bestest

  "downsample module has been moved to the theano.tensor.signal.pool module.")
# Neural Network with 79306 learnable parameters

## Layer information

  #  name      size
---  --------  --------
  0  input     1x32x32
  1  conv1     16x30x30
  2  pool1     16x15x15
  3  dropout1  16x15x15
  4  conv2     32x14x14
  5  pool2     32x7x7
  6  dropout2  32x7x7
  7  conv3     64x6x6
  8  pool3     64x3x3
  9  dropout3  64x3x3
 10  hidden    100
 11  hidden1   100
 12  output    10

  epoch    train loss    valid loss    train/val    valid acc  dur
-------  ------------  ------------  -----------  -----------  ------
      1       2.03338       1.73068      1.17490      0.37431  43.05s
      2       1.69768       1.53938      1.10283      0.43572  45.45s
      3       1.52086       1.34422      1.13141      0.51701  44.20s
      4       1.39143       1.25160      1.11172      0.55864  44.05s
      5       1.29987       1.17377      1.10743      0.58020  42.21s
      6       1.22677       1.14759      1.06899      0.59721  42.33s
      7       1.17605       1.05500      1.11474      0.63242  44.28s
      8       1.12928       1.07333      1.05213      0.62945  44.95s
      9       1.09600       1.01718      1.07748      0.64221  43.75s
     10       1.06882       0.95900      1.11451      0.66485  43.79s
     11       1.05166       0.93515      1.12459      0.67731  44.40s
     12       1.02692       0.92014      1.11606      0.67741  44.66s
     13       1.00529       0.93797      1.07177      0.67385  46.09s
     14       0.99123       0.90913      1.09030      0.68404  47.73s
     15       0.98300       0.91793      1.07090      0.68275  45.60s
     16       0.96857       0.90677      1.06816      0.68552  42.60s
     17       0.95792       0.91613      1.04561      0.68631  46.96s
     18       0.95153       0.91086      1.04465      0.69136  44.73s
     19       0.94848       0.91379      1.03797      0.68908  43.47s
     20       0.93309       0.90850      1.02707      0.68799  48.88s
     21       0.92709       0.89717      1.03336      0.69541  46.40s
     22       0.91391       0.98806      0.92495      0.66713  52.57s
     23       0.91536       0.90424      1.01230      0.69294  43.14s
     24       0.90076       0.92396      0.97489      0.68621  43.56s
     25       0.90479       0.87140      1.03831      0.69976  43.31s
     26       0.89878       0.88283      1.01807      0.69680  43.26s
     27       0.88974       0.86319      1.03076      0.70560  43.71s
     28       0.89337       0.90303      0.98929      0.69116  44.81s
     29       0.88561       0.91747      0.96527      0.68968  43.57s
     30       0.88326       0.86586      1.02009      0.70629  46.11s
     31       0.87802       0.88827      0.98846      0.69907  45.60s
     32       0.87463       0.94634      0.92423      0.68265  44.25s
     33       0.87621       0.89347      0.98069      0.69729  42.96s
     34       0.87525       0.85759      1.02059      0.70975  42.92s
     35       0.86404       0.82260      1.05038      0.71756  43.68s
     36       0.86130       0.88487      0.97336      0.70184  42.70s
     37       0.86728       0.85017      1.02013      0.71460  42.88s
     38       0.85852       0.86811      0.98895      0.70599  42.82s
     39       0.85364       0.85114      1.00293      0.70995  42.81s
     40       0.85475       0.85165      1.00364      0.70837  42.97s
     41       0.85118       0.84340      1.00923      0.71252  42.80s
     42       0.84341       0.86782      0.97188      0.70461  44.92s
Early stopping.
Best valid loss was 0.822595 at epoch 35.