Table of Contents |
---|
Overview
Code Example
Local Profiling
This guide will show our users how to use the TensorFlow Profiler to profile the execution of your TensorFlow code.
Code example
Copy and paste the following code into tf-profile.py.
Code Block | ||
---|---|---|
| ||
from datetime import datetime
import os
import tensorflow
from tensorflow.keras.datasets import mnist
from tensorflow import keras
from tensorflow.keras import layers
(train_images, train_labels), (test_images, test_labels) = mnist.load_data()
model = keras.Sequential([
layers.Dense(512, activation="relu"),
layers.Dense(10, activation="softmax")
])
model.compile(optimizer="rmsprop",
loss="sparse_categorical_crossentropy",
metrics=["accuracy"])
train_images = train_images.reshape((60000, 28 * 28))
train_images = train_images.astype("float32") / 255
test_images = test_images.reshape((10000, 28 * 28))
test_images = test_images.astype("float32") / 255
# Create a TensorBoard callback
logs = "logs/" + datetime.now().strftime("%Y%m%d-%H%M%S")
tboard_callback = tensorflow.keras.callbacks.TensorBoard(log_dir = logs,
histogram_freq = 1,
profile_batch = '10,20')
model.fit(train_images,
train_labels,
epochs=10,
batch_size=128,
callbacks = [tboard_callback]) |
The tensorflow.keras.callbacks.TensorBoard command will create a tensorboard callback and profile_batch will pick batch number 10 to batch number 20.
Local profiling on your own computer
Run the code with command
Code Block language bash python tf-profile.py
Compress the logs folder
Code Block language bash tar -zcvf ./logs.tar.gz ./logs
Download the tarball file with sftp and/or hal-ondemand.
Decompress the tarball file
Code Block language bash tar -zxvf ./logs.tar.gz
Install the tensorboard profile plugin in your python environment.
Code Block language bash pip install tensorboard_plugin_profile
Launch the tensorboard with profiler installed.
Code Block language bash tensorboard --logdir ./logs
- Open the tensorboard dashboard in your web browser. (Google Chrome is strongly recommended)
Remote Profiling on HAL system
Coming soon...
...