A large annotated corpus for learning natural language inference

Benchmark Model Rank Results
natural-language-inference-on-snli+ Unigram and bigram features#66% Test Accuracy: 78.2% Train Accuracy: 99.7
natural-language-inference-on-snli100D LSTM encoders#67% Test Accuracy: 77.6% Train Accuracy: 84.8Parameters: 220k
natural-language-inference-on-snliUnlexicalized features#68% Test Accuracy: 50.4% Train Accuracy: 49.4