在将KERAS模型与Bert类方法一起使用时,如何显示模型摘要?

发布于 2025-02-11 09:20:06 字数 3792 浏览 1 评论 0原文

我从本教程中的本教程)。

如其中所述,我用:

class Classifier(tf.keras.Model):
    def __init__(self, num_classes):
        super(Classifier, self).__init__(name="prediction")
        self.encoder = hub.KerasLayer(tfhub_handle_encoder, trainable=True)
        self.dropout = tf.keras.layers.Dropout(0.1)
        self.dense = tf.keras.layers.Dense(num_classes)

    def call(self, preprocessed_text):
        encoder_outputs = self.encoder(preprocessed_text)
        pooled_output = encoder_outputs["pooled_output"]
        x = self.dropout(pooled_output)
        x = self.dense(x)
        return x

我想在使用mode.summary()构建模型详细信息之后显示模型。不幸的是,这只会返回

模型:“预测”


图层(类型)输出形状param#

================================================ =================== keras_layer_1(keraslayer)多个109482241

dropout_10(辍学)多0

致密_10(密集)多个4614

================================================ =================== 总参数:109,486,855可训练的参数:109,486,854不可训练 参数:1


,但我想详细获取完整的图层。因此,我尝试添加一个函数以在模型中创建它:

def summary_model(self, sequence_length):
    in_id = tf.keras.layers.Input(shape=(sequence_length,), name="input_ids")
    in_mask = tf.keras.layers.Input(shape=(sequence_length,), name="input_masks")
    in_segment = tf.keras.layers.Input(shape=(sequence_length,), name="segment_ids")
    bert_inputs = [in_id, in_mask, in_segment]

    bert_output = self.encoder(bert_inputs)
    pooled_output = bert_output["pooled_output"]
    x = self.dropout(pooled_output)
    output = self.dense(x)

    model = tf.keras.models.Model(inputs=bert_inputs, outputs=output)
    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
    model.summary()

    return model

但是,这导致了与

valueerror:找不到匹配的具体功能来调用加载 从SavedModel。哥特:位置参数(总计3):

  • [< tf.tensor'输入:0'shape =(none,124)dtype = float32>,< tf.tensor'inputs_1:0'shape =(none,124)dtype = float = float32> tf.tensor'inputs_2:0' shape =(无,124)dtype = float32>]
  • false
  • 无关键字参数:{}

期望这些论点匹配以下4个选项之一:

选项1:位置参数(总计3):

  • {'input_mask':tensorSpec(shape =(none,none),dtype = tf.int32,name ='input_mask'),'input_type_ids':tensorspec(tensorspec(shape =(none,none,none),无),无),无),无) dtype = tf.int32,name ='input_type_ids'),'input_word_ids': tensorspec(shape =(无,无),dtype = tf.int32,name ='input_word_ids')}
  • false
  • 无关键字参数:{}

选项2:位置参数(总计3):

  • {'input_mask':tensorSpec(shape =(none,none),dtype = ttype = tf.int32,name ='inputs/input_mask'),'input_type_ids':tensorspec(tensorspec(shape =(无,,无,, 无),dtype = tf.int32,name ='inputs/input_type_ids'),), 'input_word_ids':tensorspec(shape =(无,无),dtype = tf.int32, name ='输入/input_word_ids')}
  • false
  • 无关键字参数:{}

选项3:位置参数(总计3):

  • {'input_mask':tensorSpec(shape =(none,none),dtype = ttype = tf.int32,name ='inputs/input_mask'),'input_type_ids':tensorspec(tensorspec(shape =(无,,无,, 无),dtype = tf.int32,name ='inputs/input_type_ids'),), 'input_word_ids':tensorspec(shape =(无,无),dtype = tf.int32, name ='输入/input_word_ids')}
  • true
  • 无关键字参数:{}

选项4:位置参数(总计3):

  • {'input_mask':tensorSpec(shape =(none,none),dtype = tf.int32,name ='input_mask'),'input_type_ids':tensorspec(tensorspec(shape =(none,none,none),无),无),无),无) dtype = tf.int32,name ='input_type_ids'),'input_word_ids': tensorspec(shape =(无,无),dtype = tf.int32,name ='input_word_ids')}
  • true
  • 无关键字参数:{}

我需要在输入形状或其他地方更改什么才能获得详细信息概括?

I went through this tutorial from this tutorial from tensorflow.

As described in there I defined my model with:

class Classifier(tf.keras.Model):
    def __init__(self, num_classes):
        super(Classifier, self).__init__(name="prediction")
        self.encoder = hub.KerasLayer(tfhub_handle_encoder, trainable=True)
        self.dropout = tf.keras.layers.Dropout(0.1)
        self.dense = tf.keras.layers.Dense(num_classes)

    def call(self, preprocessed_text):
        encoder_outputs = self.encoder(preprocessed_text)
        pooled_output = encoder_outputs["pooled_output"]
        x = self.dropout(pooled_output)
        x = self.dense(x)
        return x

I would like to display the model details after building it with mode.summary(). Unfortunately this only returns

Model: "prediction"


Layer (type) Output Shape Param #

=================================================================
keras_layer_1 (KerasLayer) multiple 109482241

dropout_10 (Dropout) multiple 0

dense_10 (Dense) multiple 4614

=================================================================
Total params: 109,486,855 Trainable params: 109,486,854 Non-trainable
params: 1


But I would like to get the full layers in detail. Therefore I tried to add a function to create it in the model:

def summary_model(self, sequence_length):
    in_id = tf.keras.layers.Input(shape=(sequence_length,), name="input_ids")
    in_mask = tf.keras.layers.Input(shape=(sequence_length,), name="input_masks")
    in_segment = tf.keras.layers.Input(shape=(sequence_length,), name="segment_ids")
    bert_inputs = [in_id, in_mask, in_segment]

    bert_output = self.encoder(bert_inputs)
    pooled_output = bert_output["pooled_output"]
    x = self.dropout(pooled_output)
    output = self.dense(x)

    model = tf.keras.models.Model(inputs=bert_inputs, outputs=output)
    model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])
    model.summary()

    return model

But this leads to an error with

ValueError: Could not find matching concrete function to call loaded
from the SavedModel. Got: Positional arguments (3 total):

  • [<tf.Tensor 'inputs:0' shape=(None, 124) dtype=float32>, <tf.Tensor 'inputs_1:0' shape=(None, 124) dtype=float32>, <tf.Tensor 'inputs_2:0'
    shape=(None, 124) dtype=float32>]
  • False
  • None Keyword arguments: {}

Expected these arguments to match one of the following 4 option(s):

Option 1: Positional arguments (3 total):

  • {'input_mask': TensorSpec(shape=(None, None), dtype=tf.int32, name='input_mask'), 'input_type_ids': TensorSpec(shape=(None, None),
    dtype=tf.int32, name='input_type_ids'), 'input_word_ids':
    TensorSpec(shape=(None, None), dtype=tf.int32, name='input_word_ids')}
  • False
  • None Keyword arguments: {}

Option 2: Positional arguments (3 total):

  • {'input_mask': TensorSpec(shape=(None, None), dtype=tf.int32, name='inputs/input_mask'), 'input_type_ids': TensorSpec(shape=(None,
    None), dtype=tf.int32, name='inputs/input_type_ids'),
    'input_word_ids': TensorSpec(shape=(None, None), dtype=tf.int32,
    name='inputs/input_word_ids')}
  • False
  • None Keyword arguments: {}

Option 3: Positional arguments (3 total):

  • {'input_mask': TensorSpec(shape=(None, None), dtype=tf.int32, name='inputs/input_mask'), 'input_type_ids': TensorSpec(shape=(None,
    None), dtype=tf.int32, name='inputs/input_type_ids'),
    'input_word_ids': TensorSpec(shape=(None, None), dtype=tf.int32,
    name='inputs/input_word_ids')}
  • True
  • None Keyword arguments: {}

Option 4: Positional arguments (3 total):

  • {'input_mask': TensorSpec(shape=(None, None), dtype=tf.int32, name='input_mask'), 'input_type_ids': TensorSpec(shape=(None, None),
    dtype=tf.int32, name='input_type_ids'), 'input_word_ids':
    TensorSpec(shape=(None, None), dtype=tf.int32, name='input_word_ids')}
  • True
  • None Keyword arguments: {}

What do I need to change in the input shape or somewhere else to be able to get the detailed summary?

如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。

扫码二维码加入Web技术交流群

发布评论

需要 登录 才能够评论, 你可以免费 注册 一个本站的账号。
列表为空,暂无数据
我们使用 Cookies 和其他技术来定制您的体验包括您的登录状态等。通过阅读我们的 隐私政策 了解更多相关信息。 单击 接受 或继续使用网站,即表示您同意使用 Cookies 和您的相关数据。
原文