Sentiment Analysis 벤치마크
Sentiment Analysis on MPQA
Accuracy
- 2018-03-29 — USE_T+DAN (w2v w.e.) : Accuracy 88.14
- 2018-05-14 — byte mLSTM7: Accuracy 88.8
- 2019-05-31 — STM+TSED+PT+2L: Accuracy 89.83
- 2021-04-29 — RoBERTa-large 355M + Entailment as Few-shot Learner: Accuracy 90.8
| Rank | Model | Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | RoBERTa-large 355M + Entailment as Few-shot Learner | 90.8 | Entailment as Few-Shot Learner | PaddlePaddle/PaddleNLP · sunyilgdx/prompts4keras · cactilab/hateguard | 2021 |
| 2 | STM+TSED+PT+2L | 89.83 | The Pupil Has Become the Master: Teacher-Student Model-Based Word Embedding Distillation with Ensemble Learning | bgshin/distill_demo | 2019 |
| 3 | byte mLSTM7 | 88.8 | A La Carte Embedding: Cheap but Effective Induction of Semantic Feature Vectors | NLPrinceton/ALaCarte | 2018 |
| 4 | USE_T+DAN (w2v w.e.) | 88.14 | Universal Sentence Encoder | facebookresearch/InferSent · facebookresearch/SentEval · textflint/textflint · +21 | 2018 |