If you use the same Theano version, it should work. Can you remove all
Theano version. Try to import it and make sure it fail. Then install the
old version.

If you want to try with a recent Theano, give me the full error message. I
could send you a diff of how to fix it. It is not sure we will support that
old pickled files in the master of Theano.

Subtensor moved of place in the code. This is probably the cause of the
error. But from memory, we have keep reference in the new version at the
old place to prevent such problems.

Fred

On Wed, Sep 21, 2016 at 10:50 AM, Thomas H <thom.hanna...@gmail.com> wrote:

> Hello,
>
> I would like to submit this puzzle to the Theano-users-group:
>
> 3 years ago we used the cPickle module to store some convolutional network
> data that had been produced using Theano (using the class called "
> LeNetConvPoolLayer").
> Here is the error we get when we now try to unpickle this data, using
> Python 2.7 (we used Canopy and Anaconda release) and up-to-date
> distributions of Theano:
>
> *AttributeError('module' object has no attribute 'Subtensor’)*
>
>
> We have tried to force the install of Theano 0.6.0 (2013 release), to no
> effect.
> After reading a thread
> <https://groups.google.com/forum/#!topic/theano-users/fOoHkYwFogc> on
> this user group describing similar troubles, we have also tried modifying
> the call to BLAS, but so far we haven't been able to circumvent the issue.
>
> Some informed advice would be greatly appreciated.
>
> Thomas
>
>
> --
>
> ---
> You received this message because you are subscribed to the Google Groups
> "theano-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to theano-users+unsubscr...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to theano-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to