Files
r2egym-gpt5-codex-160ep-1M/trainer_state.json
ModelHub XC 002044d1b4 初始化项目,由ModelHub XC社区提供模型
Model: laion/r2egym-gpt5-codex-160ep-1M
Source: Original Platform
2026-06-03 01:44:14 +08:00

3755 lines
102 KiB
JSON

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 7.0,
"eval_steps": 500,
"global_step": 1687,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.02074688796680498,
"grad_norm": 41.397027140234286,
"learning_rate": 9.467455621301776e-07,
"loss": 3.331,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.9795444011688232,
"step": 5,
"valid_targets_mean": 134.4,
"valid_targets_min": 41
},
{
"epoch": 0.04149377593360996,
"grad_norm": 35.61947240445416,
"learning_rate": 2.1301775147929e-06,
"loss": 2.722,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.5642552375793457,
"step": 10,
"valid_targets_mean": 161.3,
"valid_targets_min": 48
},
{
"epoch": 0.06224066390041494,
"grad_norm": 52.370048634783274,
"learning_rate": 3.313609467455622e-06,
"loss": 3.1562,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.396860122680664,
"step": 15,
"valid_targets_mean": 97.2,
"valid_targets_min": 66
},
{
"epoch": 0.08298755186721991,
"grad_norm": 25.358721084807172,
"learning_rate": 4.497041420118343e-06,
"loss": 2.9616,
"loss_nan_ranks": 0,
"loss_rank_avg": 3.084906578063965,
"step": 20,
"valid_targets_mean": 104.8,
"valid_targets_min": 68
},
{
"epoch": 0.1037344398340249,
"grad_norm": 14.282450323196407,
"learning_rate": 5.680473372781066e-06,
"loss": 2.4025,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.6719250679016113,
"step": 25,
"valid_targets_mean": 96.6,
"valid_targets_min": 54
},
{
"epoch": 0.12448132780082988,
"grad_norm": 6.589012266835091,
"learning_rate": 6.863905325443787e-06,
"loss": 2.3199,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.578616976737976,
"step": 30,
"valid_targets_mean": 189.2,
"valid_targets_min": 59
},
{
"epoch": 0.14522821576763487,
"grad_norm": 8.720894250976365,
"learning_rate": 8.04733727810651e-06,
"loss": 2.0466,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.2473416328430176,
"step": 35,
"valid_targets_mean": 108.4,
"valid_targets_min": 59
},
{
"epoch": 0.16597510373443983,
"grad_norm": 6.572593557764174,
"learning_rate": 9.230769230769232e-06,
"loss": 2.0725,
"loss_nan_ranks": 0,
"loss_rank_avg": 2.0674421787261963,
"step": 40,
"valid_targets_mean": 113.9,
"valid_targets_min": 58
},
{
"epoch": 0.18672199170124482,
"grad_norm": 5.530699449121817,
"learning_rate": 1.0414201183431953e-05,
"loss": 1.7968,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.739060878753662,
"step": 45,
"valid_targets_mean": 115.5,
"valid_targets_min": 62
},
{
"epoch": 0.2074688796680498,
"grad_norm": 4.37749124574813,
"learning_rate": 1.1597633136094675e-05,
"loss": 1.6917,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.5368684530258179,
"step": 50,
"valid_targets_mean": 139.1,
"valid_targets_min": 57
},
{
"epoch": 0.22821576763485477,
"grad_norm": 4.295701199410857,
"learning_rate": 1.2781065088757399e-05,
"loss": 1.5291,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.4562065601348877,
"step": 55,
"valid_targets_mean": 132.5,
"valid_targets_min": 55
},
{
"epoch": 0.24896265560165975,
"grad_norm": 4.084731189779167,
"learning_rate": 1.396449704142012e-05,
"loss": 1.5082,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.390439510345459,
"step": 60,
"valid_targets_mean": 141.6,
"valid_targets_min": 47
},
{
"epoch": 0.2697095435684647,
"grad_norm": 4.96787833573562,
"learning_rate": 1.5147928994082842e-05,
"loss": 1.3951,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.4382585287094116,
"step": 65,
"valid_targets_mean": 94.7,
"valid_targets_min": 64
},
{
"epoch": 0.29045643153526973,
"grad_norm": 4.26457221140469,
"learning_rate": 1.6331360946745562e-05,
"loss": 1.4802,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.4862592220306396,
"step": 70,
"valid_targets_mean": 133.1,
"valid_targets_min": 75
},
{
"epoch": 0.3112033195020747,
"grad_norm": 4.762762510462149,
"learning_rate": 1.7514792899408286e-05,
"loss": 1.4573,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.6193333864212036,
"step": 75,
"valid_targets_mean": 95.4,
"valid_targets_min": 62
},
{
"epoch": 0.33195020746887965,
"grad_norm": 5.211587363436592,
"learning_rate": 1.8698224852071007e-05,
"loss": 1.3158,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.5069483518600464,
"step": 80,
"valid_targets_mean": 85.4,
"valid_targets_min": 51
},
{
"epoch": 0.35269709543568467,
"grad_norm": 5.167018764133475,
"learning_rate": 1.9881656804733727e-05,
"loss": 1.3183,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.5408308506011963,
"step": 85,
"valid_targets_mean": 111.9,
"valid_targets_min": 78
},
{
"epoch": 0.37344398340248963,
"grad_norm": 4.19223324707348,
"learning_rate": 2.106508875739645e-05,
"loss": 1.1263,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0536243915557861,
"step": 90,
"valid_targets_mean": 109.9,
"valid_targets_min": 54
},
{
"epoch": 0.3941908713692946,
"grad_norm": 4.327426360795543,
"learning_rate": 2.224852071005917e-05,
"loss": 1.2161,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.2340021133422852,
"step": 95,
"valid_targets_mean": 108.8,
"valid_targets_min": 47
},
{
"epoch": 0.4149377593360996,
"grad_norm": 4.407589788387759,
"learning_rate": 2.3431952662721896e-05,
"loss": 1.179,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.28630530834198,
"step": 100,
"valid_targets_mean": 123.1,
"valid_targets_min": 78
},
{
"epoch": 0.43568464730290457,
"grad_norm": 4.476057603578654,
"learning_rate": 2.461538461538462e-05,
"loss": 1.2358,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.2336297035217285,
"step": 105,
"valid_targets_mean": 96.8,
"valid_targets_min": 61
},
{
"epoch": 0.45643153526970953,
"grad_norm": 4.275689963533272,
"learning_rate": 2.5798816568047337e-05,
"loss": 1.2222,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.3048899173736572,
"step": 110,
"valid_targets_mean": 119.9,
"valid_targets_min": 60
},
{
"epoch": 0.47717842323651455,
"grad_norm": 4.114686647723986,
"learning_rate": 2.698224852071006e-05,
"loss": 1.1238,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1282589435577393,
"step": 115,
"valid_targets_mean": 119.6,
"valid_targets_min": 57
},
{
"epoch": 0.4979253112033195,
"grad_norm": 2.848802193178965,
"learning_rate": 2.8165680473372784e-05,
"loss": 1.1292,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9970792531967163,
"step": 120,
"valid_targets_mean": 197.2,
"valid_targets_min": 70
},
{
"epoch": 0.5186721991701245,
"grad_norm": 3.917339058459324,
"learning_rate": 2.9349112426035505e-05,
"loss": 1.2792,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0999289751052856,
"step": 125,
"valid_targets_mean": 106.1,
"valid_targets_min": 50
},
{
"epoch": 0.5394190871369294,
"grad_norm": 4.973393373968101,
"learning_rate": 3.0532544378698226e-05,
"loss": 1.2547,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.5572994947433472,
"step": 130,
"valid_targets_mean": 93.6,
"valid_targets_min": 48
},
{
"epoch": 0.5601659751037344,
"grad_norm": 4.093849799808453,
"learning_rate": 3.171597633136095e-05,
"loss": 1.1989,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.289677381515503,
"step": 135,
"valid_targets_mean": 117.5,
"valid_targets_min": 56
},
{
"epoch": 0.5809128630705395,
"grad_norm": 4.130110837411625,
"learning_rate": 3.289940828402367e-05,
"loss": 1.291,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.284778356552124,
"step": 140,
"valid_targets_mean": 104.9,
"valid_targets_min": 59
},
{
"epoch": 0.6016597510373444,
"grad_norm": 5.236925712153894,
"learning_rate": 3.40828402366864e-05,
"loss": 1.2244,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.3275096416473389,
"step": 145,
"valid_targets_mean": 93.9,
"valid_targets_min": 52
},
{
"epoch": 0.6224066390041494,
"grad_norm": 4.0627237296792185,
"learning_rate": 3.5266272189349114e-05,
"loss": 1.1273,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.3301905393600464,
"step": 150,
"valid_targets_mean": 118.0,
"valid_targets_min": 82
},
{
"epoch": 0.6431535269709544,
"grad_norm": 3.6554328860688985,
"learning_rate": 3.644970414201184e-05,
"loss": 1.0308,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0490658283233643,
"step": 155,
"valid_targets_mean": 149.6,
"valid_targets_min": 53
},
{
"epoch": 0.6639004149377593,
"grad_norm": 4.388884052961649,
"learning_rate": 3.763313609467456e-05,
"loss": 0.9736,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.3347870111465454,
"step": 160,
"valid_targets_mean": 87.1,
"valid_targets_min": 46
},
{
"epoch": 0.6846473029045643,
"grad_norm": 4.283544953396232,
"learning_rate": 3.881656804733728e-05,
"loss": 1.166,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.3210680484771729,
"step": 165,
"valid_targets_mean": 119.6,
"valid_targets_min": 61
},
{
"epoch": 0.7053941908713693,
"grad_norm": 4.3275598749896504,
"learning_rate": 4e-05,
"loss": 1.0946,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.173753023147583,
"step": 170,
"valid_targets_mean": 94.1,
"valid_targets_min": 63
},
{
"epoch": 0.7261410788381742,
"grad_norm": 3.133490340424578,
"learning_rate": 3.999892923951514e-05,
"loss": 1.0991,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9916005730628967,
"step": 175,
"valid_targets_mean": 175.0,
"valid_targets_min": 65
},
{
"epoch": 0.7468879668049793,
"grad_norm": 3.820724592316591,
"learning_rate": 3.999571707271335e-05,
"loss": 1.1184,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.162503719329834,
"step": 180,
"valid_targets_mean": 116.2,
"valid_targets_min": 66
},
{
"epoch": 0.7676348547717843,
"grad_norm": 3.1153583861325584,
"learning_rate": 3.999036384354076e-05,
"loss": 1.0885,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.132462501525879,
"step": 185,
"valid_targets_mean": 155.9,
"valid_targets_min": 55
},
{
"epoch": 0.7883817427385892,
"grad_norm": 3.2268572394739867,
"learning_rate": 3.99828701252e-05,
"loss": 1.1776,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0295939445495605,
"step": 190,
"valid_targets_mean": 146.7,
"valid_targets_min": 68
},
{
"epoch": 0.8091286307053942,
"grad_norm": 4.014232602275577,
"learning_rate": 3.997323672008881e-05,
"loss": 1.1762,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.3584688901901245,
"step": 195,
"valid_targets_mean": 106.2,
"valid_targets_min": 56
},
{
"epoch": 0.8298755186721992,
"grad_norm": 2.894292449308229,
"learning_rate": 3.9961464659714154e-05,
"loss": 1.0519,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8366074562072754,
"step": 200,
"valid_targets_mean": 144.6,
"valid_targets_min": 63
},
{
"epoch": 0.8506224066390041,
"grad_norm": 3.346195548737324,
"learning_rate": 3.994755520458173e-05,
"loss": 1.2221,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.131117343902588,
"step": 205,
"valid_targets_mean": 118.6,
"valid_targets_min": 61
},
{
"epoch": 0.8713692946058091,
"grad_norm": 2.861208912009083,
"learning_rate": 3.9931509844061034e-05,
"loss": 1.1059,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.914353609085083,
"step": 210,
"valid_targets_mean": 161.1,
"valid_targets_min": 68
},
{
"epoch": 0.8921161825726142,
"grad_norm": 2.4635047614518433,
"learning_rate": 3.991333029622587e-05,
"loss": 1.0881,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7619673609733582,
"step": 215,
"valid_targets_mean": 187.9,
"valid_targets_min": 49
},
{
"epoch": 0.9128630705394191,
"grad_norm": 3.439388327039743,
"learning_rate": 3.9893018507670384e-05,
"loss": 1.0801,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1252914667129517,
"step": 220,
"valid_targets_mean": 128.8,
"valid_targets_min": 66
},
{
"epoch": 0.9336099585062241,
"grad_norm": 3.0479402830613016,
"learning_rate": 3.987057665330063e-05,
"loss": 1.1093,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.052093505859375,
"step": 225,
"valid_targets_mean": 155.6,
"valid_targets_min": 63
},
{
"epoch": 0.9543568464730291,
"grad_norm": 3.4173751722238928,
"learning_rate": 3.984600713610169e-05,
"loss": 1.043,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9762424230575562,
"step": 230,
"valid_targets_mean": 115.1,
"valid_targets_min": 66
},
{
"epoch": 0.975103734439834,
"grad_norm": 3.203880033800484,
"learning_rate": 3.981931258688038e-05,
"loss": 1.1468,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.121521234512329,
"step": 235,
"valid_targets_mean": 137.5,
"valid_targets_min": 68
},
{
"epoch": 0.995850622406639,
"grad_norm": 3.4770939413356463,
"learning_rate": 3.979049586398355e-05,
"loss": 1.1073,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1996042728424072,
"step": 240,
"valid_targets_mean": 117.0,
"valid_targets_min": 46
},
{
"epoch": 1.016597510373444,
"grad_norm": 3.930284483527447,
"learning_rate": 3.975956005299202e-05,
"loss": 1.1062,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1427793502807617,
"step": 245,
"valid_targets_mean": 93.7,
"valid_targets_min": 50
},
{
"epoch": 1.037344398340249,
"grad_norm": 3.4787980480099634,
"learning_rate": 3.972650846639019e-05,
"loss": 0.9987,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8831626176834106,
"step": 250,
"valid_targets_mean": 117.0,
"valid_targets_min": 51
},
{
"epoch": 1.058091286307054,
"grad_norm": 5.765404317940215,
"learning_rate": 3.9691344643211346e-05,
"loss": 0.9541,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1344108581542969,
"step": 255,
"valid_targets_mean": 104.3,
"valid_targets_min": 63
},
{
"epoch": 1.0788381742738589,
"grad_norm": 3.77909701623929,
"learning_rate": 3.965407234865871e-05,
"loss": 1.0478,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0921099185943604,
"step": 260,
"valid_targets_mean": 114.2,
"valid_targets_min": 59
},
{
"epoch": 1.099585062240664,
"grad_norm": 3.026006206344685,
"learning_rate": 3.9614695573702325e-05,
"loss": 0.9397,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7856882810592651,
"step": 265,
"valid_targets_mean": 144.1,
"valid_targets_min": 64
},
{
"epoch": 1.120331950207469,
"grad_norm": 2.4834572794776446,
"learning_rate": 3.957321853465163e-05,
"loss": 0.9561,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7817927598953247,
"step": 270,
"valid_targets_mean": 190.4,
"valid_targets_min": 55
},
{
"epoch": 1.1410788381742738,
"grad_norm": 3.056967616081349,
"learning_rate": 3.952964567270409e-05,
"loss": 0.9533,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8634785413742065,
"step": 275,
"valid_targets_mean": 134.5,
"valid_targets_min": 49
},
{
"epoch": 1.161825726141079,
"grad_norm": 2.5906209217918117,
"learning_rate": 3.9483981653469586e-05,
"loss": 0.9866,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7464009523391724,
"step": 280,
"valid_targets_mean": 179.6,
"valid_targets_min": 57
},
{
"epoch": 1.1825726141078838,
"grad_norm": 3.2705129782461686,
"learning_rate": 3.9436231366470836e-05,
"loss": 1.0861,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9500118494033813,
"step": 285,
"valid_targets_mean": 130.6,
"valid_targets_min": 64
},
{
"epoch": 1.2033195020746887,
"grad_norm": 2.8686194044847473,
"learning_rate": 3.93863999246199e-05,
"loss": 0.8913,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7544863224029541,
"step": 290,
"valid_targets_mean": 159.8,
"valid_targets_min": 63
},
{
"epoch": 1.2240663900414939,
"grad_norm": 2.4839411870748935,
"learning_rate": 3.933449266367066e-05,
"loss": 0.9027,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6904275417327881,
"step": 295,
"valid_targets_mean": 166.6,
"valid_targets_min": 44
},
{
"epoch": 1.2448132780082988,
"grad_norm": 3.1275981909890938,
"learning_rate": 3.92805151416475e-05,
"loss": 0.984,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0172083377838135,
"step": 300,
"valid_targets_mean": 148.0,
"valid_targets_min": 68
},
{
"epoch": 1.2655601659751037,
"grad_norm": 3.6194049895575517,
"learning_rate": 3.9224473138250186e-05,
"loss": 0.9891,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9883549213409424,
"step": 305,
"valid_targets_mean": 116.4,
"valid_targets_min": 52
},
{
"epoch": 1.2863070539419086,
"grad_norm": 3.426981706024395,
"learning_rate": 3.9166372654235e-05,
"loss": 0.8473,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7937560677528381,
"step": 310,
"valid_targets_mean": 118.4,
"valid_targets_min": 46
},
{
"epoch": 1.3070539419087137,
"grad_norm": 3.152855357782168,
"learning_rate": 3.9106219910772184e-05,
"loss": 0.9896,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8297359943389893,
"step": 315,
"valid_targets_mean": 124.4,
"valid_targets_min": 68
},
{
"epoch": 1.3278008298755186,
"grad_norm": 3.069176743069222,
"learning_rate": 3.90440213487798e-05,
"loss": 0.9571,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8529037237167358,
"step": 320,
"valid_targets_mean": 140.6,
"valid_targets_min": 58
},
{
"epoch": 1.3485477178423237,
"grad_norm": 3.3034611009209978,
"learning_rate": 3.897978362823411e-05,
"loss": 0.9725,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0133411884307861,
"step": 325,
"valid_targets_mean": 133.8,
"valid_targets_min": 65
},
{
"epoch": 1.3692946058091287,
"grad_norm": 3.5392784144202873,
"learning_rate": 3.8913513627456374e-05,
"loss": 1.0897,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0982413291931152,
"step": 330,
"valid_targets_mean": 108.6,
"valid_targets_min": 47
},
{
"epoch": 1.3900414937759336,
"grad_norm": 4.133104453790177,
"learning_rate": 3.8845218442376416e-05,
"loss": 1.0984,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1011841297149658,
"step": 335,
"valid_targets_mean": 92.4,
"valid_targets_min": 58
},
{
"epoch": 1.4107883817427385,
"grad_norm": 2.911694581277071,
"learning_rate": 3.877490538577278e-05,
"loss": 0.9807,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8329280614852905,
"step": 340,
"valid_targets_mean": 146.0,
"valid_targets_min": 69
},
{
"epoch": 1.4315352697095436,
"grad_norm": 3.7192204333227976,
"learning_rate": 3.870258198648974e-05,
"loss": 0.8938,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9809222221374512,
"step": 345,
"valid_targets_mean": 112.0,
"valid_targets_min": 73
},
{
"epoch": 1.4522821576763485,
"grad_norm": 4.515200149712003,
"learning_rate": 3.862825598863108e-05,
"loss": 0.9289,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.774828314781189,
"step": 350,
"valid_targets_mean": 201.0,
"valid_targets_min": 64
},
{
"epoch": 1.4730290456431536,
"grad_norm": 3.36292777919316,
"learning_rate": 3.855193535073097e-05,
"loss": 0.9448,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.003603458404541,
"step": 355,
"valid_targets_mean": 131.2,
"valid_targets_min": 58
},
{
"epoch": 1.4937759336099585,
"grad_norm": 3.3587994781305976,
"learning_rate": 3.847362824490173e-05,
"loss": 0.9142,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8775392174720764,
"step": 360,
"valid_targets_mean": 119.1,
"valid_targets_min": 60
},
{
"epoch": 1.5145228215767634,
"grad_norm": 4.141295017158325,
"learning_rate": 3.839334305595881e-05,
"loss": 0.9044,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9412416219711304,
"step": 365,
"valid_targets_mean": 143.9,
"valid_targets_min": 57
},
{
"epoch": 1.5352697095435683,
"grad_norm": 2.9785125564157107,
"learning_rate": 3.831108838052301e-05,
"loss": 1.0486,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7641946077346802,
"step": 370,
"valid_targets_mean": 147.4,
"valid_targets_min": 64
},
{
"epoch": 1.5560165975103735,
"grad_norm": 4.1021332056584,
"learning_rate": 3.822687302609994e-05,
"loss": 1.0625,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.2731382846832275,
"step": 375,
"valid_targets_mean": 96.9,
"valid_targets_min": 59
},
{
"epoch": 1.5767634854771784,
"grad_norm": 3.4014528210970743,
"learning_rate": 3.814070601013697e-05,
"loss": 0.99,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0162218809127808,
"step": 380,
"valid_targets_mean": 117.1,
"valid_targets_min": 67
},
{
"epoch": 1.5975103734439835,
"grad_norm": 3.998857883960084,
"learning_rate": 3.8052596559057674e-05,
"loss": 1.0098,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1427373886108398,
"step": 385,
"valid_targets_mean": 94.8,
"valid_targets_min": 63
},
{
"epoch": 1.6182572614107884,
"grad_norm": 3.573313288898261,
"learning_rate": 3.7962554107273926e-05,
"loss": 0.9996,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9803476333618164,
"step": 390,
"valid_targets_mean": 125.8,
"valid_targets_min": 64
},
{
"epoch": 1.6390041493775933,
"grad_norm": 2.627525409761934,
"learning_rate": 3.7870588296175644e-05,
"loss": 0.9104,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8428837060928345,
"step": 395,
"valid_targets_mean": 185.5,
"valid_targets_min": 54
},
{
"epoch": 1.6597510373443982,
"grad_norm": 3.8364107288625786,
"learning_rate": 3.7776708973098476e-05,
"loss": 0.9904,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.069312334060669,
"step": 400,
"valid_targets_mean": 105.6,
"valid_targets_min": 63
},
{
"epoch": 1.6804979253112033,
"grad_norm": 2.548205520795793,
"learning_rate": 3.768092619026937e-05,
"loss": 1.0073,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7841141223907471,
"step": 405,
"valid_targets_mean": 155.5,
"valid_targets_min": 62
},
{
"epoch": 1.7012448132780082,
"grad_norm": 3.382526642855817,
"learning_rate": 3.7583250203730234e-05,
"loss": 0.9565,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8623814582824707,
"step": 410,
"valid_targets_mean": 124.7,
"valid_targets_min": 69
},
{
"epoch": 1.7219917012448134,
"grad_norm": 4.314083908492832,
"learning_rate": 3.7483691472239744e-05,
"loss": 1.0132,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0424532890319824,
"step": 415,
"valid_targets_mean": 93.2,
"valid_targets_min": 60
},
{
"epoch": 1.7427385892116183,
"grad_norm": 2.5482144454375693,
"learning_rate": 3.7382260656153436e-05,
"loss": 0.8144,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6370407342910767,
"step": 420,
"valid_targets_mean": 142.7,
"valid_targets_min": 61
},
{
"epoch": 1.7634854771784232,
"grad_norm": 2.6674353520861014,
"learning_rate": 3.727896861628231e-05,
"loss": 0.9513,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7452382445335388,
"step": 425,
"valid_targets_mean": 178.5,
"valid_targets_min": 51
},
{
"epoch": 1.784232365145228,
"grad_norm": 3.494532029292696,
"learning_rate": 3.717382641272984e-05,
"loss": 0.9952,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.999313235282898,
"step": 430,
"valid_targets_mean": 100.6,
"valid_targets_min": 41
},
{
"epoch": 1.8049792531120332,
"grad_norm": 2.11108790003265,
"learning_rate": 3.7066845303707694e-05,
"loss": 0.8269,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5311782360076904,
"step": 435,
"valid_targets_mean": 236.6,
"valid_targets_min": 57
},
{
"epoch": 1.8257261410788381,
"grad_norm": 3.9114163853143737,
"learning_rate": 3.6958036744330297e-05,
"loss": 1.0451,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1796157360076904,
"step": 440,
"valid_targets_mean": 100.8,
"valid_targets_min": 65
},
{
"epoch": 1.8464730290456433,
"grad_norm": 2.966917119742618,
"learning_rate": 3.6847412385388236e-05,
"loss": 0.8965,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8532978892326355,
"step": 445,
"valid_targets_mean": 140.1,
"valid_targets_min": 61
},
{
"epoch": 1.8672199170124482,
"grad_norm": 3.599518479860152,
"learning_rate": 3.673498407210073e-05,
"loss": 0.9751,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9784261584281921,
"step": 450,
"valid_targets_mean": 105.6,
"valid_targets_min": 53
},
{
"epoch": 1.887966804979253,
"grad_norm": 3.940460990497638,
"learning_rate": 3.662076384284729e-05,
"loss": 0.9165,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.0794143676757812,
"step": 455,
"valid_targets_mean": 93.3,
"valid_targets_min": 56
},
{
"epoch": 1.908713692946058,
"grad_norm": 2.4631143719909656,
"learning_rate": 3.650476392787873e-05,
"loss": 0.9068,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7837638258934021,
"step": 460,
"valid_targets_mean": 185.1,
"valid_targets_min": 69
},
{
"epoch": 1.929460580912863,
"grad_norm": 3.725991171844277,
"learning_rate": 3.638699674800758e-05,
"loss": 0.9429,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1260584592819214,
"step": 465,
"valid_targets_mean": 108.4,
"valid_targets_min": 66
},
{
"epoch": 1.950207468879668,
"grad_norm": 3.3343720514690145,
"learning_rate": 3.6267474913278086e-05,
"loss": 1.0307,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9506238698959351,
"step": 470,
"valid_targets_mean": 127.8,
"valid_targets_min": 50
},
{
"epoch": 1.9709543568464731,
"grad_norm": 3.827770704987103,
"learning_rate": 3.614621122161603e-05,
"loss": 1.0214,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1068814992904663,
"step": 475,
"valid_targets_mean": 105.6,
"valid_targets_min": 41
},
{
"epoch": 1.991701244813278,
"grad_norm": 3.861407678400857,
"learning_rate": 3.6023218657458334e-05,
"loss": 1.0285,
"loss_nan_ranks": 0,
"loss_rank_avg": 1.1757726669311523,
"step": 480,
"valid_targets_mean": 103.8,
"valid_targets_min": 64
},
{
"epoch": 2.012448132780083,
"grad_norm": 3.1202708919801934,
"learning_rate": 3.589851039036277e-05,
"loss": 0.8016,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7584832310676575,
"step": 485,
"valid_targets_mean": 110.2,
"valid_targets_min": 56
},
{
"epoch": 2.033195020746888,
"grad_norm": 3.8354655021211514,
"learning_rate": 3.577209977359778e-05,
"loss": 0.697,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8398362398147583,
"step": 490,
"valid_targets_mean": 103.0,
"valid_targets_min": 57
},
{
"epoch": 2.0539419087136928,
"grad_norm": 4.329441765322654,
"learning_rate": 3.5644000342712695e-05,
"loss": 0.803,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9269231557846069,
"step": 495,
"valid_targets_mean": 100.8,
"valid_targets_min": 67
},
{
"epoch": 2.074688796680498,
"grad_norm": 4.4746136552596845,
"learning_rate": 3.55142258140884e-05,
"loss": 0.6451,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8513693809509277,
"step": 500,
"valid_targets_mean": 110.4,
"valid_targets_min": 59
},
{
"epoch": 2.095435684647303,
"grad_norm": 4.6090358834641565,
"learning_rate": 3.538279008346861e-05,
"loss": 0.7953,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8998103141784668,
"step": 505,
"valid_targets_mean": 84.5,
"valid_targets_min": 63
},
{
"epoch": 2.116182572614108,
"grad_norm": 3.5628641412692224,
"learning_rate": 3.524970722447197e-05,
"loss": 0.6491,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6803164482116699,
"step": 510,
"valid_targets_mean": 125.2,
"valid_targets_min": 55
},
{
"epoch": 2.136929460580913,
"grad_norm": 4.445807529078514,
"learning_rate": 3.511499148708517e-05,
"loss": 0.8269,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9091898202896118,
"step": 515,
"valid_targets_mean": 89.8,
"valid_targets_min": 55
},
{
"epoch": 2.1576763485477177,
"grad_norm": 3.696118063258985,
"learning_rate": 3.497865729613702e-05,
"loss": 0.7598,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.738641619682312,
"step": 520,
"valid_targets_mean": 128.2,
"valid_targets_min": 60
},
{
"epoch": 2.1784232365145226,
"grad_norm": 3.9014408920828285,
"learning_rate": 3.484071924975398e-05,
"loss": 0.8091,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7708579301834106,
"step": 525,
"valid_targets_mean": 115.0,
"valid_targets_min": 44
},
{
"epoch": 2.199170124481328,
"grad_norm": 4.305295137986893,
"learning_rate": 3.4701192117796964e-05,
"loss": 0.7631,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7537413835525513,
"step": 530,
"valid_targets_mean": 99.2,
"valid_targets_min": 61
},
{
"epoch": 2.219917012448133,
"grad_norm": 3.4539393109108762,
"learning_rate": 3.456009084027995e-05,
"loss": 0.696,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6141202449798584,
"step": 535,
"valid_targets_mean": 142.5,
"valid_targets_min": 67
},
{
"epoch": 2.240663900414938,
"grad_norm": 3.499033553308627,
"learning_rate": 3.441743052577014e-05,
"loss": 0.7411,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6117064952850342,
"step": 540,
"valid_targets_mean": 134.6,
"valid_targets_min": 66
},
{
"epoch": 2.2614107883817427,
"grad_norm": 4.012715094473001,
"learning_rate": 3.4273226449770314e-05,
"loss": 0.7664,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.848983645439148,
"step": 545,
"valid_targets_mean": 108.5,
"valid_targets_min": 66
},
{
"epoch": 2.2821576763485476,
"grad_norm": 4.829214040777758,
"learning_rate": 3.4127494053083086e-05,
"loss": 0.6762,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8868389129638672,
"step": 550,
"valid_targets_mean": 86.9,
"valid_targets_min": 62
},
{
"epoch": 2.3029045643153525,
"grad_norm": 3.9749139794203345,
"learning_rate": 3.398024894015764e-05,
"loss": 0.6597,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8123997449874878,
"step": 555,
"valid_targets_mean": 119.9,
"valid_targets_min": 72
},
{
"epoch": 2.323651452282158,
"grad_norm": 4.34876414676235,
"learning_rate": 3.383150687741883e-05,
"loss": 0.8247,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9735461473464966,
"step": 560,
"valid_targets_mean": 99.9,
"valid_targets_min": 52
},
{
"epoch": 2.3443983402489628,
"grad_norm": 3.912739845823611,
"learning_rate": 3.368128379157897e-05,
"loss": 0.7557,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7604776620864868,
"step": 565,
"valid_targets_mean": 119.3,
"valid_targets_min": 64
},
{
"epoch": 2.3651452282157677,
"grad_norm": 4.185361795583942,
"learning_rate": 3.3529595767932496e-05,
"loss": 0.7067,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.722848653793335,
"step": 570,
"valid_targets_mean": 133.8,
"valid_targets_min": 71
},
{
"epoch": 2.3858921161825726,
"grad_norm": 3.7895338790081166,
"learning_rate": 3.3376459048633565e-05,
"loss": 0.8143,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7524533271789551,
"step": 575,
"valid_targets_mean": 126.1,
"valid_targets_min": 53
},
{
"epoch": 2.4066390041493775,
"grad_norm": 4.1719327780746225,
"learning_rate": 3.322189003095696e-05,
"loss": 0.7022,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7856139540672302,
"step": 580,
"valid_targets_mean": 121.4,
"valid_targets_min": 55
},
{
"epoch": 2.4273858921161824,
"grad_norm": 3.7130870658654795,
"learning_rate": 3.306590526554233e-05,
"loss": 0.8357,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8768580555915833,
"step": 585,
"valid_targets_mean": 129.4,
"valid_targets_min": 47
},
{
"epoch": 2.4481327800829877,
"grad_norm": 4.362956565915498,
"learning_rate": 3.290852145462196e-05,
"loss": 0.8137,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8842595815658569,
"step": 590,
"valid_targets_mean": 99.6,
"valid_targets_min": 66
},
{
"epoch": 2.4688796680497926,
"grad_norm": 4.836605175013781,
"learning_rate": 3.274975545023242e-05,
"loss": 0.8062,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9917868375778198,
"step": 595,
"valid_targets_mean": 92.1,
"valid_targets_min": 61
},
{
"epoch": 2.4896265560165975,
"grad_norm": 4.355290465359613,
"learning_rate": 3.258962425241011e-05,
"loss": 0.8078,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9537732601165771,
"step": 600,
"valid_targets_mean": 113.8,
"valid_targets_min": 60
},
{
"epoch": 2.5103734439834025,
"grad_norm": 3.605491208223676,
"learning_rate": 3.242814500737092e-05,
"loss": 0.7583,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6148207187652588,
"step": 605,
"valid_targets_mean": 158.1,
"valid_targets_min": 37
},
{
"epoch": 2.5311203319502074,
"grad_norm": 4.3713729869832845,
"learning_rate": 3.226533500567433e-05,
"loss": 0.8238,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8725237846374512,
"step": 610,
"valid_targets_mean": 104.8,
"valid_targets_min": 58
},
{
"epoch": 2.5518672199170123,
"grad_norm": 4.489442159400303,
"learning_rate": 3.2101211680371965e-05,
"loss": 0.8316,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9407273530960083,
"step": 615,
"valid_targets_mean": 98.4,
"valid_targets_min": 57
},
{
"epoch": 2.572614107883817,
"grad_norm": 4.291817799537676,
"learning_rate": 3.193579260514097e-05,
"loss": 0.8382,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8661283254623413,
"step": 620,
"valid_targets_mean": 88.4,
"valid_targets_min": 62
},
{
"epoch": 2.5933609958506225,
"grad_norm": 3.951278391034252,
"learning_rate": 3.176909549240226e-05,
"loss": 0.797,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6531973481178284,
"step": 625,
"valid_targets_mean": 127.5,
"valid_targets_min": 56
},
{
"epoch": 2.6141078838174274,
"grad_norm": 3.6399446812749994,
"learning_rate": 3.1601138191423966e-05,
"loss": 0.7022,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.821446418762207,
"step": 630,
"valid_targets_mean": 158.7,
"valid_targets_min": 64
},
{
"epoch": 2.6348547717842323,
"grad_norm": 3.783299086234017,
"learning_rate": 3.143193868641019e-05,
"loss": 0.754,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7142607569694519,
"step": 635,
"valid_targets_mean": 128.2,
"valid_targets_min": 61
},
{
"epoch": 2.6556016597510372,
"grad_norm": 4.253882499474148,
"learning_rate": 3.1261515094575335e-05,
"loss": 0.7797,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8119747638702393,
"step": 640,
"valid_targets_mean": 93.5,
"valid_targets_min": 59
},
{
"epoch": 2.6763485477178426,
"grad_norm": 4.113442911406122,
"learning_rate": 3.108988566420417e-05,
"loss": 0.6938,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7523815631866455,
"step": 645,
"valid_targets_mean": 119.0,
"valid_targets_min": 64
},
{
"epoch": 2.6970954356846475,
"grad_norm": 3.6131999603541405,
"learning_rate": 3.0917068772697934e-05,
"loss": 0.7476,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6980721950531006,
"step": 650,
"valid_targets_mean": 146.2,
"valid_targets_min": 78
},
{
"epoch": 2.7178423236514524,
"grad_norm": 3.732929812664138,
"learning_rate": 3.074308292460646e-05,
"loss": 0.6439,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8016985654830933,
"step": 655,
"valid_targets_mean": 153.6,
"valid_targets_min": 57
},
{
"epoch": 2.7385892116182573,
"grad_norm": 3.0389793975855532,
"learning_rate": 3.056794674964685e-05,
"loss": 0.6437,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5522057414054871,
"step": 660,
"valid_targets_mean": 211.7,
"valid_targets_min": 72
},
{
"epoch": 2.759336099585062,
"grad_norm": 3.985890415354317,
"learning_rate": 3.0391679000708673e-05,
"loss": 0.7794,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7437437176704407,
"step": 665,
"valid_targets_mean": 115.2,
"valid_targets_min": 55
},
{
"epoch": 2.780082987551867,
"grad_norm": 4.76701659295223,
"learning_rate": 3.0214298551845967e-05,
"loss": 0.7425,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.880343496799469,
"step": 670,
"valid_targets_mean": 113.9,
"valid_targets_min": 71
},
{
"epoch": 2.800829875518672,
"grad_norm": 3.7563533860394323,
"learning_rate": 3.0035824396256267e-05,
"loss": 0.799,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7248696684837341,
"step": 675,
"valid_targets_mean": 106.1,
"valid_targets_min": 61
},
{
"epoch": 2.821576763485477,
"grad_norm": 3.761242240389271,
"learning_rate": 2.9856275644246903e-05,
"loss": 0.7134,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8497253656387329,
"step": 680,
"valid_targets_mean": 139.9,
"valid_targets_min": 69
},
{
"epoch": 2.8423236514522823,
"grad_norm": 4.290575353844001,
"learning_rate": 2.9675671521188766e-05,
"loss": 0.7767,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8362669944763184,
"step": 685,
"valid_targets_mean": 109.0,
"valid_targets_min": 57
},
{
"epoch": 2.863070539419087,
"grad_norm": 4.343959204903679,
"learning_rate": 2.949403136545769e-05,
"loss": 0.8742,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.9363041520118713,
"step": 690,
"valid_targets_mean": 118.2,
"valid_targets_min": 63
},
{
"epoch": 2.883817427385892,
"grad_norm": 3.8671929065161064,
"learning_rate": 2.9311374626363793e-05,
"loss": 0.6692,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7386797666549683,
"step": 695,
"valid_targets_mean": 117.1,
"valid_targets_min": 62
},
{
"epoch": 2.904564315352697,
"grad_norm": 2.4421362640071878,
"learning_rate": 2.9127720862068928e-05,
"loss": 0.7173,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.43712466955184937,
"step": 700,
"valid_targets_mean": 225.9,
"valid_targets_min": 63
},
{
"epoch": 2.9253112033195023,
"grad_norm": 3.79384332274057,
"learning_rate": 2.8943089737492465e-05,
"loss": 0.7826,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6691803932189941,
"step": 705,
"valid_targets_mean": 134.9,
"valid_targets_min": 58
},
{
"epoch": 2.9460580912863072,
"grad_norm": 5.106290015763064,
"learning_rate": 2.8757501022205653e-05,
"loss": 0.7881,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6905944347381592,
"step": 710,
"valid_targets_mean": 113.7,
"valid_targets_min": 65
},
{
"epoch": 2.966804979253112,
"grad_norm": 2.3434102426062147,
"learning_rate": 2.8570974588314767e-05,
"loss": 0.638,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38373029232025146,
"step": 715,
"valid_targets_mean": 209.8,
"valid_targets_min": 68
},
{
"epoch": 2.987551867219917,
"grad_norm": 4.177527208829087,
"learning_rate": 2.8383530408333285e-05,
"loss": 0.7669,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.8700437545776367,
"step": 720,
"valid_targets_mean": 106.6,
"valid_targets_min": 65
},
{
"epoch": 3.008298755186722,
"grad_norm": 3.3708836828475635,
"learning_rate": 2.8195188553043317e-05,
"loss": 0.6422,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5712267160415649,
"step": 725,
"valid_targets_mean": 125.0,
"valid_targets_min": 48
},
{
"epoch": 3.029045643153527,
"grad_norm": 4.331342732584377,
"learning_rate": 2.800596918934648e-05,
"loss": 0.5641,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6667015552520752,
"step": 730,
"valid_targets_mean": 108.5,
"valid_targets_min": 57
},
{
"epoch": 3.0497925311203318,
"grad_norm": 7.744354080794957,
"learning_rate": 2.7815892578104554e-05,
"loss": 0.6125,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6798431873321533,
"step": 735,
"valid_targets_mean": 95.9,
"valid_targets_min": 46
},
{
"epoch": 3.070539419087137,
"grad_norm": 4.7346730931654974,
"learning_rate": 2.762497907196996e-05,
"loss": 0.4968,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4088674783706665,
"step": 740,
"valid_targets_mean": 149.1,
"valid_targets_min": 61
},
{
"epoch": 3.091286307053942,
"grad_norm": 3.8372167911375676,
"learning_rate": 2.743324911320655e-05,
"loss": 0.591,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4837598204612732,
"step": 745,
"valid_targets_mean": 155.9,
"valid_targets_min": 48
},
{
"epoch": 3.112033195020747,
"grad_norm": 4.928852485615693,
"learning_rate": 2.724072323150069e-05,
"loss": 0.5391,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6401872038841248,
"step": 750,
"valid_targets_mean": 109.9,
"valid_targets_min": 62
},
{
"epoch": 3.132780082987552,
"grad_norm": 5.373658977233531,
"learning_rate": 2.704742204176301e-05,
"loss": 0.5404,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.7401503324508667,
"step": 755,
"valid_targets_mean": 104.4,
"valid_targets_min": 75
},
{
"epoch": 3.1535269709543567,
"grad_norm": 4.695430238717871,
"learning_rate": 2.6853366241921083e-05,
"loss": 0.546,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.507964015007019,
"step": 760,
"valid_targets_mean": 131.6,
"valid_targets_min": 52
},
{
"epoch": 3.1742738589211617,
"grad_norm": 4.940895414049785,
"learning_rate": 2.6658576610703142e-05,
"loss": 0.5017,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.44809240102767944,
"step": 765,
"valid_targets_mean": 160.5,
"valid_targets_min": 59
},
{
"epoch": 3.195020746887967,
"grad_norm": 4.475353236399673,
"learning_rate": 2.6463074005413187e-05,
"loss": 0.4833,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4449211061000824,
"step": 770,
"valid_targets_mean": 152.4,
"valid_targets_min": 72
},
{
"epoch": 3.215767634854772,
"grad_norm": 4.684709304749235,
"learning_rate": 2.6266879359697647e-05,
"loss": 0.5946,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5302928686141968,
"step": 775,
"valid_targets_mean": 129.9,
"valid_targets_min": 72
},
{
"epoch": 3.236514522821577,
"grad_norm": 4.642771055970237,
"learning_rate": 2.6070013681303933e-05,
"loss": 0.5182,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5693954229354858,
"step": 780,
"valid_targets_mean": 121.1,
"valid_targets_min": 63
},
{
"epoch": 3.2572614107883817,
"grad_norm": 4.02364920565193,
"learning_rate": 2.5872498049830973e-05,
"loss": 0.5421,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4645788073539734,
"step": 785,
"valid_targets_mean": 152.3,
"valid_targets_min": 61
},
{
"epoch": 3.2780082987551866,
"grad_norm": 4.552089103946345,
"learning_rate": 2.5674353614472084e-05,
"loss": 0.4724,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5232946276664734,
"step": 790,
"valid_targets_mean": 110.2,
"valid_targets_min": 63
},
{
"epoch": 3.2987551867219915,
"grad_norm": 3.7131472795548244,
"learning_rate": 2.5475601591750448e-05,
"loss": 0.4999,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.40184569358825684,
"step": 795,
"valid_targets_mean": 262.1,
"valid_targets_min": 49
},
{
"epoch": 3.3195020746887964,
"grad_norm": 5.143076340286301,
"learning_rate": 2.5276263263247282e-05,
"loss": 0.5116,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.527146577835083,
"step": 800,
"valid_targets_mean": 110.8,
"valid_targets_min": 62
},
{
"epoch": 3.340248962655602,
"grad_norm": 5.319558834403425,
"learning_rate": 2.5076359973323107e-05,
"loss": 0.6154,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6301993131637573,
"step": 805,
"valid_targets_mean": 96.8,
"valid_targets_min": 56
},
{
"epoch": 3.3609958506224067,
"grad_norm": 3.8762909428008725,
"learning_rate": 2.4875913126832297e-05,
"loss": 0.5952,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4537183940410614,
"step": 810,
"valid_targets_mean": 173.0,
"valid_targets_min": 57
},
{
"epoch": 3.3817427385892116,
"grad_norm": 7.622394767794957,
"learning_rate": 2.4674944186831108e-05,
"loss": 0.4865,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.46812310814857483,
"step": 815,
"valid_targets_mean": 114.8,
"valid_targets_min": 51
},
{
"epoch": 3.4024896265560165,
"grad_norm": 4.432218117678233,
"learning_rate": 2.4473474672279497e-05,
"loss": 0.504,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4488857686519623,
"step": 820,
"valid_targets_mean": 124.0,
"valid_targets_min": 61
},
{
"epoch": 3.4232365145228214,
"grad_norm": 4.55996161463432,
"learning_rate": 2.427152615573697e-05,
"loss": 0.5078,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.34916043281555176,
"step": 825,
"valid_targets_mean": 176.0,
"valid_targets_min": 59
},
{
"epoch": 3.4439834024896268,
"grad_norm": 5.114951004982656,
"learning_rate": 2.4069120261052682e-05,
"loss": 0.5275,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6819808483123779,
"step": 830,
"valid_targets_mean": 105.2,
"valid_targets_min": 71
},
{
"epoch": 3.4647302904564317,
"grad_norm": 5.28455932841826,
"learning_rate": 2.386627866105002e-05,
"loss": 0.5831,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6226446628570557,
"step": 835,
"valid_targets_mean": 134.7,
"valid_targets_min": 60
},
{
"epoch": 3.4854771784232366,
"grad_norm": 4.473738543991443,
"learning_rate": 2.3663023075205992e-05,
"loss": 0.5253,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5064611434936523,
"step": 840,
"valid_targets_mean": 167.0,
"valid_targets_min": 76
},
{
"epoch": 3.5062240663900415,
"grad_norm": 6.42051749883382,
"learning_rate": 2.3459375267325552e-05,
"loss": 0.5637,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.517977237701416,
"step": 845,
"valid_targets_mean": 111.2,
"valid_targets_min": 54
},
{
"epoch": 3.5269709543568464,
"grad_norm": 4.595241350402068,
"learning_rate": 2.325535704321126e-05,
"loss": 0.612,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5615861415863037,
"step": 850,
"valid_targets_mean": 147.6,
"valid_targets_min": 70
},
{
"epoch": 3.5477178423236513,
"grad_norm": 5.192972612859483,
"learning_rate": 2.3050990248328365e-05,
"loss": 0.5303,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5923424959182739,
"step": 855,
"valid_targets_mean": 103.9,
"valid_targets_min": 20
},
{
"epoch": 3.568464730290456,
"grad_norm": 5.3495277210724295,
"learning_rate": 2.2846296765465708e-05,
"loss": 0.5785,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6900249719619751,
"step": 860,
"valid_targets_mean": 98.2,
"valid_targets_min": 48
},
{
"epoch": 3.5892116182572615,
"grad_norm": 4.61930376588509,
"learning_rate": 2.2641298512392585e-05,
"loss": 0.5289,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5852305293083191,
"step": 865,
"valid_targets_mean": 107.1,
"valid_targets_min": 68
},
{
"epoch": 3.6099585062240664,
"grad_norm": 5.361219727707543,
"learning_rate": 2.2436017439511878e-05,
"loss": 0.519,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.630679190158844,
"step": 870,
"valid_targets_mean": 99.9,
"valid_targets_min": 48
},
{
"epoch": 3.6307053941908713,
"grad_norm": 4.72558048492487,
"learning_rate": 2.2230475527509712e-05,
"loss": 0.5643,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5127944350242615,
"step": 875,
"valid_targets_mean": 135.3,
"valid_targets_min": 69
},
{
"epoch": 3.6514522821576763,
"grad_norm": 5.72644206689818,
"learning_rate": 2.2024694785001814e-05,
"loss": 0.5333,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5506484508514404,
"step": 880,
"valid_targets_mean": 91.6,
"valid_targets_min": 56
},
{
"epoch": 3.6721991701244816,
"grad_norm": 5.396922771454308,
"learning_rate": 2.1818697246176943e-05,
"loss": 0.5145,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6444652676582336,
"step": 885,
"valid_targets_mean": 89.5,
"valid_targets_min": 69
},
{
"epoch": 3.6929460580912865,
"grad_norm": 4.709143641817256,
"learning_rate": 2.161250496843756e-05,
"loss": 0.6124,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5286765098571777,
"step": 890,
"valid_targets_mean": 118.1,
"valid_targets_min": 61
},
{
"epoch": 3.7136929460580914,
"grad_norm": 4.279711593845015,
"learning_rate": 2.1406140030037988e-05,
"loss": 0.5131,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.508799135684967,
"step": 895,
"valid_targets_mean": 155.9,
"valid_targets_min": 47
},
{
"epoch": 3.7344398340248963,
"grad_norm": 5.507064174030223,
"learning_rate": 2.119962452772039e-05,
"loss": 0.6005,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6413220763206482,
"step": 900,
"valid_targets_mean": 102.9,
"valid_targets_min": 51
},
{
"epoch": 3.7551867219917012,
"grad_norm": 4.977065659139609,
"learning_rate": 2.0992980574348687e-05,
"loss": 0.637,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.660423994064331,
"step": 905,
"valid_targets_mean": 114.2,
"valid_targets_min": 63
},
{
"epoch": 3.775933609958506,
"grad_norm": 5.507287449132665,
"learning_rate": 2.0786230296540864e-05,
"loss": 0.562,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6502384543418884,
"step": 910,
"valid_targets_mean": 95.9,
"valid_targets_min": 57
},
{
"epoch": 3.796680497925311,
"grad_norm": 3.542244490137563,
"learning_rate": 2.0579395832299688e-05,
"loss": 0.4917,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38029953837394714,
"step": 915,
"valid_targets_mean": 189.2,
"valid_targets_min": 52
},
{
"epoch": 3.817427385892116,
"grad_norm": 5.057195244837334,
"learning_rate": 2.0372499328642277e-05,
"loss": 0.6087,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6501518487930298,
"step": 920,
"valid_targets_mean": 115.8,
"valid_targets_min": 55
},
{
"epoch": 3.8381742738589213,
"grad_norm": 3.87850879824177,
"learning_rate": 2.016556293922869e-05,
"loss": 0.54,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.45556652545928955,
"step": 925,
"valid_targets_mean": 136.7,
"valid_targets_min": 61
},
{
"epoch": 3.858921161825726,
"grad_norm": 4.15102620173551,
"learning_rate": 1.9958608821989792e-05,
"loss": 0.5028,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5197992324829102,
"step": 930,
"valid_targets_mean": 172.0,
"valid_targets_min": 43
},
{
"epoch": 3.879668049792531,
"grad_norm": 3.9883443533699623,
"learning_rate": 1.9751659136754686e-05,
"loss": 0.4334,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4564608335494995,
"step": 935,
"valid_targets_mean": 171.6,
"valid_targets_min": 68
},
{
"epoch": 3.900414937759336,
"grad_norm": 3.8229294981435196,
"learning_rate": 1.9544736042877886e-05,
"loss": 0.3988,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3890151381492615,
"step": 940,
"valid_targets_mean": 203.3,
"valid_targets_min": 58
},
{
"epoch": 3.921161825726141,
"grad_norm": 4.770214843535073,
"learning_rate": 1.9337861696866643e-05,
"loss": 0.5395,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5733540058135986,
"step": 945,
"valid_targets_mean": 121.2,
"valid_targets_min": 58
},
{
"epoch": 3.9419087136929463,
"grad_norm": 5.715733351887051,
"learning_rate": 1.913105825000844e-05,
"loss": 0.6749,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.6428124904632568,
"step": 950,
"valid_targets_mean": 99.4,
"valid_targets_min": 63
},
{
"epoch": 3.962655601659751,
"grad_norm": 5.247147480424706,
"learning_rate": 1.8924347845999197e-05,
"loss": 0.5692,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4043826162815094,
"step": 955,
"valid_targets_mean": 186.8,
"valid_targets_min": 55
},
{
"epoch": 3.983402489626556,
"grad_norm": 4.693398960678804,
"learning_rate": 1.871775261857215e-05,
"loss": 0.5391,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5111123323440552,
"step": 960,
"valid_targets_mean": 134.2,
"valid_targets_min": 49
},
{
"epoch": 4.004149377593361,
"grad_norm": 4.562791344546384,
"learning_rate": 1.8511294689127887e-05,
"loss": 0.5623,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4835764467716217,
"step": 965,
"valid_targets_mean": 91.6,
"valid_targets_min": 46
},
{
"epoch": 4.024896265560166,
"grad_norm": 4.231590917124419,
"learning_rate": 1.830499616436567e-05,
"loss": 0.4125,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3629613518714905,
"step": 970,
"valid_targets_mean": 133.6,
"valid_targets_min": 56
},
{
"epoch": 4.045643153526971,
"grad_norm": 6.451063449554833,
"learning_rate": 1.8098879133916352e-05,
"loss": 0.4243,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4441721737384796,
"step": 975,
"valid_targets_mean": 104.6,
"valid_targets_min": 48
},
{
"epoch": 4.066390041493776,
"grad_norm": 7.27315245214559,
"learning_rate": 1.789296566797706e-05,
"loss": 0.4303,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.44909948110580444,
"step": 980,
"valid_targets_mean": 116.8,
"valid_targets_min": 59
},
{
"epoch": 4.087136929460581,
"grad_norm": 5.423253701096964,
"learning_rate": 1.768727781494807e-05,
"loss": 0.418,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4044135510921478,
"step": 985,
"valid_targets_mean": 116.2,
"valid_targets_min": 59
},
{
"epoch": 4.1078838174273855,
"grad_norm": 6.060822859449416,
"learning_rate": 1.7481837599071903e-05,
"loss": 0.3992,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4017379879951477,
"step": 990,
"valid_targets_mean": 126.2,
"valid_targets_min": 63
},
{
"epoch": 4.12863070539419,
"grad_norm": 6.047434247325548,
"learning_rate": 1.7276667018075073e-05,
"loss": 0.379,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4426099956035614,
"step": 995,
"valid_targets_mean": 105.3,
"valid_targets_min": 68
},
{
"epoch": 4.149377593360996,
"grad_norm": 4.238813714812382,
"learning_rate": 1.7071788040812655e-05,
"loss": 0.3745,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.30655837059020996,
"step": 1000,
"valid_targets_mean": 140.9,
"valid_targets_min": 59
},
{
"epoch": 4.170124481327801,
"grad_norm": 6.8069774791469015,
"learning_rate": 1.686722260491597e-05,
"loss": 0.3669,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.36141982674598694,
"step": 1005,
"valid_targets_mean": 105.2,
"valid_targets_min": 55
},
{
"epoch": 4.190871369294606,
"grad_norm": 6.0106894978076975,
"learning_rate": 1.6662992614443525e-05,
"loss": 0.3601,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.44953832030296326,
"step": 1010,
"valid_targets_mean": 99.1,
"valid_targets_min": 50
},
{
"epoch": 4.211618257261411,
"grad_norm": 4.13388889521017,
"learning_rate": 1.6459119937535702e-05,
"loss": 0.3598,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2533540725708008,
"step": 1015,
"valid_targets_mean": 210.5,
"valid_targets_min": 42
},
{
"epoch": 4.232365145228216,
"grad_norm": 4.270769059639924,
"learning_rate": 1.6255626404073132e-05,
"loss": 0.3857,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.35333698987960815,
"step": 1020,
"valid_targets_mean": 151.1,
"valid_targets_min": 71
},
{
"epoch": 4.253112033195021,
"grad_norm": 3.928272352429347,
"learning_rate": 1.605253380333927e-05,
"loss": 0.3545,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23659725487232208,
"step": 1025,
"valid_targets_mean": 209.1,
"valid_targets_min": 58
},
{
"epoch": 4.273858921161826,
"grad_norm": 6.520862768289659,
"learning_rate": 1.584986388168728e-05,
"loss": 0.327,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3713211119174957,
"step": 1030,
"valid_targets_mean": 106.5,
"valid_targets_min": 68
},
{
"epoch": 4.2946058091286305,
"grad_norm": 6.088618896351517,
"learning_rate": 1.5647638340211525e-05,
"loss": 0.4217,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4184168577194214,
"step": 1035,
"valid_targets_mean": 123.2,
"valid_targets_min": 57
},
{
"epoch": 4.3153526970954355,
"grad_norm": 6.739392196333286,
"learning_rate": 1.5445878832423876e-05,
"loss": 0.3972,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4932287633419037,
"step": 1040,
"valid_targets_mean": 93.1,
"valid_targets_min": 55
},
{
"epoch": 4.33609958506224,
"grad_norm": 4.816460152015017,
"learning_rate": 1.5244606961935187e-05,
"loss": 0.4227,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.33327189087867737,
"step": 1045,
"valid_targets_mean": 126.0,
"valid_targets_min": 63
},
{
"epoch": 4.356846473029045,
"grad_norm": 4.85922792161953,
"learning_rate": 1.5043844280142005e-05,
"loss": 0.3818,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3491777181625366,
"step": 1050,
"valid_targets_mean": 144.3,
"valid_targets_min": 66
},
{
"epoch": 4.377593360995851,
"grad_norm": 5.503313733342627,
"learning_rate": 1.4843612283918995e-05,
"loss": 0.3707,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3861674666404724,
"step": 1055,
"valid_targets_mean": 145.2,
"valid_targets_min": 68
},
{
"epoch": 4.398340248962656,
"grad_norm": 3.744381810663879,
"learning_rate": 1.4643932413317079e-05,
"loss": 0.3399,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2514806091785431,
"step": 1060,
"valid_targets_mean": 166.9,
"valid_targets_min": 69
},
{
"epoch": 4.419087136929461,
"grad_norm": 4.739596531448335,
"learning_rate": 1.4444826049267784e-05,
"loss": 0.3923,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.32454103231430054,
"step": 1065,
"valid_targets_mean": 124.9,
"valid_targets_min": 56
},
{
"epoch": 4.439834024896266,
"grad_norm": 5.668748766294243,
"learning_rate": 1.4246314511293777e-05,
"loss": 0.3607,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3947162330150604,
"step": 1070,
"valid_targets_mean": 139.6,
"valid_targets_min": 58
},
{
"epoch": 4.460580912863071,
"grad_norm": 5.417321720671694,
"learning_rate": 1.4048419055226146e-05,
"loss": 0.4178,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3876382112503052,
"step": 1075,
"valid_targets_mean": 137.8,
"valid_targets_min": 64
},
{
"epoch": 4.481327800829876,
"grad_norm": 5.060908418832114,
"learning_rate": 1.3851160870928317e-05,
"loss": 0.4052,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38143742084503174,
"step": 1080,
"valid_targets_mean": 117.8,
"valid_targets_min": 66
},
{
"epoch": 4.5020746887966805,
"grad_norm": 4.781433991361493,
"learning_rate": 1.3654561080027213e-05,
"loss": 0.343,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3337147533893585,
"step": 1085,
"valid_targets_mean": 156.2,
"valid_targets_min": 66
},
{
"epoch": 4.522821576763485,
"grad_norm": 6.7756272116807175,
"learning_rate": 1.345864073365157e-05,
"loss": 0.393,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.5130104422569275,
"step": 1090,
"valid_targets_mean": 103.8,
"valid_targets_min": 55
},
{
"epoch": 4.54356846473029,
"grad_norm": 6.000927566190182,
"learning_rate": 1.3263420810177902e-05,
"loss": 0.4336,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4507676661014557,
"step": 1095,
"valid_targets_mean": 105.2,
"valid_targets_min": 55
},
{
"epoch": 4.564315352697095,
"grad_norm": 6.058238312428124,
"learning_rate": 1.3068922212984188e-05,
"loss": 0.4127,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4390409588813782,
"step": 1100,
"valid_targets_mean": 97.3,
"valid_targets_min": 64
},
{
"epoch": 4.5850622406639,
"grad_norm": 4.203708752511155,
"learning_rate": 1.287516576821167e-05,
"loss": 0.4029,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.332113116979599,
"step": 1105,
"valid_targets_mean": 178.4,
"valid_targets_min": 69
},
{
"epoch": 4.605809128630705,
"grad_norm": 5.207374310292064,
"learning_rate": 1.2682172222534805e-05,
"loss": 0.3313,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4056367874145508,
"step": 1110,
"valid_targets_mean": 100.6,
"valid_targets_min": 60
},
{
"epoch": 4.62655601659751,
"grad_norm": 4.98336947480142,
"learning_rate": 1.2489962240939857e-05,
"loss": 0.3613,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2771909534931183,
"step": 1115,
"valid_targets_mean": 159.0,
"valid_targets_min": 61
},
{
"epoch": 4.647302904564316,
"grad_norm": 3.6594235056108,
"learning_rate": 1.229855640451213e-05,
"loss": 0.3179,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16788652539253235,
"step": 1120,
"valid_targets_mean": 306.2,
"valid_targets_min": 68
},
{
"epoch": 4.668049792531121,
"grad_norm": 8.687907658737734,
"learning_rate": 1.2107975208232259e-05,
"loss": 0.4198,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.379871666431427,
"step": 1125,
"valid_targets_mean": 133.1,
"valid_targets_min": 61
},
{
"epoch": 4.6887966804979255,
"grad_norm": 5.836548120395633,
"learning_rate": 1.1918239058781636e-05,
"loss": 0.3607,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3218669891357422,
"step": 1130,
"valid_targets_mean": 127.9,
"valid_targets_min": 70
},
{
"epoch": 4.70954356846473,
"grad_norm": 4.201638283395644,
"learning_rate": 1.1729368272357419e-05,
"loss": 0.3285,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23533934354782104,
"step": 1135,
"valid_targets_mean": 199.4,
"valid_targets_min": 64
},
{
"epoch": 4.730290456431535,
"grad_norm": 5.095855157677563,
"learning_rate": 1.1541383072497077e-05,
"loss": 0.3855,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.38103365898132324,
"step": 1140,
"valid_targets_mean": 140.2,
"valid_targets_min": 59
},
{
"epoch": 4.75103734439834,
"grad_norm": 6.043002128231546,
"learning_rate": 1.1354303587913003e-05,
"loss": 0.337,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.39897701144218445,
"step": 1145,
"valid_targets_mean": 124.3,
"valid_targets_min": 51
},
{
"epoch": 4.771784232365145,
"grad_norm": 5.958399057291772,
"learning_rate": 1.1168149850337136e-05,
"loss": 0.3401,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.44645410776138306,
"step": 1150,
"valid_targets_mean": 117.4,
"valid_targets_min": 55
},
{
"epoch": 4.79253112033195,
"grad_norm": 4.690423833511766,
"learning_rate": 1.0982941792376125e-05,
"loss": 0.3953,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.35815998911857605,
"step": 1155,
"valid_targets_mean": 129.4,
"valid_targets_min": 57
},
{
"epoch": 4.813278008298755,
"grad_norm": 5.286714186295266,
"learning_rate": 1.0798699245376959e-05,
"loss": 0.41,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.34337374567985535,
"step": 1160,
"valid_targets_mean": 98.7,
"valid_targets_min": 50
},
{
"epoch": 4.83402489626556,
"grad_norm": 4.8262261611258985,
"learning_rate": 1.0615441937303534e-05,
"loss": 0.3275,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.32752180099487305,
"step": 1165,
"valid_targets_mean": 132.2,
"valid_targets_min": 69
},
{
"epoch": 4.854771784232365,
"grad_norm": 5.7535756694573665,
"learning_rate": 1.0433189490624253e-05,
"loss": 0.3779,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4661560356616974,
"step": 1170,
"valid_targets_mean": 97.8,
"valid_targets_min": 64
},
{
"epoch": 4.875518672199171,
"grad_norm": 4.2819905844779775,
"learning_rate": 1.0251961420210937e-05,
"loss": 0.3465,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.33195045590400696,
"step": 1175,
"valid_targets_mean": 179.1,
"valid_targets_min": 72
},
{
"epoch": 4.8962655601659755,
"grad_norm": 3.688190995957866,
"learning_rate": 1.0071777131249237e-05,
"loss": 0.3008,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2772628664970398,
"step": 1180,
"valid_targets_mean": 161.6,
"valid_targets_min": 72
},
{
"epoch": 4.91701244813278,
"grad_norm": 5.965502494236917,
"learning_rate": 9.892655917160814e-06,
"loss": 0.3767,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.4376785159111023,
"step": 1185,
"valid_targets_mean": 114.5,
"valid_targets_min": 63
},
{
"epoch": 4.937759336099585,
"grad_norm": 5.200859052929826,
"learning_rate": 9.714616957537466e-06,
"loss": 0.3356,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.35819631814956665,
"step": 1190,
"valid_targets_mean": 124.4,
"valid_targets_min": 53
},
{
"epoch": 4.95850622406639,
"grad_norm": 4.905971401096997,
"learning_rate": 9.537679316087491e-06,
"loss": 0.3299,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.33606231212615967,
"step": 1195,
"valid_targets_mean": 158.6,
"valid_targets_min": 69
},
{
"epoch": 4.979253112033195,
"grad_norm": 5.932174753958537,
"learning_rate": 9.361861938594332e-06,
"loss": 0.3432,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.44172829389572144,
"step": 1200,
"valid_targets_mean": 106.7,
"valid_targets_min": 62
},
{
"epoch": 5.0,
"grad_norm": 5.482278806871232,
"learning_rate": 9.187183650888056e-06,
"loss": 0.3926,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.35782986879348755,
"step": 1205,
"valid_targets_mean": 116.6,
"valid_targets_min": 46
},
{
"epoch": 5.020746887966805,
"grad_norm": 4.030939056748361,
"learning_rate": 9.013663156829438e-06,
"loss": 0.2897,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24931949377059937,
"step": 1210,
"valid_targets_mean": 123.8,
"valid_targets_min": 60
},
{
"epoch": 5.04149377593361,
"grad_norm": 3.738822362764231,
"learning_rate": 8.841319036307334e-06,
"loss": 0.253,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17903779447078705,
"step": 1215,
"valid_targets_mean": 230.9,
"valid_targets_min": 47
},
{
"epoch": 5.062240663900415,
"grad_norm": 7.551485917813854,
"learning_rate": 8.670169743249143e-06,
"loss": 0.2583,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.32046300172805786,
"step": 1220,
"valid_targets_mean": 85.4,
"valid_targets_min": 54
},
{
"epoch": 5.08298755186722,
"grad_norm": 5.108187510802179,
"learning_rate": 8.50023360364487e-06,
"loss": 0.2949,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.19512224197387695,
"step": 1225,
"valid_targets_mean": 146.3,
"valid_targets_min": 53
},
{
"epoch": 5.1037344398340245,
"grad_norm": 6.161943015701863,
"learning_rate": 8.331528813584832e-06,
"loss": 0.2688,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3473849594593048,
"step": 1230,
"valid_targets_mean": 102.7,
"valid_targets_min": 59
},
{
"epoch": 5.124481327800829,
"grad_norm": 5.4077143243461325,
"learning_rate": 8.164073437311315e-06,
"loss": 0.3074,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.329645037651062,
"step": 1235,
"valid_targets_mean": 109.8,
"valid_targets_min": 50
},
{
"epoch": 5.145228215767635,
"grad_norm": 3.9507322454113507,
"learning_rate": 7.997885405284305e-06,
"loss": 0.2572,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1828637272119522,
"step": 1240,
"valid_targets_mean": 137.2,
"valid_targets_min": 64
},
{
"epoch": 5.16597510373444,
"grad_norm": 4.418001305578677,
"learning_rate": 7.83298251226158e-06,
"loss": 0.2603,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22215382754802704,
"step": 1245,
"valid_targets_mean": 156.5,
"valid_targets_min": 64
},
{
"epoch": 5.186721991701245,
"grad_norm": 4.865514725466622,
"learning_rate": 7.669382415393298e-06,
"loss": 0.2724,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.21926715970039368,
"step": 1250,
"valid_targets_mean": 142.2,
"valid_targets_min": 62
},
{
"epoch": 5.20746887966805,
"grad_norm": 4.671773078750636,
"learning_rate": 7.507102632331382e-06,
"loss": 0.2,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.21363668143749237,
"step": 1255,
"valid_targets_mean": 119.7,
"valid_targets_min": 59
},
{
"epoch": 5.228215767634855,
"grad_norm": 7.747309257352988,
"learning_rate": 7.3461605393537415e-06,
"loss": 0.2709,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.36064910888671875,
"step": 1260,
"valid_targets_mean": 103.8,
"valid_targets_min": 54
},
{
"epoch": 5.24896265560166,
"grad_norm": 5.836162881737733,
"learning_rate": 7.186573369503731e-06,
"loss": 0.2643,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22864724695682526,
"step": 1265,
"valid_targets_mean": 155.4,
"valid_targets_min": 52
},
{
"epoch": 5.269709543568465,
"grad_norm": 6.109299844374152,
"learning_rate": 7.028358210744881e-06,
"loss": 0.2853,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3166709840297699,
"step": 1270,
"valid_targets_mean": 109.1,
"valid_targets_min": 51
},
{
"epoch": 5.29045643153527,
"grad_norm": 4.832159172426314,
"learning_rate": 6.8715320041312095e-06,
"loss": 0.2802,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3006090521812439,
"step": 1275,
"valid_targets_mean": 115.7,
"valid_targets_min": 66
},
{
"epoch": 5.3112033195020745,
"grad_norm": 3.182299417076334,
"learning_rate": 6.716111541993213e-06,
"loss": 0.2687,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1729133427143097,
"step": 1280,
"valid_targets_mean": 199.3,
"valid_targets_min": 68
},
{
"epoch": 5.331950207468879,
"grad_norm": 5.136124561619918,
"learning_rate": 6.562113466139836e-06,
"loss": 0.273,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2921794056892395,
"step": 1285,
"valid_targets_mean": 98.2,
"valid_targets_min": 55
},
{
"epoch": 5.352697095435684,
"grad_norm": 6.387158477899295,
"learning_rate": 6.4095542660765145e-06,
"loss": 0.2853,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.33419498801231384,
"step": 1290,
"valid_targets_mean": 112.9,
"valid_targets_min": 49
},
{
"epoch": 5.37344398340249,
"grad_norm": 4.174496355970623,
"learning_rate": 6.258450277239545e-06,
"loss": 0.2691,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2561451494693756,
"step": 1295,
"valid_targets_mean": 127.8,
"valid_targets_min": 56
},
{
"epoch": 5.394190871369295,
"grad_norm": 4.555697459816092,
"learning_rate": 6.108817679246979e-06,
"loss": 0.2528,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22688698768615723,
"step": 1300,
"valid_targets_mean": 113.3,
"valid_targets_min": 72
},
{
"epoch": 5.4149377593361,
"grad_norm": 3.7328921616493482,
"learning_rate": 5.960672494166113e-06,
"loss": 0.2519,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18989375233650208,
"step": 1305,
"valid_targets_mean": 173.1,
"valid_targets_min": 64
},
{
"epoch": 5.435684647302905,
"grad_norm": 4.693648138618525,
"learning_rate": 5.8140305847979895e-06,
"loss": 0.2879,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24130411446094513,
"step": 1310,
"valid_targets_mean": 112.8,
"valid_targets_min": 58
},
{
"epoch": 5.45643153526971,
"grad_norm": 6.1245996173536295,
"learning_rate": 5.668907652978783e-06,
"loss": 0.287,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3436064124107361,
"step": 1315,
"valid_targets_mean": 101.9,
"valid_targets_min": 65
},
{
"epoch": 5.477178423236515,
"grad_norm": 4.880336940097785,
"learning_rate": 5.5253192378985966e-06,
"loss": 0.2254,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2438402771949768,
"step": 1320,
"valid_targets_mean": 172.6,
"valid_targets_min": 56
},
{
"epoch": 5.4979253112033195,
"grad_norm": 5.184523533528152,
"learning_rate": 5.383280714437518e-06,
"loss": 0.2503,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3002464175224304,
"step": 1325,
"valid_targets_mean": 124.2,
"valid_targets_min": 41
},
{
"epoch": 5.518672199170124,
"grad_norm": 5.31824198148948,
"learning_rate": 5.242807291519374e-06,
"loss": 0.2999,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2796890139579773,
"step": 1330,
"valid_targets_mean": 105.6,
"valid_targets_min": 52
},
{
"epoch": 5.539419087136929,
"grad_norm": 4.406452877804828,
"learning_rate": 5.103914010483206e-06,
"loss": 0.1811,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.16050797700881958,
"step": 1335,
"valid_targets_mean": 169.6,
"valid_targets_min": 51
},
{
"epoch": 5.560165975103734,
"grad_norm": 5.526397123023929,
"learning_rate": 4.966615743472709e-06,
"loss": 0.2922,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.302845299243927,
"step": 1340,
"valid_targets_mean": 102.0,
"valid_targets_min": 61
},
{
"epoch": 5.580912863070539,
"grad_norm": 4.475088302223254,
"learning_rate": 4.830927191843779e-06,
"loss": 0.2274,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22871732711791992,
"step": 1345,
"valid_targets_mean": 112.8,
"valid_targets_min": 66
},
{
"epoch": 5.601659751037344,
"grad_norm": 3.827600528927148,
"learning_rate": 4.696862884590349e-06,
"loss": 0.2435,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17285335063934326,
"step": 1350,
"valid_targets_mean": 158.9,
"valid_targets_min": 56
},
{
"epoch": 5.622406639004149,
"grad_norm": 5.486601988943072,
"learning_rate": 4.564437176788681e-06,
"loss": 0.2842,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2588198482990265,
"step": 1355,
"valid_targets_mean": 116.6,
"valid_targets_min": 55
},
{
"epoch": 5.643153526970955,
"grad_norm": 4.279231303644842,
"learning_rate": 4.433664248060295e-06,
"loss": 0.229,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2061527669429779,
"step": 1360,
"valid_targets_mean": 153.4,
"valid_targets_min": 81
},
{
"epoch": 5.66390041493776,
"grad_norm": 5.487514913708906,
"learning_rate": 4.304558101053629e-06,
"loss": 0.2755,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23738700151443481,
"step": 1365,
"valid_targets_mean": 129.2,
"valid_targets_min": 53
},
{
"epoch": 5.6846473029045645,
"grad_norm": 4.05314290527769,
"learning_rate": 4.177132559944761e-06,
"loss": 0.2214,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.196598082780838,
"step": 1370,
"valid_targets_mean": 164.5,
"valid_targets_min": 64
},
{
"epoch": 5.7053941908713695,
"grad_norm": 6.2913597721448,
"learning_rate": 4.051401268957087e-06,
"loss": 0.27,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3582676351070404,
"step": 1375,
"valid_targets_mean": 107.6,
"valid_targets_min": 65
},
{
"epoch": 5.726141078838174,
"grad_norm": 6.031743246784628,
"learning_rate": 3.927377690900436e-06,
"loss": 0.2684,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25066938996315,
"step": 1380,
"valid_targets_mean": 127.6,
"valid_targets_min": 59
},
{
"epoch": 5.746887966804979,
"grad_norm": 5.214705770189645,
"learning_rate": 3.805075105729459e-06,
"loss": 0.2639,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25866299867630005,
"step": 1385,
"valid_targets_mean": 123.2,
"valid_targets_min": 55
},
{
"epoch": 5.767634854771784,
"grad_norm": 4.524404310503156,
"learning_rate": 3.6845066091216917e-06,
"loss": 0.2862,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23066316545009613,
"step": 1390,
"valid_targets_mean": 128.7,
"valid_targets_min": 63
},
{
"epoch": 5.788381742738589,
"grad_norm": 5.097443863683129,
"learning_rate": 3.56568511107533e-06,
"loss": 0.2365,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.28775590658187866,
"step": 1395,
"valid_targets_mean": 93.3,
"valid_targets_min": 55
},
{
"epoch": 5.809128630705394,
"grad_norm": 5.670717377040933,
"learning_rate": 3.448623334526853e-06,
"loss": 0.3138,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3401385545730591,
"step": 1400,
"valid_targets_mean": 94.4,
"valid_targets_min": 59
},
{
"epoch": 5.829875518672199,
"grad_norm": 4.62592863868813,
"learning_rate": 3.333333813988726e-06,
"loss": 0.2836,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24296891689300537,
"step": 1405,
"valid_targets_mean": 131.2,
"valid_targets_min": 55
},
{
"epoch": 5.850622406639004,
"grad_norm": 5.086749493331627,
"learning_rate": 3.219828894207242e-06,
"loss": 0.2612,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2650941014289856,
"step": 1410,
"valid_targets_mean": 124.6,
"valid_targets_min": 78
},
{
"epoch": 5.87136929460581,
"grad_norm": 4.85811557310476,
"learning_rate": 3.1081207288406846e-06,
"loss": 0.2353,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2586798667907715,
"step": 1415,
"valid_targets_mean": 111.1,
"valid_targets_min": 41
},
{
"epoch": 5.8921161825726145,
"grad_norm": 2.889073243641731,
"learning_rate": 2.9982212791580044e-06,
"loss": 0.2309,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15041622519493103,
"step": 1420,
"valid_targets_mean": 211.2,
"valid_targets_min": 67
},
{
"epoch": 5.912863070539419,
"grad_norm": 4.7851155337037925,
"learning_rate": 2.890142312757982e-06,
"loss": 0.2902,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.26748985052108765,
"step": 1425,
"valid_targets_mean": 115.7,
"valid_targets_min": 47
},
{
"epoch": 5.933609958506224,
"grad_norm": 4.580807918668048,
"learning_rate": 2.7838954023092845e-06,
"loss": 0.2352,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2532392144203186,
"step": 1430,
"valid_targets_mean": 122.4,
"valid_targets_min": 60
},
{
"epoch": 5.954356846473029,
"grad_norm": 5.472491432141598,
"learning_rate": 2.679491924311226e-06,
"loss": 0.254,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.27012646198272705,
"step": 1435,
"valid_targets_mean": 116.7,
"valid_targets_min": 57
},
{
"epoch": 5.975103734439834,
"grad_norm": 4.968702559787331,
"learning_rate": 2.576943057875696e-06,
"loss": 0.2627,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23716619610786438,
"step": 1440,
"valid_targets_mean": 146.7,
"valid_targets_min": 61
},
{
"epoch": 5.995850622406639,
"grad_norm": 3.6763933325276374,
"learning_rate": 2.4762597835300815e-06,
"loss": 0.2422,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17048585414886475,
"step": 1445,
"valid_targets_mean": 182.2,
"valid_targets_min": 62
},
{
"epoch": 6.016597510373444,
"grad_norm": 3.4439805728949118,
"learning_rate": 2.377452882041551e-06,
"loss": 0.2238,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.20334039628505707,
"step": 1450,
"valid_targets_mean": 121.8,
"valid_targets_min": 48
},
{
"epoch": 6.037344398340249,
"grad_norm": 4.169982176027096,
"learning_rate": 2.280532933262678e-06,
"loss": 0.1914,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22471293807029724,
"step": 1455,
"valid_targets_mean": 117.9,
"valid_targets_min": 55
},
{
"epoch": 6.058091286307054,
"grad_norm": 4.451395467337526,
"learning_rate": 2.1855103149985934e-06,
"loss": 0.2352,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.26145243644714355,
"step": 1460,
"valid_targets_mean": 95.1,
"valid_targets_min": 59
},
{
"epoch": 6.078838174273859,
"grad_norm": 4.720931193547068,
"learning_rate": 2.0923952018957826e-06,
"loss": 0.2003,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2084977924823761,
"step": 1465,
"valid_targets_mean": 112.5,
"valid_targets_min": 70
},
{
"epoch": 6.0995850622406635,
"grad_norm": 4.908093674427741,
"learning_rate": 2.0011975643526106e-06,
"loss": 0.185,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2433365285396576,
"step": 1470,
"valid_targets_mean": 120.3,
"valid_targets_min": 67
},
{
"epoch": 6.1203319502074685,
"grad_norm": 4.960759939432554,
"learning_rate": 1.9119271674517305e-06,
"loss": 0.1719,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2257917821407318,
"step": 1475,
"valid_targets_mean": 112.6,
"valid_targets_min": 61
},
{
"epoch": 6.141078838174274,
"grad_norm": 6.960952455067493,
"learning_rate": 1.8245935699145035e-06,
"loss": 0.2254,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2723103165626526,
"step": 1480,
"valid_targets_mean": 106.7,
"valid_targets_min": 66
},
{
"epoch": 6.161825726141079,
"grad_norm": 4.632315701623262,
"learning_rate": 1.7392061230774371e-06,
"loss": 0.1786,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23765739798545837,
"step": 1485,
"valid_targets_mean": 107.6,
"valid_targets_min": 65
},
{
"epoch": 6.182572614107884,
"grad_norm": 5.756675590953208,
"learning_rate": 1.6557739698909436e-06,
"loss": 0.2062,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.26241517066955566,
"step": 1490,
"valid_targets_mean": 104.4,
"valid_targets_min": 62
},
{
"epoch": 6.203319502074689,
"grad_norm": 3.871596830434273,
"learning_rate": 1.574306043940288e-06,
"loss": 0.1752,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15879438817501068,
"step": 1495,
"valid_targets_mean": 137.4,
"valid_targets_min": 58
},
{
"epoch": 6.224066390041494,
"grad_norm": 4.069174001456709,
"learning_rate": 1.4948110684890726e-06,
"loss": 0.2002,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12537790834903717,
"step": 1500,
"valid_targets_mean": 171.4,
"valid_targets_min": 64
},
{
"epoch": 6.244813278008299,
"grad_norm": 5.581161273740655,
"learning_rate": 1.4172975555451363e-06,
"loss": 0.2326,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2820102572441101,
"step": 1505,
"valid_targets_mean": 102.7,
"valid_targets_min": 68
},
{
"epoch": 6.265560165975104,
"grad_norm": 5.524451235658226,
"learning_rate": 1.3417738049491536e-06,
"loss": 0.2402,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.28135719895362854,
"step": 1510,
"valid_targets_mean": 96.0,
"valid_targets_min": 61
},
{
"epoch": 6.286307053941909,
"grad_norm": 4.869973029745467,
"learning_rate": 1.268247903485902e-06,
"loss": 0.198,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25688737630844116,
"step": 1515,
"valid_targets_mean": 101.5,
"valid_targets_min": 62
},
{
"epoch": 6.3070539419087135,
"grad_norm": 4.314906507746581,
"learning_rate": 1.1967277240183716e-06,
"loss": 0.21,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.191103994846344,
"step": 1520,
"valid_targets_mean": 120.8,
"valid_targets_min": 60
},
{
"epoch": 6.327800829875518,
"grad_norm": 4.3035104208842085,
"learning_rate": 1.1272209246447696e-06,
"loss": 0.2066,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17241765558719635,
"step": 1525,
"valid_targets_mean": 130.2,
"valid_targets_min": 55
},
{
"epoch": 6.348547717842323,
"grad_norm": 3.91462234843459,
"learning_rate": 1.0597349478785123e-06,
"loss": 0.199,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22027219831943512,
"step": 1530,
"valid_targets_mean": 157.9,
"valid_targets_min": 59
},
{
"epoch": 6.369294605809129,
"grad_norm": 4.161379790290467,
"learning_rate": 9.942770198513218e-07,
"loss": 0.2284,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.19286894798278809,
"step": 1535,
"valid_targets_mean": 130.5,
"valid_targets_min": 55
},
{
"epoch": 6.390041493775934,
"grad_norm": 3.979992570097402,
"learning_rate": 9.308541495394751e-07,
"loss": 0.2092,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1797044277191162,
"step": 1540,
"valid_targets_mean": 127.6,
"valid_targets_min": 63
},
{
"epoch": 6.410788381742739,
"grad_norm": 4.578593444010601,
"learning_rate": 8.694731280133051e-07,
"loss": 0.2104,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25974029302597046,
"step": 1545,
"valid_targets_mean": 120.4,
"valid_targets_min": 66
},
{
"epoch": 6.431535269709544,
"grad_norm": 14.736273677745213,
"learning_rate": 8.101405277100549e-07,
"loss": 0.2306,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2802034020423889,
"step": 1550,
"valid_targets_mean": 92.9,
"valid_targets_min": 61
},
{
"epoch": 6.452282157676349,
"grad_norm": 6.111463784813684,
"learning_rate": 7.528627017301016e-07,
"loss": 0.1801,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.29020482301712036,
"step": 1555,
"valid_targets_mean": 98.5,
"valid_targets_min": 69
},
{
"epoch": 6.473029045643154,
"grad_norm": 4.419863069023317,
"learning_rate": 6.976457831567262e-07,
"loss": 0.19,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22686433792114258,
"step": 1560,
"valid_targets_mean": 121.2,
"valid_targets_min": 57
},
{
"epoch": 6.4937759336099585,
"grad_norm": 5.3614306604918,
"learning_rate": 6.444956843993754e-07,
"loss": 0.2115,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24041424691677094,
"step": 1565,
"valid_targets_mean": 99.8,
"valid_targets_min": 52
},
{
"epoch": 6.514522821576763,
"grad_norm": 10.164349086292562,
"learning_rate": 5.934180965606007e-07,
"loss": 0.2254,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2823329567909241,
"step": 1570,
"valid_targets_mean": 99.8,
"valid_targets_min": 69
},
{
"epoch": 6.535269709543568,
"grad_norm": 5.2042900023987055,
"learning_rate": 5.444184888266768e-07,
"loss": 0.2009,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.244424968957901,
"step": 1575,
"valid_targets_mean": 88.3,
"valid_targets_min": 51
},
{
"epoch": 6.556016597510373,
"grad_norm": 4.462057874376993,
"learning_rate": 4.975021078819731e-07,
"loss": 0.1839,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22408881783485413,
"step": 1580,
"valid_targets_mean": 126.3,
"valid_targets_min": 50
},
{
"epoch": 6.576763485477178,
"grad_norm": 5.410684806378169,
"learning_rate": 4.5267397734717113e-07,
"loss": 0.2637,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.27906614542007446,
"step": 1585,
"valid_targets_mean": 92.4,
"valid_targets_min": 57
},
{
"epoch": 6.597510373443983,
"grad_norm": 4.119111648737928,
"learning_rate": 4.0993889724135314e-07,
"loss": 0.2036,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1697758138179779,
"step": 1590,
"valid_targets_mean": 143.6,
"valid_targets_min": 61
},
{
"epoch": 6.618257261410788,
"grad_norm": 5.179582255288888,
"learning_rate": 3.693014434680242e-07,
"loss": 0.2089,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.24674925208091736,
"step": 1595,
"valid_targets_mean": 138.8,
"valid_targets_min": 54
},
{
"epoch": 6.639004149377593,
"grad_norm": 5.561101995379885,
"learning_rate": 3.307659673251595e-07,
"loss": 0.2244,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.177412211894989,
"step": 1600,
"valid_targets_mean": 117.6,
"valid_targets_min": 60
},
{
"epoch": 6.659751037344399,
"grad_norm": 5.195861980789606,
"learning_rate": 2.9433659503926623e-07,
"loss": 0.2327,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25790226459503174,
"step": 1605,
"valid_targets_mean": 110.1,
"valid_targets_min": 53
},
{
"epoch": 6.680497925311204,
"grad_norm": 3.911906512279716,
"learning_rate": 2.6001722732358127e-07,
"loss": 0.2035,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14186334609985352,
"step": 1610,
"valid_targets_mean": 141.2,
"valid_targets_min": 67
},
{
"epoch": 6.7012448132780085,
"grad_norm": 5.042285108929284,
"learning_rate": 2.27811538960383e-07,
"loss": 0.1913,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.2191961407661438,
"step": 1615,
"valid_targets_mean": 123.7,
"valid_targets_min": 44
},
{
"epoch": 6.721991701244813,
"grad_norm": 5.239166381330431,
"learning_rate": 1.9772297840752407e-07,
"loss": 0.1967,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.19531618058681488,
"step": 1620,
"valid_targets_mean": 111.1,
"valid_targets_min": 61
},
{
"epoch": 6.742738589211618,
"grad_norm": 4.934480610164203,
"learning_rate": 1.6975476742916886e-07,
"loss": 0.2021,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.21878521144390106,
"step": 1625,
"valid_targets_mean": 101.4,
"valid_targets_min": 50
},
{
"epoch": 6.763485477178423,
"grad_norm": 4.895769325490044,
"learning_rate": 1.43909900750836e-07,
"loss": 0.2313,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.23424255847930908,
"step": 1630,
"valid_targets_mean": 120.0,
"valid_targets_min": 60
},
{
"epoch": 6.784232365145228,
"grad_norm": 3.8671281371679322,
"learning_rate": 1.2019114573871947e-07,
"loss": 0.1937,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.17272061109542847,
"step": 1635,
"valid_targets_mean": 148.1,
"valid_targets_min": 62
},
{
"epoch": 6.804979253112033,
"grad_norm": 4.4379640165652345,
"learning_rate": 9.860104210338562e-08,
"loss": 0.1573,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.18832272291183472,
"step": 1640,
"valid_targets_mean": 119.2,
"valid_targets_min": 70
},
{
"epoch": 6.825726141078838,
"grad_norm": 3.06556856776517,
"learning_rate": 7.914190162781277e-08,
"loss": 0.1822,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.12091746181249619,
"step": 1645,
"valid_targets_mean": 182.6,
"valid_targets_min": 54
},
{
"epoch": 6.846473029045643,
"grad_norm": 4.099005450803784,
"learning_rate": 6.181580791987385e-08,
"loss": 0.2129,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1655721664428711,
"step": 1650,
"valid_targets_mean": 148.9,
"valid_targets_min": 56
},
{
"epoch": 6.867219917012449,
"grad_norm": 4.310902974974436,
"learning_rate": 4.6624616189214765e-08,
"loss": 0.1678,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.15827462077140808,
"step": 1655,
"valid_targets_mean": 143.8,
"valid_targets_min": 61
},
{
"epoch": 6.8879668049792535,
"grad_norm": 5.3122495662963685,
"learning_rate": 3.3569953048624426e-08,
"loss": 0.1979,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.25778818130493164,
"step": 1660,
"valid_targets_mean": 134.8,
"valid_targets_min": 45
},
{
"epoch": 6.908713692946058,
"grad_norm": 7.693391989415569,
"learning_rate": 2.2653216339840746e-08,
"loss": 0.2183,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.21810811758041382,
"step": 1665,
"valid_targets_mean": 94.8,
"valid_targets_min": 20
},
{
"epoch": 6.929460580912863,
"grad_norm": 5.236163106459915,
"learning_rate": 1.3875574983894802e-08,
"loss": 0.1777,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.22439393401145935,
"step": 1670,
"valid_targets_mean": 117.6,
"valid_targets_min": 62
},
{
"epoch": 6.950207468879668,
"grad_norm": 3.8797498698175685,
"learning_rate": 7.237968855937638e-09,
"loss": 0.1835,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.14645695686340332,
"step": 1675,
"valid_targets_mean": 146.5,
"valid_targets_min": 64
},
{
"epoch": 6.970954356846473,
"grad_norm": 4.609262542128345,
"learning_rate": 2.7411086846051984e-09,
"loss": 0.2133,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.1896577924489975,
"step": 1680,
"valid_targets_mean": 98.5,
"valid_targets_min": 46
},
{
"epoch": 6.991701244813278,
"grad_norm": 2.6565371089348058,
"learning_rate": 3.8547597591254147e-10,
"loss": 0.2164,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.11129122227430344,
"step": 1685,
"valid_targets_mean": 239.2,
"valid_targets_min": 57
},
{
"epoch": 7.0,
"loss_nan_ranks": 0,
"loss_rank_avg": 0.3174639642238617,
"step": 1687,
"total_flos": 22338300149760.0,
"train_loss": 0.652452801803053,
"train_runtime": 3967.747,
"train_samples_per_second": 6.796,
"train_steps_per_second": 0.425,
"valid_targets_mean": 94.8,
"valid_targets_min": 62
}
],
"logging_steps": 5,
"max_steps": 1687,
"num_input_tokens_seen": 0,
"num_train_epochs": 7,
"save_steps": 200,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 22338300149760.0,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}