教程:http://blog.csdn.net/fishseeker/article/details/61918138?utm_source=tuicool&utm_medium=referral
{SPARK_HOME}/bin/spark-submit --master ${MASTER} /home/hadoop/TensorFlowOnSpark/examples/mnist/mnist_data_setup.py --output /home/hadoop/TensorFlowOnSpark/examples/mnist/csv --format csv 问题1:pyspark要求选择python版本PYSPARK_PYTHON=/usr/bin/python2.7
在/etc/pyspark中加入 PYSPARK_PYTHON=/usr/bin/python2.7
问题二:提示不能导入numpy下面的包 要在所有节点上安装numpy