Motivational Interviewing Dialogue Generation on MindfulMI (test)
30.73BLEU-2MindfulMI_oracle
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| MindfulMI_oracleBackbone=GPT-4o2026.06 | 30.73 | 24.23 | 48.08 | 31.86 | 42.24 | 90.33 | 91.28 | — | |
| MindfulMI_oracleBackbone=LLaMA3-70B2026.06 | 26.6 | 19.52 | 43.99 | 25.77 | 35.88 | 89.2 | 90.71 | — | |
| MindfulMI_SFT+RLBackbone=GPT-4o2026.06 | 7.26 | 3.85 | 23.8 | 6.07 | 19.04 | 88.9 | 44.04 | 1.51 | |
| MindfulMI_SFTBackbone=GPT-4o2026.06 | 6.91 | 3.27 | 21.82 | 5.28 | 18.65 | 87.75 | 39.62 | 1.63 | |
| MindfulMI_SFT+RLBackbone=LLaMA3-70B2026.06 | 6.74 | 3.15 | 22.63 | 4.94 | 18.36 | 87.81 | 43.33 | 1.58 | |
| MindfulMI_SFTBackbone=LLaMA3-70B2026.06 | 6.23 | 3.03 | 21.78 | 4.52 | 17.88 | 86.11 | 38.91 | 1.66 | |
| CAMIBackbone=GPT-4o2026.06 | 5.4 | 2.18 | 22.59 | 3.87 | 17.74 | 86.15 | 30.88 | 1.92 | |
| CAMIBackbone=LLaMA3-70B2026.06 | 5.13 | 2.11 | 20.84 | 3.89 | 16.71 | 85.75 | 30.17 | 2.03 | |
| DIIRBackbone=GPT-4o2026.06 | 5.09 | 2.02 | 22.11 | 3.93 | 17.48 | 85.83 | 31.11 | 1.71 | |
| DIIRBackbone=LLaMA3-70B2026.06 | 5.08 | 1.92 | 19.67 | 4.03 | 16.45 | 84.28 | 30.39 | 1.87 | |
| CoSBackbone=GPT-4o2026.06 | 4.97 | 1.77 | 21.11 | 3.34 | 16.41 | 85.81 | 29.66 | 2.2 | |
| CoSBackbone=LLaMA3-70B2026.06 | 4.94 | 1.75 | 19.53 | 3.67 | 15.4 | 85.01 | 28.37 | 2.35 | |
| CoTBackbone=GPT-4o2026.06 | 4.8 | 1.63 | 20.95 | 3.18 | 15.55 | 84.7 | 30.24 | 2.47 | |
| CoTBackbone=LLaMA3-70B2026.06 | 4.57 | 1.56 | 20.4 | 3.05 | 15.33 | 84.87 | 29.21 | 2.63 | |
| BaseBackbone=GPT-4o2026.06 | 4.34 | 1.48 | 19.5 | 3.12 | 14.33 | 84.47 | 25.32 | 2.14 | |
| BaseBackbone=LLaMA3-70B2026.06 | 4.07 | 1.27 | 19.19 | 2.81 | 14.28 | 85.42 | 24.81 | 2.31 | |
| R1-DistBackbone=LLaMA3-70B2026.06 | 3.94 | 1.17 | 16.56 | 2.69 | 12.01 | 84.24 | 23.01 | 2.91 |