{
  "best_metric": 2.357272148132324,
  "best_model_checkpoint": "miner_id_24/checkpoint-200",
  "epoch": 0.2290294875465216,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001145147437732608,
      "grad_norm": 1.1857813596725464,
      "learning_rate": 1e-05,
      "loss": 1.7224,
      "step": 1
    },
    {
      "epoch": 0.001145147437732608,
      "eval_loss": 3.6989080905914307,
      "eval_runtime": 65.4552,
      "eval_samples_per_second": 22.473,
      "eval_steps_per_second": 5.622,
      "step": 1
    },
    {
      "epoch": 0.002290294875465216,
      "grad_norm": 1.3954263925552368,
      "learning_rate": 2e-05,
      "loss": 1.9717,
      "step": 2
    },
    {
      "epoch": 0.0034354423131978244,
      "grad_norm": 1.4567511081695557,
      "learning_rate": 3e-05,
      "loss": 1.9951,
      "step": 3
    },
    {
      "epoch": 0.004580589750930432,
      "grad_norm": 1.4046063423156738,
      "learning_rate": 4e-05,
      "loss": 2.0887,
      "step": 4
    },
    {
      "epoch": 0.005725737188663041,
      "grad_norm": 1.3008391857147217,
      "learning_rate": 5e-05,
      "loss": 2.0594,
      "step": 5
    },
    {
      "epoch": 0.006870884626395649,
      "grad_norm": 1.3616421222686768,
      "learning_rate": 6e-05,
      "loss": 2.056,
      "step": 6
    },
    {
      "epoch": 0.008016032064128256,
      "grad_norm": 1.400654911994934,
      "learning_rate": 7e-05,
      "loss": 1.9557,
      "step": 7
    },
    {
      "epoch": 0.009161179501860864,
      "grad_norm": 1.3775149583816528,
      "learning_rate": 8e-05,
      "loss": 1.8892,
      "step": 8
    },
    {
      "epoch": 0.010306326939593472,
      "grad_norm": 1.3819791078567505,
      "learning_rate": 9e-05,
      "loss": 2.197,
      "step": 9
    },
    {
      "epoch": 0.011451474377326082,
      "grad_norm": 1.2725481986999512,
      "learning_rate": 0.0001,
      "loss": 1.9847,
      "step": 10
    },
    {
      "epoch": 0.01259662181505869,
      "grad_norm": 1.3934235572814941,
      "learning_rate": 9.999316524962345e-05,
      "loss": 2.3128,
      "step": 11
    },
    {
      "epoch": 0.013741769252791297,
      "grad_norm": 1.3472585678100586,
      "learning_rate": 9.997266286704631e-05,
      "loss": 2.2015,
      "step": 12
    },
    {
      "epoch": 0.014886916690523905,
      "grad_norm": 1.4502924680709839,
      "learning_rate": 9.993849845741524e-05,
      "loss": 2.2325,
      "step": 13
    },
    {
      "epoch": 0.01603206412825651,
      "grad_norm": 1.2517640590667725,
      "learning_rate": 9.989068136093873e-05,
      "loss": 2.2105,
      "step": 14
    },
    {
      "epoch": 0.01717721156598912,
      "grad_norm": 1.3751904964447021,
      "learning_rate": 9.98292246503335e-05,
      "loss": 2.162,
      "step": 15
    },
    {
      "epoch": 0.018322359003721728,
      "grad_norm": 1.3486522436141968,
      "learning_rate": 9.975414512725057e-05,
      "loss": 2.2555,
      "step": 16
    },
    {
      "epoch": 0.019467506441454337,
      "grad_norm": 1.432662010192871,
      "learning_rate": 9.966546331768191e-05,
      "loss": 2.3353,
      "step": 17
    },
    {
      "epoch": 0.020612653879186944,
      "grad_norm": 1.4271748065948486,
      "learning_rate": 9.956320346634876e-05,
      "loss": 2.3101,
      "step": 18
    },
    {
      "epoch": 0.021757801316919553,
      "grad_norm": 1.4949047565460205,
      "learning_rate": 9.944739353007344e-05,
      "loss": 2.1739,
      "step": 19
    },
    {
      "epoch": 0.022902948754652163,
      "grad_norm": 1.4840182065963745,
      "learning_rate": 9.931806517013612e-05,
      "loss": 2.193,
      "step": 20
    },
    {
      "epoch": 0.02404809619238477,
      "grad_norm": 1.5134001970291138,
      "learning_rate": 9.917525374361912e-05,
      "loss": 2.1547,
      "step": 21
    },
    {
      "epoch": 0.02519324363011738,
      "grad_norm": 1.5771504640579224,
      "learning_rate": 9.901899829374047e-05,
      "loss": 2.1195,
      "step": 22
    },
    {
      "epoch": 0.026338391067849985,
      "grad_norm": 1.6433401107788086,
      "learning_rate": 9.884934153917997e-05,
      "loss": 2.2778,
      "step": 23
    },
    {
      "epoch": 0.027483538505582595,
      "grad_norm": 1.7171975374221802,
      "learning_rate": 9.86663298624003e-05,
      "loss": 2.3879,
      "step": 24
    },
    {
      "epoch": 0.0286286859433152,
      "grad_norm": 1.7935868501663208,
      "learning_rate": 9.847001329696653e-05,
      "loss": 2.4686,
      "step": 25
    },
    {
      "epoch": 0.02977383338104781,
      "grad_norm": 1.8529566526412964,
      "learning_rate": 9.826044551386744e-05,
      "loss": 2.4731,
      "step": 26
    },
    {
      "epoch": 0.030918980818780417,
      "grad_norm": 2.0013041496276855,
      "learning_rate": 9.803768380684242e-05,
      "loss": 2.6302,
      "step": 27
    },
    {
      "epoch": 0.03206412825651302,
      "grad_norm": 2.0277631282806396,
      "learning_rate": 9.780178907671789e-05,
      "loss": 2.6628,
      "step": 28
    },
    {
      "epoch": 0.03320927569424564,
      "grad_norm": 2.2833051681518555,
      "learning_rate": 9.755282581475769e-05,
      "loss": 2.7178,
      "step": 29
    },
    {
      "epoch": 0.03435442313197824,
      "grad_norm": 2.2559874057769775,
      "learning_rate": 9.729086208503174e-05,
      "loss": 2.4511,
      "step": 30
    },
    {
      "epoch": 0.03549957056971085,
      "grad_norm": 2.4410758018493652,
      "learning_rate": 9.701596950580806e-05,
      "loss": 2.6788,
      "step": 31
    },
    {
      "epoch": 0.036644718007443455,
      "grad_norm": 2.849975824356079,
      "learning_rate": 9.672822322997305e-05,
      "loss": 2.8659,
      "step": 32
    },
    {
      "epoch": 0.03778986544517607,
      "grad_norm": 2.968794345855713,
      "learning_rate": 9.642770192448536e-05,
      "loss": 2.7125,
      "step": 33
    },
    {
      "epoch": 0.038935012882908675,
      "grad_norm": 2.913682460784912,
      "learning_rate": 9.611448774886924e-05,
      "loss": 2.6991,
      "step": 34
    },
    {
      "epoch": 0.04008016032064128,
      "grad_norm": 3.464346408843994,
      "learning_rate": 9.578866633275288e-05,
      "loss": 2.8769,
      "step": 35
    },
    {
      "epoch": 0.04122530775837389,
      "grad_norm": 3.3060381412506104,
      "learning_rate": 9.545032675245813e-05,
      "loss": 2.8485,
      "step": 36
    },
    {
      "epoch": 0.0423704551961065,
      "grad_norm": 3.114264726638794,
      "learning_rate": 9.509956150664796e-05,
      "loss": 2.6949,
      "step": 37
    },
    {
      "epoch": 0.04351560263383911,
      "grad_norm": 3.0583348274230957,
      "learning_rate": 9.473646649103818e-05,
      "loss": 2.5889,
      "step": 38
    },
    {
      "epoch": 0.04466075007157171,
      "grad_norm": 3.425013780593872,
      "learning_rate": 9.43611409721806e-05,
      "loss": 2.6171,
      "step": 39
    },
    {
      "epoch": 0.045805897509304326,
      "grad_norm": 3.5885226726531982,
      "learning_rate": 9.397368756032445e-05,
      "loss": 2.8257,
      "step": 40
    },
    {
      "epoch": 0.04695104494703693,
      "grad_norm": 3.600555658340454,
      "learning_rate": 9.357421218136386e-05,
      "loss": 2.6852,
      "step": 41
    },
    {
      "epoch": 0.04809619238476954,
      "grad_norm": 3.9245359897613525,
      "learning_rate": 9.316282404787871e-05,
      "loss": 2.618,
      "step": 42
    },
    {
      "epoch": 0.049241339822502145,
      "grad_norm": 3.791551113128662,
      "learning_rate": 9.273963562927695e-05,
      "loss": 2.8372,
      "step": 43
    },
    {
      "epoch": 0.05038648726023476,
      "grad_norm": 4.751186370849609,
      "learning_rate": 9.230476262104677e-05,
      "loss": 2.6304,
      "step": 44
    },
    {
      "epoch": 0.051531634697967364,
      "grad_norm": 4.537441253662109,
      "learning_rate": 9.185832391312644e-05,
      "loss": 2.9015,
      "step": 45
    },
    {
      "epoch": 0.05267678213569997,
      "grad_norm": 4.935320854187012,
      "learning_rate": 9.140044155740101e-05,
      "loss": 2.9615,
      "step": 46
    },
    {
      "epoch": 0.05382192957343258,
      "grad_norm": 4.6309685707092285,
      "learning_rate": 9.093124073433463e-05,
      "loss": 2.6443,
      "step": 47
    },
    {
      "epoch": 0.05496707701116519,
      "grad_norm": 5.758911609649658,
      "learning_rate": 9.045084971874738e-05,
      "loss": 3.107,
      "step": 48
    },
    {
      "epoch": 0.056112224448897796,
      "grad_norm": 6.037940502166748,
      "learning_rate": 8.995939984474624e-05,
      "loss": 2.7087,
      "step": 49
    },
    {
      "epoch": 0.0572573718866304,
      "grad_norm": 12.22794246673584,
      "learning_rate": 8.945702546981969e-05,
      "loss": 3.4838,
      "step": 50
    },
    {
      "epoch": 0.0572573718866304,
      "eval_loss": 2.821622610092163,
      "eval_runtime": 65.8284,
      "eval_samples_per_second": 22.346,
      "eval_steps_per_second": 5.59,
      "step": 50
    },
    {
      "epoch": 0.05840251932436301,
      "grad_norm": 2.9304637908935547,
      "learning_rate": 8.894386393810563e-05,
      "loss": 1.9821,
      "step": 51
    },
    {
      "epoch": 0.05954766676209562,
      "grad_norm": 2.759567975997925,
      "learning_rate": 8.842005554284296e-05,
      "loss": 2.1817,
      "step": 52
    },
    {
      "epoch": 0.06069281419982823,
      "grad_norm": 1.9241632223129272,
      "learning_rate": 8.788574348801675e-05,
      "loss": 2.1566,
      "step": 53
    },
    {
      "epoch": 0.061837961637560834,
      "grad_norm": 1.4464048147201538,
      "learning_rate": 8.73410738492077e-05,
      "loss": 2.1203,
      "step": 54
    },
    {
      "epoch": 0.06298310907529345,
      "grad_norm": 1.1769587993621826,
      "learning_rate": 8.678619553365659e-05,
      "loss": 2.1346,
      "step": 55
    },
    {
      "epoch": 0.06412825651302605,
      "grad_norm": 1.173683762550354,
      "learning_rate": 8.622126023955446e-05,
      "loss": 2.1986,
      "step": 56
    },
    {
      "epoch": 0.06527340395075866,
      "grad_norm": 1.0991641283035278,
      "learning_rate": 8.564642241456986e-05,
      "loss": 2.1195,
      "step": 57
    },
    {
      "epoch": 0.06641855138849127,
      "grad_norm": 1.0703307390213013,
      "learning_rate": 8.506183921362443e-05,
      "loss": 2.1276,
      "step": 58
    },
    {
      "epoch": 0.06756369882622387,
      "grad_norm": 1.115053415298462,
      "learning_rate": 8.44676704559283e-05,
      "loss": 2.2543,
      "step": 59
    },
    {
      "epoch": 0.06870884626395649,
      "grad_norm": 1.0648963451385498,
      "learning_rate": 8.386407858128706e-05,
      "loss": 2.0712,
      "step": 60
    },
    {
      "epoch": 0.0698539937016891,
      "grad_norm": 1.0940132141113281,
      "learning_rate": 8.32512286056924e-05,
      "loss": 1.9888,
      "step": 61
    },
    {
      "epoch": 0.0709991411394217,
      "grad_norm": 1.093697428703308,
      "learning_rate": 8.262928807620843e-05,
      "loss": 1.9231,
      "step": 62
    },
    {
      "epoch": 0.07214428857715431,
      "grad_norm": 1.1327342987060547,
      "learning_rate": 8.199842702516583e-05,
      "loss": 2.0183,
      "step": 63
    },
    {
      "epoch": 0.07328943601488691,
      "grad_norm": 1.164723515510559,
      "learning_rate": 8.135881792367686e-05,
      "loss": 2.1144,
      "step": 64
    },
    {
      "epoch": 0.07443458345261952,
      "grad_norm": 1.3183391094207764,
      "learning_rate": 8.07106356344834e-05,
      "loss": 2.2551,
      "step": 65
    },
    {
      "epoch": 0.07557973089035214,
      "grad_norm": 1.30685293674469,
      "learning_rate": 8.005405736415126e-05,
      "loss": 2.3425,
      "step": 66
    },
    {
      "epoch": 0.07672487832808474,
      "grad_norm": 1.3706049919128418,
      "learning_rate": 7.938926261462366e-05,
      "loss": 2.2358,
      "step": 67
    },
    {
      "epoch": 0.07787002576581735,
      "grad_norm": 1.4558371305465698,
      "learning_rate": 7.871643313414718e-05,
      "loss": 2.3912,
      "step": 68
    },
    {
      "epoch": 0.07901517320354996,
      "grad_norm": 1.4907410144805908,
      "learning_rate": 7.803575286758364e-05,
      "loss": 2.2448,
      "step": 69
    },
    {
      "epoch": 0.08016032064128256,
      "grad_norm": 1.441669225692749,
      "learning_rate": 7.734740790612136e-05,
      "loss": 2.2489,
      "step": 70
    },
    {
      "epoch": 0.08130546807901518,
      "grad_norm": 1.551479697227478,
      "learning_rate": 7.66515864363997e-05,
      "loss": 2.2794,
      "step": 71
    },
    {
      "epoch": 0.08245061551674777,
      "grad_norm": 1.6066415309906006,
      "learning_rate": 7.594847868906076e-05,
      "loss": 2.5086,
      "step": 72
    },
    {
      "epoch": 0.08359576295448039,
      "grad_norm": 1.6689355373382568,
      "learning_rate": 7.52382768867422e-05,
      "loss": 2.3816,
      "step": 73
    },
    {
      "epoch": 0.084740910392213,
      "grad_norm": 1.82950758934021,
      "learning_rate": 7.452117519152542e-05,
      "loss": 2.4888,
      "step": 74
    },
    {
      "epoch": 0.0858860578299456,
      "grad_norm": 1.8634968996047974,
      "learning_rate": 7.379736965185368e-05,
      "loss": 2.402,
      "step": 75
    },
    {
      "epoch": 0.08703120526767821,
      "grad_norm": 1.8409210443496704,
      "learning_rate": 7.30670581489344e-05,
      "loss": 2.622,
      "step": 76
    },
    {
      "epoch": 0.08817635270541083,
      "grad_norm": 1.9499852657318115,
      "learning_rate": 7.233044034264034e-05,
      "loss": 2.5458,
      "step": 77
    },
    {
      "epoch": 0.08932150014314343,
      "grad_norm": 2.113006353378296,
      "learning_rate": 7.158771761692464e-05,
      "loss": 2.6393,
      "step": 78
    },
    {
      "epoch": 0.09046664758087604,
      "grad_norm": 2.140016794204712,
      "learning_rate": 7.083909302476453e-05,
      "loss": 2.6392,
      "step": 79
    },
    {
      "epoch": 0.09161179501860865,
      "grad_norm": 2.4452154636383057,
      "learning_rate": 7.008477123264848e-05,
      "loss": 2.671,
      "step": 80
    },
    {
      "epoch": 0.09275694245634125,
      "grad_norm": 2.2606778144836426,
      "learning_rate": 6.932495846462261e-05,
      "loss": 2.6456,
      "step": 81
    },
    {
      "epoch": 0.09390208989407386,
      "grad_norm": 2.6132614612579346,
      "learning_rate": 6.855986244591104e-05,
      "loss": 2.9283,
      "step": 82
    },
    {
      "epoch": 0.09504723733180646,
      "grad_norm": 2.516116142272949,
      "learning_rate": 6.778969234612584e-05,
      "loss": 2.6288,
      "step": 83
    },
    {
      "epoch": 0.09619238476953908,
      "grad_norm": 2.5448191165924072,
      "learning_rate": 6.701465872208216e-05,
      "loss": 2.5907,
      "step": 84
    },
    {
      "epoch": 0.09733753220727169,
      "grad_norm": 2.860098123550415,
      "learning_rate": 6.623497346023418e-05,
      "loss": 2.7538,
      "step": 85
    },
    {
      "epoch": 0.09848267964500429,
      "grad_norm": 2.8702075481414795,
      "learning_rate": 6.545084971874738e-05,
      "loss": 2.7157,
      "step": 86
    },
    {
      "epoch": 0.0996278270827369,
      "grad_norm": 3.0975193977355957,
      "learning_rate": 6.466250186922325e-05,
      "loss": 2.8234,
      "step": 87
    },
    {
      "epoch": 0.10077297452046952,
      "grad_norm": 3.2177658081054688,
      "learning_rate": 6.387014543809223e-05,
      "loss": 2.6295,
      "step": 88
    },
    {
      "epoch": 0.10191812195820212,
      "grad_norm": 3.4743294715881348,
      "learning_rate": 6.307399704769099e-05,
      "loss": 2.7106,
      "step": 89
    },
    {
      "epoch": 0.10306326939593473,
      "grad_norm": 3.2153120040893555,
      "learning_rate": 6.227427435703997e-05,
      "loss": 2.6216,
      "step": 90
    },
    {
      "epoch": 0.10420841683366733,
      "grad_norm": 3.6161904335021973,
      "learning_rate": 6.147119600233758e-05,
      "loss": 2.7116,
      "step": 91
    },
    {
      "epoch": 0.10535356427139994,
      "grad_norm": 3.2433080673217773,
      "learning_rate": 6.066498153718735e-05,
      "loss": 2.3979,
      "step": 92
    },
    {
      "epoch": 0.10649871170913255,
      "grad_norm": 3.324376344680786,
      "learning_rate": 5.985585137257401e-05,
      "loss": 2.5326,
      "step": 93
    },
    {
      "epoch": 0.10764385914686515,
      "grad_norm": 4.034153461456299,
      "learning_rate": 5.90440267166055e-05,
      "loss": 2.7898,
      "step": 94
    },
    {
      "epoch": 0.10878900658459777,
      "grad_norm": 4.2742767333984375,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 2.6838,
      "step": 95
    },
    {
      "epoch": 0.10993415402233038,
      "grad_norm": 3.575446128845215,
      "learning_rate": 5.74131823855921e-05,
      "loss": 2.4527,
      "step": 96
    },
    {
      "epoch": 0.11107930146006298,
      "grad_norm": 4.2994585037231445,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 2.7888,
      "step": 97
    },
    {
      "epoch": 0.11222444889779559,
      "grad_norm": 4.459681987762451,
      "learning_rate": 5.577423184847932e-05,
      "loss": 2.7611,
      "step": 98
    },
    {
      "epoch": 0.1133695963355282,
      "grad_norm": 4.7320451736450195,
      "learning_rate": 5.495227651252315e-05,
      "loss": 2.5126,
      "step": 99
    },
    {
      "epoch": 0.1145147437732608,
      "grad_norm": 7.999973773956299,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 3.1713,
      "step": 100
    },
    {
      "epoch": 0.1145147437732608,
      "eval_loss": 2.678560495376587,
      "eval_runtime": 65.8932,
      "eval_samples_per_second": 22.324,
      "eval_steps_per_second": 5.585,
      "step": 100
    },
    {
      "epoch": 0.11565989121099342,
      "grad_norm": 2.2747390270233154,
      "learning_rate": 5.330452921628497e-05,
      "loss": 1.739,
      "step": 101
    },
    {
      "epoch": 0.11680503864872602,
      "grad_norm": 2.7442595958709717,
      "learning_rate": 5.247918773366112e-05,
      "loss": 2.0246,
      "step": 102
    },
    {
      "epoch": 0.11795018608645863,
      "grad_norm": 2.145855665206909,
      "learning_rate": 5.165316846586541e-05,
      "loss": 2.1851,
      "step": 103
    },
    {
      "epoch": 0.11909533352419124,
      "grad_norm": 1.670932412147522,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 2.0612,
      "step": 104
    },
    {
      "epoch": 0.12024048096192384,
      "grad_norm": 1.3505253791809082,
      "learning_rate": 5e-05,
      "loss": 1.9727,
      "step": 105
    },
    {
      "epoch": 0.12138562839965646,
      "grad_norm": 1.1177784204483032,
      "learning_rate": 4.917330276168208e-05,
      "loss": 1.8933,
      "step": 106
    },
    {
      "epoch": 0.12253077583738907,
      "grad_norm": 1.034048080444336,
      "learning_rate": 4.834683153413459e-05,
      "loss": 1.9658,
      "step": 107
    },
    {
      "epoch": 0.12367592327512167,
      "grad_norm": 1.026869773864746,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 2.0032,
      "step": 108
    },
    {
      "epoch": 0.12482107071285428,
      "grad_norm": 1.0370545387268066,
      "learning_rate": 4.669547078371504e-05,
      "loss": 2.008,
      "step": 109
    },
    {
      "epoch": 0.1259662181505869,
      "grad_norm": 1.0483537912368774,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 2.1004,
      "step": 110
    },
    {
      "epoch": 0.1271113655883195,
      "grad_norm": 1.120125651359558,
      "learning_rate": 4.504772348747687e-05,
      "loss": 2.1231,
      "step": 111
    },
    {
      "epoch": 0.1282565130260521,
      "grad_norm": 1.1653671264648438,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 2.2569,
      "step": 112
    },
    {
      "epoch": 0.1294016604637847,
      "grad_norm": 1.1242620944976807,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 2.0203,
      "step": 113
    },
    {
      "epoch": 0.13054680790151732,
      "grad_norm": 1.2052456140518188,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 2.1744,
      "step": 114
    },
    {
      "epoch": 0.13169195533924993,
      "grad_norm": 1.223109483718872,
      "learning_rate": 4.17702704859633e-05,
      "loss": 2.1724,
      "step": 115
    },
    {
      "epoch": 0.13283710277698255,
      "grad_norm": 1.2378007173538208,
      "learning_rate": 4.095597328339452e-05,
      "loss": 2.1344,
      "step": 116
    },
    {
      "epoch": 0.13398225021471513,
      "grad_norm": 1.347692608833313,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 2.2507,
      "step": 117
    },
    {
      "epoch": 0.13512739765244774,
      "grad_norm": 1.3898274898529053,
      "learning_rate": 3.933501846281267e-05,
      "loss": 2.3259,
      "step": 118
    },
    {
      "epoch": 0.13627254509018036,
      "grad_norm": 1.3375824689865112,
      "learning_rate": 3.852880399766243e-05,
      "loss": 2.0743,
      "step": 119
    },
    {
      "epoch": 0.13741769252791297,
      "grad_norm": 1.3329682350158691,
      "learning_rate": 3.772572564296005e-05,
      "loss": 2.0794,
      "step": 120
    },
    {
      "epoch": 0.13856283996564558,
      "grad_norm": 1.4277297258377075,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 2.3247,
      "step": 121
    },
    {
      "epoch": 0.1397079874033782,
      "grad_norm": 1.5384713411331177,
      "learning_rate": 3.612985456190778e-05,
      "loss": 2.1708,
      "step": 122
    },
    {
      "epoch": 0.14085313484111078,
      "grad_norm": 1.5506294965744019,
      "learning_rate": 3.533749813077677e-05,
      "loss": 2.3587,
      "step": 123
    },
    {
      "epoch": 0.1419982822788434,
      "grad_norm": 1.7492601871490479,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 2.5121,
      "step": 124
    },
    {
      "epoch": 0.143143429716576,
      "grad_norm": 1.741735577583313,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 2.4605,
      "step": 125
    },
    {
      "epoch": 0.14428857715430862,
      "grad_norm": 1.7958412170410156,
      "learning_rate": 3.298534127791785e-05,
      "loss": 2.487,
      "step": 126
    },
    {
      "epoch": 0.14543372459204124,
      "grad_norm": 1.818452000617981,
      "learning_rate": 3.221030765387417e-05,
      "loss": 2.4381,
      "step": 127
    },
    {
      "epoch": 0.14657887202977382,
      "grad_norm": 2.2684295177459717,
      "learning_rate": 3.144013755408895e-05,
      "loss": 2.5977,
      "step": 128
    },
    {
      "epoch": 0.14772401946750643,
      "grad_norm": 2.2433760166168213,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 2.5941,
      "step": 129
    },
    {
      "epoch": 0.14886916690523905,
      "grad_norm": 2.191103219985962,
      "learning_rate": 2.991522876735154e-05,
      "loss": 2.4504,
      "step": 130
    },
    {
      "epoch": 0.15001431434297166,
      "grad_norm": 2.3713674545288086,
      "learning_rate": 2.916090697523549e-05,
      "loss": 2.6064,
      "step": 131
    },
    {
      "epoch": 0.15115946178070427,
      "grad_norm": 2.6048524379730225,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 2.6528,
      "step": 132
    },
    {
      "epoch": 0.1523046092184369,
      "grad_norm": 2.446101188659668,
      "learning_rate": 2.766955965735968e-05,
      "loss": 2.6621,
      "step": 133
    },
    {
      "epoch": 0.15344975665616947,
      "grad_norm": 2.9291086196899414,
      "learning_rate": 2.693294185106562e-05,
      "loss": 2.6843,
      "step": 134
    },
    {
      "epoch": 0.15459490409390209,
      "grad_norm": 2.703549385070801,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 2.7264,
      "step": 135
    },
    {
      "epoch": 0.1557400515316347,
      "grad_norm": 2.70150089263916,
      "learning_rate": 2.547882480847461e-05,
      "loss": 2.5466,
      "step": 136
    },
    {
      "epoch": 0.1568851989693673,
      "grad_norm": 2.891256332397461,
      "learning_rate": 2.476172311325783e-05,
      "loss": 2.86,
      "step": 137
    },
    {
      "epoch": 0.15803034640709993,
      "grad_norm": 2.9282541275024414,
      "learning_rate": 2.405152131093926e-05,
      "loss": 2.6461,
      "step": 138
    },
    {
      "epoch": 0.1591754938448325,
      "grad_norm": 2.9176509380340576,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 2.8391,
      "step": 139
    },
    {
      "epoch": 0.16032064128256512,
      "grad_norm": 3.326256275177002,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 2.8153,
      "step": 140
    },
    {
      "epoch": 0.16146578872029774,
      "grad_norm": 4.399847030639648,
      "learning_rate": 2.196424713241637e-05,
      "loss": 2.6847,
      "step": 141
    },
    {
      "epoch": 0.16261093615803035,
      "grad_norm": 3.2391726970672607,
      "learning_rate": 2.128356686585282e-05,
      "loss": 2.5672,
      "step": 142
    },
    {
      "epoch": 0.16375608359576296,
      "grad_norm": 3.579580545425415,
      "learning_rate": 2.061073738537635e-05,
      "loss": 2.8398,
      "step": 143
    },
    {
      "epoch": 0.16490123103349555,
      "grad_norm": 8.842161178588867,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 2.9203,
      "step": 144
    },
    {
      "epoch": 0.16604637847122816,
      "grad_norm": 3.746894598007202,
      "learning_rate": 1.928936436551661e-05,
      "loss": 2.6962,
      "step": 145
    },
    {
      "epoch": 0.16719152590896078,
      "grad_norm": 4.396005630493164,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 2.9049,
      "step": 146
    },
    {
      "epoch": 0.1683366733466934,
      "grad_norm": 3.7322051525115967,
      "learning_rate": 1.800157297483417e-05,
      "loss": 2.422,
      "step": 147
    },
    {
      "epoch": 0.169481820784426,
      "grad_norm": 5.1746039390563965,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 3.179,
      "step": 148
    },
    {
      "epoch": 0.17062696822215861,
      "grad_norm": 5.416689395904541,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 2.5238,
      "step": 149
    },
    {
      "epoch": 0.1717721156598912,
      "grad_norm": 8.568948745727539,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 4.4769,
      "step": 150
    },
    {
      "epoch": 0.1717721156598912,
      "eval_loss": 2.3816401958465576,
      "eval_runtime": 65.8902,
      "eval_samples_per_second": 22.325,
      "eval_steps_per_second": 5.585,
      "step": 150
    },
    {
      "epoch": 0.1729172630976238,
      "grad_norm": 0.9151042699813843,
      "learning_rate": 1.553232954407171e-05,
      "loss": 1.7747,
      "step": 151
    },
    {
      "epoch": 0.17406241053535643,
      "grad_norm": 1.0841840505599976,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 1.7702,
      "step": 152
    },
    {
      "epoch": 0.17520755797308904,
      "grad_norm": 1.1267786026000977,
      "learning_rate": 1.435357758543015e-05,
      "loss": 2.0451,
      "step": 153
    },
    {
      "epoch": 0.17635270541082165,
      "grad_norm": 1.1299494504928589,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 2.0143,
      "step": 154
    },
    {
      "epoch": 0.17749785284855424,
      "grad_norm": 1.1024436950683594,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 1.8367,
      "step": 155
    },
    {
      "epoch": 0.17864300028628685,
      "grad_norm": 1.0896661281585693,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 2.1604,
      "step": 156
    },
    {
      "epoch": 0.17978814772401946,
      "grad_norm": 0.98717200756073,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 1.8008,
      "step": 157
    },
    {
      "epoch": 0.18093329516175208,
      "grad_norm": 0.9965742230415344,
      "learning_rate": 1.157994445715706e-05,
      "loss": 1.9224,
      "step": 158
    },
    {
      "epoch": 0.1820784425994847,
      "grad_norm": 1.0607885122299194,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 2.0298,
      "step": 159
    },
    {
      "epoch": 0.1832235900372173,
      "grad_norm": 1.0798964500427246,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 2.1195,
      "step": 160
    },
    {
      "epoch": 0.1843687374749499,
      "grad_norm": 1.1053290367126465,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 2.0512,
      "step": 161
    },
    {
      "epoch": 0.1855138849126825,
      "grad_norm": 1.1297589540481567,
      "learning_rate": 9.549150281252633e-06,
      "loss": 2.1221,
      "step": 162
    },
    {
      "epoch": 0.18665903235041512,
      "grad_norm": 1.148611307144165,
      "learning_rate": 9.068759265665384e-06,
      "loss": 2.2351,
      "step": 163
    },
    {
      "epoch": 0.18780417978814773,
      "grad_norm": 1.2360886335372925,
      "learning_rate": 8.599558442598998e-06,
      "loss": 2.3199,
      "step": 164
    },
    {
      "epoch": 0.18894932722588034,
      "grad_norm": 1.2181810140609741,
      "learning_rate": 8.141676086873572e-06,
      "loss": 2.2741,
      "step": 165
    },
    {
      "epoch": 0.19009447466361293,
      "grad_norm": 1.2005771398544312,
      "learning_rate": 7.695237378953223e-06,
      "loss": 1.9889,
      "step": 166
    },
    {
      "epoch": 0.19123962210134554,
      "grad_norm": 1.393108606338501,
      "learning_rate": 7.260364370723044e-06,
      "loss": 2.3323,
      "step": 167
    },
    {
      "epoch": 0.19238476953907815,
      "grad_norm": 1.4250054359436035,
      "learning_rate": 6.837175952121306e-06,
      "loss": 2.1548,
      "step": 168
    },
    {
      "epoch": 0.19352991697681077,
      "grad_norm": 1.494267225265503,
      "learning_rate": 6.425787818636131e-06,
      "loss": 2.3836,
      "step": 169
    },
    {
      "epoch": 0.19467506441454338,
      "grad_norm": 1.4644261598587036,
      "learning_rate": 6.026312439675552e-06,
      "loss": 2.1691,
      "step": 170
    },
    {
      "epoch": 0.195820211852276,
      "grad_norm": 1.645621657371521,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 2.3764,
      "step": 171
    },
    {
      "epoch": 0.19696535929000858,
      "grad_norm": 1.5763808488845825,
      "learning_rate": 5.263533508961827e-06,
      "loss": 2.3535,
      "step": 172
    },
    {
      "epoch": 0.1981105067277412,
      "grad_norm": 1.6407052278518677,
      "learning_rate": 4.900438493352055e-06,
      "loss": 2.2467,
      "step": 173
    },
    {
      "epoch": 0.1992556541654738,
      "grad_norm": 1.699481725692749,
      "learning_rate": 4.549673247541875e-06,
      "loss": 2.2528,
      "step": 174
    },
    {
      "epoch": 0.20040080160320642,
      "grad_norm": 1.749120831489563,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 2.1888,
      "step": 175
    },
    {
      "epoch": 0.20154594904093903,
      "grad_norm": 1.974062442779541,
      "learning_rate": 3.885512251130763e-06,
      "loss": 2.6223,
      "step": 176
    },
    {
      "epoch": 0.20269109647867162,
      "grad_norm": 1.960533618927002,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 2.5293,
      "step": 177
    },
    {
      "epoch": 0.20383624391640423,
      "grad_norm": 2.0967042446136475,
      "learning_rate": 3.271776770026963e-06,
      "loss": 2.581,
      "step": 178
    },
    {
      "epoch": 0.20498139135413684,
      "grad_norm": 2.1972668170928955,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 2.5935,
      "step": 179
    },
    {
      "epoch": 0.20612653879186946,
      "grad_norm": 2.275275468826294,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 2.4806,
      "step": 180
    },
    {
      "epoch": 0.20727168622960207,
      "grad_norm": 2.3416149616241455,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 2.5221,
      "step": 181
    },
    {
      "epoch": 0.20841683366733466,
      "grad_norm": 2.363574266433716,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 2.6002,
      "step": 182
    },
    {
      "epoch": 0.20956198110506727,
      "grad_norm": 2.5128014087677,
      "learning_rate": 1.962316193157593e-06,
      "loss": 2.6096,
      "step": 183
    },
    {
      "epoch": 0.21070712854279988,
      "grad_norm": 2.7236878871917725,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 2.7113,
      "step": 184
    },
    {
      "epoch": 0.2118522759805325,
      "grad_norm": 2.779812812805176,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 2.6452,
      "step": 185
    },
    {
      "epoch": 0.2129974234182651,
      "grad_norm": 2.9389443397521973,
      "learning_rate": 1.333670137599713e-06,
      "loss": 2.6942,
      "step": 186
    },
    {
      "epoch": 0.21414257085599772,
      "grad_norm": 2.8551394939422607,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 2.6672,
      "step": 187
    },
    {
      "epoch": 0.2152877182937303,
      "grad_norm": 2.7161619663238525,
      "learning_rate": 9.810017062595322e-07,
      "loss": 2.3814,
      "step": 188
    },
    {
      "epoch": 0.21643286573146292,
      "grad_norm": 3.163106679916382,
      "learning_rate": 8.247462563808817e-07,
      "loss": 2.6353,
      "step": 189
    },
    {
      "epoch": 0.21757801316919553,
      "grad_norm": 3.059785842895508,
      "learning_rate": 6.819348298638839e-07,
      "loss": 2.5194,
      "step": 190
    },
    {
      "epoch": 0.21872316060692815,
      "grad_norm": 2.919477701187134,
      "learning_rate": 5.526064699265753e-07,
      "loss": 2.6343,
      "step": 191
    },
    {
      "epoch": 0.21986830804466076,
      "grad_norm": 3.242403030395508,
      "learning_rate": 4.367965336512403e-07,
      "loss": 2.5439,
      "step": 192
    },
    {
      "epoch": 0.22101345548239335,
      "grad_norm": 3.444347381591797,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 2.4965,
      "step": 193
    },
    {
      "epoch": 0.22215860292012596,
      "grad_norm": 3.5802149772644043,
      "learning_rate": 2.458548727494292e-07,
      "loss": 2.5986,
      "step": 194
    },
    {
      "epoch": 0.22330375035785857,
      "grad_norm": 3.4884448051452637,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 2.6199,
      "step": 195
    },
    {
      "epoch": 0.22444889779559118,
      "grad_norm": 4.175929069519043,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 3.0117,
      "step": 196
    },
    {
      "epoch": 0.2255940452333238,
      "grad_norm": 3.9975297451019287,
      "learning_rate": 6.150154258476315e-08,
      "loss": 2.9959,
      "step": 197
    },
    {
      "epoch": 0.2267391926710564,
      "grad_norm": 4.517087936401367,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 2.9329,
      "step": 198
    },
    {
      "epoch": 0.227884340108789,
      "grad_norm": 5.685300827026367,
      "learning_rate": 6.834750376549792e-09,
      "loss": 2.8613,
      "step": 199
    },
    {
      "epoch": 0.2290294875465216,
      "grad_norm": 8.818440437316895,
      "learning_rate": 0.0,
      "loss": 3.3088,
      "step": 200
    },
    {
      "epoch": 0.2290294875465216,
      "eval_loss": 2.357272148132324,
      "eval_runtime": 65.8561,
      "eval_samples_per_second": 22.337,
      "eval_steps_per_second": 5.588,
      "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": 8.28736052133888e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}