{
  "best_metric": 1.3921726942062378,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 1.019108280254777,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005095541401273885,
      "grad_norm": 1.6566962003707886,
      "learning_rate": 1e-05,
      "loss": 1.7303,
      "step": 1
    },
    {
      "epoch": 0.005095541401273885,
      "eval_loss": 3.259193181991577,
      "eval_runtime": 24.2848,
      "eval_samples_per_second": 13.63,
      "eval_steps_per_second": 3.418,
      "step": 1
    },
    {
      "epoch": 0.01019108280254777,
      "grad_norm": 1.5152714252471924,
      "learning_rate": 2e-05,
      "loss": 1.9211,
      "step": 2
    },
    {
      "epoch": 0.015286624203821656,
      "grad_norm": 1.5809290409088135,
      "learning_rate": 3e-05,
      "loss": 1.8869,
      "step": 3
    },
    {
      "epoch": 0.02038216560509554,
      "grad_norm": 1.5486772060394287,
      "learning_rate": 4e-05,
      "loss": 2.0196,
      "step": 4
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 1.5725350379943848,
      "learning_rate": 5e-05,
      "loss": 2.113,
      "step": 5
    },
    {
      "epoch": 0.030573248407643312,
      "grad_norm": 1.6562306880950928,
      "learning_rate": 6e-05,
      "loss": 2.0343,
      "step": 6
    },
    {
      "epoch": 0.035668789808917196,
      "grad_norm": 1.7832552194595337,
      "learning_rate": 7e-05,
      "loss": 2.1313,
      "step": 7
    },
    {
      "epoch": 0.04076433121019108,
      "grad_norm": 1.8613115549087524,
      "learning_rate": 8e-05,
      "loss": 2.0906,
      "step": 8
    },
    {
      "epoch": 0.045859872611464965,
      "grad_norm": 1.6480772495269775,
      "learning_rate": 9e-05,
      "loss": 1.9333,
      "step": 9
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 2.0509190559387207,
      "learning_rate": 0.0001,
      "loss": 1.8131,
      "step": 10
    },
    {
      "epoch": 0.05605095541401274,
      "grad_norm": 1.7859784364700317,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.6242,
      "step": 11
    },
    {
      "epoch": 0.061146496815286625,
      "grad_norm": 2.2067337036132812,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.5882,
      "step": 12
    },
    {
      "epoch": 0.06624203821656051,
      "grad_norm": 1.9581551551818848,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.5844,
      "step": 13
    },
    {
      "epoch": 0.07133757961783439,
      "grad_norm": 1.641848087310791,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.5202,
      "step": 14
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 1.5255184173583984,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.557,
      "step": 15
    },
    {
      "epoch": 0.08152866242038216,
      "grad_norm": 1.3865028619766235,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.5331,
      "step": 16
    },
    {
      "epoch": 0.08662420382165605,
      "grad_norm": 1.3835519552230835,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.5851,
      "step": 17
    },
    {
      "epoch": 0.09171974522292993,
      "grad_norm": 1.298506259918213,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.519,
      "step": 18
    },
    {
      "epoch": 0.09681528662420383,
      "grad_norm": 1.446386694908142,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.5427,
      "step": 19
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 1.3654260635375977,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.584,
      "step": 20
    },
    {
      "epoch": 0.1070063694267516,
      "grad_norm": 1.3002036809921265,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4703,
      "step": 21
    },
    {
      "epoch": 0.11210191082802548,
      "grad_norm": 1.2605143785476685,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.4174,
      "step": 22
    },
    {
      "epoch": 0.11719745222929936,
      "grad_norm": 1.3151603937149048,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.4381,
      "step": 23
    },
    {
      "epoch": 0.12229299363057325,
      "grad_norm": 1.4790843725204468,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.4919,
      "step": 24
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 1.4765851497650146,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.5851,
      "step": 25
    },
    {
      "epoch": 0.13248407643312102,
      "grad_norm": 1.58745276927948,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.5138,
      "step": 26
    },
    {
      "epoch": 0.1375796178343949,
      "grad_norm": 1.4464747905731201,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4522,
      "step": 27
    },
    {
      "epoch": 0.14267515923566879,
      "grad_norm": 1.5700476169586182,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4344,
      "step": 28
    },
    {
      "epoch": 0.14777070063694267,
      "grad_norm": 1.5950138568878174,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.443,
      "step": 29
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 1.480726718902588,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3856,
      "step": 30
    },
    {
      "epoch": 0.15796178343949044,
      "grad_norm": 1.416853666305542,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.4832,
      "step": 31
    },
    {
      "epoch": 0.16305732484076432,
      "grad_norm": 1.4361233711242676,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.5198,
      "step": 32
    },
    {
      "epoch": 0.1681528662420382,
      "grad_norm": 1.523002028465271,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.5438,
      "step": 33
    },
    {
      "epoch": 0.1732484076433121,
      "grad_norm": 1.562009572982788,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.6233,
      "step": 34
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 1.447596788406372,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.459,
      "step": 35
    },
    {
      "epoch": 0.18343949044585986,
      "grad_norm": 1.5640453100204468,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.5615,
      "step": 36
    },
    {
      "epoch": 0.18853503184713377,
      "grad_norm": 1.4710253477096558,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.4877,
      "step": 37
    },
    {
      "epoch": 0.19363057324840766,
      "grad_norm": 1.7064555883407593,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.5276,
      "step": 38
    },
    {
      "epoch": 0.19872611464968154,
      "grad_norm": 1.5495549440383911,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.5579,
      "step": 39
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 1.6626213788986206,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.5314,
      "step": 40
    },
    {
      "epoch": 0.2089171974522293,
      "grad_norm": 1.662790298461914,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.4474,
      "step": 41
    },
    {
      "epoch": 0.2140127388535032,
      "grad_norm": 1.5907748937606812,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.5652,
      "step": 42
    },
    {
      "epoch": 0.21910828025477708,
      "grad_norm": 2.517739772796631,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.5888,
      "step": 43
    },
    {
      "epoch": 0.22420382165605096,
      "grad_norm": 1.8781366348266602,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.5287,
      "step": 44
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 1.6947721242904663,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.557,
      "step": 45
    },
    {
      "epoch": 0.23439490445859873,
      "grad_norm": 1.8926512002944946,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.4796,
      "step": 46
    },
    {
      "epoch": 0.2394904458598726,
      "grad_norm": 1.774320363998413,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.5689,
      "step": 47
    },
    {
      "epoch": 0.2445859872611465,
      "grad_norm": 2.6031904220581055,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.5082,
      "step": 48
    },
    {
      "epoch": 0.24968152866242038,
      "grad_norm": 4.309216499328613,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.5921,
      "step": 49
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 2.6246705055236816,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.6926,
      "step": 50
    },
    {
      "epoch": 0.25477707006369427,
      "eval_loss": 1.7766997814178467,
      "eval_runtime": 24.8214,
      "eval_samples_per_second": 13.335,
      "eval_steps_per_second": 3.344,
      "step": 50
    },
    {
      "epoch": 0.25987261146496815,
      "grad_norm": 2.453874111175537,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.654,
      "step": 51
    },
    {
      "epoch": 0.26496815286624203,
      "grad_norm": 1.8163388967514038,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4758,
      "step": 52
    },
    {
      "epoch": 0.2700636942675159,
      "grad_norm": 1.19365394115448,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.3972,
      "step": 53
    },
    {
      "epoch": 0.2751592356687898,
      "grad_norm": 1.4323458671569824,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.5399,
      "step": 54
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 1.2911094427108765,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.3959,
      "step": 55
    },
    {
      "epoch": 0.28535031847133757,
      "grad_norm": 1.2983102798461914,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4503,
      "step": 56
    },
    {
      "epoch": 0.29044585987261146,
      "grad_norm": 1.282852292060852,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.4422,
      "step": 57
    },
    {
      "epoch": 0.29554140127388534,
      "grad_norm": 1.1919018030166626,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.4764,
      "step": 58
    },
    {
      "epoch": 0.3006369426751592,
      "grad_norm": 1.0220428705215454,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.4482,
      "step": 59
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": 1.004325032234192,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3913,
      "step": 60
    },
    {
      "epoch": 0.310828025477707,
      "grad_norm": 1.0730646848678589,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.5228,
      "step": 61
    },
    {
      "epoch": 0.3159235668789809,
      "grad_norm": 1.0348135232925415,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.3824,
      "step": 62
    },
    {
      "epoch": 0.32101910828025476,
      "grad_norm": 1.055734634399414,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.2925,
      "step": 63
    },
    {
      "epoch": 0.32611464968152865,
      "grad_norm": 1.2250598669052124,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.4025,
      "step": 64
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": 1.2325644493103027,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.4504,
      "step": 65
    },
    {
      "epoch": 0.3363057324840764,
      "grad_norm": 1.2184557914733887,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.4286,
      "step": 66
    },
    {
      "epoch": 0.3414012738853503,
      "grad_norm": 1.387960433959961,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.5632,
      "step": 67
    },
    {
      "epoch": 0.3464968152866242,
      "grad_norm": 1.1769790649414062,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.3765,
      "step": 68
    },
    {
      "epoch": 0.35159235668789807,
      "grad_norm": 1.2014427185058594,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4774,
      "step": 69
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": 1.1256524324417114,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.4444,
      "step": 70
    },
    {
      "epoch": 0.36178343949044584,
      "grad_norm": 1.1236120462417603,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.4237,
      "step": 71
    },
    {
      "epoch": 0.3668789808917197,
      "grad_norm": 1.0599284172058105,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3492,
      "step": 72
    },
    {
      "epoch": 0.3719745222929936,
      "grad_norm": 1.1497225761413574,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3502,
      "step": 73
    },
    {
      "epoch": 0.37707006369426754,
      "grad_norm": 1.144227385520935,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.4483,
      "step": 74
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": 1.3062736988067627,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.4855,
      "step": 75
    },
    {
      "epoch": 0.3872611464968153,
      "grad_norm": 1.2697646617889404,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.5133,
      "step": 76
    },
    {
      "epoch": 0.3923566878980892,
      "grad_norm": 1.261718988418579,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.405,
      "step": 77
    },
    {
      "epoch": 0.3974522292993631,
      "grad_norm": 1.264968752861023,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.5037,
      "step": 78
    },
    {
      "epoch": 0.40254777070063696,
      "grad_norm": 1.289087176322937,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.465,
      "step": 79
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": 1.2691148519515991,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.4851,
      "step": 80
    },
    {
      "epoch": 0.41273885350318473,
      "grad_norm": 1.3029874563217163,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.4492,
      "step": 81
    },
    {
      "epoch": 0.4178343949044586,
      "grad_norm": 1.2333258390426636,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.4576,
      "step": 82
    },
    {
      "epoch": 0.4229299363057325,
      "grad_norm": 1.300815224647522,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.4976,
      "step": 83
    },
    {
      "epoch": 0.4280254777070064,
      "grad_norm": 1.2724026441574097,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.4167,
      "step": 84
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": 1.3012876510620117,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.4261,
      "step": 85
    },
    {
      "epoch": 0.43821656050955415,
      "grad_norm": 1.2972036600112915,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3443,
      "step": 86
    },
    {
      "epoch": 0.44331210191082804,
      "grad_norm": 1.3304251432418823,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.4889,
      "step": 87
    },
    {
      "epoch": 0.4484076433121019,
      "grad_norm": 1.4206112623214722,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.4442,
      "step": 88
    },
    {
      "epoch": 0.4535031847133758,
      "grad_norm": 1.456225872039795,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.493,
      "step": 89
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": 1.5380885601043701,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.4774,
      "step": 90
    },
    {
      "epoch": 0.4636942675159236,
      "grad_norm": 1.54289710521698,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.6233,
      "step": 91
    },
    {
      "epoch": 0.46878980891719746,
      "grad_norm": 1.3833653926849365,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.4643,
      "step": 92
    },
    {
      "epoch": 0.47388535031847134,
      "grad_norm": 1.5661531686782837,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.5152,
      "step": 93
    },
    {
      "epoch": 0.4789808917197452,
      "grad_norm": 1.4505186080932617,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.4569,
      "step": 94
    },
    {
      "epoch": 0.4840764331210191,
      "grad_norm": 1.4179692268371582,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.5164,
      "step": 95
    },
    {
      "epoch": 0.489171974522293,
      "grad_norm": 1.643911361694336,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.4666,
      "step": 96
    },
    {
      "epoch": 0.4942675159235669,
      "grad_norm": 1.7628129720687866,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.5073,
      "step": 97
    },
    {
      "epoch": 0.49936305732484076,
      "grad_norm": 1.8917542695999146,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.4153,
      "step": 98
    },
    {
      "epoch": 0.5044585987261146,
      "grad_norm": 2.2565598487854004,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.6597,
      "step": 99
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 2.1938929557800293,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.6038,
      "step": 100
    },
    {
      "epoch": 0.5095541401273885,
      "eval_loss": 1.596757173538208,
      "eval_runtime": 24.8427,
      "eval_samples_per_second": 13.324,
      "eval_steps_per_second": 3.341,
      "step": 100
    },
    {
      "epoch": 0.5146496815286624,
      "grad_norm": 1.9329887628555298,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.5412,
      "step": 101
    },
    {
      "epoch": 0.5197452229299363,
      "grad_norm": 1.6277050971984863,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.5393,
      "step": 102
    },
    {
      "epoch": 0.5248407643312102,
      "grad_norm": 1.2849308252334595,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.5125,
      "step": 103
    },
    {
      "epoch": 0.5299363057324841,
      "grad_norm": 1.022396445274353,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.376,
      "step": 104
    },
    {
      "epoch": 0.535031847133758,
      "grad_norm": 0.9251585006713867,
      "learning_rate": 5e-05,
      "loss": 1.3128,
      "step": 105
    },
    {
      "epoch": 0.5401273885350318,
      "grad_norm": 0.9299936294555664,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.2795,
      "step": 106
    },
    {
      "epoch": 0.5452229299363057,
      "grad_norm": 0.995810329914093,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.377,
      "step": 107
    },
    {
      "epoch": 0.5503184713375796,
      "grad_norm": 1.081729531288147,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3628,
      "step": 108
    },
    {
      "epoch": 0.5554140127388535,
      "grad_norm": 1.1986666917800903,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4116,
      "step": 109
    },
    {
      "epoch": 0.5605095541401274,
      "grad_norm": 1.125977635383606,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.4147,
      "step": 110
    },
    {
      "epoch": 0.5656050955414013,
      "grad_norm": 1.1973488330841064,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.4281,
      "step": 111
    },
    {
      "epoch": 0.5707006369426751,
      "grad_norm": 1.1541385650634766,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.4171,
      "step": 112
    },
    {
      "epoch": 0.575796178343949,
      "grad_norm": 1.098771095275879,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.3738,
      "step": 113
    },
    {
      "epoch": 0.5808917197452229,
      "grad_norm": 1.105370283126831,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.4065,
      "step": 114
    },
    {
      "epoch": 0.5859872611464968,
      "grad_norm": 1.0938535928726196,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.4606,
      "step": 115
    },
    {
      "epoch": 0.5910828025477707,
      "grad_norm": 1.1129817962646484,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.408,
      "step": 116
    },
    {
      "epoch": 0.5961783439490446,
      "grad_norm": 1.0800514221191406,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.4246,
      "step": 117
    },
    {
      "epoch": 0.6012738853503184,
      "grad_norm": 1.0996190309524536,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.4921,
      "step": 118
    },
    {
      "epoch": 0.6063694267515923,
      "grad_norm": 1.0310497283935547,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.3317,
      "step": 119
    },
    {
      "epoch": 0.6114649681528662,
      "grad_norm": 1.1108511686325073,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.4415,
      "step": 120
    },
    {
      "epoch": 0.6165605095541401,
      "grad_norm": 1.1383891105651855,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.4281,
      "step": 121
    },
    {
      "epoch": 0.621656050955414,
      "grad_norm": 1.1201890707015991,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.511,
      "step": 122
    },
    {
      "epoch": 0.6267515923566879,
      "grad_norm": 1.2212876081466675,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.4926,
      "step": 123
    },
    {
      "epoch": 0.6318471337579618,
      "grad_norm": 1.1589946746826172,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.4489,
      "step": 124
    },
    {
      "epoch": 0.6369426751592356,
      "grad_norm": 1.1335136890411377,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3921,
      "step": 125
    },
    {
      "epoch": 0.6420382165605095,
      "grad_norm": 1.1459298133850098,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.4722,
      "step": 126
    },
    {
      "epoch": 0.6471337579617834,
      "grad_norm": 1.1157445907592773,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.4465,
      "step": 127
    },
    {
      "epoch": 0.6522292993630573,
      "grad_norm": 1.212799072265625,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.4464,
      "step": 128
    },
    {
      "epoch": 0.6573248407643312,
      "grad_norm": 1.258693814277649,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.467,
      "step": 129
    },
    {
      "epoch": 0.6624203821656051,
      "grad_norm": 1.2206029891967773,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.4743,
      "step": 130
    },
    {
      "epoch": 0.6675159235668789,
      "grad_norm": 1.2751787900924683,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.3155,
      "step": 131
    },
    {
      "epoch": 0.6726114649681528,
      "grad_norm": 1.215261459350586,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.4156,
      "step": 132
    },
    {
      "epoch": 0.6777070063694267,
      "grad_norm": 1.278517723083496,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.3962,
      "step": 133
    },
    {
      "epoch": 0.6828025477707006,
      "grad_norm": 1.306698203086853,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.4706,
      "step": 134
    },
    {
      "epoch": 0.6878980891719745,
      "grad_norm": 1.331336259841919,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.4098,
      "step": 135
    },
    {
      "epoch": 0.6929936305732484,
      "grad_norm": 1.3080174922943115,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.4591,
      "step": 136
    },
    {
      "epoch": 0.6980891719745222,
      "grad_norm": 1.3106648921966553,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.4656,
      "step": 137
    },
    {
      "epoch": 0.7031847133757961,
      "grad_norm": 1.4060168266296387,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.5567,
      "step": 138
    },
    {
      "epoch": 0.70828025477707,
      "grad_norm": 1.424262285232544,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.5551,
      "step": 139
    },
    {
      "epoch": 0.7133757961783439,
      "grad_norm": 1.417178750038147,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.4876,
      "step": 140
    },
    {
      "epoch": 0.7184713375796178,
      "grad_norm": 1.3910839557647705,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.4211,
      "step": 141
    },
    {
      "epoch": 0.7235668789808917,
      "grad_norm": 1.3794727325439453,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.5108,
      "step": 142
    },
    {
      "epoch": 0.7286624203821656,
      "grad_norm": 1.5028527975082397,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4562,
      "step": 143
    },
    {
      "epoch": 0.7337579617834394,
      "grad_norm": 1.4295216798782349,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.5037,
      "step": 144
    },
    {
      "epoch": 0.7388535031847133,
      "grad_norm": 1.4502906799316406,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.4923,
      "step": 145
    },
    {
      "epoch": 0.7439490445859872,
      "grad_norm": 1.481606364250183,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.4787,
      "step": 146
    },
    {
      "epoch": 0.7490445859872611,
      "grad_norm": 1.7646911144256592,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.4824,
      "step": 147
    },
    {
      "epoch": 0.7541401273885351,
      "grad_norm": 1.612247109413147,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.4433,
      "step": 148
    },
    {
      "epoch": 0.759235668789809,
      "grad_norm": 1.5676825046539307,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.5238,
      "step": 149
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": 1.5387394428253174,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.5192,
      "step": 150
    },
    {
      "epoch": 0.7643312101910829,
      "eval_loss": 1.4677605628967285,
      "eval_runtime": 24.8305,
      "eval_samples_per_second": 13.33,
      "eval_steps_per_second": 3.343,
      "step": 150
    },
    {
      "epoch": 0.7694267515923567,
      "grad_norm": 1.4125545024871826,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.5294,
      "step": 151
    },
    {
      "epoch": 0.7745222929936306,
      "grad_norm": 1.3933442831039429,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.5636,
      "step": 152
    },
    {
      "epoch": 0.7796178343949045,
      "grad_norm": 1.3459492921829224,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.5051,
      "step": 153
    },
    {
      "epoch": 0.7847133757961784,
      "grad_norm": 1.216903567314148,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.4011,
      "step": 154
    },
    {
      "epoch": 0.7898089171974523,
      "grad_norm": 1.1598666906356812,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.5055,
      "step": 155
    },
    {
      "epoch": 0.7949044585987262,
      "grad_norm": 1.1041417121887207,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.506,
      "step": 156
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0673881769180298,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.5288,
      "step": 157
    },
    {
      "epoch": 0.8050955414012739,
      "grad_norm": 1.0575605630874634,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.4521,
      "step": 158
    },
    {
      "epoch": 0.8101910828025478,
      "grad_norm": 1.0240464210510254,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.4156,
      "step": 159
    },
    {
      "epoch": 0.8152866242038217,
      "grad_norm": 1.0079967975616455,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.4418,
      "step": 160
    },
    {
      "epoch": 0.8203821656050956,
      "grad_norm": 0.9954540133476257,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.4436,
      "step": 161
    },
    {
      "epoch": 0.8254777070063695,
      "grad_norm": 0.9684792160987854,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4063,
      "step": 162
    },
    {
      "epoch": 0.8305732484076433,
      "grad_norm": 0.971145749092102,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.3919,
      "step": 163
    },
    {
      "epoch": 0.8356687898089172,
      "grad_norm": 1.0355074405670166,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4628,
      "step": 164
    },
    {
      "epoch": 0.8407643312101911,
      "grad_norm": 1.0215669870376587,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.39,
      "step": 165
    },
    {
      "epoch": 0.845859872611465,
      "grad_norm": 1.0762925148010254,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.4628,
      "step": 166
    },
    {
      "epoch": 0.8509554140127389,
      "grad_norm": 0.9901561141014099,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.3532,
      "step": 167
    },
    {
      "epoch": 0.8560509554140128,
      "grad_norm": 1.015710473060608,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.4046,
      "step": 168
    },
    {
      "epoch": 0.8611464968152867,
      "grad_norm": 1.0279351472854614,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.3887,
      "step": 169
    },
    {
      "epoch": 0.8662420382165605,
      "grad_norm": 1.0289154052734375,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.3454,
      "step": 170
    },
    {
      "epoch": 0.8713375796178344,
      "grad_norm": 1.0734381675720215,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.4035,
      "step": 171
    },
    {
      "epoch": 0.8764331210191083,
      "grad_norm": 1.102924108505249,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.4945,
      "step": 172
    },
    {
      "epoch": 0.8815286624203822,
      "grad_norm": 1.0636721849441528,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.5572,
      "step": 173
    },
    {
      "epoch": 0.8866242038216561,
      "grad_norm": 1.1362146139144897,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.4327,
      "step": 174
    },
    {
      "epoch": 0.89171974522293,
      "grad_norm": 1.0944325923919678,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.3863,
      "step": 175
    },
    {
      "epoch": 0.8968152866242038,
      "grad_norm": 1.1210203170776367,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.4364,
      "step": 176
    },
    {
      "epoch": 0.9019108280254777,
      "grad_norm": 1.0446661710739136,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.3869,
      "step": 177
    },
    {
      "epoch": 0.9070063694267516,
      "grad_norm": 1.1834402084350586,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.4973,
      "step": 178
    },
    {
      "epoch": 0.9121019108280255,
      "grad_norm": 1.1565152406692505,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.4824,
      "step": 179
    },
    {
      "epoch": 0.9171974522292994,
      "grad_norm": 1.1296406984329224,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.3585,
      "step": 180
    },
    {
      "epoch": 0.9222929936305733,
      "grad_norm": 1.2281200885772705,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.5283,
      "step": 181
    },
    {
      "epoch": 0.9273885350318471,
      "grad_norm": 1.1921871900558472,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.4631,
      "step": 182
    },
    {
      "epoch": 0.932484076433121,
      "grad_norm": 1.2005940675735474,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.4801,
      "step": 183
    },
    {
      "epoch": 0.9375796178343949,
      "grad_norm": 1.2575868368148804,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.4744,
      "step": 184
    },
    {
      "epoch": 0.9426751592356688,
      "grad_norm": 1.1981734037399292,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.4373,
      "step": 185
    },
    {
      "epoch": 0.9477707006369427,
      "grad_norm": 1.2592155933380127,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4686,
      "step": 186
    },
    {
      "epoch": 0.9528662420382166,
      "grad_norm": 1.3583152294158936,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.5464,
      "step": 187
    },
    {
      "epoch": 0.9579617834394905,
      "grad_norm": 1.4326961040496826,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.5275,
      "step": 188
    },
    {
      "epoch": 0.9630573248407643,
      "grad_norm": 1.4089946746826172,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.4755,
      "step": 189
    },
    {
      "epoch": 0.9681528662420382,
      "grad_norm": 1.6128218173980713,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.6237,
      "step": 190
    },
    {
      "epoch": 0.9732484076433121,
      "grad_norm": 1.5017962455749512,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.5238,
      "step": 191
    },
    {
      "epoch": 0.978343949044586,
      "grad_norm": 1.512758493423462,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.5439,
      "step": 192
    },
    {
      "epoch": 0.9834394904458599,
      "grad_norm": 1.700508952140808,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.4934,
      "step": 193
    },
    {
      "epoch": 0.9885350318471338,
      "grad_norm": 1.7387676239013672,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.6,
      "step": 194
    },
    {
      "epoch": 0.9936305732484076,
      "grad_norm": 1.9176628589630127,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.5628,
      "step": 195
    },
    {
      "epoch": 0.9987261146496815,
      "grad_norm": 2.7946009635925293,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.7433,
      "step": 196
    },
    {
      "epoch": 1.0038216560509554,
      "grad_norm": 3.4324147701263428,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.5296,
      "step": 197
    },
    {
      "epoch": 1.0089171974522293,
      "grad_norm": 0.7236244678497314,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.0954,
      "step": 198
    },
    {
      "epoch": 1.0140127388535032,
      "grad_norm": 0.7158063054084778,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.1801,
      "step": 199
    },
    {
      "epoch": 1.019108280254777,
      "grad_norm": 0.7307878136634827,
      "learning_rate": 0.0,
      "loss": 1.1353,
      "step": 200
    },
    {
      "epoch": 1.019108280254777,
      "eval_loss": 1.3921726942062378,
      "eval_runtime": 24.7946,
      "eval_samples_per_second": 13.35,
      "eval_steps_per_second": 3.347,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.01514151551828e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}