Wiki: Lesson 1

Part 1 (2018)
341

Two things:

  1. create_cnn is now deprecated.
    learn = create_cnn(data, models.resnet34, metrics=error_rate) should now be
    learn = cnn_learner(data, models.resnet34, metrics=error_rate).

  2. doc(interp.plot_top_losses) throws an error.

342

My promo code is JHOAHEV

343

**My promo code is JHOAHEV, enter it and apply you can get $10 **

344

In the function open_image we call pil2tensor which has the following lines of code:

 a = np.transpose(a, (1, 0, 2))
 a = np.transpose(a, (2, 1, 0))

i replaced it with:
a = np.transpose(a, (2, 0, 1))
and not surprisingly get the same result. Any idea what is going on here? am i missing something? Is this done for optimization in converting the dimensions?

345

I think the key here is that it’s stochastic GD. So there’s the element of chance that makes results different from time to time, but I’m new here; so I can’t be 100% sure.

346

I use a Gnome extention called system monitor that sits in the top bar. Show % GPU is a preference you can turn on.