lighthouse-scoring / data /artifacts /turn_calibration.json
AaravArora's picture
scoring service: FastAPI mounted at root, gradio at /ui
92d53f7 verified
Raw
History Blame Contribute Delete
1.92 kB
{
"temperature": 1.3422844409942627,
"bins": 15,
"val": {
"before": {
"ece": 0.062173962611503555,
"mce": 0.12007497716887894,
"brier": 0.2338956743478775,
"nll": 0.43018439412117004,
"mean_confidence": 0.9047172665596008,
"accuracy": 0.8458526532715095
},
"after": {
"ece": 0.01970834276345757,
"mce": 0.05131246692343272,
"brier": 0.22591689229011536,
"nll": 0.40334296226501465,
"mean_confidence": 0.8601455688476562,
"accuracy": 0.8458526532715095
}
},
"test": {
"before": {
"ece": 0.069850480866641,
"mce": 0.15291750606046906,
"brier": 0.2490195631980896,
"nll": 0.45989030599594116,
"mean_confidence": 0.9038535356521606,
"accuracy": 0.8340030911901082
},
"after": {
"ece": 0.02630870686493722,
"mce": 0.08176089324475322,
"brier": 0.23885364830493927,
"nll": 0.42578956484794617,
"mean_confidence": 0.8593263030052185,
"accuracy": 0.8340030911901082
}
},
"test_per_class_gap": {
"none": {
"mean_prob": 0.3464653193950653,
"base_rate": 0.377640391550747,
"gap": -0.031175072155681716
},
"distress": {
"mean_prob": 0.19325914978981018,
"base_rate": 0.18547140649149924,
"gap": 0.007787743298310945
},
"harassment": {
"mean_prob": 0.1901123970746994,
"base_rate": 0.19268418341061308,
"gap": -0.0025717863359136783
},
"identity_attack": {
"mean_prob": 0.06796493381261826,
"base_rate": 0.05151983513652756,
"gap": 0.016445098676090696
},
"threat": {
"mean_prob": 0.014295090921223164,
"base_rate": 0.0072127769191138585,
"gap": 0.007082314002109305
},
"self_harm": {
"mean_prob": 0.18790312111377716,
"base_rate": 0.18547140649149924,
"gap": 0.002431714622277925
}
}
}