Files
math-1b-sft-numinamath-bs51…/trainer_state.json

1213 lines
30 KiB
JSON
Raw Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 3358,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.005956813104988831,
"grad_norm": 2.676292657852173,
"learning_rate": 1.8811881188118813e-06,
"loss": 1.5493,
"step": 20
},
{
"epoch": 0.011913626209977662,
"grad_norm": 0.9081353545188904,
"learning_rate": 3.861386138613862e-06,
"loss": 1.2611,
"step": 40
},
{
"epoch": 0.017870439314966492,
"grad_norm": 0.38824182748794556,
"learning_rate": 5.841584158415842e-06,
"loss": 0.8778,
"step": 60
},
{
"epoch": 0.023827252419955324,
"grad_norm": 0.1888394057750702,
"learning_rate": 7.821782178217822e-06,
"loss": 0.7519,
"step": 80
},
{
"epoch": 0.029784065524944156,
"grad_norm": 0.16975219547748566,
"learning_rate": 9.801980198019802e-06,
"loss": 0.6968,
"step": 100
},
{
"epoch": 0.035740878629932984,
"grad_norm": 0.16122567653656006,
"learning_rate": 9.99924640442892e-06,
"loss": 0.6755,
"step": 120
},
{
"epoch": 0.04169769173492182,
"grad_norm": 0.15297269821166992,
"learning_rate": 9.99664167434275e-06,
"loss": 0.661,
"step": 140
},
{
"epoch": 0.04765450483991065,
"grad_norm": 0.14783619344234467,
"learning_rate": 9.99217747519571e-06,
"loss": 0.6471,
"step": 160
},
{
"epoch": 0.05361131794489948,
"grad_norm": 0.15165136754512787,
"learning_rate": 9.98585546831169e-06,
"loss": 0.6479,
"step": 180
},
{
"epoch": 0.05956813104988831,
"grad_norm": 0.14331822097301483,
"learning_rate": 9.977678006386153e-06,
"loss": 0.627,
"step": 200
},
{
"epoch": 0.06552494415487714,
"grad_norm": 0.14584220945835114,
"learning_rate": 9.967648132610596e-06,
"loss": 0.6286,
"step": 220
},
{
"epoch": 0.07148175725986597,
"grad_norm": 0.14796026051044464,
"learning_rate": 9.95576957954004e-06,
"loss": 0.6257,
"step": 240
},
{
"epoch": 0.0774385703648548,
"grad_norm": 0.14510533213615417,
"learning_rate": 9.942046767703982e-06,
"loss": 0.6233,
"step": 260
},
{
"epoch": 0.08339538346984364,
"grad_norm": 0.14004668593406677,
"learning_rate": 9.926484803961332e-06,
"loss": 0.6165,
"step": 280
},
{
"epoch": 0.08935219657483247,
"grad_norm": 0.1493811011314392,
"learning_rate": 9.909089479599916e-06,
"loss": 0.6179,
"step": 300
},
{
"epoch": 0.0953090096798213,
"grad_norm": 0.14396651089191437,
"learning_rate": 9.88986726818129e-06,
"loss": 0.6183,
"step": 320
},
{
"epoch": 0.10126582278481013,
"grad_norm": 0.1511320173740387,
"learning_rate": 9.868825323131646e-06,
"loss": 0.6051,
"step": 340
},
{
"epoch": 0.10722263588979895,
"grad_norm": 0.1451549530029297,
"learning_rate": 9.8459714750797e-06,
"loss": 0.6053,
"step": 360
},
{
"epoch": 0.11317944899478778,
"grad_norm": 0.14344023168087006,
"learning_rate": 9.821314228942587e-06,
"loss": 0.6137,
"step": 380
},
{
"epoch": 0.11913626209977662,
"grad_norm": 0.14291702210903168,
"learning_rate": 9.794862760760798e-06,
"loss": 0.6039,
"step": 400
},
{
"epoch": 0.12509307520476545,
"grad_norm": 0.1428287923336029,
"learning_rate": 9.76662691428337e-06,
"loss": 0.6084,
"step": 420
},
{
"epoch": 0.13104988830975428,
"grad_norm": 0.1444147676229477,
"learning_rate": 9.736617197304615e-06,
"loss": 0.5972,
"step": 440
},
{
"epoch": 0.1370067014147431,
"grad_norm": 0.14073446393013,
"learning_rate": 9.704844777753686e-06,
"loss": 0.5927,
"step": 460
},
{
"epoch": 0.14296351451973194,
"grad_norm": 0.14289303123950958,
"learning_rate": 9.671321479538522e-06,
"loss": 0.5938,
"step": 480
},
{
"epoch": 0.14892032762472077,
"grad_norm": 0.14584945142269135,
"learning_rate": 9.63605977814564e-06,
"loss": 0.5911,
"step": 500
},
{
"epoch": 0.1548771407297096,
"grad_norm": 0.14528514444828033,
"learning_rate": 9.59907279599747e-06,
"loss": 0.5902,
"step": 520
},
{
"epoch": 0.16083395383469842,
"grad_norm": 0.15308603644371033,
"learning_rate": 9.56037429756892e-06,
"loss": 0.5844,
"step": 540
},
{
"epoch": 0.16679076693968728,
"grad_norm": 0.14307202398777008,
"learning_rate": 9.51997868426502e-06,
"loss": 0.5966,
"step": 560
},
{
"epoch": 0.1727475800446761,
"grad_norm": 0.1525873839855194,
"learning_rate": 9.477900989061519e-06,
"loss": 0.5929,
"step": 580
},
{
"epoch": 0.17870439314966494,
"grad_norm": 0.1432172805070877,
"learning_rate": 9.43415687091047e-06,
"loss": 0.588,
"step": 600
},
{
"epoch": 0.18466120625465376,
"grad_norm": 0.1422407627105713,
"learning_rate": 9.388762608912834e-06,
"loss": 0.5839,
"step": 620
},
{
"epoch": 0.1906180193596426,
"grad_norm": 0.14125266671180725,
"learning_rate": 9.341735096260326e-06,
"loss": 0.5895,
"step": 640
},
{
"epoch": 0.19657483246463142,
"grad_norm": 0.14493770897388458,
"learning_rate": 9.293091833948709e-06,
"loss": 0.5883,
"step": 660
},
{
"epoch": 0.20253164556962025,
"grad_norm": 0.1482989341020584,
"learning_rate": 9.242850924264918e-06,
"loss": 0.5819,
"step": 680
},
{
"epoch": 0.20848845867460908,
"grad_norm": 0.1360362470149994,
"learning_rate": 9.191031064050396e-06,
"loss": 0.5741,
"step": 700
},
{
"epoch": 0.2144452717795979,
"grad_norm": 0.14374487102031708,
"learning_rate": 9.137651537743193e-06,
"loss": 0.5793,
"step": 720
},
{
"epoch": 0.22040208488458674,
"grad_norm": 0.14358563721179962,
"learning_rate": 9.082732210201371e-06,
"loss": 0.58,
"step": 740
},
{
"epoch": 0.22635889798957556,
"grad_norm": 0.14029882848262787,
"learning_rate": 9.026293519310426e-06,
"loss": 0.577,
"step": 760
},
{
"epoch": 0.23231571109456442,
"grad_norm": 0.14996567368507385,
"learning_rate": 8.968356468377454e-06,
"loss": 0.5694,
"step": 780
},
{
"epoch": 0.23827252419955325,
"grad_norm": 0.13533169031143188,
"learning_rate": 8.908942618314912e-06,
"loss": 0.5722,
"step": 800
},
{
"epoch": 0.24422933730454208,
"grad_norm": 0.14617300033569336,
"learning_rate": 8.848074079616838e-06,
"loss": 0.5724,
"step": 820
},
{
"epoch": 0.2501861504095309,
"grad_norm": 0.14078132808208466,
"learning_rate": 8.785773504130587e-06,
"loss": 0.5689,
"step": 840
},
{
"epoch": 0.2561429635145197,
"grad_norm": 0.14140498638153076,
"learning_rate": 8.722064076627082e-06,
"loss": 0.5769,
"step": 860
},
{
"epoch": 0.26209977661950856,
"grad_norm": 0.14630186557769775,
"learning_rate": 8.65696950617274e-06,
"loss": 0.5694,
"step": 880
},
{
"epoch": 0.2680565897244974,
"grad_norm": 0.14904920756816864,
"learning_rate": 8.590514017306298e-06,
"loss": 0.577,
"step": 900
},
{
"epoch": 0.2740134028294862,
"grad_norm": 0.14156410098075867,
"learning_rate": 8.522722341023795e-06,
"loss": 0.567,
"step": 920
},
{
"epoch": 0.2799702159344751,
"grad_norm": 0.1386679708957672,
"learning_rate": 8.453619705575096e-06,
"loss": 0.5637,
"step": 940
},
{
"epoch": 0.2859270290394639,
"grad_norm": 0.1374497413635254,
"learning_rate": 8.38323182707535e-06,
"loss": 0.5754,
"step": 960
},
{
"epoch": 0.29188384214445273,
"grad_norm": 0.14010567963123322,
"learning_rate": 8.31158489993489e-06,
"loss": 0.5625,
"step": 980
},
{
"epoch": 0.29784065524944153,
"grad_norm": 0.14161601662635803,
"learning_rate": 8.238705587111162e-06,
"loss": 0.5723,
"step": 1000
},
{
"epoch": 0.3037974683544304,
"grad_norm": 0.15009689331054688,
"learning_rate": 8.16462101018626e-06,
"loss": 0.566,
"step": 1020
},
{
"epoch": 0.3097542814594192,
"grad_norm": 0.14088541269302368,
"learning_rate": 8.089358739273806e-06,
"loss": 0.5627,
"step": 1040
},
{
"epoch": 0.31571109456440805,
"grad_norm": 0.13296513259410858,
"learning_rate": 8.0129467827589e-06,
"loss": 0.562,
"step": 1060
},
{
"epoch": 0.32166790766939685,
"grad_norm": 0.13788965344429016,
"learning_rate": 7.935413576874966e-06,
"loss": 0.5678,
"step": 1080
},
{
"epoch": 0.3276247207743857,
"grad_norm": 0.1438877135515213,
"learning_rate": 7.856787975121407e-06,
"loss": 0.5686,
"step": 1100
},
{
"epoch": 0.33358153387937456,
"grad_norm": 0.1439657360315323,
"learning_rate": 7.777099237525929e-06,
"loss": 0.5561,
"step": 1120
},
{
"epoch": 0.33953834698436336,
"grad_norm": 0.14364264905452728,
"learning_rate": 7.696377019755624e-06,
"loss": 0.5565,
"step": 1140
},
{
"epoch": 0.3454951600893522,
"grad_norm": 0.13935892283916473,
"learning_rate": 7.614651362080787e-06,
"loss": 0.5599,
"step": 1160
},
{
"epoch": 0.351451973194341,
"grad_norm": 0.13484328985214233,
"learning_rate": 7.531952678195621e-06,
"loss": 0.5638,
"step": 1180
},
{
"epoch": 0.3574087862993299,
"grad_norm": 0.13593600690364838,
"learning_rate": 7.4483117438999685e-06,
"loss": 0.5704,
"step": 1200
},
{
"epoch": 0.3633655994043187,
"grad_norm": 0.14042305946350098,
"learning_rate": 7.3637596856462945e-06,
"loss": 0.5692,
"step": 1220
},
{
"epoch": 0.36932241250930753,
"grad_norm": 0.13619549572467804,
"learning_rate": 7.278327968956159e-06,
"loss": 0.5565,
"step": 1240
},
{
"epoch": 0.37527922561429633,
"grad_norm": 0.14011716842651367,
"learning_rate": 7.192048386710537e-06,
"loss": 0.5606,
"step": 1260
},
{
"epoch": 0.3812360387192852,
"grad_norm": 0.13997526466846466,
"learning_rate": 7.10495304731828e-06,
"loss": 0.5669,
"step": 1280
},
{
"epoch": 0.387192851824274,
"grad_norm": 0.13427585363388062,
"learning_rate": 7.017074362767176e-06,
"loss": 0.5592,
"step": 1300
},
{
"epoch": 0.39314966492926284,
"grad_norm": 0.13454337418079376,
"learning_rate": 6.9284450365620385e-06,
"loss": 0.5566,
"step": 1320
},
{
"epoch": 0.3991064780342517,
"grad_norm": 0.13241685926914215,
"learning_rate": 6.83909805155429e-06,
"loss": 0.5672,
"step": 1340
},
{
"epoch": 0.4050632911392405,
"grad_norm": 0.1457376629114151,
"learning_rate": 6.749066657667609e-06,
"loss": 0.5598,
"step": 1360
},
{
"epoch": 0.41102010424422936,
"grad_norm": 0.13778510689735413,
"learning_rate": 6.658384359524166e-06,
"loss": 0.5535,
"step": 1380
},
{
"epoch": 0.41697691734921816,
"grad_norm": 0.13975104689598083,
"learning_rate": 6.5670849039761e-06,
"loss": 0.5571,
"step": 1400
},
{
"epoch": 0.422933730454207,
"grad_norm": 0.13337770104408264,
"learning_rate": 6.475202267546829e-06,
"loss": 0.5573,
"step": 1420
},
{
"epoch": 0.4288905435591958,
"grad_norm": 0.1365518569946289,
"learning_rate": 6.382770643786907e-06,
"loss": 0.5512,
"step": 1440
},
{
"epoch": 0.43484735666418467,
"grad_norm": 0.13856241106987,
"learning_rate": 6.289824430549092e-06,
"loss": 0.56,
"step": 1460
},
{
"epoch": 0.44080416976917347,
"grad_norm": 0.1410275250673294,
"learning_rate": 6.196398217187418e-06,
"loss": 0.5571,
"step": 1480
},
{
"epoch": 0.4467609828741623,
"grad_norm": 0.13721413910388947,
"learning_rate": 6.102526771684971e-06,
"loss": 0.5516,
"step": 1500
},
{
"epoch": 0.4527177959791511,
"grad_norm": 0.1381140947341919,
"learning_rate": 6.008245027715205e-06,
"loss": 0.5597,
"step": 1520
},
{
"epoch": 0.45867460908414,
"grad_norm": 0.1393618881702423,
"learning_rate": 5.9135880716415885e-06,
"loss": 0.565,
"step": 1540
},
{
"epoch": 0.46463142218912884,
"grad_norm": 0.13795211911201477,
"learning_rate": 5.818591129460437e-06,
"loss": 0.5485,
"step": 1560
},
{
"epoch": 0.47058823529411764,
"grad_norm": 0.13659724593162537,
"learning_rate": 5.723289553691782e-06,
"loss": 0.5503,
"step": 1580
},
{
"epoch": 0.4765450483991065,
"grad_norm": 0.1406840831041336,
"learning_rate": 5.6277188102231396e-06,
"loss": 0.5598,
"step": 1600
},
{
"epoch": 0.4825018615040953,
"grad_norm": 0.13637390732765198,
"learning_rate": 5.531914465111109e-06,
"loss": 0.567,
"step": 1620
},
{
"epoch": 0.48845867460908415,
"grad_norm": 0.13608106970787048,
"learning_rate": 5.435912171345691e-06,
"loss": 0.5534,
"step": 1640
},
{
"epoch": 0.49441548771407295,
"grad_norm": 0.14104554057121277,
"learning_rate": 5.3397476555822365e-06,
"loss": 0.5556,
"step": 1660
},
{
"epoch": 0.5003723008190618,
"grad_norm": 0.13906168937683105,
"learning_rate": 5.243456704846011e-06,
"loss": 0.5514,
"step": 1680
},
{
"epoch": 0.5063291139240507,
"grad_norm": 0.1464906483888626,
"learning_rate": 5.147075153214256e-06,
"loss": 0.5571,
"step": 1700
},
{
"epoch": 0.5122859270290394,
"grad_norm": 0.13363933563232422,
"learning_rate": 5.0506388684807805e-06,
"loss": 0.5546,
"step": 1720
},
{
"epoch": 0.5182427401340283,
"grad_norm": 0.13365721702575684,
"learning_rate": 4.954183738807972e-06,
"loss": 0.5595,
"step": 1740
},
{
"epoch": 0.5241995532390171,
"grad_norm": 0.13746888935565948,
"learning_rate": 4.857745659371249e-06,
"loss": 0.5507,
"step": 1760
},
{
"epoch": 0.530156366344006,
"grad_norm": 0.13584527373313904,
"learning_rate": 4.761360519000884e-06,
"loss": 0.5481,
"step": 1780
},
{
"epoch": 0.5361131794489948,
"grad_norm": 0.1351776421070099,
"learning_rate": 4.665064186826211e-06,
"loss": 0.5484,
"step": 1800
},
{
"epoch": 0.5420699925539836,
"grad_norm": 0.13709554076194763,
"learning_rate": 4.568892498927148e-06,
"loss": 0.5516,
"step": 1820
},
{
"epoch": 0.5480268056589724,
"grad_norm": 0.13824759423732758,
"learning_rate": 4.47288124499802e-06,
"loss": 0.5473,
"step": 1840
},
{
"epoch": 0.5539836187639613,
"grad_norm": 0.1349649280309677,
"learning_rate": 4.3770661550286485e-06,
"loss": 0.5446,
"step": 1860
},
{
"epoch": 0.5599404318689502,
"grad_norm": 0.14086101949214935,
"learning_rate": 4.281482886007665e-06,
"loss": 0.5537,
"step": 1880
},
{
"epoch": 0.5658972449739389,
"grad_norm": 0.12921996414661407,
"learning_rate": 4.186167008652975e-06,
"loss": 0.549,
"step": 1900
},
{
"epoch": 0.5718540580789278,
"grad_norm": 0.13636700809001923,
"learning_rate": 4.091153994174343e-06,
"loss": 0.5517,
"step": 1920
},
{
"epoch": 0.5778108711839166,
"grad_norm": 0.14155375957489014,
"learning_rate": 3.996479201073005e-06,
"loss": 0.5443,
"step": 1940
},
{
"epoch": 0.5837676842889055,
"grad_norm": 0.1383938044309616,
"learning_rate": 3.902177861983213e-06,
"loss": 0.5467,
"step": 1960
},
{
"epoch": 0.5897244973938943,
"grad_norm": 0.14025482535362244,
"learning_rate": 3.8082850705606387e-06,
"loss": 0.5511,
"step": 1980
},
{
"epoch": 0.5956813104988831,
"grad_norm": 0.14210323989391327,
"learning_rate": 3.7148357684224887e-06,
"loss": 0.5549,
"step": 2000
},
{
"epoch": 0.6016381236038719,
"grad_norm": 0.13119558990001678,
"learning_rate": 3.6218647321441934e-06,
"loss": 0.5472,
"step": 2020
},
{
"epoch": 0.6075949367088608,
"grad_norm": 0.13860084116458893,
"learning_rate": 3.529406560317521e-06,
"loss": 0.5501,
"step": 2040
},
{
"epoch": 0.6135517498138496,
"grad_norm": 0.133541077375412,
"learning_rate": 3.4374956606749337e-06,
"loss": 0.5461,
"step": 2060
},
{
"epoch": 0.6195085629188384,
"grad_norm": 0.133568674325943,
"learning_rate": 3.3461662372849635e-06,
"loss": 0.5409,
"step": 2080
},
{
"epoch": 0.6254653760238272,
"grad_norm": 0.13823166489601135,
"learning_rate": 3.25545227782337e-06,
"loss": 0.553,
"step": 2100
},
{
"epoch": 0.6314221891288161,
"grad_norm": 0.13361062109470367,
"learning_rate": 3.1653875409248557e-06,
"loss": 0.5451,
"step": 2120
},
{
"epoch": 0.637379002233805,
"grad_norm": 0.1389135867357254,
"learning_rate": 3.07600554361999e-06,
"loss": 0.5448,
"step": 2140
},
{
"epoch": 0.6433358153387937,
"grad_norm": 0.13915786147117615,
"learning_rate": 2.987339548862054e-06,
"loss": 0.5502,
"step": 2160
},
{
"epoch": 0.6492926284437825,
"grad_norm": 0.13328970968723297,
"learning_rate": 2.8994225531484456e-06,
"loss": 0.5491,
"step": 2180
},
{
"epoch": 0.6552494415487714,
"grad_norm": 0.1401146650314331,
"learning_rate": 2.812287274241238e-06,
"loss": 0.5442,
"step": 2200
},
{
"epoch": 0.6612062546537603,
"grad_norm": 0.13732944428920746,
"learning_rate": 2.7259661389914597e-06,
"loss": 0.5483,
"step": 2220
},
{
"epoch": 0.6671630677587491,
"grad_norm": 0.13650034368038177,
"learning_rate": 2.6404912712716585e-06,
"loss": 0.5486,
"step": 2240
},
{
"epoch": 0.6731198808637379,
"grad_norm": 0.14018245041370392,
"learning_rate": 2.5558944800211892e-06,
"loss": 0.5485,
"step": 2260
},
{
"epoch": 0.6790766939687267,
"grad_norm": 0.13473157584667206,
"learning_rate": 2.472207247408727e-06,
"loss": 0.5515,
"step": 2280
},
{
"epoch": 0.6850335070737156,
"grad_norm": 0.13380421698093414,
"learning_rate": 2.389460717116358e-06,
"loss": 0.544,
"step": 2300
},
{
"epoch": 0.6909903201787044,
"grad_norm": 0.13703712821006775,
"learning_rate": 2.30768568274967e-06,
"loss": 0.5443,
"step": 2320
},
{
"epoch": 0.6969471332836932,
"grad_norm": 0.13692112267017365,
"learning_rate": 2.2269125763780974e-06,
"loss": 0.5392,
"step": 2340
},
{
"epoch": 0.702903946388682,
"grad_norm": 0.12941767275333405,
"learning_rate": 2.147171457209807e-06,
"loss": 0.5406,
"step": 2360
},
{
"epoch": 0.7088607594936709,
"grad_norm": 0.1361764371395111,
"learning_rate": 2.0684920004053666e-06,
"loss": 0.54,
"step": 2380
},
{
"epoch": 0.7148175725986597,
"grad_norm": 0.13859207928180695,
"learning_rate": 1.990903486034308e-06,
"loss": 0.5473,
"step": 2400
},
{
"epoch": 0.7207743857036486,
"grad_norm": 0.13659511506557465,
"learning_rate": 1.914434788178737e-06,
"loss": 0.546,
"step": 2420
},
{
"epoch": 0.7267311988086373,
"grad_norm": 0.13769935071468353,
"learning_rate": 1.8391143641880188e-06,
"loss": 0.5372,
"step": 2440
},
{
"epoch": 0.7326880119136262,
"grad_norm": 0.13072194159030914,
"learning_rate": 1.7649702440885597e-06,
"loss": 0.5466,
"step": 2460
},
{
"epoch": 0.7386448250186151,
"grad_norm": 0.13650979101657867,
"learning_rate": 1.692030020152598e-06,
"loss": 0.5482,
"step": 2480
},
{
"epoch": 0.7446016381236039,
"grad_norm": 0.1302981823682785,
"learning_rate": 1.6203208366299216e-06,
"loss": 0.5512,
"step": 2500
},
{
"epoch": 0.7505584512285927,
"grad_norm": 0.13273392617702484,
"learning_rate": 1.5498693796462994e-06,
"loss": 0.5472,
"step": 2520
},
{
"epoch": 0.7565152643335815,
"grad_norm": 0.13820502161979675,
"learning_rate": 1.4807018672724066e-06,
"loss": 0.5397,
"step": 2540
},
{
"epoch": 0.7624720774385704,
"grad_norm": 0.1355142742395401,
"learning_rate": 1.4128440397669301e-06,
"loss": 0.5407,
"step": 2560
},
{
"epoch": 0.7684288905435592,
"grad_norm": 0.13458405435085297,
"learning_rate": 1.3463211499974993e-06,
"loss": 0.5424,
"step": 2580
},
{
"epoch": 0.774385703648548,
"grad_norm": 0.14045540988445282,
"learning_rate": 1.2811579540429836e-06,
"loss": 0.5463,
"step": 2600
},
{
"epoch": 0.7803425167535368,
"grad_norm": 0.14012621343135834,
"learning_rate": 1.2173787019806688e-06,
"loss": 0.5416,
"step": 2620
},
{
"epoch": 0.7862993298585257,
"grad_norm": 0.13432762026786804,
"learning_rate": 1.1550071288617486e-06,
"loss": 0.5443,
"step": 2640
},
{
"epoch": 0.7922561429635145,
"grad_norm": 0.13710802793502808,
"learning_rate": 1.0940664458784733e-06,
"loss": 0.541,
"step": 2660
},
{
"epoch": 0.7982129560685034,
"grad_norm": 0.13760127127170563,
"learning_rate": 1.0345793317262387e-06,
"loss": 0.5406,
"step": 2680
},
{
"epoch": 0.8041697691734921,
"grad_norm": 0.13593532145023346,
"learning_rate": 9.765679241638649e-07,
"loss": 0.5498,
"step": 2700
},
{
"epoch": 0.810126582278481,
"grad_norm": 0.13616913557052612,
"learning_rate": 9.200538117751595e-07,
"loss": 0.5512,
"step": 2720
},
{
"epoch": 0.8160833953834699,
"grad_norm": 0.14006003737449646,
"learning_rate": 8.650580259348595e-07,
"loss": 0.5508,
"step": 2740
},
{
"epoch": 0.8220402084884587,
"grad_norm": 0.14094527065753937,
"learning_rate": 8.116010329819418e-07,
"loss": 0.5472,
"step": 2760
},
{
"epoch": 0.8279970215934475,
"grad_norm": 0.13820116221904755,
"learning_rate": 7.597027266031992e-07,
"loss": 0.5421,
"step": 2780
},
{
"epoch": 0.8339538346984363,
"grad_norm": 0.13223175704479218,
"learning_rate": 7.093824204299299e-07,
"loss": 0.5424,
"step": 2800
},
{
"epoch": 0.8399106478034252,
"grad_norm": 0.13696220517158508,
"learning_rate": 6.60658840850486e-07,
"loss": 0.5566,
"step": 2820
},
{
"epoch": 0.845867460908414,
"grad_norm": 0.13152669370174408,
"learning_rate": 6.135501200413724e-07,
"loss": 0.5411,
"step": 2840
},
{
"epoch": 0.8518242740134029,
"grad_norm": 0.1330927312374115,
"learning_rate": 5.680737892194654e-07,
"loss": 0.5465,
"step": 2860
},
{
"epoch": 0.8577810871183916,
"grad_norm": 0.137652188539505,
"learning_rate": 5.242467721178774e-07,
"loss": 0.546,
"step": 2880
},
{
"epoch": 0.8637379002233805,
"grad_norm": 0.13697630167007446,
"learning_rate": 4.82085378687901e-07,
"loss": 0.5392,
"step": 2900
},
{
"epoch": 0.8696947133283693,
"grad_norm": 0.13946303725242615,
"learning_rate": 4.416052990293596e-07,
"loss": 0.5399,
"step": 2920
},
{
"epoch": 0.8756515264333582,
"grad_norm": 0.13377772271633148,
"learning_rate": 4.028215975516314e-07,
"loss": 0.5456,
"step": 2940
},
{
"epoch": 0.8816083395383469,
"grad_norm": 0.1395328789949417,
"learning_rate": 3.657487073675281e-07,
"loss": 0.5548,
"step": 2960
},
{
"epoch": 0.8875651526433358,
"grad_norm": 0.13429999351501465,
"learning_rate": 3.3040042492210103e-07,
"loss": 0.5388,
"step": 2980
},
{
"epoch": 0.8935219657483247,
"grad_norm": 0.14729471504688263,
"learning_rate": 2.9678990485837553e-07,
"loss": 0.5548,
"step": 3000
},
{
"epoch": 0.8994787788533135,
"grad_norm": 0.1297731250524521,
"learning_rate": 2.6492965512193934e-07,
"loss": 0.5483,
"step": 3020
},
{
"epoch": 0.9054355919583023,
"grad_norm": 0.13323907554149628,
"learning_rate": 2.348315323061845e-07,
"loss": 0.559,
"step": 3040
},
{
"epoch": 0.9113924050632911,
"grad_norm": 0.136334627866745,
"learning_rate": 2.0650673723995097e-07,
"loss": 0.5358,
"step": 3060
},
{
"epoch": 0.91734921816828,
"grad_norm": 0.13980716466903687,
"learning_rate": 1.7996581081920483e-07,
"loss": 0.5406,
"step": 3080
},
{
"epoch": 0.9233060312732688,
"grad_norm": 0.14025422930717468,
"learning_rate": 1.5521863008431725e-07,
"loss": 0.5463,
"step": 3100
},
{
"epoch": 0.9292628443782577,
"grad_norm": 0.1348225474357605,
"learning_rate": 1.3227440454437745e-07,
"loss": 0.5423,
"step": 3120
},
{
"epoch": 0.9352196574832464,
"grad_norm": 0.1378781944513321,
"learning_rate": 1.1114167274993537e-07,
"loss": 0.5572,
"step": 3140
},
{
"epoch": 0.9411764705882353,
"grad_norm": 0.1394428163766861,
"learning_rate": 9.1828299115434e-08,
"loss": 0.5473,
"step": 3160
},
{
"epoch": 0.9471332836932241,
"grad_norm": 0.1302187740802765,
"learning_rate": 7.434147099251165e-08,
"loss": 0.5434,
"step": 3180
},
{
"epoch": 0.953090096798213,
"grad_norm": 0.13509339094161987,
"learning_rate": 5.86876959952809e-08,
"loss": 0.541,
"step": 3200
},
{
"epoch": 0.9590469099032017,
"grad_norm": 0.1379985511302948,
"learning_rate": 4.4872799578554724e-08,
"loss": 0.5445,
"step": 3220
},
{
"epoch": 0.9650037230081906,
"grad_norm": 0.13398253917694092,
"learning_rate": 3.290192286994287e-08,
"loss": 0.548,
"step": 3240
},
{
"epoch": 0.9709605361131795,
"grad_norm": 0.1330421417951584,
"learning_rate": 2.2779520756611163e-08,
"loss": 0.5363,
"step": 3260
},
{
"epoch": 0.9769173492181683,
"grad_norm": 0.13323503732681274,
"learning_rate": 1.450936022742544e-08,
"loss": 0.5534,
"step": 3280
},
{
"epoch": 0.9828741623231572,
"grad_norm": 0.1395898163318634,
"learning_rate": 8.094518971088484e-09,
"loss": 0.548,
"step": 3300
},
{
"epoch": 0.9888309754281459,
"grad_norm": 0.13659007847309113,
"learning_rate": 3.537384230797303e-09,
"loss": 0.5428,
"step": 3320
},
{
"epoch": 0.9947877885331348,
"grad_norm": 0.1411222517490387,
"learning_rate": 8.396519158465621e-10,
"loss": 0.5412,
"step": 3340
},
{
"epoch": 1.0,
"step": 3358,
"total_flos": 7.06306771390772e+18,
"train_loss": 0.5769865204707152,
"train_runtime": 4593.0331,
"train_samples_per_second": 187.13,
"train_steps_per_second": 0.731
}
],
"logging_steps": 20,
"max_steps": 3358,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 7.06306771390772e+18,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}