无法使用 save/pickle 保存 tensorrflow keras 量子模型

Question

如何保存tensorflow量子模型？当我试图用量子电路保存 keras 模型时，我得到以下信息。我没有找到任何支持。 tensorflow:Layer add_circuit_2 传递了不可序列化的关键字参数 tensorflow:Layer add_circuit_2 传递了不可序列化的关键字参数 tensorflow:Layer add_circuit_2 传递了不可序列化的关键字参数

    WARNING:tensorflow:Layer add_circuit_2 was passed non-serializable keyword arguments: {'prepend': 
    cirq.Circuit([
    cirq.Moment(operations=[
    cirq.H.on(cirq.GridQubit(0, 0)),
    cirq.H.on(cirq.GridQubit(0, 1)),
    cirq.H.on(cirq.GridQubit(0, 2)),
    cirq.H.on(cirq.GridQubit(0, 3)),
    cirq.H.on(cirq.GridQubit(1, 0)),
    cirq.H.on(cirq.GridQubit(1, 1)),
    cirq.H.on(cirq.GridQubit(1, 2)),
    cirq.H.on(cirq.GridQubit(1, 3)),
    cirq.H.on(cirq.GridQubit(2, 0)),
    cirq.H.on(cirq.GridQubit(2, 1)),
    cirq.H.on(cirq.GridQubit(2, 2)),
    cirq.H.on(cirq.GridQubit(2, 3)),
    cirq.H.on(cirq.GridQubit(3, 0)),
    cirq.H.on(cirq.GridQubit(3, 1)),
    cirq.H.on(cirq.GridQubit(3, 2)),
    cirq.H.on(cirq.GridQubit(3, 3)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(0, 0), cirq.GridQubit(0, 1)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(0, 1), cirq.GridQubit(0, 2)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(0, 2), cirq.GridQubit(0, 3)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(0, 3), cirq.GridQubit(1, 0)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(1, 0), cirq.GridQubit(1, 1)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(1, 1), cirq.GridQubit(1, 2)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(1, 2), cirq.GridQubit(1, 3)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(1, 3), cirq.GridQubit(2, 0)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(2, 0), cirq.GridQubit(2, 1)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(2, 1), cirq.GridQubit(2, 2)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(2, 2), cirq.GridQubit(2, 3)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(2, 3), cirq.GridQubit(3, 0)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(3, 0), cirq.GridQubit(3, 1)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(3, 1), cirq.GridQubit(3, 2)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(3, 2), cirq.GridQubit(3, 3)),
]),
cirq.Moment(operations=[
    cirq.CZ.on(cirq.GridQubit(3, 3), cirq.GridQubit(0, 0)),
]),])}. 
They will not be included in the serialized model (and thus will be missing at deserialization time).
---------------------------------------------------------------------------
NotImplementedError                       Traceback (most recent call last)
<ipython-input-91-a74ee5c9d34d> in <module>()
----> 1 qcnn_model.save('qcnn_model.h5')

8 frames
/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/engine/base_layer.py in 
get_config(self)
497     # or that `get_config` has been overridden:
498     if len(extra_args) > 1 and hasattr(self.get_config, '_is_default'):
--> 499       raise NotImplementedError('Layers with arguments in `__init__` must '
500                                 'override `get_config`.')
501     return config

NotImplementedError: Layers with arguments in `__init__` must override `get_config`.

Answer 1

TensorFlow Quantum 尚未实现 get_config 和 load_config。我们在保存某些 Cirq 对象时遇到了一些困难，我们正在解决这个问题。

现在，如果您想保存包含量子层的模型，您可以使用

model = tf.keras.Model(...)
model.save_weights("some_path")
...
model.load_weights("some_path")

改为函数。

无法使用 save/pickle 保存 tensorrflow keras 量子模型

cannot save tensorrflow keras quantum model using save/pickle

model

save

keras

tensorflow

tensorflow-quantum