{
  "best_metric": 1.0124179124832153,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.3280032800328003,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016400164001640015,
      "grad_norm": 5.869199275970459,
      "learning_rate": 1e-05,
      "loss": 3.662,
      "step": 1
    },
    {
      "epoch": 0.0016400164001640015,
      "eval_loss": 1.2280761003494263,
      "eval_runtime": 237.6038,
      "eval_samples_per_second": 4.322,
      "eval_steps_per_second": 1.082,
      "step": 1
    },
    {
      "epoch": 0.003280032800328003,
      "grad_norm": 6.969863414764404,
      "learning_rate": 2e-05,
      "loss": 4.2734,
      "step": 2
    },
    {
      "epoch": 0.004920049200492005,
      "grad_norm": 5.851832389831543,
      "learning_rate": 3e-05,
      "loss": 4.0491,
      "step": 3
    },
    {
      "epoch": 0.006560065600656006,
      "grad_norm": 4.473809719085693,
      "learning_rate": 4e-05,
      "loss": 4.683,
      "step": 4
    },
    {
      "epoch": 0.008200082000820008,
      "grad_norm": 6.830697059631348,
      "learning_rate": 5e-05,
      "loss": 4.0391,
      "step": 5
    },
    {
      "epoch": 0.00984009840098401,
      "grad_norm": 6.396286487579346,
      "learning_rate": 6e-05,
      "loss": 4.4501,
      "step": 6
    },
    {
      "epoch": 0.011480114801148012,
      "grad_norm": 4.749326229095459,
      "learning_rate": 7e-05,
      "loss": 4.2174,
      "step": 7
    },
    {
      "epoch": 0.013120131201312012,
      "grad_norm": 4.601191520690918,
      "learning_rate": 8e-05,
      "loss": 4.1558,
      "step": 8
    },
    {
      "epoch": 0.014760147601476014,
      "grad_norm": 4.322552680969238,
      "learning_rate": 9e-05,
      "loss": 4.3853,
      "step": 9
    },
    {
      "epoch": 0.016400164001640016,
      "grad_norm": 3.4741907119750977,
      "learning_rate": 0.0001,
      "loss": 4.0951,
      "step": 10
    },
    {
      "epoch": 0.018040180401804017,
      "grad_norm": 3.2274389266967773,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.2757,
      "step": 11
    },
    {
      "epoch": 0.01968019680196802,
      "grad_norm": 3.270996332168579,
      "learning_rate": 9.997266286704631e-05,
      "loss": 4.4489,
      "step": 12
    },
    {
      "epoch": 0.02132021320213202,
      "grad_norm": 3.6034586429595947,
      "learning_rate": 9.993849845741524e-05,
      "loss": 4.0802,
      "step": 13
    },
    {
      "epoch": 0.022960229602296024,
      "grad_norm": 3.0808043479919434,
      "learning_rate": 9.989068136093873e-05,
      "loss": 4.1136,
      "step": 14
    },
    {
      "epoch": 0.024600246002460024,
      "grad_norm": 3.1643295288085938,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.5872,
      "step": 15
    },
    {
      "epoch": 0.026240262402624025,
      "grad_norm": 3.0146238803863525,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.3849,
      "step": 16
    },
    {
      "epoch": 0.02788027880278803,
      "grad_norm": 3.156456470489502,
      "learning_rate": 9.966546331768191e-05,
      "loss": 4.4676,
      "step": 17
    },
    {
      "epoch": 0.02952029520295203,
      "grad_norm": 2.756700277328491,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.4958,
      "step": 18
    },
    {
      "epoch": 0.031160311603116032,
      "grad_norm": 3.910738706588745,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.0827,
      "step": 19
    },
    {
      "epoch": 0.03280032800328003,
      "grad_norm": 3.02478289604187,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.2826,
      "step": 20
    },
    {
      "epoch": 0.03444034440344403,
      "grad_norm": 2.9688055515289307,
      "learning_rate": 9.917525374361912e-05,
      "loss": 4.4964,
      "step": 21
    },
    {
      "epoch": 0.03608036080360803,
      "grad_norm": 2.8419086933135986,
      "learning_rate": 9.901899829374047e-05,
      "loss": 4.3045,
      "step": 22
    },
    {
      "epoch": 0.03772037720377204,
      "grad_norm": 2.802485227584839,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.3695,
      "step": 23
    },
    {
      "epoch": 0.03936039360393604,
      "grad_norm": 2.7650442123413086,
      "learning_rate": 9.86663298624003e-05,
      "loss": 4.3283,
      "step": 24
    },
    {
      "epoch": 0.04100041000410004,
      "grad_norm": 2.806762933731079,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.449,
      "step": 25
    },
    {
      "epoch": 0.04264042640426404,
      "grad_norm": 2.653256416320801,
      "learning_rate": 9.826044551386744e-05,
      "loss": 4.3709,
      "step": 26
    },
    {
      "epoch": 0.04428044280442804,
      "grad_norm": 2.7254297733306885,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.6991,
      "step": 27
    },
    {
      "epoch": 0.04592045920459205,
      "grad_norm": 2.632901191711426,
      "learning_rate": 9.780178907671789e-05,
      "loss": 4.3628,
      "step": 28
    },
    {
      "epoch": 0.04756047560475605,
      "grad_norm": 2.6592962741851807,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.1369,
      "step": 29
    },
    {
      "epoch": 0.04920049200492005,
      "grad_norm": 2.6830780506134033,
      "learning_rate": 9.729086208503174e-05,
      "loss": 4.6049,
      "step": 30
    },
    {
      "epoch": 0.05084050840508405,
      "grad_norm": 2.5135135650634766,
      "learning_rate": 9.701596950580806e-05,
      "loss": 4.0797,
      "step": 31
    },
    {
      "epoch": 0.05248052480524805,
      "grad_norm": 2.5594406127929688,
      "learning_rate": 9.672822322997305e-05,
      "loss": 4.6254,
      "step": 32
    },
    {
      "epoch": 0.05412054120541206,
      "grad_norm": 2.569803476333618,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.2541,
      "step": 33
    },
    {
      "epoch": 0.05576055760557606,
      "grad_norm": 2.496262311935425,
      "learning_rate": 9.611448774886924e-05,
      "loss": 4.5,
      "step": 34
    },
    {
      "epoch": 0.05740057400574006,
      "grad_norm": 2.5223639011383057,
      "learning_rate": 9.578866633275288e-05,
      "loss": 4.3539,
      "step": 35
    },
    {
      "epoch": 0.05904059040590406,
      "grad_norm": 2.599987030029297,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.4577,
      "step": 36
    },
    {
      "epoch": 0.06068060680606806,
      "grad_norm": 2.6805663108825684,
      "learning_rate": 9.509956150664796e-05,
      "loss": 4.7178,
      "step": 37
    },
    {
      "epoch": 0.062320623206232065,
      "grad_norm": 2.485438585281372,
      "learning_rate": 9.473646649103818e-05,
      "loss": 4.4107,
      "step": 38
    },
    {
      "epoch": 0.06396063960639606,
      "grad_norm": 2.6332499980926514,
      "learning_rate": 9.43611409721806e-05,
      "loss": 4.6175,
      "step": 39
    },
    {
      "epoch": 0.06560065600656007,
      "grad_norm": 2.695639133453369,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.1513,
      "step": 40
    },
    {
      "epoch": 0.06724067240672407,
      "grad_norm": 2.7205419540405273,
      "learning_rate": 9.357421218136386e-05,
      "loss": 4.6704,
      "step": 41
    },
    {
      "epoch": 0.06888068880688807,
      "grad_norm": 2.8747119903564453,
      "learning_rate": 9.316282404787871e-05,
      "loss": 4.2848,
      "step": 42
    },
    {
      "epoch": 0.07052070520705207,
      "grad_norm": 2.8814995288848877,
      "learning_rate": 9.273963562927695e-05,
      "loss": 4.5756,
      "step": 43
    },
    {
      "epoch": 0.07216072160721607,
      "grad_norm": 2.9116883277893066,
      "learning_rate": 9.230476262104677e-05,
      "loss": 4.6236,
      "step": 44
    },
    {
      "epoch": 0.07380073800738007,
      "grad_norm": 2.986185312271118,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.7377,
      "step": 45
    },
    {
      "epoch": 0.07544075440754408,
      "grad_norm": 3.0014514923095703,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.479,
      "step": 46
    },
    {
      "epoch": 0.07708077080770807,
      "grad_norm": 3.051936149597168,
      "learning_rate": 9.093124073433463e-05,
      "loss": 4.3579,
      "step": 47
    },
    {
      "epoch": 0.07872078720787208,
      "grad_norm": 3.2363364696502686,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.7213,
      "step": 48
    },
    {
      "epoch": 0.08036080360803607,
      "grad_norm": 3.5005013942718506,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.5769,
      "step": 49
    },
    {
      "epoch": 0.08200082000820008,
      "grad_norm": 3.8495800495147705,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.1898,
      "step": 50
    },
    {
      "epoch": 0.08200082000820008,
      "eval_loss": 1.0608546733856201,
      "eval_runtime": 240.8508,
      "eval_samples_per_second": 4.264,
      "eval_steps_per_second": 1.067,
      "step": 50
    },
    {
      "epoch": 0.08364083640836409,
      "grad_norm": 3.441438913345337,
      "learning_rate": 8.894386393810563e-05,
      "loss": 3.785,
      "step": 51
    },
    {
      "epoch": 0.08528085280852808,
      "grad_norm": 2.6722733974456787,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.9664,
      "step": 52
    },
    {
      "epoch": 0.08692086920869209,
      "grad_norm": 2.544339179992676,
      "learning_rate": 8.788574348801675e-05,
      "loss": 3.7634,
      "step": 53
    },
    {
      "epoch": 0.08856088560885608,
      "grad_norm": 2.8062400817871094,
      "learning_rate": 8.73410738492077e-05,
      "loss": 3.7028,
      "step": 54
    },
    {
      "epoch": 0.09020090200902009,
      "grad_norm": 2.6475799083709717,
      "learning_rate": 8.678619553365659e-05,
      "loss": 3.8803,
      "step": 55
    },
    {
      "epoch": 0.0918409184091841,
      "grad_norm": 2.4030749797821045,
      "learning_rate": 8.622126023955446e-05,
      "loss": 3.8163,
      "step": 56
    },
    {
      "epoch": 0.09348093480934809,
      "grad_norm": 2.3919782638549805,
      "learning_rate": 8.564642241456986e-05,
      "loss": 3.6336,
      "step": 57
    },
    {
      "epoch": 0.0951209512095121,
      "grad_norm": 2.480132818222046,
      "learning_rate": 8.506183921362443e-05,
      "loss": 3.7429,
      "step": 58
    },
    {
      "epoch": 0.09676096760967609,
      "grad_norm": 2.5213751792907715,
      "learning_rate": 8.44676704559283e-05,
      "loss": 3.7879,
      "step": 59
    },
    {
      "epoch": 0.0984009840098401,
      "grad_norm": 2.4064886569976807,
      "learning_rate": 8.386407858128706e-05,
      "loss": 4.2192,
      "step": 60
    },
    {
      "epoch": 0.1000410004100041,
      "grad_norm": 2.4535820484161377,
      "learning_rate": 8.32512286056924e-05,
      "loss": 4.0685,
      "step": 61
    },
    {
      "epoch": 0.1016810168101681,
      "grad_norm": 2.422741651535034,
      "learning_rate": 8.262928807620843e-05,
      "loss": 4.1786,
      "step": 62
    },
    {
      "epoch": 0.1033210332103321,
      "grad_norm": 2.4493520259857178,
      "learning_rate": 8.199842702516583e-05,
      "loss": 4.7146,
      "step": 63
    },
    {
      "epoch": 0.1049610496104961,
      "grad_norm": 2.417475700378418,
      "learning_rate": 8.135881792367686e-05,
      "loss": 4.117,
      "step": 64
    },
    {
      "epoch": 0.1066010660106601,
      "grad_norm": 2.375321865081787,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.5069,
      "step": 65
    },
    {
      "epoch": 0.10824108241082411,
      "grad_norm": 2.4565398693084717,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.0437,
      "step": 66
    },
    {
      "epoch": 0.1098810988109881,
      "grad_norm": 2.44014835357666,
      "learning_rate": 7.938926261462366e-05,
      "loss": 4.5727,
      "step": 67
    },
    {
      "epoch": 0.11152111521115211,
      "grad_norm": 2.3664791584014893,
      "learning_rate": 7.871643313414718e-05,
      "loss": 3.7956,
      "step": 68
    },
    {
      "epoch": 0.11316113161131611,
      "grad_norm": 2.3263285160064697,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.1809,
      "step": 69
    },
    {
      "epoch": 0.11480114801148011,
      "grad_norm": 2.381890058517456,
      "learning_rate": 7.734740790612136e-05,
      "loss": 4.2684,
      "step": 70
    },
    {
      "epoch": 0.11644116441164412,
      "grad_norm": 2.294248580932617,
      "learning_rate": 7.66515864363997e-05,
      "loss": 4.1593,
      "step": 71
    },
    {
      "epoch": 0.11808118081180811,
      "grad_norm": 2.318941116333008,
      "learning_rate": 7.594847868906076e-05,
      "loss": 4.4192,
      "step": 72
    },
    {
      "epoch": 0.11972119721197212,
      "grad_norm": 2.343067169189453,
      "learning_rate": 7.52382768867422e-05,
      "loss": 4.1391,
      "step": 73
    },
    {
      "epoch": 0.12136121361213612,
      "grad_norm": 2.3993380069732666,
      "learning_rate": 7.452117519152542e-05,
      "loss": 3.9731,
      "step": 74
    },
    {
      "epoch": 0.12300123001230012,
      "grad_norm": 2.3573107719421387,
      "learning_rate": 7.379736965185368e-05,
      "loss": 4.4762,
      "step": 75
    },
    {
      "epoch": 0.12464124641246413,
      "grad_norm": 2.387101411819458,
      "learning_rate": 7.30670581489344e-05,
      "loss": 4.6123,
      "step": 76
    },
    {
      "epoch": 0.12628126281262814,
      "grad_norm": 2.2863190174102783,
      "learning_rate": 7.233044034264034e-05,
      "loss": 4.5928,
      "step": 77
    },
    {
      "epoch": 0.12792127921279212,
      "grad_norm": 2.336501359939575,
      "learning_rate": 7.158771761692464e-05,
      "loss": 4.7428,
      "step": 78
    },
    {
      "epoch": 0.12956129561295612,
      "grad_norm": 2.3301258087158203,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.1423,
      "step": 79
    },
    {
      "epoch": 0.13120131201312013,
      "grad_norm": 2.3380656242370605,
      "learning_rate": 7.008477123264848e-05,
      "loss": 4.0852,
      "step": 80
    },
    {
      "epoch": 0.13284132841328414,
      "grad_norm": 2.385219097137451,
      "learning_rate": 6.932495846462261e-05,
      "loss": 4.4596,
      "step": 81
    },
    {
      "epoch": 0.13448134481344814,
      "grad_norm": 2.293954372406006,
      "learning_rate": 6.855986244591104e-05,
      "loss": 4.4405,
      "step": 82
    },
    {
      "epoch": 0.13612136121361212,
      "grad_norm": 2.3502376079559326,
      "learning_rate": 6.778969234612584e-05,
      "loss": 4.4252,
      "step": 83
    },
    {
      "epoch": 0.13776137761377613,
      "grad_norm": 2.356443166732788,
      "learning_rate": 6.701465872208216e-05,
      "loss": 4.5532,
      "step": 84
    },
    {
      "epoch": 0.13940139401394014,
      "grad_norm": 2.2868731021881104,
      "learning_rate": 6.623497346023418e-05,
      "loss": 4.1459,
      "step": 85
    },
    {
      "epoch": 0.14104141041410415,
      "grad_norm": 2.292008399963379,
      "learning_rate": 6.545084971874738e-05,
      "loss": 4.4789,
      "step": 86
    },
    {
      "epoch": 0.14268142681426815,
      "grad_norm": 2.328474521636963,
      "learning_rate": 6.466250186922325e-05,
      "loss": 4.4808,
      "step": 87
    },
    {
      "epoch": 0.14432144321443213,
      "grad_norm": 2.4643123149871826,
      "learning_rate": 6.387014543809223e-05,
      "loss": 4.3991,
      "step": 88
    },
    {
      "epoch": 0.14596145961459614,
      "grad_norm": 2.433647632598877,
      "learning_rate": 6.307399704769099e-05,
      "loss": 4.3608,
      "step": 89
    },
    {
      "epoch": 0.14760147601476015,
      "grad_norm": 2.3895249366760254,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.5059,
      "step": 90
    },
    {
      "epoch": 0.14924149241492415,
      "grad_norm": 2.544801950454712,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.3521,
      "step": 91
    },
    {
      "epoch": 0.15088150881508816,
      "grad_norm": 2.465904474258423,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.1968,
      "step": 92
    },
    {
      "epoch": 0.15252152521525214,
      "grad_norm": 2.5939810276031494,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.4577,
      "step": 93
    },
    {
      "epoch": 0.15416154161541615,
      "grad_norm": 2.6836864948272705,
      "learning_rate": 5.90440267166055e-05,
      "loss": 4.2926,
      "step": 94
    },
    {
      "epoch": 0.15580155801558015,
      "grad_norm": 2.7345409393310547,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.5238,
      "step": 95
    },
    {
      "epoch": 0.15744157441574416,
      "grad_norm": 2.6608169078826904,
      "learning_rate": 5.74131823855921e-05,
      "loss": 4.2368,
      "step": 96
    },
    {
      "epoch": 0.15908159081590817,
      "grad_norm": 2.796182632446289,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 3.9898,
      "step": 97
    },
    {
      "epoch": 0.16072160721607215,
      "grad_norm": 3.000919818878174,
      "learning_rate": 5.577423184847932e-05,
      "loss": 4.4552,
      "step": 98
    },
    {
      "epoch": 0.16236162361623616,
      "grad_norm": 3.112570285797119,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.8275,
      "step": 99
    },
    {
      "epoch": 0.16400164001640016,
      "grad_norm": 3.4824225902557373,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.9572,
      "step": 100
    },
    {
      "epoch": 0.16400164001640016,
      "eval_loss": 1.033058524131775,
      "eval_runtime": 240.3731,
      "eval_samples_per_second": 4.273,
      "eval_steps_per_second": 1.069,
      "step": 100
    },
    {
      "epoch": 0.16564165641656417,
      "grad_norm": 2.6610124111175537,
      "learning_rate": 5.330452921628497e-05,
      "loss": 3.1483,
      "step": 101
    },
    {
      "epoch": 0.16728167281672818,
      "grad_norm": 2.432785987854004,
      "learning_rate": 5.247918773366112e-05,
      "loss": 3.4353,
      "step": 102
    },
    {
      "epoch": 0.16892168921689216,
      "grad_norm": 2.5268452167510986,
      "learning_rate": 5.165316846586541e-05,
      "loss": 4.0782,
      "step": 103
    },
    {
      "epoch": 0.17056170561705616,
      "grad_norm": 2.2998476028442383,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 3.6813,
      "step": 104
    },
    {
      "epoch": 0.17220172201722017,
      "grad_norm": 2.2520883083343506,
      "learning_rate": 5e-05,
      "loss": 4.0651,
      "step": 105
    },
    {
      "epoch": 0.17384173841738418,
      "grad_norm": 2.3039751052856445,
      "learning_rate": 4.917330276168208e-05,
      "loss": 3.6916,
      "step": 106
    },
    {
      "epoch": 0.17548175481754819,
      "grad_norm": 2.3266818523406982,
      "learning_rate": 4.834683153413459e-05,
      "loss": 3.7256,
      "step": 107
    },
    {
      "epoch": 0.17712177121771217,
      "grad_norm": 2.240109443664551,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 3.8126,
      "step": 108
    },
    {
      "epoch": 0.17876178761787617,
      "grad_norm": 2.2738168239593506,
      "learning_rate": 4.669547078371504e-05,
      "loss": 4.2271,
      "step": 109
    },
    {
      "epoch": 0.18040180401804018,
      "grad_norm": 2.248335123062134,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 3.5879,
      "step": 110
    },
    {
      "epoch": 0.1820418204182042,
      "grad_norm": 2.301835298538208,
      "learning_rate": 4.504772348747687e-05,
      "loss": 3.7979,
      "step": 111
    },
    {
      "epoch": 0.1836818368183682,
      "grad_norm": 2.276437997817993,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 3.7496,
      "step": 112
    },
    {
      "epoch": 0.18532185321853217,
      "grad_norm": 2.3297178745269775,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 3.5737,
      "step": 113
    },
    {
      "epoch": 0.18696186961869618,
      "grad_norm": 2.3600358963012695,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 3.865,
      "step": 114
    },
    {
      "epoch": 0.1886018860188602,
      "grad_norm": 2.1880991458892822,
      "learning_rate": 4.17702704859633e-05,
      "loss": 4.1837,
      "step": 115
    },
    {
      "epoch": 0.1902419024190242,
      "grad_norm": 2.239628553390503,
      "learning_rate": 4.095597328339452e-05,
      "loss": 4.1196,
      "step": 116
    },
    {
      "epoch": 0.1918819188191882,
      "grad_norm": 2.193875789642334,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 4.0053,
      "step": 117
    },
    {
      "epoch": 0.19352193521935218,
      "grad_norm": 2.304914951324463,
      "learning_rate": 3.933501846281267e-05,
      "loss": 4.2058,
      "step": 118
    },
    {
      "epoch": 0.1951619516195162,
      "grad_norm": 2.2215914726257324,
      "learning_rate": 3.852880399766243e-05,
      "loss": 4.4531,
      "step": 119
    },
    {
      "epoch": 0.1968019680196802,
      "grad_norm": 2.197253942489624,
      "learning_rate": 3.772572564296005e-05,
      "loss": 4.3224,
      "step": 120
    },
    {
      "epoch": 0.1984419844198442,
      "grad_norm": 2.223278760910034,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.3788,
      "step": 121
    },
    {
      "epoch": 0.2000820008200082,
      "grad_norm": 2.24517822265625,
      "learning_rate": 3.612985456190778e-05,
      "loss": 4.2871,
      "step": 122
    },
    {
      "epoch": 0.2017220172201722,
      "grad_norm": 2.2465004920959473,
      "learning_rate": 3.533749813077677e-05,
      "loss": 4.0847,
      "step": 123
    },
    {
      "epoch": 0.2033620336203362,
      "grad_norm": 2.2783913612365723,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 4.5374,
      "step": 124
    },
    {
      "epoch": 0.2050020500205002,
      "grad_norm": 2.2170486450195312,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 3.9275,
      "step": 125
    },
    {
      "epoch": 0.2066420664206642,
      "grad_norm": 2.3049073219299316,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.1454,
      "step": 126
    },
    {
      "epoch": 0.20828208282082822,
      "grad_norm": 2.30863356590271,
      "learning_rate": 3.221030765387417e-05,
      "loss": 4.2619,
      "step": 127
    },
    {
      "epoch": 0.2099220992209922,
      "grad_norm": 2.2851221561431885,
      "learning_rate": 3.144013755408895e-05,
      "loss": 4.1256,
      "step": 128
    },
    {
      "epoch": 0.2115621156211562,
      "grad_norm": 2.269951343536377,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 4.7476,
      "step": 129
    },
    {
      "epoch": 0.2132021320213202,
      "grad_norm": 2.1727960109710693,
      "learning_rate": 2.991522876735154e-05,
      "loss": 3.9483,
      "step": 130
    },
    {
      "epoch": 0.21484214842148422,
      "grad_norm": 2.2029051780700684,
      "learning_rate": 2.916090697523549e-05,
      "loss": 4.2011,
      "step": 131
    },
    {
      "epoch": 0.21648216482164823,
      "grad_norm": 2.241426944732666,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.237,
      "step": 132
    },
    {
      "epoch": 0.2181221812218122,
      "grad_norm": 2.2840962409973145,
      "learning_rate": 2.766955965735968e-05,
      "loss": 4.4176,
      "step": 133
    },
    {
      "epoch": 0.2197621976219762,
      "grad_norm": 2.4008915424346924,
      "learning_rate": 2.693294185106562e-05,
      "loss": 4.2021,
      "step": 134
    },
    {
      "epoch": 0.22140221402214022,
      "grad_norm": 2.265554904937744,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 4.4138,
      "step": 135
    },
    {
      "epoch": 0.22304223042230423,
      "grad_norm": 2.3229594230651855,
      "learning_rate": 2.547882480847461e-05,
      "loss": 4.5234,
      "step": 136
    },
    {
      "epoch": 0.22468224682246823,
      "grad_norm": 2.4142534732818604,
      "learning_rate": 2.476172311325783e-05,
      "loss": 4.0945,
      "step": 137
    },
    {
      "epoch": 0.22632226322263221,
      "grad_norm": 2.4058334827423096,
      "learning_rate": 2.405152131093926e-05,
      "loss": 4.3832,
      "step": 138
    },
    {
      "epoch": 0.22796227962279622,
      "grad_norm": 2.451469659805298,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.0297,
      "step": 139
    },
    {
      "epoch": 0.22960229602296023,
      "grad_norm": 2.6323137283325195,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 4.6435,
      "step": 140
    },
    {
      "epoch": 0.23124231242312424,
      "grad_norm": 2.5412402153015137,
      "learning_rate": 2.196424713241637e-05,
      "loss": 3.8432,
      "step": 141
    },
    {
      "epoch": 0.23288232882328824,
      "grad_norm": 2.6137726306915283,
      "learning_rate": 2.128356686585282e-05,
      "loss": 3.8824,
      "step": 142
    },
    {
      "epoch": 0.23452234522345222,
      "grad_norm": 2.6491565704345703,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.1124,
      "step": 143
    },
    {
      "epoch": 0.23616236162361623,
      "grad_norm": 2.6097328662872314,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.2627,
      "step": 144
    },
    {
      "epoch": 0.23780237802378024,
      "grad_norm": 2.807006359100342,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.1103,
      "step": 145
    },
    {
      "epoch": 0.23944239442394424,
      "grad_norm": 2.718845844268799,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 4.5238,
      "step": 146
    },
    {
      "epoch": 0.24108241082410825,
      "grad_norm": 2.916916608810425,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.5086,
      "step": 147
    },
    {
      "epoch": 0.24272242722427223,
      "grad_norm": 2.926973581314087,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.488,
      "step": 148
    },
    {
      "epoch": 0.24436244362443624,
      "grad_norm": 3.2567660808563232,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 4.3845,
      "step": 149
    },
    {
      "epoch": 0.24600246002460024,
      "grad_norm": 3.4107182025909424,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.0055,
      "step": 150
    },
    {
      "epoch": 0.24600246002460024,
      "eval_loss": 1.0185319185256958,
      "eval_runtime": 241.0575,
      "eval_samples_per_second": 4.26,
      "eval_steps_per_second": 1.066,
      "step": 150
    },
    {
      "epoch": 0.24764247642476425,
      "grad_norm": 2.37801456451416,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.7094,
      "step": 151
    },
    {
      "epoch": 0.24928249282492826,
      "grad_norm": 2.356282949447632,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 3.3735,
      "step": 152
    },
    {
      "epoch": 0.25092250922509224,
      "grad_norm": 2.3678855895996094,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.693,
      "step": 153
    },
    {
      "epoch": 0.2525625256252563,
      "grad_norm": 2.300280809402466,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 3.6132,
      "step": 154
    },
    {
      "epoch": 0.25420254202542025,
      "grad_norm": 2.3163371086120605,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.0216,
      "step": 155
    },
    {
      "epoch": 0.25584255842558423,
      "grad_norm": 2.3181238174438477,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 3.6766,
      "step": 156
    },
    {
      "epoch": 0.25748257482574827,
      "grad_norm": 2.2757251262664795,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 3.948,
      "step": 157
    },
    {
      "epoch": 0.25912259122591225,
      "grad_norm": 2.229625701904297,
      "learning_rate": 1.157994445715706e-05,
      "loss": 3.3743,
      "step": 158
    },
    {
      "epoch": 0.2607626076260763,
      "grad_norm": 2.2134735584259033,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 3.5337,
      "step": 159
    },
    {
      "epoch": 0.26240262402624026,
      "grad_norm": 2.276475191116333,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 4.0393,
      "step": 160
    },
    {
      "epoch": 0.26404264042640424,
      "grad_norm": 2.2348151206970215,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 3.7756,
      "step": 161
    },
    {
      "epoch": 0.2656826568265683,
      "grad_norm": 2.231473445892334,
      "learning_rate": 9.549150281252633e-06,
      "loss": 4.2008,
      "step": 162
    },
    {
      "epoch": 0.26732267322673225,
      "grad_norm": 2.216063976287842,
      "learning_rate": 9.068759265665384e-06,
      "loss": 3.9354,
      "step": 163
    },
    {
      "epoch": 0.2689626896268963,
      "grad_norm": 2.253537654876709,
      "learning_rate": 8.599558442598998e-06,
      "loss": 3.7107,
      "step": 164
    },
    {
      "epoch": 0.27060270602706027,
      "grad_norm": 2.26593017578125,
      "learning_rate": 8.141676086873572e-06,
      "loss": 4.208,
      "step": 165
    },
    {
      "epoch": 0.27224272242722425,
      "grad_norm": 2.2299046516418457,
      "learning_rate": 7.695237378953223e-06,
      "loss": 3.8224,
      "step": 166
    },
    {
      "epoch": 0.2738827388273883,
      "grad_norm": 2.169259786605835,
      "learning_rate": 7.260364370723044e-06,
      "loss": 3.8697,
      "step": 167
    },
    {
      "epoch": 0.27552275522755226,
      "grad_norm": 2.2308783531188965,
      "learning_rate": 6.837175952121306e-06,
      "loss": 4.2249,
      "step": 168
    },
    {
      "epoch": 0.2771627716277163,
      "grad_norm": 2.2986156940460205,
      "learning_rate": 6.425787818636131e-06,
      "loss": 4.2202,
      "step": 169
    },
    {
      "epoch": 0.2788027880278803,
      "grad_norm": 2.281946897506714,
      "learning_rate": 6.026312439675552e-06,
      "loss": 4.1278,
      "step": 170
    },
    {
      "epoch": 0.28044280442804426,
      "grad_norm": 2.2479248046875,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 4.1797,
      "step": 171
    },
    {
      "epoch": 0.2820828208282083,
      "grad_norm": 2.3022007942199707,
      "learning_rate": 5.263533508961827e-06,
      "loss": 4.0349,
      "step": 172
    },
    {
      "epoch": 0.28372283722837227,
      "grad_norm": 2.2037179470062256,
      "learning_rate": 4.900438493352055e-06,
      "loss": 3.451,
      "step": 173
    },
    {
      "epoch": 0.2853628536285363,
      "grad_norm": 2.3144290447235107,
      "learning_rate": 4.549673247541875e-06,
      "loss": 5.0383,
      "step": 174
    },
    {
      "epoch": 0.2870028700287003,
      "grad_norm": 2.2480649948120117,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 3.877,
      "step": 175
    },
    {
      "epoch": 0.28864288642886426,
      "grad_norm": 2.281433582305908,
      "learning_rate": 3.885512251130763e-06,
      "loss": 4.1367,
      "step": 176
    },
    {
      "epoch": 0.2902829028290283,
      "grad_norm": 2.2370946407318115,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 3.984,
      "step": 177
    },
    {
      "epoch": 0.2919229192291923,
      "grad_norm": 2.243863105773926,
      "learning_rate": 3.271776770026963e-06,
      "loss": 4.0556,
      "step": 178
    },
    {
      "epoch": 0.2935629356293563,
      "grad_norm": 2.247199535369873,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 4.2579,
      "step": 179
    },
    {
      "epoch": 0.2952029520295203,
      "grad_norm": 2.236830472946167,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 4.3736,
      "step": 180
    },
    {
      "epoch": 0.2968429684296843,
      "grad_norm": 2.3090035915374756,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.3814,
      "step": 181
    },
    {
      "epoch": 0.2984829848298483,
      "grad_norm": 2.3450686931610107,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 4.43,
      "step": 182
    },
    {
      "epoch": 0.3001230012300123,
      "grad_norm": 2.271566152572632,
      "learning_rate": 1.962316193157593e-06,
      "loss": 4.2242,
      "step": 183
    },
    {
      "epoch": 0.3017630176301763,
      "grad_norm": 2.3259236812591553,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 4.3658,
      "step": 184
    },
    {
      "epoch": 0.3034030340303403,
      "grad_norm": 2.2332334518432617,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 4.308,
      "step": 185
    },
    {
      "epoch": 0.3050430504305043,
      "grad_norm": 2.315208911895752,
      "learning_rate": 1.333670137599713e-06,
      "loss": 4.6266,
      "step": 186
    },
    {
      "epoch": 0.3066830668306683,
      "grad_norm": 2.335198402404785,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 3.9152,
      "step": 187
    },
    {
      "epoch": 0.3083230832308323,
      "grad_norm": 2.379514217376709,
      "learning_rate": 9.810017062595322e-07,
      "loss": 4.364,
      "step": 188
    },
    {
      "epoch": 0.30996309963099633,
      "grad_norm": 2.37813663482666,
      "learning_rate": 8.247462563808817e-07,
      "loss": 4.5248,
      "step": 189
    },
    {
      "epoch": 0.3116031160311603,
      "grad_norm": 2.5055665969848633,
      "learning_rate": 6.819348298638839e-07,
      "loss": 4.3659,
      "step": 190
    },
    {
      "epoch": 0.3132431324313243,
      "grad_norm": 2.5743649005889893,
      "learning_rate": 5.526064699265753e-07,
      "loss": 4.3119,
      "step": 191
    },
    {
      "epoch": 0.3148831488314883,
      "grad_norm": 2.6636548042297363,
      "learning_rate": 4.367965336512403e-07,
      "loss": 4.5536,
      "step": 192
    },
    {
      "epoch": 0.3165231652316523,
      "grad_norm": 2.4852402210235596,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 3.9755,
      "step": 193
    },
    {
      "epoch": 0.31816318163181634,
      "grad_norm": 2.659194231033325,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.2137,
      "step": 194
    },
    {
      "epoch": 0.3198031980319803,
      "grad_norm": 2.7223291397094727,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.4954,
      "step": 195
    },
    {
      "epoch": 0.3214432144321443,
      "grad_norm": 2.872950315475464,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.6446,
      "step": 196
    },
    {
      "epoch": 0.32308323083230833,
      "grad_norm": 3.0511484146118164,
      "learning_rate": 6.150154258476315e-08,
      "loss": 4.6914,
      "step": 197
    },
    {
      "epoch": 0.3247232472324723,
      "grad_norm": 3.174380302429199,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.2409,
      "step": 198
    },
    {
      "epoch": 0.32636326363263635,
      "grad_norm": 3.289060115814209,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.0412,
      "step": 199
    },
    {
      "epoch": 0.3280032800328003,
      "grad_norm": 3.7197422981262207,
      "learning_rate": 0.0,
      "loss": 4.1362,
      "step": 200
    },
    {
      "epoch": 0.3280032800328003,
      "eval_loss": 1.0124179124832153,
      "eval_runtime": 240.2098,
      "eval_samples_per_second": 4.275,
      "eval_steps_per_second": 1.07,
      "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": 5.552272274872074e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}