{
  "best_metric": 1.2767974138259888,
  "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": 24.19337272644043,
      "learning_rate": 1e-05,
      "loss": 5.4753,
      "step": 1
    },
    {
      "epoch": 0.0017006802721088435,
      "eval_loss": 1.5546903610229492,
      "eval_runtime": 11.6033,
      "eval_samples_per_second": 85.321,
      "eval_steps_per_second": 21.373,
      "step": 1
    },
    {
      "epoch": 0.003401360544217687,
      "grad_norm": 25.985258102416992,
      "learning_rate": 2e-05,
      "loss": 5.4607,
      "step": 2
    },
    {
      "epoch": 0.00510204081632653,
      "grad_norm": 27.517133712768555,
      "learning_rate": 3e-05,
      "loss": 5.9039,
      "step": 3
    },
    {
      "epoch": 0.006802721088435374,
      "grad_norm": 20.81531524658203,
      "learning_rate": 4e-05,
      "loss": 5.8029,
      "step": 4
    },
    {
      "epoch": 0.008503401360544218,
      "grad_norm": 22.034828186035156,
      "learning_rate": 5e-05,
      "loss": 5.9363,
      "step": 5
    },
    {
      "epoch": 0.01020408163265306,
      "grad_norm": 24.3472843170166,
      "learning_rate": 6e-05,
      "loss": 6.0183,
      "step": 6
    },
    {
      "epoch": 0.011904761904761904,
      "grad_norm": 22.67211151123047,
      "learning_rate": 7e-05,
      "loss": 5.9568,
      "step": 7
    },
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 27.72193717956543,
      "learning_rate": 8e-05,
      "loss": 5.6224,
      "step": 8
    },
    {
      "epoch": 0.015306122448979591,
      "grad_norm": 24.595985412597656,
      "learning_rate": 9e-05,
      "loss": 6.148,
      "step": 9
    },
    {
      "epoch": 0.017006802721088437,
      "grad_norm": 26.96007537841797,
      "learning_rate": 0.0001,
      "loss": 6.1101,
      "step": 10
    },
    {
      "epoch": 0.01870748299319728,
      "grad_norm": 27.346511840820312,
      "learning_rate": 9.999316524962345e-05,
      "loss": 6.0955,
      "step": 11
    },
    {
      "epoch": 0.02040816326530612,
      "grad_norm": 25.942846298217773,
      "learning_rate": 9.997266286704631e-05,
      "loss": 6.3446,
      "step": 12
    },
    {
      "epoch": 0.022108843537414966,
      "grad_norm": 29.56608772277832,
      "learning_rate": 9.993849845741524e-05,
      "loss": 6.0699,
      "step": 13
    },
    {
      "epoch": 0.023809523809523808,
      "grad_norm": 30.475862503051758,
      "learning_rate": 9.989068136093873e-05,
      "loss": 6.0655,
      "step": 14
    },
    {
      "epoch": 0.025510204081632654,
      "grad_norm": 31.61933135986328,
      "learning_rate": 9.98292246503335e-05,
      "loss": 6.0643,
      "step": 15
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 23.196922302246094,
      "learning_rate": 9.975414512725057e-05,
      "loss": 6.2036,
      "step": 16
    },
    {
      "epoch": 0.02891156462585034,
      "grad_norm": 29.405311584472656,
      "learning_rate": 9.966546331768191e-05,
      "loss": 5.8484,
      "step": 17
    },
    {
      "epoch": 0.030612244897959183,
      "grad_norm": 30.231185913085938,
      "learning_rate": 9.956320346634876e-05,
      "loss": 6.1691,
      "step": 18
    },
    {
      "epoch": 0.03231292517006803,
      "grad_norm": 28.007179260253906,
      "learning_rate": 9.944739353007344e-05,
      "loss": 5.9114,
      "step": 19
    },
    {
      "epoch": 0.034013605442176874,
      "grad_norm": 27.24520492553711,
      "learning_rate": 9.931806517013612e-05,
      "loss": 5.8293,
      "step": 20
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 25.410097122192383,
      "learning_rate": 9.917525374361912e-05,
      "loss": 6.3071,
      "step": 21
    },
    {
      "epoch": 0.03741496598639456,
      "grad_norm": 29.187780380249023,
      "learning_rate": 9.901899829374047e-05,
      "loss": 6.1117,
      "step": 22
    },
    {
      "epoch": 0.0391156462585034,
      "grad_norm": 27.772035598754883,
      "learning_rate": 9.884934153917997e-05,
      "loss": 6.1539,
      "step": 23
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 27.033891677856445,
      "learning_rate": 9.86663298624003e-05,
      "loss": 5.8139,
      "step": 24
    },
    {
      "epoch": 0.04251700680272109,
      "grad_norm": 33.05968475341797,
      "learning_rate": 9.847001329696653e-05,
      "loss": 5.9564,
      "step": 25
    },
    {
      "epoch": 0.04421768707482993,
      "grad_norm": 31.614299774169922,
      "learning_rate": 9.826044551386744e-05,
      "loss": 5.7707,
      "step": 26
    },
    {
      "epoch": 0.04591836734693878,
      "grad_norm": 25.837799072265625,
      "learning_rate": 9.803768380684242e-05,
      "loss": 5.6367,
      "step": 27
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 28.774608612060547,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.8827,
      "step": 28
    },
    {
      "epoch": 0.04931972789115646,
      "grad_norm": 26.06644630432129,
      "learning_rate": 9.755282581475769e-05,
      "loss": 5.7789,
      "step": 29
    },
    {
      "epoch": 0.05102040816326531,
      "grad_norm": 20.919469833374023,
      "learning_rate": 9.729086208503174e-05,
      "loss": 6.0897,
      "step": 30
    },
    {
      "epoch": 0.05272108843537415,
      "grad_norm": 24.49297332763672,
      "learning_rate": 9.701596950580806e-05,
      "loss": 5.6788,
      "step": 31
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 21.995454788208008,
      "learning_rate": 9.672822322997305e-05,
      "loss": 5.7556,
      "step": 32
    },
    {
      "epoch": 0.05612244897959184,
      "grad_norm": 24.823959350585938,
      "learning_rate": 9.642770192448536e-05,
      "loss": 5.6888,
      "step": 33
    },
    {
      "epoch": 0.05782312925170068,
      "grad_norm": 24.168792724609375,
      "learning_rate": 9.611448774886924e-05,
      "loss": 5.5849,
      "step": 34
    },
    {
      "epoch": 0.05952380952380952,
      "grad_norm": 26.622600555419922,
      "learning_rate": 9.578866633275288e-05,
      "loss": 5.6546,
      "step": 35
    },
    {
      "epoch": 0.061224489795918366,
      "grad_norm": 23.491342544555664,
      "learning_rate": 9.545032675245813e-05,
      "loss": 5.8027,
      "step": 36
    },
    {
      "epoch": 0.06292517006802721,
      "grad_norm": 29.494272232055664,
      "learning_rate": 9.509956150664796e-05,
      "loss": 5.7836,
      "step": 37
    },
    {
      "epoch": 0.06462585034013606,
      "grad_norm": 29.72945213317871,
      "learning_rate": 9.473646649103818e-05,
      "loss": 5.4765,
      "step": 38
    },
    {
      "epoch": 0.0663265306122449,
      "grad_norm": 26.936437606811523,
      "learning_rate": 9.43611409721806e-05,
      "loss": 5.5969,
      "step": 39
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 28.58720588684082,
      "learning_rate": 9.397368756032445e-05,
      "loss": 5.3136,
      "step": 40
    },
    {
      "epoch": 0.06972789115646258,
      "grad_norm": 34.12704849243164,
      "learning_rate": 9.357421218136386e-05,
      "loss": 5.5265,
      "step": 41
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 26.306472778320312,
      "learning_rate": 9.316282404787871e-05,
      "loss": 5.0378,
      "step": 42
    },
    {
      "epoch": 0.07312925170068027,
      "grad_norm": 29.366565704345703,
      "learning_rate": 9.273963562927695e-05,
      "loss": 5.3989,
      "step": 43
    },
    {
      "epoch": 0.07482993197278912,
      "grad_norm": 35.158058166503906,
      "learning_rate": 9.230476262104677e-05,
      "loss": 5.6153,
      "step": 44
    },
    {
      "epoch": 0.07653061224489796,
      "grad_norm": 29.696569442749023,
      "learning_rate": 9.185832391312644e-05,
      "loss": 5.3015,
      "step": 45
    },
    {
      "epoch": 0.0782312925170068,
      "grad_norm": 31.005678176879883,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.9801,
      "step": 46
    },
    {
      "epoch": 0.07993197278911565,
      "grad_norm": 32.18037796020508,
      "learning_rate": 9.093124073433463e-05,
      "loss": 5.0736,
      "step": 47
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 35.47964859008789,
      "learning_rate": 9.045084971874738e-05,
      "loss": 5.1036,
      "step": 48
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 33.22696304321289,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.932,
      "step": 49
    },
    {
      "epoch": 0.08503401360544217,
      "grad_norm": 38.61347961425781,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.9688,
      "step": 50
    },
    {
      "epoch": 0.08503401360544217,
      "eval_loss": 1.3801863193511963,
      "eval_runtime": 11.5991,
      "eval_samples_per_second": 85.351,
      "eval_steps_per_second": 21.381,
      "step": 50
    },
    {
      "epoch": 0.08673469387755102,
      "grad_norm": 31.136253356933594,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.8852,
      "step": 51
    },
    {
      "epoch": 0.08843537414965986,
      "grad_norm": 23.9685001373291,
      "learning_rate": 8.842005554284296e-05,
      "loss": 5.2545,
      "step": 52
    },
    {
      "epoch": 0.09013605442176871,
      "grad_norm": 26.573095321655273,
      "learning_rate": 8.788574348801675e-05,
      "loss": 5.4582,
      "step": 53
    },
    {
      "epoch": 0.09183673469387756,
      "grad_norm": 23.158388137817383,
      "learning_rate": 8.73410738492077e-05,
      "loss": 5.0829,
      "step": 54
    },
    {
      "epoch": 0.0935374149659864,
      "grad_norm": 24.976308822631836,
      "learning_rate": 8.678619553365659e-05,
      "loss": 5.4187,
      "step": 55
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 21.470502853393555,
      "learning_rate": 8.622126023955446e-05,
      "loss": 5.3496,
      "step": 56
    },
    {
      "epoch": 0.09693877551020408,
      "grad_norm": 20.489885330200195,
      "learning_rate": 8.564642241456986e-05,
      "loss": 5.4574,
      "step": 57
    },
    {
      "epoch": 0.09863945578231292,
      "grad_norm": 20.27273941040039,
      "learning_rate": 8.506183921362443e-05,
      "loss": 5.7436,
      "step": 58
    },
    {
      "epoch": 0.10034013605442177,
      "grad_norm": 22.73259162902832,
      "learning_rate": 8.44676704559283e-05,
      "loss": 5.878,
      "step": 59
    },
    {
      "epoch": 0.10204081632653061,
      "grad_norm": 19.459606170654297,
      "learning_rate": 8.386407858128706e-05,
      "loss": 5.6666,
      "step": 60
    },
    {
      "epoch": 0.10374149659863946,
      "grad_norm": 17.590333938598633,
      "learning_rate": 8.32512286056924e-05,
      "loss": 5.6417,
      "step": 61
    },
    {
      "epoch": 0.1054421768707483,
      "grad_norm": 19.19861602783203,
      "learning_rate": 8.262928807620843e-05,
      "loss": 5.502,
      "step": 62
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 17.078609466552734,
      "learning_rate": 8.199842702516583e-05,
      "loss": 5.5597,
      "step": 63
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 19.623933792114258,
      "learning_rate": 8.135881792367686e-05,
      "loss": 5.7327,
      "step": 64
    },
    {
      "epoch": 0.11054421768707483,
      "grad_norm": 19.144752502441406,
      "learning_rate": 8.07106356344834e-05,
      "loss": 5.6978,
      "step": 65
    },
    {
      "epoch": 0.11224489795918367,
      "grad_norm": 19.583158493041992,
      "learning_rate": 8.005405736415126e-05,
      "loss": 5.738,
      "step": 66
    },
    {
      "epoch": 0.11394557823129252,
      "grad_norm": 19.113208770751953,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.4139,
      "step": 67
    },
    {
      "epoch": 0.11564625850340136,
      "grad_norm": 17.07074737548828,
      "learning_rate": 7.871643313414718e-05,
      "loss": 5.4618,
      "step": 68
    },
    {
      "epoch": 0.11734693877551021,
      "grad_norm": 20.1825008392334,
      "learning_rate": 7.803575286758364e-05,
      "loss": 5.7374,
      "step": 69
    },
    {
      "epoch": 0.11904761904761904,
      "grad_norm": 20.374326705932617,
      "learning_rate": 7.734740790612136e-05,
      "loss": 5.4432,
      "step": 70
    },
    {
      "epoch": 0.12074829931972789,
      "grad_norm": 19.889938354492188,
      "learning_rate": 7.66515864363997e-05,
      "loss": 5.6169,
      "step": 71
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 23.016969680786133,
      "learning_rate": 7.594847868906076e-05,
      "loss": 5.8109,
      "step": 72
    },
    {
      "epoch": 0.12414965986394558,
      "grad_norm": 22.365962982177734,
      "learning_rate": 7.52382768867422e-05,
      "loss": 5.3963,
      "step": 73
    },
    {
      "epoch": 0.12585034013605442,
      "grad_norm": 19.952550888061523,
      "learning_rate": 7.452117519152542e-05,
      "loss": 5.8411,
      "step": 74
    },
    {
      "epoch": 0.12755102040816327,
      "grad_norm": 24.070985794067383,
      "learning_rate": 7.379736965185368e-05,
      "loss": 5.6335,
      "step": 75
    },
    {
      "epoch": 0.1292517006802721,
      "grad_norm": 18.756494522094727,
      "learning_rate": 7.30670581489344e-05,
      "loss": 5.5206,
      "step": 76
    },
    {
      "epoch": 0.13095238095238096,
      "grad_norm": 19.78859519958496,
      "learning_rate": 7.233044034264034e-05,
      "loss": 5.5298,
      "step": 77
    },
    {
      "epoch": 0.1326530612244898,
      "grad_norm": 17.80217742919922,
      "learning_rate": 7.158771761692464e-05,
      "loss": 5.5467,
      "step": 78
    },
    {
      "epoch": 0.13435374149659865,
      "grad_norm": 19.379749298095703,
      "learning_rate": 7.083909302476453e-05,
      "loss": 5.5862,
      "step": 79
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 16.74873924255371,
      "learning_rate": 7.008477123264848e-05,
      "loss": 5.3915,
      "step": 80
    },
    {
      "epoch": 0.1377551020408163,
      "grad_norm": 18.661653518676758,
      "learning_rate": 6.932495846462261e-05,
      "loss": 5.4581,
      "step": 81
    },
    {
      "epoch": 0.13945578231292516,
      "grad_norm": 15.961063385009766,
      "learning_rate": 6.855986244591104e-05,
      "loss": 5.1376,
      "step": 82
    },
    {
      "epoch": 0.141156462585034,
      "grad_norm": 14.398187637329102,
      "learning_rate": 6.778969234612584e-05,
      "loss": 5.3022,
      "step": 83
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 15.740119934082031,
      "learning_rate": 6.701465872208216e-05,
      "loss": 5.2612,
      "step": 84
    },
    {
      "epoch": 0.1445578231292517,
      "grad_norm": 16.52437973022461,
      "learning_rate": 6.623497346023418e-05,
      "loss": 5.1501,
      "step": 85
    },
    {
      "epoch": 0.14625850340136054,
      "grad_norm": 18.607925415039062,
      "learning_rate": 6.545084971874738e-05,
      "loss": 5.147,
      "step": 86
    },
    {
      "epoch": 0.14795918367346939,
      "grad_norm": 18.163801193237305,
      "learning_rate": 6.466250186922325e-05,
      "loss": 5.3176,
      "step": 87
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 18.678234100341797,
      "learning_rate": 6.387014543809223e-05,
      "loss": 5.1242,
      "step": 88
    },
    {
      "epoch": 0.15136054421768708,
      "grad_norm": 19.111909866333008,
      "learning_rate": 6.307399704769099e-05,
      "loss": 5.2016,
      "step": 89
    },
    {
      "epoch": 0.15306122448979592,
      "grad_norm": 21.934247970581055,
      "learning_rate": 6.227427435703997e-05,
      "loss": 5.0627,
      "step": 90
    },
    {
      "epoch": 0.15476190476190477,
      "grad_norm": 26.169151306152344,
      "learning_rate": 6.147119600233758e-05,
      "loss": 5.0118,
      "step": 91
    },
    {
      "epoch": 0.1564625850340136,
      "grad_norm": 26.868295669555664,
      "learning_rate": 6.066498153718735e-05,
      "loss": 5.2222,
      "step": 92
    },
    {
      "epoch": 0.15816326530612246,
      "grad_norm": 29.271276473999023,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.9448,
      "step": 93
    },
    {
      "epoch": 0.1598639455782313,
      "grad_norm": 33.750572204589844,
      "learning_rate": 5.90440267166055e-05,
      "loss": 5.093,
      "step": 94
    },
    {
      "epoch": 0.16156462585034015,
      "grad_norm": 33.64457321166992,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.7788,
      "step": 95
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 32.394657135009766,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.7954,
      "step": 96
    },
    {
      "epoch": 0.1649659863945578,
      "grad_norm": 34.656742095947266,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 4.7812,
      "step": 97
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 46.07516860961914,
      "learning_rate": 5.577423184847932e-05,
      "loss": 4.9881,
      "step": 98
    },
    {
      "epoch": 0.1683673469387755,
      "grad_norm": 38.203487396240234,
      "learning_rate": 5.495227651252315e-05,
      "loss": 5.0178,
      "step": 99
    },
    {
      "epoch": 0.17006802721088435,
      "grad_norm": 51.48576736450195,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.8706,
      "step": 100
    },
    {
      "epoch": 0.17006802721088435,
      "eval_loss": 1.350069284439087,
      "eval_runtime": 11.5413,
      "eval_samples_per_second": 85.779,
      "eval_steps_per_second": 21.488,
      "step": 100
    },
    {
      "epoch": 0.1717687074829932,
      "grad_norm": 24.463096618652344,
      "learning_rate": 5.330452921628497e-05,
      "loss": 5.0976,
      "step": 101
    },
    {
      "epoch": 0.17346938775510204,
      "grad_norm": 23.088424682617188,
      "learning_rate": 5.247918773366112e-05,
      "loss": 5.208,
      "step": 102
    },
    {
      "epoch": 0.17517006802721088,
      "grad_norm": 23.822874069213867,
      "learning_rate": 5.165316846586541e-05,
      "loss": 5.3821,
      "step": 103
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 28.965757369995117,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 5.6137,
      "step": 104
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 24.0604248046875,
      "learning_rate": 5e-05,
      "loss": 5.5528,
      "step": 105
    },
    {
      "epoch": 0.18027210884353742,
      "grad_norm": 22.2089900970459,
      "learning_rate": 4.917330276168208e-05,
      "loss": 5.5119,
      "step": 106
    },
    {
      "epoch": 0.18197278911564627,
      "grad_norm": 20.267610549926758,
      "learning_rate": 4.834683153413459e-05,
      "loss": 5.3559,
      "step": 107
    },
    {
      "epoch": 0.1836734693877551,
      "grad_norm": 24.24986457824707,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 5.5959,
      "step": 108
    },
    {
      "epoch": 0.18537414965986396,
      "grad_norm": 18.987770080566406,
      "learning_rate": 4.669547078371504e-05,
      "loss": 5.6555,
      "step": 109
    },
    {
      "epoch": 0.1870748299319728,
      "grad_norm": 18.21375846862793,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 5.385,
      "step": 110
    },
    {
      "epoch": 0.18877551020408162,
      "grad_norm": 18.386709213256836,
      "learning_rate": 4.504772348747687e-05,
      "loss": 5.5589,
      "step": 111
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 19.097965240478516,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 5.6853,
      "step": 112
    },
    {
      "epoch": 0.1921768707482993,
      "grad_norm": 16.18609046936035,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 5.4967,
      "step": 113
    },
    {
      "epoch": 0.19387755102040816,
      "grad_norm": 17.791534423828125,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 5.5527,
      "step": 114
    },
    {
      "epoch": 0.195578231292517,
      "grad_norm": 16.426427841186523,
      "learning_rate": 4.17702704859633e-05,
      "loss": 5.6107,
      "step": 115
    },
    {
      "epoch": 0.19727891156462585,
      "grad_norm": 18.30120086669922,
      "learning_rate": 4.095597328339452e-05,
      "loss": 5.7096,
      "step": 116
    },
    {
      "epoch": 0.1989795918367347,
      "grad_norm": 17.433568954467773,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 5.5259,
      "step": 117
    },
    {
      "epoch": 0.20068027210884354,
      "grad_norm": 16.40717315673828,
      "learning_rate": 3.933501846281267e-05,
      "loss": 5.6334,
      "step": 118
    },
    {
      "epoch": 0.20238095238095238,
      "grad_norm": 14.775606155395508,
      "learning_rate": 3.852880399766243e-05,
      "loss": 5.3222,
      "step": 119
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 14.336289405822754,
      "learning_rate": 3.772572564296005e-05,
      "loss": 5.3899,
      "step": 120
    },
    {
      "epoch": 0.20578231292517007,
      "grad_norm": 16.73828125,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 5.4498,
      "step": 121
    },
    {
      "epoch": 0.20748299319727892,
      "grad_norm": 17.087566375732422,
      "learning_rate": 3.612985456190778e-05,
      "loss": 5.4449,
      "step": 122
    },
    {
      "epoch": 0.20918367346938777,
      "grad_norm": 15.058612823486328,
      "learning_rate": 3.533749813077677e-05,
      "loss": 5.5024,
      "step": 123
    },
    {
      "epoch": 0.2108843537414966,
      "grad_norm": 13.776179313659668,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 5.6169,
      "step": 124
    },
    {
      "epoch": 0.21258503401360543,
      "grad_norm": 19.452163696289062,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 5.3823,
      "step": 125
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 15.750632286071777,
      "learning_rate": 3.298534127791785e-05,
      "loss": 5.3804,
      "step": 126
    },
    {
      "epoch": 0.21598639455782312,
      "grad_norm": 17.494735717773438,
      "learning_rate": 3.221030765387417e-05,
      "loss": 5.3895,
      "step": 127
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 17.02010726928711,
      "learning_rate": 3.144013755408895e-05,
      "loss": 5.4849,
      "step": 128
    },
    {
      "epoch": 0.2193877551020408,
      "grad_norm": 16.054283142089844,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 5.2059,
      "step": 129
    },
    {
      "epoch": 0.22108843537414966,
      "grad_norm": 17.07164764404297,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.2934,
      "step": 130
    },
    {
      "epoch": 0.2227891156462585,
      "grad_norm": 15.970010757446289,
      "learning_rate": 2.916090697523549e-05,
      "loss": 5.1384,
      "step": 131
    },
    {
      "epoch": 0.22448979591836735,
      "grad_norm": 16.50423240661621,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 5.1937,
      "step": 132
    },
    {
      "epoch": 0.2261904761904762,
      "grad_norm": 18.0500545501709,
      "learning_rate": 2.766955965735968e-05,
      "loss": 5.3441,
      "step": 133
    },
    {
      "epoch": 0.22789115646258504,
      "grad_norm": 16.110912322998047,
      "learning_rate": 2.693294185106562e-05,
      "loss": 5.5183,
      "step": 134
    },
    {
      "epoch": 0.22959183673469388,
      "grad_norm": 23.673166275024414,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 5.3588,
      "step": 135
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 16.812118530273438,
      "learning_rate": 2.547882480847461e-05,
      "loss": 5.3661,
      "step": 136
    },
    {
      "epoch": 0.23299319727891157,
      "grad_norm": 17.158586502075195,
      "learning_rate": 2.476172311325783e-05,
      "loss": 5.0421,
      "step": 137
    },
    {
      "epoch": 0.23469387755102042,
      "grad_norm": 17.621150970458984,
      "learning_rate": 2.405152131093926e-05,
      "loss": 4.6079,
      "step": 138
    },
    {
      "epoch": 0.23639455782312926,
      "grad_norm": 18.104022979736328,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.8402,
      "step": 139
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 18.43425750732422,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 5.1086,
      "step": 140
    },
    {
      "epoch": 0.23979591836734693,
      "grad_norm": 19.870988845825195,
      "learning_rate": 2.196424713241637e-05,
      "loss": 4.6563,
      "step": 141
    },
    {
      "epoch": 0.24149659863945577,
      "grad_norm": 22.2313232421875,
      "learning_rate": 2.128356686585282e-05,
      "loss": 5.0495,
      "step": 142
    },
    {
      "epoch": 0.24319727891156462,
      "grad_norm": 16.598520278930664,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.633,
      "step": 143
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 16.166664123535156,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.8576,
      "step": 144
    },
    {
      "epoch": 0.2465986394557823,
      "grad_norm": 24.100725173950195,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.8387,
      "step": 145
    },
    {
      "epoch": 0.24829931972789115,
      "grad_norm": 19.560876846313477,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 4.7812,
      "step": 146
    },
    {
      "epoch": 0.25,
      "grad_norm": 22.26222038269043,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.8497,
      "step": 147
    },
    {
      "epoch": 0.25170068027210885,
      "grad_norm": 21.519262313842773,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.6968,
      "step": 148
    },
    {
      "epoch": 0.2534013605442177,
      "grad_norm": 21.677261352539062,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 4.4486,
      "step": 149
    },
    {
      "epoch": 0.25510204081632654,
      "grad_norm": 28.465845108032227,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.6025,
      "step": 150
    },
    {
      "epoch": 0.25510204081632654,
      "eval_loss": 1.2885631322860718,
      "eval_runtime": 11.6028,
      "eval_samples_per_second": 85.324,
      "eval_steps_per_second": 21.374,
      "step": 150
    },
    {
      "epoch": 0.2568027210884354,
      "grad_norm": 15.359907150268555,
      "learning_rate": 1.553232954407171e-05,
      "loss": 4.8938,
      "step": 151
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 15.208480834960938,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 5.062,
      "step": 152
    },
    {
      "epoch": 0.2602040816326531,
      "grad_norm": 17.505807876586914,
      "learning_rate": 1.435357758543015e-05,
      "loss": 5.369,
      "step": 153
    },
    {
      "epoch": 0.2619047619047619,
      "grad_norm": 15.413578033447266,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 5.2143,
      "step": 154
    },
    {
      "epoch": 0.26360544217687076,
      "grad_norm": 18.26093292236328,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 5.4455,
      "step": 155
    },
    {
      "epoch": 0.2653061224489796,
      "grad_norm": 17.461668014526367,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 5.2845,
      "step": 156
    },
    {
      "epoch": 0.26700680272108845,
      "grad_norm": 18.13603401184082,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 5.4194,
      "step": 157
    },
    {
      "epoch": 0.2687074829931973,
      "grad_norm": 19.375347137451172,
      "learning_rate": 1.157994445715706e-05,
      "loss": 5.3467,
      "step": 158
    },
    {
      "epoch": 0.27040816326530615,
      "grad_norm": 18.652050018310547,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 5.5405,
      "step": 159
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 15.435720443725586,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 5.328,
      "step": 160
    },
    {
      "epoch": 0.27380952380952384,
      "grad_norm": 21.76569175720215,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 5.4896,
      "step": 161
    },
    {
      "epoch": 0.2755102040816326,
      "grad_norm": 16.876432418823242,
      "learning_rate": 9.549150281252633e-06,
      "loss": 5.6265,
      "step": 162
    },
    {
      "epoch": 0.27721088435374147,
      "grad_norm": 16.61174201965332,
      "learning_rate": 9.068759265665384e-06,
      "loss": 5.7207,
      "step": 163
    },
    {
      "epoch": 0.2789115646258503,
      "grad_norm": 19.416622161865234,
      "learning_rate": 8.599558442598998e-06,
      "loss": 5.232,
      "step": 164
    },
    {
      "epoch": 0.28061224489795916,
      "grad_norm": 16.572818756103516,
      "learning_rate": 8.141676086873572e-06,
      "loss": 5.4104,
      "step": 165
    },
    {
      "epoch": 0.282312925170068,
      "grad_norm": 16.48833465576172,
      "learning_rate": 7.695237378953223e-06,
      "loss": 5.6902,
      "step": 166
    },
    {
      "epoch": 0.28401360544217685,
      "grad_norm": 17.595455169677734,
      "learning_rate": 7.260364370723044e-06,
      "loss": 5.2924,
      "step": 167
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 18.19123077392578,
      "learning_rate": 6.837175952121306e-06,
      "loss": 5.4676,
      "step": 168
    },
    {
      "epoch": 0.28741496598639454,
      "grad_norm": 16.333417892456055,
      "learning_rate": 6.425787818636131e-06,
      "loss": 5.3097,
      "step": 169
    },
    {
      "epoch": 0.2891156462585034,
      "grad_norm": 16.5438175201416,
      "learning_rate": 6.026312439675552e-06,
      "loss": 5.487,
      "step": 170
    },
    {
      "epoch": 0.29081632653061223,
      "grad_norm": 15.9813232421875,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 5.3458,
      "step": 171
    },
    {
      "epoch": 0.2925170068027211,
      "grad_norm": 16.760990142822266,
      "learning_rate": 5.263533508961827e-06,
      "loss": 5.3999,
      "step": 172
    },
    {
      "epoch": 0.2942176870748299,
      "grad_norm": 15.316781044006348,
      "learning_rate": 4.900438493352055e-06,
      "loss": 5.4696,
      "step": 173
    },
    {
      "epoch": 0.29591836734693877,
      "grad_norm": 18.289152145385742,
      "learning_rate": 4.549673247541875e-06,
      "loss": 5.5471,
      "step": 174
    },
    {
      "epoch": 0.2976190476190476,
      "grad_norm": 16.778696060180664,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 5.5152,
      "step": 175
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 16.479299545288086,
      "learning_rate": 3.885512251130763e-06,
      "loss": 5.4456,
      "step": 176
    },
    {
      "epoch": 0.3010204081632653,
      "grad_norm": 20.302858352661133,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 5.3544,
      "step": 177
    },
    {
      "epoch": 0.30272108843537415,
      "grad_norm": 17.077611923217773,
      "learning_rate": 3.271776770026963e-06,
      "loss": 5.4747,
      "step": 178
    },
    {
      "epoch": 0.304421768707483,
      "grad_norm": 15.88514232635498,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 5.2525,
      "step": 179
    },
    {
      "epoch": 0.30612244897959184,
      "grad_norm": 18.785037994384766,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 5.2617,
      "step": 180
    },
    {
      "epoch": 0.3078231292517007,
      "grad_norm": 14.548587799072266,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 5.4835,
      "step": 181
    },
    {
      "epoch": 0.30952380952380953,
      "grad_norm": 15.249629974365234,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 5.1384,
      "step": 182
    },
    {
      "epoch": 0.3112244897959184,
      "grad_norm": 17.987117767333984,
      "learning_rate": 1.962316193157593e-06,
      "loss": 4.9743,
      "step": 183
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 14.752166748046875,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 5.1736,
      "step": 184
    },
    {
      "epoch": 0.31462585034013607,
      "grad_norm": 14.412027359008789,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 5.0739,
      "step": 185
    },
    {
      "epoch": 0.3163265306122449,
      "grad_norm": 16.77324104309082,
      "learning_rate": 1.333670137599713e-06,
      "loss": 5.1431,
      "step": 186
    },
    {
      "epoch": 0.31802721088435376,
      "grad_norm": 17.276731491088867,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 5.1948,
      "step": 187
    },
    {
      "epoch": 0.3197278911564626,
      "grad_norm": 14.923358917236328,
      "learning_rate": 9.810017062595322e-07,
      "loss": 4.8424,
      "step": 188
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 16.548255920410156,
      "learning_rate": 8.247462563808817e-07,
      "loss": 4.9962,
      "step": 189
    },
    {
      "epoch": 0.3231292517006803,
      "grad_norm": 16.286285400390625,
      "learning_rate": 6.819348298638839e-07,
      "loss": 4.9019,
      "step": 190
    },
    {
      "epoch": 0.32482993197278914,
      "grad_norm": 16.17375373840332,
      "learning_rate": 5.526064699265753e-07,
      "loss": 4.9196,
      "step": 191
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 16.75802230834961,
      "learning_rate": 4.367965336512403e-07,
      "loss": 4.3727,
      "step": 192
    },
    {
      "epoch": 0.3282312925170068,
      "grad_norm": 16.817415237426758,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 4.6064,
      "step": 193
    },
    {
      "epoch": 0.3299319727891156,
      "grad_norm": 25.76260757446289,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.6466,
      "step": 194
    },
    {
      "epoch": 0.33163265306122447,
      "grad_norm": 16.99913787841797,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.3759,
      "step": 195
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 16.4222354888916,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.4776,
      "step": 196
    },
    {
      "epoch": 0.33503401360544216,
      "grad_norm": 17.989337921142578,
      "learning_rate": 6.150154258476315e-08,
      "loss": 4.501,
      "step": 197
    },
    {
      "epoch": 0.336734693877551,
      "grad_norm": 18.40859603881836,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.6155,
      "step": 198
    },
    {
      "epoch": 0.33843537414965985,
      "grad_norm": 21.46599006652832,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.8029,
      "step": 199
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 35.6815071105957,
      "learning_rate": 0.0,
      "loss": 4.5153,
      "step": 200
    },
    {
      "epoch": 0.3401360544217687,
      "eval_loss": 1.2767974138259888,
      "eval_runtime": 11.5982,
      "eval_samples_per_second": 85.358,
      "eval_steps_per_second": 21.383,
      "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": 1.4976948111409152e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}