This is re-fined dataset based on Stanford Sentiment Treebank used in Yoon Kim (2014).
stsa.fine.*: Stanford Sentiment Treebank—an extension of MR but with train/dev/test splits provided and fine-grained labels (very positive, positive, neutral, negative, very negative). It is called SST-1 in paper.stsa.binary.*: It splits to train/test/dev and re-labels to binary labels (1=positive, 0=negative) with neutral reviews removed. It is called SST-2 in paper.