GitHub - souridas/Question-Duplicates: • Built a Siamese Network that identifies similar questions using trax, trained on Quora question & answer dataset. • Tokenized each question using nltk and formed vocab, mapped tokens to word embeddings using trax. • Used Triplet loss to get corresponding cosine similarity for each pair of questions, achieved a test set accuracy of 0.69

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
C3_W4_Assignment.ipynb		C3_W4_Assignment.ipynb

About

• Built a Siamese Network that identifies similar questions using trax, trained on Quora question & answer dataset. • Tokenized each question using nltk and formed vocab, mapped tokens to word embeddings using trax. • Used Triplet loss to get corresponding cosine similarity for each pair of questions, achieved a test set accuracy of 0.69