Seemed like we could deal with extra capacity in our network, so these are the results of adding extra MLP layers.

Loading the pickle


In [1]:
import pylearn2.utils
import pylearn2.config
import theano
import neukrill_net.dense_dataset
import neukrill_net.utils
import numpy as np
%matplotlib inline
import matplotlib.pyplot as plt
import holoviews as hl
%load_ext holoviews.ipython
import sklearn.metrics


Using gpu device 0: Tesla K40c
:0: FutureWarning: IPython widgets are experimental and may change in the future.
Welcome to the HoloViews IPython extension! (http://ioam.github.io/holoviews/)
Available magics: %compositor, %opts, %params, %view, %%labels, %%opts, %%view
<matplotlib.figure.Figure at 0x7f0a88b17310>
<matplotlib.figure.Figure at 0x7f0a88b17cd0>
<matplotlib.figure.Figure at 0x7f0a88b17ad0>

In [2]:
cd ..


/afs/inf.ed.ac.uk/user/s08/s0805516/repos/neukrill-net-work

In [3]:
settings = neukrill_net.utils.Settings("settings.json")
run_settings = neukrill_net.utils.load_run_settings(
    "run_settings/alexnet_based_norm_global_8aug_pmlp1.json", settings, force=True)

In [4]:
model = pylearn2.utils.serial.load(run_settings['alt_picklepath'])

In [25]:
def plot_monitor(model,c = 'valid_y_nll'):
    channel = model.monitor.channels[c]
    plt.title(c)
    plt.grid(which="both")
    plt.plot(channel.example_record,channel.val_record)
    return None

Adding one MLP layer

The following are the final logs from the best pickle file saved.


In [21]:
%run ~/repos/pylearn2/pylearn2/scripts/print_monitor.py /disk/scratch/neuroglycerin/models/alexnet_based_norm_global_8aug_pmlp1.pkl


epochs seen:  71
time trained:  37644.5284109
learning_rate : 0.00224997522309
momentum : 0.949993252754
total_seconds_last_epoch : 522.805847168
train_h1_kernel_norms_max : 1.93652689457
train_h1_kernel_norms_mean : 1.36644768715
train_h1_kernel_norms_min : 0.326520323753
train_h1_max_x_max_u : 18.449005127
train_h1_max_x_mean_u : 3.6366121769
train_h1_max_x_min_u : 0.0116694038734
train_h1_mean_x_max_u : 5.55913305283
train_h1_mean_x_mean_u : 0.522442698479
train_h1_mean_x_min_u : 0.00111842481419
train_h1_min_x_max_u : 1.76048529148
train_h1_min_x_mean_u : 0.0201795827597
train_h1_min_x_min_u : 0.0
train_h1_range_x_max_u : 18.409778595
train_h1_range_x_mean_u : 3.61643624306
train_h1_range_x_min_u : 0.0116694038734
train_h2_kernel_norms_max : 1.93652689457
train_h2_kernel_norms_mean : 1.90194439888
train_h2_kernel_norms_min : 1.24234855175
train_h2_max_x_max_u : 30.162984848
train_h2_max_x_mean_u : 5.85832595825
train_h2_max_x_min_u : 0.0
train_h2_mean_x_max_u : 7.68249702454
train_h2_mean_x_mean_u : 0.485629856586
train_h2_mean_x_min_u : 0.0
train_h2_min_x_max_u : 0.680022537708
train_h2_min_x_mean_u : 0.000462332915049
train_h2_min_x_min_u : 0.0
train_h2_range_x_max_u : 30.1269817352
train_h2_range_x_mean_u : 5.85786294937
train_h2_range_x_min_u : 0.0
train_h3_kernel_norms_max : 1.93652689457
train_h3_kernel_norms_mean : 1.60759186745
train_h3_kernel_norms_min : 0.17659817636
train_h3_max_x_max_u : 22.612493515
train_h3_max_x_mean_u : 2.73082137108
train_h3_max_x_min_u : 0.0
train_h3_mean_x_max_u : 5.072078228
train_h3_mean_x_mean_u : 0.214336872101
train_h3_mean_x_min_u : 0.0
train_h3_min_x_max_u : 0.894239127636
train_h3_min_x_mean_u : 0.000719975563698
train_h3_min_x_min_u : 0.0
train_h3_range_x_max_u : 22.6077594757
train_h3_range_x_mean_u : 2.73010468483
train_h3_range_x_min_u : 0.0
train_h4_kernel_norms_max : 1.93652689457
train_h4_kernel_norms_mean : 1.76954543591
train_h4_kernel_norms_min : 1.29110956192
train_h4_max_x_max_u : 10.2072410583
train_h4_max_x_mean_u : 1.34010064602
train_h4_max_x_min_u : 0.0
train_h4_mean_x_max_u : 2.61320829391
train_h4_mean_x_mean_u : 0.111384436488
train_h4_mean_x_min_u : 0.0
train_h4_min_x_max_u : 0.121199689806
train_h4_min_x_mean_u : 0.000116650931886
train_h4_min_x_min_u : 0.0
train_h4_range_x_max_u : 10.205862999
train_h4_range_x_mean_u : 1.33998453617
train_h4_range_x_min_u : 0.0
train_h5_col_norms_max : 1.69483721256
train_h5_col_norms_mean : 1.54823815823
train_h5_col_norms_min : 1.48278176785
train_h5_max_x_max_u : 6.31677007675
train_h5_max_x_mean_u : 0.992186963558
train_h5_max_x_min_u : 0.0
train_h5_mean_x_max_u : 2.6732583046
train_h5_mean_x_mean_u : 0.177114605904
train_h5_mean_x_min_u : 0.0
train_h5_min_x_max_u : 0.484212368727
train_h5_min_x_mean_u : 0.00289628538303
train_h5_min_x_min_u : 0.0
train_h5_range_x_max_u : 6.24085569382
train_h5_range_x_mean_u : 0.989290356636
train_h5_range_x_min_u : 0.0
train_h5_row_norms_max : 0.801439285278
train_h5_row_norms_mean : 0.624989390373
train_h5_row_norms_min : 0.561109364033
train_h6_col_norms_max : 1.35277760029
train_h6_col_norms_mean : 1.25154328346
train_h6_col_norms_min : 1.1571508646
train_h6_max_x_max_u : 6.93619251251
train_h6_max_x_mean_u : 1.02796196938
train_h6_max_x_min_u : 0.0
train_h6_mean_x_max_u : 3.40456819534
train_h6_mean_x_mean_u : 0.271517932415
train_h6_mean_x_min_u : 0.0
train_h6_min_x_max_u : 1.15065443516
train_h6_min_x_mean_u : 0.0172959640622
train_h6_min_x_min_u : 0.0
train_h6_range_x_max_u : 6.51557159424
train_h6_range_x_mean_u : 1.01066613197
train_h6_range_x_min_u : 0.0
train_h6_row_norms_max : 1.37594437599
train_h6_row_norms_mean : 1.25143074989
train_h6_row_norms_min : 1.16261923313
train_objective : 0.941646814346
train_term_0 : 0.677818238735
train_term_1_weight_decay : 0.264658540487
train_y_col_norms_max : 1.9365644455
train_y_col_norms_mean : 1.83844602108
train_y_col_norms_min : 1.59097707272
train_y_max_max_class : 0.979269564152
train_y_mean_max_class : 0.77555090189
train_y_min_max_class : 0.323902130127
train_y_misclass : 0.158403173089
train_y_nll : 0.483982115984
train_y_row_norms_max : 1.0110514164
train_y_row_norms_mean : 0.622707664967
train_y_row_norms_min : 0.422064244747
training_seconds_this_epoch : 321.798370361
valid_h1_kernel_norms_max : 1.93650114536
valid_h1_kernel_norms_mean : 1.36646199226
valid_h1_kernel_norms_min : 0.326523929834
valid_h1_max_x_max_u : 18.6381835938
valid_h1_max_x_mean_u : 3.73691797256
valid_h1_max_x_min_u : 0.014006097801
valid_h1_mean_x_max_u : 5.46667337418
valid_h1_mean_x_mean_u : 0.520919084549
valid_h1_mean_x_min_u : 0.00153043249156
valid_h1_min_x_max_u : 1.59043550491
valid_h1_min_x_mean_u : 0.0188718922436
valid_h1_min_x_min_u : 0.0
valid_h1_range_x_max_u : 18.5847129822
valid_h1_range_x_mean_u : 3.71804594994
valid_h1_range_x_min_u : 0.014006097801
valid_h2_kernel_norms_max : 1.93650114536
valid_h2_kernel_norms_mean : 1.90196311474
valid_h2_kernel_norms_min : 1.24235785007
valid_h2_max_x_max_u : 30.2398490906
valid_h2_max_x_mean_u : 6.04424381256
valid_h2_max_x_min_u : 0.0
valid_h2_mean_x_max_u : 7.42108869553
valid_h2_mean_x_mean_u : 0.48287665844
valid_h2_mean_x_min_u : 0.0
valid_h2_min_x_max_u : 0.581425666809
valid_h2_min_x_mean_u : 0.000439754570834
valid_h2_min_x_min_u : 0.0
valid_h2_range_x_max_u : 30.2176628113
valid_h2_range_x_mean_u : 6.0438041687
valid_h2_range_x_min_u : 0.0
valid_h3_kernel_norms_max : 1.93650114536
valid_h3_kernel_norms_mean : 1.60760188103
valid_h3_kernel_norms_min : 0.176597252488
valid_h3_max_x_max_u : 22.8060626984
valid_h3_max_x_mean_u : 2.81736016273
valid_h3_max_x_min_u : 0.0
valid_h3_mean_x_max_u : 4.87892103195
valid_h3_mean_x_mean_u : 0.213528469205
valid_h3_mean_x_min_u : 0.0
valid_h3_min_x_max_u : 0.841078341007
valid_h3_min_x_mean_u : 0.000693385198247
valid_h3_min_x_min_u : 0.0
valid_h3_range_x_max_u : 22.8060626984
valid_h3_range_x_mean_u : 2.81666779518
valid_h3_range_x_min_u : 0.0
valid_h4_kernel_norms_max : 1.93650114536
valid_h4_kernel_norms_mean : 1.76953327656
valid_h4_kernel_norms_min : 1.29111564159
valid_h4_max_x_max_u : 10.3047218323
valid_h4_max_x_mean_u : 1.39025592804
valid_h4_max_x_min_u : 0.0
valid_h4_mean_x_max_u : 2.45852470398
valid_h4_mean_x_mean_u : 0.111271314323
valid_h4_mean_x_min_u : 0.0
valid_h4_min_x_max_u : 0.108793355525
valid_h4_min_x_mean_u : 9.7477422969e-05
valid_h4_min_x_min_u : 0.0
valid_h4_range_x_max_u : 10.3046808243
valid_h4_range_x_mean_u : 1.39015841484
valid_h4_range_x_min_u : 0.0
valid_h5_col_norms_max : 1.69483196735
valid_h5_col_norms_mean : 1.5482250452
valid_h5_col_norms_min : 1.48280203342
valid_h5_max_x_max_u : 6.49928188324
valid_h5_max_x_mean_u : 1.11465549469
valid_h5_max_x_min_u : 0.0
valid_h5_mean_x_max_u : 2.48149657249
valid_h5_mean_x_mean_u : 0.177030622959
valid_h5_mean_x_min_u : 0.0
valid_h5_min_x_max_u : 0.313729315996
valid_h5_min_x_mean_u : 0.00183299754281
valid_h5_min_x_min_u : 0.0
valid_h5_range_x_max_u : 6.46065473557
valid_h5_range_x_mean_u : 1.11282265186
valid_h5_range_x_min_u : 0.0
valid_h5_row_norms_max : 0.80142647028
valid_h5_row_norms_mean : 0.624993979931
valid_h5_row_norms_min : 0.561104774475
valid_h6_col_norms_max : 1.35277795792
valid_h6_col_norms_mean : 1.25154137611
valid_h6_col_norms_min : 1.15715420246
valid_h6_max_x_max_u : 7.24401283264
valid_h6_max_x_mean_u : 1.19613695145
valid_h6_max_x_min_u : 0.0
valid_h6_mean_x_max_u : 3.18322825432
valid_h6_mean_x_mean_u : 0.270841598511
valid_h6_mean_x_min_u : 0.0
valid_h6_min_x_max_u : 0.807040452957
valid_h6_min_x_mean_u : 0.00989930145442
valid_h6_min_x_min_u : 0.0
valid_h6_range_x_max_u : 6.96368646622
valid_h6_range_x_mean_u : 1.18623745441
valid_h6_range_x_min_u : 0.0
valid_h6_row_norms_max : 1.37594091892
valid_h6_row_norms_mean : 1.25143539906
valid_h6_row_norms_min : 1.16260111332
valid_objective : 1.41862380505
valid_term_0 : 1.14650928974
valid_term_1_weight_decay : 0.264655351639
valid_y_col_norms_max : 1.93656742573
valid_y_col_norms_mean : 1.83844256401
valid_y_col_norms_min : 1.59096598625
valid_y_max_max_class : 0.98601692915
valid_y_mean_max_class : 0.751415073872
valid_y_min_max_class : 0.257199317217
valid_y_misclass : 0.270543813705
valid_y_nll : 0.920760035515
valid_y_row_norms_max : 1.01106965542
valid_y_row_norms_mean : 0.622714221478
valid_y_row_norms_min : 0.422071784735

In [19]:
plot_monitor(model,c="valid_objective")



In [20]:
plot_monitor(model,c="valid_y_nll")



In [22]:
plot_monitor(model,c="train_objective")



In [23]:
plot_monitor(model,c="train_y_nll")


A bit of overfitting happening here, but is it more than normal. Looking at the network before adding that layer:


In [29]:
run_settings = neukrill_net.utils.load_run_settings(
    "run_settings/alexnet_based_norm_global_8aug.json", settings, force=True)
old = pylearn2.utils.serial.load(run_settings['pickle abspath'])

In [30]:
plot_monitor(old,c="train_y_nll")


It drops faster and ends lower than without the MLP layer. Suspect whatever is happening here, the extra MLP layer is not having much of an effect.

Two More MLP Layers

Also ran a model adding two more MLP layers, tracking the results of this model:


In [31]:
run_settings = neukrill_net.utils.load_run_settings(
    "run_settings/alexnet_based_norm_global_8aug_pmlp2.json", settings, force=True)
twomlp = pylearn2.utils.serial.load(run_settings['alt_picklepath'])

In [32]:
%run ~/repos/pylearn2/pylearn2/scripts/print_monitor.py /disk/scratch/neuroglycerin/models/alexnet_based_norm_global_8aug_pmlp2.pkl


epochs seen:  108
time trained:  58310.8673749
learning_rate : 0.00224997522309
momentum : 0.949993252754
total_seconds_last_epoch : 523.871337891
train_h1_kernel_norms_max : 1.93652689457
train_h1_kernel_norms_mean : 1.24108326435
train_h1_kernel_norms_min : 0.346896827221
train_h1_max_x_max_u : 15.597278595
train_h1_max_x_mean_u : 3.01797914505
train_h1_max_x_min_u : 0.0175006352365
train_h1_mean_x_max_u : 4.5201010704
train_h1_mean_x_mean_u : 0.535802602768
train_h1_mean_x_min_u : 0.000964893319178
train_h1_min_x_max_u : 1.3846373558
train_h1_min_x_mean_u : 0.0361537821591
train_h1_min_x_min_u : 0.0
train_h1_range_x_max_u : 15.5755462646
train_h1_range_x_mean_u : 2.98182249069
train_h1_range_x_min_u : 0.0174981094897
train_h2_kernel_norms_max : 1.93652689457
train_h2_kernel_norms_mean : 1.91939425468
train_h2_kernel_norms_min : 1.43268847466
train_h2_max_x_max_u : 24.1306877136
train_h2_max_x_mean_u : 4.23142004013
train_h2_max_x_min_u : 0.0
train_h2_mean_x_max_u : 5.55021238327
train_h2_mean_x_mean_u : 0.320856809616
train_h2_mean_x_min_u : 0.0
train_h2_min_x_max_u : 0.285148710012
train_h2_min_x_mean_u : 0.000187351324712
train_h2_min_x_min_u : 0.0
train_h2_range_x_max_u : 24.1304149628
train_h2_range_x_mean_u : 4.23123121262
train_h2_range_x_min_u : 0.0
train_h3_kernel_norms_max : 1.93652689457
train_h3_kernel_norms_mean : 1.45233857632
train_h3_kernel_norms_min : 0.149823382497
train_h3_max_x_max_u : 18.0346679688
train_h3_max_x_mean_u : 1.80714952946
train_h3_max_x_min_u : 0.0
train_h3_mean_x_max_u : 4.98396921158
train_h3_mean_x_mean_u : 0.159110710025
train_h3_mean_x_min_u : 0.0
train_h3_min_x_max_u : 1.2547826767
train_h3_min_x_mean_u : 0.00161091249902
train_h3_min_x_min_u : 0.0
train_h3_range_x_max_u : 18.0325584412
train_h3_range_x_mean_u : 1.80553853512
train_h3_range_x_min_u : 0.0
train_h4_kernel_norms_max : 1.93652689457
train_h4_kernel_norms_mean : 1.83451616764
train_h4_kernel_norms_min : 1.45084547997
train_h4_max_x_max_u : 7.89363288879
train_h4_max_x_mean_u : 0.840265989304
train_h4_max_x_min_u : 0.0
train_h4_mean_x_max_u : 2.04898262024
train_h4_mean_x_mean_u : 0.0618543028831
train_h4_mean_x_min_u : 0.0
train_h4_min_x_max_u : 0.125375598669
train_h4_min_x_mean_u : 9.5383846201e-05
train_h4_min_x_min_u : 0.0
train_h4_range_x_max_u : 7.8920545578
train_h4_range_x_mean_u : 0.840170383453
train_h4_range_x_min_u : 0.0
train_h5_col_norms_max : 1.58895850182
train_h5_col_norms_mean : 1.44398844242
train_h5_col_norms_min : 1.38025856018
train_h5_max_x_max_u : 4.00679063797
train_h5_max_x_mean_u : 0.529235064983
train_h5_max_x_min_u : 0.0
train_h5_mean_x_max_u : 1.66993188858
train_h5_mean_x_mean_u : 0.0951632037759
train_h5_mean_x_min_u : 0.0
train_h5_min_x_max_u : 0.27304109931
train_h5_min_x_mean_u : 0.00185675511602
train_h5_min_x_min_u : 0.0
train_h5_range_x_max_u : 3.95153307915
train_h5_range_x_mean_u : 0.527378082275
train_h5_range_x_min_u : 0.0
train_h5_row_norms_max : 0.8391674757
train_h5_row_norms_mean : 0.582572638988
train_h5_row_norms_min : 0.517364919186
train_h6_col_norms_max : 1.24413371086
train_h6_col_norms_mean : 1.15448820591
train_h6_col_norms_min : 1.03166258335
train_h6_max_x_max_u : 3.90124130249
train_h6_max_x_mean_u : 0.562177181244
train_h6_max_x_min_u : 0.0
train_h6_mean_x_max_u : 1.75782036781
train_h6_mean_x_mean_u : 0.154782101512
train_h6_mean_x_min_u : 0.0
train_h6_min_x_max_u : 0.484156727791
train_h6_min_x_mean_u : 0.00996714923531
train_h6_min_x_min_u : 0.0
train_h6_range_x_max_u : 3.73738050461
train_h6_range_x_mean_u : 0.552209973335
train_h6_range_x_min_u : 0.0
train_h6_row_norms_max : 1.30450153351
train_h6_row_norms_mean : 1.15432751179
train_h6_row_norms_min : 1.06603252888
train_h7_col_norms_max : 1.70241200924
train_h7_col_norms_mean : 1.56555747986
train_h7_col_norms_min : 1.43913435936
train_h7_max_x_max_u : 6.82391500473
train_h7_max_x_mean_u : 0.765771925449
train_h7_max_x_min_u : 0.0
train_h7_mean_x_max_u : 3.41733121872
train_h7_mean_x_mean_u : 0.253536880016
train_h7_mean_x_min_u : 0.0
train_h7_min_x_max_u : 1.30108499527
train_h7_min_x_mean_u : 0.0289000999182
train_h7_min_x_min_u : 0.0
train_h7_range_x_max_u : 6.23595714569
train_h7_range_x_mean_u : 0.73687183857
train_h7_range_x_min_u : 0.0
train_h7_row_norms_max : 1.6674207449
train_h7_row_norms_mean : 1.5656106472
train_h7_row_norms_min : 1.4615020752
train_objective : 0.936310529709
train_term_0 : 0.701968312263
train_term_1_weight_decay : 0.235870480537
train_y_col_norms_max : 1.93642199039
train_y_col_norms_mean : 1.84872770309
train_y_col_norms_min : 1.62471485138
train_y_max_max_class : 0.975552499294
train_y_mean_max_class : 0.780172407627
train_y_min_max_class : 0.326435029507
train_y_misclass : 0.164150074124
train_y_nll : 0.495660811663
train_y_row_norms_max : 1.09023618698
train_y_row_norms_mean : 0.620151340961
train_y_row_norms_min : 0.417193800211
training_seconds_this_epoch : 320.167999268
valid_h1_kernel_norms_max : 1.93650114536
valid_h1_kernel_norms_mean : 1.24108839035
valid_h1_kernel_norms_min : 0.346897721291
valid_h1_max_x_max_u : 15.6933898926
valid_h1_max_x_mean_u : 3.09532427788
valid_h1_max_x_min_u : 0.0199278928339
valid_h1_mean_x_max_u : 4.45210886002
valid_h1_mean_x_mean_u : 0.534925401211
valid_h1_mean_x_min_u : 0.00122600398026
valid_h1_min_x_max_u : 1.33291780949
valid_h1_min_x_mean_u : 0.0340169891715
valid_h1_min_x_min_u : 0.0
valid_h1_range_x_max_u : 15.6834096909
valid_h1_range_x_mean_u : 3.0613079071
valid_h1_range_x_min_u : 0.0199278928339
valid_h2_kernel_norms_max : 1.93650114536
valid_h2_kernel_norms_mean : 1.91937816143
valid_h2_kernel_norms_min : 1.43270325661
valid_h2_max_x_max_u : 24.5286960602
valid_h2_max_x_mean_u : 4.37511920929
valid_h2_max_x_min_u : 0.0
valid_h2_mean_x_max_u : 5.42692470551
valid_h2_mean_x_mean_u : 0.318959027529
valid_h2_mean_x_min_u : 0.0
valid_h2_min_x_max_u : 0.293118298054
valid_h2_min_x_mean_u : 0.000167394493474
valid_h2_min_x_min_u : 0.0
valid_h2_range_x_max_u : 24.5286960602
valid_h2_range_x_mean_u : 4.37495183945
valid_h2_range_x_min_u : 0.0
valid_h3_kernel_norms_max : 1.93650114536
valid_h3_kernel_norms_mean : 1.45235693455
valid_h3_kernel_norms_min : 0.149823442101
valid_h3_max_x_max_u : 18.3407859802
valid_h3_max_x_mean_u : 1.86402249336
valid_h3_max_x_min_u : 0.0
valid_h3_mean_x_max_u : 4.83848524094
valid_h3_mean_x_mean_u : 0.158511891961
valid_h3_mean_x_min_u : 0.0
valid_h3_min_x_max_u : 1.21408545971
valid_h3_min_x_mean_u : 0.00154876732267
valid_h3_min_x_min_u : 0.0
valid_h3_range_x_max_u : 18.3407859802
valid_h3_range_x_mean_u : 1.86247348785
valid_h3_range_x_min_u : 0.0
valid_h4_kernel_norms_max : 1.93650114536
valid_h4_kernel_norms_mean : 1.83452832699
valid_h4_kernel_norms_min : 1.45084917545
valid_h4_max_x_max_u : 8.13619422913
valid_h4_max_x_mean_u : 0.881958603859
valid_h4_max_x_min_u : 0.0
valid_h4_mean_x_max_u : 1.91419696808
valid_h4_mean_x_mean_u : 0.0618683323264
valid_h4_mean_x_min_u : 0.0
valid_h4_min_x_max_u : 0.118530005217
valid_h4_min_x_mean_u : 8.67163689691e-05
valid_h4_min_x_min_u : 0.0
valid_h4_range_x_max_u : 8.13341903687
valid_h4_range_x_mean_u : 0.881872117519
valid_h4_range_x_min_u : 0.0
valid_h5_col_norms_max : 1.58894538879
valid_h5_col_norms_mean : 1.44398140907
valid_h5_col_norms_min : 1.38024270535
valid_h5_max_x_max_u : 4.2085313797
valid_h5_max_x_mean_u : 0.610977232456
valid_h5_max_x_min_u : 0.0
valid_h5_mean_x_max_u : 1.53317248821
valid_h5_mean_x_mean_u : 0.0951834768057
valid_h5_mean_x_min_u : 0.0
valid_h5_min_x_max_u : 0.170548558235
valid_h5_min_x_mean_u : 0.00108809466474
valid_h5_min_x_min_u : 0.0
valid_h5_range_x_max_u : 4.18554401398
valid_h5_range_x_mean_u : 0.609889149666
valid_h5_range_x_min_u : 0.0
valid_h5_row_norms_max : 0.839181184769
valid_h5_row_norms_mean : 0.582566082478
valid_h5_row_norms_min : 0.517359256744
valid_h6_col_norms_max : 1.244114995
valid_h6_col_norms_mean : 1.15446543694
valid_h6_col_norms_min : 1.03165102005
valid_h6_max_x_max_u : 4.14193487167
valid_h6_max_x_mean_u : 0.667279720306
valid_h6_max_x_min_u : 0.0
valid_h6_mean_x_max_u : 1.63849747181
valid_h6_mean_x_mean_u : 0.154208421707
valid_h6_mean_x_min_u : 0.0
valid_h6_min_x_max_u : 0.308727771044
valid_h6_min_x_mean_u : 0.00446361349896
valid_h6_min_x_min_u : 0.0
valid_h6_range_x_max_u : 4.08244371414
valid_h6_range_x_mean_u : 0.662816107273
valid_h6_range_x_min_u : 0.0
valid_h6_row_norms_max : 1.30448269844
valid_h6_row_norms_mean : 1.1543302536
valid_h6_row_norms_min : 1.06603121758
valid_h7_col_norms_max : 1.70239281654
valid_h7_col_norms_mean : 1.56556403637
valid_h7_col_norms_min : 1.4391554594
valid_h7_max_x_max_u : 7.13083744049
valid_h7_max_x_mean_u : 0.93351817131
valid_h7_max_x_min_u : 0.0
valid_h7_mean_x_max_u : 3.14384841919
valid_h7_mean_x_mean_u : 0.252650588751
valid_h7_mean_x_min_u : 0.0
valid_h7_min_x_max_u : 0.861000776291
valid_h7_min_x_mean_u : 0.0138010326773
valid_h7_min_x_min_u : 0.0
valid_h7_range_x_max_u : 6.89441108704
valid_h7_range_x_mean_u : 0.919717073441
valid_h7_range_x_min_u : 0.0
valid_h7_row_norms_max : 1.66741335392
valid_h7_row_norms_mean : 1.56562626362
valid_h7_row_norms_min : 1.46153271198
valid_objective : 1.43085956573
valid_term_0 : 1.20176923275
valid_term_1_weight_decay : 0.235869556665
valid_y_col_norms_max : 1.93645179272
valid_y_col_norms_mean : 1.84872829914
valid_y_col_norms_min : 1.62470257282
valid_y_max_max_class : 0.985161006451
valid_y_mean_max_class : 0.757734298706
valid_y_min_max_class : 0.256332576275
valid_y_misclass : 0.279844492674
valid_y_nll : 0.966844916344
valid_y_row_norms_max : 1.09025335312
valid_y_row_norms_mean : 0.62014991045
valid_y_row_norms_min : 0.417198300362

In [33]:
plot_monitor(twomlp, c="valid_objective")



In [34]:
plot_monitor(twomlp, c="train_objective")


Seems to be very similar to the last one. Unsure what effect adding the extra MLP layer has actually had.