[已解決] E tensorflow/stream_executor/cuda/cuda_blas.cc:510] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED

如果你的 Deep Learning Model 需要比較多的 GPU Memory 的話就會出現下面的錯誤訊息 :

E tensorflow/stream_executor/cuda/cuda_blas.cc:510] failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED



解決方法 :

請參考 : https://www.cicoria.com/keras-tensor-flow-cublas_status_not_initialized/

實際做法 :

import tensorflow as tf
from keras.backend.tensorflow_backend import set_session


config = tf.ConfigProto() 
config.gpu_options.allow_growth = True
config.log_device_placement = True
sess = tf.Session(config=config)
set_session(sess)