{
  "best_metric": 2.5549092292785645,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.9791921664626683,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004895960832313341,
      "grad_norm": 0.822125256061554,
      "learning_rate": 1e-05,
      "loss": 11.7733,
      "step": 1
    },
    {
      "epoch": 0.004895960832313341,
      "eval_loss": 3.217442750930786,
      "eval_runtime": 10.1953,
      "eval_samples_per_second": 33.741,
      "eval_steps_per_second": 8.435,
      "step": 1
    },
    {
      "epoch": 0.009791921664626682,
      "grad_norm": 0.8506293892860413,
      "learning_rate": 2e-05,
      "loss": 11.076,
      "step": 2
    },
    {
      "epoch": 0.014687882496940025,
      "grad_norm": 0.9396572113037109,
      "learning_rate": 3e-05,
      "loss": 12.0851,
      "step": 3
    },
    {
      "epoch": 0.019583843329253364,
      "grad_norm": 0.9412523508071899,
      "learning_rate": 4e-05,
      "loss": 11.8478,
      "step": 4
    },
    {
      "epoch": 0.02447980416156671,
      "grad_norm": 0.9336570501327515,
      "learning_rate": 5e-05,
      "loss": 11.2823,
      "step": 5
    },
    {
      "epoch": 0.02937576499388005,
      "grad_norm": 1.11503267288208,
      "learning_rate": 6e-05,
      "loss": 12.6901,
      "step": 6
    },
    {
      "epoch": 0.03427172582619339,
      "grad_norm": 1.039236307144165,
      "learning_rate": 7e-05,
      "loss": 12.3259,
      "step": 7
    },
    {
      "epoch": 0.03916768665850673,
      "grad_norm": 1.010493278503418,
      "learning_rate": 8e-05,
      "loss": 11.9951,
      "step": 8
    },
    {
      "epoch": 0.044063647490820076,
      "grad_norm": 1.3338216543197632,
      "learning_rate": 9e-05,
      "loss": 12.3659,
      "step": 9
    },
    {
      "epoch": 0.04895960832313342,
      "grad_norm": 1.0697107315063477,
      "learning_rate": 0.0001,
      "loss": 12.3345,
      "step": 10
    },
    {
      "epoch": 0.05385556915544676,
      "grad_norm": 1.1506353616714478,
      "learning_rate": 9.999316524962345e-05,
      "loss": 12.1538,
      "step": 11
    },
    {
      "epoch": 0.0587515299877601,
      "grad_norm": 1.155070185661316,
      "learning_rate": 9.997266286704631e-05,
      "loss": 12.3107,
      "step": 12
    },
    {
      "epoch": 0.06364749082007344,
      "grad_norm": 1.1654995679855347,
      "learning_rate": 9.993849845741524e-05,
      "loss": 11.8527,
      "step": 13
    },
    {
      "epoch": 0.06854345165238677,
      "grad_norm": 1.1552754640579224,
      "learning_rate": 9.989068136093873e-05,
      "loss": 11.9672,
      "step": 14
    },
    {
      "epoch": 0.07343941248470012,
      "grad_norm": 1.4518193006515503,
      "learning_rate": 9.98292246503335e-05,
      "loss": 11.8369,
      "step": 15
    },
    {
      "epoch": 0.07833537331701346,
      "grad_norm": 1.4227337837219238,
      "learning_rate": 9.975414512725057e-05,
      "loss": 12.5013,
      "step": 16
    },
    {
      "epoch": 0.0832313341493268,
      "grad_norm": 1.4715157747268677,
      "learning_rate": 9.966546331768191e-05,
      "loss": 12.456,
      "step": 17
    },
    {
      "epoch": 0.08812729498164015,
      "grad_norm": 1.2398803234100342,
      "learning_rate": 9.956320346634876e-05,
      "loss": 11.8697,
      "step": 18
    },
    {
      "epoch": 0.09302325581395349,
      "grad_norm": 1.473848581314087,
      "learning_rate": 9.944739353007344e-05,
      "loss": 12.4153,
      "step": 19
    },
    {
      "epoch": 0.09791921664626684,
      "grad_norm": 1.30556321144104,
      "learning_rate": 9.931806517013612e-05,
      "loss": 11.9211,
      "step": 20
    },
    {
      "epoch": 0.10281517747858017,
      "grad_norm": 1.1892075538635254,
      "learning_rate": 9.917525374361912e-05,
      "loss": 11.9629,
      "step": 21
    },
    {
      "epoch": 0.10771113831089352,
      "grad_norm": 1.2155661582946777,
      "learning_rate": 9.901899829374047e-05,
      "loss": 12.1839,
      "step": 22
    },
    {
      "epoch": 0.11260709914320685,
      "grad_norm": 1.2450673580169678,
      "learning_rate": 9.884934153917997e-05,
      "loss": 11.6361,
      "step": 23
    },
    {
      "epoch": 0.1175030599755202,
      "grad_norm": 1.290870189666748,
      "learning_rate": 9.86663298624003e-05,
      "loss": 11.6479,
      "step": 24
    },
    {
      "epoch": 0.12239902080783353,
      "grad_norm": 1.3084297180175781,
      "learning_rate": 9.847001329696653e-05,
      "loss": 11.8813,
      "step": 25
    },
    {
      "epoch": 0.12729498164014688,
      "grad_norm": 1.461325764656067,
      "learning_rate": 9.826044551386744e-05,
      "loss": 11.9869,
      "step": 26
    },
    {
      "epoch": 0.13219094247246022,
      "grad_norm": 1.3953827619552612,
      "learning_rate": 9.803768380684242e-05,
      "loss": 11.6794,
      "step": 27
    },
    {
      "epoch": 0.13708690330477355,
      "grad_norm": 1.3279074430465698,
      "learning_rate": 9.780178907671789e-05,
      "loss": 11.3979,
      "step": 28
    },
    {
      "epoch": 0.1419828641370869,
      "grad_norm": 1.4564024209976196,
      "learning_rate": 9.755282581475769e-05,
      "loss": 11.7912,
      "step": 29
    },
    {
      "epoch": 0.14687882496940025,
      "grad_norm": 1.7769300937652588,
      "learning_rate": 9.729086208503174e-05,
      "loss": 11.7995,
      "step": 30
    },
    {
      "epoch": 0.15177478580171358,
      "grad_norm": 1.665911316871643,
      "learning_rate": 9.701596950580806e-05,
      "loss": 11.5609,
      "step": 31
    },
    {
      "epoch": 0.15667074663402691,
      "grad_norm": 1.4361752271652222,
      "learning_rate": 9.672822322997305e-05,
      "loss": 11.35,
      "step": 32
    },
    {
      "epoch": 0.16156670746634028,
      "grad_norm": 1.4808471202850342,
      "learning_rate": 9.642770192448536e-05,
      "loss": 11.3955,
      "step": 33
    },
    {
      "epoch": 0.1664626682986536,
      "grad_norm": 1.3898811340332031,
      "learning_rate": 9.611448774886924e-05,
      "loss": 10.9264,
      "step": 34
    },
    {
      "epoch": 0.17135862913096694,
      "grad_norm": 1.3348588943481445,
      "learning_rate": 9.578866633275288e-05,
      "loss": 11.2693,
      "step": 35
    },
    {
      "epoch": 0.1762545899632803,
      "grad_norm": 1.4922901391983032,
      "learning_rate": 9.545032675245813e-05,
      "loss": 10.7659,
      "step": 36
    },
    {
      "epoch": 0.18115055079559364,
      "grad_norm": 1.4295297861099243,
      "learning_rate": 9.509956150664796e-05,
      "loss": 11.4559,
      "step": 37
    },
    {
      "epoch": 0.18604651162790697,
      "grad_norm": 1.3803876638412476,
      "learning_rate": 9.473646649103818e-05,
      "loss": 10.6489,
      "step": 38
    },
    {
      "epoch": 0.1909424724602203,
      "grad_norm": 1.4224756956100464,
      "learning_rate": 9.43611409721806e-05,
      "loss": 10.7751,
      "step": 39
    },
    {
      "epoch": 0.19583843329253367,
      "grad_norm": 1.5139490365982056,
      "learning_rate": 9.397368756032445e-05,
      "loss": 10.9552,
      "step": 40
    },
    {
      "epoch": 0.200734394124847,
      "grad_norm": 1.5269768238067627,
      "learning_rate": 9.357421218136386e-05,
      "loss": 10.9675,
      "step": 41
    },
    {
      "epoch": 0.20563035495716034,
      "grad_norm": 1.6002178192138672,
      "learning_rate": 9.316282404787871e-05,
      "loss": 11.0692,
      "step": 42
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.6640499830245972,
      "learning_rate": 9.273963562927695e-05,
      "loss": 11.0153,
      "step": 43
    },
    {
      "epoch": 0.21542227662178703,
      "grad_norm": 1.6430227756500244,
      "learning_rate": 9.230476262104677e-05,
      "loss": 10.5567,
      "step": 44
    },
    {
      "epoch": 0.22031823745410037,
      "grad_norm": 1.8492897748947144,
      "learning_rate": 9.185832391312644e-05,
      "loss": 10.5902,
      "step": 45
    },
    {
      "epoch": 0.2252141982864137,
      "grad_norm": 1.677350401878357,
      "learning_rate": 9.140044155740101e-05,
      "loss": 10.7521,
      "step": 46
    },
    {
      "epoch": 0.23011015911872704,
      "grad_norm": 1.7459001541137695,
      "learning_rate": 9.093124073433463e-05,
      "loss": 10.4302,
      "step": 47
    },
    {
      "epoch": 0.2350061199510404,
      "grad_norm": 1.9310698509216309,
      "learning_rate": 9.045084971874738e-05,
      "loss": 10.4866,
      "step": 48
    },
    {
      "epoch": 0.23990208078335373,
      "grad_norm": 1.9746109247207642,
      "learning_rate": 8.995939984474624e-05,
      "loss": 10.7406,
      "step": 49
    },
    {
      "epoch": 0.24479804161566707,
      "grad_norm": 2.203674077987671,
      "learning_rate": 8.945702546981969e-05,
      "loss": 10.1613,
      "step": 50
    },
    {
      "epoch": 0.24479804161566707,
      "eval_loss": 2.732313632965088,
      "eval_runtime": 10.1935,
      "eval_samples_per_second": 33.747,
      "eval_steps_per_second": 8.437,
      "step": 50
    },
    {
      "epoch": 0.24969400244798043,
      "grad_norm": 2.218334913253784,
      "learning_rate": 8.894386393810563e-05,
      "loss": 11.2629,
      "step": 51
    },
    {
      "epoch": 0.25458996328029376,
      "grad_norm": 2.2006967067718506,
      "learning_rate": 8.842005554284296e-05,
      "loss": 10.5661,
      "step": 52
    },
    {
      "epoch": 0.2594859241126071,
      "grad_norm": 2.155848264694214,
      "learning_rate": 8.788574348801675e-05,
      "loss": 10.8884,
      "step": 53
    },
    {
      "epoch": 0.26438188494492043,
      "grad_norm": 2.002484083175659,
      "learning_rate": 8.73410738492077e-05,
      "loss": 11.4403,
      "step": 54
    },
    {
      "epoch": 0.2692778457772338,
      "grad_norm": 1.7315287590026855,
      "learning_rate": 8.678619553365659e-05,
      "loss": 10.757,
      "step": 55
    },
    {
      "epoch": 0.2741738066095471,
      "grad_norm": 1.335595726966858,
      "learning_rate": 8.622126023955446e-05,
      "loss": 10.77,
      "step": 56
    },
    {
      "epoch": 0.27906976744186046,
      "grad_norm": 1.3567054271697998,
      "learning_rate": 8.564642241456986e-05,
      "loss": 10.5966,
      "step": 57
    },
    {
      "epoch": 0.2839657282741738,
      "grad_norm": 1.3059027194976807,
      "learning_rate": 8.506183921362443e-05,
      "loss": 11.3295,
      "step": 58
    },
    {
      "epoch": 0.28886168910648713,
      "grad_norm": 1.1764811277389526,
      "learning_rate": 8.44676704559283e-05,
      "loss": 10.9697,
      "step": 59
    },
    {
      "epoch": 0.2937576499388005,
      "grad_norm": 1.3196907043457031,
      "learning_rate": 8.386407858128706e-05,
      "loss": 10.8152,
      "step": 60
    },
    {
      "epoch": 0.29865361077111385,
      "grad_norm": 1.324517011642456,
      "learning_rate": 8.32512286056924e-05,
      "loss": 10.5112,
      "step": 61
    },
    {
      "epoch": 0.30354957160342716,
      "grad_norm": 1.2578760385513306,
      "learning_rate": 8.262928807620843e-05,
      "loss": 10.6921,
      "step": 62
    },
    {
      "epoch": 0.3084455324357405,
      "grad_norm": 1.2686268091201782,
      "learning_rate": 8.199842702516583e-05,
      "loss": 11.0032,
      "step": 63
    },
    {
      "epoch": 0.31334149326805383,
      "grad_norm": 1.1417449712753296,
      "learning_rate": 8.135881792367686e-05,
      "loss": 10.3947,
      "step": 64
    },
    {
      "epoch": 0.3182374541003672,
      "grad_norm": 1.1969670057296753,
      "learning_rate": 8.07106356344834e-05,
      "loss": 10.7076,
      "step": 65
    },
    {
      "epoch": 0.32313341493268055,
      "grad_norm": 1.2692161798477173,
      "learning_rate": 8.005405736415126e-05,
      "loss": 10.8158,
      "step": 66
    },
    {
      "epoch": 0.32802937576499386,
      "grad_norm": 1.202545166015625,
      "learning_rate": 7.938926261462366e-05,
      "loss": 10.7931,
      "step": 67
    },
    {
      "epoch": 0.3329253365973072,
      "grad_norm": 1.2362068891525269,
      "learning_rate": 7.871643313414718e-05,
      "loss": 11.5311,
      "step": 68
    },
    {
      "epoch": 0.3378212974296206,
      "grad_norm": 1.2878124713897705,
      "learning_rate": 7.803575286758364e-05,
      "loss": 10.5301,
      "step": 69
    },
    {
      "epoch": 0.3427172582619339,
      "grad_norm": 1.289162278175354,
      "learning_rate": 7.734740790612136e-05,
      "loss": 10.3821,
      "step": 70
    },
    {
      "epoch": 0.34761321909424725,
      "grad_norm": 1.2230243682861328,
      "learning_rate": 7.66515864363997e-05,
      "loss": 10.3741,
      "step": 71
    },
    {
      "epoch": 0.3525091799265606,
      "grad_norm": 1.18302583694458,
      "learning_rate": 7.594847868906076e-05,
      "loss": 10.9881,
      "step": 72
    },
    {
      "epoch": 0.3574051407588739,
      "grad_norm": 1.636067509651184,
      "learning_rate": 7.52382768867422e-05,
      "loss": 10.6764,
      "step": 73
    },
    {
      "epoch": 0.3623011015911873,
      "grad_norm": 1.4973101615905762,
      "learning_rate": 7.452117519152542e-05,
      "loss": 11.0905,
      "step": 74
    },
    {
      "epoch": 0.3671970624235006,
      "grad_norm": 1.2543340921401978,
      "learning_rate": 7.379736965185368e-05,
      "loss": 10.8535,
      "step": 75
    },
    {
      "epoch": 0.37209302325581395,
      "grad_norm": 1.2691313028335571,
      "learning_rate": 7.30670581489344e-05,
      "loss": 10.8121,
      "step": 76
    },
    {
      "epoch": 0.3769889840881273,
      "grad_norm": 1.2599512338638306,
      "learning_rate": 7.233044034264034e-05,
      "loss": 10.0571,
      "step": 77
    },
    {
      "epoch": 0.3818849449204406,
      "grad_norm": 1.2699353694915771,
      "learning_rate": 7.158771761692464e-05,
      "loss": 10.5324,
      "step": 78
    },
    {
      "epoch": 0.386780905752754,
      "grad_norm": 1.4043662548065186,
      "learning_rate": 7.083909302476453e-05,
      "loss": 10.8297,
      "step": 79
    },
    {
      "epoch": 0.39167686658506734,
      "grad_norm": 1.415183663368225,
      "learning_rate": 7.008477123264848e-05,
      "loss": 10.5666,
      "step": 80
    },
    {
      "epoch": 0.39657282741738065,
      "grad_norm": 1.3034484386444092,
      "learning_rate": 6.932495846462261e-05,
      "loss": 10.6123,
      "step": 81
    },
    {
      "epoch": 0.401468788249694,
      "grad_norm": 1.301733136177063,
      "learning_rate": 6.855986244591104e-05,
      "loss": 10.3449,
      "step": 82
    },
    {
      "epoch": 0.40636474908200737,
      "grad_norm": 1.5975347757339478,
      "learning_rate": 6.778969234612584e-05,
      "loss": 10.3417,
      "step": 83
    },
    {
      "epoch": 0.4112607099143207,
      "grad_norm": 1.3522671461105347,
      "learning_rate": 6.701465872208216e-05,
      "loss": 10.5421,
      "step": 84
    },
    {
      "epoch": 0.41615667074663404,
      "grad_norm": 1.494614839553833,
      "learning_rate": 6.623497346023418e-05,
      "loss": 10.8383,
      "step": 85
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 1.472745418548584,
      "learning_rate": 6.545084971874738e-05,
      "loss": 11.0545,
      "step": 86
    },
    {
      "epoch": 0.4259485924112607,
      "grad_norm": 1.4321198463439941,
      "learning_rate": 6.466250186922325e-05,
      "loss": 10.4575,
      "step": 87
    },
    {
      "epoch": 0.43084455324357407,
      "grad_norm": 1.4431061744689941,
      "learning_rate": 6.387014543809223e-05,
      "loss": 10.6799,
      "step": 88
    },
    {
      "epoch": 0.4357405140758874,
      "grad_norm": 1.5815662145614624,
      "learning_rate": 6.307399704769099e-05,
      "loss": 10.0274,
      "step": 89
    },
    {
      "epoch": 0.44063647490820074,
      "grad_norm": 1.8736461400985718,
      "learning_rate": 6.227427435703997e-05,
      "loss": 10.0374,
      "step": 90
    },
    {
      "epoch": 0.4455324357405141,
      "grad_norm": 1.7946646213531494,
      "learning_rate": 6.147119600233758e-05,
      "loss": 10.3457,
      "step": 91
    },
    {
      "epoch": 0.4504283965728274,
      "grad_norm": 1.6094313859939575,
      "learning_rate": 6.066498153718735e-05,
      "loss": 9.6412,
      "step": 92
    },
    {
      "epoch": 0.45532435740514077,
      "grad_norm": 1.8244119882583618,
      "learning_rate": 5.985585137257401e-05,
      "loss": 10.0926,
      "step": 93
    },
    {
      "epoch": 0.4602203182374541,
      "grad_norm": 1.676670789718628,
      "learning_rate": 5.90440267166055e-05,
      "loss": 9.9954,
      "step": 94
    },
    {
      "epoch": 0.46511627906976744,
      "grad_norm": 1.5506222248077393,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 10.1712,
      "step": 95
    },
    {
      "epoch": 0.4700122399020808,
      "grad_norm": 1.7559230327606201,
      "learning_rate": 5.74131823855921e-05,
      "loss": 9.247,
      "step": 96
    },
    {
      "epoch": 0.4749082007343941,
      "grad_norm": 1.8898491859436035,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 9.7197,
      "step": 97
    },
    {
      "epoch": 0.47980416156670747,
      "grad_norm": 2.0075554847717285,
      "learning_rate": 5.577423184847932e-05,
      "loss": 9.263,
      "step": 98
    },
    {
      "epoch": 0.4847001223990208,
      "grad_norm": 2.6846652030944824,
      "learning_rate": 5.495227651252315e-05,
      "loss": 9.7252,
      "step": 99
    },
    {
      "epoch": 0.48959608323133413,
      "grad_norm": 2.2351839542388916,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 8.958,
      "step": 100
    },
    {
      "epoch": 0.48959608323133413,
      "eval_loss": 2.6148223876953125,
      "eval_runtime": 10.2084,
      "eval_samples_per_second": 33.698,
      "eval_steps_per_second": 8.424,
      "step": 100
    },
    {
      "epoch": 0.4944920440636475,
      "grad_norm": 2.8941330909729004,
      "learning_rate": 5.330452921628497e-05,
      "loss": 10.7459,
      "step": 101
    },
    {
      "epoch": 0.49938800489596086,
      "grad_norm": 2.8922526836395264,
      "learning_rate": 5.247918773366112e-05,
      "loss": 10.7301,
      "step": 102
    },
    {
      "epoch": 0.5042839657282742,
      "grad_norm": 2.5065841674804688,
      "learning_rate": 5.165316846586541e-05,
      "loss": 10.9678,
      "step": 103
    },
    {
      "epoch": 0.5091799265605875,
      "grad_norm": 2.8331668376922607,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 10.9946,
      "step": 104
    },
    {
      "epoch": 0.5140758873929009,
      "grad_norm": 2.4008357524871826,
      "learning_rate": 5e-05,
      "loss": 10.9698,
      "step": 105
    },
    {
      "epoch": 0.5189718482252142,
      "grad_norm": 2.091648817062378,
      "learning_rate": 4.917330276168208e-05,
      "loss": 10.8922,
      "step": 106
    },
    {
      "epoch": 0.5238678090575275,
      "grad_norm": 1.8614530563354492,
      "learning_rate": 4.834683153413459e-05,
      "loss": 10.1095,
      "step": 107
    },
    {
      "epoch": 0.5287637698898409,
      "grad_norm": 1.6541978120803833,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 10.7482,
      "step": 108
    },
    {
      "epoch": 0.5336597307221542,
      "grad_norm": 1.503069281578064,
      "learning_rate": 4.669547078371504e-05,
      "loss": 10.2142,
      "step": 109
    },
    {
      "epoch": 0.5385556915544676,
      "grad_norm": 1.5871061086654663,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 10.8846,
      "step": 110
    },
    {
      "epoch": 0.543451652386781,
      "grad_norm": 1.438506007194519,
      "learning_rate": 4.504772348747687e-05,
      "loss": 10.365,
      "step": 111
    },
    {
      "epoch": 0.5483476132190942,
      "grad_norm": 1.1854056119918823,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 10.6455,
      "step": 112
    },
    {
      "epoch": 0.5532435740514076,
      "grad_norm": 1.2648298740386963,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 10.5801,
      "step": 113
    },
    {
      "epoch": 0.5581395348837209,
      "grad_norm": 1.160414695739746,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 10.6873,
      "step": 114
    },
    {
      "epoch": 0.5630354957160343,
      "grad_norm": 1.2953544855117798,
      "learning_rate": 4.17702704859633e-05,
      "loss": 11.0135,
      "step": 115
    },
    {
      "epoch": 0.5679314565483476,
      "grad_norm": 1.6852461099624634,
      "learning_rate": 4.095597328339452e-05,
      "loss": 10.7058,
      "step": 116
    },
    {
      "epoch": 0.572827417380661,
      "grad_norm": 1.3730648756027222,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 10.7713,
      "step": 117
    },
    {
      "epoch": 0.5777233782129743,
      "grad_norm": 1.2321871519088745,
      "learning_rate": 3.933501846281267e-05,
      "loss": 10.3063,
      "step": 118
    },
    {
      "epoch": 0.5826193390452876,
      "grad_norm": 1.377393126487732,
      "learning_rate": 3.852880399766243e-05,
      "loss": 10.4341,
      "step": 119
    },
    {
      "epoch": 0.587515299877601,
      "grad_norm": 1.338837742805481,
      "learning_rate": 3.772572564296005e-05,
      "loss": 10.2644,
      "step": 120
    },
    {
      "epoch": 0.5924112607099143,
      "grad_norm": 1.4650025367736816,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 10.421,
      "step": 121
    },
    {
      "epoch": 0.5973072215422277,
      "grad_norm": 1.3370344638824463,
      "learning_rate": 3.612985456190778e-05,
      "loss": 10.6622,
      "step": 122
    },
    {
      "epoch": 0.602203182374541,
      "grad_norm": 1.4483917951583862,
      "learning_rate": 3.533749813077677e-05,
      "loss": 10.6256,
      "step": 123
    },
    {
      "epoch": 0.6070991432068543,
      "grad_norm": 1.3943606615066528,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 10.4352,
      "step": 124
    },
    {
      "epoch": 0.6119951040391677,
      "grad_norm": 1.3855911493301392,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 10.9211,
      "step": 125
    },
    {
      "epoch": 0.616891064871481,
      "grad_norm": 1.3975025415420532,
      "learning_rate": 3.298534127791785e-05,
      "loss": 10.1287,
      "step": 126
    },
    {
      "epoch": 0.6217870257037944,
      "grad_norm": 1.3634696006774902,
      "learning_rate": 3.221030765387417e-05,
      "loss": 10.143,
      "step": 127
    },
    {
      "epoch": 0.6266829865361077,
      "grad_norm": 1.318239688873291,
      "learning_rate": 3.144013755408895e-05,
      "loss": 10.0395,
      "step": 128
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 1.3865526914596558,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 10.5748,
      "step": 129
    },
    {
      "epoch": 0.6364749082007344,
      "grad_norm": 1.4261510372161865,
      "learning_rate": 2.991522876735154e-05,
      "loss": 10.6534,
      "step": 130
    },
    {
      "epoch": 0.6413708690330477,
      "grad_norm": 1.789625883102417,
      "learning_rate": 2.916090697523549e-05,
      "loss": 10.0759,
      "step": 131
    },
    {
      "epoch": 0.6462668298653611,
      "grad_norm": 1.5815229415893555,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 9.8487,
      "step": 132
    },
    {
      "epoch": 0.6511627906976745,
      "grad_norm": 1.4309276342391968,
      "learning_rate": 2.766955965735968e-05,
      "loss": 9.9309,
      "step": 133
    },
    {
      "epoch": 0.6560587515299877,
      "grad_norm": 1.4026715755462646,
      "learning_rate": 2.693294185106562e-05,
      "loss": 10.2168,
      "step": 134
    },
    {
      "epoch": 0.6609547123623011,
      "grad_norm": 1.4140424728393555,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 9.9212,
      "step": 135
    },
    {
      "epoch": 0.6658506731946144,
      "grad_norm": 1.489136815071106,
      "learning_rate": 2.547882480847461e-05,
      "loss": 10.6646,
      "step": 136
    },
    {
      "epoch": 0.6707466340269278,
      "grad_norm": 1.5409811735153198,
      "learning_rate": 2.476172311325783e-05,
      "loss": 10.5673,
      "step": 137
    },
    {
      "epoch": 0.6756425948592412,
      "grad_norm": 1.5286870002746582,
      "learning_rate": 2.405152131093926e-05,
      "loss": 9.7692,
      "step": 138
    },
    {
      "epoch": 0.6805385556915544,
      "grad_norm": 1.512497901916504,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 9.6965,
      "step": 139
    },
    {
      "epoch": 0.6854345165238678,
      "grad_norm": 1.6204568147659302,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 9.771,
      "step": 140
    },
    {
      "epoch": 0.6903304773561811,
      "grad_norm": 1.8047850131988525,
      "learning_rate": 2.196424713241637e-05,
      "loss": 10.221,
      "step": 141
    },
    {
      "epoch": 0.6952264381884945,
      "grad_norm": 1.6464601755142212,
      "learning_rate": 2.128356686585282e-05,
      "loss": 10.5653,
      "step": 142
    },
    {
      "epoch": 0.7001223990208079,
      "grad_norm": 1.578071117401123,
      "learning_rate": 2.061073738537635e-05,
      "loss": 9.8964,
      "step": 143
    },
    {
      "epoch": 0.7050183598531212,
      "grad_norm": 1.7256425619125366,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 9.7322,
      "step": 144
    },
    {
      "epoch": 0.7099143206854345,
      "grad_norm": 1.691171407699585,
      "learning_rate": 1.928936436551661e-05,
      "loss": 9.7703,
      "step": 145
    },
    {
      "epoch": 0.7148102815177478,
      "grad_norm": 1.8565572500228882,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 9.782,
      "step": 146
    },
    {
      "epoch": 0.7197062423500612,
      "grad_norm": 1.856450080871582,
      "learning_rate": 1.800157297483417e-05,
      "loss": 9.6394,
      "step": 147
    },
    {
      "epoch": 0.7246022031823746,
      "grad_norm": 2.020888090133667,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 9.0112,
      "step": 148
    },
    {
      "epoch": 0.7294981640146879,
      "grad_norm": 2.0510852336883545,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 9.5601,
      "step": 149
    },
    {
      "epoch": 0.7343941248470012,
      "grad_norm": 2.247244358062744,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 8.7348,
      "step": 150
    },
    {
      "epoch": 0.7343941248470012,
      "eval_loss": 2.561342716217041,
      "eval_runtime": 10.2254,
      "eval_samples_per_second": 33.642,
      "eval_steps_per_second": 8.41,
      "step": 150
    },
    {
      "epoch": 0.7392900856793145,
      "grad_norm": 1.867893099784851,
      "learning_rate": 1.553232954407171e-05,
      "loss": 10.1114,
      "step": 151
    },
    {
      "epoch": 0.7441860465116279,
      "grad_norm": 1.613707423210144,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 10.8757,
      "step": 152
    },
    {
      "epoch": 0.7490820073439413,
      "grad_norm": 1.8466280698776245,
      "learning_rate": 1.435357758543015e-05,
      "loss": 10.5829,
      "step": 153
    },
    {
      "epoch": 0.7539779681762546,
      "grad_norm": 1.7909514904022217,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 10.4933,
      "step": 154
    },
    {
      "epoch": 0.758873929008568,
      "grad_norm": 1.6486133337020874,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 10.4922,
      "step": 155
    },
    {
      "epoch": 0.7637698898408812,
      "grad_norm": 1.5389719009399414,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 11.2143,
      "step": 156
    },
    {
      "epoch": 0.7686658506731946,
      "grad_norm": 1.4462549686431885,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 10.3669,
      "step": 157
    },
    {
      "epoch": 0.773561811505508,
      "grad_norm": 1.5703456401824951,
      "learning_rate": 1.157994445715706e-05,
      "loss": 10.926,
      "step": 158
    },
    {
      "epoch": 0.7784577723378213,
      "grad_norm": 1.6514225006103516,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 11.1682,
      "step": 159
    },
    {
      "epoch": 0.7833537331701347,
      "grad_norm": 1.4175182580947876,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 10.6956,
      "step": 160
    },
    {
      "epoch": 0.7882496940024479,
      "grad_norm": 1.5620766878128052,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 10.1638,
      "step": 161
    },
    {
      "epoch": 0.7931456548347613,
      "grad_norm": 1.4049431085586548,
      "learning_rate": 9.549150281252633e-06,
      "loss": 10.2151,
      "step": 162
    },
    {
      "epoch": 0.7980416156670747,
      "grad_norm": 1.2286335229873657,
      "learning_rate": 9.068759265665384e-06,
      "loss": 10.0959,
      "step": 163
    },
    {
      "epoch": 0.802937576499388,
      "grad_norm": 1.2892861366271973,
      "learning_rate": 8.599558442598998e-06,
      "loss": 9.9911,
      "step": 164
    },
    {
      "epoch": 0.8078335373317014,
      "grad_norm": 1.3772023916244507,
      "learning_rate": 8.141676086873572e-06,
      "loss": 10.2992,
      "step": 165
    },
    {
      "epoch": 0.8127294981640147,
      "grad_norm": 1.2722628116607666,
      "learning_rate": 7.695237378953223e-06,
      "loss": 10.701,
      "step": 166
    },
    {
      "epoch": 0.817625458996328,
      "grad_norm": 1.5949305295944214,
      "learning_rate": 7.260364370723044e-06,
      "loss": 10.3158,
      "step": 167
    },
    {
      "epoch": 0.8225214198286414,
      "grad_norm": 1.3147389888763428,
      "learning_rate": 6.837175952121306e-06,
      "loss": 10.9929,
      "step": 168
    },
    {
      "epoch": 0.8274173806609547,
      "grad_norm": 1.1935569047927856,
      "learning_rate": 6.425787818636131e-06,
      "loss": 10.5773,
      "step": 169
    },
    {
      "epoch": 0.8323133414932681,
      "grad_norm": 1.4854832887649536,
      "learning_rate": 6.026312439675552e-06,
      "loss": 10.3755,
      "step": 170
    },
    {
      "epoch": 0.8372093023255814,
      "grad_norm": 1.2238411903381348,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 10.0088,
      "step": 171
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 1.2479257583618164,
      "learning_rate": 5.263533508961827e-06,
      "loss": 10.51,
      "step": 172
    },
    {
      "epoch": 0.847001223990208,
      "grad_norm": 1.3344398736953735,
      "learning_rate": 4.900438493352055e-06,
      "loss": 10.2554,
      "step": 173
    },
    {
      "epoch": 0.8518971848225214,
      "grad_norm": 1.3858119249343872,
      "learning_rate": 4.549673247541875e-06,
      "loss": 10.7245,
      "step": 174
    },
    {
      "epoch": 0.8567931456548348,
      "grad_norm": 1.677107810974121,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 10.2962,
      "step": 175
    },
    {
      "epoch": 0.8616891064871481,
      "grad_norm": 1.5208059549331665,
      "learning_rate": 3.885512251130763e-06,
      "loss": 10.3727,
      "step": 176
    },
    {
      "epoch": 0.8665850673194615,
      "grad_norm": 1.3620158433914185,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 10.2428,
      "step": 177
    },
    {
      "epoch": 0.8714810281517748,
      "grad_norm": 1.3021228313446045,
      "learning_rate": 3.271776770026963e-06,
      "loss": 10.1477,
      "step": 178
    },
    {
      "epoch": 0.8763769889840881,
      "grad_norm": 1.365728735923767,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 10.6577,
      "step": 179
    },
    {
      "epoch": 0.8812729498164015,
      "grad_norm": 1.3312187194824219,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 10.2708,
      "step": 180
    },
    {
      "epoch": 0.8861689106487148,
      "grad_norm": 1.4148651361465454,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 10.4765,
      "step": 181
    },
    {
      "epoch": 0.8910648714810282,
      "grad_norm": 1.334594964981079,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 9.9097,
      "step": 182
    },
    {
      "epoch": 0.8959608323133414,
      "grad_norm": 1.4704692363739014,
      "learning_rate": 1.962316193157593e-06,
      "loss": 9.9967,
      "step": 183
    },
    {
      "epoch": 0.9008567931456548,
      "grad_norm": 1.3939872980117798,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 10.1484,
      "step": 184
    },
    {
      "epoch": 0.9057527539779682,
      "grad_norm": 1.5668071508407593,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 10.1871,
      "step": 185
    },
    {
      "epoch": 0.9106487148102815,
      "grad_norm": 1.7102683782577515,
      "learning_rate": 1.333670137599713e-06,
      "loss": 9.6067,
      "step": 186
    },
    {
      "epoch": 0.9155446756425949,
      "grad_norm": 1.5264465808868408,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 10.0599,
      "step": 187
    },
    {
      "epoch": 0.9204406364749081,
      "grad_norm": 1.370841145515442,
      "learning_rate": 9.810017062595322e-07,
      "loss": 10.0868,
      "step": 188
    },
    {
      "epoch": 0.9253365973072215,
      "grad_norm": 1.594359040260315,
      "learning_rate": 8.247462563808817e-07,
      "loss": 9.9223,
      "step": 189
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 1.6128777265548706,
      "learning_rate": 6.819348298638839e-07,
      "loss": 10.1382,
      "step": 190
    },
    {
      "epoch": 0.9351285189718482,
      "grad_norm": 1.630208134651184,
      "learning_rate": 5.526064699265753e-07,
      "loss": 9.8475,
      "step": 191
    },
    {
      "epoch": 0.9400244798041616,
      "grad_norm": 1.6013586521148682,
      "learning_rate": 4.367965336512403e-07,
      "loss": 10.0391,
      "step": 192
    },
    {
      "epoch": 0.944920440636475,
      "grad_norm": 2.0071372985839844,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 10.063,
      "step": 193
    },
    {
      "epoch": 0.9498164014687882,
      "grad_norm": 2.3870363235473633,
      "learning_rate": 2.458548727494292e-07,
      "loss": 9.9665,
      "step": 194
    },
    {
      "epoch": 0.9547123623011016,
      "grad_norm": 2.370018720626831,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 10.1208,
      "step": 195
    },
    {
      "epoch": 0.9596083231334149,
      "grad_norm": 1.9048367738723755,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 9.301,
      "step": 196
    },
    {
      "epoch": 0.9645042839657283,
      "grad_norm": 2.159170150756836,
      "learning_rate": 6.150154258476315e-08,
      "loss": 10.1897,
      "step": 197
    },
    {
      "epoch": 0.9694002447980417,
      "grad_norm": 2.348414182662964,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 9.3109,
      "step": 198
    },
    {
      "epoch": 0.9742962056303549,
      "grad_norm": 2.7795143127441406,
      "learning_rate": 6.834750376549792e-09,
      "loss": 9.5603,
      "step": 199
    },
    {
      "epoch": 0.9791921664626683,
      "grad_norm": 2.6396398544311523,
      "learning_rate": 0.0,
      "loss": 8.9188,
      "step": 200
    },
    {
      "epoch": 0.9791921664626683,
      "eval_loss": 2.5549092292785645,
      "eval_runtime": 10.2118,
      "eval_samples_per_second": 33.687,
      "eval_steps_per_second": 8.422,
      "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": 7099932280356864.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}