All Projects → yoosan → Mxnet Seq2seq

yoosan / Mxnet Seq2seq

Sequence to sequence learning with MXNET

Programming Languages

python
139335 projects - #7 most used programming language

Projects that are alternatives of or similar to Mxnet Seq2seq

Tensorflow Tutorials
텐서플로우를 기초부터 응용까지 단계별로 연습할 수 있는 소스 코드를 제공합니다
Stars: ✭ 2,096 (+4009.8%)
Mutual labels:  rnn, seq2seq
DeepLearning-Lab
Code lab for deep learning. Including rnn,seq2seq,word2vec,cross entropy,bidirectional rnn,convolution operation,pooling operation,InceptionV3,transfer learning.
Stars: ✭ 83 (+62.75%)
Mutual labels:  rnn, seq2seq
Natural Language Processing With Tensorflow
Natural Language Processing with TensorFlow, published by Packt
Stars: ✭ 222 (+335.29%)
Mutual labels:  rnn, seq2seq
Chinese Chatbot
中文聊天机器人,基于10万组对白训练而成,采用注意力机制,对一般问题都会生成一个有意义的答复。已上传模型,可直接运行,跑不起来直播吃键盘。
Stars: ✭ 124 (+143.14%)
Mutual labels:  rnn, seq2seq
Text summurization abstractive methods
Multiple implementations for abstractive text summurization , using google colab
Stars: ✭ 359 (+603.92%)
Mutual labels:  rnn, seq2seq
Poetry Seq2seq
Chinese Poetry Generation
Stars: ✭ 159 (+211.76%)
Mutual labels:  rnn, seq2seq
tensorflow-ml-nlp-tf2
텐서플로2와 머신러닝으로 시작하는 자연어처리 (로지스틱회귀부터 BERT와 GPT3까지) 실습자료
Stars: ✭ 245 (+380.39%)
Mutual labels:  rnn, 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 (+3988.24%)
Mutual labels:  rnn, seq2seq
LSTM-CTC-recaptcha
recaptcha with lstm and mxnet
Stars: ✭ 28 (-45.1%)
Mutual labels:  mxnet, rnn
GAN-RNN Timeseries-imputation
Recurrent GAN for imputation of time series data. Implemented in TensorFlow 2 on Wikipedia Web Traffic Forecast dataset from Kaggle.
Stars: ✭ 107 (+109.8%)
Mutual labels:  rnn, seq2seq
Kaggle Web Traffic
1st place solution
Stars: ✭ 1,641 (+3117.65%)
Mutual labels:  rnn, seq2seq
Seq2seq Pytorch
Sequence to Sequence Models with PyTorch
Stars: ✭ 678 (+1229.41%)
Mutual labels:  rnn, seq2seq
Pytorch Seq2seq
Tutorials on implementing a few sequence-to-sequence (seq2seq) models with PyTorch and TorchText.
Stars: ✭ 3,418 (+6601.96%)
Mutual labels:  rnn, seq2seq
Base-On-Relation-Method-Extract-News-DA-RNN-Model-For-Stock-Prediction--Pytorch
基於關聯式新聞提取方法之雙階段注意力機制模型用於股票預測
Stars: ✭ 33 (-35.29%)
Mutual labels:  rnn, seq2seq
Deeplearning
深度学习入门教程, 优秀文章, Deep Learning Tutorial
Stars: ✭ 6,783 (+13200%)
Mutual labels:  rnn, mxnet
Sockeye
Sequence-to-sequence framework with a focus on Neural Machine Translation based on Apache MXNet
Stars: ✭ 990 (+1841.18%)
Mutual labels:  mxnet, seq2seq
Boilerplate Dynet Rnn Lm
Boilerplate code for quickly getting set up to run language modeling experiments
Stars: ✭ 37 (-27.45%)
Mutual labels:  rnn
Deep Speeling
Deep Learning neural network for correcting spelling
Stars: ✭ 45 (-11.76%)
Mutual labels:  rnn
Tensorly
TensorLy: Tensor Learning in Python.
Stars: ✭ 977 (+1815.69%)
Mutual labels:  mxnet
Neural Networks
All about Neural Networks!
Stars: ✭ 34 (-33.33%)
Mutual labels:  rnn

mxnet-seq2seq

This project implements the sequence to sequence learning with mxnet for open-domain chatbot

Sequence to Sequence learning with LSTM encoder-decoder

The seq2seq encoder-decoder architecture is introduced by Sequence to Sequence Learning with Neural Networks

This implementation borrows idea from lstm_bucketing, I slightly modified it and reconstructed the embedding layer.

How to run

Firstly, process the data by

python datautils.py

then run the model by

python main.py

The architecture

We know that seq2seq encoder-decoder architecture includes two RNNs (LSTMs), one for encoding source sequence and another for decoding target sequence.

For NLP-related tasks, the sequence could be a natural language sentence. As a result, the encoder and decoder should share the word embedding layer .

The bucketing is a grate solution adapting the arbitrariness of sequence length. I padding zero to a fixed length at the encoding sequence and make buckets at the decoding phrase.

The data is formatted as:

0 0 ... 0 23 12 121 832 || 2 3432 898 7 323
0 0 ... 0 43 98 233 323 || 7 4423 833 1 232
0 0 ... 0 32 44 133 555 || 2 4534 545 6 767
---
0 0 ... 0 23 12 121 832 || 2 3432 898 7
0 0 ... 0 23 12 121 832 || 2 3432 898 7
0 0 ... 0 23 12 121 832 || 2 3432 898 7
---

The input shape for embedding layer is (batch_size, seq_len), the input shape for lstm encoder is (batch_size, seq_len, embed_dim) .

More details coming soon

For any question, please send me email.

yoosan.zhou at gmail dot com
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].