Natural Language Inference on MultiNLI matched (in-domain) RepEval 2017 (test)
76.8AccuracyOur ESIM
Evaluation Results
| Method | Links | |
|---|---|---|
| Our ESIMimplementation=authors2017.08 | 76.8 | |
| Ensembledensemble=true, RepEval 2017 participant=true2017.08 | 74.9 | |
| YixinNie-UNC-NLPRepEval 2017 participant=true2017.08 | 74.5 | |
| Single (Output Gate)gate_type=output2017.08 | 73.7 | |
| Singleensemble=false, RepEval 2017 participant=true2017.08 | 73.5 | |
| Single (Input Gate)gate_type=input, RepEval 2017 participant=true2017.08 | 73.5 | |
| Single (Forget Gate)gate_type=forget2017.08 | 72.9 | |
| Single - CharCNNcharacter_composition=removed2017.08 | 72.9 | |
| Single - Gated-Attgated_attention=removed2017.08 | 72.8 | |
| ESIM2017.08 | 72.3 | |
| RivercornersRepEval 2017 participant=true, ensemble=true2017.08 | 72.2 | |
| RivercornersRepEval 2017 participant=true, ensemble=false2017.08 | 72.1 | |
| LCT-MALTARepEval 2017 participant=true2017.08 | 70.7 | |
| Single - AbsDiff/Productelementwise_vector_comparison=removed2017.08 | 69.7 | |
| TALP-UPCRepEval 2017 participant=true2017.08 | 67.9 | |
| BiLSTM2017.08 | 66.9 | |
| Single - Word Embeddingword_level_embedding=removed2017.08 | 65.6 | |
| CBOW2017.08 | 64.8 |