We need more demo code

[Yancey1989]

PaddleCloud needs more demo training code, and we appreciate your contribution!!

You can make the code from book running on the Cloud as the following steps:

1. Configure the PaddleCloud Client as /~https://github.com/PaddlePaddle/cloud/blob/develop/doc/usage_cn.md
2. Use the clod dataset instead of the local dataset, here is a demo for preparing dataset with recordio files.
3. Submit your training job with normal mode:
   • Develop a reader for recordio files, like: /~https://github.com/PaddlePaddle/cloud/blob/develop/demo/fit_a_line/train.py#L16
   • Submit your job as the link

• Submit your training job with fault tolerant:
  • Use paddle.v2.reader.creator.cloud_reader to fetch training data from master, the demo code as:
    /~https://github.com/PaddlePaddle/cloud/blob/develop/demo/fit_a_line/train_ft.py
  • Submit your job with flag -faulttolerant

If you are interested in any one in the following, append your name at the end of the demo name

• 01.fit_a_line --- @Yancey1989
• 02.recognize_digits--- @chrisxu2016 @wanghaoshuang
• 03.image_classification --- @gongweibao
• 04.word2vec ---@typhoonzero
• 05.recommender_system --- @helinwang
• 06.understand_sentiment --- @wanghaoshuang
• 07.label_semantic_roles --- @helinwang
• 08.machine_translation --- @helinwang

[gongweibao]

03.image_classification --- @helinwang
这个给我吧，我正在帮医疗图像改造训练的流程，正好要做这个。

[helinwang]

@gongweibao 好的！