Files

1673 lines
40 KiB
JSON
Raw Permalink Normal View History

{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 234,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.004282082162451492,
"grad_norm": 2.0210089683532715,
"learning_rate": 0.0,
"loss": 1.3018,
"step": 1
},
{
"epoch": 0.008564164324902984,
"grad_norm": 1.4715334177017212,
"learning_rate": 3.3333333333333333e-06,
"loss": 1.0702,
"step": 2
},
{
"epoch": 0.012846246487354477,
"grad_norm": 1.778496503829956,
"learning_rate": 6.666666666666667e-06,
"loss": 1.24,
"step": 3
},
{
"epoch": 0.017128328649805968,
"grad_norm": 1.7433445453643799,
"learning_rate": 1e-05,
"loss": 1.2456,
"step": 4
},
{
"epoch": 0.02141041081225746,
"grad_norm": 1.7866331338882446,
"learning_rate": 1.3333333333333333e-05,
"loss": 1.2573,
"step": 5
},
{
"epoch": 0.025692492974708953,
"grad_norm": 1.7886673212051392,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.2608,
"step": 6
},
{
"epoch": 0.029974575137160443,
"grad_norm": 1.7814574241638184,
"learning_rate": 2e-05,
"loss": 1.162,
"step": 7
},
{
"epoch": 0.034256657299611935,
"grad_norm": 1.4200130701065063,
"learning_rate": 2.3333333333333336e-05,
"loss": 0.9463,
"step": 8
},
{
"epoch": 0.03853873946206343,
"grad_norm": 1.204972743988037,
"learning_rate": 2.6666666666666667e-05,
"loss": 0.9157,
"step": 9
},
{
"epoch": 0.04282082162451492,
"grad_norm": 0.9675220251083374,
"learning_rate": 3e-05,
"loss": 0.8432,
"step": 10
},
{
"epoch": 0.04710290378696641,
"grad_norm": 0.7631441950798035,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.7745,
"step": 11
},
{
"epoch": 0.051384985949417906,
"grad_norm": 0.6893360614776611,
"learning_rate": 3.6666666666666666e-05,
"loss": 0.6432,
"step": 12
},
{
"epoch": 0.055667068111869396,
"grad_norm": 0.6083648204803467,
"learning_rate": 4e-05,
"loss": 0.6216,
"step": 13
},
{
"epoch": 0.059949150274320885,
"grad_norm": 0.5643147826194763,
"learning_rate": 4.3333333333333334e-05,
"loss": 0.562,
"step": 14
},
{
"epoch": 0.06423123243677238,
"grad_norm": 0.5112939476966858,
"learning_rate": 4.666666666666667e-05,
"loss": 0.564,
"step": 15
},
{
"epoch": 0.06851331459922387,
"grad_norm": 0.4901995360851288,
"learning_rate": 5e-05,
"loss": 0.4707,
"step": 16
},
{
"epoch": 0.07279539676167536,
"grad_norm": 0.4486415386199951,
"learning_rate": 5.333333333333333e-05,
"loss": 0.4521,
"step": 17
},
{
"epoch": 0.07707747892412686,
"grad_norm": 0.40917208790779114,
"learning_rate": 5.666666666666667e-05,
"loss": 0.4595,
"step": 18
},
{
"epoch": 0.08135956108657835,
"grad_norm": 0.33339953422546387,
"learning_rate": 6e-05,
"loss": 0.3596,
"step": 19
},
{
"epoch": 0.08564164324902984,
"grad_norm": 0.4043780565261841,
"learning_rate": 6.333333333333333e-05,
"loss": 0.3969,
"step": 20
},
{
"epoch": 0.08992372541148133,
"grad_norm": 0.36939042806625366,
"learning_rate": 6.666666666666667e-05,
"loss": 0.3781,
"step": 21
},
{
"epoch": 0.09420580757393282,
"grad_norm": 0.3532072901725769,
"learning_rate": 7e-05,
"loss": 0.327,
"step": 22
},
{
"epoch": 0.09848788973638432,
"grad_norm": 0.3137285113334656,
"learning_rate": 7.333333333333333e-05,
"loss": 0.3496,
"step": 23
},
{
"epoch": 0.10276997189883581,
"grad_norm": 0.2722538411617279,
"learning_rate": 7.666666666666667e-05,
"loss": 0.3306,
"step": 24
},
{
"epoch": 0.1070520540612873,
"grad_norm": 0.2855975925922394,
"learning_rate": 8e-05,
"loss": 0.2969,
"step": 25
},
{
"epoch": 0.11133413622373879,
"grad_norm": 0.25661376118659973,
"learning_rate": 8.333333333333334e-05,
"loss": 0.296,
"step": 26
},
{
"epoch": 0.11561621838619028,
"grad_norm": 0.27822190523147583,
"learning_rate": 8.666666666666667e-05,
"loss": 0.3104,
"step": 27
},
{
"epoch": 0.11989830054864177,
"grad_norm": 0.2783198058605194,
"learning_rate": 9e-05,
"loss": 0.3535,
"step": 28
},
{
"epoch": 0.12418038271109327,
"grad_norm": 0.24854612350463867,
"learning_rate": 9.333333333333334e-05,
"loss": 0.316,
"step": 29
},
{
"epoch": 0.12846246487354476,
"grad_norm": 0.2961224913597107,
"learning_rate": 9.666666666666667e-05,
"loss": 0.3493,
"step": 30
},
{
"epoch": 0.13274454703599625,
"grad_norm": 0.28391575813293457,
"learning_rate": 0.0001,
"loss": 0.362,
"step": 31
},
{
"epoch": 0.13702662919844774,
"grad_norm": 0.3998236060142517,
"learning_rate": 9.950980392156864e-05,
"loss": 0.317,
"step": 32
},
{
"epoch": 0.14130871136089923,
"grad_norm": 0.3115137815475464,
"learning_rate": 9.901960784313727e-05,
"loss": 0.3017,
"step": 33
},
{
"epoch": 0.14559079352335072,
"grad_norm": 0.23237380385398865,
"learning_rate": 9.852941176470589e-05,
"loss": 0.3185,
"step": 34
},
{
"epoch": 0.1498728756858022,
"grad_norm": 0.205907940864563,
"learning_rate": 9.80392156862745e-05,
"loss": 0.3082,
"step": 35
},
{
"epoch": 0.15415495784825373,
"grad_norm": 0.26283031702041626,
"learning_rate": 9.754901960784314e-05,
"loss": 0.2956,
"step": 36
},
{
"epoch": 0.15843704001070522,
"grad_norm": 0.2667323350906372,
"learning_rate": 9.705882352941177e-05,
"loss": 0.3103,
"step": 37
},
{
"epoch": 0.1627191221731567,
"grad_norm": 0.20380863547325134,
"learning_rate": 9.65686274509804e-05,
"loss": 0.28,
"step": 38
},
{
"epoch": 0.1670012043356082,
"grad_norm": 0.19600191712379456,
"learning_rate": 9.607843137254903e-05,
"loss": 0.2876,
"step": 39
},
{
"epoch": 0.17128328649805968,
"grad_norm": 0.26494333148002625,
"learning_rate": 9.558823529411765e-05,
"loss": 0.296,
"step": 40
},
{
"epoch": 0.17556536866051117,
"grad_norm": 0.23206257820129395,
"learning_rate": 9.509803921568627e-05,
"loss": 0.2655,
"step": 41
},
{
"epoch": 0.17984745082296266,
"grad_norm": 0.4289762079715729,
"learning_rate": 9.46078431372549e-05,
"loss": 0.2791,
"step": 42
},
{
"epoch": 0.18412953298541415,
"grad_norm": 0.22822624444961548,
"learning_rate": 9.411764705882353e-05,
"loss": 0.2847,
"step": 43
},
{
"epoch": 0.18841161514786564,
"grad_norm": 0.22668644785881042,
"learning_rate": 9.362745098039217e-05,
"loss": 0.2687,
"step": 44
},
{
"epoch": 0.19269369731031713,
"grad_norm": 0.24102894961833954,
"learning_rate": 9.313725490196079e-05,
"loss": 0.2435,
"step": 45
},
{
"epoch": 0.19697577947276865,
"grad_norm": 0.2255445420742035,
"learning_rate": 9.264705882352942e-05,
"loss": 0.2614,
"step": 46
},
{
"epoch": 0.20125786163522014,
"grad_norm": 0.23984132707118988,
"learning_rate": 9.215686274509804e-05,
"loss": 0.2802,
"step": 47
},
{
"epoch": 0.20553994379767163,
"grad_norm": 0.252221941947937,
"learning_rate": 9.166666666666667e-05,
"loss": 0.2728,
"step": 48
},
{
"epoch": 0.20982202596012312,
"grad_norm": 0.2701578438282013,
"learning_rate": 9.11764705882353e-05,
"loss": 0.2969,
"step": 49
},
{
"epoch": 0.2141041081225746,
"grad_norm": 0.2627408504486084,
"learning_rate": 9.068627450980393e-05,
"loss": 0.2978,
"step": 50
},
{
"epoch": 0.2183861902850261,
"grad_norm": 0.23231373727321625,
"learning_rate": 9.019607843137255e-05,
"loss": 0.2749,
"step": 51
},
{
"epoch": 0.22266827244747758,
"grad_norm": 0.3100220859050751,
"learning_rate": 8.970588235294118e-05,
"loss": 0.2809,
"step": 52
},
{
"epoch": 0.22695035460992907,
"grad_norm": 0.21884219348430634,
"learning_rate": 8.921568627450981e-05,
"loss": 0.2634,
"step": 53
},
{
"epoch": 0.23123243677238056,
"grad_norm": 0.24028965830802917,
"learning_rate": 8.872549019607843e-05,
"loss": 0.3012,
"step": 54
},
{
"epoch": 0.23551451893483205,
"grad_norm": 0.22290503978729248,
"learning_rate": 8.823529411764706e-05,
"loss": 0.2506,
"step": 55
},
{
"epoch": 0.23979660109728354,
"grad_norm": 0.23921112716197968,
"learning_rate": 8.774509803921568e-05,
"loss": 0.2764,
"step": 56
},
{
"epoch": 0.24407868325973506,
"grad_norm": 0.25920459628105164,
"learning_rate": 8.725490196078432e-05,
"loss": 0.273,
"step": 57
},
{
"epoch": 0.24836076542218655,
"grad_norm": 0.23481227457523346,
"learning_rate": 8.676470588235295e-05,
"loss": 0.2967,
"step": 58
},
{
"epoch": 0.252642847584638,
"grad_norm": 0.26100870966911316,
"learning_rate": 8.627450980392158e-05,
"loss": 0.2711,
"step": 59
},
{
"epoch": 0.2569249297470895,
"grad_norm": 0.22797580063343048,
"learning_rate": 8.57843137254902e-05,
"loss": 0.2772,
"step": 60
},
{
"epoch": 0.261207011909541,
"grad_norm": 0.25839635729789734,
"learning_rate": 8.529411764705883e-05,
"loss": 0.2719,
"step": 61
},
{
"epoch": 0.2654890940719925,
"grad_norm": 0.21696603298187256,
"learning_rate": 8.480392156862745e-05,
"loss": 0.2721,
"step": 62
},
{
"epoch": 0.269771176234444,
"grad_norm": 0.2532002627849579,
"learning_rate": 8.431372549019608e-05,
"loss": 0.2907,
"step": 63
},
{
"epoch": 0.2740532583968955,
"grad_norm": 0.2296505868434906,
"learning_rate": 8.382352941176471e-05,
"loss": 0.2377,
"step": 64
},
{
"epoch": 0.278335340559347,
"grad_norm": 0.21931569278240204,
"learning_rate": 8.333333333333334e-05,
"loss": 0.2624,
"step": 65
},
{
"epoch": 0.28261742272179846,
"grad_norm": 0.2120109349489212,
"learning_rate": 8.284313725490198e-05,
"loss": 0.2458,
"step": 66
},
{
"epoch": 0.28689950488425,
"grad_norm": 0.25317978858947754,
"learning_rate": 8.23529411764706e-05,
"loss": 0.308,
"step": 67
},
{
"epoch": 0.29118158704670144,
"grad_norm": 0.2619463801383972,
"learning_rate": 8.186274509803921e-05,
"loss": 0.305,
"step": 68
},
{
"epoch": 0.29546366920915296,
"grad_norm": 0.23239323496818542,
"learning_rate": 8.137254901960785e-05,
"loss": 0.2656,
"step": 69
},
{
"epoch": 0.2997457513716044,
"grad_norm": 0.2380262017250061,
"learning_rate": 8.088235294117648e-05,
"loss": 0.2711,
"step": 70
},
{
"epoch": 0.30402783353405594,
"grad_norm": 0.2514459490776062,
"learning_rate": 8.039215686274511e-05,
"loss": 0.2953,
"step": 71
},
{
"epoch": 0.30830991569650745,
"grad_norm": 0.2481430619955063,
"learning_rate": 7.990196078431373e-05,
"loss": 0.2589,
"step": 72
},
{
"epoch": 0.3125919978589589,
"grad_norm": 0.22490905225276947,
"learning_rate": 7.941176470588235e-05,
"loss": 0.268,
"step": 73
},
{
"epoch": 0.31687408002141043,
"grad_norm": 0.2417868971824646,
"learning_rate": 7.892156862745098e-05,
"loss": 0.2452,
"step": 74
},
{
"epoch": 0.3211561621838619,
"grad_norm": 0.23763221502304077,
"learning_rate": 7.843137254901961e-05,
"loss": 0.2381,
"step": 75
},
{
"epoch": 0.3254382443463134,
"grad_norm": 0.2643090486526489,
"learning_rate": 7.794117647058824e-05,
"loss": 0.3057,
"step": 76
},
{
"epoch": 0.32972032650876487,
"grad_norm": 0.2665586769580841,
"learning_rate": 7.745098039215687e-05,
"loss": 0.2533,
"step": 77
},
{
"epoch": 0.3340024086712164,
"grad_norm": 0.22238565981388092,
"learning_rate": 7.696078431372549e-05,
"loss": 0.2449,
"step": 78
},
{
"epoch": 0.33828449083366785,
"grad_norm": 0.2617754340171814,
"learning_rate": 7.647058823529411e-05,
"loss": 0.2877,
"step": 79
},
{
"epoch": 0.34256657299611937,
"grad_norm": 0.2524866461753845,
"learning_rate": 7.598039215686274e-05,
"loss": 0.2453,
"step": 80
},
{
"epoch": 0.34684865515857083,
"grad_norm": 0.24400098621845245,
"learning_rate": 7.549019607843137e-05,
"loss": 0.2659,
"step": 81
},
{
"epoch": 0.35113073732102235,
"grad_norm": 0.2472800612449646,
"learning_rate": 7.500000000000001e-05,
"loss": 0.2618,
"step": 82
},
{
"epoch": 0.35541281948347386,
"grad_norm": 0.2767047584056854,
"learning_rate": 7.450980392156864e-05,
"loss": 0.2667,
"step": 83
},
{
"epoch": 0.3596949016459253,
"grad_norm": 0.21703694760799408,
"learning_rate": 7.401960784313726e-05,
"loss": 0.1999,
"step": 84
},
{
"epoch": 0.36397698380837684,
"grad_norm": 0.2640591561794281,
"learning_rate": 7.352941176470589e-05,
"loss": 0.2482,
"step": 85
},
{
"epoch": 0.3682590659708283,
"grad_norm": 0.2501486539840698,
"learning_rate": 7.303921568627451e-05,
"loss": 0.2409,
"step": 86
},
{
"epoch": 0.3725411481332798,
"grad_norm": 0.2676973044872284,
"learning_rate": 7.254901960784314e-05,
"loss": 0.2411,
"step": 87
},
{
"epoch": 0.3768232302957313,
"grad_norm": 0.3220224678516388,
"learning_rate": 7.205882352941177e-05,
"loss": 0.2873,
"step": 88
},
{
"epoch": 0.3811053124581828,
"grad_norm": 0.2271728515625,
"learning_rate": 7.156862745098039e-05,
"loss": 0.2601,
"step": 89
},
{
"epoch": 0.38538739462063426,
"grad_norm": 0.26562830805778503,
"learning_rate": 7.107843137254902e-05,
"loss": 0.2579,
"step": 90
},
{
"epoch": 0.3896694767830858,
"grad_norm": 0.25470057129859924,
"learning_rate": 7.058823529411765e-05,
"loss": 0.255,
"step": 91
},
{
"epoch": 0.3939515589455373,
"grad_norm": 0.24521011114120483,
"learning_rate": 7.009803921568627e-05,
"loss": 0.2347,
"step": 92
},
{
"epoch": 0.39823364110798876,
"grad_norm": 0.25197967886924744,
"learning_rate": 6.96078431372549e-05,
"loss": 0.2511,
"step": 93
},
{
"epoch": 0.4025157232704403,
"grad_norm": 0.24449393153190613,
"learning_rate": 6.911764705882354e-05,
"loss": 0.2281,
"step": 94
},
{
"epoch": 0.40679780543289173,
"grad_norm": 0.2718404531478882,
"learning_rate": 6.862745098039216e-05,
"loss": 0.2696,
"step": 95
},
{
"epoch": 0.41107988759534325,
"grad_norm": 0.26531246304512024,
"learning_rate": 6.813725490196079e-05,
"loss": 0.281,
"step": 96
},
{
"epoch": 0.4153619697577947,
"grad_norm": 0.291124552488327,
"learning_rate": 6.764705882352942e-05,
"loss": 0.2562,
"step": 97
},
{
"epoch": 0.41964405192024623,
"grad_norm": 0.2717721462249756,
"learning_rate": 6.715686274509804e-05,
"loss": 0.2724,
"step": 98
},
{
"epoch": 0.4239261340826977,
"grad_norm": 0.23931877315044403,
"learning_rate": 6.666666666666667e-05,
"loss": 0.2348,
"step": 99
},
{
"epoch": 0.4282082162451492,
"grad_norm": 0.2564341723918915,
"learning_rate": 6.61764705882353e-05,
"loss": 0.2728,
"step": 100
},
{
"epoch": 0.43249029840760067,
"grad_norm": 0.27246496081352234,
"learning_rate": 6.568627450980392e-05,
"loss": 0.2464,
"step": 101
},
{
"epoch": 0.4367723805700522,
"grad_norm": 0.27825963497161865,
"learning_rate": 6.519607843137255e-05,
"loss": 0.2662,
"step": 102
},
{
"epoch": 0.4410544627325037,
"grad_norm": 0.2940983772277832,
"learning_rate": 6.470588235294118e-05,
"loss": 0.2575,
"step": 103
},
{
"epoch": 0.44533654489495517,
"grad_norm": 0.29442298412323,
"learning_rate": 6.421568627450982e-05,
"loss": 0.251,
"step": 104
},
{
"epoch": 0.4496186270574067,
"grad_norm": 0.2557864785194397,
"learning_rate": 6.372549019607843e-05,
"loss": 0.2576,
"step": 105
},
{
"epoch": 0.45390070921985815,
"grad_norm": 0.36144790053367615,
"learning_rate": 6.323529411764705e-05,
"loss": 0.3035,
"step": 106
},
{
"epoch": 0.45818279138230966,
"grad_norm": 0.271528035402298,
"learning_rate": 6.274509803921569e-05,
"loss": 0.1997,
"step": 107
},
{
"epoch": 0.4624648735447611,
"grad_norm": 0.279249906539917,
"learning_rate": 6.225490196078432e-05,
"loss": 0.241,
"step": 108
},
{
"epoch": 0.46674695570721264,
"grad_norm": 0.27461954951286316,
"learning_rate": 6.176470588235295e-05,
"loss": 0.2898,
"step": 109
},
{
"epoch": 0.4710290378696641,
"grad_norm": 0.26777157187461853,
"learning_rate": 6.127450980392158e-05,
"loss": 0.2481,
"step": 110
},
{
"epoch": 0.4753111200321156,
"grad_norm": 0.23129451274871826,
"learning_rate": 6.078431372549019e-05,
"loss": 0.2317,
"step": 111
},
{
"epoch": 0.4795932021945671,
"grad_norm": 0.26172134280204773,
"learning_rate": 6.0294117647058825e-05,
"loss": 0.2386,
"step": 112
},
{
"epoch": 0.4838752843570186,
"grad_norm": 0.25814586877822876,
"learning_rate": 5.980392156862745e-05,
"loss": 0.24,
"step": 113
},
{
"epoch": 0.4881573665194701,
"grad_norm": 0.331678181886673,
"learning_rate": 5.931372549019608e-05,
"loss": 0.2449,
"step": 114
},
{
"epoch": 0.4924394486819216,
"grad_norm": 0.26507994532585144,
"learning_rate": 5.882352941176471e-05,
"loss": 0.2605,
"step": 115
},
{
"epoch": 0.4967215308443731,
"grad_norm": 0.28009164333343506,
"learning_rate": 5.833333333333334e-05,
"loss": 0.2306,
"step": 116
},
{
"epoch": 0.5010036130068246,
"grad_norm": 0.2864171862602234,
"learning_rate": 5.784313725490197e-05,
"loss": 0.2458,
"step": 117
},
{
"epoch": 0.505285695169276,
"grad_norm": 0.29294925928115845,
"learning_rate": 5.735294117647059e-05,
"loss": 0.269,
"step": 118
},
{
"epoch": 0.5095677773317275,
"grad_norm": 0.2571495473384857,
"learning_rate": 5.6862745098039215e-05,
"loss": 0.2242,
"step": 119
},
{
"epoch": 0.513849859494179,
"grad_norm": 0.253811776638031,
"learning_rate": 5.637254901960785e-05,
"loss": 0.2397,
"step": 120
},
{
"epoch": 0.5181319416566306,
"grad_norm": 0.2918514311313629,
"learning_rate": 5.588235294117647e-05,
"loss": 0.2526,
"step": 121
},
{
"epoch": 0.522414023819082,
"grad_norm": 0.25536906719207764,
"learning_rate": 5.5392156862745104e-05,
"loss": 0.2397,
"step": 122
},
{
"epoch": 0.5266961059815335,
"grad_norm": 0.2694036066532135,
"learning_rate": 5.490196078431373e-05,
"loss": 0.2385,
"step": 123
},
{
"epoch": 0.530978188143985,
"grad_norm": 0.3011780083179474,
"learning_rate": 5.441176470588235e-05,
"loss": 0.2947,
"step": 124
},
{
"epoch": 0.5352602703064365,
"grad_norm": 0.2834988534450531,
"learning_rate": 5.392156862745098e-05,
"loss": 0.2235,
"step": 125
},
{
"epoch": 0.539542352468888,
"grad_norm": 0.2947101891040802,
"learning_rate": 5.343137254901961e-05,
"loss": 0.2694,
"step": 126
},
{
"epoch": 0.5438244346313394,
"grad_norm": 0.243186816573143,
"learning_rate": 5.294117647058824e-05,
"loss": 0.2279,
"step": 127
},
{
"epoch": 0.548106516793791,
"grad_norm": 0.2507615089416504,
"learning_rate": 5.245098039215687e-05,
"loss": 0.2354,
"step": 128
},
{
"epoch": 0.5523885989562425,
"grad_norm": 0.2874569892883301,
"learning_rate": 5.1960784313725495e-05,
"loss": 0.2467,
"step": 129
},
{
"epoch": 0.556670681118694,
"grad_norm": 0.2643214166164398,
"learning_rate": 5.147058823529411e-05,
"loss": 0.2872,
"step": 130
},
{
"epoch": 0.5609527632811454,
"grad_norm": 0.28615084290504456,
"learning_rate": 5.0980392156862745e-05,
"loss": 0.2392,
"step": 131
},
{
"epoch": 0.5652348454435969,
"grad_norm": 0.27777400612831116,
"learning_rate": 5.049019607843137e-05,
"loss": 0.2374,
"step": 132
},
{
"epoch": 0.5695169276060484,
"grad_norm": 0.2913888692855835,
"learning_rate": 5e-05,
"loss": 0.2178,
"step": 133
},
{
"epoch": 0.5737990097685,
"grad_norm": 0.29747387766838074,
"learning_rate": 4.9509803921568634e-05,
"loss": 0.289,
"step": 134
},
{
"epoch": 0.5780810919309515,
"grad_norm": 0.270842581987381,
"learning_rate": 4.901960784313725e-05,
"loss": 0.2235,
"step": 135
},
{
"epoch": 0.5823631740934029,
"grad_norm": 0.29347699880599976,
"learning_rate": 4.8529411764705885e-05,
"loss": 0.2487,
"step": 136
},
{
"epoch": 0.5866452562558544,
"grad_norm": 0.27630671858787537,
"learning_rate": 4.803921568627452e-05,
"loss": 0.2479,
"step": 137
},
{
"epoch": 0.5909273384183059,
"grad_norm": 0.2936690151691437,
"learning_rate": 4.7549019607843135e-05,
"loss": 0.255,
"step": 138
},
{
"epoch": 0.5952094205807574,
"grad_norm": 0.2868378460407257,
"learning_rate": 4.705882352941177e-05,
"loss": 0.2355,
"step": 139
},
{
"epoch": 0.5994915027432088,
"grad_norm": 0.24357955157756805,
"learning_rate": 4.656862745098039e-05,
"loss": 0.2106,
"step": 140
},
{
"epoch": 0.6037735849056604,
"grad_norm": 0.274213045835495,
"learning_rate": 4.607843137254902e-05,
"loss": 0.2461,
"step": 141
},
{
"epoch": 0.6080556670681119,
"grad_norm": 0.2639996409416199,
"learning_rate": 4.558823529411765e-05,
"loss": 0.2141,
"step": 142
},
{
"epoch": 0.6123377492305634,
"grad_norm": 0.2821321487426758,
"learning_rate": 4.5098039215686275e-05,
"loss": 0.2479,
"step": 143
},
{
"epoch": 0.6166198313930149,
"grad_norm": 0.3077673316001892,
"learning_rate": 4.460784313725491e-05,
"loss": 0.2648,
"step": 144
},
{
"epoch": 0.6209019135554663,
"grad_norm": 0.2770426869392395,
"learning_rate": 4.411764705882353e-05,
"loss": 0.2427,
"step": 145
},
{
"epoch": 0.6251839957179178,
"grad_norm": 0.28925347328186035,
"learning_rate": 4.362745098039216e-05,
"loss": 0.2478,
"step": 146
},
{
"epoch": 0.6294660778803693,
"grad_norm": 0.27440404891967773,
"learning_rate": 4.313725490196079e-05,
"loss": 0.261,
"step": 147
},
{
"epoch": 0.6337481600428209,
"grad_norm": 0.28528398275375366,
"learning_rate": 4.2647058823529415e-05,
"loss": 0.251,
"step": 148
},
{
"epoch": 0.6380302422052723,
"grad_norm": 0.2765834331512451,
"learning_rate": 4.215686274509804e-05,
"loss": 0.216,
"step": 149
},
{
"epoch": 0.6423123243677238,
"grad_norm": 0.2953094244003296,
"learning_rate": 4.166666666666667e-05,
"loss": 0.2149,
"step": 150
},
{
"epoch": 0.6465944065301753,
"grad_norm": 0.3101818859577179,
"learning_rate": 4.11764705882353e-05,
"loss": 0.2481,
"step": 151
},
{
"epoch": 0.6508764886926268,
"grad_norm": 0.31335169076919556,
"learning_rate": 4.068627450980392e-05,
"loss": 0.2298,
"step": 152
},
{
"epoch": 0.6551585708550783,
"grad_norm": 0.29731401801109314,
"learning_rate": 4.0196078431372555e-05,
"loss": 0.2325,
"step": 153
},
{
"epoch": 0.6594406530175297,
"grad_norm": 0.3302064538002014,
"learning_rate": 3.970588235294117e-05,
"loss": 0.2686,
"step": 154
},
{
"epoch": 0.6637227351799813,
"grad_norm": 0.317256361246109,
"learning_rate": 3.9215686274509805e-05,
"loss": 0.235,
"step": 155
},
{
"epoch": 0.6680048173424328,
"grad_norm": 0.31570908427238464,
"learning_rate": 3.872549019607844e-05,
"loss": 0.2498,
"step": 156
},
{
"epoch": 0.6722868995048843,
"grad_norm": 0.3072652816772461,
"learning_rate": 3.8235294117647055e-05,
"loss": 0.2916,
"step": 157
},
{
"epoch": 0.6765689816673357,
"grad_norm": 0.3168857991695404,
"learning_rate": 3.774509803921569e-05,
"loss": 0.2578,
"step": 158
},
{
"epoch": 0.6808510638297872,
"grad_norm": 0.29152312874794006,
"learning_rate": 3.725490196078432e-05,
"loss": 0.2577,
"step": 159
},
{
"epoch": 0.6851331459922387,
"grad_norm": 0.28397178649902344,
"learning_rate": 3.6764705882352945e-05,
"loss": 0.2313,
"step": 160
},
{
"epoch": 0.6894152281546903,
"grad_norm": 0.25765499472618103,
"learning_rate": 3.627450980392157e-05,
"loss": 0.195,
"step": 161
},
{
"epoch": 0.6936973103171417,
"grad_norm": 0.27583205699920654,
"learning_rate": 3.5784313725490195e-05,
"loss": 0.2858,
"step": 162
},
{
"epoch": 0.6979793924795932,
"grad_norm": 0.24685773253440857,
"learning_rate": 3.529411764705883e-05,
"loss": 0.2081,
"step": 163
},
{
"epoch": 0.7022614746420447,
"grad_norm": 0.2538939416408539,
"learning_rate": 3.480392156862745e-05,
"loss": 0.2034,
"step": 164
},
{
"epoch": 0.7065435568044962,
"grad_norm": 0.2724453806877136,
"learning_rate": 3.431372549019608e-05,
"loss": 0.2298,
"step": 165
},
{
"epoch": 0.7108256389669477,
"grad_norm": 0.29416435956954956,
"learning_rate": 3.382352941176471e-05,
"loss": 0.2289,
"step": 166
},
{
"epoch": 0.7151077211293991,
"grad_norm": 0.28896087408065796,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.256,
"step": 167
},
{
"epoch": 0.7193898032918506,
"grad_norm": 0.26924827694892883,
"learning_rate": 3.284313725490196e-05,
"loss": 0.227,
"step": 168
},
{
"epoch": 0.7236718854543022,
"grad_norm": 0.3040426969528198,
"learning_rate": 3.235294117647059e-05,
"loss": 0.2054,
"step": 169
},
{
"epoch": 0.7279539676167537,
"grad_norm": 0.2853792607784271,
"learning_rate": 3.186274509803922e-05,
"loss": 0.237,
"step": 170
},
{
"epoch": 0.7322360497792051,
"grad_norm": 0.2915240228176117,
"learning_rate": 3.137254901960784e-05,
"loss": 0.2486,
"step": 171
},
{
"epoch": 0.7365181319416566,
"grad_norm": 0.28236696124076843,
"learning_rate": 3.0882352941176475e-05,
"loss": 0.2204,
"step": 172
},
{
"epoch": 0.7408002141041081,
"grad_norm": 0.35193923115730286,
"learning_rate": 3.0392156862745097e-05,
"loss": 0.2734,
"step": 173
},
{
"epoch": 0.7450822962665596,
"grad_norm": 0.3084484040737152,
"learning_rate": 2.9901960784313725e-05,
"loss": 0.2256,
"step": 174
},
{
"epoch": 0.7493643784290112,
"grad_norm": 0.2819043695926666,
"learning_rate": 2.9411764705882354e-05,
"loss": 0.2315,
"step": 175
},
{
"epoch": 0.7536464605914626,
"grad_norm": 0.30647221207618713,
"learning_rate": 2.8921568627450986e-05,
"loss": 0.2523,
"step": 176
},
{
"epoch": 0.7579285427539141,
"grad_norm": 0.28489452600479126,
"learning_rate": 2.8431372549019608e-05,
"loss": 0.2316,
"step": 177
},
{
"epoch": 0.7622106249163656,
"grad_norm": 0.28946006298065186,
"learning_rate": 2.7941176470588236e-05,
"loss": 0.2253,
"step": 178
},
{
"epoch": 0.7664927070788171,
"grad_norm": 0.27433091402053833,
"learning_rate": 2.7450980392156865e-05,
"loss": 0.2212,
"step": 179
},
{
"epoch": 0.7707747892412685,
"grad_norm": 0.29885029792785645,
"learning_rate": 2.696078431372549e-05,
"loss": 0.2075,
"step": 180
},
{
"epoch": 0.77505687140372,
"grad_norm": 0.3466060757637024,
"learning_rate": 2.647058823529412e-05,
"loss": 0.2638,
"step": 181
},
{
"epoch": 0.7793389535661716,
"grad_norm": 0.29057374596595764,
"learning_rate": 2.5980392156862747e-05,
"loss": 0.2169,
"step": 182
},
{
"epoch": 0.7836210357286231,
"grad_norm": 0.30510345101356506,
"learning_rate": 2.5490196078431373e-05,
"loss": 0.2495,
"step": 183
},
{
"epoch": 0.7879031178910746,
"grad_norm": 0.3553491234779358,
"learning_rate": 2.5e-05,
"loss": 0.2523,
"step": 184
},
{
"epoch": 0.792185200053526,
"grad_norm": 0.29659029841423035,
"learning_rate": 2.4509803921568626e-05,
"loss": 0.2691,
"step": 185
},
{
"epoch": 0.7964672822159775,
"grad_norm": 0.3015158474445343,
"learning_rate": 2.401960784313726e-05,
"loss": 0.2656,
"step": 186
},
{
"epoch": 0.800749364378429,
"grad_norm": 0.29579856991767883,
"learning_rate": 2.3529411764705884e-05,
"loss": 0.2326,
"step": 187
},
{
"epoch": 0.8050314465408805,
"grad_norm": 0.28458544611930847,
"learning_rate": 2.303921568627451e-05,
"loss": 0.2476,
"step": 188
},
{
"epoch": 0.809313528703332,
"grad_norm": 0.26741454005241394,
"learning_rate": 2.2549019607843138e-05,
"loss": 0.1977,
"step": 189
},
{
"epoch": 0.8135956108657835,
"grad_norm": 0.28835511207580566,
"learning_rate": 2.2058823529411766e-05,
"loss": 0.2239,
"step": 190
},
{
"epoch": 0.817877693028235,
"grad_norm": 0.30311253666877747,
"learning_rate": 2.1568627450980395e-05,
"loss": 0.193,
"step": 191
},
{
"epoch": 0.8221597751906865,
"grad_norm": 0.28291624784469604,
"learning_rate": 2.107843137254902e-05,
"loss": 0.2345,
"step": 192
},
{
"epoch": 0.8264418573531379,
"grad_norm": 0.2831307053565979,
"learning_rate": 2.058823529411765e-05,
"loss": 0.2096,
"step": 193
},
{
"epoch": 0.8307239395155894,
"grad_norm": 0.29355019330978394,
"learning_rate": 2.0098039215686277e-05,
"loss": 0.224,
"step": 194
},
{
"epoch": 0.8350060216780409,
"grad_norm": 0.3075982332229614,
"learning_rate": 1.9607843137254903e-05,
"loss": 0.2647,
"step": 195
},
{
"epoch": 0.8392881038404925,
"grad_norm": 0.31521862745285034,
"learning_rate": 1.9117647058823528e-05,
"loss": 0.2396,
"step": 196
},
{
"epoch": 0.843570186002944,
"grad_norm": 0.31922802329063416,
"learning_rate": 1.862745098039216e-05,
"loss": 0.2734,
"step": 197
},
{
"epoch": 0.8478522681653954,
"grad_norm": 0.25028201937675476,
"learning_rate": 1.8137254901960785e-05,
"loss": 0.2054,
"step": 198
},
{
"epoch": 0.8521343503278469,
"grad_norm": 0.31554773449897766,
"learning_rate": 1.7647058823529414e-05,
"loss": 0.2358,
"step": 199
},
{
"epoch": 0.8564164324902984,
"grad_norm": 0.2841096818447113,
"learning_rate": 1.715686274509804e-05,
"loss": 0.2047,
"step": 200
},
{
"epoch": 0.8606985146527499,
"grad_norm": 0.2850678563117981,
"learning_rate": 1.6666666666666667e-05,
"loss": 0.244,
"step": 201
},
{
"epoch": 0.8649805968152013,
"grad_norm": 0.3114566206932068,
"learning_rate": 1.6176470588235296e-05,
"loss": 0.2288,
"step": 202
},
{
"epoch": 0.8692626789776529,
"grad_norm": 0.2772825062274933,
"learning_rate": 1.568627450980392e-05,
"loss": 0.2378,
"step": 203
},
{
"epoch": 0.8735447611401044,
"grad_norm": 0.3019357919692993,
"learning_rate": 1.5196078431372548e-05,
"loss": 0.2317,
"step": 204
},
{
"epoch": 0.8778268433025559,
"grad_norm": 0.2756780683994293,
"learning_rate": 1.4705882352941177e-05,
"loss": 0.2005,
"step": 205
},
{
"epoch": 0.8821089254650074,
"grad_norm": 0.30061545968055725,
"learning_rate": 1.4215686274509804e-05,
"loss": 0.2267,
"step": 206
},
{
"epoch": 0.8863910076274588,
"grad_norm": 0.2663876414299011,
"learning_rate": 1.3725490196078432e-05,
"loss": 0.1906,
"step": 207
},
{
"epoch": 0.8906730897899103,
"grad_norm": 0.303177148103714,
"learning_rate": 1.323529411764706e-05,
"loss": 0.2575,
"step": 208
},
{
"epoch": 0.8949551719523618,
"grad_norm": 0.2989080846309662,
"learning_rate": 1.2745098039215686e-05,
"loss": 0.2525,
"step": 209
},
{
"epoch": 0.8992372541148134,
"grad_norm": 0.30029749870300293,
"learning_rate": 1.2254901960784313e-05,
"loss": 0.2224,
"step": 210
},
{
"epoch": 0.9035193362772648,
"grad_norm": 0.3352798521518707,
"learning_rate": 1.1764705882352942e-05,
"loss": 0.2436,
"step": 211
},
{
"epoch": 0.9078014184397163,
"grad_norm": 0.276764839887619,
"learning_rate": 1.1274509803921569e-05,
"loss": 0.1897,
"step": 212
},
{
"epoch": 0.9120835006021678,
"grad_norm": 0.3328615725040436,
"learning_rate": 1.0784313725490197e-05,
"loss": 0.2385,
"step": 213
},
{
"epoch": 0.9163655827646193,
"grad_norm": 0.3012445271015167,
"learning_rate": 1.0294117647058824e-05,
"loss": 0.2504,
"step": 214
},
{
"epoch": 0.9206476649270708,
"grad_norm": 0.3224981725215912,
"learning_rate": 9.803921568627451e-06,
"loss": 0.2347,
"step": 215
},
{
"epoch": 0.9249297470895222,
"grad_norm": 0.3013220429420471,
"learning_rate": 9.31372549019608e-06,
"loss": 0.2465,
"step": 216
},
{
"epoch": 0.9292118292519738,
"grad_norm": 0.2794628143310547,
"learning_rate": 8.823529411764707e-06,
"loss": 0.2301,
"step": 217
},
{
"epoch": 0.9334939114144253,
"grad_norm": 0.29956692457199097,
"learning_rate": 8.333333333333334e-06,
"loss": 0.2239,
"step": 218
},
{
"epoch": 0.9377759935768768,
"grad_norm": 0.35084548592567444,
"learning_rate": 7.84313725490196e-06,
"loss": 0.2684,
"step": 219
},
{
"epoch": 0.9420580757393282,
"grad_norm": 0.26515495777130127,
"learning_rate": 7.3529411764705884e-06,
"loss": 0.1934,
"step": 220
},
{
"epoch": 0.9463401579017797,
"grad_norm": 0.3398326635360718,
"learning_rate": 6.862745098039216e-06,
"loss": 0.2872,
"step": 221
},
{
"epoch": 0.9506222400642312,
"grad_norm": 0.29966697096824646,
"learning_rate": 6.372549019607843e-06,
"loss": 0.2382,
"step": 222
},
{
"epoch": 0.9549043222266828,
"grad_norm": 0.3127211630344391,
"learning_rate": 5.882352941176471e-06,
"loss": 0.2339,
"step": 223
},
{
"epoch": 0.9591864043891342,
"grad_norm": 0.2692292332649231,
"learning_rate": 5.392156862745099e-06,
"loss": 0.2214,
"step": 224
},
{
"epoch": 0.9634684865515857,
"grad_norm": 0.27468612790107727,
"learning_rate": 4.901960784313726e-06,
"loss": 0.2238,
"step": 225
},
{
"epoch": 0.9677505687140372,
"grad_norm": 0.32963308691978455,
"learning_rate": 4.411764705882353e-06,
"loss": 0.2679,
"step": 226
},
{
"epoch": 0.9720326508764887,
"grad_norm": 0.2941998243331909,
"learning_rate": 3.92156862745098e-06,
"loss": 0.2064,
"step": 227
},
{
"epoch": 0.9763147330389402,
"grad_norm": 0.34327054023742676,
"learning_rate": 3.431372549019608e-06,
"loss": 0.2577,
"step": 228
},
{
"epoch": 0.9805968152013916,
"grad_norm": 0.31347987055778503,
"learning_rate": 2.9411764705882355e-06,
"loss": 0.2431,
"step": 229
},
{
"epoch": 0.9848788973638432,
"grad_norm": 0.3206256926059723,
"learning_rate": 2.450980392156863e-06,
"loss": 0.2313,
"step": 230
},
{
"epoch": 0.9891609795262947,
"grad_norm": 0.3200533092021942,
"learning_rate": 1.96078431372549e-06,
"loss": 0.2146,
"step": 231
},
{
"epoch": 0.9934430616887462,
"grad_norm": 0.28174784779548645,
"learning_rate": 1.4705882352941177e-06,
"loss": 0.2236,
"step": 232
},
{
"epoch": 0.9977251438511976,
"grad_norm": 0.3027549386024475,
"learning_rate": 9.80392156862745e-07,
"loss": 0.1948,
"step": 233
},
{
"epoch": 1.0,
"grad_norm": 0.48666197061538696,
"learning_rate": 4.901960784313725e-07,
"loss": 0.2137,
"step": 234
}
],
"logging_steps": 1,
"max_steps": 234,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 5000,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 3.705409466956554e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}