28 lines
856 B
Python
28 lines
856 B
Python
import numpy as np
|
|
import tensorflow as tf
|
|
|
|
import keras_core
|
|
from keras_core.testing import test_case
|
|
from keras_core.utils import rng_utils
|
|
|
|
|
|
class TestRandomSeedSetting(test_case.TestCase):
|
|
def test_set_random_seed(self):
|
|
def get_model_output():
|
|
model = keras_core.Sequential(
|
|
[
|
|
keras_core.layers.Dense(10),
|
|
keras_core.layers.Dropout(0.5),
|
|
keras_core.layers.Dense(10),
|
|
]
|
|
)
|
|
x = np.random.random((32, 10)).astype("float32")
|
|
ds = tf.data.Dataset.from_tensor_slices(x).shuffle(32).batch(16)
|
|
return model.predict(ds)
|
|
|
|
rng_utils.set_random_seed(42)
|
|
y1 = get_model_output()
|
|
rng_utils.set_random_seed(42)
|
|
y2 = get_model_output()
|
|
self.assertAllClose(y1, y2)
|