Rumor Detection on Algerian Dialect Arabic-script dataset (test)
84AccuracyMarBERT + SVM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MarBERT + SVMFamily=Hybrid, C=0.01, kernel=linear2026.06 | 84 | 84 | 84 | 84 | |
| MarBERT + LRFamily=Hybrid, C=1, solver=lbfgs2026.06 | 84 | 84 | 84 | 84 | |
| MarBERT + RFFamily=Hybrid, n_estimators=2002026.06 | 84 | 84 | 84 | 84 | |
| MarBERTFamily=Transformers, Epochs=8, Batch size=16, Learning rate=enc: 2e-5 / clf: 1e-4, max len=1282026.06 | 83 | 83 | 82 | 82 | |
| DziriBERT FTFamily=Transformers, Epochs=5, Batch size=8, Learning rate=enc: 2e-5 / clf: 1e-4, max len=1282026.06 | 82 | 82 | 82 | 82 | |
| GigaBERTFamily=Transformers, Epochs=8, Batch size=16, Learning rate=enc: 2e-5 / clf: 1e-4, max len=1282026.06 | 81 | 81 | 80 | 80 | |
| AraBERTFamily=Transformers, Epochs=8, Batch size=16, Learning rate=enc: 2e-5 / clf: 1e-4, max len=1282026.06 | 81 | 81 | 82 | 81 | |
| mBERTFamily=Transformers, Epochs=8, Batch size=32, Learning rate=enc: 1e-5 / clf: 5e-5, max len=1282026.06 | 79 | 79 | 79 | 79 | |
| SVM + TF-IDFFamily=Classical ML, C=12026.06 | 77 | 76 | 77 | 76 | |
| LR + TF-IDFFamily=Classical ML, C=5, solver=liblinear2026.06 | 77 | 76 | 77 | 76 | |
| RF + TF-IDFFamily=Classical ML, n_estimators=2002026.06 | 75 | 75 | 75 | 75 | |
| BiLSTM + FastTextFamily=Deep Learning, Epochs=20, Batch size=32, Learning rate=1e-3, dim=200, patience=42026.06 | 73 | 73 | 73 | 73 | |
| BiGRU + FastTextFamily=Deep Learning, Epochs=20, Batch size=32, Learning rate=1e-3, dim=200, patience=42026.06 | 73 | 73 | 73 | 73 | |
| AraVec + BiLSTMFamily=Deep Learning, Epochs=20, Batch size=32, Learning rate=1e-3, dim=300, patience=42026.06 | 72 | 73 | 72 | 72 |