 s val 0.00 s] lr = 0.100000 loss = -0.00136296 = 0.00100978 + lam * -2.37274, offending 70295
  Forward pass
  Distances
  Train
epoch 18, times: [hn 4.07 s epoch 38.23 s val 0.00 s] lr = 0.100000 loss = -0.00137488 = 0.00101133 + lam * -2.38621, offending 69703
  Forward pass
  Distances
  Train
Valiation at epoch 19
zn_3     nbit= 14:  0.0000 0.0001 0.0013
zn_10    nbit= 32:  0.0001 0.0009 0.0068
zn_79    nbit= 64:  0.0052 0.0325 0.1418
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.03248 > 0.01797), keeping as best
epoch 19, times: [hn 4.02 s epoch 41.13 s val 10.29 s] lr = 0.100000 loss = -0.00139869 = 0.00100534 + lam * -2.40403, offending 68600
  Forward pass
  Distances
  Train
epoch 20, times: [hn 4.04 s epoch 39.95 s val 0.00 s] lr = 0.100000 loss = -0.00145269 = 0.000979535 + lam * -2.43223, offending 65970
  Forward pass
  Distances
  Train
epoch 21, times: [hn 3.97 s epoch 40.48 s val 0.00 s] lr = 0.100000 loss = -0.00144452 = 0.00100549 + lam * -2.45001, offending 66852
  Forward pass
  Distances
  Train
epoch 22, times: [hn 4.07 s epoch 40.62 s val 0.00 s] lr = 0.100000 loss = -0.00149087 = 0.000982861 + lam * -2.47373, offending 64090
  Forward pass
  Distances
  Train
epoch 23, times: [hn 4.03 s epoch 40.12 s val 0.00 s] lr = 0.100000 loss = -0.00148524 = 0.00101016 + lam * -2.4954, offending 64662
  Forward pass
  Distances
  Train
epoch 24, times: [hn 4.15 s epoch 40.25 s val 0.00 s] lr = 0.100000 loss = -0.00150118 = 0.000997417 + lam * -2.4986, offending 64105
  Forward pass
  Distances
  Train
epoch 25, times: [hn 4.04 s epoch 40.10 s val 0.00 s] lr = 0.100000 loss = -0.00151936 = 0.00099605 + lam * -2.51541, offending 63356
  Forward pass
  Distances
  Train
epoch 26, times: [hn 3.98 s epoch 40.02 s val 0.00 s] lr = 0.100000 loss = -0.00156059 = 0.00097937 + lam * -2.53996, offending 61256
  Forward pass
  Distances
  Train
epoch 27, times: [hn 4.03 s epoch 39.97 s val 0.00 s] lr = 0.100000 loss = -0.00156247 = 0.000997641 + lam * -2.56011, offending 61199
  Forward pass
  Distances
  Train
epoch 28, times: [hn 4.05 s epoch 40.50 s val 0.00 s] lr = 0.100000 loss = -0.00157917 = 0.000993955 + lam * -2.57313, offending 60406
  Forward pass
  Distances
  Train
Valiation at epoch 29
zn_3     nbit= 14:  0.0001 0.0002 0.0017
zn_10    nbit= 32:  0.0003 0.0019 0.0135
zn_79    nbit= 64:  0.0092 0.0530 0.2075
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.05303 > 0.03248), keeping as best
epoch 29, times: [hn 4.04 s epoch 40.17 s val 10.29 s] lr = 0.100000 loss = -0.00159307 = 0.00099494 + lam * -2.58801, offending 59770
  Forward pass
  Distances
  Train
epoch 30, times: [hn 4.13 s epoch 41.22 s val 0.00 s] lr = 0.100000 loss = -0.00160069 = 0.00100167 + lam * -2.60236, offending 59115
  Forward pass
  Distances
  Train
epoch 31, times: [hn 4.06 s epoch 40.95 s val 0.00 s] lr = 0.100000 loss = -0.00163313 = 0.000984158 + lam * -2.61728, offending 58243
  Forward pass
  Distances
  Train
epoch 32, times: [hn 4.07 s epoch 40.78 s val 0.00 s] lr = 0.100000 loss = -0.00165045 = 0.000989357 + lam * -2.6398, offending 57203
  Forward pass
  Distances
  Train
epoch 33, times: [hn 4.04 s epoch 40.11 s val 0.00 s] lr = 0.100000 loss = -0.00167428 = 0.000984435 + lam * -2.65871, offending 56095
  Forward pass
  Distances
  Train
epoch 34, times: [hn 4.07 s epoch 40.11 s val 0.00 s] lr = 0.100000 loss = -0.00167361 = 0.000995335 + lam * -2.66894, offending 56174
  Forward pass
  Distances
  Train
epoch 35, times: [hn 3.99 s epoch 40.71 s val 0.00 s] lr = 0.100000 loss = -0.0017393 = 0.000951308 + lam * -2.69061, offending 53126
  Forward pass
  Distances
  Train
epoch 36, times: [hn 3.97 s epoch 39.90 s val 0.00 s] lr = 0.100000 loss = -0.00174363 = 0.000977338 + lam * -2.72096, offending 53057
  Forward pass
  Distances
  Train
epoch 37, times: [hn 4.03 s epoch 40.97 s val 0.00 s] lr = 0.100000 loss = -0.00173014 = 0.00100134 + lam * -2.73148, offending 53825
  Forward pass
  Distances
  Train
epoch 38, times: [hn 4.03 s epoch 40.43 s val 0.00 s] lr = 0.100000 loss = -0.00175726 = 0.000983278 + lam * -2.74054, offending 52515
  Forward pass
  Distances
  Train
Valiation at epoch 39
zn_3     nbit= 14:  0.0001 0.0003 0.0026
zn_10    nbit= 32:  0.0004 0.0037 0.0236
zn_79    nbit= 64:  0.0180 0.0911 0.3214
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.09112 > 0.05303), keeping as best
epoch 39, times: [hn 4.04 s epoch 40.73 s val 9.59 s] lr = 0.100000 loss = -0.00180507 = 0.000958733 + lam * -2.7638, offending 50914
  Forward pass
  Distances
  Train
epoch 40, times: [hn 4.14 s epoch 40.59 s val 0.00 s] lr = 0.100000 loss = -0.00177662 = 0.000996862 + lam * -2.77349, offending 51898
  Forward pass
  Distances
  Train
epoch 41, times: [hn 4.03 s epoch 40.29 s val 0.00 s] lr = 0.100000 loss = -0.00178971 = 0.000992137 + lam * -2.78185, offending 51678
  Forward pass
  Distances
  Train
epoch 42, times: [hn 4.02 s epoch 40.16 s val 0.00 s] lr = 0.100000 loss = -0.00179674 = 0.000987956 + lam * -2.7847, offending 51017
  Forward pass
  Distances
  Train
epoch 43, times: [hn 4.06 s epoch 40.75 s val 0.00 s] lr = 0.100000 loss = -0.00181819 = 0.000979114 + lam * -2.79731, offending 50065
  Forward pass
  Distances
  Train
epoch 44, times: [hn 3.95 s epoch 40.06 s val 0.00 s] lr = 0.100000 loss = -0.00182568 = 0.000985554 + lam * -2.81123, offending 49767
  Forward pass
  Distances
  Train
epoch 45, times: [hn 3.96 s epoch 40.54 s val 0.00 s] lr = 0.100000 loss = -0.00184962 = 0.000972896 + lam * -2.82252, offending 49061
  Forward pass
  Distances
  Train
epoch 46, times: [hn 4.05 s epoch 41.07 s val 0.00 s] lr = 0.100000 loss = -0.00190659 = 0.000953258 + lam * -2.85985, offending 46606
  Forward pass
  Distances
  Train
epoch 47, times: [hn 4.04 s epoch 40.77 s val 0.00 s] lr = 0.100000 loss = -0.00189178 = 0.000978562 + lam * -2.87034, offending 47680
  Forward pass
  Distances
  Train
epoch 48, times: [hn 4.03 s epoch 40.50 s val 0.00 s] lr = 0.100000 loss = -0.00191639 = 0.000962501 + lam * -2.87889, offending 46422
  Forward pass
  Distances
  Train
Valiation at epoch 49
zn_3     nbit= 14:  0.0001 0.0004 0.0036
zn_10    nbit= 32:  0.0006 0.0051 0.0330
zn_79    nbit= 64:  0.0266 0.1292 0.4150
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.12921 > 0.09112), keeping as best
epoch 49, times: [hn 4.03 s epoch 41.37 s val 9.52 s] lr = 0.100000 loss = -0.00196612 = 0.00094136 + lam * -2.90748, offending 44407
  Forward pass
  Distances
  Train
epoch 50, times: [hn 4.04 s epoch 40.65 s val 0.00 s] lr = 0.100000 loss = -0.00194773 = 0.000968127 + lam * -2.91586, offending 45550
  Forward pass
  Distances
  Train
epoch 51, times: [hn 4.06 s epoch 40.88 s val 0.00 s] lr = 0.100000 loss = -0.00197967 = 0.000950008 + lam * -2.92968, offending 43976
  Forward pass
  Distances
  Train
epoch 52, times: [hn 3.98 s epoch 40.42 s val 0.00 s] lr = 0.100000 loss = -0.00200654 = 0.000944012 + lam * -2.95055, offending 42730
  Forward pass
  Distances
  Train
epoch 53, times: [hn 4.02 s epoch 40.68 s val 0.00 s] lr = 0.100000 loss = -0.00198378 = 0.000973331 + lam * -2.95711, offending 43658
  Forward pass
  Distances
  Train
epoch 54, times: [hn 4.03 s epoch 40.87 s val 0.00 s] lr = 0.100000 loss = -0.0020817 = 0.000902401 + lam * -2.9841, offending 40270
  Forward pass
  Distances
  Train
epoch 55, times: [hn 4.06 s epoch 41.18 s val 0.00 s] lr = 0.100000 loss = -0.00212205 = 0.000892648 + lam * -3.0147, offending 38660
  Forward pass
  Distances
  Train
epoch 56, times: [hn 4.14 s epoch 41.48 s val 0.00 s] lr = 0.100000 loss = -0.00211939 = 0.000914628 + lam * -3.03402, offending 38495
  Forward pass
  Distances
  Train
epoch 57, times: [hn 4.02 s epoch 40.73 s val 0.00 s] lr = 0.100000 loss = -0.0020322 = 0.000982638 + lam * -3.01484, offending 41524
  Forward pass
  Distances
  Train
epoch 58, times: [hn 3.95 s epoch 40.83 s val 0.00 s] lr = 0.100000 loss = -0.00205478 = 0.000945986 + lam * -3.00077, offending 40699
  Forward pass
  Distances
  Train
Valiation at epoch 59
zn_3     nbit= 14:  0.0001 0.0005 0.0044
zn_10    nbit= 32:  0.0009 0.0076 0.0451
zn_79    nbit= 64:  0.0336 0.1582 0.4774
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.15819 > 0.12921), keeping as best
epoch 59, times: [hn 4.03 s epoch 40.93 s val 9.36 s] lr = 0.100000 loss = -0.00206089 = 0.000946556 + lam * -3.00744, offending 40452
  Forward pass
  Distances
  Train
epoch 60, times: [hn 4.10 s epoch 40.68 s val 0.00 s] lr = 0.100000 loss = -0.00221433 = 0.000840529 + lam * -3.05486, offending 35331
  Forward pass
  Distances
  Train
epoch 61, times: [hn 4.05 s epoch 40.24 s val 0.00 s] lr = 0.100000 loss = -0.00225166 = 0.000849022 + lam * -3.10068, offending 34302
  Forward pass
  Distances
  Train
epoch 62, times: [hn 4.05 s epoch 40.67 s val 0.00 s] lr = 0.100000 loss = -0.00210414 = 0.000973092 + lam * -3.07723, offending 39179
  Forward pass
  Distances
  Train
epoch 63, times: [hn 3.96 s epoch 40.16 s val 0.00 s] lr = 0.100000 loss = -0.00219699 = 0.000886848 + lam * -3.08383, offending 35752
  Forward pass
  Distances
  Train
epoch 64, times: [hn 4.03 s epoch 40.35 s val 0.00 s] lr = 0.100000 loss = -0.00213349 = 0.000944226 + lam * -3.07772, offending 37898
  Forward pass
  Distances
  Train
epoch 65, times: [hn 4.08 s epoch 40.40 s val 0.00 s] lr = 0.100000 loss = -0.00225586 = 0.000844567 + lam * -3.10043, offending 33742
  Forward pass
  Distances
  Train
epoch 66, times: [hn 4.03 s epoch 40.37 s val 0.00 s] lr = 0.100000 loss = -0.00225279 = 0.000872923 + lam * -3.12571, offending 34327
  Forward pass
  Distances
  Train
epoch 67, times: [hn 4.06 s epoch 40.62 s val 0.00 s] lr = 0.100000 loss = -0.00222402 = 0.000903589 + lam * -3.12761, offending 34690
  Forward pass
  Distances
  Train
epoch 68, times: [hn 4.03 s epoch 40.29 s val 0.00 s] lr = 0.100000 loss = -0.00223894 = 0.000884721 + lam * -3.12366, offending 34213
  Forward pass
  Distances
  Train
Valiation at epoch 69
zn_3     nbit= 14:  0.0000 0.0006 0.0052
zn_10    nbit= 32:  0.0017 0.0125 0.0661
zn_79    nbit= 64:  0.0503 0.2229 0.5886
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.22289 > 0.15819), keeping as best
epoch 69, times: [hn 4.00 s epoch 40.14 s val 10.33 s] lr = 0.100000 loss = -0.00227663 = 0.000865836 + lam * -3.14246, offending 32496
  Forward pass
  Distances
  Train
epoch 70, times: [hn 4.14 s epoch 40.24 s val 0.00 s] lr = 0.100000 loss = -0.00228873 = 0.000864784 + lam * -3.15351, offending 32322
  Forward pass
  Distances
  Train
epoch 71, times: [hn 4.05 s epoch 40.48 s val 0.00 s] lr = 0.100000 loss = -0.00228147 = 0.000878784 + lam * -3.16025, offending 32770
  Forward pass
  Distances
  Train
epoch 72, times: [hn 4.05 s epoch 39.94 s val 0.00 s] lr = 0.100000 loss = -0.00212964 = 0.000982997 + lam * -3.11263, offending 37166
  Forward pass
  Distances
  Train
epoch 73, times: [hn 4.06 s epoch 40.61 s val 0.00 s] lr = 0.100000 loss = -0.00233764 = 0.000807382 + lam * -3.14502, offending 30680
  Forward pass
  Distances
  Train
epoch 74, times: [hn 4.00 s epoch 40.82 s val 0.00 s] lr = 0.100000 loss = -0.00219219 = 0.000944977 + lam * -3.13716, offending 35529
  Forward pass
  Distances
  Train
epoch 75, times: [hn 4.04 s epoch 41.37 s val 0.00 s] lr = 0.100000 loss = -0.0023247 = 0.000832232 + lam * -3.15693, offending 31073
  Forward pass
  Distances
  Train
epoch 76, times: [hn 4.04 s epoch 41.73 s val 0.00 s] lr = 0.100000 loss = -0.00244927 = 0.000761825 + lam * -3.2111, offending 27356
  Forward pass
  Distances
  Train
epoch 77, times: [hn 4.02 s epoch 39.79 s val 0.00 s] lr = 0.100000 loss = -0.00227742 = 0.000918311 + lam * -3.19573, offending 32513
  Forward pass
  Distances
  Train
epoch 78, times: [hn 4.01 s epoch 41.32 s val 0.00 s] lr = 0.100000 loss = -0.00215206 = 0.000981072 + lam * -3.13313, offending 35873
  Forward pass
  Distances
  Train
Valiation at epoch 79
zn_3     nbit= 14:  0.0001 0.0006 0.0056
zn_10    nbit= 32:  0.0018 0.0118 0.0680
zn_79    nbit= 64:  0.0542 0.2355 0.6103
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.2355 > 0.22289), keeping as best
epoch 79, times: [hn 4.03 s epoch 40.72 s val 9.52 s] lr = 0.100000 loss = -0.00231631 = 0.000834461 + lam * -3.15077, offending 31055
  Forward pass
  Distances
  Train
epoch 80, times: [hn 4.08 s epoch 40.15 s val 0.00 s] lr = 0.050000 loss = -0.00247238 = 0.000732644 + lam * -3.20502, offending 27270
  Forward pass
  Distances
  Train
epoch 81, times: [hn 4.03 s epoch 41.66 s val 0.00 s] lr = 0.050000 loss = -0.00253787 = 0.000711773 + lam * -3.24964, offending 25649
  Forward pass
  Distances
  Train
epoch 82, times: [hn 4.03 s epoch 42.01 s val 0.00 s] lr = 0.050000 loss = -0.00251448 = 0.000756036 + lam * -3.27052, offending 26632
  Forward pass
  Distances
  Train
epoch 83, times: [hn 4.06 s epoch 41.47 s val 0.00 s] lr = 0.050000 loss = -0.00261873 = 0.000683374 + lam * -3.30211, offending 23522
  Forward pass
  Distances
  Train
epoch 84, times: [hn 4.04 s epoch 41.36 s val 0.00 s] lr = 0.050000 loss = -0.00237799 = 0.000895089 + lam * -3.27308, offending 30364
  Forward pass
  Distances
  Train
epoch 85, times: [hn 4.05 s epoch 41.90 s val 0.00 s] lr = 0.050000 loss = -0.00247176 = 0.000792704 + lam * -3.26446, offending 27599
  Forward pass
  Distances
  Train
epoch 86, times: [hn 4.04 s epoch 40.98 s val 0.00 s] lr = 0.050000 loss = -0.0025632 = 0.000726702 + lam * -3.2899, offending 24670
  Forward pass
  Distances
  Train
epoch 87, times: [hn 4.11 s epoch 40.32 s val 0.00 s] lr = 0.050000 loss = -0.00240845 = 0.000865674 + lam * -3.27412, offending 28980
  Forward pass
  Distances
  Train
epoch 88, times: [hn 4.03 s epoch 40.49 s val 0.00 s] lr = 0.050000 loss = -0.00234015 = 0.000907456 + lam * -3.2476, offending 31136
  Forward pass
  Distances
  Train
Valiation at epoch 89
zn_3     nbit= 14:  0.0001 0.0009 0.0070
zn_10    nbit= 32:  0.0025 0.0186 0.0948
zn_79    nbit= 64:  0.0736 0.3030 0.7026
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.30296 > 0.2355), keeping as best
epoch 89, times: [hn 4.03 s epoch 40.98 s val 9.55 s] lr = 0.050000 loss = -0.002535 = 0.000735065 + lam * -3.27006, offending 25506
  Forward pass
  Distances
  Train
epoch 90, times: [hn 4.01 s epoch 40.34 s val 0.00 s] lr = 0.050000 loss = -0.00243746 = 0.000833402 + lam * -3.27086, offending 28582
  Forward pass
  Distances
  Train
epoch 91, times: [hn 4.05 s epoch 40.60 s val 0.00 s] lr = 0.050000 loss = -0.00252425 = 0.000761029 + lam * -3.28528, offending 25949
  Forward pass
  Distances
  Train
epoch 92, times: [hn 4.02 s epoch 41.43 s val 0.00 s] lr = 0.050000 loss = -0.0026152 = 0.000693008 + lam * -3.30821, offending 23238
  Forward pass
  Distances
  Train
epoch 93, times: [hn 4.04 s epoch 40.38 s val 0.00 s] lr = 0.050000 loss = -0.00258336 = 0.000746868 + lam * -3.33023, offending 24654
  Forward pass
  Distances
  Train
epoch 94, times: [hn 4.05 s epoch 40.22 s val 0.00 s] lr = 0.050000 loss = -0.00253238 = 0.000790239 + lam * -3.32262, offending 26039
  Forward pass
  Distances
  Train
epoch 95, times: [hn 4.06 s epoch 40.69 s val 0.00 s] lr = 0.050000 loss = -0.00240612 = 0.000885195 + lam * -3.29131, offending 29231
  Forward pass
  Distances
  Train
epoch 96, times: [hn 4.02 s epoch 40.95 s val 0.00 s] lr = 0.050000 loss = -0.00241104 = 0.000856201 + lam * -3.26724, offending 28770
  Forward pass
  Distances
  Train
epoch 97, times: [hn 4.04 s epoch 41.37 s val 0.00 s] lr = 0.050000 loss = -0.00246234 = 0.000810869 + lam * -3.27321, offending 27391
  Forward pass
  Distances
  Train
epoch 98, times: [hn 4.04 s epoch 40.17 s val 0.00 s] lr = 0.050000 loss = -0.00237155 = 0.00088615 + lam * -3.2577, offending 30144
  Forward pass
  Distances
  Train
Valiation at epoch 99
zn_3     nbit= 14:  0.0001 0.0011 0.0097
zn_10    nbit= 32:  0.0039 0.0263 0.1264
zn_79    nbit= 64:  0.0916 0.3548 0.7592
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.35478 > 0.30296), keeping as best
epoch 99, times: [hn 4.05 s epoch 40.53 s val 10.42 s] lr = 0.050000 loss = -0.00258073 = 0.000708626 + lam * -3.28935, offending 24291
  Forward pass
  Distances
  Train
epoch 100, times: [hn 4.08 s epoch 40.37 s val 0.00 s] lr = 0.050000 loss = -0.00240877 = 0.000880075 + lam * -3.28884, offending 28900
  Forward pass
  Distances
  Train
epoch 101, times: [hn 4.00 s epoch 40.70 s val 0.00 s] lr = 0.050000 loss = -0.00235 = 0.000902876 + lam * -3.25288, offending 30627
  Forward pass
  Distances
  Train
epoch 102, times: [hn 3.96 s epoch 40.87 s val 0.00 s] lr = 0.050000 loss = -0.002481 = 0.000788174 + lam * -3.26917, offending 26729
  Forward pass
  Distances
  Train
epoch 103, times: [hn 4.02 s epoch 42.46 s val 0.00 s] lr = 0.050000 loss = -0.00260607 = 0.000695254 + lam * -3.30133, offending 23527
  Forward pass
  Distances
  Train
epoch 104, times: [hn 4.04 s epoch 40.22 s val 0.00 s] lr = 0.050000 loss = -0.00240647 = 0.000887058 + lam * -3.29353, offending 28549
  Forward pass
  Distances
  Train
epoch 105, times: [hn 4.05 s epoch 39.58 s val 0.00 s] lr = 0.050000 loss = -0.00245199 = 0.000827567 + lam * -3.27956, offending 27636
  Forward pass
  Distances
  Train
epoch 106, times: [hn 4.01 s epoch 41.02 s val 0.00 s] lr = 0.050000 loss = -0.00261281 = 0.000702077 + lam * -3.31489, offending 23277
  Forward pass
  Distances
  Train
epoch 107, times: [hn 4.02 s epoch 40.06 s val 0.00 s] lr = 0.050000 loss = -0.00255683 = 0.000769398 + lam * -3.32622, offending 24921
  Forward pass
  Distances
  Train
epoch 108, times: [hn 4.03 s epoch 41.20 s val 0.00 s] lr = 0.050000 loss = -0.0026849 = 0.000669295 + lam * -3.3542, offending 21482
  Forward pass
  Distances
  Train
Valiation at epoch 109
zn_3     nbit= 14:  0.0002 0.0014 0.0101
zn_10    nbit= 32:  0.0038 0.0272 0.1308
zn_79    nbit= 64:  0.0904 0.3490 0.7513
storing test_ckpt/0.001/checkpoint.pth
epoch 109, times: [hn 4.03 s epoch 41.03 s val 9.72 s] lr = 0.050000 loss = -0.00266733 = 0.000698821 + lam * -3.36615, offending 21807
  Forward pass
  Distances
  Train
epoch 110, times: [hn 3.98 s epoch 41.75 s val 0.00 s] lr = 0.050000 loss = -0.00263875 = 0.000728527 + lam * -3.36728, offending 22986
  Forward pass
  Distances
  Train
epoch 111, times: [hn 4.11 s epoch 41.20 s val 0.00 s] lr = 0.050000 loss = -0.00249631 = 0.000844957 + lam * -3.34127, offending 26344
  Forward pass
  Distances
  Train
epoch 112, times: [hn 4.03 s epoch 40.91 s val 0.00 s] lr = 0.050000 loss = -0.00256006 = 0.000770761 + lam * -3.33082, offending 24665
  Forward pass
  Distances
  Train
epoch 113, times: [hn 4.01 s epoch 39.47 s val 0.00 s] lr = 0.050000 loss = -0.00256176 = 0.000783236 + lam * -3.345, offending 24900
  Forward pass
  Distances
  Train
epoch 114, times: [hn 4.03 s epoch 41.15 s val 0.00 s] lr = 0.050000 loss = -0.00264682 = 0.000712606 + lam * -3.35942, offending 22448
  Forward pass
  Distances
  Train
epoch 115, times: [hn 4.04 s epoch 40.79 s val 0.00 s] lr = 0.050000 loss = -0.00243297 = 0.000889904 + lam * -3.32288, offending 28255
  Forward pass
  Distances
  Train
epoch 116, times: [hn 4.06 s epoch 41.91 s val 0.00 s] lr = 0.050000 loss = -0.00253178 = 0.000794782 + lam * -3.32657, offending 25385
  Forward pass
  Distances
  Train
epoch 117, times: [hn 4.00 s epoch 40.46 s val 0.00 s] lr = 0.050000 loss = -0.00255198 = 0.000780342 + lam * -3.33232, offending 24960
  Forward pass
  Distances
  Train
epoch 118, times: [hn 4.04 s epoch 41.20 s val 0.00 s] lr = 0.050000 loss = -0.00268841 = 0.00066967 + lam * -3.35808, offending 20950
  Forward pass
  Distances
  Train
Valiation at epoch 119
zn_3     nbit= 14:  0.0001 0.0011 0.0091
zn_10    nbit= 32:  0.0034 0.0237 0.1184
zn_79    nbit= 64:  0.0822 0.3277 0.7323
storing test_ckpt/0.001/checkpoint.pth
epoch 119, times: [hn 4.03 s epoch 41.66 s val 9.39 s] lr = 0.050000 loss = -0.00265424 = 0.000714458 + lam * -3.3687, offending 22106
  Forward pass
  Distances
  Train
epoch 120, times: [hn 4.04 s epoch 40.42 s val 0.00 s] lr = 0.010000 loss = -0.00246188 = 0.000890166 + lam * -3.35205, offending 27668
  Forward pass
  Distances
  Train
epoch 121, times: [hn 4.04 s epoch 41.33 s val 0.00 s] lr = 0.010000 loss = -0.00250356 = 0.000845381 + lam * -3.34894, offending 26631
  Forward pass
  Distances
  Train
epoch 122, times: [hn 4.04 s epoch 41.20 s val 0.00 s] lr = 0.010000 loss = -0.00250886 = 0.000843989 + lam * -3.35284, offending 26829
  Forward pass
  Distances
  Train
epoch 123, times: [hn 4.05 s epoch 41.00 s val 0.00 s] lr = 0.010000 loss = -0.00268491 = 0.000670538 + lam * -3.35545, offending 21659
  Forward pass
  Distances
  Train
epoch 124, times: [hn 4.02 s epoch 41.11 s val 0.00 s] lr = 0.010000 loss = -0.00287056 = 0.000520481 + lam * -3.39104, offending 16835
  Forward pass
  Distances
  Train
epoch 125, times: [hn 4.04 s epoch 41.09 s val 0.00 s] lr = 0.010000 loss = -0.00265614 = 0.000745904 + lam * -3.40205, offending 22999
  Forward pass
  Distances
  Train
epoch 126, times: [hn 4.07 s epoch 40.92 s val 0.00 s] lr = 0.010000 loss = -0.00266996 = 0.000728622 + lam * -3.39858, offending 22700
  Forward pass
  Distances
  Train
epoch 127, times: [hn 4.05 s epoch 40.80 s val 0.00 s] lr = 0.010000 loss = -0.00272007 = 0.000683867 + lam * -3.40393, offending 21560
  Forward pass
  Distances
  Train
epoch 128, times: [hn 4.02 s epoch 40.83 s val 0.00 s] lr = 0.010000 loss = -0.00268353 = 0.000731328 + lam * -3.41486, offending 22360
  Forward pass
  Distances
  Train
Valiation at epoch 129
zn_3     nbit= 14:  0.0001 0.0012 0.0099
zn_10    nbit= 32:  0.0042 0.0280 0.1296
zn_79    nbit= 64:  0.0917 0.3555 0.7577
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.35547 > 0.35478), keeping as best
epoch 129, times: [hn 4.01 s epoch 40.46 s val 9.39 s] lr = 0.010000 loss = -0.00265526 = 0.000750833 + lam * -3.40609, offending 22916
  Forward pass
  Distances
  Train
epoch 130, times: [hn 4.14 s epoch 41.75 s val 0.00 s] lr = 0.010000 loss = -0.00267619 = 0.000720112 + lam * -3.3963, offending 22419
  Forward pass
  Distances
  Train
epoch 131, times: [hn 4.03 s epoch 39.56 s val 0.00 s] lr = 0.010000 loss = -0.00274079 = 0.000665432 + lam * -3.40622, offending 20771
  Forward pass
  Distances
  Train
epoch 132, times: [hn 4.03 s epoch 41.41 s val 0.00 s] lr = 0.010000 loss = -0.00280207 = 0.00061567 + lam * -3.41774, offending 19253
  Forward pass
  Distances
  Train
epoch 133, times: [hn 4.08 s epoch 41.41 s val 0.00 s] lr = 0.010000 loss = -0.00275714 = 0.000670556 + lam * -3.4277, offending 20851
  Forward pass
  Distances
  Train
epoch 134, times: [hn 4.04 s epoch 41.39 s val 0.00 s] lr = 0.010000 loss = -0.00273281 = 0.000693662 + lam * -3.42648, offending 21071
  Forward pass
  Distances
  Train
epoch 135, times: [hn 4.03 s epoch 41.62 s val 0.00 s] lr = 0.010000 loss = -0.00267054 = 0.000754265 + lam * -3.42481, offending 22838
  Forward pass
  Distances
  Train
epoch 136, times: [hn 4.00 s epoch 40.91 s val 0.00 s] lr = 0.010000 loss = -0.00243555 = 0.00096249 + lam * -3.39804, offending 28966
  Forward pass
  Distances
  Train
epoch 137, times: [hn 4.13 s epoch 41.99 s val 0.00 s] lr = 0.010000 loss = -0.00254905 = 0.000834254 + lam * -3.3833, offending 25736
  Forward pass
  Distances
  Train
epoch 138, times: [hn 4.05 s epoch 41.24 s val 0.00 s] lr = 0.010000 loss = -0.00279365 = 0.000607697 + lam * -3.40135, offending 19366
  Forward pass
  Distances
  Train
Valiation at epoch 139
zn_3     nbit= 14:  0.0001 0.0010 0.0094
zn_10    nbit= 32:  0.0041 0.0279 0.1298
zn_79    nbit= 64:  0.0929 0.3562 0.7594
storing test_ckpt/0.001/checkpoint.pth
zn_79,rank=10 score improves (0.35615 > 0.35547), keeping as best
epoch 139, times: [hn 4.03 s epoch 41.74 s val 10.33 s] lr = 0.010000 loss = -0.00268784 = 0.000721077 + lam * -3.40892, offending 22235
  Forward pass
  Distances
  Train
epoch 140, times: [hn 4.13 s epoch 40.56 s val 0.00 s] lr = 0.010000 loss = -0.0027821 = 0.00063587 + lam * -3.41797, offending 19541
  Forward pass
  Distances
  Train
epoch 141, times: [hn 4.05 s epoch 41.27 s val 0.00 s] lr = 0.010000 loss = -0.00274203 = 0.000680079 + lam * -3.42211, offending 20910
  Forward pass
  Distances
  Train
epoch 142, times: [hn 4.04 s epoch 41.65 s val 0.00 s] lr = 0.010000 loss = -0.00273043 = 0.000693222 + lam * -3.42365, offending 21071
  Forward pass
  Distances
  Train
epoch 143, times: [hn 4.05 s epoch 41.27 s val 0.00 s] lr = 0.010000 loss = -0.00263947 = 0.000779117 + lam * -3.41859, offending 23710
  Forward pass
  Distances
  Train
epoch 144, times: [hn 4.01 s epoch 40.72 s val 0.00 s] lr = 0.010000 loss = -0.00260661 = 0.000805094 + lam * -3.4117, offending 24010
  Forward pass
  Distances
  Train
epoch 145, times: [hn 4.02 s epoch 41.64 s val 0.00 s] lr = 0.010000 loss = -0.00270845 = 0.000698124 + lam * -3.40658, offending 21290
  Forward pass
  Distances
  Train
epoch 146, times: [hn 4.02 s epoch 41.08 s val 0.00 s] lr = 0.010000 loss = -0.00273785 = 0.000679729 + lam * -3.41758, offending 20961
  Forward pass
  Distances
  Train
epoch 147, times: [hn 4.03 s epoch 41.42 s val 0.00 s] lr = 0.010000 loss = -0.00272008 = 0.000697331 + lam * -3.41741, offending 21513
  Forward pass
  Distances
  Train
epoch 148, times: [hn 4.04 s epoch 41.48 s val 0.00 s] lr = 0.010000 loss = -0.0028138 = 0.000615015 + lam * -3.42881, offending 19003
  Forward pass
  Distances
  Train
Valiation at epoch 149
zn_3     nbit= 14:  0.0001 0.0010 0.0090
zn_10    nbit= 32:  0.0036 0.0258 0.1231
zn_79    nbit= 64:  0.0871 0.3401 0.7421
storing test_ckpt/0.001/checkpoint.pth
epoch 149, times: [hn 4.02 s epoch 41.02 s val 9.28 s] lr = 0.010000 loss = -0.00272856 = 0.000705613 + lam * -3.43418, offending 21384
  Forward pass
  Distances
  Train
epoch 150, times: [hn 4.14 s epoch 41.79 s val 0.00 s] lr = 0.010000 loss = -0.00280064 = 0.0006363 + lam * -3.43694, offending 19364
  Forward pass
  Distances
  Train
epoch 151, times: [hn 3.97 s epoch 41.18 s val 0.00 s] lr = 0.010000 loss = -0.00265455 = 0.00077967 + lam * -3.43422, offending 23260
  Forward pass
  Distances
  Train
epoch 152, times: [hn 4.04 s epoch 40.46 s val 0.00 s] lr = 0.010000 loss = -0.00269163 = 0.00073813 + lam * -3.42976, offending 22156
  Forward pass
  Distances
  Train
epoch 153, times: [hn 4.05 s epoch 40.30 s val 0.00 s] lr = 0.010000 loss = -0.00274675 = 0.000684646 + lam * -3.43139, offending 20602
  Forward pass
  Distances
  Train
epoch 154, times: [hn 4.02 s epoch 40.44 s val 0.00 s] lr = 0.010000 loss = -0.00269605 = 0.000737706 + lam * -3.43375, offending 21887
  Forward pass
  Distances
  Train
epoch 155, times: [hn 4.04 s epoch 41.35 s val 0.00 s] lr = 0.010000 loss = -0.00275733 = 0.000672064 + lam * -3.42939, offending 20554
  Forward pass
  Distances
  Train
epoch 156, times: [hn 4.04 s epoch 41.78 s val 0.00 s] lr = 0.010000 loss = -0.00270959 = 0.000724848 + lam * -3.43444, offending 21876
  Forward pass
  Distances
  Train
epoch 157, times: [hn 4.04 s epoch 40.14 s val 0.00 s] lr = 0.010000 loss = -0.00270009 = 0.000730731 + lam * -3.43082, offending 22016
  Forward pass
  Distances
  Train
epoch 158, times: [hn 4.04 s epoch 41.44 s val 0.00 s] lr = 0.010000 loss = -0.00252094 = 0.000894191 + lam * -3.41513, offending 26573
  Forward pass
  Distances
  Train
Valiation at epoch 159
zn_3     nbit= 14:  0.0001 0.0012 0.0097
zn_10    nbit= 32:  0.0038 0.0271 0.1299
zn_79    nbit= 64:  0.0917 0.3556 0.7584
storing test_ckpt/0.001/checkpoint.pth
epoch 159, times: [hn 4.07 s epoch 40.81 s val 10.44 s] lr = 0.010000 loss = -0.00267252 = 0.000737465 + lam * -3.40998, offending 22468
++ tac test_ckpt/0.001.stdout
++ grep -m1 'keeping as best'
++ grep -o '[0-9\.]*'
++ grep -o '(.*>'
+ perf=0.35615
+ echo 0.35615
0.35615
+ [[ 0.35615 > 0.0000 ]]
+ best_perf=0.35615
+ best_lambda=0.001
+ for lambda in '$lambdas'
+ mkdir -p test_ckpt/0.002
+ python -u train.py --dout 24 --save_best_criterion zn_79,rank=10 --database deep1b --lambda_uniform 0.002 --checkpoint_dir test_ckpt/0.002
+ tee test_ckpt/0.002.stdout
Namespace(batch_size=64, checkpoint_dir='test_ckpt/0.002', database='deep1b', device='cuda', dint=1024, dout=24, epochs=160, init_name='', lambda_uniform=0.00
2, lr_schedule='0.1,0.1,0.05,0.01', momentum=0.9, num_learn=500000, quantizer_train='', rank_negative=50, rank_positive=10, save_best_criterion='zn_79,rank=10
', seed=1234, size_base=1000000, val_freq=10, validation_quantizers=['zn_3', 'zn_10', 'zn_79'])
load dataset deep1b
keeping 500000/357380000 training vectors
computing training ground truth
build network
Lr schedule [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1,
0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1
, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0
.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0
5, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01,
 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01]
  Forward pass
  Distances
  Train
epoch 0, times: [hn 3.99 s epoch 55.19 s val 0.00 s] lr = 0.100000 loss = -0.00585795 = 0.00175652 + lam * -3.80723, offending 17773
  Forward pass
  Distances
  Train
epoch 1, times: [hn 4.07 s epoch 57.41 s val 0.00 s] lr = 0.100000 loss = -0.0034838 = 0.00245264 + lam * -2.96822, offending 56211
  Forward pass
  Distances
  Train
epoch 2, times: [hn 4.07 s epoch 57.20 s val 0.00 s] lr = 0.100000 loss = -0.00364436 = 0.0021228 + lam * -2.88358, offending 62854
  Forward pass
  Distances
  Train
epoch 3, times: [hn 4.03 s epoch 57.56 s val 0.00 s] lr = 0.100000 loss = -0.00355411 = 0.00213923 + lam * -2.84667, offending 71099
  Forward pass
  Distances
  Train
epoch 4, times: [hn 4.04 s epoch 55.12 s val 0.00 s] lr = 0.100000 loss = -0.00361521 = 0.00208341 + lam * -2.84931, offending 73287
  Forward pass
  Distances
  Train
epoch 5, times: [hn 4.05 s epoch 53.34 s val 0.00 s] lr = 0.100000 loss = -0.00366423 = 0.00206223 + lam * -2.86323, offending 75161
  Forward pass
  Distances
  Train
epoch 6, times: [hn 4.03 s epoch 54.38 s val 0.00 s] lr = 0.100000 loss = -0.00371333 = 0.00205965 + lam * -2.88649, offending 75809
  Forward pass
  Distances
  Train
epoch 7, times: [hn 4.03 s epoch 57.88 s val 0.00 s] lr = 0.100000 loss = -0.00379055 = 0.00203499 + lam * -2.91277, offending 74896
  Forward pass
  Distances
  Train
epoch 8, times: [hn 4.04 s epoch 55.10 s val 0.00 s] lr = 0.100000 loss = -0.00382894 = 0.00203354 + lam * -2.93124, offending 75412
  Forward pass
  Distances
  Train
Valiation at epoch 9
zn_3     nbit= 14:  0.0000 0.0003 0.0028
zn_10    nbit= 32:  0.0009 0.0073 0.0437
zn_79    nbit= 64:  0.0331 0.1581 0.4756
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.15814 > 0), keeping as best
epoch 9, times: [hn 4.08 s epoch 55.19 s val 10.01 s] lr = 0.100000 loss = -0.00391856 = 0.00200767 + lam * -2.96312, offending 73579
  Forward pass
  Distances
  Train
epoch 10, times: [hn 4.11 s epoch 53.95 s val 0.00 s] lr = 0.100000 loss = -0.00395531 = 0.00202928 + lam * -2.99229, offending 72418
  Forward pass
  Distances
  Train
epoch 11, times: [hn 4.02 s epoch 54.31 s val 0.00 s] lr = 0.100000 loss = -0.00403979 = 0.00199679 + lam * -3.01829, offending 71100
  Forward pass
  Distances
  Train
epoch 12, times: [hn 4.03 s epoch 57.55 s val 0.00 s] lr = 0.100000 loss = -0.00410755 = 0.00199602 + lam * -3.05178, offending 69239
  Forward pass
  Distances
  Train
epoch 13, times: [hn 4.17 s epoch 54.91 s val 0.00 s] lr = 0.100000 loss = -0.00412215 = 0.00202078 + lam * -3.07146, offending 69505
  Forward pass
  Distances
  Train
epoch 14, times: [hn 3.98 s epoch 50.64 s val 0.00 s] lr = 0.100000 loss = -0.00417818 = 0.00200346 + lam * -3.09082, offending 67861
  Forward pass
  Distances
  Train
epoch 15, times: [hn 4.04 s epoch 51.37 s val 0.00 s] lr = 0.100000 loss = -0.00423717 = 0.00199063 + lam * -3.1139, offending 66580
  Forward pass
  Distances
  Train
epoch 16, times: [hn 4.06 s epoch 53.19 s val 0.00 s] lr = 0.100000 loss = -0.00430782 = 0.00197357 + lam * -3.1407, offending 64886
  Forward pass
  Distances
  Train
epoch 17, times: [hn 4.06 s epoch 55.08 s val 0.00 s] lr = 0.100000 loss = -0.00430876 = 0.00199904 + lam * -3.1539, offending 65025
  Forward pass
  Distances
  Train
epoch 18, times: [hn 4.05 s epoch 49.89 s val 0.00 s] lr = 0.100000 loss = -0.00438724 = 0.00196599 + lam * -3.17661, offending 63393
  Forward pass
  Distances
  Train
Valiation at epoch 19
zn_3     nbit= 14:  0.0000 0.0004 0.0034
zn_10    nbit= 32:  0.0016 0.0123 0.0688
zn_79    nbit= 64:  0.0534 0.2344 0.6084
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.23437 > 0.15814), keeping as best
epoch 19, times: [hn 4.04 s epoch 55.62 s val 9.99 s] lr = 0.100000 loss = -0.00438699 = 0.00198257 + lam * -3.18478, offending 63419
  Forward pass
  Distances
  Train
epoch 20, times: [hn 4.27 s epoch 52.40 s val 0.00 s] lr = 0.100000 loss = -0.00446503 = 0.00195871 + lam * -3.21187, offending 61738
  Forward pass
  Distances
  Train
epoch 21, times: [hn 4.12 s epoch 51.30 s val 0.00 s] lr = 0.100000 loss = -0.0044823 = 0.00196294 + lam * -3.22262, offending 61476
  Forward pass
  Distances
  Train
epoch 22, times: [hn 4.05 s epoch 53.74 s val 0.00 s] lr = 0.100000 loss = -0.00456725 = 0.00193225 + lam * -3.24975, offending 59199
  Forward pass
  Distances
  Train
epoch 23, times: [hn 4.05 s epoch 54.78 s val 0.00 s] lr = 0.100000 loss = -0.00461641 = 0.00191901 + lam * -3.26771, offending 58068
  Forward pass
  Distances
  Train
epoch 24, times: [hn 3.98 s epoch 51.39 s val 0.00 s] lr = 0.100000 loss = -0.00458983 = 0.00195075 + lam * -3.27029, offending 58857
  Forward pass
  Distances
  Train
epoch 25, times: [hn 4.06 s epoch 60.65 s val 0.00 s] lr = 0.100000 loss = -0.00467046 = 0.00189823 + lam * -3.28434, offending 56867
  Forward pass
  Distances
  Train
epoch 26, times: [hn 4.07 s epoch 52.84 s val 0.00 s] lr = 0.100000 loss = -0.00480458 = 0.00183198 + lam * -3.31828, offending 54078
  Forward pass
  Distances
  Train
epoch 27, times: [hn 4.04 s epoch 57.42 s val 0.00 s] lr = 0.100000 loss = -0.00480334 = 0.00187152 + lam * -3.33743, offending 54270
  Forward pass
  Distances
  Train
epoch 28, times: [hn 3.99 s epoch 50.52 s val 0.00 s] lr = 0.100000 loss = -0.00485066 = 0.00184127 + lam * -3.34597, offending 53079
  Forward pass
  Distances
  Train
Valiation at epoch 29
zn_3     nbit= 14:  0.0001 0.0006 0.0054
zn_10    nbit= 32:  0.0024 0.0186 0.0967
zn_79    nbit= 64:  0.0730 0.2928 0.6898
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.29277 > 0.23437), keeping as best
epoch 29, times: [hn 4.04 s epoch 56.43 s val 9.44 s] lr = 0.100000 loss = -0.00481152 = 0.00187112 + lam * -3.34132, offending 53870
  Forward pass
  Distances
  Train
epoch 30, times: [hn 4.12 s epoch 53.14 s val 0.00 s] lr = 0.100000 loss = -0.0047848 = 0.00189665 + lam * -3.34073, offending 54685
  Forward pass
  Distances
  Train
epoch 31, times: [hn 4.05 s epoch 50.68 s val 0.00 s] lr = 0.100000 loss = -0.00495713 = 0.00176848 + lam * -3.36281, offending 50844
  Forward pass
  Distances
  Train
epoch 32, times: [hn 4.08 s epoch 48.44 s val 0.00 s] lr = 0.100000 loss = -0.00495974 = 0.00179944 + lam * -3.37959, offending 50469
  Forward pass
  Distances
  Train
epoch 33, times: [hn 4.08 s epoch 51.35 s val 0.00 s] lr = 0.100000 loss = -0.00500743 = 0.00177407 + lam * -3.39075, offending 49547
  Forward pass
  Distances
  Train
epoch 34, times: [hn 4.08 s epoch 56.09 s val 0.00 s] lr = 0.100000 loss = -0.00508896 = 0.00172275 + lam * -3.40585, offending 48099
  Forward pass
  Distances
  Train
epoch 35, times: [hn 4.07 s epoch 45.53 s val 0.00 s] lr = 0.100000 loss = -0.00520221 = 0.0016585 + lam * -3.43036, offending 45659
  Forward pass
  Distances
  Train
epoch 36, times: [hn 4.10 s epoch 53.37 s val 0.00 s] lr = 0.100000 loss = -0.00502531 = 0.00180282 + lam * -3.41406, offending 49710
  Forward pass
  Distances
  Train
epoch 37, times: [hn 4.07 s epoch 57.98 s val 0.00 s] lr = 0.100000 loss = -0.00514492 = 0.00169788 + lam * -3.4214, offending 47439
  Forward pass
  Distances
  Train
epoch 38, times: [hn 4.06 s epoch 56.15 s val 0.00 s] lr = 0.100000 loss = -0.00514714 = 0.00170963 + lam * -3.42838, offending 47023
  Forward pass
  Distances
  Train
Valiation at epoch 39
zn_3     nbit= 14:  0.0001 0.0010 0.0090
zn_10    nbit= 32:  0.0039 0.0288 0.1376
zn_79    nbit= 64:  0.0946 0.3636 0.7666
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.36359 > 0.29277), keeping as best
epoch 39, times: [hn 3.98 s epoch 57.34 s val 9.68 s] lr = 0.100000 loss = -0.00536024 = 0.00155735 + lam * -3.45879, offending 42480
  Forward pass
  Distances
  Train
epoch 40, times: [hn 4.20 s epoch 58.07 s val 0.00 s] lr = 0.100000 loss = -0.00537118 = 0.00156861 + lam * -3.46989, offending 42483
  Forward pass
  Distances
  Train
epoch 41, times: [hn 4.04 s epoch 58.58 s val 0.00 s] lr = 0.100000 loss = -0.00536963 = 0.00158203 + lam * -3.47583, offending 42432
  Forward pass
  Distances
  Train
epoch 42, times: [hn 4.04 s epoch 55.33 s val 0.00 s] lr = 0.100000 loss = -0.00538192 = 0.00157967 + lam * -3.48079, offending 42134
  Forward pass
  Distances
  Train
epoch 43, times: [hn 4.19 s epoch 58.19 s val 0.00 s] lr = 0.100000 loss = -0.0051672 = 0.00173464 + lam * -3.45092, offending 46588
  Forward pass
  Distances
  Train
epoch 44, times: [hn 4.06 s epoch 50.40 s val 0.00 s] lr = 0.100000 loss = -0.00537267 = 0.00156562 + lam * -3.46914, offending 42113
  Forward pass
  Distances
  Train
epoch 45, times: [hn 4.12 s epoch 54.38 s val 0.00 s] lr = 0.100000 loss = -0.00538563 = 0.00157039 + lam * -3.47801, offending 42017
  Forward pass
  Distances
  Train
epoch 46, times: [hn 4.12 s epoch 51.63 s val 0.00 s] lr = 0.100000 loss = -0.00538503 = 0.0015929 + lam * -3.48896, offending 41960
  Forward pass
  Distances
  Train
epoch 47, times: [hn 4.05 s epoch 53.41 s val 0.00 s] lr = 0.100000 loss = -0.00539454 = 0.00158482 + lam * -3.48968, offending 42251
  Forward pass
  Distances
  Train
epoch 48, times: [hn 4.06 s epoch 53.17 s val 0.00 s] lr = 0.100000 loss = -0.00540621 = 0.00157715 + lam * -3.49168, offending 41845
  Forward pass
  Distances
  Train
Valiation at epoch 49
zn_3     nbit= 14:  0.0001 0.0010 0.0100
zn_10    nbit= 32:  0.0055 0.0366 0.1631
zn_79    nbit= 64:  0.1068 0.3954 0.7979
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.39543 > 0.36359), keeping as best
epoch 49, times: [hn 3.99 s epoch 58.32 s val 9.54 s] lr = 0.100000 loss = -0.00562061 = 0.00142512 + lam * -3.52286, offending 37381
  Forward pass
  Distances
  Train
epoch 50, times: [hn 4.06 s epoch 58.33 s val 0.00 s] lr = 0.100000 loss = -0.00554719 = 0.00148999 + lam * -3.51859, offending 38993
  Forward pass
  Distances
  Train
epoch 51, times: [hn 4.04 s epoch 58.65 s val 0.00 s] lr = 0.100000 loss = -0.00535555 = 0.00164232 + lam * -3.49894, offending 42960
  Forward pass
  Distances
  Train
epoch 52, times: [hn 4.06 s epoch 54.58 s val 0.00 s] lr = 0.100000 loss = -0.0057231 = 0.00135345 + lam * -3.53827, offending 35193
  Forward pass
  Distances
  Train
epoch 53, times: [hn 4.14 s epoch 56.94 s val 0.00 s] lr = 0.100000 loss = -0.00537937 = 0.0016486 + lam * -3.51398, offending 42447
  Forward pass
  Distances
  Train
epoch 54, times: [hn 4.06 s epoch 52.36 s val 0.00 s] lr = 0.100000 loss = -0.00578441 = 0.00131669 + lam * -3.55055, offending 34321
  Forward pass
  Distances
  Train
epoch 55, times: [hn 4.09 s epoch 51.80 s val 0.00 s] lr = 0.100000 loss = -0.00594909 = 0.00121847 + lam * -3.58378, offending 30862
  Forward pass
  Distances
  Train
epoch 56, times: [hn 4.05 s epoch 55.92 s val 0.00 s] lr = 0.100000 loss = -0.00577593 = 0.00137636 + lam * -3.57614, offending 34374
  Forward pass
  Distances
  Train
epoch 57, times: [hn 4.04 s epoch 57.05 s val 0.00 s] lr = 0.100000 loss = -0.00545651 = 0.00160559 + lam * -3.53105, offending 40439
  Forward pass
  Distances
  Train
epoch 58, times: [hn 4.15 s epoch 54.05 s val 0.00 s] lr = 0.100000 loss = -0.00539451 = 0.00162293 + lam * -3.50872, offending 41680
  Forward pass
  Distances
  Train
Valiation at epoch 59
zn_3     nbit= 14:  0.0001 0.0015 0.0125
zn_10    nbit= 32:  0.0063 0.0403 0.1775
zn_79    nbit= 64:  0.1142 0.4168 0.8135
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.41682 > 0.39543), keeping as best
epoch 59, times: [hn 4.08 s epoch 55.65 s val 9.63 s] lr = 0.100000 loss = -0.00569123 = 0.00139188 + lam * -3.54156, offending 35729
  Forward pass
  Distances
  Train
epoch 60, times: [hn 4.14 s epoch 52.44 s val 0.00 s] lr = 0.100000 loss = -0.00567878 = 0.00143155 + lam * -3.55517, offending 36402
  Forward pass
  Distances
  Train
epoch 61, times: [hn 4.08 s epoch 58.34 s val 0.00 s] lr = 0.100000 loss = -0.00592538 = 0.00125234 + lam * -3.58886, offending 31624
  Forward pass
  Distances
  Train
epoch 62, times: [hn 4.11 s epoch 50.58 s val 0.00 s] lr = 0.100000 loss = -0.00585372 = 0.00132674 + lam * -3.59023, offending 33118
  Forward pass
  Distances
  Train
epoch 63, times: [hn 4.09 s epoch 42.98 s val 0.00 s] lr = 0.100000 loss = -0.00571894 = 0.00143478 + lam * -3.57686, offending 35742
  Forward pass
  Distances
  Train
epoch 64, times: [hn 4.11 s epoch 56.86 s val 0.00 s] lr = 0.100000 loss = -0.00578167 = 0.00137059 + lam * -3.57613, offending 33923
  Forward pass
  Distances
  Train
epoch 65, times: [hn 4.04 s epoch 42.86 s val 0.00 s] lr = 0.100000 loss = -0.00576808 = 0.00139766 + lam * -3.58287, offending 34483
  Forward pass
  Distances
  Train
epoch 66, times: [hn 4.09 s epoch 56.81 s val 0.00 s] lr = 0.100000 loss = -0.00604172 = 0.00119315 + lam * -3.61744, offending 29330
  Forward pass
  Distances
  Train
epoch 67, times: [hn 4.07 s epoch 58.71 s val 0.00 s] lr = 0.100000 loss = -0.00583087 = 0.00136912 + lam * -3.59999, offending 33149
  Forward pass
  Distances
  Train
epoch 68, times: [hn 4.08 s epoch 59.12 s val 0.00 s] lr = 0.100000 loss = -0.00562186 = 0.00151719 + lam * -3.56952, offending 36998
  Forward pass
  Distances
  Train
Valiation at epoch 69
zn_3     nbit= 14:  0.0001 0.0012 0.0108
zn_10    nbit= 32:  0.0061 0.0409 0.1790
zn_79    nbit= 64:  0.1151 0.4186 0.8159
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.41862 > 0.41682), keeping as best
epoch 69, times: [hn 4.05 s epoch 58.40 s val 9.61 s] lr = 0.100000 loss = -0.00562056 = 0.00150873 + lam * -3.56465, offending 37275
  Forward pass
  Distances
  Train
epoch 70, times: [hn 4.18 s epoch 53.94 s val 0.00 s] lr = 0.100000 loss = -0.00571588 = 0.00142718 + lam * -3.57153, offending 35285
  Forward pass
  Distances
  Train
epoch 71, times: [hn 4.05 s epoch 51.37 s val 0.00 s] lr = 0.100000 loss = -0.00601639 = 0.00120656 + lam * -3.61147, offending 29408
  Forward pass
  Distances
  Train
epoch 72, times: [hn 4.06 s epoch 56.09 s val 0.00 s] lr = 0.100000 loss = -0.0059893 = 0.00126386 + lam * -3.62658, offending 30183
  Forward pass
  Distances
  Train
epoch 73, times: [hn 4.04 s epoch 56.23 s val 0.00 s] lr = 0.100000 loss = -0.00582977 = 0.0013895 + lam * -3.60964, offending 32955
  Forward pass
  Distances
  Train
epoch 74, times: [hn 4.06 s epoch 59.32 s val 0.00 s] lr = 0.100000 loss = -0.00587728 = 0.00135558 + lam * -3.61643, offending 32168
  Forward pass
  Distances
  Train
epoch 75, times: [hn 4.08 s epoch 57.03 s val 0.00 s] lr = 0.100000 loss = -0.00580616 = 0.00140255 + lam * -3.60436, offending 33556
  Forward pass
  Distances
  Train
epoch 76, times: [hn 4.05 s epoch 57.74 s val 0.00 s] lr = 0.100000 loss = -0.00590936 = 0.00131055 + lam * -3.60995, offending 31307
  Forward pass
  Distances
  Train
epoch 77, times: [hn 4.04 s epoch 52.87 s val 0.00 s] lr = 0.100000 loss = -0.00545774 = 0.0016612 + lam * -3.55947, offending 40237
  Forward pass
  Distances
  Train
epoch 78, times: [hn 4.23 s epoch 56.06 s val 0.00 s] lr = 0.100000 loss = -0.00575904 = 0.00141514 + lam * -3.58709, offending 34569
  Forward pass
  Distances
  Train
Valiation at epoch 79
zn_3     nbit= 14:  0.0002 0.0017 0.0142
zn_10    nbit= 32:  0.0083 0.0499 0.2140
zn_79    nbit= 64:  0.1256 0.4459 0.8371
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.44587 > 0.41862), keeping as best
epoch 79, times: [hn 4.07 s epoch 55.81 s val 9.57 s] lr = 0.100000 loss = -0.00603746 = 0.00122401 + lam * -3.63073, offending 28959
  Forward pass
  Distances
  Train
epoch 80, times: [hn 4.12 s epoch 56.79 s val 0.00 s] lr = 0.050000 loss = -0.0062482 = 0.00108911 + lam * -3.66865, offending 26155
  Forward pass
  Distances
  Train
epoch 81, times: [hn 4.05 s epoch 54.48 s val 0.00 s] lr = 0.050000 loss = -0.00611955 = 0.00122833 + lam * -3.67394, offending 28859
  Forward pass
  Distances
  Train
epoch 82, times: [hn 4.05 s epoch 58.83 s val 0.00 s] lr = 0.050000 loss = -0.00609316 = 0.00124577 + lam * -3.66946, offending 29421
  Forward pass
  Distances
  Train
epoch 83, times: [hn 4.05 s epoch 55.11 s val 0.00 s] lr = 0.050000 loss = -0.00608725 = 0.00125107 + lam * -3.66916, offending 29730
  Forward pass
  Distances
  Train
epoch 84, times: [hn 4.08 s epoch 57.86 s val 0.00 s] lr = 0.050000 loss = -0.0056756 = 0.00158575 + lam * -3.63068, offending 37593
  Forward pass
  Distances
  Train
epoch 85, times: [hn 4.08 s epoch 54.30 s val 0.00 s] lr = 0.050000 loss = -0.00629972 = 0.00104302 + lam * -3.67137, offending 25108
  Forward pass
  Distances
  Train
epoch 86, times: [hn 4.06 s epoch 58.76 s val 0.00 s] lr = 0.050000 loss = -0.00583888 = 0.00146721 + lam * -3.65305, offending 34210
  Forward pass
  Distances
  Train
epoch 87, times: [hn 4.05 s epoch 56.08 s val 0.00 s] lr = 0.050000 loss = -0.00610004 = 0.0012268 + lam * -3.66342, offending 29246
  Forward pass
  Distances
  Train
epoch 88, times: [hn 4.06 s epoch 58.38 s val 0.00 s] lr = 0.050000 loss = -0.00592057 = 0.00139788 + lam * -3.65923, offending 32682
  Forward pass
  Distances
  Train
Valiation at epoch 89
zn_3     nbit= 14:  0.0003 0.0021 0.0179
zn_10    nbit= 32:  0.0111 0.0636 0.2541
zn_79    nbit= 64:  0.1423 0.4855 0.8676
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.48552 > 0.44587), keeping as best
epoch 89, times: [hn 4.07 s epoch 56.09 s val 9.59 s] lr = 0.050000 loss = -0.00622501 = 0.00113316 + lam * -3.67908, offending 26718
  Forward pass
  Distances
  Train
epoch 90, times: [hn 4.13 s epoch 53.98 s val 0.00 s] lr = 0.050000 loss = -0.0061128 = 0.00125096 + lam * -3.68188, offending 29373
  Forward pass
  Distances
  Train
epoch 91, times: [hn 4.05 s epoch 52.25 s val 0.00 s] lr = 0.050000 loss = -0.00623043 = 0.00115715 + lam * -3.69379, offending 26881
  Forward pass
  Distances
  Train
epoch 92, times: [hn 4.08 s epoch 56.31 s val 0.00 s] lr = 0.050000 loss = -0.00612689 = 0.00124593 + lam * -3.68641, offending 28780
  Forward pass
  Distances
  Train
epoch 93, times: [hn 4.05 s epoch 53.74 s val 0.00 s] lr = 0.050000 loss = -0.00634647 = 0.00106985 + lam * -3.70816, offending 24990
  Forward pass
  Distances
  Train
epoch 94, times: [hn 4.09 s epoch 57.74 s val 0.00 s] lr = 0.050000 loss = -0.00611102 = 0.00127473 + lam * -3.69287, offending 29465
  Forward pass
  Distances
  Train
epoch 95, times: [hn 4.05 s epoch 51.09 s val 0.00 s] lr = 0.050000 loss = -0.00596671 = 0.0013655 + lam * -3.6661, offending 31485
  Forward pass
  Distances
  Train
epoch 96, times: [hn 3.98 s epoch 54.26 s val 0.00 s] lr = 0.050000 loss = -0.00601874 = 0.00131976 + lam * -3.66925, offending 30359
  Forward pass
  Distances
  Train
epoch 97, times: [hn 4.04 s epoch 62.89 s val 0.00 s] lr = 0.050000 loss = -0.00595955 = 0.00137638 + lam * -3.66796, offending 31496
  Forward pass
  Distances
  Train
epoch 98, times: [hn 3.99 s epoch 57.76 s val 0.00 s] lr = 0.050000 loss = -0.00625231 = 0.00112327 + lam * -3.68779, offending 26114
  Forward pass
  Distances
  Train
Valiation at epoch 99
zn_3     nbit= 14:  0.0003 0.0027 0.0208
zn_10    nbit= 32:  0.0131 0.0749 0.2894
zn_79    nbit= 64:  0.1508 0.5059 0.8819
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.50593 > 0.48552), keeping as best
epoch 99, times: [hn 4.04 s epoch 59.77 s val 9.63 s] lr = 0.050000 loss = -0.00606476 = 0.00131027 + lam * -3.68752, offending 29773
  Forward pass
  Distances
  Train
epoch 100, times: [hn 4.18 s epoch 60.80 s val 0.00 s] lr = 0.050000 loss = -0.00571323 = 0.00156919 + lam * -3.64121, offending 36672
  Forward pass
  Distances
  Train
epoch 101, times: [hn 4.08 s epoch 56.80 s val 0.00 s] lr = 0.050000 loss = -0.00597385 = 0.00133486 + lam * -3.65435, offending 31388
  Forward pass
  Distances
  Train
epoch 102, times: [hn 4.11 s epoch 55.98 s val 0.00 s] lr = 0.050000 loss = -0.00619232 = 0.00117775 + lam * -3.68503, offending 27359
  Forward pass
  Distances
  Train
epoch 103, times: [hn 4.06 s epoch 56.67 s val 0.00 s] lr = 0.050000 loss = -0.00616468 = 0.00121528 + lam * -3.68998, offending 28143
  Forward pass
  Distances
  Train
epoch 104, times: [hn 4.07 s epoch 55.23 s val 0.00 s] lr = 0.050000 loss = -0.00616898 = 0.00122332 + lam * -3.69615, offending 27784
  Forward pass
  Distances
  Train
epoch 105, times: [hn 4.05 s epoch 56.65 s val 0.00 s] lr = 0.050000 loss = -0.00577248 = 0.00153702 + lam * -3.65475, offending 35419
  Forward pass
  Distances
  Train
epoch 106, times: [hn 4.03 s epoch 55.52 s val 0.00 s] lr = 0.050000 loss = -0.00631746 = 0.00107915 + lam * -3.69831, offending 24931
  Forward pass
  Distances
  Train
epoch 107, times: [hn 3.99 s epoch 55.76 s val 0.00 s] lr = 0.050000 loss = -0.00618491 = 0.0012255 + lam * -3.70521, offending 27754
  Forward pass
  Distances
  Train
epoch 108, times: [hn 4.07 s epoch 56.82 s val 0.00 s] lr = 0.050000 loss = -0.00602918 = 0.00133907 + lam * -3.68412, offending 30354
  Forward pass
  Distances
  Train
Valiation at epoch 109
zn_3     nbit= 14:  0.0002 0.0025 0.0201
zn_10    nbit= 32:  0.0121 0.0696 0.2712
zn_79    nbit= 64:  0.1429 0.4900 0.8664
storing test_ckpt/0.002/checkpoint.pth
epoch 109, times: [hn 4.06 s epoch 54.94 s val 9.61 s] lr = 0.050000 loss = -0.0063941 = 0.00104294 + lam * -3.71852, offending 23786
  Forward pass
  Distances
  Train
epoch 110, times: [hn 4.10 s epoch 58.80 s val 0.00 s] lr = 0.050000 loss = -0.0064049 = 0.00106436 + lam * -3.73463, offending 23723
  Forward pass
  Distances
  Train
epoch 111, times: [hn 4.09 s epoch 59.29 s val 0.00 s] lr = 0.050000 loss = -0.0062414 = 0.00119029 + lam * -3.71584, offending 26679
  Forward pass
  Distances
  Train
epoch 112, times: [hn 4.07 s epoch 53.27 s val 0.00 s] lr = 0.050000 loss = -0.0063385 = 0.00111387 + lam * -3.72618, offending 24943
  Forward pass
  Distances
  Train
epoch 113, times: [hn 4.06 s epoch 54.57 s val 0.00 s] lr = 0.050000 loss = -0.00633912 = 0.00111409 + lam * -3.7266, offending 25143
  Forward pass
  Distances
  Train
epoch 114, times: [hn 4.06 s epoch 56.31 s val 0.00 s] lr = 0.050000 loss = -0.00610228 = 0.00130546 + lam * -3.70387, offending 29235
  Forward pass
  Distances
  Train
epoch 115, times: [hn 4.05 s epoch 58.32 s val 0.00 s] lr = 0.050000 loss = -0.0061471 = 0.00125687 + lam * -3.70198, offending 28546
  Forward pass
  Distances
  Train
epoch 116, times: [hn 4.04 s epoch 54.51 s val 0.00 s] lr = 0.050000 loss = -0.00608483 = 0.00130564 + lam * -3.69524, offending 29312
  Forward pass
  Distances
  Train
epoch 117, times: [hn 4.06 s epoch 57.79 s val 0.00 s] lr = 0.050000 loss = -0.00621963 = 0.00119549 + lam * -3.70756, offending 26979
  Forward pass
  Distances
  Train
epoch 118, times: [hn 3.98 s epoch 56.98 s val 0.00 s] lr = 0.050000 loss = -0.00606324 = 0.00133146 + lam * -3.69735, offending 29568
  Forward pass
  Distances
  Train
Valiation at epoch 119
zn_3     nbit= 14:  0.0002 0.0022 0.0163
zn_10    nbit= 32:  0.0101 0.0620 0.2479
zn_79    nbit= 64:  0.1375 0.4776 0.8624
storing test_ckpt/0.002/checkpoint.pth
epoch 119, times: [hn 4.07 s epoch 55.87 s val 9.86 s] lr = 0.050000 loss = -0.00615582 = 0.00124113 + lam * -3.69848, offending 28059
  Forward pass
  Distances
  Train
epoch 120, times: [hn 4.12 s epoch 56.51 s val 0.00 s] lr = 0.010000 loss = -0.00620937 = 0.00121979 + lam * -3.71458, offending 27619
  Forward pass
  Distances
  Train
epoch 121, times: [hn 4.10 s epoch 51.83 s val 0.00 s] lr = 0.010000 loss = -0.00612991 = 0.00129794 + lam * -3.71393, offending 29406
  Forward pass
  Distances
  Train
epoch 122, times: [hn 4.08 s epoch 53.46 s val 0.00 s] lr = 0.010000 loss = -0.00635007 = 0.00110683 + lam * -3.72845, offending 25628
  Forward pass
  Distances
  Train
epoch 123, times: [hn 4.07 s epoch 57.99 s val 0.00 s] lr = 0.010000 loss = -0.00651013 = 0.000980722 + lam * -3.74542, offending 22405
  Forward pass
  Distances
  Train
epoch 124, times: [hn 4.01 s epoch 50.90 s val 0.00 s] lr = 0.010000 loss = -0.00640229 = 0.00110003 + lam * -3.75116, offending 24772
  Forward pass
  Distances
  Train
epoch 125, times: [hn 4.07 s epoch 57.82 s val 0.00 s] lr = 0.010000 loss = -0.00637417 = 0.00112786 + lam * -3.75102, offending 25414
  Forward pass
  Distances
  Train
epoch 126, times: [hn 4.18 s epoch 61.00 s val 0.00 s] lr = 0.010000 loss = -0.00608641 = 0.00138701 + lam * -3.73671, offending 30970
  Forward pass
  Distances
  Train
epoch 127, times: [hn 4.04 s epoch 57.10 s val 0.00 s] lr = 0.010000 loss = -0.00643233 = 0.00105242 + lam * -3.74238, offending 23974
  Forward pass
  Distances
  Train
epoch 128, times: [hn 4.06 s epoch 54.03 s val 0.00 s] lr = 0.010000 loss = -0.00614616 = 0.00133237 + lam * -3.73927, offending 29505
  Forward pass
  Distances
  Train
Valiation at epoch 129
zn_3     nbit= 14:  0.0004 0.0025 0.0198
zn_10    nbit= 32:  0.0122 0.0748 0.2890
zn_79    nbit= 64:  0.1525 0.5100 0.8828
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.51001 > 0.50593), keeping as best
epoch 129, times: [hn 4.18 s epoch 56.99 s val 9.89 s] lr = 0.010000 loss = -0.00631758 = 0.00115957 + lam * -3.73857, offending 26593
  Forward pass
  Distances
  Train
epoch 130, times: [hn 4.12 s epoch 58.44 s val 0.00 s] lr = 0.010000 loss = -0.00643715 = 0.00106545 + lam * -3.7513, offending 24393
  Forward pass
  Distances
  Train
epoch 131, times: [hn 4.22 s epoch 57.05 s val 0.00 s] lr = 0.010000 loss = -0.00636862 = 0.00113303 + lam * -3.75083, offending 25679
  Forward pass
  Distances
  Train
epoch 132, times: [hn 4.02 s epoch 56.18 s val 0.00 s] lr = 0.010000 loss = -0.00648763 = 0.00102723 + lam * -3.75743, offending 23374
  Forward pass
  Distances
  Train
epoch 133, times: [hn 4.05 s epoch 57.41 s val 0.00 s] lr = 0.010000 loss = -0.00633728 = 0.00116908 + lam * -3.75318, offending 26250
  Forward pass
  Distances
  Train
epoch 134, times: [hn 4.05 s epoch 56.09 s val 0.00 s] lr = 0.010000 loss = -0.00661446 = 0.000915831 + lam * -3.76514, offending 21085
  Forward pass
  Distances
  Train
epoch 135, times: [hn 4.15 s epoch 55.64 s val 0.00 s] lr = 0.010000 loss = -0.00635245 = 0.00116511 + lam * -3.75878, offending 26212
  Forward pass
  Distances
  Train
epoch 136, times: [hn 4.07 s epoch 56.19 s val 0.00 s] lr = 0.010000 loss = -0.00607127 = 0.00140903 + lam * -3.74015, offending 31056
  Forward pass
  Distances
  Train
epoch 137, times: [hn 4.06 s epoch 55.96 s val 0.00 s] lr = 0.010000 loss = -0.0061615 = 0.0013 + lam * -3.73075, offending 29325
  Forward pass
  Distances
  Train
epoch 138, times: [hn 4.06 s epoch 55.28 s val 0.00 s] lr = 0.010000 loss = -0.00641148 = 0.00108471 + lam * -3.7481, offending 24611
  Forward pass
  Distances
  Train
Valiation at epoch 139
zn_3     nbit= 14:  0.0003 0.0022 0.0186
zn_10    nbit= 32:  0.0118 0.0680 0.2676
zn_79    nbit= 64:  0.1459 0.4971 0.8737
storing test_ckpt/0.002/checkpoint.pth
epoch 139, times: [hn 4.07 s epoch 58.85 s val 9.73 s] lr = 0.010000 loss = -0.00630257 = 0.00119831 + lam * -3.75044, offending 26603
  Forward pass
  Distances
  Train
epoch 140, times: [hn 3.98 s epoch 54.03 s val 0.00 s] lr = 0.010000 loss = -0.00639225 = 0.00110789 + lam * -3.75007, offending 25256
  Forward pass
  Distances
  Train
epoch 141, times: [hn 4.14 s epoch 57.98 s val 0.00 s] lr = 0.010000 loss = -0.00652988 = 0.000999587 + lam * -3.76473, offending 22482
  Forward pass
  Distances
  Train
epoch 142, times: [hn 4.07 s epoch 62.01 s val 0.00 s] lr = 0.010000 loss = -0.0064746 = 0.00105973 + lam * -3.76716, offending 23636
  Forward pass
  Distances
  Train
epoch 143, times: [hn 4.00 s epoch 53.64 s val 0.00 s] lr = 0.010000 loss = -0.00642473 = 0.00111079 + lam * -3.76776, offending 24527
  Forward pass
  Distances
  Train
epoch 144, times: [hn 4.06 s epoch 60.67 s val 0.00 s] lr = 0.010000 loss = -0.00622191 = 0.00128589 + lam * -3.7539, offending 28344
  Forward pass
  Distances
  Train
epoch 145, times: [hn 4.05 s epoch 47.31 s val 0.00 s] lr = 0.010000 loss = -0.00633939 = 0.00115966 + lam * -3.74952, offending 26226
  Forward pass
  Distances
  Train
epoch 146, times: [hn 4.16 s epoch 59.61 s val 0.00 s] lr = 0.010000 loss = -0.00641873 = 0.00109728 + lam * -3.758, offending 24493
  Forward pass
  Distances
  Train
epoch 147, times: [hn 3.96 s epoch 59.24 s val 0.00 s] lr = 0.010000 loss = -0.00657624 = 0.000962249 + lam * -3.76925, offending 21681
  Forward pass
  Distances
  Train
epoch 148, times: [hn 4.05 s epoch 58.02 s val 0.00 s] lr = 0.010000 loss = -0.00654912 = 0.00100172 + lam * -3.77542, offending 22295
  Forward pass
  Distances
  Train
Valiation at epoch 149
zn_3     nbit= 14:  0.0002 0.0018 0.0165
zn_10    nbit= 32:  0.0105 0.0605 0.2451
zn_79    nbit= 64:  0.1389 0.4790 0.8631
storing test_ckpt/0.002/checkpoint.pth
epoch 149, times: [hn 4.04 s epoch 60.43 s val 9.54 s] lr = 0.010000 loss = -0.0062109 = 0.00130289 + lam * -3.7569, offending 28609
  Forward pass
  Distances
  Train
epoch 150, times: [hn 4.11 s epoch 52.77 s val 0.00 s] lr = 0.010000 loss = -0.00654437 = 0.000979608 + lam * -3.76199, offending 22161
  Forward pass
  Distances
  Train
epoch 151, times: [hn 4.14 s epoch 56.67 s val 0.00 s] lr = 0.010000 loss = -0.00625562 = 0.00125927 + lam * -3.75744, offending 27831
  Forward pass
  Distances
  Train
epoch 152, times: [hn 4.05 s epoch 54.43 s val 0.00 s] lr = 0.010000 loss = -0.00632093 = 0.00118031 + lam * -3.75062, offending 26476
  Forward pass
  Distances
  Train
epoch 153, times: [hn 4.05 s epoch 57.61 s val 0.00 s] lr = 0.010000 loss = -0.00648608 = 0.00103222 + lam * -3.75915, offending 23403
  Forward pass
  Distances
  Train
epoch 154, times: [hn 4.06 s epoch 54.49 s val 0.00 s] lr = 0.010000 loss = -0.00630851 = 0.00120605 + lam * -3.75728, offending 26753
  Forward pass
  Distances
  Train
^[epoch 155, times: [hn 4.05 s epoch 53.96 s val 0.00 s] lr = 0.010000 loss = -0.0065594 = 0.000971222 + lam * -3.76531, offending 21927
  Forward pass
  Distances
  Train
epoch 156, times: [hn 4.19 s epoch 57.83 s val 0.00 s] lr = 0.010000 loss = -0.00645651 = 0.00107741 + lam * -3.76696, offending 23978
  Forward pass
  Distances
  Train
epoch 157, times: [hn 4.07 s epoch 56.87 s val 0.00 s] lr = 0.010000 loss = -0.00638045 = 0.00115819 + lam * -3.76932, offending 25575
  Forward pass
  Distances
  Train
epoch 158, times: [hn 4.05 s epoch 52.18 s val 0.00 s] lr = 0.010000 loss = -0.00609558 = 0.0013848 + lam * -3.74019, offending 30580
  Forward pass
  Distances
  Train
Valiation at epoch 159
zn_3     nbit= 14:  0.0003 0.0028 0.0227
zn_10    nbit= 32:  0.0141 0.0802 0.3044
zn_79    nbit= 64:  0.1579 0.5199 0.8892
storing test_ckpt/0.002/checkpoint.pth
zn_79,rank=10 score improves (0.51987 > 0.51001), keeping as best
epoch 159, times: [hn 4.03 s epoch 46.46 s val 9.51 s] lr = 0.010000 loss = -0.00653702 = 0.000979996 + lam * -3.75851, offending 22233
++ tac test_ckpt/0.002.stdout
++ grep -m1 'keeping as best'
++ grep -o '[0-9\.]*'
++ grep -o '(.*>'
+ perf=0.51987
+ echo 0.51987
0.51987
+ [[ 0.51987 > 0.35615 ]]
+ best_perf=0.51987
+ best_lambda=0.002
+ for lambda in '$lambdas'
+ mkdir -p test_ckpt/0.005
+ python -u train.py --dout 24 --save_best_criterion zn_79,rank=10 --database deep1b --lambda_uniform 0.005 --checkpoint_dir test_ckpt/0.005
+ tee test_ckpt/0.005.stdout
Namespace(batch_size=64, checkpoint_dir='test_ckpt/0.005', database='deep1b', device='cuda', dint=1024, dout=24, epochs=160, init_name='', lambda_uniform=0.00
5, lr_schedule='0.1,0.1,0.05,0.01', momentum=0.9, num_learn=500000, quantizer_train='', rank_negative=50, rank_positive=10, save_best_criterion='zn_79,rank=10
', seed=1234, size_base=1000000, val_freq=10, validation_quantizers=['zn_3', 'zn_10', 'zn_79'])
load dataset deep1b
keeping 500000/357380000 training vectors
computing training ground truth
build network
Lr schedule [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1,
0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1
, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0
.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0
5, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01,
 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01]
  Forward pass
  Distances
  Train
epoch 0, times: [hn 3.95 s epoch 52.51 s val 0.00 s] lr = 0.100000 loss = -0.0180558 = 0.00226546 + lam * -4.06425, offending 17650
  Forward pass
  Distances
  Train
epoch 1, times: [hn 4.04 s epoch 58.18 s val 0.00 s] lr = 0.100000 loss = -0.0140318 = 0.00505507 + lam * -3.81737, offending 54327
  Forward pass
  Distances
  Train
epoch 2, times: [hn 4.08 s epoch 66.62 s val 0.00 s] lr = 0.100000 loss = -0.0140917 = 0.00509273 + lam * -3.8369, offending 59154
  Forward pass
  Distances
  Train
epoch 3, times: [hn 4.07 s epoch 51.25 s val 0.00 s] lr = 0.100000 loss = -0.0137552 = 0.00535991 + lam * -3.82303, offending 67562
  Forward pass
  Distances
  Train
epoch 4, times: [hn 4.06 s epoch 66.69 s val 0.00 s] lr = 0.100000 loss = -0.0138022 = 0.00536243 + lam * -3.83293, offending 70665
  Forward pass
  Distances
  Train
epoch 5, times: [hn 4.06 s epoch 43.23 s val 0.00 s] lr = 0.100000 loss = -0.0140113 = 0.00531436 + lam * -3.86513, offending 71385
  Forward pass
  Distances
  Train
epoch 6, times: [hn 4.05 s epoch 42.23 s val 0.00 s] lr = 0.100000 loss = -0.0140278 = 0.00543043 + lam * -3.89165, offending 73061
  Forward pass
  Distances
  Train
epoch 7, times: [hn 4.02 s epoch 41.70 s val 0.00 s] lr = 0.100000 loss = -0.0141252 = 0.0054228 + lam * -3.90961, offending 73150
  Forward pass
  Distances
  Train
epoch 8, times: [hn 4.06 s epoch 44.04 s val 0.00 s] lr = 0.100000 loss = -0.0142122 = 0.00546745 + lam * -3.93592, offending 73495
  Forward pass
  Distances
  Train
Valiation at epoch 9
zn_3     nbit= 14:  0.0009 0.0073 0.0602
zn_10    nbit= 32:  0.0281 0.1404 0.4472
zn_79    nbit= 64:  0.1742 0.5569 0.9080
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.55693 > 0), keeping as best
epoch 9, times: [hn 4.03 s epoch 41.88 s val 10.91 s] lr = 0.100000 loss = -0.0143298 = 0.00545231 + lam * -3.95643, offending 73231
  Forward pass
  Distances
  Train
epoch 10, times: [hn 4.14 s epoch 46.31 s val 0.00 s] lr = 0.100000 loss = -0.0143805 = 0.00547891 + lam * -3.97189, offending 72755
  Forward pass
  Distances
  Train
epoch 11, times: [hn 3.99 s epoch 44.65 s val 0.00 s] lr = 0.100000 loss = -0.014532 = 0.00543273 + lam * -3.99296, offending 71932
  Forward pass
  Distances
  Train
epoch 12, times: [hn 4.03 s epoch 42.86 s val 0.00 s] lr = 0.100000 loss = -0.0146589 = 0.00539247 + lam * -4.01027, offending 70754
  Forward pass
  Distances
  Train
epoch 13, times: [hn 4.06 s epoch 44.32 s val 0.00 s] lr = 0.100000 loss = -0.0147005 = 0.00538667 + lam * -4.01744, offending 71115
  Forward pass
  Distances
  Train
epoch 14, times: [hn 4.05 s epoch 43.82 s val 0.00 s] lr = 0.100000 loss = -0.014768 = 0.00534876 + lam * -4.02335, offending 70555
  Forward pass
  Distances
  Train
epoch 15, times: [hn 4.08 s epoch 45.60 s val 0.00 s] lr = 0.100000 loss = -0.01487 = 0.00524919 + lam * -4.02385, offending 69493
  Forward pass
  Distances
  Train
epoch 16, times: [hn 4.05 s epoch 43.56 s val 0.00 s] lr = 0.100000 loss = -0.0149373 = 0.00519528 + lam * -4.02651, offending 69360
  Forward pass
  Distances
  Train
epoch 17, times: [hn 4.05 s epoch 45.37 s val 0.00 s] lr = 0.100000 loss = -0.0149514 = 0.00522655 + lam * -4.0356, offending 68804
  Forward pass
  Distances
  Train
epoch 18, times: [hn 4.03 s epoch 41.76 s val 0.00 s] lr = 0.100000 loss = -0.0150749 = 0.00512665 + lam * -4.04032, offending 68651
  Forward pass
  Distances
  Train
Valiation at epoch 19
zn_3     nbit= 14:  0.0013 0.0098 0.0763
zn_10    nbit= 32:  0.0365 0.1739 0.5177
zn_79    nbit= 64:  0.1963 0.6049 0.9319
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.60493 > 0.55693), keeping as best
epoch 19, times: [hn 4.08 s epoch 42.57 s val 10.97 s] lr = 0.100000 loss = -0.0150737 = 0.0051298 + lam * -4.0407, offending 68158
  Forward pass
  Distances
  Train
epoch 20, times: [hn 4.15 s epoch 44.29 s val 0.00 s] lr = 0.100000 loss = -0.0151878 = 0.00501204 + lam * -4.03997, offending 67199
  Forward pass
  Distances
  Train
epoch 21, times: [hn 4.04 s epoch 40.38 s val 0.00 s] lr = 0.100000 loss = -0.0152632 = 0.0049683 + lam * -4.0463, offending 66918
  Forward pass
  Distances
  Train
epoch 22, times: [hn 4.06 s epoch 40.67 s val 0.00 s] lr = 0.100000 loss = -0.0152438 = 0.00499396 + lam * -4.04755, offending 66780
  Forward pass
  Distances
  Train
epoch 23, times: [hn 4.06 s epoch 40.97 s val 0.00 s] lr = 0.100000 loss = -0.015273 = 0.00497506 + lam * -4.04961, offending 66891
  Forward pass
  Distances
  Train
epoch 24, times: [hn 4.08 s epoch 40.05 s val 0.00 s] lr = 0.100000 loss = -0.0153637 = 0.00489422 + lam * -4.05159, offending 66250
  Forward pass
  Distances
  Train
epoch 25, times: [hn 3.96 s epoch 41.26 s val 0.00 s] lr = 0.100000 loss = -0.0154045 = 0.00485726 + lam * -4.05235, offending 65479
  Forward pass
  Distances
  Train
epoch 26, times: [hn 4.05 s epoch 41.16 s val 0.00 s] lr = 0.100000 loss = -0.0154195 = 0.00483691 + lam * -4.05127, offending 65825
  Forward pass
  Distances
  Train
epoch 27, times: [hn 3.97 s epoch 40.21 s val 0.00 s] lr = 0.100000 loss = -0.0154818 = 0.00478956 + lam * -4.05427, offending 65243
  Forward pass
  Distances
  Train
epoch 28, times: [hn 4.03 s epoch 40.11 s val 0.00 s] lr = 0.100000 loss = -0.0155468 = 0.00473938 + lam * -4.05723, offending 64396
  Forward pass
  Distances
  Train
Valiation at epoch 29
zn_3     nbit= 14:  0.0011 0.0104 0.0795
zn_10    nbit= 32:  0.0386 0.1847 0.5374
zn_79    nbit= 64:  0.2057 0.6216 0.9384
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.62163 > 0.60493), keeping as best
epoch 29, times: [hn 4.06 s epoch 40.06 s val 10.82 s] lr = 0.100000 loss = -0.015551 = 0.00473838 + lam * -4.05788, offending 64703
  Forward pass
  Distances
  Train
epoch 30, times: [hn 4.12 s epoch 40.51 s val 0.00 s] lr = 0.100000 loss = -0.0155334 = 0.00475537 + lam * -4.05775, offending 64452
  Forward pass
  Distances
  Train
epoch 31, times: [hn 4.04 s epoch 40.53 s val 0.00 s] lr = 0.100000 loss = -0.0155722 = 0.00472968 + lam * -4.06037, offending 64535
  Forward pass
  Distances
  Train
epoch 32, times: [hn 3.98 s epoch 40.42 s val 0.00 s] lr = 0.100000 loss = -0.0156269 = 0.00467956 + lam * -4.06129, offending 64179
  Forward pass
  Distances
  Train
epoch 33, times: [hn 4.02 s epoch 40.68 s val 0.00 s] lr = 0.100000 loss = -0.0157135 = 0.00459533 + lam * -4.06177, offending 63144
  Forward pass
  Distances
  Train
epoch 34, times: [hn 4.10 s epoch 41.03 s val 0.00 s] lr = 0.100000 loss = -0.0156518 = 0.0046648 + lam * -4.06332, offending 63582
  Forward pass
  Distances
  Train
epoch 35, times: [hn 4.06 s epoch 41.18 s val 0.00 s] lr = 0.100000 loss = -0.0157496 = 0.00457196 + lam * -4.06432, offending 62559
  Forward pass
  Distances
  Train
epoch 36, times: [hn 3.97 s epoch 40.53 s val 0.00 s] lr = 0.100000 loss = -0.0157298 = 0.0045808 + lam * -4.06213, offending 63272
  Forward pass
  Distances
  Train
epoch 37, times: [hn 4.09 s epoch 40.46 s val 0.00 s] lr = 0.100000 loss = -0.0157821 = 0.00454209 + lam * -4.06484, offending 62841
  Forward pass
  Distances
  Train
epoch 38, times: [hn 4.03 s epoch 40.01 s val 0.00 s] lr = 0.100000 loss = -0.0157302 = 0.00459537 + lam * -4.06512, offending 63516
  Forward pass
  Distances
  Train
Valiation at epoch 39
zn_3     nbit= 14:  0.0010 0.0105 0.0805
zn_10    nbit= 32:  0.0399 0.1899 0.5466
zn_79    nbit= 64:  0.2070 0.6300 0.9428
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.62998 > 0.62163), keeping as best
epoch 39, times: [hn 4.03 s epoch 40.00 s val 10.62 s] lr = 0.100000 loss = -0.0158331 = 0.00449689 + lam * -4.066, offending 62095
  Forward pass
  Distances
  Train
epoch 40, times: [hn 4.16 s epoch 40.52 s val 0.00 s] lr = 0.100000 loss = -0.0157727 = 0.00455143 + lam * -4.06482, offending 63135
  Forward pass
  Distances
  Train
epoch 41, times: [hn 4.03 s epoch 40.03 s val 0.00 s] lr = 0.100000 loss = -0.0158291 = 0.00450357 + lam * -4.06654, offending 62302
  Forward pass
  Distances
  Train
epoch 42, times: [hn 4.05 s epoch 40.63 s val 0.00 s] lr = 0.100000 loss = -0.0159174 = 0.00441564 + lam * -4.0666, offending 61446
  Forward pass
  Distances
  Train
epoch 43, times: [hn 3.98 s epoch 40.00 s val 0.00 s] lr = 0.100000 loss = -0.0159043 = 0.00443549 + lam * -4.06796, offending 61476
  Forward pass
  Distances
  Train
epoch 44, times: [hn 4.05 s epoch 39.81 s val 0.00 s] lr = 0.100000 loss = -0.0158624 = 0.00447771 + lam * -4.06802, offending 61647
  Forward pass
  Distances
  Train
epoch 45, times: [hn 4.04 s epoch 40.73 s val 0.00 s] lr = 0.100000 loss = -0.0159551 = 0.00438081 + lam * -4.06718, offending 61281
  Forward pass
  Distances
  Train
epoch 46, times: [hn 4.06 s epoch 40.45 s val 0.00 s] lr = 0.100000 loss = -0.0159456 = 0.00440497 + lam * -4.07012, offending 61425
  Forward pass
  Distances
  Train
epoch 47, times: [hn 4.03 s epoch 39.59 s val 0.00 s] lr = 0.100000 loss = -0.0159837 = 0.00436172 + lam * -4.06908, offending 61441
  Forward pass
  Distances
  Train
epoch 48, times: [hn 4.03 s epoch 40.65 s val 0.00 s] lr = 0.100000 loss = -0.015986 = 0.00435613 + lam * -4.06842, offending 60747
  Forward pass
  Distances
  Train
Valiation at epoch 49
zn_3     nbit= 14:  0.0009 0.0108 0.0842
zn_10    nbit= 32:  0.0413 0.1906 0.5467
zn_79    nbit= 64:  0.2094 0.6321 0.9445
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.6321 > 0.62998), keeping as best
epoch 49, times: [hn 4.06 s epoch 40.76 s val 10.60 s] lr = 0.100000 loss = -0.0160036 = 0.00435051 + lam * -4.07081, offending 61152
  Forward pass
  Distances
  Train
epoch 50, times: [hn 4.14 s epoch 40.80 s val 0.00 s] lr = 0.100000 loss = -0.0160032 = 0.00435018 + lam * -4.07067, offending 60626
  Forward pass
  Distances
  Train
epoch 51, times: [hn 4.04 s epoch 40.66 s val 0.00 s] lr = 0.100000 loss = -0.0160324 = 0.00432068 + lam * -4.07061, offending 60349
  Forward pass
  Distances
  Train
epoch 52, times: [hn 4.02 s epoch 40.72 s val 0.00 s] lr = 0.100000 loss = -0.0160975 = 0.00426154 + lam * -4.0718, offending 60084
  Forward pass
  Distances
  Train
epoch 53, times: [hn 4.02 s epoch 39.94 s val 0.00 s] lr = 0.100000 loss = -0.016071 = 0.00428837 + lam * -4.07187, offending 60509
  Forward pass
  Distances
  Train
epoch 54, times: [hn 3.97 s epoch 40.31 s val 0.00 s] lr = 0.100000 loss = -0.0160617 = 0.0042978 + lam * -4.0719, offending 60562
  Forward pass
  Distances
  Train
epoch 55, times: [hn 3.99 s epoch 40.06 s val 0.00 s] lr = 0.100000 loss = -0.0160994 = 0.00426249 + lam * -4.07238, offending 60171
  Forward pass
  Distances
  Train
epoch 56, times: [hn 4.04 s epoch 39.68 s val 0.00 s] lr = 0.100000 loss = -0.0161142 = 0.00425633 + lam * -4.0741, offending 60075
  Forward pass
  Distances
  Train
epoch 57, times: [hn 4.02 s epoch 40.32 s val 0.00 s] lr = 0.100000 loss = -0.0160886 = 0.00427724 + lam * -4.07317, offending 60508
  Forward pass
  Distances
  Train
epoch 58, times: [hn 3.99 s epoch 40.53 s val 0.00 s] lr = 0.100000 loss = -0.0161138 = 0.0042671 + lam * -4.07617, offending 60263
  Forward pass
  Distances
  Train
Valiation at epoch 59
zn_3     nbit= 14:  0.0013 0.0118 0.0854
zn_10    nbit= 32:  0.0416 0.1966 0.5580
zn_79    nbit= 64:  0.2128 0.6405 0.9467
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.64055 > 0.6321), keeping as best
epoch 59, times: [hn 4.07 s epoch 42.14 s val 9.85 s] lr = 0.100000 loss = -0.016122 = 0.00424173 + lam * -4.07274, offending 60020
  Forward pass
  Distances
  Train
epoch 60, times: [hn 4.14 s epoch 40.69 s val 0.00 s] lr = 0.100000 loss = -0.0161859 = 0.00419148 + lam * -4.07548, offending 59430
  Forward pass
  Distances
  Train
epoch 61, times: [hn 4.05 s epoch 39.70 s val 0.00 s] lr = 0.100000 loss = -0.0162231 = 0.00416026 + lam * -4.07666, offending 59437
  Forward pass
  Distances
  Train
epoch 62, times: [hn 3.98 s epoch 40.63 s val 0.00 s] lr = 0.100000 loss = -0.0161801 = 0.00419963 + lam * -4.07594, offending 59543
  Forward pass
  Distances
  Train
epoch 63, times: [hn 4.07 s epoch 40.24 s val 0.00 s] lr = 0.100000 loss = -0.0162103 = 0.00416134 + lam * -4.07432, offending 59286
  Forward pass
  Distances
  Train
epoch 64, times: [hn 3.97 s epoch 39.86 s val 0.00 s] lr = 0.100000 loss = -0.0162116 = 0.00415961 + lam * -4.07424, offending 59247
  Forward pass
  Distances
  Train
epoch 65, times: [hn 4.06 s epoch 39.87 s val 0.00 s] lr = 0.100000 loss = -0.0162039 = 0.00416659 + lam * -4.0741, offending 59481
  Forward pass
  Distances
  Train
epoch 66, times: [hn 4.01 s epoch 39.97 s val 0.00 s] lr = 0.100000 loss = -0.0162608 = 0.00412545 + lam * -4.07726, offending 59171
  Forward pass
  Distances
  Train
epoch 67, times: [hn 4.07 s epoch 40.11 s val 0.00 s] lr = 0.100000 loss = -0.0162673 = 0.00410937 + lam * -4.07534, offending 58815
  Forward pass
  Distances
  Train
epoch 68, times: [hn 4.11 s epoch 41.19 s val 0.00 s] lr = 0.100000 loss = -0.016291 = 0.004097 + lam * -4.07761, offending 58876
  Forward pass
  Distances
  Train
Valiation at epoch 69
zn_3     nbit= 14:  0.0011 0.0109 0.0853
zn_10    nbit= 32:  0.0433 0.2001 0.5641
zn_79    nbit= 64:  0.2135 0.6405 0.9492
storing test_ckpt/0.005/checkpoint.pth
epoch 69, times: [hn 4.03 s epoch 40.06 s val 10.56 s] lr = 0.100000 loss = -0.0162792 = 0.00410779 + lam * -4.07739, offending 58816
  Forward pass
  Distances
  Train
epoch 70, times: [hn 4.18 s epoch 40.93 s val 0.00 s] lr = 0.100000 loss = -0.0162697 = 0.0041176 + lam * -4.07745, offending 59267
  Forward pass
  Distances
  Train
epoch 71, times: [hn 3.97 s epoch 40.45 s val 0.00 s] lr = 0.100000 loss = -0.0162862 = 0.00410503 + lam * -4.07824, offending 58541
  Forward pass
  Distances
  Train
epoch 72, times: [hn 4.09 s epoch 40.35 s val 0.00 s] lr = 0.100000 loss = -0.0163177 = 0.00406657 + lam * -4.07686, offending 58673
  Forward pass
  Distances
  Train
epoch 73, times: [hn 4.07 s epoch 40.10 s val 0.00 s] lr = 0.100000 loss = -0.0163061 = 0.0040902 + lam * -4.07926, offending 58544
  Forward pass
  Distances
  Train
epoch 74, times: [hn 3.99 s epoch 40.36 s val 0.00 s] lr = 0.100000 loss = -0.0163304 = 0.00405687 + lam * -4.07746, offending 58237
  Forward pass
  Distances
  Train
epoch 75, times: [hn 4.05 s epoch 41.17 s val 0.00 s] lr = 0.100000 loss = -0.0163095 = 0.00408537 + lam * -4.07896, offending 58673
  Forward pass
  Distances
  Train
epoch 76, times: [hn 4.05 s epoch 39.91 s val 0.00 s] lr = 0.100000 loss = -0.0163576 = 0.00403597 + lam * -4.07872, offending 58275
  Forward pass
  Distances
  Train
epoch 77, times: [hn 4.04 s epoch 40.21 s val 0.00 s] lr = 0.100000 loss = -0.016354 = 0.00404342 + lam * -4.07949, offending 58263
  Forward pass
  Distances
  Train
epoch 78, times: [hn 4.04 s epoch 40.25 s val 0.00 s] lr = 0.100000 loss = -0.0163281 = 0.00406837 + lam * -4.07929, offending 58549
  Forward pass
  Distances
  Train
Valiation at epoch 79
zn_3     nbit= 14:  0.0012 0.0110 0.0849
zn_10    nbit= 32:  0.0412 0.1992 0.5635
zn_79    nbit= 64:  0.2155 0.6448 0.9486
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.64481 > 0.64055), keeping as best
epoch 79, times: [hn 4.04 s epoch 40.13 s val 10.63 s] lr = 0.100000 loss = -0.0163266 = 0.00405993 + lam * -4.07732, offending 57961
  Forward pass
  Distances
  Train
epoch 80, times: [hn 4.12 s epoch 41.02 s val 0.00 s] lr = 0.050000 loss = -0.0165108 = 0.00388988 + lam * -4.08014, offending 57534
  Forward pass
  Distances
  Train
epoch 81, times: [hn 3.97 s epoch 40.13 s val 0.00 s] lr = 0.050000 loss = -0.0165803 = 0.00382956 + lam * -4.08197, offending 56748
  Forward pass
  Distances
  Train
epoch 82, times: [hn 4.04 s epoch 41.14 s val 0.00 s] lr = 0.050000 loss = -0.0166099 = 0.00380189 + lam * -4.08236, offending 56266
  Forward pass
  Distances
  Train
epoch 83, times: [hn 4.04 s epoch 40.46 s val 0.00 s] lr = 0.050000 loss = -0.0165969 = 0.00381296 + lam * -4.08197, offending 56866
  Forward pass
  Distances
  Train
epoch 84, times: [hn 4.03 s epoch 41.54 s val 0.00 s] lr = 0.050000 loss = -0.0165904 = 0.00382221 + lam * -4.08252, offending 56571
  Forward pass
  Distances
  Train
epoch 85, times: [hn 4.05 s epoch 45.30 s val 0.00 s] lr = 0.050000 loss = -0.0167131 = 0.00370831 + lam * -4.08428, offending 55142
  Forward pass
  Distances
  Train
epoch 86, times: [hn 4.12 s epoch 40.34 s val 0.00 s] lr = 0.050000 loss = -0.0165913 = 0.00381963 + lam * -4.08218, offending 56511
  Forward pass
  Distances
  Train
epoch 87, times: [hn 4.06 s epoch 41.27 s val 0.00 s] lr = 0.050000 loss = -0.0165874 = 0.00382638 + lam * -4.08275, offending 56616
  Forward pass
  Distances
  Train
epoch 88, times: [hn 4.05 s epoch 39.90 s val 0.00 s] lr = 0.050000 loss = -0.0166473 = 0.00377137 + lam * -4.08374, offending 55486
  Forward pass
  Distances
  Train
Valiation at epoch 89
zn_3     nbit= 14:  0.0011 0.0105 0.0844
zn_10    nbit= 32:  0.0422 0.2027 0.5673
zn_79    nbit= 64:  0.2164 0.6472 0.9500
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.64722 > 0.64481), keeping as best
epoch 89, times: [hn 4.03 s epoch 40.67 s val 9.40 s] lr = 0.050000 loss = -0.0166691 = 0.00374994 + lam * -4.08381, offending 55855
  Forward pass
  Distances
  Train
epoch 90, times: [hn 4.26 s epoch 40.59 s val 0.00 s] lr = 0.050000 loss = -0.0166699 = 0.00374633 + lam * -4.08324, offending 55663
  Forward pass
  Distances
  Train
epoch 91, times: [hn 4.15 s epoch 40.89 s val 0.00 s] lr = 0.050000 loss = -0.0166935 = 0.00372486 + lam * -4.08367, offending 55156
  Forward pass
  Distances
  Train
epoch 92, times: [hn 4.03 s epoch 41.62 s val 0.00 s] lr = 0.050000 loss = -0.0166716 = 0.00375768 + lam * -4.08585, offending 55578
  Forward pass
  Distances
  Train
epoch 93, times: [hn 4.10 s epoch 41.20 s val 0.00 s] lr = 0.050000 loss = -0.0167119 = 0.00370683 + lam * -4.08374, offending 55139
  Forward pass
  Distances
  Train
epoch 94, times: [hn 4.01 s epoch 41.00 s val 0.00 s] lr = 0.050000 loss = -0.0167675 = 0.0036588 + lam * -4.08526, offending 54920
  Forward pass
  Distances
  Train
epoch 95, times: [hn 4.03 s epoch 41.08 s val 0.00 s] lr = 0.050000 loss = -0.0167079 = 0.00371304 + lam * -4.08418, offending 55681
  Forward pass
  Distances
  Train
epoch 96, times: [hn 4.05 s epoch 41.15 s val 0.00 s] lr = 0.050000 loss = -0.0167219 = 0.00370283 + lam * -4.08494, offending 55508
  Forward pass
  Distances
  Train
epoch 97, times: [hn 4.04 s epoch 41.00 s val 0.00 s] lr = 0.050000 loss = -0.0167216 = 0.00369602 + lam * -4.08352, offending 55590
  Forward pass
  Distances
  Train
epoch 98, times: [hn 4.03 s epoch 41.28 s val 0.00 s] lr = 0.050000 loss = -0.0167302 = 0.00369725 + lam * -4.0855, offending 55276
  Forward pass
  Distances
  Train
Valiation at epoch 99
zn_3     nbit= 14:  0.0011 0.0113 0.0877
zn_10    nbit= 32:  0.0449 0.2098 0.5805
zn_79    nbit= 64:  0.2203 0.6539 0.9526
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.65387 > 0.64722), keeping as best
epoch 99, times: [hn 4.06 s epoch 41.33 s val 9.76 s] lr = 0.050000 loss = -0.0167615 = 0.00366479 + lam * -4.08525, offending 54814
  Forward pass
  Distances
  Train
epoch 100, times: [hn 4.15 s epoch 40.36 s val 0.00 s] lr = 0.050000 loss = -0.0166792 = 0.00374667 + lam * -4.08518, offending 55685
  Forward pass
  Distances
  Train
epoch 101, times: [hn 4.04 s epoch 41.24 s val 0.00 s] lr = 0.050000 loss = -0.0167449 = 0.00367613 + lam * -4.08421, offending 55196
  Forward pass
  Distances
  Train
epoch 102, times: [hn 4.05 s epoch 41.02 s val 0.00 s] lr = 0.050000 loss = -0.0167688 = 0.0036576 + lam * -4.08527, offending 54967
  Forward pass
  Distances
  Train
epoch 103, times: [hn 4.04 s epoch 40.37 s val 0.00 s] lr = 0.050000 loss = -0.0167627 = 0.00366472 + lam * -4.08549, offending 55105
  Forward pass
  Distances
  Train
epoch 104, times: [hn 4.06 s epoch 41.01 s val 0.00 s] lr = 0.050000 loss = -0.0167535 = 0.0036749 + lam * -4.08568, offending 55195
  Forward pass
  Distances
  Train
epoch 105, times: [hn 4.12 s epoch 41.00 s val 0.00 s] lr = 0.050000 loss = -0.0167548 = 0.00367094 + lam * -4.08515, offending 55058
  Forward pass
  Distances
  Train
epoch 106, times: [hn 4.02 s epoch 40.75 s val 0.00 s] lr = 0.050000 loss = -0.0167978 = 0.00362827 + lam * -4.08522, offending 54700
  Forward pass
  Distances
  Train
epoch 107, times: [hn 4.04 s epoch 40.39 s val 0.00 s] lr = 0.050000 loss = -0.0167725 = 0.00365769 + lam * -4.08603, offending 55156
  Forward pass
  Distances
  Train
epoch 108, times: [hn 4.05 s epoch 40.50 s val 0.00 s] lr = 0.050000 loss = -0.0167645 = 0.00366795 + lam * -4.08648, offending 54982
  Forward pass
  Distances
  Train
Valiation at epoch 109
zn_3     nbit= 14:  0.0012 0.0114 0.0874
zn_10    nbit= 32:  0.0433 0.2082 0.5781
zn_79    nbit= 64:  0.2223 0.6528 0.9521
storing test_ckpt/0.005/checkpoint.pth
epoch 109, times: [hn 3.99 s epoch 40.38 s val 10.59 s] lr = 0.050000 loss = -0.0167876 = 0.00364865 + lam * -4.08724, offending 54714
  Forward pass
  Distances
  Train
epoch 110, times: [hn 4.14 s epoch 40.81 s val 0.00 s] lr = 0.050000 loss = -0.0167778 = 0.00365572 + lam * -4.08671, offending 54743
  Forward pass
  Distances
  Train
epoch 111, times: [hn 4.07 s epoch 40.44 s val 0.00 s] lr = 0.050000 loss = -0.0168552 = 0.00357959 + lam * -4.08696, offending 54005
  Forward pass
  Distances
  Train
epoch 112, times: [hn 4.03 s epoch 40.26 s val 0.00 s] lr = 0.050000 loss = -0.0168155 = 0.00362031 + lam * -4.08716, offending 54372
  Forward pass
  Distances
  Train
epoch 113, times: [hn 4.09 s epoch 40.49 s val 0.00 s] lr = 0.050000 loss = -0.0168339 = 0.00359876 + lam * -4.08653, offending 54401
  Forward pass
  Distances
  Train
epoch 114, times: [hn 4.09 s epoch 40.68 s val 0.00 s] lr = 0.050000 loss = -0.0168304 = 0.00360715 + lam * -4.08752, offending 54436
  Forward pass
  Distances
  Train
epoch 115, times: [hn 4.03 s epoch 40.61 s val 0.00 s] lr = 0.050000 loss = -0.0168391 = 0.00359238 + lam * -4.08629, offending 54341
  Forward pass
  Distances
  Train
epoch 116, times: [hn 4.02 s epoch 40.41 s val 0.00 s] lr = 0.050000 loss = -0.0168108 = 0.00362312 + lam * -4.08679, offending 54653
  Forward pass
  Distances
  Train
epoch 117, times: [hn 3.98 s epoch 40.28 s val 0.00 s] lr = 0.050000 loss = -0.0168627 = 0.00358087 + lam * -4.08871, offending 53995
  Forward pass
  Distances
  Train
epoch 118, times: [hn 4.03 s epoch 40.80 s val 0.00 s] lr = 0.050000 loss = -0.0168499 = 0.00359118 + lam * -4.08822, offending 54321
  Forward pass
  Distances
  Train
Valiation at epoch 119
zn_3     nbit= 14:  0.0012 0.0117 0.0872
zn_10    nbit= 32:  0.0430 0.2004 0.5680
zn_79    nbit= 64:  0.2188 0.6509 0.9507
storing test_ckpt/0.005/checkpoint.pth
epoch 119, times: [hn 4.07 s epoch 41.48 s val 10.42 s] lr = 0.050000 loss = -0.0168481 = 0.00358417 + lam * -4.08646, offending 54517
  Forward pass
  Distances
  Train
epoch 120, times: [hn 4.07 s epoch 40.05 s val 0.00 s] lr = 0.010000 loss = -0.0169306 = 0.00350677 + lam * -4.08747, offending 53545
  Forward pass
  Distances
  Train
epoch 121, times: [hn 4.04 s epoch 41.18 s val 0.00 s] lr = 0.010000 loss = -0.0169645 = 0.00347621 + lam * -4.08814, offending 53845
  Forward pass
  Distances
  Train
epoch 122, times: [hn 4.25 s epoch 41.12 s val 0.00 s] lr = 0.010000 loss = -0.0169959 = 0.00344853 + lam * -4.08888, offending 53649
  Forward pass
  Distances
  Train
epoch 123, times: [hn 4.05 s epoch 40.44 s val 0.00 s] lr = 0.010000 loss = -0.0169503 = 0.00349154 + lam * -4.08837, offending 53886
  Forward pass
  Distances
  Train
epoch 124, times: [hn 4.06 s epoch 41.15 s val 0.00 s] lr = 0.010000 loss = -0.0170285 = 0.00341661 + lam * -4.08902, offending 53011
  Forward pass
  Distances
  Train
epoch 125, times: [hn 4.00 s epoch 41.39 s val 0.00 s] lr = 0.010000 loss = -0.0170105 = 0.0034328 + lam * -4.08866, offending 52894
  Forward pass
  Distances
  Train
epoch 126, times: [hn 4.04 s epoch 40.78 s val 0.00 s] lr = 0.010000 loss = -0.0170065 = 0.00344106 + lam * -4.08952, offending 52939
  Forward pass
  Distances
  Train
epoch 127, times: [hn 4.09 s epoch 41.37 s val 0.00 s] lr = 0.010000 loss = -0.01698 = 0.00346704 + lam * -4.0894, offending 53601
  Forward pass
  Distances
  Train
epoch 128, times: [hn 4.04 s epoch 40.36 s val 0.00 s] lr = 0.010000 loss = -0.0170495 = 0.00340141 + lam * -4.09019, offending 52625
  Forward pass
  Distances
  Train
Valiation at epoch 129
zn_3     nbit= 14:  0.0011 0.0118 0.0889
zn_10    nbit= 32:  0.0437 0.2050 0.5744
zn_79    nbit= 64:  0.2194 0.6570 0.9528
storing test_ckpt/0.005/checkpoint.pth
zn_79,rank=10 score improves (0.65704 > 0.65387), keeping as best
epoch 129, times: [hn 4.02 s epoch 41.72 s val 10.06 s] lr = 0.010000 loss = -0.0170013 = 0.003443 + lam * -4.08886, offending 53313
  Forward pass
  Distances
  Train
epoch 130, times: [hn 4.20 s epoch 40.22 s val 0.00 s] lr = 0.010000 loss = -0.0170621 = 0.00338455 + lam * -4.08934, offending 52727
  Forward pass
  Distances
  Train
epoch 131, times: [hn 4.06 s epoch 41.84 s val 0.00 s] lr = 0.010000 loss = -0.0169975 = 0.00345335 + lam * -4.09017, offending 53284
  Forward pass
  Distances
  Train
epoch 132, times: [hn 4.17 s epoch 40.07 s val 0.00 s] lr = 0.010000 loss = -0.0170918 = 0.00335999 + lam * -4.09037, offending 52472
  Forward pass
  Distances
  Train
epoch 133, times: [hn 4.05 s epoch 39.67 s val 0.00 s] lr = 0.010000 loss = -0.0170416 = 0.00340462 + lam * -4.08925, offending 52629
  Forward pass
  Distances
  Train
epoch 134, times: [hn 4.06 s epoch 41.31 s val 0.00 s] lr = 0.010000 loss = -0.0170896 = 0.00336204 + lam * -4.09033, offending 52288
  Forward pass
  Distances
  Train
epoch 135, times: [hn 4.06 s epoch 41.09 s val 0.00 s] lr = 0.010000 loss = -0.0170438 = 0.00340585 + lam * -4.08992, offending 52760
  Forward pass
  Distances
  Train
epoch 136, times: [hn 4.01 s epoch 40.78 s val 0.00 s] lr = 0.010000 loss = -0.017033 = 0.00342034 + lam * -4.09067, offending 52725
  Forward pass
  Distances
  Train
epoch 137, times: [hn 4.10 s epoch 40.39 s val 0.00 s] lr = 0.010000 loss = -0.0170501 = 0.00339845 + lam * -4.0897, offending 52812
  Forward pass
  Distances
  Train
epoch 138, times: [hn 4.08 s epoch 40.96 s val 0.00 s] lr = 0.010000 loss = -0.0170803 = 0.00337062 + lam * -4.09018, offending 52581
  Forward pass
  Distances
  Train
Valiation at epoch 139
zn_3     nbit= 14:  0.0011 0.0115 0.0880
zn_10    nbit= 32:  0.0439 0.2074 0.5761
zn_79    nbit= 64:  0.2204 0.6538 0.9523
storing test_ckpt/0.005/checkpoint.pth
epoch 139, times: [hn 4.04 s epoch 40.03 s val 9.63 s] lr = 0.010000 loss = -0.0170597 = 0.00339017 + lam * -4.08998, offending 52293
  Forward pass
  Distances
  Train
epoch 140, times: [hn 3.99 s epoch 40.76 s val 0.00 s] lr = 0.010000 loss = -0.0170617 = 0.00339033 + lam * -4.09041, offending 52632
  Forward pass
  Distances
  Train
epoch 141, times: [hn 4.05 s epoch 41.18 s val 0.00 s] lr = 0.010000 loss = -0.0170062 = 0.00344693 + lam * -4.09062, offending 53159
  Forward pass
  Distances
  Train
epoch 142, times: [hn 4.02 s epoch 40.20 s val 0.00 s] lr = 0.010000 loss = -0.0170665 = 0.00338832 + lam * -4.09097, offending 52258
  Forward pass
  Distances
  Train
epoch 143, times: [hn 4.03 s epoch 40.67 s val 0.00 s] lr = 0.010000 loss = -0.0170543 = 0.00339979 + lam * -4.09081, offending 52690
  Forward pass
  Distances
  Train
epoch 144, times: [hn 4.03 s epoch 40.00 s val 0.00 s] lr = 0.010000 loss = -0.0170302 = 0.00342319 + lam * -4.09068, offending 52650
  Forward pass
  Distances
  Train
epoch 145, times: [hn 4.04 s epoch 41.66 s val 0.00 s] lr = 0.010000 loss = -0.017089 = 0.00336641 + lam * -4.09107, offending 52416
  Forward pass
  Distances
  Train
epoch 146, times: [hn 4.05 s epoch 42.06 s val 0.00 s] lr = 0.010000 loss = -0.0170437 = 0.00341104 + lam * -4.09094, offending 52889
  Forward pass
  Distances
  Train
epoch 147, times: [hn 4.07 s epoch 40.67 s val 0.00 s] lr = 0.010000 loss = -0.0170855 = 0.00336808 + lam * -4.09072, offending 52449
  Forward pass
  Distances
  Train
epoch 148, times: [hn 4.04 s epoch 40.51 s val 0.00 s] lr = 0.010000 loss = -0.017132 = 0.00332059 + lam * -4.09051, offending 51916
  Forward pass
  Distances
  Train
Valiation at epoch 149
zn_3     nbit= 14:  0.0010 0.0114 0.0878
zn_10    nbit= 32:  0.0431 0.2068 0.5759
zn_79    nbit= 64:  0.2232 0.6561 0.9519
storing test_ckpt/0.005/checkpoint.pth
epoch 149, times: [hn 4.00 s epoch 41.65 s val 9.51 s] lr = 0.010000 loss = -0.0170939 = 0.00336281 + lam * -4.09135, offending 52235
  Forward pass
  Distances
  Train
epoch 150, times: [hn 4.03 s epoch 40.60 s val 0.00 s] lr = 0.010000 loss = -0.0171265 = 0.00333072 + lam * -4.09145, offending 52138
  Forward pass
  Distances
  Train
epoch 151, times: [hn 4.05 s epoch 41.77 s val 0.00 s] lr = 0.010000 loss = -0.0171158 = 0.00334044 + lam * -4.09126, offending 51973
  Forward pass
  Distances
  Train
epoch 152, times: [hn 3.97 s epoch 41.72 s val 0.00 s] lr = 0.010000 loss = -0.0170605 = 0.0033947 + lam * -4.09105, offending 52543
  Forward pass
  Distances
  Train
epoch 153, times: [hn 4.06 s epoch 41.59 s val 0.00 s] lr = 0.010000 loss = -0.0170762 = 0.00338041 + lam * -4.09132, offending 52610
  Forward pass
  Distances
  Train
epoch 154, times: [hn 4.03 s epoch 40.47 s val 0.00 s] lr = 0.010000 loss = -0.0170395 = 0.00341423 + lam * -4.09074, offending 52728
  Forward pass
  Distances
  Train
epoch 155, times: [hn 4.03 s epoch 41.18 s val 0.00 s] lr = 0.010000 loss = -0.017141 = 0.00331739 + lam * -4.09168, offending 51699
  Forward pass
  Distances
  Train
epoch 156, times: [hn 4.06 s epoch 41.38 s val 0.00 s] lr = 0.010000 loss = -0.0171343 = 0.0033251 + lam * -4.09189, offending 52104
  Forward pass
  Distances
  Train
epoch 157, times: [hn 4.04 s epoch 40.52 s val 0.00 s] lr = 0.010000 loss = -0.0171104 = 0.00334695 + lam * -4.09146, offending 52346
  Forward pass
  Distances
  Train
epoch 158, times: [hn 4.05 s epoch 41.37 s val 0.00 s] lr = 0.010000 loss = -0.017098 = 0.00335773 + lam * -4.09116, offending 52412
  Forward pass
  Distances
  Train
Valiation at epoch 159
zn_3     nbit= 14:  0.0011 0.0113 0.0862
zn_10    nbit= 32:  0.0435 0.2034 0.5707
zn_79    nbit= 64:  0.2221 0.6559 0.9523
storing test_ckpt/0.005/checkpoint.pth
epoch 159, times: [hn 4.04 s epoch 40.53 s val 9.51 s] lr = 0.010000 loss = -0.0170846 = 0.00337214 + lam * -4.09136, offending 52449
++ tac test_ckpt/0.005.stdout
++ grep -m1 'keeping as best'
++ grep -o '(.*>'
++ grep -o '[0-9\.]*'
+ perf=0.65704
+ echo 0.65704
0.65704
+ [[ 0.65704 > 0.51987 ]]
+ best_perf=0.65704
+ best_lambda=0.005
+ for lambda in '$lambdas'
+ mkdir -p test_ckpt/0.01
+ python -u train.py --dout 24 --save_best_criterion zn_79,rank=10 --database deep1b --lambda_uniform 0.01 --checkpoint_dir test_ckpt/0.01
+ tee test_ckpt/0.01.stdout
Namespace(batch_size=64, checkpoint_dir='test_ckpt/0.01', database='deep1b', device='cuda', dint=1024, dout=24, epochs=160, init_name='', lambda_uniform=0.01,
 lr_schedule='0.1,0.1,0.05,0.01', momentum=0.9, num_learn=500000, quantizer_train='', rank_negative=50, rank_positive=10, save_best_criterion='zn_79,rank=10',
 seed=1234, size_base=1000000, val_freq=10, validation_quantizers=['zn_3', 'zn_10', 'zn_79'])
load dataset deep1b
keeping 500000/357380000 training vectors
computing training ground truth
build network
Lr schedule [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1,
0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1
, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0
.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0
5, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01,
 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01]
  Forward pass
  Distances
  Train
epoch 0, times: [hn 3.94 s epoch 55.60 s val 0.00 s] lr = 0.100000 loss = -0.0388306 = 0.00239883 + lam * -4.12294, offending 17553
  Forward pass
  Distances
  Train
epoch 1, times: [hn 4.10 s epoch 61.48 s val 0.00 s] lr = 0.100000 loss = -0.0346512 = 0.0060241 + lam * -4.06753, offending 50205
  Forward pass
  Distances
  Train
epoch 2, times: [hn 4.18 s epoch 58.05 s val 0.00 s] lr = 0.100000 loss = -0.0343554 = 0.00656543 + lam * -4.09208, offending 58269
  Forward pass
  Distances
  Train
epoch 3, times: [hn 4.05 s epoch 62.75 s val 0.00 s] lr = 0.100000 loss = -0.0340849 = 0.0067291 + lam * -4.0814, offending 64855
  Forward pass
  Distances
  Train
epoch 4, times: [hn 4.06 s epoch 56.97 s val 0.00 s] lr = 0.100000 loss = -0.0339098 = 0.00699041 + lam * -4.09002, offending 70059
  Forward pass
  Distances
  Train
epoch 5, times: [hn 4.07 s epoch 56.40 s val 0.00 s] lr = 0.100000 loss = -0.0341326 = 0.00675015 + lam * -4.08828, offending 71500
  Forward pass
  Distances
  Train
epoch 6, times: [hn 4.19 s epoch 55.19 s val 0.00 s] lr = 0.100000 loss = -0.0340648 = 0.00688159 + lam * -4.09464, offending 74573
  Forward pass
  Distances
  Train
epoch 7, times: [hn 4.05 s epoch 58.41 s val 0.00 s] lr = 0.100000 loss = -0.034261 = 0.00667349 + lam * -4.09345, offending 74352
  Forward pass
  Distances
  Train
epoch 8, times: [hn 4.11 s epoch 57.72 s val 0.00 s] lr = 0.100000 loss = -0.0343729 = 0.0066195 + lam * -4.09924, offending 75279
  Forward pass
  Distances
  Train
Valiation at epoch 9
zn_3     nbit= 14:  0.0011 0.0105 0.0843
zn_10    nbit= 32:  0.0412 0.1937 0.5510
zn_79    nbit= 64:  0.1935 0.5979 0.9289
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.59788 > 0), keeping as best
epoch 9, times: [hn 4.04 s epoch 55.18 s val 11.30 s] lr = 0.100000 loss = -0.034495 = 0.00648351 + lam * -4.09785, offending 74738
  Forward pass
  Distances
  Train
epoch 10, times: [hn 4.12 s epoch 58.56 s val 0.00 s] lr = 0.100000 loss = -0.0346548 = 0.00635188 + lam * -4.10067, offending 74238
  Forward pass
  Distances
  Train
epoch 11, times: [hn 3.98 s epoch 59.93 s val 0.00 s] lr = 0.100000 loss = -0.034777 = 0.0062543 + lam * -4.10313, offending 73793
  Forward pass
  Distances
  Train
epoch 12, times: [hn 4.07 s epoch 56.66 s val 0.00 s] lr = 0.100000 loss = -0.0349122 = 0.00612601 + lam * -4.10382, offending 72848
  Forward pass
  Distances
  Train
epoch 13, times: [hn 4.08 s epoch 40.33 s val 0.00 s] lr = 0.100000 loss = -0.0350636 = 0.00598589 + lam * -4.10495, offending 72389
  Forward pass
  Distances
  Train
epoch 14, times: [hn 4.06 s epoch 40.60 s val 0.00 s] lr = 0.100000 loss = -0.0351088 = 0.00593621 + lam * -4.1045, offending 71612
  Forward pass
  Distances
  Train
epoch 15, times: [hn 4.08 s epoch 40.17 s val 0.00 s] lr = 0.100000 loss = -0.0352209 = 0.00584642 + lam * -4.10673, offending 70858
  Forward pass
  Distances
  Train
epoch 16, times: [hn 3.98 s epoch 40.03 s val 0.00 s] lr = 0.100000 loss = -0.035386 = 0.00567089 + lam * -4.10569, offending 69684
  Forward pass
  Distances
  Train
epoch 17, times: [hn 4.00 s epoch 40.06 s val 0.00 s] lr = 0.100000 loss = -0.0354507 = 0.00563463 + lam * -4.10853, offending 69513
  Forward pass
  Distances
  Train
epoch 18, times: [hn 4.06 s epoch 40.84 s val 0.00 s] lr = 0.100000 loss = -0.0355119 = 0.00557105 + lam * -4.1083, offending 69243
  Forward pass
  Distances
  Train
Valiation at epoch 19
zn_3     nbit= 14:  0.0010 0.0109 0.0883
zn_10    nbit= 32:  0.0434 0.2045 0.5756
zn_79    nbit= 64:  0.2042 0.6230 0.9404
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.62299 > 0.59788), keeping as best
epoch 19, times: [hn 4.04 s epoch 40.39 s val 10.84 s] lr = 0.100000 loss = -0.0355356 = 0.00557008 + lam * -4.11057, offending 69276
  Forward pass
  Distances
  Train
epoch 20, times: [hn 4.04 s epoch 41.99 s val 0.00 s] lr = 0.100000 loss = -0.0355926 = 0.00550856 + lam * -4.11011, offending 68973
  Forward pass
  Distances
  Train
epoch 21, times: [hn 4.06 s epoch 40.29 s val 0.00 s] lr = 0.100000 loss = -0.0356304 = 0.00547543 + lam * -4.11058, offending 68528
  Forward pass
  Distances
  Train
epoch 22, times: [hn 4.05 s epoch 40.53 s val 0.00 s] lr = 0.100000 loss = -0.0356908 = 0.00542244 + lam * -4.11132, offending 68038
  Forward pass
  Distances
  Train
epoch 23, times: [hn 4.06 s epoch 40.36 s val 0.00 s] lr = 0.100000 loss = -0.0357497 = 0.0053748 + lam * -4.11245, offending 67676
  Forward pass
  Distances
  Train
epoch 24, times: [hn 4.04 s epoch 40.62 s val 0.00 s] lr = 0.100000 loss = -0.0358493 = 0.00527471 + lam * -4.1124, offending 67046
  Forward pass
  Distances
  Train
epoch 25, times: [hn 4.06 s epoch 40.62 s val 0.00 s] lr = 0.100000 loss = -0.0358401 = 0.00528708 + lam * -4.11272, offending 67149
  Forward pass
  Distances
  Train
epoch 26, times: [hn 3.99 s epoch 40.18 s val 0.00 s] lr = 0.100000 loss = -0.0358795 = 0.00525359 + lam * -4.11331, offending 66776
  Forward pass
  Distances
  Train
epoch 27, times: [hn 3.99 s epoch 41.03 s val 0.00 s] lr = 0.100000 loss = -0.0359552 = 0.00518765 + lam * -4.11428, offending 66240
  Forward pass
  Distances
  Train
epoch 28, times: [hn 4.03 s epoch 42.22 s val 0.00 s] lr = 0.100000 loss = -0.0359796 = 0.00516867 + lam * -4.11482, offending 65925
  Forward pass
  Distances
  Train
Valiation at epoch 29
zn_3     nbit= 14:  0.0011 0.0111 0.0886
zn_10    nbit= 32:  0.0446 0.2089 0.5833
zn_79    nbit= 64:  0.2110 0.6338 0.9434
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.63381 > 0.62299), keeping as best
epoch 29, times: [hn 4.08 s epoch 40.98 s val 10.01 s] lr = 0.100000 loss = -0.0360164 = 0.00512174 + lam * -4.11381, offending 65643
  Forward pass
  Distances
  Train
epoch 30, times: [hn 4.17 s epoch 40.61 s val 0.00 s] lr = 0.100000 loss = -0.0360299 = 0.00511035 + lam * -4.11402, offending 65614
  Forward pass
  Distances
  Train
epoch 31, times: [hn 4.06 s epoch 40.46 s val 0.00 s] lr = 0.100000 loss = -0.036088 = 0.00506627 + lam * -4.11543, offending 64958
  Forward pass
  Distances
  Train
epoch 32, times: [hn 4.08 s epoch 40.38 s val 0.00 s] lr = 0.100000 loss = -0.036117 = 0.0050361 + lam * -4.11531, offending 64934
  Forward pass
  Distances
  Train
epoch 33, times: [hn 4.09 s epoch 40.79 s val 0.00 s] lr = 0.100000 loss = -0.0362054 = 0.00494939 + lam * -4.11548, offending 64475
  Forward pass
  Distances
  Train
epoch 34, times: [hn 4.07 s epoch 40.25 s val 0.00 s] lr = 0.100000 loss = -0.0361904 = 0.00498301 + lam * -4.11734, offending 64108
  Forward pass
  Distances
  Train
epoch 35, times: [hn 4.08 s epoch 40.08 s val 0.00 s] lr = 0.100000 loss = -0.0362221 = 0.0049512 + lam * -4.11733, offending 64000
  Forward pass
  Distances
  Train
epoch 36, times: [hn 4.04 s epoch 40.45 s val 0.00 s] lr = 0.100000 loss = -0.036229 = 0.0049406 + lam * -4.11696, offending 64474
  Forward pass
  Distances
  Train
epoch 37, times: [hn 4.08 s epoch 39.96 s val 0.00 s] lr = 0.100000 loss = -0.0362454 = 0.00492242 + lam * -4.11678, offending 64181
  Forward pass
  Distances
  Train
epoch 38, times: [hn 4.07 s epoch 39.92 s val 0.00 s] lr = 0.100000 loss = -0.0362242 = 0.00494829 + lam * -4.11725, offending 64564
  Forward pass
  Distances
  Train
Valiation at epoch 39
zn_3     nbit= 14:  0.0011 0.0110 0.0914
zn_10    nbit= 32:  0.0463 0.2130 0.5879
zn_79    nbit= 64:  0.2160 0.6411 0.9475
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.64113 > 0.63381), keeping as best
epoch 39, times: [hn 4.08 s epoch 41.23 s val 10.05 s] lr = 0.100000 loss = -0.0363602 = 0.00481818 + lam * -4.11784, offending 63474
  Forward pass
  Distances
  Train
epoch 40, times: [hn 4.12 s epoch 40.57 s val 0.00 s] lr = 0.100000 loss = -0.0363323 = 0.00484858 + lam * -4.11808, offending 63668
  Forward pass
  Distances
  Train
epoch 41, times: [hn 4.04 s epoch 40.45 s val 0.00 s] lr = 0.100000 loss = -0.0363415 = 0.00483951 + lam * -4.1181, offending 63436
  Forward pass
  Distances
  Train
epoch 42, times: [hn 4.07 s epoch 40.78 s val 0.00 s] lr = 0.100000 loss = -0.0364155 = 0.00476925 + lam * -4.11848, offending 62681
  Forward pass
  Distances
  Train
epoch 43, times: [hn 4.05 s epoch 40.06 s val 0.00 s] lr = 0.100000 loss = -0.0364113 = 0.00477307 + lam * -4.11844, offending 63073
  Forward pass
  Distances
  Train
epoch 44, times: [hn 4.03 s epoch 40.19 s val 0.00 s] lr = 0.100000 loss = -0.0364122 = 0.00477536 + lam * -4.11876, offending 63003
  Forward pass
  Distances
  Train
epoch 45, times: [hn 4.04 s epoch 40.46 s val 0.00 s] lr = 0.100000 loss = -0.036464 = 0.00472504 + lam * -4.1189, offending 62444
  Forward pass
  Distances
  Train
epoch 46, times: [hn 4.05 s epoch 41.63 s val 0.00 s] lr = 0.100000 loss = -0.0364838 = 0.00471297 + lam * -4.11968, offending 62358
  Forward pass
  Distances
  Train
epoch 47, times: [hn 4.02 s epoch 42.97 s val 0.00 s] lr = 0.100000 loss = -0.0365039 = 0.00468526 + lam * -4.11892, offending 62073
  Forward pass
  Distances
  Train
epoch 48, times: [hn 4.07 s epoch 40.33 s val 0.00 s] lr = 0.100000 loss = -0.0364954 = 0.00470252 + lam * -4.11979, offending 62108
  Forward pass
  Distances
  Train
Valiation at epoch 49
zn_3     nbit= 14:  0.0011 0.0115 0.0916
zn_10    nbit= 32:  0.0461 0.2136 0.5900
zn_79    nbit= 64:  0.2158 0.6442 0.9481
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.64425 > 0.64113), keeping as best
epoch 49, times: [hn 4.03 s epoch 44.22 s val 10.81 s] lr = 0.100000 loss = -0.0365237 = 0.00468001 + lam * -4.12037, offending 62296
  Forward pass
  Distances
  Train
epoch 50, times: [hn 4.11 s epoch 55.74 s val 0.00 s] lr = 0.100000 loss = -0.0365538 = 0.00464329 + lam * -4.11971, offending 62020
  Forward pass
  Distances
  Train
epoch 51, times: [hn 4.04 s epoch 59.54 s val 0.00 s] lr = 0.100000 loss = -0.0366349 = 0.00457613 + lam * -4.12111, offending 61115
  Forward pass
  Distances
  Train
epoch 52, times: [hn 4.09 s epoch 59.24 s val 0.00 s] lr = 0.100000 loss = -0.0366483 = 0.00456312 + lam * -4.12114, offending 61255
  Forward pass
  Distances
  Train
epoch 53, times: [hn 4.09 s epoch 61.55 s val 0.00 s] lr = 0.100000 loss = -0.0365958 = 0.00461121 + lam * -4.12071, offending 61608
  Forward pass
  Distances
  Train
epoch 54, times: [hn 4.05 s epoch 60.82 s val 0.00 s] lr = 0.100000 loss = -0.0366543 = 0.00455697 + lam * -4.12112, offending 61311
  Forward pass
  Distances
  Train
epoch 55, times: [hn 4.08 s epoch 61.53 s val 0.00 s] lr = 0.100000 loss = -0.0366319 = 0.00458399 + lam * -4.12159, offending 61354
  Forward pass
  Distances
  Train
epoch 56, times: [hn 4.22 s epoch 55.89 s val 0.00 s] lr = 0.100000 loss = -0.0366261 = 0.00458436 + lam * -4.12104, offending 61270
  Forward pass
  Distances
  Train
epoch 57, times: [hn 4.04 s epoch 40.54 s val 0.00 s] lr = 0.100000 loss = -0.0366299 = 0.00457461 + lam * -4.12045, offending 61515
  Forward pass
  Distances
  Train
epoch 58, times: [hn 4.06 s epoch 40.19 s val 0.00 s] lr = 0.100000 loss = -0.0366992 = 0.00452159 + lam * -4.12208, offending 60697
  Forward pass
  Distances
  Train
Valiation at epoch 59
zn_3     nbit= 14:  0.0013 0.0112 0.0914
zn_10    nbit= 32:  0.0464 0.2162 0.5956
zn_79    nbit= 64:  0.2179 0.6498 0.9502
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.64984 > 0.64425), keeping as best
epoch 59, times: [hn 4.03 s epoch 41.27 s val 9.87 s] lr = 0.100000 loss = -0.0366942 = 0.00451689 + lam * -4.12111, offending 60685
  Forward pass
  Distances
  Train
epoch 60, times: [hn 4.22 s epoch 40.30 s val 0.00 s] lr = 0.100000 loss = -0.0367208 = 0.00450532 + lam * -4.12261, offending 60503
  Forward pass
  Distances
  Train
epoch 61, times: [hn 4.06 s epoch 38.29 s val 0.00 s] lr = 0.100000 loss = -0.0367373 = 0.00448172 + lam * -4.1219, offending 60473
  Forward pass
  Distances
  Train
epoch 62, times: [hn 4.06 s epoch 40.25 s val 0.00 s] lr = 0.100000 loss = -0.0367208 = 0.00449414 + lam * -4.12149, offending 60695
  Forward pass
  Distances
  Train
epoch 63, times: [hn 4.11 s epoch 39.76 s val 0.00 s] lr = 0.100000 loss = -0.0367404 = 0.00448772 + lam * -4.12281, offending 60518
  Forward pass
  Distances
  Train
epoch 64, times: [hn 3.97 s epoch 40.01 s val 0.00 s] lr = 0.100000 loss = -0.0367825 = 0.00444298 + lam * -4.12254, offending 60355
  Forward pass
  Distances
  Train
epoch 65, times: [hn 4.09 s epoch 39.98 s val 0.00 s] lr = 0.100000 loss = -0.0367752 = 0.00444639 + lam * -4.12216, offending 60272
  Forward pass
  Distances
  Train
epoch 66, times: [hn 3.98 s epoch 45.21 s val 0.00 s] lr = 0.100000 loss = -0.036837 = 0.00439716 + lam * -4.12341, offending 59806
  Forward pass
  Distances
  Train
epoch 67, times: [hn 4.04 s epoch 40.34 s val 0.00 s] lr = 0.100000 loss = -0.0368079 = 0.00441418 + lam * -4.12221, offending 59976
  Forward pass
  Distances
  Train
epoch 68, times: [hn 4.06 s epoch 40.46 s val 0.00 s] lr = 0.100000 loss = -0.0368168 = 0.00441081 + lam * -4.12276, offending 60168
  Forward pass
  Distances
  Train
Valiation at epoch 69
zn_3     nbit= 14:  0.0013 0.0121 0.0931
zn_10    nbit= 32:  0.0471 0.2184 0.5978
zn_79    nbit= 64:  0.2198 0.6512 0.9503
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.65117 > 0.64984), keeping as best
epoch 69, times: [hn 3.99 s epoch 39.81 s val 10.63 s] lr = 0.100000 loss = -0.0368092 = 0.00442328 + lam * -4.12325, offending 60058
  Forward pass
  Distances
  Train
epoch 70, times: [hn 4.10 s epoch 40.84 s val 0.00 s] lr = 0.100000 loss = -0.0368261 = 0.00439943 + lam * -4.12255, offending 60060
  Forward pass
  Distances
  Train
epoch 71, times: [hn 4.04 s epoch 41.36 s val 0.00 s] lr = 0.100000 loss = -0.0368682 = 0.00437563 + lam * -4.12439, offending 59729
  Forward pass
  Distances
  Train
epoch 72, times: [hn 4.13 s epoch 40.34 s val 0.00 s] lr = 0.100000 loss = -0.0369063 = 0.00433732 + lam * -4.12437, offending 59061
  Forward pass
  Distances
  Train
epoch 73, times: [hn 4.04 s epoch 41.06 s val 0.00 s] lr = 0.100000 loss = -0.0368586 = 0.00439475 + lam * -4.12533, offending 59927
  Forward pass
  Distances
  Train
epoch 74, times: [hn 4.04 s epoch 41.47 s val 0.00 s] lr = 0.100000 loss = -0.0369054 = 0.00433722 + lam * -4.12427, offending 59284
  Forward pass
  Distances
  Train
epoch 75, times: [hn 3.99 s epoch 40.37 s val 0.00 s] lr = 0.100000 loss = -0.0369061 = 0.00433399 + lam * -4.12401, offending 59147
  Forward pass
  Distances
  Train
epoch 76, times: [hn 4.06 s epoch 40.85 s val 0.00 s] lr = 0.100000 loss = -0.0369305 = 0.00431859 + lam * -4.12491, offending 59099
  Forward pass
  Distances
  Train
epoch 77, times: [hn 4.07 s epoch 40.68 s val 0.00 s] lr = 0.100000 loss = -0.0369208 = 0.00432539 + lam * -4.12462, offending 59042
  Forward pass
  Distances
  Train
epoch 78, times: [hn 4.05 s epoch 41.02 s val 0.00 s] lr = 0.100000 loss = -0.0369281 = 0.00431337 + lam * -4.12415, offending 58914
  Forward pass
  Distances
  Train
Valiation at epoch 79
zn_3     nbit= 14:  0.0012 0.0115 0.0939
zn_10    nbit= 32:  0.0474 0.2185 0.6005
zn_79    nbit= 64:  0.2220 0.6520 0.9518
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.65203 > 0.65117), keeping as best
epoch 79, times: [hn 4.04 s epoch 40.76 s val 10.57 s] lr = 0.100000 loss = -0.0369565 = 0.00428635 + lam * -4.12429, offending 58818
  Forward pass
  Distances
  Train
epoch 80, times: [hn 4.13 s epoch 41.30 s val 0.00 s] lr = 0.050000 loss = -0.0371312 = 0.00413154 + lam * -4.12628, offending 57834
  Forward pass
  Distances
  Train
epoch 81, times: [hn 4.14 s epoch 40.31 s val 0.00 s] lr = 0.050000 loss = -0.0371665 = 0.00410205 + lam * -4.12686, offending 57874
  Forward pass
  Distances
  Train
epoch 82, times: [hn 4.09 s epoch 40.74 s val 0.00 s] lr = 0.050000 loss = -0.0371635 = 0.00410146 + lam * -4.1265, offending 57575
  Forward pass
  Distances
  Train
epoch 83, times: [hn 4.04 s epoch 40.63 s val 0.00 s] lr = 0.050000 loss = -0.0371682 = 0.00409668 + lam * -4.12649, offending 57491
  Forward pass
  Distances
  Train
epoch 84, times: [hn 4.06 s epoch 40.54 s val 0.00 s] lr = 0.050000 loss = -0.0371282 = 0.00413565 + lam * -4.12638, offending 57782
  Forward pass
  Distances
  Train
epoch 85, times: [hn 4.05 s epoch 40.13 s val 0.00 s] lr = 0.050000 loss = -0.0373 = 0.00397546 + lam * -4.12755, offending 56247
  Forward pass
  Distances
  Train
epoch 86, times: [hn 4.03 s epoch 40.89 s val 0.00 s] lr = 0.050000 loss = -0.0371824 = 0.00409069 + lam * -4.12731, offending 57865
  Forward pass
  Distances
  Train
epoch 87, times: [hn 4.06 s epoch 41.19 s val 0.00 s] lr = 0.050000 loss = -0.0372129 = 0.0040628 + lam * -4.12757, offending 57334
  Forward pass
  Distances
  Train
epoch 88, times: [hn 4.09 s epoch 40.38 s val 0.00 s] lr = 0.050000 loss = -0.03727 = 0.00400763 + lam * -4.12776, offending 56918
  Forward pass
  Distances
  Train
Valiation at epoch 89
zn_3     nbit= 14:  0.0014 0.0117 0.0944
zn_10    nbit= 32:  0.0474 0.2227 0.6044
zn_79    nbit= 64:  0.2240 0.6586 0.9541
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.65864 > 0.65203), keeping as best
epoch 89, times: [hn 4.07 s epoch 40.70 s val 10.69 s] lr = 0.050000 loss = -0.0373023 = 0.00397947 + lam * -4.12818, offending 56629
  Forward pass
  Distances
  Train
epoch 90, times: [hn 4.16 s epoch 40.03 s val 0.00 s] lr = 0.050000 loss = -0.0372635 = 0.00401772 + lam * -4.12812, offending 56820
  Forward pass
  Distances
  Train
epoch 91, times: [hn 4.05 s epoch 40.22 s val 0.00 s] lr = 0.050000 loss = -0.0373106 = 0.00396456 + lam * -4.12751, offending 56483
  Forward pass
  Distances
  Train
epoch 92, times: [hn 4.07 s epoch 40.68 s val 0.00 s] lr = 0.050000 loss = -0.0372843 = 0.00400016 + lam * -4.12844, offending 56476
  Forward pass
  Distances
  Train
epoch 93, times: [hn 4.11 s epoch 41.16 s val 0.00 s] lr = 0.050000 loss = -0.0373227 = 0.00395492 + lam * -4.12777, offending 56044
  Forward pass
  Distances
  Train
epoch 94, times: [hn 4.06 s epoch 40.05 s val 0.00 s] lr = 0.050000 loss = -0.0373422 = 0.00394299 + lam * -4.12852, offending 56257
  Forward pass
  Distances
  Train
epoch 95, times: [hn 4.09 s epoch 40.13 s val 0.00 s] lr = 0.050000 loss = -0.0372601 = 0.0040202 + lam * -4.12803, offending 57042
  Forward pass
  Distances
  Train
epoch 96, times: [hn 4.06 s epoch 40.31 s val 0.00 s] lr = 0.050000 loss = -0.0373134 = 0.00396365 + lam * -4.12771, offending 56505
  Forward pass
  Distances
  Train
epoch 97, times: [hn 4.09 s epoch 40.08 s val 0.00 s] lr = 0.050000 loss = -0.0373422 = 0.00393705 + lam * -4.12793, offending 56485
  Forward pass
  Distances
  Train
epoch 98, times: [hn 4.03 s epoch 41.03 s val 0.00 s] lr = 0.050000 loss = -0.0373156 = 0.00396994 + lam * -4.12856, offending 56586
  Forward pass
  Distances
  Train
Valiation at epoch 99
zn_3     nbit= 14:  0.0012 0.0118 0.0953
zn_10    nbit= 32:  0.0491 0.2258 0.6097
zn_79    nbit= 64:  0.2245 0.6633 0.9551
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.66333 > 0.65864), keeping as best
epoch 99, times: [hn 4.05 s epoch 41.02 s val 9.70 s] lr = 0.050000 loss = -0.0373644 = 0.00392902 + lam * -4.12934, offending 55828
  Forward pass
  Distances
  Train
epoch 100, times: [hn 4.14 s epoch 39.59 s val 0.00 s] lr = 0.050000 loss = -0.0373078 = 0.00397159 + lam * -4.12794, offending 56390
  Forward pass
  Distances
  Train
epoch 101, times: [hn 3.96 s epoch 39.86 s val 0.00 s] lr = 0.050000 loss = -0.0373346 = 0.00394541 + lam * -4.128, offending 56361
  Forward pass
  Distances
  Train
epoch 102, times: [hn 4.00 s epoch 39.80 s val 0.00 s] lr = 0.050000 loss = -0.0373887 = 0.00389937 + lam * -4.1288, offending 55688
  Forward pass
  Distances
  Train
epoch 103, times: [hn 4.07 s epoch 40.48 s val 0.00 s] lr = 0.050000 loss = -0.0373829 = 0.00390825 + lam * -4.12912, offending 55884
  Forward pass
  Distances
  Train
epoch 104, times: [hn 4.14 s epoch 40.22 s val 0.00 s] lr = 0.050000 loss = -0.0373246 = 0.00395665 + lam * -4.12812, offending 56308
  Forward pass
  Distances
  Train
epoch 105, times: [hn 4.08 s epoch 40.20 s val 0.00 s] lr = 0.050000 loss = -0.0373911 = 0.0038968 + lam * -4.12879, offending 55900
  Forward pass
  Distances
  Train
epoch 106, times: [hn 3.99 s epoch 39.92 s val 0.00 s] lr = 0.050000 loss = -0.0373978 = 0.00388952 + lam * -4.12873, offending 55784
  Forward pass
  Distances
  Train
epoch 107, times: [hn 4.04 s epoch 40.50 s val 0.00 s] lr = 0.050000 loss = -0.037383 = 0.00390795 + lam * -4.1291, offending 55765
  Forward pass
  Distances
  Train
epoch 108, times: [hn 3.98 s epoch 40.02 s val 0.00 s] lr = 0.050000 loss = -0.0374041 = 0.00389 + lam * -4.12941, offending 55767
  Forward pass
  Distances
  Train
Valiation at epoch 109
zn_3     nbit= 14:  0.0013 0.0121 0.0958
zn_10    nbit= 32:  0.0497 0.2259 0.6089
zn_79    nbit= 64:  0.2252 0.6633 0.9545
storing test_ckpt/0.01/checkpoint.pth
epoch 109, times: [hn 3.98 s epoch 39.97 s val 10.56 s] lr = 0.050000 loss = -0.0374133 = 0.00388323 + lam * -4.12965, offending 55721
  Forward pass
  Distances
  Train
epoch 110, times: [hn 4.15 s epoch 40.73 s val 0.00 s] lr = 0.050000 loss = -0.0374078 = 0.00388423 + lam * -4.1292, offending 55818
  Forward pass
  Distances
  Train
epoch 111, times: [hn 4.10 s epoch 40.80 s val 0.00 s] lr = 0.050000 loss = -0.0374487 = 0.0038488 + lam * -4.12975, offending 55267
  Forward pass
  Distances
  Train
epoch 112, times: [hn 4.04 s epoch 41.35 s val 0.00 s] lr = 0.050000 loss = -0.0374284 = 0.00387069 + lam * -4.12991, offending 55342
  Forward pass
  Distances
  Train
epoch 113, times: [hn 4.03 s epoch 40.76 s val 0.00 s] lr = 0.050000 loss = -0.0374732 = 0.0038271 + lam * -4.13003, offending 55295
  Forward pass
  Distances
  Train
epoch 114, times: [hn 4.06 s epoch 40.06 s val 0.00 s] lr = 0.050000 loss = -0.0374249 = 0.00387654 + lam * -4.13015, offending 55656
  Forward pass
  Distances
  Train
epoch 115, times: [hn 4.14 s epoch 40.22 s val 0.00 s] lr = 0.050000 loss = -0.0374496 = 0.00384704 + lam * -4.12966, offending 55436
  Forward pass
  Distances
  Train
epoch 116, times: [hn 4.01 s epoch 40.78 s val 0.00 s] lr = 0.050000 loss = -0.0374379 = 0.00386093 + lam * -4.12989, offending 55450
  Forward pass
  Distances
  Train
epoch 117, times: [hn 3.97 s epoch 40.06 s val 0.00 s] lr = 0.050000 loss = -0.0374803 = 0.00382716 + lam * -4.13075, offending 55258
  Forward pass
  Distances
  Train
epoch 118, times: [hn 4.06 s epoch 41.04 s val 0.00 s] lr = 0.050000 loss = -0.0374842 = 0.00381817 + lam * -4.13024, offending 55189
  Forward pass
  Distances
  Train
Valiation at epoch 119
zn_3     nbit= 14:  0.0014 0.0116 0.0943
zn_10    nbit= 32:  0.0479 0.2224 0.6027
zn_79    nbit= 64:  0.2233 0.6580 0.9532
storing test_ckpt/0.01/checkpoint.pth
epoch 119, times: [hn 3.97 s epoch 40.08 s val 9.45 s] lr = 0.050000 loss = -0.0374331 = 0.00386145 + lam * -4.12946, offending 55704
  Forward pass
  Distances
  Train
epoch 120, times: [hn 4.18 s epoch 40.83 s val 0.00 s] lr = 0.010000 loss = -0.0375378 = 0.00375828 + lam * -4.1296, offending 54745
  Forward pass
  Distances
  Train
epoch 121, times: [hn 4.07 s epoch 47.25 s val 0.00 s] lr = 0.010000 loss = -0.0375743 = 0.00372604 + lam * -4.13004, offending 54567
  Forward pass
  Distances
  Train
epoch 122, times: [hn 4.03 s epoch 45.39 s val 0.00 s] lr = 0.010000 loss = -0.0376124 = 0.00369825 + lam * -4.13107, offending 54316
  Forward pass
  Distances
  Train
epoch 123, times: [hn 4.04 s epoch 40.84 s val 0.00 s] lr = 0.010000 loss = -0.0375644 = 0.00374585 + lam * -4.13102, offending 54626
  Forward pass
  Distances
  Train
epoch 124, times: [hn 4.05 s epoch 41.06 s val 0.00 s] lr = 0.010000 loss = -0.0376772 = 0.00363221 + lam * -4.13094, offending 53927
  Forward pass
  Distances
  Train
epoch 125, times: [hn 4.01 s epoch 40.53 s val 0.00 s] lr = 0.010000 loss = -0.0376288 = 0.00368296 + lam * -4.13117, offending 54186
  Forward pass
  Distances
  Train
epoch 126, times: [hn 3.97 s epoch 40.14 s val 0.00 s] lr = 0.010000 loss = -0.0376376 = 0.00367758 + lam * -4.13152, offending 54030
  Forward pass
  Distances
  Train
epoch 127, times: [hn 3.96 s epoch 40.41 s val 0.00 s] lr = 0.010000 loss = -0.0376348 = 0.00367859 + lam * -4.13134, offending 54236
  Forward pass
  Distances
  Train
epoch 128, times: [hn 4.06 s epoch 41.61 s val 0.00 s] lr = 0.010000 loss = -0.0376858 = 0.00363278 + lam * -4.13185, offending 53474
  Forward pass
  Distances
  Train
Valiation at epoch 129
zn_3     nbit= 14:  0.0014 0.0121 0.0952
zn_10    nbit= 32:  0.0499 0.2265 0.6106
zn_79    nbit= 64:  0.2268 0.6650 0.9549
storing test_ckpt/0.01/checkpoint.pth
zn_79,rank=10 score improves (0.66498 > 0.66333), keeping as best
epoch 129, times: [hn 4.10 s epoch 42.29 s val 10.52 s] lr = 0.010000 loss = -0.0376268 = 0.00368501 + lam * -4.13118, offending 54377
  Forward pass
  Distances
  Train
epoch 130, times: [hn 4.13 s epoch 43.12 s val 0.00 s] lr = 0.010000 loss = -0.0376536 = 0.00365656 + lam * -4.13102, offending 53836
  Forward pass
  Distances
  Train
epoch 131, times: [hn 4.06 s epoch 41.57 s val 0.00 s] lr = 0.010000 loss = -0.0376432 = 0.00367686 + lam * -4.13201, offending 54173
  Forward pass
  Distances
  Train
epoch 132, times: [hn 4.03 s epoch 41.44 s val 0.00 s] lr = 0.010000 loss = -0.0377112 = 0.00360902 + lam * -4.13202, offending 53459
  Forward pass
  Distances
  Train
epoch 133, times: [hn 4.12 s epoch 41.64 s val 0.00 s] lr = 0.010000 loss = -0.0376682 = 0.00364412 + lam * -4.13124, offending 53936
  Forward pass
  Distances
  Train
epoch 134, times: [hn 4.11 s epoch 40.70 s val 0.00 s] lr = 0.010000 loss = -0.037727 = 0.0035939 + lam * -4.13209, offending 53169
  Forward pass
  Distances
  Train
epoch 135, times: [hn 4.05 s epoch 43.87 s val 0.00 s] lr = 0.010000 loss = -0.0377155 = 0.00360513 + lam * -4.13206, offending 53348
  Forward pass
  Distances
  Train
epoch 136, times: [hn 4.12 s epoch 42.43 s val 0.00 s] lr = 0.010000 loss = -0.0376814 = 0.00364063 + lam * -4.1322, offending 53857
  Forward pass
  Distances
  Train
epoch 137, times: [hn 4.15 s epoch 41.16 s val 0.00 s] lr = 0.010000 loss = -0.0376527 = 0.00365901 + lam * -4.13117, offending 53977
  Forward pass
  Distances
  Train
epoch 138, times: [hn 4.04 s epoch 40.28 s val 0.00 s] lr = 0.010000 loss = -0.0377264 = 0.00359336 + lam * -4.13197, offending 53289
  Forward pass
  Distances
  Train
Valiation at epoch 139
zn_3     nbit= 14:  0.0013 0.0119 0.0970
zn_10    nbit= 32:  0.0490 0.2255 0.6078
zn_79    nbit= 64:  0.2256 0.6623 0.9550
storing test_ckpt/0.01/checkpoint.pth
epoch 139, times: [hn 4.10 s epoch 41.38 s val 10.19 s] lr = 0.010000 loss = -0.0376771 = 0.00364016 + lam * -4.13173, offending 53827
  Forward pass
  Distances
  Train
epoch 140, times: [hn 4.20 s epoch 40.51 s val 0.00 s] lr = 0.010000 loss = -0.0377338 = 0.00358888 + lam * -4.13227, offending 53298
  Forward pass
  Distances
  Train
epoch 141, times: [hn 4.04 s epoch 46.31 s val 0.00 s] lr = 0.010000 loss = -0.0376531 = 0.00366643 + lam * -4.13195, offending 54096
  Forward pass
  Distances
  Train
epoch 142, times: [hn 4.05 s epoch 42.17 s val 0.00 s] lr = 0.010000 loss = -0.0376967 = 0.00362874 + lam * -4.13254, offending 53450
  Forward pass
  Distances
  Train
epoch 143, times: [hn 4.10 s epoch 52.65 s val 0.00 s] lr = 0.010000 loss = -0.0377641 = 0.00356783 + lam * -4.13319, offending 52988
  Forward pass
  Distances
  Train
epoch 144, times: [hn 4.06 s epoch 59.29 s val 0.00 s] lr = 0.010000 loss = -0.0376716 = 0.00365357 + lam * -4.13251, offending 53480
  Forward pass
  Distances
  Train
epoch 145, times: [hn 4.07 s epoch 56.35 s val 0.00 s] lr = 0.010000 loss = -0.0377364 = 0.00359358 + lam * -4.133, offending 53050
  Forward pass
  Distances
  Train
epoch 146, times: [hn 4.07 s epoch 59.33 s val 0.00 s] lr = 0.010000 loss = -0.0376941 = 0.00363074 + lam * -4.13249, offending 53552
  Forward pass
  Distances
  Train
epoch 147, times: [hn 4.06 s epoch 55.20 s val 0.00 s] lr = 0.010000 loss = -0.037717 = 0.00360843 + lam * -4.13255, offending 53368
  Forward pass
  Distances
  Train
epoch 148, times: [hn 4.19 s epoch 56.47 s val 0.00 s] lr = 0.010000 loss = -0.0377535 = 0.00357167 + lam * -4.13252, offending 52882
  Forward pass
  Distances
  Train
Valiation at epoch 149
zn_3     nbit= 14:  0.0012 0.0121 0.0953
zn_10    nbit= 32:  0.0490 0.2253 0.6079
zn_79    nbit= 64:  0.2264 0.6646 0.9558
storing test_ckpt/0.01/checkpoint.pth
epoch 149, times: [hn 4.05 s epoch 55.43 s val 9.67 s] lr = 0.010000 loss = -0.0377252 = 0.00359846 + lam * -4.13237, offending 53097
  Forward pass
  Distances
  Train
epoch 150, times: [hn 4.03 s epoch 47.89 s val 0.00 s] lr = 0.010000 loss = -0.0377599 = 0.00356744 + lam * -4.13273, offending 53115
  Forward pass
  Distances
  Train
epoch 151, times: [hn 4.04 s epoch 61.73 s val 0.00 s] lr = 0.010000 loss = -0.0377572 = 0.00356828 + lam * -4.13255, offending 53105
  Forward pass
  Distances
  Train
epoch 152, times: [hn 4.05 s epoch 56.27 s val 0.00 s] lr = 0.010000 loss = -0.0377402 = 0.00358728 + lam * -4.13275, offending 53378
  Forward pass
  Distances
  Train
epoch 153, times: [hn 4.17 s epoch 59.37 s val 0.00 s] lr = 0.010000 loss = -0.0376959 = 0.00362898 + lam * -4.13249, offending 53275
  Forward pass
  Distances
  Train
epoch 154, times: [hn 4.10 s epoch 58.66 s val 0.00 s] lr = 0.010000 loss = -0.0377226 = 0.00360385 + lam * -4.13264, offending 53375
  Forward pass
  Distances
  Train
epoch 155, times: [hn 4.09 s epoch 62.83 s val 0.00 s] lr = 0.010000 loss = -0.0377617 = 0.00356784 + lam * -4.13296, offending 52795
  Forward pass
  Distances
  Train
epoch 156, times: [hn 4.07 s epoch 54.34 s val 0.00 s] lr = 0.010000 loss = -0.0377783 = 0.00355644 + lam * -4.13347, offending 52914
  Forward pass
  Distances
  Train
epoch 157, times: [hn 4.10 s epoch 54.36 s val 0.00 s] lr = 0.010000 loss = -0.0377071 = 0.0036214 + lam * -4.13285, offending 53588
  Forward pass
  Distances
  Train
epoch 158, times: [hn 4.03 s epoch 57.18 s val 0.00 s] lr = 0.010000 loss = -0.0377485 = 0.00357552 + lam * -4.1324, offending 53207
  Forward pass
  Distances
  Train
Valiation at epoch 159
zn_3     nbit= 14:  0.0014 0.0119 0.0948
zn_10    nbit= 32:  0.0472 0.2237 0.6055
zn_79    nbit= 64:  0.2252 0.6646 0.9546
storing test_ckpt/0.01/checkpoint.pth
epoch 159, times: [hn 4.18 s epoch 59.14 s val 9.58 s] lr = 0.010000 loss = -0.0377265 = 0.00359841 + lam * -4.13249, offending 53303
++ tac test_ckpt/0.01.stdout
++ grep -m1 'keeping as best'
++ grep -o '(.*>'
++ grep -o '[0-9\.]*'
+ perf=0.66498
+ echo 0.66498
0.66498
+ [[ 0.66498 > 0.65704 ]]
+ best_perf=0.66498
+ best_lambda=0.01
+ for lambda in '$lambdas'
+ mkdir -p test_ckpt/0.02
+ python -u train.py --dout 24 --save_best_criterion zn_79,rank=10 --database deep1b --lambda_uniform 0.02 --checkpoint_dir test_ckpt/0.02
+ tee test_ckpt/0.02.stdout
Namespace(batch_size=64, checkpoint_dir='test_ckpt/0.02', database='deep1b', device='cuda', dint=1024, dout=24, epochs=160, init_name='', lambda_uniform=0.02,
 lr_schedule='0.1,0.1,0.05,0.01', momentum=0.9, num_learn=500000, quantizer_train='', rank_negative=50, rank_positive=10, save_best_criterion='zn_79,rank=10',
 seed=1234, size_base=1000000, val_freq=10, validation_quantizers=['zn_3', 'zn_10', 'zn_79'])
load dataset deep1b
keeping 500000/357380000 training vectors
computing training ground truth
build network
Lr schedule [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1,
0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1
, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0
.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0
5, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01,
 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01]
  Forward pass
  Distances
  Train
epoch 0, times: [hn 4.02 s epoch 50.54 s val 0.00 s] lr = 0.100000 loss = -0.0804305 = 0.00256756 + lam * -4.1499, offending 18043
  Forward pass
  Distances
  Train
epoch 1, times: [hn 4.06 s epoch 48.71 s val 0.00 s] lr = 0.100000 loss = -0.0761247 = 0.00630061 + lam * -4.12127, offending 49386
  Forward pass
  Distances
  Train
epoch 2, times: [hn 4.07 s epoch 53.25 s val 0.00 s] lr = 0.100000 loss = -0.0757752 = 0.00682431 + lam * -4.12997, offending 58154
  Forward pass
  Distances
  Train
epoch 3, times: [hn 4.02 s epoch 52.63 s val 0.00 s] lr = 0.100000 loss = -0.0754321 = 0.00711875 + lam * -4.12754, offending 65172
  Forward pass
  Distances
  Train
epoch 4, times: [hn 4.05 s epoch 43.05 s val 0.00 s] lr = 0.100000 loss = -0.075312 = 0.00730566 + lam * -4.13088, offending 70119
  Forward pass
  Distances
  Train
epoch 5, times: [hn 4.07 s epoch 53.53 s val 0.00 s] lr = 0.100000 loss = -0.0753637 = 0.00723705 + lam * -4.13004, offending 73154
  Forward pass
  Distances
  Train
epoch 6, times: [hn 4.12 s epoch 45.47 s val 0.00 s] lr = 0.100000 loss = -0.0754022 = 0.00725845 + lam * -4.13303, offending 74730
  Forward pass
  Distances
  Train
epoch 7, times: [hn 4.03 s epoch 54.67 s val 0.00 s] lr = 0.100000 loss = -0.0755559 = 0.00712167 + lam * -4.13388, offending 75615
  Forward pass
  Distances
  Train
epoch 8, times: [hn 3.97 s epoch 46.75 s val 0.00 s] lr = 0.100000 loss = -0.075727 = 0.00700868 + lam * -4.13678, offending 75721
  Forward pass
  Distances
  Train
Valiation at epoch 9
zn_3     nbit= 14:  0.0011 0.0124 0.0945
zn_10    nbit= 32:  0.0456 0.2103 0.5784
zn_79    nbit= 64:  0.1965 0.6024 0.9276
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.60235 > 0), keeping as best
epoch 9, times: [hn 4.07 s epoch 50.22 s val 10.57 s] lr = 0.100000 loss = -0.0758759 = 0.00681491 + lam * -4.13454, offending 75308
  Forward pass
  Distances
  Train
epoch 10, times: [hn 4.16 s epoch 53.78 s val 0.00 s] lr = 0.100000 loss = -0.075993 = 0.00673497 + lam * -4.1364, offending 75358
  Forward pass
  Distances
  Train
epoch 11, times: [hn 4.25 s epoch 45.26 s val 0.00 s] lr = 0.100000 loss = -0.0761104 = 0.00664891 + lam * -4.13796, offending 74885
  Forward pass
  Distances
  Train
epoch 12, times: [hn 4.07 s epoch 42.35 s val 0.00 s] lr = 0.100000 loss = -0.0763091 = 0.00646047 + lam * -4.13848, offending 73818
  Forward pass
  Distances
  Train
epoch 13, times: [hn 4.06 s epoch 40.09 s val 0.00 s] lr = 0.100000 loss = -0.0763964 = 0.00639814 + lam * -4.13973, offending 73592
  Forward pass
  Distances
  Train
epoch 14, times: [hn 4.08 s epoch 39.16 s val 0.00 s] lr = 0.100000 loss = -0.0764489 = 0.00635176 + lam * -4.14003, offending 73279
  Forward pass
  Distances
  Train
epoch 15, times: [hn 4.03 s epoch 40.78 s val 0.00 s] lr = 0.100000 loss = -0.0765721 = 0.00625626 + lam * -4.14142, offending 72665
  Forward pass
  Distances
  Train
epoch 16, times: [hn 4.03 s epoch 40.33 s val 0.00 s] lr = 0.100000 loss = -0.0766747 = 0.0061446 + lam * -4.14096, offending 71920
  Forward pass
  Distances
  Train
epoch 17, times: [hn 4.08 s epoch 40.82 s val 0.00 s] lr = 0.100000 loss = -0.0767811 = 0.00608732 + lam * -4.14342, offending 71527
  Forward pass
  Distances
  Train
epoch 18, times: [hn 4.00 s epoch 41.41 s val 0.00 s] lr = 0.100000 loss = -0.0768102 = 0.00603145 + lam * -4.14208, offending 71245
  Forward pass
  Distances
  Train
Valiation at epoch 19
zn_3     nbit= 14:  0.0012 0.0116 0.0946
zn_10    nbit= 32:  0.0486 0.2192 0.5973
zn_79    nbit= 64:  0.2068 0.6283 0.9411
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.62826 > 0.60235), keeping as best
epoch 19, times: [hn 4.17 s epoch 41.83 s val 10.43 s] lr = 0.100000 loss = -0.0769102 = 0.00595344 + lam * -4.14318, offending 70779
  Forward pass
  Distances
  Train
epoch 20, times: [hn 4.08 s epoch 41.45 s val 0.00 s] lr = 0.100000 loss = -0.0769646 = 0.00591559 + lam * -4.14401, offending 70526
  Forward pass
  Distances
  Train
epoch 21, times: [hn 3.97 s epoch 41.02 s val 0.00 s] lr = 0.100000 loss = -0.0770523 = 0.00581315 + lam * -4.14327, offending 69701
  Forward pass
  Distances
  Train
epoch 22, times: [hn 4.03 s epoch 41.40 s val 0.00 s] lr = 0.100000 loss = -0.0771186 = 0.00577438 + lam * -4.14465, offending 69296
  Forward pass
  Distances
  Train
epoch 23, times: [hn 4.01 s epoch 41.46 s val 0.00 s] lr = 0.100000 loss = -0.0771985 = 0.00571047 + lam * -4.14545, offending 69045
  Forward pass
  Distances
  Train
epoch 24, times: [hn 4.03 s epoch 40.89 s val 0.00 s] lr = 0.100000 loss = -0.077273 = 0.00563255 + lam * -4.14528, offending 68446
  Forward pass
  Distances
  Train
epoch 25, times: [hn 4.05 s epoch 40.30 s val 0.00 s] lr = 0.100000 loss = -0.0772622 = 0.00564103 + lam * -4.14516, offending 68464
  Forward pass
  Distances
  Train
epoch 26, times: [hn 4.06 s epoch 41.23 s val 0.00 s] lr = 0.100000 loss = -0.0773196 = 0.00560283 + lam * -4.14612, offending 68227
  Forward pass
  Distances
  Train
epoch 27, times: [hn 4.10 s epoch 40.91 s val 0.00 s] lr = 0.100000 loss = -0.0773816 = 0.00553805 + lam * -4.14598, offending 67748
  Forward pass
  Distances
  Train
epoch 28, times: [hn 4.08 s epoch 41.41 s val 0.00 s] lr = 0.100000 loss = -0.077427 = 0.00551766 + lam * -4.14723, offending 67442
  Forward pass
  Distances
  Train
Valiation at epoch 29
zn_3     nbit= 14:  0.0013 0.0122 0.0972
zn_10    nbit= 32:  0.0508 0.2236 0.6035
zn_79    nbit= 64:  0.2139 0.6374 0.9438
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.63738 > 0.62826), keeping as best
epoch 29, times: [hn 4.02 s epoch 41.33 s val 9.86 s] lr = 0.100000 loss = -0.077478 = 0.00543815 + lam * -4.14581, offending 67101
  Forward pass
  Distances
  Train
epoch 30, times: [hn 4.07 s epoch 41.75 s val 0.00 s] lr = 0.100000 loss = -0.0775014 = 0.00543927 + lam * -4.14704, offending 67252
  Forward pass
  Distances
  Train
epoch 31, times: [hn 4.02 s epoch 41.34 s val 0.00 s] lr = 0.100000 loss = -0.077528 = 0.00541981 + lam * -4.14739, offending 66776
  Forward pass
  Distances
  Train
epoch 32, times: [hn 4.04 s epoch 41.03 s val 0.00 s] lr = 0.100000 loss = -0.0775489 = 0.00540614 + lam * -4.14775, offending 66575
  Forward pass
  Distances
  Train
epoch 33, times: [hn 4.11 s epoch 55.80 s val 0.00 s] lr = 0.100000 loss = -0.0776501 = 0.00530298 + lam * -4.14766, offending 66214
  Forward pass
  Distances
  Train
epoch 34, times: [hn 4.01 s epoch 49.37 s val 0.00 s] lr = 0.100000 loss = -0.0775987 = 0.00536641 + lam * -4.14825, offending 66212
  Forward pass
  Distances
  Train
epoch 35, times: [hn 3.99 s epoch 57.92 s val 0.00 s] lr = 0.100000 loss = -0.0776448 = 0.00532186 + lam * -4.14833, offending 65644
  Forward pass
  Distances
  Train
epoch 36, times: [hn 4.07 s epoch 59.82 s val 0.00 s] lr = 0.100000 loss = -0.0776927 = 0.00529682 + lam * -4.14948, offending 65858
  Forward pass
  Distances
  Train
epoch 37, times: [hn 4.06 s epoch 54.36 s val 0.00 s] lr = 0.100000 loss = -0.0777139 = 0.00525316 + lam * -4.14835, offending 65373
  Forward pass
  Distances
  Train
epoch 38, times: [hn 4.06 s epoch 54.19 s val 0.00 s] lr = 0.100000 loss = -0.0777013 = 0.00526815 + lam * -4.14847, offending 65556
  Forward pass
  Distances
  Train
Valiation at epoch 39
zn_3     nbit= 14:  0.0014 0.0121 0.0990
zn_10    nbit= 32:  0.0496 0.2284 0.6124
zn_79    nbit= 64:  0.2156 0.6446 0.9479
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.64456 > 0.63738), keeping as best
epoch 39, times: [hn 4.06 s epoch 53.15 s val 9.84 s] lr = 0.100000 loss = -0.0778481 = 0.0051395 + lam * -4.14938, offending 64598
  Forward pass
  Distances
  Train
epoch 40, times: [hn 4.15 s epoch 48.46 s val 0.00 s] lr = 0.100000 loss = -0.0777681 = 0.00521206 + lam * -4.14901, offending 65189
  Forward pass
  Distances
  Train
epoch 41, times: [hn 4.07 s epoch 48.88 s val 0.00 s] lr = 0.100000 loss = -0.077866 = 0.00512901 + lam * -4.14975, offending 64473
  Forward pass
  Distances
  Train
epoch 42, times: [hn 4.05 s epoch 56.30 s val 0.00 s] lr = 0.100000 loss = -0.0778645 = 0.0051339 + lam * -4.14992, offending 64370
  Forward pass
  Distances
  Train
epoch 43, times: [hn 4.27 s epoch 59.65 s val 0.00 s] lr = 0.100000 loss = -0.077828 = 0.0051696 + lam * -4.14988, offending 64901
  Forward pass
  Distances
  Train
epoch 44, times: [hn 4.00 s epoch 55.81 s val 0.00 s] lr = 0.100000 loss = -0.0778963 = 0.00510673 + lam * -4.15015, offending 63974
  Forward pass
  Distances
  Train
epoch 45, times: [hn 4.09 s epoch 52.69 s val 0.00 s] lr = 0.100000 loss = -0.0779451 = 0.00506434 + lam * -4.15047, offending 64033
  Forward pass
  Distances
  Train
epoch 46, times: [hn 4.07 s epoch 57.51 s val 0.00 s] lr = 0.100000 loss = -0.0779362 = 0.00507526 + lam * -4.15057, offending 63953
  Forward pass
  Distances
  Train
epoch 47, times: [hn 4.06 s epoch 57.30 s val 0.00 s] lr = 0.100000 loss = -0.078003 = 0.005005 + lam * -4.1504, offending 63499
  Forward pass
  Distances
  Train
epoch 48, times: [hn 4.04 s epoch 52.22 s val 0.00 s] lr = 0.100000 loss = -0.077989 = 0.00502757 + lam * -4.15083, offending 63791
  Forward pass
  Distances
  Train
Valiation at epoch 49
zn_3     nbit= 14:  0.0012 0.0125 0.1002
zn_10    nbit= 32:  0.0505 0.2295 0.6125
zn_79    nbit= 64:  0.2168 0.6446 0.9473
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.64462 > 0.64456), keeping as best
epoch 49, times: [hn 4.13 s epoch 53.41 s val 9.65 s] lr = 0.100000 loss = -0.0780338 = 0.00498299 + lam * -4.15084, offending 63517
  Forward pass
  Distances
  Train
epoch 50, times: [hn 4.19 s epoch 57.29 s val 0.00 s] lr = 0.100000 loss = -0.0781153 = 0.00491211 + lam * -4.15137, offending 62666
  Forward pass
  Distances
  Train
epoch 51, times: [hn 4.07 s epoch 49.50 s val 0.00 s] lr = 0.100000 loss = -0.0780676 = 0.00497049 + lam * -4.15191, offending 63226
  Forward pass
  Distances
  Train
epoch 52, times: [hn 4.06 s epoch 57.41 s val 0.00 s] lr = 0.100000 loss = -0.0781251 = 0.0049018 + lam * -4.15135, offending 62668
  Forward pass
  Distances
  Train
epoch 53, times: [hn 4.05 s epoch 56.86 s val 0.00 s] lr = 0.100000 loss = -0.0780818 = 0.00495987 + lam * -4.15208, offending 63441
  Forward pass
  Distances
  Train
epoch 54, times: [hn 4.04 s epoch 58.35 s val 0.00 s] lr = 0.100000 loss = -0.0781177 = 0.00491142 + lam * -4.15146, offending 62878
  Forward pass
  Distances
  Train
epoch 55, times: [hn 4.08 s epoch 54.17 s val 0.00 s] lr = 0.100000 loss = -0.078115 = 0.00490639 + lam * -4.15107, offending 62885
  Forward pass
  Distances
  Train
epoch 56, times: [hn 4.06 s epoch 54.61 s val 0.00 s] lr = 0.100000 loss = -0.0781579 = 0.00487607 + lam * -4.1517, offending 62629
  Forward pass
  Distances
  Train
epoch 57, times: [hn 4.24 s epoch 57.28 s val 0.00 s] lr = 0.100000 loss = -0.0781363 = 0.00487979 + lam * -4.1508, offending 62650
  Forward pass
  Distances
  Train
epoch 58, times: [hn 3.99 s epoch 59.29 s val 0.00 s] lr = 0.100000 loss = -0.0782283 = 0.004817 + lam * -4.15226, offending 62032
  Forward pass
  Distances
  Train
Valiation at epoch 59
zn_3     nbit= 14:  0.0012 0.0128 0.1021
zn_10    nbit= 32:  0.0520 0.2351 0.6182
zn_79    nbit= 64:  0.2199 0.6513 0.9498
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.65133 > 0.64462), keeping as best
epoch 59, times: [hn 4.17 s epoch 55.20 s val 9.93 s] lr = 0.100000 loss = -0.0782032 = 0.00483783 + lam * -4.15205, offending 61911
  Forward pass
  Distances
  Train
epoch 60, times: [hn 4.13 s epoch 56.46 s val 0.00 s] lr = 0.100000 loss = -0.0782075 = 0.00484088 + lam * -4.15242, offending 62426
  Forward pass
  Distances
  Train
epoch 61, times: [hn 4.03 s epoch 63.73 s val 0.00 s] lr = 0.100000 loss = -0.0782767 = 0.00477678 + lam * -4.15267, offending 61610
  Forward pass
  Distances
  Train
epoch 62, times: [hn 4.09 s epoch 63.97 s val 0.00 s] lr = 0.100000 loss = -0.0782585 = 0.0047874 + lam * -4.1523, offending 62226
  Forward pass
  Distances
  Train
epoch 63, times: [hn 4.06 s epoch 61.01 s val 0.00 s] lr = 0.100000 loss = -0.0782411 = 0.00480494 + lam * -4.1523, offending 62060
  Forward pass
  Distances
  Train
epoch 64, times: [hn 4.05 s epoch 49.57 s val 0.00 s] lr = 0.100000 loss = -0.0782487 = 0.0048089 + lam * -4.15288, offending 62115
  Forward pass
  Distances
  Train
epoch 65, times: [hn 4.04 s epoch 53.45 s val 0.00 s] lr = 0.100000 loss = -0.0783337 = 0.00472417 + lam * -4.15289, offending 61327
  Forward pass
  Distances
  Train
epoch 66, times: [hn 4.05 s epoch 50.47 s val 0.00 s] lr = 0.100000 loss = -0.0783735 = 0.00469867 + lam * -4.15361, offending 61282
  Forward pass
  Distances
  Train
epoch 67, times: [hn 4.07 s epoch 55.59 s val 0.00 s] lr = 0.100000 loss = -0.078326 = 0.00472577 + lam * -4.15259, offending 61245
  Forward pass
  Distances
  Train
epoch 68, times: [hn 4.07 s epoch 58.97 s val 0.00 s] lr = 0.100000 loss = -0.0783707 = 0.00470154 + lam * -4.15361, offending 61273
  Forward pass
  Distances
  Train
Valiation at epoch 69
zn_3     nbit= 14:  0.0013 0.0125 0.1015
zn_10    nbit= 32:  0.0507 0.2329 0.6212
zn_79    nbit= 64:  0.2218 0.6526 0.9504
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.65263 > 0.65133), keeping as best
epoch 69, times: [hn 4.05 s epoch 53.20 s val 9.96 s] lr = 0.100000 loss = -0.0783555 = 0.0047047 + lam * -4.15301, offending 61303
  Forward pass
  Distances
  Train
epoch 70, times: [hn 4.06 s epoch 55.19 s val 0.00 s] lr = 0.100000 loss = -0.0783273 = 0.00474302 + lam * -4.15352, offending 61569
  Forward pass
  Distances
  Train
epoch 71, times: [hn 4.05 s epoch 56.42 s val 0.00 s] lr = 0.100000 loss = -0.0783654 = 0.00472873 + lam * -4.1547, offending 61394
  Forward pass
  Distances
  Train
epoch 72, times: [hn 4.06 s epoch 46.84 s val 0.00 s] lr = 0.100000 loss = -0.0784196 = 0.00466164 + lam * -4.15406, offending 60945
  Forward pass
  Distances
  Train
epoch 73, times: [hn 3.97 s epoch 48.80 s val 0.00 s] lr = 0.100000 loss = -0.0783726 = 0.00471487 + lam * -4.15437, offending 61036
  Forward pass
  Distances
  Train
epoch 74, times: [hn 4.05 s epoch 51.81 s val 0.00 s] lr = 0.100000 loss = -0.0784191 = 0.00465683 + lam * -4.1538, offending 61020
  Forward pass
  Distances
  Train
epoch 75, times: [hn 4.21 s epoch 54.63 s val 0.00 s] lr = 0.100000 loss = -0.0784389 = 0.00464387 + lam * -4.15414, offending 60764
  Forward pass
  Distances
  Train
epoch 76, times: [hn 4.07 s epoch 52.77 s val 0.00 s] lr = 0.100000 loss = -0.0784501 = 0.00463605 + lam * -4.15431, offending 60712
  Forward pass
  Distances
  Train
epoch 77, times: [hn 4.10 s epoch 56.40 s val 0.00 s] lr = 0.100000 loss = -0.0784663 = 0.00461885 + lam * -4.15426, offending 60499
  Forward pass
  Distances
  Train
epoch 78, times: [hn 4.07 s epoch 54.41 s val 0.00 s] lr = 0.100000 loss = -0.0784515 = 0.00463306 + lam * -4.15423, offending 60560
  Forward pass
  Distances
  Train
Valiation at epoch 79
zn_3     nbit= 14:  0.0015 0.0122 0.1018
zn_10    nbit= 32:  0.0527 0.2354 0.6227
zn_79    nbit= 64:  0.2218 0.6539 0.9506
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.65393 > 0.65263), keeping as best
epoch 79, times: [hn 4.19 s epoch 58.29 s val 9.95 s] lr = 0.100000 loss = -0.0784953 = 0.00459732 + lam * -4.15463, offending 60295
  Forward pass
  Distances
  Train
epoch 80, times: [hn 4.12 s epoch 57.00 s val 0.00 s] lr = 0.050000 loss = -0.0786254 = 0.00448272 + lam * -4.15541, offending 59983
  Forward pass
  Distances
  Train
epoch 81, times: [hn 4.06 s epoch 53.50 s val 0.00 s] lr = 0.050000 loss = -0.0787055 = 0.00441559 + lam * -4.15605, offending 59427
  Forward pass
  Distances
  Train
epoch 82, times: [hn 4.05 s epoch 55.46 s val 0.00 s] lr = 0.050000 loss = -0.07878 = 0.00435403 + lam * -4.1567, offending 58671
  Forward pass
  Distances
  Train
epoch 83, times: [hn 4.06 s epoch 55.35 s val 0.00 s] lr = 0.050000 loss = -0.0787422 = 0.00437216 + lam * -4.15572, offending 59104
  Forward pass
  Distances
  Train
epoch 84, times: [hn 4.19 s epoch 51.81 s val 0.00 s] lr = 0.050000 loss = -0.0787031 = 0.0044093 + lam * -4.15562, offending 59348
  Forward pass
  Distances
  Train
epoch 85, times: [hn 4.07 s epoch 57.52 s val 0.00 s] lr = 0.050000 loss = -0.0788475 = 0.00428073 + lam * -4.15641, offending 58085
  Forward pass
  Distances
  Train
epoch 86, times: [hn 4.06 s epoch 55.20 s val 0.00 s] lr = 0.050000 loss = -0.0787583 = 0.00436084 + lam * -4.15596, offending 58798
  Forward pass
  Distances
  Train
epoch 87, times: [hn 4.03 s epoch 52.91 s val 0.00 s] lr = 0.050000 loss = -0.0787763 = 0.00434853 + lam * -4.15624, offending 58764
  Forward pass
  Distances
  Train
epoch 88, times: [hn 3.99 s epoch 56.17 s val 0.00 s] lr = 0.050000 loss = -0.0788232 = 0.00431108 + lam * -4.15672, offending 58328
  Forward pass
  Distances
  Train
Valiation at epoch 89
zn_3     nbit= 14:  0.0012 0.0125 0.1015
zn_10    nbit= 32:  0.0525 0.2384 0.6258
zn_79    nbit= 64:  0.2236 0.6582 0.9521
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.6582 > 0.65393), keeping as best
epoch 89, times: [hn 4.06 s epoch 53.38 s val 9.84 s] lr = 0.050000 loss = -0.0788609 = 0.00428576 + lam * -4.15733, offending 58052
  Forward pass
  Distances
  Train
epoch 90, times: [hn 4.13 s epoch 56.84 s val 0.00 s] lr = 0.050000 loss = -0.0788459 = 0.00428373 + lam * -4.15648, offending 58283
  Forward pass
  Distances
  Train
epoch 91, times: [hn 4.05 s epoch 50.47 s val 0.00 s] lr = 0.050000 loss = -0.0788635 = 0.00427688 + lam * -4.15702, offending 58155
  Forward pass
  Distances
  Train
epoch 92, times: [hn 4.05 s epoch 55.20 s val 0.00 s] lr = 0.050000 loss = -0.0788603 = 0.00428634 + lam * -4.15733, offending 58141
  Forward pass
  Distances
  Train
epoch 93, times: [hn 4.05 s epoch 55.55 s val 0.00 s] lr = 0.050000 loss = -0.0789434 = 0.00420911 + lam * -4.15763, offending 57296
  Forward pass
  Distances
  Train
epoch 94, times: [hn 4.04 s epoch 53.65 s val 0.00 s] lr = 0.050000 loss = -0.0789498 = 0.00420209 + lam * -4.15759, offending 57458
  Forward pass
  Distances
  Train
epoch 95, times: [hn 4.05 s epoch 58.96 s val 0.00 s] lr = 0.050000 loss = -0.0788947 = 0.00424246 + lam * -4.15686, offending 57984
  Forward pass
  Distances
  Train
epoch 96, times: [hn 4.04 s epoch 55.59 s val 0.00 s] lr = 0.050000 loss = -0.0789185 = 0.0042302 + lam * -4.15744, offending 57920
  Forward pass
  Distances
  Train
epoch 97, times: [hn 4.05 s epoch 58.41 s val 0.00 s] lr = 0.050000 loss = -0.078878 = 0.00425725 + lam * -4.15676, offending 57939
  Forward pass
  Distances
  Train
epoch 98, times: [hn 4.06 s epoch 52.40 s val 0.00 s] lr = 0.050000 loss = -0.0788626 = 0.00429499 + lam * -4.15788, offending 58384
  Forward pass
  Distances
  Train
Valiation at epoch 99
zn_3     nbit= 14:  0.0014 0.0127 0.1029
zn_10    nbit= 32:  0.0535 0.2397 0.6292
zn_79    nbit= 64:  0.2261 0.6612 0.9544
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.66124 > 0.6582), keeping as best
epoch 99, times: [hn 4.10 s epoch 54.18 s val 9.98 s] lr = 0.050000 loss = -0.0789851 = 0.00417802 + lam * -4.15815, offending 57290
  Forward pass
  Distances
  Train
epoch 100, times: [hn 4.15 s epoch 58.49 s val 0.00 s] lr = 0.050000 loss = -0.0788865 = 0.0042511 + lam * -4.15688, offending 58001
  Forward pass
  Distances
  Train
epoch 101, times: [hn 4.07 s epoch 52.71 s val 0.00 s] lr = 0.050000 loss = -0.0789384 = 0.00420927 + lam * -4.15738, offending 57439
  Forward pass
  Distances
  Train
epoch 102, times: [hn 4.07 s epoch 52.19 s val 0.00 s] lr = 0.050000 loss = -0.0789653 = 0.00419137 + lam * -4.15784, offending 56970
  Forward pass
  Distances
  Train
epoch 103, times: [hn 4.04 s epoch 56.55 s val 0.00 s] lr = 0.050000 loss = -0.0789748 = 0.00417612 + lam * -4.15755, offending 57219
  Forward pass
  Distances
  Train
epoch 104, times: [hn 4.00 s epoch 55.88 s val 0.00 s] lr = 0.050000 loss = -0.0789315 = 0.00421418 + lam * -4.15728, offending 57668
  Forward pass
  Distances
  Train
epoch 105, times: [hn 4.07 s epoch 55.15 s val 0.00 s] lr = 0.050000 loss = -0.0789942 = 0.00415893 + lam * -4.15766, offending 56750
  Forward pass
  Distances
  Train
epoch 106, times: [hn 4.06 s epoch 49.24 s val 0.00 s] lr = 0.050000 loss = -0.0789537 = 0.00419328 + lam * -4.15735, offending 57157
  Forward pass
  Distances
  Train
epoch 107, times: [hn 4.07 s epoch 53.55 s val 0.00 s] lr = 0.050000 loss = -0.0789343 = 0.00421539 + lam * -4.15748, offending 57650
  Forward pass
  Distances
  Train
epoch 108, times: [hn 4.04 s epoch 55.64 s val 0.00 s] lr = 0.050000 loss = -0.0789493 = 0.00420185 + lam * -4.15756, offending 57328
  Forward pass
  Distances
  Train
Valiation at epoch 109
zn_3     nbit= 14:  0.0014 0.0127 0.1028
zn_10    nbit= 32:  0.0549 0.2404 0.6321
zn_79    nbit= 64:  0.2263 0.6618 0.9529
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.66182 > 0.66124), keeping as best
epoch 109, times: [hn 4.13 s epoch 52.49 s val 10.01 s] lr = 0.050000 loss = -0.0789786 = 0.00416925 + lam * -4.15739, offending 57229
  Forward pass
  Distances
  Train
epoch 110, times: [hn 4.13 s epoch 46.61 s val 0.00 s] lr = 0.050000 loss = -0.078975 = 0.00418207 + lam * -4.15785, offending 57545
  Forward pass
  Distances
  Train
epoch 111, times: [hn 4.03 s epoch 57.61 s val 0.00 s] lr = 0.050000 loss = -0.079049 = 0.00410672 + lam * -4.15779, offending 56685
  Forward pass
  Distances
  Train
epoch 112, times: [hn 4.05 s epoch 50.91 s val 0.00 s] lr = 0.050000 loss = -0.0790423 = 0.0041316 + lam * -4.1587, offending 56860
  Forward pass
  Distances
  Train
epoch 113, times: [hn 4.04 s epoch 53.97 s val 0.00 s] lr = 0.050000 loss = -0.0790623 = 0.00410668 + lam * -4.15845, offending 56714
  Forward pass
  Distances
  Train
epoch 114, times: [hn 4.06 s epoch 51.42 s val 0.00 s] lr = 0.050000 loss = -0.0790243 = 0.00414819 + lam * -4.15863, offending 56802
  Forward pass
  Distances
  Train
epoch 115, times: [hn 4.08 s epoch 53.69 s val 0.00 s] lr = 0.050000 loss = -0.0790338 = 0.00413321 + lam * -4.15835, offending 56852
  Forward pass
  Distances
  Train
epoch 116, times: [hn 4.08 s epoch 50.23 s val 0.00 s] lr = 0.050000 loss = -0.0790415 = 0.00413085 + lam * -4.15862, offending 57049
  Forward pass
  Distances
  Train
epoch 117, times: [hn 4.02 s epoch 54.43 s val 0.00 s] lr = 0.050000 loss = -0.0790543 = 0.00412129 + lam * -4.15878, offending 56963
  Forward pass
  Distances
  Train
epoch 118, times: [hn 4.04 s epoch 48.26 s val 0.00 s] lr = 0.050000 loss = -0.0790523 = 0.00411141 + lam * -4.15819, offending 56454
  Forward pass
  Distances
  Train
Valiation at epoch 119
zn_3     nbit= 14:  0.0015 0.0134 0.1035
zn_10    nbit= 32:  0.0521 0.2353 0.6250
zn_79    nbit= 64:  0.2254 0.6616 0.9535
storing test_ckpt/0.02/checkpoint.pth
epoch 119, times: [hn 4.04 s epoch 53.12 s val 9.60 s] lr = 0.050000 loss = -0.0790143 = 0.00414699 + lam * -4.15807, offending 57208
  Forward pass
  Distances
  Train
epoch 120, times: [hn 4.15 s epoch 55.82 s val 0.00 s] lr = 0.010000 loss = -0.0791322 = 0.00403971 + lam * -4.1586, offending 56132
  Forward pass
  Distances
  Train
epoch 121, times: [hn 4.05 s epoch 58.65 s val 0.00 s] lr = 0.010000 loss = -0.0791921 = 0.0039774 + lam * -4.15847, offending 55728
  Forward pass
  Distances
  Train
epoch 122, times: [hn 4.14 s epoch 51.89 s val 0.00 s] lr = 0.010000 loss = -0.0792193 = 0.00396435 + lam * -4.15918, offending 55840
  Forward pass
  Distances
  Train
epoch 123, times: [hn 4.07 s epoch 47.98 s val 0.00 s] lr = 0.010000 loss = -0.079204 = 0.00398292 + lam * -4.15935, offending 55909
  Forward pass
  Distances
  Train
epoch 124, times: [hn 4.05 s epoch 53.63 s val 0.00 s] lr = 0.010000 loss = -0.0792986 = 0.00388633 + lam * -4.15925, offending 55172
  Forward pass
  Distances
  Train
epoch 125, times: [hn 4.05 s epoch 51.31 s val 0.00 s] lr = 0.010000 loss = -0.0792449 = 0.00393592 + lam * -4.15904, offending 55612
  Forward pass
  Distances
  Train
epoch 126, times: [hn 4.04 s epoch 53.77 s val 0.00 s] lr = 0.010000 loss = -0.0792558 = 0.0039297 + lam * -4.15927, offending 55571
  Forward pass
  Distances
  Train
epoch 127, times: [hn 4.06 s epoch 53.16 s val 0.00 s] lr = 0.010000 loss = -0.0792241 = 0.00396173 + lam * -4.15929, offending 55696
  Forward pass
  Distances
  Train
epoch 128, times: [hn 4.04 s epoch 50.20 s val 0.00 s] lr = 0.010000 loss = -0.0792729 = 0.00392836 + lam * -4.16006, offending 55059
  Forward pass
  Distances
  Train
Valiation at epoch 129
zn_3     nbit= 14:  0.0015 0.0130 0.1031
zn_10    nbit= 32:  0.0538 0.2446 0.6333
zn_79    nbit= 64:  0.2284 0.6647 0.9541
storing test_ckpt/0.02/checkpoint.pth
zn_79,rank=10 score improves (0.66471 > 0.66182), keeping as best
epoch 129, times: [hn 4.17 s epoch 52.78 s val 9.82 s] lr = 0.010000 loss = -0.0792488 = 0.00394092 + lam * -4.15949, offending 55712
  Forward pass
  Distances
  Train
epoch 130, times: [hn 4.15 s epoch 52.53 s val 0.00 s] lr = 0.010000 loss = -0.079212 = 0.00397315 + lam * -4.15926, offending 55580
  Forward pass
  Distances
  Train
epoch 131, times: [hn 4.10 s epoch 48.28 s val 0.00 s] lr = 0.010000 loss = -0.0792781 = 0.00391176 + lam * -4.15949, offending 55234
  Forward pass
  Distances
  Train
epoch 132, times: [hn 4.03 s epoch 52.01 s val 0.00 s] lr = 0.010000 loss = -0.0793007 = 0.00389158 + lam * -4.15962, offending 55109
  Forward pass
  Distances
  Train
epoch 133, times: [hn 4.02 s epoch 44.79 s val 0.00 s] lr = 0.010000 loss = -0.0792929 = 0.00389802 + lam * -4.15954, offending 55184
  Forward pass
  Distances
  Train
epoch 134, times: [hn 4.10 s epoch 55.67 s val 0.00 s] lr = 0.010000 loss = -0.0793511 = 0.00385616 + lam * -4.16036, offending 54762
  Forward pass
  Distances
  Train
epoch 135, times: [hn 4.05 s epoch 57.06 s val 0.00 s] lr = 0.010000 loss = -0.0793227 = 0.00387606 + lam * -4.15994, offending 54922
  Forward pass
  Distances
  Train
epoch 136, times: [hn 4.05 s epoch 51.45 s val 0.00 s] lr = 0.010000 loss = -0.0792344 = 0.00397065 + lam * -4.16025, offending 55449
  Forward pass
  Distances
  Train
epoch 137, times: [hn 4.05 s epoch 54.38 s val 0.00 s] lr = 0.010000 loss = -0.0793015 = 0.00388287 + lam * -4.15922, offending 54876
  Forward pass
  Distances
  Train
epoch 138, times: [hn 4.06 s epoch 57.36 s val 0.00 s] lr = 0.010000 loss = -0.0793277 = 0.00387094 + lam * -4.15993, offending 54795
  Forward pass
  Distances
  Train
Valiation at epoch 139
zn_3     nbit= 14:  0.0013 0.0132 0.1042
zn_10    nbit= 32:  0.0527 0.2382 0.6255
zn_79    nbit= 64:  0.2259 0.6635 0.9545
storing test_ckpt/0.02/checkpoint.pth
epoch 139, times: [hn 4.05 s epoch 44.51 s val 9.72 s] lr = 0.010000 loss = -0.0793093 = 0.00388971 + lam * -4.15995, offending 55071
  Forward pass
  Distances
  Train
epoch 140, times: [hn 4.13 s epoch 55.01 s val 0.00 s] lr = 0.010000 loss = -0.079337 = 0.0038596 + lam * -4.15983, offending 54587
  Forward pass
  Distances
  Train
epoch 141, times: [hn 4.04 s epoch 51.45 s val 0.00 s] lr = 0.010000 loss = -0.0792535 = 0.00394336 + lam * -4.15984, offending 55546
  Forward pass
  Distances
  Train
epoch 142, times: [hn 4.06 s epoch 50.85 s val 0.00 s] lr = 0.010000 loss = -0.0793415 = 0.00386206 + lam * -4.16018, offending 54698
  Forward pass
  Distances
  Train
epoch 143, times: [hn 4.06 s epoch 51.54 s val 0.00 s] lr = 0.010000 loss = -0.0793552 = 0.00384975 + lam * -4.16025, offending 54708
  Forward pass
  Distances
  Train
epoch 144, times: [hn 4.04 s epoch 48.26 s val 0.00 s] lr = 0.010000 loss = -0.0793219 = 0.0038842 + lam * -4.16031, offending 54800
  Forward pass
  Distances
  Train
epoch 145, times: [hn 4.05 s epoch 50.48 s val 0.00 s] lr = 0.010000 loss = -0.0793566 = 0.00385893 + lam * -4.16078, offending 54493
  Forward pass
  Distances
  Train
epoch 146, times: [hn 4.04 s epoch 54.15 s val 0.00 s] lr = 0.010000 loss = -0.0793305 = 0.00387234 + lam * -4.16014, offending 55234
  Forward pass
  Distances
  Train
epoch 147, times: [hn 4.04 s epoch 45.46 s val 0.00 s] lr = 0.010000 loss = -0.0793638 = 0.00384336 + lam * -4.16036, offending 54524
  Forward pass
  Distances
  Train
epoch 148, times: [hn 4.04 s epoch 55.87 s val 0.00 s] lr = 0.010000 loss = -0.0793959 = 0.00381287 + lam * -4.16044, offending 54246
  Forward pass
  Distances
  Train
Valiation at epoch 149
zn_3     nbit= 14:  0.0013 0.0129 0.1040
zn_10    nbit= 32:  0.0539 0.2406 0.6291
zn_79    nbit= 64:  0.2272 0.6642 0.9552
storing test_ckpt/0.02/checkpoint.pth
epoch 149, times: [hn 4.05 s epoch 48.09 s val 9.99 s] lr = 0.010000 loss = -0.0793678 = 0.00384299 + lam * -4.16054, offending 54527
  Forward pass
  Distances
  Train
epoch 150, times: [hn 4.03 s epoch 44.51 s val 0.00 s] lr = 0.010000 loss = -0.0793849 = 0.00383087 + lam * -4.16079, offending 54614
  Forward pass
  Distances
  Train
epoch 151, times: [hn 4.05 s epoch 53.04 s val 0.00 s] lr = 0.010000 loss = -0.0793342 = 0.0038728 + lam * -4.16035, offending 54880
  Forward pass
  Distances
  Train
epoch 152, times: [hn 4.05 s epoch 47.54 s val 0.00 s] lr = 0.010000 loss = -0.0793551 = 0.00385226 + lam * -4.16037, offending 55147
  Forward pass
  Distances
  Train
epoch 153, times: [hn 4.04 s epoch 47.43 s val 0.00 s] lr = 0.010000 loss = -0.0793588 = 0.00384845 + lam * -4.16036, offending 54577
  Forward pass
  Distances
  Train
epoch 154, times: [hn 4.01 s epoch 48.72 s val 0.00 s] lr = 0.010000 loss = -0.0793399 = 0.00385716 + lam * -4.15985, offending 54435
  Forward pass
  Distances
  Train
epoch 155, times: [hn 4.02 s epoch 49.61 s val 0.00 s] lr = 0.010000 loss = -0.0793829 = 0.00383023 + lam * -4.16066, offending 54131
  Forward pass
  Distances
  Train
epoch 156, times: [hn 3.97 s epoch 53.14 s val 0.00 s] lr = 0.010000 loss = -0.0794226 = 0.00379752 + lam * -4.16101, offending 54176
  Forward pass
  Distances
  Train
epoch 157, times: [hn 4.07 s epoch 54.50 s val 0.00 s] lr = 0.010000 loss = -0.0793504 = 0.00385288 + lam * -4.16016, offending 54814
  Forward pass
  Distances
  Train
epoch 158, times: [hn 4.03 s epoch 51.57 s val 0.00 s] lr = 0.010000 loss = -0.0793613 = 0.00384946 + lam * -4.16054, offending 54627
  Forward pass
  Distances
  Train
Valiation at epoch 159
zn_3     nbit= 14:  0.0014 0.0126 0.1032
zn_10    nbit= 32:  0.0527 0.2367 0.6254
zn_79    nbit= 64:  0.2263 0.6626 0.9545
storing test_ckpt/0.02/checkpoint.pth
epoch 159, times: [hn 4.04 s epoch 46.70 s val 9.76 s] lr = 0.010000 loss = -0.0793505 = 0.00385487 + lam * -4.16027, offending 54873
++ tac test_ckpt/0.02.stdout
++ grep -m1 'keeping as best'
++ grep -o '[0-9\.]*'
++ grep -o '(.*>'
+ perf=0.66471
+ echo 0.66471
0.66471
+ [[ 0.66471 > 0.66498 ]]
+ for lambda in '$lambdas'
+ mkdir -p test_ckpt/0.05
+ python -u train.py --dout 24 --save_best_criterion zn_79,rank=10 --database deep1b --lambda_uniform 0.05 --checkpoint_dir test_ckpt/0.05
+ tee test_ckpt/0.05.stdout
Namespace(batch_size=64, checkpoint_dir='test_ckpt/0.05', database='deep1b', device='cuda', dint=1024, dout=24, epochs=160, init_name='', lambda_uniform=0.05,
 lr_schedule='0.1,0.1,0.05,0.01', momentum=0.9, num_learn=500000, quantizer_train='', rank_negative=50, rank_positive=10, save_best_criterion='zn_79,rank=10',
 seed=1234, size_base=1000000, val_freq=10, validation_quantizers=['zn_3', 'zn_10', 'zn_79'])
load dataset deep1b
keeping 500000/357380000 training vectors
computing training ground truth
build network
Lr schedule [0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1,
0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1
, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0
.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.0
5, 0.05, 0.05, 0.05, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01,
 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01, 0.01]
  Forward pass
  Distances
  Train
epoch 0, times: [hn 3.96 s epoch 53.04 s val 0.00 s] lr = 0.100000 loss = -0.20596 = 0.00278932 + lam * -4.17499, offending 18434
  Forward pass
  Distances
  Train
epoch 1, times: [hn 4.06 s epoch 52.01 s val 0.00 s] lr = 0.100000 loss = -0.201598 = 0.00644614 + lam * -4.16089, offending 47229
  Forward pass
  Distances
  Train
epoch 2, times: [hn 4.09 s epoch 48.00 s val 0.00 s] lr = 0.100000 loss = -0.201049 = 0.00722991 + lam * -4.16558, offending 57258
  Forward pass
  Distances
  Train
epoch 3, times: [hn 4.04 s epoch 51.06 s val 0.00 s] lr = 0.100000 loss = -0.200535 = 0.00763391 + lam * -4.16337, offending 64865
  Forward pass
  Distances
  Train
epoch 4, times: [hn 4.20 s epoch 56.89 s val 0.00 s] lr = 0.100000 loss = -0.200368 = 0.00787357 + lam * -4.16483, offending 69520
  Forward pass
  Distances
  Train
epoch 5, times: [hn 4.05 s epoch 56.75 s val 0.00 s] lr = 0.100000 loss = -0.200369 = 0.0078809 + lam * -4.165, offending 72712
  Forward pass
  Distances
  Train
epoch 6, times: [hn 4.05 s epoch 56.24 s val 0.00 s] lr = 0.100000 loss = -0.200456 = 0.00788203 + lam * -4.16677, offending 75300
  Forward pass
  Distances
  Train
epoch 7, times: [hn 4.10 s epoch 54.56 s val 0.00 s] lr = 0.100000 loss = -0.200448 = 0.00787535 + lam * -4.16647, offending 77009
  Forward pass
  Distances
  Train
epoch 8, times: [hn 4.03 s epoch 50.78 s val 0.00 s] lr = 0.100000 loss = -0.200643 = 0.00773296 + lam * -4.16751, offending 77792
  Forward pass
  Distances
  Train
Valiation at epoch 9
zn_3     nbit= 14:  0.0012 0.0117 0.0962
zn_10    nbit= 32:  0.0499 0.2252 0.6020
zn_79    nbit= 64:  0.1961 0.5949 0.9215
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.59491 > 0), keeping as best
epoch 9, times: [hn 3.98 s epoch 58.09 s val 10.09 s] lr = 0.100000 loss = -0.200726 = 0.00768934 + lam * -4.16831, offending 78165
  Forward pass
  Distances
  Train
epoch 10, times: [hn 4.15 s epoch 53.52 s val 0.00 s] lr = 0.100000 loss = -0.200916 = 0.00752349 + lam * -4.16879, offending 77790
  Forward pass
  Distances
  Train
epoch 11, times: [hn 4.06 s epoch 48.64 s val 0.00 s] lr = 0.100000 loss = -0.201082 = 0.00739277 + lam * -4.1695, offending 76858
  Forward pass
  Distances
  Train
epoch 12, times: [hn 4.03 s epoch 54.09 s val 0.00 s] lr = 0.100000 loss = -0.201245 = 0.00726043 + lam * -4.17011, offending 76674
  Forward pass
  Distances
  Train
epoch 13, times: [hn 4.03 s epoch 56.40 s val 0.00 s] lr = 0.100000 loss = -0.201363 = 0.00717334 + lam * -4.17073, offending 76474
  Forward pass
  Distances
  Train
epoch 14, times: [hn 4.11 s epoch 50.43 s val 0.00 s] lr = 0.100000 loss = -0.201491 = 0.00704904 + lam * -4.1708, offending 75970
  Forward pass
  Distances
  Train
epoch 15, times: [hn 4.04 s epoch 55.07 s val 0.00 s] lr = 0.100000 loss = -0.201563 = 0.00700236 + lam * -4.1713, offending 75212
  Forward pass
  Distances
  Train
epoch 16, times: [hn 4.06 s epoch 51.97 s val 0.00 s] lr = 0.100000 loss = -0.201718 = 0.00685893 + lam * -4.17154, offending 75082
  Forward pass
  Distances
  Train
epoch 17, times: [hn 4.02 s epoch 55.68 s val 0.00 s] lr = 0.100000 loss = -0.201813 = 0.00680462 + lam * -4.17235, offending 75035
  Forward pass
  Distances
  Train
epoch 18, times: [hn 4.05 s epoch 56.74 s val 0.00 s] lr = 0.100000 loss = -0.201837 = 0.00675611 + lam * -4.17186, offending 74583
  Forward pass
  Distances
  Train
Valiation at epoch 19
zn_3     nbit= 14:  0.0012 0.0124 0.1016
zn_10    nbit= 32:  0.0511 0.2317 0.6191
zn_79    nbit= 64:  0.2072 0.6259 0.9371
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.62588 > 0.59491), keeping as best
epoch 19, times: [hn 4.04 s epoch 45.13 s val 10.31 s] lr = 0.100000 loss = -0.201968 = 0.00668408 + lam * -4.17304, offending 74209
  Forward pass
  Distances
  Train
epoch 20, times: [hn 4.17 s epoch 55.61 s val 0.00 s] lr = 0.100000 loss = -0.202045 = 0.00662065 + lam * -4.1733, offending 74053
  Forward pass
  Distances
  Train
epoch 21, times: [hn 4.04 s epoch 53.57 s val 0.00 s] lr = 0.100000 loss = -0.2021 = 0.00655141 + lam * -4.17303, offending 73493
  Forward pass
  Distances
  Train
epoch 22, times: [hn 4.17 s epoch 56.63 s val 0.00 s] lr = 0.100000 loss = -0.202228 = 0.006497 + lam * -4.17451, offending 72658
  Forward pass
  Distances
  Train
epoch 23, times: [hn 4.02 s epoch 57.88 s val 0.00 s] lr = 0.100000 loss = -0.202259 = 0.0064521 + lam * -4.17422, offending 72896
  Forward pass
  Distances
  Train
epoch 24, times: [hn 4.05 s epoch 54.92 s val 0.00 s] lr = 0.100000 loss = -0.202353 = 0.00638088 + lam * -4.17468, offending 72213
  Forward pass
  Distances
  Train
epoch 25, times: [hn 4.01 s epoch 60.97 s val 0.00 s] lr = 0.100000 loss = -0.202402 = 0.00632537 + lam * -4.17456, offending 71797
  Forward pass
  Distances
  Train
epoch 26, times: [hn 4.17 s epoch 55.68 s val 0.00 s] lr = 0.100000 loss = -0.202431 = 0.00632562 + lam * -4.17514, offending 72407
  Forward pass
  Distances
  Train
epoch 27, times: [hn 4.04 s epoch 50.30 s val 0.00 s] lr = 0.100000 loss = -0.20245 = 0.00628439 + lam * -4.1747, offending 71659
  Forward pass
  Distances
  Train
epoch 28, times: [hn 4.05 s epoch 52.25 s val 0.00 s] lr = 0.100000 loss = -0.202537 = 0.00625082 + lam * -4.17576, offending 71637
  Forward pass
  Distances
  Train
Valiation at epoch 29
zn_3     nbit= 14:  0.0014 0.0123 0.1032
zn_10    nbit= 32:  0.0525 0.2386 0.6242
zn_79    nbit= 64:  0.2115 0.6344 0.9399
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.63444 > 0.62588), keeping as best
epoch 29, times: [hn 4.06 s epoch 48.92 s val 9.75 s] lr = 0.100000 loss = -0.202582 = 0.00619452 + lam * -4.17553, offending 70872
  Forward pass
  Distances
  Train
epoch 30, times: [hn 4.02 s epoch 59.00 s val 0.00 s] lr = 0.100000 loss = -0.202644 = 0.00615329 + lam * -4.17594, offending 70515
  Forward pass
  Distances
  Train
epoch 31, times: [hn 4.06 s epoch 50.67 s val 0.00 s] lr = 0.100000 loss = -0.202711 = 0.00608914 + lam * -4.17601, offending 70355
  Forward pass
  Distances
  Train
epoch 32, times: [hn 3.99 s epoch 57.24 s val 0.00 s] lr = 0.100000 loss = -0.202748 = 0.0060661 + lam * -4.17629, offending 70422
  Forward pass
  Distances
  Train
epoch 33, times: [hn 4.08 s epoch 59.19 s val 0.00 s] lr = 0.100000 loss = -0.202825 = 0.00599622 + lam * -4.17642, offending 69480
  Forward pass
  Distances
  Train
epoch 34, times: [hn 4.08 s epoch 57.51 s val 0.00 s] lr = 0.100000 loss = -0.202787 = 0.00603659 + lam * -4.17648, offending 69973
  Forward pass
  Distances
  Train
epoch 35, times: [hn 4.05 s epoch 41.47 s val 0.00 s] lr = 0.100000 loss = -0.202935 = 0.00591645 + lam * -4.17702, offending 68597
  Forward pass
  Distances
  Train
epoch 36, times: [hn 4.04 s epoch 41.30 s val 0.00 s] lr = 0.100000 loss = -0.202919 = 0.00595622 + lam * -4.1775, offending 69143
  Forward pass
  Distances
  Train
epoch 37, times: [hn 4.05 s epoch 41.69 s val 0.00 s] lr = 0.100000 loss = -0.202904 = 0.00593619 + lam * -4.1768, offending 69339
  Forward pass
  Distances
  Train
epoch 38, times: [hn 4.04 s epoch 41.45 s val 0.00 s] lr = 0.100000 loss = -0.202917 = 0.00591136 + lam * -4.17656, offending 69236
  Forward pass
  Distances
  Train
Valiation at epoch 39
zn_3     nbit= 14:  0.0011 0.0119 0.1026
zn_10    nbit= 32:  0.0545 0.2420 0.6330
zn_79    nbit= 64:  0.2179 0.6424 0.9425
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.64245 > 0.63444), keeping as best
epoch 39, times: [hn 4.02 s epoch 40.91 s val 10.01 s] lr = 0.100000 loss = -0.20307 = 0.00583693 + lam * -4.17814, offending 68832
  Forward pass
  Distances
  Train
epoch 40, times: [hn 4.07 s epoch 41.87 s val 0.00 s] lr = 0.100000 loss = -0.203049 = 0.00584613 + lam * -4.1779, offending 68782
  Forward pass
  Distances
  Train
epoch 41, times: [hn 4.05 s epoch 40.97 s val 0.00 s] lr = 0.100000 loss = -0.203039 = 0.00585032 + lam * -4.17779, offending 68801
  Forward pass
  Distances
  Train
epoch 42, times: [hn 4.06 s epoch 41.68 s val 0.00 s] lr = 0.100000 loss = -0.2031 = 0.00581355 + lam * -4.17826, offending 68134
  Forward pass
  Distances
  Train
epoch 43, times: [hn 4.07 s epoch 41.22 s val 0.00 s] lr = 0.100000 loss = -0.203125 = 0.00577664 + lam * -4.17802, offending 68315
  Forward pass
  Distances
  Train
epoch 44, times: [hn 4.03 s epoch 41.46 s val 0.00 s] lr = 0.100000 loss = -0.203098 = 0.00579865 + lam * -4.17793, offending 68395
  Forward pass
  Distances
  Train
epoch 45, times: [hn 4.03 s epoch 41.29 s val 0.00 s] lr = 0.100000 loss = -0.203222 = 0.00570279 + lam * -4.17849, offending 67722
  Forward pass
  Distances
  Train
epoch 46, times: [hn 4.04 s epoch 41.90 s val 0.00 s] lr = 0.100000 loss = -0.203238 = 0.00567971 + lam * -4.17835, offending 67540
  Forward pass
  Distances
  Train
epoch 47, times: [hn 4.05 s epoch 41.34 s val 0.00 s] lr = 0.100000 loss = -0.203249 = 0.00567687 + lam * -4.17851, offending 67478
  Forward pass
  Distances
  Train
epoch 48, times: [hn 4.04 s epoch 41.30 s val 0.00 s] lr = 0.100000 loss = -0.203256 = 0.0056916 + lam * -4.17895, offending 67485
  Forward pass
  Distances
  Train
Valiation at epoch 49
zn_3     nbit= 14:  0.0013 0.0127 0.1056
zn_10    nbit= 32:  0.0540 0.2427 0.6321
zn_79    nbit= 64:  0.2168 0.6449 0.9453
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.64491 > 0.64245), keeping as best
epoch 49, times: [hn 4.04 s epoch 39.42 s val 9.60 s] lr = 0.100000 loss = -0.203279 = 0.0056775 + lam * -4.17913, offending 67484
  Forward pass
  Distances
  Train
epoch 50, times: [hn 4.03 s epoch 40.40 s val 0.00 s] lr = 0.100000 loss = -0.203355 = 0.00561467 + lam * -4.17939, offending 67068
  Forward pass
  Distances
  Train
epoch 51, times: [hn 4.05 s epoch 40.01 s val 0.00 s] lr = 0.100000 loss = -0.203366 = 0.0056202 + lam * -4.17972, offending 67116
  Forward pass
  Distances
  Train
epoch 52, times: [hn 4.04 s epoch 40.41 s val 0.00 s] lr = 0.100000 loss = -0.203384 = 0.00558386 + lam * -4.17935, offending 66659
  Forward pass
  Distances
  Train
epoch 53, times: [hn 4.13 s epoch 39.31 s val 0.00 s] lr = 0.100000 loss = -0.203359 = 0.00561664 + lam * -4.17951, offending 66851
  Forward pass
  Distances
  Train
epoch 54, times: [hn 4.05 s epoch 33.92 s val 0.00 s] lr = 0.100000 loss = -0.203408 = 0.00555294 + lam * -4.17921, offending 66667
  Forward pass
  Distances
  Train
epoch 55, times: [hn 4.08 s epoch 40.66 s val 0.00 s] lr = 0.100000 loss = -0.20338 = 0.00558229 + lam * -4.17924, offending 66941
  Forward pass
  Distances
  Train
epoch 56, times: [hn 4.00 s epoch 39.79 s val 0.00 s] lr = 0.100000 loss = -0.203471 = 0.00549283 + lam * -4.17928, offending 66190
  Forward pass
  Distances
  Train
epoch 57, times: [hn 3.96 s epoch 39.91 s val 0.00 s] lr = 0.100000 loss = -0.203449 = 0.0055047 + lam * -4.17907, offending 66298
  Forward pass
  Distances
  Train
epoch 58, times: [hn 4.06 s epoch 40.20 s val 0.00 s] lr = 0.100000 loss = -0.203506 = 0.00549042 + lam * -4.17993, offending 66256
  Forward pass
  Distances
  Train
Valiation at epoch 59
zn_3     nbit= 14:  0.0012 0.0123 0.1042
zn_10    nbit= 32:  0.0556 0.2466 0.6366
zn_79    nbit= 64:  0.2193 0.6491 0.9467
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.64907 > 0.64491), keeping as best
epoch 59, times: [hn 4.05 s epoch 39.93 s val 9.52 s] lr = 0.100000 loss = -0.203527 = 0.00544964 + lam * -4.17953, offending 65693
  Forward pass
  Distances
  Train
epoch 60, times: [hn 4.11 s epoch 40.32 s val 0.00 s] lr = 0.100000 loss = -0.203511 = 0.00550607 + lam * -4.18034, offending 66167
  Forward pass
  Distances
  Train
epoch 61, times: [hn 4.05 s epoch 40.58 s val 0.00 s] lr = 0.100000 loss = -0.203547 = 0.00545602 + lam * -4.18006, offending 65970
  Forward pass
  Distances
  Train
epoch 62, times: [hn 4.05 s epoch 40.83 s val 0.00 s] lr = 0.100000 loss = -0.203545 = 0.00545683 + lam * -4.18003, offending 65996
  Forward pass
  Distances
  Train
epoch 63, times: [hn 4.03 s epoch 40.29 s val 0.00 s] lr = 0.100000 loss = -0.203571 = 0.00543243 + lam * -4.18007, offending 65420
  Forward pass
  Distances
  Train
epoch 64, times: [hn 4.05 s epoch 40.29 s val 0.00 s] lr = 0.100000 loss = -0.20358 = 0.00543251 + lam * -4.18025, offending 65465
  Forward pass
  Distances
  Train
epoch 65, times: [hn 3.98 s epoch 40.50 s val 0.00 s] lr = 0.100000 loss = -0.203669 = 0.00535062 + lam * -4.18039, offending 65162
  Forward pass
  Distances
  Train
epoch 66, times: [hn 3.99 s epoch 40.84 s val 0.00 s] lr = 0.100000 loss = -0.20368 = 0.00536175 + lam * -4.18084, offending 65315
  Forward pass
  Distances
  Train
epoch 67, times: [hn 4.08 s epoch 40.51 s val 0.00 s] lr = 0.100000 loss = -0.20364 = 0.00536571 + lam * -4.18011, offending 65286
  Forward pass
  Distances
  Train
epoch 68, times: [hn 3.99 s epoch 40.10 s val 0.00 s] lr = 0.100000 loss = -0.20371 = 0.0053224 + lam * -4.18065, offending 64906
  Forward pass
  Distances
  Train
Valiation at epoch 69
zn_3     nbit= 14:  0.0015 0.0125 0.1051
zn_10    nbit= 32:  0.0561 0.2505 0.6406
zn_79    nbit= 64:  0.2207 0.6483 0.9462
storing test_ckpt/0.05/checkpoint.pth
epoch 69, times: [hn 4.05 s epoch 40.42 s val 9.62 s] lr = 0.100000 loss = -0.203733 = 0.00530525 + lam * -4.18077, offending 64686
  Forward pass
  Distances
  Train
epoch 70, times: [hn 4.08 s epoch 39.97 s val 0.00 s] lr = 0.100000 loss = -0.203669 = 0.00535509 + lam * -4.18047, offending 65256
  Forward pass
  Distances
  Train
epoch 71, times: [hn 4.04 s epoch 40.14 s val 0.00 s] lr = 0.100000 loss = -0.203769 = 0.0052993 + lam * -4.18137, offending 64711
  Forward pass
  Distances
  Train
epoch 72, times: [hn 4.06 s epoch 39.65 s val 0.00 s] lr = 0.100000 loss = -0.20383 = 0.00524225 + lam * -4.18144, offending 64189
  Forward pass
  Distances
  Train
epoch 73, times: [hn 4.05 s epoch 39.78 s val 0.00 s] lr = 0.100000 loss = -0.203764 = 0.00530412 + lam * -4.18137, offending 64520
  Forward pass
  Distances
  Train
epoch 74, times: [hn 4.03 s epoch 39.81 s val 0.00 s] lr = 0.100000 loss = -0.203763 = 0.0052744 + lam * -4.18074, offending 64415
  Forward pass
  Distances
  Train
epoch 75, times: [hn 4.00 s epoch 40.45 s val 0.00 s] lr = 0.100000 loss = -0.203796 = 0.00525521 + lam * -4.18103, offending 64454
  Forward pass
  Distances
  Train
epoch 76, times: [hn 3.98 s epoch 41.44 s val 0.00 s] lr = 0.100000 loss = -0.203817 = 0.00524033 + lam * -4.18115, offending 64290
  Forward pass
  Distances
  Train
epoch 77, times: [hn 4.00 s epoch 41.37 s val 0.00 s] lr = 0.100000 loss = -0.203884 = 0.00520481 + lam * -4.18178, offending 63898
  Forward pass
  Distances
  Train
epoch 78, times: [hn 4.03 s epoch 40.75 s val 0.00 s] lr = 0.100000 loss = -0.203837 = 0.00523998 + lam * -4.18154, offending 64228
  Forward pass
  Distances
  Train
Valiation at epoch 79
zn_3     nbit= 14:  0.0013 0.0133 0.1048
zn_10    nbit= 32:  0.0555 0.2471 0.6386
zn_79    nbit= 64:  0.2193 0.6487 0.9458
storing test_ckpt/0.05/checkpoint.pth
epoch 79, times: [hn 4.05 s epoch 41.17 s val 9.59 s] lr = 0.100000 loss = -0.203861 = 0.00520067 + lam * -4.18123, offending 63786
  Forward pass
  Distances
  Train
epoch 80, times: [hn 4.09 s epoch 41.72 s val 0.00 s] lr = 0.050000 loss = -0.204007 = 0.00510177 + lam * -4.18217, offending 63493
  Forward pass
  Distances
  Train
epoch 81, times: [hn 4.06 s epoch 41.00 s val 0.00 s] lr = 0.050000 loss = -0.204175 = 0.00496454 + lam * -4.18279, offending 62691
  Forward pass
  Distances
  Train
epoch 82, times: [hn 3.99 s epoch 40.87 s val 0.00 s] lr = 0.050000 loss = -0.20412 = 0.00501938 + lam * -4.18279, offending 62543
  Forward pass
  Distances
  Train
epoch 83, times: [hn 4.05 s epoch 41.18 s val 0.00 s] lr = 0.050000 loss = -0.20413 = 0.00498964 + lam * -4.18239, offending 63198
  Forward pass
  Distances
  Train
epoch 84, times: [hn 4.02 s epoch 39.11 s val 0.00 s] lr = 0.050000 loss = -0.204108 = 0.00500217 + lam * -4.18221, offending 63114
  Forward pass
  Distances
  Train
epoch 85, times: [hn 4.04 s epoch 40.84 s val 0.00 s] lr = 0.050000 loss = -0.204297 = 0.00485919 + lam * -4.18313, offending 61693
  Forward pass
  Distances
  Train
epoch 86, times: [hn 3.95 s epoch 40.82 s val 0.00 s] lr = 0.050000 loss = -0.204122 = 0.00500472 + lam * -4.18254, offending 62913
  Forward pass
  Distances
  Train
epoch 87, times: [hn 4.03 s epoch 40.12 s val 0.00 s] lr = 0.050000 loss = -0.204188 = 0.00496773 + lam * -4.18311, offending 62721
  Forward pass
  Distances
  Train
epoch 88, times: [hn 4.07 s epoch 41.53 s val 0.00 s] lr = 0.050000 loss = -0.204234 = 0.0049125 + lam * -4.18293, offending 62263
  Forward pass
  Distances
  Train
Valiation at epoch 89
zn_3     nbit= 14:  0.0013 0.0122 0.1061
zn_10    nbit= 32:  0.0564 0.2494 0.6449
zn_79    nbit= 64:  0.2233 0.6580 0.9487
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.65798 > 0.64907), keeping as best
epoch 89, times: [hn 4.08 s epoch 41.42 s val 9.77 s] lr = 0.050000 loss = -0.204291 = 0.00488888 + lam * -4.18359, offending 61774
  Forward pass
  Distances
  Train
epoch 90, times: [hn 4.13 s epoch 40.94 s val 0.00 s] lr = 0.050000 loss = -0.204286 = 0.0048695 + lam * -4.18311, offending 61815
  Forward pass
  Distances
  Train
epoch 91, times: [hn 4.08 s epoch 40.86 s val 0.00 s] lr = 0.050000 loss = -0.204271 = 0.00489003 + lam * -4.18322, offending 62182
  Forward pass
  Distances
  Train
epoch 92, times: [hn 4.06 s epoch 41.09 s val 0.00 s] lr = 0.050000 loss = -0.204306 = 0.00487624 + lam * -4.18365, offending 61886
  Forward pass
  Distances
  Train
epoch 93, times: [hn 4.04 s epoch 41.40 s val 0.00 s] lr = 0.050000 loss = -0.204374 = 0.00480414 + lam * -4.18356, offending 61205
  Forward pass
  Distances
  Train
epoch 94, times: [hn 3.99 s epoch 41.30 s val 0.00 s] lr = 0.050000 loss = -0.204337 = 0.0048484 + lam * -4.18372, offending 61757
  Forward pass
  Distances
  Train
epoch 95, times: [hn 4.03 s epoch 41.74 s val 0.00 s] lr = 0.050000 loss = -0.204284 = 0.00488038 + lam * -4.18329, offending 62076
  Forward pass
  Distances
  Train
epoch 96, times: [hn 4.12 s epoch 41.79 s val 0.00 s] lr = 0.050000 loss = -0.20434 = 0.00482894 + lam * -4.18338, offending 61645
  Forward pass
  Distances
  Train
epoch 97, times: [hn 4.09 s epoch 40.98 s val 0.00 s] lr = 0.050000 loss = -0.204311 = 0.00486421 + lam * -4.18351, offending 61758
  Forward pass
  Distances
  Train
epoch 98, times: [hn 3.99 s epoch 41.07 s val 0.00 s] lr = 0.050000 loss = -0.204359 = 0.00483631 + lam * -4.1839, offending 61840
  Forward pass
  Distances
  Train
Valiation at epoch 99
zn_3     nbit= 14:  0.0011 0.0124 0.1046
zn_10    nbit= 32:  0.0569 0.2526 0.6429
zn_79    nbit= 64:  0.2233 0.6582 0.9484
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.65822 > 0.65798), keeping as best
epoch 99, times: [hn 3.98 s epoch 41.10 s val 9.83 s] lr = 0.050000 loss = -0.204463 = 0.00474913 + lam * -4.18424, offending 60633
  Forward pass
  Distances
  Train
epoch 100, times: [hn 4.15 s epoch 41.26 s val 0.00 s] lr = 0.050000 loss = -0.20434 = 0.00483775 + lam * -4.18356, offending 61655
  Forward pass
  Distances
  Train
epoch 101, times: [hn 4.04 s epoch 40.79 s val 0.00 s] lr = 0.050000 loss = -0.20434 = 0.00483049 + lam * -4.1834, offending 61587
  Forward pass
  Distances
  Train
epoch 102, times: [hn 4.02 s epoch 41.30 s val 0.00 s] lr = 0.050000 loss = -0.204432 = 0.00477266 + lam * -4.18409, offending 60774
  Forward pass
  Distances
  Train
epoch 103, times: [hn 4.04 s epoch 41.44 s val 0.00 s] lr = 0.050000 loss = -0.204401 = 0.00477912 + lam * -4.1836, offending 61153
  Forward pass
  Distances
  Train
epoch 104, times: [hn 3.97 s epoch 41.36 s val 0.00 s] lr = 0.050000 loss = -0.204384 = 0.00481158 + lam * -4.1839, offending 61429
  Forward pass
  Distances
  Train
epoch 105, times: [hn 4.04 s epoch 41.11 s val 0.00 s] lr = 0.050000 loss = -0.204478 = 0.00472109 + lam * -4.18397, offending 60157
  Forward pass
  Distances
  Train
epoch 106, times: [hn 4.07 s epoch 41.96 s val 0.00 s] lr = 0.050000 loss = -0.204439 = 0.0047544 + lam * -4.18386, offending 60631
  Forward pass
  Distances
  Train
epoch 107, times: [hn 4.05 s epoch 39.63 s val 0.00 s] lr = 0.050000 loss = -0.204397 = 0.00479433 + lam * -4.18383, offending 61313
  Forward pass
  Distances
  Train
epoch 108, times: [hn 4.03 s epoch 39.43 s val 0.00 s] lr = 0.050000 loss = -0.204444 = 0.00474744 + lam * -4.18383, offending 60916
  Forward pass
  Distances
  Train
Valiation at epoch 109
zn_3     nbit= 14:  0.0013 0.0124 0.1053
zn_10    nbit= 32:  0.0556 0.2531 0.6455
zn_79    nbit= 64:  0.2241 0.6566 0.9489
storing test_ckpt/0.05/checkpoint.pth
epoch 109, times: [hn 3.96 s epoch 40.75 s val 9.65 s] lr = 0.050000 loss = -0.204467 = 0.00473348 + lam * -4.18401, offending 60880
  Forward pass
  Distances
  Train
epoch 110, times: [hn 4.05 s epoch 41.76 s val 0.00 s] lr = 0.050000 loss = -0.204499 = 0.00471546 + lam * -4.18428, offending 60452
  Forward pass
  Distances
  Train
epoch 111, times: [hn 4.05 s epoch 41.52 s val 0.00 s] lr = 0.050000 loss = -0.204492 = 0.00470025 + lam * -4.18384, offending 60514
  Forward pass
  Distances
  Train
epoch 112, times: [hn 4.02 s epoch 40.92 s val 0.00 s] lr = 0.050000 loss = -0.204506 = 0.00472366 + lam * -4.18459, offending 60529
  Forward pass
  Distances
  Train
epoch 113, times: [hn 4.01 s epoch 40.69 s val 0.00 s] lr = 0.050000 loss = -0.204557 = 0.00465917 + lam * -4.18432, offending 60396
  Forward pass
  Distances
  Train
epoch 114, times: [hn 3.98 s epoch 40.51 s val 0.00 s] lr = 0.050000 loss = -0.204486 = 0.00472202 + lam * -4.18416, offending 60697
  Forward pass
  Distances
  Train
epoch 115, times: [hn 4.04 s epoch 41.20 s val 0.00 s] lr = 0.050000 loss = -0.204512 = 0.00470748 + lam * -4.18439, offending 60425
  Forward pass
  Distances
  Train
epoch 116, times: [hn 3.94 s epoch 41.40 s val 0.00 s] lr = 0.050000 loss = -0.204506 = 0.00471305 + lam * -4.18437, offending 60651
  Forward pass
  Distances
  Train
epoch 117, times: [hn 4.05 s epoch 41.31 s val 0.00 s] lr = 0.050000 loss = -0.20451 = 0.00472223 + lam * -4.18465, offending 60540
  Forward pass
  Distances
  Train
epoch 118, times: [hn 3.99 s epoch 41.12 s val 0.00 s] lr = 0.050000 loss = -0.204571 = 0.00465568 + lam * -4.18453, offending 59758
  Forward pass
  Distances
  Train
Valiation at epoch 119
zn_3     nbit= 14:  0.0012 0.0126 0.1056
zn_10    nbit= 32:  0.0555 0.2506 0.6430
zn_79    nbit= 64:  0.2232 0.6573 0.9487
storing test_ckpt/0.05/checkpoint.pth
epoch 119, times: [hn 4.08 s epoch 41.21 s val 9.66 s] lr = 0.050000 loss = -0.204482 = 0.00472778 + lam * -4.18419, offending 60857
  Forward pass
  Distances
  Train
epoch 120, times: [hn 4.11 s epoch 41.43 s val 0.00 s] lr = 0.010000 loss = -0.204619 = 0.00460951 + lam * -4.18457, offending 60218
  Forward pass
  Distances
  Train
epoch 121, times: [hn 4.08 s epoch 41.56 s val 0.00 s] lr = 0.010000 loss = -0.204675 = 0.00455776 + lam * -4.18466, offending 59698
  Forward pass
  Distances
  Train
epoch 122, times: [hn 4.00 s epoch 41.91 s val 0.00 s] lr = 0.010000 loss = -0.204784 = 0.00447394 + lam * -4.18515, offending 58987
  Forward pass
  Distances
  Train
epoch 123, times: [hn 4.02 s epoch 41.27 s val 0.00 s] lr = 0.010000 loss = -0.204705 = 0.00454264 + lam * -4.18496, offending 59662
  Forward pass
  Distances
  Train
epoch 124, times: [hn 3.98 s epoch 41.28 s val 0.00 s] lr = 0.010000 loss = -0.204802 = 0.00445531 + lam * -4.18514, offending 58967
  Forward pass
  Distances
  Train
epoch 125, times: [hn 4.04 s epoch 41.10 s val 0.00 s] lr = 0.010000 loss = -0.204748 = 0.00449077 + lam * -4.18478, offending 58977
  Forward pass
  Distances
  Train
epoch 126, times: [hn 4.04 s epoch 40.54 s val 0.00 s] lr = 0.010000 loss = -0.204741 = 0.00451856 + lam * -4.1852, offending 59524
  Forward pass
  Distances
  Train
epoch 127, times: [hn 4.05 s epoch 40.56 s val 0.00 s] lr = 0.010000 loss = -0.20472 = 0.00451945 + lam * -4.18478, offending 59238
  Forward pass
  Distances
  Train
epoch 128, times: [hn 4.05 s epoch 41.32 s val 0.00 s] lr = 0.010000 loss = -0.204803 = 0.00447539 + lam * -4.18556, offending 58848
  Forward pass
  Distances
  Train
Valiation at epoch 129
zn_3     nbit= 14:  0.0012 0.0120 0.1056
zn_10    nbit= 32:  0.0567 0.2557 0.6499
zn_79    nbit= 64:  0.2269 0.6620 0.9501
storing test_ckpt/0.05/checkpoint.pth
zn_79,rank=10 score improves (0.66201 > 0.65822), keeping as best
epoch 129, times: [hn 4.05 s epoch 41.22 s val 9.65 s] lr = 0.010000 loss = -0.204767 = 0.00450423 + lam * -4.18543, offending 59335
  Forward pass
  Distances
  Train
epoch 130, times: [hn 4.11 s epoch 41.63 s val 0.00 s] lr = 0.010000 loss = -0.20473 = 0.00452641 + lam * -4.18513, offending 59469
  Forward pass
  Distances
  Train
epoch 131, times: [hn 4.04 s epoch 41.39 s val 0.00 s] lr = 0.010000 loss = -0.204772 = 0.00449188 + lam * -4.18527, offending 59437
  Forward pass
  Distances
  Train
epoch 132, times: [hn 4.02 s epoch 41.63 s val 0.00 s] lr = 0.010000 loss = -0.204812 = 0.00444615 + lam * -4.18516, offending 58851
  Forward pass
  Distances
  Train
epoch 133, times: [hn 4.04 s epoch 41.55 s val 0.00 s] lr = 0.010000 loss = -0.204788 = 0.00447241 + lam * -4.18521, offending 59138
  Forward pass
  Distances
  Train
epoch 134, times: [hn 4.03 s epoch 41.72 s val 0.00 s] lr = 0.010000 loss = -0.204889 = 0.00439595 + lam * -4.18571, offending 57855
  Forward pass
  Distances
  Train
epoch 135, times: [hn 4.05 s epoch 41.06 s val 0.00 s] lr = 0.010000 loss = -0.204815 = 0.0044593 + lam * -4.18549, offending 58875
  Forward pass
  Distances
  Train
epoch 136, times: [hn 4.03 s epoch 41.04 s val 0.00 s] lr = 0.010000 loss = -0.204792 = 0.0044874 + lam * -4.18558, offending 59206
  Forward pass
  Distances
  Train
epoch 137, times: [hn 4.05 s epoch 41.25 s val 0.00 s] lr = 0.010000 loss = -0.204789 = 0.0044616 + lam * -4.18501, offending 59164
  Forward pass
  Distances
  Train
epoch 138, times: [hn 4.04 s epoch 41.31 s val 0.00 s] lr = 0.010000 loss = -0.204884 = 0.0043891 + lam * -4.18545, offending 58399
  Forward pass
  Distances
  Train
Valiation at epoch 139
zn_3     nbit= 14:  0.0014 0.0123 0.1056
zn_10    nbit= 32:  0.0558 0.2518 0.6448
zn_79    nbit= 64:  0.2245 0.6602 0.9498
storing test_ckpt/0.05/checkpoint.pth
epoch 139, times: [hn 4.04 s epoch 41.19 s val 9.69 s] lr = 0.010000 loss = -0.204844 = 0.00443422 + lam * -4.18556, offending 58633
  Forward pass
  Distances
  Train
epoch 140, times: [hn 3.99 s epoch 40.79 s val 0.00 s] lr = 0.010000 loss = -0.204883 = 0.00438002 + lam * -4.18525, offending 58486
  Forward pass
  Distances
  Train
epoch 141, times: [hn 4.06 s epoch 41.29 s val 0.00 s] lr = 0.010000 loss = -0.204767 = 0.00450066 + lam * -4.18534, offending 59073
  Forward pass
  Distances
  Train
epoch 142, times: [hn 3.99 s epoch 41.21 s val 0.00 s] lr = 0.010000 loss = -0.20485 = 0.00443076 + lam * -4.18561, offending 58509
  Forward pass
  Distances
  Train
epoch 143, times: [hn 4.07 s epoch 41.31 s val 0.00 s] lr = 0.010000 loss = -0.204893 = 0.00439633 + lam * -4.18579, offending 58321
  Forward pass
  Distances
  Train
epoch 144, times: [hn 4.05 s epoch 41.51 s val 0.00 s] lr = 0.010000 loss = -0.204787 = 0.00447807 + lam * -4.18531, offending 58905
  Forward pass
  Distances
  Train
epoch 145, times: [hn 4.05 s epoch 41.57 s val 0.00 s] lr = 0.010000 loss = -0.204882 = 0.00440797 + lam * -4.1858, offending 58392
  Forward pass
  Distances
  Train
epoch 146, times: [hn 4.04 s epoch 39.17 s val 0.00 s] lr = 0.010000 loss = -0.20482 = 0.00445774 + lam * -4.18555, offending 58997
  Forward pass
  Distances
  Train
epoch 147, times: [hn 4.05 s epoch 41.11 s val 0.00 s] lr = 0.010000 loss = -0.20493 = 0.00435847 + lam * -4.18576, offending 57908
  Forward pass
  Distances
  Train
epoch 148, times: [hn 4.03 s epoch 40.97 s val 0.00 s] lr = 0.010000 loss = -0.204932 = 0.00437047 + lam * -4.18604, offending 57825
  Forward pass
  Distances
  Train
Valiation at epoch 149
zn_3     nbit= 14:  0.0015 0.0127 0.1072
zn_10    nbit= 32:  0.0570 0.2548 0.6473
zn_79    nbit= 64:  0.2261 0.6614 0.9497
storing test_ckpt/0.05/checkpoint.pth
epoch 149, times: [hn 4.04 s epoch 41.45 s val 9.50 s] lr = 0.010000 loss = -0.204933 = 0.00436243 + lam * -4.1859, offending 57809
  Forward pass
  Distances
  Train
epoch 150, times: [hn 4.10 s epoch 41.07 s val 0.00 s] lr = 0.010000 loss = -0.204915 = 0.00438756 + lam * -4.18605, offending 58274
  Forward pass
  Distances
  Train
epoch 151, times: [hn 4.07 s epoch 41.01 s val 0.00 s] lr = 0.010000 loss = -0.204875 = 0.00441306 + lam * -4.18577, offending 58213
  Forward pass
  Distances
  Train
epoch 152, times: [hn 4.06 s epoch 41.09 s val 0.00 s] lr = 0.010000 loss = -0.204933 = 0.00436406 + lam * -4.18594, offending 57855
  Forward pass
  Distances
  Train
epoch 153, times: [hn 4.05 s epoch 41.24 s val 0.00 s] lr = 0.010000 loss = -0.204853 = 0.00442276 + lam * -4.18551, offending 58486
  Forward pass
  Distances
  Train
epoch 154, times: [hn 4.06 s epoch 41.22 s val 0.00 s] lr = 0.010000 loss = -0.204831 = 0.00443468 + lam * -4.18532, offending 58388
  Forward pass
  Distances
  Train
epoch 155, times: [hn 4.02 s epoch 41.12 s val 0.00 s] lr = 0.010000 loss = -0.204889 = 0.00439998 + lam * -4.18577, offending 58570
  Forward pass
  Distances
  Train
epoch 156, times: [hn 4.03 s epoch 41.25 s val 0.00 s] lr = 0.010000 loss = -0.204961 = 0.00434008 + lam * -4.18602, offending 57881
  Forward pass
  Distances
  Train
epoch 157, times: [hn 4.02 s epoch 40.09 s val 0.00 s] lr = 0.010000 loss = -0.2049 = 0.00438528 + lam * -4.18571, offending 58297
  Forward pass
  Distances
  Train
epoch 158, times: [hn 4.02 s epoch 39.82 s val 0.00 s] lr = 0.010000 loss = -0.204869 = 0.00442732 + lam * -4.18593, offending 58335
  Forward pass
  Distances
  Train
Valiation at epoch 159
zn_3     nbit= 14:  0.0013 0.0125 0.1077
zn_10    nbit= 32:  0.0559 0.2522 0.6441
zn_79    nbit= 64:  0.2251 0.6620 0.9512
storing test_ckpt/0.05/checkpoint.pth
epoch 159, times: [hn 4.04 s epoch 41.15 s val 9.69 s] lr = 0.010000 loss = -0.20487 = 0.0044128 + lam * -4.18566, offending 58746
++ tac test_ckpt/0.05.stdout
++ grep -m1 'keeping as best'
++ grep -o '[0-9\.]*'
++ grep -o '(.*>'
+ perf=0.66201
+ echo 0.66201
0.66201
+ [[ 0.66201 > 0.66498 ]]
+ echo 'Best value of lambda: 0.01'
Best value of lambda: 0.01
+ python eval.py --database deep1b --quantizer zn_79 --ckpt-path test_ckpt/0.01/checkpoint.pth.best
Loading net
zn_79    nbit= 64:  0.2291 0.6636 0.9541