In [25]:

    

import numpy as np
import cPickle as pickle
import matplotlib.pyplot as plt
plt.style.use('bmh')
%matplotlib inline


    

In [31]:

    

LNHS_FILE = '../ln_hs_training_losses.pkl'
lnhs_losses = pickle.load(open(LN_HS_FILE, 'rb'))
SIMPLE_FILE = '../training_losses.pkl'
simple_losses = pickle.load(open(SIMPLE_FILE, 'rb'))
# VAL_LNHS_FILE = '../ln_hs_validation_losses.pkl'
# val_lnhs_losses = pickle.load(open(VAL_LNHS_FILE, 'rb'))
# VAL_SIMPLE_LOSS = '../validation_losses.pkl'
# val_simple_losses = pickle.load(open(VAL_SIMPLE_LOSS, 'rb'))
all_lossess = {'Training - LN HS':lnhs_losses,
               'Training - Simple':simple_losses}#,
#                'Validation - LN HS':val_lnhs_losses,
#                'Validation - Simple':val_simple_losses}
plots = ln_hs_losses.keys()


    

In [33]:

    

fig = plt.figure(figsize=(16, 24))
for i, pl in enumerate(plots):
    ax = fig.add_subplot(len(plots), 1, i + 1)
    max_mean = -np.infty
    for loss in all_lossess.keys():
        for ep in all_lossess[loss][pl].keys():
            ax.plot(np.linspace(ep - 1, ep, len(all_lossess[loss][pl][ep])), all_lossess[loss][pl][ep], 
                    label='{} {}'.format(loss, pl), alpha=0.8)
            max_mean = np.max([max_mean, np.mean(all_lossess[loss][pl][ep])])
    ax.set_ylim([0, 2 * max_mean])
    ax.set_title(pl.capitalize())
    ax.legend()
plt.show()


    

In [ ]: