def test_random_non_multiple(self):
""" Test indices generator when number of samples
is not a multiple of the batch size
and when generation is random
"""
shuffle = True
batch_size = 3
n = 10
np.random.seed(0)
n_batches = int(np.ceil(n * 1. / batch_size))
sample_indices = []
batch_indices = []
for batch in indices_generator(shuffle, batch_size, n):
new_sample_indices, new_batch_index = batch
sample_indices += new_sample_indices.tolist()
batch_indices.append(new_batch_index)
assert sample_indices != np.arange(n).tolist()
assert batch_indices != np.arange(n_batches).tolist()
assert set(sample_indices) == set(np.arange(n))
assert set(batch_indices) == set(np.arange(n_batches))
def get_epoch_train(self):
assert self.on_memory, "only for data loaded on memory"
for (batch, idx) in indices_generator(shuffle=True,
batch_size=Cfg.batch_size,
n=int(self.n_train)):
yield self._X_train[batch], self._y_train[batch], idx
def test_deterministic_multiple(self):
""" Test indices generator when number of samples
is a multiple of the batch size
and when generation is deterministic
"""
shuffle = False
batch_size = 5
n = 10
n_batches = n / batch_size
sample_indices = []
batch_indices = []
for batch in indices_generator(shuffle, batch_size, n):
new_sample_indices, new_batch_index = batch
sample_indices += new_sample_indices.tolist()
batch_indices.append(new_batch_index)
assert sample_indices == np.arange(n).tolist()
assert batch_indices == np.arange(n_batches).tolist()
def get_epoch_test(self):
for (batch, idx) in indices_generator(shuffle=False,
batch_size=Cfg.batch_size,
n=self.n_test):
yield self._X_test[batch], self._y_test[batch], idx