Detecting duplicate questions from Quora (R/Keras) Created mainly for muni.cz students (Bi8700)
Simple modification of an approach proposed at https://blogs.rstudio.com/tensorflow/posts/2018-01-09-keras-duplicate-questions-quora/
- contains an example of exploiting word embeddings (FastText), the model is based on siamese architecture.
Trained model can be downloaded from: https://www.dropbox.com/s/5ctq55d51ltqn2r/model-question-pairs-weights.tar.xz?dl=0