TensorFlow Profiler: モデルパフォーマンスをプロファイリングする

from datetime import datetime
from packaging import version

import os

pip install -U tensorboard_plugin_profile

import tensorflow as tf

print("TensorFlow version: ", tf.__version__)

TensorFlow version:  2.2.0-dev20200405

device_name = tf.test.gpu_device_name()
if not device_name:
  raise SystemError('GPU device not found')
print('Found GPU at: {}'.format(device_name))

Found GPU at: /device:GPU:0

import tensorflow_datasets as tfds
tfds.disable_progress_bar()

(ds_train, ds_test), ds_info = tfds.load(
    'mnist',
    split=['train', 'test'],
    shuffle_files=True,
    as_supervised=True,
    with_info=True,
)

WARNING:absl:Dataset mnist is hosted on GCS. It will automatically be downloaded to your
local data directory. If you'd instead prefer to read directly from our public
GCS bucket (recommended if you're running on GCP), you can instead set
data_dir=gs://tfds-data/datasets.
Downloading and preparing dataset mnist/3.0.0 (download: 11.06 MiB, generated: Unknown size, total: 11.06 MiB) to /root/tensorflow_datasets/mnist/3.0.0...
Dataset mnist downloaded and prepared to /root/tensorflow_datasets/mnist/3.0.0. Subsequent calls will reuse this data.

def normalize_img(image, label):
  """Normalizes images: `uint8` -> `float32`."""
  return tf.cast(image, tf.float32) / 255., label

ds_train = ds_train.map(normalize_img)
ds_train = ds_train.batch(128)

ds_test = ds_test.map(normalize_img)
ds_test = ds_test.batch(128)

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28, 1)),
  tf.keras.layers.Dense(128,activation='relu'),
  tf.keras.layers.Dense(10, activation='softmax')
])
model.compile(
    loss='sparse_categorical_crossentropy',
    optimizer=tf.keras.optimizers.Adam(0.001),
    metrics=['accuracy']
)

# Create a TensorBoard callback
logs = "logs/" + datetime.now().strftime("%Y%m%d-%H%M%S")

tboard_callback = tf.keras.callbacks.TensorBoard(log_dir = logs,
                                                 histogram_freq = 1,
                                                 profile_batch = '500,520')

model.fit(ds_train,
          epochs=2,
          validation_data=ds_test,
          callbacks = [tboard_callback])

Epoch 1/2
469/469 [==============================] - 11s 22ms/step - loss: 0.3684 - accuracy: 0.8981 - val_loss: 0.1971 - val_accuracy: 0.9436
Epoch 2/2
 50/469 [==>...........................] - ETA: 9s - loss: 0.2014 - accuracy: 0.9439WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/summary_ops_v2.py:1271: stop (from tensorflow.python.eager.profiler) is deprecated and will be removed after 2020-07-01.
Instructions for updating:
use `tf.profiler.experimental.stop` instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/summary_ops_v2.py:1271: stop (from tensorflow.python.eager.profiler) is deprecated and will be removed after 2020-07-01.
Instructions for updating:
use `tf.profiler.experimental.stop` instead.
469/469 [==============================] - 11s 24ms/step - loss: 0.1685 - accuracy: 0.9525 - val_loss: 0.1376 - val_accuracy: 0.9595
<tensorflow.python.keras.callbacks.History at 0x7f23919a6a58>

# Load the TensorBoard notebook extension.
%load_ext tensorboard

# Launch TensorBoard and navigate to the Profile tab to view performance profile
%tensorboard --logdir=logs

<IPython.core.display.Javascript object>

(ds_train, ds_test), ds_info = tfds.load(
    'mnist',
    split=['train', 'test'],
    shuffle_files=True,
    as_supervised=True,
    with_info=True,
)

ds_train = ds_train.map(normalize_img)
ds_train = ds_train.batch(128)
ds_train = ds_train.cache()
ds_train = ds_train.prefetch(tf.data.experimental.AUTOTUNE)

ds_test = ds_test.map(normalize_img)
ds_test = ds_test.batch(128)
ds_test = ds_test.cache()
ds_test = ds_test.prefetch(tf.data.experimental.AUTOTUNE)

model.fit(ds_train,
          epochs=2,
          validation_data=ds_test,
          callbacks = [tboard_callback])

Epoch 1/2
469/469 [==============================] - 10s 22ms/step - loss: 0.1194 - accuracy: 0.9658 - val_loss: 0.1116 - val_accuracy: 0.9680
Epoch 2/2
469/469 [==============================] - 1s 3ms/step - loss: 0.0918 - accuracy: 0.9740 - val_loss: 0.0979 - val_accuracy: 0.9712
<tensorflow.python.keras.callbacks.History at 0x7f23908762b0>

%tensorboard --logdir=logs

Reusing TensorBoard on port 6006 (pid 750), started 0:00:12 ago. (Use '!kill 750' to kill it.)
<IPython.core.display.Javascript object>