Here are the examples of the python api numpy.random.randn.astype taken from open source projects. By voting up you can indicate which examples are most useful and appropriate.
119 Examples
3
Example 1
Project: treeano Source File: partition_axis_test.py
def test_partition_axis_node():
# just testing that it runs
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(4, 8, 12, 16, 20)),
partition_axis.PartitionAxisNode("pa",
split_idx=2,
num_splits=4,
channel_axis=3)]
).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(4, 8, 12, 16, 20).astype(fX)
ans = x[:, :, :, 8:12, :]
res = fn(x)[0]
nt.assert_equal(ans.shape, network["pa"].get_vw("default").shape)
np.testing.assert_equal(res, ans)
3
Example 2
def test_basic(self):
"""Just check that the Variational layer can compile and run"""
nb_samples, input_dim, output_dim = 3, 10, 5
layer = VariationalDense(input_dim=input_dim, output_dim=output_dim,
batch_size=nb_samples)
X = layer.get_input()
Y1 = layer.get_output(train=True)
Y2 = layer.get_output(train=False)
F = theano.function([X], [Y1, Y2])
y1, y2 = F(np.random.randn(nb_samples, input_dim).astype(floatX))
assert y1.shape == (nb_samples, output_dim)
assert y2.shape == (nb_samples, output_dim)
3
Example 3
Project: treeano Source File: simple_test.py
def test_apply_node():
network = tn.SequentialNode("s", [
tn.InputNode("in", shape=(3, 4, 5)),
tn.ApplyNode("a", fn=T.sum, shape_fn=lambda x: ()),
]).network()
fn = network.function(["in"], ["s"])
x = np.random.randn(3, 4, 5).astype(fX)
np.testing.assert_allclose(fn(x)[0],
x.sum(),
rtol=1e-5)
3
Example 4
Project: treeano Source File: lrn_test.py
def _test_localresponse_normalization_fn(fn, shape=(3, 4, 5, 6), **kwargs):
vw = treeano.VariableWrapper("foo", variable=T.tensor4(), shape=shape)
new_kwargs = dict(
# use a big value of alpha so mistakes involving alpha show up strong
alpha=1.5,
k=2,
beta=0.75,
n=5,
)
new_kwargs.update(kwargs)
fn = theano.function([vw.variable], [fn(vw, **new_kwargs)])
x = np.random.randn(*shape).astype(fX)
res, = fn(x)
ans = ground_truth_normalizer(x, **new_kwargs)
np.testing.assert_allclose(ans, res, rtol=1e-5)
3
Example 5
Project: treeano Source File: theanode_test.py
def test_repeat_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("in", shape=(3,)),
tn.RepeatNode("r", repeats=2, axis=0)]
).network()
fn = network.function(["in"], ["s"])
x = np.random.randn(3).astype(fX)
np.testing.assert_allclose(np.repeat(x, 2, 0),
fn(x)[0])
3
Example 6
Project: treeano Source File: simple_test.py
def test_reference_node():
network = tn.SequentialNode("s", [
tn.InputNode("input1", shape=(3, 4, 5)),
tn.InputNode("input2", shape=(5, 4, 3)),
tn.ReferenceNode("ref", reference="input1"),
]).network()
fn = network.function(["input1"], ["ref"])
x = np.random.randn(3, 4, 5).astype(fX)
np.testing.assert_allclose(fn(x)[0], x)
3
Example 7
Project: treeano Source File: gradient_test.py
def test_gradient_reversal():
v = np.random.randn(3, 4).astype(fX)
m = T.matrix()
s1 = m.sum()
g1 = T.grad(s1, m)
s2 = ttg.gradient_reversal(s1)
g2 = T.grad(s2, m)
g1_res, g2_res, s1_res, s2_res = theano.function([m], [g1, g2, s1, s2])(v)
np.testing.assert_allclose(v.sum(), s1_res, rtol=1e-5)
np.testing.assert_equal(s1_res, s2_res)
np.testing.assert_equal(np.ones((3, 4), dtype=fX), g1_res)
np.testing.assert_equal(g1_res, -g2_res)
3
Example 8
Project: lda2vec Source File: test_fake_data.py
def test_orthogonal_matrix_covariance():
msg = "Orthogonal matrix should have less covariance than a random matrix"
orth = Variable(fake_data.orthogonal_matrix([20, 20]).astype('float32'))
rand = Variable(np.random.randn(20, 20).astype('float32'))
orth_cc = cross_covariance(orth, orth).data
rand_cc = cross_covariance(rand, rand).data
assert orth_cc < rand_cc, msg
3
Example 9
Project: pylearn2 Source File: test_autoencoder.py
def test_autoencoder_logistic_linear_tied():
data = np.random.randn(10, 5).astype(config.floatX)
ae = Autoencoder(5, 7, act_enc='sigmoid', act_dec='linear',
tied_weights=True)
w = ae.weights.get_value()
ae.hidbias.set_value(np.random.randn(7).astype(config.floatX))
hb = ae.hidbias.get_value()
ae.visbias.set_value(np.random.randn(5).astype(config.floatX))
vb = ae.visbias.get_value()
d = tensor.matrix()
result = np.dot(1. / (1 + np.exp(-hb - np.dot(data, w))), w.T) + vb
ff = theano.function([d], ae.reconstruct(d))
assert _allclose(ff(data), result)
3
Example 10
def test_variable2():
s = treeano.core.variable.VariableWrapper("foo",
shape=(3, 4, 5),
is_shared=True,
inits=[])
assert s.value.sum() == 0
x = np.random.randn(3, 4, 5).astype(theano.config.floatX)
s.value = x
assert np.allclose(s.value, x)
try:
s.value = np.random.randn(5, 4, 3)
except:
pass
else:
assert False
3
Example 11
Project: treeano Source File: irregular_length_test.py
def test_irregular_length_attention_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("l", shape=(None,)),
tn.InputNode("i", shape=(None, 3)),
irregular_length.irregular_length_attention_node(
"foo",
lengths_reference="l",
num_units=3,
output_units=None)]
).network()
nt.assert_equal((None, 3), network["foo"].get_vw("default").shape)
fn = network.function(["i", "l"], ["s"])
x = np.random.randn(15, 3).astype(fX)
l = np.array([2, 3, 7, 3], dtype=fX)
res = fn(x, l)[0].shape
ans = (4, 3)
nt.assert_equal(ans, res)
3
Example 12
Project: treeano Source File: upsample_test.py
def test_repeat_n_d_node2():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(3, 4, 5)),
tn.RepeatNDNode("r", upsample_factor=(1, 1, 1))]).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(3, 4, 5).astype(fX)
np.testing.assert_equal(x,
fn(x)[0])
3
Example 13
Project: treeano Source File: downsample_test.py
def test_global_mean_pool_2d_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(6, 5, 4, 3)),
tn.GlobalMeanPool2DNode("gp")]
).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(6, 5, 4, 3).astype(fX)
ans = x.mean(axis=(2, 3))
np.testing.assert_allclose(ans,
fn(x)[0],
rtol=1e-5,
atol=1e-7)
3
Example 14
Project: treeano Source File: lrn_test.py
def test_local_response_normalization_node_shape():
for ndim in [2, 3, 4, 5, 6]:
shape = (3,) * ndim
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=shape),
lrn.LocalResponseNormalizationNode("lrn")]
).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(*shape).astype(fX)
res = fn(x)[0].shape
np.testing.assert_equal(shape, res)
3
Example 15
Project: lda2vec Source File: test_dirichlet_likelihood.py
def test_concentration():
""" Test that alpha > 1.0 on a dense vector has a higher likelihood
than alpha < 1.0 on a dense vector, and test that a sparse vector
has the opposite character. """
dense = np.random.randn(5, 10).astype('float32')
sparse = np.random.randn(5, 10).astype('float32')
sparse[:, 1:] /= 1e5
weights = Variable(dense)
dhl_dense_10 = dirichlet_likelihood(weights, alpha=10.0).data
dhl_dense_01 = dirichlet_likelihood(weights, alpha=0.1).data
weights = Variable(sparse)
dhl_sparse_10 = dirichlet_likelihood(weights, alpha=10.0).data
dhl_sparse_01 = dirichlet_likelihood(weights, alpha=0.1).data
msg = "Sparse vector has higher likelihood than dense with alpha=0.1"
assert dhl_sparse_01 > dhl_dense_01, msg
msg = "Dense vector has higher likelihood than sparse with alpha=10.0"
assert dhl_dense_10 > dhl_sparse_10, msg
3
Example 16
Project: treeano Source File: utils_test.py
def test_stable_softmax_grad():
x = theano.shared(np.random.randn(50, 50).astype(fX))
s1 = T.nnet.softmax(x)
s2 = treeano.utils.stable_softmax(
x.reshape([50, 1, 5, 10]),
axis=(2, 3)
).reshape([50, 50])
np.testing.assert_allclose(s1.eval(), s2.eval(), rtol=1e-5)
g1 = T.grad(s1[:10, :10].sum(), x)
g2 = T.grad(s2[:10, :10].sum(), x)
np.testing.assert_allclose(g1.eval(), g2.eval(), rtol=1e-5)
3
Example 17
Project: treeano Source File: simple_test.py
def test_add_bias_node():
network = tn.SequentialNode("s", [
tn.InputNode("in", shape=(3, 4, 5)),
tn.AddBiasNode("b", broadcastable_axes=())
]).network()
bias_var = network["b"].get_vw("bias")
fn = network.function(["in"], ["s"])
x = np.random.randn(3, 4, 5).astype(fX)
y = np.random.randn(3, 4, 5).astype(fX)
# test that bias is 0 initially
np.testing.assert_allclose(fn(x)[0], x)
# set bias_var value to new value
bias_var.value = y
# test that adding works
np.testing.assert_allclose(fn(x)[0], x + y)
3
Example 18
Project: hessianfree Source File: test_gpu.py
def test_rnn_calc_G(dtype):
inputs = np.random.randn(1000, 10, 1).astype(dtype)
rnn = hf.RNNet([1, 10, 1], debug=(dtype == np.float64), use_GPU=True)
rnn.cache_minibatch(inputs, inputs)
rnn.optimizer = hf.opt.HessianFree()
v = np.random.randn(rnn.W.size).astype(dtype)
gpu_Gv = rnn.GPU_calc_G(v)
cpu_Gv = rnn.calc_G(v)
assert np.allclose(gpu_Gv, cpu_Gv, rtol=1e-4)
3
Example 19
Project: tvb-library Source File: _numba_tests.py
def _run_n_node(self, n_node):
weights = numpy.random.randn(n_node, n_node).astype('f')
state = numpy.random.randn(n_node, 2, self.n_thread).astype('f')
out = numpy.zeros((n_node, self.n_thread)).astype('f')
offset = 0.5
cf = cu_simple_cfun(offset, 1)
@self.jit_and_run(out, weights, state)
def kernel(out, weights, state):
t = cuda.blockDim.x * cuda.blockIdx.x + cuda.threadIdx.x
for i in range(weights.shape[0]):
out[i, t] = cf(weights, state, i, t)
expected = weights.dot(state[:, 1] + offset)
ok = numpy.allclose(expected, out, 1e-4, 1e-5)
se = numpy.sqrt((expected - out)**2)
numpy.testing.assert_allclose(out, expected, 1e-4, 1e-5)
3
Example 20
Project: parakeet Source File: test_simple_regression.py
def test_simple_regression():
N = 10
x = np.random.randn(N).astype('float64')
slope = 903.29
offset = 102.1
y = slope * x + offset
expect(fit_simple_regression, [x,y], (slope,offset))
3
Example 21
Project: treeano Source File: stochastic_test.py
def test_gaussian_dropout_node():
def make_network(p):
return tn.SequentialNode("s", [
tn.InputNode("i", shape=(3, 4, 5)),
tn.GaussianDropoutNode("do", p=p)
]).network()
x = np.random.randn(3, 4, 5).astype(fX)
fn1 = make_network(0).function(["i"], ["s"])
np.testing.assert_allclose(fn1(x)[0], x)
@nt.raises(AssertionError)
def test_not_identity():
fn2 = make_network(0.5).function(["i"], ["s"])
np.testing.assert_allclose(fn2(x)[0], x)
test_not_identity()
3
Example 22
Project: treeano Source File: theanode_test.py
def test_reshape_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("in", shape=(3, 4, 5)),
tn.ReshapeNode("r", shape=(5, 12))]
).network()
fn = network.function(["in"], ["s"])
x = np.random.randn(3, 4, 5).astype(fX)
res = fn(x)[0]
np.testing.assert_allclose(res,
x.reshape(5, 12))
3
Example 23
Project: kaggle-galaxies Source File: realtime_augmentation.py
def post_augment_brightness_gen(data_gen, std=0.5):
for target_arrays, chunk_size in data_gen:
labels = target_arrays.pop()
stds = np.random.randn(chunk_size).astype('float32') * std
noise = stds[:, None] * colour_channel_weights[None, :]
target_arrays = [np.clip(t + noise[:, None, None, :], 0, 1) for t in target_arrays]
target_arrays.append(labels)
yield target_arrays, chunk_size
3
Example 24
Project: treeano Source File: theanode_test.py
def test_dimshuffle_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("in", shape=(3, 4, 5)),
tn.DimshuffleNode("r", pattern=(1, "x", 0, 2))]
).network()
fn = network.function(["in"], ["s"])
x = np.random.randn(3, 4, 5).astype(fX)
ans = T.constant(x).dimshuffle(1, "x", 0, 2).eval()
res = fn(x)[0]
np.testing.assert_equal(res.shape, ans.shape)
np.testing.assert_equal(res, ans)
3
Example 25
def test_dense_node():
network = tn.SequentialNode(
"seq",
[tn.InputNode("in", shape=(3, 4, 5)),
tn.DenseNode("fc1", num_units=6),
tn.DenseNode("fc2", num_units=7),
tn.DenseNode("fc3", num_units=8)]
).network()
x = np.random.randn(3, 4, 5).astype(fX)
fn = network.function(["in"], ["fc3"])
res = fn(x)[0]
nt.assert_equal(res.shape, (3, 8))
3
Example 26
Project: treeano Source File: gradnet_test.py
def test_grad_net_interpolation_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(1, 10)),
gradnet.GradNetInterpolationNode(
"gradnet",
{"early": tn.ReLUNode("r"),
"late": tn.TanhNode("t")},
late_gate=0.5)]
).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(1, 10).astype(fX)
ans = 0.5 * np.clip(x, 0, np.inf) + 0.5 * np.tanh(x)
np.testing.assert_allclose(ans, fn(x)[0], rtol=1e-5)
3
Example 27
Project: treeano Source File: activation_transformation_test.py
def test_concatenate_negation_node():
# just testing that it runs
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(10, 10)),
activation_transformation.ConcatenateNegationNode("a")]).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(10, 10).astype(fX)
ans = np.concatenate([x, -x], axis=1)
np.testing.assert_allclose(ans, fn(x)[0])
3
Example 28
Project: treeano Source File: downsample_test.py
def test_custom_global_pool_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(6, 5, 4, 3)),
tn.CustomGlobalPoolNode("gp", pool_function=T.mean)]
).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(6, 5, 4, 3).astype(fX)
ans = x.mean(axis=(2, 3))
np.testing.assert_allclose(ans,
fn(x)[0],
rtol=1e-5,
atol=1e-7)
3
Example 29
Project: treeano Source File: kumaraswamy_unit_test.py
def test_kumaraswamy_unit_node():
# just testing that it runs
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(100,)),
ku.KumaraswamyUnitNode("k")]).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(100).astype(fX)
fn(x)
3
Example 30
Project: deep_recommend_system Source File: sparse_tensor_dense_matmul_grad_test.py
Function: randomtensor
Function: randomtensor
def _randomTensor(self, size, np_dtype, adjoint=False, sparse=False):
n, m = size
x = np.random.randn(n, m).astype(np_dtype)
if adjoint:
x = x.transpose()
if sparse:
return self._sparsify(x)
else:
return tf.constant(x, dtype=np_dtype)
3
Example 31
Project: treeano Source File: lrn_test.py
def test_local_response_normalization_2d_node_shape():
shape = (3, 4, 5, 6)
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=shape),
lrn.LocalResponseNormalization2DNode("lrn")]
).network()
fn = network.function(["i"], ["s"])
x = np.random.randn(*shape).astype(fX)
res = fn(x)[0].shape
np.testing.assert_equal(shape, res)
3
Example 32
Project: treeano Source File: monitor_test.py
def test_monitor_variance_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("x", shape=(3, 4, 5)),
tn.MonitorVarianceNode("mv")]).network()
vw = network["mv"].get_vw("var")
x = np.random.randn(3, 4, 5).astype(fX)
ans = x.var()
fn = network.function(["x"], [vw.variable])
np.testing.assert_allclose(fn(x), [ans], rtol=1e-5)
3
Example 33
Project: treeano Source File: paired_conv_test.py
def test_paired_conv_2d_with_bias_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(3, 4, 5, 6)),
paired_conv.PairedConvNode(
"c",
{"conv": tn.Conv2DWithBiasNode("c_conv"),
"separator": tn.IdentityNode("sep")},
filter_size=(2, 2),
num_filters=7,
pad="same")]
).network()
fn = network.function(["i"], ["s"])
res = fn(np.random.randn(3, 4, 5, 6).astype(fX))[0]
np.testing.assert_equal((3, 7, 5, 6), res.shape)
3
Example 34
def init_params(self, mu0, sigma_mean0, sigma_std0, sigma_min, sigma_max):
'''
Initialize parameters
mu = random normal with std mu0, mean 0
Sigma = random normal with mean sigma_mean0, std sigma_std0,
and min / max of sigma_min, sigma_max
'''
self.mu = mu0 * np.random.randn(self.N, self.K).astype(DTYPE)
self.Sigma = np.random.randn(*self.Sigma.shape).astype(DTYPE)
self.Sigma *= sigma_std0
self.Sigma += sigma_mean0
self.Sigma = np.maximum(sigma_min, np.minimum(self.Sigma, sigma_max))
3
Example 35
Project: treeano Source File: recurrent_convolution_test.py
Function: test_default_recurrent_conv_2d_node
Function: test_default_recurrent_conv_2d_node
def test_default_recurrent_conv_2d_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(3, 4, 5, 6)),
rcl.DefaultRecurrentConv2DNode("a",
num_filters=7,
filter_size=(3, 3),
pad="same")]
).network()
fn = network.function(["i"], ["s"])
res = fn(np.random.randn(3, 4, 5, 6).astype(fX))[0]
np.testing.assert_equal((3, 7, 5, 6), res.shape)
3
Example 36
Project: treeano Source File: simple_test.py
def test_send_to_node():
network = tn.ContainerNode("c", [
tn.SequentialNode(
"s1",
[tn.InputNode("in", shape=(3, 4, 5)),
tn.SendToNode("stn1", reference="s2")]),
tn.SequentialNode(
"s2",
[tn.SendToNode("stn2", reference="stn3")]),
tn.SequentialNode(
"s3",
[tn.SendToNode("stn3", reference="i")]),
tn.IdentityNode("i"),
]).network()
fn = network.function(["in"], ["i"])
x = np.random.randn(3, 4, 5).astype(fX)
np.testing.assert_allclose(fn(x)[0], x)
3
Example 37
Project: treeano Source File: utils_test.py
def _clone_test_case(clone_fn):
x = T.matrix("x")
y = T.matrix("y")
x_shape = x.shape
sample_y = np.random.randn(4, 5).astype(theano.config.floatX)
srng = theano.tensor.shared_randomstreams.RandomStreams()
mask = srng.binomial(n=1, p=0.5, size=x_shape)
mask2 = clone_fn([mask], replace={x: y})[0]
mask2.eval({y: sample_y}) # ERROR
3
Example 38
def test_embed():
""" Test that embedding is treated like a Variable"""
embed_dense = L.EmbedID(5, 10)
embed_sparse = L.EmbedID(5, 10)
embed_dense.W.data[:] = np.random.randn(5, 10).astype('float32')
embed_sparse.W.data[:] = np.random.randn(5, 10).astype('float32')
embed_sparse.W.data[:, 1:] /= 1e5
dhl_dense_01 = dirichlet_likelihood(embed_dense, alpha=0.1).data
dhl_sparse_01 = dirichlet_likelihood(embed_sparse, alpha=0.1).data
msg = "Sparse vector has higher likelihood than dense with alpha=0.1"
assert dhl_sparse_01 > dhl_dense_01, msg
3
Example 39
Project: hessianfree Source File: test_gpu.py
def test_ff_calc_G(dtype):
inputs = np.random.randn(1000, 1).astype(dtype)
ff = hf.FFNet([1, 10, 1], debug=(dtype == np.float64), use_GPU=True)
ff.cache_minibatch(inputs, inputs)
v = np.random.randn(ff.W.size).astype(dtype)
gpu_Gv = ff.GPU_calc_G(v)
cpu_Gv = ff.calc_G(v)
assert np.allclose(gpu_Gv, cpu_Gv, rtol=1e-4)
3
Example 40
Project: treeano Source File: simple_test.py
def test_linear_mapping_node():
network = tn.SequentialNode("s", [
tn.InputNode("in", shape=(3, 4, 5)),
tn.LinearMappingNode("linear", output_dim=6),
]).network()
weight_var = network["linear"].get_vw("weight")
fn = network.function(["in"], ["s"])
x = np.random.randn(3, 4, 5).astype(fX)
W = np.random.randn(5, 6).astype(fX)
# test that weight is 0 initially
np.testing.assert_allclose(fn(x)[0], np.zeros((3, 4, 6)))
# set weight_var value to new value
weight_var.value = W
# test that adding works
np.testing.assert_allclose(np.dot(x, W), fn(x)[0], rtol=1e-4, atol=1e-7)
3
Example 41
Project: seya Source File: test_apply.py
def test_apply_model(self):
"""Test keras.models.Sequential.__call__"""
nb_samples, input_dim, output_dim = 3, 10, 5
model = Sequential()
model.add(Dense(output_dim=output_dim, input_dim=input_dim))
model.compile('sgd', 'mse')
X = K.placeholder(ndim=2)
Y = apply_model(model, X)
F = theano.function([X], Y)
x = np.random.randn(nb_samples, input_dim).astype(floatX)
y1 = F(x)
y2 = model.predict(x)
# results of __call__ should match model.predict
assert_allclose(y1, y2)
3
Example 42
Project: attention-lvcsr Source File: test_types.py
def test_cdata():
if not theano.config.cxx:
raise SkipTest("G++ not available, so we need to skip this test.")
i = TensorType('float32', (False,))()
c = ProdOp()(i)
i2 = GetOp()(c)
mode = None
if theano.config.mode == "FAST_COMPILE":
mode = "FAST_RUN"
# This should be a passthrough function for vectors
f = theano.function([i], i2, mode=mode)
v = numpy.random.randn(9).astype('float32')
v2 = f(v)
assert (v2 == v).all()
3
Example 43
Project: GPflow Source File: test_transforms.py
def setUp(self):
tf.reset_default_graph()
self.x = tf.placeholder(float_type)
self.x_np = np.random.randn(10).astype(np_float_type)
self.session = tf.Session()
self.transforms = [C() for C in GPflow.transforms.Transform.__subclasses__()]
self.transforms.append(GPflow.transforms.Logistic(7.3, 19.4))
3
Example 44
Project: treeano Source File: stochastic_test.py
def test_dropout_node():
def make_network(p):
return tn.SequentialNode("s", [
tn.InputNode("i", shape=(3, 4, 5)),
tn.DropoutNode("do", p=p)
]).network()
x = np.random.randn(3, 4, 5).astype(fX)
fn1 = make_network(0).function(["i"], ["s"])
np.testing.assert_allclose(fn1(x)[0], x)
@nt.raises(AssertionError)
def test_not_identity():
fn2 = make_network(0.5).function(["i"], ["s"])
np.testing.assert_allclose(fn2(x)[0], x)
test_not_identity()
3
Example 45
Project: treeano Source File: fns_test.py
def test_transform_node_data_postwalk():
network1 = tn.InputNode("i", shape=(3, 4, 5)).network()
def change_it_up(obj):
if obj == (3, 4, 5):
return (6, 7, 8)
elif obj == "i":
return "foo"
else:
return obj
network2 = canopy.transforms.transform_node_data_postwalk(network1,
change_it_up)
x = np.random.randn(6, 7, 8).astype(fX)
fn = network2.function(["foo"], ["foo"])
np.testing.assert_equal(x, fn(x)[0])
3
Example 46
Project: chainer Source File: test_det.py
def det_scaling(self, gpu=False):
scaling = numpy.random.randn(1).astype('float32')
if gpu:
cx = cuda.to_gpu(self.x)
sx = cuda.to_gpu(scaling * self.x)
else:
cx = self.x
sx = scaling * self.x
c = float(scaling ** self.x.shape[1])
cxv = chainer.Variable(cx)
sxv = chainer.Variable(sx)
cxd = self.det(cxv)
sxd = self.det(sxv)
testing.assert_allclose(cxd.data * c, sxd.data)
3
Example 47
def test_sort(self):
x = tensor.matrix()
self._compile_and_check(
[x],
[sort(x)],
[np.random.randn(10, 40).astype(theano.config.floatX)],
SortOp)
self._compile_and_check(
[x],
[sort(x, axis=None)],
[np.random.randn(10, 40).astype(theano.config.floatX)],
SortOp)
3
Example 48
Project: kaggle-galaxies Source File: realtime_augmentation.py
def post_augment_gaussian_noise_gen(data_gen, std=0.1):
"""
Adds gaussian noise. Note that this is not entirely correct, the correct way would be to do it
before downsampling, so the regular image and the rot45 image have the same noise pattern.
But this is easier.
"""
for target_arrays, chunk_size in data_gen:
labels = target_arrays.pop()
noise = np.random.randn(*target_arrays[0].shape).astype('float32') * std
target_arrays = [np.clip(t + noise, 0, 1) for t in target_arrays]
target_arrays.append(labels)
yield target_arrays, chunk_size
3
Example 49
Project: treeano Source File: composite_test.py
def test_dense_combine_node():
network = tn.SequentialNode(
"seq",
[tn.InputNode("in", shape=(3, 4, 5)),
tn.DenseCombineNode("fc1", [tn.IdentityNode("i1")], num_units=6),
tn.DenseCombineNode("fc2", [tn.IdentityNode("i2")], num_units=7),
tn.DenseCombineNode("fc3", [tn.IdentityNode("i3")], num_units=8)]
).network()
x = np.random.randn(3, 4, 5).astype(fX)
fn = network.function(["in"], ["fc3"])
res = fn(x)[0]
nt.assert_equal(res.shape, (3, 8))
3
Example 50
def test_default_recurrent_conv_2d_node():
network = tn.SequentialNode(
"s",
[tn.InputNode("i", shape=(3, 4, 5, 6)),
deconv_upsample.DeconvUpsample2DNode(
"a",
num_filters=7,
upsample_factor=(2, 2),
filter_size=(3, 3),
)]
).network()
fn = network.function(["i"], ["s"])
res = fn(np.random.randn(3, 4, 5, 6).astype(fX))[0]
np.testing.assert_equal((3, 7, 10, 12), res.shape)
np.testing.assert_equal((3, 7, 10, 12),
network['a'].get_vw('default').shape)