Last active
August 29, 2015 13:59
-
-
Save Atcold/10662301 to your computer and use it in GitHub Desktop.
Training
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| ==> Train 87 | |
| [================================ 515/515 ===================================>] ETA: 0ms | Step: 693ms | |
| ======> Time to learn 1 iteration = 357.17 sec | |
| ======> Time to train 1 sample = 5.41 ms | |
| ======> Train CE error: 3.32 | |
| ======> Train accuracy: 25.648% | |
| Conv layer 1 | |
| max L2 weights norm: 4.315349 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 3888, big grads : 0 | |
| Conv layer 2 | |
| max L2 weights norm: 3.357420 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 6400, big grads : 0 | |
| Conv layer 3 | |
| max L2 weights norm: 3.059696 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 2304, big grads : 0 | |
| Conv layer 4 | |
| max L2 weights norm: 5.916777 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 2304, big grads : 0 | |
| Conv layer 5 | |
| max L2 weights norm: 5.614500 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 2304, big grads : 0 | |
| Linear layer 1 | |
| max L2 weights norm: 3.783963 | |
| #small weights: 37, big weights: 0 | |
| #small grads : 409600, big grads : 0 | |
| Linear layer 2 | |
| max L2 weights norm: 2.079521 | |
| #small weights: 3, big weights: 0 | |
| #small grads : 60398, big grads : 0 | |
| Output layer | |
| max L2 weights norm: 1.500999 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 10361, big grads : 0 | |
| train_confusion.totalValid: 0.25647754854369, prevTrainAcc: 0.85417172330097 | |
| >>>>>>>>>>>>>>><<<<<<<<<<<<<<< | |
| >>> Drop in training > 50% <<< | |
| >>>>>>>>>>>>>>><<<<<<<<<<<<<<< | |
| ==> Train 88 | |
| [================================ 515/515 ===================================>] ETA: 0ms | Step: 692ms | |
| ======> Time to learn 1 iteration = 356.99 sec | |
| ======> Time to train 1 sample = 5.41 ms | |
| ======> Train CE error: 1.22 | |
| ======> Train accuracy: 85.413% | |
| Conv layer 1 | |
| max L2 weights norm: 2.452757 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 23, big grads : 0 | |
| Conv layer 2 | |
| max L2 weights norm: 2.758021 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 2726, big grads : 0 | |
| Conv layer 3 | |
| max L2 weights norm: 2.899910 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 799, big grads : 0 | |
| Conv layer 4 | |
| max L2 weights norm: 2.841064 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 146, big grads : 0 | |
| Conv layer 5 | |
| max L2 weights norm: 3.378338 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 11, big grads : 0 | |
| Linear layer 1 | |
| max L2 weights norm: 3.801837 | |
| #small weights: 32, big weights: 0 | |
| #small grads : 22902, big grads : 0 | |
| Linear layer 2 | |
| max L2 weights norm: 2.094819 | |
| #small weights: 6, big weights: 0 | |
| #small grads : 14103, big grads : 0 | |
| Output layer | |
| max L2 weights norm: 1.549240 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 711, big grads : 0 | |
| train_confusion.totalValid: 0.85412621359223, prevTrainAcc: 0.25647754854369 | |
| ==> Train 96 | |
| [================================ 515/515 ===================================>] ETA: 0ms | Step: 693ms | |
| ======> Time to learn 1 iteration = 356.78 sec | |
| ======> Time to train 1 sample = 5.41 ms | |
| ======> Train CE error: 4.00 | |
| ======> Train accuracy: 3.609% | |
| Conv layer 1 | |
| max L2 weights norm: 3.228836 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 585, big grads : 0 | |
| Conv layer 2 | |
| max L2 weights norm: 3.343111 | |
| #small weights: 2, big weights: 0 | |
| #small grads : 4029, big grads : 0 | |
| Conv layer 3 | |
| max L2 weights norm: 3.478521 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 915, big grads : 0 | |
| Conv layer 4 | |
| max L2 weights norm: 2.825122 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 382, big grads : 0 | |
| Conv layer 5 | |
| max L2 weights norm: 3.765861 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 632, big grads : 0 | |
| Linear layer 1 | |
| max L2 weights norm: 3.923441 | |
| #small weights: 36, big weights: 0 | |
| #small grads : 306043, big grads : 0 | |
| Linear layer 2 | |
| max L2 weights norm: 2.134979 | |
| #small weights: 4, big weights: 0 | |
| #small grads : 34045, big grads : 0 | |
| Output layer | |
| max L2 weights norm: 1.830988 | |
| #small weights: 3, big weights: 0 | |
| #small grads : 5272, big grads : 0 | |
| train_confusion.totalValid: 0.036089199029126, prevTrainAcc: 0.58793992718447 | |
| >>>>>>>>>>>>>>><<<<<<<<<<<<<<< | |
| >>> Drop in training > 50% <<< | |
| >>>>>>>>>>>>>>><<<<<<<<<<<<<<< | |
| ==> Train 97 | |
| [================================ 515/515 ===================================>] ETA: 0ms | Step: 693ms | |
| ======> Time to learn 1 iteration = 357.17 sec | |
| ======> Time to train 1 sample = 5.41 ms | |
| ======> Train CE error: 3.95 | |
| ======> Train accuracy: 3.924% | |
| Conv layer 1 | |
| max L2 weights norm: 1.932102 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 221, big grads : 0 | |
| Conv layer 2 | |
| max L2 weights norm: 3.344407 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 3583, big grads : 0 | |
| Conv layer 3 | |
| max L2 weights norm: 3.106975 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 1144, big grads : 0 | |
| Conv layer 4 | |
| max L2 weights norm: 2.824525 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 992, big grads : 0 | |
| Conv layer 5 | |
| max L2 weights norm: 3.766861 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 862, big grads : 0 | |
| Linear layer 1 | |
| max L2 weights norm: 3.915259 | |
| #small weights: 41, big weights: 0 | |
| #small grads : 292320, big grads : 0 | |
| Linear layer 2 | |
| max L2 weights norm: 2.135637 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 31244, big grads : 0 | |
| Output layer | |
| max L2 weights norm: 1.538128 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 3757, big grads : 0 | |
| train_confusion.totalValid: 0.039244538834951, prevTrainAcc: 0.036089199029126 | |
| ==> Train 114 | |
| [================================ 515/515 ===================================>] ETA: 0ms | Step: 693ms | |
| ======> Time to learn 1 iteration = 356.73 sec | |
| ======> Time to train 1 sample = 5.40 ms | |
| ======> Train CE error: 3.01 | |
| ======> Train accuracy: 33.255% | |
| Conv layer 1 | |
| max L2 weights norm: 2.264421 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 77, big grads : 0 | |
| Conv layer 2 | |
| max L2 weights norm: 2.846607 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 4254, big grads : 0 | |
| Conv layer 3 | |
| max L2 weights norm: 3.038891 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 636, big grads : 0 | |
| Conv layer 4 | |
| max L2 weights norm: 3.229704 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 590, big grads : 0 | |
| Conv layer 5 | |
| max L2 weights norm: 3.786046 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 842, big grads : 0 | |
| Linear layer 1 | |
| max L2 weights norm: 3.940843 | |
| #small weights: 29, big weights: 0 | |
| #small grads : 30524, big grads : 0 | |
| Linear layer 2 | |
| max L2 weights norm: 2.142980 | |
| #small weights: 4, big weights: 0 | |
| #small grads : 13841, big grads : 0 | |
| Output layer | |
| max L2 weights norm: 1.444991 | |
| #small weights: 2, big weights: 0 | |
| #small grads : 16, big grads : 0 | |
| train_confusion.totalValid: 0.33255461165049, prevTrainAcc: 0.66914441747573 | |
| >>>>>>>>>>>>>>><<<<<<<<<<<<<<< | |
| >>> Drop in training > 50% <<< | |
| >>>>>>>>>>>>>>><<<<<<<<<<<<<<< | |
| ==> Train 115 | |
| [================================ 515/515 ===================================>] ETA: 0ms | Step: 693ms | |
| ======> Time to learn 1 iteration = 357.10 sec | |
| ======> Time to train 1 sample = 5.41 ms | |
| ======> Train CE error: 1.82 | |
| ======> Train accuracy: 69.974% | |
| Conv layer 1 | |
| max L2 weights norm: 1.847098 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 128, big grads : 0 | |
| Conv layer 2 | |
| max L2 weights norm: 2.777326 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 3682, big grads : 0 | |
| Conv layer 3 | |
| max L2 weights norm: 3.028112 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 1339, big grads : 0 | |
| Conv layer 4 | |
| max L2 weights norm: 3.239054 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 1234, big grads : 0 | |
| Conv layer 5 | |
| max L2 weights norm: 3.785908 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 886, big grads : 0 | |
| Linear layer 1 | |
| max L2 weights norm: 3.940366 | |
| #small weights: 46, big weights: 0 | |
| #small grads : 45391, big grads : 0 | |
| Linear layer 2 | |
| max L2 weights norm: 2.141491 | |
| #small weights: 1, big weights: 0 | |
| #small grads : 8998, big grads : 0 | |
| Output layer | |
| max L2 weights norm: 1.412055 | |
| #small weights: 0, big weights: 0 | |
| #small grads : 143, big grads : 0 | |
| train_confusion.totalValid: 0.69974211165049, prevTrainAcc: 0.33255461165049 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment