Datasets:
File size: 2,646 Bytes
3e93229 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 | {
"schema_version": "v0.2",
"dataset_name": "eval-trap-stability-manifold-benchmark-v0.2",
"pretty_name": "Eval Trap Stability Manifold Benchmark v0.2",
"task_type": "binary_classification",
"benchmark_type": "multi_surface_stability_eval_trap",
"description": "Synthetic benchmark for exposing evaluation traps in which high-confidence predictions occur in mathematically unstable system states across multiple competing stability surfaces.",
"stability_manifold": {
"surfaces": {
"baseline_surface": "S1 = buffer - (pressure * coupling) - (k * lag)",
"coupling_surface": "S2 = buffer - (pressure * coupling^2) - (k * lag)",
"lag_surface": "S3 = buffer - (pressure * coupling) - (k * lag^2)"
},
"stability_rule": "label_stable = 1 if min(S1, S2, S3) >= 0 else 0"
},
"constants": {
"k": 1.0,
"boundary_eps": 0.02,
"high_content_score_threshold": 0.9
},
"columns": {
"scenario_id": {
"type": "string",
"description": "Unique scenario identifier"
},
"pressure": {
"type": "float",
"description": "External system load"
},
"buffer": {
"type": "float",
"description": "Remaining stabilizing capacity"
},
"lag": {
"type": "float",
"description": "Control response delay"
},
"coupling": {
"type": "float",
"description": "Subsystem coupling strength"
},
"content_score": {
"type": "float",
"description": "Primary model confidence or affinity score"
},
"regime_hint": {
"type": "string",
"allowed_values": ["baseline", "coupling", "lag"],
"description": "Human-readable hint for dominant instability regime"
},
"label_stable": {
"type": "integer",
"allowed_values": [0, 1],
"description": "Ground truth stability label where 1 = stable and 0 = unstable"
}
},
"prediction_contract": {
"required_columns": ["scenario_id", "prediction"],
"prediction_allowed_values": [0, 1],
"row_alignment": "scenario_id",
"scenario_id_must_be_unique": true
},
"splits": {
"train": "Training examples spanning all three instability regimes",
"in_domain_test": "Evaluation examples drawn from same distribution family as train",
"distribution_shift": "Evaluation examples with shifted parameter ranges",
"boundary_trap": "Near-manifold seam examples designed to trigger false rescue behaviour"
},
"diagnostic_regions": {
"collapse_zone": "min(S1, S2, S3) < -boundary_eps",
"near_boundary": "|min(S1, S2, S3)| <= boundary_eps",
"safe_margin": "min(S1, S2, S3) > boundary_eps"
}
} |