Files
deepseek-math-7B-finetuned/trainer_state.json

4221 lines
103 KiB
JSON
Raw Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 1000,
"global_step": 5910,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0016921905406548777,
"grad_norm": 10.694635585504944,
"learning_rate": 1.5202702702702704e-06,
"loss": 2.6757,
"step": 10
},
{
"epoch": 0.0033843810813097555,
"grad_norm": 5.468837725172288,
"learning_rate": 3.2094594594594594e-06,
"loss": 2.3662,
"step": 20
},
{
"epoch": 0.005076571621964634,
"grad_norm": 3.9786531895093464,
"learning_rate": 4.898648648648649e-06,
"loss": 1.7146,
"step": 30
},
{
"epoch": 0.006768762162619511,
"grad_norm": 4.092947192038235,
"learning_rate": 6.587837837837837e-06,
"loss": 1.1047,
"step": 40
},
{
"epoch": 0.008460952703274389,
"grad_norm": 1.2729172309708834,
"learning_rate": 8.277027027027027e-06,
"loss": 0.7679,
"step": 50
},
{
"epoch": 0.010153143243929267,
"grad_norm": 1.7680997715941034,
"learning_rate": 9.966216216216217e-06,
"loss": 0.5549,
"step": 60
},
{
"epoch": 0.011845333784584144,
"grad_norm": 1.1591949561413044,
"learning_rate": 1.1655405405405405e-05,
"loss": 0.47,
"step": 70
},
{
"epoch": 0.013537524325239022,
"grad_norm": 2.4431810338212294,
"learning_rate": 1.3344594594594596e-05,
"loss": 0.3916,
"step": 80
},
{
"epoch": 0.0152297148658939,
"grad_norm": 0.6200366624872841,
"learning_rate": 1.5033783783783784e-05,
"loss": 0.3494,
"step": 90
},
{
"epoch": 0.016921905406548778,
"grad_norm": 2.7144356083329577,
"learning_rate": 1.6722972972972976e-05,
"loss": 0.314,
"step": 100
},
{
"epoch": 0.018614095947203656,
"grad_norm": 10.919270559463541,
"learning_rate": 1.8412162162162163e-05,
"loss": 0.2798,
"step": 110
},
{
"epoch": 0.020306286487858535,
"grad_norm": 0.7816749283383329,
"learning_rate": 2.0101351351351353e-05,
"loss": 0.2524,
"step": 120
},
{
"epoch": 0.02199847702851341,
"grad_norm": 0.5347062004824426,
"learning_rate": 2.179054054054054e-05,
"loss": 0.2211,
"step": 130
},
{
"epoch": 0.023690667569168287,
"grad_norm": 0.550712559745511,
"learning_rate": 2.347972972972973e-05,
"loss": 0.194,
"step": 140
},
{
"epoch": 0.025382858109823166,
"grad_norm": 1.7649680694845573,
"learning_rate": 2.516891891891892e-05,
"loss": 0.1843,
"step": 150
},
{
"epoch": 0.027075048650478044,
"grad_norm": 0.28948675660488626,
"learning_rate": 2.685810810810811e-05,
"loss": 0.1682,
"step": 160
},
{
"epoch": 0.028767239191132922,
"grad_norm": 0.705365592374216,
"learning_rate": 2.8547297297297298e-05,
"loss": 0.1592,
"step": 170
},
{
"epoch": 0.0304594297317878,
"grad_norm": 0.2759031943476703,
"learning_rate": 3.0236486486486488e-05,
"loss": 0.1512,
"step": 180
},
{
"epoch": 0.03215162027244268,
"grad_norm": 0.5008419380887897,
"learning_rate": 3.192567567567568e-05,
"loss": 0.1409,
"step": 190
},
{
"epoch": 0.033843810813097557,
"grad_norm": 0.4082014798546979,
"learning_rate": 3.3614864864864865e-05,
"loss": 0.1369,
"step": 200
},
{
"epoch": 0.035536001353752435,
"grad_norm": 0.4331239703891574,
"learning_rate": 3.530405405405405e-05,
"loss": 0.1305,
"step": 210
},
{
"epoch": 0.03722819189440731,
"grad_norm": 0.7433873463817868,
"learning_rate": 3.6993243243243245e-05,
"loss": 0.1303,
"step": 220
},
{
"epoch": 0.03892038243506219,
"grad_norm": 0.16778442150676437,
"learning_rate": 3.868243243243244e-05,
"loss": 0.1246,
"step": 230
},
{
"epoch": 0.04061257297571707,
"grad_norm": 0.35113896210837,
"learning_rate": 4.0371621621621626e-05,
"loss": 0.1185,
"step": 240
},
{
"epoch": 0.04230476351637194,
"grad_norm": 0.3406408467196104,
"learning_rate": 4.206081081081081e-05,
"loss": 0.1161,
"step": 250
},
{
"epoch": 0.04399695405702682,
"grad_norm": 0.12184403312626964,
"learning_rate": 4.375e-05,
"loss": 0.1143,
"step": 260
},
{
"epoch": 0.0456891445976817,
"grad_norm": 0.48130830284213894,
"learning_rate": 4.543918918918919e-05,
"loss": 0.1107,
"step": 270
},
{
"epoch": 0.047381335138336575,
"grad_norm": 0.1397912854760459,
"learning_rate": 4.712837837837838e-05,
"loss": 0.1086,
"step": 280
},
{
"epoch": 0.04907352567899145,
"grad_norm": 0.21714992035410935,
"learning_rate": 4.881756756756757e-05,
"loss": 0.1065,
"step": 290
},
{
"epoch": 0.05076571621964633,
"grad_norm": 0.16001664743021524,
"learning_rate": 4.999996477042944e-05,
"loss": 0.1034,
"step": 300
},
{
"epoch": 0.05245790676030121,
"grad_norm": 0.8804255832512046,
"learning_rate": 4.9999338469715e-05,
"loss": 0.104,
"step": 310
},
{
"epoch": 0.05415009730095609,
"grad_norm": 0.9028193459885879,
"learning_rate": 4.999792931222984e-05,
"loss": 0.1031,
"step": 320
},
{
"epoch": 0.055842287841610966,
"grad_norm": 0.1319892308116438,
"learning_rate": 4.999573734210188e-05,
"loss": 0.1028,
"step": 330
},
{
"epoch": 0.057534478382265844,
"grad_norm": 0.15388338624325235,
"learning_rate": 4.999276262797288e-05,
"loss": 0.1011,
"step": 340
},
{
"epoch": 0.05922666892292072,
"grad_norm": 0.14484465011182138,
"learning_rate": 4.9989005262996316e-05,
"loss": 0.0967,
"step": 350
},
{
"epoch": 0.0609188594635756,
"grad_norm": 0.17756232882047332,
"learning_rate": 4.998446536483443e-05,
"loss": 0.0981,
"step": 360
},
{
"epoch": 0.06261105000423048,
"grad_norm": 0.12393175323273149,
"learning_rate": 4.99791430756546e-05,
"loss": 0.0964,
"step": 370
},
{
"epoch": 0.06430324054488536,
"grad_norm": 0.11847906667492378,
"learning_rate": 4.997303856212483e-05,
"loss": 0.0967,
"step": 380
},
{
"epoch": 0.06599543108554023,
"grad_norm": 0.11677513187207987,
"learning_rate": 4.996615201540857e-05,
"loss": 0.0937,
"step": 390
},
{
"epoch": 0.06768762162619511,
"grad_norm": 0.16786236289813727,
"learning_rate": 4.9958483651158726e-05,
"loss": 0.0933,
"step": 400
},
{
"epoch": 0.06937981216684999,
"grad_norm": 0.17570863177817475,
"learning_rate": 4.995003370951088e-05,
"loss": 0.0901,
"step": 410
},
{
"epoch": 0.07107200270750487,
"grad_norm": 0.4695745650589833,
"learning_rate": 4.994080245507578e-05,
"loss": 0.1135,
"step": 420
},
{
"epoch": 0.07276419324815975,
"grad_norm": 0.34206988428476703,
"learning_rate": 4.993079017693111e-05,
"loss": 0.1178,
"step": 430
},
{
"epoch": 0.07445638378881463,
"grad_norm": 0.11253495532967475,
"learning_rate": 4.991999718861233e-05,
"loss": 0.1006,
"step": 440
},
{
"epoch": 0.0761485743294695,
"grad_norm": 0.17020660653200337,
"learning_rate": 4.9908423828102955e-05,
"loss": 0.0928,
"step": 450
},
{
"epoch": 0.07784076487012438,
"grad_norm": 0.32652051080347055,
"learning_rate": 4.9896070457823906e-05,
"loss": 0.0948,
"step": 460
},
{
"epoch": 0.07953295541077926,
"grad_norm": 0.22826141624908175,
"learning_rate": 4.988293746462222e-05,
"loss": 0.0922,
"step": 470
},
{
"epoch": 0.08122514595143414,
"grad_norm": 0.08457851013030424,
"learning_rate": 4.986902525975887e-05,
"loss": 0.0902,
"step": 480
},
{
"epoch": 0.082917336492089,
"grad_norm": 0.13070319396995603,
"learning_rate": 4.9854334278895946e-05,
"loss": 0.0888,
"step": 490
},
{
"epoch": 0.08460952703274388,
"grad_norm": 0.08107677504927116,
"learning_rate": 4.9838864982082975e-05,
"loss": 0.0881,
"step": 500
},
{
"epoch": 0.08630171757339876,
"grad_norm": 0.07915509750705131,
"learning_rate": 4.982261785374251e-05,
"loss": 0.0866,
"step": 510
},
{
"epoch": 0.08799390811405364,
"grad_norm": 0.10289073933459165,
"learning_rate": 4.980559340265502e-05,
"loss": 0.0866,
"step": 520
},
{
"epoch": 0.08968609865470852,
"grad_norm": 0.07241250513416626,
"learning_rate": 4.978779216194285e-05,
"loss": 0.085,
"step": 530
},
{
"epoch": 0.0913782891953634,
"grad_norm": 0.16151179785966988,
"learning_rate": 4.976921468905364e-05,
"loss": 0.0837,
"step": 540
},
{
"epoch": 0.09307047973601827,
"grad_norm": 0.09934569083291507,
"learning_rate": 4.9749861565742815e-05,
"loss": 0.0851,
"step": 550
},
{
"epoch": 0.09476267027667315,
"grad_norm": 0.11266079537528949,
"learning_rate": 4.972973339805534e-05,
"loss": 0.0849,
"step": 560
},
{
"epoch": 0.09645486081732803,
"grad_norm": 0.10115984698126272,
"learning_rate": 4.970883081630678e-05,
"loss": 0.0839,
"step": 570
},
{
"epoch": 0.0981470513579829,
"grad_norm": 0.13143827636360655,
"learning_rate": 4.968715447506358e-05,
"loss": 0.0879,
"step": 580
},
{
"epoch": 0.09983924189863778,
"grad_norm": 0.11830270941860212,
"learning_rate": 4.9664705053122504e-05,
"loss": 0.0854,
"step": 590
},
{
"epoch": 0.10153143243929266,
"grad_norm": 0.12875728575977066,
"learning_rate": 4.9641483253489466e-05,
"loss": 0.0832,
"step": 600
},
{
"epoch": 0.10322362297994754,
"grad_norm": 0.10739564648963017,
"learning_rate": 4.961748980335739e-05,
"loss": 0.0833,
"step": 610
},
{
"epoch": 0.10491581352060242,
"grad_norm": 0.14309128949862512,
"learning_rate": 4.9592725454083574e-05,
"loss": 0.0838,
"step": 620
},
{
"epoch": 0.1066080040612573,
"grad_norm": 0.12267374041957275,
"learning_rate": 4.956719098116608e-05,
"loss": 0.0827,
"step": 630
},
{
"epoch": 0.10830019460191218,
"grad_norm": 0.08756197387140273,
"learning_rate": 4.9540887184219464e-05,
"loss": 0.0817,
"step": 640
},
{
"epoch": 0.10999238514256705,
"grad_norm": 0.08530144788542346,
"learning_rate": 4.951381488694974e-05,
"loss": 0.0833,
"step": 650
},
{
"epoch": 0.11168457568322193,
"grad_norm": 0.10786268889452347,
"learning_rate": 4.94859749371286e-05,
"loss": 0.0822,
"step": 660
},
{
"epoch": 0.11337676622387681,
"grad_norm": 0.13303735264657252,
"learning_rate": 4.9457368206566824e-05,
"loss": 0.0828,
"step": 670
},
{
"epoch": 0.11506895676453169,
"grad_norm": 0.10498614065579724,
"learning_rate": 4.942799559108704e-05,
"loss": 0.0838,
"step": 680
},
{
"epoch": 0.11676114730518657,
"grad_norm": 0.07539770457350665,
"learning_rate": 4.939785801049562e-05,
"loss": 0.0814,
"step": 690
},
{
"epoch": 0.11845333784584144,
"grad_norm": 0.10844883466530295,
"learning_rate": 4.9366956408553904e-05,
"loss": 0.0809,
"step": 700
},
{
"epoch": 0.12014552838649632,
"grad_norm": 0.12370988442629981,
"learning_rate": 4.933529175294862e-05,
"loss": 0.0817,
"step": 710
},
{
"epoch": 0.1218377189271512,
"grad_norm": 0.11743981033369338,
"learning_rate": 4.9302865035261605e-05,
"loss": 0.0784,
"step": 720
},
{
"epoch": 0.12352990946780608,
"grad_norm": 0.20181605466697428,
"learning_rate": 4.926967727093876e-05,
"loss": 0.0798,
"step": 730
},
{
"epoch": 0.12522210000846096,
"grad_norm": 0.08226166438986647,
"learning_rate": 4.923572949925822e-05,
"loss": 0.0793,
"step": 740
},
{
"epoch": 0.12691429054911582,
"grad_norm": 0.11591499828890973,
"learning_rate": 4.920102278329783e-05,
"loss": 0.078,
"step": 750
},
{
"epoch": 0.1286064810897707,
"grad_norm": 0.08463858496133438,
"learning_rate": 4.916555820990185e-05,
"loss": 0.0806,
"step": 760
},
{
"epoch": 0.13029867163042558,
"grad_norm": 0.0823183614063425,
"learning_rate": 4.912933688964694e-05,
"loss": 0.0791,
"step": 770
},
{
"epoch": 0.13199086217108047,
"grad_norm": 0.06903458377489168,
"learning_rate": 4.909235995680732e-05,
"loss": 0.0788,
"step": 780
},
{
"epoch": 0.13368305271173533,
"grad_norm": 0.06787693470278572,
"learning_rate": 4.905462856931935e-05,
"loss": 0.0804,
"step": 790
},
{
"epoch": 0.13537524325239023,
"grad_norm": 0.134721426792912,
"learning_rate": 4.9016143908745156e-05,
"loss": 0.0836,
"step": 800
},
{
"epoch": 0.1370674337930451,
"grad_norm": 0.0912981203210339,
"learning_rate": 4.897690718023572e-05,
"loss": 0.0827,
"step": 810
},
{
"epoch": 0.13875962433369998,
"grad_norm": 0.08369836162397677,
"learning_rate": 4.89369196124931e-05,
"loss": 0.0803,
"step": 820
},
{
"epoch": 0.14045181487435485,
"grad_norm": 0.169803701399313,
"learning_rate": 4.889618245773196e-05,
"loss": 0.0794,
"step": 830
},
{
"epoch": 0.14214400541500974,
"grad_norm": 0.07119075753357014,
"learning_rate": 4.885469699164036e-05,
"loss": 0.0802,
"step": 840
},
{
"epoch": 0.1438361959556646,
"grad_norm": 0.06351731211273606,
"learning_rate": 4.881246451333977e-05,
"loss": 0.0774,
"step": 850
},
{
"epoch": 0.1455283864963195,
"grad_norm": 0.06124274539973317,
"learning_rate": 4.8769486345344484e-05,
"loss": 0.0761,
"step": 860
},
{
"epoch": 0.14722057703697436,
"grad_norm": 0.07023551698086875,
"learning_rate": 4.872576383352008e-05,
"loss": 0.0765,
"step": 870
},
{
"epoch": 0.14891276757762925,
"grad_norm": 0.16009397731187566,
"learning_rate": 4.868129834704137e-05,
"loss": 0.0781,
"step": 880
},
{
"epoch": 0.15060495811828412,
"grad_norm": 0.06081894615883028,
"learning_rate": 4.8636091278349496e-05,
"loss": 0.0766,
"step": 890
},
{
"epoch": 0.152297148658939,
"grad_norm": 0.06452658575654371,
"learning_rate": 4.85901440431083e-05,
"loss": 0.0761,
"step": 900
},
{
"epoch": 0.15398933919959387,
"grad_norm": 0.08278811218744207,
"learning_rate": 4.8543458080160026e-05,
"loss": 0.0769,
"step": 910
},
{
"epoch": 0.15568152974024876,
"grad_norm": 0.06287734218381943,
"learning_rate": 4.8496034851480254e-05,
"loss": 0.0775,
"step": 920
},
{
"epoch": 0.15737372028090363,
"grad_norm": 0.058151816865827446,
"learning_rate": 4.84478758421321e-05,
"loss": 0.0764,
"step": 930
},
{
"epoch": 0.15906591082155852,
"grad_norm": 0.07784539609339237,
"learning_rate": 4.839898256021974e-05,
"loss": 0.0746,
"step": 940
},
{
"epoch": 0.16075810136221338,
"grad_norm": 0.05880857043420918,
"learning_rate": 4.834935653684114e-05,
"loss": 0.0764,
"step": 950
},
{
"epoch": 0.16245029190286828,
"grad_norm": 0.06575591069801254,
"learning_rate": 4.829899932604018e-05,
"loss": 0.0752,
"step": 960
},
{
"epoch": 0.16414248244352314,
"grad_norm": 0.07909220036883742,
"learning_rate": 4.824791250475792e-05,
"loss": 0.0778,
"step": 970
},
{
"epoch": 0.165834672984178,
"grad_norm": 0.06518291660732925,
"learning_rate": 4.819609767278325e-05,
"loss": 0.0752,
"step": 980
},
{
"epoch": 0.1675268635248329,
"grad_norm": 0.06170503836015974,
"learning_rate": 4.814355645270277e-05,
"loss": 0.0742,
"step": 990
},
{
"epoch": 0.16921905406548776,
"grad_norm": 0.07594097779988289,
"learning_rate": 4.8090290489850045e-05,
"loss": 0.0744,
"step": 1000
},
{
"epoch": 0.16921905406548776,
"eval_loss": 0.07673268765211105,
"eval_runtime": 511.1676,
"eval_samples_per_second": 11.417,
"eval_steps_per_second": 0.714,
"step": 1000
},
{
"epoch": 0.17091124460614265,
"grad_norm": 0.06564849989119345,
"learning_rate": 4.803630145225397e-05,
"loss": 0.0748,
"step": 1010
},
{
"epoch": 0.17260343514679752,
"grad_norm": 0.06841035995921553,
"learning_rate": 4.798159103058665e-05,
"loss": 0.0773,
"step": 1020
},
{
"epoch": 0.1742956256874524,
"grad_norm": 0.06831669570907373,
"learning_rate": 4.7926160938110354e-05,
"loss": 0.0745,
"step": 1030
},
{
"epoch": 0.17598781622810727,
"grad_norm": 0.06972114399837329,
"learning_rate": 4.7870012910623966e-05,
"loss": 0.0747,
"step": 1040
},
{
"epoch": 0.17768000676876217,
"grad_norm": 0.05638352782721958,
"learning_rate": 4.781314870640854e-05,
"loss": 0.0744,
"step": 1050
},
{
"epoch": 0.17937219730941703,
"grad_norm": 0.061738079578726124,
"learning_rate": 4.77555701061723e-05,
"loss": 0.0731,
"step": 1060
},
{
"epoch": 0.18106438785007192,
"grad_norm": 0.14191313167025663,
"learning_rate": 4.769727891299481e-05,
"loss": 0.0751,
"step": 1070
},
{
"epoch": 0.1827565783907268,
"grad_norm": 0.06309678507754822,
"learning_rate": 4.763827695227061e-05,
"loss": 0.0753,
"step": 1080
},
{
"epoch": 0.18444876893138168,
"grad_norm": 0.08105644186847036,
"learning_rate": 4.7578566071651966e-05,
"loss": 0.0729,
"step": 1090
},
{
"epoch": 0.18614095947203654,
"grad_norm": 0.0816054498455632,
"learning_rate": 4.751814814099104e-05,
"loss": 0.0748,
"step": 1100
},
{
"epoch": 0.18783315001269144,
"grad_norm": 0.06011302998196863,
"learning_rate": 4.7457025052281325e-05,
"loss": 0.0752,
"step": 1110
},
{
"epoch": 0.1895253405533463,
"grad_norm": 0.06819034137917843,
"learning_rate": 4.7395198719598444e-05,
"loss": 0.0746,
"step": 1120
},
{
"epoch": 0.1912175310940012,
"grad_norm": 0.06686364961497757,
"learning_rate": 4.733267107904014e-05,
"loss": 0.0746,
"step": 1130
},
{
"epoch": 0.19290972163465606,
"grad_norm": 0.06531159650073481,
"learning_rate": 4.726944408866571e-05,
"loss": 0.0736,
"step": 1140
},
{
"epoch": 0.19460191217531095,
"grad_norm": 0.09837590090024338,
"learning_rate": 4.720551972843463e-05,
"loss": 0.0737,
"step": 1150
},
{
"epoch": 0.1962941027159658,
"grad_norm": 0.056392384436672174,
"learning_rate": 4.7140900000144614e-05,
"loss": 0.0744,
"step": 1160
},
{
"epoch": 0.1979862932566207,
"grad_norm": 0.05608738724669037,
"learning_rate": 4.707558692736888e-05,
"loss": 0.0761,
"step": 1170
},
{
"epoch": 0.19967848379727557,
"grad_norm": 0.058672403290893446,
"learning_rate": 4.7009582555392784e-05,
"loss": 0.0745,
"step": 1180
},
{
"epoch": 0.20137067433793046,
"grad_norm": 0.07513105765503325,
"learning_rate": 4.694288895114982e-05,
"loss": 0.0778,
"step": 1190
},
{
"epoch": 0.20306286487858533,
"grad_norm": 0.05908390351829059,
"learning_rate": 4.687550820315682e-05,
"loss": 0.0743,
"step": 1200
},
{
"epoch": 0.20475505541924022,
"grad_norm": 0.08250715072702297,
"learning_rate": 4.680744242144861e-05,
"loss": 0.0731,
"step": 1210
},
{
"epoch": 0.20644724595989508,
"grad_norm": 0.19875652662247484,
"learning_rate": 4.673869373751193e-05,
"loss": 0.0759,
"step": 1220
},
{
"epoch": 0.20813943650054997,
"grad_norm": 0.06584054043670805,
"learning_rate": 4.666926430421864e-05,
"loss": 0.0738,
"step": 1230
},
{
"epoch": 0.20983162704120484,
"grad_norm": 0.06621861948935144,
"learning_rate": 4.659915629575836e-05,
"loss": 0.0745,
"step": 1240
},
{
"epoch": 0.21152381758185973,
"grad_norm": 0.3973719341157724,
"learning_rate": 4.652837190757033e-05,
"loss": 0.0754,
"step": 1250
},
{
"epoch": 0.2132160081225146,
"grad_norm": 0.06578070678107657,
"learning_rate": 4.645691335627473e-05,
"loss": 0.0753,
"step": 1260
},
{
"epoch": 0.2149081986631695,
"grad_norm": 0.13407195184232457,
"learning_rate": 4.6384782879603195e-05,
"loss": 0.0763,
"step": 1270
},
{
"epoch": 0.21660038920382435,
"grad_norm": 0.07749999972151454,
"learning_rate": 4.6311982736328774e-05,
"loss": 0.0749,
"step": 1280
},
{
"epoch": 0.21829257974447921,
"grad_norm": 0.1348296648759177,
"learning_rate": 4.623851520619519e-05,
"loss": 0.0718,
"step": 1290
},
{
"epoch": 0.2199847702851341,
"grad_norm": 0.06412351475313544,
"learning_rate": 4.616438258984546e-05,
"loss": 0.0716,
"step": 1300
},
{
"epoch": 0.22167696082578897,
"grad_norm": 0.07773080866566949,
"learning_rate": 4.608958720874984e-05,
"loss": 0.0761,
"step": 1310
},
{
"epoch": 0.22336915136644386,
"grad_norm": 0.058113920630413575,
"learning_rate": 4.601413140513312e-05,
"loss": 0.0719,
"step": 1320
},
{
"epoch": 0.22506134190709873,
"grad_norm": 0.05046319118860472,
"learning_rate": 4.59380175419013e-05,
"loss": 0.0712,
"step": 1330
},
{
"epoch": 0.22675353244775362,
"grad_norm": 0.08020181422226004,
"learning_rate": 4.586124800256759e-05,
"loss": 0.0732,
"step": 1340
},
{
"epoch": 0.22844572298840848,
"grad_norm": 0.054965711497122836,
"learning_rate": 4.5783825191177744e-05,
"loss": 0.0717,
"step": 1350
},
{
"epoch": 0.23013791352906338,
"grad_norm": 0.05321613596916164,
"learning_rate": 4.5705751532234783e-05,
"loss": 0.0721,
"step": 1360
},
{
"epoch": 0.23183010406971824,
"grad_norm": 0.06049740319415766,
"learning_rate": 4.562702947062313e-05,
"loss": 0.0708,
"step": 1370
},
{
"epoch": 0.23352229461037313,
"grad_norm": 0.04982219999527112,
"learning_rate": 4.554766147153195e-05,
"loss": 0.0707,
"step": 1380
},
{
"epoch": 0.235214485151028,
"grad_norm": 0.05701663143844104,
"learning_rate": 4.546765002037804e-05,
"loss": 0.0697,
"step": 1390
},
{
"epoch": 0.2369066756916829,
"grad_norm": 0.053103231318294586,
"learning_rate": 4.538699762272795e-05,
"loss": 0.0723,
"step": 1400
},
{
"epoch": 0.23859886623233775,
"grad_norm": 0.055725823646977125,
"learning_rate": 4.530570680421955e-05,
"loss": 0.0721,
"step": 1410
},
{
"epoch": 0.24029105677299264,
"grad_norm": 0.057250590847969914,
"learning_rate": 4.522378011048288e-05,
"loss": 0.0725,
"step": 1420
},
{
"epoch": 0.2419832473136475,
"grad_norm": 0.05095728407569466,
"learning_rate": 4.514122010706052e-05,
"loss": 0.0721,
"step": 1430
},
{
"epoch": 0.2436754378543024,
"grad_norm": 0.06212821775161869,
"learning_rate": 4.5058029379327185e-05,
"loss": 0.0709,
"step": 1440
},
{
"epoch": 0.24536762839495727,
"grad_norm": 0.21631049482013637,
"learning_rate": 4.497421053240879e-05,
"loss": 0.0706,
"step": 1450
},
{
"epoch": 0.24705981893561216,
"grad_norm": 0.07565802363631137,
"learning_rate": 4.488976619110086e-05,
"loss": 0.0712,
"step": 1460
},
{
"epoch": 0.24875200947626702,
"grad_norm": 0.06253795088772365,
"learning_rate": 4.480469899978633e-05,
"loss": 0.0712,
"step": 1470
},
{
"epoch": 0.2504442000169219,
"grad_norm": 0.05364846922936465,
"learning_rate": 4.471901162235278e-05,
"loss": 0.0736,
"step": 1480
},
{
"epoch": 0.2521363905575768,
"grad_norm": 0.06637650481807221,
"learning_rate": 4.463270674210896e-05,
"loss": 0.0694,
"step": 1490
},
{
"epoch": 0.25382858109823164,
"grad_norm": 0.07736235493242273,
"learning_rate": 4.454578706170075e-05,
"loss": 0.0713,
"step": 1500
},
{
"epoch": 0.25552077163888653,
"grad_norm": 0.04937058370172276,
"learning_rate": 4.445825530302661e-05,
"loss": 0.0704,
"step": 1510
},
{
"epoch": 0.2572129621795414,
"grad_norm": 0.0722464540291525,
"learning_rate": 4.437011420715226e-05,
"loss": 0.0709,
"step": 1520
},
{
"epoch": 0.2589051527201963,
"grad_norm": 0.05726634753272463,
"learning_rate": 4.4281366534224886e-05,
"loss": 0.0722,
"step": 1530
},
{
"epoch": 0.26059734326085116,
"grad_norm": 0.059314086608780604,
"learning_rate": 4.41920150633867e-05,
"loss": 0.0735,
"step": 1540
},
{
"epoch": 0.26228953380150605,
"grad_norm": 0.0652600707198846,
"learning_rate": 4.410206259268789e-05,
"loss": 0.072,
"step": 1550
},
{
"epoch": 0.26398172434216094,
"grad_norm": 0.05188864537865943,
"learning_rate": 4.401151193899903e-05,
"loss": 0.0702,
"step": 1560
},
{
"epoch": 0.26567391488281583,
"grad_norm": 0.05327487452586731,
"learning_rate": 4.392036593792283e-05,
"loss": 0.0717,
"step": 1570
},
{
"epoch": 0.26736610542347067,
"grad_norm": 0.0675398562148978,
"learning_rate": 4.3828627443705405e-05,
"loss": 0.0704,
"step": 1580
},
{
"epoch": 0.26905829596412556,
"grad_norm": 0.053079769376812545,
"learning_rate": 4.3736299329146806e-05,
"loss": 0.0693,
"step": 1590
},
{
"epoch": 0.27075048650478045,
"grad_norm": 0.05516421758288954,
"learning_rate": 4.364338448551113e-05,
"loss": 0.0699,
"step": 1600
},
{
"epoch": 0.2724426770454353,
"grad_norm": 0.0703312957568337,
"learning_rate": 4.354988582243595e-05,
"loss": 0.0712,
"step": 1610
},
{
"epoch": 0.2741348675860902,
"grad_norm": 0.06053705624900413,
"learning_rate": 4.3455806267841204e-05,
"loss": 0.0702,
"step": 1620
},
{
"epoch": 0.2758270581267451,
"grad_norm": 0.08840827741734174,
"learning_rate": 4.3361148767837486e-05,
"loss": 0.0704,
"step": 1630
},
{
"epoch": 0.27751924866739996,
"grad_norm": 0.07137187853494739,
"learning_rate": 4.3265916286633834e-05,
"loss": 0.072,
"step": 1640
},
{
"epoch": 0.2792114392080548,
"grad_norm": 0.06084132746936776,
"learning_rate": 4.317011180644487e-05,
"loss": 0.0694,
"step": 1650
},
{
"epoch": 0.2809036297487097,
"grad_norm": 0.07940918458739425,
"learning_rate": 4.3073738327397416e-05,
"loss": 0.0713,
"step": 1660
},
{
"epoch": 0.2825958202893646,
"grad_norm": 0.06286093659263682,
"learning_rate": 4.297679886743656e-05,
"loss": 0.0703,
"step": 1670
},
{
"epoch": 0.2842880108300195,
"grad_norm": 0.05479160831316014,
"learning_rate": 4.2879296462231135e-05,
"loss": 0.0701,
"step": 1680
},
{
"epoch": 0.2859802013706743,
"grad_norm": 0.05322078581119646,
"learning_rate": 4.278123416507864e-05,
"loss": 0.0699,
"step": 1690
},
{
"epoch": 0.2876723919113292,
"grad_norm": 0.06654438784616207,
"learning_rate": 4.268261504680968e-05,
"loss": 0.0692,
"step": 1700
},
{
"epoch": 0.2893645824519841,
"grad_norm": 0.06983098565171621,
"learning_rate": 4.258344219569175e-05,
"loss": 0.0725,
"step": 1710
},
{
"epoch": 0.291056772992639,
"grad_norm": 0.04844618892455936,
"learning_rate": 4.248371871733254e-05,
"loss": 0.0699,
"step": 1720
},
{
"epoch": 0.2927489635332938,
"grad_norm": 0.05616734118678198,
"learning_rate": 4.23834477345827e-05,
"loss": 0.0705,
"step": 1730
},
{
"epoch": 0.2944411540739487,
"grad_norm": 0.045709943185862495,
"learning_rate": 4.2282632387438035e-05,
"loss": 0.0675,
"step": 1740
},
{
"epoch": 0.2961333446146036,
"grad_norm": 0.054558690456903106,
"learning_rate": 4.218127583294116e-05,
"loss": 0.0691,
"step": 1750
},
{
"epoch": 0.2978255351552585,
"grad_norm": 0.04863778129988292,
"learning_rate": 4.207938124508266e-05,
"loss": 0.0701,
"step": 1760
},
{
"epoch": 0.29951772569591334,
"grad_norm": 0.05135137708341738,
"learning_rate": 4.197695181470168e-05,
"loss": 0.0706,
"step": 1770
},
{
"epoch": 0.30120991623656823,
"grad_norm": 0.058337351283158756,
"learning_rate": 4.187399074938604e-05,
"loss": 0.0695,
"step": 1780
},
{
"epoch": 0.3029021067772231,
"grad_norm": 0.04656843018904869,
"learning_rate": 4.1770501273371724e-05,
"loss": 0.0683,
"step": 1790
},
{
"epoch": 0.304594297317878,
"grad_norm": 0.050754466580127326,
"learning_rate": 4.166648662744198e-05,
"loss": 0.068,
"step": 1800
},
{
"epoch": 0.30628648785853285,
"grad_norm": 0.05242780883524395,
"learning_rate": 4.156195006882578e-05,
"loss": 0.0683,
"step": 1810
},
{
"epoch": 0.30797867839918774,
"grad_norm": 0.06100798456746341,
"learning_rate": 4.145689487109586e-05,
"loss": 0.0669,
"step": 1820
},
{
"epoch": 0.30967086893984264,
"grad_norm": 0.062434160562195674,
"learning_rate": 4.13513243240662e-05,
"loss": 0.0705,
"step": 1830
},
{
"epoch": 0.31136305948049753,
"grad_norm": 0.16671882401373178,
"learning_rate": 4.124524173368894e-05,
"loss": 0.0682,
"step": 1840
},
{
"epoch": 0.31305525002115236,
"grad_norm": 0.060801114321219464,
"learning_rate": 4.1138650421951e-05,
"loss": 0.0689,
"step": 1850
},
{
"epoch": 0.31474744056180726,
"grad_norm": 0.0618662558280838,
"learning_rate": 4.1031553726769875e-05,
"loss": 0.0698,
"step": 1860
},
{
"epoch": 0.31643963110246215,
"grad_norm": 0.04900895328350808,
"learning_rate": 4.0923955001889236e-05,
"loss": 0.069,
"step": 1870
},
{
"epoch": 0.31813182164311704,
"grad_norm": 0.05257117328321432,
"learning_rate": 4.0815857616773856e-05,
"loss": 0.0686,
"step": 1880
},
{
"epoch": 0.3198240121837719,
"grad_norm": 0.048444286213805594,
"learning_rate": 4.07072649565041e-05,
"loss": 0.0691,
"step": 1890
},
{
"epoch": 0.32151620272442677,
"grad_norm": 0.05507475885048762,
"learning_rate": 4.059818042166993e-05,
"loss": 0.069,
"step": 1900
},
{
"epoch": 0.32320839326508166,
"grad_norm": 0.05180762836176336,
"learning_rate": 4.0488607428264395e-05,
"loss": 0.0678,
"step": 1910
},
{
"epoch": 0.32490058380573655,
"grad_norm": 0.06036750076898302,
"learning_rate": 4.037854940757668e-05,
"loss": 0.0681,
"step": 1920
},
{
"epoch": 0.3265927743463914,
"grad_norm": 0.04558721388234883,
"learning_rate": 4.026800980608467e-05,
"loss": 0.0681,
"step": 1930
},
{
"epoch": 0.3282849648870463,
"grad_norm": 0.05756573141073445,
"learning_rate": 4.0156992085346965e-05,
"loss": 0.0708,
"step": 1940
},
{
"epoch": 0.3299771554277012,
"grad_norm": 0.04849131348131361,
"learning_rate": 4.004549972189454e-05,
"loss": 0.0688,
"step": 1950
},
{
"epoch": 0.331669345968356,
"grad_norm": 0.04456546119692637,
"learning_rate": 3.9933536207121834e-05,
"loss": 0.0672,
"step": 1960
},
{
"epoch": 0.3333615365090109,
"grad_norm": 0.04787607056647239,
"learning_rate": 3.982110504717746e-05,
"loss": 0.0675,
"step": 1970
},
{
"epoch": 0.3350537270496658,
"grad_norm": 0.05123648792416722,
"learning_rate": 3.9708209762854364e-05,
"loss": 0.0679,
"step": 1980
},
{
"epoch": 0.3367459175903207,
"grad_norm": 0.04948133371753341,
"learning_rate": 3.959485388947961e-05,
"loss": 0.0673,
"step": 1990
},
{
"epoch": 0.3384381081309755,
"grad_norm": 0.055543461509569875,
"learning_rate": 3.948104097680365e-05,
"loss": 0.068,
"step": 2000
},
{
"epoch": 0.3384381081309755,
"eval_loss": 0.068834587931633,
"eval_runtime": 510.4033,
"eval_samples_per_second": 11.434,
"eval_steps_per_second": 0.715,
"step": 2000
},
{
"epoch": 0.3401302986716304,
"grad_norm": 0.049787944565199964,
"learning_rate": 3.936677458888915e-05,
"loss": 0.0674,
"step": 2010
},
{
"epoch": 0.3418224892122853,
"grad_norm": 0.05766782096468548,
"learning_rate": 3.925205830399944e-05,
"loss": 0.0693,
"step": 2020
},
{
"epoch": 0.3435146797529402,
"grad_norm": 0.052150108059753564,
"learning_rate": 3.913689571448638e-05,
"loss": 0.0695,
"step": 2030
},
{
"epoch": 0.34520687029359504,
"grad_norm": 0.051279806071734366,
"learning_rate": 3.902129042667789e-05,
"loss": 0.067,
"step": 2040
},
{
"epoch": 0.34689906083424993,
"grad_norm": 0.04703167897556351,
"learning_rate": 3.89052460607651e-05,
"loss": 0.0671,
"step": 2050
},
{
"epoch": 0.3485912513749048,
"grad_norm": 0.08011314324337529,
"learning_rate": 3.878876625068883e-05,
"loss": 0.0684,
"step": 2060
},
{
"epoch": 0.3502834419155597,
"grad_norm": 0.0599016423669008,
"learning_rate": 3.867185464402594e-05,
"loss": 0.0666,
"step": 2070
},
{
"epoch": 0.35197563245621455,
"grad_norm": 0.05624253465950071,
"learning_rate": 3.855451490187503e-05,
"loss": 0.0679,
"step": 2080
},
{
"epoch": 0.35366782299686944,
"grad_norm": 0.04860541604617138,
"learning_rate": 3.843675069874177e-05,
"loss": 0.0653,
"step": 2090
},
{
"epoch": 0.35536001353752433,
"grad_norm": 0.06061045632387045,
"learning_rate": 3.831856572242392e-05,
"loss": 0.0676,
"step": 2100
},
{
"epoch": 0.3570522040781792,
"grad_norm": 0.08405843059471316,
"learning_rate": 3.8199963673895746e-05,
"loss": 0.0687,
"step": 2110
},
{
"epoch": 0.35874439461883406,
"grad_norm": 0.05094252500973247,
"learning_rate": 3.808094826719221e-05,
"loss": 0.067,
"step": 2120
},
{
"epoch": 0.36043658515948895,
"grad_norm": 0.04747569632830337,
"learning_rate": 3.796152322929259e-05,
"loss": 0.0678,
"step": 2130
},
{
"epoch": 0.36212877570014385,
"grad_norm": 0.043989513939970984,
"learning_rate": 3.784169230000386e-05,
"loss": 0.067,
"step": 2140
},
{
"epoch": 0.36382096624079874,
"grad_norm": 0.0641170858011972,
"learning_rate": 3.772145923184347e-05,
"loss": 0.0671,
"step": 2150
},
{
"epoch": 0.3655131567814536,
"grad_norm": 0.058065017298541036,
"learning_rate": 3.760082778992191e-05,
"loss": 0.0657,
"step": 2160
},
{
"epoch": 0.36720534732210847,
"grad_norm": 0.06762932993491448,
"learning_rate": 3.747980175182478e-05,
"loss": 0.0675,
"step": 2170
},
{
"epoch": 0.36889753786276336,
"grad_norm": 0.050683973509453334,
"learning_rate": 3.73583849074945e-05,
"loss": 0.0687,
"step": 2180
},
{
"epoch": 0.37058972840341825,
"grad_norm": 0.05691150704673455,
"learning_rate": 3.723658105911163e-05,
"loss": 0.067,
"step": 2190
},
{
"epoch": 0.3722819189440731,
"grad_norm": 0.049353064847232904,
"learning_rate": 3.71143940209758e-05,
"loss": 0.0661,
"step": 2200
},
{
"epoch": 0.373974109484728,
"grad_norm": 0.04522483534797012,
"learning_rate": 3.699182761938625e-05,
"loss": 0.0658,
"step": 2210
},
{
"epoch": 0.37566630002538287,
"grad_norm": 0.04990413273190803,
"learning_rate": 3.6868885692522057e-05,
"loss": 0.066,
"step": 2220
},
{
"epoch": 0.37735849056603776,
"grad_norm": 0.04573719324320986,
"learning_rate": 3.674557209032187e-05,
"loss": 0.0656,
"step": 2230
},
{
"epoch": 0.3790506811066926,
"grad_norm": 0.050884676936902004,
"learning_rate": 3.6621890674363426e-05,
"loss": 0.0667,
"step": 2240
},
{
"epoch": 0.3807428716473475,
"grad_norm": 0.055475437641100425,
"learning_rate": 3.649784531774254e-05,
"loss": 0.0661,
"step": 2250
},
{
"epoch": 0.3824350621880024,
"grad_norm": 0.05714257795792713,
"learning_rate": 3.637343990495192e-05,
"loss": 0.0649,
"step": 2260
},
{
"epoch": 0.3841272527286572,
"grad_norm": 0.04538322983324413,
"learning_rate": 3.624867833175943e-05,
"loss": 0.0659,
"step": 2270
},
{
"epoch": 0.3858194432693121,
"grad_norm": 0.0775738037403566,
"learning_rate": 3.6123564505086135e-05,
"loss": 0.0684,
"step": 2280
},
{
"epoch": 0.387511633809967,
"grad_norm": 0.05103392966267644,
"learning_rate": 3.599810234288399e-05,
"loss": 0.0676,
"step": 2290
},
{
"epoch": 0.3892038243506219,
"grad_norm": 0.04160967674017053,
"learning_rate": 3.587229577401305e-05,
"loss": 0.0654,
"step": 2300
},
{
"epoch": 0.39089601489127673,
"grad_norm": 0.050046338195732655,
"learning_rate": 3.574614873811855e-05,
"loss": 0.0661,
"step": 2310
},
{
"epoch": 0.3925882054319316,
"grad_norm": 0.042803100772230124,
"learning_rate": 3.5619665185507464e-05,
"loss": 0.065,
"step": 2320
},
{
"epoch": 0.3942803959725865,
"grad_norm": 0.04098179977984561,
"learning_rate": 3.549284907702481e-05,
"loss": 0.0672,
"step": 2330
},
{
"epoch": 0.3959725865132414,
"grad_norm": 0.0833163915825678,
"learning_rate": 3.536570438392967e-05,
"loss": 0.0683,
"step": 2340
},
{
"epoch": 0.39766477705389625,
"grad_norm": 0.0683950887033814,
"learning_rate": 3.523823508777071e-05,
"loss": 0.0674,
"step": 2350
},
{
"epoch": 0.39935696759455114,
"grad_norm": 0.04505033643926753,
"learning_rate": 3.511044518026167e-05,
"loss": 0.0671,
"step": 2360
},
{
"epoch": 0.40104915813520603,
"grad_norm": 0.059210977545472145,
"learning_rate": 3.4982338663156165e-05,
"loss": 0.068,
"step": 2370
},
{
"epoch": 0.4027413486758609,
"grad_norm": 0.04398059690534138,
"learning_rate": 3.485391954812255e-05,
"loss": 0.0687,
"step": 2380
},
{
"epoch": 0.40443353921651576,
"grad_norm": 0.06861155312080358,
"learning_rate": 3.4725191856618175e-05,
"loss": 0.0664,
"step": 2390
},
{
"epoch": 0.40612572975717065,
"grad_norm": 0.10610432826858647,
"learning_rate": 3.4596159619763505e-05,
"loss": 0.0656,
"step": 2400
},
{
"epoch": 0.40781792029782554,
"grad_norm": 0.06686306525419287,
"learning_rate": 3.4466826878215873e-05,
"loss": 0.0667,
"step": 2410
},
{
"epoch": 0.40951011083848043,
"grad_norm": 0.08182070866352631,
"learning_rate": 3.433719768204294e-05,
"loss": 0.0678,
"step": 2420
},
{
"epoch": 0.41120230137913527,
"grad_norm": 0.04548575127896905,
"learning_rate": 3.4207276090595894e-05,
"loss": 0.0657,
"step": 2430
},
{
"epoch": 0.41289449191979016,
"grad_norm": 0.051588152232755674,
"learning_rate": 3.407706617238226e-05,
"loss": 0.0651,
"step": 2440
},
{
"epoch": 0.41458668246044506,
"grad_norm": 0.04271294534713993,
"learning_rate": 3.394657200493861e-05,
"loss": 0.0669,
"step": 2450
},
{
"epoch": 0.41627887300109995,
"grad_norm": 0.06272198755653845,
"learning_rate": 3.381579767470277e-05,
"loss": 0.0651,
"step": 2460
},
{
"epoch": 0.4179710635417548,
"grad_norm": 0.060222498731354906,
"learning_rate": 3.3684747276885877e-05,
"loss": 0.0656,
"step": 2470
},
{
"epoch": 0.4196632540824097,
"grad_norm": 0.04518261992722282,
"learning_rate": 3.35534249153442e-05,
"loss": 0.0662,
"step": 2480
},
{
"epoch": 0.42135544462306457,
"grad_norm": 0.044678753192329106,
"learning_rate": 3.342183470245054e-05,
"loss": 0.0653,
"step": 2490
},
{
"epoch": 0.42304763516371946,
"grad_norm": 0.05059086049314136,
"learning_rate": 3.328998075896551e-05,
"loss": 0.0656,
"step": 2500
},
{
"epoch": 0.4247398257043743,
"grad_norm": 0.049512227276864045,
"learning_rate": 3.315786721390845e-05,
"loss": 0.0661,
"step": 2510
},
{
"epoch": 0.4264320162450292,
"grad_norm": 0.04345046465851396,
"learning_rate": 3.302549820442818e-05,
"loss": 0.0652,
"step": 2520
},
{
"epoch": 0.4281242067856841,
"grad_norm": 0.047954387938694894,
"learning_rate": 3.2892877875673396e-05,
"loss": 0.0643,
"step": 2530
},
{
"epoch": 0.429816397326339,
"grad_norm": 0.04230879386994909,
"learning_rate": 3.276001038066287e-05,
"loss": 0.0663,
"step": 2540
},
{
"epoch": 0.4315085878669938,
"grad_norm": 0.04680145886971475,
"learning_rate": 3.262689988015544e-05,
"loss": 0.0644,
"step": 2550
},
{
"epoch": 0.4332007784076487,
"grad_norm": 0.06333383895835933,
"learning_rate": 3.2493550542519656e-05,
"loss": 0.0637,
"step": 2560
},
{
"epoch": 0.4348929689483036,
"grad_norm": 0.042297227144947644,
"learning_rate": 3.23599665436033e-05,
"loss": 0.0662,
"step": 2570
},
{
"epoch": 0.43658515948895843,
"grad_norm": 0.045119411089813447,
"learning_rate": 3.222615206660258e-05,
"loss": 0.0653,
"step": 2580
},
{
"epoch": 0.4382773500296133,
"grad_norm": 0.045597809939061776,
"learning_rate": 3.2092111301931156e-05,
"loss": 0.0659,
"step": 2590
},
{
"epoch": 0.4399695405702682,
"grad_norm": 0.053728017312180965,
"learning_rate": 3.19578484470889e-05,
"loss": 0.0654,
"step": 2600
},
{
"epoch": 0.4416617311109231,
"grad_norm": 0.05589207401233193,
"learning_rate": 3.182336770653047e-05,
"loss": 0.0648,
"step": 2610
},
{
"epoch": 0.44335392165157794,
"grad_norm": 0.04316275995071727,
"learning_rate": 3.1688673291533636e-05,
"loss": 0.0639,
"step": 2620
},
{
"epoch": 0.44504611219223283,
"grad_norm": 0.04828683592102072,
"learning_rate": 3.155376942006739e-05,
"loss": 0.0673,
"step": 2630
},
{
"epoch": 0.4467383027328877,
"grad_norm": 0.049621002112458006,
"learning_rate": 3.1418660316659895e-05,
"loss": 0.0642,
"step": 2640
},
{
"epoch": 0.4484304932735426,
"grad_norm": 0.043864672184891956,
"learning_rate": 3.128335021226616e-05,
"loss": 0.0658,
"step": 2650
},
{
"epoch": 0.45012268381419746,
"grad_norm": 0.12079837341691556,
"learning_rate": 3.114784334413556e-05,
"loss": 0.0659,
"step": 2660
},
{
"epoch": 0.45181487435485235,
"grad_norm": 0.04893223382916863,
"learning_rate": 3.101214395567915e-05,
"loss": 0.0663,
"step": 2670
},
{
"epoch": 0.45350706489550724,
"grad_norm": 0.05406447618020935,
"learning_rate": 3.0876256296336775e-05,
"loss": 0.0667,
"step": 2680
},
{
"epoch": 0.45519925543616213,
"grad_norm": 0.06126331158408281,
"learning_rate": 3.074018462144401e-05,
"loss": 0.0648,
"step": 2690
},
{
"epoch": 0.45689144597681697,
"grad_norm": 0.05326753653993471,
"learning_rate": 3.060393319209888e-05,
"loss": 0.0644,
"step": 2700
},
{
"epoch": 0.45858363651747186,
"grad_norm": 0.041891934750803474,
"learning_rate": 3.0467506275028453e-05,
"loss": 0.0652,
"step": 2710
},
{
"epoch": 0.46027582705812675,
"grad_norm": 0.041668200417168566,
"learning_rate": 3.0330908142455207e-05,
"loss": 0.0646,
"step": 2720
},
{
"epoch": 0.46196801759878164,
"grad_norm": 0.062080962661839156,
"learning_rate": 3.019414307196324e-05,
"loss": 0.067,
"step": 2730
},
{
"epoch": 0.4636602081394365,
"grad_norm": 0.0479026207674969,
"learning_rate": 3.0057215346364354e-05,
"loss": 0.0646,
"step": 2740
},
{
"epoch": 0.4653523986800914,
"grad_norm": 0.03993096878242372,
"learning_rate": 2.9920129253563878e-05,
"loss": 0.0639,
"step": 2750
},
{
"epoch": 0.46704458922074626,
"grad_norm": 0.04415715300148828,
"learning_rate": 2.9782889086426453e-05,
"loss": 0.0637,
"step": 2760
},
{
"epoch": 0.46873677976140116,
"grad_norm": 0.056148067231621204,
"learning_rate": 2.9645499142641565e-05,
"loss": 0.0652,
"step": 2770
},
{
"epoch": 0.470428970302056,
"grad_norm": 0.0446658338752171,
"learning_rate": 2.950796372458896e-05,
"loss": 0.0638,
"step": 2780
},
{
"epoch": 0.4721211608427109,
"grad_norm": 0.04196989056777316,
"learning_rate": 2.9370287139203956e-05,
"loss": 0.063,
"step": 2790
},
{
"epoch": 0.4738133513833658,
"grad_norm": 0.04918235227295245,
"learning_rate": 2.9232473697842488e-05,
"loss": 0.0647,
"step": 2800
},
{
"epoch": 0.47550554192402067,
"grad_norm": 0.043127336350349786,
"learning_rate": 2.90945277161462e-05,
"loss": 0.0659,
"step": 2810
},
{
"epoch": 0.4771977324646755,
"grad_norm": 0.0495117471794782,
"learning_rate": 2.8956453513907223e-05,
"loss": 0.0663,
"step": 2820
},
{
"epoch": 0.4788899230053304,
"grad_norm": 0.057171959042341354,
"learning_rate": 2.8818255414932928e-05,
"loss": 0.067,
"step": 2830
},
{
"epoch": 0.4805821135459853,
"grad_norm": 0.04980676972740161,
"learning_rate": 2.867993774691054e-05,
"loss": 0.0648,
"step": 2840
},
{
"epoch": 0.4822743040866402,
"grad_norm": 0.04249633703186063,
"learning_rate": 2.8541504841271587e-05,
"loss": 0.0637,
"step": 2850
},
{
"epoch": 0.483966494627295,
"grad_norm": 0.05665632593137514,
"learning_rate": 2.840296103305628e-05,
"loss": 0.0668,
"step": 2860
},
{
"epoch": 0.4856586851679499,
"grad_norm": 0.05266985454136702,
"learning_rate": 2.8264310660777748e-05,
"loss": 0.0667,
"step": 2870
},
{
"epoch": 0.4873508757086048,
"grad_norm": 0.0511324941441528,
"learning_rate": 2.8125558066286195e-05,
"loss": 0.0647,
"step": 2880
},
{
"epoch": 0.48904306624925964,
"grad_norm": 0.05312502776248933,
"learning_rate": 2.798670759463292e-05,
"loss": 0.064,
"step": 2890
},
{
"epoch": 0.49073525678991453,
"grad_norm": 0.04391494206313917,
"learning_rate": 2.784776359393424e-05,
"loss": 0.0651,
"step": 2900
},
{
"epoch": 0.4924274473305694,
"grad_norm": 0.06734068102837011,
"learning_rate": 2.7708730415235375e-05,
"loss": 0.0648,
"step": 2910
},
{
"epoch": 0.4941196378712243,
"grad_norm": 0.06131607163909934,
"learning_rate": 2.756961241237413e-05,
"loss": 0.0634,
"step": 2920
},
{
"epoch": 0.49581182841187915,
"grad_norm": 0.04484727409853693,
"learning_rate": 2.7430413941844614e-05,
"loss": 0.0634,
"step": 2930
},
{
"epoch": 0.49750401895253404,
"grad_norm": 0.04388586118373513,
"learning_rate": 2.729113936266077e-05,
"loss": 0.0646,
"step": 2940
},
{
"epoch": 0.49919620949318894,
"grad_norm": 0.04779079797700071,
"learning_rate": 2.7151793036219892e-05,
"loss": 0.0636,
"step": 2950
},
{
"epoch": 0.5008884000338438,
"grad_norm": 0.042132793383735184,
"learning_rate": 2.701237932616607e-05,
"loss": 0.0634,
"step": 2960
},
{
"epoch": 0.5025805905744987,
"grad_norm": 0.04333795678793624,
"learning_rate": 2.6872902598253476e-05,
"loss": 0.0638,
"step": 2970
},
{
"epoch": 0.5042727811151536,
"grad_norm": 0.04114374823242134,
"learning_rate": 2.6733367220209742e-05,
"loss": 0.0653,
"step": 2980
},
{
"epoch": 0.5059649716558084,
"grad_norm": 0.05153114257848755,
"learning_rate": 2.6593777561599097e-05,
"loss": 0.0644,
"step": 2990
},
{
"epoch": 0.5076571621964633,
"grad_norm": 0.04892744580099628,
"learning_rate": 2.6454137993685585e-05,
"loss": 0.0637,
"step": 3000
},
{
"epoch": 0.5076571621964633,
"eval_loss": 0.06484896689653397,
"eval_runtime": 510.0728,
"eval_samples_per_second": 11.442,
"eval_steps_per_second": 0.716,
"step": 3000
},
{
"epoch": 0.5093493527371182,
"grad_norm": 0.03827028525994092,
"learning_rate": 2.6314452889296177e-05,
"loss": 0.0619,
"step": 3010
},
{
"epoch": 0.5110415432777731,
"grad_norm": 0.0447714654314853,
"learning_rate": 2.6174726622683786e-05,
"loss": 0.0645,
"step": 3020
},
{
"epoch": 0.5127337338184279,
"grad_norm": 0.04776881468049298,
"learning_rate": 2.6034963569390376e-05,
"loss": 0.0648,
"step": 3030
},
{
"epoch": 0.5144259243590829,
"grad_norm": 0.055302925535470435,
"learning_rate": 2.589516810610984e-05,
"loss": 0.0639,
"step": 3040
},
{
"epoch": 0.5161181148997377,
"grad_norm": 0.054676744990593905,
"learning_rate": 2.5755344610551026e-05,
"loss": 0.0626,
"step": 3050
},
{
"epoch": 0.5178103054403926,
"grad_norm": 0.04340851199789463,
"learning_rate": 2.56154974613006e-05,
"loss": 0.0647,
"step": 3060
},
{
"epoch": 0.5195024959810475,
"grad_norm": 0.041530575120781094,
"learning_rate": 2.5475631037685958e-05,
"loss": 0.0657,
"step": 3070
},
{
"epoch": 0.5211946865217023,
"grad_norm": 0.049711901848570585,
"learning_rate": 2.533574971963807e-05,
"loss": 0.066,
"step": 3080
},
{
"epoch": 0.5228868770623573,
"grad_norm": 0.10323962667241846,
"learning_rate": 2.51958578875543e-05,
"loss": 0.064,
"step": 3090
},
{
"epoch": 0.5245790676030121,
"grad_norm": 0.04024374620040581,
"learning_rate": 2.5055959922161314e-05,
"loss": 0.0628,
"step": 3100
},
{
"epoch": 0.5262712581436669,
"grad_norm": 0.08104874965147474,
"learning_rate": 2.4916060204377792e-05,
"loss": 0.0642,
"step": 3110
},
{
"epoch": 0.5279634486843219,
"grad_norm": 0.060782927356135996,
"learning_rate": 2.4776163115177325e-05,
"loss": 0.064,
"step": 3120
},
{
"epoch": 0.5296556392249767,
"grad_norm": 0.0407557653674574,
"learning_rate": 2.4636273035451164e-05,
"loss": 0.0634,
"step": 3130
},
{
"epoch": 0.5313478297656317,
"grad_norm": 0.041925585349924056,
"learning_rate": 2.4496394345871083e-05,
"loss": 0.0647,
"step": 3140
},
{
"epoch": 0.5330400203062865,
"grad_norm": 0.05522138688168146,
"learning_rate": 2.4356531426752144e-05,
"loss": 0.0627,
"step": 3150
},
{
"epoch": 0.5347322108469413,
"grad_norm": 0.05790854041482495,
"learning_rate": 2.4216688657915574e-05,
"loss": 0.0637,
"step": 3160
},
{
"epoch": 0.5364244013875963,
"grad_norm": 0.11506562373097944,
"learning_rate": 2.4076870418551605e-05,
"loss": 0.0644,
"step": 3170
},
{
"epoch": 0.5381165919282511,
"grad_norm": 0.0391327164925947,
"learning_rate": 2.3937081087082288e-05,
"loss": 0.0633,
"step": 3180
},
{
"epoch": 0.539808782468906,
"grad_norm": 0.04263224399320138,
"learning_rate": 2.379732504102446e-05,
"loss": 0.0637,
"step": 3190
},
{
"epoch": 0.5415009730095609,
"grad_norm": 0.04357043184545777,
"learning_rate": 2.365760665685259e-05,
"loss": 0.0632,
"step": 3200
},
{
"epoch": 0.5431931635502157,
"grad_norm": 0.05572076585199387,
"learning_rate": 2.3517930309861782e-05,
"loss": 0.0635,
"step": 3210
},
{
"epoch": 0.5448853540908706,
"grad_norm": 0.04103324104627984,
"learning_rate": 2.3378300374030728e-05,
"loss": 0.064,
"step": 3220
},
{
"epoch": 0.5465775446315255,
"grad_norm": 0.04659985964738097,
"learning_rate": 2.323872122188476e-05,
"loss": 0.0648,
"step": 3230
},
{
"epoch": 0.5482697351721804,
"grad_norm": 0.05481556818403658,
"learning_rate": 2.3099197224358918e-05,
"loss": 0.0649,
"step": 3240
},
{
"epoch": 0.5499619257128353,
"grad_norm": 0.039166322811640106,
"learning_rate": 2.2959732750661053e-05,
"loss": 0.0618,
"step": 3250
},
{
"epoch": 0.5516541162534901,
"grad_norm": 0.037501216437416725,
"learning_rate": 2.282033216813504e-05,
"loss": 0.0628,
"step": 3260
},
{
"epoch": 0.553346306794145,
"grad_norm": 0.04650705196029865,
"learning_rate": 2.2680999842123977e-05,
"loss": 0.0644,
"step": 3270
},
{
"epoch": 0.5550384973347999,
"grad_norm": 0.07242939350286805,
"learning_rate": 2.2541740135833518e-05,
"loss": 0.0621,
"step": 3280
},
{
"epoch": 0.5567306878754548,
"grad_norm": 0.04184444624793964,
"learning_rate": 2.2402557410195204e-05,
"loss": 0.0641,
"step": 3290
},
{
"epoch": 0.5584228784161096,
"grad_norm": 0.05656291302870268,
"learning_rate": 2.226345602372994e-05,
"loss": 0.0631,
"step": 3300
},
{
"epoch": 0.5601150689567646,
"grad_norm": 0.04667395228952371,
"learning_rate": 2.2124440332411487e-05,
"loss": 0.0649,
"step": 3310
},
{
"epoch": 0.5618072594974194,
"grad_norm": 0.04468882476329186,
"learning_rate": 2.1985514689530037e-05,
"loss": 0.0621,
"step": 3320
},
{
"epoch": 0.5634994500380743,
"grad_norm": 0.06117915651429709,
"learning_rate": 2.1846683445555924e-05,
"loss": 0.0638,
"step": 3330
},
{
"epoch": 0.5651916405787292,
"grad_norm": 0.050888050485787734,
"learning_rate": 2.1707950948003343e-05,
"loss": 0.0641,
"step": 3340
},
{
"epoch": 0.566883831119384,
"grad_norm": 0.04656120131241164,
"learning_rate": 2.1569321541294278e-05,
"loss": 0.0625,
"step": 3350
},
{
"epoch": 0.568576021660039,
"grad_norm": 0.044858092163292354,
"learning_rate": 2.1430799566622364e-05,
"loss": 0.0634,
"step": 3360
},
{
"epoch": 0.5702682122006938,
"grad_norm": 0.040507953187460526,
"learning_rate": 2.1292389361817025e-05,
"loss": 0.0626,
"step": 3370
},
{
"epoch": 0.5719604027413486,
"grad_norm": 0.04070591562339343,
"learning_rate": 2.11540952612076e-05,
"loss": 0.064,
"step": 3380
},
{
"epoch": 0.5736525932820036,
"grad_norm": 0.037899404381374464,
"learning_rate": 2.10159215954876e-05,
"loss": 0.0659,
"step": 3390
},
{
"epoch": 0.5753447838226584,
"grad_norm": 0.057542354523080985,
"learning_rate": 2.087787269157911e-05,
"loss": 0.0611,
"step": 3400
},
{
"epoch": 0.5770369743633134,
"grad_norm": 0.05951860446928795,
"learning_rate": 2.0739952872497285e-05,
"loss": 0.0667,
"step": 3410
},
{
"epoch": 0.5787291649039682,
"grad_norm": 0.05630511962528299,
"learning_rate": 2.060216645721498e-05,
"loss": 0.0629,
"step": 3420
},
{
"epoch": 0.580421355444623,
"grad_norm": 0.04316226696055459,
"learning_rate": 2.046451776052748e-05,
"loss": 0.0634,
"step": 3430
},
{
"epoch": 0.582113545985278,
"grad_norm": 0.04002750092220964,
"learning_rate": 2.0327011092917438e-05,
"loss": 0.0625,
"step": 3440
},
{
"epoch": 0.5838057365259328,
"grad_norm": 0.04170769790210492,
"learning_rate": 2.0189650760419814e-05,
"loss": 0.0632,
"step": 3450
},
{
"epoch": 0.5854979270665877,
"grad_norm": 0.047539094930624526,
"learning_rate": 2.0052441064487078e-05,
"loss": 0.0626,
"step": 3460
},
{
"epoch": 0.5871901176072426,
"grad_norm": 0.03729179829269829,
"learning_rate": 1.991538630185452e-05,
"loss": 0.0637,
"step": 3470
},
{
"epoch": 0.5888823081478974,
"grad_norm": 0.03842077360884552,
"learning_rate": 1.9778490764405645e-05,
"loss": 0.0636,
"step": 3480
},
{
"epoch": 0.5905744986885524,
"grad_norm": 0.0540382808860073,
"learning_rate": 1.964175873903784e-05,
"loss": 0.0604,
"step": 3490
},
{
"epoch": 0.5922666892292072,
"grad_norm": 0.045484618110890426,
"learning_rate": 1.9505194507528065e-05,
"loss": 0.0641,
"step": 3500
},
{
"epoch": 0.5939588797698621,
"grad_norm": 0.04209112283118814,
"learning_rate": 1.9368802346398825e-05,
"loss": 0.0617,
"step": 3510
},
{
"epoch": 0.595651070310517,
"grad_norm": 0.044140845525802254,
"learning_rate": 1.9232586526784214e-05,
"loss": 0.0617,
"step": 3520
},
{
"epoch": 0.5973432608511718,
"grad_norm": 0.04022093386340225,
"learning_rate": 1.909655131429616e-05,
"loss": 0.0628,
"step": 3530
},
{
"epoch": 0.5990354513918267,
"grad_norm": 0.039204676979806695,
"learning_rate": 1.8960700968890884e-05,
"loss": 0.0626,
"step": 3540
},
{
"epoch": 0.6007276419324816,
"grad_norm": 0.04032097676990017,
"learning_rate": 1.8825039744735454e-05,
"loss": 0.0604,
"step": 3550
},
{
"epoch": 0.6024198324731365,
"grad_norm": 0.037031648022000194,
"learning_rate": 1.868957189007461e-05,
"loss": 0.0648,
"step": 3560
},
{
"epoch": 0.6041120230137913,
"grad_norm": 0.06651838571942774,
"learning_rate": 1.8554301647097673e-05,
"loss": 0.0642,
"step": 3570
},
{
"epoch": 0.6058042135544462,
"grad_norm": 0.060153242610817975,
"learning_rate": 1.841923325180578e-05,
"loss": 0.0624,
"step": 3580
},
{
"epoch": 0.6074964040951011,
"grad_norm": 0.040053202164083125,
"learning_rate": 1.8284370933879156e-05,
"loss": 0.0621,
"step": 3590
},
{
"epoch": 0.609188594635756,
"grad_norm": 0.07249923202850005,
"learning_rate": 1.8149718916544695e-05,
"loss": 0.0627,
"step": 3600
},
{
"epoch": 0.6108807851764109,
"grad_norm": 0.046299394065382554,
"learning_rate": 1.8015281416443723e-05,
"loss": 0.0622,
"step": 3610
},
{
"epoch": 0.6125729757170657,
"grad_norm": 0.038318676884301246,
"learning_rate": 1.7881062643499915e-05,
"loss": 0.0623,
"step": 3620
},
{
"epoch": 0.6142651662577207,
"grad_norm": 0.04454541005029023,
"learning_rate": 1.774706680078752e-05,
"loss": 0.0618,
"step": 3630
},
{
"epoch": 0.6159573567983755,
"grad_norm": 0.06330907469180895,
"learning_rate": 1.7613298084399644e-05,
"loss": 0.0637,
"step": 3640
},
{
"epoch": 0.6176495473390303,
"grad_norm": 0.03846146553504766,
"learning_rate": 1.7479760683316968e-05,
"loss": 0.0624,
"step": 3650
},
{
"epoch": 0.6193417378796853,
"grad_norm": 0.04492754131015157,
"learning_rate": 1.734645877927649e-05,
"loss": 0.0614,
"step": 3660
},
{
"epoch": 0.6210339284203401,
"grad_norm": 0.04137185456903909,
"learning_rate": 1.7213396546640586e-05,
"loss": 0.0613,
"step": 3670
},
{
"epoch": 0.6227261189609951,
"grad_norm": 0.045821100046964706,
"learning_rate": 1.708057815226632e-05,
"loss": 0.0631,
"step": 3680
},
{
"epoch": 0.6244183095016499,
"grad_norm": 0.05220608786215871,
"learning_rate": 1.6948007755374896e-05,
"loss": 0.0621,
"step": 3690
},
{
"epoch": 0.6261105000423047,
"grad_norm": 0.040150100263605354,
"learning_rate": 1.681568950742151e-05,
"loss": 0.0597,
"step": 3700
},
{
"epoch": 0.6278026905829597,
"grad_norm": 0.050096371298430406,
"learning_rate": 1.6683627551965227e-05,
"loss": 0.0631,
"step": 3710
},
{
"epoch": 0.6294948811236145,
"grad_norm": 0.04708310547831978,
"learning_rate": 1.6551826024539338e-05,
"loss": 0.0612,
"step": 3720
},
{
"epoch": 0.6311870716642694,
"grad_norm": 0.05819848922580848,
"learning_rate": 1.642028905252178e-05,
"loss": 0.0622,
"step": 3730
},
{
"epoch": 0.6328792622049243,
"grad_norm": 0.044477518796261835,
"learning_rate": 1.628902075500591e-05,
"loss": 0.0616,
"step": 3740
},
{
"epoch": 0.6345714527455791,
"grad_norm": 0.04163371007970935,
"learning_rate": 1.6158025242671536e-05,
"loss": 0.0635,
"step": 3750
},
{
"epoch": 0.6362636432862341,
"grad_norm": 0.04785576633752719,
"learning_rate": 1.6027306617656128e-05,
"loss": 0.0612,
"step": 3760
},
{
"epoch": 0.6379558338268889,
"grad_norm": 0.040661631937929356,
"learning_rate": 1.589686897342646e-05,
"loss": 0.0625,
"step": 3770
},
{
"epoch": 0.6396480243675438,
"grad_norm": 0.03663280847925467,
"learning_rate": 1.576671639465031e-05,
"loss": 0.0601,
"step": 3780
},
{
"epoch": 0.6413402149081987,
"grad_norm": 0.05278926880359737,
"learning_rate": 1.5636852957068644e-05,
"loss": 0.0628,
"step": 3790
},
{
"epoch": 0.6430324054488535,
"grad_norm": 0.0626218729547267,
"learning_rate": 1.5507282727367928e-05,
"loss": 0.0614,
"step": 3800
},
{
"epoch": 0.6447245959895084,
"grad_norm": 0.04609502805038381,
"learning_rate": 1.5378009763052777e-05,
"loss": 0.0624,
"step": 3810
},
{
"epoch": 0.6464167865301633,
"grad_norm": 0.054090280773029784,
"learning_rate": 1.5249038112318947e-05,
"loss": 0.0612,
"step": 3820
},
{
"epoch": 0.6481089770708182,
"grad_norm": 0.04427290412563348,
"learning_rate": 1.5120371813926487e-05,
"loss": 0.065,
"step": 3830
},
{
"epoch": 0.6498011676114731,
"grad_norm": 0.04264932370581012,
"learning_rate": 1.4992014897073348e-05,
"loss": 0.0606,
"step": 3840
},
{
"epoch": 0.6514933581521279,
"grad_norm": 0.041930232477467966,
"learning_rate": 1.4863971381269131e-05,
"loss": 0.059,
"step": 3850
},
{
"epoch": 0.6531855486927828,
"grad_norm": 0.09838372088966947,
"learning_rate": 1.4736245276209281e-05,
"loss": 0.0628,
"step": 3860
},
{
"epoch": 0.6548777392334377,
"grad_norm": 0.04380330148367012,
"learning_rate": 1.4608840581649486e-05,
"loss": 0.0606,
"step": 3870
},
{
"epoch": 0.6565699297740926,
"grad_norm": 0.03910023814128401,
"learning_rate": 1.4481761287280412e-05,
"loss": 0.0601,
"step": 3880
},
{
"epoch": 0.6582621203147474,
"grad_norm": 0.04280843336379621,
"learning_rate": 1.4355011372602816e-05,
"loss": 0.0619,
"step": 3890
},
{
"epoch": 0.6599543108554023,
"grad_norm": 0.03759092479271448,
"learning_rate": 1.4228594806802872e-05,
"loss": 0.0602,
"step": 3900
},
{
"epoch": 0.6616465013960572,
"grad_norm": 0.038239787295855615,
"learning_rate": 1.4102515548627926e-05,
"loss": 0.062,
"step": 3910
},
{
"epoch": 0.663338691936712,
"grad_norm": 0.03720954492963802,
"learning_rate": 1.3976777546262471e-05,
"loss": 0.0606,
"step": 3920
},
{
"epoch": 0.665030882477367,
"grad_norm": 0.05379213882933423,
"learning_rate": 1.3851384737204579e-05,
"loss": 0.0623,
"step": 3930
},
{
"epoch": 0.6667230730180218,
"grad_norm": 0.036898058920349905,
"learning_rate": 1.3726341048142522e-05,
"loss": 0.0602,
"step": 3940
},
{
"epoch": 0.6684152635586768,
"grad_norm": 0.061650567852263326,
"learning_rate": 1.3601650394831856e-05,
"loss": 0.0618,
"step": 3950
},
{
"epoch": 0.6701074540993316,
"grad_norm": 0.03847325655911707,
"learning_rate": 1.3477316681972807e-05,
"loss": 0.0609,
"step": 3960
},
{
"epoch": 0.6717996446399864,
"grad_norm": 0.03919965173077874,
"learning_rate": 1.3353343803087954e-05,
"loss": 0.0605,
"step": 3970
},
{
"epoch": 0.6734918351806414,
"grad_norm": 0.03591517879554738,
"learning_rate": 1.3229735640400326e-05,
"loss": 0.0602,
"step": 3980
},
{
"epoch": 0.6751840257212962,
"grad_norm": 0.04923138061071831,
"learning_rate": 1.3106496064711826e-05,
"loss": 0.0619,
"step": 3990
},
{
"epoch": 0.676876216261951,
"grad_norm": 0.04108664171417808,
"learning_rate": 1.2983628935282036e-05,
"loss": 0.0601,
"step": 4000
},
{
"epoch": 0.676876216261951,
"eval_loss": 0.06200229004025459,
"eval_runtime": 510.0751,
"eval_samples_per_second": 11.441,
"eval_steps_per_second": 0.716,
"step": 4000
},
{
"epoch": 0.678568406802606,
"grad_norm": 0.11484641754138697,
"learning_rate": 1.2861138099707337e-05,
"loss": 0.0625,
"step": 4010
},
{
"epoch": 0.6802605973432608,
"grad_norm": 0.044560475982201864,
"learning_rate": 1.2739027393800428e-05,
"loss": 0.0611,
"step": 4020
},
{
"epoch": 0.6819527878839158,
"grad_norm": 0.04406556361797706,
"learning_rate": 1.2617300641470203e-05,
"loss": 0.062,
"step": 4030
},
{
"epoch": 0.6836449784245706,
"grad_norm": 0.04408584577401301,
"learning_rate": 1.2495961654602009e-05,
"loss": 0.0627,
"step": 4040
},
{
"epoch": 0.6853371689652255,
"grad_norm": 0.03958417410687766,
"learning_rate": 1.2375014232938315e-05,
"loss": 0.0616,
"step": 4050
},
{
"epoch": 0.6870293595058804,
"grad_norm": 0.035822674640071474,
"learning_rate": 1.2254462163959634e-05,
"loss": 0.0604,
"step": 4060
},
{
"epoch": 0.6887215500465352,
"grad_norm": 0.06018269251041653,
"learning_rate": 1.2134309222766019e-05,
"loss": 0.0614,
"step": 4070
},
{
"epoch": 0.6904137405871901,
"grad_norm": 0.04751555384103322,
"learning_rate": 1.2014559171958762e-05,
"loss": 0.0631,
"step": 4080
},
{
"epoch": 0.692105931127845,
"grad_norm": 0.051412001695713284,
"learning_rate": 1.189521576152261e-05,
"loss": 0.0606,
"step": 4090
},
{
"epoch": 0.6937981216684999,
"grad_norm": 0.04464799982613195,
"learning_rate": 1.1776282728708347e-05,
"loss": 0.0629,
"step": 4100
},
{
"epoch": 0.6954903122091548,
"grad_norm": 0.034676586872487244,
"learning_rate": 1.1657763797915705e-05,
"loss": 0.062,
"step": 4110
},
{
"epoch": 0.6971825027498096,
"grad_norm": 0.03676686505714596,
"learning_rate": 1.1539662680576821e-05,
"loss": 0.0597,
"step": 4120
},
{
"epoch": 0.6988746932904645,
"grad_norm": 0.03935775238828102,
"learning_rate": 1.142198307503989e-05,
"loss": 0.0601,
"step": 4130
},
{
"epoch": 0.7005668838311194,
"grad_norm": 0.048748585113061856,
"learning_rate": 1.130472866645349e-05,
"loss": 0.0602,
"step": 4140
},
{
"epoch": 0.7022590743717743,
"grad_norm": 0.04352616391957919,
"learning_rate": 1.1187903126651092e-05,
"loss": 0.0617,
"step": 4150
},
{
"epoch": 0.7039512649124291,
"grad_norm": 0.05586980209838057,
"learning_rate": 1.1071510114036088e-05,
"loss": 0.06,
"step": 4160
},
{
"epoch": 0.705643455453084,
"grad_norm": 0.06321872531261899,
"learning_rate": 1.0955553273467242e-05,
"loss": 0.0628,
"step": 4170
},
{
"epoch": 0.7073356459937389,
"grad_norm": 0.04824662715626538,
"learning_rate": 1.0840036236144544e-05,
"loss": 0.0614,
"step": 4180
},
{
"epoch": 0.7090278365343937,
"grad_norm": 0.03881831680715111,
"learning_rate": 1.0724962619495524e-05,
"loss": 0.061,
"step": 4190
},
{
"epoch": 0.7107200270750487,
"grad_norm": 0.04860518239162666,
"learning_rate": 1.0610336027061916e-05,
"loss": 0.063,
"step": 4200
},
{
"epoch": 0.7124122176157035,
"grad_norm": 0.04172676301238284,
"learning_rate": 1.0496160048386878e-05,
"loss": 0.0616,
"step": 4210
},
{
"epoch": 0.7141044081563585,
"grad_norm": 0.04891476379742556,
"learning_rate": 1.0382438258902535e-05,
"loss": 0.0597,
"step": 4220
},
{
"epoch": 0.7157965986970133,
"grad_norm": 0.03754595452221889,
"learning_rate": 1.0269174219818023e-05,
"loss": 0.0616,
"step": 4230
},
{
"epoch": 0.7174887892376681,
"grad_norm": 0.04429893729588766,
"learning_rate": 1.0156371478008011e-05,
"loss": 0.0616,
"step": 4240
},
{
"epoch": 0.7191809797783231,
"grad_norm": 0.04241523368377547,
"learning_rate": 1.0044033565901561e-05,
"loss": 0.0622,
"step": 4250
},
{
"epoch": 0.7208731703189779,
"grad_norm": 0.04291686361048184,
"learning_rate": 9.93216400137158e-06,
"loss": 0.0598,
"step": 4260
},
{
"epoch": 0.7225653608596327,
"grad_norm": 0.06478614105438416,
"learning_rate": 9.820766287624602e-06,
"loss": 0.062,
"step": 4270
},
{
"epoch": 0.7242575514002877,
"grad_norm": 0.04342650645863712,
"learning_rate": 9.709843913091104e-06,
"loss": 0.062,
"step": 4280
},
{
"epoch": 0.7259497419409425,
"grad_norm": 0.03668999755359844,
"learning_rate": 9.599400351316293e-06,
"loss": 0.0594,
"step": 4290
},
{
"epoch": 0.7276419324815975,
"grad_norm": 0.03642295331884263,
"learning_rate": 9.489439060851273e-06,
"loss": 0.0597,
"step": 4300
},
{
"epoch": 0.7293341230222523,
"grad_norm": 0.04341851185033999,
"learning_rate": 9.379963485144821e-06,
"loss": 0.0616,
"step": 4310
},
{
"epoch": 0.7310263135629071,
"grad_norm": 0.04686512023251531,
"learning_rate": 9.270977052435443e-06,
"loss": 0.061,
"step": 4320
},
{
"epoch": 0.7327185041035621,
"grad_norm": 0.03995328123375408,
"learning_rate": 9.162483175644146e-06,
"loss": 0.0608,
"step": 4330
},
{
"epoch": 0.7344106946442169,
"grad_norm": 0.04294963733096995,
"learning_rate": 9.054485252267452e-06,
"loss": 0.0606,
"step": 4340
},
{
"epoch": 0.7361028851848718,
"grad_norm": 0.0437589139668772,
"learning_rate": 8.946986664271099e-06,
"loss": 0.0642,
"step": 4350
},
{
"epoch": 0.7377950757255267,
"grad_norm": 0.03905118332207717,
"learning_rate": 8.839990777984044e-06,
"loss": 0.0596,
"step": 4360
},
{
"epoch": 0.7394872662661816,
"grad_norm": 0.03704612749001196,
"learning_rate": 8.733500943993109e-06,
"loss": 0.061,
"step": 4370
},
{
"epoch": 0.7411794568068365,
"grad_norm": 0.035749262648839396,
"learning_rate": 8.627520497038047e-06,
"loss": 0.0608,
"step": 4380
},
{
"epoch": 0.7428716473474913,
"grad_norm": 0.059541120977181304,
"learning_rate": 8.52205275590708e-06,
"loss": 0.0582,
"step": 4390
},
{
"epoch": 0.7445638378881462,
"grad_norm": 0.03703828371062853,
"learning_rate": 8.417101023333032e-06,
"loss": 0.0577,
"step": 4400
},
{
"epoch": 0.7462560284288011,
"grad_norm": 0.037618856979218726,
"learning_rate": 8.312668585889835e-06,
"loss": 0.0592,
"step": 4410
},
{
"epoch": 0.747948218969456,
"grad_norm": 0.04065683081081544,
"learning_rate": 8.20875871388965e-06,
"loss": 0.0607,
"step": 4420
},
{
"epoch": 0.7496404095101108,
"grad_norm": 0.036781880174871856,
"learning_rate": 8.105374661280471e-06,
"loss": 0.0586,
"step": 4430
},
{
"epoch": 0.7513326000507657,
"grad_norm": 0.03820920245804256,
"learning_rate": 8.002519665544167e-06,
"loss": 0.0592,
"step": 4440
},
{
"epoch": 0.7530247905914206,
"grad_norm": 0.043018760405495626,
"learning_rate": 7.900196947595187e-06,
"loss": 0.0609,
"step": 4450
},
{
"epoch": 0.7547169811320755,
"grad_norm": 0.044726424078343936,
"learning_rate": 7.79840971167959e-06,
"loss": 0.0585,
"step": 4460
},
{
"epoch": 0.7564091716727304,
"grad_norm": 0.04875439653707478,
"learning_rate": 7.697161145274817e-06,
"loss": 0.0603,
"step": 4470
},
{
"epoch": 0.7581013622133852,
"grad_norm": 0.037114350479566995,
"learning_rate": 7.596454418989776e-06,
"loss": 0.062,
"step": 4480
},
{
"epoch": 0.7597935527540401,
"grad_norm": 0.0459543150162269,
"learning_rate": 7.496292686465631e-06,
"loss": 0.0579,
"step": 4490
},
{
"epoch": 0.761485743294695,
"grad_norm": 0.036596180011920276,
"learning_rate": 7.396679084277008e-06,
"loss": 0.0594,
"step": 4500
},
{
"epoch": 0.7631779338353498,
"grad_norm": 0.04781504560895781,
"learning_rate": 7.29761673183374e-06,
"loss": 0.0604,
"step": 4510
},
{
"epoch": 0.7648701243760048,
"grad_norm": 0.05432780898062741,
"learning_rate": 7.199108731283269e-06,
"loss": 0.0587,
"step": 4520
},
{
"epoch": 0.7665623149166596,
"grad_norm": 0.04252217624080214,
"learning_rate": 7.101158167413416e-06,
"loss": 0.0595,
"step": 4530
},
{
"epoch": 0.7682545054573144,
"grad_norm": 0.05277459414378154,
"learning_rate": 7.003768107555844e-06,
"loss": 0.0615,
"step": 4540
},
{
"epoch": 0.7699466959979694,
"grad_norm": 0.06241975823896846,
"learning_rate": 6.906941601489958e-06,
"loss": 0.0595,
"step": 4550
},
{
"epoch": 0.7716388865386242,
"grad_norm": 0.03703441794263418,
"learning_rate": 6.810681681347428e-06,
"loss": 0.061,
"step": 4560
},
{
"epoch": 0.7733310770792792,
"grad_norm": 0.059755176618706676,
"learning_rate": 6.714991361517217e-06,
"loss": 0.0616,
"step": 4570
},
{
"epoch": 0.775023267619934,
"grad_norm": 0.04901953037595079,
"learning_rate": 6.619873638551216e-06,
"loss": 0.0618,
"step": 4580
},
{
"epoch": 0.7767154581605888,
"grad_norm": 0.05512517804134823,
"learning_rate": 6.525331491070391e-06,
"loss": 0.0626,
"step": 4590
},
{
"epoch": 0.7784076487012438,
"grad_norm": 0.037390553382658766,
"learning_rate": 6.4313678796714845e-06,
"loss": 0.0589,
"step": 4600
},
{
"epoch": 0.7800998392418986,
"grad_norm": 0.044235384961872896,
"learning_rate": 6.337985746834332e-06,
"loss": 0.0611,
"step": 4610
},
{
"epoch": 0.7817920297825535,
"grad_norm": 0.05144950655823148,
"learning_rate": 6.245188016829706e-06,
"loss": 0.0635,
"step": 4620
},
{
"epoch": 0.7834842203232084,
"grad_norm": 0.04484412283088257,
"learning_rate": 6.152977595627768e-06,
"loss": 0.0603,
"step": 4630
},
{
"epoch": 0.7851764108638633,
"grad_norm": 0.04134987164905719,
"learning_rate": 6.061357370807017e-06,
"loss": 0.0619,
"step": 4640
},
{
"epoch": 0.7868686014045182,
"grad_norm": 0.047814203539871335,
"learning_rate": 5.970330211463904e-06,
"loss": 0.0608,
"step": 4650
},
{
"epoch": 0.788560791945173,
"grad_norm": 0.05307745701463773,
"learning_rate": 5.879898968122987e-06,
"loss": 0.0603,
"step": 4660
},
{
"epoch": 0.7902529824858279,
"grad_norm": 0.036890676663936736,
"learning_rate": 5.790066472647637e-06,
"loss": 0.0593,
"step": 4670
},
{
"epoch": 0.7919451730264828,
"grad_norm": 0.0765386792024549,
"learning_rate": 5.7008355381513975e-06,
"loss": 0.0598,
"step": 4680
},
{
"epoch": 0.7936373635671377,
"grad_norm": 0.03811935562690127,
"learning_rate": 5.612208958909848e-06,
"loss": 0.0587,
"step": 4690
},
{
"epoch": 0.7953295541077925,
"grad_norm": 0.036126732537436174,
"learning_rate": 5.524189510273137e-06,
"loss": 0.0592,
"step": 4700
},
{
"epoch": 0.7970217446484474,
"grad_norm": 0.04388869708382243,
"learning_rate": 5.436779948579041e-06,
"loss": 0.0592,
"step": 4710
},
{
"epoch": 0.7987139351891023,
"grad_norm": 0.03715886623159806,
"learning_rate": 5.349983011066687e-06,
"loss": 0.06,
"step": 4720
},
{
"epoch": 0.8004061257297572,
"grad_norm": 0.03999842074335387,
"learning_rate": 5.263801415790817e-06,
"loss": 0.0623,
"step": 4730
},
{
"epoch": 0.8020983162704121,
"grad_norm": 0.04307858023050033,
"learning_rate": 5.178237861536647e-06,
"loss": 0.0616,
"step": 4740
},
{
"epoch": 0.8037905068110669,
"grad_norm": 0.057944246540194345,
"learning_rate": 5.093295027735392e-06,
"loss": 0.0587,
"step": 4750
},
{
"epoch": 0.8054826973517218,
"grad_norm": 0.04271848976367691,
"learning_rate": 5.008975574380337e-06,
"loss": 0.0605,
"step": 4760
},
{
"epoch": 0.8071748878923767,
"grad_norm": 0.03917832779114474,
"learning_rate": 4.925282141943557e-06,
"loss": 0.0605,
"step": 4770
},
{
"epoch": 0.8088670784330315,
"grad_norm": 0.038648040689787284,
"learning_rate": 4.842217351293207e-06,
"loss": 0.0576,
"step": 4780
},
{
"epoch": 0.8105592689736865,
"grad_norm": 0.04101393519370689,
"learning_rate": 4.759783803611476e-06,
"loss": 0.0587,
"step": 4790
},
{
"epoch": 0.8122514595143413,
"grad_norm": 0.04460833891541299,
"learning_rate": 4.6779840803131055e-06,
"loss": 0.06,
"step": 4800
},
{
"epoch": 0.8139436500549961,
"grad_norm": 0.036743006364765206,
"learning_rate": 4.596820742964559e-06,
"loss": 0.0597,
"step": 4810
},
{
"epoch": 0.8156358405956511,
"grad_norm": 0.03803732353743312,
"learning_rate": 4.516296333203829e-06,
"loss": 0.0592,
"step": 4820
},
{
"epoch": 0.8173280311363059,
"grad_norm": 0.03673281040176844,
"learning_rate": 4.436413372660814e-06,
"loss": 0.0586,
"step": 4830
},
{
"epoch": 0.8190202216769609,
"grad_norm": 0.03982742259760073,
"learning_rate": 4.357174362878369e-06,
"loss": 0.0602,
"step": 4840
},
{
"epoch": 0.8207124122176157,
"grad_norm": 0.03697109633677963,
"learning_rate": 4.278581785233957e-06,
"loss": 0.0609,
"step": 4850
},
{
"epoch": 0.8224046027582705,
"grad_norm": 0.0573065309747207,
"learning_rate": 4.20063810086197e-06,
"loss": 0.06,
"step": 4860
},
{
"epoch": 0.8240967932989255,
"grad_norm": 0.0455554580021282,
"learning_rate": 4.123345750576649e-06,
"loss": 0.0581,
"step": 4870
},
{
"epoch": 0.8257889838395803,
"grad_norm": 0.05207951890158224,
"learning_rate": 4.046707154795617e-06,
"loss": 0.059,
"step": 4880
},
{
"epoch": 0.8274811743802352,
"grad_norm": 0.0488756519104218,
"learning_rate": 3.97072471346413e-06,
"loss": 0.06,
"step": 4890
},
{
"epoch": 0.8291733649208901,
"grad_norm": 0.03617734911907982,
"learning_rate": 3.895400805979888e-06,
"loss": 0.0578,
"step": 4900
},
{
"epoch": 0.830865555461545,
"grad_norm": 0.03996411149095108,
"learning_rate": 3.820737791118553e-06,
"loss": 0.0592,
"step": 4910
},
{
"epoch": 0.8325577460021999,
"grad_norm": 0.04414735041579236,
"learning_rate": 3.7467380069598574e-06,
"loss": 0.0604,
"step": 4920
},
{
"epoch": 0.8342499365428547,
"grad_norm": 0.04148041703755401,
"learning_rate": 3.6734037708144097e-06,
"loss": 0.0604,
"step": 4930
},
{
"epoch": 0.8359421270835096,
"grad_norm": 0.04799457131190261,
"learning_rate": 3.6007373791511033e-06,
"loss": 0.0612,
"step": 4940
},
{
"epoch": 0.8376343176241645,
"grad_norm": 0.03885006014011949,
"learning_rate": 3.528741107525216e-06,
"loss": 0.0608,
"step": 4950
},
{
"epoch": 0.8393265081648194,
"grad_norm": 0.03817155033176637,
"learning_rate": 3.457417210507166e-06,
"loss": 0.0596,
"step": 4960
},
{
"epoch": 0.8410186987054742,
"grad_norm": 0.03731023374790798,
"learning_rate": 3.386767921611872e-06,
"loss": 0.0588,
"step": 4970
},
{
"epoch": 0.8427108892461291,
"grad_norm": 0.041531537308173175,
"learning_rate": 3.3167954532288574e-06,
"loss": 0.0587,
"step": 4980
},
{
"epoch": 0.844403079786784,
"grad_norm": 0.05687260120116034,
"learning_rate": 3.2475019965529095e-06,
"loss": 0.0599,
"step": 4990
},
{
"epoch": 0.8460952703274389,
"grad_norm": 0.03765539617352035,
"learning_rate": 3.1788897215155268e-06,
"loss": 0.0608,
"step": 5000
},
{
"epoch": 0.8460952703274389,
"eval_loss": 0.06013359874486923,
"eval_runtime": 510.524,
"eval_samples_per_second": 11.431,
"eval_steps_per_second": 0.715,
"step": 5000
},
{
"epoch": 0.8477874608680938,
"grad_norm": 0.038309364458009816,
"learning_rate": 3.1109607767169286e-06,
"loss": 0.0585,
"step": 5010
},
{
"epoch": 0.8494796514087486,
"grad_norm": 0.03927952804348791,
"learning_rate": 3.0437172893587774e-06,
"loss": 0.059,
"step": 5020
},
{
"epoch": 0.8511718419494035,
"grad_norm": 0.041197176100621125,
"learning_rate": 2.9771613651775665e-06,
"loss": 0.0602,
"step": 5030
},
{
"epoch": 0.8528640324900584,
"grad_norm": 0.04992272222336298,
"learning_rate": 2.9112950883786766e-06,
"loss": 0.0597,
"step": 5040
},
{
"epoch": 0.8545562230307132,
"grad_norm": 0.04000073843152877,
"learning_rate": 2.8461205215711246e-06,
"loss": 0.0617,
"step": 5050
},
{
"epoch": 0.8562484135713682,
"grad_norm": 0.04160266596509446,
"learning_rate": 2.781639705702943e-06,
"loss": 0.0619,
"step": 5060
},
{
"epoch": 0.857940604112023,
"grad_norm": 0.05213779650162722,
"learning_rate": 2.71785465999731e-06,
"loss": 0.0609,
"step": 5070
},
{
"epoch": 0.859632794652678,
"grad_norm": 0.035078052471428256,
"learning_rate": 2.6547673818892615e-06,
"loss": 0.0603,
"step": 5080
},
{
"epoch": 0.8613249851933328,
"grad_norm": 0.039507454080708694,
"learning_rate": 2.5923798469631916e-06,
"loss": 0.0608,
"step": 5090
},
{
"epoch": 0.8630171757339876,
"grad_norm": 0.03875575324620654,
"learning_rate": 2.530694008890963e-06,
"loss": 0.0589,
"step": 5100
},
{
"epoch": 0.8647093662746426,
"grad_norm": 0.04050751075443286,
"learning_rate": 2.4697117993707247e-06,
"loss": 0.0608,
"step": 5110
},
{
"epoch": 0.8664015568152974,
"grad_norm": 0.03517717295387998,
"learning_rate": 2.409435128066445e-06,
"loss": 0.0598,
"step": 5120
},
{
"epoch": 0.8680937473559522,
"grad_norm": 0.036996821632751614,
"learning_rate": 2.3498658825480717e-06,
"loss": 0.0598,
"step": 5130
},
{
"epoch": 0.8697859378966072,
"grad_norm": 0.03883966277953073,
"learning_rate": 2.2910059282324534e-06,
"loss": 0.0589,
"step": 5140
},
{
"epoch": 0.871478128437262,
"grad_norm": 0.039833006966338826,
"learning_rate": 2.2328571083249207e-06,
"loss": 0.0595,
"step": 5150
},
{
"epoch": 0.8731703189779169,
"grad_norm": 0.036529968234169014,
"learning_rate": 2.175421243761541e-06,
"loss": 0.0575,
"step": 5160
},
{
"epoch": 0.8748625095185718,
"grad_norm": 0.03443644169221473,
"learning_rate": 2.1187001331521433e-06,
"loss": 0.0584,
"step": 5170
},
{
"epoch": 0.8765547000592266,
"grad_norm": 0.042313635722485456,
"learning_rate": 2.0626955527239283e-06,
"loss": 0.0594,
"step": 5180
},
{
"epoch": 0.8782468905998816,
"grad_norm": 0.04205370483511292,
"learning_rate": 2.007409256265916e-06,
"loss": 0.0596,
"step": 5190
},
{
"epoch": 0.8799390811405364,
"grad_norm": 0.06952118465028437,
"learning_rate": 1.952842975073971e-06,
"loss": 0.0587,
"step": 5200
},
{
"epoch": 0.8816312716811913,
"grad_norm": 0.04117806502556948,
"learning_rate": 1.8989984178966246e-06,
"loss": 0.0614,
"step": 5210
},
{
"epoch": 0.8833234622218462,
"grad_norm": 0.058478390973884284,
"learning_rate": 1.845877270881538e-06,
"loss": 0.0608,
"step": 5220
},
{
"epoch": 0.885015652762501,
"grad_norm": 0.039511810377841905,
"learning_rate": 1.7934811975227033e-06,
"loss": 0.0605,
"step": 5230
},
{
"epoch": 0.8867078433031559,
"grad_norm": 0.04219145977899964,
"learning_rate": 1.7418118386083843e-06,
"loss": 0.0591,
"step": 5240
},
{
"epoch": 0.8884000338438108,
"grad_norm": 0.037168579182890245,
"learning_rate": 1.6908708121696842e-06,
"loss": 0.0588,
"step": 5250
},
{
"epoch": 0.8900922243844657,
"grad_norm": 0.03429587442786015,
"learning_rate": 1.6406597134299178e-06,
"loss": 0.0587,
"step": 5260
},
{
"epoch": 0.8917844149251206,
"grad_norm": 0.04132605827036987,
"learning_rate": 1.5911801147546369e-06,
"loss": 0.06,
"step": 5270
},
{
"epoch": 0.8934766054657755,
"grad_norm": 0.03501424713836602,
"learning_rate": 1.5424335656023886e-06,
"loss": 0.0598,
"step": 5280
},
{
"epoch": 0.8951687960064303,
"grad_norm": 0.03685759685435392,
"learning_rate": 1.4944215924762168e-06,
"loss": 0.0577,
"step": 5290
},
{
"epoch": 0.8968609865470852,
"grad_norm": 0.03698566286048149,
"learning_rate": 1.4471456988758292e-06,
"loss": 0.0592,
"step": 5300
},
{
"epoch": 0.8985531770877401,
"grad_norm": 0.03725409439290792,
"learning_rate": 1.4006073652505419e-06,
"loss": 0.0599,
"step": 5310
},
{
"epoch": 0.9002453676283949,
"grad_norm": 0.043456891428921354,
"learning_rate": 1.3548080489528963e-06,
"loss": 0.0599,
"step": 5320
},
{
"epoch": 0.9019375581690499,
"grad_norm": 0.03733579656039839,
"learning_rate": 1.3097491841930403e-06,
"loss": 0.0585,
"step": 5330
},
{
"epoch": 0.9036297487097047,
"grad_norm": 0.03775814252068895,
"learning_rate": 1.2654321819938004e-06,
"loss": 0.058,
"step": 5340
},
{
"epoch": 0.9053219392503596,
"grad_norm": 0.03324948771468124,
"learning_rate": 1.2218584301465107e-06,
"loss": 0.059,
"step": 5350
},
{
"epoch": 0.9070141297910145,
"grad_norm": 0.03920515209486877,
"learning_rate": 1.179029293167555e-06,
"loss": 0.0579,
"step": 5360
},
{
"epoch": 0.9087063203316693,
"grad_norm": 0.04820685014256461,
"learning_rate": 1.1369461122556018e-06,
"loss": 0.0588,
"step": 5370
},
{
"epoch": 0.9103985108723243,
"grad_norm": 0.036669201705327836,
"learning_rate": 1.0956102052496558e-06,
"loss": 0.059,
"step": 5380
},
{
"epoch": 0.9120907014129791,
"grad_norm": 0.06147241492633053,
"learning_rate": 1.0550228665877543e-06,
"loss": 0.0601,
"step": 5390
},
{
"epoch": 0.9137828919536339,
"grad_norm": 0.035151975378681186,
"learning_rate": 1.015185367266447e-06,
"loss": 0.0578,
"step": 5400
},
{
"epoch": 0.9154750824942889,
"grad_norm": 0.04332126418524112,
"learning_rate": 9.76098954800983e-07,
"loss": 0.0591,
"step": 5410
},
{
"epoch": 0.9171672730349437,
"grad_norm": 0.03574485899781292,
"learning_rate": 9.377648531862598e-07,
"loss": 0.0591,
"step": 5420
},
{
"epoch": 0.9188594635755987,
"grad_norm": 0.03710560618301365,
"learning_rate": 9.001842628584806e-07,
"loss": 0.0592,
"step": 5430
},
{
"epoch": 0.9205516541162535,
"grad_norm": 0.04017156375680633,
"learning_rate": 8.633583606575679e-07,
"loss": 0.0629,
"step": 5440
},
{
"epoch": 0.9222438446569083,
"grad_norm": 0.05141445478540544,
"learning_rate": 8.272882997903186e-07,
"loss": 0.0588,
"step": 5450
},
{
"epoch": 0.9239360351975633,
"grad_norm": 0.037371118355001766,
"learning_rate": 7.919752097942712e-07,
"loss": 0.0589,
"step": 5460
},
{
"epoch": 0.9256282257382181,
"grad_norm": 0.06332229086085672,
"learning_rate": 7.574201965023508e-07,
"loss": 0.0613,
"step": 5470
},
{
"epoch": 0.927320416278873,
"grad_norm": 0.038810424325564884,
"learning_rate": 7.236243420082334e-07,
"loss": 0.0578,
"step": 5480
},
{
"epoch": 0.9290126068195279,
"grad_norm": 0.04166578888884816,
"learning_rate": 6.905887046324693e-07,
"loss": 0.0575,
"step": 5490
},
{
"epoch": 0.9307047973601827,
"grad_norm": 0.041431921917415765,
"learning_rate": 6.583143188893303e-07,
"loss": 0.0604,
"step": 5500
},
{
"epoch": 0.9323969879008376,
"grad_norm": 0.035477182990891386,
"learning_rate": 6.268021954544096e-07,
"loss": 0.0582,
"step": 5510
},
{
"epoch": 0.9340891784414925,
"grad_norm": 0.038958165621434504,
"learning_rate": 5.960533211329917e-07,
"loss": 0.0602,
"step": 5520
},
{
"epoch": 0.9357813689821474,
"grad_norm": 0.03397886542856727,
"learning_rate": 5.660686588291303e-07,
"loss": 0.0587,
"step": 5530
},
{
"epoch": 0.9374735595228023,
"grad_norm": 0.037968200391315654,
"learning_rate": 5.368491475155141e-07,
"loss": 0.0588,
"step": 5540
},
{
"epoch": 0.9391657500634571,
"grad_norm": 0.038852001885395664,
"learning_rate": 5.083957022040398e-07,
"loss": 0.0574,
"step": 5550
},
{
"epoch": 0.940857940604112,
"grad_norm": 0.03745419242162026,
"learning_rate": 4.807092139171771e-07,
"loss": 0.0591,
"step": 5560
},
{
"epoch": 0.9425501311447669,
"grad_norm": 0.03907096330901171,
"learning_rate": 4.5379054966006585e-07,
"loss": 0.059,
"step": 5570
},
{
"epoch": 0.9442423216854218,
"grad_norm": 0.04968241184370766,
"learning_rate": 4.2764055239335186e-07,
"loss": 0.06,
"step": 5580
},
{
"epoch": 0.9459345122260766,
"grad_norm": 0.04149978746162696,
"learning_rate": 4.022600410067995e-07,
"loss": 0.0606,
"step": 5590
},
{
"epoch": 0.9476267027667316,
"grad_norm": 0.036708205150157824,
"learning_rate": 3.7764981029365097e-07,
"loss": 0.0583,
"step": 5600
},
{
"epoch": 0.9493188933073864,
"grad_norm": 0.03989442790693185,
"learning_rate": 3.538106309257272e-07,
"loss": 0.059,
"step": 5610
},
{
"epoch": 0.9510110838480413,
"grad_norm": 0.042628169535712294,
"learning_rate": 3.3074324942929667e-07,
"loss": 0.0596,
"step": 5620
},
{
"epoch": 0.9527032743886962,
"grad_norm": 0.04178066045162929,
"learning_rate": 3.084483881617084e-07,
"loss": 0.0587,
"step": 5630
},
{
"epoch": 0.954395464929351,
"grad_norm": 0.03999187608785915,
"learning_rate": 2.8692674528876527e-07,
"loss": 0.0584,
"step": 5640
},
{
"epoch": 0.956087655470006,
"grad_norm": 0.03753331187888784,
"learning_rate": 2.661789947628501e-07,
"loss": 0.06,
"step": 5650
},
{
"epoch": 0.9577798460106608,
"grad_norm": 0.03798306906458078,
"learning_rate": 2.4620578630183134e-07,
"loss": 0.0595,
"step": 5660
},
{
"epoch": 0.9594720365513156,
"grad_norm": 0.041922810904808615,
"learning_rate": 2.270077453687236e-07,
"loss": 0.0576,
"step": 5670
},
{
"epoch": 0.9611642270919706,
"grad_norm": 0.03793657385286506,
"learning_rate": 2.08585473152087e-07,
"loss": 0.0607,
"step": 5680
},
{
"epoch": 0.9628564176326254,
"grad_norm": 0.040585934400225614,
"learning_rate": 1.9093954654720582e-07,
"loss": 0.0581,
"step": 5690
},
{
"epoch": 0.9645486081732804,
"grad_norm": 0.03702989382494412,
"learning_rate": 1.7407051813803087e-07,
"loss": 0.0586,
"step": 5700
},
{
"epoch": 0.9662407987139352,
"grad_norm": 0.03734149052120102,
"learning_rate": 1.5797891617986282e-07,
"loss": 0.059,
"step": 5710
},
{
"epoch": 0.96793298925459,
"grad_norm": 0.035570939343692515,
"learning_rate": 1.4266524458281526e-07,
"loss": 0.0574,
"step": 5720
},
{
"epoch": 0.969625179795245,
"grad_norm": 0.034389309413600175,
"learning_rate": 1.281299828960386e-07,
"loss": 0.0586,
"step": 5730
},
{
"epoch": 0.9713173703358998,
"grad_norm": 0.0409938258757912,
"learning_rate": 1.1437358629270146e-07,
"loss": 0.0608,
"step": 5740
},
{
"epoch": 0.9730095608765547,
"grad_norm": 0.0355315124005057,
"learning_rate": 1.0139648555572978e-07,
"loss": 0.0581,
"step": 5750
},
{
"epoch": 0.9747017514172096,
"grad_norm": 0.03763186545296007,
"learning_rate": 8.919908706432056e-08,
"loss": 0.0577,
"step": 5760
},
{
"epoch": 0.9763939419578644,
"grad_norm": 0.03794455726139563,
"learning_rate": 7.778177278122412e-08,
"loss": 0.0592,
"step": 5770
},
{
"epoch": 0.9780861324985193,
"grad_norm": 0.04508110131875037,
"learning_rate": 6.714490024077591e-08,
"loss": 0.0589,
"step": 5780
},
{
"epoch": 0.9797783230391742,
"grad_norm": 0.04170037599854637,
"learning_rate": 5.7288802537691645e-08,
"loss": 0.0601,
"step": 5790
},
{
"epoch": 0.9814705135798291,
"grad_norm": 0.03461557528272302,
"learning_rate": 4.821378831665335e-08,
"loss": 0.0592,
"step": 5800
},
{
"epoch": 0.983162704120484,
"grad_norm": 0.036396171830752964,
"learning_rate": 3.992014176263381e-08,
"loss": 0.0599,
"step": 5810
},
{
"epoch": 0.9848548946611388,
"grad_norm": 0.03346203310487427,
"learning_rate": 3.2408122592000876e-08,
"loss": 0.0593,
"step": 5820
},
{
"epoch": 0.9865470852017937,
"grad_norm": 0.0381442787084737,
"learning_rate": 2.5677966044379553e-08,
"loss": 0.0595,
"step": 5830
},
{
"epoch": 0.9882392757424486,
"grad_norm": 0.045855633648734115,
"learning_rate": 1.9729882875296757e-08,
"loss": 0.059,
"step": 5840
},
{
"epoch": 0.9899314662831035,
"grad_norm": 0.03559538703643374,
"learning_rate": 1.4564059349561621e-08,
"loss": 0.0598,
"step": 5850
},
{
"epoch": 0.9916236568237583,
"grad_norm": 0.036840407623005736,
"learning_rate": 1.0180657235459023e-08,
"loss": 0.06,
"step": 5860
},
{
"epoch": 0.9933158473644133,
"grad_norm": 0.037312840018662065,
"learning_rate": 6.579813799656442e-09,
"loss": 0.0593,
"step": 5870
},
{
"epoch": 0.9950080379050681,
"grad_norm": 0.03925209952479138,
"learning_rate": 3.761641802924043e-09,
"loss": 0.0591,
"step": 5880
},
{
"epoch": 0.996700228445723,
"grad_norm": 0.042163126698245304,
"learning_rate": 1.7262294966013949e-09,
"loss": 0.0593,
"step": 5890
},
{
"epoch": 0.9983924189863779,
"grad_norm": 0.03639727948907427,
"learning_rate": 4.736406198274645e-10,
"loss": 0.0591,
"step": 5900
},
{
"epoch": 1.0,
"grad_norm": 0.07202303845764885,
"learning_rate": 3.914397544990678e-12,
"loss": 0.0586,
"step": 5910
},
{
"epoch": 1.0,
"step": 5910,
"total_flos": 4644350000627712.0,
"train_loss": 0.08661497140495507,
"train_runtime": 70638.1483,
"train_samples_per_second": 2.677,
"train_steps_per_second": 0.084
}
],
"logging_steps": 10,
"max_steps": 5910,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 1000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 4644350000627712.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}