Skip to content

Commit

Permalink
timegan typo
Browse files Browse the repository at this point in the history
  • Loading branch information
@stefan-jansen
stefan-jansen committed May 28, 2021
1 parent fda9828 commit 909af01
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion 21_gans_for_synthetic_time_series/02_TimeGAN_TF2.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -950,7 +950,7 @@
" with tf.GradientTape() as tape:\n",
" h = embedder(x)\n",
" h_hat_supervised = supervisor(h)\n",
" g_loss_s = mse(h[:, 1:, :], h_hat_supervised[:, -1:, :])\n",
" g_loss_s = mse(h[:, 1:, :], h_hat_supervised[:, :-1, :])\n",
"\n",
" var_list = supervisor.trainable_variables\n",
" gradients = tape.gradient(g_loss_s, var_list)\n",
Expand Down

1 comment on commit 909af01

@Salbinus
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hi,
I think the supervised loss is also involved when the generator is trained in the 3rd Phase.
So I'm wondering if the loss in the corresponding functions (train_generator & train_embedder) shouldn't be updated as well?

Thanks and best

Please sign in to comment.