Context-aware turn-taking on Friends
100I1Qwen3-4b-instruct
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Qwen3-4b-instructEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 100 | 90.7 | 38.33 | 6.03 | 43.13 | 36.82 | 51.48 | |
| Qwen3-4b-instructEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 100 | 90.7 | 38.33 | 6.03 | 43.13 | 36.82 | 51.48 | |
| Qwen3-8bEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 100 | 87.21 | 43.33 | 6.44 | 42.68 | 37 | 50.7 | |
| Qwen3-8bEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 100 | 87.21 | 43.33 | 6.44 | 42.68 | 37 | 50.7 | |
| gpt-oss-20bEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 98.06 | 81.4 | 58.33 | 21.62 | 49.89 | 48 | 55.92 | |
| gpt-oss-20bEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 98.06 | 81.4 | 58.33 | 21.62 | 49.89 | 48 | 55.92 | |
| LLaMA3.1-8b-instructEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 97.09 | 85.27 | 50 | 16.01 | 47.34 | 44.23 | 54.21 | |
| LLaMA3.1-8b-instructEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 97.09 | 85.27 | 50 | 16.01 | 47.34 | 44.23 | 54.21 | |
| gpt-5.2Evaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 94.17 | 84.88 | 60 | 18.71 | 49 | 46.63 | 55.41 | |
| gpt-5.2Evaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 94.17 | 84.88 | 60 | 18.71 | 49 | 46.63 | 55.41 | |
| Qwen2.5-7bEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 92.23 | 77.52 | 55 | 24.95 | 49.67 | 48.39 | 55 | |
| Qwen2.5-7bEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 92.23 | 77.52 | 55 | 24.95 | 49.67 | 48.39 | 55 | |
| Mistral-7b-instructEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 89.32 | 84.88 | 26.67 | 18.71 | 46.23 | 43.3 | 52.87 | |
| Mistral-7b-instructEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 89.32 | 84.88 | 26.67 | 18.71 | 46.23 | 43.3 | 52.87 | |
| gemini-3.1-proEvaluation Protocol=Zero-shot, System Prompt Frequency=Single2026.03 | 82.52 | 80.62 | 86.67 | 34.1 | 56.43 | 56.11 | 60.54 | |
| gemini-3.1-proEvaluation Protocol=Zero-shot, System Prompt Frequency=Repeated twice2026.03 | 82.52 | 80.62 | 86.67 | 34.1 | 56.43 | 56.11 | 60.54 |