https://github.com/tmulc18/seq2seq

Create a sequence to sequence model that translate English words into pig Latin
https://github.com/tmulc18/seq2seq

Last synced: about 2 months ago
JSON representation

Create a sequence to sequence model that translate English words into pig Latin

• Host: GitHub
• URL: https://github.com/tmulc18/seq2seq
• Owner: tmulc18
• Created: 2017-03-27T03:42:31.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2017-10-12T06:25:37.000Z (over 7 years ago)
• Last Synced: 2024-10-29T21:10:09.933Z (3 months ago)
• Language: Jupyter Notebook
• Size: 355 KB
• Stars: 0
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Seq2Seq

Create a sequence to sequence model that translate English words into pig Latin 

## Download parameters

https://drive.google.com/open?id=0B3NC6C-8OM5BaV9vRUpjcl9WZmc

## How to use

Seq2Seq 3Layers.ipynb is used to train the encoder and decoder

Seq2Seq 3Layers (inference).ipynb will use the parameters stored in params.p to run inference on a given input sequence

## Implementation details

Followed the famous Sequence to Sequence paper but this implemenation was only 3 LSTM cells deep each with only 500 nodes