Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Multitask Deep Learning for Sentiment Analysis using Character-Level Language Model, Bi-LSTMs for POS Tag, Chunking and Unsupervised Dependency Parsing. Inspired by this great article https://arxiv.org/abs/1611.01587

Stars: ✭ 93 (-44.31%)

Mutual labels: lstm, recurrent-neural-networks

Pytorch Learners Tutorial

PyTorch tutorial for learners

Stars: ✭ 97 (-41.92%)

Mutual labels: lstm, recurrent-neural-networks

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 (-25.75%)

Mutual labels: lstm, seq2seq

Lstm Ctc Ocr

using rnn (lstm or gru) and ctc to convert line image into text, based on torch7 and warp-ctc

Stars: ✭ 70 (-58.08%)

Mutual labels: lstm, recurrent-neural-networks

Rnn Text Classification Tf

Tensorflow Implementation of Recurrent Neural Network (Vanilla, LSTM, GRU) for Text Classification

Stars: ✭ 114 (-31.74%)

Mutual labels: lstm, recurrent-neural-networks

Image Caption Generator

A neural network to generate captions for an image using CNN and RNN with BEAM Search.

Stars: ✭ 126 (-24.55%)

Mutual labels: lstm, recurrent-neural-networks

Language Translation

Neural machine translator for English2German translation.

Stars: ✭ 82 (-50.9%)

Mutual labels: lstm, recurrent-neural-networks

Tensorflow seq2seq chatbot

Stars: ✭ 81 (-51.5%)

Mutual labels: lstm, seq2seq

Nspm

🤖 Neural SPARQL Machines for Knowledge Graph Question Answering.

Stars: ✭ 156 (-6.59%)

Mutual labels: lstm, seq2seq

Ai Reading Materials

Some of the ML and DL related reading materials, research papers that I've read

Stars: ✭ 79 (-52.69%)

Mutual labels: lstm, recurrent-neural-networks

Linear Attention Recurrent Neural Network

A recurrent attention module consisting of an LSTM cell which can query its own past cell states by the means of windowed multi-head attention. The formulas are derived from the BN-LSTM and the Transformer Network. The LARNN cell with attention can be easily used inside a loop on the cell state, just like any other RNN. (LARNN)

Stars: ✭ 119 (-28.74%)

Mutual labels: lstm, recurrent-neural-networks

Document Classifier Lstm

A bidirectional LSTM with attention for multiclass/multilabel text classification.

Stars: ✭ 136 (-18.56%)

Mutual labels: lstm, recurrent-neural-networks

Stockprediction

Plain Stock Close-Price Prediction via Graves LSTM RNNs

Stars: ✭ 134 (-19.76%)

Mutual labels: lstm, recurrent-neural-networks

Keras Lmu

Keras implementation of Legendre Memory Units

Stars: ✭ 160 (-4.19%)

Mutual labels: lstm, recurrent-neural-networks

Poetry Seq2seq

Chinese Poetry Generation

Stars: ✭ 159 (-4.79%)

Mutual labels: lstm, seq2seq

View All Similar Projects ➔

News summarization

News summarization using sequence to sequence model in TensorFlow.

Introduction

This repository is a demonstration of abstractive summarization of news article exploiting TensorFlow sequence to sequence model. This model incorporates attention mechanism and uses LSTM cell as both encoder and decoder.

This model is trained on one million Associated Press Worldstream news stories from English Gigaword second edition. The examples below are based on the model trained on AWS EC2 g2.2xlarge instance for 10 epochs, which took around 20 hours.

For more detailed information, please see our project research paper: Headline Generation Using Recurrent Neural Network.

Examples

News 1

News: A roadside bomb killed five people Thursday near a shelter used as a police recruiting center in northeast Baghdad, police said.

Actual headline: Iraqi police: Bomb kills 5 near police recruiting center in northeast Baghdad

Predicted headline: URGENT Explosion kills five people in Baghdad

News 2

News: The euro hit a record high against the dollar Monday in Asia as concerns over the U.S. subprime mortgage crisis remain a heavy weight on the greenback.

Actual headline: Euro hits record high versus dollar in Asian trading

Predicted headline: Euro hits record high against dollar

How to run

For demonstration, we use the sample file (a very small portion of English Gigaword) from LDC as our dataset to train our model. If you want to reproduce the results like the above examples, larger training set is necessary. You can download the trained model parameters which was trained on a larger portion on Gigaword by following the instructions in the Download vocabs and trained model parameters section below. The whole English Gigaword can be obtained from university libraries.

Pre-req

Install Python 3
Download deep-news-summarization

$ git clone https://github.com/hengluchang/deep-news-summarization.git

Install TensorFlow 0.12, pandas, Numpy, nltk, and requests

$ pip install -r requirements.txt

Create two folders named "working_dir" and "output" under the deep-news-summarization folder.

$ cd deep-news-summarization
$ mkdir -p working_dir output

Download vocabs and trained model parameters

Run download_vocabs_and_trained_params.py file. This will download encoder and decoder vocabularies and trained model parameters to working_dir folder.

$ python download_vocabs_and_trained_params.py ./working_dir

Go to Interactive testing section below to reproduce the results as the examples above.

Train your own summarizer

Set "mode = train" in seq2seq.ini file.
Run split_data.py file to split the dataset into training, evaluation, and testing sets. train_enc.txt, eval_enc.txt, test_enc, train_dec.txt, eval_dec.txt, and test_dec.txt total of six files will be created under ./dataset.

$ python split_data.py

Run execute.py file. This will create vocab80000_enc.txt, vocab80000_dec.txt, and checkpoint data under ./working_dir. If you use your own dataset, optimizing bucket sizes to minimize padding in execute.py file can help to get better results. Also, keep training the model until the preplexity of the evaluation sets are under 10 for better performances.

$ python execute.py

Testing

Set "mode = test" in seq2seq.ini file.
Run execute.py file. This will read the model parameters (seq2seq.ckpt-XXXXX) into your model and create predicted_test_headline.txt under ./output.

$ python execute.py

Run evaluation.py file to get BLEU scores between actual headlines and predicted headlines. This will create BLEU.txt file.

$ python evaluation.py

Interactive testing

Set "mode = interactive" in seq2seq.ini file.
Run execute.py. This will read the model parameters (seq2seq.ckpt-XXXXX) into your model and ask user for an input.

$ python execute.py

References

Sequence-to-Sequence Models: TensorFlow's tutorial using seq2seq_model.py.
Chatbots with Seq2Seq: I adpoted most of the code from this blog.
Speakeasy chatbot: A blog with benchmarked training step-time using various computing engines.
Generating News Headlines with Recurrent Neural Networks: A related work in generating news haedlines.
Evaluation and preplexity: A youtube video explaining preplexity.
Understanding LSTM Networks: A blog explaning LSTM.

Research Paper References

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].

Stars: ✭ 167

Visit Git Page 🔗Visit User Page 🔗Visit Issues Page (6) 🔗