tiny-jev-support-v03-rl / evaluation.json
sainath's picture
Upload 17 files
f412230 verified
Raw History Blame Contribute Delete
92.1 kB
{
"scope": "Exploratory existing splits; test has previously informed development",
"calibration": {
"global": 1.2589254117941675,
"by_question": {
"150440942f8fd3e04c6ffcb60c0215ffb1265e7d27e071f3ac453fd9817a85d4": 1.7378008287493754,
"2a95b662ebe5bb2aa612f55ab9490ec200c83aec6e0ce1e9c439d769cb39a93c": 0.1,
"330b97e5d96bf0d453555d62465367c41f1c5eb57ca53a1f3e8015c8ebe8548a": 0.1,
"59786fb640773138e83576a42b1baff87c3a12c6924070c867a17f7541f8fd50": 1.0
},
"minimum_examples": 50,
"unknown_question_policy": "global temperature; calibration not established"
},
"splits": {
"validation": {
"uncalibrated": {
"overall": {
"count": 800,
"accuracy": 0.93125,
"macro_f1": 0.9441014135596143,
"nll": 0.1488269731871437,
"brier": 0.0975886054619946,
"ece_10_bins": 0.017362924292683653,
"oracle_expected_nll": 0.14322728173785357,
"oracle_distribution_mse": 0.012396862257624472,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 14,
"confidence": 0.5606423233236585,
"accuracy": 0.42857142857142855
},
{
"lower": 0.6,
"upper": 0.7,
"count": 38,
"confidence": 0.6494223318601909,
"accuracy": 0.7105263157894737
},
{
"lower": 0.7,
"upper": 0.8,
"count": 48,
"confidence": 0.7586801064511141,
"accuracy": 0.6875
},
{
"lower": 0.8,
"upper": 0.9,
"count": 76,
"confidence": 0.8506408341621098,
"accuracy": 0.7763157894736842
},
{
"lower": 0.9,
"upper": 1.0,
"count": 624,
"confidence": 0.9925415913263956,
"accuracy": 0.9935897435897436
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 200,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 400,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 600,
"error_rate": 0.0033333333333332993
},
{
"coverage": 1.0,
"count": 800,
"error_rate": 0.06874999999999998
}
]
},
"by_task": {
"department": {
"count": 200,
"accuracy": 0.995,
"macro_f1": 0.9948307988219031,
"nll": 0.02237316982799771,
"brier": 0.010520559354480222,
"ece_10_bins": 0.009969844520092013,
"oracle_expected_nll": 0.02237316982799771,
"oracle_distribution_mse": 0.010520559354480222,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 1,
"confidence": 0.6584500074386597,
"accuracy": 1.0
},
{
"lower": 0.7,
"upper": 0.8,
"count": 1,
"confidence": 0.7848847508430481,
"accuracy": 1.0
},
{
"lower": 0.8,
"upper": 0.9,
"count": 5,
"confidence": 0.8563984274864197,
"accuracy": 0.8
},
{
"lower": 0.9,
"upper": 1.0,
"count": 193,
"confidence": 0.9940139299229637,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0050000000000000044
}
]
},
"escalation": {
"count": 200,
"accuracy": 0.73,
"macro_f1": 0.7282608695652174,
"nll": 0.5679232407765421,
"brier": 0.3794293531759783,
"ece_10_bins": 0.07154568254947667,
"oracle_expected_nll": 0.5455244749793817,
"oracle_distribution_mse": 0.0386623803584977,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 14,
"confidence": 0.5606423233236585,
"accuracy": 0.42857142857142855
},
{
"lower": 0.6,
"upper": 0.7,
"count": 37,
"confidence": 0.6491783406283405,
"accuracy": 0.7027027027027027
},
{
"lower": 0.7,
"upper": 0.8,
"count": 47,
"confidence": 0.7581225608257537,
"accuracy": 0.6808510638297872
},
{
"lower": 0.8,
"upper": 0.9,
"count": 71,
"confidence": 0.8502353698434965,
"accuracy": 0.7746478873239436
},
{
"lower": 0.9,
"upper": 1.0,
"count": 31,
"confidence": 0.9187506764165817,
"accuracy": 0.8709677419354839
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.14
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.18999999999999995
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.23333333333333328
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.27
}
]
},
"refund": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 9.63352862168806e-05,
"brier": 1.9762492451412852e-08,
"ece_10_bins": 9.633034467693147e-05,
"oracle_expected_nll": 9.63352862168806e-05,
"oracle_distribution_mse": 1.9762492451412852e-08,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 0.9999036696553231,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
},
"severity": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.004915146857817981,
"brier": 0.0004044895550275171,
"ece_10_bins": 0.00480370193719859,
"oracle_expected_nll": 0.004915146857817981,
"oracle_distribution_mse": 0.0004044895550275171,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 0.9951962980628014,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
}
}
},
"global": {
"overall": {
"count": 800,
"accuracy": 0.93125,
"macro_f1": 0.9441014135596143,
"nll": 0.1497719984221153,
"brier": 0.09681936358318226,
"ece_10_bins": 0.00701747298240663,
"oracle_expected_nll": 0.14532400513188456,
"oracle_distribution_mse": 0.01255885871902599,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 25,
"confidence": 0.565165548324585,
"accuracy": 0.52
},
{
"lower": 0.6,
"upper": 0.7,
"count": 42,
"confidence": 0.6500179114795867,
"accuracy": 0.6666666666666666
},
{
"lower": 0.7,
"upper": 0.8,
"count": 70,
"confidence": 0.7533530805792127,
"accuracy": 0.7571428571428571
},
{
"lower": 0.8,
"upper": 0.9,
"count": 80,
"confidence": 0.8482251234352589,
"accuracy": 0.85
},
{
"lower": 0.9,
"upper": 1.0,
"count": 583,
"confidence": 0.9942052873493672,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 200,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 400,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 600,
"error_rate": 0.0033333333333332993
},
{
"coverage": 1.0,
"count": 800,
"error_rate": 0.06874999999999998
}
]
},
"by_task": {
"department": {
"count": 200,
"accuracy": 0.995,
"macro_f1": 0.9948307988219031,
"nll": 0.029297618117778645,
"brier": 0.011847873858568251,
"ece_10_bins": 0.0152635979652405,
"oracle_expected_nll": 0.029297618117778645,
"oracle_distribution_mse": 0.011847873858568251,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 1,
"confidence": 0.6272245645523071,
"accuracy": 1.0
},
{
"lower": 0.7,
"upper": 0.8,
"count": 2,
"confidence": 0.7533551752567291,
"accuracy": 1.0
},
{
"lower": 0.8,
"upper": 0.9,
"count": 9,
"confidence": 0.839881075753106,
"accuracy": 0.8888888888888888
},
{
"lower": 0.9,
"upper": 1.0,
"count": 188,
"confidence": 0.9907149777133414,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0050000000000000044
}
]
},
"escalation": {
"count": 200,
"accuracy": 0.73,
"macro_f1": 0.7282608695652174,
"nll": 0.558092843042443,
"brier": 0.37415878347224946,
"ece_10_bins": 0.014070594906806932,
"oracle_expected_nll": 0.54030086988152,
"oracle_distribution_mse": 0.03711676401562454,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 25,
"confidence": 0.565165548324585,
"accuracy": 0.52
},
{
"lower": 0.6,
"upper": 0.7,
"count": 41,
"confidence": 0.6505738467704959,
"accuracy": 0.6585365853658537
},
{
"lower": 0.7,
"upper": 0.8,
"count": 68,
"confidence": 0.7533530189710504,
"accuracy": 0.75
},
{
"lower": 0.8,
"upper": 0.9,
"count": 65,
"confidence": 0.8466363714291499,
"accuracy": 0.8307692307692308
},
{
"lower": 0.9,
"upper": 1.0,
"count": 1,
"confidence": 0.9008614420890808,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.14
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.18999999999999995
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.23333333333333328
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.27
}
]
},
"refund": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.0006419544265103104,
"brier": 8.562553287209336e-07,
"ece_10_bins": 0.0006417402625084057,
"oracle_expected_nll": 0.0006419544265103104,
"oracle_distribution_mse": 8.562553287209336e-07,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 0.9993582597374916,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
},
"severity": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.01105557810172927,
"brier": 0.0012699407465824586,
"ece_10_bins": 0.010687653124332395,
"oracle_expected_nll": 0.01105557810172927,
"oracle_distribution_mse": 0.0012699407465824586,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 6,
"confidence": 0.8779526750246683,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 194,
"confidence": 0.9927564604380696,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
}
}
},
"per_question": {
"overall": {
"count": 800,
"accuracy": 0.93125,
"macro_f1": 0.9441014135596143,
"nll": 0.1545261240547399,
"brier": 0.09900029128593654,
"ece_10_bins": 0.015734072476625435,
"oracle_expected_nll": 0.14892643260544974,
"oracle_distribution_mse": 0.013808548081566407,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 15,
"confidence": 0.5626973231633504,
"accuracy": 0.4666666666666667
},
{
"lower": 0.6,
"upper": 0.7,
"count": 38,
"confidence": 0.6498527918991289,
"accuracy": 0.7105263157894737
},
{
"lower": 0.7,
"upper": 0.8,
"count": 56,
"confidence": 0.7574487828782627,
"accuracy": 0.7142857142857143
},
{
"lower": 0.8,
"upper": 0.9,
"count": 81,
"confidence": 0.8516954691321762,
"accuracy": 0.8024691358024691
},
{
"lower": 0.9,
"upper": 1.0,
"count": 610,
"confidence": 0.9894479682211016,
"accuracy": 0.9934426229508196
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 200,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 400,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 600,
"error_rate": 0.00666666666666671
},
{
"coverage": 1.0,
"count": 800,
"error_rate": 0.06874999999999998
}
]
},
"by_task": {
"department": {
"count": 200,
"accuracy": 0.995,
"macro_f1": 0.9948307988219031,
"nll": 0.05018125544241732,
"brier": 0.016571811967767927,
"ece_10_bins": 0.03622871160507202,
"oracle_expected_nll": 0.05018125544241732,
"oracle_distribution_mse": 0.016571811967767927,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 1,
"confidence": 0.5914673209190369,
"accuracy": 1.0
},
{
"lower": 0.6,
"upper": 0.7,
"count": 1,
"confidence": 0.6748074889183044,
"accuracy": 1.0
},
{
"lower": 0.7,
"upper": 0.8,
"count": 9,
"confidence": 0.7539301647080315,
"accuracy": 0.8888888888888888
},
{
"lower": 0.8,
"upper": 0.9,
"count": 10,
"confidence": 0.8620621740818024,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 179,
"confidence": 0.9781116739997651,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0050000000000000044
}
]
},
"escalation": {
"count": 200,
"accuracy": 0.73,
"macro_f1": 0.7282608695652174,
"nll": 0.5679232407765421,
"brier": 0.3794293531759783,
"ece_10_bins": 0.07154568254947667,
"oracle_expected_nll": 0.5455244749793817,
"oracle_distribution_mse": 0.0386623803584977,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 14,
"confidence": 0.5606423233236585,
"accuracy": 0.42857142857142855
},
{
"lower": 0.6,
"upper": 0.7,
"count": 37,
"confidence": 0.6491783406283405,
"accuracy": 0.7027027027027027
},
{
"lower": 0.7,
"upper": 0.8,
"count": 47,
"confidence": 0.7581225608257537,
"accuracy": 0.6808510638297872
},
{
"lower": 0.8,
"upper": 0.9,
"count": 71,
"confidence": 0.8502353698434965,
"accuracy": 0.7746478873239436
},
{
"lower": 0.9,
"upper": 1.0,
"count": 31,
"confidence": 0.9187506764165817,
"accuracy": 0.8709677419354839
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.14
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.18999999999999995
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.23333333333333328
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.27
}
]
},
"refund": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.0,
"brier": 4.3804362436076516e-76,
"ece_10_bins": 0.0,
"oracle_expected_nll": 0.0,
"oracle_distribution_mse": 4.3804362436076516e-76,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 1.0,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
},
"severity": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.0,
"brier": 7.613875406184365e-24,
"ece_10_bins": 0.0,
"oracle_expected_nll": 0.0,
"oracle_distribution_mse": 7.613875406184365e-24,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 1.0,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
}
}
}
},
"test": {
"uncalibrated": {
"overall": {
"count": 800,
"accuracy": 0.925,
"macro_f1": 0.9392552790275188,
"nll": 0.15805856801203613,
"brier": 0.10519841661327942,
"ece_10_bins": 0.025654533132910683,
"oracle_expected_nll": 0.138174928971103,
"oracle_distribution_mse": 0.01090862890356645,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 17,
"confidence": 0.5612775718464571,
"accuracy": 0.23529411764705882
},
{
"lower": 0.6,
"upper": 0.7,
"count": 16,
"confidence": 0.6538418717682362,
"accuracy": 0.75
},
{
"lower": 0.7,
"upper": 0.8,
"count": 64,
"confidence": 0.7603475973010063,
"accuracy": 0.765625
},
{
"lower": 0.8,
"upper": 0.9,
"count": 75,
"confidence": 0.8518771576881409,
"accuracy": 0.68
},
{
"lower": 0.9,
"upper": 1.0,
"count": 628,
"confidence": 0.9939726705953573,
"accuracy": 0.9936305732484076
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 200,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 400,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 600,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 800,
"error_rate": 0.07499999999999996
}
]
},
"by_task": {
"department": {
"count": 200,
"accuracy": 0.995,
"macro_f1": 0.9944254755766148,
"nll": 0.013614306728828655,
"brier": 0.006972022766819956,
"ece_10_bins": 0.00483260512351995,
"oracle_expected_nll": 0.013614306728828655,
"oracle_distribution_mse": 0.006972022766819956,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 3,
"confidence": 0.6343480944633484,
"accuracy": 0.6666666666666666
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 3,
"confidence": 0.8595178723335266,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 194,
"confidence": 0.9976901086335329,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0050000000000000044
}
]
},
"escalation": {
"count": 200,
"accuracy": 0.705,
"macro_f1": 0.7046381817726715,
"nll": 0.6145763126252539,
"brier": 0.4129853920524647,
"ece_10_bins": 0.11028719127178192,
"oracle_expected_nll": 0.5350417564615215,
"oracle_distribution_mse": 0.03582624121361282,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 17,
"confidence": 0.5612775718464571,
"accuracy": 0.23529411764705882
},
{
"lower": 0.6,
"upper": 0.7,
"count": 13,
"confidence": 0.6583404357616718,
"accuracy": 0.7692307692307693
},
{
"lower": 0.7,
"upper": 0.8,
"count": 64,
"confidence": 0.7603475973010063,
"accuracy": 0.765625
},
{
"lower": 0.8,
"upper": 0.9,
"count": 69,
"confidence": 0.8522450759790946,
"accuracy": 0.6521739130434783
},
{
"lower": 0.9,
"upper": 1.0,
"count": 37,
"confidence": 0.9170670589885196,
"accuracy": 0.8918918918918919
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.19999999999999996
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.26
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.2733333333333333
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.29500000000000004
}
]
},
"refund": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 9.923051486518732e-05,
"brier": 2.306917757809561e-08,
"ece_10_bins": 9.922474622725552e-05,
"oracle_expected_nll": 9.923051486518732e-05,
"oracle_distribution_mse": 2.306917757809561e-08,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 0.9999007752537727,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
},
"severity": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.003944422179196725,
"brier": 0.000836228564655437,
"ece_10_bins": 0.00370678603649136,
"oracle_expected_nll": 0.003944422179196725,
"oracle_distribution_mse": 0.000836228564655437,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 3,
"confidence": 0.8357743223508199,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 197,
"confidence": 0.9987376640895902,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
}
}
},
"global": {
"overall": {
"count": 800,
"accuracy": 0.925,
"macro_f1": 0.9392552790275188,
"nll": 0.15451136706332724,
"brier": 0.10279470684407294,
"ece_10_bins": 0.02014885999262334,
"oracle_expected_nll": 0.1387172312235814,
"oracle_distribution_mse": 0.010962450171169045,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 22,
"confidence": 0.5578833384947344,
"accuracy": 0.36363636363636365
},
{
"lower": 0.6,
"upper": 0.7,
"count": 24,
"confidence": 0.6626042500138283,
"accuracy": 0.7916666666666666
},
{
"lower": 0.7,
"upper": 0.8,
"count": 86,
"confidence": 0.7429546443528907,
"accuracy": 0.7441860465116279
},
{
"lower": 0.8,
"upper": 0.9,
"count": 79,
"confidence": 0.8482364740552781,
"accuracy": 0.759493670886076
},
{
"lower": 0.9,
"upper": 1.0,
"count": 589,
"confidence": 0.9972299244286451,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 200,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 400,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 600,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 800,
"error_rate": 0.07499999999999996
}
]
},
"by_task": {
"department": {
"count": 200,
"accuracy": 0.995,
"macro_f1": 0.9944254755766148,
"nll": 0.017469275119756628,
"brier": 0.008068493609724506,
"ece_10_bins": 0.010246815681457555,
"oracle_expected_nll": 0.017469275119756628,
"oracle_distribution_mse": 0.008068493609724506,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 2,
"confidence": 0.5836484432220459,
"accuracy": 0.5
},
{
"lower": 0.6,
"upper": 0.7,
"count": 1,
"confidence": 0.6414172649383545,
"accuracy": 1.0
},
{
"lower": 0.7,
"upper": 0.8,
"count": 1,
"confidence": 0.7566628456115723,
"accuracy": 1.0
},
{
"lower": 0.8,
"upper": 0.9,
"count": 4,
"confidence": 0.8481354713439941,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 192,
"confidence": 0.9964964153865973,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0050000000000000044
}
]
},
"escalation": {
"count": 200,
"accuracy": 0.705,
"macro_f1": 0.7046381817726715,
"nll": 0.5920632934974356,
"brier": 0.4015789899739018,
"ece_10_bins": 0.07626515358686448,
"oracle_expected_nll": 0.5288867501384523,
"oracle_distribution_mse": 0.03424996328228628,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 20,
"confidence": 0.5553068280220032,
"accuracy": 0.35
},
{
"lower": 0.6,
"upper": 0.7,
"count": 23,
"confidence": 0.6635254232779794,
"accuracy": 0.782608695652174
},
{
"lower": 0.7,
"upper": 0.8,
"count": 83,
"confidence": 0.7421774620033172,
"accuracy": 0.7349397590361446
},
{
"lower": 0.8,
"upper": 0.9,
"count": 74,
"confidence": 0.848746615487176,
"accuracy": 0.7432432432432432
},
{
"lower": 0.9,
"upper": 1.0,
"count": 0,
"confidence": null,
"accuracy": null
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.19999999999999996
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.26
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.2733333333333333
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.29500000000000004
}
]
},
"refund": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.0006544187157477001,
"brier": 9.303064614080757e-07,
"ece_10_bins": 0.0006541860103607133,
"oracle_expected_nll": 0.0006544187157477001,
"oracle_distribution_mse": 9.303064614080757e-07,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 0.9993458139896393,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
},
"severity": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.007858480920368994,
"brier": 0.0015304134862039825,
"ece_10_bins": 0.007402259111404373,
"oracle_expected_nll": 0.007858480920368994,
"oracle_distribution_mse": 0.0015304134862039825,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 2,
"confidence": 0.7683536112308502,
"accuracy": 1.0
},
{
"lower": 0.8,
"upper": 0.9,
"count": 1,
"confidence": 0.8108900189399719,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 197,
"confidence": 0.9957967052604947,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
}
}
},
"per_question": {
"overall": {
"count": 800,
"accuracy": 0.925,
"macro_f1": 0.9392552790275188,
"nll": 0.16079165909777055,
"brier": 0.10593323494171354,
"ece_10_bins": 0.02878171861171729,
"oracle_expected_nll": 0.14090802005683742,
"oracle_distribution_mse": 0.01164344723200057,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 19,
"confidence": 0.5593467480257938,
"accuracy": 0.2631578947368421
},
{
"lower": 0.6,
"upper": 0.7,
"count": 15,
"confidence": 0.6569308956464132,
"accuracy": 0.8
},
{
"lower": 0.7,
"upper": 0.8,
"count": 68,
"confidence": 0.7610192018396714,
"accuracy": 0.7794117647058824
},
{
"lower": 0.8,
"upper": 0.9,
"count": 73,
"confidence": 0.8525561878125961,
"accuracy": 0.6712328767123288
},
{
"lower": 0.9,
"upper": 1.0,
"count": 625,
"confidence": 0.9923768733978271,
"accuracy": 0.9936
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 200,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 400,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 600,
"error_rate": 0.0016666666666667052
},
{
"coverage": 1.0,
"count": 800,
"error_rate": 0.07499999999999996
}
]
},
"by_task": {
"department": {
"count": 200,
"accuracy": 0.995,
"macro_f1": 0.9944254755766148,
"nll": 0.02859032138164205,
"brier": 0.010747547714388071,
"ece_10_bins": 0.01983757793903353,
"oracle_expected_nll": 0.02859032138164205,
"oracle_distribution_mse": 0.010747547714388071,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 2,
"confidence": 0.5429347455501556,
"accuracy": 0.5
},
{
"lower": 0.6,
"upper": 0.7,
"count": 2,
"confidence": 0.6477688848972321,
"accuracy": 1.0
},
{
"lower": 0.7,
"upper": 0.8,
"count": 4,
"confidence": 0.771764874458313,
"accuracy": 1.0
},
{
"lower": 0.8,
"upper": 0.9,
"count": 4,
"confidence": 0.8579228669404984,
"accuracy": 1.0
},
{
"lower": 0.9,
"upper": 1.0,
"count": 188,
"confidence": 0.990979070042042,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0050000000000000044
}
]
},
"escalation": {
"count": 200,
"accuracy": 0.705,
"macro_f1": 0.7046381817726715,
"nll": 0.6145763126252539,
"brier": 0.4129853920524647,
"ece_10_bins": 0.11028719127178192,
"oracle_expected_nll": 0.5350417564615215,
"oracle_distribution_mse": 0.03582624121361282,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 17,
"confidence": 0.5612775718464571,
"accuracy": 0.23529411764705882
},
{
"lower": 0.6,
"upper": 0.7,
"count": 13,
"confidence": 0.6583404357616718,
"accuracy": 0.7692307692307693
},
{
"lower": 0.7,
"upper": 0.8,
"count": 64,
"confidence": 0.7603475973010063,
"accuracy": 0.765625
},
{
"lower": 0.8,
"upper": 0.9,
"count": 69,
"confidence": 0.8522450759790946,
"accuracy": 0.6521739130434783
},
{
"lower": 0.9,
"upper": 1.0,
"count": 37,
"confidence": 0.9170670589885196,
"accuracy": 0.8918918918918919
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.19999999999999996
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.26
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.2733333333333333
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.29500000000000004
}
]
},
"refund": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 0.0,
"brier": 6.297547947730489e-70,
"ece_10_bins": 0.0,
"oracle_expected_nll": 0.0,
"oracle_distribution_mse": 6.297547947730489e-70,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 1.0,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
},
"severity": {
"count": 200,
"accuracy": 1.0,
"macro_f1": 1.0,
"nll": 2.384186146287072e-09,
"brier": 1.3848676160326649e-15,
"ece_10_bins": 2.38418573772492e-09,
"oracle_expected_nll": 2.384186146287072e-09,
"oracle_distribution_mse": 1.3848676160326649e-15,
"reliability": [
{
"lower": 0.0,
"upper": 0.1,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.1,
"upper": 0.2,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.2,
"upper": 0.3,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.3,
"upper": 0.4,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.4,
"upper": 0.5,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.5,
"upper": 0.6,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.6,
"upper": 0.7,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.7,
"upper": 0.8,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.8,
"upper": 0.9,
"count": 0,
"confidence": null,
"accuracy": null
},
{
"lower": 0.9,
"upper": 1.0,
"count": 200,
"confidence": 0.9999999976158143,
"accuracy": 1.0
}
],
"risk_coverage": [
{
"coverage": 0.25,
"count": 50,
"error_rate": 0.0
},
{
"coverage": 0.5,
"count": 100,
"error_rate": 0.0
},
{
"coverage": 0.75,
"count": 150,
"error_rate": 0.0
},
{
"coverage": 1.0,
"count": 200,
"error_rate": 0.0
}
]
}
}
}
}
}
}