def testFeaturesStats(self):
num_epochs = 5
total_records = num_epochs * self._num_records
batch_size = 2
stats_aggregator = stats_ops.StatsAggregator()
dataset = self.make_batch_feature(
filenames=self.test_filenames[0],
num_epochs=num_epochs,
batch_size=batch_size,
shuffle=True,
shuffle_seed=5,
drop_final_batch=True).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
for _ in range(total_records // batch_size):
sess.run(next_element)
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasCount(
sess.run(summary_t), "record_stats:features", total_records)
self._assertSummaryHasCount(
sess.run(summary_t), "record_stats:feature-values", total_records)
self._assertSummaryHasSum(
sess.run(summary_t), "record_stats:features", total_records * 3)
self._assertSummaryHasSum(
sess.run(summary_t), "record_stats:feature-values",
self._sum_keywords(1) * num_epochs + 2 * total_records)
def testBytesProduced(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(100).map(
lambda x: array_ops.tile([x], ops.convert_to_tensor([x]))).apply(
stats_ops.bytes_produced_stats("bytes_produced")).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
expected_sum = 0.0
for i in range(100):
self.assertAllEqual(
np.array([i] * i, dtype=np.int64), sess.run(next_element))
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str, "bytes_produced", float(i + 1))
expected_sum += i * 8.0
self._assertSummaryHasSum(summary_str, "bytes_produced", expected_sum)
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str, "bytes_produced", 100.0)
self._assertSummaryHasSum(summary_str, "bytes_produced", expected_sum)
def testLatencyStatsOptimization(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.from_tensors(1).apply(
optimization.assert_next(
["LatencyStats", "Map", "LatencyStats", "Prefetch",
"LatencyStats"])).map(lambda x: x * x).prefetch(1).apply(
optimization.optimize(["latency_all_edges"])).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
get_next = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
self.assertEqual(1 * 1, sess.run(get_next))
with self.assertRaises(errors.OutOfRangeError):
sess.run(get_next)
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str,
"record_latency_TensorDataset/_1", 1)
self._assertSummaryHasCount(summary_str, "record_latency_MapDataset/_4",
1)
self._assertSummaryHasCount(summary_str,
"record_latency_PrefetchDataset/_6", 1)
def testLatencyStatsOptimization(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.from_tensors(1).apply(
optimization.assert_next([
"LatencyStats", "Map", "LatencyStats", "Prefetch",
"LatencyStats"
])).map(lambda x: x * x).prefetch(1).apply(
optimization.optimize(["latency_all_edges"])).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
get_next = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
self.assertEqual(1 * 1, sess.run(get_next))
with self.assertRaises(errors.OutOfRangeError):
sess.run(get_next)
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str,
"record_latency_TensorDataset/_1", 1)
self._assertSummaryHasCount(summary_str,
"record_latency_MapDataset/_4", 1)
self._assertSummaryHasCount(summary_str,
"record_latency_PrefetchDataset/_6", 1)
def testBytesProduced(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(100).map(
lambda x: array_ops.tile([x], ops.convert_to_tensor([x]))).apply(
stats_ops.bytes_produced_stats("bytes_produced")).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
expected_sum = 0.0
for i in range(100):
self.assertAllEqual(
np.array([i] * i, dtype=np.int64), sess.run(next_element))
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str, "bytes_produced", float(i + 1))
expected_sum += i * 8.0
self._assertSummaryHasSum(summary_str, "bytes_produced", expected_sum)
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str, "bytes_produced", 100.0)
self._assertSummaryHasSum(summary_str, "bytes_produced", expected_sum)
def testFeaturesStats(self):
num_epochs = 5
total_records = num_epochs * self._num_records
batch_size = 2
stats_aggregator = stats_ops.StatsAggregator()
dataset = self.make_batch_feature(
filenames=self.test_filenames[0],
num_epochs=num_epochs,
batch_size=batch_size,
shuffle=True,
shuffle_seed=5,
drop_final_batch=True).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
for _ in range(total_records // batch_size):
sess.run(next_element)
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasCount(sess.run(summary_t),
"record_stats:features", total_records)
self._assertSummaryHasCount(sess.run(summary_t),
"record_stats:feature-values",
total_records)
self._assertSummaryHasSum(sess.run(summary_t),
"record_stats:features",
total_records * 3)
self._assertSummaryHasSum(
sess.run(summary_t), "record_stats:feature-values",
self._sum_keywords(1) * num_epochs + 2 * total_records)
def testPrefetchBufferUtilization(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(100).map(
lambda x: array_ops.tile([x], ops.convert_to_tensor([x]))).prefetch(
-1).apply(stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.cached_session() as sess:
sess.run(iterator.initializer)
for i in range(100):
self.assertAllEqual(
np.array([i] * i, dtype=np.int64), sess.run(next_element))
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str, "Prefetch::buffer_utilization",
float(i + 1))
self._assertSummaryContains(summary_str, "Prefetch::buffer_capacity")
self._assertSummaryContains(summary_str, "Prefetch::buffer_size")
self._assertSummaryHasRange(summary_str, "Prefetch::buffer_utilization",
0, 1)
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
summary_str = sess.run(summary_t)
self._assertSummaryHasCount(summary_str, "Prefetch::buffer_utilization",
100)
def testLatencyStats(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(100).apply(
stats_ops.latency_stats("record_latency")).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
for i in range(100):
self.assertEqual(i, sess.run(next_element))
self._assertSummaryHasCount(
sess.run(summary_t), "record_latency", float(i + 1))
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasCount(sess.run(summary_t), "record_latency", 100.0)
def testLatencyStats(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(100).apply(
stats_ops.latency_stats("record_latency")).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
for i in range(100):
self.assertEqual(i, sess.run(next_element))
self._assertSummaryHasCount(
sess.run(summary_t), "record_latency", float(i + 1))
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasCount(sess.run(summary_t), "record_latency", 100.0)
def testMultipleIteratorsSameAggregator(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(100).apply(
stats_ops.latency_stats("record_latency")).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator_0 = dataset.make_initializable_iterator()
iterator_1 = dataset.make_initializable_iterator()
next_element = iterator_0.get_next() + iterator_1.get_next()
summary_t = stats_aggregator.get_summary()
with self.cached_session() as sess:
sess.run([iterator_0.initializer, iterator_1.initializer])
for i in range(100):
self.assertEqual(i * 2, sess.run(next_element))
self._assertSummaryHasCount(
sess.run(summary_t), "record_latency", float(2 * (i + 1)))
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasCount(sess.run(summary_t), "record_latency", 200.0)
def testFilteredElementsStats(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(101).filter(
lambda x: math_ops.equal(math_ops.mod(x, 3), 0)).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
for i in range(34):
self.assertEqual(i * 3, sess.run(next_element))
if i is not 0:
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::dropped_elements", float(i * 2))
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::filtered_elements", float(i + 1))
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::dropped_elements", 67.0)
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::filtered_elements", 34.0)
def testFilteredElementsStats(self):
stats_aggregator = stats_ops.StatsAggregator()
dataset = dataset_ops.Dataset.range(101).filter(
lambda x: math_ops.equal(math_ops.mod(x, 3), 0)).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
iterator = dataset.make_initializable_iterator()
next_element = iterator.get_next()
summary_t = stats_aggregator.get_summary()
with self.test_session() as sess:
sess.run(iterator.initializer)
for i in range(34):
self.assertEqual(i * 3, sess.run(next_element))
if i is not 0:
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::dropped_elements", float(i * 2))
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::filtered_elements", float(i + 1))
with self.assertRaises(errors.OutOfRangeError):
sess.run(next_element)
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::dropped_elements", 67.0)
self._assertSummaryHasScalarValue(
sess.run(summary_t), "Filter::filtered_elements", 34.0)
def _build_dataset_stats_aggregator(self):
stats_aggregator = stats_ops.StatsAggregator()
return dataset_ops.Dataset.range(10).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
def _build_dataset_stats_aggregator(self):
stats_aggregator = stats_ops.StatsAggregator()
return dataset_ops.Dataset.range(10).apply(
stats_ops.set_stats_aggregator(stats_aggregator))
