Validation timing detection on M-TESC English
58.41Overall Macro-F1MEGUMI
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MEGUMI2026.06 | 58.41 | 45.07 | 41.56 | 43.24 | |
| XLM-RoBERTa2026.06 | 54.83 | 39.94 | 52.09 | 45.21 | |
| GPT 4o NanoEvaluation Protocol=Zero-shot2026.06 | 53.74 | 39.29 | 60.8 | 47.73 | |
| GPT 4o NanoEvaluation Protocol=3-shot2026.06 | 53.7 | 39.24 | 60.57 | 47.62 | |
| BERT2026.06 | 53.29 | 38.29 | 41.35 | 39.76 | |
| ModernBERT2026.06 | 52.8 | 37.67 | 46.77 | 41.73 | |
| Llama 3.1 8bEvaluation Protocol=LoRA2026.06 | 50.18 | 33.33 | 31.17 | 32.21 | |
| Llama 3.1 8bEvaluation Protocol=Zero-shot2026.06 | 49.07 | 38.55 | 81.64 | 52.37 | |
| Random Baseline2026.06 | 48.84 | 34.29 | 50.27 | 40.76 | |
| GPT 4o NanoEvaluation Protocol=CoT2026.06 | 46.66 | 36.53 | 82.08 | 50.56 | |
| Llama 3.1 8bEvaluation Protocol=3-shot2026.06 | 34.15 | 35.41 | 95.93 | 51.73 |