def on_monitor(self, model, dataset, algorithm):
if self.rescale == 'none':
global_rescale = False
patch_rescale = False
elif self.rescale == 'global':
global_rescale = True
patch_rescale = False
elif self.rescale == 'individual':
global_rescale = False
patch_rescale = True
else:
assert False
# implement saving from show_examples
examples = dataset.get_batch_topo(self.rows*self.cols)
examples = dataset.adjust_for_viewer(examples)
if global_rescale:
examples /= np.abs(examples).max()
if examples.shape[3] == 1:
is_color = False
elif examples.shape[3] == 3:
is_color = True
pv = patch_viewer.PatchViewer( (self.rows, self.cols), examples.shape[1:3], is_color=is_color)
for i in xrange(self.rows*self.cols):
pv.add_patch(examples[i,:,:,:], activation = 0.0, rescale = patch_rescale)
# pv.show()
pv.save(self.save_path % len(model.monitor.channels[model.monitor.channels.keys()[0]].val_record))
print "dataset image saved"
def show_facetube(x):
# implement saving from show_examples
examples = x
rows = int(np.sqrt(examples.shape[0]))
cols = int(examples.shape[0] / rows + 1)
examples /= np.abs(examples).max()
if examples.shape[3] == 1:
is_color = False
elif examples.shape[3] == 3:
is_color = True
else:
print examples.shape[3], "?"
#take 36 images
if rows >= 6:
rows = 6
cols = 6
idx = np.arange(0.0, examples.shape[0], examples.shape[0] / (0. + 36))
idx = [int(id) for id in idx]
else:
idx = np.arange(examples.shape[0])
pv = patch_viewer.PatchViewer((rows, cols),
examples.shape[1:3],
is_color=is_color)
for i in idx:
pv.add_patch(examples[i, :, :, :], activation=0.0, rescale=False)
pv.show()
def on_monitor(self, model, dataset, algorithm):
self.count += 1
if self.count % self.freq != 0:
return None
import matplotlib.pyplot as plt
if self.rescale == 'none':
global_rescale = False
patch_rescale = False
elif self.rescale == 'global':
global_rescale = True
patch_rescale = False
elif self.rescale == 'individual':
global_rescale = False
patch_rescale = True
else:
assert False
# implement saving from show_examples
examples = dataset.get_batch_design(self.rows*self.cols)
X = examples
input_space = dataset.transformer.input_space
if input_space:
# needs reshaping
if len(input_space.axes) != len(X.shape):
X = X.reshape([X.shape[0]]+list(input_space.shape)+[input_space.num_channels])
# dimension transposition
else:
# How can we detect axes of X?
X = input_space.convert_numpy(X,input_space.axes,default)
examples = X#.astype('float32')
if global_rescale:
examples /= np.abs(examples).max()
if examples.shape[3] == 1:
is_color = False
elif examples.shape[3] == 3:
is_color = True
pv = patch_viewer.PatchViewer( (self.rows, self.cols), examples.shape[1:3], is_color=is_color)
for i in xrange(self.rows*self.cols):
pv.add_patch(examples[i,:,:,:], activation = 0.0, rescale = patch_rescale)
pv.save(self.save_path % self.count)
print "dataset image saved"
def show_random_examples():
example_cnt = test.X.shape[0]
pv = patch_viewer.PatchViewer(
grid_shape=(1, test.clip_shape_ds[0]),
patch_shape=[test.clip_shape_ds[1], test.clip_shape_ds[2]])
while True:
i = np.random.randint(example_cnt)
print 'i =', i
print 'rain_bits =', test.y[i]
print 'pred_bits =', test.y_pred[i]
track_frames = test.X[i].reshape((3, 12, 12))
print 'track_frames[-1].sum() =', track_frames[-1].sum()
for j in range(track_frames.shape[0]):
pv.add_patch(track_frames[j].astype('float32'), activation=0)
pv.show()
def get_weights_report(model_path=None,
model=None,
rescale='individual',
border=False,
norm_sort=False,
dataset=None):
"""
Returns a PatchViewer displaying a grid of filter weights
Parameters
----------
model_path : str
Filepath of the model to make the report on.
rescale : str
A string specifying how to rescale the filter images:
- 'individual' (default) : scale each filter so that it
uses as much as possible of the dynamic range
of the display under the constraint that 0
is gray and no value gets clipped
- 'global' : scale the whole ensemble of weights
- 'none' : don't rescale
dataset : pylearn2.datasets.dataset.Dataset
Dataset object to do view conversion for displaying the weights. If
not provided one will be loaded from the model's dataset_yaml_src.
Returns
-------
WRITEME
"""
if model is None:
logger.info('making weights report')
logger.info('loading model')
model = serial.load(model_path)
logger.info('loading done')
else:
assert model_path is None
assert model is not None
if rescale == 'none':
global_rescale = False
patch_rescale = False
elif rescale == 'global':
global_rescale = True
patch_rescale = False
elif rescale == 'individual':
global_rescale = False
patch_rescale = True
else:
raise ValueError('rescale=' + rescale +
", must be 'none', 'global', or 'individual'")
if isinstance(model, dict):
#assume this was a saved matlab dictionary
del model['__version__']
del model['__header__']
del model['__globals__']
keys = [key for key in model \
if hasattr(model[key], 'ndim') and model[key].ndim == 2]
if len(keys) > 2:
key = None
while key not in keys:
logger.info('Which is the weights?')
for key in keys:
logger.info('\t{0}'.format(key))
key = input()
else:
key, = keys
weights = model[key]
norms = np.sqrt(np.square(weights).sum(axis=1))
logger.info('min norm: {0}'.format(norms.min()))
logger.info('mean norm: {0}'.format(norms.mean()))
logger.info('max norm: {0}'.format(norms.max()))
return patch_viewer.make_viewer(weights,
is_color=weights.shape[1] % 3 == 0)
weights_view = None
W = None
try:
weights_view = model.get_weights_topo()
h = weights_view.shape[0]
except NotImplementedError:
if dataset is None:
logger.info('loading dataset...')
control.push_load_data(False)
dataset = yaml_parse.load(model.dataset_yaml_src)
control.pop_load_data()
logger.info('...done')
try:
W = model.get_weights()
except AttributeError as e:
reraise_as(AttributeError("""
Encountered an AttributeError while trying to call get_weights on a model.
This probably means you need to implement get_weights for this model class,
but look at the original exception to be sure.
If this is an older model class, it may have weights stored as weightsShared,
etc.
Original exception: """+str(e)))
if W is None and weights_view is None:
raise ValueError("model doesn't support any weights interfaces")
if weights_view is None:
weights_format = model.get_weights_format()
assert hasattr(weights_format,'__iter__')
assert len(weights_format) == 2
assert weights_format[0] in ['v','h']
assert weights_format[1] in ['v','h']
assert weights_format[0] != weights_format[1]
if weights_format[0] == 'v':
W = W.T
h = W.shape[0]
if norm_sort:
norms = np.sqrt(1e-8+np.square(W).sum(axis=1))
norm_prop = norms / norms.max()
weights_view = dataset.get_weights_view(W)
assert weights_view.shape[0] == h
try:
hr, hc = model.get_weights_view_shape()
except NotImplementedError:
hr = int(np.ceil(np.sqrt(h)))
hc = hr
if 'hidShape' in dir(model):
hr, hc = model.hidShape
pv = patch_viewer.PatchViewer(grid_shape=(hr, hc),
patch_shape=weights_view.shape[1:3],
is_color = weights_view.shape[-1] == 3)
if global_rescale:
weights_view /= np.abs(weights_view).max()
if norm_sort:
logger.info('sorting weights by decreasing norm')
idx = sorted( range(h), key=lambda l : - norm_prop[l] )
else:
idx = range(h)
if border:
act = 0
else:
act = None
for i in range(0,h):
patch = weights_view[idx[i],...]
pv.add_patch(patch, rescale=patch_rescale, activation=act)
abs_weights = np.abs(weights_view)
logger.info('smallest enc weight magnitude: {0}'.format(abs_weights.min()))
logger.info('mean enc weight magnitude: {0}'.format(abs_weights.mean()))
logger.info('max enc weight magnitude: {0}'.format(abs_weights.max()))
if W is not None:
norms = np.sqrt(np.square(W).sum(axis=1))
assert norms.shape == (h,)
logger.info('min norm: {0}'.format(norms.min()))
logger.info('mean norm: {0}'.format(norms.mean()))
logger.info('max norm: {0}'.format(norms.max()))
return pv
print 'this dataset has ' + str(len(examples.shape) -
2) + ' topological dimensions'
quit(-1)
#
if examples.shape[3] == 1:
is_color = False
elif examples.shape[3] == 3:
is_color = True
else:
print 'got unknown image format with ' + str(
examples.shape[3]) + ' channels'
print 'supported formats are 1 channel greyscale or three channel RGB'
quit(-1)
#
print examples.shape[1:3]
pv = patch_viewer.PatchViewer((rows, cols),
examples.shape[1:3],
is_color=is_color)
for i in xrange(rows * cols):
pv.add_patch(examples[i, :, :, :], activation=0.0, rescale=patch_rescale)
#
if out is None:
pv.show()
else:
pv.save(out)
def show_examples(path, rows, cols, rescale='global', out=None):
"""
.. todo::
WRITEME
Parameters
----------
path : string
The pickle or YAML file to show examples of
rows : int
WRITEME
cols : int
WRITEME
rescale : {'rescale', 'global', 'individual'}
Default is 'rescale', WRITEME
out : string, optional
WRITEME
"""
if rescale == 'none':
global_rescale = False
patch_rescale = False
elif rescale == 'global':
global_rescale = True
patch_rescale = False
elif rescale == 'individual':
global_rescale = False
patch_rescale = True
if path.endswith('.pkl'):
from pylearn2.utils import serial
obj = serial.load(path)
elif path.endswith('.yaml'):
print 'Building dataset from yaml...'
obj = yaml_parse.load_path(path)
print '...done'
else:
obj = yaml_parse.load(path)
if hasattr(obj, 'get_batch_topo'):
# obj is a Dataset
dataset = obj
examples = dataset.get_batch_topo(rows * cols)
else:
# obj is a Model
model = obj
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
theano_rng = RandomStreams(42)
design_examples_var = model.random_design_matrix(batch_size=rows *
cols,
theano_rng=theano_rng)
from theano import function
print 'compiling sampling function'
f = function([], design_examples_var)
print 'sampling'
design_examples = f()
print 'loading dataset'
dataset = yaml_parse.load(model.dataset_yaml_src)
examples = dataset.get_topological_view(design_examples)
norms = np.asarray([
np.sqrt(np.sum(np.square(examples[i, :])))
for i in xrange(examples.shape[0])
])
print 'norms of examples: '
print '\tmin: ', norms.min()
print '\tmean: ', norms.mean()
print '\tmax: ', norms.max()
print 'range of elements of examples', (examples.min(), examples.max())
print 'dtype: ', examples.dtype
examples = dataset.adjust_for_viewer(examples)
if global_rescale:
examples /= np.abs(examples).max()
if len(examples.shape) != 4:
print 'sorry, view_examples.py only supports image examples for now.'
print 'this dataset has ' + str(len(examples.shape) - 2),
print 'topological dimensions'
quit(-1)
if examples.shape[3] == 1:
is_color = False
elif examples.shape[3] == 3:
is_color = True
else:
print 'got unknown image format with ' + str(examples.shape[3]),
print 'channels'
print 'supported formats are 1 channel greyscale or three channel RGB'
quit(-1)
print examples.shape[1:3]
pv = patch_viewer.PatchViewer((rows, cols),
examples.shape[1:3],
is_color=is_color)
for i in xrange(rows * cols):
pv.add_patch(examples[i, :, :, :],
activation=0.0,
rescale=patch_rescale)
if out is None:
pv.show()
else:
pv.save(out)
def main(options, positional_args):
"""
.. todo::
WRITEME
"""
assert len(positional_args) == 1
path, = positional_args
out = options.out
rescale = options.rescale
if rescale == 'none':
global_rescale = False
patch_rescale = False
elif rescale == 'global':
global_rescale = True
patch_rescale = False
elif rescale == 'individual':
global_rescale = False
patch_rescale = True
else:
assert False
if path.endswith('.pkl'):
from pylearn2.utils import serial
obj = serial.load(path)
elif path.endswith('.yaml'):
print 'Building dataset from yaml...'
obj = yaml_parse.load_path(path)
print '...done'
else:
obj = yaml_parse.load(path)
rows = options.rows
cols = options.cols
if hasattr(obj, 'get_batch_topo'):
#obj is a Dataset
dataset = obj
examples = dataset.get_batch_topo(rows * cols)
else:
#obj is a Model
model = obj
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
theano_rng = RandomStreams(42)
design_examples_var = model.random_design_matrix(batch_size=rows *
cols,
theano_rng=theano_rng)
from theano import function
print 'compiling sampling function'
f = function([], design_examples_var)
print 'sampling'
design_examples = f()
print 'loading dataset'
dataset = yaml_parse.load(model.dataset_yaml_src)
examples = dataset.get_topological_view(design_examples)
norms = N.asarray([
N.sqrt(N.sum(N.square(examples[i, :])))
for i in xrange(examples.shape[0])
])
print 'norms of examples: '
print '\tmin: ', norms.min()
print '\tmean: ', norms.mean()
print '\tmax: ', norms.max()
print 'range of elements of examples', (examples.min(), examples.max())
print 'dtype: ', examples.dtype
examples = dataset.adjust_for_viewer(examples)
if global_rescale:
examples /= N.abs(examples).max()
if len(examples.shape) != 4:
print 'sorry, view_examples.py only supports image examples for now.'
print 'this dataset has ' + str(len(examples.shape) -
2) + ' topological dimensions'
quit(-1)
is_color = False
assert examples.shape[3] == 2
print examples.shape[1:3]
pv = patch_viewer.PatchViewer((rows, cols * 2),
examples.shape[1:3],
is_color=is_color)
for i in xrange(rows * cols):
# Load patches in backwards order for easier cross-eyed viewing
# (Ian can't do the magic eye thing where you focus your eyes past the screen, must
# focus eyes in front of screen)
pv.add_patch(examples[i, :, :, 1],
activation=0.0,
rescale=patch_rescale)
pv.add_patch(examples[i, :, :, 0],
activation=0.0,
rescale=patch_rescale)
if out is None:
pv.show()
else:
pv.save(out)
if norm_sort:
norms = np.sqrt(1e-8 + np.square(W).sum(axis=1))
norm_prop = norms / norms.max()
weights_view = dataset.get_weights_view(W)
assert weights_view.shape[0] == h
try:
hr, hc = model.get_weights_view_shape()
except NotImplementedError:
hr = int(np.ceil(np.sqrt(h)))
hc = hr
if 'hidShape' in dir(model):
hr, hc = model.hidShape
pv = patch_viewer.PatchViewer(grid_shape=(hr, hc),
patch_shape=weights_view.shape[1:3],
is_color=weights_view.shape[-1] == 3)
if global_rescale:
weights_view /= np.abs(weights_view).max()
if norm_sort:
print 'sorting weights by decreasing norm'
idx = sorted(range(h), key=lambda l: -norm_prop[l])
else:
idx = range(h)
if border:
act = 0
else:
act = None
def main(options, positional_args):
assert len(positional_args) == 1
path ,= positional_args
out = options.out
rescale = options.rescale
if rescale == 'none':
global_rescale = False
patch_rescale = False
elif rescale == 'global':
global_rescale = True
patch_rescale = False
elif rescale == 'individual':
global_rescale = False
patch_rescale = True
else:
assert False
if path.endswith('.pkl'):
from pylearn2.utils import serial
obj = serial.load(path)
elif path.endswith('.yaml'):
print 'Building dataset from yaml...'
obj =yaml_parse.load_path(path)
print '...done'
else:
obj = yaml_parse.load(path)
rows = options.rows
cols = options.cols
if hasattr(obj,'get_batch_topo'):
#obj is a Dataset
dataset = obj
examples = dataset.get_batch_topo(rows*cols)
else:
#obj is a Model
model = obj
from theano.sandbox.rng_mrg import MRG_RandomStreams as RandomStreams
theano_rng = RandomStreams(42)
design_examples_var = model.random_design_matrix(batch_size = rows * cols, theano_rng = theano_rng)
from theano import function
print 'compiling sampling function'
f = function([],design_examples_var)
print 'sampling'
design_examples = f()
print 'loading dataset'
dataset = yaml_parse.load(model.dataset_yaml_src)
examples = dataset.get_topological_view(design_examples)
norms = N.asarray( [
N.sqrt(N.sum(N.square(examples[i,:])))
for i in xrange(examples.shape[0])
])
print 'norms of examples: '
print '\tmin: ',norms.min()
print '\tmean: ',norms.mean()
print '\tmax: ',norms.max()
print 'range of elements of examples',(examples.min(),examples.max())
print 'dtype: ', examples.dtype
examples = dataset.adjust_for_viewer(examples)
if global_rescale:
examples /= N.abs(examples).max()
if len(examples.shape) != 4:
print 'sorry, view_examples.py only supports image examples for now.'
print 'this dataset has '+str(len(examples.shape)-2)+' topological dimensions'
quit(-1)
#
if examples.shape[3] == 1:
is_color = False
elif examples.shape[3] == 3:
is_color = True
else:
print 'got unknown image format with '+str(examples.shape[3])+' channels'
print 'supported formats are 1 channel greyscale or three channel RGB'
quit(-1)
#
print examples.shape[1:3]
pv = patch_viewer.PatchViewer( (rows, cols), examples.shape[1:3], is_color = is_color)
for i in xrange(rows*cols):
pv.add_patch(examples[i,:,:,:], activation = 0.0, rescale = patch_rescale)
#
if out is None:
pv.show()
else:
pv.save(out)
num_pieces = 1,
kernel_shape = (4, 4),
pool_shape = (1, 1),
pool_stride=(1, 1),
irange = 0.05)
deconv = Deconv(layer_name = 'deconv',
num_channels = 1,
kernel_shape = (4, 4),
irange = 0.05)
mlp = MLP(input_space =input_space,
layers = [conv, deconv])
mlp.layers[1].transformer._filters.set_value(mlp.layers[0].transformer._filters.get_value())
x = input_space.get_theano_batch()
out = mlp.fprop(x)
f = theano.function([x], out)
data = MNIST('test')
data_specs = (input_space, 'features')
iter = data.iterator(mode = 'sequential', batch_size = 2, data_specs = data_specs)
pv = patch_viewer.PatchViewer((10, 10), (28, 28))
for item in iter:
res = f(item)
pv.add_patch(item[0,:,:,0])
pv.add_patch(res[0,:,:,0])
pv.show()
break
def get_weights_report(model_path, rescale = True):
print 'making weights report'
print 'loading model'
p = serial.load(model_path)
print 'loading done'
dataset = yaml_parse.load(p.dataset_yaml_src)
if hasattr(p,'get_weights'):
p.weights = p.get_weights()
if 'weightsShared' in dir(p):
p.weights = p.weightsShared.get_value()
if 'W' in dir(p):
if hasattr(p.W,'__array__'):
warnings.warn('model.W is an ndarray; I can figure out how to display this but that seems like a sign of a bad bug')
p.weights = p.W
from theano import shared
p.W = shared(p.W)
else:
p.weights = p.W.get_value()
if 'D' in dir(p):
p.decWeightsShared = p.D
if 'enc_weights_shared' in dir(p):
p.weights = p.enc_weights_shared.get_value()
if 'W' in dir(p) and len(p.W.get_value().shape) == 3:
W = p.W.get_value()
nh , nv, ns = W.shape
pv = patch_viewer.PatchViewer(grid_shape=(nh,ns), patch_shape= dataset.view_shape()[0:2])
for i in range(0,nh):
for k in range(0,ns):
patch = W[i,:,k]
patch = dataset.vec_to_view(patch, weights = True)
pv.add_patch( patch, rescale = rescale)
#
#
elif len(p.weights.shape) == 2:
if hasattr(p,'get_weights_format'):
p.weights_format = p.get_weights_format
assert type(p.weights_format()) == type([])
assert len(p.weights_format()) == 2
assert p.weights_format()[0] in ['v','h']
assert p.weights_format()[1] in ['v','h']
assert p.weights_format()[0] != p.weights_format()[1]
if p.weights_format()[0] == 'v':
p.weights = p.weights.transpose()
h = p.weights.shape[0]
hr = int(N.ceil(N.sqrt(h)))
hc = hr
if 'hidShape' in dir(p):
hr, hc = p.hidShape
pv = patch_viewer.PatchViewer(grid_shape=(hr,hc), patch_shape=dataset.view_shape()[0:2],
is_color = dataset.view_shape()[2] == 3)
weights_mat = p.weights
assert weights_mat.shape[0] == h
weights_view = dataset.get_weights_view(weights_mat)
assert weights_view.shape[0] == h
#print 'weights_view shape '+str(weights_view.shape)
for i in range(0,h):
patch = weights_view[i,...]
pv.add_patch( patch, rescale = rescale)
else:
e = p.weights
d = p.dec_weights_shared.value
h = e.shape[0]
if len(e.shape) == 8:
raise Exception("get_weights_report doesn't support tiled convolution yet, use the show_weights8 app")
if e.shape[4] != 1:
raise Exception('weights shape: '+str(e.shape))
shape = e.shape[1:3]
dur = e.shape[3]
show_dec = id(e) != id(d)
pv = PatchViewer.PatchViewer( grid_shape = ((1+show_dec)*h,dur), patch_shape=shape)
for i in range(0,h):
pv.addVid( e[i,:,:,:,0], rescale = rescale)
if show_dec:
pv.addVid( d[i,:,:,:,0], rescale = rescale)
print 'smallest enc weight magnitude: '+str(N.abs(p.weights).min())
print 'mean enc weight magnitude: '+str(N.abs(p.weights).mean())
print 'max enc weight magnitude: '+str(N.abs(p.weights).max())
return pv
weights_view = dataset.get_weights_view(W)
assert weights_view.shape[0] == h
try:
hr, hc = model.get_weights_view_shape()
except NotImplementedError:
hr = int(np.ceil(np.sqrt(h)))
hc = hr
if 'hidShape' in dir(model):
hr, hc = model.hidShape
if (weights_view.shape[-1] == 1 and len(channels) != 1):
channels = [0]
pv = patch_viewer.PatchViewer(grid_shape=(hr, hc),
patch_shape=weights_view.shape[1:3],
is_color=len(channels) == 3)
if global_rescale:
weights_view /= np.abs(weights_view).max()
if norm_sort:
logger.info('sorting weights by decreasing norm')
idx = sorted(range(h), key=lambda l: -norm_prop[l])
else:
idx = range(h)
if border:
act = 0
else:
act = None
import sys
from pylearn2.utils import serial
from pylearn2.gui import patch_viewer
model_path = sys.argv[1]
height = int(sys.argv[2])
width = int(sys.argv[3])
model = serial.load(model_path)
W, = model.layers[0].transformer.get_params()
W = W.get_value()
assert height * width == W.shape[0] * W.shape[3]
pv = patch_viewer.PatchViewer(grid_shape=(height, width),
patch_shape=[W.shape[1], W.shape[2]])
for i in range(W.shape[3]):
for j in range(W.shape[0]):
pv.add_patch(W[j, :, :, i])
pv.show()
h = W.shape[0]
if norm_sort:
norms = np.sqrt(1e-8 + np.square(W).sum(axis=1))
norm_prop = norms / norms.max()
weights_view = dataset.get_weights_view(W)
assert weights_view.shape[0] == h
#print 'weights_view shape '+str(weights_view.shape)
hr = int(np.ceil(np.sqrt(h)))
hc = hr
if 'hidShape' in dir(model):
hr, hc = model.hidShape
pv = patch_viewer.PatchViewer(grid_shape=(hr, hc),
patch_shape=weights_view.shape[1:3],
is_color=weights_view.shape[-1] == 3)
if global_rescale:
weights_view /= np.abs(weights_view).max()
if norm_sort:
print 'sorting weights by decreasing norm'
idx = sorted(range(h), key=lambda l: -norm_prop[l])
else:
idx = range(h)
if border:
act = 0
else:
act = None
