TF 1.0 support and small numpy casting issue
govg opened this issue · 1 comments
govg commented
Hi,
Thanks for releasing your code. I just had two quick notes
-
The function tf.mul is deprecated in favor of tf.multiply. Are you working with the earlier version of TF, by any chance? If so, I will ignore this
-
I get an error that a float object can't be interpreted as an index. It arises from line 36 in the nvlda.py file, where you initialize with np.ones((1,float)) instead of ((1,int)). I have changed this, and it seems to work after that.
Thanks and regards,
Govind
akashgit commented
Thanks Govind, yes I need to update the code for TF 1 (higher) atm it is needs the older version. Feel free to reopen if needed.