lunny/keras
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'add-1' of https://github.com/dbonadiman/keras into dbonadiman-add-1

Browse Source
This commit is contained in:
Francois Chollet 2015-11-08 18:07:20 -08:00
commit 1760ed6cd7
2 changed files with 109 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
examples/imdb_bidirectional_lstm.py Normal file
View File

@ -0,0 +1,64 @@
from __future__ import absolute_import
from __future__ import print_function
import numpy as np
np.random.seed(1337) # for reproducibility
from keras.preprocessing import sequence
from keras.utils.np_utils import accuracy
from keras.models import Graph
from keras.layers.core import Dense, Dropout
from keras.layers.embeddings import Embedding
from keras.layers.recurrent import LSTM
from keras.datasets import imdb
'''
Train a Bidirectional LSTM on the IMDB sentiment classification task.
The dataset is actually too small for bidirectional LSTM to be of any advantage
compared to simpler, much faster methods such as TF-IDF+LogReg.
Bidirectional LSTM may be not suited for a simple text classification task.
Notes:
- RNNs are tricky. And in particular Bidirectional RNNs you may experiment
with different Recurrent layers and different parameters to find the best configuration
for your task.
GPU command:
THEANO_FLAGS=mode=FAST_RUN,device=gpu,floatX=float32 python imdb_bidirectional_lstm.py
Output after 4 epochs on CPU: ~0.8146
'''
max_features = 20000
maxlen = 100 # cut texts after this number of words (among top max_features most common words)
batch_size = 32
print("Loading data...")
(X_train, y_train), (X_test, y_test) = imdb.load_data(nb_words=max_features, test_split=0.2)
print(len(X_train), 'train sequences')
print(len(X_test), 'test sequences')
print("Pad sequences (samples x time)")
X_train = sequence.pad_sequences(X_train, maxlen=maxlen)
X_test = sequence.pad_sequences(X_test, maxlen=maxlen)
print('X_train shape:', X_train.shape)
print('X_test shape:', X_test.shape)
y_train = np.array(y_train)
y_test = np.array(y_test)
print('Build model...')
model = Graph()
model.add_input(name='input', input_shape=(1,), dtype=int)
model.add_node(Embedding(max_features, 128, input_length=maxlen), name='embedding', input='input')
model.add_node(LSTM(64), name='forward', input='embedding' ) # You can change these two layers with GRU
model.add_node(LSTM(64, go_backwards=True), name='backward', input='embedding' )
model.add_node(Dropout(0.5), name='dropout', inputs=['forward', 'backward'])
model.add_node(Dense(1, activation='sigmoid'), name='sigmoid', input='dropout')
model.add_output(name='output', input='sigmoid')
# try using different optimizers and different optimizer configs
model.compile('adam', {'output':'binary_crossentropy'})
print("Train...")
model.fit({'input':X_train, 'output':y_train}, batch_size=batch_size, nb_epoch=4)
acc = accuracy(y_test, np.round(np.array(model.predict({'input':X_test}, batch_size=batch_size)['output'])))
print('Test accuracy:', acc)

68
keras/layers/recurrent.py
View File

@ -55,7 +55,8 @@ class SimpleRNN(Recurrent):
'''
def __init__(self, output_dim,
init='glorot_uniform', inner_init='orthogonal', activation='sigmoid', weights=None,
truncate_gradient=-1, return_sequences=False, input_dim=None, input_length=None, **kwargs):
truncate_gradient=-1, return_sequences=False, input_dim=None,
input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -63,6 +64,7 @@ class SimpleRNN(Recurrent):
self.activation = activations.get(activation)
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -107,7 +109,8 @@ class SimpleRNN(Recurrent):
# initialization of the output. Input to _step with default tap=-1.
outputs_info=T.unbroadcast(alloc_zeros_matrix(X.shape[1], self.output_dim), 1),
non_sequences=self.U, # static inputs to _step
truncate_gradient=self.truncate_gradient)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
@ -122,7 +125,8 @@ class SimpleRNN(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(SimpleRNN, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
@ -141,7 +145,7 @@ class SimpleDeepRNN(Recurrent):
init='glorot_uniform', inner_init='orthogonal',
activation='sigmoid', inner_activation='hard_sigmoid',
weights=None, truncate_gradient=-1, return_sequences=False,
input_dim=None, input_length=None, **kwargs):
input_dim=None, input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -151,6 +155,7 @@ class SimpleDeepRNN(Recurrent):
self.depth = depth
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -202,8 +207,8 @@ class SimpleDeepRNN(Recurrent):
taps=[(-i-1) for i in range(self.depth)]
)],
non_sequences=self.Us,
truncate_gradient=self.truncate_gradient
)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
@ -220,7 +225,8 @@ class SimpleDeepRNN(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(SimpleDeepRNN, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
@ -251,7 +257,7 @@ class GRU(Recurrent):
init='glorot_uniform', inner_init='orthogonal',
activation='sigmoid', inner_activation='hard_sigmoid',
weights=None, truncate_gradient=-1, return_sequences=False,
input_dim=None, input_length=None, **kwargs):
input_dim=None, input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -260,6 +266,7 @@ class GRU(Recurrent):
self.truncate_gradient = truncate_gradient
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -317,7 +324,8 @@ class GRU(Recurrent):
sequences=[x_z, x_r, x_h, padded_mask],
outputs_info=T.unbroadcast(alloc_zeros_matrix(X.shape[1], self.output_dim), 1),
non_sequences=[self.U_z, self.U_r, self.U_h],
truncate_gradient=self.truncate_gradient)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
@ -333,7 +341,8 @@ class GRU(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(GRU, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
@ -367,7 +376,7 @@ class LSTM(Recurrent):
init='glorot_uniform', inner_init='orthogonal', forget_bias_init='one',
activation='tanh', inner_activation='hard_sigmoid',
weights=None, truncate_gradient=-1, return_sequences=False,
input_dim=None, input_length=None, **kwargs):
input_dim=None, input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -377,6 +386,7 @@ class LSTM(Recurrent):
self.truncate_gradient = truncate_gradient
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -447,7 +457,8 @@ class LSTM(Recurrent):
T.unbroadcast(alloc_zeros_matrix(X.shape[1], self.output_dim), 1)
],
non_sequences=[self.U_i, self.U_f, self.U_o, self.U_c],
truncate_gradient=self.truncate_gradient)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
@ -464,7 +475,8 @@ class LSTM(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(LSTM, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
@ -493,7 +505,7 @@ class JZS1(Recurrent):
init='glorot_uniform', inner_init='orthogonal',
activation='tanh', inner_activation='sigmoid',
weights=None, truncate_gradient=-1, return_sequences=False,
input_dim=None, input_length=None, **kwargs):
input_dim=None, input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -502,6 +514,7 @@ class JZS1(Recurrent):
self.truncate_gradient = truncate_gradient
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -566,7 +579,8 @@ class JZS1(Recurrent):
sequences=[x_z, x_r, x_h, padded_mask],
outputs_info=T.unbroadcast(alloc_zeros_matrix(X.shape[1], self.output_dim), 1),
non_sequences=[self.U_r, self.U_h],
truncate_gradient=self.truncate_gradient)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
return outputs[-1]
@ -581,7 +595,8 @@ class JZS1(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(JZS1, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
@ -610,7 +625,7 @@ class JZS2(Recurrent):
init='glorot_uniform', inner_init='orthogonal',
activation='tanh', inner_activation='sigmoid',
weights=None, truncate_gradient=-1, return_sequences=False,
input_dim=None, input_length=None, **kwargs):
input_dim=None, input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -619,6 +634,7 @@ class JZS2(Recurrent):
self.truncate_gradient = truncate_gradient
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -684,7 +700,9 @@ class JZS2(Recurrent):
sequences=[x_z, x_r, x_h, padded_mask],
outputs_info=T.unbroadcast(alloc_zeros_matrix(X.shape[1], self.output_dim), 1),
non_sequences=[self.U_z, self.U_r, self.U_h],
truncate_gradient=self.truncate_gradient)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
return outputs[-1]
@ -699,7 +717,8 @@ class JZS2(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(JZS2, self).get_config()
return dict(list(base_config.items()) + list(config.items()))
@ -728,7 +747,7 @@ class JZS3(Recurrent):
init='glorot_uniform', inner_init='orthogonal',
activation='tanh', inner_activation='sigmoid',
weights=None, truncate_gradient=-1, return_sequences=False,
input_dim=None, input_length=None, **kwargs):
input_dim=None, input_length=None, go_backwards=False, **kwargs):
self.output_dim = output_dim
self.init = initializations.get(init)
self.inner_init = initializations.get(inner_init)
@ -737,6 +756,7 @@ class JZS3(Recurrent):
self.truncate_gradient = truncate_gradient
self.return_sequences = return_sequences
self.initial_weights = weights
self.go_backwards = go_backwards
self.input_dim = input_dim
self.input_length = input_length
@ -794,8 +814,9 @@ class JZS3(Recurrent):
sequences=[x_z, x_r, x_h, padded_mask],
outputs_info=T.unbroadcast(alloc_zeros_matrix(X.shape[1], self.output_dim), 1),
non_sequences=[self.U_z, self.U_r, self.U_h],
truncate_gradient=self.truncate_gradient
)
truncate_gradient=self.truncate_gradient,
go_backwards=self.go_backwards)
if self.return_sequences:
return outputs.dimshuffle((1, 0, 2))
return outputs[-1]
@ -810,6 +831,7 @@ class JZS3(Recurrent):
"truncate_gradient": self.truncate_gradient,
"return_sequences": self.return_sequences,
"input_dim": self.input_dim,
"input_length": self.input_length}
"input_length": self.input_length,
"go_backwards": self.go_backwards}
base_config = super(JZS3, self).get_config()
return dict(list(base_config.items()) + list(config.items()))