Created by: duanyuzhuo

模式：cpu本地 描述：模型训练阶段，网络构建成功，后续报错如下

Traceback (most recent call last):
  File "./thirdparty/train.py", line 147, in <module>
    dssm_trainer.train()
  File "./thirdparty/train.py", line 137, in train
    num_passes=self.model_conf.get_int("train", "PASS_NUM")
  File "/home/disk3/duanyuzhuo/dyz_file/fromTian/cluster_train_cpu_nordma/paddle/python27-gcc482/lib/python2.7/site-packages/paddle/v2/trainer.py", line 153, in train
    in_args = feeder(data_batch)
  File "paddle/python27-gcc482/lib/python2.7/site-packages/py_paddle/dataprovider_converter.py", line 282, in __call__
    return self.convert(dat, argument)
  File "paddle/python27-gcc482/lib/python2.7/site-packages/paddle/v2/data_feeder.py", line 134, in convert
    return DataProviderConverter.convert(self, reorder_data(dat), argument)
  File "paddle/python27-gcc482/lib/python2.7/site-packages/py_paddle/dataprovider_converter.py", line 267, in convert
    scanner.pre_scan(each_step)
  File "paddle/python27-gcc482/lib/python2.7/site-packages/py_paddle/dataprovider_converter.py", line 224, in pre_scan
    for each in dat:
TypeError: 'int' object is not iterable

因为报错定位到paddle脚本，通过参考其它issue（地址：https://github.com/PaddlePaddle/Paddle/issues/2062） 定位到应该是数据读取的问题，看起来是序列数据获取时并非是序列导致，排查过数据本身可以确定要feed进去的数据并非不可迭代。需要的话可以提供脚本和运行环境。 辛苦排查！