{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9788972089857046,
  "eval_steps": 500,
  "global_step": 273,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.010891763104152484,
      "grad_norm": 5.917168704218648,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.8738,
      "step": 1
    },
    {
      "epoch": 0.021783526208304968,
      "grad_norm": 5.848189651981173,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.8637,
      "step": 2
    },
    {
      "epoch": 0.03267528931245745,
      "grad_norm": 5.4410485869996865,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.8524,
      "step": 3
    },
    {
      "epoch": 0.043567052416609936,
      "grad_norm": 2.423684029964203,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.783,
      "step": 4
    },
    {
      "epoch": 0.05445881552076242,
      "grad_norm": 4.0235503087845155,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.7805,
      "step": 5
    },
    {
      "epoch": 0.0653505786249149,
      "grad_norm": 4.14511666016107,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.7643,
      "step": 6
    },
    {
      "epoch": 0.0762423417290674,
      "grad_norm": 4.42039479030743,
      "learning_rate": 2e-05,
      "loss": 0.7512,
      "step": 7
    },
    {
      "epoch": 0.08713410483321987,
      "grad_norm": 2.857807259160155,
      "learning_rate": 2.2857142857142858e-05,
      "loss": 0.7182,
      "step": 8
    },
    {
      "epoch": 0.09802586793737236,
      "grad_norm": 2.690272592093101,
      "learning_rate": 2.5714285714285718e-05,
      "loss": 0.6835,
      "step": 9
    },
    {
      "epoch": 0.10891763104152484,
      "grad_norm": 2.0185553871545165,
      "learning_rate": 2.8571428571428574e-05,
      "loss": 0.666,
      "step": 10
    },
    {
      "epoch": 0.11980939414567733,
      "grad_norm": 1.6526659811622688,
      "learning_rate": 3.142857142857143e-05,
      "loss": 0.6551,
      "step": 11
    },
    {
      "epoch": 0.1307011572498298,
      "grad_norm": 1.3684051055691528,
      "learning_rate": 3.4285714285714284e-05,
      "loss": 0.6374,
      "step": 12
    },
    {
      "epoch": 0.1415929203539823,
      "grad_norm": 1.3936888061646935,
      "learning_rate": 3.714285714285715e-05,
      "loss": 0.6142,
      "step": 13
    },
    {
      "epoch": 0.1524846834581348,
      "grad_norm": 1.3347032728939032,
      "learning_rate": 4e-05,
      "loss": 0.6085,
      "step": 14
    },
    {
      "epoch": 0.16337644656228728,
      "grad_norm": 1.6041646981350726,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.6168,
      "step": 15
    },
    {
      "epoch": 0.17426820966643974,
      "grad_norm": 1.250156217488812,
      "learning_rate": 4.5714285714285716e-05,
      "loss": 0.5904,
      "step": 16
    },
    {
      "epoch": 0.18515997277059223,
      "grad_norm": 1.363684815879282,
      "learning_rate": 4.857142857142857e-05,
      "loss": 0.5983,
      "step": 17
    },
    {
      "epoch": 0.19605173587474473,
      "grad_norm": 1.160866980807094,
      "learning_rate": 5.1428571428571436e-05,
      "loss": 0.5762,
      "step": 18
    },
    {
      "epoch": 0.20694349897889722,
      "grad_norm": 1.8320088500371636,
      "learning_rate": 5.4285714285714295e-05,
      "loss": 0.5794,
      "step": 19
    },
    {
      "epoch": 0.21783526208304968,
      "grad_norm": 1.310331415981151,
      "learning_rate": 5.714285714285715e-05,
      "loss": 0.5723,
      "step": 20
    },
    {
      "epoch": 0.22872702518720217,
      "grad_norm": 2.0403045692700275,
      "learning_rate": 6.000000000000001e-05,
      "loss": 0.576,
      "step": 21
    },
    {
      "epoch": 0.23961878829135466,
      "grad_norm": 1.0753217952290004,
      "learning_rate": 6.285714285714286e-05,
      "loss": 0.5657,
      "step": 22
    },
    {
      "epoch": 0.25051055139550715,
      "grad_norm": 2.0101370583197076,
      "learning_rate": 6.571428571428571e-05,
      "loss": 0.5751,
      "step": 23
    },
    {
      "epoch": 0.2614023144996596,
      "grad_norm": 1.8269660980505025,
      "learning_rate": 6.857142857142857e-05,
      "loss": 0.5622,
      "step": 24
    },
    {
      "epoch": 0.27229407760381213,
      "grad_norm": 1.333190530233097,
      "learning_rate": 7.142857142857143e-05,
      "loss": 0.553,
      "step": 25
    },
    {
      "epoch": 0.2831858407079646,
      "grad_norm": 1.5278657539191043,
      "learning_rate": 7.42857142857143e-05,
      "loss": 0.5746,
      "step": 26
    },
    {
      "epoch": 0.29407760381211706,
      "grad_norm": 1.4753801604023495,
      "learning_rate": 7.714285714285715e-05,
      "loss": 0.5504,
      "step": 27
    },
    {
      "epoch": 0.3049693669162696,
      "grad_norm": 1.4736990838732313,
      "learning_rate": 8e-05,
      "loss": 0.5542,
      "step": 28
    },
    {
      "epoch": 0.31586113002042204,
      "grad_norm": 2.3486277040401893,
      "learning_rate": 7.999671154713278e-05,
      "loss": 0.5507,
      "step": 29
    },
    {
      "epoch": 0.32675289312457456,
      "grad_norm": 1.1844138236301467,
      "learning_rate": 7.99868467292272e-05,
      "loss": 0.5496,
      "step": 30
    },
    {
      "epoch": 0.337644656228727,
      "grad_norm": 2.4920955827056748,
      "learning_rate": 7.997040716828271e-05,
      "loss": 0.5575,
      "step": 31
    },
    {
      "epoch": 0.3485364193328795,
      "grad_norm": 1.7099423006847476,
      "learning_rate": 7.994739556733538e-05,
      "loss": 0.5596,
      "step": 32
    },
    {
      "epoch": 0.359428182437032,
      "grad_norm": 2.1342906902090504,
      "learning_rate": 7.991781571001347e-05,
      "loss": 0.5337,
      "step": 33
    },
    {
      "epoch": 0.37031994554118447,
      "grad_norm": 1.7483277996825424,
      "learning_rate": 7.988167245991528e-05,
      "loss": 0.5492,
      "step": 34
    },
    {
      "epoch": 0.381211708645337,
      "grad_norm": 1.7968259576328387,
      "learning_rate": 7.983897175980957e-05,
      "loss": 0.5314,
      "step": 35
    },
    {
      "epoch": 0.39210347174948945,
      "grad_norm": 1.559300515616932,
      "learning_rate": 7.97897206306583e-05,
      "loss": 0.5307,
      "step": 36
    },
    {
      "epoch": 0.4029952348536419,
      "grad_norm": 1.0889578452931736,
      "learning_rate": 7.973392717046233e-05,
      "loss": 0.5408,
      "step": 37
    },
    {
      "epoch": 0.41388699795779443,
      "grad_norm": 1.313871076023579,
      "learning_rate": 7.967160055292984e-05,
      "loss": 0.5147,
      "step": 38
    },
    {
      "epoch": 0.4247787610619469,
      "grad_norm": 1.583430729115433,
      "learning_rate": 7.960275102596809e-05,
      "loss": 0.5216,
      "step": 39
    },
    {
      "epoch": 0.43567052416609936,
      "grad_norm": 1.3933127292778362,
      "learning_rate": 7.952738990999824e-05,
      "loss": 0.5187,
      "step": 40
    },
    {
      "epoch": 0.4465622872702519,
      "grad_norm": 1.4991075926399846,
      "learning_rate": 7.94455295960942e-05,
      "loss": 0.5123,
      "step": 41
    },
    {
      "epoch": 0.45745405037440434,
      "grad_norm": 1.2790673350427315,
      "learning_rate": 7.93571835439452e-05,
      "loss": 0.5255,
      "step": 42
    },
    {
      "epoch": 0.46834581347855686,
      "grad_norm": 1.656174041977428,
      "learning_rate": 7.926236627964262e-05,
      "loss": 0.5224,
      "step": 43
    },
    {
      "epoch": 0.4792375765827093,
      "grad_norm": 1.3680863968048569,
      "learning_rate": 7.916109339329173e-05,
      "loss": 0.5104,
      "step": 44
    },
    {
      "epoch": 0.4901293396868618,
      "grad_norm": 1.1248700116366424,
      "learning_rate": 7.905338153644818e-05,
      "loss": 0.5036,
      "step": 45
    },
    {
      "epoch": 0.5010211027910143,
      "grad_norm": 1.4820737812996578,
      "learning_rate": 7.89392484193802e-05,
      "loss": 0.5169,
      "step": 46
    },
    {
      "epoch": 0.5119128658951668,
      "grad_norm": 1.1533791684607533,
      "learning_rate": 7.881871280815659e-05,
      "loss": 0.5071,
      "step": 47
    },
    {
      "epoch": 0.5228046289993192,
      "grad_norm": 1.9492363638377506,
      "learning_rate": 7.869179452156118e-05,
      "loss": 0.5093,
      "step": 48
    },
    {
      "epoch": 0.5336963921034718,
      "grad_norm": 0.9617512636578286,
      "learning_rate": 7.855851442783414e-05,
      "loss": 0.501,
      "step": 49
    },
    {
      "epoch": 0.5445881552076243,
      "grad_norm": 1.0817397757606,
      "learning_rate": 7.841889444124078e-05,
      "loss": 0.4997,
      "step": 50
    },
    {
      "epoch": 0.5554799183117767,
      "grad_norm": 2.1099922597533074,
      "learning_rate": 7.827295751846836e-05,
      "loss": 0.5151,
      "step": 51
    },
    {
      "epoch": 0.5663716814159292,
      "grad_norm": 1.0752679988230807,
      "learning_rate": 7.81207276548515e-05,
      "loss": 0.512,
      "step": 52
    },
    {
      "epoch": 0.5772634445200817,
      "grad_norm": 2.033228507248604,
      "learning_rate": 7.796222988042676e-05,
      "loss": 0.5247,
      "step": 53
    },
    {
      "epoch": 0.5881552076242341,
      "grad_norm": 1.4290947325043002,
      "learning_rate": 7.779749025581717e-05,
      "loss": 0.5161,
      "step": 54
    },
    {
      "epoch": 0.5990469707283866,
      "grad_norm": 1.6107474254316703,
      "learning_rate": 7.762653586794731e-05,
      "loss": 0.5113,
      "step": 55
    },
    {
      "epoch": 0.6099387338325392,
      "grad_norm": 1.7403793303180215,
      "learning_rate": 7.74493948255895e-05,
      "loss": 0.5104,
      "step": 56
    },
    {
      "epoch": 0.6208304969366917,
      "grad_norm": 1.3273260664695445,
      "learning_rate": 7.726609625474218e-05,
      "loss": 0.5149,
      "step": 57
    },
    {
      "epoch": 0.6317222600408441,
      "grad_norm": 1.2835468728323596,
      "learning_rate": 7.707667029384088e-05,
      "loss": 0.4985,
      "step": 58
    },
    {
      "epoch": 0.6426140231449966,
      "grad_norm": 1.3493586560972042,
      "learning_rate": 7.688114808880283e-05,
      "loss": 0.515,
      "step": 59
    },
    {
      "epoch": 0.6535057862491491,
      "grad_norm": 0.8317273919287447,
      "learning_rate": 7.667956178790582e-05,
      "loss": 0.5059,
      "step": 60
    },
    {
      "epoch": 0.6643975493533015,
      "grad_norm": 1.1983241891992835,
      "learning_rate": 7.647194453650228e-05,
      "loss": 0.4935,
      "step": 61
    },
    {
      "epoch": 0.675289312457454,
      "grad_norm": 1.1964694002526848,
      "learning_rate": 7.625833047156953e-05,
      "loss": 0.5058,
      "step": 62
    },
    {
      "epoch": 0.6861810755616066,
      "grad_norm": 1.1878899519661879,
      "learning_rate": 7.603875471609677e-05,
      "loss": 0.4903,
      "step": 63
    },
    {
      "epoch": 0.697072838665759,
      "grad_norm": 1.2583807783190593,
      "learning_rate": 7.581325337331013e-05,
      "loss": 0.497,
      "step": 64
    },
    {
      "epoch": 0.7079646017699115,
      "grad_norm": 0.9383067837973975,
      "learning_rate": 7.558186352073648e-05,
      "loss": 0.4874,
      "step": 65
    },
    {
      "epoch": 0.718856364874064,
      "grad_norm": 1.114271813149658,
      "learning_rate": 7.534462320410702e-05,
      "loss": 0.4957,
      "step": 66
    },
    {
      "epoch": 0.7297481279782164,
      "grad_norm": 0.9889746448795311,
      "learning_rate": 7.510157143110172e-05,
      "loss": 0.4976,
      "step": 67
    },
    {
      "epoch": 0.7406398910823689,
      "grad_norm": 1.5156805885931242,
      "learning_rate": 7.485274816493558e-05,
      "loss": 0.4959,
      "step": 68
    },
    {
      "epoch": 0.7515316541865215,
      "grad_norm": 0.9749471928652824,
      "learning_rate": 7.459819431778775e-05,
      "loss": 0.4796,
      "step": 69
    },
    {
      "epoch": 0.762423417290674,
      "grad_norm": 0.8404206560752534,
      "learning_rate": 7.433795174407465e-05,
      "loss": 0.4762,
      "step": 70
    },
    {
      "epoch": 0.7733151803948264,
      "grad_norm": 1.2857273953478379,
      "learning_rate": 7.407206323356818e-05,
      "loss": 0.4955,
      "step": 71
    },
    {
      "epoch": 0.7842069434989789,
      "grad_norm": 1.5254696719564749,
      "learning_rate": 7.380057250436006e-05,
      "loss": 0.4945,
      "step": 72
    },
    {
      "epoch": 0.7950987066031314,
      "grad_norm": 1.0225507119621404,
      "learning_rate": 7.352352419567362e-05,
      "loss": 0.4883,
      "step": 73
    },
    {
      "epoch": 0.8059904697072838,
      "grad_norm": 1.3824406570348136,
      "learning_rate": 7.324096386052416e-05,
      "loss": 0.4906,
      "step": 74
    },
    {
      "epoch": 0.8168822328114363,
      "grad_norm": 0.9568602258759329,
      "learning_rate": 7.295293795822887e-05,
      "loss": 0.4838,
      "step": 75
    },
    {
      "epoch": 0.8277739959155889,
      "grad_norm": 1.2765315330970022,
      "learning_rate": 7.265949384676795e-05,
      "loss": 0.4815,
      "step": 76
    },
    {
      "epoch": 0.8386657590197413,
      "grad_norm": 1.2279324061842216,
      "learning_rate": 7.236067977499791e-05,
      "loss": 0.4855,
      "step": 77
    },
    {
      "epoch": 0.8495575221238938,
      "grad_norm": 0.9629770077059606,
      "learning_rate": 7.205654487471826e-05,
      "loss": 0.4811,
      "step": 78
    },
    {
      "epoch": 0.8604492852280463,
      "grad_norm": 1.295116932433786,
      "learning_rate": 7.174713915259331e-05,
      "loss": 0.4889,
      "step": 79
    },
    {
      "epoch": 0.8713410483321987,
      "grad_norm": 0.9924719057218118,
      "learning_rate": 7.143251348192971e-05,
      "loss": 0.488,
      "step": 80
    },
    {
      "epoch": 0.8822328114363512,
      "grad_norm": 1.3773423039136135,
      "learning_rate": 7.111271959431189e-05,
      "loss": 0.4859,
      "step": 81
    },
    {
      "epoch": 0.8931245745405038,
      "grad_norm": 0.9416135398394887,
      "learning_rate": 7.078781007109625e-05,
      "loss": 0.4803,
      "step": 82
    },
    {
      "epoch": 0.9040163376446563,
      "grad_norm": 1.0674313281470116,
      "learning_rate": 7.045783833476538e-05,
      "loss": 0.4733,
      "step": 83
    },
    {
      "epoch": 0.9149081007488087,
      "grad_norm": 1.000464101793752,
      "learning_rate": 7.012285864014445e-05,
      "loss": 0.4726,
      "step": 84
    },
    {
      "epoch": 0.9257998638529612,
      "grad_norm": 1.3944228987734595,
      "learning_rate": 6.978292606548029e-05,
      "loss": 0.4817,
      "step": 85
    },
    {
      "epoch": 0.9366916269571137,
      "grad_norm": 0.9210477820477416,
      "learning_rate": 6.943809650338541e-05,
      "loss": 0.4856,
      "step": 86
    },
    {
      "epoch": 0.9475833900612661,
      "grad_norm": 1.4702724614133884,
      "learning_rate": 6.908842665164789e-05,
      "loss": 0.4893,
      "step": 87
    },
    {
      "epoch": 0.9584751531654186,
      "grad_norm": 0.9392028605617796,
      "learning_rate": 6.873397400390911e-05,
      "loss": 0.4749,
      "step": 88
    },
    {
      "epoch": 0.9693669162695712,
      "grad_norm": 1.33752177947439,
      "learning_rate": 6.837479684021032e-05,
      "loss": 0.4642,
      "step": 89
    },
    {
      "epoch": 0.9802586793737236,
      "grad_norm": 1.0674472905397991,
      "learning_rate": 6.80109542174102e-05,
      "loss": 0.4721,
      "step": 90
    },
    {
      "epoch": 0.9911504424778761,
      "grad_norm": 0.8855006430820443,
      "learning_rate": 6.76425059594746e-05,
      "loss": 0.471,
      "step": 91
    },
    {
      "epoch": 1.0047651463580667,
      "grad_norm": 0.7477549645216865,
      "learning_rate": 6.726951264763998e-05,
      "loss": 0.4723,
      "step": 92
    },
    {
      "epoch": 1.0156569094622192,
      "grad_norm": 0.8956363744921413,
      "learning_rate": 6.689203561045268e-05,
      "loss": 0.4562,
      "step": 93
    },
    {
      "epoch": 1.0265486725663717,
      "grad_norm": 1.060602408653099,
      "learning_rate": 6.651013691368492e-05,
      "loss": 0.4599,
      "step": 94
    },
    {
      "epoch": 1.0374404356705242,
      "grad_norm": 1.2755687426821307,
      "learning_rate": 6.612387935012995e-05,
      "loss": 0.4586,
      "step": 95
    },
    {
      "epoch": 1.0483321987746765,
      "grad_norm": 0.7866789226297869,
      "learning_rate": 6.573332642927737e-05,
      "loss": 0.4564,
      "step": 96
    },
    {
      "epoch": 1.059223961878829,
      "grad_norm": 1.0671580779917749,
      "learning_rate": 6.53385423668708e-05,
      "loss": 0.4532,
      "step": 97
    },
    {
      "epoch": 1.0701157249829816,
      "grad_norm": 1.8705582630594135,
      "learning_rate": 6.493959207434934e-05,
      "loss": 0.4518,
      "step": 98
    },
    {
      "epoch": 1.081007488087134,
      "grad_norm": 0.8747747648963227,
      "learning_rate": 6.453654114817467e-05,
      "loss": 0.4602,
      "step": 99
    },
    {
      "epoch": 1.0918992511912866,
      "grad_norm": 2.6708530213907915,
      "learning_rate": 6.412945585904545e-05,
      "loss": 0.4706,
      "step": 100
    },
    {
      "epoch": 1.1027910142954391,
      "grad_norm": 2.059630403847963,
      "learning_rate": 6.371840314100104e-05,
      "loss": 0.4746,
      "step": 101
    },
    {
      "epoch": 1.1136827773995917,
      "grad_norm": 2.100904334906369,
      "learning_rate": 6.330345058041585e-05,
      "loss": 0.4645,
      "step": 102
    },
    {
      "epoch": 1.124574540503744,
      "grad_norm": 1.9589727338791063,
      "learning_rate": 6.288466640488679e-05,
      "loss": 0.4574,
      "step": 103
    },
    {
      "epoch": 1.1354663036078965,
      "grad_norm": 1.250427733328813,
      "learning_rate": 6.2462119472015e-05,
      "loss": 0.453,
      "step": 104
    },
    {
      "epoch": 1.146358066712049,
      "grad_norm": 1.244933711184395,
      "learning_rate": 6.20358792580841e-05,
      "loss": 0.4562,
      "step": 105
    },
    {
      "epoch": 1.1572498298162015,
      "grad_norm": 0.978598635089721,
      "learning_rate": 6.160601584663681e-05,
      "loss": 0.4544,
      "step": 106
    },
    {
      "epoch": 1.168141592920354,
      "grad_norm": 1.0133074820438104,
      "learning_rate": 6.11725999169515e-05,
      "loss": 0.4567,
      "step": 107
    },
    {
      "epoch": 1.1790333560245065,
      "grad_norm": 0.9079803477398684,
      "learning_rate": 6.0735702732421015e-05,
      "loss": 0.4537,
      "step": 108
    },
    {
      "epoch": 1.189925119128659,
      "grad_norm": 0.47916750157430577,
      "learning_rate": 6.029539612883529e-05,
      "loss": 0.4514,
      "step": 109
    },
    {
      "epoch": 1.2008168822328114,
      "grad_norm": 0.737478359754882,
      "learning_rate": 5.9851752502570015e-05,
      "loss": 0.4462,
      "step": 110
    },
    {
      "epoch": 1.2117086453369639,
      "grad_norm": 0.5899481616784151,
      "learning_rate": 5.940484479868288e-05,
      "loss": 0.4456,
      "step": 111
    },
    {
      "epoch": 1.2226004084411164,
      "grad_norm": 0.766606804098348,
      "learning_rate": 5.895474649891995e-05,
      "loss": 0.4441,
      "step": 112
    },
    {
      "epoch": 1.233492171545269,
      "grad_norm": 0.6645207203467242,
      "learning_rate": 5.8501531609633424e-05,
      "loss": 0.4463,
      "step": 113
    },
    {
      "epoch": 1.2443839346494214,
      "grad_norm": 0.681340548554224,
      "learning_rate": 5.8045274649613386e-05,
      "loss": 0.4424,
      "step": 114
    },
    {
      "epoch": 1.2552756977535737,
      "grad_norm": 0.7119900691253739,
      "learning_rate": 5.7586050637835295e-05,
      "loss": 0.4459,
      "step": 115
    },
    {
      "epoch": 1.2661674608577265,
      "grad_norm": 0.49479084006465096,
      "learning_rate": 5.7123935081125034e-05,
      "loss": 0.4408,
      "step": 116
    },
    {
      "epoch": 1.2770592239618788,
      "grad_norm": 0.7116517379970301,
      "learning_rate": 5.6659003961743965e-05,
      "loss": 0.4455,
      "step": 117
    },
    {
      "epoch": 1.2879509870660313,
      "grad_norm": 0.6527860650845239,
      "learning_rate": 5.619133372489575e-05,
      "loss": 0.4362,
      "step": 118
    },
    {
      "epoch": 1.2988427501701838,
      "grad_norm": 0.5221316831765591,
      "learning_rate": 5.572100126615695e-05,
      "loss": 0.4483,
      "step": 119
    },
    {
      "epoch": 1.3097345132743363,
      "grad_norm": 0.36665207601249333,
      "learning_rate": 5.524808391883367e-05,
      "loss": 0.4488,
      "step": 120
    },
    {
      "epoch": 1.3206262763784888,
      "grad_norm": 0.41335991497113145,
      "learning_rate": 5.477265944124626e-05,
      "loss": 0.4466,
      "step": 121
    },
    {
      "epoch": 1.3315180394826411,
      "grad_norm": 0.4288901779920896,
      "learning_rate": 5.429480600394405e-05,
      "loss": 0.4455,
      "step": 122
    },
    {
      "epoch": 1.3424098025867937,
      "grad_norm": 0.4476404528186165,
      "learning_rate": 5.381460217685231e-05,
      "loss": 0.4409,
      "step": 123
    },
    {
      "epoch": 1.3533015656909462,
      "grad_norm": 0.3975466345144197,
      "learning_rate": 5.333212691635368e-05,
      "loss": 0.4368,
      "step": 124
    },
    {
      "epoch": 1.3641933287950987,
      "grad_norm": 0.3833210592542072,
      "learning_rate": 5.2847459552305834e-05,
      "loss": 0.4414,
      "step": 125
    },
    {
      "epoch": 1.3750850918992512,
      "grad_norm": 0.38028400797503575,
      "learning_rate": 5.23606797749979e-05,
      "loss": 0.4342,
      "step": 126
    },
    {
      "epoch": 1.3859768550034037,
      "grad_norm": 0.29336709544411704,
      "learning_rate": 5.1871867622047624e-05,
      "loss": 0.4349,
      "step": 127
    },
    {
      "epoch": 1.3968686181075562,
      "grad_norm": 0.4021055378411946,
      "learning_rate": 5.13811034652413e-05,
      "loss": 0.4359,
      "step": 128
    },
    {
      "epoch": 1.4077603812117085,
      "grad_norm": 0.36564910628203584,
      "learning_rate": 5.088846799731885e-05,
      "loss": 0.4449,
      "step": 129
    },
    {
      "epoch": 1.418652144315861,
      "grad_norm": 0.3687189406492797,
      "learning_rate": 5.039404221870612e-05,
      "loss": 0.4384,
      "step": 130
    },
    {
      "epoch": 1.4295439074200136,
      "grad_norm": 0.29150702350417407,
      "learning_rate": 4.989790742419658e-05,
      "loss": 0.4391,
      "step": 131
    },
    {
      "epoch": 1.440435670524166,
      "grad_norm": 0.28941702462294044,
      "learning_rate": 4.940014518958461e-05,
      "loss": 0.435,
      "step": 132
    },
    {
      "epoch": 1.4513274336283186,
      "grad_norm": 0.3046809035055879,
      "learning_rate": 4.890083735825258e-05,
      "loss": 0.4423,
      "step": 133
    },
    {
      "epoch": 1.4622191967324711,
      "grad_norm": 0.3901636771063769,
      "learning_rate": 4.8400066027713974e-05,
      "loss": 0.4373,
      "step": 134
    },
    {
      "epoch": 1.4731109598366237,
      "grad_norm": 0.2975253048975319,
      "learning_rate": 4.789791353611469e-05,
      "loss": 0.441,
      "step": 135
    },
    {
      "epoch": 1.484002722940776,
      "grad_norm": 0.2861058200191958,
      "learning_rate": 4.7394462448694756e-05,
      "loss": 0.4392,
      "step": 136
    },
    {
      "epoch": 1.4948944860449285,
      "grad_norm": 0.30291271659415053,
      "learning_rate": 4.688979554421276e-05,
      "loss": 0.4437,
      "step": 137
    },
    {
      "epoch": 1.505786249149081,
      "grad_norm": 0.27992103529188517,
      "learning_rate": 4.6383995801335176e-05,
      "loss": 0.4448,
      "step": 138
    },
    {
      "epoch": 1.5166780122532335,
      "grad_norm": 0.25747738933655456,
      "learning_rate": 4.5877146384992725e-05,
      "loss": 0.4431,
      "step": 139
    },
    {
      "epoch": 1.527569775357386,
      "grad_norm": 0.341071325514537,
      "learning_rate": 4.5369330632706223e-05,
      "loss": 0.4331,
      "step": 140
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.2602954860252923,
      "learning_rate": 4.486063204088402e-05,
      "loss": 0.4433,
      "step": 141
    },
    {
      "epoch": 1.549353301565691,
      "grad_norm": 0.23767705409963544,
      "learning_rate": 4.435113425109324e-05,
      "loss": 0.4337,
      "step": 142
    },
    {
      "epoch": 1.5602450646698434,
      "grad_norm": 0.29322114556856593,
      "learning_rate": 4.3840921036307274e-05,
      "loss": 0.4387,
      "step": 143
    },
    {
      "epoch": 1.5711368277739959,
      "grad_norm": 0.2171415398561475,
      "learning_rate": 4.333007628713158e-05,
      "loss": 0.4333,
      "step": 144
    },
    {
      "epoch": 1.5820285908781484,
      "grad_norm": 0.26753092128368694,
      "learning_rate": 4.281868399801016e-05,
      "loss": 0.4375,
      "step": 145
    },
    {
      "epoch": 1.592920353982301,
      "grad_norm": 0.2577848543389177,
      "learning_rate": 4.230682825341498e-05,
      "loss": 0.4331,
      "step": 146
    },
    {
      "epoch": 1.6038121170864534,
      "grad_norm": 0.23037795102518452,
      "learning_rate": 4.17945932140206e-05,
      "loss": 0.4432,
      "step": 147
    },
    {
      "epoch": 1.6147038801906057,
      "grad_norm": 0.22432493774064247,
      "learning_rate": 4.128206310286622e-05,
      "loss": 0.433,
      "step": 148
    },
    {
      "epoch": 1.6255956432947585,
      "grad_norm": 0.27660950965609943,
      "learning_rate": 4.0769322191507485e-05,
      "loss": 0.4291,
      "step": 149
    },
    {
      "epoch": 1.6364874063989108,
      "grad_norm": 0.29639803872823944,
      "learning_rate": 4.025645478616045e-05,
      "loss": 0.4479,
      "step": 150
    },
    {
      "epoch": 1.6473791695030633,
      "grad_norm": 0.22551842554518517,
      "learning_rate": 3.974354521383956e-05,
      "loss": 0.4305,
      "step": 151
    },
    {
      "epoch": 1.6582709326072158,
      "grad_norm": 0.21855552569143427,
      "learning_rate": 3.923067780849252e-05,
      "loss": 0.4341,
      "step": 152
    },
    {
      "epoch": 1.6691626957113683,
      "grad_norm": 0.25345471682479176,
      "learning_rate": 3.87179368971338e-05,
      "loss": 0.4332,
      "step": 153
    },
    {
      "epoch": 1.6800544588155208,
      "grad_norm": 0.3294043510791402,
      "learning_rate": 3.820540678597942e-05,
      "loss": 0.4324,
      "step": 154
    },
    {
      "epoch": 1.6909462219196731,
      "grad_norm": 0.2915052857817396,
      "learning_rate": 3.769317174658503e-05,
      "loss": 0.4289,
      "step": 155
    },
    {
      "epoch": 1.7018379850238259,
      "grad_norm": 0.20123186425930717,
      "learning_rate": 3.718131600198984e-05,
      "loss": 0.4414,
      "step": 156
    },
    {
      "epoch": 1.7127297481279782,
      "grad_norm": 0.23019840765145663,
      "learning_rate": 3.666992371286843e-05,
      "loss": 0.4341,
      "step": 157
    },
    {
      "epoch": 1.7236215112321307,
      "grad_norm": 0.2075296642104422,
      "learning_rate": 3.615907896369273e-05,
      "loss": 0.4391,
      "step": 158
    },
    {
      "epoch": 1.7345132743362832,
      "grad_norm": 0.20499145565748494,
      "learning_rate": 3.564886574890677e-05,
      "loss": 0.4368,
      "step": 159
    },
    {
      "epoch": 1.7454050374404355,
      "grad_norm": 0.2270936672913207,
      "learning_rate": 3.5139367959115986e-05,
      "loss": 0.4362,
      "step": 160
    },
    {
      "epoch": 1.7562968005445883,
      "grad_norm": 0.20695924414455472,
      "learning_rate": 3.4630669367293797e-05,
      "loss": 0.4321,
      "step": 161
    },
    {
      "epoch": 1.7671885636487406,
      "grad_norm": 0.20145634393976244,
      "learning_rate": 3.412285361500729e-05,
      "loss": 0.4327,
      "step": 162
    },
    {
      "epoch": 1.778080326752893,
      "grad_norm": 0.1954369301373391,
      "learning_rate": 3.3616004198664845e-05,
      "loss": 0.4324,
      "step": 163
    },
    {
      "epoch": 1.7889720898570456,
      "grad_norm": 0.20481400956545093,
      "learning_rate": 3.311020445578725e-05,
      "loss": 0.4377,
      "step": 164
    },
    {
      "epoch": 1.799863852961198,
      "grad_norm": 0.2162909320824147,
      "learning_rate": 3.260553755130525e-05,
      "loss": 0.4299,
      "step": 165
    },
    {
      "epoch": 1.8107556160653506,
      "grad_norm": 0.1811829339048361,
      "learning_rate": 3.210208646388532e-05,
      "loss": 0.43,
      "step": 166
    },
    {
      "epoch": 1.821647379169503,
      "grad_norm": 0.2086574503584923,
      "learning_rate": 3.1599933972286026e-05,
      "loss": 0.4346,
      "step": 167
    },
    {
      "epoch": 1.8325391422736557,
      "grad_norm": 0.22526815431719788,
      "learning_rate": 3.109916264174743e-05,
      "loss": 0.4288,
      "step": 168
    },
    {
      "epoch": 1.843430905377808,
      "grad_norm": 0.18015782113002662,
      "learning_rate": 3.0599854810415393e-05,
      "loss": 0.428,
      "step": 169
    },
    {
      "epoch": 1.8543226684819605,
      "grad_norm": 0.20990114701229817,
      "learning_rate": 3.0102092575803435e-05,
      "loss": 0.4333,
      "step": 170
    },
    {
      "epoch": 1.865214431586113,
      "grad_norm": 0.1988483391924283,
      "learning_rate": 2.9605957781293893e-05,
      "loss": 0.4346,
      "step": 171
    },
    {
      "epoch": 1.8761061946902655,
      "grad_norm": 0.19078378305907726,
      "learning_rate": 2.911153200268116e-05,
      "loss": 0.4251,
      "step": 172
    },
    {
      "epoch": 1.886997957794418,
      "grad_norm": 0.23446180498320415,
      "learning_rate": 2.8618896534758707e-05,
      "loss": 0.4377,
      "step": 173
    },
    {
      "epoch": 1.8978897208985703,
      "grad_norm": 0.1761870449513099,
      "learning_rate": 2.8128132377952376e-05,
      "loss": 0.4227,
      "step": 174
    },
    {
      "epoch": 1.908781484002723,
      "grad_norm": 0.2332557884998122,
      "learning_rate": 2.7639320225002108e-05,
      "loss": 0.4325,
      "step": 175
    },
    {
      "epoch": 1.9196732471068754,
      "grad_norm": 0.2211426717672741,
      "learning_rate": 2.715254044769418e-05,
      "loss": 0.4308,
      "step": 176
    },
    {
      "epoch": 1.9305650102110279,
      "grad_norm": 0.18328106510038045,
      "learning_rate": 2.666787308364634e-05,
      "loss": 0.4322,
      "step": 177
    },
    {
      "epoch": 1.9414567733151804,
      "grad_norm": 0.2288520554831422,
      "learning_rate": 2.6185397823147703e-05,
      "loss": 0.4263,
      "step": 178
    },
    {
      "epoch": 1.952348536419333,
      "grad_norm": 0.17271019835077814,
      "learning_rate": 2.5705193996055977e-05,
      "loss": 0.4231,
      "step": 179
    },
    {
      "epoch": 1.9632402995234854,
      "grad_norm": 0.21114296576465388,
      "learning_rate": 2.5227340558753755e-05,
      "loss": 0.4323,
      "step": 180
    },
    {
      "epoch": 1.9741320626276377,
      "grad_norm": 0.1708320475165614,
      "learning_rate": 2.4751916081166336e-05,
      "loss": 0.4318,
      "step": 181
    },
    {
      "epoch": 1.9850238257317905,
      "grad_norm": 0.20243633225986957,
      "learning_rate": 2.427899873384306e-05,
      "loss": 0.4284,
      "step": 182
    },
    {
      "epoch": 1.9959155888359428,
      "grad_norm": 0.17214298347005028,
      "learning_rate": 2.3808666275104248e-05,
      "loss": 0.4303,
      "step": 183
    },
    {
      "epoch": 2.0095302927161334,
      "grad_norm": 0.2371948537766053,
      "learning_rate": 2.334099603825605e-05,
      "loss": 0.4195,
      "step": 184
    },
    {
      "epoch": 2.020422055820286,
      "grad_norm": 0.20683435017217672,
      "learning_rate": 2.2876064918874993e-05,
      "loss": 0.4056,
      "step": 185
    },
    {
      "epoch": 2.0313138189244384,
      "grad_norm": 0.2132114122172093,
      "learning_rate": 2.241394936216472e-05,
      "loss": 0.4033,
      "step": 186
    },
    {
      "epoch": 2.0422055820285907,
      "grad_norm": 0.19087121395561507,
      "learning_rate": 2.1954725350386614e-05,
      "loss": 0.405,
      "step": 187
    },
    {
      "epoch": 2.0530973451327434,
      "grad_norm": 0.21040243392909405,
      "learning_rate": 2.14984683903666e-05,
      "loss": 0.4094,
      "step": 188
    },
    {
      "epoch": 2.0639891082368957,
      "grad_norm": 0.16468816702883052,
      "learning_rate": 2.1045253501080058e-05,
      "loss": 0.3988,
      "step": 189
    },
    {
      "epoch": 2.0748808713410485,
      "grad_norm": 0.17523456140653074,
      "learning_rate": 2.0595155201317115e-05,
      "loss": 0.4009,
      "step": 190
    },
    {
      "epoch": 2.085772634445201,
      "grad_norm": 0.19773800473297834,
      "learning_rate": 2.0148247497430012e-05,
      "loss": 0.4016,
      "step": 191
    },
    {
      "epoch": 2.096664397549353,
      "grad_norm": 0.1647636735264003,
      "learning_rate": 1.970460387116472e-05,
      "loss": 0.3948,
      "step": 192
    },
    {
      "epoch": 2.107556160653506,
      "grad_norm": 0.18353188744441096,
      "learning_rate": 1.9264297267579e-05,
      "loss": 0.3995,
      "step": 193
    },
    {
      "epoch": 2.118447923757658,
      "grad_norm": 0.1804931450578559,
      "learning_rate": 1.8827400083048503e-05,
      "loss": 0.404,
      "step": 194
    },
    {
      "epoch": 2.129339686861811,
      "grad_norm": 0.16921208996470288,
      "learning_rate": 1.8393984153363203e-05,
      "loss": 0.3886,
      "step": 195
    },
    {
      "epoch": 2.140231449965963,
      "grad_norm": 0.15129405097453685,
      "learning_rate": 1.7964120741915905e-05,
      "loss": 0.4082,
      "step": 196
    },
    {
      "epoch": 2.151123213070116,
      "grad_norm": 0.17224554916641183,
      "learning_rate": 1.753788052798501e-05,
      "loss": 0.3994,
      "step": 197
    },
    {
      "epoch": 2.162014976174268,
      "grad_norm": 0.15770956818834928,
      "learning_rate": 1.7115333595113225e-05,
      "loss": 0.403,
      "step": 198
    },
    {
      "epoch": 2.172906739278421,
      "grad_norm": 0.1716039599905281,
      "learning_rate": 1.669654941958416e-05,
      "loss": 0.4032,
      "step": 199
    },
    {
      "epoch": 2.1837985023825732,
      "grad_norm": 0.16454710212660165,
      "learning_rate": 1.628159685899897e-05,
      "loss": 0.405,
      "step": 200
    },
    {
      "epoch": 2.1946902654867255,
      "grad_norm": 0.13993922901377312,
      "learning_rate": 1.5870544140954543e-05,
      "loss": 0.4077,
      "step": 201
    },
    {
      "epoch": 2.2055820285908783,
      "grad_norm": 0.16301323668704157,
      "learning_rate": 1.5463458851825345e-05,
      "loss": 0.4053,
      "step": 202
    },
    {
      "epoch": 2.2164737916950306,
      "grad_norm": 0.15989174020128882,
      "learning_rate": 1.5060407925650662e-05,
      "loss": 0.4036,
      "step": 203
    },
    {
      "epoch": 2.2273655547991833,
      "grad_norm": 0.14155165998806302,
      "learning_rate": 1.466145763312922e-05,
      "loss": 0.4021,
      "step": 204
    },
    {
      "epoch": 2.2382573179033356,
      "grad_norm": 0.13197910914100647,
      "learning_rate": 1.426667357072265e-05,
      "loss": 0.403,
      "step": 205
    },
    {
      "epoch": 2.249149081007488,
      "grad_norm": 0.1502501179431283,
      "learning_rate": 1.3876120649870051e-05,
      "loss": 0.4068,
      "step": 206
    },
    {
      "epoch": 2.2600408441116406,
      "grad_norm": 0.12632548576221497,
      "learning_rate": 1.3489863086315085e-05,
      "loss": 0.3954,
      "step": 207
    },
    {
      "epoch": 2.270932607215793,
      "grad_norm": 0.12310652443499624,
      "learning_rate": 1.3107964389547326e-05,
      "loss": 0.4056,
      "step": 208
    },
    {
      "epoch": 2.2818243703199457,
      "grad_norm": 0.13232393309386553,
      "learning_rate": 1.2730487352360026e-05,
      "loss": 0.4028,
      "step": 209
    },
    {
      "epoch": 2.292716133424098,
      "grad_norm": 0.12337150415039137,
      "learning_rate": 1.2357494040525416e-05,
      "loss": 0.4125,
      "step": 210
    },
    {
      "epoch": 2.3036078965282503,
      "grad_norm": 0.11483009808175632,
      "learning_rate": 1.1989045782589815e-05,
      "loss": 0.404,
      "step": 211
    },
    {
      "epoch": 2.314499659632403,
      "grad_norm": 0.1213039797258653,
      "learning_rate": 1.1625203159789686e-05,
      "loss": 0.398,
      "step": 212
    },
    {
      "epoch": 2.3253914227365553,
      "grad_norm": 0.1296712824505999,
      "learning_rate": 1.1266025996090902e-05,
      "loss": 0.4103,
      "step": 213
    },
    {
      "epoch": 2.336283185840708,
      "grad_norm": 0.12739388021409695,
      "learning_rate": 1.0911573348352107e-05,
      "loss": 0.4054,
      "step": 214
    },
    {
      "epoch": 2.3471749489448603,
      "grad_norm": 0.11766621480801555,
      "learning_rate": 1.0561903496614603e-05,
      "loss": 0.4036,
      "step": 215
    },
    {
      "epoch": 2.358066712049013,
      "grad_norm": 0.12339813858079209,
      "learning_rate": 1.0217073934519726e-05,
      "loss": 0.4012,
      "step": 216
    },
    {
      "epoch": 2.3689584751531654,
      "grad_norm": 0.11647805921713034,
      "learning_rate": 9.877141359855567e-06,
      "loss": 0.4109,
      "step": 217
    },
    {
      "epoch": 2.379850238257318,
      "grad_norm": 0.109735463051235,
      "learning_rate": 9.542161665234623e-06,
      "loss": 0.3991,
      "step": 218
    },
    {
      "epoch": 2.3907420013614704,
      "grad_norm": 0.1181837766256319,
      "learning_rate": 9.212189928903758e-06,
      "loss": 0.4008,
      "step": 219
    },
    {
      "epoch": 2.4016337644656227,
      "grad_norm": 0.12247494016760661,
      "learning_rate": 8.887280405688106e-06,
      "loss": 0.4044,
      "step": 220
    },
    {
      "epoch": 2.4125255275697755,
      "grad_norm": 0.11192215755173132,
      "learning_rate": 8.567486518070306e-06,
      "loss": 0.3992,
      "step": 221
    },
    {
      "epoch": 2.4234172906739277,
      "grad_norm": 0.109297303198497,
      "learning_rate": 8.252860847406712e-06,
      "loss": 0.3948,
      "step": 222
    },
    {
      "epoch": 2.4343090537780805,
      "grad_norm": 0.11284991227332902,
      "learning_rate": 7.943455125281741e-06,
      "loss": 0.3995,
      "step": 223
    },
    {
      "epoch": 2.445200816882233,
      "grad_norm": 0.11548250618562617,
      "learning_rate": 7.639320225002106e-06,
      "loss": 0.407,
      "step": 224
    },
    {
      "epoch": 2.456092579986385,
      "grad_norm": 0.116001123064691,
      "learning_rate": 7.340506153232052e-06,
      "loss": 0.4,
      "step": 225
    },
    {
      "epoch": 2.466984343090538,
      "grad_norm": 0.10992561869419722,
      "learning_rate": 7.047062041771133e-06,
      "loss": 0.4045,
      "step": 226
    },
    {
      "epoch": 2.47787610619469,
      "grad_norm": 0.10955112339301888,
      "learning_rate": 6.759036139475843e-06,
      "loss": 0.3964,
      "step": 227
    },
    {
      "epoch": 2.488767869298843,
      "grad_norm": 0.11910747258190017,
      "learning_rate": 6.476475804326377e-06,
      "loss": 0.4025,
      "step": 228
    },
    {
      "epoch": 2.499659632402995,
      "grad_norm": 0.116651055560925,
      "learning_rate": 6.199427495639963e-06,
      "loss": 0.3994,
      "step": 229
    },
    {
      "epoch": 2.5105513955071475,
      "grad_norm": 0.1138861345434715,
      "learning_rate": 5.927936766431836e-06,
      "loss": 0.4022,
      "step": 230
    },
    {
      "epoch": 2.5214431586113,
      "grad_norm": 0.09848824385777599,
      "learning_rate": 5.662048255925357e-06,
      "loss": 0.3988,
      "step": 231
    },
    {
      "epoch": 2.532334921715453,
      "grad_norm": 0.10150492531984864,
      "learning_rate": 5.40180568221226e-06,
      "loss": 0.4047,
      "step": 232
    },
    {
      "epoch": 2.5432266848196052,
      "grad_norm": 0.11056977593577616,
      "learning_rate": 5.147251835064424e-06,
      "loss": 0.4057,
      "step": 233
    },
    {
      "epoch": 2.5541184479237575,
      "grad_norm": 0.10655441819972683,
      "learning_rate": 4.898428568898288e-06,
      "loss": 0.3971,
      "step": 234
    },
    {
      "epoch": 2.5650102110279103,
      "grad_norm": 0.10580636847333873,
      "learning_rate": 4.65537679589299e-06,
      "loss": 0.4045,
      "step": 235
    },
    {
      "epoch": 2.5759019741320626,
      "grad_norm": 0.09908356652521799,
      "learning_rate": 4.418136479263533e-06,
      "loss": 0.4004,
      "step": 236
    },
    {
      "epoch": 2.5867937372362153,
      "grad_norm": 0.09287282151185479,
      "learning_rate": 4.186746626689879e-06,
      "loss": 0.4006,
      "step": 237
    },
    {
      "epoch": 2.5976855003403676,
      "grad_norm": 0.11618401418689138,
      "learning_rate": 3.961245283903239e-06,
      "loss": 0.4025,
      "step": 238
    },
    {
      "epoch": 2.60857726344452,
      "grad_norm": 0.09813772046749265,
      "learning_rate": 3.7416695284304737e-06,
      "loss": 0.4051,
      "step": 239
    },
    {
      "epoch": 2.6194690265486726,
      "grad_norm": 0.08852232603950476,
      "learning_rate": 3.5280554634977217e-06,
      "loss": 0.4013,
      "step": 240
    },
    {
      "epoch": 2.630360789652825,
      "grad_norm": 0.09394838490800748,
      "learning_rate": 3.320438212094197e-06,
      "loss": 0.4059,
      "step": 241
    },
    {
      "epoch": 2.6412525527569777,
      "grad_norm": 0.09059231002024923,
      "learning_rate": 3.1188519111971804e-06,
      "loss": 0.4039,
      "step": 242
    },
    {
      "epoch": 2.65214431586113,
      "grad_norm": 0.09411366384619929,
      "learning_rate": 2.9233297061591346e-06,
      "loss": 0.3973,
      "step": 243
    },
    {
      "epoch": 2.6630360789652823,
      "grad_norm": 0.09939434350976686,
      "learning_rate": 2.733903745257838e-06,
      "loss": 0.4049,
      "step": 244
    },
    {
      "epoch": 2.673927842069435,
      "grad_norm": 0.09359826061681908,
      "learning_rate": 2.550605174410512e-06,
      "loss": 0.4017,
      "step": 245
    },
    {
      "epoch": 2.6848196051735873,
      "grad_norm": 0.09354919223068338,
      "learning_rate": 2.373464132052701e-06,
      "loss": 0.4115,
      "step": 246
    },
    {
      "epoch": 2.69571136827774,
      "grad_norm": 0.09746026612098238,
      "learning_rate": 2.202509744182835e-06,
      "loss": 0.3982,
      "step": 247
    },
    {
      "epoch": 2.7066031313818923,
      "grad_norm": 0.09168148902821703,
      "learning_rate": 2.0377701195732545e-06,
      "loss": 0.4071,
      "step": 248
    },
    {
      "epoch": 2.717494894486045,
      "grad_norm": 0.09074247743593482,
      "learning_rate": 1.879272345148513e-06,
      "loss": 0.4101,
      "step": 249
    },
    {
      "epoch": 2.7283866575901974,
      "grad_norm": 0.09117996023348725,
      "learning_rate": 1.727042481531651e-06,
      "loss": 0.4035,
      "step": 250
    },
    {
      "epoch": 2.73927842069435,
      "grad_norm": 0.09180074130441847,
      "learning_rate": 1.5811055587592283e-06,
      "loss": 0.4009,
      "step": 251
    },
    {
      "epoch": 2.7501701837985024,
      "grad_norm": 0.08657516729277243,
      "learning_rate": 1.4414855721658705e-06,
      "loss": 0.4041,
      "step": 252
    },
    {
      "epoch": 2.7610619469026547,
      "grad_norm": 0.09146051585946402,
      "learning_rate": 1.3082054784388221e-06,
      "loss": 0.4055,
      "step": 253
    },
    {
      "epoch": 2.7719537100068075,
      "grad_norm": 0.09305195530663764,
      "learning_rate": 1.1812871918434143e-06,
      "loss": 0.4027,
      "step": 254
    },
    {
      "epoch": 2.7828454731109598,
      "grad_norm": 0.09703421946821408,
      "learning_rate": 1.0607515806198142e-06,
      "loss": 0.4079,
      "step": 255
    },
    {
      "epoch": 2.7937372362151125,
      "grad_norm": 0.08906174726324725,
      "learning_rate": 9.466184635518361e-07,
      "loss": 0.4017,
      "step": 256
    },
    {
      "epoch": 2.804628999319265,
      "grad_norm": 0.08574571986172967,
      "learning_rate": 8.389066067082852e-07,
      "loss": 0.3967,
      "step": 257
    },
    {
      "epoch": 2.815520762423417,
      "grad_norm": 0.0880676756279348,
      "learning_rate": 7.376337203573824e-07,
      "loss": 0.406,
      "step": 258
    },
    {
      "epoch": 2.82641252552757,
      "grad_norm": 0.08789941779545143,
      "learning_rate": 6.428164560548134e-07,
      "loss": 0.4076,
      "step": 259
    },
    {
      "epoch": 2.837304288631722,
      "grad_norm": 0.08553769033856681,
      "learning_rate": 5.544704039058025e-07,
      "loss": 0.3983,
      "step": 260
    },
    {
      "epoch": 2.848196051735875,
      "grad_norm": 0.08010149417951533,
      "learning_rate": 4.7261009000177274e-07,
      "loss": 0.3918,
      "step": 261
    },
    {
      "epoch": 2.859087814840027,
      "grad_norm": 0.07962434672916259,
      "learning_rate": 3.972489740319274e-07,
      "loss": 0.3924,
      "step": 262
    },
    {
      "epoch": 2.8699795779441795,
      "grad_norm": 0.08482433839475353,
      "learning_rate": 3.283994470701579e-07,
      "loss": 0.4057,
      "step": 263
    },
    {
      "epoch": 2.880871341048332,
      "grad_norm": 0.08350491927147591,
      "learning_rate": 2.66072829537678e-07,
      "loss": 0.4099,
      "step": 264
    },
    {
      "epoch": 2.891763104152485,
      "grad_norm": 0.0880658902471283,
      "learning_rate": 2.102793693417038e-07,
      "loss": 0.4073,
      "step": 265
    },
    {
      "epoch": 2.9026548672566372,
      "grad_norm": 0.08328293731348216,
      "learning_rate": 1.6102824019043728e-07,
      "loss": 0.4002,
      "step": 266
    },
    {
      "epoch": 2.9135466303607895,
      "grad_norm": 0.07841175895615185,
      "learning_rate": 1.1832754008472614e-07,
      "loss": 0.389,
      "step": 267
    },
    {
      "epoch": 2.9244383934649423,
      "grad_norm": 0.08287793053425582,
      "learning_rate": 8.21842899865466e-08,
      "loss": 0.3955,
      "step": 268
    },
    {
      "epoch": 2.9353301565690946,
      "grad_norm": 0.08049459789709469,
      "learning_rate": 5.260443266462467e-08,
      "loss": 0.3949,
      "step": 269
    },
    {
      "epoch": 2.9462219196732473,
      "grad_norm": 0.086710676105483,
      "learning_rate": 2.9592831717293326e-08,
      "loss": 0.4072,
      "step": 270
    },
    {
      "epoch": 2.9571136827773996,
      "grad_norm": 0.08154316295610879,
      "learning_rate": 1.3153270772807702e-08,
      "loss": 0.3967,
      "step": 271
    },
    {
      "epoch": 2.968005445881552,
      "grad_norm": 0.08705352570078856,
      "learning_rate": 3.2884528672294523e-09,
      "loss": 0.395,
      "step": 272
    },
    {
      "epoch": 2.9788972089857046,
      "grad_norm": 0.08297445165644558,
      "learning_rate": 0.0,
      "loss": 0.4097,
      "step": 273
    },
    {
      "epoch": 2.9788972089857046,
      "step": 273,
      "total_flos": 7.02510214541004e+18,
      "train_loss": 0.464617241979082,
      "train_runtime": 57723.5744,
      "train_samples_per_second": 2.443,
      "train_steps_per_second": 0.005
    }
  ],
  "logging_steps": 1,
  "max_steps": 273,
  "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": 7.02510214541004e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}