{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.0,
"eval_steps": 500,
"global_step": 33984,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.011770244821092278,
"grad_norm": 4.125,
"learning_rate": 1.9607843137254903e-05,
"loss": 0.6291,
"step": 100
},
{
"epoch": 0.023540489642184557,
"grad_norm": 13.0625,
"learning_rate": 3.9215686274509805e-05,
"loss": 0.6276,
"step": 200
},
{
"epoch": 0.03531073446327684,
"grad_norm": 4.09375,
"learning_rate": 5.882352941176471e-05,
"loss": 0.6124,
"step": 300
},
{
"epoch": 0.047080979284369114,
"grad_norm": 12.9375,
"learning_rate": 7.843137254901961e-05,
"loss": 0.5819,
"step": 400
},
{
"epoch": 0.05885122410546139,
"grad_norm": 2.296875,
"learning_rate": 9.80392156862745e-05,
"loss": 0.5751,
"step": 500
},
{
"epoch": 0.07062146892655367,
"grad_norm": 8.9375,
"learning_rate": 0.00011764705882352942,
"loss": 0.5796,
"step": 600
},
{
"epoch": 0.08239171374764595,
"grad_norm": 8.5625,
"learning_rate": 0.0001372549019607843,
"loss": 0.5595,
"step": 700
},
{
"epoch": 0.09416195856873823,
"grad_norm": 3.5625,
"learning_rate": 0.00015686274509803922,
"loss": 0.539,
"step": 800
},
{
"epoch": 0.1059322033898305,
"grad_norm": 8.3125,
"learning_rate": 0.00017647058823529413,
"loss": 0.5309,
"step": 900
},
{
"epoch": 0.11770244821092278,
"grad_norm": 11.6875,
"learning_rate": 0.000196078431372549,
"loss": 0.5169,
"step": 1000
},
{
"epoch": 0.12947269303201506,
"grad_norm": 15.25,
"learning_rate": 0.00019999709351685367,
"loss": 0.5094,
"step": 1100
},
{
"epoch": 0.14124293785310735,
"grad_norm": 2.734375,
"learning_rate": 0.00019998528621863396,
"loss": 0.5228,
"step": 1200
},
{
"epoch": 0.1530131826741996,
"grad_norm": 2.59375,
"learning_rate": 0.00019996439752174486,
"loss": 0.51,
"step": 1300
},
{
"epoch": 0.1647834274952919,
"grad_norm": 14.25,
"learning_rate": 0.00019993442932345136,
"loss": 0.5085,
"step": 1400
},
{
"epoch": 0.1765536723163842,
"grad_norm": 6.15625,
"learning_rate": 0.0001998953843456855,
"loss": 0.4933,
"step": 1500
},
{
"epoch": 0.18832391713747645,
"grad_norm": 18.0,
"learning_rate": 0.00019984726613479922,
"loss": 0.4831,
"step": 1600
},
{
"epoch": 0.20009416195856874,
"grad_norm": 5.34375,
"learning_rate": 0.0001997900790612421,
"loss": 0.4831,
"step": 1700
},
{
"epoch": 0.211864406779661,
"grad_norm": 19.625,
"learning_rate": 0.00019972382831916457,
"loss": 0.4857,
"step": 1800
},
{
"epoch": 0.2236346516007533,
"grad_norm": 6.59375,
"learning_rate": 0.000199648519925946,
"loss": 0.4772,
"step": 1900
},
{
"epoch": 0.23540489642184556,
"grad_norm": 5.21875,
"learning_rate": 0.00019956416072164826,
"loss": 0.4953,
"step": 2000
},
{
"epoch": 0.24717514124293785,
"grad_norm": 16.25,
"learning_rate": 0.00019947075836839438,
"loss": 0.4906,
"step": 2100
},
{
"epoch": 0.2589453860640301,
"grad_norm": 11.125,
"learning_rate": 0.00019936832134967265,
"loss": 0.4731,
"step": 2200
},
{
"epoch": 0.2707156308851224,
"grad_norm": 6.40625,
"learning_rate": 0.00019925685896956616,
"loss": 0.472,
"step": 2300
},
{
"epoch": 0.2824858757062147,
"grad_norm": 4.96875,
"learning_rate": 0.00019913638135190756,
"loss": 0.4788,
"step": 2400
},
{
"epoch": 0.294256120527307,
"grad_norm": 3.9375,
"learning_rate": 0.00019900689943935971,
"loss": 0.4721,
"step": 2500
},
{
"epoch": 0.3060263653483992,
"grad_norm": 8.6875,
"learning_rate": 0.0001988684249924217,
"loss": 0.469,
"step": 2600
},
{
"epoch": 0.3177966101694915,
"grad_norm": 10.75,
"learning_rate": 0.00019872097058836076,
"loss": 0.4496,
"step": 2700
},
{
"epoch": 0.3295668549905838,
"grad_norm": 3.0,
"learning_rate": 0.00019856454962006973,
"loss": 0.4528,
"step": 2800
},
{
"epoch": 0.3413370998116761,
"grad_norm": 20.0,
"learning_rate": 0.00019839917629485077,
"loss": 0.4763,
"step": 2900
},
{
"epoch": 0.3531073446327684,
"grad_norm": 8.6875,
"learning_rate": 0.0001982248656331249,
"loss": 0.4762,
"step": 3000
},
{
"epoch": 0.3648775894538606,
"grad_norm": 9.0625,
"learning_rate": 0.0001980416334670678,
"loss": 0.4758,
"step": 3100
},
{
"epoch": 0.3766478342749529,
"grad_norm": 9.8125,
"learning_rate": 0.00019784949643917164,
"loss": 0.4649,
"step": 3200
},
{
"epoch": 0.3884180790960452,
"grad_norm": 27.875,
"learning_rate": 0.0001976484720007337,
"loss": 0.4669,
"step": 3300
},
{
"epoch": 0.4001883239171375,
"grad_norm": 2.296875,
"learning_rate": 0.00019743857841027115,
"loss": 0.463,
"step": 3400
},
{
"epoch": 0.4119585687382298,
"grad_norm": 13.1875,
"learning_rate": 0.00019721983473186278,
"loss": 0.4717,
"step": 3500
},
{
"epoch": 0.423728813559322,
"grad_norm": 2.390625,
"learning_rate": 0.00019699226083341742,
"loss": 0.4766,
"step": 3600
},
{
"epoch": 0.4354990583804143,
"grad_norm": 7.28125,
"learning_rate": 0.00019675587738486936,
"loss": 0.471,
"step": 3700
},
{
"epoch": 0.4472693032015066,
"grad_norm": 2.109375,
"learning_rate": 0.00019651070585630104,
"loss": 0.4591,
"step": 3800
},
{
"epoch": 0.4590395480225989,
"grad_norm": 3.796875,
"learning_rate": 0.00019625676851599288,
"loss": 0.4778,
"step": 3900
},
{
"epoch": 0.4708097928436911,
"grad_norm": 12.4375,
"learning_rate": 0.0001959940884284008,
"loss": 0.4474,
"step": 4000
},
{
"epoch": 0.4825800376647834,
"grad_norm": 8.8125,
"learning_rate": 0.00019572268945206126,
"loss": 0.4669,
"step": 4100
},
{
"epoch": 0.4943502824858757,
"grad_norm": 2.96875,
"learning_rate": 0.00019544259623742428,
"loss": 0.4736,
"step": 4200
},
{
"epoch": 0.506120527306968,
"grad_norm": 3.796875,
"learning_rate": 0.00019515383422461454,
"loss": 0.4709,
"step": 4300
},
{
"epoch": 0.5178907721280602,
"grad_norm": 11.5,
"learning_rate": 0.0001948564296411207,
"loss": 0.4612,
"step": 4400
},
{
"epoch": 0.5296610169491526,
"grad_norm": 2.328125,
"learning_rate": 0.0001945504094994132,
"loss": 0.4586,
"step": 4500
},
{
"epoch": 0.5414312617702448,
"grad_norm": 9.6875,
"learning_rate": 0.00019423580159449083,
"loss": 0.4488,
"step": 4600
},
{
"epoch": 0.553201506591337,
"grad_norm": 9.25,
"learning_rate": 0.00019391263450135607,
"loss": 0.4721,
"step": 4700
},
{
"epoch": 0.5649717514124294,
"grad_norm": 6.28125,
"learning_rate": 0.00019358093757241996,
"loss": 0.4507,
"step": 4800
},
{
"epoch": 0.5767419962335216,
"grad_norm": 6.84375,
"learning_rate": 0.00019324074093483577,
"loss": 0.4579,
"step": 4900
},
{
"epoch": 0.588512241054614,
"grad_norm": 5.3125,
"learning_rate": 0.0001928920754877629,
"loss": 0.4778,
"step": 5000
},
{
"epoch": 0.6002824858757062,
"grad_norm": 1.6875,
"learning_rate": 0.0001925349728995602,
"loss": 0.4615,
"step": 5100
},
{
"epoch": 0.6120527306967984,
"grad_norm": 9.4375,
"learning_rate": 0.00019216946560490973,
"loss": 0.4628,
"step": 5200
},
{
"epoch": 0.6238229755178908,
"grad_norm": 3.078125,
"learning_rate": 0.00019179558680187077,
"loss": 0.4768,
"step": 5300
},
{
"epoch": 0.635593220338983,
"grad_norm": 12.8125,
"learning_rate": 0.00019141337044886457,
"loss": 0.4498,
"step": 5400
},
{
"epoch": 0.6473634651600754,
"grad_norm": 3.46875,
"learning_rate": 0.0001910228512615899,
"loss": 0.4535,
"step": 5500
},
{
"epoch": 0.6591337099811676,
"grad_norm": 14.5625,
"learning_rate": 0.00019062406470987,
"loss": 0.4484,
"step": 5600
},
{
"epoch": 0.6709039548022598,
"grad_norm": 13.0,
"learning_rate": 0.00019021704701443083,
"loss": 0.4697,
"step": 5700
},
{
"epoch": 0.6826741996233522,
"grad_norm": 10.25,
"learning_rate": 0.00018980183514361154,
"loss": 0.462,
"step": 5800
},
{
"epoch": 0.6944444444444444,
"grad_norm": 2.4375,
"learning_rate": 0.00018937846681000631,
"loss": 0.4562,
"step": 5900
},
{
"epoch": 0.7062146892655368,
"grad_norm": 4.3125,
"learning_rate": 0.0001889469804670393,
"loss": 0.4509,
"step": 6000
},
{
"epoch": 0.717984934086629,
"grad_norm": 14.625,
"learning_rate": 0.000188507415305472,
"loss": 0.4401,
"step": 6100
},
{
"epoch": 0.7297551789077212,
"grad_norm": 5.96875,
"learning_rate": 0.0001880598112498435,
"loss": 0.4609,
"step": 6200
},
{
"epoch": 0.7415254237288136,
"grad_norm": 12.25,
"learning_rate": 0.00018760420895484446,
"loss": 0.4496,
"step": 6300
},
{
"epoch": 0.7532956685499058,
"grad_norm": 3.03125,
"learning_rate": 0.00018714064980162435,
"loss": 0.4541,
"step": 6400
},
{
"epoch": 0.7650659133709982,
"grad_norm": 15.4375,
"learning_rate": 0.00018666917589403297,
"loss": 0.4544,
"step": 6500
},
{
"epoch": 0.7768361581920904,
"grad_norm": 5.21875,
"learning_rate": 0.00018618983005479637,
"loss": 0.4532,
"step": 6600
},
{
"epoch": 0.7886064030131826,
"grad_norm": 4.96875,
"learning_rate": 0.00018570265582162728,
"loss": 0.454,
"step": 6700
},
{
"epoch": 0.800376647834275,
"grad_norm": 8.4375,
"learning_rate": 0.00018520769744327062,
"loss": 0.4576,
"step": 6800
},
{
"epoch": 0.8121468926553672,
"grad_norm": 2.640625,
"learning_rate": 0.00018470499987548473,
"loss": 0.4578,
"step": 6900
},
{
"epoch": 0.8239171374764596,
"grad_norm": 10.4375,
"learning_rate": 0.00018419460877695795,
"loss": 0.4544,
"step": 7000
},
{
"epoch": 0.8356873822975518,
"grad_norm": 25.25,
"learning_rate": 0.00018367657050516163,
"loss": 0.463,
"step": 7100
},
{
"epoch": 0.847457627118644,
"grad_norm": 1.921875,
"learning_rate": 0.00018315093211213962,
"loss": 0.4597,
"step": 7200
},
{
"epoch": 0.8592278719397364,
"grad_norm": 6.875,
"learning_rate": 0.00018261774134023471,
"loss": 0.4571,
"step": 7300
},
{
"epoch": 0.8709981167608286,
"grad_norm": 11.1875,
"learning_rate": 0.00018207704661775216,
"loss": 0.4545,
"step": 7400
},
{
"epoch": 0.882768361581921,
"grad_norm": 2.140625,
"learning_rate": 0.00018152889705456117,
"loss": 0.4429,
"step": 7500
},
{
"epoch": 0.8945386064030132,
"grad_norm": 4.84375,
"learning_rate": 0.00018097334243763453,
"loss": 0.4668,
"step": 7600
},
{
"epoch": 0.9063088512241054,
"grad_norm": 13.0,
"learning_rate": 0.00018041043322652626,
"loss": 0.4599,
"step": 7700
},
{
"epoch": 0.9180790960451978,
"grad_norm": 6.46875,
"learning_rate": 0.0001798402205487888,
"loss": 0.4561,
"step": 7800
},
{
"epoch": 0.92984934086629,
"grad_norm": 10.75,
"learning_rate": 0.00017926275619532906,
"loss": 0.4467,
"step": 7900
},
{
"epoch": 0.9416195856873822,
"grad_norm": 16.125,
"learning_rate": 0.00017867809261570448,
"loss": 0.4487,
"step": 8000
},
{
"epoch": 0.9533898305084746,
"grad_norm": 8.4375,
"learning_rate": 0.00017808628291335912,
"loss": 0.4647,
"step": 8100
},
{
"epoch": 0.9651600753295668,
"grad_norm": 9.125,
"learning_rate": 0.0001774873808408004,
"loss": 0.4515,
"step": 8200
},
{
"epoch": 0.9769303201506592,
"grad_norm": 27.375,
"learning_rate": 0.00017688144079471707,
"loss": 0.4537,
"step": 8300
},
{
"epoch": 0.9887005649717514,
"grad_norm": 1.9765625,
"learning_rate": 0.0001762685178110382,
"loss": 0.45,
"step": 8400
},
{
"epoch": 1.0004708097928436,
"grad_norm": 29.75,
"learning_rate": 0.0001756486675599347,
"loss": 0.4565,
"step": 8500
},
{
"epoch": 1.012241054613936,
"grad_norm": 1.84375,
"learning_rate": 0.00017502194634076278,
"loss": 0.4486,
"step": 8600
},
{
"epoch": 1.0240112994350283,
"grad_norm": 4.3125,
"learning_rate": 0.00017438841107695046,
"loss": 0.4556,
"step": 8700
},
{
"epoch": 1.0357815442561205,
"grad_norm": 15.5625,
"learning_rate": 0.00017374811931082743,
"loss": 0.4599,
"step": 8800
},
{
"epoch": 1.0475517890772128,
"grad_norm": 3.78125,
"learning_rate": 0.0001731011291983986,
"loss": 0.4479,
"step": 8900
},
{
"epoch": 1.0593220338983051,
"grad_norm": 18.875,
"learning_rate": 0.00017244749950406186,
"loss": 0.4488,
"step": 9000
},
{
"epoch": 1.0710922787193973,
"grad_norm": 9.75,
"learning_rate": 0.00017178728959527076,
"loss": 0.4458,
"step": 9100
},
{
"epoch": 1.0828625235404896,
"grad_norm": 5.0625,
"learning_rate": 0.00017112055943714234,
"loss": 0.4352,
"step": 9200
},
{
"epoch": 1.094632768361582,
"grad_norm": 15.0625,
"learning_rate": 0.00017044736958701058,
"loss": 0.4359,
"step": 9300
},
{
"epoch": 1.1064030131826743,
"grad_norm": 4.28125,
"learning_rate": 0.00016976778118892619,
"loss": 0.4564,
"step": 9400
},
{
"epoch": 1.1181732580037664,
"grad_norm": 3.703125,
"learning_rate": 0.00016908185596810292,
"loss": 0.4508,
"step": 9500
},
{
"epoch": 1.1299435028248588,
"grad_norm": 11.1875,
"learning_rate": 0.00016838965622531157,
"loss": 0.4489,
"step": 9600
},
{
"epoch": 1.1417137476459511,
"grad_norm": 8.0625,
"learning_rate": 0.00016769124483122098,
"loss": 0.4513,
"step": 9700
},
{
"epoch": 1.1534839924670433,
"grad_norm": 16.5,
"learning_rate": 0.00016698668522068788,
"loss": 0.4562,
"step": 9800
},
{
"epoch": 1.1652542372881356,
"grad_norm": 5.75,
"learning_rate": 0.00016627604138699515,
"loss": 0.4606,
"step": 9900
},
{
"epoch": 1.177024482109228,
"grad_norm": 7.34375,
"learning_rate": 0.00016555937787603967,
"loss": 0.4457,
"step": 10000
},
{
"epoch": 1.18879472693032,
"grad_norm": 10.3125,
"learning_rate": 0.00016483675978046946,
"loss": 0.447,
"step": 10100
},
{
"epoch": 1.2005649717514124,
"grad_norm": 19.75,
"learning_rate": 0.00016410825273377192,
"loss": 0.452,
"step": 10200
},
{
"epoch": 1.2123352165725048,
"grad_norm": 6.65625,
"learning_rate": 0.0001633739229043121,
"loss": 0.4449,
"step": 10300
},
{
"epoch": 1.2241054613935969,
"grad_norm": 31.0,
"learning_rate": 0.00016263383698932306,
"loss": 0.4555,
"step": 10400
},
{
"epoch": 1.2358757062146892,
"grad_norm": 7.1875,
"learning_rate": 0.00016188806220884786,
"loss": 0.4664,
"step": 10500
},
{
"epoch": 1.2476459510357816,
"grad_norm": 10.4375,
"learning_rate": 0.00016113666629963413,
"loss": 0.4559,
"step": 10600
},
{
"epoch": 1.2594161958568737,
"grad_norm": 8.4375,
"learning_rate": 0.00016037971750898172,
"loss": 0.4493,
"step": 10700
},
{
"epoch": 1.271186440677966,
"grad_norm": 8.875,
"learning_rate": 0.00015961728458854397,
"loss": 0.4556,
"step": 10800
},
{
"epoch": 1.2829566854990584,
"grad_norm": 3.296875,
"learning_rate": 0.00015884943678808324,
"loss": 0.4399,
"step": 10900
},
{
"epoch": 1.2947269303201507,
"grad_norm": 6.9375,
"learning_rate": 0.00015807624384918116,
"loss": 0.4572,
"step": 11000
},
{
"epoch": 1.3064971751412429,
"grad_norm": 4.5625,
"learning_rate": 0.00015729777599890395,
"loss": 0.4358,
"step": 11100
},
{
"epoch": 1.3182674199623352,
"grad_norm": 10.625,
"learning_rate": 0.00015651410394342422,
"loss": 0.4469,
"step": 11200
},
{
"epoch": 1.3300376647834276,
"grad_norm": 8.9375,
"learning_rate": 0.00015572529886159872,
"loss": 0.4488,
"step": 11300
},
{
"epoch": 1.34180790960452,
"grad_norm": 9.0,
"learning_rate": 0.00015493143239850329,
"loss": 0.4443,
"step": 11400
},
{
"epoch": 1.353578154425612,
"grad_norm": 7.1875,
"learning_rate": 0.00015413257665892566,
"loss": 0.4427,
"step": 11500
},
{
"epoch": 1.3653483992467044,
"grad_norm": 16.5,
"learning_rate": 0.00015332880420081647,
"loss": 0.468,
"step": 11600
},
{
"epoch": 1.3771186440677967,
"grad_norm": 2.390625,
"learning_rate": 0.00015252018802869866,
"loss": 0.4567,
"step": 11700
},
{
"epoch": 1.3888888888888888,
"grad_norm": 3.421875,
"learning_rate": 0.000151706801587037,
"loss": 0.4427,
"step": 11800
},
{
"epoch": 1.4006591337099812,
"grad_norm": 20.375,
"learning_rate": 0.00015088871875356723,
"loss": 0.454,
"step": 11900
},
{
"epoch": 1.4124293785310735,
"grad_norm": 12.25,
"learning_rate": 0.00015006601383258584,
"loss": 0.4537,
"step": 12000
},
{
"epoch": 1.4241996233521657,
"grad_norm": 3.84375,
"learning_rate": 0.00014923876154820134,
"loss": 0.4528,
"step": 12100
},
{
"epoch": 1.435969868173258,
"grad_norm": 2.953125,
"learning_rate": 0.00014840703703754716,
"loss": 0.4481,
"step": 12200
},
{
"epoch": 1.4477401129943503,
"grad_norm": 1.9296875,
"learning_rate": 0.00014757091584395726,
"loss": 0.4497,
"step": 12300
},
{
"epoch": 1.4595103578154425,
"grad_norm": 3.375,
"learning_rate": 0.00014673047391010463,
"loss": 0.453,
"step": 12400
},
{
"epoch": 1.4712806026365348,
"grad_norm": 5.4375,
"learning_rate": 0.00014588578757110358,
"loss": 0.4604,
"step": 12500
},
{
"epoch": 1.4830508474576272,
"grad_norm": 10.1875,
"learning_rate": 0.00014503693354757667,
"loss": 0.462,
"step": 12600
},
{
"epoch": 1.4948210922787193,
"grad_norm": 8.25,
"learning_rate": 0.000144183988938686,
"loss": 0.4348,
"step": 12700
},
{
"epoch": 1.5065913370998116,
"grad_norm": 3.6875,
"learning_rate": 0.00014332703121513088,
"loss": 0.4358,
"step": 12800
},
{
"epoch": 1.518361581920904,
"grad_norm": 7.75,
"learning_rate": 0.00014246613821211108,
"loss": 0.4546,
"step": 12900
},
{
"epoch": 1.530131826741996,
"grad_norm": 7.5625,
"learning_rate": 0.00014160138812225738,
"loss": 0.448,
"step": 13000
},
{
"epoch": 1.5419020715630887,
"grad_norm": 1.875,
"learning_rate": 0.00014073285948852963,
"loss": 0.4559,
"step": 13100
},
{
"epoch": 1.5536723163841808,
"grad_norm": 4.46875,
"learning_rate": 0.00013986063119708275,
"loss": 0.4503,
"step": 13200
},
{
"epoch": 1.565442561205273,
"grad_norm": 1.796875,
"learning_rate": 0.00013898478247010177,
"loss": 0.4443,
"step": 13300
},
{
"epoch": 1.5772128060263655,
"grad_norm": 15.3125,
"learning_rate": 0.00013810539285860624,
"loss": 0.4559,
"step": 13400
},
{
"epoch": 1.5889830508474576,
"grad_norm": 3.390625,
"learning_rate": 0.000137222542235225,
"loss": 0.4448,
"step": 13500
},
{
"epoch": 1.60075329566855,
"grad_norm": 8.6875,
"learning_rate": 0.00013633631078694127,
"loss": 0.4529,
"step": 13600
},
{
"epoch": 1.6125235404896423,
"grad_norm": 2.6875,
"learning_rate": 0.00013544677900780976,
"loss": 0.4487,
"step": 13700
},
{
"epoch": 1.6242937853107344,
"grad_norm": 6.40625,
"learning_rate": 0.0001345540276916455,
"loss": 0.4717,
"step": 13800
},
{
"epoch": 1.6360640301318268,
"grad_norm": 5.9375,
"learning_rate": 0.00013365813792468539,
"loss": 0.4464,
"step": 13900
},
{
"epoch": 1.6478342749529191,
"grad_norm": 20.25,
"learning_rate": 0.00013275919107822364,
"loss": 0.4522,
"step": 14000
},
{
"epoch": 1.6596045197740112,
"grad_norm": 4.0,
"learning_rate": 0.0001318572688012209,
"loss": 0.4284,
"step": 14100
},
{
"epoch": 1.6713747645951036,
"grad_norm": 15.5625,
"learning_rate": 0.0001309524530128882,
"loss": 0.436,
"step": 14200
},
{
"epoch": 1.683145009416196,
"grad_norm": 7.1875,
"learning_rate": 0.0001300448258952466,
"loss": 0.4545,
"step": 14300
},
{
"epoch": 1.694915254237288,
"grad_norm": 7.90625,
"learning_rate": 0.00012913446988566273,
"loss": 0.4605,
"step": 14400
},
{
"epoch": 1.7066854990583804,
"grad_norm": 8.3125,
"learning_rate": 0.00012822146766936125,
"loss": 0.4539,
"step": 14500
},
{
"epoch": 1.7184557438794728,
"grad_norm": 4.75,
"learning_rate": 0.00012730590217191479,
"loss": 0.4354,
"step": 14600
},
{
"epoch": 1.7302259887005649,
"grad_norm": 9.1875,
"learning_rate": 0.00012638785655171196,
"loss": 0.4569,
"step": 14700
},
{
"epoch": 1.7419962335216572,
"grad_norm": 2.0,
"learning_rate": 0.00012546741419240445,
"loss": 0.438,
"step": 14800
},
{
"epoch": 1.7537664783427496,
"grad_norm": 4.5625,
"learning_rate": 0.00012454465869533328,
"loss": 0.4541,
"step": 14900
},
{
"epoch": 1.7655367231638417,
"grad_norm": 3.3125,
"learning_rate": 0.0001236196738719357,
"loss": 0.4317,
"step": 15000
},
{
"epoch": 1.777306967984934,
"grad_norm": 16.625,
"learning_rate": 0.00012269254373613287,
"loss": 0.4593,
"step": 15100
},
{
"epoch": 1.7890772128060264,
"grad_norm": 7.1875,
"learning_rate": 0.00012176335249669888,
"loss": 0.431,
"step": 15200
},
{
"epoch": 1.8008474576271185,
"grad_norm": 11.125,
"learning_rate": 0.00012083218454961237,
"loss": 0.4362,
"step": 15300
},
{
"epoch": 1.812617702448211,
"grad_norm": 6.78125,
"learning_rate": 0.00011989912447039113,
"loss": 0.4351,
"step": 15400
},
{
"epoch": 1.8243879472693032,
"grad_norm": 14.625,
"learning_rate": 0.00011896425700641042,
"loss": 0.4485,
"step": 15500
},
{
"epoch": 1.8361581920903953,
"grad_norm": 6.59375,
"learning_rate": 0.00011802766706920533,
"loss": 0.4403,
"step": 15600
},
{
"epoch": 1.847928436911488,
"grad_norm": 4.15625,
"learning_rate": 0.00011708943972675866,
"loss": 0.4438,
"step": 15700
},
{
"epoch": 1.85969868173258,
"grad_norm": 18.5,
"learning_rate": 0.00011614966019577442,
"loss": 0.4447,
"step": 15800
},
{
"epoch": 1.8714689265536724,
"grad_norm": 6.6875,
"learning_rate": 0.00011520841383393774,
"loss": 0.4475,
"step": 15900
},
{
"epoch": 1.8832391713747647,
"grad_norm": 7.6875,
"learning_rate": 0.00011426578613216217,
"loss": 0.4433,
"step": 16000
},
{
"epoch": 1.8950094161958568,
"grad_norm": 6.625,
"learning_rate": 0.00011332186270682473,
"loss": 0.4472,
"step": 16100
},
{
"epoch": 1.9067796610169492,
"grad_norm": 5.46875,
"learning_rate": 0.00011237672929198944,
"loss": 0.4343,
"step": 16200
},
{
"epoch": 1.9185499058380415,
"grad_norm": 2.625,
"learning_rate": 0.00011143047173162057,
"loss": 0.4347,
"step": 16300
},
{
"epoch": 1.9303201506591336,
"grad_norm": 2.171875,
"learning_rate": 0.00011048317597178561,
"loss": 0.4326,
"step": 16400
},
{
"epoch": 1.942090395480226,
"grad_norm": 6.28125,
"learning_rate": 0.00010953492805284882,
"loss": 0.4489,
"step": 16500
},
{
"epoch": 1.9538606403013183,
"grad_norm": 14.4375,
"learning_rate": 0.0001085858141016566,
"loss": 0.4524,
"step": 16600
},
{
"epoch": 1.9656308851224105,
"grad_norm": 9.75,
"learning_rate": 0.00010763592032371482,
"loss": 0.4456,
"step": 16700
},
{
"epoch": 1.9774011299435028,
"grad_norm": 2.734375,
"learning_rate": 0.00010668533299535885,
"loss": 0.4479,
"step": 16800
},
{
"epoch": 1.9891713747645952,
"grad_norm": 4.0625,
"learning_rate": 0.00010573413845591748,
"loss": 0.4495,
"step": 16900
},
{
"epoch": 2.0009416195856873,
"grad_norm": 11.0,
"learning_rate": 0.00010478242309987088,
"loss": 0.4326,
"step": 17000
},
{
"epoch": 2.01271186440678,
"grad_norm": 7.28125,
"learning_rate": 0.00010383027336900355,
"loss": 0.4483,
"step": 17100
},
{
"epoch": 2.024482109227872,
"grad_norm": 10.3125,
"learning_rate": 0.00010287777574455315,
"loss": 0.4369,
"step": 17200
},
{
"epoch": 2.036252354048964,
"grad_norm": 8.625,
"learning_rate": 0.00010192501673935557,
"loss": 0.4239,
"step": 17300
},
{
"epoch": 2.0480225988700567,
"grad_norm": 9.1875,
"learning_rate": 0.00010097208288998727,
"loss": 0.4482,
"step": 17400
},
{
"epoch": 2.059792843691149,
"grad_norm": 5.34375,
"learning_rate": 0.00010001906074890527,
"loss": 0.4355,
"step": 17500
},
{
"epoch": 2.071563088512241,
"grad_norm": 17.75,
"learning_rate": 9.906603687658595e-05,
"loss": 0.4389,
"step": 17600
},
{
"epoch": 2.0833333333333335,
"grad_norm": 1.75,
"learning_rate": 9.811309783366291e-05,
"loss": 0.4386,
"step": 17700
},
{
"epoch": 2.0951035781544256,
"grad_norm": 11.9375,
"learning_rate": 9.716033017306491e-05,
"loss": 0.4553,
"step": 17800
},
{
"epoch": 2.1068738229755177,
"grad_norm": 7.15625,
"learning_rate": 9.620782043215454e-05,
"loss": 0.4537,
"step": 17900
},
{
"epoch": 2.1186440677966103,
"grad_norm": 5.375,
"learning_rate": 9.52556551248682e-05,
"loss": 0.4468,
"step": 18000
},
{
"epoch": 2.1304143126177024,
"grad_norm": 5.125,
"learning_rate": 9.430392073385824e-05,
"loss": 0.4374,
"step": 18100
},
{
"epoch": 2.1421845574387945,
"grad_norm": 4.03125,
"learning_rate": 9.33527037026382e-05,
"loss": 0.4431,
"step": 18200
},
{
"epoch": 2.153954802259887,
"grad_norm": 8.875,
"learning_rate": 9.240209042773105e-05,
"loss": 0.4401,
"step": 18300
},
{
"epoch": 2.1657250470809792,
"grad_norm": 5.25,
"learning_rate": 9.145216725082219e-05,
"loss": 0.447,
"step": 18400
},
{
"epoch": 2.1774952919020714,
"grad_norm": 10.125,
"learning_rate": 9.050302045091731e-05,
"loss": 0.4367,
"step": 18500
},
{
"epoch": 2.189265536723164,
"grad_norm": 4.75,
"learning_rate": 8.955473623650578e-05,
"loss": 0.4416,
"step": 18600
},
{
"epoch": 2.201035781544256,
"grad_norm": 7.0,
"learning_rate": 8.860740073773057e-05,
"loss": 0.4385,
"step": 18700
},
{
"epoch": 2.2128060263653486,
"grad_norm": 1.8828125,
"learning_rate": 8.766109999856553e-05,
"loss": 0.4508,
"step": 18800
},
{
"epoch": 2.2245762711864407,
"grad_norm": 2.65625,
"learning_rate": 8.671591996899974e-05,
"loss": 0.4432,
"step": 18900
},
{
"epoch": 2.236346516007533,
"grad_norm": 4.0625,
"learning_rate": 8.57719464972314e-05,
"loss": 0.4423,
"step": 19000
},
{
"epoch": 2.2481167608286254,
"grad_norm": 2.1875,
"learning_rate": 8.482926532187022e-05,
"loss": 0.4476,
"step": 19100
},
{
"epoch": 2.2598870056497176,
"grad_norm": 16.25,
"learning_rate": 8.388796206415004e-05,
"loss": 0.4417,
"step": 19200
},
{
"epoch": 2.2716572504708097,
"grad_norm": 4.5,
"learning_rate": 8.294812222015206e-05,
"loss": 0.4419,
"step": 19300
},
{
"epoch": 2.2834274952919023,
"grad_norm": 3.609375,
"learning_rate": 8.200983115303953e-05,
"loss": 0.4184,
"step": 19400
},
{
"epoch": 2.2951977401129944,
"grad_norm": 1.515625,
"learning_rate": 8.107317408530441e-05,
"loss": 0.4314,
"step": 19500
},
{
"epoch": 2.3069679849340865,
"grad_norm": 3.671875,
"learning_rate": 8.013823609102673e-05,
"loss": 0.4603,
"step": 19600
},
{
"epoch": 2.318738229755179,
"grad_norm": 11.3125,
"learning_rate": 7.920510208814779e-05,
"loss": 0.4465,
"step": 19700
},
{
"epoch": 2.330508474576271,
"grad_norm": 23.625,
"learning_rate": 7.827385683075701e-05,
"loss": 0.4501,
"step": 19800
},
{
"epoch": 2.3422787193973633,
"grad_norm": 9.375,
"learning_rate": 7.734458490139404e-05,
"loss": 0.4451,
"step": 19900
},
{
"epoch": 2.354048964218456,
"grad_norm": 9.1875,
"learning_rate": 7.641737070336649e-05,
"loss": 0.455,
"step": 20000
},
{
"epoch": 2.365819209039548,
"grad_norm": 5.46875,
"learning_rate": 7.549229845308362e-05,
"loss": 0.4444,
"step": 20100
},
{
"epoch": 2.37758945386064,
"grad_norm": 5.09375,
"learning_rate": 7.456945217240721e-05,
"loss": 0.4364,
"step": 20200
},
{
"epoch": 2.3893596986817327,
"grad_norm": 4.90625,
"learning_rate": 7.36489156810202e-05,
"loss": 0.4502,
"step": 20300
},
{
"epoch": 2.401129943502825,
"grad_norm": 2.8125,
"learning_rate": 7.273077258881342e-05,
"loss": 0.4303,
"step": 20400
},
{
"epoch": 2.412900188323917,
"grad_norm": 7.34375,
"learning_rate": 7.181510628829158e-05,
"loss": 0.4454,
"step": 20500
},
{
"epoch": 2.4246704331450095,
"grad_norm": 4.625,
"learning_rate": 7.090199994699911e-05,
"loss": 0.452,
"step": 20600
},
{
"epoch": 2.4364406779661016,
"grad_norm": 2.609375,
"learning_rate": 6.999153649996595e-05,
"loss": 0.4465,
"step": 20700
},
{
"epoch": 2.4482109227871938,
"grad_norm": 4.53125,
"learning_rate": 6.9083798642175e-05,
"loss": 0.4435,
"step": 20800
},
{
"epoch": 2.4599811676082863,
"grad_norm": 8.9375,
"learning_rate": 6.81788688210512e-05,
"loss": 0.421,
"step": 20900
},
{
"epoch": 2.4717514124293785,
"grad_norm": 16.375,
"learning_rate": 6.727682922897282e-05,
"loss": 0.4429,
"step": 21000
},
{
"epoch": 2.4835216572504706,
"grad_norm": 11.875,
"learning_rate": 6.637776179580641e-05,
"loss": 0.4327,
"step": 21100
},
{
"epoch": 2.495291902071563,
"grad_norm": 5.40625,
"learning_rate": 6.548174818146514e-05,
"loss": 0.4435,
"step": 21200
},
{
"epoch": 2.5070621468926553,
"grad_norm": 11.0,
"learning_rate": 6.458886976849183e-05,
"loss": 0.4429,
"step": 21300
},
{
"epoch": 2.5188323917137474,
"grad_norm": 2.109375,
"learning_rate": 6.369920765466746e-05,
"loss": 0.4286,
"step": 21400
},
{
"epoch": 2.53060263653484,
"grad_norm": 9.0625,
"learning_rate": 6.28128426456449e-05,
"loss": 0.4445,
"step": 21500
},
{
"epoch": 2.542372881355932,
"grad_norm": 15.375,
"learning_rate": 6.19298552476098e-05,
"loss": 0.4619,
"step": 21600
},
{
"epoch": 2.554143126177024,
"grad_norm": 2.03125,
"learning_rate": 6.105032565996836e-05,
"loss": 0.4357,
"step": 21700
},
{
"epoch": 2.565913370998117,
"grad_norm": 23.75,
"learning_rate": 6.017433376806296e-05,
"loss": 0.4436,
"step": 21800
},
{
"epoch": 2.577683615819209,
"grad_norm": 2.1875,
"learning_rate": 5.9301959135916496e-05,
"loss": 0.4567,
"step": 21900
},
{
"epoch": 2.5894538606403015,
"grad_norm": 4.09375,
"learning_rate": 5.84332809990058e-05,
"loss": 0.4446,
"step": 22000
},
{
"epoch": 2.6012241054613936,
"grad_norm": 6.03125,
"learning_rate": 5.756837825706474e-05,
"loss": 0.4351,
"step": 22100
},
{
"epoch": 2.6129943502824857,
"grad_norm": 3.578125,
"learning_rate": 5.670732946691808e-05,
"loss": 0.4467,
"step": 22200
},
{
"epoch": 2.6247645951035783,
"grad_norm": 8.125,
"learning_rate": 5.5850212835346375e-05,
"loss": 0.4417,
"step": 22300
},
{
"epoch": 2.6365348399246704,
"grad_norm": 11.9375,
"learning_rate": 5.499710621198262e-05,
"loss": 0.4498,
"step": 22400
},
{
"epoch": 2.648305084745763,
"grad_norm": 13.6875,
"learning_rate": 5.414808708224135e-05,
"loss": 0.4454,
"step": 22500
},
{
"epoch": 2.660075329566855,
"grad_norm": 2.359375,
"learning_rate": 5.330323256028101e-05,
"loss": 0.4596,
"step": 22600
},
{
"epoch": 2.6718455743879472,
"grad_norm": 6.6875,
"learning_rate": 5.2462619381999635e-05,
"loss": 0.4407,
"step": 22700
},
{
"epoch": 2.68361581920904,
"grad_norm": 3.046875,
"learning_rate": 5.162632389806523e-05,
"loss": 0.4448,
"step": 22800
},
{
"epoch": 2.695386064030132,
"grad_norm": 7.6875,
"learning_rate": 5.079442206698121e-05,
"loss": 0.4162,
"step": 22900
},
{
"epoch": 2.707156308851224,
"grad_norm": 12.1875,
"learning_rate": 4.996698944818699e-05,
"loss": 0.4467,
"step": 23000
},
{
"epoch": 2.7189265536723166,
"grad_norm": 6.03125,
"learning_rate": 4.914410119519528e-05,
"loss": 0.4397,
"step": 23100
},
{
"epoch": 2.7306967984934087,
"grad_norm": 2.1875,
"learning_rate": 4.832583204876626e-05,
"loss": 0.4434,
"step": 23200
},
{
"epoch": 2.742467043314501,
"grad_norm": 14.5,
"learning_rate": 4.7512256330118645e-05,
"loss": 0.4518,
"step": 23300
},
{
"epoch": 2.7542372881355934,
"grad_norm": 3.015625,
"learning_rate": 4.670344793417967e-05,
"loss": 0.4398,
"step": 23400
},
{
"epoch": 2.7660075329566856,
"grad_norm": 4.4375,
"learning_rate": 4.589948032287334e-05,
"loss": 0.4414,
"step": 23500
},
{
"epoch": 2.7777777777777777,
"grad_norm": 7.96875,
"learning_rate": 4.5100426518447905e-05,
"loss": 0.4478,
"step": 23600
},
{
"epoch": 2.7895480225988702,
"grad_norm": 5.5625,
"learning_rate": 4.430635909684356e-05,
"loss": 0.4585,
"step": 23700
},
{
"epoch": 2.8013182674199624,
"grad_norm": 6.03125,
"learning_rate": 4.351735018110066e-05,
"loss": 0.454,
"step": 23800
},
{
"epoch": 2.8130885122410545,
"grad_norm": 8.6875,
"learning_rate": 4.273347143480877e-05,
"loss": 0.4386,
"step": 23900
},
{
"epoch": 2.824858757062147,
"grad_norm": 2.390625,
"learning_rate": 4.1954794055597756e-05,
"loss": 0.4277,
"step": 24000
},
{
"epoch": 2.836629001883239,
"grad_norm": 10.1875,
"learning_rate": 4.118138876867119e-05,
"loss": 0.4457,
"step": 24100
},
{
"epoch": 2.8483992467043313,
"grad_norm": 2.84375,
"learning_rate": 4.0413325820382373e-05,
"loss": 0.4538,
"step": 24200
},
{
"epoch": 2.860169491525424,
"grad_norm": 1.7109375,
"learning_rate": 3.965067497185416e-05,
"loss": 0.4337,
"step": 24300
},
{
"epoch": 2.871939736346516,
"grad_norm": 18.375,
"learning_rate": 3.889350549264275e-05,
"loss": 0.4471,
"step": 24400
},
{
"epoch": 2.883709981167608,
"grad_norm": 9.125,
"learning_rate": 3.814188615444602e-05,
"loss": 0.4391,
"step": 24500
},
{
"epoch": 2.8954802259887007,
"grad_norm": 2.40625,
"learning_rate": 3.739588522485736e-05,
"loss": 0.4405,
"step": 24600
},
{
"epoch": 2.907250470809793,
"grad_norm": 3.5,
"learning_rate": 3.665557046116491e-05,
"loss": 0.4375,
"step": 24700
},
{
"epoch": 2.919020715630885,
"grad_norm": 3.703125,
"learning_rate": 3.592100910419738e-05,
"loss": 0.448,
"step": 24800
},
{
"epoch": 2.9307909604519775,
"grad_norm": 8.875,
"learning_rate": 3.519226787221692e-05,
"loss": 0.4356,
"step": 24900
},
{
"epoch": 2.9425612052730696,
"grad_norm": 12.125,
"learning_rate": 3.446941295485905e-05,
"loss": 0.4543,
"step": 25000
},
{
"epoch": 2.9543314500941618,
"grad_norm": 4.09375,
"learning_rate": 3.375251000712094e-05,
"loss": 0.4476,
"step": 25100
},
{
"epoch": 2.9661016949152543,
"grad_norm": 11.875,
"learning_rate": 3.304162414339814e-05,
"loss": 0.4476,
"step": 25200
},
{
"epoch": 2.9778719397363465,
"grad_norm": 10.625,
"learning_rate": 3.23368199315704e-05,
"loss": 0.4519,
"step": 25300
},
{
"epoch": 2.9896421845574386,
"grad_norm": 4.03125,
"learning_rate": 3.163816138713708e-05,
"loss": 0.4443,
"step": 25400
},
{
"epoch": 3.001412429378531,
"grad_norm": 15.0,
"learning_rate": 3.094571196740299e-05,
"loss": 0.4411,
"step": 25500
},
{
"epoch": 3.0131826741996233,
"grad_norm": 12.3125,
"learning_rate": 3.0259534565714488e-05,
"loss": 0.4403,
"step": 25600
},
{
"epoch": 3.024952919020716,
"grad_norm": 7.875,
"learning_rate": 2.957969150574712e-05,
"loss": 0.435,
"step": 25700
},
{
"epoch": 3.036723163841808,
"grad_norm": 12.5625,
"learning_rate": 2.8906244535845072e-05,
"loss": 0.444,
"step": 25800
},
{
"epoch": 3.0484934086629,
"grad_norm": 8.4375,
"learning_rate": 2.823925482341252e-05,
"loss": 0.4496,
"step": 25900
},
{
"epoch": 3.0602636534839927,
"grad_norm": 2.3125,
"learning_rate": 2.7578782949358073e-05,
"loss": 0.4441,
"step": 26000
},
{
"epoch": 3.0720338983050848,
"grad_norm": 10.125,
"learning_rate": 2.692488890259235e-05,
"loss": 0.4524,
"step": 26100
},
{
"epoch": 3.083804143126177,
"grad_norm": 1.96875,
"learning_rate": 2.6277632074579338e-05,
"loss": 0.4281,
"step": 26200
},
{
"epoch": 3.0955743879472695,
"grad_norm": 3.6875,
"learning_rate": 2.5637071253942002e-05,
"loss": 0.4364,
"step": 26300
},
{
"epoch": 3.1073446327683616,
"grad_norm": 2.203125,
"learning_rate": 2.5003264621122802e-05,
"loss": 0.4425,
"step": 26400
},
{
"epoch": 3.1191148775894537,
"grad_norm": 6.09375,
"learning_rate": 2.4376269743099102e-05,
"loss": 0.4227,
"step": 26500
},
{
"epoch": 3.1308851224105463,
"grad_norm": 6.65625,
"learning_rate": 2.375614356815461e-05,
"loss": 0.4615,
"step": 26600
},
{
"epoch": 3.1426553672316384,
"grad_norm": 5.375,
"learning_rate": 2.314294242070706e-05,
"loss": 0.4553,
"step": 26700
},
{
"epoch": 3.1544256120527305,
"grad_norm": 5.5,
"learning_rate": 2.2536721996192168e-05,
"loss": 0.4414,
"step": 26800
},
{
"epoch": 3.166195856873823,
"grad_norm": 2.4375,
"learning_rate": 2.19375373560051e-05,
"loss": 0.4525,
"step": 26900
},
{
"epoch": 3.1779661016949152,
"grad_norm": 14.625,
"learning_rate": 2.1345442922499394e-05,
"loss": 0.454,
"step": 27000
},
{
"epoch": 3.1897363465160073,
"grad_norm": 5.8125,
"learning_rate": 2.076049247404388e-05,
"loss": 0.4483,
"step": 27100
},
{
"epoch": 3.2015065913371,
"grad_norm": 9.1875,
"learning_rate": 2.018273914013814e-05,
"loss": 0.4437,
"step": 27200
},
{
"epoch": 3.213276836158192,
"grad_norm": 5.3125,
"learning_rate": 1.9612235396586976e-05,
"loss": 0.4423,
"step": 27300
},
{
"epoch": 3.225047080979284,
"grad_norm": 13.8125,
"learning_rate": 1.904903306073401e-05,
"loss": 0.4485,
"step": 27400
},
{
"epoch": 3.2368173258003767,
"grad_norm": 9.0625,
"learning_rate": 1.8493183286755423e-05,
"loss": 0.4435,
"step": 27500
},
{
"epoch": 3.248587570621469,
"grad_norm": 4.15625,
"learning_rate": 1.794473656101364e-05,
"loss": 0.4339,
"step": 27600
},
{
"epoch": 3.2603578154425614,
"grad_norm": 10.25,
"learning_rate": 1.7403742697471824e-05,
"loss": 0.4312,
"step": 27700
},
{
"epoch": 3.2721280602636535,
"grad_norm": 5.34375,
"learning_rate": 1.687025083316942e-05,
"loss": 0.4342,
"step": 27800
},
{
"epoch": 3.2838983050847457,
"grad_norm": 4.84375,
"learning_rate": 1.6344309423759142e-05,
"loss": 0.4368,
"step": 27900
},
{
"epoch": 3.2956685499058382,
"grad_norm": 13.625,
"learning_rate": 1.5825966239105882e-05,
"loss": 0.4312,
"step": 28000
},
{
"epoch": 3.3074387947269304,
"grad_norm": 10.25,
"learning_rate": 1.5315268358947987e-05,
"loss": 0.4367,
"step": 28100
},
{
"epoch": 3.3192090395480225,
"grad_norm": 3.546875,
"learning_rate": 1.4812262168620939e-05,
"loss": 0.4364,
"step": 28200
},
{
"epoch": 3.330979284369115,
"grad_norm": 8.1875,
"learning_rate": 1.4316993354844444e-05,
"loss": 0.4529,
"step": 28300
},
{
"epoch": 3.342749529190207,
"grad_norm": 3.0625,
"learning_rate": 1.3829506901572853e-05,
"loss": 0.4415,
"step": 28400
},
{
"epoch": 3.3545197740112993,
"grad_norm": 3.046875,
"learning_rate": 1.3349847085909262e-05,
"loss": 0.4292,
"step": 28500
},
{
"epoch": 3.366290018832392,
"grad_norm": 3.234375,
"learning_rate": 1.2878057474084004e-05,
"loss": 0.4493,
"step": 28600
},
{
"epoch": 3.378060263653484,
"grad_norm": 18.125,
"learning_rate": 1.2414180917497743e-05,
"loss": 0.4343,
"step": 28700
},
{
"epoch": 3.389830508474576,
"grad_norm": 2.078125,
"learning_rate": 1.1958259548829153e-05,
"loss": 0.4554,
"step": 28800
},
{
"epoch": 3.4016007532956687,
"grad_norm": 11.4375,
"learning_rate": 1.1510334778208332e-05,
"loss": 0.4496,
"step": 28900
},
{
"epoch": 3.413370998116761,
"grad_norm": 5.25,
"learning_rate": 1.1070447289455632e-05,
"loss": 0.4372,
"step": 29000
},
{
"epoch": 3.425141242937853,
"grad_norm": 14.9375,
"learning_rate": 1.0638637036386279e-05,
"loss": 0.4254,
"step": 29100
},
{
"epoch": 3.4369114877589455,
"grad_norm": 3.53125,
"learning_rate": 1.0214943239181606e-05,
"loss": 0.4426,
"step": 29200
},
{
"epoch": 3.4486817325800376,
"grad_norm": 6.0625,
"learning_rate": 9.799404380826827e-06,
"loss": 0.4429,
"step": 29300
},
{
"epoch": 3.4604519774011298,
"grad_norm": 6.0625,
"learning_rate": 9.392058203615583e-06,
"loss": 0.4349,
"step": 29400
},
{
"epoch": 3.4722222222222223,
"grad_norm": 2.921875,
"learning_rate": 8.992941705722014e-06,
"loss": 0.4466,
"step": 29500
},
{
"epoch": 3.4839924670433144,
"grad_norm": 7.03125,
"learning_rate": 8.602091137840318e-06,
"loss": 0.4381,
"step": 29600
},
{
"epoch": 3.4957627118644066,
"grad_norm": 3.46875,
"learning_rate": 8.219541999892155e-06,
"loss": 0.4479,
"step": 29700
},
{
"epoch": 3.507532956685499,
"grad_norm": 11.4375,
"learning_rate": 7.845329037802229e-06,
"loss": 0.4385,
"step": 29800
},
{
"epoch": 3.5193032015065913,
"grad_norm": 1.78125,
"learning_rate": 7.479486240342604e-06,
"loss": 0.4394,
"step": 29900
},
{
"epoch": 3.5310734463276834,
"grad_norm": 5.0625,
"learning_rate": 7.122046836045348e-06,
"loss": 0.4451,
"step": 30000
},
{
"epoch": 3.542843691148776,
"grad_norm": 6.34375,
"learning_rate": 6.773043290184644e-06,
"loss": 0.4482,
"step": 30100
},
{
"epoch": 3.554613935969868,
"grad_norm": 1.1953125,
"learning_rate": 6.432507301828028e-06,
"loss": 0.4402,
"step": 30200
},
{
"epoch": 3.56638418079096,
"grad_norm": 5.65625,
"learning_rate": 6.100469800957154e-06,
"loss": 0.4393,
"step": 30300
},
{
"epoch": 3.5781544256120528,
"grad_norm": 6.5,
"learning_rate": 5.7769609456585825e-06,
"loss": 0.424,
"step": 30400
},
{
"epoch": 3.589924670433145,
"grad_norm": 2.15625,
"learning_rate": 5.462010119384664e-06,
"loss": 0.4531,
"step": 30500
},
{
"epoch": 3.601694915254237,
"grad_norm": 2.125,
"learning_rate": 5.155645928284558e-06,
"loss": 0.4494,
"step": 30600
},
{
"epoch": 3.6134651600753296,
"grad_norm": 6.96875,
"learning_rate": 4.857896198606138e-06,
"loss": 0.4412,
"step": 30700
},
{
"epoch": 3.6252354048964217,
"grad_norm": 15.8125,
"learning_rate": 4.568787974168554e-06,
"loss": 0.4351,
"step": 30800
},
{
"epoch": 3.6370056497175143,
"grad_norm": 4.46875,
"learning_rate": 4.288347513905944e-06,
"loss": 0.4331,
"step": 30900
},
{
"epoch": 3.6487758945386064,
"grad_norm": 7.3125,
"learning_rate": 4.0166002894823864e-06,
"loss": 0.4526,
"step": 31000
},
{
"epoch": 3.6605461393596985,
"grad_norm": 2.078125,
"learning_rate": 3.7535709829783607e-06,
"loss": 0.4527,
"step": 31100
},
{
"epoch": 3.672316384180791,
"grad_norm": 5.0,
"learning_rate": 3.499283484648952e-06,
"loss": 0.446,
"step": 31200
},
{
"epoch": 3.684086629001883,
"grad_norm": 9.0625,
"learning_rate": 3.2537608907540206e-06,
"loss": 0.4392,
"step": 31300
},
{
"epoch": 3.695856873822976,
"grad_norm": 9.1875,
"learning_rate": 3.0170255014603244e-06,
"loss": 0.4518,
"step": 31400
},
{
"epoch": 3.707627118644068,
"grad_norm": 6.5,
"learning_rate": 2.789098818816105e-06,
"loss": 0.4497,
"step": 31500
},
{
"epoch": 3.71939736346516,
"grad_norm": 8.25,
"learning_rate": 2.5700015447981507e-06,
"loss": 0.4575,
"step": 31600
},
{
"epoch": 3.7311676082862526,
"grad_norm": 5.40625,
"learning_rate": 2.359753579431423e-06,
"loss": 0.4382,
"step": 31700
},
{
"epoch": 3.7429378531073447,
"grad_norm": 6.0,
"learning_rate": 2.158374018981635e-06,
"loss": 0.4399,
"step": 31800
},
{
"epoch": 3.754708097928437,
"grad_norm": 3.984375,
"learning_rate": 1.9658811542207835e-06,
"loss": 0.4402,
"step": 31900
},
{
"epoch": 3.7664783427495294,
"grad_norm": 8.5,
"learning_rate": 1.7822924687658227e-06,
"loss": 0.4379,
"step": 32000
},
{
"epoch": 3.7782485875706215,
"grad_norm": 6.5,
"learning_rate": 1.60762463749069e-06,
"loss": 0.4482,
"step": 32100
},
{
"epoch": 3.7900188323917137,
"grad_norm": 3.546875,
"learning_rate": 1.4418935250117837e-06,
"loss": 0.4401,
"step": 32200
},
{
"epoch": 3.8017890772128062,
"grad_norm": 7.53125,
"learning_rate": 1.2851141842469937e-06,
"loss": 0.4488,
"step": 32300
},
{
"epoch": 3.8135593220338984,
"grad_norm": 2.796875,
"learning_rate": 1.1373008550485175e-06,
"loss": 0.4399,
"step": 32400
},
{
"epoch": 3.8253295668549905,
"grad_norm": 8.75,
"learning_rate": 9.9846696290945e-07,
"loss": 0.4341,
"step": 32500
},
{
"epoch": 3.837099811676083,
"grad_norm": 16.875,
"learning_rate": 8.686251177444371e-07,
"loss": 0.4489,
"step": 32600
},
{
"epoch": 3.848870056497175,
"grad_norm": 4.625,
"learning_rate": 7.477871127442803e-07,
"loss": 0.4342,
"step": 32700
},
{
"epoch": 3.8606403013182673,
"grad_norm": 6.8125,
"learning_rate": 6.35963923304872e-07,
"loss": 0.427,
"step": 32800
},
{
"epoch": 3.87241054613936,
"grad_norm": 4.125,
"learning_rate": 5.331657060302586e-07,
"loss": 0.4401,
"step": 32900
},
{
"epoch": 3.884180790960452,
"grad_norm": 18.0,
"learning_rate": 4.3940179781019055e-07,
"loss": 0.4432,
"step": 33000
},
{
"epoch": 3.895951035781544,
"grad_norm": 2.328125,
"learning_rate": 3.546807149720555e-07,
"loss": 0.439,
"step": 33100
},
{
"epoch": 3.9077212806026367,
"grad_norm": 3.171875,
"learning_rate": 2.7901015250737515e-07,
"loss": 0.4474,
"step": 33200
},
{
"epoch": 3.919491525423729,
"grad_norm": 7.40625,
"learning_rate": 2.123969833728534e-07,
"loss": 0.4568,
"step": 33300
},
{
"epoch": 3.931261770244821,
"grad_norm": 7.65625,
"learning_rate": 1.5484725786619748e-07,
"loss": 0.4505,
"step": 33400
},
{
"epoch": 3.9430320150659135,
"grad_norm": 11.125,
"learning_rate": 1.0636620307650225e-07,
"loss": 0.4382,
"step": 33500
},
{
"epoch": 3.9548022598870056,
"grad_norm": 3.703125,
"learning_rate": 6.695822240952998e-08,
"loss": 0.4416,
"step": 33600
},
{
"epoch": 3.9665725047080977,
"grad_norm": 1.734375,
"learning_rate": 3.6626895187785726e-08,
"loss": 0.4507,
"step": 33700
},
{
"epoch": 3.9783427495291903,
"grad_norm": 8.875,
"learning_rate": 1.5374976325366332e-08,
"loss": 0.4207,
"step": 33800
},
{
"epoch": 3.9901129943502824,
"grad_norm": 8.5,
"learning_rate": 3.204396077749472e-09,
"loss": 0.4692,
"step": 33900
}
],
"logging_steps": 100,
"max_steps": 33984,
"num_input_tokens_seen": 0,
"num_train_epochs": 4,
"save_steps": 4000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 1.2131684208845605e+19,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}