toelt-llc / course-zhaw-dlcourse-autumn2018 Goto Github PK

Ciao Umberto,

In some the Jupyter notebook (Week 4, 5), we pass the learning_rate to compute accuracy:

for example:

print ("Accuracy:", accuracy.eval({X: train, Y: labels_, learning_rate: 0.001}, session = sessb))

but here it is not needed, right ?

Is there some reason to pass the learning_rate in such cases ?

Thanks

Ciao Umberto,

In the juyper notebook, I found that

plt.rc('font', family='arial')
plt.rc('xtick', labelsize='x-small')
plt.rc('ytick', labelsize='x-small')

plt.tight_layout()
fig = plt.figure(figsize=(8, 5))
ax = fig.add_subplot(1, 1, 1)

plt.tick_params(labelsize=16)

ax.plot(x, y, ls='solid', color = 'black')
ax.set_xlabel(xlab, fontsize = 16)
ax.set_ylabel(ylab, fontsize = 16)

will display a verbose info about the object on top of the plot like this one:

"<Figure size 432x288 with 0 Axes>"

by putting:

plt.tight_layout()

at the of the block with solve this minor feature:

plt.rc('font', family='arial')
plt.rc('xtick', labelsize='x-small')
plt.rc('ytick', labelsize='x-small')

fig = plt.figure(figsize=(8, 5))
ax = fig.add_subplot(1, 1, 1)

plt.tick_params(labelsize=16)

ax.plot(x, y, ls='solid', color = 'black')
ax.set_xlabel(xlab, fontsize = 16)
ax.set_ylabel(ylab, fontsize = 16)
plt.tight_layout()

My 2 cents
Cheers
Fabien