training-a-chatbot-using-weights-and-biases

This GitHub project trains a chatbot using the Cornell Movie-Dialogs Corpus with PyTorch. It covers dataset pre-processing, implementing a seq2seq model with Luong attention, training with mini-batches, greedy-search decoding, and chatbot interaction. A hands-on for understanding and applying NLP and weights and biases.

Chatbot Training with Cornell Movie Dialogs Corpus

This repository guides you through training a chatbot using the Cornell Movie Dialogs Corpus in a PyTorch environment, based on a detailed PyTorch Chatbot Tutorial. The project focuses on a sequence-to-sequence model with Luong attention mechanisms, leveraging the PyTorch framework for an end-to-end chatbot training experience.

Objective

• To understand and implement the sequence-to-sequence model architecture using PyTorch.
• To learn the process of data pre-processing, training, and interacting with a chatbot model.

Steps

1. Training the Model

   • Make a copy of the tutorial notebook and execute the training and evaluation steps in your local Google Colab environment.

2. Hyperparameter Optimization

   • Integrate Weights and Biases (W&B) for hyperparameter optimization.
   • Set up a hyperparameter sweep with W&B using Random Search strategy for learning rate, optimizer, clip, teacher forcing ratio, and decoder learning ratio.

References

• Cornell Movie Dialogs Corpus
• Weights and Biases Framework Video Tutorial
• Weights and Biases Framework Colab project
• PyTorch Profiler Example
• PyTorch TorchScript tutorial

Contributions

We welcome contributions and suggestions on how to improve this project. Feel free to fork the repository, make changes, and submit pull requests.

License

This project is licensed under the MIT License - see the LICENSE.md file for details.