ap-sql-v1 / run_summary.json
samrat-kar's picture
Add ap-sql-v1 LoRA adapter
b0922bc verified
Raw History Blame Contribute Delete
52 kB
{
"base_model": "Qwen/Qwen2.5-Coder-7B-Instruct",
"rows_train": 1441,
"rows_val": 95,
"steps": 273,
"train_loss": 0.01574120909053851,
"seconds": 9101.8,
"peak_gpu_memory_gb": 17.54,
"lora_r": 16,
"lora_alpha": 32,
"target_modules": [
"q_proj",
"k_proj",
"v_proj",
"o_proj",
"gate_proj",
"up_proj",
"down_proj"
],
"log_history": [
{
"loss": 0.362233966588974,
"grad_norm": 0.35578352212905884,
"learning_rate": 0.0,
"epoch": 0.011103400416377515,
"step": 1
},
{
"loss": 0.21971727907657623,
"grad_norm": 0.2704537808895111,
"learning_rate": 6.666666666666667e-05,
"epoch": 0.02220680083275503,
"step": 2
},
{
"loss": 0.2119903415441513,
"grad_norm": 0.23446440696716309,
"learning_rate": 0.00013333333333333334,
"epoch": 0.033310201249132546,
"step": 3
},
{
"loss": 0.3059872090816498,
"grad_norm": 0.2956233620643616,
"learning_rate": 0.0002,
"epoch": 0.04441360166551006,
"step": 4
},
{
"loss": 0.2478485405445099,
"grad_norm": 0.30406656861305237,
"learning_rate": 0.00019999323080037624,
"epoch": 0.055517002081887576,
"step": 5
},
{
"loss": 0.11736467480659485,
"grad_norm": 0.19965699315071106,
"learning_rate": 0.00019997292411794618,
"epoch": 0.06662040249826509,
"step": 6
},
{
"loss": 0.1486368626356125,
"grad_norm": 0.23600247502326965,
"learning_rate": 0.0001999390827019096,
"epoch": 0.0777238029146426,
"step": 7
},
{
"loss": 0.09324199706315994,
"grad_norm": 0.2227136790752411,
"learning_rate": 0.0001998917111338525,
"epoch": 0.08882720333102012,
"step": 8
},
{
"loss": 0.07419925928115845,
"grad_norm": 0.212725430727005,
"learning_rate": 0.00019983081582712685,
"epoch": 0.09993060374739764,
"step": 9
},
{
"loss": 0.09104721248149872,
"grad_norm": 0.2687018811702728,
"learning_rate": 0.00019975640502598244,
"epoch": 0.11103400416377515,
"step": 10
},
{
"loss": 0.07975094020366669,
"grad_norm": 0.2289663851261139,
"learning_rate": 0.00019966848880445062,
"epoch": 0.12213740458015267,
"step": 11
},
{
"loss": 0.07311172783374786,
"grad_norm": 0.2065844088792801,
"learning_rate": 0.00019956707906498044,
"epoch": 0.13324080499653018,
"step": 12
},
{
"loss": 0.06214028224349022,
"grad_norm": 0.16729624569416046,
"learning_rate": 0.00019945218953682734,
"epoch": 0.1443442054129077,
"step": 13
},
{
"loss": 0.039421241730451584,
"grad_norm": 0.14760349690914154,
"learning_rate": 0.00019932383577419432,
"epoch": 0.1554476058292852,
"step": 14
},
{
"loss": 0.04088294133543968,
"grad_norm": 0.1884099245071411,
"learning_rate": 0.00019918203515412617,
"epoch": 0.16655100624566274,
"step": 15
},
{
"loss": 0.03639013320207596,
"grad_norm": 0.14482857286930084,
"learning_rate": 0.00019902680687415705,
"epoch": 0.17765440666204024,
"step": 16
},
{
"loss": 0.05018017068505287,
"grad_norm": 0.20465002954006195,
"learning_rate": 0.00019885817194971117,
"epoch": 0.18875780707841777,
"step": 17
},
{
"loss": 0.09407485276460648,
"grad_norm": 0.208365797996521,
"learning_rate": 0.00019867615321125795,
"epoch": 0.19986120749479527,
"step": 18
},
{
"loss": 0.0462566502392292,
"grad_norm": 0.2098541110754013,
"learning_rate": 0.00019848077530122083,
"epoch": 0.2109646079111728,
"step": 19
},
{
"loss": 0.03283077850937843,
"grad_norm": 0.16920238733291626,
"learning_rate": 0.00019827206467064133,
"epoch": 0.2220680083275503,
"step": 20
},
{
"loss": 0.031955886632204056,
"grad_norm": 0.12768959999084473,
"learning_rate": 0.00019805004957559793,
"epoch": 0.23317140874392783,
"step": 21
},
{
"loss": 0.034089263528585434,
"grad_norm": 0.13755914568901062,
"learning_rate": 0.00019781476007338058,
"epoch": 0.24427480916030533,
"step": 22
},
{
"loss": 0.02859678864479065,
"grad_norm": 0.1348200887441635,
"learning_rate": 0.00019756622801842143,
"epoch": 0.25537820957668284,
"step": 23
},
{
"loss": 0.02967454306781292,
"grad_norm": 0.17000287771224976,
"learning_rate": 0.00019730448705798239,
"epoch": 0.26648160999306036,
"step": 24
},
{
"loss": 0.03603748977184296,
"grad_norm": 0.1551555097103119,
"learning_rate": 0.00019702957262759965,
"epoch": 0.2775850104094379,
"step": 25
},
{
"loss": 0.06082950532436371,
"grad_norm": 0.24252957105636597,
"learning_rate": 0.00019674152194628638,
"epoch": 0.2886884108258154,
"step": 26
},
{
"loss": 0.03331135958433151,
"grad_norm": 0.16082148253917694,
"learning_rate": 0.0001964403740114939,
"epoch": 0.2997918112421929,
"step": 27
},
{
"loss": 0.016830770298838615,
"grad_norm": 0.11476831883192062,
"learning_rate": 0.0001961261695938319,
"epoch": 0.3108952116585704,
"step": 28
},
{
"loss": 0.020966561511158943,
"grad_norm": 0.12785278260707855,
"learning_rate": 0.0001957989512315489,
"epoch": 0.32199861207494795,
"step": 29
},
{
"loss": 0.019154518842697144,
"grad_norm": 0.18446354568004608,
"learning_rate": 0.0001954587632247732,
"epoch": 0.3331020124913255,
"step": 30
},
{
"loss": 0.015241650864481926,
"grad_norm": 0.12455496937036514,
"learning_rate": 0.00019510565162951537,
"epoch": 0.34420541290770296,
"step": 31
},
{
"loss": 0.02742624841630459,
"grad_norm": 0.16151973605155945,
"learning_rate": 0.00019473966425143292,
"epoch": 0.3553088133240805,
"step": 32
},
{
"loss": 0.03055701218545437,
"grad_norm": 0.1681850254535675,
"learning_rate": 0.00019436085063935835,
"epoch": 0.366412213740458,
"step": 33
},
{
"loss": 0.03377750143408775,
"grad_norm": 0.18554241955280304,
"learning_rate": 0.00019396926207859084,
"epoch": 0.37751561415683554,
"step": 34
},
{
"loss": 0.011026578955352306,
"grad_norm": 0.08266780525445938,
"learning_rate": 0.00019356495158395315,
"epoch": 0.3886190145732131,
"step": 35
},
{
"loss": 0.011678414419293404,
"grad_norm": 0.0954771563410759,
"learning_rate": 0.00019314797389261424,
"epoch": 0.39972241498959055,
"step": 36
},
{
"loss": 0.003908986691385508,
"grad_norm": 0.057436637580394745,
"learning_rate": 0.00019271838545667876,
"epoch": 0.4108258154059681,
"step": 37
},
{
"loss": 0.03971560299396515,
"grad_norm": 0.1711757332086563,
"learning_rate": 0.00019227624443554425,
"epoch": 0.4219292158223456,
"step": 38
},
{
"loss": 0.026749372482299805,
"grad_norm": 0.15317557752132416,
"learning_rate": 0.00019182161068802741,
"epoch": 0.43303261623872313,
"step": 39
},
{
"loss": 0.005962643772363663,
"grad_norm": 0.07149907946586609,
"learning_rate": 0.0001913545457642601,
"epoch": 0.4441360166551006,
"step": 40
},
{
"loss": 0.034220557659864426,
"grad_norm": 0.13144898414611816,
"learning_rate": 0.00019087511289735644,
"epoch": 0.45523941707147814,
"step": 41
},
{
"loss": 0.013403044082224369,
"grad_norm": 0.10616762936115265,
"learning_rate": 0.00019038337699485208,
"epoch": 0.46634281748785567,
"step": 42
},
{
"loss": 0.0515558123588562,
"grad_norm": 0.25350019335746765,
"learning_rate": 0.0001898794046299167,
"epoch": 0.4774462179042332,
"step": 43
},
{
"loss": 0.006211064290255308,
"grad_norm": 0.07005950063467026,
"learning_rate": 0.00018936326403234125,
"epoch": 0.48854961832061067,
"step": 44
},
{
"loss": 0.03814203664660454,
"grad_norm": 0.18773649632930756,
"learning_rate": 0.00018883502507930042,
"epoch": 0.4996530187369882,
"step": 45
},
{
"loss": 0.004700696095824242,
"grad_norm": 0.04619601368904114,
"learning_rate": 0.00018829475928589271,
"epoch": 0.5107564191533657,
"step": 46
},
{
"loss": 0.01920146495103836,
"grad_norm": 0.21882829070091248,
"learning_rate": 0.0001877425397954582,
"epoch": 0.5218598195697433,
"step": 47
},
{
"loss": 0.015273750759661198,
"grad_norm": 0.179578959941864,
"learning_rate": 0.00018717844136967624,
"epoch": 0.5329632199861207,
"step": 48
},
{
"loss": 0.02806437574326992,
"grad_norm": 0.15267197787761688,
"learning_rate": 0.00018660254037844388,
"epoch": 0.5440666204024983,
"step": 49
},
{
"loss": 0.02385113015770912,
"grad_norm": 0.20542870461940765,
"learning_rate": 0.00018601491478953657,
"epoch": 0.5551700208188758,
"step": 50
},
{
"loss": 0.028099708259105682,
"grad_norm": 0.1347873955965042,
"learning_rate": 0.00018541564415805258,
"epoch": 0.5662734212352533,
"step": 51
},
{
"loss": 0.019417954608798027,
"grad_norm": 0.09850087761878967,
"learning_rate": 0.0001848048096156426,
"epoch": 0.5773768216516308,
"step": 52
},
{
"loss": 0.004227581899613142,
"grad_norm": 0.06576558947563171,
"learning_rate": 0.00018418249385952575,
"epoch": 0.5884802220680083,
"step": 53
},
{
"loss": 0.005577925592660904,
"grad_norm": 0.06200479716062546,
"learning_rate": 0.00018354878114129367,
"epoch": 0.5995836224843858,
"step": 54
},
{
"loss": 0.049879446625709534,
"grad_norm": 0.3281773328781128,
"learning_rate": 0.00018290375725550417,
"epoch": 0.6106870229007634,
"step": 55
},
{
"loss": 0.006307754199951887,
"grad_norm": 0.06183058023452759,
"learning_rate": 0.00018224750952806624,
"epoch": 0.6217904233171409,
"step": 56
},
{
"loss": 0.04659178853034973,
"grad_norm": 0.20692244172096252,
"learning_rate": 0.00018158012680441723,
"epoch": 0.6328938237335184,
"step": 57
},
{
"loss": 0.022197065874934196,
"grad_norm": 0.1442933827638626,
"learning_rate": 0.00018090169943749476,
"epoch": 0.6439972241498959,
"step": 58
},
{
"loss": 0.011002053506672382,
"grad_norm": 0.11611606925725937,
"learning_rate": 0.0001802123192755044,
"epoch": 0.6551006245662734,
"step": 59
},
{
"loss": 0.005552457645535469,
"grad_norm": 0.06183777004480362,
"learning_rate": 0.0001795120796494848,
"epoch": 0.666204024982651,
"step": 60
},
{
"loss": 0.00588051276281476,
"grad_norm": 0.07396812736988068,
"learning_rate": 0.00017880107536067218,
"epoch": 0.6773074253990284,
"step": 61
},
{
"loss": 0.005040598101913929,
"grad_norm": 0.06021937355399132,
"learning_rate": 0.00017807940266766593,
"epoch": 0.6884108258154059,
"step": 62
},
{
"loss": 0.006408208981156349,
"grad_norm": 0.0630301982164383,
"learning_rate": 0.0001773471592733964,
"epoch": 0.6995142262317835,
"step": 63
},
{
"loss": 0.03109092079102993,
"grad_norm": 0.1972433626651764,
"learning_rate": 0.0001766044443118978,
"epoch": 0.710617626648161,
"step": 64
},
{
"loss": 0.01442437432706356,
"grad_norm": 0.2614825665950775,
"learning_rate": 0.00017585135833488692,
"epoch": 0.7217210270645386,
"step": 65
},
{
"loss": 0.02434384636580944,
"grad_norm": 0.11953062564134598,
"learning_rate": 0.00017508800329814995,
"epoch": 0.732824427480916,
"step": 66
},
{
"loss": 0.003504230175167322,
"grad_norm": 0.056314583867788315,
"learning_rate": 0.00017431448254773944,
"epoch": 0.7439278278972935,
"step": 67
},
{
"loss": 0.01270079892128706,
"grad_norm": 0.08628875762224197,
"learning_rate": 0.0001735309008059829,
"epoch": 0.7550312283136711,
"step": 68
},
{
"loss": 0.013141262345016003,
"grad_norm": 0.09234391897916794,
"learning_rate": 0.00017273736415730488,
"epoch": 0.7661346287300486,
"step": 69
},
{
"loss": 0.021891677752137184,
"grad_norm": 0.1135958731174469,
"learning_rate": 0.0001719339800338651,
"epoch": 0.7772380291464261,
"step": 70
},
{
"loss": 0.03220357373356819,
"grad_norm": 0.12099183350801468,
"learning_rate": 0.00017112085720101373,
"epoch": 0.7883414295628036,
"step": 71
},
{
"loss": 0.006556101609021425,
"grad_norm": 0.1038985624909401,
"learning_rate": 0.0001702981057425662,
"epoch": 0.7994448299791811,
"step": 72
},
{
"loss": 0.0022207903675734997,
"grad_norm": 0.02890835516154766,
"learning_rate": 0.00016946583704589973,
"epoch": 0.8105482303955587,
"step": 73
},
{
"loss": 0.006729573477059603,
"grad_norm": 0.07326466590166092,
"learning_rate": 0.0001686241637868734,
"epoch": 0.8216516308119362,
"step": 74
},
{
"loss": 0.009240851737558842,
"grad_norm": 0.09405697882175446,
"learning_rate": 0.00016777319991457325,
"epoch": 0.8327550312283136,
"step": 75
},
{
"loss": 0.0017095255898311734,
"grad_norm": 0.018018698319792747,
"learning_rate": 0.00016691306063588583,
"epoch": 0.8438584316446912,
"step": 76
},
{
"loss": 0.003485671477392316,
"grad_norm": 0.09172848612070084,
"learning_rate": 0.00016604386239990078,
"epoch": 0.8549618320610687,
"step": 77
},
{
"loss": 0.006044326350092888,
"grad_norm": 0.08144538104534149,
"learning_rate": 0.00016516572288214552,
"epoch": 0.8660652324774463,
"step": 78
},
{
"loss": 0.019570548087358475,
"grad_norm": 0.12051013112068176,
"learning_rate": 0.00016427876096865394,
"epoch": 0.8771686328938237,
"step": 79
},
{
"loss": 0.029929179698228836,
"grad_norm": 0.12691369652748108,
"learning_rate": 0.00016338309673987101,
"epoch": 0.8882720333102012,
"step": 80
},
{
"loss": 0.009485170245170593,
"grad_norm": 0.07689705491065979,
"learning_rate": 0.000162478851454396,
"epoch": 0.8993754337265788,
"step": 81
},
{
"loss": 0.007876642979681492,
"grad_norm": 0.07503333687782288,
"learning_rate": 0.0001615661475325658,
"epoch": 0.9104788341429563,
"step": 82
},
{
"loss": 0.005961703136563301,
"grad_norm": 0.05012199282646179,
"learning_rate": 0.00016064510853988138,
"epoch": 0.9215822345593337,
"step": 83
},
{
"loss": 0.020787648856639862,
"grad_norm": 0.1281776875257492,
"learning_rate": 0.00015971585917027862,
"epoch": 0.9326856349757113,
"step": 84
},
{
"loss": 0.000465249118860811,
"grad_norm": 0.008060788735747337,
"learning_rate": 0.00015877852522924732,
"epoch": 0.9437890353920888,
"step": 85
},
{
"loss": 0.016680452972650528,
"grad_norm": 0.08592799305915833,
"learning_rate": 0.00015783323361679864,
"epoch": 0.9548924358084664,
"step": 86
},
{
"loss": 0.009535298682749271,
"grad_norm": 0.0659300684928894,
"learning_rate": 0.00015688011231028518,
"epoch": 0.9659958362248439,
"step": 87
},
{
"loss": 0.01739584095776081,
"grad_norm": 0.1285824328660965,
"learning_rate": 0.0001559192903470747,
"epoch": 0.9770992366412213,
"step": 88
},
{
"loss": 0.013622242957353592,
"grad_norm": 0.1113409548997879,
"learning_rate": 0.0001549508978070806,
"epoch": 0.9882026370575989,
"step": 89
},
{
"loss": 0.012220436707139015,
"grad_norm": 0.0880318358540535,
"learning_rate": 0.0001539750657951513,
"epoch": 0.9993060374739764,
"step": 90
},
{
"loss": 0.001605391502380371,
"grad_norm": 0.08987312018871307,
"learning_rate": 0.0001529919264233205,
"epoch": 1.0,
"step": 91
},
{
"loss": 0.005238894373178482,
"grad_norm": 0.05442668870091438,
"learning_rate": 0.00015200161279292155,
"epoch": 1.0111034004163775,
"step": 92
},
{
"loss": 0.003774596843868494,
"grad_norm": 0.051331259310245514,
"learning_rate": 0.00015100425897656753,
"epoch": 1.022206800832755,
"step": 93
},
{
"loss": 0.0033705425448715687,
"grad_norm": 0.13363640010356903,
"learning_rate": 0.00015000000000000001,
"epoch": 1.0333102012491326,
"step": 94
},
{
"loss": 0.004251929931342602,
"grad_norm": 0.04953952506184578,
"learning_rate": 0.0001489889718238087,
"epoch": 1.0444136016655101,
"step": 95
},
{
"loss": 0.0062233032658696175,
"grad_norm": 0.08337115496397018,
"learning_rate": 0.00014797131132502465,
"epoch": 1.0555170020818876,
"step": 96
},
{
"loss": 0.00801208708435297,
"grad_norm": 0.07984863221645355,
"learning_rate": 0.00014694715627858908,
"epoch": 1.066620402498265,
"step": 97
},
{
"loss": 0.002963041188195348,
"grad_norm": 0.05126333609223366,
"learning_rate": 0.00014591664533870118,
"epoch": 1.0777238029146425,
"step": 98
},
{
"loss": 0.008451717905700207,
"grad_norm": 0.0980011373758316,
"learning_rate": 0.00014487991802004623,
"epoch": 1.0888272033310202,
"step": 99
},
{
"loss": 0.0033840062096714973,
"grad_norm": 0.09044548869132996,
"learning_rate": 0.00014383711467890774,
"epoch": 1.0999306037473977,
"step": 100
},
{
"loss": 0.01799582503736019,
"grad_norm": 0.15540874004364014,
"learning_rate": 0.00014278837649416544,
"epoch": 1.1110340041637752,
"step": 101
},
{
"loss": 0.005631173960864544,
"grad_norm": 0.19040288031101227,
"learning_rate": 0.0001417338454481818,
"epoch": 1.1221374045801527,
"step": 102
},
{
"loss": 0.006984387058764696,
"grad_norm": 0.09872876852750778,
"learning_rate": 0.00014067366430758004,
"epoch": 1.1332408049965301,
"step": 103
},
{
"loss": 0.0009191720164380968,
"grad_norm": 0.015590984374284744,
"learning_rate": 0.0001396079766039157,
"epoch": 1.1443442054129078,
"step": 104
},
{
"loss": 0.0014551086351275444,
"grad_norm": 0.05302764102816582,
"learning_rate": 0.00013853692661424484,
"epoch": 1.1554476058292853,
"step": 105
},
{
"loss": 0.0014657770516350865,
"grad_norm": 0.03392955660820007,
"learning_rate": 0.00013746065934159123,
"epoch": 1.1665510062456628,
"step": 106
},
{
"loss": 0.0008271295810118318,
"grad_norm": 0.039895109832286835,
"learning_rate": 0.00013637932049531516,
"epoch": 1.1776544066620402,
"step": 107
},
{
"loss": 0.0202441718429327,
"grad_norm": 0.3854893147945404,
"learning_rate": 0.00013529305647138687,
"epoch": 1.1887578070784177,
"step": 108
},
{
"loss": 0.0037672286853194237,
"grad_norm": 0.09699325263500214,
"learning_rate": 0.00013420201433256689,
"epoch": 1.1998612074947952,
"step": 109
},
{
"loss": 0.0019683365244418383,
"grad_norm": 0.038851648569107056,
"learning_rate": 0.0001331063417884958,
"epoch": 1.2109646079111729,
"step": 110
},
{
"loss": 0.003456352511420846,
"grad_norm": 0.059991780668497086,
"learning_rate": 0.00013200618717569714,
"epoch": 1.2220680083275504,
"step": 111
},
{
"loss": 0.010525468736886978,
"grad_norm": 0.0657477006316185,
"learning_rate": 0.00013090169943749476,
"epoch": 1.2331714087439278,
"step": 112
},
{
"loss": 0.0004884903901256621,
"grad_norm": 0.011057283729314804,
"learning_rate": 0.0001297930281038482,
"epoch": 1.2442748091603053,
"step": 113
},
{
"loss": 0.001567146391607821,
"grad_norm": 0.02416708879172802,
"learning_rate": 0.00012868032327110904,
"epoch": 1.2553782095766828,
"step": 114
},
{
"loss": 0.0027189922984689474,
"grad_norm": 0.027829235419631004,
"learning_rate": 0.0001275637355816999,
"epoch": 1.2664816099930603,
"step": 115
},
{
"loss": 0.0011546590831130743,
"grad_norm": 0.020313648506999016,
"learning_rate": 0.00012644341620372023,
"epoch": 1.277585010409438,
"step": 116
},
{
"loss": 0.008269840851426125,
"grad_norm": 0.06646428257226944,
"learning_rate": 0.0001253195168104802,
"epoch": 1.2886884108258154,
"step": 117
},
{
"loss": 0.006803714670240879,
"grad_norm": 0.04964790865778923,
"learning_rate": 0.00012419218955996676,
"epoch": 1.299791811242193,
"step": 118
},
{
"loss": 0.000572250341065228,
"grad_norm": 0.012129321694374084,
"learning_rate": 0.00012306158707424403,
"epoch": 1.3108952116585704,
"step": 119
},
{
"loss": 0.0030361602548509836,
"grad_norm": 0.0791798084974289,
"learning_rate": 0.00012192786241879033,
"epoch": 1.3219986120749478,
"step": 120
},
{
"loss": 0.0032888527493923903,
"grad_norm": 0.04054338485002518,
"learning_rate": 0.00012079116908177593,
"epoch": 1.3331020124913255,
"step": 121
},
{
"loss": 0.004093429539352655,
"grad_norm": 0.044847872108221054,
"learning_rate": 0.00011965166095328301,
"epoch": 1.344205412907703,
"step": 122
},
{
"loss": 0.0018080237787216902,
"grad_norm": 0.030936224386096,
"learning_rate": 0.00011850949230447145,
"epoch": 1.3553088133240805,
"step": 123
},
{
"loss": 0.0028136586770415306,
"grad_norm": 0.036087218672037125,
"learning_rate": 0.00011736481776669306,
"epoch": 1.366412213740458,
"step": 124
},
{
"loss": 0.00630409549921751,
"grad_norm": 0.09387701749801636,
"learning_rate": 0.00011621779231055676,
"epoch": 1.3775156141568354,
"step": 125
},
{
"loss": 0.0028767576441168785,
"grad_norm": 0.10042953491210938,
"learning_rate": 0.00011506857122494831,
"epoch": 1.3886190145732131,
"step": 126
},
{
"loss": 0.0023748106323182583,
"grad_norm": 0.031763847917318344,
"learning_rate": 0.00011391731009600654,
"epoch": 1.3997224149895906,
"step": 127
},
{
"loss": 0.0006115291034802794,
"grad_norm": 0.013790605589747429,
"learning_rate": 0.00011276416478605949,
"epoch": 1.410825815405968,
"step": 128
},
{
"loss": 0.0006597705651074648,
"grad_norm": 0.023788312450051308,
"learning_rate": 0.00011160929141252303,
"epoch": 1.4219292158223455,
"step": 129
},
{
"loss": 0.0027555343694984913,
"grad_norm": 0.049149271100759506,
"learning_rate": 0.00011045284632676536,
"epoch": 1.433032616238723,
"step": 130
},
{
"loss": 0.0011079043615609407,
"grad_norm": 0.017563462257385254,
"learning_rate": 0.00010929498609293924,
"epoch": 1.4441360166551007,
"step": 131
},
{
"loss": 0.00791978556662798,
"grad_norm": 0.13089224696159363,
"learning_rate": 0.00010813586746678583,
"epoch": 1.4552394170714782,
"step": 132
},
{
"loss": 0.0012466092593967915,
"grad_norm": 0.03258095309138298,
"learning_rate": 0.00010697564737441252,
"epoch": 1.4663428174878557,
"step": 133
},
{
"loss": 0.0012967974180355668,
"grad_norm": 0.03307458758354187,
"learning_rate": 0.00010581448289104758,
"epoch": 1.4774462179042331,
"step": 134
},
{
"loss": 0.007309555076062679,
"grad_norm": 0.08119885623455048,
"learning_rate": 0.0001046525312197747,
"epoch": 1.4885496183206106,
"step": 135
},
{
"loss": 0.0024388134479522705,
"grad_norm": 0.0365670882165432,
"learning_rate": 0.00010348994967025012,
"epoch": 1.4996530187369883,
"step": 136
},
{
"loss": 0.009372331202030182,
"grad_norm": 0.11805802583694458,
"learning_rate": 0.00010232689563740563,
"epoch": 1.5107564191533656,
"step": 137
},
{
"loss": 0.0108938692137599,
"grad_norm": 0.08489461988210678,
"learning_rate": 0.00010116352658013973,
"epoch": 1.5218598195697433,
"step": 138
},
{
"loss": 0.0026753286365419626,
"grad_norm": 0.057914867997169495,
"learning_rate": 0.0001,
"epoch": 1.5329632199861207,
"step": 139
},
{
"loss": 0.004938989877700806,
"grad_norm": 0.0661998838186264,
"learning_rate": 9.883647341986032e-05,
"epoch": 1.5440666204024982,
"step": 140
},
{
"loss": 0.0006187134422361851,
"grad_norm": 0.012344218790531158,
"learning_rate": 9.767310436259438e-05,
"epoch": 1.555170020818876,
"step": 141
},
{
"loss": 0.0019832479301840067,
"grad_norm": 0.07194311171770096,
"learning_rate": 9.651005032974994e-05,
"epoch": 1.5662734212352531,
"step": 142
},
{
"loss": 0.023575376719236374,
"grad_norm": 0.10246355086565018,
"learning_rate": 9.534746878022534e-05,
"epoch": 1.5773768216516308,
"step": 143
},
{
"loss": 0.0016534080496057868,
"grad_norm": 0.027751371264457703,
"learning_rate": 9.418551710895243e-05,
"epoch": 1.5884802220680083,
"step": 144
},
{
"loss": 0.010032778605818748,
"grad_norm": 0.08701073378324509,
"learning_rate": 9.302435262558747e-05,
"epoch": 1.5995836224843858,
"step": 145
},
{
"loss": 0.004825797863304615,
"grad_norm": 0.07686074078083038,
"learning_rate": 9.186413253321418e-05,
"epoch": 1.6106870229007635,
"step": 146
},
{
"loss": 0.007475260179489851,
"grad_norm": 0.0755486860871315,
"learning_rate": 9.070501390706079e-05,
"epoch": 1.6217904233171407,
"step": 147
},
{
"loss": 0.0029515489004552364,
"grad_norm": 0.04138359799981117,
"learning_rate": 8.954715367323468e-05,
"epoch": 1.6328938237335184,
"step": 148
},
{
"loss": 0.02156146615743637,
"grad_norm": 0.14557036757469177,
"learning_rate": 8.839070858747697e-05,
"epoch": 1.643997224149896,
"step": 149
},
{
"loss": 0.0013141734525561333,
"grad_norm": 0.033466584980487823,
"learning_rate": 8.723583521394054e-05,
"epoch": 1.6551006245662734,
"step": 150
},
{
"loss": 0.009912682697176933,
"grad_norm": 0.10021013021469116,
"learning_rate": 8.608268990399349e-05,
"epoch": 1.666204024982651,
"step": 151
},
{
"loss": 0.002557775005698204,
"grad_norm": 0.06423213332891464,
"learning_rate": 8.49314287750517e-05,
"epoch": 1.6773074253990283,
"step": 152
},
{
"loss": 0.007192865014076233,
"grad_norm": 0.05610248073935509,
"learning_rate": 8.378220768944327e-05,
"epoch": 1.688410825815406,
"step": 153
},
{
"loss": 0.00357620557770133,
"grad_norm": 0.0698416456580162,
"learning_rate": 8.263518223330697e-05,
"epoch": 1.6995142262317835,
"step": 154
},
{
"loss": 0.001066345488652587,
"grad_norm": 0.022768152877688408,
"learning_rate": 8.149050769552856e-05,
"epoch": 1.710617626648161,
"step": 155
},
{
"loss": 0.002087992150336504,
"grad_norm": 0.05063783749938011,
"learning_rate": 8.034833904671698e-05,
"epoch": 1.7217210270645387,
"step": 156
},
{
"loss": 0.005414406768977642,
"grad_norm": 0.07123545557260513,
"learning_rate": 7.920883091822408e-05,
"epoch": 1.732824427480916,
"step": 157
},
{
"loss": 0.0005238516023382545,
"grad_norm": 0.013095811009407043,
"learning_rate": 7.807213758120966e-05,
"epoch": 1.7439278278972936,
"step": 158
},
{
"loss": 0.0023119491524994373,
"grad_norm": 0.02543606422841549,
"learning_rate": 7.693841292575598e-05,
"epoch": 1.755031228313671,
"step": 159
},
{
"loss": 0.0012464482570067048,
"grad_norm": 0.023485898971557617,
"learning_rate": 7.580781044003324e-05,
"epoch": 1.7661346287300486,
"step": 160
},
{
"loss": 0.010121936909854412,
"grad_norm": 0.11865287274122238,
"learning_rate": 7.468048318951983e-05,
"epoch": 1.7772380291464263,
"step": 161
},
{
"loss": 0.010622553527355194,
"grad_norm": 0.10067789256572723,
"learning_rate": 7.35565837962798e-05,
"epoch": 1.7883414295628035,
"step": 162
},
{
"loss": 0.01093344110995531,
"grad_norm": 0.11214911192655563,
"learning_rate": 7.243626441830009e-05,
"epoch": 1.7994448299791812,
"step": 163
},
{
"loss": 0.002597471931949258,
"grad_norm": 0.03298800811171532,
"learning_rate": 7.131967672889101e-05,
"epoch": 1.8105482303955587,
"step": 164
},
{
"loss": 0.0004867906682193279,
"grad_norm": 0.009381928481161594,
"learning_rate": 7.02069718961518e-05,
"epoch": 1.8216516308119362,
"step": 165
},
{
"loss": 0.0009496217826381326,
"grad_norm": 0.03668565675616264,
"learning_rate": 6.909830056250527e-05,
"epoch": 1.8327550312283136,
"step": 166
},
{
"loss": 0.005561809055507183,
"grad_norm": 0.06403116881847382,
"learning_rate": 6.799381282430284e-05,
"epoch": 1.843858431644691,
"step": 167
},
{
"loss": 0.00022523854568134993,
"grad_norm": 0.004085203632712364,
"learning_rate": 6.68936582115042e-05,
"epoch": 1.8549618320610688,
"step": 168
},
{
"loss": 0.0006040821899659932,
"grad_norm": 0.016563741490244865,
"learning_rate": 6.579798566743314e-05,
"epoch": 1.8660652324774463,
"step": 169
},
{
"loss": 0.006220725830644369,
"grad_norm": 0.09190915524959564,
"learning_rate": 6.470694352861312e-05,
"epoch": 1.8771686328938237,
"step": 170
},
{
"loss": 0.0005163773312233388,
"grad_norm": 0.012748060747981071,
"learning_rate": 6.362067950468489e-05,
"epoch": 1.8882720333102012,
"step": 171
},
{
"loss": 0.009463150054216385,
"grad_norm": 0.08283476531505585,
"learning_rate": 6.25393406584088e-05,
"epoch": 1.8993754337265787,
"step": 172
},
{
"loss": 0.005616231355816126,
"grad_norm": 0.10440317541360855,
"learning_rate": 6.146307338575519e-05,
"epoch": 1.9104788341429564,
"step": 173
},
{
"loss": 0.0025631391908973455,
"grad_norm": 0.03647289052605629,
"learning_rate": 6.039202339608432e-05,
"epoch": 1.9215822345593336,
"step": 174
},
{
"loss": 0.0007256513345055282,
"grad_norm": 0.018309885635972023,
"learning_rate": 5.9326335692419995e-05,
"epoch": 1.9326856349757113,
"step": 175
},
{
"loss": 0.006048131268471479,
"grad_norm": 0.08872969448566437,
"learning_rate": 5.8266154551818216e-05,
"epoch": 1.9437890353920888,
"step": 176
},
{
"loss": 0.0048399316146969795,
"grad_norm": 0.0693540945649147,
"learning_rate": 5.72116235058346e-05,
"epoch": 1.9548924358084663,
"step": 177
},
{
"loss": 0.00079789396841079,
"grad_norm": 0.022740449756383896,
"learning_rate": 5.616288532109225e-05,
"epoch": 1.965995836224844,
"step": 178
},
{
"loss": 0.0002554625098127872,
"grad_norm": 0.004780248738825321,
"learning_rate": 5.5120081979953785e-05,
"epoch": 1.9770992366412212,
"step": 179
},
{
"loss": 0.004320482723414898,
"grad_norm": 0.051174018532037735,
"learning_rate": 5.4083354661298814e-05,
"epoch": 1.988202637057599,
"step": 180
},
{
"loss": 0.00403692526742816,
"grad_norm": 0.04460834339261055,
"learning_rate": 5.305284372141095e-05,
"epoch": 1.9993060374739764,
"step": 181
},
{
"loss": 0.000192473380593583,
"grad_norm": 0.02723965048789978,
"learning_rate": 5.2028688674975415e-05,
"epoch": 2.0,
"step": 182
},
{
"loss": 0.0031282880809158087,
"grad_norm": 0.07167559117078781,
"learning_rate": 5.101102817619131e-05,
"epoch": 2.0111034004163777,
"step": 183
},
{
"loss": 0.001440710504539311,
"grad_norm": 0.034675925970077515,
"learning_rate": 5.000000000000002e-05,
"epoch": 2.022206800832755,
"step": 184
},
{
"loss": 0.002487765159457922,
"grad_norm": 0.04422704502940178,
"learning_rate": 4.899574102343247e-05,
"epoch": 2.0333102012491326,
"step": 185
},
{
"loss": 0.00018012276268564165,
"grad_norm": 0.003625614335760474,
"learning_rate": 4.799838720707846e-05,
"epoch": 2.04441360166551,
"step": 186
},
{
"loss": 0.0008509792387485504,
"grad_norm": 0.0641360729932785,
"learning_rate": 4.700807357667952e-05,
"epoch": 2.0555170020818876,
"step": 187
},
{
"loss": 0.0012335458304733038,
"grad_norm": 0.03804023936390877,
"learning_rate": 4.6024934204848745e-05,
"epoch": 2.0666204024982653,
"step": 188
},
{
"loss": 0.0007197751547209918,
"grad_norm": 0.01487179659307003,
"learning_rate": 4.50491021929194e-05,
"epoch": 2.0777238029146425,
"step": 189
},
{
"loss": 0.000570658827200532,
"grad_norm": 0.012829587794840336,
"learning_rate": 4.4080709652925336e-05,
"epoch": 2.0888272033310202,
"step": 190
},
{
"loss": 0.0003685185220092535,
"grad_norm": 0.00644276337698102,
"learning_rate": 4.3119887689714844e-05,
"epoch": 2.0999306037473975,
"step": 191
},
{
"loss": 0.00217076507396996,
"grad_norm": 0.03593655303120613,
"learning_rate": 4.216676638320135e-05,
"epoch": 2.111034004163775,
"step": 192
},
{
"loss": 0.0021631934214383364,
"grad_norm": 0.038568392395973206,
"learning_rate": 4.12214747707527e-05,
"epoch": 2.122137404580153,
"step": 193
},
{
"loss": 0.00018508080393075943,
"grad_norm": 0.004468716215342283,
"learning_rate": 4.028414082972141e-05,
"epoch": 2.13324080499653,
"step": 194
},
{
"loss": 0.0009128937381319702,
"grad_norm": 0.027085695415735245,
"learning_rate": 3.935489146011869e-05,
"epoch": 2.144344205412908,
"step": 195
},
{
"loss": 0.0002966481843031943,
"grad_norm": 0.006275610998272896,
"learning_rate": 3.843385246743417e-05,
"epoch": 2.155447605829285,
"step": 196
},
{
"loss": 0.0011582208098843694,
"grad_norm": 0.02687346190214157,
"learning_rate": 3.7521148545604e-05,
"epoch": 2.1665510062456628,
"step": 197
},
{
"loss": 0.0008262763731181622,
"grad_norm": 0.014363622292876244,
"learning_rate": 3.661690326012897e-05,
"epoch": 2.1776544066620405,
"step": 198
},
{
"loss": 0.0003042164898943156,
"grad_norm": 0.01361521054059267,
"learning_rate": 3.5721239031346066e-05,
"epoch": 2.1887578070784177,
"step": 199
},
{
"loss": 0.00021733059838879853,
"grad_norm": 0.00866472814232111,
"learning_rate": 3.483427711785449e-05,
"epoch": 2.1998612074947954,
"step": 200
},
{
"loss": 0.00023889444128144532,
"grad_norm": 0.00470183277502656,
"learning_rate": 3.395613760009925e-05,
"epoch": 2.2109646079111727,
"step": 201
},
{
"loss": 0.0022749893832951784,
"grad_norm": 0.03821490705013275,
"learning_rate": 3.308693936411421e-05,
"epoch": 2.2220680083275504,
"step": 202
},
{
"loss": 0.00029496397473849356,
"grad_norm": 0.007405790500342846,
"learning_rate": 3.222680008542678e-05,
"epoch": 2.233171408743928,
"step": 203
},
{
"loss": 0.00033837920636869967,
"grad_norm": 0.008731233887374401,
"learning_rate": 3.137583621312665e-05,
"epoch": 2.2442748091603053,
"step": 204
},
{
"loss": 0.0007719028508290648,
"grad_norm": 0.014901618473231792,
"learning_rate": 3.053416295410026e-05,
"epoch": 2.255378209576683,
"step": 205
},
{
"loss": 0.0014813048765063286,
"grad_norm": 0.04358435794711113,
"learning_rate": 2.9701894257433826e-05,
"epoch": 2.2664816099930603,
"step": 206
},
{
"loss": 0.0005430534947663546,
"grad_norm": 0.011528179049491882,
"learning_rate": 2.8879142798986292e-05,
"epoch": 2.277585010409438,
"step": 207
},
{
"loss": 0.00034370424691587687,
"grad_norm": 0.004893212113529444,
"learning_rate": 2.8066019966134904e-05,
"epoch": 2.2886884108258156,
"step": 208
},
{
"loss": 0.0010527092963457108,
"grad_norm": 0.0205328818410635,
"learning_rate": 2.7262635842695127e-05,
"epoch": 2.299791811242193,
"step": 209
},
{
"loss": 0.0008116448880173266,
"grad_norm": 0.021442309021949768,
"learning_rate": 2.6469099194017143e-05,
"epoch": 2.3108952116585706,
"step": 210
},
{
"loss": 0.0037238351069390774,
"grad_norm": 0.04724681004881859,
"learning_rate": 2.5685517452260567e-05,
"epoch": 2.321998612074948,
"step": 211
},
{
"loss": 0.00570558849722147,
"grad_norm": 0.06557858735322952,
"learning_rate": 2.491199670185008e-05,
"epoch": 2.3331020124913255,
"step": 212
},
{
"loss": 0.0011785834794864058,
"grad_norm": 0.02593190036714077,
"learning_rate": 2.4148641665113113e-05,
"epoch": 2.344205412907703,
"step": 213
},
{
"loss": 0.0018704122630879283,
"grad_norm": 0.05805144086480141,
"learning_rate": 2.339555568810221e-05,
"epoch": 2.3553088133240805,
"step": 214
},
{
"loss": 0.0009465268230997026,
"grad_norm": 0.029634978622198105,
"learning_rate": 2.265284072660362e-05,
"epoch": 2.366412213740458,
"step": 215
},
{
"loss": 0.0006902794120833278,
"grad_norm": 0.017646944150328636,
"learning_rate": 2.192059733233408e-05,
"epoch": 2.3775156141568354,
"step": 216
},
{
"loss": 0.0008653043187223375,
"grad_norm": 0.02128649689257145,
"learning_rate": 2.119892463932781e-05,
"epoch": 2.388619014573213,
"step": 217
},
{
"loss": 0.0009620467899367213,
"grad_norm": 0.01664072461426258,
"learning_rate": 2.0487920350515212e-05,
"epoch": 2.3997224149895904,
"step": 218
},
{
"loss": 0.00021245694370009005,
"grad_norm": 0.004011219833046198,
"learning_rate": 1.9787680724495617e-05,
"epoch": 2.410825815405968,
"step": 219
},
{
"loss": 0.000662539794575423,
"grad_norm": 0.013697715476155281,
"learning_rate": 1.9098300562505266e-05,
"epoch": 2.4219292158223458,
"step": 220
},
{
"loss": 0.0013410050887614489,
"grad_norm": 0.03492460772395134,
"learning_rate": 1.8419873195582814e-05,
"epoch": 2.433032616238723,
"step": 221
},
{
"loss": 0.0006607146933674812,
"grad_norm": 0.016314877197146416,
"learning_rate": 1.775249047193377e-05,
"epoch": 2.4441360166551007,
"step": 222
},
{
"loss": 0.00267712096683681,
"grad_norm": 0.03245789557695389,
"learning_rate": 1.7096242744495837e-05,
"epoch": 2.455239417071478,
"step": 223
},
{
"loss": 0.0009540443425066769,
"grad_norm": 0.01580522023141384,
"learning_rate": 1.6451218858706374e-05,
"epoch": 2.4663428174878557,
"step": 224
},
{
"loss": 0.0002631157694850117,
"grad_norm": 0.006987827364355326,
"learning_rate": 1.5817506140474247e-05,
"epoch": 2.4774462179042334,
"step": 225
},
{
"loss": 0.0005279219476506114,
"grad_norm": 0.011882264167070389,
"learning_rate": 1.5195190384357404e-05,
"epoch": 2.4885496183206106,
"step": 226
},
{
"loss": 0.0005659055314026773,
"grad_norm": 0.01122165285050869,
"learning_rate": 1.458435584194745e-05,
"epoch": 2.4996530187369883,
"step": 227
},
{
"loss": 0.0006534359999932349,
"grad_norm": 0.015417223796248436,
"learning_rate": 1.3985085210463477e-05,
"epoch": 2.5107564191533656,
"step": 228
},
{
"loss": 0.000537577667273581,
"grad_norm": 0.014984376728534698,
"learning_rate": 1.339745962155613e-05,
"epoch": 2.5218598195697433,
"step": 229
},
{
"loss": 0.0014547649770975113,
"grad_norm": 0.04610738530755043,
"learning_rate": 1.2821558630323772e-05,
"epoch": 2.5329632199861205,
"step": 230
},
{
"loss": 0.0030281252693384886,
"grad_norm": 0.06278136372566223,
"learning_rate": 1.2257460204541794e-05,
"epoch": 2.544066620402498,
"step": 231
},
{
"loss": 4.744660691358149e-05,
"grad_norm": 0.0010382796172052622,
"learning_rate": 1.1705240714107302e-05,
"epoch": 2.555170020818876,
"step": 232
},
{
"loss": 0.0011553519871085882,
"grad_norm": 0.03249562904238701,
"learning_rate": 1.116497492069961e-05,
"epoch": 2.566273421235253,
"step": 233
},
{
"loss": 0.0005710968980565667,
"grad_norm": 0.040762804448604584,
"learning_rate": 1.0636735967658784e-05,
"epoch": 2.577376821651631,
"step": 234
},
{
"loss": 0.0016874882858246565,
"grad_norm": 0.03226013481616974,
"learning_rate": 1.0120595370083318e-05,
"epoch": 2.588480222068008,
"step": 235
},
{
"loss": 0.0004933075397275388,
"grad_norm": 0.008815680630505085,
"learning_rate": 9.616623005147951e-06,
"epoch": 2.599583622484386,
"step": 236
},
{
"loss": 0.0004388394590932876,
"grad_norm": 0.008908948861062527,
"learning_rate": 9.124887102643575e-06,
"epoch": 2.6106870229007635,
"step": 237
},
{
"loss": 0.00035047775600105524,
"grad_norm": 0.008026096038520336,
"learning_rate": 8.645454235739903e-06,
"epoch": 2.6217904233171407,
"step": 238
},
{
"loss": 0.004668767563998699,
"grad_norm": 0.09909072518348694,
"learning_rate": 8.178389311972612e-06,
"epoch": 2.6328938237335184,
"step": 239
},
{
"loss": 0.0018479757709428668,
"grad_norm": 0.03939471393823624,
"learning_rate": 7.72375556445577e-06,
"epoch": 2.6439972241498957,
"step": 240
},
{
"loss": 0.001170073519460857,
"grad_norm": 0.07461482286453247,
"learning_rate": 7.281614543321269e-06,
"epoch": 2.6551006245662734,
"step": 241
},
{
"loss": 0.001040186150930822,
"grad_norm": 0.018851246684789658,
"learning_rate": 6.852026107385756e-06,
"epoch": 2.666204024982651,
"step": 242
},
{
"loss": 0.0003154140431433916,
"grad_norm": 0.011888038367033005,
"learning_rate": 6.435048416046863e-06,
"epoch": 2.6773074253990283,
"step": 243
},
{
"loss": 0.002783727366477251,
"grad_norm": 0.06753791868686676,
"learning_rate": 6.030737921409169e-06,
"epoch": 2.688410825815406,
"step": 244
},
{
"loss": 8.915372018236667e-05,
"grad_norm": 0.0017917046789079905,
"learning_rate": 5.639149360641649e-06,
"epoch": 2.6995142262317833,
"step": 245
},
{
"loss": 0.001064301235601306,
"grad_norm": 0.025877000764012337,
"learning_rate": 5.26033574856708e-06,
"epoch": 2.710617626648161,
"step": 246
},
{
"loss": 0.00019584805704653263,
"grad_norm": 0.013628424145281315,
"learning_rate": 4.8943483704846475e-06,
"epoch": 2.7217210270645387,
"step": 247
},
{
"loss": 0.0002667722583282739,
"grad_norm": 0.009531455114483833,
"learning_rate": 4.541236775226809e-06,
"epoch": 2.732824427480916,
"step": 248
},
{
"loss": 0.0014836416812613606,
"grad_norm": 0.03923977538943291,
"learning_rate": 4.20104876845111e-06,
"epoch": 2.7439278278972936,
"step": 249
},
{
"loss": 0.0010388277005404234,
"grad_norm": 0.029680265113711357,
"learning_rate": 3.873830406168111e-06,
"epoch": 2.755031228313671,
"step": 250
},
{
"loss": 0.000178680318640545,
"grad_norm": 0.00518755754455924,
"learning_rate": 3.5596259885061102e-06,
"epoch": 2.7661346287300486,
"step": 251
},
{
"loss": 0.0009329643216915429,
"grad_norm": 0.024954449385404587,
"learning_rate": 3.2584780537136207e-06,
"epoch": 2.7772380291464263,
"step": 252
},
{
"loss": 0.0006331746117211878,
"grad_norm": 0.012082848697900772,
"learning_rate": 2.970427372400353e-06,
"epoch": 2.7883414295628035,
"step": 253
},
{
"loss": 0.00015647141844965518,
"grad_norm": 0.004532885737717152,
"learning_rate": 2.6955129420176196e-06,
"epoch": 2.799444829979181,
"step": 254
},
{
"loss": 0.0001516131014795974,
"grad_norm": 0.0039943912997841835,
"learning_rate": 2.433771981578581e-06,
"epoch": 2.8105482303955585,
"step": 255
},
{
"loss": 0.0005077925743535161,
"grad_norm": 0.014085155911743641,
"learning_rate": 2.1852399266194314e-06,
"epoch": 2.821651630811936,
"step": 256
},
{
"loss": 0.0010776021517813206,
"grad_norm": 0.03057239204645157,
"learning_rate": 1.9499504244020693e-06,
"epoch": 2.832755031228314,
"step": 257
},
{
"loss": 0.001233020331710577,
"grad_norm": 0.0293893925845623,
"learning_rate": 1.7279353293586765e-06,
"epoch": 2.843858431644691,
"step": 258
},
{
"loss": 0.00045728698023594916,
"grad_norm": 0.012387805618345737,
"learning_rate": 1.5192246987791981e-06,
"epoch": 2.854961832061069,
"step": 259
},
{
"loss": 0.006937120109796524,
"grad_norm": 0.051831986755132675,
"learning_rate": 1.323846788742078e-06,
"epoch": 2.866065232477446,
"step": 260
},
{
"loss": 0.0007000218611210585,
"grad_norm": 0.02848062850534916,
"learning_rate": 1.14182805028884e-06,
"epoch": 2.8771686328938237,
"step": 261
},
{
"loss": 0.003899730509147048,
"grad_norm": 0.06617487221956253,
"learning_rate": 9.731931258429638e-07,
"epoch": 2.8882720333102014,
"step": 262
},
{
"loss": 0.002479749033227563,
"grad_norm": 0.042154788970947266,
"learning_rate": 8.17964845873831e-07,
"epoch": 2.8993754337265787,
"step": 263
},
{
"loss": 0.0009774853242561221,
"grad_norm": 0.029534008353948593,
"learning_rate": 6.761642258056978e-07,
"epoch": 2.9104788341429564,
"step": 264
},
{
"loss": 0.0017800165805965662,
"grad_norm": 0.04713388904929161,
"learning_rate": 5.478104631726711e-07,
"epoch": 2.9215822345593336,
"step": 265
},
{
"loss": 0.00025485060177743435,
"grad_norm": 0.0057479869574308395,
"learning_rate": 4.329209350195651e-07,
"epoch": 2.9326856349757113,
"step": 266
},
{
"loss": 0.003669393016025424,
"grad_norm": 0.07603989541530609,
"learning_rate": 3.315111955493944e-07,
"epoch": 2.943789035392089,
"step": 267
},
{
"loss": 0.0010033966973423958,
"grad_norm": 0.026915691792964935,
"learning_rate": 2.4359497401758024e-07,
"epoch": 2.9548924358084663,
"step": 268
},
{
"loss": 0.0012706794077530503,
"grad_norm": 0.023083670064806938,
"learning_rate": 1.6918417287318245e-07,
"epoch": 2.965995836224844,
"step": 269
},
{
"loss": 0.0004130644374527037,
"grad_norm": 0.00994869228452444,
"learning_rate": 1.0828886614754341e-07,
"epoch": 2.9770992366412212,
"step": 270
},
{
"loss": 0.0012230968568474054,
"grad_norm": 0.026720473542809486,
"learning_rate": 6.09172980904238e-08,
"epoch": 2.988202637057599,
"step": 271
},
{
"loss": 0.000913719180971384,
"grad_norm": 0.01431743148714304,
"learning_rate": 2.7075882053828605e-08,
"epoch": 2.9993060374739766,
"step": 272
},
{
"loss": 0.01702476479113102,
"grad_norm": 0.5101261734962463,
"learning_rate": 6.769199623779532e-09,
"epoch": 3.0,
"step": 273
},
{
"train_runtime": 9101.4738,
"train_samples_per_second": 0.475,
"train_steps_per_second": 0.03,
"total_flos": 5.568150387086623e+17,
"train_loss": 0.01574120909053851,
"epoch": 3.0,
"step": 273
}
]
}