Autokeras: Autokeras 1.0.3 not running in google Colab

[oscaralvaro]

Bug Description

When trying to reproduce MNIST example using the latest master-branch version (which solves issue #1095), the fit method get's stuck and never finishes. Below the content of notebook I used.

Bug Reproduction

Google colab notebook content:

%tensorflow_version 2.x
import tensorflow as tf
print(tf.__version__) # prints 2.2.0-rc4

!pip install  -v "/content/drive/My Drive/autokeras-master.tar.gz"
.
.
.
.
#Successfully installed autokeras-1.0.3 colorama-0.4.3 keras-tuner-1.0.1 terminaltables-3.1.0
#Cleaning up...
#Removed build tracker '/tmp/pip-req-tracker-c06gu3n1'

tf.debugging.set_log_device_placement(True)

import autokeras as ak
from tensorflow.keras.datasets import mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
print(x_train.shape)  # (60000, 28, 28)
print(y_train.shape)  # (60000,)
print(y_train[:3])  # array([7, 2, 1], dtype=uint8)

clf = ak.ImageClassifier(max_trials=3)

clf.fit(x_train, y_train, epochs=10, verbose=2)

Output:
Executing op TensorSliceDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op TensorSliceDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op ZipDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op OptimizeDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op ModelDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op ReduceDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op TakeDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op SkipDataset in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op BatchDatasetV2 in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op BatchDatasetV2 in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op Fill in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarIsInitializedOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op LogicalNot in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op Assert in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op AssignVariableOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op Fill in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op VarIsInitializedOp in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op AssignVariableOp in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op RandomUniform in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op Sub in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op Mul in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op Add in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op DestroyResourceOp in device /job:localhost/replica:0/task:0/device:GPU:0
Executing op DestroyResourceOp in device /job:localhost/replica:0/task:0/device:CPU:0

then the cell just hangs forever

Setup Details

Include the details about the versions of:

• OS type and version: Google colab
• Python: 3.6.9
• autokeras: master
• keras-tuner: keras-tuner-1.0.1
• scikit-learn: 0.22.2.post1
• numpy: 1.18.3
• pandas: 1.0.3
• tensorflow: 2.2.0-rc4

[oscaralvaro]

When using a prior version of both tensorflow (2.1.0) an autokeras (1.0.2) this error doesn't happen

[ruti1012]

I'm having the same issue using autokeras 1.0.3 (latest commit) on colab with the ImageClassifier.

[qingquansong]

When using a prior version of both tensorflow (2.1.0) an autokeras (1.0.2) this error doesn't happen

It might because of the modification of kerastuner. Please try to install a fairly latest version of kerastuner (note: the __version__ attribute of this version may not be correct):

KerasTuner: pip install git+git://github.com/keras-team/keras-tuner.git@d2d69cba21a0b482a85ce2a38893e2322e139c01

TensorFlow: 2.2.0 (or the default 2.2.0rc4 on current Google colab)
AutoKeras: 1.0.3 master branch

Thanks.

[oscaralvaro]

Thank you QQ for the tip. For now I will stick to TF 2.1.0 and AK 1.0.2, as is working so far.

I wanted to reported anyway so you guys are aware of the issue.

[ruti1012]

Using the latest version of KerasTuner fixed the problem, thanks!

[cibic89]

this happens to me too on Google Colab and on local machine using this Keras Tuner:
KerasTuner: pip install git+git://github.com/keras-team/keras-tuner.git@d2d69cba21a0b482a85ce2a38893e2322e139c01 --no-dependencies

versions:
TensorFlow: default 2.2.0rc4 on current Google colab
AutoKeras: 1.0.3 master branch

[haifeng-jin]

We just release the real 1.0.3. You may try this https://colab.research.google.com/github/keras-team/autokeras/blob/master/docs/ipynb/image_classification.ipynb