Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Evaluate score on minibatches #86

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Evaluate score on minibatches #86

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
44 changes: 25 additions & 19 deletions breze/learn/base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -246,27 +246,28 @@ def _make_loss_functions(self, mode=None, givens=None,

return f_loss, f_d_loss

def _make_args(self, X, Z, imp_weight=None):
def _make_args(self, X, Z, imp_weight=None, n_cycles=False):
batch_size = getattr(self, 'batch_size', None)
if batch_size is None:
X, Z = cast_array_to_local_type(X), cast_array_to_local_type(Z)
times = n_cycles if n_cycles else None
if imp_weight is not None:
imp_weight = cast_array_to_local_type(imp_weight)
data = itertools.repeat([X, Z, imp_weight])
data = itertools.repeat([X, Z, imp_weight], times=times)
else:
data = itertools.repeat([X, Z])
data = itertools.repeat([X, Z], times=times)
elif batch_size < 1:
raise ValueError('need strictly positive batch size')
else:
if imp_weight is not None:
data = iter_minibatches([X, Z, imp_weight], self.batch_size,
list(self.sample_dim) + [self.sample_dim[0]])
list(self.sample_dim) + [self.sample_dim[0]], n_cycles=n_cycles)
data = ((cast_array_to_local_type(x),
cast_array_to_local_type(z),
cast_array_to_local_type(w)) for x, z, w in data)
else:
data = iter_minibatches([X, Z], self.batch_size,
self.sample_dim)
self.sample_dim, n_cycles=n_cycles)

data = ((cast_array_to_local_type(x),
cast_array_to_local_type(z)) for x, z in data)
Expand Down Expand Up @@ -373,16 +374,17 @@ def score(self, X, Z, imp_weight=None):
l : scalar
Score of the model.
"""
X = cast_array_to_local_type(X)
Z = cast_array_to_local_type(Z)
if imp_weight is not None:
imp_weight = cast_array_to_local_type(imp_weight)
if self.f_score is None:
self.f_score = self._make_score_function(
imp_weight=(imp_weight is not None))
if imp_weight is None:
return self.f_score(X, Z)
return self.f_score(X, Z, imp_weight)

score = 0
sample_count = 0
for arg in self._make_args(X, Z, imp_weight, n_cycles=1):
samples_in_batch = int(arg[0][0].shape[self.sample_dim[0]])
score += self.f_score(*arg[0]) * samples_in_batch
sample_count += samples_in_batch
return score / sample_count


class UnsupervisedModel(Model, BrezeWrapperBase):
Expand Down Expand Up @@ -492,7 +494,7 @@ def fit(self, X, W=None):
if i + 1 >= self.max_iter:
break

def _make_args(self, X, W=None):
def _make_args(self, X, W=None, n_cycles=False):
batch_size = getattr(self, 'batch_size', None)
use_imp_weight = W is not None
if self.use_imp_weight != use_imp_weight:
Expand All @@ -507,11 +509,12 @@ def _make_args(self, X, W=None):
sample_dim.append(sample_dim[0])

if batch_size is None:
data = itertools.repeat(item)
times = n_cycles if n_cycles else None
data = itertools.repeat(item, times=times)
elif batch_size < 1:
raise ValueError('need strictly positive batch size')
else:
data = iter_minibatches(item, self.batch_size, sample_dim)
data = iter_minibatches(item, self.batch_size, sample_dim, n_cycles=n_cycles)
if use_imp_weight:
data = ((cast_array_to_local_type(x), cast_array_to_local_type(w))
for x, w in data)
Expand Down Expand Up @@ -542,13 +545,16 @@ def score(self, X, W=None):
l : scalar
Score of the model.
"""
X = cast_array_to_local_type(X)
if self.f_score is None:
self.f_score = self._make_score_function()
args = [X] if W is None else [X, W]
l = self.f_score(*args)

return l
score = 0
sample_count = 0
for arg in self._make_args(X, W, n_cycles=1):
samples_in_batch = int(arg[0][0].shape[self.sample_dim[0]])
score += self.f_score(*arg[0]) * samples_in_batch
sample_count += samples_in_batch
return score / sample_count


class TransformBrezeWrapperMixin(object):
Expand Down