Created by: taojiucheng

def merge_dict(dict, item):
    temp= {}
    list = ['category', 'item_type', 'layout', 'pos', 'cuid', 'province', 'net_type']
    for i in range(len(list)):
        if i > 3:
            temp[list[i]] = dict[list[i]]
        else:
            '''
            print dict[list[i]]
            print item[list[i]]
            '''
            temp[list[i]] = fluid.layers.sequence_concat(input=[dict[list[i]], item[list[i]]])
    return temp

函数的输入是两个字典，需要对不同的key对应的value(是一个lodtensor)进行拼接，concat的输入打印出来是这样的

<paddle.fluid.core.LoDTensor object at 0x7fc70982cd18>
<paddle.fluid.core.LoDTensor object at 0x7fc70982cc38>

输入转成numpy array后是[[1]]、[[1]] 调用op后报错如下：

File "eval_temp.py", line 410, in merge_dict
    temp[list[i]] = fluid.layers.sequence_concat(input=[dict[list[i]], item[list[i]]])
  File "/home/user/paddle_release_home/python/lib/python2.7/site-packages/paddle/fluid/layers/nn.py", line 2149, in sequence_concat
    out = helper.create_variable_for_type_inference(dtype=helper.input_dtype())
  File "/home/user/paddle_release_home/python/lib/python2.7/site-packages/paddle/fluid/layer_helper.py", line 102, in input_dtype
    inputs = self.multiple_input(input_param_name)
  File "/home/user/paddle_release_home/python/lib/python2.7/site-packages/paddle/fluid/layer_helper.py", line 64, in multiple_input
    raise type_error
TypeError: Input of sequence_concat layer should be Variable or sequence of Variable

函数的输出组织成字典的形式直接通过feed传给exe，想请问下是我使用的方法不对还是什么问题呢？