Skip to content

This is the solution for Quora Question Pairs problem on Kaggle. It is implemented using a Siamese Network.

Notifications You must be signed in to change notification settings

aditya1601/Quora_Question_Pairs

Repository files navigation

Quora Question Pairs

This is my submission for the Quora Question Pairs problem on Kaggle. Link to the problem can be found here.

Instructions

  • ONLY Download the Jupyter Notebook from the repository.(If you want to clone, you need to use Git LFS, because "submission.csv" > 50MB)
  • Place it in a working directory
  • In the working directory, extract the train.csv and test.csv files which can be downloaded from here.
  • Open the jupyter notebook. (It is recommended to use a GPU for training).
  • pip install everything which is in the import libraries section.
  • Run all the cells.

Architecture used

A Siamese Network is used to find out the similarity between questions. The feature vector for each question is calculated using gensim's Doc2Vec implementation.

The Siamese Network

Siamese Network This is a Siamese Network. The architecture of the base network is shown below.

The Base Network (Residual NN)

Base Network (Residual)


About

This is the solution for Quora Question Pairs problem on Kaggle. It is implemented using a Siamese Network.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published