I think here must be updated in the padding section
Hi @hetekce
When using pad_sequences
function, you should directly pass the ragged tensor without converting it to a NumPy array.
By converting it to a NumPy array, you lose the ragged tensor structure, and you code may not return expected results.
Hope it helps!
Hi,
actually it raises an error without converting it to a NumPy array. Moreover, you can see that usage in Lab3, which was created by Deeplearning.AI team
I just wanted to share this issue hier for Deeplarning.AI team to update the notebook. If you look at the picture carefully, they create a ragged_sequences in the previous cell and not passing it into the second cell (see that i have uncommented). Instead they use sequences
variable.
Have a good day!
@hetekce
Thanks for bringing this up. The staff have been informed to fix the notebook.
Thanks for reporting the issue!