Created by: Emma-Ding

之前使用的v1版本，model还用了tensor_layer，使用如下，现在迁到fluid上，想问问如何实现类似的操作

`user_emb1 = fc_layer(name="user_emb1", input=concat_layer(input=user_emb_vec), size=256, act=ReluActivation()) user_emb2 = fc_layer(name="user_emb2", input=user_emb1, size=128, act=ReluActivation()) user_emb3 = fc_layer(name="user_emb3", input=user_emb2, size=64, act=ReluActivation()) user_emb4 = fc_layer(name="user_emb4", input=user_emb3, size=32) user_emb5 = fc_layer(name="user_emb5", input=user_emb4, size=16)

user_cross1 = fc_layer(name="user_cross1", input=concat_layer(input=user_emb_vec), size=128, act=LinearActivation())
user_cross2_raw = tensor_layer(name="user_cross2_raw", a=user_cross1, b=user_cross1, size=128, act=LinearActivation())
user_cross2 = user_cross2_raw + user_cross1

user_cross3_raw = tensor_layer(name="user_cross3_raw", a=user_cross2, b=user_cross1, size=128, act=LinearActivation())
user_cross3 = user_cross3_raw + user_cross2`