Topic Shift Detection on Topic Shift Dataset ΔT > 512 v1.0 (test)
70.3PrecisionNonlinear Naive Bayes With Attention
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Nonlinear Naive Bayes With AttentionMechanism=Attention-based NLU, Token Gap Range=ΔT > 5122026.02 | 70.3 | 98 | 78.3 | 81.9 | |
| NSPModel Type=BERT-based NSP, Token Gap Range=ΔT > 5122026.02 | 58.8 | 91.7 | 63.7 | 71.7 |