udrasht/neural-pos-tagger

This repository includes an implementation of a neural sequence model, such as FFNN or LSTM, designed to tag words in sentences with the correct part-of-speech (POS) tags.

Neural-POS-Tagger

Explanation:

• 2022201020_assignment2_report.pdf: This file contains the report for the assignment.
• INLP_Assignment.pdf: This file contains problem statement.
• README.md: This file provides instructions and explanations for running the code.
• UD_English-Atis: This directory contains the dataset files in CoNLL-U format.
• pos_tagger.py: The main Python script implementing the POS tagger using FFNN and LSTM models.
• trained_model: This folder contain .pt file for lstm and ffnn model.
• requirements.txt: File listing the Python dependencies required to run the code.
• Notebook_files: folder contain Notebook(.ipynb) file for both the models(FFNN, LSTM(RNN)) with the help of these file i create pos_tagger.py file .

Requirements

• Python 3
• PyTorch
• scikit-learn
• pandas
• seaborn

Usage

1. Install the required dependencies:

   $ pip install -r requirements.txt
   

2. Execute the pos_tagger.py script:

   $ python pos_tagger.py <model_type>
   

   Replace <model_type> with -f for FFNN or -r for LSTM.

   Example:

• To run the POS tagger with the Feed Forward Neural Network (FFNN) model, use the following command:

  $ python pos_tagger.py -f
  

  This command will prompt you to enter a sentence for POS tagging.

• To run the POS tagger with the Recurrent Neural Network (RNN) model, use the following command:

  $ python pos_tagger.py -r
  

  This command will prompt you to enter a sentence for POS tagging.

3. Follow the prompts to input sentences for POS tagging. Press q exit the program.

Notes

• Ensure that the dataset files and the saved model file (lstm_model1.pt,ffnn_model_best.pt) are placed in the specified directories as mentioned in the directory structure.
• The model type (-f for FFNN, -r for LSTM) must be specified as a command-line argument.
• The paths to the dataset files and the saved model file are hardcoded in the script and should be adjusted if necessary.