AmazonMLChallenge2021

This repo contains the solutions by team Panaroma for the Amazon ML challenge 2021. We managed to fetch an overall team rank of 16th among 3000+ teams. We have used multilingual BERT and did an end-to-end training for 3 epochs over the entire dataset after some preprocessing which involved removal of duplicates,handling missing values. We also tried XLM-Roberta for the given task. We managed to fetch 67.465 score on the public leaderboard through our approach. The link to the leaderboard https://www.hackerearth.com/challenges/competitive/amazon-ml-challenge/leaderboard/