ValueError：Python输入与input_signature不兼容：

如何解决ValueError：Python输入与input_signature不兼容：

系统信息

• OS平台和发行版：CentOS Linux版本7.7.1908 -TensorFlow版本：2.3.0

我正在关注以下示例：https://www.tensorflow.org/tutorials/text/image_captioning?hl=en

它可以正常工作并保存检查点，我现在想将其转换为TF Lite模型。

以下是完整转换代码的链接：https://colab.research.google.com/drive/1GJkGcwWvDAWMooTsECzuSRUSPbirADhb?usp=sharing

这里是完整火车代码的链接： https://colab.research.google.com/drive/1X2d9WW1EMEzN8Rgva3rtjevP0T_jFccj?usp=sharing

我也关注isssue#32999

这是我要保存的内容，它们将转换推理图：

@tf.function
def evaluate(image):
    hidden = decoder.reset_states(batch_size=1)

    temp_input = tf.expand_dims(load_image(image)[0],0)
    img_tensor_val = image_features_extract_model(temp_input)
    img_tensor_val = tf.reshape(img_tensor_val,(img_tensor_val.shape[0],-1,img_tensor_val.shape[3]))

    features = encoder(img_tensor_val)

    dec_input = tf.expand_dims([tokenizer.word_index['<start>']],0)
    result = []

    for i in range(max_length):
        predictions,hidden,attention_weights = decoder(dec_input,features,hidden)

        predicted_id = tf.random.categorical(predictions,1)[0][0]
        # print(tokenizer.index_word)
        print(predicted_id,predicted_id.dtype)

        # for key,value in tokenizer.index_word.items():
        #     key = tf.convert_to_tensor(key)
        #     tf.dtypes.cast(key,tf.int64)
        #     print(key)

        # print(tokenizer.index_word)

        result.append(predicted_id)

        # if tokenizer.index_word[predicted_id] == '<end>':
        #     return result

        dec_input = tf.expand_dims([predicted_id],0)

    return result

export_dir = "./"
tflite_enc_input = ''
ckpt.f = evaluate
to_save = evaluate.get_concrete_function('')

converter = tf.lite.TFLiteConverter.from_concrete_functions([to_save])
tflite_model = converter.convert()

但是我得到这个错误

ValueError: in user code:

    convert2savedmodel.py:310 evaluate  *
        predictions,hidden)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/keras/engine/base_layer.py:985 __call__  **
        outputs = call_fn(inputs,*args,**kwargs)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py:780 __call__
        result = self._call(*args,**kwds)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/eager/def_function.py:840 _call
        return self._stateless_fn(*args,**kwds)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/eager/function.py:2828 __call__
        graph_function,args,kwargs = self._maybe_define_function(args,kwargs)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/eager/function.py:3171 _maybe_define_function
        *args,**kwargs)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/eager/function.py:2622 canonicalize_function_inputs
        self._flat_input_signature)
    /share/nishome/19930072_0/miniconda3/envs/tf2.3/lib/python3.7/site-packages/tensorflow/python/eager/function.py:2713 _convert_inputs_to_signature
        format_error_message(inputs,input_signature))

    ValueError: Python inputs incompatible with input_signature:
      inputs: (
        Tensor("ExpandDims_1:0",shape=(1,1),dtype=int32),Tensor("cnn__encoder/StatefulPartitionedCall:0",64,256),dtype=float32),Tensor("zeros:0",512),dtype=float32))
      input_signature: (
        TensorSpec(shape=(1,dtype=tf.int64,name=None),TensorSpec(shape=(1,dtype=tf.float32,name=None))

编码器型号：

class CNN_Encoder(tf.keras.Model):
    def __init__(self,embedding):
        super(CNN_Encoder,self).__init__()
        # shape after fc == (batch_size,embedding_dim)
        self.fc = tf.keras.layers.Dense(embedding_dim)

    @tf.function(input_signature=[tf.TensorSpec(shape=(1,features_shape),dtype=tf.dtypes.float32)])
    def call(self,x):
        x = self.fc(x)
        x = tf.nn.relu(x)
        return x

解码器型号：

class RNN_Decoder(tf.keras.Model):
    def __init__(self,embedding_dim,units,vocab_size):
        super(RNN_Decoder,self).__init__()
        self.units = units

        self.embedding = tf.keras.layers.Embedding(vocab_size,embedding_dim)
        self.gru = tf.keras.layers.GRU(self.units,return_sequences=True,return_state=True,recurrent_initializer='glorot_uniform',unroll = True)
        self.fc1 = tf.keras.layers.Dense(self.units)
        self.fc2 = tf.keras.layers.Dense(vocab_size)

        self.attention = BahdanauAttention(self.units)


    @tf.function(input_signature=[tf.TensorSpec(shape=[1,1],dtype=tf.int64),tf.TensorSpec(shape=[1,256],dtype=tf.float32),512],dtype=tf.float32)])
    def call(self,x,hidden):

        context_vector,attention_weights = self.attention(features,hidden)

        #x shape after passing through embedding == (batch_size,1,embedding_dim)
        x = self.embedding(x)

        #x shape after concatenation == (batch_size,embedding_dim + hidden_size)
        x = tf.concat([tf.expand_dims(context_vector,x],axis=-1)


        output,state = self.gru(x)

        #shape == (batch_size,max_length,hidden_size)
        x = self.fc1(output)

        #x shape == (batch_size,hidden_size)
        x = tf.reshape(x,(-1,x.shape[2]))

        # output shape == (batch_size * max_length,vocab)
        x = self.fc2(x)

        return x,state,attention_weights

    def reset_states(self,batch_size):
        return tf.zeros((batch_size,self.units))

我只是将tf.function更改为int32，如下所示：

@tf.function(input_signature=[tf.TensorSpec(shape=[1,dtype=tf.int32),dtype=tf.float32)])

但是出现另一个错误：

ValueError：Python输入与input_signature不兼容：

Tensor("ExpandDims_2:0",dtype=int64),Tensor("rnn__decoder/StatefulPartitionedCall:1",dtype=float32))
input_signature: (
TensorSpec(shape=(1,dtype=tf.int32,name=None))```
Why the dtypes of inputs change from int64 to int32?