Created by: AoZhang

错误日志如下：

[INFO 2016-12-30 16:27:47,058 networks.py:1145] The input order is [word, label]
[INFO 2016-12-30 16:27:47,058 networks.py:1152] The output order is [word, label, __fc_layer_3__]
I1230 16:27:47.066862  7230 Trainer.cpp:149] trainer: in testing mode
I1230 16:27:47.066907  7230 Trainer.cpp:156] trainer mode: Testing
I1230 16:27:47.142230  7230 PyDataProvider2.cpp:257] loading dataprovider dataprovider::predict_process
I1230 16:27:47.143563  7230 GradientMachine.cpp:123] Loading parameters from ./output/pass-00029
F1230 16:27:54.450764  7230 Matrix.cpp:3391] Check failed: idx < height_ (1 vs. 1)
*** Check failure stack trace: ***
    @           0xec7d7c  google::LogMessage::Fail()
    @           0xec7cd4  google::LogMessage::SendToLog()
    @           0xec7769  google::LogMessage::Flush()
    @           0xeca52a  google::LogMessageFatal::~LogMessageFatal()
    @           0x811468  paddle::CpuMatrix::printOneRow()
    @           0x757deb  paddle::Tester::printOutput()
    @           0x7587b0  paddle::Tester::testOneBatch()
    @           0x7593d7  paddle::Tester::testOneBatchById()
    @           0x7596e7  paddle::Tester::testOnePass()
    @           0x759add  paddle::Tester::test()
    @           0x58ee87  main
    @     0x7ff413f34bd5  __libc_start_main
    @           0x59ad49  (unknown)

dataprovider和配置文件跟训练基本一致，预估的配置里改了几个地方：

1. batch_size改成1了
2. 输出部分 outputs([word, label, output])

预估的启动命令：

model=./output/pass-00029
paddle train \
        --config ./predictor_config.py \
        --init_model_path=$model \
        --job=test \
        --use_gpu=false\
        --config_args=is_predict=1 \
        --predict_output_dir=./output.predict