Diffusers
Safetensors
HY / confidence_validation /summary.json
Cccccz's picture
Upload batch 46: 223 files (7.45 GiB)
ba798d3 verified
Raw History Blame Contribute Delete
1.8 kB
{
"c1_validation_metrics": {
"P1": {
"log_mae": 0.17691564559936523,
"model": "Confidence C1",
"pearson": 0.8829622268676758,
"spearman": 0.8590718873187204,
"split": "validation",
"step": "P1"
},
"P2": {
"log_mae": 0.14503394067287445,
"model": "Confidence C1",
"pearson": 0.9074602127075195,
"spearman": 0.8631690385373965,
"split": "validation",
"step": "P2"
},
"P3": {
"log_mae": 0.15804709494113922,
"model": "Confidence C1",
"pearson": 0.8872363567352295,
"spearman": 0.8700319930976419,
"split": "validation",
"step": "P3"
},
"overall": {
"log_mae": 0.15999889373779297,
"model": "Confidence C1",
"pearson": 0.96539306640625,
"spearman": 0.9529179491248462,
"split": "validation",
"step": "overall"
}
},
"gain_vs_random": 0.48957529207015615,
"oracle_gap": 0.007935705508866102,
"protocol": "WorldPlay fulltrain100 (8400 transitions); independent validation25 (2100 transitions); no validation tuning",
"remaining_risk_at_33pct": {
"Confidence C0": 0.008430036716163158,
"Confidence C1": 0.004878982435911894,
"Confidence C2": 0.004925357177853584,
"Oracle": 0.004840569104999304,
"Random": 0.009558672141283751,
"Timestep-only": 0.005005753133445978
},
"routing_at_33pct": {
"P1": {
"selection_fraction": 0.012857142857142857,
"within_step_fallback_rate": 0.012857142857142857
},
"P2": {
"selection_fraction": 0.05714285714285714,
"within_step_fallback_rate": 0.05714285714285714
},
"P3": {
"selection_fraction": 0.93,
"within_step_fallback_rate": 0.93
}
},
"train_samples": 8400,
"validation_samples": 2100
}