Tensorflow2.0 + conda安裝記錄

------------恢復內容開始------------

tf2.0不僅應用的動態圖，在代碼量上比1.4大大優化，安裝相比於tf1.4之類也要簡單一點，特別是有anaconda輔助的時候。

首先是經典anaconda創建環境：

conda create -n tf2.0 python=3.7

 

 

 然后是激活環境：

conda activate tf2.0

pip直接安裝tf2.0:

pip install tensorflow-gpu==2.0

 

成功安裝這些依賴包之后視為安裝完成。針對tf的gpu版本，使用anaconda可以實現多版本的tf共存。在tf2.0環境之下執行如下指令：

conda install cudnn=7.6.0
conda install cudatoolkit=10.0.130

安裝cudnn和cuda即可。

由於我平常使用pycharm，所以還需要將anaconda的環境導入pycharm中。點擊file->setting->Project:tf2.0->Project Interpreter:

 

在右上角添加新的conda existing environment:

 

最后找一個tf2.0的示例程序檢測一下是否成功運行：

from __future__ import absolute_import, division, print_function, unicode_literals

# 安裝 TensorFlow

import tensorflow as tf
mnist = tf.keras.datasets.mnist

(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(128, activation='relu'),
  tf.keras.layers.Dropout(0.2),
  tf.keras.layers.Dense(10, activation='softmax')
])

model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])

model.fit(x_train, y_train, epochs=5)

model.evaluate(x_test,  y_test, verbose=2)

...
42016/60000 [====================>.........] - ETA: 0s - loss: 0.0737 - accuracy: 0.9772
43136/60000 [====================>.........] - ETA: 0s - loss: 0.0737 - accuracy: 0.9771
44256/60000 [=====================>........] - ETA: 0s - loss: 0.0740 - accuracy: 0.9769
45312/60000 [=====================>........] - ETA: 0s - loss: 0.0742 - accuracy: 0.9767
46432/60000 [======================>.......] - ETA: 0s - loss: 0.0740 - accuracy: 0.9768
47584/60000 [======================>.......] - ETA: 0s - loss: 0.0741 - accuracy: 0.9768
48736/60000 [=======================>......] - ETA: 0s - loss: 0.0739 - accuracy: 0.9769
49824/60000 [=======================>......] - ETA: 0s - loss: 0.0736 - accuracy: 0.9770
50912/60000 [========================>.....] - ETA: 0s - loss: 0.0739 - accuracy: 0.9768
52096/60000 [=========================>....] - ETA: 0s - loss: 0.0738 - accuracy: 0.9769
53248/60000 [=========================>....] - ETA: 0s - loss: 0.0736 - accuracy: 0.9769
54272/60000 [==========================>...] - ETA: 0s - loss: 0.0735 - accuracy: 0.9769
55392/60000 [==========================>...] - ETA: 0s - loss: 0.0734 - accuracy: 0.9769
56480/60000 [===========================>..] - ETA: 0s - loss: 0.0733 - accuracy: 0.9770
57568/60000 [===========================>..] - ETA: 0s - loss: 0.0733 - accuracy: 0.9769
58656/60000 [============================>.] - ETA: 0s - loss: 0.0740 - accuracy: 0.9767
59776/60000 [============================>.] - ETA: 0s - loss: 0.0739 - accuracy: 0.9768
60000/60000 [==============================] - 3s 47us/sample - loss: 0.0740 - accuracy: 0.9768
10000/1 - 0s - loss: 0.0387 - accuracy: 0.9776