/WSDM_Spotify

4'th Place Approach to the 2019 WSDM Spotify Skip Prediction Challenge

Primary LanguageJupyter Notebook

ACM WSDM Spotify Skip Prediction Challenge 4th place solution: Utilizing Attention Mechanism for Sequential Skip Prediction [paper]

Team Name: learner6
Team Members: Zhaoyue Cheng, Maksims Volkovs

Environment and Requirements

The model is implemented and tested on the following Environment:

  • Intel(R) Xeon(R) CPU E5-2620 v4 @ 2.10GHz
  • 256GB RAM
  • Nvidia Titan Xp

With Requirements:

  • Python 3.6
  • Numpy
  • Pandas
  • Pytorch 0.4.1 with Cuda 9.0

Preprocessing and Executing

  • All the required code are uploaded, instructions to be updated soon
  • The network we used is in folder models/rnn_model, we code up multiple different models but only be able to test one of them due to time constraints, the one we are using for the leaderboard submission is RNNModelAtt1SeperateEncoder in file rnn_model which corresponds to the Network Architecture image shown below

Network Architecture

Image of Architecture