Text Classification on AG (test)
0.0445Test Error RateXLNet
Evaluation results
| Method | Links | ||
|---|---|---|---|
| XLNetArchitecture=24-layer, Model size=Large2019.06 | 0.0445 | — | |
| Mixed VAT2019.06 | 0.0495 | — | |
| ULMFiT2018.01 | 0.0501 | — | |
| ULMFIT2019.06 | 0.0501 | — | |
| CNN2018.01 | 0.0657 | — | |
| CNN2019.06 | 0.0657 | — | |
| DPCNN2018.01 | 0.0687 | — | |
| DPCNN2019.06 | 0.0687 | — | |
| Char-level CNN2018.01 | 0.0951 | — | |
| BERT-FiTfine-tuning=true2018.10 | — | 0.9475 | |
| CCCapsNetfine-tuning=false, trained from scratch=true2018.10 | — | 0.9239 | |
| DPCNNfine-tuning=true2018.10 | — | 0.9313 | |
| GLCClassifier=Pre-trained BERT-base, Number of clean labels=400, Noise types=2, Noise levels=10, k (SGD steps)=52019.11 | — | 0.8388 | |
| MLCClassifier=Pre-trained BERT-base, Number of clean labels=400, Noise types=2, Noise levels=10, k (SGD steps)=52019.11 | — | 0.8527 | |
| MW-NetClassifier=Pre-trained BERT-base, Number of clean labels=400, Noise types=2, Noise levels=10, k (SGD steps)=52019.11 | — | 0.7591 | |
| Region Emb.fine-tuning=false, trained from scratch=true2018.10 | — | 0.928 | |
| ULMFiTfine-tuning=true2018.10 | — | 0.9499 | |
| XLNetfine-tuning=true2018.10 | — | 0.9555 |