CODEWITHSUNDEEP
pythontensorflowkeras
ibrahim sharaby
ibrahim sharaby

Reputation: 21

Error while creating a log file using TensorBoard with Keras

I'm getting an error while creating a log file using TensorBoard with Keras.

The code

import pandas as pd
from keras.callbacks import TensorBoard
from keras.models import Sequential
from keras.layers import *

training_data_df = pd.read_csv("sales_data_training_scaled.csv")

X = training_data_df.drop('total_earnings', axis=1).values
Y = training_data_df[['total_earnings']].values

# Define the model
model = Sequential()
model.add(Dense(50, input_dim=9, activation='relu', name='layer_1'))
model.add(Dense(100, activation='relu', name='layer_2'))
model.add(Dense(50, activation='relu', name='layer_3'))
model.add(Dense(1, activation='linear', name='output_layer'))
model.compile(loss='mean_squared_error', optimizer='adam')

# Create a TensorBoard logger
logger = TensorBoard(
    log_dir='logs',
    histogram_freq=5,
    write_graph=True
)

# Train the model
model.fit(
    X,
    Y,
    epochs=50,
    shuffle=True,
    verbose=2,
    callbacks=[logger]
)

# Load the separate test data set

test_data_df = pd.read_csv("sales_data_test_scaled.csv")
X_test = test_data_df.drop('total_earnings', axis=1).values
Y_test = test_data_df[['total_earnings']].values
test_error_rate = model.evaluate(X_test, Y_test, verbose=0)
print(test_error_rate)

Then I got this error:

Traceback (most recent call last):

File "E:/Building.Deep.Learning.Applications.with.Keras.2.0/Exercise Files/06/model_logging final.py", line 34, in callbacks=[logger]

File "C:\Python3.6.4\lib\site-packages\keras\engine\training.py", line 1041, in fit steps_per_epoch=steps_per_epoch)

File "C:\Python3.6.4\lib\site-packages\keras\engine\training_arrays.py", line 219, in fit_loop callbacks.on_epoch_end(epoch, epoch_logs)

File "C:\Python3.6.4\lib\site-packages\keras\callbacks.py", line 77, in on_epoch_end callback.on_epoch_end(epoch, logs)

File "C:\Python3.6.4\lib\site-packages\keras\callbacks.py", line 865, in on_epoch_end

raise ValueError("If printing histograms, validation_data must be " ValueError: If printing histograms, validation_data must be provided, and cannot be a generator.

Upvotes: 2

Views: 2235

Answers (3)

Shimon Doodkin
Shimon Doodkin

Reputation: 4589

one solution is to turn off histograms with histogram_freq=0 :

logger = TensorBoard( log_dir='logs', histogram_freq=0, write_graph=True )

another solution is to specify not generator validation data

another solution is to create a wrapper for tensor board like here: https://github.com/keras-team/keras/issues/3358

see answer here: Keras autoencoder with Tensorflow Dataset API and logging to Tensorboard

Upvotes: 0

Shiva
Shiva

Reputation: 651

First load X_test, Y_test data and use them with validation_data arg in model.fit. Working code below.

import pandas as pd
from keras.callbacks import TensorBoard
from keras.models import Sequential
from keras.layers import *

training_data_df = pd.read_csv("sales_data_training_scaled.csv")

X = training_data_df.drop('total_earnings', axis=1).values
Y = training_data_df[['total_earnings']].values

# Define the model
model = Sequential()
model.add(Dense(50, input_dim=9, activation='relu', name='layer_1'))
model.add(Dense(100, activation='relu', name='layer_2'))
model.add(Dense(50, activation='relu', name='layer_3'))
model.add(Dense(1, activation='linear', name='output_layer'))
model.compile(loss='mean_squared_error', optimizer='adam')

# Create a TensorBoard logger
logger = TensorBoard(
    log_dir='logs',
    histogram_freq=5,
    write_graph=True
)

# Load the separate test data set.
# >>> Setup X_test, Y_test before using in model.fit below. <<<
test_data_df = pd.read_csv("sales_data_test_scaled.csv")
X_test = test_data_df.drop('total_earnings', axis=1).values
Y_test = test_data_df[['total_earnings']].values

# Train the model
model.fit(
    X,
    Y,
    epochs=50,
    shuffle=True,
    verbose=2,
    callbacks=[logger],
    validation_data=(X_test, Y_test)   # <<< Add this.
)

# Evaluate
test_error_rate = model.evaluate(X_test, Y_test, verbose=0)
print(test_error_rate)

Upvotes: 1

VegardKT
VegardKT

Reputation: 1246

Move your validation into your .fit function like so:

# Train the model
model.fit(
X,
Y,
epochs=50,
shuffle=True,
verbose=2,
validation_data=(X_test, Y_test),
callbacks=[logger]
)

When you do it after the .fit function like you are, the logger cannot see the validation data.

You can also set histogram_freq=0 if that doesnt work. Your histogram will not work then though.

Upvotes: 4

Related Questions