当升级脚本失败时，如何将tensorflow 1.0代码转换为tensorflow 2.0

Question

我有以下tensorflow 1.0代码：

import tensorflow as tf
feature_cols = tf.contrib.learn.infer_real_valued_columns_from_input(X_train)
dnn_clf = tf.contrib.learn.DNNClassifier(hidden_units=[300,100], n_classes=10,
feature_columns=feature_cols)
dnn_clf = tf.contrib.learn.SKCompat(dnn_clf)  # if TensorFlow >= 1.1
dnn_clf.fit(X_train, y_train, batch_size=50, steps=40000)

当我尝试运行转换器时：

tf_upgrade_v2 --infile mlp.py --outfile mlp2.py

我得到以下错误：

--------------------------------------------------------------------------------
File: mlp.py
--------------------------------------------------------------------------------
mlp.py:3:15: ERROR: Using member tf.contrib.learn.infer_real_valued_columns_from_input in 
deprecated module tf.contrib. tf.contrib.learn.infer_real_valued_columns_from_input cannot be converted automatically. tf.contrib will not be distributed with TensorFlow 2.0, please 
consider an alternative in non-contrib TensorFlow, a community-maintained repository such as 
tensorflow/addons, or fork the required code.
mlp.py:4:10: ERROR: Using member tf.contrib.learn.DNNClassifier in deprecated module tf.contrib. tf.contrib.learn.DNNClassifier cannot be converted automatically. tf.contrib will not be distributed with TensorFlow 2.0, please consider an alternative in non-contrib TensorFlow, a community-maintained repository such as tensorflow/addons, or fork the required code.
mlp.py:6:10: ERROR: Using member tf.contrib.learn.SKCompat in deprecated module tf.contrib. tf.contrib.learn.SKCompat cannot be converted automatically. tf.contrib will not be distributed with TensorFlow 2.0, please consider an alternative in non-contrib TensorFlow, a community-maintained repository such as tensorflow/addons, or fork the required code.

我的问题是：当升级脚本失败时，如何将tensorflow 1.0代码转换为tensorflow 2.0

Answer 1

一些Tensorflow 1.x库被移到Tensorflow/addons，而在Tensorflow 2.x中，很少有库被Tensorflow.Estimator替换。tf_upgrade_v2脚本不会将代码从Tfv1.x完全升级到Tfv2.x。此脚本将通过将现有的TensorFlow 1.x Python脚本转换为TensorFlow 2.0来加速升级过程。

在这种情况下，tf.contrib.learn被Estimator取代。

TF2.x兼容代码：替换

tf.contrib.learn.DNNLinearCombinedRegressor

带有

tf.estimator.DNNLinearCombinedRegressor

查看此链接了解更多信息