{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4151770268711798,
  "eval_steps": 5,
  "global_step": 4500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0,
      "eval_GEN Loss": 1.2385475635528564,
      "eval_GEN top-5 accuracy": 0.9149495835160018,
      "eval_PRM Accuracy": 0.2169811320754717,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.35658914728682173,
      "eval_PRM Loss": 0.6931641101837158,
      "eval_PRM NPV": 0.2169811320754717,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 1.8721153736114502,
      "eval_runtime": 55.5872,
      "eval_samples_per_second": 1.169,
      "eval_steps_per_second": 0.09,
      "step": 0
    },
    {
      "epoch": 9.226156152692884e-05,
      "grad_norm": 3.1269080771685585,
      "learning_rate": 3.109452736318408e-09,
      "loss": 1.6746,
      "step": 1
    },
    {
      "epoch": 0.00018452312305385768,
      "grad_norm": 3.5449940973885044,
      "learning_rate": 6.218905472636816e-09,
      "loss": 1.8898,
      "step": 2
    },
    {
      "epoch": 0.0002767846845807865,
      "grad_norm": 3.4603091206773233,
      "learning_rate": 9.328358208955224e-09,
      "loss": 1.5738,
      "step": 3
    },
    {
      "epoch": 0.00036904624610771536,
      "grad_norm": 4.577609127214851,
      "learning_rate": 1.2437810945273633e-08,
      "loss": 1.93,
      "step": 4
    },
    {
      "epoch": 0.0004613078076346442,
      "grad_norm": 3.117899016828671,
      "learning_rate": 1.554726368159204e-08,
      "loss": 1.8182,
      "step": 5
    },
    {
      "epoch": 0.0004613078076346442,
      "eval_GEN Loss": 1.2392983436584473,
      "eval_GEN top-5 accuracy": 0.9149495835160018,
      "eval_PRM Accuracy": 0.2169811320754717,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.35658914728682173,
      "eval_PRM Loss": 0.6931981444358826,
      "eval_PRM NPV": 0.2169811320754717,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 1.873557686805725,
      "eval_runtime": 55.9315,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 5
    },
    {
      "epoch": 0.000553569369161573,
      "grad_norm": 5.835077403604924,
      "learning_rate": 1.8656716417910447e-08,
      "loss": 1.8551,
      "step": 6
    },
    {
      "epoch": 0.0006458309306885019,
      "grad_norm": 4.210580345606465,
      "learning_rate": 2.176616915422886e-08,
      "loss": 1.728,
      "step": 7
    },
    {
      "epoch": 0.0007380924922154307,
      "grad_norm": 3.9832144368906084,
      "learning_rate": 2.4875621890547265e-08,
      "loss": 1.7872,
      "step": 8
    },
    {
      "epoch": 0.0008303540537423596,
      "grad_norm": 4.290185811653636,
      "learning_rate": 2.798507462686567e-08,
      "loss": 1.9864,
      "step": 9
    },
    {
      "epoch": 0.0009226156152692884,
      "grad_norm": 4.043726650440571,
      "learning_rate": 3.109452736318408e-08,
      "loss": 1.8669,
      "step": 10
    },
    {
      "epoch": 0.0009226156152692884,
      "eval_GEN Loss": 1.2399370670318604,
      "eval_GEN top-5 accuracy": 0.9140727750986409,
      "eval_PRM Accuracy": 0.2169811320754717,
      "eval_PRM F1": 0.0,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.35658914728682173,
      "eval_PRM Loss": 0.6931888461112976,
      "eval_PRM NPV": 0.2169811320754717,
      "eval_PRM Precision": 0.0,
      "eval_PRM Recall": 0.0,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 1.8730769157409668,
      "eval_runtime": 55.8813,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 10
    },
    {
      "epoch": 0.0010148771767962172,
      "grad_norm": 4.406640040740555,
      "learning_rate": 3.420398009950249e-08,
      "loss": 1.9705,
      "step": 11
    },
    {
      "epoch": 0.001107138738323146,
      "grad_norm": 3.001575732066938,
      "learning_rate": 3.7313432835820895e-08,
      "loss": 2.0106,
      "step": 12
    },
    {
      "epoch": 0.0011994002998500749,
      "grad_norm": 5.2380060824960655,
      "learning_rate": 4.042288557213931e-08,
      "loss": 1.9344,
      "step": 13
    },
    {
      "epoch": 0.0012916618613770037,
      "grad_norm": 4.727236576460403,
      "learning_rate": 4.353233830845772e-08,
      "loss": 1.8549,
      "step": 14
    },
    {
      "epoch": 0.0013839234229039326,
      "grad_norm": 3.570054790174049,
      "learning_rate": 4.664179104477612e-08,
      "loss": 1.7543,
      "step": 15
    },
    {
      "epoch": 0.0013839234229039326,
      "eval_GEN Loss": 1.2388570308685303,
      "eval_GEN top-5 accuracy": 0.9149495835160018,
      "eval_PRM Accuracy": 0.22641509433962265,
      "eval_PRM F1": 0.023809523809523808,
      "eval_PRM F1 AUC": 0.5060240963855421,
      "eval_PRM F1 Neg": 0.359375,
      "eval_PRM Loss": 0.6931596994400024,
      "eval_PRM NPV": 0.21904761904761905,
      "eval_PRM Precision": 1.0,
      "eval_PRM Recall": 0.012048192771084338,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 1.8725961446762085,
      "eval_runtime": 56.0124,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 15
    },
    {
      "epoch": 0.0014761849844308614,
      "grad_norm": 4.270529049520865,
      "learning_rate": 4.975124378109453e-08,
      "loss": 2.2358,
      "step": 16
    },
    {
      "epoch": 0.0015684465459577903,
      "grad_norm": 2.6485702290657556,
      "learning_rate": 5.286069651741294e-08,
      "loss": 2.0089,
      "step": 17
    },
    {
      "epoch": 0.0016607081074847191,
      "grad_norm": 5.330399703975629,
      "learning_rate": 5.597014925373134e-08,
      "loss": 1.7244,
      "step": 18
    },
    {
      "epoch": 0.001752969669011648,
      "grad_norm": 4.505723463929839,
      "learning_rate": 5.9079601990049754e-08,
      "loss": 1.9108,
      "step": 19
    },
    {
      "epoch": 0.0018452312305385769,
      "grad_norm": 3.5337375234196995,
      "learning_rate": 6.218905472636817e-08,
      "loss": 1.7413,
      "step": 20
    },
    {
      "epoch": 0.0018452312305385769,
      "eval_GEN Loss": 1.2375545501708984,
      "eval_GEN top-5 accuracy": 0.9145111793073214,
      "eval_PRM Accuracy": 0.25471698113207547,
      "eval_PRM F1": 0.09195402298850575,
      "eval_PRM F1 AUC": 0.5240963855421686,
      "eval_PRM F1 Neg": 0.368,
      "eval_PRM Loss": 0.693073034286499,
      "eval_PRM NPV": 0.22549019607843138,
      "eval_PRM Precision": 1.0,
      "eval_PRM Recall": 0.04819277108433735,
      "eval_PRM Specificty": 1.0,
      "eval_loss": 1.8711538314819336,
      "eval_runtime": 55.6906,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 20
    },
    {
      "epoch": 0.0019374927920655057,
      "grad_norm": 4.773455261650772,
      "learning_rate": 6.529850746268657e-08,
      "loss": 1.5142,
      "step": 21
    },
    {
      "epoch": 0.0020297543535924343,
      "grad_norm": 4.704180183206908,
      "learning_rate": 6.840796019900498e-08,
      "loss": 1.9425,
      "step": 22
    },
    {
      "epoch": 0.002122015915119363,
      "grad_norm": 5.212381712859903,
      "learning_rate": 7.151741293532339e-08,
      "loss": 1.7149,
      "step": 23
    },
    {
      "epoch": 0.002214277476646292,
      "grad_norm": 4.804344322732823,
      "learning_rate": 7.462686567164179e-08,
      "loss": 1.7484,
      "step": 24
    },
    {
      "epoch": 0.002306539038173221,
      "grad_norm": 3.4852690113725977,
      "learning_rate": 7.77363184079602e-08,
      "loss": 2.1117,
      "step": 25
    },
    {
      "epoch": 0.002306539038173221,
      "eval_GEN Loss": 1.2359493970870972,
      "eval_GEN top-5 accuracy": 0.9149495835160018,
      "eval_PRM Accuracy": 0.3113207547169811,
      "eval_PRM F1": 0.23157894736842105,
      "eval_PRM F1 AUC": 0.5445259298061813,
      "eval_PRM F1 Neg": 0.37606837606837606,
      "eval_PRM Loss": 0.6930320262908936,
      "eval_PRM NPV": 0.23404255319148937,
      "eval_PRM Precision": 0.9166666666666666,
      "eval_PRM Recall": 0.13253012048192772,
      "eval_PRM Specificty": 0.9565217391304348,
      "eval_loss": 1.871634602546692,
      "eval_runtime": 56.0441,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 25
    },
    {
      "epoch": 0.0023988005997001498,
      "grad_norm": 5.621336774755585,
      "learning_rate": 8.084577114427861e-08,
      "loss": 2.0755,
      "step": 26
    },
    {
      "epoch": 0.0024910621612270786,
      "grad_norm": 5.278482792916371,
      "learning_rate": 8.395522388059703e-08,
      "loss": 1.853,
      "step": 27
    },
    {
      "epoch": 0.0025833237227540075,
      "grad_norm": 3.429298179970435,
      "learning_rate": 8.706467661691544e-08,
      "loss": 1.8945,
      "step": 28
    },
    {
      "epoch": 0.0026755852842809363,
      "grad_norm": 3.7644884791494992,
      "learning_rate": 9.017412935323384e-08,
      "loss": 1.9576,
      "step": 29
    },
    {
      "epoch": 0.002767846845807865,
      "grad_norm": 4.516102493942929,
      "learning_rate": 9.328358208955224e-08,
      "loss": 2.0927,
      "step": 30
    },
    {
      "epoch": 0.002767846845807865,
      "eval_GEN Loss": 1.228596806526184,
      "eval_GEN top-5 accuracy": 0.9140727750986409,
      "eval_PRM Accuracy": 0.42452830188679247,
      "eval_PRM F1": 0.48739495798319327,
      "eval_PRM F1 AUC": 0.5225248821372446,
      "eval_PRM F1 Neg": 0.34408602150537637,
      "eval_PRM Loss": 0.6929349303245544,
      "eval_PRM NPV": 0.22857142857142856,
      "eval_PRM Precision": 0.8055555555555556,
      "eval_PRM Recall": 0.3493975903614458,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 1.8620191812515259,
      "eval_runtime": 55.8959,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 30
    },
    {
      "epoch": 0.002860108407334794,
      "grad_norm": 2.5973567522779573,
      "learning_rate": 9.639303482587065e-08,
      "loss": 1.9186,
      "step": 31
    },
    {
      "epoch": 0.002952369968861723,
      "grad_norm": 3.964938026808683,
      "learning_rate": 9.950248756218906e-08,
      "loss": 1.9025,
      "step": 32
    },
    {
      "epoch": 0.0030446315303886517,
      "grad_norm": 4.359962855655254,
      "learning_rate": 1.0261194029850747e-07,
      "loss": 1.6659,
      "step": 33
    },
    {
      "epoch": 0.0031368930919155806,
      "grad_norm": 4.6732889936639985,
      "learning_rate": 1.0572139303482589e-07,
      "loss": 1.6562,
      "step": 34
    },
    {
      "epoch": 0.0032291546534425094,
      "grad_norm": 4.754796330549508,
      "learning_rate": 1.088308457711443e-07,
      "loss": 1.7985,
      "step": 35
    },
    {
      "epoch": 0.0032291546534425094,
      "eval_GEN Loss": 1.2255665063858032,
      "eval_GEN top-5 accuracy": 0.9145111793073214,
      "eval_PRM Accuracy": 0.4811320754716981,
      "eval_PRM F1": 0.5736434108527132,
      "eval_PRM F1 AUC": 0.5272393923520168,
      "eval_PRM F1 Neg": 0.3373493975903614,
      "eval_PRM Loss": 0.692743718624115,
      "eval_PRM NPV": 0.23333333333333334,
      "eval_PRM Precision": 0.8043478260869565,
      "eval_PRM Recall": 0.4457831325301205,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 1.8596153259277344,
      "eval_runtime": 56.8707,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 35
    },
    {
      "epoch": 0.0033214162149694383,
      "grad_norm": 4.724528772133822,
      "learning_rate": 1.1194029850746268e-07,
      "loss": 2.1671,
      "step": 36
    },
    {
      "epoch": 0.003413677776496367,
      "grad_norm": 3.9150096808620876,
      "learning_rate": 1.150497512437811e-07,
      "loss": 1.6174,
      "step": 37
    },
    {
      "epoch": 0.003505939338023296,
      "grad_norm": 5.773492084076176,
      "learning_rate": 1.1815920398009951e-07,
      "loss": 1.8857,
      "step": 38
    },
    {
      "epoch": 0.003598200899550225,
      "grad_norm": 5.451277117015242,
      "learning_rate": 1.2126865671641792e-07,
      "loss": 2.0441,
      "step": 39
    },
    {
      "epoch": 0.0036904624610771537,
      "grad_norm": 2.696966431348262,
      "learning_rate": 1.2437810945273633e-07,
      "loss": 1.8209,
      "step": 40
    },
    {
      "epoch": 0.0036904624610771537,
      "eval_GEN Loss": 1.2069568634033203,
      "eval_GEN top-5 accuracy": 0.9149495835160018,
      "eval_PRM Accuracy": 0.660377358490566,
      "eval_PRM F1": 0.7831325301204819,
      "eval_PRM F1 AUC": 0.5002619172341539,
      "eval_PRM F1 Neg": 0.21739130434782608,
      "eval_PRM Loss": 0.6919686794281006,
      "eval_PRM NPV": 0.21739130434782608,
      "eval_PRM Precision": 0.7831325301204819,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.8411058187484741,
      "eval_runtime": 56.0246,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 40
    },
    {
      "epoch": 0.0037827240226040826,
      "grad_norm": 3.493866530933666,
      "learning_rate": 1.2748756218905472e-07,
      "loss": 1.6863,
      "step": 41
    },
    {
      "epoch": 0.0038749855841310114,
      "grad_norm": 2.6161801908482523,
      "learning_rate": 1.3059701492537313e-07,
      "loss": 1.7593,
      "step": 42
    },
    {
      "epoch": 0.00396724714565794,
      "grad_norm": 5.694591249676084,
      "learning_rate": 1.3370646766169154e-07,
      "loss": 1.6312,
      "step": 43
    },
    {
      "epoch": 0.004059508707184869,
      "grad_norm": 3.1568086255048233,
      "learning_rate": 1.3681592039800996e-07,
      "loss": 1.597,
      "step": 44
    },
    {
      "epoch": 0.0041517702687117975,
      "grad_norm": 3.587678655299929,
      "learning_rate": 1.3992537313432837e-07,
      "loss": 1.7693,
      "step": 45
    },
    {
      "epoch": 0.0041517702687117975,
      "eval_GEN Loss": 1.1930245161056519,
      "eval_GEN top-5 accuracy": 0.9149495835160018,
      "eval_PRM Accuracy": 0.6886792452830188,
      "eval_PRM F1": 0.8092485549132948,
      "eval_PRM F1 AUC": 0.48690413829229967,
      "eval_PRM F1 Neg": 0.15384615384615385,
      "eval_PRM Loss": 0.691718339920044,
      "eval_PRM NPV": 0.1875,
      "eval_PRM Precision": 0.7777777777777778,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.13043478260869565,
      "eval_loss": 1.8271634578704834,
      "eval_runtime": 55.8771,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 45
    },
    {
      "epoch": 0.004244031830238726,
      "grad_norm": 2.5237286778823944,
      "learning_rate": 1.4303482587064678e-07,
      "loss": 1.7863,
      "step": 46
    },
    {
      "epoch": 0.004336293391765655,
      "grad_norm": 5.26890923882605,
      "learning_rate": 1.461442786069652e-07,
      "loss": 1.8039,
      "step": 47
    },
    {
      "epoch": 0.004428554953292584,
      "grad_norm": 4.171052115799442,
      "learning_rate": 1.4925373134328358e-07,
      "loss": 1.7905,
      "step": 48
    },
    {
      "epoch": 0.004520816514819513,
      "grad_norm": 6.515718983622325,
      "learning_rate": 1.52363184079602e-07,
      "loss": 1.9001,
      "step": 49
    },
    {
      "epoch": 0.004613078076346442,
      "grad_norm": 4.372461415549234,
      "learning_rate": 1.554726368159204e-07,
      "loss": 1.8444,
      "step": 50
    },
    {
      "epoch": 0.004613078076346442,
      "eval_GEN Loss": 1.1805022954940796,
      "eval_GEN top-5 accuracy": 0.9153879877246821,
      "eval_PRM Accuracy": 0.6509433962264151,
      "eval_PRM F1": 0.783625730994152,
      "eval_PRM F1 AUC": 0.44709271870089057,
      "eval_PRM F1 Neg": 0.0975609756097561,
      "eval_PRM Loss": 0.69174724817276,
      "eval_PRM NPV": 0.1111111111111111,
      "eval_PRM Precision": 0.7613636363636364,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.08695652173913043,
      "eval_loss": 1.817307710647583,
      "eval_runtime": 55.9481,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 50
    },
    {
      "epoch": 0.004705339637873371,
      "grad_norm": 4.193248151423293,
      "learning_rate": 1.5858208955223882e-07,
      "loss": 2.1622,
      "step": 51
    },
    {
      "epoch": 0.0047976011994002995,
      "grad_norm": 4.623631435876188,
      "learning_rate": 1.6169154228855723e-07,
      "loss": 2.0667,
      "step": 52
    },
    {
      "epoch": 0.004889862760927228,
      "grad_norm": 3.345737382972426,
      "learning_rate": 1.6480099502487564e-07,
      "loss": 1.7507,
      "step": 53
    },
    {
      "epoch": 0.004982124322454157,
      "grad_norm": 3.9041233882558597,
      "learning_rate": 1.6791044776119405e-07,
      "loss": 1.818,
      "step": 54
    },
    {
      "epoch": 0.005074385883981086,
      "grad_norm": 5.298607880934922,
      "learning_rate": 1.7101990049751244e-07,
      "loss": 1.5183,
      "step": 55
    },
    {
      "epoch": 0.005074385883981086,
      "eval_GEN Loss": 1.1278165578842163,
      "eval_GEN top-5 accuracy": 0.9158263919333626,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6889498233795166,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.7629808187484741,
      "eval_runtime": 55.9126,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 55
    },
    {
      "epoch": 0.005166647445508015,
      "grad_norm": 3.336218883854023,
      "learning_rate": 1.7412935323383088e-07,
      "loss": 1.6167,
      "step": 56
    },
    {
      "epoch": 0.005258909007034944,
      "grad_norm": 3.8759513202112656,
      "learning_rate": 1.7723880597014926e-07,
      "loss": 1.7826,
      "step": 57
    },
    {
      "epoch": 0.005351170568561873,
      "grad_norm": 3.127856786853475,
      "learning_rate": 1.8034825870646767e-07,
      "loss": 1.9113,
      "step": 58
    },
    {
      "epoch": 0.0054434321300888015,
      "grad_norm": 4.326349376650393,
      "learning_rate": 1.834577114427861e-07,
      "loss": 1.7121,
      "step": 59
    },
    {
      "epoch": 0.00553569369161573,
      "grad_norm": 4.624249003919422,
      "learning_rate": 1.8656716417910447e-07,
      "loss": 1.5453,
      "step": 60
    },
    {
      "epoch": 0.00553569369161573,
      "eval_GEN Loss": 1.0909538269042969,
      "eval_GEN top-5 accuracy": 0.9158263919333626,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6880059242248535,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.7301682233810425,
      "eval_runtime": 56.7905,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 60
    },
    {
      "epoch": 0.005627955253142659,
      "grad_norm": 3.486124204854497,
      "learning_rate": 1.896766169154229e-07,
      "loss": 1.7168,
      "step": 61
    },
    {
      "epoch": 0.005720216814669588,
      "grad_norm": 4.489173258726614,
      "learning_rate": 1.927860696517413e-07,
      "loss": 1.6848,
      "step": 62
    },
    {
      "epoch": 0.005812478376196517,
      "grad_norm": 3.9397917707556602,
      "learning_rate": 1.9589552238805974e-07,
      "loss": 1.4471,
      "step": 63
    },
    {
      "epoch": 0.005904739937723446,
      "grad_norm": 2.6644880049876924,
      "learning_rate": 1.9900497512437812e-07,
      "loss": 1.7052,
      "step": 64
    },
    {
      "epoch": 0.005997001499250375,
      "grad_norm": 4.2575966052447285,
      "learning_rate": 2.0211442786069656e-07,
      "loss": 1.7001,
      "step": 65
    },
    {
      "epoch": 0.005997001499250375,
      "eval_GEN Loss": 1.064455270767212,
      "eval_GEN top-5 accuracy": 0.918456817185445,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6875361204147339,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.7044471502304077,
      "eval_runtime": 56.9189,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 65
    },
    {
      "epoch": 0.0060892630607773035,
      "grad_norm": 3.2355348663472263,
      "learning_rate": 2.0522388059701495e-07,
      "loss": 1.6682,
      "step": 66
    },
    {
      "epoch": 0.006181524622304232,
      "grad_norm": 3.071024405471419,
      "learning_rate": 2.0833333333333333e-07,
      "loss": 1.6072,
      "step": 67
    },
    {
      "epoch": 0.006273786183831161,
      "grad_norm": 3.554680940774165,
      "learning_rate": 2.1144278606965177e-07,
      "loss": 1.7963,
      "step": 68
    },
    {
      "epoch": 0.00636604774535809,
      "grad_norm": 2.5298879543430677,
      "learning_rate": 2.1455223880597016e-07,
      "loss": 1.6359,
      "step": 69
    },
    {
      "epoch": 0.006458309306885019,
      "grad_norm": 6.083087924089791,
      "learning_rate": 2.176616915422886e-07,
      "loss": 1.8998,
      "step": 70
    },
    {
      "epoch": 0.006458309306885019,
      "eval_GEN Loss": 1.0378422737121582,
      "eval_GEN top-5 accuracy": 0.9197720298114862,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6867654323577881,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.6774038076400757,
      "eval_runtime": 55.8194,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 70
    },
    {
      "epoch": 0.006550570868411948,
      "grad_norm": 2.9042617792127965,
      "learning_rate": 2.2077114427860698e-07,
      "loss": 1.8414,
      "step": 71
    },
    {
      "epoch": 0.006642832429938877,
      "grad_norm": 4.244590383180499,
      "learning_rate": 2.2388059701492537e-07,
      "loss": 1.6526,
      "step": 72
    },
    {
      "epoch": 0.0067350939914658054,
      "grad_norm": 3.3203920579098694,
      "learning_rate": 2.269900497512438e-07,
      "loss": 1.5533,
      "step": 73
    },
    {
      "epoch": 0.006827355552992734,
      "grad_norm": 1.744830315172512,
      "learning_rate": 2.300995024875622e-07,
      "loss": 1.3191,
      "step": 74
    },
    {
      "epoch": 0.006919617114519663,
      "grad_norm": 4.326451307986983,
      "learning_rate": 2.3320895522388063e-07,
      "loss": 1.5463,
      "step": 75
    },
    {
      "epoch": 0.006919617114519663,
      "eval_GEN Loss": 0.9482924938201904,
      "eval_GEN top-5 accuracy": 0.9224024550635687,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6790199875831604,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.5992788076400757,
      "eval_runtime": 56.734,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 75
    },
    {
      "epoch": 0.007011878676046592,
      "grad_norm": 2.1644395245838948,
      "learning_rate": 2.3631840796019902e-07,
      "loss": 1.5132,
      "step": 76
    },
    {
      "epoch": 0.007104140237573521,
      "grad_norm": 5.274809047495626,
      "learning_rate": 2.394278606965174e-07,
      "loss": 1.434,
      "step": 77
    },
    {
      "epoch": 0.00719640179910045,
      "grad_norm": 2.813602299901219,
      "learning_rate": 2.4253731343283584e-07,
      "loss": 1.4597,
      "step": 78
    },
    {
      "epoch": 0.007288663360627379,
      "grad_norm": 2.023967194882434,
      "learning_rate": 2.4564676616915423e-07,
      "loss": 1.4597,
      "step": 79
    },
    {
      "epoch": 0.007380924922154307,
      "grad_norm": 2.3835753067138805,
      "learning_rate": 2.4875621890547267e-07,
      "loss": 1.4664,
      "step": 80
    },
    {
      "epoch": 0.007380924922154307,
      "eval_GEN Loss": 0.8692818284034729,
      "eval_GEN top-5 accuracy": 0.9285401139850943,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6763344407081604,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.526802897453308,
      "eval_runtime": 56.8248,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 80
    },
    {
      "epoch": 0.007473186483681236,
      "grad_norm": 3.2167315704974158,
      "learning_rate": 2.5186567164179105e-07,
      "loss": 1.4135,
      "step": 81
    },
    {
      "epoch": 0.007565448045208165,
      "grad_norm": 2.8205981850032686,
      "learning_rate": 2.5497512437810944e-07,
      "loss": 1.4354,
      "step": 82
    },
    {
      "epoch": 0.007657709606735094,
      "grad_norm": 2.518582934298941,
      "learning_rate": 2.580845771144279e-07,
      "loss": 1.4236,
      "step": 83
    },
    {
      "epoch": 0.007749971168262023,
      "grad_norm": 4.025192935451012,
      "learning_rate": 2.6119402985074626e-07,
      "loss": 1.5447,
      "step": 84
    },
    {
      "epoch": 0.00784223272978895,
      "grad_norm": 3.937779371079708,
      "learning_rate": 2.643034825870647e-07,
      "loss": 1.5491,
      "step": 85
    },
    {
      "epoch": 0.00784223272978895,
      "eval_GEN Loss": 0.8156515955924988,
      "eval_GEN top-5 accuracy": 0.9368697939500219,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6749014854431152,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.4822115898132324,
      "eval_runtime": 56.7923,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 85
    },
    {
      "epoch": 0.00793449429131588,
      "grad_norm": 4.373961075167894,
      "learning_rate": 2.674129353233831e-07,
      "loss": 1.5427,
      "step": 86
    },
    {
      "epoch": 0.008026755852842809,
      "grad_norm": 4.003591369571609,
      "learning_rate": 2.7052238805970147e-07,
      "loss": 1.5683,
      "step": 87
    },
    {
      "epoch": 0.008119017414369737,
      "grad_norm": 1.9910457980372962,
      "learning_rate": 2.736318407960199e-07,
      "loss": 1.508,
      "step": 88
    },
    {
      "epoch": 0.008211278975896666,
      "grad_norm": 3.5811749558032298,
      "learning_rate": 2.767412935323383e-07,
      "loss": 1.434,
      "step": 89
    },
    {
      "epoch": 0.008303540537423595,
      "grad_norm": 2.949039054324564,
      "learning_rate": 2.7985074626865674e-07,
      "loss": 1.2979,
      "step": 90
    },
    {
      "epoch": 0.008303540537423595,
      "eval_GEN Loss": 0.7697263956069946,
      "eval_GEN top-5 accuracy": 0.9421306444541867,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6727759838104248,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.4397836923599243,
      "eval_runtime": 56.0328,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 90
    },
    {
      "epoch": 0.008395802098950524,
      "grad_norm": 4.300210380913143,
      "learning_rate": 2.829601990049751e-07,
      "loss": 1.3892,
      "step": 91
    },
    {
      "epoch": 0.008488063660477453,
      "grad_norm": 4.497865551632025,
      "learning_rate": 2.8606965174129356e-07,
      "loss": 1.3393,
      "step": 92
    },
    {
      "epoch": 0.008580325222004382,
      "grad_norm": 3.681475742018798,
      "learning_rate": 2.8917910447761195e-07,
      "loss": 1.3857,
      "step": 93
    },
    {
      "epoch": 0.00867258678353131,
      "grad_norm": 4.2811633470511605,
      "learning_rate": 2.922885572139304e-07,
      "loss": 1.5033,
      "step": 94
    },
    {
      "epoch": 0.00876484834505824,
      "grad_norm": 2.6063769382613136,
      "learning_rate": 2.9539800995024877e-07,
      "loss": 1.3467,
      "step": 95
    },
    {
      "epoch": 0.00876484834505824,
      "eval_GEN Loss": 0.7327077984809875,
      "eval_GEN top-5 accuracy": 0.9465146865409908,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6708540916442871,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.4075721502304077,
      "eval_runtime": 55.936,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 95
    },
    {
      "epoch": 0.008857109906585168,
      "grad_norm": 2.280241426078729,
      "learning_rate": 2.9850746268656716e-07,
      "loss": 1.4608,
      "step": 96
    },
    {
      "epoch": 0.008949371468112097,
      "grad_norm": 3.237155480971528,
      "learning_rate": 3.016169154228856e-07,
      "loss": 1.3498,
      "step": 97
    },
    {
      "epoch": 0.009041633029639026,
      "grad_norm": 4.845276940236311,
      "learning_rate": 3.04726368159204e-07,
      "loss": 1.4412,
      "step": 98
    },
    {
      "epoch": 0.009133894591165955,
      "grad_norm": 2.6876818129375284,
      "learning_rate": 3.078358208955224e-07,
      "loss": 1.4279,
      "step": 99
    },
    {
      "epoch": 0.009226156152692884,
      "grad_norm": 4.503626692399455,
      "learning_rate": 3.109452736318408e-07,
      "loss": 1.4015,
      "step": 100
    },
    {
      "epoch": 0.009226156152692884,
      "eval_GEN Loss": 0.6929720044136047,
      "eval_GEN top-5 accuracy": 0.9504603244191144,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6667302250862122,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.3634614944458008,
      "eval_runtime": 56.041,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 100
    },
    {
      "epoch": 0.009318417714219812,
      "grad_norm": 3.162280791834292,
      "learning_rate": 3.1405472636815924e-07,
      "loss": 1.386,
      "step": 101
    },
    {
      "epoch": 0.009410679275746741,
      "grad_norm": 3.5415551551296294,
      "learning_rate": 3.1716417910447763e-07,
      "loss": 1.3551,
      "step": 102
    },
    {
      "epoch": 0.00950294083727367,
      "grad_norm": 2.59884729682728,
      "learning_rate": 3.2027363184079607e-07,
      "loss": 1.3806,
      "step": 103
    },
    {
      "epoch": 0.009595202398800599,
      "grad_norm": 2.654981055039507,
      "learning_rate": 3.2338308457711446e-07,
      "loss": 1.321,
      "step": 104
    },
    {
      "epoch": 0.009687463960327528,
      "grad_norm": 2.2640850649522335,
      "learning_rate": 3.2649253731343284e-07,
      "loss": 1.3313,
      "step": 105
    },
    {
      "epoch": 0.009687463960327528,
      "eval_GEN Loss": 0.6564959287643433,
      "eval_GEN top-5 accuracy": 0.958790004384042,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6540261507034302,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.3162260055541992,
      "eval_runtime": 56.1397,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 105
    },
    {
      "epoch": 0.009779725521854457,
      "grad_norm": 2.6718127266221456,
      "learning_rate": 3.296019900497513e-07,
      "loss": 1.2695,
      "step": 106
    },
    {
      "epoch": 0.009871987083381386,
      "grad_norm": 2.3077499382198243,
      "learning_rate": 3.3271144278606967e-07,
      "loss": 1.3066,
      "step": 107
    },
    {
      "epoch": 0.009964248644908314,
      "grad_norm": 3.7464571642834916,
      "learning_rate": 3.358208955223881e-07,
      "loss": 1.2675,
      "step": 108
    },
    {
      "epoch": 0.010056510206435243,
      "grad_norm": 2.244849109531541,
      "learning_rate": 3.389303482587065e-07,
      "loss": 1.3297,
      "step": 109
    },
    {
      "epoch": 0.010148771767962172,
      "grad_norm": 3.3919027523850445,
      "learning_rate": 3.420398009950249e-07,
      "loss": 1.2619,
      "step": 110
    },
    {
      "epoch": 0.010148771767962172,
      "eval_GEN Loss": 0.632244348526001,
      "eval_GEN top-5 accuracy": 0.9644892590968873,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.64864581823349,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2882211208343506,
      "eval_runtime": 55.9108,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 110
    },
    {
      "epoch": 0.010241033329489101,
      "grad_norm": 1.487790956636087,
      "learning_rate": 3.451492537313433e-07,
      "loss": 1.1671,
      "step": 111
    },
    {
      "epoch": 0.01033329489101603,
      "grad_norm": 1.949272382635497,
      "learning_rate": 3.4825870646766175e-07,
      "loss": 1.4099,
      "step": 112
    },
    {
      "epoch": 0.010425556452542959,
      "grad_norm": 1.8034335034787636,
      "learning_rate": 3.5136815920398014e-07,
      "loss": 1.1681,
      "step": 113
    },
    {
      "epoch": 0.010517818014069888,
      "grad_norm": 2.3584905317190623,
      "learning_rate": 3.544776119402985e-07,
      "loss": 1.3049,
      "step": 114
    },
    {
      "epoch": 0.010610079575596816,
      "grad_norm": 4.7832735810400075,
      "learning_rate": 3.575870646766169e-07,
      "loss": 1.19,
      "step": 115
    },
    {
      "epoch": 0.010610079575596816,
      "eval_GEN Loss": 0.6159378290176392,
      "eval_GEN top-5 accuracy": 0.9675580885576501,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.646122932434082,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2694711685180664,
      "eval_runtime": 56.8869,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 115
    },
    {
      "epoch": 0.010702341137123745,
      "grad_norm": 3.159208513024856,
      "learning_rate": 3.6069651741293535e-07,
      "loss": 1.2334,
      "step": 116
    },
    {
      "epoch": 0.010794602698650674,
      "grad_norm": 2.7159029094213345,
      "learning_rate": 3.638059701492538e-07,
      "loss": 1.2048,
      "step": 117
    },
    {
      "epoch": 0.010886864260177603,
      "grad_norm": 5.919509251949845,
      "learning_rate": 3.669154228855722e-07,
      "loss": 1.1744,
      "step": 118
    },
    {
      "epoch": 0.010979125821704532,
      "grad_norm": 3.387517453611791,
      "learning_rate": 3.7002487562189056e-07,
      "loss": 1.239,
      "step": 119
    },
    {
      "epoch": 0.01107138738323146,
      "grad_norm": 3.0146841098526593,
      "learning_rate": 3.7313432835820895e-07,
      "loss": 1.29,
      "step": 120
    },
    {
      "epoch": 0.01107138738323146,
      "eval_GEN Loss": 0.6037746667861938,
      "eval_GEN top-5 accuracy": 0.9693117053923718,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6428813338279724,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2550480365753174,
      "eval_runtime": 56.9415,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 120
    },
    {
      "epoch": 0.01116364894475839,
      "grad_norm": 2.3794434955477186,
      "learning_rate": 3.7624378109452744e-07,
      "loss": 1.1949,
      "step": 121
    },
    {
      "epoch": 0.011255910506285318,
      "grad_norm": 3.322725738719409,
      "learning_rate": 3.793532338308458e-07,
      "loss": 1.2007,
      "step": 122
    },
    {
      "epoch": 0.011348172067812247,
      "grad_norm": 3.2082068601478473,
      "learning_rate": 3.824626865671642e-07,
      "loss": 1.2183,
      "step": 123
    },
    {
      "epoch": 0.011440433629339176,
      "grad_norm": 2.1721049460512023,
      "learning_rate": 3.855721393034826e-07,
      "loss": 1.2992,
      "step": 124
    },
    {
      "epoch": 0.011532695190866105,
      "grad_norm": 4.970164595238515,
      "learning_rate": 3.88681592039801e-07,
      "loss": 1.3881,
      "step": 125
    },
    {
      "epoch": 0.011532695190866105,
      "eval_GEN Loss": 0.5954346656799316,
      "eval_GEN top-5 accuracy": 0.9693117053923718,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.637204647064209,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2435096502304077,
      "eval_runtime": 56.9017,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 125
    },
    {
      "epoch": 0.011624956752393034,
      "grad_norm": 2.140319140328948,
      "learning_rate": 3.9179104477611947e-07,
      "loss": 1.3041,
      "step": 126
    },
    {
      "epoch": 0.011717218313919963,
      "grad_norm": 1.6493112799847287,
      "learning_rate": 3.9490049751243786e-07,
      "loss": 1.2208,
      "step": 127
    },
    {
      "epoch": 0.011809479875446892,
      "grad_norm": 2.361177425828381,
      "learning_rate": 3.9800995024875624e-07,
      "loss": 1.2167,
      "step": 128
    },
    {
      "epoch": 0.01190174143697382,
      "grad_norm": 1.575569515428191,
      "learning_rate": 4.0111940298507463e-07,
      "loss": 1.1579,
      "step": 129
    },
    {
      "epoch": 0.01199400299850075,
      "grad_norm": 1.954657491756902,
      "learning_rate": 4.042288557213931e-07,
      "loss": 1.2077,
      "step": 130
    },
    {
      "epoch": 0.01199400299850075,
      "eval_GEN Loss": 0.5868880748748779,
      "eval_GEN top-5 accuracy": 0.9697501096010521,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6340000033378601,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2335336208343506,
      "eval_runtime": 56.7979,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 130
    },
    {
      "epoch": 0.012086264560027678,
      "grad_norm": 3.3479998825199906,
      "learning_rate": 4.073383084577115e-07,
      "loss": 1.0645,
      "step": 131
    },
    {
      "epoch": 0.012178526121554607,
      "grad_norm": 3.8275086734185106,
      "learning_rate": 4.104477611940299e-07,
      "loss": 1.3601,
      "step": 132
    },
    {
      "epoch": 0.012270787683081536,
      "grad_norm": 1.6417110032987758,
      "learning_rate": 4.135572139303483e-07,
      "loss": 1.226,
      "step": 133
    },
    {
      "epoch": 0.012363049244608465,
      "grad_norm": 4.253174971807062,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 1.187,
      "step": 134
    },
    {
      "epoch": 0.012455310806135394,
      "grad_norm": 2.4885860148602372,
      "learning_rate": 4.1977611940298516e-07,
      "loss": 1.2166,
      "step": 135
    },
    {
      "epoch": 0.012455310806135394,
      "eval_GEN Loss": 0.5787835121154785,
      "eval_GEN top-5 accuracy": 0.9706269180184129,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.630656361579895,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2252403497695923,
      "eval_runtime": 56.7719,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 135
    },
    {
      "epoch": 0.012547572367662322,
      "grad_norm": 3.8927966197390105,
      "learning_rate": 4.2288557213930354e-07,
      "loss": 1.3302,
      "step": 136
    },
    {
      "epoch": 0.012639833929189251,
      "grad_norm": 2.322490627211395,
      "learning_rate": 4.2599502487562193e-07,
      "loss": 1.2212,
      "step": 137
    },
    {
      "epoch": 0.01273209549071618,
      "grad_norm": 4.256376338969038,
      "learning_rate": 4.291044776119403e-07,
      "loss": 1.1828,
      "step": 138
    },
    {
      "epoch": 0.012824357052243109,
      "grad_norm": 2.197689682736893,
      "learning_rate": 4.322139303482587e-07,
      "loss": 1.148,
      "step": 139
    },
    {
      "epoch": 0.012916618613770038,
      "grad_norm": 2.524441418608464,
      "learning_rate": 4.353233830845772e-07,
      "loss": 1.2342,
      "step": 140
    },
    {
      "epoch": 0.012916618613770038,
      "eval_GEN Loss": 0.572195291519165,
      "eval_GEN top-5 accuracy": 0.9710653222270934,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6265490055084229,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2159855365753174,
      "eval_runtime": 56.7698,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 140
    },
    {
      "epoch": 0.013008880175296967,
      "grad_norm": 1.8499524568910997,
      "learning_rate": 4.384328358208956e-07,
      "loss": 1.1273,
      "step": 141
    },
    {
      "epoch": 0.013101141736823895,
      "grad_norm": 4.305504027393469,
      "learning_rate": 4.4154228855721396e-07,
      "loss": 1.2465,
      "step": 142
    },
    {
      "epoch": 0.013193403298350824,
      "grad_norm": 3.3734445402391686,
      "learning_rate": 4.4465174129353235e-07,
      "loss": 1.2308,
      "step": 143
    },
    {
      "epoch": 0.013285664859877753,
      "grad_norm": 1.3413274073357973,
      "learning_rate": 4.4776119402985074e-07,
      "loss": 1.1872,
      "step": 144
    },
    {
      "epoch": 0.013377926421404682,
      "grad_norm": 1.341728169252834,
      "learning_rate": 4.5087064676616923e-07,
      "loss": 1.115,
      "step": 145
    },
    {
      "epoch": 0.013377926421404682,
      "eval_GEN Loss": 0.5650824308395386,
      "eval_GEN top-5 accuracy": 0.9710653222270934,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6218520402908325,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2105768918991089,
      "eval_runtime": 56.9056,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 145
    },
    {
      "epoch": 0.013470187982931611,
      "grad_norm": 3.6616799188033853,
      "learning_rate": 4.539800995024876e-07,
      "loss": 1.1553,
      "step": 146
    },
    {
      "epoch": 0.01356244954445854,
      "grad_norm": 2.5202344461470774,
      "learning_rate": 4.57089552238806e-07,
      "loss": 1.1569,
      "step": 147
    },
    {
      "epoch": 0.013654711105985469,
      "grad_norm": 4.192930250780877,
      "learning_rate": 4.601990049751244e-07,
      "loss": 1.1474,
      "step": 148
    },
    {
      "epoch": 0.013746972667512397,
      "grad_norm": 3.937700910615772,
      "learning_rate": 4.6330845771144277e-07,
      "loss": 1.17,
      "step": 149
    },
    {
      "epoch": 0.013839234229039326,
      "grad_norm": 1.5356498681091049,
      "learning_rate": 4.6641791044776126e-07,
      "loss": 1.0327,
      "step": 150
    },
    {
      "epoch": 0.013839234229039326,
      "eval_GEN Loss": 0.5591273903846741,
      "eval_GEN top-5 accuracy": 0.9723805348531346,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6177559494972229,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.2018028497695923,
      "eval_runtime": 55.7981,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 150
    },
    {
      "epoch": 0.013931495790566255,
      "grad_norm": 3.426476007703366,
      "learning_rate": 4.6952736318407965e-07,
      "loss": 1.0698,
      "step": 151
    },
    {
      "epoch": 0.014023757352093184,
      "grad_norm": 3.84524761100555,
      "learning_rate": 4.7263681592039803e-07,
      "loss": 1.1573,
      "step": 152
    },
    {
      "epoch": 0.014116018913620113,
      "grad_norm": 2.700267632870109,
      "learning_rate": 4.757462686567164e-07,
      "loss": 1.1669,
      "step": 153
    },
    {
      "epoch": 0.014208280475147042,
      "grad_norm": 2.5411199119199406,
      "learning_rate": 4.788557213930348e-07,
      "loss": 1.127,
      "step": 154
    },
    {
      "epoch": 0.01430054203667397,
      "grad_norm": 2.630794237531673,
      "learning_rate": 4.819651741293534e-07,
      "loss": 1.1384,
      "step": 155
    },
    {
      "epoch": 0.01430054203667397,
      "eval_GEN Loss": 0.5535362362861633,
      "eval_GEN top-5 accuracy": 0.9723805348531346,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6125956177711487,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1959134340286255,
      "eval_runtime": 56.0126,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 155
    },
    {
      "epoch": 0.0143928035982009,
      "grad_norm": 3.0187035031997502,
      "learning_rate": 4.850746268656717e-07,
      "loss": 1.3768,
      "step": 156
    },
    {
      "epoch": 0.014485065159727828,
      "grad_norm": 2.911667717269186,
      "learning_rate": 4.881840796019901e-07,
      "loss": 1.2243,
      "step": 157
    },
    {
      "epoch": 0.014577326721254757,
      "grad_norm": 3.201497950034923,
      "learning_rate": 4.912935323383085e-07,
      "loss": 1.2479,
      "step": 158
    },
    {
      "epoch": 0.014669588282781686,
      "grad_norm": 4.372336363957918,
      "learning_rate": 4.944029850746269e-07,
      "loss": 1.0542,
      "step": 159
    },
    {
      "epoch": 0.014761849844308615,
      "grad_norm": 1.7042290709906678,
      "learning_rate": 4.975124378109453e-07,
      "loss": 1.1225,
      "step": 160
    },
    {
      "epoch": 0.014761849844308615,
      "eval_GEN Loss": 0.5486795902252197,
      "eval_GEN top-5 accuracy": 0.9723805348531346,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6068178415298462,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1885817050933838,
      "eval_runtime": 55.8918,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 160
    },
    {
      "epoch": 0.014854111405835544,
      "grad_norm": 2.2313977327125945,
      "learning_rate": 5.006218905472638e-07,
      "loss": 1.2895,
      "step": 161
    },
    {
      "epoch": 0.014946372967362473,
      "grad_norm": 1.6765634851215816,
      "learning_rate": 5.037313432835821e-07,
      "loss": 1.1747,
      "step": 162
    },
    {
      "epoch": 0.015038634528889401,
      "grad_norm": 2.3646330611673934,
      "learning_rate": 5.068407960199005e-07,
      "loss": 1.0631,
      "step": 163
    },
    {
      "epoch": 0.01513089609041633,
      "grad_norm": 2.6665593945894877,
      "learning_rate": 5.099502487562189e-07,
      "loss": 1.0604,
      "step": 164
    },
    {
      "epoch": 0.015223157651943259,
      "grad_norm": 3.6891562290617297,
      "learning_rate": 5.130597014925374e-07,
      "loss": 1.2219,
      "step": 165
    },
    {
      "epoch": 0.015223157651943259,
      "eval_GEN Loss": 0.5440877676010132,
      "eval_GEN top-5 accuracy": 0.972818939061815,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.6010922193527222,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1805288791656494,
      "eval_runtime": 55.9587,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 165
    },
    {
      "epoch": 0.015315419213470188,
      "grad_norm": 3.499835361395495,
      "learning_rate": 5.161691542288558e-07,
      "loss": 1.1838,
      "step": 166
    },
    {
      "epoch": 0.015407680774997117,
      "grad_norm": 1.9058665102788184,
      "learning_rate": 5.192786069651742e-07,
      "loss": 1.1222,
      "step": 167
    },
    {
      "epoch": 0.015499942336524046,
      "grad_norm": 1.7497038975039427,
      "learning_rate": 5.223880597014925e-07,
      "loss": 1.1473,
      "step": 168
    },
    {
      "epoch": 0.015592203898050975,
      "grad_norm": 1.376208125069818,
      "learning_rate": 5.25497512437811e-07,
      "loss": 1.2389,
      "step": 169
    },
    {
      "epoch": 0.0156844654595779,
      "grad_norm": 1.506267183082681,
      "learning_rate": 5.286069651741294e-07,
      "loss": 1.1814,
      "step": 170
    },
    {
      "epoch": 0.0156844654595779,
      "eval_GEN Loss": 0.5395631790161133,
      "eval_GEN top-5 accuracy": 0.9719421306444542,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5953467488288879,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1742788553237915,
      "eval_runtime": 55.8564,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 170
    },
    {
      "epoch": 0.01577672702110483,
      "grad_norm": 1.4669207809936677,
      "learning_rate": 5.317164179104478e-07,
      "loss": 1.104,
      "step": 171
    },
    {
      "epoch": 0.01586898858263176,
      "grad_norm": 3.187983413829535,
      "learning_rate": 5.348258706467662e-07,
      "loss": 1.1191,
      "step": 172
    },
    {
      "epoch": 0.015961250144158688,
      "grad_norm": 2.263443121489794,
      "learning_rate": 5.379353233830846e-07,
      "loss": 1.3148,
      "step": 173
    },
    {
      "epoch": 0.016053511705685617,
      "grad_norm": 1.6708278469196651,
      "learning_rate": 5.410447761194029e-07,
      "loss": 1.0963,
      "step": 174
    },
    {
      "epoch": 0.016145773267212546,
      "grad_norm": 1.2433902795383085,
      "learning_rate": 5.441542288557215e-07,
      "loss": 1.1433,
      "step": 175
    },
    {
      "epoch": 0.016145773267212546,
      "eval_GEN Loss": 0.5355656147003174,
      "eval_GEN top-5 accuracy": 0.972818939061815,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5899839401245117,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1680288314819336,
      "eval_runtime": 56.0484,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 175
    },
    {
      "epoch": 0.016238034828739475,
      "grad_norm": 3.123639407344628,
      "learning_rate": 5.472636815920398e-07,
      "loss": 1.0163,
      "step": 176
    },
    {
      "epoch": 0.016330296390266404,
      "grad_norm": 2.0634700843488134,
      "learning_rate": 5.503731343283583e-07,
      "loss": 1.0857,
      "step": 177
    },
    {
      "epoch": 0.016422557951793332,
      "grad_norm": 1.3954752682063847,
      "learning_rate": 5.534825870646766e-07,
      "loss": 1.157,
      "step": 178
    },
    {
      "epoch": 0.01651481951332026,
      "grad_norm": 2.3945957166778307,
      "learning_rate": 5.565920398009951e-07,
      "loss": 1.1593,
      "step": 179
    },
    {
      "epoch": 0.01660708107484719,
      "grad_norm": 2.0141835860911446,
      "learning_rate": 5.597014925373135e-07,
      "loss": 1.1564,
      "step": 180
    },
    {
      "epoch": 0.01660708107484719,
      "eval_GEN Loss": 0.5322163701057434,
      "eval_GEN top-5 accuracy": 0.9723805348531346,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5848449468612671,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1637018918991089,
      "eval_runtime": 56.2358,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 180
    },
    {
      "epoch": 0.01669934263637412,
      "grad_norm": 1.7479021559057548,
      "learning_rate": 5.628109452736319e-07,
      "loss": 1.1356,
      "step": 181
    },
    {
      "epoch": 0.016791604197901048,
      "grad_norm": 2.6506406095324624,
      "learning_rate": 5.659203980099502e-07,
      "loss": 1.0324,
      "step": 182
    },
    {
      "epoch": 0.016883865759427977,
      "grad_norm": 2.142269070565063,
      "learning_rate": 5.690298507462687e-07,
      "loss": 1.1757,
      "step": 183
    },
    {
      "epoch": 0.016976127320954906,
      "grad_norm": 1.392911091467902,
      "learning_rate": 5.721393034825871e-07,
      "loss": 1.1192,
      "step": 184
    },
    {
      "epoch": 0.017068388882481834,
      "grad_norm": 3.234694851702116,
      "learning_rate": 5.752487562189056e-07,
      "loss": 0.9247,
      "step": 185
    },
    {
      "epoch": 0.017068388882481834,
      "eval_GEN Loss": 0.5290143489837646,
      "eval_GEN top-5 accuracy": 0.9736957474791758,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5798476934432983,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1593749523162842,
      "eval_runtime": 55.7916,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 185
    },
    {
      "epoch": 0.017160650444008763,
      "grad_norm": 3.262504360848584,
      "learning_rate": 5.783582089552239e-07,
      "loss": 1.1936,
      "step": 186
    },
    {
      "epoch": 0.017252912005535692,
      "grad_norm": 1.2351033356893326,
      "learning_rate": 5.814676616915423e-07,
      "loss": 1.1101,
      "step": 187
    },
    {
      "epoch": 0.01734517356706262,
      "grad_norm": 2.9221099184657735,
      "learning_rate": 5.845771144278608e-07,
      "loss": 1.1287,
      "step": 188
    },
    {
      "epoch": 0.01743743512858955,
      "grad_norm": 1.2464805163577264,
      "learning_rate": 5.876865671641792e-07,
      "loss": 1.1525,
      "step": 189
    },
    {
      "epoch": 0.01752969669011648,
      "grad_norm": 1.719012479550381,
      "learning_rate": 5.907960199004975e-07,
      "loss": 1.2659,
      "step": 190
    },
    {
      "epoch": 0.01752969669011648,
      "eval_GEN Loss": 0.526759922504425,
      "eval_GEN top-5 accuracy": 0.972818939061815,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5763874650001526,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1551682949066162,
      "eval_runtime": 55.8238,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 190
    },
    {
      "epoch": 0.017621958251643408,
      "grad_norm": 2.036814904860174,
      "learning_rate": 5.93905472636816e-07,
      "loss": 1.1664,
      "step": 191
    },
    {
      "epoch": 0.017714219813170336,
      "grad_norm": 1.9093392405912946,
      "learning_rate": 5.970149253731343e-07,
      "loss": 1.0943,
      "step": 192
    },
    {
      "epoch": 0.017806481374697265,
      "grad_norm": 1.569505782183973,
      "learning_rate": 6.001243781094528e-07,
      "loss": 1.0698,
      "step": 193
    },
    {
      "epoch": 0.017898742936224194,
      "grad_norm": 1.5582990636644043,
      "learning_rate": 6.032338308457712e-07,
      "loss": 1.2057,
      "step": 194
    },
    {
      "epoch": 0.017991004497751123,
      "grad_norm": 1.8711371605271108,
      "learning_rate": 6.063432835820896e-07,
      "loss": 1.2447,
      "step": 195
    },
    {
      "epoch": 0.017991004497751123,
      "eval_GEN Loss": 0.5234596133232117,
      "eval_GEN top-5 accuracy": 0.9736957474791758,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5738641619682312,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1536058187484741,
      "eval_runtime": 56.0063,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 195
    },
    {
      "epoch": 0.018083266059278052,
      "grad_norm": 3.374081738855087,
      "learning_rate": 6.09452736318408e-07,
      "loss": 1.0396,
      "step": 196
    },
    {
      "epoch": 0.01817552762080498,
      "grad_norm": 2.2035285361727652,
      "learning_rate": 6.125621890547264e-07,
      "loss": 1.1203,
      "step": 197
    },
    {
      "epoch": 0.01826778918233191,
      "grad_norm": 2.467382046802569,
      "learning_rate": 6.156716417910448e-07,
      "loss": 1.0652,
      "step": 198
    },
    {
      "epoch": 0.01836005074385884,
      "grad_norm": 1.3431188681413289,
      "learning_rate": 6.187810945273633e-07,
      "loss": 1.1854,
      "step": 199
    },
    {
      "epoch": 0.018452312305385767,
      "grad_norm": 1.850275067257811,
      "learning_rate": 6.218905472636816e-07,
      "loss": 1.0531,
      "step": 200
    },
    {
      "epoch": 0.018452312305385767,
      "eval_GEN Loss": 0.5223154425621033,
      "eval_GEN top-5 accuracy": 0.9736957474791758,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5719572901725769,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1526442766189575,
      "eval_runtime": 55.998,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 200
    },
    {
      "epoch": 0.018544573866912696,
      "grad_norm": 1.8369164323211342,
      "learning_rate": 6.25e-07,
      "loss": 1.0348,
      "step": 201
    },
    {
      "epoch": 0.018636835428439625,
      "grad_norm": 4.042908754628373,
      "learning_rate": 6.281094527363185e-07,
      "loss": 1.2664,
      "step": 202
    },
    {
      "epoch": 0.018729096989966554,
      "grad_norm": 1.8835558742943022,
      "learning_rate": 6.312189054726368e-07,
      "loss": 1.1774,
      "step": 203
    },
    {
      "epoch": 0.018821358551493483,
      "grad_norm": 1.5364103259559745,
      "learning_rate": 6.343283582089553e-07,
      "loss": 1.2211,
      "step": 204
    },
    {
      "epoch": 0.01891362011302041,
      "grad_norm": 1.6463932144779783,
      "learning_rate": 6.374378109452736e-07,
      "loss": 1.2084,
      "step": 205
    },
    {
      "epoch": 0.01891362011302041,
      "eval_GEN Loss": 0.5205348134040833,
      "eval_GEN top-5 accuracy": 0.9754493643138974,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5690513849258423,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1484375,
      "eval_runtime": 56.0259,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 205
    },
    {
      "epoch": 0.01900588167454734,
      "grad_norm": 3.022050229054206,
      "learning_rate": 6.405472636815921e-07,
      "loss": 1.0556,
      "step": 206
    },
    {
      "epoch": 0.01909814323607427,
      "grad_norm": 3.5266726313156043,
      "learning_rate": 6.436567164179106e-07,
      "loss": 1.1038,
      "step": 207
    },
    {
      "epoch": 0.019190404797601198,
      "grad_norm": 1.2466144988035346,
      "learning_rate": 6.467661691542289e-07,
      "loss": 1.0974,
      "step": 208
    },
    {
      "epoch": 0.019282666359128127,
      "grad_norm": 1.962282295115707,
      "learning_rate": 6.498756218905473e-07,
      "loss": 1.0972,
      "step": 209
    },
    {
      "epoch": 0.019374927920655056,
      "grad_norm": 2.0357105644300755,
      "learning_rate": 6.529850746268657e-07,
      "loss": 1.1944,
      "step": 210
    },
    {
      "epoch": 0.019374927920655056,
      "eval_GEN Loss": 0.5192821025848389,
      "eval_GEN top-5 accuracy": 0.9736957474791758,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5659216046333313,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1461538076400757,
      "eval_runtime": 55.9503,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 210
    },
    {
      "epoch": 0.019467189482181985,
      "grad_norm": 2.506291963518388,
      "learning_rate": 6.560945273631841e-07,
      "loss": 1.1764,
      "step": 211
    },
    {
      "epoch": 0.019559451043708913,
      "grad_norm": 1.1739892120429067,
      "learning_rate": 6.592039800995026e-07,
      "loss": 1.1399,
      "step": 212
    },
    {
      "epoch": 0.019651712605235842,
      "grad_norm": 1.6811487213429959,
      "learning_rate": 6.623134328358209e-07,
      "loss": 1.1451,
      "step": 213
    },
    {
      "epoch": 0.01974397416676277,
      "grad_norm": 2.340710465050252,
      "learning_rate": 6.654228855721393e-07,
      "loss": 1.0598,
      "step": 214
    },
    {
      "epoch": 0.0198362357282897,
      "grad_norm": 2.8002553717549903,
      "learning_rate": 6.685323383084577e-07,
      "loss": 1.0842,
      "step": 215
    },
    {
      "epoch": 0.0198362357282897,
      "eval_GEN Loss": 0.5170443058013916,
      "eval_GEN top-5 accuracy": 0.9745725558965366,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5624172687530518,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.143990397453308,
      "eval_runtime": 55.9108,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 215
    },
    {
      "epoch": 0.01992849728981663,
      "grad_norm": 3.214947687116344,
      "learning_rate": 6.716417910447762e-07,
      "loss": 1.1459,
      "step": 216
    },
    {
      "epoch": 0.020020758851343558,
      "grad_norm": 3.029276370298952,
      "learning_rate": 6.747512437810946e-07,
      "loss": 1.0847,
      "step": 217
    },
    {
      "epoch": 0.020113020412870487,
      "grad_norm": 1.663984961233634,
      "learning_rate": 6.77860696517413e-07,
      "loss": 1.2899,
      "step": 218
    },
    {
      "epoch": 0.020205281974397415,
      "grad_norm": 1.4544759741871345,
      "learning_rate": 6.809701492537314e-07,
      "loss": 1.1231,
      "step": 219
    },
    {
      "epoch": 0.020297543535924344,
      "grad_norm": 2.203854407410451,
      "learning_rate": 6.840796019900498e-07,
      "loss": 1.1777,
      "step": 220
    },
    {
      "epoch": 0.020297543535924344,
      "eval_GEN Loss": 0.5135334730148315,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5585988163948059,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.137379765510559,
      "eval_runtime": 56.0845,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 220
    },
    {
      "epoch": 0.020389805097451273,
      "grad_norm": 2.0670558917920747,
      "learning_rate": 6.871890547263682e-07,
      "loss": 1.1395,
      "step": 221
    },
    {
      "epoch": 0.020482066658978202,
      "grad_norm": 2.8771722472921004,
      "learning_rate": 6.902985074626866e-07,
      "loss": 0.9456,
      "step": 222
    },
    {
      "epoch": 0.02057432822050513,
      "grad_norm": 1.3261488963028663,
      "learning_rate": 6.93407960199005e-07,
      "loss": 1.0876,
      "step": 223
    },
    {
      "epoch": 0.02066658978203206,
      "grad_norm": 1.5730855667159154,
      "learning_rate": 6.965174129353235e-07,
      "loss": 1.0239,
      "step": 224
    },
    {
      "epoch": 0.02075885134355899,
      "grad_norm": 3.621240004877932,
      "learning_rate": 6.996268656716417e-07,
      "loss": 0.8469,
      "step": 225
    },
    {
      "epoch": 0.02075885134355899,
      "eval_GEN Loss": 0.5120242834091187,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5553549528121948,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1325721740722656,
      "eval_runtime": 56.002,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 225
    },
    {
      "epoch": 0.020851112905085917,
      "grad_norm": 1.484888422594663,
      "learning_rate": 7.027363184079603e-07,
      "loss": 1.0099,
      "step": 226
    },
    {
      "epoch": 0.020943374466612846,
      "grad_norm": 1.3755576735455195,
      "learning_rate": 7.058457711442787e-07,
      "loss": 1.1258,
      "step": 227
    },
    {
      "epoch": 0.021035636028139775,
      "grad_norm": 1.2243039274674814,
      "learning_rate": 7.08955223880597e-07,
      "loss": 1.0561,
      "step": 228
    },
    {
      "epoch": 0.021127897589666704,
      "grad_norm": 2.2021697636941155,
      "learning_rate": 7.120646766169155e-07,
      "loss": 1.1327,
      "step": 229
    },
    {
      "epoch": 0.021220159151193633,
      "grad_norm": 3.22855391632117,
      "learning_rate": 7.151741293532338e-07,
      "loss": 1.088,
      "step": 230
    },
    {
      "epoch": 0.021220159151193633,
      "eval_GEN Loss": 0.5095731616020203,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5523665547370911,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1322115659713745,
      "eval_runtime": 55.9755,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 230
    },
    {
      "epoch": 0.02131242071272056,
      "grad_norm": 2.1624933493323497,
      "learning_rate": 7.182835820895523e-07,
      "loss": 0.9234,
      "step": 231
    },
    {
      "epoch": 0.02140468227424749,
      "grad_norm": 1.9286550194070369,
      "learning_rate": 7.213930348258707e-07,
      "loss": 1.1006,
      "step": 232
    },
    {
      "epoch": 0.02149694383577442,
      "grad_norm": 1.9879188571883055,
      "learning_rate": 7.24502487562189e-07,
      "loss": 1.0354,
      "step": 233
    },
    {
      "epoch": 0.02158920539730135,
      "grad_norm": 1.7033235635672739,
      "learning_rate": 7.276119402985076e-07,
      "loss": 1.1862,
      "step": 234
    },
    {
      "epoch": 0.021681466958828277,
      "grad_norm": 2.971431635426475,
      "learning_rate": 7.307213930348259e-07,
      "loss": 0.8153,
      "step": 235
    },
    {
      "epoch": 0.021681466958828277,
      "eval_GEN Loss": 0.5085355639457703,
      "eval_GEN top-5 accuracy": 0.9763261727312582,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5487057566642761,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.131009578704834,
      "eval_runtime": 56.2583,
      "eval_samples_per_second": 1.155,
      "eval_steps_per_second": 0.089,
      "step": 235
    },
    {
      "epoch": 0.021773728520355206,
      "grad_norm": 2.3170545130056195,
      "learning_rate": 7.338308457711443e-07,
      "loss": 0.9272,
      "step": 236
    },
    {
      "epoch": 0.021865990081882135,
      "grad_norm": 2.5063211979179623,
      "learning_rate": 7.369402985074628e-07,
      "loss": 1.1512,
      "step": 237
    },
    {
      "epoch": 0.021958251643409064,
      "grad_norm": 2.6050090210752845,
      "learning_rate": 7.400497512437811e-07,
      "loss": 1.1128,
      "step": 238
    },
    {
      "epoch": 0.022050513204935993,
      "grad_norm": 0.9438638212182147,
      "learning_rate": 7.431592039800996e-07,
      "loss": 1.0492,
      "step": 239
    },
    {
      "epoch": 0.02214277476646292,
      "grad_norm": 1.7367803016092296,
      "learning_rate": 7.462686567164179e-07,
      "loss": 1.0682,
      "step": 240
    },
    {
      "epoch": 0.02214277476646292,
      "eval_GEN Loss": 0.5049706697463989,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5450387001037598,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1257210969924927,
      "eval_runtime": 55.9463,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 240
    },
    {
      "epoch": 0.02223503632798985,
      "grad_norm": 1.0587270737449548,
      "learning_rate": 7.493781094527363e-07,
      "loss": 1.0032,
      "step": 241
    },
    {
      "epoch": 0.02232729788951678,
      "grad_norm": 1.5482171931689053,
      "learning_rate": 7.524875621890549e-07,
      "loss": 1.1029,
      "step": 242
    },
    {
      "epoch": 0.022419559451043708,
      "grad_norm": 1.5518736555259462,
      "learning_rate": 7.555970149253732e-07,
      "loss": 1.0056,
      "step": 243
    },
    {
      "epoch": 0.022511821012570637,
      "grad_norm": 2.8331292567857047,
      "learning_rate": 7.587064676616916e-07,
      "loss": 1.0184,
      "step": 244
    },
    {
      "epoch": 0.022604082574097566,
      "grad_norm": 2.549284096825949,
      "learning_rate": 7.6181592039801e-07,
      "loss": 0.9846,
      "step": 245
    },
    {
      "epoch": 0.022604082574097566,
      "eval_GEN Loss": 0.5028582215309143,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5410569310188293,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.126201868057251,
      "eval_runtime": 55.9978,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 245
    },
    {
      "epoch": 0.022696344135624495,
      "grad_norm": 2.7637017707401457,
      "learning_rate": 7.649253731343284e-07,
      "loss": 1.1036,
      "step": 246
    },
    {
      "epoch": 0.022788605697151423,
      "grad_norm": 1.7165914392239319,
      "learning_rate": 7.680348258706469e-07,
      "loss": 1.0846,
      "step": 247
    },
    {
      "epoch": 0.022880867258678352,
      "grad_norm": 1.7964868558424398,
      "learning_rate": 7.711442786069652e-07,
      "loss": 1.1794,
      "step": 248
    },
    {
      "epoch": 0.02297312882020528,
      "grad_norm": 1.5067101117393125,
      "learning_rate": 7.742537313432836e-07,
      "loss": 1.1901,
      "step": 249
    },
    {
      "epoch": 0.02306539038173221,
      "grad_norm": 1.885967326150524,
      "learning_rate": 7.77363184079602e-07,
      "loss": 1.1061,
      "step": 250
    },
    {
      "epoch": 0.02306539038173221,
      "eval_GEN Loss": 0.49827855825424194,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5379000902175903,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.120192289352417,
      "eval_runtime": 56.2878,
      "eval_samples_per_second": 1.155,
      "eval_steps_per_second": 0.089,
      "step": 250
    },
    {
      "epoch": 0.02315765194325914,
      "grad_norm": 1.7081726615257915,
      "learning_rate": 7.804726368159204e-07,
      "loss": 0.9803,
      "step": 251
    },
    {
      "epoch": 0.023249913504786068,
      "grad_norm": 1.962242829716037,
      "learning_rate": 7.835820895522389e-07,
      "loss": 1.1248,
      "step": 252
    },
    {
      "epoch": 0.023342175066312996,
      "grad_norm": 1.3568711442449137,
      "learning_rate": 7.866915422885573e-07,
      "loss": 1.1357,
      "step": 253
    },
    {
      "epoch": 0.023434436627839925,
      "grad_norm": 1.847285767372738,
      "learning_rate": 7.898009950248757e-07,
      "loss": 1.1333,
      "step": 254
    },
    {
      "epoch": 0.023526698189366854,
      "grad_norm": 2.2990666027216897,
      "learning_rate": 7.929104477611942e-07,
      "loss": 0.8898,
      "step": 255
    },
    {
      "epoch": 0.023526698189366854,
      "eval_GEN Loss": 0.49851059913635254,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5344753861427307,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1207932233810425,
      "eval_runtime": 55.867,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 255
    },
    {
      "epoch": 0.023618959750893783,
      "grad_norm": 1.9308137024234377,
      "learning_rate": 7.960199004975125e-07,
      "loss": 1.1893,
      "step": 256
    },
    {
      "epoch": 0.023711221312420712,
      "grad_norm": 2.4493692503226265,
      "learning_rate": 7.991293532338309e-07,
      "loss": 1.1574,
      "step": 257
    },
    {
      "epoch": 0.02380348287394764,
      "grad_norm": 1.553760673770195,
      "learning_rate": 8.022388059701493e-07,
      "loss": 1.0206,
      "step": 258
    },
    {
      "epoch": 0.02389574443547457,
      "grad_norm": 2.010889516326646,
      "learning_rate": 8.053482587064677e-07,
      "loss": 1.1219,
      "step": 259
    },
    {
      "epoch": 0.0239880059970015,
      "grad_norm": 2.2948511162812566,
      "learning_rate": 8.084577114427862e-07,
      "loss": 1.0648,
      "step": 260
    },
    {
      "epoch": 0.0239880059970015,
      "eval_GEN Loss": 0.499536395072937,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5314286947250366,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.123798131942749,
      "eval_runtime": 56.2249,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 260
    },
    {
      "epoch": 0.024080267558528427,
      "grad_norm": 1.3984976010614756,
      "learning_rate": 8.115671641791046e-07,
      "loss": 1.0773,
      "step": 261
    },
    {
      "epoch": 0.024172529120055356,
      "grad_norm": 2.7309847518077404,
      "learning_rate": 8.14676616915423e-07,
      "loss": 1.1736,
      "step": 262
    },
    {
      "epoch": 0.024264790681582285,
      "grad_norm": 1.3141521911071716,
      "learning_rate": 8.177860696517413e-07,
      "loss": 0.9211,
      "step": 263
    },
    {
      "epoch": 0.024357052243109214,
      "grad_norm": 2.1621667375131914,
      "learning_rate": 8.208955223880598e-07,
      "loss": 1.1967,
      "step": 264
    },
    {
      "epoch": 0.024449313804636143,
      "grad_norm": 1.65439325524114,
      "learning_rate": 8.240049751243782e-07,
      "loss": 0.9722,
      "step": 265
    },
    {
      "epoch": 0.024449313804636143,
      "eval_GEN Loss": 0.49999383091926575,
      "eval_GEN top-5 accuracy": 0.9763261727312582,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5308574438095093,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1251201629638672,
      "eval_runtime": 55.8955,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 265
    },
    {
      "epoch": 0.02454157536616307,
      "grad_norm": 2.623145414454357,
      "learning_rate": 8.271144278606966e-07,
      "loss": 1.1088,
      "step": 266
    },
    {
      "epoch": 0.02463383692769,
      "grad_norm": 2.288403387365106,
      "learning_rate": 8.30223880597015e-07,
      "loss": 1.0018,
      "step": 267
    },
    {
      "epoch": 0.02472609848921693,
      "grad_norm": 1.4482698240184462,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.0952,
      "step": 268
    },
    {
      "epoch": 0.024818360050743858,
      "grad_norm": 1.1600946139990198,
      "learning_rate": 8.364427860696518e-07,
      "loss": 0.9762,
      "step": 269
    },
    {
      "epoch": 0.024910621612270787,
      "grad_norm": 1.4967658760734532,
      "learning_rate": 8.395522388059703e-07,
      "loss": 1.2571,
      "step": 270
    },
    {
      "epoch": 0.024910621612270787,
      "eval_GEN Loss": 0.5007842779159546,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5283689498901367,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.125,
      "eval_runtime": 55.9944,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 270
    },
    {
      "epoch": 0.025002883173797716,
      "grad_norm": 2.328523158673554,
      "learning_rate": 8.426616915422886e-07,
      "loss": 1.0343,
      "step": 271
    },
    {
      "epoch": 0.025095144735324645,
      "grad_norm": 1.4909803648560656,
      "learning_rate": 8.457711442786071e-07,
      "loss": 1.104,
      "step": 272
    },
    {
      "epoch": 0.025187406296851574,
      "grad_norm": 1.8657451248588977,
      "learning_rate": 8.488805970149254e-07,
      "loss": 0.9584,
      "step": 273
    },
    {
      "epoch": 0.025279667858378502,
      "grad_norm": 1.0806269013565903,
      "learning_rate": 8.519900497512439e-07,
      "loss": 0.9994,
      "step": 274
    },
    {
      "epoch": 0.02537192941990543,
      "grad_norm": 1.8828202710100699,
      "learning_rate": 8.550995024875623e-07,
      "loss": 1.0252,
      "step": 275
    },
    {
      "epoch": 0.02537192941990543,
      "eval_GEN Loss": 0.4993970990180969,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5242939591407776,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1192307472229004,
      "eval_runtime": 55.9406,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 275
    },
    {
      "epoch": 0.02546419098143236,
      "grad_norm": 1.9910428682832662,
      "learning_rate": 8.582089552238806e-07,
      "loss": 0.9602,
      "step": 276
    },
    {
      "epoch": 0.02555645254295929,
      "grad_norm": 2.205952921717709,
      "learning_rate": 8.613184079601991e-07,
      "loss": 1.1484,
      "step": 277
    },
    {
      "epoch": 0.025648714104486218,
      "grad_norm": 1.6616398335829814,
      "learning_rate": 8.644278606965174e-07,
      "loss": 1.0244,
      "step": 278
    },
    {
      "epoch": 0.025740975666013147,
      "grad_norm": 1.6161884507818356,
      "learning_rate": 8.675373134328359e-07,
      "loss": 0.8702,
      "step": 279
    },
    {
      "epoch": 0.025833237227540076,
      "grad_norm": 1.1562188681419945,
      "learning_rate": 8.706467661691544e-07,
      "loss": 1.0532,
      "step": 280
    },
    {
      "epoch": 0.025833237227540076,
      "eval_GEN Loss": 0.49678125977516174,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5206115245819092,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.115625023841858,
      "eval_runtime": 55.937,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 280
    },
    {
      "epoch": 0.025925498789067004,
      "grad_norm": 1.9707863917005386,
      "learning_rate": 8.737562189054727e-07,
      "loss": 1.0873,
      "step": 281
    },
    {
      "epoch": 0.026017760350593933,
      "grad_norm": 4.278966152269661,
      "learning_rate": 8.768656716417912e-07,
      "loss": 1.277,
      "step": 282
    },
    {
      "epoch": 0.026110021912120862,
      "grad_norm": 1.2908293416193641,
      "learning_rate": 8.799751243781095e-07,
      "loss": 0.9749,
      "step": 283
    },
    {
      "epoch": 0.02620228347364779,
      "grad_norm": 1.2684817107225679,
      "learning_rate": 8.830845771144279e-07,
      "loss": 1.0421,
      "step": 284
    },
    {
      "epoch": 0.02629454503517472,
      "grad_norm": 1.5940989867178927,
      "learning_rate": 8.861940298507464e-07,
      "loss": 1.1096,
      "step": 285
    },
    {
      "epoch": 0.02629454503517472,
      "eval_GEN Loss": 0.4946899712085724,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5191790461540222,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1149038076400757,
      "eval_runtime": 56.0251,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 285
    },
    {
      "epoch": 0.02638680659670165,
      "grad_norm": 2.055428248373644,
      "learning_rate": 8.893034825870647e-07,
      "loss": 0.9623,
      "step": 286
    },
    {
      "epoch": 0.026479068158228578,
      "grad_norm": 1.9264725448015343,
      "learning_rate": 8.924129353233831e-07,
      "loss": 0.9619,
      "step": 287
    },
    {
      "epoch": 0.026571329719755506,
      "grad_norm": 2.4195160243952674,
      "learning_rate": 8.955223880597015e-07,
      "loss": 1.2104,
      "step": 288
    },
    {
      "epoch": 0.026663591281282435,
      "grad_norm": 1.9500252731013148,
      "learning_rate": 8.9863184079602e-07,
      "loss": 1.1173,
      "step": 289
    },
    {
      "epoch": 0.026755852842809364,
      "grad_norm": 1.1811091155512554,
      "learning_rate": 9.017412935323385e-07,
      "loss": 0.985,
      "step": 290
    },
    {
      "epoch": 0.026755852842809364,
      "eval_GEN Loss": 0.4933905303478241,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5175021290779114,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1144230365753174,
      "eval_runtime": 57.0475,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 290
    },
    {
      "epoch": 0.026848114404336293,
      "grad_norm": 1.2546086202256017,
      "learning_rate": 9.048507462686568e-07,
      "loss": 1.0852,
      "step": 291
    },
    {
      "epoch": 0.026940375965863222,
      "grad_norm": 3.1134612905257883,
      "learning_rate": 9.079601990049752e-07,
      "loss": 1.1482,
      "step": 292
    },
    {
      "epoch": 0.02703263752739015,
      "grad_norm": 1.400022154726557,
      "learning_rate": 9.110696517412936e-07,
      "loss": 1.0331,
      "step": 293
    },
    {
      "epoch": 0.02712489908891708,
      "grad_norm": 1.0246627685545686,
      "learning_rate": 9.14179104477612e-07,
      "loss": 1.0233,
      "step": 294
    },
    {
      "epoch": 0.02721716065044401,
      "grad_norm": 2.7396386216256223,
      "learning_rate": 9.172885572139304e-07,
      "loss": 1.1775,
      "step": 295
    },
    {
      "epoch": 0.02721716065044401,
      "eval_GEN Loss": 0.49512484669685364,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5191892385482788,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.118749976158142,
      "eval_runtime": 56.9037,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 295
    },
    {
      "epoch": 0.027309422211970937,
      "grad_norm": 1.5007117831839922,
      "learning_rate": 9.203980099502488e-07,
      "loss": 0.976,
      "step": 296
    },
    {
      "epoch": 0.027401683773497866,
      "grad_norm": 1.5792917701848133,
      "learning_rate": 9.235074626865673e-07,
      "loss": 1.2102,
      "step": 297
    },
    {
      "epoch": 0.027493945335024795,
      "grad_norm": 3.098253837116703,
      "learning_rate": 9.266169154228855e-07,
      "loss": 1.2062,
      "step": 298
    },
    {
      "epoch": 0.027586206896551724,
      "grad_norm": 2.44234681946125,
      "learning_rate": 9.297263681592041e-07,
      "loss": 1.1697,
      "step": 299
    },
    {
      "epoch": 0.027678468458078653,
      "grad_norm": 1.3977209211043808,
      "learning_rate": 9.328358208955225e-07,
      "loss": 1.0619,
      "step": 300
    },
    {
      "epoch": 0.027678468458078653,
      "eval_GEN Loss": 0.4945972263813019,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5235617756843567,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1182692050933838,
      "eval_runtime": 56.9471,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 300
    },
    {
      "epoch": 0.02777073001960558,
      "grad_norm": 1.7627879947578182,
      "learning_rate": 9.359452736318409e-07,
      "loss": 0.9506,
      "step": 301
    },
    {
      "epoch": 0.02786299158113251,
      "grad_norm": 2.5988992466232568,
      "learning_rate": 9.390547263681593e-07,
      "loss": 0.8767,
      "step": 302
    },
    {
      "epoch": 0.02795525314265944,
      "grad_norm": 4.085254431764655,
      "learning_rate": 9.421641791044776e-07,
      "loss": 1.1512,
      "step": 303
    },
    {
      "epoch": 0.028047514704186368,
      "grad_norm": 2.6136174838400663,
      "learning_rate": 9.452736318407961e-07,
      "loss": 1.103,
      "step": 304
    },
    {
      "epoch": 0.028139776265713297,
      "grad_norm": 2.108395179560001,
      "learning_rate": 9.483830845771146e-07,
      "loss": 1.1541,
      "step": 305
    },
    {
      "epoch": 0.028139776265713297,
      "eval_GEN Loss": 0.4926488697528839,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5267210006713867,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.113942265510559,
      "eval_runtime": 56.8595,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 305
    },
    {
      "epoch": 0.028232037827240226,
      "grad_norm": 2.495153869564682,
      "learning_rate": 9.514925373134328e-07,
      "loss": 0.9752,
      "step": 306
    },
    {
      "epoch": 0.028324299388767155,
      "grad_norm": 2.387247198562086,
      "learning_rate": 9.546019900497513e-07,
      "loss": 0.8936,
      "step": 307
    },
    {
      "epoch": 0.028416560950294083,
      "grad_norm": 1.460754878627914,
      "learning_rate": 9.577114427860696e-07,
      "loss": 0.9544,
      "step": 308
    },
    {
      "epoch": 0.028508822511821012,
      "grad_norm": 2.0395098214067002,
      "learning_rate": 9.608208955223882e-07,
      "loss": 0.9313,
      "step": 309
    },
    {
      "epoch": 0.02860108407334794,
      "grad_norm": 1.437449786362347,
      "learning_rate": 9.639303482587067e-07,
      "loss": 1.0574,
      "step": 310
    },
    {
      "epoch": 0.02860108407334794,
      "eval_GEN Loss": 0.49192938208580017,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.527777373790741,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1115384101867676,
      "eval_runtime": 56.9538,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 310
    },
    {
      "epoch": 0.02869334563487487,
      "grad_norm": 1.5733885494961437,
      "learning_rate": 9.670398009950248e-07,
      "loss": 0.9618,
      "step": 311
    },
    {
      "epoch": 0.0287856071964018,
      "grad_norm": 2.604042706490351,
      "learning_rate": 9.701492537313434e-07,
      "loss": 0.9865,
      "step": 312
    },
    {
      "epoch": 0.028877868757928728,
      "grad_norm": 2.1168350541971095,
      "learning_rate": 9.732587064676617e-07,
      "loss": 1.0898,
      "step": 313
    },
    {
      "epoch": 0.028970130319455657,
      "grad_norm": 1.3268050898742347,
      "learning_rate": 9.763681592039802e-07,
      "loss": 1.1057,
      "step": 314
    },
    {
      "epoch": 0.029062391880982585,
      "grad_norm": 1.9214658491996879,
      "learning_rate": 9.794776119402986e-07,
      "loss": 1.1069,
      "step": 315
    },
    {
      "epoch": 0.029062391880982585,
      "eval_GEN Loss": 0.4930776357650757,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5243564248085022,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1070913076400757,
      "eval_runtime": 56.8272,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 315
    },
    {
      "epoch": 0.029154653442509514,
      "grad_norm": 1.391531246930202,
      "learning_rate": 9.82587064676617e-07,
      "loss": 1.0834,
      "step": 316
    },
    {
      "epoch": 0.029246915004036443,
      "grad_norm": 1.7684232974539065,
      "learning_rate": 9.856965174129355e-07,
      "loss": 0.9377,
      "step": 317
    },
    {
      "epoch": 0.029339176565563372,
      "grad_norm": 1.394772844206612,
      "learning_rate": 9.888059701492538e-07,
      "loss": 1.0288,
      "step": 318
    },
    {
      "epoch": 0.0294314381270903,
      "grad_norm": 1.954988795164839,
      "learning_rate": 9.919154228855721e-07,
      "loss": 1.0052,
      "step": 319
    },
    {
      "epoch": 0.02952369968861723,
      "grad_norm": 2.2258500441193436,
      "learning_rate": 9.950248756218907e-07,
      "loss": 0.9812,
      "step": 320
    },
    {
      "epoch": 0.02952369968861723,
      "eval_GEN Loss": 0.49087533354759216,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5189715623855591,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.1018028259277344,
      "eval_runtime": 57.0008,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 320
    },
    {
      "epoch": 0.02961596125014416,
      "grad_norm": 2.3727144188815235,
      "learning_rate": 9.98134328358209e-07,
      "loss": 1.1373,
      "step": 321
    },
    {
      "epoch": 0.029708222811671087,
      "grad_norm": 1.8414136026249512,
      "learning_rate": 1.0012437810945275e-06,
      "loss": 1.0329,
      "step": 322
    },
    {
      "epoch": 0.029800484373198016,
      "grad_norm": 2.183721854667889,
      "learning_rate": 1.0043532338308459e-06,
      "loss": 1.1187,
      "step": 323
    },
    {
      "epoch": 0.029892745934724945,
      "grad_norm": 2.709076311722809,
      "learning_rate": 1.0074626865671642e-06,
      "loss": 1.0809,
      "step": 324
    },
    {
      "epoch": 0.029985007496251874,
      "grad_norm": 2.0096601825645157,
      "learning_rate": 1.0105721393034828e-06,
      "loss": 1.0652,
      "step": 325
    },
    {
      "epoch": 0.029985007496251874,
      "eval_GEN Loss": 0.4871484637260437,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.521902322769165,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.095312476158142,
      "eval_runtime": 56.9629,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 325
    },
    {
      "epoch": 0.030077269057778803,
      "grad_norm": 1.4424187514027385,
      "learning_rate": 1.013681592039801e-06,
      "loss": 0.978,
      "step": 326
    },
    {
      "epoch": 0.03016953061930573,
      "grad_norm": 2.009952390028351,
      "learning_rate": 1.0167910447761194e-06,
      "loss": 1.045,
      "step": 327
    },
    {
      "epoch": 0.03026179218083266,
      "grad_norm": 1.6533431602033455,
      "learning_rate": 1.0199004975124378e-06,
      "loss": 1.1309,
      "step": 328
    },
    {
      "epoch": 0.03035405374235959,
      "grad_norm": 1.1151714587863297,
      "learning_rate": 1.0230099502487563e-06,
      "loss": 1.0358,
      "step": 329
    },
    {
      "epoch": 0.030446315303886518,
      "grad_norm": 3.18926319432028,
      "learning_rate": 1.0261194029850748e-06,
      "loss": 1.2152,
      "step": 330
    },
    {
      "epoch": 0.030446315303886518,
      "eval_GEN Loss": 0.4860176146030426,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5272826552391052,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.092548131942749,
      "eval_runtime": 56.2808,
      "eval_samples_per_second": 1.155,
      "eval_steps_per_second": 0.089,
      "step": 330
    },
    {
      "epoch": 0.030538576865413447,
      "grad_norm": 2.5222962270206155,
      "learning_rate": 1.0292288557213932e-06,
      "loss": 0.913,
      "step": 331
    },
    {
      "epoch": 0.030630838426940376,
      "grad_norm": 2.71533521054579,
      "learning_rate": 1.0323383084577115e-06,
      "loss": 0.9796,
      "step": 332
    },
    {
      "epoch": 0.030723099988467305,
      "grad_norm": 1.4512743704884454,
      "learning_rate": 1.0354477611940298e-06,
      "loss": 1.0495,
      "step": 333
    },
    {
      "epoch": 0.030815361549994234,
      "grad_norm": 1.649164586450439,
      "learning_rate": 1.0385572139303484e-06,
      "loss": 1.1252,
      "step": 334
    },
    {
      "epoch": 0.030907623111521162,
      "grad_norm": 1.4892318199359345,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 1.0085,
      "step": 335
    },
    {
      "epoch": 0.030907623111521162,
      "eval_GEN Loss": 0.4881289303302765,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5235222578048706,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.0868990421295166,
      "eval_runtime": 55.9602,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 335
    },
    {
      "epoch": 0.03099988467304809,
      "grad_norm": 1.532233030413437,
      "learning_rate": 1.044776119402985e-06,
      "loss": 1.1126,
      "step": 336
    },
    {
      "epoch": 0.03109214623457502,
      "grad_norm": 1.9975806321345762,
      "learning_rate": 1.0478855721393036e-06,
      "loss": 1.1373,
      "step": 337
    },
    {
      "epoch": 0.03118440779610195,
      "grad_norm": 1.922418107580109,
      "learning_rate": 1.050995024875622e-06,
      "loss": 1.1442,
      "step": 338
    },
    {
      "epoch": 0.03127666935762888,
      "grad_norm": 2.5195638644586813,
      "learning_rate": 1.0541044776119405e-06,
      "loss": 0.8737,
      "step": 339
    },
    {
      "epoch": 0.0313689309191558,
      "grad_norm": 1.935740382197344,
      "learning_rate": 1.0572139303482588e-06,
      "loss": 0.9634,
      "step": 340
    },
    {
      "epoch": 0.0313689309191558,
      "eval_GEN Loss": 0.4911785125732422,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.523034930229187,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.090504765510559,
      "eval_runtime": 55.9064,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 340
    },
    {
      "epoch": 0.031461192480682736,
      "grad_norm": 1.9486286152465724,
      "learning_rate": 1.0603233830845771e-06,
      "loss": 0.9855,
      "step": 341
    },
    {
      "epoch": 0.03155345404220966,
      "grad_norm": 1.8546603370864803,
      "learning_rate": 1.0634328358208957e-06,
      "loss": 0.9511,
      "step": 342
    },
    {
      "epoch": 0.03164571560373659,
      "grad_norm": 1.7799476055488155,
      "learning_rate": 1.066542288557214e-06,
      "loss": 1.2302,
      "step": 343
    },
    {
      "epoch": 0.03173797716526352,
      "grad_norm": 1.680439411630123,
      "learning_rate": 1.0696517412935323e-06,
      "loss": 1.0727,
      "step": 344
    },
    {
      "epoch": 0.03183023872679045,
      "grad_norm": 1.6792308304460757,
      "learning_rate": 1.0727611940298509e-06,
      "loss": 0.9026,
      "step": 345
    },
    {
      "epoch": 0.03183023872679045,
      "eval_GEN Loss": 0.49160081148147583,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.518447756767273,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.0872596502304077,
      "eval_runtime": 56.2021,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 345
    },
    {
      "epoch": 0.031922500288317376,
      "grad_norm": 2.1885529669090316,
      "learning_rate": 1.0758706467661692e-06,
      "loss": 0.995,
      "step": 346
    },
    {
      "epoch": 0.03201476184984431,
      "grad_norm": 2.742387418896271,
      "learning_rate": 1.0789800995024878e-06,
      "loss": 0.7945,
      "step": 347
    },
    {
      "epoch": 0.032107023411371234,
      "grad_norm": 1.4403393167685103,
      "learning_rate": 1.0820895522388059e-06,
      "loss": 1.0051,
      "step": 348
    },
    {
      "epoch": 0.032199284972898166,
      "grad_norm": 1.371955107543284,
      "learning_rate": 1.0851990049751244e-06,
      "loss": 1.1058,
      "step": 349
    },
    {
      "epoch": 0.03229154653442509,
      "grad_norm": 2.8464898439478494,
      "learning_rate": 1.088308457711443e-06,
      "loss": 0.8464,
      "step": 350
    },
    {
      "epoch": 0.03229154653442509,
      "eval_GEN Loss": 0.493305504322052,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.511089026927948,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.0890624523162842,
      "eval_runtime": 55.8793,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 350
    },
    {
      "epoch": 0.032383808095952024,
      "grad_norm": 1.5514112449284048,
      "learning_rate": 1.0914179104477613e-06,
      "loss": 1.0468,
      "step": 351
    },
    {
      "epoch": 0.03247606965747895,
      "grad_norm": 1.6275512940031138,
      "learning_rate": 1.0945273631840796e-06,
      "loss": 1.0689,
      "step": 352
    },
    {
      "epoch": 0.03256833121900588,
      "grad_norm": 1.8446916567920402,
      "learning_rate": 1.0976368159203982e-06,
      "loss": 1.0027,
      "step": 353
    },
    {
      "epoch": 0.03266059278053281,
      "grad_norm": 2.2633912193599217,
      "learning_rate": 1.1007462686567165e-06,
      "loss": 1.1833,
      "step": 354
    },
    {
      "epoch": 0.03275285434205974,
      "grad_norm": 1.8695264550249542,
      "learning_rate": 1.1038557213930349e-06,
      "loss": 1.0588,
      "step": 355
    },
    {
      "epoch": 0.03275285434205974,
      "eval_GEN Loss": 0.4953005909919739,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8783068783068783,
      "eval_PRM F1 AUC": 0.5,
      "eval_PRM F1 Neg": 0.0,
      "eval_PRM Loss": 0.5123229026794434,
      "eval_PRM NPV": 0.0,
      "eval_PRM Precision": 0.7830188679245284,
      "eval_PRM Recall": 1.0,
      "eval_PRM Specificty": 0.0,
      "eval_loss": 1.087499976158142,
      "eval_runtime": 56.1679,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 355
    },
    {
      "epoch": 0.032845115903586665,
      "grad_norm": 1.8335053955860532,
      "learning_rate": 1.1069651741293532e-06,
      "loss": 1.037,
      "step": 356
    },
    {
      "epoch": 0.0329373774651136,
      "grad_norm": 1.6486659896008766,
      "learning_rate": 1.1100746268656717e-06,
      "loss": 1.1072,
      "step": 357
    },
    {
      "epoch": 0.03302963902664052,
      "grad_norm": 1.572195389553314,
      "learning_rate": 1.1131840796019903e-06,
      "loss": 0.9697,
      "step": 358
    },
    {
      "epoch": 0.033121900588167455,
      "grad_norm": 2.7687689049610777,
      "learning_rate": 1.1162935323383086e-06,
      "loss": 0.916,
      "step": 359
    },
    {
      "epoch": 0.03321416214969438,
      "grad_norm": 1.7728085056791567,
      "learning_rate": 1.119402985074627e-06,
      "loss": 1.0788,
      "step": 360
    },
    {
      "epoch": 0.03321416214969438,
      "eval_GEN Loss": 0.4948984682559967,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8839779005524862,
      "eval_PRM F1 AUC": 0.5906233630172866,
      "eval_PRM F1 Neg": 0.3225806451612903,
      "eval_PRM Loss": 0.5221074223518372,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.963855421686747,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.0902644395828247,
      "eval_runtime": 56.9761,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 360
    },
    {
      "epoch": 0.03330642371122131,
      "grad_norm": 1.5701375146458443,
      "learning_rate": 1.1225124378109453e-06,
      "loss": 1.0978,
      "step": 361
    },
    {
      "epoch": 0.03339868527274824,
      "grad_norm": 2.1204690039490304,
      "learning_rate": 1.1256218905472638e-06,
      "loss": 1.0687,
      "step": 362
    },
    {
      "epoch": 0.03349094683427517,
      "grad_norm": 1.601233790604872,
      "learning_rate": 1.1287313432835822e-06,
      "loss": 0.9983,
      "step": 363
    },
    {
      "epoch": 0.033583208395802096,
      "grad_norm": 1.8898677542781388,
      "learning_rate": 1.1318407960199005e-06,
      "loss": 0.9063,
      "step": 364
    },
    {
      "epoch": 0.03367546995732903,
      "grad_norm": 2.2649661149457727,
      "learning_rate": 1.134950248756219e-06,
      "loss": 0.9561,
      "step": 365
    },
    {
      "epoch": 0.03367546995732903,
      "eval_GEN Loss": 0.4931755065917969,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8839779005524862,
      "eval_PRM F1 AUC": 0.5906233630172866,
      "eval_PRM F1 Neg": 0.3225806451612903,
      "eval_PRM Loss": 0.5178951025009155,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.963855421686747,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.088641881942749,
      "eval_runtime": 56.9362,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 365
    },
    {
      "epoch": 0.033767731518855953,
      "grad_norm": 2.5696075162247354,
      "learning_rate": 1.1380597014925374e-06,
      "loss": 0.8822,
      "step": 366
    },
    {
      "epoch": 0.033859993080382886,
      "grad_norm": 1.4433836855901425,
      "learning_rate": 1.141169154228856e-06,
      "loss": 1.0123,
      "step": 367
    },
    {
      "epoch": 0.03395225464190981,
      "grad_norm": 1.4558018932634749,
      "learning_rate": 1.1442786069651742e-06,
      "loss": 1.0535,
      "step": 368
    },
    {
      "epoch": 0.034044516203436744,
      "grad_norm": 3.175675569714823,
      "learning_rate": 1.1473880597014926e-06,
      "loss": 1.2205,
      "step": 369
    },
    {
      "epoch": 0.03413677776496367,
      "grad_norm": 1.7693144783390349,
      "learning_rate": 1.1504975124378111e-06,
      "loss": 0.9804,
      "step": 370
    },
    {
      "epoch": 0.03413677776496367,
      "eval_GEN Loss": 0.4947631359100342,
      "eval_GEN top-5 accuracy": 0.9754493643138974,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8839779005524862,
      "eval_PRM F1 AUC": 0.5906233630172866,
      "eval_PRM F1 Neg": 0.3225806451612903,
      "eval_PRM Loss": 0.5144875645637512,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.963855421686747,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.0914663076400757,
      "eval_runtime": 56.9019,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 370
    },
    {
      "epoch": 0.0342290393264906,
      "grad_norm": 1.7432989957940266,
      "learning_rate": 1.1536069651741295e-06,
      "loss": 0.9745,
      "step": 371
    },
    {
      "epoch": 0.03432130088801753,
      "grad_norm": 1.62302846536514,
      "learning_rate": 1.1567164179104478e-06,
      "loss": 0.9044,
      "step": 372
    },
    {
      "epoch": 0.03441356244954446,
      "grad_norm": 2.1506916661651183,
      "learning_rate": 1.1598258706467663e-06,
      "loss": 1.1325,
      "step": 373
    },
    {
      "epoch": 0.034505824011071384,
      "grad_norm": 1.7935705697116229,
      "learning_rate": 1.1629353233830847e-06,
      "loss": 0.9483,
      "step": 374
    },
    {
      "epoch": 0.03459808557259832,
      "grad_norm": 1.763019620661752,
      "learning_rate": 1.1660447761194032e-06,
      "loss": 0.9373,
      "step": 375
    },
    {
      "epoch": 0.03459808557259832,
      "eval_GEN Loss": 0.4980253577232361,
      "eval_GEN top-5 accuracy": 0.9754493643138974,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8901098901098901,
      "eval_PRM F1 AUC": 0.5966474594028287,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM Loss": 0.513366162776947,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.8181818181818182,
      "eval_PRM Recall": 0.9759036144578314,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.0947115421295166,
      "eval_runtime": 56.9597,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 375
    },
    {
      "epoch": 0.03469034713412524,
      "grad_norm": 1.768254678969684,
      "learning_rate": 1.1691542288557215e-06,
      "loss": 1.0563,
      "step": 376
    },
    {
      "epoch": 0.034782608695652174,
      "grad_norm": 1.8553706418639913,
      "learning_rate": 1.1722636815920399e-06,
      "loss": 1.2008,
      "step": 377
    },
    {
      "epoch": 0.0348748702571791,
      "grad_norm": 1.4481012889071891,
      "learning_rate": 1.1753731343283584e-06,
      "loss": 1.0245,
      "step": 378
    },
    {
      "epoch": 0.03496713181870603,
      "grad_norm": 1.7256487619893692,
      "learning_rate": 1.1784825870646768e-06,
      "loss": 1.002,
      "step": 379
    },
    {
      "epoch": 0.03505939338023296,
      "grad_norm": 2.5508092701959217,
      "learning_rate": 1.181592039800995e-06,
      "loss": 0.9065,
      "step": 380
    },
    {
      "epoch": 0.03505939338023296,
      "eval_GEN Loss": 0.495575875043869,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8839779005524862,
      "eval_PRM F1 AUC": 0.5906233630172866,
      "eval_PRM F1 Neg": 0.3225806451612903,
      "eval_PRM Loss": 0.5119543075561523,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.963855421686747,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.0922476053237915,
      "eval_runtime": 57.038,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 380
    },
    {
      "epoch": 0.03515165494175989,
      "grad_norm": 1.4722059115710762,
      "learning_rate": 1.1847014925373134e-06,
      "loss": 0.9706,
      "step": 381
    },
    {
      "epoch": 0.035243916503286815,
      "grad_norm": 1.536297412134017,
      "learning_rate": 1.187810945273632e-06,
      "loss": 0.9951,
      "step": 382
    },
    {
      "epoch": 0.03533617806481375,
      "grad_norm": 1.5079787215711309,
      "learning_rate": 1.1909203980099505e-06,
      "loss": 1.0306,
      "step": 383
    },
    {
      "epoch": 0.03542843962634067,
      "grad_norm": 1.5422442471822861,
      "learning_rate": 1.1940298507462686e-06,
      "loss": 1.0272,
      "step": 384
    },
    {
      "epoch": 0.035520701187867605,
      "grad_norm": 1.657448677437209,
      "learning_rate": 1.1971393034825872e-06,
      "loss": 0.838,
      "step": 385
    },
    {
      "epoch": 0.035520701187867605,
      "eval_GEN Loss": 0.49375835061073303,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8839779005524862,
      "eval_PRM F1 AUC": 0.5906233630172866,
      "eval_PRM F1 Neg": 0.3225806451612903,
      "eval_PRM Loss": 0.500721275806427,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.963855421686747,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.0851562023162842,
      "eval_runtime": 56.0198,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 385
    },
    {
      "epoch": 0.03561296274939453,
      "grad_norm": 2.249630661061084,
      "learning_rate": 1.2002487562189055e-06,
      "loss": 1.1123,
      "step": 386
    },
    {
      "epoch": 0.03570522431092146,
      "grad_norm": 1.5608366240192577,
      "learning_rate": 1.203358208955224e-06,
      "loss": 0.9374,
      "step": 387
    },
    {
      "epoch": 0.03579748587244839,
      "grad_norm": 2.726231550430038,
      "learning_rate": 1.2064676616915424e-06,
      "loss": 1.1078,
      "step": 388
    },
    {
      "epoch": 0.03588974743397532,
      "grad_norm": 1.358167054618328,
      "learning_rate": 1.2095771144278607e-06,
      "loss": 1.0385,
      "step": 389
    },
    {
      "epoch": 0.035982008995502246,
      "grad_norm": 1.4329854592023732,
      "learning_rate": 1.2126865671641793e-06,
      "loss": 0.9826,
      "step": 390
    },
    {
      "epoch": 0.035982008995502246,
      "eval_GEN Loss": 0.4905219078063965,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8839779005524862,
      "eval_PRM F1 AUC": 0.5906233630172866,
      "eval_PRM F1 Neg": 0.3225806451612903,
      "eval_PRM Loss": 0.4987432360649109,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.8163265306122449,
      "eval_PRM Recall": 0.963855421686747,
      "eval_PRM Specificty": 0.21739130434782608,
      "eval_loss": 1.0786057710647583,
      "eval_runtime": 55.9471,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 390
    },
    {
      "epoch": 0.03607427055702918,
      "grad_norm": 1.8050566080096448,
      "learning_rate": 1.2157960199004976e-06,
      "loss": 0.8972,
      "step": 391
    },
    {
      "epoch": 0.036166532118556104,
      "grad_norm": 2.425921846861029,
      "learning_rate": 1.218905472636816e-06,
      "loss": 1.075,
      "step": 392
    },
    {
      "epoch": 0.036258793680083036,
      "grad_norm": 2.7081453428234146,
      "learning_rate": 1.2220149253731345e-06,
      "loss": 0.7915,
      "step": 393
    },
    {
      "epoch": 0.03635105524160996,
      "grad_norm": 1.4076303908682901,
      "learning_rate": 1.2251243781094528e-06,
      "loss": 0.9477,
      "step": 394
    },
    {
      "epoch": 0.036443316803136894,
      "grad_norm": 1.5556668203320005,
      "learning_rate": 1.2282338308457713e-06,
      "loss": 0.9539,
      "step": 395
    },
    {
      "epoch": 0.036443316803136894,
      "eval_GEN Loss": 0.4896683394908905,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8926553672316384,
      "eval_PRM F1 AUC": 0.6498166579360922,
      "eval_PRM F1 Neg": 0.45714285714285713,
      "eval_PRM Loss": 0.5040706396102905,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8404255319148937,
      "eval_PRM Recall": 0.9518072289156626,
      "eval_PRM Specificty": 0.34782608695652173,
      "eval_loss": 1.0819711685180664,
      "eval_runtime": 56.9164,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 395
    },
    {
      "epoch": 0.03653557836466382,
      "grad_norm": 2.2235428048145374,
      "learning_rate": 1.2313432835820897e-06,
      "loss": 1.1407,
      "step": 396
    },
    {
      "epoch": 0.03662783992619075,
      "grad_norm": 1.77863726581506,
      "learning_rate": 1.234452736318408e-06,
      "loss": 1.0715,
      "step": 397
    },
    {
      "epoch": 0.03672010148771768,
      "grad_norm": 1.4604250196141944,
      "learning_rate": 1.2375621890547266e-06,
      "loss": 0.9055,
      "step": 398
    },
    {
      "epoch": 0.03681236304924461,
      "grad_norm": 2.200986884891539,
      "learning_rate": 1.2406716417910449e-06,
      "loss": 1.2049,
      "step": 399
    },
    {
      "epoch": 0.036904624610771534,
      "grad_norm": 1.1855948742026567,
      "learning_rate": 1.2437810945273632e-06,
      "loss": 0.9039,
      "step": 400
    },
    {
      "epoch": 0.036904624610771534,
      "eval_GEN Loss": 0.4911164343357086,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8914285714285715,
      "eval_PRM F1 AUC": 0.6655316919853327,
      "eval_PRM F1 Neg": 0.4864864864864865,
      "eval_PRM Loss": 0.5084080100059509,
      "eval_PRM NPV": 0.6428571428571429,
      "eval_PRM Precision": 0.8478260869565217,
      "eval_PRM Recall": 0.9397590361445783,
      "eval_PRM Specificty": 0.391304347826087,
      "eval_loss": 1.0823317766189575,
      "eval_runtime": 56.9601,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 400
    },
    {
      "epoch": 0.03699688617229847,
      "grad_norm": 3.4674853596817603,
      "learning_rate": 1.2468905472636816e-06,
      "loss": 1.1359,
      "step": 401
    },
    {
      "epoch": 0.03708914773382539,
      "grad_norm": 2.0138986976213227,
      "learning_rate": 1.25e-06,
      "loss": 1.0893,
      "step": 402
    },
    {
      "epoch": 0.037181409295352325,
      "grad_norm": 1.7482583349421175,
      "learning_rate": 1.2499999716807582e-06,
      "loss": 0.9825,
      "step": 403
    },
    {
      "epoch": 0.03727367085687925,
      "grad_norm": 2.0783198737242836,
      "learning_rate": 1.249999886723035e-06,
      "loss": 0.9184,
      "step": 404
    },
    {
      "epoch": 0.03736593241840618,
      "grad_norm": 1.3697661624020034,
      "learning_rate": 1.2499997451268387e-06,
      "loss": 0.866,
      "step": 405
    },
    {
      "epoch": 0.03736593241840618,
      "eval_GEN Loss": 0.49147337675094604,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.872093023255814,
      "eval_PRM F1 AUC": 0.6474594028287062,
      "eval_PRM F1 Neg": 0.45,
      "eval_PRM Loss": 0.5320136547088623,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.8426966292134831,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.391304347826087,
      "eval_loss": 1.0904446840286255,
      "eval_runtime": 56.9151,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 405
    },
    {
      "epoch": 0.03745819397993311,
      "grad_norm": 2.125000262304555,
      "learning_rate": 1.2499995468921814e-06,
      "loss": 1.1024,
      "step": 406
    },
    {
      "epoch": 0.03755045554146004,
      "grad_norm": 3.4249056948683068,
      "learning_rate": 1.2499992920190814e-06,
      "loss": 0.9481,
      "step": 407
    },
    {
      "epoch": 0.037642717102986965,
      "grad_norm": 1.7365912485637587,
      "learning_rate": 1.2499989805075618e-06,
      "loss": 0.9993,
      "step": 408
    },
    {
      "epoch": 0.0377349786645139,
      "grad_norm": 1.8287690837900874,
      "learning_rate": 1.249998612357651e-06,
      "loss": 0.9735,
      "step": 409
    },
    {
      "epoch": 0.03782724022604082,
      "grad_norm": 2.443614782953873,
      "learning_rate": 1.249998187569382e-06,
      "loss": 0.9865,
      "step": 410
    },
    {
      "epoch": 0.03782724022604082,
      "eval_GEN Loss": 0.48897504806518555,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8850574712643678,
      "eval_PRM F1 AUC": 0.6595075955997904,
      "eval_PRM F1 Neg": 0.47368421052631576,
      "eval_PRM Loss": 0.5099429488182068,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8461538461538461,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.391304347826087,
      "eval_loss": 1.0731971263885498,
      "eval_runtime": 55.9536,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 410
    },
    {
      "epoch": 0.037919501787567755,
      "grad_norm": 3.4338451669618215,
      "learning_rate": 1.2499977061427935e-06,
      "loss": 1.1171,
      "step": 411
    },
    {
      "epoch": 0.03801176334909468,
      "grad_norm": 2.0289882614581076,
      "learning_rate": 1.2499971680779291e-06,
      "loss": 0.8247,
      "step": 412
    },
    {
      "epoch": 0.03810402491062161,
      "grad_norm": 2.298527207458775,
      "learning_rate": 1.2499965733748375e-06,
      "loss": 1.0174,
      "step": 413
    },
    {
      "epoch": 0.03819628647214854,
      "grad_norm": 1.3183253127786805,
      "learning_rate": 1.2499959220335727e-06,
      "loss": 0.8969,
      "step": 414
    },
    {
      "epoch": 0.03828854803367547,
      "grad_norm": 3.9704666165868616,
      "learning_rate": 1.249995214054194e-06,
      "loss": 1.1768,
      "step": 415
    },
    {
      "epoch": 0.03828854803367547,
      "eval_GEN Loss": 0.486381858587265,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8850574712643678,
      "eval_PRM F1 AUC": 0.6595075955997904,
      "eval_PRM F1 Neg": 0.47368421052631576,
      "eval_PRM Loss": 0.49420467019081116,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8461538461538461,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.391304347826087,
      "eval_loss": 1.0578124523162842,
      "eval_runtime": 56.0141,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 415
    },
    {
      "epoch": 0.038380809595202396,
      "grad_norm": 1.6175496053472633,
      "learning_rate": 1.2499944494367649e-06,
      "loss": 0.9617,
      "step": 416
    },
    {
      "epoch": 0.03847307115672933,
      "grad_norm": 2.8215401472618398,
      "learning_rate": 1.2499936281813551e-06,
      "loss": 0.888,
      "step": 417
    },
    {
      "epoch": 0.038565332718256254,
      "grad_norm": 3.067168938139133,
      "learning_rate": 1.2499927502880391e-06,
      "loss": 0.9159,
      "step": 418
    },
    {
      "epoch": 0.038657594279783186,
      "grad_norm": 1.4636998556443914,
      "learning_rate": 1.249991815756896e-06,
      "loss": 0.9951,
      "step": 419
    },
    {
      "epoch": 0.03874985584131011,
      "grad_norm": 2.244478144941869,
      "learning_rate": 1.2499908245880114e-06,
      "loss": 1.071,
      "step": 420
    },
    {
      "epoch": 0.03874985584131011,
      "eval_GEN Loss": 0.4856526851654053,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8850574712643678,
      "eval_PRM F1 AUC": 0.6595075955997904,
      "eval_PRM F1 Neg": 0.47368421052631576,
      "eval_PRM Loss": 0.48444485664367676,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.8461538461538461,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.391304347826087,
      "eval_loss": 1.0461539030075073,
      "eval_runtime": 56.0607,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 420
    },
    {
      "epoch": 0.038842117402837044,
      "grad_norm": 1.986588212280402,
      "learning_rate": 1.249989776781474e-06,
      "loss": 0.9483,
      "step": 421
    },
    {
      "epoch": 0.03893437896436397,
      "grad_norm": 1.2109181066291579,
      "learning_rate": 1.2499886723373795e-06,
      "loss": 0.8441,
      "step": 422
    },
    {
      "epoch": 0.0390266405258909,
      "grad_norm": 1.7634910222087619,
      "learning_rate": 1.2499875112558277e-06,
      "loss": 0.9839,
      "step": 423
    },
    {
      "epoch": 0.03911890208741783,
      "grad_norm": 1.7202561747692766,
      "learning_rate": 1.2499862935369238e-06,
      "loss": 1.0435,
      "step": 424
    },
    {
      "epoch": 0.03921116364894476,
      "grad_norm": 1.394069525167447,
      "learning_rate": 1.2499850191807783e-06,
      "loss": 0.9949,
      "step": 425
    },
    {
      "epoch": 0.03921116364894476,
      "eval_GEN Loss": 0.48652511835098267,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8786127167630058,
      "eval_PRM F1 AUC": 0.6534834992142482,
      "eval_PRM F1 Neg": 0.46153846153846156,
      "eval_PRM Loss": 0.4757426679134369,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.8444444444444444,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.391304347826087,
      "eval_loss": 1.0393028259277344,
      "eval_runtime": 56.0661,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 425
    },
    {
      "epoch": 0.039303425210471685,
      "grad_norm": 1.7432266075385632,
      "learning_rate": 1.2499836881875066e-06,
      "loss": 0.9469,
      "step": 426
    },
    {
      "epoch": 0.03939568677199862,
      "grad_norm": 1.7296291983439422,
      "learning_rate": 1.2499823005572294e-06,
      "loss": 0.9209,
      "step": 427
    },
    {
      "epoch": 0.03948794833352554,
      "grad_norm": 2.2430648621312077,
      "learning_rate": 1.2499808562900723e-06,
      "loss": 1.1083,
      "step": 428
    },
    {
      "epoch": 0.039580209895052475,
      "grad_norm": 2.4839328433585277,
      "learning_rate": 1.2499793553861664e-06,
      "loss": 0.8595,
      "step": 429
    },
    {
      "epoch": 0.0396724714565794,
      "grad_norm": 1.5598120753532012,
      "learning_rate": 1.2499777978456475e-06,
      "loss": 0.9752,
      "step": 430
    },
    {
      "epoch": 0.0396724714565794,
      "eval_GEN Loss": 0.48572012782096863,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.6511262441068623,
      "eval_PRM F1 Neg": 0.45454545454545453,
      "eval_PRM Loss": 0.4870435893535614,
      "eval_PRM NPV": 0.47619047619047616,
      "eval_PRM Precision": 0.8470588235294118,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.43478260869565216,
      "eval_loss": 1.043870210647583,
      "eval_runtime": 56.0713,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 430
    },
    {
      "epoch": 0.03976473301810633,
      "grad_norm": 2.1524630412012176,
      "learning_rate": 1.2499761836686567e-06,
      "loss": 1.2483,
      "step": 431
    },
    {
      "epoch": 0.03985699457963326,
      "grad_norm": 1.684216300151307,
      "learning_rate": 1.2499745128553406e-06,
      "loss": 0.9628,
      "step": 432
    },
    {
      "epoch": 0.03994925614116019,
      "grad_norm": 1.8621080525821148,
      "learning_rate": 1.2499727854058505e-06,
      "loss": 1.1252,
      "step": 433
    },
    {
      "epoch": 0.040041517702687116,
      "grad_norm": 1.5520774876746433,
      "learning_rate": 1.2499710013203426e-06,
      "loss": 1.1255,
      "step": 434
    },
    {
      "epoch": 0.04013377926421405,
      "grad_norm": 2.613192298018756,
      "learning_rate": 1.249969160598979e-06,
      "loss": 1.0245,
      "step": 435
    },
    {
      "epoch": 0.04013377926421405,
      "eval_GEN Loss": 0.48517999053001404,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.84472049689441,
      "eval_PRM F1 AUC": 0.6922472498690415,
      "eval_PRM F1 Neg": 0.5098039215686274,
      "eval_PRM Loss": 0.5124436616897583,
      "eval_PRM NPV": 0.4642857142857143,
      "eval_PRM Precision": 0.8717948717948718,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 1.0579928159713745,
      "eval_runtime": 55.9343,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 435
    },
    {
      "epoch": 0.04022604082574097,
      "grad_norm": 1.5090261556949027,
      "learning_rate": 1.2499672632419262e-06,
      "loss": 0.9561,
      "step": 436
    },
    {
      "epoch": 0.040318302387267906,
      "grad_norm": 1.7390946058193508,
      "learning_rate": 1.2499653092493567e-06,
      "loss": 0.9918,
      "step": 437
    },
    {
      "epoch": 0.04041056394879483,
      "grad_norm": 2.765509328605111,
      "learning_rate": 1.2499632986214468e-06,
      "loss": 0.9601,
      "step": 438
    },
    {
      "epoch": 0.04050282551032176,
      "grad_norm": 1.5586441018546473,
      "learning_rate": 1.249961231358379e-06,
      "loss": 0.9044,
      "step": 439
    },
    {
      "epoch": 0.04059508707184869,
      "grad_norm": 2.2748830104553357,
      "learning_rate": 1.249959107460341e-06,
      "loss": 0.9474,
      "step": 440
    },
    {
      "epoch": 0.04059508707184869,
      "eval_GEN Loss": 0.48855283856391907,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.47612351179122925,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 1.0473557710647583,
      "eval_runtime": 56.2396,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 440
    },
    {
      "epoch": 0.04068734863337562,
      "grad_norm": 1.8201253581714998,
      "learning_rate": 1.249956926927525e-06,
      "loss": 0.8294,
      "step": 441
    },
    {
      "epoch": 0.040779610194902546,
      "grad_norm": 1.9508496557784138,
      "learning_rate": 1.2499546897601284e-06,
      "loss": 0.8903,
      "step": 442
    },
    {
      "epoch": 0.04087187175642948,
      "grad_norm": 2.3625577338752297,
      "learning_rate": 1.249952395958354e-06,
      "loss": 0.7296,
      "step": 443
    },
    {
      "epoch": 0.040964133317956404,
      "grad_norm": 1.6663025871239967,
      "learning_rate": 1.24995004552241e-06,
      "loss": 0.9373,
      "step": 444
    },
    {
      "epoch": 0.041056394879483336,
      "grad_norm": 1.6603489043388613,
      "learning_rate": 1.2499476384525092e-06,
      "loss": 0.9188,
      "step": 445
    },
    {
      "epoch": 0.041056394879483336,
      "eval_GEN Loss": 0.4943142831325531,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8863636363636364,
      "eval_PRM F1 AUC": 0.6437925615505501,
      "eval_PRM F1 Neg": 0.4444444444444444,
      "eval_PRM Loss": 0.43712368607521057,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.8387096774193549,
      "eval_PRM Recall": 0.9397590361445783,
      "eval_PRM Specificty": 0.34782608695652173,
      "eval_loss": 1.0341947078704834,
      "eval_runtime": 56.2179,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 445
    },
    {
      "epoch": 0.04114865644101026,
      "grad_norm": 2.9126230204462487,
      "learning_rate": 1.2499451747488695e-06,
      "loss": 1.0148,
      "step": 446
    },
    {
      "epoch": 0.041240918002537194,
      "grad_norm": 1.6767556924452518,
      "learning_rate": 1.2499426544117146e-06,
      "loss": 1.0057,
      "step": 447
    },
    {
      "epoch": 0.04133317956406412,
      "grad_norm": 1.2491079330246846,
      "learning_rate": 1.2499400774412725e-06,
      "loss": 0.9206,
      "step": 448
    },
    {
      "epoch": 0.04142544112559105,
      "grad_norm": 1.6211871447944266,
      "learning_rate": 1.249937443837777e-06,
      "loss": 0.8217,
      "step": 449
    },
    {
      "epoch": 0.04151770268711798,
      "grad_norm": 1.5443566788338265,
      "learning_rate": 1.2499347536014667e-06,
      "loss": 0.8358,
      "step": 450
    },
    {
      "epoch": 0.04151770268711798,
      "eval_GEN Loss": 0.48962274193763733,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8875739644970414,
      "eval_PRM F1 AUC": 0.7126767941330541,
      "eval_PRM F1 Neg": 0.5581395348837209,
      "eval_PRM Loss": 0.4641878008842468,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.872093023255814,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 1.0381009578704834,
      "eval_runtime": 57.0011,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 450
    },
    {
      "epoch": 0.04160996424864491,
      "grad_norm": 1.4733450427095798,
      "learning_rate": 1.2499320067325854e-06,
      "loss": 0.9267,
      "step": 451
    },
    {
      "epoch": 0.041702225810171835,
      "grad_norm": 2.9303870595244903,
      "learning_rate": 1.249929203231382e-06,
      "loss": 0.8684,
      "step": 452
    },
    {
      "epoch": 0.04179448737169877,
      "grad_norm": 1.785876435180517,
      "learning_rate": 1.2499263430981106e-06,
      "loss": 0.9132,
      "step": 453
    },
    {
      "epoch": 0.04188674893322569,
      "grad_norm": 2.148315533111999,
      "learning_rate": 1.2499234263330302e-06,
      "loss": 0.8028,
      "step": 454
    },
    {
      "epoch": 0.041979010494752625,
      "grad_norm": 2.0582297728234127,
      "learning_rate": 1.2499204529364053e-06,
      "loss": 0.7296,
      "step": 455
    },
    {
      "epoch": 0.041979010494752625,
      "eval_GEN Loss": 0.48988714814186096,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8764044943820225,
      "eval_PRM F1 AUC": 0.6003143006809848,
      "eval_PRM F1 Neg": 0.35294117647058826,
      "eval_PRM Loss": 0.45193052291870117,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.8210526315789474,
      "eval_PRM Recall": 0.9397590361445783,
      "eval_PRM Specificty": 0.2608695652173913,
      "eval_loss": 1.038401484489441,
      "eval_runtime": 56.9886,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 455
    },
    {
      "epoch": 0.04207127205627955,
      "grad_norm": 1.9190244382111306,
      "learning_rate": 1.2499174229085053e-06,
      "loss": 1.0398,
      "step": 456
    },
    {
      "epoch": 0.04216353361780648,
      "grad_norm": 1.8902843000632146,
      "learning_rate": 1.249914336249605e-06,
      "loss": 0.951,
      "step": 457
    },
    {
      "epoch": 0.04225579517933341,
      "grad_norm": 1.8539637597556553,
      "learning_rate": 1.2499111929599836e-06,
      "loss": 0.921,
      "step": 458
    },
    {
      "epoch": 0.04234805674086034,
      "grad_norm": 2.476016793580202,
      "learning_rate": 1.2499079930399266e-06,
      "loss": 0.8745,
      "step": 459
    },
    {
      "epoch": 0.042440318302387266,
      "grad_norm": 1.8361355698224726,
      "learning_rate": 1.2499047364897234e-06,
      "loss": 0.7503,
      "step": 460
    },
    {
      "epoch": 0.042440318302387266,
      "eval_GEN Loss": 0.4883134067058563,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8636363636363636,
      "eval_PRM F1 AUC": 0.5882661079099004,
      "eval_PRM F1 Neg": 0.3333333333333333,
      "eval_PRM Loss": 0.46063748002052307,
      "eval_PRM NPV": 0.46153846153846156,
      "eval_PRM Precision": 0.8172043010752689,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.2608695652173913,
      "eval_loss": 1.0429086685180664,
      "eval_runtime": 56.9469,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 460
    },
    {
      "epoch": 0.0425325798639142,
      "grad_norm": 2.4697307017515753,
      "learning_rate": 1.2499014233096694e-06,
      "loss": 0.8739,
      "step": 461
    },
    {
      "epoch": 0.04262484142544112,
      "grad_norm": 2.1777696172759966,
      "learning_rate": 1.2498980535000649e-06,
      "loss": 1.0281,
      "step": 462
    },
    {
      "epoch": 0.042717102986968056,
      "grad_norm": 1.9585664156482705,
      "learning_rate": 1.249894627061215e-06,
      "loss": 0.8492,
      "step": 463
    },
    {
      "epoch": 0.04280936454849498,
      "grad_norm": 1.8737817400264731,
      "learning_rate": 1.2498911439934306e-06,
      "loss": 0.934,
      "step": 464
    },
    {
      "epoch": 0.04290162611002191,
      "grad_norm": 1.6298637608830269,
      "learning_rate": 1.249887604297027e-06,
      "loss": 0.8942,
      "step": 465
    },
    {
      "epoch": 0.04290162611002191,
      "eval_GEN Loss": 0.49040842056274414,
      "eval_GEN top-5 accuracy": 0.9758877685225779,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8681318681318682,
      "eval_PRM F1 AUC": 0.5411210057621791,
      "eval_PRM F1 Neg": 0.2,
      "eval_PRM Loss": 0.4450353980064392,
      "eval_PRM NPV": 0.42857142857142855,
      "eval_PRM Precision": 0.797979797979798,
      "eval_PRM Recall": 0.9518072289156626,
      "eval_PRM Specificty": 0.13043478260869565,
      "eval_loss": 1.0331730842590332,
      "eval_runtime": 56.9373,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 465
    },
    {
      "epoch": 0.04299388767154884,
      "grad_norm": 3.533649562896908,
      "learning_rate": 1.2498840079723252e-06,
      "loss": 0.8994,
      "step": 466
    },
    {
      "epoch": 0.04308614923307577,
      "grad_norm": 1.8933966875151385,
      "learning_rate": 1.249880355019651e-06,
      "loss": 0.9002,
      "step": 467
    },
    {
      "epoch": 0.0431784107946027,
      "grad_norm": 1.6618181663533436,
      "learning_rate": 1.2498766454393354e-06,
      "loss": 0.8512,
      "step": 468
    },
    {
      "epoch": 0.04327067235612963,
      "grad_norm": 2.2357348374085952,
      "learning_rate": 1.2498728792317146e-06,
      "loss": 0.8073,
      "step": 469
    },
    {
      "epoch": 0.043362933917656554,
      "grad_norm": 1.9890022894412644,
      "learning_rate": 1.2498690563971302e-06,
      "loss": 0.6942,
      "step": 470
    },
    {
      "epoch": 0.043362933917656554,
      "eval_GEN Loss": 0.49270206689834595,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.474351167678833,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 1.0350961685180664,
      "eval_runtime": 56.9081,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 470
    },
    {
      "epoch": 0.04345519547918349,
      "grad_norm": 1.5430350756935,
      "learning_rate": 1.2498651769359282e-06,
      "loss": 0.9192,
      "step": 471
    },
    {
      "epoch": 0.04354745704071041,
      "grad_norm": 1.903783246574929,
      "learning_rate": 1.24986124084846e-06,
      "loss": 0.7874,
      "step": 472
    },
    {
      "epoch": 0.043639718602237344,
      "grad_norm": 2.5498924226341617,
      "learning_rate": 1.249857248135083e-06,
      "loss": 1.0717,
      "step": 473
    },
    {
      "epoch": 0.04373198016376427,
      "grad_norm": 1.7294727852998102,
      "learning_rate": 1.2498531987961584e-06,
      "loss": 0.9471,
      "step": 474
    },
    {
      "epoch": 0.0438242417252912,
      "grad_norm": 2.1361973305253974,
      "learning_rate": 1.2498490928320537e-06,
      "loss": 0.9701,
      "step": 475
    },
    {
      "epoch": 0.0438242417252912,
      "eval_GEN Loss": 0.4921441972255707,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.437370240688324,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9977163672447205,
      "eval_runtime": 56.8242,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 475
    },
    {
      "epoch": 0.04391650328681813,
      "grad_norm": 1.5290437108743213,
      "learning_rate": 1.2498449302431404e-06,
      "loss": 0.9156,
      "step": 476
    },
    {
      "epoch": 0.04400876484834506,
      "grad_norm": 3.9898809086088347,
      "learning_rate": 1.2498407110297961e-06,
      "loss": 1.045,
      "step": 477
    },
    {
      "epoch": 0.044101026409871985,
      "grad_norm": 1.713146486887783,
      "learning_rate": 1.249836435192403e-06,
      "loss": 1.0169,
      "step": 478
    },
    {
      "epoch": 0.04419328797139892,
      "grad_norm": 1.7998586465129913,
      "learning_rate": 1.2498321027313486e-06,
      "loss": 0.8223,
      "step": 479
    },
    {
      "epoch": 0.04428554953292584,
      "grad_norm": 2.3953689065770876,
      "learning_rate": 1.2498277136470255e-06,
      "loss": 0.7211,
      "step": 480
    },
    {
      "epoch": 0.04428554953292584,
      "eval_GEN Loss": 0.4898214042186737,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.4590703248977661,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9987980723381042,
      "eval_runtime": 56.7968,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 480
    },
    {
      "epoch": 0.044377811094452775,
      "grad_norm": 2.392990239491769,
      "learning_rate": 1.2498232679398317e-06,
      "loss": 0.8219,
      "step": 481
    },
    {
      "epoch": 0.0444700726559797,
      "grad_norm": 1.6793203035636983,
      "learning_rate": 1.2498187656101698e-06,
      "loss": 0.8014,
      "step": 482
    },
    {
      "epoch": 0.04456233421750663,
      "grad_norm": 2.296906461119564,
      "learning_rate": 1.2498142066584478e-06,
      "loss": 0.8739,
      "step": 483
    },
    {
      "epoch": 0.04465459577903356,
      "grad_norm": 2.466338328508299,
      "learning_rate": 1.249809591085079e-06,
      "loss": 0.8223,
      "step": 484
    },
    {
      "epoch": 0.04474685734056049,
      "grad_norm": 2.2045465950681216,
      "learning_rate": 1.2498049188904814e-06,
      "loss": 0.873,
      "step": 485
    },
    {
      "epoch": 0.04474685734056049,
      "eval_GEN Loss": 0.490390807390213,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4337656497955322,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9806490540504456,
      "eval_runtime": 56.8854,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 485
    },
    {
      "epoch": 0.044839118902087416,
      "grad_norm": 1.8767930377025188,
      "learning_rate": 1.2498001900750788e-06,
      "loss": 0.7161,
      "step": 486
    },
    {
      "epoch": 0.04493138046361435,
      "grad_norm": 1.5220048281800544,
      "learning_rate": 1.2497954046392995e-06,
      "loss": 0.9338,
      "step": 487
    },
    {
      "epoch": 0.045023642025141274,
      "grad_norm": 2.5084112217548404,
      "learning_rate": 1.2497905625835773e-06,
      "loss": 0.9706,
      "step": 488
    },
    {
      "epoch": 0.045115903586668206,
      "grad_norm": 1.8840675032339782,
      "learning_rate": 1.2497856639083508e-06,
      "loss": 0.7622,
      "step": 489
    },
    {
      "epoch": 0.04520816514819513,
      "grad_norm": 1.6871894439217352,
      "learning_rate": 1.249780708614064e-06,
      "loss": 0.8241,
      "step": 490
    },
    {
      "epoch": 0.04520816514819513,
      "eval_GEN Loss": 0.4915316104888916,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.41830411553382874,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9688100814819336,
      "eval_runtime": 57.0794,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 490
    },
    {
      "epoch": 0.045300426709722064,
      "grad_norm": 1.7444544329434182,
      "learning_rate": 1.2497756967011658e-06,
      "loss": 0.7967,
      "step": 491
    },
    {
      "epoch": 0.04539268827124899,
      "grad_norm": 2.95337755170944,
      "learning_rate": 1.2497706281701107e-06,
      "loss": 1.0388,
      "step": 492
    },
    {
      "epoch": 0.04548494983277592,
      "grad_norm": 1.9349385229854092,
      "learning_rate": 1.249765503021358e-06,
      "loss": 0.8927,
      "step": 493
    },
    {
      "epoch": 0.04557721139430285,
      "grad_norm": 1.6823368055068961,
      "learning_rate": 1.2497603212553718e-06,
      "loss": 0.8645,
      "step": 494
    },
    {
      "epoch": 0.04566947295582978,
      "grad_norm": 1.1607320411290694,
      "learning_rate": 1.249755082872622e-06,
      "loss": 0.8193,
      "step": 495
    },
    {
      "epoch": 0.04566947295582978,
      "eval_GEN Loss": 0.48993173241615295,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4339875280857086,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9669471383094788,
      "eval_runtime": 57.1224,
      "eval_samples_per_second": 1.138,
      "eval_steps_per_second": 0.088,
      "step": 495
    },
    {
      "epoch": 0.045761734517356704,
      "grad_norm": 2.5211869469282977,
      "learning_rate": 1.2497497878735831e-06,
      "loss": 0.9542,
      "step": 496
    },
    {
      "epoch": 0.04585399607888364,
      "grad_norm": 1.8091471907694843,
      "learning_rate": 1.2497444362587352e-06,
      "loss": 0.7601,
      "step": 497
    },
    {
      "epoch": 0.04594625764041056,
      "grad_norm": 1.7687992668461012,
      "learning_rate": 1.249739028028563e-06,
      "loss": 0.7659,
      "step": 498
    },
    {
      "epoch": 0.046038519201937494,
      "grad_norm": 1.4396309371176,
      "learning_rate": 1.2497335631835569e-06,
      "loss": 0.878,
      "step": 499
    },
    {
      "epoch": 0.04613078076346442,
      "grad_norm": 1.9668782595980827,
      "learning_rate": 1.2497280417242117e-06,
      "loss": 0.8463,
      "step": 500
    },
    {
      "epoch": 0.04613078076346442,
      "eval_GEN Loss": 0.49217748641967773,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4544593095779419,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9750000238418579,
      "eval_runtime": 56.8259,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 500
    },
    {
      "epoch": 0.04622304232499135,
      "grad_norm": 2.6448659618140278,
      "learning_rate": 1.2497224636510282e-06,
      "loss": 0.9017,
      "step": 501
    },
    {
      "epoch": 0.04631530388651828,
      "grad_norm": 2.865230099490025,
      "learning_rate": 1.2497168289645117e-06,
      "loss": 0.5919,
      "step": 502
    },
    {
      "epoch": 0.04640756544804521,
      "grad_norm": 3.5435121302194768,
      "learning_rate": 1.2497111376651728e-06,
      "loss": 1.0819,
      "step": 503
    },
    {
      "epoch": 0.046499827009572135,
      "grad_norm": 2.1265478178651334,
      "learning_rate": 1.2497053897535275e-06,
      "loss": 0.9577,
      "step": 504
    },
    {
      "epoch": 0.04659208857109907,
      "grad_norm": 2.3646329293378567,
      "learning_rate": 1.249699585230096e-06,
      "loss": 0.7973,
      "step": 505
    },
    {
      "epoch": 0.04659208857109907,
      "eval_GEN Loss": 0.49786898493766785,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.4246281087398529,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9762019515037537,
      "eval_runtime": 55.8521,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 505
    },
    {
      "epoch": 0.04668435013262599,
      "grad_norm": 1.7789669025076071,
      "learning_rate": 1.2496937240954053e-06,
      "loss": 0.9174,
      "step": 506
    },
    {
      "epoch": 0.046776611694152925,
      "grad_norm": 1.4940974641561018,
      "learning_rate": 1.2496878063499859e-06,
      "loss": 0.7763,
      "step": 507
    },
    {
      "epoch": 0.04686887325567985,
      "grad_norm": 2.240464981910987,
      "learning_rate": 1.249681831994374e-06,
      "loss": 0.9162,
      "step": 508
    },
    {
      "epoch": 0.04696113481720678,
      "grad_norm": 1.7331241048782036,
      "learning_rate": 1.2496758010291115e-06,
      "loss": 0.7101,
      "step": 509
    },
    {
      "epoch": 0.04705339637873371,
      "grad_norm": 2.1776111323371206,
      "learning_rate": 1.2496697134547445e-06,
      "loss": 0.9251,
      "step": 510
    },
    {
      "epoch": 0.04705339637873371,
      "eval_GEN Loss": 0.5021944642066956,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.42142724990844727,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9699519276618958,
      "eval_runtime": 55.9135,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 510
    },
    {
      "epoch": 0.04714565794026064,
      "grad_norm": 1.8149877200200812,
      "learning_rate": 1.249663569271825e-06,
      "loss": 0.8276,
      "step": 511
    },
    {
      "epoch": 0.047237919501787566,
      "grad_norm": 2.311685094068292,
      "learning_rate": 1.2496573684809093e-06,
      "loss": 1.0577,
      "step": 512
    },
    {
      "epoch": 0.0473301810633145,
      "grad_norm": 1.8430300506162633,
      "learning_rate": 1.24965111108256e-06,
      "loss": 1.0098,
      "step": 513
    },
    {
      "epoch": 0.047422442624841424,
      "grad_norm": 1.7150755345180988,
      "learning_rate": 1.2496447970773435e-06,
      "loss": 0.8558,
      "step": 514
    },
    {
      "epoch": 0.047514704186368356,
      "grad_norm": 2.322171852550018,
      "learning_rate": 1.2496384264658322e-06,
      "loss": 1.1605,
      "step": 515
    },
    {
      "epoch": 0.047514704186368356,
      "eval_GEN Loss": 0.5072044730186462,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7169811320754716,
      "eval_PRM F1": 0.7945205479452054,
      "eval_PRM F1 AUC": 0.7407019381875328,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.5061312317848206,
      "eval_PRM NPV": 0.4186046511627907,
      "eval_PRM Precision": 0.9206349206349206,
      "eval_PRM Recall": 0.6987951807228916,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 1.0418870449066162,
      "eval_runtime": 56.1244,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 515
    },
    {
      "epoch": 0.04760696574789528,
      "grad_norm": 2.376452841356278,
      "learning_rate": 1.249631999248604e-06,
      "loss": 0.8696,
      "step": 516
    },
    {
      "epoch": 0.047699227309422214,
      "grad_norm": 2.1697200720346226,
      "learning_rate": 1.2496255154262405e-06,
      "loss": 0.7282,
      "step": 517
    },
    {
      "epoch": 0.04779148887094914,
      "grad_norm": 2.231618573506481,
      "learning_rate": 1.2496189749993296e-06,
      "loss": 1.0144,
      "step": 518
    },
    {
      "epoch": 0.04788375043247607,
      "grad_norm": 1.4477856564989626,
      "learning_rate": 1.249612377968464e-06,
      "loss": 0.7972,
      "step": 519
    },
    {
      "epoch": 0.047976011994003,
      "grad_norm": 2.522904308533631,
      "learning_rate": 1.2496057243342416e-06,
      "loss": 0.9651,
      "step": 520
    },
    {
      "epoch": 0.047976011994003,
      "eval_GEN Loss": 0.5013367533683777,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.44258493185043335,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9829326868057251,
      "eval_runtime": 56.1787,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 520
    },
    {
      "epoch": 0.04806827355552993,
      "grad_norm": 1.3827215268108803,
      "learning_rate": 1.2495990140972653e-06,
      "loss": 0.9284,
      "step": 521
    },
    {
      "epoch": 0.048160535117056855,
      "grad_norm": 1.9755875110206902,
      "learning_rate": 1.2495922472581435e-06,
      "loss": 0.8925,
      "step": 522
    },
    {
      "epoch": 0.04825279667858379,
      "grad_norm": 2.0814499114539373,
      "learning_rate": 1.2495854238174887e-06,
      "loss": 1.1083,
      "step": 523
    },
    {
      "epoch": 0.04834505824011071,
      "grad_norm": 2.5417062148450147,
      "learning_rate": 1.24957854377592e-06,
      "loss": 1.0285,
      "step": 524
    },
    {
      "epoch": 0.048437319801637645,
      "grad_norm": 1.4908946062519384,
      "learning_rate": 1.2495716071340602e-06,
      "loss": 0.8482,
      "step": 525
    },
    {
      "epoch": 0.048437319801637645,
      "eval_GEN Loss": 0.4994891285896301,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.44036853313446045,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9760817289352417,
      "eval_runtime": 55.9269,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 525
    },
    {
      "epoch": 0.04852958136316457,
      "grad_norm": 1.5067261309429942,
      "learning_rate": 1.2495646138925384e-06,
      "loss": 0.863,
      "step": 526
    },
    {
      "epoch": 0.0486218429246915,
      "grad_norm": 1.7909389832906948,
      "learning_rate": 1.2495575640519886e-06,
      "loss": 0.8381,
      "step": 527
    },
    {
      "epoch": 0.04871410448621843,
      "grad_norm": 1.603287958124817,
      "learning_rate": 1.2495504576130488e-06,
      "loss": 0.8497,
      "step": 528
    },
    {
      "epoch": 0.04880636604774536,
      "grad_norm": 2.4102436997096057,
      "learning_rate": 1.2495432945763638e-06,
      "loss": 0.8235,
      "step": 529
    },
    {
      "epoch": 0.048898627609272285,
      "grad_norm": 1.6308890605210715,
      "learning_rate": 1.2495360749425821e-06,
      "loss": 0.9034,
      "step": 530
    },
    {
      "epoch": 0.048898627609272285,
      "eval_GEN Loss": 0.49671459197998047,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8795180722891566,
      "eval_PRM F1 AUC": 0.7223677317967523,
      "eval_PRM F1 Neg": 0.5652173913043478,
      "eval_PRM Loss": 0.4215235710144043,
      "eval_PRM NPV": 0.5652173913043478,
      "eval_PRM Precision": 0.8795180722891566,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9649639129638672,
      "eval_runtime": 55.754,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 530
    },
    {
      "epoch": 0.04899088917079922,
      "grad_norm": 1.6207713329635076,
      "learning_rate": 1.2495287987123587e-06,
      "loss": 0.9246,
      "step": 531
    },
    {
      "epoch": 0.04908315073232614,
      "grad_norm": 2.137962559846118,
      "learning_rate": 1.2495214658863521e-06,
      "loss": 1.0508,
      "step": 532
    },
    {
      "epoch": 0.049175412293853075,
      "grad_norm": 3.0740623866229284,
      "learning_rate": 1.2495140764652274e-06,
      "loss": 0.9697,
      "step": 533
    },
    {
      "epoch": 0.04926767385538,
      "grad_norm": 1.915824797340991,
      "learning_rate": 1.2495066304496542e-06,
      "loss": 0.7938,
      "step": 534
    },
    {
      "epoch": 0.04935993541690693,
      "grad_norm": 2.0017572967949744,
      "learning_rate": 1.2494991278403068e-06,
      "loss": 0.8639,
      "step": 535
    },
    {
      "epoch": 0.04935993541690693,
      "eval_GEN Loss": 0.49528592824935913,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.874251497005988,
      "eval_PRM F1 AUC": 0.7006286013619696,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM Loss": 0.4145190715789795,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.8690476190476191,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9653846025466919,
      "eval_runtime": 55.8675,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 535
    },
    {
      "epoch": 0.04945219697843386,
      "grad_norm": 1.7101567064092154,
      "learning_rate": 1.2494915686378658e-06,
      "loss": 0.7717,
      "step": 536
    },
    {
      "epoch": 0.04954445853996079,
      "grad_norm": 1.8633417800502488,
      "learning_rate": 1.2494839528430158e-06,
      "loss": 0.9391,
      "step": 537
    },
    {
      "epoch": 0.049636720101487716,
      "grad_norm": 2.255765287910322,
      "learning_rate": 1.2494762804564472e-06,
      "loss": 0.9261,
      "step": 538
    },
    {
      "epoch": 0.04972898166301465,
      "grad_norm": 2.097067196958236,
      "learning_rate": 1.2494685514788548e-06,
      "loss": 0.8661,
      "step": 539
    },
    {
      "epoch": 0.049821243224541574,
      "grad_norm": 1.8120378173288358,
      "learning_rate": 1.2494607659109396e-06,
      "loss": 0.9078,
      "step": 540
    },
    {
      "epoch": 0.049821243224541574,
      "eval_GEN Loss": 0.500647246837616,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM Loss": 0.44509997963905334,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9778845906257629,
      "eval_runtime": 55.9283,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 540
    },
    {
      "epoch": 0.049913504786068506,
      "grad_norm": 1.9209686059123097,
      "learning_rate": 1.2494529237534067e-06,
      "loss": 0.9968,
      "step": 541
    },
    {
      "epoch": 0.05000576634759543,
      "grad_norm": 2.791385333364772,
      "learning_rate": 1.249445025006967e-06,
      "loss": 0.7469,
      "step": 542
    },
    {
      "epoch": 0.050098027909122364,
      "grad_norm": 2.070234368370357,
      "learning_rate": 1.2494370696723362e-06,
      "loss": 0.9128,
      "step": 543
    },
    {
      "epoch": 0.05019028947064929,
      "grad_norm": 1.8867484348866532,
      "learning_rate": 1.2494290577502354e-06,
      "loss": 0.8872,
      "step": 544
    },
    {
      "epoch": 0.05028255103217622,
      "grad_norm": 2.5572474728796784,
      "learning_rate": 1.2494209892413905e-06,
      "loss": 0.8109,
      "step": 545
    },
    {
      "epoch": 0.05028255103217622,
      "eval_GEN Loss": 0.5057682394981384,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8627450980392157,
      "eval_PRM F1 AUC": 0.8106338397066527,
      "eval_PRM F1 Neg": 0.6440677966101694,
      "eval_PRM Loss": 0.4570285379886627,
      "eval_PRM NPV": 0.5277777777777778,
      "eval_PRM Precision": 0.9428571428571428,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.9915865659713745,
      "eval_runtime": 55.772,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 545
    },
    {
      "epoch": 0.05037481259370315,
      "grad_norm": 1.5454658736067963,
      "learning_rate": 1.2494128641465328e-06,
      "loss": 0.9491,
      "step": 546
    },
    {
      "epoch": 0.05046707415523008,
      "grad_norm": 2.0036633039639886,
      "learning_rate": 1.2494046824663983e-06,
      "loss": 0.8376,
      "step": 547
    },
    {
      "epoch": 0.050559335716757005,
      "grad_norm": 1.4598488584859581,
      "learning_rate": 1.2493964442017286e-06,
      "loss": 0.8602,
      "step": 548
    },
    {
      "epoch": 0.05065159727828394,
      "grad_norm": 2.1035556200520196,
      "learning_rate": 1.2493881493532705e-06,
      "loss": 0.9633,
      "step": 549
    },
    {
      "epoch": 0.05074385883981086,
      "grad_norm": 2.1527492930053396,
      "learning_rate": 1.2493797979217755e-06,
      "loss": 0.8665,
      "step": 550
    },
    {
      "epoch": 0.05074385883981086,
      "eval_GEN Loss": 0.5002301931381226,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.4020197093486786,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9468749761581421,
      "eval_runtime": 56.4,
      "eval_samples_per_second": 1.152,
      "eval_steps_per_second": 0.089,
      "step": 550
    },
    {
      "epoch": 0.050836120401337795,
      "grad_norm": 1.833646828545211,
      "learning_rate": 1.2493713899080004e-06,
      "loss": 0.821,
      "step": 551
    },
    {
      "epoch": 0.05092838196286472,
      "grad_norm": 1.30148826318895,
      "learning_rate": 1.2493629253127071e-06,
      "loss": 0.8375,
      "step": 552
    },
    {
      "epoch": 0.05102064352439165,
      "grad_norm": 1.4962814588034477,
      "learning_rate": 1.2493544041366627e-06,
      "loss": 0.868,
      "step": 553
    },
    {
      "epoch": 0.05111290508591858,
      "grad_norm": 1.1764107105303518,
      "learning_rate": 1.2493458263806397e-06,
      "loss": 0.8314,
      "step": 554
    },
    {
      "epoch": 0.05120516664744551,
      "grad_norm": 1.8363744781281612,
      "learning_rate": 1.2493371920454148e-06,
      "loss": 0.7943,
      "step": 555
    },
    {
      "epoch": 0.05120516664744551,
      "eval_GEN Loss": 0.5007193684577942,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8674698795180723,
      "eval_PRM F1 AUC": 0.6946045049764275,
      "eval_PRM F1 Neg": 0.5217391304347826,
      "eval_PRM Loss": 0.4065876603126526,
      "eval_PRM NPV": 0.5217391304347826,
      "eval_PRM Precision": 0.8674698795180723,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9555288553237915,
      "eval_runtime": 55.9094,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 555
    },
    {
      "epoch": 0.051297428208972436,
      "grad_norm": 1.550613414062432,
      "learning_rate": 1.249328501131771e-06,
      "loss": 0.8468,
      "step": 556
    },
    {
      "epoch": 0.05138968977049937,
      "grad_norm": 2.7132984825720463,
      "learning_rate": 1.2493197536404958e-06,
      "loss": 0.8685,
      "step": 557
    },
    {
      "epoch": 0.05148195133202629,
      "grad_norm": 1.942546454934521,
      "learning_rate": 1.2493109495723817e-06,
      "loss": 0.929,
      "step": 558
    },
    {
      "epoch": 0.051574212893553226,
      "grad_norm": 1.747050475065823,
      "learning_rate": 1.249302088928227e-06,
      "loss": 0.8059,
      "step": 559
    },
    {
      "epoch": 0.05166647445508015,
      "grad_norm": 2.179161275340723,
      "learning_rate": 1.2492931717088338e-06,
      "loss": 0.8583,
      "step": 560
    },
    {
      "epoch": 0.05166647445508015,
      "eval_GEN Loss": 0.4996907711029053,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4268765449523926,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9620192050933838,
      "eval_runtime": 56.3711,
      "eval_samples_per_second": 1.153,
      "eval_steps_per_second": 0.089,
      "step": 560
    },
    {
      "epoch": 0.05175873601660708,
      "grad_norm": 3.458615325297117,
      "learning_rate": 1.2492841979150112e-06,
      "loss": 1.1285,
      "step": 561
    },
    {
      "epoch": 0.05185099757813401,
      "grad_norm": 1.580788882347539,
      "learning_rate": 1.2492751675475717e-06,
      "loss": 0.9045,
      "step": 562
    },
    {
      "epoch": 0.05194325913966094,
      "grad_norm": 1.7784183829994715,
      "learning_rate": 1.2492660806073339e-06,
      "loss": 0.9334,
      "step": 563
    },
    {
      "epoch": 0.052035520701187866,
      "grad_norm": 2.05324686189782,
      "learning_rate": 1.2492569370951214e-06,
      "loss": 0.9087,
      "step": 564
    },
    {
      "epoch": 0.0521277822627148,
      "grad_norm": 2.1872773094388736,
      "learning_rate": 1.2492477370117626e-06,
      "loss": 0.9456,
      "step": 565
    },
    {
      "epoch": 0.0521277822627148,
      "eval_GEN Loss": 0.5036789178848267,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8289473684210527,
      "eval_PRM F1 AUC": 0.749083289680461,
      "eval_PRM F1 Neg": 0.5666666666666667,
      "eval_PRM Loss": 0.47351765632629395,
      "eval_PRM NPV": 0.4594594594594595,
      "eval_PRM Precision": 0.9130434782608695,
      "eval_PRM Recall": 0.7590361445783133,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9850360751152039,
      "eval_runtime": 56.0195,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 565
    },
    {
      "epoch": 0.052220043824241724,
      "grad_norm": 2.250551754667014,
      "learning_rate": 1.2492384803580913e-06,
      "loss": 1.0084,
      "step": 566
    },
    {
      "epoch": 0.052312305385768657,
      "grad_norm": 1.796614062913743,
      "learning_rate": 1.2492291671349464e-06,
      "loss": 0.8762,
      "step": 567
    },
    {
      "epoch": 0.05240456694729558,
      "grad_norm": 2.4404571448323558,
      "learning_rate": 1.2492197973431716e-06,
      "loss": 0.9033,
      "step": 568
    },
    {
      "epoch": 0.052496828508822514,
      "grad_norm": 2.0002215859856256,
      "learning_rate": 1.2492103709836166e-06,
      "loss": 0.8474,
      "step": 569
    },
    {
      "epoch": 0.05258909007034944,
      "grad_norm": 2.031138317769336,
      "learning_rate": 1.249200888057135e-06,
      "loss": 0.8613,
      "step": 570
    },
    {
      "epoch": 0.05258909007034944,
      "eval_GEN Loss": 0.5067321062088013,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8535031847133758,
      "eval_PRM F1 AUC": 0.751440544787847,
      "eval_PRM F1 Neg": 0.5818181818181818,
      "eval_PRM Loss": 0.45650914311408997,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9054054054054054,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9834134578704834,
      "eval_runtime": 55.9727,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 570
    },
    {
      "epoch": 0.05268135163187637,
      "grad_norm": 1.2268314366957531,
      "learning_rate": 1.2491913485645862e-06,
      "loss": 0.8199,
      "step": 571
    },
    {
      "epoch": 0.0527736131934033,
      "grad_norm": 2.776007518800226,
      "learning_rate": 1.2491817525068354e-06,
      "loss": 0.7782,
      "step": 572
    },
    {
      "epoch": 0.05286587475493023,
      "grad_norm": 1.7140385408739764,
      "learning_rate": 1.2491720998847515e-06,
      "loss": 0.8507,
      "step": 573
    },
    {
      "epoch": 0.052958136316457155,
      "grad_norm": 1.504065091867208,
      "learning_rate": 1.2491623906992092e-06,
      "loss": 0.7926,
      "step": 574
    },
    {
      "epoch": 0.05305039787798409,
      "grad_norm": 3.8341411486065207,
      "learning_rate": 1.2491526249510887e-06,
      "loss": 1.2054,
      "step": 575
    },
    {
      "epoch": 0.05305039787798409,
      "eval_GEN Loss": 0.5083337426185608,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.4059189558029175,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 1.0139423608779907,
      "eval_runtime": 55.9813,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 575
    },
    {
      "epoch": 0.05314265943951101,
      "grad_norm": 2.428060017967457,
      "learning_rate": 1.2491428026412752e-06,
      "loss": 0.8207,
      "step": 576
    },
    {
      "epoch": 0.053234921001037945,
      "grad_norm": 1.6934940756786587,
      "learning_rate": 1.2491329237706585e-06,
      "loss": 0.7227,
      "step": 577
    },
    {
      "epoch": 0.05332718256256487,
      "grad_norm": 2.866311618541313,
      "learning_rate": 1.2491229883401335e-06,
      "loss": 0.755,
      "step": 578
    },
    {
      "epoch": 0.0534194441240918,
      "grad_norm": 1.4945832523138576,
      "learning_rate": 1.249112996350601e-06,
      "loss": 0.8642,
      "step": 579
    },
    {
      "epoch": 0.05351170568561873,
      "grad_norm": 2.2047133714725686,
      "learning_rate": 1.2491029478029667e-06,
      "loss": 0.7895,
      "step": 580
    },
    {
      "epoch": 0.05351170568561873,
      "eval_GEN Loss": 0.5080254673957825,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.40413254499435425,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9838341474533081,
      "eval_runtime": 57.004,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 580
    },
    {
      "epoch": 0.05360396724714566,
      "grad_norm": 3.386791216845856,
      "learning_rate": 1.2490928426981406e-06,
      "loss": 1.0675,
      "step": 581
    },
    {
      "epoch": 0.053696228808672586,
      "grad_norm": 1.8008239620645539,
      "learning_rate": 1.2490826810370388e-06,
      "loss": 0.8744,
      "step": 582
    },
    {
      "epoch": 0.05378849037019952,
      "grad_norm": 2.1841412190942178,
      "learning_rate": 1.249072462820582e-06,
      "loss": 0.7634,
      "step": 583
    },
    {
      "epoch": 0.053880751931726444,
      "grad_norm": 1.8399717770459827,
      "learning_rate": 1.2490621880496968e-06,
      "loss": 0.953,
      "step": 584
    },
    {
      "epoch": 0.053973013493253376,
      "grad_norm": 1.8292775229199802,
      "learning_rate": 1.2490518567253136e-06,
      "loss": 0.9233,
      "step": 585
    },
    {
      "epoch": 0.053973013493253376,
      "eval_GEN Loss": 0.5101640820503235,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7264150943396226,
      "eval_PRM F1": 0.8053691275167785,
      "eval_PRM F1 AUC": 0.7310110005238345,
      "eval_PRM F1 Neg": 0.5396825396825397,
      "eval_PRM Loss": 0.48764801025390625,
      "eval_PRM NPV": 0.425,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.7228915662650602,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 1.0290865898132324,
      "eval_runtime": 55.8138,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 585
    },
    {
      "epoch": 0.0540652750547803,
      "grad_norm": 2.3623630200412364,
      "learning_rate": 1.2490414688483688e-06,
      "loss": 0.8771,
      "step": 586
    },
    {
      "epoch": 0.054157536616307234,
      "grad_norm": 2.680394664733209,
      "learning_rate": 1.2490310244198036e-06,
      "loss": 0.9148,
      "step": 587
    },
    {
      "epoch": 0.05424979817783416,
      "grad_norm": 2.3588791333181307,
      "learning_rate": 1.2490205234405652e-06,
      "loss": 0.8595,
      "step": 588
    },
    {
      "epoch": 0.05434205973936109,
      "grad_norm": 2.099221453060737,
      "learning_rate": 1.2490099659116045e-06,
      "loss": 0.88,
      "step": 589
    },
    {
      "epoch": 0.05443432130088802,
      "grad_norm": 1.8877239678915219,
      "learning_rate": 1.2489993518338784e-06,
      "loss": 0.7525,
      "step": 590
    },
    {
      "epoch": 0.05443432130088802,
      "eval_GEN Loss": 0.5080305337905884,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7169811320754716,
      "eval_PRM F1": 0.7945205479452054,
      "eval_PRM F1 AUC": 0.7407019381875328,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.5066564083099365,
      "eval_PRM NPV": 0.4186046511627907,
      "eval_PRM Precision": 0.9206349206349206,
      "eval_PRM Recall": 0.6987951807228916,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 1.0357571840286255,
      "eval_runtime": 55.6673,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 590
    },
    {
      "epoch": 0.05452658286241495,
      "grad_norm": 2.0831530454994596,
      "learning_rate": 1.248988681208349e-06,
      "loss": 0.8641,
      "step": 591
    },
    {
      "epoch": 0.054618844423941874,
      "grad_norm": 2.434562035432444,
      "learning_rate": 1.2489779540359831e-06,
      "loss": 0.9395,
      "step": 592
    },
    {
      "epoch": 0.05471110598546881,
      "grad_norm": 1.9821337764100029,
      "learning_rate": 1.2489671703177527e-06,
      "loss": 0.9168,
      "step": 593
    },
    {
      "epoch": 0.05480336754699573,
      "grad_norm": 1.6688384237423188,
      "learning_rate": 1.2489563300546353e-06,
      "loss": 0.9708,
      "step": 594
    },
    {
      "epoch": 0.054895629108522664,
      "grad_norm": 2.071824994069014,
      "learning_rate": 1.2489454332476133e-06,
      "loss": 1.0497,
      "step": 595
    },
    {
      "epoch": 0.054895629108522664,
      "eval_GEN Loss": 0.5029069185256958,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.45642784237861633,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9888221025466919,
      "eval_runtime": 55.781,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 595
    },
    {
      "epoch": 0.05498789067004959,
      "grad_norm": 1.5930169744134677,
      "learning_rate": 1.2489344798976737e-06,
      "loss": 0.8281,
      "step": 596
    },
    {
      "epoch": 0.05508015223157652,
      "grad_norm": 1.6643989876991003,
      "learning_rate": 1.2489234700058098e-06,
      "loss": 0.9325,
      "step": 597
    },
    {
      "epoch": 0.05517241379310345,
      "grad_norm": 2.053707907675361,
      "learning_rate": 1.2489124035730189e-06,
      "loss": 0.8368,
      "step": 598
    },
    {
      "epoch": 0.05526467535463038,
      "grad_norm": 1.3682926993227886,
      "learning_rate": 1.248901280600304e-06,
      "loss": 0.7565,
      "step": 599
    },
    {
      "epoch": 0.055356936916157305,
      "grad_norm": 1.3096687655332433,
      "learning_rate": 1.2488901010886727e-06,
      "loss": 0.8928,
      "step": 600
    },
    {
      "epoch": 0.055356936916157305,
      "eval_GEN Loss": 0.5018876194953918,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4348255693912506,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9718149304389954,
      "eval_runtime": 55.8031,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 600
    },
    {
      "epoch": 0.05544919847768424,
      "grad_norm": 1.6840601316309256,
      "learning_rate": 1.2488788650391388e-06,
      "loss": 0.9862,
      "step": 601
    },
    {
      "epoch": 0.05554146003921116,
      "grad_norm": 1.6312387559443748,
      "learning_rate": 1.2488675724527198e-06,
      "loss": 0.7968,
      "step": 602
    },
    {
      "epoch": 0.055633721600738095,
      "grad_norm": 2.4815635062361445,
      "learning_rate": 1.2488562233304396e-06,
      "loss": 0.9873,
      "step": 603
    },
    {
      "epoch": 0.05572598316226502,
      "grad_norm": 2.0552191331719247,
      "learning_rate": 1.2488448176733262e-06,
      "loss": 0.8915,
      "step": 604
    },
    {
      "epoch": 0.05581824472379195,
      "grad_norm": 2.117906169476475,
      "learning_rate": 1.2488333554824138e-06,
      "loss": 0.9145,
      "step": 605
    },
    {
      "epoch": 0.05581824472379195,
      "eval_GEN Loss": 0.5006439685821533,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.44969943165779114,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9799278974533081,
      "eval_runtime": 55.7109,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 605
    },
    {
      "epoch": 0.05591050628531888,
      "grad_norm": 2.038318905132651,
      "learning_rate": 1.2488218367587406e-06,
      "loss": 0.8549,
      "step": 606
    },
    {
      "epoch": 0.05600276784684581,
      "grad_norm": 1.778697446036608,
      "learning_rate": 1.2488102615033509e-06,
      "loss": 0.7878,
      "step": 607
    },
    {
      "epoch": 0.056095029408372736,
      "grad_norm": 1.8828030591591378,
      "learning_rate": 1.2487986297172933e-06,
      "loss": 0.7999,
      "step": 608
    },
    {
      "epoch": 0.05618729096989967,
      "grad_norm": 2.1397135575934847,
      "learning_rate": 1.248786941401622e-06,
      "loss": 0.9599,
      "step": 609
    },
    {
      "epoch": 0.056279552531426594,
      "grad_norm": 2.161450949773939,
      "learning_rate": 1.248775196557396e-06,
      "loss": 0.7918,
      "step": 610
    },
    {
      "epoch": 0.056279552531426594,
      "eval_GEN Loss": 0.49783971905708313,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.43421003222465515,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9634615182876587,
      "eval_runtime": 55.7168,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 610
    },
    {
      "epoch": 0.056371814092953526,
      "grad_norm": 1.7052262220432919,
      "learning_rate": 1.24876339518568e-06,
      "loss": 0.785,
      "step": 611
    },
    {
      "epoch": 0.05646407565448045,
      "grad_norm": 1.844819283964889,
      "learning_rate": 1.2487515372875434e-06,
      "loss": 0.8631,
      "step": 612
    },
    {
      "epoch": 0.056556337216007384,
      "grad_norm": 2.002989390981208,
      "learning_rate": 1.2487396228640606e-06,
      "loss": 0.7582,
      "step": 613
    },
    {
      "epoch": 0.05664859877753431,
      "grad_norm": 3.118346437681846,
      "learning_rate": 1.2487276519163114e-06,
      "loss": 0.9991,
      "step": 614
    },
    {
      "epoch": 0.05674086033906124,
      "grad_norm": 2.5488199308499677,
      "learning_rate": 1.2487156244453807e-06,
      "loss": 0.9039,
      "step": 615
    },
    {
      "epoch": 0.05674086033906124,
      "eval_GEN Loss": 0.4985102713108063,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.4201836884021759,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9504807591438293,
      "eval_runtime": 55.8043,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 615
    },
    {
      "epoch": 0.05683312190058817,
      "grad_norm": 3.8890403160471334,
      "learning_rate": 1.2487035404523583e-06,
      "loss": 1.1244,
      "step": 616
    },
    {
      "epoch": 0.0569253834621151,
      "grad_norm": 2.0172657598100177,
      "learning_rate": 1.2486913999383393e-06,
      "loss": 0.8988,
      "step": 617
    },
    {
      "epoch": 0.057017645023642025,
      "grad_norm": 1.4561923092029698,
      "learning_rate": 1.248679202904424e-06,
      "loss": 0.7902,
      "step": 618
    },
    {
      "epoch": 0.05710990658516896,
      "grad_norm": 2.702825562484095,
      "learning_rate": 1.2486669493517176e-06,
      "loss": 0.993,
      "step": 619
    },
    {
      "epoch": 0.05720216814669588,
      "grad_norm": 2.2481441663700052,
      "learning_rate": 1.2486546392813307e-06,
      "loss": 0.8837,
      "step": 620
    },
    {
      "epoch": 0.05720216814669588,
      "eval_GEN Loss": 0.4994538128376007,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8354430379746836,
      "eval_PRM F1 AUC": 0.7019381875327397,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM Loss": 0.45006895065307617,
      "eval_PRM NPV": 0.45161290322580644,
      "eval_PRM Precision": 0.88,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9386418461799622,
      "eval_runtime": 55.781,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 620
    },
    {
      "epoch": 0.057294429708222815,
      "grad_norm": 1.5164571122984896,
      "learning_rate": 1.2486422726943786e-06,
      "loss": 0.8415,
      "step": 621
    },
    {
      "epoch": 0.05738669126974974,
      "grad_norm": 1.770771183189257,
      "learning_rate": 1.2486298495919822e-06,
      "loss": 0.8257,
      "step": 622
    },
    {
      "epoch": 0.05747895283127667,
      "grad_norm": 1.6682291902578132,
      "learning_rate": 1.2486173699752672e-06,
      "loss": 0.821,
      "step": 623
    },
    {
      "epoch": 0.0575712143928036,
      "grad_norm": 2.108982914483475,
      "learning_rate": 1.2486048338453646e-06,
      "loss": 0.9856,
      "step": 624
    },
    {
      "epoch": 0.05766347595433053,
      "grad_norm": 1.810862754522022,
      "learning_rate": 1.2485922412034102e-06,
      "loss": 0.786,
      "step": 625
    },
    {
      "epoch": 0.05766347595433053,
      "eval_GEN Loss": 0.5000378489494324,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.43361374735832214,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.932812511920929,
      "eval_runtime": 55.7417,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 625
    },
    {
      "epoch": 0.057755737515857455,
      "grad_norm": 1.772217409354434,
      "learning_rate": 1.2485795920505455e-06,
      "loss": 0.6705,
      "step": 626
    },
    {
      "epoch": 0.05784799907738439,
      "grad_norm": 2.001067166174967,
      "learning_rate": 1.2485668863879167e-06,
      "loss": 1.0233,
      "step": 627
    },
    {
      "epoch": 0.05794026063891131,
      "grad_norm": 1.4840923781348787,
      "learning_rate": 1.248554124216675e-06,
      "loss": 0.9001,
      "step": 628
    },
    {
      "epoch": 0.058032522200438245,
      "grad_norm": 1.8101679414620182,
      "learning_rate": 1.248541305537977e-06,
      "loss": 0.7187,
      "step": 629
    },
    {
      "epoch": 0.05812478376196517,
      "grad_norm": 2.073748668928117,
      "learning_rate": 1.2485284303529847e-06,
      "loss": 0.8885,
      "step": 630
    },
    {
      "epoch": 0.05812478376196517,
      "eval_GEN Loss": 0.501082181930542,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8622754491017964,
      "eval_PRM F1 AUC": 0.6728653745416449,
      "eval_PRM F1 Neg": 0.4888888888888889,
      "eval_PRM Loss": 0.41623395681381226,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8571428571428571,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9604567289352417,
      "eval_runtime": 55.9232,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 630
    },
    {
      "epoch": 0.0582170453234921,
      "grad_norm": 1.993221305531676,
      "learning_rate": 1.2485154986628646e-06,
      "loss": 0.8211,
      "step": 631
    },
    {
      "epoch": 0.05830930688501903,
      "grad_norm": 1.6783005576180885,
      "learning_rate": 1.2485025104687885e-06,
      "loss": 0.8251,
      "step": 632
    },
    {
      "epoch": 0.05840156844654596,
      "grad_norm": 1.7378727446242044,
      "learning_rate": 1.248489465771933e-06,
      "loss": 0.8405,
      "step": 633
    },
    {
      "epoch": 0.058493830008072886,
      "grad_norm": 1.9168838737841298,
      "learning_rate": 1.2484763645734816e-06,
      "loss": 0.913,
      "step": 634
    },
    {
      "epoch": 0.05858609156959982,
      "grad_norm": 1.5552224413451192,
      "learning_rate": 1.24846320687462e-06,
      "loss": 0.9587,
      "step": 635
    },
    {
      "epoch": 0.05858609156959982,
      "eval_GEN Loss": 0.49984222650527954,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8690476190476191,
      "eval_PRM F1 AUC": 0.678889470927187,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM Loss": 0.40613850951194763,
      "eval_PRM NPV": 0.5238095238095238,
      "eval_PRM Precision": 0.8588235294117647,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9527644515037537,
      "eval_runtime": 56.0062,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 635
    },
    {
      "epoch": 0.058678353131126744,
      "grad_norm": 2.752900272064645,
      "learning_rate": 1.2484499926765416e-06,
      "loss": 1.0392,
      "step": 636
    },
    {
      "epoch": 0.058770614692653676,
      "grad_norm": 2.478031843513477,
      "learning_rate": 1.2484367219804435e-06,
      "loss": 0.7721,
      "step": 637
    },
    {
      "epoch": 0.0588628762541806,
      "grad_norm": 1.750905595546406,
      "learning_rate": 1.2484233947875282e-06,
      "loss": 0.8989,
      "step": 638
    },
    {
      "epoch": 0.058955137815707534,
      "grad_norm": 1.7510339767346743,
      "learning_rate": 1.2484100110990035e-06,
      "loss": 0.9025,
      "step": 639
    },
    {
      "epoch": 0.05904739937723446,
      "grad_norm": 2.2048689696788135,
      "learning_rate": 1.2483965709160827e-06,
      "loss": 0.8448,
      "step": 640
    },
    {
      "epoch": 0.05904739937723446,
      "eval_GEN Loss": 0.5029840469360352,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8606060606060606,
      "eval_PRM F1 AUC": 0.6885804085908852,
      "eval_PRM F1 Neg": 0.5106382978723404,
      "eval_PRM Loss": 0.39927953481674194,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8658536585365854,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9294471144676208,
      "eval_runtime": 55.8663,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 640
    },
    {
      "epoch": 0.05913966093876139,
      "grad_norm": 1.5209893142275832,
      "learning_rate": 1.248383074239983e-06,
      "loss": 0.7128,
      "step": 641
    },
    {
      "epoch": 0.05923192250028832,
      "grad_norm": 1.843063551533251,
      "learning_rate": 1.2483695210719282e-06,
      "loss": 0.792,
      "step": 642
    },
    {
      "epoch": 0.05932418406181525,
      "grad_norm": 1.4352538619861959,
      "learning_rate": 1.248355911413146e-06,
      "loss": 0.6026,
      "step": 643
    },
    {
      "epoch": 0.059416445623342175,
      "grad_norm": 3.812000845948826,
      "learning_rate": 1.2483422452648702e-06,
      "loss": 1.0718,
      "step": 644
    },
    {
      "epoch": 0.05950870718486911,
      "grad_norm": 3.590813734052817,
      "learning_rate": 1.2483285226283387e-06,
      "loss": 0.9815,
      "step": 645
    },
    {
      "epoch": 0.05950870718486911,
      "eval_GEN Loss": 0.502841591835022,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8674698795180723,
      "eval_PRM F1 AUC": 0.6946045049764275,
      "eval_PRM F1 Neg": 0.5217391304347826,
      "eval_PRM Loss": 0.39608943462371826,
      "eval_PRM NPV": 0.5217391304347826,
      "eval_PRM Precision": 0.8674698795180723,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9320913553237915,
      "eval_runtime": 55.9501,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 645
    },
    {
      "epoch": 0.05960096874639603,
      "grad_norm": 1.9928220967621102,
      "learning_rate": 1.2483147435047954e-06,
      "loss": 0.7131,
      "step": 646
    },
    {
      "epoch": 0.059693230307922965,
      "grad_norm": 1.9365920045835188,
      "learning_rate": 1.248300907895489e-06,
      "loss": 0.9459,
      "step": 647
    },
    {
      "epoch": 0.05978549186944989,
      "grad_norm": 2.6623701009527387,
      "learning_rate": 1.2482870158016733e-06,
      "loss": 0.7441,
      "step": 648
    },
    {
      "epoch": 0.05987775343097682,
      "grad_norm": 2.6690979217425994,
      "learning_rate": 1.2482730672246069e-06,
      "loss": 0.8438,
      "step": 649
    },
    {
      "epoch": 0.05997001499250375,
      "grad_norm": 1.74071030127736,
      "learning_rate": 1.2482590621655544e-06,
      "loss": 0.8071,
      "step": 650
    },
    {
      "epoch": 0.05997001499250375,
      "eval_GEN Loss": 0.5032788515090942,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4073921740055084,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9045072197914124,
      "eval_runtime": 55.6871,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 650
    },
    {
      "epoch": 0.06006227655403068,
      "grad_norm": 1.867697574685573,
      "learning_rate": 1.2482450006257845e-06,
      "loss": 0.7068,
      "step": 651
    },
    {
      "epoch": 0.060154538115557606,
      "grad_norm": 3.150656478669109,
      "learning_rate": 1.2482308826065717e-06,
      "loss": 1.0362,
      "step": 652
    },
    {
      "epoch": 0.06024679967708454,
      "grad_norm": 2.0723154529102277,
      "learning_rate": 1.2482167081091953e-06,
      "loss": 0.8605,
      "step": 653
    },
    {
      "epoch": 0.06033906123861146,
      "grad_norm": 1.5733706218276544,
      "learning_rate": 1.24820247713494e-06,
      "loss": 0.8048,
      "step": 654
    },
    {
      "epoch": 0.060431322800138396,
      "grad_norm": 1.9533228833495297,
      "learning_rate": 1.248188189685095e-06,
      "loss": 0.8584,
      "step": 655
    },
    {
      "epoch": 0.060431322800138396,
      "eval_GEN Loss": 0.5047159194946289,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM Loss": 0.43611615896224976,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.907932698726654,
      "eval_runtime": 55.742,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 655
    },
    {
      "epoch": 0.06052358436166532,
      "grad_norm": 1.8164491320202634,
      "learning_rate": 1.2481738457609554e-06,
      "loss": 0.7616,
      "step": 656
    },
    {
      "epoch": 0.06061584592319225,
      "grad_norm": 2.9805106768035183,
      "learning_rate": 1.2481594453638213e-06,
      "loss": 1.0346,
      "step": 657
    },
    {
      "epoch": 0.06070810748471918,
      "grad_norm": 2.198359986050082,
      "learning_rate": 1.248144988494997e-06,
      "loss": 0.9404,
      "step": 658
    },
    {
      "epoch": 0.06080036904624611,
      "grad_norm": 1.7843818714451314,
      "learning_rate": 1.2481304751557932e-06,
      "loss": 0.7976,
      "step": 659
    },
    {
      "epoch": 0.060892630607773036,
      "grad_norm": 1.852373355865221,
      "learning_rate": 1.248115905347525e-06,
      "loss": 0.6315,
      "step": 660
    },
    {
      "epoch": 0.060892630607773036,
      "eval_GEN Loss": 0.5038718581199646,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.379349946975708,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.887620210647583,
      "eval_runtime": 55.8929,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 660
    },
    {
      "epoch": 0.06098489216929997,
      "grad_norm": 2.5190164743513677,
      "learning_rate": 1.2481012790715125e-06,
      "loss": 0.9121,
      "step": 661
    },
    {
      "epoch": 0.061077153730826894,
      "grad_norm": 2.1642822732493094,
      "learning_rate": 1.2480865963290814e-06,
      "loss": 0.9153,
      "step": 662
    },
    {
      "epoch": 0.061169415292353826,
      "grad_norm": 2.4869822741610736,
      "learning_rate": 1.2480718571215622e-06,
      "loss": 0.8426,
      "step": 663
    },
    {
      "epoch": 0.06126167685388075,
      "grad_norm": 1.4804113543475204,
      "learning_rate": 1.2480570614502905e-06,
      "loss": 0.7338,
      "step": 664
    },
    {
      "epoch": 0.061353938415407684,
      "grad_norm": 2.5825031570366948,
      "learning_rate": 1.2480422093166072e-06,
      "loss": 0.988,
      "step": 665
    },
    {
      "epoch": 0.061353938415407684,
      "eval_GEN Loss": 0.5030157566070557,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3597264587879181,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.891406238079071,
      "eval_runtime": 55.8048,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 665
    },
    {
      "epoch": 0.06144619997693461,
      "grad_norm": 2.4865553091482386,
      "learning_rate": 1.2480273007218583e-06,
      "loss": 0.7469,
      "step": 666
    },
    {
      "epoch": 0.06153846153846154,
      "grad_norm": 1.4962681330808214,
      "learning_rate": 1.2480123356673947e-06,
      "loss": 0.7616,
      "step": 667
    },
    {
      "epoch": 0.06163072309998847,
      "grad_norm": 3.3098298083873035,
      "learning_rate": 1.2479973141545727e-06,
      "loss": 1.007,
      "step": 668
    },
    {
      "epoch": 0.0617229846615154,
      "grad_norm": 1.9786754768724968,
      "learning_rate": 1.2479822361847532e-06,
      "loss": 0.9201,
      "step": 669
    },
    {
      "epoch": 0.061815246223042325,
      "grad_norm": 2.2883225948352206,
      "learning_rate": 1.2479671017593031e-06,
      "loss": 0.9133,
      "step": 670
    },
    {
      "epoch": 0.061815246223042325,
      "eval_GEN Loss": 0.5008916258811951,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM Loss": 0.388599693775177,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8749399185180664,
      "eval_runtime": 55.8347,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 670
    },
    {
      "epoch": 0.06190750778456926,
      "grad_norm": 1.6824160258294973,
      "learning_rate": 1.2479519108795937e-06,
      "loss": 0.8843,
      "step": 671
    },
    {
      "epoch": 0.06199976934609618,
      "grad_norm": 1.8734800223296395,
      "learning_rate": 1.2479366635470014e-06,
      "loss": 0.9205,
      "step": 672
    },
    {
      "epoch": 0.06209203090762311,
      "grad_norm": 2.410707480946803,
      "learning_rate": 1.2479213597629084e-06,
      "loss": 1.0086,
      "step": 673
    },
    {
      "epoch": 0.06218429246915004,
      "grad_norm": 2.0368417016154154,
      "learning_rate": 1.247905999528701e-06,
      "loss": 1.0038,
      "step": 674
    },
    {
      "epoch": 0.062276554030676966,
      "grad_norm": 1.6606314268932567,
      "learning_rate": 1.2478905828457716e-06,
      "loss": 0.9871,
      "step": 675
    },
    {
      "epoch": 0.062276554030676966,
      "eval_GEN Loss": 0.5003483295440674,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8533333333333334,
      "eval_PRM F1 AUC": 0.8203247773703509,
      "eval_PRM F1 Neg": 0.6451612903225806,
      "eval_PRM Loss": 0.4477854371070862,
      "eval_PRM NPV": 0.5128205128205128,
      "eval_PRM Precision": 0.9552238805970149,
      "eval_PRM Recall": 0.7710843373493976,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.9173076748847961,
      "eval_runtime": 55.7039,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 675
    },
    {
      "epoch": 0.0623688155922039,
      "grad_norm": 2.4488961881294107,
      "learning_rate": 1.247875109715517e-06,
      "loss": 0.9534,
      "step": 676
    },
    {
      "epoch": 0.062461077153730823,
      "grad_norm": 2.0301141579032165,
      "learning_rate": 1.2478595801393397e-06,
      "loss": 0.9554,
      "step": 677
    },
    {
      "epoch": 0.06255333871525776,
      "grad_norm": 1.7484711809717781,
      "learning_rate": 1.247843994118647e-06,
      "loss": 0.9209,
      "step": 678
    },
    {
      "epoch": 0.06264560027678469,
      "grad_norm": 2.031267928666826,
      "learning_rate": 1.2478283516548509e-06,
      "loss": 0.9142,
      "step": 679
    },
    {
      "epoch": 0.0627378618383116,
      "grad_norm": 1.884594680835046,
      "learning_rate": 1.2478126527493695e-06,
      "loss": 0.7914,
      "step": 680
    },
    {
      "epoch": 0.0627378618383116,
      "eval_GEN Loss": 0.49689099192619324,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8860759493670886,
      "eval_PRM F1 AUC": 0.8129910948140389,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3846823275089264,
      "eval_PRM NPV": 0.5806451612903226,
      "eval_PRM Precision": 0.9333333333333333,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.879687488079071,
      "eval_runtime": 56.6869,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 680
    },
    {
      "epoch": 0.06283012339983854,
      "grad_norm": 2.353693418975392,
      "learning_rate": 1.247796897403625e-06,
      "loss": 0.7576,
      "step": 681
    },
    {
      "epoch": 0.06292238496136547,
      "grad_norm": 2.459026074737489,
      "learning_rate": 1.2477810856190454e-06,
      "loss": 0.9668,
      "step": 682
    },
    {
      "epoch": 0.0630146465228924,
      "grad_norm": 1.5088873303678316,
      "learning_rate": 1.2477652173970633e-06,
      "loss": 0.7493,
      "step": 683
    },
    {
      "epoch": 0.06310690808441932,
      "grad_norm": 2.066196603628531,
      "learning_rate": 1.2477492927391172e-06,
      "loss": 0.996,
      "step": 684
    },
    {
      "epoch": 0.06319916964594625,
      "grad_norm": 2.563516949224603,
      "learning_rate": 1.2477333116466501e-06,
      "loss": 1.1314,
      "step": 685
    },
    {
      "epoch": 0.06319916964594625,
      "eval_GEN Loss": 0.49459177255630493,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.35469162464141846,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9004807472229004,
      "eval_runtime": 56.8087,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 685
    },
    {
      "epoch": 0.06329143120747319,
      "grad_norm": 1.7918439395513766,
      "learning_rate": 1.2477172741211097e-06,
      "loss": 0.7747,
      "step": 686
    },
    {
      "epoch": 0.06338369276900012,
      "grad_norm": 1.6754035191224634,
      "learning_rate": 1.2477011801639502e-06,
      "loss": 0.8361,
      "step": 687
    },
    {
      "epoch": 0.06347595433052704,
      "grad_norm": 2.9061349298415404,
      "learning_rate": 1.2476850297766294e-06,
      "loss": 0.9363,
      "step": 688
    },
    {
      "epoch": 0.06356821589205397,
      "grad_norm": 1.8680575195107672,
      "learning_rate": 1.247668822960611e-06,
      "loss": 0.7602,
      "step": 689
    },
    {
      "epoch": 0.0636604774535809,
      "grad_norm": 2.0860233465747173,
      "learning_rate": 1.247652559717364e-06,
      "loss": 0.7808,
      "step": 690
    },
    {
      "epoch": 0.0636604774535809,
      "eval_GEN Loss": 0.4925038814544678,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.35282284021377563,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.884495198726654,
      "eval_runtime": 56.5919,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 690
    },
    {
      "epoch": 0.06375273901510783,
      "grad_norm": 1.8577118351161288,
      "learning_rate": 1.2476362400483617e-06,
      "loss": 0.8205,
      "step": 691
    },
    {
      "epoch": 0.06384500057663475,
      "grad_norm": 1.4626529248849887,
      "learning_rate": 1.2476198639550833e-06,
      "loss": 0.7822,
      "step": 692
    },
    {
      "epoch": 0.06393726213816169,
      "grad_norm": 2.315929103420837,
      "learning_rate": 1.247603431439013e-06,
      "loss": 0.8778,
      "step": 693
    },
    {
      "epoch": 0.06402952369968862,
      "grad_norm": 1.4148957871947343,
      "learning_rate": 1.2475869425016396e-06,
      "loss": 0.7385,
      "step": 694
    },
    {
      "epoch": 0.06412178526121555,
      "grad_norm": 1.9927127988576974,
      "learning_rate": 1.2475703971444575e-06,
      "loss": 0.7152,
      "step": 695
    },
    {
      "epoch": 0.06412178526121555,
      "eval_GEN Loss": 0.49013301730155945,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.36663177609443665,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8553485870361328,
      "eval_runtime": 56.8218,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 695
    },
    {
      "epoch": 0.06421404682274247,
      "grad_norm": 2.207444043712501,
      "learning_rate": 1.2475537953689662e-06,
      "loss": 0.713,
      "step": 696
    },
    {
      "epoch": 0.0643063083842694,
      "grad_norm": 1.886201426859166,
      "learning_rate": 1.2475371371766702e-06,
      "loss": 0.8012,
      "step": 697
    },
    {
      "epoch": 0.06439856994579633,
      "grad_norm": 2.5680702309053136,
      "learning_rate": 1.2475204225690785e-06,
      "loss": 0.9574,
      "step": 698
    },
    {
      "epoch": 0.06449083150732327,
      "grad_norm": 2.658370152428033,
      "learning_rate": 1.2475036515477065e-06,
      "loss": 1.0898,
      "step": 699
    },
    {
      "epoch": 0.06458309306885018,
      "grad_norm": 2.197277787521508,
      "learning_rate": 1.2474868241140737e-06,
      "loss": 0.928,
      "step": 700
    },
    {
      "epoch": 0.06458309306885018,
      "eval_GEN Loss": 0.4897848963737488,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.34977298974990845,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8474759459495544,
      "eval_runtime": 55.9244,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 700
    },
    {
      "epoch": 0.06467535463037712,
      "grad_norm": 2.42990179685859,
      "learning_rate": 1.2474699402697052e-06,
      "loss": 0.8279,
      "step": 701
    },
    {
      "epoch": 0.06476761619190405,
      "grad_norm": 2.472823147937655,
      "learning_rate": 1.2474530000161308e-06,
      "loss": 0.9482,
      "step": 702
    },
    {
      "epoch": 0.06485987775343098,
      "grad_norm": 2.446117329514279,
      "learning_rate": 1.2474360033548858e-06,
      "loss": 0.7692,
      "step": 703
    },
    {
      "epoch": 0.0649521393149579,
      "grad_norm": 1.973986088568778,
      "learning_rate": 1.2474189502875105e-06,
      "loss": 0.8697,
      "step": 704
    },
    {
      "epoch": 0.06504440087648483,
      "grad_norm": 2.7409896385756114,
      "learning_rate": 1.2474018408155502e-06,
      "loss": 0.8734,
      "step": 705
    },
    {
      "epoch": 0.06504440087648483,
      "eval_GEN Loss": 0.49004632234573364,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.3554302752017975,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8493990302085876,
      "eval_runtime": 56.0916,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 705
    },
    {
      "epoch": 0.06513666243801176,
      "grad_norm": 2.219226330814617,
      "learning_rate": 1.2473846749405556e-06,
      "loss": 1.0019,
      "step": 706
    },
    {
      "epoch": 0.0652289239995387,
      "grad_norm": 1.8740037144485404,
      "learning_rate": 1.2473674526640817e-06,
      "loss": 0.8388,
      "step": 707
    },
    {
      "epoch": 0.06532118556106561,
      "grad_norm": 2.2348521337067724,
      "learning_rate": 1.2473501739876899e-06,
      "loss": 1.0764,
      "step": 708
    },
    {
      "epoch": 0.06541344712259255,
      "grad_norm": 1.9185201204998532,
      "learning_rate": 1.2473328389129457e-06,
      "loss": 0.8188,
      "step": 709
    },
    {
      "epoch": 0.06550570868411948,
      "grad_norm": 1.7479011149182606,
      "learning_rate": 1.2473154474414201e-06,
      "loss": 0.8953,
      "step": 710
    },
    {
      "epoch": 0.06550570868411948,
      "eval_GEN Loss": 0.4939427077770233,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8789808917197452,
      "eval_PRM F1 AUC": 0.8069669984284967,
      "eval_PRM F1 Neg": 0.6545454545454545,
      "eval_PRM Loss": 0.3669225573539734,
      "eval_PRM NPV": 0.5625,
      "eval_PRM Precision": 0.9324324324324325,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8590745329856873,
      "eval_runtime": 55.7164,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 710
    },
    {
      "epoch": 0.06559797024564641,
      "grad_norm": 1.9154297915300535,
      "learning_rate": 1.247297999574689e-06,
      "loss": 0.8618,
      "step": 711
    },
    {
      "epoch": 0.06569023180717333,
      "grad_norm": 2.020703353307214,
      "learning_rate": 1.2472804953143337e-06,
      "loss": 0.8877,
      "step": 712
    },
    {
      "epoch": 0.06578249336870026,
      "grad_norm": 1.8524179286156006,
      "learning_rate": 1.2472629346619405e-06,
      "loss": 0.7848,
      "step": 713
    },
    {
      "epoch": 0.0658747549302272,
      "grad_norm": 1.9779243978117849,
      "learning_rate": 1.2472453176191007e-06,
      "loss": 0.8378,
      "step": 714
    },
    {
      "epoch": 0.06596701649175413,
      "grad_norm": 2.1877741410284397,
      "learning_rate": 1.2472276441874107e-06,
      "loss": 0.8562,
      "step": 715
    },
    {
      "epoch": 0.06596701649175413,
      "eval_GEN Loss": 0.49638262391090393,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3564877510070801,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.859375,
      "eval_runtime": 55.895,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 715
    },
    {
      "epoch": 0.06605927805328105,
      "grad_norm": 2.4350153073154597,
      "learning_rate": 1.2472099143684723e-06,
      "loss": 0.8768,
      "step": 716
    },
    {
      "epoch": 0.06615153961480798,
      "grad_norm": 1.4549457216393873,
      "learning_rate": 1.247192128163892e-06,
      "loss": 0.7692,
      "step": 717
    },
    {
      "epoch": 0.06624380117633491,
      "grad_norm": 2.505367115067907,
      "learning_rate": 1.2471742855752816e-06,
      "loss": 0.8329,
      "step": 718
    },
    {
      "epoch": 0.06633606273786184,
      "grad_norm": 1.8190506564213158,
      "learning_rate": 1.2471563866042583e-06,
      "loss": 0.9022,
      "step": 719
    },
    {
      "epoch": 0.06642832429938876,
      "grad_norm": 1.7817688080675171,
      "learning_rate": 1.247138431252444e-06,
      "loss": 0.9805,
      "step": 720
    },
    {
      "epoch": 0.06642832429938876,
      "eval_GEN Loss": 0.4984198212623596,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.34813085198402405,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8671875,
      "eval_runtime": 55.8061,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 720
    },
    {
      "epoch": 0.06652058586091569,
      "grad_norm": 2.7336873193573243,
      "learning_rate": 1.2471204195214656e-06,
      "loss": 0.8606,
      "step": 721
    },
    {
      "epoch": 0.06661284742244263,
      "grad_norm": 2.3306084563214933,
      "learning_rate": 1.2471023514129557e-06,
      "loss": 0.7781,
      "step": 722
    },
    {
      "epoch": 0.06670510898396956,
      "grad_norm": 1.7219822318993079,
      "learning_rate": 1.2470842269285517e-06,
      "loss": 0.8055,
      "step": 723
    },
    {
      "epoch": 0.06679737054549648,
      "grad_norm": 1.9205101088879777,
      "learning_rate": 1.2470660460698954e-06,
      "loss": 0.9165,
      "step": 724
    },
    {
      "epoch": 0.06688963210702341,
      "grad_norm": 2.9798148380574,
      "learning_rate": 1.2470478088386351e-06,
      "loss": 0.9771,
      "step": 725
    },
    {
      "epoch": 0.06688963210702341,
      "eval_GEN Loss": 0.49796873331069946,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.34290528297424316,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8899038434028625,
      "eval_runtime": 56.1946,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 725
    },
    {
      "epoch": 0.06698189366855034,
      "grad_norm": 2.662186811434316,
      "learning_rate": 1.2470295152364233e-06,
      "loss": 0.9973,
      "step": 726
    },
    {
      "epoch": 0.06707415523007727,
      "grad_norm": 2.2845077888095373,
      "learning_rate": 1.2470111652649179e-06,
      "loss": 0.9237,
      "step": 727
    },
    {
      "epoch": 0.06716641679160419,
      "grad_norm": 2.720090139434416,
      "learning_rate": 1.2469927589257813e-06,
      "loss": 0.7501,
      "step": 728
    },
    {
      "epoch": 0.06725867835313112,
      "grad_norm": 1.522941660785064,
      "learning_rate": 1.2469742962206818e-06,
      "loss": 0.8071,
      "step": 729
    },
    {
      "epoch": 0.06735093991465806,
      "grad_norm": 1.4019671017176252,
      "learning_rate": 1.2469557771512929e-06,
      "loss": 0.7039,
      "step": 730
    },
    {
      "epoch": 0.06735093991465806,
      "eval_GEN Loss": 0.4958108067512512,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.36516496539115906,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8774639368057251,
      "eval_runtime": 55.7974,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 730
    },
    {
      "epoch": 0.06744320147618499,
      "grad_norm": 1.1680217200435885,
      "learning_rate": 1.2469372017192924e-06,
      "loss": 0.8015,
      "step": 731
    },
    {
      "epoch": 0.06753546303771191,
      "grad_norm": 1.5781252944765307,
      "learning_rate": 1.2469185699263637e-06,
      "loss": 0.8491,
      "step": 732
    },
    {
      "epoch": 0.06762772459923884,
      "grad_norm": 1.8145479068747337,
      "learning_rate": 1.2468998817741953e-06,
      "loss": 0.7989,
      "step": 733
    },
    {
      "epoch": 0.06771998616076577,
      "grad_norm": 1.6535751360344864,
      "learning_rate": 1.2468811372644807e-06,
      "loss": 0.9262,
      "step": 734
    },
    {
      "epoch": 0.0678122477222927,
      "grad_norm": 1.9278342852110244,
      "learning_rate": 1.2468623363989185e-06,
      "loss": 1.0176,
      "step": 735
    },
    {
      "epoch": 0.0678122477222927,
      "eval_GEN Loss": 0.49498119950294495,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8481012658227848,
      "eval_PRM F1 AUC": 0.7297014143530645,
      "eval_PRM F1 Neg": 0.5555555555555556,
      "eval_PRM Loss": 0.4146951138973236,
      "eval_PRM NPV": 0.4838709677419355,
      "eval_PRM Precision": 0.8933333333333333,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8856971263885498,
      "eval_runtime": 55.6896,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 735
    },
    {
      "epoch": 0.06790450928381962,
      "grad_norm": 1.6121518007170748,
      "learning_rate": 1.2468434791792128e-06,
      "loss": 0.9168,
      "step": 736
    },
    {
      "epoch": 0.06799677084534655,
      "grad_norm": 2.38794987956524,
      "learning_rate": 1.246824565607072e-06,
      "loss": 0.9095,
      "step": 737
    },
    {
      "epoch": 0.06808903240687349,
      "grad_norm": 1.813456431482765,
      "learning_rate": 1.2468055956842105e-06,
      "loss": 0.9157,
      "step": 738
    },
    {
      "epoch": 0.06818129396840042,
      "grad_norm": 1.4495805711914043,
      "learning_rate": 1.2467865694123468e-06,
      "loss": 0.5399,
      "step": 739
    },
    {
      "epoch": 0.06827355552992734,
      "grad_norm": 2.3167972690740153,
      "learning_rate": 1.2467674867932058e-06,
      "loss": 0.7926,
      "step": 740
    },
    {
      "epoch": 0.06827355552992734,
      "eval_GEN Loss": 0.49450138211250305,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8407643312101911,
      "eval_PRM F1 AUC": 0.7236773179675223,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.43435630202293396,
      "eval_PRM NPV": 0.46875,
      "eval_PRM Precision": 0.8918918918918919,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8999999761581421,
      "eval_runtime": 55.9371,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 740
    },
    {
      "epoch": 0.06836581709145427,
      "grad_norm": 1.5071480108442745,
      "learning_rate": 1.2467483478285164e-06,
      "loss": 0.7966,
      "step": 741
    },
    {
      "epoch": 0.0684580786529812,
      "grad_norm": 2.702874301518555,
      "learning_rate": 1.2467291525200132e-06,
      "loss": 1.0614,
      "step": 742
    },
    {
      "epoch": 0.06855034021450813,
      "grad_norm": 2.365685824108943,
      "learning_rate": 1.2467099008694353e-06,
      "loss": 0.9154,
      "step": 743
    },
    {
      "epoch": 0.06864260177603505,
      "grad_norm": 2.95301854010953,
      "learning_rate": 1.2466905928785278e-06,
      "loss": 0.8738,
      "step": 744
    },
    {
      "epoch": 0.06873486333756199,
      "grad_norm": 1.5476261126743078,
      "learning_rate": 1.24667122854904e-06,
      "loss": 0.6732,
      "step": 745
    },
    {
      "epoch": 0.06873486333756199,
      "eval_GEN Loss": 0.4933946132659912,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8354430379746836,
      "eval_PRM F1 AUC": 0.7019381875327397,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM Loss": 0.4207013249397278,
      "eval_PRM NPV": 0.45161290322580644,
      "eval_PRM Precision": 0.88,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9192307591438293,
      "eval_runtime": 55.8318,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 745
    },
    {
      "epoch": 0.06882712489908892,
      "grad_norm": 2.3640495382258315,
      "learning_rate": 1.2466518078827272e-06,
      "loss": 0.9631,
      "step": 746
    },
    {
      "epoch": 0.06891938646061585,
      "grad_norm": 2.8018933198562173,
      "learning_rate": 1.246632330881349e-06,
      "loss": 0.859,
      "step": 747
    },
    {
      "epoch": 0.06901164802214277,
      "grad_norm": 1.8883189644042988,
      "learning_rate": 1.2466127975466704e-06,
      "loss": 0.9006,
      "step": 748
    },
    {
      "epoch": 0.0691039095836697,
      "grad_norm": 1.6165626373161106,
      "learning_rate": 1.2465932078804618e-06,
      "loss": 0.6278,
      "step": 749
    },
    {
      "epoch": 0.06919617114519663,
      "grad_norm": 2.5763895171342583,
      "learning_rate": 1.2465735618844982e-06,
      "loss": 0.7422,
      "step": 750
    },
    {
      "epoch": 0.06919617114519663,
      "eval_GEN Loss": 0.4946337640285492,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8407643312101911,
      "eval_PRM F1 AUC": 0.7236773179675223,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.4318975508213043,
      "eval_PRM NPV": 0.46875,
      "eval_PRM Precision": 0.8918918918918919,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9271033406257629,
      "eval_runtime": 55.804,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 750
    },
    {
      "epoch": 0.06928843270672357,
      "grad_norm": 1.6546090660530792,
      "learning_rate": 1.2465538595605603e-06,
      "loss": 0.8247,
      "step": 751
    },
    {
      "epoch": 0.06938069426825048,
      "grad_norm": 2.1096823021207367,
      "learning_rate": 1.2465341009104332e-06,
      "loss": 0.8506,
      "step": 752
    },
    {
      "epoch": 0.06947295582977742,
      "grad_norm": 2.1639271288843616,
      "learning_rate": 1.2465142859359071e-06,
      "loss": 1.0005,
      "step": 753
    },
    {
      "epoch": 0.06956521739130435,
      "grad_norm": 2.036608387513846,
      "learning_rate": 1.2464944146387788e-06,
      "loss": 0.7255,
      "step": 754
    },
    {
      "epoch": 0.06965747895283128,
      "grad_norm": 2.231108365933131,
      "learning_rate": 1.2464744870208482e-06,
      "loss": 0.8253,
      "step": 755
    },
    {
      "epoch": 0.06965747895283128,
      "eval_GEN Loss": 0.49510422348976135,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8481012658227848,
      "eval_PRM F1 AUC": 0.7297014143530645,
      "eval_PRM F1 Neg": 0.5555555555555556,
      "eval_PRM Loss": 0.42670556902885437,
      "eval_PRM NPV": 0.4838709677419355,
      "eval_PRM Precision": 0.8933333333333333,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9216346144676208,
      "eval_runtime": 55.7729,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 755
    },
    {
      "epoch": 0.0697497405143582,
      "grad_norm": 2.458747699242366,
      "learning_rate": 1.2464545030839211e-06,
      "loss": 0.88,
      "step": 756
    },
    {
      "epoch": 0.06984200207588513,
      "grad_norm": 1.7411311492077937,
      "learning_rate": 1.2464344628298089e-06,
      "loss": 0.779,
      "step": 757
    },
    {
      "epoch": 0.06993426363741206,
      "grad_norm": 1.7288453764623393,
      "learning_rate": 1.2464143662603276e-06,
      "loss": 0.7594,
      "step": 758
    },
    {
      "epoch": 0.070026525198939,
      "grad_norm": 1.8324765053708996,
      "learning_rate": 1.2463942133772982e-06,
      "loss": 0.7864,
      "step": 759
    },
    {
      "epoch": 0.07011878676046591,
      "grad_norm": 2.4382269154942757,
      "learning_rate": 1.2463740041825472e-06,
      "loss": 0.8951,
      "step": 760
    },
    {
      "epoch": 0.07011878676046591,
      "eval_GEN Loss": 0.4967345595359802,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.4055047035217285,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9228966236114502,
      "eval_runtime": 55.7798,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 760
    },
    {
      "epoch": 0.07021104832199285,
      "grad_norm": 1.8272948263291933,
      "learning_rate": 1.2463537386779058e-06,
      "loss": 0.949,
      "step": 761
    },
    {
      "epoch": 0.07030330988351978,
      "grad_norm": 1.7831105003221732,
      "learning_rate": 1.2463334168652106e-06,
      "loss": 0.9366,
      "step": 762
    },
    {
      "epoch": 0.07039557144504671,
      "grad_norm": 1.4386035838480709,
      "learning_rate": 1.2463130387463033e-06,
      "loss": 0.7774,
      "step": 763
    },
    {
      "epoch": 0.07048783300657363,
      "grad_norm": 2.5212138362814307,
      "learning_rate": 1.2462926043230303e-06,
      "loss": 0.9012,
      "step": 764
    },
    {
      "epoch": 0.07058009456810056,
      "grad_norm": 1.7618420271161546,
      "learning_rate": 1.2462721135972436e-06,
      "loss": 0.8087,
      "step": 765
    },
    {
      "epoch": 0.07058009456810056,
      "eval_GEN Loss": 0.4948992133140564,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8354430379746836,
      "eval_PRM F1 AUC": 0.7019381875327397,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM Loss": 0.3998696208000183,
      "eval_PRM NPV": 0.45161290322580644,
      "eval_PRM Precision": 0.88,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.901562511920929,
      "eval_runtime": 55.6741,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 765
    },
    {
      "epoch": 0.0706723561296275,
      "grad_norm": 1.9743598065521322,
      "learning_rate": 1.2462515665708e-06,
      "loss": 0.6323,
      "step": 766
    },
    {
      "epoch": 0.07076461769115443,
      "grad_norm": 1.8900912865651107,
      "learning_rate": 1.2462309632455616e-06,
      "loss": 0.8452,
      "step": 767
    },
    {
      "epoch": 0.07085687925268135,
      "grad_norm": 1.9386372731860115,
      "learning_rate": 1.2462103036233957e-06,
      "loss": 0.8129,
      "step": 768
    },
    {
      "epoch": 0.07094914081420828,
      "grad_norm": 1.5221375336860836,
      "learning_rate": 1.246189587706174e-06,
      "loss": 0.7972,
      "step": 769
    },
    {
      "epoch": 0.07104140237573521,
      "grad_norm": 1.946764142267064,
      "learning_rate": 1.2461688154957744e-06,
      "loss": 0.759,
      "step": 770
    },
    {
      "epoch": 0.07104140237573521,
      "eval_GEN Loss": 0.4942178428173065,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.39047446846961975,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8980769515037537,
      "eval_runtime": 55.8653,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 770
    },
    {
      "epoch": 0.07113366393726214,
      "grad_norm": 2.1233654864649485,
      "learning_rate": 1.2461479869940787e-06,
      "loss": 0.8307,
      "step": 771
    },
    {
      "epoch": 0.07122592549878906,
      "grad_norm": 2.017896776743582,
      "learning_rate": 1.2461271022029748e-06,
      "loss": 1.0371,
      "step": 772
    },
    {
      "epoch": 0.071318187060316,
      "grad_norm": 1.7703832595630926,
      "learning_rate": 1.2461061611243553e-06,
      "loss": 0.9259,
      "step": 773
    },
    {
      "epoch": 0.07141044862184293,
      "grad_norm": 2.8705218440160527,
      "learning_rate": 1.246085163760118e-06,
      "loss": 0.8962,
      "step": 774
    },
    {
      "epoch": 0.07150271018336986,
      "grad_norm": 2.6644405002345675,
      "learning_rate": 1.246064110112165e-06,
      "loss": 0.9875,
      "step": 775
    },
    {
      "epoch": 0.07150271018336986,
      "eval_GEN Loss": 0.4935784339904785,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.3880944550037384,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8998197317123413,
      "eval_runtime": 55.8179,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 775
    },
    {
      "epoch": 0.07159497174489678,
      "grad_norm": 2.039120209572376,
      "learning_rate": 1.2460430001824053e-06,
      "loss": 0.8678,
      "step": 776
    },
    {
      "epoch": 0.07168723330642371,
      "grad_norm": 1.8770441809021905,
      "learning_rate": 1.2460218339727512e-06,
      "loss": 0.7401,
      "step": 777
    },
    {
      "epoch": 0.07177949486795064,
      "grad_norm": 2.367333899626303,
      "learning_rate": 1.2460006114851212e-06,
      "loss": 1.145,
      "step": 778
    },
    {
      "epoch": 0.07187175642947757,
      "grad_norm": 1.5591185213212513,
      "learning_rate": 1.245979332721438e-06,
      "loss": 0.9962,
      "step": 779
    },
    {
      "epoch": 0.07196401799100449,
      "grad_norm": 2.1569361671091802,
      "learning_rate": 1.2459579976836303e-06,
      "loss": 0.9726,
      "step": 780
    },
    {
      "epoch": 0.07196401799100449,
      "eval_GEN Loss": 0.49650710821151733,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.3951421082019806,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9027644395828247,
      "eval_runtime": 56.9022,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 780
    },
    {
      "epoch": 0.07205627955253142,
      "grad_norm": 1.8591473145629258,
      "learning_rate": 1.2459366063736318e-06,
      "loss": 0.8798,
      "step": 781
    },
    {
      "epoch": 0.07214854111405836,
      "grad_norm": 2.414556386299106,
      "learning_rate": 1.2459151587933801e-06,
      "loss": 0.7836,
      "step": 782
    },
    {
      "epoch": 0.07224080267558529,
      "grad_norm": 1.7918474384343825,
      "learning_rate": 1.2458936549448198e-06,
      "loss": 0.7399,
      "step": 783
    },
    {
      "epoch": 0.07233306423711221,
      "grad_norm": 1.9349954353770182,
      "learning_rate": 1.2458720948298992e-06,
      "loss": 0.8301,
      "step": 784
    },
    {
      "epoch": 0.07242532579863914,
      "grad_norm": 2.243189487230969,
      "learning_rate": 1.2458504784505717e-06,
      "loss": 0.8181,
      "step": 785
    },
    {
      "epoch": 0.07242532579863914,
      "eval_GEN Loss": 0.49928420782089233,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.37937721610069275,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9114182591438293,
      "eval_runtime": 56.9808,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 785
    },
    {
      "epoch": 0.07251758736016607,
      "grad_norm": 2.163418551734592,
      "learning_rate": 1.245828805808797e-06,
      "loss": 0.5935,
      "step": 786
    },
    {
      "epoch": 0.072609848921693,
      "grad_norm": 2.1430336298179045,
      "learning_rate": 1.2458070769065385e-06,
      "loss": 0.9858,
      "step": 787
    },
    {
      "epoch": 0.07270211048321992,
      "grad_norm": 2.756131543633587,
      "learning_rate": 1.2457852917457656e-06,
      "loss": 0.919,
      "step": 788
    },
    {
      "epoch": 0.07279437204474686,
      "grad_norm": 3.10291318997362,
      "learning_rate": 1.2457634503284525e-06,
      "loss": 0.7908,
      "step": 789
    },
    {
      "epoch": 0.07288663360627379,
      "grad_norm": 1.9828118890125443,
      "learning_rate": 1.2457415526565784e-06,
      "loss": 0.9122,
      "step": 790
    },
    {
      "epoch": 0.07288663360627379,
      "eval_GEN Loss": 0.4999357759952545,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37910550832748413,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9312499761581421,
      "eval_runtime": 56.6552,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 790
    },
    {
      "epoch": 0.07297889516780072,
      "grad_norm": 1.8759492012191048,
      "learning_rate": 1.2457195987321278e-06,
      "loss": 1.0204,
      "step": 791
    },
    {
      "epoch": 0.07307115672932764,
      "grad_norm": 1.7356039772475023,
      "learning_rate": 1.2456975885570899e-06,
      "loss": 0.8332,
      "step": 792
    },
    {
      "epoch": 0.07316341829085457,
      "grad_norm": 2.2456585233178714,
      "learning_rate": 1.2456755221334597e-06,
      "loss": 0.8643,
      "step": 793
    },
    {
      "epoch": 0.0732556798523815,
      "grad_norm": 1.6329985065250308,
      "learning_rate": 1.2456533994632366e-06,
      "loss": 0.5548,
      "step": 794
    },
    {
      "epoch": 0.07334794141390844,
      "grad_norm": 2.2492550574621477,
      "learning_rate": 1.2456312205484256e-06,
      "loss": 0.8706,
      "step": 795
    },
    {
      "epoch": 0.07334794141390844,
      "eval_GEN Loss": 0.4971840977668762,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.39162924885749817,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9308894276618958,
      "eval_runtime": 57.1847,
      "eval_samples_per_second": 1.137,
      "eval_steps_per_second": 0.087,
      "step": 795
    },
    {
      "epoch": 0.07344020297543535,
      "grad_norm": 2.33775263480178,
      "learning_rate": 1.2456089853910366e-06,
      "loss": 0.8396,
      "step": 796
    },
    {
      "epoch": 0.07353246453696229,
      "grad_norm": 1.9459634424275452,
      "learning_rate": 1.2455866939930844e-06,
      "loss": 0.924,
      "step": 797
    },
    {
      "epoch": 0.07362472609848922,
      "grad_norm": 2.163106136718416,
      "learning_rate": 1.2455643463565893e-06,
      "loss": 0.6806,
      "step": 798
    },
    {
      "epoch": 0.07371698766001615,
      "grad_norm": 1.5412218241088955,
      "learning_rate": 1.245541942483576e-06,
      "loss": 0.8529,
      "step": 799
    },
    {
      "epoch": 0.07380924922154307,
      "grad_norm": 2.0771376002130038,
      "learning_rate": 1.2455194823760756e-06,
      "loss": 0.9164,
      "step": 800
    },
    {
      "epoch": 0.07380924922154307,
      "eval_GEN Loss": 0.4945929944515228,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.4143304228782654,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9093149304389954,
      "eval_runtime": 57.0837,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 800
    },
    {
      "epoch": 0.07390151078307,
      "grad_norm": 2.203026010894973,
      "learning_rate": 1.2454969660361228e-06,
      "loss": 0.912,
      "step": 801
    },
    {
      "epoch": 0.07399377234459693,
      "grad_norm": 1.6362508475412942,
      "learning_rate": 1.2454743934657582e-06,
      "loss": 0.6309,
      "step": 802
    },
    {
      "epoch": 0.07408603390612387,
      "grad_norm": 1.8245987540417024,
      "learning_rate": 1.2454517646670273e-06,
      "loss": 0.9523,
      "step": 803
    },
    {
      "epoch": 0.07417829546765078,
      "grad_norm": 2.565586744284615,
      "learning_rate": 1.245429079641981e-06,
      "loss": 0.7819,
      "step": 804
    },
    {
      "epoch": 0.07427055702917772,
      "grad_norm": 1.4786286126417236,
      "learning_rate": 1.245406338392675e-06,
      "loss": 0.8991,
      "step": 805
    },
    {
      "epoch": 0.07427055702917772,
      "eval_GEN Loss": 0.4920736849308014,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.4271751940250397,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.90234375,
      "eval_runtime": 56.8199,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 805
    },
    {
      "epoch": 0.07436281859070465,
      "grad_norm": 2.533897967555276,
      "learning_rate": 1.2453835409211701e-06,
      "loss": 0.7043,
      "step": 806
    },
    {
      "epoch": 0.07445508015223158,
      "grad_norm": 1.8240446733405726,
      "learning_rate": 1.2453606872295321e-06,
      "loss": 0.9383,
      "step": 807
    },
    {
      "epoch": 0.0745473417137585,
      "grad_norm": 2.5124258386859206,
      "learning_rate": 1.2453377773198321e-06,
      "loss": 0.8692,
      "step": 808
    },
    {
      "epoch": 0.07463960327528543,
      "grad_norm": 2.7066494905774383,
      "learning_rate": 1.2453148111941465e-06,
      "loss": 0.9071,
      "step": 809
    },
    {
      "epoch": 0.07473186483681236,
      "grad_norm": 1.408810146832989,
      "learning_rate": 1.2452917888545563e-06,
      "loss": 0.7177,
      "step": 810
    },
    {
      "epoch": 0.07473186483681236,
      "eval_GEN Loss": 0.4875481426715851,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.3897683620452881,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9242788553237915,
      "eval_runtime": 56.6804,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 810
    },
    {
      "epoch": 0.0748241263983393,
      "grad_norm": 1.8834786631308178,
      "learning_rate": 1.2452687103031478e-06,
      "loss": 0.875,
      "step": 811
    },
    {
      "epoch": 0.07491638795986622,
      "grad_norm": 1.5732525514551252,
      "learning_rate": 1.2452455755420124e-06,
      "loss": 0.6362,
      "step": 812
    },
    {
      "epoch": 0.07500864952139315,
      "grad_norm": 2.282625772188802,
      "learning_rate": 1.2452223845732467e-06,
      "loss": 1.0403,
      "step": 813
    },
    {
      "epoch": 0.07510091108292008,
      "grad_norm": 2.104306217863785,
      "learning_rate": 1.2451991373989522e-06,
      "loss": 0.7805,
      "step": 814
    },
    {
      "epoch": 0.07519317264444701,
      "grad_norm": 1.633308231347707,
      "learning_rate": 1.245175834021236e-06,
      "loss": 0.7325,
      "step": 815
    },
    {
      "epoch": 0.07519317264444701,
      "eval_GEN Loss": 0.4881250858306885,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.874251497005988,
      "eval_PRM F1 AUC": 0.7006286013619696,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM Loss": 0.3997836709022522,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.8690476190476191,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9693509340286255,
      "eval_runtime": 56.7256,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 815
    },
    {
      "epoch": 0.07528543420597393,
      "grad_norm": 2.0255801243871656,
      "learning_rate": 1.2451524744422091e-06,
      "loss": 0.9489,
      "step": 816
    },
    {
      "epoch": 0.07537769576750086,
      "grad_norm": 2.9147138178744822,
      "learning_rate": 1.245129058663989e-06,
      "loss": 1.0011,
      "step": 817
    },
    {
      "epoch": 0.0754699573290278,
      "grad_norm": 1.4163130766664815,
      "learning_rate": 1.2451055866886974e-06,
      "loss": 0.6281,
      "step": 818
    },
    {
      "epoch": 0.07556221889055473,
      "grad_norm": 1.5326618856461074,
      "learning_rate": 1.2450820585184618e-06,
      "loss": 0.861,
      "step": 819
    },
    {
      "epoch": 0.07565448045208165,
      "grad_norm": 2.2389560997418094,
      "learning_rate": 1.2450584741554137e-06,
      "loss": 0.9153,
      "step": 820
    },
    {
      "epoch": 0.07565448045208165,
      "eval_GEN Loss": 0.48855963349342346,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8554216867469879,
      "eval_PRM F1 AUC": 0.6668412781561027,
      "eval_PRM F1 Neg": 0.4782608695652174,
      "eval_PRM Loss": 0.4276357591152191,
      "eval_PRM NPV": 0.4782608695652174,
      "eval_PRM Precision": 0.8554216867469879,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9779447317123413,
      "eval_runtime": 56.9661,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 820
    },
    {
      "epoch": 0.07574674201360858,
      "grad_norm": 2.416810091664023,
      "learning_rate": 1.2450348336016907e-06,
      "loss": 0.8267,
      "step": 821
    },
    {
      "epoch": 0.07583900357513551,
      "grad_norm": 2.376343332622679,
      "learning_rate": 1.2450111368594353e-06,
      "loss": 0.9099,
      "step": 822
    },
    {
      "epoch": 0.07593126513666244,
      "grad_norm": 1.5609925923787278,
      "learning_rate": 1.2449873839307946e-06,
      "loss": 0.7421,
      "step": 823
    },
    {
      "epoch": 0.07602352669818936,
      "grad_norm": 2.0250215905922775,
      "learning_rate": 1.2449635748179214e-06,
      "loss": 0.7673,
      "step": 824
    },
    {
      "epoch": 0.0761157882597163,
      "grad_norm": 2.381198063048954,
      "learning_rate": 1.2449397095229733e-06,
      "loss": 0.8431,
      "step": 825
    },
    {
      "epoch": 0.0761157882597163,
      "eval_GEN Loss": 0.48923951387405396,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8606060606060606,
      "eval_PRM F1 AUC": 0.6885804085908852,
      "eval_PRM F1 Neg": 0.5106382978723404,
      "eval_PRM Loss": 0.4137907922267914,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8658536585365854,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9695913195610046,
      "eval_runtime": 56.8733,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 825
    },
    {
      "epoch": 0.07620804982124323,
      "grad_norm": 2.3543419232709706,
      "learning_rate": 1.2449157880481128e-06,
      "loss": 0.7274,
      "step": 826
    },
    {
      "epoch": 0.07630031138277016,
      "grad_norm": 1.5937985167746698,
      "learning_rate": 1.244891810395508e-06,
      "loss": 0.8499,
      "step": 827
    },
    {
      "epoch": 0.07639257294429708,
      "grad_norm": 1.944719113741731,
      "learning_rate": 1.2448677765673314e-06,
      "loss": 0.8374,
      "step": 828
    },
    {
      "epoch": 0.07648483450582401,
      "grad_norm": 2.5598233470175784,
      "learning_rate": 1.2448436865657613e-06,
      "loss": 0.8337,
      "step": 829
    },
    {
      "epoch": 0.07657709606735094,
      "grad_norm": 1.9857330422071102,
      "learning_rate": 1.2448195403929806e-06,
      "loss": 0.8968,
      "step": 830
    },
    {
      "epoch": 0.07657709606735094,
      "eval_GEN Loss": 0.4866617023944855,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.874251497005988,
      "eval_PRM F1 AUC": 0.7006286013619696,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM Loss": 0.40177038311958313,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.8690476190476191,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9986177682876587,
      "eval_runtime": 56.8728,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 830
    },
    {
      "epoch": 0.07666935762887787,
      "grad_norm": 2.135657820144189,
      "learning_rate": 1.2447953380511777e-06,
      "loss": 0.7917,
      "step": 831
    },
    {
      "epoch": 0.07676161919040479,
      "grad_norm": 2.611450488338792,
      "learning_rate": 1.2447710795425455e-06,
      "loss": 0.9619,
      "step": 832
    },
    {
      "epoch": 0.07685388075193172,
      "grad_norm": 1.8469631495600116,
      "learning_rate": 1.2447467648692827e-06,
      "loss": 0.747,
      "step": 833
    },
    {
      "epoch": 0.07694614231345866,
      "grad_norm": 1.6536962967012152,
      "learning_rate": 1.2447223940335923e-06,
      "loss": 0.6887,
      "step": 834
    },
    {
      "epoch": 0.07703840387498559,
      "grad_norm": 2.2203734977559035,
      "learning_rate": 1.2446979670376833e-06,
      "loss": 1.023,
      "step": 835
    },
    {
      "epoch": 0.07703840387498559,
      "eval_GEN Loss": 0.4862686097621918,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.4109131097793579,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9471153616905212,
      "eval_runtime": 56.6381,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 835
    },
    {
      "epoch": 0.07713066543651251,
      "grad_norm": 1.8404547050667157,
      "learning_rate": 1.244673483883769e-06,
      "loss": 0.8868,
      "step": 836
    },
    {
      "epoch": 0.07722292699803944,
      "grad_norm": 1.6246090485457525,
      "learning_rate": 1.2446489445740683e-06,
      "loss": 0.7601,
      "step": 837
    },
    {
      "epoch": 0.07731518855956637,
      "grad_norm": 1.585357583735492,
      "learning_rate": 1.244624349110805e-06,
      "loss": 0.8628,
      "step": 838
    },
    {
      "epoch": 0.0774074501210933,
      "grad_norm": 2.995908072129442,
      "learning_rate": 1.2445996974962078e-06,
      "loss": 0.8864,
      "step": 839
    },
    {
      "epoch": 0.07749971168262022,
      "grad_norm": 2.71225165118734,
      "learning_rate": 1.2445749897325106e-06,
      "loss": 0.8182,
      "step": 840
    },
    {
      "epoch": 0.07749971168262022,
      "eval_GEN Loss": 0.4852362275123596,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8407643312101911,
      "eval_PRM F1 AUC": 0.7236773179675223,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.4437427222728729,
      "eval_PRM NPV": 0.46875,
      "eval_PRM Precision": 0.8918918918918919,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9168870449066162,
      "eval_runtime": 56.7321,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 840
    },
    {
      "epoch": 0.07759197324414716,
      "grad_norm": 2.4546522919332583,
      "learning_rate": 1.2445502258219528e-06,
      "loss": 0.8936,
      "step": 841
    },
    {
      "epoch": 0.07768423480567409,
      "grad_norm": 1.7161177572020176,
      "learning_rate": 1.2445254057667783e-06,
      "loss": 0.8069,
      "step": 842
    },
    {
      "epoch": 0.07777649636720102,
      "grad_norm": 1.5790672985944891,
      "learning_rate": 1.2445005295692364e-06,
      "loss": 0.7169,
      "step": 843
    },
    {
      "epoch": 0.07786875792872794,
      "grad_norm": 2.013552860185453,
      "learning_rate": 1.2444755972315813e-06,
      "loss": 0.8245,
      "step": 844
    },
    {
      "epoch": 0.07796101949025487,
      "grad_norm": 2.1417414885459904,
      "learning_rate": 1.2444506087560725e-06,
      "loss": 0.8883,
      "step": 845
    },
    {
      "epoch": 0.07796101949025487,
      "eval_GEN Loss": 0.4868348240852356,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7452830188679245,
      "eval_PRM F1": 0.8280254777070064,
      "eval_PRM F1 AUC": 0.6959140911471975,
      "eval_PRM F1 Neg": 0.509090909090909,
      "eval_PRM Loss": 0.4720483124256134,
      "eval_PRM NPV": 0.4375,
      "eval_PRM Precision": 0.8783783783783784,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9293269515037537,
      "eval_runtime": 56.9438,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 845
    },
    {
      "epoch": 0.0780532810517818,
      "grad_norm": 1.8469790639648116,
      "learning_rate": 1.2444255641449746e-06,
      "loss": 0.8698,
      "step": 846
    },
    {
      "epoch": 0.07814554261330874,
      "grad_norm": 2.755446596183262,
      "learning_rate": 1.2444004634005572e-06,
      "loss": 0.7381,
      "step": 847
    },
    {
      "epoch": 0.07823780417483565,
      "grad_norm": 2.007426586044603,
      "learning_rate": 1.2443753065250947e-06,
      "loss": 0.696,
      "step": 848
    },
    {
      "epoch": 0.07833006573636259,
      "grad_norm": 1.6971518291669285,
      "learning_rate": 1.2443500935208673e-06,
      "loss": 0.7656,
      "step": 849
    },
    {
      "epoch": 0.07842232729788952,
      "grad_norm": 1.9358908564041117,
      "learning_rate": 1.244324824390159e-06,
      "loss": 0.8341,
      "step": 850
    },
    {
      "epoch": 0.07842232729788952,
      "eval_GEN Loss": 0.4880904257297516,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8354430379746836,
      "eval_PRM F1 AUC": 0.7019381875327397,
      "eval_PRM F1 Neg": 0.5185185185185185,
      "eval_PRM Loss": 0.4468076825141907,
      "eval_PRM NPV": 0.45161290322580644,
      "eval_PRM Precision": 0.88,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.943088948726654,
      "eval_runtime": 56.6384,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 850
    },
    {
      "epoch": 0.07851458885941645,
      "grad_norm": 1.9723772554221473,
      "learning_rate": 1.2442994991352604e-06,
      "loss": 0.7479,
      "step": 851
    },
    {
      "epoch": 0.07860685042094337,
      "grad_norm": 1.5806776439548738,
      "learning_rate": 1.2442741177584669e-06,
      "loss": 0.7837,
      "step": 852
    },
    {
      "epoch": 0.0786991119824703,
      "grad_norm": 3.5942277704877754,
      "learning_rate": 1.2442486802620776e-06,
      "loss": 0.8349,
      "step": 853
    },
    {
      "epoch": 0.07879137354399723,
      "grad_norm": 1.9765224349151842,
      "learning_rate": 1.2442231866483983e-06,
      "loss": 0.6078,
      "step": 854
    },
    {
      "epoch": 0.07888363510552417,
      "grad_norm": 3.481110541921185,
      "learning_rate": 1.2441976369197393e-06,
      "loss": 1.1207,
      "step": 855
    },
    {
      "epoch": 0.07888363510552417,
      "eval_GEN Loss": 0.4901587963104248,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.4273208677768707,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9876201748847961,
      "eval_runtime": 56.9564,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 855
    },
    {
      "epoch": 0.07897589666705108,
      "grad_norm": 1.5284528662332821,
      "learning_rate": 1.2441720310784157e-06,
      "loss": 0.7793,
      "step": 856
    },
    {
      "epoch": 0.07906815822857802,
      "grad_norm": 1.5133521433785049,
      "learning_rate": 1.244146369126748e-06,
      "loss": 0.6137,
      "step": 857
    },
    {
      "epoch": 0.07916041979010495,
      "grad_norm": 2.130285114985017,
      "learning_rate": 1.244120651067062e-06,
      "loss": 0.7553,
      "step": 858
    },
    {
      "epoch": 0.07925268135163188,
      "grad_norm": 2.0300028976770537,
      "learning_rate": 1.2440948769016879e-06,
      "loss": 0.6827,
      "step": 859
    },
    {
      "epoch": 0.0793449429131588,
      "grad_norm": 1.6635017425255172,
      "learning_rate": 1.2440690466329617e-06,
      "loss": 0.7518,
      "step": 860
    },
    {
      "epoch": 0.0793449429131588,
      "eval_GEN Loss": 0.490712970495224,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8518518518518519,
      "eval_PRM F1 AUC": 0.6982713462545836,
      "eval_PRM F1 Neg": 0.52,
      "eval_PRM Loss": 0.4272684156894684,
      "eval_PRM NPV": 0.48148148148148145,
      "eval_PRM Precision": 0.8734177215189873,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9701322317123413,
      "eval_runtime": 56.7977,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 860
    },
    {
      "epoch": 0.07943720447468573,
      "grad_norm": 2.504983146550541,
      "learning_rate": 1.244043160263224e-06,
      "loss": 0.7125,
      "step": 861
    },
    {
      "epoch": 0.07952946603621266,
      "grad_norm": 1.4474241072175298,
      "learning_rate": 1.2440172177948207e-06,
      "loss": 0.5333,
      "step": 862
    },
    {
      "epoch": 0.0796217275977396,
      "grad_norm": 1.6976002637301968,
      "learning_rate": 1.2439912192301027e-06,
      "loss": 0.7309,
      "step": 863
    },
    {
      "epoch": 0.07971398915926652,
      "grad_norm": 2.610885285690451,
      "learning_rate": 1.2439651645714263e-06,
      "loss": 0.9831,
      "step": 864
    },
    {
      "epoch": 0.07980625072079345,
      "grad_norm": 2.2277261457261646,
      "learning_rate": 1.2439390538211522e-06,
      "loss": 0.8019,
      "step": 865
    },
    {
      "epoch": 0.07980625072079345,
      "eval_GEN Loss": 0.49128982424736023,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.43341773748397827,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9430288672447205,
      "eval_runtime": 56.5487,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 865
    },
    {
      "epoch": 0.07989851228232038,
      "grad_norm": 2.0245463264697907,
      "learning_rate": 1.2439128869816468e-06,
      "loss": 0.8017,
      "step": 866
    },
    {
      "epoch": 0.07999077384384731,
      "grad_norm": 1.6616743102186426,
      "learning_rate": 1.2438866640552816e-06,
      "loss": 0.8502,
      "step": 867
    },
    {
      "epoch": 0.08008303540537423,
      "grad_norm": 1.9057241421259494,
      "learning_rate": 1.2438603850444328e-06,
      "loss": 0.8247,
      "step": 868
    },
    {
      "epoch": 0.08017529696690116,
      "grad_norm": 2.440056653745764,
      "learning_rate": 1.2438340499514817e-06,
      "loss": 0.8581,
      "step": 869
    },
    {
      "epoch": 0.0802675585284281,
      "grad_norm": 2.0871783888188493,
      "learning_rate": 1.243807658778815e-06,
      "loss": 0.8389,
      "step": 870
    },
    {
      "epoch": 0.0802675585284281,
      "eval_GEN Loss": 0.4900650382041931,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.42185112833976746,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.939723551273346,
      "eval_runtime": 56.657,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 870
    },
    {
      "epoch": 0.08035982008995503,
      "grad_norm": 2.167709794516737,
      "learning_rate": 1.243781211528824e-06,
      "loss": 0.8505,
      "step": 871
    },
    {
      "epoch": 0.08045208165148195,
      "grad_norm": 2.6216805950295323,
      "learning_rate": 1.243754708203906e-06,
      "loss": 0.9081,
      "step": 872
    },
    {
      "epoch": 0.08054434321300888,
      "grad_norm": 1.5561520009829226,
      "learning_rate": 1.243728148806462e-06,
      "loss": 0.8009,
      "step": 873
    },
    {
      "epoch": 0.08063660477453581,
      "grad_norm": 1.7864929931020381,
      "learning_rate": 1.2437015333388996e-06,
      "loss": 0.7929,
      "step": 874
    },
    {
      "epoch": 0.08072886633606274,
      "grad_norm": 2.4580123714254465,
      "learning_rate": 1.2436748618036302e-06,
      "loss": 0.8353,
      "step": 875
    },
    {
      "epoch": 0.08072886633606274,
      "eval_GEN Loss": 0.49164655804634094,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8427672955974843,
      "eval_PRM F1 AUC": 0.707962283918282,
      "eval_PRM F1 Neg": 0.5283018867924528,
      "eval_PRM Loss": 0.44592949748039246,
      "eval_PRM NPV": 0.4666666666666667,
      "eval_PRM Precision": 0.881578947368421,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9358773827552795,
      "eval_runtime": 56.9863,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 875
    },
    {
      "epoch": 0.08082112789758966,
      "grad_norm": 1.8016659935710604,
      "learning_rate": 1.2436481342030712e-06,
      "loss": 0.6954,
      "step": 876
    },
    {
      "epoch": 0.0809133894591166,
      "grad_norm": 1.6708701334542464,
      "learning_rate": 1.2436213505396445e-06,
      "loss": 0.8919,
      "step": 877
    },
    {
      "epoch": 0.08100565102064353,
      "grad_norm": 1.5697716770569945,
      "learning_rate": 1.243594510815777e-06,
      "loss": 0.9492,
      "step": 878
    },
    {
      "epoch": 0.08109791258217046,
      "grad_norm": 1.7560814536415021,
      "learning_rate": 1.2435676150339016e-06,
      "loss": 0.8383,
      "step": 879
    },
    {
      "epoch": 0.08119017414369738,
      "grad_norm": 2.2993878777535506,
      "learning_rate": 1.243540663196455e-06,
      "loss": 0.7426,
      "step": 880
    },
    {
      "epoch": 0.08119017414369738,
      "eval_GEN Loss": 0.4915597438812256,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8427672955974843,
      "eval_PRM F1 AUC": 0.707962283918282,
      "eval_PRM F1 Neg": 0.5283018867924528,
      "eval_PRM Loss": 0.45518842339515686,
      "eval_PRM NPV": 0.4666666666666667,
      "eval_PRM Precision": 0.881578947368421,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9466345906257629,
      "eval_runtime": 56.7621,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 880
    },
    {
      "epoch": 0.08128243570522431,
      "grad_norm": 2.425938989001129,
      "learning_rate": 1.2435136553058802e-06,
      "loss": 0.9346,
      "step": 881
    },
    {
      "epoch": 0.08137469726675124,
      "grad_norm": 2.1066464670940035,
      "learning_rate": 1.2434865913646242e-06,
      "loss": 0.957,
      "step": 882
    },
    {
      "epoch": 0.08146695882827817,
      "grad_norm": 2.2420383948380653,
      "learning_rate": 1.2434594713751398e-06,
      "loss": 1.0105,
      "step": 883
    },
    {
      "epoch": 0.08155922038980509,
      "grad_norm": 2.1763898304110336,
      "learning_rate": 1.2434322953398846e-06,
      "loss": 0.6929,
      "step": 884
    },
    {
      "epoch": 0.08165148195133203,
      "grad_norm": 1.790863741197223,
      "learning_rate": 1.2434050632613214e-06,
      "loss": 0.6883,
      "step": 885
    },
    {
      "epoch": 0.08165148195133203,
      "eval_GEN Loss": 0.49333953857421875,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8427672955974843,
      "eval_PRM F1 AUC": 0.707962283918282,
      "eval_PRM F1 Neg": 0.5283018867924528,
      "eval_PRM Loss": 0.4510464668273926,
      "eval_PRM NPV": 0.4666666666666667,
      "eval_PRM Precision": 0.881578947368421,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9476562738418579,
      "eval_runtime": 55.8007,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 885
    },
    {
      "epoch": 0.08174374351285896,
      "grad_norm": 1.6524067640391467,
      "learning_rate": 1.243377775141918e-06,
      "loss": 0.7504,
      "step": 886
    },
    {
      "epoch": 0.08183600507438589,
      "grad_norm": 1.7549163735268902,
      "learning_rate": 1.2433504309841475e-06,
      "loss": 0.8734,
      "step": 887
    },
    {
      "epoch": 0.08192826663591281,
      "grad_norm": 1.887443067589934,
      "learning_rate": 1.2433230307904873e-06,
      "loss": 0.6943,
      "step": 888
    },
    {
      "epoch": 0.08202052819743974,
      "grad_norm": 1.5869128030746127,
      "learning_rate": 1.243295574563421e-06,
      "loss": 0.7765,
      "step": 889
    },
    {
      "epoch": 0.08211278975896667,
      "grad_norm": 1.5400141425545875,
      "learning_rate": 1.2432680623054363e-06,
      "loss": 0.7727,
      "step": 890
    },
    {
      "epoch": 0.08211278975896667,
      "eval_GEN Loss": 0.4944790005683899,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4337035119533539,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9912259578704834,
      "eval_runtime": 55.9741,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 890
    },
    {
      "epoch": 0.0822050513204936,
      "grad_norm": 1.979917657565039,
      "learning_rate": 1.2432404940190268e-06,
      "loss": 0.7471,
      "step": 891
    },
    {
      "epoch": 0.08229731288202052,
      "grad_norm": 3.301646440722295,
      "learning_rate": 1.2432128697066906e-06,
      "loss": 0.9991,
      "step": 892
    },
    {
      "epoch": 0.08238957444354746,
      "grad_norm": 1.6350141598657844,
      "learning_rate": 1.243185189370931e-06,
      "loss": 0.722,
      "step": 893
    },
    {
      "epoch": 0.08248183600507439,
      "grad_norm": 3.8710426053518883,
      "learning_rate": 1.2431574530142566e-06,
      "loss": 0.8502,
      "step": 894
    },
    {
      "epoch": 0.08257409756660132,
      "grad_norm": 1.7868172321055178,
      "learning_rate": 1.2431296606391808e-06,
      "loss": 0.64,
      "step": 895
    },
    {
      "epoch": 0.08257409756660132,
      "eval_GEN Loss": 0.4964035153388977,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8606060606060606,
      "eval_PRM F1 AUC": 0.6885804085908852,
      "eval_PRM F1 Neg": 0.5106382978723404,
      "eval_PRM Loss": 0.43730512261390686,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8658536585365854,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 1.0329326391220093,
      "eval_runtime": 56.2022,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 895
    },
    {
      "epoch": 0.08266635912812824,
      "grad_norm": 2.8442898880016956,
      "learning_rate": 1.243101812248222e-06,
      "loss": 0.7484,
      "step": 896
    },
    {
      "epoch": 0.08275862068965517,
      "grad_norm": 1.9643221892455205,
      "learning_rate": 1.2430739078439043e-06,
      "loss": 0.6905,
      "step": 897
    },
    {
      "epoch": 0.0828508822511821,
      "grad_norm": 1.9929314050835263,
      "learning_rate": 1.2430459474287561e-06,
      "loss": 0.9033,
      "step": 898
    },
    {
      "epoch": 0.08294314381270904,
      "grad_norm": 2.12943498187798,
      "learning_rate": 1.2430179310053115e-06,
      "loss": 0.9438,
      "step": 899
    },
    {
      "epoch": 0.08303540537423595,
      "grad_norm": 2.636941537564535,
      "learning_rate": 1.2429898585761089e-06,
      "loss": 0.8624,
      "step": 900
    },
    {
      "epoch": 0.08303540537423595,
      "eval_GEN Loss": 0.49784040451049805,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.43831467628479004,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9543870091438293,
      "eval_runtime": 55.8343,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 900
    },
    {
      "epoch": 0.08312766693576289,
      "grad_norm": 2.4751278736954565,
      "learning_rate": 1.2429617301436927e-06,
      "loss": 0.8424,
      "step": 901
    },
    {
      "epoch": 0.08321992849728982,
      "grad_norm": 2.4505556014641576,
      "learning_rate": 1.242933545710612e-06,
      "loss": 0.9397,
      "step": 902
    },
    {
      "epoch": 0.08331219005881675,
      "grad_norm": 2.168359211408005,
      "learning_rate": 1.2429053052794208e-06,
      "loss": 0.9011,
      "step": 903
    },
    {
      "epoch": 0.08340445162034367,
      "grad_norm": 1.8215197239340088,
      "learning_rate": 1.2428770088526778e-06,
      "loss": 0.8597,
      "step": 904
    },
    {
      "epoch": 0.0834967131818706,
      "grad_norm": 1.4998595951461988,
      "learning_rate": 1.242848656432948e-06,
      "loss": 0.7586,
      "step": 905
    },
    {
      "epoch": 0.0834967131818706,
      "eval_GEN Loss": 0.4991404116153717,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.442852646112442,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9323317408561707,
      "eval_runtime": 55.809,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 905
    },
    {
      "epoch": 0.08358897474339753,
      "grad_norm": 2.448892605638469,
      "learning_rate": 1.2428202480228004e-06,
      "loss": 0.7082,
      "step": 906
    },
    {
      "epoch": 0.08368123630492447,
      "grad_norm": 2.0823511958510625,
      "learning_rate": 1.2427917836248095e-06,
      "loss": 0.8262,
      "step": 907
    },
    {
      "epoch": 0.08377349786645139,
      "grad_norm": 2.4923033408350985,
      "learning_rate": 1.2427632632415548e-06,
      "loss": 0.897,
      "step": 908
    },
    {
      "epoch": 0.08386575942797832,
      "grad_norm": 1.9198283211883294,
      "learning_rate": 1.2427346868756207e-06,
      "loss": 0.6094,
      "step": 909
    },
    {
      "epoch": 0.08395802098950525,
      "grad_norm": 2.0695183395171526,
      "learning_rate": 1.242706054529597e-06,
      "loss": 0.8529,
      "step": 910
    },
    {
      "epoch": 0.08395802098950525,
      "eval_GEN Loss": 0.4996287524700165,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.42069101333618164,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9454928040504456,
      "eval_runtime": 55.8764,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 910
    },
    {
      "epoch": 0.08405028255103218,
      "grad_norm": 1.778329862058544,
      "learning_rate": 1.2426773662060785e-06,
      "loss": 0.9364,
      "step": 911
    },
    {
      "epoch": 0.0841425441125591,
      "grad_norm": 1.4975012988149345,
      "learning_rate": 1.2426486219076647e-06,
      "loss": 0.7888,
      "step": 912
    },
    {
      "epoch": 0.08423480567408603,
      "grad_norm": 3.5460235175487056,
      "learning_rate": 1.2426198216369605e-06,
      "loss": 0.9086,
      "step": 913
    },
    {
      "epoch": 0.08432706723561297,
      "grad_norm": 1.4721525607859052,
      "learning_rate": 1.2425909653965759e-06,
      "loss": 0.7795,
      "step": 914
    },
    {
      "epoch": 0.0844193287971399,
      "grad_norm": 2.1330276691274728,
      "learning_rate": 1.242562053189126e-06,
      "loss": 0.7734,
      "step": 915
    },
    {
      "epoch": 0.0844193287971399,
      "eval_GEN Loss": 0.4994935095310211,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.4029008150100708,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9498196840286255,
      "eval_runtime": 55.8242,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 915
    },
    {
      "epoch": 0.08451159035866682,
      "grad_norm": 1.4857650700445841,
      "learning_rate": 1.2425330850172308e-06,
      "loss": 0.5915,
      "step": 916
    },
    {
      "epoch": 0.08460385192019375,
      "grad_norm": 2.5305604505637347,
      "learning_rate": 1.2425040608835154e-06,
      "loss": 0.8846,
      "step": 917
    },
    {
      "epoch": 0.08469611348172068,
      "grad_norm": 1.6612318143106672,
      "learning_rate": 1.24247498079061e-06,
      "loss": 0.7023,
      "step": 918
    },
    {
      "epoch": 0.08478837504324761,
      "grad_norm": 1.7376350497879325,
      "learning_rate": 1.24244584474115e-06,
      "loss": 0.8275,
      "step": 919
    },
    {
      "epoch": 0.08488063660477453,
      "grad_norm": 1.540593975695022,
      "learning_rate": 1.2424166527377757e-06,
      "loss": 0.7477,
      "step": 920
    },
    {
      "epoch": 0.08488063660477453,
      "eval_GEN Loss": 0.500395655632019,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.39217761158943176,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9426081776618958,
      "eval_runtime": 55.8493,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 920
    },
    {
      "epoch": 0.08497289816630146,
      "grad_norm": 2.9010946783351184,
      "learning_rate": 1.2423874047831326e-06,
      "loss": 0.9336,
      "step": 921
    },
    {
      "epoch": 0.0850651597278284,
      "grad_norm": 2.4788366402222763,
      "learning_rate": 1.2423581008798707e-06,
      "loss": 0.8536,
      "step": 922
    },
    {
      "epoch": 0.08515742128935533,
      "grad_norm": 1.7633558621672578,
      "learning_rate": 1.2423287410306462e-06,
      "loss": 0.7852,
      "step": 923
    },
    {
      "epoch": 0.08524968285088225,
      "grad_norm": 1.5975516766632853,
      "learning_rate": 1.2422993252381195e-06,
      "loss": 0.8795,
      "step": 924
    },
    {
      "epoch": 0.08534194441240918,
      "grad_norm": 1.9115089431244583,
      "learning_rate": 1.2422698535049563e-06,
      "loss": 0.8315,
      "step": 925
    },
    {
      "epoch": 0.08534194441240918,
      "eval_GEN Loss": 0.49908775091171265,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.39852482080459595,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9270432591438293,
      "eval_runtime": 55.8449,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 925
    },
    {
      "epoch": 0.08543420597393611,
      "grad_norm": 2.376237572356329,
      "learning_rate": 1.2422403258338275e-06,
      "loss": 0.9664,
      "step": 926
    },
    {
      "epoch": 0.08552646753546304,
      "grad_norm": 1.7224706922713144,
      "learning_rate": 1.2422107422274086e-06,
      "loss": 0.9022,
      "step": 927
    },
    {
      "epoch": 0.08561872909698996,
      "grad_norm": 1.7754361881681706,
      "learning_rate": 1.242181102688381e-06,
      "loss": 0.9474,
      "step": 928
    },
    {
      "epoch": 0.0857109906585169,
      "grad_norm": 2.44837687350161,
      "learning_rate": 1.2421514072194302e-06,
      "loss": 0.914,
      "step": 929
    },
    {
      "epoch": 0.08580325222004383,
      "grad_norm": 2.06445320624774,
      "learning_rate": 1.2421216558232474e-06,
      "loss": 0.9022,
      "step": 930
    },
    {
      "epoch": 0.08580325222004383,
      "eval_GEN Loss": 0.49789413809776306,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.4136628806591034,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9203726053237915,
      "eval_runtime": 55.885,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 930
    },
    {
      "epoch": 0.08589551378157076,
      "grad_norm": 2.694014836265426,
      "learning_rate": 1.242091848502529e-06,
      "loss": 0.8918,
      "step": 931
    },
    {
      "epoch": 0.08598777534309768,
      "grad_norm": 1.6161159496644908,
      "learning_rate": 1.2420619852599758e-06,
      "loss": 0.9424,
      "step": 932
    },
    {
      "epoch": 0.08608003690462461,
      "grad_norm": 2.2565358418890367,
      "learning_rate": 1.2420320660982944e-06,
      "loss": 0.7794,
      "step": 933
    },
    {
      "epoch": 0.08617229846615154,
      "grad_norm": 2.1082931556063547,
      "learning_rate": 1.2420020910201958e-06,
      "loss": 0.8659,
      "step": 934
    },
    {
      "epoch": 0.08626456002767847,
      "grad_norm": 1.616588521436696,
      "learning_rate": 1.2419720600283966e-06,
      "loss": 0.8662,
      "step": 935
    },
    {
      "epoch": 0.08626456002767847,
      "eval_GEN Loss": 0.4949224293231964,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM Loss": 0.42087623476982117,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9172475934028625,
      "eval_runtime": 55.6459,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 935
    },
    {
      "epoch": 0.0863568215892054,
      "grad_norm": 2.2089355564102737,
      "learning_rate": 1.2419419731256182e-06,
      "loss": 0.753,
      "step": 936
    },
    {
      "epoch": 0.08644908315073233,
      "grad_norm": 1.693462533837462,
      "learning_rate": 1.241911830314587e-06,
      "loss": 0.636,
      "step": 937
    },
    {
      "epoch": 0.08654134471225926,
      "grad_norm": 2.3068665088424654,
      "learning_rate": 1.2418816315980348e-06,
      "loss": 0.6942,
      "step": 938
    },
    {
      "epoch": 0.08663360627378619,
      "grad_norm": 1.5593779345836822,
      "learning_rate": 1.2418513769786982e-06,
      "loss": 0.8544,
      "step": 939
    },
    {
      "epoch": 0.08672586783531311,
      "grad_norm": 2.0972921187878755,
      "learning_rate": 1.241821066459319e-06,
      "loss": 0.817,
      "step": 940
    },
    {
      "epoch": 0.08672586783531311,
      "eval_GEN Loss": 0.49334296584129333,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.40967661142349243,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9105167984962463,
      "eval_runtime": 55.8201,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 940
    },
    {
      "epoch": 0.08681812939684004,
      "grad_norm": 3.396534800564875,
      "learning_rate": 1.2417907000426435e-06,
      "loss": 1.0753,
      "step": 941
    },
    {
      "epoch": 0.08691039095836697,
      "grad_norm": 1.4537018426734836,
      "learning_rate": 1.241760277731424e-06,
      "loss": 0.8919,
      "step": 942
    },
    {
      "epoch": 0.0870026525198939,
      "grad_norm": 2.1880471890812188,
      "learning_rate": 1.2417297995284176e-06,
      "loss": 0.9027,
      "step": 943
    },
    {
      "epoch": 0.08709491408142082,
      "grad_norm": 2.1748698411085376,
      "learning_rate": 1.241699265436386e-06,
      "loss": 0.9034,
      "step": 944
    },
    {
      "epoch": 0.08718717564294776,
      "grad_norm": 2.0742997899946753,
      "learning_rate": 1.241668675458096e-06,
      "loss": 0.9607,
      "step": 945
    },
    {
      "epoch": 0.08718717564294776,
      "eval_GEN Loss": 0.49398431181907654,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.40069958567619324,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9083533883094788,
      "eval_runtime": 55.8051,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 945
    },
    {
      "epoch": 0.08727943720447469,
      "grad_norm": 1.68528997969868,
      "learning_rate": 1.2416380295963202e-06,
      "loss": 0.7965,
      "step": 946
    },
    {
      "epoch": 0.08737169876600162,
      "grad_norm": 1.6895554469366851,
      "learning_rate": 1.2416073278538354e-06,
      "loss": 0.6948,
      "step": 947
    },
    {
      "epoch": 0.08746396032752854,
      "grad_norm": 1.863158177970021,
      "learning_rate": 1.2415765702334239e-06,
      "loss": 0.8556,
      "step": 948
    },
    {
      "epoch": 0.08755622188905547,
      "grad_norm": 1.8045272780669495,
      "learning_rate": 1.2415457567378735e-06,
      "loss": 0.7606,
      "step": 949
    },
    {
      "epoch": 0.0876484834505824,
      "grad_norm": 2.46053200857306,
      "learning_rate": 1.241514887369976e-06,
      "loss": 0.6749,
      "step": 950
    },
    {
      "epoch": 0.0876484834505824,
      "eval_GEN Loss": 0.4946034848690033,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.40132224559783936,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9060096144676208,
      "eval_runtime": 55.8463,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 950
    },
    {
      "epoch": 0.08774074501210934,
      "grad_norm": 1.7453560866782651,
      "learning_rate": 1.2414839621325288e-06,
      "loss": 0.8106,
      "step": 951
    },
    {
      "epoch": 0.08783300657363625,
      "grad_norm": 2.327415032945926,
      "learning_rate": 1.2414529810283348e-06,
      "loss": 1.0471,
      "step": 952
    },
    {
      "epoch": 0.08792526813516319,
      "grad_norm": 1.5499009356140918,
      "learning_rate": 1.2414219440602012e-06,
      "loss": 0.7972,
      "step": 953
    },
    {
      "epoch": 0.08801752969669012,
      "grad_norm": 2.675616280773306,
      "learning_rate": 1.2413908512309408e-06,
      "loss": 0.8065,
      "step": 954
    },
    {
      "epoch": 0.08810979125821705,
      "grad_norm": 2.68357802713316,
      "learning_rate": 1.2413597025433712e-06,
      "loss": 0.9762,
      "step": 955
    },
    {
      "epoch": 0.08810979125821705,
      "eval_GEN Loss": 0.49337396025657654,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.405143678188324,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9100961685180664,
      "eval_runtime": 55.9096,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 955
    },
    {
      "epoch": 0.08820205281974397,
      "grad_norm": 1.704720640234973,
      "learning_rate": 1.2413284980003154e-06,
      "loss": 0.7031,
      "step": 956
    },
    {
      "epoch": 0.0882943143812709,
      "grad_norm": 2.1369641797328924,
      "learning_rate": 1.241297237604601e-06,
      "loss": 1.029,
      "step": 957
    },
    {
      "epoch": 0.08838657594279783,
      "grad_norm": 2.0579810955328384,
      "learning_rate": 1.2412659213590607e-06,
      "loss": 0.9636,
      "step": 958
    },
    {
      "epoch": 0.08847883750432477,
      "grad_norm": 1.2458359657878737,
      "learning_rate": 1.2412345492665326e-06,
      "loss": 0.582,
      "step": 959
    },
    {
      "epoch": 0.08857109906585169,
      "grad_norm": 1.906453396979652,
      "learning_rate": 1.2412031213298597e-06,
      "loss": 0.9057,
      "step": 960
    },
    {
      "epoch": 0.08857109906585169,
      "eval_GEN Loss": 0.49197861552238464,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4226519763469696,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.914843738079071,
      "eval_runtime": 55.9482,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 960
    },
    {
      "epoch": 0.08866336062737862,
      "grad_norm": 2.544046625569359,
      "learning_rate": 1.24117163755189e-06,
      "loss": 0.7873,
      "step": 961
    },
    {
      "epoch": 0.08875562218890555,
      "grad_norm": 2.315145397439504,
      "learning_rate": 1.2411400979354769e-06,
      "loss": 0.8408,
      "step": 962
    },
    {
      "epoch": 0.08884788375043248,
      "grad_norm": 1.4001561441990045,
      "learning_rate": 1.241108502483478e-06,
      "loss": 0.6308,
      "step": 963
    },
    {
      "epoch": 0.0889401453119594,
      "grad_norm": 2.018304603971791,
      "learning_rate": 1.2410768511987568e-06,
      "loss": 0.654,
      "step": 964
    },
    {
      "epoch": 0.08903240687348633,
      "grad_norm": 1.914867302083205,
      "learning_rate": 1.2410451440841817e-06,
      "loss": 0.7151,
      "step": 965
    },
    {
      "epoch": 0.08903240687348633,
      "eval_GEN Loss": 0.493167906999588,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4310265779495239,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.926682710647583,
      "eval_runtime": 55.9206,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 965
    },
    {
      "epoch": 0.08912466843501327,
      "grad_norm": 1.5606454945037835,
      "learning_rate": 1.241013381142626e-06,
      "loss": 0.8283,
      "step": 966
    },
    {
      "epoch": 0.0892169299965402,
      "grad_norm": 5.223071926553168,
      "learning_rate": 1.240981562376968e-06,
      "loss": 0.7292,
      "step": 967
    },
    {
      "epoch": 0.08930919155806712,
      "grad_norm": 1.5113562039472168,
      "learning_rate": 1.2409496877900914e-06,
      "loss": 0.7259,
      "step": 968
    },
    {
      "epoch": 0.08940145311959405,
      "grad_norm": 2.5214760068160142,
      "learning_rate": 1.2409177573848846e-06,
      "loss": 0.8092,
      "step": 969
    },
    {
      "epoch": 0.08949371468112098,
      "grad_norm": 2.0447430483082933,
      "learning_rate": 1.2408857711642412e-06,
      "loss": 0.9398,
      "step": 970
    },
    {
      "epoch": 0.08949371468112098,
      "eval_GEN Loss": 0.49091872572898865,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.42805832624435425,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9384615421295166,
      "eval_runtime": 55.9109,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 970
    },
    {
      "epoch": 0.08958597624264791,
      "grad_norm": 1.570560978823499,
      "learning_rate": 1.2408537291310595e-06,
      "loss": 0.7814,
      "step": 971
    },
    {
      "epoch": 0.08967823780417483,
      "grad_norm": 2.2696024022161714,
      "learning_rate": 1.2408216312882437e-06,
      "loss": 0.9953,
      "step": 972
    },
    {
      "epoch": 0.08977049936570176,
      "grad_norm": 1.4593275503611438,
      "learning_rate": 1.2407894776387023e-06,
      "loss": 0.7604,
      "step": 973
    },
    {
      "epoch": 0.0898627609272287,
      "grad_norm": 2.3007610096129274,
      "learning_rate": 1.240757268185349e-06,
      "loss": 0.9001,
      "step": 974
    },
    {
      "epoch": 0.08995502248875563,
      "grad_norm": 2.2448516620611816,
      "learning_rate": 1.2407250029311032e-06,
      "loss": 1.0175,
      "step": 975
    },
    {
      "epoch": 0.08995502248875563,
      "eval_GEN Loss": 0.49143147468566895,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8461538461538461,
      "eval_PRM F1 AUC": 0.7454164484023049,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.46443119645118713,
      "eval_PRM NPV": 0.48484848484848486,
      "eval_PRM Precision": 0.9041095890410958,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9454326629638672,
      "eval_runtime": 55.7019,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 975
    },
    {
      "epoch": 0.09004728405028255,
      "grad_norm": 2.210950991950348,
      "learning_rate": 1.2406926818788884e-06,
      "loss": 0.8523,
      "step": 976
    },
    {
      "epoch": 0.09013954561180948,
      "grad_norm": 1.6662669941233985,
      "learning_rate": 1.2406603050316334e-06,
      "loss": 0.7193,
      "step": 977
    },
    {
      "epoch": 0.09023180717333641,
      "grad_norm": 2.6173364182712073,
      "learning_rate": 1.2406278723922727e-06,
      "loss": 0.9668,
      "step": 978
    },
    {
      "epoch": 0.09032406873486334,
      "grad_norm": 2.578586937335509,
      "learning_rate": 1.2405953839637453e-06,
      "loss": 0.9479,
      "step": 979
    },
    {
      "epoch": 0.09041633029639026,
      "grad_norm": 3.7347857967613374,
      "learning_rate": 1.240562839748995e-06,
      "loss": 0.8276,
      "step": 980
    },
    {
      "epoch": 0.09041633029639026,
      "eval_GEN Loss": 0.4923248887062073,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8461538461538461,
      "eval_PRM F1 AUC": 0.7454164484023049,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4941073954105377,
      "eval_PRM NPV": 0.48484848484848486,
      "eval_PRM Precision": 0.9041095890410958,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.96484375,
      "eval_runtime": 56.725,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 980
    },
    {
      "epoch": 0.0905085918579172,
      "grad_norm": 2.638901943907216,
      "learning_rate": 1.2405302397509713e-06,
      "loss": 0.7783,
      "step": 981
    },
    {
      "epoch": 0.09060085341944413,
      "grad_norm": 2.1376484634956294,
      "learning_rate": 1.2404975839726285e-06,
      "loss": 0.9446,
      "step": 982
    },
    {
      "epoch": 0.09069311498097106,
      "grad_norm": 1.9396715223441883,
      "learning_rate": 1.2404648724169258e-06,
      "loss": 0.9514,
      "step": 983
    },
    {
      "epoch": 0.09078537654249798,
      "grad_norm": 1.7578422227243191,
      "learning_rate": 1.2404321050868276e-06,
      "loss": 0.7279,
      "step": 984
    },
    {
      "epoch": 0.09087763810402491,
      "grad_norm": 1.8642313172315839,
      "learning_rate": 1.2403992819853033e-06,
      "loss": 0.7766,
      "step": 985
    },
    {
      "epoch": 0.09087763810402491,
      "eval_GEN Loss": 0.4914352297782898,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.45686113834381104,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9616586565971375,
      "eval_runtime": 56.8705,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 985
    },
    {
      "epoch": 0.09096989966555184,
      "grad_norm": 1.9382213328481714,
      "learning_rate": 1.2403664031153274e-06,
      "loss": 0.9468,
      "step": 986
    },
    {
      "epoch": 0.09106216122707878,
      "grad_norm": 1.7634110586373042,
      "learning_rate": 1.2403334684798796e-06,
      "loss": 0.9205,
      "step": 987
    },
    {
      "epoch": 0.0911544227886057,
      "grad_norm": 1.533406115187923,
      "learning_rate": 1.2403004780819441e-06,
      "loss": 0.8005,
      "step": 988
    },
    {
      "epoch": 0.09124668435013263,
      "grad_norm": 2.0331736276058496,
      "learning_rate": 1.240267431924511e-06,
      "loss": 0.8553,
      "step": 989
    },
    {
      "epoch": 0.09133894591165956,
      "grad_norm": 1.5357846151003822,
      "learning_rate": 1.2402343300105746e-06,
      "loss": 0.7107,
      "step": 990
    },
    {
      "epoch": 0.09133894591165956,
      "eval_GEN Loss": 0.49173447489738464,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4222582280635834,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9712740182876587,
      "eval_runtime": 56.771,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 990
    },
    {
      "epoch": 0.09143120747318649,
      "grad_norm": 1.786324163487861,
      "learning_rate": 1.2402011723431348e-06,
      "loss": 0.7631,
      "step": 991
    },
    {
      "epoch": 0.09152346903471341,
      "grad_norm": 1.2584966725554103,
      "learning_rate": 1.2401679589251967e-06,
      "loss": 0.7694,
      "step": 992
    },
    {
      "epoch": 0.09161573059624034,
      "grad_norm": 1.7125418298951514,
      "learning_rate": 1.2401346897597696e-06,
      "loss": 0.7168,
      "step": 993
    },
    {
      "epoch": 0.09170799215776727,
      "grad_norm": 4.6677885891605255,
      "learning_rate": 1.2401013648498687e-06,
      "loss": 1.0056,
      "step": 994
    },
    {
      "epoch": 0.0918002537192942,
      "grad_norm": 2.096807475712761,
      "learning_rate": 1.240067984198514e-06,
      "loss": 1.0276,
      "step": 995
    },
    {
      "epoch": 0.0918002537192942,
      "eval_GEN Loss": 0.494640052318573,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4195377826690674,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9857572317123413,
      "eval_runtime": 56.692,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 995
    },
    {
      "epoch": 0.09189251528082112,
      "grad_norm": 1.3429032844712543,
      "learning_rate": 1.2400345478087306e-06,
      "loss": 0.736,
      "step": 996
    },
    {
      "epoch": 0.09198477684234806,
      "grad_norm": 1.5727366523808957,
      "learning_rate": 1.240001055683548e-06,
      "loss": 0.7833,
      "step": 997
    },
    {
      "epoch": 0.09207703840387499,
      "grad_norm": 1.3380310355741822,
      "learning_rate": 1.2399675078260021e-06,
      "loss": 0.6486,
      "step": 998
    },
    {
      "epoch": 0.09216929996540192,
      "grad_norm": 1.6976269992176125,
      "learning_rate": 1.2399339042391326e-06,
      "loss": 0.8231,
      "step": 999
    },
    {
      "epoch": 0.09226156152692884,
      "grad_norm": 1.3615470809281198,
      "learning_rate": 1.2399002449259845e-06,
      "loss": 0.7108,
      "step": 1000
    },
    {
      "epoch": 0.09226156152692884,
      "eval_GEN Loss": 0.49521714448928833,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8606060606060606,
      "eval_PRM F1 AUC": 0.6885804085908852,
      "eval_PRM F1 Neg": 0.5106382978723404,
      "eval_PRM Loss": 0.42610013484954834,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8658536585365854,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9921875,
      "eval_runtime": 56.8655,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1000
    },
    {
      "epoch": 0.09235382308845577,
      "grad_norm": 3.1917917311136788,
      "learning_rate": 1.2398665298896086e-06,
      "loss": 0.9791,
      "step": 1001
    },
    {
      "epoch": 0.0924460846499827,
      "grad_norm": 1.2435109862977927,
      "learning_rate": 1.2398327591330599e-06,
      "loss": 0.6088,
      "step": 1002
    },
    {
      "epoch": 0.09253834621150964,
      "grad_norm": 1.6375665664287087,
      "learning_rate": 1.2397989326593988e-06,
      "loss": 0.7379,
      "step": 1003
    },
    {
      "epoch": 0.09263060777303656,
      "grad_norm": 1.8619757019705074,
      "learning_rate": 1.2397650504716907e-06,
      "loss": 0.9372,
      "step": 1004
    },
    {
      "epoch": 0.09272286933456349,
      "grad_norm": 2.0115412373645825,
      "learning_rate": 1.239731112573006e-06,
      "loss": 0.8765,
      "step": 1005
    },
    {
      "epoch": 0.09272286933456349,
      "eval_GEN Loss": 0.4953429698944092,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4615454077720642,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9923076629638672,
      "eval_runtime": 56.4484,
      "eval_samples_per_second": 1.151,
      "eval_steps_per_second": 0.089,
      "step": 1005
    },
    {
      "epoch": 0.09281513089609042,
      "grad_norm": 1.8999889847203806,
      "learning_rate": 1.2396971189664202e-06,
      "loss": 0.8761,
      "step": 1006
    },
    {
      "epoch": 0.09290739245761735,
      "grad_norm": 1.5805870123343921,
      "learning_rate": 1.2396630696550143e-06,
      "loss": 0.5916,
      "step": 1007
    },
    {
      "epoch": 0.09299965401914427,
      "grad_norm": 1.7449783247013633,
      "learning_rate": 1.2396289646418733e-06,
      "loss": 0.7875,
      "step": 1008
    },
    {
      "epoch": 0.0930919155806712,
      "grad_norm": 1.6141993052024273,
      "learning_rate": 1.2395948039300882e-06,
      "loss": 0.7338,
      "step": 1009
    },
    {
      "epoch": 0.09318417714219814,
      "grad_norm": 1.8277815860684463,
      "learning_rate": 1.2395605875227542e-06,
      "loss": 0.7331,
      "step": 1010
    },
    {
      "epoch": 0.09318417714219814,
      "eval_GEN Loss": 0.4921077489852905,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8407643312101911,
      "eval_PRM F1 AUC": 0.7236773179675223,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.46621182560920715,
      "eval_PRM NPV": 0.46875,
      "eval_PRM Precision": 0.8918918918918919,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9893629550933838,
      "eval_runtime": 55.7417,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 1010
    },
    {
      "epoch": 0.09327643870372505,
      "grad_norm": 1.856528550567078,
      "learning_rate": 1.239526315422973e-06,
      "loss": 0.744,
      "step": 1011
    },
    {
      "epoch": 0.09336870026525199,
      "grad_norm": 1.5903192366250005,
      "learning_rate": 1.2394919876338494e-06,
      "loss": 0.8565,
      "step": 1012
    },
    {
      "epoch": 0.09346096182677892,
      "grad_norm": 2.241953738337567,
      "learning_rate": 1.2394576041584946e-06,
      "loss": 0.8878,
      "step": 1013
    },
    {
      "epoch": 0.09355322338830585,
      "grad_norm": 1.701638442750919,
      "learning_rate": 1.239423165000025e-06,
      "loss": 0.8068,
      "step": 1014
    },
    {
      "epoch": 0.09364548494983277,
      "grad_norm": 2.8186542530982095,
      "learning_rate": 1.2393886701615607e-06,
      "loss": 0.8443,
      "step": 1015
    },
    {
      "epoch": 0.09364548494983277,
      "eval_GEN Loss": 0.4892708659172058,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.46163588762283325,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9852163195610046,
      "eval_runtime": 55.7734,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 1015
    },
    {
      "epoch": 0.0937377465113597,
      "grad_norm": 1.7717725144662118,
      "learning_rate": 1.239354119646228e-06,
      "loss": 0.8047,
      "step": 1016
    },
    {
      "epoch": 0.09383000807288663,
      "grad_norm": 1.8512229970673009,
      "learning_rate": 1.2393195134571582e-06,
      "loss": 0.6764,
      "step": 1017
    },
    {
      "epoch": 0.09392226963441357,
      "grad_norm": 2.54185236358672,
      "learning_rate": 1.239284851597487e-06,
      "loss": 1.0241,
      "step": 1018
    },
    {
      "epoch": 0.09401453119594048,
      "grad_norm": 2.8263319126756703,
      "learning_rate": 1.2392501340703557e-06,
      "loss": 0.8761,
      "step": 1019
    },
    {
      "epoch": 0.09410679275746742,
      "grad_norm": 2.0255142275269042,
      "learning_rate": 1.2392153608789103e-06,
      "loss": 0.9198,
      "step": 1020
    },
    {
      "epoch": 0.09410679275746742,
      "eval_GEN Loss": 0.4873213768005371,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.46061861515045166,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9809495210647583,
      "eval_runtime": 55.9561,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1020
    },
    {
      "epoch": 0.09419905431899435,
      "grad_norm": 2.6503757314184373,
      "learning_rate": 1.2391805320263023e-06,
      "loss": 0.7316,
      "step": 1021
    },
    {
      "epoch": 0.09429131588052128,
      "grad_norm": 1.7216110902862465,
      "learning_rate": 1.2391456475156876e-06,
      "loss": 0.8213,
      "step": 1022
    },
    {
      "epoch": 0.0943835774420482,
      "grad_norm": 1.9917097770601746,
      "learning_rate": 1.239110707350228e-06,
      "loss": 0.5839,
      "step": 1023
    },
    {
      "epoch": 0.09447583900357513,
      "grad_norm": 2.115845903684835,
      "learning_rate": 1.2390757115330889e-06,
      "loss": 0.8723,
      "step": 1024
    },
    {
      "epoch": 0.09456810056510206,
      "grad_norm": 1.985374323924196,
      "learning_rate": 1.2390406600674427e-06,
      "loss": 0.76,
      "step": 1025
    },
    {
      "epoch": 0.09456810056510206,
      "eval_GEN Loss": 0.4882165491580963,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.45601755380630493,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9912259578704834,
      "eval_runtime": 55.8091,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 1025
    },
    {
      "epoch": 0.094660362126629,
      "grad_norm": 2.0338184228229776,
      "learning_rate": 1.239005552956465e-06,
      "loss": 0.7357,
      "step": 1026
    },
    {
      "epoch": 0.09475262368815592,
      "grad_norm": 1.6369499769315794,
      "learning_rate": 1.238970390203338e-06,
      "loss": 0.8611,
      "step": 1027
    },
    {
      "epoch": 0.09484488524968285,
      "grad_norm": 2.3335893973512776,
      "learning_rate": 1.2389351718112476e-06,
      "loss": 1.027,
      "step": 1028
    },
    {
      "epoch": 0.09493714681120978,
      "grad_norm": 2.1131556795739788,
      "learning_rate": 1.2388998977833857e-06,
      "loss": 0.6919,
      "step": 1029
    },
    {
      "epoch": 0.09502940837273671,
      "grad_norm": 3.9803852639076336,
      "learning_rate": 1.2388645681229486e-06,
      "loss": 0.9372,
      "step": 1030
    },
    {
      "epoch": 0.09502940837273671,
      "eval_GEN Loss": 0.4888576865196228,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.45159152150154114,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9866586327552795,
      "eval_runtime": 55.8556,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1030
    },
    {
      "epoch": 0.09512166993426363,
      "grad_norm": 2.674260347679863,
      "learning_rate": 1.2388291828331383e-06,
      "loss": 0.6835,
      "step": 1031
    },
    {
      "epoch": 0.09521393149579056,
      "grad_norm": 1.4120791571535167,
      "learning_rate": 1.238793741917161e-06,
      "loss": 0.7706,
      "step": 1032
    },
    {
      "epoch": 0.0953061930573175,
      "grad_norm": 1.7594840517952317,
      "learning_rate": 1.238758245378229e-06,
      "loss": 0.7101,
      "step": 1033
    },
    {
      "epoch": 0.09539845461884443,
      "grad_norm": 2.4218671387046653,
      "learning_rate": 1.2387226932195587e-06,
      "loss": 0.8659,
      "step": 1034
    },
    {
      "epoch": 0.09549071618037135,
      "grad_norm": 1.7849933037963401,
      "learning_rate": 1.2386870854443719e-06,
      "loss": 0.8784,
      "step": 1035
    },
    {
      "epoch": 0.09549071618037135,
      "eval_GEN Loss": 0.48858630657196045,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.46425315737724304,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9759014248847961,
      "eval_runtime": 55.7373,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 1035
    },
    {
      "epoch": 0.09558297774189828,
      "grad_norm": 2.115845187930869,
      "learning_rate": 1.2386514220558953e-06,
      "loss": 0.6927,
      "step": 1036
    },
    {
      "epoch": 0.09567523930342521,
      "grad_norm": 1.6759229154986803,
      "learning_rate": 1.2386157030573608e-06,
      "loss": 0.7528,
      "step": 1037
    },
    {
      "epoch": 0.09576750086495214,
      "grad_norm": 1.7471625660255927,
      "learning_rate": 1.238579928452006e-06,
      "loss": 0.8052,
      "step": 1038
    },
    {
      "epoch": 0.09585976242647906,
      "grad_norm": 1.8974624367917863,
      "learning_rate": 1.238544098243072e-06,
      "loss": 0.8325,
      "step": 1039
    },
    {
      "epoch": 0.095952023988006,
      "grad_norm": 2.272234491171637,
      "learning_rate": 1.2385082124338058e-06,
      "loss": 0.8701,
      "step": 1040
    },
    {
      "epoch": 0.095952023988006,
      "eval_GEN Loss": 0.4879036843776703,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.46317258477211,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9808894395828247,
      "eval_runtime": 55.8278,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1040
    },
    {
      "epoch": 0.09604428554953293,
      "grad_norm": 1.9479029261980405,
      "learning_rate": 1.23847227102746e-06,
      "loss": 0.8885,
      "step": 1041
    },
    {
      "epoch": 0.09613654711105986,
      "grad_norm": 1.5555454438192333,
      "learning_rate": 1.2384362740272913e-06,
      "loss": 0.6559,
      "step": 1042
    },
    {
      "epoch": 0.09622880867258678,
      "grad_norm": 1.5042057654066376,
      "learning_rate": 1.238400221436562e-06,
      "loss": 0.7191,
      "step": 1043
    },
    {
      "epoch": 0.09632107023411371,
      "grad_norm": 1.897573278510063,
      "learning_rate": 1.238364113258539e-06,
      "loss": 0.8611,
      "step": 1044
    },
    {
      "epoch": 0.09641333179564064,
      "grad_norm": 1.6709830910703745,
      "learning_rate": 1.2383279494964948e-06,
      "loss": 0.746,
      "step": 1045
    },
    {
      "epoch": 0.09641333179564064,
      "eval_GEN Loss": 0.4869081974029541,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8554216867469879,
      "eval_PRM F1 AUC": 0.6668412781561027,
      "eval_PRM F1 Neg": 0.4782608695652174,
      "eval_PRM Loss": 0.4486120343208313,
      "eval_PRM NPV": 0.4782608695652174,
      "eval_PRM Precision": 0.8554216867469879,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9861778616905212,
      "eval_runtime": 55.9687,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1045
    },
    {
      "epoch": 0.09650559335716757,
      "grad_norm": 1.7310991194741818,
      "learning_rate": 1.2382917301537061e-06,
      "loss": 0.7717,
      "step": 1046
    },
    {
      "epoch": 0.09659785491869449,
      "grad_norm": 1.5023268924529263,
      "learning_rate": 1.2382554552334558e-06,
      "loss": 0.6981,
      "step": 1047
    },
    {
      "epoch": 0.09669011648022142,
      "grad_norm": 3.330774683342622,
      "learning_rate": 1.2382191247390307e-06,
      "loss": 0.945,
      "step": 1048
    },
    {
      "epoch": 0.09678237804174836,
      "grad_norm": 1.3849945323861854,
      "learning_rate": 1.2381827386737234e-06,
      "loss": 0.75,
      "step": 1049
    },
    {
      "epoch": 0.09687463960327529,
      "grad_norm": 2.560950054271451,
      "learning_rate": 1.238146297040831e-06,
      "loss": 0.8057,
      "step": 1050
    },
    {
      "epoch": 0.09687463960327529,
      "eval_GEN Loss": 0.4846045970916748,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.863905325443787,
      "eval_PRM F1 AUC": 0.6571503404924044,
      "eval_PRM F1 Neg": 0.46511627906976744,
      "eval_PRM Loss": 0.4420788288116455,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8488372093023255,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.43478260869565216,
      "eval_loss": 1.0069711208343506,
      "eval_runtime": 56.3492,
      "eval_samples_per_second": 1.154,
      "eval_steps_per_second": 0.089,
      "step": 1050
    },
    {
      "epoch": 0.09696690116480221,
      "grad_norm": 1.2051478050447069,
      "learning_rate": 1.2381097998436562e-06,
      "loss": 0.7035,
      "step": 1051
    },
    {
      "epoch": 0.09705916272632914,
      "grad_norm": 1.6971767462504381,
      "learning_rate": 1.2380732470855062e-06,
      "loss": 0.7376,
      "step": 1052
    },
    {
      "epoch": 0.09715142428785607,
      "grad_norm": 2.5051400000537924,
      "learning_rate": 1.2380366387696935e-06,
      "loss": 0.9175,
      "step": 1053
    },
    {
      "epoch": 0.097243685849383,
      "grad_norm": 1.9790729909551097,
      "learning_rate": 1.2379999748995358e-06,
      "loss": 0.7782,
      "step": 1054
    },
    {
      "epoch": 0.09733594741090992,
      "grad_norm": 1.5143617905400066,
      "learning_rate": 1.2379632554783553e-06,
      "loss": 0.5763,
      "step": 1055
    },
    {
      "epoch": 0.09733594741090992,
      "eval_GEN Loss": 0.4831518232822418,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8554216867469879,
      "eval_PRM F1 AUC": 0.6668412781561027,
      "eval_PRM F1 Neg": 0.4782608695652174,
      "eval_PRM Loss": 0.4305679500102997,
      "eval_PRM NPV": 0.4782608695652174,
      "eval_PRM Precision": 0.8554216867469879,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9697115421295166,
      "eval_runtime": 56.2017,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 1055
    },
    {
      "epoch": 0.09742820897243686,
      "grad_norm": 2.1927628921543154,
      "learning_rate": 1.23792648050948e-06,
      "loss": 0.9713,
      "step": 1056
    },
    {
      "epoch": 0.09752047053396379,
      "grad_norm": 1.7522898262035198,
      "learning_rate": 1.2378896499962423e-06,
      "loss": 0.7826,
      "step": 1057
    },
    {
      "epoch": 0.09761273209549072,
      "grad_norm": 2.464980681638915,
      "learning_rate": 1.2378527639419795e-06,
      "loss": 0.9874,
      "step": 1058
    },
    {
      "epoch": 0.09770499365701764,
      "grad_norm": 2.5801590379761583,
      "learning_rate": 1.237815822350035e-06,
      "loss": 0.8722,
      "step": 1059
    },
    {
      "epoch": 0.09779725521854457,
      "grad_norm": 1.6227529227120872,
      "learning_rate": 1.2377788252237558e-06,
      "loss": 0.7129,
      "step": 1060
    },
    {
      "epoch": 0.09779725521854457,
      "eval_GEN Loss": 0.48260170221328735,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4441496431827545,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9358773827552795,
      "eval_runtime": 56.0349,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1060
    },
    {
      "epoch": 0.0978895167800715,
      "grad_norm": 2.0915950164198045,
      "learning_rate": 1.237741772566495e-06,
      "loss": 0.8365,
      "step": 1061
    },
    {
      "epoch": 0.09798177834159844,
      "grad_norm": 2.209002448171688,
      "learning_rate": 1.2377046643816104e-06,
      "loss": 0.879,
      "step": 1062
    },
    {
      "epoch": 0.09807403990312535,
      "grad_norm": 2.0767325247321,
      "learning_rate": 1.2376675006724644e-06,
      "loss": 0.8205,
      "step": 1063
    },
    {
      "epoch": 0.09816630146465229,
      "grad_norm": 1.6383269233868047,
      "learning_rate": 1.2376302814424255e-06,
      "loss": 0.7769,
      "step": 1064
    },
    {
      "epoch": 0.09825856302617922,
      "grad_norm": 1.335123538600553,
      "learning_rate": 1.237593006694866e-06,
      "loss": 0.7526,
      "step": 1065
    },
    {
      "epoch": 0.09825856302617922,
      "eval_GEN Loss": 0.4857008755207062,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM Loss": 0.48533496260643005,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.954026460647583,
      "eval_runtime": 55.9249,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1065
    },
    {
      "epoch": 0.09835082458770615,
      "grad_norm": 1.5668643122578736,
      "learning_rate": 1.2375556764331641e-06,
      "loss": 0.6694,
      "step": 1066
    },
    {
      "epoch": 0.09844308614923307,
      "grad_norm": 2.265192148242357,
      "learning_rate": 1.2375182906607025e-06,
      "loss": 0.8344,
      "step": 1067
    },
    {
      "epoch": 0.09853534771076,
      "grad_norm": 3.331332787875964,
      "learning_rate": 1.2374808493808693e-06,
      "loss": 0.914,
      "step": 1068
    },
    {
      "epoch": 0.09862760927228693,
      "grad_norm": 3.0449674069849713,
      "learning_rate": 1.2374433525970576e-06,
      "loss": 0.9626,
      "step": 1069
    },
    {
      "epoch": 0.09871987083381387,
      "grad_norm": 3.495290514797651,
      "learning_rate": 1.2374058003126651e-06,
      "loss": 0.7489,
      "step": 1070
    },
    {
      "epoch": 0.09871987083381387,
      "eval_GEN Loss": 0.48816636204719543,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM Loss": 0.4851211607456207,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9559495449066162,
      "eval_runtime": 55.9455,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1070
    },
    {
      "epoch": 0.09881213239534078,
      "grad_norm": 2.768920860049011,
      "learning_rate": 1.2373681925310952e-06,
      "loss": 0.8713,
      "step": 1071
    },
    {
      "epoch": 0.09890439395686772,
      "grad_norm": 2.326404515885457,
      "learning_rate": 1.2373305292557558e-06,
      "loss": 0.5683,
      "step": 1072
    },
    {
      "epoch": 0.09899665551839465,
      "grad_norm": 1.707817515880418,
      "learning_rate": 1.23729281049006e-06,
      "loss": 0.6518,
      "step": 1073
    },
    {
      "epoch": 0.09908891707992158,
      "grad_norm": 2.1661936872492045,
      "learning_rate": 1.2372550362374258e-06,
      "loss": 0.7883,
      "step": 1074
    },
    {
      "epoch": 0.0991811786414485,
      "grad_norm": 1.5784542421180987,
      "learning_rate": 1.2372172065012768e-06,
      "loss": 0.7391,
      "step": 1075
    },
    {
      "epoch": 0.0991811786414485,
      "eval_GEN Loss": 0.48966479301452637,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.42942070960998535,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9418870210647583,
      "eval_runtime": 55.9779,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1075
    },
    {
      "epoch": 0.09927344020297543,
      "grad_norm": 2.318419152657595,
      "learning_rate": 1.2371793212850406e-06,
      "loss": 1.0127,
      "step": 1076
    },
    {
      "epoch": 0.09936570176450236,
      "grad_norm": 1.5615375184288336,
      "learning_rate": 1.237141380592151e-06,
      "loss": 0.7858,
      "step": 1077
    },
    {
      "epoch": 0.0994579633260293,
      "grad_norm": 3.6175963867550807,
      "learning_rate": 1.2371033844260457e-06,
      "loss": 0.85,
      "step": 1078
    },
    {
      "epoch": 0.09955022488755622,
      "grad_norm": 2.180342768126879,
      "learning_rate": 1.2370653327901685e-06,
      "loss": 0.779,
      "step": 1079
    },
    {
      "epoch": 0.09964248644908315,
      "grad_norm": 1.7449433978213127,
      "learning_rate": 1.2370272256879673e-06,
      "loss": 0.8243,
      "step": 1080
    },
    {
      "epoch": 0.09964248644908315,
      "eval_GEN Loss": 0.49025124311447144,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4073673486709595,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9507211446762085,
      "eval_runtime": 56.1036,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1080
    },
    {
      "epoch": 0.09973474801061008,
      "grad_norm": 2.920875514089996,
      "learning_rate": 1.2369890631228956e-06,
      "loss": 1.0256,
      "step": 1081
    },
    {
      "epoch": 0.09982700957213701,
      "grad_norm": 2.224332763254484,
      "learning_rate": 1.2369508450984115e-06,
      "loss": 0.6702,
      "step": 1082
    },
    {
      "epoch": 0.09991927113366393,
      "grad_norm": 1.545276626196894,
      "learning_rate": 1.2369125716179789e-06,
      "loss": 0.7714,
      "step": 1083
    },
    {
      "epoch": 0.10001153269519086,
      "grad_norm": 1.6531163183487245,
      "learning_rate": 1.2368742426850658e-06,
      "loss": 0.7989,
      "step": 1084
    },
    {
      "epoch": 0.1001037942567178,
      "grad_norm": 2.400449458139153,
      "learning_rate": 1.2368358583031457e-06,
      "loss": 0.9417,
      "step": 1085
    },
    {
      "epoch": 0.1001037942567178,
      "eval_GEN Loss": 0.4873465299606323,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.41614651679992676,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9508413672447205,
      "eval_runtime": 56.9944,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1085
    },
    {
      "epoch": 0.10019605581824473,
      "grad_norm": 2.397450377181036,
      "learning_rate": 1.236797418475697e-06,
      "loss": 0.9024,
      "step": 1086
    },
    {
      "epoch": 0.10028831737977165,
      "grad_norm": 2.774157868181651,
      "learning_rate": 1.2367589232062033e-06,
      "loss": 0.9406,
      "step": 1087
    },
    {
      "epoch": 0.10038057894129858,
      "grad_norm": 2.155396098393847,
      "learning_rate": 1.236720372498153e-06,
      "loss": 0.8102,
      "step": 1088
    },
    {
      "epoch": 0.10047284050282551,
      "grad_norm": 1.566442610547236,
      "learning_rate": 1.2366817663550399e-06,
      "loss": 0.669,
      "step": 1089
    },
    {
      "epoch": 0.10056510206435244,
      "grad_norm": 1.6382905235056355,
      "learning_rate": 1.2366431047803622e-06,
      "loss": 0.7896,
      "step": 1090
    },
    {
      "epoch": 0.10056510206435244,
      "eval_GEN Loss": 0.4887427091598511,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.45235133171081543,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9478966593742371,
      "eval_runtime": 56.2997,
      "eval_samples_per_second": 1.155,
      "eval_steps_per_second": 0.089,
      "step": 1090
    },
    {
      "epoch": 0.10065736362587936,
      "grad_norm": 3.260548589865754,
      "learning_rate": 1.2366043877776235e-06,
      "loss": 0.93,
      "step": 1091
    },
    {
      "epoch": 0.1007496251874063,
      "grad_norm": 2.8646590819893207,
      "learning_rate": 1.2365656153503327e-06,
      "loss": 0.6845,
      "step": 1092
    },
    {
      "epoch": 0.10084188674893323,
      "grad_norm": 2.0288586179259758,
      "learning_rate": 1.236526787502003e-06,
      "loss": 0.8515,
      "step": 1093
    },
    {
      "epoch": 0.10093414831046016,
      "grad_norm": 1.5112978349865815,
      "learning_rate": 1.2364879042361535e-06,
      "loss": 0.7404,
      "step": 1094
    },
    {
      "epoch": 0.10102640987198708,
      "grad_norm": 2.401128397609629,
      "learning_rate": 1.2364489655563075e-06,
      "loss": 0.66,
      "step": 1095
    },
    {
      "epoch": 0.10102640987198708,
      "eval_GEN Loss": 0.4885111153125763,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.46584510803222656,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9509615302085876,
      "eval_runtime": 57.106,
      "eval_samples_per_second": 1.138,
      "eval_steps_per_second": 0.088,
      "step": 1095
    },
    {
      "epoch": 0.10111867143351401,
      "grad_norm": 1.5690924072635593,
      "learning_rate": 1.2364099714659937e-06,
      "loss": 0.7624,
      "step": 1096
    },
    {
      "epoch": 0.10121093299504094,
      "grad_norm": 2.1418884280774155,
      "learning_rate": 1.236370921968746e-06,
      "loss": 1.0047,
      "step": 1097
    },
    {
      "epoch": 0.10130319455656787,
      "grad_norm": 1.8642136781223946,
      "learning_rate": 1.2363318170681032e-06,
      "loss": 0.8175,
      "step": 1098
    },
    {
      "epoch": 0.10139545611809479,
      "grad_norm": 2.6736263314289292,
      "learning_rate": 1.2362926567676087e-06,
      "loss": 0.9793,
      "step": 1099
    },
    {
      "epoch": 0.10148771767962173,
      "grad_norm": 1.752541227854694,
      "learning_rate": 1.2362534410708114e-06,
      "loss": 0.7323,
      "step": 1100
    },
    {
      "epoch": 0.10148771767962173,
      "eval_GEN Loss": 0.4882470965385437,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4444061517715454,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9581129550933838,
      "eval_runtime": 55.8385,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1100
    },
    {
      "epoch": 0.10157997924114866,
      "grad_norm": 2.503064476777125,
      "learning_rate": 1.2362141699812652e-06,
      "loss": 0.778,
      "step": 1101
    },
    {
      "epoch": 0.10167224080267559,
      "grad_norm": 1.5878985869751334,
      "learning_rate": 1.236174843502529e-06,
      "loss": 0.7759,
      "step": 1102
    },
    {
      "epoch": 0.10176450236420251,
      "grad_norm": 1.9126008356736544,
      "learning_rate": 1.2361354616381664e-06,
      "loss": 0.7474,
      "step": 1103
    },
    {
      "epoch": 0.10185676392572944,
      "grad_norm": 1.5539796327415196,
      "learning_rate": 1.2360960243917463e-06,
      "loss": 0.749,
      "step": 1104
    },
    {
      "epoch": 0.10194902548725637,
      "grad_norm": 1.960197152011404,
      "learning_rate": 1.2360565317668425e-06,
      "loss": 0.7112,
      "step": 1105
    },
    {
      "epoch": 0.10194902548725637,
      "eval_GEN Loss": 0.4906383156776428,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.44833284616470337,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9697115421295166,
      "eval_runtime": 56.0608,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1105
    },
    {
      "epoch": 0.1020412870487833,
      "grad_norm": 2.0062356883556047,
      "learning_rate": 1.236016983767034e-06,
      "loss": 0.9052,
      "step": 1106
    },
    {
      "epoch": 0.10213354861031022,
      "grad_norm": 2.407371182829471,
      "learning_rate": 1.2359773803959047e-06,
      "loss": 0.8707,
      "step": 1107
    },
    {
      "epoch": 0.10222581017183716,
      "grad_norm": 1.8258105151103796,
      "learning_rate": 1.2359377216570437e-06,
      "loss": 0.6605,
      "step": 1108
    },
    {
      "epoch": 0.10231807173336409,
      "grad_norm": 1.8117603448203083,
      "learning_rate": 1.2358980075540444e-06,
      "loss": 0.5328,
      "step": 1109
    },
    {
      "epoch": 0.10241033329489102,
      "grad_norm": 2.3949952588731342,
      "learning_rate": 1.2358582380905063e-06,
      "loss": 0.8625,
      "step": 1110
    },
    {
      "epoch": 0.10241033329489102,
      "eval_GEN Loss": 0.4929998815059662,
      "eval_GEN top-5 accuracy": 0.9767645769399387,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.450009286403656,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9662860631942749,
      "eval_runtime": 55.8382,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1110
    },
    {
      "epoch": 0.10250259485641794,
      "grad_norm": 2.267304274320555,
      "learning_rate": 1.2358184132700331e-06,
      "loss": 1.0628,
      "step": 1111
    },
    {
      "epoch": 0.10259485641794487,
      "grad_norm": 1.9153714077184445,
      "learning_rate": 1.2357785330962338e-06,
      "loss": 0.8216,
      "step": 1112
    },
    {
      "epoch": 0.1026871179794718,
      "grad_norm": 2.1349724599465776,
      "learning_rate": 1.2357385975727226e-06,
      "loss": 0.8565,
      "step": 1113
    },
    {
      "epoch": 0.10277937954099874,
      "grad_norm": 1.5515261839089143,
      "learning_rate": 1.2356986067031181e-06,
      "loss": 0.7103,
      "step": 1114
    },
    {
      "epoch": 0.10287164110252565,
      "grad_norm": 2.18568346598404,
      "learning_rate": 1.2356585604910448e-06,
      "loss": 0.9461,
      "step": 1115
    },
    {
      "epoch": 0.10287164110252565,
      "eval_GEN Loss": 0.4928361177444458,
      "eval_GEN top-5 accuracy": 0.9776413853572994,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.4473930597305298,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9716346263885498,
      "eval_runtime": 55.9383,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1115
    },
    {
      "epoch": 0.10296390266405259,
      "grad_norm": 1.9593718521236188,
      "learning_rate": 1.2356184589401317e-06,
      "loss": 0.701,
      "step": 1116
    },
    {
      "epoch": 0.10305616422557952,
      "grad_norm": 1.5112039103193984,
      "learning_rate": 1.2355783020540125e-06,
      "loss": 0.4822,
      "step": 1117
    },
    {
      "epoch": 0.10314842578710645,
      "grad_norm": 3.492610895074985,
      "learning_rate": 1.2355380898363265e-06,
      "loss": 1.0495,
      "step": 1118
    },
    {
      "epoch": 0.10324068734863337,
      "grad_norm": 2.779790989181864,
      "learning_rate": 1.235497822290718e-06,
      "loss": 0.8498,
      "step": 1119
    },
    {
      "epoch": 0.1033329489101603,
      "grad_norm": 2.495259404393216,
      "learning_rate": 1.2354574994208358e-06,
      "loss": 0.8314,
      "step": 1120
    },
    {
      "epoch": 0.1033329489101603,
      "eval_GEN Loss": 0.49186062812805176,
      "eval_GEN top-5 accuracy": 0.977202981148619,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.4566497206687927,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9811899065971375,
      "eval_runtime": 56.1218,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 1120
    },
    {
      "epoch": 0.10342521047168723,
      "grad_norm": 2.638511333717374,
      "learning_rate": 1.235417121230334e-06,
      "loss": 1.0844,
      "step": 1121
    },
    {
      "epoch": 0.10351747203321417,
      "grad_norm": 2.157032359175817,
      "learning_rate": 1.235376687722872e-06,
      "loss": 0.9147,
      "step": 1122
    },
    {
      "epoch": 0.10360973359474109,
      "grad_norm": 1.3307503885733896,
      "learning_rate": 1.2353361989021137e-06,
      "loss": 0.7644,
      "step": 1123
    },
    {
      "epoch": 0.10370199515626802,
      "grad_norm": 1.9289836445089643,
      "learning_rate": 1.2352956547717285e-06,
      "loss": 0.8233,
      "step": 1124
    },
    {
      "epoch": 0.10379425671779495,
      "grad_norm": 1.6397876995873373,
      "learning_rate": 1.2352550553353903e-06,
      "loss": 0.5665,
      "step": 1125
    },
    {
      "epoch": 0.10379425671779495,
      "eval_GEN Loss": 0.4904106557369232,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8518518518518519,
      "eval_PRM F1 AUC": 0.6982713462545836,
      "eval_PRM F1 Neg": 0.52,
      "eval_PRM Loss": 0.46971702575683594,
      "eval_PRM NPV": 0.48148148148148145,
      "eval_PRM Precision": 0.8734177215189873,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9793269038200378,
      "eval_runtime": 56.005,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1125
    },
    {
      "epoch": 0.10388651827932188,
      "grad_norm": 2.223510724143778,
      "learning_rate": 1.2352144005967787e-06,
      "loss": 0.7505,
      "step": 1126
    },
    {
      "epoch": 0.1039787798408488,
      "grad_norm": 3.043622424099526,
      "learning_rate": 1.2351736905595774e-06,
      "loss": 0.9179,
      "step": 1127
    },
    {
      "epoch": 0.10407104140237573,
      "grad_norm": 2.3547098805165105,
      "learning_rate": 1.235132925227476e-06,
      "loss": 0.8413,
      "step": 1128
    },
    {
      "epoch": 0.10416330296390267,
      "grad_norm": 1.9630806022172627,
      "learning_rate": 1.2350921046041683e-06,
      "loss": 0.7343,
      "step": 1129
    },
    {
      "epoch": 0.1042555645254296,
      "grad_norm": 1.946989665759252,
      "learning_rate": 1.2350512286933538e-06,
      "loss": 0.8474,
      "step": 1130
    },
    {
      "epoch": 0.1042555645254296,
      "eval_GEN Loss": 0.4904525578022003,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8375,
      "eval_PRM F1 AUC": 0.6862231534834993,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM Loss": 0.486596941947937,
      "eval_PRM NPV": 0.4482758620689655,
      "eval_PRM Precision": 0.8701298701298701,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9783653616905212,
      "eval_runtime": 57.2988,
      "eval_samples_per_second": 1.134,
      "eval_steps_per_second": 0.087,
      "step": 1130
    },
    {
      "epoch": 0.10434782608695652,
      "grad_norm": 2.2566667342476494,
      "learning_rate": 1.2350102974987368e-06,
      "loss": 0.9051,
      "step": 1131
    },
    {
      "epoch": 0.10444008764848345,
      "grad_norm": 1.190516769690317,
      "learning_rate": 1.2349693110240265e-06,
      "loss": 0.7958,
      "step": 1132
    },
    {
      "epoch": 0.10453234921001038,
      "grad_norm": 2.5848385443398807,
      "learning_rate": 1.2349282692729373e-06,
      "loss": 0.8176,
      "step": 1133
    },
    {
      "epoch": 0.10462461077153731,
      "grad_norm": 1.6079555069768316,
      "learning_rate": 1.234887172249188e-06,
      "loss": 0.8849,
      "step": 1134
    },
    {
      "epoch": 0.10471687233306423,
      "grad_norm": 1.8259966634702556,
      "learning_rate": 1.2348460199565035e-06,
      "loss": 0.7843,
      "step": 1135
    },
    {
      "epoch": 0.10471687233306423,
      "eval_GEN Loss": 0.49019476771354675,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7547169811320755,
      "eval_PRM F1": 0.8375,
      "eval_PRM F1 AUC": 0.6862231534834993,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM Loss": 0.49282306432724,
      "eval_PRM NPV": 0.4482758620689655,
      "eval_PRM Precision": 0.8701298701298701,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9757211804389954,
      "eval_runtime": 56.0892,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1135
    },
    {
      "epoch": 0.10480913389459116,
      "grad_norm": 2.578494402269661,
      "learning_rate": 1.2348048123986125e-06,
      "loss": 0.8173,
      "step": 1136
    },
    {
      "epoch": 0.1049013954561181,
      "grad_norm": 1.7756464307745905,
      "learning_rate": 1.2347635495792494e-06,
      "loss": 1.0041,
      "step": 1137
    },
    {
      "epoch": 0.10499365701764503,
      "grad_norm": 2.242358836794997,
      "learning_rate": 1.234722231502154e-06,
      "loss": 0.8198,
      "step": 1138
    },
    {
      "epoch": 0.10508591857917195,
      "grad_norm": 1.562835763782735,
      "learning_rate": 1.2346808581710702e-06,
      "loss": 0.7372,
      "step": 1139
    },
    {
      "epoch": 0.10517818014069888,
      "grad_norm": 1.7829047388278465,
      "learning_rate": 1.2346394295897473e-06,
      "loss": 0.7671,
      "step": 1140
    },
    {
      "epoch": 0.10517818014069888,
      "eval_GEN Loss": 0.4887770712375641,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8427672955974843,
      "eval_PRM F1 AUC": 0.707962283918282,
      "eval_PRM F1 Neg": 0.5283018867924528,
      "eval_PRM Loss": 0.5005943179130554,
      "eval_PRM NPV": 0.4666666666666667,
      "eval_PRM Precision": 0.881578947368421,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9809495210647583,
      "eval_runtime": 55.797,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 1140
    },
    {
      "epoch": 0.10527044170222581,
      "grad_norm": 2.4676921484264422,
      "learning_rate": 1.2345979457619397e-06,
      "loss": 1.0341,
      "step": 1141
    },
    {
      "epoch": 0.10536270326375274,
      "grad_norm": 2.3281434201296265,
      "learning_rate": 1.2345564066914068e-06,
      "loss": 0.9362,
      "step": 1142
    },
    {
      "epoch": 0.10545496482527966,
      "grad_norm": 1.8217499538697424,
      "learning_rate": 1.234514812381913e-06,
      "loss": 0.8462,
      "step": 1143
    },
    {
      "epoch": 0.1055472263868066,
      "grad_norm": 1.998244371373172,
      "learning_rate": 1.2344731628372273e-06,
      "loss": 0.8503,
      "step": 1144
    },
    {
      "epoch": 0.10563948794833353,
      "grad_norm": 1.8376465866184033,
      "learning_rate": 1.2344314580611242e-06,
      "loss": 0.7209,
      "step": 1145
    },
    {
      "epoch": 0.10563948794833353,
      "eval_GEN Loss": 0.48478788137435913,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.4554228186607361,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9803485870361328,
      "eval_runtime": 55.9927,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1145
    },
    {
      "epoch": 0.10573174950986046,
      "grad_norm": 1.5757690964316267,
      "learning_rate": 1.2343896980573832e-06,
      "loss": 0.8978,
      "step": 1146
    },
    {
      "epoch": 0.10582401107138738,
      "grad_norm": 2.4409144721834375,
      "learning_rate": 1.2343478828297887e-06,
      "loss": 1.0605,
      "step": 1147
    },
    {
      "epoch": 0.10591627263291431,
      "grad_norm": 2.052028030357802,
      "learning_rate": 1.2343060123821296e-06,
      "loss": 0.8907,
      "step": 1148
    },
    {
      "epoch": 0.10600853419444124,
      "grad_norm": 1.544575693393687,
      "learning_rate": 1.234264086718201e-06,
      "loss": 0.6599,
      "step": 1149
    },
    {
      "epoch": 0.10610079575596817,
      "grad_norm": 2.1519084900119787,
      "learning_rate": 1.2342221058418017e-06,
      "loss": 0.9246,
      "step": 1150
    },
    {
      "epoch": 0.10610079575596817,
      "eval_GEN Loss": 0.48479387164115906,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.6511262441068623,
      "eval_PRM F1 Neg": 0.45454545454545453,
      "eval_PRM Loss": 0.43649429082870483,
      "eval_PRM NPV": 0.47619047619047616,
      "eval_PRM Precision": 0.8470588235294118,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.43478260869565216,
      "eval_loss": 1.0037260055541992,
      "eval_runtime": 55.917,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1150
    },
    {
      "epoch": 0.1061930573174951,
      "grad_norm": 1.7989726386356726,
      "learning_rate": 1.234180069756736e-06,
      "loss": 0.7731,
      "step": 1151
    },
    {
      "epoch": 0.10628531887902203,
      "grad_norm": 1.8165689721538687,
      "learning_rate": 1.2341379784668137e-06,
      "loss": 0.7234,
      "step": 1152
    },
    {
      "epoch": 0.10637758044054896,
      "grad_norm": 2.6393223009208033,
      "learning_rate": 1.234095831975849e-06,
      "loss": 0.9134,
      "step": 1153
    },
    {
      "epoch": 0.10646984200207589,
      "grad_norm": 2.770264310554098,
      "learning_rate": 1.2340536302876614e-06,
      "loss": 0.9177,
      "step": 1154
    },
    {
      "epoch": 0.10656210356360281,
      "grad_norm": 1.7140804450288918,
      "learning_rate": 1.234011373406075e-06,
      "loss": 0.7026,
      "step": 1155
    },
    {
      "epoch": 0.10656210356360281,
      "eval_GEN Loss": 0.486250638961792,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8674698795180723,
      "eval_PRM F1 AUC": 0.6946045049764275,
      "eval_PRM F1 Neg": 0.5217391304347826,
      "eval_PRM Loss": 0.40797463059425354,
      "eval_PRM NPV": 0.5217391304347826,
      "eval_PRM Precision": 0.8674698795180723,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.959254801273346,
      "eval_runtime": 55.9408,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1155
    },
    {
      "epoch": 0.10665436512512974,
      "grad_norm": 1.726417072871133,
      "learning_rate": 1.2339690613349194e-06,
      "loss": 0.9421,
      "step": 1156
    },
    {
      "epoch": 0.10674662668665667,
      "grad_norm": 1.7498272511515074,
      "learning_rate": 1.233926694078029e-06,
      "loss": 0.8111,
      "step": 1157
    },
    {
      "epoch": 0.1068388882481836,
      "grad_norm": 2.1298247243358723,
      "learning_rate": 1.233884271639243e-06,
      "loss": 1.0038,
      "step": 1158
    },
    {
      "epoch": 0.10693114980971052,
      "grad_norm": 2.31391672660952,
      "learning_rate": 1.2338417940224059e-06,
      "loss": 0.7578,
      "step": 1159
    },
    {
      "epoch": 0.10702341137123746,
      "grad_norm": 1.9063754681354979,
      "learning_rate": 1.2337992612313672e-06,
      "loss": 0.8946,
      "step": 1160
    },
    {
      "epoch": 0.10702341137123746,
      "eval_GEN Loss": 0.4875490963459015,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.43092072010040283,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9337740540504456,
      "eval_runtime": 55.9026,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1160
    },
    {
      "epoch": 0.10711567293276439,
      "grad_norm": 2.2177275668005723,
      "learning_rate": 1.233756673269981e-06,
      "loss": 0.8565,
      "step": 1161
    },
    {
      "epoch": 0.10720793449429132,
      "grad_norm": 2.305173590342485,
      "learning_rate": 1.2337140301421071e-06,
      "loss": 0.8235,
      "step": 1162
    },
    {
      "epoch": 0.10730019605581824,
      "grad_norm": 2.0110986777526,
      "learning_rate": 1.2336713318516095e-06,
      "loss": 0.6888,
      "step": 1163
    },
    {
      "epoch": 0.10739245761734517,
      "grad_norm": 1.7485914014280124,
      "learning_rate": 1.2336285784023578e-06,
      "loss": 0.9129,
      "step": 1164
    },
    {
      "epoch": 0.1074847191788721,
      "grad_norm": 1.929861511939592,
      "learning_rate": 1.2335857697982266e-06,
      "loss": 0.8243,
      "step": 1165
    },
    {
      "epoch": 0.1074847191788721,
      "eval_GEN Loss": 0.4870619773864746,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.44409653544425964,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9309495091438293,
      "eval_runtime": 56.1366,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 1165
    },
    {
      "epoch": 0.10757698074039904,
      "grad_norm": 1.980310529716834,
      "learning_rate": 1.2335429060430944e-06,
      "loss": 0.8317,
      "step": 1166
    },
    {
      "epoch": 0.10766924230192595,
      "grad_norm": 1.656414726821583,
      "learning_rate": 1.2334999871408467e-06,
      "loss": 0.7256,
      "step": 1167
    },
    {
      "epoch": 0.10776150386345289,
      "grad_norm": 1.4769578060692212,
      "learning_rate": 1.233457013095372e-06,
      "loss": 0.8265,
      "step": 1168
    },
    {
      "epoch": 0.10785376542497982,
      "grad_norm": 1.6487135203109253,
      "learning_rate": 1.2334139839105654e-06,
      "loss": 0.8368,
      "step": 1169
    },
    {
      "epoch": 0.10794602698650675,
      "grad_norm": 1.5763624855331888,
      "learning_rate": 1.2333708995903258e-06,
      "loss": 0.6424,
      "step": 1170
    },
    {
      "epoch": 0.10794602698650675,
      "eval_GEN Loss": 0.48684194684028625,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.42577382922172546,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9293269515037537,
      "eval_runtime": 56.9057,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1170
    },
    {
      "epoch": 0.10803828854803367,
      "grad_norm": 2.069352072052707,
      "learning_rate": 1.2333277601385577e-06,
      "loss": 0.822,
      "step": 1171
    },
    {
      "epoch": 0.1081305501095606,
      "grad_norm": 1.5312709494364685,
      "learning_rate": 1.2332845655591704e-06,
      "loss": 0.6668,
      "step": 1172
    },
    {
      "epoch": 0.10822281167108753,
      "grad_norm": 2.0578395624832018,
      "learning_rate": 1.2332413158560784e-06,
      "loss": 0.6411,
      "step": 1173
    },
    {
      "epoch": 0.10831507323261447,
      "grad_norm": 2.02697975392182,
      "learning_rate": 1.2331980110332007e-06,
      "loss": 0.9499,
      "step": 1174
    },
    {
      "epoch": 0.10840733479414139,
      "grad_norm": 1.677973701213988,
      "learning_rate": 1.2331546510944622e-06,
      "loss": 0.8123,
      "step": 1175
    },
    {
      "epoch": 0.10840733479414139,
      "eval_GEN Loss": 0.48646190762519836,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.40674087405204773,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9456730484962463,
      "eval_runtime": 57.0068,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1175
    },
    {
      "epoch": 0.10849959635566832,
      "grad_norm": 1.3564191365190188,
      "learning_rate": 1.2331112360437918e-06,
      "loss": 0.7495,
      "step": 1176
    },
    {
      "epoch": 0.10859185791719525,
      "grad_norm": 2.548023510607541,
      "learning_rate": 1.2330677658851242e-06,
      "loss": 1.0494,
      "step": 1177
    },
    {
      "epoch": 0.10868411947872218,
      "grad_norm": 2.4895815250280777,
      "learning_rate": 1.2330242406223985e-06,
      "loss": 0.9173,
      "step": 1178
    },
    {
      "epoch": 0.1087763810402491,
      "grad_norm": 2.93590344431322,
      "learning_rate": 1.2329806602595589e-06,
      "loss": 0.8843,
      "step": 1179
    },
    {
      "epoch": 0.10886864260177603,
      "grad_norm": 1.9049200362528216,
      "learning_rate": 1.2329370248005551e-06,
      "loss": 0.8401,
      "step": 1180
    },
    {
      "epoch": 0.10886864260177603,
      "eval_GEN Loss": 0.48844602704048157,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4213868975639343,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9299278855323792,
      "eval_runtime": 55.8828,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1180
    },
    {
      "epoch": 0.10896090416330297,
      "grad_norm": 1.7688347547982666,
      "learning_rate": 1.2328933342493411e-06,
      "loss": 0.8114,
      "step": 1181
    },
    {
      "epoch": 0.1090531657248299,
      "grad_norm": 1.7359552501254916,
      "learning_rate": 1.2328495886098762e-06,
      "loss": 0.7648,
      "step": 1182
    },
    {
      "epoch": 0.10914542728635682,
      "grad_norm": 1.612046766686986,
      "learning_rate": 1.232805787886125e-06,
      "loss": 0.7457,
      "step": 1183
    },
    {
      "epoch": 0.10923768884788375,
      "grad_norm": 1.9636722783651244,
      "learning_rate": 1.2327619320820566e-06,
      "loss": 0.917,
      "step": 1184
    },
    {
      "epoch": 0.10932995040941068,
      "grad_norm": 1.5754703622264092,
      "learning_rate": 1.2327180212016452e-06,
      "loss": 0.6563,
      "step": 1185
    },
    {
      "epoch": 0.10932995040941068,
      "eval_GEN Loss": 0.49035125970840454,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8441558441558441,
      "eval_PRM F1 AUC": 0.7611314824515453,
      "eval_PRM F1 Neg": 0.5862068965517241,
      "eval_PRM Loss": 0.4824090898036957,
      "eval_PRM NPV": 0.4857142857142857,
      "eval_PRM Precision": 0.9154929577464789,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9477764368057251,
      "eval_runtime": 55.8827,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1185
    },
    {
      "epoch": 0.10942221197093761,
      "grad_norm": 2.2896371509113744,
      "learning_rate": 1.2326740552488702e-06,
      "loss": 0.8701,
      "step": 1186
    },
    {
      "epoch": 0.10951447353246453,
      "grad_norm": 2.0582193165405154,
      "learning_rate": 1.2326300342277159e-06,
      "loss": 0.8605,
      "step": 1187
    },
    {
      "epoch": 0.10960673509399146,
      "grad_norm": 2.1309555411326198,
      "learning_rate": 1.2325859581421713e-06,
      "loss": 0.8683,
      "step": 1188
    },
    {
      "epoch": 0.1096989966555184,
      "grad_norm": 2.510452992249232,
      "learning_rate": 1.2325418269962308e-06,
      "loss": 0.903,
      "step": 1189
    },
    {
      "epoch": 0.10979125821704533,
      "grad_norm": 2.2442105538117354,
      "learning_rate": 1.2324976407938939e-06,
      "loss": 1.0334,
      "step": 1190
    },
    {
      "epoch": 0.10979125821704533,
      "eval_GEN Loss": 0.48855337500572205,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8589743589743589,
      "eval_PRM F1 AUC": 0.7731796752226295,
      "eval_PRM F1 Neg": 0.6071428571428571,
      "eval_PRM Loss": 0.4656677544116974,
      "eval_PRM NPV": 0.5151515151515151,
      "eval_PRM Precision": 0.9178082191780822,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9493990540504456,
      "eval_runtime": 56.0119,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1190
    },
    {
      "epoch": 0.10988351977857225,
      "grad_norm": 2.022699428884022,
      "learning_rate": 1.2324533995391643e-06,
      "loss": 0.8244,
      "step": 1191
    },
    {
      "epoch": 0.10997578134009918,
      "grad_norm": 1.9424420201266335,
      "learning_rate": 1.2324091032360517e-06,
      "loss": 0.8637,
      "step": 1192
    },
    {
      "epoch": 0.11006804290162611,
      "grad_norm": 2.3609370608937383,
      "learning_rate": 1.23236475188857e-06,
      "loss": 0.7843,
      "step": 1193
    },
    {
      "epoch": 0.11016030446315304,
      "grad_norm": 1.9832704742742902,
      "learning_rate": 1.2323203455007384e-06,
      "loss": 0.6733,
      "step": 1194
    },
    {
      "epoch": 0.11025256602467996,
      "grad_norm": 1.7324730654333904,
      "learning_rate": 1.2322758840765813e-06,
      "loss": 0.6238,
      "step": 1195
    },
    {
      "epoch": 0.11025256602467996,
      "eval_GEN Loss": 0.48724260926246643,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8622754491017964,
      "eval_PRM F1 AUC": 0.6728653745416449,
      "eval_PRM F1 Neg": 0.4888888888888889,
      "eval_PRM Loss": 0.4369560778141022,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8571428571428571,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9841346144676208,
      "eval_runtime": 56.161,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 1195
    },
    {
      "epoch": 0.1103448275862069,
      "grad_norm": 3.3504049718983215,
      "learning_rate": 1.2322313676201274e-06,
      "loss": 0.8395,
      "step": 1196
    },
    {
      "epoch": 0.11043708914773383,
      "grad_norm": 1.8281556673261714,
      "learning_rate": 1.2321867961354116e-06,
      "loss": 0.7803,
      "step": 1197
    },
    {
      "epoch": 0.11052935070926076,
      "grad_norm": 2.1794012034206585,
      "learning_rate": 1.2321421696264723e-06,
      "loss": 0.9177,
      "step": 1198
    },
    {
      "epoch": 0.11062161227078768,
      "grad_norm": 2.4232356631219707,
      "learning_rate": 1.2320974880973542e-06,
      "loss": 0.7455,
      "step": 1199
    },
    {
      "epoch": 0.11071387383231461,
      "grad_norm": 2.466812411105342,
      "learning_rate": 1.2320527515521059e-06,
      "loss": 0.8103,
      "step": 1200
    },
    {
      "epoch": 0.11071387383231461,
      "eval_GEN Loss": 0.4868861436843872,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.863905325443787,
      "eval_PRM F1 AUC": 0.6571503404924044,
      "eval_PRM F1 Neg": 0.46511627906976744,
      "eval_PRM Loss": 0.45209363102912903,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8488372093023255,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.43478260869565216,
      "eval_loss": 1.0098557472229004,
      "eval_runtime": 56.1683,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 1200
    },
    {
      "epoch": 0.11080613539384154,
      "grad_norm": 2.3272115875739146,
      "learning_rate": 1.232007959994782e-06,
      "loss": 0.7054,
      "step": 1201
    },
    {
      "epoch": 0.11089839695536848,
      "grad_norm": 2.583595119552973,
      "learning_rate": 1.231963113429441e-06,
      "loss": 0.9144,
      "step": 1202
    },
    {
      "epoch": 0.1109906585168954,
      "grad_norm": 3.306092969142411,
      "learning_rate": 1.2319182118601475e-06,
      "loss": 0.8067,
      "step": 1203
    },
    {
      "epoch": 0.11108292007842233,
      "grad_norm": 2.5640626066477172,
      "learning_rate": 1.2318732552909703e-06,
      "loss": 0.9441,
      "step": 1204
    },
    {
      "epoch": 0.11117518163994926,
      "grad_norm": 1.6918386714801197,
      "learning_rate": 1.2318282437259835e-06,
      "loss": 0.7275,
      "step": 1205
    },
    {
      "epoch": 0.11117518163994926,
      "eval_GEN Loss": 0.4903247058391571,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8518518518518519,
      "eval_PRM F1 AUC": 0.6982713462545836,
      "eval_PRM F1 Neg": 0.52,
      "eval_PRM Loss": 0.4829825162887573,
      "eval_PRM NPV": 0.48148148148148145,
      "eval_PRM Precision": 0.8734177215189873,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9960336685180664,
      "eval_runtime": 56.0394,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1205
    },
    {
      "epoch": 0.11126744320147619,
      "grad_norm": 1.9142145845356089,
      "learning_rate": 1.2317831771692658e-06,
      "loss": 0.9109,
      "step": 1206
    },
    {
      "epoch": 0.11135970476300311,
      "grad_norm": 2.2061703937594603,
      "learning_rate": 1.231738055624902e-06,
      "loss": 0.8353,
      "step": 1207
    },
    {
      "epoch": 0.11145196632453004,
      "grad_norm": 1.6339986391626806,
      "learning_rate": 1.2316928790969803e-06,
      "loss": 0.7552,
      "step": 1208
    },
    {
      "epoch": 0.11154422788605697,
      "grad_norm": 1.9839433452397386,
      "learning_rate": 1.231647647589595e-06,
      "loss": 0.8809,
      "step": 1209
    },
    {
      "epoch": 0.1116364894475839,
      "grad_norm": 1.6948509211816858,
      "learning_rate": 1.231602361106845e-06,
      "loss": 0.7515,
      "step": 1210
    },
    {
      "epoch": 0.1116364894475839,
      "eval_GEN Loss": 0.49288755655288696,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.84472049689441,
      "eval_PRM F1 AUC": 0.6922472498690415,
      "eval_PRM F1 Neg": 0.5098039215686274,
      "eval_PRM Loss": 0.4933969974517822,
      "eval_PRM NPV": 0.4642857142857143,
      "eval_PRM Precision": 0.8717948717948718,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9946514368057251,
      "eval_runtime": 57.0308,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1210
    },
    {
      "epoch": 0.11172875100911082,
      "grad_norm": 2.333241413499883,
      "learning_rate": 1.2315570196528342e-06,
      "loss": 0.7236,
      "step": 1211
    },
    {
      "epoch": 0.11182101257063776,
      "grad_norm": 1.8259354114271136,
      "learning_rate": 1.2315116232316715e-06,
      "loss": 0.7274,
      "step": 1212
    },
    {
      "epoch": 0.11191327413216469,
      "grad_norm": 1.5460668876861083,
      "learning_rate": 1.2314661718474708e-06,
      "loss": 0.7628,
      "step": 1213
    },
    {
      "epoch": 0.11200553569369162,
      "grad_norm": 3.979263775837841,
      "learning_rate": 1.2314206655043512e-06,
      "loss": 1.0967,
      "step": 1214
    },
    {
      "epoch": 0.11209779725521854,
      "grad_norm": 1.803237941124396,
      "learning_rate": 1.2313751042064362e-06,
      "loss": 0.673,
      "step": 1215
    },
    {
      "epoch": 0.11209779725521854,
      "eval_GEN Loss": 0.4911290109157562,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.48653483390808105,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9960336685180664,
      "eval_runtime": 56.7757,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1215
    },
    {
      "epoch": 0.11219005881674547,
      "grad_norm": 2.0604165445059595,
      "learning_rate": 1.2313294879578548e-06,
      "loss": 0.7396,
      "step": 1216
    },
    {
      "epoch": 0.1122823203782724,
      "grad_norm": 1.8054316251819142,
      "learning_rate": 1.2312838167627408e-06,
      "loss": 0.9521,
      "step": 1217
    },
    {
      "epoch": 0.11237458193979934,
      "grad_norm": 2.0219527025493234,
      "learning_rate": 1.2312380906252334e-06,
      "loss": 0.7238,
      "step": 1218
    },
    {
      "epoch": 0.11246684350132626,
      "grad_norm": 1.9476637258107512,
      "learning_rate": 1.2311923095494755e-06,
      "loss": 0.6648,
      "step": 1219
    },
    {
      "epoch": 0.11255910506285319,
      "grad_norm": 1.9191471098412223,
      "learning_rate": 1.2311464735396166e-06,
      "loss": 0.7794,
      "step": 1220
    },
    {
      "epoch": 0.11255910506285319,
      "eval_GEN Loss": 0.48921579122543335,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.47596776485443115,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9862980842590332,
      "eval_runtime": 56.7988,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1220
    },
    {
      "epoch": 0.11265136662438012,
      "grad_norm": 1.910437132627685,
      "learning_rate": 1.2311005825998102e-06,
      "loss": 0.9121,
      "step": 1221
    },
    {
      "epoch": 0.11274362818590705,
      "grad_norm": 1.5798561671650437,
      "learning_rate": 1.2310546367342149e-06,
      "loss": 0.8384,
      "step": 1222
    },
    {
      "epoch": 0.11283588974743397,
      "grad_norm": 2.1210775495187972,
      "learning_rate": 1.2310086359469945e-06,
      "loss": 0.7574,
      "step": 1223
    },
    {
      "epoch": 0.1129281513089609,
      "grad_norm": 2.0998204169465673,
      "learning_rate": 1.2309625802423176e-06,
      "loss": 0.9323,
      "step": 1224
    },
    {
      "epoch": 0.11302041287048784,
      "grad_norm": 3.8016711561536463,
      "learning_rate": 1.230916469624358e-06,
      "loss": 0.8167,
      "step": 1225
    },
    {
      "epoch": 0.11302041287048784,
      "eval_GEN Loss": 0.48783984780311584,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4506700038909912,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9697115421295166,
      "eval_runtime": 56.7983,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1225
    },
    {
      "epoch": 0.11311267443201477,
      "grad_norm": 2.3009203906219935,
      "learning_rate": 1.2308703040972942e-06,
      "loss": 0.9366,
      "step": 1226
    },
    {
      "epoch": 0.11320493599354169,
      "grad_norm": 2.26488523449271,
      "learning_rate": 1.2308240836653096e-06,
      "loss": 0.7221,
      "step": 1227
    },
    {
      "epoch": 0.11329719755506862,
      "grad_norm": 1.8704250858740405,
      "learning_rate": 1.230777808332593e-06,
      "loss": 0.6154,
      "step": 1228
    },
    {
      "epoch": 0.11338945911659555,
      "grad_norm": 2.0560324644077466,
      "learning_rate": 1.230731478103338e-06,
      "loss": 0.8106,
      "step": 1229
    },
    {
      "epoch": 0.11348172067812248,
      "grad_norm": 1.495671876208774,
      "learning_rate": 1.230685092981743e-06,
      "loss": 0.7048,
      "step": 1230
    },
    {
      "epoch": 0.11348172067812248,
      "eval_GEN Loss": 0.4879695177078247,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.42036473751068115,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9837740659713745,
      "eval_runtime": 56.9633,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1230
    },
    {
      "epoch": 0.1135739822396494,
      "grad_norm": 1.9992911268614089,
      "learning_rate": 1.2306386529720115e-06,
      "loss": 0.6861,
      "step": 1231
    },
    {
      "epoch": 0.11366624380117633,
      "grad_norm": 2.1625798494717157,
      "learning_rate": 1.2305921580783519e-06,
      "loss": 0.8691,
      "step": 1232
    },
    {
      "epoch": 0.11375850536270327,
      "grad_norm": 1.7572110133120686,
      "learning_rate": 1.2305456083049778e-06,
      "loss": 0.6188,
      "step": 1233
    },
    {
      "epoch": 0.1138507669242302,
      "grad_norm": 3.990292125990343,
      "learning_rate": 1.2304990036561077e-06,
      "loss": 1.122,
      "step": 1234
    },
    {
      "epoch": 0.11394302848575712,
      "grad_norm": 1.362868875893105,
      "learning_rate": 1.2304523441359645e-06,
      "loss": 0.6572,
      "step": 1235
    },
    {
      "epoch": 0.11394302848575712,
      "eval_GEN Loss": 0.4927229881286621,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.41139060258865356,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9867788553237915,
      "eval_runtime": 55.8913,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1235
    },
    {
      "epoch": 0.11403529004728405,
      "grad_norm": 1.4789390013239685,
      "learning_rate": 1.230405629748777e-06,
      "loss": 0.7417,
      "step": 1236
    },
    {
      "epoch": 0.11412755160881098,
      "grad_norm": 1.8020256171051858,
      "learning_rate": 1.2303588604987785e-06,
      "loss": 0.7417,
      "step": 1237
    },
    {
      "epoch": 0.11421981317033791,
      "grad_norm": 1.9066295153227304,
      "learning_rate": 1.230312036390207e-06,
      "loss": 0.7737,
      "step": 1238
    },
    {
      "epoch": 0.11431207473186483,
      "grad_norm": 2.4963181073538094,
      "learning_rate": 1.2302651574273063e-06,
      "loss": 1.054,
      "step": 1239
    },
    {
      "epoch": 0.11440433629339176,
      "grad_norm": 1.8251129584340515,
      "learning_rate": 1.230218223614324e-06,
      "loss": 0.9767,
      "step": 1240
    },
    {
      "epoch": 0.11440433629339176,
      "eval_GEN Loss": 0.49484437704086304,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.4209262430667877,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9575120210647583,
      "eval_runtime": 55.8379,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1240
    },
    {
      "epoch": 0.1144965978549187,
      "grad_norm": 1.997936736599995,
      "learning_rate": 1.2301712349555139e-06,
      "loss": 0.7246,
      "step": 1241
    },
    {
      "epoch": 0.11458885941644563,
      "grad_norm": 1.4723677395700392,
      "learning_rate": 1.230124191455134e-06,
      "loss": 0.8091,
      "step": 1242
    },
    {
      "epoch": 0.11468112097797255,
      "grad_norm": 1.8669420475096228,
      "learning_rate": 1.2300770931174472e-06,
      "loss": 0.8244,
      "step": 1243
    },
    {
      "epoch": 0.11477338253949948,
      "grad_norm": 1.976773884958207,
      "learning_rate": 1.2300299399467217e-06,
      "loss": 0.7619,
      "step": 1244
    },
    {
      "epoch": 0.11486564410102641,
      "grad_norm": 1.7786316282322259,
      "learning_rate": 1.229982731947231e-06,
      "loss": 0.5921,
      "step": 1245
    },
    {
      "epoch": 0.11486564410102641,
      "eval_GEN Loss": 0.4953586459159851,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.44311127066612244,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9525240659713745,
      "eval_runtime": 55.9117,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1245
    },
    {
      "epoch": 0.11495790566255334,
      "grad_norm": 1.7248180124838284,
      "learning_rate": 1.2299354691232527e-06,
      "loss": 0.7807,
      "step": 1246
    },
    {
      "epoch": 0.11505016722408026,
      "grad_norm": 1.9239963241109062,
      "learning_rate": 1.22988815147907e-06,
      "loss": 0.7448,
      "step": 1247
    },
    {
      "epoch": 0.1151424287856072,
      "grad_norm": 2.1630647471198636,
      "learning_rate": 1.2298407790189709e-06,
      "loss": 0.9987,
      "step": 1248
    },
    {
      "epoch": 0.11523469034713413,
      "grad_norm": 1.8775242072610066,
      "learning_rate": 1.2297933517472484e-06,
      "loss": 0.8983,
      "step": 1249
    },
    {
      "epoch": 0.11532695190866106,
      "grad_norm": 1.9481258755149686,
      "learning_rate": 1.2297458696682005e-06,
      "loss": 0.9304,
      "step": 1250
    },
    {
      "epoch": 0.11532695190866106,
      "eval_GEN Loss": 0.49161943793296814,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4523923993110657,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.95703125,
      "eval_runtime": 56.0817,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1250
    },
    {
      "epoch": 0.11541921347018798,
      "grad_norm": 1.8349139608614415,
      "learning_rate": 1.22969833278613e-06,
      "loss": 0.6716,
      "step": 1251
    },
    {
      "epoch": 0.11551147503171491,
      "grad_norm": 1.5689091622911395,
      "learning_rate": 1.2296507411053446e-06,
      "loss": 0.7903,
      "step": 1252
    },
    {
      "epoch": 0.11560373659324184,
      "grad_norm": 1.864361596538834,
      "learning_rate": 1.2296030946301574e-06,
      "loss": 0.8044,
      "step": 1253
    },
    {
      "epoch": 0.11569599815476878,
      "grad_norm": 1.9153860482212435,
      "learning_rate": 1.2295553933648863e-06,
      "loss": 0.9074,
      "step": 1254
    },
    {
      "epoch": 0.1157882597162957,
      "grad_norm": 1.9224706653837949,
      "learning_rate": 1.2295076373138536e-06,
      "loss": 0.922,
      "step": 1255
    },
    {
      "epoch": 0.1157882597162957,
      "eval_GEN Loss": 0.49123042821884155,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.44373172521591187,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9667067527770996,
      "eval_runtime": 55.9882,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1255
    },
    {
      "epoch": 0.11588052127782263,
      "grad_norm": 1.5471438731740057,
      "learning_rate": 1.2294598264813873e-06,
      "loss": 0.9188,
      "step": 1256
    },
    {
      "epoch": 0.11597278283934956,
      "grad_norm": 1.7555395101693638,
      "learning_rate": 1.2294119608718203e-06,
      "loss": 0.839,
      "step": 1257
    },
    {
      "epoch": 0.11606504440087649,
      "grad_norm": 1.6358990907131337,
      "learning_rate": 1.22936404048949e-06,
      "loss": 0.8239,
      "step": 1258
    },
    {
      "epoch": 0.11615730596240341,
      "grad_norm": 1.8715056530450735,
      "learning_rate": 1.2293160653387388e-06,
      "loss": 0.8721,
      "step": 1259
    },
    {
      "epoch": 0.11624956752393034,
      "grad_norm": 1.7281595737040045,
      "learning_rate": 1.2292680354239148e-06,
      "loss": 0.7831,
      "step": 1260
    },
    {
      "epoch": 0.11624956752393034,
      "eval_GEN Loss": 0.4905729293823242,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4472057521343231,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9637620449066162,
      "eval_runtime": 56.0599,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1260
    },
    {
      "epoch": 0.11634182908545727,
      "grad_norm": 1.7609236740955532,
      "learning_rate": 1.2292199507493702e-06,
      "loss": 0.7484,
      "step": 1261
    },
    {
      "epoch": 0.1164340906469842,
      "grad_norm": 2.118717397078278,
      "learning_rate": 1.2291718113194627e-06,
      "loss": 0.8318,
      "step": 1262
    },
    {
      "epoch": 0.11652635220851112,
      "grad_norm": 1.8373932256046257,
      "learning_rate": 1.2291236171385545e-06,
      "loss": 0.7109,
      "step": 1263
    },
    {
      "epoch": 0.11661861377003806,
      "grad_norm": 1.7655496164280557,
      "learning_rate": 1.2290753682110135e-06,
      "loss": 0.7079,
      "step": 1264
    },
    {
      "epoch": 0.11671087533156499,
      "grad_norm": 2.5227832491472744,
      "learning_rate": 1.2290270645412116e-06,
      "loss": 0.7442,
      "step": 1265
    },
    {
      "epoch": 0.11671087533156499,
      "eval_GEN Loss": 0.48896458745002747,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.44111210107803345,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9471153616905212,
      "eval_runtime": 56.9285,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1265
    },
    {
      "epoch": 0.11680313689309192,
      "grad_norm": 1.7873531252217434,
      "learning_rate": 1.2289787061335263e-06,
      "loss": 0.7228,
      "step": 1266
    },
    {
      "epoch": 0.11689539845461884,
      "grad_norm": 2.8849635823358573,
      "learning_rate": 1.2289302929923402e-06,
      "loss": 0.8533,
      "step": 1267
    },
    {
      "epoch": 0.11698766001614577,
      "grad_norm": 2.4305007851692153,
      "learning_rate": 1.2288818251220403e-06,
      "loss": 0.8249,
      "step": 1268
    },
    {
      "epoch": 0.1170799215776727,
      "grad_norm": 1.6413552127999673,
      "learning_rate": 1.2288333025270185e-06,
      "loss": 0.7109,
      "step": 1269
    },
    {
      "epoch": 0.11717218313919964,
      "grad_norm": 2.397754173705168,
      "learning_rate": 1.2287847252116726e-06,
      "loss": 0.8528,
      "step": 1270
    },
    {
      "epoch": 0.11717218313919964,
      "eval_GEN Loss": 0.48914703726768494,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.40857037901878357,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9649038314819336,
      "eval_runtime": 55.9966,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1270
    },
    {
      "epoch": 0.11726444470072656,
      "grad_norm": 1.749884991453997,
      "learning_rate": 1.2287360931804048e-06,
      "loss": 0.9046,
      "step": 1271
    },
    {
      "epoch": 0.11735670626225349,
      "grad_norm": 3.3203167016211377,
      "learning_rate": 1.2286874064376217e-06,
      "loss": 0.9641,
      "step": 1272
    },
    {
      "epoch": 0.11744896782378042,
      "grad_norm": 1.78784685017401,
      "learning_rate": 1.2286386649877358e-06,
      "loss": 0.6087,
      "step": 1273
    },
    {
      "epoch": 0.11754122938530735,
      "grad_norm": 2.207589169562766,
      "learning_rate": 1.2285898688351638e-06,
      "loss": 0.788,
      "step": 1274
    },
    {
      "epoch": 0.11763349094683427,
      "grad_norm": 1.9101249677078853,
      "learning_rate": 1.2285410179843277e-06,
      "loss": 0.7491,
      "step": 1275
    },
    {
      "epoch": 0.11763349094683427,
      "eval_GEN Loss": 0.4912085235118866,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.39622050523757935,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.957932710647583,
      "eval_runtime": 57.1497,
      "eval_samples_per_second": 1.137,
      "eval_steps_per_second": 0.087,
      "step": 1275
    },
    {
      "epoch": 0.1177257525083612,
      "grad_norm": 2.6422083951311666,
      "learning_rate": 1.2284921124396547e-06,
      "loss": 0.7895,
      "step": 1276
    },
    {
      "epoch": 0.11781801406988814,
      "grad_norm": 2.46197267519697,
      "learning_rate": 1.2284431522055766e-06,
      "loss": 0.7814,
      "step": 1277
    },
    {
      "epoch": 0.11791027563141507,
      "grad_norm": 1.8120570948314536,
      "learning_rate": 1.2283941372865303e-06,
      "loss": 0.7436,
      "step": 1278
    },
    {
      "epoch": 0.11800253719294199,
      "grad_norm": 2.684960415278719,
      "learning_rate": 1.2283450676869573e-06,
      "loss": 0.7525,
      "step": 1279
    },
    {
      "epoch": 0.11809479875446892,
      "grad_norm": 2.506761052462561,
      "learning_rate": 1.2282959434113045e-06,
      "loss": 0.7506,
      "step": 1280
    },
    {
      "epoch": 0.11809479875446892,
      "eval_GEN Loss": 0.4923955202102661,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.4056079089641571,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9118990302085876,
      "eval_runtime": 56.9534,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1280
    },
    {
      "epoch": 0.11818706031599585,
      "grad_norm": 1.6064649037889431,
      "learning_rate": 1.2282467644640239e-06,
      "loss": 0.6396,
      "step": 1281
    },
    {
      "epoch": 0.11827932187752278,
      "grad_norm": 1.940776110107445,
      "learning_rate": 1.228197530849572e-06,
      "loss": 0.8193,
      "step": 1282
    },
    {
      "epoch": 0.1183715834390497,
      "grad_norm": 1.7846565497956695,
      "learning_rate": 1.2281482425724103e-06,
      "loss": 0.7188,
      "step": 1283
    },
    {
      "epoch": 0.11846384500057663,
      "grad_norm": 2.2620249600204994,
      "learning_rate": 1.2280988996370057e-06,
      "loss": 0.9059,
      "step": 1284
    },
    {
      "epoch": 0.11855610656210357,
      "grad_norm": 1.7475948681226834,
      "learning_rate": 1.2280495020478291e-06,
      "loss": 0.7448,
      "step": 1285
    },
    {
      "epoch": 0.11855610656210357,
      "eval_GEN Loss": 0.4951961636543274,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4066463112831116,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8948317170143127,
      "eval_runtime": 56.9624,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1285
    },
    {
      "epoch": 0.1186483681236305,
      "grad_norm": 1.326016769754368,
      "learning_rate": 1.2280000498093578e-06,
      "loss": 0.6742,
      "step": 1286
    },
    {
      "epoch": 0.11874062968515742,
      "grad_norm": 2.292162155898998,
      "learning_rate": 1.2279505429260729e-06,
      "loss": 0.803,
      "step": 1287
    },
    {
      "epoch": 0.11883289124668435,
      "grad_norm": 1.563686091518793,
      "learning_rate": 1.2279009814024605e-06,
      "loss": 0.7024,
      "step": 1288
    },
    {
      "epoch": 0.11892515280821128,
      "grad_norm": 2.5605781601401514,
      "learning_rate": 1.2278513652430123e-06,
      "loss": 0.836,
      "step": 1289
    },
    {
      "epoch": 0.11901741436973821,
      "grad_norm": 1.9957556185976306,
      "learning_rate": 1.2278016944522245e-06,
      "loss": 0.8837,
      "step": 1290
    },
    {
      "epoch": 0.11901741436973821,
      "eval_GEN Loss": 0.49639612436294556,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.38042151927948,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8947115540504456,
      "eval_runtime": 56.8549,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1290
    },
    {
      "epoch": 0.11910967593126513,
      "grad_norm": 3.0767161496384716,
      "learning_rate": 1.2277519690345982e-06,
      "loss": 0.9519,
      "step": 1291
    },
    {
      "epoch": 0.11920193749279206,
      "grad_norm": 2.022954445214826,
      "learning_rate": 1.2277021889946397e-06,
      "loss": 0.6947,
      "step": 1292
    },
    {
      "epoch": 0.119294199054319,
      "grad_norm": 1.9583881679622657,
      "learning_rate": 1.2276523543368602e-06,
      "loss": 0.7825,
      "step": 1293
    },
    {
      "epoch": 0.11938646061584593,
      "grad_norm": 1.5783633172665636,
      "learning_rate": 1.2276024650657758e-06,
      "loss": 0.6771,
      "step": 1294
    },
    {
      "epoch": 0.11947872217737285,
      "grad_norm": 1.6031041132358899,
      "learning_rate": 1.2275525211859076e-06,
      "loss": 0.6923,
      "step": 1295
    },
    {
      "epoch": 0.11947872217737285,
      "eval_GEN Loss": 0.49745118618011475,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3748480975627899,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9041466116905212,
      "eval_runtime": 56.9659,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1295
    },
    {
      "epoch": 0.11957098373889978,
      "grad_norm": 1.573106670247981,
      "learning_rate": 1.2275025227017813e-06,
      "loss": 0.8175,
      "step": 1296
    },
    {
      "epoch": 0.11966324530042671,
      "grad_norm": 2.8140865304331673,
      "learning_rate": 1.2274524696179283e-06,
      "loss": 1.1144,
      "step": 1297
    },
    {
      "epoch": 0.11975550686195365,
      "grad_norm": 1.5626355365290727,
      "learning_rate": 1.227402361938884e-06,
      "loss": 0.6706,
      "step": 1298
    },
    {
      "epoch": 0.11984776842348056,
      "grad_norm": 2.2106948655961323,
      "learning_rate": 1.2273521996691895e-06,
      "loss": 0.7366,
      "step": 1299
    },
    {
      "epoch": 0.1199400299850075,
      "grad_norm": 3.611276867720812,
      "learning_rate": 1.2273019828133908e-06,
      "loss": 0.8502,
      "step": 1300
    },
    {
      "epoch": 0.1199400299850075,
      "eval_GEN Loss": 0.49392223358154297,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.37433359026908875,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9100360870361328,
      "eval_runtime": 56.978,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1300
    },
    {
      "epoch": 0.12003229154653443,
      "grad_norm": 2.1935284161123323,
      "learning_rate": 1.227251711376038e-06,
      "loss": 0.9244,
      "step": 1301
    },
    {
      "epoch": 0.12012455310806136,
      "grad_norm": 2.28938953997289,
      "learning_rate": 1.2272013853616874e-06,
      "loss": 0.6428,
      "step": 1302
    },
    {
      "epoch": 0.12021681466958828,
      "grad_norm": 1.8831071642607708,
      "learning_rate": 1.2271510047748994e-06,
      "loss": 0.5655,
      "step": 1303
    },
    {
      "epoch": 0.12030907623111521,
      "grad_norm": 2.7674605841492,
      "learning_rate": 1.2271005696202394e-06,
      "loss": 0.8868,
      "step": 1304
    },
    {
      "epoch": 0.12040133779264214,
      "grad_norm": 2.1277466206939586,
      "learning_rate": 1.227050079902278e-06,
      "loss": 0.8592,
      "step": 1305
    },
    {
      "epoch": 0.12040133779264214,
      "eval_GEN Loss": 0.49284905195236206,
      "eval_GEN top-5 accuracy": 0.9780797895659799,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.398231565952301,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8812500238418579,
      "eval_runtime": 56.8489,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1305
    },
    {
      "epoch": 0.12049359935416908,
      "grad_norm": 3.230113473102956,
      "learning_rate": 1.226999535625591e-06,
      "loss": 1.0419,
      "step": 1306
    },
    {
      "epoch": 0.120585860915696,
      "grad_norm": 1.6201613850118866,
      "learning_rate": 1.2269489367947582e-06,
      "loss": 0.8235,
      "step": 1307
    },
    {
      "epoch": 0.12067812247722293,
      "grad_norm": 2.264023013262882,
      "learning_rate": 1.2268982834143656e-06,
      "loss": 0.8541,
      "step": 1308
    },
    {
      "epoch": 0.12077038403874986,
      "grad_norm": 2.421431406370077,
      "learning_rate": 1.2268475754890029e-06,
      "loss": 0.962,
      "step": 1309
    },
    {
      "epoch": 0.12086264560027679,
      "grad_norm": 1.7229390248536531,
      "learning_rate": 1.2267968130232657e-06,
      "loss": 0.712,
      "step": 1310
    },
    {
      "epoch": 0.12086264560027679,
      "eval_GEN Loss": 0.49430614709854126,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8552631578947368,
      "eval_PRM F1 AUC": 0.8046097433211105,
      "eval_PRM F1 Neg": 0.6333333333333333,
      "eval_PRM Loss": 0.47042927145957947,
      "eval_PRM NPV": 0.5135135135135135,
      "eval_PRM Precision": 0.9420289855072463,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.9013221263885498,
      "eval_runtime": 57.0128,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1310
    },
    {
      "epoch": 0.12095490716180371,
      "grad_norm": 2.2250996507214276,
      "learning_rate": 1.226745996021754e-06,
      "loss": 0.8493,
      "step": 1311
    },
    {
      "epoch": 0.12104716872333064,
      "grad_norm": 1.8707936283098114,
      "learning_rate": 1.226695124489073e-06,
      "loss": 0.6961,
      "step": 1312
    },
    {
      "epoch": 0.12113943028485757,
      "grad_norm": 3.3417424110119347,
      "learning_rate": 1.2266441984298327e-06,
      "loss": 0.9266,
      "step": 1313
    },
    {
      "epoch": 0.1212316918463845,
      "grad_norm": 2.5732522872035446,
      "learning_rate": 1.2265932178486483e-06,
      "loss": 0.7325,
      "step": 1314
    },
    {
      "epoch": 0.12132395340791143,
      "grad_norm": 1.9018229284694772,
      "learning_rate": 1.2265421827501395e-06,
      "loss": 0.7876,
      "step": 1315
    },
    {
      "epoch": 0.12132395340791143,
      "eval_GEN Loss": 0.49222859740257263,
      "eval_GEN top-5 accuracy": 0.9789565979833407,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8774193548387097,
      "eval_PRM F1 AUC": 0.8226820324777371,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.4281691312789917,
      "eval_PRM NPV": 0.5588235294117647,
      "eval_PRM Precision": 0.9444444444444444,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.8835336565971375,
      "eval_runtime": 56.8326,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1315
    },
    {
      "epoch": 0.12141621496943836,
      "grad_norm": 1.8488596501663879,
      "learning_rate": 1.2264910931389312e-06,
      "loss": 0.7547,
      "step": 1316
    },
    {
      "epoch": 0.12150847653096529,
      "grad_norm": 1.8484299160489088,
      "learning_rate": 1.2264399490196533e-06,
      "loss": 0.8028,
      "step": 1317
    },
    {
      "epoch": 0.12160073809249222,
      "grad_norm": 1.5443884201729963,
      "learning_rate": 1.2263887503969405e-06,
      "loss": 0.6124,
      "step": 1318
    },
    {
      "epoch": 0.12169299965401914,
      "grad_norm": 1.5278949700900137,
      "learning_rate": 1.2263374972754324e-06,
      "loss": 0.7439,
      "step": 1319
    },
    {
      "epoch": 0.12178526121554607,
      "grad_norm": 1.8514423421923552,
      "learning_rate": 1.226286189659774e-06,
      "loss": 0.8699,
      "step": 1320
    },
    {
      "epoch": 0.12178526121554607,
      "eval_GEN Loss": 0.49053096771240234,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.38080504536628723,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9111778736114502,
      "eval_runtime": 56.9873,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1320
    },
    {
      "epoch": 0.121877522777073,
      "grad_norm": 2.565714757651536,
      "learning_rate": 1.2262348275546145e-06,
      "loss": 0.8825,
      "step": 1321
    },
    {
      "epoch": 0.12196978433859994,
      "grad_norm": 1.7304304013800949,
      "learning_rate": 1.226183410964609e-06,
      "loss": 0.7343,
      "step": 1322
    },
    {
      "epoch": 0.12206204590012686,
      "grad_norm": 2.51550071315553,
      "learning_rate": 1.226131939894416e-06,
      "loss": 1.048,
      "step": 1323
    },
    {
      "epoch": 0.12215430746165379,
      "grad_norm": 2.3891342523973447,
      "learning_rate": 1.2260804143487007e-06,
      "loss": 0.8047,
      "step": 1324
    },
    {
      "epoch": 0.12224656902318072,
      "grad_norm": 1.8475248227200738,
      "learning_rate": 1.226028834332132e-06,
      "loss": 0.6824,
      "step": 1325
    },
    {
      "epoch": 0.12224656902318072,
      "eval_GEN Loss": 0.49069076776504517,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3912678062915802,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9151442050933838,
      "eval_runtime": 56.8241,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1325
    },
    {
      "epoch": 0.12233883058470765,
      "grad_norm": 1.779257449054373,
      "learning_rate": 1.2259771998493845e-06,
      "loss": 0.5389,
      "step": 1326
    },
    {
      "epoch": 0.12243109214623457,
      "grad_norm": 3.5142895414815505,
      "learning_rate": 1.2259255109051369e-06,
      "loss": 1.0975,
      "step": 1327
    },
    {
      "epoch": 0.1225233537077615,
      "grad_norm": 2.0668360555397824,
      "learning_rate": 1.2258737675040738e-06,
      "loss": 0.818,
      "step": 1328
    },
    {
      "epoch": 0.12261561526928844,
      "grad_norm": 1.6280634119087998,
      "learning_rate": 1.225821969650884e-06,
      "loss": 0.9491,
      "step": 1329
    },
    {
      "epoch": 0.12270787683081537,
      "grad_norm": 1.9001001064855316,
      "learning_rate": 1.2257701173502618e-06,
      "loss": 0.8077,
      "step": 1330
    },
    {
      "epoch": 0.12270787683081537,
      "eval_GEN Loss": 0.48902031779289246,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3988769054412842,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9060696959495544,
      "eval_runtime": 56.8036,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1330
    },
    {
      "epoch": 0.12280013839234229,
      "grad_norm": 1.6537615151434861,
      "learning_rate": 1.2257182106069056e-06,
      "loss": 0.6946,
      "step": 1331
    },
    {
      "epoch": 0.12289239995386922,
      "grad_norm": 1.8871678927422304,
      "learning_rate": 1.22566624942552e-06,
      "loss": 0.7479,
      "step": 1332
    },
    {
      "epoch": 0.12298466151539615,
      "grad_norm": 1.979539107160251,
      "learning_rate": 1.2256142338108132e-06,
      "loss": 0.7868,
      "step": 1333
    },
    {
      "epoch": 0.12307692307692308,
      "grad_norm": 2.4620381895051557,
      "learning_rate": 1.2255621637674991e-06,
      "loss": 0.9231,
      "step": 1334
    },
    {
      "epoch": 0.12316918463845,
      "grad_norm": 2.5053203833222666,
      "learning_rate": 1.2255100393002967e-06,
      "loss": 0.8815,
      "step": 1335
    },
    {
      "epoch": 0.12316918463845,
      "eval_GEN Loss": 0.4872274696826935,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4308062493801117,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.907151460647583,
      "eval_runtime": 57.2071,
      "eval_samples_per_second": 1.136,
      "eval_steps_per_second": 0.087,
      "step": 1335
    },
    {
      "epoch": 0.12326144619997693,
      "grad_norm": 2.4154779428773003,
      "learning_rate": 1.225457860413929e-06,
      "loss": 0.8436,
      "step": 1336
    },
    {
      "epoch": 0.12335370776150387,
      "grad_norm": 1.8829189925644858,
      "learning_rate": 1.225405627113125e-06,
      "loss": 0.8892,
      "step": 1337
    },
    {
      "epoch": 0.1234459693230308,
      "grad_norm": 1.7127274025772998,
      "learning_rate": 1.2253533394026178e-06,
      "loss": 0.4481,
      "step": 1338
    },
    {
      "epoch": 0.12353823088455772,
      "grad_norm": 2.03664475177838,
      "learning_rate": 1.2253009972871463e-06,
      "loss": 0.7433,
      "step": 1339
    },
    {
      "epoch": 0.12363049244608465,
      "grad_norm": 1.7966903993830157,
      "learning_rate": 1.2252486007714535e-06,
      "loss": 0.8347,
      "step": 1340
    },
    {
      "epoch": 0.12363049244608465,
      "eval_GEN Loss": 0.48570749163627625,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.4023730754852295,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9155048131942749,
      "eval_runtime": 56.9631,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1340
    },
    {
      "epoch": 0.12372275400761158,
      "grad_norm": 1.5100872404160781,
      "learning_rate": 1.2251961498602876e-06,
      "loss": 0.7208,
      "step": 1341
    },
    {
      "epoch": 0.12381501556913851,
      "grad_norm": 1.1805249230992214,
      "learning_rate": 1.2251436445584019e-06,
      "loss": 0.5913,
      "step": 1342
    },
    {
      "epoch": 0.12390727713066543,
      "grad_norm": 2.000992335213143,
      "learning_rate": 1.2250910848705546e-06,
      "loss": 0.8445,
      "step": 1343
    },
    {
      "epoch": 0.12399953869219237,
      "grad_norm": 1.5996161674342857,
      "learning_rate": 1.2250384708015087e-06,
      "loss": 0.7581,
      "step": 1344
    },
    {
      "epoch": 0.1240918002537193,
      "grad_norm": 3.241294099197593,
      "learning_rate": 1.2249858023560317e-06,
      "loss": 1.0236,
      "step": 1345
    },
    {
      "epoch": 0.1240918002537193,
      "eval_GEN Loss": 0.4876135289669037,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8928571428571429,
      "eval_PRM F1 AUC": 0.7344159245678366,
      "eval_PRM F1 Neg": 0.5909090909090909,
      "eval_PRM Loss": 0.39364543557167053,
      "eval_PRM NPV": 0.6190476190476191,
      "eval_PRM Precision": 0.8823529411764706,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9313101172447205,
      "eval_runtime": 56.8033,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1345
    },
    {
      "epoch": 0.12418406181524622,
      "grad_norm": 1.9905229496189492,
      "learning_rate": 1.2249330795388971e-06,
      "loss": 0.8514,
      "step": 1346
    },
    {
      "epoch": 0.12427632337677315,
      "grad_norm": 1.571750154821204,
      "learning_rate": 1.2248803023548827e-06,
      "loss": 0.7193,
      "step": 1347
    },
    {
      "epoch": 0.12436858493830008,
      "grad_norm": 1.685205589806922,
      "learning_rate": 1.224827470808771e-06,
      "loss": 0.5701,
      "step": 1348
    },
    {
      "epoch": 0.12446084649982701,
      "grad_norm": 2.394225354378885,
      "learning_rate": 1.2247745849053497e-06,
      "loss": 0.9537,
      "step": 1349
    },
    {
      "epoch": 0.12455310806135393,
      "grad_norm": 2.3157824689347515,
      "learning_rate": 1.2247216446494115e-06,
      "loss": 0.6846,
      "step": 1350
    },
    {
      "epoch": 0.12455310806135393,
      "eval_GEN Loss": 0.490536630153656,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.40498775243759155,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.920192301273346,
      "eval_runtime": 56.8933,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1350
    },
    {
      "epoch": 0.12464536962288086,
      "grad_norm": 1.3484877227150613,
      "learning_rate": 1.2246686500457539e-06,
      "loss": 0.7361,
      "step": 1351
    },
    {
      "epoch": 0.1247376311844078,
      "grad_norm": 1.462469573473979,
      "learning_rate": 1.2246156010991791e-06,
      "loss": 0.7746,
      "step": 1352
    },
    {
      "epoch": 0.12482989274593473,
      "grad_norm": 2.441359574282289,
      "learning_rate": 1.224562497814495e-06,
      "loss": 0.68,
      "step": 1353
    },
    {
      "epoch": 0.12492215430746165,
      "grad_norm": 1.4352457642433032,
      "learning_rate": 1.2245093401965136e-06,
      "loss": 0.6796,
      "step": 1354
    },
    {
      "epoch": 0.1250144158689886,
      "grad_norm": 2.0719272539069866,
      "learning_rate": 1.224456128250052e-06,
      "loss": 0.822,
      "step": 1355
    },
    {
      "epoch": 0.1250144158689886,
      "eval_GEN Loss": 0.493169367313385,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.40531107783317566,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9220553040504456,
      "eval_runtime": 55.9278,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1355
    },
    {
      "epoch": 0.1251066774305155,
      "grad_norm": 1.6547964849612078,
      "learning_rate": 1.2244028619799326e-06,
      "loss": 0.7411,
      "step": 1356
    },
    {
      "epoch": 0.12519893899204243,
      "grad_norm": 1.9780433262016586,
      "learning_rate": 1.2243495413909823e-06,
      "loss": 0.6415,
      "step": 1357
    },
    {
      "epoch": 0.12529120055356938,
      "grad_norm": 1.6968738629275064,
      "learning_rate": 1.2242961664880332e-06,
      "loss": 0.6864,
      "step": 1358
    },
    {
      "epoch": 0.1253834621150963,
      "grad_norm": 3.751259488620392,
      "learning_rate": 1.2242427372759222e-06,
      "loss": 0.9207,
      "step": 1359
    },
    {
      "epoch": 0.1254757236766232,
      "grad_norm": 1.2764892118359368,
      "learning_rate": 1.2241892537594912e-06,
      "loss": 0.6415,
      "step": 1360
    },
    {
      "epoch": 0.1254757236766232,
      "eval_GEN Loss": 0.492832213640213,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.4038274884223938,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9268028736114502,
      "eval_runtime": 56.1999,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 1360
    },
    {
      "epoch": 0.12556798523815016,
      "grad_norm": 2.1026123520145767,
      "learning_rate": 1.2241357159435868e-06,
      "loss": 0.8027,
      "step": 1361
    },
    {
      "epoch": 0.12566024679967708,
      "grad_norm": 2.5842829641384437,
      "learning_rate": 1.2240821238330608e-06,
      "loss": 0.7074,
      "step": 1362
    },
    {
      "epoch": 0.12575250836120402,
      "grad_norm": 1.9007473539853121,
      "learning_rate": 1.2240284774327698e-06,
      "loss": 0.873,
      "step": 1363
    },
    {
      "epoch": 0.12584476992273094,
      "grad_norm": 2.5262611850282233,
      "learning_rate": 1.2239747767475752e-06,
      "loss": 0.6985,
      "step": 1364
    },
    {
      "epoch": 0.12593703148425786,
      "grad_norm": 2.2135095209477913,
      "learning_rate": 1.2239210217823435e-06,
      "loss": 0.9369,
      "step": 1365
    },
    {
      "epoch": 0.12593703148425786,
      "eval_GEN Loss": 0.49305102229118347,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.4163629710674286,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9364783763885498,
      "eval_runtime": 55.9229,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1365
    },
    {
      "epoch": 0.1260292930457848,
      "grad_norm": 1.8745471552444068,
      "learning_rate": 1.2238672125419462e-06,
      "loss": 0.6284,
      "step": 1366
    },
    {
      "epoch": 0.12612155460731173,
      "grad_norm": 2.8458876213877007,
      "learning_rate": 1.2238133490312594e-06,
      "loss": 0.8194,
      "step": 1367
    },
    {
      "epoch": 0.12621381616883864,
      "grad_norm": 1.9038505262417882,
      "learning_rate": 1.2237594312551644e-06,
      "loss": 0.8271,
      "step": 1368
    },
    {
      "epoch": 0.1263060777303656,
      "grad_norm": 1.6360322936279517,
      "learning_rate": 1.2237054592185475e-06,
      "loss": 0.8412,
      "step": 1369
    },
    {
      "epoch": 0.1263983392918925,
      "grad_norm": 2.344156408789055,
      "learning_rate": 1.2236514329262992e-06,
      "loss": 0.8236,
      "step": 1370
    },
    {
      "epoch": 0.1263983392918925,
      "eval_GEN Loss": 0.4930301010608673,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4327784776687622,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9435697197914124,
      "eval_runtime": 55.9203,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1370
    },
    {
      "epoch": 0.12649060085341945,
      "grad_norm": 1.6354513496716774,
      "learning_rate": 1.2235973523833156e-06,
      "loss": 0.6719,
      "step": 1371
    },
    {
      "epoch": 0.12658286241494637,
      "grad_norm": 2.511174037923046,
      "learning_rate": 1.223543217594498e-06,
      "loss": 0.7847,
      "step": 1372
    },
    {
      "epoch": 0.1266751239764733,
      "grad_norm": 2.3701663942762456,
      "learning_rate": 1.2234890285647518e-06,
      "loss": 1.0791,
      "step": 1373
    },
    {
      "epoch": 0.12676738553800024,
      "grad_norm": 1.7896499207358794,
      "learning_rate": 1.2234347852989881e-06,
      "loss": 0.9398,
      "step": 1374
    },
    {
      "epoch": 0.12685964709952716,
      "grad_norm": 2.4965286537941127,
      "learning_rate": 1.2233804878021218e-06,
      "loss": 0.9747,
      "step": 1375
    },
    {
      "epoch": 0.12685964709952716,
      "eval_GEN Loss": 0.4926159977912903,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4483198821544647,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9524038434028625,
      "eval_runtime": 56.9883,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1375
    },
    {
      "epoch": 0.12695190866105407,
      "grad_norm": 1.4924837656210077,
      "learning_rate": 1.223326136079074e-06,
      "loss": 0.6207,
      "step": 1376
    },
    {
      "epoch": 0.12704417022258102,
      "grad_norm": 3.2142046158407096,
      "learning_rate": 1.22327173013477e-06,
      "loss": 0.8013,
      "step": 1377
    },
    {
      "epoch": 0.12713643178410794,
      "grad_norm": 1.8665806346977394,
      "learning_rate": 1.22321726997414e-06,
      "loss": 0.7902,
      "step": 1378
    },
    {
      "epoch": 0.12722869334563489,
      "grad_norm": 1.6012907009929764,
      "learning_rate": 1.2231627556021196e-06,
      "loss": 0.8071,
      "step": 1379
    },
    {
      "epoch": 0.1273209549071618,
      "grad_norm": 1.6471852536792064,
      "learning_rate": 1.2231081870236487e-06,
      "loss": 0.6864,
      "step": 1380
    },
    {
      "epoch": 0.1273209549071618,
      "eval_GEN Loss": 0.4910682141780853,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.42695653438568115,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9421274065971375,
      "eval_runtime": 56.876,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1380
    },
    {
      "epoch": 0.12741321646868872,
      "grad_norm": 3.352318014291917,
      "learning_rate": 1.2230535642436722e-06,
      "loss": 0.8315,
      "step": 1381
    },
    {
      "epoch": 0.12750547803021567,
      "grad_norm": 1.856451818849259,
      "learning_rate": 1.2229988872671405e-06,
      "loss": 0.806,
      "step": 1382
    },
    {
      "epoch": 0.1275977395917426,
      "grad_norm": 2.5806734605644337,
      "learning_rate": 1.2229441560990085e-06,
      "loss": 0.9162,
      "step": 1383
    },
    {
      "epoch": 0.1276900011532695,
      "grad_norm": 1.4628711909970085,
      "learning_rate": 1.222889370744236e-06,
      "loss": 0.8331,
      "step": 1384
    },
    {
      "epoch": 0.12778226271479645,
      "grad_norm": 1.6707117807192426,
      "learning_rate": 1.2228345312077875e-06,
      "loss": 0.7961,
      "step": 1385
    },
    {
      "epoch": 0.12778226271479645,
      "eval_GEN Loss": 0.4901101589202881,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4238015413284302,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9219951629638672,
      "eval_runtime": 57.0429,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 1385
    },
    {
      "epoch": 0.12787452427632337,
      "grad_norm": 1.7394289318327385,
      "learning_rate": 1.2227796374946326e-06,
      "loss": 0.6722,
      "step": 1386
    },
    {
      "epoch": 0.12796678583785032,
      "grad_norm": 1.6752054740440538,
      "learning_rate": 1.222724689609746e-06,
      "loss": 0.9017,
      "step": 1387
    },
    {
      "epoch": 0.12805904739937723,
      "grad_norm": 2.077628593692242,
      "learning_rate": 1.2226696875581075e-06,
      "loss": 0.814,
      "step": 1388
    },
    {
      "epoch": 0.12815130896090415,
      "grad_norm": 2.8099615886878313,
      "learning_rate": 1.2226146313447009e-06,
      "loss": 0.9671,
      "step": 1389
    },
    {
      "epoch": 0.1282435705224311,
      "grad_norm": 2.0630872164343446,
      "learning_rate": 1.222559520974516e-06,
      "loss": 0.8071,
      "step": 1390
    },
    {
      "epoch": 0.1282435705224311,
      "eval_GEN Loss": 0.48906949162483215,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.40765708684921265,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9087740182876587,
      "eval_runtime": 57.318,
      "eval_samples_per_second": 1.134,
      "eval_steps_per_second": 0.087,
      "step": 1390
    },
    {
      "epoch": 0.12833583208395802,
      "grad_norm": 1.457275792782061,
      "learning_rate": 1.2225043564525465e-06,
      "loss": 0.6686,
      "step": 1391
    },
    {
      "epoch": 0.12842809364548494,
      "grad_norm": 2.1525328381332693,
      "learning_rate": 1.222449137783792e-06,
      "loss": 0.8122,
      "step": 1392
    },
    {
      "epoch": 0.12852035520701188,
      "grad_norm": 2.2700897179949786,
      "learning_rate": 1.222393864973256e-06,
      "loss": 1.0304,
      "step": 1393
    },
    {
      "epoch": 0.1286126167685388,
      "grad_norm": 2.6915181959518706,
      "learning_rate": 1.2223385380259478e-06,
      "loss": 0.8429,
      "step": 1394
    },
    {
      "epoch": 0.12870487833006575,
      "grad_norm": 2.2887944308865253,
      "learning_rate": 1.2222831569468808e-06,
      "loss": 0.7829,
      "step": 1395
    },
    {
      "epoch": 0.12870487833006575,
      "eval_GEN Loss": 0.48873692750930786,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8606060606060606,
      "eval_PRM F1 AUC": 0.6885804085908852,
      "eval_PRM F1 Neg": 0.5106382978723404,
      "eval_PRM Loss": 0.4033927321434021,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8658536585365854,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9007812738418579,
      "eval_runtime": 55.9162,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1395
    },
    {
      "epoch": 0.12879713989159267,
      "grad_norm": 1.3869280208113692,
      "learning_rate": 1.2222277217410742e-06,
      "loss": 0.5959,
      "step": 1396
    },
    {
      "epoch": 0.12888940145311958,
      "grad_norm": 2.0303859235387063,
      "learning_rate": 1.2221722324135513e-06,
      "loss": 0.7309,
      "step": 1397
    },
    {
      "epoch": 0.12898166301464653,
      "grad_norm": 2.2558662821989826,
      "learning_rate": 1.2221166889693408e-06,
      "loss": 0.9391,
      "step": 1398
    },
    {
      "epoch": 0.12907392457617345,
      "grad_norm": 2.499820992030088,
      "learning_rate": 1.222061091413476e-06,
      "loss": 1.0213,
      "step": 1399
    },
    {
      "epoch": 0.12916618613770037,
      "grad_norm": 2.1620822515755163,
      "learning_rate": 1.2220054397509956e-06,
      "loss": 0.5987,
      "step": 1400
    },
    {
      "epoch": 0.12916618613770037,
      "eval_GEN Loss": 0.48808538913726807,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8941176470588236,
      "eval_PRM F1 AUC": 0.7187008905185961,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.394230455160141,
      "eval_PRM NPV": 0.631578947368421,
      "eval_PRM Precision": 0.8735632183908046,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9165865182876587,
      "eval_runtime": 55.8693,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1400
    },
    {
      "epoch": 0.1292584476992273,
      "grad_norm": 1.8023173226208795,
      "learning_rate": 1.2219497339869422e-06,
      "loss": 0.8016,
      "step": 1401
    },
    {
      "epoch": 0.12935070926075423,
      "grad_norm": 1.4360042587144939,
      "learning_rate": 1.2218939741263645e-06,
      "loss": 0.7909,
      "step": 1402
    },
    {
      "epoch": 0.12944297082228118,
      "grad_norm": 2.2181041510646957,
      "learning_rate": 1.2218381601743152e-06,
      "loss": 0.82,
      "step": 1403
    },
    {
      "epoch": 0.1295352323838081,
      "grad_norm": 2.3711314095201597,
      "learning_rate": 1.2217822921358525e-06,
      "loss": 0.7794,
      "step": 1404
    },
    {
      "epoch": 0.12962749394533502,
      "grad_norm": 3.077543869031744,
      "learning_rate": 1.221726370016039e-06,
      "loss": 1.0432,
      "step": 1405
    },
    {
      "epoch": 0.12962749394533502,
      "eval_GEN Loss": 0.4886215031147003,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8875739644970414,
      "eval_PRM F1 AUC": 0.7126767941330541,
      "eval_PRM F1 Neg": 0.5581395348837209,
      "eval_PRM Loss": 0.39263656735420227,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.872093023255814,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9164062738418579,
      "eval_runtime": 56.9822,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1405
    },
    {
      "epoch": 0.12971975550686196,
      "grad_norm": 1.493004898420434,
      "learning_rate": 1.2216703938199427e-06,
      "loss": 0.7421,
      "step": 1406
    },
    {
      "epoch": 0.12981201706838888,
      "grad_norm": 3.166827384383026,
      "learning_rate": 1.2216143635526363e-06,
      "loss": 0.9538,
      "step": 1407
    },
    {
      "epoch": 0.1299042786299158,
      "grad_norm": 2.080544775842973,
      "learning_rate": 1.221558279219197e-06,
      "loss": 0.8079,
      "step": 1408
    },
    {
      "epoch": 0.12999654019144274,
      "grad_norm": 2.0712815496356627,
      "learning_rate": 1.2215021408247073e-06,
      "loss": 0.8785,
      "step": 1409
    },
    {
      "epoch": 0.13008880175296966,
      "grad_norm": 2.2258005591246155,
      "learning_rate": 1.2214459483742548e-06,
      "loss": 0.9552,
      "step": 1410
    },
    {
      "epoch": 0.13008880175296966,
      "eval_GEN Loss": 0.4911350607872009,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8536585365853658,
      "eval_PRM F1 AUC": 0.6825563122053432,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM Loss": 0.39618173241615295,
      "eval_PRM NPV": 0.48,
      "eval_PRM Precision": 0.8641975308641975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.8948917984962463,
      "eval_runtime": 56.9222,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1410
    },
    {
      "epoch": 0.1301810633144966,
      "grad_norm": 1.725193134210622,
      "learning_rate": 1.2213897018729317e-06,
      "loss": 0.7585,
      "step": 1411
    },
    {
      "epoch": 0.13027332487602353,
      "grad_norm": 2.4539206060217524,
      "learning_rate": 1.221333401325835e-06,
      "loss": 0.8237,
      "step": 1412
    },
    {
      "epoch": 0.13036558643755045,
      "grad_norm": 2.4648471038553588,
      "learning_rate": 1.2212770467380668e-06,
      "loss": 1.0025,
      "step": 1413
    },
    {
      "epoch": 0.1304578479990774,
      "grad_norm": 1.943709308158613,
      "learning_rate": 1.2212206381147344e-06,
      "loss": 0.8606,
      "step": 1414
    },
    {
      "epoch": 0.1305501095606043,
      "grad_norm": 1.6151719683712205,
      "learning_rate": 1.221164175460949e-06,
      "loss": 0.8879,
      "step": 1415
    },
    {
      "epoch": 0.1305501095606043,
      "eval_GEN Loss": 0.49076810479164124,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.4118606150150299,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8805288672447205,
      "eval_runtime": 55.9309,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1415
    },
    {
      "epoch": 0.13064237112213123,
      "grad_norm": 1.8935609703014025,
      "learning_rate": 1.2211076587818276e-06,
      "loss": 1.0113,
      "step": 1416
    },
    {
      "epoch": 0.13073463268365818,
      "grad_norm": 1.7868360237988912,
      "learning_rate": 1.221051088082492e-06,
      "loss": 0.7498,
      "step": 1417
    },
    {
      "epoch": 0.1308268942451851,
      "grad_norm": 2.6406930623671157,
      "learning_rate": 1.2209944633680686e-06,
      "loss": 0.7203,
      "step": 1418
    },
    {
      "epoch": 0.13091915580671204,
      "grad_norm": 1.6732757519263721,
      "learning_rate": 1.2209377846436888e-06,
      "loss": 0.6282,
      "step": 1419
    },
    {
      "epoch": 0.13101141736823896,
      "grad_norm": 1.4389012831970156,
      "learning_rate": 1.220881051914489e-06,
      "loss": 0.6431,
      "step": 1420
    },
    {
      "epoch": 0.13101141736823896,
      "eval_GEN Loss": 0.4901523292064667,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4123425781726837,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8772836327552795,
      "eval_runtime": 55.8691,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1420
    },
    {
      "epoch": 0.13110367892976588,
      "grad_norm": 1.9077427061357575,
      "learning_rate": 1.2208242651856102e-06,
      "loss": 0.7044,
      "step": 1421
    },
    {
      "epoch": 0.13119594049129282,
      "grad_norm": 1.6201455051099196,
      "learning_rate": 1.2207674244621986e-06,
      "loss": 0.8841,
      "step": 1422
    },
    {
      "epoch": 0.13128820205281974,
      "grad_norm": 2.61301669288379,
      "learning_rate": 1.2207105297494054e-06,
      "loss": 0.907,
      "step": 1423
    },
    {
      "epoch": 0.13138046361434666,
      "grad_norm": 1.5551235813856161,
      "learning_rate": 1.2206535810523865e-06,
      "loss": 0.7483,
      "step": 1424
    },
    {
      "epoch": 0.1314727251758736,
      "grad_norm": 1.9306786971149377,
      "learning_rate": 1.220596578376302e-06,
      "loss": 0.7005,
      "step": 1425
    },
    {
      "epoch": 0.1314727251758736,
      "eval_GEN Loss": 0.48764145374298096,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.3840732276439667,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8736779093742371,
      "eval_runtime": 56.0107,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1425
    },
    {
      "epoch": 0.13156498673740052,
      "grad_norm": 1.4270304714741135,
      "learning_rate": 1.2205395217263186e-06,
      "loss": 0.6763,
      "step": 1426
    },
    {
      "epoch": 0.13165724829892747,
      "grad_norm": 2.30788141561209,
      "learning_rate": 1.220482411107606e-06,
      "loss": 0.9192,
      "step": 1427
    },
    {
      "epoch": 0.1317495098604544,
      "grad_norm": 2.7212851138373955,
      "learning_rate": 1.2204252465253403e-06,
      "loss": 0.9717,
      "step": 1428
    },
    {
      "epoch": 0.1318417714219813,
      "grad_norm": 2.230544584496275,
      "learning_rate": 1.2203680279847014e-06,
      "loss": 0.8579,
      "step": 1429
    },
    {
      "epoch": 0.13193403298350825,
      "grad_norm": 1.640629308733161,
      "learning_rate": 1.2203107554908746e-06,
      "loss": 0.8892,
      "step": 1430
    },
    {
      "epoch": 0.13193403298350825,
      "eval_GEN Loss": 0.48576074838638306,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3708195686340332,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8837740421295166,
      "eval_runtime": 55.9978,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1430
    },
    {
      "epoch": 0.13202629454503517,
      "grad_norm": 2.8181086708700076,
      "learning_rate": 1.2202534290490503e-06,
      "loss": 0.7244,
      "step": 1431
    },
    {
      "epoch": 0.1321185561065621,
      "grad_norm": 2.2999219725297078,
      "learning_rate": 1.2201960486644233e-06,
      "loss": 0.6805,
      "step": 1432
    },
    {
      "epoch": 0.13221081766808904,
      "grad_norm": 3.036926360636184,
      "learning_rate": 1.2201386143421933e-06,
      "loss": 0.7745,
      "step": 1433
    },
    {
      "epoch": 0.13230307922961596,
      "grad_norm": 1.626070332283681,
      "learning_rate": 1.2200811260875653e-06,
      "loss": 0.7557,
      "step": 1434
    },
    {
      "epoch": 0.1323953407911429,
      "grad_norm": 2.9045959136378876,
      "learning_rate": 1.2200235839057491e-06,
      "loss": 0.9921,
      "step": 1435
    },
    {
      "epoch": 0.1323953407911429,
      "eval_GEN Loss": 0.48395654559135437,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3801724314689636,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8667067289352417,
      "eval_runtime": 55.9602,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1435
    },
    {
      "epoch": 0.13248760235266982,
      "grad_norm": 3.534488933557012,
      "learning_rate": 1.2199659878019593e-06,
      "loss": 1.0597,
      "step": 1436
    },
    {
      "epoch": 0.13257986391419674,
      "grad_norm": 1.9046011313670044,
      "learning_rate": 1.219908337781415e-06,
      "loss": 0.8453,
      "step": 1437
    },
    {
      "epoch": 0.13267212547572368,
      "grad_norm": 2.0757607966541696,
      "learning_rate": 1.2198506338493406e-06,
      "loss": 0.8751,
      "step": 1438
    },
    {
      "epoch": 0.1327643870372506,
      "grad_norm": 2.128683664799301,
      "learning_rate": 1.2197928760109658e-06,
      "loss": 0.7596,
      "step": 1439
    },
    {
      "epoch": 0.13285664859877752,
      "grad_norm": 2.487992956455477,
      "learning_rate": 1.2197350642715241e-06,
      "loss": 0.8282,
      "step": 1440
    },
    {
      "epoch": 0.13285664859877752,
      "eval_GEN Loss": 0.4832429587841034,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.4173552095890045,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8638221025466919,
      "eval_runtime": 56.0343,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1440
    },
    {
      "epoch": 0.13294891016030447,
      "grad_norm": 2.2825361847454717,
      "learning_rate": 1.219677198636255e-06,
      "loss": 0.8686,
      "step": 1441
    },
    {
      "epoch": 0.13304117172183139,
      "grad_norm": 2.6788643263216123,
      "learning_rate": 1.2196192791104017e-06,
      "loss": 0.6179,
      "step": 1442
    },
    {
      "epoch": 0.13313343328335833,
      "grad_norm": 2.1012184128946236,
      "learning_rate": 1.2195613056992137e-06,
      "loss": 0.6893,
      "step": 1443
    },
    {
      "epoch": 0.13322569484488525,
      "grad_norm": 1.5552346137341364,
      "learning_rate": 1.2195032784079444e-06,
      "loss": 0.7452,
      "step": 1444
    },
    {
      "epoch": 0.13331795640641217,
      "grad_norm": 1.606597977368422,
      "learning_rate": 1.219445197241852e-06,
      "loss": 0.621,
      "step": 1445
    },
    {
      "epoch": 0.13331795640641217,
      "eval_GEN Loss": 0.4831272065639496,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.441954642534256,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8729567527770996,
      "eval_runtime": 57.0335,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1445
    },
    {
      "epoch": 0.13341021796793912,
      "grad_norm": 1.89406867652153,
      "learning_rate": 1.2193870622062004e-06,
      "loss": 0.7446,
      "step": 1446
    },
    {
      "epoch": 0.13350247952946603,
      "grad_norm": 3.2388273648052297,
      "learning_rate": 1.2193288733062574e-06,
      "loss": 0.921,
      "step": 1447
    },
    {
      "epoch": 0.13359474109099295,
      "grad_norm": 3.0643457878600135,
      "learning_rate": 1.2192706305472965e-06,
      "loss": 0.9296,
      "step": 1448
    },
    {
      "epoch": 0.1336870026525199,
      "grad_norm": 1.6532173075171206,
      "learning_rate": 1.2192123339345957e-06,
      "loss": 0.786,
      "step": 1449
    },
    {
      "epoch": 0.13377926421404682,
      "grad_norm": 1.717418204393591,
      "learning_rate": 1.2191539834734378e-06,
      "loss": 0.7298,
      "step": 1450
    },
    {
      "epoch": 0.13377926421404682,
      "eval_GEN Loss": 0.48509472608566284,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.4426087737083435,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8865384459495544,
      "eval_runtime": 55.9,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1450
    },
    {
      "epoch": 0.13387152577557376,
      "grad_norm": 1.7870663859688258,
      "learning_rate": 1.2190955791691108e-06,
      "loss": 0.6784,
      "step": 1451
    },
    {
      "epoch": 0.13396378733710068,
      "grad_norm": 1.7480168937710523,
      "learning_rate": 1.2190371210269075e-06,
      "loss": 0.7037,
      "step": 1452
    },
    {
      "epoch": 0.1340560488986276,
      "grad_norm": 1.8071013731005843,
      "learning_rate": 1.218978609052125e-06,
      "loss": 0.8095,
      "step": 1453
    },
    {
      "epoch": 0.13414831046015455,
      "grad_norm": 3.023351955770835,
      "learning_rate": 1.218920043250066e-06,
      "loss": 0.991,
      "step": 1454
    },
    {
      "epoch": 0.13424057202168146,
      "grad_norm": 2.034217737011871,
      "learning_rate": 1.2188614236260378e-06,
      "loss": 0.7052,
      "step": 1455
    },
    {
      "epoch": 0.13424057202168146,
      "eval_GEN Loss": 0.48578301072120667,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.43944641947746277,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8933894038200378,
      "eval_runtime": 57.0451,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 1455
    },
    {
      "epoch": 0.13433283358320838,
      "grad_norm": 1.7513297937120134,
      "learning_rate": 1.2188027501853527e-06,
      "loss": 0.9649,
      "step": 1456
    },
    {
      "epoch": 0.13442509514473533,
      "grad_norm": 2.1796927369703702,
      "learning_rate": 1.218744022933328e-06,
      "loss": 0.6808,
      "step": 1457
    },
    {
      "epoch": 0.13451735670626225,
      "grad_norm": 1.5648907930871407,
      "learning_rate": 1.2186852418752851e-06,
      "loss": 0.8168,
      "step": 1458
    },
    {
      "epoch": 0.1346096182677892,
      "grad_norm": 2.659296652822949,
      "learning_rate": 1.218626407016551e-06,
      "loss": 0.8799,
      "step": 1459
    },
    {
      "epoch": 0.1347018798293161,
      "grad_norm": 1.6903040043421167,
      "learning_rate": 1.2185675183624577e-06,
      "loss": 0.9248,
      "step": 1460
    },
    {
      "epoch": 0.1347018798293161,
      "eval_GEN Loss": 0.4877380132675171,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.42827802896499634,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.907151460647583,
      "eval_runtime": 57.0855,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 1460
    },
    {
      "epoch": 0.13479414139084303,
      "grad_norm": 2.2399104560871024,
      "learning_rate": 1.2185085759183416e-06,
      "loss": 0.9153,
      "step": 1461
    },
    {
      "epoch": 0.13488640295236998,
      "grad_norm": 1.5843009173873546,
      "learning_rate": 1.2184495796895441e-06,
      "loss": 0.585,
      "step": 1462
    },
    {
      "epoch": 0.1349786645138969,
      "grad_norm": 1.540416191368777,
      "learning_rate": 1.2183905296814116e-06,
      "loss": 0.7705,
      "step": 1463
    },
    {
      "epoch": 0.13507092607542381,
      "grad_norm": 1.5800918821585193,
      "learning_rate": 1.2183314258992953e-06,
      "loss": 0.8058,
      "step": 1464
    },
    {
      "epoch": 0.13516318763695076,
      "grad_norm": 2.6540944037546295,
      "learning_rate": 1.2182722683485511e-06,
      "loss": 0.8014,
      "step": 1465
    },
    {
      "epoch": 0.13516318763695076,
      "eval_GEN Loss": 0.4897516071796417,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.43161651492118835,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.909254789352417,
      "eval_runtime": 57.121,
      "eval_samples_per_second": 1.138,
      "eval_steps_per_second": 0.088,
      "step": 1465
    },
    {
      "epoch": 0.13525544919847768,
      "grad_norm": 2.555886509779221,
      "learning_rate": 1.2182130570345404e-06,
      "loss": 0.8484,
      "step": 1466
    },
    {
      "epoch": 0.13534771076000462,
      "grad_norm": 1.6730850578977305,
      "learning_rate": 1.2181537919626284e-06,
      "loss": 0.8389,
      "step": 1467
    },
    {
      "epoch": 0.13543997232153154,
      "grad_norm": 2.340466897164336,
      "learning_rate": 1.2180944731381864e-06,
      "loss": 0.6125,
      "step": 1468
    },
    {
      "epoch": 0.13553223388305846,
      "grad_norm": 2.3124459269800606,
      "learning_rate": 1.2180351005665897e-06,
      "loss": 0.8138,
      "step": 1469
    },
    {
      "epoch": 0.1356244954445854,
      "grad_norm": 1.8353511605355965,
      "learning_rate": 1.2179756742532187e-06,
      "loss": 0.7342,
      "step": 1470
    },
    {
      "epoch": 0.1356244954445854,
      "eval_GEN Loss": 0.49066561460494995,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.41019701957702637,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9172475934028625,
      "eval_runtime": 57.0038,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1470
    },
    {
      "epoch": 0.13571675700611233,
      "grad_norm": 2.8703132324935763,
      "learning_rate": 1.2179161942034587e-06,
      "loss": 0.7313,
      "step": 1471
    },
    {
      "epoch": 0.13580901856763924,
      "grad_norm": 1.4823664860463994,
      "learning_rate": 1.2178566604226999e-06,
      "loss": 0.7206,
      "step": 1472
    },
    {
      "epoch": 0.1359012801291662,
      "grad_norm": 2.5045616762517238,
      "learning_rate": 1.2177970729163374e-06,
      "loss": 0.6631,
      "step": 1473
    },
    {
      "epoch": 0.1359935416906931,
      "grad_norm": 1.797659550845175,
      "learning_rate": 1.217737431689771e-06,
      "loss": 0.845,
      "step": 1474
    },
    {
      "epoch": 0.13608580325222006,
      "grad_norm": 1.7472275102991048,
      "learning_rate": 1.2176777367484058e-06,
      "loss": 0.7978,
      "step": 1475
    },
    {
      "epoch": 0.13608580325222006,
      "eval_GEN Loss": 0.48971661925315857,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3996686339378357,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9207331538200378,
      "eval_runtime": 56.8658,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1475
    },
    {
      "epoch": 0.13617806481374697,
      "grad_norm": 2.3145155051067117,
      "learning_rate": 1.217617988097651e-06,
      "loss": 1.0098,
      "step": 1476
    },
    {
      "epoch": 0.1362703263752739,
      "grad_norm": 1.7676781476663201,
      "learning_rate": 1.2175581857429212e-06,
      "loss": 0.7219,
      "step": 1477
    },
    {
      "epoch": 0.13636258793680084,
      "grad_norm": 2.1027275664909144,
      "learning_rate": 1.2174983296896362e-06,
      "loss": 0.6974,
      "step": 1478
    },
    {
      "epoch": 0.13645484949832776,
      "grad_norm": 1.677411819562967,
      "learning_rate": 1.2174384199432198e-06,
      "loss": 0.6773,
      "step": 1479
    },
    {
      "epoch": 0.13654711105985468,
      "grad_norm": 1.401378609734081,
      "learning_rate": 1.2173784565091013e-06,
      "loss": 0.6816,
      "step": 1480
    },
    {
      "epoch": 0.13654711105985468,
      "eval_GEN Loss": 0.4899117350578308,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.41072383522987366,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9126802682876587,
      "eval_runtime": 56.8851,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1480
    },
    {
      "epoch": 0.13663937262138162,
      "grad_norm": 3.1739626128379514,
      "learning_rate": 1.2173184393927146e-06,
      "loss": 1.0342,
      "step": 1481
    },
    {
      "epoch": 0.13673163418290854,
      "grad_norm": 2.1981313502484006,
      "learning_rate": 1.2172583685994987e-06,
      "loss": 0.9485,
      "step": 1482
    },
    {
      "epoch": 0.1368238957444355,
      "grad_norm": 1.2383898111735496,
      "learning_rate": 1.217198244134897e-06,
      "loss": 0.7046,
      "step": 1483
    },
    {
      "epoch": 0.1369161573059624,
      "grad_norm": 2.5278187196321085,
      "learning_rate": 1.2171380660043586e-06,
      "loss": 0.6783,
      "step": 1484
    },
    {
      "epoch": 0.13700841886748932,
      "grad_norm": 1.9326352047414406,
      "learning_rate": 1.2170778342133365e-06,
      "loss": 0.8354,
      "step": 1485
    },
    {
      "epoch": 0.13700841886748932,
      "eval_GEN Loss": 0.4875543415546417,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.4324727952480316,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9129206538200378,
      "eval_runtime": 56.9608,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1485
    },
    {
      "epoch": 0.13710068042901627,
      "grad_norm": 2.016188506648634,
      "learning_rate": 1.217017548767289e-06,
      "loss": 0.8474,
      "step": 1486
    },
    {
      "epoch": 0.1371929419905432,
      "grad_norm": 1.8314060510452763,
      "learning_rate": 1.2169572096716794e-06,
      "loss": 0.9059,
      "step": 1487
    },
    {
      "epoch": 0.1372852035520701,
      "grad_norm": 1.6984189460823083,
      "learning_rate": 1.2168968169319758e-06,
      "loss": 0.6674,
      "step": 1488
    },
    {
      "epoch": 0.13737746511359705,
      "grad_norm": 2.6395713682345163,
      "learning_rate": 1.216836370553651e-06,
      "loss": 0.8628,
      "step": 1489
    },
    {
      "epoch": 0.13746972667512397,
      "grad_norm": 1.9000009573257397,
      "learning_rate": 1.2167758705421827e-06,
      "loss": 0.911,
      "step": 1490
    },
    {
      "epoch": 0.13746972667512397,
      "eval_GEN Loss": 0.4862188994884491,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.42939531803131104,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9112980961799622,
      "eval_runtime": 56.904,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1490
    },
    {
      "epoch": 0.13756198823665092,
      "grad_norm": 1.940560206516826,
      "learning_rate": 1.2167153169030536e-06,
      "loss": 1.0167,
      "step": 1491
    },
    {
      "epoch": 0.13765424979817784,
      "grad_norm": 1.7804113668872108,
      "learning_rate": 1.216654709641751e-06,
      "loss": 0.8197,
      "step": 1492
    },
    {
      "epoch": 0.13774651135970475,
      "grad_norm": 1.870396973825811,
      "learning_rate": 1.2165940487637672e-06,
      "loss": 0.8109,
      "step": 1493
    },
    {
      "epoch": 0.1378387729212317,
      "grad_norm": 1.783129166529002,
      "learning_rate": 1.2165333342745997e-06,
      "loss": 0.8711,
      "step": 1494
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 1.6537272729529997,
      "learning_rate": 1.2164725661797505e-06,
      "loss": 0.6381,
      "step": 1495
    },
    {
      "epoch": 0.13793103448275862,
      "eval_GEN Loss": 0.48459392786026,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.4264398217201233,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9077523946762085,
      "eval_runtime": 55.9849,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1495
    },
    {
      "epoch": 0.13802329604428554,
      "grad_norm": 1.7188956810123366,
      "learning_rate": 1.2164117444847262e-06,
      "loss": 0.8818,
      "step": 1496
    },
    {
      "epoch": 0.13811555760581248,
      "grad_norm": 1.968972988950428,
      "learning_rate": 1.2163508691950387e-06,
      "loss": 0.813,
      "step": 1497
    },
    {
      "epoch": 0.1382078191673394,
      "grad_norm": 2.160000707263368,
      "learning_rate": 1.2162899403162046e-06,
      "loss": 0.5748,
      "step": 1498
    },
    {
      "epoch": 0.13830008072886635,
      "grad_norm": 2.382010283993841,
      "learning_rate": 1.2162289578537457e-06,
      "loss": 0.7955,
      "step": 1499
    },
    {
      "epoch": 0.13839234229039327,
      "grad_norm": 1.495369182403971,
      "learning_rate": 1.2161679218131877e-06,
      "loss": 0.7968,
      "step": 1500
    },
    {
      "epoch": 0.13839234229039327,
      "eval_GEN Loss": 0.48590993881225586,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.41839560866355896,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9126201868057251,
      "eval_runtime": 55.8859,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1500
    },
    {
      "epoch": 0.13848460385192018,
      "grad_norm": 1.7812087696351404,
      "learning_rate": 1.216106832200062e-06,
      "loss": 0.732,
      "step": 1501
    },
    {
      "epoch": 0.13857686541344713,
      "grad_norm": 2.1991333275716305,
      "learning_rate": 1.2160456890199052e-06,
      "loss": 0.9506,
      "step": 1502
    },
    {
      "epoch": 0.13866912697497405,
      "grad_norm": 2.8191440678976973,
      "learning_rate": 1.2159844922782575e-06,
      "loss": 0.8977,
      "step": 1503
    },
    {
      "epoch": 0.13876138853650097,
      "grad_norm": 2.3006810473425205,
      "learning_rate": 1.2159232419806649e-06,
      "loss": 0.8017,
      "step": 1504
    },
    {
      "epoch": 0.13885365009802791,
      "grad_norm": 2.5897149997967532,
      "learning_rate": 1.2158619381326778e-06,
      "loss": 0.902,
      "step": 1505
    },
    {
      "epoch": 0.13885365009802791,
      "eval_GEN Loss": 0.48909762501716614,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.42517611384391785,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9254207015037537,
      "eval_runtime": 55.9005,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1505
    },
    {
      "epoch": 0.13894591165955483,
      "grad_norm": 1.6344563329401594,
      "learning_rate": 1.215800580739852e-06,
      "loss": 0.8441,
      "step": 1506
    },
    {
      "epoch": 0.13903817322108178,
      "grad_norm": 2.1475173191522052,
      "learning_rate": 1.2157391698077477e-06,
      "loss": 0.9089,
      "step": 1507
    },
    {
      "epoch": 0.1391304347826087,
      "grad_norm": 2.188858630147719,
      "learning_rate": 1.21567770534193e-06,
      "loss": 0.9531,
      "step": 1508
    },
    {
      "epoch": 0.13922269634413562,
      "grad_norm": 1.2498198125794533,
      "learning_rate": 1.2156161873479688e-06,
      "loss": 0.6935,
      "step": 1509
    },
    {
      "epoch": 0.13931495790566256,
      "grad_norm": 3.1222719593071524,
      "learning_rate": 1.2155546158314391e-06,
      "loss": 1.0682,
      "step": 1510
    },
    {
      "epoch": 0.13931495790566256,
      "eval_GEN Loss": 0.4886327385902405,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.43589362502098083,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9278846383094788,
      "eval_runtime": 56.8024,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1510
    },
    {
      "epoch": 0.13940721946718948,
      "grad_norm": 1.6876532513309759,
      "learning_rate": 1.2154929907979207e-06,
      "loss": 0.9661,
      "step": 1511
    },
    {
      "epoch": 0.1394994810287164,
      "grad_norm": 2.3098677426370844,
      "learning_rate": 1.2154313122529979e-06,
      "loss": 0.9075,
      "step": 1512
    },
    {
      "epoch": 0.13959174259024335,
      "grad_norm": 1.5925941919284827,
      "learning_rate": 1.2153695802022603e-06,
      "loss": 0.7232,
      "step": 1513
    },
    {
      "epoch": 0.13968400415177026,
      "grad_norm": 2.07332708079127,
      "learning_rate": 1.215307794651302e-06,
      "loss": 0.6794,
      "step": 1514
    },
    {
      "epoch": 0.1397762657132972,
      "grad_norm": 1.5762825942969272,
      "learning_rate": 1.2152459556057221e-06,
      "loss": 0.6641,
      "step": 1515
    },
    {
      "epoch": 0.1397762657132972,
      "eval_GEN Loss": 0.4870279133319855,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4536207914352417,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9316706657409668,
      "eval_runtime": 56.648,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 1515
    },
    {
      "epoch": 0.13986852727482413,
      "grad_norm": 1.480708156546866,
      "learning_rate": 1.2151840630711249e-06,
      "loss": 0.7197,
      "step": 1516
    },
    {
      "epoch": 0.13996078883635105,
      "grad_norm": 2.5720173893902976,
      "learning_rate": 1.2151221170531186e-06,
      "loss": 0.9796,
      "step": 1517
    },
    {
      "epoch": 0.140053050397878,
      "grad_norm": 1.5842240024501602,
      "learning_rate": 1.2150601175573173e-06,
      "loss": 0.6834,
      "step": 1518
    },
    {
      "epoch": 0.1401453119594049,
      "grad_norm": 1.9219813009605233,
      "learning_rate": 1.2149980645893397e-06,
      "loss": 0.8441,
      "step": 1519
    },
    {
      "epoch": 0.14023757352093183,
      "grad_norm": 1.9267743882787396,
      "learning_rate": 1.2149359581548084e-06,
      "loss": 0.8574,
      "step": 1520
    },
    {
      "epoch": 0.14023757352093183,
      "eval_GEN Loss": 0.4875936806201935,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.46150749921798706,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9364783763885498,
      "eval_runtime": 56.8273,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1520
    },
    {
      "epoch": 0.14032983508245878,
      "grad_norm": 2.5934660350263643,
      "learning_rate": 1.2148737982593522e-06,
      "loss": 0.8217,
      "step": 1521
    },
    {
      "epoch": 0.1404220966439857,
      "grad_norm": 1.7778069422397285,
      "learning_rate": 1.2148115849086039e-06,
      "loss": 0.8128,
      "step": 1522
    },
    {
      "epoch": 0.14051435820551264,
      "grad_norm": 1.9120964058598204,
      "learning_rate": 1.2147493181082014e-06,
      "loss": 0.6522,
      "step": 1523
    },
    {
      "epoch": 0.14060661976703956,
      "grad_norm": 1.7316037392769772,
      "learning_rate": 1.2146869978637874e-06,
      "loss": 0.8323,
      "step": 1524
    },
    {
      "epoch": 0.14069888132856648,
      "grad_norm": 2.2877225371734435,
      "learning_rate": 1.2146246241810094e-06,
      "loss": 0.7117,
      "step": 1525
    },
    {
      "epoch": 0.14069888132856648,
      "eval_GEN Loss": 0.48820412158966064,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8518518518518519,
      "eval_PRM F1 AUC": 0.6982713462545836,
      "eval_PRM F1 Neg": 0.52,
      "eval_PRM Loss": 0.45991775393486023,
      "eval_PRM NPV": 0.48148148148148145,
      "eval_PRM Precision": 0.8734177215189873,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9534855484962463,
      "eval_runtime": 56.8446,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1525
    },
    {
      "epoch": 0.14079114289009342,
      "grad_norm": 1.3918306824776139,
      "learning_rate": 1.21456219706552e-06,
      "loss": 0.715,
      "step": 1526
    },
    {
      "epoch": 0.14088340445162034,
      "grad_norm": 1.4411090798512138,
      "learning_rate": 1.2144997165229764e-06,
      "loss": 0.7589,
      "step": 1527
    },
    {
      "epoch": 0.14097566601314726,
      "grad_norm": 1.7806133121953431,
      "learning_rate": 1.2144371825590404e-06,
      "loss": 0.7439,
      "step": 1528
    },
    {
      "epoch": 0.1410679275746742,
      "grad_norm": 3.8133716362518717,
      "learning_rate": 1.2143745951793792e-06,
      "loss": 0.9046,
      "step": 1529
    },
    {
      "epoch": 0.14116018913620113,
      "grad_norm": 1.8334556072347568,
      "learning_rate": 1.2143119543896645e-06,
      "loss": 0.8661,
      "step": 1530
    },
    {
      "epoch": 0.14116018913620113,
      "eval_GEN Loss": 0.48858997225761414,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8518518518518519,
      "eval_PRM F1 AUC": 0.6982713462545836,
      "eval_PRM F1 Neg": 0.52,
      "eval_PRM Loss": 0.4668982923030853,
      "eval_PRM NPV": 0.48148148148148145,
      "eval_PRM Precision": 0.8734177215189873,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9585336446762085,
      "eval_runtime": 56.8617,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1530
    },
    {
      "epoch": 0.14125245069772807,
      "grad_norm": 2.1977085918394224,
      "learning_rate": 1.214249260195573e-06,
      "loss": 0.8644,
      "step": 1531
    },
    {
      "epoch": 0.141344712259255,
      "grad_norm": 4.325783037130388,
      "learning_rate": 1.2141865126027859e-06,
      "loss": 1.0033,
      "step": 1532
    },
    {
      "epoch": 0.1414369738207819,
      "grad_norm": 1.920480883036466,
      "learning_rate": 1.2141237116169897e-06,
      "loss": 0.9399,
      "step": 1533
    },
    {
      "epoch": 0.14152923538230885,
      "grad_norm": 2.1268793254030114,
      "learning_rate": 1.2140608572438753e-06,
      "loss": 0.9536,
      "step": 1534
    },
    {
      "epoch": 0.14162149694383577,
      "grad_norm": 1.6851987626291,
      "learning_rate": 1.213997949489139e-06,
      "loss": 0.8608,
      "step": 1535
    },
    {
      "epoch": 0.14162149694383577,
      "eval_GEN Loss": 0.4883213937282562,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.4781765937805176,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.957932710647583,
      "eval_runtime": 56.8715,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1535
    },
    {
      "epoch": 0.1417137585053627,
      "grad_norm": 1.8035419546136695,
      "learning_rate": 1.2139349883584813e-06,
      "loss": 0.6354,
      "step": 1536
    },
    {
      "epoch": 0.14180602006688964,
      "grad_norm": 2.2774576257316537,
      "learning_rate": 1.2138719738576082e-06,
      "loss": 0.8053,
      "step": 1537
    },
    {
      "epoch": 0.14189828162841656,
      "grad_norm": 1.7243372664491499,
      "learning_rate": 1.2138089059922298e-06,
      "loss": 0.819,
      "step": 1538
    },
    {
      "epoch": 0.1419905431899435,
      "grad_norm": 1.7654585989184457,
      "learning_rate": 1.2137457847680615e-06,
      "loss": 1.1385,
      "step": 1539
    },
    {
      "epoch": 0.14208280475147042,
      "grad_norm": 1.791716423273646,
      "learning_rate": 1.2136826101908235e-06,
      "loss": 0.6427,
      "step": 1540
    },
    {
      "epoch": 0.14208280475147042,
      "eval_GEN Loss": 0.48769626021385193,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8518518518518519,
      "eval_PRM F1 AUC": 0.6982713462545836,
      "eval_PRM F1 Neg": 0.52,
      "eval_PRM Loss": 0.4586150050163269,
      "eval_PRM NPV": 0.48148148148148145,
      "eval_PRM Precision": 0.8734177215189873,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9599158763885498,
      "eval_runtime": 56.7674,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1540
    },
    {
      "epoch": 0.14217506631299734,
      "grad_norm": 2.6113178827100123,
      "learning_rate": 1.2136193822662408e-06,
      "loss": 0.8815,
      "step": 1541
    },
    {
      "epoch": 0.14226732787452429,
      "grad_norm": 2.27231942686477,
      "learning_rate": 1.213556101000043e-06,
      "loss": 0.8198,
      "step": 1542
    },
    {
      "epoch": 0.1423595894360512,
      "grad_norm": 1.7440396558376248,
      "learning_rate": 1.213492766397965e-06,
      "loss": 0.681,
      "step": 1543
    },
    {
      "epoch": 0.14245185099757812,
      "grad_norm": 1.9289662225514435,
      "learning_rate": 1.2134293784657463e-06,
      "loss": 0.9392,
      "step": 1544
    },
    {
      "epoch": 0.14254411255910507,
      "grad_norm": 2.015619872344053,
      "learning_rate": 1.2133659372091313e-06,
      "loss": 0.9288,
      "step": 1545
    },
    {
      "epoch": 0.14254411255910507,
      "eval_GEN Loss": 0.4858710467815399,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8606060606060606,
      "eval_PRM F1 AUC": 0.6885804085908852,
      "eval_PRM F1 Neg": 0.5106382978723404,
      "eval_PRM Loss": 0.4427182078361511,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8658536585365854,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9689303040504456,
      "eval_runtime": 56.714,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 1545
    },
    {
      "epoch": 0.142636374120632,
      "grad_norm": 1.7083452678678301,
      "learning_rate": 1.213302442633869e-06,
      "loss": 0.7585,
      "step": 1546
    },
    {
      "epoch": 0.14272863568215893,
      "grad_norm": 1.7437547134384943,
      "learning_rate": 1.213238894745713e-06,
      "loss": 0.5721,
      "step": 1547
    },
    {
      "epoch": 0.14282089724368585,
      "grad_norm": 1.6264445760393105,
      "learning_rate": 1.2131752935504223e-06,
      "loss": 0.804,
      "step": 1548
    },
    {
      "epoch": 0.14291315880521277,
      "grad_norm": 3.0483329518966347,
      "learning_rate": 1.2131116390537612e-06,
      "loss": 0.9359,
      "step": 1549
    },
    {
      "epoch": 0.14300542036673972,
      "grad_norm": 1.4994529197445892,
      "learning_rate": 1.2130479312614973e-06,
      "loss": 0.7795,
      "step": 1550
    },
    {
      "epoch": 0.14300542036673972,
      "eval_GEN Loss": 0.4825584888458252,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8622754491017964,
      "eval_PRM F1 AUC": 0.6728653745416449,
      "eval_PRM F1 Neg": 0.4888888888888889,
      "eval_PRM Loss": 0.4122474491596222,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8571428571428571,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9410457015037537,
      "eval_runtime": 56.6267,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 1550
    },
    {
      "epoch": 0.14309768192826663,
      "grad_norm": 2.03221205648812,
      "learning_rate": 1.2129841701794044e-06,
      "loss": 0.5885,
      "step": 1551
    },
    {
      "epoch": 0.14318994348979355,
      "grad_norm": 1.796864093783121,
      "learning_rate": 1.2129203558132604e-06,
      "loss": 0.6886,
      "step": 1552
    },
    {
      "epoch": 0.1432822050513205,
      "grad_norm": 1.5117740616104873,
      "learning_rate": 1.2128564881688486e-06,
      "loss": 0.8127,
      "step": 1553
    },
    {
      "epoch": 0.14337446661284742,
      "grad_norm": 1.9128950440003294,
      "learning_rate": 1.2127925672519564e-06,
      "loss": 0.6823,
      "step": 1554
    },
    {
      "epoch": 0.14346672817437436,
      "grad_norm": 2.221095276125673,
      "learning_rate": 1.2127285930683766e-06,
      "loss": 0.9162,
      "step": 1555
    },
    {
      "epoch": 0.14346672817437436,
      "eval_GEN Loss": 0.48015135526657104,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4164435565471649,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9243990182876587,
      "eval_runtime": 56.8141,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1555
    },
    {
      "epoch": 0.14355898973590128,
      "grad_norm": 1.6199425103379719,
      "learning_rate": 1.2126645656239064e-06,
      "loss": 0.929,
      "step": 1556
    },
    {
      "epoch": 0.1436512512974282,
      "grad_norm": 1.617495758978593,
      "learning_rate": 1.2126004849243485e-06,
      "loss": 0.7732,
      "step": 1557
    },
    {
      "epoch": 0.14374351285895515,
      "grad_norm": 2.5232182449527536,
      "learning_rate": 1.2125363509755098e-06,
      "loss": 0.8469,
      "step": 1558
    },
    {
      "epoch": 0.14383577442048207,
      "grad_norm": 1.8377351016379688,
      "learning_rate": 1.212472163783202e-06,
      "loss": 0.6864,
      "step": 1559
    },
    {
      "epoch": 0.14392803598200898,
      "grad_norm": 1.561310611356702,
      "learning_rate": 1.212407923353242e-06,
      "loss": 0.5072,
      "step": 1560
    },
    {
      "epoch": 0.14392803598200898,
      "eval_GEN Loss": 0.47932034730911255,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.41043820977211,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.926682710647583,
      "eval_runtime": 56.9817,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1560
    },
    {
      "epoch": 0.14402029754353593,
      "grad_norm": 2.1865153012788716,
      "learning_rate": 1.2123436296914514e-06,
      "loss": 0.8404,
      "step": 1561
    },
    {
      "epoch": 0.14411255910506285,
      "grad_norm": 1.4396843867341758,
      "learning_rate": 1.2122792828036566e-06,
      "loss": 0.7448,
      "step": 1562
    },
    {
      "epoch": 0.1442048206665898,
      "grad_norm": 2.4406187017441305,
      "learning_rate": 1.212214882695689e-06,
      "loss": 0.7268,
      "step": 1563
    },
    {
      "epoch": 0.1442970822281167,
      "grad_norm": 2.331673626795673,
      "learning_rate": 1.2121504293733843e-06,
      "loss": 0.7616,
      "step": 1564
    },
    {
      "epoch": 0.14438934378964363,
      "grad_norm": 1.4222004672390411,
      "learning_rate": 1.2120859228425835e-06,
      "loss": 0.6549,
      "step": 1565
    },
    {
      "epoch": 0.14438934378964363,
      "eval_GEN Loss": 0.4810382127761841,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8928571428571429,
      "eval_PRM F1 AUC": 0.7344159245678366,
      "eval_PRM F1 Neg": 0.5909090909090909,
      "eval_PRM Loss": 0.39960551261901855,
      "eval_PRM NPV": 0.6190476190476191,
      "eval_PRM Precision": 0.8823529411764706,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9530048370361328,
      "eval_runtime": 56.7626,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1565
    },
    {
      "epoch": 0.14448160535117058,
      "grad_norm": 1.5144381910161673,
      "learning_rate": 1.2120213631091322e-06,
      "loss": 0.6392,
      "step": 1566
    },
    {
      "epoch": 0.1445738669126975,
      "grad_norm": 1.5191413241489855,
      "learning_rate": 1.211956750178881e-06,
      "loss": 0.724,
      "step": 1567
    },
    {
      "epoch": 0.14466612847422441,
      "grad_norm": 1.5331338319160501,
      "learning_rate": 1.2118920840576853e-06,
      "loss": 0.7858,
      "step": 1568
    },
    {
      "epoch": 0.14475839003575136,
      "grad_norm": 1.6681354517322655,
      "learning_rate": 1.2118273647514054e-06,
      "loss": 0.7343,
      "step": 1569
    },
    {
      "epoch": 0.14485065159727828,
      "grad_norm": 2.6359620666061736,
      "learning_rate": 1.2117625922659057e-06,
      "loss": 0.7797,
      "step": 1570
    },
    {
      "epoch": 0.14485065159727828,
      "eval_GEN Loss": 0.48034995794296265,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9058823529411765,
      "eval_PRM F1 AUC": 0.746464117338921,
      "eval_PRM F1 Neg": 0.6190476190476191,
      "eval_PRM Loss": 0.39335888624191284,
      "eval_PRM NPV": 0.6842105263157895,
      "eval_PRM Precision": 0.8850574712643678,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9628605842590332,
      "eval_runtime": 56.8155,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1570
    },
    {
      "epoch": 0.14494291315880523,
      "grad_norm": 1.815111611079528,
      "learning_rate": 1.2116977666070564e-06,
      "loss": 0.8399,
      "step": 1571
    },
    {
      "epoch": 0.14503517472033214,
      "grad_norm": 2.564024454609857,
      "learning_rate": 1.211632887780732e-06,
      "loss": 0.816,
      "step": 1572
    },
    {
      "epoch": 0.14512743628185906,
      "grad_norm": 2.1408366599577984,
      "learning_rate": 1.211567955792812e-06,
      "loss": 0.6796,
      "step": 1573
    },
    {
      "epoch": 0.145219697843386,
      "grad_norm": 1.6763765475188022,
      "learning_rate": 1.2115029706491804e-06,
      "loss": 0.6421,
      "step": 1574
    },
    {
      "epoch": 0.14531195940491293,
      "grad_norm": 2.6073873782420476,
      "learning_rate": 1.2114379323557266e-06,
      "loss": 0.7641,
      "step": 1575
    },
    {
      "epoch": 0.14531195940491293,
      "eval_GEN Loss": 0.477909654378891,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8795180722891566,
      "eval_PRM F1 AUC": 0.7223677317967523,
      "eval_PRM F1 Neg": 0.5652173913043478,
      "eval_PRM Loss": 0.3995104730129242,
      "eval_PRM NPV": 0.5652173913043478,
      "eval_PRM Precision": 0.8795180722891566,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9445913434028625,
      "eval_runtime": 56.7706,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1575
    },
    {
      "epoch": 0.14540422096643985,
      "grad_norm": 1.7960350045434368,
      "learning_rate": 1.2113728409183441e-06,
      "loss": 1.0443,
      "step": 1576
    },
    {
      "epoch": 0.1454964825279668,
      "grad_norm": 2.2135505447891433,
      "learning_rate": 1.211307696342932e-06,
      "loss": 0.6969,
      "step": 1577
    },
    {
      "epoch": 0.1455887440894937,
      "grad_norm": 1.8917194674305324,
      "learning_rate": 1.2112424986353934e-06,
      "loss": 0.8017,
      "step": 1578
    },
    {
      "epoch": 0.14568100565102066,
      "grad_norm": 1.7442166552375302,
      "learning_rate": 1.211177247801637e-06,
      "loss": 0.9089,
      "step": 1579
    },
    {
      "epoch": 0.14577326721254757,
      "grad_norm": 1.9222757837622877,
      "learning_rate": 1.2111119438475758e-06,
      "loss": 0.7088,
      "step": 1580
    },
    {
      "epoch": 0.14577326721254757,
      "eval_GEN Loss": 0.4757411479949951,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8862275449101796,
      "eval_PRM F1 AUC": 0.7283918281822945,
      "eval_PRM F1 Neg": 0.5777777777777777,
      "eval_PRM Loss": 0.39801672101020813,
      "eval_PRM NPV": 0.5909090909090909,
      "eval_PRM Precision": 0.8809523809523809,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9429086446762085,
      "eval_runtime": 56.9673,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1580
    },
    {
      "epoch": 0.1458655287740745,
      "grad_norm": 1.8373254696524093,
      "learning_rate": 1.2110465867791275e-06,
      "loss": 0.9386,
      "step": 1581
    },
    {
      "epoch": 0.14595779033560144,
      "grad_norm": 2.084647192251181,
      "learning_rate": 1.2109811766022153e-06,
      "loss": 0.8144,
      "step": 1582
    },
    {
      "epoch": 0.14605005189712836,
      "grad_norm": 2.3870932416535933,
      "learning_rate": 1.2109157133227662e-06,
      "loss": 0.9397,
      "step": 1583
    },
    {
      "epoch": 0.14614231345865528,
      "grad_norm": 1.777806736041665,
      "learning_rate": 1.2108501969467132e-06,
      "loss": 0.7087,
      "step": 1584
    },
    {
      "epoch": 0.14623457502018222,
      "grad_norm": 2.276862066058128,
      "learning_rate": 1.2107846274799932e-06,
      "loss": 0.71,
      "step": 1585
    },
    {
      "epoch": 0.14623457502018222,
      "eval_GEN Loss": 0.47742316126823425,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.6969617600838135,
      "eval_PRM F1 Neg": 0.5365853658536586,
      "eval_PRM Loss": 0.39713582396507263,
      "eval_PRM NPV": 0.6111111111111112,
      "eval_PRM Precision": 0.8636363636363636,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.95703125,
      "eval_runtime": 56.7561,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1585
    },
    {
      "epoch": 0.14632683658170914,
      "grad_norm": 2.3895934805555155,
      "learning_rate": 1.2107190049285478e-06,
      "loss": 0.7321,
      "step": 1586
    },
    {
      "epoch": 0.1464190981432361,
      "grad_norm": 1.4390572125615302,
      "learning_rate": 1.2106533292983247e-06,
      "loss": 0.7152,
      "step": 1587
    },
    {
      "epoch": 0.146511359704763,
      "grad_norm": 2.096151784511048,
      "learning_rate": 1.2105876005952749e-06,
      "loss": 0.898,
      "step": 1588
    },
    {
      "epoch": 0.14660362126628992,
      "grad_norm": 1.6431998302958275,
      "learning_rate": 1.210521818825355e-06,
      "loss": 0.7141,
      "step": 1589
    },
    {
      "epoch": 0.14669588282781687,
      "grad_norm": 1.7248784796130232,
      "learning_rate": 1.2104559839945261e-06,
      "loss": 1.0005,
      "step": 1590
    },
    {
      "epoch": 0.14669588282781687,
      "eval_GEN Loss": 0.47937461733818054,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.874251497005988,
      "eval_PRM F1 AUC": 0.7006286013619696,
      "eval_PRM F1 Neg": 0.5333333333333333,
      "eval_PRM Loss": 0.40157634019851685,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.8690476190476191,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9463942050933838,
      "eval_runtime": 56.8058,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1590
    },
    {
      "epoch": 0.1467881443893438,
      "grad_norm": 1.4031303818781538,
      "learning_rate": 1.2103900961087547e-06,
      "loss": 0.6271,
      "step": 1591
    },
    {
      "epoch": 0.1468804059508707,
      "grad_norm": 1.744541248195711,
      "learning_rate": 1.2103241551740114e-06,
      "loss": 0.6399,
      "step": 1592
    },
    {
      "epoch": 0.14697266751239765,
      "grad_norm": 1.4339104648237544,
      "learning_rate": 1.2102581611962715e-06,
      "loss": 0.7929,
      "step": 1593
    },
    {
      "epoch": 0.14706492907392457,
      "grad_norm": 1.4630138411701694,
      "learning_rate": 1.2101921141815163e-06,
      "loss": 0.7397,
      "step": 1594
    },
    {
      "epoch": 0.14715719063545152,
      "grad_norm": 1.4826477252971642,
      "learning_rate": 1.2101260141357302e-06,
      "loss": 0.8408,
      "step": 1595
    },
    {
      "epoch": 0.14715719063545152,
      "eval_GEN Loss": 0.48221975564956665,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8674698795180723,
      "eval_PRM F1 AUC": 0.6946045049764275,
      "eval_PRM F1 Neg": 0.5217391304347826,
      "eval_PRM Loss": 0.4121546149253845,
      "eval_PRM NPV": 0.5217391304347826,
      "eval_PRM Precision": 0.8674698795180723,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.948437511920929,
      "eval_runtime": 56.0388,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1595
    },
    {
      "epoch": 0.14724945219697844,
      "grad_norm": 1.8944713991231408,
      "learning_rate": 1.210059861064904e-06,
      "loss": 0.8403,
      "step": 1596
    },
    {
      "epoch": 0.14734171375850535,
      "grad_norm": 1.4630907452130664,
      "learning_rate": 1.2099936549750318e-06,
      "loss": 0.7839,
      "step": 1597
    },
    {
      "epoch": 0.1474339753200323,
      "grad_norm": 1.3507923187890327,
      "learning_rate": 1.2099273958721142e-06,
      "loss": 0.6461,
      "step": 1598
    },
    {
      "epoch": 0.14752623688155922,
      "grad_norm": 1.7681718189083817,
      "learning_rate": 1.2098610837621552e-06,
      "loss": 0.9491,
      "step": 1599
    },
    {
      "epoch": 0.14761849844308614,
      "grad_norm": 1.7883526947799715,
      "learning_rate": 1.2097947186511641e-06,
      "loss": 0.9205,
      "step": 1600
    },
    {
      "epoch": 0.14761849844308614,
      "eval_GEN Loss": 0.48528167605400085,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.42186102271080017,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9498798251152039,
      "eval_runtime": 55.7531,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 1600
    },
    {
      "epoch": 0.14771076000461308,
      "grad_norm": 1.7554929817074767,
      "learning_rate": 1.2097283005451555e-06,
      "loss": 0.913,
      "step": 1601
    },
    {
      "epoch": 0.14780302156614,
      "grad_norm": 1.539366146174878,
      "learning_rate": 1.2096618294501476e-06,
      "loss": 0.7369,
      "step": 1602
    },
    {
      "epoch": 0.14789528312766695,
      "grad_norm": 1.5570817115421292,
      "learning_rate": 1.2095953053721645e-06,
      "loss": 0.673,
      "step": 1603
    },
    {
      "epoch": 0.14798754468919387,
      "grad_norm": 1.4719215916664423,
      "learning_rate": 1.2095287283172346e-06,
      "loss": 0.6456,
      "step": 1604
    },
    {
      "epoch": 0.14807980625072079,
      "grad_norm": 1.6580542575424595,
      "learning_rate": 1.2094620982913914e-06,
      "loss": 0.7629,
      "step": 1605
    },
    {
      "epoch": 0.14807980625072079,
      "eval_GEN Loss": 0.4843921363353729,
      "eval_GEN top-5 accuracy": 0.9785181937746602,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.429315984249115,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9508413672447205,
      "eval_runtime": 55.8532,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1605
    },
    {
      "epoch": 0.14817206781224773,
      "grad_norm": 1.4995954682971415,
      "learning_rate": 1.209395415300673e-06,
      "loss": 0.729,
      "step": 1606
    },
    {
      "epoch": 0.14826432937377465,
      "grad_norm": 1.8353140264749694,
      "learning_rate": 1.209328679351122e-06,
      "loss": 0.6965,
      "step": 1607
    },
    {
      "epoch": 0.14835659093530157,
      "grad_norm": 2.320723749388007,
      "learning_rate": 1.2092618904487863e-06,
      "loss": 1.1059,
      "step": 1608
    },
    {
      "epoch": 0.14844885249682852,
      "grad_norm": 1.591344939349978,
      "learning_rate": 1.2091950485997184e-06,
      "loss": 0.7887,
      "step": 1609
    },
    {
      "epoch": 0.14854111405835543,
      "grad_norm": 2.3945639293802827,
      "learning_rate": 1.2091281538099758e-06,
      "loss": 0.8983,
      "step": 1610
    },
    {
      "epoch": 0.14854111405835543,
      "eval_GEN Loss": 0.48569273948669434,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.4210745394229889,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9552884697914124,
      "eval_runtime": 56.7096,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 1610
    },
    {
      "epoch": 0.14863337561988238,
      "grad_norm": 2.5894034595953426,
      "learning_rate": 1.2090612060856205e-06,
      "loss": 0.918,
      "step": 1611
    },
    {
      "epoch": 0.1487256371814093,
      "grad_norm": 1.8055335993350563,
      "learning_rate": 1.2089942054327193e-06,
      "loss": 0.9302,
      "step": 1612
    },
    {
      "epoch": 0.14881789874293622,
      "grad_norm": 1.431276838781587,
      "learning_rate": 1.208927151857344e-06,
      "loss": 0.6177,
      "step": 1613
    },
    {
      "epoch": 0.14891016030446316,
      "grad_norm": 1.8548376484175946,
      "learning_rate": 1.2088600453655712e-06,
      "loss": 0.7329,
      "step": 1614
    },
    {
      "epoch": 0.14900242186599008,
      "grad_norm": 1.2618433104244229,
      "learning_rate": 1.2087928859634819e-06,
      "loss": 0.8308,
      "step": 1615
    },
    {
      "epoch": 0.14900242186599008,
      "eval_GEN Loss": 0.48376229405403137,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8658536585365854,
      "eval_PRM F1 AUC": 0.710319539025668,
      "eval_PRM F1 Neg": 0.5416666666666666,
      "eval_PRM Loss": 0.412661612033844,
      "eval_PRM NPV": 0.52,
      "eval_PRM Precision": 0.8765432098765432,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9467548131942749,
      "eval_runtime": 56.9104,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1615
    },
    {
      "epoch": 0.149094683427517,
      "grad_norm": 1.4320796997565808,
      "learning_rate": 1.2087256736571625e-06,
      "loss": 0.6618,
      "step": 1616
    },
    {
      "epoch": 0.14918694498904395,
      "grad_norm": 2.8879524205470233,
      "learning_rate": 1.2086584084527036e-06,
      "loss": 0.9691,
      "step": 1617
    },
    {
      "epoch": 0.14927920655057086,
      "grad_norm": 2.0813575595306837,
      "learning_rate": 1.208591090356201e-06,
      "loss": 0.7889,
      "step": 1618
    },
    {
      "epoch": 0.1493714681120978,
      "grad_norm": 1.7154256082863044,
      "learning_rate": 1.2085237193737555e-06,
      "loss": 0.821,
      "step": 1619
    },
    {
      "epoch": 0.14946372967362473,
      "grad_norm": 1.8068920307391616,
      "learning_rate": 1.208456295511472e-06,
      "loss": 0.9439,
      "step": 1620
    },
    {
      "epoch": 0.14946372967362473,
      "eval_GEN Loss": 0.48404887318611145,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4239797294139862,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9418269395828247,
      "eval_runtime": 56.9264,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1620
    },
    {
      "epoch": 0.14955599123515165,
      "grad_norm": 2.1943708935820743,
      "learning_rate": 1.2083888187754606e-06,
      "loss": 0.6457,
      "step": 1621
    },
    {
      "epoch": 0.1496482527966786,
      "grad_norm": 2.466683965417046,
      "learning_rate": 1.208321289171836e-06,
      "loss": 0.9365,
      "step": 1622
    },
    {
      "epoch": 0.1497405143582055,
      "grad_norm": 1.9127624121626094,
      "learning_rate": 1.2082537067067183e-06,
      "loss": 0.8982,
      "step": 1623
    },
    {
      "epoch": 0.14983277591973243,
      "grad_norm": 2.025036769928188,
      "learning_rate": 1.2081860713862315e-06,
      "loss": 0.9021,
      "step": 1624
    },
    {
      "epoch": 0.14992503748125938,
      "grad_norm": 1.7945257966905834,
      "learning_rate": 1.2081183832165048e-06,
      "loss": 0.705,
      "step": 1625
    },
    {
      "epoch": 0.14992503748125938,
      "eval_GEN Loss": 0.4825240969657898,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.4197624921798706,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9364182949066162,
      "eval_runtime": 56.7028,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 1625
    },
    {
      "epoch": 0.1500172990427863,
      "grad_norm": 1.631956597393466,
      "learning_rate": 1.2080506422036725e-06,
      "loss": 0.8216,
      "step": 1626
    },
    {
      "epoch": 0.15010956060431324,
      "grad_norm": 1.4820572458087988,
      "learning_rate": 1.2079828483538734e-06,
      "loss": 0.7541,
      "step": 1627
    },
    {
      "epoch": 0.15020182216584016,
      "grad_norm": 1.7679517466246568,
      "learning_rate": 1.2079150016732508e-06,
      "loss": 0.7515,
      "step": 1628
    },
    {
      "epoch": 0.15029408372736708,
      "grad_norm": 1.7491647994721775,
      "learning_rate": 1.2078471021679532e-06,
      "loss": 0.9263,
      "step": 1629
    },
    {
      "epoch": 0.15038634528889402,
      "grad_norm": 1.6872250249087157,
      "learning_rate": 1.2077791498441337e-06,
      "loss": 0.7216,
      "step": 1630
    },
    {
      "epoch": 0.15038634528889402,
      "eval_GEN Loss": 0.4820396602153778,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.41102948784828186,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9518629908561707,
      "eval_runtime": 56.7591,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1630
    },
    {
      "epoch": 0.15047860685042094,
      "grad_norm": 1.513845787668661,
      "learning_rate": 1.2077111447079506e-06,
      "loss": 0.7099,
      "step": 1631
    },
    {
      "epoch": 0.15057086841194786,
      "grad_norm": 1.4587680435549852,
      "learning_rate": 1.207643086765566e-06,
      "loss": 0.8275,
      "step": 1632
    },
    {
      "epoch": 0.1506631299734748,
      "grad_norm": 1.7551382662965955,
      "learning_rate": 1.2075749760231483e-06,
      "loss": 0.7261,
      "step": 1633
    },
    {
      "epoch": 0.15075539153500173,
      "grad_norm": 2.864420119803085,
      "learning_rate": 1.207506812486869e-06,
      "loss": 0.9541,
      "step": 1634
    },
    {
      "epoch": 0.15084765309652867,
      "grad_norm": 1.5404400297296554,
      "learning_rate": 1.2074385961629053e-06,
      "loss": 0.7804,
      "step": 1635
    },
    {
      "epoch": 0.15084765309652867,
      "eval_GEN Loss": 0.48046940565109253,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.6969617600838135,
      "eval_PRM F1 Neg": 0.5365853658536586,
      "eval_PRM Loss": 0.40919992327690125,
      "eval_PRM NPV": 0.6111111111111112,
      "eval_PRM Precision": 0.8636363636363636,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9668869972229004,
      "eval_runtime": 56.8026,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1635
    },
    {
      "epoch": 0.1509399146580556,
      "grad_norm": 1.2536688091569381,
      "learning_rate": 1.2073703270574398e-06,
      "loss": 0.7099,
      "step": 1636
    },
    {
      "epoch": 0.1510321762195825,
      "grad_norm": 1.421850236026778,
      "learning_rate": 1.2073020051766579e-06,
      "loss": 0.7492,
      "step": 1637
    },
    {
      "epoch": 0.15112443778110946,
      "grad_norm": 3.2116489289162726,
      "learning_rate": 1.207233630526752e-06,
      "loss": 0.9474,
      "step": 1638
    },
    {
      "epoch": 0.15121669934263637,
      "grad_norm": 1.8012319667097618,
      "learning_rate": 1.2071652031139182e-06,
      "loss": 0.6945,
      "step": 1639
    },
    {
      "epoch": 0.1513089609041633,
      "grad_norm": 1.6378723565772295,
      "learning_rate": 1.2070967229443573e-06,
      "loss": 0.8106,
      "step": 1640
    },
    {
      "epoch": 0.1513089609041633,
      "eval_GEN Loss": 0.479536235332489,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.392453134059906,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9292067289352417,
      "eval_runtime": 56.8961,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1640
    },
    {
      "epoch": 0.15140122246569024,
      "grad_norm": 1.56630901885544,
      "learning_rate": 1.2070281900242751e-06,
      "loss": 0.7959,
      "step": 1641
    },
    {
      "epoch": 0.15149348402721716,
      "grad_norm": 1.430543264390956,
      "learning_rate": 1.2069596043598821e-06,
      "loss": 0.6623,
      "step": 1642
    },
    {
      "epoch": 0.1515857455887441,
      "grad_norm": 1.9462579376769658,
      "learning_rate": 1.206890965957394e-06,
      "loss": 0.8359,
      "step": 1643
    },
    {
      "epoch": 0.15167800715027102,
      "grad_norm": 1.9999967322955061,
      "learning_rate": 1.2068222748230305e-06,
      "loss": 0.7238,
      "step": 1644
    },
    {
      "epoch": 0.15177026871179794,
      "grad_norm": 1.3361769691062042,
      "learning_rate": 1.2067535309630166e-06,
      "loss": 0.6875,
      "step": 1645
    },
    {
      "epoch": 0.15177026871179794,
      "eval_GEN Loss": 0.47986242175102234,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.36502107977867126,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9144230484962463,
      "eval_runtime": 56.7669,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1645
    },
    {
      "epoch": 0.15186253027332489,
      "grad_norm": 2.5340417095182044,
      "learning_rate": 1.206684734383582e-06,
      "loss": 0.9313,
      "step": 1646
    },
    {
      "epoch": 0.1519547918348518,
      "grad_norm": 2.947386800871364,
      "learning_rate": 1.2066158850909613e-06,
      "loss": 1.0766,
      "step": 1647
    },
    {
      "epoch": 0.15204705339637872,
      "grad_norm": 1.5150065894360136,
      "learning_rate": 1.2065469830913935e-06,
      "loss": 0.867,
      "step": 1648
    },
    {
      "epoch": 0.15213931495790567,
      "grad_norm": 1.5827785765301632,
      "learning_rate": 1.206478028391123e-06,
      "loss": 0.5487,
      "step": 1649
    },
    {
      "epoch": 0.1522315765194326,
      "grad_norm": 1.9882035168587877,
      "learning_rate": 1.206409020996398e-06,
      "loss": 0.784,
      "step": 1650
    },
    {
      "epoch": 0.1522315765194326,
      "eval_GEN Loss": 0.4792032539844513,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9058823529411765,
      "eval_PRM F1 AUC": 0.746464117338921,
      "eval_PRM F1 Neg": 0.6190476190476191,
      "eval_PRM Loss": 0.3730213940143585,
      "eval_PRM NPV": 0.6842105263157895,
      "eval_PRM Precision": 0.8850574712643678,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9349759817123413,
      "eval_runtime": 57.0221,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1650
    },
    {
      "epoch": 0.15232383808095953,
      "grad_norm": 2.136656197409622,
      "learning_rate": 1.2063399609134724e-06,
      "loss": 0.9674,
      "step": 1651
    },
    {
      "epoch": 0.15241609964248645,
      "grad_norm": 1.5599202081446026,
      "learning_rate": 1.2062708481486046e-06,
      "loss": 0.7025,
      "step": 1652
    },
    {
      "epoch": 0.15250836120401337,
      "grad_norm": 2.03873936217497,
      "learning_rate": 1.2062016827080574e-06,
      "loss": 1.0646,
      "step": 1653
    },
    {
      "epoch": 0.15260062276554032,
      "grad_norm": 2.245857856496522,
      "learning_rate": 1.2061324645980993e-06,
      "loss": 0.7655,
      "step": 1654
    },
    {
      "epoch": 0.15269288432706724,
      "grad_norm": 2.3068424191925017,
      "learning_rate": 1.2060631938250022e-06,
      "loss": 0.8513,
      "step": 1655
    },
    {
      "epoch": 0.15269288432706724,
      "eval_GEN Loss": 0.480029433965683,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.9005847953216374,
      "eval_PRM F1 AUC": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.5853658536585366,
      "eval_PRM Loss": 0.38942280411720276,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9539663195610046,
      "eval_runtime": 56.8457,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1655
    },
    {
      "epoch": 0.15278514588859415,
      "grad_norm": 1.5602624685968338,
      "learning_rate": 1.205993870395044e-06,
      "loss": 0.6726,
      "step": 1656
    },
    {
      "epoch": 0.1528774074501211,
      "grad_norm": 1.963139289755594,
      "learning_rate": 1.2059244943145066e-06,
      "loss": 0.8696,
      "step": 1657
    },
    {
      "epoch": 0.15296966901164802,
      "grad_norm": 1.6254674761629366,
      "learning_rate": 1.2058550655896774e-06,
      "loss": 0.7386,
      "step": 1658
    },
    {
      "epoch": 0.15306193057317496,
      "grad_norm": 1.87372899223456,
      "learning_rate": 1.2057855842268477e-06,
      "loss": 0.8901,
      "step": 1659
    },
    {
      "epoch": 0.15315419213470188,
      "grad_norm": 1.6275426038225607,
      "learning_rate": 1.2057160502323142e-06,
      "loss": 0.8991,
      "step": 1660
    },
    {
      "epoch": 0.15315419213470188,
      "eval_GEN Loss": 0.48058921098709106,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.9005847953216374,
      "eval_PRM F1 AUC": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.5853658536585366,
      "eval_PRM Loss": 0.39074864983558655,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9556490182876587,
      "eval_runtime": 56.6544,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 1660
    },
    {
      "epoch": 0.1532464536962288,
      "grad_norm": 2.462114718297041,
      "learning_rate": 1.2056464636123782e-06,
      "loss": 0.9514,
      "step": 1661
    },
    {
      "epoch": 0.15333871525775575,
      "grad_norm": 2.388658099297588,
      "learning_rate": 1.2055768243733458e-06,
      "loss": 0.933,
      "step": 1662
    },
    {
      "epoch": 0.15343097681928267,
      "grad_norm": 2.6046711615357627,
      "learning_rate": 1.2055071325215275e-06,
      "loss": 0.8297,
      "step": 1663
    },
    {
      "epoch": 0.15352323838080958,
      "grad_norm": 1.5160597348850142,
      "learning_rate": 1.2054373880632395e-06,
      "loss": 0.8085,
      "step": 1664
    },
    {
      "epoch": 0.15361549994233653,
      "grad_norm": 1.511132567278672,
      "learning_rate": 1.2053675910048014e-06,
      "loss": 0.5161,
      "step": 1665
    },
    {
      "epoch": 0.15361549994233653,
      "eval_GEN Loss": 0.4798246920108795,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8862275449101796,
      "eval_PRM F1 AUC": 0.7283918281822945,
      "eval_PRM F1 Neg": 0.5777777777777777,
      "eval_PRM Loss": 0.4006507694721222,
      "eval_PRM NPV": 0.5909090909090909,
      "eval_PRM Precision": 0.8809523809523809,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9478966593742371,
      "eval_runtime": 56.9298,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1665
    },
    {
      "epoch": 0.15370776150386345,
      "grad_norm": 1.7971993479517718,
      "learning_rate": 1.2052977413525388e-06,
      "loss": 0.7776,
      "step": 1666
    },
    {
      "epoch": 0.1538000230653904,
      "grad_norm": 1.3934391838398303,
      "learning_rate": 1.2052278391127816e-06,
      "loss": 0.6859,
      "step": 1667
    },
    {
      "epoch": 0.15389228462691731,
      "grad_norm": 2.6120998571193597,
      "learning_rate": 1.2051578842918643e-06,
      "loss": 0.617,
      "step": 1668
    },
    {
      "epoch": 0.15398454618844423,
      "grad_norm": 2.384936624570837,
      "learning_rate": 1.205087876896126e-06,
      "loss": 0.7993,
      "step": 1669
    },
    {
      "epoch": 0.15407680774997118,
      "grad_norm": 1.794813745419985,
      "learning_rate": 1.2050178169319117e-06,
      "loss": 0.7012,
      "step": 1670
    },
    {
      "epoch": 0.15407680774997118,
      "eval_GEN Loss": 0.47999176383018494,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.41702377796173096,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9520432949066162,
      "eval_runtime": 56.8304,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1670
    },
    {
      "epoch": 0.1541690693114981,
      "grad_norm": 1.8947530793166678,
      "learning_rate": 1.2049477044055697e-06,
      "loss": 0.728,
      "step": 1671
    },
    {
      "epoch": 0.15426133087302502,
      "grad_norm": 1.6667324756944912,
      "learning_rate": 1.2048775393234538e-06,
      "loss": 0.6683,
      "step": 1672
    },
    {
      "epoch": 0.15435359243455196,
      "grad_norm": 1.8403445494935375,
      "learning_rate": 1.2048073216919227e-06,
      "loss": 0.6996,
      "step": 1673
    },
    {
      "epoch": 0.15444585399607888,
      "grad_norm": 1.996166589754953,
      "learning_rate": 1.2047370515173393e-06,
      "loss": 0.7859,
      "step": 1674
    },
    {
      "epoch": 0.15453811555760583,
      "grad_norm": 2.0348565896472466,
      "learning_rate": 1.2046667288060718e-06,
      "loss": 0.796,
      "step": 1675
    },
    {
      "epoch": 0.15453811555760583,
      "eval_GEN Loss": 0.48171553015708923,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8588957055214724,
      "eval_PRM F1 AUC": 0.7042954426401258,
      "eval_PRM F1 Neg": 0.5306122448979592,
      "eval_PRM Loss": 0.4134211540222168,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9649639129638672,
      "eval_runtime": 56.8471,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1675
    },
    {
      "epoch": 0.15463037711913274,
      "grad_norm": 1.9811196370473396,
      "learning_rate": 1.2045963535644931e-06,
      "loss": 0.7303,
      "step": 1676
    },
    {
      "epoch": 0.15472263868065966,
      "grad_norm": 2.0680009877363728,
      "learning_rate": 1.2045259257989803e-06,
      "loss": 0.9992,
      "step": 1677
    },
    {
      "epoch": 0.1548149002421866,
      "grad_norm": 2.3231930006003707,
      "learning_rate": 1.204455445515916e-06,
      "loss": 0.8537,
      "step": 1678
    },
    {
      "epoch": 0.15490716180371353,
      "grad_norm": 1.50076283457596,
      "learning_rate": 1.2043849127216873e-06,
      "loss": 0.7387,
      "step": 1679
    },
    {
      "epoch": 0.15499942336524045,
      "grad_norm": 3.237347355630976,
      "learning_rate": 1.2043143274226856e-06,
      "loss": 0.9481,
      "step": 1680
    },
    {
      "epoch": 0.15499942336524045,
      "eval_GEN Loss": 0.4792095720767975,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.9005847953216374,
      "eval_PRM F1 AUC": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.5853658536585366,
      "eval_PRM Loss": 0.400808721780777,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9655048251152039,
      "eval_runtime": 56.653,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 1680
    },
    {
      "epoch": 0.1550916849267674,
      "grad_norm": 1.7110570452735685,
      "learning_rate": 1.204243689625308e-06,
      "loss": 0.7252,
      "step": 1681
    },
    {
      "epoch": 0.1551839464882943,
      "grad_norm": 1.5904298016958394,
      "learning_rate": 1.2041729993359551e-06,
      "loss": 0.8541,
      "step": 1682
    },
    {
      "epoch": 0.15527620804982123,
      "grad_norm": 1.6387993124820592,
      "learning_rate": 1.2041022565610337e-06,
      "loss": 0.7684,
      "step": 1683
    },
    {
      "epoch": 0.15536846961134818,
      "grad_norm": 1.8078395242211844,
      "learning_rate": 1.2040314613069543e-06,
      "loss": 0.6376,
      "step": 1684
    },
    {
      "epoch": 0.1554607311728751,
      "grad_norm": 1.9072669000614397,
      "learning_rate": 1.2039606135801322e-06,
      "loss": 0.8818,
      "step": 1685
    },
    {
      "epoch": 0.1554607311728751,
      "eval_GEN Loss": 0.476083368062973,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.9005847953216374,
      "eval_PRM F1 AUC": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.5853658536585366,
      "eval_PRM Loss": 0.39639008045196533,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9598557949066162,
      "eval_runtime": 56.9089,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1685
    },
    {
      "epoch": 0.15555299273440204,
      "grad_norm": 3.375773958150725,
      "learning_rate": 1.203889713386988e-06,
      "loss": 0.7987,
      "step": 1686
    },
    {
      "epoch": 0.15564525429592896,
      "grad_norm": 2.099490335827624,
      "learning_rate": 1.2038187607339472e-06,
      "loss": 0.888,
      "step": 1687
    },
    {
      "epoch": 0.15573751585745588,
      "grad_norm": 1.5885592506757007,
      "learning_rate": 1.203747755627439e-06,
      "loss": 0.8049,
      "step": 1688
    },
    {
      "epoch": 0.15582977741898282,
      "grad_norm": 3.767814518316457,
      "learning_rate": 1.203676698073898e-06,
      "loss": 0.7145,
      "step": 1689
    },
    {
      "epoch": 0.15592203898050974,
      "grad_norm": 1.3371675311778843,
      "learning_rate": 1.203605588079764e-06,
      "loss": 0.7191,
      "step": 1690
    },
    {
      "epoch": 0.15592203898050974,
      "eval_GEN Loss": 0.4749113619327545,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.9005847953216374,
      "eval_PRM F1 AUC": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.5853658536585366,
      "eval_PRM Loss": 0.3965228796005249,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9549278616905212,
      "eval_runtime": 56.8121,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1690
    },
    {
      "epoch": 0.15601430054203666,
      "grad_norm": 2.000046315884686,
      "learning_rate": 1.2035344256514808e-06,
      "loss": 0.7516,
      "step": 1691
    },
    {
      "epoch": 0.1561065621035636,
      "grad_norm": 1.6137698284927622,
      "learning_rate": 1.2034632107954975e-06,
      "loss": 0.6281,
      "step": 1692
    },
    {
      "epoch": 0.15619882366509052,
      "grad_norm": 2.3084275818615816,
      "learning_rate": 1.2033919435182674e-06,
      "loss": 0.9654,
      "step": 1693
    },
    {
      "epoch": 0.15629108522661747,
      "grad_norm": 2.2810687982767837,
      "learning_rate": 1.2033206238262488e-06,
      "loss": 0.8687,
      "step": 1694
    },
    {
      "epoch": 0.1563833467881444,
      "grad_norm": 1.8474844628890328,
      "learning_rate": 1.2032492517259051e-06,
      "loss": 0.7925,
      "step": 1695
    },
    {
      "epoch": 0.1563833467881444,
      "eval_GEN Loss": 0.47422850131988525,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.9005847953216374,
      "eval_PRM F1 AUC": 0.7247249869041383,
      "eval_PRM F1 Neg": 0.5853658536585366,
      "eval_PRM Loss": 0.40184977650642395,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.875,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9668269157409668,
      "eval_runtime": 56.8617,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1695
    },
    {
      "epoch": 0.1564756083496713,
      "grad_norm": 3.3225835899501224,
      "learning_rate": 1.203177827223704e-06,
      "loss": 0.9558,
      "step": 1696
    },
    {
      "epoch": 0.15656786991119825,
      "grad_norm": 3.003528462283362,
      "learning_rate": 1.2031063503261182e-06,
      "loss": 1.0253,
      "step": 1697
    },
    {
      "epoch": 0.15666013147272517,
      "grad_norm": 1.869436446365027,
      "learning_rate": 1.203034821039625e-06,
      "loss": 0.8881,
      "step": 1698
    },
    {
      "epoch": 0.1567523930342521,
      "grad_norm": 2.4244399451612337,
      "learning_rate": 1.2029632393707064e-06,
      "loss": 0.9702,
      "step": 1699
    },
    {
      "epoch": 0.15684465459577904,
      "grad_norm": 1.4551539004376537,
      "learning_rate": 1.2028916053258494e-06,
      "loss": 0.6471,
      "step": 1700
    },
    {
      "epoch": 0.15684465459577904,
      "eval_GEN Loss": 0.4727160930633545,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.4090442657470703,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9594951868057251,
      "eval_runtime": 57.034,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1700
    },
    {
      "epoch": 0.15693691615730596,
      "grad_norm": 1.633623089933077,
      "learning_rate": 1.2028199189115455e-06,
      "loss": 0.8667,
      "step": 1701
    },
    {
      "epoch": 0.1570291777188329,
      "grad_norm": 2.680415118185708,
      "learning_rate": 1.2027481801342912e-06,
      "loss": 0.8986,
      "step": 1702
    },
    {
      "epoch": 0.15712143928035982,
      "grad_norm": 1.6584278448257652,
      "learning_rate": 1.2026763890005871e-06,
      "loss": 0.8192,
      "step": 1703
    },
    {
      "epoch": 0.15721370084188674,
      "grad_norm": 1.8433897939640338,
      "learning_rate": 1.2026045455169396e-06,
      "loss": 0.9292,
      "step": 1704
    },
    {
      "epoch": 0.15730596240341368,
      "grad_norm": 1.712124135850333,
      "learning_rate": 1.2025326496898587e-06,
      "loss": 0.814,
      "step": 1705
    },
    {
      "epoch": 0.15730596240341368,
      "eval_GEN Loss": 0.47189226746559143,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.413433700799942,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9611778855323792,
      "eval_runtime": 56.8385,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1705
    },
    {
      "epoch": 0.1573982239649406,
      "grad_norm": 1.5400809782769067,
      "learning_rate": 1.2024607015258605e-06,
      "loss": 0.7189,
      "step": 1706
    },
    {
      "epoch": 0.15749048552646752,
      "grad_norm": 2.0792363943929373,
      "learning_rate": 1.2023887010314642e-06,
      "loss": 0.7638,
      "step": 1707
    },
    {
      "epoch": 0.15758274708799447,
      "grad_norm": 1.8314031543374316,
      "learning_rate": 1.2023166482131952e-06,
      "loss": 0.7807,
      "step": 1708
    },
    {
      "epoch": 0.1576750086495214,
      "grad_norm": 1.1599471168352014,
      "learning_rate": 1.2022445430775827e-06,
      "loss": 0.5619,
      "step": 1709
    },
    {
      "epoch": 0.15776727021104833,
      "grad_norm": 1.919625624387616,
      "learning_rate": 1.2021723856311613e-06,
      "loss": 0.7799,
      "step": 1710
    },
    {
      "epoch": 0.15776727021104833,
      "eval_GEN Loss": 0.4721803367137909,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8757396449704142,
      "eval_PRM F1 AUC": 0.6849135673127291,
      "eval_PRM F1 Neg": 0.5116279069767442,
      "eval_PRM Loss": 0.4043463468551636,
      "eval_PRM NPV": 0.55,
      "eval_PRM Precision": 0.8604651162790697,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9524038434028625,
      "eval_runtime": 56.8285,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1710
    },
    {
      "epoch": 0.15785953177257525,
      "grad_norm": 1.7158155907304815,
      "learning_rate": 1.2021001758804696e-06,
      "loss": 0.7479,
      "step": 1711
    },
    {
      "epoch": 0.15795179333410217,
      "grad_norm": 1.7809784027850344,
      "learning_rate": 1.2020279138320517e-06,
      "loss": 0.7212,
      "step": 1712
    },
    {
      "epoch": 0.15804405489562912,
      "grad_norm": 1.2951346321259123,
      "learning_rate": 1.201955599492456e-06,
      "loss": 0.5891,
      "step": 1713
    },
    {
      "epoch": 0.15813631645715603,
      "grad_norm": 1.094419927256266,
      "learning_rate": 1.2018832328682357e-06,
      "loss": 0.5057,
      "step": 1714
    },
    {
      "epoch": 0.15822857801868295,
      "grad_norm": 1.18666422405567,
      "learning_rate": 1.2018108139659488e-06,
      "loss": 0.6807,
      "step": 1715
    },
    {
      "epoch": 0.15822857801868295,
      "eval_GEN Loss": 0.4728214144706726,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.3869480490684509,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9214543104171753,
      "eval_runtime": 56.0349,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1715
    },
    {
      "epoch": 0.1583208395802099,
      "grad_norm": 1.3590830697605831,
      "learning_rate": 1.2017383427921582e-06,
      "loss": 0.6337,
      "step": 1716
    },
    {
      "epoch": 0.15841310114173682,
      "grad_norm": 1.8654387537840311,
      "learning_rate": 1.201665819353431e-06,
      "loss": 0.7712,
      "step": 1717
    },
    {
      "epoch": 0.15850536270326376,
      "grad_norm": 1.3261619385314058,
      "learning_rate": 1.2015932436563396e-06,
      "loss": 0.6758,
      "step": 1718
    },
    {
      "epoch": 0.15859762426479068,
      "grad_norm": 1.360766445942538,
      "learning_rate": 1.2015206157074607e-06,
      "loss": 0.6135,
      "step": 1719
    },
    {
      "epoch": 0.1586898858263176,
      "grad_norm": 3.092527225707507,
      "learning_rate": 1.2014479355133765e-06,
      "loss": 0.8822,
      "step": 1720
    },
    {
      "epoch": 0.1586898858263176,
      "eval_GEN Loss": 0.4744291305541992,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3729347288608551,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8841947317123413,
      "eval_runtime": 56.9052,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1720
    },
    {
      "epoch": 0.15878214738784455,
      "grad_norm": 1.7859672639020925,
      "learning_rate": 1.2013752030806729e-06,
      "loss": 0.6137,
      "step": 1721
    },
    {
      "epoch": 0.15887440894937147,
      "grad_norm": 1.8132191474772663,
      "learning_rate": 1.2013024184159412e-06,
      "loss": 0.8266,
      "step": 1722
    },
    {
      "epoch": 0.15896667051089838,
      "grad_norm": 1.6281369509399972,
      "learning_rate": 1.2012295815257772e-06,
      "loss": 0.7941,
      "step": 1723
    },
    {
      "epoch": 0.15905893207242533,
      "grad_norm": 2.240640289607103,
      "learning_rate": 1.2011566924167814e-06,
      "loss": 0.8291,
      "step": 1724
    },
    {
      "epoch": 0.15915119363395225,
      "grad_norm": 2.2741129229217854,
      "learning_rate": 1.2010837510955594e-06,
      "loss": 0.7531,
      "step": 1725
    },
    {
      "epoch": 0.15915119363395225,
      "eval_GEN Loss": 0.4747401177883148,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3753626346588135,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8645432591438293,
      "eval_runtime": 56.7347,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 1725
    },
    {
      "epoch": 0.1592434551954792,
      "grad_norm": 1.8661920082388164,
      "learning_rate": 1.201010757568721e-06,
      "loss": 0.6207,
      "step": 1726
    },
    {
      "epoch": 0.1593357167570061,
      "grad_norm": 1.9780752738830902,
      "learning_rate": 1.2009377118428811e-06,
      "loss": 0.8295,
      "step": 1727
    },
    {
      "epoch": 0.15942797831853303,
      "grad_norm": 2.706376669079482,
      "learning_rate": 1.2008646139246594e-06,
      "loss": 0.7115,
      "step": 1728
    },
    {
      "epoch": 0.15952023988005998,
      "grad_norm": 1.9100448997257597,
      "learning_rate": 1.2007914638206797e-06,
      "loss": 0.7295,
      "step": 1729
    },
    {
      "epoch": 0.1596125014415869,
      "grad_norm": 2.0075331662337668,
      "learning_rate": 1.2007182615375714e-06,
      "loss": 0.8302,
      "step": 1730
    },
    {
      "epoch": 0.1596125014415869,
      "eval_GEN Loss": 0.4771677553653717,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8875739644970414,
      "eval_PRM F1 AUC": 0.7126767941330541,
      "eval_PRM F1 Neg": 0.5581395348837209,
      "eval_PRM Loss": 0.36156365275382996,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.872093023255814,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.8718149065971375,
      "eval_runtime": 57.0068,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 1730
    },
    {
      "epoch": 0.15970476300311381,
      "grad_norm": 2.2282521166062117,
      "learning_rate": 1.2006450070819681e-06,
      "loss": 0.7787,
      "step": 1731
    },
    {
      "epoch": 0.15979702456464076,
      "grad_norm": 2.727345528912454,
      "learning_rate": 1.2005717004605081e-06,
      "loss": 0.8815,
      "step": 1732
    },
    {
      "epoch": 0.15988928612616768,
      "grad_norm": 5.409008356411552,
      "learning_rate": 1.2004983416798346e-06,
      "loss": 1.3479,
      "step": 1733
    },
    {
      "epoch": 0.15998154768769463,
      "grad_norm": 1.6377853480910112,
      "learning_rate": 1.2004249307465956e-06,
      "loss": 0.7857,
      "step": 1734
    },
    {
      "epoch": 0.16007380924922154,
      "grad_norm": 1.961215619307502,
      "learning_rate": 1.2003514676674435e-06,
      "loss": 0.9155,
      "step": 1735
    },
    {
      "epoch": 0.16007380924922154,
      "eval_GEN Loss": 0.4761819541454315,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8875739644970414,
      "eval_PRM F1 AUC": 0.7126767941330541,
      "eval_PRM F1 Neg": 0.5581395348837209,
      "eval_PRM Loss": 0.3655785322189331,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.872093023255814,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.871874988079071,
      "eval_runtime": 56.9603,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1735
    },
    {
      "epoch": 0.16016607081074846,
      "grad_norm": 2.0968914042084403,
      "learning_rate": 1.200277952449036e-06,
      "loss": 0.8153,
      "step": 1736
    },
    {
      "epoch": 0.1602583323722754,
      "grad_norm": 2.408793464839001,
      "learning_rate": 1.2002043850980346e-06,
      "loss": 0.7876,
      "step": 1737
    },
    {
      "epoch": 0.16035059393380233,
      "grad_norm": 2.041860953980221,
      "learning_rate": 1.2001307656211069e-06,
      "loss": 0.6351,
      "step": 1738
    },
    {
      "epoch": 0.16044285549532925,
      "grad_norm": 1.9127806526581694,
      "learning_rate": 1.2000570940249237e-06,
      "loss": 0.7659,
      "step": 1739
    },
    {
      "epoch": 0.1605351170568562,
      "grad_norm": 3.3468723163460665,
      "learning_rate": 1.1999833703161614e-06,
      "loss": 0.9352,
      "step": 1740
    },
    {
      "epoch": 0.1605351170568562,
      "eval_GEN Loss": 0.4743192195892334,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8875739644970414,
      "eval_PRM F1 AUC": 0.7126767941330541,
      "eval_PRM F1 Neg": 0.5581395348837209,
      "eval_PRM Loss": 0.3662451207637787,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.872093023255814,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.8649038672447205,
      "eval_runtime": 57.2074,
      "eval_samples_per_second": 1.136,
      "eval_steps_per_second": 0.087,
      "step": 1740
    },
    {
      "epoch": 0.1606273786183831,
      "grad_norm": 3.228071781396564,
      "learning_rate": 1.1999095945015013e-06,
      "loss": 0.9619,
      "step": 1741
    },
    {
      "epoch": 0.16071964017991006,
      "grad_norm": 1.9599230495336764,
      "learning_rate": 1.1998357665876288e-06,
      "loss": 0.7585,
      "step": 1742
    },
    {
      "epoch": 0.16081190174143697,
      "grad_norm": 1.9625157070892876,
      "learning_rate": 1.199761886581234e-06,
      "loss": 0.9341,
      "step": 1743
    },
    {
      "epoch": 0.1609041633029639,
      "grad_norm": 2.6344561913816547,
      "learning_rate": 1.1996879544890125e-06,
      "loss": 0.7793,
      "step": 1744
    },
    {
      "epoch": 0.16099642486449084,
      "grad_norm": 2.375414692367507,
      "learning_rate": 1.199613970317664e-06,
      "loss": 0.7729,
      "step": 1745
    },
    {
      "epoch": 0.16099642486449084,
      "eval_GEN Loss": 0.47493529319763184,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9056603773584906,
      "eval_PRM F1 AUC": 0.8467784180199058,
      "eval_PRM F1 Neg": 0.7169811320754716,
      "eval_PRM Loss": 0.38084328174591064,
      "eval_PRM NPV": 0.6333333333333333,
      "eval_PRM Precision": 0.9473684210526315,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.846875011920929,
      "eval_runtime": 56.7965,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 1745
    },
    {
      "epoch": 0.16108868642601776,
      "grad_norm": 2.346254146722747,
      "learning_rate": 1.1995399340738932e-06,
      "loss": 0.8282,
      "step": 1746
    },
    {
      "epoch": 0.16118094798754468,
      "grad_norm": 1.682342164816387,
      "learning_rate": 1.1994658457644091e-06,
      "loss": 0.8964,
      "step": 1747
    },
    {
      "epoch": 0.16127320954907162,
      "grad_norm": 1.9917021102123866,
      "learning_rate": 1.1993917053959258e-06,
      "loss": 0.7675,
      "step": 1748
    },
    {
      "epoch": 0.16136547111059854,
      "grad_norm": 1.5307292646365027,
      "learning_rate": 1.1993175129751622e-06,
      "loss": 0.6664,
      "step": 1749
    },
    {
      "epoch": 0.1614577326721255,
      "grad_norm": 2.145969974023176,
      "learning_rate": 1.1992432685088414e-06,
      "loss": 0.8581,
      "step": 1750
    },
    {
      "epoch": 0.1614577326721255,
      "eval_GEN Loss": 0.4752160906791687,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8987341772151899,
      "eval_PRM F1 AUC": 0.8407543216343635,
      "eval_PRM F1 Neg": 0.7037037037037037,
      "eval_PRM Loss": 0.3944178521633148,
      "eval_PRM NPV": 0.6129032258064516,
      "eval_PRM Precision": 0.9466666666666667,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.8573317527770996,
      "eval_runtime": 56.9403,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1750
    },
    {
      "epoch": 0.1615499942336524,
      "grad_norm": 1.857511557807682,
      "learning_rate": 1.199168972003692e-06,
      "loss": 0.7344,
      "step": 1751
    },
    {
      "epoch": 0.16164225579517932,
      "grad_norm": 3.3110136809654356,
      "learning_rate": 1.1990946234664464e-06,
      "loss": 0.8864,
      "step": 1752
    },
    {
      "epoch": 0.16173451735670627,
      "grad_norm": 1.851308184913895,
      "learning_rate": 1.1990202229038426e-06,
      "loss": 0.821,
      "step": 1753
    },
    {
      "epoch": 0.1618267789182332,
      "grad_norm": 1.7270899120519583,
      "learning_rate": 1.1989457703226222e-06,
      "loss": 0.8285,
      "step": 1754
    },
    {
      "epoch": 0.1619190404797601,
      "grad_norm": 1.8888581494402377,
      "learning_rate": 1.198871265729533e-06,
      "loss": 0.8772,
      "step": 1755
    },
    {
      "epoch": 0.1619190404797601,
      "eval_GEN Loss": 0.47463130950927734,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3658333420753479,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8574519157409668,
      "eval_runtime": 56.0056,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1755
    },
    {
      "epoch": 0.16201130204128705,
      "grad_norm": 1.6795665178601533,
      "learning_rate": 1.1987967091313266e-06,
      "loss": 0.85,
      "step": 1756
    },
    {
      "epoch": 0.16210356360281397,
      "grad_norm": 1.901449510428993,
      "learning_rate": 1.198722100534759e-06,
      "loss": 0.8562,
      "step": 1757
    },
    {
      "epoch": 0.16219582516434092,
      "grad_norm": 3.4299953114132395,
      "learning_rate": 1.1986474399465916e-06,
      "loss": 1.1412,
      "step": 1758
    },
    {
      "epoch": 0.16228808672586784,
      "grad_norm": 2.148886315304666,
      "learning_rate": 1.1985727273735904e-06,
      "loss": 0.9179,
      "step": 1759
    },
    {
      "epoch": 0.16238034828739475,
      "grad_norm": 1.6700772269498403,
      "learning_rate": 1.1984979628225258e-06,
      "loss": 0.8421,
      "step": 1760
    },
    {
      "epoch": 0.16238034828739475,
      "eval_GEN Loss": 0.4730831980705261,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3528091013431549,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8771033883094788,
      "eval_runtime": 56.0585,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1760
    },
    {
      "epoch": 0.1624726098489217,
      "grad_norm": 1.5933767095295974,
      "learning_rate": 1.198423146300173e-06,
      "loss": 0.7713,
      "step": 1761
    },
    {
      "epoch": 0.16256487141044862,
      "grad_norm": 2.0012070258657455,
      "learning_rate": 1.198348277813312e-06,
      "loss": 0.9833,
      "step": 1762
    },
    {
      "epoch": 0.16265713297197554,
      "grad_norm": 1.7347135422048094,
      "learning_rate": 1.1982733573687279e-06,
      "loss": 0.7167,
      "step": 1763
    },
    {
      "epoch": 0.16274939453350248,
      "grad_norm": 1.9715518457252303,
      "learning_rate": 1.1981983849732096e-06,
      "loss": 0.7582,
      "step": 1764
    },
    {
      "epoch": 0.1628416560950294,
      "grad_norm": 2.396025379859153,
      "learning_rate": 1.1981233606335515e-06,
      "loss": 0.9252,
      "step": 1765
    },
    {
      "epoch": 0.1628416560950294,
      "eval_GEN Loss": 0.47371742129325867,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3666206896305084,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8832331895828247,
      "eval_runtime": 55.9256,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1765
    },
    {
      "epoch": 0.16293391765655635,
      "grad_norm": 1.67230443249086,
      "learning_rate": 1.1980482843565524e-06,
      "loss": 0.4898,
      "step": 1766
    },
    {
      "epoch": 0.16302617921808327,
      "grad_norm": 2.6798053789935015,
      "learning_rate": 1.1979731561490158e-06,
      "loss": 0.7325,
      "step": 1767
    },
    {
      "epoch": 0.16311844077961019,
      "grad_norm": 1.2788674533092632,
      "learning_rate": 1.1978979760177499e-06,
      "loss": 0.6708,
      "step": 1768
    },
    {
      "epoch": 0.16321070234113713,
      "grad_norm": 3.2214528116004915,
      "learning_rate": 1.1978227439695675e-06,
      "loss": 0.7314,
      "step": 1769
    },
    {
      "epoch": 0.16330296390266405,
      "grad_norm": 1.4528340184239323,
      "learning_rate": 1.1977474600112865e-06,
      "loss": 0.6578,
      "step": 1770
    },
    {
      "epoch": 0.16330296390266405,
      "eval_GEN Loss": 0.4762537181377411,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3963039815425873,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8923677802085876,
      "eval_runtime": 55.9566,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1770
    },
    {
      "epoch": 0.16339522546419097,
      "grad_norm": 2.2097031063476797,
      "learning_rate": 1.197672124149729e-06,
      "loss": 0.8733,
      "step": 1771
    },
    {
      "epoch": 0.16348748702571791,
      "grad_norm": 1.8180984617525522,
      "learning_rate": 1.1975967363917225e-06,
      "loss": 0.8292,
      "step": 1772
    },
    {
      "epoch": 0.16357974858724483,
      "grad_norm": 1.3777427473261157,
      "learning_rate": 1.1975212967440983e-06,
      "loss": 0.8133,
      "step": 1773
    },
    {
      "epoch": 0.16367201014877178,
      "grad_norm": 2.904575933076612,
      "learning_rate": 1.197445805213693e-06,
      "loss": 0.8565,
      "step": 1774
    },
    {
      "epoch": 0.1637642717102987,
      "grad_norm": 2.2467603792144306,
      "learning_rate": 1.197370261807348e-06,
      "loss": 0.7633,
      "step": 1775
    },
    {
      "epoch": 0.1637642717102987,
      "eval_GEN Loss": 0.47734469175338745,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.44640040397644043,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9230168461799622,
      "eval_runtime": 55.9677,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1775
    },
    {
      "epoch": 0.16385653327182562,
      "grad_norm": 2.054697081292661,
      "learning_rate": 1.1972946665319088e-06,
      "loss": 0.867,
      "step": 1776
    },
    {
      "epoch": 0.16394879483335256,
      "grad_norm": 2.8727010273222864,
      "learning_rate": 1.1972190193942259e-06,
      "loss": 0.8253,
      "step": 1777
    },
    {
      "epoch": 0.16404105639487948,
      "grad_norm": 2.716670343247038,
      "learning_rate": 1.1971433204011552e-06,
      "loss": 1.0008,
      "step": 1778
    },
    {
      "epoch": 0.1641333179564064,
      "grad_norm": 3.8480085615947606,
      "learning_rate": 1.197067569559556e-06,
      "loss": 1.0191,
      "step": 1779
    },
    {
      "epoch": 0.16422557951793335,
      "grad_norm": 1.5197868351541262,
      "learning_rate": 1.1969917668762932e-06,
      "loss": 0.816,
      "step": 1780
    },
    {
      "epoch": 0.16422557951793335,
      "eval_GEN Loss": 0.47710874676704407,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.4292348325252533,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.922776460647583,
      "eval_runtime": 55.9198,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1780
    },
    {
      "epoch": 0.16431784107946026,
      "grad_norm": 2.996241200808312,
      "learning_rate": 1.1969159123582361e-06,
      "loss": 0.7155,
      "step": 1781
    },
    {
      "epoch": 0.1644101026409872,
      "grad_norm": 1.4503014184697487,
      "learning_rate": 1.1968400060122589e-06,
      "loss": 0.6684,
      "step": 1782
    },
    {
      "epoch": 0.16450236420251413,
      "grad_norm": 2.321984586660832,
      "learning_rate": 1.1967640478452402e-06,
      "loss": 0.777,
      "step": 1783
    },
    {
      "epoch": 0.16459462576404105,
      "grad_norm": 1.6149520618062305,
      "learning_rate": 1.1966880378640636e-06,
      "loss": 0.7009,
      "step": 1784
    },
    {
      "epoch": 0.164686887325568,
      "grad_norm": 1.4362764575132554,
      "learning_rate": 1.1966119760756173e-06,
      "loss": 0.7981,
      "step": 1785
    },
    {
      "epoch": 0.164686887325568,
      "eval_GEN Loss": 0.47585561871528625,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.40282073616981506,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9460336565971375,
      "eval_runtime": 55.8059,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 1785
    },
    {
      "epoch": 0.1647791488870949,
      "grad_norm": 1.68124564818914,
      "learning_rate": 1.1965358624867936e-06,
      "loss": 0.8926,
      "step": 1786
    },
    {
      "epoch": 0.16487141044862183,
      "grad_norm": 1.8189750538289764,
      "learning_rate": 1.196459697104491e-06,
      "loss": 0.7107,
      "step": 1787
    },
    {
      "epoch": 0.16496367201014878,
      "grad_norm": 3.8822325320822824,
      "learning_rate": 1.1963834799356106e-06,
      "loss": 0.9042,
      "step": 1788
    },
    {
      "epoch": 0.1650559335716757,
      "grad_norm": 2.377530408699936,
      "learning_rate": 1.19630721098706e-06,
      "loss": 0.7527,
      "step": 1789
    },
    {
      "epoch": 0.16514819513320264,
      "grad_norm": 1.4515146490342556,
      "learning_rate": 1.1962308902657508e-06,
      "loss": 0.7672,
      "step": 1790
    },
    {
      "epoch": 0.16514819513320264,
      "eval_GEN Loss": 0.47496604919433594,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8757396449704142,
      "eval_PRM F1 AUC": 0.6849135673127291,
      "eval_PRM F1 Neg": 0.5116279069767442,
      "eval_PRM Loss": 0.40148061513900757,
      "eval_PRM NPV": 0.55,
      "eval_PRM Precision": 0.8604651162790697,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9661658406257629,
      "eval_runtime": 55.8316,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1790
    },
    {
      "epoch": 0.16524045669472956,
      "grad_norm": 1.8367418287024868,
      "learning_rate": 1.1961545177785994e-06,
      "loss": 0.9241,
      "step": 1791
    },
    {
      "epoch": 0.16533271825625648,
      "grad_norm": 1.7373451874372527,
      "learning_rate": 1.1960780935325263e-06,
      "loss": 0.7596,
      "step": 1792
    },
    {
      "epoch": 0.16542497981778342,
      "grad_norm": 2.199842245342328,
      "learning_rate": 1.1960016175344574e-06,
      "loss": 1.0013,
      "step": 1793
    },
    {
      "epoch": 0.16551724137931034,
      "grad_norm": 1.3624274925812592,
      "learning_rate": 1.1959250897913235e-06,
      "loss": 0.6842,
      "step": 1794
    },
    {
      "epoch": 0.16560950294083726,
      "grad_norm": 2.105499786133481,
      "learning_rate": 1.195848510310059e-06,
      "loss": 0.887,
      "step": 1795
    },
    {
      "epoch": 0.16560950294083726,
      "eval_GEN Loss": 0.4728744924068451,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8809523809523809,
      "eval_PRM F1 AUC": 0.7066526977475118,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.4052446782588959,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.8705882352941177,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9497596025466919,
      "eval_runtime": 56.9528,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1795
    },
    {
      "epoch": 0.1657017645023642,
      "grad_norm": 2.4826299160827947,
      "learning_rate": 1.195771879097604e-06,
      "loss": 0.8808,
      "step": 1796
    },
    {
      "epoch": 0.16579402606389113,
      "grad_norm": 1.4297200589498609,
      "learning_rate": 1.1956951961609032e-06,
      "loss": 0.7984,
      "step": 1797
    },
    {
      "epoch": 0.16588628762541807,
      "grad_norm": 1.4074620516650813,
      "learning_rate": 1.195618461506905e-06,
      "loss": 0.606,
      "step": 1798
    },
    {
      "epoch": 0.165978549186945,
      "grad_norm": 1.8470885961957082,
      "learning_rate": 1.1955416751425638e-06,
      "loss": 0.922,
      "step": 1799
    },
    {
      "epoch": 0.1660708107484719,
      "grad_norm": 1.6546661085060437,
      "learning_rate": 1.1954648370748382e-06,
      "loss": 0.7227,
      "step": 1800
    },
    {
      "epoch": 0.1660708107484719,
      "eval_GEN Loss": 0.47097355127334595,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4237675070762634,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9322115182876587,
      "eval_runtime": 56.9229,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 1800
    },
    {
      "epoch": 0.16616307230999885,
      "grad_norm": 1.6426354604079443,
      "learning_rate": 1.1953879473106909e-06,
      "loss": 0.746,
      "step": 1801
    },
    {
      "epoch": 0.16625533387152577,
      "grad_norm": 2.074416629403325,
      "learning_rate": 1.19531100585709e-06,
      "loss": 0.8324,
      "step": 1802
    },
    {
      "epoch": 0.1663475954330527,
      "grad_norm": 1.7389515438506715,
      "learning_rate": 1.1952340127210084e-06,
      "loss": 0.7606,
      "step": 1803
    },
    {
      "epoch": 0.16643985699457964,
      "grad_norm": 1.2344952399735567,
      "learning_rate": 1.1951569679094225e-06,
      "loss": 0.64,
      "step": 1804
    },
    {
      "epoch": 0.16653211855610656,
      "grad_norm": 1.265859398615304,
      "learning_rate": 1.195079871429315e-06,
      "loss": 0.6751,
      "step": 1805
    },
    {
      "epoch": 0.16653211855610656,
      "eval_GEN Loss": 0.4726194739341736,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4417702555656433,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9378004670143127,
      "eval_runtime": 56.0617,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1805
    },
    {
      "epoch": 0.1666243801176335,
      "grad_norm": 2.996394260517124,
      "learning_rate": 1.1950027232876723e-06,
      "loss": 0.649,
      "step": 1806
    },
    {
      "epoch": 0.16671664167916042,
      "grad_norm": 2.070350182240329,
      "learning_rate": 1.1949255234914855e-06,
      "loss": 0.6666,
      "step": 1807
    },
    {
      "epoch": 0.16680890324068734,
      "grad_norm": 2.5651078496165804,
      "learning_rate": 1.1948482720477506e-06,
      "loss": 0.8245,
      "step": 1808
    },
    {
      "epoch": 0.16690116480221429,
      "grad_norm": 2.818405435354929,
      "learning_rate": 1.1947709689634685e-06,
      "loss": 0.8456,
      "step": 1809
    },
    {
      "epoch": 0.1669934263637412,
      "grad_norm": 1.3221976867543705,
      "learning_rate": 1.1946936142456443e-06,
      "loss": 0.6072,
      "step": 1810
    },
    {
      "epoch": 0.1669934263637412,
      "eval_GEN Loss": 0.47267863154411316,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.4132683575153351,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9447716474533081,
      "eval_runtime": 55.9272,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1810
    },
    {
      "epoch": 0.16708568792526812,
      "grad_norm": 1.57219569330212,
      "learning_rate": 1.194616207901288e-06,
      "loss": 0.7223,
      "step": 1811
    },
    {
      "epoch": 0.16717794948679507,
      "grad_norm": 1.4754263132976029,
      "learning_rate": 1.1945387499374146e-06,
      "loss": 0.6975,
      "step": 1812
    },
    {
      "epoch": 0.167270211048322,
      "grad_norm": 2.8838211570500176,
      "learning_rate": 1.1944612403610429e-06,
      "loss": 0.9909,
      "step": 1813
    },
    {
      "epoch": 0.16736247260984893,
      "grad_norm": 2.0838869934665407,
      "learning_rate": 1.1943836791791974e-06,
      "loss": 0.8582,
      "step": 1814
    },
    {
      "epoch": 0.16745473417137585,
      "grad_norm": 2.0720505816262813,
      "learning_rate": 1.1943060663989065e-06,
      "loss": 0.7503,
      "step": 1815
    },
    {
      "epoch": 0.16745473417137585,
      "eval_GEN Loss": 0.4740515649318695,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8690476190476191,
      "eval_PRM F1 AUC": 0.678889470927187,
      "eval_PRM F1 Neg": 0.5,
      "eval_PRM Loss": 0.40221935510635376,
      "eval_PRM NPV": 0.5238095238095238,
      "eval_PRM Precision": 0.8588235294117647,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.965624988079071,
      "eval_runtime": 55.9164,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1815
    },
    {
      "epoch": 0.16754699573290277,
      "grad_norm": 1.9489886302628043,
      "learning_rate": 1.194228402027204e-06,
      "loss": 0.6032,
      "step": 1816
    },
    {
      "epoch": 0.16763925729442972,
      "grad_norm": 3.146227514371221,
      "learning_rate": 1.1941506860711277e-06,
      "loss": 0.8231,
      "step": 1817
    },
    {
      "epoch": 0.16773151885595664,
      "grad_norm": 1.8486025174842518,
      "learning_rate": 1.1940729185377204e-06,
      "loss": 0.7736,
      "step": 1818
    },
    {
      "epoch": 0.16782378041748355,
      "grad_norm": 2.3156176425509485,
      "learning_rate": 1.1939950994340295e-06,
      "loss": 0.8739,
      "step": 1819
    },
    {
      "epoch": 0.1679160419790105,
      "grad_norm": 1.8496325085265781,
      "learning_rate": 1.1939172287671069e-06,
      "loss": 0.7784,
      "step": 1820
    },
    {
      "epoch": 0.1679160419790105,
      "eval_GEN Loss": 0.47243618965148926,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8727272727272727,
      "eval_PRM F1 AUC": 0.7163436354112102,
      "eval_PRM F1 Neg": 0.5531914893617021,
      "eval_PRM Loss": 0.40494734048843384,
      "eval_PRM NPV": 0.5416666666666666,
      "eval_PRM Precision": 0.8780487804878049,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.9459735751152039,
      "eval_runtime": 55.886,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1820
    },
    {
      "epoch": 0.16800830354053742,
      "grad_norm": 1.369839127474054,
      "learning_rate": 1.1938393065440094e-06,
      "loss": 0.8119,
      "step": 1821
    },
    {
      "epoch": 0.16810056510206436,
      "grad_norm": 1.5572494870346232,
      "learning_rate": 1.1937613327717989e-06,
      "loss": 0.6656,
      "step": 1822
    },
    {
      "epoch": 0.16819282666359128,
      "grad_norm": 1.2733382757256089,
      "learning_rate": 1.193683307457541e-06,
      "loss": 0.6123,
      "step": 1823
    },
    {
      "epoch": 0.1682850882251182,
      "grad_norm": 2.302630167731284,
      "learning_rate": 1.1936052306083069e-06,
      "loss": 0.6936,
      "step": 1824
    },
    {
      "epoch": 0.16837734978664515,
      "grad_norm": 2.0470012487601275,
      "learning_rate": 1.1935271022311715e-06,
      "loss": 0.9461,
      "step": 1825
    },
    {
      "epoch": 0.16837734978664515,
      "eval_GEN Loss": 0.47079581022262573,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4052063822746277,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9320312738418579,
      "eval_runtime": 55.9966,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1825
    },
    {
      "epoch": 0.16846961134817207,
      "grad_norm": 2.6111676028904975,
      "learning_rate": 1.1934489223332154e-06,
      "loss": 0.924,
      "step": 1826
    },
    {
      "epoch": 0.16856187290969898,
      "grad_norm": 2.276340324668247,
      "learning_rate": 1.1933706909215232e-06,
      "loss": 0.8264,
      "step": 1827
    },
    {
      "epoch": 0.16865413447122593,
      "grad_norm": 1.3305571111012198,
      "learning_rate": 1.1932924080031843e-06,
      "loss": 0.7021,
      "step": 1828
    },
    {
      "epoch": 0.16874639603275285,
      "grad_norm": 1.7884137397764654,
      "learning_rate": 1.193214073585293e-06,
      "loss": 0.617,
      "step": 1829
    },
    {
      "epoch": 0.1688386575942798,
      "grad_norm": 1.9236606276039037,
      "learning_rate": 1.1931356876749479e-06,
      "loss": 0.7649,
      "step": 1830
    },
    {
      "epoch": 0.1688386575942798,
      "eval_GEN Loss": 0.4701881408691406,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3850785493850708,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9339542984962463,
      "eval_runtime": 55.9442,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1830
    },
    {
      "epoch": 0.1689309191558067,
      "grad_norm": 2.432346492635802,
      "learning_rate": 1.1930572502792524e-06,
      "loss": 1.0868,
      "step": 1831
    },
    {
      "epoch": 0.16902318071733363,
      "grad_norm": 1.7381174837752686,
      "learning_rate": 1.1929787614053148e-06,
      "loss": 0.6999,
      "step": 1832
    },
    {
      "epoch": 0.16911544227886058,
      "grad_norm": 3.1065036645173594,
      "learning_rate": 1.1929002210602479e-06,
      "loss": 0.8944,
      "step": 1833
    },
    {
      "epoch": 0.1692077038403875,
      "grad_norm": 1.3985172252832394,
      "learning_rate": 1.192821629251169e-06,
      "loss": 0.7605,
      "step": 1834
    },
    {
      "epoch": 0.16929996540191442,
      "grad_norm": 2.0996990775788413,
      "learning_rate": 1.1927429859852003e-06,
      "loss": 0.7781,
      "step": 1835
    },
    {
      "epoch": 0.16929996540191442,
      "eval_GEN Loss": 0.46855273842811584,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8941176470588236,
      "eval_PRM F1 AUC": 0.7187008905185961,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.3788681626319885,
      "eval_PRM NPV": 0.631578947368421,
      "eval_PRM Precision": 0.8735632183908046,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.5217391304347826,
      "eval_loss": 0.9432091116905212,
      "eval_runtime": 55.8578,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1835
    },
    {
      "epoch": 0.16939222696344136,
      "grad_norm": 1.373389670403736,
      "learning_rate": 1.1926642912694685e-06,
      "loss": 0.7181,
      "step": 1836
    },
    {
      "epoch": 0.16948448852496828,
      "grad_norm": 1.3460589965050005,
      "learning_rate": 1.1925855451111053e-06,
      "loss": 0.7506,
      "step": 1837
    },
    {
      "epoch": 0.16957675008649523,
      "grad_norm": 2.1119266670809163,
      "learning_rate": 1.1925067475172466e-06,
      "loss": 0.9014,
      "step": 1838
    },
    {
      "epoch": 0.16966901164802214,
      "grad_norm": 1.7106487417678713,
      "learning_rate": 1.1924278984950331e-06,
      "loss": 0.6861,
      "step": 1839
    },
    {
      "epoch": 0.16976127320954906,
      "grad_norm": 2.2103833265963777,
      "learning_rate": 1.1923489980516104e-06,
      "loss": 0.7988,
      "step": 1840
    },
    {
      "epoch": 0.16976127320954906,
      "eval_GEN Loss": 0.46745893359184265,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.36948809027671814,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9295673370361328,
      "eval_runtime": 55.898,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1840
    },
    {
      "epoch": 0.169853534771076,
      "grad_norm": 2.5020171248511134,
      "learning_rate": 1.1922700461941283e-06,
      "loss": 0.6389,
      "step": 1841
    },
    {
      "epoch": 0.16994579633260293,
      "grad_norm": 2.3921345275085897,
      "learning_rate": 1.1921910429297418e-06,
      "loss": 0.8788,
      "step": 1842
    },
    {
      "epoch": 0.17003805789412985,
      "grad_norm": 1.0725633407914312,
      "learning_rate": 1.1921119882656101e-06,
      "loss": 0.597,
      "step": 1843
    },
    {
      "epoch": 0.1701303194556568,
      "grad_norm": 1.9758289520461825,
      "learning_rate": 1.1920328822088974e-06,
      "loss": 0.65,
      "step": 1844
    },
    {
      "epoch": 0.1702225810171837,
      "grad_norm": 1.3922603490803653,
      "learning_rate": 1.1919537247667726e-06,
      "loss": 0.668,
      "step": 1845
    },
    {
      "epoch": 0.1702225810171837,
      "eval_GEN Loss": 0.46774396300315857,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3799689710140228,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9161057472229004,
      "eval_runtime": 56.1623,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 1845
    },
    {
      "epoch": 0.17031484257871066,
      "grad_norm": 1.6231177902314144,
      "learning_rate": 1.1918745159464087e-06,
      "loss": 0.6615,
      "step": 1846
    },
    {
      "epoch": 0.17040710414023758,
      "grad_norm": 1.3090450478129656,
      "learning_rate": 1.191795255754984e-06,
      "loss": 0.7573,
      "step": 1847
    },
    {
      "epoch": 0.1704993657017645,
      "grad_norm": 2.794326829308904,
      "learning_rate": 1.191715944199681e-06,
      "loss": 0.7382,
      "step": 1848
    },
    {
      "epoch": 0.17059162726329144,
      "grad_norm": 3.159342566083108,
      "learning_rate": 1.191636581287687e-06,
      "loss": 1.1812,
      "step": 1849
    },
    {
      "epoch": 0.17068388882481836,
      "grad_norm": 1.5542403126104076,
      "learning_rate": 1.1915571670261944e-06,
      "loss": 0.4285,
      "step": 1850
    },
    {
      "epoch": 0.17068388882481836,
      "eval_GEN Loss": 0.4687207341194153,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.38126498460769653,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9199519157409668,
      "eval_runtime": 55.8179,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 1850
    },
    {
      "epoch": 0.17077615038634528,
      "grad_norm": 1.6366686188141122,
      "learning_rate": 1.1914777014223993e-06,
      "loss": 0.6588,
      "step": 1851
    },
    {
      "epoch": 0.17086841194787222,
      "grad_norm": 1.763222407711819,
      "learning_rate": 1.1913981844835033e-06,
      "loss": 0.7652,
      "step": 1852
    },
    {
      "epoch": 0.17096067350939914,
      "grad_norm": 2.261970091002136,
      "learning_rate": 1.1913186162167122e-06,
      "loss": 0.8263,
      "step": 1853
    },
    {
      "epoch": 0.1710529350709261,
      "grad_norm": 2.233581755609803,
      "learning_rate": 1.191238996629237e-06,
      "loss": 0.7478,
      "step": 1854
    },
    {
      "epoch": 0.171145196632453,
      "grad_norm": 2.5150143585753444,
      "learning_rate": 1.1911593257282926e-06,
      "loss": 0.9442,
      "step": 1855
    },
    {
      "epoch": 0.171145196632453,
      "eval_GEN Loss": 0.4689250588417053,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.37381306290626526,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9309495091438293,
      "eval_runtime": 55.769,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 1855
    },
    {
      "epoch": 0.17123745819397992,
      "grad_norm": 1.9653967057939061,
      "learning_rate": 1.1910796035210986e-06,
      "loss": 0.7906,
      "step": 1856
    },
    {
      "epoch": 0.17132971975550687,
      "grad_norm": 2.4563107770390133,
      "learning_rate": 1.1909998300148802e-06,
      "loss": 0.7522,
      "step": 1857
    },
    {
      "epoch": 0.1714219813170338,
      "grad_norm": 2.425855312863495,
      "learning_rate": 1.1909200052168662e-06,
      "loss": 1.12,
      "step": 1858
    },
    {
      "epoch": 0.1715142428785607,
      "grad_norm": 1.9460311578544538,
      "learning_rate": 1.1908401291342906e-06,
      "loss": 0.8324,
      "step": 1859
    },
    {
      "epoch": 0.17160650444008765,
      "grad_norm": 1.5972274760046872,
      "learning_rate": 1.1907602017743919e-06,
      "loss": 0.8474,
      "step": 1860
    },
    {
      "epoch": 0.17160650444008765,
      "eval_GEN Loss": 0.46865659952163696,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.38174164295196533,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9336538314819336,
      "eval_runtime": 55.9742,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1860
    },
    {
      "epoch": 0.17169876600161457,
      "grad_norm": 1.7692435542544476,
      "learning_rate": 1.1906802231444132e-06,
      "loss": 0.5984,
      "step": 1861
    },
    {
      "epoch": 0.17179102756314152,
      "grad_norm": 1.2937208992413296,
      "learning_rate": 1.190600193251602e-06,
      "loss": 0.6311,
      "step": 1862
    },
    {
      "epoch": 0.17188328912466844,
      "grad_norm": 2.1915470807528123,
      "learning_rate": 1.1905201121032114e-06,
      "loss": 1.0597,
      "step": 1863
    },
    {
      "epoch": 0.17197555068619536,
      "grad_norm": 2.094873538641951,
      "learning_rate": 1.1904399797064978e-06,
      "loss": 0.6461,
      "step": 1864
    },
    {
      "epoch": 0.1720678122477223,
      "grad_norm": 1.9349259284452442,
      "learning_rate": 1.1903597960687235e-06,
      "loss": 0.7687,
      "step": 1865
    },
    {
      "epoch": 0.1720678122477223,
      "eval_GEN Loss": 0.4682375192642212,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3880854845046997,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9427884817123413,
      "eval_runtime": 56.1305,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 1865
    },
    {
      "epoch": 0.17216007380924922,
      "grad_norm": 1.6799371021089247,
      "learning_rate": 1.1902795611971545e-06,
      "loss": 0.8063,
      "step": 1866
    },
    {
      "epoch": 0.17225233537077614,
      "grad_norm": 2.0228644674773713,
      "learning_rate": 1.1901992750990618e-06,
      "loss": 0.8166,
      "step": 1867
    },
    {
      "epoch": 0.17234459693230308,
      "grad_norm": 2.5252102679387276,
      "learning_rate": 1.1901189377817212e-06,
      "loss": 0.6701,
      "step": 1868
    },
    {
      "epoch": 0.17243685849383,
      "grad_norm": 1.9980916127907438,
      "learning_rate": 1.190038549252413e-06,
      "loss": 0.809,
      "step": 1869
    },
    {
      "epoch": 0.17252912005535695,
      "grad_norm": 2.0079522762005446,
      "learning_rate": 1.189958109518422e-06,
      "loss": 1.0007,
      "step": 1870
    },
    {
      "epoch": 0.17252912005535695,
      "eval_GEN Loss": 0.4692970812320709,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.41786956787109375,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9440504908561707,
      "eval_runtime": 55.8881,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1870
    },
    {
      "epoch": 0.17262138161688387,
      "grad_norm": 1.8787687357136489,
      "learning_rate": 1.1898776185870378e-06,
      "loss": 0.8742,
      "step": 1871
    },
    {
      "epoch": 0.1727136431784108,
      "grad_norm": 2.2622161929407705,
      "learning_rate": 1.189797076465555e-06,
      "loss": 0.8112,
      "step": 1872
    },
    {
      "epoch": 0.17280590473993773,
      "grad_norm": 1.4773040619683082,
      "learning_rate": 1.1897164831612717e-06,
      "loss": 0.6426,
      "step": 1873
    },
    {
      "epoch": 0.17289816630146465,
      "grad_norm": 2.7782129345098676,
      "learning_rate": 1.1896358386814922e-06,
      "loss": 0.862,
      "step": 1874
    },
    {
      "epoch": 0.17299042786299157,
      "grad_norm": 3.2884494924228718,
      "learning_rate": 1.189555143033524e-06,
      "loss": 0.6958,
      "step": 1875
    },
    {
      "epoch": 0.17299042786299157,
      "eval_GEN Loss": 0.4688989818096161,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.4337483048439026,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9493389129638672,
      "eval_runtime": 56.0161,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1875
    },
    {
      "epoch": 0.17308268942451852,
      "grad_norm": 1.9003758440236473,
      "learning_rate": 1.1894743962246802e-06,
      "loss": 0.8281,
      "step": 1876
    },
    {
      "epoch": 0.17317495098604543,
      "grad_norm": 2.4667028650650504,
      "learning_rate": 1.189393598262278e-06,
      "loss": 0.8671,
      "step": 1877
    },
    {
      "epoch": 0.17326721254757238,
      "grad_norm": 1.9819340573548156,
      "learning_rate": 1.1893127491536397e-06,
      "loss": 0.5571,
      "step": 1878
    },
    {
      "epoch": 0.1733594741090993,
      "grad_norm": 1.9869611101078306,
      "learning_rate": 1.1892318489060918e-06,
      "loss": 0.8948,
      "step": 1879
    },
    {
      "epoch": 0.17345173567062622,
      "grad_norm": 2.054294013985498,
      "learning_rate": 1.1891508975269655e-06,
      "loss": 0.7908,
      "step": 1880
    },
    {
      "epoch": 0.17345173567062622,
      "eval_GEN Loss": 0.4685158133506775,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8862275449101796,
      "eval_PRM F1 AUC": 0.7283918281822945,
      "eval_PRM F1 Neg": 0.5777777777777777,
      "eval_PRM Loss": 0.3908511698246002,
      "eval_PRM NPV": 0.5909090909090909,
      "eval_PRM Precision": 0.8809523809523809,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.5652173913043478,
      "eval_loss": 0.951442301273346,
      "eval_runtime": 56.7739,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 1880
    },
    {
      "epoch": 0.17354399723215316,
      "grad_norm": 1.6717171649021387,
      "learning_rate": 1.189069895023597e-06,
      "loss": 0.7849,
      "step": 1881
    },
    {
      "epoch": 0.17363625879368008,
      "grad_norm": 1.7825294071368827,
      "learning_rate": 1.1889888414033268e-06,
      "loss": 0.791,
      "step": 1882
    },
    {
      "epoch": 0.173728520355207,
      "grad_norm": 2.3399100576380185,
      "learning_rate": 1.1889077366735e-06,
      "loss": 0.8288,
      "step": 1883
    },
    {
      "epoch": 0.17382078191673395,
      "grad_norm": 1.4659728050258447,
      "learning_rate": 1.1888265808414667e-06,
      "loss": 0.7501,
      "step": 1884
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 1.55020720573834,
      "learning_rate": 1.188745373914581e-06,
      "loss": 0.7398,
      "step": 1885
    },
    {
      "epoch": 0.17391304347826086,
      "eval_GEN Loss": 0.4690574109554291,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8823529411764706,
      "eval_PRM F1 AUC": 0.6909376636982714,
      "eval_PRM F1 Neg": 0.5238095238095238,
      "eval_PRM Loss": 0.3869512975215912,
      "eval_PRM NPV": 0.5789473684210527,
      "eval_PRM Precision": 0.8620689655172413,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9814903736114502,
      "eval_runtime": 55.822,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1885
    },
    {
      "epoch": 0.1740053050397878,
      "grad_norm": 2.3344400330851114,
      "learning_rate": 1.1886641159002022e-06,
      "loss": 0.7529,
      "step": 1886
    },
    {
      "epoch": 0.17409756660131473,
      "grad_norm": 1.982474628280561,
      "learning_rate": 1.1885828068056937e-06,
      "loss": 0.7663,
      "step": 1887
    },
    {
      "epoch": 0.17418982816284165,
      "grad_norm": 2.180397432523735,
      "learning_rate": 1.1885014466384245e-06,
      "loss": 0.655,
      "step": 1888
    },
    {
      "epoch": 0.1742820897243686,
      "grad_norm": 3.8786558250419008,
      "learning_rate": 1.1884200354057672e-06,
      "loss": 0.9249,
      "step": 1889
    },
    {
      "epoch": 0.1743743512858955,
      "grad_norm": 1.4128292216687506,
      "learning_rate": 1.1883385731150995e-06,
      "loss": 0.794,
      "step": 1890
    },
    {
      "epoch": 0.1743743512858955,
      "eval_GEN Loss": 0.46872907876968384,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8823529411764706,
      "eval_PRM F1 AUC": 0.6909376636982714,
      "eval_PRM F1 Neg": 0.5238095238095238,
      "eval_PRM Loss": 0.3835473656654358,
      "eval_PRM NPV": 0.5789473684210527,
      "eval_PRM Precision": 0.8620689655172413,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.4782608695652174,
      "eval_loss": 0.9583533406257629,
      "eval_runtime": 56.173,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 1890
    },
    {
      "epoch": 0.17446661284742243,
      "grad_norm": 1.7512735734999927,
      "learning_rate": 1.1882570597738032e-06,
      "loss": 0.8348,
      "step": 1891
    },
    {
      "epoch": 0.17455887440894938,
      "grad_norm": 2.5214015504072105,
      "learning_rate": 1.1881754953892661e-06,
      "loss": 0.7224,
      "step": 1892
    },
    {
      "epoch": 0.1746511359704763,
      "grad_norm": 1.6795900667309693,
      "learning_rate": 1.1880938799688787e-06,
      "loss": 0.8449,
      "step": 1893
    },
    {
      "epoch": 0.17474339753200324,
      "grad_norm": 1.7118345168670674,
      "learning_rate": 1.188012213520038e-06,
      "loss": 0.7201,
      "step": 1894
    },
    {
      "epoch": 0.17483565909353016,
      "grad_norm": 1.584372664384906,
      "learning_rate": 1.1879304960501441e-06,
      "loss": 0.7297,
      "step": 1895
    },
    {
      "epoch": 0.17483565909353016,
      "eval_GEN Loss": 0.4695640206336975,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3987809717655182,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.918870210647583,
      "eval_runtime": 55.7564,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 1895
    },
    {
      "epoch": 0.17492792065505708,
      "grad_norm": 2.6131805911009676,
      "learning_rate": 1.1878487275666025e-06,
      "loss": 0.7928,
      "step": 1896
    },
    {
      "epoch": 0.17502018221658402,
      "grad_norm": 1.5966418591815272,
      "learning_rate": 1.1877669080768235e-06,
      "loss": 0.7113,
      "step": 1897
    },
    {
      "epoch": 0.17511244377811094,
      "grad_norm": 1.6367611060894423,
      "learning_rate": 1.1876850375882213e-06,
      "loss": 0.6772,
      "step": 1898
    },
    {
      "epoch": 0.17520470533963786,
      "grad_norm": 2.0825135677424855,
      "learning_rate": 1.1876031161082154e-06,
      "loss": 0.8016,
      "step": 1899
    },
    {
      "epoch": 0.1752969669011648,
      "grad_norm": 1.9678959563769876,
      "learning_rate": 1.1875211436442293e-06,
      "loss": 0.8319,
      "step": 1900
    },
    {
      "epoch": 0.1752969669011648,
      "eval_GEN Loss": 0.4692772924900055,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4279230535030365,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9149038195610046,
      "eval_runtime": 55.7671,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 1900
    },
    {
      "epoch": 0.17538922846269173,
      "grad_norm": 2.3252179355430083,
      "learning_rate": 1.187439120203692e-06,
      "loss": 0.5821,
      "step": 1901
    },
    {
      "epoch": 0.17548149002421867,
      "grad_norm": 2.9514616238302844,
      "learning_rate": 1.1873570457940364e-06,
      "loss": 0.8697,
      "step": 1902
    },
    {
      "epoch": 0.1755737515857456,
      "grad_norm": 1.8577708695461148,
      "learning_rate": 1.1872749204227e-06,
      "loss": 0.6237,
      "step": 1903
    },
    {
      "epoch": 0.1756660131472725,
      "grad_norm": 1.6169552802513891,
      "learning_rate": 1.1871927440971254e-06,
      "loss": 0.696,
      "step": 1904
    },
    {
      "epoch": 0.17575827470879946,
      "grad_norm": 2.739107800746727,
      "learning_rate": 1.1871105168247592e-06,
      "loss": 0.6307,
      "step": 1905
    },
    {
      "epoch": 0.17575827470879946,
      "eval_GEN Loss": 0.46751323342323303,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.38527750968933105,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8990384340286255,
      "eval_runtime": 56.2352,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 1905
    },
    {
      "epoch": 0.17585053627032637,
      "grad_norm": 1.683112032608916,
      "learning_rate": 1.1870282386130536e-06,
      "loss": 0.6837,
      "step": 1906
    },
    {
      "epoch": 0.1759427978318533,
      "grad_norm": 1.5973651948445948,
      "learning_rate": 1.186945909469464e-06,
      "loss": 0.6812,
      "step": 1907
    },
    {
      "epoch": 0.17603505939338024,
      "grad_norm": 2.3245973463415366,
      "learning_rate": 1.1868635294014517e-06,
      "loss": 0.9884,
      "step": 1908
    },
    {
      "epoch": 0.17612732095490716,
      "grad_norm": 1.817496038080407,
      "learning_rate": 1.186781098416482e-06,
      "loss": 0.7388,
      "step": 1909
    },
    {
      "epoch": 0.1762195825164341,
      "grad_norm": 1.3311852279017433,
      "learning_rate": 1.1866986165220248e-06,
      "loss": 0.6314,
      "step": 1910
    },
    {
      "epoch": 0.1762195825164341,
      "eval_GEN Loss": 0.4685840606689453,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.35463541746139526,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9057692289352417,
      "eval_runtime": 55.8633,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1910
    },
    {
      "epoch": 0.17631184407796102,
      "grad_norm": 1.7186034726360713,
      "learning_rate": 1.186616083725555e-06,
      "loss": 0.7453,
      "step": 1911
    },
    {
      "epoch": 0.17640410563948794,
      "grad_norm": 2.224287522579521,
      "learning_rate": 1.1865335000345515e-06,
      "loss": 0.608,
      "step": 1912
    },
    {
      "epoch": 0.1764963672010149,
      "grad_norm": 1.7476775322804345,
      "learning_rate": 1.1864508654564986e-06,
      "loss": 0.5619,
      "step": 1913
    },
    {
      "epoch": 0.1765886287625418,
      "grad_norm": 2.0527580222396797,
      "learning_rate": 1.1863681799988844e-06,
      "loss": 0.7012,
      "step": 1914
    },
    {
      "epoch": 0.17668089032406872,
      "grad_norm": 2.3313585606950946,
      "learning_rate": 1.186285443669202e-06,
      "loss": 0.8966,
      "step": 1915
    },
    {
      "epoch": 0.17668089032406872,
      "eval_GEN Loss": 0.46925368905067444,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.34594297409057617,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9025240540504456,
      "eval_runtime": 57.0764,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 1915
    },
    {
      "epoch": 0.17677315188559567,
      "grad_norm": 2.535215320237209,
      "learning_rate": 1.1862026564749492e-06,
      "loss": 0.9057,
      "step": 1916
    },
    {
      "epoch": 0.1768654134471226,
      "grad_norm": 3.5084157619272656,
      "learning_rate": 1.1861198184236286e-06,
      "loss": 0.9853,
      "step": 1917
    },
    {
      "epoch": 0.17695767500864953,
      "grad_norm": 2.060703280233832,
      "learning_rate": 1.1860369295227468e-06,
      "loss": 0.8597,
      "step": 1918
    },
    {
      "epoch": 0.17704993657017645,
      "grad_norm": 1.5742432988107469,
      "learning_rate": 1.1859539897798152e-06,
      "loss": 0.8047,
      "step": 1919
    },
    {
      "epoch": 0.17714219813170337,
      "grad_norm": 1.7976768419993143,
      "learning_rate": 1.1858709992023501e-06,
      "loss": 0.7721,
      "step": 1920
    },
    {
      "epoch": 0.17714219813170337,
      "eval_GEN Loss": 0.4682477116584778,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3484565317630768,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8586538434028625,
      "eval_runtime": 55.9293,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1920
    },
    {
      "epoch": 0.17723445969323032,
      "grad_norm": 2.343535713364003,
      "learning_rate": 1.1857879577978724e-06,
      "loss": 0.5613,
      "step": 1921
    },
    {
      "epoch": 0.17732672125475724,
      "grad_norm": 1.757333883870705,
      "learning_rate": 1.185704865573907e-06,
      "loss": 0.7187,
      "step": 1922
    },
    {
      "epoch": 0.17741898281628415,
      "grad_norm": 2.0400670424123133,
      "learning_rate": 1.1856217225379842e-06,
      "loss": 0.8537,
      "step": 1923
    },
    {
      "epoch": 0.1775112443778111,
      "grad_norm": 1.7479042814278654,
      "learning_rate": 1.1855385286976387e-06,
      "loss": 0.7641,
      "step": 1924
    },
    {
      "epoch": 0.17760350593933802,
      "grad_norm": 2.417533290690098,
      "learning_rate": 1.1854552840604092e-06,
      "loss": 0.7124,
      "step": 1925
    },
    {
      "epoch": 0.17760350593933802,
      "eval_GEN Loss": 0.4683780074119568,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.3596678078174591,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.854687511920929,
      "eval_runtime": 56.0748,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1925
    },
    {
      "epoch": 0.17769576750086497,
      "grad_norm": 1.641541075462338,
      "learning_rate": 1.18537198863384e-06,
      "loss": 0.8352,
      "step": 1926
    },
    {
      "epoch": 0.17778802906239188,
      "grad_norm": 1.3910642781393008,
      "learning_rate": 1.1852886424254788e-06,
      "loss": 0.6602,
      "step": 1927
    },
    {
      "epoch": 0.1778802906239188,
      "grad_norm": 2.232958456514103,
      "learning_rate": 1.185205245442879e-06,
      "loss": 0.8616,
      "step": 1928
    },
    {
      "epoch": 0.17797255218544575,
      "grad_norm": 2.3839649191841006,
      "learning_rate": 1.185121797693598e-06,
      "loss": 0.631,
      "step": 1929
    },
    {
      "epoch": 0.17806481374697267,
      "grad_norm": 1.8444372479741982,
      "learning_rate": 1.1850382991851982e-06,
      "loss": 0.7821,
      "step": 1930
    },
    {
      "epoch": 0.17806481374697267,
      "eval_GEN Loss": 0.4683499038219452,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.3657971918582916,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8576322197914124,
      "eval_runtime": 55.8482,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1930
    },
    {
      "epoch": 0.17815707530849959,
      "grad_norm": 2.25169689693887,
      "learning_rate": 1.1849547499252462e-06,
      "loss": 0.9682,
      "step": 1931
    },
    {
      "epoch": 0.17824933687002653,
      "grad_norm": 1.947708892171765,
      "learning_rate": 1.1848711499213134e-06,
      "loss": 0.6699,
      "step": 1932
    },
    {
      "epoch": 0.17834159843155345,
      "grad_norm": 2.7578340026793176,
      "learning_rate": 1.1847874991809757e-06,
      "loss": 0.7486,
      "step": 1933
    },
    {
      "epoch": 0.1784338599930804,
      "grad_norm": 1.8144475086471232,
      "learning_rate": 1.1847037977118136e-06,
      "loss": 0.7841,
      "step": 1934
    },
    {
      "epoch": 0.17852612155460731,
      "grad_norm": 2.1749335875167395,
      "learning_rate": 1.1846200455214128e-06,
      "loss": 0.8312,
      "step": 1935
    },
    {
      "epoch": 0.17852612155460731,
      "eval_GEN Loss": 0.4700009822845459,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.3598850667476654,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8522836565971375,
      "eval_runtime": 56.0167,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1935
    },
    {
      "epoch": 0.17861838311613423,
      "grad_norm": 1.8202686727472561,
      "learning_rate": 1.184536242617362e-06,
      "loss": 0.7958,
      "step": 1936
    },
    {
      "epoch": 0.17871064467766118,
      "grad_norm": 2.3130305737826484,
      "learning_rate": 1.1844523890072566e-06,
      "loss": 0.7722,
      "step": 1937
    },
    {
      "epoch": 0.1788029062391881,
      "grad_norm": 1.9812235010674377,
      "learning_rate": 1.1843684846986949e-06,
      "loss": 0.7294,
      "step": 1938
    },
    {
      "epoch": 0.17889516780071502,
      "grad_norm": 3.1377665851712013,
      "learning_rate": 1.1842845296992809e-06,
      "loss": 0.902,
      "step": 1939
    },
    {
      "epoch": 0.17898742936224196,
      "grad_norm": 2.0456290093446277,
      "learning_rate": 1.1842005240166222e-06,
      "loss": 0.8287,
      "step": 1940
    },
    {
      "epoch": 0.17898742936224196,
      "eval_GEN Loss": 0.46932679414749146,
      "eval_GEN top-5 accuracy": 0.9850942569048663,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9202453987730062,
      "eval_PRM F1 AUC": 0.8431115767417497,
      "eval_PRM F1 Neg": 0.7346938775510204,
      "eval_PRM Loss": 0.34539997577667236,
      "eval_PRM NPV": 0.6923076923076923,
      "eval_PRM Precision": 0.9375,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.844531238079071,
      "eval_runtime": 55.9681,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1940
    },
    {
      "epoch": 0.17907969092376888,
      "grad_norm": 1.7501401490304878,
      "learning_rate": 1.184116467658332e-06,
      "loss": 0.598,
      "step": 1941
    },
    {
      "epoch": 0.17917195248529583,
      "grad_norm": 2.751844920732577,
      "learning_rate": 1.1840323606320273e-06,
      "loss": 0.8756,
      "step": 1942
    },
    {
      "epoch": 0.17926421404682275,
      "grad_norm": 1.7486262039835792,
      "learning_rate": 1.1839482029453302e-06,
      "loss": 0.6555,
      "step": 1943
    },
    {
      "epoch": 0.17935647560834966,
      "grad_norm": 3.3452249547582564,
      "learning_rate": 1.1838639946058673e-06,
      "loss": 0.8685,
      "step": 1944
    },
    {
      "epoch": 0.1794487371698766,
      "grad_norm": 1.8288374287804032,
      "learning_rate": 1.1837797356212693e-06,
      "loss": 0.8282,
      "step": 1945
    },
    {
      "epoch": 0.1794487371698766,
      "eval_GEN Loss": 0.4705554246902466,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.328626811504364,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8490384817123413,
      "eval_runtime": 56.3921,
      "eval_samples_per_second": 1.153,
      "eval_steps_per_second": 0.089,
      "step": 1945
    },
    {
      "epoch": 0.17954099873140353,
      "grad_norm": 1.943313902167551,
      "learning_rate": 1.1836954259991724e-06,
      "loss": 0.917,
      "step": 1946
    },
    {
      "epoch": 0.17963326029293045,
      "grad_norm": 1.5216936319454786,
      "learning_rate": 1.1836110657472162e-06,
      "loss": 0.6325,
      "step": 1947
    },
    {
      "epoch": 0.1797255218544574,
      "grad_norm": 2.5124030329025335,
      "learning_rate": 1.1835266548730465e-06,
      "loss": 0.7952,
      "step": 1948
    },
    {
      "epoch": 0.1798177834159843,
      "grad_norm": 1.6176394503610152,
      "learning_rate": 1.1834421933843118e-06,
      "loss": 0.7563,
      "step": 1949
    },
    {
      "epoch": 0.17991004497751126,
      "grad_norm": 2.4086105514681106,
      "learning_rate": 1.1833576812886666e-06,
      "loss": 0.847,
      "step": 1950
    },
    {
      "epoch": 0.17991004497751126,
      "eval_GEN Loss": 0.46942463517189026,
      "eval_GEN top-5 accuracy": 0.9850942569048663,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.3282986879348755,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8774038553237915,
      "eval_runtime": 56.8631,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1950
    },
    {
      "epoch": 0.18000230653903818,
      "grad_norm": 4.602223665033289,
      "learning_rate": 1.1832731185937695e-06,
      "loss": 0.8888,
      "step": 1951
    },
    {
      "epoch": 0.1800945681005651,
      "grad_norm": 1.8659980358360455,
      "learning_rate": 1.1831885053072837e-06,
      "loss": 0.7075,
      "step": 1952
    },
    {
      "epoch": 0.18018682966209204,
      "grad_norm": 1.8337120619647553,
      "learning_rate": 1.183103841436877e-06,
      "loss": 0.7455,
      "step": 1953
    },
    {
      "epoch": 0.18027909122361896,
      "grad_norm": 1.8044581737282863,
      "learning_rate": 1.1830191269902216e-06,
      "loss": 0.8378,
      "step": 1954
    },
    {
      "epoch": 0.18037135278514588,
      "grad_norm": 1.843116592608778,
      "learning_rate": 1.1829343619749944e-06,
      "loss": 0.7311,
      "step": 1955
    },
    {
      "epoch": 0.18037135278514588,
      "eval_GEN Loss": 0.4702025353908539,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.33778148889541626,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8728365302085876,
      "eval_runtime": 56.9483,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 1955
    },
    {
      "epoch": 0.18046361434667282,
      "grad_norm": 1.5054335447800735,
      "learning_rate": 1.1828495463988774e-06,
      "loss": 0.6745,
      "step": 1956
    },
    {
      "epoch": 0.18055587590819974,
      "grad_norm": 1.920825500204467,
      "learning_rate": 1.1827646802695563e-06,
      "loss": 0.8852,
      "step": 1957
    },
    {
      "epoch": 0.1806481374697267,
      "grad_norm": 2.0527177750313452,
      "learning_rate": 1.182679763594722e-06,
      "loss": 0.9191,
      "step": 1958
    },
    {
      "epoch": 0.1807403990312536,
      "grad_norm": 1.6131473811718557,
      "learning_rate": 1.1825947963820698e-06,
      "loss": 0.7022,
      "step": 1959
    },
    {
      "epoch": 0.18083266059278053,
      "grad_norm": 1.5671800926925206,
      "learning_rate": 1.1825097786392994e-06,
      "loss": 0.6837,
      "step": 1960
    },
    {
      "epoch": 0.18083266059278053,
      "eval_GEN Loss": 0.4705111086368561,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9156626506024096,
      "eval_PRM F1 AUC": 0.8056574122577265,
      "eval_PRM F1 Neg": 0.6956521739130435,
      "eval_PRM Loss": 0.34896120429039,
      "eval_PRM NPV": 0.6956521739130435,
      "eval_PRM Precision": 0.9156626506024096,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8716346025466919,
      "eval_runtime": 56.8601,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 1960
    },
    {
      "epoch": 0.18092492215430747,
      "grad_norm": 2.79084900873371,
      "learning_rate": 1.1824247103741154e-06,
      "loss": 0.7971,
      "step": 1961
    },
    {
      "epoch": 0.1810171837158344,
      "grad_norm": 1.9170428272343674,
      "learning_rate": 1.1823395915942265e-06,
      "loss": 0.7246,
      "step": 1962
    },
    {
      "epoch": 0.1811094452773613,
      "grad_norm": 1.9462537574889882,
      "learning_rate": 1.1822544223073467e-06,
      "loss": 0.8289,
      "step": 1963
    },
    {
      "epoch": 0.18120170683888825,
      "grad_norm": 1.9375282632531803,
      "learning_rate": 1.1821692025211942e-06,
      "loss": 0.76,
      "step": 1964
    },
    {
      "epoch": 0.18129396840041517,
      "grad_norm": 1.3435109446325704,
      "learning_rate": 1.1820839322434913e-06,
      "loss": 0.6949,
      "step": 1965
    },
    {
      "epoch": 0.18129396840041517,
      "eval_GEN Loss": 0.4705446660518646,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.37570104002952576,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8603365421295166,
      "eval_runtime": 56.0616,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 1965
    },
    {
      "epoch": 0.18138622996194212,
      "grad_norm": 2.409717875745877,
      "learning_rate": 1.1819986114819656e-06,
      "loss": 0.8287,
      "step": 1966
    },
    {
      "epoch": 0.18147849152346904,
      "grad_norm": 2.018430092064614,
      "learning_rate": 1.1819132402443491e-06,
      "loss": 0.8305,
      "step": 1967
    },
    {
      "epoch": 0.18157075308499596,
      "grad_norm": 1.6910998307111729,
      "learning_rate": 1.1818278185383783e-06,
      "loss": 0.774,
      "step": 1968
    },
    {
      "epoch": 0.1816630146465229,
      "grad_norm": 2.1156338275691637,
      "learning_rate": 1.181742346371794e-06,
      "loss": 0.632,
      "step": 1969
    },
    {
      "epoch": 0.18175527620804982,
      "grad_norm": 1.8875932222931249,
      "learning_rate": 1.1816568237523419e-06,
      "loss": 0.668,
      "step": 1970
    },
    {
      "epoch": 0.18175527620804982,
      "eval_GEN Loss": 0.4716153144836426,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.39824846386909485,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8655648827552795,
      "eval_runtime": 55.9045,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1970
    },
    {
      "epoch": 0.18184753776957674,
      "grad_norm": 1.8218580804105595,
      "learning_rate": 1.1815712506877725e-06,
      "loss": 0.6293,
      "step": 1971
    },
    {
      "epoch": 0.18193979933110369,
      "grad_norm": 2.323150747517259,
      "learning_rate": 1.18148562718584e-06,
      "loss": 0.7472,
      "step": 1972
    },
    {
      "epoch": 0.1820320608926306,
      "grad_norm": 2.2830565197763906,
      "learning_rate": 1.181399953254304e-06,
      "loss": 0.6765,
      "step": 1973
    },
    {
      "epoch": 0.18212432245415755,
      "grad_norm": 2.81192357035052,
      "learning_rate": 1.1813142289009286e-06,
      "loss": 0.809,
      "step": 1974
    },
    {
      "epoch": 0.18221658401568447,
      "grad_norm": 1.9730415783176882,
      "learning_rate": 1.1812284541334821e-06,
      "loss": 0.8976,
      "step": 1975
    },
    {
      "epoch": 0.18221658401568447,
      "eval_GEN Loss": 0.4716103971004486,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3583613634109497,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8695913553237915,
      "eval_runtime": 55.9433,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 1975
    },
    {
      "epoch": 0.1823088455772114,
      "grad_norm": 2.8985150365969434,
      "learning_rate": 1.1811426289597377e-06,
      "loss": 0.8941,
      "step": 1976
    },
    {
      "epoch": 0.18240110713873833,
      "grad_norm": 2.1626277533488376,
      "learning_rate": 1.1810567533874728e-06,
      "loss": 0.7578,
      "step": 1977
    },
    {
      "epoch": 0.18249336870026525,
      "grad_norm": 2.357520483642952,
      "learning_rate": 1.1809708274244695e-06,
      "loss": 0.8202,
      "step": 1978
    },
    {
      "epoch": 0.18258563026179217,
      "grad_norm": 3.1136973651999953,
      "learning_rate": 1.1808848510785149e-06,
      "loss": 1.128,
      "step": 1979
    },
    {
      "epoch": 0.18267789182331912,
      "grad_norm": 2.472151221670113,
      "learning_rate": 1.1807988243574e-06,
      "loss": 0.9272,
      "step": 1980
    },
    {
      "epoch": 0.18267789182331912,
      "eval_GEN Loss": 0.4723505675792694,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3514373302459717,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8723557591438293,
      "eval_runtime": 55.8327,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 1980
    },
    {
      "epoch": 0.18277015338484603,
      "grad_norm": 1.7355618460112476,
      "learning_rate": 1.1807127472689208e-06,
      "loss": 0.6718,
      "step": 1981
    },
    {
      "epoch": 0.18286241494637298,
      "grad_norm": 1.7950478995183463,
      "learning_rate": 1.180626619820878e-06,
      "loss": 0.6722,
      "step": 1982
    },
    {
      "epoch": 0.1829546765078999,
      "grad_norm": 2.4822181820471343,
      "learning_rate": 1.180540442021076e-06,
      "loss": 0.8901,
      "step": 1983
    },
    {
      "epoch": 0.18304693806942682,
      "grad_norm": 1.6620784138258664,
      "learning_rate": 1.1804542138773247e-06,
      "loss": 0.7279,
      "step": 1984
    },
    {
      "epoch": 0.18313919963095376,
      "grad_norm": 1.669328072447358,
      "learning_rate": 1.1803679353974384e-06,
      "loss": 0.7729,
      "step": 1985
    },
    {
      "epoch": 0.18313919963095376,
      "eval_GEN Loss": 0.47229859232902527,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.35772424936294556,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8501201868057251,
      "eval_runtime": 56.0208,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 1985
    },
    {
      "epoch": 0.18323146119248068,
      "grad_norm": 1.334830857646573,
      "learning_rate": 1.1802816065892356e-06,
      "loss": 0.6131,
      "step": 1986
    },
    {
      "epoch": 0.1833237227540076,
      "grad_norm": 2.405568253699568,
      "learning_rate": 1.1801952274605397e-06,
      "loss": 0.7406,
      "step": 1987
    },
    {
      "epoch": 0.18341598431553455,
      "grad_norm": 2.1917242170043476,
      "learning_rate": 1.1801087980191784e-06,
      "loss": 0.7511,
      "step": 1988
    },
    {
      "epoch": 0.18350824587706147,
      "grad_norm": 2.554704922098671,
      "learning_rate": 1.1800223182729839e-06,
      "loss": 0.896,
      "step": 1989
    },
    {
      "epoch": 0.1836005074385884,
      "grad_norm": 2.3679975677847698,
      "learning_rate": 1.1799357882297934e-06,
      "loss": 0.9535,
      "step": 1990
    },
    {
      "epoch": 0.1836005074385884,
      "eval_GEN Loss": 0.47487273812294006,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8930817610062893,
      "eval_PRM F1 AUC": 0.819015191199581,
      "eval_PRM F1 Neg": 0.6792452830188679,
      "eval_PRM Loss": 0.3714655339717865,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9342105263157895,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.845973551273346,
      "eval_runtime": 55.9976,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 1990
    },
    {
      "epoch": 0.18369276900011533,
      "grad_norm": 1.6180444080504108,
      "learning_rate": 1.1798492078974484e-06,
      "loss": 0.6486,
      "step": 1991
    },
    {
      "epoch": 0.18378503056164225,
      "grad_norm": 1.683006147370453,
      "learning_rate": 1.1797625772837946e-06,
      "loss": 0.859,
      "step": 1992
    },
    {
      "epoch": 0.1838772921231692,
      "grad_norm": 2.2510671580043202,
      "learning_rate": 1.1796758963966828e-06,
      "loss": 0.7019,
      "step": 1993
    },
    {
      "epoch": 0.1839695536846961,
      "grad_norm": 1.8567289016022117,
      "learning_rate": 1.1795891652439684e-06,
      "loss": 0.8697,
      "step": 1994
    },
    {
      "epoch": 0.18406181524622303,
      "grad_norm": 2.422870701730181,
      "learning_rate": 1.1795023838335107e-06,
      "loss": 0.7371,
      "step": 1995
    },
    {
      "epoch": 0.18406181524622303,
      "eval_GEN Loss": 0.4758482873439789,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM Loss": 0.3763881325721741,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8481971025466919,
      "eval_runtime": 55.8838,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 1995
    },
    {
      "epoch": 0.18415407680774998,
      "grad_norm": 1.8054001363809202,
      "learning_rate": 1.1794155521731744e-06,
      "loss": 0.6933,
      "step": 1996
    },
    {
      "epoch": 0.1842463383692769,
      "grad_norm": 1.5414872633375913,
      "learning_rate": 1.1793286702708279e-06,
      "loss": 0.8097,
      "step": 1997
    },
    {
      "epoch": 0.18433859993080384,
      "grad_norm": 2.322123260692024,
      "learning_rate": 1.1792417381343448e-06,
      "loss": 0.9594,
      "step": 1998
    },
    {
      "epoch": 0.18443086149233076,
      "grad_norm": 2.403147298475866,
      "learning_rate": 1.1791547557716031e-06,
      "loss": 0.6702,
      "step": 1999
    },
    {
      "epoch": 0.18452312305385768,
      "grad_norm": 1.8937608761817435,
      "learning_rate": 1.1790677231904852e-06,
      "loss": 0.7323,
      "step": 2000
    },
    {
      "epoch": 0.18452312305385768,
      "eval_GEN Loss": 0.47482290863990784,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9135802469135802,
      "eval_PRM F1 AUC": 0.8370874803562075,
      "eval_PRM F1 Neg": 0.72,
      "eval_PRM Loss": 0.34736916422843933,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9367088607594937,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8415865302085876,
      "eval_runtime": 55.9308,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2000
    },
    {
      "epoch": 0.18461538461538463,
      "grad_norm": 1.7218811898237494,
      "learning_rate": 1.1789806403988781e-06,
      "loss": 0.6016,
      "step": 2001
    },
    {
      "epoch": 0.18470764617691154,
      "grad_norm": 1.6092138907105,
      "learning_rate": 1.1788935074046735e-06,
      "loss": 0.8096,
      "step": 2002
    },
    {
      "epoch": 0.18479990773843846,
      "grad_norm": 1.5971983932629983,
      "learning_rate": 1.1788063242157671e-06,
      "loss": 0.7057,
      "step": 2003
    },
    {
      "epoch": 0.1848921692999654,
      "grad_norm": 1.3621746398913959,
      "learning_rate": 1.1787190908400602e-06,
      "loss": 0.5874,
      "step": 2004
    },
    {
      "epoch": 0.18498443086149233,
      "grad_norm": 3.8139246079298226,
      "learning_rate": 1.1786318072854574e-06,
      "loss": 0.8478,
      "step": 2005
    },
    {
      "epoch": 0.18498443086149233,
      "eval_GEN Loss": 0.47527387738227844,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.33743923902511597,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8596153855323792,
      "eval_runtime": 56.9064,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2005
    },
    {
      "epoch": 0.18507669242301927,
      "grad_norm": 2.781499716829389,
      "learning_rate": 1.1785444735598692e-06,
      "loss": 0.8955,
      "step": 2006
    },
    {
      "epoch": 0.1851689539845462,
      "grad_norm": 2.206461851230783,
      "learning_rate": 1.1784570896712091e-06,
      "loss": 0.861,
      "step": 2007
    },
    {
      "epoch": 0.1852612155460731,
      "grad_norm": 1.6405028562717892,
      "learning_rate": 1.1783696556273966e-06,
      "loss": 0.7969,
      "step": 2008
    },
    {
      "epoch": 0.18535347710760006,
      "grad_norm": 1.937622062292657,
      "learning_rate": 1.1782821714363548e-06,
      "loss": 0.7386,
      "step": 2009
    },
    {
      "epoch": 0.18544573866912697,
      "grad_norm": 1.871704342430374,
      "learning_rate": 1.178194637106012e-06,
      "loss": 0.7642,
      "step": 2010
    },
    {
      "epoch": 0.18544573866912697,
      "eval_GEN Loss": 0.4762713313102722,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3472242057323456,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8661057949066162,
      "eval_runtime": 56.8425,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2010
    },
    {
      "epoch": 0.1855380002306539,
      "grad_norm": 1.300223142727639,
      "learning_rate": 1.1781070526443002e-06,
      "loss": 0.6246,
      "step": 2011
    },
    {
      "epoch": 0.18563026179218084,
      "grad_norm": 1.4839254987161208,
      "learning_rate": 1.1780194180591567e-06,
      "loss": 0.712,
      "step": 2012
    },
    {
      "epoch": 0.18572252335370776,
      "grad_norm": 1.3638374277079743,
      "learning_rate": 1.1779317333585232e-06,
      "loss": 0.5256,
      "step": 2013
    },
    {
      "epoch": 0.1858147849152347,
      "grad_norm": 3.1800809335198794,
      "learning_rate": 1.1778439985503456e-06,
      "loss": 0.781,
      "step": 2014
    },
    {
      "epoch": 0.18590704647676162,
      "grad_norm": 2.8899657971448702,
      "learning_rate": 1.1777562136425747e-06,
      "loss": 1.0692,
      "step": 2015
    },
    {
      "epoch": 0.18590704647676162,
      "eval_GEN Loss": 0.4769386053085327,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.35975176095962524,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8862980604171753,
      "eval_runtime": 56.8541,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2015
    },
    {
      "epoch": 0.18599930803828854,
      "grad_norm": 2.145923227050257,
      "learning_rate": 1.1776683786431655e-06,
      "loss": 0.7344,
      "step": 2016
    },
    {
      "epoch": 0.1860915695998155,
      "grad_norm": 1.5044043557834834,
      "learning_rate": 1.1775804935600781e-06,
      "loss": 0.7167,
      "step": 2017
    },
    {
      "epoch": 0.1861838311613424,
      "grad_norm": 1.742271416763045,
      "learning_rate": 1.1774925584012766e-06,
      "loss": 0.7182,
      "step": 2018
    },
    {
      "epoch": 0.18627609272286932,
      "grad_norm": 2.3624915019513644,
      "learning_rate": 1.1774045731747299e-06,
      "loss": 0.8248,
      "step": 2019
    },
    {
      "epoch": 0.18636835428439627,
      "grad_norm": 2.1935117177206647,
      "learning_rate": 1.177316537888411e-06,
      "loss": 0.7486,
      "step": 2020
    },
    {
      "epoch": 0.18636835428439627,
      "eval_GEN Loss": 0.4771266579627991,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3712487518787384,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9055288434028625,
      "eval_runtime": 57.2909,
      "eval_samples_per_second": 1.135,
      "eval_steps_per_second": 0.087,
      "step": 2020
    },
    {
      "epoch": 0.1864606158459232,
      "grad_norm": 1.7514578539412164,
      "learning_rate": 1.1772284525502983e-06,
      "loss": 0.7124,
      "step": 2021
    },
    {
      "epoch": 0.1865528774074501,
      "grad_norm": 1.6268295502484904,
      "learning_rate": 1.177140317168374e-06,
      "loss": 0.6699,
      "step": 2022
    },
    {
      "epoch": 0.18664513896897705,
      "grad_norm": 1.7886518441477073,
      "learning_rate": 1.177052131750625e-06,
      "loss": 0.6993,
      "step": 2023
    },
    {
      "epoch": 0.18673740053050397,
      "grad_norm": 5.2219209888016325,
      "learning_rate": 1.1769638963050428e-06,
      "loss": 1.0932,
      "step": 2024
    },
    {
      "epoch": 0.18682966209203092,
      "grad_norm": 2.1651132863048232,
      "learning_rate": 1.1768756108396237e-06,
      "loss": 0.7829,
      "step": 2025
    },
    {
      "epoch": 0.18682966209203092,
      "eval_GEN Loss": 0.4762979745864868,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.37799975275993347,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9197115302085876,
      "eval_runtime": 56.8767,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2025
    },
    {
      "epoch": 0.18692192365355784,
      "grad_norm": 2.4120105181234504,
      "learning_rate": 1.176787275362368e-06,
      "loss": 0.9647,
      "step": 2026
    },
    {
      "epoch": 0.18701418521508476,
      "grad_norm": 1.4392912028263638,
      "learning_rate": 1.1766988898812808e-06,
      "loss": 0.7379,
      "step": 2027
    },
    {
      "epoch": 0.1871064467766117,
      "grad_norm": 1.59708899434638,
      "learning_rate": 1.176610454404372e-06,
      "loss": 0.7667,
      "step": 2028
    },
    {
      "epoch": 0.18719870833813862,
      "grad_norm": 1.3651238541526256,
      "learning_rate": 1.1765219689396553e-06,
      "loss": 0.6794,
      "step": 2029
    },
    {
      "epoch": 0.18729096989966554,
      "grad_norm": 2.079074025639708,
      "learning_rate": 1.17643343349515e-06,
      "loss": 0.8413,
      "step": 2030
    },
    {
      "epoch": 0.18729096989966554,
      "eval_GEN Loss": 0.47625789046287537,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3893580138683319,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9141826629638672,
      "eval_runtime": 56.9139,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2030
    },
    {
      "epoch": 0.18738323146119248,
      "grad_norm": 1.3647385935413638,
      "learning_rate": 1.1763448480788786e-06,
      "loss": 0.6232,
      "step": 2031
    },
    {
      "epoch": 0.1874754930227194,
      "grad_norm": 3.510711336384845,
      "learning_rate": 1.1762562126988696e-06,
      "loss": 1.1275,
      "step": 2032
    },
    {
      "epoch": 0.18756775458424635,
      "grad_norm": 1.7667114211392427,
      "learning_rate": 1.176167527363155e-06,
      "loss": 0.7605,
      "step": 2033
    },
    {
      "epoch": 0.18766001614577327,
      "grad_norm": 1.5509237422154474,
      "learning_rate": 1.176078792079771e-06,
      "loss": 0.7539,
      "step": 2034
    },
    {
      "epoch": 0.18775227770730019,
      "grad_norm": 1.929023562753418,
      "learning_rate": 1.1759900068567599e-06,
      "loss": 0.8275,
      "step": 2035
    },
    {
      "epoch": 0.18775227770730019,
      "eval_GEN Loss": 0.47494471073150635,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.40141040086746216,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9050480723381042,
      "eval_runtime": 55.9735,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2035
    },
    {
      "epoch": 0.18784453926882713,
      "grad_norm": 1.5786079669231259,
      "learning_rate": 1.1759011717021668e-06,
      "loss": 0.7178,
      "step": 2036
    },
    {
      "epoch": 0.18793680083035405,
      "grad_norm": 3.363351897038916,
      "learning_rate": 1.1758122866240425e-06,
      "loss": 0.9335,
      "step": 2037
    },
    {
      "epoch": 0.18802906239188097,
      "grad_norm": 1.7553134317965489,
      "learning_rate": 1.1757233516304419e-06,
      "loss": 0.7692,
      "step": 2038
    },
    {
      "epoch": 0.18812132395340792,
      "grad_norm": 1.5074484864053896,
      "learning_rate": 1.1756343667294243e-06,
      "loss": 0.841,
      "step": 2039
    },
    {
      "epoch": 0.18821358551493483,
      "grad_norm": 1.9001341763009556,
      "learning_rate": 1.1755453319290535e-06,
      "loss": 0.6041,
      "step": 2040
    },
    {
      "epoch": 0.18821358551493483,
      "eval_GEN Loss": 0.47478625178337097,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3914642632007599,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9007812738418579,
      "eval_runtime": 56.0471,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2040
    },
    {
      "epoch": 0.18830584707646178,
      "grad_norm": 1.887618824502614,
      "learning_rate": 1.1754562472373984e-06,
      "loss": 0.61,
      "step": 2041
    },
    {
      "epoch": 0.1883981086379887,
      "grad_norm": 2.537116970872114,
      "learning_rate": 1.1753671126625315e-06,
      "loss": 0.9013,
      "step": 2042
    },
    {
      "epoch": 0.18849037019951562,
      "grad_norm": 1.749885410614083,
      "learning_rate": 1.1752779282125307e-06,
      "loss": 0.7976,
      "step": 2043
    },
    {
      "epoch": 0.18858263176104256,
      "grad_norm": 1.7703292830260156,
      "learning_rate": 1.1751886938954779e-06,
      "loss": 0.691,
      "step": 2044
    },
    {
      "epoch": 0.18867489332256948,
      "grad_norm": 1.9204534035808125,
      "learning_rate": 1.1750994097194595e-06,
      "loss": 0.6879,
      "step": 2045
    },
    {
      "epoch": 0.18867489332256948,
      "eval_GEN Loss": 0.4729349613189697,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.37438541650772095,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8925480842590332,
      "eval_runtime": 55.9191,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2045
    },
    {
      "epoch": 0.1887671548840964,
      "grad_norm": 1.5905092183160414,
      "learning_rate": 1.175010075692567e-06,
      "loss": 0.7958,
      "step": 2046
    },
    {
      "epoch": 0.18885941644562335,
      "grad_norm": 2.599172986193549,
      "learning_rate": 1.1749206918228956e-06,
      "loss": 1.0229,
      "step": 2047
    },
    {
      "epoch": 0.18895167800715026,
      "grad_norm": 1.5407394588399215,
      "learning_rate": 1.1748312581185453e-06,
      "loss": 0.6175,
      "step": 2048
    },
    {
      "epoch": 0.1890439395686772,
      "grad_norm": 1.7075539827400186,
      "learning_rate": 1.1747417745876211e-06,
      "loss": 0.7414,
      "step": 2049
    },
    {
      "epoch": 0.18913620113020413,
      "grad_norm": 1.7729999806091692,
      "learning_rate": 1.174652241238232e-06,
      "loss": 0.8014,
      "step": 2050
    },
    {
      "epoch": 0.18913620113020413,
      "eval_GEN Loss": 0.4732709228992462,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3624724745750427,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.893750011920929,
      "eval_runtime": 56.1189,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 2050
    },
    {
      "epoch": 0.18922846269173105,
      "grad_norm": 1.3814801364156035,
      "learning_rate": 1.1745626580784916e-06,
      "loss": 0.7294,
      "step": 2051
    },
    {
      "epoch": 0.189320724253258,
      "grad_norm": 2.742469468717978,
      "learning_rate": 1.174473025116518e-06,
      "loss": 0.9501,
      "step": 2052
    },
    {
      "epoch": 0.1894129858147849,
      "grad_norm": 1.899074295936426,
      "learning_rate": 1.1743833423604344e-06,
      "loss": 0.6854,
      "step": 2053
    },
    {
      "epoch": 0.18950524737631183,
      "grad_norm": 1.805086519608829,
      "learning_rate": 1.1742936098183672e-06,
      "loss": 0.6486,
      "step": 2054
    },
    {
      "epoch": 0.18959750893783878,
      "grad_norm": 2.569618963834657,
      "learning_rate": 1.1742038274984485e-06,
      "loss": 0.7753,
      "step": 2055
    },
    {
      "epoch": 0.18959750893783878,
      "eval_GEN Loss": 0.47299501299858093,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.36366239190101624,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8921874761581421,
      "eval_runtime": 56.0044,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2055
    },
    {
      "epoch": 0.1896897704993657,
      "grad_norm": 2.914152850560168,
      "learning_rate": 1.1741139954088147e-06,
      "loss": 0.8671,
      "step": 2056
    },
    {
      "epoch": 0.18978203206089264,
      "grad_norm": 1.4479355516929415,
      "learning_rate": 1.174024113557606e-06,
      "loss": 0.7495,
      "step": 2057
    },
    {
      "epoch": 0.18987429362241956,
      "grad_norm": 1.3999078074787001,
      "learning_rate": 1.1739341819529683e-06,
      "loss": 0.6818,
      "step": 2058
    },
    {
      "epoch": 0.18996655518394648,
      "grad_norm": 1.797137528146229,
      "learning_rate": 1.1738442006030507e-06,
      "loss": 0.6997,
      "step": 2059
    },
    {
      "epoch": 0.19005881674547342,
      "grad_norm": 2.181809965715182,
      "learning_rate": 1.173754169516008e-06,
      "loss": 0.9243,
      "step": 2060
    },
    {
      "epoch": 0.19005881674547342,
      "eval_GEN Loss": 0.47521984577178955,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.38565847277641296,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.893629789352417,
      "eval_runtime": 55.9352,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2060
    },
    {
      "epoch": 0.19015107830700034,
      "grad_norm": 1.6489695067822996,
      "learning_rate": 1.1736640886999987e-06,
      "loss": 0.8201,
      "step": 2061
    },
    {
      "epoch": 0.19024333986852726,
      "grad_norm": 2.2030014674323226,
      "learning_rate": 1.1735739581631858e-06,
      "loss": 0.8497,
      "step": 2062
    },
    {
      "epoch": 0.1903356014300542,
      "grad_norm": 2.391613576562548,
      "learning_rate": 1.1734837779137374e-06,
      "loss": 0.6548,
      "step": 2063
    },
    {
      "epoch": 0.19042786299158113,
      "grad_norm": 2.438905114414268,
      "learning_rate": 1.1733935479598258e-06,
      "loss": 0.8612,
      "step": 2064
    },
    {
      "epoch": 0.19052012455310807,
      "grad_norm": 1.907891784799133,
      "learning_rate": 1.1733032683096279e-06,
      "loss": 0.7451,
      "step": 2065
    },
    {
      "epoch": 0.19052012455310807,
      "eval_GEN Loss": 0.476781964302063,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.3958512544631958,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9001802802085876,
      "eval_runtime": 56.1362,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 2065
    },
    {
      "epoch": 0.190612386114635,
      "grad_norm": 1.412398127994599,
      "learning_rate": 1.1732129389713246e-06,
      "loss": 0.7208,
      "step": 2066
    },
    {
      "epoch": 0.1907046476761619,
      "grad_norm": 1.5377625095076017,
      "learning_rate": 1.1731225599531018e-06,
      "loss": 0.7473,
      "step": 2067
    },
    {
      "epoch": 0.19079690923768886,
      "grad_norm": 1.739774552853579,
      "learning_rate": 1.1730321312631499e-06,
      "loss": 0.7613,
      "step": 2068
    },
    {
      "epoch": 0.19088917079921577,
      "grad_norm": 1.5628178567314877,
      "learning_rate": 1.1729416529096635e-06,
      "loss": 0.6486,
      "step": 2069
    },
    {
      "epoch": 0.1909814323607427,
      "grad_norm": 1.8349339622653353,
      "learning_rate": 1.1728511249008424e-06,
      "loss": 0.9405,
      "step": 2070
    },
    {
      "epoch": 0.1909814323607427,
      "eval_GEN Loss": 0.47572657465934753,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3906554877758026,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9009014368057251,
      "eval_runtime": 55.8639,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2070
    },
    {
      "epoch": 0.19107369392226964,
      "grad_norm": 1.6341536898808608,
      "learning_rate": 1.1727605472448899e-06,
      "loss": 0.641,
      "step": 2071
    },
    {
      "epoch": 0.19116595548379656,
      "grad_norm": 1.5734194760396665,
      "learning_rate": 1.1726699199500142e-06,
      "loss": 0.6824,
      "step": 2072
    },
    {
      "epoch": 0.1912582170453235,
      "grad_norm": 1.899392202542072,
      "learning_rate": 1.1725792430244286e-06,
      "loss": 0.8438,
      "step": 2073
    },
    {
      "epoch": 0.19135047860685042,
      "grad_norm": 1.4655415320012528,
      "learning_rate": 1.17248851647635e-06,
      "loss": 0.7546,
      "step": 2074
    },
    {
      "epoch": 0.19144274016837734,
      "grad_norm": 1.5406032119092525,
      "learning_rate": 1.1723977403140002e-06,
      "loss": 0.7013,
      "step": 2075
    },
    {
      "epoch": 0.19144274016837734,
      "eval_GEN Loss": 0.4752981960773468,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3990924656391144,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9125601053237915,
      "eval_runtime": 55.9667,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2075
    },
    {
      "epoch": 0.1915350017299043,
      "grad_norm": 2.1057681799535932,
      "learning_rate": 1.1723069145456055e-06,
      "loss": 0.7732,
      "step": 2076
    },
    {
      "epoch": 0.1916272632914312,
      "grad_norm": 1.8863658332607263,
      "learning_rate": 1.172216039179397e-06,
      "loss": 0.7382,
      "step": 2077
    },
    {
      "epoch": 0.19171952485295812,
      "grad_norm": 2.2750369010593197,
      "learning_rate": 1.1721251142236095e-06,
      "loss": 0.713,
      "step": 2078
    },
    {
      "epoch": 0.19181178641448507,
      "grad_norm": 2.5443842796035256,
      "learning_rate": 1.172034139686483e-06,
      "loss": 0.7111,
      "step": 2079
    },
    {
      "epoch": 0.191904047976012,
      "grad_norm": 2.0015549144460234,
      "learning_rate": 1.1719431155762617e-06,
      "loss": 0.8692,
      "step": 2080
    },
    {
      "epoch": 0.191904047976012,
      "eval_GEN Loss": 0.4749683737754822,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.4088994860649109,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9153845906257629,
      "eval_runtime": 56.0366,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2080
    },
    {
      "epoch": 0.19199630953753893,
      "grad_norm": 1.723692974059648,
      "learning_rate": 1.1718520419011943e-06,
      "loss": 0.7381,
      "step": 2081
    },
    {
      "epoch": 0.19208857109906585,
      "grad_norm": 1.6204546727085558,
      "learning_rate": 1.171760918669534e-06,
      "loss": 0.6291,
      "step": 2082
    },
    {
      "epoch": 0.19218083266059277,
      "grad_norm": 1.7591591876718784,
      "learning_rate": 1.171669745889539e-06,
      "loss": 0.6556,
      "step": 2083
    },
    {
      "epoch": 0.19227309422211972,
      "grad_norm": 1.6837994993989869,
      "learning_rate": 1.1715785235694709e-06,
      "loss": 0.5594,
      "step": 2084
    },
    {
      "epoch": 0.19236535578364664,
      "grad_norm": 2.1498111899108165,
      "learning_rate": 1.1714872517175967e-06,
      "loss": 0.976,
      "step": 2085
    },
    {
      "epoch": 0.19236535578364664,
      "eval_GEN Loss": 0.47523412108421326,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.4029233455657959,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9121394157409668,
      "eval_runtime": 55.9498,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2085
    },
    {
      "epoch": 0.19245761734517355,
      "grad_norm": 1.594466370015717,
      "learning_rate": 1.1713959303421877e-06,
      "loss": 0.7225,
      "step": 2086
    },
    {
      "epoch": 0.1925498789067005,
      "grad_norm": 1.2921041280679968,
      "learning_rate": 1.1713045594515194e-06,
      "loss": 0.6543,
      "step": 2087
    },
    {
      "epoch": 0.19264214046822742,
      "grad_norm": 1.603007946484057,
      "learning_rate": 1.171213139053872e-06,
      "loss": 0.757,
      "step": 2088
    },
    {
      "epoch": 0.19273440202975436,
      "grad_norm": 2.7035496344344776,
      "learning_rate": 1.1711216691575302e-06,
      "loss": 0.9921,
      "step": 2089
    },
    {
      "epoch": 0.19282666359128128,
      "grad_norm": 2.627501808473552,
      "learning_rate": 1.171030149770783e-06,
      "loss": 0.7437,
      "step": 2090
    },
    {
      "epoch": 0.19282666359128128,
      "eval_GEN Loss": 0.47421345114707947,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.37928611040115356,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9138821959495544,
      "eval_runtime": 55.8172,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2090
    },
    {
      "epoch": 0.1929189251528082,
      "grad_norm": 2.939675116898713,
      "learning_rate": 1.1709385809019243e-06,
      "loss": 0.9881,
      "step": 2091
    },
    {
      "epoch": 0.19301118671433515,
      "grad_norm": 1.5382624829545013,
      "learning_rate": 1.170846962559252e-06,
      "loss": 0.5454,
      "step": 2092
    },
    {
      "epoch": 0.19310344827586207,
      "grad_norm": 2.412140666845858,
      "learning_rate": 1.1707552947510689e-06,
      "loss": 0.8449,
      "step": 2093
    },
    {
      "epoch": 0.19319570983738898,
      "grad_norm": 2.050495972674451,
      "learning_rate": 1.170663577485682e-06,
      "loss": 0.8667,
      "step": 2094
    },
    {
      "epoch": 0.19328797139891593,
      "grad_norm": 2.412449004816114,
      "learning_rate": 1.1705718107714025e-06,
      "loss": 0.8212,
      "step": 2095
    },
    {
      "epoch": 0.19328797139891593,
      "eval_GEN Loss": 0.47555992007255554,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.37556973099708557,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9121994972229004,
      "eval_runtime": 56.066,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 2095
    },
    {
      "epoch": 0.19338023296044285,
      "grad_norm": 1.8618620941503157,
      "learning_rate": 1.1704799946165471e-06,
      "loss": 0.9453,
      "step": 2096
    },
    {
      "epoch": 0.1934724945219698,
      "grad_norm": 1.7248835980945905,
      "learning_rate": 1.1703881290294357e-06,
      "loss": 0.6785,
      "step": 2097
    },
    {
      "epoch": 0.19356475608349671,
      "grad_norm": 2.42521747941105,
      "learning_rate": 1.1702962140183938e-06,
      "loss": 0.9181,
      "step": 2098
    },
    {
      "epoch": 0.19365701764502363,
      "grad_norm": 1.6569213548155524,
      "learning_rate": 1.1702042495917507e-06,
      "loss": 0.4116,
      "step": 2099
    },
    {
      "epoch": 0.19374927920655058,
      "grad_norm": 2.445742982708023,
      "learning_rate": 1.1701122357578402e-06,
      "loss": 0.8711,
      "step": 2100
    },
    {
      "epoch": 0.19374927920655058,
      "eval_GEN Loss": 0.47657743096351624,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.38126426935195923,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.904026448726654,
      "eval_runtime": 55.9303,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2100
    },
    {
      "epoch": 0.1938415407680775,
      "grad_norm": 2.489860100516864,
      "learning_rate": 1.170020172525001e-06,
      "loss": 0.8869,
      "step": 2101
    },
    {
      "epoch": 0.19393380232960442,
      "grad_norm": 1.7185828996157118,
      "learning_rate": 1.1699280599015758e-06,
      "loss": 0.8002,
      "step": 2102
    },
    {
      "epoch": 0.19402606389113136,
      "grad_norm": 1.730053667769409,
      "learning_rate": 1.1698358978959121e-06,
      "loss": 0.6431,
      "step": 2103
    },
    {
      "epoch": 0.19411832545265828,
      "grad_norm": 2.210130976296359,
      "learning_rate": 1.1697436865163617e-06,
      "loss": 0.8812,
      "step": 2104
    },
    {
      "epoch": 0.19421058701418523,
      "grad_norm": 1.7501335028584264,
      "learning_rate": 1.1696514257712812e-06,
      "loss": 0.6264,
      "step": 2105
    },
    {
      "epoch": 0.19421058701418523,
      "eval_GEN Loss": 0.47559112310409546,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.37901872396469116,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9000601172447205,
      "eval_runtime": 55.8737,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2105
    },
    {
      "epoch": 0.19430284857571214,
      "grad_norm": 1.518035137093612,
      "learning_rate": 1.169559115669031e-06,
      "loss": 0.8045,
      "step": 2106
    },
    {
      "epoch": 0.19439511013723906,
      "grad_norm": 1.8605848033256487,
      "learning_rate": 1.1694667562179766e-06,
      "loss": 0.7365,
      "step": 2107
    },
    {
      "epoch": 0.194487371698766,
      "grad_norm": 2.2692257263639997,
      "learning_rate": 1.1693743474264879e-06,
      "loss": 0.8885,
      "step": 2108
    },
    {
      "epoch": 0.19457963326029293,
      "grad_norm": 1.7389934815117432,
      "learning_rate": 1.1692818893029386e-06,
      "loss": 0.61,
      "step": 2109
    },
    {
      "epoch": 0.19467189482181985,
      "grad_norm": 1.8801892562984148,
      "learning_rate": 1.1691893818557081e-06,
      "loss": 0.7616,
      "step": 2110
    },
    {
      "epoch": 0.19467189482181985,
      "eval_GEN Loss": 0.4751238226890564,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3778570294380188,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8924880027770996,
      "eval_runtime": 56.1755,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 2110
    },
    {
      "epoch": 0.1947641563833468,
      "grad_norm": 2.9935999868900574,
      "learning_rate": 1.1690968250931793e-06,
      "loss": 0.8215,
      "step": 2111
    },
    {
      "epoch": 0.1948564179448737,
      "grad_norm": 1.240026989214338,
      "learning_rate": 1.1690042190237397e-06,
      "loss": 0.7554,
      "step": 2112
    },
    {
      "epoch": 0.19494867950640066,
      "grad_norm": 1.663347006712576,
      "learning_rate": 1.1689115636557817e-06,
      "loss": 0.7469,
      "step": 2113
    },
    {
      "epoch": 0.19504094106792758,
      "grad_norm": 2.1324165398294737,
      "learning_rate": 1.1688188589977012e-06,
      "loss": 0.878,
      "step": 2114
    },
    {
      "epoch": 0.1951332026294545,
      "grad_norm": 1.5318585856133733,
      "learning_rate": 1.1687261050578999e-06,
      "loss": 0.7284,
      "step": 2115
    },
    {
      "epoch": 0.1951332026294545,
      "eval_GEN Loss": 0.47557395696640015,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.38053131103515625,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8861178159713745,
      "eval_runtime": 55.9378,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2115
    },
    {
      "epoch": 0.19522546419098144,
      "grad_norm": 1.9406484809626303,
      "learning_rate": 1.1686333018447831e-06,
      "loss": 0.7447,
      "step": 2116
    },
    {
      "epoch": 0.19531772575250836,
      "grad_norm": 1.4661856513402853,
      "learning_rate": 1.168540449366761e-06,
      "loss": 0.7673,
      "step": 2117
    },
    {
      "epoch": 0.19540998731403528,
      "grad_norm": 1.5532965239645762,
      "learning_rate": 1.1684475476322477e-06,
      "loss": 0.6529,
      "step": 2118
    },
    {
      "epoch": 0.19550224887556222,
      "grad_norm": 1.7328866001668306,
      "learning_rate": 1.1683545966496624e-06,
      "loss": 0.7228,
      "step": 2119
    },
    {
      "epoch": 0.19559451043708914,
      "grad_norm": 2.653230913036073,
      "learning_rate": 1.1682615964274282e-06,
      "loss": 0.9043,
      "step": 2120
    },
    {
      "epoch": 0.19559451043708914,
      "eval_GEN Loss": 0.4752795100212097,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3887932300567627,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8825721144676208,
      "eval_runtime": 56.2164,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 2120
    },
    {
      "epoch": 0.1956867719986161,
      "grad_norm": 1.834892901245111,
      "learning_rate": 1.1681685469739728e-06,
      "loss": 0.736,
      "step": 2121
    },
    {
      "epoch": 0.195779033560143,
      "grad_norm": 2.268595314023195,
      "learning_rate": 1.168075448297729e-06,
      "loss": 0.7198,
      "step": 2122
    },
    {
      "epoch": 0.19587129512166992,
      "grad_norm": 2.8209869109637893,
      "learning_rate": 1.167982300407133e-06,
      "loss": 0.9889,
      "step": 2123
    },
    {
      "epoch": 0.19596355668319687,
      "grad_norm": 2.2758613417274995,
      "learning_rate": 1.1678891033106265e-06,
      "loss": 0.7325,
      "step": 2124
    },
    {
      "epoch": 0.1960558182447238,
      "grad_norm": 2.5891500558623437,
      "learning_rate": 1.167795857016655e-06,
      "loss": 0.7827,
      "step": 2125
    },
    {
      "epoch": 0.1960558182447238,
      "eval_GEN Loss": 0.4752276539802551,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3605865240097046,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8686298131942749,
      "eval_runtime": 56.9304,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2125
    },
    {
      "epoch": 0.1961480798062507,
      "grad_norm": 1.7321506869110495,
      "learning_rate": 1.1677025615336684e-06,
      "loss": 0.8529,
      "step": 2126
    },
    {
      "epoch": 0.19624034136777765,
      "grad_norm": 1.6446856506620964,
      "learning_rate": 1.1676092168701214e-06,
      "loss": 0.812,
      "step": 2127
    },
    {
      "epoch": 0.19633260292930457,
      "grad_norm": 2.035225255341396,
      "learning_rate": 1.1675158230344731e-06,
      "loss": 0.6864,
      "step": 2128
    },
    {
      "epoch": 0.19642486449083152,
      "grad_norm": 1.6727419464219047,
      "learning_rate": 1.1674223800351871e-06,
      "loss": 0.6947,
      "step": 2129
    },
    {
      "epoch": 0.19651712605235844,
      "grad_norm": 2.7270117150880293,
      "learning_rate": 1.1673288878807312e-06,
      "loss": 0.9565,
      "step": 2130
    },
    {
      "epoch": 0.19651712605235844,
      "eval_GEN Loss": 0.4728906452655792,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM Loss": 0.34505847096443176,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8583533763885498,
      "eval_runtime": 56.8477,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2130
    },
    {
      "epoch": 0.19660938761388536,
      "grad_norm": 2.315561439377134,
      "learning_rate": 1.1672353465795778e-06,
      "loss": 0.8579,
      "step": 2131
    },
    {
      "epoch": 0.1967016491754123,
      "grad_norm": 2.445484447216849,
      "learning_rate": 1.1671417561402037e-06,
      "loss": 0.7155,
      "step": 2132
    },
    {
      "epoch": 0.19679391073693922,
      "grad_norm": 1.4171824307104404,
      "learning_rate": 1.1670481165710906e-06,
      "loss": 0.786,
      "step": 2133
    },
    {
      "epoch": 0.19688617229846614,
      "grad_norm": 1.9530654960217884,
      "learning_rate": 1.1669544278807239e-06,
      "loss": 0.791,
      "step": 2134
    },
    {
      "epoch": 0.19697843385999309,
      "grad_norm": 1.9259482942527117,
      "learning_rate": 1.1668606900775936e-06,
      "loss": 0.855,
      "step": 2135
    },
    {
      "epoch": 0.19697843385999309,
      "eval_GEN Loss": 0.4720461666584015,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.34648963809013367,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8544471263885498,
      "eval_runtime": 56.7497,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2135
    },
    {
      "epoch": 0.19707069542152,
      "grad_norm": 1.7540937429449306,
      "learning_rate": 1.166766903170195e-06,
      "loss": 0.6871,
      "step": 2136
    },
    {
      "epoch": 0.19716295698304695,
      "grad_norm": 2.2431423540689486,
      "learning_rate": 1.1666730671670269e-06,
      "loss": 0.7345,
      "step": 2137
    },
    {
      "epoch": 0.19725521854457387,
      "grad_norm": 2.180562741419492,
      "learning_rate": 1.1665791820765926e-06,
      "loss": 0.6921,
      "step": 2138
    },
    {
      "epoch": 0.1973474801061008,
      "grad_norm": 1.869450975392008,
      "learning_rate": 1.1664852479074005e-06,
      "loss": 0.8264,
      "step": 2139
    },
    {
      "epoch": 0.19743974166762773,
      "grad_norm": 1.57533144178452,
      "learning_rate": 1.1663912646679628e-06,
      "loss": 0.6432,
      "step": 2140
    },
    {
      "epoch": 0.19743974166762773,
      "eval_GEN Loss": 0.47172442078590393,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.3543279767036438,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8531851172447205,
      "eval_runtime": 55.8033,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2140
    },
    {
      "epoch": 0.19753200322915465,
      "grad_norm": 1.2884555954442063,
      "learning_rate": 1.1662972323667967e-06,
      "loss": 0.5873,
      "step": 2141
    },
    {
      "epoch": 0.19762426479068157,
      "grad_norm": 2.5057557700609068,
      "learning_rate": 1.1662031510124233e-06,
      "loss": 0.7394,
      "step": 2142
    },
    {
      "epoch": 0.19771652635220852,
      "grad_norm": 2.1793536972246894,
      "learning_rate": 1.1661090206133686e-06,
      "loss": 0.8169,
      "step": 2143
    },
    {
      "epoch": 0.19780878791373543,
      "grad_norm": 1.7726817338950984,
      "learning_rate": 1.1660148411781625e-06,
      "loss": 0.824,
      "step": 2144
    },
    {
      "epoch": 0.19790104947526238,
      "grad_norm": 2.0722325893696443,
      "learning_rate": 1.16592061271534e-06,
      "loss": 0.5855,
      "step": 2145
    },
    {
      "epoch": 0.19790104947526238,
      "eval_GEN Loss": 0.4713197350502014,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3694514036178589,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8676682710647583,
      "eval_runtime": 55.8797,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2145
    },
    {
      "epoch": 0.1979933110367893,
      "grad_norm": 1.4638687069641676,
      "learning_rate": 1.1658263352334402e-06,
      "loss": 0.684,
      "step": 2146
    },
    {
      "epoch": 0.19808557259831622,
      "grad_norm": 1.3822063982717827,
      "learning_rate": 1.1657320087410068e-06,
      "loss": 0.6227,
      "step": 2147
    },
    {
      "epoch": 0.19817783415984316,
      "grad_norm": 1.709733050362052,
      "learning_rate": 1.1656376332465873e-06,
      "loss": 0.6928,
      "step": 2148
    },
    {
      "epoch": 0.19827009572137008,
      "grad_norm": 1.8248242738528617,
      "learning_rate": 1.1655432087587348e-06,
      "loss": 0.8613,
      "step": 2149
    },
    {
      "epoch": 0.198362357282897,
      "grad_norm": 1.9798829605323542,
      "learning_rate": 1.165448735286006e-06,
      "loss": 0.7774,
      "step": 2150
    },
    {
      "epoch": 0.198362357282897,
      "eval_GEN Loss": 0.4713093638420105,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3774649500846863,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8806490302085876,
      "eval_runtime": 55.8184,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2150
    },
    {
      "epoch": 0.19845461884442395,
      "grad_norm": 2.9023380907319134,
      "learning_rate": 1.1653542128369615e-06,
      "loss": 1.0227,
      "step": 2151
    },
    {
      "epoch": 0.19854688040595087,
      "grad_norm": 1.9566433766921913,
      "learning_rate": 1.1652596414201683e-06,
      "loss": 0.6662,
      "step": 2152
    },
    {
      "epoch": 0.1986391419674778,
      "grad_norm": 3.0263774573118845,
      "learning_rate": 1.1651650210441958e-06,
      "loss": 1.0691,
      "step": 2153
    },
    {
      "epoch": 0.19873140352900473,
      "grad_norm": 1.8388460908019544,
      "learning_rate": 1.1650703517176188e-06,
      "loss": 0.5693,
      "step": 2154
    },
    {
      "epoch": 0.19882366509053165,
      "grad_norm": 1.7077486572942435,
      "learning_rate": 1.1649756334490165e-06,
      "loss": 0.7307,
      "step": 2155
    },
    {
      "epoch": 0.19882366509053165,
      "eval_GEN Loss": 0.4734862446784973,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4129568040370941,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8912860751152039,
      "eval_runtime": 55.758,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 2155
    },
    {
      "epoch": 0.1989159266520586,
      "grad_norm": 2.1519996732380497,
      "learning_rate": 1.1648808662469725e-06,
      "loss": 0.7858,
      "step": 2156
    },
    {
      "epoch": 0.1990081882135855,
      "grad_norm": 2.8792455478180288,
      "learning_rate": 1.1647860501200744e-06,
      "loss": 0.8351,
      "step": 2157
    },
    {
      "epoch": 0.19910044977511243,
      "grad_norm": 2.5978872367652084,
      "learning_rate": 1.164691185076915e-06,
      "loss": 0.7649,
      "step": 2158
    },
    {
      "epoch": 0.19919271133663938,
      "grad_norm": 2.4542299987549216,
      "learning_rate": 1.1645962711260908e-06,
      "loss": 0.861,
      "step": 2159
    },
    {
      "epoch": 0.1992849728981663,
      "grad_norm": 1.826676457982078,
      "learning_rate": 1.1645013082762034e-06,
      "loss": 0.7304,
      "step": 2160
    },
    {
      "epoch": 0.1992849728981663,
      "eval_GEN Loss": 0.4735404849052429,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4202714264392853,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.899218738079071,
      "eval_runtime": 55.8929,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2160
    },
    {
      "epoch": 0.19937723445969324,
      "grad_norm": 2.0889619371874173,
      "learning_rate": 1.164406296535858e-06,
      "loss": 0.875,
      "step": 2161
    },
    {
      "epoch": 0.19946949602122016,
      "grad_norm": 2.213142481426363,
      "learning_rate": 1.1643112359136648e-06,
      "loss": 0.6971,
      "step": 2162
    },
    {
      "epoch": 0.19956175758274708,
      "grad_norm": 2.0919058344979153,
      "learning_rate": 1.1642161264182389e-06,
      "loss": 0.7692,
      "step": 2163
    },
    {
      "epoch": 0.19965401914427403,
      "grad_norm": 2.0040945903130782,
      "learning_rate": 1.1641209680581986e-06,
      "loss": 0.7919,
      "step": 2164
    },
    {
      "epoch": 0.19974628070580094,
      "grad_norm": 2.67666508515558,
      "learning_rate": 1.1640257608421679e-06,
      "loss": 1.0232,
      "step": 2165
    },
    {
      "epoch": 0.19974628070580094,
      "eval_GEN Loss": 0.4732702374458313,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.41842952370643616,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9045072197914124,
      "eval_runtime": 55.9137,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2165
    },
    {
      "epoch": 0.19983854226732786,
      "grad_norm": 2.211300331233957,
      "learning_rate": 1.163930504778774e-06,
      "loss": 0.8335,
      "step": 2166
    },
    {
      "epoch": 0.1999308038288548,
      "grad_norm": 2.2516906475534624,
      "learning_rate": 1.1638351998766495e-06,
      "loss": 1.0556,
      "step": 2167
    },
    {
      "epoch": 0.20002306539038173,
      "grad_norm": 1.854903417938329,
      "learning_rate": 1.1637398461444313e-06,
      "loss": 0.7604,
      "step": 2168
    },
    {
      "epoch": 0.20011532695190867,
      "grad_norm": 2.695239187375436,
      "learning_rate": 1.16364444359076e-06,
      "loss": 0.9138,
      "step": 2169
    },
    {
      "epoch": 0.2002075885134356,
      "grad_norm": 2.417092201907459,
      "learning_rate": 1.1635489922242816e-06,
      "loss": 0.8951,
      "step": 2170
    },
    {
      "epoch": 0.2002075885134356,
      "eval_GEN Loss": 0.47371259331703186,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4227430522441864,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9177283644676208,
      "eval_runtime": 55.7797,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2170
    },
    {
      "epoch": 0.2002998500749625,
      "grad_norm": 2.8940595211064455,
      "learning_rate": 1.1634534920536458e-06,
      "loss": 0.926,
      "step": 2171
    },
    {
      "epoch": 0.20039211163648946,
      "grad_norm": 1.5796208126736675,
      "learning_rate": 1.1633579430875067e-06,
      "loss": 0.8173,
      "step": 2172
    },
    {
      "epoch": 0.20048437319801637,
      "grad_norm": 2.2360021997772366,
      "learning_rate": 1.1632623453345238e-06,
      "loss": 0.8244,
      "step": 2173
    },
    {
      "epoch": 0.2005766347595433,
      "grad_norm": 1.9058751759846053,
      "learning_rate": 1.1631666988033595e-06,
      "loss": 0.7873,
      "step": 2174
    },
    {
      "epoch": 0.20066889632107024,
      "grad_norm": 1.637412668967031,
      "learning_rate": 1.1630710035026824e-06,
      "loss": 0.8754,
      "step": 2175
    },
    {
      "epoch": 0.20066889632107024,
      "eval_GEN Loss": 0.4731568992137909,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.3976535201072693,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9191105961799622,
      "eval_runtime": 55.7887,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2175
    },
    {
      "epoch": 0.20076115788259716,
      "grad_norm": 2.0792545831549614,
      "learning_rate": 1.1629752594411636e-06,
      "loss": 0.8187,
      "step": 2176
    },
    {
      "epoch": 0.2008534194441241,
      "grad_norm": 1.5427235907101737,
      "learning_rate": 1.1628794666274803e-06,
      "loss": 0.7435,
      "step": 2177
    },
    {
      "epoch": 0.20094568100565102,
      "grad_norm": 1.9341408509913365,
      "learning_rate": 1.1627836250703127e-06,
      "loss": 0.842,
      "step": 2178
    },
    {
      "epoch": 0.20103794256717794,
      "grad_norm": 2.5845751856843986,
      "learning_rate": 1.162687734778347e-06,
      "loss": 0.6199,
      "step": 2179
    },
    {
      "epoch": 0.2011302041287049,
      "grad_norm": 1.7097335293652327,
      "learning_rate": 1.1625917957602723e-06,
      "loss": 0.6996,
      "step": 2180
    },
    {
      "epoch": 0.2011302041287049,
      "eval_GEN Loss": 0.4752475619316101,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.40738368034362793,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9316105842590332,
      "eval_runtime": 55.9206,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2180
    },
    {
      "epoch": 0.2012224656902318,
      "grad_norm": 1.7254864136798387,
      "learning_rate": 1.1624958080247826e-06,
      "loss": 0.6264,
      "step": 2181
    },
    {
      "epoch": 0.20131472725175872,
      "grad_norm": 1.7315952013574136,
      "learning_rate": 1.1623997715805771e-06,
      "loss": 0.6824,
      "step": 2182
    },
    {
      "epoch": 0.20140698881328567,
      "grad_norm": 1.5971953206050904,
      "learning_rate": 1.1623036864363582e-06,
      "loss": 0.7546,
      "step": 2183
    },
    {
      "epoch": 0.2014992503748126,
      "grad_norm": 2.0592919744913183,
      "learning_rate": 1.1622075526008337e-06,
      "loss": 0.636,
      "step": 2184
    },
    {
      "epoch": 0.20159151193633953,
      "grad_norm": 1.4963490463559308,
      "learning_rate": 1.162111370082715e-06,
      "loss": 0.6686,
      "step": 2185
    },
    {
      "epoch": 0.20159151193633953,
      "eval_GEN Loss": 0.475746750831604,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3980647027492523,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9399038553237915,
      "eval_runtime": 56.1855,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 2185
    },
    {
      "epoch": 0.20168377349786645,
      "grad_norm": 1.7118715761439545,
      "learning_rate": 1.1620151388907186e-06,
      "loss": 0.8563,
      "step": 2186
    },
    {
      "epoch": 0.20177603505939337,
      "grad_norm": 1.8824076807560626,
      "learning_rate": 1.1619188590335651e-06,
      "loss": 0.9051,
      "step": 2187
    },
    {
      "epoch": 0.20186829662092032,
      "grad_norm": 1.502251399123566,
      "learning_rate": 1.1618225305199794e-06,
      "loss": 0.563,
      "step": 2188
    },
    {
      "epoch": 0.20196055818244724,
      "grad_norm": 2.339386555095602,
      "learning_rate": 1.161726153358691e-06,
      "loss": 0.9121,
      "step": 2189
    },
    {
      "epoch": 0.20205281974397415,
      "grad_norm": 1.9191486206770145,
      "learning_rate": 1.1616297275584338e-06,
      "loss": 0.8298,
      "step": 2190
    },
    {
      "epoch": 0.20205281974397415,
      "eval_GEN Loss": 0.4742852747440338,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3946433961391449,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9516226053237915,
      "eval_runtime": 56.8234,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2190
    },
    {
      "epoch": 0.2021450813055011,
      "grad_norm": 1.4872003622550216,
      "learning_rate": 1.1615332531279459e-06,
      "loss": 0.7534,
      "step": 2191
    },
    {
      "epoch": 0.20223734286702802,
      "grad_norm": 1.8293995126466744,
      "learning_rate": 1.1614367300759702e-06,
      "loss": 0.7144,
      "step": 2192
    },
    {
      "epoch": 0.20232960442855497,
      "grad_norm": 2.424816644760713,
      "learning_rate": 1.1613401584112535e-06,
      "loss": 0.9931,
      "step": 2193
    },
    {
      "epoch": 0.20242186599008188,
      "grad_norm": 1.4672034085405155,
      "learning_rate": 1.1612435381425478e-06,
      "loss": 0.6609,
      "step": 2194
    },
    {
      "epoch": 0.2025141275516088,
      "grad_norm": 3.1144754000810835,
      "learning_rate": 1.1611468692786083e-06,
      "loss": 0.8067,
      "step": 2195
    },
    {
      "epoch": 0.2025141275516088,
      "eval_GEN Loss": 0.47315630316734314,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.39846715331077576,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9524038434028625,
      "eval_runtime": 56.0352,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2195
    },
    {
      "epoch": 0.20260638911313575,
      "grad_norm": 1.5713749669294763,
      "learning_rate": 1.1610501518281957e-06,
      "loss": 0.9158,
      "step": 2196
    },
    {
      "epoch": 0.20269865067466267,
      "grad_norm": 2.8028927388889344,
      "learning_rate": 1.1609533858000747e-06,
      "loss": 0.7719,
      "step": 2197
    },
    {
      "epoch": 0.20279091223618959,
      "grad_norm": 1.6739765872226005,
      "learning_rate": 1.160856571203014e-06,
      "loss": 0.7266,
      "step": 2198
    },
    {
      "epoch": 0.20288317379771653,
      "grad_norm": 1.615812143284236,
      "learning_rate": 1.1607597080457876e-06,
      "loss": 0.8073,
      "step": 2199
    },
    {
      "epoch": 0.20297543535924345,
      "grad_norm": 2.141367380534688,
      "learning_rate": 1.160662796337173e-06,
      "loss": 0.8588,
      "step": 2200
    },
    {
      "epoch": 0.20297543535924345,
      "eval_GEN Loss": 0.4725123941898346,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.4200517237186432,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.940625011920929,
      "eval_runtime": 56.0489,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2200
    },
    {
      "epoch": 0.2030676969207704,
      "grad_norm": 1.8213880271221528,
      "learning_rate": 1.1605658360859528e-06,
      "loss": 0.6891,
      "step": 2201
    },
    {
      "epoch": 0.20315995848229731,
      "grad_norm": 1.656531375970182,
      "learning_rate": 1.1604688273009136e-06,
      "loss": 0.7833,
      "step": 2202
    },
    {
      "epoch": 0.20325222004382423,
      "grad_norm": 2.263318720055018,
      "learning_rate": 1.1603717699908461e-06,
      "loss": 0.7284,
      "step": 2203
    },
    {
      "epoch": 0.20334448160535118,
      "grad_norm": 2.576418203657882,
      "learning_rate": 1.1602746641645462e-06,
      "loss": 0.7333,
      "step": 2204
    },
    {
      "epoch": 0.2034367431668781,
      "grad_norm": 1.767694725912544,
      "learning_rate": 1.160177509830814e-06,
      "loss": 0.6952,
      "step": 2205
    },
    {
      "epoch": 0.2034367431668781,
      "eval_GEN Loss": 0.47379887104034424,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.44554200768470764,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9538461565971375,
      "eval_runtime": 55.7983,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2205
    },
    {
      "epoch": 0.20352900472840502,
      "grad_norm": 2.023565969537719,
      "learning_rate": 1.160080306998453e-06,
      "loss": 0.5573,
      "step": 2206
    },
    {
      "epoch": 0.20362126628993196,
      "grad_norm": 1.5327095711767684,
      "learning_rate": 1.1599830556762728e-06,
      "loss": 0.7337,
      "step": 2207
    },
    {
      "epoch": 0.20371352785145888,
      "grad_norm": 2.4388722426346257,
      "learning_rate": 1.1598857558730858e-06,
      "loss": 0.8009,
      "step": 2208
    },
    {
      "epoch": 0.20380578941298583,
      "grad_norm": 1.7281652578654474,
      "learning_rate": 1.1597884075977097e-06,
      "loss": 0.6855,
      "step": 2209
    },
    {
      "epoch": 0.20389805097451275,
      "grad_norm": 2.320746802377084,
      "learning_rate": 1.1596910108589665e-06,
      "loss": 0.864,
      "step": 2210
    },
    {
      "epoch": 0.20389805097451275,
      "eval_GEN Loss": 0.4759262502193451,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.4454600512981415,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9655048251152039,
      "eval_runtime": 55.8888,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2210
    },
    {
      "epoch": 0.20399031253603966,
      "grad_norm": 1.5772755489697625,
      "learning_rate": 1.1595935656656823e-06,
      "loss": 0.6484,
      "step": 2211
    },
    {
      "epoch": 0.2040825740975666,
      "grad_norm": 1.9593363728966164,
      "learning_rate": 1.1594960720266875e-06,
      "loss": 0.8232,
      "step": 2212
    },
    {
      "epoch": 0.20417483565909353,
      "grad_norm": 3.061828812767506,
      "learning_rate": 1.1593985299508173e-06,
      "loss": 0.7763,
      "step": 2213
    },
    {
      "epoch": 0.20426709722062045,
      "grad_norm": 2.034800705615405,
      "learning_rate": 1.1593009394469113e-06,
      "loss": 0.6538,
      "step": 2214
    },
    {
      "epoch": 0.2043593587821474,
      "grad_norm": 2.2677067288477204,
      "learning_rate": 1.1592033005238132e-06,
      "loss": 0.8398,
      "step": 2215
    },
    {
      "epoch": 0.2043593587821474,
      "eval_GEN Loss": 0.4771953225135803,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.42658737301826477,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9716346263885498,
      "eval_runtime": 57.2219,
      "eval_samples_per_second": 1.136,
      "eval_steps_per_second": 0.087,
      "step": 2215
    },
    {
      "epoch": 0.2044516203436743,
      "grad_norm": 1.8709461722271208,
      "learning_rate": 1.159105613190371e-06,
      "loss": 0.6145,
      "step": 2216
    },
    {
      "epoch": 0.20454388190520126,
      "grad_norm": 1.6362598916212179,
      "learning_rate": 1.1590078774554375e-06,
      "loss": 0.5606,
      "step": 2217
    },
    {
      "epoch": 0.20463614346672818,
      "grad_norm": 1.5807983911321843,
      "learning_rate": 1.1589100933278698e-06,
      "loss": 0.5946,
      "step": 2218
    },
    {
      "epoch": 0.2047284050282551,
      "grad_norm": 1.9108688609185664,
      "learning_rate": 1.158812260816529e-06,
      "loss": 0.5409,
      "step": 2219
    },
    {
      "epoch": 0.20482066658978204,
      "grad_norm": 1.9265345321607514,
      "learning_rate": 1.1587143799302806e-06,
      "loss": 0.6977,
      "step": 2220
    },
    {
      "epoch": 0.20482066658978204,
      "eval_GEN Loss": 0.4785206615924835,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.41944700479507446,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9948317408561707,
      "eval_runtime": 56.9385,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2220
    },
    {
      "epoch": 0.20491292815130896,
      "grad_norm": 2.395886053095591,
      "learning_rate": 1.1586164506779952e-06,
      "loss": 0.828,
      "step": 2221
    },
    {
      "epoch": 0.20500518971283588,
      "grad_norm": 2.119859533382076,
      "learning_rate": 1.1585184730685472e-06,
      "loss": 0.7004,
      "step": 2222
    },
    {
      "epoch": 0.20509745127436282,
      "grad_norm": 1.5753261897183282,
      "learning_rate": 1.1584204471108151e-06,
      "loss": 0.6569,
      "step": 2223
    },
    {
      "epoch": 0.20518971283588974,
      "grad_norm": 2.784071443294006,
      "learning_rate": 1.1583223728136828e-06,
      "loss": 0.8896,
      "step": 2224
    },
    {
      "epoch": 0.2052819743974167,
      "grad_norm": 3.5920705062442426,
      "learning_rate": 1.1582242501860375e-06,
      "loss": 0.9065,
      "step": 2225
    },
    {
      "epoch": 0.2052819743974167,
      "eval_GEN Loss": 0.4771478474140167,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.4128265082836151,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 1.0010817050933838,
      "eval_runtime": 56.9999,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 2225
    },
    {
      "epoch": 0.2053742359589436,
      "grad_norm": 2.849917161964412,
      "learning_rate": 1.1581260792367714e-06,
      "loss": 0.9079,
      "step": 2226
    },
    {
      "epoch": 0.20546649752047053,
      "grad_norm": 4.075805587999574,
      "learning_rate": 1.1580278599747807e-06,
      "loss": 0.9575,
      "step": 2227
    },
    {
      "epoch": 0.20555875908199747,
      "grad_norm": 1.5820298116715013,
      "learning_rate": 1.1579295924089665e-06,
      "loss": 0.5648,
      "step": 2228
    },
    {
      "epoch": 0.2056510206435244,
      "grad_norm": 1.8614520135578403,
      "learning_rate": 1.1578312765482335e-06,
      "loss": 0.874,
      "step": 2229
    },
    {
      "epoch": 0.2057432822050513,
      "grad_norm": 1.973629767428031,
      "learning_rate": 1.1577329124014918e-06,
      "loss": 0.8192,
      "step": 2230
    },
    {
      "epoch": 0.2057432822050513,
      "eval_GEN Loss": 0.4752151370048523,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.40567606687545776,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9876201748847961,
      "eval_runtime": 57.1134,
      "eval_samples_per_second": 1.138,
      "eval_steps_per_second": 0.088,
      "step": 2230
    },
    {
      "epoch": 0.20583554376657826,
      "grad_norm": 1.959528236745921,
      "learning_rate": 1.157634499977655e-06,
      "loss": 0.7145,
      "step": 2231
    },
    {
      "epoch": 0.20592780532810517,
      "grad_norm": 3.7674519565202194,
      "learning_rate": 1.1575360392856414e-06,
      "loss": 0.6961,
      "step": 2232
    },
    {
      "epoch": 0.20602006688963212,
      "grad_norm": 2.9903737460475868,
      "learning_rate": 1.1574375303343738e-06,
      "loss": 1.0235,
      "step": 2233
    },
    {
      "epoch": 0.20611232845115904,
      "grad_norm": 2.062107256056617,
      "learning_rate": 1.157338973132779e-06,
      "loss": 0.8793,
      "step": 2234
    },
    {
      "epoch": 0.20620459001268596,
      "grad_norm": 1.5714567565958513,
      "learning_rate": 1.1572403676897886e-06,
      "loss": 0.6572,
      "step": 2235
    },
    {
      "epoch": 0.20620459001268596,
      "eval_GEN Loss": 0.47319895029067993,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.41750359535217285,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9718149304389954,
      "eval_runtime": 56.9057,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2235
    },
    {
      "epoch": 0.2062968515742129,
      "grad_norm": 2.102201250010656,
      "learning_rate": 1.1571417140143384e-06,
      "loss": 0.8417,
      "step": 2236
    },
    {
      "epoch": 0.20638911313573982,
      "grad_norm": 1.4546035367495007,
      "learning_rate": 1.1570430121153683e-06,
      "loss": 0.6773,
      "step": 2237
    },
    {
      "epoch": 0.20648137469726674,
      "grad_norm": 2.257513931694606,
      "learning_rate": 1.1569442620018232e-06,
      "loss": 0.8263,
      "step": 2238
    },
    {
      "epoch": 0.20657363625879369,
      "grad_norm": 1.5450495346865374,
      "learning_rate": 1.1568454636826515e-06,
      "loss": 0.6472,
      "step": 2239
    },
    {
      "epoch": 0.2066658978203206,
      "grad_norm": 2.2640080839540087,
      "learning_rate": 1.1567466171668067e-06,
      "loss": 0.8704,
      "step": 2240
    },
    {
      "epoch": 0.2066658978203206,
      "eval_GEN Loss": 0.47217726707458496,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4422333836555481,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9549278616905212,
      "eval_runtime": 56.8929,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2240
    },
    {
      "epoch": 0.20675815938184755,
      "grad_norm": 1.9422457799858017,
      "learning_rate": 1.156647722463247e-06,
      "loss": 0.6511,
      "step": 2241
    },
    {
      "epoch": 0.20685042094337447,
      "grad_norm": 1.882138731890447,
      "learning_rate": 1.1565487795809334e-06,
      "loss": 0.7802,
      "step": 2242
    },
    {
      "epoch": 0.2069426825049014,
      "grad_norm": 2.08874878682335,
      "learning_rate": 1.1564497885288328e-06,
      "loss": 0.7902,
      "step": 2243
    },
    {
      "epoch": 0.20703494406642833,
      "grad_norm": 1.557435516696818,
      "learning_rate": 1.156350749315916e-06,
      "loss": 0.7547,
      "step": 2244
    },
    {
      "epoch": 0.20712720562795525,
      "grad_norm": 1.3826535986615829,
      "learning_rate": 1.1562516619511576e-06,
      "loss": 0.7163,
      "step": 2245
    },
    {
      "epoch": 0.20712720562795525,
      "eval_GEN Loss": 0.47124239802360535,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4405944049358368,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9440504908561707,
      "eval_runtime": 57.0393,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 2245
    },
    {
      "epoch": 0.20721946718948217,
      "grad_norm": 2.6485078662914034,
      "learning_rate": 1.1561525264435378e-06,
      "loss": 0.7523,
      "step": 2246
    },
    {
      "epoch": 0.20731172875100912,
      "grad_norm": 2.150377140764201,
      "learning_rate": 1.1560533428020398e-06,
      "loss": 0.9366,
      "step": 2247
    },
    {
      "epoch": 0.20740399031253604,
      "grad_norm": 1.5380322833366769,
      "learning_rate": 1.155954111035652e-06,
      "loss": 0.6323,
      "step": 2248
    },
    {
      "epoch": 0.20749625187406298,
      "grad_norm": 2.707062319921174,
      "learning_rate": 1.1558548311533671e-06,
      "loss": 1.0883,
      "step": 2249
    },
    {
      "epoch": 0.2075885134355899,
      "grad_norm": 1.7728842055382055,
      "learning_rate": 1.1557555031641815e-06,
      "loss": 0.8393,
      "step": 2250
    },
    {
      "epoch": 0.2075885134355899,
      "eval_GEN Loss": 0.46939241886138916,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4045702815055847,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9229567050933838,
      "eval_runtime": 56.2359,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 2250
    },
    {
      "epoch": 0.20768077499711682,
      "grad_norm": 1.9400906411710708,
      "learning_rate": 1.1556561270770971e-06,
      "loss": 0.8392,
      "step": 2251
    },
    {
      "epoch": 0.20777303655864376,
      "grad_norm": 1.713949094373851,
      "learning_rate": 1.155556702901119e-06,
      "loss": 0.6385,
      "step": 2252
    },
    {
      "epoch": 0.20786529812017068,
      "grad_norm": 1.4296639261220656,
      "learning_rate": 1.1554572306452575e-06,
      "loss": 0.5572,
      "step": 2253
    },
    {
      "epoch": 0.2079575596816976,
      "grad_norm": 2.0860711101621385,
      "learning_rate": 1.1553577103185268e-06,
      "loss": 0.6838,
      "step": 2254
    },
    {
      "epoch": 0.20804982124322455,
      "grad_norm": 1.2807061636561392,
      "learning_rate": 1.1552581419299455e-06,
      "loss": 0.7326,
      "step": 2255
    },
    {
      "epoch": 0.20804982124322455,
      "eval_GEN Loss": 0.46940556168556213,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.37396758794784546,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9145432710647583,
      "eval_runtime": 56.0935,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 2255
    },
    {
      "epoch": 0.20814208280475147,
      "grad_norm": 1.6619205021563814,
      "learning_rate": 1.1551585254885366e-06,
      "loss": 0.6068,
      "step": 2256
    },
    {
      "epoch": 0.2082343443662784,
      "grad_norm": 1.758571705646242,
      "learning_rate": 1.155058861003328e-06,
      "loss": 0.7124,
      "step": 2257
    },
    {
      "epoch": 0.20832660592780533,
      "grad_norm": 2.3864128225982273,
      "learning_rate": 1.1549591484833509e-06,
      "loss": 0.7266,
      "step": 2258
    },
    {
      "epoch": 0.20841886748933225,
      "grad_norm": 2.0267324513331517,
      "learning_rate": 1.1548593879376417e-06,
      "loss": 0.6844,
      "step": 2259
    },
    {
      "epoch": 0.2085111290508592,
      "grad_norm": 1.520344931273364,
      "learning_rate": 1.1547595793752404e-06,
      "loss": 0.6734,
      "step": 2260
    },
    {
      "epoch": 0.2085111290508592,
      "eval_GEN Loss": 0.4700671434402466,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3706658184528351,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9083533883094788,
      "eval_runtime": 56.1305,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 2260
    },
    {
      "epoch": 0.2086033906123861,
      "grad_norm": 1.5288111304880814,
      "learning_rate": 1.1546597228051924e-06,
      "loss": 0.5556,
      "step": 2261
    },
    {
      "epoch": 0.20869565217391303,
      "grad_norm": 1.6684899236360808,
      "learning_rate": 1.1545598182365467e-06,
      "loss": 0.6458,
      "step": 2262
    },
    {
      "epoch": 0.20878791373543998,
      "grad_norm": 1.6059246926166975,
      "learning_rate": 1.1544598656783566e-06,
      "loss": 0.7448,
      "step": 2263
    },
    {
      "epoch": 0.2088801752969669,
      "grad_norm": 1.6885814137936577,
      "learning_rate": 1.1543598651396803e-06,
      "loss": 0.6189,
      "step": 2264
    },
    {
      "epoch": 0.20897243685849384,
      "grad_norm": 2.0709593140019833,
      "learning_rate": 1.1542598166295795e-06,
      "loss": 0.8351,
      "step": 2265
    },
    {
      "epoch": 0.20897243685849384,
      "eval_GEN Loss": 0.47002270817756653,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.3875349462032318,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9081730842590332,
      "eval_runtime": 56.9324,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2265
    },
    {
      "epoch": 0.20906469842002076,
      "grad_norm": 1.4131631275588865,
      "learning_rate": 1.1541597201571213e-06,
      "loss": 0.6641,
      "step": 2266
    },
    {
      "epoch": 0.20915695998154768,
      "grad_norm": 1.802653364706875,
      "learning_rate": 1.1540595757313762e-06,
      "loss": 0.6437,
      "step": 2267
    },
    {
      "epoch": 0.20924922154307463,
      "grad_norm": 1.5221604622967477,
      "learning_rate": 1.1539593833614197e-06,
      "loss": 0.5606,
      "step": 2268
    },
    {
      "epoch": 0.20934148310460154,
      "grad_norm": 2.8260828422142095,
      "learning_rate": 1.153859143056331e-06,
      "loss": 0.6902,
      "step": 2269
    },
    {
      "epoch": 0.20943374466612846,
      "grad_norm": 2.305636405276973,
      "learning_rate": 1.1537588548251949e-06,
      "loss": 0.6833,
      "step": 2270
    },
    {
      "epoch": 0.20943374466612846,
      "eval_GEN Loss": 0.46963343024253845,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.40641361474990845,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.918749988079071,
      "eval_runtime": 56.8849,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2270
    },
    {
      "epoch": 0.2095260062276554,
      "grad_norm": 3.34761906162355,
      "learning_rate": 1.1536585186770986e-06,
      "loss": 0.8558,
      "step": 2271
    },
    {
      "epoch": 0.20961826778918233,
      "grad_norm": 1.6061109119553205,
      "learning_rate": 1.1535581346211352e-06,
      "loss": 0.9361,
      "step": 2272
    },
    {
      "epoch": 0.20971052935070927,
      "grad_norm": 2.5392476724228845,
      "learning_rate": 1.153457702666402e-06,
      "loss": 0.9529,
      "step": 2273
    },
    {
      "epoch": 0.2098027909122362,
      "grad_norm": 1.9642139317507186,
      "learning_rate": 1.1533572228219998e-06,
      "loss": 0.6781,
      "step": 2274
    },
    {
      "epoch": 0.2098950524737631,
      "grad_norm": 2.1817493632736245,
      "learning_rate": 1.1532566950970344e-06,
      "loss": 0.7433,
      "step": 2275
    },
    {
      "epoch": 0.2098950524737631,
      "eval_GEN Loss": 0.4717455208301544,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.40705302357673645,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9230769276618958,
      "eval_runtime": 55.9526,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2275
    },
    {
      "epoch": 0.20998731403529006,
      "grad_norm": 2.07245401836506,
      "learning_rate": 1.1531561195006157e-06,
      "loss": 0.8597,
      "step": 2276
    },
    {
      "epoch": 0.21007957559681698,
      "grad_norm": 1.8762935610064746,
      "learning_rate": 1.1530554960418586e-06,
      "loss": 0.7134,
      "step": 2277
    },
    {
      "epoch": 0.2101718371583439,
      "grad_norm": 2.0624062792250513,
      "learning_rate": 1.1529548247298807e-06,
      "loss": 0.7134,
      "step": 2278
    },
    {
      "epoch": 0.21026409871987084,
      "grad_norm": 1.5135048395455495,
      "learning_rate": 1.152854105573806e-06,
      "loss": 0.5582,
      "step": 2279
    },
    {
      "epoch": 0.21035636028139776,
      "grad_norm": 1.8250532253357739,
      "learning_rate": 1.1527533385827611e-06,
      "loss": 0.6407,
      "step": 2280
    },
    {
      "epoch": 0.21035636028139776,
      "eval_GEN Loss": 0.4707697927951813,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3882807195186615,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9154447317123413,
      "eval_runtime": 56.2476,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 2280
    },
    {
      "epoch": 0.2104486218429247,
      "grad_norm": 2.1871585738069617,
      "learning_rate": 1.1526525237658781e-06,
      "loss": 0.7079,
      "step": 2281
    },
    {
      "epoch": 0.21054088340445162,
      "grad_norm": 4.136395112624742,
      "learning_rate": 1.1525516611322929e-06,
      "loss": 1.043,
      "step": 2282
    },
    {
      "epoch": 0.21063314496597854,
      "grad_norm": 1.7781751336623937,
      "learning_rate": 1.1524507506911457e-06,
      "loss": 0.9093,
      "step": 2283
    },
    {
      "epoch": 0.2107254065275055,
      "grad_norm": 1.468557325789731,
      "learning_rate": 1.1523497924515812e-06,
      "loss": 0.6357,
      "step": 2284
    },
    {
      "epoch": 0.2108176680890324,
      "grad_norm": 4.01144786534884,
      "learning_rate": 1.1522487864227485e-06,
      "loss": 1.1312,
      "step": 2285
    },
    {
      "epoch": 0.2108176680890324,
      "eval_GEN Loss": 0.47120070457458496,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3929106593132019,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9097355604171753,
      "eval_runtime": 55.9853,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2285
    },
    {
      "epoch": 0.21090992965055932,
      "grad_norm": 3.2513916000655785,
      "learning_rate": 1.152147732613801e-06,
      "loss": 0.6032,
      "step": 2286
    },
    {
      "epoch": 0.21100219121208627,
      "grad_norm": 1.460368053452147,
      "learning_rate": 1.1520466310338961e-06,
      "loss": 0.7242,
      "step": 2287
    },
    {
      "epoch": 0.2110944527736132,
      "grad_norm": 2.680184935014927,
      "learning_rate": 1.151945481692196e-06,
      "loss": 0.9445,
      "step": 2288
    },
    {
      "epoch": 0.21118671433514014,
      "grad_norm": 1.9445908018721438,
      "learning_rate": 1.1518442845978668e-06,
      "loss": 0.8572,
      "step": 2289
    },
    {
      "epoch": 0.21127897589666705,
      "grad_norm": 2.2245065211449035,
      "learning_rate": 1.1517430397600794e-06,
      "loss": 0.8897,
      "step": 2290
    },
    {
      "epoch": 0.21127897589666705,
      "eval_GEN Loss": 0.4726479649543762,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.42766568064689636,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9161959290504456,
      "eval_runtime": 56.8666,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2290
    },
    {
      "epoch": 0.21137123745819397,
      "grad_norm": 1.89519577801565,
      "learning_rate": 1.1516417471880083e-06,
      "loss": 0.7204,
      "step": 2291
    },
    {
      "epoch": 0.21146349901972092,
      "grad_norm": 2.519218451131417,
      "learning_rate": 1.1515404068908336e-06,
      "loss": 0.8184,
      "step": 2292
    },
    {
      "epoch": 0.21155576058124784,
      "grad_norm": 2.7804067284655263,
      "learning_rate": 1.151439018877738e-06,
      "loss": 0.9123,
      "step": 2293
    },
    {
      "epoch": 0.21164802214277476,
      "grad_norm": 1.5898426208901604,
      "learning_rate": 1.1513375831579102e-06,
      "loss": 0.7653,
      "step": 2294
    },
    {
      "epoch": 0.2117402837043017,
      "grad_norm": 1.475260349760314,
      "learning_rate": 1.1512360997405423e-06,
      "loss": 0.646,
      "step": 2295
    },
    {
      "epoch": 0.2117402837043017,
      "eval_GEN Loss": 0.47331321239471436,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM Loss": 0.45182713866233826,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9334735870361328,
      "eval_runtime": 56.8636,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2295
    },
    {
      "epoch": 0.21183254526582862,
      "grad_norm": 2.1198426529587864,
      "learning_rate": 1.1511345686348303e-06,
      "loss": 0.7455,
      "step": 2296
    },
    {
      "epoch": 0.21192480682735557,
      "grad_norm": 1.9830299211465507,
      "learning_rate": 1.1510329898499757e-06,
      "loss": 0.9839,
      "step": 2297
    },
    {
      "epoch": 0.21201706838888248,
      "grad_norm": 1.9188414390512405,
      "learning_rate": 1.1509313633951835e-06,
      "loss": 0.8163,
      "step": 2298
    },
    {
      "epoch": 0.2121093299504094,
      "grad_norm": 2.4727749001007933,
      "learning_rate": 1.1508296892796637e-06,
      "loss": 0.8645,
      "step": 2299
    },
    {
      "epoch": 0.21220159151193635,
      "grad_norm": 1.5287857215679712,
      "learning_rate": 1.1507279675126292e-06,
      "loss": 0.7882,
      "step": 2300
    },
    {
      "epoch": 0.21220159151193635,
      "eval_GEN Loss": 0.473702609539032,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM Loss": 0.449951171875,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9365684986114502,
      "eval_runtime": 56.9733,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 2300
    },
    {
      "epoch": 0.21229385307346327,
      "grad_norm": 2.0629844309652077,
      "learning_rate": 1.1506261981032993e-06,
      "loss": 0.7907,
      "step": 2301
    },
    {
      "epoch": 0.2123861146349902,
      "grad_norm": 1.9978895455166346,
      "learning_rate": 1.1505243810608956e-06,
      "loss": 0.7675,
      "step": 2302
    },
    {
      "epoch": 0.21247837619651713,
      "grad_norm": 2.0449103375414825,
      "learning_rate": 1.1504225163946455e-06,
      "loss": 0.947,
      "step": 2303
    },
    {
      "epoch": 0.21257063775804405,
      "grad_norm": 1.7881179619619427,
      "learning_rate": 1.1503206041137798e-06,
      "loss": 0.7616,
      "step": 2304
    },
    {
      "epoch": 0.212662899319571,
      "grad_norm": 1.4617832613411683,
      "learning_rate": 1.1502186442275343e-06,
      "loss": 0.7394,
      "step": 2305
    },
    {
      "epoch": 0.212662899319571,
      "eval_GEN Loss": 0.4710967242717743,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4045696556568146,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9139423370361328,
      "eval_runtime": 56.7701,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2305
    },
    {
      "epoch": 0.21275516088109792,
      "grad_norm": 1.7463932326806049,
      "learning_rate": 1.1501166367451487e-06,
      "loss": 0.6597,
      "step": 2306
    },
    {
      "epoch": 0.21284742244262483,
      "grad_norm": 1.8822327415848314,
      "learning_rate": 1.1500145816758665e-06,
      "loss": 0.8181,
      "step": 2307
    },
    {
      "epoch": 0.21293968400415178,
      "grad_norm": 2.800533451323056,
      "learning_rate": 1.1499124790289366e-06,
      "loss": 0.6983,
      "step": 2308
    },
    {
      "epoch": 0.2130319455656787,
      "grad_norm": 2.5893522569938074,
      "learning_rate": 1.1498103288136117e-06,
      "loss": 0.8804,
      "step": 2309
    },
    {
      "epoch": 0.21312420712720562,
      "grad_norm": 2.004701548344563,
      "learning_rate": 1.1497081310391487e-06,
      "loss": 0.7371,
      "step": 2310
    },
    {
      "epoch": 0.21312420712720562,
      "eval_GEN Loss": 0.4717312753200531,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.38857412338256836,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9256009459495544,
      "eval_runtime": 56.7822,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2310
    },
    {
      "epoch": 0.21321646868873256,
      "grad_norm": 1.915192617305031,
      "learning_rate": 1.1496058857148092e-06,
      "loss": 0.6161,
      "step": 2311
    },
    {
      "epoch": 0.21330873025025948,
      "grad_norm": 2.398367890800583,
      "learning_rate": 1.1495035928498583e-06,
      "loss": 0.6897,
      "step": 2312
    },
    {
      "epoch": 0.21340099181178643,
      "grad_norm": 2.056290982286057,
      "learning_rate": 1.1494012524535663e-06,
      "loss": 0.9089,
      "step": 2313
    },
    {
      "epoch": 0.21349325337331335,
      "grad_norm": 1.5304632400591565,
      "learning_rate": 1.1492988645352076e-06,
      "loss": 0.6597,
      "step": 2314
    },
    {
      "epoch": 0.21358551493484026,
      "grad_norm": 1.6754110826954896,
      "learning_rate": 1.1491964291040603e-06,
      "loss": 0.6768,
      "step": 2315
    },
    {
      "epoch": 0.21358551493484026,
      "eval_GEN Loss": 0.4733254611492157,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4184282124042511,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9504807591438293,
      "eval_runtime": 57.0241,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 2315
    },
    {
      "epoch": 0.2136777764963672,
      "grad_norm": 1.7498588420362595,
      "learning_rate": 1.1490939461694076e-06,
      "loss": 0.6434,
      "step": 2316
    },
    {
      "epoch": 0.21377003805789413,
      "grad_norm": 1.828200640622191,
      "learning_rate": 1.1489914157405366e-06,
      "loss": 0.8032,
      "step": 2317
    },
    {
      "epoch": 0.21386229961942105,
      "grad_norm": 1.83234122485386,
      "learning_rate": 1.1488888378267386e-06,
      "loss": 0.7374,
      "step": 2318
    },
    {
      "epoch": 0.213954561180948,
      "grad_norm": 1.7295731226131061,
      "learning_rate": 1.1487862124373094e-06,
      "loss": 0.7779,
      "step": 2319
    },
    {
      "epoch": 0.2140468227424749,
      "grad_norm": 1.5441769824026699,
      "learning_rate": 1.1486835395815495e-06,
      "loss": 0.6952,
      "step": 2320
    },
    {
      "epoch": 0.2140468227424749,
      "eval_GEN Loss": 0.4737255573272705,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.4278686046600342,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9647836685180664,
      "eval_runtime": 55.8031,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2320
    },
    {
      "epoch": 0.21413908430400186,
      "grad_norm": 1.511777944246195,
      "learning_rate": 1.1485808192687628e-06,
      "loss": 0.5728,
      "step": 2321
    },
    {
      "epoch": 0.21423134586552878,
      "grad_norm": 2.1065699580074853,
      "learning_rate": 1.148478051508258e-06,
      "loss": 0.9463,
      "step": 2322
    },
    {
      "epoch": 0.2143236074270557,
      "grad_norm": 3.2278463260050407,
      "learning_rate": 1.1483752363093483e-06,
      "loss": 0.7803,
      "step": 2323
    },
    {
      "epoch": 0.21441586898858264,
      "grad_norm": 1.5233719435589255,
      "learning_rate": 1.1482723736813511e-06,
      "loss": 0.7441,
      "step": 2324
    },
    {
      "epoch": 0.21450813055010956,
      "grad_norm": 1.5885137468512758,
      "learning_rate": 1.1481694636335874e-06,
      "loss": 0.6965,
      "step": 2325
    },
    {
      "epoch": 0.21450813055010956,
      "eval_GEN Loss": 0.4758618175983429,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.4119797348976135,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9652644395828247,
      "eval_runtime": 55.8377,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2325
    },
    {
      "epoch": 0.21460039211163648,
      "grad_norm": 1.5039896057405835,
      "learning_rate": 1.1480665061753838e-06,
      "loss": 0.7052,
      "step": 2326
    },
    {
      "epoch": 0.21469265367316342,
      "grad_norm": 1.7498110914358747,
      "learning_rate": 1.1479635013160698e-06,
      "loss": 0.7319,
      "step": 2327
    },
    {
      "epoch": 0.21478491523469034,
      "grad_norm": 2.132755590620228,
      "learning_rate": 1.1478604490649802e-06,
      "loss": 0.7302,
      "step": 2328
    },
    {
      "epoch": 0.2148771767962173,
      "grad_norm": 2.5800827059678726,
      "learning_rate": 1.1477573494314536e-06,
      "loss": 0.9263,
      "step": 2329
    },
    {
      "epoch": 0.2149694383577442,
      "grad_norm": 1.5882991923459204,
      "learning_rate": 1.1476542024248335e-06,
      "loss": 0.7195,
      "step": 2330
    },
    {
      "epoch": 0.2149694383577442,
      "eval_GEN Loss": 0.4753049910068512,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.40260228514671326,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.960156261920929,
      "eval_runtime": 55.9356,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2330
    },
    {
      "epoch": 0.21506169991927113,
      "grad_norm": 1.7885748342602523,
      "learning_rate": 1.1475510080544665e-06,
      "loss": 0.7798,
      "step": 2331
    },
    {
      "epoch": 0.21515396148079807,
      "grad_norm": 1.480208733609352,
      "learning_rate": 1.147447766329705e-06,
      "loss": 0.6837,
      "step": 2332
    },
    {
      "epoch": 0.215246223042325,
      "grad_norm": 1.895530242420699,
      "learning_rate": 1.1473444772599045e-06,
      "loss": 0.783,
      "step": 2333
    },
    {
      "epoch": 0.2153384846038519,
      "grad_norm": 2.431044283870173,
      "learning_rate": 1.1472411408544252e-06,
      "loss": 0.6952,
      "step": 2334
    },
    {
      "epoch": 0.21543074616537886,
      "grad_norm": 1.7642384605238612,
      "learning_rate": 1.1471377571226316e-06,
      "loss": 0.6747,
      "step": 2335
    },
    {
      "epoch": 0.21543074616537886,
      "eval_GEN Loss": 0.47444283962249756,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4047582149505615,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9581730961799622,
      "eval_runtime": 55.8988,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2335
    },
    {
      "epoch": 0.21552300772690577,
      "grad_norm": 1.868259521679867,
      "learning_rate": 1.1470343260738928e-06,
      "loss": 0.7325,
      "step": 2336
    },
    {
      "epoch": 0.21561526928843272,
      "grad_norm": 2.4300969412096998,
      "learning_rate": 1.1469308477175817e-06,
      "loss": 0.825,
      "step": 2337
    },
    {
      "epoch": 0.21570753084995964,
      "grad_norm": 1.993711784578015,
      "learning_rate": 1.1468273220630756e-06,
      "loss": 0.7888,
      "step": 2338
    },
    {
      "epoch": 0.21579979241148656,
      "grad_norm": 1.6353502032591918,
      "learning_rate": 1.1467237491197559e-06,
      "loss": 0.7462,
      "step": 2339
    },
    {
      "epoch": 0.2158920539730135,
      "grad_norm": 3.1713808076121976,
      "learning_rate": 1.146620128897009e-06,
      "loss": 0.9708,
      "step": 2340
    },
    {
      "epoch": 0.2158920539730135,
      "eval_GEN Loss": 0.47516658902168274,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.85,
      "eval_PRM F1 AUC": 0.7139863803038241,
      "eval_PRM F1 Neg": 0.5384615384615384,
      "eval_PRM Loss": 0.42991694808006287,
      "eval_PRM NPV": 0.4827586206896552,
      "eval_PRM Precision": 0.8831168831168831,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9573917984962463,
      "eval_runtime": 55.8666,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2340
    },
    {
      "epoch": 0.21598431553454042,
      "grad_norm": 1.9589790408647827,
      "learning_rate": 1.1465164614042251e-06,
      "loss": 0.7606,
      "step": 2341
    },
    {
      "epoch": 0.21607657709606734,
      "grad_norm": 2.0162196837094744,
      "learning_rate": 1.1464127466507987e-06,
      "loss": 0.6981,
      "step": 2342
    },
    {
      "epoch": 0.2161688386575943,
      "grad_norm": 2.2410233024067265,
      "learning_rate": 1.1463089846461283e-06,
      "loss": 0.8722,
      "step": 2343
    },
    {
      "epoch": 0.2162611002191212,
      "grad_norm": 1.8122832657732915,
      "learning_rate": 1.1462051753996172e-06,
      "loss": 0.8332,
      "step": 2344
    },
    {
      "epoch": 0.21635336178064815,
      "grad_norm": 1.8244146059988295,
      "learning_rate": 1.1461013189206728e-06,
      "loss": 0.8179,
      "step": 2345
    },
    {
      "epoch": 0.21635336178064815,
      "eval_GEN Loss": 0.4750309884548187,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8535031847133758,
      "eval_PRM F1 AUC": 0.751440544787847,
      "eval_PRM F1 Neg": 0.5818181818181818,
      "eval_PRM Loss": 0.4460814893245697,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9054054054054054,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9593148827552795,
      "eval_runtime": 55.9823,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2345
    },
    {
      "epoch": 0.21644562334217507,
      "grad_norm": 1.4819709836535895,
      "learning_rate": 1.1459974152187064e-06,
      "loss": 0.6879,
      "step": 2346
    },
    {
      "epoch": 0.216537884903702,
      "grad_norm": 2.5475880117354106,
      "learning_rate": 1.1458934643031344e-06,
      "loss": 0.6995,
      "step": 2347
    },
    {
      "epoch": 0.21663014646522893,
      "grad_norm": 1.7298597891054333,
      "learning_rate": 1.1457894661833767e-06,
      "loss": 0.6828,
      "step": 2348
    },
    {
      "epoch": 0.21672240802675585,
      "grad_norm": 1.5181012040835442,
      "learning_rate": 1.1456854208688578e-06,
      "loss": 0.7195,
      "step": 2349
    },
    {
      "epoch": 0.21681466958828277,
      "grad_norm": 1.906492177727322,
      "learning_rate": 1.1455813283690064e-06,
      "loss": 0.7811,
      "step": 2350
    },
    {
      "epoch": 0.21681466958828277,
      "eval_GEN Loss": 0.47427472472190857,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8535031847133758,
      "eval_PRM F1 AUC": 0.751440544787847,
      "eval_PRM F1 Neg": 0.5818181818181818,
      "eval_PRM Loss": 0.4521903991699219,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9054054054054054,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9660456776618958,
      "eval_runtime": 56.8725,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2350
    },
    {
      "epoch": 0.21690693114980972,
      "grad_norm": 2.0952804451135987,
      "learning_rate": 1.145477188693256e-06,
      "loss": 0.684,
      "step": 2351
    },
    {
      "epoch": 0.21699919271133664,
      "grad_norm": 2.386165354308659,
      "learning_rate": 1.145373001851043e-06,
      "loss": 0.8854,
      "step": 2352
    },
    {
      "epoch": 0.21709145427286355,
      "grad_norm": 1.8874466244002326,
      "learning_rate": 1.1452687678518097e-06,
      "loss": 0.6785,
      "step": 2353
    },
    {
      "epoch": 0.2171837158343905,
      "grad_norm": 2.0608842425595477,
      "learning_rate": 1.1451644867050022e-06,
      "loss": 0.804,
      "step": 2354
    },
    {
      "epoch": 0.21727597739591742,
      "grad_norm": 2.022665852118557,
      "learning_rate": 1.1450601584200696e-06,
      "loss": 0.7699,
      "step": 2355
    },
    {
      "epoch": 0.21727597739591742,
      "eval_GEN Loss": 0.47283825278282166,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.43678930401802063,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9777644276618958,
      "eval_runtime": 56.002,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2355
    },
    {
      "epoch": 0.21736823895744437,
      "grad_norm": 2.0400495938169145,
      "learning_rate": 1.144955783006467e-06,
      "loss": 0.9566,
      "step": 2356
    },
    {
      "epoch": 0.21746050051897128,
      "grad_norm": 2.1740673572995015,
      "learning_rate": 1.144851360473653e-06,
      "loss": 0.8656,
      "step": 2357
    },
    {
      "epoch": 0.2175527620804982,
      "grad_norm": 2.42001296340199,
      "learning_rate": 1.1447468908310904e-06,
      "loss": 0.7008,
      "step": 2358
    },
    {
      "epoch": 0.21764502364202515,
      "grad_norm": 1.9788688615523606,
      "learning_rate": 1.1446423740882464e-06,
      "loss": 0.8722,
      "step": 2359
    },
    {
      "epoch": 0.21773728520355207,
      "grad_norm": 2.5999805764933424,
      "learning_rate": 1.1445378102545926e-06,
      "loss": 0.7861,
      "step": 2360
    },
    {
      "epoch": 0.21773728520355207,
      "eval_GEN Loss": 0.4706652760505676,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.42590969800949097,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9721153974533081,
      "eval_runtime": 56.1296,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 2360
    },
    {
      "epoch": 0.21782954676507899,
      "grad_norm": 1.559569118407366,
      "learning_rate": 1.1444331993396046e-06,
      "loss": 0.7351,
      "step": 2361
    },
    {
      "epoch": 0.21792180832660593,
      "grad_norm": 2.1665835047710367,
      "learning_rate": 1.1443285413527626e-06,
      "loss": 0.8115,
      "step": 2362
    },
    {
      "epoch": 0.21801406988813285,
      "grad_norm": 1.5822610252059668,
      "learning_rate": 1.1442238363035506e-06,
      "loss": 0.7499,
      "step": 2363
    },
    {
      "epoch": 0.2181063314496598,
      "grad_norm": 1.5238898164541468,
      "learning_rate": 1.1441190842014574e-06,
      "loss": 0.7761,
      "step": 2364
    },
    {
      "epoch": 0.21819859301118671,
      "grad_norm": 1.985694895982136,
      "learning_rate": 1.1440142850559758e-06,
      "loss": 0.7319,
      "step": 2365
    },
    {
      "epoch": 0.21819859301118671,
      "eval_GEN Loss": 0.468374639749527,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.4251801371574402,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9659855961799622,
      "eval_runtime": 55.9352,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2365
    },
    {
      "epoch": 0.21829085457271363,
      "grad_norm": 1.808616661632616,
      "learning_rate": 1.1439094388766026e-06,
      "loss": 0.6165,
      "step": 2366
    },
    {
      "epoch": 0.21838311613424058,
      "grad_norm": 1.8714653109491266,
      "learning_rate": 1.1438045456728392e-06,
      "loss": 0.682,
      "step": 2367
    },
    {
      "epoch": 0.2184753776957675,
      "grad_norm": 2.3334762045797404,
      "learning_rate": 1.1436996054541912e-06,
      "loss": 1.0488,
      "step": 2368
    },
    {
      "epoch": 0.21856763925729442,
      "grad_norm": 1.553192305658262,
      "learning_rate": 1.1435946182301686e-06,
      "loss": 0.8035,
      "step": 2369
    },
    {
      "epoch": 0.21865990081882136,
      "grad_norm": 1.6675862371845878,
      "learning_rate": 1.1434895840102856e-06,
      "loss": 0.6578,
      "step": 2370
    },
    {
      "epoch": 0.21865990081882136,
      "eval_GEN Loss": 0.46896758675575256,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.42500072717666626,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9605769515037537,
      "eval_runtime": 55.8721,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2370
    },
    {
      "epoch": 0.21875216238034828,
      "grad_norm": 1.6936454210685636,
      "learning_rate": 1.1433845028040603e-06,
      "loss": 0.7876,
      "step": 2371
    },
    {
      "epoch": 0.21884442394187523,
      "grad_norm": 3.188558427741426,
      "learning_rate": 1.1432793746210152e-06,
      "loss": 0.9185,
      "step": 2372
    },
    {
      "epoch": 0.21893668550340215,
      "grad_norm": 2.0587332732713484,
      "learning_rate": 1.1431741994706776e-06,
      "loss": 0.8059,
      "step": 2373
    },
    {
      "epoch": 0.21902894706492906,
      "grad_norm": 2.838131575363757,
      "learning_rate": 1.1430689773625783e-06,
      "loss": 0.8441,
      "step": 2374
    },
    {
      "epoch": 0.219121208626456,
      "grad_norm": 2.296253791344834,
      "learning_rate": 1.1429637083062528e-06,
      "loss": 0.5491,
      "step": 2375
    },
    {
      "epoch": 0.219121208626456,
      "eval_GEN Loss": 0.4697877764701843,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.41090792417526245,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9453125,
      "eval_runtime": 55.8973,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2375
    },
    {
      "epoch": 0.21921347018798293,
      "grad_norm": 2.1846944804918462,
      "learning_rate": 1.142858392311241e-06,
      "loss": 0.6736,
      "step": 2376
    },
    {
      "epoch": 0.21930573174950985,
      "grad_norm": 1.9274668959793566,
      "learning_rate": 1.1427530293870865e-06,
      "loss": 0.793,
      "step": 2377
    },
    {
      "epoch": 0.2193979933110368,
      "grad_norm": 1.5422715509371179,
      "learning_rate": 1.1426476195433372e-06,
      "loss": 0.684,
      "step": 2378
    },
    {
      "epoch": 0.2194902548725637,
      "grad_norm": 1.502657644965345,
      "learning_rate": 1.142542162789546e-06,
      "loss": 0.6823,
      "step": 2379
    },
    {
      "epoch": 0.21958251643409066,
      "grad_norm": 1.7822182835928675,
      "learning_rate": 1.1424366591352694e-06,
      "loss": 0.7874,
      "step": 2380
    },
    {
      "epoch": 0.21958251643409066,
      "eval_GEN Loss": 0.46932539343833923,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.38606709241867065,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9384615421295166,
      "eval_runtime": 55.9097,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2380
    },
    {
      "epoch": 0.21967477799561758,
      "grad_norm": 1.7523445852540822,
      "learning_rate": 1.1423311085900684e-06,
      "loss": 0.7415,
      "step": 2381
    },
    {
      "epoch": 0.2197670395571445,
      "grad_norm": 1.7913442418646242,
      "learning_rate": 1.142225511163508e-06,
      "loss": 0.8646,
      "step": 2382
    },
    {
      "epoch": 0.21985930111867144,
      "grad_norm": 1.761396358655263,
      "learning_rate": 1.1421198668651579e-06,
      "loss": 0.884,
      "step": 2383
    },
    {
      "epoch": 0.21995156268019836,
      "grad_norm": 2.0539365928270605,
      "learning_rate": 1.1420141757045911e-06,
      "loss": 0.591,
      "step": 2384
    },
    {
      "epoch": 0.22004382424172528,
      "grad_norm": 2.3418539255499304,
      "learning_rate": 1.141908437691386e-06,
      "loss": 0.9255,
      "step": 2385
    },
    {
      "epoch": 0.22004382424172528,
      "eval_GEN Loss": 0.4676920473575592,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3769882023334503,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9442908763885498,
      "eval_runtime": 56.1242,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 2385
    },
    {
      "epoch": 0.22013608580325222,
      "grad_norm": 1.5027133667777839,
      "learning_rate": 1.1418026528351248e-06,
      "loss": 0.6969,
      "step": 2386
    },
    {
      "epoch": 0.22022834736477914,
      "grad_norm": 1.8817816249958745,
      "learning_rate": 1.1416968211453934e-06,
      "loss": 0.8139,
      "step": 2387
    },
    {
      "epoch": 0.2203206089263061,
      "grad_norm": 1.7124071370104497,
      "learning_rate": 1.1415909426317832e-06,
      "loss": 0.8992,
      "step": 2388
    },
    {
      "epoch": 0.220412870487833,
      "grad_norm": 1.978229345626964,
      "learning_rate": 1.1414850173038885e-06,
      "loss": 0.5924,
      "step": 2389
    },
    {
      "epoch": 0.22050513204935993,
      "grad_norm": 1.6016314739029147,
      "learning_rate": 1.1413790451713085e-06,
      "loss": 0.6232,
      "step": 2390
    },
    {
      "epoch": 0.22050513204935993,
      "eval_GEN Loss": 0.46786993741989136,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3706422448158264,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9359976053237915,
      "eval_runtime": 55.9104,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2390
    },
    {
      "epoch": 0.22059739361088687,
      "grad_norm": 2.2803210506354565,
      "learning_rate": 1.1412730262436467e-06,
      "loss": 0.8871,
      "step": 2391
    },
    {
      "epoch": 0.2206896551724138,
      "grad_norm": 1.7584831030043306,
      "learning_rate": 1.1411669605305107e-06,
      "loss": 0.6859,
      "step": 2392
    },
    {
      "epoch": 0.2207819167339407,
      "grad_norm": 2.260883576001255,
      "learning_rate": 1.141060848041512e-06,
      "loss": 0.7366,
      "step": 2393
    },
    {
      "epoch": 0.22087417829546765,
      "grad_norm": 1.6312673244843356,
      "learning_rate": 1.1409546887862671e-06,
      "loss": 0.5583,
      "step": 2394
    },
    {
      "epoch": 0.22096643985699457,
      "grad_norm": 1.6042762461377682,
      "learning_rate": 1.1408484827743963e-06,
      "loss": 0.7177,
      "step": 2395
    },
    {
      "epoch": 0.22096643985699457,
      "eval_GEN Loss": 0.4680069386959076,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.37928250432014465,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9259615540504456,
      "eval_runtime": 56.3175,
      "eval_samples_per_second": 1.154,
      "eval_steps_per_second": 0.089,
      "step": 2395
    },
    {
      "epoch": 0.22105870141852152,
      "grad_norm": 1.527794619670946,
      "learning_rate": 1.140742230015524e-06,
      "loss": 0.8465,
      "step": 2396
    },
    {
      "epoch": 0.22115096298004844,
      "grad_norm": 1.969434610489511,
      "learning_rate": 1.1406359305192789e-06,
      "loss": 0.6828,
      "step": 2397
    },
    {
      "epoch": 0.22124322454157536,
      "grad_norm": 1.5875569813257355,
      "learning_rate": 1.1405295842952944e-06,
      "loss": 0.7225,
      "step": 2398
    },
    {
      "epoch": 0.2213354861031023,
      "grad_norm": 2.4401147860126207,
      "learning_rate": 1.140423191353207e-06,
      "loss": 0.7447,
      "step": 2399
    },
    {
      "epoch": 0.22142774766462922,
      "grad_norm": 1.6427035393739329,
      "learning_rate": 1.1403167517026591e-06,
      "loss": 0.6974,
      "step": 2400
    },
    {
      "epoch": 0.22142774766462922,
      "eval_GEN Loss": 0.46850404143333435,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.38373351097106934,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9254807829856873,
      "eval_runtime": 55.9352,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2400
    },
    {
      "epoch": 0.22152000922615614,
      "grad_norm": 1.7451701920144373,
      "learning_rate": 1.1402102653532961e-06,
      "loss": 0.8668,
      "step": 2401
    },
    {
      "epoch": 0.22161227078768309,
      "grad_norm": 2.151270156869049,
      "learning_rate": 1.140103732314768e-06,
      "loss": 0.8334,
      "step": 2402
    },
    {
      "epoch": 0.22170453234921,
      "grad_norm": 1.6133499639668547,
      "learning_rate": 1.1399971525967285e-06,
      "loss": 0.5765,
      "step": 2403
    },
    {
      "epoch": 0.22179679391073695,
      "grad_norm": 1.2951764413763402,
      "learning_rate": 1.1398905262088366e-06,
      "loss": 0.5789,
      "step": 2404
    },
    {
      "epoch": 0.22188905547226387,
      "grad_norm": 1.9143571737959348,
      "learning_rate": 1.1397838531607548e-06,
      "loss": 0.8101,
      "step": 2405
    },
    {
      "epoch": 0.22188905547226387,
      "eval_GEN Loss": 0.46997690200805664,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3785282075405121,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9268629550933838,
      "eval_runtime": 55.9975,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2405
    },
    {
      "epoch": 0.2219813170337908,
      "grad_norm": 2.1369828854602892,
      "learning_rate": 1.1396771334621499e-06,
      "loss": 0.7544,
      "step": 2406
    },
    {
      "epoch": 0.22207357859531773,
      "grad_norm": 1.3784225486612767,
      "learning_rate": 1.1395703671226932e-06,
      "loss": 0.7144,
      "step": 2407
    },
    {
      "epoch": 0.22216584015684465,
      "grad_norm": 1.6688989208818583,
      "learning_rate": 1.1394635541520598e-06,
      "loss": 0.6259,
      "step": 2408
    },
    {
      "epoch": 0.22225810171837157,
      "grad_norm": 1.8506620044841888,
      "learning_rate": 1.1393566945599293e-06,
      "loss": 0.7966,
      "step": 2409
    },
    {
      "epoch": 0.22235036327989852,
      "grad_norm": 2.238252442399522,
      "learning_rate": 1.1392497883559853e-06,
      "loss": 0.8908,
      "step": 2410
    },
    {
      "epoch": 0.22235036327989852,
      "eval_GEN Loss": 0.4711766839027405,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.37810245156288147,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9131009578704834,
      "eval_runtime": 55.9686,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2410
    },
    {
      "epoch": 0.22244262484142543,
      "grad_norm": 1.6627933985265864,
      "learning_rate": 1.1391428355499163e-06,
      "loss": 0.7435,
      "step": 2411
    },
    {
      "epoch": 0.22253488640295238,
      "grad_norm": 1.6940717176891962,
      "learning_rate": 1.1390358361514143e-06,
      "loss": 0.8278,
      "step": 2412
    },
    {
      "epoch": 0.2226271479644793,
      "grad_norm": 2.7938614253432323,
      "learning_rate": 1.1389287901701757e-06,
      "loss": 0.7211,
      "step": 2413
    },
    {
      "epoch": 0.22271940952600622,
      "grad_norm": 1.690730641332566,
      "learning_rate": 1.1388216976159012e-06,
      "loss": 0.528,
      "step": 2414
    },
    {
      "epoch": 0.22281167108753316,
      "grad_norm": 2.131417765946894,
      "learning_rate": 1.1387145584982956e-06,
      "loss": 0.9121,
      "step": 2415
    },
    {
      "epoch": 0.22281167108753316,
      "eval_GEN Loss": 0.47082820534706116,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.38620659708976746,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.907932698726654,
      "eval_runtime": 56.9834,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 2415
    },
    {
      "epoch": 0.22290393264906008,
      "grad_norm": 1.665691644810356,
      "learning_rate": 1.1386073728270682e-06,
      "loss": 0.734,
      "step": 2416
    },
    {
      "epoch": 0.222996194210587,
      "grad_norm": 1.6187270870803299,
      "learning_rate": 1.1385001406119322e-06,
      "loss": 0.6109,
      "step": 2417
    },
    {
      "epoch": 0.22308845577211395,
      "grad_norm": 1.9602103537950144,
      "learning_rate": 1.1383928618626052e-06,
      "loss": 0.7229,
      "step": 2418
    },
    {
      "epoch": 0.22318071733364087,
      "grad_norm": 2.0472504753973286,
      "learning_rate": 1.138285536588809e-06,
      "loss": 0.7685,
      "step": 2419
    },
    {
      "epoch": 0.2232729788951678,
      "grad_norm": 2.481401439851354,
      "learning_rate": 1.1381781648002695e-06,
      "loss": 0.7794,
      "step": 2420
    },
    {
      "epoch": 0.2232729788951678,
      "eval_GEN Loss": 0.4718153476715088,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.39808425307273865,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9167668223381042,
      "eval_runtime": 56.8258,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2420
    },
    {
      "epoch": 0.22336524045669473,
      "grad_norm": 2.152102989463276,
      "learning_rate": 1.138070746506717e-06,
      "loss": 0.9804,
      "step": 2421
    },
    {
      "epoch": 0.22345750201822165,
      "grad_norm": 1.9955954711734134,
      "learning_rate": 1.1379632817178859e-06,
      "loss": 0.8168,
      "step": 2422
    },
    {
      "epoch": 0.2235497635797486,
      "grad_norm": 1.599732125977681,
      "learning_rate": 1.1378557704435147e-06,
      "loss": 0.5893,
      "step": 2423
    },
    {
      "epoch": 0.2236420251412755,
      "grad_norm": 2.055255500617136,
      "learning_rate": 1.1377482126933463e-06,
      "loss": 0.891,
      "step": 2424
    },
    {
      "epoch": 0.22373428670280243,
      "grad_norm": 2.0444524277785803,
      "learning_rate": 1.137640608477128e-06,
      "loss": 0.7909,
      "step": 2425
    },
    {
      "epoch": 0.22373428670280243,
      "eval_GEN Loss": 0.4728257954120636,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.4107135236263275,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9212139248847961,
      "eval_runtime": 56.9273,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2425
    },
    {
      "epoch": 0.22382654826432938,
      "grad_norm": 1.6192087375379063,
      "learning_rate": 1.1375329578046105e-06,
      "loss": 0.8534,
      "step": 2426
    },
    {
      "epoch": 0.2239188098258563,
      "grad_norm": 2.132404090605897,
      "learning_rate": 1.1374252606855498e-06,
      "loss": 0.7363,
      "step": 2427
    },
    {
      "epoch": 0.22401107138738324,
      "grad_norm": 2.1744719112179864,
      "learning_rate": 1.1373175171297055e-06,
      "loss": 0.8048,
      "step": 2428
    },
    {
      "epoch": 0.22410333294891016,
      "grad_norm": 1.6479747531841462,
      "learning_rate": 1.1372097271468413e-06,
      "loss": 0.9014,
      "step": 2429
    },
    {
      "epoch": 0.22419559451043708,
      "grad_norm": 1.7625154033683774,
      "learning_rate": 1.1371018907467256e-06,
      "loss": 0.855,
      "step": 2430
    },
    {
      "epoch": 0.22419559451043708,
      "eval_GEN Loss": 0.47299402952194214,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8589743589743589,
      "eval_PRM F1 AUC": 0.7731796752226295,
      "eval_PRM F1 Neg": 0.6071428571428571,
      "eval_PRM Loss": 0.43708139657974243,
      "eval_PRM NPV": 0.5151515151515151,
      "eval_PRM Precision": 0.9178082191780822,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9305288195610046,
      "eval_runtime": 56.8176,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2430
    },
    {
      "epoch": 0.22428785607196403,
      "grad_norm": 2.3010725350981294,
      "learning_rate": 1.1369940079391303e-06,
      "loss": 0.9764,
      "step": 2431
    },
    {
      "epoch": 0.22438011763349094,
      "grad_norm": 2.0818426827837357,
      "learning_rate": 1.136886078733832e-06,
      "loss": 0.719,
      "step": 2432
    },
    {
      "epoch": 0.22447237919501786,
      "grad_norm": 1.7541157290491092,
      "learning_rate": 1.1367781031406119e-06,
      "loss": 0.7939,
      "step": 2433
    },
    {
      "epoch": 0.2245646407565448,
      "grad_norm": 3.2390284510728558,
      "learning_rate": 1.1366700811692542e-06,
      "loss": 0.8502,
      "step": 2434
    },
    {
      "epoch": 0.22465690231807173,
      "grad_norm": 2.3228636799465607,
      "learning_rate": 1.1365620128295484e-06,
      "loss": 0.698,
      "step": 2435
    },
    {
      "epoch": 0.22465690231807173,
      "eval_GEN Loss": 0.47129055857658386,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8516129032258064,
      "eval_PRM F1 AUC": 0.7671555788370874,
      "eval_PRM F1 Neg": 0.5964912280701754,
      "eval_PRM Loss": 0.4372565746307373,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9166666666666666,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9247596263885498,
      "eval_runtime": 56.8818,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2435
    },
    {
      "epoch": 0.22474916387959867,
      "grad_norm": 2.425877722634202,
      "learning_rate": 1.1364538981312879e-06,
      "loss": 0.6756,
      "step": 2436
    },
    {
      "epoch": 0.2248414254411256,
      "grad_norm": 2.361565498188788,
      "learning_rate": 1.13634573708427e-06,
      "loss": 0.656,
      "step": 2437
    },
    {
      "epoch": 0.2249336870026525,
      "grad_norm": 1.6936773950644115,
      "learning_rate": 1.1362375296982966e-06,
      "loss": 0.7373,
      "step": 2438
    },
    {
      "epoch": 0.22502594856417946,
      "grad_norm": 2.388291506864521,
      "learning_rate": 1.1361292759831738e-06,
      "loss": 0.7895,
      "step": 2439
    },
    {
      "epoch": 0.22511821012570638,
      "grad_norm": 2.4285604544608903,
      "learning_rate": 1.1360209759487112e-06,
      "loss": 0.8819,
      "step": 2440
    },
    {
      "epoch": 0.22511821012570638,
      "eval_GEN Loss": 0.4684344232082367,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.36903172731399536,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9042067527770996,
      "eval_runtime": 56.941,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2440
    },
    {
      "epoch": 0.2252104716872333,
      "grad_norm": 1.9214486009218754,
      "learning_rate": 1.1359126296047235e-06,
      "loss": 0.7505,
      "step": 2441
    },
    {
      "epoch": 0.22530273324876024,
      "grad_norm": 1.4059886334090042,
      "learning_rate": 1.1358042369610289e-06,
      "loss": 0.6201,
      "step": 2442
    },
    {
      "epoch": 0.22539499481028716,
      "grad_norm": 1.3026528887742264,
      "learning_rate": 1.1356957980274504e-06,
      "loss": 0.5657,
      "step": 2443
    },
    {
      "epoch": 0.2254872563718141,
      "grad_norm": 2.1622229210816104,
      "learning_rate": 1.135587312813815e-06,
      "loss": 0.6993,
      "step": 2444
    },
    {
      "epoch": 0.22557951793334102,
      "grad_norm": 1.5415557007956615,
      "learning_rate": 1.1354787813299536e-06,
      "loss": 0.6928,
      "step": 2445
    },
    {
      "epoch": 0.22557951793334102,
      "eval_GEN Loss": 0.469711035490036,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3571073114871979,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9175480604171753,
      "eval_runtime": 56.7865,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2445
    },
    {
      "epoch": 0.22567177949486794,
      "grad_norm": 1.3915882131267747,
      "learning_rate": 1.1353702035857017e-06,
      "loss": 0.7732,
      "step": 2446
    },
    {
      "epoch": 0.2257640410563949,
      "grad_norm": 2.878040973623829,
      "learning_rate": 1.1352615795908983e-06,
      "loss": 0.8507,
      "step": 2447
    },
    {
      "epoch": 0.2258563026179218,
      "grad_norm": 1.8932740611767567,
      "learning_rate": 1.1351529093553876e-06,
      "loss": 0.6389,
      "step": 2448
    },
    {
      "epoch": 0.22594856417944872,
      "grad_norm": 1.8058416639527972,
      "learning_rate": 1.1350441928890171e-06,
      "loss": 0.7784,
      "step": 2449
    },
    {
      "epoch": 0.22604082574097567,
      "grad_norm": 1.3383035069198335,
      "learning_rate": 1.134935430201639e-06,
      "loss": 0.6901,
      "step": 2450
    },
    {
      "epoch": 0.22604082574097567,
      "eval_GEN Loss": 0.4697313904762268,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3636733591556549,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9028846025466919,
      "eval_runtime": 56.8545,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2450
    },
    {
      "epoch": 0.2261330873025026,
      "grad_norm": 1.9378913317121984,
      "learning_rate": 1.1348266213031097e-06,
      "loss": 0.7741,
      "step": 2451
    },
    {
      "epoch": 0.22622534886402954,
      "grad_norm": 2.034692154429518,
      "learning_rate": 1.1347177662032894e-06,
      "loss": 0.8802,
      "step": 2452
    },
    {
      "epoch": 0.22631761042555645,
      "grad_norm": 1.5020953975044546,
      "learning_rate": 1.134608864912043e-06,
      "loss": 0.5675,
      "step": 2453
    },
    {
      "epoch": 0.22640987198708337,
      "grad_norm": 1.618387271348409,
      "learning_rate": 1.1344999174392388e-06,
      "loss": 0.7853,
      "step": 2454
    },
    {
      "epoch": 0.22650213354861032,
      "grad_norm": 1.4706788156565047,
      "learning_rate": 1.1343909237947504e-06,
      "loss": 0.634,
      "step": 2455
    },
    {
      "epoch": 0.22650213354861032,
      "eval_GEN Loss": 0.4696446359157562,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.36343905329704285,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8920673131942749,
      "eval_runtime": 56.955,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 2455
    },
    {
      "epoch": 0.22659439511013724,
      "grad_norm": 2.202525362172754,
      "learning_rate": 1.1342818839884548e-06,
      "loss": 0.6124,
      "step": 2456
    },
    {
      "epoch": 0.22668665667166416,
      "grad_norm": 2.554850650073834,
      "learning_rate": 1.134172798030233e-06,
      "loss": 0.9653,
      "step": 2457
    },
    {
      "epoch": 0.2267789182331911,
      "grad_norm": 1.410417704172557,
      "learning_rate": 1.1340636659299707e-06,
      "loss": 0.6578,
      "step": 2458
    },
    {
      "epoch": 0.22687117979471802,
      "grad_norm": 2.234493248027032,
      "learning_rate": 1.133954487697558e-06,
      "loss": 0.9514,
      "step": 2459
    },
    {
      "epoch": 0.22696344135624497,
      "grad_norm": 2.0401035606875633,
      "learning_rate": 1.1338452633428883e-06,
      "loss": 0.5441,
      "step": 2460
    },
    {
      "epoch": 0.22696344135624497,
      "eval_GEN Loss": 0.4691702723503113,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3740851879119873,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8882812261581421,
      "eval_runtime": 56.7323,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2460
    },
    {
      "epoch": 0.22705570291777188,
      "grad_norm": 1.7006937438662142,
      "learning_rate": 1.13373599287586e-06,
      "loss": 0.8534,
      "step": 2461
    },
    {
      "epoch": 0.2271479644792988,
      "grad_norm": 1.3400719713400748,
      "learning_rate": 1.1336266763063752e-06,
      "loss": 0.5771,
      "step": 2462
    },
    {
      "epoch": 0.22724022604082575,
      "grad_norm": 1.9203227243680183,
      "learning_rate": 1.1335173136443406e-06,
      "loss": 0.6458,
      "step": 2463
    },
    {
      "epoch": 0.22733248760235267,
      "grad_norm": 2.1478223804179666,
      "learning_rate": 1.1334079048996662e-06,
      "loss": 0.9295,
      "step": 2464
    },
    {
      "epoch": 0.22742474916387959,
      "grad_norm": 1.8610640901338107,
      "learning_rate": 1.1332984500822676e-06,
      "loss": 0.8251,
      "step": 2465
    },
    {
      "epoch": 0.22742474916387959,
      "eval_GEN Loss": 0.46802860498428345,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3707696497440338,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8885817527770996,
      "eval_runtime": 57.1267,
      "eval_samples_per_second": 1.138,
      "eval_steps_per_second": 0.088,
      "step": 2465
    },
    {
      "epoch": 0.22751701072540653,
      "grad_norm": 1.824966759349408,
      "learning_rate": 1.1331889492020634e-06,
      "loss": 0.8261,
      "step": 2466
    },
    {
      "epoch": 0.22760927228693345,
      "grad_norm": 1.4536672716986374,
      "learning_rate": 1.1330794022689764e-06,
      "loss": 0.6273,
      "step": 2467
    },
    {
      "epoch": 0.2277015338484604,
      "grad_norm": 2.0134294555857015,
      "learning_rate": 1.1329698092929345e-06,
      "loss": 0.8769,
      "step": 2468
    },
    {
      "epoch": 0.22779379540998732,
      "grad_norm": 1.9646142059554255,
      "learning_rate": 1.1328601702838688e-06,
      "loss": 0.7944,
      "step": 2469
    },
    {
      "epoch": 0.22788605697151423,
      "grad_norm": 1.5708958311354917,
      "learning_rate": 1.1327504852517152e-06,
      "loss": 0.6498,
      "step": 2470
    },
    {
      "epoch": 0.22788605697151423,
      "eval_GEN Loss": 0.4714735448360443,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.397542804479599,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9019230604171753,
      "eval_runtime": 56.9012,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2470
    },
    {
      "epoch": 0.22797831853304118,
      "grad_norm": 1.3681737730047887,
      "learning_rate": 1.1326407542064132e-06,
      "loss": 0.497,
      "step": 2471
    },
    {
      "epoch": 0.2280705800945681,
      "grad_norm": 1.97168956436627,
      "learning_rate": 1.1325309771579075e-06,
      "loss": 0.7414,
      "step": 2472
    },
    {
      "epoch": 0.22816284165609502,
      "grad_norm": 2.3031316138910194,
      "learning_rate": 1.1324211541161454e-06,
      "loss": 0.855,
      "step": 2473
    },
    {
      "epoch": 0.22825510321762196,
      "grad_norm": 2.1849471605093425,
      "learning_rate": 1.1323112850910794e-06,
      "loss": 0.7153,
      "step": 2474
    },
    {
      "epoch": 0.22834736477914888,
      "grad_norm": 2.230071753179375,
      "learning_rate": 1.1322013700926665e-06,
      "loss": 0.7911,
      "step": 2475
    },
    {
      "epoch": 0.22834736477914888,
      "eval_GEN Loss": 0.47294795513153076,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8589743589743589,
      "eval_PRM F1 AUC": 0.7731796752226295,
      "eval_PRM F1 Neg": 0.6071428571428571,
      "eval_PRM Loss": 0.4190768599510193,
      "eval_PRM NPV": 0.5151515151515151,
      "eval_PRM Precision": 0.9178082191780822,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9164663553237915,
      "eval_runtime": 56.9171,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2475
    },
    {
      "epoch": 0.22843962634067583,
      "grad_norm": 1.982015033997405,
      "learning_rate": 1.1320914091308672e-06,
      "loss": 0.816,
      "step": 2476
    },
    {
      "epoch": 0.22853188790220275,
      "grad_norm": 1.6435419692673323,
      "learning_rate": 1.1319814022156461e-06,
      "loss": 0.6268,
      "step": 2477
    },
    {
      "epoch": 0.22862414946372966,
      "grad_norm": 1.5840100620675945,
      "learning_rate": 1.1318713493569724e-06,
      "loss": 0.7047,
      "step": 2478
    },
    {
      "epoch": 0.2287164110252566,
      "grad_norm": 2.0689373763994925,
      "learning_rate": 1.131761250564819e-06,
      "loss": 0.6856,
      "step": 2479
    },
    {
      "epoch": 0.22880867258678353,
      "grad_norm": 2.643783420100991,
      "learning_rate": 1.1316511058491635e-06,
      "loss": 0.874,
      "step": 2480
    },
    {
      "epoch": 0.22880867258678353,
      "eval_GEN Loss": 0.47014084458351135,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8407643312101911,
      "eval_PRM F1 AUC": 0.7236773179675223,
      "eval_PRM F1 Neg": 0.5454545454545454,
      "eval_PRM Loss": 0.419371634721756,
      "eval_PRM NPV": 0.46875,
      "eval_PRM Precision": 0.8918918918918919,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9097355604171753,
      "eval_runtime": 56.6946,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2480
    },
    {
      "epoch": 0.22890093414831045,
      "grad_norm": 1.4845504230546336,
      "learning_rate": 1.1315409152199875e-06,
      "loss": 0.6945,
      "step": 2481
    },
    {
      "epoch": 0.2289931957098374,
      "grad_norm": 1.3983407481004397,
      "learning_rate": 1.1314306786872764e-06,
      "loss": 0.6221,
      "step": 2482
    },
    {
      "epoch": 0.2290854572713643,
      "grad_norm": 1.5308073226456993,
      "learning_rate": 1.13132039626102e-06,
      "loss": 0.7287,
      "step": 2483
    },
    {
      "epoch": 0.22917771883289126,
      "grad_norm": 1.6513885081879696,
      "learning_rate": 1.1312100679512123e-06,
      "loss": 0.7563,
      "step": 2484
    },
    {
      "epoch": 0.22926998039441818,
      "grad_norm": 1.637595673797351,
      "learning_rate": 1.1310996937678514e-06,
      "loss": 0.6138,
      "step": 2485
    },
    {
      "epoch": 0.22926998039441818,
      "eval_GEN Loss": 0.46911635994911194,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.3957677185535431,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9058894515037537,
      "eval_runtime": 56.8024,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2485
    },
    {
      "epoch": 0.2293622419559451,
      "grad_norm": 2.141175588712988,
      "learning_rate": 1.1309892737209398e-06,
      "loss": 0.6231,
      "step": 2486
    },
    {
      "epoch": 0.22945450351747204,
      "grad_norm": 3.5286845366659665,
      "learning_rate": 1.1308788078204837e-06,
      "loss": 0.7897,
      "step": 2487
    },
    {
      "epoch": 0.22954676507899896,
      "grad_norm": 2.3754392821441126,
      "learning_rate": 1.1307682960764937e-06,
      "loss": 0.9385,
      "step": 2488
    },
    {
      "epoch": 0.22963902664052588,
      "grad_norm": 2.075474213820218,
      "learning_rate": 1.1306577384989848e-06,
      "loss": 0.7251,
      "step": 2489
    },
    {
      "epoch": 0.22973128820205282,
      "grad_norm": 1.573113395506502,
      "learning_rate": 1.1305471350979755e-06,
      "loss": 0.6645,
      "step": 2490
    },
    {
      "epoch": 0.22973128820205282,
      "eval_GEN Loss": 0.4681898057460785,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.38737961649894714,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9082331657409668,
      "eval_runtime": 56.8407,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2490
    },
    {
      "epoch": 0.22982354976357974,
      "grad_norm": 1.7327869966982956,
      "learning_rate": 1.1304364858834894e-06,
      "loss": 0.5968,
      "step": 2491
    },
    {
      "epoch": 0.2299158113251067,
      "grad_norm": 1.546646575464143,
      "learning_rate": 1.1303257908655529e-06,
      "loss": 0.5915,
      "step": 2492
    },
    {
      "epoch": 0.2300080728866336,
      "grad_norm": 1.6666122378775579,
      "learning_rate": 1.130215050054198e-06,
      "loss": 0.793,
      "step": 2493
    },
    {
      "epoch": 0.23010033444816053,
      "grad_norm": 2.4377169421469485,
      "learning_rate": 1.1301042634594602e-06,
      "loss": 0.574,
      "step": 2494
    },
    {
      "epoch": 0.23019259600968747,
      "grad_norm": 1.894979391573054,
      "learning_rate": 1.129993431091379e-06,
      "loss": 0.6417,
      "step": 2495
    },
    {
      "epoch": 0.23019259600968747,
      "eval_GEN Loss": 0.4674574136734009,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7200104766893662,
      "eval_PRM F1 Neg": 0.5490196078431373,
      "eval_PRM Loss": 0.3983570337295532,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8846153846153846,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9028846025466919,
      "eval_runtime": 56.7877,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2495
    },
    {
      "epoch": 0.2302848575712144,
      "grad_norm": 2.3448281942223703,
      "learning_rate": 1.1298825529599983e-06,
      "loss": 0.8902,
      "step": 2496
    },
    {
      "epoch": 0.2303771191327413,
      "grad_norm": 1.7789743158712312,
      "learning_rate": 1.1297716290753657e-06,
      "loss": 0.6988,
      "step": 2497
    },
    {
      "epoch": 0.23046938069426826,
      "grad_norm": 1.518338773618905,
      "learning_rate": 1.1296606594475337e-06,
      "loss": 0.4898,
      "step": 2498
    },
    {
      "epoch": 0.23056164225579517,
      "grad_norm": 2.0993593440032865,
      "learning_rate": 1.1295496440865583e-06,
      "loss": 0.7423,
      "step": 2499
    },
    {
      "epoch": 0.23065390381732212,
      "grad_norm": 2.3951483049527615,
      "learning_rate": 1.1294385830025e-06,
      "loss": 0.7233,
      "step": 2500
    },
    {
      "epoch": 0.23065390381732212,
      "eval_GEN Loss": 0.468417227268219,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4037969708442688,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8917668461799622,
      "eval_runtime": 56.7159,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2500
    },
    {
      "epoch": 0.23074616537884904,
      "grad_norm": 2.212728059817675,
      "learning_rate": 1.1293274762054233e-06,
      "loss": 0.8199,
      "step": 2501
    },
    {
      "epoch": 0.23083842694037596,
      "grad_norm": 2.6001781670563138,
      "learning_rate": 1.129216323705397e-06,
      "loss": 1.0106,
      "step": 2502
    },
    {
      "epoch": 0.2309306885019029,
      "grad_norm": 1.6724983686291832,
      "learning_rate": 1.1291051255124937e-06,
      "loss": 0.7285,
      "step": 2503
    },
    {
      "epoch": 0.23102295006342982,
      "grad_norm": 2.6611634155114245,
      "learning_rate": 1.1289938816367903e-06,
      "loss": 0.6679,
      "step": 2504
    },
    {
      "epoch": 0.23111521162495674,
      "grad_norm": 2.067860591355526,
      "learning_rate": 1.1288825920883685e-06,
      "loss": 0.8011,
      "step": 2505
    },
    {
      "epoch": 0.23111521162495674,
      "eval_GEN Loss": 0.46988198161125183,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.38036635518074036,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8838942050933838,
      "eval_runtime": 56.0063,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2505
    },
    {
      "epoch": 0.2312074731864837,
      "grad_norm": 1.7813382417817287,
      "learning_rate": 1.1287712568773127e-06,
      "loss": 0.7122,
      "step": 2506
    },
    {
      "epoch": 0.2312997347480106,
      "grad_norm": 2.059914193923338,
      "learning_rate": 1.1286598760137126e-06,
      "loss": 0.796,
      "step": 2507
    },
    {
      "epoch": 0.23139199630953755,
      "grad_norm": 2.3880312965231476,
      "learning_rate": 1.128548449507662e-06,
      "loss": 0.831,
      "step": 2508
    },
    {
      "epoch": 0.23148425787106447,
      "grad_norm": 1.8289964252066055,
      "learning_rate": 1.1284369773692581e-06,
      "loss": 0.6621,
      "step": 2509
    },
    {
      "epoch": 0.2315765194325914,
      "grad_norm": 1.472177833646282,
      "learning_rate": 1.1283254596086031e-06,
      "loss": 0.655,
      "step": 2510
    },
    {
      "epoch": 0.2315765194325914,
      "eval_GEN Loss": 0.4698483645915985,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.362417072057724,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8776442408561707,
      "eval_runtime": 55.6801,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 2510
    },
    {
      "epoch": 0.23166878099411833,
      "grad_norm": 1.5573238422258162,
      "learning_rate": 1.1282138962358026e-06,
      "loss": 0.6579,
      "step": 2511
    },
    {
      "epoch": 0.23176104255564525,
      "grad_norm": 1.7342550230597888,
      "learning_rate": 1.1281022872609668e-06,
      "loss": 0.7244,
      "step": 2512
    },
    {
      "epoch": 0.23185330411717217,
      "grad_norm": 1.676035826200078,
      "learning_rate": 1.1279906326942097e-06,
      "loss": 0.4781,
      "step": 2513
    },
    {
      "epoch": 0.23194556567869912,
      "grad_norm": 1.630407863274102,
      "learning_rate": 1.12787893254565e-06,
      "loss": 0.6057,
      "step": 2514
    },
    {
      "epoch": 0.23203782724022604,
      "grad_norm": 2.1285004105290954,
      "learning_rate": 1.1277671868254097e-06,
      "loss": 0.6775,
      "step": 2515
    },
    {
      "epoch": 0.23203782724022604,
      "eval_GEN Loss": 0.4699053168296814,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.355672687292099,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8748798370361328,
      "eval_runtime": 55.7112,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 2515
    },
    {
      "epoch": 0.23213008880175298,
      "grad_norm": 1.9006796685899903,
      "learning_rate": 1.1276553955436155e-06,
      "loss": 0.7582,
      "step": 2516
    },
    {
      "epoch": 0.2322223503632799,
      "grad_norm": 1.582411726868668,
      "learning_rate": 1.1275435587103983e-06,
      "loss": 0.6562,
      "step": 2517
    },
    {
      "epoch": 0.23231461192480682,
      "grad_norm": 2.2044870251500774,
      "learning_rate": 1.1274316763358927e-06,
      "loss": 0.5809,
      "step": 2518
    },
    {
      "epoch": 0.23240687348633376,
      "grad_norm": 2.2165562061612465,
      "learning_rate": 1.1273197484302377e-06,
      "loss": 0.7458,
      "step": 2519
    },
    {
      "epoch": 0.23249913504786068,
      "grad_norm": 2.065663963935201,
      "learning_rate": 1.1272077750035767e-06,
      "loss": 0.723,
      "step": 2520
    },
    {
      "epoch": 0.23249913504786068,
      "eval_GEN Loss": 0.46835580468177795,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.354735404253006,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8722355961799622,
      "eval_runtime": 55.774,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2520
    },
    {
      "epoch": 0.2325913966093876,
      "grad_norm": 2.313014337748344,
      "learning_rate": 1.1270957560660563e-06,
      "loss": 0.8063,
      "step": 2521
    },
    {
      "epoch": 0.23268365817091455,
      "grad_norm": 3.635399712281739,
      "learning_rate": 1.1269836916278284e-06,
      "loss": 0.8438,
      "step": 2522
    },
    {
      "epoch": 0.23277591973244147,
      "grad_norm": 1.927833728108103,
      "learning_rate": 1.1268715816990483e-06,
      "loss": 0.7967,
      "step": 2523
    },
    {
      "epoch": 0.2328681812939684,
      "grad_norm": 1.9774224623083527,
      "learning_rate": 1.1267594262898754e-06,
      "loss": 0.7281,
      "step": 2524
    },
    {
      "epoch": 0.23296044285549533,
      "grad_norm": 1.4849486701625527,
      "learning_rate": 1.1266472254104735e-06,
      "loss": 0.7488,
      "step": 2525
    },
    {
      "epoch": 0.23296044285549533,
      "eval_GEN Loss": 0.46815571188926697,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.39794889092445374,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8714542984962463,
      "eval_runtime": 56.7078,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2525
    },
    {
      "epoch": 0.23305270441702225,
      "grad_norm": 2.052214522194372,
      "learning_rate": 1.1265349790710102e-06,
      "loss": 0.6389,
      "step": 2526
    },
    {
      "epoch": 0.2331449659785492,
      "grad_norm": 1.8405345488527973,
      "learning_rate": 1.126422687281658e-06,
      "loss": 0.7683,
      "step": 2527
    },
    {
      "epoch": 0.23323722754007611,
      "grad_norm": 1.7569126218126419,
      "learning_rate": 1.1263103500525927e-06,
      "loss": 0.6766,
      "step": 2528
    },
    {
      "epoch": 0.23332948910160303,
      "grad_norm": 2.2530742113486104,
      "learning_rate": 1.1261979673939942e-06,
      "loss": 0.9568,
      "step": 2529
    },
    {
      "epoch": 0.23342175066312998,
      "grad_norm": 2.326025848352322,
      "learning_rate": 1.1260855393160471e-06,
      "loss": 0.7399,
      "step": 2530
    },
    {
      "epoch": 0.23342175066312998,
      "eval_GEN Loss": 0.4674154818058014,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.41416069865226746,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8833533525466919,
      "eval_runtime": 56.7922,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2530
    },
    {
      "epoch": 0.2335140122246569,
      "grad_norm": 1.6557122109170073,
      "learning_rate": 1.1259730658289396e-06,
      "loss": 0.5562,
      "step": 2531
    },
    {
      "epoch": 0.23360627378618384,
      "grad_norm": 1.442127422746816,
      "learning_rate": 1.1258605469428643e-06,
      "loss": 0.718,
      "step": 2532
    },
    {
      "epoch": 0.23369853534771076,
      "grad_norm": 2.168232468969206,
      "learning_rate": 1.125747982668018e-06,
      "loss": 0.9217,
      "step": 2533
    },
    {
      "epoch": 0.23379079690923768,
      "grad_norm": 1.8672615603219742,
      "learning_rate": 1.1256353730146011e-06,
      "loss": 0.8414,
      "step": 2534
    },
    {
      "epoch": 0.23388305847076463,
      "grad_norm": 2.6723027112594555,
      "learning_rate": 1.125522717992819e-06,
      "loss": 0.9236,
      "step": 2535
    },
    {
      "epoch": 0.23388305847076463,
      "eval_GEN Loss": 0.4678378701210022,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4094238579273224,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8882211446762085,
      "eval_runtime": 57.3303,
      "eval_samples_per_second": 1.134,
      "eval_steps_per_second": 0.087,
      "step": 2535
    },
    {
      "epoch": 0.23397532003229154,
      "grad_norm": 2.5711676954810976,
      "learning_rate": 1.12541001761288e-06,
      "loss": 0.8562,
      "step": 2536
    },
    {
      "epoch": 0.23406758159381846,
      "grad_norm": 1.49570424655103,
      "learning_rate": 1.125297271884998e-06,
      "loss": 0.695,
      "step": 2537
    },
    {
      "epoch": 0.2341598431553454,
      "grad_norm": 3.4188011926568325,
      "learning_rate": 1.1251844808193895e-06,
      "loss": 0.8133,
      "step": 2538
    },
    {
      "epoch": 0.23425210471687233,
      "grad_norm": 2.772937083538868,
      "learning_rate": 1.125071644426276e-06,
      "loss": 0.8046,
      "step": 2539
    },
    {
      "epoch": 0.23434436627839927,
      "grad_norm": 1.4829643971961517,
      "learning_rate": 1.124958762715883e-06,
      "loss": 0.5224,
      "step": 2540
    },
    {
      "epoch": 0.23434436627839927,
      "eval_GEN Loss": 0.46604621410369873,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.39644289016723633,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8786057829856873,
      "eval_runtime": 56.8861,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2540
    },
    {
      "epoch": 0.2344366278399262,
      "grad_norm": 1.6763105657342332,
      "learning_rate": 1.1248458356984398e-06,
      "loss": 0.6448,
      "step": 2541
    },
    {
      "epoch": 0.2345288894014531,
      "grad_norm": 1.7961755994124575,
      "learning_rate": 1.1247328633841806e-06,
      "loss": 0.6953,
      "step": 2542
    },
    {
      "epoch": 0.23462115096298006,
      "grad_norm": 1.6026609778166998,
      "learning_rate": 1.1246198457833423e-06,
      "loss": 0.5873,
      "step": 2543
    },
    {
      "epoch": 0.23471341252450698,
      "grad_norm": 1.8692006613808,
      "learning_rate": 1.1245067829061675e-06,
      "loss": 0.9321,
      "step": 2544
    },
    {
      "epoch": 0.2348056740860339,
      "grad_norm": 1.6154700578899546,
      "learning_rate": 1.1243936747629017e-06,
      "loss": 0.6672,
      "step": 2545
    },
    {
      "epoch": 0.2348056740860339,
      "eval_GEN Loss": 0.4653279781341553,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.3707880973815918,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8753004670143127,
      "eval_runtime": 56.8048,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2545
    },
    {
      "epoch": 0.23489793564756084,
      "grad_norm": 1.9530299830184554,
      "learning_rate": 1.124280521363795e-06,
      "loss": 0.8186,
      "step": 2546
    },
    {
      "epoch": 0.23499019720908776,
      "grad_norm": 2.0571922856662845,
      "learning_rate": 1.1241673227191016e-06,
      "loss": 0.7337,
      "step": 2547
    },
    {
      "epoch": 0.2350824587706147,
      "grad_norm": 1.953611207398242,
      "learning_rate": 1.1240540788390798e-06,
      "loss": 0.7153,
      "step": 2548
    },
    {
      "epoch": 0.23517472033214162,
      "grad_norm": 1.9533462404809976,
      "learning_rate": 1.1239407897339915e-06,
      "loss": 0.5951,
      "step": 2549
    },
    {
      "epoch": 0.23526698189366854,
      "grad_norm": 4.451501371893454,
      "learning_rate": 1.1238274554141037e-06,
      "loss": 0.9408,
      "step": 2550
    },
    {
      "epoch": 0.23526698189366854,
      "eval_GEN Loss": 0.4650691747665405,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3606521785259247,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8812500238418579,
      "eval_runtime": 57.4661,
      "eval_samples_per_second": 1.131,
      "eval_steps_per_second": 0.087,
      "step": 2550
    },
    {
      "epoch": 0.2353592434551955,
      "grad_norm": 2.2577344799305603,
      "learning_rate": 1.123714075889687e-06,
      "loss": 0.7341,
      "step": 2551
    },
    {
      "epoch": 0.2354515050167224,
      "grad_norm": 1.935955057499266,
      "learning_rate": 1.1236006511710154e-06,
      "loss": 0.8388,
      "step": 2552
    },
    {
      "epoch": 0.23554376657824933,
      "grad_norm": 1.904958157934222,
      "learning_rate": 1.123487181268368e-06,
      "loss": 0.6223,
      "step": 2553
    },
    {
      "epoch": 0.23563602813977627,
      "grad_norm": 2.4034295050685004,
      "learning_rate": 1.1233736661920278e-06,
      "loss": 0.7952,
      "step": 2554
    },
    {
      "epoch": 0.2357282897013032,
      "grad_norm": 3.2878261423784254,
      "learning_rate": 1.1232601059522814e-06,
      "loss": 0.8085,
      "step": 2555
    },
    {
      "epoch": 0.2357282897013032,
      "eval_GEN Loss": 0.46591365337371826,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.3971644639968872,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8887619972229004,
      "eval_runtime": 56.7474,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2555
    },
    {
      "epoch": 0.23582055126283014,
      "grad_norm": 2.413711180678156,
      "learning_rate": 1.12314650055942e-06,
      "loss": 0.7785,
      "step": 2556
    },
    {
      "epoch": 0.23591281282435705,
      "grad_norm": 1.578646643660678,
      "learning_rate": 1.1230328500237386e-06,
      "loss": 0.7172,
      "step": 2557
    },
    {
      "epoch": 0.23600507438588397,
      "grad_norm": 1.7671334029133194,
      "learning_rate": 1.1229191543555363e-06,
      "loss": 0.85,
      "step": 2558
    },
    {
      "epoch": 0.23609733594741092,
      "grad_norm": 2.0075239914500527,
      "learning_rate": 1.1228054135651164e-06,
      "loss": 0.7113,
      "step": 2559
    },
    {
      "epoch": 0.23618959750893784,
      "grad_norm": 2.8377217672071002,
      "learning_rate": 1.1226916276627866e-06,
      "loss": 1.0561,
      "step": 2560
    },
    {
      "epoch": 0.23618959750893784,
      "eval_GEN Loss": 0.46483534574508667,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.42682313919067383,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9055288434028625,
      "eval_runtime": 56.6655,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 2560
    },
    {
      "epoch": 0.23628185907046476,
      "grad_norm": 1.7992243342898053,
      "learning_rate": 1.1225777966588582e-06,
      "loss": 0.7715,
      "step": 2561
    },
    {
      "epoch": 0.2363741206319917,
      "grad_norm": 1.5924146515434936,
      "learning_rate": 1.1224639205636465e-06,
      "loss": 0.7376,
      "step": 2562
    },
    {
      "epoch": 0.23646638219351862,
      "grad_norm": 2.5928456411016514,
      "learning_rate": 1.1223499993874712e-06,
      "loss": 0.7723,
      "step": 2563
    },
    {
      "epoch": 0.23655864375504557,
      "grad_norm": 2.432199496348407,
      "learning_rate": 1.1222360331406564e-06,
      "loss": 0.8685,
      "step": 2564
    },
    {
      "epoch": 0.23665090531657249,
      "grad_norm": 1.3741751143949394,
      "learning_rate": 1.1221220218335293e-06,
      "loss": 0.6125,
      "step": 2565
    },
    {
      "epoch": 0.23665090531657249,
      "eval_GEN Loss": 0.4643940031528473,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.41885513067245483,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9077523946762085,
      "eval_runtime": 57.0156,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 2565
    },
    {
      "epoch": 0.2367431668780994,
      "grad_norm": 1.4487965371124216,
      "learning_rate": 1.1220079654764224e-06,
      "loss": 0.7297,
      "step": 2566
    },
    {
      "epoch": 0.23683542843962635,
      "grad_norm": 2.158227852156897,
      "learning_rate": 1.121893864079671e-06,
      "loss": 0.7006,
      "step": 2567
    },
    {
      "epoch": 0.23692769000115327,
      "grad_norm": 2.543070925070056,
      "learning_rate": 1.121779717653616e-06,
      "loss": 0.8689,
      "step": 2568
    },
    {
      "epoch": 0.2370199515626802,
      "grad_norm": 1.3737590732668938,
      "learning_rate": 1.1216655262086008e-06,
      "loss": 0.6451,
      "step": 2569
    },
    {
      "epoch": 0.23711221312420713,
      "grad_norm": 2.0399396054569614,
      "learning_rate": 1.1215512897549738e-06,
      "loss": 0.8344,
      "step": 2570
    },
    {
      "epoch": 0.23711221312420713,
      "eval_GEN Loss": 0.46495646238327026,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.3905562162399292,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9021634459495544,
      "eval_runtime": 56.7579,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2570
    },
    {
      "epoch": 0.23720447468573405,
      "grad_norm": 1.9272540358650754,
      "learning_rate": 1.1214370083030874e-06,
      "loss": 0.6872,
      "step": 2571
    },
    {
      "epoch": 0.237296736247261,
      "grad_norm": 2.0792286041612966,
      "learning_rate": 1.1213226818632979e-06,
      "loss": 0.9057,
      "step": 2572
    },
    {
      "epoch": 0.23738899780878792,
      "grad_norm": 1.7602523207797398,
      "learning_rate": 1.1212083104459656e-06,
      "loss": 0.7097,
      "step": 2573
    },
    {
      "epoch": 0.23748125937031483,
      "grad_norm": 2.7092024520314406,
      "learning_rate": 1.1210938940614554e-06,
      "loss": 1.0002,
      "step": 2574
    },
    {
      "epoch": 0.23757352093184178,
      "grad_norm": 2.213368663770566,
      "learning_rate": 1.1209794327201354e-06,
      "loss": 0.9512,
      "step": 2575
    },
    {
      "epoch": 0.23757352093184178,
      "eval_GEN Loss": 0.46629077196121216,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.3744640648365021,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9001802802085876,
      "eval_runtime": 57.154,
      "eval_samples_per_second": 1.137,
      "eval_steps_per_second": 0.087,
      "step": 2575
    },
    {
      "epoch": 0.2376657824933687,
      "grad_norm": 2.2257461915290966,
      "learning_rate": 1.1208649264323787e-06,
      "loss": 0.7274,
      "step": 2576
    },
    {
      "epoch": 0.23775804405489562,
      "grad_norm": 2.0154960156775914,
      "learning_rate": 1.120750375208562e-06,
      "loss": 0.7006,
      "step": 2577
    },
    {
      "epoch": 0.23785030561642256,
      "grad_norm": 1.8514038743508354,
      "learning_rate": 1.1206357790590657e-06,
      "loss": 0.6589,
      "step": 2578
    },
    {
      "epoch": 0.23794256717794948,
      "grad_norm": 1.812508552994726,
      "learning_rate": 1.120521137994275e-06,
      "loss": 0.6732,
      "step": 2579
    },
    {
      "epoch": 0.23803482873947643,
      "grad_norm": 1.7726608492956808,
      "learning_rate": 1.1204064520245788e-06,
      "loss": 0.7692,
      "step": 2580
    },
    {
      "epoch": 0.23803482873947643,
      "eval_GEN Loss": 0.4679611623287201,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3795567750930786,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8996995091438293,
      "eval_runtime": 56.7412,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2580
    },
    {
      "epoch": 0.23812709030100335,
      "grad_norm": 1.9729873787395944,
      "learning_rate": 1.1202917211603703e-06,
      "loss": 0.8437,
      "step": 2581
    },
    {
      "epoch": 0.23821935186253027,
      "grad_norm": 1.8435487070001206,
      "learning_rate": 1.1201769454120464e-06,
      "loss": 0.7759,
      "step": 2582
    },
    {
      "epoch": 0.2383116134240572,
      "grad_norm": 2.0289952064123127,
      "learning_rate": 1.1200621247900083e-06,
      "loss": 0.6987,
      "step": 2583
    },
    {
      "epoch": 0.23840387498558413,
      "grad_norm": 2.0958208286160525,
      "learning_rate": 1.119947259304661e-06,
      "loss": 0.8447,
      "step": 2584
    },
    {
      "epoch": 0.23849613654711105,
      "grad_norm": 1.9227664424655957,
      "learning_rate": 1.1198323489664143e-06,
      "loss": 0.685,
      "step": 2585
    },
    {
      "epoch": 0.23849613654711105,
      "eval_GEN Loss": 0.4695647358894348,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4069802165031433,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9093149304389954,
      "eval_runtime": 55.9755,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2585
    },
    {
      "epoch": 0.238588398108638,
      "grad_norm": 2.6346044059746117,
      "learning_rate": 1.1197173937856812e-06,
      "loss": 0.8283,
      "step": 2586
    },
    {
      "epoch": 0.2386806596701649,
      "grad_norm": 2.2779477792534593,
      "learning_rate": 1.119602393772879e-06,
      "loss": 0.7428,
      "step": 2587
    },
    {
      "epoch": 0.23877292123169186,
      "grad_norm": 2.657376334867618,
      "learning_rate": 1.1194873489384294e-06,
      "loss": 0.874,
      "step": 2588
    },
    {
      "epoch": 0.23886518279321878,
      "grad_norm": 1.428451456630398,
      "learning_rate": 1.119372259292758e-06,
      "loss": 0.783,
      "step": 2589
    },
    {
      "epoch": 0.2389574443547457,
      "grad_norm": 1.778757753593459,
      "learning_rate": 1.1192571248462943e-06,
      "loss": 0.7061,
      "step": 2590
    },
    {
      "epoch": 0.2389574443547457,
      "eval_GEN Loss": 0.4712638258934021,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7888947092718702,
      "eval_PRM F1 Neg": 0.6206896551724138,
      "eval_PRM Loss": 0.45570430159568787,
      "eval_PRM NPV": 0.5142857142857142,
      "eval_PRM Precision": 0.9295774647887324,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9371995329856873,
      "eval_runtime": 55.6965,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 2590
    },
    {
      "epoch": 0.23904970591627264,
      "grad_norm": 1.735152051528127,
      "learning_rate": 1.1191419456094718e-06,
      "loss": 0.6781,
      "step": 2591
    },
    {
      "epoch": 0.23914196747779956,
      "grad_norm": 2.7383848006946643,
      "learning_rate": 1.1190267215927287e-06,
      "loss": 0.7268,
      "step": 2592
    },
    {
      "epoch": 0.23923422903932648,
      "grad_norm": 1.9042119975874754,
      "learning_rate": 1.1189114528065063e-06,
      "loss": 0.7572,
      "step": 2593
    },
    {
      "epoch": 0.23932649060085343,
      "grad_norm": 2.088345537741797,
      "learning_rate": 1.1187961392612506e-06,
      "loss": 0.7748,
      "step": 2594
    },
    {
      "epoch": 0.23941875216238034,
      "grad_norm": 2.757349713158077,
      "learning_rate": 1.1186807809674115e-06,
      "loss": 0.6148,
      "step": 2595
    },
    {
      "epoch": 0.23941875216238034,
      "eval_GEN Loss": 0.4711546003818512,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8571428571428571,
      "eval_PRM F1 AUC": 0.7888947092718702,
      "eval_PRM F1 Neg": 0.6206896551724138,
      "eval_PRM Loss": 0.45487111806869507,
      "eval_PRM NPV": 0.5142857142857142,
      "eval_PRM Precision": 0.9295774647887324,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9346153736114502,
      "eval_runtime": 56.6127,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 2595
    },
    {
      "epoch": 0.2395110137239073,
      "grad_norm": 1.808332122177196,
      "learning_rate": 1.118565377935443e-06,
      "loss": 0.7489,
      "step": 2596
    },
    {
      "epoch": 0.2396032752854342,
      "grad_norm": 1.417073802622359,
      "learning_rate": 1.1184499301758032e-06,
      "loss": 0.5061,
      "step": 2597
    },
    {
      "epoch": 0.23969553684696113,
      "grad_norm": 1.9466946699476475,
      "learning_rate": 1.1183344376989538e-06,
      "loss": 0.7785,
      "step": 2598
    },
    {
      "epoch": 0.23978779840848807,
      "grad_norm": 2.303712894332619,
      "learning_rate": 1.1182189005153612e-06,
      "loss": 0.9928,
      "step": 2599
    },
    {
      "epoch": 0.239880059970015,
      "grad_norm": 2.699353816427195,
      "learning_rate": 1.1181033186354957e-06,
      "loss": 0.7598,
      "step": 2600
    },
    {
      "epoch": 0.239880059970015,
      "eval_GEN Loss": 0.4719969928264618,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4061139225959778,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9103365540504456,
      "eval_runtime": 55.8192,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2600
    },
    {
      "epoch": 0.2399723215315419,
      "grad_norm": 1.3226400337027335,
      "learning_rate": 1.117987692069831e-06,
      "loss": 0.6029,
      "step": 2601
    },
    {
      "epoch": 0.24006458309306886,
      "grad_norm": 1.8055411209427468,
      "learning_rate": 1.117872020828846e-06,
      "loss": 0.8511,
      "step": 2602
    },
    {
      "epoch": 0.24015684465459577,
      "grad_norm": 2.523980707739512,
      "learning_rate": 1.1177563049230224e-06,
      "loss": 0.8335,
      "step": 2603
    },
    {
      "epoch": 0.24024910621612272,
      "grad_norm": 2.6537671700111254,
      "learning_rate": 1.117640544362847e-06,
      "loss": 0.8793,
      "step": 2604
    },
    {
      "epoch": 0.24034136777764964,
      "grad_norm": 2.6910494722363367,
      "learning_rate": 1.1175247391588097e-06,
      "loss": 0.6902,
      "step": 2605
    },
    {
      "epoch": 0.24034136777764964,
      "eval_GEN Loss": 0.4716975688934326,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3727628290653229,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9069110751152039,
      "eval_runtime": 55.7799,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2605
    },
    {
      "epoch": 0.24043362933917656,
      "grad_norm": 1.7087716381305673,
      "learning_rate": 1.1174088893214056e-06,
      "loss": 0.5984,
      "step": 2606
    },
    {
      "epoch": 0.2405258909007035,
      "grad_norm": 1.6349844486450962,
      "learning_rate": 1.117292994861133e-06,
      "loss": 0.751,
      "step": 2607
    },
    {
      "epoch": 0.24061815246223042,
      "grad_norm": 2.257379986588271,
      "learning_rate": 1.1171770557884942e-06,
      "loss": 0.6488,
      "step": 2608
    },
    {
      "epoch": 0.24071041402375734,
      "grad_norm": 1.4797117470245391,
      "learning_rate": 1.1170610721139957e-06,
      "loss": 0.6943,
      "step": 2609
    },
    {
      "epoch": 0.2408026755852843,
      "grad_norm": 3.203300606249343,
      "learning_rate": 1.1169450438481486e-06,
      "loss": 0.7245,
      "step": 2610
    },
    {
      "epoch": 0.2408026755852843,
      "eval_GEN Loss": 0.47212257981300354,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3651526868343353,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9094951748847961,
      "eval_runtime": 55.776,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2610
    },
    {
      "epoch": 0.2408949371468112,
      "grad_norm": 2.082562661039052,
      "learning_rate": 1.1168289710014673e-06,
      "loss": 0.6943,
      "step": 2611
    },
    {
      "epoch": 0.24098719870833815,
      "grad_norm": 1.3461386274799059,
      "learning_rate": 1.1167128535844705e-06,
      "loss": 0.7148,
      "step": 2612
    },
    {
      "epoch": 0.24107946026986507,
      "grad_norm": 1.7466275712776993,
      "learning_rate": 1.1165966916076807e-06,
      "loss": 0.8748,
      "step": 2613
    },
    {
      "epoch": 0.241171721831392,
      "grad_norm": 1.4778306964718693,
      "learning_rate": 1.116480485081625e-06,
      "loss": 0.6777,
      "step": 2614
    },
    {
      "epoch": 0.24126398339291893,
      "grad_norm": 1.5605011569521596,
      "learning_rate": 1.1163642340168342e-06,
      "loss": 0.696,
      "step": 2615
    },
    {
      "epoch": 0.24126398339291893,
      "eval_GEN Loss": 0.4711157977581024,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37030795216560364,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9044471383094788,
      "eval_runtime": 55.8132,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2615
    },
    {
      "epoch": 0.24135624495444585,
      "grad_norm": 2.2301855181418073,
      "learning_rate": 1.116247938423843e-06,
      "loss": 0.7758,
      "step": 2616
    },
    {
      "epoch": 0.24144850651597277,
      "grad_norm": 2.111904500075404,
      "learning_rate": 1.1161315983131907e-06,
      "loss": 0.6658,
      "step": 2617
    },
    {
      "epoch": 0.24154076807749972,
      "grad_norm": 1.5123236063854808,
      "learning_rate": 1.1160152136954198e-06,
      "loss": 0.7382,
      "step": 2618
    },
    {
      "epoch": 0.24163302963902664,
      "grad_norm": 1.7181890502465444,
      "learning_rate": 1.1158987845810771e-06,
      "loss": 0.622,
      "step": 2619
    },
    {
      "epoch": 0.24172529120055358,
      "grad_norm": 1.7510919721248044,
      "learning_rate": 1.115782310980714e-06,
      "loss": 0.7016,
      "step": 2620
    },
    {
      "epoch": 0.24172529120055358,
      "eval_GEN Loss": 0.47017866373062134,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37671229243278503,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9063100814819336,
      "eval_runtime": 55.8576,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2620
    },
    {
      "epoch": 0.2418175527620805,
      "grad_norm": 1.8378109776070277,
      "learning_rate": 1.1156657929048855e-06,
      "loss": 0.7029,
      "step": 2621
    },
    {
      "epoch": 0.24190981432360742,
      "grad_norm": 2.321164617975305,
      "learning_rate": 1.1155492303641503e-06,
      "loss": 0.692,
      "step": 2622
    },
    {
      "epoch": 0.24200207588513437,
      "grad_norm": 2.664759457326069,
      "learning_rate": 1.1154326233690718e-06,
      "loss": 0.7498,
      "step": 2623
    },
    {
      "epoch": 0.24209433744666128,
      "grad_norm": 2.6055498615935666,
      "learning_rate": 1.115315971930217e-06,
      "loss": 0.9187,
      "step": 2624
    },
    {
      "epoch": 0.2421865990081882,
      "grad_norm": 1.9355539873623118,
      "learning_rate": 1.1151992760581573e-06,
      "loss": 0.609,
      "step": 2625
    },
    {
      "epoch": 0.2421865990081882,
      "eval_GEN Loss": 0.47074583172798157,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.3882634937763214,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9135817289352417,
      "eval_runtime": 55.6479,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 2625
    },
    {
      "epoch": 0.24227886056971515,
      "grad_norm": 2.944471657945971,
      "learning_rate": 1.1150825357634672e-06,
      "loss": 0.8762,
      "step": 2626
    },
    {
      "epoch": 0.24237112213124207,
      "grad_norm": 2.200905130844562,
      "learning_rate": 1.1149657510567265e-06,
      "loss": 0.6878,
      "step": 2627
    },
    {
      "epoch": 0.242463383692769,
      "grad_norm": 2.28866226747744,
      "learning_rate": 1.1148489219485183e-06,
      "loss": 0.6965,
      "step": 2628
    },
    {
      "epoch": 0.24255564525429593,
      "grad_norm": 1.824038878062565,
      "learning_rate": 1.1147320484494299e-06,
      "loss": 0.7871,
      "step": 2629
    },
    {
      "epoch": 0.24264790681582285,
      "grad_norm": 1.7826891608940783,
      "learning_rate": 1.114615130570052e-06,
      "loss": 0.742,
      "step": 2630
    },
    {
      "epoch": 0.24264790681582285,
      "eval_GEN Loss": 0.4712772071361542,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.39751917123794556,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9146033525466919,
      "eval_runtime": 56.0668,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 2630
    },
    {
      "epoch": 0.2427401683773498,
      "grad_norm": 1.3443934187863702,
      "learning_rate": 1.1144981683209808e-06,
      "loss": 0.5388,
      "step": 2631
    },
    {
      "epoch": 0.24283242993887671,
      "grad_norm": 1.9350901119904973,
      "learning_rate": 1.1143811617128147e-06,
      "loss": 0.7367,
      "step": 2632
    },
    {
      "epoch": 0.24292469150040363,
      "grad_norm": 2.0198018563883497,
      "learning_rate": 1.1142641107561576e-06,
      "loss": 0.6642,
      "step": 2633
    },
    {
      "epoch": 0.24301695306193058,
      "grad_norm": 1.7564955951482246,
      "learning_rate": 1.1141470154616165e-06,
      "loss": 0.7376,
      "step": 2634
    },
    {
      "epoch": 0.2431092146234575,
      "grad_norm": 2.2912139567429244,
      "learning_rate": 1.1140298758398032e-06,
      "loss": 0.7932,
      "step": 2635
    },
    {
      "epoch": 0.2431092146234575,
      "eval_GEN Loss": 0.4723615348339081,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8641975308641975,
      "eval_PRM F1 AUC": 0.7260345730749084,
      "eval_PRM F1 Neg": 0.56,
      "eval_PRM Loss": 0.4061235785484314,
      "eval_PRM NPV": 0.5185185185185185,
      "eval_PRM Precision": 0.8860759493670886,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9222956895828247,
      "eval_runtime": 55.6989,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 2635
    },
    {
      "epoch": 0.24320147618498444,
      "grad_norm": 2.6350049683050263,
      "learning_rate": 1.1139126919013327e-06,
      "loss": 0.7523,
      "step": 2636
    },
    {
      "epoch": 0.24329373774651136,
      "grad_norm": 2.6376997258443313,
      "learning_rate": 1.1137954636568243e-06,
      "loss": 0.9642,
      "step": 2637
    },
    {
      "epoch": 0.24338599930803828,
      "grad_norm": 1.5119685825236164,
      "learning_rate": 1.1136781911169018e-06,
      "loss": 0.5982,
      "step": 2638
    },
    {
      "epoch": 0.24347826086956523,
      "grad_norm": 1.713111181227738,
      "learning_rate": 1.1135608742921925e-06,
      "loss": 0.6589,
      "step": 2639
    },
    {
      "epoch": 0.24357052243109215,
      "grad_norm": 1.8382115876879144,
      "learning_rate": 1.1134435131933276e-06,
      "loss": 0.7064,
      "step": 2640
    },
    {
      "epoch": 0.24357052243109215,
      "eval_GEN Loss": 0.47442442178726196,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.41116347908973694,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9266225695610046,
      "eval_runtime": 56.8658,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2640
    },
    {
      "epoch": 0.24366278399261906,
      "grad_norm": 1.517008188562679,
      "learning_rate": 1.1133261078309429e-06,
      "loss": 0.6288,
      "step": 2641
    },
    {
      "epoch": 0.243755045554146,
      "grad_norm": 1.6833885463981062,
      "learning_rate": 1.1132086582156775e-06,
      "loss": 0.8063,
      "step": 2642
    },
    {
      "epoch": 0.24384730711567293,
      "grad_norm": 1.9506178692626712,
      "learning_rate": 1.1130911643581752e-06,
      "loss": 0.9221,
      "step": 2643
    },
    {
      "epoch": 0.24393956867719988,
      "grad_norm": 1.711158475549289,
      "learning_rate": 1.1129736262690833e-06,
      "loss": 0.7147,
      "step": 2644
    },
    {
      "epoch": 0.2440318302387268,
      "grad_norm": 2.244195921152605,
      "learning_rate": 1.112856043959053e-06,
      "loss": 0.7238,
      "step": 2645
    },
    {
      "epoch": 0.2440318302387268,
      "eval_GEN Loss": 0.47486352920532227,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.40899932384490967,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9401442408561707,
      "eval_runtime": 56.5917,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 2645
    },
    {
      "epoch": 0.2441240918002537,
      "grad_norm": 1.6849552676106292,
      "learning_rate": 1.1127384174387405e-06,
      "loss": 0.5591,
      "step": 2646
    },
    {
      "epoch": 0.24421635336178066,
      "grad_norm": 2.903197551076138,
      "learning_rate": 1.1126207467188047e-06,
      "loss": 0.9402,
      "step": 2647
    },
    {
      "epoch": 0.24430861492330758,
      "grad_norm": 2.36574246101209,
      "learning_rate": 1.1125030318099092e-06,
      "loss": 0.7487,
      "step": 2648
    },
    {
      "epoch": 0.2444008764848345,
      "grad_norm": 1.5393744570309666,
      "learning_rate": 1.1123852727227218e-06,
      "loss": 0.8016,
      "step": 2649
    },
    {
      "epoch": 0.24449313804636144,
      "grad_norm": 1.868008692844655,
      "learning_rate": 1.1122674694679137e-06,
      "loss": 0.6475,
      "step": 2650
    },
    {
      "epoch": 0.24449313804636144,
      "eval_GEN Loss": 0.4760426878929138,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4073770344257355,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9487980604171753,
      "eval_runtime": 56.8027,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2650
    },
    {
      "epoch": 0.24458539960788836,
      "grad_norm": 2.734933908646507,
      "learning_rate": 1.1121496220561605e-06,
      "loss": 0.985,
      "step": 2651
    },
    {
      "epoch": 0.2446776611694153,
      "grad_norm": 2.0199063915080253,
      "learning_rate": 1.112031730498142e-06,
      "loss": 0.7998,
      "step": 2652
    },
    {
      "epoch": 0.24476992273094222,
      "grad_norm": 1.7830298994819749,
      "learning_rate": 1.1119137948045413e-06,
      "loss": 0.8221,
      "step": 2653
    },
    {
      "epoch": 0.24486218429246914,
      "grad_norm": 3.3780892989823603,
      "learning_rate": 1.1117958149860465e-06,
      "loss": 0.9432,
      "step": 2654
    },
    {
      "epoch": 0.2449544458539961,
      "grad_norm": 2.0519511551282648,
      "learning_rate": 1.1116777910533484e-06,
      "loss": 0.7193,
      "step": 2655
    },
    {
      "epoch": 0.2449544458539961,
      "eval_GEN Loss": 0.47731730341911316,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.43130752444267273,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9389122724533081,
      "eval_runtime": 55.7727,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2655
    },
    {
      "epoch": 0.245046707415523,
      "grad_norm": 1.7867197941851463,
      "learning_rate": 1.1115597230171427e-06,
      "loss": 0.6895,
      "step": 2656
    },
    {
      "epoch": 0.24513896897704993,
      "grad_norm": 1.8496498318116188,
      "learning_rate": 1.111441610888129e-06,
      "loss": 0.7608,
      "step": 2657
    },
    {
      "epoch": 0.24523123053857687,
      "grad_norm": 2.9975046480207412,
      "learning_rate": 1.111323454677011e-06,
      "loss": 0.6447,
      "step": 2658
    },
    {
      "epoch": 0.2453234921001038,
      "grad_norm": 2.435834494320346,
      "learning_rate": 1.1112052543944963e-06,
      "loss": 0.8602,
      "step": 2659
    },
    {
      "epoch": 0.24541575366163074,
      "grad_norm": 1.5755190483418144,
      "learning_rate": 1.1110870100512958e-06,
      "loss": 0.531,
      "step": 2660
    },
    {
      "epoch": 0.24541575366163074,
      "eval_GEN Loss": 0.4768896996974945,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.43830370903015137,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.934495210647583,
      "eval_runtime": 55.8348,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2660
    },
    {
      "epoch": 0.24550801522315766,
      "grad_norm": 1.7338215646981916,
      "learning_rate": 1.1109687216581255e-06,
      "loss": 0.666,
      "step": 2661
    },
    {
      "epoch": 0.24560027678468457,
      "grad_norm": 2.60583301671582,
      "learning_rate": 1.1108503892257048e-06,
      "loss": 0.7092,
      "step": 2662
    },
    {
      "epoch": 0.24569253834621152,
      "grad_norm": 2.6521807938255275,
      "learning_rate": 1.1107320127647567e-06,
      "loss": 1.0368,
      "step": 2663
    },
    {
      "epoch": 0.24578479990773844,
      "grad_norm": 1.9397235093993084,
      "learning_rate": 1.1106135922860093e-06,
      "loss": 0.8096,
      "step": 2664
    },
    {
      "epoch": 0.24587706146926536,
      "grad_norm": 1.4033374000064165,
      "learning_rate": 1.1104951278001937e-06,
      "loss": 0.7326,
      "step": 2665
    },
    {
      "epoch": 0.24587706146926536,
      "eval_GEN Loss": 0.4765397012233734,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4285373389720917,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9275240302085876,
      "eval_runtime": 55.7565,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 2665
    },
    {
      "epoch": 0.2459693230307923,
      "grad_norm": 2.164335109214383,
      "learning_rate": 1.1103766193180456e-06,
      "loss": 0.8181,
      "step": 2666
    },
    {
      "epoch": 0.24606158459231922,
      "grad_norm": 2.2196893197651812,
      "learning_rate": 1.1102580668503043e-06,
      "loss": 0.7924,
      "step": 2667
    },
    {
      "epoch": 0.24615384615384617,
      "grad_norm": 2.993252069894412,
      "learning_rate": 1.1101394704077132e-06,
      "loss": 0.9642,
      "step": 2668
    },
    {
      "epoch": 0.24624610771537309,
      "grad_norm": 2.27327253053909,
      "learning_rate": 1.1100208300010195e-06,
      "loss": 0.8564,
      "step": 2669
    },
    {
      "epoch": 0.2463383692769,
      "grad_norm": 2.3821446459764504,
      "learning_rate": 1.1099021456409748e-06,
      "loss": 0.7544,
      "step": 2670
    },
    {
      "epoch": 0.2463383692769,
      "eval_GEN Loss": 0.4759509563446045,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.42510223388671875,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9228966236114502,
      "eval_runtime": 56.0279,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2670
    },
    {
      "epoch": 0.24643063083842695,
      "grad_norm": 1.7964367007397872,
      "learning_rate": 1.1097834173383344e-06,
      "loss": 0.7891,
      "step": 2671
    },
    {
      "epoch": 0.24652289239995387,
      "grad_norm": 1.7203876762332726,
      "learning_rate": 1.1096646451038578e-06,
      "loss": 0.6537,
      "step": 2672
    },
    {
      "epoch": 0.2466151539614808,
      "grad_norm": 2.3866574552580597,
      "learning_rate": 1.109545828948308e-06,
      "loss": 0.7328,
      "step": 2673
    },
    {
      "epoch": 0.24670741552300773,
      "grad_norm": 1.7511228802379015,
      "learning_rate": 1.1094269688824526e-06,
      "loss": 0.8012,
      "step": 2674
    },
    {
      "epoch": 0.24679967708453465,
      "grad_norm": 1.6420429647856971,
      "learning_rate": 1.1093080649170626e-06,
      "loss": 0.7032,
      "step": 2675
    },
    {
      "epoch": 0.24679967708453465,
      "eval_GEN Loss": 0.4752283990383148,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.410010427236557,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.921875,
      "eval_runtime": 56.7526,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2675
    },
    {
      "epoch": 0.2468919386460616,
      "grad_norm": 2.92803193347504,
      "learning_rate": 1.1091891170629137e-06,
      "loss": 0.8658,
      "step": 2676
    },
    {
      "epoch": 0.24698420020758852,
      "grad_norm": 1.2843911845081415,
      "learning_rate": 1.1090701253307848e-06,
      "loss": 0.6142,
      "step": 2677
    },
    {
      "epoch": 0.24707646176911544,
      "grad_norm": 1.7815618811696259,
      "learning_rate": 1.1089510897314591e-06,
      "loss": 0.6174,
      "step": 2678
    },
    {
      "epoch": 0.24716872333064238,
      "grad_norm": 1.6848760870756567,
      "learning_rate": 1.108832010275724e-06,
      "loss": 0.7942,
      "step": 2679
    },
    {
      "epoch": 0.2472609848921693,
      "grad_norm": 2.0652980233879488,
      "learning_rate": 1.1087128869743704e-06,
      "loss": 0.8283,
      "step": 2680
    },
    {
      "epoch": 0.2472609848921693,
      "eval_GEN Loss": 0.4764530658721924,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.39405161142349243,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9256610870361328,
      "eval_runtime": 55.736,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 2680
    },
    {
      "epoch": 0.24735324645369622,
      "grad_norm": 1.4959808394336118,
      "learning_rate": 1.1085937198381938e-06,
      "loss": 0.7071,
      "step": 2681
    },
    {
      "epoch": 0.24744550801522316,
      "grad_norm": 1.5492094935392113,
      "learning_rate": 1.1084745088779932e-06,
      "loss": 0.6962,
      "step": 2682
    },
    {
      "epoch": 0.24753776957675008,
      "grad_norm": 1.5212333996460836,
      "learning_rate": 1.1083552541045715e-06,
      "loss": 0.6939,
      "step": 2683
    },
    {
      "epoch": 0.24763003113827703,
      "grad_norm": 1.6342928925896765,
      "learning_rate": 1.1082359555287359e-06,
      "loss": 0.7901,
      "step": 2684
    },
    {
      "epoch": 0.24772229269980395,
      "grad_norm": 2.1132965754143966,
      "learning_rate": 1.1081166131612971e-06,
      "loss": 0.8711,
      "step": 2685
    },
    {
      "epoch": 0.24772229269980395,
      "eval_GEN Loss": 0.47772204875946045,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3826387822628021,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9354567527770996,
      "eval_runtime": 55.8083,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2685
    },
    {
      "epoch": 0.24781455426133087,
      "grad_norm": 1.8408714340145624,
      "learning_rate": 1.1079972270130706e-06,
      "loss": 0.725,
      "step": 2686
    },
    {
      "epoch": 0.2479068158228578,
      "grad_norm": 2.1413222276832324,
      "learning_rate": 1.1078777970948751e-06,
      "loss": 0.6882,
      "step": 2687
    },
    {
      "epoch": 0.24799907738438473,
      "grad_norm": 1.8030187154881656,
      "learning_rate": 1.1077583234175338e-06,
      "loss": 0.7754,
      "step": 2688
    },
    {
      "epoch": 0.24809133894591165,
      "grad_norm": 1.5547666932982365,
      "learning_rate": 1.107638805991873e-06,
      "loss": 0.7013,
      "step": 2689
    },
    {
      "epoch": 0.2481836005074386,
      "grad_norm": 2.0074387054007024,
      "learning_rate": 1.1075192448287242e-06,
      "loss": 0.6715,
      "step": 2690
    },
    {
      "epoch": 0.2481836005074386,
      "eval_GEN Loss": 0.4780990481376648,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.37563762068748474,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9396634697914124,
      "eval_runtime": 55.9202,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2690
    },
    {
      "epoch": 0.2482758620689655,
      "grad_norm": 1.7498174218088007,
      "learning_rate": 1.1073996399389219e-06,
      "loss": 0.6374,
      "step": 2691
    },
    {
      "epoch": 0.24836812363049243,
      "grad_norm": 2.4867012202192145,
      "learning_rate": 1.1072799913333048e-06,
      "loss": 0.5798,
      "step": 2692
    },
    {
      "epoch": 0.24846038519201938,
      "grad_norm": 1.5050903166311473,
      "learning_rate": 1.1071602990227156e-06,
      "loss": 0.5502,
      "step": 2693
    },
    {
      "epoch": 0.2485526467535463,
      "grad_norm": 3.2452340997446245,
      "learning_rate": 1.1070405630180014e-06,
      "loss": 0.8691,
      "step": 2694
    },
    {
      "epoch": 0.24864490831507324,
      "grad_norm": 2.1210822441781514,
      "learning_rate": 1.1069207833300128e-06,
      "loss": 0.711,
      "step": 2695
    },
    {
      "epoch": 0.24864490831507324,
      "eval_GEN Loss": 0.47814249992370605,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.37390992045402527,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9502403736114502,
      "eval_runtime": 55.9602,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2695
    },
    {
      "epoch": 0.24873716987660016,
      "grad_norm": 2.8177571574365814,
      "learning_rate": 1.106800959969604e-06,
      "loss": 0.9125,
      "step": 2696
    },
    {
      "epoch": 0.24882943143812708,
      "grad_norm": 1.5027747160508305,
      "learning_rate": 1.106681092947634e-06,
      "loss": 0.5993,
      "step": 2697
    },
    {
      "epoch": 0.24892169299965403,
      "grad_norm": 2.1972824734459353,
      "learning_rate": 1.106561182274965e-06,
      "loss": 0.6126,
      "step": 2698
    },
    {
      "epoch": 0.24901395456118094,
      "grad_norm": 1.87921303262548,
      "learning_rate": 1.106441227962464e-06,
      "loss": 0.6475,
      "step": 2699
    },
    {
      "epoch": 0.24910621612270786,
      "grad_norm": 2.269059349742849,
      "learning_rate": 1.1063212300210007e-06,
      "loss": 0.7593,
      "step": 2700
    },
    {
      "epoch": 0.24910621612270786,
      "eval_GEN Loss": 0.4765802025794983,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3797445595264435,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9288461804389954,
      "eval_runtime": 55.8275,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2700
    },
    {
      "epoch": 0.2491984776842348,
      "grad_norm": 2.133584930517032,
      "learning_rate": 1.1062011884614502e-06,
      "loss": 0.7743,
      "step": 2701
    },
    {
      "epoch": 0.24929073924576173,
      "grad_norm": 2.759934439164502,
      "learning_rate": 1.1060811032946904e-06,
      "loss": 0.8687,
      "step": 2702
    },
    {
      "epoch": 0.24938300080728867,
      "grad_norm": 1.501543511252636,
      "learning_rate": 1.1059609745316038e-06,
      "loss": 0.5996,
      "step": 2703
    },
    {
      "epoch": 0.2494752623688156,
      "grad_norm": 2.765476358871736,
      "learning_rate": 1.105840802183077e-06,
      "loss": 0.8542,
      "step": 2704
    },
    {
      "epoch": 0.2495675239303425,
      "grad_norm": 2.192084663737546,
      "learning_rate": 1.1057205862599995e-06,
      "loss": 0.8602,
      "step": 2705
    },
    {
      "epoch": 0.2495675239303425,
      "eval_GEN Loss": 0.47709906101226807,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.40841448307037354,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.922656238079071,
      "eval_runtime": 55.886,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2705
    },
    {
      "epoch": 0.24965978549186946,
      "grad_norm": 1.82215323380972,
      "learning_rate": 1.1056003267732657e-06,
      "loss": 0.7283,
      "step": 2706
    },
    {
      "epoch": 0.24975204705339638,
      "grad_norm": 1.9918539551359982,
      "learning_rate": 1.105480023733774e-06,
      "loss": 0.7047,
      "step": 2707
    },
    {
      "epoch": 0.2498443086149233,
      "grad_norm": 2.1372829755698053,
      "learning_rate": 1.1053596771524263e-06,
      "loss": 0.7286,
      "step": 2708
    },
    {
      "epoch": 0.24993657017645024,
      "grad_norm": 3.666956765802162,
      "learning_rate": 1.1052392870401285e-06,
      "loss": 0.5377,
      "step": 2709
    },
    {
      "epoch": 0.2500288317379772,
      "grad_norm": 2.434505687115984,
      "learning_rate": 1.1051188534077906e-06,
      "loss": 0.669,
      "step": 2710
    },
    {
      "epoch": 0.2500288317379772,
      "eval_GEN Loss": 0.47733473777770996,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4218302369117737,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9270432591438293,
      "eval_runtime": 55.9367,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2710
    },
    {
      "epoch": 0.2501210932995041,
      "grad_norm": 2.58324077058937,
      "learning_rate": 1.1049983762663267e-06,
      "loss": 0.9696,
      "step": 2711
    },
    {
      "epoch": 0.250213354861031,
      "grad_norm": 1.8620362834184567,
      "learning_rate": 1.104877855626654e-06,
      "loss": 0.7172,
      "step": 2712
    },
    {
      "epoch": 0.25030561642255794,
      "grad_norm": 2.60333731709298,
      "learning_rate": 1.104757291499695e-06,
      "loss": 0.7288,
      "step": 2713
    },
    {
      "epoch": 0.25039787798408486,
      "grad_norm": 2.0264868894723147,
      "learning_rate": 1.104636683896375e-06,
      "loss": 0.7463,
      "step": 2714
    },
    {
      "epoch": 0.25049013954561183,
      "grad_norm": 2.05589004516307,
      "learning_rate": 1.1045160328276237e-06,
      "loss": 0.6319,
      "step": 2715
    },
    {
      "epoch": 0.25049013954561183,
      "eval_GEN Loss": 0.4764898121356964,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4027074873447418,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9216946959495544,
      "eval_runtime": 56.0182,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2715
    },
    {
      "epoch": 0.25058240110713875,
      "grad_norm": 2.5289463158075356,
      "learning_rate": 1.104395338304375e-06,
      "loss": 0.6668,
      "step": 2716
    },
    {
      "epoch": 0.25067466266866567,
      "grad_norm": 1.5864082678088869,
      "learning_rate": 1.1042746003375656e-06,
      "loss": 0.6903,
      "step": 2717
    },
    {
      "epoch": 0.2507669242301926,
      "grad_norm": 1.6947751046940056,
      "learning_rate": 1.104153818938138e-06,
      "loss": 0.7553,
      "step": 2718
    },
    {
      "epoch": 0.2508591857917195,
      "grad_norm": 2.44322897534241,
      "learning_rate": 1.104032994117037e-06,
      "loss": 0.851,
      "step": 2719
    },
    {
      "epoch": 0.2509514473532464,
      "grad_norm": 2.141264688497023,
      "learning_rate": 1.103912125885212e-06,
      "loss": 0.8212,
      "step": 2720
    },
    {
      "epoch": 0.2509514473532464,
      "eval_GEN Loss": 0.4742036759853363,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37442681193351746,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.924098551273346,
      "eval_runtime": 55.7235,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 2720
    },
    {
      "epoch": 0.2510437089147734,
      "grad_norm": 1.6571002455330712,
      "learning_rate": 1.1037912142536165e-06,
      "loss": 0.5541,
      "step": 2721
    },
    {
      "epoch": 0.2511359704763003,
      "grad_norm": 2.9809860574731837,
      "learning_rate": 1.1036702592332074e-06,
      "loss": 0.6998,
      "step": 2722
    },
    {
      "epoch": 0.25122823203782724,
      "grad_norm": 1.468966340414337,
      "learning_rate": 1.103549260834946e-06,
      "loss": 0.4936,
      "step": 2723
    },
    {
      "epoch": 0.25132049359935416,
      "grad_norm": 2.289771045783806,
      "learning_rate": 1.1034282190697973e-06,
      "loss": 0.8155,
      "step": 2724
    },
    {
      "epoch": 0.2514127551608811,
      "grad_norm": 1.5916603472113702,
      "learning_rate": 1.1033071339487305e-06,
      "loss": 0.7593,
      "step": 2725
    },
    {
      "epoch": 0.2514127551608811,
      "eval_GEN Loss": 0.4734470546245575,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37095266580581665,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9146033525466919,
      "eval_runtime": 55.9357,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2725
    },
    {
      "epoch": 0.25150501672240805,
      "grad_norm": 1.9223544384193163,
      "learning_rate": 1.1031860054827182e-06,
      "loss": 0.8488,
      "step": 2726
    },
    {
      "epoch": 0.25159727828393497,
      "grad_norm": 1.5396407340450482,
      "learning_rate": 1.1030648336827374e-06,
      "loss": 0.6941,
      "step": 2727
    },
    {
      "epoch": 0.2516895398454619,
      "grad_norm": 2.6191322420455068,
      "learning_rate": 1.1029436185597688e-06,
      "loss": 0.989,
      "step": 2728
    },
    {
      "epoch": 0.2517818014069888,
      "grad_norm": 1.4063027010558389,
      "learning_rate": 1.1028223601247974e-06,
      "loss": 0.7228,
      "step": 2729
    },
    {
      "epoch": 0.2518740629685157,
      "grad_norm": 1.7045344682664259,
      "learning_rate": 1.1027010583888115e-06,
      "loss": 0.6418,
      "step": 2730
    },
    {
      "epoch": 0.2518740629685157,
      "eval_GEN Loss": 0.4710525870323181,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.3637698292732239,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9052283763885498,
      "eval_runtime": 55.9242,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2730
    },
    {
      "epoch": 0.2519663245300427,
      "grad_norm": 1.687606690212869,
      "learning_rate": 1.102579713362804e-06,
      "loss": 0.7405,
      "step": 2731
    },
    {
      "epoch": 0.2520585860915696,
      "grad_norm": 2.197511379073901,
      "learning_rate": 1.1024583250577708e-06,
      "loss": 0.7448,
      "step": 2732
    },
    {
      "epoch": 0.25215084765309653,
      "grad_norm": 3.2422250647843427,
      "learning_rate": 1.1023368934847128e-06,
      "loss": 1.1846,
      "step": 2733
    },
    {
      "epoch": 0.25224310921462345,
      "grad_norm": 2.7578343184098104,
      "learning_rate": 1.1022154186546341e-06,
      "loss": 0.9751,
      "step": 2734
    },
    {
      "epoch": 0.25233537077615037,
      "grad_norm": 1.4020187852952832,
      "learning_rate": 1.102093900578543e-06,
      "loss": 0.707,
      "step": 2735
    },
    {
      "epoch": 0.25233537077615037,
      "eval_GEN Loss": 0.47052672505378723,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3734349310398102,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8960937261581421,
      "eval_runtime": 55.9148,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2735
    },
    {
      "epoch": 0.2524276323376773,
      "grad_norm": 2.546163967016158,
      "learning_rate": 1.1019723392674518e-06,
      "loss": 0.8057,
      "step": 2736
    },
    {
      "epoch": 0.25251989389920426,
      "grad_norm": 2.3967268437186138,
      "learning_rate": 1.1018507347323767e-06,
      "loss": 0.8003,
      "step": 2737
    },
    {
      "epoch": 0.2526121554607312,
      "grad_norm": 1.8438201721412526,
      "learning_rate": 1.101729086984337e-06,
      "loss": 0.7421,
      "step": 2738
    },
    {
      "epoch": 0.2527044170222581,
      "grad_norm": 1.6678431054044824,
      "learning_rate": 1.1016073960343573e-06,
      "loss": 0.7248,
      "step": 2739
    },
    {
      "epoch": 0.252796678583785,
      "grad_norm": 2.2137463759365152,
      "learning_rate": 1.101485661893465e-06,
      "loss": 0.6184,
      "step": 2740
    },
    {
      "epoch": 0.252796678583785,
      "eval_GEN Loss": 0.47182321548461914,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4018365740776062,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9039663672447205,
      "eval_runtime": 55.9513,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2740
    },
    {
      "epoch": 0.25288894014531194,
      "grad_norm": 1.864168833974342,
      "learning_rate": 1.1013638845726921e-06,
      "loss": 0.7604,
      "step": 2741
    },
    {
      "epoch": 0.2529812017068389,
      "grad_norm": 1.9727405405879306,
      "learning_rate": 1.1012420640830743e-06,
      "loss": 0.7417,
      "step": 2742
    },
    {
      "epoch": 0.25307346326836583,
      "grad_norm": 2.198680636928872,
      "learning_rate": 1.101120200435651e-06,
      "loss": 0.6413,
      "step": 2743
    },
    {
      "epoch": 0.25316572482989275,
      "grad_norm": 1.7728143302191022,
      "learning_rate": 1.1009982936414656e-06,
      "loss": 0.7255,
      "step": 2744
    },
    {
      "epoch": 0.25325798639141966,
      "grad_norm": 1.987240472964129,
      "learning_rate": 1.1008763437115657e-06,
      "loss": 0.6966,
      "step": 2745
    },
    {
      "epoch": 0.25325798639141966,
      "eval_GEN Loss": 0.4717996120452881,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3923601508140564,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9013221263885498,
      "eval_runtime": 55.8897,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2745
    },
    {
      "epoch": 0.2533502479529466,
      "grad_norm": 1.4273274115572978,
      "learning_rate": 1.1007543506570023e-06,
      "loss": 0.7564,
      "step": 2746
    },
    {
      "epoch": 0.25344250951447356,
      "grad_norm": 2.6709177785801677,
      "learning_rate": 1.100632314488831e-06,
      "loss": 0.6231,
      "step": 2747
    },
    {
      "epoch": 0.2535347710760005,
      "grad_norm": 1.7995659784526166,
      "learning_rate": 1.1005102352181106e-06,
      "loss": 0.7487,
      "step": 2748
    },
    {
      "epoch": 0.2536270326375274,
      "grad_norm": 1.6121025402103923,
      "learning_rate": 1.1003881128559043e-06,
      "loss": 0.682,
      "step": 2749
    },
    {
      "epoch": 0.2537192941990543,
      "grad_norm": 2.0586458948042443,
      "learning_rate": 1.1002659474132787e-06,
      "loss": 0.6513,
      "step": 2750
    },
    {
      "epoch": 0.2537192941990543,
      "eval_GEN Loss": 0.47003883123397827,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.36587613821029663,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8927283883094788,
      "eval_runtime": 56.0661,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 2750
    },
    {
      "epoch": 0.25381155576058123,
      "grad_norm": 1.7331050204059888,
      "learning_rate": 1.1001437389013052e-06,
      "loss": 0.7722,
      "step": 2751
    },
    {
      "epoch": 0.25390381732210815,
      "grad_norm": 1.563777794889261,
      "learning_rate": 1.100021487331058e-06,
      "loss": 0.5947,
      "step": 2752
    },
    {
      "epoch": 0.2539960788836351,
      "grad_norm": 1.8759858025217275,
      "learning_rate": 1.099899192713616e-06,
      "loss": 0.8462,
      "step": 2753
    },
    {
      "epoch": 0.25408834044516204,
      "grad_norm": 1.922353800268907,
      "learning_rate": 1.0997768550600615e-06,
      "loss": 0.693,
      "step": 2754
    },
    {
      "epoch": 0.25418060200668896,
      "grad_norm": 1.5572991964844718,
      "learning_rate": 1.0996544743814813e-06,
      "loss": 0.7165,
      "step": 2755
    },
    {
      "epoch": 0.25418060200668896,
      "eval_GEN Loss": 0.4701164662837982,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3527683615684509,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.897536039352417,
      "eval_runtime": 55.8188,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2755
    },
    {
      "epoch": 0.2542728635682159,
      "grad_norm": 1.6046248906656375,
      "learning_rate": 1.0995320506889654e-06,
      "loss": 0.6619,
      "step": 2756
    },
    {
      "epoch": 0.2543651251297428,
      "grad_norm": 1.4209949872231915,
      "learning_rate": 1.0994095839936081e-06,
      "loss": 0.6936,
      "step": 2757
    },
    {
      "epoch": 0.25445738669126977,
      "grad_norm": 1.534915694918809,
      "learning_rate": 1.0992870743065077e-06,
      "loss": 0.532,
      "step": 2758
    },
    {
      "epoch": 0.2545496482527967,
      "grad_norm": 1.8044306504046457,
      "learning_rate": 1.0991645216387661e-06,
      "loss": 0.7963,
      "step": 2759
    },
    {
      "epoch": 0.2546419098143236,
      "grad_norm": 1.4632734763269986,
      "learning_rate": 1.0990419260014891e-06,
      "loss": 0.5536,
      "step": 2760
    },
    {
      "epoch": 0.2546419098143236,
      "eval_GEN Loss": 0.47240176796913147,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.34533143043518066,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8927884697914124,
      "eval_runtime": 55.7199,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 2760
    },
    {
      "epoch": 0.2547341713758505,
      "grad_norm": 1.7222787821212286,
      "learning_rate": 1.0989192874057865e-06,
      "loss": 0.6526,
      "step": 2761
    },
    {
      "epoch": 0.25482643293737745,
      "grad_norm": 1.7703129961636446,
      "learning_rate": 1.0987966058627726e-06,
      "loss": 0.7849,
      "step": 2762
    },
    {
      "epoch": 0.2549186944989044,
      "grad_norm": 1.5422890636705469,
      "learning_rate": 1.098673881383564e-06,
      "loss": 0.6029,
      "step": 2763
    },
    {
      "epoch": 0.25501095606043134,
      "grad_norm": 1.6050379066340041,
      "learning_rate": 1.098551113979283e-06,
      "loss": 0.4569,
      "step": 2764
    },
    {
      "epoch": 0.25510321762195826,
      "grad_norm": 1.3368182001234166,
      "learning_rate": 1.0984283036610544e-06,
      "loss": 0.6307,
      "step": 2765
    },
    {
      "epoch": 0.25510321762195826,
      "eval_GEN Loss": 0.4736599028110504,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3503721356391907,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8943509459495544,
      "eval_runtime": 55.9476,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2765
    },
    {
      "epoch": 0.2551954791834852,
      "grad_norm": 2.679884884446318,
      "learning_rate": 1.0983054504400078e-06,
      "loss": 0.9728,
      "step": 2766
    },
    {
      "epoch": 0.2552877407450121,
      "grad_norm": 2.2610303912303893,
      "learning_rate": 1.0981825543272765e-06,
      "loss": 0.7604,
      "step": 2767
    },
    {
      "epoch": 0.255380002306539,
      "grad_norm": 2.6415929410892214,
      "learning_rate": 1.098059615333997e-06,
      "loss": 0.9016,
      "step": 2768
    },
    {
      "epoch": 0.255472263868066,
      "grad_norm": 1.8350088078906566,
      "learning_rate": 1.0979366334713108e-06,
      "loss": 0.7414,
      "step": 2769
    },
    {
      "epoch": 0.2555645254295929,
      "grad_norm": 1.7231897799566085,
      "learning_rate": 1.0978136087503625e-06,
      "loss": 0.727,
      "step": 2770
    },
    {
      "epoch": 0.2555645254295929,
      "eval_GEN Loss": 0.4734903573989868,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3658936619758606,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8900841474533081,
      "eval_runtime": 55.934,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2770
    },
    {
      "epoch": 0.2556567869911198,
      "grad_norm": 1.6561298745041995,
      "learning_rate": 1.0976905411823005e-06,
      "loss": 0.7006,
      "step": 2771
    },
    {
      "epoch": 0.25574904855264674,
      "grad_norm": 1.5465264004793344,
      "learning_rate": 1.097567430778278e-06,
      "loss": 0.6357,
      "step": 2772
    },
    {
      "epoch": 0.25584131011417366,
      "grad_norm": 1.4634836315657354,
      "learning_rate": 1.0974442775494506e-06,
      "loss": 0.7187,
      "step": 2773
    },
    {
      "epoch": 0.25593357167570063,
      "grad_norm": 2.37903174551299,
      "learning_rate": 1.0973210815069795e-06,
      "loss": 0.8919,
      "step": 2774
    },
    {
      "epoch": 0.25602583323722755,
      "grad_norm": 1.9867145247951667,
      "learning_rate": 1.0971978426620285e-06,
      "loss": 0.5644,
      "step": 2775
    },
    {
      "epoch": 0.25602583323722755,
      "eval_GEN Loss": 0.47412750124931335,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.37998804450035095,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8872596025466919,
      "eval_runtime": 56.6649,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 2775
    },
    {
      "epoch": 0.25611809479875447,
      "grad_norm": 1.6259013407828804,
      "learning_rate": 1.0970745610257657e-06,
      "loss": 0.8081,
      "step": 2776
    },
    {
      "epoch": 0.2562103563602814,
      "grad_norm": 1.7748196718121532,
      "learning_rate": 1.096951236609363e-06,
      "loss": 0.7647,
      "step": 2777
    },
    {
      "epoch": 0.2563026179218083,
      "grad_norm": 1.779199920781924,
      "learning_rate": 1.0968278694239965e-06,
      "loss": 0.7141,
      "step": 2778
    },
    {
      "epoch": 0.2563948794833353,
      "grad_norm": 1.8238775569929553,
      "learning_rate": 1.096704459480846e-06,
      "loss": 0.8336,
      "step": 2779
    },
    {
      "epoch": 0.2564871410448622,
      "grad_norm": 2.5219035745327996,
      "learning_rate": 1.0965810067910946e-06,
      "loss": 0.7671,
      "step": 2780
    },
    {
      "epoch": 0.2564871410448622,
      "eval_GEN Loss": 0.47480443120002747,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.38208648562431335,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8912860751152039,
      "eval_runtime": 55.9268,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2780
    },
    {
      "epoch": 0.2565794026063891,
      "grad_norm": 2.010591698276529,
      "learning_rate": 1.09645751136593e-06,
      "loss": 0.7778,
      "step": 2781
    },
    {
      "epoch": 0.25667166416791604,
      "grad_norm": 1.9932117067457162,
      "learning_rate": 1.096333973216544e-06,
      "loss": 0.9167,
      "step": 2782
    },
    {
      "epoch": 0.25676392572944295,
      "grad_norm": 1.6171203505033056,
      "learning_rate": 1.0962103923541311e-06,
      "loss": 0.8603,
      "step": 2783
    },
    {
      "epoch": 0.2568561872909699,
      "grad_norm": 2.4121571177480354,
      "learning_rate": 1.096086768789891e-06,
      "loss": 0.836,
      "step": 2784
    },
    {
      "epoch": 0.25694844885249685,
      "grad_norm": 1.5901849529773644,
      "learning_rate": 1.0959631025350264e-06,
      "loss": 0.6084,
      "step": 2785
    },
    {
      "epoch": 0.25694844885249685,
      "eval_GEN Loss": 0.47598758339881897,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.388836145401001,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8961538672447205,
      "eval_runtime": 56.9302,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2785
    },
    {
      "epoch": 0.25704071041402377,
      "grad_norm": 1.8847158556124448,
      "learning_rate": 1.0958393936007442e-06,
      "loss": 0.7882,
      "step": 2786
    },
    {
      "epoch": 0.2571329719755507,
      "grad_norm": 1.5407724113318073,
      "learning_rate": 1.0957156419982548e-06,
      "loss": 0.6125,
      "step": 2787
    },
    {
      "epoch": 0.2572252335370776,
      "grad_norm": 1.6609345073163515,
      "learning_rate": 1.0955918477387731e-06,
      "loss": 0.5919,
      "step": 2788
    },
    {
      "epoch": 0.2573174950986045,
      "grad_norm": 1.7724364631933782,
      "learning_rate": 1.0954680108335174e-06,
      "loss": 0.7188,
      "step": 2789
    },
    {
      "epoch": 0.2574097566601315,
      "grad_norm": 1.9579215759808466,
      "learning_rate": 1.0953441312937102e-06,
      "loss": 0.5865,
      "step": 2790
    },
    {
      "epoch": 0.2574097566601315,
      "eval_GEN Loss": 0.47702980041503906,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.39262256026268005,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9006009697914124,
      "eval_runtime": 56.8136,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2790
    },
    {
      "epoch": 0.2575020182216584,
      "grad_norm": 1.802557422520963,
      "learning_rate": 1.0952202091305774e-06,
      "loss": 0.8458,
      "step": 2791
    },
    {
      "epoch": 0.25759427978318533,
      "grad_norm": 1.54787936179994,
      "learning_rate": 1.0950962443553492e-06,
      "loss": 0.6077,
      "step": 2792
    },
    {
      "epoch": 0.25768654134471225,
      "grad_norm": 1.584416561131339,
      "learning_rate": 1.0949722369792592e-06,
      "loss": 0.6189,
      "step": 2793
    },
    {
      "epoch": 0.25777880290623917,
      "grad_norm": 2.352430888381316,
      "learning_rate": 1.0948481870135454e-06,
      "loss": 0.9219,
      "step": 2794
    },
    {
      "epoch": 0.25787106446776614,
      "grad_norm": 2.6680729850490867,
      "learning_rate": 1.0947240944694496e-06,
      "loss": 0.6239,
      "step": 2795
    },
    {
      "epoch": 0.25787106446776614,
      "eval_GEN Loss": 0.47783082723617554,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.37871477007865906,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9094350934028625,
      "eval_runtime": 55.8842,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2795
    },
    {
      "epoch": 0.25796332602929306,
      "grad_norm": 1.4334642233541819,
      "learning_rate": 1.0945999593582166e-06,
      "loss": 0.6609,
      "step": 2796
    },
    {
      "epoch": 0.25805558759082,
      "grad_norm": 1.348389367047391,
      "learning_rate": 1.0944757816910965e-06,
      "loss": 0.6109,
      "step": 2797
    },
    {
      "epoch": 0.2581478491523469,
      "grad_norm": 1.9822240292242792,
      "learning_rate": 1.0943515614793419e-06,
      "loss": 0.7317,
      "step": 2798
    },
    {
      "epoch": 0.2582401107138738,
      "grad_norm": 1.4757451030987536,
      "learning_rate": 1.0942272987342098e-06,
      "loss": 0.6135,
      "step": 2799
    },
    {
      "epoch": 0.25833237227540073,
      "grad_norm": 2.7201681269827813,
      "learning_rate": 1.0941029934669616e-06,
      "loss": 0.7775,
      "step": 2800
    },
    {
      "epoch": 0.25833237227540073,
      "eval_GEN Loss": 0.4769158661365509,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3735000193119049,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9156851172447205,
      "eval_runtime": 56.0778,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 2800
    },
    {
      "epoch": 0.2584246338369277,
      "grad_norm": 1.320850616664563,
      "learning_rate": 1.0939786456888616e-06,
      "loss": 0.5955,
      "step": 2801
    },
    {
      "epoch": 0.2585168953984546,
      "grad_norm": 1.9698847411133535,
      "learning_rate": 1.0938542554111788e-06,
      "loss": 0.6053,
      "step": 2802
    },
    {
      "epoch": 0.25860915695998155,
      "grad_norm": 2.674289754523188,
      "learning_rate": 1.093729822645185e-06,
      "loss": 0.8347,
      "step": 2803
    },
    {
      "epoch": 0.25870141852150846,
      "grad_norm": 1.9187299052654343,
      "learning_rate": 1.0936053474021573e-06,
      "loss": 0.8464,
      "step": 2804
    },
    {
      "epoch": 0.2587936800830354,
      "grad_norm": 1.7308216274669346,
      "learning_rate": 1.0934808296933753e-06,
      "loss": 0.6464,
      "step": 2805
    },
    {
      "epoch": 0.2587936800830354,
      "eval_GEN Loss": 0.4770454168319702,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3920973539352417,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9135817289352417,
      "eval_runtime": 55.8704,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2805
    },
    {
      "epoch": 0.25888594164456236,
      "grad_norm": 2.050243820309279,
      "learning_rate": 1.0933562695301228e-06,
      "loss": 0.749,
      "step": 2806
    },
    {
      "epoch": 0.2589782032060893,
      "grad_norm": 1.5088742536349635,
      "learning_rate": 1.0932316669236882e-06,
      "loss": 0.5941,
      "step": 2807
    },
    {
      "epoch": 0.2590704647676162,
      "grad_norm": 3.192567161532349,
      "learning_rate": 1.093107021885363e-06,
      "loss": 0.8434,
      "step": 2808
    },
    {
      "epoch": 0.2591627263291431,
      "grad_norm": 2.8380429039134536,
      "learning_rate": 1.0929823344264426e-06,
      "loss": 0.7969,
      "step": 2809
    },
    {
      "epoch": 0.25925498789067003,
      "grad_norm": 2.367740112491841,
      "learning_rate": 1.0928576045582266e-06,
      "loss": 0.6597,
      "step": 2810
    },
    {
      "epoch": 0.25925498789067003,
      "eval_GEN Loss": 0.4770470857620239,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4092256426811218,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9203125238418579,
      "eval_runtime": 55.8472,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2810
    },
    {
      "epoch": 0.259347249452197,
      "grad_norm": 1.665398423253935,
      "learning_rate": 1.0927328322920178e-06,
      "loss": 0.77,
      "step": 2811
    },
    {
      "epoch": 0.2594395110137239,
      "grad_norm": 1.750131626629132,
      "learning_rate": 1.0926080176391235e-06,
      "loss": 0.5612,
      "step": 2812
    },
    {
      "epoch": 0.25953177257525084,
      "grad_norm": 1.6182853219142004,
      "learning_rate": 1.0924831606108546e-06,
      "loss": 0.7637,
      "step": 2813
    },
    {
      "epoch": 0.25962403413677776,
      "grad_norm": 2.144579481980696,
      "learning_rate": 1.092358261218526e-06,
      "loss": 0.6083,
      "step": 2814
    },
    {
      "epoch": 0.2597162956983047,
      "grad_norm": 1.6875025106559716,
      "learning_rate": 1.0922333194734562e-06,
      "loss": 0.6488,
      "step": 2815
    },
    {
      "epoch": 0.2597162956983047,
      "eval_GEN Loss": 0.4769200384616852,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.42085930705070496,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9325721263885498,
      "eval_runtime": 55.7213,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 2815
    },
    {
      "epoch": 0.2598085572598316,
      "grad_norm": 2.162660529928559,
      "learning_rate": 1.0921083353869674e-06,
      "loss": 0.8474,
      "step": 2816
    },
    {
      "epoch": 0.25990081882135857,
      "grad_norm": 1.466975487936104,
      "learning_rate": 1.091983308970386e-06,
      "loss": 0.6649,
      "step": 2817
    },
    {
      "epoch": 0.2599930803828855,
      "grad_norm": 1.947681954358698,
      "learning_rate": 1.0918582402350421e-06,
      "loss": 0.7354,
      "step": 2818
    },
    {
      "epoch": 0.2600853419444124,
      "grad_norm": 1.4949756812404982,
      "learning_rate": 1.0917331291922698e-06,
      "loss": 0.5357,
      "step": 2819
    },
    {
      "epoch": 0.2601776035059393,
      "grad_norm": 1.7323684241199255,
      "learning_rate": 1.0916079758534067e-06,
      "loss": 0.6904,
      "step": 2820
    },
    {
      "epoch": 0.2601776035059393,
      "eval_GEN Loss": 0.47724971175193787,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.42346706986427307,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9420673251152039,
      "eval_runtime": 55.7956,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2820
    },
    {
      "epoch": 0.26026986506746624,
      "grad_norm": 2.285875956263543,
      "learning_rate": 1.0914827802297942e-06,
      "loss": 0.8076,
      "step": 2821
    },
    {
      "epoch": 0.2603621266289932,
      "grad_norm": 1.8603760629637214,
      "learning_rate": 1.091357542332778e-06,
      "loss": 0.7745,
      "step": 2822
    },
    {
      "epoch": 0.26045438819052014,
      "grad_norm": 2.7703884501992952,
      "learning_rate": 1.091232262173707e-06,
      "loss": 0.8429,
      "step": 2823
    },
    {
      "epoch": 0.26054664975204705,
      "grad_norm": 2.629327540371188,
      "learning_rate": 1.0911069397639349e-06,
      "loss": 0.9216,
      "step": 2824
    },
    {
      "epoch": 0.260638911313574,
      "grad_norm": 1.870606733976828,
      "learning_rate": 1.090981575114818e-06,
      "loss": 0.8966,
      "step": 2825
    },
    {
      "epoch": 0.260638911313574,
      "eval_GEN Loss": 0.4756940007209778,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.4186084568500519,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9485576748847961,
      "eval_runtime": 55.7228,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 2825
    },
    {
      "epoch": 0.2607311728751009,
      "grad_norm": 2.6340260628856034,
      "learning_rate": 1.0908561682377175e-06,
      "loss": 0.8419,
      "step": 2826
    },
    {
      "epoch": 0.26082343443662787,
      "grad_norm": 1.94939204128999,
      "learning_rate": 1.0907307191439977e-06,
      "loss": 0.7387,
      "step": 2827
    },
    {
      "epoch": 0.2609156959981548,
      "grad_norm": 1.5741812293126676,
      "learning_rate": 1.090605227845027e-06,
      "loss": 0.7373,
      "step": 2828
    },
    {
      "epoch": 0.2610079575596817,
      "grad_norm": 2.2443092110742726,
      "learning_rate": 1.0904796943521777e-06,
      "loss": 0.8435,
      "step": 2829
    },
    {
      "epoch": 0.2611002191212086,
      "grad_norm": 1.434500517995226,
      "learning_rate": 1.0903541186768257e-06,
      "loss": 0.6617,
      "step": 2830
    },
    {
      "epoch": 0.2611002191212086,
      "eval_GEN Loss": 0.47352904081344604,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.4016229212284088,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9518029093742371,
      "eval_runtime": 55.8548,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2830
    },
    {
      "epoch": 0.26119248068273554,
      "grad_norm": 1.7298664370626673,
      "learning_rate": 1.0902285008303513e-06,
      "loss": 0.7555,
      "step": 2831
    },
    {
      "epoch": 0.26128474224426246,
      "grad_norm": 1.4444042079177812,
      "learning_rate": 1.0901028408241377e-06,
      "loss": 0.5976,
      "step": 2832
    },
    {
      "epoch": 0.26137700380578943,
      "grad_norm": 1.6604419780132995,
      "learning_rate": 1.0899771386695726e-06,
      "loss": 0.6701,
      "step": 2833
    },
    {
      "epoch": 0.26146926536731635,
      "grad_norm": 3.146589065133363,
      "learning_rate": 1.0898513943780471e-06,
      "loss": 0.6214,
      "step": 2834
    },
    {
      "epoch": 0.26156152692884327,
      "grad_norm": 1.968668778021192,
      "learning_rate": 1.0897256079609567e-06,
      "loss": 0.6522,
      "step": 2835
    },
    {
      "epoch": 0.26156152692884327,
      "eval_GEN Loss": 0.4726296663284302,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3989977240562439,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9579927921295166,
      "eval_runtime": 55.89,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2835
    },
    {
      "epoch": 0.2616537884903702,
      "grad_norm": 3.026666023218241,
      "learning_rate": 1.0895997794297003e-06,
      "loss": 0.9134,
      "step": 2836
    },
    {
      "epoch": 0.2617460500518971,
      "grad_norm": 1.7298089701732546,
      "learning_rate": 1.0894739087956806e-06,
      "loss": 0.5006,
      "step": 2837
    },
    {
      "epoch": 0.2618383116134241,
      "grad_norm": 1.4286580177198507,
      "learning_rate": 1.089347996070304e-06,
      "loss": 0.5675,
      "step": 2838
    },
    {
      "epoch": 0.261930573174951,
      "grad_norm": 3.652190565418445,
      "learning_rate": 1.0892220412649814e-06,
      "loss": 0.8784,
      "step": 2839
    },
    {
      "epoch": 0.2620228347364779,
      "grad_norm": 1.7002966343821668,
      "learning_rate": 1.0890960443911265e-06,
      "loss": 0.6068,
      "step": 2840
    },
    {
      "epoch": 0.2620228347364779,
      "eval_GEN Loss": 0.4713214039802551,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.41650906205177307,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9401442408561707,
      "eval_runtime": 55.952,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2840
    },
    {
      "epoch": 0.26211509629800483,
      "grad_norm": 2.1625294193006255,
      "learning_rate": 1.0889700054601576e-06,
      "loss": 0.7479,
      "step": 2841
    },
    {
      "epoch": 0.26220735785953175,
      "grad_norm": 1.9285517920868451,
      "learning_rate": 1.0888439244834965e-06,
      "loss": 0.714,
      "step": 2842
    },
    {
      "epoch": 0.2622996194210587,
      "grad_norm": 1.5857584967914014,
      "learning_rate": 1.088717801472569e-06,
      "loss": 0.6862,
      "step": 2843
    },
    {
      "epoch": 0.26239188098258565,
      "grad_norm": 1.4191435423319874,
      "learning_rate": 1.0885916364388043e-06,
      "loss": 0.8066,
      "step": 2844
    },
    {
      "epoch": 0.26248414254411256,
      "grad_norm": 1.4633345884881437,
      "learning_rate": 1.0884654293936358e-06,
      "loss": 0.6089,
      "step": 2845
    },
    {
      "epoch": 0.26248414254411256,
      "eval_GEN Loss": 0.4720485210418701,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.426534503698349,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9353966116905212,
      "eval_runtime": 56.0563,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 2845
    },
    {
      "epoch": 0.2625764041056395,
      "grad_norm": 2.2858576838558835,
      "learning_rate": 1.0883391803485005e-06,
      "loss": 0.6386,
      "step": 2846
    },
    {
      "epoch": 0.2626686656671664,
      "grad_norm": 1.9808166992748992,
      "learning_rate": 1.0882128893148395e-06,
      "loss": 0.7891,
      "step": 2847
    },
    {
      "epoch": 0.2627609272286933,
      "grad_norm": 1.7137051599153532,
      "learning_rate": 1.0880865563040975e-06,
      "loss": 0.6102,
      "step": 2848
    },
    {
      "epoch": 0.2628531887902203,
      "grad_norm": 1.6459237540588072,
      "learning_rate": 1.0879601813277226e-06,
      "loss": 0.8092,
      "step": 2849
    },
    {
      "epoch": 0.2629454503517472,
      "grad_norm": 2.1743503929510166,
      "learning_rate": 1.0878337643971677e-06,
      "loss": 0.8703,
      "step": 2850
    },
    {
      "epoch": 0.2629454503517472,
      "eval_GEN Loss": 0.47063061594963074,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4229276776313782,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.931911051273346,
      "eval_runtime": 55.8547,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2850
    },
    {
      "epoch": 0.26303771191327413,
      "grad_norm": 2.0289772824087042,
      "learning_rate": 1.0877073055238883e-06,
      "loss": 0.9132,
      "step": 2851
    },
    {
      "epoch": 0.26312997347480105,
      "grad_norm": 2.0769986421133293,
      "learning_rate": 1.0875808047193444e-06,
      "loss": 0.7412,
      "step": 2852
    },
    {
      "epoch": 0.26322223503632797,
      "grad_norm": 1.6719285616073285,
      "learning_rate": 1.087454261995e-06,
      "loss": 0.5139,
      "step": 2853
    },
    {
      "epoch": 0.26331449659785494,
      "grad_norm": 1.7787465322995637,
      "learning_rate": 1.0873276773623227e-06,
      "loss": 0.8152,
      "step": 2854
    },
    {
      "epoch": 0.26340675815938186,
      "grad_norm": 2.1741799027194255,
      "learning_rate": 1.0872010508327833e-06,
      "loss": 0.8449,
      "step": 2855
    },
    {
      "epoch": 0.26340675815938186,
      "eval_GEN Loss": 0.47170206904411316,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.42667096853256226,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9335336685180664,
      "eval_runtime": 55.912,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2855
    },
    {
      "epoch": 0.2634990197209088,
      "grad_norm": 1.7974457261666446,
      "learning_rate": 1.0870743824178572e-06,
      "loss": 0.8676,
      "step": 2856
    },
    {
      "epoch": 0.2635912812824357,
      "grad_norm": 1.8658790318280039,
      "learning_rate": 1.086947672129023e-06,
      "loss": 0.8661,
      "step": 2857
    },
    {
      "epoch": 0.2636835428439626,
      "grad_norm": 2.184235811563245,
      "learning_rate": 1.086820919977764e-06,
      "loss": 0.7951,
      "step": 2858
    },
    {
      "epoch": 0.2637758044054896,
      "grad_norm": 2.122415369907746,
      "learning_rate": 1.0866941259755663e-06,
      "loss": 0.6642,
      "step": 2859
    },
    {
      "epoch": 0.2638680659670165,
      "grad_norm": 2.7901658251737365,
      "learning_rate": 1.0865672901339202e-06,
      "loss": 0.7244,
      "step": 2860
    },
    {
      "epoch": 0.2638680659670165,
      "eval_GEN Loss": 0.4719868302345276,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.42815595865249634,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9323317408561707,
      "eval_runtime": 55.8947,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2860
    },
    {
      "epoch": 0.2639603275285434,
      "grad_norm": 1.4985642027008574,
      "learning_rate": 1.0864404124643197e-06,
      "loss": 0.4886,
      "step": 2861
    },
    {
      "epoch": 0.26405258909007034,
      "grad_norm": 1.6611065077938427,
      "learning_rate": 1.0863134929782626e-06,
      "loss": 0.8357,
      "step": 2862
    },
    {
      "epoch": 0.26414485065159726,
      "grad_norm": 1.5424880746667047,
      "learning_rate": 1.0861865316872506e-06,
      "loss": 0.7578,
      "step": 2863
    },
    {
      "epoch": 0.2642371122131242,
      "grad_norm": 1.5086276870700677,
      "learning_rate": 1.0860595286027893e-06,
      "loss": 0.6162,
      "step": 2864
    },
    {
      "epoch": 0.26432937377465116,
      "grad_norm": 1.7995003060467978,
      "learning_rate": 1.0859324837363878e-06,
      "loss": 0.7894,
      "step": 2865
    },
    {
      "epoch": 0.26432937377465116,
      "eval_GEN Loss": 0.4730812609195709,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.42739221453666687,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9325721263885498,
      "eval_runtime": 55.8967,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2865
    },
    {
      "epoch": 0.2644216353361781,
      "grad_norm": 2.2670008145067673,
      "learning_rate": 1.085805397099559e-06,
      "loss": 0.7332,
      "step": 2866
    },
    {
      "epoch": 0.264513896897705,
      "grad_norm": 1.9072589631028012,
      "learning_rate": 1.08567826870382e-06,
      "loss": 0.6485,
      "step": 2867
    },
    {
      "epoch": 0.2646061584592319,
      "grad_norm": 1.600773972719319,
      "learning_rate": 1.0855510985606912e-06,
      "loss": 0.6992,
      "step": 2868
    },
    {
      "epoch": 0.26469842002075883,
      "grad_norm": 2.2154277191973315,
      "learning_rate": 1.0854238866816966e-06,
      "loss": 0.6674,
      "step": 2869
    },
    {
      "epoch": 0.2647906815822858,
      "grad_norm": 2.246468231463725,
      "learning_rate": 1.085296633078365e-06,
      "loss": 0.7329,
      "step": 2870
    },
    {
      "epoch": 0.2647906815822858,
      "eval_GEN Loss": 0.47351667284965515,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.41970211267471313,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.931370198726654,
      "eval_runtime": 55.7497,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 2870
    },
    {
      "epoch": 0.2648829431438127,
      "grad_norm": 2.0524643362930917,
      "learning_rate": 1.0851693377622276e-06,
      "loss": 0.765,
      "step": 2871
    },
    {
      "epoch": 0.26497520470533964,
      "grad_norm": 1.6777409247684603,
      "learning_rate": 1.085042000744821e-06,
      "loss": 0.7887,
      "step": 2872
    },
    {
      "epoch": 0.26506746626686656,
      "grad_norm": 2.9044502055453307,
      "learning_rate": 1.084914622037684e-06,
      "loss": 0.9233,
      "step": 2873
    },
    {
      "epoch": 0.2651597278283935,
      "grad_norm": 1.851984411139579,
      "learning_rate": 1.0847872016523598e-06,
      "loss": 0.6793,
      "step": 2874
    },
    {
      "epoch": 0.26525198938992045,
      "grad_norm": 1.7689138325551346,
      "learning_rate": 1.084659739600396e-06,
      "loss": 0.8068,
      "step": 2875
    },
    {
      "epoch": 0.26525198938992045,
      "eval_GEN Loss": 0.47261348366737366,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.41723376512527466,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9238581657409668,
      "eval_runtime": 56.662,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 2875
    },
    {
      "epoch": 0.26534425095144737,
      "grad_norm": 2.518988655945643,
      "learning_rate": 1.0845322358933429e-06,
      "loss": 0.8823,
      "step": 2876
    },
    {
      "epoch": 0.2654365125129743,
      "grad_norm": 1.6922890281696157,
      "learning_rate": 1.0844046905427552e-06,
      "loss": 0.6131,
      "step": 2877
    },
    {
      "epoch": 0.2655287740745012,
      "grad_norm": 1.5298582902353595,
      "learning_rate": 1.0842771035601916e-06,
      "loss": 0.5772,
      "step": 2878
    },
    {
      "epoch": 0.2656210356360281,
      "grad_norm": 2.140596771900157,
      "learning_rate": 1.0841494749572135e-06,
      "loss": 0.7481,
      "step": 2879
    },
    {
      "epoch": 0.26571329719755504,
      "grad_norm": 1.4039039760746899,
      "learning_rate": 1.0840218047453877e-06,
      "loss": 0.6066,
      "step": 2880
    },
    {
      "epoch": 0.26571329719755504,
      "eval_GEN Loss": 0.4720955193042755,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.38698023557662964,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9079927802085876,
      "eval_runtime": 56.5785,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 2880
    },
    {
      "epoch": 0.265805558759082,
      "grad_norm": 1.734679531546747,
      "learning_rate": 1.0838940929362832e-06,
      "loss": 0.7649,
      "step": 2881
    },
    {
      "epoch": 0.26589782032060894,
      "grad_norm": 1.4436070519957807,
      "learning_rate": 1.0837663395414737e-06,
      "loss": 0.5511,
      "step": 2882
    },
    {
      "epoch": 0.26599008188213585,
      "grad_norm": 1.5306854758668027,
      "learning_rate": 1.0836385445725366e-06,
      "loss": 0.8169,
      "step": 2883
    },
    {
      "epoch": 0.26608234344366277,
      "grad_norm": 2.209621320462489,
      "learning_rate": 1.0835107080410525e-06,
      "loss": 0.7476,
      "step": 2884
    },
    {
      "epoch": 0.2661746050051897,
      "grad_norm": 2.333483988165698,
      "learning_rate": 1.0833828299586064e-06,
      "loss": 0.8471,
      "step": 2885
    },
    {
      "epoch": 0.2661746050051897,
      "eval_GEN Loss": 0.4727107882499695,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.37869298458099365,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9090144038200378,
      "eval_runtime": 56.8512,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2885
    },
    {
      "epoch": 0.26626686656671666,
      "grad_norm": 3.621223958701815,
      "learning_rate": 1.0832549103367868e-06,
      "loss": 0.9463,
      "step": 2886
    },
    {
      "epoch": 0.2663591281282436,
      "grad_norm": 2.5282885683733642,
      "learning_rate": 1.083126949187186e-06,
      "loss": 0.7999,
      "step": 2887
    },
    {
      "epoch": 0.2664513896897705,
      "grad_norm": 1.9973503048168053,
      "learning_rate": 1.0829989465213996e-06,
      "loss": 0.6615,
      "step": 2888
    },
    {
      "epoch": 0.2665436512512974,
      "grad_norm": 1.758681115083525,
      "learning_rate": 1.0828709023510282e-06,
      "loss": 0.7291,
      "step": 2889
    },
    {
      "epoch": 0.26663591281282434,
      "grad_norm": 1.7853998571874048,
      "learning_rate": 1.0827428166876746e-06,
      "loss": 0.7475,
      "step": 2890
    },
    {
      "epoch": 0.26663591281282434,
      "eval_GEN Loss": 0.4736158549785614,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3929135203361511,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9142428040504456,
      "eval_runtime": 56.6963,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2890
    },
    {
      "epoch": 0.2667281743743513,
      "grad_norm": 2.0202502120712547,
      "learning_rate": 1.0826146895429465e-06,
      "loss": 0.7987,
      "step": 2891
    },
    {
      "epoch": 0.26682043593587823,
      "grad_norm": 2.0584889741402552,
      "learning_rate": 1.082486520928455e-06,
      "loss": 0.7831,
      "step": 2892
    },
    {
      "epoch": 0.26691269749740515,
      "grad_norm": 1.7347886982583645,
      "learning_rate": 1.082358310855815e-06,
      "loss": 0.659,
      "step": 2893
    },
    {
      "epoch": 0.26700495905893207,
      "grad_norm": 2.2671950749564704,
      "learning_rate": 1.0822300593366446e-06,
      "loss": 0.87,
      "step": 2894
    },
    {
      "epoch": 0.267097220620459,
      "grad_norm": 1.6796338733481695,
      "learning_rate": 1.082101766382567e-06,
      "loss": 0.6606,
      "step": 2895
    },
    {
      "epoch": 0.267097220620459,
      "eval_GEN Loss": 0.4737313985824585,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4053592085838318,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9170072078704834,
      "eval_runtime": 56.834,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2895
    },
    {
      "epoch": 0.2671894821819859,
      "grad_norm": 1.8186297016083144,
      "learning_rate": 1.0819734320052077e-06,
      "loss": 0.6944,
      "step": 2896
    },
    {
      "epoch": 0.2672817437435129,
      "grad_norm": 2.2232199693453305,
      "learning_rate": 1.0818450562161966e-06,
      "loss": 0.7168,
      "step": 2897
    },
    {
      "epoch": 0.2673740053050398,
      "grad_norm": 2.4038255462991507,
      "learning_rate": 1.0817166390271674e-06,
      "loss": 0.7652,
      "step": 2898
    },
    {
      "epoch": 0.2674662668665667,
      "grad_norm": 1.9710086581823223,
      "learning_rate": 1.0815881804497576e-06,
      "loss": 0.823,
      "step": 2899
    },
    {
      "epoch": 0.26755852842809363,
      "grad_norm": 1.4910994291652362,
      "learning_rate": 1.081459680495608e-06,
      "loss": 0.7373,
      "step": 2900
    },
    {
      "epoch": 0.26755852842809363,
      "eval_GEN Loss": 0.47203004360198975,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.410864919424057,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9215144515037537,
      "eval_runtime": 56.7335,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 2900
    },
    {
      "epoch": 0.26765078998962055,
      "grad_norm": 2.1131906187932836,
      "learning_rate": 1.081331139176364e-06,
      "loss": 0.6806,
      "step": 2901
    },
    {
      "epoch": 0.2677430515511475,
      "grad_norm": 2.7948668407010326,
      "learning_rate": 1.0812025565036738e-06,
      "loss": 0.8329,
      "step": 2902
    },
    {
      "epoch": 0.26783531311267444,
      "grad_norm": 2.6533122459357092,
      "learning_rate": 1.0810739324891897e-06,
      "loss": 0.7192,
      "step": 2903
    },
    {
      "epoch": 0.26792757467420136,
      "grad_norm": 2.0983627377891305,
      "learning_rate": 1.080945267144568e-06,
      "loss": 0.7468,
      "step": 2904
    },
    {
      "epoch": 0.2680198362357283,
      "grad_norm": 3.007420205172924,
      "learning_rate": 1.0808165604814686e-06,
      "loss": 1.0453,
      "step": 2905
    },
    {
      "epoch": 0.2680198362357283,
      "eval_GEN Loss": 0.47039514780044556,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.4074372351169586,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9245192408561707,
      "eval_runtime": 56.786,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 2905
    },
    {
      "epoch": 0.2681120977972552,
      "grad_norm": 1.5998505247895043,
      "learning_rate": 1.0806878125115549e-06,
      "loss": 0.6214,
      "step": 2906
    },
    {
      "epoch": 0.2682043593587822,
      "grad_norm": 1.7106725387046147,
      "learning_rate": 1.0805590232464944e-06,
      "loss": 0.7344,
      "step": 2907
    },
    {
      "epoch": 0.2682966209203091,
      "grad_norm": 1.9177576355349442,
      "learning_rate": 1.0804301926979584e-06,
      "loss": 0.6649,
      "step": 2908
    },
    {
      "epoch": 0.268388882481836,
      "grad_norm": 1.3196856273651647,
      "learning_rate": 1.0803013208776212e-06,
      "loss": 0.6409,
      "step": 2909
    },
    {
      "epoch": 0.26848114404336293,
      "grad_norm": 1.9568369630543458,
      "learning_rate": 1.0801724077971618e-06,
      "loss": 0.8207,
      "step": 2910
    },
    {
      "epoch": 0.26848114404336293,
      "eval_GEN Loss": 0.4692898094654083,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3971266746520996,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9300480484962463,
      "eval_runtime": 56.8504,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2910
    },
    {
      "epoch": 0.26857340560488985,
      "grad_norm": 2.3157908592391068,
      "learning_rate": 1.0800434534682623e-06,
      "loss": 0.8243,
      "step": 2911
    },
    {
      "epoch": 0.26866566716641677,
      "grad_norm": 1.613711002272769,
      "learning_rate": 1.0799144579026087e-06,
      "loss": 0.6867,
      "step": 2912
    },
    {
      "epoch": 0.26875792872794374,
      "grad_norm": 2.4738510202088015,
      "learning_rate": 1.0797854211118911e-06,
      "loss": 0.7694,
      "step": 2913
    },
    {
      "epoch": 0.26885019028947066,
      "grad_norm": 2.2261598878042395,
      "learning_rate": 1.0796563431078027e-06,
      "loss": 0.8152,
      "step": 2914
    },
    {
      "epoch": 0.2689424518509976,
      "grad_norm": 1.761019883546548,
      "learning_rate": 1.0795272239020407e-06,
      "loss": 0.8802,
      "step": 2915
    },
    {
      "epoch": 0.2689424518509976,
      "eval_GEN Loss": 0.46886858344078064,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.39003583788871765,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9323918223381042,
      "eval_runtime": 56.6345,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 2915
    },
    {
      "epoch": 0.2690347134125245,
      "grad_norm": 2.6407705364832723,
      "learning_rate": 1.0793980635063064e-06,
      "loss": 0.8825,
      "step": 2916
    },
    {
      "epoch": 0.2691269749740514,
      "grad_norm": 1.7562121403030946,
      "learning_rate": 1.0792688619323042e-06,
      "loss": 0.7568,
      "step": 2917
    },
    {
      "epoch": 0.2692192365355784,
      "grad_norm": 1.8160496091454899,
      "learning_rate": 1.0791396191917429e-06,
      "loss": 0.747,
      "step": 2918
    },
    {
      "epoch": 0.2693114980971053,
      "grad_norm": 1.4703350220943332,
      "learning_rate": 1.0790103352963343e-06,
      "loss": 0.687,
      "step": 2919
    },
    {
      "epoch": 0.2694037596586322,
      "grad_norm": 2.53177968868352,
      "learning_rate": 1.0788810102577946e-06,
      "loss": 0.6893,
      "step": 2920
    },
    {
      "epoch": 0.2694037596586322,
      "eval_GEN Loss": 0.4682959020137787,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.39357900619506836,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9245192408561707,
      "eval_runtime": 56.8544,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 2920
    },
    {
      "epoch": 0.26949602122015914,
      "grad_norm": 1.8516885475416047,
      "learning_rate": 1.0787516440878435e-06,
      "loss": 0.7163,
      "step": 2921
    },
    {
      "epoch": 0.26958828278168606,
      "grad_norm": 2.1160565704273266,
      "learning_rate": 1.0786222367982037e-06,
      "loss": 0.8189,
      "step": 2922
    },
    {
      "epoch": 0.26968054434321304,
      "grad_norm": 1.9040697952784618,
      "learning_rate": 1.0784927884006032e-06,
      "loss": 0.7959,
      "step": 2923
    },
    {
      "epoch": 0.26977280590473995,
      "grad_norm": 1.4444408649716918,
      "learning_rate": 1.0783632989067725e-06,
      "loss": 0.6232,
      "step": 2924
    },
    {
      "epoch": 0.2698650674662669,
      "grad_norm": 1.3227319492381608,
      "learning_rate": 1.0782337683284457e-06,
      "loss": 0.5944,
      "step": 2925
    },
    {
      "epoch": 0.2698650674662669,
      "eval_GEN Loss": 0.4667108356952667,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.4003351628780365,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9219951629638672,
      "eval_runtime": 56.8948,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 2925
    },
    {
      "epoch": 0.2699573290277938,
      "grad_norm": 1.901285890344741,
      "learning_rate": 1.0781041966773615e-06,
      "loss": 0.8238,
      "step": 2926
    },
    {
      "epoch": 0.2700495905893207,
      "grad_norm": 1.880259543140357,
      "learning_rate": 1.077974583965262e-06,
      "loss": 0.8227,
      "step": 2927
    },
    {
      "epoch": 0.27014185215084763,
      "grad_norm": 1.6942444576860036,
      "learning_rate": 1.0778449302038925e-06,
      "loss": 0.8278,
      "step": 2928
    },
    {
      "epoch": 0.2702341137123746,
      "grad_norm": 1.7801397377465484,
      "learning_rate": 1.0777152354050026e-06,
      "loss": 0.6676,
      "step": 2929
    },
    {
      "epoch": 0.2703263752739015,
      "grad_norm": 2.2657342008091192,
      "learning_rate": 1.0775854995803456e-06,
      "loss": 0.7701,
      "step": 2930
    },
    {
      "epoch": 0.2703263752739015,
      "eval_GEN Loss": 0.4680368900299072,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.4020960330963135,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9251803159713745,
      "eval_runtime": 55.9556,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2930
    },
    {
      "epoch": 0.27041863683542844,
      "grad_norm": 1.6407330308061099,
      "learning_rate": 1.0774557227416783e-06,
      "loss": 0.5789,
      "step": 2931
    },
    {
      "epoch": 0.27051089839695536,
      "grad_norm": 1.6021232520345274,
      "learning_rate": 1.077325904900761e-06,
      "loss": 0.6688,
      "step": 2932
    },
    {
      "epoch": 0.2706031599584823,
      "grad_norm": 1.479719404476177,
      "learning_rate": 1.0771960460693588e-06,
      "loss": 0.7653,
      "step": 2933
    },
    {
      "epoch": 0.27069542152000925,
      "grad_norm": 1.4609981980198374,
      "learning_rate": 1.0770661462592385e-06,
      "loss": 0.6051,
      "step": 2934
    },
    {
      "epoch": 0.27078768308153617,
      "grad_norm": 1.0992705582228601,
      "learning_rate": 1.0769362054821728e-06,
      "loss": 0.5368,
      "step": 2935
    },
    {
      "epoch": 0.27078768308153617,
      "eval_GEN Loss": 0.4699912965297699,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.4076942801475525,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9277043342590332,
      "eval_runtime": 55.7894,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 2935
    },
    {
      "epoch": 0.2708799446430631,
      "grad_norm": 1.4378030605591086,
      "learning_rate": 1.0768062237499365e-06,
      "loss": 0.729,
      "step": 2936
    },
    {
      "epoch": 0.27097220620459,
      "grad_norm": 1.6161195381033198,
      "learning_rate": 1.0766762010743093e-06,
      "loss": 0.5311,
      "step": 2937
    },
    {
      "epoch": 0.2710644677661169,
      "grad_norm": 1.7086682571920622,
      "learning_rate": 1.0765461374670738e-06,
      "loss": 0.8596,
      "step": 2938
    },
    {
      "epoch": 0.2711567293276439,
      "grad_norm": 2.069676595962534,
      "learning_rate": 1.0764160329400167e-06,
      "loss": 0.5081,
      "step": 2939
    },
    {
      "epoch": 0.2712489908891708,
      "grad_norm": 1.7704229224283745,
      "learning_rate": 1.076285887504928e-06,
      "loss": 0.84,
      "step": 2940
    },
    {
      "epoch": 0.2712489908891708,
      "eval_GEN Loss": 0.47121885418891907,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.4136001765727997,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9354567527770996,
      "eval_runtime": 56.1669,
      "eval_samples_per_second": 1.157,
      "eval_steps_per_second": 0.089,
      "step": 2940
    },
    {
      "epoch": 0.27134125245069773,
      "grad_norm": 2.1653525888719827,
      "learning_rate": 1.0761557011736018e-06,
      "loss": 0.6645,
      "step": 2941
    },
    {
      "epoch": 0.27143351401222465,
      "grad_norm": 2.854016505791687,
      "learning_rate": 1.0760254739578358e-06,
      "loss": 0.9773,
      "step": 2942
    },
    {
      "epoch": 0.27152577557375157,
      "grad_norm": 1.657128474618374,
      "learning_rate": 1.0758952058694313e-06,
      "loss": 0.7026,
      "step": 2943
    },
    {
      "epoch": 0.2716180371352785,
      "grad_norm": 1.915486511031709,
      "learning_rate": 1.075764896920194e-06,
      "loss": 0.8005,
      "step": 2944
    },
    {
      "epoch": 0.27171029869680546,
      "grad_norm": 2.2045174048833958,
      "learning_rate": 1.0756345471219319e-06,
      "loss": 0.8616,
      "step": 2945
    },
    {
      "epoch": 0.27171029869680546,
      "eval_GEN Loss": 0.4716050326824188,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.4085181653499603,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9379807710647583,
      "eval_runtime": 55.8446,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 2945
    },
    {
      "epoch": 0.2718025602583324,
      "grad_norm": 1.9954431213998831,
      "learning_rate": 1.0755041564864577e-06,
      "loss": 0.678,
      "step": 2946
    },
    {
      "epoch": 0.2718948218198593,
      "grad_norm": 1.985754389146779,
      "learning_rate": 1.075373725025588e-06,
      "loss": 0.7281,
      "step": 2947
    },
    {
      "epoch": 0.2719870833813862,
      "grad_norm": 2.7574089347205066,
      "learning_rate": 1.0752432527511421e-06,
      "loss": 0.8584,
      "step": 2948
    },
    {
      "epoch": 0.27207934494291314,
      "grad_norm": 2.2764212700991937,
      "learning_rate": 1.0751127396749445e-06,
      "loss": 0.7423,
      "step": 2949
    },
    {
      "epoch": 0.2721716065044401,
      "grad_norm": 1.6483397800493493,
      "learning_rate": 1.0749821858088215e-06,
      "loss": 0.7096,
      "step": 2950
    },
    {
      "epoch": 0.2721716065044401,
      "eval_GEN Loss": 0.4714268743991852,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.4028923511505127,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9395432472229004,
      "eval_runtime": 55.8972,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 2950
    },
    {
      "epoch": 0.27226386806596703,
      "grad_norm": 2.456810661767572,
      "learning_rate": 1.0748515911646047e-06,
      "loss": 0.8599,
      "step": 2951
    },
    {
      "epoch": 0.27235612962749395,
      "grad_norm": 1.614728262049482,
      "learning_rate": 1.0747209557541286e-06,
      "loss": 0.5507,
      "step": 2952
    },
    {
      "epoch": 0.27244839118902087,
      "grad_norm": 2.3540574691967384,
      "learning_rate": 1.0745902795892315e-06,
      "loss": 0.7493,
      "step": 2953
    },
    {
      "epoch": 0.2725406527505478,
      "grad_norm": 2.230432170743013,
      "learning_rate": 1.0744595626817557e-06,
      "loss": 0.763,
      "step": 2954
    },
    {
      "epoch": 0.27263291431207476,
      "grad_norm": 2.2265999218489303,
      "learning_rate": 1.074328805043547e-06,
      "loss": 0.7823,
      "step": 2955
    },
    {
      "epoch": 0.27263291431207476,
      "eval_GEN Loss": 0.470962792634964,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.40174680948257446,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9305288195610046,
      "eval_runtime": 56.1549,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 2955
    },
    {
      "epoch": 0.2727251758736017,
      "grad_norm": 1.8371295485989962,
      "learning_rate": 1.0741980066864545e-06,
      "loss": 0.6203,
      "step": 2956
    },
    {
      "epoch": 0.2728174374351286,
      "grad_norm": 2.1076505355641943,
      "learning_rate": 1.0740671676223319e-06,
      "loss": 0.8231,
      "step": 2957
    },
    {
      "epoch": 0.2729096989966555,
      "grad_norm": 1.6193518676565852,
      "learning_rate": 1.0739362878630355e-06,
      "loss": 0.5592,
      "step": 2958
    },
    {
      "epoch": 0.27300196055818243,
      "grad_norm": 1.6979882137639126,
      "learning_rate": 1.073805367420426e-06,
      "loss": 0.6869,
      "step": 2959
    },
    {
      "epoch": 0.27309422211970935,
      "grad_norm": 1.3944447627316257,
      "learning_rate": 1.073674406306368e-06,
      "loss": 0.517,
      "step": 2960
    },
    {
      "epoch": 0.27309422211970935,
      "eval_GEN Loss": 0.4710898995399475,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3922373950481415,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9268028736114502,
      "eval_runtime": 56.0061,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 2960
    },
    {
      "epoch": 0.2731864836812363,
      "grad_norm": 2.980974654172515,
      "learning_rate": 1.0735434045327287e-06,
      "loss": 0.7722,
      "step": 2961
    },
    {
      "epoch": 0.27327874524276324,
      "grad_norm": 1.8456992095491502,
      "learning_rate": 1.07341236211138e-06,
      "loss": 0.7864,
      "step": 2962
    },
    {
      "epoch": 0.27337100680429016,
      "grad_norm": 1.8784025473858066,
      "learning_rate": 1.0732812790541977e-06,
      "loss": 0.721,
      "step": 2963
    },
    {
      "epoch": 0.2734632683658171,
      "grad_norm": 2.066693689693836,
      "learning_rate": 1.07315015537306e-06,
      "loss": 0.8374,
      "step": 2964
    },
    {
      "epoch": 0.273555529927344,
      "grad_norm": 1.7295788004467523,
      "learning_rate": 1.0730189910798497e-06,
      "loss": 0.446,
      "step": 2965
    },
    {
      "epoch": 0.273555529927344,
      "eval_GEN Loss": 0.4710273742675781,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3842709958553314,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9241586327552795,
      "eval_runtime": 55.947,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 2965
    },
    {
      "epoch": 0.273647791488871,
      "grad_norm": 1.9123825715438492,
      "learning_rate": 1.0728877861864534e-06,
      "loss": 0.8068,
      "step": 2966
    },
    {
      "epoch": 0.2737400530503979,
      "grad_norm": 2.253684877950857,
      "learning_rate": 1.072756540704761e-06,
      "loss": 0.7399,
      "step": 2967
    },
    {
      "epoch": 0.2738323146119248,
      "grad_norm": 2.3267420812374087,
      "learning_rate": 1.0726252546466659e-06,
      "loss": 0.6644,
      "step": 2968
    },
    {
      "epoch": 0.27392457617345173,
      "grad_norm": 2.1594839660895744,
      "learning_rate": 1.072493928024066e-06,
      "loss": 0.9195,
      "step": 2969
    },
    {
      "epoch": 0.27401683773497865,
      "grad_norm": 1.615247046523298,
      "learning_rate": 1.0723625608488617e-06,
      "loss": 0.6362,
      "step": 2970
    },
    {
      "epoch": 0.27401683773497865,
      "eval_GEN Loss": 0.4714062213897705,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM Loss": 0.3864392936229706,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9131009578704834,
      "eval_runtime": 56.0699,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 2970
    },
    {
      "epoch": 0.2741090992965056,
      "grad_norm": 3.6703664972373864,
      "learning_rate": 1.072231153132958e-06,
      "loss": 1.1565,
      "step": 2971
    },
    {
      "epoch": 0.27420136085803254,
      "grad_norm": 1.665237552315827,
      "learning_rate": 1.0720997048882632e-06,
      "loss": 0.5983,
      "step": 2972
    },
    {
      "epoch": 0.27429362241955946,
      "grad_norm": 3.053703069908747,
      "learning_rate": 1.0719682161266894e-06,
      "loss": 0.5336,
      "step": 2973
    },
    {
      "epoch": 0.2743858839810864,
      "grad_norm": 1.7050088474819154,
      "learning_rate": 1.0718366868601526e-06,
      "loss": 0.8571,
      "step": 2974
    },
    {
      "epoch": 0.2744781455426133,
      "grad_norm": 1.8257498753652626,
      "learning_rate": 1.0717051171005715e-06,
      "loss": 0.8708,
      "step": 2975
    },
    {
      "epoch": 0.2744781455426133,
      "eval_GEN Loss": 0.4726153016090393,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.41426679491996765,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.915745198726654,
      "eval_runtime": 56.6605,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 2975
    },
    {
      "epoch": 0.2745704071041402,
      "grad_norm": 2.0781769843727673,
      "learning_rate": 1.0715735068598698e-06,
      "loss": 0.9534,
      "step": 2976
    },
    {
      "epoch": 0.2746626686656672,
      "grad_norm": 1.3498716611053658,
      "learning_rate": 1.071441856149974e-06,
      "loss": 0.6658,
      "step": 2977
    },
    {
      "epoch": 0.2747549302271941,
      "grad_norm": 2.3352739141914896,
      "learning_rate": 1.0713101649828143e-06,
      "loss": 0.8961,
      "step": 2978
    },
    {
      "epoch": 0.274847191788721,
      "grad_norm": 1.893398868696727,
      "learning_rate": 1.0711784333703252e-06,
      "loss": 0.827,
      "step": 2979
    },
    {
      "epoch": 0.27493945335024794,
      "grad_norm": 2.4254582277827876,
      "learning_rate": 1.071046661324444e-06,
      "loss": 0.7849,
      "step": 2980
    },
    {
      "epoch": 0.27493945335024794,
      "eval_GEN Loss": 0.47261038422584534,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4217749238014221,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9133413434028625,
      "eval_runtime": 56.6778,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 2980
    },
    {
      "epoch": 0.27503171491177486,
      "grad_norm": 2.051485330504249,
      "learning_rate": 1.0709148488571123e-06,
      "loss": 0.6959,
      "step": 2981
    },
    {
      "epoch": 0.27512397647330183,
      "grad_norm": 1.8124372530176573,
      "learning_rate": 1.070782995980275e-06,
      "loss": 0.6654,
      "step": 2982
    },
    {
      "epoch": 0.27521623803482875,
      "grad_norm": 1.789274532211207,
      "learning_rate": 1.070651102705881e-06,
      "loss": 0.8282,
      "step": 2983
    },
    {
      "epoch": 0.27530849959635567,
      "grad_norm": 2.4247167739441218,
      "learning_rate": 1.0705191690458826e-06,
      "loss": 0.6691,
      "step": 2984
    },
    {
      "epoch": 0.2754007611578826,
      "grad_norm": 1.9362840055640174,
      "learning_rate": 1.0703871950122362e-06,
      "loss": 0.7744,
      "step": 2985
    },
    {
      "epoch": 0.2754007611578826,
      "eval_GEN Loss": 0.4706338346004486,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.4156058728694916,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9146634340286255,
      "eval_runtime": 56.5735,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 2985
    },
    {
      "epoch": 0.2754930227194095,
      "grad_norm": 1.7661738455074483,
      "learning_rate": 1.0702551806169009e-06,
      "loss": 0.6302,
      "step": 2986
    },
    {
      "epoch": 0.2755852842809365,
      "grad_norm": 2.179368130907011,
      "learning_rate": 1.07012312587184e-06,
      "loss": 0.826,
      "step": 2987
    },
    {
      "epoch": 0.2756775458424634,
      "grad_norm": 1.3227171925333114,
      "learning_rate": 1.0699910307890213e-06,
      "loss": 0.5974,
      "step": 2988
    },
    {
      "epoch": 0.2757698074039903,
      "grad_norm": 1.5215068251231954,
      "learning_rate": 1.0698588953804146e-06,
      "loss": 0.7154,
      "step": 2989
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 1.8611650992188542,
      "learning_rate": 1.0697267196579946e-06,
      "loss": 0.7443,
      "step": 2990
    },
    {
      "epoch": 0.27586206896551724,
      "eval_GEN Loss": 0.4698081314563751,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3971595764160156,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9191105961799622,
      "eval_runtime": 56.8346,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 2990
    },
    {
      "epoch": 0.27595433052704416,
      "grad_norm": 1.5731395456884663,
      "learning_rate": 1.0695945036337394e-06,
      "loss": 0.5874,
      "step": 2991
    },
    {
      "epoch": 0.2760465920885711,
      "grad_norm": 1.8458591760521923,
      "learning_rate": 1.0694622473196303e-06,
      "loss": 0.8392,
      "step": 2992
    },
    {
      "epoch": 0.27613885365009805,
      "grad_norm": 1.5301326642721071,
      "learning_rate": 1.069329950727653e-06,
      "loss": 0.7545,
      "step": 2993
    },
    {
      "epoch": 0.27623111521162497,
      "grad_norm": 1.4687202099750942,
      "learning_rate": 1.0691976138697959e-06,
      "loss": 0.7071,
      "step": 2994
    },
    {
      "epoch": 0.2763233767731519,
      "grad_norm": 1.6107935879435766,
      "learning_rate": 1.069065236758052e-06,
      "loss": 0.5573,
      "step": 2995
    },
    {
      "epoch": 0.2763233767731519,
      "eval_GEN Loss": 0.47077441215515137,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3876074254512787,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9328725934028625,
      "eval_runtime": 56.6805,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 2995
    },
    {
      "epoch": 0.2764156383346788,
      "grad_norm": 3.191370588523095,
      "learning_rate": 1.0689328194044176e-06,
      "loss": 0.8169,
      "step": 2996
    },
    {
      "epoch": 0.2765078998962057,
      "grad_norm": 1.2001886420614896,
      "learning_rate": 1.0688003618208922e-06,
      "loss": 0.5587,
      "step": 2997
    },
    {
      "epoch": 0.2766001614577327,
      "grad_norm": 2.323862866865989,
      "learning_rate": 1.0686678640194791e-06,
      "loss": 0.7312,
      "step": 2998
    },
    {
      "epoch": 0.2766924230192596,
      "grad_norm": 2.321497592497748,
      "learning_rate": 1.068535326012186e-06,
      "loss": 0.8557,
      "step": 2999
    },
    {
      "epoch": 0.27678468458078653,
      "grad_norm": 2.0297776260028892,
      "learning_rate": 1.0684027478110238e-06,
      "loss": 0.7923,
      "step": 3000
    },
    {
      "epoch": 0.27678468458078653,
      "eval_GEN Loss": 0.47090959548950195,
      "eval_GEN top-5 accuracy": 0.9850942569048663,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.38966938853263855,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9231970906257629,
      "eval_runtime": 56.8399,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3000
    },
    {
      "epoch": 0.27687694614231345,
      "grad_norm": 1.6618185518380078,
      "learning_rate": 1.0682701294280066e-06,
      "loss": 0.7444,
      "step": 3001
    },
    {
      "epoch": 0.27696920770384037,
      "grad_norm": 2.476708643456072,
      "learning_rate": 1.0681374708751525e-06,
      "loss": 0.901,
      "step": 3002
    },
    {
      "epoch": 0.27706146926536734,
      "grad_norm": 1.6390607184185797,
      "learning_rate": 1.0680047721644831e-06,
      "loss": 0.6479,
      "step": 3003
    },
    {
      "epoch": 0.27715373082689426,
      "grad_norm": 1.2077098188787156,
      "learning_rate": 1.0678720333080242e-06,
      "loss": 0.6578,
      "step": 3004
    },
    {
      "epoch": 0.2772459923884212,
      "grad_norm": 2.095117818126065,
      "learning_rate": 1.0677392543178045e-06,
      "loss": 0.8181,
      "step": 3005
    },
    {
      "epoch": 0.2772459923884212,
      "eval_GEN Loss": 0.47230812907218933,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.4182332158088684,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9130408763885498,
      "eval_runtime": 55.9194,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3005
    },
    {
      "epoch": 0.2773382539499481,
      "grad_norm": 2.6604775050617984,
      "learning_rate": 1.0676064352058566e-06,
      "loss": 0.6083,
      "step": 3006
    },
    {
      "epoch": 0.277430515511475,
      "grad_norm": 1.3858070793868345,
      "learning_rate": 1.0674735759842166e-06,
      "loss": 0.5896,
      "step": 3007
    },
    {
      "epoch": 0.27752277707300194,
      "grad_norm": 3.880363472013846,
      "learning_rate": 1.0673406766649252e-06,
      "loss": 0.8189,
      "step": 3008
    },
    {
      "epoch": 0.2776150386345289,
      "grad_norm": 2.310164277193226,
      "learning_rate": 1.067207737260025e-06,
      "loss": 0.7813,
      "step": 3009
    },
    {
      "epoch": 0.27770730019605583,
      "grad_norm": 2.441424253660931,
      "learning_rate": 1.067074757781564e-06,
      "loss": 0.7209,
      "step": 3010
    },
    {
      "epoch": 0.27770730019605583,
      "eval_GEN Loss": 0.4707789719104767,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.40797895193099976,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9105167984962463,
      "eval_runtime": 55.7052,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3010
    },
    {
      "epoch": 0.27779956175758275,
      "grad_norm": 1.7815883948680813,
      "learning_rate": 1.0669417382415923e-06,
      "loss": 0.7369,
      "step": 3011
    },
    {
      "epoch": 0.27789182331910967,
      "grad_norm": 1.5110397906159032,
      "learning_rate": 1.0668086786521648e-06,
      "loss": 0.6315,
      "step": 3012
    },
    {
      "epoch": 0.2779840848806366,
      "grad_norm": 1.9341888127529423,
      "learning_rate": 1.0666755790253393e-06,
      "loss": 0.6519,
      "step": 3013
    },
    {
      "epoch": 0.27807634644216356,
      "grad_norm": 1.9835501065707783,
      "learning_rate": 1.0665424393731777e-06,
      "loss": 0.7808,
      "step": 3014
    },
    {
      "epoch": 0.2781686080036905,
      "grad_norm": 2.3280147083623075,
      "learning_rate": 1.0664092597077455e-06,
      "loss": 0.8808,
      "step": 3015
    },
    {
      "epoch": 0.2781686080036905,
      "eval_GEN Loss": 0.4707986116409302,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.38820645213127136,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9114783406257629,
      "eval_runtime": 56.7691,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3015
    },
    {
      "epoch": 0.2782608695652174,
      "grad_norm": 2.8357232108307215,
      "learning_rate": 1.066276040041111e-06,
      "loss": 0.861,
      "step": 3016
    },
    {
      "epoch": 0.2783531311267443,
      "grad_norm": 1.8295674414500154,
      "learning_rate": 1.0661427803853473e-06,
      "loss": 0.6751,
      "step": 3017
    },
    {
      "epoch": 0.27844539268827123,
      "grad_norm": 2.2622370843638877,
      "learning_rate": 1.0660094807525305e-06,
      "loss": 0.7406,
      "step": 3018
    },
    {
      "epoch": 0.2785376542497982,
      "grad_norm": 2.195596812019375,
      "learning_rate": 1.0658761411547404e-06,
      "loss": 0.7446,
      "step": 3019
    },
    {
      "epoch": 0.2786299158113251,
      "grad_norm": 1.621731708015415,
      "learning_rate": 1.0657427616040605e-06,
      "loss": 0.6217,
      "step": 3020
    },
    {
      "epoch": 0.2786299158113251,
      "eval_GEN Loss": 0.47051486372947693,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.4024726450443268,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9075720906257629,
      "eval_runtime": 56.7361,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3020
    },
    {
      "epoch": 0.27872217737285204,
      "grad_norm": 2.049016492117123,
      "learning_rate": 1.0656093421125777e-06,
      "loss": 0.8602,
      "step": 3021
    },
    {
      "epoch": 0.27881443893437896,
      "grad_norm": 2.0497393145955685,
      "learning_rate": 1.0654758826923827e-06,
      "loss": 0.7315,
      "step": 3022
    },
    {
      "epoch": 0.2789067004959059,
      "grad_norm": 2.122846118333475,
      "learning_rate": 1.0653423833555701e-06,
      "loss": 0.7259,
      "step": 3023
    },
    {
      "epoch": 0.2789989620574328,
      "grad_norm": 2.1648262068114836,
      "learning_rate": 1.0652088441142375e-06,
      "loss": 0.6142,
      "step": 3024
    },
    {
      "epoch": 0.27909122361895977,
      "grad_norm": 1.8756445899355763,
      "learning_rate": 1.0650752649804866e-06,
      "loss": 0.818,
      "step": 3025
    },
    {
      "epoch": 0.27909122361895977,
      "eval_GEN Loss": 0.4718669354915619,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.3990402817726135,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9111778736114502,
      "eval_runtime": 55.7786,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3025
    },
    {
      "epoch": 0.2791834851804867,
      "grad_norm": 1.4974941619338655,
      "learning_rate": 1.0649416459664224e-06,
      "loss": 0.6817,
      "step": 3026
    },
    {
      "epoch": 0.2792757467420136,
      "grad_norm": 2.7454544109571524,
      "learning_rate": 1.064807987084154e-06,
      "loss": 0.7396,
      "step": 3027
    },
    {
      "epoch": 0.2793680083035405,
      "grad_norm": 1.5025185893054946,
      "learning_rate": 1.0646742883457932e-06,
      "loss": 0.6328,
      "step": 3028
    },
    {
      "epoch": 0.27946026986506745,
      "grad_norm": 1.691085999858614,
      "learning_rate": 1.0645405497634566e-06,
      "loss": 0.6876,
      "step": 3029
    },
    {
      "epoch": 0.2795525314265944,
      "grad_norm": 3.3312380732250904,
      "learning_rate": 1.0644067713492635e-06,
      "loss": 0.8739,
      "step": 3030
    },
    {
      "epoch": 0.2795525314265944,
      "eval_GEN Loss": 0.4710805118083954,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3732966184616089,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9143028855323792,
      "eval_runtime": 55.7757,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3030
    },
    {
      "epoch": 0.27964479298812134,
      "grad_norm": 1.8158589611092544,
      "learning_rate": 1.064272953115337e-06,
      "loss": 0.7647,
      "step": 3031
    },
    {
      "epoch": 0.27973705454964826,
      "grad_norm": 3.9458975882161513,
      "learning_rate": 1.064139095073804e-06,
      "loss": 1.2141,
      "step": 3032
    },
    {
      "epoch": 0.2798293161111752,
      "grad_norm": 2.5882358601325253,
      "learning_rate": 1.0640051972367955e-06,
      "loss": 0.6481,
      "step": 3033
    },
    {
      "epoch": 0.2799215776727021,
      "grad_norm": 1.9084181803913842,
      "learning_rate": 1.0638712596164445e-06,
      "loss": 0.6903,
      "step": 3034
    },
    {
      "epoch": 0.280013839234229,
      "grad_norm": 2.644414617579794,
      "learning_rate": 1.0637372822248892e-06,
      "loss": 0.9132,
      "step": 3035
    },
    {
      "epoch": 0.280013839234229,
      "eval_GEN Loss": 0.46965330839157104,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9047619047619048,
      "eval_PRM F1 AUC": 0.7621791513881613,
      "eval_PRM F1 Neg": 0.6363636363636364,
      "eval_PRM Loss": 0.36662760376930237,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.8941176470588236,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9252403974533081,
      "eval_runtime": 55.7503,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 3035
    },
    {
      "epoch": 0.280106100795756,
      "grad_norm": 1.9822324253651207,
      "learning_rate": 1.063603265074271e-06,
      "loss": 0.6281,
      "step": 3036
    },
    {
      "epoch": 0.2801983623572829,
      "grad_norm": 1.923125657012463,
      "learning_rate": 1.0634692081767344e-06,
      "loss": 0.7492,
      "step": 3037
    },
    {
      "epoch": 0.2802906239188098,
      "grad_norm": 2.3429376529638755,
      "learning_rate": 1.0633351115444281e-06,
      "loss": 0.8975,
      "step": 3038
    },
    {
      "epoch": 0.28038288548033674,
      "grad_norm": 2.050496352175447,
      "learning_rate": 1.063200975189504e-06,
      "loss": 0.5983,
      "step": 3039
    },
    {
      "epoch": 0.28047514704186366,
      "grad_norm": 2.4762319883475437,
      "learning_rate": 1.0630667991241178e-06,
      "loss": 1.0166,
      "step": 3040
    },
    {
      "epoch": 0.28047514704186366,
      "eval_GEN Loss": 0.46793049573898315,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3690236210823059,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9129807949066162,
      "eval_runtime": 56.4612,
      "eval_samples_per_second": 1.151,
      "eval_steps_per_second": 0.089,
      "step": 3040
    },
    {
      "epoch": 0.28056740860339063,
      "grad_norm": 2.302922509664043,
      "learning_rate": 1.0629325833604288e-06,
      "loss": 0.6984,
      "step": 3041
    },
    {
      "epoch": 0.28065967016491755,
      "grad_norm": 1.7904631579843462,
      "learning_rate": 1.0627983279105996e-06,
      "loss": 0.7855,
      "step": 3042
    },
    {
      "epoch": 0.28075193172644447,
      "grad_norm": 1.7484699054593347,
      "learning_rate": 1.062664032786797e-06,
      "loss": 0.6728,
      "step": 3043
    },
    {
      "epoch": 0.2808441932879714,
      "grad_norm": 2.1338057527343115,
      "learning_rate": 1.0625296980011909e-06,
      "loss": 0.8728,
      "step": 3044
    },
    {
      "epoch": 0.2809364548494983,
      "grad_norm": 1.4213221012684427,
      "learning_rate": 1.0623953235659547e-06,
      "loss": 0.6779,
      "step": 3045
    },
    {
      "epoch": 0.2809364548494983,
      "eval_GEN Loss": 0.4686720371246338,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.381797730922699,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9061899185180664,
      "eval_runtime": 56.9058,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3045
    },
    {
      "epoch": 0.2810287164110253,
      "grad_norm": 2.1135693186589215,
      "learning_rate": 1.0622609094932658e-06,
      "loss": 0.7909,
      "step": 3046
    },
    {
      "epoch": 0.2811209779725522,
      "grad_norm": 1.7274339732855157,
      "learning_rate": 1.0621264557953052e-06,
      "loss": 0.6717,
      "step": 3047
    },
    {
      "epoch": 0.2812132395340791,
      "grad_norm": 1.9125030429274472,
      "learning_rate": 1.061991962484257e-06,
      "loss": 0.6906,
      "step": 3048
    },
    {
      "epoch": 0.28130550109560604,
      "grad_norm": 2.144051575041993,
      "learning_rate": 1.0618574295723095e-06,
      "loss": 0.6694,
      "step": 3049
    },
    {
      "epoch": 0.28139776265713295,
      "grad_norm": 1.84832154181422,
      "learning_rate": 1.0617228570716538e-06,
      "loss": 0.5958,
      "step": 3050
    },
    {
      "epoch": 0.28139776265713295,
      "eval_GEN Loss": 0.4657774865627289,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.38704806566238403,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.900661051273346,
      "eval_runtime": 56.6353,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 3050
    },
    {
      "epoch": 0.2814900242186599,
      "grad_norm": 2.31598315278293,
      "learning_rate": 1.0615882449944856e-06,
      "loss": 0.7841,
      "step": 3051
    },
    {
      "epoch": 0.28158228578018685,
      "grad_norm": 2.1371664471350713,
      "learning_rate": 1.0614535933530033e-06,
      "loss": 0.7341,
      "step": 3052
    },
    {
      "epoch": 0.28167454734171377,
      "grad_norm": 2.4419606666641105,
      "learning_rate": 1.0613189021594096e-06,
      "loss": 0.9996,
      "step": 3053
    },
    {
      "epoch": 0.2817668089032407,
      "grad_norm": 2.1011073182541566,
      "learning_rate": 1.06118417142591e-06,
      "loss": 0.6853,
      "step": 3054
    },
    {
      "epoch": 0.2818590704647676,
      "grad_norm": 2.086353803274598,
      "learning_rate": 1.0610494011647143e-06,
      "loss": 0.6906,
      "step": 3055
    },
    {
      "epoch": 0.2818590704647676,
      "eval_GEN Loss": 0.4659087359905243,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3741576671600342,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9114783406257629,
      "eval_runtime": 56.863,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3055
    },
    {
      "epoch": 0.2819513320262945,
      "grad_norm": 2.871683684942679,
      "learning_rate": 1.0609145913880355e-06,
      "loss": 0.7057,
      "step": 3056
    },
    {
      "epoch": 0.2820435935878215,
      "grad_norm": 1.8595216782239863,
      "learning_rate": 1.0607797421080902e-06,
      "loss": 0.6267,
      "step": 3057
    },
    {
      "epoch": 0.2821358551493484,
      "grad_norm": 2.536029962732933,
      "learning_rate": 1.0606448533370988e-06,
      "loss": 0.7527,
      "step": 3058
    },
    {
      "epoch": 0.28222811671087533,
      "grad_norm": 1.487910259640829,
      "learning_rate": 1.0605099250872852e-06,
      "loss": 0.7512,
      "step": 3059
    },
    {
      "epoch": 0.28232037827240225,
      "grad_norm": 2.0658230844092147,
      "learning_rate": 1.0603749573708765e-06,
      "loss": 0.8208,
      "step": 3060
    },
    {
      "epoch": 0.28232037827240225,
      "eval_GEN Loss": 0.4652542173862457,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.375637412071228,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9095553159713745,
      "eval_runtime": 57.3191,
      "eval_samples_per_second": 1.134,
      "eval_steps_per_second": 0.087,
      "step": 3060
    },
    {
      "epoch": 0.28241263983392917,
      "grad_norm": 1.7626753024284254,
      "learning_rate": 1.060239950200104e-06,
      "loss": 0.8132,
      "step": 3061
    },
    {
      "epoch": 0.28250490139545614,
      "grad_norm": 1.451956880870587,
      "learning_rate": 1.0601049035872021e-06,
      "loss": 0.5288,
      "step": 3062
    },
    {
      "epoch": 0.28259716295698306,
      "grad_norm": 3.030208795825869,
      "learning_rate": 1.0599698175444092e-06,
      "loss": 0.7612,
      "step": 3063
    },
    {
      "epoch": 0.28268942451851,
      "grad_norm": 2.182971082205315,
      "learning_rate": 1.0598346920839664e-06,
      "loss": 0.6718,
      "step": 3064
    },
    {
      "epoch": 0.2827816860800369,
      "grad_norm": 1.7677658952444775,
      "learning_rate": 1.0596995272181195e-06,
      "loss": 0.7722,
      "step": 3065
    },
    {
      "epoch": 0.2827816860800369,
      "eval_GEN Loss": 0.46604636311531067,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.37874099612236023,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.907151460647583,
      "eval_runtime": 56.6518,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3065
    },
    {
      "epoch": 0.2828739476415638,
      "grad_norm": 2.2423082403953045,
      "learning_rate": 1.0595643229591174e-06,
      "loss": 0.675,
      "step": 3066
    },
    {
      "epoch": 0.28296620920309073,
      "grad_norm": 1.5534307236488345,
      "learning_rate": 1.059429079319212e-06,
      "loss": 0.6772,
      "step": 3067
    },
    {
      "epoch": 0.2830584707646177,
      "grad_norm": 1.9817521697600702,
      "learning_rate": 1.05929379631066e-06,
      "loss": 0.6152,
      "step": 3068
    },
    {
      "epoch": 0.2831507323261446,
      "grad_norm": 1.55341060271511,
      "learning_rate": 1.0591584739457203e-06,
      "loss": 0.7239,
      "step": 3069
    },
    {
      "epoch": 0.28324299388767155,
      "grad_norm": 2.2594796869473357,
      "learning_rate": 1.0590231122366565e-06,
      "loss": 0.7845,
      "step": 3070
    },
    {
      "epoch": 0.28324299388767155,
      "eval_GEN Loss": 0.4675382971763611,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3763037919998169,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9029446840286255,
      "eval_runtime": 56.8095,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3070
    },
    {
      "epoch": 0.28333525544919846,
      "grad_norm": 1.5711109656493212,
      "learning_rate": 1.0588877111957348e-06,
      "loss": 0.5559,
      "step": 3071
    },
    {
      "epoch": 0.2834275170107254,
      "grad_norm": 2.3612479186297133,
      "learning_rate": 1.0587522708352262e-06,
      "loss": 0.9523,
      "step": 3072
    },
    {
      "epoch": 0.28351977857225236,
      "grad_norm": 2.175565693130009,
      "learning_rate": 1.058616791167404e-06,
      "loss": 0.7909,
      "step": 3073
    },
    {
      "epoch": 0.2836120401337793,
      "grad_norm": 1.7017775914052056,
      "learning_rate": 1.0584812722045455e-06,
      "loss": 0.7176,
      "step": 3074
    },
    {
      "epoch": 0.2837043016953062,
      "grad_norm": 2.6788065268577643,
      "learning_rate": 1.0583457139589317e-06,
      "loss": 0.9553,
      "step": 3075
    },
    {
      "epoch": 0.2837043016953062,
      "eval_GEN Loss": 0.46606746315956116,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3762892484664917,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9116586446762085,
      "eval_runtime": 56.7424,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3075
    },
    {
      "epoch": 0.2837965632568331,
      "grad_norm": 2.914097442064055,
      "learning_rate": 1.0582101164428474e-06,
      "loss": 0.869,
      "step": 3076
    },
    {
      "epoch": 0.28388882481836003,
      "grad_norm": 2.88039190454395,
      "learning_rate": 1.0580744796685804e-06,
      "loss": 0.9041,
      "step": 3077
    },
    {
      "epoch": 0.283981086379887,
      "grad_norm": 2.363975874657006,
      "learning_rate": 1.0579388036484223e-06,
      "loss": 0.932,
      "step": 3078
    },
    {
      "epoch": 0.2840733479414139,
      "grad_norm": 2.0158679523647995,
      "learning_rate": 1.0578030883946685e-06,
      "loss": 0.7301,
      "step": 3079
    },
    {
      "epoch": 0.28416560950294084,
      "grad_norm": 1.8537815026079734,
      "learning_rate": 1.0576673339196176e-06,
      "loss": 0.8477,
      "step": 3080
    },
    {
      "epoch": 0.28416560950294084,
      "eval_GEN Loss": 0.46597346663475037,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3801872730255127,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8949519395828247,
      "eval_runtime": 56.6931,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3080
    },
    {
      "epoch": 0.28425787106446776,
      "grad_norm": 1.8994480286322213,
      "learning_rate": 1.0575315402355718e-06,
      "loss": 0.6243,
      "step": 3081
    },
    {
      "epoch": 0.2843501326259947,
      "grad_norm": 3.2779745014244988,
      "learning_rate": 1.057395707354837e-06,
      "loss": 1.0052,
      "step": 3082
    },
    {
      "epoch": 0.2844423941875216,
      "grad_norm": 2.492639473823188,
      "learning_rate": 1.0572598352897227e-06,
      "loss": 0.8754,
      "step": 3083
    },
    {
      "epoch": 0.28453465574904857,
      "grad_norm": 1.900689999724326,
      "learning_rate": 1.0571239240525414e-06,
      "loss": 0.7231,
      "step": 3084
    },
    {
      "epoch": 0.2846269173105755,
      "grad_norm": 2.2859324540427357,
      "learning_rate": 1.0569879736556102e-06,
      "loss": 0.859,
      "step": 3085
    },
    {
      "epoch": 0.2846269173105755,
      "eval_GEN Loss": 0.46712419390678406,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.4000728130340576,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8842548131942749,
      "eval_runtime": 56.7942,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3085
    },
    {
      "epoch": 0.2847191788721024,
      "grad_norm": 2.1509835400542827,
      "learning_rate": 1.056851984111249e-06,
      "loss": 0.8304,
      "step": 3086
    },
    {
      "epoch": 0.2848114404336293,
      "grad_norm": 2.1363475596955626,
      "learning_rate": 1.056715955431781e-06,
      "loss": 0.5765,
      "step": 3087
    },
    {
      "epoch": 0.28490370199515624,
      "grad_norm": 1.7308635146031295,
      "learning_rate": 1.0565798876295337e-06,
      "loss": 0.6607,
      "step": 3088
    },
    {
      "epoch": 0.2849959635566832,
      "grad_norm": 1.582608132372354,
      "learning_rate": 1.0564437807168376e-06,
      "loss": 0.8734,
      "step": 3089
    },
    {
      "epoch": 0.28508822511821014,
      "grad_norm": 1.4671379084193517,
      "learning_rate": 1.0563076347060273e-06,
      "loss": 0.6062,
      "step": 3090
    },
    {
      "epoch": 0.28508822511821014,
      "eval_GEN Loss": 0.4661886692047119,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.38512077927589417,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8808293342590332,
      "eval_runtime": 56.7829,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3090
    },
    {
      "epoch": 0.28518048667973706,
      "grad_norm": 1.7688010433446533,
      "learning_rate": 1.0561714496094398e-06,
      "loss": 0.727,
      "step": 3091
    },
    {
      "epoch": 0.285272748241264,
      "grad_norm": 1.8775239804600636,
      "learning_rate": 1.056035225439417e-06,
      "loss": 0.6714,
      "step": 3092
    },
    {
      "epoch": 0.2853650098027909,
      "grad_norm": 1.9746959633598722,
      "learning_rate": 1.0558989622083038e-06,
      "loss": 0.8284,
      "step": 3093
    },
    {
      "epoch": 0.28545727136431787,
      "grad_norm": 1.5507223235366274,
      "learning_rate": 1.0557626599284481e-06,
      "loss": 0.6109,
      "step": 3094
    },
    {
      "epoch": 0.2855495329258448,
      "grad_norm": 1.7766831703362023,
      "learning_rate": 1.0556263186122022e-06,
      "loss": 0.8638,
      "step": 3095
    },
    {
      "epoch": 0.2855495329258448,
      "eval_GEN Loss": 0.4653873145580292,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3613568842411041,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8912259340286255,
      "eval_runtime": 55.9366,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3095
    },
    {
      "epoch": 0.2856417944873717,
      "grad_norm": 1.7554203841913272,
      "learning_rate": 1.0554899382719217e-06,
      "loss": 0.7518,
      "step": 3096
    },
    {
      "epoch": 0.2857340560488986,
      "grad_norm": 1.6844967445903287,
      "learning_rate": 1.0553535189199655e-06,
      "loss": 0.7656,
      "step": 3097
    },
    {
      "epoch": 0.28582631761042554,
      "grad_norm": 2.089667784863428,
      "learning_rate": 1.0552170605686957e-06,
      "loss": 0.8665,
      "step": 3098
    },
    {
      "epoch": 0.28591857917195246,
      "grad_norm": 1.6438299141886128,
      "learning_rate": 1.055080563230479e-06,
      "loss": 0.6618,
      "step": 3099
    },
    {
      "epoch": 0.28601084073347943,
      "grad_norm": 1.6914329780792354,
      "learning_rate": 1.0549440269176843e-06,
      "loss": 0.654,
      "step": 3100
    },
    {
      "epoch": 0.28601084073347943,
      "eval_GEN Loss": 0.4667896032333374,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3566409647464752,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8939903974533081,
      "eval_runtime": 55.8686,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3100
    },
    {
      "epoch": 0.28610310229500635,
      "grad_norm": 1.5334234556289796,
      "learning_rate": 1.0548074516426855e-06,
      "loss": 0.6701,
      "step": 3101
    },
    {
      "epoch": 0.28619536385653327,
      "grad_norm": 1.6714199493367639,
      "learning_rate": 1.0546708374178585e-06,
      "loss": 0.7958,
      "step": 3102
    },
    {
      "epoch": 0.2862876254180602,
      "grad_norm": 1.6422071550644641,
      "learning_rate": 1.0545341842555843e-06,
      "loss": 0.7805,
      "step": 3103
    },
    {
      "epoch": 0.2863798869795871,
      "grad_norm": 1.9429238807507643,
      "learning_rate": 1.0543974921682462e-06,
      "loss": 0.6407,
      "step": 3104
    },
    {
      "epoch": 0.2864721485411141,
      "grad_norm": 1.8132138305768448,
      "learning_rate": 1.0542607611682312e-06,
      "loss": 0.6952,
      "step": 3105
    },
    {
      "epoch": 0.2864721485411141,
      "eval_GEN Loss": 0.465804785490036,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.35782474279403687,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8822115659713745,
      "eval_runtime": 55.9472,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3105
    },
    {
      "epoch": 0.286564410102641,
      "grad_norm": 2.124122330337371,
      "learning_rate": 1.0541239912679306e-06,
      "loss": 0.6771,
      "step": 3106
    },
    {
      "epoch": 0.2866566716641679,
      "grad_norm": 1.554192333664422,
      "learning_rate": 1.0539871824797384e-06,
      "loss": 0.7065,
      "step": 3107
    },
    {
      "epoch": 0.28674893322569484,
      "grad_norm": 1.629955964226237,
      "learning_rate": 1.0538503348160523e-06,
      "loss": 0.7569,
      "step": 3108
    },
    {
      "epoch": 0.28684119478722175,
      "grad_norm": 1.5532291249377288,
      "learning_rate": 1.053713448289274e-06,
      "loss": 0.6305,
      "step": 3109
    },
    {
      "epoch": 0.2869334563487487,
      "grad_norm": 2.141857916910032,
      "learning_rate": 1.0535765229118084e-06,
      "loss": 0.7892,
      "step": 3110
    },
    {
      "epoch": 0.2869334563487487,
      "eval_GEN Loss": 0.4660208523273468,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.36910420656204224,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8732572197914124,
      "eval_runtime": 55.8231,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3110
    },
    {
      "epoch": 0.28702571791027565,
      "grad_norm": 2.1212484068932773,
      "learning_rate": 1.0534395586960634e-06,
      "loss": 0.7813,
      "step": 3111
    },
    {
      "epoch": 0.28711797947180256,
      "grad_norm": 2.087905572005798,
      "learning_rate": 1.0533025556544513e-06,
      "loss": 0.8203,
      "step": 3112
    },
    {
      "epoch": 0.2872102410333295,
      "grad_norm": 1.812317216913645,
      "learning_rate": 1.0531655137993876e-06,
      "loss": 0.7788,
      "step": 3113
    },
    {
      "epoch": 0.2873025025948564,
      "grad_norm": 1.8474439368782816,
      "learning_rate": 1.0530284331432911e-06,
      "loss": 0.7616,
      "step": 3114
    },
    {
      "epoch": 0.2873947641563833,
      "grad_norm": 1.8675715485348674,
      "learning_rate": 1.0528913136985841e-06,
      "loss": 0.8438,
      "step": 3115
    },
    {
      "epoch": 0.2873947641563833,
      "eval_GEN Loss": 0.46344852447509766,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.38797926902770996,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8698317408561707,
      "eval_runtime": 56.3747,
      "eval_samples_per_second": 1.153,
      "eval_steps_per_second": 0.089,
      "step": 3115
    },
    {
      "epoch": 0.2874870257179103,
      "grad_norm": 2.0242343361431367,
      "learning_rate": 1.0527541554776926e-06,
      "loss": 0.7094,
      "step": 3116
    },
    {
      "epoch": 0.2875792872794372,
      "grad_norm": 1.9460536316360604,
      "learning_rate": 1.0526169584930464e-06,
      "loss": 0.5945,
      "step": 3117
    },
    {
      "epoch": 0.28767154884096413,
      "grad_norm": 1.9319594951384043,
      "learning_rate": 1.0524797227570785e-06,
      "loss": 0.6974,
      "step": 3118
    },
    {
      "epoch": 0.28776381040249105,
      "grad_norm": 1.719496726115059,
      "learning_rate": 1.0523424482822251e-06,
      "loss": 0.7208,
      "step": 3119
    },
    {
      "epoch": 0.28785607196401797,
      "grad_norm": 2.424005694128972,
      "learning_rate": 1.0522051350809265e-06,
      "loss": 0.7763,
      "step": 3120
    },
    {
      "epoch": 0.28785607196401797,
      "eval_GEN Loss": 0.4616316556930542,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.38041043281555176,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.870312511920929,
      "eval_runtime": 56.6768,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3120
    },
    {
      "epoch": 0.28794833352554494,
      "grad_norm": 1.4211401390054537,
      "learning_rate": 1.0520677831656258e-06,
      "loss": 0.8117,
      "step": 3121
    },
    {
      "epoch": 0.28804059508707186,
      "grad_norm": 1.7596771331547327,
      "learning_rate": 1.0519303925487708e-06,
      "loss": 0.5949,
      "step": 3122
    },
    {
      "epoch": 0.2881328566485988,
      "grad_norm": 2.363982976402196,
      "learning_rate": 1.0517929632428113e-06,
      "loss": 0.7793,
      "step": 3123
    },
    {
      "epoch": 0.2882251182101257,
      "grad_norm": 1.9629677355913673,
      "learning_rate": 1.051655495260202e-06,
      "loss": 0.9329,
      "step": 3124
    },
    {
      "epoch": 0.2883173797716526,
      "grad_norm": 2.2197297120392405,
      "learning_rate": 1.0515179886134e-06,
      "loss": 0.8808,
      "step": 3125
    },
    {
      "epoch": 0.2883173797716526,
      "eval_GEN Loss": 0.45907220244407654,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3719245195388794,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8778846263885498,
      "eval_runtime": 55.6551,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 3125
    },
    {
      "epoch": 0.2884096413331796,
      "grad_norm": 1.4296998827142253,
      "learning_rate": 1.0513804433148664e-06,
      "loss": 0.6082,
      "step": 3126
    },
    {
      "epoch": 0.2885019028947065,
      "grad_norm": 2.0088710776882372,
      "learning_rate": 1.0512428593770662e-06,
      "loss": 0.8761,
      "step": 3127
    },
    {
      "epoch": 0.2885941644562334,
      "grad_norm": 2.556149882217034,
      "learning_rate": 1.051105236812467e-06,
      "loss": 0.8749,
      "step": 3128
    },
    {
      "epoch": 0.28868642601776034,
      "grad_norm": 1.8383744387669663,
      "learning_rate": 1.0509675756335404e-06,
      "loss": 0.9021,
      "step": 3129
    },
    {
      "epoch": 0.28877868757928726,
      "grad_norm": 1.8068564555818132,
      "learning_rate": 1.0508298758527617e-06,
      "loss": 0.7847,
      "step": 3130
    },
    {
      "epoch": 0.28877868757928726,
      "eval_GEN Loss": 0.4589958190917969,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.369075745344162,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.892848551273346,
      "eval_runtime": 55.9093,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3130
    },
    {
      "epoch": 0.2888709491408142,
      "grad_norm": 1.7482976395088272,
      "learning_rate": 1.0506921374826094e-06,
      "loss": 0.7627,
      "step": 3131
    },
    {
      "epoch": 0.28896321070234116,
      "grad_norm": 1.8152660651430073,
      "learning_rate": 1.0505543605355656e-06,
      "loss": 0.6659,
      "step": 3132
    },
    {
      "epoch": 0.2890554722638681,
      "grad_norm": 3.3737528732931783,
      "learning_rate": 1.0504165450241158e-06,
      "loss": 0.833,
      "step": 3133
    },
    {
      "epoch": 0.289147733825395,
      "grad_norm": 1.942320949150712,
      "learning_rate": 1.050278690960749e-06,
      "loss": 0.6873,
      "step": 3134
    },
    {
      "epoch": 0.2892399953869219,
      "grad_norm": 2.2186790493947837,
      "learning_rate": 1.050140798357958e-06,
      "loss": 0.7874,
      "step": 3135
    },
    {
      "epoch": 0.2892399953869219,
      "eval_GEN Loss": 0.45963695645332336,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.36822742223739624,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8883413672447205,
      "eval_runtime": 55.8123,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3135
    },
    {
      "epoch": 0.28933225694844883,
      "grad_norm": 1.8656051278043004,
      "learning_rate": 1.0500028672282384e-06,
      "loss": 0.623,
      "step": 3136
    },
    {
      "epoch": 0.2894245185099758,
      "grad_norm": 1.8888042933587164,
      "learning_rate": 1.0498648975840901e-06,
      "loss": 0.6907,
      "step": 3137
    },
    {
      "epoch": 0.2895167800715027,
      "grad_norm": 1.2742909023044926,
      "learning_rate": 1.049726889438016e-06,
      "loss": 0.5956,
      "step": 3138
    },
    {
      "epoch": 0.28960904163302964,
      "grad_norm": 1.7209415378099318,
      "learning_rate": 1.049588842802523e-06,
      "loss": 0.6488,
      "step": 3139
    },
    {
      "epoch": 0.28970130319455656,
      "grad_norm": 1.5859824419279278,
      "learning_rate": 1.0494507576901206e-06,
      "loss": 0.5687,
      "step": 3140
    },
    {
      "epoch": 0.28970130319455656,
      "eval_GEN Loss": 0.46031397581100464,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3708396553993225,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8823317289352417,
      "eval_runtime": 55.804,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3140
    },
    {
      "epoch": 0.2897935647560835,
      "grad_norm": 1.4086078227801055,
      "learning_rate": 1.0493126341133221e-06,
      "loss": 0.6124,
      "step": 3141
    },
    {
      "epoch": 0.28988582631761045,
      "grad_norm": 1.3737896057678165,
      "learning_rate": 1.049174472084645e-06,
      "loss": 0.756,
      "step": 3142
    },
    {
      "epoch": 0.28997808787913737,
      "grad_norm": 1.4575605104274383,
      "learning_rate": 1.0490362716166097e-06,
      "loss": 0.6639,
      "step": 3143
    },
    {
      "epoch": 0.2900703494406643,
      "grad_norm": 1.7526819456268004,
      "learning_rate": 1.04889803272174e-06,
      "loss": 0.7761,
      "step": 3144
    },
    {
      "epoch": 0.2901626110021912,
      "grad_norm": 2.067018200441921,
      "learning_rate": 1.0487597554125633e-06,
      "loss": 0.8508,
      "step": 3145
    },
    {
      "epoch": 0.2901626110021912,
      "eval_GEN Loss": 0.46307626366615295,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3669787645339966,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8828125,
      "eval_runtime": 55.9242,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3145
    },
    {
      "epoch": 0.2902548725637181,
      "grad_norm": 1.904528284618996,
      "learning_rate": 1.0486214397016107e-06,
      "loss": 0.5534,
      "step": 3146
    },
    {
      "epoch": 0.29034713412524504,
      "grad_norm": 1.9020526562960165,
      "learning_rate": 1.0484830856014162e-06,
      "loss": 0.848,
      "step": 3147
    },
    {
      "epoch": 0.290439395686772,
      "grad_norm": 2.155351680336442,
      "learning_rate": 1.048344693124518e-06,
      "loss": 0.7462,
      "step": 3148
    },
    {
      "epoch": 0.29053165724829894,
      "grad_norm": 2.550682080895001,
      "learning_rate": 1.0482062622834575e-06,
      "loss": 0.771,
      "step": 3149
    },
    {
      "epoch": 0.29062391880982585,
      "grad_norm": 1.8111298565882687,
      "learning_rate": 1.0480677930907792e-06,
      "loss": 0.7119,
      "step": 3150
    },
    {
      "epoch": 0.29062391880982585,
      "eval_GEN Loss": 0.4651535451412201,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.35730528831481934,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8854567408561707,
      "eval_runtime": 55.7264,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 3150
    },
    {
      "epoch": 0.2907161803713528,
      "grad_norm": 2.3965045367411784,
      "learning_rate": 1.0479292855590314e-06,
      "loss": 0.6272,
      "step": 3151
    },
    {
      "epoch": 0.2908084419328797,
      "grad_norm": 1.2022092351837308,
      "learning_rate": 1.0477907397007664e-06,
      "loss": 0.6686,
      "step": 3152
    },
    {
      "epoch": 0.29090070349440666,
      "grad_norm": 1.9397325181218765,
      "learning_rate": 1.047652155528539e-06,
      "loss": 0.7637,
      "step": 3153
    },
    {
      "epoch": 0.2909929650559336,
      "grad_norm": 1.5083879696158757,
      "learning_rate": 1.047513533054908e-06,
      "loss": 0.7001,
      "step": 3154
    },
    {
      "epoch": 0.2910852266174605,
      "grad_norm": 1.5808087887051612,
      "learning_rate": 1.0473748722924354e-06,
      "loss": 0.5169,
      "step": 3155
    },
    {
      "epoch": 0.2910852266174605,
      "eval_GEN Loss": 0.46677032113075256,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3591814935207367,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8782451748847961,
      "eval_runtime": 55.9761,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3155
    },
    {
      "epoch": 0.2911774881789874,
      "grad_norm": 1.6359547268654968,
      "learning_rate": 1.0472361732536874e-06,
      "loss": 0.7349,
      "step": 3156
    },
    {
      "epoch": 0.29126974974051434,
      "grad_norm": 1.8549432198455764,
      "learning_rate": 1.0470974359512324e-06,
      "loss": 0.6993,
      "step": 3157
    },
    {
      "epoch": 0.2913620113020413,
      "grad_norm": 1.653857682227101,
      "learning_rate": 1.0469586603976433e-06,
      "loss": 0.6384,
      "step": 3158
    },
    {
      "epoch": 0.29145427286356823,
      "grad_norm": 1.548072397896846,
      "learning_rate": 1.0468198466054965e-06,
      "loss": 0.5188,
      "step": 3159
    },
    {
      "epoch": 0.29154653442509515,
      "grad_norm": 5.077027236249122,
      "learning_rate": 1.0466809945873713e-06,
      "loss": 1.0489,
      "step": 3160
    },
    {
      "epoch": 0.29154653442509515,
      "eval_GEN Loss": 0.46819600462913513,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3675070106983185,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8615384697914124,
      "eval_runtime": 55.9124,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3160
    },
    {
      "epoch": 0.29163879598662207,
      "grad_norm": 1.4133852381883283,
      "learning_rate": 1.0465421043558505e-06,
      "loss": 0.7251,
      "step": 3161
    },
    {
      "epoch": 0.291731057548149,
      "grad_norm": 1.919934809858561,
      "learning_rate": 1.0464031759235207e-06,
      "loss": 0.7187,
      "step": 3162
    },
    {
      "epoch": 0.2918233191096759,
      "grad_norm": 1.986475543429513,
      "learning_rate": 1.046264209302972e-06,
      "loss": 0.6051,
      "step": 3163
    },
    {
      "epoch": 0.2919155806712029,
      "grad_norm": 1.4775431135751924,
      "learning_rate": 1.0461252045067972e-06,
      "loss": 0.6428,
      "step": 3164
    },
    {
      "epoch": 0.2920078422327298,
      "grad_norm": 2.4297431446091817,
      "learning_rate": 1.0459861615475936e-06,
      "loss": 0.8346,
      "step": 3165
    },
    {
      "epoch": 0.2920078422327298,
      "eval_GEN Loss": 0.4699571132659912,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.40065595507621765,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8694711327552795,
      "eval_runtime": 55.948,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3165
    },
    {
      "epoch": 0.2921001037942567,
      "grad_norm": 1.7331737289138733,
      "learning_rate": 1.0458470804379617e-06,
      "loss": 0.7742,
      "step": 3166
    },
    {
      "epoch": 0.29219236535578363,
      "grad_norm": 1.867585373874945,
      "learning_rate": 1.0457079611905047e-06,
      "loss": 0.6015,
      "step": 3167
    },
    {
      "epoch": 0.29228462691731055,
      "grad_norm": 2.1300608669779955,
      "learning_rate": 1.04556880381783e-06,
      "loss": 0.8298,
      "step": 3168
    },
    {
      "epoch": 0.2923768884788375,
      "grad_norm": 1.5229898910347492,
      "learning_rate": 1.0454296083325483e-06,
      "loss": 0.7539,
      "step": 3169
    },
    {
      "epoch": 0.29246915004036445,
      "grad_norm": 3.596303188964942,
      "learning_rate": 1.045290374747274e-06,
      "loss": 0.8435,
      "step": 3170
    },
    {
      "epoch": 0.29246915004036445,
      "eval_GEN Loss": 0.46960684657096863,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.39612385630607605,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8713341355323792,
      "eval_runtime": 55.9532,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3170
    },
    {
      "epoch": 0.29256141160189136,
      "grad_norm": 1.964153866959998,
      "learning_rate": 1.0451511030746242e-06,
      "loss": 0.7423,
      "step": 3171
    },
    {
      "epoch": 0.2926536731634183,
      "grad_norm": 2.411175488046264,
      "learning_rate": 1.0450117933272201e-06,
      "loss": 0.7087,
      "step": 3172
    },
    {
      "epoch": 0.2927459347249452,
      "grad_norm": 2.634525391910467,
      "learning_rate": 1.0448724455176862e-06,
      "loss": 0.8788,
      "step": 3173
    },
    {
      "epoch": 0.2928381962864722,
      "grad_norm": 1.7843384106790685,
      "learning_rate": 1.0447330596586504e-06,
      "loss": 0.7808,
      "step": 3174
    },
    {
      "epoch": 0.2929304578479991,
      "grad_norm": 1.966600069213244,
      "learning_rate": 1.044593635762744e-06,
      "loss": 0.6539,
      "step": 3175
    },
    {
      "epoch": 0.2929304578479991,
      "eval_GEN Loss": 0.46833521127700806,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.39416956901550293,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8832932710647583,
      "eval_runtime": 55.9285,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3175
    },
    {
      "epoch": 0.293022719409526,
      "grad_norm": 1.764432799519812,
      "learning_rate": 1.044454173842602e-06,
      "loss": 0.7885,
      "step": 3176
    },
    {
      "epoch": 0.29311498097105293,
      "grad_norm": 2.04802878259487,
      "learning_rate": 1.0443146739108625e-06,
      "loss": 0.6674,
      "step": 3177
    },
    {
      "epoch": 0.29320724253257985,
      "grad_norm": 1.4919837660601845,
      "learning_rate": 1.0441751359801673e-06,
      "loss": 0.5971,
      "step": 3178
    },
    {
      "epoch": 0.29329950409410677,
      "grad_norm": 1.957245487506922,
      "learning_rate": 1.0440355600631612e-06,
      "loss": 0.863,
      "step": 3179
    },
    {
      "epoch": 0.29339176565563374,
      "grad_norm": 2.3287094530422547,
      "learning_rate": 1.0438959461724934e-06,
      "loss": 0.7705,
      "step": 3180
    },
    {
      "epoch": 0.29339176565563374,
      "eval_GEN Loss": 0.4665025472640991,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3920274078845978,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8968750238418579,
      "eval_runtime": 55.9439,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3180
    },
    {
      "epoch": 0.29348402721716066,
      "grad_norm": 1.9833444517471517,
      "learning_rate": 1.0437562943208153e-06,
      "loss": 0.8876,
      "step": 3181
    },
    {
      "epoch": 0.2935762887786876,
      "grad_norm": 1.8946601399084904,
      "learning_rate": 1.0436166045207831e-06,
      "loss": 0.6933,
      "step": 3182
    },
    {
      "epoch": 0.2936685503402145,
      "grad_norm": 2.8117872189601822,
      "learning_rate": 1.0434768767850551e-06,
      "loss": 0.9753,
      "step": 3183
    },
    {
      "epoch": 0.2937608119017414,
      "grad_norm": 3.414644451572609,
      "learning_rate": 1.0433371111262938e-06,
      "loss": 1.0414,
      "step": 3184
    },
    {
      "epoch": 0.2938530734632684,
      "grad_norm": 1.705567029138378,
      "learning_rate": 1.043197307557165e-06,
      "loss": 0.6504,
      "step": 3185
    },
    {
      "epoch": 0.2938530734632684,
      "eval_GEN Loss": 0.46626728773117065,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.39432328939437866,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8966947197914124,
      "eval_runtime": 55.9209,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3185
    },
    {
      "epoch": 0.2939453350247953,
      "grad_norm": 2.3853810340231756,
      "learning_rate": 1.0430574660903379e-06,
      "loss": 0.826,
      "step": 3186
    },
    {
      "epoch": 0.2940375965863222,
      "grad_norm": 1.4601027958840533,
      "learning_rate": 1.0429175867384853e-06,
      "loss": 0.7381,
      "step": 3187
    },
    {
      "epoch": 0.29412985814784914,
      "grad_norm": 3.0209128877522353,
      "learning_rate": 1.042777669514283e-06,
      "loss": 0.8718,
      "step": 3188
    },
    {
      "epoch": 0.29422211970937606,
      "grad_norm": 2.85446344423048,
      "learning_rate": 1.0426377144304107e-06,
      "loss": 0.8125,
      "step": 3189
    },
    {
      "epoch": 0.29431438127090304,
      "grad_norm": 1.8311409864746164,
      "learning_rate": 1.0424977214995515e-06,
      "loss": 0.7096,
      "step": 3190
    },
    {
      "epoch": 0.29431438127090304,
      "eval_GEN Loss": 0.46524283289909363,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.39024388790130615,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8878605961799622,
      "eval_runtime": 55.9145,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3190
    },
    {
      "epoch": 0.29440664283242995,
      "grad_norm": 1.4732126770264173,
      "learning_rate": 1.0423576907343916e-06,
      "loss": 0.7715,
      "step": 3191
    },
    {
      "epoch": 0.2944989043939569,
      "grad_norm": 1.4891954160985086,
      "learning_rate": 1.0422176221476208e-06,
      "loss": 0.6787,
      "step": 3192
    },
    {
      "epoch": 0.2945911659554838,
      "grad_norm": 2.4310716826356407,
      "learning_rate": 1.0420775157519325e-06,
      "loss": 0.7358,
      "step": 3193
    },
    {
      "epoch": 0.2946834275170107,
      "grad_norm": 1.847219074555379,
      "learning_rate": 1.041937371560023e-06,
      "loss": 0.7859,
      "step": 3194
    },
    {
      "epoch": 0.29477568907853763,
      "grad_norm": 1.6234422563789452,
      "learning_rate": 1.0417971895845927e-06,
      "loss": 0.7301,
      "step": 3195
    },
    {
      "epoch": 0.29477568907853763,
      "eval_GEN Loss": 0.4664727747440338,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3986499607563019,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8802283406257629,
      "eval_runtime": 56.0516,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 3195
    },
    {
      "epoch": 0.2948679506400646,
      "grad_norm": 2.2291279703305173,
      "learning_rate": 1.0416569698383451e-06,
      "loss": 0.8245,
      "step": 3196
    },
    {
      "epoch": 0.2949602122015915,
      "grad_norm": 2.3734572538934553,
      "learning_rate": 1.041516712333987e-06,
      "loss": 0.8111,
      "step": 3197
    },
    {
      "epoch": 0.29505247376311844,
      "grad_norm": 1.7671502648734991,
      "learning_rate": 1.041376417084229e-06,
      "loss": 0.796,
      "step": 3198
    },
    {
      "epoch": 0.29514473532464536,
      "grad_norm": 1.883649292075704,
      "learning_rate": 1.0412360841017845e-06,
      "loss": 0.7656,
      "step": 3199
    },
    {
      "epoch": 0.2952369968861723,
      "grad_norm": 2.680206645958656,
      "learning_rate": 1.041095713399371e-06,
      "loss": 0.7452,
      "step": 3200
    },
    {
      "epoch": 0.2952369968861723,
      "eval_GEN Loss": 0.4674176573753357,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4074541926383972,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8787259459495544,
      "eval_runtime": 55.9267,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3200
    },
    {
      "epoch": 0.29532925844769925,
      "grad_norm": 1.9446061260422907,
      "learning_rate": 1.040955304989709e-06,
      "loss": 0.7373,
      "step": 3201
    },
    {
      "epoch": 0.29542152000922617,
      "grad_norm": 2.3552602313247095,
      "learning_rate": 1.0408148588855225e-06,
      "loss": 0.655,
      "step": 3202
    },
    {
      "epoch": 0.2955137815707531,
      "grad_norm": 2.6776142622557515,
      "learning_rate": 1.0406743750995388e-06,
      "loss": 0.6728,
      "step": 3203
    },
    {
      "epoch": 0.29560604313228,
      "grad_norm": 2.496971892181065,
      "learning_rate": 1.0405338536444893e-06,
      "loss": 0.8394,
      "step": 3204
    },
    {
      "epoch": 0.2956983046938069,
      "grad_norm": 2.0945994420792324,
      "learning_rate": 1.0403932945331078e-06,
      "loss": 0.6992,
      "step": 3205
    },
    {
      "epoch": 0.2956983046938069,
      "eval_GEN Loss": 0.46679702401161194,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.4008890986442566,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8759615421295166,
      "eval_runtime": 55.741,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 3205
    },
    {
      "epoch": 0.2957905662553339,
      "grad_norm": 1.4268433586044347,
      "learning_rate": 1.040252697778132e-06,
      "loss": 0.637,
      "step": 3206
    },
    {
      "epoch": 0.2958828278168608,
      "grad_norm": 1.9604369515361877,
      "learning_rate": 1.0401120633923031e-06,
      "loss": 0.7567,
      "step": 3207
    },
    {
      "epoch": 0.29597508937838773,
      "grad_norm": 2.050256242253183,
      "learning_rate": 1.0399713913883655e-06,
      "loss": 0.8661,
      "step": 3208
    },
    {
      "epoch": 0.29606735093991465,
      "grad_norm": 1.6955652138476502,
      "learning_rate": 1.0398306817790677e-06,
      "loss": 0.8202,
      "step": 3209
    },
    {
      "epoch": 0.29615961250144157,
      "grad_norm": 1.6589527435088736,
      "learning_rate": 1.0396899345771603e-06,
      "loss": 0.8077,
      "step": 3210
    },
    {
      "epoch": 0.29615961250144157,
      "eval_GEN Loss": 0.46665677428245544,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.4116832911968231,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8700721263885498,
      "eval_runtime": 56.2772,
      "eval_samples_per_second": 1.155,
      "eval_steps_per_second": 0.089,
      "step": 3210
    },
    {
      "epoch": 0.2962518740629685,
      "grad_norm": 2.179169601372341,
      "learning_rate": 1.039549149795398e-06,
      "loss": 0.7368,
      "step": 3211
    },
    {
      "epoch": 0.29634413562449546,
      "grad_norm": 2.490690970558911,
      "learning_rate": 1.0394083274465397e-06,
      "loss": 0.6379,
      "step": 3212
    },
    {
      "epoch": 0.2964363971860224,
      "grad_norm": 2.5464766250650746,
      "learning_rate": 1.0392674675433462e-06,
      "loss": 0.813,
      "step": 3213
    },
    {
      "epoch": 0.2965286587475493,
      "grad_norm": 1.8716356283686209,
      "learning_rate": 1.0391265700985827e-06,
      "loss": 0.7216,
      "step": 3214
    },
    {
      "epoch": 0.2966209203090762,
      "grad_norm": 1.7010034716123574,
      "learning_rate": 1.0389856351250178e-06,
      "loss": 0.647,
      "step": 3215
    },
    {
      "epoch": 0.2966209203090762,
      "eval_GEN Loss": 0.4681289494037628,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4168323874473572,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8778846263885498,
      "eval_runtime": 56.2273,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 3215
    },
    {
      "epoch": 0.29671318187060314,
      "grad_norm": 1.3811160230917578,
      "learning_rate": 1.038844662635423e-06,
      "loss": 0.5191,
      "step": 3216
    },
    {
      "epoch": 0.2968054434321301,
      "grad_norm": 1.5464390579174991,
      "learning_rate": 1.0387036526425734e-06,
      "loss": 0.7829,
      "step": 3217
    },
    {
      "epoch": 0.29689770499365703,
      "grad_norm": 2.517594632348134,
      "learning_rate": 1.0385626051592476e-06,
      "loss": 0.9745,
      "step": 3218
    },
    {
      "epoch": 0.29698996655518395,
      "grad_norm": 2.413440548805422,
      "learning_rate": 1.0384215201982275e-06,
      "loss": 0.8237,
      "step": 3219
    },
    {
      "epoch": 0.29708222811671087,
      "grad_norm": 1.6389357500069879,
      "learning_rate": 1.0382803977722986e-06,
      "loss": 0.5575,
      "step": 3220
    },
    {
      "epoch": 0.29708222811671087,
      "eval_GEN Loss": 0.4676324129104614,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.40068355202674866,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8912860751152039,
      "eval_runtime": 55.9035,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3220
    },
    {
      "epoch": 0.2971744896782378,
      "grad_norm": 2.1137986501791888,
      "learning_rate": 1.0381392378942495e-06,
      "loss": 0.735,
      "step": 3221
    },
    {
      "epoch": 0.29726675123976476,
      "grad_norm": 2.168557091076219,
      "learning_rate": 1.0379980405768726e-06,
      "loss": 0.7403,
      "step": 3222
    },
    {
      "epoch": 0.2973590128012917,
      "grad_norm": 3.216875359654587,
      "learning_rate": 1.0378568058329628e-06,
      "loss": 0.8687,
      "step": 3223
    },
    {
      "epoch": 0.2974512743628186,
      "grad_norm": 2.0052480433898565,
      "learning_rate": 1.0377155336753196e-06,
      "loss": 0.8577,
      "step": 3224
    },
    {
      "epoch": 0.2975435359243455,
      "grad_norm": 1.4313688871535097,
      "learning_rate": 1.0375742241167452e-06,
      "loss": 0.6885,
      "step": 3225
    },
    {
      "epoch": 0.2975435359243455,
      "eval_GEN Loss": 0.46804577112197876,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.3963159918785095,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9114783406257629,
      "eval_runtime": 56.0151,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 3225
    },
    {
      "epoch": 0.29763579748587243,
      "grad_norm": 2.187857154681799,
      "learning_rate": 1.0374328771700452e-06,
      "loss": 0.865,
      "step": 3226
    },
    {
      "epoch": 0.29772805904739935,
      "grad_norm": 1.6542876255566346,
      "learning_rate": 1.0372914928480285e-06,
      "loss": 0.7568,
      "step": 3227
    },
    {
      "epoch": 0.2978203206089263,
      "grad_norm": 1.2735996968453465,
      "learning_rate": 1.0371500711635079e-06,
      "loss": 0.5604,
      "step": 3228
    },
    {
      "epoch": 0.29791258217045324,
      "grad_norm": 2.874541431609378,
      "learning_rate": 1.037008612129299e-06,
      "loss": 0.8126,
      "step": 3229
    },
    {
      "epoch": 0.29800484373198016,
      "grad_norm": 1.3027060538651356,
      "learning_rate": 1.0368671157582211e-06,
      "loss": 0.6454,
      "step": 3230
    },
    {
      "epoch": 0.29800484373198016,
      "eval_GEN Loss": 0.46702277660369873,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.38811028003692627,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9039062261581421,
      "eval_runtime": 56.1379,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 3230
    },
    {
      "epoch": 0.2980971052935071,
      "grad_norm": 1.3655562260481116,
      "learning_rate": 1.0367255820630972e-06,
      "loss": 0.74,
      "step": 3231
    },
    {
      "epoch": 0.298189366855034,
      "grad_norm": 1.7104187736498002,
      "learning_rate": 1.0365840110567528e-06,
      "loss": 0.6899,
      "step": 3232
    },
    {
      "epoch": 0.298281628416561,
      "grad_norm": 2.070311453084554,
      "learning_rate": 1.0364424027520175e-06,
      "loss": 0.6575,
      "step": 3233
    },
    {
      "epoch": 0.2983738899780879,
      "grad_norm": 1.3532262791037883,
      "learning_rate": 1.036300757161724e-06,
      "loss": 0.7087,
      "step": 3234
    },
    {
      "epoch": 0.2984661515396148,
      "grad_norm": 2.9280333355882613,
      "learning_rate": 1.0361590742987084e-06,
      "loss": 0.9321,
      "step": 3235
    },
    {
      "epoch": 0.2984661515396148,
      "eval_GEN Loss": 0.46798866987228394,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.38410934805870056,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8912259340286255,
      "eval_runtime": 55.8978,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3235
    },
    {
      "epoch": 0.29855841310114173,
      "grad_norm": 1.7575149604597236,
      "learning_rate": 1.0360173541758105e-06,
      "loss": 0.6251,
      "step": 3236
    },
    {
      "epoch": 0.29865067466266865,
      "grad_norm": 1.7941693097692035,
      "learning_rate": 1.035875596805873e-06,
      "loss": 0.7648,
      "step": 3237
    },
    {
      "epoch": 0.2987429362241956,
      "grad_norm": 1.422349157133686,
      "learning_rate": 1.0357338022017422e-06,
      "loss": 0.7285,
      "step": 3238
    },
    {
      "epoch": 0.29883519778572254,
      "grad_norm": 2.204233576169532,
      "learning_rate": 1.0355919703762677e-06,
      "loss": 0.6954,
      "step": 3239
    },
    {
      "epoch": 0.29892745934724946,
      "grad_norm": 2.9978934632184164,
      "learning_rate": 1.0354501013423023e-06,
      "loss": 0.776,
      "step": 3240
    },
    {
      "epoch": 0.29892745934724946,
      "eval_GEN Loss": 0.46850425004959106,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.3801555633544922,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8795071840286255,
      "eval_runtime": 57.0482,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 3240
    },
    {
      "epoch": 0.2990197209087764,
      "grad_norm": 2.0307286634079444,
      "learning_rate": 1.035308195112703e-06,
      "loss": 0.7,
      "step": 3241
    },
    {
      "epoch": 0.2991119824703033,
      "grad_norm": 3.218966889947449,
      "learning_rate": 1.035166251700329e-06,
      "loss": 0.9292,
      "step": 3242
    },
    {
      "epoch": 0.2992042440318302,
      "grad_norm": 2.209032735963684,
      "learning_rate": 1.0350242711180439e-06,
      "loss": 0.8464,
      "step": 3243
    },
    {
      "epoch": 0.2992965055933572,
      "grad_norm": 2.379495123563452,
      "learning_rate": 1.0348822533787139e-06,
      "loss": 0.9501,
      "step": 3244
    },
    {
      "epoch": 0.2993887671548841,
      "grad_norm": 1.6734295398994734,
      "learning_rate": 1.0347401984952089e-06,
      "loss": 0.8336,
      "step": 3245
    },
    {
      "epoch": 0.2993887671548841,
      "eval_GEN Loss": 0.46801918745040894,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.40139010548591614,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.865504801273346,
      "eval_runtime": 56.979,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3245
    },
    {
      "epoch": 0.299481028716411,
      "grad_norm": 2.06947352453802,
      "learning_rate": 1.034598106480402e-06,
      "loss": 0.7311,
      "step": 3246
    },
    {
      "epoch": 0.29957329027793794,
      "grad_norm": 1.9702157908877436,
      "learning_rate": 1.03445597734717e-06,
      "loss": 0.8078,
      "step": 3247
    },
    {
      "epoch": 0.29966555183946486,
      "grad_norm": 1.6355725513816324,
      "learning_rate": 1.0343138111083931e-06,
      "loss": 0.7881,
      "step": 3248
    },
    {
      "epoch": 0.29975781340099183,
      "grad_norm": 2.2578121541255713,
      "learning_rate": 1.0341716077769543e-06,
      "loss": 0.8894,
      "step": 3249
    },
    {
      "epoch": 0.29985007496251875,
      "grad_norm": 1.9801506616358524,
      "learning_rate": 1.0340293673657402e-06,
      "loss": 0.7919,
      "step": 3250
    },
    {
      "epoch": 0.29985007496251875,
      "eval_GEN Loss": 0.46785077452659607,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.41465696692466736,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8686298131942749,
      "eval_runtime": 56.6402,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 3250
    },
    {
      "epoch": 0.29994233652404567,
      "grad_norm": 3.444329472431624,
      "learning_rate": 1.0338870898876412e-06,
      "loss": 1.1695,
      "step": 3251
    },
    {
      "epoch": 0.3000345980855726,
      "grad_norm": 2.937429164958739,
      "learning_rate": 1.0337447753555504e-06,
      "loss": 0.7983,
      "step": 3252
    },
    {
      "epoch": 0.3001268596470995,
      "grad_norm": 2.2419866762563423,
      "learning_rate": 1.0336024237823647e-06,
      "loss": 0.9784,
      "step": 3253
    },
    {
      "epoch": 0.3002191212086265,
      "grad_norm": 1.8267692641547941,
      "learning_rate": 1.0334600351809842e-06,
      "loss": 0.695,
      "step": 3254
    },
    {
      "epoch": 0.3003113827701534,
      "grad_norm": 1.7169402593841039,
      "learning_rate": 1.0333176095643124e-06,
      "loss": 0.7023,
      "step": 3255
    },
    {
      "epoch": 0.3003113827701534,
      "eval_GEN Loss": 0.46702641248703003,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.38843604922294617,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8678485751152039,
      "eval_runtime": 55.8029,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3255
    },
    {
      "epoch": 0.3004036443316803,
      "grad_norm": 1.9036497474696479,
      "learning_rate": 1.0331751469452561e-06,
      "loss": 0.7492,
      "step": 3256
    },
    {
      "epoch": 0.30049590589320724,
      "grad_norm": 1.5382110250732495,
      "learning_rate": 1.0330326473367255e-06,
      "loss": 0.6062,
      "step": 3257
    },
    {
      "epoch": 0.30058816745473416,
      "grad_norm": 2.05641696513611,
      "learning_rate": 1.0328901107516342e-06,
      "loss": 0.7757,
      "step": 3258
    },
    {
      "epoch": 0.3006804290162611,
      "grad_norm": 2.116337995374133,
      "learning_rate": 1.032747537202899e-06,
      "loss": 1.0547,
      "step": 3259
    },
    {
      "epoch": 0.30077269057778805,
      "grad_norm": 1.9551700046569036,
      "learning_rate": 1.0326049267034403e-06,
      "loss": 0.7711,
      "step": 3260
    },
    {
      "epoch": 0.30077269057778805,
      "eval_GEN Loss": 0.4676561951637268,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3697621822357178,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8673076629638672,
      "eval_runtime": 55.946,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3260
    },
    {
      "epoch": 0.30086495213931497,
      "grad_norm": 1.6620577709301876,
      "learning_rate": 1.0324622792661813e-06,
      "loss": 0.647,
      "step": 3261
    },
    {
      "epoch": 0.3009572137008419,
      "grad_norm": 1.6303362595302524,
      "learning_rate": 1.0323195949040493e-06,
      "loss": 0.6007,
      "step": 3262
    },
    {
      "epoch": 0.3010494752623688,
      "grad_norm": 1.5566984769084449,
      "learning_rate": 1.0321768736299745e-06,
      "loss": 0.752,
      "step": 3263
    },
    {
      "epoch": 0.3011417368238957,
      "grad_norm": 1.900052620726974,
      "learning_rate": 1.0320341154568904e-06,
      "loss": 0.7988,
      "step": 3264
    },
    {
      "epoch": 0.3012339983854227,
      "grad_norm": 3.0201460667688313,
      "learning_rate": 1.0318913203977344e-06,
      "loss": 0.9369,
      "step": 3265
    },
    {
      "epoch": 0.3012339983854227,
      "eval_GEN Loss": 0.46747279167175293,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.3642984926700592,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8667067289352417,
      "eval_runtime": 55.6597,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 3265
    },
    {
      "epoch": 0.3013262599469496,
      "grad_norm": 1.5204419144766583,
      "learning_rate": 1.0317484884654464e-06,
      "loss": 0.7844,
      "step": 3266
    },
    {
      "epoch": 0.30141852150847653,
      "grad_norm": 2.310936576885017,
      "learning_rate": 1.03160561967297e-06,
      "loss": 0.8933,
      "step": 3267
    },
    {
      "epoch": 0.30151078307000345,
      "grad_norm": 1.820555318156068,
      "learning_rate": 1.0314627140332522e-06,
      "loss": 0.7743,
      "step": 3268
    },
    {
      "epoch": 0.30160304463153037,
      "grad_norm": 1.8115708513353699,
      "learning_rate": 1.0313197715592438e-06,
      "loss": 0.8415,
      "step": 3269
    },
    {
      "epoch": 0.30169530619305734,
      "grad_norm": 1.9909249657601442,
      "learning_rate": 1.0311767922638978e-06,
      "loss": 0.863,
      "step": 3270
    },
    {
      "epoch": 0.30169530619305734,
      "eval_GEN Loss": 0.4707719385623932,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.38755959272384644,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8667067289352417,
      "eval_runtime": 56.9362,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3270
    },
    {
      "epoch": 0.30178756775458426,
      "grad_norm": 1.244021862801661,
      "learning_rate": 1.0310337761601719e-06,
      "loss": 0.6429,
      "step": 3271
    },
    {
      "epoch": 0.3018798293161112,
      "grad_norm": 1.6657812673684578,
      "learning_rate": 1.0308907232610258e-06,
      "loss": 0.6845,
      "step": 3272
    },
    {
      "epoch": 0.3019720908776381,
      "grad_norm": 1.9383551294255283,
      "learning_rate": 1.0307476335794237e-06,
      "loss": 0.7033,
      "step": 3273
    },
    {
      "epoch": 0.302064352439165,
      "grad_norm": 1.5038797483073527,
      "learning_rate": 1.0306045071283321e-06,
      "loss": 0.6858,
      "step": 3274
    },
    {
      "epoch": 0.30215661400069194,
      "grad_norm": 1.6482822932377652,
      "learning_rate": 1.030461343920722e-06,
      "loss": 0.7536,
      "step": 3275
    },
    {
      "epoch": 0.30215661400069194,
      "eval_GEN Loss": 0.4728904962539673,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.39366382360458374,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8687499761581421,
      "eval_runtime": 56.7359,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3275
    },
    {
      "epoch": 0.3022488755622189,
      "grad_norm": 1.9541821455417878,
      "learning_rate": 1.0303181439695664e-06,
      "loss": 0.7107,
      "step": 3276
    },
    {
      "epoch": 0.30234113712374583,
      "grad_norm": 1.7631724194084861,
      "learning_rate": 1.0301749072878428e-06,
      "loss": 0.7293,
      "step": 3277
    },
    {
      "epoch": 0.30243339868527275,
      "grad_norm": 1.956013345066894,
      "learning_rate": 1.0300316338885314e-06,
      "loss": 0.7992,
      "step": 3278
    },
    {
      "epoch": 0.30252566024679967,
      "grad_norm": 1.9077567510106757,
      "learning_rate": 1.0298883237846157e-06,
      "loss": 0.8802,
      "step": 3279
    },
    {
      "epoch": 0.3026179218083266,
      "grad_norm": 1.6983129042163072,
      "learning_rate": 1.0297449769890827e-06,
      "loss": 0.7191,
      "step": 3280
    },
    {
      "epoch": 0.3026179218083266,
      "eval_GEN Loss": 0.4703946113586426,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.3676469624042511,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8661658763885498,
      "eval_runtime": 56.9354,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3280
    },
    {
      "epoch": 0.30271018336985356,
      "grad_norm": 2.4167596748010736,
      "learning_rate": 1.029601593514923e-06,
      "loss": 0.8724,
      "step": 3281
    },
    {
      "epoch": 0.3028024449313805,
      "grad_norm": 1.9284718370881728,
      "learning_rate": 1.02945817337513e-06,
      "loss": 0.771,
      "step": 3282
    },
    {
      "epoch": 0.3028947064929074,
      "grad_norm": 2.0896592797305265,
      "learning_rate": 1.0293147165827005e-06,
      "loss": 0.726,
      "step": 3283
    },
    {
      "epoch": 0.3029869680544343,
      "grad_norm": 1.4631848407087988,
      "learning_rate": 1.0291712231506352e-06,
      "loss": 0.7402,
      "step": 3284
    },
    {
      "epoch": 0.30307922961596123,
      "grad_norm": 2.023432144881425,
      "learning_rate": 1.0290276930919374e-06,
      "loss": 0.7985,
      "step": 3285
    },
    {
      "epoch": 0.30307922961596123,
      "eval_GEN Loss": 0.4699096381664276,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3604976236820221,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8707332015037537,
      "eval_runtime": 56.8859,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3285
    },
    {
      "epoch": 0.3031714911774882,
      "grad_norm": 1.623194573926711,
      "learning_rate": 1.0288841264196142e-06,
      "loss": 0.5966,
      "step": 3286
    },
    {
      "epoch": 0.3032637527390151,
      "grad_norm": 2.5588576795049978,
      "learning_rate": 1.0287405231466757e-06,
      "loss": 0.9355,
      "step": 3287
    },
    {
      "epoch": 0.30335601430054204,
      "grad_norm": 2.076081250269867,
      "learning_rate": 1.0285968832861354e-06,
      "loss": 0.8369,
      "step": 3288
    },
    {
      "epoch": 0.30344827586206896,
      "grad_norm": 2.405456077218641,
      "learning_rate": 1.0284532068510103e-06,
      "loss": 0.8733,
      "step": 3289
    },
    {
      "epoch": 0.3035405374235959,
      "grad_norm": 1.2966798455161543,
      "learning_rate": 1.0283094938543206e-06,
      "loss": 0.6689,
      "step": 3290
    },
    {
      "epoch": 0.3035405374235959,
      "eval_GEN Loss": 0.47063034772872925,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3636990785598755,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8725360631942749,
      "eval_runtime": 56.8365,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3290
    },
    {
      "epoch": 0.3036327989851228,
      "grad_norm": 2.0413652043800243,
      "learning_rate": 1.0281657443090897e-06,
      "loss": 0.8375,
      "step": 3291
    },
    {
      "epoch": 0.30372506054664977,
      "grad_norm": 1.6862004076391688,
      "learning_rate": 1.0280219582283446e-06,
      "loss": 0.7106,
      "step": 3292
    },
    {
      "epoch": 0.3038173221081767,
      "grad_norm": 1.572203241693701,
      "learning_rate": 1.0278781356251151e-06,
      "loss": 0.7421,
      "step": 3293
    },
    {
      "epoch": 0.3039095836697036,
      "grad_norm": 1.4160268666443627,
      "learning_rate": 1.0277342765124348e-06,
      "loss": 0.5468,
      "step": 3294
    },
    {
      "epoch": 0.3040018452312305,
      "grad_norm": 1.6878246187805812,
      "learning_rate": 1.0275903809033405e-06,
      "loss": 0.8052,
      "step": 3295
    },
    {
      "epoch": 0.3040018452312305,
      "eval_GEN Loss": 0.4707462191581726,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.36769601702690125,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8847355842590332,
      "eval_runtime": 56.7304,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3295
    },
    {
      "epoch": 0.30409410679275745,
      "grad_norm": 1.614324008379159,
      "learning_rate": 1.0274464488108723e-06,
      "loss": 0.7218,
      "step": 3296
    },
    {
      "epoch": 0.3041863683542844,
      "grad_norm": 2.084914313864402,
      "learning_rate": 1.0273024802480733e-06,
      "loss": 0.6896,
      "step": 3297
    },
    {
      "epoch": 0.30427862991581134,
      "grad_norm": 3.964611332592801,
      "learning_rate": 1.0271584752279905e-06,
      "loss": 0.9625,
      "step": 3298
    },
    {
      "epoch": 0.30437089147733826,
      "grad_norm": 2.240553154010285,
      "learning_rate": 1.0270144337636733e-06,
      "loss": 0.9378,
      "step": 3299
    },
    {
      "epoch": 0.3044631530388652,
      "grad_norm": 1.8621243038663489,
      "learning_rate": 1.026870355868176e-06,
      "loss": 0.7919,
      "step": 3300
    },
    {
      "epoch": 0.3044631530388652,
      "eval_GEN Loss": 0.47090718150138855,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.38071757555007935,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8900841474533081,
      "eval_runtime": 55.7203,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3300
    },
    {
      "epoch": 0.3045554146003921,
      "grad_norm": 1.9486981087268203,
      "learning_rate": 1.0267262415545536e-06,
      "loss": 0.6998,
      "step": 3301
    },
    {
      "epoch": 0.30464767616191907,
      "grad_norm": 2.5574249284977015,
      "learning_rate": 1.0265820908358677e-06,
      "loss": 0.6515,
      "step": 3302
    },
    {
      "epoch": 0.304739937723446,
      "grad_norm": 2.5832467387808142,
      "learning_rate": 1.02643790372518e-06,
      "loss": 0.8226,
      "step": 3303
    },
    {
      "epoch": 0.3048321992849729,
      "grad_norm": 1.937257837319866,
      "learning_rate": 1.026293680235558e-06,
      "loss": 0.7538,
      "step": 3304
    },
    {
      "epoch": 0.3049244608464998,
      "grad_norm": 1.6625397420449046,
      "learning_rate": 1.026149420380071e-06,
      "loss": 0.7756,
      "step": 3305
    },
    {
      "epoch": 0.3049244608464998,
      "eval_GEN Loss": 0.47270333766937256,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.39520299434661865,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8859975934028625,
      "eval_runtime": 55.7971,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3305
    },
    {
      "epoch": 0.30501672240802674,
      "grad_norm": 2.268252691157371,
      "learning_rate": 1.026005124171792e-06,
      "loss": 0.7261,
      "step": 3306
    },
    {
      "epoch": 0.30510898396955366,
      "grad_norm": 1.4806585163162858,
      "learning_rate": 1.0258607916237972e-06,
      "loss": 0.6518,
      "step": 3307
    },
    {
      "epoch": 0.30520124553108063,
      "grad_norm": 1.436677787171373,
      "learning_rate": 1.0257164227491668e-06,
      "loss": 0.6294,
      "step": 3308
    },
    {
      "epoch": 0.30529350709260755,
      "grad_norm": 2.2437822009554726,
      "learning_rate": 1.0255720175609834e-06,
      "loss": 0.6874,
      "step": 3309
    },
    {
      "epoch": 0.30538576865413447,
      "grad_norm": 1.8269910046246456,
      "learning_rate": 1.0254275760723335e-06,
      "loss": 0.5265,
      "step": 3310
    },
    {
      "epoch": 0.30538576865413447,
      "eval_GEN Loss": 0.4729827046394348,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4143664538860321,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8935697078704834,
      "eval_runtime": 55.8389,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3310
    },
    {
      "epoch": 0.3054780302156614,
      "grad_norm": 2.395724923886861,
      "learning_rate": 1.025283098296306e-06,
      "loss": 0.7263,
      "step": 3311
    },
    {
      "epoch": 0.3055702917771883,
      "grad_norm": 2.230094583216984,
      "learning_rate": 1.0251385842459942e-06,
      "loss": 0.7754,
      "step": 3312
    },
    {
      "epoch": 0.3056625533387153,
      "grad_norm": 1.868791460671676,
      "learning_rate": 1.0249940339344942e-06,
      "loss": 0.7799,
      "step": 3313
    },
    {
      "epoch": 0.3057548149002422,
      "grad_norm": 2.468894125372381,
      "learning_rate": 1.024849447374905e-06,
      "loss": 0.8185,
      "step": 3314
    },
    {
      "epoch": 0.3058470764617691,
      "grad_norm": 2.0452050694746498,
      "learning_rate": 1.0247048245803296e-06,
      "loss": 0.7051,
      "step": 3315
    },
    {
      "epoch": 0.3058470764617691,
      "eval_GEN Loss": 0.4728299081325531,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.4197731018066406,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9040865302085876,
      "eval_runtime": 55.8304,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3315
    },
    {
      "epoch": 0.30593933802329604,
      "grad_norm": 1.805030598113026,
      "learning_rate": 1.024560165563874e-06,
      "loss": 0.6042,
      "step": 3316
    },
    {
      "epoch": 0.30603159958482296,
      "grad_norm": 1.829706059984909,
      "learning_rate": 1.024415470338647e-06,
      "loss": 0.6795,
      "step": 3317
    },
    {
      "epoch": 0.30612386114634993,
      "grad_norm": 2.9942106735190213,
      "learning_rate": 1.0242707389177615e-06,
      "loss": 1.025,
      "step": 3318
    },
    {
      "epoch": 0.30621612270787685,
      "grad_norm": 1.8334212605152038,
      "learning_rate": 1.0241259713143334e-06,
      "loss": 0.7262,
      "step": 3319
    },
    {
      "epoch": 0.30630838426940377,
      "grad_norm": 1.6526923793308834,
      "learning_rate": 1.0239811675414813e-06,
      "loss": 0.4432,
      "step": 3320
    },
    {
      "epoch": 0.30630838426940377,
      "eval_GEN Loss": 0.4724779427051544,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.42770978808403015,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9074519276618958,
      "eval_runtime": 56.6413,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 3320
    },
    {
      "epoch": 0.3064006458309307,
      "grad_norm": 2.137500298581433,
      "learning_rate": 1.0238363276123277e-06,
      "loss": 0.7667,
      "step": 3321
    },
    {
      "epoch": 0.3064929073924576,
      "grad_norm": 1.8812189534134063,
      "learning_rate": 1.0236914515399988e-06,
      "loss": 0.6562,
      "step": 3322
    },
    {
      "epoch": 0.3065851689539845,
      "grad_norm": 2.006870280497996,
      "learning_rate": 1.0235465393376226e-06,
      "loss": 0.9133,
      "step": 3323
    },
    {
      "epoch": 0.3066774305155115,
      "grad_norm": 1.8062423441610895,
      "learning_rate": 1.0234015910183317e-06,
      "loss": 0.7146,
      "step": 3324
    },
    {
      "epoch": 0.3067696920770384,
      "grad_norm": 2.2441966271930838,
      "learning_rate": 1.0232566065952619e-06,
      "loss": 0.7222,
      "step": 3325
    },
    {
      "epoch": 0.3067696920770384,
      "eval_GEN Loss": 0.47074562311172485,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.41788098216056824,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9084134697914124,
      "eval_runtime": 55.8037,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3325
    },
    {
      "epoch": 0.30686195363856533,
      "grad_norm": 2.6136074390789745,
      "learning_rate": 1.023111586081551e-06,
      "loss": 0.8835,
      "step": 3326
    },
    {
      "epoch": 0.30695421520009225,
      "grad_norm": 1.8722610066994665,
      "learning_rate": 1.0229665294903419e-06,
      "loss": 0.7502,
      "step": 3327
    },
    {
      "epoch": 0.30704647676161917,
      "grad_norm": 1.6319415044337886,
      "learning_rate": 1.0228214368347794e-06,
      "loss": 0.5874,
      "step": 3328
    },
    {
      "epoch": 0.30713873832314614,
      "grad_norm": 2.8342384197846364,
      "learning_rate": 1.0226763081280119e-06,
      "loss": 1.125,
      "step": 3329
    },
    {
      "epoch": 0.30723099988467306,
      "grad_norm": 1.9241778615562308,
      "learning_rate": 1.0225311433831917e-06,
      "loss": 0.6292,
      "step": 3330
    },
    {
      "epoch": 0.30723099988467306,
      "eval_GEN Loss": 0.47109854221343994,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.4101719856262207,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9084134697914124,
      "eval_runtime": 55.8197,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3330
    },
    {
      "epoch": 0.3073232614462,
      "grad_norm": 2.1434093132793293,
      "learning_rate": 1.0223859426134733e-06,
      "loss": 0.6021,
      "step": 3331
    },
    {
      "epoch": 0.3074155230077269,
      "grad_norm": 1.6934957483109931,
      "learning_rate": 1.0222407058320156e-06,
      "loss": 0.572,
      "step": 3332
    },
    {
      "epoch": 0.3075077845692538,
      "grad_norm": 1.7503604521794125,
      "learning_rate": 1.0220954330519796e-06,
      "loss": 0.8133,
      "step": 3333
    },
    {
      "epoch": 0.3076000461307808,
      "grad_norm": 1.8703807772651815,
      "learning_rate": 1.0219501242865307e-06,
      "loss": 0.7132,
      "step": 3334
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 1.9980591641510483,
      "learning_rate": 1.0218047795488365e-06,
      "loss": 0.8978,
      "step": 3335
    },
    {
      "epoch": 0.3076923076923077,
      "eval_GEN Loss": 0.4707014560699463,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.4088593125343323,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9044471383094788,
      "eval_runtime": 55.6871,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3335
    },
    {
      "epoch": 0.30778456925383463,
      "grad_norm": 2.07492655187952,
      "learning_rate": 1.0216593988520683e-06,
      "loss": 0.7707,
      "step": 3336
    },
    {
      "epoch": 0.30787683081536155,
      "grad_norm": 1.8991136494571181,
      "learning_rate": 1.0215139822094018e-06,
      "loss": 0.7475,
      "step": 3337
    },
    {
      "epoch": 0.30796909237688846,
      "grad_norm": 1.3872993182792597,
      "learning_rate": 1.0213685296340134e-06,
      "loss": 0.5416,
      "step": 3338
    },
    {
      "epoch": 0.3080613539384154,
      "grad_norm": 1.3707842988170422,
      "learning_rate": 1.0212230411390855e-06,
      "loss": 0.552,
      "step": 3339
    },
    {
      "epoch": 0.30815361549994236,
      "grad_norm": 2.2669664923232897,
      "learning_rate": 1.0210775167378015e-06,
      "loss": 0.6874,
      "step": 3340
    },
    {
      "epoch": 0.30815361549994236,
      "eval_GEN Loss": 0.4694972336292267,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.40061259269714355,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8984375,
      "eval_runtime": 56.5324,
      "eval_samples_per_second": 1.15,
      "eval_steps_per_second": 0.088,
      "step": 3340
    },
    {
      "epoch": 0.3082458770614693,
      "grad_norm": 2.0212913292135943,
      "learning_rate": 1.0209319564433499e-06,
      "loss": 0.8697,
      "step": 3341
    },
    {
      "epoch": 0.3083381386229962,
      "grad_norm": 2.1789344103685298,
      "learning_rate": 1.0207863602689208e-06,
      "loss": 0.8567,
      "step": 3342
    },
    {
      "epoch": 0.3084304001845231,
      "grad_norm": 1.98778489061429,
      "learning_rate": 1.0206407282277089e-06,
      "loss": 0.7202,
      "step": 3343
    },
    {
      "epoch": 0.30852266174605003,
      "grad_norm": 2.7202012428272497,
      "learning_rate": 1.0204950603329118e-06,
      "loss": 0.6787,
      "step": 3344
    },
    {
      "epoch": 0.308614923307577,
      "grad_norm": 2.1318058328549205,
      "learning_rate": 1.0203493565977292e-06,
      "loss": 0.7727,
      "step": 3345
    },
    {
      "epoch": 0.308614923307577,
      "eval_GEN Loss": 0.4690205752849579,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.387918084859848,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9064303040504456,
      "eval_runtime": 56.7077,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3345
    },
    {
      "epoch": 0.3087071848691039,
      "grad_norm": 1.890544659623137,
      "learning_rate": 1.020203617035366e-06,
      "loss": 0.7057,
      "step": 3346
    },
    {
      "epoch": 0.30879944643063084,
      "grad_norm": 1.646076173241973,
      "learning_rate": 1.020057841659029e-06,
      "loss": 0.73,
      "step": 3347
    },
    {
      "epoch": 0.30889170799215776,
      "grad_norm": 1.527831774418119,
      "learning_rate": 1.0199120304819285e-06,
      "loss": 0.7791,
      "step": 3348
    },
    {
      "epoch": 0.3089839695536847,
      "grad_norm": 1.3464500653717142,
      "learning_rate": 1.019766183517278e-06,
      "loss": 0.6014,
      "step": 3349
    },
    {
      "epoch": 0.30907623111521165,
      "grad_norm": 3.5756884753754457,
      "learning_rate": 1.0196203007782948e-06,
      "loss": 0.7607,
      "step": 3350
    },
    {
      "epoch": 0.30907623111521165,
      "eval_GEN Loss": 0.46669140458106995,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3780517876148224,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.9027042984962463,
      "eval_runtime": 56.5699,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 3350
    },
    {
      "epoch": 0.30916849267673857,
      "grad_norm": 2.7250270759632174,
      "learning_rate": 1.0194743822781988e-06,
      "loss": 1.1092,
      "step": 3351
    },
    {
      "epoch": 0.3092607542382655,
      "grad_norm": 3.136282312388678,
      "learning_rate": 1.0193284280302132e-06,
      "loss": 1.0014,
      "step": 3352
    },
    {
      "epoch": 0.3093530157997924,
      "grad_norm": 2.1291434451106217,
      "learning_rate": 1.0191824380475649e-06,
      "loss": 0.7561,
      "step": 3353
    },
    {
      "epoch": 0.3094452773613193,
      "grad_norm": 3.082858778453155,
      "learning_rate": 1.0190364123434834e-06,
      "loss": 1.0409,
      "step": 3354
    },
    {
      "epoch": 0.30953753892284624,
      "grad_norm": 2.3365254181050146,
      "learning_rate": 1.0188903509312022e-06,
      "loss": 0.6982,
      "step": 3355
    },
    {
      "epoch": 0.30953753892284624,
      "eval_GEN Loss": 0.46679937839508057,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8711656441717791,
      "eval_PRM F1 AUC": 0.7320586694604505,
      "eval_PRM F1 Neg": 0.5714285714285714,
      "eval_PRM Loss": 0.39128080010414124,
      "eval_PRM NPV": 0.5384615384615384,
      "eval_PRM Precision": 0.8875,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8769230842590332,
      "eval_runtime": 56.8006,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3355
    },
    {
      "epoch": 0.3096298004843732,
      "grad_norm": 1.9106366259885774,
      "learning_rate": 1.0187442538239574e-06,
      "loss": 0.9559,
      "step": 3356
    },
    {
      "epoch": 0.30972206204590014,
      "grad_norm": 2.0395205025758694,
      "learning_rate": 1.0185981210349887e-06,
      "loss": 0.6558,
      "step": 3357
    },
    {
      "epoch": 0.30981432360742706,
      "grad_norm": 1.7719361919041747,
      "learning_rate": 1.0184519525775385e-06,
      "loss": 0.6844,
      "step": 3358
    },
    {
      "epoch": 0.309906585168954,
      "grad_norm": 2.0228887938339666,
      "learning_rate": 1.0183057484648531e-06,
      "loss": 0.6582,
      "step": 3359
    },
    {
      "epoch": 0.3099988467304809,
      "grad_norm": 1.9374935217996054,
      "learning_rate": 1.0181595087101818e-06,
      "loss": 0.7496,
      "step": 3360
    },
    {
      "epoch": 0.3099988467304809,
      "eval_GEN Loss": 0.4677060544490814,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.4080377221107483,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8750600814819336,
      "eval_runtime": 56.7766,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3360
    },
    {
      "epoch": 0.31009110829200787,
      "grad_norm": 1.6071211254353064,
      "learning_rate": 1.018013233326777e-06,
      "loss": 0.7742,
      "step": 3361
    },
    {
      "epoch": 0.3101833698535348,
      "grad_norm": 2.0562964741189953,
      "learning_rate": 1.0178669223278942e-06,
      "loss": 0.7995,
      "step": 3362
    },
    {
      "epoch": 0.3102756314150617,
      "grad_norm": 1.7809982036684668,
      "learning_rate": 1.0177205757267926e-06,
      "loss": 0.7282,
      "step": 3363
    },
    {
      "epoch": 0.3103678929765886,
      "grad_norm": 1.5973330668416594,
      "learning_rate": 1.0175741935367344e-06,
      "loss": 0.7552,
      "step": 3364
    },
    {
      "epoch": 0.31046015453811554,
      "grad_norm": 1.8496922122601893,
      "learning_rate": 1.0174277757709847e-06,
      "loss": 0.7035,
      "step": 3365
    },
    {
      "epoch": 0.31046015453811554,
      "eval_GEN Loss": 0.4686073362827301,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4041784107685089,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.868870198726654,
      "eval_runtime": 56.9633,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3365
    },
    {
      "epoch": 0.31055241609964246,
      "grad_norm": 2.0865713160728694,
      "learning_rate": 1.0172813224428123e-06,
      "loss": 0.9894,
      "step": 3366
    },
    {
      "epoch": 0.31064467766116943,
      "grad_norm": 1.8065729357525429,
      "learning_rate": 1.017134833565489e-06,
      "loss": 0.8664,
      "step": 3367
    },
    {
      "epoch": 0.31073693922269635,
      "grad_norm": 2.4333008525534088,
      "learning_rate": 1.0169883091522898e-06,
      "loss": 0.8904,
      "step": 3368
    },
    {
      "epoch": 0.31082920078422327,
      "grad_norm": 1.8428756429351756,
      "learning_rate": 1.016841749216493e-06,
      "loss": 0.9325,
      "step": 3369
    },
    {
      "epoch": 0.3109214623457502,
      "grad_norm": 1.3092401645417482,
      "learning_rate": 1.0166951537713804e-06,
      "loss": 0.7078,
      "step": 3370
    },
    {
      "epoch": 0.3109214623457502,
      "eval_GEN Loss": 0.4676709473133087,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3755190968513489,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8591946959495544,
      "eval_runtime": 56.7841,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3370
    },
    {
      "epoch": 0.3110137239072771,
      "grad_norm": 1.7672206805137922,
      "learning_rate": 1.0165485228302362e-06,
      "loss": 0.7657,
      "step": 3371
    },
    {
      "epoch": 0.3111059854688041,
      "grad_norm": 1.6200604261966551,
      "learning_rate": 1.0164018564063485e-06,
      "loss": 0.6887,
      "step": 3372
    },
    {
      "epoch": 0.311198247030331,
      "grad_norm": 2.204190312458895,
      "learning_rate": 1.0162551545130088e-06,
      "loss": 0.637,
      "step": 3373
    },
    {
      "epoch": 0.3112905085918579,
      "grad_norm": 2.870443599677136,
      "learning_rate": 1.0161084171635109e-06,
      "loss": 0.836,
      "step": 3374
    },
    {
      "epoch": 0.31138277015338484,
      "grad_norm": 1.7490066154481514,
      "learning_rate": 1.0159616443711526e-06,
      "loss": 0.7755,
      "step": 3375
    },
    {
      "epoch": 0.31138277015338484,
      "eval_GEN Loss": 0.4669856131076813,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3453295826911926,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8633413314819336,
      "eval_runtime": 56.7946,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3375
    },
    {
      "epoch": 0.31147503171491175,
      "grad_norm": 2.0361920885017017,
      "learning_rate": 1.0158148361492348e-06,
      "loss": 0.7277,
      "step": 3376
    },
    {
      "epoch": 0.31156729327643873,
      "grad_norm": 1.4250053539163856,
      "learning_rate": 1.0156679925110614e-06,
      "loss": 0.6848,
      "step": 3377
    },
    {
      "epoch": 0.31165955483796565,
      "grad_norm": 1.5666054968783123,
      "learning_rate": 1.0155211134699395e-06,
      "loss": 0.6273,
      "step": 3378
    },
    {
      "epoch": 0.31175181639949257,
      "grad_norm": 2.3737875587370185,
      "learning_rate": 1.0153741990391797e-06,
      "loss": 0.7456,
      "step": 3379
    },
    {
      "epoch": 0.3118440779610195,
      "grad_norm": 1.853789005686633,
      "learning_rate": 1.0152272492320956e-06,
      "loss": 0.6138,
      "step": 3380
    },
    {
      "epoch": 0.3118440779610195,
      "eval_GEN Loss": 0.46676862239837646,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.33773162961006165,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8675480484962463,
      "eval_runtime": 56.6869,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3380
    },
    {
      "epoch": 0.3119363395225464,
      "grad_norm": 1.7541052082128803,
      "learning_rate": 1.0150802640620037e-06,
      "loss": 0.7671,
      "step": 3381
    },
    {
      "epoch": 0.3120286010840733,
      "grad_norm": 1.8550924751502285,
      "learning_rate": 1.0149332435422245e-06,
      "loss": 0.7803,
      "step": 3382
    },
    {
      "epoch": 0.3121208626456003,
      "grad_norm": 1.5843097412727327,
      "learning_rate": 1.014786187686081e-06,
      "loss": 0.6949,
      "step": 3383
    },
    {
      "epoch": 0.3122131242071272,
      "grad_norm": 1.7730409501494113,
      "learning_rate": 1.0146390965068994e-06,
      "loss": 0.6834,
      "step": 3384
    },
    {
      "epoch": 0.31230538576865413,
      "grad_norm": 2.512328584502214,
      "learning_rate": 1.0144919700180097e-06,
      "loss": 0.7811,
      "step": 3385
    },
    {
      "epoch": 0.31230538576865413,
      "eval_GEN Loss": 0.4673357307910919,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8982035928143712,
      "eval_PRM F1 AUC": 0.7561550550026193,
      "eval_PRM F1 Neg": 0.6222222222222222,
      "eval_PRM Loss": 0.334909588098526,
      "eval_PRM NPV": 0.6363636363636364,
      "eval_PRM Precision": 0.8928571428571429,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8689903616905212,
      "eval_runtime": 56.8249,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3385
    },
    {
      "epoch": 0.31239764733018105,
      "grad_norm": 2.6287097619865736,
      "learning_rate": 1.0143448082327448e-06,
      "loss": 0.7828,
      "step": 3386
    },
    {
      "epoch": 0.31248990889170797,
      "grad_norm": 1.8687824661604415,
      "learning_rate": 1.0141976111644403e-06,
      "loss": 0.5685,
      "step": 3387
    },
    {
      "epoch": 0.31258217045323494,
      "grad_norm": 1.9189808352920166,
      "learning_rate": 1.0140503788264358e-06,
      "loss": 0.6937,
      "step": 3388
    },
    {
      "epoch": 0.31267443201476186,
      "grad_norm": 1.9459741266297461,
      "learning_rate": 1.0139031112320736e-06,
      "loss": 0.8864,
      "step": 3389
    },
    {
      "epoch": 0.3127666935762888,
      "grad_norm": 2.6156046369506694,
      "learning_rate": 1.013755808394699e-06,
      "loss": 0.7837,
      "step": 3390
    },
    {
      "epoch": 0.3127666935762888,
      "eval_GEN Loss": 0.4673900902271271,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.34583035111427307,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8541466593742371,
      "eval_runtime": 56.8622,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3390
    },
    {
      "epoch": 0.3128589551378157,
      "grad_norm": 2.354927366443173,
      "learning_rate": 1.0136084703276617e-06,
      "loss": 0.7306,
      "step": 3391
    },
    {
      "epoch": 0.3129512166993426,
      "grad_norm": 2.255573155195867,
      "learning_rate": 1.0134610970443127e-06,
      "loss": 0.6164,
      "step": 3392
    },
    {
      "epoch": 0.3130434782608696,
      "grad_norm": 2.8555387428391694,
      "learning_rate": 1.0133136885580077e-06,
      "loss": 0.6752,
      "step": 3393
    },
    {
      "epoch": 0.3131357398223965,
      "grad_norm": 2.1770738403624286,
      "learning_rate": 1.0131662448821052e-06,
      "loss": 0.723,
      "step": 3394
    },
    {
      "epoch": 0.3132280013839234,
      "grad_norm": 2.0981190343643608,
      "learning_rate": 1.0130187660299665e-06,
      "loss": 0.7703,
      "step": 3395
    },
    {
      "epoch": 0.3132280013839234,
      "eval_GEN Loss": 0.4663728177547455,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.3642297089099884,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8474158644676208,
      "eval_runtime": 57.0,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 3395
    },
    {
      "epoch": 0.31332026294545035,
      "grad_norm": 3.527970621621397,
      "learning_rate": 1.0128712520149567e-06,
      "loss": 0.8934,
      "step": 3396
    },
    {
      "epoch": 0.31341252450697726,
      "grad_norm": 3.170039347899062,
      "learning_rate": 1.0127237028504432e-06,
      "loss": 0.8189,
      "step": 3397
    },
    {
      "epoch": 0.3135047860685042,
      "grad_norm": 2.0768709791999744,
      "learning_rate": 1.0125761185497977e-06,
      "loss": 0.5339,
      "step": 3398
    },
    {
      "epoch": 0.31359704763003116,
      "grad_norm": 1.8826166264845288,
      "learning_rate": 1.012428499126394e-06,
      "loss": 0.6593,
      "step": 3399
    },
    {
      "epoch": 0.3136893091915581,
      "grad_norm": 1.9129527924954723,
      "learning_rate": 1.01228084459361e-06,
      "loss": 0.7743,
      "step": 3400
    },
    {
      "epoch": 0.3136893091915581,
      "eval_GEN Loss": 0.4669395387172699,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.38741618394851685,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8521634340286255,
      "eval_runtime": 56.8911,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3400
    },
    {
      "epoch": 0.313781570753085,
      "grad_norm": 2.502354480838246,
      "learning_rate": 1.0121331549648264e-06,
      "loss": 0.8719,
      "step": 3401
    },
    {
      "epoch": 0.3138738323146119,
      "grad_norm": 1.4679716181941183,
      "learning_rate": 1.0119854302534268e-06,
      "loss": 0.5931,
      "step": 3402
    },
    {
      "epoch": 0.31396609387613883,
      "grad_norm": 2.656745236494864,
      "learning_rate": 1.0118376704727984e-06,
      "loss": 0.6946,
      "step": 3403
    },
    {
      "epoch": 0.3140583554376658,
      "grad_norm": 1.5085630730545219,
      "learning_rate": 1.0116898756363316e-06,
      "loss": 0.6718,
      "step": 3404
    },
    {
      "epoch": 0.3141506169991927,
      "grad_norm": 2.0771395260898204,
      "learning_rate": 1.0115420457574195e-06,
      "loss": 0.6816,
      "step": 3405
    },
    {
      "epoch": 0.3141506169991927,
      "eval_GEN Loss": 0.4653070867061615,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3639722764492035,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8503605723381042,
      "eval_runtime": 56.9927,
      "eval_samples_per_second": 1.14,
      "eval_steps_per_second": 0.088,
      "step": 3405
    },
    {
      "epoch": 0.31424287856071964,
      "grad_norm": 1.333892759760426,
      "learning_rate": 1.0113941808494587e-06,
      "loss": 0.5261,
      "step": 3406
    },
    {
      "epoch": 0.31433514012224656,
      "grad_norm": 2.4276151214316184,
      "learning_rate": 1.011246280925849e-06,
      "loss": 0.8444,
      "step": 3407
    },
    {
      "epoch": 0.3144274016837735,
      "grad_norm": 2.210272519313882,
      "learning_rate": 1.0110983459999935e-06,
      "loss": 0.7401,
      "step": 3408
    },
    {
      "epoch": 0.31451966324530045,
      "grad_norm": 2.2593652637655706,
      "learning_rate": 1.0109503760852983e-06,
      "loss": 0.8228,
      "step": 3409
    },
    {
      "epoch": 0.31461192480682737,
      "grad_norm": 1.7818353199493056,
      "learning_rate": 1.0108023711951722e-06,
      "loss": 0.8208,
      "step": 3410
    },
    {
      "epoch": 0.31461192480682737,
      "eval_GEN Loss": 0.4647001624107361,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.891566265060241,
      "eval_PRM F1 AUC": 0.7501309586170771,
      "eval_PRM F1 Neg": 0.6086956521739131,
      "eval_PRM Loss": 0.3483923077583313,
      "eval_PRM NPV": 0.6086956521739131,
      "eval_PRM Precision": 0.891566265060241,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8742788434028625,
      "eval_runtime": 55.6935,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3410
    },
    {
      "epoch": 0.3147041863683543,
      "grad_norm": 2.012667795539502,
      "learning_rate": 1.0106543313430282e-06,
      "loss": 0.6787,
      "step": 3411
    },
    {
      "epoch": 0.3147964479298812,
      "grad_norm": 1.3123381621678762,
      "learning_rate": 1.0105062565422814e-06,
      "loss": 0.5917,
      "step": 3412
    },
    {
      "epoch": 0.3148887094914081,
      "grad_norm": 2.2622176665884255,
      "learning_rate": 1.010358146806351e-06,
      "loss": 0.806,
      "step": 3413
    },
    {
      "epoch": 0.31498097105293504,
      "grad_norm": 1.527314938773981,
      "learning_rate": 1.0102100021486589e-06,
      "loss": 0.4999,
      "step": 3414
    },
    {
      "epoch": 0.315073232614462,
      "grad_norm": 1.916729805183037,
      "learning_rate": 1.0100618225826298e-06,
      "loss": 0.5403,
      "step": 3415
    },
    {
      "epoch": 0.315073232614462,
      "eval_GEN Loss": 0.465240478515625,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.35986462235450745,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8895432949066162,
      "eval_runtime": 55.815,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3415
    },
    {
      "epoch": 0.31516549417598894,
      "grad_norm": 1.716122950973823,
      "learning_rate": 1.0099136081216923e-06,
      "loss": 0.6357,
      "step": 3416
    },
    {
      "epoch": 0.31525775573751585,
      "grad_norm": 2.1558103027639692,
      "learning_rate": 1.009765358779278e-06,
      "loss": 0.6351,
      "step": 3417
    },
    {
      "epoch": 0.3153500172990428,
      "grad_norm": 1.8971406082319706,
      "learning_rate": 1.009617074568821e-06,
      "loss": 0.6116,
      "step": 3418
    },
    {
      "epoch": 0.3154422788605697,
      "grad_norm": 2.920437480370756,
      "learning_rate": 1.0094687555037596e-06,
      "loss": 0.7252,
      "step": 3419
    },
    {
      "epoch": 0.31553454042209667,
      "grad_norm": 2.8320797851427493,
      "learning_rate": 1.0093204015975341e-06,
      "loss": 0.9,
      "step": 3420
    },
    {
      "epoch": 0.31553454042209667,
      "eval_GEN Loss": 0.4643435776233673,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37148717045783997,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8838341236114502,
      "eval_runtime": 55.7998,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3420
    },
    {
      "epoch": 0.3156268019836236,
      "grad_norm": 1.51356421307728,
      "learning_rate": 1.009172012863589e-06,
      "loss": 0.4982,
      "step": 3421
    },
    {
      "epoch": 0.3157190635451505,
      "grad_norm": 1.534661922114155,
      "learning_rate": 1.0090235893153717e-06,
      "loss": 0.6358,
      "step": 3422
    },
    {
      "epoch": 0.3158113251066774,
      "grad_norm": 1.8494959772780213,
      "learning_rate": 1.0088751309663316e-06,
      "loss": 0.6841,
      "step": 3423
    },
    {
      "epoch": 0.31590358666820434,
      "grad_norm": 1.6758762109155423,
      "learning_rate": 1.0087266378299234e-06,
      "loss": 0.7718,
      "step": 3424
    },
    {
      "epoch": 0.3159958482297313,
      "grad_norm": 2.1952365542300885,
      "learning_rate": 1.0085781099196032e-06,
      "loss": 0.7814,
      "step": 3425
    },
    {
      "epoch": 0.3159958482297313,
      "eval_GEN Loss": 0.46513885259628296,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.39377743005752563,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8839543461799622,
      "eval_runtime": 56.4961,
      "eval_samples_per_second": 1.151,
      "eval_steps_per_second": 0.089,
      "step": 3425
    },
    {
      "epoch": 0.31608810979125823,
      "grad_norm": 3.3753403499604553,
      "learning_rate": 1.0084295472488308e-06,
      "loss": 1.0128,
      "step": 3426
    },
    {
      "epoch": 0.31618037135278515,
      "grad_norm": 2.0482485230569725,
      "learning_rate": 1.0082809498310695e-06,
      "loss": 0.7715,
      "step": 3427
    },
    {
      "epoch": 0.31627263291431207,
      "grad_norm": 2.360636547456829,
      "learning_rate": 1.0081323176797852e-06,
      "loss": 0.7996,
      "step": 3428
    },
    {
      "epoch": 0.316364894475839,
      "grad_norm": 2.053635802687062,
      "learning_rate": 1.0079836508084471e-06,
      "loss": 0.9527,
      "step": 3429
    },
    {
      "epoch": 0.3164571560373659,
      "grad_norm": 1.9656706975676381,
      "learning_rate": 1.0078349492305278e-06,
      "loss": 0.7118,
      "step": 3430
    },
    {
      "epoch": 0.3164571560373659,
      "eval_GEN Loss": 0.4659498631954193,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8607594936708861,
      "eval_PRM F1 AUC": 0.7574646411733892,
      "eval_PRM F1 Neg": 0.5925925925925926,
      "eval_PRM Loss": 0.43402910232543945,
      "eval_PRM NPV": 0.5161290322580645,
      "eval_PRM Precision": 0.9066666666666666,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8809494972229004,
      "eval_runtime": 56.7131,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3430
    },
    {
      "epoch": 0.3165494175988929,
      "grad_norm": 1.9034363516133948,
      "learning_rate": 1.0076862129595027e-06,
      "loss": 0.6944,
      "step": 3431
    },
    {
      "epoch": 0.3166416791604198,
      "grad_norm": 2.5491627377452257,
      "learning_rate": 1.0075374420088508e-06,
      "loss": 0.6056,
      "step": 3432
    },
    {
      "epoch": 0.3167339407219467,
      "grad_norm": 2.6899026807620006,
      "learning_rate": 1.0073886363920538e-06,
      "loss": 0.8779,
      "step": 3433
    },
    {
      "epoch": 0.31682620228347363,
      "grad_norm": 1.481629692279984,
      "learning_rate": 1.0072397961225964e-06,
      "loss": 0.7051,
      "step": 3434
    },
    {
      "epoch": 0.31691846384500055,
      "grad_norm": 2.977962078464129,
      "learning_rate": 1.0070909212139674e-06,
      "loss": 0.8423,
      "step": 3435
    },
    {
      "epoch": 0.31691846384500055,
      "eval_GEN Loss": 0.46746936440467834,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8496732026143791,
      "eval_PRM F1 AUC": 0.782870612886328,
      "eval_PRM F1 Neg": 0.6101694915254238,
      "eval_PRM Loss": 0.450547456741333,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9285714285714286,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8827524185180664,
      "eval_runtime": 56.8121,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3435
    },
    {
      "epoch": 0.3170107254065275,
      "grad_norm": 2.3996413961854057,
      "learning_rate": 1.0069420116796574e-06,
      "loss": 0.6885,
      "step": 3436
    },
    {
      "epoch": 0.31710298696805445,
      "grad_norm": 2.1142899971653937,
      "learning_rate": 1.0067930675331613e-06,
      "loss": 0.8298,
      "step": 3437
    },
    {
      "epoch": 0.31719524852958136,
      "grad_norm": 1.8630473519628383,
      "learning_rate": 1.0066440887879762e-06,
      "loss": 0.7965,
      "step": 3438
    },
    {
      "epoch": 0.3172875100911083,
      "grad_norm": 2.6474044745535057,
      "learning_rate": 1.0064950754576032e-06,
      "loss": 0.8672,
      "step": 3439
    },
    {
      "epoch": 0.3173797716526352,
      "grad_norm": 2.046198469978302,
      "learning_rate": 1.0063460275555462e-06,
      "loss": 0.6864,
      "step": 3440
    },
    {
      "epoch": 0.3173797716526352,
      "eval_GEN Loss": 0.4668099284172058,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8535031847133758,
      "eval_PRM F1 AUC": 0.751440544787847,
      "eval_PRM F1 Neg": 0.5818181818181818,
      "eval_PRM Loss": 0.4373442232608795,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9054054054054054,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8778245449066162,
      "eval_runtime": 56.7758,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3440
    },
    {
      "epoch": 0.3174720332141622,
      "grad_norm": 2.091925694187833,
      "learning_rate": 1.0061969450953116e-06,
      "loss": 0.8362,
      "step": 3441
    },
    {
      "epoch": 0.3175642947756891,
      "grad_norm": 1.8035522488555062,
      "learning_rate": 1.0060478280904102e-06,
      "loss": 0.7548,
      "step": 3442
    },
    {
      "epoch": 0.317656556337216,
      "grad_norm": 1.8014604359281734,
      "learning_rate": 1.0058986765543543e-06,
      "loss": 0.7336,
      "step": 3443
    },
    {
      "epoch": 0.31774881789874293,
      "grad_norm": 1.6621129855361083,
      "learning_rate": 1.0057494905006612e-06,
      "loss": 0.7264,
      "step": 3444
    },
    {
      "epoch": 0.31784107946026985,
      "grad_norm": 1.9453823959241852,
      "learning_rate": 1.00560026994285e-06,
      "loss": 0.7352,
      "step": 3445
    },
    {
      "epoch": 0.31784107946026985,
      "eval_GEN Loss": 0.46572715044021606,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8589743589743589,
      "eval_PRM F1 AUC": 0.7731796752226295,
      "eval_PRM F1 Neg": 0.6071428571428571,
      "eval_PRM Loss": 0.4288994073867798,
      "eval_PRM NPV": 0.5151515151515151,
      "eval_PRM Precision": 0.9178082191780822,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.872776448726654,
      "eval_runtime": 56.7986,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3445
    },
    {
      "epoch": 0.31793334102179677,
      "grad_norm": 2.276878407395933,
      "learning_rate": 1.0054510148944432e-06,
      "loss": 0.8751,
      "step": 3446
    },
    {
      "epoch": 0.31802560258332374,
      "grad_norm": 2.2816585194345493,
      "learning_rate": 1.0053017253689664e-06,
      "loss": 0.8072,
      "step": 3447
    },
    {
      "epoch": 0.31811786414485066,
      "grad_norm": 2.775210014993187,
      "learning_rate": 1.0051524013799489e-06,
      "loss": 0.7209,
      "step": 3448
    },
    {
      "epoch": 0.3182101257063776,
      "grad_norm": 2.021413380796253,
      "learning_rate": 1.0050030429409222e-06,
      "loss": 0.6319,
      "step": 3449
    },
    {
      "epoch": 0.3183023872679045,
      "grad_norm": 1.798017525701329,
      "learning_rate": 1.0048536500654215e-06,
      "loss": 0.993,
      "step": 3450
    },
    {
      "epoch": 0.3183023872679045,
      "eval_GEN Loss": 0.46543627977371216,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.40401512384414673,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8692307472229004,
      "eval_runtime": 56.8161,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3450
    },
    {
      "epoch": 0.3183946488294314,
      "grad_norm": 1.629701791307313,
      "learning_rate": 1.0047042227669855e-06,
      "loss": 0.6564,
      "step": 3451
    },
    {
      "epoch": 0.3184869103909584,
      "grad_norm": 2.318404719343471,
      "learning_rate": 1.0045547610591549e-06,
      "loss": 0.7442,
      "step": 3452
    },
    {
      "epoch": 0.3185791719524853,
      "grad_norm": 1.7955599986456374,
      "learning_rate": 1.0044052649554747e-06,
      "loss": 0.7654,
      "step": 3453
    },
    {
      "epoch": 0.3186714335140122,
      "grad_norm": 1.972705877313899,
      "learning_rate": 1.004255734469492e-06,
      "loss": 0.7959,
      "step": 3454
    },
    {
      "epoch": 0.31876369507553914,
      "grad_norm": 1.2881381625310782,
      "learning_rate": 1.0041061696147578e-06,
      "loss": 0.5447,
      "step": 3455
    },
    {
      "epoch": 0.31876369507553914,
      "eval_GEN Loss": 0.46501624584198,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8780487804878049,
      "eval_PRM F1 AUC": 0.7380827658459927,
      "eval_PRM F1 Neg": 0.5833333333333334,
      "eval_PRM Loss": 0.37184834480285645,
      "eval_PRM NPV": 0.56,
      "eval_PRM Precision": 0.8888888888888888,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8751201629638672,
      "eval_runtime": 56.8625,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3455
    },
    {
      "epoch": 0.31885595663706606,
      "grad_norm": 2.7123745967657538,
      "learning_rate": 1.0039565704048255e-06,
      "loss": 0.7356,
      "step": 3456
    },
    {
      "epoch": 0.31894821819859304,
      "grad_norm": 1.858272651565408,
      "learning_rate": 1.0038069368532525e-06,
      "loss": 0.6798,
      "step": 3457
    },
    {
      "epoch": 0.31904047976011995,
      "grad_norm": 1.5476194325166135,
      "learning_rate": 1.0036572689735987e-06,
      "loss": 0.7346,
      "step": 3458
    },
    {
      "epoch": 0.3191327413216469,
      "grad_norm": 2.0370774214217233,
      "learning_rate": 1.003507566779427e-06,
      "loss": 0.7689,
      "step": 3459
    },
    {
      "epoch": 0.3192250028831738,
      "grad_norm": 2.663822577447917,
      "learning_rate": 1.003357830284304e-06,
      "loss": 0.9592,
      "step": 3460
    },
    {
      "epoch": 0.3192250028831738,
      "eval_GEN Loss": 0.46541598439216614,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8848484848484849,
      "eval_PRM F1 AUC": 0.7441068622315348,
      "eval_PRM F1 Neg": 0.5957446808510638,
      "eval_PRM Loss": 0.36349013447761536,
      "eval_PRM NPV": 0.5833333333333334,
      "eval_PRM Precision": 0.8902439024390244,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6086956521739131,
      "eval_loss": 0.8764423131942749,
      "eval_runtime": 55.832,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3460
    },
    {
      "epoch": 0.3193172644447007,
      "grad_norm": 1.5500941830841264,
      "learning_rate": 1.0032080595017987e-06,
      "loss": 0.8309,
      "step": 3461
    },
    {
      "epoch": 0.31940952600622763,
      "grad_norm": 2.4884640035515204,
      "learning_rate": 1.003058254445484e-06,
      "loss": 0.8815,
      "step": 3462
    },
    {
      "epoch": 0.3195017875677546,
      "grad_norm": 1.5222473484521577,
      "learning_rate": 1.002908415128935e-06,
      "loss": 0.6373,
      "step": 3463
    },
    {
      "epoch": 0.3195940491292815,
      "grad_norm": 1.7115155752891467,
      "learning_rate": 1.0027585415657306e-06,
      "loss": 0.6498,
      "step": 3464
    },
    {
      "epoch": 0.31968631069080844,
      "grad_norm": 1.7464842802665954,
      "learning_rate": 1.0026086337694526e-06,
      "loss": 0.7887,
      "step": 3465
    },
    {
      "epoch": 0.31968631069080844,
      "eval_GEN Loss": 0.46864184737205505,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.369650274515152,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8722355961799622,
      "eval_runtime": 56.7646,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3465
    },
    {
      "epoch": 0.31977857225233536,
      "grad_norm": 1.7329564540020683,
      "learning_rate": 1.0024586917536858e-06,
      "loss": 0.7487,
      "step": 3466
    },
    {
      "epoch": 0.3198708338138623,
      "grad_norm": 3.370463703304283,
      "learning_rate": 1.0023087155320183e-06,
      "loss": 0.7237,
      "step": 3467
    },
    {
      "epoch": 0.31996309537538925,
      "grad_norm": 1.8017335990526395,
      "learning_rate": 1.002158705118041e-06,
      "loss": 0.8718,
      "step": 3468
    },
    {
      "epoch": 0.32005535693691617,
      "grad_norm": 2.1581579821709336,
      "learning_rate": 1.0020086605253482e-06,
      "loss": 0.6968,
      "step": 3469
    },
    {
      "epoch": 0.3201476184984431,
      "grad_norm": 1.867968093790674,
      "learning_rate": 1.0018585817675373e-06,
      "loss": 0.8581,
      "step": 3470
    },
    {
      "epoch": 0.3201476184984431,
      "eval_GEN Loss": 0.4694361984729767,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3823939859867096,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8747596144676208,
      "eval_runtime": 56.912,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3470
    },
    {
      "epoch": 0.32023988005997,
      "grad_norm": 2.755711413920739,
      "learning_rate": 1.0017084688582086e-06,
      "loss": 0.8308,
      "step": 3471
    },
    {
      "epoch": 0.3203321416214969,
      "grad_norm": 1.8441340179149308,
      "learning_rate": 1.0015583218109652e-06,
      "loss": 0.8914,
      "step": 3472
    },
    {
      "epoch": 0.3204244031830239,
      "grad_norm": 2.2674160438948654,
      "learning_rate": 1.001408140639414e-06,
      "loss": 0.6273,
      "step": 3473
    },
    {
      "epoch": 0.3205166647445508,
      "grad_norm": 1.7966698742268874,
      "learning_rate": 1.0012579253571648e-06,
      "loss": 0.6454,
      "step": 3474
    },
    {
      "epoch": 0.32060892630607774,
      "grad_norm": 2.2228832871488713,
      "learning_rate": 1.0011076759778301e-06,
      "loss": 0.7326,
      "step": 3475
    },
    {
      "epoch": 0.32060892630607774,
      "eval_GEN Loss": 0.46892285346984863,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.38079655170440674,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8813701868057251,
      "eval_runtime": 55.8428,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3475
    },
    {
      "epoch": 0.32070118786760465,
      "grad_norm": 2.4632508087198017,
      "learning_rate": 1.000957392515026e-06,
      "loss": 0.8512,
      "step": 3476
    },
    {
      "epoch": 0.32079344942913157,
      "grad_norm": 1.8050158259780842,
      "learning_rate": 1.0008070749823709e-06,
      "loss": 0.5992,
      "step": 3477
    },
    {
      "epoch": 0.3208857109906585,
      "grad_norm": 2.481792641995227,
      "learning_rate": 1.0006567233934873e-06,
      "loss": 0.8449,
      "step": 3478
    },
    {
      "epoch": 0.32097797255218546,
      "grad_norm": 2.5834799184142945,
      "learning_rate": 1.000506337762e-06,
      "loss": 0.7446,
      "step": 3479
    },
    {
      "epoch": 0.3210702341137124,
      "grad_norm": 1.488416660909454,
      "learning_rate": 1.0003559181015374e-06,
      "loss": 0.6129,
      "step": 3480
    },
    {
      "epoch": 0.3210702341137124,
      "eval_GEN Loss": 0.4685995578765869,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.37271106243133545,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.885036051273346,
      "eval_runtime": 56.9734,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3480
    },
    {
      "epoch": 0.3211624956752393,
      "grad_norm": 1.6589345037028045,
      "learning_rate": 1.000205464425731e-06,
      "loss": 0.5114,
      "step": 3481
    },
    {
      "epoch": 0.3212547572367662,
      "grad_norm": 1.4823346507551198,
      "learning_rate": 1.0000549767482142e-06,
      "loss": 0.6256,
      "step": 3482
    },
    {
      "epoch": 0.32134701879829314,
      "grad_norm": 2.1792902858080847,
      "learning_rate": 9.999044550826256e-07,
      "loss": 0.8143,
      "step": 3483
    },
    {
      "epoch": 0.3214392803598201,
      "grad_norm": 1.9011944299609913,
      "learning_rate": 9.997538994426049e-07,
      "loss": 0.755,
      "step": 3484
    },
    {
      "epoch": 0.32153154192134703,
      "grad_norm": 1.4360295660088076,
      "learning_rate": 9.99603309841796e-07,
      "loss": 0.6985,
      "step": 3485
    },
    {
      "epoch": 0.32153154192134703,
      "eval_GEN Loss": 0.4663397967815399,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3620414733886719,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9073317050933838,
      "eval_runtime": 57.0555,
      "eval_samples_per_second": 1.139,
      "eval_steps_per_second": 0.088,
      "step": 3485
    },
    {
      "epoch": 0.32162380348287395,
      "grad_norm": 1.4470063670363587,
      "learning_rate": 9.994526862938455e-07,
      "loss": 0.574,
      "step": 3486
    },
    {
      "epoch": 0.32171606504440087,
      "grad_norm": 2.356573665782338,
      "learning_rate": 9.993020288124033e-07,
      "loss": 0.5472,
      "step": 3487
    },
    {
      "epoch": 0.3218083266059278,
      "grad_norm": 1.380305689048948,
      "learning_rate": 9.99151337411122e-07,
      "loss": 0.6889,
      "step": 3488
    },
    {
      "epoch": 0.32190058816745476,
      "grad_norm": 5.131841186139524,
      "learning_rate": 9.990006121036575e-07,
      "loss": 1.1968,
      "step": 3489
    },
    {
      "epoch": 0.3219928497289817,
      "grad_norm": 1.48050417658812,
      "learning_rate": 9.98849852903669e-07,
      "loss": 0.7084,
      "step": 3490
    },
    {
      "epoch": 0.3219928497289817,
      "eval_GEN Loss": 0.4651165306568146,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3570702373981476,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9067908525466919,
      "eval_runtime": 56.7204,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3490
    },
    {
      "epoch": 0.3220851112905086,
      "grad_norm": 1.8579928368458347,
      "learning_rate": 9.986990598248184e-07,
      "loss": 0.7977,
      "step": 3491
    },
    {
      "epoch": 0.3221773728520355,
      "grad_norm": 2.893219702474659,
      "learning_rate": 9.985482328807706e-07,
      "loss": 0.8473,
      "step": 3492
    },
    {
      "epoch": 0.32226963441356243,
      "grad_norm": 1.6827002563814335,
      "learning_rate": 9.983973720851942e-07,
      "loss": 0.6579,
      "step": 3493
    },
    {
      "epoch": 0.32236189597508935,
      "grad_norm": 1.989817046564749,
      "learning_rate": 9.982464774517601e-07,
      "loss": 0.6656,
      "step": 3494
    },
    {
      "epoch": 0.3224541575366163,
      "grad_norm": 2.4968862552563533,
      "learning_rate": 9.980955489941426e-07,
      "loss": 0.6077,
      "step": 3495
    },
    {
      "epoch": 0.3224541575366163,
      "eval_GEN Loss": 0.4649212956428528,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3645148277282715,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8651442527770996,
      "eval_runtime": 56.6942,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3495
    },
    {
      "epoch": 0.32254641909814324,
      "grad_norm": 1.9968034514774233,
      "learning_rate": 9.979445867260192e-07,
      "loss": 0.7793,
      "step": 3496
    },
    {
      "epoch": 0.32263868065967016,
      "grad_norm": 1.8670585681050127,
      "learning_rate": 9.977935906610705e-07,
      "loss": 0.6814,
      "step": 3497
    },
    {
      "epoch": 0.3227309422211971,
      "grad_norm": 2.5632636171198646,
      "learning_rate": 9.976425608129797e-07,
      "loss": 0.5964,
      "step": 3498
    },
    {
      "epoch": 0.322823203782724,
      "grad_norm": 3.296842722098701,
      "learning_rate": 9.974914971954335e-07,
      "loss": 0.8559,
      "step": 3499
    },
    {
      "epoch": 0.322915465344251,
      "grad_norm": 2.117938633792717,
      "learning_rate": 9.973403998221216e-07,
      "loss": 0.8139,
      "step": 3500
    },
    {
      "epoch": 0.322915465344251,
      "eval_GEN Loss": 0.4646627604961395,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3864864408969879,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.851832926273346,
      "eval_runtime": 56.6424,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 3500
    },
    {
      "epoch": 0.3230077269057779,
      "grad_norm": 1.7625420107198013,
      "learning_rate": 9.971892687067365e-07,
      "loss": 0.7544,
      "step": 3501
    },
    {
      "epoch": 0.3230999884673048,
      "grad_norm": 1.543818768500952,
      "learning_rate": 9.970381038629739e-07,
      "loss": 0.6944,
      "step": 3502
    },
    {
      "epoch": 0.32319225002883173,
      "grad_norm": 1.7411614102137767,
      "learning_rate": 9.96886905304533e-07,
      "loss": 0.7008,
      "step": 3503
    },
    {
      "epoch": 0.32328451159035865,
      "grad_norm": 2.4201877683372226,
      "learning_rate": 9.967356730451153e-07,
      "loss": 0.6991,
      "step": 3504
    },
    {
      "epoch": 0.3233767731518856,
      "grad_norm": 1.5902999389333115,
      "learning_rate": 9.965844070984259e-07,
      "loss": 0.5897,
      "step": 3505
    },
    {
      "epoch": 0.3233767731518856,
      "eval_GEN Loss": 0.4636280834674835,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3915148973464966,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8522836565971375,
      "eval_runtime": 56.9852,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3505
    },
    {
      "epoch": 0.32346903471341254,
      "grad_norm": 1.8223841583214533,
      "learning_rate": 9.964331074781724e-07,
      "loss": 0.7535,
      "step": 3506
    },
    {
      "epoch": 0.32356129627493946,
      "grad_norm": 1.8371545605290758,
      "learning_rate": 9.962817741980662e-07,
      "loss": 0.753,
      "step": 3507
    },
    {
      "epoch": 0.3236535578364664,
      "grad_norm": 1.706081604023497,
      "learning_rate": 9.96130407271821e-07,
      "loss": 0.5756,
      "step": 3508
    },
    {
      "epoch": 0.3237458193979933,
      "grad_norm": 1.5201240848822422,
      "learning_rate": 9.959790067131545e-07,
      "loss": 0.5854,
      "step": 3509
    },
    {
      "epoch": 0.3238380809595202,
      "grad_norm": 1.5433507169008192,
      "learning_rate": 9.958275725357864e-07,
      "loss": 0.7748,
      "step": 3510
    },
    {
      "epoch": 0.3238380809595202,
      "eval_GEN Loss": 0.4637582302093506,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.37704744935035706,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8530648946762085,
      "eval_runtime": 56.2812,
      "eval_samples_per_second": 1.155,
      "eval_steps_per_second": 0.089,
      "step": 3510
    },
    {
      "epoch": 0.3239303425210472,
      "grad_norm": 1.9682506816690448,
      "learning_rate": 9.956761047534398e-07,
      "loss": 0.6819,
      "step": 3511
    },
    {
      "epoch": 0.3240226040825741,
      "grad_norm": 2.234304513029621,
      "learning_rate": 9.955246033798412e-07,
      "loss": 0.8584,
      "step": 3512
    },
    {
      "epoch": 0.324114865644101,
      "grad_norm": 2.273810968616627,
      "learning_rate": 9.9537306842872e-07,
      "loss": 0.7094,
      "step": 3513
    },
    {
      "epoch": 0.32420712720562794,
      "grad_norm": 1.7473241154353563,
      "learning_rate": 9.95221499913808e-07,
      "loss": 0.6667,
      "step": 3514
    },
    {
      "epoch": 0.32429938876715486,
      "grad_norm": 2.204837392640544,
      "learning_rate": 9.950698978488413e-07,
      "loss": 0.7814,
      "step": 3515
    },
    {
      "epoch": 0.32429938876715486,
      "eval_GEN Loss": 0.46310362219810486,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.36104968190193176,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8682692050933838,
      "eval_runtime": 56.0223,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 3515
    },
    {
      "epoch": 0.32439165032868184,
      "grad_norm": 3.279985801957001,
      "learning_rate": 9.949182622475578e-07,
      "loss": 0.754,
      "step": 3516
    },
    {
      "epoch": 0.32448391189020875,
      "grad_norm": 1.6149360580825767,
      "learning_rate": 9.94766593123699e-07,
      "loss": 0.7358,
      "step": 3517
    },
    {
      "epoch": 0.3245761734517357,
      "grad_norm": 2.603168287879418,
      "learning_rate": 9.946148904910097e-07,
      "loss": 0.6986,
      "step": 3518
    },
    {
      "epoch": 0.3246684350132626,
      "grad_norm": 1.5952717176591018,
      "learning_rate": 9.94463154363237e-07,
      "loss": 0.7918,
      "step": 3519
    },
    {
      "epoch": 0.3247606965747895,
      "grad_norm": 3.494653024342418,
      "learning_rate": 9.94311384754132e-07,
      "loss": 0.8076,
      "step": 3520
    },
    {
      "epoch": 0.3247606965747895,
      "eval_GEN Loss": 0.46455931663513184,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.36879345774650574,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8652644157409668,
      "eval_runtime": 55.8298,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3520
    },
    {
      "epoch": 0.3248529581363165,
      "grad_norm": 1.6094419502732449,
      "learning_rate": 9.94159581677448e-07,
      "loss": 0.6275,
      "step": 3521
    },
    {
      "epoch": 0.3249452196978434,
      "grad_norm": 2.2631956566098324,
      "learning_rate": 9.940077451469415e-07,
      "loss": 0.7404,
      "step": 3522
    },
    {
      "epoch": 0.3250374812593703,
      "grad_norm": 1.967997279588366,
      "learning_rate": 9.938558751763722e-07,
      "loss": 0.7002,
      "step": 3523
    },
    {
      "epoch": 0.32512974282089724,
      "grad_norm": 1.4462636431295166,
      "learning_rate": 9.937039717795033e-07,
      "loss": 0.6084,
      "step": 3524
    },
    {
      "epoch": 0.32522200438242416,
      "grad_norm": 3.3220946326326035,
      "learning_rate": 9.935520349700998e-07,
      "loss": 1.0089,
      "step": 3525
    },
    {
      "epoch": 0.32522200438242416,
      "eval_GEN Loss": 0.46530255675315857,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.38551005721092224,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8630408644676208,
      "eval_runtime": 55.9978,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3525
    },
    {
      "epoch": 0.3253142659439511,
      "grad_norm": 1.2702398349924198,
      "learning_rate": 9.934000647619308e-07,
      "loss": 0.6857,
      "step": 3526
    },
    {
      "epoch": 0.32540652750547805,
      "grad_norm": 1.4903996596380185,
      "learning_rate": 9.932480611687682e-07,
      "loss": 0.5613,
      "step": 3527
    },
    {
      "epoch": 0.32549878906700497,
      "grad_norm": 1.615800743168112,
      "learning_rate": 9.930960242043866e-07,
      "loss": 0.7292,
      "step": 3528
    },
    {
      "epoch": 0.3255910506285319,
      "grad_norm": 1.754218562582439,
      "learning_rate": 9.929439538825638e-07,
      "loss": 0.7225,
      "step": 3529
    },
    {
      "epoch": 0.3256833121900588,
      "grad_norm": 1.47978871431795,
      "learning_rate": 9.927918502170807e-07,
      "loss": 0.5319,
      "step": 3530
    },
    {
      "epoch": 0.3256833121900588,
      "eval_GEN Loss": 0.46523532271385193,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.39558637142181396,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8748798370361328,
      "eval_runtime": 55.8385,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3530
    },
    {
      "epoch": 0.3257755737515857,
      "grad_norm": 1.44946982046061,
      "learning_rate": 9.926397132217215e-07,
      "loss": 0.7779,
      "step": 3531
    },
    {
      "epoch": 0.3258678353131127,
      "grad_norm": 1.3606095037000285,
      "learning_rate": 9.924875429102726e-07,
      "loss": 0.5909,
      "step": 3532
    },
    {
      "epoch": 0.3259600968746396,
      "grad_norm": 1.8284814268112917,
      "learning_rate": 9.923353392965242e-07,
      "loss": 0.5984,
      "step": 3533
    },
    {
      "epoch": 0.32605235843616653,
      "grad_norm": 2.3443389162735913,
      "learning_rate": 9.92183102394269e-07,
      "loss": 0.747,
      "step": 3534
    },
    {
      "epoch": 0.32614461999769345,
      "grad_norm": 1.8918598611991704,
      "learning_rate": 9.920308322173034e-07,
      "loss": 0.7048,
      "step": 3535
    },
    {
      "epoch": 0.32614461999769345,
      "eval_GEN Loss": 0.46586254239082336,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.40054383873939514,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8873196840286255,
      "eval_runtime": 56.3702,
      "eval_samples_per_second": 1.153,
      "eval_steps_per_second": 0.089,
      "step": 3535
    },
    {
      "epoch": 0.32623688155922037,
      "grad_norm": 2.1254200252964237,
      "learning_rate": 9.918785287794258e-07,
      "loss": 0.6929,
      "step": 3536
    },
    {
      "epoch": 0.32632914312074734,
      "grad_norm": 2.922619342579277,
      "learning_rate": 9.917261920944386e-07,
      "loss": 0.8515,
      "step": 3537
    },
    {
      "epoch": 0.32642140468227426,
      "grad_norm": 1.6413771771773444,
      "learning_rate": 9.915738221761467e-07,
      "loss": 0.62,
      "step": 3538
    },
    {
      "epoch": 0.3265136662438012,
      "grad_norm": 2.4277621445063735,
      "learning_rate": 9.91421419038358e-07,
      "loss": 0.6826,
      "step": 3539
    },
    {
      "epoch": 0.3266059278053281,
      "grad_norm": 1.8688089590377583,
      "learning_rate": 9.912689826948835e-07,
      "loss": 0.6656,
      "step": 3540
    },
    {
      "epoch": 0.3266059278053281,
      "eval_GEN Loss": 0.46760138869285583,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.41551968455314636,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8883413672447205,
      "eval_runtime": 55.8784,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3540
    },
    {
      "epoch": 0.326698189366855,
      "grad_norm": 2.3097141831984227,
      "learning_rate": 9.911165131595372e-07,
      "loss": 0.8522,
      "step": 3541
    },
    {
      "epoch": 0.32679045092838194,
      "grad_norm": 2.0738725418081287,
      "learning_rate": 9.909640104461363e-07,
      "loss": 0.8364,
      "step": 3542
    },
    {
      "epoch": 0.3268827124899089,
      "grad_norm": 2.6397397055896996,
      "learning_rate": 9.908114745685006e-07,
      "loss": 0.8954,
      "step": 3543
    },
    {
      "epoch": 0.32697497405143583,
      "grad_norm": 2.4376903133080288,
      "learning_rate": 9.906589055404534e-07,
      "loss": 0.8489,
      "step": 3544
    },
    {
      "epoch": 0.32706723561296275,
      "grad_norm": 1.9557592094536642,
      "learning_rate": 9.905063033758204e-07,
      "loss": 0.7317,
      "step": 3545
    },
    {
      "epoch": 0.32706723561296275,
      "eval_GEN Loss": 0.46769118309020996,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.43777814507484436,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8790565133094788,
      "eval_runtime": 55.8401,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3545
    },
    {
      "epoch": 0.32715949717448967,
      "grad_norm": 2.887833227549036,
      "learning_rate": 9.903536680884312e-07,
      "loss": 0.7556,
      "step": 3546
    },
    {
      "epoch": 0.3272517587360166,
      "grad_norm": 2.0135855167794174,
      "learning_rate": 9.902009996921174e-07,
      "loss": 0.8285,
      "step": 3547
    },
    {
      "epoch": 0.32734402029754356,
      "grad_norm": 1.6496526739939865,
      "learning_rate": 9.900482982007143e-07,
      "loss": 0.7358,
      "step": 3548
    },
    {
      "epoch": 0.3274362818590705,
      "grad_norm": 1.8934874830584503,
      "learning_rate": 9.898955636280596e-07,
      "loss": 0.7606,
      "step": 3549
    },
    {
      "epoch": 0.3275285434205974,
      "grad_norm": 1.6708224328009713,
      "learning_rate": 9.897427959879946e-07,
      "loss": 0.6658,
      "step": 3550
    },
    {
      "epoch": 0.3275285434205974,
      "eval_GEN Loss": 0.470447838306427,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8441558441558441,
      "eval_PRM F1 AUC": 0.7611314824515453,
      "eval_PRM F1 Neg": 0.5862068965517241,
      "eval_PRM Loss": 0.46784281730651855,
      "eval_PRM NPV": 0.4857142857142857,
      "eval_PRM Precision": 0.9154929577464789,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8864783644676208,
      "eval_runtime": 55.7936,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3550
    },
    {
      "epoch": 0.3276208049821243,
      "grad_norm": 1.9864574515885203,
      "learning_rate": 9.895899952943635e-07,
      "loss": 0.9907,
      "step": 3551
    },
    {
      "epoch": 0.32771306654365123,
      "grad_norm": 2.0995437303723774,
      "learning_rate": 9.89437161561013e-07,
      "loss": 0.7183,
      "step": 3552
    },
    {
      "epoch": 0.3278053281051782,
      "grad_norm": 2.986297370252215,
      "learning_rate": 9.892842948017933e-07,
      "loss": 0.7068,
      "step": 3553
    },
    {
      "epoch": 0.3278975896667051,
      "grad_norm": 2.518680116744529,
      "learning_rate": 9.891313950305576e-07,
      "loss": 0.7512,
      "step": 3554
    },
    {
      "epoch": 0.32798985122823204,
      "grad_norm": 2.800470026143454,
      "learning_rate": 9.889784622611614e-07,
      "loss": 0.6876,
      "step": 3555
    },
    {
      "epoch": 0.32798985122823204,
      "eval_GEN Loss": 0.4716262221336365,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8516129032258064,
      "eval_PRM F1 AUC": 0.7671555788370874,
      "eval_PRM F1 Neg": 0.5964912280701754,
      "eval_PRM Loss": 0.4461728632450104,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9166666666666666,
      "eval_PRM Recall": 0.7951807228915663,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8704928159713745,
      "eval_runtime": 55.986,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3555
    },
    {
      "epoch": 0.32808211278975896,
      "grad_norm": 1.7422711145725653,
      "learning_rate": 9.888254965074643e-07,
      "loss": 0.6865,
      "step": 3556
    },
    {
      "epoch": 0.3281743743512859,
      "grad_norm": 2.3397409083732446,
      "learning_rate": 9.886724977833278e-07,
      "loss": 0.7051,
      "step": 3557
    },
    {
      "epoch": 0.3282666359128128,
      "grad_norm": 1.9108475400470184,
      "learning_rate": 9.885194661026174e-07,
      "loss": 0.6331,
      "step": 3558
    },
    {
      "epoch": 0.3283588974743398,
      "grad_norm": 1.5522411601505866,
      "learning_rate": 9.883664014792005e-07,
      "loss": 0.6853,
      "step": 3559
    },
    {
      "epoch": 0.3284511590358667,
      "grad_norm": 2.229268100481141,
      "learning_rate": 9.882133039269486e-07,
      "loss": 0.7715,
      "step": 3560
    },
    {
      "epoch": 0.3284511590358667,
      "eval_GEN Loss": 0.46998584270477295,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3751089572906494,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8559495210647583,
      "eval_runtime": 55.8954,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3560
    },
    {
      "epoch": 0.3285434205973936,
      "grad_norm": 2.098409479358096,
      "learning_rate": 9.880601734597354e-07,
      "loss": 0.648,
      "step": 3561
    },
    {
      "epoch": 0.32863568215892053,
      "grad_norm": 1.5486454764490598,
      "learning_rate": 9.879070100914375e-07,
      "loss": 0.7608,
      "step": 3562
    },
    {
      "epoch": 0.32872794372044745,
      "grad_norm": 1.4930135376285247,
      "learning_rate": 9.877538138359355e-07,
      "loss": 0.6318,
      "step": 3563
    },
    {
      "epoch": 0.3288202052819744,
      "grad_norm": 3.033861758280288,
      "learning_rate": 9.876005847071117e-07,
      "loss": 0.8673,
      "step": 3564
    },
    {
      "epoch": 0.32891246684350134,
      "grad_norm": 1.8478406721587446,
      "learning_rate": 9.874473227188523e-07,
      "loss": 0.6408,
      "step": 3565
    },
    {
      "epoch": 0.32891246684350134,
      "eval_GEN Loss": 0.4700247645378113,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.35278984904289246,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8729567527770996,
      "eval_runtime": 56.8694,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3565
    },
    {
      "epoch": 0.32900472840502826,
      "grad_norm": 1.721789085268667,
      "learning_rate": 9.87294027885046e-07,
      "loss": 0.6808,
      "step": 3566
    },
    {
      "epoch": 0.3290969899665552,
      "grad_norm": 1.929773975651704,
      "learning_rate": 9.871407002195845e-07,
      "loss": 0.6625,
      "step": 3567
    },
    {
      "epoch": 0.3291892515280821,
      "grad_norm": 2.4115324080600087,
      "learning_rate": 9.869873397363628e-07,
      "loss": 0.7695,
      "step": 3568
    },
    {
      "epoch": 0.32928151308960907,
      "grad_norm": 2.0794827013150656,
      "learning_rate": 9.868339464492785e-07,
      "loss": 0.7534,
      "step": 3569
    },
    {
      "epoch": 0.329373774651136,
      "grad_norm": 2.420888531379509,
      "learning_rate": 9.866805203722327e-07,
      "loss": 0.9365,
      "step": 3570
    },
    {
      "epoch": 0.329373774651136,
      "eval_GEN Loss": 0.467998743057251,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3475819230079651,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8640024065971375,
      "eval_runtime": 56.7779,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3570
    },
    {
      "epoch": 0.3294660362126629,
      "grad_norm": 2.0227342219705027,
      "learning_rate": 9.865270615191285e-07,
      "loss": 0.6998,
      "step": 3571
    },
    {
      "epoch": 0.3295582977741898,
      "grad_norm": 1.6543561218331455,
      "learning_rate": 9.863735699038732e-07,
      "loss": 0.6135,
      "step": 3572
    },
    {
      "epoch": 0.32965055933571674,
      "grad_norm": 2.194018305326554,
      "learning_rate": 9.86220045540376e-07,
      "loss": 0.9116,
      "step": 3573
    },
    {
      "epoch": 0.32974282089724366,
      "grad_norm": 1.8454937454144151,
      "learning_rate": 9.860664884425499e-07,
      "loss": 0.7856,
      "step": 3574
    },
    {
      "epoch": 0.32983508245877063,
      "grad_norm": 1.634498692436031,
      "learning_rate": 9.859128986243102e-07,
      "loss": 0.7193,
      "step": 3575
    },
    {
      "epoch": 0.32983508245877063,
      "eval_GEN Loss": 0.46865424513816833,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.34768542647361755,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8395432829856873,
      "eval_runtime": 56.7471,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3575
    },
    {
      "epoch": 0.32992734402029755,
      "grad_norm": 1.357543461498226,
      "learning_rate": 9.857592760995755e-07,
      "loss": 0.6006,
      "step": 3576
    },
    {
      "epoch": 0.33001960558182447,
      "grad_norm": 2.224208403060692,
      "learning_rate": 9.856056208822674e-07,
      "loss": 0.8296,
      "step": 3577
    },
    {
      "epoch": 0.3301118671433514,
      "grad_norm": 1.8696798446162632,
      "learning_rate": 9.854519329863102e-07,
      "loss": 0.6758,
      "step": 3578
    },
    {
      "epoch": 0.3302041287048783,
      "grad_norm": 2.034689805212518,
      "learning_rate": 9.852982124256317e-07,
      "loss": 0.6788,
      "step": 3579
    },
    {
      "epoch": 0.3302963902664053,
      "grad_norm": 1.4513004451549032,
      "learning_rate": 9.85144459214162e-07,
      "loss": 0.7563,
      "step": 3580
    },
    {
      "epoch": 0.3302963902664053,
      "eval_GEN Loss": 0.46821364760398865,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3536064326763153,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8288461565971375,
      "eval_runtime": 55.7502,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 3580
    },
    {
      "epoch": 0.3303886518279322,
      "grad_norm": 2.093938427424334,
      "learning_rate": 9.849906733658344e-07,
      "loss": 0.8765,
      "step": 3581
    },
    {
      "epoch": 0.3304809133894591,
      "grad_norm": 1.5655414643833065,
      "learning_rate": 9.848368548945855e-07,
      "loss": 0.7538,
      "step": 3582
    },
    {
      "epoch": 0.33057317495098604,
      "grad_norm": 1.5621782867141305,
      "learning_rate": 9.846830038143544e-07,
      "loss": 0.677,
      "step": 3583
    },
    {
      "epoch": 0.33066543651251296,
      "grad_norm": 1.7419478060144966,
      "learning_rate": 9.845291201390833e-07,
      "loss": 0.7433,
      "step": 3584
    },
    {
      "epoch": 0.33075769807403993,
      "grad_norm": 1.4566245925145112,
      "learning_rate": 9.843752038827174e-07,
      "loss": 0.6689,
      "step": 3585
    },
    {
      "epoch": 0.33075769807403993,
      "eval_GEN Loss": 0.467538058757782,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.3611035943031311,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8266226053237915,
      "eval_runtime": 55.9986,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3585
    },
    {
      "epoch": 0.33084995963556685,
      "grad_norm": 1.7653443118561036,
      "learning_rate": 9.842212550592048e-07,
      "loss": 0.5293,
      "step": 3586
    },
    {
      "epoch": 0.33094222119709377,
      "grad_norm": 1.9766834292208788,
      "learning_rate": 9.840672736824968e-07,
      "loss": 0.6617,
      "step": 3587
    },
    {
      "epoch": 0.3310344827586207,
      "grad_norm": 2.154507425593431,
      "learning_rate": 9.839132597665472e-07,
      "loss": 0.7231,
      "step": 3588
    },
    {
      "epoch": 0.3311267443201476,
      "grad_norm": 1.9482214572827599,
      "learning_rate": 9.837592133253131e-07,
      "loss": 0.827,
      "step": 3589
    },
    {
      "epoch": 0.3312190058816745,
      "grad_norm": 1.5324463255142962,
      "learning_rate": 9.836051343727542e-07,
      "loss": 0.6762,
      "step": 3590
    },
    {
      "epoch": 0.3312190058816745,
      "eval_GEN Loss": 0.4666840732097626,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3566901385784149,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8296874761581421,
      "eval_runtime": 55.9346,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3590
    },
    {
      "epoch": 0.3313112674432015,
      "grad_norm": 2.1318209293480193,
      "learning_rate": 9.83451022922834e-07,
      "loss": 0.7061,
      "step": 3591
    },
    {
      "epoch": 0.3314035290047284,
      "grad_norm": 2.0365177450658285,
      "learning_rate": 9.832968789895175e-07,
      "loss": 0.616,
      "step": 3592
    },
    {
      "epoch": 0.33149579056625533,
      "grad_norm": 1.8458554826737847,
      "learning_rate": 9.83142702586774e-07,
      "loss": 0.7136,
      "step": 3593
    },
    {
      "epoch": 0.33158805212778225,
      "grad_norm": 1.8249708828529854,
      "learning_rate": 9.82988493728575e-07,
      "loss": 0.6292,
      "step": 3594
    },
    {
      "epoch": 0.33168031368930917,
      "grad_norm": 1.4049328479545602,
      "learning_rate": 9.828342524288952e-07,
      "loss": 0.6941,
      "step": 3595
    },
    {
      "epoch": 0.33168031368930917,
      "eval_GEN Loss": 0.4676535129547119,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.35014253854751587,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8380408883094788,
      "eval_runtime": 55.8272,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3595
    },
    {
      "epoch": 0.33177257525083614,
      "grad_norm": 1.918620717524742,
      "learning_rate": 9.826799787017125e-07,
      "loss": 0.6852,
      "step": 3596
    },
    {
      "epoch": 0.33186483681236306,
      "grad_norm": 2.7821418769492015,
      "learning_rate": 9.82525672561007e-07,
      "loss": 0.7714,
      "step": 3597
    },
    {
      "epoch": 0.33195709837389,
      "grad_norm": 2.7324268536076075,
      "learning_rate": 9.82371334020762e-07,
      "loss": 0.6198,
      "step": 3598
    },
    {
      "epoch": 0.3320493599354169,
      "grad_norm": 1.7560132191259046,
      "learning_rate": 9.822169630949646e-07,
      "loss": 0.7944,
      "step": 3599
    },
    {
      "epoch": 0.3321416214969438,
      "grad_norm": 1.9811400790997766,
      "learning_rate": 9.820625597976034e-07,
      "loss": 0.757,
      "step": 3600
    },
    {
      "epoch": 0.3321416214969438,
      "eval_GEN Loss": 0.46804919838905334,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.35448625683784485,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8407752513885498,
      "eval_runtime": 56.8251,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3600
    },
    {
      "epoch": 0.3322338830584708,
      "grad_norm": 2.317114525595054,
      "learning_rate": 9.819081241426712e-07,
      "loss": 0.8489,
      "step": 3601
    },
    {
      "epoch": 0.3323261446199977,
      "grad_norm": 2.3501109658797543,
      "learning_rate": 9.81753656144163e-07,
      "loss": 0.9152,
      "step": 3602
    },
    {
      "epoch": 0.33241840618152463,
      "grad_norm": 2.659171281274666,
      "learning_rate": 9.815991558160767e-07,
      "loss": 0.7636,
      "step": 3603
    },
    {
      "epoch": 0.33251066774305155,
      "grad_norm": 1.5818614548927388,
      "learning_rate": 9.81444623172414e-07,
      "loss": 0.7166,
      "step": 3604
    },
    {
      "epoch": 0.33260292930457847,
      "grad_norm": 1.9309032159798014,
      "learning_rate": 9.812900582271782e-07,
      "loss": 0.7325,
      "step": 3605
    },
    {
      "epoch": 0.33260292930457847,
      "eval_GEN Loss": 0.47041070461273193,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3725634813308716,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8425480723381042,
      "eval_runtime": 56.7459,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3605
    },
    {
      "epoch": 0.3326951908661054,
      "grad_norm": 1.997389389637752,
      "learning_rate": 9.811354609943768e-07,
      "loss": 0.8185,
      "step": 3606
    },
    {
      "epoch": 0.33278745242763236,
      "grad_norm": 2.4516550611031187,
      "learning_rate": 9.80980831488019e-07,
      "loss": 0.8558,
      "step": 3607
    },
    {
      "epoch": 0.3328797139891593,
      "grad_norm": 3.098445188537611,
      "learning_rate": 9.808261697221182e-07,
      "loss": 0.821,
      "step": 3608
    },
    {
      "epoch": 0.3329719755506862,
      "grad_norm": 3.3512603523443776,
      "learning_rate": 9.806714757106896e-07,
      "loss": 0.7201,
      "step": 3609
    },
    {
      "epoch": 0.3330642371122131,
      "grad_norm": 1.7652329363446022,
      "learning_rate": 9.805167494677522e-07,
      "loss": 0.6832,
      "step": 3610
    },
    {
      "epoch": 0.3330642371122131,
      "eval_GEN Loss": 0.471610963344574,
      "eval_GEN top-5 accuracy": 0.9850942569048663,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.39141222834587097,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8512319922447205,
      "eval_runtime": 56.7684,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3610
    },
    {
      "epoch": 0.33315649867374003,
      "grad_norm": 1.725382749587872,
      "learning_rate": 9.803619910073273e-07,
      "loss": 0.6388,
      "step": 3611
    },
    {
      "epoch": 0.333248760235267,
      "grad_norm": 1.3562386428929896,
      "learning_rate": 9.802072003434395e-07,
      "loss": 0.4694,
      "step": 3612
    },
    {
      "epoch": 0.3333410217967939,
      "grad_norm": 2.178595676424367,
      "learning_rate": 9.80052377490116e-07,
      "loss": 0.9785,
      "step": 3613
    },
    {
      "epoch": 0.33343328335832084,
      "grad_norm": 1.9166366676661377,
      "learning_rate": 9.798975224613874e-07,
      "loss": 0.7763,
      "step": 3614
    },
    {
      "epoch": 0.33352554491984776,
      "grad_norm": 2.886258933560402,
      "learning_rate": 9.797426352712865e-07,
      "loss": 0.9123,
      "step": 3615
    },
    {
      "epoch": 0.33352554491984776,
      "eval_GEN Loss": 0.474600613117218,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8496732026143791,
      "eval_PRM F1 AUC": 0.782870612886328,
      "eval_PRM F1 Neg": 0.6101694915254238,
      "eval_PRM Loss": 0.4381133019924164,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9285714285714286,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8804687261581421,
      "eval_runtime": 55.5869,
      "eval_samples_per_second": 1.169,
      "eval_steps_per_second": 0.09,
      "step": 3615
    },
    {
      "epoch": 0.3336178064813747,
      "grad_norm": 2.193058438376044,
      "learning_rate": 9.795877159338495e-07,
      "loss": 0.8594,
      "step": 3616
    },
    {
      "epoch": 0.33371006804290165,
      "grad_norm": 2.1883940438320395,
      "learning_rate": 9.794327644631158e-07,
      "loss": 0.8318,
      "step": 3617
    },
    {
      "epoch": 0.33380232960442857,
      "grad_norm": 1.6312277262767367,
      "learning_rate": 9.792777808731272e-07,
      "loss": 0.8312,
      "step": 3618
    },
    {
      "epoch": 0.3338945911659555,
      "grad_norm": 2.1066961426840316,
      "learning_rate": 9.791227651779281e-07,
      "loss": 0.5322,
      "step": 3619
    },
    {
      "epoch": 0.3339868527274824,
      "grad_norm": 1.4272026742780817,
      "learning_rate": 9.789677173915668e-07,
      "loss": 0.6554,
      "step": 3620
    },
    {
      "epoch": 0.3339868527274824,
      "eval_GEN Loss": 0.4727225601673126,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.42424073815345764,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8775841593742371,
      "eval_runtime": 55.7186,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3620
    },
    {
      "epoch": 0.3340791142890093,
      "grad_norm": 2.145798214258014,
      "learning_rate": 9.788126375280938e-07,
      "loss": 0.8423,
      "step": 3621
    },
    {
      "epoch": 0.33417137585053625,
      "grad_norm": 2.9563684798156435,
      "learning_rate": 9.786575256015626e-07,
      "loss": 0.8395,
      "step": 3622
    },
    {
      "epoch": 0.3342636374120632,
      "grad_norm": 1.7965145492137091,
      "learning_rate": 9.7850238162603e-07,
      "loss": 0.7023,
      "step": 3623
    },
    {
      "epoch": 0.33435589897359014,
      "grad_norm": 1.7984494777458995,
      "learning_rate": 9.78347205615555e-07,
      "loss": 0.5726,
      "step": 3624
    },
    {
      "epoch": 0.33444816053511706,
      "grad_norm": 1.406845092088345,
      "learning_rate": 9.781919975842e-07,
      "loss": 0.7181,
      "step": 3625
    },
    {
      "epoch": 0.33444816053511706,
      "eval_GEN Loss": 0.47157713770866394,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.39016249775886536,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8611177802085876,
      "eval_runtime": 56.817,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3625
    },
    {
      "epoch": 0.334540422096644,
      "grad_norm": 1.5375369962792949,
      "learning_rate": 9.780367575460303e-07,
      "loss": 0.7015,
      "step": 3626
    },
    {
      "epoch": 0.3346326836581709,
      "grad_norm": 2.5355376845655746,
      "learning_rate": 9.778814855151138e-07,
      "loss": 0.8713,
      "step": 3627
    },
    {
      "epoch": 0.33472494521969787,
      "grad_norm": 1.564816302027679,
      "learning_rate": 9.77726181505522e-07,
      "loss": 0.8189,
      "step": 3628
    },
    {
      "epoch": 0.3348172067812248,
      "grad_norm": 2.1185382668919535,
      "learning_rate": 9.775708455313283e-07,
      "loss": 0.9248,
      "step": 3629
    },
    {
      "epoch": 0.3349094683427517,
      "grad_norm": 2.394410579705335,
      "learning_rate": 9.774154776066093e-07,
      "loss": 0.8632,
      "step": 3630
    },
    {
      "epoch": 0.3349094683427517,
      "eval_GEN Loss": 0.4705427289009094,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.35791870951652527,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8482271432876587,
      "eval_runtime": 56.6399,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 3630
    },
    {
      "epoch": 0.3350017299042786,
      "grad_norm": 2.375203554586488,
      "learning_rate": 9.772600777454452e-07,
      "loss": 0.7267,
      "step": 3631
    },
    {
      "epoch": 0.33509399146580554,
      "grad_norm": 1.5536616085963402,
      "learning_rate": 9.771046459619185e-07,
      "loss": 0.7144,
      "step": 3632
    },
    {
      "epoch": 0.3351862530273325,
      "grad_norm": 2.492059887652298,
      "learning_rate": 9.769491822701144e-07,
      "loss": 0.8739,
      "step": 3633
    },
    {
      "epoch": 0.33527851458885943,
      "grad_norm": 1.7183116395532245,
      "learning_rate": 9.767936866841215e-07,
      "loss": 0.6874,
      "step": 3634
    },
    {
      "epoch": 0.33537077615038635,
      "grad_norm": 1.7403832046999108,
      "learning_rate": 9.76638159218031e-07,
      "loss": 0.7693,
      "step": 3635
    },
    {
      "epoch": 0.33537077615038635,
      "eval_GEN Loss": 0.4694426655769348,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3402249217033386,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8430889248847961,
      "eval_runtime": 56.7164,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3635
    },
    {
      "epoch": 0.33546303771191327,
      "grad_norm": 1.259890347045147,
      "learning_rate": 9.76482599885937e-07,
      "loss": 0.5989,
      "step": 3636
    },
    {
      "epoch": 0.3355552992734402,
      "grad_norm": 1.5214473482294408,
      "learning_rate": 9.763270087019366e-07,
      "loss": 0.5695,
      "step": 3637
    },
    {
      "epoch": 0.3356475608349671,
      "grad_norm": 2.1584421759239234,
      "learning_rate": 9.761713856801297e-07,
      "loss": 0.8237,
      "step": 3638
    },
    {
      "epoch": 0.3357398223964941,
      "grad_norm": 2.7809056361146203,
      "learning_rate": 9.76015730834619e-07,
      "loss": 0.8319,
      "step": 3639
    },
    {
      "epoch": 0.335832083958021,
      "grad_norm": 2.4080382572933536,
      "learning_rate": 9.758600441795104e-07,
      "loss": 0.7404,
      "step": 3640
    },
    {
      "epoch": 0.335832083958021,
      "eval_GEN Loss": 0.4696483612060547,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.32540377974510193,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8377704620361328,
      "eval_runtime": 56.5814,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 3640
    },
    {
      "epoch": 0.3359243455195479,
      "grad_norm": 2.0434367395612214,
      "learning_rate": 9.75704325728912e-07,
      "loss": 0.727,
      "step": 3641
    },
    {
      "epoch": 0.33601660708107484,
      "grad_norm": 2.4873363449877135,
      "learning_rate": 9.75548575496936e-07,
      "loss": 0.6249,
      "step": 3642
    },
    {
      "epoch": 0.33610886864260175,
      "grad_norm": 1.7978561149299006,
      "learning_rate": 9.753927934976963e-07,
      "loss": 0.6698,
      "step": 3643
    },
    {
      "epoch": 0.33620113020412873,
      "grad_norm": 1.6167632990347218,
      "learning_rate": 9.752369797453098e-07,
      "loss": 0.625,
      "step": 3644
    },
    {
      "epoch": 0.33629339176565565,
      "grad_norm": 1.5494300264674061,
      "learning_rate": 9.75081134253897e-07,
      "loss": 0.6728,
      "step": 3645
    },
    {
      "epoch": 0.33629339176565565,
      "eval_GEN Loss": 0.46915918588638306,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.32989072799682617,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8176382184028625,
      "eval_runtime": 56.6609,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3645
    },
    {
      "epoch": 0.33638565332718257,
      "grad_norm": 1.8593922337502635,
      "learning_rate": 9.749252570375807e-07,
      "loss": 0.6463,
      "step": 3646
    },
    {
      "epoch": 0.3364779148887095,
      "grad_norm": 2.1989756024765894,
      "learning_rate": 9.747693481104871e-07,
      "loss": 0.7323,
      "step": 3647
    },
    {
      "epoch": 0.3365701764502364,
      "grad_norm": 1.6801286671949187,
      "learning_rate": 9.746134074867445e-07,
      "loss": 0.7104,
      "step": 3648
    },
    {
      "epoch": 0.3366624380117634,
      "grad_norm": 2.827493765635504,
      "learning_rate": 9.744574351804847e-07,
      "loss": 0.8494,
      "step": 3649
    },
    {
      "epoch": 0.3367546995732903,
      "grad_norm": 2.021652059715509,
      "learning_rate": 9.743014312058419e-07,
      "loss": 0.7701,
      "step": 3650
    },
    {
      "epoch": 0.3367546995732903,
      "eval_GEN Loss": 0.46834027767181396,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.35319727659225464,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8159555196762085,
      "eval_runtime": 55.7016,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3650
    },
    {
      "epoch": 0.3368469611348172,
      "grad_norm": 1.5189083616034036,
      "learning_rate": 9.741453955769537e-07,
      "loss": 0.5413,
      "step": 3651
    },
    {
      "epoch": 0.33693922269634413,
      "grad_norm": 2.198977083488279,
      "learning_rate": 9.739893283079602e-07,
      "loss": 0.5491,
      "step": 3652
    },
    {
      "epoch": 0.33703148425787105,
      "grad_norm": 2.098928406083978,
      "learning_rate": 9.738332294130042e-07,
      "loss": 0.8467,
      "step": 3653
    },
    {
      "epoch": 0.33712374581939797,
      "grad_norm": 2.177492651983075,
      "learning_rate": 9.736770989062323e-07,
      "loss": 0.6042,
      "step": 3654
    },
    {
      "epoch": 0.33721600738092494,
      "grad_norm": 1.915251927667097,
      "learning_rate": 9.735209368017925e-07,
      "loss": 0.8345,
      "step": 3655
    },
    {
      "epoch": 0.33721600738092494,
      "eval_GEN Loss": 0.46902939677238464,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.89171974522293,
      "eval_PRM F1 AUC": 0.8347302252488213,
      "eval_PRM F1 Neg": 0.6909090909090909,
      "eval_PRM Loss": 0.3785856068134308,
      "eval_PRM NPV": 0.59375,
      "eval_PRM Precision": 0.9459459459459459,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8260869565217391,
      "eval_loss": 0.828906238079071,
      "eval_runtime": 56.2284,
      "eval_samples_per_second": 1.156,
      "eval_steps_per_second": 0.089,
      "step": 3655
    },
    {
      "epoch": 0.33730826894245186,
      "grad_norm": 2.489470358440496,
      "learning_rate": 9.733647431138372e-07,
      "loss": 0.7829,
      "step": 3656
    },
    {
      "epoch": 0.3374005305039788,
      "grad_norm": 2.248705874883505,
      "learning_rate": 9.732085178565203e-07,
      "loss": 0.7285,
      "step": 3657
    },
    {
      "epoch": 0.3374927920655057,
      "grad_norm": 2.1206303218948874,
      "learning_rate": 9.730522610439993e-07,
      "loss": 0.7276,
      "step": 3658
    },
    {
      "epoch": 0.3375850536270326,
      "grad_norm": 1.904699035757697,
      "learning_rate": 9.72895972690435e-07,
      "loss": 0.5263,
      "step": 3659
    },
    {
      "epoch": 0.3376773151885596,
      "grad_norm": 1.7465993965482869,
      "learning_rate": 9.727396528099895e-07,
      "loss": 0.6408,
      "step": 3660
    },
    {
      "epoch": 0.3376773151885596,
      "eval_GEN Loss": 0.468569815158844,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8903225806451613,
      "eval_PRM F1 AUC": 0.8504452592980618,
      "eval_PRM F1 Neg": 0.7017543859649122,
      "eval_PRM Loss": 0.38954806327819824,
      "eval_PRM NPV": 0.5882352941176471,
      "eval_PRM Precision": 0.9583333333333334,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.8366886973381042,
      "eval_runtime": 55.7754,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3660
    },
    {
      "epoch": 0.3377695767500865,
      "grad_norm": 1.7376465982070755,
      "learning_rate": 9.725833014168299e-07,
      "loss": 0.7115,
      "step": 3661
    },
    {
      "epoch": 0.3378618383116134,
      "grad_norm": 2.417376726737251,
      "learning_rate": 9.72426918525124e-07,
      "loss": 1.0539,
      "step": 3662
    },
    {
      "epoch": 0.33795409987314035,
      "grad_norm": 1.7637093071233916,
      "learning_rate": 9.72270504149044e-07,
      "loss": 0.6651,
      "step": 3663
    },
    {
      "epoch": 0.33804636143466726,
      "grad_norm": 1.565703912701807,
      "learning_rate": 9.721140583027642e-07,
      "loss": 0.7624,
      "step": 3664
    },
    {
      "epoch": 0.33813862299619424,
      "grad_norm": 2.122687048201629,
      "learning_rate": 9.719575810004622e-07,
      "loss": 0.9355,
      "step": 3665
    },
    {
      "epoch": 0.33813862299619424,
      "eval_GEN Loss": 0.46805840730667114,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.8974358974358975,
      "eval_PRM F1 AUC": 0.8564693556836039,
      "eval_PRM F1 Neg": 0.7142857142857143,
      "eval_PRM Loss": 0.3779523968696594,
      "eval_PRM NPV": 0.6060606060606061,
      "eval_PRM Precision": 0.958904109589041,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.8695652173913043,
      "eval_loss": 0.8263221383094788,
      "eval_runtime": 55.6559,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 3665
    },
    {
      "epoch": 0.33823088455772116,
      "grad_norm": 4.193905375584633,
      "learning_rate": 9.718010722563181e-07,
      "loss": 1.0032,
      "step": 3666
    },
    {
      "epoch": 0.3383231461192481,
      "grad_norm": 2.1995600724755193,
      "learning_rate": 9.716445320845147e-07,
      "loss": 0.7828,
      "step": 3667
    },
    {
      "epoch": 0.338415407680775,
      "grad_norm": 1.7227909694563461,
      "learning_rate": 9.714879604992386e-07,
      "loss": 0.73,
      "step": 3668
    },
    {
      "epoch": 0.3385076692423019,
      "grad_norm": 3.2031034818587494,
      "learning_rate": 9.71331357514678e-07,
      "loss": 0.8766,
      "step": 3669
    },
    {
      "epoch": 0.33859993080382883,
      "grad_norm": 2.2793518104362716,
      "learning_rate": 9.711747231450245e-07,
      "loss": 0.6005,
      "step": 3670
    },
    {
      "epoch": 0.33859993080382883,
      "eval_GEN Loss": 0.46660399436950684,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.35452428460121155,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8137319684028625,
      "eval_runtime": 55.8921,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3670
    },
    {
      "epoch": 0.3386921923653558,
      "grad_norm": 2.450212002364278,
      "learning_rate": 9.710180574044727e-07,
      "loss": 0.9707,
      "step": 3671
    },
    {
      "epoch": 0.3387844539268827,
      "grad_norm": 1.9184393432890179,
      "learning_rate": 9.7086136030722e-07,
      "loss": 0.7585,
      "step": 3672
    },
    {
      "epoch": 0.33887671548840964,
      "grad_norm": 1.5332319234122729,
      "learning_rate": 9.707046318674664e-07,
      "loss": 0.681,
      "step": 3673
    },
    {
      "epoch": 0.33896897704993656,
      "grad_norm": 1.7981134862562713,
      "learning_rate": 9.70547872099415e-07,
      "loss": 0.6154,
      "step": 3674
    },
    {
      "epoch": 0.3390612386114635,
      "grad_norm": 1.3864572001784805,
      "learning_rate": 9.703910810172714e-07,
      "loss": 0.6337,
      "step": 3675
    },
    {
      "epoch": 0.3390612386114635,
      "eval_GEN Loss": 0.4660557210445404,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3518015444278717,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8154447078704834,
      "eval_runtime": 56.1294,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 3675
    },
    {
      "epoch": 0.33915350017299045,
      "grad_norm": 1.8251635328400626,
      "learning_rate": 9.702342586352444e-07,
      "loss": 0.7104,
      "step": 3676
    },
    {
      "epoch": 0.33924576173451737,
      "grad_norm": 1.7706215082444872,
      "learning_rate": 9.700774049675456e-07,
      "loss": 0.7321,
      "step": 3677
    },
    {
      "epoch": 0.3393380232960443,
      "grad_norm": 1.363043612442319,
      "learning_rate": 9.699205200283893e-07,
      "loss": 0.658,
      "step": 3678
    },
    {
      "epoch": 0.3394302848575712,
      "grad_norm": 1.395396675215874,
      "learning_rate": 9.697636038319924e-07,
      "loss": 0.5778,
      "step": 3679
    },
    {
      "epoch": 0.3395225464190981,
      "grad_norm": 1.2753565109487026,
      "learning_rate": 9.69606656392575e-07,
      "loss": 0.5915,
      "step": 3680
    },
    {
      "epoch": 0.3395225464190981,
      "eval_GEN Loss": 0.46370846033096313,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3440261483192444,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8190504908561707,
      "eval_runtime": 56.0384,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 3680
    },
    {
      "epoch": 0.3396148079806251,
      "grad_norm": 1.5388305527427382,
      "learning_rate": 9.6944967772436e-07,
      "loss": 0.6732,
      "step": 3681
    },
    {
      "epoch": 0.339707069542152,
      "grad_norm": 2.5860578343426774,
      "learning_rate": 9.69292667841573e-07,
      "loss": 0.7758,
      "step": 3682
    },
    {
      "epoch": 0.33979933110367894,
      "grad_norm": 2.217120260060514,
      "learning_rate": 9.691356267584426e-07,
      "loss": 0.7643,
      "step": 3683
    },
    {
      "epoch": 0.33989159266520586,
      "grad_norm": 1.6719881907051264,
      "learning_rate": 9.689785544892e-07,
      "loss": 0.6452,
      "step": 3684
    },
    {
      "epoch": 0.3399838542267328,
      "grad_norm": 2.2701623261076125,
      "learning_rate": 9.688214510480792e-07,
      "loss": 0.9379,
      "step": 3685
    },
    {
      "epoch": 0.3399838542267328,
      "eval_GEN Loss": 0.4624996781349182,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3344188332557678,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8329927921295166,
      "eval_runtime": 55.8232,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3685
    },
    {
      "epoch": 0.3400761157882597,
      "grad_norm": 2.1755879254878963,
      "learning_rate": 9.686643164493176e-07,
      "loss": 0.8055,
      "step": 3686
    },
    {
      "epoch": 0.34016837734978667,
      "grad_norm": 2.186986244273936,
      "learning_rate": 9.685071507071546e-07,
      "loss": 0.7312,
      "step": 3687
    },
    {
      "epoch": 0.3402606389113136,
      "grad_norm": 2.4219525196443996,
      "learning_rate": 9.68349953835833e-07,
      "loss": 0.8389,
      "step": 3688
    },
    {
      "epoch": 0.3403529004728405,
      "grad_norm": 1.811163087316862,
      "learning_rate": 9.681927258495978e-07,
      "loss": 0.6604,
      "step": 3689
    },
    {
      "epoch": 0.3404451620343674,
      "grad_norm": 2.2280629579605535,
      "learning_rate": 9.68035466762698e-07,
      "loss": 0.7746,
      "step": 3690
    },
    {
      "epoch": 0.3404451620343674,
      "eval_GEN Loss": 0.4619635045528412,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3435130715370178,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8375601172447205,
      "eval_runtime": 55.8567,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3690
    },
    {
      "epoch": 0.34053742359589434,
      "grad_norm": 1.486801093589663,
      "learning_rate": 9.67878176589384e-07,
      "loss": 0.6528,
      "step": 3691
    },
    {
      "epoch": 0.3406296851574213,
      "grad_norm": 2.02550157026884,
      "learning_rate": 9.677208553439102e-07,
      "loss": 0.8417,
      "step": 3692
    },
    {
      "epoch": 0.34072194671894823,
      "grad_norm": 1.7136683126158871,
      "learning_rate": 9.675635030405328e-07,
      "loss": 0.6631,
      "step": 3693
    },
    {
      "epoch": 0.34081420828047515,
      "grad_norm": 1.8907001519565731,
      "learning_rate": 9.674061196935117e-07,
      "loss": 0.6518,
      "step": 3694
    },
    {
      "epoch": 0.34090646984200207,
      "grad_norm": 1.6914300147416752,
      "learning_rate": 9.67248705317109e-07,
      "loss": 0.6074,
      "step": 3695
    },
    {
      "epoch": 0.34090646984200207,
      "eval_GEN Loss": 0.46258804202079773,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3438650965690613,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8478966355323792,
      "eval_runtime": 56.0644,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 3695
    },
    {
      "epoch": 0.340998731403529,
      "grad_norm": 1.5959064666900813,
      "learning_rate": 9.670912599255897e-07,
      "loss": 0.6259,
      "step": 3696
    },
    {
      "epoch": 0.3410909929650559,
      "grad_norm": 2.026588840209758,
      "learning_rate": 9.669337835332223e-07,
      "loss": 0.5374,
      "step": 3697
    },
    {
      "epoch": 0.3411832545265829,
      "grad_norm": 2.0114116905077544,
      "learning_rate": 9.66776276154277e-07,
      "loss": 0.6961,
      "step": 3698
    },
    {
      "epoch": 0.3412755160881098,
      "grad_norm": 2.0283224825638366,
      "learning_rate": 9.666187378030278e-07,
      "loss": 0.5546,
      "step": 3699
    },
    {
      "epoch": 0.3413677776496367,
      "grad_norm": 2.928884210405768,
      "learning_rate": 9.66461168493751e-07,
      "loss": 0.9221,
      "step": 3700
    },
    {
      "epoch": 0.3413677776496367,
      "eval_GEN Loss": 0.46336492896080017,
      "eval_GEN top-5 accuracy": 0.9850942569048663,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.33952993154525757,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8602163195610046,
      "eval_runtime": 55.9332,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3700
    },
    {
      "epoch": 0.34146003921116364,
      "grad_norm": 2.471443110899123,
      "learning_rate": 9.66303568240725e-07,
      "loss": 0.8619,
      "step": 3701
    },
    {
      "epoch": 0.34155230077269055,
      "grad_norm": 1.1715910962923328,
      "learning_rate": 9.66145937058233e-07,
      "loss": 0.553,
      "step": 3702
    },
    {
      "epoch": 0.3416445623342175,
      "grad_norm": 1.501347001425055,
      "learning_rate": 9.65988274960559e-07,
      "loss": 0.5746,
      "step": 3703
    },
    {
      "epoch": 0.34173682389574445,
      "grad_norm": 1.464144404758925,
      "learning_rate": 9.65830581961991e-07,
      "loss": 0.4682,
      "step": 3704
    },
    {
      "epoch": 0.34182908545727136,
      "grad_norm": 2.9602733712186162,
      "learning_rate": 9.656728580768188e-07,
      "loss": 0.9073,
      "step": 3705
    },
    {
      "epoch": 0.34182908545727136,
      "eval_GEN Loss": 0.4640003442764282,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3416084349155426,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8563100695610046,
      "eval_runtime": 55.9781,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3705
    },
    {
      "epoch": 0.3419213470187983,
      "grad_norm": 1.39742700047925,
      "learning_rate": 9.655151033193362e-07,
      "loss": 0.5994,
      "step": 3706
    },
    {
      "epoch": 0.3420136085803252,
      "grad_norm": 1.3997881250587596,
      "learning_rate": 9.653573177038392e-07,
      "loss": 0.5794,
      "step": 3707
    },
    {
      "epoch": 0.3421058701418522,
      "grad_norm": 1.9537256665259948,
      "learning_rate": 9.65199501244626e-07,
      "loss": 0.7756,
      "step": 3708
    },
    {
      "epoch": 0.3421981317033791,
      "grad_norm": 2.5251039110342335,
      "learning_rate": 9.650416539559987e-07,
      "loss": 0.7615,
      "step": 3709
    },
    {
      "epoch": 0.342290393264906,
      "grad_norm": 1.9751014507249347,
      "learning_rate": 9.648837758522614e-07,
      "loss": 0.5872,
      "step": 3710
    },
    {
      "epoch": 0.342290393264906,
      "eval_GEN Loss": 0.46506547927856445,
      "eval_GEN top-5 accuracy": 0.9842174484875055,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.36560991406440735,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8573317527770996,
      "eval_runtime": 56.0097,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3710
    },
    {
      "epoch": 0.34238265482643293,
      "grad_norm": 2.274600642817432,
      "learning_rate": 9.647258669477213e-07,
      "loss": 0.8695,
      "step": 3711
    },
    {
      "epoch": 0.34247491638795985,
      "grad_norm": 1.91019450167604,
      "learning_rate": 9.645679272566883e-07,
      "loss": 0.7842,
      "step": 3712
    },
    {
      "epoch": 0.34256717794948677,
      "grad_norm": 1.5952385364835358,
      "learning_rate": 9.644099567934756e-07,
      "loss": 0.8352,
      "step": 3713
    },
    {
      "epoch": 0.34265943951101374,
      "grad_norm": 2.710953429833622,
      "learning_rate": 9.64251955572398e-07,
      "loss": 0.6237,
      "step": 3714
    },
    {
      "epoch": 0.34275170107254066,
      "grad_norm": 1.366794952745796,
      "learning_rate": 9.640939236077742e-07,
      "loss": 0.7175,
      "step": 3715
    },
    {
      "epoch": 0.34275170107254066,
      "eval_GEN Loss": 0.46673673391342163,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3873176872730255,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8762019276618958,
      "eval_runtime": 56.7214,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3715
    },
    {
      "epoch": 0.3428439626340676,
      "grad_norm": 3.2775756515553356,
      "learning_rate": 9.639358609139257e-07,
      "loss": 0.8002,
      "step": 3716
    },
    {
      "epoch": 0.3429362241955945,
      "grad_norm": 2.67874937881373,
      "learning_rate": 9.637777675051757e-07,
      "loss": 0.9112,
      "step": 3717
    },
    {
      "epoch": 0.3430284857571214,
      "grad_norm": 2.214272219994824,
      "learning_rate": 9.636196433958515e-07,
      "loss": 0.7867,
      "step": 3718
    },
    {
      "epoch": 0.3431207473186484,
      "grad_norm": 1.9175322900304457,
      "learning_rate": 9.634614886002817e-07,
      "loss": 0.8526,
      "step": 3719
    },
    {
      "epoch": 0.3432130088801753,
      "grad_norm": 1.9718312198041474,
      "learning_rate": 9.633033031327993e-07,
      "loss": 0.7541,
      "step": 3720
    },
    {
      "epoch": 0.3432130088801753,
      "eval_GEN Loss": 0.46820056438446045,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.4053679406642914,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.885817289352417,
      "eval_runtime": 55.9795,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3720
    },
    {
      "epoch": 0.3433052704417022,
      "grad_norm": 1.6607748272890903,
      "learning_rate": 9.63145087007739e-07,
      "loss": 0.6884,
      "step": 3721
    },
    {
      "epoch": 0.34339753200322914,
      "grad_norm": 1.8858171279161577,
      "learning_rate": 9.629868402394387e-07,
      "loss": 0.7745,
      "step": 3722
    },
    {
      "epoch": 0.34348979356475606,
      "grad_norm": 1.7962143787665013,
      "learning_rate": 9.62828562842239e-07,
      "loss": 0.9024,
      "step": 3723
    },
    {
      "epoch": 0.34358205512628304,
      "grad_norm": 2.544460439921917,
      "learning_rate": 9.626702548304832e-07,
      "loss": 0.7605,
      "step": 3724
    },
    {
      "epoch": 0.34367431668780996,
      "grad_norm": 2.881148068779642,
      "learning_rate": 9.625119162185173e-07,
      "loss": 0.7746,
      "step": 3725
    },
    {
      "epoch": 0.34367431668780996,
      "eval_GEN Loss": 0.4682743549346924,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.41063398122787476,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8886418342590332,
      "eval_runtime": 55.8897,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3725
    },
    {
      "epoch": 0.3437665782493369,
      "grad_norm": 2.211281306245446,
      "learning_rate": 9.623535470206905e-07,
      "loss": 0.905,
      "step": 3726
    },
    {
      "epoch": 0.3438588398108638,
      "grad_norm": 1.6261235331180999,
      "learning_rate": 9.62195147251354e-07,
      "loss": 0.658,
      "step": 3727
    },
    {
      "epoch": 0.3439511013723907,
      "grad_norm": 1.3763894325034005,
      "learning_rate": 9.620367169248628e-07,
      "loss": 0.5018,
      "step": 3728
    },
    {
      "epoch": 0.34404336293391763,
      "grad_norm": 1.6154059462172627,
      "learning_rate": 9.618782560555736e-07,
      "loss": 0.6924,
      "step": 3729
    },
    {
      "epoch": 0.3441356244954446,
      "grad_norm": 2.6467165804514696,
      "learning_rate": 9.617197646578466e-07,
      "loss": 0.7522,
      "step": 3730
    },
    {
      "epoch": 0.3441356244954446,
      "eval_GEN Loss": 0.468242347240448,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.4047353267669678,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8870792984962463,
      "eval_runtime": 55.8541,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3730
    },
    {
      "epoch": 0.3442278860569715,
      "grad_norm": 1.593684201672445,
      "learning_rate": 9.615612427460444e-07,
      "loss": 0.5861,
      "step": 3731
    },
    {
      "epoch": 0.34432014761849844,
      "grad_norm": 2.1543309100245502,
      "learning_rate": 9.61402690334533e-07,
      "loss": 0.7387,
      "step": 3732
    },
    {
      "epoch": 0.34441240918002536,
      "grad_norm": 1.4882698916355077,
      "learning_rate": 9.6124410743768e-07,
      "loss": 0.6851,
      "step": 3733
    },
    {
      "epoch": 0.3445046707415523,
      "grad_norm": 1.8572118230979693,
      "learning_rate": 9.610854940698568e-07,
      "loss": 0.5976,
      "step": 3734
    },
    {
      "epoch": 0.34459693230307925,
      "grad_norm": 1.6671442899708826,
      "learning_rate": 9.609268502454373e-07,
      "loss": 0.6387,
      "step": 3735
    },
    {
      "epoch": 0.34459693230307925,
      "eval_GEN Loss": 0.46616464853286743,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3937951624393463,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8822716474533081,
      "eval_runtime": 55.7213,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 3735
    },
    {
      "epoch": 0.34468919386460617,
      "grad_norm": 2.0680544806719037,
      "learning_rate": 9.607681759787976e-07,
      "loss": 0.5404,
      "step": 3736
    },
    {
      "epoch": 0.3447814554261331,
      "grad_norm": 1.772288908488974,
      "learning_rate": 9.606094712843173e-07,
      "loss": 0.7036,
      "step": 3737
    },
    {
      "epoch": 0.34487371698766,
      "grad_norm": 1.9102702479273743,
      "learning_rate": 9.604507361763787e-07,
      "loss": 0.6893,
      "step": 3738
    },
    {
      "epoch": 0.3449659785491869,
      "grad_norm": 1.7583148862631657,
      "learning_rate": 9.602919706693661e-07,
      "loss": 0.6487,
      "step": 3739
    },
    {
      "epoch": 0.3450582401107139,
      "grad_norm": 1.2407806825683938,
      "learning_rate": 9.601331747776673e-07,
      "loss": 0.5756,
      "step": 3740
    },
    {
      "epoch": 0.3450582401107139,
      "eval_GEN Loss": 0.4655279517173767,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3840682804584503,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8846153616905212,
      "eval_runtime": 55.8923,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3740
    },
    {
      "epoch": 0.3451505016722408,
      "grad_norm": 1.845289086473611,
      "learning_rate": 9.599743485156733e-07,
      "loss": 0.7607,
      "step": 3741
    },
    {
      "epoch": 0.34524276323376774,
      "grad_norm": 1.974588145971051,
      "learning_rate": 9.598154918977758e-07,
      "loss": 0.6877,
      "step": 3742
    },
    {
      "epoch": 0.34533502479529465,
      "grad_norm": 2.734223985194787,
      "learning_rate": 9.59656604938372e-07,
      "loss": 0.7999,
      "step": 3743
    },
    {
      "epoch": 0.3454272863568216,
      "grad_norm": 1.2711755505902111,
      "learning_rate": 9.594976876518595e-07,
      "loss": 0.661,
      "step": 3744
    },
    {
      "epoch": 0.3455195479183485,
      "grad_norm": 1.3860884880801732,
      "learning_rate": 9.593387400526401e-07,
      "loss": 0.5841,
      "step": 3745
    },
    {
      "epoch": 0.3455195479183485,
      "eval_GEN Loss": 0.46540525555610657,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3680097758769989,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8952524065971375,
      "eval_runtime": 55.8634,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3745
    },
    {
      "epoch": 0.34561180947987546,
      "grad_norm": 2.0360075972208596,
      "learning_rate": 9.59179762155118e-07,
      "loss": 0.9218,
      "step": 3746
    },
    {
      "epoch": 0.3457040710414024,
      "grad_norm": 1.9912416761743457,
      "learning_rate": 9.590207539736998e-07,
      "loss": 0.6224,
      "step": 3747
    },
    {
      "epoch": 0.3457963326029293,
      "grad_norm": 1.410263018673505,
      "learning_rate": 9.58861715522795e-07,
      "loss": 0.6918,
      "step": 3748
    },
    {
      "epoch": 0.3458885941644562,
      "grad_norm": 3.480801931827441,
      "learning_rate": 9.587026468168164e-07,
      "loss": 0.8224,
      "step": 3749
    },
    {
      "epoch": 0.34598085572598314,
      "grad_norm": 3.476373522330977,
      "learning_rate": 9.585435478701784e-07,
      "loss": 0.8239,
      "step": 3750
    },
    {
      "epoch": 0.34598085572598314,
      "eval_GEN Loss": 0.4648699164390564,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3654617369174957,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9065504670143127,
      "eval_runtime": 55.9693,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3750
    },
    {
      "epoch": 0.3460731172875101,
      "grad_norm": 2.2131471786098253,
      "learning_rate": 9.58384418697299e-07,
      "loss": 0.5293,
      "step": 3751
    },
    {
      "epoch": 0.34616537884903703,
      "grad_norm": 2.211190767403979,
      "learning_rate": 9.58225259312599e-07,
      "loss": 0.8615,
      "step": 3752
    },
    {
      "epoch": 0.34625764041056395,
      "grad_norm": 2.277051565992359,
      "learning_rate": 9.580660697305014e-07,
      "loss": 0.6596,
      "step": 3753
    },
    {
      "epoch": 0.34634990197209087,
      "grad_norm": 1.7825016885152825,
      "learning_rate": 9.579068499654324e-07,
      "loss": 0.6554,
      "step": 3754
    },
    {
      "epoch": 0.3464421635336178,
      "grad_norm": 1.5833402800799643,
      "learning_rate": 9.577476000318205e-07,
      "loss": 0.4765,
      "step": 3755
    },
    {
      "epoch": 0.3464421635336178,
      "eval_GEN Loss": 0.46510693430900574,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3725627660751343,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8920673131942749,
      "eval_runtime": 55.8662,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3755
    },
    {
      "epoch": 0.34653442509514476,
      "grad_norm": 1.7563207722624166,
      "learning_rate": 9.575883199440972e-07,
      "loss": 0.4881,
      "step": 3756
    },
    {
      "epoch": 0.3466266866566717,
      "grad_norm": 2.880408204576627,
      "learning_rate": 9.574290097166973e-07,
      "loss": 0.8167,
      "step": 3757
    },
    {
      "epoch": 0.3467189482181986,
      "grad_norm": 1.4951138351055484,
      "learning_rate": 9.57269669364057e-07,
      "loss": 0.7277,
      "step": 3758
    },
    {
      "epoch": 0.3468112097797255,
      "grad_norm": 2.755994265752733,
      "learning_rate": 9.571102989006162e-07,
      "loss": 0.8306,
      "step": 3759
    },
    {
      "epoch": 0.34690347134125243,
      "grad_norm": 3.5789846195289416,
      "learning_rate": 9.569508983408176e-07,
      "loss": 0.9367,
      "step": 3760
    },
    {
      "epoch": 0.34690347134125243,
      "eval_GEN Loss": 0.46539977192878723,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.39090612530708313,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8852163553237915,
      "eval_runtime": 55.9406,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3760
    },
    {
      "epoch": 0.34699573290277935,
      "grad_norm": 2.4155404325311904,
      "learning_rate": 9.56791467699106e-07,
      "loss": 0.6552,
      "step": 3761
    },
    {
      "epoch": 0.3470879944643063,
      "grad_norm": 1.5826400078873153,
      "learning_rate": 9.566320069899292e-07,
      "loss": 0.6167,
      "step": 3762
    },
    {
      "epoch": 0.34718025602583324,
      "grad_norm": 2.280185495672841,
      "learning_rate": 9.56472516227738e-07,
      "loss": 0.7432,
      "step": 3763
    },
    {
      "epoch": 0.34727251758736016,
      "grad_norm": 2.279201341934873,
      "learning_rate": 9.56312995426986e-07,
      "loss": 0.7819,
      "step": 3764
    },
    {
      "epoch": 0.3473647791488871,
      "grad_norm": 2.6630987225038747,
      "learning_rate": 9.561534446021283e-07,
      "loss": 0.9754,
      "step": 3765
    },
    {
      "epoch": 0.3473647791488871,
      "eval_GEN Loss": 0.4660956561565399,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.40761876106262207,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8894831538200378,
      "eval_runtime": 55.9735,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3765
    },
    {
      "epoch": 0.347457040710414,
      "grad_norm": 1.993435733786602,
      "learning_rate": 9.559938637676246e-07,
      "loss": 0.6586,
      "step": 3766
    },
    {
      "epoch": 0.347549302271941,
      "grad_norm": 2.533630176587448,
      "learning_rate": 9.558342529379359e-07,
      "loss": 0.8528,
      "step": 3767
    },
    {
      "epoch": 0.3476415638334679,
      "grad_norm": 2.5306174761243816,
      "learning_rate": 9.556746121275261e-07,
      "loss": 0.5803,
      "step": 3768
    },
    {
      "epoch": 0.3477338253949948,
      "grad_norm": 1.3563241511337962,
      "learning_rate": 9.555149413508627e-07,
      "loss": 0.6045,
      "step": 3769
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 1.3482186086924404,
      "learning_rate": 9.553552406224153e-07,
      "loss": 0.6691,
      "step": 3770
    },
    {
      "epoch": 0.34782608695652173,
      "eval_GEN Loss": 0.4648943245410919,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.41272056102752686,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8916466236114502,
      "eval_runtime": 56.9343,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3770
    },
    {
      "epoch": 0.34791834851804865,
      "grad_norm": 2.0529576917360686,
      "learning_rate": 9.551955099566558e-07,
      "loss": 0.7308,
      "step": 3771
    },
    {
      "epoch": 0.3480106100795756,
      "grad_norm": 2.4063712600360883,
      "learning_rate": 9.550357493680595e-07,
      "loss": 0.6653,
      "step": 3772
    },
    {
      "epoch": 0.34810287164110254,
      "grad_norm": 2.5305538893807475,
      "learning_rate": 9.548759588711041e-07,
      "loss": 0.7102,
      "step": 3773
    },
    {
      "epoch": 0.34819513320262946,
      "grad_norm": 2.1691344668854846,
      "learning_rate": 9.5471613848027e-07,
      "loss": 0.6277,
      "step": 3774
    },
    {
      "epoch": 0.3482873947641564,
      "grad_norm": 1.6769005741622578,
      "learning_rate": 9.545562882100406e-07,
      "loss": 0.7539,
      "step": 3775
    },
    {
      "epoch": 0.3482873947641564,
      "eval_GEN Loss": 0.46490785479545593,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3942284882068634,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8865384459495544,
      "eval_runtime": 55.8577,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3775
    },
    {
      "epoch": 0.3483796563256833,
      "grad_norm": 2.0073840928156366,
      "learning_rate": 9.543964080749017e-07,
      "loss": 0.7097,
      "step": 3776
    },
    {
      "epoch": 0.3484719178872102,
      "grad_norm": 2.1929671167175124,
      "learning_rate": 9.542364980893414e-07,
      "loss": 0.6984,
      "step": 3777
    },
    {
      "epoch": 0.3485641794487372,
      "grad_norm": 3.6134093049420746,
      "learning_rate": 9.54076558267852e-07,
      "loss": 0.9162,
      "step": 3778
    },
    {
      "epoch": 0.3486564410102641,
      "grad_norm": 2.037670869481623,
      "learning_rate": 9.539165886249262e-07,
      "loss": 0.7101,
      "step": 3779
    },
    {
      "epoch": 0.348748702571791,
      "grad_norm": 2.2242157983921924,
      "learning_rate": 9.537565891750619e-07,
      "loss": 0.6847,
      "step": 3780
    },
    {
      "epoch": 0.348748702571791,
      "eval_GEN Loss": 0.46401163935661316,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.39052385091781616,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8843749761581421,
      "eval_runtime": 55.8697,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3780
    },
    {
      "epoch": 0.34884096413331794,
      "grad_norm": 1.9904305979917003,
      "learning_rate": 9.53596559932758e-07,
      "loss": 0.7894,
      "step": 3781
    },
    {
      "epoch": 0.34893322569484486,
      "grad_norm": 1.9684031179267463,
      "learning_rate": 9.534365009125165e-07,
      "loss": 0.4988,
      "step": 3782
    },
    {
      "epoch": 0.34902548725637184,
      "grad_norm": 2.3661066800252395,
      "learning_rate": 9.532764121288423e-07,
      "loss": 0.6426,
      "step": 3783
    },
    {
      "epoch": 0.34911774881789875,
      "grad_norm": 1.990318659947277,
      "learning_rate": 9.531162935962432e-07,
      "loss": 0.7695,
      "step": 3784
    },
    {
      "epoch": 0.3492100103794257,
      "grad_norm": 1.9400922917227312,
      "learning_rate": 9.529561453292288e-07,
      "loss": 0.712,
      "step": 3785
    },
    {
      "epoch": 0.3492100103794257,
      "eval_GEN Loss": 0.4638323485851288,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3923055827617645,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8871995210647583,
      "eval_runtime": 55.7818,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 3785
    },
    {
      "epoch": 0.3493022719409526,
      "grad_norm": 1.6983564404606992,
      "learning_rate": 9.527959673423126e-07,
      "loss": 0.7138,
      "step": 3786
    },
    {
      "epoch": 0.3493945335024795,
      "grad_norm": 1.7608052182230653,
      "learning_rate": 9.526357596500097e-07,
      "loss": 0.8493,
      "step": 3787
    },
    {
      "epoch": 0.3494867950640065,
      "grad_norm": 2.095177570576328,
      "learning_rate": 9.524755222668383e-07,
      "loss": 0.5445,
      "step": 3788
    },
    {
      "epoch": 0.3495790566255334,
      "grad_norm": 2.1300396783995064,
      "learning_rate": 9.5231525520732e-07,
      "loss": 0.7375,
      "step": 3789
    },
    {
      "epoch": 0.3496713181870603,
      "grad_norm": 2.076996752497041,
      "learning_rate": 9.521549584859779e-07,
      "loss": 0.7043,
      "step": 3790
    },
    {
      "epoch": 0.3496713181870603,
      "eval_GEN Loss": 0.46403998136520386,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.40035301446914673,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8885817527770996,
      "eval_runtime": 55.9117,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3790
    },
    {
      "epoch": 0.34976357974858724,
      "grad_norm": 2.721758709724885,
      "learning_rate": 9.519946321173385e-07,
      "loss": 0.9223,
      "step": 3791
    },
    {
      "epoch": 0.34985584131011416,
      "grad_norm": 1.5456915103968132,
      "learning_rate": 9.518342761159307e-07,
      "loss": 0.6306,
      "step": 3792
    },
    {
      "epoch": 0.3499481028716411,
      "grad_norm": 1.9476559346686992,
      "learning_rate": 9.516738904962867e-07,
      "loss": 0.622,
      "step": 3793
    },
    {
      "epoch": 0.35004036443316805,
      "grad_norm": 1.8450391455544923,
      "learning_rate": 9.515134752729405e-07,
      "loss": 0.8644,
      "step": 3794
    },
    {
      "epoch": 0.35013262599469497,
      "grad_norm": 2.0371267793957806,
      "learning_rate": 9.51353030460429e-07,
      "loss": 0.7757,
      "step": 3795
    },
    {
      "epoch": 0.35013262599469497,
      "eval_GEN Loss": 0.46473821997642517,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8625,
      "eval_PRM F1 AUC": 0.7417496071241488,
      "eval_PRM F1 Neg": 0.5769230769230769,
      "eval_PRM Loss": 0.39523544907569885,
      "eval_PRM NPV": 0.5172413793103449,
      "eval_PRM Precision": 0.8961038961038961,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8787259459495544,
      "eval_runtime": 56.8412,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3795
    },
    {
      "epoch": 0.3502248875562219,
      "grad_norm": 2.100192006301843,
      "learning_rate": 9.511925560732924e-07,
      "loss": 0.6447,
      "step": 3796
    },
    {
      "epoch": 0.3503171491177488,
      "grad_norm": 1.4656661365226142,
      "learning_rate": 9.510320521260728e-07,
      "loss": 0.6536,
      "step": 3797
    },
    {
      "epoch": 0.3504094106792757,
      "grad_norm": 1.9324349060004573,
      "learning_rate": 9.508715186333156e-07,
      "loss": 0.8792,
      "step": 3798
    },
    {
      "epoch": 0.3505016722408027,
      "grad_norm": 1.9929081691027095,
      "learning_rate": 9.507109556095682e-07,
      "loss": 0.7737,
      "step": 3799
    },
    {
      "epoch": 0.3505939338023296,
      "grad_norm": 2.4839335273626837,
      "learning_rate": 9.505503630693816e-07,
      "loss": 0.9378,
      "step": 3800
    },
    {
      "epoch": 0.3505939338023296,
      "eval_GEN Loss": 0.4664175510406494,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8553459119496856,
      "eval_PRM F1 AUC": 0.7357255107386066,
      "eval_PRM F1 Neg": 0.5660377358490566,
      "eval_PRM Loss": 0.4022744596004486,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.8947368421052632,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8781850934028625,
      "eval_runtime": 56.1148,
      "eval_samples_per_second": 1.158,
      "eval_steps_per_second": 0.089,
      "step": 3800
    },
    {
      "epoch": 0.35068619536385653,
      "grad_norm": 2.151126844136183,
      "learning_rate": 9.503897410273085e-07,
      "loss": 0.7538,
      "step": 3801
    },
    {
      "epoch": 0.35077845692538345,
      "grad_norm": 1.7029214297304305,
      "learning_rate": 9.502290894979049e-07,
      "loss": 0.7354,
      "step": 3802
    },
    {
      "epoch": 0.35087071848691037,
      "grad_norm": 3.72531159054256,
      "learning_rate": 9.500684084957293e-07,
      "loss": 0.8023,
      "step": 3803
    },
    {
      "epoch": 0.35096298004843735,
      "grad_norm": 2.883998651974907,
      "learning_rate": 9.49907698035343e-07,
      "loss": 0.8637,
      "step": 3804
    },
    {
      "epoch": 0.35105524160996426,
      "grad_norm": 2.5533777808372102,
      "learning_rate": 9.497469581313096e-07,
      "loss": 0.6717,
      "step": 3805
    },
    {
      "epoch": 0.35105524160996426,
      "eval_GEN Loss": 0.4656904339790344,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8535031847133758,
      "eval_PRM F1 AUC": 0.751440544787847,
      "eval_PRM F1 Neg": 0.5818181818181818,
      "eval_PRM Loss": 0.410686194896698,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9054054054054054,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8751803040504456,
      "eval_runtime": 55.9079,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3805
    },
    {
      "epoch": 0.3511475031714912,
      "grad_norm": 1.6926813226920387,
      "learning_rate": 9.495861887981955e-07,
      "loss": 0.8087,
      "step": 3806
    },
    {
      "epoch": 0.3512397647330181,
      "grad_norm": 1.2478912525061865,
      "learning_rate": 9.494253900505704e-07,
      "loss": 0.5283,
      "step": 3807
    },
    {
      "epoch": 0.351332026294545,
      "grad_norm": 1.8740959615924406,
      "learning_rate": 9.492645619030057e-07,
      "loss": 0.642,
      "step": 3808
    },
    {
      "epoch": 0.35142428785607194,
      "grad_norm": 2.04318850754419,
      "learning_rate": 9.491037043700759e-07,
      "loss": 0.5957,
      "step": 3809
    },
    {
      "epoch": 0.3515165494175989,
      "grad_norm": 1.899982599882739,
      "learning_rate": 9.489428174663584e-07,
      "loss": 0.7386,
      "step": 3810
    },
    {
      "epoch": 0.3515165494175989,
      "eval_GEN Loss": 0.4658967852592468,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4093562066555023,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8658654093742371,
      "eval_runtime": 56.707,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3810
    },
    {
      "epoch": 0.35160881097912583,
      "grad_norm": 2.5564895762346453,
      "learning_rate": 9.487819012064326e-07,
      "loss": 0.9171,
      "step": 3811
    },
    {
      "epoch": 0.35170107254065275,
      "grad_norm": 2.1461505395937053,
      "learning_rate": 9.486209556048816e-07,
      "loss": 0.6051,
      "step": 3812
    },
    {
      "epoch": 0.35179333410217967,
      "grad_norm": 1.9175809793688028,
      "learning_rate": 9.4845998067629e-07,
      "loss": 0.7453,
      "step": 3813
    },
    {
      "epoch": 0.3518855956637066,
      "grad_norm": 1.7937262195464165,
      "learning_rate": 9.48298976435246e-07,
      "loss": 0.7153,
      "step": 3814
    },
    {
      "epoch": 0.35197785722523356,
      "grad_norm": 1.6516314850301779,
      "learning_rate": 9.481379428963397e-07,
      "loss": 0.7662,
      "step": 3815
    },
    {
      "epoch": 0.35197785722523356,
      "eval_GEN Loss": 0.463291198015213,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.38919463753700256,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8502404093742371,
      "eval_runtime": 56.6495,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3815
    },
    {
      "epoch": 0.3520701187867605,
      "grad_norm": 1.6259084608320094,
      "learning_rate": 9.479768800741646e-07,
      "loss": 0.84,
      "step": 3816
    },
    {
      "epoch": 0.3521623803482874,
      "grad_norm": 2.452537410992221,
      "learning_rate": 9.47815787983316e-07,
      "loss": 0.8509,
      "step": 3817
    },
    {
      "epoch": 0.3522546419098143,
      "grad_norm": 2.069276426972907,
      "learning_rate": 9.476546666383928e-07,
      "loss": 0.6947,
      "step": 3818
    },
    {
      "epoch": 0.35234690347134123,
      "grad_norm": 2.4313907613121177,
      "learning_rate": 9.474935160539955e-07,
      "loss": 0.9207,
      "step": 3819
    },
    {
      "epoch": 0.3524391650328682,
      "grad_norm": 2.635050207075767,
      "learning_rate": 9.473323362447286e-07,
      "loss": 0.7986,
      "step": 3820
    },
    {
      "epoch": 0.3524391650328682,
      "eval_GEN Loss": 0.46222686767578125,
      "eval_GEN top-5 accuracy": 0.9837790442788251,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.37923020124435425,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8461538553237915,
      "eval_runtime": 55.8756,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3820
    },
    {
      "epoch": 0.3525314265943951,
      "grad_norm": 1.8978964523398498,
      "learning_rate": 9.47171127225198e-07,
      "loss": 0.8523,
      "step": 3821
    },
    {
      "epoch": 0.35262368815592204,
      "grad_norm": 1.614683666230754,
      "learning_rate": 9.470098890100124e-07,
      "loss": 0.6918,
      "step": 3822
    },
    {
      "epoch": 0.35271594971744896,
      "grad_norm": 1.6858822643243787,
      "learning_rate": 9.46848621613784e-07,
      "loss": 0.8118,
      "step": 3823
    },
    {
      "epoch": 0.3528082112789759,
      "grad_norm": 1.3720323494444129,
      "learning_rate": 9.466873250511269e-07,
      "loss": 0.5211,
      "step": 3824
    },
    {
      "epoch": 0.3529004728405028,
      "grad_norm": 3.001293814190167,
      "learning_rate": 9.46525999336658e-07,
      "loss": 0.8726,
      "step": 3825
    },
    {
      "epoch": 0.3529004728405028,
      "eval_GEN Loss": 0.4613455533981323,
      "eval_GEN top-5 accuracy": 0.9846558526961858,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3843279182910919,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8443509340286255,
      "eval_runtime": 55.8878,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3825
    },
    {
      "epoch": 0.3529927344020298,
      "grad_norm": 1.6061883282502658,
      "learning_rate": 9.463646444849971e-07,
      "loss": 0.6558,
      "step": 3826
    },
    {
      "epoch": 0.3530849959635567,
      "grad_norm": 2.499435859612238,
      "learning_rate": 9.46203260510766e-07,
      "loss": 0.6363,
      "step": 3827
    },
    {
      "epoch": 0.3531772575250836,
      "grad_norm": 2.466810532714648,
      "learning_rate": 9.460418474285901e-07,
      "loss": 0.7952,
      "step": 3828
    },
    {
      "epoch": 0.35326951908661053,
      "grad_norm": 1.7470667316172535,
      "learning_rate": 9.458804052530966e-07,
      "loss": 0.7722,
      "step": 3829
    },
    {
      "epoch": 0.35336178064813745,
      "grad_norm": 1.4695160320235972,
      "learning_rate": 9.457189339989158e-07,
      "loss": 0.6944,
      "step": 3830
    },
    {
      "epoch": 0.35336178064813745,
      "eval_GEN Loss": 0.461611270904541,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.39079543948173523,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8508413434028625,
      "eval_runtime": 56.066,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 3830
    },
    {
      "epoch": 0.3534540422096644,
      "grad_norm": 1.5954380009008133,
      "learning_rate": 9.455574336806803e-07,
      "loss": 0.7239,
      "step": 3831
    },
    {
      "epoch": 0.35354630377119134,
      "grad_norm": 2.5123736388523277,
      "learning_rate": 9.453959043130255e-07,
      "loss": 0.7351,
      "step": 3832
    },
    {
      "epoch": 0.35363856533271826,
      "grad_norm": 1.332911708200968,
      "learning_rate": 9.452343459105898e-07,
      "loss": 0.5641,
      "step": 3833
    },
    {
      "epoch": 0.3537308268942452,
      "grad_norm": 1.4504415008958897,
      "learning_rate": 9.450727584880134e-07,
      "loss": 0.7043,
      "step": 3834
    },
    {
      "epoch": 0.3538230884557721,
      "grad_norm": 1.3033330344002674,
      "learning_rate": 9.449111420599398e-07,
      "loss": 0.6206,
      "step": 3835
    },
    {
      "epoch": 0.3538230884557721,
      "eval_GEN Loss": 0.4629392623901367,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.38266974687576294,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8508413434028625,
      "eval_runtime": 55.9445,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3835
    },
    {
      "epoch": 0.35391535001729907,
      "grad_norm": 2.173836572408765,
      "learning_rate": 9.447494966410151e-07,
      "loss": 0.7043,
      "step": 3836
    },
    {
      "epoch": 0.354007611578826,
      "grad_norm": 1.9238904080345363,
      "learning_rate": 9.445878222458876e-07,
      "loss": 0.5655,
      "step": 3837
    },
    {
      "epoch": 0.3540998731403529,
      "grad_norm": 2.7406387439384097,
      "learning_rate": 9.444261188892086e-07,
      "loss": 0.7306,
      "step": 3838
    },
    {
      "epoch": 0.3541921347018798,
      "grad_norm": 1.7635468735584554,
      "learning_rate": 9.44264386585632e-07,
      "loss": 0.7511,
      "step": 3839
    },
    {
      "epoch": 0.35428439626340674,
      "grad_norm": 1.9160124904326394,
      "learning_rate": 9.44102625349814e-07,
      "loss": 0.7338,
      "step": 3840
    },
    {
      "epoch": 0.35428439626340674,
      "eval_GEN Loss": 0.461733341217041,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.36684173345565796,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8509014248847961,
      "eval_runtime": 56.9064,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3840
    },
    {
      "epoch": 0.35437665782493366,
      "grad_norm": 1.8609823494273454,
      "learning_rate": 9.439408351964141e-07,
      "loss": 0.8417,
      "step": 3841
    },
    {
      "epoch": 0.35446891938646063,
      "grad_norm": 1.815245733887611,
      "learning_rate": 9.437790161400934e-07,
      "loss": 0.6562,
      "step": 3842
    },
    {
      "epoch": 0.35456118094798755,
      "grad_norm": 2.025458500947324,
      "learning_rate": 9.436171681955167e-07,
      "loss": 0.674,
      "step": 3843
    },
    {
      "epoch": 0.35465344250951447,
      "grad_norm": 2.2212581093030765,
      "learning_rate": 9.434552913773506e-07,
      "loss": 0.784,
      "step": 3844
    },
    {
      "epoch": 0.3547457040710414,
      "grad_norm": 2.529955778467214,
      "learning_rate": 9.432933857002649e-07,
      "loss": 0.8232,
      "step": 3845
    },
    {
      "epoch": 0.3547457040710414,
      "eval_GEN Loss": 0.46213412284851074,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3619709312915802,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8529447317123413,
      "eval_runtime": 56.8045,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3845
    },
    {
      "epoch": 0.3548379656325683,
      "grad_norm": 2.8878755615180696,
      "learning_rate": 9.431314511789314e-07,
      "loss": 0.7746,
      "step": 3846
    },
    {
      "epoch": 0.3549302271940953,
      "grad_norm": 1.9593453625378463,
      "learning_rate": 9.429694878280252e-07,
      "loss": 0.7861,
      "step": 3847
    },
    {
      "epoch": 0.3550224887556222,
      "grad_norm": 1.7402377194876628,
      "learning_rate": 9.428074956622234e-07,
      "loss": 0.7562,
      "step": 3848
    },
    {
      "epoch": 0.3551147503171491,
      "grad_norm": 2.8108155560139965,
      "learning_rate": 9.42645474696206e-07,
      "loss": 0.7331,
      "step": 3849
    },
    {
      "epoch": 0.35520701187867604,
      "grad_norm": 1.953666290338613,
      "learning_rate": 9.42483424944656e-07,
      "loss": 0.6774,
      "step": 3850
    },
    {
      "epoch": 0.35520701187867604,
      "eval_GEN Loss": 0.46292129158973694,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3741525709629059,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8564302921295166,
      "eval_runtime": 55.8635,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3850
    },
    {
      "epoch": 0.35529927344020296,
      "grad_norm": 3.6255077078826496,
      "learning_rate": 9.42321346422258e-07,
      "loss": 0.8206,
      "step": 3851
    },
    {
      "epoch": 0.35539153500172993,
      "grad_norm": 1.6079824700403849,
      "learning_rate": 9.421592391437002e-07,
      "loss": 0.6385,
      "step": 3852
    },
    {
      "epoch": 0.35548379656325685,
      "grad_norm": 2.1655772457246814,
      "learning_rate": 9.419971031236731e-07,
      "loss": 0.9775,
      "step": 3853
    },
    {
      "epoch": 0.35557605812478377,
      "grad_norm": 1.8286955908035634,
      "learning_rate": 9.418349383768693e-07,
      "loss": 0.8645,
      "step": 3854
    },
    {
      "epoch": 0.3556683196863107,
      "grad_norm": 1.6657469796828948,
      "learning_rate": 9.416727449179849e-07,
      "loss": 0.5622,
      "step": 3855
    },
    {
      "epoch": 0.3556683196863107,
      "eval_GEN Loss": 0.4636772572994232,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3858485221862793,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8675480484962463,
      "eval_runtime": 55.8769,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3855
    },
    {
      "epoch": 0.3557605812478376,
      "grad_norm": 2.1092599758930834,
      "learning_rate": 9.41510522761718e-07,
      "loss": 0.9397,
      "step": 3856
    },
    {
      "epoch": 0.3558528428093645,
      "grad_norm": 1.6385850685683812,
      "learning_rate": 9.413482719227693e-07,
      "loss": 0.7014,
      "step": 3857
    },
    {
      "epoch": 0.3559451043708915,
      "grad_norm": 2.1391797324635426,
      "learning_rate": 9.41185992415842e-07,
      "loss": 0.7022,
      "step": 3858
    },
    {
      "epoch": 0.3560373659324184,
      "grad_norm": 1.6717476091751462,
      "learning_rate": 9.410236842556425e-07,
      "loss": 0.6209,
      "step": 3859
    },
    {
      "epoch": 0.35612962749394533,
      "grad_norm": 1.737214758222355,
      "learning_rate": 9.408613474568792e-07,
      "loss": 0.6674,
      "step": 3860
    },
    {
      "epoch": 0.35612962749394533,
      "eval_GEN Loss": 0.46808555722236633,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.39053481817245483,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8746995329856873,
      "eval_runtime": 55.8204,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 3860
    },
    {
      "epoch": 0.35622188905547225,
      "grad_norm": 1.634835933116288,
      "learning_rate": 9.406989820342635e-07,
      "loss": 0.7859,
      "step": 3861
    },
    {
      "epoch": 0.35631415061699917,
      "grad_norm": 2.1055771986113143,
      "learning_rate": 9.405365880025094e-07,
      "loss": 0.6726,
      "step": 3862
    },
    {
      "epoch": 0.35640641217852614,
      "grad_norm": 2.431314053750086,
      "learning_rate": 9.403741653763326e-07,
      "loss": 0.7384,
      "step": 3863
    },
    {
      "epoch": 0.35649867374005306,
      "grad_norm": 3.055052876372201,
      "learning_rate": 9.402117141704528e-07,
      "loss": 0.882,
      "step": 3864
    },
    {
      "epoch": 0.35659093530158,
      "grad_norm": 2.2987037223477986,
      "learning_rate": 9.400492343995913e-07,
      "loss": 0.6937,
      "step": 3865
    },
    {
      "epoch": 0.35659093530158,
      "eval_GEN Loss": 0.46811652183532715,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3976913392543793,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8783053159713745,
      "eval_runtime": 55.9134,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3865
    },
    {
      "epoch": 0.3566831968631069,
      "grad_norm": 1.88775680309827,
      "learning_rate": 9.398867260784722e-07,
      "loss": 0.7951,
      "step": 3866
    },
    {
      "epoch": 0.3567754584246338,
      "grad_norm": 2.757702026301362,
      "learning_rate": 9.397241892218224e-07,
      "loss": 0.7589,
      "step": 3867
    },
    {
      "epoch": 0.3568677199861608,
      "grad_norm": 1.4183440981857498,
      "learning_rate": 9.395616238443711e-07,
      "loss": 0.6209,
      "step": 3868
    },
    {
      "epoch": 0.3569599815476877,
      "grad_norm": 2.116785998997551,
      "learning_rate": 9.393990299608505e-07,
      "loss": 0.4923,
      "step": 3869
    },
    {
      "epoch": 0.35705224310921463,
      "grad_norm": 2.0178708897587248,
      "learning_rate": 9.392364075859947e-07,
      "loss": 0.8115,
      "step": 3870
    },
    {
      "epoch": 0.35705224310921463,
      "eval_GEN Loss": 0.46893763542175293,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.38415005803108215,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8784855604171753,
      "eval_runtime": 56.6675,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 3870
    },
    {
      "epoch": 0.35714450467074155,
      "grad_norm": 1.9452389432296426,
      "learning_rate": 9.390737567345412e-07,
      "loss": 0.6038,
      "step": 3871
    },
    {
      "epoch": 0.35723676623226847,
      "grad_norm": 2.6029200983402556,
      "learning_rate": 9.389110774212292e-07,
      "loss": 0.8236,
      "step": 3872
    },
    {
      "epoch": 0.3573290277937954,
      "grad_norm": 2.134219518901736,
      "learning_rate": 9.387483696608017e-07,
      "loss": 0.6429,
      "step": 3873
    },
    {
      "epoch": 0.35742128935532236,
      "grad_norm": 1.2882575551910274,
      "learning_rate": 9.385856334680028e-07,
      "loss": 0.5704,
      "step": 3874
    },
    {
      "epoch": 0.3575135509168493,
      "grad_norm": 2.2756337927689088,
      "learning_rate": 9.384228688575804e-07,
      "loss": 0.7023,
      "step": 3875
    },
    {
      "epoch": 0.3575135509168493,
      "eval_GEN Loss": 0.4678434431552887,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3739252984523773,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8676081895828247,
      "eval_runtime": 56.8406,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3875
    },
    {
      "epoch": 0.3576058124783762,
      "grad_norm": 3.145844345187359,
      "learning_rate": 9.382600758442842e-07,
      "loss": 0.8728,
      "step": 3876
    },
    {
      "epoch": 0.3576980740399031,
      "grad_norm": 2.7632525316118164,
      "learning_rate": 9.380972544428668e-07,
      "loss": 0.8656,
      "step": 3877
    },
    {
      "epoch": 0.35779033560143003,
      "grad_norm": 1.8978531612523957,
      "learning_rate": 9.379344046680834e-07,
      "loss": 0.5401,
      "step": 3878
    },
    {
      "epoch": 0.357882597162957,
      "grad_norm": 2.388514387758438,
      "learning_rate": 9.377715265346918e-07,
      "loss": 0.6398,
      "step": 3879
    },
    {
      "epoch": 0.3579748587244839,
      "grad_norm": 1.8780245639190891,
      "learning_rate": 9.376086200574521e-07,
      "loss": 0.6647,
      "step": 3880
    },
    {
      "epoch": 0.3579748587244839,
      "eval_GEN Loss": 0.46916478872299194,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8695652173913043,
      "eval_PRM F1 AUC": 0.747773703509691,
      "eval_PRM F1 Neg": 0.5882352941176471,
      "eval_PRM Loss": 0.3886345326900482,
      "eval_PRM NPV": 0.5357142857142857,
      "eval_PRM Precision": 0.8974358974358975,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8644831776618958,
      "eval_runtime": 56.9015,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3880
    },
    {
      "epoch": 0.35806712028601084,
      "grad_norm": 2.009493156742869,
      "learning_rate": 9.374456852511271e-07,
      "loss": 0.6761,
      "step": 3881
    },
    {
      "epoch": 0.35815938184753776,
      "grad_norm": 1.7169391858657355,
      "learning_rate": 9.372827221304824e-07,
      "loss": 0.7489,
      "step": 3882
    },
    {
      "epoch": 0.3582516434090647,
      "grad_norm": 1.477651972755526,
      "learning_rate": 9.371197307102858e-07,
      "loss": 0.7054,
      "step": 3883
    },
    {
      "epoch": 0.35834390497059165,
      "grad_norm": 1.4775545777510137,
      "learning_rate": 9.369567110053079e-07,
      "loss": 0.659,
      "step": 3884
    },
    {
      "epoch": 0.35843616653211857,
      "grad_norm": 2.361819598256709,
      "learning_rate": 9.367936630303219e-07,
      "loss": 0.6494,
      "step": 3885
    },
    {
      "epoch": 0.35843616653211857,
      "eval_GEN Loss": 0.4692051410675049,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.3907117545604706,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8585336804389954,
      "eval_runtime": 56.947,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3885
    },
    {
      "epoch": 0.3585284280936455,
      "grad_norm": 2.1015243521485867,
      "learning_rate": 9.366305868001032e-07,
      "loss": 0.6841,
      "step": 3886
    },
    {
      "epoch": 0.3586206896551724,
      "grad_norm": 1.5275645613236177,
      "learning_rate": 9.364674823294305e-07,
      "loss": 0.7102,
      "step": 3887
    },
    {
      "epoch": 0.3587129512166993,
      "grad_norm": 1.694547908499395,
      "learning_rate": 9.363043496330841e-07,
      "loss": 0.6289,
      "step": 3888
    },
    {
      "epoch": 0.35880521277822625,
      "grad_norm": 2.6472250028389914,
      "learning_rate": 9.361411887258478e-07,
      "loss": 0.7378,
      "step": 3889
    },
    {
      "epoch": 0.3588974743397532,
      "grad_norm": 2.0570534130896982,
      "learning_rate": 9.359779996225069e-07,
      "loss": 0.849,
      "step": 3890
    },
    {
      "epoch": 0.3588974743397532,
      "eval_GEN Loss": 0.4700799286365509,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3856019377708435,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8542067408561707,
      "eval_runtime": 55.9946,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3890
    },
    {
      "epoch": 0.35898973590128014,
      "grad_norm": 2.4627103209735792,
      "learning_rate": 9.358147823378504e-07,
      "loss": 0.7538,
      "step": 3891
    },
    {
      "epoch": 0.35908199746280706,
      "grad_norm": 1.889462017915708,
      "learning_rate": 9.356515368866691e-07,
      "loss": 0.7394,
      "step": 3892
    },
    {
      "epoch": 0.359174259024334,
      "grad_norm": 1.5176644901269198,
      "learning_rate": 9.354882632837564e-07,
      "loss": 0.6269,
      "step": 3893
    },
    {
      "epoch": 0.3592665205858609,
      "grad_norm": 2.443968729888847,
      "learning_rate": 9.353249615439086e-07,
      "loss": 0.8084,
      "step": 3894
    },
    {
      "epoch": 0.35935878214738787,
      "grad_norm": 1.6250856842066927,
      "learning_rate": 9.351616316819244e-07,
      "loss": 0.6622,
      "step": 3895
    },
    {
      "epoch": 0.35935878214738787,
      "eval_GEN Loss": 0.4713021218776703,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.389230340719223,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8591646552085876,
      "eval_runtime": 56.9,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3895
    },
    {
      "epoch": 0.3594510437089148,
      "grad_norm": 2.135737398545149,
      "learning_rate": 9.349982737126049e-07,
      "loss": 0.8483,
      "step": 3896
    },
    {
      "epoch": 0.3595433052704417,
      "grad_norm": 3.0544410601012912,
      "learning_rate": 9.348348876507539e-07,
      "loss": 0.997,
      "step": 3897
    },
    {
      "epoch": 0.3596355668319686,
      "grad_norm": 1.7750545970717169,
      "learning_rate": 9.346714735111777e-07,
      "loss": 0.7438,
      "step": 3898
    },
    {
      "epoch": 0.35972782839349554,
      "grad_norm": 2.354262303495788,
      "learning_rate": 9.345080313086852e-07,
      "loss": 0.9604,
      "step": 3899
    },
    {
      "epoch": 0.3598200899550225,
      "grad_norm": 1.5586055281824707,
      "learning_rate": 9.343445610580876e-07,
      "loss": 0.5026,
      "step": 3900
    },
    {
      "epoch": 0.3598200899550225,
      "eval_GEN Loss": 0.4718244969844818,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.3874286711215973,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8618389368057251,
      "eval_runtime": 55.9624,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 3900
    },
    {
      "epoch": 0.35991235151654943,
      "grad_norm": 2.532291224005924,
      "learning_rate": 9.34181062774199e-07,
      "loss": 0.8086,
      "step": 3901
    },
    {
      "epoch": 0.36000461307807635,
      "grad_norm": 1.4441988268323078,
      "learning_rate": 9.34017536471836e-07,
      "loss": 0.769,
      "step": 3902
    },
    {
      "epoch": 0.36009687463960327,
      "grad_norm": 1.5550775104131975,
      "learning_rate": 9.338539821658173e-07,
      "loss": 0.6358,
      "step": 3903
    },
    {
      "epoch": 0.3601891362011302,
      "grad_norm": 2.1635188110029695,
      "learning_rate": 9.336903998709645e-07,
      "loss": 0.8817,
      "step": 3904
    },
    {
      "epoch": 0.3602813977626571,
      "grad_norm": 2.2757405363598853,
      "learning_rate": 9.33526789602102e-07,
      "loss": 0.8107,
      "step": 3905
    },
    {
      "epoch": 0.3602813977626571,
      "eval_GEN Loss": 0.47220349311828613,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3897966742515564,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8663762211799622,
      "eval_runtime": 55.8937,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3905
    },
    {
      "epoch": 0.3603736593241841,
      "grad_norm": 2.2308007808204837,
      "learning_rate": 9.33363151374056e-07,
      "loss": 0.8066,
      "step": 3906
    },
    {
      "epoch": 0.360465920885711,
      "grad_norm": 1.3362641067355674,
      "learning_rate": 9.331994852016558e-07,
      "loss": 0.4111,
      "step": 3907
    },
    {
      "epoch": 0.3605581824472379,
      "grad_norm": 2.5460885774899906,
      "learning_rate": 9.330357910997333e-07,
      "loss": 0.6938,
      "step": 3908
    },
    {
      "epoch": 0.36065044400876484,
      "grad_norm": 2.2605960671612153,
      "learning_rate": 9.328720690831225e-07,
      "loss": 0.772,
      "step": 3909
    },
    {
      "epoch": 0.36074270557029176,
      "grad_norm": 1.792022086468857,
      "learning_rate": 9.327083191666603e-07,
      "loss": 0.8155,
      "step": 3910
    },
    {
      "epoch": 0.36074270557029176,
      "eval_GEN Loss": 0.4752283990383148,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.39145511388778687,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8692908883094788,
      "eval_runtime": 55.8883,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 3910
    },
    {
      "epoch": 0.36083496713181873,
      "grad_norm": 1.9763901690157581,
      "learning_rate": 9.325445413651857e-07,
      "loss": 0.7397,
      "step": 3911
    },
    {
      "epoch": 0.36092722869334565,
      "grad_norm": 2.225812758965922,
      "learning_rate": 9.323807356935408e-07,
      "loss": 0.7975,
      "step": 3912
    },
    {
      "epoch": 0.36101949025487257,
      "grad_norm": 1.4682897173636202,
      "learning_rate": 9.322169021665697e-07,
      "loss": 0.6978,
      "step": 3913
    },
    {
      "epoch": 0.3611117518163995,
      "grad_norm": 1.9344383215245098,
      "learning_rate": 9.320530407991195e-07,
      "loss": 0.7764,
      "step": 3914
    },
    {
      "epoch": 0.3612040133779264,
      "grad_norm": 1.2722725998123767,
      "learning_rate": 9.318891516060394e-07,
      "loss": 0.5773,
      "step": 3915
    },
    {
      "epoch": 0.3612040133779264,
      "eval_GEN Loss": 0.4741431474685669,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.38687625527381897,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8719651699066162,
      "eval_runtime": 55.7692,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 3915
    },
    {
      "epoch": 0.3612962749394534,
      "grad_norm": 1.8212294838929985,
      "learning_rate": 9.317252346021813e-07,
      "loss": 0.5912,
      "step": 3916
    },
    {
      "epoch": 0.3613885365009803,
      "grad_norm": 1.7452254458039675,
      "learning_rate": 9.315612898023996e-07,
      "loss": 0.6987,
      "step": 3917
    },
    {
      "epoch": 0.3614807980625072,
      "grad_norm": 1.31477492305167,
      "learning_rate": 9.313973172215512e-07,
      "loss": 0.6211,
      "step": 3918
    },
    {
      "epoch": 0.36157305962403413,
      "grad_norm": 2.0074989374716616,
      "learning_rate": 9.31233316874496e-07,
      "loss": 0.8034,
      "step": 3919
    },
    {
      "epoch": 0.36166532118556105,
      "grad_norm": 1.6624007379242667,
      "learning_rate": 9.310692887760953e-07,
      "loss": 0.6212,
      "step": 3920
    },
    {
      "epoch": 0.36166532118556105,
      "eval_GEN Loss": 0.4754544496536255,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.38960573077201843,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8695312738418579,
      "eval_runtime": 55.9215,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3920
    },
    {
      "epoch": 0.36175758274708797,
      "grad_norm": 2.9188768585500684,
      "learning_rate": 9.30905232941214e-07,
      "loss": 0.6554,
      "step": 3921
    },
    {
      "epoch": 0.36184984430861494,
      "grad_norm": 2.8486299249502864,
      "learning_rate": 9.307411493847189e-07,
      "loss": 0.7328,
      "step": 3922
    },
    {
      "epoch": 0.36194210587014186,
      "grad_norm": 3.0359147936149955,
      "learning_rate": 9.305770381214797e-07,
      "loss": 0.9836,
      "step": 3923
    },
    {
      "epoch": 0.3620343674316688,
      "grad_norm": 2.232953980129854,
      "learning_rate": 9.304128991663682e-07,
      "loss": 0.7897,
      "step": 3924
    },
    {
      "epoch": 0.3621266289931957,
      "grad_norm": 2.5579285213508447,
      "learning_rate": 9.302487325342592e-07,
      "loss": 0.8184,
      "step": 3925
    },
    {
      "epoch": 0.3621266289931957,
      "eval_GEN Loss": 0.4731970429420471,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3874712586402893,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8670673370361328,
      "eval_runtime": 55.9538,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3925
    },
    {
      "epoch": 0.3622188905547226,
      "grad_norm": 1.544779658763646,
      "learning_rate": 9.300845382400295e-07,
      "loss": 0.6434,
      "step": 3926
    },
    {
      "epoch": 0.3623111521162496,
      "grad_norm": 2.533379720921697,
      "learning_rate": 9.299203162985588e-07,
      "loss": 0.814,
      "step": 3927
    },
    {
      "epoch": 0.3624034136777765,
      "grad_norm": 2.0187043322722165,
      "learning_rate": 9.297560667247291e-07,
      "loss": 0.6568,
      "step": 3928
    },
    {
      "epoch": 0.3624956752393034,
      "grad_norm": 2.1658462527501814,
      "learning_rate": 9.295917895334247e-07,
      "loss": 0.8179,
      "step": 3929
    },
    {
      "epoch": 0.36258793680083035,
      "grad_norm": 1.8097854359582977,
      "learning_rate": 9.294274847395333e-07,
      "loss": 0.6307,
      "step": 3930
    },
    {
      "epoch": 0.36258793680083035,
      "eval_GEN Loss": 0.47248291969299316,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3808736503124237,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8524940013885498,
      "eval_runtime": 55.9452,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3930
    },
    {
      "epoch": 0.36268019836235726,
      "grad_norm": 3.1236716350800844,
      "learning_rate": 9.292631523579436e-07,
      "loss": 0.8585,
      "step": 3931
    },
    {
      "epoch": 0.36277245992388424,
      "grad_norm": 1.7941913378758887,
      "learning_rate": 9.290987924035483e-07,
      "loss": 0.7959,
      "step": 3932
    },
    {
      "epoch": 0.36286472148541116,
      "grad_norm": 1.5539593321714393,
      "learning_rate": 9.289344048912419e-07,
      "loss": 0.5842,
      "step": 3933
    },
    {
      "epoch": 0.3629569830469381,
      "grad_norm": 1.6377698814107418,
      "learning_rate": 9.287699898359212e-07,
      "loss": 0.5516,
      "step": 3934
    },
    {
      "epoch": 0.363049244608465,
      "grad_norm": 1.5011577632910293,
      "learning_rate": 9.28605547252486e-07,
      "loss": 0.493,
      "step": 3935
    },
    {
      "epoch": 0.363049244608465,
      "eval_GEN Loss": 0.4696495532989502,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.3636634349822998,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8429387211799622,
      "eval_runtime": 55.9414,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 3935
    },
    {
      "epoch": 0.3631415061699919,
      "grad_norm": 2.185620317280508,
      "learning_rate": 9.284410771558381e-07,
      "loss": 0.64,
      "step": 3936
    },
    {
      "epoch": 0.36323376773151883,
      "grad_norm": 1.1611957577343748,
      "learning_rate": 9.282765795608823e-07,
      "loss": 0.5867,
      "step": 3937
    },
    {
      "epoch": 0.3633260292930458,
      "grad_norm": 2.29939808391568,
      "learning_rate": 9.281120544825253e-07,
      "loss": 0.7806,
      "step": 3938
    },
    {
      "epoch": 0.3634182908545727,
      "grad_norm": 1.7476494741268465,
      "learning_rate": 9.27947501935677e-07,
      "loss": 0.6418,
      "step": 3939
    },
    {
      "epoch": 0.36351055241609964,
      "grad_norm": 1.832308747311918,
      "learning_rate": 9.277829219352492e-07,
      "loss": 0.7227,
      "step": 3940
    },
    {
      "epoch": 0.36351055241609964,
      "eval_GEN Loss": 0.4686095714569092,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.35803502798080444,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8316105604171753,
      "eval_runtime": 56.9301,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3940
    },
    {
      "epoch": 0.36360281397762656,
      "grad_norm": 1.756504456086922,
      "learning_rate": 9.276183144961563e-07,
      "loss": 0.5178,
      "step": 3941
    },
    {
      "epoch": 0.3636950755391535,
      "grad_norm": 1.539716152384563,
      "learning_rate": 9.274536796333155e-07,
      "loss": 0.6088,
      "step": 3942
    },
    {
      "epoch": 0.36378733710068045,
      "grad_norm": 2.2323749504744295,
      "learning_rate": 9.272890173616461e-07,
      "loss": 0.6462,
      "step": 3943
    },
    {
      "epoch": 0.36387959866220737,
      "grad_norm": 2.7944215350707444,
      "learning_rate": 9.271243276960703e-07,
      "loss": 0.9325,
      "step": 3944
    },
    {
      "epoch": 0.3639718602237343,
      "grad_norm": 2.2827568375840417,
      "learning_rate": 9.269596106515123e-07,
      "loss": 0.7828,
      "step": 3945
    },
    {
      "epoch": 0.3639718602237343,
      "eval_GEN Loss": 0.46773645281791687,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.3533172607421875,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8210036158561707,
      "eval_runtime": 56.8395,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3945
    },
    {
      "epoch": 0.3640641217852612,
      "grad_norm": 1.2240595302825128,
      "learning_rate": 9.267948662428991e-07,
      "loss": 0.5274,
      "step": 3946
    },
    {
      "epoch": 0.3641563833467881,
      "grad_norm": 3.213220609306924,
      "learning_rate": 9.2663009448516e-07,
      "loss": 0.8502,
      "step": 3947
    },
    {
      "epoch": 0.3642486449083151,
      "grad_norm": 2.392812082105232,
      "learning_rate": 9.264652953932269e-07,
      "loss": 0.63,
      "step": 3948
    },
    {
      "epoch": 0.364340906469842,
      "grad_norm": 1.5045056292338777,
      "learning_rate": 9.263004689820346e-07,
      "loss": 0.5596,
      "step": 3949
    },
    {
      "epoch": 0.36443316803136894,
      "grad_norm": 2.274918683669336,
      "learning_rate": 9.261356152665194e-07,
      "loss": 0.6024,
      "step": 3950
    },
    {
      "epoch": 0.36443316803136894,
      "eval_GEN Loss": 0.4670032262802124,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9221556886227545,
      "eval_PRM F1 AUC": 0.8116815086432687,
      "eval_PRM F1 Neg": 0.7111111111111111,
      "eval_PRM Loss": 0.3519865870475769,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9166666666666666,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8191105723381042,
      "eval_runtime": 56.9892,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3950
    },
    {
      "epoch": 0.36452542959289586,
      "grad_norm": 1.9811678560162718,
      "learning_rate": 9.259707342616206e-07,
      "loss": 0.6069,
      "step": 3951
    },
    {
      "epoch": 0.3646176911544228,
      "grad_norm": 1.4454342477217352,
      "learning_rate": 9.258058259822801e-07,
      "loss": 0.5096,
      "step": 3952
    },
    {
      "epoch": 0.3647099527159497,
      "grad_norm": 2.3156618693135798,
      "learning_rate": 9.256408904434424e-07,
      "loss": 0.6777,
      "step": 3953
    },
    {
      "epoch": 0.36480221427747667,
      "grad_norm": 1.740709241229069,
      "learning_rate": 9.254759276600536e-07,
      "loss": 0.706,
      "step": 3954
    },
    {
      "epoch": 0.3648944758390036,
      "grad_norm": 2.606479521447348,
      "learning_rate": 9.253109376470637e-07,
      "loss": 0.8412,
      "step": 3955
    },
    {
      "epoch": 0.3648944758390036,
      "eval_GEN Loss": 0.468524694442749,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.3395959734916687,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8205228447914124,
      "eval_runtime": 56.8418,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3955
    },
    {
      "epoch": 0.3649867374005305,
      "grad_norm": 2.7606277682607976,
      "learning_rate": 9.251459204194237e-07,
      "loss": 0.8132,
      "step": 3956
    },
    {
      "epoch": 0.3650789989620574,
      "grad_norm": 3.0246155566793966,
      "learning_rate": 9.24980875992088e-07,
      "loss": 0.5843,
      "step": 3957
    },
    {
      "epoch": 0.36517126052358434,
      "grad_norm": 2.1177339370604957,
      "learning_rate": 9.248158043800131e-07,
      "loss": 0.7132,
      "step": 3958
    },
    {
      "epoch": 0.3652635220851113,
      "grad_norm": 2.0518771132659035,
      "learning_rate": 9.246507055981581e-07,
      "loss": 0.7696,
      "step": 3959
    },
    {
      "epoch": 0.36535578364663823,
      "grad_norm": 2.0204041923069322,
      "learning_rate": 9.244855796614847e-07,
      "loss": 0.6573,
      "step": 3960
    },
    {
      "epoch": 0.36535578364663823,
      "eval_GEN Loss": 0.4692714214324951,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.34468531608581543,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8237680196762085,
      "eval_runtime": 56.9049,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 3960
    },
    {
      "epoch": 0.36544804520816515,
      "grad_norm": 2.3531646256363543,
      "learning_rate": 9.243204265849565e-07,
      "loss": 0.7047,
      "step": 3961
    },
    {
      "epoch": 0.36554030676969207,
      "grad_norm": 2.190645571443471,
      "learning_rate": 9.241552463835401e-07,
      "loss": 0.7353,
      "step": 3962
    },
    {
      "epoch": 0.365632568331219,
      "grad_norm": 1.836449679894377,
      "learning_rate": 9.239900390722044e-07,
      "loss": 0.6638,
      "step": 3963
    },
    {
      "epoch": 0.36572482989274596,
      "grad_norm": 1.8075356136350886,
      "learning_rate": 9.238248046659208e-07,
      "loss": 0.8139,
      "step": 3964
    },
    {
      "epoch": 0.3658170914542729,
      "grad_norm": 2.071607684536743,
      "learning_rate": 9.23659543179663e-07,
      "loss": 0.8313,
      "step": 3965
    },
    {
      "epoch": 0.3658170914542729,
      "eval_GEN Loss": 0.4681418836116791,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.34498780965805054,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8307992815971375,
      "eval_runtime": 56.9457,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 3965
    },
    {
      "epoch": 0.3659093530157998,
      "grad_norm": 1.6615308373019935,
      "learning_rate": 9.234942546284073e-07,
      "loss": 0.6671,
      "step": 3966
    },
    {
      "epoch": 0.3660016145773267,
      "grad_norm": 2.407311589772096,
      "learning_rate": 9.233289390271324e-07,
      "loss": 0.7313,
      "step": 3967
    },
    {
      "epoch": 0.36609387613885364,
      "grad_norm": 1.78091081399118,
      "learning_rate": 9.231635963908195e-07,
      "loss": 0.7081,
      "step": 3968
    },
    {
      "epoch": 0.36618613770038055,
      "grad_norm": 1.3524491992351055,
      "learning_rate": 9.229982267344521e-07,
      "loss": 0.4901,
      "step": 3969
    },
    {
      "epoch": 0.36627839926190753,
      "grad_norm": 2.4125619013458626,
      "learning_rate": 9.228328300730163e-07,
      "loss": 0.7864,
      "step": 3970
    },
    {
      "epoch": 0.36627839926190753,
      "eval_GEN Loss": 0.46698760986328125,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9166666666666666,
      "eval_PRM F1 AUC": 0.7899423782084861,
      "eval_PRM F1 Neg": 0.6818181818181818,
      "eval_PRM Loss": 0.34183862805366516,
      "eval_PRM NPV": 0.7142857142857143,
      "eval_PRM Precision": 0.9058823529411765,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8408052921295166,
      "eval_runtime": 56.717,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 3970
    },
    {
      "epoch": 0.36637066082343445,
      "grad_norm": 2.0102619639630483,
      "learning_rate": 9.226674064215009e-07,
      "loss": 0.6951,
      "step": 3971
    },
    {
      "epoch": 0.36646292238496136,
      "grad_norm": 2.1899655139354675,
      "learning_rate": 9.225019557948963e-07,
      "loss": 0.879,
      "step": 3972
    },
    {
      "epoch": 0.3665551839464883,
      "grad_norm": 2.9473848516990353,
      "learning_rate": 9.223364782081965e-07,
      "loss": 1.1008,
      "step": 3973
    },
    {
      "epoch": 0.3666474455080152,
      "grad_norm": 1.734542962304058,
      "learning_rate": 9.221709736763969e-07,
      "loss": 0.571,
      "step": 3974
    },
    {
      "epoch": 0.3667397070695422,
      "grad_norm": 2.087318540950687,
      "learning_rate": 9.220054422144958e-07,
      "loss": 0.7517,
      "step": 3975
    },
    {
      "epoch": 0.3667397070695422,
      "eval_GEN Loss": 0.4662870168685913,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9221556886227545,
      "eval_PRM F1 AUC": 0.8116815086432687,
      "eval_PRM F1 Neg": 0.7111111111111111,
      "eval_PRM Loss": 0.3461471498012543,
      "eval_PRM NPV": 0.7272727272727273,
      "eval_PRM Precision": 0.9166666666666666,
      "eval_PRM Recall": 0.927710843373494,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8326321840286255,
      "eval_runtime": 56.8207,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3975
    },
    {
      "epoch": 0.3668319686310691,
      "grad_norm": 4.192616207272796,
      "learning_rate": 9.218398838374941e-07,
      "loss": 1.0464,
      "step": 3976
    },
    {
      "epoch": 0.366924230192596,
      "grad_norm": 1.3163866020779345,
      "learning_rate": 9.216742985603949e-07,
      "loss": 0.7344,
      "step": 3977
    },
    {
      "epoch": 0.36701649175412293,
      "grad_norm": 1.7421045252575347,
      "learning_rate": 9.215086863982038e-07,
      "loss": 0.7164,
      "step": 3978
    },
    {
      "epoch": 0.36710875331564985,
      "grad_norm": 1.9426012337940428,
      "learning_rate": 9.213430473659288e-07,
      "loss": 0.7645,
      "step": 3979
    },
    {
      "epoch": 0.3672010148771768,
      "grad_norm": 2.078511095833166,
      "learning_rate": 9.211773814785805e-07,
      "loss": 0.7956,
      "step": 3980
    },
    {
      "epoch": 0.3672010148771768,
      "eval_GEN Loss": 0.4650685787200928,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM Loss": 0.354384183883667,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8284855484962463,
      "eval_runtime": 56.7813,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 3980
    },
    {
      "epoch": 0.36729327643870374,
      "grad_norm": 1.4226300487337655,
      "learning_rate": 9.210116887511715e-07,
      "loss": 0.6856,
      "step": 3981
    },
    {
      "epoch": 0.36738553800023066,
      "grad_norm": 1.5802065341474425,
      "learning_rate": 9.208459691987173e-07,
      "loss": 0.7379,
      "step": 3982
    },
    {
      "epoch": 0.3674777995617576,
      "grad_norm": 2.931374876411286,
      "learning_rate": 9.20680222836236e-07,
      "loss": 0.8796,
      "step": 3983
    },
    {
      "epoch": 0.3675700611232845,
      "grad_norm": 1.664174322429993,
      "learning_rate": 9.205144496787473e-07,
      "loss": 0.5219,
      "step": 3984
    },
    {
      "epoch": 0.3676623226848114,
      "grad_norm": 2.3045945175787406,
      "learning_rate": 9.203486497412741e-07,
      "loss": 0.6681,
      "step": 3985
    },
    {
      "epoch": 0.3676623226848114,
      "eval_GEN Loss": 0.4651217758655548,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.360866516828537,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8353966474533081,
      "eval_runtime": 56.5806,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 3985
    },
    {
      "epoch": 0.3677545842463384,
      "grad_norm": 1.437534499562714,
      "learning_rate": 9.201828230388414e-07,
      "loss": 0.5612,
      "step": 3986
    },
    {
      "epoch": 0.3678468458078653,
      "grad_norm": 2.0339165764923592,
      "learning_rate": 9.200169695864766e-07,
      "loss": 0.7082,
      "step": 3987
    },
    {
      "epoch": 0.3679391073693922,
      "grad_norm": 2.4339029712856015,
      "learning_rate": 9.198510893992096e-07,
      "loss": 0.9734,
      "step": 3988
    },
    {
      "epoch": 0.36803136893091914,
      "grad_norm": 1.8425620431468062,
      "learning_rate": 9.196851824920728e-07,
      "loss": 0.7213,
      "step": 3989
    },
    {
      "epoch": 0.36812363049244606,
      "grad_norm": 2.0436216851794606,
      "learning_rate": 9.195192488801009e-07,
      "loss": 0.6996,
      "step": 3990
    },
    {
      "epoch": 0.36812363049244606,
      "eval_GEN Loss": 0.46527743339538574,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.36600261926651,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8408653736114502,
      "eval_runtime": 56.7965,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 3990
    },
    {
      "epoch": 0.36821589205397304,
      "grad_norm": 2.8902439169409244,
      "learning_rate": 9.193532885783311e-07,
      "loss": 0.9879,
      "step": 3991
    },
    {
      "epoch": 0.36830815361549996,
      "grad_norm": 1.6074402482743189,
      "learning_rate": 9.191873016018032e-07,
      "loss": 0.755,
      "step": 3992
    },
    {
      "epoch": 0.3684004151770269,
      "grad_norm": 2.707070551204786,
      "learning_rate": 9.190212879655585e-07,
      "loss": 0.8225,
      "step": 3993
    },
    {
      "epoch": 0.3684926767385538,
      "grad_norm": 2.0578697298196187,
      "learning_rate": 9.188552476846424e-07,
      "loss": 0.6969,
      "step": 3994
    },
    {
      "epoch": 0.3685849383000807,
      "grad_norm": 1.2209581399760836,
      "learning_rate": 9.186891807741008e-07,
      "loss": 0.6195,
      "step": 3995
    },
    {
      "epoch": 0.3685849383000807,
      "eval_GEN Loss": 0.4651360511779785,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.3803725242614746,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8537259697914124,
      "eval_runtime": 56.8539,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 3995
    },
    {
      "epoch": 0.3686771998616077,
      "grad_norm": 1.9462806070740368,
      "learning_rate": 9.185230872489836e-07,
      "loss": 0.6783,
      "step": 3996
    },
    {
      "epoch": 0.3687694614231346,
      "grad_norm": 1.515197633726276,
      "learning_rate": 9.183569671243422e-07,
      "loss": 0.7117,
      "step": 3997
    },
    {
      "epoch": 0.3688617229846615,
      "grad_norm": 2.8910756193567004,
      "learning_rate": 9.181908204152307e-07,
      "loss": 1.06,
      "step": 3998
    },
    {
      "epoch": 0.36895398454618844,
      "grad_norm": 1.6565018381522563,
      "learning_rate": 9.180246471367056e-07,
      "loss": 0.7391,
      "step": 3999
    },
    {
      "epoch": 0.36904624610771536,
      "grad_norm": 1.8025853766831315,
      "learning_rate": 9.178584473038256e-07,
      "loss": 0.7724,
      "step": 4000
    },
    {
      "epoch": 0.36904624610771536,
      "eval_GEN Loss": 0.46650370955467224,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3997230529785156,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8706430196762085,
      "eval_runtime": 56.8088,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 4000
    },
    {
      "epoch": 0.3691385076692423,
      "grad_norm": 2.682410159310335,
      "learning_rate": 9.176922209316523e-07,
      "loss": 0.8458,
      "step": 4001
    },
    {
      "epoch": 0.36923076923076925,
      "grad_norm": 1.8513303134741645,
      "learning_rate": 9.175259680352491e-07,
      "loss": 0.8483,
      "step": 4002
    },
    {
      "epoch": 0.36932303079229617,
      "grad_norm": 2.0211716726128124,
      "learning_rate": 9.173596886296824e-07,
      "loss": 0.7502,
      "step": 4003
    },
    {
      "epoch": 0.3694152923538231,
      "grad_norm": 1.6796074854229206,
      "learning_rate": 9.171933827300203e-07,
      "loss": 0.7305,
      "step": 4004
    },
    {
      "epoch": 0.36950755391535,
      "grad_norm": 2.065852070488085,
      "learning_rate": 9.17027050351334e-07,
      "loss": 0.6767,
      "step": 4005
    },
    {
      "epoch": 0.36950755391535,
      "eval_GEN Loss": 0.4657479524612427,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.39866435527801514,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8692307472229004,
      "eval_runtime": 56.9144,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 4005
    },
    {
      "epoch": 0.3695998154768769,
      "grad_norm": 1.7698874648907141,
      "learning_rate": 9.168606915086968e-07,
      "loss": 0.7763,
      "step": 4006
    },
    {
      "epoch": 0.3696920770384039,
      "grad_norm": 3.00804928931779,
      "learning_rate": 9.166943062171843e-07,
      "loss": 0.8135,
      "step": 4007
    },
    {
      "epoch": 0.3697843385999308,
      "grad_norm": 2.1118877457957024,
      "learning_rate": 9.165278944918747e-07,
      "loss": 0.8485,
      "step": 4008
    },
    {
      "epoch": 0.36987660016145774,
      "grad_norm": 2.0106584675711883,
      "learning_rate": 9.163614563478484e-07,
      "loss": 0.9577,
      "step": 4009
    },
    {
      "epoch": 0.36996886172298465,
      "grad_norm": 1.728129128202833,
      "learning_rate": 9.161949918001884e-07,
      "loss": 0.7102,
      "step": 4010
    },
    {
      "epoch": 0.36996886172298465,
      "eval_GEN Loss": 0.4645037055015564,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.39239707589149475,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8724759817123413,
      "eval_runtime": 55.8278,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4010
    },
    {
      "epoch": 0.3700611232845116,
      "grad_norm": 2.035905752614218,
      "learning_rate": 9.160285008639798e-07,
      "loss": 0.8883,
      "step": 4011
    },
    {
      "epoch": 0.37015338484603855,
      "grad_norm": 1.3860527719352767,
      "learning_rate": 9.158619835543104e-07,
      "loss": 0.5338,
      "step": 4012
    },
    {
      "epoch": 0.37024564640756547,
      "grad_norm": 1.4082787162490455,
      "learning_rate": 9.156954398862701e-07,
      "loss": 0.6993,
      "step": 4013
    },
    {
      "epoch": 0.3703379079690924,
      "grad_norm": 2.892460946245489,
      "learning_rate": 9.155288698749514e-07,
      "loss": 0.9446,
      "step": 4014
    },
    {
      "epoch": 0.3704301695306193,
      "grad_norm": 1.8448389644174532,
      "learning_rate": 9.153622735354497e-07,
      "loss": 0.7462,
      "step": 4015
    },
    {
      "epoch": 0.3704301695306193,
      "eval_GEN Loss": 0.4658695459365845,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.39387214183807373,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8804687261581421,
      "eval_runtime": 55.7695,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4015
    },
    {
      "epoch": 0.3705224310921462,
      "grad_norm": 1.838572471588768,
      "learning_rate": 9.151956508828613e-07,
      "loss": 0.6251,
      "step": 4016
    },
    {
      "epoch": 0.37061469265367314,
      "grad_norm": 1.7592334646815513,
      "learning_rate": 9.150290019322864e-07,
      "loss": 0.7358,
      "step": 4017
    },
    {
      "epoch": 0.3707069542152001,
      "grad_norm": 1.6043460622418197,
      "learning_rate": 9.148623266988267e-07,
      "loss": 0.5459,
      "step": 4018
    },
    {
      "epoch": 0.37079921577672703,
      "grad_norm": 2.28665835970567,
      "learning_rate": 9.14695625197587e-07,
      "loss": 0.6591,
      "step": 4019
    },
    {
      "epoch": 0.37089147733825395,
      "grad_norm": 1.9493097046410903,
      "learning_rate": 9.145288974436733e-07,
      "loss": 0.8836,
      "step": 4020
    },
    {
      "epoch": 0.37089147733825395,
      "eval_GEN Loss": 0.4661194086074829,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3922044634819031,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8952524065971375,
      "eval_runtime": 55.7277,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4020
    },
    {
      "epoch": 0.37098373889978087,
      "grad_norm": 2.071579706138531,
      "learning_rate": 9.143621434521956e-07,
      "loss": 0.6997,
      "step": 4021
    },
    {
      "epoch": 0.3710760004613078,
      "grad_norm": 1.8797242045101994,
      "learning_rate": 9.14195363238265e-07,
      "loss": 0.7907,
      "step": 4022
    },
    {
      "epoch": 0.37116826202283476,
      "grad_norm": 1.4780756985737438,
      "learning_rate": 9.140285568169951e-07,
      "loss": 0.5715,
      "step": 4023
    },
    {
      "epoch": 0.3712605235843617,
      "grad_norm": 2.169615330791163,
      "learning_rate": 9.138617242035026e-07,
      "loss": 0.6418,
      "step": 4024
    },
    {
      "epoch": 0.3713527851458886,
      "grad_norm": 2.4617614692052947,
      "learning_rate": 9.136948654129058e-07,
      "loss": 0.8067,
      "step": 4025
    },
    {
      "epoch": 0.3713527851458886,
      "eval_GEN Loss": 0.4669303297996521,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3980518877506256,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9057692289352417,
      "eval_runtime": 55.7951,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4025
    },
    {
      "epoch": 0.3714450467074155,
      "grad_norm": 1.6853666388182196,
      "learning_rate": 9.135279804603262e-07,
      "loss": 0.5899,
      "step": 4026
    },
    {
      "epoch": 0.37153730826894243,
      "grad_norm": 1.6027268662219922,
      "learning_rate": 9.133610693608866e-07,
      "loss": 0.7735,
      "step": 4027
    },
    {
      "epoch": 0.3716295698304694,
      "grad_norm": 3.3010178510482437,
      "learning_rate": 9.131941321297131e-07,
      "loss": 0.8266,
      "step": 4028
    },
    {
      "epoch": 0.3717218313919963,
      "grad_norm": 1.6191175013110048,
      "learning_rate": 9.130271687819338e-07,
      "loss": 0.7277,
      "step": 4029
    },
    {
      "epoch": 0.37181409295352325,
      "grad_norm": 3.9012333777277113,
      "learning_rate": 9.12860179332679e-07,
      "loss": 0.711,
      "step": 4030
    },
    {
      "epoch": 0.37181409295352325,
      "eval_GEN Loss": 0.46693354845046997,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.3991369307041168,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9065504670143127,
      "eval_runtime": 55.904,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4030
    },
    {
      "epoch": 0.37190635451505016,
      "grad_norm": 2.566631053448556,
      "learning_rate": 9.126931637970817e-07,
      "loss": 0.6825,
      "step": 4031
    },
    {
      "epoch": 0.3719986160765771,
      "grad_norm": 1.8342602799097993,
      "learning_rate": 9.12526122190277e-07,
      "loss": 0.5844,
      "step": 4032
    },
    {
      "epoch": 0.372090877638104,
      "grad_norm": 1.7245018786356503,
      "learning_rate": 9.123590545274028e-07,
      "loss": 0.6231,
      "step": 4033
    },
    {
      "epoch": 0.372183139199631,
      "grad_norm": 1.2882972868034825,
      "learning_rate": 9.121919608235984e-07,
      "loss": 0.6102,
      "step": 4034
    },
    {
      "epoch": 0.3722754007611579,
      "grad_norm": 2.4318892523648015,
      "learning_rate": 9.120248410940066e-07,
      "loss": 0.9743,
      "step": 4035
    },
    {
      "epoch": 0.3722754007611579,
      "eval_GEN Loss": 0.4661226272583008,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3899044096469879,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9005408883094788,
      "eval_runtime": 55.8406,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4035
    },
    {
      "epoch": 0.3723676623226848,
      "grad_norm": 1.4129661653841254,
      "learning_rate": 9.11857695353772e-07,
      "loss": 0.6628,
      "step": 4036
    },
    {
      "epoch": 0.37245992388421173,
      "grad_norm": 2.235675663639896,
      "learning_rate": 9.116905236180412e-07,
      "loss": 0.6884,
      "step": 4037
    },
    {
      "epoch": 0.37255218544573865,
      "grad_norm": 2.7216412016621345,
      "learning_rate": 9.115233259019643e-07,
      "loss": 1.123,
      "step": 4038
    },
    {
      "epoch": 0.3726444470072656,
      "grad_norm": 2.4642576140626895,
      "learning_rate": 9.113561022206924e-07,
      "loss": 0.828,
      "step": 4039
    },
    {
      "epoch": 0.37273670856879254,
      "grad_norm": 1.8259694350493174,
      "learning_rate": 9.111888525893799e-07,
      "loss": 0.6149,
      "step": 4040
    },
    {
      "epoch": 0.37273670856879254,
      "eval_GEN Loss": 0.46511706709861755,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.38533416390419006,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9043269157409668,
      "eval_runtime": 56.8252,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 4040
    },
    {
      "epoch": 0.37282897013031946,
      "grad_norm": 2.4773444919848977,
      "learning_rate": 9.110215770231829e-07,
      "loss": 0.7007,
      "step": 4041
    },
    {
      "epoch": 0.3729212316918464,
      "grad_norm": 1.7804524882857788,
      "learning_rate": 9.108542755372607e-07,
      "loss": 0.6445,
      "step": 4042
    },
    {
      "epoch": 0.3730134932533733,
      "grad_norm": 2.072669769630183,
      "learning_rate": 9.10686948146774e-07,
      "loss": 0.8563,
      "step": 4043
    },
    {
      "epoch": 0.3731057548149002,
      "grad_norm": 2.693136566311807,
      "learning_rate": 9.105195948668863e-07,
      "loss": 0.8057,
      "step": 4044
    },
    {
      "epoch": 0.3731980163764272,
      "grad_norm": 1.891017567824246,
      "learning_rate": 9.103522157127635e-07,
      "loss": 0.6851,
      "step": 4045
    },
    {
      "epoch": 0.3731980163764272,
      "eval_GEN Loss": 0.4658222496509552,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.3822549283504486,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.892848551273346,
      "eval_runtime": 56.6543,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 4045
    },
    {
      "epoch": 0.3732902779379541,
      "grad_norm": 1.649263312835569,
      "learning_rate": 9.101848106995738e-07,
      "loss": 0.6073,
      "step": 4046
    },
    {
      "epoch": 0.373382539499481,
      "grad_norm": 2.1316612619140054,
      "learning_rate": 9.100173798424877e-07,
      "loss": 0.6932,
      "step": 4047
    },
    {
      "epoch": 0.37347480106100794,
      "grad_norm": 1.9378131597038677,
      "learning_rate": 9.098499231566779e-07,
      "loss": 0.6467,
      "step": 4048
    },
    {
      "epoch": 0.37356706262253486,
      "grad_norm": 2.0166981219815403,
      "learning_rate": 9.096824406573199e-07,
      "loss": 0.6986,
      "step": 4049
    },
    {
      "epoch": 0.37365932418406184,
      "grad_norm": 1.3359555592984909,
      "learning_rate": 9.095149323595908e-07,
      "loss": 0.6596,
      "step": 4050
    },
    {
      "epoch": 0.37365932418406184,
      "eval_GEN Loss": 0.46569550037384033,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.375729501247406,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.887499988079071,
      "eval_runtime": 56.7661,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 4050
    },
    {
      "epoch": 0.37375158574558875,
      "grad_norm": 2.0655599871047152,
      "learning_rate": 9.093473982786708e-07,
      "loss": 0.8894,
      "step": 4051
    },
    {
      "epoch": 0.3738438473071157,
      "grad_norm": 2.774807538190179,
      "learning_rate": 9.091798384297419e-07,
      "loss": 0.9404,
      "step": 4052
    },
    {
      "epoch": 0.3739361088686426,
      "grad_norm": 2.466072038830056,
      "learning_rate": 9.090122528279888e-07,
      "loss": 0.7547,
      "step": 4053
    },
    {
      "epoch": 0.3740283704301695,
      "grad_norm": 2.0877020203222023,
      "learning_rate": 9.088446414885985e-07,
      "loss": 0.7237,
      "step": 4054
    },
    {
      "epoch": 0.3741206319916965,
      "grad_norm": 2.263149423748166,
      "learning_rate": 9.086770044267599e-07,
      "loss": 0.6551,
      "step": 4055
    },
    {
      "epoch": 0.3741206319916965,
      "eval_GEN Loss": 0.46623334288597107,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.37452223896980286,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8838942050933838,
      "eval_runtime": 56.7906,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 4055
    },
    {
      "epoch": 0.3742128935532234,
      "grad_norm": 1.950672893187621,
      "learning_rate": 9.085093416576644e-07,
      "loss": 0.5224,
      "step": 4056
    },
    {
      "epoch": 0.3743051551147503,
      "grad_norm": 1.5851451956760336,
      "learning_rate": 9.083416531965064e-07,
      "loss": 0.7405,
      "step": 4057
    },
    {
      "epoch": 0.37439741667627724,
      "grad_norm": 1.6121406269473215,
      "learning_rate": 9.081739390584818e-07,
      "loss": 0.6958,
      "step": 4058
    },
    {
      "epoch": 0.37448967823780416,
      "grad_norm": 1.9542873854807654,
      "learning_rate": 9.08006199258789e-07,
      "loss": 0.7563,
      "step": 4059
    },
    {
      "epoch": 0.3745819397993311,
      "grad_norm": 1.659723881225212,
      "learning_rate": 9.078384338126289e-07,
      "loss": 0.6368,
      "step": 4060
    },
    {
      "epoch": 0.3745819397993311,
      "eval_GEN Loss": 0.46463996171951294,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3785744309425354,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8866586685180664,
      "eval_runtime": 55.9519,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 4060
    },
    {
      "epoch": 0.37467420136085805,
      "grad_norm": 1.8295808467681853,
      "learning_rate": 9.076706427352048e-07,
      "loss": 0.5662,
      "step": 4061
    },
    {
      "epoch": 0.37476646292238497,
      "grad_norm": 3.345993918974571,
      "learning_rate": 9.075028260417221e-07,
      "loss": 0.8254,
      "step": 4062
    },
    {
      "epoch": 0.3748587244839119,
      "grad_norm": 1.5304809935686723,
      "learning_rate": 9.073349837473888e-07,
      "loss": 0.7073,
      "step": 4063
    },
    {
      "epoch": 0.3749509860454388,
      "grad_norm": 1.932737370444257,
      "learning_rate": 9.071671158674148e-07,
      "loss": 0.7327,
      "step": 4064
    },
    {
      "epoch": 0.3750432476069657,
      "grad_norm": 1.7632726646771022,
      "learning_rate": 9.069992224170125e-07,
      "loss": 0.6399,
      "step": 4065
    },
    {
      "epoch": 0.3750432476069657,
      "eval_GEN Loss": 0.46530988812446594,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.38563913106918335,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8853365182876587,
      "eval_runtime": 55.9535,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 4065
    },
    {
      "epoch": 0.3751355091684927,
      "grad_norm": 2.008937633115089,
      "learning_rate": 9.068313034113968e-07,
      "loss": 0.6322,
      "step": 4066
    },
    {
      "epoch": 0.3752277707300196,
      "grad_norm": 1.5091400584678478,
      "learning_rate": 9.06663358865785e-07,
      "loss": 0.59,
      "step": 4067
    },
    {
      "epoch": 0.37532003229154653,
      "grad_norm": 1.9643058749280151,
      "learning_rate": 9.064953887953961e-07,
      "loss": 0.7106,
      "step": 4068
    },
    {
      "epoch": 0.37541229385307345,
      "grad_norm": 1.4066710687651958,
      "learning_rate": 9.06327393215452e-07,
      "loss": 0.4696,
      "step": 4069
    },
    {
      "epoch": 0.37550455541460037,
      "grad_norm": 2.515416684612912,
      "learning_rate": 9.061593721411767e-07,
      "loss": 0.7084,
      "step": 4070
    },
    {
      "epoch": 0.37550455541460037,
      "eval_GEN Loss": 0.4660487771034241,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.36429563164711,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8866586685180664,
      "eval_runtime": 55.7371,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4070
    },
    {
      "epoch": 0.37559681697612735,
      "grad_norm": 7.281703882839622,
      "learning_rate": 9.059913255877966e-07,
      "loss": 0.8893,
      "step": 4071
    },
    {
      "epoch": 0.37568907853765426,
      "grad_norm": 1.9450188003044784,
      "learning_rate": 9.058232535705403e-07,
      "loss": 0.8208,
      "step": 4072
    },
    {
      "epoch": 0.3757813400991812,
      "grad_norm": 1.5293624406192017,
      "learning_rate": 9.056551561046387e-07,
      "loss": 0.7075,
      "step": 4073
    },
    {
      "epoch": 0.3758736016607081,
      "grad_norm": 2.7747055999034633,
      "learning_rate": 9.05487033205325e-07,
      "loss": 0.6516,
      "step": 4074
    },
    {
      "epoch": 0.375965863222235,
      "grad_norm": 1.841774860399856,
      "learning_rate": 9.05318884887835e-07,
      "loss": 0.6616,
      "step": 4075
    },
    {
      "epoch": 0.375965863222235,
      "eval_GEN Loss": 0.46514856815338135,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.3565382659435272,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8983774185180664,
      "eval_runtime": 56.7211,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4075
    },
    {
      "epoch": 0.37605812478376194,
      "grad_norm": 2.144210112325349,
      "learning_rate": 9.051507111674063e-07,
      "loss": 0.8158,
      "step": 4076
    },
    {
      "epoch": 0.3761503863452889,
      "grad_norm": 2.20490626849351,
      "learning_rate": 9.049825120592792e-07,
      "loss": 0.8346,
      "step": 4077
    },
    {
      "epoch": 0.37624264790681583,
      "grad_norm": 1.397526771085783,
      "learning_rate": 9.048142875786962e-07,
      "loss": 0.6113,
      "step": 4078
    },
    {
      "epoch": 0.37633490946834275,
      "grad_norm": 1.6255512559752765,
      "learning_rate": 9.046460377409021e-07,
      "loss": 0.5746,
      "step": 4079
    },
    {
      "epoch": 0.37642717102986967,
      "grad_norm": 2.1719524743768552,
      "learning_rate": 9.04477762561144e-07,
      "loss": 0.7364,
      "step": 4080
    },
    {
      "epoch": 0.37642717102986967,
      "eval_GEN Loss": 0.4640125334262848,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.3585696220397949,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8894230723381042,
      "eval_runtime": 55.8127,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4080
    },
    {
      "epoch": 0.3765194325913966,
      "grad_norm": 1.7551920488598083,
      "learning_rate": 9.04309462054671e-07,
      "loss": 0.6205,
      "step": 4081
    },
    {
      "epoch": 0.37661169415292356,
      "grad_norm": 1.841775913433435,
      "learning_rate": 9.041411362367347e-07,
      "loss": 0.6243,
      "step": 4082
    },
    {
      "epoch": 0.3767039557144505,
      "grad_norm": 2.609938241896056,
      "learning_rate": 9.039727851225896e-07,
      "loss": 1.1076,
      "step": 4083
    },
    {
      "epoch": 0.3767962172759774,
      "grad_norm": 2.012470815006629,
      "learning_rate": 9.038044087274913e-07,
      "loss": 0.5505,
      "step": 4084
    },
    {
      "epoch": 0.3768884788375043,
      "grad_norm": 3.057574371135215,
      "learning_rate": 9.036360070666988e-07,
      "loss": 0.8766,
      "step": 4085
    },
    {
      "epoch": 0.3768884788375043,
      "eval_GEN Loss": 0.4655226469039917,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3777284324169159,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8847355842590332,
      "eval_runtime": 55.8848,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4085
    },
    {
      "epoch": 0.37698074039903123,
      "grad_norm": 1.2289862852034075,
      "learning_rate": 9.034675801554728e-07,
      "loss": 0.5911,
      "step": 4086
    },
    {
      "epoch": 0.3770730019605582,
      "grad_norm": 2.1253223280119684,
      "learning_rate": 9.032991280090763e-07,
      "loss": 0.9118,
      "step": 4087
    },
    {
      "epoch": 0.3771652635220851,
      "grad_norm": 1.580319589302749,
      "learning_rate": 9.031306506427748e-07,
      "loss": 0.7123,
      "step": 4088
    },
    {
      "epoch": 0.37725752508361204,
      "grad_norm": 2.0879649675225225,
      "learning_rate": 9.029621480718359e-07,
      "loss": 0.7451,
      "step": 4089
    },
    {
      "epoch": 0.37734978664513896,
      "grad_norm": 1.8203296833292217,
      "learning_rate": 9.027936203115298e-07,
      "loss": 0.6666,
      "step": 4090
    },
    {
      "epoch": 0.37734978664513896,
      "eval_GEN Loss": 0.4692361354827881,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.7830188679245284,
      "eval_PRM F1": 0.8535031847133758,
      "eval_PRM F1 AUC": 0.751440544787847,
      "eval_PRM F1 Neg": 0.5818181818181818,
      "eval_PRM Loss": 0.4262109398841858,
      "eval_PRM NPV": 0.5,
      "eval_PRM Precision": 0.9054054054054054,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9024038314819336,
      "eval_runtime": 55.7702,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4090
    },
    {
      "epoch": 0.3774420482066659,
      "grad_norm": 1.7255796283932037,
      "learning_rate": 9.026250673771285e-07,
      "loss": 0.6584,
      "step": 4091
    },
    {
      "epoch": 0.3775343097681928,
      "grad_norm": 1.8623634582837343,
      "learning_rate": 9.024564892839065e-07,
      "loss": 0.8439,
      "step": 4092
    },
    {
      "epoch": 0.3776265713297198,
      "grad_norm": 1.4866862015032798,
      "learning_rate": 9.022878860471408e-07,
      "loss": 0.5778,
      "step": 4093
    },
    {
      "epoch": 0.3777188328912467,
      "grad_norm": 1.9813739983608938,
      "learning_rate": 9.021192576821103e-07,
      "loss": 0.6114,
      "step": 4094
    },
    {
      "epoch": 0.3778110944527736,
      "grad_norm": 3.0584135468221385,
      "learning_rate": 9.019506042040965e-07,
      "loss": 0.6117,
      "step": 4095
    },
    {
      "epoch": 0.3778110944527736,
      "eval_GEN Loss": 0.4708370268344879,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.7641509433962265,
      "eval_PRM F1": 0.8366013071895425,
      "eval_PRM F1 AUC": 0.7551073860660031,
      "eval_PRM F1 Neg": 0.576271186440678,
      "eval_PRM Loss": 0.4815226197242737,
      "eval_PRM NPV": 0.4722222222222222,
      "eval_PRM Precision": 0.9142857142857143,
      "eval_PRM Recall": 0.7710843373493976,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9449519515037537,
      "eval_runtime": 55.6971,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 4095
    },
    {
      "epoch": 0.37790335601430053,
      "grad_norm": 2.525534773816227,
      "learning_rate": 9.017819256283831e-07,
      "loss": 0.7888,
      "step": 4096
    },
    {
      "epoch": 0.37799561757582745,
      "grad_norm": 1.6094735277602292,
      "learning_rate": 9.01613221970256e-07,
      "loss": 0.6046,
      "step": 4097
    },
    {
      "epoch": 0.3780878791373544,
      "grad_norm": 1.9296375765041347,
      "learning_rate": 9.014444932450031e-07,
      "loss": 0.7918,
      "step": 4098
    },
    {
      "epoch": 0.37818014069888134,
      "grad_norm": 1.7958187386170053,
      "learning_rate": 9.012757394679153e-07,
      "loss": 0.6623,
      "step": 4099
    },
    {
      "epoch": 0.37827240226040826,
      "grad_norm": 1.8015992592485275,
      "learning_rate": 9.011069606542851e-07,
      "loss": 0.697,
      "step": 4100
    },
    {
      "epoch": 0.37827240226040826,
      "eval_GEN Loss": 0.47058019042015076,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7735849056603774,
      "eval_PRM F1": 0.8441558441558441,
      "eval_PRM F1 AUC": 0.7611314824515453,
      "eval_PRM F1 Neg": 0.5862068965517241,
      "eval_PRM Loss": 0.46750107407569885,
      "eval_PRM NPV": 0.4857142857142857,
      "eval_PRM Precision": 0.9154929577464789,
      "eval_PRM Recall": 0.7831325301204819,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9304086565971375,
      "eval_runtime": 55.7299,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4100
    },
    {
      "epoch": 0.3783646638219352,
      "grad_norm": 1.7976359852623367,
      "learning_rate": 9.009381568194076e-07,
      "loss": 0.688,
      "step": 4101
    },
    {
      "epoch": 0.3784569253834621,
      "grad_norm": 2.14261377974147,
      "learning_rate": 9.007693279785798e-07,
      "loss": 0.6121,
      "step": 4102
    },
    {
      "epoch": 0.37854918694498907,
      "grad_norm": 2.1940832758947146,
      "learning_rate": 9.006004741471015e-07,
      "loss": 0.8052,
      "step": 4103
    },
    {
      "epoch": 0.378641448506516,
      "grad_norm": 2.102796633400336,
      "learning_rate": 9.004315953402746e-07,
      "loss": 0.8437,
      "step": 4104
    },
    {
      "epoch": 0.3787337100680429,
      "grad_norm": 1.6218420515095666,
      "learning_rate": 9.002626915734026e-07,
      "loss": 0.518,
      "step": 4105
    },
    {
      "epoch": 0.3787337100680429,
      "eval_GEN Loss": 0.47079312801361084,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.7924528301886793,
      "eval_PRM F1": 0.8589743589743589,
      "eval_PRM F1 AUC": 0.7731796752226295,
      "eval_PRM F1 Neg": 0.6071428571428571,
      "eval_PRM Loss": 0.4295346140861511,
      "eval_PRM NPV": 0.5151515151515151,
      "eval_PRM Precision": 0.9178082191780822,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9019832015037537,
      "eval_runtime": 56.0146,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 4105
    },
    {
      "epoch": 0.3788259716295698,
      "grad_norm": 2.400523109204971,
      "learning_rate": 9.000937628617926e-07,
      "loss": 0.8537,
      "step": 4106
    },
    {
      "epoch": 0.37891823319109674,
      "grad_norm": 1.8319235433145722,
      "learning_rate": 8.999248092207526e-07,
      "loss": 0.7021,
      "step": 4107
    },
    {
      "epoch": 0.37901049475262366,
      "grad_norm": 1.986099336811533,
      "learning_rate": 8.997558306655937e-07,
      "loss": 0.7189,
      "step": 4108
    },
    {
      "epoch": 0.37910275631415064,
      "grad_norm": 2.403008577479829,
      "learning_rate": 8.99586827211629e-07,
      "loss": 0.9662,
      "step": 4109
    },
    {
      "epoch": 0.37919501787567755,
      "grad_norm": 1.9345838871761694,
      "learning_rate": 8.994177988741737e-07,
      "loss": 0.6802,
      "step": 4110
    },
    {
      "epoch": 0.37919501787567755,
      "eval_GEN Loss": 0.4700886607170105,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.3806597888469696,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8706730604171753,
      "eval_runtime": 55.8094,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4110
    },
    {
      "epoch": 0.37928727943720447,
      "grad_norm": 2.9720549147061597,
      "learning_rate": 8.992487456685458e-07,
      "loss": 0.9286,
      "step": 4111
    },
    {
      "epoch": 0.3793795409987314,
      "grad_norm": 1.5057151845691115,
      "learning_rate": 8.990796676100645e-07,
      "loss": 0.4774,
      "step": 4112
    },
    {
      "epoch": 0.3794718025602583,
      "grad_norm": 2.4999280095476464,
      "learning_rate": 8.989105647140526e-07,
      "loss": 0.8233,
      "step": 4113
    },
    {
      "epoch": 0.3795640641217853,
      "grad_norm": 2.237538056428383,
      "learning_rate": 8.98741436995834e-07,
      "loss": 0.7955,
      "step": 4114
    },
    {
      "epoch": 0.3796563256833122,
      "grad_norm": 2.5053640123480925,
      "learning_rate": 8.985722844707356e-07,
      "loss": 0.6693,
      "step": 4115
    },
    {
      "epoch": 0.3796563256833122,
      "eval_GEN Loss": 0.46913740038871765,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.3650398552417755,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8587139248847961,
      "eval_runtime": 55.7723,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4115
    },
    {
      "epoch": 0.3797485872448391,
      "grad_norm": 1.9776617660935796,
      "learning_rate": 8.984031071540864e-07,
      "loss": 0.7515,
      "step": 4116
    },
    {
      "epoch": 0.37984084880636604,
      "grad_norm": 2.2762681584190285,
      "learning_rate": 8.982339050612171e-07,
      "loss": 0.6703,
      "step": 4117
    },
    {
      "epoch": 0.37993311036789296,
      "grad_norm": 2.202246794691524,
      "learning_rate": 8.98064678207461e-07,
      "loss": 0.8285,
      "step": 4118
    },
    {
      "epoch": 0.38002537192941993,
      "grad_norm": 2.3502799803474153,
      "learning_rate": 8.978954266081541e-07,
      "loss": 0.6768,
      "step": 4119
    },
    {
      "epoch": 0.38011763349094685,
      "grad_norm": 3.0643027733717756,
      "learning_rate": 8.977261502786342e-07,
      "loss": 0.7997,
      "step": 4120
    },
    {
      "epoch": 0.38011763349094685,
      "eval_GEN Loss": 0.47055891156196594,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.379611611366272,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8648437261581421,
      "eval_runtime": 55.7657,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4120
    },
    {
      "epoch": 0.38020989505247377,
      "grad_norm": 2.1003669521683968,
      "learning_rate": 8.975568492342412e-07,
      "loss": 0.7221,
      "step": 4121
    },
    {
      "epoch": 0.3803021566140007,
      "grad_norm": 1.5552526360183851,
      "learning_rate": 8.973875234903177e-07,
      "loss": 0.6073,
      "step": 4122
    },
    {
      "epoch": 0.3803944181755276,
      "grad_norm": 1.7268954960628977,
      "learning_rate": 8.972181730622079e-07,
      "loss": 0.6531,
      "step": 4123
    },
    {
      "epoch": 0.3804866797370545,
      "grad_norm": 1.693727855069528,
      "learning_rate": 8.970487979652588e-07,
      "loss": 0.5369,
      "step": 4124
    },
    {
      "epoch": 0.3805789412985815,
      "grad_norm": 1.507390905963885,
      "learning_rate": 8.968793982148194e-07,
      "loss": 0.5774,
      "step": 4125
    },
    {
      "epoch": 0.3805789412985815,
      "eval_GEN Loss": 0.4713898301124573,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.3831806182861328,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8662259578704834,
      "eval_runtime": 55.8614,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4125
    },
    {
      "epoch": 0.3806712028601084,
      "grad_norm": 2.209484094137634,
      "learning_rate": 8.96709973826241e-07,
      "loss": 0.6745,
      "step": 4126
    },
    {
      "epoch": 0.38076346442163533,
      "grad_norm": 2.186873075109473,
      "learning_rate": 8.965405248148772e-07,
      "loss": 0.4689,
      "step": 4127
    },
    {
      "epoch": 0.38085572598316225,
      "grad_norm": 2.6099540206379634,
      "learning_rate": 8.963710511960835e-07,
      "loss": 0.7421,
      "step": 4128
    },
    {
      "epoch": 0.38094798754468917,
      "grad_norm": 1.9645379617710306,
      "learning_rate": 8.96201552985218e-07,
      "loss": 0.8676,
      "step": 4129
    },
    {
      "epoch": 0.38104024910621614,
      "grad_norm": 1.6447165659097023,
      "learning_rate": 8.96032030197641e-07,
      "loss": 0.5708,
      "step": 4130
    },
    {
      "epoch": 0.38104024910621614,
      "eval_GEN Loss": 0.4711868166923523,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM Loss": 0.3539343774318695,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8549879789352417,
      "eval_runtime": 55.9042,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4130
    },
    {
      "epoch": 0.38113251066774306,
      "grad_norm": 1.520369438128281,
      "learning_rate": 8.958624828487148e-07,
      "loss": 0.7368,
      "step": 4131
    },
    {
      "epoch": 0.38122477222927,
      "grad_norm": 2.2056337831280066,
      "learning_rate": 8.956929109538038e-07,
      "loss": 0.8261,
      "step": 4132
    },
    {
      "epoch": 0.3813170337907969,
      "grad_norm": 1.4795051528939969,
      "learning_rate": 8.955233145282754e-07,
      "loss": 0.7209,
      "step": 4133
    },
    {
      "epoch": 0.3814092953523238,
      "grad_norm": 1.367917648382338,
      "learning_rate": 8.953536935874983e-07,
      "loss": 0.7128,
      "step": 4134
    },
    {
      "epoch": 0.3815015569138508,
      "grad_norm": 1.8337092081931965,
      "learning_rate": 8.95184048146844e-07,
      "loss": 0.8696,
      "step": 4135
    },
    {
      "epoch": 0.3815015569138508,
      "eval_GEN Loss": 0.47096556425094604,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3369731903076172,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.854687511920929,
      "eval_runtime": 55.9978,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 4135
    },
    {
      "epoch": 0.3815938184753777,
      "grad_norm": 1.6270411670323788,
      "learning_rate": 8.950143782216861e-07,
      "loss": 0.6367,
      "step": 4136
    },
    {
      "epoch": 0.38168608003690463,
      "grad_norm": 2.613811982352718,
      "learning_rate": 8.948446838274e-07,
      "loss": 0.8066,
      "step": 4137
    },
    {
      "epoch": 0.38177834159843155,
      "grad_norm": 2.4712859869770427,
      "learning_rate": 8.94674964979364e-07,
      "loss": 0.725,
      "step": 4138
    },
    {
      "epoch": 0.38187060315995847,
      "grad_norm": 1.2268644952870233,
      "learning_rate": 8.945052216929584e-07,
      "loss": 0.5812,
      "step": 4139
    },
    {
      "epoch": 0.3819628647214854,
      "grad_norm": 2.153654454794922,
      "learning_rate": 8.943354539835653e-07,
      "loss": 0.8481,
      "step": 4140
    },
    {
      "epoch": 0.3819628647214854,
      "eval_GEN Loss": 0.4700216054916382,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3401414155960083,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8580529093742371,
      "eval_runtime": 56.0493,
      "eval_samples_per_second": 1.16,
      "eval_steps_per_second": 0.089,
      "step": 4140
    },
    {
      "epoch": 0.38205512628301236,
      "grad_norm": 2.980288290905818,
      "learning_rate": 8.941656618665692e-07,
      "loss": 0.7908,
      "step": 4141
    },
    {
      "epoch": 0.3821473878445393,
      "grad_norm": 2.176296309902741,
      "learning_rate": 8.939958453573573e-07,
      "loss": 0.7233,
      "step": 4142
    },
    {
      "epoch": 0.3822396494060662,
      "grad_norm": 1.7614135698499784,
      "learning_rate": 8.938260044713185e-07,
      "loss": 0.7508,
      "step": 4143
    },
    {
      "epoch": 0.3823319109675931,
      "grad_norm": 1.6930084883845857,
      "learning_rate": 8.93656139223844e-07,
      "loss": 0.7988,
      "step": 4144
    },
    {
      "epoch": 0.38242417252912003,
      "grad_norm": 1.6486405498598211,
      "learning_rate": 8.934862496303273e-07,
      "loss": 0.851,
      "step": 4145
    },
    {
      "epoch": 0.38242417252912003,
      "eval_GEN Loss": 0.4703010022640228,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8773584905660378,
      "eval_PRM F1": 0.9212121212121213,
      "eval_PRM F1 AUC": 0.8273965426925091,
      "eval_PRM F1 Neg": 0.723404255319149,
      "eval_PRM Loss": 0.35920852422714233,
      "eval_PRM NPV": 0.7083333333333334,
      "eval_PRM Precision": 0.926829268292683,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8649038672447205,
      "eval_runtime": 55.7304,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4145
    },
    {
      "epoch": 0.382516434090647,
      "grad_norm": 1.9797619992575017,
      "learning_rate": 8.933163357061641e-07,
      "loss": 0.7228,
      "step": 4146
    },
    {
      "epoch": 0.3826086956521739,
      "grad_norm": 2.793289457320529,
      "learning_rate": 8.931463974667524e-07,
      "loss": 0.7135,
      "step": 4147
    },
    {
      "epoch": 0.38270095721370084,
      "grad_norm": 2.006307614165737,
      "learning_rate": 8.929764349274917e-07,
      "loss": 0.6977,
      "step": 4148
    },
    {
      "epoch": 0.38279321877522776,
      "grad_norm": 1.911785679074507,
      "learning_rate": 8.928064481037849e-07,
      "loss": 0.6887,
      "step": 4149
    },
    {
      "epoch": 0.3828854803367547,
      "grad_norm": 1.516171077232002,
      "learning_rate": 8.926364370110361e-07,
      "loss": 0.6743,
      "step": 4150
    },
    {
      "epoch": 0.3828854803367547,
      "eval_GEN Loss": 0.4698607325553894,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.3694307208061218,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8732572197914124,
      "eval_runtime": 56.5443,
      "eval_samples_per_second": 1.15,
      "eval_steps_per_second": 0.088,
      "step": 4150
    },
    {
      "epoch": 0.38297774189828165,
      "grad_norm": 1.5397195893431548,
      "learning_rate": 8.924664016646523e-07,
      "loss": 0.7933,
      "step": 4151
    },
    {
      "epoch": 0.3830700034598086,
      "grad_norm": 2.1903763474763815,
      "learning_rate": 8.922963420800421e-07,
      "loss": 0.8532,
      "step": 4152
    },
    {
      "epoch": 0.3831622650213355,
      "grad_norm": 2.8023344938705628,
      "learning_rate": 8.921262582726166e-07,
      "loss": 0.8403,
      "step": 4153
    },
    {
      "epoch": 0.3832545265828624,
      "grad_norm": 2.078904830604637,
      "learning_rate": 8.919561502577892e-07,
      "loss": 0.7517,
      "step": 4154
    },
    {
      "epoch": 0.3833467881443893,
      "grad_norm": 2.189586301332461,
      "learning_rate": 8.917860180509752e-07,
      "loss": 0.8015,
      "step": 4155
    },
    {
      "epoch": 0.3833467881443893,
      "eval_GEN Loss": 0.46900999546051025,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.38497164845466614,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8823317289352417,
      "eval_runtime": 55.8541,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4155
    },
    {
      "epoch": 0.38343904970591625,
      "grad_norm": 2.0327216013773035,
      "learning_rate": 8.916158616675924e-07,
      "loss": 0.7306,
      "step": 4156
    },
    {
      "epoch": 0.3835313112674432,
      "grad_norm": 2.107789450922101,
      "learning_rate": 8.914456811230604e-07,
      "loss": 0.6397,
      "step": 4157
    },
    {
      "epoch": 0.38362357282897014,
      "grad_norm": 2.73424299657904,
      "learning_rate": 8.912754764328014e-07,
      "loss": 0.8332,
      "step": 4158
    },
    {
      "epoch": 0.38371583439049706,
      "grad_norm": 2.7737737665339637,
      "learning_rate": 8.911052476122398e-07,
      "loss": 0.8803,
      "step": 4159
    },
    {
      "epoch": 0.383808095952024,
      "grad_norm": 1.9201754705277472,
      "learning_rate": 8.909349946768015e-07,
      "loss": 0.7676,
      "step": 4160
    },
    {
      "epoch": 0.383808095952024,
      "eval_GEN Loss": 0.4681911766529083,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.3734659254550934,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8832331895828247,
      "eval_runtime": 55.866,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4160
    },
    {
      "epoch": 0.3839003575135509,
      "grad_norm": 2.369577124317477,
      "learning_rate": 8.907647176419157e-07,
      "loss": 0.814,
      "step": 4161
    },
    {
      "epoch": 0.38399261907507787,
      "grad_norm": 2.2861983181404213,
      "learning_rate": 8.905944165230127e-07,
      "loss": 0.7954,
      "step": 4162
    },
    {
      "epoch": 0.3840848806366048,
      "grad_norm": 2.3950837743453772,
      "learning_rate": 8.904240913355259e-07,
      "loss": 0.7816,
      "step": 4163
    },
    {
      "epoch": 0.3841771421981317,
      "grad_norm": 2.5310327958778487,
      "learning_rate": 8.902537420948898e-07,
      "loss": 0.8308,
      "step": 4164
    },
    {
      "epoch": 0.3842694037596586,
      "grad_norm": 2.6523252994872504,
      "learning_rate": 8.900833688165423e-07,
      "loss": 0.971,
      "step": 4165
    },
    {
      "epoch": 0.3842694037596586,
      "eval_GEN Loss": 0.4663677215576172,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.36391881108283997,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8916466236114502,
      "eval_runtime": 55.8576,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4165
    },
    {
      "epoch": 0.38436166532118554,
      "grad_norm": 1.724650213178172,
      "learning_rate": 8.899129715159225e-07,
      "loss": 0.7285,
      "step": 4166
    },
    {
      "epoch": 0.3844539268827125,
      "grad_norm": 1.6841176991837141,
      "learning_rate": 8.897425502084722e-07,
      "loss": 0.7494,
      "step": 4167
    },
    {
      "epoch": 0.38454618844423943,
      "grad_norm": 1.9759139217235067,
      "learning_rate": 8.895721049096354e-07,
      "loss": 0.7731,
      "step": 4168
    },
    {
      "epoch": 0.38463845000576635,
      "grad_norm": 1.7685707146533716,
      "learning_rate": 8.89401635634858e-07,
      "loss": 0.6805,
      "step": 4169
    },
    {
      "epoch": 0.38473071156729327,
      "grad_norm": 1.545108224647599,
      "learning_rate": 8.892311423995882e-07,
      "loss": 0.7398,
      "step": 4170
    },
    {
      "epoch": 0.38473071156729327,
      "eval_GEN Loss": 0.4657104015350342,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.36361831426620483,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8933293223381042,
      "eval_runtime": 55.8136,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4170
    },
    {
      "epoch": 0.3848229731288202,
      "grad_norm": 1.5057504669971054,
      "learning_rate": 8.890606252192764e-07,
      "loss": 0.5024,
      "step": 4171
    },
    {
      "epoch": 0.3849152346903471,
      "grad_norm": 2.081063713196889,
      "learning_rate": 8.88890084109375e-07,
      "loss": 0.7642,
      "step": 4172
    },
    {
      "epoch": 0.3850074962518741,
      "grad_norm": 2.046901548970188,
      "learning_rate": 8.88719519085339e-07,
      "loss": 0.6805,
      "step": 4173
    },
    {
      "epoch": 0.385099757813401,
      "grad_norm": 1.7701919851806927,
      "learning_rate": 8.88548930162625e-07,
      "loss": 0.7638,
      "step": 4174
    },
    {
      "epoch": 0.3851920193749279,
      "grad_norm": 2.263116391467018,
      "learning_rate": 8.883783173566922e-07,
      "loss": 0.7666,
      "step": 4175
    },
    {
      "epoch": 0.3851920193749279,
      "eval_GEN Loss": 0.46534600853919983,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9101796407185628,
      "eval_PRM F1 AUC": 0.7839182818229439,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3683016896247864,
      "eval_PRM NPV": 0.6818181818181818,
      "eval_PRM Precision": 0.9047619047619048,
      "eval_PRM Recall": 0.9156626506024096,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8954927921295166,
      "eval_runtime": 55.9185,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 4175
    },
    {
      "epoch": 0.38528428093645484,
      "grad_norm": 2.207970619843975,
      "learning_rate": 8.882076806830015e-07,
      "loss": 0.5522,
      "step": 4176
    },
    {
      "epoch": 0.38537654249798176,
      "grad_norm": 2.4600937725481153,
      "learning_rate": 8.880370201570167e-07,
      "loss": 0.7578,
      "step": 4177
    },
    {
      "epoch": 0.38546880405950873,
      "grad_norm": 1.815296054932152,
      "learning_rate": 8.87866335794203e-07,
      "loss": 0.7262,
      "step": 4178
    },
    {
      "epoch": 0.38556106562103565,
      "grad_norm": 1.9574633905051928,
      "learning_rate": 8.876956276100282e-07,
      "loss": 0.6068,
      "step": 4179
    },
    {
      "epoch": 0.38565332718256257,
      "grad_norm": 2.4599120766660088,
      "learning_rate": 8.87524895619962e-07,
      "loss": 0.7273,
      "step": 4180
    },
    {
      "epoch": 0.38565332718256257,
      "eval_GEN Loss": 0.46767744421958923,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3833829462528229,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9011418223381042,
      "eval_runtime": 56.0967,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 4180
    },
    {
      "epoch": 0.3857455887440895,
      "grad_norm": 1.7663893952240763,
      "learning_rate": 8.873541398394768e-07,
      "loss": 0.7064,
      "step": 4181
    },
    {
      "epoch": 0.3858378503056164,
      "grad_norm": 1.7272816864428557,
      "learning_rate": 8.871833602840463e-07,
      "loss": 0.6884,
      "step": 4182
    },
    {
      "epoch": 0.3859301118671434,
      "grad_norm": 1.7081083896742395,
      "learning_rate": 8.870125569691471e-07,
      "loss": 0.6074,
      "step": 4183
    },
    {
      "epoch": 0.3860223734286703,
      "grad_norm": 1.9922988697969741,
      "learning_rate": 8.868417299102575e-07,
      "loss": 0.722,
      "step": 4184
    },
    {
      "epoch": 0.3861146349901972,
      "grad_norm": 2.414018439640081,
      "learning_rate": 8.866708791228583e-07,
      "loss": 0.8671,
      "step": 4185
    },
    {
      "epoch": 0.3861146349901972,
      "eval_GEN Loss": 0.4676385223865509,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3864862620830536,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9075120091438293,
      "eval_runtime": 56.0006,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 4185
    },
    {
      "epoch": 0.38620689655172413,
      "grad_norm": 2.2164509051288137,
      "learning_rate": 8.86500004622432e-07,
      "loss": 0.5784,
      "step": 4186
    },
    {
      "epoch": 0.38629915811325105,
      "grad_norm": 1.7056810662656736,
      "learning_rate": 8.863291064244639e-07,
      "loss": 0.5829,
      "step": 4187
    },
    {
      "epoch": 0.38639141967477797,
      "grad_norm": 2.400412623022726,
      "learning_rate": 8.861581845444407e-07,
      "loss": 0.8468,
      "step": 4188
    },
    {
      "epoch": 0.38648368123630494,
      "grad_norm": 1.9693845638268446,
      "learning_rate": 8.859872389978518e-07,
      "loss": 0.7192,
      "step": 4189
    },
    {
      "epoch": 0.38657594279783186,
      "grad_norm": 1.9040475462580282,
      "learning_rate": 8.858162698001885e-07,
      "loss": 0.728,
      "step": 4190
    },
    {
      "epoch": 0.38657594279783186,
      "eval_GEN Loss": 0.4676743149757385,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.38475629687309265,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9117187261581421,
      "eval_runtime": 55.7488,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4190
    },
    {
      "epoch": 0.3866682043593588,
      "grad_norm": 1.9798730968837608,
      "learning_rate": 8.856452769669441e-07,
      "loss": 0.8242,
      "step": 4191
    },
    {
      "epoch": 0.3867604659208857,
      "grad_norm": 1.77718143310619,
      "learning_rate": 8.854742605136147e-07,
      "loss": 0.878,
      "step": 4192
    },
    {
      "epoch": 0.3868527274824126,
      "grad_norm": 2.41595112775984,
      "learning_rate": 8.853032204556977e-07,
      "loss": 0.5646,
      "step": 4193
    },
    {
      "epoch": 0.3869449890439396,
      "grad_norm": 2.116837203567898,
      "learning_rate": 8.85132156808693e-07,
      "loss": 0.7877,
      "step": 4194
    },
    {
      "epoch": 0.3870372506054665,
      "grad_norm": 1.9309305362091516,
      "learning_rate": 8.84961069588103e-07,
      "loss": 0.7074,
      "step": 4195
    },
    {
      "epoch": 0.3870372506054665,
      "eval_GEN Loss": 0.4683523178100586,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.381059855222702,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9173678159713745,
      "eval_runtime": 55.7508,
      "eval_samples_per_second": 1.166,
      "eval_steps_per_second": 0.09,
      "step": 4195
    },
    {
      "epoch": 0.38712951216699343,
      "grad_norm": 1.9911867343954848,
      "learning_rate": 8.847899588094316e-07,
      "loss": 0.6731,
      "step": 4196
    },
    {
      "epoch": 0.38722177372852035,
      "grad_norm": 2.8252080754802202,
      "learning_rate": 8.846188244881854e-07,
      "loss": 1.019,
      "step": 4197
    },
    {
      "epoch": 0.38731403529004726,
      "grad_norm": 1.9017921564465072,
      "learning_rate": 8.844476666398726e-07,
      "loss": 0.6573,
      "step": 4198
    },
    {
      "epoch": 0.38740629685157424,
      "grad_norm": 1.648562886711212,
      "learning_rate": 8.842764852800037e-07,
      "loss": 0.7175,
      "step": 4199
    },
    {
      "epoch": 0.38749855841310116,
      "grad_norm": 1.899182383830107,
      "learning_rate": 8.841052804240918e-07,
      "loss": 0.6292,
      "step": 4200
    },
    {
      "epoch": 0.38749855841310116,
      "eval_GEN Loss": 0.4674999713897705,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.38112860918045044,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9100961685180664,
      "eval_runtime": 55.9871,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 4200
    },
    {
      "epoch": 0.3875908199746281,
      "grad_norm": 3.086778704607036,
      "learning_rate": 8.839340520876513e-07,
      "loss": 0.9341,
      "step": 4201
    },
    {
      "epoch": 0.387683081536155,
      "grad_norm": 1.420937228764487,
      "learning_rate": 8.837628002861998e-07,
      "loss": 0.6994,
      "step": 4202
    },
    {
      "epoch": 0.3877753430976819,
      "grad_norm": 1.621592627710753,
      "learning_rate": 8.835915250352557e-07,
      "loss": 0.6528,
      "step": 4203
    },
    {
      "epoch": 0.38786760465920883,
      "grad_norm": 1.815620082216712,
      "learning_rate": 8.834202263503409e-07,
      "loss": 0.7354,
      "step": 4204
    },
    {
      "epoch": 0.3879598662207358,
      "grad_norm": 2.5642367575220777,
      "learning_rate": 8.83248904246978e-07,
      "loss": 0.6432,
      "step": 4205
    },
    {
      "epoch": 0.3879598662207358,
      "eval_GEN Loss": 0.46722376346588135,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3848364055156708,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9116586446762085,
      "eval_runtime": 55.8921,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4205
    },
    {
      "epoch": 0.3880521277822627,
      "grad_norm": 2.057249050716057,
      "learning_rate": 8.830775587406932e-07,
      "loss": 0.7836,
      "step": 4206
    },
    {
      "epoch": 0.38814438934378964,
      "grad_norm": 3.066725636991236,
      "learning_rate": 8.829061898470137e-07,
      "loss": 0.9165,
      "step": 4207
    },
    {
      "epoch": 0.38823665090531656,
      "grad_norm": 2.269660349015463,
      "learning_rate": 8.827347975814693e-07,
      "loss": 0.8813,
      "step": 4208
    },
    {
      "epoch": 0.3883289124668435,
      "grad_norm": 1.8156204829884148,
      "learning_rate": 8.82563381959592e-07,
      "loss": 0.5474,
      "step": 4209
    },
    {
      "epoch": 0.38842117402837045,
      "grad_norm": 1.4701347270248641,
      "learning_rate": 8.823919429969156e-07,
      "loss": 0.7296,
      "step": 4210
    },
    {
      "epoch": 0.38842117402837045,
      "eval_GEN Loss": 0.46739426255226135,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.4045126438140869,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.920192301273346,
      "eval_runtime": 55.82,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4210
    },
    {
      "epoch": 0.38851343558989737,
      "grad_norm": 2.4347157445639453,
      "learning_rate": 8.822204807089763e-07,
      "loss": 0.5665,
      "step": 4211
    },
    {
      "epoch": 0.3886056971514243,
      "grad_norm": 1.6792201907571038,
      "learning_rate": 8.820489951113117e-07,
      "loss": 0.8638,
      "step": 4212
    },
    {
      "epoch": 0.3886979587129512,
      "grad_norm": 2.130828845031335,
      "learning_rate": 8.81877486219463e-07,
      "loss": 0.6532,
      "step": 4213
    },
    {
      "epoch": 0.3887902202744781,
      "grad_norm": 1.8535774897395798,
      "learning_rate": 8.81705954048972e-07,
      "loss": 0.7654,
      "step": 4214
    },
    {
      "epoch": 0.3888824818360051,
      "grad_norm": 1.5685077404719894,
      "learning_rate": 8.815343986153833e-07,
      "loss": 0.6149,
      "step": 4215
    },
    {
      "epoch": 0.3888824818360051,
      "eval_GEN Loss": 0.46671146154403687,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8679245283018868,
      "eval_PRM F1 AUC": 0.7634887375589314,
      "eval_PRM F1 Neg": 0.6037735849056604,
      "eval_PRM Loss": 0.39625248312950134,
      "eval_PRM NPV": 0.5333333333333333,
      "eval_PRM Precision": 0.9078947368421053,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9140625,
      "eval_runtime": 55.6395,
      "eval_samples_per_second": 1.168,
      "eval_steps_per_second": 0.09,
      "step": 4215
    },
    {
      "epoch": 0.388974743397532,
      "grad_norm": 1.6526513609658287,
      "learning_rate": 8.813628199342438e-07,
      "loss": 0.6009,
      "step": 4216
    },
    {
      "epoch": 0.38906700495905894,
      "grad_norm": 1.757727787369412,
      "learning_rate": 8.81191218021102e-07,
      "loss": 0.6424,
      "step": 4217
    },
    {
      "epoch": 0.38915926652058586,
      "grad_norm": 2.169229011572423,
      "learning_rate": 8.810195928915087e-07,
      "loss": 0.6784,
      "step": 4218
    },
    {
      "epoch": 0.3892515280821128,
      "grad_norm": 1.6219806215029597,
      "learning_rate": 8.808479445610169e-07,
      "loss": 0.5808,
      "step": 4219
    },
    {
      "epoch": 0.3893437896436397,
      "grad_norm": 1.4709005221147395,
      "learning_rate": 8.806762730451818e-07,
      "loss": 0.7293,
      "step": 4220
    },
    {
      "epoch": 0.3893437896436397,
      "eval_GEN Loss": 0.46757838129997253,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.382976233959198,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9084134697914124,
      "eval_runtime": 55.7814,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4220
    },
    {
      "epoch": 0.38943605120516667,
      "grad_norm": 1.604322000558313,
      "learning_rate": 8.805045783595604e-07,
      "loss": 0.6584,
      "step": 4221
    },
    {
      "epoch": 0.3895283127666936,
      "grad_norm": 2.038467337841438,
      "learning_rate": 8.803328605197117e-07,
      "loss": 0.7671,
      "step": 4222
    },
    {
      "epoch": 0.3896205743282205,
      "grad_norm": 1.878972099042478,
      "learning_rate": 8.801611195411975e-07,
      "loss": 0.7579,
      "step": 4223
    },
    {
      "epoch": 0.3897128358897474,
      "grad_norm": 1.5431423004238445,
      "learning_rate": 8.799893554395808e-07,
      "loss": 0.7222,
      "step": 4224
    },
    {
      "epoch": 0.38980509745127434,
      "grad_norm": 2.2025219864565972,
      "learning_rate": 8.798175682304275e-07,
      "loss": 0.8172,
      "step": 4225
    },
    {
      "epoch": 0.38980509745127434,
      "eval_GEN Loss": 0.4683484733104706,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3735651969909668,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9147836565971375,
      "eval_runtime": 56.9154,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 4225
    },
    {
      "epoch": 0.3898973590128013,
      "grad_norm": 2.045514727852902,
      "learning_rate": 8.796457579293049e-07,
      "loss": 0.7462,
      "step": 4226
    },
    {
      "epoch": 0.38998962057432823,
      "grad_norm": 1.6115629716649735,
      "learning_rate": 8.794739245517833e-07,
      "loss": 0.8266,
      "step": 4227
    },
    {
      "epoch": 0.39008188213585515,
      "grad_norm": 1.7415746057289088,
      "learning_rate": 8.793020681134337e-07,
      "loss": 0.7017,
      "step": 4228
    },
    {
      "epoch": 0.39017414369738207,
      "grad_norm": 2.6234889182620753,
      "learning_rate": 8.791301886298306e-07,
      "loss": 0.6513,
      "step": 4229
    },
    {
      "epoch": 0.390266405258909,
      "grad_norm": 1.6981577867104116,
      "learning_rate": 8.789582861165498e-07,
      "loss": 0.594,
      "step": 4230
    },
    {
      "epoch": 0.390266405258909,
      "eval_GEN Loss": 0.46845388412475586,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3746455907821655,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9098557829856873,
      "eval_runtime": 56.7228,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4230
    },
    {
      "epoch": 0.39035866682043596,
      "grad_norm": 2.773731049775948,
      "learning_rate": 8.787863605891693e-07,
      "loss": 1.1777,
      "step": 4231
    },
    {
      "epoch": 0.3904509283819629,
      "grad_norm": 1.696418373981368,
      "learning_rate": 8.786144120632695e-07,
      "loss": 0.6604,
      "step": 4232
    },
    {
      "epoch": 0.3905431899434898,
      "grad_norm": 2.1182461696003143,
      "learning_rate": 8.784424405544323e-07,
      "loss": 0.8694,
      "step": 4233
    },
    {
      "epoch": 0.3906354515050167,
      "grad_norm": 1.4293058448827018,
      "learning_rate": 8.782704460782423e-07,
      "loss": 0.5743,
      "step": 4234
    },
    {
      "epoch": 0.39072771306654364,
      "grad_norm": 1.8843160214318244,
      "learning_rate": 8.780984286502859e-07,
      "loss": 0.661,
      "step": 4235
    },
    {
      "epoch": 0.39072771306654364,
      "eval_GEN Loss": 0.4686843454837799,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8902439024390244,
      "eval_PRM F1 AUC": 0.7658459926663174,
      "eval_PRM F1 Neg": 0.625,
      "eval_PRM Loss": 0.3805236518383026,
      "eval_PRM NPV": 0.6,
      "eval_PRM Precision": 0.9012345679012346,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9052884578704834,
      "eval_runtime": 56.7354,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4235
    },
    {
      "epoch": 0.39081997462807055,
      "grad_norm": 2.3980605896043388,
      "learning_rate": 8.779263882861513e-07,
      "loss": 0.6525,
      "step": 4236
    },
    {
      "epoch": 0.39091223618959753,
      "grad_norm": 1.3055808070959023,
      "learning_rate": 8.777543250014295e-07,
      "loss": 0.6436,
      "step": 4237
    },
    {
      "epoch": 0.39100449775112445,
      "grad_norm": 2.0331653865520964,
      "learning_rate": 8.775822388117128e-07,
      "loss": 0.6717,
      "step": 4238
    },
    {
      "epoch": 0.39109675931265137,
      "grad_norm": 1.9885650635117162,
      "learning_rate": 8.77410129732596e-07,
      "loss": 0.8394,
      "step": 4239
    },
    {
      "epoch": 0.3911890208741783,
      "grad_norm": 1.6864053757155355,
      "learning_rate": 8.77237997779676e-07,
      "loss": 0.6439,
      "step": 4240
    },
    {
      "epoch": 0.3911890208741783,
      "eval_GEN Loss": 0.46901586651802063,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.3903825581073761,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.900120198726654,
      "eval_runtime": 56.4945,
      "eval_samples_per_second": 1.151,
      "eval_steps_per_second": 0.089,
      "step": 4240
    },
    {
      "epoch": 0.3912812824357052,
      "grad_norm": 1.6301848760171647,
      "learning_rate": 8.770658429685516e-07,
      "loss": 0.5913,
      "step": 4241
    },
    {
      "epoch": 0.3913735439972322,
      "grad_norm": 2.0210589593852935,
      "learning_rate": 8.768936653148238e-07,
      "loss": 0.6881,
      "step": 4242
    },
    {
      "epoch": 0.3914658055587591,
      "grad_norm": 2.2252430452388317,
      "learning_rate": 8.767214648340956e-07,
      "loss": 0.6681,
      "step": 4243
    },
    {
      "epoch": 0.391558067120286,
      "grad_norm": 2.453319884352639,
      "learning_rate": 8.765492415419718e-07,
      "loss": 0.9377,
      "step": 4244
    },
    {
      "epoch": 0.39165032868181293,
      "grad_norm": 2.2995127505561657,
      "learning_rate": 8.763769954540602e-07,
      "loss": 0.6208,
      "step": 4245
    },
    {
      "epoch": 0.39165032868181293,
      "eval_GEN Loss": 0.4683094620704651,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4013063609600067,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8981971144676208,
      "eval_runtime": 55.9065,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4245
    },
    {
      "epoch": 0.39174259024333985,
      "grad_norm": 1.5032394570002048,
      "learning_rate": 8.762047265859694e-07,
      "loss": 0.6121,
      "step": 4246
    },
    {
      "epoch": 0.3918348518048668,
      "grad_norm": 1.5480770372463208,
      "learning_rate": 8.760324349533107e-07,
      "loss": 0.6737,
      "step": 4247
    },
    {
      "epoch": 0.39192711336639374,
      "grad_norm": 3.29304573792603,
      "learning_rate": 8.758601205716979e-07,
      "loss": 1.124,
      "step": 4248
    },
    {
      "epoch": 0.39201937492792066,
      "grad_norm": 2.828185598759477,
      "learning_rate": 8.75687783456746e-07,
      "loss": 0.7595,
      "step": 4249
    },
    {
      "epoch": 0.3921116364894476,
      "grad_norm": 1.9740521344126285,
      "learning_rate": 8.755154236240724e-07,
      "loss": 0.7065,
      "step": 4250
    },
    {
      "epoch": 0.3921116364894476,
      "eval_GEN Loss": 0.46747711300849915,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.40354177355766296,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.889723539352417,
      "eval_runtime": 55.8998,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4250
    },
    {
      "epoch": 0.3922038980509745,
      "grad_norm": 3.759399027897542,
      "learning_rate": 8.75343041089297e-07,
      "loss": 1.1175,
      "step": 4251
    },
    {
      "epoch": 0.3922961596125014,
      "grad_norm": 1.6637726403596287,
      "learning_rate": 8.751706358680412e-07,
      "loss": 0.8956,
      "step": 4252
    },
    {
      "epoch": 0.3923884211740284,
      "grad_norm": 1.6850903286112668,
      "learning_rate": 8.749982079759284e-07,
      "loss": 0.6962,
      "step": 4253
    },
    {
      "epoch": 0.3924806827355553,
      "grad_norm": 1.7528438887076614,
      "learning_rate": 8.748257574285847e-07,
      "loss": 0.8169,
      "step": 4254
    },
    {
      "epoch": 0.3925729442970822,
      "grad_norm": 1.5810883783134633,
      "learning_rate": 8.746532842416374e-07,
      "loss": 0.6327,
      "step": 4255
    },
    {
      "epoch": 0.3925729442970822,
      "eval_GEN Loss": 0.4670736491680145,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.41287872195243835,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8895432949066162,
      "eval_runtime": 56.7809,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 4255
    },
    {
      "epoch": 0.39266520585860915,
      "grad_norm": 2.8406146197077455,
      "learning_rate": 8.744807884307167e-07,
      "loss": 0.7154,
      "step": 4256
    },
    {
      "epoch": 0.39275746742013606,
      "grad_norm": 1.2857420565541273,
      "learning_rate": 8.743082700114544e-07,
      "loss": 0.5915,
      "step": 4257
    },
    {
      "epoch": 0.39284972898166304,
      "grad_norm": 1.637782652915373,
      "learning_rate": 8.74135728999484e-07,
      "loss": 0.7289,
      "step": 4258
    },
    {
      "epoch": 0.39294199054318996,
      "grad_norm": 2.969165663626796,
      "learning_rate": 8.739631654104418e-07,
      "loss": 0.8184,
      "step": 4259
    },
    {
      "epoch": 0.3930342521047169,
      "grad_norm": 2.559728727344599,
      "learning_rate": 8.737905792599656e-07,
      "loss": 0.6754,
      "step": 4260
    },
    {
      "epoch": 0.3930342521047169,
      "eval_GEN Loss": 0.46754637360572815,
      "eval_GEN top-5 accuracy": 0.9833406400701447,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.43243032693862915,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8934495449066162,
      "eval_runtime": 55.7105,
      "eval_samples_per_second": 1.167,
      "eval_steps_per_second": 0.09,
      "step": 4260
    },
    {
      "epoch": 0.3931265136662438,
      "grad_norm": 2.5493950798759752,
      "learning_rate": 8.736179705636956e-07,
      "loss": 0.7258,
      "step": 4261
    },
    {
      "epoch": 0.3932187752277707,
      "grad_norm": 2.392558144440463,
      "learning_rate": 8.734453393372737e-07,
      "loss": 0.8199,
      "step": 4262
    },
    {
      "epoch": 0.3933110367892977,
      "grad_norm": 3.0183404373499054,
      "learning_rate": 8.73272685596344e-07,
      "loss": 0.765,
      "step": 4263
    },
    {
      "epoch": 0.3934032983508246,
      "grad_norm": 2.0420228995141523,
      "learning_rate": 8.73100009356553e-07,
      "loss": 0.8191,
      "step": 4264
    },
    {
      "epoch": 0.3934955599123515,
      "grad_norm": 1.6300875443257699,
      "learning_rate": 8.729273106335484e-07,
      "loss": 0.5293,
      "step": 4265
    },
    {
      "epoch": 0.3934955599123515,
      "eval_GEN Loss": 0.4678018093109131,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.875,
      "eval_PRM F1 AUC": 0.7695128339444735,
      "eval_PRM F1 Neg": 0.6153846153846154,
      "eval_PRM Loss": 0.42081284523010254,
      "eval_PRM NPV": 0.5517241379310345,
      "eval_PRM Precision": 0.9090909090909091,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8900841474533081,
      "eval_runtime": 55.9135,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4265
    },
    {
      "epoch": 0.39358782147387844,
      "grad_norm": 2.6738385856087556,
      "learning_rate": 8.727545894429808e-07,
      "loss": 0.8442,
      "step": 4266
    },
    {
      "epoch": 0.39368008303540536,
      "grad_norm": 2.977650023710774,
      "learning_rate": 8.725818458005023e-07,
      "loss": 0.7996,
      "step": 4267
    },
    {
      "epoch": 0.3937723445969323,
      "grad_norm": 2.202312702165346,
      "learning_rate": 8.724090797217673e-07,
      "loss": 0.7535,
      "step": 4268
    },
    {
      "epoch": 0.39386460615845925,
      "grad_norm": 2.2462239745983648,
      "learning_rate": 8.722362912224319e-07,
      "loss": 0.8528,
      "step": 4269
    },
    {
      "epoch": 0.39395686771998617,
      "grad_norm": 1.4743924556776011,
      "learning_rate": 8.72063480318155e-07,
      "loss": 0.6928,
      "step": 4270
    },
    {
      "epoch": 0.39395686771998617,
      "eval_GEN Loss": 0.46791863441467285,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.4018362760543823,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8888221383094788,
      "eval_runtime": 55.8428,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4270
    },
    {
      "epoch": 0.3940491292815131,
      "grad_norm": 1.6676823465996011,
      "learning_rate": 8.718906470245962e-07,
      "loss": 0.6344,
      "step": 4271
    },
    {
      "epoch": 0.39414139084304,
      "grad_norm": 1.6338925494223546,
      "learning_rate": 8.717177913574185e-07,
      "loss": 0.5008,
      "step": 4272
    },
    {
      "epoch": 0.3942336524045669,
      "grad_norm": 2.31864908227619,
      "learning_rate": 8.715449133322862e-07,
      "loss": 0.8714,
      "step": 4273
    },
    {
      "epoch": 0.3943259139660939,
      "grad_norm": 2.967127583133976,
      "learning_rate": 8.713720129648658e-07,
      "loss": 0.8467,
      "step": 4274
    },
    {
      "epoch": 0.3944181755276208,
      "grad_norm": 1.7684710300278648,
      "learning_rate": 8.711990902708259e-07,
      "loss": 0.8083,
      "step": 4275
    },
    {
      "epoch": 0.3944181755276208,
      "eval_GEN Loss": 0.46796780824661255,
      "eval_GEN top-5 accuracy": 0.9829022358614643,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3863297998905182,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8980769515037537,
      "eval_runtime": 55.8049,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4275
    },
    {
      "epoch": 0.39451043708914774,
      "grad_norm": 2.1835758697785215,
      "learning_rate": 8.710261452658367e-07,
      "loss": 0.7628,
      "step": 4276
    },
    {
      "epoch": 0.39460269865067465,
      "grad_norm": 1.616844244186754,
      "learning_rate": 8.708531779655712e-07,
      "loss": 0.656,
      "step": 4277
    },
    {
      "epoch": 0.3946949602122016,
      "grad_norm": 1.9467251957577303,
      "learning_rate": 8.706801883857036e-07,
      "loss": 0.5198,
      "step": 4278
    },
    {
      "epoch": 0.39478722177372855,
      "grad_norm": 2.0203125928456545,
      "learning_rate": 8.705071765419107e-07,
      "loss": 0.7089,
      "step": 4279
    },
    {
      "epoch": 0.39487948333525547,
      "grad_norm": 2.401128089853153,
      "learning_rate": 8.703341424498709e-07,
      "loss": 0.5473,
      "step": 4280
    },
    {
      "epoch": 0.39487948333525547,
      "eval_GEN Loss": 0.469140887260437,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.37683579325675964,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8948917984962463,
      "eval_runtime": 56.0903,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 4280
    },
    {
      "epoch": 0.3949717448967824,
      "grad_norm": 2.4335425958896266,
      "learning_rate": 8.701610861252651e-07,
      "loss": 0.6486,
      "step": 4281
    },
    {
      "epoch": 0.3950640064583093,
      "grad_norm": 1.9253960622065953,
      "learning_rate": 8.699880075837758e-07,
      "loss": 0.6372,
      "step": 4282
    },
    {
      "epoch": 0.3951562680198362,
      "grad_norm": 2.1450626740156618,
      "learning_rate": 8.698149068410874e-07,
      "loss": 0.8274,
      "step": 4283
    },
    {
      "epoch": 0.39524852958136314,
      "grad_norm": 1.8894892390947697,
      "learning_rate": 8.69641783912887e-07,
      "loss": 0.5552,
      "step": 4284
    },
    {
      "epoch": 0.3953407911428901,
      "grad_norm": 1.3202127166939235,
      "learning_rate": 8.694686388148629e-07,
      "loss": 0.5809,
      "step": 4285
    },
    {
      "epoch": 0.3953407911428901,
      "eval_GEN Loss": 0.4691178798675537,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3823791444301605,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8769831657409668,
      "eval_runtime": 55.9286,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 4285
    },
    {
      "epoch": 0.39543305270441703,
      "grad_norm": 1.7775162075571096,
      "learning_rate": 8.692954715627061e-07,
      "loss": 0.7479,
      "step": 4286
    },
    {
      "epoch": 0.39552531426594395,
      "grad_norm": 2.333735615942076,
      "learning_rate": 8.691222821721091e-07,
      "loss": 0.7185,
      "step": 4287
    },
    {
      "epoch": 0.39561757582747087,
      "grad_norm": 2.8226766334007642,
      "learning_rate": 8.689490706587665e-07,
      "loss": 0.7122,
      "step": 4288
    },
    {
      "epoch": 0.3957098373889978,
      "grad_norm": 2.0178110729702547,
      "learning_rate": 8.687758370383753e-07,
      "loss": 0.8036,
      "step": 4289
    },
    {
      "epoch": 0.39580209895052476,
      "grad_norm": 2.5008606023518145,
      "learning_rate": 8.686025813266341e-07,
      "loss": 0.9469,
      "step": 4290
    },
    {
      "epoch": 0.39580209895052476,
      "eval_GEN Loss": 0.46987855434417725,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3912084996700287,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8719350695610046,
      "eval_runtime": 55.8113,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4290
    },
    {
      "epoch": 0.3958943605120517,
      "grad_norm": 2.3268250789852067,
      "learning_rate": 8.684293035392434e-07,
      "loss": 0.8212,
      "step": 4291
    },
    {
      "epoch": 0.3959866220735786,
      "grad_norm": 1.9505158873446584,
      "learning_rate": 8.682560036919061e-07,
      "loss": 0.6496,
      "step": 4292
    },
    {
      "epoch": 0.3960788836351055,
      "grad_norm": 2.038437527865328,
      "learning_rate": 8.68082681800327e-07,
      "loss": 0.7373,
      "step": 4293
    },
    {
      "epoch": 0.39617114519663243,
      "grad_norm": 2.4737729408867306,
      "learning_rate": 8.679093378802126e-07,
      "loss": 0.7354,
      "step": 4294
    },
    {
      "epoch": 0.3962634067581594,
      "grad_norm": 1.6244962839446495,
      "learning_rate": 8.677359719472715e-07,
      "loss": 0.6132,
      "step": 4295
    },
    {
      "epoch": 0.3962634067581594,
      "eval_GEN Loss": 0.47019505500793457,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8888888888888888,
      "eval_PRM F1 AUC": 0.7815610267155579,
      "eval_PRM F1 Neg": 0.64,
      "eval_PRM Loss": 0.3758028745651245,
      "eval_PRM NPV": 0.5925925925925926,
      "eval_PRM Precision": 0.9113924050632911,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8687499761581421,
      "eval_runtime": 55.7957,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4295
    },
    {
      "epoch": 0.3963556683196863,
      "grad_norm": 1.5006721132422796,
      "learning_rate": 8.675625840172145e-07,
      "loss": 0.6793,
      "step": 4296
    },
    {
      "epoch": 0.39644792988121325,
      "grad_norm": 1.854110476334864,
      "learning_rate": 8.673891741057545e-07,
      "loss": 0.7255,
      "step": 4297
    },
    {
      "epoch": 0.39654019144274016,
      "grad_norm": 2.881951406084966,
      "learning_rate": 8.672157422286059e-07,
      "loss": 0.9449,
      "step": 4298
    },
    {
      "epoch": 0.3966324530042671,
      "grad_norm": 2.6298327970311997,
      "learning_rate": 8.670422884014857e-07,
      "loss": 0.8994,
      "step": 4299
    },
    {
      "epoch": 0.396724714565794,
      "grad_norm": 2.422473869873427,
      "learning_rate": 8.668688126401122e-07,
      "loss": 0.9399,
      "step": 4300
    },
    {
      "epoch": 0.396724714565794,
      "eval_GEN Loss": 0.4692126512527466,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.36371347308158875,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8774639368057251,
      "eval_runtime": 55.8896,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4300
    },
    {
      "epoch": 0.396816976127321,
      "grad_norm": 1.5792953443899793,
      "learning_rate": 8.666953149602061e-07,
      "loss": 0.6925,
      "step": 4301
    },
    {
      "epoch": 0.3969092376888479,
      "grad_norm": 1.2419722108732951,
      "learning_rate": 8.665217953774902e-07,
      "loss": 0.6279,
      "step": 4302
    },
    {
      "epoch": 0.3970014992503748,
      "grad_norm": 3.743828938173356,
      "learning_rate": 8.663482539076891e-07,
      "loss": 0.6552,
      "step": 4303
    },
    {
      "epoch": 0.39709376081190173,
      "grad_norm": 2.2608987667639786,
      "learning_rate": 8.661746905665295e-07,
      "loss": 0.7684,
      "step": 4304
    },
    {
      "epoch": 0.39718602237342865,
      "grad_norm": 2.48220334487363,
      "learning_rate": 8.660011053697396e-07,
      "loss": 0.7858,
      "step": 4305
    },
    {
      "epoch": 0.39718602237342865,
      "eval_GEN Loss": 0.4675985872745514,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.896969696969697,
      "eval_PRM F1 AUC": 0.7718700890518595,
      "eval_PRM F1 Neg": 0.6382978723404256,
      "eval_PRM Loss": 0.3623906970024109,
      "eval_PRM NPV": 0.625,
      "eval_PRM Precision": 0.9024390243902439,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8757211565971375,
      "eval_runtime": 55.8885,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4305
    },
    {
      "epoch": 0.3972782839349556,
      "grad_norm": 1.9106094571790158,
      "learning_rate": 8.658274983330501e-07,
      "loss": 0.6983,
      "step": 4306
    },
    {
      "epoch": 0.39737054549648254,
      "grad_norm": 2.021054454586861,
      "learning_rate": 8.656538694721938e-07,
      "loss": 0.6352,
      "step": 4307
    },
    {
      "epoch": 0.39746280705800946,
      "grad_norm": 1.8533564134638605,
      "learning_rate": 8.65480218802905e-07,
      "loss": 0.6699,
      "step": 4308
    },
    {
      "epoch": 0.3975550686195364,
      "grad_norm": 1.8649460331187122,
      "learning_rate": 8.653065463409203e-07,
      "loss": 0.6021,
      "step": 4309
    },
    {
      "epoch": 0.3976473301810633,
      "grad_norm": 1.8993497965328716,
      "learning_rate": 8.65132852101978e-07,
      "loss": 0.8226,
      "step": 4310
    },
    {
      "epoch": 0.3976473301810633,
      "eval_GEN Loss": 0.46792805194854736,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3694693446159363,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8582331538200378,
      "eval_runtime": 55.9861,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 4310
    },
    {
      "epoch": 0.39773959174259027,
      "grad_norm": 1.1810875212470133,
      "learning_rate": 8.649591361018189e-07,
      "loss": 0.6184,
      "step": 4311
    },
    {
      "epoch": 0.3978318533041172,
      "grad_norm": 2.0411580430573415,
      "learning_rate": 8.647853983561849e-07,
      "loss": 0.7681,
      "step": 4312
    },
    {
      "epoch": 0.3979241148656441,
      "grad_norm": 1.7961702318560229,
      "learning_rate": 8.646116388808208e-07,
      "loss": 0.7263,
      "step": 4313
    },
    {
      "epoch": 0.398016376427171,
      "grad_norm": 3.8066442571781756,
      "learning_rate": 8.64437857691473e-07,
      "loss": 0.8646,
      "step": 4314
    },
    {
      "epoch": 0.39810863798869794,
      "grad_norm": 1.7079718234420176,
      "learning_rate": 8.642640548038894e-07,
      "loss": 0.7741,
      "step": 4315
    },
    {
      "epoch": 0.39810863798869794,
      "eval_GEN Loss": 0.46880170702934265,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.3784363567829132,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8519831895828247,
      "eval_runtime": 55.8455,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4315
    },
    {
      "epoch": 0.39820089955022486,
      "grad_norm": 1.693772350603163,
      "learning_rate": 8.640902302338205e-07,
      "loss": 0.7377,
      "step": 4316
    },
    {
      "epoch": 0.39829316111175184,
      "grad_norm": 1.9772742178441343,
      "learning_rate": 8.63916383997019e-07,
      "loss": 0.6117,
      "step": 4317
    },
    {
      "epoch": 0.39838542267327876,
      "grad_norm": 2.2263447589393395,
      "learning_rate": 8.637425161092385e-07,
      "loss": 0.7982,
      "step": 4318
    },
    {
      "epoch": 0.3984776842348057,
      "grad_norm": 2.169691279340369,
      "learning_rate": 8.635686265862353e-07,
      "loss": 0.8237,
      "step": 4319
    },
    {
      "epoch": 0.3985699457963326,
      "grad_norm": 3.8853240209915243,
      "learning_rate": 8.633947154437676e-07,
      "loss": 0.6362,
      "step": 4320
    },
    {
      "epoch": 0.3985699457963326,
      "eval_GEN Loss": 0.46980276703834534,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4024483561515808,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8568509817123413,
      "eval_runtime": 55.8756,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4320
    },
    {
      "epoch": 0.3986622073578595,
      "grad_norm": 1.3914618434889319,
      "learning_rate": 8.632207826975957e-07,
      "loss": 0.5795,
      "step": 4321
    },
    {
      "epoch": 0.3987544689193865,
      "grad_norm": 1.915776930310122,
      "learning_rate": 8.630468283634814e-07,
      "loss": 0.705,
      "step": 4322
    },
    {
      "epoch": 0.3988467304809134,
      "grad_norm": 1.7215899315176006,
      "learning_rate": 8.62872852457189e-07,
      "loss": 0.8159,
      "step": 4323
    },
    {
      "epoch": 0.3989389920424403,
      "grad_norm": 1.545074800659904,
      "learning_rate": 8.626988549944841e-07,
      "loss": 0.5778,
      "step": 4324
    },
    {
      "epoch": 0.39903125360396724,
      "grad_norm": 1.7586549280698154,
      "learning_rate": 8.625248359911349e-07,
      "loss": 0.8274,
      "step": 4325
    },
    {
      "epoch": 0.39903125360396724,
      "eval_GEN Loss": 0.46910834312438965,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.8662420382165605,
      "eval_PRM F1 AUC": 0.7792037716081718,
      "eval_PRM F1 Neg": 0.6181818181818182,
      "eval_PRM Loss": 0.4126182198524475,
      "eval_PRM NPV": 0.53125,
      "eval_PRM Precision": 0.918918918918919,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8543269038200378,
      "eval_runtime": 55.8429,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4325
    },
    {
      "epoch": 0.39912351516549416,
      "grad_norm": 2.5558426758543007,
      "learning_rate": 8.623507954629112e-07,
      "loss": 0.638,
      "step": 4326
    },
    {
      "epoch": 0.39921577672702113,
      "grad_norm": 1.5081553023369632,
      "learning_rate": 8.621767334255848e-07,
      "loss": 0.6766,
      "step": 4327
    },
    {
      "epoch": 0.39930803828854805,
      "grad_norm": 2.1463099975729816,
      "learning_rate": 8.620026498949296e-07,
      "loss": 0.7741,
      "step": 4328
    },
    {
      "epoch": 0.39940029985007497,
      "grad_norm": 1.9034757117327512,
      "learning_rate": 8.618285448867211e-07,
      "loss": 0.6781,
      "step": 4329
    },
    {
      "epoch": 0.3994925614116019,
      "grad_norm": 1.2363998587003104,
      "learning_rate": 8.616544184167372e-07,
      "loss": 0.5225,
      "step": 4330
    },
    {
      "epoch": 0.3994925614116019,
      "eval_GEN Loss": 0.47018128633499146,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.40099209547042847,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8483173251152039,
      "eval_runtime": 56.7683,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 4330
    },
    {
      "epoch": 0.3995848229731288,
      "grad_norm": 2.6054869448047797,
      "learning_rate": 8.614802705007573e-07,
      "loss": 0.8293,
      "step": 4331
    },
    {
      "epoch": 0.3996770845346557,
      "grad_norm": 1.871319476311996,
      "learning_rate": 8.613061011545631e-07,
      "loss": 0.6276,
      "step": 4332
    },
    {
      "epoch": 0.3997693460961827,
      "grad_norm": 2.2925421434459423,
      "learning_rate": 8.611319103939381e-07,
      "loss": 0.9295,
      "step": 4333
    },
    {
      "epoch": 0.3998616076577096,
      "grad_norm": 2.4538391934055537,
      "learning_rate": 8.609576982346676e-07,
      "loss": 0.8343,
      "step": 4334
    },
    {
      "epoch": 0.39995386921923654,
      "grad_norm": 2.8113169959474975,
      "learning_rate": 8.60783464692539e-07,
      "loss": 0.8452,
      "step": 4335
    },
    {
      "epoch": 0.39995386921923654,
      "eval_GEN Loss": 0.4706370234489441,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.3908641040325165,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8507211804389954,
      "eval_runtime": 56.6316,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 4335
    },
    {
      "epoch": 0.40004613078076345,
      "grad_norm": 2.5281605112095114,
      "learning_rate": 8.606092097833418e-07,
      "loss": 0.9098,
      "step": 4336
    },
    {
      "epoch": 0.40013839234229037,
      "grad_norm": 1.3100590711659552,
      "learning_rate": 8.604349335228673e-07,
      "loss": 0.652,
      "step": 4337
    },
    {
      "epoch": 0.40023065390381735,
      "grad_norm": 1.5234659517323896,
      "learning_rate": 8.602606359269083e-07,
      "loss": 0.7689,
      "step": 4338
    },
    {
      "epoch": 0.40032291546534426,
      "grad_norm": 3.0874861229794472,
      "learning_rate": 8.600863170112604e-07,
      "loss": 0.9095,
      "step": 4339
    },
    {
      "epoch": 0.4004151770268712,
      "grad_norm": 2.235667731018532,
      "learning_rate": 8.599119767917204e-07,
      "loss": 0.7663,
      "step": 4340
    },
    {
      "epoch": 0.4004151770268712,
      "eval_GEN Loss": 0.4703901708126068,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.39742788672447205,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.849879801273346,
      "eval_runtime": 56.5996,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 4340
    },
    {
      "epoch": 0.4005074385883981,
      "grad_norm": 1.9462855893287139,
      "learning_rate": 8.597376152840872e-07,
      "loss": 0.5899,
      "step": 4341
    },
    {
      "epoch": 0.400599700149925,
      "grad_norm": 2.6000624996684785,
      "learning_rate": 8.59563232504162e-07,
      "loss": 0.8631,
      "step": 4342
    },
    {
      "epoch": 0.400691961711452,
      "grad_norm": 2.156921172353974,
      "learning_rate": 8.593888284677472e-07,
      "loss": 0.5716,
      "step": 4343
    },
    {
      "epoch": 0.4007842232729789,
      "grad_norm": 2.584212344365474,
      "learning_rate": 8.59214403190648e-07,
      "loss": 0.8445,
      "step": 4344
    },
    {
      "epoch": 0.40087648483450583,
      "grad_norm": 2.03840427685523,
      "learning_rate": 8.590399566886708e-07,
      "loss": 0.6369,
      "step": 4345
    },
    {
      "epoch": 0.40087648483450583,
      "eval_GEN Loss": 0.4684564769268036,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.40281203389167786,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.848557710647583,
      "eval_runtime": 55.942,
      "eval_samples_per_second": 1.162,
      "eval_steps_per_second": 0.089,
      "step": 4345
    },
    {
      "epoch": 0.40096874639603275,
      "grad_norm": 1.7097548289773044,
      "learning_rate": 8.588654889776244e-07,
      "loss": 0.7837,
      "step": 4346
    },
    {
      "epoch": 0.40106100795755967,
      "grad_norm": 1.613053005537647,
      "learning_rate": 8.586910000733193e-07,
      "loss": 0.7572,
      "step": 4347
    },
    {
      "epoch": 0.4011532695190866,
      "grad_norm": 2.1600841435389917,
      "learning_rate": 8.585164899915679e-07,
      "loss": 0.8231,
      "step": 4348
    },
    {
      "epoch": 0.40124553108061356,
      "grad_norm": 2.3169260190547933,
      "learning_rate": 8.583419587481845e-07,
      "loss": 0.8244,
      "step": 4349
    },
    {
      "epoch": 0.4013377926421405,
      "grad_norm": 2.4666178968342956,
      "learning_rate": 8.581674063589858e-07,
      "loss": 0.7386,
      "step": 4350
    },
    {
      "epoch": 0.4013377926421405,
      "eval_GEN Loss": 0.46762922406196594,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.4029407203197479,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.849879801273346,
      "eval_runtime": 56.6279,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 4350
    },
    {
      "epoch": 0.4014300542036674,
      "grad_norm": 1.9500004709727101,
      "learning_rate": 8.579928328397897e-07,
      "loss": 0.7464,
      "step": 4351
    },
    {
      "epoch": 0.4015223157651943,
      "grad_norm": 1.7715356063353256,
      "learning_rate": 8.578182382064161e-07,
      "loss": 0.6496,
      "step": 4352
    },
    {
      "epoch": 0.40161457732672123,
      "grad_norm": 1.7966537489965584,
      "learning_rate": 8.576436224746874e-07,
      "loss": 0.7461,
      "step": 4353
    },
    {
      "epoch": 0.4017068388882482,
      "grad_norm": 1.4498825680301908,
      "learning_rate": 8.574689856604274e-07,
      "loss": 0.5625,
      "step": 4354
    },
    {
      "epoch": 0.4017991004497751,
      "grad_norm": 1.805573182684093,
      "learning_rate": 8.572943277794623e-07,
      "loss": 0.683,
      "step": 4355
    },
    {
      "epoch": 0.4017991004497751,
      "eval_GEN Loss": 0.46739768981933594,
      "eval_GEN top-5 accuracy": 0.9807102148180622,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8805031446540881,
      "eval_PRM F1 AUC": 0.7912519643792562,
      "eval_PRM F1 Neg": 0.6415094339622641,
      "eval_PRM Loss": 0.3970527946949005,
      "eval_PRM NPV": 0.5666666666666667,
      "eval_PRM Precision": 0.9210526315789473,
      "eval_PRM Recall": 0.8433734939759037,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8538461327552795,
      "eval_runtime": 56.8637,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 4355
    },
    {
      "epoch": 0.40189136201130204,
      "grad_norm": 1.9613282217487513,
      "learning_rate": 8.571196488476191e-07,
      "loss": 0.6686,
      "step": 4356
    },
    {
      "epoch": 0.40198362357282896,
      "grad_norm": 1.632307050440881,
      "learning_rate": 8.569449488807282e-07,
      "loss": 0.7813,
      "step": 4357
    },
    {
      "epoch": 0.4020758851343559,
      "grad_norm": 1.6378752783854558,
      "learning_rate": 8.56770227894621e-07,
      "loss": 0.4462,
      "step": 4358
    },
    {
      "epoch": 0.40216814669588286,
      "grad_norm": 2.5864864685607944,
      "learning_rate": 8.56595485905131e-07,
      "loss": 0.7576,
      "step": 4359
    },
    {
      "epoch": 0.4022604082574098,
      "grad_norm": 1.7330514206039742,
      "learning_rate": 8.564207229280934e-07,
      "loss": 0.7924,
      "step": 4360
    },
    {
      "epoch": 0.4022604082574098,
      "eval_GEN Loss": 0.46744897961616516,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.38540875911712646,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8606970906257629,
      "eval_runtime": 56.7162,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4360
    },
    {
      "epoch": 0.4023526698189367,
      "grad_norm": 3.369996026573113,
      "learning_rate": 8.562459389793456e-07,
      "loss": 0.7206,
      "step": 4361
    },
    {
      "epoch": 0.4024449313804636,
      "grad_norm": 3.2931655708702103,
      "learning_rate": 8.560711340747269e-07,
      "loss": 1.0202,
      "step": 4362
    },
    {
      "epoch": 0.40253719294199053,
      "grad_norm": 1.8059033468820964,
      "learning_rate": 8.558963082300784e-07,
      "loss": 0.7273,
      "step": 4363
    },
    {
      "epoch": 0.40262945450351745,
      "grad_norm": 1.8184119178783473,
      "learning_rate": 8.557214614612428e-07,
      "loss": 0.8668,
      "step": 4364
    },
    {
      "epoch": 0.4027217160650444,
      "grad_norm": 2.9247755605565113,
      "learning_rate": 8.555465937840654e-07,
      "loss": 0.7438,
      "step": 4365
    },
    {
      "epoch": 0.4027217160650444,
      "eval_GEN Loss": 0.46761518716812134,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8765432098765432,
      "eval_PRM F1 AUC": 0.7537977998952332,
      "eval_PRM F1 Neg": 0.6,
      "eval_PRM Loss": 0.38552287220954895,
      "eval_PRM NPV": 0.5555555555555556,
      "eval_PRM Precision": 0.8987341772151899,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8652644157409668,
      "eval_runtime": 56.6237,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 4365
    },
    {
      "epoch": 0.40281397762657134,
      "grad_norm": 2.007352052654947,
      "learning_rate": 8.553717052143927e-07,
      "loss": 0.5653,
      "step": 4366
    },
    {
      "epoch": 0.40290623918809826,
      "grad_norm": 2.4348150487070366,
      "learning_rate": 8.551967957680735e-07,
      "loss": 0.7362,
      "step": 4367
    },
    {
      "epoch": 0.4029985007496252,
      "grad_norm": 1.6882944930720185,
      "learning_rate": 8.550218654609583e-07,
      "loss": 0.6152,
      "step": 4368
    },
    {
      "epoch": 0.4030907623111521,
      "grad_norm": 2.728792418299751,
      "learning_rate": 8.548469143088995e-07,
      "loss": 0.9368,
      "step": 4369
    },
    {
      "epoch": 0.40318302387267907,
      "grad_norm": 1.7312906967832777,
      "learning_rate": 8.546719423277515e-07,
      "loss": 0.6434,
      "step": 4370
    },
    {
      "epoch": 0.40318302387267907,
      "eval_GEN Loss": 0.4671454131603241,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8834355828220859,
      "eval_PRM F1 AUC": 0.7598218962807753,
      "eval_PRM F1 Neg": 0.6122448979591837,
      "eval_PRM Loss": 0.37842366099357605,
      "eval_PRM NPV": 0.5769230769230769,
      "eval_PRM Precision": 0.9,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.8683894276618958,
      "eval_runtime": 55.9002,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4370
    },
    {
      "epoch": 0.403275285434206,
      "grad_norm": 2.387194276491699,
      "learning_rate": 8.544969495333708e-07,
      "loss": 0.7262,
      "step": 4371
    },
    {
      "epoch": 0.4033675469957329,
      "grad_norm": 2.3504103136965036,
      "learning_rate": 8.54321935941615e-07,
      "loss": 0.8849,
      "step": 4372
    },
    {
      "epoch": 0.4034598085572598,
      "grad_norm": 2.5378532715092943,
      "learning_rate": 8.541469015683448e-07,
      "loss": 0.6914,
      "step": 4373
    },
    {
      "epoch": 0.40355207011878674,
      "grad_norm": 1.3334952623466765,
      "learning_rate": 8.539718464294214e-07,
      "loss": 0.5388,
      "step": 4374
    },
    {
      "epoch": 0.40364433168031366,
      "grad_norm": 2.5910527360789968,
      "learning_rate": 8.537967705407091e-07,
      "loss": 1.1671,
      "step": 4375
    },
    {
      "epoch": 0.40364433168031366,
      "eval_GEN Loss": 0.4680883288383484,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.39346766471862793,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.876682698726654,
      "eval_runtime": 56.8994,
      "eval_samples_per_second": 1.142,
      "eval_steps_per_second": 0.088,
      "step": 4375
    },
    {
      "epoch": 0.40373659324184064,
      "grad_norm": 2.4608254281800583,
      "learning_rate": 8.536216739180732e-07,
      "loss": 0.8573,
      "step": 4376
    },
    {
      "epoch": 0.40382885480336755,
      "grad_norm": 1.6992940977093691,
      "learning_rate": 8.534465565773812e-07,
      "loss": 0.8575,
      "step": 4377
    },
    {
      "epoch": 0.4039211163648945,
      "grad_norm": 2.17954121982997,
      "learning_rate": 8.53271418534503e-07,
      "loss": 0.9639,
      "step": 4378
    },
    {
      "epoch": 0.4040133779264214,
      "grad_norm": 1.8256402776695555,
      "learning_rate": 8.530962598053093e-07,
      "loss": 0.6011,
      "step": 4379
    },
    {
      "epoch": 0.4041056394879483,
      "grad_norm": 1.9087405972380291,
      "learning_rate": 8.529210804056736e-07,
      "loss": 0.6906,
      "step": 4380
    },
    {
      "epoch": 0.4041056394879483,
      "eval_GEN Loss": 0.46784546971321106,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.41346266865730286,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8895432949066162,
      "eval_runtime": 56.7067,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4380
    },
    {
      "epoch": 0.4041979010494753,
      "grad_norm": 1.8445393058249546,
      "learning_rate": 8.527458803514707e-07,
      "loss": 0.8251,
      "step": 4381
    },
    {
      "epoch": 0.4042901626110022,
      "grad_norm": 1.9618027964644604,
      "learning_rate": 8.525706596585778e-07,
      "loss": 0.8287,
      "step": 4382
    },
    {
      "epoch": 0.4043824241725291,
      "grad_norm": 1.9320467612487213,
      "learning_rate": 8.523954183428734e-07,
      "loss": 0.801,
      "step": 4383
    },
    {
      "epoch": 0.40447468573405604,
      "grad_norm": 1.875394469172402,
      "learning_rate": 8.522201564202384e-07,
      "loss": 0.8412,
      "step": 4384
    },
    {
      "epoch": 0.40456694729558296,
      "grad_norm": 2.4927860706774556,
      "learning_rate": 8.520448739065551e-07,
      "loss": 0.7267,
      "step": 4385
    },
    {
      "epoch": 0.40456694729558296,
      "eval_GEN Loss": 0.4678288698196411,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.4132847785949707,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8905649185180664,
      "eval_runtime": 56.7675,
      "eval_samples_per_second": 1.145,
      "eval_steps_per_second": 0.088,
      "step": 4385
    },
    {
      "epoch": 0.40465920885710993,
      "grad_norm": 1.7263428219561536,
      "learning_rate": 8.51869570817708e-07,
      "loss": 0.6153,
      "step": 4386
    },
    {
      "epoch": 0.40475147041863685,
      "grad_norm": 1.4988849618998097,
      "learning_rate": 8.516942471695832e-07,
      "loss": 0.6473,
      "step": 4387
    },
    {
      "epoch": 0.40484373198016377,
      "grad_norm": 1.1853685593104997,
      "learning_rate": 8.51518902978069e-07,
      "loss": 0.51,
      "step": 4388
    },
    {
      "epoch": 0.4049359935416907,
      "grad_norm": 2.2161008965613775,
      "learning_rate": 8.513435382590553e-07,
      "loss": 0.8579,
      "step": 4389
    },
    {
      "epoch": 0.4050282551032176,
      "grad_norm": 1.9920668011206768,
      "learning_rate": 8.511681530284338e-07,
      "loss": 0.7394,
      "step": 4390
    },
    {
      "epoch": 0.4050282551032176,
      "eval_GEN Loss": 0.46759864687919617,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.8207547169811321,
      "eval_PRM F1": 0.8819875776397516,
      "eval_PRM F1 AUC": 0.7755369303300157,
      "eval_PRM F1 Neg": 0.6274509803921569,
      "eval_PRM Loss": 0.395404577255249,
      "eval_PRM NPV": 0.5714285714285714,
      "eval_PRM Precision": 0.9102564102564102,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8879206776618958,
      "eval_runtime": 56.6619,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 4390
    },
    {
      "epoch": 0.4051205166647445,
      "grad_norm": 2.2142864018699857,
      "learning_rate": 8.509927473020984e-07,
      "loss": 0.7945,
      "step": 4391
    },
    {
      "epoch": 0.4052127782262715,
      "grad_norm": 3.7178164515061827,
      "learning_rate": 8.508173210959443e-07,
      "loss": 0.8368,
      "step": 4392
    },
    {
      "epoch": 0.4053050397877984,
      "grad_norm": 2.126597112537908,
      "learning_rate": 8.506418744258691e-07,
      "loss": 0.7517,
      "step": 4393
    },
    {
      "epoch": 0.40539730134932533,
      "grad_norm": 1.7946472581849136,
      "learning_rate": 8.504664073077723e-07,
      "loss": 0.841,
      "step": 4394
    },
    {
      "epoch": 0.40548956291085225,
      "grad_norm": 1.4681017207467861,
      "learning_rate": 8.502909197575545e-07,
      "loss": 0.749,
      "step": 4395
    },
    {
      "epoch": 0.40548956291085225,
      "eval_GEN Loss": 0.46796053647994995,
      "eval_GEN top-5 accuracy": 0.979395002192021,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.3973378539085388,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8907451629638672,
      "eval_runtime": 56.8194,
      "eval_samples_per_second": 1.144,
      "eval_steps_per_second": 0.088,
      "step": 4395
    },
    {
      "epoch": 0.40558182447237917,
      "grad_norm": 1.7169143722621392,
      "learning_rate": 8.501154117911194e-07,
      "loss": 0.8347,
      "step": 4396
    },
    {
      "epoch": 0.40567408603390614,
      "grad_norm": 2.1505651863819844,
      "learning_rate": 8.49939883424371e-07,
      "loss": 0.7512,
      "step": 4397
    },
    {
      "epoch": 0.40576634759543306,
      "grad_norm": 2.8080987594474296,
      "learning_rate": 8.497643346732164e-07,
      "loss": 0.8967,
      "step": 4398
    },
    {
      "epoch": 0.40585860915696,
      "grad_norm": 1.7603215902984848,
      "learning_rate": 8.49588765553564e-07,
      "loss": 0.7488,
      "step": 4399
    },
    {
      "epoch": 0.4059508707184869,
      "grad_norm": 3.248644485469129,
      "learning_rate": 8.494131760813243e-07,
      "loss": 0.6555,
      "step": 4400
    },
    {
      "epoch": 0.4059508707184869,
      "eval_GEN Loss": 0.46874508261680603,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.4028520882129669,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8941105604171753,
      "eval_runtime": 56.6489,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 4400
    },
    {
      "epoch": 0.4060431322800138,
      "grad_norm": 2.126199976435892,
      "learning_rate": 8.49237566272409e-07,
      "loss": 0.748,
      "step": 4401
    },
    {
      "epoch": 0.4061353938415408,
      "grad_norm": 1.876211486711498,
      "learning_rate": 8.490619361427327e-07,
      "loss": 0.6085,
      "step": 4402
    },
    {
      "epoch": 0.4062276554030677,
      "grad_norm": 1.90944988052703,
      "learning_rate": 8.488862857082112e-07,
      "loss": 0.5791,
      "step": 4403
    },
    {
      "epoch": 0.40631991696459463,
      "grad_norm": 3.087267682102724,
      "learning_rate": 8.487106149847619e-07,
      "loss": 0.7401,
      "step": 4404
    },
    {
      "epoch": 0.40641217852612155,
      "grad_norm": 1.6621465576501164,
      "learning_rate": 8.485349239883045e-07,
      "loss": 0.6956,
      "step": 4405
    },
    {
      "epoch": 0.40641217852612155,
      "eval_GEN Loss": 0.4685569703578949,
      "eval_GEN top-5 accuracy": 0.9798334064007015,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8944099378881988,
      "eval_PRM F1 AUC": 0.8033001571503404,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.4055129587650299,
      "eval_PRM NPV": 0.6071428571428571,
      "eval_PRM Precision": 0.9230769230769231,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8938701748847961,
      "eval_runtime": 56.6806,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 4405
    },
    {
      "epoch": 0.40650444008764847,
      "grad_norm": 1.4851470168327767,
      "learning_rate": 8.483592127347605e-07,
      "loss": 0.661,
      "step": 4406
    },
    {
      "epoch": 0.4065967016491754,
      "grad_norm": 1.8451475631657672,
      "learning_rate": 8.481834812400532e-07,
      "loss": 0.7586,
      "step": 4407
    },
    {
      "epoch": 0.40668896321070236,
      "grad_norm": 1.3547269587281008,
      "learning_rate": 8.480077295201072e-07,
      "loss": 0.7059,
      "step": 4408
    },
    {
      "epoch": 0.4067812247722293,
      "grad_norm": 1.7124027834867268,
      "learning_rate": 8.478319575908502e-07,
      "loss": 0.6812,
      "step": 4409
    },
    {
      "epoch": 0.4068734863337562,
      "grad_norm": 1.759564729600565,
      "learning_rate": 8.4765616546821e-07,
      "loss": 0.7851,
      "step": 4410
    },
    {
      "epoch": 0.4068734863337562,
      "eval_GEN Loss": 0.4692930281162262,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.42200595140457153,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9019230604171753,
      "eval_runtime": 56.7105,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4410
    },
    {
      "epoch": 0.4069657478952831,
      "grad_norm": 1.7811880694067443,
      "learning_rate": 8.474803531681181e-07,
      "loss": 0.5887,
      "step": 4411
    },
    {
      "epoch": 0.40705800945681003,
      "grad_norm": 2.2451517550932794,
      "learning_rate": 8.473045207065062e-07,
      "loss": 0.7215,
      "step": 4412
    },
    {
      "epoch": 0.407150271018337,
      "grad_norm": 1.7316909628745827,
      "learning_rate": 8.471286680993085e-07,
      "loss": 0.7175,
      "step": 4413
    },
    {
      "epoch": 0.4072425325798639,
      "grad_norm": 2.340231309531403,
      "learning_rate": 8.469527953624615e-07,
      "loss": 0.6224,
      "step": 4414
    },
    {
      "epoch": 0.40733479414139084,
      "grad_norm": 1.4651810419410551,
      "learning_rate": 8.467769025119027e-07,
      "loss": 0.5748,
      "step": 4415
    },
    {
      "epoch": 0.40733479414139084,
      "eval_GEN Loss": 0.4682878255844116,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8301886792452831,
      "eval_PRM F1": 0.8875,
      "eval_PRM F1 AUC": 0.7972760607647982,
      "eval_PRM F1 Neg": 0.6538461538461539,
      "eval_PRM Loss": 0.4185051918029785,
      "eval_PRM NPV": 0.5862068965517241,
      "eval_PRM Precision": 0.922077922077922,
      "eval_PRM Recall": 0.8554216867469879,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.9013822078704834,
      "eval_runtime": 56.5889,
      "eval_samples_per_second": 1.149,
      "eval_steps_per_second": 0.088,
      "step": 4415
    },
    {
      "epoch": 0.40742705570291776,
      "grad_norm": 1.879025493176752,
      "learning_rate": 8.46600989563572e-07,
      "loss": 0.7935,
      "step": 4416
    },
    {
      "epoch": 0.4075193172644447,
      "grad_norm": 2.327842369918227,
      "learning_rate": 8.464250565334107e-07,
      "loss": 0.5894,
      "step": 4417
    },
    {
      "epoch": 0.40761157882597165,
      "grad_norm": 1.3246416414931381,
      "learning_rate": 8.462491034373623e-07,
      "loss": 0.5294,
      "step": 4418
    },
    {
      "epoch": 0.4077038403874986,
      "grad_norm": 3.7774430830015397,
      "learning_rate": 8.46073130291372e-07,
      "loss": 0.706,
      "step": 4419
    },
    {
      "epoch": 0.4077961019490255,
      "grad_norm": 2.650456065546193,
      "learning_rate": 8.458971371113865e-07,
      "loss": 0.882,
      "step": 4420
    },
    {
      "epoch": 0.4077961019490255,
      "eval_GEN Loss": 0.467420756816864,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.39714956283569336,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.89453125,
      "eval_runtime": 56.6992,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4420
    },
    {
      "epoch": 0.4078883635105524,
      "grad_norm": 2.0157016281789484,
      "learning_rate": 8.457211239133549e-07,
      "loss": 0.9023,
      "step": 4421
    },
    {
      "epoch": 0.40798062507207933,
      "grad_norm": 2.223748213372893,
      "learning_rate": 8.455450907132276e-07,
      "loss": 0.6436,
      "step": 4422
    },
    {
      "epoch": 0.40807288663360625,
      "grad_norm": 3.16632598211609,
      "learning_rate": 8.453690375269571e-07,
      "loss": 0.987,
      "step": 4423
    },
    {
      "epoch": 0.4081651481951332,
      "grad_norm": 2.002340916306112,
      "learning_rate": 8.451929643704973e-07,
      "loss": 0.758,
      "step": 4424
    },
    {
      "epoch": 0.40825740975666014,
      "grad_norm": 1.4163943534286552,
      "learning_rate": 8.450168712598047e-07,
      "loss": 0.6064,
      "step": 4425
    },
    {
      "epoch": 0.40825740975666014,
      "eval_GEN Loss": 0.4673185646533966,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9024390243902439,
      "eval_PRM F1 AUC": 0.7936092194866422,
      "eval_PRM F1 Neg": 0.6666666666666666,
      "eval_PRM Loss": 0.3878537714481354,
      "eval_PRM NPV": 0.64,
      "eval_PRM Precision": 0.9135802469135802,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.8930889368057251,
      "eval_runtime": 56.7088,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4425
    },
    {
      "epoch": 0.40834967131818706,
      "grad_norm": 1.7478277707364132,
      "learning_rate": 8.448407582108369e-07,
      "loss": 0.7904,
      "step": 4426
    },
    {
      "epoch": 0.408441932879714,
      "grad_norm": 2.2197129291864166,
      "learning_rate": 8.446646252395533e-07,
      "loss": 1.0033,
      "step": 4427
    },
    {
      "epoch": 0.4085341944412409,
      "grad_norm": 1.675542293324628,
      "learning_rate": 8.444884723619159e-07,
      "loss": 0.8066,
      "step": 4428
    },
    {
      "epoch": 0.40862645600276787,
      "grad_norm": 1.4592189769777335,
      "learning_rate": 8.443122995938876e-07,
      "loss": 0.6287,
      "step": 4429
    },
    {
      "epoch": 0.4087187175642948,
      "grad_norm": 1.3523548018805331,
      "learning_rate": 8.441361069514334e-07,
      "loss": 0.6749,
      "step": 4430
    },
    {
      "epoch": 0.4087187175642948,
      "eval_GEN Loss": 0.4690874218940735,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.39502543210983276,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8984375,
      "eval_runtime": 56.7137,
      "eval_samples_per_second": 1.146,
      "eval_steps_per_second": 0.088,
      "step": 4430
    },
    {
      "epoch": 0.4088109791258217,
      "grad_norm": 1.5013281547125283,
      "learning_rate": 8.439598944505204e-07,
      "loss": 0.6776,
      "step": 4431
    },
    {
      "epoch": 0.4089032406873486,
      "grad_norm": 1.7706508398410616,
      "learning_rate": 8.437836621071171e-07,
      "loss": 0.6414,
      "step": 4432
    },
    {
      "epoch": 0.40899550224887554,
      "grad_norm": 2.4259695608999476,
      "learning_rate": 8.436074099371938e-07,
      "loss": 0.7821,
      "step": 4433
    },
    {
      "epoch": 0.4090877638104025,
      "grad_norm": 2.524478691095849,
      "learning_rate": 8.434311379567231e-07,
      "loss": 0.7882,
      "step": 4434
    },
    {
      "epoch": 0.40918002537192943,
      "grad_norm": 1.722788404617185,
      "learning_rate": 8.432548461816788e-07,
      "loss": 0.6921,
      "step": 4435
    },
    {
      "epoch": 0.40918002537192943,
      "eval_GEN Loss": 0.4683658182621002,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.38892853260040283,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.890625,
      "eval_runtime": 56.9599,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 4435
    },
    {
      "epoch": 0.40927228693345635,
      "grad_norm": 1.7227399959850518,
      "learning_rate": 8.430785346280368e-07,
      "loss": 0.6833,
      "step": 4436
    },
    {
      "epoch": 0.40936454849498327,
      "grad_norm": 2.7846131252207473,
      "learning_rate": 8.429022033117748e-07,
      "loss": 1.0043,
      "step": 4437
    },
    {
      "epoch": 0.4094568100565102,
      "grad_norm": 2.058279534043985,
      "learning_rate": 8.42725852248872e-07,
      "loss": 0.7642,
      "step": 4438
    },
    {
      "epoch": 0.4095490716180371,
      "grad_norm": 2.8036887462122917,
      "learning_rate": 8.425494814553098e-07,
      "loss": 0.7572,
      "step": 4439
    },
    {
      "epoch": 0.4096413331795641,
      "grad_norm": 1.5305309942984735,
      "learning_rate": 8.423730909470712e-07,
      "loss": 0.7385,
      "step": 4440
    },
    {
      "epoch": 0.4096413331795641,
      "eval_GEN Loss": 0.46863633394241333,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8679245283018868,
      "eval_PRM F1": 0.9146341463414634,
      "eval_PRM F1 AUC": 0.8213724463069669,
      "eval_PRM F1 Neg": 0.7083333333333334,
      "eval_PRM Loss": 0.3912132680416107,
      "eval_PRM NPV": 0.68,
      "eval_PRM Precision": 0.9259259259259259,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8941707015037537,
      "eval_runtime": 55.83,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4440
    },
    {
      "epoch": 0.409733594741091,
      "grad_norm": 1.6620025701406809,
      "learning_rate": 8.421966807401409e-07,
      "loss": 0.6896,
      "step": 4441
    },
    {
      "epoch": 0.4098258563026179,
      "grad_norm": 2.3311673457438267,
      "learning_rate": 8.420202508505056e-07,
      "loss": 0.7278,
      "step": 4442
    },
    {
      "epoch": 0.40991811786414484,
      "grad_norm": 2.0409251633390197,
      "learning_rate": 8.418438012941535e-07,
      "loss": 0.6658,
      "step": 4443
    },
    {
      "epoch": 0.41001037942567176,
      "grad_norm": 2.7458604685369465,
      "learning_rate": 8.41667332087075e-07,
      "loss": 0.8577,
      "step": 4444
    },
    {
      "epoch": 0.41010264098719873,
      "grad_norm": 1.809262802391263,
      "learning_rate": 8.414908432452616e-07,
      "loss": 0.6504,
      "step": 4445
    },
    {
      "epoch": 0.41010264098719873,
      "eval_GEN Loss": 0.4685811698436737,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9079754601226994,
      "eval_PRM F1 AUC": 0.8153483499214248,
      "eval_PRM F1 Neg": 0.6938775510204082,
      "eval_PRM Loss": 0.39829114079475403,
      "eval_PRM NPV": 0.6538461538461539,
      "eval_PRM Precision": 0.925,
      "eval_PRM Recall": 0.891566265060241,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8966947197914124,
      "eval_runtime": 56.0773,
      "eval_samples_per_second": 1.159,
      "eval_steps_per_second": 0.089,
      "step": 4445
    },
    {
      "epoch": 0.41019490254872565,
      "grad_norm": 2.2159807326316163,
      "learning_rate": 8.413143347847073e-07,
      "loss": 0.8437,
      "step": 4446
    },
    {
      "epoch": 0.41028716411025257,
      "grad_norm": 1.815086239177233,
      "learning_rate": 8.411378067214075e-07,
      "loss": 0.7837,
      "step": 4447
    },
    {
      "epoch": 0.4103794256717795,
      "grad_norm": 1.8169003421483951,
      "learning_rate": 8.409612590713595e-07,
      "loss": 0.814,
      "step": 4448
    },
    {
      "epoch": 0.4104716872333064,
      "grad_norm": 1.5581465180457919,
      "learning_rate": 8.407846918505621e-07,
      "loss": 0.7114,
      "step": 4449
    },
    {
      "epoch": 0.4105639487948334,
      "grad_norm": 1.7951832995862964,
      "learning_rate": 8.406081050750164e-07,
      "loss": 0.7132,
      "step": 4450
    },
    {
      "epoch": 0.4105639487948334,
      "eval_GEN Loss": 0.4693397581577301,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9,
      "eval_PRM F1 AUC": 0.8250392875851232,
      "eval_PRM F1 Neg": 0.6923076923076923,
      "eval_PRM Loss": 0.4170929193496704,
      "eval_PRM NPV": 0.6206896551724138,
      "eval_PRM Precision": 0.935064935064935,
      "eval_PRM Recall": 0.8674698795180723,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9039062261581421,
      "eval_runtime": 55.981,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 4450
    },
    {
      "epoch": 0.4106562103563603,
      "grad_norm": 2.0050657877483835,
      "learning_rate": 8.40431498760725e-07,
      "loss": 0.6324,
      "step": 4451
    },
    {
      "epoch": 0.4107484719178872,
      "grad_norm": 2.1479722246348203,
      "learning_rate": 8.402548729236919e-07,
      "loss": 0.576,
      "step": 4452
    },
    {
      "epoch": 0.41084073347941413,
      "grad_norm": 2.2416873439526057,
      "learning_rate": 8.400782275799236e-07,
      "loss": 0.7593,
      "step": 4453
    },
    {
      "epoch": 0.41093299504094105,
      "grad_norm": 1.9037926701573586,
      "learning_rate": 8.399015627454274e-07,
      "loss": 0.7573,
      "step": 4454
    },
    {
      "epoch": 0.41102525660246797,
      "grad_norm": 2.282854272157789,
      "learning_rate": 8.397248784362136e-07,
      "loss": 0.8826,
      "step": 4455
    },
    {
      "epoch": 0.41102525660246797,
      "eval_GEN Loss": 0.4691080152988434,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.4137914180755615,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.905348539352417,
      "eval_runtime": 55.8514,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4455
    },
    {
      "epoch": 0.41111751816399494,
      "grad_norm": 2.0067909166136464,
      "learning_rate": 8.395481746682932e-07,
      "loss": 0.7274,
      "step": 4456
    },
    {
      "epoch": 0.41120977972552186,
      "grad_norm": 2.1478142640649573,
      "learning_rate": 8.393714514576797e-07,
      "loss": 0.8989,
      "step": 4457
    },
    {
      "epoch": 0.4113020412870488,
      "grad_norm": 3.3050958758486906,
      "learning_rate": 8.391947088203877e-07,
      "loss": 0.756,
      "step": 4458
    },
    {
      "epoch": 0.4113943028485757,
      "grad_norm": 1.4580142202762119,
      "learning_rate": 8.390179467724341e-07,
      "loss": 0.744,
      "step": 4459
    },
    {
      "epoch": 0.4114865644101026,
      "grad_norm": 2.115557227177532,
      "learning_rate": 8.388411653298373e-07,
      "loss": 0.7162,
      "step": 4460
    },
    {
      "epoch": 0.4114865644101026,
      "eval_GEN Loss": 0.46775931119918823,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.40120822191238403,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9041466116905212,
      "eval_runtime": 55.8126,
      "eval_samples_per_second": 1.165,
      "eval_steps_per_second": 0.09,
      "step": 4460
    },
    {
      "epoch": 0.4115788259716296,
      "grad_norm": 2.704404467676901,
      "learning_rate": 8.386643645086174e-07,
      "loss": 0.7605,
      "step": 4461
    },
    {
      "epoch": 0.4116710875331565,
      "grad_norm": 1.8194182036124926,
      "learning_rate": 8.384875443247965e-07,
      "loss": 0.7232,
      "step": 4462
    },
    {
      "epoch": 0.41176334909468343,
      "grad_norm": 1.4935980073225366,
      "learning_rate": 8.383107047943984e-07,
      "loss": 0.6165,
      "step": 4463
    },
    {
      "epoch": 0.41185561065621035,
      "grad_norm": 1.6614528684593,
      "learning_rate": 8.381338459334484e-07,
      "loss": 0.5703,
      "step": 4464
    },
    {
      "epoch": 0.41194787221773727,
      "grad_norm": 1.2808233276497625,
      "learning_rate": 8.379569677579738e-07,
      "loss": 0.5915,
      "step": 4465
    },
    {
      "epoch": 0.41194787221773727,
      "eval_GEN Loss": 0.46657660603523254,
      "eval_GEN top-5 accuracy": 0.9811486190267427,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.39101603627204895,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9128004908561707,
      "eval_runtime": 55.9853,
      "eval_samples_per_second": 1.161,
      "eval_steps_per_second": 0.089,
      "step": 4465
    },
    {
      "epoch": 0.41204013377926424,
      "grad_norm": 2.186578060323299,
      "learning_rate": 8.377800702840034e-07,
      "loss": 0.8973,
      "step": 4466
    },
    {
      "epoch": 0.41213239534079116,
      "grad_norm": 1.9318827540981358,
      "learning_rate": 8.376031535275685e-07,
      "loss": 0.662,
      "step": 4467
    },
    {
      "epoch": 0.4122246569023181,
      "grad_norm": 3.6880093608066518,
      "learning_rate": 8.374262175047008e-07,
      "loss": 0.9352,
      "step": 4468
    },
    {
      "epoch": 0.412316918463845,
      "grad_norm": 1.4699628358500672,
      "learning_rate": 8.372492622314351e-07,
      "loss": 0.4356,
      "step": 4469
    },
    {
      "epoch": 0.4124091800253719,
      "grad_norm": 2.2722085263208913,
      "learning_rate": 8.370722877238074e-07,
      "loss": 0.8579,
      "step": 4470
    },
    {
      "epoch": 0.4124091800253719,
      "eval_GEN Loss": 0.4665497839450836,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9036144578313253,
      "eval_PRM F1 AUC": 0.7778941854374019,
      "eval_PRM F1 Neg": 0.6521739130434783,
      "eval_PRM Loss": 0.39095500111579895,
      "eval_PRM NPV": 0.6521739130434783,
      "eval_PRM Precision": 0.9036144578313253,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6521739130434783,
      "eval_loss": 0.9246995449066162,
      "eval_runtime": 55.8651,
      "eval_samples_per_second": 1.164,
      "eval_steps_per_second": 0.09,
      "step": 4470
    },
    {
      "epoch": 0.41250144158689883,
      "grad_norm": 2.2356962555940334,
      "learning_rate": 8.368952939978551e-07,
      "loss": 0.833,
      "step": 4471
    },
    {
      "epoch": 0.4125937031484258,
      "grad_norm": 1.8751205604035595,
      "learning_rate": 8.367182810696174e-07,
      "loss": 0.6344,
      "step": 4472
    },
    {
      "epoch": 0.4126859647099527,
      "grad_norm": 3.220241121045629,
      "learning_rate": 8.365412489551361e-07,
      "loss": 0.9262,
      "step": 4473
    },
    {
      "epoch": 0.41277822627147964,
      "grad_norm": 1.3901277316331762,
      "learning_rate": 8.363641976704537e-07,
      "loss": 0.6542,
      "step": 4474
    },
    {
      "epoch": 0.41287048783300656,
      "grad_norm": 1.596711024441536,
      "learning_rate": 8.361871272316149e-07,
      "loss": 0.6882,
      "step": 4475
    },
    {
      "epoch": 0.41287048783300656,
      "eval_GEN Loss": 0.46529778838157654,
      "eval_GEN top-5 accuracy": 0.9820254274441035,
      "eval_PRM Accuracy": 0.8584905660377359,
      "eval_PRM F1": 0.9090909090909091,
      "eval_PRM F1 AUC": 0.7996333158721844,
      "eval_PRM F1 Neg": 0.6808510638297872,
      "eval_PRM Loss": 0.3983197808265686,
      "eval_PRM NPV": 0.6666666666666666,
      "eval_PRM Precision": 0.9146341463414634,
      "eval_PRM Recall": 0.9036144578313253,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9189903736114502,
      "eval_runtime": 55.9033,
      "eval_samples_per_second": 1.163,
      "eval_steps_per_second": 0.089,
      "step": 4475
    },
    {
      "epoch": 0.4129627493945335,
      "grad_norm": 1.6681143143459427,
      "learning_rate": 8.360100376546665e-07,
      "loss": 0.7299,
      "step": 4476
    },
    {
      "epoch": 0.41305501095606045,
      "grad_norm": 2.2424161084535594,
      "learning_rate": 8.358329289556562e-07,
      "loss": 0.673,
      "step": 4477
    },
    {
      "epoch": 0.41314727251758737,
      "grad_norm": 1.7662827683428823,
      "learning_rate": 8.356558011506341e-07,
      "loss": 0.7465,
      "step": 4478
    },
    {
      "epoch": 0.4132395340791143,
      "grad_norm": 3.055652146740076,
      "learning_rate": 8.354786542556515e-07,
      "loss": 0.8324,
      "step": 4479
    },
    {
      "epoch": 0.4133317956406412,
      "grad_norm": 1.488356917683893,
      "learning_rate": 8.353014882867623e-07,
      "loss": 0.4947,
      "step": 4480
    },
    {
      "epoch": 0.4133317956406412,
      "eval_GEN Loss": 0.46626776456832886,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.839622641509434,
      "eval_PRM F1": 0.8957055214723927,
      "eval_PRM F1 AUC": 0.7875851231011,
      "eval_PRM F1 Neg": 0.6530612244897959,
      "eval_PRM Loss": 0.40798261761665344,
      "eval_PRM NPV": 0.6153846153846154,
      "eval_PRM Precision": 0.9125,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.6956521739130435,
      "eval_loss": 0.9106971025466919,
      "eval_runtime": 56.9463,
      "eval_samples_per_second": 1.141,
      "eval_steps_per_second": 0.088,
      "step": 4480
    },
    {
      "epoch": 0.4134240572021681,
      "grad_norm": 1.7835403412196758,
      "learning_rate": 8.35124303260021e-07,
      "loss": 0.7146,
      "step": 4481
    },
    {
      "epoch": 0.4135163187636951,
      "grad_norm": 1.9581232939708986,
      "learning_rate": 8.349470991914846e-07,
      "loss": 0.77,
      "step": 4482
    },
    {
      "epoch": 0.413608580325222,
      "grad_norm": 1.5995569070562858,
      "learning_rate": 8.347698760972117e-07,
      "loss": 0.5544,
      "step": 4483
    },
    {
      "epoch": 0.41370084188674894,
      "grad_norm": 2.5398514864931605,
      "learning_rate": 8.345926339932624e-07,
      "loss": 0.8174,
      "step": 4484
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 1.5217580094923997,
      "learning_rate": 8.344153728956987e-07,
      "loss": 0.6762,
      "step": 4485
    },
    {
      "epoch": 0.41379310344827586,
      "eval_GEN Loss": 0.4669657051563263,
      "eval_GEN top-5 accuracy": 0.9824638316527838,
      "eval_PRM Accuracy": 0.8490566037735849,
      "eval_PRM F1": 0.9012345679012346,
      "eval_PRM F1 AUC": 0.8093242535358826,
      "eval_PRM F1 Neg": 0.68,
      "eval_PRM Loss": 0.41092225909233093,
      "eval_PRM NPV": 0.6296296296296297,
      "eval_PRM Precision": 0.9240506329113924,
      "eval_PRM Recall": 0.8795180722891566,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8997595906257629,
      "eval_runtime": 56.6675,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 4485
    },
    {
      "epoch": 0.4138853650098028,
      "grad_norm": 3.042243171827483,
      "learning_rate": 8.342380928205842e-07,
      "loss": 0.7689,
      "step": 4486
    },
    {
      "epoch": 0.4139776265713297,
      "grad_norm": 1.5485432781574762,
      "learning_rate": 8.340607937839845e-07,
      "loss": 0.8441,
      "step": 4487
    },
    {
      "epoch": 0.41406988813285667,
      "grad_norm": 1.756016456470045,
      "learning_rate": 8.338834758019666e-07,
      "loss": 0.6179,
      "step": 4488
    },
    {
      "epoch": 0.4141621496943836,
      "grad_norm": 2.4625968719934614,
      "learning_rate": 8.337061388905994e-07,
      "loss": 0.8448,
      "step": 4489
    },
    {
      "epoch": 0.4142544112559105,
      "grad_norm": 1.935119372912926,
      "learning_rate": 8.335287830659532e-07,
      "loss": 0.7571,
      "step": 4490
    },
    {
      "epoch": 0.4142544112559105,
      "eval_GEN Loss": 0.46851304173469543,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8717948717948718,
      "eval_PRM F1 AUC": 0.8009429020429545,
      "eval_PRM F1 Neg": 0.6428571428571429,
      "eval_PRM Loss": 0.4371475577354431,
      "eval_PRM NPV": 0.5454545454545454,
      "eval_PRM Precision": 0.9315068493150684,
      "eval_PRM Recall": 0.8192771084337349,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.9039062261581421,
      "eval_runtime": 56.6122,
      "eval_samples_per_second": 1.148,
      "eval_steps_per_second": 0.088,
      "step": 4490
    },
    {
      "epoch": 0.4143466728174374,
      "grad_norm": 2.2362447481289847,
      "learning_rate": 8.333514083441005e-07,
      "loss": 0.6513,
      "step": 4491
    },
    {
      "epoch": 0.41443893437896434,
      "grad_norm": 1.5410883047355917,
      "learning_rate": 8.331740147411154e-07,
      "loss": 0.6805,
      "step": 4492
    },
    {
      "epoch": 0.4145311959404913,
      "grad_norm": 2.7168498366099927,
      "learning_rate": 8.329966022730731e-07,
      "loss": 0.802,
      "step": 4493
    },
    {
      "epoch": 0.41462345750201823,
      "grad_norm": 2.572989119387312,
      "learning_rate": 8.328191709560517e-07,
      "loss": 0.8211,
      "step": 4494
    },
    {
      "epoch": 0.41471571906354515,
      "grad_norm": 2.550119540256824,
      "learning_rate": 8.326417208061296e-07,
      "loss": 0.5406,
      "step": 4495
    },
    {
      "epoch": 0.41471571906354515,
      "eval_GEN Loss": 0.4688943326473236,
      "eval_GEN top-5 accuracy": 0.981587023235423,
      "eval_PRM Accuracy": 0.8018867924528302,
      "eval_PRM F1": 0.864516129032258,
      "eval_PRM F1 AUC": 0.7949188056574124,
      "eval_PRM F1 Neg": 0.631578947368421,
      "eval_PRM Loss": 0.44194895029067993,
      "eval_PRM NPV": 0.5294117647058824,
      "eval_PRM Precision": 0.9305555555555556,
      "eval_PRM Recall": 0.8072289156626506,
      "eval_PRM Specificty": 0.782608695652174,
      "eval_loss": 0.8975961804389954,
      "eval_runtime": 56.8679,
      "eval_samples_per_second": 1.143,
      "eval_steps_per_second": 0.088,
      "step": 4495
    },
    {
      "epoch": 0.41480798062507207,
      "grad_norm": 2.7199340244458647,
      "learning_rate": 8.324642518393882e-07,
      "loss": 0.9646,
      "step": 4496
    },
    {
      "epoch": 0.414900242186599,
      "grad_norm": 1.891792622852327,
      "learning_rate": 8.322867640719094e-07,
      "loss": 0.8376,
      "step": 4497
    },
    {
      "epoch": 0.41499250374812596,
      "grad_norm": 2.6360234208519358,
      "learning_rate": 8.321092575197778e-07,
      "loss": 0.8396,
      "step": 4498
    },
    {
      "epoch": 0.4150847653096529,
      "grad_norm": 1.7398535721206243,
      "learning_rate": 8.319317321990796e-07,
      "loss": 0.668,
      "step": 4499
    },
    {
      "epoch": 0.4151770268711798,
      "grad_norm": 2.429001589129579,
      "learning_rate": 8.317541881259019e-07,
      "loss": 0.744,
      "step": 4500
    },
    {
      "epoch": 0.4151770268711798,
      "eval_GEN Loss": 0.46817731857299805,
      "eval_GEN top-5 accuracy": 0.9802718106093818,
      "eval_PRM Accuracy": 0.8113207547169812,
      "eval_PRM F1": 0.8734177215189873,
      "eval_PRM F1 AUC": 0.785227867993714,
      "eval_PRM F1 Neg": 0.6296296296296297,
      "eval_PRM Loss": 0.42457708716392517,
      "eval_PRM NPV": 0.5483870967741935,
      "eval_PRM Precision": 0.92,
      "eval_PRM Recall": 0.8313253012048193,
      "eval_PRM Specificty": 0.7391304347826086,
      "eval_loss": 0.8905048370361328,
      "eval_runtime": 56.6838,
      "eval_samples_per_second": 1.147,
      "eval_steps_per_second": 0.088,
      "step": 4500
    }
  ],
  "logging_steps": 1,
  "max_steps": 10838,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 141745119559680.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}