训练神经网络的时候有多个GPU，如何同时调用？

坑：未填

参考：https://blog.csdn.net/Houchaoqun_XMU/article/details/79422369

 

from keras.utils import multi_gpu_model


model = inception_v4_model(img_height, img_width, channel, num_classes, dropout_keep_prob=0.2)
parallel_model = multi_gpu_model(model, gpus=2)