{
  "best_metric": 2.3127670288085938,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.03305238803503553,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016526194017517766,
      "grad_norm": 1.594321608543396,
      "learning_rate": 1e-05,
      "loss": 2.4819,
      "step": 1
    },
    {
      "epoch": 0.00016526194017517766,
      "eval_loss": 3.2956507205963135,
      "eval_runtime": 298.3784,
      "eval_samples_per_second": 34.155,
      "eval_steps_per_second": 8.539,
      "step": 1
    },
    {
      "epoch": 0.0003305238803503553,
      "grad_norm": 1.8362373113632202,
      "learning_rate": 2e-05,
      "loss": 2.474,
      "step": 2
    },
    {
      "epoch": 0.0004957858205255329,
      "grad_norm": 2.0558102130889893,
      "learning_rate": 3e-05,
      "loss": 2.5933,
      "step": 3
    },
    {
      "epoch": 0.0006610477607007106,
      "grad_norm": 1.5579073429107666,
      "learning_rate": 4e-05,
      "loss": 2.42,
      "step": 4
    },
    {
      "epoch": 0.0008263097008758883,
      "grad_norm": 2.087095260620117,
      "learning_rate": 5e-05,
      "loss": 2.5554,
      "step": 5
    },
    {
      "epoch": 0.0009915716410510659,
      "grad_norm": 2.1744778156280518,
      "learning_rate": 6e-05,
      "loss": 2.7619,
      "step": 6
    },
    {
      "epoch": 0.0011568335812262437,
      "grad_norm": 2.49554705619812,
      "learning_rate": 7e-05,
      "loss": 2.7488,
      "step": 7
    },
    {
      "epoch": 0.0013220955214014213,
      "grad_norm": 2.244565010070801,
      "learning_rate": 8e-05,
      "loss": 2.6717,
      "step": 8
    },
    {
      "epoch": 0.001487357461576599,
      "grad_norm": 1.63128662109375,
      "learning_rate": 9e-05,
      "loss": 2.5385,
      "step": 9
    },
    {
      "epoch": 0.0016526194017517765,
      "grad_norm": 1.5725308656692505,
      "learning_rate": 0.0001,
      "loss": 2.5384,
      "step": 10
    },
    {
      "epoch": 0.0018178813419269541,
      "grad_norm": 2.130584478378296,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.8119,
      "step": 11
    },
    {
      "epoch": 0.0019831432821021317,
      "grad_norm": 1.7600144147872925,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.3037,
      "step": 12
    },
    {
      "epoch": 0.0021484052222773093,
      "grad_norm": 1.920034646987915,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.515,
      "step": 13
    },
    {
      "epoch": 0.0023136671624524874,
      "grad_norm": 1.5650461912155151,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.5834,
      "step": 14
    },
    {
      "epoch": 0.002478929102627665,
      "grad_norm": 1.933816909790039,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.5565,
      "step": 15
    },
    {
      "epoch": 0.0026441910428028426,
      "grad_norm": 1.9310052394866943,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.3907,
      "step": 16
    },
    {
      "epoch": 0.00280945298297802,
      "grad_norm": 2.0040125846862793,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.4109,
      "step": 17
    },
    {
      "epoch": 0.002974714923153198,
      "grad_norm": 1.7697948217391968,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.4493,
      "step": 18
    },
    {
      "epoch": 0.0031399768633283754,
      "grad_norm": 1.7867257595062256,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.6587,
      "step": 19
    },
    {
      "epoch": 0.003305238803503553,
      "grad_norm": 1.7235246896743774,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.4252,
      "step": 20
    },
    {
      "epoch": 0.0034705007436787306,
      "grad_norm": 1.6747289896011353,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.5149,
      "step": 21
    },
    {
      "epoch": 0.0036357626838539082,
      "grad_norm": 1.5265719890594482,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.3857,
      "step": 22
    },
    {
      "epoch": 0.0038010246240290863,
      "grad_norm": 1.76333749294281,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.6925,
      "step": 23
    },
    {
      "epoch": 0.0039662865642042635,
      "grad_norm": 1.7076408863067627,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.5469,
      "step": 24
    },
    {
      "epoch": 0.0041315485043794415,
      "grad_norm": 1.6395620107650757,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.4239,
      "step": 25
    },
    {
      "epoch": 0.004296810444554619,
      "grad_norm": 1.6570926904678345,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.3486,
      "step": 26
    },
    {
      "epoch": 0.004462072384729797,
      "grad_norm": 1.6805280447006226,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.395,
      "step": 27
    },
    {
      "epoch": 0.004627334324904975,
      "grad_norm": 1.8204387426376343,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.4308,
      "step": 28
    },
    {
      "epoch": 0.004792596265080152,
      "grad_norm": 1.7184354066848755,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.3483,
      "step": 29
    },
    {
      "epoch": 0.00495785820525533,
      "grad_norm": 2.094357490539551,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.4923,
      "step": 30
    },
    {
      "epoch": 0.005123120145430507,
      "grad_norm": 1.8471448421478271,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.4706,
      "step": 31
    },
    {
      "epoch": 0.005288382085605685,
      "grad_norm": 2.124126434326172,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.5576,
      "step": 32
    },
    {
      "epoch": 0.005453644025780862,
      "grad_norm": 1.85663902759552,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.4436,
      "step": 33
    },
    {
      "epoch": 0.00561890596595604,
      "grad_norm": 1.9243381023406982,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.7482,
      "step": 34
    },
    {
      "epoch": 0.005784167906131218,
      "grad_norm": 2.0597476959228516,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.5625,
      "step": 35
    },
    {
      "epoch": 0.005949429846306396,
      "grad_norm": 1.9038246870040894,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.6131,
      "step": 36
    },
    {
      "epoch": 0.006114691786481574,
      "grad_norm": 1.9353429079055786,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.4935,
      "step": 37
    },
    {
      "epoch": 0.006279953726656751,
      "grad_norm": 1.7112752199172974,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.6013,
      "step": 38
    },
    {
      "epoch": 0.006445215666831929,
      "grad_norm": 2.026658535003662,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.5474,
      "step": 39
    },
    {
      "epoch": 0.006610477607007106,
      "grad_norm": 1.8708208799362183,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.5254,
      "step": 40
    },
    {
      "epoch": 0.006775739547182284,
      "grad_norm": 1.9076058864593506,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.5782,
      "step": 41
    },
    {
      "epoch": 0.006941001487357461,
      "grad_norm": 1.8290849924087524,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.591,
      "step": 42
    },
    {
      "epoch": 0.007106263427532639,
      "grad_norm": 2.0316317081451416,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.6142,
      "step": 43
    },
    {
      "epoch": 0.0072715253677078165,
      "grad_norm": 1.8418331146240234,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.3819,
      "step": 44
    },
    {
      "epoch": 0.0074367873078829945,
      "grad_norm": 2.284447431564331,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.6443,
      "step": 45
    },
    {
      "epoch": 0.007602049248058173,
      "grad_norm": 2.4346821308135986,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.5674,
      "step": 46
    },
    {
      "epoch": 0.00776731118823335,
      "grad_norm": 2.468438148498535,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.5521,
      "step": 47
    },
    {
      "epoch": 0.007932573128408527,
      "grad_norm": 2.068291664123535,
      "learning_rate": 9.045084971874738e-05,
      "loss": 2.391,
      "step": 48
    },
    {
      "epoch": 0.008097835068583706,
      "grad_norm": 2.661067008972168,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.8375,
      "step": 49
    },
    {
      "epoch": 0.008263097008758883,
      "grad_norm": 2.414609670639038,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.7058,
      "step": 50
    },
    {
      "epoch": 0.008263097008758883,
      "eval_loss": 2.5834617614746094,
      "eval_runtime": 298.8144,
      "eval_samples_per_second": 34.105,
      "eval_steps_per_second": 8.527,
      "step": 50
    },
    {
      "epoch": 0.00842835894893406,
      "grad_norm": 3.0396018028259277,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.2944,
      "step": 51
    },
    {
      "epoch": 0.008593620889109237,
      "grad_norm": 2.1757869720458984,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.2936,
      "step": 52
    },
    {
      "epoch": 0.008758882829284416,
      "grad_norm": 1.5016415119171143,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.3452,
      "step": 53
    },
    {
      "epoch": 0.008924144769459593,
      "grad_norm": 1.3061988353729248,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.2717,
      "step": 54
    },
    {
      "epoch": 0.00908940670963477,
      "grad_norm": 1.2057137489318848,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1364,
      "step": 55
    },
    {
      "epoch": 0.00925466864980995,
      "grad_norm": 1.2943191528320312,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.38,
      "step": 56
    },
    {
      "epoch": 0.009419930589985127,
      "grad_norm": 1.4102344512939453,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.3099,
      "step": 57
    },
    {
      "epoch": 0.009585192530160304,
      "grad_norm": 1.5076066255569458,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.1414,
      "step": 58
    },
    {
      "epoch": 0.009750454470335481,
      "grad_norm": 1.3501635789871216,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.2923,
      "step": 59
    },
    {
      "epoch": 0.00991571641051066,
      "grad_norm": 1.1580036878585815,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.1756,
      "step": 60
    },
    {
      "epoch": 0.010080978350685837,
      "grad_norm": 1.290929913520813,
      "learning_rate": 8.32512286056924e-05,
      "loss": 2.3641,
      "step": 61
    },
    {
      "epoch": 0.010246240290861014,
      "grad_norm": 1.1682236194610596,
      "learning_rate": 8.262928807620843e-05,
      "loss": 2.1672,
      "step": 62
    },
    {
      "epoch": 0.010411502231036193,
      "grad_norm": 1.239505648612976,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.3083,
      "step": 63
    },
    {
      "epoch": 0.01057676417121137,
      "grad_norm": 1.268985629081726,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.0984,
      "step": 64
    },
    {
      "epoch": 0.010742026111386548,
      "grad_norm": 1.3137507438659668,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.4505,
      "step": 65
    },
    {
      "epoch": 0.010907288051561725,
      "grad_norm": 1.4477914571762085,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.2489,
      "step": 66
    },
    {
      "epoch": 0.011072549991736904,
      "grad_norm": 1.4892327785491943,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.2383,
      "step": 67
    },
    {
      "epoch": 0.01123781193191208,
      "grad_norm": 1.4834392070770264,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3078,
      "step": 68
    },
    {
      "epoch": 0.011403073872087258,
      "grad_norm": 1.5023738145828247,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.546,
      "step": 69
    },
    {
      "epoch": 0.011568335812262435,
      "grad_norm": 1.404241919517517,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.48,
      "step": 70
    },
    {
      "epoch": 0.011733597752437614,
      "grad_norm": 1.3176933526992798,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.2899,
      "step": 71
    },
    {
      "epoch": 0.011898859692612791,
      "grad_norm": 1.3202505111694336,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.3109,
      "step": 72
    },
    {
      "epoch": 0.012064121632787968,
      "grad_norm": 1.3403223752975464,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.3758,
      "step": 73
    },
    {
      "epoch": 0.012229383572963147,
      "grad_norm": 1.6575223207473755,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.3825,
      "step": 74
    },
    {
      "epoch": 0.012394645513138325,
      "grad_norm": 1.2987710237503052,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.2338,
      "step": 75
    },
    {
      "epoch": 0.012559907453313502,
      "grad_norm": 1.4904484748840332,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.5069,
      "step": 76
    },
    {
      "epoch": 0.012725169393488679,
      "grad_norm": 1.4264194965362549,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.3788,
      "step": 77
    },
    {
      "epoch": 0.012890431333663858,
      "grad_norm": 1.6009210348129272,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.2594,
      "step": 78
    },
    {
      "epoch": 0.013055693273839035,
      "grad_norm": 1.6254149675369263,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.57,
      "step": 79
    },
    {
      "epoch": 0.013220955214014212,
      "grad_norm": 1.6565309762954712,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.2785,
      "step": 80
    },
    {
      "epoch": 0.013386217154189391,
      "grad_norm": 1.3984935283660889,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.036,
      "step": 81
    },
    {
      "epoch": 0.013551479094364568,
      "grad_norm": 1.9303289651870728,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.4273,
      "step": 82
    },
    {
      "epoch": 0.013716741034539745,
      "grad_norm": 1.6424449682235718,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.1854,
      "step": 83
    },
    {
      "epoch": 0.013882002974714923,
      "grad_norm": 1.7099841833114624,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.3934,
      "step": 84
    },
    {
      "epoch": 0.014047264914890101,
      "grad_norm": 1.7815921306610107,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.4998,
      "step": 85
    },
    {
      "epoch": 0.014212526855065279,
      "grad_norm": 2.057406187057495,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.7939,
      "step": 86
    },
    {
      "epoch": 0.014377788795240456,
      "grad_norm": 1.9033821821212769,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.9531,
      "step": 87
    },
    {
      "epoch": 0.014543050735415633,
      "grad_norm": 2.0324106216430664,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.4984,
      "step": 88
    },
    {
      "epoch": 0.014708312675590812,
      "grad_norm": 1.8717880249023438,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.1211,
      "step": 89
    },
    {
      "epoch": 0.014873574615765989,
      "grad_norm": 1.696349859237671,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.2934,
      "step": 90
    },
    {
      "epoch": 0.015038836555941166,
      "grad_norm": 1.9078240394592285,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.4263,
      "step": 91
    },
    {
      "epoch": 0.015204098496116345,
      "grad_norm": 1.867570161819458,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.5714,
      "step": 92
    },
    {
      "epoch": 0.015369360436291522,
      "grad_norm": 1.8307164907455444,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.3361,
      "step": 93
    },
    {
      "epoch": 0.0155346223764667,
      "grad_norm": 1.7124042510986328,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.3351,
      "step": 94
    },
    {
      "epoch": 0.015699884316641877,
      "grad_norm": 1.8100595474243164,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.5527,
      "step": 95
    },
    {
      "epoch": 0.015865146256817054,
      "grad_norm": 1.7517802715301514,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.392,
      "step": 96
    },
    {
      "epoch": 0.01603040819699223,
      "grad_norm": 1.9452416896820068,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.5972,
      "step": 97
    },
    {
      "epoch": 0.01619567013716741,
      "grad_norm": 2.082277297973633,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.5602,
      "step": 98
    },
    {
      "epoch": 0.01636093207734259,
      "grad_norm": 2.1542720794677734,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.5349,
      "step": 99
    },
    {
      "epoch": 0.016526194017517766,
      "grad_norm": 3.09428334236145,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 2.8479,
      "step": 100
    },
    {
      "epoch": 0.016526194017517766,
      "eval_loss": 2.4071338176727295,
      "eval_runtime": 298.8093,
      "eval_samples_per_second": 34.105,
      "eval_steps_per_second": 8.527,
      "step": 100
    },
    {
      "epoch": 0.016691455957692943,
      "grad_norm": 1.3711987733840942,
      "learning_rate": 5.330452921628497e-05,
      "loss": 2.2082,
      "step": 101
    },
    {
      "epoch": 0.01685671789786812,
      "grad_norm": 1.3217147588729858,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.1227,
      "step": 102
    },
    {
      "epoch": 0.017021979838043298,
      "grad_norm": 1.2512484788894653,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.2301,
      "step": 103
    },
    {
      "epoch": 0.017187241778218475,
      "grad_norm": 1.229514241218567,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.1412,
      "step": 104
    },
    {
      "epoch": 0.017352503718393655,
      "grad_norm": 1.2918286323547363,
      "learning_rate": 5e-05,
      "loss": 2.0837,
      "step": 105
    },
    {
      "epoch": 0.017517765658568833,
      "grad_norm": 1.0894007682800293,
      "learning_rate": 4.917330276168208e-05,
      "loss": 2.1457,
      "step": 106
    },
    {
      "epoch": 0.01768302759874401,
      "grad_norm": 1.4407212734222412,
      "learning_rate": 4.834683153413459e-05,
      "loss": 2.2183,
      "step": 107
    },
    {
      "epoch": 0.017848289538919187,
      "grad_norm": 1.1538491249084473,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.0557,
      "step": 108
    },
    {
      "epoch": 0.018013551479094364,
      "grad_norm": 1.211962342262268,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.1331,
      "step": 109
    },
    {
      "epoch": 0.01817881341926954,
      "grad_norm": 1.2235366106033325,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.1654,
      "step": 110
    },
    {
      "epoch": 0.01834407535944472,
      "grad_norm": 1.2009706497192383,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.152,
      "step": 111
    },
    {
      "epoch": 0.0185093372996199,
      "grad_norm": 1.2473372220993042,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.2714,
      "step": 112
    },
    {
      "epoch": 0.018674599239795076,
      "grad_norm": 1.2723076343536377,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.2163,
      "step": 113
    },
    {
      "epoch": 0.018839861179970253,
      "grad_norm": 1.4569956064224243,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.3087,
      "step": 114
    },
    {
      "epoch": 0.01900512312014543,
      "grad_norm": 1.3537229299545288,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.235,
      "step": 115
    },
    {
      "epoch": 0.019170385060320608,
      "grad_norm": 1.2044799327850342,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.0298,
      "step": 116
    },
    {
      "epoch": 0.019335647000495785,
      "grad_norm": 1.4252375364303589,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.3789,
      "step": 117
    },
    {
      "epoch": 0.019500908940670962,
      "grad_norm": 1.425341248512268,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.3746,
      "step": 118
    },
    {
      "epoch": 0.019666170880846143,
      "grad_norm": 1.3347434997558594,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.1806,
      "step": 119
    },
    {
      "epoch": 0.01983143282102132,
      "grad_norm": 1.4959675073623657,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.1389,
      "step": 120
    },
    {
      "epoch": 0.019996694761196497,
      "grad_norm": 1.3700460195541382,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.3567,
      "step": 121
    },
    {
      "epoch": 0.020161956701371674,
      "grad_norm": 1.342644453048706,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.3131,
      "step": 122
    },
    {
      "epoch": 0.02032721864154685,
      "grad_norm": 1.4528932571411133,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.2284,
      "step": 123
    },
    {
      "epoch": 0.02049248058172203,
      "grad_norm": 1.587868571281433,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.3817,
      "step": 124
    },
    {
      "epoch": 0.020657742521897206,
      "grad_norm": 1.4122402667999268,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.1179,
      "step": 125
    },
    {
      "epoch": 0.020823004462072386,
      "grad_norm": 1.3969868421554565,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.076,
      "step": 126
    },
    {
      "epoch": 0.020988266402247564,
      "grad_norm": 1.4013752937316895,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.2594,
      "step": 127
    },
    {
      "epoch": 0.02115352834242274,
      "grad_norm": 1.3998066186904907,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.2329,
      "step": 128
    },
    {
      "epoch": 0.021318790282597918,
      "grad_norm": 1.5217996835708618,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.3423,
      "step": 129
    },
    {
      "epoch": 0.021484052222773095,
      "grad_norm": 1.6962171792984009,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.5244,
      "step": 130
    },
    {
      "epoch": 0.021649314162948272,
      "grad_norm": 1.340029239654541,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.0709,
      "step": 131
    },
    {
      "epoch": 0.02181457610312345,
      "grad_norm": 1.635557770729065,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.417,
      "step": 132
    },
    {
      "epoch": 0.021979838043298627,
      "grad_norm": 1.4261572360992432,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.3794,
      "step": 133
    },
    {
      "epoch": 0.022145099983473807,
      "grad_norm": 1.4236623048782349,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.2441,
      "step": 134
    },
    {
      "epoch": 0.022310361923648984,
      "grad_norm": 1.4902772903442383,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.2744,
      "step": 135
    },
    {
      "epoch": 0.02247562386382416,
      "grad_norm": 1.6812496185302734,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.2461,
      "step": 136
    },
    {
      "epoch": 0.02264088580399934,
      "grad_norm": 1.6441234350204468,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.2716,
      "step": 137
    },
    {
      "epoch": 0.022806147744174516,
      "grad_norm": 1.6247007846832275,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.4147,
      "step": 138
    },
    {
      "epoch": 0.022971409684349693,
      "grad_norm": 1.7661762237548828,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.2377,
      "step": 139
    },
    {
      "epoch": 0.02313667162452487,
      "grad_norm": 1.6346014738082886,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.2171,
      "step": 140
    },
    {
      "epoch": 0.02330193356470005,
      "grad_norm": 2.0629405975341797,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.3785,
      "step": 141
    },
    {
      "epoch": 0.023467195504875228,
      "grad_norm": 1.7220429182052612,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.2389,
      "step": 142
    },
    {
      "epoch": 0.023632457445050405,
      "grad_norm": 1.9484666585922241,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.5385,
      "step": 143
    },
    {
      "epoch": 0.023797719385225583,
      "grad_norm": 1.6129279136657715,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.0664,
      "step": 144
    },
    {
      "epoch": 0.02396298132540076,
      "grad_norm": 1.9319655895233154,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.4997,
      "step": 145
    },
    {
      "epoch": 0.024128243265575937,
      "grad_norm": 1.8399348258972168,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.5167,
      "step": 146
    },
    {
      "epoch": 0.024293505205751114,
      "grad_norm": 1.7930209636688232,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.2799,
      "step": 147
    },
    {
      "epoch": 0.024458767145926295,
      "grad_norm": 1.9988473653793335,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 2.3344,
      "step": 148
    },
    {
      "epoch": 0.024624029086101472,
      "grad_norm": 2.2595012187957764,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.7397,
      "step": 149
    },
    {
      "epoch": 0.02478929102627665,
      "grad_norm": 2.4940292835235596,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 2.8146,
      "step": 150
    },
    {
      "epoch": 0.02478929102627665,
      "eval_loss": 2.332402229309082,
      "eval_runtime": 298.8452,
      "eval_samples_per_second": 34.101,
      "eval_steps_per_second": 8.526,
      "step": 150
    },
    {
      "epoch": 0.024954552966451826,
      "grad_norm": 1.0474474430084229,
      "learning_rate": 1.553232954407171e-05,
      "loss": 2.1341,
      "step": 151
    },
    {
      "epoch": 0.025119814906627003,
      "grad_norm": 1.2102326154708862,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.9576,
      "step": 152
    },
    {
      "epoch": 0.02528507684680218,
      "grad_norm": 1.1026294231414795,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.9797,
      "step": 153
    },
    {
      "epoch": 0.025450338786977358,
      "grad_norm": 1.2603615522384644,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.0646,
      "step": 154
    },
    {
      "epoch": 0.02561560072715254,
      "grad_norm": 1.139230728149414,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 2.0735,
      "step": 155
    },
    {
      "epoch": 0.025780862667327716,
      "grad_norm": 1.157387614250183,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.9056,
      "step": 156
    },
    {
      "epoch": 0.025946124607502893,
      "grad_norm": 1.1281324625015259,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 2.2624,
      "step": 157
    },
    {
      "epoch": 0.02611138654767807,
      "grad_norm": 1.256828784942627,
      "learning_rate": 1.157994445715706e-05,
      "loss": 2.1557,
      "step": 158
    },
    {
      "epoch": 0.026276648487853247,
      "grad_norm": 1.351871132850647,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.3004,
      "step": 159
    },
    {
      "epoch": 0.026441910428028424,
      "grad_norm": 1.1952428817749023,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.3049,
      "step": 160
    },
    {
      "epoch": 0.0266071723682036,
      "grad_norm": 1.232858419418335,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.1426,
      "step": 161
    },
    {
      "epoch": 0.026772434308378782,
      "grad_norm": 1.241575837135315,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.1978,
      "step": 162
    },
    {
      "epoch": 0.02693769624855396,
      "grad_norm": 1.3630142211914062,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.3025,
      "step": 163
    },
    {
      "epoch": 0.027102958188729136,
      "grad_norm": 1.2903443574905396,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.2243,
      "step": 164
    },
    {
      "epoch": 0.027268220128904314,
      "grad_norm": 1.4098113775253296,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.1784,
      "step": 165
    },
    {
      "epoch": 0.02743348206907949,
      "grad_norm": 1.2299137115478516,
      "learning_rate": 7.695237378953223e-06,
      "loss": 2.1628,
      "step": 166
    },
    {
      "epoch": 0.027598744009254668,
      "grad_norm": 1.313812494277954,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.231,
      "step": 167
    },
    {
      "epoch": 0.027764005949429845,
      "grad_norm": 1.3522822856903076,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.2089,
      "step": 168
    },
    {
      "epoch": 0.027929267889605022,
      "grad_norm": 1.5807390213012695,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.2813,
      "step": 169
    },
    {
      "epoch": 0.028094529829780203,
      "grad_norm": 1.413523554801941,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.1175,
      "step": 170
    },
    {
      "epoch": 0.02825979176995538,
      "grad_norm": 1.3760771751403809,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.1888,
      "step": 171
    },
    {
      "epoch": 0.028425053710130557,
      "grad_norm": 1.3485667705535889,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.1243,
      "step": 172
    },
    {
      "epoch": 0.028590315650305734,
      "grad_norm": 1.4076825380325317,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.1172,
      "step": 173
    },
    {
      "epoch": 0.02875557759048091,
      "grad_norm": 1.4966665506362915,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.3045,
      "step": 174
    },
    {
      "epoch": 0.02892083953065609,
      "grad_norm": 1.390042781829834,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.0924,
      "step": 175
    },
    {
      "epoch": 0.029086101470831266,
      "grad_norm": 1.5770869255065918,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.1363,
      "step": 176
    },
    {
      "epoch": 0.029251363411006447,
      "grad_norm": 1.498537302017212,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.325,
      "step": 177
    },
    {
      "epoch": 0.029416625351181624,
      "grad_norm": 1.6831473112106323,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.3059,
      "step": 178
    },
    {
      "epoch": 0.0295818872913568,
      "grad_norm": 1.4050158262252808,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.1108,
      "step": 179
    },
    {
      "epoch": 0.029747149231531978,
      "grad_norm": 1.6712894439697266,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.2501,
      "step": 180
    },
    {
      "epoch": 0.029912411171707155,
      "grad_norm": 1.8563475608825684,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.4876,
      "step": 181
    },
    {
      "epoch": 0.030077673111882332,
      "grad_norm": 1.5725376605987549,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.1577,
      "step": 182
    },
    {
      "epoch": 0.03024293505205751,
      "grad_norm": 1.4795066118240356,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.0373,
      "step": 183
    },
    {
      "epoch": 0.03040819699223269,
      "grad_norm": 1.7696070671081543,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.3244,
      "step": 184
    },
    {
      "epoch": 0.030573458932407867,
      "grad_norm": 1.731056809425354,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.3391,
      "step": 185
    },
    {
      "epoch": 0.030738720872583045,
      "grad_norm": 1.7064017057418823,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.2635,
      "step": 186
    },
    {
      "epoch": 0.030903982812758222,
      "grad_norm": 1.615507960319519,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.0639,
      "step": 187
    },
    {
      "epoch": 0.0310692447529334,
      "grad_norm": 1.7159605026245117,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.3516,
      "step": 188
    },
    {
      "epoch": 0.031234506693108576,
      "grad_norm": 1.7344233989715576,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.2243,
      "step": 189
    },
    {
      "epoch": 0.03139976863328375,
      "grad_norm": 1.6940995454788208,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.2338,
      "step": 190
    },
    {
      "epoch": 0.031565030573458934,
      "grad_norm": 1.8071060180664062,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.3151,
      "step": 191
    },
    {
      "epoch": 0.03173029251363411,
      "grad_norm": 1.8728740215301514,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.3952,
      "step": 192
    },
    {
      "epoch": 0.03189555445380929,
      "grad_norm": 1.7236273288726807,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.2221,
      "step": 193
    },
    {
      "epoch": 0.03206081639398446,
      "grad_norm": 2.014632225036621,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.4298,
      "step": 194
    },
    {
      "epoch": 0.03222607833415964,
      "grad_norm": 1.8893264532089233,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.4412,
      "step": 195
    },
    {
      "epoch": 0.03239134027433482,
      "grad_norm": 2.013986110687256,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 2.2003,
      "step": 196
    },
    {
      "epoch": 0.03255660221451,
      "grad_norm": 1.9729688167572021,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.3512,
      "step": 197
    },
    {
      "epoch": 0.03272186415468518,
      "grad_norm": 2.01000714302063,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.3895,
      "step": 198
    },
    {
      "epoch": 0.03288712609486035,
      "grad_norm": 2.135226011276245,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.2905,
      "step": 199
    },
    {
      "epoch": 0.03305238803503553,
      "grad_norm": 2.7688214778900146,
      "learning_rate": 0.0,
      "loss": 2.8933,
      "step": 200
    },
    {
      "epoch": 0.03305238803503553,
      "eval_loss": 2.3127670288085938,
      "eval_runtime": 298.8974,
      "eval_samples_per_second": 34.095,
      "eval_steps_per_second": 8.525,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.4428946137088e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}