Laughter Reasoning on SMILE-Next
0.2427BLEU-4LLaMA3
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| LLaMA3Input Modality=LLM (Our approach)2026.05 | 0.2427 | 0.2328 | 0.4168 | 0.7828 | |
| Vicuna-v1.5Input Modality=LLM (Our approach)2026.05 | 0.2354 | 0.2311 | 0.4191 | 0.7907 | |
| Qwen2.5Input Modality=LLM (Our approach)2026.05 | 0.2303 | 0.2238 | 0.4097 | 0.7666 | |
| Video-LLaVAInput Modality=Visual LLM2026.05 | 0.1769 | 0.1818 | 0.3419 | 0.6567 | |
| Qwen2.5-Omni-7BInput Modality=Audio-Visual LLM2026.05 | 0.1683 | 0.1919 | 0.3549 | 0.7221 | |
| Qwen2.5-VLInput Modality=Visual LLM2026.05 | 0.1665 | 0.1764 | 0.3284 | 0.658 | |
| MiniCPM-o-v2.6Input Modality=Audio-Visual LLM2026.05 | 0.0802 | 0.1185 | 0.2632 | 0.5732 |