Please click to confirm and resubmit


To generate more treasures, Gary is trying to train a generator using the GAN framework.

He has a bunch of real treasures as training data, and the initial graphs of both the generator (G) and the critic (C).

He first feeds some real treasures to C, and obtains its outputs:

logits_true = []

Then he feeds random noises to G to generate fake treasures, which he then feeds to C and obtains the outputs:

logits_gen = []

Assume the true treatures have label 0, can you help Gary compute the loss for G and C?

g_loss =
c_loss =

> Congratulations! ^10

> All treasures have been found and you have reached the end of this adventure,^10 where the lions weep and dreams are born.^10

> Please leave your contacts if you wish to take on another adventure with us!

> Name:
> Email:
> Mobile:

> Almost there! ^10

> You have found most of our treasures but some have been stolen by Gollum.^10 You efforts deserve some reward, you can leave your contacts here and we'll be in touch.^10

> However, we encourage you to slip inside the eye of your mind, you might find a better path to the end!

> Name:
> Email:
> Mobile:

Try Again!