autokeras Building Automodel model fails when using two ImageInput()s

Bug Description

Building Automodel multimodal classification model fails with two ImageInputs. I get the following error ~5 times ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique. ending in a RuntimeError: Too many failed attempts to build model.

Are two ImageInputs() not yet supported?

Bug Reproduction

Code for reproducing the bug:

import autokeras as ak
from tensorflow import keras
import helpers as utils
import numpy as np

metrics = ['accuracy', keras.metrics.TopKCategoricalAccuracy(k=5)]

clf = ak.AutoModel(
        inputs=[ak.ImageInput(name='img1'), ak.ImageInput(name='img2')],
        outputs=ak.ClassificationHead(loss='binary_crossentropy', metrics=metrics),
        objective='val_accuracy',
        project_name="ears",
        tuner="hyperband"
)

at = AutoImageClassificationTrain('./Dataset/')
X, Y = at.create_x_and_y()

(pairTrain, labelTrain) = utils.make_pairs(X, Y)
# (pairVal, labelVal) = utils.make_pairs(valX, valY)


clf.fit(x=[pairTrain[:, 0], pairTrain[:, 1]], y=labelTrain[:], epochs=50, batch_size=32, validation_split = 0.2, metrics=top_ks)
model = clf.export_model()
try:
    model.save("ear_discriminator", save_format="tf")
except Exception:
    model.save("ear_discriminator.h5")

print('\n\n\n[INFO] Completed succesfully!')

Data used by the code:

Two input RGB images as described.

Expected Behavior

Setup Details

Include the details about the versions of:

OS type and version: Ubuntu 18.04.5 LTS
Python: 3.8.5
autokeras: 1.0.12
keras-tuner: 1.0.2
scikit-learn: 0.23.2
numpy: 1.19.2
pandas: 1.1.3
tensorflow: 2.4.1

Additional context

Autokeras worked fine 3-4 days ago for a simple ImageClassification task. But, this seems to be something within Autokeras instantiating two ResNet-50s while initiating search?

Below is the error trace

Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.
Invalid model 0/5
Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.
Invalid model 1/5
Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.
Invalid model 2/5
Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.
Invalid model 3/5
Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.
Invalid model 4/5
Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.
Invalid model 5/5
Traceback (most recent call last):
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 104, in build
    model = self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/graph.py", line 254, in build
    model = tf.keras.Model(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 120, in __init__
    self._init_graph_network(inputs, outputs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/training/tracking/base.py", line 517, in _method_wrapper
    result = method(self, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 203, in _init_graph_network
    nodes, nodes_by_depth, layers, _ = _map_graph_network(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/tensorflow/python/keras/engine/functional.py", line 1000, in _map_graph_network
    raise ValueError('The name "' + name + '" is used ' +
ValueError: The name "resnet50" is used 2 times in the model. All layer names should be unique.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "auto_modeler.py", line 25, in <module>
    clf.fit(x=[pairTrain[:, 0], pairTrain[:, 1]], y=labelTrain[:], epochs=50, batch_size=32, validation_split = 0.2, metrics=top_ks)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/auto_model.py", line 273, in fit
    self.tuner.search(
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/autokeras/engine/tuner.py", line 178, in search
    self.hypermodel.build(hp)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 64, in _build_wrapper
    return self._build(hp, *args, **kwargs)
  File "/home/saurav/anaconda3/lib/python3.8/site-packages/kerastuner/engine/hypermodel.py", line 112, in build
    raise RuntimeError(
RuntimeError: Too many failed attempts to build model.

Apr 22 '21 02:04 Saurav-K-Aryal

Any updates on this issue? I'm having the same problem with the following setup:

OS type and version: Ubuntu 18.04.5 LTS
Python: 3.7.10
autokeras: 1.0.14
keras-tuner: 1.0.2
scikit-learn: 0.24.2
numpy: 1.18.5
pandas: 1.2.4
tensorflow: 2.3.0

Jun 15 '21 04:06 janTOffermann

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.