YY_pdd: spark的这个persist算子设计违背了函数式设计原则。一个函数即有返回值，又改变的RDD的存储级别。这是一个不纯粹的函数。我被坑了

小雷震天: 很有帮助，感谢分享

Jetuser-data: 这个图片迁移深度学习代码十个跑九个报错

Jetuser-data: Instructions for updating: If using Keras pass *_constraint arguments to layers. Traceback (most recent call last): File "neural_style.py", line 205, in main() File "neural_style.py", line 176, in main checkpoint_iterations=options.checkpoint_iterations File "C:\Users\Administrator\stylize.py", line 117, in stylize _, height, width, number = map(lambda j: j.value, layer.get_shape()) # 117 File "C:\Users\Administrator\stylize.py", line 117, in _, height, width, number = map(lambda j: j.value, layer.get_shape()) # 117 AttributeError: 'int' object has no attribute 'value'

weixin_43919751 回复 夏天的天: 把global _vgg_params改成_vgg_params=None