In [41]:

    

import pylearn2.utils
import pylearn2.config
import pylearn2.gui
import numpy as np
%matplotlib inline
import matplotlib.pyplot as plt
import os.path


    

In [36]:

    

model = pylearn2.utils.serial.load(os.path.expandvars('${DATA_DIR}/plankton/models/fewer_conv_channels_no_dropout_recent.pkl'))


    

In [37]:

    

print('## Model structure summary\n')
print(model)
params = model.get_params() 
n_params = {p.name : p.get_value().size for p in params}
total_params = sum(n_params.values())
print('\n## Number of parameters\n')
print('  ' + '\n  '.join(['{0} : {1} ({2:.1f}%)'.format(k, v, 100.*v/total_params) 
                          for k, v in sorted(n_params.items(), key=lambda x: x[0])]))
print('\nTotal : {0}'.format(total_params))


    

    




## Model structure summary

h1
	Input space: Conv2DSpace(shape=(64, 64), num_channels=1, axes=('b', 0, 1, 'c'), dtype=float32)
	Total input dimension: 4096
h2
	Input space: Conv2DSpace(shape=(34, 34), num_channels=48, axes=('b', 'c', 0, 1), dtype=float32)
	Total input dimension: 55488
h3
	Input space: Conv2DSpace(shape=(18, 18), num_channels=48, axes=('b', 'c', 0, 1), dtype=float32)
	Total input dimension: 15552
h4
	Input space: Conv2DSpace(shape=(10, 10), num_channels=48, axes=('b', 'c', 0, 1), dtype=float32)
	Total input dimension: 4800
h5
	Input space: VectorSpace(dim=512, dtype=float32)
	Total input dimension: 512
y
	Input space: VectorSpace(dim=512, dtype=float32)
	Total input dimension: 512

## Number of parameters

  h1_W : 1200 (0.0%)
  h1_b : 221952 (7.1%)
  h2_W : 20736 (0.7%)
  h2_b : 62208 (2.0%)
  h3_W : 20736 (0.7%)
  h3_b : 19200 (0.6%)
  h4_W : 2457600 (78.5%)
  h4_b : 512 (0.0%)
  h5_W : 262144 (8.4%)
  h5_b : 512 (0.0%)
  softmax_W : 61952 (2.0%)
  softmax_b : 121 (0.0%)

Total : 3128873

In [40]:

    

tr = np.array(model.monitor.channels['valid_y_y_1_nll'].time_record) / 3600.
fig = plt.figure()
ax1 = fig.add_subplot(111)
ax1.plot(model.monitor.channels['valid_y_y_1_nll'].val_record)
ax1.plot(model.monitor.channels['train_y_y_1_nll'].val_record)
ax1.set_xlabel('Epochs')
ax1.legend(['Valid', 'Train'])
ax1.set_ylabel('NLL')
ax1.grid(True)
ax2 = ax1.twiny()
ax2.set_xticks(np.arange(0,tr.shape[0],20))
ax2.set_xticklabels(['{0:.2f}'.format(t) for t in tr[::20]])
ax2.set_xlabel('Hours')


    

    
Out[40]:





<matplotlib.text.Text at 0x7f8c366a1510>






    

In [6]:

    

plt.plot(model.monitor.channels['learning_rate'].val_record)


    

    
Out[6]:





[<matplotlib.lines.Line2D at 0x7f8c37c1ffd0>]






    

In [7]:

    

h1_W_up_norms = np.array([float(v) for v in model.monitor.channels['mean_update_h1_W_kernel_norm_mean'].val_record])
h1_W_norms = np.array([float(v) for v in model.monitor.channels['valid_h1_kernel_norms_mean'].val_record])
plt.plot(h1_W_norms / h1_W_up_norms)
plt.show()
plt.plot(model.monitor.channels['valid_h1_kernel_norms_mean'].val_record)
plt.plot(model.monitor.channels['valid_h1_kernel_norms_max'].val_record)


    

    













    
Out[7]:





[<matplotlib.lines.Line2D at 0x7f8c37b3af90>]






    

In [8]:

    

h2_W_up_norms = np.array([float(v) for v in model.monitor.channels['mean_update_h2_W_kernel_norm_mean'].val_record])
h2_W_norms = np.array([float(v) for v in model.monitor.channels['valid_h2_kernel_norms_mean'].val_record])
plt.plot(h2_W_norms / h2_W_up_norms)
plt.show()
plt.plot(model.monitor.channels['valid_h2_kernel_norms_mean'].val_record)
plt.plot(model.monitor.channels['valid_h2_kernel_norms_max'].val_record)


    

    













    
Out[8]:





[<matplotlib.lines.Line2D at 0x7f8c379c5c10>]






    

In [9]:

    

h3_W_up_norms = np.array([float(v) for v in model.monitor.channels['mean_update_h3_W_kernel_norm_mean'].val_record])
h3_W_norms = np.array([float(v) for v in model.monitor.channels['valid_h3_kernel_norms_mean'].val_record])
plt.plot(h3_W_norms / h3_W_up_norms)
plt.show()
plt.plot(model.monitor.channels['valid_h3_kernel_norms_mean'].val_record)
plt.plot(model.monitor.channels['valid_h3_kernel_norms_max'].val_record)


    

    













    
Out[9]:





[<matplotlib.lines.Line2D at 0x7f8c37d0abd0>]






    

In [10]:

    

h4_W_up_norms = np.array([float(v) for v in model.monitor.channels['mean_update_h4_W_col_norm_mean'].val_record])
h4_W_norms = np.array([float(v) for v in model.monitor.channels['valid_h4_col_norms_mean'].val_record])
plt.plot(h4_W_norms / h4_W_up_norms)
plt.show()
plt.plot(model.monitor.channels['valid_h4_col_norms_mean'].val_record)
plt.plot(model.monitor.channels['valid_h4_col_norms_max'].val_record)


    

    













    
Out[10]:





[<matplotlib.lines.Line2D at 0x7f8c37849f10>]






    

In [11]:

    

h5_W_up_norms = np.array([float(v) for v in model.monitor.channels['mean_update_h5_W_col_norm_mean'].val_record])
h5_W_norms = np.array([float(v) for v in model.monitor.channels['valid_h5_col_norms_mean'].val_record])
plt.plot(h5_W_norms / h5_W_up_norms)
plt.show()
plt.plot(model.monitor.channels['valid_h5_col_norms_mean'].val_record)
plt.plot(model.monitor.channels['valid_h5_col_norms_max'].val_record)


    

    













    
Out[11]:





[<matplotlib.lines.Line2D at 0x7f8c376c3050>]






    

In [ ]: