Can I use TensorBoard with Google Colab?

Question

Is there any way to use TensorBoard when training a TensorFlow model on Google Colab?

Answer 1

TensorBoard for TensorFlow running on Google Colab using tensorboardcolab. This uses ngrok internally for tunnelling.

1. Install TensorBoardColab

!pip install tensorboardcolab

2. Create a tensorboardcolab object

tbc = TensorBoardColab()

This automatically creates a TensorBoard link that can be used. This Tensorboard is reading the data at './Graph'

3. Create a FileWriter pointing to this location

summary_writer = tbc.get_writer()

tensorboardcolab library has the method that returns FileWriter object pointing to above './Graph' location.

4. Start adding summary information to Event files at './Graph' location using summary_writer object

You can add scalar info or graph or histogram data.

Reference: https://github.com/taomanwai/tensorboardcolab

Answer 2

Simple and easiest way I have found so far:

Get setup_google_colab.py file using wget

!wget https://raw.githubusercontent.com/hse-aml/intro-to- dl/master/setup_google_colab.py -O setup_google_colab.py
import setup_google_colab

To run tensorboard in background, expose port and click on the link.
I am assuming that you have proper added value to visualize in your summary and then merge all summaries.

import os
os.system("tensorboard --logdir=./logs --host 0.0.0.0 --port 6006 &")
setup_google_colab.expose_port_on_colab(6006)

After running above statements you will prompted with a link like:

Open https://a1b2c34d5.ngrok.io to access your 6006 port

Refer following git for further help:

https://github.com/MUmarAmanat/MLWithTensorflow/blob/master/colab_tensorboard.ipynb

Answer 3

There is an alternative solution but we have to use TFv2.0 preview. So if you don't have problems with the migration try this:

install tfv2.0 for GPU or CPU (TPU no available yet)

CPU
tf-nightly-2.0-preview
GPU
tf-nightly-gpu-2.0-preview

%%capture
!pip install -q tf-nightly-gpu-2.0-preview
# Load the TensorBoard notebook extension
# %load_ext tensorboard.notebook # For older versions
%load_ext tensorboard

import TensorBoard as usual:

from tensorflow.keras.callbacks import TensorBoard

Clean or Create folder where to save the logs (run this lines before run the training fit())

# Clear any logs from previous runs
import time

!rm -R ./logs/ # rf
log_dir="logs/fit/{}".format(time.strftime("%Y%m%d-%H%M%S", time.gmtime()))
tensorboard = TensorBoard(log_dir=log_dir, histogram_freq=1)

Have fun with TensorBoard! :)

%tensorboard --logdir logs/fit

Here the official colab notebook and the repo on github

New TFv2.0 alpha release:

CPU
!pip install -q tensorflow==2.0.0-alpha0 GPU
!pip install -q tensorflow-gpu==2.0.0-alpha0

Answer 4

Here is how you can display your models inline on Google Colab. Below is a very simple example that displays a placeholder:

from IPython.display import clear_output, Image, display, HTML
import tensorflow as tf
import numpy as np
from google.colab import files

def strip_consts(graph_def, max_const_size=32):
    """Strip large constant values from graph_def."""
    strip_def = tf.GraphDef()
    for n0 in graph_def.node:
        n = strip_def.node.add() 
        n.MergeFrom(n0)
        if n.op == 'Const':
            tensor = n.attr['value'].tensor
            size = len(tensor.tensor_content)
            if size > max_const_size:
                tensor.tensor_content = "<stripped %d bytes>"%size
    return strip_def

def show_graph(graph_def, max_const_size=32):
    """Visualize TensorFlow graph."""
    if hasattr(graph_def, 'as_graph_def'):
        graph_def = graph_def.as_graph_def()
    strip_def = strip_consts(graph_def, max_const_size=max_const_size)
    code = """
        <script>
          function load() {{
            document.getElementById("{id}").pbtxt = {data};
          }}
        </script>
        <link rel="import" href="https://tensorboard.appspot.com/tf-graph-basic.build.html" onload=load()>
        <div style="height:600px">
          <tf-graph-basic id="{id}"></tf-graph-basic>
        </div>
    """.format(data=repr(str(strip_def)), id='graph'+str(np.random.rand()))

    iframe = """
        <iframe seamless style="width:1200px;height:620px;border:0" srcdoc="{}"></iframe>
    """.format(code.replace('"', '&quot;'))
    display(HTML(iframe))


"""Create a sample tensor"""
sample_placeholder= tf.placeholder(dtype=tf.float32) 
"""Show it"""
graph_def = tf.get_default_graph().as_graph_def()
show_graph(graph_def)

Currently, you cannot run a Tensorboard service on Google Colab the way you run it locally. Also, you cannot export your entire log to your Drive via something like summary_writer = tf.summary.FileWriter('./logs', graph_def=sess.graph_def) so that you could then download it and look at it locally.

Answer 5

I tried but did not get the result but when used as below, got the results

import tensorboardcolab as tb
tbc = tb.TensorBoardColab()

after this open the link from the output.

import tensorflow as tf
import numpy as np

Explicitly create a Graph object

graph = tf.Graph()
with graph.as_default()

Complete example :

with tf.name_scope("variables"):
    # Variable to keep track of how many times the graph has been run
    global_step = tf.Variable(0, dtype=tf.int32, name="global_step")

    # Increments the above `global_step` Variable, should be run whenever the graph is run
    increment_step = global_step.assign_add(1)

    # Variable that keeps track of previous output value:
    previous_value = tf.Variable(0.0, dtype=tf.float32, name="previous_value")

# Primary transformation Operations
with tf.name_scope("exercise_transformation"):

    # Separate input layer
    with tf.name_scope("input"):
        # Create input placeholder- takes in a Vector 
        a = tf.placeholder(tf.float32, shape=[None], name="input_placeholder_a")

    # Separate middle layer
    with tf.name_scope("intermediate_layer"):
        b = tf.reduce_prod(a, name="product_b")
        c = tf.reduce_sum(a, name="sum_c")

    # Separate output layer
    with tf.name_scope("output"):
        d = tf.add(b, c, name="add_d")
        output = tf.subtract(d, previous_value, name="output")
        update_prev = previous_value.assign(output)

# Summary Operations
with tf.name_scope("summaries"):
    tf.summary.scalar('output', output)  # Creates summary for output node
    tf.summary.scalar('product of inputs', b, )
    tf.summary.scalar('sum of inputs', c)

# Global Variables and Operations
with tf.name_scope("global_ops"):
    # Initialization Op
    init = tf.initialize_all_variables()
    # Collect all summary Ops in graph
    merged_summaries = tf.summary.merge_all()

# Start a Session, using the explicitly created Graph
sess = tf.Session(graph=graph)

# Open a SummaryWriter to save summaries
writer = tf.summary.FileWriter('./Graph', sess.graph)

# Initialize Variables
sess.run(init)

def run_graph(input_tensor):
    """
    Helper function; runs the graph with given input tensor and saves summaries
    """
    feed_dict = {a: input_tensor}
    output, summary, step = sess.run([update_prev, merged_summaries, increment_step], feed_dict=feed_dict)
    writer.add_summary(summary, global_step=step)


# Run the graph with various inputs
run_graph([2,8])
run_graph([3,1,3,3])
run_graph([8])
run_graph([1,2,3])
run_graph([11,4])
run_graph([4,1])
run_graph([7,3,1])
run_graph([6,3])
run_graph([0,2])
run_graph([4,5,6])

# Writes the summaries to disk
writer.flush()

# Flushes the summaries to disk and closes the SummaryWriter
writer.close()

# Close the session
sess.close()

# To start TensorBoard after running this file, execute the following command:
# $ tensorboard --logdir='./improved_graph'

Answer 6

I make use of google drive's back-up and sync https://www.google.com/drive/download/backup-and-sync/. The event files, which are prediodically saved in my google drive during training, are automatically synchronised to a folder on my own computer. Let's call this folder logs. To access the visualizations in tensorboard I open the command prompt, navigate to the synchronized google drive folder, and type: tensorboard --logdir=logs.

So, by automatically syncing my drive with my computer (using back-up and sync), I can use tensorboard as if I am training on my own computer.

Edit: Here is a notebook that might be helpful. https://colab.research.google.com/gist/MartijnCa/961c5f4c774930f4bdd32d51829da6f6/tensorboard-with-google-drive-backup-and-sync.ipynb