answer-emojis / checkpoint-363 /trainer_state.json
evayzh's picture
Training in progress, epoch 2, checkpoint
6c48917 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.97319587628866,
"eval_steps": 61,
"global_step": 363,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.008247422680412371,
"grad_norm": 2.2737574577331543,
"learning_rate": 2e-05,
"loss": 1.0155,
"step": 1
},
{
"epoch": 0.008247422680412371,
"eval_loss": 1.2302311658859253,
"eval_runtime": 295.4772,
"eval_samples_per_second": 2.031,
"eval_steps_per_second": 1.015,
"step": 1
},
{
"epoch": 0.016494845360824743,
"grad_norm": 2.203125476837158,
"learning_rate": 4e-05,
"loss": 1.1137,
"step": 2
},
{
"epoch": 0.024742268041237112,
"grad_norm": 1.7233558893203735,
"learning_rate": 6e-05,
"loss": 0.9453,
"step": 3
},
{
"epoch": 0.032989690721649485,
"grad_norm": 2.38301420211792,
"learning_rate": 8e-05,
"loss": 1.1091,
"step": 4
},
{
"epoch": 0.041237113402061855,
"grad_norm": 1.6754157543182373,
"learning_rate": 0.0001,
"loss": 0.9209,
"step": 5
},
{
"epoch": 0.049484536082474224,
"grad_norm": 1.6663925647735596,
"learning_rate": 0.00012,
"loss": 0.9145,
"step": 6
},
{
"epoch": 0.0577319587628866,
"grad_norm": 1.6516679525375366,
"learning_rate": 0.00014,
"loss": 0.8456,
"step": 7
},
{
"epoch": 0.06597938144329897,
"grad_norm": 1.4858132600784302,
"learning_rate": 0.00016,
"loss": 0.8257,
"step": 8
},
{
"epoch": 0.07422680412371134,
"grad_norm": 1.2829698324203491,
"learning_rate": 0.00018,
"loss": 0.7396,
"step": 9
},
{
"epoch": 0.08247422680412371,
"grad_norm": 0.8841626644134521,
"learning_rate": 0.0002,
"loss": 0.6599,
"step": 10
},
{
"epoch": 0.09072164948453608,
"grad_norm": 0.6680086255073547,
"learning_rate": 0.0001999960397967811,
"loss": 0.6368,
"step": 11
},
{
"epoch": 0.09896907216494845,
"grad_norm": 0.9510122537612915,
"learning_rate": 0.00019998415950078858,
"loss": 0.7061,
"step": 12
},
{
"epoch": 0.10721649484536082,
"grad_norm": 0.8464949131011963,
"learning_rate": 0.00019996436005299012,
"loss": 0.6674,
"step": 13
},
{
"epoch": 0.1154639175257732,
"grad_norm": 0.8255585432052612,
"learning_rate": 0.00019993664302158255,
"loss": 0.6971,
"step": 14
},
{
"epoch": 0.12371134020618557,
"grad_norm": 0.7613081932067871,
"learning_rate": 0.00019990101060186733,
"loss": 0.6621,
"step": 15
},
{
"epoch": 0.13195876288659794,
"grad_norm": 0.6413361430168152,
"learning_rate": 0.00019985746561607698,
"loss": 0.6336,
"step": 16
},
{
"epoch": 0.1402061855670103,
"grad_norm": 0.7216547727584839,
"learning_rate": 0.0001998060115131513,
"loss": 0.7165,
"step": 17
},
{
"epoch": 0.14845360824742268,
"grad_norm": 0.7801871299743652,
"learning_rate": 0.00019974665236846442,
"loss": 0.7449,
"step": 18
},
{
"epoch": 0.15670103092783505,
"grad_norm": 0.6427637338638306,
"learning_rate": 0.00019967939288350182,
"loss": 0.6344,
"step": 19
},
{
"epoch": 0.16494845360824742,
"grad_norm": 0.6753759980201721,
"learning_rate": 0.00019960423838548814,
"loss": 0.735,
"step": 20
},
{
"epoch": 0.1731958762886598,
"grad_norm": 0.6213053464889526,
"learning_rate": 0.00019952119482696503,
"loss": 0.6675,
"step": 21
},
{
"epoch": 0.18144329896907216,
"grad_norm": 0.6505969762802124,
"learning_rate": 0.00019943026878531983,
"loss": 0.6129,
"step": 22
},
{
"epoch": 0.18969072164948453,
"grad_norm": 0.6324624419212341,
"learning_rate": 0.0001993314674622646,
"loss": 0.6289,
"step": 23
},
{
"epoch": 0.1979381443298969,
"grad_norm": 0.6155864000320435,
"learning_rate": 0.00019922479868326578,
"loss": 0.6358,
"step": 24
},
{
"epoch": 0.20618556701030927,
"grad_norm": 0.6395933628082275,
"learning_rate": 0.0001991102708969241,
"loss": 0.6225,
"step": 25
},
{
"epoch": 0.21443298969072164,
"grad_norm": 0.5889492034912109,
"learning_rate": 0.00019898789317430575,
"loss": 0.6674,
"step": 26
},
{
"epoch": 0.22268041237113403,
"grad_norm": 0.5262942910194397,
"learning_rate": 0.00019885767520822376,
"loss": 0.6077,
"step": 27
},
{
"epoch": 0.2309278350515464,
"grad_norm": 0.5878570675849915,
"learning_rate": 0.0001987196273124703,
"loss": 0.6145,
"step": 28
},
{
"epoch": 0.23917525773195877,
"grad_norm": 0.5845457315444946,
"learning_rate": 0.00019857376042099983,
"loss": 0.6435,
"step": 29
},
{
"epoch": 0.24742268041237114,
"grad_norm": 0.6508314609527588,
"learning_rate": 0.00019842008608706295,
"loss": 0.6799,
"step": 30
},
{
"epoch": 0.2556701030927835,
"grad_norm": 0.5770230293273926,
"learning_rate": 0.00019825861648229152,
"loss": 0.6092,
"step": 31
},
{
"epoch": 0.2639175257731959,
"grad_norm": 0.5438909530639648,
"learning_rate": 0.00019808936439573454,
"loss": 0.5952,
"step": 32
},
{
"epoch": 0.2721649484536082,
"grad_norm": 0.5276155471801758,
"learning_rate": 0.00019791234323284513,
"loss": 0.5899,
"step": 33
},
{
"epoch": 0.2804123711340206,
"grad_norm": 0.6430350542068481,
"learning_rate": 0.00019772756701441887,
"loss": 0.6581,
"step": 34
},
{
"epoch": 0.28865979381443296,
"grad_norm": 0.6087874174118042,
"learning_rate": 0.0001975350503754833,
"loss": 0.654,
"step": 35
},
{
"epoch": 0.29690721649484536,
"grad_norm": 0.610534131526947,
"learning_rate": 0.00019733480856413868,
"loss": 0.6655,
"step": 36
},
{
"epoch": 0.30515463917525776,
"grad_norm": 0.5277837514877319,
"learning_rate": 0.0001971268574403503,
"loss": 0.6228,
"step": 37
},
{
"epoch": 0.3134020618556701,
"grad_norm": 0.6729940176010132,
"learning_rate": 0.00019691121347469235,
"loss": 0.8324,
"step": 38
},
{
"epoch": 0.3216494845360825,
"grad_norm": 0.541552722454071,
"learning_rate": 0.00019668789374704338,
"loss": 0.54,
"step": 39
},
{
"epoch": 0.32989690721649484,
"grad_norm": 0.5642793774604797,
"learning_rate": 0.0001964569159452335,
"loss": 0.641,
"step": 40
},
{
"epoch": 0.33814432989690724,
"grad_norm": 0.5639921426773071,
"learning_rate": 0.00019621829836364337,
"loss": 0.6042,
"step": 41
},
{
"epoch": 0.3463917525773196,
"grad_norm": 0.566059410572052,
"learning_rate": 0.00019597205990175525,
"loss": 0.6364,
"step": 42
},
{
"epoch": 0.354639175257732,
"grad_norm": 0.5911106467247009,
"learning_rate": 0.00019571822006265622,
"loss": 0.6777,
"step": 43
},
{
"epoch": 0.3628865979381443,
"grad_norm": 0.5940718054771423,
"learning_rate": 0.00019545679895149315,
"loss": 0.713,
"step": 44
},
{
"epoch": 0.3711340206185567,
"grad_norm": 0.46311718225479126,
"learning_rate": 0.0001951878172738806,
"loss": 0.5298,
"step": 45
},
{
"epoch": 0.37938144329896906,
"grad_norm": 0.5446880459785461,
"learning_rate": 0.00019491129633426068,
"loss": 0.546,
"step": 46
},
{
"epoch": 0.38762886597938145,
"grad_norm": 0.6126728653907776,
"learning_rate": 0.00019462725803421566,
"loss": 0.5725,
"step": 47
},
{
"epoch": 0.3958762886597938,
"grad_norm": 0.49817898869514465,
"learning_rate": 0.0001943357248707334,
"loss": 0.6224,
"step": 48
},
{
"epoch": 0.4041237113402062,
"grad_norm": 0.4865269362926483,
"learning_rate": 0.0001940367199344253,
"loss": 0.544,
"step": 49
},
{
"epoch": 0.41237113402061853,
"grad_norm": 0.48718348145484924,
"learning_rate": 0.00019373026690769763,
"loss": 0.6132,
"step": 50
},
{
"epoch": 0.42061855670103093,
"grad_norm": 0.5080978274345398,
"learning_rate": 0.0001934163900628756,
"loss": 0.6278,
"step": 51
},
{
"epoch": 0.4288659793814433,
"grad_norm": 0.4930504262447357,
"learning_rate": 0.00019309511426028104,
"loss": 0.602,
"step": 52
},
{
"epoch": 0.43711340206185567,
"grad_norm": 0.4977351725101471,
"learning_rate": 0.00019276646494626332,
"loss": 0.5516,
"step": 53
},
{
"epoch": 0.44536082474226807,
"grad_norm": 0.4758591055870056,
"learning_rate": 0.00019243046815118386,
"loss": 0.5136,
"step": 54
},
{
"epoch": 0.4536082474226804,
"grad_norm": 0.4887770712375641,
"learning_rate": 0.00019208715048735445,
"loss": 0.6265,
"step": 55
},
{
"epoch": 0.4618556701030928,
"grad_norm": 0.5064010620117188,
"learning_rate": 0.00019173653914692946,
"loss": 0.6524,
"step": 56
},
{
"epoch": 0.47010309278350515,
"grad_norm": 0.5001596808433533,
"learning_rate": 0.00019137866189975202,
"loss": 0.622,
"step": 57
},
{
"epoch": 0.47835051546391755,
"grad_norm": 0.4940868616104126,
"learning_rate": 0.00019101354709115468,
"loss": 0.6362,
"step": 58
},
{
"epoch": 0.4865979381443299,
"grad_norm": 0.4839000701904297,
"learning_rate": 0.00019064122363971427,
"loss": 0.6223,
"step": 59
},
{
"epoch": 0.4948453608247423,
"grad_norm": 0.4475175440311432,
"learning_rate": 0.00019026172103496137,
"loss": 0.5625,
"step": 60
},
{
"epoch": 0.5030927835051546,
"grad_norm": 0.4716227352619171,
"learning_rate": 0.0001898750693350447,
"loss": 0.5161,
"step": 61
},
{
"epoch": 0.5030927835051546,
"eval_loss": 0.5744044780731201,
"eval_runtime": 296.9417,
"eval_samples_per_second": 2.021,
"eval_steps_per_second": 1.01,
"step": 61
},
{
"epoch": 0.511340206185567,
"grad_norm": 0.47013187408447266,
"learning_rate": 0.00018948129916435046,
"loss": 0.592,
"step": 62
},
{
"epoch": 0.5195876288659794,
"grad_norm": 0.44856756925582886,
"learning_rate": 0.00018908044171107657,
"loss": 0.5981,
"step": 63
},
{
"epoch": 0.5278350515463918,
"grad_norm": 0.4443698525428772,
"learning_rate": 0.00018867252872476257,
"loss": 0.6068,
"step": 64
},
{
"epoch": 0.5360824742268041,
"grad_norm": 0.423221617937088,
"learning_rate": 0.00018825759251377483,
"loss": 0.537,
"step": 65
},
{
"epoch": 0.5443298969072164,
"grad_norm": 0.4647126793861389,
"learning_rate": 0.00018783566594274783,
"loss": 0.5896,
"step": 66
},
{
"epoch": 0.5525773195876289,
"grad_norm": 0.4287152588367462,
"learning_rate": 0.00018740678242998077,
"loss": 0.5424,
"step": 67
},
{
"epoch": 0.5608247422680412,
"grad_norm": 0.4562716484069824,
"learning_rate": 0.00018697097594479103,
"loss": 0.5571,
"step": 68
},
{
"epoch": 0.5690721649484536,
"grad_norm": 0.37922483682632446,
"learning_rate": 0.0001865282810048235,
"loss": 0.5337,
"step": 69
},
{
"epoch": 0.5773195876288659,
"grad_norm": 0.4227437973022461,
"learning_rate": 0.0001860787326733168,
"loss": 0.5989,
"step": 70
},
{
"epoch": 0.5855670103092784,
"grad_norm": 0.39743122458457947,
"learning_rate": 0.0001856223665563258,
"loss": 0.537,
"step": 71
},
{
"epoch": 0.5938144329896907,
"grad_norm": 0.41971203684806824,
"learning_rate": 0.00018515921879990187,
"loss": 0.5353,
"step": 72
},
{
"epoch": 0.6020618556701031,
"grad_norm": 0.4010031819343567,
"learning_rate": 0.00018468932608722973,
"loss": 0.5954,
"step": 73
},
{
"epoch": 0.6103092783505155,
"grad_norm": 0.469838947057724,
"learning_rate": 0.000184212725635722,
"loss": 0.6942,
"step": 74
},
{
"epoch": 0.6185567010309279,
"grad_norm": 0.381673663854599,
"learning_rate": 0.00018372945519407158,
"loss": 0.557,
"step": 75
},
{
"epoch": 0.6268041237113402,
"grad_norm": 0.3736635446548462,
"learning_rate": 0.00018323955303926163,
"loss": 0.5174,
"step": 76
},
{
"epoch": 0.6350515463917525,
"grad_norm": 0.4390040338039398,
"learning_rate": 0.00018274305797353395,
"loss": 0.6249,
"step": 77
},
{
"epoch": 0.643298969072165,
"grad_norm": 0.4298490285873413,
"learning_rate": 0.00018224000932131568,
"loss": 0.5682,
"step": 78
},
{
"epoch": 0.6515463917525773,
"grad_norm": 0.3995118737220764,
"learning_rate": 0.00018173044692610467,
"loss": 0.5261,
"step": 79
},
{
"epoch": 0.6597938144329897,
"grad_norm": 0.41546231508255005,
"learning_rate": 0.00018121441114731367,
"loss": 0.5296,
"step": 80
},
{
"epoch": 0.668041237113402,
"grad_norm": 0.4129379093647003,
"learning_rate": 0.0001806919428570737,
"loss": 0.5576,
"step": 81
},
{
"epoch": 0.6762886597938145,
"grad_norm": 0.4260002672672272,
"learning_rate": 0.00018016308343699687,
"loss": 0.6042,
"step": 82
},
{
"epoch": 0.6845360824742268,
"grad_norm": 0.3927561640739441,
"learning_rate": 0.00017962787477489878,
"loss": 0.5349,
"step": 83
},
{
"epoch": 0.6927835051546392,
"grad_norm": 0.39806240797042847,
"learning_rate": 0.00017908635926148069,
"loss": 0.5386,
"step": 84
},
{
"epoch": 0.7010309278350515,
"grad_norm": 0.3762768805027008,
"learning_rate": 0.00017853857978697223,
"loss": 0.545,
"step": 85
},
{
"epoch": 0.709278350515464,
"grad_norm": 0.4306787848472595,
"learning_rate": 0.00017798457973773417,
"loss": 0.6305,
"step": 86
},
{
"epoch": 0.7175257731958763,
"grad_norm": 0.3811737596988678,
"learning_rate": 0.00017742440299282203,
"loss": 0.539,
"step": 87
},
{
"epoch": 0.7257731958762886,
"grad_norm": 0.39163580536842346,
"learning_rate": 0.00017685809392051083,
"loss": 0.5286,
"step": 88
},
{
"epoch": 0.734020618556701,
"grad_norm": 0.4237455725669861,
"learning_rate": 0.00017628569737478076,
"loss": 0.5801,
"step": 89
},
{
"epoch": 0.7422680412371134,
"grad_norm": 0.41603776812553406,
"learning_rate": 0.00017570725869176467,
"loss": 0.6001,
"step": 90
},
{
"epoch": 0.7505154639175258,
"grad_norm": 0.3811546862125397,
"learning_rate": 0.00017512282368615728,
"loss": 0.5942,
"step": 91
},
{
"epoch": 0.7587628865979381,
"grad_norm": 0.3982357680797577,
"learning_rate": 0.00017453243864758638,
"loss": 0.5655,
"step": 92
},
{
"epoch": 0.7670103092783506,
"grad_norm": 0.43069708347320557,
"learning_rate": 0.00017393615033694656,
"loss": 0.5251,
"step": 93
},
{
"epoch": 0.7752577319587629,
"grad_norm": 0.3952259421348572,
"learning_rate": 0.0001733340059826956,
"loss": 0.5462,
"step": 94
},
{
"epoch": 0.7835051546391752,
"grad_norm": 0.40667715668678284,
"learning_rate": 0.00017272605327711365,
"loss": 0.5303,
"step": 95
},
{
"epoch": 0.7917525773195876,
"grad_norm": 0.3917592465877533,
"learning_rate": 0.000172112340372526,
"loss": 0.612,
"step": 96
},
{
"epoch": 0.8,
"grad_norm": 0.37455278635025024,
"learning_rate": 0.00017149291587748898,
"loss": 0.4831,
"step": 97
},
{
"epoch": 0.8082474226804124,
"grad_norm": 0.3805007040500641,
"learning_rate": 0.00017086782885294025,
"loss": 0.5237,
"step": 98
},
{
"epoch": 0.8164948453608247,
"grad_norm": 0.4057518541812897,
"learning_rate": 0.0001702371288083127,
"loss": 0.611,
"step": 99
},
{
"epoch": 0.8247422680412371,
"grad_norm": 0.424902081489563,
"learning_rate": 0.00016960086569761332,
"loss": 0.5071,
"step": 100
},
{
"epoch": 0.8329896907216495,
"grad_norm": 0.41015303134918213,
"learning_rate": 0.0001689590899154664,
"loss": 0.6073,
"step": 101
},
{
"epoch": 0.8412371134020619,
"grad_norm": 0.3922909200191498,
"learning_rate": 0.00016831185229312237,
"loss": 0.5923,
"step": 102
},
{
"epoch": 0.8494845360824742,
"grad_norm": 0.43730470538139343,
"learning_rate": 0.0001676592040944315,
"loss": 0.5937,
"step": 103
},
{
"epoch": 0.8577319587628865,
"grad_norm": 0.3726069927215576,
"learning_rate": 0.0001670011970117838,
"loss": 0.4912,
"step": 104
},
{
"epoch": 0.865979381443299,
"grad_norm": 0.404433012008667,
"learning_rate": 0.00016633788316201454,
"loss": 0.474,
"step": 105
},
{
"epoch": 0.8742268041237113,
"grad_norm": 0.41312918066978455,
"learning_rate": 0.0001656693150822766,
"loss": 0.5545,
"step": 106
},
{
"epoch": 0.8824742268041237,
"grad_norm": 0.42073437571525574,
"learning_rate": 0.0001649955457258792,
"loss": 0.5488,
"step": 107
},
{
"epoch": 0.8907216494845361,
"grad_norm": 0.39594408869743347,
"learning_rate": 0.00016431662845809388,
"loss": 0.5798,
"step": 108
},
{
"epoch": 0.8989690721649485,
"grad_norm": 0.41331595182418823,
"learning_rate": 0.00016363261705192757,
"loss": 0.5188,
"step": 109
},
{
"epoch": 0.9072164948453608,
"grad_norm": 0.3879910111427307,
"learning_rate": 0.00016294356568386369,
"loss": 0.5676,
"step": 110
},
{
"epoch": 0.9154639175257732,
"grad_norm": 0.41248220205307007,
"learning_rate": 0.00016224952892957123,
"loss": 0.541,
"step": 111
},
{
"epoch": 0.9237113402061856,
"grad_norm": 0.3406890034675598,
"learning_rate": 0.0001615505617595819,
"loss": 0.4902,
"step": 112
},
{
"epoch": 0.931958762886598,
"grad_norm": 0.3741547167301178,
"learning_rate": 0.00016084671953493643,
"loss": 0.5247,
"step": 113
},
{
"epoch": 0.9402061855670103,
"grad_norm": 0.37796300649642944,
"learning_rate": 0.00016013805800279976,
"loss": 0.5091,
"step": 114
},
{
"epoch": 0.9484536082474226,
"grad_norm": 0.3957986533641815,
"learning_rate": 0.00015942463329204546,
"loss": 0.5809,
"step": 115
},
{
"epoch": 0.9567010309278351,
"grad_norm": 0.38373249769210815,
"learning_rate": 0.00015870650190881022,
"loss": 0.4693,
"step": 116
},
{
"epoch": 0.9649484536082474,
"grad_norm": 0.4250248074531555,
"learning_rate": 0.00015798372073201836,
"loss": 0.5668,
"step": 117
},
{
"epoch": 0.9731958762886598,
"grad_norm": 0.37807247042655945,
"learning_rate": 0.00015725634700887678,
"loss": 0.5159,
"step": 118
},
{
"epoch": 0.9814432989690721,
"grad_norm": 0.3726092278957367,
"learning_rate": 0.00015652443835034068,
"loss": 0.5605,
"step": 119
},
{
"epoch": 0.9896907216494846,
"grad_norm": 0.38076040148735046,
"learning_rate": 0.0001557880527265505,
"loss": 0.537,
"step": 120
},
{
"epoch": 0.9979381443298969,
"grad_norm": 0.37176623940467834,
"learning_rate": 0.00015504724846224064,
"loss": 0.5603,
"step": 121
},
{
"epoch": 1.0061855670103093,
"grad_norm": 0.36700350046157837,
"learning_rate": 0.00015430208423211975,
"loss": 0.5398,
"step": 122
},
{
"epoch": 1.0061855670103093,
"eval_loss": 0.5378922820091248,
"eval_runtime": 296.9545,
"eval_samples_per_second": 2.021,
"eval_steps_per_second": 1.01,
"step": 122
},
{
"epoch": 1.0041237113402062,
"grad_norm": 0.35800787806510925,
"learning_rate": 0.00015355261905622343,
"loss": 0.5065,
"step": 123
},
{
"epoch": 1.0123711340206185,
"grad_norm": 0.32381781935691833,
"learning_rate": 0.0001527989122952398,
"loss": 0.4246,
"step": 124
},
{
"epoch": 1.0206185567010309,
"grad_norm": 0.34609419107437134,
"learning_rate": 0.00015204102364580765,
"loss": 0.4654,
"step": 125
},
{
"epoch": 1.0288659793814432,
"grad_norm": 0.3360902965068817,
"learning_rate": 0.00015127901313578831,
"loss": 0.424,
"step": 126
},
{
"epoch": 1.0371134020618558,
"grad_norm": 0.3538791835308075,
"learning_rate": 0.00015051294111951134,
"loss": 0.4261,
"step": 127
},
{
"epoch": 1.045360824742268,
"grad_norm": 0.36259981989860535,
"learning_rate": 0.000149742868272994,
"loss": 0.4162,
"step": 128
},
{
"epoch": 1.0536082474226804,
"grad_norm": 0.3865598440170288,
"learning_rate": 0.00014896885558913562,
"loss": 0.4767,
"step": 129
},
{
"epoch": 1.0618556701030928,
"grad_norm": 0.34752511978149414,
"learning_rate": 0.00014819096437288664,
"loss": 0.4461,
"step": 130
},
{
"epoch": 1.0701030927835051,
"grad_norm": 0.3833150565624237,
"learning_rate": 0.000147409256236393,
"loss": 0.469,
"step": 131
},
{
"epoch": 1.0783505154639175,
"grad_norm": 0.3767507076263428,
"learning_rate": 0.0001466237930941163,
"loss": 0.4165,
"step": 132
},
{
"epoch": 1.0865979381443298,
"grad_norm": 0.47273167967796326,
"learning_rate": 0.00014583463715792984,
"loss": 0.6179,
"step": 133
},
{
"epoch": 1.0948453608247424,
"grad_norm": 0.3700537383556366,
"learning_rate": 0.00014504185093219116,
"loss": 0.4509,
"step": 134
},
{
"epoch": 1.1030927835051547,
"grad_norm": 0.4104892611503601,
"learning_rate": 0.0001442454972087915,
"loss": 0.4542,
"step": 135
},
{
"epoch": 1.111340206185567,
"grad_norm": 0.39110255241394043,
"learning_rate": 0.00014344563906218256,
"loss": 0.4679,
"step": 136
},
{
"epoch": 1.1195876288659794,
"grad_norm": 0.42041921615600586,
"learning_rate": 0.0001426423398443803,
"loss": 0.4453,
"step": 137
},
{
"epoch": 1.1278350515463917,
"grad_norm": 0.4177539050579071,
"learning_rate": 0.0001418356631799478,
"loss": 0.4112,
"step": 138
},
{
"epoch": 1.136082474226804,
"grad_norm": 0.4446333944797516,
"learning_rate": 0.00014102567296095551,
"loss": 0.4735,
"step": 139
},
{
"epoch": 1.1443298969072164,
"grad_norm": 0.39587873220443726,
"learning_rate": 0.00014021243334192082,
"loss": 0.4636,
"step": 140
},
{
"epoch": 1.1525773195876288,
"grad_norm": 0.415682852268219,
"learning_rate": 0.00013939600873472694,
"loss": 0.4534,
"step": 141
},
{
"epoch": 1.1608247422680413,
"grad_norm": 0.3811276853084564,
"learning_rate": 0.00013857646380352102,
"loss": 0.4555,
"step": 142
},
{
"epoch": 1.1690721649484537,
"grad_norm": 0.3821997046470642,
"learning_rate": 0.00013775386345959246,
"loss": 0.4381,
"step": 143
},
{
"epoch": 1.177319587628866,
"grad_norm": 0.3770444989204407,
"learning_rate": 0.00013692827285623197,
"loss": 0.4064,
"step": 144
},
{
"epoch": 1.1855670103092784,
"grad_norm": 0.43066269159317017,
"learning_rate": 0.0001360997573835708,
"loss": 0.4532,
"step": 145
},
{
"epoch": 1.1938144329896907,
"grad_norm": 0.36327502131462097,
"learning_rate": 0.00013526838266340177,
"loss": 0.3665,
"step": 146
},
{
"epoch": 1.202061855670103,
"grad_norm": 0.40168052911758423,
"learning_rate": 0.00013443421454398174,
"loss": 0.4727,
"step": 147
},
{
"epoch": 1.2103092783505154,
"grad_norm": 0.39729252457618713,
"learning_rate": 0.00013359731909481616,
"loss": 0.4236,
"step": 148
},
{
"epoch": 1.218556701030928,
"grad_norm": 0.39847901463508606,
"learning_rate": 0.00013275776260142608,
"loss": 0.4036,
"step": 149
},
{
"epoch": 1.2268041237113403,
"grad_norm": 0.3883432447910309,
"learning_rate": 0.00013191561156009803,
"loss": 0.4061,
"step": 150
},
{
"epoch": 1.2350515463917526,
"grad_norm": 0.3763248026371002,
"learning_rate": 0.0001310709326726173,
"loss": 0.3907,
"step": 151
},
{
"epoch": 1.243298969072165,
"grad_norm": 0.4451892673969269,
"learning_rate": 0.00013022379284098487,
"loss": 0.483,
"step": 152
},
{
"epoch": 1.2515463917525773,
"grad_norm": 0.4433348774909973,
"learning_rate": 0.00012937425916211852,
"loss": 0.4228,
"step": 153
},
{
"epoch": 1.2597938144329897,
"grad_norm": 0.42469412088394165,
"learning_rate": 0.00012852239892253842,
"loss": 0.4543,
"step": 154
},
{
"epoch": 1.268041237113402,
"grad_norm": 0.41997918486595154,
"learning_rate": 0.00012766827959303787,
"loss": 0.3975,
"step": 155
},
{
"epoch": 1.2762886597938143,
"grad_norm": 0.4060366749763489,
"learning_rate": 0.00012681196882333916,
"loss": 0.4245,
"step": 156
},
{
"epoch": 1.2845360824742267,
"grad_norm": 0.4799356162548065,
"learning_rate": 0.0001259535344367357,
"loss": 0.3823,
"step": 157
},
{
"epoch": 1.2927835051546392,
"grad_norm": 0.387761652469635,
"learning_rate": 0.00012509304442471985,
"loss": 0.4591,
"step": 158
},
{
"epoch": 1.3010309278350516,
"grad_norm": 0.4064452350139618,
"learning_rate": 0.0001242305669415979,
"loss": 0.439,
"step": 159
},
{
"epoch": 1.309278350515464,
"grad_norm": 0.39760932326316833,
"learning_rate": 0.00012336617029909205,
"loss": 0.4017,
"step": 160
},
{
"epoch": 1.3175257731958763,
"grad_norm": 0.43200570344924927,
"learning_rate": 0.00012249992296092956,
"loss": 0.4423,
"step": 161
},
{
"epoch": 1.3257731958762886,
"grad_norm": 0.43232297897338867,
"learning_rate": 0.00012163189353742035,
"loss": 0.4342,
"step": 162
},
{
"epoch": 1.334020618556701,
"grad_norm": 0.43833640217781067,
"learning_rate": 0.00012076215078002278,
"loss": 0.5219,
"step": 163
},
{
"epoch": 1.3422680412371135,
"grad_norm": 0.40365973114967346,
"learning_rate": 0.0001198907635758982,
"loss": 0.4123,
"step": 164
},
{
"epoch": 1.3505154639175259,
"grad_norm": 0.43913137912750244,
"learning_rate": 0.00011901780094245483,
"loss": 0.4149,
"step": 165
},
{
"epoch": 1.3587628865979382,
"grad_norm": 0.4219210147857666,
"learning_rate": 0.00011814333202188126,
"loss": 0.4127,
"step": 166
},
{
"epoch": 1.3670103092783505,
"grad_norm": 0.42331403493881226,
"learning_rate": 0.0001172674260756702,
"loss": 0.3926,
"step": 167
},
{
"epoch": 1.3752577319587629,
"grad_norm": 0.3824922442436218,
"learning_rate": 0.00011639015247913261,
"loss": 0.3781,
"step": 168
},
{
"epoch": 1.3835051546391752,
"grad_norm": 0.4475767910480499,
"learning_rate": 0.0001155115807159029,
"loss": 0.5629,
"step": 169
},
{
"epoch": 1.3917525773195876,
"grad_norm": 0.36695027351379395,
"learning_rate": 0.00011463178037243554,
"loss": 0.3622,
"step": 170
},
{
"epoch": 1.4,
"grad_norm": 0.41651466488838196,
"learning_rate": 0.0001137508211324936,
"loss": 0.4043,
"step": 171
},
{
"epoch": 1.4082474226804123,
"grad_norm": 0.4297608733177185,
"learning_rate": 0.00011286877277162943,
"loss": 0.4575,
"step": 172
},
{
"epoch": 1.4164948453608248,
"grad_norm": 0.441707968711853,
"learning_rate": 0.00011198570515165822,
"loss": 0.4435,
"step": 173
},
{
"epoch": 1.4247422680412372,
"grad_norm": 0.4173315465450287,
"learning_rate": 0.00011110168821512452,
"loss": 0.4367,
"step": 174
},
{
"epoch": 1.4329896907216495,
"grad_norm": 0.42195403575897217,
"learning_rate": 0.00011021679197976274,
"loss": 0.4258,
"step": 175
},
{
"epoch": 1.4412371134020618,
"grad_norm": 0.394330769777298,
"learning_rate": 0.00010933108653295128,
"loss": 0.4329,
"step": 176
},
{
"epoch": 1.4494845360824742,
"grad_norm": 0.4263797998428345,
"learning_rate": 0.00010844464202616127,
"loss": 0.475,
"step": 177
},
{
"epoch": 1.4577319587628865,
"grad_norm": 0.36980873346328735,
"learning_rate": 0.00010755752866940062,
"loss": 0.3833,
"step": 178
},
{
"epoch": 1.465979381443299,
"grad_norm": 0.42231741547584534,
"learning_rate": 0.0001066698167256527,
"loss": 0.4716,
"step": 179
},
{
"epoch": 1.4742268041237114,
"grad_norm": 0.40805140137672424,
"learning_rate": 0.00010578157650531146,
"loss": 0.4483,
"step": 180
},
{
"epoch": 1.4824742268041238,
"grad_norm": 0.4024035334587097,
"learning_rate": 0.00010489287836061246,
"loss": 0.4319,
"step": 181
},
{
"epoch": 1.4907216494845361,
"grad_norm": 0.38521435856819153,
"learning_rate": 0.00010400379268006082,
"loss": 0.4082,
"step": 182
},
{
"epoch": 1.4989690721649485,
"grad_norm": 0.4192781448364258,
"learning_rate": 0.00010311438988285598,
"loss": 0.4614,
"step": 183
},
{
"epoch": 1.4989690721649485,
"eval_loss": 0.5294703245162964,
"eval_runtime": 297.0142,
"eval_samples_per_second": 2.02,
"eval_steps_per_second": 1.01,
"step": 183
},
{
"epoch": 1.5072164948453608,
"grad_norm": 0.4073619246482849,
"learning_rate": 0.00010222474041331436,
"loss": 0.4335,
"step": 184
},
{
"epoch": 1.5154639175257731,
"grad_norm": 0.3790091574192047,
"learning_rate": 0.0001013349147352898,
"loss": 0.3892,
"step": 185
},
{
"epoch": 1.5237113402061855,
"grad_norm": 0.39920473098754883,
"learning_rate": 0.00010044498332659264,
"loss": 0.416,
"step": 186
},
{
"epoch": 1.5319587628865978,
"grad_norm": 0.37349697947502136,
"learning_rate": 9.955501667340741e-05,
"loss": 0.3943,
"step": 187
},
{
"epoch": 1.5402061855670102,
"grad_norm": 0.3844608962535858,
"learning_rate": 9.866508526471023e-05,
"loss": 0.4057,
"step": 188
},
{
"epoch": 1.5484536082474227,
"grad_norm": 0.40029478073120117,
"learning_rate": 9.77752595866857e-05,
"loss": 0.429,
"step": 189
},
{
"epoch": 1.556701030927835,
"grad_norm": 0.42949914932250977,
"learning_rate": 9.688561011714404e-05,
"loss": 0.4247,
"step": 190
},
{
"epoch": 1.5649484536082474,
"grad_norm": 0.3950637876987457,
"learning_rate": 9.599620731993922e-05,
"loss": 0.4358,
"step": 191
},
{
"epoch": 1.5731958762886598,
"grad_norm": 0.4211474359035492,
"learning_rate": 9.510712163938755e-05,
"loss": 0.4485,
"step": 192
},
{
"epoch": 1.5814432989690723,
"grad_norm": 0.41278165578842163,
"learning_rate": 9.421842349468855e-05,
"loss": 0.458,
"step": 193
},
{
"epoch": 1.5896907216494847,
"grad_norm": 0.4086206555366516,
"learning_rate": 9.333018327434731e-05,
"loss": 0.4295,
"step": 194
},
{
"epoch": 1.597938144329897,
"grad_norm": 0.38980671763420105,
"learning_rate": 9.244247133059938e-05,
"loss": 0.3858,
"step": 195
},
{
"epoch": 1.6061855670103093,
"grad_norm": 0.4090917706489563,
"learning_rate": 9.155535797383874e-05,
"loss": 0.3678,
"step": 196
},
{
"epoch": 1.6144329896907217,
"grad_norm": 0.4124603271484375,
"learning_rate": 9.066891346704875e-05,
"loss": 0.4203,
"step": 197
},
{
"epoch": 1.622680412371134,
"grad_norm": 0.4070824384689331,
"learning_rate": 8.978320802023731e-05,
"loss": 0.4558,
"step": 198
},
{
"epoch": 1.6309278350515464,
"grad_norm": 0.376213401556015,
"learning_rate": 8.88983117848755e-05,
"loss": 0.3928,
"step": 199
},
{
"epoch": 1.6391752577319587,
"grad_norm": 0.4122854769229889,
"learning_rate": 8.801429484834183e-05,
"loss": 0.4521,
"step": 200
},
{
"epoch": 1.647422680412371,
"grad_norm": 0.38308247923851013,
"learning_rate": 8.713122722837058e-05,
"loss": 0.3727,
"step": 201
},
{
"epoch": 1.6556701030927834,
"grad_norm": 0.4036957919597626,
"learning_rate": 8.624917886750638e-05,
"loss": 0.4497,
"step": 202
},
{
"epoch": 1.6639175257731957,
"grad_norm": 0.44214046001434326,
"learning_rate": 8.536821962756447e-05,
"loss": 0.5122,
"step": 203
},
{
"epoch": 1.6721649484536083,
"grad_norm": 0.40875983238220215,
"learning_rate": 8.448841928409711e-05,
"loss": 0.4393,
"step": 204
},
{
"epoch": 1.6804123711340206,
"grad_norm": 0.42527636885643005,
"learning_rate": 8.360984752086743e-05,
"loss": 0.4545,
"step": 205
},
{
"epoch": 1.688659793814433,
"grad_norm": 0.433173269033432,
"learning_rate": 8.273257392432981e-05,
"loss": 0.4867,
"step": 206
},
{
"epoch": 1.6969072164948453,
"grad_norm": 0.386598140001297,
"learning_rate": 8.185666797811878e-05,
"loss": 0.4253,
"step": 207
},
{
"epoch": 1.705154639175258,
"grad_norm": 0.40319326519966125,
"learning_rate": 8.09821990575452e-05,
"loss": 0.3935,
"step": 208
},
{
"epoch": 1.7134020618556702,
"grad_norm": 0.4310265779495239,
"learning_rate": 8.010923642410184e-05,
"loss": 0.4184,
"step": 209
},
{
"epoch": 1.7216494845360826,
"grad_norm": 0.3964579999446869,
"learning_rate": 7.923784921997726e-05,
"loss": 0.4449,
"step": 210
},
{
"epoch": 1.729896907216495,
"grad_norm": 0.39828160405158997,
"learning_rate": 7.836810646257971e-05,
"loss": 0.4227,
"step": 211
},
{
"epoch": 1.7381443298969073,
"grad_norm": 0.3750438392162323,
"learning_rate": 7.750007703907046e-05,
"loss": 0.3746,
"step": 212
},
{
"epoch": 1.7463917525773196,
"grad_norm": 0.40668681263923645,
"learning_rate": 7.663382970090795e-05,
"loss": 0.4351,
"step": 213
},
{
"epoch": 1.754639175257732,
"grad_norm": 0.4102826714515686,
"learning_rate": 7.57694330584021e-05,
"loss": 0.3999,
"step": 214
},
{
"epoch": 1.7628865979381443,
"grad_norm": 0.4256535470485687,
"learning_rate": 7.490695557528016e-05,
"loss": 0.4321,
"step": 215
},
{
"epoch": 1.7711340206185566,
"grad_norm": 0.3946252167224884,
"learning_rate": 7.404646556326433e-05,
"loss": 0.4323,
"step": 216
},
{
"epoch": 1.779381443298969,
"grad_norm": 0.3808107078075409,
"learning_rate": 7.318803117666084e-05,
"loss": 0.3942,
"step": 217
},
{
"epoch": 1.7876288659793813,
"grad_norm": 0.420895516872406,
"learning_rate": 7.233172040696216e-05,
"loss": 0.4466,
"step": 218
},
{
"epoch": 1.7958762886597937,
"grad_norm": 0.3973442614078522,
"learning_rate": 7.14776010774616e-05,
"loss": 0.4455,
"step": 219
},
{
"epoch": 1.8041237113402062,
"grad_norm": 0.40194034576416016,
"learning_rate": 7.062574083788152e-05,
"loss": 0.4622,
"step": 220
},
{
"epoch": 1.8123711340206186,
"grad_norm": 0.37952494621276855,
"learning_rate": 6.977620715901514e-05,
"loss": 0.3899,
"step": 221
},
{
"epoch": 1.820618556701031,
"grad_norm": 0.43268251419067383,
"learning_rate": 6.892906732738271e-05,
"loss": 0.4555,
"step": 222
},
{
"epoch": 1.8288659793814435,
"grad_norm": 0.39225122332572937,
"learning_rate": 6.8084388439902e-05,
"loss": 0.3727,
"step": 223
},
{
"epoch": 1.8371134020618558,
"grad_norm": 0.39572715759277344,
"learning_rate": 6.724223739857392e-05,
"loss": 0.4696,
"step": 224
},
{
"epoch": 1.8453608247422681,
"grad_norm": 0.4312278628349304,
"learning_rate": 6.640268090518385e-05,
"loss": 0.492,
"step": 225
},
{
"epoch": 1.8536082474226805,
"grad_norm": 0.40482354164123535,
"learning_rate": 6.556578545601829e-05,
"loss": 0.4342,
"step": 226
},
{
"epoch": 1.8618556701030928,
"grad_norm": 0.3752698302268982,
"learning_rate": 6.473161733659828e-05,
"loss": 0.4197,
"step": 227
},
{
"epoch": 1.8701030927835052,
"grad_norm": 0.41621121764183044,
"learning_rate": 6.390024261642922e-05,
"loss": 0.4895,
"step": 228
},
{
"epoch": 1.8783505154639175,
"grad_norm": 0.3930753767490387,
"learning_rate": 6.307172714376808e-05,
"loss": 0.4508,
"step": 229
},
{
"epoch": 1.8865979381443299,
"grad_norm": 0.4012429714202881,
"learning_rate": 6.224613654040753e-05,
"loss": 0.4536,
"step": 230
},
{
"epoch": 1.8948453608247422,
"grad_norm": 0.3710065186023712,
"learning_rate": 6.142353619647903e-05,
"loss": 0.4206,
"step": 231
},
{
"epoch": 1.9030927835051545,
"grad_norm": 0.3984532058238983,
"learning_rate": 6.0603991265273074e-05,
"loss": 0.394,
"step": 232
},
{
"epoch": 1.9113402061855669,
"grad_norm": 0.3717309832572937,
"learning_rate": 5.978756665807917e-05,
"loss": 0.4004,
"step": 233
},
{
"epoch": 1.9195876288659792,
"grad_norm": 0.3788721561431885,
"learning_rate": 5.897432703904453e-05,
"loss": 0.4225,
"step": 234
},
{
"epoch": 1.9278350515463918,
"grad_norm": 0.4055582284927368,
"learning_rate": 5.8164336820052203e-05,
"loss": 0.4771,
"step": 235
},
{
"epoch": 1.9360824742268041,
"grad_norm": 0.39383062720298767,
"learning_rate": 5.735766015561971e-05,
"loss": 0.4054,
"step": 236
},
{
"epoch": 1.9443298969072165,
"grad_norm": 0.39192789793014526,
"learning_rate": 5.65543609378175e-05,
"loss": 0.4554,
"step": 237
},
{
"epoch": 1.9525773195876288,
"grad_norm": 0.43132880330085754,
"learning_rate": 5.5754502791208504e-05,
"loss": 0.4845,
"step": 238
},
{
"epoch": 1.9608247422680414,
"grad_norm": 0.3810337483882904,
"learning_rate": 5.495814906780886e-05,
"loss": 0.4253,
"step": 239
},
{
"epoch": 1.9690721649484537,
"grad_norm": 0.38002529740333557,
"learning_rate": 5.4165362842070185e-05,
"loss": 0.4291,
"step": 240
},
{
"epoch": 1.977319587628866,
"grad_norm": 0.3775567412376404,
"learning_rate": 5.3376206905883694e-05,
"loss": 0.397,
"step": 241
},
{
"epoch": 1.9855670103092784,
"grad_norm": 0.36817148327827454,
"learning_rate": 5.259074376360701e-05,
"loss": 0.3816,
"step": 242
},
{
"epoch": 1.9938144329896907,
"grad_norm": 0.4288536608219147,
"learning_rate": 5.18090356271134e-05,
"loss": 0.4502,
"step": 243
},
{
"epoch": 2.002061855670103,
"grad_norm": 0.39343351125717163,
"learning_rate": 5.1031144410864384e-05,
"loss": 0.4323,
"step": 244
},
{
"epoch": 2.002061855670103,
"eval_loss": 0.5178368091583252,
"eval_runtime": 297.015,
"eval_samples_per_second": 2.02,
"eval_steps_per_second": 1.01,
"step": 244
},
{
"epoch": 2.0103092783505154,
"grad_norm": 0.5457305312156677,
"learning_rate": 5.0257131727006016e-05,
"loss": 0.3386,
"step": 245
},
{
"epoch": 2.0082474226804123,
"grad_norm": 0.3745405972003937,
"learning_rate": 4.9487058880488656e-05,
"loss": 0.3649,
"step": 246
},
{
"epoch": 2.0164948453608247,
"grad_norm": 0.3878428637981415,
"learning_rate": 4.87209868642117e-05,
"loss": 0.3819,
"step": 247
},
{
"epoch": 2.024742268041237,
"grad_norm": 0.3813370168209076,
"learning_rate": 4.795897635419235e-05,
"loss": 0.3593,
"step": 248
},
{
"epoch": 2.0329896907216494,
"grad_norm": 0.3491398096084595,
"learning_rate": 4.720108770476024e-05,
"loss": 0.3451,
"step": 249
},
{
"epoch": 2.0412371134020617,
"grad_norm": 0.3811284303665161,
"learning_rate": 4.6447380943776575e-05,
"loss": 0.3804,
"step": 250
},
{
"epoch": 2.049484536082474,
"grad_norm": 0.3748043477535248,
"learning_rate": 4.56979157678803e-05,
"loss": 0.3629,
"step": 251
},
{
"epoch": 2.0577319587628864,
"grad_norm": 0.325930655002594,
"learning_rate": 4.495275153775937e-05,
"loss": 0.2997,
"step": 252
},
{
"epoch": 2.065979381443299,
"grad_norm": 0.3703972399234772,
"learning_rate": 4.4211947273449494e-05,
"loss": 0.3476,
"step": 253
},
{
"epoch": 2.0742268041237115,
"grad_norm": 0.3672904670238495,
"learning_rate": 4.347556164965934e-05,
"loss": 0.3728,
"step": 254
},
{
"epoch": 2.082474226804124,
"grad_norm": 0.37608975172042847,
"learning_rate": 4.274365299112323e-05,
"loss": 0.3471,
"step": 255
},
{
"epoch": 2.090721649484536,
"grad_norm": 0.3415372669696808,
"learning_rate": 4.2016279267981664e-05,
"loss": 0.3589,
"step": 256
},
{
"epoch": 2.0989690721649485,
"grad_norm": 0.3874717354774475,
"learning_rate": 4.129349809118981e-05,
"loss": 0.3728,
"step": 257
},
{
"epoch": 2.107216494845361,
"grad_norm": 0.3430662155151367,
"learning_rate": 4.057536670795459e-05,
"loss": 0.3566,
"step": 258
},
{
"epoch": 2.1154639175257732,
"grad_norm": 0.39854690432548523,
"learning_rate": 3.9861941997200245e-05,
"loss": 0.4037,
"step": 259
},
{
"epoch": 2.1237113402061856,
"grad_norm": 0.34778809547424316,
"learning_rate": 3.915328046506357e-05,
"loss": 0.3291,
"step": 260
},
{
"epoch": 2.131958762886598,
"grad_norm": 0.383894145488739,
"learning_rate": 3.8449438240418134e-05,
"loss": 0.3737,
"step": 261
},
{
"epoch": 2.1402061855670103,
"grad_norm": 0.37021082639694214,
"learning_rate": 3.775047107042883e-05,
"loss": 0.3602,
"step": 262
},
{
"epoch": 2.1484536082474226,
"grad_norm": 0.40833646059036255,
"learning_rate": 3.705643431613634e-05,
"loss": 0.4031,
"step": 263
},
{
"epoch": 2.156701030927835,
"grad_norm": 0.3483593761920929,
"learning_rate": 3.636738294807245e-05,
"loss": 0.3332,
"step": 264
},
{
"epoch": 2.1649484536082473,
"grad_norm": 0.4144445061683655,
"learning_rate": 3.568337154190614e-05,
"loss": 0.4228,
"step": 265
},
{
"epoch": 2.1731958762886596,
"grad_norm": 0.37662217020988464,
"learning_rate": 3.500445427412077e-05,
"loss": 0.3613,
"step": 266
},
{
"epoch": 2.181443298969072,
"grad_norm": 0.36589595675468445,
"learning_rate": 3.433068491772341e-05,
"loss": 0.3088,
"step": 267
},
{
"epoch": 2.1896907216494848,
"grad_norm": 0.37732839584350586,
"learning_rate": 3.366211683798549e-05,
"loss": 0.3393,
"step": 268
},
{
"epoch": 2.197938144329897,
"grad_norm": 0.34662479162216187,
"learning_rate": 3.299880298821625e-05,
"loss": 0.3447,
"step": 269
},
{
"epoch": 2.2061855670103094,
"grad_norm": 0.3836705982685089,
"learning_rate": 3.23407959055685e-05,
"loss": 0.3733,
"step": 270
},
{
"epoch": 2.2144329896907218,
"grad_norm": 0.3639705777168274,
"learning_rate": 3.1688147706877666e-05,
"loss": 0.3425,
"step": 271
},
{
"epoch": 2.222680412371134,
"grad_norm": 0.3514609634876251,
"learning_rate": 3.1040910084533614e-05,
"loss": 0.3195,
"step": 272
},
{
"epoch": 2.2309278350515465,
"grad_norm": 0.3559933006763458,
"learning_rate": 3.039913430238672e-05,
"loss": 0.3442,
"step": 273
},
{
"epoch": 2.239175257731959,
"grad_norm": 0.38752618432044983,
"learning_rate": 2.9762871191687313e-05,
"loss": 0.3778,
"step": 274
},
{
"epoch": 2.247422680412371,
"grad_norm": 0.3630293309688568,
"learning_rate": 2.913217114705975e-05,
"loss": 0.3355,
"step": 275
},
{
"epoch": 2.2556701030927835,
"grad_norm": 0.3738977015018463,
"learning_rate": 2.850708412251103e-05,
"loss": 0.3583,
"step": 276
},
{
"epoch": 2.263917525773196,
"grad_norm": 0.38425007462501526,
"learning_rate": 2.7887659627474017e-05,
"loss": 0.356,
"step": 277
},
{
"epoch": 2.272164948453608,
"grad_norm": 0.3420022428035736,
"learning_rate": 2.7273946722886366e-05,
"loss": 0.3227,
"step": 278
},
{
"epoch": 2.2804123711340205,
"grad_norm": 0.3774506449699402,
"learning_rate": 2.6665994017304407e-05,
"loss": 0.3707,
"step": 279
},
{
"epoch": 2.288659793814433,
"grad_norm": 0.347858190536499,
"learning_rate": 2.6063849663053475e-05,
"loss": 0.3151,
"step": 280
},
{
"epoch": 2.296907216494845,
"grad_norm": 0.39050960540771484,
"learning_rate": 2.5467561352413648e-05,
"loss": 0.3882,
"step": 281
},
{
"epoch": 2.3051546391752575,
"grad_norm": 0.35410475730895996,
"learning_rate": 2.4877176313842753e-05,
"loss": 0.3098,
"step": 282
},
{
"epoch": 2.31340206185567,
"grad_norm": 0.3665927052497864,
"learning_rate": 2.4292741308235345e-05,
"loss": 0.3504,
"step": 283
},
{
"epoch": 2.3216494845360827,
"grad_norm": 0.33465299010276794,
"learning_rate": 2.3714302625219243e-05,
"loss": 0.2926,
"step": 284
},
{
"epoch": 2.329896907216495,
"grad_norm": 0.4150053560733795,
"learning_rate": 2.3141906079489183e-05,
"loss": 0.3964,
"step": 285
},
{
"epoch": 2.3381443298969073,
"grad_norm": 0.33678460121154785,
"learning_rate": 2.2575597007177984e-05,
"loss": 0.326,
"step": 286
},
{
"epoch": 2.3463917525773197,
"grad_norm": 0.3581312894821167,
"learning_rate": 2.2015420262265863e-05,
"loss": 0.3698,
"step": 287
},
{
"epoch": 2.354639175257732,
"grad_norm": 0.37406647205352783,
"learning_rate": 2.1461420213027772e-05,
"loss": 0.3661,
"step": 288
},
{
"epoch": 2.3628865979381444,
"grad_norm": 0.35271313786506653,
"learning_rate": 2.0913640738519335e-05,
"loss": 0.3609,
"step": 289
},
{
"epoch": 2.3711340206185567,
"grad_norm": 0.3584725558757782,
"learning_rate": 2.0372125225101234e-05,
"loss": 0.3293,
"step": 290
},
{
"epoch": 2.379381443298969,
"grad_norm": 0.4022175073623657,
"learning_rate": 1.983691656300314e-05,
"loss": 0.3965,
"step": 291
},
{
"epoch": 2.3876288659793814,
"grad_norm": 0.3425813317298889,
"learning_rate": 1.930805714292634e-05,
"loss": 0.3163,
"step": 292
},
{
"epoch": 2.3958762886597937,
"grad_norm": 0.35875558853149414,
"learning_rate": 1.8785588852686376e-05,
"loss": 0.3114,
"step": 293
},
{
"epoch": 2.404123711340206,
"grad_norm": 0.35764867067337036,
"learning_rate": 1.8269553073895375e-05,
"loss": 0.3512,
"step": 294
},
{
"epoch": 2.4123711340206184,
"grad_norm": 0.3454972505569458,
"learning_rate": 1.7759990678684335e-05,
"loss": 0.3095,
"step": 295
},
{
"epoch": 2.4206185567010308,
"grad_norm": 0.3647129535675049,
"learning_rate": 1.7256942026466072e-05,
"loss": 0.3368,
"step": 296
},
{
"epoch": 2.428865979381443,
"grad_norm": 0.4362187385559082,
"learning_rate": 1.6760446960738364e-05,
"loss": 0.4902,
"step": 297
},
{
"epoch": 2.437113402061856,
"grad_norm": 0.3899203836917877,
"learning_rate": 1.6270544805928424e-05,
"loss": 0.3338,
"step": 298
},
{
"epoch": 2.4453608247422682,
"grad_norm": 0.3600149154663086,
"learning_rate": 1.5787274364278004e-05,
"loss": 0.328,
"step": 299
},
{
"epoch": 2.4536082474226806,
"grad_norm": 0.37898537516593933,
"learning_rate": 1.5310673912770312e-05,
"loss": 0.3389,
"step": 300
},
{
"epoch": 2.461855670103093,
"grad_norm": 0.3836730420589447,
"learning_rate": 1.4840781200098152e-05,
"loss": 0.3568,
"step": 301
},
{
"epoch": 2.4701030927835053,
"grad_norm": 0.3963097333908081,
"learning_rate": 1.4377633443674233e-05,
"loss": 0.3803,
"step": 302
},
{
"epoch": 2.4783505154639176,
"grad_norm": 0.3930961787700653,
"learning_rate": 1.392126732668323e-05,
"loss": 0.4055,
"step": 303
},
{
"epoch": 2.48659793814433,
"grad_norm": 0.35550636053085327,
"learning_rate": 1.3471718995176507e-05,
"loss": 0.3519,
"step": 304
},
{
"epoch": 2.4948453608247423,
"grad_norm": 0.387653112411499,
"learning_rate": 1.3029024055209015e-05,
"loss": 0.3823,
"step": 305
},
{
"epoch": 2.4948453608247423,
"eval_loss": 0.5238527059555054,
"eval_runtime": 297.0109,
"eval_samples_per_second": 2.02,
"eval_steps_per_second": 1.01,
"step": 305
},
{
"epoch": 2.5030927835051546,
"grad_norm": 0.3949449360370636,
"learning_rate": 1.2593217570019267e-05,
"loss": 0.3438,
"step": 306
},
{
"epoch": 2.511340206185567,
"grad_norm": 0.3976804316043854,
"learning_rate": 1.2164334057252203e-05,
"loss": 0.3745,
"step": 307
},
{
"epoch": 2.5195876288659793,
"grad_norm": 0.3557458221912384,
"learning_rate": 1.174240748622516e-05,
"loss": 0.3153,
"step": 308
},
{
"epoch": 2.5278350515463917,
"grad_norm": 0.3675740361213684,
"learning_rate": 1.1327471275237456e-05,
"loss": 0.3308,
"step": 309
},
{
"epoch": 2.536082474226804,
"grad_norm": 0.37261641025543213,
"learning_rate": 1.0919558288923426e-05,
"loss": 0.3351,
"step": 310
},
{
"epoch": 2.5443298969072163,
"grad_norm": 0.3633248507976532,
"learning_rate": 1.0518700835649553e-05,
"loss": 0.3134,
"step": 311
},
{
"epoch": 2.5525773195876287,
"grad_norm": 0.4315129816532135,
"learning_rate": 1.0124930664955301e-05,
"loss": 0.3815,
"step": 312
},
{
"epoch": 2.560824742268041,
"grad_norm": 0.38921600580215454,
"learning_rate": 9.73827896503865e-06,
"loss": 0.3801,
"step": 313
},
{
"epoch": 2.5690721649484534,
"grad_norm": 0.3557523787021637,
"learning_rate": 9.358776360285759e-06,
"loss": 0.3299,
"step": 314
},
{
"epoch": 2.5773195876288657,
"grad_norm": 0.4037410318851471,
"learning_rate": 8.986452908845322e-06,
"loss": 0.3893,
"step": 315
},
{
"epoch": 2.5855670103092785,
"grad_norm": 0.37199708819389343,
"learning_rate": 8.621338100247988e-06,
"loss": 0.3183,
"step": 316
},
{
"epoch": 2.593814432989691,
"grad_norm": 0.3822597563266754,
"learning_rate": 8.26346085307057e-06,
"loss": 0.3908,
"step": 317
},
{
"epoch": 2.602061855670103,
"grad_norm": 0.354440301656723,
"learning_rate": 7.91284951264557e-06,
"loss": 0.3391,
"step": 318
},
{
"epoch": 2.6103092783505155,
"grad_norm": 0.37611883878707886,
"learning_rate": 7.569531848816147e-06,
"loss": 0.3647,
"step": 319
},
{
"epoch": 2.618556701030928,
"grad_norm": 0.3572885990142822,
"learning_rate": 7.233535053736706e-06,
"loss": 0.3623,
"step": 320
},
{
"epoch": 2.62680412371134,
"grad_norm": 0.37367528676986694,
"learning_rate": 6.90488573971898e-06,
"loss": 0.3666,
"step": 321
},
{
"epoch": 2.6350515463917525,
"grad_norm": 0.37292131781578064,
"learning_rate": 6.583609937124435e-06,
"loss": 0.3615,
"step": 322
},
{
"epoch": 2.643298969072165,
"grad_norm": 0.38947710394859314,
"learning_rate": 6.269733092302399e-06,
"loss": 0.404,
"step": 323
},
{
"epoch": 2.6515463917525772,
"grad_norm": 0.4026634395122528,
"learning_rate": 5.963280065574694e-06,
"loss": 0.3609,
"step": 324
},
{
"epoch": 2.6597938144329896,
"grad_norm": 0.3776322305202484,
"learning_rate": 5.664275129266605e-06,
"loss": 0.3513,
"step": 325
},
{
"epoch": 2.668041237113402,
"grad_norm": 0.3595254123210907,
"learning_rate": 5.372741965784323e-06,
"loss": 0.3285,
"step": 326
},
{
"epoch": 2.6762886597938147,
"grad_norm": 0.35534557700157166,
"learning_rate": 5.088703665739336e-06,
"loss": 0.3453,
"step": 327
},
{
"epoch": 2.684536082474227,
"grad_norm": 0.36769574880599976,
"learning_rate": 4.812182726119397e-06,
"loss": 0.3498,
"step": 328
},
{
"epoch": 2.6927835051546394,
"grad_norm": 0.38923129439353943,
"learning_rate": 4.543201048506851e-06,
"loss": 0.3701,
"step": 329
},
{
"epoch": 2.7010309278350517,
"grad_norm": 0.36807262897491455,
"learning_rate": 4.2817799373437994e-06,
"loss": 0.3208,
"step": 330
},
{
"epoch": 2.709278350515464,
"grad_norm": 0.385692298412323,
"learning_rate": 4.027940098244753e-06,
"loss": 0.3754,
"step": 331
},
{
"epoch": 2.7175257731958764,
"grad_norm": 0.3878878653049469,
"learning_rate": 3.7817016363566493e-06,
"loss": 0.3812,
"step": 332
},
{
"epoch": 2.7257731958762887,
"grad_norm": 0.3546048104763031,
"learning_rate": 3.54308405476651e-06,
"loss": 0.314,
"step": 333
},
{
"epoch": 2.734020618556701,
"grad_norm": 0.39490392804145813,
"learning_rate": 3.312106252956626e-06,
"loss": 0.3538,
"step": 334
},
{
"epoch": 2.7422680412371134,
"grad_norm": 0.39301711320877075,
"learning_rate": 3.0887865253076632e-06,
"loss": 0.3851,
"step": 335
},
{
"epoch": 2.7505154639175258,
"grad_norm": 0.3987005054950714,
"learning_rate": 2.873142559649722e-06,
"loss": 0.3945,
"step": 336
},
{
"epoch": 2.758762886597938,
"grad_norm": 0.377178430557251,
"learning_rate": 2.6651914358613252e-06,
"loss": 0.3813,
"step": 337
},
{
"epoch": 2.7670103092783505,
"grad_norm": 0.35405266284942627,
"learning_rate": 2.464949624516688e-06,
"loss": 0.3046,
"step": 338
},
{
"epoch": 2.775257731958763,
"grad_norm": 0.3652137219905853,
"learning_rate": 2.272432985581119e-06,
"loss": 0.3396,
"step": 339
},
{
"epoch": 2.783505154639175,
"grad_norm": 0.3721199631690979,
"learning_rate": 2.0876567671548773e-06,
"loss": 0.3039,
"step": 340
},
{
"epoch": 2.7917525773195875,
"grad_norm": 0.3970593512058258,
"learning_rate": 1.910635604265465e-06,
"loss": 0.4095,
"step": 341
},
{
"epoch": 2.8,
"grad_norm": 0.3831520080566406,
"learning_rate": 1.7413835177084835e-06,
"loss": 0.359,
"step": 342
},
{
"epoch": 2.808247422680412,
"grad_norm": 0.3413088917732239,
"learning_rate": 1.5799139129370588e-06,
"loss": 0.3073,
"step": 343
},
{
"epoch": 2.8164948453608245,
"grad_norm": 0.3681793510913849,
"learning_rate": 1.4262395790001881e-06,
"loss": 0.3917,
"step": 344
},
{
"epoch": 2.824742268041237,
"grad_norm": 0.3547656536102295,
"learning_rate": 1.2803726875296963e-06,
"loss": 0.3308,
"step": 345
},
{
"epoch": 2.8329896907216496,
"grad_norm": 0.38389670848846436,
"learning_rate": 1.142324791776239e-06,
"loss": 0.4037,
"step": 346
},
{
"epoch": 2.841237113402062,
"grad_norm": 0.37809357047080994,
"learning_rate": 1.01210682569427e-06,
"loss": 0.3697,
"step": 347
},
{
"epoch": 2.8494845360824743,
"grad_norm": 0.435940146446228,
"learning_rate": 8.897291030759314e-07,
"loss": 0.3713,
"step": 348
},
{
"epoch": 2.8577319587628867,
"grad_norm": 0.36410847306251526,
"learning_rate": 7.752013167342531e-07,
"loss": 0.3464,
"step": 349
},
{
"epoch": 2.865979381443299,
"grad_norm": 0.3589307963848114,
"learning_rate": 6.68532537735389e-07,
"loss": 0.3365,
"step": 350
},
{
"epoch": 2.8742268041237113,
"grad_norm": 0.3795320689678192,
"learning_rate": 5.697312146801915e-07,
"loss": 0.3551,
"step": 351
},
{
"epoch": 2.8824742268041237,
"grad_norm": 0.3537892997264862,
"learning_rate": 4.788051730349907e-07,
"loss": 0.3312,
"step": 352
},
{
"epoch": 2.890721649484536,
"grad_norm": 0.37003251910209656,
"learning_rate": 3.9576161451186923e-07,
"loss": 0.365,
"step": 353
},
{
"epoch": 2.8989690721649484,
"grad_norm": 0.33176755905151367,
"learning_rate": 3.2060711649817277e-07,
"loss": 0.3017,
"step": 354
},
{
"epoch": 2.9072164948453607,
"grad_norm": 0.39390304684638977,
"learning_rate": 2.5334763153559424e-07,
"loss": 0.3607,
"step": 355
},
{
"epoch": 2.915463917525773,
"grad_norm": 0.35808899998664856,
"learning_rate": 1.939884868487174e-07,
"loss": 0.3327,
"step": 356
},
{
"epoch": 2.923711340206186,
"grad_norm": 0.3784313499927521,
"learning_rate": 1.4253438392304307e-07,
"loss": 0.3194,
"step": 357
},
{
"epoch": 2.931958762886598,
"grad_norm": 0.3837621808052063,
"learning_rate": 9.898939813267616e-08,
"loss": 0.3583,
"step": 358
},
{
"epoch": 2.9402061855670105,
"grad_norm": 0.35086172819137573,
"learning_rate": 6.335697841747257e-08,
"loss": 0.324,
"step": 359
},
{
"epoch": 2.948453608247423,
"grad_norm": 0.3730827867984772,
"learning_rate": 3.563994700988005e-08,
"loss": 0.3437,
"step": 360
},
{
"epoch": 2.956701030927835,
"grad_norm": 0.32978203892707825,
"learning_rate": 1.5840499211439152e-08,
"loss": 0.3002,
"step": 361
},
{
"epoch": 2.9649484536082475,
"grad_norm": 0.39566338062286377,
"learning_rate": 3.960203218911218e-09,
"loss": 0.3865,
"step": 362
},
{
"epoch": 2.97319587628866,
"grad_norm": 0.3649386465549469,
"learning_rate": 0.0,
"loss": 0.3111,
"step": 363
}
],
"logging_steps": 1,
"max_steps": 363,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"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": 4.8645179782987776e+17,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}