Commentary Generation on Race Japanese
0.64AlignmentGPT-4.1
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4.1Decoding Strategy=Feedback (ICL)2026.03 | 0.64 | 0.28 | 2.6 | |
| GPT-4.1Decoding Strategy=Feedback2026.03 | 0.62 | 0.28 | 2.5 | |
| GPT-4.1Decoding Strategy=Realtime2026.03 | 0.62 | 0.28 | 2.5 | |
| Qwen2.5-VL-7B-InstructDecoding Strategy=stateless2026.03 | 0.6 | 0.13 | 2.5 | |
| LLaVA-NeXT-Video-7B-hfDecoding Strategy=stateless2026.03 | 0.51 | 0.14 | 0.1 | |
| Qwen2.5-VL-7B-InstructDecoding Strategy=Realtime2026.03 | 0.3 | 0.29 | 3 | |
| LLaVA-NeXT-Video-7B-hfDecoding Strategy=Feedback (ICL)2026.03 | 0.22 | 0.28 | 5.6 | |
| Qwen2.5-VL-7B-InstructDecoding Strategy=Feedback2026.03 | 0.21 | 0.29 | 4.4 | |
| LLaVA-NeXT-Video-7B-hfDecoding Strategy=Feedback2026.03 | 0.2 | 0.13 | 0.8 | |
| Qwen2.5-VL-7B-InstructDecoding Strategy=Feedback (ICL)2026.03 | 0.2 | 0.29 | 5.4 | |
| LLaVA-NeXT-Video-7B-hfDecoding Strategy=Realtime2026.03 | 0.19 | 0.23 | 0.5 | |
| GPT-4.1Decoding Strategy=stateless2026.03 | 0.15 | 0.25 | 7.1 |