对话情绪识别
对话情绪识别（Emotion Detection，简称EmoTect），专注于识别智能对话场景中用户的情绪，针对智能对话场景中的用户文本，自动判断该文本的情绪类别并给出相应的置信度，情绪类型分为积极、消极、中性。
对话情绪识别适用于聊天、客服等多个场景，能够帮助企业更好地把握对话质量、改善产品的用户交互体验，也能分析客服服务质量、降低人工质检成本。本项目代码来源于PaddleNLP，可通过 AI开放平台-对话情绪识别 线上体验。
下载安装命令
## CPU版本安装命令
pip install -f https://paddlepaddle.org.cn/pip/oschina/cpu paddlepaddle
## GPU版本安装命令
pip install -f https://paddlepaddle.org.cn/pip/oschina/gpu paddlepaddle-gpu 注意
本项目代码包含多个文件, Fork并使用GPU环境来运行后, 才能看到项目完整代码, 并正确运行:
并请检查相关参数设置, 例如use_gpu, fluid.CUDAPlace(0)等处是否设置正确.
In[1]
# 解压训练数据
!cd data/data9740/ && unzip -qo 对话情绪识别.zip
!cd data/data9740/ && mv data/* .
!cd data/data9740/ && mv train.tsv train.txt
!cd data/data9740/ && mv test.tsv test.txt
!cd data/data9740/ && mv infer.tsv infer.txt
!cd data/data9740/ && mv dev.tsv dev.txt In[1]
# 代码的整体结构如下
# |---data/ # 训练数据
# |---logs/ # 日志文件
# |---pretrained_model/ # 预训练参数
# |---train_model/ # 训练过程中保存的参数，分散形式
# |---infer_model/ # 固化后的推理模型，参数为分散形式
# |---work/
# | |---reader.py # 自定义的数据读取函数
# | |---model.py # 模型结构
# | |---utils.py # 日志，模型所需工具函数相关
# | |---preprocess.py # 数据预处理，数据增强相关
# | |---loss.py # 优化器、学习率变化，
# |---config.py # 配置文件，强烈建议先看看配置文件
# |---train.py
# |---freeze.py
# |---eval.py
# |---infer.py
#
# 介于该项目比较简单，所以并未完全按照这个结构开发， In[17]
# 建议先看看 config.py 文件，里面集中了绝大多数配置
# 运行开始训练
!python train.py2019-08-06 15:22:31,978 - [1958] [line:133] - INFO: start train text classification, train params: {'vocabulary': 'vocab.txt', 'momentum_strategy': {'learning_rate': 0.001, 'lr_epochs': [6, 8, 10], 'lr_decay': [1, 0.5, 0.25, 0.1]}, 'train_file_list': 'train.txt', 'pretrained_model_dir': './pretrained_model', 'pretrained': False, 'label_file': 'label_list.txt', 'use_gpu': True, 'sgd_strategy': {'learning_rate': 0.001, 'lr_epochs': [6, 8, 10], 'lr_decay': [1, 0.5, 0.25, 0.1]}, 'save_model_dir': './train_model', 'label_dict': {'消极': 0, '中性': 1, '积极': 2}, 'save_freeze_dir': './infer_model', 'sample_count': 9656, 'data_dir': 'data/data9740', 'model_type': 'cnn_net', 'train_batch_size': 128, 'rsm_strategy': {'learning_rate': 0.001, 'lr_epochs': [6, 8, 10], 'lr_decay': [1, 0.5, 0.25, 0.1]}, 'class_dim': 3, 'adam_strategy': {'learning_rate': 0.001}, 'continue_train': False, 'sample_frequency': 10, 'num_epochs': 12}
2019-08-06 15:22:31,978 - [1958] [line:134] - INFO: create place, use gpu: True
2019-08-06 15:22:32,548 - [1958] [line:139] - INFO: define input data tensor
2019-08-06 15:22:32,549 - [1958] [line:144] - INFO: build custom reader
2019-08-06 15:22:32,578 - [1958] [line:148] - INFO: build network cnn_net
2019-08-06 15:22:32,591 - [1958] [line:166] - INFO: build optimizer
W0806 15:22:33.450701 1958 device_context.cc:259] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 9.2, Runtime API Version: 9.0
W0806 15:22:33.454759 1958 device_context.cc:267] device: 0, cuDNN Version: 7.3.
2019-08-06 15:22:33,471 - [1958] [line:181] - INFO: current pass: 0, start read text
2019-08-06 15:22:33,483 - [1958] [line:202] - INFO: pass 0, batch 0, accuracy:0.3671875 loss:1.0980240106582642 time:0.01 sec
2019-08-06 15:22:33,571 - [1958] [line:202] - INFO: pass 0, batch 10, accuracy:0.7734375 loss:0.8332474231719971 time:0.01 sec
2019-08-06 15:22:33,660 - [1958] [line:202] - INFO: pass 0, batch 20, accuracy:0.7109375 loss:0.7695658206939697 time:0.01 sec
2019-08-06 15:22:33,747 - [1958] [line:202] - INFO: pass 0, batch 30, accuracy:0.796875 loss:0.5332558155059814 time:0.01 sec
2019-08-06 15:22:33,833 - [1958] [line:202] - INFO: pass 0, batch 40, accuracy:0.7734375 loss:0.5938246846199036 time:0.01 sec
2019-08-06 15:22:33,921 - [1958] [line:202] - INFO: pass 0, batch 50, accuracy:0.7734375 loss:0.5801756978034973 time:0.01 sec
2019-08-06 15:22:34,009 - [1958] [line:202] - INFO: pass 0, batch 60, accuracy:0.84375 loss:0.47657740116119385 time:0.01 sec
2019-08-06 15:22:34,096 - [1958] [line:202] - INFO: pass 0, batch 70, accuracy:0.8359375 loss:0.4089566469192505 time:0.01 sec
2019-08-06 15:22:34,138 - [1958] [line:209] - INFO: pass 0 train result, current pass mean accuracy:0.7945480707444643 loss:0.5929135970379177
2019-08-06 15:22:34,139 - [1958] [line:211] - INFO: temp save pass 0 train result, current best pass mean accuracy:0.7945480707444643
2019-08-06 15:22:37,070 - [1958] [line:181] - INFO: current pass: 1, start read text
2019-08-06 15:22:37,082 - [1958] [line:202] - INFO: pass 1, batch 0, accuracy:0.875 loss:0.3282725214958191 time:0.01 sec
2019-08-06 15:22:37,178 - [1958] [line:202] - INFO: pass 1, batch 10, accuracy:0.90625 loss:0.27343350648880005 time:0.01 sec
2019-08-06 15:22:37,274 - [1958] [line:202] - INFO: pass 1, batch 20, accuracy:0.921875 loss:0.18297576904296875 time:0.01 sec
2019-08-06 15:22:37,366 - [1958] [line:202] - INFO: pass 1, batch 30, accuracy:0.890625 loss:0.2546702027320862 time:0.01 sec
2019-08-06 15:22:37,455 - [1958] [line:202] - INFO: pass 1, batch 40, accuracy:0.9140625 loss:0.2037762850522995 time:0.01 sec
2019-08-06 15:22:37,541 - [1958] [line:202] - INFO: pass 1, batch 50, accuracy:0.9375 loss:0.23568695783615112 time:0.01 sec
2019-08-06 15:22:37,627 - [1958] [line:202] - INFO: pass 1, batch 60, accuracy:0.9296875 loss:0.18198511004447937 time:0.01 sec
2019-08-06 15:22:37,716 - [1958] [line:202] - INFO: pass 1, batch 70, accuracy:0.8984375 loss:0.29754745960235596 time:0.01 sec
2019-08-06 15:22:37,758 - [1958] [line:209] - INFO: pass 1 train result, current pass mean accuracy:0.9100553228666908 loss:0.24150564776439415
2019-08-06 15:22:37,759 - [1958] [line:211] - INFO: temp save pass 1 train result, current best pass mean accuracy:0.9100553228666908
2019-08-06 15:22:40,647 - [1958] [line:181] - INFO: current pass: 2, start read text
2019-08-06 15:22:40,660 - [1958] [line:202] - INFO: pass 2, batch 0, accuracy:0.9453125 loss:0.1696481853723526 time:0.01 sec
2019-08-06 15:22:40,748 - [1958] [line:202] - INFO: pass 2, batch 10, accuracy:0.984375 loss:0.12824317812919617 time:0.01 sec
2019-08-06 15:22:40,835 - [1958] [line:202] - INFO: pass 2, batch 20, accuracy:0.9453125 loss:0.13505560159683228 time:0.01 sec
2019-08-06 15:22:40,924 - [1958] [line:202] - INFO: pass 2, batch 30, accuracy:0.9453125 loss:0.15139944851398468 time:0.01 sec
2019-08-06 15:22:41,011 - [1958] [line:202] - INFO: pass 2, batch 40, accuracy:0.9765625 loss:0.08035554736852646 time:0.01 sec
2019-08-06 15:22:41,098 - [1958] [line:202] - INFO: pass 2, batch 50, accuracy:0.9609375 loss:0.13403186202049255 time:0.01 sec
2019-08-06 15:22:41,188 - [1958] [line:202] - INFO: pass 2, batch 60, accuracy:0.96875 loss:0.10372276604175568 time:0.01 sec
2019-08-06 15:22:41,275 - [1958] [line:202] - INFO: pass 2, batch 70, accuracy:0.953125 loss:0.13950803875923157 time:0.01 sec
2019-08-06 15:22:41,317 - [1958] [line:209] - INFO: pass 2 train result, current pass mean accuracy:0.9557995662877434 loss:0.1242725053115895
2019-08-06 15:22:41,318 - [1958] [line:211] - INFO: temp save pass 2 train result, current best pass mean accuracy:0.9557995662877434
2019-08-06 15:22:44,193 - [1958] [line:181] - INFO: current pass: 3, start read text
2019-08-06 15:22:44,205 - [1958] [line:202] - INFO: pass 3, batch 0, accuracy:0.984375 loss:0.04495285078883171 time:0.01 sec
2019-08-06 15:22:44,294 - [1958] [line:202] - INFO: pass 3, batch 10, accuracy:0.9609375 loss:0.09051218628883362 time:0.01 sec
2019-08-06 15:22:44,384 - [1958] [line:202] - INFO: pass 3, batch 20, accuracy:0.9765625 loss:0.04519175738096237 time:0.01 sec
2019-08-06 15:22:44,471 - [1958] [line:202] - INFO: pass 3, batch 30, accuracy:0.984375 loss:0.05740939825773239 time:0.01 sec
2019-08-06 15:22:44,559 - [1958] [line:202] - INFO: pass 3, batch 40, accuracy:0.984375 loss:0.0364103764295578 time:0.01 sec
2019-08-06 15:22:44,651 - [1958] [line:202] - INFO: pass 3, batch 50, accuracy:0.96875 loss:0.06522298604249954 time:0.01 sec
2019-08-06 15:22:44,739 - [1958] [line:202] - INFO: pass 3, batch 60, accuracy:0.984375 loss:0.05128682404756546 time:0.01 sec
2019-08-06 15:22:44,828 - [1958] [line:202] - INFO: pass 3, batch 70, accuracy:0.9921875 loss:0.040056630969047546 time:0.01 sec
2019-08-06 15:22:44,870 - [1958] [line:209] - INFO: pass 3 train result, current pass mean accuracy:0.9774203794567209 loss:0.07036846854086769
2019-08-06 15:22:44,870 - [1958] [line:211] - INFO: temp save pass 3 train result, current best pass mean accuracy:0.9774203794567209
2019-08-06 15:22:47,790 - [1958] [line:181] - INFO: current pass: 4, start read text
2019-08-06 15:22:47,803 - [1958] [line:202] - INFO: pass 4, batch 0, accuracy:0.9921875 loss:0.023129595443606377 time:0.01 sec
2019-08-06 15:22:47,894 - [1958] [line:202] - INFO: pass 4, batch 10, accuracy:0.9609375 loss:0.07010433077812195 time:0.01 sec
2019-08-06 15:22:47,983 - [1958] [line:202] - INFO: pass 4, batch 20, accuracy:0.984375 loss:0.062000859528779984 time:0.01 sec
2019-08-06 15:22:48,084 - [1958] [line:202] - INFO: pass 4, batch 30, accuracy:0.9765625 loss:0.05033635348081589 time:0.01 sec
2019-08-06 15:22:48,179 - [1958] [line:202] - INFO: pass 4, batch 40, accuracy:0.984375 loss:0.02487669512629509 time:0.01 sec
2019-08-06 15:22:48,268 - [1958] [line:202] - INFO: pass 4, batch 50, accuracy:0.984375 loss:0.05061167851090431 time:0.01 sec
2019-08-06 15:22:48,359 - [1958] [line:202] - INFO: pass 4, batch 60, accuracy:0.9921875 loss:0.01879894733428955 time:0.01 sec
2019-08-06 15:22:48,450 - [1958] [line:202] - INFO: pass 4, batch 70, accuracy:0.984375 loss:0.04964970424771309 time:0.01 sec
2019-08-06 15:22:48,493 - [1958] [line:209] - INFO: pass 4 train result, current pass mean accuracy:0.9861225321104652 loss:0.043323556145064925
2019-08-06 15:22:48,493 - [1958] [line:211] - INFO: temp save pass 4 train result, current best pass mean accuracy:0.9861225321104652
2019-08-06 15:22:51,377 - [1958] [line:181] - INFO: current pass: 5, start read text
2019-08-06 15:22:51,389 - [1958] [line:202] - INFO: pass 5, batch 0, accuracy:0.9921875 loss:0.026656800881028175 time:0.01 sec
2019-08-06 15:22:51,478 - [1958] [line:202] - INFO: pass 5, batch 10, accuracy:0.9921875 loss:0.01348726823925972 time:0.01 sec
2019-08-06 15:22:51,570 - [1958] [line:202] - INFO: pass 5, batch 20, accuracy:1.0 loss:0.01214015856385231 time:0.01 sec
2019-08-06 15:22:51,658 - [1958] [line:202] - INFO: pass 5, batch 30, accuracy:1.0 loss:0.006573010236024857 time:0.01 sec
2019-08-06 15:22:51,747 - [1958] [line:202] - INFO: pass 5, batch 40, accuracy:0.9765625 loss:0.06301312148571014 time:0.01 sec
2019-08-06 15:22:51,838 - [1958] [line:202] - INFO: pass 5, batch 50, accuracy:0.9765625 loss:0.05193619802594185 time:0.01 sec
2019-08-06 15:22:51,927 - [1958] [line:202] - INFO: pass 5, batch 60, accuracy:1.0 loss:0.01177458930760622 time:0.01 sec
2019-08-06 15:22:52,014 - [1958] [line:202] - INFO: pass 5, batch 70, accuracy:1.0 loss:0.003710896708071232 time:0.01 sec
2019-08-06 15:22:52,057 - [1958] [line:209] - INFO: pass 5 train result, current pass mean accuracy:0.9912623347420442 loss:0.026174830622039735
2019-08-06 15:22:52,057 - [1958] [line:211] - INFO: temp save pass 5 train result, current best pass mean accuracy:0.9912623347420442
2019-08-06 15:22:54,947 - [1958] [line:181] - INFO: current pass: 6, start read text
2019-08-06 15:22:54,959 - [1958] [line:202] - INFO: pass 6, batch 0, accuracy:1.0 loss:0.00326349213719368 time:0.01 sec
2019-08-06 15:22:55,058 - [1958] [line:202] - INFO: pass 6, batch 10, accuracy:0.9921875 loss:0.011361930519342422 time:0.01 sec
2019-08-06 15:22:55,147 - [1958] [line:202] - INFO: pass 6, batch 20, accuracy:1.0 loss:0.00248183635994792 time:0.01 sec
2019-08-06 15:22:55,235 - [1958] [line:202] - INFO: pass 6, batch 30, accuracy:1.0 loss:0.011630130931735039 time:0.01 sec
2019-08-06 15:22:55,326 - [1958] [line:202] - INFO: pass 6, batch 40, accuracy:0.984375 loss:0.02832399122416973 time:0.01 sec
2019-08-06 15:22:55,415 - [1958] [line:202] - INFO: pass 6, batch 50, accuracy:1.0 loss:0.0035656006075441837 time:0.01 sec
2019-08-06 15:22:55,504 - [1958] [line:202] - INFO: pass 6, batch 60, accuracy:0.96875 loss:0.06978510320186615 time:0.01 sec
2019-08-06 15:22:55,595 - [1958] [line:202] - INFO: pass 6, batch 70, accuracy:0.9921875 loss:0.01564416103065014 time:0.01 sec
2019-08-06 15:22:55,637 - [1958] [line:209] - INFO: pass 6 train result, current pass mean accuracy:0.9947574005315178 loss:0.01621751293331679
2019-08-06 15:22:55,637 - [1958] [line:211] - INFO: temp save pass 6 train result, current best pass mean accuracy:0.9947574005315178
2019-08-06 15:22:58,516 - [1958] [line:181] - INFO: current pass: 7, start read text
2019-08-06 15:22:58,529 - [1958] [line:202] - INFO: pass 7, batch 0, accuracy:0.9921875 loss:0.012981493026018143 time:0.01 sec
2019-08-06 15:22:58,619 - [1958] [line:202] - INFO: pass 7, batch 10, accuracy:0.9921875 loss:0.010562659241259098 time:0.01 sec
2019-08-06 15:22:58,706 - [1958] [line:202] - INFO: pass 7, batch 20, accuracy:1.0 loss:0.005756759084761143 time:0.01 sec
2019-08-06 15:22:58,796 - [1958] [line:202] - INFO: pass 7, batch 30, accuracy:1.0 loss:0.0029212036170065403 time:0.01 sec
2019-08-06 15:22:58,885 - [1958] [line:202] - INFO: pass 7, batch 40, accuracy:1.0 loss:0.005461801774799824 time:0.01 sec
2019-08-06 15:22:58,973 - [1958] [line:202] - INFO: pass 7, batch 50, accuracy:1.0 loss:0.00487672770395875 time:0.01 sec
2019-08-06 15:22:59,065 - [1958] [line:202] - INFO: pass 7, batch 60, accuracy:1.0 loss:0.0028506112284958363 time:0.01 sec
2019-08-06 15:22:59,156 - [1958] [line:202] - INFO: pass 7, batch 70, accuracy:0.9921875 loss:0.037087175995111465 time:0.01 sec
2019-08-06 15:22:59,199 - [1958] [line:209] - INFO: pass 7 train result, current pass mean accuracy:0.9974300978999389 loss:0.010514583155255471
2019-08-06 15:22:59,199 - [1958] [line:211] - INFO: temp save pass 7 train result, current best pass mean accuracy:0.9974300978999389
2019-08-06 15:23:02,144 - [1958] [line:181] - INFO: current pass: 8, start read text
2019-08-06 15:23:02,157 - [1958] [line:202] - INFO: pass 8, batch 0, accuracy:1.0 loss:0.0029630300123244524 time:0.01 sec
2019-08-06 15:23:02,247 - [1958] [line:202] - INFO: pass 8, batch 10, accuracy:1.0 loss:0.006094495300203562 time:0.01 sec
2019-08-06 15:23:02,339 - [1958] [line:202] - INFO: pass 8, batch 20, accuracy:1.0 loss:0.0022179321385920048 time:0.01 sec
2019-08-06 15:23:02,427 - [1958] [line:202] - INFO: pass 8, batch 30, accuracy:0.9921875 loss:0.010981489904224873 time:0.01 sec
2019-08-06 15:23:02,516 - [1958] [line:202] - INFO: pass 8, batch 40, accuracy:1.0 loss:0.007036641705781221 time:0.01 sec
2019-08-06 15:23:02,607 - [1958] [line:202] - INFO: pass 8, batch 50, accuracy:1.0 loss:0.007084527984261513 time:0.01 sec
2019-08-06 15:23:02,696 - [1958] [line:202] - INFO: pass 8, batch 60, accuracy:1.0 loss:0.0032452098093926907 time:0.01 sec
2019-08-06 15:23:02,784 - [1958] [line:202] - INFO: pass 8, batch 70, accuracy:1.0 loss:0.0033507051412016153 time:0.01 sec
2019-08-06 15:23:02,829 - [1958] [line:209] - INFO: pass 8 train result, current pass mean accuracy:0.998869242636781 loss:0.005227179575932065
2019-08-06 15:23:02,829 - [1958] [line:211] - INFO: temp save pass 8 train result, current best pass mean accuracy:0.998869242636781
2019-08-06 15:23:05,721 - [1958] [line:181] - INFO: current pass: 9, start read text
2019-08-06 15:23:05,734 - [1958] [line:202] - INFO: pass 9, batch 0, accuracy:1.0 loss:0.0022285240702331066 time:0.01 sec
2019-08-06 15:23:05,823 - [1958] [line:202] - INFO: pass 9, batch 10, accuracy:1.0 loss:0.0009064251207746565 time:0.01 sec
2019-08-06 15:23:05,910 - [1958] [line:202] - INFO: pass 9, batch 20, accuracy:1.0 loss:0.002235196763649583 time:0.01 sec
2019-08-06 15:23:06,000 - [1958] [line:202] - INFO: pass 9, batch 30, accuracy:1.0 loss:0.001683577778749168 time:0.01 sec
2019-08-06 15:23:06,087 - [1958] [line:202] - INFO: pass 9, batch 40, accuracy:1.0 loss:0.00042652207775972784 time:0.01 sec
2019-08-06 15:23:06,174 - [1958] [line:202] - INFO: pass 9, batch 50, accuracy:1.0 loss:0.001178225502371788 time:0.01 sec
2019-08-06 15:23:06,261 - [1958] [line:202] - INFO: pass 9, batch 60, accuracy:1.0 loss:0.0008675195858813822 time:0.01 sec
2019-08-06 15:23:06,350 - [1958] [line:202] - INFO: pass 9, batch 70, accuracy:1.0 loss:0.00487190717831254 time:0.01 sec
2019-08-06 15:23:06,392 - [1958] [line:209] - INFO: pass 9 train result, current pass mean accuracy:0.9991776307946757 loss:0.0031451306632860857
2019-08-06 15:23:06,392 - [1958] [line:211] - INFO: temp save pass 9 train result, current best pass mean accuracy:0.9991776307946757
2019-08-06 15:23:09,326 - [1958] [line:181] - INFO: current pass: 10, start read text
2019-08-06 15:23:09,339 - [1958] [line:202] - INFO: pass 10, batch 0, accuracy:1.0 loss:0.00043672695755958557 time:0.01 sec
2019-08-06 15:23:09,428 - [1958] [line:202] - INFO: pass 10, batch 10, accuracy:1.0 loss:0.0017140342388302088 time:0.01 sec
2019-08-06 15:23:09,518 - [1958] [line:202] - INFO: pass 10, batch 20, accuracy:1.0 loss:0.0005519241094589233 time:0.01 sec
2019-08-06 15:23:09,606 - [1958] [line:202] - INFO: pass 10, batch 30, accuracy:1.0 loss:0.0009649908752180636 time:0.01 sec
2019-08-06 15:23:09,694 - [1958] [line:202] - INFO: pass 10, batch 40, accuracy:1.0 loss:0.0011842334643006325 time:0.01 sec
2019-08-06 15:23:09,784 - [1958] [line:202] - INFO: pass 10, batch 50, accuracy:1.0 loss:0.0012068767100572586 time:0.01 sec
2019-08-06 15:23:09,871 - [1958] [line:202] - INFO: pass 10, batch 60, accuracy:1.0 loss:0.00045044749276712537 time:0.01 sec
2019-08-06 15:23:09,958 - [1958] [line:202] - INFO: pass 10, batch 70, accuracy:1.0 loss:0.0010997994104400277 time:0.01 sec
2019-08-06 15:23:10,000 - [1958] [line:209] - INFO: pass 10 train result, current pass mean accuracy:0.9995551728888562 loss:0.0022376098725209502
2019-08-06 15:23:10,000 - [1958] [line:211] - INFO: temp save pass 10 train result, current best pass mean accuracy:0.9995551728888562
2019-08-06 15:23:12,868 - [1958] [line:181] - INFO: current pass: 11, start read text
2019-08-06 15:23:12,880 - [1958] [line:202] - INFO: pass 11, batch 0, accuracy:1.0 loss:0.0011764491209760308 time:0.01 sec
2019-08-06 15:23:12,972 - [1958] [line:202] - INFO: pass 11, batch 10, accuracy:1.0 loss:0.0012356986990198493 time:0.01 sec
2019-08-06 15:23:13,059 - [1958] [line:202] - INFO: pass 11, batch 20, accuracy:1.0 loss:0.0037445956841111183 time:0.01 sec
2019-08-06 15:23:13,146 - [1958] [line:202] - INFO: pass 11, batch 30, accuracy:1.0 loss:0.0006758287781849504 time:0.01 sec
2019-08-06 15:23:13,235 - [1958] [line:202] - INFO: pass 11, batch 40, accuracy:1.0 loss:0.0007359752780757844 time:0.01 sec
2019-08-06 15:23:13,322 - [1958] [line:202] - INFO: pass 11, batch 50, accuracy:1.0 loss:0.0007078530034050345 time:0.01 sec
2019-08-06 15:23:13,409 - [1958] [line:202] - INFO: pass 11, batch 60, accuracy:1.0 loss:0.003935564309358597 time:0.01 sec
2019-08-06 15:23:13,498 - [1958] [line:202] - INFO: pass 11, batch 70, accuracy:0.9921875 loss:0.015181416645646095 time:0.01 sec
2019-08-06 15:23:13,539 - [1958] [line:209] - INFO: pass 11 train result, current pass mean accuracy:0.9993832228999389 loss:0.0019590824032141092
2019-08-06 15:23:13,540 - [1958] [line:216] - INFO: training till last epcho, end training In[7]
# 固化训练好的参数
!python freeze.py2019-08-06 11:42:48,712 - [360] [line:46] - INFO: build network cnn_net to freeze 2019-08-06 11:42:49,681 - [360] [line:66] - INFO: freeze cnn_net end
In[18]
# 验证训练的结果
!python eval.pyW0806 15:34:57.842015 2027 device_context.cc:259] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 9.2, Runtime API Version: 9.0 W0806 15:34:57.845950 2027 device_context.cc:267] device: 0, cuDNN Version: 7.3. total eval count:1036 cost time:1.32 sec predict accuracy:0.8397683397683398
In[2]
# 测试预测一下
!python infer.pyW0906 15:00:01.984120 98 device_context.cc:259] Please NOTE: device: 0, CUDA Capability: 70, Driver API Version: 9.2, Runtime API Version: 9.0 W0906 15:00:01.987993 98 device_context.cc:267] device: 0, cuDNN Version: 7.3. infer text:向宇波 就是 喜欢 刘夏 可爱 运动 独立 infer result:2
点击链接，使用AI Studio一键上手实践项目吧：https://aistudio.baidu.com/aistudio/projectdetail/122291
下载安装命令
## CPU版本安装命令
pip install -f https://paddlepaddle.org.cn/pip/oschina/cpu paddlepaddle
## GPU版本安装命令
pip install -f https://paddlepaddle.org.cn/pip/oschina/gpu paddlepaddle-gpu>> 访问 PaddlePaddle 官网，了解更多相关内容。