{
  "best_metric": 2.9099483489990234,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3766478342749529,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0018832391713747645,
      "grad_norm": 2.8433897495269775,
      "learning_rate": 1e-05,
      "loss": 3.1762,
      "step": 1
    },
    {
      "epoch": 0.0018832391713747645,
      "eval_loss": 4.379508018493652,
      "eval_runtime": 82.8028,
      "eval_samples_per_second": 10.809,
      "eval_steps_per_second": 2.705,
      "step": 1
    },
    {
      "epoch": 0.003766478342749529,
      "grad_norm": 2.323256254196167,
      "learning_rate": 2e-05,
      "loss": 3.401,
      "step": 2
    },
    {
      "epoch": 0.005649717514124294,
      "grad_norm": 2.690786600112915,
      "learning_rate": 3e-05,
      "loss": 3.4446,
      "step": 3
    },
    {
      "epoch": 0.007532956685499058,
      "grad_norm": 2.3593900203704834,
      "learning_rate": 4e-05,
      "loss": 3.4328,
      "step": 4
    },
    {
      "epoch": 0.009416195856873822,
      "grad_norm": 1.8206570148468018,
      "learning_rate": 5e-05,
      "loss": 3.5544,
      "step": 5
    },
    {
      "epoch": 0.011299435028248588,
      "grad_norm": 1.5169888734817505,
      "learning_rate": 6e-05,
      "loss": 3.3046,
      "step": 6
    },
    {
      "epoch": 0.013182674199623353,
      "grad_norm": 1.9192094802856445,
      "learning_rate": 7e-05,
      "loss": 3.325,
      "step": 7
    },
    {
      "epoch": 0.015065913370998116,
      "grad_norm": 1.4936283826828003,
      "learning_rate": 8e-05,
      "loss": 3.4199,
      "step": 8
    },
    {
      "epoch": 0.01694915254237288,
      "grad_norm": 1.4846974611282349,
      "learning_rate": 9e-05,
      "loss": 3.4149,
      "step": 9
    },
    {
      "epoch": 0.018832391713747645,
      "grad_norm": 1.5412017107009888,
      "learning_rate": 0.0001,
      "loss": 3.2844,
      "step": 10
    },
    {
      "epoch": 0.02071563088512241,
      "grad_norm": 1.6269813776016235,
      "learning_rate": 9.999316524962345e-05,
      "loss": 3.402,
      "step": 11
    },
    {
      "epoch": 0.022598870056497175,
      "grad_norm": 2.0466113090515137,
      "learning_rate": 9.997266286704631e-05,
      "loss": 3.1431,
      "step": 12
    },
    {
      "epoch": 0.02448210922787194,
      "grad_norm": 1.5508036613464355,
      "learning_rate": 9.993849845741524e-05,
      "loss": 3.3318,
      "step": 13
    },
    {
      "epoch": 0.026365348399246705,
      "grad_norm": 1.4174089431762695,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.0651,
      "step": 14
    },
    {
      "epoch": 0.02824858757062147,
      "grad_norm": 1.5283914804458618,
      "learning_rate": 9.98292246503335e-05,
      "loss": 3.2137,
      "step": 15
    },
    {
      "epoch": 0.030131826741996232,
      "grad_norm": 1.5180612802505493,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.0092,
      "step": 16
    },
    {
      "epoch": 0.032015065913371,
      "grad_norm": 1.9057605266571045,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.3436,
      "step": 17
    },
    {
      "epoch": 0.03389830508474576,
      "grad_norm": 1.8316656351089478,
      "learning_rate": 9.956320346634876e-05,
      "loss": 3.0831,
      "step": 18
    },
    {
      "epoch": 0.035781544256120526,
      "grad_norm": 1.723305106163025,
      "learning_rate": 9.944739353007344e-05,
      "loss": 3.1291,
      "step": 19
    },
    {
      "epoch": 0.03766478342749529,
      "grad_norm": 1.768608570098877,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.9144,
      "step": 20
    },
    {
      "epoch": 0.03954802259887006,
      "grad_norm": 1.706481695175171,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.8884,
      "step": 21
    },
    {
      "epoch": 0.04143126177024482,
      "grad_norm": 1.553443193435669,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.8753,
      "step": 22
    },
    {
      "epoch": 0.04331450094161959,
      "grad_norm": 2.097810983657837,
      "learning_rate": 9.884934153917997e-05,
      "loss": 3.1215,
      "step": 23
    },
    {
      "epoch": 0.04519774011299435,
      "grad_norm": 1.7343829870224,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.9433,
      "step": 24
    },
    {
      "epoch": 0.047080979284369114,
      "grad_norm": 1.7442926168441772,
      "learning_rate": 9.847001329696653e-05,
      "loss": 3.0032,
      "step": 25
    },
    {
      "epoch": 0.04896421845574388,
      "grad_norm": 1.9430217742919922,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.9892,
      "step": 26
    },
    {
      "epoch": 0.05084745762711865,
      "grad_norm": 2.0032362937927246,
      "learning_rate": 9.803768380684242e-05,
      "loss": 3.0039,
      "step": 27
    },
    {
      "epoch": 0.05273069679849341,
      "grad_norm": 1.802256464958191,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.7603,
      "step": 28
    },
    {
      "epoch": 0.054613935969868174,
      "grad_norm": 1.8950775861740112,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.8566,
      "step": 29
    },
    {
      "epoch": 0.05649717514124294,
      "grad_norm": 1.7536348104476929,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.8173,
      "step": 30
    },
    {
      "epoch": 0.0583804143126177,
      "grad_norm": 1.7853409051895142,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.9015,
      "step": 31
    },
    {
      "epoch": 0.060263653483992465,
      "grad_norm": 1.743138313293457,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.7302,
      "step": 32
    },
    {
      "epoch": 0.062146892655367235,
      "grad_norm": 1.975548267364502,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.8849,
      "step": 33
    },
    {
      "epoch": 0.064030131826742,
      "grad_norm": 1.9865808486938477,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.7483,
      "step": 34
    },
    {
      "epoch": 0.06591337099811675,
      "grad_norm": 1.8429542779922485,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.9041,
      "step": 35
    },
    {
      "epoch": 0.06779661016949153,
      "grad_norm": 1.9406712055206299,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.8061,
      "step": 36
    },
    {
      "epoch": 0.0696798493408663,
      "grad_norm": 1.9041534662246704,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.9197,
      "step": 37
    },
    {
      "epoch": 0.07156308851224105,
      "grad_norm": 1.850974678993225,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.615,
      "step": 38
    },
    {
      "epoch": 0.07344632768361582,
      "grad_norm": 2.2933380603790283,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.8775,
      "step": 39
    },
    {
      "epoch": 0.07532956685499058,
      "grad_norm": 1.9177305698394775,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.8434,
      "step": 40
    },
    {
      "epoch": 0.07721280602636535,
      "grad_norm": 1.9171340465545654,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.4944,
      "step": 41
    },
    {
      "epoch": 0.07909604519774012,
      "grad_norm": 2.1136105060577393,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.4957,
      "step": 42
    },
    {
      "epoch": 0.08097928436911488,
      "grad_norm": 2.5157737731933594,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.6001,
      "step": 43
    },
    {
      "epoch": 0.08286252354048965,
      "grad_norm": 2.361225128173828,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.9205,
      "step": 44
    },
    {
      "epoch": 0.0847457627118644,
      "grad_norm": 3.3204503059387207,
      "learning_rate": 9.185832391312644e-05,
      "loss": 3.1481,
      "step": 45
    },
    {
      "epoch": 0.08662900188323917,
      "grad_norm": 2.994537830352783,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.9183,
      "step": 46
    },
    {
      "epoch": 0.08851224105461393,
      "grad_norm": 2.2447259426116943,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.9877,
      "step": 47
    },
    {
      "epoch": 0.0903954802259887,
      "grad_norm": 2.787060260772705,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.8197,
      "step": 48
    },
    {
      "epoch": 0.09227871939736347,
      "grad_norm": 2.2366034984588623,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.6518,
      "step": 49
    },
    {
      "epoch": 0.09416195856873823,
      "grad_norm": 2.402299404144287,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.8353,
      "step": 50
    },
    {
      "epoch": 0.09416195856873823,
      "eval_loss": 3.3923280239105225,
      "eval_runtime": 84.2895,
      "eval_samples_per_second": 10.618,
      "eval_steps_per_second": 2.658,
      "step": 50
    },
    {
      "epoch": 0.096045197740113,
      "grad_norm": 11.350125312805176,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.0026,
      "step": 51
    },
    {
      "epoch": 0.09792843691148775,
      "grad_norm": 6.891103267669678,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.5141,
      "step": 52
    },
    {
      "epoch": 0.09981167608286252,
      "grad_norm": 2.3344533443450928,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.3384,
      "step": 53
    },
    {
      "epoch": 0.1016949152542373,
      "grad_norm": 1.4267592430114746,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.3727,
      "step": 54
    },
    {
      "epoch": 0.10357815442561205,
      "grad_norm": 1.1526230573654175,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.1816,
      "step": 55
    },
    {
      "epoch": 0.10546139359698682,
      "grad_norm": 1.235979437828064,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.2251,
      "step": 56
    },
    {
      "epoch": 0.10734463276836158,
      "grad_norm": 1.0604145526885986,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.1276,
      "step": 57
    },
    {
      "epoch": 0.10922787193973635,
      "grad_norm": 1.084671139717102,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.1245,
      "step": 58
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 1.1658869981765747,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.3567,
      "step": 59
    },
    {
      "epoch": 0.11299435028248588,
      "grad_norm": 1.0910725593566895,
      "learning_rate": 8.386407858128706e-05,
      "loss": 3.0536,
      "step": 60
    },
    {
      "epoch": 0.11487758945386065,
      "grad_norm": 1.2034207582473755,
      "learning_rate": 8.32512286056924e-05,
      "loss": 3.1525,
      "step": 61
    },
    {
      "epoch": 0.1167608286252354,
      "grad_norm": 1.2226837873458862,
      "learning_rate": 8.262928807620843e-05,
      "loss": 3.1705,
      "step": 62
    },
    {
      "epoch": 0.11864406779661017,
      "grad_norm": 1.2016551494598389,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.9398,
      "step": 63
    },
    {
      "epoch": 0.12052730696798493,
      "grad_norm": 1.204183578491211,
      "learning_rate": 8.135881792367686e-05,
      "loss": 3.0235,
      "step": 64
    },
    {
      "epoch": 0.1224105461393597,
      "grad_norm": 1.2860832214355469,
      "learning_rate": 8.07106356344834e-05,
      "loss": 3.0672,
      "step": 65
    },
    {
      "epoch": 0.12429378531073447,
      "grad_norm": 1.2328332662582397,
      "learning_rate": 8.005405736415126e-05,
      "loss": 3.0955,
      "step": 66
    },
    {
      "epoch": 0.12617702448210924,
      "grad_norm": 1.6168620586395264,
      "learning_rate": 7.938926261462366e-05,
      "loss": 3.0177,
      "step": 67
    },
    {
      "epoch": 0.128060263653484,
      "grad_norm": 1.2841657400131226,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.7914,
      "step": 68
    },
    {
      "epoch": 0.12994350282485875,
      "grad_norm": 1.4681788682937622,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.8915,
      "step": 69
    },
    {
      "epoch": 0.1318267419962335,
      "grad_norm": 1.3547992706298828,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.847,
      "step": 70
    },
    {
      "epoch": 0.1337099811676083,
      "grad_norm": 1.4908517599105835,
      "learning_rate": 7.66515864363997e-05,
      "loss": 3.0859,
      "step": 71
    },
    {
      "epoch": 0.13559322033898305,
      "grad_norm": 1.3843392133712769,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.9605,
      "step": 72
    },
    {
      "epoch": 0.1374764595103578,
      "grad_norm": 1.5052849054336548,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.8122,
      "step": 73
    },
    {
      "epoch": 0.1393596986817326,
      "grad_norm": 1.432239294052124,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.0778,
      "step": 74
    },
    {
      "epoch": 0.14124293785310735,
      "grad_norm": 1.6122084856033325,
      "learning_rate": 7.379736965185368e-05,
      "loss": 3.0425,
      "step": 75
    },
    {
      "epoch": 0.1431261770244821,
      "grad_norm": 1.4387708902359009,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.8623,
      "step": 76
    },
    {
      "epoch": 0.14500941619585686,
      "grad_norm": 1.54473078250885,
      "learning_rate": 7.233044034264034e-05,
      "loss": 3.0656,
      "step": 77
    },
    {
      "epoch": 0.14689265536723164,
      "grad_norm": 1.8318265676498413,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.682,
      "step": 78
    },
    {
      "epoch": 0.1487758945386064,
      "grad_norm": 1.5669119358062744,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.9117,
      "step": 79
    },
    {
      "epoch": 0.15065913370998116,
      "grad_norm": 1.5926783084869385,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.7855,
      "step": 80
    },
    {
      "epoch": 0.15254237288135594,
      "grad_norm": 1.6019879579544067,
      "learning_rate": 6.932495846462261e-05,
      "loss": 3.0604,
      "step": 81
    },
    {
      "epoch": 0.1544256120527307,
      "grad_norm": 1.6384003162384033,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.7671,
      "step": 82
    },
    {
      "epoch": 0.15630885122410546,
      "grad_norm": 1.7119642496109009,
      "learning_rate": 6.778969234612584e-05,
      "loss": 3.0026,
      "step": 83
    },
    {
      "epoch": 0.15819209039548024,
      "grad_norm": 1.6655210256576538,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.6533,
      "step": 84
    },
    {
      "epoch": 0.160075329566855,
      "grad_norm": 1.9186038970947266,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.9363,
      "step": 85
    },
    {
      "epoch": 0.16195856873822975,
      "grad_norm": 1.489404559135437,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.6277,
      "step": 86
    },
    {
      "epoch": 0.1638418079096045,
      "grad_norm": 1.7440699338912964,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.8285,
      "step": 87
    },
    {
      "epoch": 0.1657250470809793,
      "grad_norm": 1.6308974027633667,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.7187,
      "step": 88
    },
    {
      "epoch": 0.16760828625235405,
      "grad_norm": 1.5655220746994019,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.8688,
      "step": 89
    },
    {
      "epoch": 0.1694915254237288,
      "grad_norm": 1.77265465259552,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.6857,
      "step": 90
    },
    {
      "epoch": 0.1713747645951036,
      "grad_norm": 1.9447436332702637,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.9053,
      "step": 91
    },
    {
      "epoch": 0.17325800376647835,
      "grad_norm": 1.775681495666504,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.7148,
      "step": 92
    },
    {
      "epoch": 0.1751412429378531,
      "grad_norm": 1.7115436792373657,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.7373,
      "step": 93
    },
    {
      "epoch": 0.17702448210922786,
      "grad_norm": 1.9007091522216797,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.6979,
      "step": 94
    },
    {
      "epoch": 0.17890772128060264,
      "grad_norm": 2.017606258392334,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6975,
      "step": 95
    },
    {
      "epoch": 0.1807909604519774,
      "grad_norm": 1.9475010633468628,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.821,
      "step": 96
    },
    {
      "epoch": 0.18267419962335216,
      "grad_norm": 2.0956406593322754,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.9663,
      "step": 97
    },
    {
      "epoch": 0.18455743879472694,
      "grad_norm": 2.372631788253784,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.9039,
      "step": 98
    },
    {
      "epoch": 0.1864406779661017,
      "grad_norm": 2.242234468460083,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.8717,
      "step": 99
    },
    {
      "epoch": 0.18832391713747645,
      "grad_norm": 2.3908538818359375,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.9875,
      "step": 100
    },
    {
      "epoch": 0.18832391713747645,
      "eval_loss": 3.0859832763671875,
      "eval_runtime": 84.1173,
      "eval_samples_per_second": 10.64,
      "eval_steps_per_second": 2.663,
      "step": 100
    },
    {
      "epoch": 0.1902071563088512,
      "grad_norm": 4.014430999755859,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.0592,
      "step": 101
    },
    {
      "epoch": 0.192090395480226,
      "grad_norm": 3.2044429779052734,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.3896,
      "step": 102
    },
    {
      "epoch": 0.19397363465160075,
      "grad_norm": 2.226583957672119,
      "learning_rate": 5.165316846586541e-05,
      "loss": 3.3422,
      "step": 103
    },
    {
      "epoch": 0.1958568738229755,
      "grad_norm": 1.3197691440582275,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.2024,
      "step": 104
    },
    {
      "epoch": 0.1977401129943503,
      "grad_norm": 1.0132977962493896,
      "learning_rate": 5e-05,
      "loss": 3.2219,
      "step": 105
    },
    {
      "epoch": 0.19962335216572505,
      "grad_norm": 0.9062100052833557,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.1935,
      "step": 106
    },
    {
      "epoch": 0.2015065913370998,
      "grad_norm": 0.9697667360305786,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.8305,
      "step": 107
    },
    {
      "epoch": 0.2033898305084746,
      "grad_norm": 0.9510912299156189,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.2501,
      "step": 108
    },
    {
      "epoch": 0.20527306967984935,
      "grad_norm": 0.9355286359786987,
      "learning_rate": 4.669547078371504e-05,
      "loss": 3.164,
      "step": 109
    },
    {
      "epoch": 0.2071563088512241,
      "grad_norm": 0.9343602061271667,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.916,
      "step": 110
    },
    {
      "epoch": 0.20903954802259886,
      "grad_norm": 1.0070860385894775,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.0341,
      "step": 111
    },
    {
      "epoch": 0.21092278719397364,
      "grad_norm": 1.0562351942062378,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.0562,
      "step": 112
    },
    {
      "epoch": 0.2128060263653484,
      "grad_norm": 1.1873866319656372,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.9973,
      "step": 113
    },
    {
      "epoch": 0.21468926553672316,
      "grad_norm": 1.0758321285247803,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.8368,
      "step": 114
    },
    {
      "epoch": 0.21657250470809794,
      "grad_norm": 1.1357245445251465,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.9367,
      "step": 115
    },
    {
      "epoch": 0.2184557438794727,
      "grad_norm": 1.3519351482391357,
      "learning_rate": 4.095597328339452e-05,
      "loss": 3.0099,
      "step": 116
    },
    {
      "epoch": 0.22033898305084745,
      "grad_norm": 1.2024457454681396,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.7763,
      "step": 117
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1.2387433052062988,
      "learning_rate": 3.933501846281267e-05,
      "loss": 3.0198,
      "step": 118
    },
    {
      "epoch": 0.224105461393597,
      "grad_norm": 1.313709020614624,
      "learning_rate": 3.852880399766243e-05,
      "loss": 3.1109,
      "step": 119
    },
    {
      "epoch": 0.22598870056497175,
      "grad_norm": 1.244250774383545,
      "learning_rate": 3.772572564296005e-05,
      "loss": 3.167,
      "step": 120
    },
    {
      "epoch": 0.2278719397363465,
      "grad_norm": 1.196662187576294,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.8915,
      "step": 121
    },
    {
      "epoch": 0.2297551789077213,
      "grad_norm": 1.2813066244125366,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.8115,
      "step": 122
    },
    {
      "epoch": 0.23163841807909605,
      "grad_norm": 1.3406072854995728,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.9674,
      "step": 123
    },
    {
      "epoch": 0.2335216572504708,
      "grad_norm": 1.3253072500228882,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.9662,
      "step": 124
    },
    {
      "epoch": 0.23540489642184556,
      "grad_norm": 1.1978328227996826,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.6883,
      "step": 125
    },
    {
      "epoch": 0.23728813559322035,
      "grad_norm": 1.5642106533050537,
      "learning_rate": 3.298534127791785e-05,
      "loss": 3.0166,
      "step": 126
    },
    {
      "epoch": 0.2391713747645951,
      "grad_norm": 1.3431531190872192,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.8157,
      "step": 127
    },
    {
      "epoch": 0.24105461393596986,
      "grad_norm": 1.359119176864624,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.8429,
      "step": 128
    },
    {
      "epoch": 0.24293785310734464,
      "grad_norm": 1.3198179006576538,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.6613,
      "step": 129
    },
    {
      "epoch": 0.2448210922787194,
      "grad_norm": 1.302690863609314,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.7254,
      "step": 130
    },
    {
      "epoch": 0.24670433145009416,
      "grad_norm": 1.565119981765747,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.501,
      "step": 131
    },
    {
      "epoch": 0.24858757062146894,
      "grad_norm": 1.453204870223999,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.8787,
      "step": 132
    },
    {
      "epoch": 0.2504708097928437,
      "grad_norm": 1.467968463897705,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.8343,
      "step": 133
    },
    {
      "epoch": 0.2523540489642185,
      "grad_norm": 1.429507851600647,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.7667,
      "step": 134
    },
    {
      "epoch": 0.2542372881355932,
      "grad_norm": 1.5315501689910889,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.8579,
      "step": 135
    },
    {
      "epoch": 0.256120527306968,
      "grad_norm": 1.46918785572052,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.7235,
      "step": 136
    },
    {
      "epoch": 0.2580037664783427,
      "grad_norm": 1.4308747053146362,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.6879,
      "step": 137
    },
    {
      "epoch": 0.2598870056497175,
      "grad_norm": 1.5765674114227295,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.7142,
      "step": 138
    },
    {
      "epoch": 0.2617702448210923,
      "grad_norm": 1.8339365720748901,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.9516,
      "step": 139
    },
    {
      "epoch": 0.263653483992467,
      "grad_norm": 1.5653728246688843,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.6766,
      "step": 140
    },
    {
      "epoch": 0.2655367231638418,
      "grad_norm": 1.8216297626495361,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.8411,
      "step": 141
    },
    {
      "epoch": 0.2674199623352166,
      "grad_norm": 1.9197458028793335,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.0476,
      "step": 142
    },
    {
      "epoch": 0.2693032015065913,
      "grad_norm": 1.7129372358322144,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.7656,
      "step": 143
    },
    {
      "epoch": 0.2711864406779661,
      "grad_norm": 1.7168159484863281,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.6073,
      "step": 144
    },
    {
      "epoch": 0.2730696798493409,
      "grad_norm": 1.5696486234664917,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.5049,
      "step": 145
    },
    {
      "epoch": 0.2749529190207156,
      "grad_norm": 2.275179147720337,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.6699,
      "step": 146
    },
    {
      "epoch": 0.2768361581920904,
      "grad_norm": 1.80905020236969,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.7997,
      "step": 147
    },
    {
      "epoch": 0.2787193973634652,
      "grad_norm": 1.8971529006958008,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.7801,
      "step": 148
    },
    {
      "epoch": 0.2806026365348399,
      "grad_norm": 1.895287036895752,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.8401,
      "step": 149
    },
    {
      "epoch": 0.2824858757062147,
      "grad_norm": 1.9081578254699707,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.6307,
      "step": 150
    },
    {
      "epoch": 0.2824858757062147,
      "eval_loss": 2.999302387237549,
      "eval_runtime": 84.1828,
      "eval_samples_per_second": 10.632,
      "eval_steps_per_second": 2.661,
      "step": 150
    },
    {
      "epoch": 0.2843691148775895,
      "grad_norm": 1.9292227029800415,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.1515,
      "step": 151
    },
    {
      "epoch": 0.2862523540489642,
      "grad_norm": 2.149834632873535,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 3.1294,
      "step": 152
    },
    {
      "epoch": 0.288135593220339,
      "grad_norm": 2.3186590671539307,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.202,
      "step": 153
    },
    {
      "epoch": 0.2900188323917137,
      "grad_norm": 2.065333843231201,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.3281,
      "step": 154
    },
    {
      "epoch": 0.2919020715630885,
      "grad_norm": 1.923057198524475,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 3.2773,
      "step": 155
    },
    {
      "epoch": 0.2937853107344633,
      "grad_norm": 1.8536986112594604,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.1431,
      "step": 156
    },
    {
      "epoch": 0.295668549905838,
      "grad_norm": 1.6561652421951294,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 3.1784,
      "step": 157
    },
    {
      "epoch": 0.2975517890772128,
      "grad_norm": 1.33123779296875,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.0747,
      "step": 158
    },
    {
      "epoch": 0.2994350282485876,
      "grad_norm": 1.2324334383010864,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.2008,
      "step": 159
    },
    {
      "epoch": 0.3013182674199623,
      "grad_norm": 1.0965245962142944,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 3.1245,
      "step": 160
    },
    {
      "epoch": 0.3032015065913371,
      "grad_norm": 1.0899035930633545,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.9268,
      "step": 161
    },
    {
      "epoch": 0.3050847457627119,
      "grad_norm": 1.0179719924926758,
      "learning_rate": 9.549150281252633e-06,
      "loss": 3.06,
      "step": 162
    },
    {
      "epoch": 0.3069679849340866,
      "grad_norm": 1.043216347694397,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.0953,
      "step": 163
    },
    {
      "epoch": 0.3088512241054614,
      "grad_norm": 0.9929958581924438,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.8708,
      "step": 164
    },
    {
      "epoch": 0.3107344632768362,
      "grad_norm": 1.0903116464614868,
      "learning_rate": 8.141676086873572e-06,
      "loss": 3.0618,
      "step": 165
    },
    {
      "epoch": 0.3126177024482109,
      "grad_norm": 1.0406426191329956,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.8713,
      "step": 166
    },
    {
      "epoch": 0.3145009416195857,
      "grad_norm": 1.081437110900879,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.9906,
      "step": 167
    },
    {
      "epoch": 0.3163841807909605,
      "grad_norm": 1.1000030040740967,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.8243,
      "step": 168
    },
    {
      "epoch": 0.3182674199623352,
      "grad_norm": 1.1690813302993774,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.7323,
      "step": 169
    },
    {
      "epoch": 0.32015065913371,
      "grad_norm": 1.3410359621047974,
      "learning_rate": 6.026312439675552e-06,
      "loss": 3.083,
      "step": 170
    },
    {
      "epoch": 0.3220338983050847,
      "grad_norm": 1.1199862957000732,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.6038,
      "step": 171
    },
    {
      "epoch": 0.3239171374764595,
      "grad_norm": 1.1899114847183228,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.7865,
      "step": 172
    },
    {
      "epoch": 0.3258003766478343,
      "grad_norm": 1.2084773778915405,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.9007,
      "step": 173
    },
    {
      "epoch": 0.327683615819209,
      "grad_norm": 1.2817610502243042,
      "learning_rate": 4.549673247541875e-06,
      "loss": 3.0293,
      "step": 174
    },
    {
      "epoch": 0.3295668549905838,
      "grad_norm": 1.3755650520324707,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.8946,
      "step": 175
    },
    {
      "epoch": 0.3314500941619586,
      "grad_norm": 1.3017992973327637,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.8591,
      "step": 176
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.3509595394134521,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.6648,
      "step": 177
    },
    {
      "epoch": 0.3352165725047081,
      "grad_norm": 1.4948245286941528,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.9096,
      "step": 178
    },
    {
      "epoch": 0.3370998116760829,
      "grad_norm": 1.4310879707336426,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.7543,
      "step": 179
    },
    {
      "epoch": 0.3389830508474576,
      "grad_norm": 1.2463889122009277,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.6284,
      "step": 180
    },
    {
      "epoch": 0.3408662900188324,
      "grad_norm": 1.578369140625,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.9463,
      "step": 181
    },
    {
      "epoch": 0.3427495291902072,
      "grad_norm": 1.5463346242904663,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.6064,
      "step": 182
    },
    {
      "epoch": 0.3446327683615819,
      "grad_norm": 1.7626255750656128,
      "learning_rate": 1.962316193157593e-06,
      "loss": 3.1057,
      "step": 183
    },
    {
      "epoch": 0.3465160075329567,
      "grad_norm": 1.6831303834915161,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.7708,
      "step": 184
    },
    {
      "epoch": 0.3483992467043315,
      "grad_norm": 1.8432420492172241,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.9415,
      "step": 185
    },
    {
      "epoch": 0.3502824858757062,
      "grad_norm": 1.8837491273880005,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.9082,
      "step": 186
    },
    {
      "epoch": 0.352165725047081,
      "grad_norm": 1.6048381328582764,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.697,
      "step": 187
    },
    {
      "epoch": 0.3540489642184557,
      "grad_norm": 1.7277355194091797,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.972,
      "step": 188
    },
    {
      "epoch": 0.3559322033898305,
      "grad_norm": 1.6392277479171753,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.5739,
      "step": 189
    },
    {
      "epoch": 0.3578154425612053,
      "grad_norm": 1.695238471031189,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.6498,
      "step": 190
    },
    {
      "epoch": 0.35969868173258,
      "grad_norm": 1.9994577169418335,
      "learning_rate": 5.526064699265753e-07,
      "loss": 3.1411,
      "step": 191
    },
    {
      "epoch": 0.3615819209039548,
      "grad_norm": 1.9724178314208984,
      "learning_rate": 4.367965336512403e-07,
      "loss": 3.0422,
      "step": 192
    },
    {
      "epoch": 0.3634651600753296,
      "grad_norm": 1.897047996520996,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.7169,
      "step": 193
    },
    {
      "epoch": 0.3653483992467043,
      "grad_norm": 2.0710525512695312,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.7913,
      "step": 194
    },
    {
      "epoch": 0.3672316384180791,
      "grad_norm": 1.9151296615600586,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.5889,
      "step": 195
    },
    {
      "epoch": 0.3691148775894539,
      "grad_norm": 2.0966222286224365,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.885,
      "step": 196
    },
    {
      "epoch": 0.3709981167608286,
      "grad_norm": 2.2905309200286865,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.8489,
      "step": 197
    },
    {
      "epoch": 0.3728813559322034,
      "grad_norm": 2.56792950630188,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 3.1196,
      "step": 198
    },
    {
      "epoch": 0.3747645951035782,
      "grad_norm": 2.9147226810455322,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.9975,
      "step": 199
    },
    {
      "epoch": 0.3766478342749529,
      "grad_norm": 2.9878008365631104,
      "learning_rate": 0.0,
      "loss": 3.1973,
      "step": 200
    },
    {
      "epoch": 0.3766478342749529,
      "eval_loss": 2.9099483489990234,
      "eval_runtime": 84.2571,
      "eval_samples_per_second": 10.622,
      "eval_steps_per_second": 2.659,
      "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": 3.458761861544018e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}