Legal Document Classification on Bangladeshi legal document dataset (held-out evaluation set)
67.96AccuracyBiGRU + KAN
Evaluation results
| Method | Links | |
|---|---|---|
| BiGRU + KANCategory=Recurrent2026.05 | 67.96 | |
| Random ForestCategory=Classical ML2026.05 | 62 | |
| SVMCategory=Classical ML2026.05 | 62 | |
| Logistic RegressionCategory=Classical ML2026.05 | 59 | |
| KNNCategory=Classical ML2026.05 | 58 | |
| BiGRU (w/o KAN)Category=Recurrent2026.05 | 57.34 | |
| BiLSTM (w/o KAN)Category=Recurrent2026.05 | 51.88 | |
| Naive BayesCategory=Classical ML2026.05 | 48 | |
| LongformerCategory=PLMs2026.05 | 41.73 | |
| T5Category=PLMs2026.05 | 41.01 | |
| Legal-BERTCategory=PLMs2026.05 | 38.85 | |
| BERTCategory=PLMs2026.05 | 38.13 | |
| RoBERTaCategory=PLMs2026.05 | 37.41 |