Created by: ws441572091

保存模型时，feed的list中包含输入和labels fluid.io.save_inference_model(tmp_params_dirname,['words','labels'],[crf_decode],exe) 预测的reader为 yield words，labels 数据例子：[42, 14, 127, 47, 6, 17, 79, 13, 60, 236, 16, 1111, 624, 3], [12, 6, 6, 0, 0, 0, 8, 5, 5, 5, 0, 7, 1, 0] 预测的代码如下，运行没有问题

现在希望保存模型时，feed的list中只包含输入

fluid.io.save_inference_model(tmp_params_dirname,['words'],[crf_decode],exe) 预测的reader修为yield words 数据例子：[42, 14, 127, 47, 6, 17, 79, 13, 60, 236, 16, 1111, 624, 3] 报错：The number of fields in data (14) does not match len(feed_list) (1)

在外面加了一层list之后，数据例子：[[42, 14, 127, 47, 6, 17, 79, 13, 60, 236, 16, 1111, 624, 3]] 报错：The Input(Label) should be a 2-D tensor with the 2nd dimensions fixed to 1.