{
  "best_metric": 0.23894764482975006,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.7835455435847208,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0039177277179236044,
      "grad_norm": 35.1519660949707,
      "learning_rate": 1e-05,
      "loss": 3.3377,
      "step": 1
    },
    {
      "epoch": 0.0039177277179236044,
      "eval_loss": 1.376975655555725,
      "eval_runtime": 30.7914,
      "eval_samples_per_second": 13.965,
      "eval_steps_per_second": 3.507,
      "step": 1
    },
    {
      "epoch": 0.007835455435847209,
      "grad_norm": 37.591468811035156,
      "learning_rate": 2e-05,
      "loss": 3.5823,
      "step": 2
    },
    {
      "epoch": 0.011753183153770812,
      "grad_norm": 27.038301467895508,
      "learning_rate": 3e-05,
      "loss": 3.1019,
      "step": 3
    },
    {
      "epoch": 0.015670910871694418,
      "grad_norm": 17.632797241210938,
      "learning_rate": 4e-05,
      "loss": 2.424,
      "step": 4
    },
    {
      "epoch": 0.019588638589618023,
      "grad_norm": 13.900057792663574,
      "learning_rate": 5e-05,
      "loss": 1.7934,
      "step": 5
    },
    {
      "epoch": 0.023506366307541625,
      "grad_norm": 6.731367111206055,
      "learning_rate": 6e-05,
      "loss": 1.5624,
      "step": 6
    },
    {
      "epoch": 0.02742409402546523,
      "grad_norm": 4.34248161315918,
      "learning_rate": 7e-05,
      "loss": 1.2948,
      "step": 7
    },
    {
      "epoch": 0.031341821743388835,
      "grad_norm": 3.769124746322632,
      "learning_rate": 8e-05,
      "loss": 1.3975,
      "step": 8
    },
    {
      "epoch": 0.03525954946131244,
      "grad_norm": 3.391641855239868,
      "learning_rate": 9e-05,
      "loss": 1.2989,
      "step": 9
    },
    {
      "epoch": 0.039177277179236046,
      "grad_norm": 3.0701184272766113,
      "learning_rate": 0.0001,
      "loss": 1.0842,
      "step": 10
    },
    {
      "epoch": 0.043095004897159644,
      "grad_norm": 3.9934017658233643,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.0914,
      "step": 11
    },
    {
      "epoch": 0.04701273261508325,
      "grad_norm": 3.2494006156921387,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.1219,
      "step": 12
    },
    {
      "epoch": 0.050930460333006855,
      "grad_norm": 3.7557196617126465,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.1797,
      "step": 13
    },
    {
      "epoch": 0.05484818805093046,
      "grad_norm": 3.097635507583618,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.215,
      "step": 14
    },
    {
      "epoch": 0.058765915768854066,
      "grad_norm": 3.3909478187561035,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.0895,
      "step": 15
    },
    {
      "epoch": 0.06268364348677767,
      "grad_norm": 3.7466847896575928,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.3652,
      "step": 16
    },
    {
      "epoch": 0.06660137120470128,
      "grad_norm": 3.1385905742645264,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.1725,
      "step": 17
    },
    {
      "epoch": 0.07051909892262488,
      "grad_norm": 3.1092734336853027,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.3138,
      "step": 18
    },
    {
      "epoch": 0.07443682664054849,
      "grad_norm": 2.706139326095581,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.0156,
      "step": 19
    },
    {
      "epoch": 0.07835455435847209,
      "grad_norm": 3.3206448554992676,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3558,
      "step": 20
    },
    {
      "epoch": 0.08227228207639568,
      "grad_norm": 2.7940611839294434,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.2443,
      "step": 21
    },
    {
      "epoch": 0.08619000979431929,
      "grad_norm": 2.728489875793457,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.1225,
      "step": 22
    },
    {
      "epoch": 0.0901077375122429,
      "grad_norm": 2.7424354553222656,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.118,
      "step": 23
    },
    {
      "epoch": 0.0940254652301665,
      "grad_norm": 2.959437847137451,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.1265,
      "step": 24
    },
    {
      "epoch": 0.0979431929480901,
      "grad_norm": 3.068791151046753,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.111,
      "step": 25
    },
    {
      "epoch": 0.10186092066601371,
      "grad_norm": 3.436812400817871,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.2612,
      "step": 26
    },
    {
      "epoch": 0.10577864838393732,
      "grad_norm": 2.8119099140167236,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.9175,
      "step": 27
    },
    {
      "epoch": 0.10969637610186092,
      "grad_norm": 2.9515905380249023,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.1611,
      "step": 28
    },
    {
      "epoch": 0.11361410381978453,
      "grad_norm": 3.0629842281341553,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.0789,
      "step": 29
    },
    {
      "epoch": 0.11753183153770813,
      "grad_norm": 3.72255277633667,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.1929,
      "step": 30
    },
    {
      "epoch": 0.12144955925563174,
      "grad_norm": 2.680954694747925,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.9458,
      "step": 31
    },
    {
      "epoch": 0.12536728697355534,
      "grad_norm": 2.4375383853912354,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.949,
      "step": 32
    },
    {
      "epoch": 0.12928501469147893,
      "grad_norm": 2.8537662029266357,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.1638,
      "step": 33
    },
    {
      "epoch": 0.13320274240940255,
      "grad_norm": 2.751599073410034,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.1644,
      "step": 34
    },
    {
      "epoch": 0.13712047012732614,
      "grad_norm": 2.9511516094207764,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.052,
      "step": 35
    },
    {
      "epoch": 0.14103819784524976,
      "grad_norm": 2.903895616531372,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.0966,
      "step": 36
    },
    {
      "epoch": 0.14495592556317335,
      "grad_norm": 2.5975160598754883,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.033,
      "step": 37
    },
    {
      "epoch": 0.14887365328109697,
      "grad_norm": 2.664412021636963,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.1325,
      "step": 38
    },
    {
      "epoch": 0.15279138099902057,
      "grad_norm": 2.3869800567626953,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.8831,
      "step": 39
    },
    {
      "epoch": 0.15670910871694418,
      "grad_norm": 3.5770905017852783,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.0356,
      "step": 40
    },
    {
      "epoch": 0.16062683643486778,
      "grad_norm": 3.232245922088623,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.1462,
      "step": 41
    },
    {
      "epoch": 0.16454456415279137,
      "grad_norm": 2.72615647315979,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.0273,
      "step": 42
    },
    {
      "epoch": 0.168462291870715,
      "grad_norm": 2.8978590965270996,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.021,
      "step": 43
    },
    {
      "epoch": 0.17238001958863858,
      "grad_norm": 3.248000144958496,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.093,
      "step": 44
    },
    {
      "epoch": 0.1762977473065622,
      "grad_norm": 2.8521573543548584,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.0163,
      "step": 45
    },
    {
      "epoch": 0.1802154750244858,
      "grad_norm": 3.390756845474243,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.8542,
      "step": 46
    },
    {
      "epoch": 0.1841332027424094,
      "grad_norm": 2.9699442386627197,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.9035,
      "step": 47
    },
    {
      "epoch": 0.188050930460333,
      "grad_norm": 2.9843027591705322,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.0058,
      "step": 48
    },
    {
      "epoch": 0.19196865817825662,
      "grad_norm": 4.668224334716797,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.7376,
      "step": 49
    },
    {
      "epoch": 0.1958863858961802,
      "grad_norm": 2.9681031703948975,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.7555,
      "step": 50
    },
    {
      "epoch": 0.1958863858961802,
      "eval_loss": 0.3203987777233124,
      "eval_runtime": 31.553,
      "eval_samples_per_second": 13.628,
      "eval_steps_per_second": 3.423,
      "step": 50
    },
    {
      "epoch": 0.19980411361410383,
      "grad_norm": 7.285029888153076,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.5949,
      "step": 51
    },
    {
      "epoch": 0.20372184133202742,
      "grad_norm": 4.377089500427246,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.3685,
      "step": 52
    },
    {
      "epoch": 0.20763956904995104,
      "grad_norm": 2.7384912967681885,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.329,
      "step": 53
    },
    {
      "epoch": 0.21155729676787463,
      "grad_norm": 2.206838846206665,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.1793,
      "step": 54
    },
    {
      "epoch": 0.21547502448579825,
      "grad_norm": 2.329073905944824,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.0442,
      "step": 55
    },
    {
      "epoch": 0.21939275220372184,
      "grad_norm": 2.0647895336151123,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.0533,
      "step": 56
    },
    {
      "epoch": 0.22331047992164543,
      "grad_norm": 2.3976070880889893,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.1446,
      "step": 57
    },
    {
      "epoch": 0.22722820763956905,
      "grad_norm": 2.5454862117767334,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.2031,
      "step": 58
    },
    {
      "epoch": 0.23114593535749264,
      "grad_norm": 2.18048357963562,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.2731,
      "step": 59
    },
    {
      "epoch": 0.23506366307541626,
      "grad_norm": 2.0146234035491943,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.1083,
      "step": 60
    },
    {
      "epoch": 0.23898139079333985,
      "grad_norm": 2.171734094619751,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.9576,
      "step": 61
    },
    {
      "epoch": 0.24289911851126347,
      "grad_norm": 2.0504679679870605,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.1194,
      "step": 62
    },
    {
      "epoch": 0.24681684622918706,
      "grad_norm": 2.211573362350464,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.0986,
      "step": 63
    },
    {
      "epoch": 0.2507345739471107,
      "grad_norm": 2.2392194271087646,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.1292,
      "step": 64
    },
    {
      "epoch": 0.2546523016650343,
      "grad_norm": 2.342341661453247,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.1437,
      "step": 65
    },
    {
      "epoch": 0.25857002938295787,
      "grad_norm": 2.102189064025879,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.0247,
      "step": 66
    },
    {
      "epoch": 0.2624877571008815,
      "grad_norm": 2.1631100177764893,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.0929,
      "step": 67
    },
    {
      "epoch": 0.2664054848188051,
      "grad_norm": 2.0671801567077637,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.1522,
      "step": 68
    },
    {
      "epoch": 0.2703232125367287,
      "grad_norm": 2.249319314956665,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.0299,
      "step": 69
    },
    {
      "epoch": 0.2742409402546523,
      "grad_norm": 2.0861968994140625,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.9798,
      "step": 70
    },
    {
      "epoch": 0.2781586679725759,
      "grad_norm": 2.284118890762329,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.9014,
      "step": 71
    },
    {
      "epoch": 0.2820763956904995,
      "grad_norm": 2.335890054702759,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.9852,
      "step": 72
    },
    {
      "epoch": 0.2859941234084231,
      "grad_norm": 2.15493106842041,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.024,
      "step": 73
    },
    {
      "epoch": 0.2899118511263467,
      "grad_norm": 2.6023154258728027,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.134,
      "step": 74
    },
    {
      "epoch": 0.2938295788442703,
      "grad_norm": 2.412675142288208,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.1444,
      "step": 75
    },
    {
      "epoch": 0.29774730656219395,
      "grad_norm": 2.2884745597839355,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.0542,
      "step": 76
    },
    {
      "epoch": 0.30166503428011754,
      "grad_norm": 2.557800531387329,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.3114,
      "step": 77
    },
    {
      "epoch": 0.30558276199804113,
      "grad_norm": 2.212139129638672,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.0496,
      "step": 78
    },
    {
      "epoch": 0.3095004897159647,
      "grad_norm": 2.3875274658203125,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.1647,
      "step": 79
    },
    {
      "epoch": 0.31341821743388837,
      "grad_norm": 2.358973979949951,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.9839,
      "step": 80
    },
    {
      "epoch": 0.31733594515181196,
      "grad_norm": 2.8181726932525635,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.17,
      "step": 81
    },
    {
      "epoch": 0.32125367286973555,
      "grad_norm": 2.4572412967681885,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.9679,
      "step": 82
    },
    {
      "epoch": 0.32517140058765914,
      "grad_norm": 2.355374574661255,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.9826,
      "step": 83
    },
    {
      "epoch": 0.32908912830558273,
      "grad_norm": 2.5143699645996094,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.1583,
      "step": 84
    },
    {
      "epoch": 0.3330068560235064,
      "grad_norm": 2.4585354328155518,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.9229,
      "step": 85
    },
    {
      "epoch": 0.33692458374143,
      "grad_norm": 3.429137706756592,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.8921,
      "step": 86
    },
    {
      "epoch": 0.34084231145935356,
      "grad_norm": 2.3490748405456543,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.9492,
      "step": 87
    },
    {
      "epoch": 0.34476003917727716,
      "grad_norm": 2.655750036239624,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.9217,
      "step": 88
    },
    {
      "epoch": 0.3486777668952008,
      "grad_norm": 2.411027193069458,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.8969,
      "step": 89
    },
    {
      "epoch": 0.3525954946131244,
      "grad_norm": 2.8227200508117676,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.2383,
      "step": 90
    },
    {
      "epoch": 0.356513222331048,
      "grad_norm": 2.5691335201263428,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0072,
      "step": 91
    },
    {
      "epoch": 0.3604309500489716,
      "grad_norm": 2.5621490478515625,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.0163,
      "step": 92
    },
    {
      "epoch": 0.3643486777668952,
      "grad_norm": 2.4902713298797607,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.9825,
      "step": 93
    },
    {
      "epoch": 0.3682664054848188,
      "grad_norm": 2.6015195846557617,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.0142,
      "step": 94
    },
    {
      "epoch": 0.3721841332027424,
      "grad_norm": 2.901930093765259,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.0073,
      "step": 95
    },
    {
      "epoch": 0.376101860920666,
      "grad_norm": 2.4084105491638184,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.87,
      "step": 96
    },
    {
      "epoch": 0.38001958863858964,
      "grad_norm": 2.4517323970794678,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.8646,
      "step": 97
    },
    {
      "epoch": 0.38393731635651324,
      "grad_norm": 2.4089605808258057,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.7959,
      "step": 98
    },
    {
      "epoch": 0.3878550440744368,
      "grad_norm": 2.4464828968048096,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.6847,
      "step": 99
    },
    {
      "epoch": 0.3917727717923604,
      "grad_norm": 3.361382484436035,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.8682,
      "step": 100
    },
    {
      "epoch": 0.3917727717923604,
      "eval_loss": 0.2728331387042999,
      "eval_runtime": 31.565,
      "eval_samples_per_second": 13.623,
      "eval_steps_per_second": 3.422,
      "step": 100
    },
    {
      "epoch": 0.395690499510284,
      "grad_norm": 3.1312220096588135,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4848,
      "step": 101
    },
    {
      "epoch": 0.39960822722820766,
      "grad_norm": 2.4794907569885254,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.1429,
      "step": 102
    },
    {
      "epoch": 0.40352595494613125,
      "grad_norm": 2.022571086883545,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.1495,
      "step": 103
    },
    {
      "epoch": 0.40744368266405484,
      "grad_norm": 1.588888168334961,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.9335,
      "step": 104
    },
    {
      "epoch": 0.41136141038197843,
      "grad_norm": 1.7831252813339233,
      "learning_rate": 5e-05,
      "loss": 1.1477,
      "step": 105
    },
    {
      "epoch": 0.4152791380999021,
      "grad_norm": 1.7243947982788086,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.1025,
      "step": 106
    },
    {
      "epoch": 0.41919686581782567,
      "grad_norm": 1.7146291732788086,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.0639,
      "step": 107
    },
    {
      "epoch": 0.42311459353574926,
      "grad_norm": 2.0036494731903076,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.0837,
      "step": 108
    },
    {
      "epoch": 0.42703232125367285,
      "grad_norm": 2.0686757564544678,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2024,
      "step": 109
    },
    {
      "epoch": 0.4309500489715965,
      "grad_norm": 1.8652163743972778,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1025,
      "step": 110
    },
    {
      "epoch": 0.4348677766895201,
      "grad_norm": 1.8024530410766602,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.9686,
      "step": 111
    },
    {
      "epoch": 0.4387855044074437,
      "grad_norm": 2.0305209159851074,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.0767,
      "step": 112
    },
    {
      "epoch": 0.4427032321253673,
      "grad_norm": 1.802006483078003,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.9561,
      "step": 113
    },
    {
      "epoch": 0.44662095984329087,
      "grad_norm": 2.171335220336914,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.2262,
      "step": 114
    },
    {
      "epoch": 0.4505386875612145,
      "grad_norm": 2.009345769882202,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.9849,
      "step": 115
    },
    {
      "epoch": 0.4544564152791381,
      "grad_norm": 1.8234909772872925,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.963,
      "step": 116
    },
    {
      "epoch": 0.4583741429970617,
      "grad_norm": 2.2594306468963623,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.0953,
      "step": 117
    },
    {
      "epoch": 0.4622918707149853,
      "grad_norm": 1.9550275802612305,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.9998,
      "step": 118
    },
    {
      "epoch": 0.46620959843290893,
      "grad_norm": 2.1157915592193604,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.0168,
      "step": 119
    },
    {
      "epoch": 0.4701273261508325,
      "grad_norm": 1.942893624305725,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.8811,
      "step": 120
    },
    {
      "epoch": 0.4740450538687561,
      "grad_norm": 2.085742235183716,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.9816,
      "step": 121
    },
    {
      "epoch": 0.4779627815866797,
      "grad_norm": 2.3129210472106934,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2215,
      "step": 122
    },
    {
      "epoch": 0.48188050930460335,
      "grad_norm": 1.9363096952438354,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.8749,
      "step": 123
    },
    {
      "epoch": 0.48579823702252695,
      "grad_norm": 2.096863031387329,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.9519,
      "step": 124
    },
    {
      "epoch": 0.48971596474045054,
      "grad_norm": 1.9756804704666138,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.9019,
      "step": 125
    },
    {
      "epoch": 0.49363369245837413,
      "grad_norm": 2.351135730743408,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.2106,
      "step": 126
    },
    {
      "epoch": 0.4975514201762977,
      "grad_norm": 2.1129064559936523,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0321,
      "step": 127
    },
    {
      "epoch": 0.5014691478942214,
      "grad_norm": 2.019407272338867,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.01,
      "step": 128
    },
    {
      "epoch": 0.5053868756121449,
      "grad_norm": 2.3377411365509033,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.1417,
      "step": 129
    },
    {
      "epoch": 0.5093046033300686,
      "grad_norm": 1.9798047542572021,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.8732,
      "step": 130
    },
    {
      "epoch": 0.5132223310479922,
      "grad_norm": 2.208526372909546,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.9831,
      "step": 131
    },
    {
      "epoch": 0.5171400587659157,
      "grad_norm": 2.229703426361084,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.809,
      "step": 132
    },
    {
      "epoch": 0.5210577864838394,
      "grad_norm": 2.2709872722625732,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.8777,
      "step": 133
    },
    {
      "epoch": 0.524975514201763,
      "grad_norm": 2.0026984214782715,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.959,
      "step": 134
    },
    {
      "epoch": 0.5288932419196866,
      "grad_norm": 2.0388407707214355,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.9683,
      "step": 135
    },
    {
      "epoch": 0.5328109696376102,
      "grad_norm": 2.274142026901245,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.9447,
      "step": 136
    },
    {
      "epoch": 0.5367286973555337,
      "grad_norm": 2.741628885269165,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.9562,
      "step": 137
    },
    {
      "epoch": 0.5406464250734574,
      "grad_norm": 2.4420840740203857,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.9921,
      "step": 138
    },
    {
      "epoch": 0.544564152791381,
      "grad_norm": 2.174839735031128,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.9309,
      "step": 139
    },
    {
      "epoch": 0.5484818805093046,
      "grad_norm": 2.241736650466919,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.0526,
      "step": 140
    },
    {
      "epoch": 0.5523996082272282,
      "grad_norm": 2.2623815536499023,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.8288,
      "step": 141
    },
    {
      "epoch": 0.5563173359451518,
      "grad_norm": 2.475973129272461,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.9575,
      "step": 142
    },
    {
      "epoch": 0.5602350636630754,
      "grad_norm": 2.2730438709259033,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.9225,
      "step": 143
    },
    {
      "epoch": 0.564152791380999,
      "grad_norm": 2.3677737712860107,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.8787,
      "step": 144
    },
    {
      "epoch": 0.5680705190989226,
      "grad_norm": 2.1602799892425537,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.8815,
      "step": 145
    },
    {
      "epoch": 0.5719882468168462,
      "grad_norm": 2.3288028240203857,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.812,
      "step": 146
    },
    {
      "epoch": 0.5759059745347699,
      "grad_norm": 2.4011189937591553,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.846,
      "step": 147
    },
    {
      "epoch": 0.5798237022526934,
      "grad_norm": 2.1641552448272705,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.6728,
      "step": 148
    },
    {
      "epoch": 0.5837414299706171,
      "grad_norm": 2.5565390586853027,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.7585,
      "step": 149
    },
    {
      "epoch": 0.5876591576885406,
      "grad_norm": 2.993711233139038,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.751,
      "step": 150
    },
    {
      "epoch": 0.5876591576885406,
      "eval_loss": 0.24573403596878052,
      "eval_runtime": 31.5546,
      "eval_samples_per_second": 13.627,
      "eval_steps_per_second": 3.423,
      "step": 150
    },
    {
      "epoch": 0.5915768854064642,
      "grad_norm": 2.052823305130005,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.2849,
      "step": 151
    },
    {
      "epoch": 0.5954946131243879,
      "grad_norm": 1.8187516927719116,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.1651,
      "step": 152
    },
    {
      "epoch": 0.5994123408423114,
      "grad_norm": 1.9232217073440552,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.158,
      "step": 153
    },
    {
      "epoch": 0.6033300685602351,
      "grad_norm": 1.8911371231079102,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.011,
      "step": 154
    },
    {
      "epoch": 0.6072477962781586,
      "grad_norm": 1.883107304573059,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.0794,
      "step": 155
    },
    {
      "epoch": 0.6111655239960823,
      "grad_norm": 1.7157275676727295,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.9902,
      "step": 156
    },
    {
      "epoch": 0.6150832517140059,
      "grad_norm": 11.430994987487793,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.095,
      "step": 157
    },
    {
      "epoch": 0.6190009794319294,
      "grad_norm": 1.923728346824646,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.0466,
      "step": 158
    },
    {
      "epoch": 0.6229187071498531,
      "grad_norm": 1.690798044204712,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.0216,
      "step": 159
    },
    {
      "epoch": 0.6268364348677767,
      "grad_norm": 1.8792833089828491,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.9898,
      "step": 160
    },
    {
      "epoch": 0.6307541625857003,
      "grad_norm": 2.0197887420654297,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.1489,
      "step": 161
    },
    {
      "epoch": 0.6346718903036239,
      "grad_norm": 1.739636778831482,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.9464,
      "step": 162
    },
    {
      "epoch": 0.6385896180215475,
      "grad_norm": 1.642520546913147,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.8664,
      "step": 163
    },
    {
      "epoch": 0.6425073457394711,
      "grad_norm": 1.674628734588623,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.8505,
      "step": 164
    },
    {
      "epoch": 0.6464250734573947,
      "grad_norm": 1.931117057800293,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.1382,
      "step": 165
    },
    {
      "epoch": 0.6503428011753183,
      "grad_norm": 1.7734427452087402,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0118,
      "step": 166
    },
    {
      "epoch": 0.6542605288932419,
      "grad_norm": 2.02400803565979,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1354,
      "step": 167
    },
    {
      "epoch": 0.6581782566111655,
      "grad_norm": 1.73727285861969,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.9586,
      "step": 168
    },
    {
      "epoch": 0.6620959843290891,
      "grad_norm": 1.7932664155960083,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.9289,
      "step": 169
    },
    {
      "epoch": 0.6660137120470128,
      "grad_norm": 1.734366774559021,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9243,
      "step": 170
    },
    {
      "epoch": 0.6699314397649363,
      "grad_norm": 1.8200660943984985,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.9855,
      "step": 171
    },
    {
      "epoch": 0.67384916748286,
      "grad_norm": 1.886230707168579,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.1023,
      "step": 172
    },
    {
      "epoch": 0.6777668952007836,
      "grad_norm": 1.8692189455032349,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.9193,
      "step": 173
    },
    {
      "epoch": 0.6816846229187071,
      "grad_norm": 2.0463931560516357,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.8724,
      "step": 174
    },
    {
      "epoch": 0.6856023506366308,
      "grad_norm": 2.1118369102478027,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.1919,
      "step": 175
    },
    {
      "epoch": 0.6895200783545543,
      "grad_norm": 1.8366163969039917,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9332,
      "step": 176
    },
    {
      "epoch": 0.693437806072478,
      "grad_norm": 1.9711445569992065,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.8643,
      "step": 177
    },
    {
      "epoch": 0.6973555337904016,
      "grad_norm": 1.955785870552063,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.963,
      "step": 178
    },
    {
      "epoch": 0.7012732615083251,
      "grad_norm": 3.3038651943206787,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.9136,
      "step": 179
    },
    {
      "epoch": 0.7051909892262488,
      "grad_norm": 2.139598846435547,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.9532,
      "step": 180
    },
    {
      "epoch": 0.7091087169441724,
      "grad_norm": 1.8191031217575073,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.795,
      "step": 181
    },
    {
      "epoch": 0.713026444662096,
      "grad_norm": 2.069441318511963,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.9831,
      "step": 182
    },
    {
      "epoch": 0.7169441723800196,
      "grad_norm": 2.1862785816192627,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.996,
      "step": 183
    },
    {
      "epoch": 0.7208619000979432,
      "grad_norm": 2.2842917442321777,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.097,
      "step": 184
    },
    {
      "epoch": 0.7247796278158668,
      "grad_norm": 1.955565333366394,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.785,
      "step": 185
    },
    {
      "epoch": 0.7286973555337904,
      "grad_norm": 1.749302864074707,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.7255,
      "step": 186
    },
    {
      "epoch": 0.732615083251714,
      "grad_norm": 2.4262523651123047,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.8931,
      "step": 187
    },
    {
      "epoch": 0.7365328109696376,
      "grad_norm": 2.24680495262146,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.8067,
      "step": 188
    },
    {
      "epoch": 0.7404505386875612,
      "grad_norm": 1.8985466957092285,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8182,
      "step": 189
    },
    {
      "epoch": 0.7443682664054848,
      "grad_norm": 2.6636481285095215,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.0733,
      "step": 190
    },
    {
      "epoch": 0.7482859941234085,
      "grad_norm": 2.3171799182891846,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.9825,
      "step": 191
    },
    {
      "epoch": 0.752203721841332,
      "grad_norm": 2.5546414852142334,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.0336,
      "step": 192
    },
    {
      "epoch": 0.7561214495592556,
      "grad_norm": 2.307732105255127,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.8529,
      "step": 193
    },
    {
      "epoch": 0.7600391772771793,
      "grad_norm": 2.2367029190063477,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.7411,
      "step": 194
    },
    {
      "epoch": 0.7639569049951028,
      "grad_norm": 2.188103437423706,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.7011,
      "step": 195
    },
    {
      "epoch": 0.7678746327130265,
      "grad_norm": 2.367469310760498,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.8039,
      "step": 196
    },
    {
      "epoch": 0.77179236043095,
      "grad_norm": 2.4637210369110107,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.8507,
      "step": 197
    },
    {
      "epoch": 0.7757100881488737,
      "grad_norm": 2.821525812149048,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.856,
      "step": 198
    },
    {
      "epoch": 0.7796278158667973,
      "grad_norm": 2.412360191345215,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.5984,
      "step": 199
    },
    {
      "epoch": 0.7835455435847208,
      "grad_norm": 3.7647109031677246,
      "learning_rate": 0.0,
      "loss": 0.9927,
      "step": 200
    },
    {
      "epoch": 0.7835455435847208,
      "eval_loss": 0.23894764482975006,
      "eval_runtime": 31.5711,
      "eval_samples_per_second": 13.62,
      "eval_steps_per_second": 3.421,
      "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": 2.863232968556544e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}