OOM when allocating tensor with shape[300,14,14,256] and type float on /job:localhost/replica:0/tas

訓練mask-R-cnn模型，訓練時將model.train(dataset_train, dataset_val,learning_rate=config.LEARNING_RATE,epochs=40,layers='heads')中的layers=‘heads’修改成layers='4+'，結果報錯 OOM when allocating tensor with shape[300,256,

>>阅读原文<<