{
  "best_metric": 1.1545281410217285,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.1695993216027136,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008479966080135679,
      "grad_norm": 4.362088203430176,
      "learning_rate": 1e-05,
      "loss": 4.4454,
      "step": 1
    },
    {
      "epoch": 0.0008479966080135679,
      "eval_loss": 1.4747374057769775,
      "eval_runtime": 366.7923,
      "eval_samples_per_second": 5.417,
      "eval_steps_per_second": 1.355,
      "step": 1
    },
    {
      "epoch": 0.0016959932160271359,
      "grad_norm": 4.764973163604736,
      "learning_rate": 2e-05,
      "loss": 4.7478,
      "step": 2
    },
    {
      "epoch": 0.0025439898240407037,
      "grad_norm": 4.6787567138671875,
      "learning_rate": 3e-05,
      "loss": 4.6386,
      "step": 3
    },
    {
      "epoch": 0.0033919864320542717,
      "grad_norm": 3.80454683303833,
      "learning_rate": 4e-05,
      "loss": 4.8356,
      "step": 4
    },
    {
      "epoch": 0.00423998304006784,
      "grad_norm": 3.849437952041626,
      "learning_rate": 5e-05,
      "loss": 4.9256,
      "step": 5
    },
    {
      "epoch": 0.005087979648081407,
      "grad_norm": 4.223307132720947,
      "learning_rate": 6e-05,
      "loss": 4.7853,
      "step": 6
    },
    {
      "epoch": 0.005935976256094976,
      "grad_norm": 3.8130242824554443,
      "learning_rate": 7e-05,
      "loss": 4.5596,
      "step": 7
    },
    {
      "epoch": 0.0067839728641085435,
      "grad_norm": 3.2439517974853516,
      "learning_rate": 8e-05,
      "loss": 4.8954,
      "step": 8
    },
    {
      "epoch": 0.007631969472122111,
      "grad_norm": 3.0985023975372314,
      "learning_rate": 9e-05,
      "loss": 4.5665,
      "step": 9
    },
    {
      "epoch": 0.00847996608013568,
      "grad_norm": 2.9138731956481934,
      "learning_rate": 0.0001,
      "loss": 5.007,
      "step": 10
    },
    {
      "epoch": 0.009327962688149248,
      "grad_norm": 3.0814642906188965,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.7177,
      "step": 11
    },
    {
      "epoch": 0.010175959296162815,
      "grad_norm": 2.865621328353882,
      "learning_rate": 9.997266286704631e-05,
      "loss": 5.0019,
      "step": 12
    },
    {
      "epoch": 0.011023955904176383,
      "grad_norm": 2.9789087772369385,
      "learning_rate": 9.993849845741524e-05,
      "loss": 5.5366,
      "step": 13
    },
    {
      "epoch": 0.011871952512189952,
      "grad_norm": 2.9014413356781006,
      "learning_rate": 9.989068136093873e-05,
      "loss": 5.031,
      "step": 14
    },
    {
      "epoch": 0.012719949120203519,
      "grad_norm": 2.8634445667266846,
      "learning_rate": 9.98292246503335e-05,
      "loss": 4.9463,
      "step": 15
    },
    {
      "epoch": 0.013567945728217087,
      "grad_norm": 3.059725761413574,
      "learning_rate": 9.975414512725057e-05,
      "loss": 4.8471,
      "step": 16
    },
    {
      "epoch": 0.014415942336230655,
      "grad_norm": 2.7378642559051514,
      "learning_rate": 9.966546331768191e-05,
      "loss": 5.3401,
      "step": 17
    },
    {
      "epoch": 0.015263938944244222,
      "grad_norm": 4.3096394538879395,
      "learning_rate": 9.956320346634876e-05,
      "loss": 4.8012,
      "step": 18
    },
    {
      "epoch": 0.016111935552257792,
      "grad_norm": 3.4539849758148193,
      "learning_rate": 9.944739353007344e-05,
      "loss": 4.8003,
      "step": 19
    },
    {
      "epoch": 0.01695993216027136,
      "grad_norm": 3.3515334129333496,
      "learning_rate": 9.931806517013612e-05,
      "loss": 4.6889,
      "step": 20
    },
    {
      "epoch": 0.017807928768284926,
      "grad_norm": 3.2630534172058105,
      "learning_rate": 9.917525374361912e-05,
      "loss": 5.204,
      "step": 21
    },
    {
      "epoch": 0.018655925376298496,
      "grad_norm": 3.223846197128296,
      "learning_rate": 9.901899829374047e-05,
      "loss": 5.0531,
      "step": 22
    },
    {
      "epoch": 0.019503921984312063,
      "grad_norm": 2.922436475753784,
      "learning_rate": 9.884934153917997e-05,
      "loss": 4.8231,
      "step": 23
    },
    {
      "epoch": 0.02035191859232563,
      "grad_norm": 2.912968873977661,
      "learning_rate": 9.86663298624003e-05,
      "loss": 4.6071,
      "step": 24
    },
    {
      "epoch": 0.0211999152003392,
      "grad_norm": 3.0953924655914307,
      "learning_rate": 9.847001329696653e-05,
      "loss": 4.4957,
      "step": 25
    },
    {
      "epoch": 0.022047911808352767,
      "grad_norm": 3.4393253326416016,
      "learning_rate": 9.826044551386744e-05,
      "loss": 4.7432,
      "step": 26
    },
    {
      "epoch": 0.022895908416366333,
      "grad_norm": 3.0531647205352783,
      "learning_rate": 9.803768380684242e-05,
      "loss": 4.58,
      "step": 27
    },
    {
      "epoch": 0.023743905024379904,
      "grad_norm": 2.8929591178894043,
      "learning_rate": 9.780178907671789e-05,
      "loss": 5.2947,
      "step": 28
    },
    {
      "epoch": 0.02459190163239347,
      "grad_norm": 2.7970383167266846,
      "learning_rate": 9.755282581475769e-05,
      "loss": 4.4972,
      "step": 29
    },
    {
      "epoch": 0.025439898240407037,
      "grad_norm": 2.789538860321045,
      "learning_rate": 9.729086208503174e-05,
      "loss": 4.6292,
      "step": 30
    },
    {
      "epoch": 0.026287894848420607,
      "grad_norm": 2.835469961166382,
      "learning_rate": 9.701596950580806e-05,
      "loss": 5.2461,
      "step": 31
    },
    {
      "epoch": 0.027135891456434174,
      "grad_norm": 2.9661502838134766,
      "learning_rate": 9.672822322997305e-05,
      "loss": 5.4213,
      "step": 32
    },
    {
      "epoch": 0.02798388806444774,
      "grad_norm": 3.0827746391296387,
      "learning_rate": 9.642770192448536e-05,
      "loss": 4.6553,
      "step": 33
    },
    {
      "epoch": 0.02883188467246131,
      "grad_norm": 3.0094544887542725,
      "learning_rate": 9.611448774886924e-05,
      "loss": 4.2848,
      "step": 34
    },
    {
      "epoch": 0.029679881280474878,
      "grad_norm": 2.9767343997955322,
      "learning_rate": 9.578866633275288e-05,
      "loss": 5.0974,
      "step": 35
    },
    {
      "epoch": 0.030527877888488444,
      "grad_norm": 2.970184326171875,
      "learning_rate": 9.545032675245813e-05,
      "loss": 4.7893,
      "step": 36
    },
    {
      "epoch": 0.03137587449650201,
      "grad_norm": 3.15069580078125,
      "learning_rate": 9.509956150664796e-05,
      "loss": 5.0775,
      "step": 37
    },
    {
      "epoch": 0.032223871104515585,
      "grad_norm": 3.233588933944702,
      "learning_rate": 9.473646649103818e-05,
      "loss": 4.9845,
      "step": 38
    },
    {
      "epoch": 0.03307186771252915,
      "grad_norm": 3.172109365463257,
      "learning_rate": 9.43611409721806e-05,
      "loss": 5.1913,
      "step": 39
    },
    {
      "epoch": 0.03391986432054272,
      "grad_norm": 3.247112512588501,
      "learning_rate": 9.397368756032445e-05,
      "loss": 4.9318,
      "step": 40
    },
    {
      "epoch": 0.034767860928556285,
      "grad_norm": 3.2850818634033203,
      "learning_rate": 9.357421218136386e-05,
      "loss": 4.6583,
      "step": 41
    },
    {
      "epoch": 0.03561585753656985,
      "grad_norm": 3.4368865489959717,
      "learning_rate": 9.316282404787871e-05,
      "loss": 4.9475,
      "step": 42
    },
    {
      "epoch": 0.03646385414458342,
      "grad_norm": 3.4902491569519043,
      "learning_rate": 9.273963562927695e-05,
      "loss": 5.2225,
      "step": 43
    },
    {
      "epoch": 0.03731185075259699,
      "grad_norm": 3.5062997341156006,
      "learning_rate": 9.230476262104677e-05,
      "loss": 4.7834,
      "step": 44
    },
    {
      "epoch": 0.03815984736061056,
      "grad_norm": 3.91382098197937,
      "learning_rate": 9.185832391312644e-05,
      "loss": 4.6601,
      "step": 45
    },
    {
      "epoch": 0.039007843968624126,
      "grad_norm": 4.155858993530273,
      "learning_rate": 9.140044155740101e-05,
      "loss": 4.7702,
      "step": 46
    },
    {
      "epoch": 0.03985584057663769,
      "grad_norm": 3.78229022026062,
      "learning_rate": 9.093124073433463e-05,
      "loss": 5.0361,
      "step": 47
    },
    {
      "epoch": 0.04070383718465126,
      "grad_norm": 4.286041259765625,
      "learning_rate": 9.045084971874738e-05,
      "loss": 4.8115,
      "step": 48
    },
    {
      "epoch": 0.041551833792664826,
      "grad_norm": 4.295352935791016,
      "learning_rate": 8.995939984474624e-05,
      "loss": 4.3678,
      "step": 49
    },
    {
      "epoch": 0.0423998304006784,
      "grad_norm": 5.1579999923706055,
      "learning_rate": 8.945702546981969e-05,
      "loss": 4.8081,
      "step": 50
    },
    {
      "epoch": 0.0423998304006784,
      "eval_loss": 1.2079874277114868,
      "eval_runtime": 368.6217,
      "eval_samples_per_second": 5.39,
      "eval_steps_per_second": 1.348,
      "step": 50
    },
    {
      "epoch": 0.043247827008691966,
      "grad_norm": 2.945244073867798,
      "learning_rate": 8.894386393810563e-05,
      "loss": 4.3313,
      "step": 51
    },
    {
      "epoch": 0.04409582361670553,
      "grad_norm": 2.7942020893096924,
      "learning_rate": 8.842005554284296e-05,
      "loss": 3.9297,
      "step": 52
    },
    {
      "epoch": 0.0449438202247191,
      "grad_norm": 2.118166923522949,
      "learning_rate": 8.788574348801675e-05,
      "loss": 4.3762,
      "step": 53
    },
    {
      "epoch": 0.04579181683273267,
      "grad_norm": 1.9930267333984375,
      "learning_rate": 8.73410738492077e-05,
      "loss": 4.8376,
      "step": 54
    },
    {
      "epoch": 0.04663981344074624,
      "grad_norm": 2.2635605335235596,
      "learning_rate": 8.678619553365659e-05,
      "loss": 4.4252,
      "step": 55
    },
    {
      "epoch": 0.04748781004875981,
      "grad_norm": 2.0338568687438965,
      "learning_rate": 8.622126023955446e-05,
      "loss": 4.1499,
      "step": 56
    },
    {
      "epoch": 0.048335806656773374,
      "grad_norm": 2.1100118160247803,
      "learning_rate": 8.564642241456986e-05,
      "loss": 4.4142,
      "step": 57
    },
    {
      "epoch": 0.04918380326478694,
      "grad_norm": 2.1581358909606934,
      "learning_rate": 8.506183921362443e-05,
      "loss": 4.3231,
      "step": 58
    },
    {
      "epoch": 0.05003179987280051,
      "grad_norm": 2.1608479022979736,
      "learning_rate": 8.44676704559283e-05,
      "loss": 4.3675,
      "step": 59
    },
    {
      "epoch": 0.050879796480814074,
      "grad_norm": 2.024601697921753,
      "learning_rate": 8.386407858128706e-05,
      "loss": 4.975,
      "step": 60
    },
    {
      "epoch": 0.05172779308882765,
      "grad_norm": 2.0843992233276367,
      "learning_rate": 8.32512286056924e-05,
      "loss": 4.706,
      "step": 61
    },
    {
      "epoch": 0.052575789696841214,
      "grad_norm": 2.1028475761413574,
      "learning_rate": 8.262928807620843e-05,
      "loss": 5.2285,
      "step": 62
    },
    {
      "epoch": 0.05342378630485478,
      "grad_norm": 2.36572003364563,
      "learning_rate": 8.199842702516583e-05,
      "loss": 4.849,
      "step": 63
    },
    {
      "epoch": 0.05427178291286835,
      "grad_norm": 2.0823628902435303,
      "learning_rate": 8.135881792367686e-05,
      "loss": 5.1388,
      "step": 64
    },
    {
      "epoch": 0.055119779520881915,
      "grad_norm": 2.204245090484619,
      "learning_rate": 8.07106356344834e-05,
      "loss": 4.8233,
      "step": 65
    },
    {
      "epoch": 0.05596777612889548,
      "grad_norm": 2.037562608718872,
      "learning_rate": 8.005405736415126e-05,
      "loss": 4.6374,
      "step": 66
    },
    {
      "epoch": 0.056815772736909055,
      "grad_norm": 2.247715711593628,
      "learning_rate": 7.938926261462366e-05,
      "loss": 5.0396,
      "step": 67
    },
    {
      "epoch": 0.05766376934492262,
      "grad_norm": 2.378948211669922,
      "learning_rate": 7.871643313414718e-05,
      "loss": 5.002,
      "step": 68
    },
    {
      "epoch": 0.05851176595293619,
      "grad_norm": 2.3862147331237793,
      "learning_rate": 7.803575286758364e-05,
      "loss": 4.7315,
      "step": 69
    },
    {
      "epoch": 0.059359762560949755,
      "grad_norm": 2.530724287033081,
      "learning_rate": 7.734740790612136e-05,
      "loss": 4.9347,
      "step": 70
    },
    {
      "epoch": 0.06020775916896332,
      "grad_norm": 2.448453903198242,
      "learning_rate": 7.66515864363997e-05,
      "loss": 4.4421,
      "step": 71
    },
    {
      "epoch": 0.06105575577697689,
      "grad_norm": 2.2494592666625977,
      "learning_rate": 7.594847868906076e-05,
      "loss": 4.3259,
      "step": 72
    },
    {
      "epoch": 0.06190375238499046,
      "grad_norm": 2.4794130325317383,
      "learning_rate": 7.52382768867422e-05,
      "loss": 4.7365,
      "step": 73
    },
    {
      "epoch": 0.06275174899300402,
      "grad_norm": 2.7880704402923584,
      "learning_rate": 7.452117519152542e-05,
      "loss": 4.6844,
      "step": 74
    },
    {
      "epoch": 0.0635997456010176,
      "grad_norm": 2.77188777923584,
      "learning_rate": 7.379736965185368e-05,
      "loss": 4.9251,
      "step": 75
    },
    {
      "epoch": 0.06444774220903117,
      "grad_norm": 2.695709705352783,
      "learning_rate": 7.30670581489344e-05,
      "loss": 4.4193,
      "step": 76
    },
    {
      "epoch": 0.06529573881704473,
      "grad_norm": 2.6865248680114746,
      "learning_rate": 7.233044034264034e-05,
      "loss": 4.4448,
      "step": 77
    },
    {
      "epoch": 0.0661437354250583,
      "grad_norm": 2.293822765350342,
      "learning_rate": 7.158771761692464e-05,
      "loss": 4.9784,
      "step": 78
    },
    {
      "epoch": 0.06699173203307186,
      "grad_norm": 2.4043514728546143,
      "learning_rate": 7.083909302476453e-05,
      "loss": 4.9327,
      "step": 79
    },
    {
      "epoch": 0.06783972864108544,
      "grad_norm": 2.4730265140533447,
      "learning_rate": 7.008477123264848e-05,
      "loss": 5.0296,
      "step": 80
    },
    {
      "epoch": 0.06868772524909901,
      "grad_norm": 2.3908400535583496,
      "learning_rate": 6.932495846462261e-05,
      "loss": 5.096,
      "step": 81
    },
    {
      "epoch": 0.06953572185711257,
      "grad_norm": 2.6292006969451904,
      "learning_rate": 6.855986244591104e-05,
      "loss": 4.7537,
      "step": 82
    },
    {
      "epoch": 0.07038371846512614,
      "grad_norm": 2.5612385272979736,
      "learning_rate": 6.778969234612584e-05,
      "loss": 5.1523,
      "step": 83
    },
    {
      "epoch": 0.0712317150731397,
      "grad_norm": 2.547719717025757,
      "learning_rate": 6.701465872208216e-05,
      "loss": 4.634,
      "step": 84
    },
    {
      "epoch": 0.07207971168115328,
      "grad_norm": 2.726158857345581,
      "learning_rate": 6.623497346023418e-05,
      "loss": 4.921,
      "step": 85
    },
    {
      "epoch": 0.07292770828916684,
      "grad_norm": 2.6802666187286377,
      "learning_rate": 6.545084971874738e-05,
      "loss": 5.0294,
      "step": 86
    },
    {
      "epoch": 0.07377570489718041,
      "grad_norm": 2.8893790245056152,
      "learning_rate": 6.466250186922325e-05,
      "loss": 4.9306,
      "step": 87
    },
    {
      "epoch": 0.07462370150519398,
      "grad_norm": 2.926038980484009,
      "learning_rate": 6.387014543809223e-05,
      "loss": 5.296,
      "step": 88
    },
    {
      "epoch": 0.07547169811320754,
      "grad_norm": 2.9031198024749756,
      "learning_rate": 6.307399704769099e-05,
      "loss": 4.6424,
      "step": 89
    },
    {
      "epoch": 0.07631969472122112,
      "grad_norm": 2.91731595993042,
      "learning_rate": 6.227427435703997e-05,
      "loss": 4.9993,
      "step": 90
    },
    {
      "epoch": 0.07716769132923468,
      "grad_norm": 3.06075119972229,
      "learning_rate": 6.147119600233758e-05,
      "loss": 4.9737,
      "step": 91
    },
    {
      "epoch": 0.07801568793724825,
      "grad_norm": 2.9869635105133057,
      "learning_rate": 6.066498153718735e-05,
      "loss": 4.7087,
      "step": 92
    },
    {
      "epoch": 0.07886368454526183,
      "grad_norm": 3.2028849124908447,
      "learning_rate": 5.985585137257401e-05,
      "loss": 4.7985,
      "step": 93
    },
    {
      "epoch": 0.07971168115327538,
      "grad_norm": 3.2854199409484863,
      "learning_rate": 5.90440267166055e-05,
      "loss": 5.0808,
      "step": 94
    },
    {
      "epoch": 0.08055967776128896,
      "grad_norm": 3.3411152362823486,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 4.2881,
      "step": 95
    },
    {
      "epoch": 0.08140767436930252,
      "grad_norm": 3.3091275691986084,
      "learning_rate": 5.74131823855921e-05,
      "loss": 5.1494,
      "step": 96
    },
    {
      "epoch": 0.08225567097731609,
      "grad_norm": 3.4057445526123047,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 5.0678,
      "step": 97
    },
    {
      "epoch": 0.08310366758532965,
      "grad_norm": 3.694821834564209,
      "learning_rate": 5.577423184847932e-05,
      "loss": 5.1944,
      "step": 98
    },
    {
      "epoch": 0.08395166419334323,
      "grad_norm": 3.8164803981781006,
      "learning_rate": 5.495227651252315e-05,
      "loss": 4.8733,
      "step": 99
    },
    {
      "epoch": 0.0847996608013568,
      "grad_norm": 5.192181587219238,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 4.9282,
      "step": 100
    },
    {
      "epoch": 0.0847996608013568,
      "eval_loss": 1.1804490089416504,
      "eval_runtime": 368.6207,
      "eval_samples_per_second": 5.39,
      "eval_steps_per_second": 1.348,
      "step": 100
    },
    {
      "epoch": 0.08564765740937036,
      "grad_norm": 2.422438144683838,
      "learning_rate": 5.330452921628497e-05,
      "loss": 4.3436,
      "step": 101
    },
    {
      "epoch": 0.08649565401738393,
      "grad_norm": 2.2779412269592285,
      "learning_rate": 5.247918773366112e-05,
      "loss": 4.4265,
      "step": 102
    },
    {
      "epoch": 0.08734365062539749,
      "grad_norm": 2.0400304794311523,
      "learning_rate": 5.165316846586541e-05,
      "loss": 4.1714,
      "step": 103
    },
    {
      "epoch": 0.08819164723341107,
      "grad_norm": 1.880078911781311,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 4.1855,
      "step": 104
    },
    {
      "epoch": 0.08903964384142464,
      "grad_norm": 1.8325084447860718,
      "learning_rate": 5e-05,
      "loss": 4.3204,
      "step": 105
    },
    {
      "epoch": 0.0898876404494382,
      "grad_norm": 1.8795162439346313,
      "learning_rate": 4.917330276168208e-05,
      "loss": 4.4312,
      "step": 106
    },
    {
      "epoch": 0.09073563705745177,
      "grad_norm": 1.8876075744628906,
      "learning_rate": 4.834683153413459e-05,
      "loss": 4.4647,
      "step": 107
    },
    {
      "epoch": 0.09158363366546533,
      "grad_norm": 1.8472651243209839,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 4.6483,
      "step": 108
    },
    {
      "epoch": 0.0924316302734789,
      "grad_norm": 1.9912683963775635,
      "learning_rate": 4.669547078371504e-05,
      "loss": 4.7963,
      "step": 109
    },
    {
      "epoch": 0.09327962688149248,
      "grad_norm": 2.042919635772705,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 5.1027,
      "step": 110
    },
    {
      "epoch": 0.09412762348950604,
      "grad_norm": 2.073084592819214,
      "learning_rate": 4.504772348747687e-05,
      "loss": 4.7349,
      "step": 111
    },
    {
      "epoch": 0.09497562009751961,
      "grad_norm": 2.118652105331421,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 4.3654,
      "step": 112
    },
    {
      "epoch": 0.09582361670553317,
      "grad_norm": 2.092261791229248,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 4.7062,
      "step": 113
    },
    {
      "epoch": 0.09667161331354675,
      "grad_norm": 2.0549087524414062,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 4.8884,
      "step": 114
    },
    {
      "epoch": 0.09751960992156031,
      "grad_norm": 1.9667161703109741,
      "learning_rate": 4.17702704859633e-05,
      "loss": 5.0113,
      "step": 115
    },
    {
      "epoch": 0.09836760652957388,
      "grad_norm": 2.183830499649048,
      "learning_rate": 4.095597328339452e-05,
      "loss": 4.6784,
      "step": 116
    },
    {
      "epoch": 0.09921560313758745,
      "grad_norm": 2.051776885986328,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 5.0518,
      "step": 117
    },
    {
      "epoch": 0.10006359974560101,
      "grad_norm": 2.446779251098633,
      "learning_rate": 3.933501846281267e-05,
      "loss": 4.695,
      "step": 118
    },
    {
      "epoch": 0.10091159635361459,
      "grad_norm": 2.3001549243927,
      "learning_rate": 3.852880399766243e-05,
      "loss": 4.3205,
      "step": 119
    },
    {
      "epoch": 0.10175959296162815,
      "grad_norm": 2.7006070613861084,
      "learning_rate": 3.772572564296005e-05,
      "loss": 4.8592,
      "step": 120
    },
    {
      "epoch": 0.10260758956964172,
      "grad_norm": 2.1881604194641113,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 4.6239,
      "step": 121
    },
    {
      "epoch": 0.1034555861776553,
      "grad_norm": 2.380777597427368,
      "learning_rate": 3.612985456190778e-05,
      "loss": 5.3121,
      "step": 122
    },
    {
      "epoch": 0.10430358278566886,
      "grad_norm": 2.426774024963379,
      "learning_rate": 3.533749813077677e-05,
      "loss": 5.1067,
      "step": 123
    },
    {
      "epoch": 0.10515157939368243,
      "grad_norm": 2.5026845932006836,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 4.1341,
      "step": 124
    },
    {
      "epoch": 0.10599957600169599,
      "grad_norm": 2.3385419845581055,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 4.9654,
      "step": 125
    },
    {
      "epoch": 0.10684757260970956,
      "grad_norm": 2.7509677410125732,
      "learning_rate": 3.298534127791785e-05,
      "loss": 4.3466,
      "step": 126
    },
    {
      "epoch": 0.10769556921772312,
      "grad_norm": 2.9490578174591064,
      "learning_rate": 3.221030765387417e-05,
      "loss": 5.1551,
      "step": 127
    },
    {
      "epoch": 0.1085435658257367,
      "grad_norm": 2.265023946762085,
      "learning_rate": 3.144013755408895e-05,
      "loss": 4.798,
      "step": 128
    },
    {
      "epoch": 0.10939156243375027,
      "grad_norm": 2.2808432579040527,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 4.588,
      "step": 129
    },
    {
      "epoch": 0.11023955904176383,
      "grad_norm": 2.5440714359283447,
      "learning_rate": 2.991522876735154e-05,
      "loss": 5.2281,
      "step": 130
    },
    {
      "epoch": 0.1110875556497774,
      "grad_norm": 2.297882556915283,
      "learning_rate": 2.916090697523549e-05,
      "loss": 4.8398,
      "step": 131
    },
    {
      "epoch": 0.11193555225779096,
      "grad_norm": 2.478205919265747,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 4.3081,
      "step": 132
    },
    {
      "epoch": 0.11278354886580454,
      "grad_norm": 2.669485330581665,
      "learning_rate": 2.766955965735968e-05,
      "loss": 4.8905,
      "step": 133
    },
    {
      "epoch": 0.11363154547381811,
      "grad_norm": 2.624321222305298,
      "learning_rate": 2.693294185106562e-05,
      "loss": 4.794,
      "step": 134
    },
    {
      "epoch": 0.11447954208183167,
      "grad_norm": 2.6909666061401367,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 5.1577,
      "step": 135
    },
    {
      "epoch": 0.11532753868984524,
      "grad_norm": 2.5199904441833496,
      "learning_rate": 2.547882480847461e-05,
      "loss": 4.4238,
      "step": 136
    },
    {
      "epoch": 0.1161755352978588,
      "grad_norm": 2.583390712738037,
      "learning_rate": 2.476172311325783e-05,
      "loss": 4.7589,
      "step": 137
    },
    {
      "epoch": 0.11702353190587238,
      "grad_norm": 3.0719287395477295,
      "learning_rate": 2.405152131093926e-05,
      "loss": 5.1474,
      "step": 138
    },
    {
      "epoch": 0.11787152851388595,
      "grad_norm": 3.1999142169952393,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 4.6627,
      "step": 139
    },
    {
      "epoch": 0.11871952512189951,
      "grad_norm": 2.7626335620880127,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 5.1589,
      "step": 140
    },
    {
      "epoch": 0.11956752172991308,
      "grad_norm": 2.6776087284088135,
      "learning_rate": 2.196424713241637e-05,
      "loss": 4.9263,
      "step": 141
    },
    {
      "epoch": 0.12041551833792664,
      "grad_norm": 2.8333215713500977,
      "learning_rate": 2.128356686585282e-05,
      "loss": 5.1223,
      "step": 142
    },
    {
      "epoch": 0.12126351494594022,
      "grad_norm": 2.761094093322754,
      "learning_rate": 2.061073738537635e-05,
      "loss": 4.9352,
      "step": 143
    },
    {
      "epoch": 0.12211151155395378,
      "grad_norm": 3.142958402633667,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 4.7899,
      "step": 144
    },
    {
      "epoch": 0.12295950816196735,
      "grad_norm": 3.279120445251465,
      "learning_rate": 1.928936436551661e-05,
      "loss": 4.433,
      "step": 145
    },
    {
      "epoch": 0.12380750476998093,
      "grad_norm": 3.364237070083618,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 4.5097,
      "step": 146
    },
    {
      "epoch": 0.12465550137799448,
      "grad_norm": 3.385071039199829,
      "learning_rate": 1.800157297483417e-05,
      "loss": 4.753,
      "step": 147
    },
    {
      "epoch": 0.12550349798600804,
      "grad_norm": 3.5379798412323,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 4.9349,
      "step": 148
    },
    {
      "epoch": 0.12635149459402162,
      "grad_norm": 4.043220043182373,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 5.0459,
      "step": 149
    },
    {
      "epoch": 0.1271994912020352,
      "grad_norm": 4.804100036621094,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.3865,
      "step": 150
    },
    {
      "epoch": 0.1271994912020352,
      "eval_loss": 1.166787028312683,
      "eval_runtime": 368.1394,
      "eval_samples_per_second": 5.397,
      "eval_steps_per_second": 1.35,
      "step": 150
    },
    {
      "epoch": 0.12804748781004877,
      "grad_norm": 1.7919011116027832,
      "learning_rate": 1.553232954407171e-05,
      "loss": 3.9199,
      "step": 151
    },
    {
      "epoch": 0.12889548441806234,
      "grad_norm": 1.9280797243118286,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 4.4191,
      "step": 152
    },
    {
      "epoch": 0.12974348102607589,
      "grad_norm": 1.8384506702423096,
      "learning_rate": 1.435357758543015e-05,
      "loss": 3.5721,
      "step": 153
    },
    {
      "epoch": 0.13059147763408946,
      "grad_norm": 1.9051966667175293,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 4.1426,
      "step": 154
    },
    {
      "epoch": 0.13143947424210303,
      "grad_norm": 2.434952974319458,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 4.1037,
      "step": 155
    },
    {
      "epoch": 0.1322874708501166,
      "grad_norm": 1.8974602222442627,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 4.0044,
      "step": 156
    },
    {
      "epoch": 0.13313546745813018,
      "grad_norm": 1.922234058380127,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 4.5226,
      "step": 157
    },
    {
      "epoch": 0.13398346406614373,
      "grad_norm": 1.8575645685195923,
      "learning_rate": 1.157994445715706e-05,
      "loss": 4.7192,
      "step": 158
    },
    {
      "epoch": 0.1348314606741573,
      "grad_norm": 1.8708919286727905,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 4.3725,
      "step": 159
    },
    {
      "epoch": 0.13567945728217087,
      "grad_norm": 1.8756542205810547,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 4.3824,
      "step": 160
    },
    {
      "epoch": 0.13652745389018445,
      "grad_norm": 1.9296820163726807,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 5.2585,
      "step": 161
    },
    {
      "epoch": 0.13737545049819802,
      "grad_norm": 1.8444390296936035,
      "learning_rate": 9.549150281252633e-06,
      "loss": 5.4111,
      "step": 162
    },
    {
      "epoch": 0.13822344710621157,
      "grad_norm": 1.8733844757080078,
      "learning_rate": 9.068759265665384e-06,
      "loss": 5.4443,
      "step": 163
    },
    {
      "epoch": 0.13907144371422514,
      "grad_norm": 2.009781837463379,
      "learning_rate": 8.599558442598998e-06,
      "loss": 5.3441,
      "step": 164
    },
    {
      "epoch": 0.13991944032223871,
      "grad_norm": 2.064223289489746,
      "learning_rate": 8.141676086873572e-06,
      "loss": 4.3621,
      "step": 165
    },
    {
      "epoch": 0.1407674369302523,
      "grad_norm": 2.1569154262542725,
      "learning_rate": 7.695237378953223e-06,
      "loss": 4.9552,
      "step": 166
    },
    {
      "epoch": 0.14161543353826583,
      "grad_norm": 2.096059560775757,
      "learning_rate": 7.260364370723044e-06,
      "loss": 4.446,
      "step": 167
    },
    {
      "epoch": 0.1424634301462794,
      "grad_norm": 2.1552207469940186,
      "learning_rate": 6.837175952121306e-06,
      "loss": 5.1089,
      "step": 168
    },
    {
      "epoch": 0.14331142675429298,
      "grad_norm": 2.1895458698272705,
      "learning_rate": 6.425787818636131e-06,
      "loss": 4.8667,
      "step": 169
    },
    {
      "epoch": 0.14415942336230655,
      "grad_norm": 2.401721239089966,
      "learning_rate": 6.026312439675552e-06,
      "loss": 4.6076,
      "step": 170
    },
    {
      "epoch": 0.14500741997032013,
      "grad_norm": 2.207214593887329,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 4.4114,
      "step": 171
    },
    {
      "epoch": 0.14585541657833367,
      "grad_norm": 2.435170888900757,
      "learning_rate": 5.263533508961827e-06,
      "loss": 4.6796,
      "step": 172
    },
    {
      "epoch": 0.14670341318634725,
      "grad_norm": 2.510496139526367,
      "learning_rate": 4.900438493352055e-06,
      "loss": 5.1973,
      "step": 173
    },
    {
      "epoch": 0.14755140979436082,
      "grad_norm": 2.4800169467926025,
      "learning_rate": 4.549673247541875e-06,
      "loss": 4.6695,
      "step": 174
    },
    {
      "epoch": 0.1483994064023744,
      "grad_norm": 2.383852005004883,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 4.8739,
      "step": 175
    },
    {
      "epoch": 0.14924740301038797,
      "grad_norm": 2.5435712337493896,
      "learning_rate": 3.885512251130763e-06,
      "loss": 4.8202,
      "step": 176
    },
    {
      "epoch": 0.15009539961840151,
      "grad_norm": 2.2736432552337646,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 4.8861,
      "step": 177
    },
    {
      "epoch": 0.1509433962264151,
      "grad_norm": 2.577716827392578,
      "learning_rate": 3.271776770026963e-06,
      "loss": 4.5959,
      "step": 178
    },
    {
      "epoch": 0.15179139283442866,
      "grad_norm": 2.394709825515747,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 5.2788,
      "step": 179
    },
    {
      "epoch": 0.15263938944244224,
      "grad_norm": 2.184279441833496,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 5.1552,
      "step": 180
    },
    {
      "epoch": 0.1534873860504558,
      "grad_norm": 2.4257490634918213,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 4.7869,
      "step": 181
    },
    {
      "epoch": 0.15433538265846936,
      "grad_norm": 2.295222759246826,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 4.3428,
      "step": 182
    },
    {
      "epoch": 0.15518337926648293,
      "grad_norm": 2.3843867778778076,
      "learning_rate": 1.962316193157593e-06,
      "loss": 5.0517,
      "step": 183
    },
    {
      "epoch": 0.1560313758744965,
      "grad_norm": 2.3849644660949707,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 4.6856,
      "step": 184
    },
    {
      "epoch": 0.15687937248251008,
      "grad_norm": 2.5821375846862793,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 4.898,
      "step": 185
    },
    {
      "epoch": 0.15772736909052365,
      "grad_norm": 2.6807963848114014,
      "learning_rate": 1.333670137599713e-06,
      "loss": 4.6862,
      "step": 186
    },
    {
      "epoch": 0.1585753656985372,
      "grad_norm": 2.4151716232299805,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 4.7081,
      "step": 187
    },
    {
      "epoch": 0.15942336230655077,
      "grad_norm": 2.5919737815856934,
      "learning_rate": 9.810017062595322e-07,
      "loss": 5.1155,
      "step": 188
    },
    {
      "epoch": 0.16027135891456434,
      "grad_norm": 2.806579113006592,
      "learning_rate": 8.247462563808817e-07,
      "loss": 4.5753,
      "step": 189
    },
    {
      "epoch": 0.16111935552257792,
      "grad_norm": 2.768523931503296,
      "learning_rate": 6.819348298638839e-07,
      "loss": 4.6045,
      "step": 190
    },
    {
      "epoch": 0.1619673521305915,
      "grad_norm": 3.063878059387207,
      "learning_rate": 5.526064699265753e-07,
      "loss": 4.5476,
      "step": 191
    },
    {
      "epoch": 0.16281534873860504,
      "grad_norm": 2.8875696659088135,
      "learning_rate": 4.367965336512403e-07,
      "loss": 5.0934,
      "step": 192
    },
    {
      "epoch": 0.1636633453466186,
      "grad_norm": 2.884516954421997,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 4.6816,
      "step": 193
    },
    {
      "epoch": 0.16451134195463218,
      "grad_norm": 2.838547468185425,
      "learning_rate": 2.458548727494292e-07,
      "loss": 4.3913,
      "step": 194
    },
    {
      "epoch": 0.16535933856264576,
      "grad_norm": 3.1088197231292725,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 4.5998,
      "step": 195
    },
    {
      "epoch": 0.1662073351706593,
      "grad_norm": 3.092036485671997,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 4.7223,
      "step": 196
    },
    {
      "epoch": 0.16705533177867288,
      "grad_norm": 3.4671244621276855,
      "learning_rate": 6.150154258476315e-08,
      "loss": 5.162,
      "step": 197
    },
    {
      "epoch": 0.16790332838668645,
      "grad_norm": 3.778698682785034,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 4.36,
      "step": 198
    },
    {
      "epoch": 0.16875132499470002,
      "grad_norm": 3.9721970558166504,
      "learning_rate": 6.834750376549792e-09,
      "loss": 4.5964,
      "step": 199
    },
    {
      "epoch": 0.1695993216027136,
      "grad_norm": 5.212419509887695,
      "learning_rate": 0.0,
      "loss": 4.5932,
      "step": 200
    },
    {
      "epoch": 0.1695993216027136,
      "eval_loss": 1.1545281410217285,
      "eval_runtime": 368.2929,
      "eval_samples_per_second": 5.395,
      "eval_steps_per_second": 1.349,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.568461143693066e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}