Project Tasks Overview

1. Text Preprocessing:

   • Dialogsum Data: Preprocessing is complete and ready for model training.
   • Email Thread Dataset (Kaggle): Inconsistent and requires further cleaning before it can be used for training.

2. Vectorization: We will be employing different vectorization techniques using pretrained embeddings and fine-tuning where necessary:

   • GloVe: Both the use of pretrained vectors and further fine-tuning.
   • BERT: Starting with a pretrained model and fine-tuning on our specific datasets.
   • BART: Utilizing a pretrained model with additional fine-tuning.
   • T5: Potential exploration of T5 embeddings for vectorization.

3. Model Implementation: We have been instructed to implement models based on the following architectures:

   • Transformer-Based Solution: To leverage state-of-the-art capabilities.
   • LSTM/GRU: To compare performance with transformer-based models.
   • R-CNN: To explore the capabilities of convolutional approaches in sequence modeling.

4. Evaluation Metrics:

   • Use the ROUGE score to evaluate the models' performance initially.

Team Allocation

• Dipankar:
  • Embedding: GloVe
  • Models: R-CNN, Transformer

<<<<<<< HEAD

• Pinak:
  • Embedding: Bart

=======

93926ad (Moved things around. Created a corpus. Check data/processed.)

Next Steps for Team Members:

• Each member needs to choose an embedding technique and at least one model to implement.
• Coordinate who will explore LSTM/GRU and who will investigate T5 embeddings.
• Determine who will be responsible for fine-tuning BERT and BART.
• Create your own branches and work on them.
• Let everyone know of your choices on the group.