pip install Keras-TextClassificationstep2: download and unzip the dir of 'data.rar', 地址: https://pan.baidu.com/s/1I3vydhmFEQ9nuPG2fDou8Q 提取码: rket
cover the dir of data to anaconda, like '/anaconda/3.5.1/envs/tensorflow13/Lib/site-packages/keras_textclassification/data'
step3: goto # Train&Usage(调用) and Predict&Usage(调用)- Albert-fineture
- Xlnet-fineture
- Bert-fineture
- FastText
- TextCNN
- charCNN
- TextRNN
- TextRCNN
- TextDCNN
- TextDPCNN
- TextVDCNN
- TextCRNN
- DeepMoji
- SelfAttention
- HAN
- CapsuleNet
- Transformer-encode
- 1. 进入keras_textclassification/m01_FastText目录，
- 2. 训练: 运行 train.py, 例如: python train.py
- 3. 预测: 运行 predict.py, 例如: python predict.py
- 说明: 默认不带pre train的random embedding，训练和验证语料只有100条，完整语料移步下面data查看下载
- bert,word2vec,random样例在test/目录下, 注意word2vec(char or word), random-word, bert(chinese_L-12_H-768_A-12)未全部加载,需要下载
- multi_multi_class/目录下以text-cnn为例进行多标签分类实例，转化为multi-onehot标签类别，分类则取一定阀值的类
- sentence_similarity/目录下以bert为例进行两个句子文本相似度计算,数据格式如data/sim_webank/目录下所示
- predict_bert_text_cnn.py
- tet_char_bert_embedding.py
- tet_char_bert_embedding.py
- tet_char_xlnet_embedding.py
- tet_char_random_embedding.py
- tet_char_word2vec_embedding.py
- tet_word_random_embedding.py
- tet_word_word2vec_embedding.py
- 数据下载
** github项目中只是上传部分数据，需要的前往链接: https://pan.baidu.com/s/1I3vydhmFEQ9nuPG2fDou8Q 提取码: rket
- baidu_qa_2019（百度qa问答语料，只取title作为分类样本，17个类，有一个是空''，已经压缩上传）
- baike_qa_train.csv
- baike_qa_valid.csv
- byte_multi_news（今日头条2018新闻标题多标签语料，1070个标签，fate233爬取, 地址为: [byte_multi_news](https://github.com/fate233/toutiao-multilevel-text-classfication-dataset)）
-labels.csv
-train.csv
-valid.csv
- embeddings
- chinese_L-12_H-768_A-12/(取谷歌预训练好点的模型,已经压缩上传,
keras-bert还可以加载百度版ernie(需转换，[https://github.com/ArthurRizar/tensorflow_ernie](https://github.com/ArthurRizar/tensorflow_ernie)),
哈工大版bert-wwm(tf框架，[https://github.com/ymcui/Chinese-BERT-wwm](https://github.com/ymcui/Chinese-BERT-wwm))
- albert_base_zh/(brightmart训练的albert, 地址为https://github.com/brightmart/albert_zh)
- chinese_xlnet_mid_L-24_H-768_A-12/(哈工大预训练的中文xlnet模型[https://github.com/ymcui/Chinese-PreTrained-XLNet],24层)
- term_char.txt(已经上传, 项目中已全, wiki字典, 还可以用新华字典什么的)
- term_word.txt(未上传, 项目中只有部分, 可参考词向量的)
- w2v_model_merge_short.vec(未上传, 项目中只有部分, 词向量, 可以用自己的)
- w2v_model_wiki_char.vec(已上传百度网盘, 项目中只有部分, 自己训练的维基百科字向量, 可以用自己的)
- model
- fast_text/预训练模型存放地址
-
- 构建了base基类(网络(graph)、向量嵌入(词、字、句子embedding)),后边的具体模型继承它们，代码简单
-
- keras_layers存放一些常用的layer, conf存放项目数据、模型的地址, data存放数据和语料, data_preprocess为数据预处理模块,
- FastText: Bag of Tricks for Efficient Text Classification
- TextCNN： Convolutional Neural Networks for Sentence Classification
- charCNN-kim： Character-Aware Neural Language Models
- charCNN-zhang: Character-level Convolutional Networks for Text Classification
- TextRNN： Recurrent Neural Network for Text Classification with Multi-Task Learning
- RCNN： Recurrent Convolutional Neural Networks for Text Classification
- DCNN: A Convolutional Neural Network for Modelling Sentences
- DPCNN: Deep Pyramid Convolutional Neural Networks for Text Categorization
- VDCNN: Very Deep Convolutional Networks
- CRNN: A C-LSTM Neural Network for Text Classification
- DeepMoji: Using millions of emojio ccurrences to learn any-domain represent ations for detecting sentiment, emotion and sarcasm
- SelfAttention: Attention Is All You Need
- HAN: Hierarchical Attention Networks for Document Classification
- CapsuleNet: Dynamic Routing Between Capsules
- Transformer(encode or decode): Attention Is All You Need
- Bert: BERT: Pre-trainingofDeepBidirectionalTransformersfor LanguageUnderstanding
- Xlnet: XLNet: Generalized Autoregressive Pretraining for Language Understanding
- Albert: ALBERT: A LITE BERT FOR SELF-SUPERVISED LEARNING OF LANGUAGE REPRESENTATIONS
- 文本分类项目: https://github.com/mosu027/TextClassification
- 文本分类看山杯: https://github.com/brightmart/text_classification
- Kashgari项目: https://github.com/BrikerMan/Kashgari
- 文本分类Ipty : https://github.com/lpty/classifier
- keras文本分类: https://github.com/ShawnyXiao/TextClassification-Keras
- keras文本分类: https://github.com/AlexYangLi/TextClassification
- CapsuleNet模型: https://github.com/bojone/Capsule
- transformer模型: https://github.com/CyberZHG/keras-transformer
- keras_albert_model: https://github.com/TinkerMob/keras_albert_model
from keras_textclassification import train
train(graph='TextCNN', # 必填, 算法名, 可选"ALBERT","BERT","XLNET","FASTTEXT","TEXTCNN","CHARCNN",
# "TEXTRNN","RCNN","DCNN","DPCNN","VDCNN","CRNN","DEEPMOJI",
# "SELFATTENTION", "HAN","CAPSULE","TRANSFORMER"
label=17, # 必填, 类别数, 训练集和测试集合必须一样
path_train_data=None, # 必填, 训练数据文件, csv格式, 必须含'label,ques'头文件, 详见keras_textclassification/data
path_dev_data=None, # 必填, 测试数据文件, csv格式, 必须含'label,ques'头文件, 详见keras_textclassification/data
rate=1, # 可填, 训练数据选取比例
hyper_parameters=None) # 可填, json格式, 超参数, 默认embedding为'char','random'# 适配linux
import pathlib
import sys
import os
project_path = str(pathlib.Path(os.path.abspath(__file__)).parent.parent.parent)
sys.path.append(project_path)
from keras_textclassification.conf.path_config import path_model_dir
# 数据预处理, 删除文件目录下文件
from keras_textclassification.data_preprocess.text_preprocess import PreprocessText, delete_file
# 模型图
from keras_textclassification.m02_TextCNN.graph import TextCNNGraph as Graph
# 计算时间
import time
# 可配置地址
# path_model_dir = 'Y:/tet_keras_textclassification/'
path_model = path_model_dir + '/textcnn.model'
path_fineture = path_model_dir + '/fineture.embedding'
path_hyper_parameters = path_model_dir + '/hyper_parameters.json'
# 输入训练验证文件地址,sample数据集label填17
# path_train = path_model_dir + 'data/train.csv'
# path_valid = path_model_dir + 'data/val.csv'
# # or 输入训练/预测list, 这时候label选择填3
path_train = ['游戏,斩 魔仙 者 称号 怎么 得来 的', '文化,我爱你 古文 怎么 说', '健康,牙龈 包住 牙齿 怎么办']
path_valid = ['娱乐,李克勤 什么 歌 好听', '电脑,UPS 电源 工作 原理', '文化,我爱你 古文 怎么 说 的 呢']
# 会删除存在的model目录下的所有文件
# path_model_dir = 'Y:/tet_keras_textclassification/model/'
def train(hyper_parameters=None, rate=1.0):
if not hyper_parameters:
# 可配置参数
hyper_parameters = {
'len_max': 50, # 句子最大长度, 固定推荐20-50, bert越长会越慢, 占用空间也会变大, 本地win10-4G设为20就好, 过大小心OOM
'embed_size': 300, # 字/词向量维度, bert取768, word取300, char可以更小些
'vocab_size': 20000, # 这里随便填的，会根据代码里修改
'trainable': True, # embedding是静态的还是动态的, 即控制可不可以微调
'level_type': 'char', # 级别, 最小单元, 字/词, 填 'char' or 'word', 注意:word2vec模式下训练语料要首先切好
'embedding_type': 'random', # 级别, 嵌入类型, 还可以填'xlnet'、'random'、 'bert'、 'albert' or 'word2vec"
'gpu_memory_fraction': 0.66, #gpu使用率
'model': {'label': 3, # 类别数
'batch_size': 5, # 批处理尺寸, 感觉原则上越大越好,尤其是样本不均衡的时候, batch_size设置影响比较大
'dropout': 0.5, # 随机失活, 概率
'decay_step': 100, # 学习率衰减step, 每N个step衰减一次
'decay_rate': 0.9, # 学习率衰减系数, 乘法
'epochs': 20, # 训练最大轮次
'patience': 3, # 早停,2-3就好
'lr': 5e-5, # 学习率,bert取5e-5,其他取1e-3, 对训练会有比较大的影响, 如果准确率一直上不去,可以考虑调这个参数
'l2': 1e-9, # l2正则化
'activate_classify': 'softmax', # 最后一个layer, 即分类激活函数
'loss': 'categorical_crossentropy', # 损失函数
'metrics': 'accuracy', # 保存更好模型的评价标准
'is_training': True, # 训练后者是测试模型
'model_path': path_model,
# 模型地址, loss降低则保存的依据, save_best_only=True, save_weights_only=True
'path_hyper_parameters': path_hyper_parameters, # 模型(包括embedding)，超参数地址,
'path_fineture': path_fineture, # 保存embedding trainable地址, 例如字向量、词向量、bert向量等
},
'embedding': {'layer_indexes': [12], # bert取的层数
# 'corpus_path': '', # embedding预训练数据地址,不配则会默认取conf里边默认的地址
},
'data':{'train_data': path_train, # 训练数据
'val_data': path_valid # 验证数据
},
}
# 删除先前存在的模型和embedding微调模型等
delete_file(path_model_dir)
time_start = time.time()
# graph初始化
graph = Graph(hyper_parameters)
print("graph init ok!")
ra_ed = graph.word_embedding
# 数据预处理
pt = PreprocessText()
x_train, y_train = pt.preprocess_label_ques_to_idx(hyper_parameters['embedding_type'],
hyper_parameters['data']['train_data'],
ra_ed, rate=rate, shuffle=True)
x_val, y_val = pt.preprocess_label_ques_to_idx(hyper_parameters['embedding_type'],
hyper_parameters['data']['val_data'],
ra_ed, rate=rate, shuffle=True)
print("data propress ok!")
print(len(y_train))
# 训练
graph.fit(x_train, y_train, x_val, y_val)
print("耗时:" + str(time.time()-time_start))
if __name__=="__main__":
train(rate=1)
# 注意: 4G的1050Ti的GPU、win10下batch_size=32,len_max=20, gpu<=0.87, 应该就可以bert-fineture了。
# 全量数据训练一轮(batch_size=32),就能达到80%准确率(验证集), 效果还是不错的
# win10下出现过错误,gpu、len_max、batch_size配小一点就好:ailed to allocate 3.56G (3822520832 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY: out of memory# 适配linux
import pathlib
import sys
import os
project_path = str(pathlib.Path(os.path.abspath(__file__)).parent.parent.parent)
sys.path.append(project_path)
# 数据预处理, 删除文件目录下文件
from keras_textclassification.data_preprocess.text_preprocess import PreprocessText, read_and_process, load_json
# 模型图
from keras_textclassification.m02_TextCNN.graph import TextCNNGraph as Graph
from keras_textclassification.conf.path_config import path_model_dir
# 计算时间
import time
import numpy as np
def pred_input(path_hyper_parameter):
# 输入预测
# 加载超参数
hyper_parameters = load_json(path_hyper_parameter)
pt = PreprocessText()
# 模式初始化和加载
graph = Graph(hyper_parameters)
graph.load_model()
ra_ed = graph.word_embedding
ques = '我要打王者荣耀'
# str to token
ques_embed = ra_ed.sentence2idx(ques)
if hyper_parameters['embedding_type'] == 'bert':
x_val_1 = np.array([ques_embed[0]])
x_val_2 = np.array([ques_embed[1]])
x_val = [x_val_1, x_val_2]
else:
x_val = ques_embed
# 预测
pred = graph.predict(x_val)
# 取id to label and pred
pre = pt.prereocess_idx(pred[0])
print(pre)
while True:
print("请输入: ")
ques = input()
ques_embed = ra_ed.sentence2idx(ques)
print(ques_embed)
if hyper_parameters['embedding_type'] == 'bert':
x_val_1 = np.array([ques_embed[0]])
x_val_2 = np.array([ques_embed[1]])
x_val = [x_val_1, x_val_2]
else:
x_val = ques_embed
pred = graph.predict(x_val)
pre = pt.prereocess_idx(pred[0])
print(pre)
if __name__=="__main__":
# 可输入 input 预测
pred_input(path_hyper_parameter=path_model_dir + '/hyper_parameters.json')*希望对你有所帮助!