Error when Converting a TensorFlow.js Layers model (JSON + binary weight file(s)) to a Keras HDF5 model file.

To get help from the community, we encourage using Stack Overflow and the tensorflow.js tag.

TensorFlow.js version

1.0.0

Browser version

Chrome 72.0.3626.121 (Official Build) (64-bit)

Describe the problem or feature request

I imported tfjs by a script in HTML: <script src="https://cdn.jsdelivr.net/npm/@tensorflow/tfjs@1.0.0/dist/tf.min.js"></script> I trained a MNIST model and save it as a JSON file and a bin file. But Error is raised when using tensorflowjs_converter to convert the model into Keras h5 model. I get:

I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2019-03-21 11:28:50.356788: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2592000000 Hz
2019-03-21 11:28:50.356952: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x5545eb0 executing computations on platform Host. Devices:
2019-03-21 11:28:50.356967: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): <undefined>, <undefined>
Traceback (most recent call last):
  File "/home/bigdog/.local/bin/tensorflowjs_converter", line 11, in <module>
    sys.exit(main())
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflowjs/converters/converter.py", line 368, in main
    FLAGS.output_path)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflowjs/converters/converter.py", line 169, in dispatch_tensorflowjs_to_keras_h5_conversion
    model = keras_tfjs_loader.load_keras_model(config_json_path)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflowjs/converters/keras_tfjs_loader.py", line 218, in load_keras_model
    use_unique_name_scope=use_unique_name_scope)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflowjs/converters/keras_tfjs_loader.py", line 65, in _deserialize_keras_model
    model = keras.models.model_from_json(json.dumps(model_topology_json))
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/saving/model_config.py", line 96, in model_from_json
    return deserialize(config, custom_objects=custom_objects)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/layers/serialization.py", line 88, in deserialize
    printable_module_name='layer')
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/utils/generic_utils.py", line 192, in deserialize_keras_object
    list(custom_objects.items())))
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/engine/sequential.py", line 351, in from_config
    custom_objects=custom_objects)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/layers/serialization.py", line 88, in deserialize
    printable_module_name='layer')
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/utils/generic_utils.py", line 194, in deserialize_keras_object
    return cls.from_config(cls_config)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/engine/base_layer.py", line 437, in from_config
    return cls(**config)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/layers/convolutional.py", line 477, in __init__
    kernel_initializer=initializers.get(kernel_initializer),
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/initializers.py", line 194, in get
    return deserialize(identifier)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/initializers.py", line 186, in deserialize
    printable_module_name='initializer')
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/keras/utils/generic_utils.py", line 194, in deserialize_keras_object
    return cls.from_config(cls_config)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/ops/init_ops_v2.py", line 89, in from_config
    return cls(**config)
  File "/home/bigdog/.local/lib/python3.5/site-packages/tensorflow/python/ops/init_ops_v2.py", line 390, in __init__
    raise ValueError("Invalid `mode` argument:", mode)
ValueError: ('Invalid `mode` argument:', None)

Code to reproduce the bug / link to feature request

tensorflowjs_converter --input_format tfjs_layers_model --output_format keras mnist-cnn.json result/

tfjs_layers_model file attached for reference

browser_trained.zip

[caisq]

Thanks for the additional information!

There are a number of things going on here.

1. TensorFlow.js doesn’t populate the “mode” field of VarianceScaling initializer with the default “fan_in” value properly. Instead, it leaves it as “null”.
2. The latest version of tf-2.0-nightly-preview doesn’t handle distribution: normal properly. For that I’ve filed b/129177721 to notify the keras team. cc @yhliang @fchollet @karmel

@1egit The current work around is as follows: install and use tensorflowjs version 0.8.5: `pip install tensorflowjs==0.8.5". This should install the latest stable release of tensorflow (1.13.1), which doesn’t have this issue.

Let me know if you have more questions.

[eightmillimeter]

Hi all,

I’m facing the similar issue:

Traceback (most recent call last): File “/home/burak/.local/bin/tensorflowjs_converter”, line 11, in <module> sys.exit(main()) File “/home/burak/.local/lib/python2.7/site-packages/tensorflowjs/converters/converter.py”, line 352, in main strip_debug_ops=FLAGS.strip_debug_ops) File “/home/burak/.local/lib/python2.7/site-packages/tensorflowjs/converters/tf_saved_model_conversion_pb.py”, line 329, in convert_tf_frozen_model graph = load_graph(frozen_model_path, output_node_names) File “/home/burak/.local/lib/python2.7/site-packages/tensorflowjs/converters/tf_saved_model_conversion_pb.py”, line 63, in load_graph graph_def.ParseFromString(f.read()) File “/home/burak/.local/lib/python2.7/site-packages/tensorflow/python/lib/io/file_io.py”, line 125, in read self._preread_check() File “/home/burak/.local/lib/python2.7/site-packages/tensorflow/python/lib/io/file_io.py”, line 85, in _preread_check compat.as_bytes(self.__name), 1024 * 512, status) File “/home/burak/.local/lib/python2.7/site-packages/tensorflow/python/framework/errors_impl.py”, line 528, in exit c_api.TF_GetCode(self.status.status)) tensorflow.python.framework.errors_impl.NotFoundError: output/saved_model.pb; No such file or directory

I couldn’t solve it. Can anyone help?

PS: First time asking question here, sorry if I forgetting something.