CoolFace
Modelpublic

Cccccz/HY

sourceHugging Faceupdated 11d agoView on Hugging Face
0likes
summary.json64 linesDownload Raw Back to confidence_validation
1{2  "c1_validation_metrics": {3    "P1": {4      "log_mae": 0.17691564559936523,5      "model": "Confidence C1",6      "pearson": 0.8829622268676758,7      "spearman": 0.8590718873187204,8      "split": "validation",9      "step": "P1"10    },11    "P2": {12      "log_mae": 0.14503394067287445,13      "model": "Confidence C1",14      "pearson": 0.9074602127075195,15      "spearman": 0.8631690385373965,16      "split": "validation",17      "step": "P2"18    },19    "P3": {20      "log_mae": 0.15804709494113922,21      "model": "Confidence C1",22      "pearson": 0.8872363567352295,23      "spearman": 0.8700319930976419,24      "split": "validation",25      "step": "P3"26    },27    "overall": {28      "log_mae": 0.15999889373779297,29      "model": "Confidence C1",30      "pearson": 0.96539306640625,31      "spearman": 0.9529179491248462,32      "split": "validation",33      "step": "overall"34    }35  },36  "gain_vs_random": 0.48957529207015615,37  "oracle_gap": 0.007935705508866102,38  "protocol": "WorldPlay fulltrain100 (8400 transitions); independent validation25 (2100 transitions); no validation tuning",39  "remaining_risk_at_33pct": {40    "Confidence C0": 0.008430036716163158,41    "Confidence C1": 0.004878982435911894,42    "Confidence C2": 0.004925357177853584,43    "Oracle": 0.004840569104999304,44    "Random": 0.009558672141283751,45    "Timestep-only": 0.00500575313344597846  },47  "routing_at_33pct": {48    "P1": {49      "selection_fraction": 0.012857142857142857,50      "within_step_fallback_rate": 0.01285714285714285751    },52    "P2": {53      "selection_fraction": 0.05714285714285714,54      "within_step_fallback_rate": 0.0571428571428571455    },56    "P3": {57      "selection_fraction": 0.93,58      "within_step_fallback_rate": 0.9359    }60  },61  "train_samples": 8400,62  "validation_samples": 210063}64