This is CS6910 course assignment-3 at IIT Madras. Here you will find detailed information about the assignment. In this assignment, I created a Recurrent Neural Network to transliterate a word from English to Bengali (how we type while chatting with our friends on WhatsApp etc). Different types of cells such as vanilla RNN, LSTM, GRU have been implemented to improve the accuracy of the model. Along with that, an attention network has been added in the model to further increase the accuracy of the model.

The model has been trained by using Aksharantar dataset released by AI4Bharat.

Here is the detailed wandb report for this assignment.

In this assignment, I used these libraies:

import torch
import random
import pandas as pd
import torch.nn as nn
from torch import optim
from torch.utils.data import TensorDataset
from torch.utils.data import DataLoader
from torch.nn.utils.rnn import pad_sequence
import argparse

If you don't have these packages installed then use this command to install:

pip install pytorch
pip install pandas

If you want to train and use this model for your dataset then download the train.py file. Hyperparameters of the model, train and test dataset path can be mentioned using command line arguments. Here are the supported command line arguments: