Slopus-9B / adapter_phase2 /trainer_state.json
Kukedlc's picture
Add adapter_phase2/trainer_state.json (ckpt-758 final phase 2)
906797a verified
Raw
History Blame Contribute Delete
141 kB
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"eval_steps": 500,
"global_step": 758,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.002638522427440633,
"grad_norm": 0.3394036889076233,
"learning_rate": 0.0,
"loss": 1.0827826261520386,
"step": 1
},
{
"epoch": 0.005277044854881266,
"grad_norm": 0.2860078513622284,
"learning_rate": 5e-06,
"loss": 1.102963924407959,
"step": 2
},
{
"epoch": 0.0079155672823219,
"grad_norm": 0.3322666585445404,
"learning_rate": 1e-05,
"loss": 0.9182360172271729,
"step": 3
},
{
"epoch": 0.010554089709762533,
"grad_norm": 0.29622846841812134,
"learning_rate": 1.5e-05,
"loss": 0.9905988574028015,
"step": 4
},
{
"epoch": 0.013192612137203167,
"grad_norm": 0.26111382246017456,
"learning_rate": 2e-05,
"loss": 1.096787691116333,
"step": 5
},
{
"epoch": 0.0158311345646438,
"grad_norm": 0.2614597976207733,
"learning_rate": 2.5e-05,
"loss": 1.105420708656311,
"step": 6
},
{
"epoch": 0.018469656992084433,
"grad_norm": 0.2569218873977661,
"learning_rate": 3e-05,
"loss": 0.9514756202697754,
"step": 7
},
{
"epoch": 0.021108179419525065,
"grad_norm": 0.24397169053554535,
"learning_rate": 3.5e-05,
"loss": 0.9208998680114746,
"step": 8
},
{
"epoch": 0.023746701846965697,
"grad_norm": 0.30534666776657104,
"learning_rate": 4e-05,
"loss": 1.0918810367584229,
"step": 9
},
{
"epoch": 0.026385224274406333,
"grad_norm": 0.2684849202632904,
"learning_rate": 4.5e-05,
"loss": 1.0292065143585205,
"step": 10
},
{
"epoch": 0.029023746701846966,
"grad_norm": 0.26211386919021606,
"learning_rate": 5e-05,
"loss": 1.0240130424499512,
"step": 11
},
{
"epoch": 0.0316622691292876,
"grad_norm": 0.2565249502658844,
"learning_rate": 5.500000000000001e-05,
"loss": 0.9769070148468018,
"step": 12
},
{
"epoch": 0.03430079155672823,
"grad_norm": 0.26444166898727417,
"learning_rate": 6e-05,
"loss": 1.0853519439697266,
"step": 13
},
{
"epoch": 0.036939313984168866,
"grad_norm": 0.2618539035320282,
"learning_rate": 6.500000000000001e-05,
"loss": 1.0640870332717896,
"step": 14
},
{
"epoch": 0.0395778364116095,
"grad_norm": 0.2667272388935089,
"learning_rate": 7e-05,
"loss": 0.998728334903717,
"step": 15
},
{
"epoch": 0.04221635883905013,
"grad_norm": 0.2519418001174927,
"learning_rate": 7.500000000000001e-05,
"loss": 0.9445884227752686,
"step": 16
},
{
"epoch": 0.044854881266490766,
"grad_norm": 0.23243127763271332,
"learning_rate": 8e-05,
"loss": 0.9943759441375732,
"step": 17
},
{
"epoch": 0.047493403693931395,
"grad_norm": 0.24636130034923553,
"learning_rate": 8.5e-05,
"loss": 1.087374210357666,
"step": 18
},
{
"epoch": 0.05013192612137203,
"grad_norm": 0.2173345983028412,
"learning_rate": 9e-05,
"loss": 1.0086473226547241,
"step": 19
},
{
"epoch": 0.052770448548812667,
"grad_norm": 0.20626968145370483,
"learning_rate": 9.5e-05,
"loss": 0.8200538158416748,
"step": 20
},
{
"epoch": 0.055408970976253295,
"grad_norm": 0.23201459646224976,
"learning_rate": 0.0001,
"loss": 0.9757561683654785,
"step": 21
},
{
"epoch": 0.05804749340369393,
"grad_norm": 0.20844975113868713,
"learning_rate": 9.999954697061305e-05,
"loss": 0.9280185103416443,
"step": 22
},
{
"epoch": 0.06068601583113457,
"grad_norm": 0.20567265152931213,
"learning_rate": 9.999818789066165e-05,
"loss": 1.0713424682617188,
"step": 23
},
{
"epoch": 0.0633245382585752,
"grad_norm": 0.22238284349441528,
"learning_rate": 9.999592278477388e-05,
"loss": 0.9708124995231628,
"step": 24
},
{
"epoch": 0.06596306068601583,
"grad_norm": 0.2274278849363327,
"learning_rate": 9.999275169399614e-05,
"loss": 1.0282939672470093,
"step": 25
},
{
"epoch": 0.06860158311345646,
"grad_norm": 0.2159455567598343,
"learning_rate": 9.998867467579234e-05,
"loss": 1.1293445825576782,
"step": 26
},
{
"epoch": 0.0712401055408971,
"grad_norm": 0.1889648586511612,
"learning_rate": 9.998369180404283e-05,
"loss": 0.8767735958099365,
"step": 27
},
{
"epoch": 0.07387862796833773,
"grad_norm": 0.2028406262397766,
"learning_rate": 9.997780316904311e-05,
"loss": 1.0634520053863525,
"step": 28
},
{
"epoch": 0.07651715039577836,
"grad_norm": 0.180164635181427,
"learning_rate": 9.997100887750215e-05,
"loss": 0.8971686363220215,
"step": 29
},
{
"epoch": 0.079155672823219,
"grad_norm": 0.18992403149604797,
"learning_rate": 9.99633090525405e-05,
"loss": 0.9242538809776306,
"step": 30
},
{
"epoch": 0.08179419525065963,
"grad_norm": 0.1975812315940857,
"learning_rate": 9.995470383368808e-05,
"loss": 1.0072821378707886,
"step": 31
},
{
"epoch": 0.08443271767810026,
"grad_norm": 0.21048954129219055,
"learning_rate": 9.994519337688152e-05,
"loss": 1.0022131204605103,
"step": 32
},
{
"epoch": 0.0870712401055409,
"grad_norm": 0.20481035113334656,
"learning_rate": 9.99347778544615e-05,
"loss": 1.032679796218872,
"step": 33
},
{
"epoch": 0.08970976253298153,
"grad_norm": 0.1966608613729477,
"learning_rate": 9.992345745516953e-05,
"loss": 1.0260268449783325,
"step": 34
},
{
"epoch": 0.09234828496042216,
"grad_norm": 0.19506685435771942,
"learning_rate": 9.991123238414455e-05,
"loss": 1.0653270483016968,
"step": 35
},
{
"epoch": 0.09498680738786279,
"grad_norm": 0.19783715903759003,
"learning_rate": 9.989810286291923e-05,
"loss": 1.0186381340026855,
"step": 36
},
{
"epoch": 0.09762532981530343,
"grad_norm": 0.19126851856708527,
"learning_rate": 9.98840691294159e-05,
"loss": 0.9795345067977905,
"step": 37
},
{
"epoch": 0.10026385224274406,
"grad_norm": 0.19075506925582886,
"learning_rate": 9.986913143794232e-05,
"loss": 0.9185978770256042,
"step": 38
},
{
"epoch": 0.10290237467018469,
"grad_norm": 0.20359137654304504,
"learning_rate": 9.985329005918702e-05,
"loss": 1.068305253982544,
"step": 39
},
{
"epoch": 0.10554089709762533,
"grad_norm": 0.17329703271389008,
"learning_rate": 9.983654528021442e-05,
"loss": 0.8745554685592651,
"step": 40
},
{
"epoch": 0.10817941952506596,
"grad_norm": 0.184177964925766,
"learning_rate": 9.981889740445958e-05,
"loss": 1.0511523485183716,
"step": 41
},
{
"epoch": 0.11081794195250659,
"grad_norm": 0.17558042705059052,
"learning_rate": 9.980034675172274e-05,
"loss": 1.0206934213638306,
"step": 42
},
{
"epoch": 0.11345646437994723,
"grad_norm": 0.1741347759962082,
"learning_rate": 9.978089365816357e-05,
"loss": 0.993105947971344,
"step": 43
},
{
"epoch": 0.11609498680738786,
"grad_norm": 0.18105603754520416,
"learning_rate": 9.976053847629496e-05,
"loss": 0.9645117521286011,
"step": 44
},
{
"epoch": 0.11873350923482849,
"grad_norm": 0.18442459404468536,
"learning_rate": 9.973928157497674e-05,
"loss": 1.0399397611618042,
"step": 45
},
{
"epoch": 0.12137203166226913,
"grad_norm": 0.1740163266658783,
"learning_rate": 9.971712333940895e-05,
"loss": 1.0997076034545898,
"step": 46
},
{
"epoch": 0.12401055408970976,
"grad_norm": 0.16615112125873566,
"learning_rate": 9.969406417112489e-05,
"loss": 0.9800876379013062,
"step": 47
},
{
"epoch": 0.1266490765171504,
"grad_norm": 0.17227014899253845,
"learning_rate": 9.967010448798375e-05,
"loss": 0.9780592918395996,
"step": 48
},
{
"epoch": 0.12928759894459102,
"grad_norm": 0.20567218959331512,
"learning_rate": 9.964524472416319e-05,
"loss": 1.0072335004806519,
"step": 49
},
{
"epoch": 0.13192612137203166,
"grad_norm": 0.18364901840686798,
"learning_rate": 9.961948533015133e-05,
"loss": 1.0017499923706055,
"step": 50
},
{
"epoch": 0.1345646437994723,
"grad_norm": 0.1748533993959427,
"learning_rate": 9.95928267727387e-05,
"loss": 0.9483516216278076,
"step": 51
},
{
"epoch": 0.13720316622691292,
"grad_norm": 0.17033548653125763,
"learning_rate": 9.956526953500964e-05,
"loss": 0.9018915891647339,
"step": 52
},
{
"epoch": 0.13984168865435356,
"grad_norm": 0.1777356117963791,
"learning_rate": 9.953681411633376e-05,
"loss": 0.9655129909515381,
"step": 53
},
{
"epoch": 0.1424802110817942,
"grad_norm": 0.1858673244714737,
"learning_rate": 9.950746103235663e-05,
"loss": 1.0894356966018677,
"step": 54
},
{
"epoch": 0.14511873350923482,
"grad_norm": 0.16702871024608612,
"learning_rate": 9.947721081499068e-05,
"loss": 0.9454880952835083,
"step": 55
},
{
"epoch": 0.14775725593667546,
"grad_norm": 0.16913406550884247,
"learning_rate": 9.944606401240539e-05,
"loss": 0.9937676191329956,
"step": 56
},
{
"epoch": 0.1503957783641161,
"grad_norm": 0.17091500759124756,
"learning_rate": 9.941402118901744e-05,
"loss": 0.9700629711151123,
"step": 57
},
{
"epoch": 0.15303430079155672,
"grad_norm": 0.18050940334796906,
"learning_rate": 9.938108292548044e-05,
"loss": 1.009847640991211,
"step": 58
},
{
"epoch": 0.15567282321899736,
"grad_norm": 0.1656666100025177,
"learning_rate": 9.934724981867446e-05,
"loss": 0.9251805543899536,
"step": 59
},
{
"epoch": 0.158311345646438,
"grad_norm": 0.17462797462940216,
"learning_rate": 9.931252248169518e-05,
"loss": 0.9192997813224792,
"step": 60
},
{
"epoch": 0.16094986807387862,
"grad_norm": 0.1910332590341568,
"learning_rate": 9.927690154384273e-05,
"loss": 0.9398717880249023,
"step": 61
},
{
"epoch": 0.16358839050131926,
"grad_norm": 0.17913269996643066,
"learning_rate": 9.924038765061042e-05,
"loss": 0.9283158779144287,
"step": 62
},
{
"epoch": 0.1662269129287599,
"grad_norm": 0.17269554734230042,
"learning_rate": 9.920298146367286e-05,
"loss": 0.9414621591567993,
"step": 63
},
{
"epoch": 0.16886543535620052,
"grad_norm": 0.15802136063575745,
"learning_rate": 9.916468366087417e-05,
"loss": 1.0821268558502197,
"step": 64
},
{
"epoch": 0.17150395778364116,
"grad_norm": 0.16915935277938843,
"learning_rate": 9.912549493621554e-05,
"loss": 0.909949541091919,
"step": 65
},
{
"epoch": 0.1741424802110818,
"grad_norm": 0.181443452835083,
"learning_rate": 9.908541599984276e-05,
"loss": 1.0008492469787598,
"step": 66
},
{
"epoch": 0.17678100263852242,
"grad_norm": 0.1813468486070633,
"learning_rate": 9.904444757803321e-05,
"loss": 0.9849015474319458,
"step": 67
},
{
"epoch": 0.17941952506596306,
"grad_norm": 0.18925133347511292,
"learning_rate": 9.900259041318289e-05,
"loss": 1.089128017425537,
"step": 68
},
{
"epoch": 0.1820580474934037,
"grad_norm": 0.17448389530181885,
"learning_rate": 9.895984526379281e-05,
"loss": 1.0046229362487793,
"step": 69
},
{
"epoch": 0.18469656992084432,
"grad_norm": 0.16590456664562225,
"learning_rate": 9.891621290445534e-05,
"loss": 0.9553288221359253,
"step": 70
},
{
"epoch": 0.18733509234828497,
"grad_norm": 0.18540161848068237,
"learning_rate": 9.887169412584011e-05,
"loss": 1.0598280429840088,
"step": 71
},
{
"epoch": 0.18997361477572558,
"grad_norm": 0.1836412101984024,
"learning_rate": 9.882628973467972e-05,
"loss": 1.0192971229553223,
"step": 72
},
{
"epoch": 0.19261213720316622,
"grad_norm": 0.1572759449481964,
"learning_rate": 9.878000055375512e-05,
"loss": 1.0670673847198486,
"step": 73
},
{
"epoch": 0.19525065963060687,
"grad_norm": 0.16847407817840576,
"learning_rate": 9.873282742188066e-05,
"loss": 1.0559749603271484,
"step": 74
},
{
"epoch": 0.19788918205804748,
"grad_norm": 0.18431080877780914,
"learning_rate": 9.868477119388896e-05,
"loss": 1.0605112314224243,
"step": 75
},
{
"epoch": 0.20052770448548812,
"grad_norm": 0.17507214844226837,
"learning_rate": 9.863583274061534e-05,
"loss": 0.9671310186386108,
"step": 76
},
{
"epoch": 0.20316622691292877,
"grad_norm": 0.169069305062294,
"learning_rate": 9.858601294888213e-05,
"loss": 0.9900617003440857,
"step": 77
},
{
"epoch": 0.20580474934036938,
"grad_norm": 0.1741606891155243,
"learning_rate": 9.853531272148249e-05,
"loss": 1.0460593700408936,
"step": 78
},
{
"epoch": 0.20844327176781002,
"grad_norm": 0.18498769402503967,
"learning_rate": 9.848373297716414e-05,
"loss": 1.0282113552093506,
"step": 79
},
{
"epoch": 0.21108179419525067,
"grad_norm": 0.15732872486114502,
"learning_rate": 9.84312746506127e-05,
"loss": 0.9001719951629639,
"step": 80
},
{
"epoch": 0.21372031662269128,
"grad_norm": 0.18386222422122955,
"learning_rate": 9.837793869243468e-05,
"loss": 0.9183194637298584,
"step": 81
},
{
"epoch": 0.21635883905013192,
"grad_norm": 0.17975150048732758,
"learning_rate": 9.832372606914037e-05,
"loss": 1.0881237983703613,
"step": 82
},
{
"epoch": 0.21899736147757257,
"grad_norm": 0.17720025777816772,
"learning_rate": 9.82686377631262e-05,
"loss": 0.9983437061309814,
"step": 83
},
{
"epoch": 0.22163588390501318,
"grad_norm": 0.18643102049827576,
"learning_rate": 9.821267477265705e-05,
"loss": 1.1010072231292725,
"step": 84
},
{
"epoch": 0.22427440633245382,
"grad_norm": 0.19078285992145538,
"learning_rate": 9.815583811184808e-05,
"loss": 1.091160774230957,
"step": 85
},
{
"epoch": 0.22691292875989447,
"grad_norm": 0.1700907051563263,
"learning_rate": 9.80981288106464e-05,
"loss": 0.9641087055206299,
"step": 86
},
{
"epoch": 0.22955145118733508,
"grad_norm": 0.1585739403963089,
"learning_rate": 9.803954791481239e-05,
"loss": 0.9505077600479126,
"step": 87
},
{
"epoch": 0.23218997361477572,
"grad_norm": 0.16782094538211823,
"learning_rate": 9.798009648590072e-05,
"loss": 1.0017526149749756,
"step": 88
},
{
"epoch": 0.23482849604221637,
"grad_norm": 0.17637847363948822,
"learning_rate": 9.791977560124119e-05,
"loss": 0.9929439425468445,
"step": 89
},
{
"epoch": 0.23746701846965698,
"grad_norm": 0.17673365771770477,
"learning_rate": 9.785858635391914e-05,
"loss": 1.0124274492263794,
"step": 90
},
{
"epoch": 0.24010554089709762,
"grad_norm": 0.16427874565124512,
"learning_rate": 9.779652985275562e-05,
"loss": 0.8820173144340515,
"step": 91
},
{
"epoch": 0.24274406332453827,
"grad_norm": 0.1866939812898636,
"learning_rate": 9.773360722228742e-05,
"loss": 0.9582796096801758,
"step": 92
},
{
"epoch": 0.24538258575197888,
"grad_norm": 0.16584990918636322,
"learning_rate": 9.766981960274653e-05,
"loss": 0.9931353330612183,
"step": 93
},
{
"epoch": 0.24802110817941952,
"grad_norm": 0.1827409565448761,
"learning_rate": 9.760516815003964e-05,
"loss": 1.0762715339660645,
"step": 94
},
{
"epoch": 0.25065963060686014,
"grad_norm": 0.1706336885690689,
"learning_rate": 9.753965403572703e-05,
"loss": 0.9621766805648804,
"step": 95
},
{
"epoch": 0.2532981530343008,
"grad_norm": 0.17512862384319305,
"learning_rate": 9.747327844700147e-05,
"loss": 1.0493756532669067,
"step": 96
},
{
"epoch": 0.2559366754617414,
"grad_norm": 0.1861727088689804,
"learning_rate": 9.740604258666668e-05,
"loss": 0.9710463285446167,
"step": 97
},
{
"epoch": 0.25857519788918204,
"grad_norm": 0.17469093203544617,
"learning_rate": 9.733794767311544e-05,
"loss": 1.0105197429656982,
"step": 98
},
{
"epoch": 0.2612137203166227,
"grad_norm": 0.1550728976726532,
"learning_rate": 9.726899494030768e-05,
"loss": 0.8987333178520203,
"step": 99
},
{
"epoch": 0.2638522427440633,
"grad_norm": 0.17990083992481232,
"learning_rate": 9.719918563774792e-05,
"loss": 0.8861106038093567,
"step": 100
},
{
"epoch": 0.26649076517150394,
"grad_norm": 0.15576201677322388,
"learning_rate": 9.71285210304628e-05,
"loss": 1.059525728225708,
"step": 101
},
{
"epoch": 0.2691292875989446,
"grad_norm": 0.16039283573627472,
"learning_rate": 9.705700239897809e-05,
"loss": 0.9966175556182861,
"step": 102
},
{
"epoch": 0.2717678100263852,
"grad_norm": 0.1752789467573166,
"learning_rate": 9.698463103929542e-05,
"loss": 0.9518905878067017,
"step": 103
},
{
"epoch": 0.27440633245382584,
"grad_norm": 0.18624742329120636,
"learning_rate": 9.691140826286893e-05,
"loss": 0.9436296224594116,
"step": 104
},
{
"epoch": 0.2770448548812665,
"grad_norm": 0.160239577293396,
"learning_rate": 9.683733539658139e-05,
"loss": 1.0474563837051392,
"step": 105
},
{
"epoch": 0.2796833773087071,
"grad_norm": 0.1599990278482437,
"learning_rate": 9.676241378272021e-05,
"loss": 1.0836262702941895,
"step": 106
},
{
"epoch": 0.28232189973614774,
"grad_norm": 0.1549750417470932,
"learning_rate": 9.66866447789531e-05,
"loss": 0.9756717085838318,
"step": 107
},
{
"epoch": 0.2849604221635884,
"grad_norm": 0.1627299189567566,
"learning_rate": 9.661002975830349e-05,
"loss": 0.9459734559059143,
"step": 108
},
{
"epoch": 0.287598944591029,
"grad_norm": 0.15503612160682678,
"learning_rate": 9.653257010912559e-05,
"loss": 0.9717468023300171,
"step": 109
},
{
"epoch": 0.29023746701846964,
"grad_norm": 0.16452868282794952,
"learning_rate": 9.645426723507929e-05,
"loss": 0.9126809239387512,
"step": 110
},
{
"epoch": 0.2928759894459103,
"grad_norm": 0.17188014090061188,
"learning_rate": 9.637512255510475e-05,
"loss": 0.9921525120735168,
"step": 111
},
{
"epoch": 0.2955145118733509,
"grad_norm": 0.1495610475540161,
"learning_rate": 9.629513750339655e-05,
"loss": 0.866377592086792,
"step": 112
},
{
"epoch": 0.29815303430079154,
"grad_norm": 0.17267285287380219,
"learning_rate": 9.621431352937789e-05,
"loss": 0.9987406730651855,
"step": 113
},
{
"epoch": 0.3007915567282322,
"grad_norm": 0.15371519327163696,
"learning_rate": 9.613265209767417e-05,
"loss": 0.8675357103347778,
"step": 114
},
{
"epoch": 0.3034300791556728,
"grad_norm": 0.169174462556839,
"learning_rate": 9.605015468808651e-05,
"loss": 0.9573705792427063,
"step": 115
},
{
"epoch": 0.30606860158311344,
"grad_norm": 0.17276419699192047,
"learning_rate": 9.596682279556498e-05,
"loss": 1.0629740953445435,
"step": 116
},
{
"epoch": 0.3087071240105541,
"grad_norm": 0.16400578618049622,
"learning_rate": 9.58826579301814e-05,
"loss": 0.9535355567932129,
"step": 117
},
{
"epoch": 0.3113456464379947,
"grad_norm": 0.18149444460868835,
"learning_rate": 9.579766161710209e-05,
"loss": 1.0182714462280273,
"step": 118
},
{
"epoch": 0.31398416886543534,
"grad_norm": 0.17230141162872314,
"learning_rate": 9.571183539656011e-05,
"loss": 1.0627704858779907,
"step": 119
},
{
"epoch": 0.316622691292876,
"grad_norm": 0.16222770512104034,
"learning_rate": 9.56251808238275e-05,
"loss": 1.0022966861724854,
"step": 120
},
{
"epoch": 0.31926121372031663,
"grad_norm": 0.1661306470632553,
"learning_rate": 9.553769946918697e-05,
"loss": 0.9902384877204895,
"step": 121
},
{
"epoch": 0.32189973614775724,
"grad_norm": 0.1522214263677597,
"learning_rate": 9.544939291790351e-05,
"loss": 1.0071738958358765,
"step": 122
},
{
"epoch": 0.3245382585751979,
"grad_norm": 0.1561179757118225,
"learning_rate": 9.536026277019561e-05,
"loss": 1.0049107074737549,
"step": 123
},
{
"epoch": 0.32717678100263853,
"grad_norm": 0.16295814514160156,
"learning_rate": 9.527031064120632e-05,
"loss": 0.9579784870147705,
"step": 124
},
{
"epoch": 0.32981530343007914,
"grad_norm": 0.1570494920015335,
"learning_rate": 9.517953816097396e-05,
"loss": 0.9788990020751953,
"step": 125
},
{
"epoch": 0.3324538258575198,
"grad_norm": 0.1672319918870926,
"learning_rate": 9.508794697440257e-05,
"loss": 1.018532395362854,
"step": 126
},
{
"epoch": 0.33509234828496043,
"grad_norm": 0.1804421991109848,
"learning_rate": 9.499553874123212e-05,
"loss": 0.9944267272949219,
"step": 127
},
{
"epoch": 0.33773087071240104,
"grad_norm": 0.17225806415081024,
"learning_rate": 9.490231513600843e-05,
"loss": 1.0111069679260254,
"step": 128
},
{
"epoch": 0.3403693931398417,
"grad_norm": 0.15516908466815948,
"learning_rate": 9.480827784805278e-05,
"loss": 0.9605324268341064,
"step": 129
},
{
"epoch": 0.34300791556728233,
"grad_norm": 0.16040503978729248,
"learning_rate": 9.47134285814314e-05,
"loss": 0.933555006980896,
"step": 130
},
{
"epoch": 0.34564643799472294,
"grad_norm": 0.18396888673305511,
"learning_rate": 9.461776905492446e-05,
"loss": 1.0374318361282349,
"step": 131
},
{
"epoch": 0.3482849604221636,
"grad_norm": 0.17078086733818054,
"learning_rate": 9.452130100199503e-05,
"loss": 0.9799578785896301,
"step": 132
},
{
"epoch": 0.35092348284960423,
"grad_norm": 0.1631857305765152,
"learning_rate": 9.442402617075765e-05,
"loss": 0.9599941968917847,
"step": 133
},
{
"epoch": 0.35356200527704484,
"grad_norm": 0.15046699345111847,
"learning_rate": 9.43259463239466e-05,
"loss": 0.9483698010444641,
"step": 134
},
{
"epoch": 0.3562005277044855,
"grad_norm": 0.14928267896175385,
"learning_rate": 9.422706323888397e-05,
"loss": 0.8940687775611877,
"step": 135
},
{
"epoch": 0.35883905013192613,
"grad_norm": 0.17583905160427094,
"learning_rate": 9.412737870744752e-05,
"loss": 0.8684873580932617,
"step": 136
},
{
"epoch": 0.36147757255936674,
"grad_norm": 0.16805775463581085,
"learning_rate": 9.402689453603815e-05,
"loss": 1.027605652809143,
"step": 137
},
{
"epoch": 0.3641160949868074,
"grad_norm": 0.16120663285255432,
"learning_rate": 9.392561254554713e-05,
"loss": 0.9406588673591614,
"step": 138
},
{
"epoch": 0.36675461741424803,
"grad_norm": 0.16707223653793335,
"learning_rate": 9.382353457132317e-05,
"loss": 1.0448381900787354,
"step": 139
},
{
"epoch": 0.36939313984168864,
"grad_norm": 0.17459368705749512,
"learning_rate": 9.372066246313921e-05,
"loss": 0.9596197605133057,
"step": 140
},
{
"epoch": 0.3720316622691293,
"grad_norm": 0.16556183993816376,
"learning_rate": 9.361699808515876e-05,
"loss": 1.0335559844970703,
"step": 141
},
{
"epoch": 0.37467018469656993,
"grad_norm": 0.16449083387851715,
"learning_rate": 9.351254331590215e-05,
"loss": 0.9463608860969543,
"step": 142
},
{
"epoch": 0.37730870712401055,
"grad_norm": 0.16412568092346191,
"learning_rate": 9.340730004821266e-05,
"loss": 0.9579424858093262,
"step": 143
},
{
"epoch": 0.37994722955145116,
"grad_norm": 0.176985964179039,
"learning_rate": 9.330127018922194e-05,
"loss": 1.0462236404418945,
"step": 144
},
{
"epoch": 0.38258575197889183,
"grad_norm": 0.1623564213514328,
"learning_rate": 9.31944556603157e-05,
"loss": 0.9870291948318481,
"step": 145
},
{
"epoch": 0.38522427440633245,
"grad_norm": 0.18084806203842163,
"learning_rate": 9.308685839709879e-05,
"loss": 1.0054789781570435,
"step": 146
},
{
"epoch": 0.38786279683377306,
"grad_norm": 0.18217523396015167,
"learning_rate": 9.297848034936006e-05,
"loss": 1.019425868988037,
"step": 147
},
{
"epoch": 0.39050131926121373,
"grad_norm": 0.15862634778022766,
"learning_rate": 9.286932348103716e-05,
"loss": 0.8727700710296631,
"step": 148
},
{
"epoch": 0.39313984168865435,
"grad_norm": 0.17286421358585358,
"learning_rate": 9.275938977018081e-05,
"loss": 1.0189862251281738,
"step": 149
},
{
"epoch": 0.39577836411609496,
"grad_norm": 0.16580720245838165,
"learning_rate": 9.264868120891912e-05,
"loss": 1.0655592679977417,
"step": 150
},
{
"epoch": 0.39841688654353563,
"grad_norm": 0.16368559002876282,
"learning_rate": 9.253719980342135e-05,
"loss": 0.937400221824646,
"step": 151
},
{
"epoch": 0.40105540897097625,
"grad_norm": 0.1721360981464386,
"learning_rate": 9.24249475738616e-05,
"loss": 0.9234015941619873,
"step": 152
},
{
"epoch": 0.40369393139841686,
"grad_norm": 0.18208572268486023,
"learning_rate": 9.231192655438221e-05,
"loss": 1.0558464527130127,
"step": 153
},
{
"epoch": 0.40633245382585753,
"grad_norm": 0.1552550047636032,
"learning_rate": 9.219813879305692e-05,
"loss": 0.9326491355895996,
"step": 154
},
{
"epoch": 0.40897097625329815,
"grad_norm": 0.1624125838279724,
"learning_rate": 9.208358635185373e-05,
"loss": 1.020031452178955,
"step": 155
},
{
"epoch": 0.41160949868073876,
"grad_norm": 0.1575743854045868,
"learning_rate": 9.19682713065975e-05,
"loss": 1.0257234573364258,
"step": 156
},
{
"epoch": 0.41424802110817943,
"grad_norm": 0.14020493626594543,
"learning_rate": 9.185219574693242e-05,
"loss": 0.9181210994720459,
"step": 157
},
{
"epoch": 0.41688654353562005,
"grad_norm": 0.18366360664367676,
"learning_rate": 9.173536177628409e-05,
"loss": 1.1496880054473877,
"step": 158
},
{
"epoch": 0.41952506596306066,
"grad_norm": 0.15704800188541412,
"learning_rate": 9.161777151182136e-05,
"loss": 0.9523612260818481,
"step": 159
},
{
"epoch": 0.42216358839050133,
"grad_norm": 0.20316386222839355,
"learning_rate": 9.149942708441808e-05,
"loss": 1.1875518560409546,
"step": 160
},
{
"epoch": 0.42480211081794195,
"grad_norm": 0.16364015638828278,
"learning_rate": 9.138033063861436e-05,
"loss": 0.9598340392112732,
"step": 161
},
{
"epoch": 0.42744063324538256,
"grad_norm": 0.17661993205547333,
"learning_rate": 9.126048433257779e-05,
"loss": 1.0217077732086182,
"step": 162
},
{
"epoch": 0.43007915567282323,
"grad_norm": 0.1741223782300949,
"learning_rate": 9.113989033806434e-05,
"loss": 1.0106780529022217,
"step": 163
},
{
"epoch": 0.43271767810026385,
"grad_norm": 0.1821887344121933,
"learning_rate": 9.101855084037894e-05,
"loss": 0.992047905921936,
"step": 164
},
{
"epoch": 0.43535620052770446,
"grad_norm": 0.1628859043121338,
"learning_rate": 9.089646803833589e-05,
"loss": 1.083019733428955,
"step": 165
},
{
"epoch": 0.43799472295514513,
"grad_norm": 0.1696838140487671,
"learning_rate": 9.077364414421909e-05,
"loss": 1.0616456270217896,
"step": 166
},
{
"epoch": 0.44063324538258575,
"grad_norm": 0.17563055455684662,
"learning_rate": 9.065008138374189e-05,
"loss": 0.981091320514679,
"step": 167
},
{
"epoch": 0.44327176781002636,
"grad_norm": 0.16527481377124786,
"learning_rate": 9.052578199600675e-05,
"loss": 0.9541459083557129,
"step": 168
},
{
"epoch": 0.44591029023746703,
"grad_norm": 0.1794591099023819,
"learning_rate": 9.040074823346465e-05,
"loss": 0.9892167448997498,
"step": 169
},
{
"epoch": 0.44854881266490765,
"grad_norm": 0.1672830432653427,
"learning_rate": 9.027498236187439e-05,
"loss": 1.0351946353912354,
"step": 170
},
{
"epoch": 0.45118733509234826,
"grad_norm": 0.17655089497566223,
"learning_rate": 9.014848666026138e-05,
"loss": 0.9711981415748596,
"step": 171
},
{
"epoch": 0.45382585751978893,
"grad_norm": 0.1557280719280243,
"learning_rate": 9.002126342087642e-05,
"loss": 0.9778252840042114,
"step": 172
},
{
"epoch": 0.45646437994722955,
"grad_norm": 0.17430546879768372,
"learning_rate": 8.989331494915417e-05,
"loss": 1.0428410768508911,
"step": 173
},
{
"epoch": 0.45910290237467016,
"grad_norm": 0.1841093748807907,
"learning_rate": 8.976464356367134e-05,
"loss": 0.9610801935195923,
"step": 174
},
{
"epoch": 0.46174142480211083,
"grad_norm": 0.18140369653701782,
"learning_rate": 8.963525159610465e-05,
"loss": 1.0488272905349731,
"step": 175
},
{
"epoch": 0.46437994722955145,
"grad_norm": 0.16312600672245026,
"learning_rate": 8.950514139118868e-05,
"loss": 0.995346188545227,
"step": 176
},
{
"epoch": 0.46701846965699206,
"grad_norm": 0.18155968189239502,
"learning_rate": 8.937431530667328e-05,
"loss": 1.089247226715088,
"step": 177
},
{
"epoch": 0.46965699208443273,
"grad_norm": 0.16854368150234222,
"learning_rate": 8.92427757132809e-05,
"loss": 1.0536508560180664,
"step": 178
},
{
"epoch": 0.47229551451187335,
"grad_norm": 0.16431039571762085,
"learning_rate": 8.911052499466357e-05,
"loss": 0.9526739716529846,
"step": 179
},
{
"epoch": 0.47493403693931396,
"grad_norm": 0.16609320044517517,
"learning_rate": 8.897756554735977e-05,
"loss": 1.0196548700332642,
"step": 180
},
{
"epoch": 0.47757255936675463,
"grad_norm": 0.16544099152088165,
"learning_rate": 8.884389978075098e-05,
"loss": 0.9600881338119507,
"step": 181
},
{
"epoch": 0.48021108179419525,
"grad_norm": 0.18530991673469543,
"learning_rate": 8.870953011701804e-05,
"loss": 1.0501227378845215,
"step": 182
},
{
"epoch": 0.48284960422163586,
"grad_norm": 0.147592693567276,
"learning_rate": 8.857445899109715e-05,
"loss": 1.0376373529434204,
"step": 183
},
{
"epoch": 0.48548812664907653,
"grad_norm": 0.19425112009048462,
"learning_rate": 8.843868885063593e-05,
"loss": 1.0233935117721558,
"step": 184
},
{
"epoch": 0.48812664907651715,
"grad_norm": 0.15668635070323944,
"learning_rate": 8.83022221559489e-05,
"loss": 1.0277446508407593,
"step": 185
},
{
"epoch": 0.49076517150395776,
"grad_norm": 0.1669091135263443,
"learning_rate": 8.8165061379973e-05,
"loss": 1.032301425933838,
"step": 186
},
{
"epoch": 0.49340369393139843,
"grad_norm": 0.1691151261329651,
"learning_rate": 8.80272090082227e-05,
"loss": 0.9066123366355896,
"step": 187
},
{
"epoch": 0.49604221635883905,
"grad_norm": 0.17567206919193268,
"learning_rate": 8.788866753874503e-05,
"loss": 1.017852544784546,
"step": 188
},
{
"epoch": 0.49868073878627966,
"grad_norm": 0.17546191811561584,
"learning_rate": 8.774943948207426e-05,
"loss": 1.0644317865371704,
"step": 189
},
{
"epoch": 0.5013192612137203,
"grad_norm": 0.16563469171524048,
"learning_rate": 8.760952736118644e-05,
"loss": 1.0079925060272217,
"step": 190
},
{
"epoch": 0.503957783641161,
"grad_norm": 0.15652592480182648,
"learning_rate": 8.746893371145366e-05,
"loss": 0.9726508855819702,
"step": 191
},
{
"epoch": 0.5065963060686016,
"grad_norm": 0.16774515807628632,
"learning_rate": 8.732766108059813e-05,
"loss": 1.0333598852157593,
"step": 192
},
{
"epoch": 0.5092348284960422,
"grad_norm": 0.19023677706718445,
"learning_rate": 8.718571202864598e-05,
"loss": 1.0274595022201538,
"step": 193
},
{
"epoch": 0.5118733509234829,
"grad_norm": 0.16863608360290527,
"learning_rate": 8.70430891278809e-05,
"loss": 0.990319013595581,
"step": 194
},
{
"epoch": 0.5145118733509235,
"grad_norm": 0.1866767555475235,
"learning_rate": 8.689979496279746e-05,
"loss": 1.0066261291503906,
"step": 195
},
{
"epoch": 0.5171503957783641,
"grad_norm": 0.1698773354291916,
"learning_rate": 8.675583213005443e-05,
"loss": 1.0301073789596558,
"step": 196
},
{
"epoch": 0.5197889182058048,
"grad_norm": 0.18543866276741028,
"learning_rate": 8.661120323842751e-05,
"loss": 1.0110327005386353,
"step": 197
},
{
"epoch": 0.5224274406332454,
"grad_norm": 0.17273621261119843,
"learning_rate": 8.646591090876224e-05,
"loss": 0.942511796951294,
"step": 198
},
{
"epoch": 0.525065963060686,
"grad_norm": 0.17858357727527618,
"learning_rate": 8.631995777392645e-05,
"loss": 0.9492037892341614,
"step": 199
},
{
"epoch": 0.5277044854881267,
"grad_norm": 0.21972784399986267,
"learning_rate": 8.61733464787625e-05,
"loss": 1.125957727432251,
"step": 200
},
{
"epoch": 0.5303430079155673,
"grad_norm": 0.17801721394062042,
"learning_rate": 8.602607968003935e-05,
"loss": 1.0186785459518433,
"step": 201
},
{
"epoch": 0.5329815303430079,
"grad_norm": 0.1811058521270752,
"learning_rate": 8.587816004640456e-05,
"loss": 1.0295569896697998,
"step": 202
},
{
"epoch": 0.5356200527704486,
"grad_norm": 0.1558198630809784,
"learning_rate": 8.572959025833573e-05,
"loss": 0.882567286491394,
"step": 203
},
{
"epoch": 0.5382585751978892,
"grad_norm": 0.1586204618215561,
"learning_rate": 8.558037300809208e-05,
"loss": 0.8986984491348267,
"step": 204
},
{
"epoch": 0.5408970976253298,
"grad_norm": 0.17992497980594635,
"learning_rate": 8.543051099966558e-05,
"loss": 0.9992472529411316,
"step": 205
},
{
"epoch": 0.5435356200527705,
"grad_norm": 0.16778439283370972,
"learning_rate": 8.528000694873198e-05,
"loss": 0.9697674512863159,
"step": 206
},
{
"epoch": 0.5461741424802111,
"grad_norm": 0.15742550790309906,
"learning_rate": 8.512886358260162e-05,
"loss": 0.9573698043823242,
"step": 207
},
{
"epoch": 0.5488126649076517,
"grad_norm": 0.16890263557434082,
"learning_rate": 8.49770836401699e-05,
"loss": 0.8758108615875244,
"step": 208
},
{
"epoch": 0.5514511873350924,
"grad_norm": 0.16583888232707977,
"learning_rate": 8.482466987186785e-05,
"loss": 0.9687983989715576,
"step": 209
},
{
"epoch": 0.554089709762533,
"grad_norm": 0.17507483065128326,
"learning_rate": 8.467162503961208e-05,
"loss": 1.0307694673538208,
"step": 210
},
{
"epoch": 0.5567282321899736,
"grad_norm": 0.1639733910560608,
"learning_rate": 8.451795191675488e-05,
"loss": 0.9487868547439575,
"step": 211
},
{
"epoch": 0.5593667546174143,
"grad_norm": 0.18524755537509918,
"learning_rate": 8.436365328803385e-05,
"loss": 1.0524779558181763,
"step": 212
},
{
"epoch": 0.5620052770448549,
"grad_norm": 0.1726824939250946,
"learning_rate": 8.420873194952152e-05,
"loss": 1.0479933023452759,
"step": 213
},
{
"epoch": 0.5646437994722955,
"grad_norm": 0.15832193195819855,
"learning_rate": 8.405319070857467e-05,
"loss": 0.9252246022224426,
"step": 214
},
{
"epoch": 0.5672823218997362,
"grad_norm": 0.18013589084148407,
"learning_rate": 8.389703238378339e-05,
"loss": 1.035888671875,
"step": 215
},
{
"epoch": 0.5699208443271768,
"grad_norm": 0.1560203582048416,
"learning_rate": 8.374025980492011e-05,
"loss": 0.9540398120880127,
"step": 216
},
{
"epoch": 0.5725593667546174,
"grad_norm": 0.16869907081127167,
"learning_rate": 8.358287581288822e-05,
"loss": 1.0841728448867798,
"step": 217
},
{
"epoch": 0.575197889182058,
"grad_norm": 0.1753094494342804,
"learning_rate": 8.342488325967068e-05,
"loss": 1.0053566694259644,
"step": 218
},
{
"epoch": 0.5778364116094987,
"grad_norm": 0.16862048208713531,
"learning_rate": 8.326628500827826e-05,
"loss": 0.9311268329620361,
"step": 219
},
{
"epoch": 0.5804749340369393,
"grad_norm": 0.1801312416791916,
"learning_rate": 8.310708393269772e-05,
"loss": 0.9063538908958435,
"step": 220
},
{
"epoch": 0.58311345646438,
"grad_norm": 0.18925103545188904,
"learning_rate": 8.294728291783966e-05,
"loss": 1.0815038681030273,
"step": 221
},
{
"epoch": 0.5857519788918206,
"grad_norm": 0.18007421493530273,
"learning_rate": 8.278688485948633e-05,
"loss": 1.0415452718734741,
"step": 222
},
{
"epoch": 0.5883905013192612,
"grad_norm": 0.1658685952425003,
"learning_rate": 8.262589266423908e-05,
"loss": 0.9547388553619385,
"step": 223
},
{
"epoch": 0.5910290237467019,
"grad_norm": 0.16324631869792938,
"learning_rate": 8.246430924946575e-05,
"loss": 0.900030791759491,
"step": 224
},
{
"epoch": 0.5936675461741425,
"grad_norm": 0.17047886550426483,
"learning_rate": 8.230213754324773e-05,
"loss": 0.914841890335083,
"step": 225
},
{
"epoch": 0.5963060686015831,
"grad_norm": 0.18901900947093964,
"learning_rate": 8.213938048432697e-05,
"loss": 0.8218327760696411,
"step": 226
},
{
"epoch": 0.5989445910290238,
"grad_norm": 0.18023498356342316,
"learning_rate": 8.197604102205271e-05,
"loss": 1.0853064060211182,
"step": 227
},
{
"epoch": 0.6015831134564644,
"grad_norm": 0.16877271234989166,
"learning_rate": 8.181212211632799e-05,
"loss": 0.9420395493507385,
"step": 228
},
{
"epoch": 0.604221635883905,
"grad_norm": 0.1619059145450592,
"learning_rate": 8.16476267375561e-05,
"loss": 0.9768050909042358,
"step": 229
},
{
"epoch": 0.6068601583113457,
"grad_norm": 0.18686461448669434,
"learning_rate": 8.14825578665866e-05,
"loss": 1.000288486480713,
"step": 230
},
{
"epoch": 0.6094986807387863,
"grad_norm": 0.1876915544271469,
"learning_rate": 8.131691849466153e-05,
"loss": 1.0954421758651733,
"step": 231
},
{
"epoch": 0.6121372031662269,
"grad_norm": 0.17862121760845184,
"learning_rate": 8.115071162336098e-05,
"loss": 0.9122456908226013,
"step": 232
},
{
"epoch": 0.6147757255936676,
"grad_norm": 0.17977792024612427,
"learning_rate": 8.098394026454885e-05,
"loss": 0.9491496086120605,
"step": 233
},
{
"epoch": 0.6174142480211082,
"grad_norm": 0.1632521152496338,
"learning_rate": 8.081660744031819e-05,
"loss": 0.9856275916099548,
"step": 234
},
{
"epoch": 0.6200527704485488,
"grad_norm": 0.17910513281822205,
"learning_rate": 8.064871618293646e-05,
"loss": 1.0011181831359863,
"step": 235
},
{
"epoch": 0.6226912928759895,
"grad_norm": 0.1681145429611206,
"learning_rate": 8.048026953479062e-05,
"loss": 0.973308801651001,
"step": 236
},
{
"epoch": 0.6253298153034301,
"grad_norm": 0.18468503654003143,
"learning_rate": 8.03112705483319e-05,
"loss": 1.0287890434265137,
"step": 237
},
{
"epoch": 0.6279683377308707,
"grad_norm": 0.16042198240756989,
"learning_rate": 8.014172228602064e-05,
"loss": 0.9670297503471375,
"step": 238
},
{
"epoch": 0.6306068601583114,
"grad_norm": 0.16996483504772186,
"learning_rate": 7.997162782027061e-05,
"loss": 0.9653870463371277,
"step": 239
},
{
"epoch": 0.633245382585752,
"grad_norm": 0.1739293932914734,
"learning_rate": 7.98009902333935e-05,
"loss": 0.9817367792129517,
"step": 240
},
{
"epoch": 0.6358839050131926,
"grad_norm": 0.17517700791358948,
"learning_rate": 7.962981261754294e-05,
"loss": 0.9925550222396851,
"step": 241
},
{
"epoch": 0.6385224274406333,
"grad_norm": 0.1885872781276703,
"learning_rate": 7.945809807465857e-05,
"loss": 0.9885467886924744,
"step": 242
},
{
"epoch": 0.6411609498680739,
"grad_norm": 0.1705755591392517,
"learning_rate": 7.928584971640974e-05,
"loss": 1.0115675926208496,
"step": 243
},
{
"epoch": 0.6437994722955145,
"grad_norm": 0.15275037288665771,
"learning_rate": 7.911307066413919e-05,
"loss": 0.9338511228561401,
"step": 244
},
{
"epoch": 0.6464379947229552,
"grad_norm": 0.17367307841777802,
"learning_rate": 7.893976404880643e-05,
"loss": 0.9358500242233276,
"step": 245
},
{
"epoch": 0.6490765171503958,
"grad_norm": 0.18579082190990448,
"learning_rate": 7.876593301093104e-05,
"loss": 0.993777871131897,
"step": 246
},
{
"epoch": 0.6517150395778364,
"grad_norm": 0.17416690289974213,
"learning_rate": 7.859158070053577e-05,
"loss": 0.9911141991615295,
"step": 247
},
{
"epoch": 0.6543535620052771,
"grad_norm": 0.17322957515716553,
"learning_rate": 7.841671027708944e-05,
"loss": 0.9777154922485352,
"step": 248
},
{
"epoch": 0.6569920844327177,
"grad_norm": 0.14296574890613556,
"learning_rate": 7.824132490944967e-05,
"loss": 0.8306946158409119,
"step": 249
},
{
"epoch": 0.6596306068601583,
"grad_norm": 0.18208211660385132,
"learning_rate": 7.80654277758055e-05,
"loss": 0.9833745956420898,
"step": 250
},
{
"epoch": 0.662269129287599,
"grad_norm": 0.15957476198673248,
"learning_rate": 7.788902206361973e-05,
"loss": 0.9916603565216064,
"step": 251
},
{
"epoch": 0.6649076517150396,
"grad_norm": 0.17746901512145996,
"learning_rate": 7.771211096957125e-05,
"loss": 0.9748624563217163,
"step": 252
},
{
"epoch": 0.6675461741424802,
"grad_norm": 0.19569706916809082,
"learning_rate": 7.7534697699497e-05,
"loss": 0.9759517908096313,
"step": 253
},
{
"epoch": 0.6701846965699209,
"grad_norm": 0.1637251377105713,
"learning_rate": 7.735678546833402e-05,
"loss": 0.9450954794883728,
"step": 254
},
{
"epoch": 0.6728232189973615,
"grad_norm": 0.18049630522727966,
"learning_rate": 7.717837750006106e-05,
"loss": 1.0253992080688477,
"step": 255
},
{
"epoch": 0.6754617414248021,
"grad_norm": 0.165897935628891,
"learning_rate": 7.699947702764021e-05,
"loss": 1.0213006734848022,
"step": 256
},
{
"epoch": 0.6781002638522428,
"grad_norm": 0.20414817333221436,
"learning_rate": 7.682008729295833e-05,
"loss": 1.0483496189117432,
"step": 257
},
{
"epoch": 0.6807387862796834,
"grad_norm": 0.16246487200260162,
"learning_rate": 7.664021154676829e-05,
"loss": 0.9613660573959351,
"step": 258
},
{
"epoch": 0.683377308707124,
"grad_norm": 0.18064430356025696,
"learning_rate": 7.645985304863003e-05,
"loss": 0.9091204404830933,
"step": 259
},
{
"epoch": 0.6860158311345647,
"grad_norm": 0.17883849143981934,
"learning_rate": 7.627901506685157e-05,
"loss": 0.9444289207458496,
"step": 260
},
{
"epoch": 0.6886543535620053,
"grad_norm": 0.17278419435024261,
"learning_rate": 7.609770087842969e-05,
"loss": 0.8510349988937378,
"step": 261
},
{
"epoch": 0.6912928759894459,
"grad_norm": 0.15988567471504211,
"learning_rate": 7.591591376899062e-05,
"loss": 0.9418317079544067,
"step": 262
},
{
"epoch": 0.6939313984168866,
"grad_norm": 0.1863398402929306,
"learning_rate": 7.573365703273046e-05,
"loss": 0.9580644369125366,
"step": 263
},
{
"epoch": 0.6965699208443272,
"grad_norm": 0.16767530143260956,
"learning_rate": 7.555093397235552e-05,
"loss": 0.8833745121955872,
"step": 264
},
{
"epoch": 0.6992084432717678,
"grad_norm": 0.17919766902923584,
"learning_rate": 7.536774789902246e-05,
"loss": 0.9687992334365845,
"step": 265
},
{
"epoch": 0.7018469656992085,
"grad_norm": 0.20329569280147552,
"learning_rate": 7.518410213227822e-05,
"loss": 0.8355344533920288,
"step": 266
},
{
"epoch": 0.7044854881266491,
"grad_norm": 0.17873340845108032,
"learning_rate": 7.500000000000001e-05,
"loss": 0.9391046166419983,
"step": 267
},
{
"epoch": 0.7071240105540897,
"grad_norm": 0.17777030169963837,
"learning_rate": 7.481544483833485e-05,
"loss": 1.0823737382888794,
"step": 268
},
{
"epoch": 0.7097625329815304,
"grad_norm": 0.18217693269252777,
"learning_rate": 7.463043999163919e-05,
"loss": 1.016803503036499,
"step": 269
},
{
"epoch": 0.712401055408971,
"grad_norm": 0.1759069412946701,
"learning_rate": 7.444498881241835e-05,
"loss": 1.019862413406372,
"step": 270
},
{
"epoch": 0.7150395778364116,
"grad_norm": 0.17746379971504211,
"learning_rate": 7.425909466126568e-05,
"loss": 0.996162474155426,
"step": 271
},
{
"epoch": 0.7176781002638523,
"grad_norm": 0.1615501195192337,
"learning_rate": 7.407276090680172e-05,
"loss": 1.0143609046936035,
"step": 272
},
{
"epoch": 0.7203166226912929,
"grad_norm": 0.17791925370693207,
"learning_rate": 7.388599092561315e-05,
"loss": 0.9809026718139648,
"step": 273
},
{
"epoch": 0.7229551451187335,
"grad_norm": 0.17972691357135773,
"learning_rate": 7.369878810219154e-05,
"loss": 1.0634626150131226,
"step": 274
},
{
"epoch": 0.7255936675461742,
"grad_norm": 0.18457219004631042,
"learning_rate": 7.351115582887211e-05,
"loss": 0.9123151302337646,
"step": 275
},
{
"epoch": 0.7282321899736148,
"grad_norm": 0.1787765920162201,
"learning_rate": 7.332309750577222e-05,
"loss": 1.0113319158554077,
"step": 276
},
{
"epoch": 0.7308707124010554,
"grad_norm": 0.15324430167675018,
"learning_rate": 7.313461654072973e-05,
"loss": 0.8716138601303101,
"step": 277
},
{
"epoch": 0.7335092348284961,
"grad_norm": 0.17701661586761475,
"learning_rate": 7.29457163492413e-05,
"loss": 0.9495022296905518,
"step": 278
},
{
"epoch": 0.7361477572559367,
"grad_norm": 0.1860339343547821,
"learning_rate": 7.275640035440045e-05,
"loss": 1.1002585887908936,
"step": 279
},
{
"epoch": 0.7387862796833773,
"grad_norm": 0.1833198368549347,
"learning_rate": 7.256667198683552e-05,
"loss": 1.0140724182128906,
"step": 280
},
{
"epoch": 0.741424802110818,
"grad_norm": 0.16927815973758698,
"learning_rate": 7.237653468464756e-05,
"loss": 0.8884044885635376,
"step": 281
},
{
"epoch": 0.7440633245382586,
"grad_norm": 0.1619526445865631,
"learning_rate": 7.218599189334799e-05,
"loss": 0.8716824054718018,
"step": 282
},
{
"epoch": 0.7467018469656992,
"grad_norm": 0.1803620606660843,
"learning_rate": 7.199504706579617e-05,
"loss": 0.9060261845588684,
"step": 283
},
{
"epoch": 0.7493403693931399,
"grad_norm": 0.1543535739183426,
"learning_rate": 7.180370366213684e-05,
"loss": 0.9115622043609619,
"step": 284
},
{
"epoch": 0.7519788918205804,
"grad_norm": 0.18664343655109406,
"learning_rate": 7.161196514973734e-05,
"loss": 0.9853682518005371,
"step": 285
},
{
"epoch": 0.7546174142480211,
"grad_norm": 0.165468230843544,
"learning_rate": 7.141983500312498e-05,
"loss": 0.9828847646713257,
"step": 286
},
{
"epoch": 0.7572559366754618,
"grad_norm": 0.16414625942707062,
"learning_rate": 7.12273167039238e-05,
"loss": 0.9459795951843262,
"step": 287
},
{
"epoch": 0.7598944591029023,
"grad_norm": 0.16700607538223267,
"learning_rate": 7.10344137407917e-05,
"loss": 0.9463711977005005,
"step": 288
},
{
"epoch": 0.762532981530343,
"grad_norm": 0.18241791427135468,
"learning_rate": 7.084112960935716e-05,
"loss": 1.0513451099395752,
"step": 289
},
{
"epoch": 0.7651715039577837,
"grad_norm": 0.1688919961452484,
"learning_rate": 7.064746781215577e-05,
"loss": 0.9387044310569763,
"step": 290
},
{
"epoch": 0.7678100263852242,
"grad_norm": 0.19025741517543793,
"learning_rate": 7.045343185856701e-05,
"loss": 1.0054974555969238,
"step": 291
},
{
"epoch": 0.7704485488126649,
"grad_norm": 0.15950188040733337,
"learning_rate": 7.02590252647504e-05,
"loss": 0.958078920841217,
"step": 292
},
{
"epoch": 0.7730870712401056,
"grad_norm": 0.16371142864227295,
"learning_rate": 7.006425155358195e-05,
"loss": 0.8427443504333496,
"step": 293
},
{
"epoch": 0.7757255936675461,
"grad_norm": 0.16640605032444,
"learning_rate": 6.986911425459028e-05,
"loss": 0.9656301736831665,
"step": 294
},
{
"epoch": 0.7783641160949868,
"grad_norm": 0.15756769478321075,
"learning_rate": 6.967361690389258e-05,
"loss": 0.8850897550582886,
"step": 295
},
{
"epoch": 0.7810026385224275,
"grad_norm": 0.1860872358083725,
"learning_rate": 6.947776304413071e-05,
"loss": 1.0678696632385254,
"step": 296
},
{
"epoch": 0.783641160949868,
"grad_norm": 0.1724071353673935,
"learning_rate": 6.92815562244068e-05,
"loss": 0.9517515897750854,
"step": 297
},
{
"epoch": 0.7862796833773087,
"grad_norm": 0.163114994764328,
"learning_rate": 6.908500000021904e-05,
"loss": 1.0018560886383057,
"step": 298
},
{
"epoch": 0.7889182058047494,
"grad_norm": 0.1724291294813156,
"learning_rate": 6.88880979333973e-05,
"loss": 0.8515908718109131,
"step": 299
},
{
"epoch": 0.7915567282321899,
"grad_norm": 0.18454426527023315,
"learning_rate": 6.869085359203844e-05,
"loss": 1.0833189487457275,
"step": 300
},
{
"epoch": 0.7941952506596306,
"grad_norm": 0.19565890729427338,
"learning_rate": 6.849327055044183e-05,
"loss": 0.9864209294319153,
"step": 301
},
{
"epoch": 0.7968337730870713,
"grad_norm": 0.1698215752840042,
"learning_rate": 6.82953523890444e-05,
"loss": 1.0138895511627197,
"step": 302
},
{
"epoch": 0.7994722955145118,
"grad_norm": 0.18076743185520172,
"learning_rate": 6.809710269435589e-05,
"loss": 0.9143779277801514,
"step": 303
},
{
"epoch": 0.8021108179419525,
"grad_norm": 0.18112292885780334,
"learning_rate": 6.789852505889383e-05,
"loss": 0.9987108707427979,
"step": 304
},
{
"epoch": 0.8047493403693932,
"grad_norm": 0.19905203580856323,
"learning_rate": 6.769962308111839e-05,
"loss": 1.0594322681427002,
"step": 305
},
{
"epoch": 0.8073878627968337,
"grad_norm": 0.15984803438186646,
"learning_rate": 6.750040036536718e-05,
"loss": 0.8765107989311218,
"step": 306
},
{
"epoch": 0.8100263852242744,
"grad_norm": 0.18265588581562042,
"learning_rate": 6.730086052179004e-05,
"loss": 1.0807573795318604,
"step": 307
},
{
"epoch": 0.8126649076517151,
"grad_norm": 0.17338073253631592,
"learning_rate": 6.710100716628344e-05,
"loss": 1.0164018869400024,
"step": 308
},
{
"epoch": 0.8153034300791556,
"grad_norm": 0.16966766119003296,
"learning_rate": 6.690084392042513e-05,
"loss": 0.8892185688018799,
"step": 309
},
{
"epoch": 0.8179419525065963,
"grad_norm": 0.1647440642118454,
"learning_rate": 6.670037441140843e-05,
"loss": 0.9149801731109619,
"step": 310
},
{
"epoch": 0.820580474934037,
"grad_norm": 0.1590905487537384,
"learning_rate": 6.649960227197647e-05,
"loss": 0.9656392335891724,
"step": 311
},
{
"epoch": 0.8232189973614775,
"grad_norm": 0.1711450070142746,
"learning_rate": 6.629853114035642e-05,
"loss": 0.8801575899124146,
"step": 312
},
{
"epoch": 0.8258575197889182,
"grad_norm": 0.16500213742256165,
"learning_rate": 6.609716466019356e-05,
"loss": 0.962440013885498,
"step": 313
},
{
"epoch": 0.8284960422163589,
"grad_norm": 0.16399310529232025,
"learning_rate": 6.589550648048517e-05,
"loss": 0.9615823030471802,
"step": 314
},
{
"epoch": 0.8311345646437994,
"grad_norm": 0.1439095288515091,
"learning_rate": 6.569356025551454e-05,
"loss": 0.8974040150642395,
"step": 315
},
{
"epoch": 0.8337730870712401,
"grad_norm": 0.1773672252893448,
"learning_rate": 6.549132964478465e-05,
"loss": 0.8947254419326782,
"step": 316
},
{
"epoch": 0.8364116094986808,
"grad_norm": 0.1713113933801651,
"learning_rate": 6.528881831295188e-05,
"loss": 0.9751154184341431,
"step": 317
},
{
"epoch": 0.8390501319261213,
"grad_norm": 0.16960115730762482,
"learning_rate": 6.508602992975963e-05,
"loss": 0.887660026550293,
"step": 318
},
{
"epoch": 0.841688654353562,
"grad_norm": 0.16344815492630005,
"learning_rate": 6.488296816997173e-05,
"loss": 1.0093176364898682,
"step": 319
},
{
"epoch": 0.8443271767810027,
"grad_norm": 0.1804485321044922,
"learning_rate": 6.467963671330602e-05,
"loss": 0.9559671878814697,
"step": 320
},
{
"epoch": 0.8469656992084432,
"grad_norm": 0.1776222586631775,
"learning_rate": 6.447603924436744e-05,
"loss": 0.9938176870346069,
"step": 321
},
{
"epoch": 0.8496042216358839,
"grad_norm": 0.18694733083248138,
"learning_rate": 6.427217945258151e-05,
"loss": 1.0674500465393066,
"step": 322
},
{
"epoch": 0.8522427440633246,
"grad_norm": 0.19572079181671143,
"learning_rate": 6.406806103212725e-05,
"loss": 1.019213318824768,
"step": 323
},
{
"epoch": 0.8548812664907651,
"grad_norm": 0.18069222569465637,
"learning_rate": 6.38636876818704e-05,
"loss": 1.1204936504364014,
"step": 324
},
{
"epoch": 0.8575197889182058,
"grad_norm": 0.18144075572490692,
"learning_rate": 6.36590631052963e-05,
"loss": 0.9554729461669922,
"step": 325
},
{
"epoch": 0.8601583113456465,
"grad_norm": 0.17216528952121735,
"learning_rate": 6.34541910104428e-05,
"loss": 0.9567139148712158,
"step": 326
},
{
"epoch": 0.862796833773087,
"grad_norm": 0.16113649308681488,
"learning_rate": 6.32490751098331e-05,
"loss": 0.8932434320449829,
"step": 327
},
{
"epoch": 0.8654353562005277,
"grad_norm": 0.17543751001358032,
"learning_rate": 6.304371912040839e-05,
"loss": 1.0658743381500244,
"step": 328
},
{
"epoch": 0.8680738786279684,
"grad_norm": 0.1963348686695099,
"learning_rate": 6.283812676346063e-05,
"loss": 0.9329290390014648,
"step": 329
},
{
"epoch": 0.8707124010554089,
"grad_norm": 0.14915871620178223,
"learning_rate": 6.263230176456498e-05,
"loss": 0.8152093291282654,
"step": 330
},
{
"epoch": 0.8733509234828496,
"grad_norm": 0.16528934240341187,
"learning_rate": 6.242624785351236e-05,
"loss": 0.9133073091506958,
"step": 331
},
{
"epoch": 0.8759894459102903,
"grad_norm": 0.16276095807552338,
"learning_rate": 6.221996876424185e-05,
"loss": 0.9373147487640381,
"step": 332
},
{
"epoch": 0.8786279683377308,
"grad_norm": 0.1769801527261734,
"learning_rate": 6.201346823477303e-05,
"loss": 0.9239742159843445,
"step": 333
},
{
"epoch": 0.8812664907651715,
"grad_norm": 0.166982963681221,
"learning_rate": 6.180675000713825e-05,
"loss": 0.9344125986099243,
"step": 334
},
{
"epoch": 0.8839050131926122,
"grad_norm": 0.17170056700706482,
"learning_rate": 6.159981782731474e-05,
"loss": 0.9131274223327637,
"step": 335
},
{
"epoch": 0.8865435356200527,
"grad_norm": 0.1550513505935669,
"learning_rate": 6.139267544515689e-05,
"loss": 0.9826679229736328,
"step": 336
},
{
"epoch": 0.8891820580474934,
"grad_norm": 0.16641782224178314,
"learning_rate": 6.118532661432812e-05,
"loss": 0.9474678039550781,
"step": 337
},
{
"epoch": 0.8918205804749341,
"grad_norm": 0.1698181927204132,
"learning_rate": 6.097777509223299e-05,
"loss": 0.9482673406600952,
"step": 338
},
{
"epoch": 0.8944591029023746,
"grad_norm": 0.17008566856384277,
"learning_rate": 6.0770024639949074e-05,
"loss": 0.9680395126342773,
"step": 339
},
{
"epoch": 0.8970976253298153,
"grad_norm": 0.15967540442943573,
"learning_rate": 6.056207902215874e-05,
"loss": 1.001111388206482,
"step": 340
},
{
"epoch": 0.899736147757256,
"grad_norm": 0.1756800413131714,
"learning_rate": 6.0353942007081046e-05,
"loss": 0.9691710472106934,
"step": 341
},
{
"epoch": 0.9023746701846965,
"grad_norm": 0.1710166037082672,
"learning_rate": 6.014561736640334e-05,
"loss": 0.945921778678894,
"step": 342
},
{
"epoch": 0.9050131926121372,
"grad_norm": 0.17799116671085358,
"learning_rate": 5.993710887521302e-05,
"loss": 1.0534354448318481,
"step": 343
},
{
"epoch": 0.9076517150395779,
"grad_norm": 0.1695486456155777,
"learning_rate": 5.9728420311929014e-05,
"loss": 0.9974007606506348,
"step": 344
},
{
"epoch": 0.9102902374670184,
"grad_norm": 0.176923930644989,
"learning_rate": 5.951955545823342e-05,
"loss": 0.8604757785797119,
"step": 345
},
{
"epoch": 0.9129287598944591,
"grad_norm": 0.17479252815246582,
"learning_rate": 5.93105180990029e-05,
"loss": 0.9862606525421143,
"step": 346
},
{
"epoch": 0.9155672823218998,
"grad_norm": 0.18432128429412842,
"learning_rate": 5.9101312022240106e-05,
"loss": 1.0248565673828125,
"step": 347
},
{
"epoch": 0.9182058047493403,
"grad_norm": 0.18314237892627716,
"learning_rate": 5.889194101900509e-05,
"loss": 0.9174256324768066,
"step": 348
},
{
"epoch": 0.920844327176781,
"grad_norm": 0.16578571498394012,
"learning_rate": 5.868240888334653e-05,
"loss": 1.056714415550232,
"step": 349
},
{
"epoch": 0.9234828496042217,
"grad_norm": 0.15476450324058533,
"learning_rate": 5.847271941223301e-05,
"loss": 0.9293104410171509,
"step": 350
},
{
"epoch": 0.9261213720316622,
"grad_norm": 0.1833016276359558,
"learning_rate": 5.826287640548425e-05,
"loss": 0.9992409348487854,
"step": 351
},
{
"epoch": 0.9287598944591029,
"grad_norm": 0.17842108011245728,
"learning_rate": 5.8052883665702195e-05,
"loss": 1.0228012800216675,
"step": 352
},
{
"epoch": 0.9313984168865436,
"grad_norm": 0.18546055257320404,
"learning_rate": 5.784274499820214e-05,
"loss": 0.9743684530258179,
"step": 353
},
{
"epoch": 0.9340369393139841,
"grad_norm": 0.16940274834632874,
"learning_rate": 5.7632464210943726e-05,
"loss": 0.9181085824966431,
"step": 354
},
{
"epoch": 0.9366754617414248,
"grad_norm": 0.1520531177520752,
"learning_rate": 5.742204511446203e-05,
"loss": 0.9358412623405457,
"step": 355
},
{
"epoch": 0.9393139841688655,
"grad_norm": 0.1511743813753128,
"learning_rate": 5.72114915217984e-05,
"loss": 0.932067334651947,
"step": 356
},
{
"epoch": 0.941952506596306,
"grad_norm": 0.1603623628616333,
"learning_rate": 5.700080724843147e-05,
"loss": 0.8600016832351685,
"step": 357
},
{
"epoch": 0.9445910290237467,
"grad_norm": 0.16928866505622864,
"learning_rate": 5.678999611220786e-05,
"loss": 0.9700044393539429,
"step": 358
},
{
"epoch": 0.9472295514511874,
"grad_norm": 0.1810542345046997,
"learning_rate": 5.657906193327325e-05,
"loss": 0.9956720471382141,
"step": 359
},
{
"epoch": 0.9498680738786279,
"grad_norm": 0.16913573443889618,
"learning_rate": 5.636800853400285e-05,
"loss": 0.9498183727264404,
"step": 360
},
{
"epoch": 0.9525065963060686,
"grad_norm": 0.17175836861133575,
"learning_rate": 5.6156839738932343e-05,
"loss": 0.9880730509757996,
"step": 361
},
{
"epoch": 0.9551451187335093,
"grad_norm": 0.1765371859073639,
"learning_rate": 5.594555937468856e-05,
"loss": 0.9926971197128296,
"step": 362
},
{
"epoch": 0.9577836411609498,
"grad_norm": 0.17631809413433075,
"learning_rate": 5.573417126992003e-05,
"loss": 1.1096311807632446,
"step": 363
},
{
"epoch": 0.9604221635883905,
"grad_norm": 0.18647527694702148,
"learning_rate": 5.5522679255227695e-05,
"loss": 0.9738677740097046,
"step": 364
},
{
"epoch": 0.9630606860158312,
"grad_norm": 0.17723622918128967,
"learning_rate": 5.531108716309547e-05,
"loss": 1.0370627641677856,
"step": 365
},
{
"epoch": 0.9656992084432717,
"grad_norm": 0.18259158730506897,
"learning_rate": 5.509939882782077e-05,
"loss": 1.0807445049285889,
"step": 366
},
{
"epoch": 0.9683377308707124,
"grad_norm": 0.15763477981090546,
"learning_rate": 5.4887618085445094e-05,
"loss": 0.958760142326355,
"step": 367
},
{
"epoch": 0.9709762532981531,
"grad_norm": 0.15909765660762787,
"learning_rate": 5.467574877368441e-05,
"loss": 0.873651921749115,
"step": 368
},
{
"epoch": 0.9736147757255936,
"grad_norm": 0.16833601891994476,
"learning_rate": 5.446379473185972e-05,
"loss": 0.9816885590553284,
"step": 369
},
{
"epoch": 0.9762532981530343,
"grad_norm": 0.15526193380355835,
"learning_rate": 5.42517598008274e-05,
"loss": 0.9115396738052368,
"step": 370
},
{
"epoch": 0.978891820580475,
"grad_norm": 0.16502319276332855,
"learning_rate": 5.4039647822909624e-05,
"loss": 0.8796002864837646,
"step": 371
},
{
"epoch": 0.9815303430079155,
"grad_norm": 0.16136875748634338,
"learning_rate": 5.38274626418248e-05,
"loss": 0.9705801010131836,
"step": 372
},
{
"epoch": 0.9841688654353562,
"grad_norm": 0.15776164829730988,
"learning_rate": 5.361520810261779e-05,
"loss": 0.9282939434051514,
"step": 373
},
{
"epoch": 0.9868073878627969,
"grad_norm": 0.1755094826221466,
"learning_rate": 5.340288805159036e-05,
"loss": 0.8166096806526184,
"step": 374
},
{
"epoch": 0.9894459102902374,
"grad_norm": 0.15046261250972748,
"learning_rate": 5.319050633623142e-05,
"loss": 0.8774311542510986,
"step": 375
},
{
"epoch": 0.9920844327176781,
"grad_norm": 0.17574267089366913,
"learning_rate": 5.297806680514731e-05,
"loss": 0.9345486164093018,
"step": 376
},
{
"epoch": 0.9947229551451188,
"grad_norm": 0.17158278822898865,
"learning_rate": 5.2765573307992036e-05,
"loss": 0.9240652322769165,
"step": 377
},
{
"epoch": 0.9973614775725593,
"grad_norm": 0.17372801899909973,
"learning_rate": 5.2553029695397525e-05,
"loss": 0.9987330436706543,
"step": 378
},
{
"epoch": 1.0,
"grad_norm": 0.18322864174842834,
"learning_rate": 5.234043981890394e-05,
"loss": 0.9055143594741821,
"step": 379
},
{
"epoch": 1.0026385224274406,
"grad_norm": 0.163027822971344,
"learning_rate": 5.212780753088968e-05,
"loss": 0.8689513206481934,
"step": 380
},
{
"epoch": 1.0052770448548813,
"grad_norm": 0.16747920215129852,
"learning_rate": 5.191513668450178e-05,
"loss": 0.8572540283203125,
"step": 381
},
{
"epoch": 1.007915567282322,
"grad_norm": 0.15519443154335022,
"learning_rate": 5.170243113358594e-05,
"loss": 0.7752639651298523,
"step": 382
},
{
"epoch": 1.0105540897097625,
"grad_norm": 0.17219293117523193,
"learning_rate": 5.14896947326168e-05,
"loss": 0.8173640966415405,
"step": 383
},
{
"epoch": 1.0131926121372032,
"grad_norm": 0.18401013314723969,
"learning_rate": 5.127693133662801e-05,
"loss": 0.9488857388496399,
"step": 384
},
{
"epoch": 1.0158311345646438,
"grad_norm": 0.16421173512935638,
"learning_rate": 5.1064144801142374e-05,
"loss": 0.8519411087036133,
"step": 385
},
{
"epoch": 1.0184696569920844,
"grad_norm": 0.16793890297412872,
"learning_rate": 5.0851338982102084e-05,
"loss": 0.8834291696548462,
"step": 386
},
{
"epoch": 1.0211081794195251,
"grad_norm": 0.15045003592967987,
"learning_rate": 5.0638517735798696e-05,
"loss": 0.7745985984802246,
"step": 387
},
{
"epoch": 1.0237467018469657,
"grad_norm": 0.18770800530910492,
"learning_rate": 5.042568491880339e-05,
"loss": 0.8939248919487,
"step": 388
},
{
"epoch": 1.0263852242744063,
"grad_norm": 0.1821233332157135,
"learning_rate": 5.021284438789694e-05,
"loss": 0.8552336096763611,
"step": 389
},
{
"epoch": 1.029023746701847,
"grad_norm": 0.16420002281665802,
"learning_rate": 5e-05,
"loss": 0.8692670464515686,
"step": 390
},
{
"epoch": 1.0316622691292876,
"grad_norm": 0.18629834055900574,
"learning_rate": 4.9787155612103074e-05,
"loss": 0.8282995820045471,
"step": 391
},
{
"epoch": 1.0343007915567282,
"grad_norm": 0.1892436146736145,
"learning_rate": 4.957431508119663e-05,
"loss": 0.8892487287521362,
"step": 392
},
{
"epoch": 1.036939313984169,
"grad_norm": 0.22068382799625397,
"learning_rate": 4.936148226420132e-05,
"loss": 0.8855545520782471,
"step": 393
},
{
"epoch": 1.0395778364116095,
"grad_norm": 0.1730453372001648,
"learning_rate": 4.914866101789792e-05,
"loss": 0.9536787271499634,
"step": 394
},
{
"epoch": 1.04221635883905,
"grad_norm": 0.17769935727119446,
"learning_rate": 4.893585519885764e-05,
"loss": 0.841202974319458,
"step": 395
},
{
"epoch": 1.0448548812664908,
"grad_norm": 0.16897647082805634,
"learning_rate": 4.8723068663372006e-05,
"loss": 0.8201183080673218,
"step": 396
},
{
"epoch": 1.0474934036939314,
"grad_norm": 0.16851696372032166,
"learning_rate": 4.851030526738321e-05,
"loss": 0.7748086452484131,
"step": 397
},
{
"epoch": 1.050131926121372,
"grad_norm": 0.1909930258989334,
"learning_rate": 4.829756886641407e-05,
"loss": 0.9193738698959351,
"step": 398
},
{
"epoch": 1.0527704485488127,
"grad_norm": 0.18687032163143158,
"learning_rate": 4.8084863315498234e-05,
"loss": 0.838584303855896,
"step": 399
},
{
"epoch": 1.0554089709762533,
"grad_norm": 0.1934981346130371,
"learning_rate": 4.787219246911034e-05,
"loss": 0.9033339619636536,
"step": 400
},
{
"epoch": 1.0580474934036939,
"grad_norm": 0.18443354964256287,
"learning_rate": 4.765956018109607e-05,
"loss": 0.8763784766197205,
"step": 401
},
{
"epoch": 1.0606860158311346,
"grad_norm": 0.18538866937160492,
"learning_rate": 4.744697030460248e-05,
"loss": 0.8398501873016357,
"step": 402
},
{
"epoch": 1.0633245382585752,
"grad_norm": 0.16640767455101013,
"learning_rate": 4.723442669200798e-05,
"loss": 0.8269528150558472,
"step": 403
},
{
"epoch": 1.0659630606860158,
"grad_norm": 0.22288216650485992,
"learning_rate": 4.702193319485271e-05,
"loss": 0.8861086368560791,
"step": 404
},
{
"epoch": 1.0686015831134565,
"grad_norm": 0.18035729229450226,
"learning_rate": 4.680949366376858e-05,
"loss": 0.824222981929779,
"step": 405
},
{
"epoch": 1.071240105540897,
"grad_norm": 0.17725308239459991,
"learning_rate": 4.659711194840964e-05,
"loss": 0.9309662580490112,
"step": 406
},
{
"epoch": 1.0738786279683377,
"grad_norm": 0.19792447984218597,
"learning_rate": 4.638479189738224e-05,
"loss": 0.9101674556732178,
"step": 407
},
{
"epoch": 1.0765171503957784,
"grad_norm": 0.19369834661483765,
"learning_rate": 4.6172537358175214e-05,
"loss": 0.8125869631767273,
"step": 408
},
{
"epoch": 1.079155672823219,
"grad_norm": 0.2080707997083664,
"learning_rate": 4.5960352177090395e-05,
"loss": 0.9006462097167969,
"step": 409
},
{
"epoch": 1.0817941952506596,
"grad_norm": 0.1985592097043991,
"learning_rate": 4.5748240199172614e-05,
"loss": 0.8294681310653687,
"step": 410
},
{
"epoch": 1.0844327176781003,
"grad_norm": 0.18108457326889038,
"learning_rate": 4.5536205268140294e-05,
"loss": 0.8876621127128601,
"step": 411
},
{
"epoch": 1.087071240105541,
"grad_norm": 0.21750716865062714,
"learning_rate": 4.532425122631559e-05,
"loss": 0.868391752243042,
"step": 412
},
{
"epoch": 1.0897097625329815,
"grad_norm": 0.19873136281967163,
"learning_rate": 4.511238191455491e-05,
"loss": 0.8306460380554199,
"step": 413
},
{
"epoch": 1.0923482849604222,
"grad_norm": 0.18541795015335083,
"learning_rate": 4.4900601172179244e-05,
"loss": 0.9179568290710449,
"step": 414
},
{
"epoch": 1.0949868073878628,
"grad_norm": 0.18384666740894318,
"learning_rate": 4.468891283690454e-05,
"loss": 0.8310732841491699,
"step": 415
},
{
"epoch": 1.0976253298153034,
"grad_norm": 0.1964699774980545,
"learning_rate": 4.447732074477233e-05,
"loss": 0.9393026828765869,
"step": 416
},
{
"epoch": 1.1002638522427441,
"grad_norm": 0.19017209112644196,
"learning_rate": 4.4265828730079987e-05,
"loss": 0.8362956047058105,
"step": 417
},
{
"epoch": 1.1029023746701847,
"grad_norm": 0.19348829984664917,
"learning_rate": 4.405444062531145e-05,
"loss": 0.7968118190765381,
"step": 418
},
{
"epoch": 1.1055408970976253,
"grad_norm": 0.17324209213256836,
"learning_rate": 4.3843160261067655e-05,
"loss": 0.7850878834724426,
"step": 419
},
{
"epoch": 1.108179419525066,
"grad_norm": 0.17320138216018677,
"learning_rate": 4.363199146599717e-05,
"loss": 0.9237335324287415,
"step": 420
},
{
"epoch": 1.1108179419525066,
"grad_norm": 0.19056344032287598,
"learning_rate": 4.342093806672678e-05,
"loss": 0.8847521543502808,
"step": 421
},
{
"epoch": 1.1134564643799472,
"grad_norm": 0.17850638926029205,
"learning_rate": 4.321000388779214e-05,
"loss": 0.8789336681365967,
"step": 422
},
{
"epoch": 1.116094986807388,
"grad_norm": 0.17765560746192932,
"learning_rate": 4.2999192751568564e-05,
"loss": 0.8255523443222046,
"step": 423
},
{
"epoch": 1.1187335092348285,
"grad_norm": 0.17237840592861176,
"learning_rate": 4.2788508478201606e-05,
"loss": 0.8108619451522827,
"step": 424
},
{
"epoch": 1.121372031662269,
"grad_norm": 0.18996207416057587,
"learning_rate": 4.2577954885537986e-05,
"loss": 0.8163636922836304,
"step": 425
},
{
"epoch": 1.1240105540897098,
"grad_norm": 0.19661268591880798,
"learning_rate": 4.236753578905627e-05,
"loss": 0.8180580139160156,
"step": 426
},
{
"epoch": 1.1266490765171504,
"grad_norm": 0.1983136236667633,
"learning_rate": 4.215725500179787e-05,
"loss": 0.8661148548126221,
"step": 427
},
{
"epoch": 1.129287598944591,
"grad_norm": 0.20327302813529968,
"learning_rate": 4.194711633429782e-05,
"loss": 0.8483480215072632,
"step": 428
},
{
"epoch": 1.1319261213720317,
"grad_norm": 0.18003475666046143,
"learning_rate": 4.1737123594515756e-05,
"loss": 0.7864603996276855,
"step": 429
},
{
"epoch": 1.1345646437994723,
"grad_norm": 0.19587212800979614,
"learning_rate": 4.152728058776701e-05,
"loss": 0.8808155059814453,
"step": 430
},
{
"epoch": 1.1372031662269129,
"grad_norm": 0.19418397545814514,
"learning_rate": 4.131759111665349e-05,
"loss": 0.832161545753479,
"step": 431
},
{
"epoch": 1.1398416886543536,
"grad_norm": 0.2255297303199768,
"learning_rate": 4.110805898099493e-05,
"loss": 0.8426797986030579,
"step": 432
},
{
"epoch": 1.1424802110817942,
"grad_norm": 0.19156090915203094,
"learning_rate": 4.089868797775989e-05,
"loss": 0.8604530096054077,
"step": 433
},
{
"epoch": 1.1451187335092348,
"grad_norm": 0.20741666853427887,
"learning_rate": 4.06894819009971e-05,
"loss": 0.9090075492858887,
"step": 434
},
{
"epoch": 1.1477572559366755,
"grad_norm": 0.19898787140846252,
"learning_rate": 4.0480444541766576e-05,
"loss": 0.8166146874427795,
"step": 435
},
{
"epoch": 1.150395778364116,
"grad_norm": 0.1938183307647705,
"learning_rate": 4.0271579688071e-05,
"loss": 0.7139468193054199,
"step": 436
},
{
"epoch": 1.1530343007915567,
"grad_norm": 0.19528253376483917,
"learning_rate": 4.0062891124787e-05,
"loss": 0.8504555821418762,
"step": 437
},
{
"epoch": 1.1556728232189974,
"grad_norm": 0.1629706174135208,
"learning_rate": 3.9854382633596664e-05,
"loss": 0.7654178142547607,
"step": 438
},
{
"epoch": 1.158311345646438,
"grad_norm": 0.18239827454090118,
"learning_rate": 3.964605799291897e-05,
"loss": 0.8295183777809143,
"step": 439
},
{
"epoch": 1.1609498680738786,
"grad_norm": 0.20417547225952148,
"learning_rate": 3.943792097784127e-05,
"loss": 0.8282760977745056,
"step": 440
},
{
"epoch": 1.1635883905013193,
"grad_norm": 0.16850367188453674,
"learning_rate": 3.922997536005094e-05,
"loss": 0.7826701998710632,
"step": 441
},
{
"epoch": 1.16622691292876,
"grad_norm": 0.23355093598365784,
"learning_rate": 3.9022224907767005e-05,
"loss": 0.8789860010147095,
"step": 442
},
{
"epoch": 1.1688654353562005,
"grad_norm": 0.21100738644599915,
"learning_rate": 3.8814673385671894e-05,
"loss": 0.8798761367797852,
"step": 443
},
{
"epoch": 1.1715039577836412,
"grad_norm": 0.19496478140354156,
"learning_rate": 3.8607324554843136e-05,
"loss": 0.8681943416595459,
"step": 444
},
{
"epoch": 1.1741424802110818,
"grad_norm": 0.2121591866016388,
"learning_rate": 3.840018217268527e-05,
"loss": 0.8384369015693665,
"step": 445
},
{
"epoch": 1.1767810026385224,
"grad_norm": 0.1898183971643448,
"learning_rate": 3.819324999286177e-05,
"loss": 0.7906219959259033,
"step": 446
},
{
"epoch": 1.1794195250659631,
"grad_norm": 0.20216920971870422,
"learning_rate": 3.7986531765226964e-05,
"loss": 0.7790259122848511,
"step": 447
},
{
"epoch": 1.1820580474934037,
"grad_norm": 0.20957563817501068,
"learning_rate": 3.778003123575815e-05,
"loss": 0.7760137915611267,
"step": 448
},
{
"epoch": 1.1846965699208443,
"grad_norm": 0.20613044500350952,
"learning_rate": 3.757375214648764e-05,
"loss": 0.8447249531745911,
"step": 449
},
{
"epoch": 1.187335092348285,
"grad_norm": 0.2005365639925003,
"learning_rate": 3.7367698235435036e-05,
"loss": 0.8000772595405579,
"step": 450
},
{
"epoch": 1.1899736147757256,
"grad_norm": 0.18416504561901093,
"learning_rate": 3.716187323653939e-05,
"loss": 0.8127347230911255,
"step": 451
},
{
"epoch": 1.1926121372031662,
"grad_norm": 0.18120023608207703,
"learning_rate": 3.695628087959162e-05,
"loss": 0.8312699794769287,
"step": 452
},
{
"epoch": 1.195250659630607,
"grad_norm": 0.19914136826992035,
"learning_rate": 3.675092489016693e-05,
"loss": 0.8587290048599243,
"step": 453
},
{
"epoch": 1.1978891820580475,
"grad_norm": 0.18168587982654572,
"learning_rate": 3.6545808989557205e-05,
"loss": 0.8895556926727295,
"step": 454
},
{
"epoch": 1.200527704485488,
"grad_norm": 0.19216057658195496,
"learning_rate": 3.634093689470371e-05,
"loss": 0.8902896642684937,
"step": 455
},
{
"epoch": 1.2031662269129288,
"grad_norm": 0.19391824305057526,
"learning_rate": 3.61363123181296e-05,
"loss": 0.7836971282958984,
"step": 456
},
{
"epoch": 1.2058047493403694,
"grad_norm": 0.22049692273139954,
"learning_rate": 3.5931938967872766e-05,
"loss": 0.9629299640655518,
"step": 457
},
{
"epoch": 1.20844327176781,
"grad_norm": 0.20522050559520721,
"learning_rate": 3.5727820547418524e-05,
"loss": 0.8854247331619263,
"step": 458
},
{
"epoch": 1.2110817941952507,
"grad_norm": 0.19162432849407196,
"learning_rate": 3.5523960755632574e-05,
"loss": 0.8424211740493774,
"step": 459
},
{
"epoch": 1.2137203166226913,
"grad_norm": 0.19570188224315643,
"learning_rate": 3.5320363286694015e-05,
"loss": 0.7317659854888916,
"step": 460
},
{
"epoch": 1.2163588390501319,
"grad_norm": 0.2107311189174652,
"learning_rate": 3.5117031830028274e-05,
"loss": 0.8849686980247498,
"step": 461
},
{
"epoch": 1.2189973614775726,
"grad_norm": 0.20445716381072998,
"learning_rate": 3.4913970070240386e-05,
"loss": 0.8883921504020691,
"step": 462
},
{
"epoch": 1.2216358839050132,
"grad_norm": 0.2084072083234787,
"learning_rate": 3.471118168704811e-05,
"loss": 0.8633065819740295,
"step": 463
},
{
"epoch": 1.2242744063324538,
"grad_norm": 0.2052319049835205,
"learning_rate": 3.450867035521536e-05,
"loss": 0.8934541344642639,
"step": 464
},
{
"epoch": 1.2269129287598945,
"grad_norm": 0.19645145535469055,
"learning_rate": 3.4306439744485454e-05,
"loss": 0.9101189970970154,
"step": 465
},
{
"epoch": 1.229551451187335,
"grad_norm": 0.2135409414768219,
"learning_rate": 3.4104493519514844e-05,
"loss": 0.833043098449707,
"step": 466
},
{
"epoch": 1.2321899736147757,
"grad_norm": 0.18406689167022705,
"learning_rate": 3.390283533980646e-05,
"loss": 0.8629342317581177,
"step": 467
},
{
"epoch": 1.2348284960422165,
"grad_norm": 0.2001754194498062,
"learning_rate": 3.370146885964358e-05,
"loss": 0.873695969581604,
"step": 468
},
{
"epoch": 1.237467018469657,
"grad_norm": 0.2148927003145218,
"learning_rate": 3.350039772802354e-05,
"loss": 0.8796666264533997,
"step": 469
},
{
"epoch": 1.2401055408970976,
"grad_norm": 0.21805380284786224,
"learning_rate": 3.3299625588591564e-05,
"loss": 0.9472431540489197,
"step": 470
},
{
"epoch": 1.2427440633245384,
"grad_norm": 0.1728621870279312,
"learning_rate": 3.309915607957487e-05,
"loss": 0.7903483510017395,
"step": 471
},
{
"epoch": 1.245382585751979,
"grad_norm": 0.1994401216506958,
"learning_rate": 3.289899283371657e-05,
"loss": 0.8357045650482178,
"step": 472
},
{
"epoch": 1.2480211081794195,
"grad_norm": 0.1856645941734314,
"learning_rate": 3.269913947820998e-05,
"loss": 0.8685092926025391,
"step": 473
},
{
"epoch": 1.2506596306068603,
"grad_norm": 0.18104133009910583,
"learning_rate": 3.249959963463283e-05,
"loss": 0.8382564783096313,
"step": 474
},
{
"epoch": 1.2532981530343008,
"grad_norm": 0.1977917104959488,
"learning_rate": 3.2300376918881624e-05,
"loss": 0.7892318964004517,
"step": 475
},
{
"epoch": 1.2559366754617414,
"grad_norm": 0.19494417309761047,
"learning_rate": 3.210147494110618e-05,
"loss": 0.8779850602149963,
"step": 476
},
{
"epoch": 1.258575197889182,
"grad_norm": 0.18986152112483978,
"learning_rate": 3.1902897305644095e-05,
"loss": 0.8703120350837708,
"step": 477
},
{
"epoch": 1.2612137203166227,
"grad_norm": 0.21342100203037262,
"learning_rate": 3.170464761095562e-05,
"loss": 0.8629742860794067,
"step": 478
},
{
"epoch": 1.2638522427440633,
"grad_norm": 0.19701190292835236,
"learning_rate": 3.1506729449558184e-05,
"loss": 0.8151880502700806,
"step": 479
},
{
"epoch": 1.266490765171504,
"grad_norm": 0.19951196014881134,
"learning_rate": 3.130914640796157e-05,
"loss": 0.7900158762931824,
"step": 480
},
{
"epoch": 1.2691292875989446,
"grad_norm": 0.1988476812839508,
"learning_rate": 3.1111902066602724e-05,
"loss": 0.8788620233535767,
"step": 481
},
{
"epoch": 1.2717678100263852,
"grad_norm": 0.2119346261024475,
"learning_rate": 3.091499999978097e-05,
"loss": 0.9024869799613953,
"step": 482
},
{
"epoch": 1.2744063324538257,
"grad_norm": 0.1857406198978424,
"learning_rate": 3.071844377559323e-05,
"loss": 0.8004682064056396,
"step": 483
},
{
"epoch": 1.2770448548812665,
"grad_norm": 0.2080564647912979,
"learning_rate": 3.052223695586929e-05,
"loss": 0.7997844815254211,
"step": 484
},
{
"epoch": 1.279683377308707,
"grad_norm": 0.20487792789936066,
"learning_rate": 3.0326383096107426e-05,
"loss": 0.8369828462600708,
"step": 485
},
{
"epoch": 1.2823218997361479,
"grad_norm": 0.21704550087451935,
"learning_rate": 3.013088574540974e-05,
"loss": 0.8362990617752075,
"step": 486
},
{
"epoch": 1.2849604221635884,
"grad_norm": 0.1796228289604187,
"learning_rate": 2.9935748446418066e-05,
"loss": 0.7657400369644165,
"step": 487
},
{
"epoch": 1.287598944591029,
"grad_norm": 0.19341661036014557,
"learning_rate": 2.9740974735249627e-05,
"loss": 0.8564615845680237,
"step": 488
},
{
"epoch": 1.2902374670184695,
"grad_norm": 0.19471068680286407,
"learning_rate": 2.9546568141433006e-05,
"loss": 0.8282036781311035,
"step": 489
},
{
"epoch": 1.2928759894459103,
"grad_norm": 0.19298681616783142,
"learning_rate": 2.9352532187844252e-05,
"loss": 0.7633144855499268,
"step": 490
},
{
"epoch": 1.2955145118733509,
"grad_norm": 0.2185523808002472,
"learning_rate": 2.915887039064287e-05,
"loss": 0.8637824058532715,
"step": 491
},
{
"epoch": 1.2981530343007917,
"grad_norm": 0.18757079541683197,
"learning_rate": 2.8965586259208295e-05,
"loss": 0.9021022915840149,
"step": 492
},
{
"epoch": 1.3007915567282322,
"grad_norm": 0.22943295538425446,
"learning_rate": 2.8772683296076196e-05,
"loss": 0.7836156487464905,
"step": 493
},
{
"epoch": 1.3034300791556728,
"grad_norm": 0.22361497581005096,
"learning_rate": 2.858016499687503e-05,
"loss": 0.8892571926116943,
"step": 494
},
{
"epoch": 1.3060686015831133,
"grad_norm": 0.17517295479774475,
"learning_rate": 2.8388034850262646e-05,
"loss": 0.7846842408180237,
"step": 495
},
{
"epoch": 1.3087071240105541,
"grad_norm": 0.21480333805084229,
"learning_rate": 2.8196296337863193e-05,
"loss": 0.8207566142082214,
"step": 496
},
{
"epoch": 1.3113456464379947,
"grad_norm": 0.20812678337097168,
"learning_rate": 2.8004952934203838e-05,
"loss": 0.8578101396560669,
"step": 497
},
{
"epoch": 1.3139841688654355,
"grad_norm": 0.20664533972740173,
"learning_rate": 2.7814008106652012e-05,
"loss": 0.8509771823883057,
"step": 498
},
{
"epoch": 1.316622691292876,
"grad_norm": 0.21508260071277618,
"learning_rate": 2.762346531535246e-05,
"loss": 0.8185523748397827,
"step": 499
},
{
"epoch": 1.3192612137203166,
"grad_norm": 0.1851697713136673,
"learning_rate": 2.7433328013164495e-05,
"loss": 0.7895690202713013,
"step": 500
},
{
"epoch": 1.3218997361477571,
"grad_norm": 0.21966880559921265,
"learning_rate": 2.7243599645599576e-05,
"loss": 0.8147685527801514,
"step": 501
},
{
"epoch": 1.324538258575198,
"grad_norm": 0.2082214504480362,
"learning_rate": 2.7054283650758682e-05,
"loss": 0.9062052369117737,
"step": 502
},
{
"epoch": 1.3271767810026385,
"grad_norm": 0.18478548526763916,
"learning_rate": 2.6865383459270265e-05,
"loss": 0.8946532011032104,
"step": 503
},
{
"epoch": 1.3298153034300793,
"grad_norm": 0.2242600917816162,
"learning_rate": 2.6676902494227795e-05,
"loss": 0.8830220699310303,
"step": 504
},
{
"epoch": 1.3324538258575198,
"grad_norm": 0.22042159736156464,
"learning_rate": 2.6488844171127903e-05,
"loss": 0.8637486696243286,
"step": 505
},
{
"epoch": 1.3350923482849604,
"grad_norm": 0.19908133149147034,
"learning_rate": 2.6301211897808464e-05,
"loss": 0.9240211248397827,
"step": 506
},
{
"epoch": 1.337730870712401,
"grad_norm": 0.21462836861610413,
"learning_rate": 2.6114009074386846e-05,
"loss": 0.8997042179107666,
"step": 507
},
{
"epoch": 1.3403693931398417,
"grad_norm": 0.2102992832660675,
"learning_rate": 2.5927239093198274e-05,
"loss": 0.8191858530044556,
"step": 508
},
{
"epoch": 1.3430079155672823,
"grad_norm": 0.21195337176322937,
"learning_rate": 2.574090533873431e-05,
"loss": 0.8745005130767822,
"step": 509
},
{
"epoch": 1.345646437994723,
"grad_norm": 0.17979484796524048,
"learning_rate": 2.555501118758167e-05,
"loss": 0.8337305784225464,
"step": 510
},
{
"epoch": 1.3482849604221636,
"grad_norm": 0.19970622658729553,
"learning_rate": 2.5369560008360828e-05,
"loss": 0.8389545679092407,
"step": 511
},
{
"epoch": 1.3509234828496042,
"grad_norm": 0.2224525660276413,
"learning_rate": 2.518455516166517e-05,
"loss": 0.9300051927566528,
"step": 512
},
{
"epoch": 1.3535620052770447,
"grad_norm": 0.2070772796869278,
"learning_rate": 2.500000000000001e-05,
"loss": 0.8477452993392944,
"step": 513
},
{
"epoch": 1.3562005277044855,
"grad_norm": 0.20911110937595367,
"learning_rate": 2.4815897867721784e-05,
"loss": 0.8734545707702637,
"step": 514
},
{
"epoch": 1.358839050131926,
"grad_norm": 0.21647144854068756,
"learning_rate": 2.4632252100977566e-05,
"loss": 0.8972655534744263,
"step": 515
},
{
"epoch": 1.3614775725593669,
"grad_norm": 0.21526023745536804,
"learning_rate": 2.4449066027644475e-05,
"loss": 0.8118870258331299,
"step": 516
},
{
"epoch": 1.3641160949868074,
"grad_norm": 0.20327945053577423,
"learning_rate": 2.4266342967269552e-05,
"loss": 0.855424165725708,
"step": 517
},
{
"epoch": 1.366754617414248,
"grad_norm": 0.19717800617218018,
"learning_rate": 2.4084086231009395e-05,
"loss": 0.7565803527832031,
"step": 518
},
{
"epoch": 1.3693931398416885,
"grad_norm": 0.20523078739643097,
"learning_rate": 2.3902299121570333e-05,
"loss": 0.9261975288391113,
"step": 519
},
{
"epoch": 1.3720316622691293,
"grad_norm": 0.219309002161026,
"learning_rate": 2.3720984933148444e-05,
"loss": 0.7988795042037964,
"step": 520
},
{
"epoch": 1.3746701846965699,
"grad_norm": 0.24325881898403168,
"learning_rate": 2.354014695136997e-05,
"loss": 0.8413968086242676,
"step": 521
},
{
"epoch": 1.3773087071240107,
"grad_norm": 0.19859999418258667,
"learning_rate": 2.3359788453231724e-05,
"loss": 0.8811598420143127,
"step": 522
},
{
"epoch": 1.3799472295514512,
"grad_norm": 0.2395762950181961,
"learning_rate": 2.317991270704167e-05,
"loss": 0.8950267434120178,
"step": 523
},
{
"epoch": 1.3825857519788918,
"grad_norm": 0.17747214436531067,
"learning_rate": 2.30005229723598e-05,
"loss": 0.7522149085998535,
"step": 524
},
{
"epoch": 1.3852242744063323,
"grad_norm": 0.2008499652147293,
"learning_rate": 2.282162249993895e-05,
"loss": 0.727147102355957,
"step": 525
},
{
"epoch": 1.3878627968337731,
"grad_norm": 0.20609167218208313,
"learning_rate": 2.2643214531665978e-05,
"loss": 0.8256455659866333,
"step": 526
},
{
"epoch": 1.3905013192612137,
"grad_norm": 0.20032787322998047,
"learning_rate": 2.246530230050301e-05,
"loss": 0.8437114357948303,
"step": 527
},
{
"epoch": 1.3931398416886545,
"grad_norm": 0.21618762612342834,
"learning_rate": 2.228788903042877e-05,
"loss": 0.8061575293540955,
"step": 528
},
{
"epoch": 1.395778364116095,
"grad_norm": 0.2020416408777237,
"learning_rate": 2.211097793638029e-05,
"loss": 0.7951303124427795,
"step": 529
},
{
"epoch": 1.3984168865435356,
"grad_norm": 0.21808432042598724,
"learning_rate": 2.19345722241945e-05,
"loss": 0.893155038356781,
"step": 530
},
{
"epoch": 1.4010554089709761,
"grad_norm": 0.20245902240276337,
"learning_rate": 2.175867509055033e-05,
"loss": 0.8358583450317383,
"step": 531
},
{
"epoch": 1.403693931398417,
"grad_norm": 0.20912860333919525,
"learning_rate": 2.158328972291056e-05,
"loss": 0.8978977799415588,
"step": 532
},
{
"epoch": 1.4063324538258575,
"grad_norm": 0.22441457211971283,
"learning_rate": 2.1408419299464245e-05,
"loss": 0.861559271812439,
"step": 533
},
{
"epoch": 1.4089709762532983,
"grad_norm": 0.19903819262981415,
"learning_rate": 2.1234066989068972e-05,
"loss": 0.8497825860977173,
"step": 534
},
{
"epoch": 1.4116094986807388,
"grad_norm": 0.2019970715045929,
"learning_rate": 2.106023595119358e-05,
"loss": 0.8910925388336182,
"step": 535
},
{
"epoch": 1.4142480211081794,
"grad_norm": 0.19506564736366272,
"learning_rate": 2.0886929335860826e-05,
"loss": 0.8591622710227966,
"step": 536
},
{
"epoch": 1.41688654353562,
"grad_norm": 0.19065196812152863,
"learning_rate": 2.071415028359026e-05,
"loss": 0.8114444017410278,
"step": 537
},
{
"epoch": 1.4195250659630607,
"grad_norm": 0.20525790750980377,
"learning_rate": 2.0541901925341445e-05,
"loss": 0.8627693057060242,
"step": 538
},
{
"epoch": 1.4221635883905013,
"grad_norm": 0.1845739483833313,
"learning_rate": 2.0370187382457068e-05,
"loss": 0.784498929977417,
"step": 539
},
{
"epoch": 1.424802110817942,
"grad_norm": 0.19489313662052155,
"learning_rate": 2.0199009766606507e-05,
"loss": 0.8950122594833374,
"step": 540
},
{
"epoch": 1.4274406332453826,
"grad_norm": 0.19720782339572906,
"learning_rate": 2.0028372179729403e-05,
"loss": 0.8929563760757446,
"step": 541
},
{
"epoch": 1.4300791556728232,
"grad_norm": 0.2457919865846634,
"learning_rate": 1.985827771397938e-05,
"loss": 0.8557782173156738,
"step": 542
},
{
"epoch": 1.4327176781002637,
"grad_norm": 0.212116539478302,
"learning_rate": 1.9688729451668114e-05,
"loss": 0.8671682476997375,
"step": 543
},
{
"epoch": 1.4353562005277045,
"grad_norm": 0.20607347786426544,
"learning_rate": 1.9519730465209384e-05,
"loss": 0.8468962907791138,
"step": 544
},
{
"epoch": 1.437994722955145,
"grad_norm": 0.19764231145381927,
"learning_rate": 1.935128381706355e-05,
"loss": 0.8730595707893372,
"step": 545
},
{
"epoch": 1.4406332453825859,
"grad_norm": 0.20810900628566742,
"learning_rate": 1.9183392559681812e-05,
"loss": 0.905620813369751,
"step": 546
},
{
"epoch": 1.4432717678100264,
"grad_norm": 0.20794421434402466,
"learning_rate": 1.901605973545116e-05,
"loss": 0.8467898964881897,
"step": 547
},
{
"epoch": 1.445910290237467,
"grad_norm": 0.1914779394865036,
"learning_rate": 1.8849288376639017e-05,
"loss": 0.8068687915802002,
"step": 548
},
{
"epoch": 1.4485488126649075,
"grad_norm": 0.21812552213668823,
"learning_rate": 1.868308150533847e-05,
"loss": 0.8541518449783325,
"step": 549
},
{
"epoch": 1.4511873350923483,
"grad_norm": 0.2212463617324829,
"learning_rate": 1.8517442133413403e-05,
"loss": 0.8813215494155884,
"step": 550
},
{
"epoch": 1.4538258575197889,
"grad_norm": 0.22319260239601135,
"learning_rate": 1.8352373262443916e-05,
"loss": 0.9005920886993408,
"step": 551
},
{
"epoch": 1.4564643799472297,
"grad_norm": 0.22040770947933197,
"learning_rate": 1.818787788367202e-05,
"loss": 0.8013516664505005,
"step": 552
},
{
"epoch": 1.4591029023746702,
"grad_norm": 0.21650803089141846,
"learning_rate": 1.8023958977947304e-05,
"loss": 0.8534271717071533,
"step": 553
},
{
"epoch": 1.4617414248021108,
"grad_norm": 0.2045343667268753,
"learning_rate": 1.7860619515673033e-05,
"loss": 0.7753493785858154,
"step": 554
},
{
"epoch": 1.4643799472295513,
"grad_norm": 0.20727676153182983,
"learning_rate": 1.7697862456752273e-05,
"loss": 0.8886493444442749,
"step": 555
},
{
"epoch": 1.4670184696569921,
"grad_norm": 0.19541321694850922,
"learning_rate": 1.7535690750534266e-05,
"loss": 0.7950649261474609,
"step": 556
},
{
"epoch": 1.4696569920844327,
"grad_norm": 0.21428488194942474,
"learning_rate": 1.7374107335760936e-05,
"loss": 0.8809351921081543,
"step": 557
},
{
"epoch": 1.4722955145118735,
"grad_norm": 0.1980544626712799,
"learning_rate": 1.7213115140513686e-05,
"loss": 0.7901041507720947,
"step": 558
},
{
"epoch": 1.474934036939314,
"grad_norm": 0.19511975347995758,
"learning_rate": 1.7052717082160346e-05,
"loss": 0.8931927680969238,
"step": 559
},
{
"epoch": 1.4775725593667546,
"grad_norm": 0.20257903635501862,
"learning_rate": 1.689291606730228e-05,
"loss": 0.7760272026062012,
"step": 560
},
{
"epoch": 1.4802110817941951,
"grad_norm": 0.2089790254831314,
"learning_rate": 1.673371499172174e-05,
"loss": 0.8708549737930298,
"step": 561
},
{
"epoch": 1.482849604221636,
"grad_norm": 0.2033015638589859,
"learning_rate": 1.6575116740329317e-05,
"loss": 0.8020316362380981,
"step": 562
},
{
"epoch": 1.4854881266490765,
"grad_norm": 0.19399622082710266,
"learning_rate": 1.6417124187111775e-05,
"loss": 0.8268682956695557,
"step": 563
},
{
"epoch": 1.4881266490765173,
"grad_norm": 0.2089431881904602,
"learning_rate": 1.6259740195079903e-05,
"loss": 0.8855108618736267,
"step": 564
},
{
"epoch": 1.4907651715039578,
"grad_norm": 0.2224927544593811,
"learning_rate": 1.610296761621662e-05,
"loss": 0.825925350189209,
"step": 565
},
{
"epoch": 1.4934036939313984,
"grad_norm": 0.22050341963768005,
"learning_rate": 1.594680929142535e-05,
"loss": 0.896023690700531,
"step": 566
},
{
"epoch": 1.496042216358839,
"grad_norm": 0.193280890583992,
"learning_rate": 1.5791268050478486e-05,
"loss": 0.7469483017921448,
"step": 567
},
{
"epoch": 1.4986807387862797,
"grad_norm": 0.20433108508586884,
"learning_rate": 1.5636346711966155e-05,
"loss": 0.8419564962387085,
"step": 568
},
{
"epoch": 1.5013192612137203,
"grad_norm": 0.2134133279323578,
"learning_rate": 1.5482048083245114e-05,
"loss": 0.8411462903022766,
"step": 569
},
{
"epoch": 1.503957783641161,
"grad_norm": 0.2517402470111847,
"learning_rate": 1.532837496038792e-05,
"loss": 0.8998865485191345,
"step": 570
},
{
"epoch": 1.5065963060686016,
"grad_norm": 0.19719058275222778,
"learning_rate": 1.517533012813217e-05,
"loss": 0.790407121181488,
"step": 571
},
{
"epoch": 1.5092348284960422,
"grad_norm": 0.19815827906131744,
"learning_rate": 1.5022916359830113e-05,
"loss": 0.827681303024292,
"step": 572
},
{
"epoch": 1.5118733509234827,
"grad_norm": 0.20996305346488953,
"learning_rate": 1.4871136417398406e-05,
"loss": 0.8366416692733765,
"step": 573
},
{
"epoch": 1.5145118733509235,
"grad_norm": 0.20357419550418854,
"learning_rate": 1.4719993051268022e-05,
"loss": 0.7767808437347412,
"step": 574
},
{
"epoch": 1.517150395778364,
"grad_norm": 0.19412170350551605,
"learning_rate": 1.4569489000334436e-05,
"loss": 0.8966915011405945,
"step": 575
},
{
"epoch": 1.5197889182058049,
"grad_norm": 0.21949054300785065,
"learning_rate": 1.4419626991907925e-05,
"loss": 0.9867626428604126,
"step": 576
},
{
"epoch": 1.5224274406332454,
"grad_norm": 0.21542461216449738,
"learning_rate": 1.427040974166427e-05,
"loss": 0.7484675645828247,
"step": 577
},
{
"epoch": 1.525065963060686,
"grad_norm": 0.20638151466846466,
"learning_rate": 1.412183995359544e-05,
"loss": 0.8688530921936035,
"step": 578
},
{
"epoch": 1.5277044854881265,
"grad_norm": 0.21876366436481476,
"learning_rate": 1.3973920319960655e-05,
"loss": 0.9383907318115234,
"step": 579
},
{
"epoch": 1.5303430079155673,
"grad_norm": 0.21015901863574982,
"learning_rate": 1.3826653521237525e-05,
"loss": 0.7594534754753113,
"step": 580
},
{
"epoch": 1.5329815303430079,
"grad_norm": 0.19950366020202637,
"learning_rate": 1.3680042226073552e-05,
"loss": 0.8834112882614136,
"step": 581
},
{
"epoch": 1.5356200527704487,
"grad_norm": 0.21755766868591309,
"learning_rate": 1.3534089091237755e-05,
"loss": 0.8213707208633423,
"step": 582
},
{
"epoch": 1.5382585751978892,
"grad_norm": 0.19696074724197388,
"learning_rate": 1.3388796761572492e-05,
"loss": 0.9167644381523132,
"step": 583
},
{
"epoch": 1.5408970976253298,
"grad_norm": 0.21788553893566132,
"learning_rate": 1.3244167869945589e-05,
"loss": 0.931348443031311,
"step": 584
},
{
"epoch": 1.5435356200527703,
"grad_norm": 0.2014394998550415,
"learning_rate": 1.310020503720254e-05,
"loss": 0.8700617551803589,
"step": 585
},
{
"epoch": 1.5461741424802111,
"grad_norm": 0.21630780398845673,
"learning_rate": 1.295691087211912e-05,
"loss": 0.8227249979972839,
"step": 586
},
{
"epoch": 1.5488126649076517,
"grad_norm": 0.18821114301681519,
"learning_rate": 1.2814287971354022e-05,
"loss": 0.8128796815872192,
"step": 587
},
{
"epoch": 1.5514511873350925,
"grad_norm": 0.2557939887046814,
"learning_rate": 1.2672338919401866e-05,
"loss": 0.8653614521026611,
"step": 588
},
{
"epoch": 1.554089709762533,
"grad_norm": 0.1947825402021408,
"learning_rate": 1.253106628854635e-05,
"loss": 0.9020968675613403,
"step": 589
},
{
"epoch": 1.5567282321899736,
"grad_norm": 0.21898962557315826,
"learning_rate": 1.2390472638813571e-05,
"loss": 0.7656817436218262,
"step": 590
},
{
"epoch": 1.5593667546174141,
"grad_norm": 0.21510836482048035,
"learning_rate": 1.2250560517925746e-05,
"loss": 0.8295509815216064,
"step": 591
},
{
"epoch": 1.562005277044855,
"grad_norm": 0.2284301370382309,
"learning_rate": 1.211133246125497e-05,
"loss": 0.8284837603569031,
"step": 592
},
{
"epoch": 1.5646437994722955,
"grad_norm": 0.20992207527160645,
"learning_rate": 1.1972790991777311e-05,
"loss": 0.8653608560562134,
"step": 593
},
{
"epoch": 1.5672823218997363,
"grad_norm": 0.22688503563404083,
"learning_rate": 1.183493862002702e-05,
"loss": 0.8686569929122925,
"step": 594
},
{
"epoch": 1.5699208443271768,
"grad_norm": 0.19538283348083496,
"learning_rate": 1.1697777844051105e-05,
"loss": 0.7942227125167847,
"step": 595
},
{
"epoch": 1.5725593667546174,
"grad_norm": 0.21061819791793823,
"learning_rate": 1.1561311149364073e-05,
"loss": 0.84868323802948,
"step": 596
},
{
"epoch": 1.575197889182058,
"grad_norm": 0.21847675740718842,
"learning_rate": 1.1425541008902851e-05,
"loss": 0.8898189067840576,
"step": 597
},
{
"epoch": 1.5778364116094987,
"grad_norm": 0.20205287635326385,
"learning_rate": 1.1290469882981986e-05,
"loss": 0.6871122121810913,
"step": 598
},
{
"epoch": 1.5804749340369393,
"grad_norm": 0.209647536277771,
"learning_rate": 1.1156100219249022e-05,
"loss": 0.7839260101318359,
"step": 599
},
{
"epoch": 1.58311345646438,
"grad_norm": 0.19462355971336365,
"learning_rate": 1.102243445264025e-05,
"loss": 0.7594859600067139,
"step": 600
},
{
"epoch": 1.5857519788918206,
"grad_norm": 0.20564399659633636,
"learning_rate": 1.0889475005336446e-05,
"loss": 0.9194256067276001,
"step": 601
},
{
"epoch": 1.5883905013192612,
"grad_norm": 0.2016313374042511,
"learning_rate": 1.075722428671911e-05,
"loss": 0.8529200553894043,
"step": 602
},
{
"epoch": 1.5910290237467017,
"grad_norm": 0.22161997854709625,
"learning_rate": 1.0625684693326727e-05,
"loss": 0.8794209957122803,
"step": 603
},
{
"epoch": 1.5936675461741425,
"grad_norm": 0.19965285062789917,
"learning_rate": 1.0494858608811326e-05,
"loss": 0.7318451404571533,
"step": 604
},
{
"epoch": 1.596306068601583,
"grad_norm": 0.21324901282787323,
"learning_rate": 1.036474840389537e-05,
"loss": 0.7938931584358215,
"step": 605
},
{
"epoch": 1.5989445910290239,
"grad_norm": 0.2236800491809845,
"learning_rate": 1.0235356436328675e-05,
"loss": 0.9312845468521118,
"step": 606
},
{
"epoch": 1.6015831134564644,
"grad_norm": 0.20095610618591309,
"learning_rate": 1.0106685050845838e-05,
"loss": 0.9334725737571716,
"step": 607
},
{
"epoch": 1.604221635883905,
"grad_norm": 0.24465373158454895,
"learning_rate": 9.978736579123577e-06,
"loss": 0.9391512274742126,
"step": 608
},
{
"epoch": 1.6068601583113455,
"grad_norm": 0.21828067302703857,
"learning_rate": 9.851513339738628e-06,
"loss": 0.9064762592315674,
"step": 609
},
{
"epoch": 1.6094986807387863,
"grad_norm": 0.21165606379508972,
"learning_rate": 9.72501763812561e-06,
"loss": 0.8413673639297485,
"step": 610
},
{
"epoch": 1.6121372031662269,
"grad_norm": 0.19530081748962402,
"learning_rate": 9.599251766535345e-06,
"loss": 0.8129094839096069,
"step": 611
},
{
"epoch": 1.6147757255936677,
"grad_norm": 0.19908255338668823,
"learning_rate": 9.474218003993273e-06,
"loss": 0.7808138132095337,
"step": 612
},
{
"epoch": 1.6174142480211082,
"grad_norm": 0.1975618451833725,
"learning_rate": 9.349918616258114e-06,
"loss": 0.7855852842330933,
"step": 613
},
{
"epoch": 1.6200527704485488,
"grad_norm": 0.20149685442447662,
"learning_rate": 9.226355855780921e-06,
"loss": 0.774482011795044,
"step": 614
},
{
"epoch": 1.6226912928759893,
"grad_norm": 0.21338312327861786,
"learning_rate": 9.103531961664118e-06,
"loss": 0.8815280199050903,
"step": 615
},
{
"epoch": 1.6253298153034301,
"grad_norm": 0.2008296102285385,
"learning_rate": 8.981449159621074e-06,
"loss": 0.8243716359138489,
"step": 616
},
{
"epoch": 1.6279683377308707,
"grad_norm": 0.21292272210121155,
"learning_rate": 8.860109661935674e-06,
"loss": 0.8072988390922546,
"step": 617
},
{
"epoch": 1.6306068601583115,
"grad_norm": 0.23465301096439362,
"learning_rate": 8.73951566742221e-06,
"loss": 0.8511773347854614,
"step": 618
},
{
"epoch": 1.633245382585752,
"grad_norm": 0.46375322341918945,
"learning_rate": 8.619669361385663e-06,
"loss": 0.853812575340271,
"step": 619
},
{
"epoch": 1.6358839050131926,
"grad_norm": 0.18248288333415985,
"learning_rate": 8.500572915581922e-06,
"loss": 0.7355204224586487,
"step": 620
},
{
"epoch": 1.6385224274406331,
"grad_norm": 0.18261535465717316,
"learning_rate": 8.38222848817864e-06,
"loss": 0.8264297246932983,
"step": 621
},
{
"epoch": 1.641160949868074,
"grad_norm": 0.20754438638687134,
"learning_rate": 8.264638223715916e-06,
"loss": 0.8225191235542297,
"step": 622
},
{
"epoch": 1.6437994722955145,
"grad_norm": 0.20159278810024261,
"learning_rate": 8.14780425306758e-06,
"loss": 0.87016361951828,
"step": 623
},
{
"epoch": 1.6464379947229553,
"grad_norm": 0.22935034334659576,
"learning_rate": 8.031728693402502e-06,
"loss": 0.8048298358917236,
"step": 624
},
{
"epoch": 1.6490765171503958,
"grad_norm": 0.18488283455371857,
"learning_rate": 7.91641364814628e-06,
"loss": 0.8106614351272583,
"step": 625
},
{
"epoch": 1.6517150395778364,
"grad_norm": 0.21160657703876495,
"learning_rate": 7.801861206943089e-06,
"loss": 0.8310866951942444,
"step": 626
},
{
"epoch": 1.654353562005277,
"grad_norm": 0.1970130354166031,
"learning_rate": 7.688073445617799e-06,
"loss": 0.7170044779777527,
"step": 627
},
{
"epoch": 1.6569920844327177,
"grad_norm": 0.20660647749900818,
"learning_rate": 7.5750524261384235e-06,
"loss": 0.7831541299819946,
"step": 628
},
{
"epoch": 1.6596306068601583,
"grad_norm": 0.20981073379516602,
"learning_rate": 7.462800196578662e-06,
"loss": 0.8566669225692749,
"step": 629
},
{
"epoch": 1.662269129287599,
"grad_norm": 0.23708082735538483,
"learning_rate": 7.35131879108088e-06,
"loss": 0.8529813289642334,
"step": 630
},
{
"epoch": 1.6649076517150396,
"grad_norm": 0.9549576640129089,
"learning_rate": 7.240610229819195e-06,
"loss": 0.8493508100509644,
"step": 631
},
{
"epoch": 1.6675461741424802,
"grad_norm": 0.2114357352256775,
"learning_rate": 7.130676518962859e-06,
"loss": 0.8748669624328613,
"step": 632
},
{
"epoch": 1.6701846965699207,
"grad_norm": 0.19402459263801575,
"learning_rate": 7.0215196506399515e-06,
"loss": 0.8187092542648315,
"step": 633
},
{
"epoch": 1.6728232189973615,
"grad_norm": 0.20854152739048004,
"learning_rate": 6.913141602901213e-06,
"loss": 0.9096065163612366,
"step": 634
},
{
"epoch": 1.675461741424802,
"grad_norm": 0.21093755960464478,
"learning_rate": 6.8055443396842945e-06,
"loss": 0.8709210157394409,
"step": 635
},
{
"epoch": 1.6781002638522429,
"grad_norm": 0.21466106176376343,
"learning_rate": 6.698729810778065e-06,
"loss": 0.8059540390968323,
"step": 636
},
{
"epoch": 1.6807387862796834,
"grad_norm": 0.19374695420265198,
"learning_rate": 6.592699951787362e-06,
"loss": 0.8607715368270874,
"step": 637
},
{
"epoch": 1.683377308707124,
"grad_norm": 0.2216903418302536,
"learning_rate": 6.487456684097848e-06,
"loss": 0.8703226447105408,
"step": 638
},
{
"epoch": 1.6860158311345645,
"grad_norm": 0.24131982028484344,
"learning_rate": 6.3830019148412525e-06,
"loss": 0.8075211048126221,
"step": 639
},
{
"epoch": 1.6886543535620053,
"grad_norm": 0.19988761842250824,
"learning_rate": 6.279337536860785e-06,
"loss": 0.8634479641914368,
"step": 640
},
{
"epoch": 1.6912928759894459,
"grad_norm": 0.21736286580562592,
"learning_rate": 6.17646542867682e-06,
"loss": 0.8513948321342468,
"step": 641
},
{
"epoch": 1.6939313984168867,
"grad_norm": 0.21482457220554352,
"learning_rate": 6.07438745445289e-06,
"loss": 0.8792531490325928,
"step": 642
},
{
"epoch": 1.6965699208443272,
"grad_norm": 0.20397554337978363,
"learning_rate": 5.973105463961865e-06,
"loss": 0.8307263851165771,
"step": 643
},
{
"epoch": 1.6992084432717678,
"grad_norm": 0.20772746205329895,
"learning_rate": 5.872621292552477e-06,
"loss": 0.8588171005249023,
"step": 644
},
{
"epoch": 1.7018469656992083,
"grad_norm": 0.21618413925170898,
"learning_rate": 5.772936761116027e-06,
"loss": 0.8800046443939209,
"step": 645
},
{
"epoch": 1.7044854881266491,
"grad_norm": 0.20802974700927734,
"learning_rate": 5.6740536760534144e-06,
"loss": 0.9347285032272339,
"step": 646
},
{
"epoch": 1.7071240105540897,
"grad_norm": 0.22136233747005463,
"learning_rate": 5.575973829242364e-06,
"loss": 0.8545072078704834,
"step": 647
},
{
"epoch": 1.7097625329815305,
"grad_norm": 0.2067175656557083,
"learning_rate": 5.478698998004967e-06,
"loss": 0.8864841461181641,
"step": 648
},
{
"epoch": 1.712401055408971,
"grad_norm": 0.23648934066295624,
"learning_rate": 5.382230945075556e-06,
"loss": 0.86419677734375,
"step": 649
},
{
"epoch": 1.7150395778364116,
"grad_norm": 0.20306909084320068,
"learning_rate": 5.286571418568615e-06,
"loss": 0.8809700012207031,
"step": 650
},
{
"epoch": 1.7176781002638521,
"grad_norm": 0.2015990912914276,
"learning_rate": 5.191722151947226e-06,
"loss": 0.7576265931129456,
"step": 651
},
{
"epoch": 1.720316622691293,
"grad_norm": 0.20526087284088135,
"learning_rate": 5.097684863991575e-06,
"loss": 0.8124517798423767,
"step": 652
},
{
"epoch": 1.7229551451187335,
"grad_norm": 0.21329709887504578,
"learning_rate": 5.004461258767873e-06,
"loss": 0.8556035757064819,
"step": 653
},
{
"epoch": 1.7255936675461743,
"grad_norm": 0.21723578870296478,
"learning_rate": 4.912053025597429e-06,
"loss": 0.8384562134742737,
"step": 654
},
{
"epoch": 1.7282321899736148,
"grad_norm": 0.19574065506458282,
"learning_rate": 4.820461839026047e-06,
"loss": 0.8278510570526123,
"step": 655
},
{
"epoch": 1.7308707124010554,
"grad_norm": 0.1882028728723526,
"learning_rate": 4.729689358793693e-06,
"loss": 0.8415632843971252,
"step": 656
},
{
"epoch": 1.733509234828496,
"grad_norm": 0.22890862822532654,
"learning_rate": 4.639737229804403e-06,
"loss": 0.8932334184646606,
"step": 657
},
{
"epoch": 1.7361477572559367,
"grad_norm": 0.2388426512479782,
"learning_rate": 4.5506070820964976e-06,
"loss": 0.7936335802078247,
"step": 658
},
{
"epoch": 1.7387862796833773,
"grad_norm": 0.2018558830022812,
"learning_rate": 4.462300530813024e-06,
"loss": 0.7627620100975037,
"step": 659
},
{
"epoch": 1.741424802110818,
"grad_norm": 0.2154654711484909,
"learning_rate": 4.374819176172501e-06,
"loss": 0.8787714838981628,
"step": 660
},
{
"epoch": 1.7440633245382586,
"grad_norm": 0.2038903832435608,
"learning_rate": 4.2881646034398925e-06,
"loss": 0.8130042552947998,
"step": 661
},
{
"epoch": 1.7467018469656992,
"grad_norm": 0.21601106226444244,
"learning_rate": 4.20233838289793e-06,
"loss": 0.835135281085968,
"step": 662
},
{
"epoch": 1.7493403693931397,
"grad_norm": 0.2100323736667633,
"learning_rate": 4.117342069818603e-06,
"loss": 0.8817960023880005,
"step": 663
},
{
"epoch": 1.7519788918205803,
"grad_norm": 0.20168212056159973,
"learning_rate": 4.033177204435024e-06,
"loss": 0.8514724969863892,
"step": 664
},
{
"epoch": 1.754617414248021,
"grad_norm": 0.2164795845746994,
"learning_rate": 3.949845311913492e-06,
"loss": 0.8425164818763733,
"step": 665
},
{
"epoch": 1.7572559366754619,
"grad_norm": 0.19507786631584167,
"learning_rate": 3.8673479023258465e-06,
"loss": 0.8335729241371155,
"step": 666
},
{
"epoch": 1.7598944591029024,
"grad_norm": 0.23975177109241486,
"learning_rate": 3.7856864706221185e-06,
"loss": 0.8719772696495056,
"step": 667
},
{
"epoch": 1.762532981530343,
"grad_norm": 0.21782037615776062,
"learning_rate": 3.7048624966034506e-06,
"loss": 0.8989812135696411,
"step": 668
},
{
"epoch": 1.7651715039577835,
"grad_norm": 0.1859140247106552,
"learning_rate": 3.6248774448952695e-06,
"loss": 0.8056320548057556,
"step": 669
},
{
"epoch": 1.767810026385224,
"grad_norm": 0.19920724630355835,
"learning_rate": 3.5457327649207172e-06,
"loss": 0.7947468757629395,
"step": 670
},
{
"epoch": 1.770448548812665,
"grad_norm": 0.19092382490634918,
"learning_rate": 3.467429890874424e-06,
"loss": 0.7448170781135559,
"step": 671
},
{
"epoch": 1.7730870712401057,
"grad_norm": 0.22142529487609863,
"learning_rate": 3.389970241696516e-06,
"loss": 0.9054186344146729,
"step": 672
},
{
"epoch": 1.7757255936675462,
"grad_norm": 0.22562207281589508,
"learning_rate": 3.3133552210468875e-06,
"loss": 0.9187839031219482,
"step": 673
},
{
"epoch": 1.7783641160949868,
"grad_norm": 0.21036329865455627,
"learning_rate": 3.2375862172797864e-06,
"loss": 0.7974717020988464,
"step": 674
},
{
"epoch": 1.7810026385224274,
"grad_norm": 0.21586009860038757,
"learning_rate": 3.162664603418608e-06,
"loss": 0.9025945663452148,
"step": 675
},
{
"epoch": 1.783641160949868,
"grad_norm": 0.20773616433143616,
"learning_rate": 3.0885917371310745e-06,
"loss": 0.8064232468605042,
"step": 676
},
{
"epoch": 1.7862796833773087,
"grad_norm": 0.2322840839624405,
"learning_rate": 3.0153689607045845e-06,
"loss": 0.7908546924591064,
"step": 677
},
{
"epoch": 1.7889182058047495,
"grad_norm": 0.2061799168586731,
"learning_rate": 2.942997601021924e-06,
"loss": 0.8989397287368774,
"step": 678
},
{
"epoch": 1.79155672823219,
"grad_norm": 0.2287965565919876,
"learning_rate": 2.871478969537206e-06,
"loss": 0.9402959942817688,
"step": 679
},
{
"epoch": 1.7941952506596306,
"grad_norm": 0.22069700062274933,
"learning_rate": 2.800814362252091e-06,
"loss": 0.7985097169876099,
"step": 680
},
{
"epoch": 1.7968337730870712,
"grad_norm": 0.20467771589756012,
"learning_rate": 2.731005059692332e-06,
"loss": 0.7937331199645996,
"step": 681
},
{
"epoch": 1.7994722955145117,
"grad_norm": 0.21517489850521088,
"learning_rate": 2.662052326884551e-06,
"loss": 0.8889937996864319,
"step": 682
},
{
"epoch": 1.8021108179419525,
"grad_norm": 0.20879219472408295,
"learning_rate": 2.5939574133333312e-06,
"loss": 0.875966489315033,
"step": 683
},
{
"epoch": 1.8047493403693933,
"grad_norm": 0.20481455326080322,
"learning_rate": 2.5267215529985342e-06,
"loss": 0.7772852778434753,
"step": 684
},
{
"epoch": 1.8073878627968338,
"grad_norm": 0.23448669910430908,
"learning_rate": 2.4603459642729863e-06,
"loss": 0.9144713878631592,
"step": 685
},
{
"epoch": 1.8100263852242744,
"grad_norm": 0.22476829588413239,
"learning_rate": 2.394831849960377e-06,
"loss": 0.802951455116272,
"step": 686
},
{
"epoch": 1.812664907651715,
"grad_norm": 0.20005077123641968,
"learning_rate": 2.330180397253473e-06,
"loss": 0.7209995985031128,
"step": 687
},
{
"epoch": 1.8153034300791555,
"grad_norm": 0.21807320415973663,
"learning_rate": 2.266392777712595e-06,
"loss": 0.8040324449539185,
"step": 688
},
{
"epoch": 1.8179419525065963,
"grad_norm": 0.21830332279205322,
"learning_rate": 2.203470147244385e-06,
"loss": 0.875383734703064,
"step": 689
},
{
"epoch": 1.820580474934037,
"grad_norm": 0.23793306946754456,
"learning_rate": 2.1414136460808807e-06,
"loss": 0.7538058161735535,
"step": 690
},
{
"epoch": 1.8232189973614776,
"grad_norm": 0.22125472128391266,
"learning_rate": 2.0802243987588066e-06,
"loss": 0.8075780868530273,
"step": 691
},
{
"epoch": 1.8258575197889182,
"grad_norm": 0.21494165062904358,
"learning_rate": 2.019903514099275e-06,
"loss": 0.8103631138801575,
"step": 692
},
{
"epoch": 1.8284960422163588,
"grad_norm": 0.23122619092464447,
"learning_rate": 1.9604520851876198e-06,
"loss": 0.8613567352294922,
"step": 693
},
{
"epoch": 1.8311345646437993,
"grad_norm": 0.20822040736675262,
"learning_rate": 1.901871189353599e-06,
"loss": 0.9021971225738525,
"step": 694
},
{
"epoch": 1.83377308707124,
"grad_norm": 0.20075157284736633,
"learning_rate": 1.8441618881519184e-06,
"loss": 0.8062754273414612,
"step": 695
},
{
"epoch": 1.8364116094986809,
"grad_norm": 0.20415347814559937,
"learning_rate": 1.7873252273429509e-06,
"loss": 0.811295747756958,
"step": 696
},
{
"epoch": 1.8390501319261214,
"grad_norm": 0.21688245236873627,
"learning_rate": 1.7313622368738014e-06,
"loss": 0.8866380453109741,
"step": 697
},
{
"epoch": 1.841688654353562,
"grad_norm": 0.19460809230804443,
"learning_rate": 1.6762739308596343e-06,
"loss": 0.8552433252334595,
"step": 698
},
{
"epoch": 1.8443271767810026,
"grad_norm": 0.1838616132736206,
"learning_rate": 1.6220613075653202e-06,
"loss": 0.6805065274238586,
"step": 699
},
{
"epoch": 1.8469656992084431,
"grad_norm": 0.1913633495569229,
"learning_rate": 1.5687253493873067e-06,
"loss": 0.7490788698196411,
"step": 700
},
{
"epoch": 1.849604221635884,
"grad_norm": 0.2015620470046997,
"learning_rate": 1.51626702283586e-06,
"loss": 0.8849630355834961,
"step": 701
},
{
"epoch": 1.8522427440633247,
"grad_norm": 0.2095753252506256,
"learning_rate": 1.464687278517518e-06,
"loss": 0.8486800193786621,
"step": 702
},
{
"epoch": 1.8548812664907652,
"grad_norm": 0.21090112626552582,
"learning_rate": 1.4139870511178766e-06,
"loss": 0.8101059794425964,
"step": 703
},
{
"epoch": 1.8575197889182058,
"grad_norm": 0.22010932862758636,
"learning_rate": 1.3641672593846632e-06,
"loss": 0.8158517479896545,
"step": 704
},
{
"epoch": 1.8601583113456464,
"grad_norm": 0.20055443048477173,
"learning_rate": 1.3152288061110518e-06,
"loss": 0.7629213333129883,
"step": 705
},
{
"epoch": 1.862796833773087,
"grad_norm": 0.19355174899101257,
"learning_rate": 1.2671725781193467e-06,
"loss": 0.7851141691207886,
"step": 706
},
{
"epoch": 1.8654353562005277,
"grad_norm": 0.2218313366174698,
"learning_rate": 1.2199994462448904e-06,
"loss": 0.8358639478683472,
"step": 707
},
{
"epoch": 1.8680738786279685,
"grad_norm": 0.20099952816963196,
"learning_rate": 1.1737102653202826e-06,
"loss": 0.8785051107406616,
"step": 708
},
{
"epoch": 1.870712401055409,
"grad_norm": 0.20789465308189392,
"learning_rate": 1.128305874159896e-06,
"loss": 0.7248390316963196,
"step": 709
},
{
"epoch": 1.8733509234828496,
"grad_norm": 0.20585159957408905,
"learning_rate": 1.0837870955446638e-06,
"loss": 0.8331854343414307,
"step": 710
},
{
"epoch": 1.8759894459102902,
"grad_norm": 0.21308377385139465,
"learning_rate": 1.040154736207194e-06,
"loss": 0.860977292060852,
"step": 711
},
{
"epoch": 1.8786279683377307,
"grad_norm": 0.21850626170635223,
"learning_rate": 9.974095868171162e-07,
"loss": 0.8206614255905151,
"step": 712
},
{
"epoch": 1.8812664907651715,
"grad_norm": 0.19836367666721344,
"learning_rate": 9.555524219667989e-07,
"loss": 0.8025310635566711,
"step": 713
},
{
"epoch": 1.8839050131926123,
"grad_norm": 0.21136832237243652,
"learning_rate": 9.145840001572537e-07,
"loss": 0.8504902124404907,
"step": 714
},
{
"epoch": 1.8865435356200528,
"grad_norm": 0.23634563386440277,
"learning_rate": 8.745050637844532e-07,
"loss": 0.8698995113372803,
"step": 715
},
{
"epoch": 1.8891820580474934,
"grad_norm": 0.2007746547460556,
"learning_rate": 8.353163391258301e-07,
"loss": 0.8317679166793823,
"step": 716
},
{
"epoch": 1.891820580474934,
"grad_norm": 0.20997071266174316,
"learning_rate": 7.970185363271431e-07,
"loss": 0.7819387912750244,
"step": 717
},
{
"epoch": 1.8944591029023745,
"grad_norm": 0.21593962609767914,
"learning_rate": 7.596123493895991e-07,
"loss": 0.8462232947349548,
"step": 718
},
{
"epoch": 1.8970976253298153,
"grad_norm": 0.20098824799060822,
"learning_rate": 7.230984561572729e-07,
"loss": 0.835645318031311,
"step": 719
},
{
"epoch": 1.899736147757256,
"grad_norm": 0.20634238421916962,
"learning_rate": 6.874775183048299e-07,
"loss": 0.8471999764442444,
"step": 720
},
{
"epoch": 1.9023746701846966,
"grad_norm": 0.18705205619335175,
"learning_rate": 6.527501813255344e-07,
"loss": 0.8066807389259338,
"step": 721
},
{
"epoch": 1.9050131926121372,
"grad_norm": 0.22316007316112518,
"learning_rate": 6.189170745195649e-07,
"loss": 0.8412566184997559,
"step": 722
},
{
"epoch": 1.9076517150395778,
"grad_norm": 0.20048147439956665,
"learning_rate": 5.859788109825793e-07,
"loss": 0.7970318794250488,
"step": 723
},
{
"epoch": 1.9102902374670183,
"grad_norm": 0.22987322509288788,
"learning_rate": 5.53935987594617e-07,
"loss": 0.7696294188499451,
"step": 724
},
{
"epoch": 1.912928759894459,
"grad_norm": 0.23662720620632172,
"learning_rate": 5.227891850093314e-07,
"loss": 0.8058530688285828,
"step": 725
},
{
"epoch": 1.9155672823218999,
"grad_norm": 0.2005518525838852,
"learning_rate": 4.925389676433745e-07,
"loss": 0.7893189191818237,
"step": 726
},
{
"epoch": 1.9182058047493404,
"grad_norm": 0.22957301139831543,
"learning_rate": 4.6318588366625616e-07,
"loss": 0.9222890138626099,
"step": 727
},
{
"epoch": 1.920844327176781,
"grad_norm": 0.20035627484321594,
"learning_rate": 4.347304649903572e-07,
"loss": 0.8075400590896606,
"step": 728
},
{
"epoch": 1.9234828496042216,
"grad_norm": 0.22035662829875946,
"learning_rate": 4.071732272613149e-07,
"loss": 0.800722599029541,
"step": 729
},
{
"epoch": 1.9261213720316621,
"grad_norm": 0.21661067008972168,
"learning_rate": 3.805146698486695e-07,
"loss": 0.8441472053527832,
"step": 730
},
{
"epoch": 1.928759894459103,
"grad_norm": 0.2007487416267395,
"learning_rate": 3.5475527583681e-07,
"loss": 0.720578134059906,
"step": 731
},
{
"epoch": 1.9313984168865437,
"grad_norm": 0.2147989273071289,
"learning_rate": 3.2989551201624835e-07,
"loss": 0.8720710277557373,
"step": 732
},
{
"epoch": 1.9340369393139842,
"grad_norm": 0.2244662344455719,
"learning_rate": 3.059358288751202e-07,
"loss": 0.8538236021995544,
"step": 733
},
{
"epoch": 1.9366754617414248,
"grad_norm": 0.1987634301185608,
"learning_rate": 2.828766605910471e-07,
"loss": 0.8031460642814636,
"step": 734
},
{
"epoch": 1.9393139841688654,
"grad_norm": 0.23431123793125153,
"learning_rate": 2.6071842502326527e-07,
"loss": 0.9187484979629517,
"step": 735
},
{
"epoch": 1.941952506596306,
"grad_norm": 0.20140373706817627,
"learning_rate": 2.3946152370505346e-07,
"loss": 0.8281949758529663,
"step": 736
},
{
"epoch": 1.9445910290237467,
"grad_norm": 0.20692074298858643,
"learning_rate": 2.1910634183644474e-07,
"loss": 0.7750309705734253,
"step": 737
},
{
"epoch": 1.9472295514511875,
"grad_norm": 0.2009861320257187,
"learning_rate": 1.996532482772595e-07,
"loss": 0.8492032885551453,
"step": 738
},
{
"epoch": 1.949868073878628,
"grad_norm": 0.21607014536857605,
"learning_rate": 1.811025955404333e-07,
"loss": 0.8605877161026001,
"step": 739
},
{
"epoch": 1.9525065963060686,
"grad_norm": 0.21618352830410004,
"learning_rate": 1.6345471978558845e-07,
"loss": 0.8067154884338379,
"step": 740
},
{
"epoch": 1.9551451187335092,
"grad_norm": 0.1969093680381775,
"learning_rate": 1.4670994081297795e-07,
"loss": 0.8418698310852051,
"step": 741
},
{
"epoch": 1.9577836411609497,
"grad_norm": 0.21148380637168884,
"learning_rate": 1.3086856205768438e-07,
"loss": 0.9082729816436768,
"step": 742
},
{
"epoch": 1.9604221635883905,
"grad_norm": 0.21570627391338348,
"learning_rate": 1.1593087058410779e-07,
"loss": 0.828383207321167,
"step": 743
},
{
"epoch": 1.9630606860158313,
"grad_norm": 0.22988225519657135,
"learning_rate": 1.0189713708078085e-07,
"loss": 0.8422545194625854,
"step": 744
},
{
"epoch": 1.9656992084432718,
"grad_norm": 0.21034526824951172,
"learning_rate": 8.876761585545068e-08,
"loss": 0.8603432774543762,
"step": 745
},
{
"epoch": 1.9683377308707124,
"grad_norm": 0.23274272680282593,
"learning_rate": 7.65425448304713e-08,
"loss": 0.9002218246459961,
"step": 746
},
{
"epoch": 1.970976253298153,
"grad_norm": 0.19577914476394653,
"learning_rate": 6.522214553850159e-08,
"loss": 0.8663936853408813,
"step": 747
},
{
"epoch": 1.9736147757255935,
"grad_norm": 0.2175392359495163,
"learning_rate": 5.4806623118486275e-08,
"loss": 0.8117172122001648,
"step": 748
},
{
"epoch": 1.9762532981530343,
"grad_norm": 0.2106153666973114,
"learning_rate": 4.529616631193112e-08,
"loss": 0.9605646133422852,
"step": 749
},
{
"epoch": 1.978891820580475,
"grad_norm": 0.20242197811603546,
"learning_rate": 3.669094745950008e-08,
"loss": 0.7868301272392273,
"step": 750
},
{
"epoch": 1.9815303430079156,
"grad_norm": 0.22185878455638885,
"learning_rate": 2.899112249786229e-08,
"loss": 0.9537394046783447,
"step": 751
},
{
"epoch": 1.9841688654353562,
"grad_norm": 0.2320915013551712,
"learning_rate": 2.219683095690539e-08,
"loss": 0.9150735139846802,
"step": 752
},
{
"epoch": 1.9868073878627968,
"grad_norm": 0.24962309002876282,
"learning_rate": 1.6308195957182027e-08,
"loss": 0.769099235534668,
"step": 753
},
{
"epoch": 1.9894459102902373,
"grad_norm": 0.20675364136695862,
"learning_rate": 1.1325324207667188e-08,
"loss": 0.8421361446380615,
"step": 754
},
{
"epoch": 1.992084432717678,
"grad_norm": 0.1924310028553009,
"learning_rate": 7.248306003865279e-09,
"loss": 0.8723808526992798,
"step": 755
},
{
"epoch": 1.9947229551451189,
"grad_norm": 0.2054050862789154,
"learning_rate": 4.07721522613369e-09,
"loss": 0.8893437385559082,
"step": 756
},
{
"epoch": 1.9973614775725594,
"grad_norm": 0.21500982344150543,
"learning_rate": 1.8121093383671738e-09,
"loss": 0.9416911602020264,
"step": 757
},
{
"epoch": 2.0,
"grad_norm": 0.2577999234199524,
"learning_rate": 4.5302938694868903e-10,
"loss": 0.8910604119300842,
"step": 758
}
],
"logging_steps": 1,
"max_steps": 758,
"num_input_tokens_seen": 0,
"num_train_epochs": 2,
"save_steps": 100,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 8.294513101283913e+18,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}