{
  "best_metric": 1.2877659797668457,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.1702127659574468,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.000851063829787234,
      "grad_norm": 1.8748613595962524,
      "learning_rate": 1e-05,
      "loss": 1.743,
      "step": 1
    },
    {
      "epoch": 0.000851063829787234,
      "eval_loss": 1.891872525215149,
      "eval_runtime": 158.022,
      "eval_samples_per_second": 12.524,
      "eval_steps_per_second": 3.132,
      "step": 1
    },
    {
      "epoch": 0.001702127659574468,
      "grad_norm": 2.367863893508911,
      "learning_rate": 2e-05,
      "loss": 1.7684,
      "step": 2
    },
    {
      "epoch": 0.002553191489361702,
      "grad_norm": 2.3297040462493896,
      "learning_rate": 3e-05,
      "loss": 1.8837,
      "step": 3
    },
    {
      "epoch": 0.003404255319148936,
      "grad_norm": 1.8379721641540527,
      "learning_rate": 4e-05,
      "loss": 1.8312,
      "step": 4
    },
    {
      "epoch": 0.00425531914893617,
      "grad_norm": 1.386456847190857,
      "learning_rate": 5e-05,
      "loss": 1.7847,
      "step": 5
    },
    {
      "epoch": 0.005106382978723404,
      "grad_norm": 1.1972447633743286,
      "learning_rate": 6e-05,
      "loss": 1.7449,
      "step": 6
    },
    {
      "epoch": 0.005957446808510639,
      "grad_norm": 1.271209955215454,
      "learning_rate": 7e-05,
      "loss": 1.6197,
      "step": 7
    },
    {
      "epoch": 0.006808510638297872,
      "grad_norm": 1.3694218397140503,
      "learning_rate": 8e-05,
      "loss": 1.6423,
      "step": 8
    },
    {
      "epoch": 0.007659574468085106,
      "grad_norm": 1.2521004676818848,
      "learning_rate": 9e-05,
      "loss": 1.554,
      "step": 9
    },
    {
      "epoch": 0.00851063829787234,
      "grad_norm": 1.1916066408157349,
      "learning_rate": 0.0001,
      "loss": 1.5668,
      "step": 10
    },
    {
      "epoch": 0.009361702127659575,
      "grad_norm": 1.3646974563598633,
      "learning_rate": 9.999316524962345e-05,
      "loss": 1.5826,
      "step": 11
    },
    {
      "epoch": 0.010212765957446808,
      "grad_norm": 1.2432794570922852,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.6031,
      "step": 12
    },
    {
      "epoch": 0.011063829787234043,
      "grad_norm": 1.0825303792953491,
      "learning_rate": 9.993849845741524e-05,
      "loss": 1.4893,
      "step": 13
    },
    {
      "epoch": 0.011914893617021277,
      "grad_norm": 1.0660406351089478,
      "learning_rate": 9.989068136093873e-05,
      "loss": 1.4527,
      "step": 14
    },
    {
      "epoch": 0.01276595744680851,
      "grad_norm": 1.009065866470337,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.3977,
      "step": 15
    },
    {
      "epoch": 0.013617021276595745,
      "grad_norm": 1.0018229484558105,
      "learning_rate": 9.975414512725057e-05,
      "loss": 1.446,
      "step": 16
    },
    {
      "epoch": 0.01446808510638298,
      "grad_norm": 1.0209919214248657,
      "learning_rate": 9.966546331768191e-05,
      "loss": 1.4836,
      "step": 17
    },
    {
      "epoch": 0.015319148936170212,
      "grad_norm": 0.9393747448921204,
      "learning_rate": 9.956320346634876e-05,
      "loss": 1.3585,
      "step": 18
    },
    {
      "epoch": 0.016170212765957447,
      "grad_norm": 0.9613560438156128,
      "learning_rate": 9.944739353007344e-05,
      "loss": 1.356,
      "step": 19
    },
    {
      "epoch": 0.01702127659574468,
      "grad_norm": 1.0050835609436035,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.3822,
      "step": 20
    },
    {
      "epoch": 0.017872340425531916,
      "grad_norm": 0.9947201013565063,
      "learning_rate": 9.917525374361912e-05,
      "loss": 1.4136,
      "step": 21
    },
    {
      "epoch": 0.01872340425531915,
      "grad_norm": 0.9691237211227417,
      "learning_rate": 9.901899829374047e-05,
      "loss": 1.3146,
      "step": 22
    },
    {
      "epoch": 0.01957446808510638,
      "grad_norm": 0.9850197434425354,
      "learning_rate": 9.884934153917997e-05,
      "loss": 1.469,
      "step": 23
    },
    {
      "epoch": 0.020425531914893616,
      "grad_norm": 1.5736337900161743,
      "learning_rate": 9.86663298624003e-05,
      "loss": 1.5199,
      "step": 24
    },
    {
      "epoch": 0.02127659574468085,
      "grad_norm": 1.0585840940475464,
      "learning_rate": 9.847001329696653e-05,
      "loss": 1.4728,
      "step": 25
    },
    {
      "epoch": 0.022127659574468085,
      "grad_norm": 1.0158225297927856,
      "learning_rate": 9.826044551386744e-05,
      "loss": 1.4406,
      "step": 26
    },
    {
      "epoch": 0.02297872340425532,
      "grad_norm": 0.9874927401542664,
      "learning_rate": 9.803768380684242e-05,
      "loss": 1.4329,
      "step": 27
    },
    {
      "epoch": 0.023829787234042554,
      "grad_norm": 0.9586259126663208,
      "learning_rate": 9.780178907671789e-05,
      "loss": 1.4521,
      "step": 28
    },
    {
      "epoch": 0.02468085106382979,
      "grad_norm": 0.9385019540786743,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.3147,
      "step": 29
    },
    {
      "epoch": 0.02553191489361702,
      "grad_norm": 0.9428925514221191,
      "learning_rate": 9.729086208503174e-05,
      "loss": 1.3566,
      "step": 30
    },
    {
      "epoch": 0.026382978723404255,
      "grad_norm": 0.9508212804794312,
      "learning_rate": 9.701596950580806e-05,
      "loss": 1.3666,
      "step": 31
    },
    {
      "epoch": 0.02723404255319149,
      "grad_norm": 0.9178204536437988,
      "learning_rate": 9.672822322997305e-05,
      "loss": 1.3106,
      "step": 32
    },
    {
      "epoch": 0.028085106382978724,
      "grad_norm": 0.9037249088287354,
      "learning_rate": 9.642770192448536e-05,
      "loss": 1.3662,
      "step": 33
    },
    {
      "epoch": 0.02893617021276596,
      "grad_norm": 0.9893576502799988,
      "learning_rate": 9.611448774886924e-05,
      "loss": 1.3251,
      "step": 34
    },
    {
      "epoch": 0.029787234042553193,
      "grad_norm": 1.0144944190979004,
      "learning_rate": 9.578866633275288e-05,
      "loss": 1.4219,
      "step": 35
    },
    {
      "epoch": 0.030638297872340424,
      "grad_norm": 0.9419549107551575,
      "learning_rate": 9.545032675245813e-05,
      "loss": 1.3484,
      "step": 36
    },
    {
      "epoch": 0.03148936170212766,
      "grad_norm": 0.957611620426178,
      "learning_rate": 9.509956150664796e-05,
      "loss": 1.3355,
      "step": 37
    },
    {
      "epoch": 0.03234042553191489,
      "grad_norm": 0.9596093893051147,
      "learning_rate": 9.473646649103818e-05,
      "loss": 1.3806,
      "step": 38
    },
    {
      "epoch": 0.033191489361702124,
      "grad_norm": 0.9163219928741455,
      "learning_rate": 9.43611409721806e-05,
      "loss": 1.4021,
      "step": 39
    },
    {
      "epoch": 0.03404255319148936,
      "grad_norm": 0.9538283348083496,
      "learning_rate": 9.397368756032445e-05,
      "loss": 1.3568,
      "step": 40
    },
    {
      "epoch": 0.03489361702127659,
      "grad_norm": 1.006136178970337,
      "learning_rate": 9.357421218136386e-05,
      "loss": 1.3723,
      "step": 41
    },
    {
      "epoch": 0.03574468085106383,
      "grad_norm": 0.987679660320282,
      "learning_rate": 9.316282404787871e-05,
      "loss": 1.2594,
      "step": 42
    },
    {
      "epoch": 0.03659574468085106,
      "grad_norm": 0.9679043292999268,
      "learning_rate": 9.273963562927695e-05,
      "loss": 1.1836,
      "step": 43
    },
    {
      "epoch": 0.0374468085106383,
      "grad_norm": 0.8755620121955872,
      "learning_rate": 9.230476262104677e-05,
      "loss": 1.3037,
      "step": 44
    },
    {
      "epoch": 0.03829787234042553,
      "grad_norm": 0.8599547743797302,
      "learning_rate": 9.185832391312644e-05,
      "loss": 1.1144,
      "step": 45
    },
    {
      "epoch": 0.03914893617021276,
      "grad_norm": 0.8587265610694885,
      "learning_rate": 9.140044155740101e-05,
      "loss": 1.1981,
      "step": 46
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8599922060966492,
      "learning_rate": 9.093124073433463e-05,
      "loss": 1.1404,
      "step": 47
    },
    {
      "epoch": 0.04085106382978723,
      "grad_norm": 0.8876784443855286,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.2432,
      "step": 48
    },
    {
      "epoch": 0.04170212765957447,
      "grad_norm": 0.9077394604682922,
      "learning_rate": 8.995939984474624e-05,
      "loss": 1.1024,
      "step": 49
    },
    {
      "epoch": 0.0425531914893617,
      "grad_norm": 0.8578384518623352,
      "learning_rate": 8.945702546981969e-05,
      "loss": 1.2343,
      "step": 50
    },
    {
      "epoch": 0.0425531914893617,
      "eval_loss": 1.340824007987976,
      "eval_runtime": 159.6856,
      "eval_samples_per_second": 12.393,
      "eval_steps_per_second": 3.1,
      "step": 50
    },
    {
      "epoch": 0.04340425531914894,
      "grad_norm": 0.8624850511550903,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.4815,
      "step": 51
    },
    {
      "epoch": 0.04425531914893617,
      "grad_norm": 0.8273656964302063,
      "learning_rate": 8.842005554284296e-05,
      "loss": 1.4696,
      "step": 52
    },
    {
      "epoch": 0.0451063829787234,
      "grad_norm": 0.7026848793029785,
      "learning_rate": 8.788574348801675e-05,
      "loss": 1.4004,
      "step": 53
    },
    {
      "epoch": 0.04595744680851064,
      "grad_norm": 0.7106903791427612,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.382,
      "step": 54
    },
    {
      "epoch": 0.04680851063829787,
      "grad_norm": 0.6895958185195923,
      "learning_rate": 8.678619553365659e-05,
      "loss": 1.4084,
      "step": 55
    },
    {
      "epoch": 0.04765957446808511,
      "grad_norm": 0.7255818247795105,
      "learning_rate": 8.622126023955446e-05,
      "loss": 1.4782,
      "step": 56
    },
    {
      "epoch": 0.04851063829787234,
      "grad_norm": 0.7112653851509094,
      "learning_rate": 8.564642241456986e-05,
      "loss": 1.3669,
      "step": 57
    },
    {
      "epoch": 0.04936170212765958,
      "grad_norm": 0.7303239107131958,
      "learning_rate": 8.506183921362443e-05,
      "loss": 1.3724,
      "step": 58
    },
    {
      "epoch": 0.05021276595744681,
      "grad_norm": 0.7271257638931274,
      "learning_rate": 8.44676704559283e-05,
      "loss": 1.3484,
      "step": 59
    },
    {
      "epoch": 0.05106382978723404,
      "grad_norm": 0.7479179501533508,
      "learning_rate": 8.386407858128706e-05,
      "loss": 1.3721,
      "step": 60
    },
    {
      "epoch": 0.05191489361702128,
      "grad_norm": 0.7274481058120728,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.3921,
      "step": 61
    },
    {
      "epoch": 0.05276595744680851,
      "grad_norm": 0.7331058979034424,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.2515,
      "step": 62
    },
    {
      "epoch": 0.05361702127659575,
      "grad_norm": 0.7577350735664368,
      "learning_rate": 8.199842702516583e-05,
      "loss": 1.3553,
      "step": 63
    },
    {
      "epoch": 0.05446808510638298,
      "grad_norm": 0.7397419214248657,
      "learning_rate": 8.135881792367686e-05,
      "loss": 1.3651,
      "step": 64
    },
    {
      "epoch": 0.05531914893617021,
      "grad_norm": 0.7716683149337769,
      "learning_rate": 8.07106356344834e-05,
      "loss": 1.3785,
      "step": 65
    },
    {
      "epoch": 0.05617021276595745,
      "grad_norm": 0.7625778317451477,
      "learning_rate": 8.005405736415126e-05,
      "loss": 1.2758,
      "step": 66
    },
    {
      "epoch": 0.05702127659574468,
      "grad_norm": 0.7820684909820557,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.3865,
      "step": 67
    },
    {
      "epoch": 0.05787234042553192,
      "grad_norm": 0.7896472811698914,
      "learning_rate": 7.871643313414718e-05,
      "loss": 1.3674,
      "step": 68
    },
    {
      "epoch": 0.05872340425531915,
      "grad_norm": 0.7673555612564087,
      "learning_rate": 7.803575286758364e-05,
      "loss": 1.4017,
      "step": 69
    },
    {
      "epoch": 0.059574468085106386,
      "grad_norm": 0.7671334147453308,
      "learning_rate": 7.734740790612136e-05,
      "loss": 1.3585,
      "step": 70
    },
    {
      "epoch": 0.06042553191489362,
      "grad_norm": 0.7786324620246887,
      "learning_rate": 7.66515864363997e-05,
      "loss": 1.4051,
      "step": 71
    },
    {
      "epoch": 0.06127659574468085,
      "grad_norm": 0.7414199709892273,
      "learning_rate": 7.594847868906076e-05,
      "loss": 1.3497,
      "step": 72
    },
    {
      "epoch": 0.062127659574468086,
      "grad_norm": 0.7384480834007263,
      "learning_rate": 7.52382768867422e-05,
      "loss": 1.3227,
      "step": 73
    },
    {
      "epoch": 0.06297872340425532,
      "grad_norm": 0.7712032794952393,
      "learning_rate": 7.452117519152542e-05,
      "loss": 1.2783,
      "step": 74
    },
    {
      "epoch": 0.06382978723404255,
      "grad_norm": 0.7809745073318481,
      "learning_rate": 7.379736965185368e-05,
      "loss": 1.2586,
      "step": 75
    },
    {
      "epoch": 0.06468085106382979,
      "grad_norm": 0.824797511100769,
      "learning_rate": 7.30670581489344e-05,
      "loss": 1.3505,
      "step": 76
    },
    {
      "epoch": 0.06553191489361702,
      "grad_norm": 0.7680295705795288,
      "learning_rate": 7.233044034264034e-05,
      "loss": 1.293,
      "step": 77
    },
    {
      "epoch": 0.06638297872340425,
      "grad_norm": 0.8489116430282593,
      "learning_rate": 7.158771761692464e-05,
      "loss": 1.3227,
      "step": 78
    },
    {
      "epoch": 0.06723404255319149,
      "grad_norm": 0.8415547609329224,
      "learning_rate": 7.083909302476453e-05,
      "loss": 1.3854,
      "step": 79
    },
    {
      "epoch": 0.06808510638297872,
      "grad_norm": 0.807040810585022,
      "learning_rate": 7.008477123264848e-05,
      "loss": 1.3454,
      "step": 80
    },
    {
      "epoch": 0.06893617021276596,
      "grad_norm": 0.8339987397193909,
      "learning_rate": 6.932495846462261e-05,
      "loss": 1.4252,
      "step": 81
    },
    {
      "epoch": 0.06978723404255319,
      "grad_norm": 0.7896285057067871,
      "learning_rate": 6.855986244591104e-05,
      "loss": 1.2538,
      "step": 82
    },
    {
      "epoch": 0.07063829787234042,
      "grad_norm": 0.8138987421989441,
      "learning_rate": 6.778969234612584e-05,
      "loss": 1.2803,
      "step": 83
    },
    {
      "epoch": 0.07148936170212766,
      "grad_norm": 0.8562771081924438,
      "learning_rate": 6.701465872208216e-05,
      "loss": 1.3412,
      "step": 84
    },
    {
      "epoch": 0.07234042553191489,
      "grad_norm": 0.8194132447242737,
      "learning_rate": 6.623497346023418e-05,
      "loss": 1.3244,
      "step": 85
    },
    {
      "epoch": 0.07319148936170213,
      "grad_norm": 0.8392623066902161,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.3477,
      "step": 86
    },
    {
      "epoch": 0.07404255319148936,
      "grad_norm": 0.8646876215934753,
      "learning_rate": 6.466250186922325e-05,
      "loss": 1.3786,
      "step": 87
    },
    {
      "epoch": 0.0748936170212766,
      "grad_norm": 0.8488941788673401,
      "learning_rate": 6.387014543809223e-05,
      "loss": 1.3181,
      "step": 88
    },
    {
      "epoch": 0.07574468085106383,
      "grad_norm": 0.8375950455665588,
      "learning_rate": 6.307399704769099e-05,
      "loss": 1.3133,
      "step": 89
    },
    {
      "epoch": 0.07659574468085106,
      "grad_norm": 0.8652522563934326,
      "learning_rate": 6.227427435703997e-05,
      "loss": 1.3316,
      "step": 90
    },
    {
      "epoch": 0.0774468085106383,
      "grad_norm": 0.9034956097602844,
      "learning_rate": 6.147119600233758e-05,
      "loss": 1.3505,
      "step": 91
    },
    {
      "epoch": 0.07829787234042553,
      "grad_norm": 0.893261194229126,
      "learning_rate": 6.066498153718735e-05,
      "loss": 1.231,
      "step": 92
    },
    {
      "epoch": 0.07914893617021276,
      "grad_norm": 0.7510685324668884,
      "learning_rate": 5.985585137257401e-05,
      "loss": 1.2056,
      "step": 93
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7444056868553162,
      "learning_rate": 5.90440267166055e-05,
      "loss": 1.1075,
      "step": 94
    },
    {
      "epoch": 0.08085106382978724,
      "grad_norm": 0.7763978838920593,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 1.1477,
      "step": 95
    },
    {
      "epoch": 0.08170212765957446,
      "grad_norm": 0.7534809708595276,
      "learning_rate": 5.74131823855921e-05,
      "loss": 1.1346,
      "step": 96
    },
    {
      "epoch": 0.0825531914893617,
      "grad_norm": 0.7667890191078186,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 1.1447,
      "step": 97
    },
    {
      "epoch": 0.08340425531914894,
      "grad_norm": 0.7440046072006226,
      "learning_rate": 5.577423184847932e-05,
      "loss": 1.1686,
      "step": 98
    },
    {
      "epoch": 0.08425531914893616,
      "grad_norm": 0.7858494520187378,
      "learning_rate": 5.495227651252315e-05,
      "loss": 1.1549,
      "step": 99
    },
    {
      "epoch": 0.0851063829787234,
      "grad_norm": 0.9045118093490601,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 1.1898,
      "step": 100
    },
    {
      "epoch": 0.0851063829787234,
      "eval_loss": 1.3029175996780396,
      "eval_runtime": 159.5546,
      "eval_samples_per_second": 12.403,
      "eval_steps_per_second": 3.102,
      "step": 100
    },
    {
      "epoch": 0.08595744680851064,
      "grad_norm": 0.6465333104133606,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.4366,
      "step": 101
    },
    {
      "epoch": 0.08680851063829788,
      "grad_norm": 0.6552577614784241,
      "learning_rate": 5.247918773366112e-05,
      "loss": 1.4214,
      "step": 102
    },
    {
      "epoch": 0.0876595744680851,
      "grad_norm": 0.6319261789321899,
      "learning_rate": 5.165316846586541e-05,
      "loss": 1.349,
      "step": 103
    },
    {
      "epoch": 0.08851063829787234,
      "grad_norm": 0.6403036713600159,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 1.3772,
      "step": 104
    },
    {
      "epoch": 0.08936170212765958,
      "grad_norm": 0.6934713125228882,
      "learning_rate": 5e-05,
      "loss": 1.4058,
      "step": 105
    },
    {
      "epoch": 0.0902127659574468,
      "grad_norm": 0.6659963726997375,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.4235,
      "step": 106
    },
    {
      "epoch": 0.09106382978723404,
      "grad_norm": 0.6600536704063416,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.4097,
      "step": 107
    },
    {
      "epoch": 0.09191489361702128,
      "grad_norm": 0.7010683417320251,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 1.3924,
      "step": 108
    },
    {
      "epoch": 0.09276595744680852,
      "grad_norm": 0.6982723474502563,
      "learning_rate": 4.669547078371504e-05,
      "loss": 1.4562,
      "step": 109
    },
    {
      "epoch": 0.09361702127659574,
      "grad_norm": 0.7002760767936707,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 1.3803,
      "step": 110
    },
    {
      "epoch": 0.09446808510638298,
      "grad_norm": 0.6749926805496216,
      "learning_rate": 4.504772348747687e-05,
      "loss": 1.3515,
      "step": 111
    },
    {
      "epoch": 0.09531914893617022,
      "grad_norm": 0.7307199239730835,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 1.3318,
      "step": 112
    },
    {
      "epoch": 0.09617021276595744,
      "grad_norm": 0.7253100275993347,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 1.374,
      "step": 113
    },
    {
      "epoch": 0.09702127659574468,
      "grad_norm": 0.7308505773544312,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 1.2943,
      "step": 114
    },
    {
      "epoch": 0.09787234042553192,
      "grad_norm": 0.7035881876945496,
      "learning_rate": 4.17702704859633e-05,
      "loss": 1.3687,
      "step": 115
    },
    {
      "epoch": 0.09872340425531916,
      "grad_norm": 0.7168084979057312,
      "learning_rate": 4.095597328339452e-05,
      "loss": 1.2595,
      "step": 116
    },
    {
      "epoch": 0.09957446808510638,
      "grad_norm": 0.715908408164978,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 1.3488,
      "step": 117
    },
    {
      "epoch": 0.10042553191489362,
      "grad_norm": 0.7147959470748901,
      "learning_rate": 3.933501846281267e-05,
      "loss": 1.2683,
      "step": 118
    },
    {
      "epoch": 0.10127659574468086,
      "grad_norm": 0.7234799861907959,
      "learning_rate": 3.852880399766243e-05,
      "loss": 1.2676,
      "step": 119
    },
    {
      "epoch": 0.10212765957446808,
      "grad_norm": 0.7318155765533447,
      "learning_rate": 3.772572564296005e-05,
      "loss": 1.4023,
      "step": 120
    },
    {
      "epoch": 0.10297872340425532,
      "grad_norm": 0.740513801574707,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 1.2806,
      "step": 121
    },
    {
      "epoch": 0.10382978723404256,
      "grad_norm": 0.7200966477394104,
      "learning_rate": 3.612985456190778e-05,
      "loss": 1.2499,
      "step": 122
    },
    {
      "epoch": 0.1046808510638298,
      "grad_norm": 0.7381619215011597,
      "learning_rate": 3.533749813077677e-05,
      "loss": 1.2613,
      "step": 123
    },
    {
      "epoch": 0.10553191489361702,
      "grad_norm": 0.7533549070358276,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.3025,
      "step": 124
    },
    {
      "epoch": 0.10638297872340426,
      "grad_norm": 0.7795563340187073,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 1.3883,
      "step": 125
    },
    {
      "epoch": 0.1072340425531915,
      "grad_norm": 0.7705385684967041,
      "learning_rate": 3.298534127791785e-05,
      "loss": 1.3243,
      "step": 126
    },
    {
      "epoch": 0.10808510638297872,
      "grad_norm": 0.7889993786811829,
      "learning_rate": 3.221030765387417e-05,
      "loss": 1.3566,
      "step": 127
    },
    {
      "epoch": 0.10893617021276596,
      "grad_norm": 0.7757228016853333,
      "learning_rate": 3.144013755408895e-05,
      "loss": 1.356,
      "step": 128
    },
    {
      "epoch": 0.1097872340425532,
      "grad_norm": 0.7150682806968689,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 1.1898,
      "step": 129
    },
    {
      "epoch": 0.11063829787234042,
      "grad_norm": 0.7610926628112793,
      "learning_rate": 2.991522876735154e-05,
      "loss": 1.2685,
      "step": 130
    },
    {
      "epoch": 0.11148936170212766,
      "grad_norm": 0.7644234895706177,
      "learning_rate": 2.916090697523549e-05,
      "loss": 1.313,
      "step": 131
    },
    {
      "epoch": 0.1123404255319149,
      "grad_norm": 0.7740320563316345,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 1.381,
      "step": 132
    },
    {
      "epoch": 0.11319148936170213,
      "grad_norm": 0.7785450220108032,
      "learning_rate": 2.766955965735968e-05,
      "loss": 1.3293,
      "step": 133
    },
    {
      "epoch": 0.11404255319148936,
      "grad_norm": 0.7868967652320862,
      "learning_rate": 2.693294185106562e-05,
      "loss": 1.3033,
      "step": 134
    },
    {
      "epoch": 0.1148936170212766,
      "grad_norm": 0.7834706902503967,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 1.322,
      "step": 135
    },
    {
      "epoch": 0.11574468085106383,
      "grad_norm": 0.7756061553955078,
      "learning_rate": 2.547882480847461e-05,
      "loss": 1.3527,
      "step": 136
    },
    {
      "epoch": 0.11659574468085106,
      "grad_norm": 0.7716322541236877,
      "learning_rate": 2.476172311325783e-05,
      "loss": 1.2393,
      "step": 137
    },
    {
      "epoch": 0.1174468085106383,
      "grad_norm": 0.8269044756889343,
      "learning_rate": 2.405152131093926e-05,
      "loss": 1.2158,
      "step": 138
    },
    {
      "epoch": 0.11829787234042553,
      "grad_norm": 0.8297855257987976,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 1.3039,
      "step": 139
    },
    {
      "epoch": 0.11914893617021277,
      "grad_norm": 0.8258456587791443,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 1.3295,
      "step": 140
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8581304550170898,
      "learning_rate": 2.196424713241637e-05,
      "loss": 1.3879,
      "step": 141
    },
    {
      "epoch": 0.12085106382978723,
      "grad_norm": 0.8501140475273132,
      "learning_rate": 2.128356686585282e-05,
      "loss": 1.3223,
      "step": 142
    },
    {
      "epoch": 0.12170212765957447,
      "grad_norm": 0.7281916737556458,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.1589,
      "step": 143
    },
    {
      "epoch": 0.1225531914893617,
      "grad_norm": 0.7303036451339722,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 1.1733,
      "step": 144
    },
    {
      "epoch": 0.12340425531914893,
      "grad_norm": 0.7178941965103149,
      "learning_rate": 1.928936436551661e-05,
      "loss": 1.1137,
      "step": 145
    },
    {
      "epoch": 0.12425531914893617,
      "grad_norm": 0.7589569687843323,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 1.1218,
      "step": 146
    },
    {
      "epoch": 0.1251063829787234,
      "grad_norm": 0.7542309165000916,
      "learning_rate": 1.800157297483417e-05,
      "loss": 1.0927,
      "step": 147
    },
    {
      "epoch": 0.12595744680851065,
      "grad_norm": 0.7093368172645569,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 1.0782,
      "step": 148
    },
    {
      "epoch": 0.12680851063829787,
      "grad_norm": 0.7346543073654175,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 1.1137,
      "step": 149
    },
    {
      "epoch": 0.1276595744680851,
      "grad_norm": 0.9323696494102478,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 1.1791,
      "step": 150
    },
    {
      "epoch": 0.1276595744680851,
      "eval_loss": 1.297571063041687,
      "eval_runtime": 159.5915,
      "eval_samples_per_second": 12.4,
      "eval_steps_per_second": 3.102,
      "step": 150
    },
    {
      "epoch": 0.12851063829787235,
      "grad_norm": 1.0139029026031494,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.4873,
      "step": 151
    },
    {
      "epoch": 0.12936170212765957,
      "grad_norm": 0.72251296043396,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.4423,
      "step": 152
    },
    {
      "epoch": 0.1302127659574468,
      "grad_norm": 0.7016530632972717,
      "learning_rate": 1.435357758543015e-05,
      "loss": 1.3479,
      "step": 153
    },
    {
      "epoch": 0.13106382978723405,
      "grad_norm": 0.7013909220695496,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 1.4226,
      "step": 154
    },
    {
      "epoch": 0.13191489361702127,
      "grad_norm": 0.7218390107154846,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.4675,
      "step": 155
    },
    {
      "epoch": 0.1327659574468085,
      "grad_norm": 0.6750871539115906,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 1.3774,
      "step": 156
    },
    {
      "epoch": 0.13361702127659575,
      "grad_norm": 0.6665737628936768,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.3668,
      "step": 157
    },
    {
      "epoch": 0.13446808510638297,
      "grad_norm": 0.7015026807785034,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.402,
      "step": 158
    },
    {
      "epoch": 0.13531914893617022,
      "grad_norm": 0.6711598634719849,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 1.4126,
      "step": 159
    },
    {
      "epoch": 0.13617021276595745,
      "grad_norm": 0.6730396151542664,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 1.3807,
      "step": 160
    },
    {
      "epoch": 0.13702127659574467,
      "grad_norm": 0.669445276260376,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 1.3275,
      "step": 161
    },
    {
      "epoch": 0.13787234042553193,
      "grad_norm": 0.6813521981239319,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.3058,
      "step": 162
    },
    {
      "epoch": 0.13872340425531915,
      "grad_norm": 0.6964238286018372,
      "learning_rate": 9.068759265665384e-06,
      "loss": 1.2427,
      "step": 163
    },
    {
      "epoch": 0.13957446808510637,
      "grad_norm": 0.7479780316352844,
      "learning_rate": 8.599558442598998e-06,
      "loss": 1.4141,
      "step": 164
    },
    {
      "epoch": 0.14042553191489363,
      "grad_norm": 0.7196533679962158,
      "learning_rate": 8.141676086873572e-06,
      "loss": 1.3808,
      "step": 165
    },
    {
      "epoch": 0.14127659574468085,
      "grad_norm": 0.7272875308990479,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.3039,
      "step": 166
    },
    {
      "epoch": 0.14212765957446807,
      "grad_norm": 0.7312141060829163,
      "learning_rate": 7.260364370723044e-06,
      "loss": 1.2731,
      "step": 167
    },
    {
      "epoch": 0.14297872340425533,
      "grad_norm": 0.7293948531150818,
      "learning_rate": 6.837175952121306e-06,
      "loss": 1.2566,
      "step": 168
    },
    {
      "epoch": 0.14382978723404255,
      "grad_norm": 0.7320953607559204,
      "learning_rate": 6.425787818636131e-06,
      "loss": 1.2843,
      "step": 169
    },
    {
      "epoch": 0.14468085106382977,
      "grad_norm": 1.2721083164215088,
      "learning_rate": 6.026312439675552e-06,
      "loss": 1.33,
      "step": 170
    },
    {
      "epoch": 0.14553191489361703,
      "grad_norm": 0.7685941457748413,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 1.3429,
      "step": 171
    },
    {
      "epoch": 0.14638297872340425,
      "grad_norm": 0.7606873512268066,
      "learning_rate": 5.263533508961827e-06,
      "loss": 1.2729,
      "step": 172
    },
    {
      "epoch": 0.1472340425531915,
      "grad_norm": 0.7444884777069092,
      "learning_rate": 4.900438493352055e-06,
      "loss": 1.266,
      "step": 173
    },
    {
      "epoch": 0.14808510638297873,
      "grad_norm": 0.8201679587364197,
      "learning_rate": 4.549673247541875e-06,
      "loss": 1.3772,
      "step": 174
    },
    {
      "epoch": 0.14893617021276595,
      "grad_norm": 0.8054002523422241,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 1.3901,
      "step": 175
    },
    {
      "epoch": 0.1497872340425532,
      "grad_norm": 0.7705247402191162,
      "learning_rate": 3.885512251130763e-06,
      "loss": 1.2899,
      "step": 176
    },
    {
      "epoch": 0.15063829787234043,
      "grad_norm": 0.764252245426178,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 1.3036,
      "step": 177
    },
    {
      "epoch": 0.15148936170212765,
      "grad_norm": 0.8219606280326843,
      "learning_rate": 3.271776770026963e-06,
      "loss": 1.4145,
      "step": 178
    },
    {
      "epoch": 0.1523404255319149,
      "grad_norm": 0.7749344706535339,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 1.2505,
      "step": 179
    },
    {
      "epoch": 0.15319148936170213,
      "grad_norm": 0.8066408634185791,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 1.3183,
      "step": 180
    },
    {
      "epoch": 0.15404255319148935,
      "grad_norm": 0.8209425806999207,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.3196,
      "step": 181
    },
    {
      "epoch": 0.1548936170212766,
      "grad_norm": 0.8286488652229309,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 1.3868,
      "step": 182
    },
    {
      "epoch": 0.15574468085106383,
      "grad_norm": 0.8415066599845886,
      "learning_rate": 1.962316193157593e-06,
      "loss": 1.2648,
      "step": 183
    },
    {
      "epoch": 0.15659574468085105,
      "grad_norm": 0.8078544735908508,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 1.3087,
      "step": 184
    },
    {
      "epoch": 0.1574468085106383,
      "grad_norm": 0.8198104500770569,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 1.2721,
      "step": 185
    },
    {
      "epoch": 0.15829787234042553,
      "grad_norm": 0.8817468285560608,
      "learning_rate": 1.333670137599713e-06,
      "loss": 1.4314,
      "step": 186
    },
    {
      "epoch": 0.15914893617021278,
      "grad_norm": 0.8821167945861816,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 1.3568,
      "step": 187
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8515989780426025,
      "learning_rate": 9.810017062595322e-07,
      "loss": 1.242,
      "step": 188
    },
    {
      "epoch": 0.16085106382978723,
      "grad_norm": 0.9171628355979919,
      "learning_rate": 8.247462563808817e-07,
      "loss": 1.3265,
      "step": 189
    },
    {
      "epoch": 0.16170212765957448,
      "grad_norm": 1.0031181573867798,
      "learning_rate": 6.819348298638839e-07,
      "loss": 1.4505,
      "step": 190
    },
    {
      "epoch": 0.1625531914893617,
      "grad_norm": 1.0394679307937622,
      "learning_rate": 5.526064699265753e-07,
      "loss": 1.2682,
      "step": 191
    },
    {
      "epoch": 0.16340425531914893,
      "grad_norm": 0.8244553804397583,
      "learning_rate": 4.367965336512403e-07,
      "loss": 1.2625,
      "step": 192
    },
    {
      "epoch": 0.16425531914893618,
      "grad_norm": 0.7695810794830322,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 1.1723,
      "step": 193
    },
    {
      "epoch": 0.1651063829787234,
      "grad_norm": 0.7537410259246826,
      "learning_rate": 2.458548727494292e-07,
      "loss": 1.0678,
      "step": 194
    },
    {
      "epoch": 0.16595744680851063,
      "grad_norm": 0.7461527585983276,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 1.1333,
      "step": 195
    },
    {
      "epoch": 0.16680851063829788,
      "grad_norm": 0.7646304368972778,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 1.1396,
      "step": 196
    },
    {
      "epoch": 0.1676595744680851,
      "grad_norm": 0.6909725069999695,
      "learning_rate": 6.150154258476315e-08,
      "loss": 1.1358,
      "step": 197
    },
    {
      "epoch": 0.16851063829787233,
      "grad_norm": 0.752728283405304,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 1.1791,
      "step": 198
    },
    {
      "epoch": 0.16936170212765958,
      "grad_norm": 0.786652684211731,
      "learning_rate": 6.834750376549792e-09,
      "loss": 1.1234,
      "step": 199
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 0.7679680585861206,
      "learning_rate": 0.0,
      "loss": 1.1415,
      "step": 200
    },
    {
      "epoch": 0.1702127659574468,
      "eval_loss": 1.2877659797668457,
      "eval_runtime": 159.3703,
      "eval_samples_per_second": 12.418,
      "eval_steps_per_second": 3.106,
      "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": 1.5466286572948685e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}