解决 failed to allocate 4.67G (501

错误如下

failed to allocate 4.67G (5019415296 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY
failed to allocate 4.21G (4517473792 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY
failed to allocate 3.79G (4065726208 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY
failed to allocate 3.41G (3659153408 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY

原因：没有指定GPU，导致内存不足
解决办法：
第一步：需要指定GPU，代码头部添加如下代码：

import os
os.environ["CUDA_VISIBLE_DEVICES"] = "1"

第二步：限制当前脚本可用显存，代码头部添加第一行，session 语句进行如第二行的修改

gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=0.333)
sess = tf.Session(config=tf.ConfigProto(gpu_options=gpu_options))

有时执行第一步后程序也可以正常运行

参考：

1. https://blog.csdn.net/wangkun1340378/article/details/72782593
2. https://blog.csdn.net/py496980138/article/details/78843217