{
  "best_metric": 0.6790642738342285,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.013402355463972794,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.701177731986397e-05,
      "grad_norm": 8.344467163085938,
      "learning_rate": 1e-05,
      "loss": 2.5618,
      "step": 1
    },
    {
      "epoch": 6.701177731986397e-05,
      "eval_loss": 0.8288924694061279,
      "eval_runtime": 1982.1937,
      "eval_samples_per_second": 12.68,
      "eval_steps_per_second": 3.17,
      "step": 1
    },
    {
      "epoch": 0.00013402355463972793,
      "grad_norm": 7.046421527862549,
      "learning_rate": 2e-05,
      "loss": 2.107,
      "step": 2
    },
    {
      "epoch": 0.0002010353319595919,
      "grad_norm": 7.221210479736328,
      "learning_rate": 3e-05,
      "loss": 2.7472,
      "step": 3
    },
    {
      "epoch": 0.00026804710927945586,
      "grad_norm": 5.236713409423828,
      "learning_rate": 4e-05,
      "loss": 2.1887,
      "step": 4
    },
    {
      "epoch": 0.00033505888659931984,
      "grad_norm": 5.94106388092041,
      "learning_rate": 5e-05,
      "loss": 2.5344,
      "step": 5
    },
    {
      "epoch": 0.0004020706639191838,
      "grad_norm": 5.348053932189941,
      "learning_rate": 6e-05,
      "loss": 2.7533,
      "step": 6
    },
    {
      "epoch": 0.00046908244123904775,
      "grad_norm": 4.864931106567383,
      "learning_rate": 7e-05,
      "loss": 2.4738,
      "step": 7
    },
    {
      "epoch": 0.0005360942185589117,
      "grad_norm": 4.8689398765563965,
      "learning_rate": 8e-05,
      "loss": 2.6612,
      "step": 8
    },
    {
      "epoch": 0.0006031059958787757,
      "grad_norm": 4.846255302429199,
      "learning_rate": 9e-05,
      "loss": 2.3118,
      "step": 9
    },
    {
      "epoch": 0.0006701177731986397,
      "grad_norm": 6.144360542297363,
      "learning_rate": 0.0001,
      "loss": 2.492,
      "step": 10
    },
    {
      "epoch": 0.0007371295505185036,
      "grad_norm": 7.286844253540039,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.6034,
      "step": 11
    },
    {
      "epoch": 0.0008041413278383676,
      "grad_norm": 4.436187744140625,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.6801,
      "step": 12
    },
    {
      "epoch": 0.0008711531051582316,
      "grad_norm": 4.1784210205078125,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.6066,
      "step": 13
    },
    {
      "epoch": 0.0009381648824780955,
      "grad_norm": 5.016221523284912,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.8911,
      "step": 14
    },
    {
      "epoch": 0.0010051766597979595,
      "grad_norm": 4.402395725250244,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.751,
      "step": 15
    },
    {
      "epoch": 0.0010721884371178235,
      "grad_norm": 133.68438720703125,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.2188,
      "step": 16
    },
    {
      "epoch": 0.0011392002144376874,
      "grad_norm": 5.27332878112793,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.646,
      "step": 17
    },
    {
      "epoch": 0.0012062119917575513,
      "grad_norm": 4.458730697631836,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.4707,
      "step": 18
    },
    {
      "epoch": 0.0012732237690774154,
      "grad_norm": 4.768811225891113,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.7653,
      "step": 19
    },
    {
      "epoch": 0.0013402355463972794,
      "grad_norm": 5.060022830963135,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.9524,
      "step": 20
    },
    {
      "epoch": 0.0014072473237171433,
      "grad_norm": 4.475470066070557,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.5964,
      "step": 21
    },
    {
      "epoch": 0.0014742591010370072,
      "grad_norm": 4.7729058265686035,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.4069,
      "step": 22
    },
    {
      "epoch": 0.0015412708783568711,
      "grad_norm": 5.6212849617004395,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.7651,
      "step": 23
    },
    {
      "epoch": 0.0016082826556767353,
      "grad_norm": 5.773212909698486,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.7976,
      "step": 24
    },
    {
      "epoch": 0.0016752944329965992,
      "grad_norm": 6.008873462677002,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.995,
      "step": 25
    },
    {
      "epoch": 0.0017423062103164631,
      "grad_norm": 5.016881465911865,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.9344,
      "step": 26
    },
    {
      "epoch": 0.001809317987636327,
      "grad_norm": 5.705882549285889,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.0932,
      "step": 27
    },
    {
      "epoch": 0.001876329764956191,
      "grad_norm": 5.702162742614746,
      "learning_rate": 9.780178907671789e-05,
      "loss": 3.0524,
      "step": 28
    },
    {
      "epoch": 0.001943341542276055,
      "grad_norm": 5.03911018371582,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.6615,
      "step": 29
    },
    {
      "epoch": 0.002010353319595919,
      "grad_norm": 5.445006847381592,
      "learning_rate": 9.729086208503174e-05,
      "loss": 3.0964,
      "step": 30
    },
    {
      "epoch": 0.002077365096915783,
      "grad_norm": 4.916455268859863,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.771,
      "step": 31
    },
    {
      "epoch": 0.002144376874235647,
      "grad_norm": 5.7100510597229,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.8505,
      "step": 32
    },
    {
      "epoch": 0.002211388651555511,
      "grad_norm": 5.968597888946533,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.9487,
      "step": 33
    },
    {
      "epoch": 0.0022784004288753748,
      "grad_norm": 8.009086608886719,
      "learning_rate": 9.611448774886924e-05,
      "loss": 3.1886,
      "step": 34
    },
    {
      "epoch": 0.0023454122061952387,
      "grad_norm": 6.9888081550598145,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.8914,
      "step": 35
    },
    {
      "epoch": 0.0024124239835151026,
      "grad_norm": 7.901605606079102,
      "learning_rate": 9.545032675245813e-05,
      "loss": 3.1196,
      "step": 36
    },
    {
      "epoch": 0.0024794357608349665,
      "grad_norm": 7.282350540161133,
      "learning_rate": 9.509956150664796e-05,
      "loss": 3.626,
      "step": 37
    },
    {
      "epoch": 0.002546447538154831,
      "grad_norm": 7.514776706695557,
      "learning_rate": 9.473646649103818e-05,
      "loss": 3.5363,
      "step": 38
    },
    {
      "epoch": 0.002613459315474695,
      "grad_norm": 7.731841087341309,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.8775,
      "step": 39
    },
    {
      "epoch": 0.0026804710927945587,
      "grad_norm": 8.245630264282227,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.9931,
      "step": 40
    },
    {
      "epoch": 0.0027474828701144227,
      "grad_norm": 7.0467400550842285,
      "learning_rate": 9.357421218136386e-05,
      "loss": 3.1278,
      "step": 41
    },
    {
      "epoch": 0.0028144946474342866,
      "grad_norm": 9.853047370910645,
      "learning_rate": 9.316282404787871e-05,
      "loss": 3.2163,
      "step": 42
    },
    {
      "epoch": 0.0028815064247541505,
      "grad_norm": 8.23642635345459,
      "learning_rate": 9.273963562927695e-05,
      "loss": 3.358,
      "step": 43
    },
    {
      "epoch": 0.0029485182020740144,
      "grad_norm": 9.7107572555542,
      "learning_rate": 9.230476262104677e-05,
      "loss": 3.2123,
      "step": 44
    },
    {
      "epoch": 0.0030155299793938784,
      "grad_norm": 8.807342529296875,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.2306,
      "step": 45
    },
    {
      "epoch": 0.0030825417567137423,
      "grad_norm": 10.767407417297363,
      "learning_rate": 9.140044155740101e-05,
      "loss": 3.7151,
      "step": 46
    },
    {
      "epoch": 0.003149553534033606,
      "grad_norm": 13.164258003234863,
      "learning_rate": 9.093124073433463e-05,
      "loss": 3.5732,
      "step": 47
    },
    {
      "epoch": 0.0032165653113534706,
      "grad_norm": 13.115236282348633,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.7065,
      "step": 48
    },
    {
      "epoch": 0.0032835770886733345,
      "grad_norm": 12.329370498657227,
      "learning_rate": 8.995939984474624e-05,
      "loss": 3.5499,
      "step": 49
    },
    {
      "epoch": 0.0033505888659931984,
      "grad_norm": 31.36498260498047,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.3818,
      "step": 50
    },
    {
      "epoch": 0.0033505888659931984,
      "eval_loss": 0.7828689217567444,
      "eval_runtime": 1992.6514,
      "eval_samples_per_second": 12.613,
      "eval_steps_per_second": 3.154,
      "step": 50
    },
    {
      "epoch": 0.0034176006433130623,
      "grad_norm": 4.389657974243164,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9921,
      "step": 51
    },
    {
      "epoch": 0.0034846124206329263,
      "grad_norm": 3.457416296005249,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.0497,
      "step": 52
    },
    {
      "epoch": 0.00355162419795279,
      "grad_norm": 3.0380311012268066,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.4211,
      "step": 53
    },
    {
      "epoch": 0.003618635975272654,
      "grad_norm": 3.076496124267578,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.4153,
      "step": 54
    },
    {
      "epoch": 0.003685647752592518,
      "grad_norm": 3.4271836280822754,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.73,
      "step": 55
    },
    {
      "epoch": 0.003752659529912382,
      "grad_norm": 3.4942305088043213,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.6226,
      "step": 56
    },
    {
      "epoch": 0.003819671307232246,
      "grad_norm": 3.492516279220581,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.2824,
      "step": 57
    },
    {
      "epoch": 0.00388668308455211,
      "grad_norm": 3.6336874961853027,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.6809,
      "step": 58
    },
    {
      "epoch": 0.003953694861871974,
      "grad_norm": 2.9908275604248047,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.3927,
      "step": 59
    },
    {
      "epoch": 0.004020706639191838,
      "grad_norm": 3.518939971923828,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.4082,
      "step": 60
    },
    {
      "epoch": 0.004087718416511702,
      "grad_norm": 3.265479803085327,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.1751,
      "step": 61
    },
    {
      "epoch": 0.004154730193831566,
      "grad_norm": 3.3653993606567383,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.4896,
      "step": 62
    },
    {
      "epoch": 0.0042217419711514294,
      "grad_norm": 3.851893901824951,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.5697,
      "step": 63
    },
    {
      "epoch": 0.004288753748471294,
      "grad_norm": 3.4626893997192383,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.3114,
      "step": 64
    },
    {
      "epoch": 0.004355765525791158,
      "grad_norm": 3.7448463439941406,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.271,
      "step": 65
    },
    {
      "epoch": 0.004422777303111022,
      "grad_norm": 4.006443500518799,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.7517,
      "step": 66
    },
    {
      "epoch": 0.004489789080430886,
      "grad_norm": 3.7098660469055176,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.5202,
      "step": 67
    },
    {
      "epoch": 0.0045568008577507495,
      "grad_norm": 3.427579641342163,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3707,
      "step": 68
    },
    {
      "epoch": 0.004623812635070614,
      "grad_norm": 3.6153299808502197,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.4862,
      "step": 69
    },
    {
      "epoch": 0.004690824412390477,
      "grad_norm": 3.9386746883392334,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.7582,
      "step": 70
    },
    {
      "epoch": 0.004757836189710342,
      "grad_norm": 4.337942600250244,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.6046,
      "step": 71
    },
    {
      "epoch": 0.004824847967030205,
      "grad_norm": 4.4766764640808105,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.9247,
      "step": 72
    },
    {
      "epoch": 0.00489185974435007,
      "grad_norm": 4.415731906890869,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.7555,
      "step": 73
    },
    {
      "epoch": 0.004958871521669933,
      "grad_norm": 5.77033805847168,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.9811,
      "step": 74
    },
    {
      "epoch": 0.005025883298989797,
      "grad_norm": 4.642111778259277,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.6796,
      "step": 75
    },
    {
      "epoch": 0.005092895076309662,
      "grad_norm": 4.44441556930542,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.7302,
      "step": 76
    },
    {
      "epoch": 0.005159906853629525,
      "grad_norm": 3.9535281658172607,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.7586,
      "step": 77
    },
    {
      "epoch": 0.00522691863094939,
      "grad_norm": 4.526594638824463,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.7493,
      "step": 78
    },
    {
      "epoch": 0.005293930408269253,
      "grad_norm": 4.743769645690918,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.9715,
      "step": 79
    },
    {
      "epoch": 0.0053609421855891175,
      "grad_norm": 4.42484712600708,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.4965,
      "step": 80
    },
    {
      "epoch": 0.005427953962908981,
      "grad_norm": 4.793690204620361,
      "learning_rate": 6.932495846462261e-05,
      "loss": 3.2403,
      "step": 81
    },
    {
      "epoch": 0.005494965740228845,
      "grad_norm": 5.552221775054932,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.7791,
      "step": 82
    },
    {
      "epoch": 0.005561977517548709,
      "grad_norm": 5.081087589263916,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.7269,
      "step": 83
    },
    {
      "epoch": 0.005628989294868573,
      "grad_norm": 5.697877407073975,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.5947,
      "step": 84
    },
    {
      "epoch": 0.0056960010721884375,
      "grad_norm": 6.151732444763184,
      "learning_rate": 6.623497346023418e-05,
      "loss": 3.2169,
      "step": 85
    },
    {
      "epoch": 0.005763012849508301,
      "grad_norm": 6.099040985107422,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.58,
      "step": 86
    },
    {
      "epoch": 0.005830024626828165,
      "grad_norm": 7.048328876495361,
      "learning_rate": 6.466250186922325e-05,
      "loss": 3.2879,
      "step": 87
    },
    {
      "epoch": 0.005897036404148029,
      "grad_norm": 6.532546043395996,
      "learning_rate": 6.387014543809223e-05,
      "loss": 3.1044,
      "step": 88
    },
    {
      "epoch": 0.005964048181467893,
      "grad_norm": 6.472437381744385,
      "learning_rate": 6.307399704769099e-05,
      "loss": 3.0868,
      "step": 89
    },
    {
      "epoch": 0.006031059958787757,
      "grad_norm": 6.901464462280273,
      "learning_rate": 6.227427435703997e-05,
      "loss": 3.1882,
      "step": 90
    },
    {
      "epoch": 0.006098071736107621,
      "grad_norm": 6.474289894104004,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.9062,
      "step": 91
    },
    {
      "epoch": 0.006165083513427485,
      "grad_norm": 6.908431529998779,
      "learning_rate": 6.066498153718735e-05,
      "loss": 3.8009,
      "step": 92
    },
    {
      "epoch": 0.006232095290747349,
      "grad_norm": 6.610859394073486,
      "learning_rate": 5.985585137257401e-05,
      "loss": 3.0205,
      "step": 93
    },
    {
      "epoch": 0.006299107068067212,
      "grad_norm": 7.842891693115234,
      "learning_rate": 5.90440267166055e-05,
      "loss": 3.0723,
      "step": 94
    },
    {
      "epoch": 0.006366118845387077,
      "grad_norm": 7.9277777671813965,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6804,
      "step": 95
    },
    {
      "epoch": 0.006433130622706941,
      "grad_norm": 8.31147289276123,
      "learning_rate": 5.74131823855921e-05,
      "loss": 3.4852,
      "step": 96
    },
    {
      "epoch": 0.006500142400026805,
      "grad_norm": 8.773941993713379,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.2839,
      "step": 97
    },
    {
      "epoch": 0.006567154177346669,
      "grad_norm": 9.821691513061523,
      "learning_rate": 5.577423184847932e-05,
      "loss": 3.0602,
      "step": 98
    },
    {
      "epoch": 0.0066341659546665325,
      "grad_norm": 12.05204963684082,
      "learning_rate": 5.495227651252315e-05,
      "loss": 3.0884,
      "step": 99
    },
    {
      "epoch": 0.006701177731986397,
      "grad_norm": 16.014434814453125,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.0637,
      "step": 100
    },
    {
      "epoch": 0.006701177731986397,
      "eval_loss": 0.7381904125213623,
      "eval_runtime": 1991.5974,
      "eval_samples_per_second": 12.62,
      "eval_steps_per_second": 3.155,
      "step": 100
    },
    {
      "epoch": 0.00676818950930626,
      "grad_norm": 3.4336345195770264,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.2992,
      "step": 101
    },
    {
      "epoch": 0.006835201286626125,
      "grad_norm": 2.7011988162994385,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.2436,
      "step": 102
    },
    {
      "epoch": 0.006902213063945988,
      "grad_norm": 2.623509407043457,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.6087,
      "step": 103
    },
    {
      "epoch": 0.0069692248412658525,
      "grad_norm": 2.6322193145751953,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.6776,
      "step": 104
    },
    {
      "epoch": 0.007036236618585716,
      "grad_norm": 2.598435878753662,
      "learning_rate": 5e-05,
      "loss": 2.4349,
      "step": 105
    },
    {
      "epoch": 0.00710324839590558,
      "grad_norm": 2.986523151397705,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.6434,
      "step": 106
    },
    {
      "epoch": 0.007170260173225445,
      "grad_norm": 2.7807955741882324,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.3435,
      "step": 107
    },
    {
      "epoch": 0.007237271950545308,
      "grad_norm": 3.3481178283691406,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.6426,
      "step": 108
    },
    {
      "epoch": 0.007304283727865173,
      "grad_norm": 3.0154755115509033,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.2395,
      "step": 109
    },
    {
      "epoch": 0.007371295505185036,
      "grad_norm": 3.805389881134033,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.3782,
      "step": 110
    },
    {
      "epoch": 0.0074383072825049005,
      "grad_norm": 3.2649686336517334,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.1778,
      "step": 111
    },
    {
      "epoch": 0.007505319059824764,
      "grad_norm": 3.6283347606658936,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.5709,
      "step": 112
    },
    {
      "epoch": 0.007572330837144628,
      "grad_norm": 3.476731777191162,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.5964,
      "step": 113
    },
    {
      "epoch": 0.007639342614464492,
      "grad_norm": 3.379956007003784,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.4988,
      "step": 114
    },
    {
      "epoch": 0.007706354391784356,
      "grad_norm": 3.687070846557617,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.6037,
      "step": 115
    },
    {
      "epoch": 0.00777336616910422,
      "grad_norm": 3.569629192352295,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.33,
      "step": 116
    },
    {
      "epoch": 0.007840377946424084,
      "grad_norm": 3.511577844619751,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.7223,
      "step": 117
    },
    {
      "epoch": 0.007907389723743947,
      "grad_norm": 3.7060160636901855,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.6752,
      "step": 118
    },
    {
      "epoch": 0.007974401501063813,
      "grad_norm": 3.7573060989379883,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.7462,
      "step": 119
    },
    {
      "epoch": 0.008041413278383676,
      "grad_norm": 4.448085784912109,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.836,
      "step": 120
    },
    {
      "epoch": 0.00810842505570354,
      "grad_norm": 3.845568895339966,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.4517,
      "step": 121
    },
    {
      "epoch": 0.008175436833023403,
      "grad_norm": 3.9906766414642334,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.8422,
      "step": 122
    },
    {
      "epoch": 0.008242448610343268,
      "grad_norm": 4.061316967010498,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.745,
      "step": 123
    },
    {
      "epoch": 0.008309460387663132,
      "grad_norm": 4.211214065551758,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.57,
      "step": 124
    },
    {
      "epoch": 0.008376472164982995,
      "grad_norm": 4.14617919921875,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.5636,
      "step": 125
    },
    {
      "epoch": 0.008443483942302859,
      "grad_norm": 4.362563133239746,
      "learning_rate": 3.298534127791785e-05,
      "loss": 3.0273,
      "step": 126
    },
    {
      "epoch": 0.008510495719622724,
      "grad_norm": 4.220723628997803,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.6382,
      "step": 127
    },
    {
      "epoch": 0.008577507496942588,
      "grad_norm": 4.606372356414795,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.8578,
      "step": 128
    },
    {
      "epoch": 0.008644519274262451,
      "grad_norm": 4.473711967468262,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.9329,
      "step": 129
    },
    {
      "epoch": 0.008711531051582316,
      "grad_norm": 4.153161525726318,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.5251,
      "step": 130
    },
    {
      "epoch": 0.00877854282890218,
      "grad_norm": 4.623281002044678,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.7656,
      "step": 131
    },
    {
      "epoch": 0.008845554606222043,
      "grad_norm": 4.800572872161865,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.6164,
      "step": 132
    },
    {
      "epoch": 0.008912566383541907,
      "grad_norm": 5.332072734832764,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.707,
      "step": 133
    },
    {
      "epoch": 0.008979578160861772,
      "grad_norm": 5.535910606384277,
      "learning_rate": 2.693294185106562e-05,
      "loss": 3.1558,
      "step": 134
    },
    {
      "epoch": 0.009046589938181636,
      "grad_norm": 6.015562057495117,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.7852,
      "step": 135
    },
    {
      "epoch": 0.009113601715501499,
      "grad_norm": 6.302767753601074,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.924,
      "step": 136
    },
    {
      "epoch": 0.009180613492821363,
      "grad_norm": 5.9889817237854,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.9211,
      "step": 137
    },
    {
      "epoch": 0.009247625270141228,
      "grad_norm": 6.719918251037598,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.9759,
      "step": 138
    },
    {
      "epoch": 0.009314637047461091,
      "grad_norm": 6.2145466804504395,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.76,
      "step": 139
    },
    {
      "epoch": 0.009381648824780955,
      "grad_norm": 6.185413360595703,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 3.1391,
      "step": 140
    },
    {
      "epoch": 0.00944866060210082,
      "grad_norm": 6.555198669433594,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.5508,
      "step": 141
    },
    {
      "epoch": 0.009515672379420683,
      "grad_norm": 7.662604331970215,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.0451,
      "step": 142
    },
    {
      "epoch": 0.009582684156740547,
      "grad_norm": 7.461409568786621,
      "learning_rate": 2.061073738537635e-05,
      "loss": 3.3342,
      "step": 143
    },
    {
      "epoch": 0.00964969593406041,
      "grad_norm": 8.082277297973633,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 3.191,
      "step": 144
    },
    {
      "epoch": 0.009716707711380276,
      "grad_norm": 8.397353172302246,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.8982,
      "step": 145
    },
    {
      "epoch": 0.00978371948870014,
      "grad_norm": 8.452592849731445,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 3.6459,
      "step": 146
    },
    {
      "epoch": 0.009850731266020003,
      "grad_norm": 9.31845760345459,
      "learning_rate": 1.800157297483417e-05,
      "loss": 3.3051,
      "step": 147
    },
    {
      "epoch": 0.009917743043339866,
      "grad_norm": 10.629938125610352,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.6277,
      "step": 148
    },
    {
      "epoch": 0.009984754820659731,
      "grad_norm": 11.908143997192383,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 3.9846,
      "step": 149
    },
    {
      "epoch": 0.010051766597979595,
      "grad_norm": 16.358522415161133,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 3.1578,
      "step": 150
    },
    {
      "epoch": 0.010051766597979595,
      "eval_loss": 0.6861550807952881,
      "eval_runtime": 1994.0732,
      "eval_samples_per_second": 12.604,
      "eval_steps_per_second": 3.151,
      "step": 150
    },
    {
      "epoch": 0.010118778375299458,
      "grad_norm": 2.392547845840454,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.0178,
      "step": 151
    },
    {
      "epoch": 0.010185790152619324,
      "grad_norm": 2.4158225059509277,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9237,
      "step": 152
    },
    {
      "epoch": 0.010252801929939187,
      "grad_norm": 2.4237442016601562,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.0154,
      "step": 153
    },
    {
      "epoch": 0.01031981370725905,
      "grad_norm": 2.6341869831085205,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.2634,
      "step": 154
    },
    {
      "epoch": 0.010386825484578914,
      "grad_norm": 3.080211877822876,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.6441,
      "step": 155
    },
    {
      "epoch": 0.01045383726189878,
      "grad_norm": 3.1427195072174072,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.4366,
      "step": 156
    },
    {
      "epoch": 0.010520849039218643,
      "grad_norm": 2.933468818664551,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.9356,
      "step": 157
    },
    {
      "epoch": 0.010587860816538506,
      "grad_norm": 2.947983503341675,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.353,
      "step": 158
    },
    {
      "epoch": 0.010654872593858371,
      "grad_norm": 2.8385581970214844,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.1686,
      "step": 159
    },
    {
      "epoch": 0.010721884371178235,
      "grad_norm": 3.2543482780456543,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.3608,
      "step": 160
    },
    {
      "epoch": 0.010788896148498098,
      "grad_norm": 3.2085723876953125,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.6463,
      "step": 161
    },
    {
      "epoch": 0.010855907925817962,
      "grad_norm": 3.250600576400757,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.7104,
      "step": 162
    },
    {
      "epoch": 0.010922919703137827,
      "grad_norm": 3.3636810779571533,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.6361,
      "step": 163
    },
    {
      "epoch": 0.01098993148045769,
      "grad_norm": 3.599681854248047,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.5543,
      "step": 164
    },
    {
      "epoch": 0.011056943257777554,
      "grad_norm": 3.432981014251709,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.627,
      "step": 165
    },
    {
      "epoch": 0.011123955035097418,
      "grad_norm": 3.219128131866455,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.5214,
      "step": 166
    },
    {
      "epoch": 0.011190966812417283,
      "grad_norm": 3.520371437072754,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.5199,
      "step": 167
    },
    {
      "epoch": 0.011257978589737146,
      "grad_norm": 3.379117250442505,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.6324,
      "step": 168
    },
    {
      "epoch": 0.01132499036705701,
      "grad_norm": 3.5783040523529053,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.84,
      "step": 169
    },
    {
      "epoch": 0.011392002144376875,
      "grad_norm": 3.4731979370117188,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.5246,
      "step": 170
    },
    {
      "epoch": 0.011459013921696739,
      "grad_norm": 3.354538917541504,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.522,
      "step": 171
    },
    {
      "epoch": 0.011526025699016602,
      "grad_norm": 3.5728278160095215,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.8172,
      "step": 172
    },
    {
      "epoch": 0.011593037476336466,
      "grad_norm": 4.089961528778076,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.9409,
      "step": 173
    },
    {
      "epoch": 0.01166004925365633,
      "grad_norm": 4.006018161773682,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.6685,
      "step": 174
    },
    {
      "epoch": 0.011727061030976194,
      "grad_norm": 3.9705677032470703,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.5336,
      "step": 175
    },
    {
      "epoch": 0.011794072808296058,
      "grad_norm": 3.8912458419799805,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.2566,
      "step": 176
    },
    {
      "epoch": 0.011861084585615921,
      "grad_norm": 4.209029197692871,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.9324,
      "step": 177
    },
    {
      "epoch": 0.011928096362935786,
      "grad_norm": 4.272871494293213,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.5612,
      "step": 178
    },
    {
      "epoch": 0.01199510814025565,
      "grad_norm": 4.781076431274414,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 3.0724,
      "step": 179
    },
    {
      "epoch": 0.012062119917575513,
      "grad_norm": 4.252874851226807,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.7622,
      "step": 180
    },
    {
      "epoch": 0.012129131694895379,
      "grad_norm": 4.639196395874023,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5305,
      "step": 181
    },
    {
      "epoch": 0.012196143472215242,
      "grad_norm": 4.713561058044434,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 3.0388,
      "step": 182
    },
    {
      "epoch": 0.012263155249535106,
      "grad_norm": 4.903632164001465,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.9792,
      "step": 183
    },
    {
      "epoch": 0.01233016702685497,
      "grad_norm": 5.172715663909912,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.7801,
      "step": 184
    },
    {
      "epoch": 0.012397178804174834,
      "grad_norm": 5.697531223297119,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.918,
      "step": 185
    },
    {
      "epoch": 0.012464190581494698,
      "grad_norm": 5.048072338104248,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.3824,
      "step": 186
    },
    {
      "epoch": 0.012531202358814561,
      "grad_norm": 5.522087574005127,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.4603,
      "step": 187
    },
    {
      "epoch": 0.012598214136134425,
      "grad_norm": 5.683063983917236,
      "learning_rate": 9.810017062595322e-07,
      "loss": 3.5254,
      "step": 188
    },
    {
      "epoch": 0.01266522591345429,
      "grad_norm": 5.502018928527832,
      "learning_rate": 8.247462563808817e-07,
      "loss": 3.891,
      "step": 189
    },
    {
      "epoch": 0.012732237690774154,
      "grad_norm": 5.893340110778809,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.8505,
      "step": 190
    },
    {
      "epoch": 0.012799249468094017,
      "grad_norm": 5.934426307678223,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.8021,
      "step": 191
    },
    {
      "epoch": 0.012866261245413882,
      "grad_norm": 7.878473281860352,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.527,
      "step": 192
    },
    {
      "epoch": 0.012933273022733746,
      "grad_norm": 6.568013668060303,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.5976,
      "step": 193
    },
    {
      "epoch": 0.01300028480005361,
      "grad_norm": 6.732630252838135,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.7497,
      "step": 194
    },
    {
      "epoch": 0.013067296577373473,
      "grad_norm": 8.31375503540039,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.8449,
      "step": 195
    },
    {
      "epoch": 0.013134308354693338,
      "grad_norm": 8.19385814666748,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.3041,
      "step": 196
    },
    {
      "epoch": 0.013201320132013201,
      "grad_norm": 9.54801082611084,
      "learning_rate": 6.150154258476315e-08,
      "loss": 3.3472,
      "step": 197
    },
    {
      "epoch": 0.013268331909333065,
      "grad_norm": 8.8629150390625,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.2483,
      "step": 198
    },
    {
      "epoch": 0.013335343686652928,
      "grad_norm": 9.573737144470215,
      "learning_rate": 6.834750376549792e-09,
      "loss": 3.4225,
      "step": 199
    },
    {
      "epoch": 0.013402355463972794,
      "grad_norm": 16.116779327392578,
      "learning_rate": 0.0,
      "loss": 3.4212,
      "step": 200
    },
    {
      "epoch": 0.013402355463972794,
      "eval_loss": 0.6790642738342285,
      "eval_runtime": 1993.2252,
      "eval_samples_per_second": 12.61,
      "eval_steps_per_second": 3.153,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.969324210670797e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}