Seq2Seq-Workshop

Seq2Seq workshop materials

Presentation slides:

NLP Basics

Intro to Sequence to Sequence Learning

Dependencies

• Python 3.6
• Pytorch 0.4.1
• MeCab with neologd dictionary
  • Instructions for Mac
  • Instructions for Ubuntu

Installation

1. Create a conda environment:

   conda env create -n seq2seq python=3.6
   

2. Install dependencies

   pip install spacy
   pip install mecab-python3
   
   

3. Install Spacy en model:

   python -m spacy download en
   

4. Install Pytorch 0.4.1:

   pip install torch==0.4.1
   

Data

• English-Japanese Translation Corpus: (ftp://ftp.monash.edu/pub/nihongo/examples.utf.gz)
• COCO Image Captioning Dataset (2014 train/val set)