Metaphor Detection on TroFi
89.7F1 ScoreLAG
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| LAGLLM=Claude 3.5 Sonnet2025.04 | 89.7 | 84.6 | — | — | |
| MetaPRO2025.04 | 79 | 70 | — | — | |
| TSI CMT*2025.04 | 66 | 66.8 | — | — | |
| Claude Sonnet 4.0Training Status=Fine-tuned on VUA-182026.04 | 65.1 | 60.2 | 52.6 | 85.2 | |
| Claude Sonnet 4.0Evaluation protocol=zero-shot, Evaluation target=target words2026.04 | 65.1 | — | — | — | |
| ConceptFrameMetTraining Status=Fine-tuned on VUA-182026.04 | 63.4 | 60.5 | 53.2 | 78.7 | |
| ConceptFrameMetEvaluation protocol=fine-tuned, Training dataset=VUA-18 (train), Evaluation target=target words2026.04 | 63.4 | — | — | — | |
| Gemini 2.5Training Status=Fine-tuned on VUA-182026.04 | 63.3 | 60.1 | 52.8 | 78.9 | |
| Gemini 2.5Evaluation protocol=zero-shot, Evaluation target=target words2026.04 | 63.3 | — | — | — | |
| MelBertTraining Status=Fine-tuned on VUA-182026.04 | 63.1 | 59.3 | 52.1 | 80.1 | |
| MelBertEvaluation protocol=fine-tuned, Training dataset=VUA-18 (train), Evaluation target=target words2026.04 | 63.1 | — | — | — | |
| FrameBertTraining Status=Fine-tuned on VUA-182026.04 | 62 | 59.5 | 52.4 | 76.1 | |
| FrameBertEvaluation protocol=fine-tuned, Training dataset=VUA-18 (train), Evaluation target=target words2026.04 | 62 | — | — | — | |
| Random baselineTraining Status=Fine-tuned on VUA-182026.04 | 46.6 | 50 | 43.6 | 50.1 | |
| Random baseline2026.04 | 46.6 | — | — | — |