learn.ops.split_squeeze(1, time_steps, X) and tf.unpack()

[ghost]

Hi,
I am upgrading the example to the latest TF.
In your comments you mention that split.squeeze should be replaced by tf.unpak, however there is no tf.unpack in the latest TF. Can you please clarify what you had in mind?

Thanks,

[pusj]

Hellom @QuantScientist ,

This function seems to exist, however, some tweaks are needed.

http:https://git.hiddenunit.com/hakan/tersorflow/commit/eacabf76a0e1fb3b43026a14ad92e4d972534639#3b7cf8bc73b283ca27055142f9b4eba7c97eb063

Other than this, did you experience other issues in updating the code for TF 1.0? I plan to convert this piece of excellent code myself, starting with this in one or two weeks...

BR
Staffan

[tucakov]

tf.nn.rnn_cell was replaced by tf.contrib.rnn.

Now I am still trying to find a way to update the split_squeeze

[pusj]

@tucakov , after a few simple tweaks I got this code working (updated for TF 1.0): https://github.com/akash13singh/LSTM_TimeSeries

There are however quite a few warnings of functions that will become obsolete and this script is only running on 1 thread (and not on the gpu) but hopefully I will improve both of these items...

[tucakov]

@pusj cool! I'll have a look!!!

[pusj]

@tucakov ,

I have now done a few further updates and added these the the repository I just created...:

https://github.com/pusj/LSTM-Time-Series-Analysis-using-Tensorflow

Please feel free to add comments to the two issues I've added in the issue list