Add summary.writer before invoking trace_export

PiperOrigin-RevId: 236422330

tensorflow/python/keras/callbacks.py

@@ -1140,7 +1140,7 @@ class TensorBoard(Callback):
         your training.
       profile_batch: Profile the batch to sample compute characteristics. By
         default, it will profile the second batch. Set profile_batch=0 to
-        disable profiling.
+        disable profiling. Must run in TensorFlow eager mode.
 
   Raises:
       ValueError: If histogram_freq is set and no validation data is provided.
@@ -1274,16 +1274,10 @@ class TensorBoard(Callback):
       self._samples_seen_at_last_write = self._samples_seen
     self._total_batches_seen += 1
     if self._is_tracing:
-      # TODO(b/126388999): Remove step info in the summary name.
-      summary_ops_v2.trace_export(
-          name='batch_%d' % self._total_batches_seen,
-          step=self._total_batches_seen,
-          profiler_outdir=self.log_dir)
-      self._is_tracing = False
+      self._log_trace()
     elif (not self._is_tracing and
           self._total_batches_seen == self._profile_batch - 1):
-      summary_ops_v2.trace_on(graph=True, profiler=True)
-      self._is_tracing = True
+      self._enable_trace()
 
   def on_epoch_end(self, epoch, logs=None):
     """Runs metrics and histogram summaries at epoch end."""
@@ -1294,13 +1288,24 @@ class TensorBoard(Callback):
       self._log_weights(epoch)
 
   def on_train_end(self, logs=None):
-    self._close_writers()
     if self._is_tracing:
-      # TODO(b/126388999): Remove step info in the summary name.
-      summary_ops_v2.trace_export(
-          name='batch_%d' % self._total_batches_seen,
-          step=self._total_batches_seen,
-          profiler_outdir=self.log_dir)
+      self._log_trace()
+    self._close_writers()
+
+  def _enable_trace(self):
+    if context.executing_eagerly():
+      summary_ops_v2.trace_on(graph=True, profiler=True)
+      self._is_tracing = True
+
+  def _log_trace(self):
+    if context.executing_eagerly():
+      with self._train_writer.as_default(), \
+          summary_ops_v2.always_record_summaries():
+        # TODO(b/126388999): Remove step info in the summary name.
+        summary_ops_v2.trace_export(
+            name='batch_%d' % self._total_batches_seen,
+            step=self._total_batches_seen,
+            profiler_outdir=self.log_dir)
       self._is_tracing = False
 
   def _log_metrics(self, logs, prefix, step):

tensorflow/python/keras/callbacks_test.py

@@ -1236,17 +1236,28 @@ class TestTensorBoardV2(keras_parameterized.TestCase):
 
 
 # Note that this test specifies model_type explicitly.
-class TestTensorBoardV2WriteModelTest(test.TestCase):
+class TestTensorBoardV2NonParameterizedTest(keras_parameterized.TestCase):
 
   def setUp(self):
-    super(TestTensorBoardV2WriteModelTest, self).setUp()
+    super(TestTensorBoardV2NonParameterizedTest, self).setUp()
     self.logdir = os.path.join(self.get_temp_dir(), 'tb')
     self.train_dir = os.path.join(self.logdir, 'train')
     self.validation_dir = os.path.join(self.logdir, 'validation')
 
+  def _get_seq_model(self):
+    model = keras.models.Sequential([
+        keras.layers.Conv2D(8, (3, 3), input_shape=(10, 10, 1)),
+        keras.layers.Flatten(),
+        keras.layers.Dense(1),
+    ])
+    model.compile('sgd', 'mse', run_eagerly=testing_utils.should_run_eagerly())
+    return model
+
   def fitModelAndAssertKerasModelWritten(self, model):
     x, y = np.ones((10, 10, 10, 1)), np.ones((10, 1))
-    tb_cbk = keras.callbacks.TensorBoard(self.logdir, write_graph=True)
+    tb_cbk = keras.callbacks.TensorBoard(self.logdir,
+                                         write_graph=True,
+                                         profile_batch=0)
     model.fit(
         x,
         y,
@@ -1289,6 +1300,74 @@ class TestTensorBoardV2WriteModelTest(test.TestCase):
     model.compile('sgd', 'mse', run_eagerly=False)
     self.fitModelAndAssertKerasModelWritten(model)
 
+  # TODO(b/126944683): Put parameterization in the class when graph is fixed.
+  @keras_parameterized.run_all_keras_modes(always_skip_v1=True)
+  def test_TensorBoard_autoTrace(self):
+    model = self._get_seq_model()
+    x, y = np.ones((10, 10, 10, 1)), np.ones((10, 1))
+    tb_cbk = keras.callbacks.TensorBoard(
+        self.logdir, histogram_freq=1, profile_batch=1, write_graph=False)
+
+    model.fit(
+        x,
+        y,
+        batch_size=2,
+        epochs=2,
+        validation_data=(x, y),
+        callbacks=[tb_cbk])
+    summary_file = list_summaries(self.logdir)
+
+    self.assertEqual(
+        summary_file.tensors,
+        {
+            _ObservedSummary(logdir=self.train_dir, tag=u'batch_1'),
+        },
+    )
+
+  # TODO(b/126944683): Put parameterization in the class when graph is fixed.
+  @keras_parameterized.run_all_keras_modes(always_skip_v1=True)
+  def test_TensorBoard_autoTrace_tagNameWithBatchNum(self):
+    model = self._get_seq_model()
+    x, y = np.ones((10, 10, 10, 1)), np.ones((10, 1))
+    tb_cbk = keras.callbacks.TensorBoard(
+        self.logdir, histogram_freq=1, profile_batch=2, write_graph=False)
+
+    model.fit(
+        x,
+        y,
+        batch_size=2,
+        epochs=2,
+        validation_data=(x, y),
+        callbacks=[tb_cbk])
+    summary_file = list_summaries(self.logdir)
+
+    self.assertEqual(
+        summary_file.tensors,
+        {
+            _ObservedSummary(logdir=self.train_dir, tag=u'batch_2'),
+        },
+    )
+
+  # TODO(b/126944683): Put parameterization in the class when graph is fixed.
+  @keras_parameterized.run_all_keras_modes(always_skip_v1=True)
+  def test_TensorBoard_autoTrace_profile_batch_largerThanBatchCount(self):
+    model = self._get_seq_model()
+    x, y = np.ones((10, 10, 10, 1)), np.ones((10, 1))
+    tb_cbk = keras.callbacks.TensorBoard(
+        self.logdir, histogram_freq=1, profile_batch=10000, write_graph=False)
+
+    model.fit(
+        x,
+        y,
+        batch_size=2,
+        epochs=2,
+        validation_data=(x, y),
+        callbacks=[tb_cbk])
+    summary_file = list_summaries(self.logdir)
+
+    # Enabled trace only on the 10000th batch, thus it should be empty.
+    self.assertEmpty(summary_file.tensors)
+
 
 if __name__ == '__main__':
   test.main()