Python 将Tensorflow预处理添加到现有Keras模型中（用于Tensorflow服务）

我想在导出的Keras模型中包含我的自定义预处理逻辑，以便在Tensorflow服务中使用

我的预处理执行字符串标记化，并使用外部字典将每个标记转换为索引，以便输入到嵌入层：

模型架构和培训：

由于该模型将用于Tensorflow服务，因此我希望将所有预处理逻辑合并到模型本身（在导出的模型文件中编码）

Q：如何仅使用Keras库进行此操作？

我发现了如何结合Keras和Tensorflow的解释。但我仍然不确定如何将所有内容导出为一个模型

我知道Tensorflow有内置的字符串拆分，而且

使用Tensorflow操作的预处理逻辑：

---

Q：我如何使用这些Tensorflow预定义的预处理操作和我的Keras层来同时进行训练，然后将模型导出为“黑匣子”以用于Tensorflow服务？

我想出来了，所以我将在这里回答我自己的问题

要点如下：

import tensorflow as tf
from keras import backend as K
from keras.models import Sequential, Embedding, LSTM, Dense
from tensorflow.contrib.session_bundle import exporter
from tensorflow.contrib.lookup import HashTable, TextFileInitializer

# Initialize Keras with Tensorflow session
sess = tf.Session()
K.set_session(sess)

# Token to index lookup dictionary
token_to_idx_path = '...'
token_to_idx_dict = HashTable(TextFileInitializer(token_to_idx_path, tf.string, 0, tf.int64, 1, delimiter='\t'), 0)

maxlen = ...

# Pre-processing sub-graph using Tensorflow operations
input = tf.placeholder(tf.string, name='input')
sparse_tokenized_input = tf.string_split(input)
tokenized_input = tf.sparse_tensor_to_dense(sparse_tokenized_input, default_value='')
token_idxs = token_to_idx_dict.lookup(tokenized_input)
token_idxs_padded = tf.pad(token_idxs, [[0,0],[0,maxlen]])
token_idxs_embedding = tf.slice(token_idxs_padded, [0,0], [-1,maxlen])

# Initialize Keras model
model = Sequential()
e = Embedding(max_features, 128, input_length=maxlen)
e.set_input(token_idxs_embedding)
model.add(e)
model.add(LSTM(128, activation='sigmoid'))
model.add(Dense(num_classes, activation='softmax'))

# Load weights from previously trained Keras model
weights_path = '...'
model.load_weights(weights_path)

K.set_learning_phase(0)

# Export model in Tensorflow format
# (Official tutorial: https://github.com/tensorflow/serving/blob/master/tensorflow_serving/g3doc/serving_basic.md)
saver = tf.train.Saver(sharded=True)
model_exporter = exporter.Exporter(saver)
signature = exporter.classification_signature(input_tensor=model.input, scores_tensor=model.output)
model_exporter.init(sess.graph.as_graph_def(), default_graph_signature=signature)
model_dir = '...'
model_version = 1
model_exporter.export(model_dir, tf.constant(model_version), sess)

# Input example
with sess.as_default():
    token_to_idx_dict.init.run()
    sess.run(model.output, feed_dict={input: ["this is a raw input example"]})

首先，（在单独的代码文件中）我仅使用Keras和我自己的预处理函数训练模型，将Keras模型权重文件和我的令牌导出到索引字典

然后，我复制了Keras模型架构，将输入设置为预处理的tensor输出，从先前训练的Keras模型加载权重文件，并将其夹在Tensorflow预处理操作和Tensorflow导出器之间

最终产品：

import tensorflow as tf
from keras import backend as K
from keras.models import Sequential, Embedding, LSTM, Dense
from tensorflow.contrib.session_bundle import exporter
from tensorflow.contrib.lookup import HashTable, TextFileInitializer

# Initialize Keras with Tensorflow session
sess = tf.Session()
K.set_session(sess)

# Token to index lookup dictionary
token_to_idx_path = '...'
token_to_idx_dict = HashTable(TextFileInitializer(token_to_idx_path, tf.string, 0, tf.int64, 1, delimiter='\t'), 0)

maxlen = ...

# Pre-processing sub-graph using Tensorflow operations
input = tf.placeholder(tf.string, name='input')
sparse_tokenized_input = tf.string_split(input)
tokenized_input = tf.sparse_tensor_to_dense(sparse_tokenized_input, default_value='')
token_idxs = token_to_idx_dict.lookup(tokenized_input)
token_idxs_padded = tf.pad(token_idxs, [[0,0],[0,maxlen]])
token_idxs_embedding = tf.slice(token_idxs_padded, [0,0], [-1,maxlen])

# Initialize Keras model
model = Sequential()
e = Embedding(max_features, 128, input_length=maxlen)
e.set_input(token_idxs_embedding)
model.add(e)
model.add(LSTM(128, activation='sigmoid'))
model.add(Dense(num_classes, activation='softmax'))

# Load weights from previously trained Keras model
weights_path = '...'
model.load_weights(weights_path)

K.set_learning_phase(0)

# Export model in Tensorflow format
# (Official tutorial: https://github.com/tensorflow/serving/blob/master/tensorflow_serving/g3doc/serving_basic.md)
saver = tf.train.Saver(sharded=True)
model_exporter = exporter.Exporter(saver)
signature = exporter.classification_signature(input_tensor=model.input, scores_tensor=model.output)
model_exporter.init(sess.graph.as_graph_def(), default_graph_signature=signature)
model_dir = '...'
model_version = 1
model_exporter.export(model_dir, tf.constant(model_version), sess)

# Input example
with sess.as_default():
    token_to_idx_dict.init.run()
    sess.run(model.output, feed_dict={input: ["this is a raw input example"]})

---

公认的答案非常有用，但是它使用了@Qululu提到的过时Keras API和过时的TF服务API（导出器），并且它没有显示如何导出模型，以便其输入是原始TF占位符（相对于Keras model.input，这是后预处理）。以下是TFV1.4和Keras 2.1.2的一个版本：

sess = tf.Session()
K.set_session(sess)

K._LEARNING_PHASE = tf.constant(0)
K.set_learning_phase(0)

max_features = 5000
max_lens = 500

dict_table = tf.contrib.lookup.HashTable(tf.contrib.lookup.TextFileInitializer("vocab.txt",tf.string, 0, tf.int64, TextFileIndex.LINE_NUMBER, vocab_size=max_features, delimiter=" "), 0)

x_input = tf.placeholder(tf.string, name='x_input', shape=(None,))
sparse_tokenized_input = tf.string_split(x_input)
tokenized_input = tf.sparse_tensor_to_dense(sparse_tokenized_input, default_value='')
token_idxs = dict_table.lookup(tokenized_input)
token_idxs_padded = tf.pad(token_idxs, [[0,0],[0, max_lens]])
token_idxs_embedding = tf.slice(token_idxs_padded, [0,0], [-1, max_lens])

model = Sequential()
model.add(InputLayer(input_tensor=token_idxs_embedding, input_shape=(None, max_lens)))

 ...REST OF MODEL...

model.load_weights("model.h5")

x_info = tf.saved_model.utils.build_tensor_info(x_input)
y_info = tf.saved_model.utils.build_tensor_info(model.output)

prediction_signature = tf.saved_model.signature_def_utils.build_signature_def(inputs={"text": x_info}, outputs={"prediction":y_info}, method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME)

builder = saved_model_builder.SavedModelBuilder("/path/to/model")

legacy_init_op = tf.group(tf.tables_initializer(), name='legacy_init_op')

init_op = tf.group(tf.global_variables_initializer(), tf.local_variables_initializer())
sess.run(init_op)


# Add the meta_graph and the variables to the builder
builder.add_meta_graph_and_variables(
  sess, [tag_constants.SERVING],
  signature_def_map={
       signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
           prediction_signature,
  },
  legacy_init_op=legacy_init_op)

builder.save()  

---

更新使用Tensorflow进行推理的预处理是一项CPU操作，如果模型部署在GPU服务器上，则无法有效执行。GPU的延迟非常严重，吞吐量非常低。因此，为了在客户端过程中进行有效的预处理，我们放弃了这种方法。

找到了解决方案？@OphirYoktan请参见下面的答案。仅供参考，层方法

set_input（）

仅适用于Keras版本1.1.1。之后，它被移除了。在以后的版本中，我不知道如何将层的输入设置为Tensorflow张量。如果有人这样做，请发表评论。Hi@Qululu，在Keras 2.0+中，您现在可以使用Keras模型/层自动调用Tensorflow张量/占位符（就像您通常使用Keras层/张量等一样）。。。例如，请参阅此官方页面：。。。希望这有帮助！；）如何使用tf预处理训练模型？如何调用.fit（）以及tf占位符是如何输入的？目前似乎不支持该占位符。为了调用.fit（），我最终从模型中删除了InputLayer，并这样调用了fit（）：

model.fit（token\u idxs\u embedding.eval（session=sess，feed\u dict={x\u input:x\u train}），y\u train…

。这具体化了TF占位符。为了进行推断，我将InputLayer放回模型并保存了它。您是如何调用model.fit（）的这里？输入只是字符串列表吗？我把原始字符串放在哪里处理？@chattrat423这是一个用于导出模型进行推理的脚本，这样Tensorflow可以代替客户端进行预处理。对于培训，您可以使用仅限Keras的版本，而不需要Tensorflow预处理代码（您可以在任何python库中进行预处理）。谢谢，我真的很欣赏这种清晰。您是否也有一个后处理示例，我们可以将预测的标签索引转换为字符串标签形式？我正在尝试为tensorflow创建此示例，用于我的Kerasmodel@chattrat423我不确定我是否理解您的问题。当您将模型导出为Sav时edModel，您应该使用

tf.saved_model.signature_def_utils.build_signature_def

定义输出签名，如代码所示。这是特定于模型的，应该适合模型的输出层。客户端应该提取预测从TF Serving调用并解析它们。如果您的SavedModel已成功保存，请编写一个小python脚本来调用它并打印响应，它将向您显示需要解析的数据结构。@chattrat423我没有做过类似的操作（即使用Tensorflow进行“后处理”）.如前所述，我发现最好在客户端流程中进行所有预处理/后处理，而不是在Tensorflow中。

import tensorflow as tf
from keras import backend as K
from keras.models import Sequential, Embedding, LSTM, Dense
from tensorflow.contrib.session_bundle import exporter
from tensorflow.contrib.lookup import HashTable, TextFileInitializer

# Initialize Keras with Tensorflow session
sess = tf.Session()
K.set_session(sess)

# Token to index lookup dictionary
token_to_idx_path = '...'
token_to_idx_dict = HashTable(TextFileInitializer(token_to_idx_path, tf.string, 0, tf.int64, 1, delimiter='\t'), 0)

maxlen = ...

# Pre-processing sub-graph using Tensorflow operations
input = tf.placeholder(tf.string, name='input')
sparse_tokenized_input = tf.string_split(input)
tokenized_input = tf.sparse_tensor_to_dense(sparse_tokenized_input, default_value='')
token_idxs = token_to_idx_dict.lookup(tokenized_input)
token_idxs_padded = tf.pad(token_idxs, [[0,0],[0,maxlen]])
token_idxs_embedding = tf.slice(token_idxs_padded, [0,0], [-1,maxlen])

# Initialize Keras model
model = Sequential()
e = Embedding(max_features, 128, input_length=maxlen)
e.set_input(token_idxs_embedding)
model.add(e)
model.add(LSTM(128, activation='sigmoid'))
model.add(Dense(num_classes, activation='softmax'))

# Load weights from previously trained Keras model
weights_path = '...'
model.load_weights(weights_path)

K.set_learning_phase(0)

# Export model in Tensorflow format
# (Official tutorial: https://github.com/tensorflow/serving/blob/master/tensorflow_serving/g3doc/serving_basic.md)
saver = tf.train.Saver(sharded=True)
model_exporter = exporter.Exporter(saver)
signature = exporter.classification_signature(input_tensor=model.input, scores_tensor=model.output)
model_exporter.init(sess.graph.as_graph_def(), default_graph_signature=signature)
model_dir = '...'
model_version = 1
model_exporter.export(model_dir, tf.constant(model_version), sess)

# Input example
with sess.as_default():
    token_to_idx_dict.init.run()
    sess.run(model.output, feed_dict={input: ["this is a raw input example"]})

sess = tf.Session()
K.set_session(sess)

K._LEARNING_PHASE = tf.constant(0)
K.set_learning_phase(0)

max_features = 5000
max_lens = 500

dict_table = tf.contrib.lookup.HashTable(tf.contrib.lookup.TextFileInitializer("vocab.txt",tf.string, 0, tf.int64, TextFileIndex.LINE_NUMBER, vocab_size=max_features, delimiter=" "), 0)

x_input = tf.placeholder(tf.string, name='x_input', shape=(None,))
sparse_tokenized_input = tf.string_split(x_input)
tokenized_input = tf.sparse_tensor_to_dense(sparse_tokenized_input, default_value='')
token_idxs = dict_table.lookup(tokenized_input)
token_idxs_padded = tf.pad(token_idxs, [[0,0],[0, max_lens]])
token_idxs_embedding = tf.slice(token_idxs_padded, [0,0], [-1, max_lens])

model = Sequential()
model.add(InputLayer(input_tensor=token_idxs_embedding, input_shape=(None, max_lens)))

 ...REST OF MODEL...

model.load_weights("model.h5")

x_info = tf.saved_model.utils.build_tensor_info(x_input)
y_info = tf.saved_model.utils.build_tensor_info(model.output)

prediction_signature = tf.saved_model.signature_def_utils.build_signature_def(inputs={"text": x_info}, outputs={"prediction":y_info}, method_name=tf.saved_model.signature_constants.PREDICT_METHOD_NAME)

builder = saved_model_builder.SavedModelBuilder("/path/to/model")

legacy_init_op = tf.group(tf.tables_initializer(), name='legacy_init_op')

init_op = tf.group(tf.global_variables_initializer(), tf.local_variables_initializer())
sess.run(init_op)


# Add the meta_graph and the variables to the builder
builder.add_meta_graph_and_variables(
  sess, [tag_constants.SERVING],
  signature_def_map={
       signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
           prediction_signature,
  },
  legacy_init_op=legacy_init_op)

builder.save()