qev-450m-mlx / reports /evaluation.json
twainsk's picture
Publish Qev v0.4.0 qev-450m with unchanged LFM2.5 foundation
2aa0473 verified
Raw History Blame Contribute Delete
15.9 kB
{
"label": "vl-a",
"backbone": "lfm2-vl",
"attention": "causal",
"calibration": {
"raw": {
"n": 1120,
"accuracy": 0.8848214285714285,
"nll": 0.8286656505531844,
"brier": 0.20143608167937646,
"ece": 0.09297216033922746,
"confident_error_rate": 0.07678571428571429,
"score_mae": 0.4503401108564636
},
"calibrated": {
"n": 1120,
"accuracy": 0.8848214285714285,
"nll": 0.3321787220851725,
"brier": 0.1661111770689924,
"ece": 0.02083344159724505,
"confident_error_rate": 0.015178571428571428,
"score_mae": 0.511491432084392
},
"temperature": 4.5947934199881395,
"by_source": {
"agnews": {
"n": 40,
"accuracy": 0.85,
"nll": 0.4613964179525684,
"brier": 0.23150962591792573,
"ece": 0.09537901553524014,
"confident_error_rate": 0.075
},
"agnews_yn": {
"n": 80,
"accuracy": 0.9375,
"nll": 0.18578394333721132,
"brier": 0.0901551691310664,
"ece": 0.030280076568509166,
"confident_error_rate": 0.025
},
"amazon": {
"n": 40,
"accuracy": 0.55,
"nll": 1.063923220607017,
"brier": 0.5988881204390519,
"ece": 0.14293638736581393,
"confident_error_rate": 0.025,
"score_mae": 0.5729569605099538
},
"banking77": {
"n": 40,
"accuracy": 0.675,
"nll": 1.5402265483550865,
"brier": 0.5359581656242952,
"ece": 0.2201891612204594,
"confident_error_rate": 0.0
},
"boolq": {
"n": 40,
"accuracy": 0.775,
"nll": 0.4769649130344238,
"brier": 0.31620540680281184,
"ece": 0.09690336717231163,
"confident_error_rate": 0.025
},
"dbpedia14": {
"n": 40,
"accuracy": 1.0,
"nll": 0.12480092447696775,
"brier": 0.048435455765173344,
"ece": 0.06948473292062886,
"confident_error_rate": 0.0
},
"imdb": {
"n": 40,
"accuracy": 0.875,
"nll": 0.2765451001786277,
"brier": 0.1442027427269653,
"ece": 0.0644099496689776,
"confident_error_rate": 0.025
},
"mnli": {
"n": 40,
"accuracy": 0.75,
"nll": 0.5184540211783258,
"brier": 0.3223852601810737,
"ece": 0.1619117743485849,
"confident_error_rate": 0.0
},
"sst5": {
"n": 40,
"accuracy": 0.5,
"nll": 1.2298145317333087,
"brier": 0.657291300878368,
"ece": 0.140258883357918,
"confident_error_rate": 0.0,
"score_mae": 0.6459058613680704
},
"synthetic_approval_en": {
"n": 13,
"accuracy": 1.0,
"nll": 0.02713503023395747,
"brier": 0.012400901610244864,
"ece": 0.023288187036937025,
"confident_error_rate": 0.0
},
"synthetic_approval_zh": {
"n": 13,
"accuracy": 0.9230769230769231,
"nll": 0.5102563080984519,
"brier": 0.15344398526112696,
"ece": 0.07446202415132108,
"confident_error_rate": 0.07692307692307693
},
"synthetic_deadline_en": {
"n": 12,
"accuracy": 0.9166666666666666,
"nll": 0.2372234019882158,
"brier": 0.12317158447362637,
"ece": 0.10171254195717513,
"confident_error_rate": 0.0,
"score_mae": 0.1977927131161766
},
"synthetic_deadline_zh": {
"n": 12,
"accuracy": 0.8333333333333334,
"nll": 0.2709692456015684,
"brier": 0.15859877650471121,
"ece": 0.18334363956648586,
"confident_error_rate": 0.0,
"score_mae": 0.19353121308906318
},
"synthetic_membership_en": {
"n": 12,
"accuracy": 1.0,
"nll": 0.4864467399805366,
"brier": 0.2567791241792929,
"ece": 0.37032491734394646,
"confident_error_rate": 0.0,
"score_mae": 0.38166059727082313
},
"synthetic_membership_zh": {
"n": 12,
"accuracy": 0.75,
"nll": 0.4737501553972594,
"brier": 0.2732967828693059,
"ece": 0.06227235426146951,
"confident_error_rate": 0.0,
"score_mae": 0.2467490356606777
},
"synthetic_refund_en": {
"n": 13,
"accuracy": 1.0,
"nll": 0.00465089068399359,
"brier": 0.00010111006670935923,
"ece": 0.004625338642054944,
"confident_error_rate": 0.0
},
"synthetic_refund_zh": {
"n": 13,
"accuracy": 0.9230769230769231,
"nll": 0.08045327525689482,
"brier": 0.048408113827161446,
"ece": 0.06068902817670701,
"confident_error_rate": 0.0
},
"synthetic_snake_observable_teacher": {
"n": 500,
"accuracy": 0.996,
"nll": 0.02869776433137775,
"brier": 0.009226250091244111,
"ece": 0.02180775425742663,
"confident_error_rate": 0.0
},
"trec": {
"n": 40,
"accuracy": 0.85,
"nll": 0.5164713056059863,
"brier": 0.24584333997497865,
"ece": 0.08697971292754486,
"confident_error_rate": 0.075
},
"yelp": {
"n": 40,
"accuracy": 0.425,
"nll": 1.419889190687788,
"brier": 0.7752482539353915,
"ece": 0.2837610425917585,
"confident_error_rate": 0.075,
"score_mae": 0.6234811251354
},
"yelp_yn": {
"n": 40,
"accuracy": 0.9,
"nll": 0.29940020197899064,
"brier": 0.16628785487628298,
"ece": 0.07293930838007276,
"confident_error_rate": 0.05
}
},
"by_language": {
"en": {
"n": 1070,
"accuracy": 0.8859813084112149,
"nll": 0.3321722535885867,
"brier": 0.1665772844080706,
"ece": 0.021660332496098857,
"confident_error_rate": 0.014953271028037384,
"score_mae": 0.5600499833693122
},
"zh": {
"n": 50,
"accuracy": 0.86,
"nll": 0.3323171479121088,
"brier": 0.15613648001271913,
"ece": 0.057765111700233476,
"confident_error_rate": 0.02,
"score_mae": 0.2201401243748704
}
},
"by_qtype": {
"choice": {
"n": 726,
"accuracy": 0.9476584022038568,
"nll": 0.20356603067047876,
"brier": 0.08558454886155338,
"ece": 0.021395347973956688,
"confident_error_rate": 0.009641873278236915
},
"noul": {
"n": 226,
"accuracy": 0.8938053097345132,
"nll": 0.25701539040659355,
"brier": 0.14562333459003557,
"ece": 0.03266069127275644,
"confident_error_rate": 0.02654867256637168
},
"score": {
"n": 168,
"accuracy": 0.6011904761904762,
"nll": 0.9890818585519972,
"brier": 0.5416622751573553,
"ece": 0.10589658337562606,
"confident_error_rate": 0.023809523809523808,
"score_mae": 0.511491432084392
}
},
"snake": {
"n": 500,
"accuracy": 0.996,
"nll": 0.02869776433137775,
"brier": 0.009226250091244111,
"ece": 0.02180775425742663,
"confident_error_rate": 0.0
},
"general": {
"n": 620,
"accuracy": 0.7951612903225806,
"nll": 0.576921429951136,
"brier": 0.2926312794704023,
"ece": 0.0518043898294689,
"confident_error_rate": 0.027419354838709678,
"score_mae": 0.511491432084392
}
},
"development": {
"raw": {
"n": 1380,
"accuracy": 0.8608695652173913,
"nll": 1.037557116940451,
"brier": 0.24184774106538062,
"ece": 0.11305555889037969,
"confident_error_rate": 0.09492753623188406,
"score_mae": 0.4030307561703975
},
"calibrated": {
"n": 1380,
"accuracy": 0.8608695652173913,
"nll": 0.3907753465146853,
"brier": 0.19660894207985088,
"ece": 0.0236135519422816,
"confident_error_rate": 0.018115942028985508,
"score_mae": 0.4692499422681266
},
"temperature": 4.5947934199881395,
"by_source": {
"agnews": {
"n": 60,
"accuracy": 0.8166666666666667,
"nll": 0.5432987762626141,
"brier": 0.2771836739526094,
"ece": 0.13067113910729217,
"confident_error_rate": 0.06666666666666667
},
"agnews_yn": {
"n": 120,
"accuracy": 0.9333333333333333,
"nll": 0.2474484161086575,
"brier": 0.12688493305603707,
"ece": 0.03359069154418503,
"confident_error_rate": 0.041666666666666664
},
"amazon": {
"n": 60,
"accuracy": 0.5833333333333334,
"nll": 1.019207187139469,
"brier": 0.5524721077332212,
"ece": 0.13899730979015376,
"confident_error_rate": 0.0,
"score_mae": 0.5781072710895353
},
"banking77": {
"n": 60,
"accuracy": 0.6833333333333333,
"nll": 1.386287052884947,
"brier": 0.4274681156509871,
"ece": 0.1742433197260666,
"confident_error_rate": 0.0
},
"boolq": {
"n": 60,
"accuracy": 0.6333333333333333,
"nll": 0.7842629466005272,
"brier": 0.5372472386575228,
"ece": 0.21517449578016706,
"confident_error_rate": 0.06666666666666667
},
"dbpedia14": {
"n": 60,
"accuracy": 0.9666666666666667,
"nll": 0.17627788368667227,
"brier": 0.05372020233696572,
"ece": 0.027486791887844184,
"confident_error_rate": 0.016666666666666666
},
"imdb": {
"n": 60,
"accuracy": 0.8,
"nll": 0.5042462474442398,
"brier": 0.27497683228600267,
"ece": 0.11728494718294821,
"confident_error_rate": 0.06666666666666667
},
"mnli": {
"n": 60,
"accuracy": 0.6333333333333333,
"nll": 0.8209178273118785,
"brier": 0.47354704169414424,
"ece": 0.1454240547092434,
"confident_error_rate": 0.016666666666666666
},
"sst5": {
"n": 60,
"accuracy": 0.5,
"nll": 1.09419991411571,
"brier": 0.6078575013753795,
"ece": 0.10528637320636364,
"confident_error_rate": 0.0,
"score_mae": 0.5626427438598316
},
"synthetic_approval_en": {
"n": 13,
"accuracy": 1.0,
"nll": 0.00305151055912005,
"brier": 9.100589788026809e-05,
"ece": 0.003028460775100883,
"confident_error_rate": 0.0
},
"synthetic_approval_zh": {
"n": 13,
"accuracy": 0.9230769230769231,
"nll": 0.31252450790157693,
"brier": 0.14840695576773935,
"ece": 0.07158792870719877,
"confident_error_rate": 0.07692307692307693
},
"synthetic_deadline_en": {
"n": 12,
"accuracy": 0.9166666666666666,
"nll": 0.22089136851581057,
"brier": 0.10851671633317528,
"ece": 0.09197092561596329,
"confident_error_rate": 0.0,
"score_mae": 0.17273660444065683
},
"synthetic_deadline_zh": {
"n": 12,
"accuracy": 0.9166666666666666,
"nll": 0.22471943817596504,
"brier": 0.09609427381476522,
"ece": 0.14289170677922886,
"confident_error_rate": 0.0,
"score_mae": 0.12739659503185494
},
"synthetic_membership_en": {
"n": 12,
"accuracy": 0.8333333333333334,
"nll": 0.5289100636414515,
"brier": 0.3247949977066678,
"ece": 0.3540032822410373,
"confident_error_rate": 0.0,
"score_mae": 0.3658395223985365
},
"synthetic_membership_zh": {
"n": 12,
"accuracy": 0.9166666666666666,
"nll": 0.3384060300915314,
"brier": 0.17768819657830845,
"ece": 0.1240372424832984,
"confident_error_rate": 0.0,
"score_mae": 0.14854067532162193
},
"synthetic_refund_en": {
"n": 13,
"accuracy": 1.0,
"nll": 0.004772849934686132,
"brier": 9.814589364790191e-05,
"ece": 0.004748060980601321,
"confident_error_rate": 0.0
},
"synthetic_refund_zh": {
"n": 13,
"accuracy": 1.0,
"nll": 0.026827711733167094,
"brier": 0.002599816753125005,
"ece": 0.02614938190979532,
"confident_error_rate": 0.0
},
"synthetic_snake_observable_teacher": {
"n": 500,
"accuracy": 0.988,
"nll": 0.03911276666514134,
"brier": 0.015846521973586804,
"ece": 0.022786703593587133,
"confident_error_rate": 0.0
},
"trec": {
"n": 60,
"accuracy": 0.9166666666666666,
"nll": 0.3414991446020253,
"brier": 0.13705633897022337,
"ece": 0.08460856930544054,
"confident_error_rate": 0.016666666666666666
},
"yelp": {
"n": 60,
"accuracy": 0.6833333333333333,
"nll": 0.9475770833309305,
"brier": 0.4856780158774344,
"ece": 0.09408258083704575,
"confident_error_rate": 0.03333333333333333,
"score_mae": 0.47949708623098025
},
"yelp_yn": {
"n": 60,
"accuracy": 0.9166666666666666,
"nll": 0.21141537958578643,
"brier": 0.13479642958918142,
"ece": 0.05396240895190205,
"confident_error_rate": 0.03333333333333333
}
},
"by_language": {
"en": {
"n": 1330,
"accuracy": 0.8578947368421053,
"nll": 0.3970683411397024,
"brier": 0.20005403186669637,
"ece": 0.026465524784310385,
"confident_error_rate": 0.01804511278195489,
"score_mae": 0.5082242136906429
},
"zh": {
"n": 50,
"accuracy": 0.94,
"nll": 0.22338168948923257,
"brier": 0.10496955374976241,
"ece": 0.05224892565568175,
"confident_error_rate": 0.02,
"score_mae": 0.13796863517673844
}
},
"by_qtype": {
"choice": {
"n": 826,
"accuracy": 0.9200968523002422,
"nll": 0.26604807827778204,
"brier": 0.11137077099847731,
"ece": 0.01219164077100229,
"confident_error_rate": 0.009685230024213076
},
"noul": {
"n": 326,
"accuracy": 0.8558282208588958,
"nll": 0.3684051891182644,
"brier": 0.2211119494266715,
"ece": 0.074143327637379,
"confident_error_rate": 0.046012269938650305
},
"score": {
"n": 228,
"accuracy": 0.6535087719298246,
"nll": 0.8746235696502792,
"brier": 0.4703753847033207,
"ece": 0.027926810584187323,
"confident_error_rate": 0.008771929824561403,
"score_mae": 0.4692499422681266
}
},
"snake": {
"n": 500,
"accuracy": 0.988,
"nll": 0.03911276666514134,
"brier": 0.015846521973586804,
"ece": 0.022786703593587133,
"confident_error_rate": 0.0
},
"general": {
"n": 880,
"accuracy": 0.7886363636363637,
"nll": 0.5905836305201081,
"brier": 0.2993148625947737,
"ece": 0.04887231156165163,
"confident_error_rate": 0.028409090909090908,
"score_mae": 0.4692499422681266
}
},
"training_seconds": 1089.3942992687225,
"optimizer_steps": 4476,
"forward_tokens": 22269908,
"snake_closed_loop": {
"complete": true,
"food": {
"count": 20,
"mean": 41.4,
"p50": 41.5,
"p95": 50,
"max": 53
},
"collisions": 0,
"terminal_reasons": {
"max_steps": 20
},
"prediction_call_ms": {
"count": 1500,
"mean": 47.461315778669665,
"p50": 48.8280649999524,
"p95": 51.10321799998019,
"max": 57.466298000008464
}
}
}