benanne / morb Goto Github PK

I can't figure out how to correct this error:

ValueError: the batch size in the image (2) at run time is different than at build time (10) for the ConvOp.
Apply node that caused the error: ConvOp{('imshp', (1, 28, 28)),('kshp', (28, 28)),('nkern', 50),('bsize', 10),('dx', 1),('dy', 1),('out_mode', 'valid'),('unroll_batch', 5),('unroll_kern', 2),('unroll_patch', False),('imshp_logical', (1, 28, 28)),('kshp_logical', (28, 28)),('kshp_logical_top_aligned', True)}(Subtensor{int64:int64:}.0, Subtensor{::, ::, ::-1, ::-1}.0)
Inputs shapes: [(2, 1, 28, 28), (50, 1, 28, 28)]
Inputs strides: [(6272, 6272, 224, 8), (6272, 6272, -224, -8)]

I made mb_size=1, and it works but i'd like to work with larger minibatches, so can you please help me out here?

Hi,
Thank you for your post,it helps me a lot.
I tried to train Guassianbinary RBM with my data,which consists of real number ranging from -100 to 100. But the value of MSE monitor keeps increasing. I konw the triaing of RBM uses free energy function to do backpropagation, so I want to ask if it is okay for value of MSE monitor to keep incresing while training.

Best,
Amos

The Monitor baseclass doesn't do anything useful - monitors should just be Python functions that potentially take a StatsCollector and a Units instance as arguments. Should it turn out to be necessary to make them 'richer', then it's better to use annotations, which is more in line with Samplers and ActivationFunctions.

Problem: I meet a import error in morb/init.py when I excute 'from morb import base, units, parameters, stats, updaters, trainers, monitors'

• I download all the code file.When I run the code , it raise the import error.
• Then I change 'import base' to 'from . import base' in the morb/init.py . It do not raise import error.But it raise ModuleNotFoundError.

ImportError information: 
 File "D:mydir\example.py", line 1, in <module>
    from morb import base, units, parameters, stats, updaters, trainers, monitors
  File "D:mydir\morb\__init__.py", line 1, in <module>
    import base
ImportError: No module named 'base'

ModuleNotFoundError information: 
 File "D:mydir\example.py", line 1, in <module>
    from morb import base, units, parameters, stats, updaters, trainers, monitors
  File "D:mydir\morb\__init__.py", line 7, in <module>
    from .import updaters
  File "D:mydir\morb\updaters.py", line 2, in <module>
    import samplers
ModuleNotFoundError: No module named 'samplers'

First off, thanks for this awesome library! I'm new to DeepLearning and Theano, but your lib makes it so much easier.

I was wondering if there's a way to save (pickle) MORB models? I tried modifying example_basic.py by adding the following lines at the end:

import cPickle
with open('model.pkl', 'wb') as f:
    cPickle.dump(rbm, f, protocol=cPickle.HIGHEST_PROTOCOL)

but I get the following trace:

...
Epoch 49
MSE = 0.0026
Took 6.69 seconds
Traceback (most recent call last):
  File "example_basic.py", line 61, in <module>
    cPickle.dump(rbm, f, protocol=cPickle.HIGHEST_PROTOCOL)
cPickle.PicklingError: Can't pickle <type 'function'>: attribute lookup __builtin__.function failed

If I remove the protocol=cPickle.HIGHEST_PROTOCOL argument I get:

Epoch 49
MSE = 0.0026
Took 6.96 seconds
Traceback (most recent call last):
  File "example_basic.py", line 61, in <module>
    cPickle.dump(rbm, f)
  File "/<removed path for privacy>/python2.7/copy_reg.py", line 70, in _reduce_ex
    raise TypeError, "can't pickle %s objects" % base.__name__
TypeError: can't pickle function objects

Is there another way to save MORB models for later use?

Thanks,
Jorge