All Projects → yanwii → Seq2seq

yanwii / Seq2seq

Licence: apache-2.0
基于Pytorch的中文聊天机器人 集成BeamSearch算法

Programming Languages

python
139335 projects - #7 most used programming language

Projects that are alternatives of or similar to Seq2seq

minimal-nmt
A minimal nmt example to serve as an seq2seq+attention reference.
Stars: ✭ 36 (-82%)
Mutual labels:  seq2seq, beam-search
neural-chat
An AI chatbot using seq2seq
Stars: ✭ 30 (-85%)
Mutual labels:  seq2seq, beam-search
beam search
Beam search for neural network sequence to sequence (encoder-decoder) models.
Stars: ✭ 31 (-84.5%)
Mutual labels:  seq2seq, beam-search
Poetry Seq2seq
Chinese Poetry Generation
Stars: ✭ 159 (-20.5%)
Mutual labels:  beam-search, seq2seq
Im2latex
Image to LaTeX (Seq2seq + Attention with Beam Search) - Tensorflow
Stars: ✭ 342 (+71%)
Mutual labels:  beam-search, seq2seq
Nlp made easy
Explains nlp building blocks in a simple manner.
Stars: ✭ 232 (+16%)
Mutual labels:  beam-search, seq2seq
transformer
Neutron: A pytorch based implementation of Transformer and its variants.
Stars: ✭ 60 (-70%)
Mutual labels:  seq2seq, beam-search
Dynamic Seq2seq
seq2seq中文聊天机器人
Stars: ✭ 303 (+51.5%)
Mutual labels:  chatbots, seq2seq
Pytorch Chatbot
Pytorch seq2seq chatbot
Stars: ✭ 336 (+68%)
Mutual labels:  beam-search, seq2seq
Seq2seq chatbot
基于seq2seq模型的简单对话系统的tf实现,具有embedding、attention、beam_search等功能,数据集是Cornell Movie Dialogs
Stars: ✭ 308 (+54%)
Mutual labels:  beam-search, seq2seq
Seq2seq chatbot new
基于seq2seq模型的简单对话系统的tf实现,具有embedding、attention、beam_search等功能,数据集是Cornell Movie Dialogs
Stars: ✭ 144 (-28%)
Mutual labels:  beam-search, seq2seq
Tf Seq2seq
Sequence to sequence learning using TensorFlow.
Stars: ✭ 387 (+93.5%)
Mutual labels:  beam-search, seq2seq
Neural conversation models
Tensorflow based Neural Conversation Models
Stars: ✭ 192 (-4%)
Mutual labels:  chatbots, beam-search
Tensorflow Tutorials
텐서플로우를 기초부터 응용까지 단계별로 연습할 수 있는 소스 코드를 제공합니다
Stars: ✭ 2,096 (+948%)
Mutual labels:  seq2seq
Nspm
🤖 Neural SPARQL Machines for Knowledge Graph Question Answering.
Stars: ✭ 156 (-22%)
Mutual labels:  seq2seq
Nlp pytorch project
Embedding, NMT, Text_Classification, Text_Generation, NER etc.
Stars: ✭ 153 (-23.5%)
Mutual labels:  seq2seq
Kospeech
Open-Source Toolkit for End-to-End Korean Automatic Speech Recognition.
Stars: ✭ 190 (-5%)
Mutual labels:  seq2seq
Tensorflow Ml Nlp
텐서플로우와 머신러닝으로 시작하는 자연어처리(로지스틱회귀부터 트랜스포머 챗봇까지)
Stars: ✭ 176 (-12%)
Mutual labels:  seq2seq
Awesome Speech Recognition Speech Synthesis Papers
Automatic Speech Recognition (ASR), Speaker Verification, Speech Synthesis, Text-to-Speech (TTS), Language Modelling, Singing Voice Synthesis (SVS), Voice Conversion (VC)
Stars: ✭ 2,085 (+942.5%)
Mutual labels:  seq2seq
Botonic
Build chatbots and conversational experiences using React
Stars: ✭ 144 (-28%)
Mutual labels:  chatbots

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)
Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].