Python 如何在TF2中命名ops和tensor创建以帮助记录/调试？

我正在使用Tensorflow 2.0a和

tf.debug.set\u log\u device\u placement（True）

来记录每个操作的位置。但有些行动相当神秘。。对我来说，了解这些操作对应的是什么很重要，因为它们在CPU上运行，可能会减慢我的训练速度

特别是：

VarIsInitializedOp、AssignVariableOp、ReadVariableOp

这些操作是否有一个清晰的映射到我可以在代码中标记的特定tf调用

Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:CPU:0
2019-05-30 16:20:56.082274: I tensorflow/core/common_runtime/eager/execute.cc:394] Executing op VarHandleOp in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op VarIsInitializedOp in device /job:localhost/replica:0/task:0/device:CPU:0
2019-05-30 16:20:56.082418: I tensorflow/core/common_runtime/eager/execute.cc:394] Executing op VarIsInitializedOp in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op AssignVariableOp in device /job:localhost/replica:0/task:0/device:CPU:0
2019-05-30 16:20:56.084034: I tensorflow/core/common_runtime/eager/execute.cc:394] Executing op AssignVariableOp in device /job:localhost/replica:0/task:0/device:CPU:0
Executing op ReadVariableOp in device /job:localhost/replica:0/task:0/device:CPU:0
2019-05-30 16:20:56.258670: I tensorflow/core/common_runtime/eager/execute.cc:394] Executing op ReadVariableOp in device /job:localhost/replica:0/task:0/device:CPU:0