[Tensorflow] memory leak

에피소드 100번마다 model을 save 시키고 있는데, 수천 번의 episode를 실행하면 model save 시 MemoryError 발생하는 문제가 있다.

그래서 googling을 해봤다.

Memory leak when continuously run assign op

https://github.com/tensorflow/tensorflow/issues/4151

=> tf.assign을 loop 밖에서 하고, placeholder를 사용할것.

Memory Leak Running simple feed_dict graph #9091

https://github.com/tensorflow/tensorflow/issues/9091

Memory leak in zeros_like/Tile #13221

https://github.com/tensorflow/tensorflow/issues/13221

Memory leak with TensorFlow

https://stackoverflow.com/questions/44327803/memory-leak-with-tensorflow

How to get current CPU and RAM usage in Python?

https://stackoverflow.com/questions/276052/how-to-get-current-cpu-and-ram-usage-in-python

tensorflow - memory leak?

https://stackoverflow.com/questions/38714557/tensorflow-memory-leak

https://www.tensorflow.org/programmers_guide/saved_model

[참고] Saving and Restoring

https://www.tensorflow.org/programmers_guide/saved_model

[참고] TensorFlow: saving/restoring and mixing multiple models

https://blog.metaflow.fr/tensorflow-saving-restoring-and-mixing-multiple-models-c4c94d5d7125

[참고] TensorFlow: How to freeze a model and serve it with a python API

https://blog.metaflow.fr/tensorflow-how-to-freeze-a-model-and-serve-it-with-a-python-api-d4f3596b3adc