验证集loss为nan

[laiheshui]

作者您好，我在复现时出现验证集和测试集loss为nan情况，请问可能是什么原因造成的？
非常感谢！
Epoch: 0001 train_loss= 2.07928 train_acc= 0.23030 val_loss= nan val_acc= 0.32117 time= 2.65600
Epoch: 0002 train_loss= 2.01573 train_acc= 0.77091 val_loss= nan val_acc= 0.32117 time= 2.52879
Epoch: 0003 train_loss= 1.91171 train_acc= 0.75977 val_loss= nan val_acc= 0.32117 time= 2.45211
Epoch: 0004 train_loss= 1.78018 train_acc= 0.76200 val_loss= nan val_acc= 0.32117 time= 2.43472
Epoch: 0005 train_loss= 1.63068 train_acc= 0.74985 val_loss= nan val_acc= 0.32117 time= 2.50082
Epoch: 0006 train_loss= 1.48861 train_acc= 0.74742 val_loss= nan val_acc= 0.32117 time= 2.56341
Epoch: 0007 train_loss= 1.38061 train_acc= 0.73000 val_loss= nan val_acc= 0.32117 time= 2.46355
Epoch: 0008 train_loss= 1.29031 train_acc= 0.67389 val_loss= nan val_acc= 0.32117 time= 2.63429
Epoch: 0009 train_loss= 1.22601 train_acc= 0.59186 val_loss= nan val_acc= 0.32117 time= 2.51430
Epoch: 0010 train_loss= 1.16189 train_acc= 0.56391 val_loss= nan val_acc= 0.32117 time= 2.33937
Epoch: 0011 train_loss= 1.09515 train_acc= 0.57545 val_loss= nan val_acc= 0.32117 time= 2.54391
Epoch: 0012 train_loss= 1.03070 train_acc= 0.60867 val_loss= nan val_acc= 0.32117 time= 2.50400
Epoch: 0013 train_loss= 0.95400 train_acc= 0.68159 val_loss= nan val_acc= 0.32117 time= 2.34988
Epoch: 0014 train_loss= 0.88769 train_acc= 0.75491 val_loss= nan val_acc= 0.32117 time= 2.64189
Epoch: 0015 train_loss= 0.82062 train_acc= 0.78043 val_loss= nan val_acc= 0.32117 time= 2.40145
Epoch: 0016 train_loss= 0.77129 train_acc= 0.78104 val_loss= nan val_acc= 0.32117 time= 2.48341
Epoch: 0017 train_loss= 0.72617 train_acc= 0.78266 val_loss= nan val_acc= 0.32117 time= 2.45733
Epoch: 0018 train_loss= 0.68917 train_acc= 0.79036 val_loss= nan val_acc= 0.32117 time= 2.84410
Epoch: 0019 train_loss= 0.66309 train_acc= 0.80778 val_loss= nan val_acc= 0.32117 time= 2.53790
Epoch: 0020 train_loss= 0.63271 train_acc= 0.83857 val_loss= nan val_acc= 0.32117 time= 2.36713
Epoch: 0021 train_loss= 0.60310 train_acc= 0.85335 val_loss= nan val_acc= 0.32117 time= 2.40726
Epoch: 0022 train_loss= 0.57907 train_acc= 0.86328 val_loss= nan val_acc= 0.32117 time= 2.35343
Epoch: 0023 train_loss= 0.55026 train_acc= 0.86773 val_loss= nan val_acc= 0.32117 time= 2.36145
Epoch: 0024 train_loss= 0.52539 train_acc= 0.87503 val_loss= nan val_acc= 0.32117 time= 2.54975
Epoch: 0025 train_loss= 0.50089 train_acc= 0.87989 val_loss= nan val_acc= 0.32117 time= 2.34190
Epoch: 0026 train_loss= 0.47744 train_acc= 0.88211 val_loss= nan val_acc= 0.32117 time= 2.42881
Epoch: 0027 train_loss= 0.45458 train_acc= 0.89285 val_loss= nan val_acc= 0.32117 time= 2.45021
Epoch: 0028 train_loss= 0.43473 train_acc= 0.89143 val_loss= nan val_acc= 0.32117 time= 2.50036
Epoch: 0029 train_loss= 0.40996 train_acc= 0.90257 val_loss= nan val_acc= 0.32117 time= 3.01274
Epoch: 0030 train_loss= 0.39648 train_acc= 0.90196 val_loss= nan val_acc= 0.32117 time= 5.04990
Epoch: 0031 train_loss= 0.37891 train_acc= 0.90379 val_loss= nan val_acc= 0.32117 time= 5.03217
Epoch: 0032 train_loss= 0.36177 train_acc= 0.91169 val_loss= nan val_acc= 0.32117 time= 2.40506
Epoch: 0033 train_loss= 0.34688 train_acc= 0.91533 val_loss= nan val_acc= 0.32117 time= 2.30696
Epoch: 0034 train_loss= 0.32800 train_acc= 0.92121 val_loss= nan val_acc= 0.32117 time= 2.43240
Epoch: 0035 train_loss= 0.31175 train_acc= 0.92506 val_loss= nan val_acc= 0.32117 time= 2.31022
Epoch: 0036 train_loss= 0.29964 train_acc= 0.92870 val_loss= nan val_acc= 0.32117 time= 2.39205
Epoch: 0037 train_loss= 0.28558 train_acc= 0.93296 val_loss= nan val_acc= 0.32117 time= 2.40226
Epoch: 0038 train_loss= 0.27217 train_acc= 0.93741 val_loss= nan val_acc= 0.32117 time= 2.46458
Epoch: 0039 train_loss= 0.25874 train_acc= 0.93944 val_loss= nan val_acc= 0.32117 time= 2.50741
Epoch: 0040 train_loss= 0.24668 train_acc= 0.94166 val_loss= nan val_acc= 0.32117 time= 2.40061
Epoch: 0041 train_loss= 0.23751 train_acc= 0.94470 val_loss= nan val_acc= 0.32117 time= 2.44478
Epoch: 0042 train_loss= 0.22154 train_acc= 0.95017 val_loss= nan val_acc= 0.32117 time= 2.37206
Epoch: 0043 train_loss= 0.21236 train_acc= 0.95402 val_loss= nan val_acc= 0.32117 time= 2.44820
Epoch: 0044 train_loss= 0.19870 train_acc= 0.95503 val_loss= nan val_acc= 0.32117 time= 2.39960
Epoch: 0045 train_loss= 0.19212 train_acc= 0.95686 val_loss= nan val_acc= 0.32117 time= 2.75499
Epoch: 0046 train_loss= 0.18108 train_acc= 0.96010 val_loss= nan val_acc= 0.32117 time= 2.42976
Epoch: 0047 train_loss= 0.17043 train_acc= 0.96273 val_loss= nan val_acc= 0.32117 time= 2.52861
Epoch: 0048 train_loss= 0.16526 train_acc= 0.96172 val_loss= nan val_acc= 0.32117 time= 2.52610
Epoch: 0049 train_loss= 0.15962 train_acc= 0.96415 val_loss= nan val_acc= 0.32117 time= 2.42181
Epoch: 0050 train_loss= 0.15042 train_acc= 0.96597 val_loss= nan val_acc= 0.32117 time= 2.64140
Epoch: 0051 train_loss= 0.14265 train_acc= 0.96617 val_loss= nan val_acc= 0.32117 time= 2.57291
Epoch: 0052 train_loss= 0.13148 train_acc= 0.97043 val_loss= nan val_acc= 0.32117 time= 2.40228
Epoch: 0053 train_loss= 0.12872 train_acc= 0.97124 val_loss= nan val_acc= 0.32117 time= 2.51607
Epoch: 0054 train_loss= 0.12205 train_acc= 0.97225 val_loss= nan val_acc= 0.32117 time= 2.55258
Epoch: 0055 train_loss= 0.11693 train_acc= 0.97428 val_loss= nan val_acc= 0.32117 time= 2.64202
Epoch: 0056 train_loss= 0.11393 train_acc= 0.97286 val_loss= nan val_acc= 0.32117 time= 2.55240
Epoch: 0057 train_loss= 0.10748 train_acc= 0.97509 val_loss= nan val_acc= 0.32117 time= 2.36710
Epoch: 0058 train_loss= 0.10186 train_acc= 0.97590 val_loss= nan val_acc= 0.32117 time= 2.47185
Epoch: 0059 train_loss= 0.09597 train_acc= 0.97812 val_loss= nan val_acc= 0.32117 time= 2.31466
Epoch: 0060 train_loss= 0.09626 train_acc= 0.97873 val_loss= nan val_acc= 0.32117 time= 2.30629
Epoch: 0061 train_loss= 0.08930 train_acc= 0.97893 val_loss= nan val_acc= 0.32117 time= 2.26337
Epoch: 0062 train_loss= 0.08495 train_acc= 0.98197 val_loss= nan val_acc= 0.32117 time= 2.36974
Epoch: 0063 train_loss= 0.08541 train_acc= 0.97995 val_loss= nan val_acc= 0.32117 time= 2.42046
Epoch: 0064 train_loss= 0.08166 train_acc= 0.98157 val_loss= nan val_acc= 0.32117 time= 2.44093
Epoch: 0065 train_loss= 0.07523 train_acc= 0.98298 val_loss= nan val_acc= 0.32117 time= 2.23188
Epoch: 0066 train_loss= 0.07568 train_acc= 0.98298 val_loss= nan val_acc= 0.32117 time= 2.36673
Epoch: 0067 train_loss= 0.07194 train_acc= 0.98157 val_loss= nan val_acc= 0.32117 time= 2.46660
Epoch: 0068 train_loss= 0.07120 train_acc= 0.98298 val_loss= nan val_acc= 0.32117 time= 2.50214
Epoch: 0069 train_loss= 0.06753 train_acc= 0.98481 val_loss= nan val_acc= 0.32117 time= 2.38757
Epoch: 0070 train_loss= 0.06485 train_acc= 0.98400 val_loss= nan val_acc= 0.32117 time= 2.42465
Epoch: 0071 train_loss= 0.06490 train_acc= 0.98440 val_loss= nan val_acc= 0.32117 time= 2.35781
Epoch: 0072 train_loss= 0.06169 train_acc= 0.98623 val_loss= nan val_acc= 0.32117 time= 2.34908
Epoch: 0073 train_loss= 0.05782 train_acc= 0.98805 val_loss= nan val_acc= 0.32117 time= 2.38237
Epoch: 0074 train_loss= 0.05849 train_acc= 0.98764 val_loss= nan val_acc= 0.32117 time= 2.55141
Epoch: 0075 train_loss= 0.05523 train_acc= 0.98805 val_loss= nan val_acc= 0.32117 time= 2.40528
Epoch: 0076 train_loss= 0.05511 train_acc= 0.98744 val_loss= nan val_acc= 0.32117 time= 2.42923
Epoch: 0077 train_loss= 0.05031 train_acc= 0.98947 val_loss= nan val_acc= 0.32117 time= 2.38595
Epoch: 0078 train_loss= 0.04973 train_acc= 0.98967 val_loss= nan val_acc= 0.32117 time= 2.31613
Epoch: 0079 train_loss= 0.04927 train_acc= 0.98785 val_loss= nan val_acc= 0.32117 time= 2.46006
Epoch: 0080 train_loss= 0.04808 train_acc= 0.98967 val_loss= nan val_acc= 0.32117 time= 2.59593
Epoch: 0081 train_loss= 0.04598 train_acc= 0.99109 val_loss= nan val_acc= 0.32117 time= 2.44770
Epoch: 0082 train_loss= 0.04472 train_acc= 0.99028 val_loss= nan val_acc= 0.32117 time= 2.60915
Epoch: 0083 train_loss= 0.04416 train_acc= 0.99129 val_loss= nan val_acc= 0.32117 time= 2.55736
Epoch: 0084 train_loss= 0.04580 train_acc= 0.98987 val_loss= nan val_acc= 0.32117 time= 2.48007
Epoch: 0085 train_loss= 0.04266 train_acc= 0.99169 val_loss= nan val_acc= 0.32117 time= 2.55349
Epoch: 0086 train_loss= 0.04031 train_acc= 0.99190 val_loss= nan val_acc= 0.32117 time= 2.54955
Epoch: 0087 train_loss= 0.03944 train_acc= 0.99190 val_loss= nan val_acc= 0.32117 time= 2.33307
Epoch: 0088 train_loss= 0.04018 train_acc= 0.99190 val_loss= nan val_acc= 0.32117 time= 2.49014
Epoch: 0089 train_loss= 0.03806 train_acc= 0.99230 val_loss= nan val_acc= 0.32117 time= 2.42449
Epoch: 0090 train_loss= 0.03667 train_acc= 0.99230 val_loss= nan val_acc= 0.32117 time= 2.67252
Epoch: 0091 train_loss= 0.03799 train_acc= 0.99291 val_loss= nan val_acc= 0.32117 time= 2.57552
Epoch: 0092 train_loss= 0.03489 train_acc= 0.99210 val_loss= nan val_acc= 0.32117 time= 2.41641
Epoch: 0093 train_loss= 0.03438 train_acc= 0.99433 val_loss= nan val_acc= 0.32117 time= 2.54833
Epoch: 0094 train_loss= 0.03502 train_acc= 0.99413 val_loss= nan val_acc= 0.32117 time= 2.47512
Epoch: 0095 train_loss= 0.03404 train_acc= 0.99291 val_loss= nan val_acc= 0.32117 time= 2.61559
Epoch: 0096 train_loss= 0.03194 train_acc= 0.99271 val_loss= nan val_acc= 0.32117 time= 2.32975
Epoch: 0097 train_loss= 0.03148 train_acc= 0.99413 val_loss= nan val_acc= 0.32117 time= 2.54627
Epoch: 0098 train_loss= 0.03027 train_acc= 0.99473 val_loss= nan val_acc= 0.32117 time= 2.60389
Epoch: 0099 train_loss= 0.03030 train_acc= 0.99433 val_loss= nan val_acc= 0.32117 time= 2.67167
Epoch: 0100 train_loss= 0.02928 train_acc= 0.99433 val_loss= nan val_acc= 0.32117 time= 2.45379
Epoch: 0101 train_loss= 0.02864 train_acc= 0.99494 val_loss= nan val_acc= 0.32117 time= 2.40216
Epoch: 0102 train_loss= 0.02695 train_acc= 0.99453 val_loss= nan val_acc= 0.32117 time= 2.38281
Epoch: 0103 train_loss= 0.02705 train_acc= 0.99534 val_loss= nan val_acc= 0.32117 time= 2.45374
Epoch: 0104 train_loss= 0.02878 train_acc= 0.99494 val_loss= nan val_acc= 0.32117 time= 2.33660
Epoch: 0105 train_loss= 0.02560 train_acc= 0.99494 val_loss= nan val_acc= 0.32117 time= 2.46241
Epoch: 0106 train_loss= 0.02520 train_acc= 0.99514 val_loss= nan val_acc= 0.32117 time= 2.53520
Epoch: 0107 train_loss= 0.02649 train_acc= 0.99433 val_loss= nan val_acc= 0.32117 time= 2.53673
Epoch: 0108 train_loss= 0.02551 train_acc= 0.99494 val_loss= nan val_acc= 0.32117 time= 2.58581
Epoch: 0109 train_loss= 0.02476 train_acc= 0.99514 val_loss= nan val_acc= 0.32117 time= 2.59459
Epoch: 0110 train_loss= 0.02362 train_acc= 0.99473 val_loss= nan val_acc= 0.32117 time= 2.40176
Epoch: 0111 train_loss= 0.02432 train_acc= 0.99514 val_loss= nan val_acc= 0.32117 time= 2.38593
Epoch: 0112 train_loss= 0.02211 train_acc= 0.99656 val_loss= nan val_acc= 0.32117 time= 2.43226
Epoch: 0113 train_loss= 0.02265 train_acc= 0.99615 val_loss= nan val_acc= 0.32117 time= 2.31928
Epoch: 0114 train_loss= 0.02149 train_acc= 0.99635 val_loss= nan val_acc= 0.32117 time= 2.65328
Epoch: 0115 train_loss= 0.02208 train_acc= 0.99554 val_loss= nan val_acc= 0.32117 time= 2.51178
Epoch: 0116 train_loss= 0.02103 train_acc= 0.99656 val_loss= nan val_acc= 0.32117 time= 2.51010
Epoch: 0117 train_loss= 0.02052 train_acc= 0.99635 val_loss= nan val_acc= 0.32117 time= 2.61634
Epoch: 0118 train_loss= 0.02000 train_acc= 0.99635 val_loss= nan val_acc= 0.32117 time= 2.51400
Epoch: 0119 train_loss= 0.01968 train_acc= 0.99595 val_loss= nan val_acc= 0.32117 time= 2.62318
Epoch: 0120 train_loss= 0.01959 train_acc= 0.99656 val_loss= nan val_acc= 0.32117 time= 2.54576
Epoch: 0121 train_loss= 0.01953 train_acc= 0.99615 val_loss= nan val_acc= 0.32117 time= 2.49061
Epoch: 0122 train_loss= 0.01817 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.48426
Epoch: 0123 train_loss= 0.01852 train_acc= 0.99696 val_loss= nan val_acc= 0.32117 time= 2.39592
Epoch: 0124 train_loss= 0.01804 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.46380
Epoch: 0125 train_loss= 0.01751 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.48109
Epoch: 0126 train_loss= 0.01786 train_acc= 0.99737 val_loss= nan val_acc= 0.32117 time= 2.46414
Epoch: 0127 train_loss= 0.01748 train_acc= 0.99696 val_loss= nan val_acc= 0.32117 time= 2.30126
Epoch: 0128 train_loss= 0.01651 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.57740
Epoch: 0129 train_loss= 0.01745 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.37804
Epoch: 0130 train_loss= 0.01800 train_acc= 0.99696 val_loss= nan val_acc= 0.32117 time= 2.39581
Epoch: 0131 train_loss= 0.01570 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.34141
Epoch: 0132 train_loss= 0.01707 train_acc= 0.99656 val_loss= nan val_acc= 0.32117 time= 2.42393
Epoch: 0133 train_loss= 0.01635 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.29725
Epoch: 0134 train_loss= 0.01489 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.49231
Epoch: 0135 train_loss= 0.01614 train_acc= 0.99696 val_loss= nan val_acc= 0.32117 time= 2.35138
Epoch: 0136 train_loss= 0.01557 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.20584
Epoch: 0137 train_loss= 0.01477 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.36589
Epoch: 0138 train_loss= 0.01502 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.36767
Epoch: 0139 train_loss= 0.01441 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.51541
Epoch: 0140 train_loss= 0.01434 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.54680
Epoch: 0141 train_loss= 0.01463 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.46273
Epoch: 0142 train_loss= 0.01367 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.46355
Epoch: 0143 train_loss= 0.01399 train_acc= 0.99696 val_loss= nan val_acc= 0.32117 time= 2.49396
Epoch: 0144 train_loss= 0.01383 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.39091
Epoch: 0145 train_loss= 0.01264 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.54578
Epoch: 0146 train_loss= 0.01309 train_acc= 0.99737 val_loss= nan val_acc= 0.32117 time= 2.34139
Epoch: 0147 train_loss= 0.01279 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.36023
Epoch: 0148 train_loss= 0.01259 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.41666
Epoch: 0149 train_loss= 0.01323 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.55719
Epoch: 0150 train_loss= 0.01221 train_acc= 0.99737 val_loss= nan val_acc= 0.32117 time= 2.31748
Epoch: 0151 train_loss= 0.01226 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.25113
Epoch: 0152 train_loss= 0.01157 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.34178
Epoch: 0153 train_loss= 0.01168 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.54862
Epoch: 0154 train_loss= 0.01303 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.38928
Epoch: 0155 train_loss= 0.01297 train_acc= 0.99696 val_loss= nan val_acc= 0.32117 time= 2.59955
Epoch: 0156 train_loss= 0.01169 train_acc= 0.99716 val_loss= nan val_acc= 0.32117 time= 2.41674
Epoch: 0157 train_loss= 0.01113 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.31280
Epoch: 0158 train_loss= 0.01156 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.43748
Epoch: 0159 train_loss= 0.01103 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.41130
Epoch: 0160 train_loss= 0.01104 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.37633
Epoch: 0161 train_loss= 0.01056 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.52587
Epoch: 0162 train_loss= 0.01047 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.55117
Epoch: 0163 train_loss= 0.01022 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.43270
Epoch: 0164 train_loss= 0.01067 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.69437
Epoch: 0165 train_loss= 0.00991 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.44006
Epoch: 0166 train_loss= 0.01034 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.41904
Epoch: 0167 train_loss= 0.01026 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.22751
Epoch: 0168 train_loss= 0.01039 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.59061
Epoch: 0169 train_loss= 0.01004 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.32812
Epoch: 0170 train_loss= 0.01007 train_acc= 0.99858 val_loss= nan val_acc= 0.32117 time= 2.65901
Epoch: 0171 train_loss= 0.01011 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.42729
Epoch: 0172 train_loss= 0.00957 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.33333
Epoch: 0173 train_loss= 0.01010 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.37304
Epoch: 0174 train_loss= 0.01038 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.43272
Epoch: 0175 train_loss= 0.00918 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.62734
Epoch: 0176 train_loss= 0.00945 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.33482
Epoch: 0177 train_loss= 0.00963 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.39062
Epoch: 0178 train_loss= 0.00976 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.28254
Epoch: 0179 train_loss= 0.00888 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.38773
Epoch: 0180 train_loss= 0.00962 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.76122
Epoch: 0181 train_loss= 0.00987 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.54829
Epoch: 0182 train_loss= 0.00910 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.46872
Epoch: 0183 train_loss= 0.01060 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.36129
Epoch: 0184 train_loss= 0.00933 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.47207
Epoch: 0185 train_loss= 0.00859 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.60163
Epoch: 0186 train_loss= 0.00892 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.26298
Epoch: 0187 train_loss= 0.00914 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.37304
Epoch: 0188 train_loss= 0.00936 train_acc= 0.99737 val_loss= nan val_acc= 0.32117 time= 2.45225
Epoch: 0189 train_loss= 0.00873 train_acc= 0.99757 val_loss= nan val_acc= 0.32117 time= 2.39466
Epoch: 0190 train_loss= 0.00936 train_acc= 0.99777 val_loss= nan val_acc= 0.32117 time= 2.53997
Epoch: 0191 train_loss= 0.00832 train_acc= 0.99878 val_loss= nan val_acc= 0.32117 time= 2.59912
Epoch: 0192 train_loss= 0.00849 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.52384
Epoch: 0193 train_loss= 0.00842 train_acc= 0.99878 val_loss= nan val_acc= 0.32117 time= 2.54080
Epoch: 0194 train_loss= 0.00830 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.56574
Epoch: 0195 train_loss= 0.00794 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.45157
Epoch: 0196 train_loss= 0.00850 train_acc= 0.99818 val_loss= nan val_acc= 0.32117 time= 2.58926
Epoch: 0197 train_loss= 0.00881 train_acc= 0.99838 val_loss= nan val_acc= 0.32117 time= 2.45261
Epoch: 0198 train_loss= 0.00817 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.46599
Epoch: 0199 train_loss= 0.00806 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.52751
Epoch: 0200 train_loss= 0.00857 train_acc= 0.99797 val_loss= nan val_acc= 0.32117 time= 2.34088
Optimization Finished!
Test set results: cost= nan accuracy= 0.31795 time= 0.81085

[wuladapang]

作者您好，我也发生了相同的问题。在浏览前面的问题时，看您提到过，通过降低学习率做尝试，但并没有效果。整体的复现也是按照readme来的，请问有可能是什么原因造成这种情况的呢？
非常感谢您的回复。

[uncle-tou]

是否曾因为版本不符而修改过TensorFlow的一些语句？我也有类似的问题，重新修改相关语句后已成功解决。

[tianranlan]

是否曾因为版本不符而修改过TensorFlow的一些语句？我也有类似的问题，重新修改相关语句后已成功解决。

你好，可以请教一下是怎么解决的吗