{
  "best_metric": 2.1015098094940186,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3401360544217687,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017006802721088435,
      "grad_norm": 6.714749813079834,
      "learning_rate": 1e-05,
      "loss": 8.7146,
      "step": 1
    },
    {
      "epoch": 0.0017006802721088435,
      "eval_loss": 2.2044851779937744,
      "eval_runtime": 166.4736,
      "eval_samples_per_second": 5.947,
      "eval_steps_per_second": 1.49,
      "step": 1
    },
    {
      "epoch": 0.003401360544217687,
      "grad_norm": 6.183635234832764,
      "learning_rate": 2e-05,
      "loss": 8.4547,
      "step": 2
    },
    {
      "epoch": 0.00510204081632653,
      "grad_norm": 6.1161065101623535,
      "learning_rate": 3e-05,
      "loss": 8.7953,
      "step": 3
    },
    {
      "epoch": 0.006802721088435374,
      "grad_norm": 4.83544921875,
      "learning_rate": 4e-05,
      "loss": 8.4317,
      "step": 4
    },
    {
      "epoch": 0.008503401360544218,
      "grad_norm": 5.279078960418701,
      "learning_rate": 5e-05,
      "loss": 8.3697,
      "step": 5
    },
    {
      "epoch": 0.01020408163265306,
      "grad_norm": 5.119232177734375,
      "learning_rate": 6e-05,
      "loss": 8.7176,
      "step": 6
    },
    {
      "epoch": 0.011904761904761904,
      "grad_norm": 4.606754302978516,
      "learning_rate": 7e-05,
      "loss": 8.421,
      "step": 7
    },
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 5.059332847595215,
      "learning_rate": 8e-05,
      "loss": 8.2958,
      "step": 8
    },
    {
      "epoch": 0.015306122448979591,
      "grad_norm": 4.8666486740112305,
      "learning_rate": 9e-05,
      "loss": 8.6875,
      "step": 9
    },
    {
      "epoch": 0.017006802721088437,
      "grad_norm": 4.661813735961914,
      "learning_rate": 0.0001,
      "loss": 8.7021,
      "step": 10
    },
    {
      "epoch": 0.01870748299319728,
      "grad_norm": 4.167578220367432,
      "learning_rate": 9.999316524962345e-05,
      "loss": 7.7501,
      "step": 11
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 4.497811794281006,
      "learning_rate": 9.997266286704631e-05,
      "loss": 8.6324,
      "step": 12
    },
    {
      "epoch": 0.022108843537414966,
      "grad_norm": 4.054053783416748,
      "learning_rate": 9.993849845741524e-05,
      "loss": 8.6611,
      "step": 13
    },
    {
      "epoch": 0.023809523809523808,
      "grad_norm": 3.894209384918213,
      "learning_rate": 9.989068136093873e-05,
      "loss": 7.9581,
      "step": 14
    },
    {
      "epoch": 0.025510204081632654,
      "grad_norm": 3.855048179626465,
      "learning_rate": 9.98292246503335e-05,
      "loss": 8.4279,
      "step": 15
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 3.5461373329162598,
      "learning_rate": 9.975414512725057e-05,
      "loss": 8.1087,
      "step": 16
    },
    {
      "epoch": 0.02891156462585034,
      "grad_norm": 3.5614609718322754,
      "learning_rate": 9.966546331768191e-05,
      "loss": 8.522,
      "step": 17
    },
    {
      "epoch": 0.030612244897959183,
      "grad_norm": 3.8800792694091797,
      "learning_rate": 9.956320346634876e-05,
      "loss": 8.4559,
      "step": 18
    },
    {
      "epoch": 0.03231292517006803,
      "grad_norm": 3.53694224357605,
      "learning_rate": 9.944739353007344e-05,
      "loss": 8.2512,
      "step": 19
    },
    {
      "epoch": 0.034013605442176874,
      "grad_norm": 3.4773006439208984,
      "learning_rate": 9.931806517013612e-05,
      "loss": 8.5001,
      "step": 20
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 3.359560251235962,
      "learning_rate": 9.917525374361912e-05,
      "loss": 7.5962,
      "step": 21
    },
    {
      "epoch": 0.03741496598639456,
      "grad_norm": 3.146897554397583,
      "learning_rate": 9.901899829374047e-05,
      "loss": 7.5821,
      "step": 22
    },
    {
      "epoch": 0.0391156462585034,
      "grad_norm": 3.0724167823791504,
      "learning_rate": 9.884934153917997e-05,
      "loss": 7.9377,
      "step": 23
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 3.0712780952453613,
      "learning_rate": 9.86663298624003e-05,
      "loss": 7.8861,
      "step": 24
    },
    {
      "epoch": 0.04251700680272109,
      "grad_norm": 3.401433229446411,
      "learning_rate": 9.847001329696653e-05,
      "loss": 8.0017,
      "step": 25
    },
    {
      "epoch": 0.04421768707482993,
      "grad_norm": 3.4345617294311523,
      "learning_rate": 9.826044551386744e-05,
      "loss": 8.5047,
      "step": 26
    },
    {
      "epoch": 0.04591836734693878,
      "grad_norm": 3.3811564445495605,
      "learning_rate": 9.803768380684242e-05,
      "loss": 8.3609,
      "step": 27
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 3.4209554195404053,
      "learning_rate": 9.780178907671789e-05,
      "loss": 8.3527,
      "step": 28
    },
    {
      "epoch": 0.04931972789115646,
      "grad_norm": 3.5917084217071533,
      "learning_rate": 9.755282581475769e-05,
      "loss": 8.4742,
      "step": 29
    },
    {
      "epoch": 0.05102040816326531,
      "grad_norm": 3.58453369140625,
      "learning_rate": 9.729086208503174e-05,
      "loss": 8.6524,
      "step": 30
    },
    {
      "epoch": 0.05272108843537415,
      "grad_norm": 3.7753114700317383,
      "learning_rate": 9.701596950580806e-05,
      "loss": 9.1421,
      "step": 31
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 3.782766819000244,
      "learning_rate": 9.672822322997305e-05,
      "loss": 8.5608,
      "step": 32
    },
    {
      "epoch": 0.05612244897959184,
      "grad_norm": 3.8492214679718018,
      "learning_rate": 9.642770192448536e-05,
      "loss": 8.9691,
      "step": 33
    },
    {
      "epoch": 0.05782312925170068,
      "grad_norm": 3.81087327003479,
      "learning_rate": 9.611448774886924e-05,
      "loss": 8.4944,
      "step": 34
    },
    {
      "epoch": 0.05952380952380952,
      "grad_norm": 4.038293838500977,
      "learning_rate": 9.578866633275288e-05,
      "loss": 9.0974,
      "step": 35
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 3.9375903606414795,
      "learning_rate": 9.545032675245813e-05,
      "loss": 8.6506,
      "step": 36
    },
    {
      "epoch": 0.06292517006802721,
      "grad_norm": 4.183244705200195,
      "learning_rate": 9.509956150664796e-05,
      "loss": 8.6248,
      "step": 37
    },
    {
      "epoch": 0.06462585034013606,
      "grad_norm": 4.185721397399902,
      "learning_rate": 9.473646649103818e-05,
      "loss": 8.5399,
      "step": 38
    },
    {
      "epoch": 0.0663265306122449,
      "grad_norm": 4.3026556968688965,
      "learning_rate": 9.43611409721806e-05,
      "loss": 8.4857,
      "step": 39
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 4.410216808319092,
      "learning_rate": 9.397368756032445e-05,
      "loss": 8.5795,
      "step": 40
    },
    {
      "epoch": 0.06972789115646258,
      "grad_norm": 4.4675092697143555,
      "learning_rate": 9.357421218136386e-05,
      "loss": 8.4653,
      "step": 41
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 4.690471649169922,
      "learning_rate": 9.316282404787871e-05,
      "loss": 8.9399,
      "step": 42
    },
    {
      "epoch": 0.07312925170068027,
      "grad_norm": 4.525552749633789,
      "learning_rate": 9.273963562927695e-05,
      "loss": 8.368,
      "step": 43
    },
    {
      "epoch": 0.07482993197278912,
      "grad_norm": 4.87044095993042,
      "learning_rate": 9.230476262104677e-05,
      "loss": 8.754,
      "step": 44
    },
    {
      "epoch": 0.07653061224489796,
      "grad_norm": 4.972063064575195,
      "learning_rate": 9.185832391312644e-05,
      "loss": 8.5943,
      "step": 45
    },
    {
      "epoch": 0.0782312925170068,
      "grad_norm": 5.44361686706543,
      "learning_rate": 9.140044155740101e-05,
      "loss": 8.8404,
      "step": 46
    },
    {
      "epoch": 0.07993197278911565,
      "grad_norm": 5.5264434814453125,
      "learning_rate": 9.093124073433463e-05,
      "loss": 8.7216,
      "step": 47
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 5.9806318283081055,
      "learning_rate": 9.045084971874738e-05,
      "loss": 9.1827,
      "step": 48
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 6.1380438804626465,
      "learning_rate": 8.995939984474624e-05,
      "loss": 9.0687,
      "step": 49
    },
    {
      "epoch": 0.08503401360544217,
      "grad_norm": 6.707734107971191,
      "learning_rate": 8.945702546981969e-05,
      "loss": 9.309,
      "step": 50
    },
    {
      "epoch": 0.08503401360544217,
      "eval_loss": 2.154860019683838,
      "eval_runtime": 168.3251,
      "eval_samples_per_second": 5.881,
      "eval_steps_per_second": 1.473,
      "step": 50
    },
    {
      "epoch": 0.08673469387755102,
      "grad_norm": 3.603348731994629,
      "learning_rate": 8.894386393810563e-05,
      "loss": 7.9831,
      "step": 51
    },
    {
      "epoch": 0.08843537414965986,
      "grad_norm": 3.376866340637207,
      "learning_rate": 8.842005554284296e-05,
      "loss": 8.2731,
      "step": 52
    },
    {
      "epoch": 0.09013605442176871,
      "grad_norm": 2.9752745628356934,
      "learning_rate": 8.788574348801675e-05,
      "loss": 8.3384,
      "step": 53
    },
    {
      "epoch": 0.09183673469387756,
      "grad_norm": 2.9328720569610596,
      "learning_rate": 8.73410738492077e-05,
      "loss": 8.4844,
      "step": 54
    },
    {
      "epoch": 0.0935374149659864,
      "grad_norm": 2.7755463123321533,
      "learning_rate": 8.678619553365659e-05,
      "loss": 7.8982,
      "step": 55
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 2.9773318767547607,
      "learning_rate": 8.622126023955446e-05,
      "loss": 8.0223,
      "step": 56
    },
    {
      "epoch": 0.09693877551020408,
      "grad_norm": 3.0061182975769043,
      "learning_rate": 8.564642241456986e-05,
      "loss": 8.49,
      "step": 57
    },
    {
      "epoch": 0.09863945578231292,
      "grad_norm": 2.942753791809082,
      "learning_rate": 8.506183921362443e-05,
      "loss": 8.5193,
      "step": 58
    },
    {
      "epoch": 0.10034013605442177,
      "grad_norm": 2.744061231613159,
      "learning_rate": 8.44676704559283e-05,
      "loss": 7.9305,
      "step": 59
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 2.8421335220336914,
      "learning_rate": 8.386407858128706e-05,
      "loss": 8.4879,
      "step": 60
    },
    {
      "epoch": 0.10374149659863946,
      "grad_norm": 2.866511106491089,
      "learning_rate": 8.32512286056924e-05,
      "loss": 8.6973,
      "step": 61
    },
    {
      "epoch": 0.1054421768707483,
      "grad_norm": 2.857673406600952,
      "learning_rate": 8.262928807620843e-05,
      "loss": 8.2819,
      "step": 62
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 2.8629140853881836,
      "learning_rate": 8.199842702516583e-05,
      "loss": 8.25,
      "step": 63
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 2.75886607170105,
      "learning_rate": 8.135881792367686e-05,
      "loss": 8.3936,
      "step": 64
    },
    {
      "epoch": 0.11054421768707483,
      "grad_norm": 2.746535062789917,
      "learning_rate": 8.07106356344834e-05,
      "loss": 8.1376,
      "step": 65
    },
    {
      "epoch": 0.11224489795918367,
      "grad_norm": 2.8505587577819824,
      "learning_rate": 8.005405736415126e-05,
      "loss": 8.3909,
      "step": 66
    },
    {
      "epoch": 0.11394557823129252,
      "grad_norm": 2.7247655391693115,
      "learning_rate": 7.938926261462366e-05,
      "loss": 8.1479,
      "step": 67
    },
    {
      "epoch": 0.11564625850340136,
      "grad_norm": 2.76584792137146,
      "learning_rate": 7.871643313414718e-05,
      "loss": 8.4541,
      "step": 68
    },
    {
      "epoch": 0.11734693877551021,
      "grad_norm": 2.764802932739258,
      "learning_rate": 7.803575286758364e-05,
      "loss": 8.6158,
      "step": 69
    },
    {
      "epoch": 0.11904761904761904,
      "grad_norm": 2.6732802391052246,
      "learning_rate": 7.734740790612136e-05,
      "loss": 8.4331,
      "step": 70
    },
    {
      "epoch": 0.12074829931972789,
      "grad_norm": 2.681309938430786,
      "learning_rate": 7.66515864363997e-05,
      "loss": 8.138,
      "step": 71
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 2.6904282569885254,
      "learning_rate": 7.594847868906076e-05,
      "loss": 8.0172,
      "step": 72
    },
    {
      "epoch": 0.12414965986394558,
      "grad_norm": 2.6603457927703857,
      "learning_rate": 7.52382768867422e-05,
      "loss": 7.8828,
      "step": 73
    },
    {
      "epoch": 0.12585034013605442,
      "grad_norm": 2.763792037963867,
      "learning_rate": 7.452117519152542e-05,
      "loss": 8.2998,
      "step": 74
    },
    {
      "epoch": 0.12755102040816327,
      "grad_norm": 2.6464648246765137,
      "learning_rate": 7.379736965185368e-05,
      "loss": 8.1136,
      "step": 75
    },
    {
      "epoch": 0.1292517006802721,
      "grad_norm": 2.75872802734375,
      "learning_rate": 7.30670581489344e-05,
      "loss": 8.2918,
      "step": 76
    },
    {
      "epoch": 0.13095238095238096,
      "grad_norm": 2.8654608726501465,
      "learning_rate": 7.233044034264034e-05,
      "loss": 8.2762,
      "step": 77
    },
    {
      "epoch": 0.1326530612244898,
      "grad_norm": 3.0308353900909424,
      "learning_rate": 7.158771761692464e-05,
      "loss": 8.2789,
      "step": 78
    },
    {
      "epoch": 0.13435374149659865,
      "grad_norm": 3.077528715133667,
      "learning_rate": 7.083909302476453e-05,
      "loss": 8.5138,
      "step": 79
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 3.1231086254119873,
      "learning_rate": 7.008477123264848e-05,
      "loss": 8.545,
      "step": 80
    },
    {
      "epoch": 0.1377551020408163,
      "grad_norm": 3.171665668487549,
      "learning_rate": 6.932495846462261e-05,
      "loss": 8.292,
      "step": 81
    },
    {
      "epoch": 0.13945578231292516,
      "grad_norm": 3.3295700550079346,
      "learning_rate": 6.855986244591104e-05,
      "loss": 8.8624,
      "step": 82
    },
    {
      "epoch": 0.141156462585034,
      "grad_norm": 3.192854881286621,
      "learning_rate": 6.778969234612584e-05,
      "loss": 8.4102,
      "step": 83
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 3.304255962371826,
      "learning_rate": 6.701465872208216e-05,
      "loss": 8.2563,
      "step": 84
    },
    {
      "epoch": 0.1445578231292517,
      "grad_norm": 3.5270016193389893,
      "learning_rate": 6.623497346023418e-05,
      "loss": 8.2568,
      "step": 85
    },
    {
      "epoch": 0.14625850340136054,
      "grad_norm": 3.503565788269043,
      "learning_rate": 6.545084971874738e-05,
      "loss": 8.5318,
      "step": 86
    },
    {
      "epoch": 0.14795918367346939,
      "grad_norm": 3.811997652053833,
      "learning_rate": 6.466250186922325e-05,
      "loss": 8.5162,
      "step": 87
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 3.7655692100524902,
      "learning_rate": 6.387014543809223e-05,
      "loss": 8.5464,
      "step": 88
    },
    {
      "epoch": 0.15136054421768708,
      "grad_norm": 3.7764742374420166,
      "learning_rate": 6.307399704769099e-05,
      "loss": 8.2092,
      "step": 89
    },
    {
      "epoch": 0.15306122448979592,
      "grad_norm": 4.0026068687438965,
      "learning_rate": 6.227427435703997e-05,
      "loss": 8.8043,
      "step": 90
    },
    {
      "epoch": 0.15476190476190477,
      "grad_norm": 3.9808855056762695,
      "learning_rate": 6.147119600233758e-05,
      "loss": 8.202,
      "step": 91
    },
    {
      "epoch": 0.1564625850340136,
      "grad_norm": 4.2311506271362305,
      "learning_rate": 6.066498153718735e-05,
      "loss": 8.7583,
      "step": 92
    },
    {
      "epoch": 0.15816326530612246,
      "grad_norm": 4.375192642211914,
      "learning_rate": 5.985585137257401e-05,
      "loss": 8.7502,
      "step": 93
    },
    {
      "epoch": 0.1598639455782313,
      "grad_norm": 4.594576358795166,
      "learning_rate": 5.90440267166055e-05,
      "loss": 8.7413,
      "step": 94
    },
    {
      "epoch": 0.16156462585034015,
      "grad_norm": 4.622951984405518,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 8.3798,
      "step": 95
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 4.7610182762146,
      "learning_rate": 5.74131823855921e-05,
      "loss": 8.2725,
      "step": 96
    },
    {
      "epoch": 0.1649659863945578,
      "grad_norm": 4.730075359344482,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 8.2059,
      "step": 97
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 5.103707313537598,
      "learning_rate": 5.577423184847932e-05,
      "loss": 8.5788,
      "step": 98
    },
    {
      "epoch": 0.1683673469387755,
      "grad_norm": 5.315412521362305,
      "learning_rate": 5.495227651252315e-05,
      "loss": 8.5764,
      "step": 99
    },
    {
      "epoch": 0.17006802721088435,
      "grad_norm": 5.606744766235352,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 9.0685,
      "step": 100
    },
    {
      "epoch": 0.17006802721088435,
      "eval_loss": 2.1324100494384766,
      "eval_runtime": 167.3885,
      "eval_samples_per_second": 5.914,
      "eval_steps_per_second": 1.482,
      "step": 100
    },
    {
      "epoch": 0.1717687074829932,
      "grad_norm": 3.411219835281372,
      "learning_rate": 5.330452921628497e-05,
      "loss": 8.1978,
      "step": 101
    },
    {
      "epoch": 0.17346938775510204,
      "grad_norm": 2.9712109565734863,
      "learning_rate": 5.247918773366112e-05,
      "loss": 8.1988,
      "step": 102
    },
    {
      "epoch": 0.17517006802721088,
      "grad_norm": 2.6267952919006348,
      "learning_rate": 5.165316846586541e-05,
      "loss": 8.0431,
      "step": 103
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 2.626807928085327,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 7.7013,
      "step": 104
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 2.6987144947052,
      "learning_rate": 5e-05,
      "loss": 8.2586,
      "step": 105
    },
    {
      "epoch": 0.18027210884353742,
      "grad_norm": 2.6826634407043457,
      "learning_rate": 4.917330276168208e-05,
      "loss": 8.1472,
      "step": 106
    },
    {
      "epoch": 0.18197278911564627,
      "grad_norm": 2.735485315322876,
      "learning_rate": 4.834683153413459e-05,
      "loss": 8.4179,
      "step": 107
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 2.5861611366271973,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 7.8594,
      "step": 108
    },
    {
      "epoch": 0.18537414965986396,
      "grad_norm": 2.5558035373687744,
      "learning_rate": 4.669547078371504e-05,
      "loss": 8.0071,
      "step": 109
    },
    {
      "epoch": 0.1870748299319728,
      "grad_norm": 2.7161362171173096,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 8.5373,
      "step": 110
    },
    {
      "epoch": 0.18877551020408162,
      "grad_norm": 2.7265706062316895,
      "learning_rate": 4.504772348747687e-05,
      "loss": 8.5177,
      "step": 111
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 2.675358295440674,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 8.2526,
      "step": 112
    },
    {
      "epoch": 0.1921768707482993,
      "grad_norm": 2.5557126998901367,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 8.3611,
      "step": 113
    },
    {
      "epoch": 0.19387755102040816,
      "grad_norm": 2.600491523742676,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 8.0857,
      "step": 114
    },
    {
      "epoch": 0.195578231292517,
      "grad_norm": 2.675493001937866,
      "learning_rate": 4.17702704859633e-05,
      "loss": 8.4282,
      "step": 115
    },
    {
      "epoch": 0.19727891156462585,
      "grad_norm": 2.6874914169311523,
      "learning_rate": 4.095597328339452e-05,
      "loss": 8.7172,
      "step": 116
    },
    {
      "epoch": 0.1989795918367347,
      "grad_norm": 2.6319267749786377,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 8.4087,
      "step": 117
    },
    {
      "epoch": 0.20068027210884354,
      "grad_norm": 2.5473825931549072,
      "learning_rate": 3.933501846281267e-05,
      "loss": 7.9771,
      "step": 118
    },
    {
      "epoch": 0.20238095238095238,
      "grad_norm": 2.4717583656311035,
      "learning_rate": 3.852880399766243e-05,
      "loss": 8.2495,
      "step": 119
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 2.6229114532470703,
      "learning_rate": 3.772572564296005e-05,
      "loss": 7.9508,
      "step": 120
    },
    {
      "epoch": 0.20578231292517007,
      "grad_norm": 2.567268133163452,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 8.2602,
      "step": 121
    },
    {
      "epoch": 0.20748299319727892,
      "grad_norm": 2.5883162021636963,
      "learning_rate": 3.612985456190778e-05,
      "loss": 7.905,
      "step": 122
    },
    {
      "epoch": 0.20918367346938777,
      "grad_norm": 2.4426956176757812,
      "learning_rate": 3.533749813077677e-05,
      "loss": 7.5359,
      "step": 123
    },
    {
      "epoch": 0.2108843537414966,
      "grad_norm": 2.634551525115967,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 8.4191,
      "step": 124
    },
    {
      "epoch": 0.21258503401360543,
      "grad_norm": 2.5540385246276855,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 7.9974,
      "step": 125
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 2.6500864028930664,
      "learning_rate": 3.298534127791785e-05,
      "loss": 8.1055,
      "step": 126
    },
    {
      "epoch": 0.21598639455782312,
      "grad_norm": 2.833156108856201,
      "learning_rate": 3.221030765387417e-05,
      "loss": 8.7102,
      "step": 127
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 2.714055299758911,
      "learning_rate": 3.144013755408895e-05,
      "loss": 8.1279,
      "step": 128
    },
    {
      "epoch": 0.2193877551020408,
      "grad_norm": 2.905270576477051,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 8.287,
      "step": 129
    },
    {
      "epoch": 0.22108843537414966,
      "grad_norm": 3.030742883682251,
      "learning_rate": 2.991522876735154e-05,
      "loss": 8.6168,
      "step": 130
    },
    {
      "epoch": 0.2227891156462585,
      "grad_norm": 3.0595264434814453,
      "learning_rate": 2.916090697523549e-05,
      "loss": 8.6429,
      "step": 131
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 3.1110522747039795,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 8.8154,
      "step": 132
    },
    {
      "epoch": 0.2261904761904762,
      "grad_norm": 3.2264280319213867,
      "learning_rate": 2.766955965735968e-05,
      "loss": 8.5175,
      "step": 133
    },
    {
      "epoch": 0.22789115646258504,
      "grad_norm": 3.237546920776367,
      "learning_rate": 2.693294185106562e-05,
      "loss": 8.4187,
      "step": 134
    },
    {
      "epoch": 0.22959183673469388,
      "grad_norm": 3.30430269241333,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 8.5076,
      "step": 135
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 3.4622321128845215,
      "learning_rate": 2.547882480847461e-05,
      "loss": 8.7037,
      "step": 136
    },
    {
      "epoch": 0.23299319727891157,
      "grad_norm": 3.5023865699768066,
      "learning_rate": 2.476172311325783e-05,
      "loss": 8.5119,
      "step": 137
    },
    {
      "epoch": 0.23469387755102042,
      "grad_norm": 3.5723392963409424,
      "learning_rate": 2.405152131093926e-05,
      "loss": 8.5055,
      "step": 138
    },
    {
      "epoch": 0.23639455782312926,
      "grad_norm": 3.758561611175537,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 8.4222,
      "step": 139
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 3.675050735473633,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 8.4037,
      "step": 140
    },
    {
      "epoch": 0.23979591836734693,
      "grad_norm": 3.7652249336242676,
      "learning_rate": 2.196424713241637e-05,
      "loss": 8.4925,
      "step": 141
    },
    {
      "epoch": 0.24149659863945577,
      "grad_norm": 3.909127712249756,
      "learning_rate": 2.128356686585282e-05,
      "loss": 8.5039,
      "step": 142
    },
    {
      "epoch": 0.24319727891156462,
      "grad_norm": 4.169961929321289,
      "learning_rate": 2.061073738537635e-05,
      "loss": 8.816,
      "step": 143
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 4.12620735168457,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 8.3357,
      "step": 144
    },
    {
      "epoch": 0.2465986394557823,
      "grad_norm": 4.372957706451416,
      "learning_rate": 1.928936436551661e-05,
      "loss": 8.3081,
      "step": 145
    },
    {
      "epoch": 0.24829931972789115,
      "grad_norm": 4.840909957885742,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 8.0614,
      "step": 146
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.817358493804932,
      "learning_rate": 1.800157297483417e-05,
      "loss": 8.3252,
      "step": 147
    },
    {
      "epoch": 0.25170068027210885,
      "grad_norm": 5.278514862060547,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 8.4868,
      "step": 148
    },
    {
      "epoch": 0.2534013605442177,
      "grad_norm": 5.515327453613281,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 8.8076,
      "step": 149
    },
    {
      "epoch": 0.25510204081632654,
      "grad_norm": 6.306919574737549,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 9.1399,
      "step": 150
    },
    {
      "epoch": 0.25510204081632654,
      "eval_loss": 2.108788013458252,
      "eval_runtime": 167.2192,
      "eval_samples_per_second": 5.92,
      "eval_steps_per_second": 1.483,
      "step": 150
    },
    {
      "epoch": 0.2568027210884354,
      "grad_norm": 2.715665102005005,
      "learning_rate": 1.553232954407171e-05,
      "loss": 8.2424,
      "step": 151
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 2.711014747619629,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 8.4406,
      "step": 152
    },
    {
      "epoch": 0.2602040816326531,
      "grad_norm": 2.667208433151245,
      "learning_rate": 1.435357758543015e-05,
      "loss": 8.1946,
      "step": 153
    },
    {
      "epoch": 0.2619047619047619,
      "grad_norm": 2.706900119781494,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 8.2611,
      "step": 154
    },
    {
      "epoch": 0.26360544217687076,
      "grad_norm": 2.7188560962677,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 8.4274,
      "step": 155
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 2.6436047554016113,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 8.2176,
      "step": 156
    },
    {
      "epoch": 0.26700680272108845,
      "grad_norm": 2.6086480617523193,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 8.2724,
      "step": 157
    },
    {
      "epoch": 0.2687074829931973,
      "grad_norm": 2.588721513748169,
      "learning_rate": 1.157994445715706e-05,
      "loss": 7.8066,
      "step": 158
    },
    {
      "epoch": 0.27040816326530615,
      "grad_norm": 2.639631748199463,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 8.3854,
      "step": 159
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 2.5398762226104736,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 8.1485,
      "step": 160
    },
    {
      "epoch": 0.27380952380952384,
      "grad_norm": 2.547229290008545,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 8.0766,
      "step": 161
    },
    {
      "epoch": 0.2755102040816326,
      "grad_norm": 2.4030380249023438,
      "learning_rate": 9.549150281252633e-06,
      "loss": 8.0231,
      "step": 162
    },
    {
      "epoch": 0.27721088435374147,
      "grad_norm": 2.533580780029297,
      "learning_rate": 9.068759265665384e-06,
      "loss": 8.3442,
      "step": 163
    },
    {
      "epoch": 0.2789115646258503,
      "grad_norm": 2.5191545486450195,
      "learning_rate": 8.599558442598998e-06,
      "loss": 8.14,
      "step": 164
    },
    {
      "epoch": 0.28061224489795916,
      "grad_norm": 2.4945812225341797,
      "learning_rate": 8.141676086873572e-06,
      "loss": 8.2592,
      "step": 165
    },
    {
      "epoch": 0.282312925170068,
      "grad_norm": 2.445951223373413,
      "learning_rate": 7.695237378953223e-06,
      "loss": 8.2397,
      "step": 166
    },
    {
      "epoch": 0.28401360544217685,
      "grad_norm": 2.4862051010131836,
      "learning_rate": 7.260364370723044e-06,
      "loss": 8.2873,
      "step": 167
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 2.424083709716797,
      "learning_rate": 6.837175952121306e-06,
      "loss": 8.0432,
      "step": 168
    },
    {
      "epoch": 0.28741496598639454,
      "grad_norm": 2.550328016281128,
      "learning_rate": 6.425787818636131e-06,
      "loss": 8.1391,
      "step": 169
    },
    {
      "epoch": 0.2891156462585034,
      "grad_norm": 2.4685375690460205,
      "learning_rate": 6.026312439675552e-06,
      "loss": 7.9183,
      "step": 170
    },
    {
      "epoch": 0.29081632653061223,
      "grad_norm": 2.3770382404327393,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 8.1037,
      "step": 171
    },
    {
      "epoch": 0.2925170068027211,
      "grad_norm": 2.423503875732422,
      "learning_rate": 5.263533508961827e-06,
      "loss": 7.9687,
      "step": 172
    },
    {
      "epoch": 0.2942176870748299,
      "grad_norm": 2.5000953674316406,
      "learning_rate": 4.900438493352055e-06,
      "loss": 7.861,
      "step": 173
    },
    {
      "epoch": 0.29591836734693877,
      "grad_norm": 2.5570833683013916,
      "learning_rate": 4.549673247541875e-06,
      "loss": 8.2333,
      "step": 174
    },
    {
      "epoch": 0.2976190476190476,
      "grad_norm": 2.5112290382385254,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 8.1147,
      "step": 175
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 2.4931840896606445,
      "learning_rate": 3.885512251130763e-06,
      "loss": 8.5374,
      "step": 176
    },
    {
      "epoch": 0.3010204081632653,
      "grad_norm": 2.687894582748413,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 8.1714,
      "step": 177
    },
    {
      "epoch": 0.30272108843537415,
      "grad_norm": 2.73854660987854,
      "learning_rate": 3.271776770026963e-06,
      "loss": 8.3724,
      "step": 178
    },
    {
      "epoch": 0.304421768707483,
      "grad_norm": 2.7159829139709473,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 8.1649,
      "step": 179
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 2.9380297660827637,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 8.1434,
      "step": 180
    },
    {
      "epoch": 0.3078231292517007,
      "grad_norm": 2.9060769081115723,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 8.5208,
      "step": 181
    },
    {
      "epoch": 0.30952380952380953,
      "grad_norm": 3.0917787551879883,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 9.0192,
      "step": 182
    },
    {
      "epoch": 0.3112244897959184,
      "grad_norm": 3.146559953689575,
      "learning_rate": 1.962316193157593e-06,
      "loss": 8.8936,
      "step": 183
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 3.101686954498291,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 8.4705,
      "step": 184
    },
    {
      "epoch": 0.31462585034013607,
      "grad_norm": 3.419445037841797,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 8.1825,
      "step": 185
    },
    {
      "epoch": 0.3163265306122449,
      "grad_norm": 3.2799603939056396,
      "learning_rate": 1.333670137599713e-06,
      "loss": 8.3592,
      "step": 186
    },
    {
      "epoch": 0.31802721088435376,
      "grad_norm": 3.2579667568206787,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 8.5877,
      "step": 187
    },
    {
      "epoch": 0.3197278911564626,
      "grad_norm": 3.5122814178466797,
      "learning_rate": 9.810017062595322e-07,
      "loss": 8.6479,
      "step": 188
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 3.426347494125366,
      "learning_rate": 8.247462563808817e-07,
      "loss": 8.2555,
      "step": 189
    },
    {
      "epoch": 0.3231292517006803,
      "grad_norm": 3.559950351715088,
      "learning_rate": 6.819348298638839e-07,
      "loss": 8.403,
      "step": 190
    },
    {
      "epoch": 0.32482993197278914,
      "grad_norm": 3.6049587726593018,
      "learning_rate": 5.526064699265753e-07,
      "loss": 8.1693,
      "step": 191
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 3.805424451828003,
      "learning_rate": 4.367965336512403e-07,
      "loss": 8.7501,
      "step": 192
    },
    {
      "epoch": 0.3282312925170068,
      "grad_norm": 3.8504488468170166,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 8.2466,
      "step": 193
    },
    {
      "epoch": 0.3299319727891156,
      "grad_norm": 3.989088296890259,
      "learning_rate": 2.458548727494292e-07,
      "loss": 8.4433,
      "step": 194
    },
    {
      "epoch": 0.33163265306122447,
      "grad_norm": 4.311855316162109,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 8.5139,
      "step": 195
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 4.428941249847412,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 8.403,
      "step": 196
    },
    {
      "epoch": 0.33503401360544216,
      "grad_norm": 4.9349284172058105,
      "learning_rate": 6.150154258476315e-08,
      "loss": 8.7689,
      "step": 197
    },
    {
      "epoch": 0.336734693877551,
      "grad_norm": 5.077785968780518,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 8.8967,
      "step": 198
    },
    {
      "epoch": 0.33843537414965985,
      "grad_norm": 5.027740955352783,
      "learning_rate": 6.834750376549792e-09,
      "loss": 8.4149,
      "step": 199
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 5.754498481750488,
      "learning_rate": 0.0,
      "loss": 8.9504,
      "step": 200
    },
    {
      "epoch": 0.3401360544217687,
      "eval_loss": 2.1015098094940186,
      "eval_runtime": 167.6677,
      "eval_samples_per_second": 5.905,
      "eval_steps_per_second": 1.479,
      "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": 4.318036492120228e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}