{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9865410497981157,
  "eval_steps": 500,
  "global_step": 184,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010767160161507403,
      "grad_norm": 0.7117969989776611,
      "learning_rate": 1e-05,
      "loss": 88.6269,
      "step": 1
    },
    {
      "epoch": 0.021534320323014805,
      "grad_norm": 0.6731469035148621,
      "learning_rate": 2e-05,
      "loss": 88.5874,
      "step": 2
    },
    {
      "epoch": 0.03230148048452221,
      "grad_norm": 0.684374213218689,
      "learning_rate": 3e-05,
      "loss": 88.5793,
      "step": 3
    },
    {
      "epoch": 0.04306864064602961,
      "grad_norm": 0.6637169122695923,
      "learning_rate": 4e-05,
      "loss": 88.5999,
      "step": 4
    },
    {
      "epoch": 0.05383580080753701,
      "grad_norm": 0.7389045357704163,
      "learning_rate": 5e-05,
      "loss": 88.5472,
      "step": 5
    },
    {
      "epoch": 0.06460296096904442,
      "grad_norm": 0.7229312658309937,
      "learning_rate": 6e-05,
      "loss": 88.619,
      "step": 6
    },
    {
      "epoch": 0.07537012113055182,
      "grad_norm": 0.6590586304664612,
      "learning_rate": 7e-05,
      "loss": 88.558,
      "step": 7
    },
    {
      "epoch": 0.08613728129205922,
      "grad_norm": 0.7481277585029602,
      "learning_rate": 8e-05,
      "loss": 88.5616,
      "step": 8
    },
    {
      "epoch": 0.09690444145356662,
      "grad_norm": 0.7117283940315247,
      "learning_rate": 9e-05,
      "loss": 88.5806,
      "step": 9
    },
    {
      "epoch": 0.10767160161507403,
      "grad_norm": 0.6721596121788025,
      "learning_rate": 0.0001,
      "loss": 88.5172,
      "step": 10
    },
    {
      "epoch": 0.11843876177658143,
      "grad_norm": 0.6705605387687683,
      "learning_rate": 9.999185052823207e-05,
      "loss": 88.6148,
      "step": 11
    },
    {
      "epoch": 0.12920592193808883,
      "grad_norm": 0.7348136305809021,
      "learning_rate": 9.996740476948385e-05,
      "loss": 88.4994,
      "step": 12
    },
    {
      "epoch": 0.13997308209959622,
      "grad_norm": 0.7699621319770813,
      "learning_rate": 9.992667069255619e-05,
      "loss": 88.5175,
      "step": 13
    },
    {
      "epoch": 0.15074024226110364,
      "grad_norm": 0.6699883937835693,
      "learning_rate": 9.98696615758975e-05,
      "loss": 88.6153,
      "step": 14
    },
    {
      "epoch": 0.16150740242261102,
      "grad_norm": 0.708699107170105,
      "learning_rate": 9.979639600327522e-05,
      "loss": 88.512,
      "step": 15
    },
    {
      "epoch": 0.17227456258411844,
      "grad_norm": 0.7570011615753174,
      "learning_rate": 9.970689785771798e-05,
      "loss": 88.5852,
      "step": 16
    },
    {
      "epoch": 0.18304172274562583,
      "grad_norm": 0.7698582410812378,
      "learning_rate": 9.960119631373022e-05,
      "loss": 88.4942,
      "step": 17
    },
    {
      "epoch": 0.19380888290713325,
      "grad_norm": 0.7012806534767151,
      "learning_rate": 9.947932582778188e-05,
      "loss": 88.5333,
      "step": 18
    },
    {
      "epoch": 0.20457604306864063,
      "grad_norm": 0.7035436630249023,
      "learning_rate": 9.934132612707632e-05,
      "loss": 88.4619,
      "step": 19
    },
    {
      "epoch": 0.21534320323014805,
      "grad_norm": 0.7284402847290039,
      "learning_rate": 9.918724219660013e-05,
      "loss": 88.5454,
      "step": 20
    },
    {
      "epoch": 0.22611036339165544,
      "grad_norm": 0.8240225911140442,
      "learning_rate": 9.901712426445901e-05,
      "loss": 88.485,
      "step": 21
    },
    {
      "epoch": 0.23687752355316286,
      "grad_norm": 0.6969294548034668,
      "learning_rate": 9.883102778550434e-05,
      "loss": 88.4997,
      "step": 22
    },
    {
      "epoch": 0.24764468371467024,
      "grad_norm": 0.7305822968482971,
      "learning_rate": 9.862901342325617e-05,
      "loss": 88.4881,
      "step": 23
    },
    {
      "epoch": 0.25841184387617766,
      "grad_norm": 0.64059978723526,
      "learning_rate": 9.841114703012817e-05,
      "loss": 88.5312,
      "step": 24
    },
    {
      "epoch": 0.2691790040376851,
      "grad_norm": 0.6767567992210388,
      "learning_rate": 9.817749962596115e-05,
      "loss": 88.4213,
      "step": 25
    },
    {
      "epoch": 0.27994616419919244,
      "grad_norm": 0.6672939658164978,
      "learning_rate": 9.792814737487207e-05,
      "loss": 88.5844,
      "step": 26
    },
    {
      "epoch": 0.29071332436069985,
      "grad_norm": 0.7302522659301758,
      "learning_rate": 9.766317156042615e-05,
      "loss": 88.4343,
      "step": 27
    },
    {
      "epoch": 0.30148048452220727,
      "grad_norm": 0.6843910813331604,
      "learning_rate": 9.738265855914013e-05,
      "loss": 88.4801,
      "step": 28
    },
    {
      "epoch": 0.3122476446837147,
      "grad_norm": 0.6865822076797485,
      "learning_rate": 9.708669981232541e-05,
      "loss": 88.5386,
      "step": 29
    },
    {
      "epoch": 0.32301480484522205,
      "grad_norm": 0.6347126960754395,
      "learning_rate": 9.677539179628005e-05,
      "loss": 88.5193,
      "step": 30
    },
    {
      "epoch": 0.33378196500672946,
      "grad_norm": 0.6892479658126831,
      "learning_rate": 9.644883599083958e-05,
      "loss": 88.5409,
      "step": 31
    },
    {
      "epoch": 0.3445491251682369,
      "grad_norm": 0.7092834115028381,
      "learning_rate": 9.610713884629666e-05,
      "loss": 88.4388,
      "step": 32
    },
    {
      "epoch": 0.3553162853297443,
      "grad_norm": 0.6895610690116882,
      "learning_rate": 9.57504117487006e-05,
      "loss": 88.5119,
      "step": 33
    },
    {
      "epoch": 0.36608344549125166,
      "grad_norm": 0.7082483172416687,
      "learning_rate": 9.537877098354786e-05,
      "loss": 88.5405,
      "step": 34
    },
    {
      "epoch": 0.3768506056527591,
      "grad_norm": 0.6627039909362793,
      "learning_rate": 9.499233769787535e-05,
      "loss": 88.4298,
      "step": 35
    },
    {
      "epoch": 0.3876177658142665,
      "grad_norm": 0.7085216641426086,
      "learning_rate": 9.459123786076912e-05,
      "loss": 88.4722,
      "step": 36
    },
    {
      "epoch": 0.3983849259757739,
      "grad_norm": 0.6436282396316528,
      "learning_rate": 9.417560222230115e-05,
      "loss": 88.4651,
      "step": 37
    },
    {
      "epoch": 0.40915208613728127,
      "grad_norm": 0.6832641959190369,
      "learning_rate": 9.374556627090749e-05,
      "loss": 88.5401,
      "step": 38
    },
    {
      "epoch": 0.4199192462987887,
      "grad_norm": 0.6698402762413025,
      "learning_rate": 9.330127018922194e-05,
      "loss": 88.4947,
      "step": 39
    },
    {
      "epoch": 0.4306864064602961,
      "grad_norm": 0.753197431564331,
      "learning_rate": 9.284285880837946e-05,
      "loss": 88.4373,
      "step": 40
    },
    {
      "epoch": 0.4414535666218035,
      "grad_norm": 0.6734095215797424,
      "learning_rate": 9.237048156080432e-05,
      "loss": 88.4169,
      "step": 41
    },
    {
      "epoch": 0.4522207267833109,
      "grad_norm": 0.6761934161186218,
      "learning_rate": 9.188429243149824e-05,
      "loss": 88.4011,
      "step": 42
    },
    {
      "epoch": 0.4629878869448183,
      "grad_norm": 0.7175827622413635,
      "learning_rate": 9.138444990784453e-05,
      "loss": 88.4714,
      "step": 43
    },
    {
      "epoch": 0.4737550471063257,
      "grad_norm": 0.7111787796020508,
      "learning_rate": 9.087111692794459e-05,
      "loss": 88.4071,
      "step": 44
    },
    {
      "epoch": 0.4845222072678331,
      "grad_norm": 0.6621391773223877,
      "learning_rate": 9.034446082750352e-05,
      "loss": 88.4675,
      "step": 45
    },
    {
      "epoch": 0.4952893674293405,
      "grad_norm": 0.6450015306472778,
      "learning_rate": 8.980465328528219e-05,
      "loss": 88.3995,
      "step": 46
    },
    {
      "epoch": 0.506056527590848,
      "grad_norm": 0.6297629475593567,
      "learning_rate": 8.925187026713362e-05,
      "loss": 88.4825,
      "step": 47
    },
    {
      "epoch": 0.5168236877523553,
      "grad_norm": 0.6011308431625366,
      "learning_rate": 8.868629196864182e-05,
      "loss": 88.4623,
      "step": 48
    },
    {
      "epoch": 0.5275908479138627,
      "grad_norm": 0.731069803237915,
      "learning_rate": 8.810810275638183e-05,
      "loss": 88.3527,
      "step": 49
    },
    {
      "epoch": 0.5383580080753702,
      "grad_norm": 0.6550359129905701,
      "learning_rate": 8.751749110782012e-05,
      "loss": 88.429,
      "step": 50
    },
    {
      "epoch": 0.5491251682368775,
      "grad_norm": 0.5967565774917603,
      "learning_rate": 8.691464954987493e-05,
      "loss": 88.467,
      "step": 51
    },
    {
      "epoch": 0.5598923283983849,
      "grad_norm": 0.6432732343673706,
      "learning_rate": 8.629977459615655e-05,
      "loss": 88.4472,
      "step": 52
    },
    {
      "epoch": 0.5706594885598923,
      "grad_norm": 0.5874126553535461,
      "learning_rate": 8.567306668290799e-05,
      "loss": 88.4731,
      "step": 53
    },
    {
      "epoch": 0.5814266487213997,
      "grad_norm": 0.6305550336837769,
      "learning_rate": 8.503473010366713e-05,
      "loss": 88.4714,
      "step": 54
    },
    {
      "epoch": 0.5921938088829072,
      "grad_norm": 0.6246164441108704,
      "learning_rate": 8.438497294267117e-05,
      "loss": 88.5134,
      "step": 55
    },
    {
      "epoch": 0.6029609690444145,
      "grad_norm": 0.6057698726654053,
      "learning_rate": 8.37240070070257e-05,
      "loss": 88.5225,
      "step": 56
    },
    {
      "epoch": 0.6137281292059219,
      "grad_norm": 0.6253562569618225,
      "learning_rate": 8.305204775766003e-05,
      "loss": 88.4513,
      "step": 57
    },
    {
      "epoch": 0.6244952893674294,
      "grad_norm": 0.6293994188308716,
      "learning_rate": 8.236931423909138e-05,
      "loss": 88.4613,
      "step": 58
    },
    {
      "epoch": 0.6352624495289367,
      "grad_norm": 0.6635474562644958,
      "learning_rate": 8.16760290080212e-05,
      "loss": 88.4207,
      "step": 59
    },
    {
      "epoch": 0.6460296096904441,
      "grad_norm": 0.641502320766449,
      "learning_rate": 8.097241806078615e-05,
      "loss": 88.4173,
      "step": 60
    },
    {
      "epoch": 0.6567967698519516,
      "grad_norm": 0.6065200567245483,
      "learning_rate": 8.025871075968828e-05,
      "loss": 88.4221,
      "step": 61
    },
    {
      "epoch": 0.6675639300134589,
      "grad_norm": 0.7158219218254089,
      "learning_rate": 7.953513975822755e-05,
      "loss": 88.3222,
      "step": 62
    },
    {
      "epoch": 0.6783310901749664,
      "grad_norm": 0.6504823565483093,
      "learning_rate": 7.880194092526199e-05,
      "loss": 88.3827,
      "step": 63
    },
    {
      "epoch": 0.6890982503364738,
      "grad_norm": 0.624383807182312,
      "learning_rate": 7.805935326811912e-05,
      "loss": 88.4309,
      "step": 64
    },
    {
      "epoch": 0.6998654104979811,
      "grad_norm": 0.6950279474258423,
      "learning_rate": 7.730761885468485e-05,
      "loss": 88.4252,
      "step": 65
    },
    {
      "epoch": 0.7106325706594886,
      "grad_norm": 0.7349709868431091,
      "learning_rate": 7.654698273449435e-05,
      "loss": 88.2957,
      "step": 66
    },
    {
      "epoch": 0.721399730820996,
      "grad_norm": 0.6542060971260071,
      "learning_rate": 7.577769285885109e-05,
      "loss": 88.3716,
      "step": 67
    },
    {
      "epoch": 0.7321668909825033,
      "grad_norm": 0.6578887104988098,
      "learning_rate": 7.500000000000001e-05,
      "loss": 88.3229,
      "step": 68
    },
    {
      "epoch": 0.7429340511440108,
      "grad_norm": 0.7284409403800964,
      "learning_rate": 7.421415766938097e-05,
      "loss": 88.3006,
      "step": 69
    },
    {
      "epoch": 0.7537012113055181,
      "grad_norm": 0.6003475785255432,
      "learning_rate": 7.342042203498951e-05,
      "loss": 88.3455,
      "step": 70
    },
    {
      "epoch": 0.7644683714670256,
      "grad_norm": 0.6060341596603394,
      "learning_rate": 7.261905183787136e-05,
      "loss": 88.3768,
      "step": 71
    },
    {
      "epoch": 0.775235531628533,
      "grad_norm": 0.6123959422111511,
      "learning_rate": 7.181030830777837e-05,
      "loss": 88.3185,
      "step": 72
    },
    {
      "epoch": 0.7860026917900403,
      "grad_norm": 0.5798596143722534,
      "learning_rate": 7.099445507801323e-05,
      "loss": 88.4075,
      "step": 73
    },
    {
      "epoch": 0.7967698519515478,
      "grad_norm": 0.6350551843643188,
      "learning_rate": 7.017175809949044e-05,
      "loss": 88.3893,
      "step": 74
    },
    {
      "epoch": 0.8075370121130552,
      "grad_norm": 0.6014510989189148,
      "learning_rate": 6.934248555404198e-05,
      "loss": 88.3887,
      "step": 75
    },
    {
      "epoch": 0.8183041722745625,
      "grad_norm": 0.5696172714233398,
      "learning_rate": 6.850690776699573e-05,
      "loss": 88.3717,
      "step": 76
    },
    {
      "epoch": 0.82907133243607,
      "grad_norm": 0.5863236784934998,
      "learning_rate": 6.766529711905513e-05,
      "loss": 88.4043,
      "step": 77
    },
    {
      "epoch": 0.8398384925975774,
      "grad_norm": 0.6391366720199585,
      "learning_rate": 6.681792795750875e-05,
      "loss": 88.3141,
      "step": 78
    },
    {
      "epoch": 0.8506056527590848,
      "grad_norm": 0.639610230922699,
      "learning_rate": 6.5965076506799e-05,
      "loss": 88.3615,
      "step": 79
    },
    {
      "epoch": 0.8613728129205922,
      "grad_norm": 0.5825693607330322,
      "learning_rate": 6.510702077847863e-05,
      "loss": 88.3598,
      "step": 80
    },
    {
      "epoch": 0.8721399730820996,
      "grad_norm": 0.6000289916992188,
      "learning_rate": 6.4244040480585e-05,
      "loss": 88.3163,
      "step": 81
    },
    {
      "epoch": 0.882907133243607,
      "grad_norm": 0.6992833614349365,
      "learning_rate": 6.337641692646106e-05,
      "loss": 88.3116,
      "step": 82
    },
    {
      "epoch": 0.8936742934051144,
      "grad_norm": 0.5917866230010986,
      "learning_rate": 6.250443294305315e-05,
      "loss": 88.3113,
      "step": 83
    },
    {
      "epoch": 0.9044414535666218,
      "grad_norm": 0.6197730898857117,
      "learning_rate": 6.162837277871553e-05,
      "loss": 88.3008,
      "step": 84
    },
    {
      "epoch": 0.9152086137281292,
      "grad_norm": 0.5347660779953003,
      "learning_rate": 6.0748522010551215e-05,
      "loss": 88.3428,
      "step": 85
    },
    {
      "epoch": 0.9259757738896366,
      "grad_norm": 0.625930905342102,
      "learning_rate": 5.9865167451320005e-05,
      "loss": 88.4048,
      "step": 86
    },
    {
      "epoch": 0.9367429340511441,
      "grad_norm": 0.613715648651123,
      "learning_rate": 5.897859705594359e-05,
      "loss": 88.3623,
      "step": 87
    },
    {
      "epoch": 0.9475100942126514,
      "grad_norm": 0.6773898601531982,
      "learning_rate": 5.808909982763825e-05,
      "loss": 88.305,
      "step": 88
    },
    {
      "epoch": 0.9582772543741588,
      "grad_norm": 0.6399573087692261,
      "learning_rate": 5.719696572370595e-05,
      "loss": 88.2482,
      "step": 89
    },
    {
      "epoch": 0.9690444145356663,
      "grad_norm": 0.6839354038238525,
      "learning_rate": 5.6302485561014475e-05,
      "loss": 88.281,
      "step": 90
    },
    {
      "epoch": 0.9798115746971736,
      "grad_norm": 0.6810314655303955,
      "learning_rate": 5.540595092119709e-05,
      "loss": 88.2934,
      "step": 91
    },
    {
      "epoch": 0.990578734858681,
      "grad_norm": 0.6228731870651245,
      "learning_rate": 5.4507654055603275e-05,
      "loss": 88.3096,
      "step": 92
    },
    {
      "epoch": 0.990578734858681,
      "eval_loss": 11.04428768157959,
      "eval_runtime": 0.759,
      "eval_samples_per_second": 206.84,
      "eval_steps_per_second": 52.698,
      "step": 92
    },
    {
      "epoch": 1.0067294751009421,
      "grad_norm": 0.5612848997116089,
      "learning_rate": 5.360788779003082e-05,
      "loss": 88.4213,
      "step": 93
    },
    {
      "epoch": 1.0174966352624495,
      "grad_norm": 0.6439865827560425,
      "learning_rate": 5.270694542927088e-05,
      "loss": 88.3083,
      "step": 94
    },
    {
      "epoch": 1.0282637954239569,
      "grad_norm": 0.5426231026649475,
      "learning_rate": 5.180512066149682e-05,
      "loss": 88.3768,
      "step": 95
    },
    {
      "epoch": 1.0390309555854644,
      "grad_norm": 0.6521313190460205,
      "learning_rate": 5.090270746252802e-05,
      "loss": 88.2851,
      "step": 96
    },
    {
      "epoch": 1.0497981157469718,
      "grad_norm": 0.601963996887207,
      "learning_rate": 5e-05,
      "loss": 88.3373,
      "step": 97
    },
    {
      "epoch": 1.0605652759084792,
      "grad_norm": 0.6257451176643372,
      "learning_rate": 4.909729253747197e-05,
      "loss": 88.3497,
      "step": 98
    },
    {
      "epoch": 1.0713324360699865,
      "grad_norm": 0.6289570927619934,
      "learning_rate": 4.819487933850319e-05,
      "loss": 88.3188,
      "step": 99
    },
    {
      "epoch": 1.0820995962314939,
      "grad_norm": 0.6000854969024658,
      "learning_rate": 4.729305457072913e-05,
      "loss": 88.3107,
      "step": 100
    },
    {
      "epoch": 1.0928667563930015,
      "grad_norm": 0.7118406295776367,
      "learning_rate": 4.63921122099692e-05,
      "loss": 88.2865,
      "step": 101
    },
    {
      "epoch": 1.1036339165545088,
      "grad_norm": 0.5987817049026489,
      "learning_rate": 4.549234594439674e-05,
      "loss": 88.2681,
      "step": 102
    },
    {
      "epoch": 1.1144010767160162,
      "grad_norm": 0.5965442657470703,
      "learning_rate": 4.4594049078802925e-05,
      "loss": 88.3694,
      "step": 103
    },
    {
      "epoch": 1.1251682368775235,
      "grad_norm": 0.5947834253311157,
      "learning_rate": 4.3697514438985536e-05,
      "loss": 88.3057,
      "step": 104
    },
    {
      "epoch": 1.135935397039031,
      "grad_norm": 0.6357874870300293,
      "learning_rate": 4.280303427629404e-05,
      "loss": 88.2947,
      "step": 105
    },
    {
      "epoch": 1.1467025572005383,
      "grad_norm": 0.5863412618637085,
      "learning_rate": 4.1910900172361764e-05,
      "loss": 88.342,
      "step": 106
    },
    {
      "epoch": 1.1574697173620458,
      "grad_norm": 0.6012663245201111,
      "learning_rate": 4.1021402944056416e-05,
      "loss": 88.2721,
      "step": 107
    },
    {
      "epoch": 1.1682368775235532,
      "grad_norm": 0.6170640587806702,
      "learning_rate": 4.0134832548680006e-05,
      "loss": 88.3084,
      "step": 108
    },
    {
      "epoch": 1.1790040376850606,
      "grad_norm": 0.6174831390380859,
      "learning_rate": 3.92514779894488e-05,
      "loss": 88.3112,
      "step": 109
    },
    {
      "epoch": 1.189771197846568,
      "grad_norm": 0.5705954432487488,
      "learning_rate": 3.8371627221284495e-05,
      "loss": 88.3226,
      "step": 110
    },
    {
      "epoch": 1.2005383580080753,
      "grad_norm": 0.6390698552131653,
      "learning_rate": 3.7495567056946855e-05,
      "loss": 88.2471,
      "step": 111
    },
    {
      "epoch": 1.2113055181695827,
      "grad_norm": 0.643191397190094,
      "learning_rate": 3.6623583073538966e-05,
      "loss": 88.1814,
      "step": 112
    },
    {
      "epoch": 1.2220726783310902,
      "grad_norm": 0.6732293963432312,
      "learning_rate": 3.5755959519415005e-05,
      "loss": 88.2546,
      "step": 113
    },
    {
      "epoch": 1.2328398384925976,
      "grad_norm": 0.6384011507034302,
      "learning_rate": 3.489297922152136e-05,
      "loss": 88.2384,
      "step": 114
    },
    {
      "epoch": 1.243606998654105,
      "grad_norm": 0.5936709046363831,
      "learning_rate": 3.403492349320101e-05,
      "loss": 88.2589,
      "step": 115
    },
    {
      "epoch": 1.2543741588156123,
      "grad_norm": 0.5669568777084351,
      "learning_rate": 3.3182072042491244e-05,
      "loss": 88.2816,
      "step": 116
    },
    {
      "epoch": 1.2651413189771197,
      "grad_norm": 0.5536919236183167,
      "learning_rate": 3.2334702880944886e-05,
      "loss": 88.3449,
      "step": 117
    },
    {
      "epoch": 1.2759084791386273,
      "grad_norm": 0.5760201811790466,
      "learning_rate": 3.149309223300428e-05,
      "loss": 88.3198,
      "step": 118
    },
    {
      "epoch": 1.2866756393001346,
      "grad_norm": 0.62739497423172,
      "learning_rate": 3.065751444595805e-05,
      "loss": 88.2775,
      "step": 119
    },
    {
      "epoch": 1.297442799461642,
      "grad_norm": 0.5816006064414978,
      "learning_rate": 2.982824190050958e-05,
      "loss": 88.3048,
      "step": 120
    },
    {
      "epoch": 1.3082099596231493,
      "grad_norm": 0.6178335547447205,
      "learning_rate": 2.900554492198677e-05,
      "loss": 88.3031,
      "step": 121
    },
    {
      "epoch": 1.3189771197846567,
      "grad_norm": 0.6052290797233582,
      "learning_rate": 2.8189691692221627e-05,
      "loss": 88.2896,
      "step": 122
    },
    {
      "epoch": 1.3297442799461643,
      "grad_norm": 0.5566094517707825,
      "learning_rate": 2.738094816212866e-05,
      "loss": 88.3939,
      "step": 123
    },
    {
      "epoch": 1.3405114401076716,
      "grad_norm": 0.6167682409286499,
      "learning_rate": 2.65795779650105e-05,
      "loss": 88.283,
      "step": 124
    },
    {
      "epoch": 1.351278600269179,
      "grad_norm": 0.6485788822174072,
      "learning_rate": 2.5785842330619038e-05,
      "loss": 88.341,
      "step": 125
    },
    {
      "epoch": 1.3620457604306864,
      "grad_norm": 0.6236302256584167,
      "learning_rate": 2.500000000000001e-05,
      "loss": 88.3239,
      "step": 126
    },
    {
      "epoch": 1.3728129205921937,
      "grad_norm": 0.6133517622947693,
      "learning_rate": 2.422230714114891e-05,
      "loss": 88.257,
      "step": 127
    },
    {
      "epoch": 1.3835800807537013,
      "grad_norm": 0.5628191828727722,
      "learning_rate": 2.3453017265505673e-05,
      "loss": 88.2815,
      "step": 128
    },
    {
      "epoch": 1.3943472409152087,
      "grad_norm": 0.6054531931877136,
      "learning_rate": 2.269238114531515e-05,
      "loss": 88.2568,
      "step": 129
    },
    {
      "epoch": 1.405114401076716,
      "grad_norm": 0.6800547242164612,
      "learning_rate": 2.194064673188089e-05,
      "loss": 88.2577,
      "step": 130
    },
    {
      "epoch": 1.4158815612382234,
      "grad_norm": 0.6266992092132568,
      "learning_rate": 2.1198059074738024e-05,
      "loss": 88.2155,
      "step": 131
    },
    {
      "epoch": 1.4266487213997308,
      "grad_norm": 0.6254451274871826,
      "learning_rate": 2.0464860241772455e-05,
      "loss": 88.2093,
      "step": 132
    },
    {
      "epoch": 1.4374158815612383,
      "grad_norm": 0.6577990055084229,
      "learning_rate": 1.9741289240311755e-05,
      "loss": 88.1898,
      "step": 133
    },
    {
      "epoch": 1.4481830417227457,
      "grad_norm": 0.6361428499221802,
      "learning_rate": 1.902758193921385e-05,
      "loss": 88.3488,
      "step": 134
    },
    {
      "epoch": 1.458950201884253,
      "grad_norm": 0.6057702302932739,
      "learning_rate": 1.832397099197882e-05,
      "loss": 88.2169,
      "step": 135
    },
    {
      "epoch": 1.4697173620457604,
      "grad_norm": 0.5825548768043518,
      "learning_rate": 1.7630685760908622e-05,
      "loss": 88.2397,
      "step": 136
    },
    {
      "epoch": 1.4804845222072678,
      "grad_norm": 0.7038293480873108,
      "learning_rate": 1.6947952242339992e-05,
      "loss": 88.1477,
      "step": 137
    },
    {
      "epoch": 1.4912516823687754,
      "grad_norm": 0.6376841068267822,
      "learning_rate": 1.6275992992974308e-05,
      "loss": 88.2545,
      "step": 138
    },
    {
      "epoch": 1.5020188425302825,
      "grad_norm": 0.653458833694458,
      "learning_rate": 1.561502705732883e-05,
      "loss": 88.2055,
      "step": 139
    },
    {
      "epoch": 1.51278600269179,
      "grad_norm": 0.6237602829933167,
      "learning_rate": 1.4965269896332885e-05,
      "loss": 88.2087,
      "step": 140
    },
    {
      "epoch": 1.5235531628532974,
      "grad_norm": 0.6579408645629883,
      "learning_rate": 1.4326933317092e-05,
      "loss": 88.2702,
      "step": 141
    },
    {
      "epoch": 1.5343203230148048,
      "grad_norm": 0.5924363136291504,
      "learning_rate": 1.3700225403843469e-05,
      "loss": 88.248,
      "step": 142
    },
    {
      "epoch": 1.5450874831763124,
      "grad_norm": 0.6045029759407043,
      "learning_rate": 1.3085350450125072e-05,
      "loss": 88.3215,
      "step": 143
    },
    {
      "epoch": 1.5558546433378195,
      "grad_norm": 0.6474866271018982,
      "learning_rate": 1.2482508892179884e-05,
      "loss": 88.2769,
      "step": 144
    },
    {
      "epoch": 1.5666218034993271,
      "grad_norm": 0.6447649002075195,
      "learning_rate": 1.1891897243618182e-05,
      "loss": 88.2294,
      "step": 145
    },
    {
      "epoch": 1.5773889636608345,
      "grad_norm": 0.6493783593177795,
      "learning_rate": 1.1313708031358183e-05,
      "loss": 88.2699,
      "step": 146
    },
    {
      "epoch": 1.5881561238223418,
      "grad_norm": 0.6726663112640381,
      "learning_rate": 1.0748129732866391e-05,
      "loss": 88.2096,
      "step": 147
    },
    {
      "epoch": 1.5989232839838494,
      "grad_norm": 0.5635113716125488,
      "learning_rate": 1.0195346714717813e-05,
      "loss": 88.3316,
      "step": 148
    },
    {
      "epoch": 1.6096904441453566,
      "grad_norm": 0.6250278949737549,
      "learning_rate": 9.65553917249648e-06,
      "loss": 88.2534,
      "step": 149
    },
    {
      "epoch": 1.6204576043068641,
      "grad_norm": 0.61742103099823,
      "learning_rate": 9.12888307205541e-06,
      "loss": 88.3115,
      "step": 150
    },
    {
      "epoch": 1.6312247644683715,
      "grad_norm": 0.6726629137992859,
      "learning_rate": 8.615550092155478e-06,
      "loss": 88.1615,
      "step": 151
    },
    {
      "epoch": 1.6419919246298789,
      "grad_norm": 0.6055439114570618,
      "learning_rate": 8.115707568501768e-06,
      "loss": 88.2359,
      "step": 152
    },
    {
      "epoch": 1.6527590847913862,
      "grad_norm": 0.646351158618927,
      "learning_rate": 7.629518439195671e-06,
      "loss": 88.2837,
      "step": 153
    },
    {
      "epoch": 1.6635262449528936,
      "grad_norm": 0.6116978526115417,
      "learning_rate": 7.157141191620548e-06,
      "loss": 88.2912,
      "step": 154
    },
    {
      "epoch": 1.6742934051144012,
      "grad_norm": 0.6652332544326782,
      "learning_rate": 6.698729810778065e-06,
      "loss": 88.2366,
      "step": 155
    },
    {
      "epoch": 1.6850605652759085,
      "grad_norm": 0.7036592364311218,
      "learning_rate": 6.2544337290925185e-06,
      "loss": 88.2283,
      "step": 156
    },
    {
      "epoch": 1.695827725437416,
      "grad_norm": 0.6081019043922424,
      "learning_rate": 5.824397777698859e-06,
      "loss": 88.2434,
      "step": 157
    },
    {
      "epoch": 1.7065948855989233,
      "grad_norm": 0.6562557220458984,
      "learning_rate": 5.408762139230888e-06,
      "loss": 88.1598,
      "step": 158
    },
    {
      "epoch": 1.7173620457604306,
      "grad_norm": 0.6159968376159668,
      "learning_rate": 5.007662302124672e-06,
      "loss": 88.2584,
      "step": 159
    },
    {
      "epoch": 1.7281292059219382,
      "grad_norm": 0.6619474291801453,
      "learning_rate": 4.621229016452156e-06,
      "loss": 88.18,
      "step": 160
    },
    {
      "epoch": 1.7388963660834453,
      "grad_norm": 0.6214718222618103,
      "learning_rate": 4.249588251299391e-06,
      "loss": 88.2279,
      "step": 161
    },
    {
      "epoch": 1.749663526244953,
      "grad_norm": 0.6064915657043457,
      "learning_rate": 3.892861153703342e-06,
      "loss": 88.2215,
      "step": 162
    },
    {
      "epoch": 1.7604306864064603,
      "grad_norm": 0.5849805474281311,
      "learning_rate": 3.551164009160429e-06,
      "loss": 88.2112,
      "step": 163
    },
    {
      "epoch": 1.7711978465679676,
      "grad_norm": 0.6270782351493835,
      "learning_rate": 3.2246082037199532e-06,
      "loss": 88.2335,
      "step": 164
    },
    {
      "epoch": 1.7819650067294752,
      "grad_norm": 0.5952561497688293,
      "learning_rate": 2.9133001876746004e-06,
      "loss": 88.2984,
      "step": 165
    },
    {
      "epoch": 1.7927321668909824,
      "grad_norm": 0.6238298416137695,
      "learning_rate": 2.6173414408598827e-06,
      "loss": 88.2918,
      "step": 166
    },
    {
      "epoch": 1.80349932705249,
      "grad_norm": 0.6672609448432922,
      "learning_rate": 2.3368284395738684e-06,
      "loss": 88.2529,
      "step": 167
    },
    {
      "epoch": 1.8142664872139973,
      "grad_norm": 0.5070897340774536,
      "learning_rate": 2.0718526251279346e-06,
      "loss": 88.3376,
      "step": 168
    },
    {
      "epoch": 1.8250336473755047,
      "grad_norm": 0.6033185124397278,
      "learning_rate": 1.8225003740388547e-06,
      "loss": 88.3511,
      "step": 169
    },
    {
      "epoch": 1.8358008075370122,
      "grad_norm": 0.5488670468330383,
      "learning_rate": 1.5888529698718346e-06,
      "loss": 88.3084,
      "step": 170
    },
    {
      "epoch": 1.8465679676985194,
      "grad_norm": 0.6001074910163879,
      "learning_rate": 1.3709865767438435e-06,
      "loss": 88.2296,
      "step": 171
    },
    {
      "epoch": 1.857335127860027,
      "grad_norm": 0.5917924642562866,
      "learning_rate": 1.1689722144956671e-06,
      "loss": 88.3425,
      "step": 172
    },
    {
      "epoch": 1.8681022880215343,
      "grad_norm": 0.5938112735748291,
      "learning_rate": 9.82875735540989e-07,
      "loss": 88.2656,
      "step": 173
    },
    {
      "epoch": 1.8788694481830417,
      "grad_norm": 0.6183878779411316,
      "learning_rate": 8.127578033998662e-07,
      "loss": 88.2139,
      "step": 174
    },
    {
      "epoch": 1.8896366083445493,
      "grad_norm": 0.6551972031593323,
      "learning_rate": 6.58673872923693e-07,
      "loss": 88.2028,
      "step": 175
    },
    {
      "epoch": 1.9004037685060564,
      "grad_norm": 0.6331955194473267,
      "learning_rate": 5.206741722181386e-07,
      "loss": 88.2651,
      "step": 176
    },
    {
      "epoch": 1.911170928667564,
      "grad_norm": 0.6921940445899963,
      "learning_rate": 3.9880368626978304e-07,
      "loss": 88.2543,
      "step": 177
    },
    {
      "epoch": 1.9219380888290714,
      "grad_norm": 0.6135378479957581,
      "learning_rate": 2.9310214228202013e-07,
      "loss": 88.269,
      "step": 178
    },
    {
      "epoch": 1.9327052489905787,
      "grad_norm": 0.5835920572280884,
      "learning_rate": 2.0360399672478824e-07,
      "loss": 88.267,
      "step": 179
    },
    {
      "epoch": 1.9434724091520863,
      "grad_norm": 0.5504326224327087,
      "learning_rate": 1.3033842410251075e-07,
      "loss": 88.3445,
      "step": 180
    },
    {
      "epoch": 1.9542395693135934,
      "grad_norm": 0.6638001799583435,
      "learning_rate": 7.332930744380906e-08,
      "loss": 88.2083,
      "step": 181
    },
    {
      "epoch": 1.965006729475101,
      "grad_norm": 0.69869065284729,
      "learning_rate": 3.259523051615254e-08,
      "loss": 88.262,
      "step": 182
    },
    {
      "epoch": 1.9757738896366084,
      "grad_norm": 0.6452627778053284,
      "learning_rate": 8.149471767937567e-09,
      "loss": 88.2167,
      "step": 183
    },
    {
      "epoch": 1.9865410497981157,
      "grad_norm": 0.6180316805839539,
      "learning_rate": 0.0,
      "loss": 88.3043,
      "step": 184
    },
    {
      "epoch": 1.9865410497981157,
      "eval_loss": 11.03515911102295,
      "eval_runtime": 0.3851,
      "eval_samples_per_second": 407.704,
      "eval_steps_per_second": 103.874,
      "step": 184
    }
  ],
  "logging_steps": 1,
  "max_steps": 184,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 1943745331200.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}