Discriminative Improvements to Distributional Sentence Similarity

Open paper

Benchmark	Model	Rank	Results
semantic-textual-similarity-on-mrpc	TF-KLD	–	Accuracy: 80.4%F1: 85.9%