def evaluate(model, X_test, Y_test, eval_size, batch_size):
N_test = X_test.shape[0] if eval_size is None else eval_size
if N_test > X_test.shape[0]:
raise ValueError('Test size can be no larger than {}'.format(
X_test.shape[0]))
model.no_grad()
# TODO(beam2d): make chx.array(0, dtype=...) work
total_loss = chx.zeros((), dtype=chx.float32)
num_correct = chx.zeros((), dtype=chx.int64)
for i in range(0, N_test, batch_size):
x = X_test[i:min(i + batch_size, N_test)]
t = Y_test[i:min(i + batch_size, N_test)]
y = model(x)
total_loss += compute_loss(y, t) * batch_size
num_correct += (y.argmax(axis=1).astype(t.dtype) == t).astype(
chx.int32).sum()
model.require_grad()
mean_loss = float(total_loss) / N_test
accuracy = int(num_correct) / N_test
return mean_loss, accuracy
def new_linear_params(n_in, n_out):
W = np.random.randn(n_out, n_in).astype(
np.float32) # TODO(beam2d): not supported in chx
W /= np.sqrt(n_in) # TODO(beam2d): not supported in chx
W = chx.array(W)
b = chx.zeros(n_out, dtype=chx.float32)
return W, b
def new_linear_params(n_in, n_out):
W = np.random.randn(n_out, n_in).astype(
np.float32) # TODO(beam2d): not supported in chx
W /= np.sqrt(n_in) # TODO(beam2d): not supported in chx
W = chx.array(W)
b = chx.zeros(n_out, dtype=chx.float32)
return W, b
def forward_chainerx(self, inputs):
# Computes the forward pass in ChainerX.
#
# In case of an acceptable error, the error is stored and a dummy
# output array is returned.
#
# The detected errors are checked in `check_forward_outputs`, and
# also in `_create_test_entry_function` to skip
# backward/double-backward tests.
accept_errors = self.__get_accept_errors()
try:
outputs = self.forward_xp(inputs, chainerx)
self.__forward_error_chainerx = 'ok'
except accept_errors as e:
# Keep detected error
self.__forward_error_chainerx = e
# A dummy output array is returned
y = chainerx.zeros((0, ), 'float32')
outputs = y,
return outputs
def forward_chainerx(self, inputs):
# Computes the forward pass in ChainerX.
#
# In case of an acceptable error, the error is stored and a dummy
# output array is returned.
#
# The detected errors are checked in `check_forward_outputs`, and
# also in `_create_test_entry_function` to skip
# backward/double-backward tests.
accept_errors = self.__get_accept_errors()
try:
outputs = self.forward_xp(inputs, chainerx)
self.__forward_error_chainerx = 'ok'
except accept_errors as e:
# Keep detected error
self.__forward_error_chainerx = e
# A dummy output array is returned
y = chainerx.zeros((0,), 'float32')
outputs = y,
return outputs
def test_zeros_with_device(device):
a = chainerx.zeros((2,), 'float32', device)
b = chainerx.zeros((2,), 'float32')
chainerx.testing.assert_array_equal_ex(a, b)
array_utils.check_device(a, device)
def test_zeros_with_device(device):
a = chainerx.zeros((2,), 'float32', device)
b = chainerx.zeros((2,), 'float32')
chainerx.testing.assert_array_equal_ex(a, b)
array_utils.check_device(a, device)
def __init__(self, n_in, n_out):
W = np.random.randn(n_in, n_out).astype(np.float32)
W /= np.sqrt(n_in)
self.W = chx.array(W)
self.b = chx.zeros((n_out, ), dtype=chx.float32)
def __init__(self, size, dtype=chx.float32):
shape = size,
self.avg_mean = chx.zeros(shape, dtype)
self.avg_var = chx.zeros(shape, dtype)
self.gamma = chx.ones(shape, dtype)
self.beta = chx.zeros(shape, dtype)
def __init__(self, n_in, n_out):
W = np.random.randn(n_in, n_out).astype(np.float32)
W /= np.sqrt(n_in)
self.W = chx.array(W)
self.b = chx.zeros((n_out,), dtype=chx.float32)
def __init__(self, size, dtype=chx.float32):
shape = size,
self.avg_mean = chx.zeros(shape, dtype)
self.avg_var = chx.zeros(shape, dtype)
self.gamma = chx.ones(shape, dtype)
self.beta = chx.zeros(shape, dtype)
