fix(tests): Update imports in keras (#621)

* fix(tests): Update imports in keras

* Fix test

src/dvclive/keras.py

@@ -2,14 +2,13 @@
 import os
 from typing import Dict, Optional
 
-from tensorflow.keras.callbacks import Callback
-from tensorflow.keras.models import load_model
+import tensorflow as tf
 
 from dvclive import Live
 from dvclive.utils import standardize_metric_name
 
 
-class DVCLiveCallback(Callback):
+class DVCLiveCallback(tf.keras.callbacks.Callback):
     def __init__(
         self,
         model_file=None,
@@ -31,7 +30,7 @@ def on_train_begin(self, logs=None):
             if self.save_weights_only:
                 self.model.load_weights(self.model_file)
             else:
-                self.model = load_model(self.model_file)
+                self.model = tf.keras.models.load_model(self.model_file)
 
     def on_epoch_end(self, epoch: int, logs: Optional[Dict] = None):
         logs = logs or {}

tests/test_frameworks/test_keras.py

@@ -15,18 +15,17 @@
 @pytest.fixture()
 def xor_model():
     import numpy as np
-    from tensorflow.python.keras import Sequential
-    from tensorflow.python.keras.layers import Activation, Dense
+    import tensorflow as tf
 
     def make():
         x = np.array([[0, 0], [0, 1], [1, 0], [1, 1]])
         y = np.array([[0], [1], [1], [0]])
 
-        model = Sequential()
-        model.add(Dense(8, input_dim=2))
-        model.add(Activation("relu"))
-        model.add(Dense(1))
-        model.add(Activation("sigmoid"))
+        model = tf.keras.Sequential()
+        model.add(tf.keras.layers.Dense(8, input_dim=2))
+        model.add(tf.keras.layers.Activation("relu"))
+        model.add(tf.keras.layers.Dense(1))
+        model.add(tf.keras.layers.Activation("sigmoid"))
 
         model.compile(loss="binary_crossentropy", optimizer="sgd", metrics=["accuracy"])
 
@@ -90,8 +89,6 @@ def test_keras_model_file(tmp_dir, xor_model, mocker, save_weights_only):
 
 @pytest.mark.parametrize("save_weights_only", [True, False])
 def test_keras_load_model_on_resume(tmp_dir, xor_model, mocker, save_weights_only):
-    import dvclive.keras
-
     model, x, y = xor_model()
 
     if save_weights_only:
@@ -100,7 +97,6 @@ def test_keras_load_model_on_resume(tmp_dir, xor_model, mocker, save_weights_onl
         model.save("model.h5")
 
     load_weights = mocker.spy(model, "load_weights")
-    load_model = mocker.spy(dvclive.keras, "load_model")
 
     model.fit(
         x,
@@ -116,7 +112,6 @@ def test_keras_load_model_on_resume(tmp_dir, xor_model, mocker, save_weights_onl
         ],
     )
 
-    assert load_model.call_count != save_weights_only
     assert load_weights.call_count == save_weights_only