All Projects → alesee → Abstractive Text Summarization

alesee / Abstractive Text Summarization

PyTorch implementation/experiments on Abstractive Text Summarization using Sequence-to-sequence RNNs and Beyond paper.

Projects that are alternatives of or similar to Abstractive Text Summarization

Pandas
pandas cheetsheet
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Ysda deeplearning17
Yandex SDA classes on deep learning. Version of year 2017
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Nestedtensor
[Prototype] Tools for the concurrent manipulation of variably sized Tensors.
Stars: ✭ 119 (+0%)
Mutual labels:  jupyter-notebook
Vae Tensorflow
A Tensorflow implementation of a Variational Autoencoder for the deep learning course at the University of Southern California (USC).
Stars: ✭ 117 (-1.68%)
Mutual labels:  jupyter-notebook
Qiskit Tutorials
A collection of Jupyter notebooks showing how to use the Qiskit SDK
Stars: ✭ 1,777 (+1393.28%)
Mutual labels:  jupyter-notebook
Senato.py
A scraper for the data made available by the Italian Senate, and a cluster analysis to detect similar amendments.
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Amazonsagemakercourse
SageMaker Course Material
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Pydatadc 2018 Tidy
PyData 2018 tutorial for tidying data
Stars: ✭ 119 (+0%)
Mutual labels:  jupyter-notebook
Planet Amazon Deforestation
The open source repository for the Kaggle Amazon forest devastation competition https://www.kaggle.com/c/planet-understanding-the-amazon-from-space
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Deeplearning With Tensorflow Notes
龙曲良《TensorFlow深度学习》学习笔记及代码,采用TensorFlow2.0.0版本
Stars: ✭ 119 (+0%)
Mutual labels:  jupyter-notebook
Reinforcementlearning Atarigame
Pytorch LSTM RNN for reinforcement learning to play Atari games from OpenAI Universe. We also use Google Deep Mind's Asynchronous Advantage Actor-Critic (A3C) Algorithm. This is much superior and efficient than DQN and obsoletes it. Can play on many games
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Statistical Learning Method
《统计学习方法》笔记-基于Python算法实现
Stars: ✭ 1,643 (+1280.67%)
Mutual labels:  jupyter-notebook
Midi Dataset
Code for creating a dataset of MIDI ground truth
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Bitcoin Price Prediction Using Sentiment Analysis
Predicts real-time bitcoin price using twitter and reddit sentiment, and sends out notifications via SMS.
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Chromagan
Official Implementation of ChromaGAN: An Adversarial Approach for Picture Colorization
Stars: ✭ 117 (-1.68%)
Mutual labels:  jupyter-notebook
Teach Me Quantum
⚛ 10 week Practical Course on Quantum Information Science and Quantum Computing - with Qiskit and IBMQX
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Tensorflow shiny
A R/Shiny app for interactive RNN tensorflow models
Stars: ✭ 118 (-0.84%)
Mutual labels:  jupyter-notebook
Automunge
Artificial Learning, Intelligent Machines
Stars: ✭ 119 (+0%)
Mutual labels:  jupyter-notebook
Texture Synthesis Nonparametric Sampling
Implementation of "Texture Synthesis with Non-Parametric Sampling" paper by Alexei A. Efros and Thomas K. Leung
Stars: ✭ 119 (+0%)
Mutual labels:  jupyter-notebook
Ds salary proj
Repo for the data science salary prediction of the Data Science Project From Scratch video on my youtube
Stars: ✭ 116 (-2.52%)
Mutual labels:  jupyter-notebook

abstractive-text-summarization

This repository and notebook contains code for in-progress implementation/experiments on Abstractive Text Summarization using Sequence-to-sequence RNNs and Beyond paper.

Requirements

  1. Create conda environment

conda env create -f environment.yml --gpu

conda env create -f environment-cpu.yml --cpu

  1. Activate environment

source activate abs-sum -- gpu

source activate abs-sum-cpu -- cpu

  1. Install dependencies (PyTorch, Fastai, etc) via:

pip install -r requirements.txt

  1. Download spacy english module

python -m spacy download en

Dataset

The dataset used is a subset of the gigaword dataset and can be found here.

It contains 3,803,955 parallel source & target examples for training and 189,649 examples for validation.

After downloading, we created article-title pairs, saved in tabular datset format (.csv) and extracted a sample subset (80,000 for training & 20,000 for validation). This data preparation can be found here.

An example article-title pair looks like this:

source: the algerian cabinet chaired by president abdelaziz bouteflika on sunday adopted the #### finance bill predicated on an oil price of ## dollars a barrel and a growth rate of #.# percent , it was announced here .

target: algeria adopts #### finance bill with oil put at ## dollars a barrel

Experimenting on the complete dataset (3M) would take a really long time (also $$$$). So in order to train and experiment faster we use our sample subset of 80,000.

Current Features

  • model architecture supports LSTM & GRU (biLSTM-to-uniLSTM or biGRU-to-uniGRU)
  • implements batch data processing
  • implements attention mechanism (Bahdanau et al. & Luong et al.(global dot))
  • implements scheduled sampling (teacher forcing)
  • implements tied embeddings
  • initializes encoder-decoder with pretrained vectors (glove.6B.200d)
  • implements custom training callbacks (tensorboard visualization for PyTorch, save best model & log checkpoint)
  • implements attention plots

To-Do

  • Implement additional linguistic features embeddings
  • Implement generator-pointer switch and replace unknown words by selecting source token with the highest attention score.
  • Implement large vocabulary trick
  • Implement sentence level attention
  • Implement beam search during inference
  • implement rouge evaluation

Baseline Training & Validation Loss

alt text

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].