偶然发现当在jupyter notebook中执行如下代码时会产生些错乱的结果
import tensorflow as tf
import os
MODEL_DIR = 'C:/Users/pradmin/Desktop/迁移学习相关/inception_dec_2015/'
MODEL_FILE = 'scr.pb'
MODEL_FILE = 'tensorflow_inception_graph.pb'
with tf.gfile.FastGFile(MODEL_DIR+MODEL_FILE, 'rb') as f:
graph_def = tf.GraphDef()
graph_def.ParseFromString(f.read())
tf.import_graph_def(graph_def,name='')
with tf.Session() as sess:
print(os.path.join(MODEL_DIR, MODEL_FILE))
#softmax_tensor = sess.graph.get_tensor_by_name('final_training_ops/y:0')
softmax_tensor = sess.graph.get_tensor_by_name('pool_3/_reshape:0')
print(softmax_tensor)
在两个pb文件中tensor是不一样的,scr.pb中是由张量'final_training_ops/y:0' 的,而没有'pool_3/_reshape:0'
tensorflow_inception_graph.pb则相反,但是在两个pb文件都执行一遍的情况下,原本没有的张量却都可以打印出来,
尝试在Linux系统下的shell环境下运行,则为出现该种情况,这可能是jupyter notebook的一个bug,有知道原因的大神可以告知下。