Trajectory quality correlation analysis on Teacher-generated Reasoning Datasets Qwen-2.5-3B student
0.85Spearman Correlation (Abs)Rank-Surprisal Ratio
Evaluation Results
| Method | Links | |
|---|---|---|
| Rank-Surprisal RatioMetric Category=Student-Specific2026.01 | 0.85 | |
| Rule-based QualityMetric Category=Student-Agnostic2026.01 | 0.75 | |
| Avg-Surpr_localMetric Category=Student-Specific2026.01 | 0.72 | |
| Avg-SurprisalMetric Category=Student-Specific2026.01 | 0.7 | |
| G-NormMetric Category=Student-Specific2026.01 | 0.7 | |
| GRACEMetric Category=Student-Specific2026.01 | 0.69 | |
| Avg-RankMetric Category=Student-Specific2026.01 | 0.62 | |
| Influence ScoreMetric Category=Student-Specific2026.01 | 0.59 | |
| Avg-Token LengthMetric Category=Student-Agnostic2026.01 | 0.47 | |
| LLM-judged QualityMetric Category=Student-Agnostic2026.01 | 0.4 | |
| Teacher ParamsMetric Category=Student-Agnostic2026.01 | 0.26 | |
| Verified AccuracyMetric Category=Student-Agnostic2026.01 | 0.1 | |
| Teacher PerformanceMetric Category=Student-Agnostic2026.01 | 0.03 |