anhassan / Quora-Question-Similarity-Using-BERT-Embeddings-with-Siamese-Networks-as-Feature-Extractors

Proposed a model architecture which learns to classify duplicate question pairs based on highly contextualized sentence representations. Achieved by using both individual and combined sentence embeddings based on Word2Vec, GloVe and FastText word embeddings with CNN(convolutional neural networks) and DNN(Dense neural network) Siamese networks an…
14Updated last year

Related projects: