{
  "best_metric": 1.6314605474472046,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.4035087719298245,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.007017543859649123,
      "grad_norm": 0.5051743388175964,
      "learning_rate": 1e-05,
      "loss": 2.1471,
      "step": 1
    },
    {
      "epoch": 0.007017543859649123,
      "eval_loss": 2.996117115020752,
      "eval_runtime": 3.271,
      "eval_samples_per_second": 73.372,
      "eval_steps_per_second": 18.343,
      "step": 1
    },
    {
      "epoch": 0.014035087719298246,
      "grad_norm": 0.7480506300926208,
      "learning_rate": 2e-05,
      "loss": 2.7872,
      "step": 2
    },
    {
      "epoch": 0.021052631578947368,
      "grad_norm": 0.7710661888122559,
      "learning_rate": 3e-05,
      "loss": 2.9703,
      "step": 3
    },
    {
      "epoch": 0.028070175438596492,
      "grad_norm": 0.771657407283783,
      "learning_rate": 4e-05,
      "loss": 2.9018,
      "step": 4
    },
    {
      "epoch": 0.03508771929824561,
      "grad_norm": 0.8080238699913025,
      "learning_rate": 5e-05,
      "loss": 2.978,
      "step": 5
    },
    {
      "epoch": 0.042105263157894736,
      "grad_norm": 0.7472100257873535,
      "learning_rate": 6e-05,
      "loss": 2.847,
      "step": 6
    },
    {
      "epoch": 0.04912280701754386,
      "grad_norm": 0.7656800150871277,
      "learning_rate": 7e-05,
      "loss": 2.8421,
      "step": 7
    },
    {
      "epoch": 0.056140350877192984,
      "grad_norm": 0.7420645952224731,
      "learning_rate": 8e-05,
      "loss": 2.8205,
      "step": 8
    },
    {
      "epoch": 0.06315789473684211,
      "grad_norm": 0.7433472275733948,
      "learning_rate": 9e-05,
      "loss": 2.833,
      "step": 9
    },
    {
      "epoch": 0.07017543859649122,
      "grad_norm": 0.6302344799041748,
      "learning_rate": 0.0001,
      "loss": 2.8698,
      "step": 10
    },
    {
      "epoch": 0.07719298245614035,
      "grad_norm": 0.643413245677948,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.6993,
      "step": 11
    },
    {
      "epoch": 0.08421052631578947,
      "grad_norm": 0.5664858222007751,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.6516,
      "step": 12
    },
    {
      "epoch": 0.0912280701754386,
      "grad_norm": 0.6126735806465149,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.5619,
      "step": 13
    },
    {
      "epoch": 0.09824561403508772,
      "grad_norm": 0.5654096603393555,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.6035,
      "step": 14
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 0.5275979042053223,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.5133,
      "step": 15
    },
    {
      "epoch": 0.11228070175438597,
      "grad_norm": 0.5334392786026001,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.5076,
      "step": 16
    },
    {
      "epoch": 0.11929824561403508,
      "grad_norm": 0.5243760347366333,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.4712,
      "step": 17
    },
    {
      "epoch": 0.12631578947368421,
      "grad_norm": 0.5230803489685059,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.1886,
      "step": 18
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 0.5679934620857239,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.4104,
      "step": 19
    },
    {
      "epoch": 0.14035087719298245,
      "grad_norm": 0.5256866812705994,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.2721,
      "step": 20
    },
    {
      "epoch": 0.14736842105263157,
      "grad_norm": 0.5222025513648987,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.2976,
      "step": 21
    },
    {
      "epoch": 0.1543859649122807,
      "grad_norm": 0.4829598367214203,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.2854,
      "step": 22
    },
    {
      "epoch": 0.16140350877192983,
      "grad_norm": 0.4910241663455963,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.262,
      "step": 23
    },
    {
      "epoch": 0.16842105263157894,
      "grad_norm": 0.483591228723526,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.1533,
      "step": 24
    },
    {
      "epoch": 0.17543859649122806,
      "grad_norm": 0.47343581914901733,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.2513,
      "step": 25
    },
    {
      "epoch": 0.1824561403508772,
      "grad_norm": 0.42493799328804016,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.0971,
      "step": 26
    },
    {
      "epoch": 0.18947368421052632,
      "grad_norm": 0.4053170382976532,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.1213,
      "step": 27
    },
    {
      "epoch": 0.19649122807017544,
      "grad_norm": 0.44475165009498596,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.1416,
      "step": 28
    },
    {
      "epoch": 0.20350877192982456,
      "grad_norm": 0.4029024541378021,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.0392,
      "step": 29
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.42991721630096436,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.103,
      "step": 30
    },
    {
      "epoch": 0.21754385964912282,
      "grad_norm": 0.39374154806137085,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.1059,
      "step": 31
    },
    {
      "epoch": 0.22456140350877193,
      "grad_norm": 0.41946330666542053,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.0418,
      "step": 32
    },
    {
      "epoch": 0.23157894736842105,
      "grad_norm": 0.4189073145389557,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.988,
      "step": 33
    },
    {
      "epoch": 0.23859649122807017,
      "grad_norm": 0.45518365502357483,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1021,
      "step": 34
    },
    {
      "epoch": 0.24561403508771928,
      "grad_norm": 0.5687738060951233,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.1288,
      "step": 35
    },
    {
      "epoch": 0.25263157894736843,
      "grad_norm": 0.4418773055076599,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.7965,
      "step": 36
    },
    {
      "epoch": 0.2596491228070175,
      "grad_norm": 0.387173593044281,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.9161,
      "step": 37
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 0.3450778126716614,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.7996,
      "step": 38
    },
    {
      "epoch": 0.2736842105263158,
      "grad_norm": 0.4097825884819031,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.9505,
      "step": 39
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 0.3776609003543854,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.8893,
      "step": 40
    },
    {
      "epoch": 0.28771929824561404,
      "grad_norm": 0.37853214144706726,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.9843,
      "step": 41
    },
    {
      "epoch": 0.29473684210526313,
      "grad_norm": 0.3810253441333771,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.9512,
      "step": 42
    },
    {
      "epoch": 0.3017543859649123,
      "grad_norm": 0.318187415599823,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.843,
      "step": 43
    },
    {
      "epoch": 0.3087719298245614,
      "grad_norm": 0.33174723386764526,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.9003,
      "step": 44
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 0.3459150195121765,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.9464,
      "step": 45
    },
    {
      "epoch": 0.32280701754385965,
      "grad_norm": 0.335860937833786,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.0076,
      "step": 46
    },
    {
      "epoch": 0.3298245614035088,
      "grad_norm": 0.3969382643699646,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.0486,
      "step": 47
    },
    {
      "epoch": 0.3368421052631579,
      "grad_norm": 0.4115678668022156,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.8113,
      "step": 48
    },
    {
      "epoch": 0.34385964912280703,
      "grad_norm": 0.35497140884399414,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.8439,
      "step": 49
    },
    {
      "epoch": 0.3508771929824561,
      "grad_norm": 0.34050214290618896,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.8737,
      "step": 50
    },
    {
      "epoch": 0.3508771929824561,
      "eval_loss": 1.8202682733535767,
      "eval_runtime": 3.2667,
      "eval_samples_per_second": 73.469,
      "eval_steps_per_second": 18.367,
      "step": 50
    },
    {
      "epoch": 0.35789473684210527,
      "grad_norm": 0.41991162300109863,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.0116,
      "step": 51
    },
    {
      "epoch": 0.3649122807017544,
      "grad_norm": 0.3535086214542389,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.9474,
      "step": 52
    },
    {
      "epoch": 0.3719298245614035,
      "grad_norm": 0.34679949283599854,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.8764,
      "step": 53
    },
    {
      "epoch": 0.37894736842105264,
      "grad_norm": 0.33437803387641907,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.7393,
      "step": 54
    },
    {
      "epoch": 0.38596491228070173,
      "grad_norm": 0.3942807912826538,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.7551,
      "step": 55
    },
    {
      "epoch": 0.3929824561403509,
      "grad_norm": 0.3683478534221649,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.7336,
      "step": 56
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3137025237083435,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.7814,
      "step": 57
    },
    {
      "epoch": 0.4070175438596491,
      "grad_norm": 0.3721928298473358,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.8202,
      "step": 58
    },
    {
      "epoch": 0.41403508771929826,
      "grad_norm": 0.3355889618396759,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.8055,
      "step": 59
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.33667171001434326,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.7569,
      "step": 60
    },
    {
      "epoch": 0.4280701754385965,
      "grad_norm": 0.35266298055648804,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.7909,
      "step": 61
    },
    {
      "epoch": 0.43508771929824563,
      "grad_norm": 0.3281211256980896,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.6775,
      "step": 62
    },
    {
      "epoch": 0.4421052631578947,
      "grad_norm": 0.3804435729980469,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.7461,
      "step": 63
    },
    {
      "epoch": 0.44912280701754387,
      "grad_norm": 0.3238387405872345,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.6718,
      "step": 64
    },
    {
      "epoch": 0.45614035087719296,
      "grad_norm": 0.3669014573097229,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.8564,
      "step": 65
    },
    {
      "epoch": 0.4631578947368421,
      "grad_norm": 0.34674087166786194,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.6807,
      "step": 66
    },
    {
      "epoch": 0.47017543859649125,
      "grad_norm": 0.37510576844215393,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.7713,
      "step": 67
    },
    {
      "epoch": 0.47719298245614034,
      "grad_norm": 0.38316330313682556,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.77,
      "step": 68
    },
    {
      "epoch": 0.4842105263157895,
      "grad_norm": 0.3848925530910492,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.7144,
      "step": 69
    },
    {
      "epoch": 0.49122807017543857,
      "grad_norm": 0.5103507041931152,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.882,
      "step": 70
    },
    {
      "epoch": 0.4982456140350877,
      "grad_norm": 0.5755714178085327,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.4924,
      "step": 71
    },
    {
      "epoch": 0.5052631578947369,
      "grad_norm": 0.4698088467121124,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.6765,
      "step": 72
    },
    {
      "epoch": 0.512280701754386,
      "grad_norm": 0.5318788886070251,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.7607,
      "step": 73
    },
    {
      "epoch": 0.519298245614035,
      "grad_norm": 0.36463528871536255,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.572,
      "step": 74
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 0.32310420274734497,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.7084,
      "step": 75
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.3288344144821167,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.7272,
      "step": 76
    },
    {
      "epoch": 0.5403508771929825,
      "grad_norm": 0.3782610297203064,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.7809,
      "step": 77
    },
    {
      "epoch": 0.5473684210526316,
      "grad_norm": 0.3587157726287842,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.7432,
      "step": 78
    },
    {
      "epoch": 0.5543859649122806,
      "grad_norm": 0.44228455424308777,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.8167,
      "step": 79
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 0.41848447918891907,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.8394,
      "step": 80
    },
    {
      "epoch": 0.5684210526315789,
      "grad_norm": 0.3970872461795807,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.8224,
      "step": 81
    },
    {
      "epoch": 0.5754385964912281,
      "grad_norm": 0.34348198771476746,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.7333,
      "step": 82
    },
    {
      "epoch": 0.5824561403508772,
      "grad_norm": 0.32690948247909546,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.7369,
      "step": 83
    },
    {
      "epoch": 0.5894736842105263,
      "grad_norm": 0.3117843568325043,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.939,
      "step": 84
    },
    {
      "epoch": 0.5964912280701754,
      "grad_norm": 0.32997897267341614,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.7508,
      "step": 85
    },
    {
      "epoch": 0.6035087719298246,
      "grad_norm": 0.3241843283176422,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.7476,
      "step": 86
    },
    {
      "epoch": 0.6105263157894737,
      "grad_norm": 0.33179211616516113,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.7788,
      "step": 87
    },
    {
      "epoch": 0.6175438596491228,
      "grad_norm": 0.34552618861198425,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.7396,
      "step": 88
    },
    {
      "epoch": 0.624561403508772,
      "grad_norm": 0.3433299660682678,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.665,
      "step": 89
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.31586334109306335,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.7527,
      "step": 90
    },
    {
      "epoch": 0.6385964912280702,
      "grad_norm": 0.3767358064651489,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.7517,
      "step": 91
    },
    {
      "epoch": 0.6456140350877193,
      "grad_norm": 0.32300201058387756,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.6961,
      "step": 92
    },
    {
      "epoch": 0.6526315789473685,
      "grad_norm": 0.32934755086898804,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.654,
      "step": 93
    },
    {
      "epoch": 0.6596491228070176,
      "grad_norm": 0.36839109659194946,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.8184,
      "step": 94
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.35935112833976746,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.6511,
      "step": 95
    },
    {
      "epoch": 0.6736842105263158,
      "grad_norm": 0.3269180655479431,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.6021,
      "step": 96
    },
    {
      "epoch": 0.6807017543859649,
      "grad_norm": 0.35928547382354736,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.6459,
      "step": 97
    },
    {
      "epoch": 0.6877192982456141,
      "grad_norm": 0.3492797613143921,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.6723,
      "step": 98
    },
    {
      "epoch": 0.6947368421052632,
      "grad_norm": 0.35333722829818726,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.7135,
      "step": 99
    },
    {
      "epoch": 0.7017543859649122,
      "grad_norm": 0.32469287514686584,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.7398,
      "step": 100
    },
    {
      "epoch": 0.7017543859649122,
      "eval_loss": 1.672231912612915,
      "eval_runtime": 3.273,
      "eval_samples_per_second": 73.327,
      "eval_steps_per_second": 18.332,
      "step": 100
    },
    {
      "epoch": 0.7087719298245614,
      "grad_norm": 0.3625134825706482,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.6476,
      "step": 101
    },
    {
      "epoch": 0.7157894736842105,
      "grad_norm": 0.4348846971988678,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.6806,
      "step": 102
    },
    {
      "epoch": 0.7228070175438597,
      "grad_norm": 0.3913838565349579,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.6648,
      "step": 103
    },
    {
      "epoch": 0.7298245614035088,
      "grad_norm": 0.39361119270324707,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.6907,
      "step": 104
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 0.512360692024231,
      "learning_rate": 5e-05,
      "loss": 1.7565,
      "step": 105
    },
    {
      "epoch": 0.743859649122807,
      "grad_norm": 0.6451547145843506,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.5585,
      "step": 106
    },
    {
      "epoch": 0.7508771929824561,
      "grad_norm": 0.5572435855865479,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.5856,
      "step": 107
    },
    {
      "epoch": 0.7578947368421053,
      "grad_norm": 0.5742175579071045,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.7551,
      "step": 108
    },
    {
      "epoch": 0.7649122807017544,
      "grad_norm": 0.5322205424308777,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.6704,
      "step": 109
    },
    {
      "epoch": 0.7719298245614035,
      "grad_norm": 0.5029052495956421,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.7025,
      "step": 110
    },
    {
      "epoch": 0.7789473684210526,
      "grad_norm": 0.3639456331729889,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.672,
      "step": 111
    },
    {
      "epoch": 0.7859649122807018,
      "grad_norm": 0.4021667242050171,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.8145,
      "step": 112
    },
    {
      "epoch": 0.7929824561403509,
      "grad_norm": 0.3488664925098419,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.6062,
      "step": 113
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2994093596935272,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.6521,
      "step": 114
    },
    {
      "epoch": 0.8070175438596491,
      "grad_norm": 0.3659916818141937,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.6726,
      "step": 115
    },
    {
      "epoch": 0.8140350877192982,
      "grad_norm": 0.36561453342437744,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.5975,
      "step": 116
    },
    {
      "epoch": 0.8210526315789474,
      "grad_norm": 0.3824693560600281,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.7223,
      "step": 117
    },
    {
      "epoch": 0.8280701754385965,
      "grad_norm": 0.38324618339538574,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.6543,
      "step": 118
    },
    {
      "epoch": 0.8350877192982457,
      "grad_norm": 0.4729977548122406,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.7257,
      "step": 119
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.37248340249061584,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.704,
      "step": 120
    },
    {
      "epoch": 0.8491228070175438,
      "grad_norm": 0.37331002950668335,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.7647,
      "step": 121
    },
    {
      "epoch": 0.856140350877193,
      "grad_norm": 0.37321344017982483,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.71,
      "step": 122
    },
    {
      "epoch": 0.8631578947368421,
      "grad_norm": 0.353610634803772,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.5953,
      "step": 123
    },
    {
      "epoch": 0.8701754385964913,
      "grad_norm": 0.34108981490135193,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.7361,
      "step": 124
    },
    {
      "epoch": 0.8771929824561403,
      "grad_norm": 0.35363441705703735,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.6088,
      "step": 125
    },
    {
      "epoch": 0.8842105263157894,
      "grad_norm": 0.3262953460216522,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.7244,
      "step": 126
    },
    {
      "epoch": 0.8912280701754386,
      "grad_norm": 0.33338385820388794,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.6388,
      "step": 127
    },
    {
      "epoch": 0.8982456140350877,
      "grad_norm": 0.32990458607673645,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.6955,
      "step": 128
    },
    {
      "epoch": 0.9052631578947369,
      "grad_norm": 0.33621975779533386,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.595,
      "step": 129
    },
    {
      "epoch": 0.9122807017543859,
      "grad_norm": 0.32537221908569336,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.6951,
      "step": 130
    },
    {
      "epoch": 0.9192982456140351,
      "grad_norm": 0.350075364112854,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.7636,
      "step": 131
    },
    {
      "epoch": 0.9263157894736842,
      "grad_norm": 0.30533602833747864,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.5656,
      "step": 132
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 0.33069756627082825,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.6085,
      "step": 133
    },
    {
      "epoch": 0.9403508771929825,
      "grad_norm": 0.3743823766708374,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.6508,
      "step": 134
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 0.3827040493488312,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.644,
      "step": 135
    },
    {
      "epoch": 0.9543859649122807,
      "grad_norm": 0.39691340923309326,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.6019,
      "step": 136
    },
    {
      "epoch": 0.9614035087719298,
      "grad_norm": 0.371954083442688,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.566,
      "step": 137
    },
    {
      "epoch": 0.968421052631579,
      "grad_norm": 0.3848728537559509,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.6273,
      "step": 138
    },
    {
      "epoch": 0.9754385964912281,
      "grad_norm": 0.45733651518821716,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.6613,
      "step": 139
    },
    {
      "epoch": 0.9824561403508771,
      "grad_norm": 0.5124295353889465,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.6957,
      "step": 140
    },
    {
      "epoch": 0.9894736842105263,
      "grad_norm": 0.4083141088485718,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.6724,
      "step": 141
    },
    {
      "epoch": 0.9964912280701754,
      "grad_norm": 0.3951103687286377,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.5819,
      "step": 142
    },
    {
      "epoch": 1.0035087719298246,
      "grad_norm": 0.6020079851150513,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.5424,
      "step": 143
    },
    {
      "epoch": 1.0105263157894737,
      "grad_norm": 0.4805242121219635,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.4509,
      "step": 144
    },
    {
      "epoch": 1.0175438596491229,
      "grad_norm": 0.42455071210861206,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.6233,
      "step": 145
    },
    {
      "epoch": 1.024561403508772,
      "grad_norm": 0.4013636112213135,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.6855,
      "step": 146
    },
    {
      "epoch": 1.0315789473684212,
      "grad_norm": 0.405375212430954,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.5721,
      "step": 147
    },
    {
      "epoch": 1.03859649122807,
      "grad_norm": 0.39946234226226807,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.5711,
      "step": 148
    },
    {
      "epoch": 1.0456140350877192,
      "grad_norm": 0.3341842293739319,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.6196,
      "step": 149
    },
    {
      "epoch": 1.0526315789473684,
      "grad_norm": 0.38158881664276123,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.7341,
      "step": 150
    },
    {
      "epoch": 1.0526315789473684,
      "eval_loss": 1.6389861106872559,
      "eval_runtime": 3.2692,
      "eval_samples_per_second": 73.413,
      "eval_steps_per_second": 18.353,
      "step": 150
    },
    {
      "epoch": 1.0596491228070175,
      "grad_norm": 0.33268940448760986,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.6369,
      "step": 151
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 0.31823140382766724,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.5936,
      "step": 152
    },
    {
      "epoch": 1.0736842105263158,
      "grad_norm": 0.3143119812011719,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.7404,
      "step": 153
    },
    {
      "epoch": 1.080701754385965,
      "grad_norm": 0.34440138936042786,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.672,
      "step": 154
    },
    {
      "epoch": 1.087719298245614,
      "grad_norm": 0.32974791526794434,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.7157,
      "step": 155
    },
    {
      "epoch": 1.0947368421052632,
      "grad_norm": 0.3708537220954895,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.5872,
      "step": 156
    },
    {
      "epoch": 1.1017543859649124,
      "grad_norm": 0.3071533143520355,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.765,
      "step": 157
    },
    {
      "epoch": 1.1087719298245613,
      "grad_norm": 0.3035176396369934,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.7027,
      "step": 158
    },
    {
      "epoch": 1.1157894736842104,
      "grad_norm": 0.3310808837413788,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.6664,
      "step": 159
    },
    {
      "epoch": 1.1228070175438596,
      "grad_norm": 0.34847113490104675,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.635,
      "step": 160
    },
    {
      "epoch": 1.1298245614035087,
      "grad_norm": 0.3538299798965454,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.6584,
      "step": 161
    },
    {
      "epoch": 1.1368421052631579,
      "grad_norm": 0.32887282967567444,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.5728,
      "step": 162
    },
    {
      "epoch": 1.143859649122807,
      "grad_norm": 0.3403266966342926,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.6041,
      "step": 163
    },
    {
      "epoch": 1.1508771929824562,
      "grad_norm": 0.3529265224933624,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.6052,
      "step": 164
    },
    {
      "epoch": 1.1578947368421053,
      "grad_norm": 0.3816332519054413,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.553,
      "step": 165
    },
    {
      "epoch": 1.1649122807017545,
      "grad_norm": 0.3339821696281433,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.7165,
      "step": 166
    },
    {
      "epoch": 1.1719298245614036,
      "grad_norm": 0.4401732087135315,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.6222,
      "step": 167
    },
    {
      "epoch": 1.1789473684210527,
      "grad_norm": 0.34751060605049133,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.5905,
      "step": 168
    },
    {
      "epoch": 1.1859649122807017,
      "grad_norm": 0.32923898100852966,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.5593,
      "step": 169
    },
    {
      "epoch": 1.1929824561403508,
      "grad_norm": 0.36460188031196594,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.5299,
      "step": 170
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.3814084231853485,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.6562,
      "step": 171
    },
    {
      "epoch": 1.207017543859649,
      "grad_norm": 0.37295228242874146,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.67,
      "step": 172
    },
    {
      "epoch": 1.2140350877192982,
      "grad_norm": 0.4124593138694763,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.6914,
      "step": 173
    },
    {
      "epoch": 1.2210526315789474,
      "grad_norm": 0.4529687762260437,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.5548,
      "step": 174
    },
    {
      "epoch": 1.2280701754385965,
      "grad_norm": 0.45611923933029175,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.7082,
      "step": 175
    },
    {
      "epoch": 1.2350877192982457,
      "grad_norm": 0.4975593686103821,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.607,
      "step": 176
    },
    {
      "epoch": 1.2421052631578948,
      "grad_norm": 0.6174046993255615,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.6878,
      "step": 177
    },
    {
      "epoch": 1.2491228070175437,
      "grad_norm": 0.41679346561431885,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.5936,
      "step": 178
    },
    {
      "epoch": 1.256140350877193,
      "grad_norm": 0.3796026110649109,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.541,
      "step": 179
    },
    {
      "epoch": 1.263157894736842,
      "grad_norm": 0.31601041555404663,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.5436,
      "step": 180
    },
    {
      "epoch": 1.2701754385964912,
      "grad_norm": 0.3836706876754761,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.565,
      "step": 181
    },
    {
      "epoch": 1.2771929824561403,
      "grad_norm": 0.3156001567840576,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.5774,
      "step": 182
    },
    {
      "epoch": 1.2842105263157895,
      "grad_norm": 0.349017858505249,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.6324,
      "step": 183
    },
    {
      "epoch": 1.2912280701754386,
      "grad_norm": 0.3165048062801361,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.6538,
      "step": 184
    },
    {
      "epoch": 1.2982456140350878,
      "grad_norm": 0.37328585982322693,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.6926,
      "step": 185
    },
    {
      "epoch": 1.305263157894737,
      "grad_norm": 0.34521886706352234,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.6421,
      "step": 186
    },
    {
      "epoch": 1.312280701754386,
      "grad_norm": 0.33264097571372986,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.6639,
      "step": 187
    },
    {
      "epoch": 1.3192982456140352,
      "grad_norm": 0.36932653188705444,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.655,
      "step": 188
    },
    {
      "epoch": 1.3263157894736843,
      "grad_norm": 0.32533422112464905,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.7196,
      "step": 189
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 0.31627508997917175,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.5735,
      "step": 190
    },
    {
      "epoch": 1.3403508771929824,
      "grad_norm": 0.3094812333583832,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.5687,
      "step": 191
    },
    {
      "epoch": 1.3473684210526315,
      "grad_norm": 0.31595465540885925,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.7733,
      "step": 192
    },
    {
      "epoch": 1.3543859649122807,
      "grad_norm": 0.3201944828033447,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.7411,
      "step": 193
    },
    {
      "epoch": 1.3614035087719298,
      "grad_norm": 0.3263149857521057,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.6366,
      "step": 194
    },
    {
      "epoch": 1.368421052631579,
      "grad_norm": 0.3188920021057129,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.6258,
      "step": 195
    },
    {
      "epoch": 1.3754385964912281,
      "grad_norm": 0.30912092328071594,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.7647,
      "step": 196
    },
    {
      "epoch": 1.3824561403508773,
      "grad_norm": 0.3271566927433014,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.5736,
      "step": 197
    },
    {
      "epoch": 1.3894736842105262,
      "grad_norm": 0.32316580414772034,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.5551,
      "step": 198
    },
    {
      "epoch": 1.3964912280701753,
      "grad_norm": 0.3162274658679962,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.6287,
      "step": 199
    },
    {
      "epoch": 1.4035087719298245,
      "grad_norm": 0.32728272676467896,
      "learning_rate": 0.0,
      "loss": 1.688,
      "step": 200
    },
    {
      "epoch": 1.4035087719298245,
      "eval_loss": 1.6314605474472046,
      "eval_runtime": 3.297,
      "eval_samples_per_second": 72.794,
      "eval_steps_per_second": 18.198,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "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": 4944352090521600.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}