Skip to content
/ seq2seq Public

基于Pytorch的中文聊天机器人 集成BeamSearch算法

License

Notifications You must be signed in to change notification settings

yanwii/seq2seq

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

16 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

基于Pytorch的中文聊天机器人 集成BeamSearch算法

Pytorch 厉害了!


Requirements:
Python3
Pytorch
Jieba分词


Pytorch 安装

    python2.7
    pip2 install http://download.pytorch.org/whl/cu80/torch-0.2.0.post3-cp27-cp27mu-manylinux1_x86_64.whl 
    pip2 install torchvision 

    python3.5
    pip3 install http://download.pytorch.org/whl/cu80/torch-0.2.0.post3-cp35-cp35m-manylinux1_x86_64.whl 
    pip3 install torchvision
    
    python3.6
    pip3 install http://download.pytorch.org/whl/cu80/torch-0.2.0.post3-cp36-cp36m-manylinux1_x86_64.whl 
    pip3 install torchvision

关于BeamSearch算法

很经典的贪心算法,在很多领域都有应用。

在这个引用中 我们引入了惩罚因子


用法

    # 准备数据
    python3 preprocessing.py
    # 训练
    python3 seq2seq.py train
    # 预测
    python3 seq2seq.py predict
    # 重新训练
    python3 seq2seq.py retrain

以下是k=5时的结果, 越接近1,结果越好

    me > 我是谁
    drop [3, 1], 1
    drop [1, 6, 1], 2
    drop [7, 6, 1], 3
    drop [4, 5, 6, 1], 4
    drop [7, 6, 8, 1], 5
    ai >  __UNK__ -1.92623626371
    ai >   -1.41548742168
    ai >  关你 -1.83084125204
    ai >  我是你 0.0647218796512
    ai >  关你屁事 -0.311924366579

Status

2017-09-23 Update

    修复
    ValueError: Expected 2 or 4 dimensions (got 1)

About

基于Pytorch的中文聊天机器人 集成BeamSearch算法

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages