The following repository is the codebase for our NLP Project on Natural Language Inference. We have implemented various models to classify the relationship between the premise and the hypothesis into one of three categories: entailment, contradiction, or neutral.

The code for the project is available on the following link: https://github.com/100smdok/NLP_Project_SNLI

The files are as follows:

• BERT.ipynb - Implements a BERT classification model on the SNLI dataset
• GRU.ipynb - Implements a GRU classification model on the SNLI dataset
• LSTM.ipynb - Implements an LSTM classification model on the SNLI dataset
• RNN_relu.ipynb - Implements an RNN classification model (activation is relu) on the SNLI dataset
• RNN_sigmoid.ipynb - Implements an RNN classification model (activation is sigmoid) on the SNLI dataset
• RNN_softmax.ipynb - Implements an RNN classification model (activation is softmax) on the SNLI dataset

The dataset is available at the following link: https://drive.google.com/drive/folders/1UIYb08D2zYVA56hM41S6y6HgixaTF9Fs?usp=sharing

The checkpoints are available at the following link: https://drive.google.com/drive/folders/10mPe7khOux173jTUttgIwyDNzgPw8tOq?usp=sharing

The final models are available at this link: https://drive.google.com/drive/folders/14uxKOzSoTwSg4l0-IuF_AIm1_HVPiiqD?usp=sharing

The glove embeddings produced from the SNLI training set are available at the following link: https://drive.google.com/drive/folders/1qGzP7nSpLlycGTuq-m5L4X6DQnqp7h1e?usp=sharing

Download the dataset from the above link and change the dataset's location in the following notebooks and hit RUN!!

Each notebook will produce checkpoints, final models, Classification Reports, and a Confusion Matrix Plot.

• Saurav Chhatani
• Balaji Patukala
• Parth Maradia