All-but-the-Top: Simple and Effective Postprocessing for Word Representations

Benchmark Model Rank Results
sentiment-analysis-on-mrGRU-RNN-WORD2VEC#9Accuracy: 78.26
text-classification-on-trec-6GRU-RNN-GLOVE#12Error: 7.0
subjectivity-analysis-on-subjGRU-RNN-GLOVE#14Accuracy: 91.85
sentiment-analysis-on-sst-5-fine-grained-classificationGRU-RNN-WORD2VEC#21Accuracy: 45.02