julia-1-swe / validation_metrics.json
niko0xdev's picture
zorch2 Colab best checkpoint (best_overall) + final eval r118/r119/r121
084bcea verified
Raw History Blame Contribute Delete
1.88 kB
{
"count": 3564,
"accuracy": 0.995230078563412,
"nll": 0.02104884277840532,
"ece": 0.0031197859783365185,
"per_decision": {
"role": {
"count": 1188,
"accuracy": 0.9949494949494949,
"nll": 0.023152108122646918,
"ece": 0.0036363091934409454
},
"action": {
"count": 1188,
"accuracy": 0.9932659932659933,
"nll": 0.026944733532553087,
"ece": 0.004396045850182216
},
"risk": {
"count": 1188,
"accuracy": 0.9974747474747475,
"nll": 0.013049686680015953,
"ece": 0.002284457286198937
}
},
"confidence_policy": [
{
"threshold": 0.5,
"coverage": 0.9988776655443322,
"accuracy": 0.9960674157303371,
"count": 3560
},
{
"threshold": 0.6,
"coverage": 0.9983164983164983,
"accuracy": 0.9963462619449128,
"count": 3558
},
{
"threshold": 0.7,
"coverage": 0.9977553310886644,
"accuracy": 0.9966254218222722,
"count": 3556
},
{
"threshold": 0.75,
"coverage": 0.9971941638608305,
"accuracy": 0.9971862689926843,
"count": 3554
},
{
"threshold": 0.8,
"coverage": 0.9966329966329966,
"accuracy": 0.9974662162162162,
"count": 3552
},
{
"threshold": 0.85,
"coverage": 0.9957912457912458,
"accuracy": 0.9980276134122288,
"count": 3549
},
{
"threshold": 0.9,
"coverage": 0.994668911335578,
"accuracy": 0.9983074753173484,
"count": 3545
},
{
"threshold": 0.95,
"coverage": 0.9943883277216611,
"accuracy": 0.9983069977426636,
"count": 3544
}
],
"task_count": 1188,
"task_exact_match": 0.9865319865319865,
"task_head_accuracy": {
"role": 0.9949494949494949,
"action": 0.9932659932659933,
"risk": 0.9974747474747475
}
}