{
  "best_metric": 0.8850148916244507,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.8429926238145417,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004214963119072708,
      "grad_norm": 41.612220764160156,
      "learning_rate": 1e-05,
      "loss": 7.2132,
      "step": 1
    },
    {
      "epoch": 0.004214963119072708,
      "eval_loss": 8.23657512664795,
      "eval_runtime": 6.4615,
      "eval_samples_per_second": 61.905,
      "eval_steps_per_second": 15.476,
      "step": 1
    },
    {
      "epoch": 0.008429926238145416,
      "grad_norm": 51.6939697265625,
      "learning_rate": 2e-05,
      "loss": 7.8276,
      "step": 2
    },
    {
      "epoch": 0.012644889357218124,
      "grad_norm": 30.254961013793945,
      "learning_rate": 3e-05,
      "loss": 7.1067,
      "step": 3
    },
    {
      "epoch": 0.01685985247629083,
      "grad_norm": 25.40729331970215,
      "learning_rate": 4e-05,
      "loss": 6.709,
      "step": 4
    },
    {
      "epoch": 0.02107481559536354,
      "grad_norm": 13.173149108886719,
      "learning_rate": 5e-05,
      "loss": 6.1096,
      "step": 5
    },
    {
      "epoch": 0.02528977871443625,
      "grad_norm": 11.012861251831055,
      "learning_rate": 6e-05,
      "loss": 5.8745,
      "step": 6
    },
    {
      "epoch": 0.029504741833508957,
      "grad_norm": 8.849925994873047,
      "learning_rate": 7e-05,
      "loss": 5.4415,
      "step": 7
    },
    {
      "epoch": 0.03371970495258166,
      "grad_norm": 9.389341354370117,
      "learning_rate": 8e-05,
      "loss": 4.8593,
      "step": 8
    },
    {
      "epoch": 0.037934668071654375,
      "grad_norm": 6.585432529449463,
      "learning_rate": 9e-05,
      "loss": 4.5175,
      "step": 9
    },
    {
      "epoch": 0.04214963119072708,
      "grad_norm": 4.894410610198975,
      "learning_rate": 0.0001,
      "loss": 4.0759,
      "step": 10
    },
    {
      "epoch": 0.04636459430979979,
      "grad_norm": 4.423712730407715,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.0971,
      "step": 11
    },
    {
      "epoch": 0.0505795574288725,
      "grad_norm": 4.499648571014404,
      "learning_rate": 9.997266286704631e-05,
      "loss": 3.6501,
      "step": 12
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 5.415881156921387,
      "learning_rate": 9.993849845741524e-05,
      "loss": 3.5986,
      "step": 13
    },
    {
      "epoch": 0.059009483667017915,
      "grad_norm": 4.25494384765625,
      "learning_rate": 9.989068136093873e-05,
      "loss": 3.3206,
      "step": 14
    },
    {
      "epoch": 0.06322444678609063,
      "grad_norm": 3.8527588844299316,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.8618,
      "step": 15
    },
    {
      "epoch": 0.06743940990516333,
      "grad_norm": 3.8190791606903076,
      "learning_rate": 9.975414512725057e-05,
      "loss": 3.1456,
      "step": 16
    },
    {
      "epoch": 0.07165437302423604,
      "grad_norm": 4.118343353271484,
      "learning_rate": 9.966546331768191e-05,
      "loss": 3.1063,
      "step": 17
    },
    {
      "epoch": 0.07586933614330875,
      "grad_norm": 3.4795942306518555,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.7793,
      "step": 18
    },
    {
      "epoch": 0.08008429926238145,
      "grad_norm": 4.708921432495117,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.83,
      "step": 19
    },
    {
      "epoch": 0.08429926238145416,
      "grad_norm": 3.6379666328430176,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.617,
      "step": 20
    },
    {
      "epoch": 0.08851422550052687,
      "grad_norm": 6.122097492218018,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.629,
      "step": 21
    },
    {
      "epoch": 0.09272918861959958,
      "grad_norm": 3.8281404972076416,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.7323,
      "step": 22
    },
    {
      "epoch": 0.09694415173867228,
      "grad_norm": 3.2188069820404053,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2853,
      "step": 23
    },
    {
      "epoch": 0.101159114857745,
      "grad_norm": 3.772247552871704,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.5451,
      "step": 24
    },
    {
      "epoch": 0.1053740779768177,
      "grad_norm": 3.3952317237854004,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.4106,
      "step": 25
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 3.4862706661224365,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.4085,
      "step": 26
    },
    {
      "epoch": 0.11380400421496312,
      "grad_norm": 3.5459954738616943,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.4619,
      "step": 27
    },
    {
      "epoch": 0.11801896733403583,
      "grad_norm": 3.4695639610290527,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.0973,
      "step": 28
    },
    {
      "epoch": 0.12223393045310854,
      "grad_norm": 4.246074676513672,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.2455,
      "step": 29
    },
    {
      "epoch": 0.12644889357218125,
      "grad_norm": 3.55924654006958,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.0533,
      "step": 30
    },
    {
      "epoch": 0.13066385669125394,
      "grad_norm": 3.641091823577881,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.0145,
      "step": 31
    },
    {
      "epoch": 0.13487881981032665,
      "grad_norm": 3.1307079792022705,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.7619,
      "step": 32
    },
    {
      "epoch": 0.13909378292939936,
      "grad_norm": 4.551141262054443,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.4398,
      "step": 33
    },
    {
      "epoch": 0.14330874604847207,
      "grad_norm": 3.9455955028533936,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.1323,
      "step": 34
    },
    {
      "epoch": 0.1475237091675448,
      "grad_norm": 3.8257951736450195,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.2129,
      "step": 35
    },
    {
      "epoch": 0.1517386722866175,
      "grad_norm": 3.752977132797241,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.8777,
      "step": 36
    },
    {
      "epoch": 0.1559536354056902,
      "grad_norm": 3.572842597961426,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.6792,
      "step": 37
    },
    {
      "epoch": 0.1601685985247629,
      "grad_norm": 3.5856552124023438,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.8626,
      "step": 38
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 3.6300437450408936,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.8331,
      "step": 39
    },
    {
      "epoch": 0.16859852476290832,
      "grad_norm": 3.802783727645874,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.7212,
      "step": 40
    },
    {
      "epoch": 0.17281348788198103,
      "grad_norm": 3.6040241718292236,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.5991,
      "step": 41
    },
    {
      "epoch": 0.17702845100105374,
      "grad_norm": 4.066723346710205,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.785,
      "step": 42
    },
    {
      "epoch": 0.18124341412012646,
      "grad_norm": 3.7321507930755615,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.722,
      "step": 43
    },
    {
      "epoch": 0.18545837723919917,
      "grad_norm": 4.020709037780762,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5768,
      "step": 44
    },
    {
      "epoch": 0.18967334035827185,
      "grad_norm": 4.00160026550293,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.4513,
      "step": 45
    },
    {
      "epoch": 0.19388830347734456,
      "grad_norm": 4.807865619659424,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.7363,
      "step": 46
    },
    {
      "epoch": 0.19810326659641728,
      "grad_norm": 3.86297345161438,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5316,
      "step": 47
    },
    {
      "epoch": 0.20231822971549,
      "grad_norm": 4.611600875854492,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.2584,
      "step": 48
    },
    {
      "epoch": 0.2065331928345627,
      "grad_norm": 4.2867021560668945,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.3393,
      "step": 49
    },
    {
      "epoch": 0.2107481559536354,
      "grad_norm": 6.130868911743164,
      "learning_rate": 8.945702546981969e-05,
      "loss": 2.0587,
      "step": 50
    },
    {
      "epoch": 0.2107481559536354,
      "eval_loss": 1.554897427558899,
      "eval_runtime": 6.4666,
      "eval_samples_per_second": 61.857,
      "eval_steps_per_second": 15.464,
      "step": 50
    },
    {
      "epoch": 0.21496311907270813,
      "grad_norm": 4.251574516296387,
      "learning_rate": 8.894386393810563e-05,
      "loss": 2.1604,
      "step": 51
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 3.9960267543792725,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.0034,
      "step": 52
    },
    {
      "epoch": 0.22339304531085352,
      "grad_norm": 3.5435032844543457,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1025,
      "step": 53
    },
    {
      "epoch": 0.22760800842992623,
      "grad_norm": 2.965231418609619,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.6474,
      "step": 54
    },
    {
      "epoch": 0.23182297154899895,
      "grad_norm": 2.7395787239074707,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.5813,
      "step": 55
    },
    {
      "epoch": 0.23603793466807166,
      "grad_norm": 3.470393657684326,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.8214,
      "step": 56
    },
    {
      "epoch": 0.24025289778714437,
      "grad_norm": 3.1501359939575195,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.604,
      "step": 57
    },
    {
      "epoch": 0.24446786090621708,
      "grad_norm": 3.092348337173462,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.6645,
      "step": 58
    },
    {
      "epoch": 0.24868282402528977,
      "grad_norm": 2.8015921115875244,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.6998,
      "step": 59
    },
    {
      "epoch": 0.2528977871443625,
      "grad_norm": 2.905069589614868,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.5711,
      "step": 60
    },
    {
      "epoch": 0.2571127502634352,
      "grad_norm": 2.585193634033203,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.3652,
      "step": 61
    },
    {
      "epoch": 0.2613277133825079,
      "grad_norm": 3.029945135116577,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.4766,
      "step": 62
    },
    {
      "epoch": 0.2655426765015806,
      "grad_norm": 2.936624050140381,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.4975,
      "step": 63
    },
    {
      "epoch": 0.2697576396206533,
      "grad_norm": 2.7380998134613037,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4037,
      "step": 64
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 2.9049437046051025,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.5488,
      "step": 65
    },
    {
      "epoch": 0.2781875658587987,
      "grad_norm": 2.983566999435425,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.3697,
      "step": 66
    },
    {
      "epoch": 0.28240252897787144,
      "grad_norm": 3.108199119567871,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.4842,
      "step": 67
    },
    {
      "epoch": 0.28661749209694415,
      "grad_norm": 2.9537482261657715,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.2233,
      "step": 68
    },
    {
      "epoch": 0.29083245521601686,
      "grad_norm": 3.0543484687805176,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4292,
      "step": 69
    },
    {
      "epoch": 0.2950474183350896,
      "grad_norm": 3.0750956535339355,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4593,
      "step": 70
    },
    {
      "epoch": 0.2992623814541623,
      "grad_norm": 3.2659246921539307,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.2492,
      "step": 71
    },
    {
      "epoch": 0.303477344573235,
      "grad_norm": 3.628136396408081,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.5088,
      "step": 72
    },
    {
      "epoch": 0.3076923076923077,
      "grad_norm": 3.5745816230773926,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.5352,
      "step": 73
    },
    {
      "epoch": 0.3119072708113804,
      "grad_norm": 2.7928030490875244,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.121,
      "step": 74
    },
    {
      "epoch": 0.31612223393045313,
      "grad_norm": 3.3540618419647217,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.2916,
      "step": 75
    },
    {
      "epoch": 0.3203371970495258,
      "grad_norm": 3.366863965988159,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.388,
      "step": 76
    },
    {
      "epoch": 0.3245521601685985,
      "grad_norm": 3.5833253860473633,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.4952,
      "step": 77
    },
    {
      "epoch": 0.3287671232876712,
      "grad_norm": 3.5157740116119385,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.1532,
      "step": 78
    },
    {
      "epoch": 0.3329820864067439,
      "grad_norm": 3.7377912998199463,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3759,
      "step": 79
    },
    {
      "epoch": 0.33719704952581664,
      "grad_norm": 3.495645761489868,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.1822,
      "step": 80
    },
    {
      "epoch": 0.34141201264488935,
      "grad_norm": 3.4798645973205566,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.1689,
      "step": 81
    },
    {
      "epoch": 0.34562697576396206,
      "grad_norm": 3.490549087524414,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.3261,
      "step": 82
    },
    {
      "epoch": 0.3498419388830348,
      "grad_norm": 4.399184226989746,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2789,
      "step": 83
    },
    {
      "epoch": 0.3540569020021075,
      "grad_norm": 3.5043818950653076,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.2317,
      "step": 84
    },
    {
      "epoch": 0.3582718651211802,
      "grad_norm": 3.1400296688079834,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.9662,
      "step": 85
    },
    {
      "epoch": 0.3624868282402529,
      "grad_norm": 3.659299612045288,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.1803,
      "step": 86
    },
    {
      "epoch": 0.3667017913593256,
      "grad_norm": 4.160400867462158,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.413,
      "step": 87
    },
    {
      "epoch": 0.37091675447839834,
      "grad_norm": 3.50260853767395,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.0956,
      "step": 88
    },
    {
      "epoch": 0.37513171759747105,
      "grad_norm": 3.959489345550537,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.2424,
      "step": 89
    },
    {
      "epoch": 0.3793466807165437,
      "grad_norm": 3.793694019317627,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.0903,
      "step": 90
    },
    {
      "epoch": 0.3835616438356164,
      "grad_norm": 3.3994905948638916,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.0591,
      "step": 91
    },
    {
      "epoch": 0.38777660695468913,
      "grad_norm": 3.680428981781006,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.0473,
      "step": 92
    },
    {
      "epoch": 0.39199157007376184,
      "grad_norm": 3.4609715938568115,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.8781,
      "step": 93
    },
    {
      "epoch": 0.39620653319283455,
      "grad_norm": 3.00770902633667,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.6652,
      "step": 94
    },
    {
      "epoch": 0.40042149631190727,
      "grad_norm": 4.031708240509033,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1705,
      "step": 95
    },
    {
      "epoch": 0.40463645943098,
      "grad_norm": 4.026821613311768,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.8865,
      "step": 96
    },
    {
      "epoch": 0.4088514225500527,
      "grad_norm": 3.490584373474121,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.654,
      "step": 97
    },
    {
      "epoch": 0.4130663856691254,
      "grad_norm": 4.305174350738525,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.8525,
      "step": 98
    },
    {
      "epoch": 0.4172813487881981,
      "grad_norm": 3.9346580505371094,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.9059,
      "step": 99
    },
    {
      "epoch": 0.4214963119072708,
      "grad_norm": 4.256545066833496,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.9487,
      "step": 100
    },
    {
      "epoch": 0.4214963119072708,
      "eval_loss": 1.090325117111206,
      "eval_runtime": 6.446,
      "eval_samples_per_second": 62.054,
      "eval_steps_per_second": 15.513,
      "step": 100
    },
    {
      "epoch": 0.42571127502634354,
      "grad_norm": 3.412214994430542,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.2593,
      "step": 101
    },
    {
      "epoch": 0.42992623814541625,
      "grad_norm": 3.409604072570801,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.3612,
      "step": 102
    },
    {
      "epoch": 0.43414120126448896,
      "grad_norm": 3.3964297771453857,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.3978,
      "step": 103
    },
    {
      "epoch": 0.4383561643835616,
      "grad_norm": 3.2675745487213135,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.528,
      "step": 104
    },
    {
      "epoch": 0.44257112750263433,
      "grad_norm": 2.9116034507751465,
      "learning_rate": 5e-05,
      "loss": 1.2795,
      "step": 105
    },
    {
      "epoch": 0.44678609062170704,
      "grad_norm": 2.8685240745544434,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2282,
      "step": 106
    },
    {
      "epoch": 0.45100105374077976,
      "grad_norm": 3.022886037826538,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.3268,
      "step": 107
    },
    {
      "epoch": 0.45521601685985247,
      "grad_norm": 3.373258352279663,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.2447,
      "step": 108
    },
    {
      "epoch": 0.4594309799789252,
      "grad_norm": 3.2787013053894043,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.2876,
      "step": 109
    },
    {
      "epoch": 0.4636459430979979,
      "grad_norm": 2.9773242473602295,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.1625,
      "step": 110
    },
    {
      "epoch": 0.4678609062170706,
      "grad_norm": 3.4932281970977783,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.4073,
      "step": 111
    },
    {
      "epoch": 0.4720758693361433,
      "grad_norm": 2.868907928466797,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.1351,
      "step": 112
    },
    {
      "epoch": 0.47629083245521603,
      "grad_norm": 3.098670721054077,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.3266,
      "step": 113
    },
    {
      "epoch": 0.48050579557428874,
      "grad_norm": 2.9290401935577393,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.0518,
      "step": 114
    },
    {
      "epoch": 0.48472075869336145,
      "grad_norm": 2.9318532943725586,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.1591,
      "step": 115
    },
    {
      "epoch": 0.48893572181243417,
      "grad_norm": 3.1286261081695557,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.2829,
      "step": 116
    },
    {
      "epoch": 0.4931506849315068,
      "grad_norm": 2.7056784629821777,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.9676,
      "step": 117
    },
    {
      "epoch": 0.49736564805057953,
      "grad_norm": 3.082429885864258,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2765,
      "step": 118
    },
    {
      "epoch": 0.5015806111696522,
      "grad_norm": 2.8888089656829834,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2716,
      "step": 119
    },
    {
      "epoch": 0.505795574288725,
      "grad_norm": 2.764641523361206,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.9778,
      "step": 120
    },
    {
      "epoch": 0.5100105374077977,
      "grad_norm": 2.8492064476013184,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.0074,
      "step": 121
    },
    {
      "epoch": 0.5142255005268704,
      "grad_norm": 2.8319928646087646,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2063,
      "step": 122
    },
    {
      "epoch": 0.5184404636459431,
      "grad_norm": 3.0168888568878174,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.174,
      "step": 123
    },
    {
      "epoch": 0.5226554267650158,
      "grad_norm": 3.1107656955718994,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.2505,
      "step": 124
    },
    {
      "epoch": 0.5268703898840885,
      "grad_norm": 2.8055074214935303,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.0315,
      "step": 125
    },
    {
      "epoch": 0.5310853530031612,
      "grad_norm": 3.23095703125,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.9506,
      "step": 126
    },
    {
      "epoch": 0.5353003161222339,
      "grad_norm": 2.8599140644073486,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.0006,
      "step": 127
    },
    {
      "epoch": 0.5395152792413066,
      "grad_norm": 3.1262266635894775,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.2086,
      "step": 128
    },
    {
      "epoch": 0.5437302423603794,
      "grad_norm": 3.4063544273376465,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.0033,
      "step": 129
    },
    {
      "epoch": 0.547945205479452,
      "grad_norm": 3.177546262741089,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2051,
      "step": 130
    },
    {
      "epoch": 0.5521601685985248,
      "grad_norm": 2.8110849857330322,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.9197,
      "step": 131
    },
    {
      "epoch": 0.5563751317175974,
      "grad_norm": 3.1420490741729736,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.8664,
      "step": 132
    },
    {
      "epoch": 0.5605900948366702,
      "grad_norm": 3.2633495330810547,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.0824,
      "step": 133
    },
    {
      "epoch": 0.5648050579557429,
      "grad_norm": 3.8799312114715576,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.1485,
      "step": 134
    },
    {
      "epoch": 0.5690200210748156,
      "grad_norm": 3.4397261142730713,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.026,
      "step": 135
    },
    {
      "epoch": 0.5732349841938883,
      "grad_norm": 3.4484407901763916,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.0107,
      "step": 136
    },
    {
      "epoch": 0.5774499473129611,
      "grad_norm": 3.4492568969726562,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.106,
      "step": 137
    },
    {
      "epoch": 0.5816649104320337,
      "grad_norm": 3.4244260787963867,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.8723,
      "step": 138
    },
    {
      "epoch": 0.5858798735511064,
      "grad_norm": 2.6440515518188477,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.6067,
      "step": 139
    },
    {
      "epoch": 0.5900948366701791,
      "grad_norm": 3.1771254539489746,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.7691,
      "step": 140
    },
    {
      "epoch": 0.5943097997892518,
      "grad_norm": 2.9901037216186523,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.7327,
      "step": 141
    },
    {
      "epoch": 0.5985247629083246,
      "grad_norm": 3.6130969524383545,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.0231,
      "step": 142
    },
    {
      "epoch": 0.6027397260273972,
      "grad_norm": 3.418724298477173,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.8567,
      "step": 143
    },
    {
      "epoch": 0.60695468914647,
      "grad_norm": 3.110846519470215,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.7971,
      "step": 144
    },
    {
      "epoch": 0.6111696522655427,
      "grad_norm": 3.287153720855713,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.6426,
      "step": 145
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 3.4254863262176514,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.7087,
      "step": 146
    },
    {
      "epoch": 0.6195995785036881,
      "grad_norm": 4.266956329345703,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.9496,
      "step": 147
    },
    {
      "epoch": 0.6238145416227608,
      "grad_norm": 3.553666591644287,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.8529,
      "step": 148
    },
    {
      "epoch": 0.6280295047418335,
      "grad_norm": 3.952604055404663,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.7811,
      "step": 149
    },
    {
      "epoch": 0.6322444678609063,
      "grad_norm": 5.144307613372803,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.8062,
      "step": 150
    },
    {
      "epoch": 0.6322444678609063,
      "eval_loss": 0.9287216663360596,
      "eval_runtime": 6.4552,
      "eval_samples_per_second": 61.965,
      "eval_steps_per_second": 15.491,
      "step": 150
    },
    {
      "epoch": 0.6364594309799789,
      "grad_norm": 2.7988481521606445,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.1912,
      "step": 151
    },
    {
      "epoch": 0.6406743940990516,
      "grad_norm": 3.2067959308624268,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.1908,
      "step": 152
    },
    {
      "epoch": 0.6448893572181243,
      "grad_norm": 3.812892198562622,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.2148,
      "step": 153
    },
    {
      "epoch": 0.649104320337197,
      "grad_norm": 3.4481053352355957,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.2177,
      "step": 154
    },
    {
      "epoch": 0.6533192834562698,
      "grad_norm": 3.6395459175109863,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.1517,
      "step": 155
    },
    {
      "epoch": 0.6575342465753424,
      "grad_norm": 3.1742465496063232,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.0774,
      "step": 156
    },
    {
      "epoch": 0.6617492096944152,
      "grad_norm": 3.301126718521118,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.0928,
      "step": 157
    },
    {
      "epoch": 0.6659641728134879,
      "grad_norm": 2.7524447441101074,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.092,
      "step": 158
    },
    {
      "epoch": 0.6701791359325606,
      "grad_norm": 3.0300285816192627,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.0336,
      "step": 159
    },
    {
      "epoch": 0.6743940990516333,
      "grad_norm": 3.1219213008880615,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.2394,
      "step": 160
    },
    {
      "epoch": 0.678609062170706,
      "grad_norm": 3.193305492401123,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.249,
      "step": 161
    },
    {
      "epoch": 0.6828240252897787,
      "grad_norm": 2.8801887035369873,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.0987,
      "step": 162
    },
    {
      "epoch": 0.6870389884088515,
      "grad_norm": 2.6667888164520264,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.9403,
      "step": 163
    },
    {
      "epoch": 0.6912539515279241,
      "grad_norm": 3.5246291160583496,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.999,
      "step": 164
    },
    {
      "epoch": 0.6954689146469969,
      "grad_norm": 3.30068039894104,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.2046,
      "step": 165
    },
    {
      "epoch": 0.6996838777660696,
      "grad_norm": 2.9631736278533936,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.0144,
      "step": 166
    },
    {
      "epoch": 0.7038988408851422,
      "grad_norm": 3.124361753463745,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.1043,
      "step": 167
    },
    {
      "epoch": 0.708113804004215,
      "grad_norm": 3.4386749267578125,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.0658,
      "step": 168
    },
    {
      "epoch": 0.7123287671232876,
      "grad_norm": 3.0321691036224365,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.9244,
      "step": 169
    },
    {
      "epoch": 0.7165437302423604,
      "grad_norm": 3.0845730304718018,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.9411,
      "step": 170
    },
    {
      "epoch": 0.7207586933614331,
      "grad_norm": 3.4167368412017822,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.8987,
      "step": 171
    },
    {
      "epoch": 0.7249736564805058,
      "grad_norm": 3.1037421226501465,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.8812,
      "step": 172
    },
    {
      "epoch": 0.7291886195995785,
      "grad_norm": 3.3272879123687744,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.8962,
      "step": 173
    },
    {
      "epoch": 0.7334035827186512,
      "grad_norm": 2.8732645511627197,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.0336,
      "step": 174
    },
    {
      "epoch": 0.7376185458377239,
      "grad_norm": 2.975170612335205,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.0367,
      "step": 175
    },
    {
      "epoch": 0.7418335089567967,
      "grad_norm": 2.992979049682617,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.9468,
      "step": 176
    },
    {
      "epoch": 0.7460484720758693,
      "grad_norm": 3.0731427669525146,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.9874,
      "step": 177
    },
    {
      "epoch": 0.7502634351949421,
      "grad_norm": 3.1039509773254395,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.9825,
      "step": 178
    },
    {
      "epoch": 0.7544783983140148,
      "grad_norm": 3.490957260131836,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.1308,
      "step": 179
    },
    {
      "epoch": 0.7586933614330874,
      "grad_norm": 3.0487220287323,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.749,
      "step": 180
    },
    {
      "epoch": 0.7629083245521602,
      "grad_norm": 3.4461748600006104,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.0737,
      "step": 181
    },
    {
      "epoch": 0.7671232876712328,
      "grad_norm": 3.135690927505493,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.9565,
      "step": 182
    },
    {
      "epoch": 0.7713382507903056,
      "grad_norm": 3.355886459350586,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.9347,
      "step": 183
    },
    {
      "epoch": 0.7755532139093783,
      "grad_norm": 3.083711862564087,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.925,
      "step": 184
    },
    {
      "epoch": 0.779768177028451,
      "grad_norm": 3.7594337463378906,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.1239,
      "step": 185
    },
    {
      "epoch": 0.7839831401475237,
      "grad_norm": 3.505735158920288,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.9493,
      "step": 186
    },
    {
      "epoch": 0.7881981032665965,
      "grad_norm": 3.5608115196228027,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.0724,
      "step": 187
    },
    {
      "epoch": 0.7924130663856691,
      "grad_norm": 3.57678484916687,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.7925,
      "step": 188
    },
    {
      "epoch": 0.7966280295047419,
      "grad_norm": 3.389411687850952,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.8195,
      "step": 189
    },
    {
      "epoch": 0.8008429926238145,
      "grad_norm": 3.5880331993103027,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.8296,
      "step": 190
    },
    {
      "epoch": 0.8050579557428873,
      "grad_norm": 3.352307081222534,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.7803,
      "step": 191
    },
    {
      "epoch": 0.80927291886196,
      "grad_norm": 3.363117218017578,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.8002,
      "step": 192
    },
    {
      "epoch": 0.8134878819810326,
      "grad_norm": 3.484776496887207,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.8929,
      "step": 193
    },
    {
      "epoch": 0.8177028451001054,
      "grad_norm": 3.153294801712036,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.8148,
      "step": 194
    },
    {
      "epoch": 0.821917808219178,
      "grad_norm": 4.189630031585693,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.8427,
      "step": 195
    },
    {
      "epoch": 0.8261327713382508,
      "grad_norm": 3.6086337566375732,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.5954,
      "step": 196
    },
    {
      "epoch": 0.8303477344573235,
      "grad_norm": 4.1201348304748535,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.7051,
      "step": 197
    },
    {
      "epoch": 0.8345626975763962,
      "grad_norm": 4.5860700607299805,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.706,
      "step": 198
    },
    {
      "epoch": 0.8387776606954689,
      "grad_norm": 5.123166561126709,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.0983,
      "step": 199
    },
    {
      "epoch": 0.8429926238145417,
      "grad_norm": 4.927257061004639,
      "learning_rate": 0.0,
      "loss": 0.8509,
      "step": 200
    },
    {
      "epoch": 0.8429926238145417,
      "eval_loss": 0.8850148916244507,
      "eval_runtime": 6.4599,
      "eval_samples_per_second": 61.92,
      "eval_steps_per_second": 15.48,
      "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": 1.3317552734208e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}