A large annotated corpus for learning natural language inference

3.3 Sentence embeddings and NLI

1. How were neural sentence-embedding models evaluated on SNLI?
   Each sentence is encoded separately into a 100-d vector (no cross-sentence context); the two vectors are concatenated and fed to three 200-d tanh layers and a 3-way softmax, trained jointly. Word embeddings start from 300-d GloVe and are fine-tuned.
   • Bowman et al., 2015, A large annotated corpus for learning natural language inference (Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP)), §3.3, pp. 637–638
2. How did the neural sentence encoders compare on SNLI?
   Test accuracy: sum of word vectors 75.3%, plain RNN 72.2%, LSTM 77.6%, essentially equal to the lexicalized classifier (78.2%), with a much smaller train–test gap, suggesting higher-capacity neural models would pay off.
   • Bowman et al., 2015, A large annotated corpus for learning natural language inference (Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing (EMNLP)), §3.3, pp. 637–638