Feedback Generation Evaluation on Expert Evaluation Feedback Quality
4.01AccuracyPsyScore
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PsyScoreModel=PsyScore (Full)2026.06 | 4.01 | 4.13 | 4.21 | 4.17 | 4.25 | |
| PsyScore-No-IRTAblation=No IRT layer2026.06 | 3.65 | 4.03 | 3.48 | 3.84 | 3.77 | |
| Qwen-3Model=Qwen-32026.06 | 3.13 | 2.91 | 2.91 | 2.7 | 3.11 | |
| GPT-4oModel=GPT-4o2026.06 | 3.1 | 2.88 | 2.79 | 2.55 | 3.15 | |
| Llama-4Model=Llama-42026.06 | 2.88 | 2.42 | 2.66 | 2.34 | 2.62 |